TESTING ppc build based on known good base revision

This commit is actually a cherry-pick of #6643 but for testing purposes
only (see Issue #6653)

.circleci/config.yml

@@ -0,0 +1,56 @@
+version: 2.1
+
+orbs:
+  win: circleci/windows@2.4.0
+
+executors:
+  windows-2xlarge:
+    machine:
+      image: 'windows-server-2019-vs2019:stable'
+      resource_class: windows.2xlarge
+      shell: bash.exe
+
+jobs:
+  build:
+    executor: windows-2xlarge
+    
+    environment:
+      THIRDPARTY_HOME: C:/Users/circleci/thirdparty
+      CMAKE_HOME: C:/Users/circleci/thirdparty/cmake-3.16.4-win64-x64
+      CMAKE_BIN: C:/Users/circleci/thirdparty/cmake-3.16.4-win64-x64/bin/cmake.exe
+      CMAKE_GENERATOR: Visual Studio 16 2019
+      SNAPPY_HOME: C:/Users/circleci/thirdparty/snappy-1.1.7
+      SNAPPY_INCLUDE: C:/Users/circleci/thirdparty/snappy-1.1.7;C:/Users/circleci/thirdparty/snappy-1.1.7/build
+      SNAPPY_LIB_DEBUG: C:/Users/circleci/thirdparty/snappy-1.1.7/build/Debug/snappy.lib
+
+    steps:
+      - checkout
+      - run:
+          name: "Install thirdparty dependencies"
+          command: |
+            mkdir ${THIRDPARTY_HOME}
+            cd ${THIRDPARTY_HOME}
+            echo "Installing CMake..."
+            curl --fail --silent --show-error --output cmake-3.16.4-win64-x64.zip --location https://github.com/Kitware/CMake/releases/download/v3.16.4/cmake-3.16.4-win64-x64.zip
+            unzip -q cmake-3.16.4-win64-x64.zip
+            echo "Building Snappy dependency..."
+            curl --fail --silent --show-error --output snappy-1.1.7.zip --location https://github.com/google/snappy/archive/1.1.7.zip
+            unzip -q snappy-1.1.7.zip
+            cd snappy-1.1.7
+            mkdir build
+            cd build
+            ${CMAKE_BIN} -G "${CMAKE_GENERATOR}" ..
+            msbuild.exe Snappy.sln -maxCpuCount -property:Configuration=Debug -property:Platform=x64
+      - run:
+          name: "Build RocksDB"
+          command: |
+            mkdir build
+            cd build
+            ${CMAKE_BIN} -G "${CMAKE_GENERATOR}" -DCMAKE_BUILD_TYPE=Debug -DOPTDBG=1 -DPORTABLE=1 -DSNAPPY=1 -DJNI=1 ..
+            cd ..
+            msbuild.exe build/rocksdb.sln -maxCpuCount -property:Configuration=Debug -property:Platform=x64
+      - run:
+          name: "Test RocksDB"
+          shell: powershell.exe
+          command: |
+            build_tools\run_ci_db_test.ps1 -SuiteRun db_basic_test,db_test,db_test2,env_basic_test,env_test,db_merge_operand_test -Concurrency 16

.gitignore

@@ -34,6 +34,7 @@ manifest_dump
 sst_dump
 blob_dump
 block_cache_trace_analyzer
+db_with_timestamp_basic_test
 tools/block_cache_analyzer/*.pyc
 column_aware_encoding_exp
 util/build_version.cc
@@ -52,6 +53,8 @@ trace_analyzer
 trace_analyzer_test
 block_cache_trace_analyzer
 .DS_Store
+.vs
+.vscode
 
 java/out
 java/target
@@ -80,3 +83,5 @@ fbcode/
 fbcode
 buckifier/*.pyc
 buckifier/__pycache__
+
+compile_commands.json

.travis.yml

@@ -3,28 +3,39 @@ language: cpp
 os:
   - linux
   - osx
+arch:
+  - amd64
+  - arm64
+  - ppc64le
 compiler:
   - clang
   - gcc
-osx_image: xcode8.3
-jdk:
-  - openjdk7
+osx_image: xcode9.4
 cache:
   - ccache
-  - apt
 
 addons:
   apt:
     sources:
       - ubuntu-toolchain-r-test
     packages:
-      - curl
-      - g++-8
-      - libbz2-dev
       - libgflags-dev
+      - libbz2-dev
+      - liblz4-dev
       - libsnappy-dev
-      - mingw-w64
+      - liblzma-dev  # xv
+      - libzstd-dev
       - zlib1g-dev
+  homebrew:
+    update: true
+    packages:
+      - ccache
+      - gflags
+      - lz4
+      - snappy
+      - xz
+      - zstd
+
 env:
   - TEST_GROUP=platform_dependent # 16-18 minutes
   - TEST_GROUP=1 # 33-35 minutes
@@ -43,37 +54,135 @@ env:
 
 matrix:
   exclude:
-  - os: osx
-    env: TEST_GROUP=1
-  - os: osx
-    env: TEST_GROUP=2
-  - os: osx
-    env: TEST_GROUP=3
-  - os: osx
-    env: TEST_GROUP=4
   - os: osx
     env: JOB_NAME=cmake-gcc8
-  - os : osx
+  - os: osx
     env: JOB_NAME=cmake-mingw
-  - os : linux
-    compiler: clang
-  - os : osx
+  - os: osx
+    arch: ppc64le
+  - os: osx
     compiler: gcc
+  - os : linux
+    arch: arm64
+    env: JOB_NAME=cmake-mingw
+  - os: linux
+    arch: ppc64le
+    env: JOB_NAME=cmake-mingw
+  - os: linux
+    compiler: clang
+  # Exclude most osx, arm64 and ppc64le tests for pull requests, but build in branches
+  - if: type != pull_request
+    os: osx
+    env: TEST_GROUP=1
+  - if: type != pull_request
+    os : linux
+    arch: arm64
+    env: TEST_GROUP=1
+  - if: type != pull_request
+    os: linux
+    arch: ppc64le
+    env: TEST_GROUP=1
+  - if: type != pull_request
+    os: osx
+    env: TEST_GROUP=2
+  - if: type != pull_request
+    os : linux
+    arch: arm64
+    env: TEST_GROUP=2
+  - if: type != pull_request
+    os: linux
+    arch: ppc64le
+    env: TEST_GROUP=2
+  - if: type != pull_request
+    os: osx
+    env: TEST_GROUP=3
+  - if: type != pull_request
+    os : linux
+    arch: arm64
+    env: TEST_GROUP=3
+  - if: type != pull_request
+    os: linux
+    arch: ppc64le
+    env: TEST_GROUP=3
+  - if: type != pull_request
+    os: osx
+    env: TEST_GROUP=4
+  - if: type != pull_request
+    os : linux
+    arch: arm64
+    env: TEST_GROUP=4
+  - if: type != pull_request
+    os: linux
+    arch: ppc64le
+    env: TEST_GROUP=4
+  - if: type != pull_request
+    os : linux
+    arch: arm64
+    env: JOB_NAME=java-test
+  - if: type != pull_request
+    os: linux
+    arch: ppc64le
+    env: JOB_NAME=java-test
+  - if: type != pull_request
+    os : linux
+    arch: arm64
+    env: JOB_NAME=lite-build
+  - if: type != pull_request
+    os: linux
+    arch: ppc64le
+    env: JOB_NAME=lite-build
+  - if: type != pull_request
+    os : linux
+    arch: arm64
+    env: JOB_NAME=examples
+  - if: type != pull_request
+    os: linux
+    arch: ppc64le
+    env: JOB_NAME=examples
+  - if: type != pull_request
+    os : linux
+    arch: arm64
+    env: JOB_NAME=cmake-gcc8
+  - if: type != pull_request
+    os: linux
+    arch: ppc64le
+    env: JOB_NAME=cmake-gcc8
 
-# https://docs.travis-ci.com/user/caching/#ccache-cache
 install:
   - if [ "${TRAVIS_OS_NAME}" == osx ]; then
-      brew install ccache zstd lz4 snappy xz;
       PATH=$PATH:/usr/local/opt/ccache/libexec;
     fi
   - if [ "${JOB_NAME}" == cmake-gcc8 ]; then
+      sudo apt-get install -y g++-8;
       CC=gcc-8 && CXX=g++-8;
     fi
-  - if [[ "${JOB_NAME}" == cmake* ]] && [ "${TRAVIS_OS_NAME}" == linux ]; then
-      mkdir cmake-dist && curl -sfSL https://github.com/Kitware/CMake/releases/download/v3.14.5/cmake-3.14.5-Linux-x86_64.tar.gz | tar --strip-components=1 -C cmake-dist -xz && export PATH=$PWD/cmake-dist/bin:$PATH;
+  - if [ "${JOB_NAME}" == cmake-mingw ]; then
+      sudo apt-get install -y mingw-w64 ;
     fi
-  - if [[ "${JOB_NAME}" == java_test ]]; then
-      java -version && echo "JAVA_HOME=${JAVA_HOME}";
+  - if [[ "${JOB_NAME}" == cmake* ]] && [ "${TRAVIS_OS_NAME}" == linux ]; then
+      CMAKE_DIST_URL="https://rocksdb-deps.s3-us-west-2.amazonaws.com/cmake/cmake-3.14.5-Linux-$(uname -m).tar.bz2";
+      TAR_OPT="--strip-components=1 -xj";
+      if [ "aarch64" == "$(uname -m)" ]; then
+        sudo apt-get install -y libuv1 librhash0;
+        sudo apt-get upgrade -y libstdc++6;
+      fi;
+      mkdir cmake-dist && curl --silent --fail --show-error --location "${CMAKE_DIST_URL}" | tar -C cmake-dist ${TAR_OPT} && export PATH=$PWD/cmake-dist/bin:$PATH;
+    fi
+  - |
+    if [[ "${JOB_NAME}" == java_test || "${JOB_NAME}" == cmake* ]]; then
+      # Ensure JDK 8
+      if [ "${TRAVIS_OS_NAME}" == osx ]; then
+        brew tap AdoptOpenJDK/openjdk
+        brew cask install adoptopenjdk8
+        export JAVA_HOME=$(/usr/libexec/java_home)
+      else
+        sudo apt-get install -y openjdk-8-jdk
+        export PATH=/usr/lib/jvm/java-8-openjdk-$(dpkg --print-architecture)/bin:$PATH
+        export JAVA_HOME=/usr/lib/jvm/java-8-openjdk-$(dpkg --print-architecture)
+      fi
+      echo "JAVA_HOME=${JAVA_HOME}"
+      which java && java -version
+      which javac && javac -version
     fi
 
 before_script:
@@ -82,7 +191,7 @@ before_script:
   - ulimit -n 8192
 
 script:
-  - ${CXX} --version
+  - date; ${CXX} --version
   - if [ `command -v ccache` ]; then ccache -C; fi
   - case $TEST_GROUP in
     platform_dependent)
@@ -92,7 +201,7 @@ script:
       OPT=-DTRAVIS V=1 ROCKSDBTESTS_START=db_block_cache_test ROCKSDBTESTS_END=db_iter_test make -j4 check_some
       ;;
     2)
-      OPT=-DTRAVIS V=1 make -j4 tools && OPT=-DTRAVIS V=1 ROCKSDBTESTS_START=db_iter_test ROCKSDBTESTS_END=options_file_test make -j4 check_some
+      OPT="-DTRAVIS -DROCKSDB_NAMESPACE=alternative_rocksdb_ns" V=1 make -j4 tools && OPT="-DTRAVIS -DROCKSDB_NAMESPACE=alternative_rocksdb_ns" V=1 ROCKSDBTESTS_START=db_iter_test ROCKSDBTESTS_END=options_file_test make -j4 check_some
       ;;
     3)
       OPT=-DTRAVIS V=1 ROCKSDBTESTS_START=options_file_test ROCKSDBTESTS_END=write_prepared_transaction_test make -j4 check_some
@@ -113,7 +222,7 @@ script:
       ;;
     cmake-mingw)
       sudo update-alternatives --set x86_64-w64-mingw32-g++ /usr/bin/x86_64-w64-mingw32-g++-posix;
-      mkdir build && cd build && cmake -DJNI=1 .. -DCMAKE_C_COMPILER=x86_64-w64-mingw32-gcc -DCMAKE_CXX_COMPILER=x86_64-w64-mingw32-g++ -DCMAKE_SYSTEM_NAME=Windows && make -j4 rocksdb rocksdbjni
+      mkdir build && cd build && cmake -DJNI=1 -DWITH_GFLAGS=OFF .. -DCMAKE_C_COMPILER=x86_64-w64-mingw32-gcc -DCMAKE_CXX_COMPILER=x86_64-w64-mingw32-g++ -DCMAKE_SYSTEM_NAME=Windows && make -j4 rocksdb rocksdbjni
       ;;
     cmake*)
       mkdir build && cd build && cmake -DJNI=1 .. -DCMAKE_BUILD_TYPE=Release && make -j4 rocksdb rocksdbjni

CMakeLists.txt

@@ -14,7 +14,7 @@
 #        cd build
 # 3. Run cmake to generate project files for Windows, add more options to enable required third-party libraries.
 #    See thirdparty.inc for more information.
-#        sample command: cmake -G "Visual Studio 15 Win64" -DWITH_GFLAGS=1 -DWITH_SNAPPY=1 -DWITH_JEMALLOC=1 -DWITH_JNI=1 ..
+#        sample command: cmake -G "Visual Studio 15 Win64" -DCMAKE_BUILD_TYPE=Release -DWITH_GFLAGS=1 -DWITH_SNAPPY=1 -DWITH_JEMALLOC=1 -DWITH_JNI=1 ..
 # 4. Then build the project in debug mode (you may want to add /m[:<N>] flag to run msbuild in <N> parallel threads
 #                                          or simply /m to use all avail cores)
 #        msbuild rocksdb.sln
@@ -45,6 +45,16 @@ if(POLICY CMP0042)
   cmake_policy(SET CMP0042 NEW)
 endif()
 
+if(NOT CMAKE_BUILD_TYPE)
+  if(EXISTS "${CMAKE_SOURCE_DIR}/.git")
+    set(default_build_type "Debug")
+  else()
+    set(default_build_type "RelWithDebInfo")
+  endif()
+  set(CMAKE_BUILD_TYPE "${default_build_type}" CACHE STRING
+    "Default BUILD_TYPE is ${default_build_type}" FORCE)
+endif()
+
 find_program(CCACHE_FOUND ccache)
 if(CCACHE_FOUND)
   set_property(GLOBAL PROPERTY RULE_LAUNCH_COMPILE ccache)
@@ -62,15 +72,22 @@ if (WITH_WINDOWS_UTF8_FILENAMES)
 endif()
 # third-party/folly is only validated to work on Linux and Windows for now.
 # So only turn it on there by default.
-if(CMAKE_SYSTEM_NAME MATCHES "Linux" OR CMAKE_SYSTEM_NAME MATCHES "Windows")
-  option(WITH_FOLLY_DISTRIBUTED_MUTEX "build with folly::DistributedMutex" ON)
+if(CMAKE_SYSTEM_NAME MATCHES "Linux|Windows")
+  if(MSVC AND MSVC_VERSION LESS 1910)
+    # Folly does not compile with MSVC older than VS2017
+    option(WITH_FOLLY_DISTRIBUTED_MUTEX "build with folly::DistributedMutex" OFF)
+  else()
+    option(WITH_FOLLY_DISTRIBUTED_MUTEX "build with folly::DistributedMutex" ON)
+  endif()
 else()
   option(WITH_FOLLY_DISTRIBUTED_MUTEX "build with folly::DistributedMutex" OFF)
 endif()
+
+include(CMakeDependentOption)
+CMAKE_DEPENDENT_OPTION(WITH_GFLAGS "build with GFlags" ON
+  "NOT MSVC;NOT MINGW" OFF)
+
 if(MSVC)
-  # Defaults currently different for GFLAGS.
-  #  We will address find_package work a little later
-  option(WITH_GFLAGS "build with GFlags" OFF)
   option(WITH_XPRESS "build with windows built in compression" OFF)
   include(${CMAKE_CURRENT_SOURCE_DIR}/thirdparty.inc)
 else()
@@ -87,14 +104,11 @@ else()
   endif()
 
   # No config file for this
-  option(WITH_GFLAGS "build with GFlags" ON)
   if(WITH_GFLAGS)
-    find_package(gflags)
-    if(gflags_FOUND)
-      add_definitions(-DGFLAGS=1)
-      include_directories(${gflags_INCLUDE_DIR})
-      list(APPEND THIRDPARTY_LIBS ${gflags_LIBRARIES})
-    endif()
+    find_package(gflags REQUIRED)
+    add_definitions(-DGFLAGS=1)
+    include_directories(${gflags_INCLUDE_DIR})
+    list(APPEND THIRDPARTY_LIBS gflags::gflags)
   endif()
 
   if(WITH_SNAPPY)
@@ -174,7 +188,7 @@ else()
   set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -W -Wextra -Wall")
   set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wsign-compare -Wshadow -Wno-unused-parameter -Wno-unused-variable -Woverloaded-virtual -Wnon-virtual-dtor -Wno-missing-field-initializers -Wno-strict-aliasing")
   if(MINGW)
-    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wno-format")
+    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wno-format -fno-asynchronous-unwind-tables")
     add_definitions(-D_POSIX_C_SOURCE=1)
   endif()
   set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")
@@ -189,16 +203,25 @@ else()
 endif()
 
 include(CheckCCompilerFlag)
-if(CMAKE_SYSTEM_PROCESSOR MATCHES "ppc64le")
+if(CMAKE_SYSTEM_PROCESSOR MATCHES "^(powerpc|ppc)64")
+  CHECK_C_COMPILER_FLAG("-mcpu=power9" HAS_POWER9)
+  if(HAS_POWER9)
+    set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -mcpu=power9 -mtune=power9")
+    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mcpu=power9 -mtune=power9")
+  else()
+    CHECK_C_COMPILER_FLAG("-mcpu=power8" HAS_POWER8)
+    if(HAS_POWER8)
+      set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -mcpu=power8 -mtune=power8")
+      set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mcpu=power8 -mtune=power8")
+    endif(HAS_POWER8)
+  endif(HAS_POWER9)
   CHECK_C_COMPILER_FLAG("-maltivec" HAS_ALTIVEC)
   if(HAS_ALTIVEC)
     message(STATUS " HAS_ALTIVEC yes")
     set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -maltivec")
     set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -maltivec")
-    set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -mcpu=power8")
-    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mcpu=power8")
   endif(HAS_ALTIVEC)
-endif(CMAKE_SYSTEM_PROCESSOR MATCHES "ppc64le")
+endif(CMAKE_SYSTEM_PROCESSOR MATCHES "^(powerpc|ppc)64")
 
 if(CMAKE_SYSTEM_PROCESSOR MATCHES "aarch64|AARCH64")
         CHECK_C_COMPILER_FLAG("-march=armv8-a+crc+crypto" HAS_ARMV8_CRC)
@@ -221,7 +244,7 @@ else()
   if(MSVC)
     set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /arch:AVX2")
   else()
-    if(NOT HAVE_POWER8 AND NOT HAS_ARMV8_CRC)
+    if(NOT CMAKE_SYSTEM_PROCESSOR MATCHES "^(powerpc|ppc)64" AND NOT HAS_ARMV8_CRC)
       set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=native")
     endif()
   endif()
@@ -467,6 +490,11 @@ if(HAVE_SCHED_GETCPU)
   add_definitions(-DROCKSDB_SCHED_GETCPU_PRESENT)
 endif()
 
+check_cxx_symbol_exists(getauxval auvx.h HAVE_AUXV_GETAUXVAL)
+if(HAVE_AUXV_GETAUXVAL)
+  add_definitions(-DROCKSDB_AUXV_GETAUXVAL_PRESENT)
+endif()
+
 include_directories(${PROJECT_SOURCE_DIR})
 include_directories(${PROJECT_SOURCE_DIR}/include)
 include_directories(SYSTEM ${PROJECT_SOURCE_DIR}/third-party/gtest-1.8.1/fused-src)
@@ -482,6 +510,8 @@ set(SOURCES
         cache/lru_cache.cc
         cache/sharded_cache.cc
         db/arena_wrapped_db_iter.cc
+        db/blob/blob_file_addition.cc
+        db/blob/blob_file_garbage.cc
         db/builder.cc
         db/c.cc
         db/column_family.cc
@@ -545,6 +575,7 @@ set(SOURCES
         env/env_chroot.cc
         env/env_encryption.cc
         env/env_hdfs.cc
+        env/file_system.cc
         env/mock_env.cc
         file/delete_scheduler.cc
         file/file_prefetch_buffer.cc
@@ -589,12 +620,15 @@ set(SOURCES
         options/options_sanity_check.cc
         port/stack_trace.cc
         table/adaptive/adaptive_table_factory.cc
+        table/block_based/binary_search_index_reader.cc
         table/block_based/block.cc
         table/block_based/block_based_filter_block.cc
         table/block_based/block_based_table_builder.cc
         table/block_based/block_based_table_factory.cc
+        table/block_based/block_based_table_iterator.cc
         table/block_based/block_based_table_reader.cc
         table/block_based/block_builder.cc
+        table/block_based/block_prefetcher.cc
         table/block_based/block_prefix_index.cc
         table/block_based/data_block_hash_index.cc
         table/block_based/data_block_footer.cc
@@ -602,9 +636,14 @@ set(SOURCES
         table/block_based/filter_policy.cc
         table/block_based/flush_block_policy.cc
         table/block_based/full_filter_block.cc
+        table/block_based/hash_index_reader.cc
         table/block_based/index_builder.cc
+        table/block_based/index_reader_common.cc
         table/block_based/parsed_full_filter_block.cc
         table/block_based/partitioned_filter_block.cc
+        table/block_based/partitioned_index_iterator.cc
+        table/block_based/partitioned_index_reader.cc
+        table/block_based/reader_common.cc
         table/block_based/uncompression_dict_reader.cc
         table/block_fetcher.cc
         table/cuckoo/cuckoo_table_builder.cc
@@ -631,7 +670,6 @@ set(SOURCES
         test_util/testutil.cc
         test_util/transaction_test_util.cc
         tools/block_cache_analyzer/block_cache_trace_analyzer.cc
-        tools/db_bench_tool.cc
         tools/dump/db_dump_tool.cc
         tools/ldb_cmd.cc
         tools/ldb_tool.cc
@@ -651,6 +689,7 @@ set(SOURCES
         util/random.cc
         util/rate_limiter.cc
         util/slice.cc
+        util/file_checksum_helper.cc
         util/status.cc
         util/string_util.cc
         util/thread_local.cc
@@ -719,11 +758,11 @@ if(HAVE_SSE42 AND NOT MSVC)
     PROPERTIES COMPILE_FLAGS "-msse4.2 -mpclmul")
 endif()
 
-if(HAVE_POWER8)
+if(CMAKE_SYSTEM_PROCESSOR MATCHES "^(powerpc|ppc)64")
   list(APPEND SOURCES
     util/crc32c_ppc.c
     util/crc32c_ppc_asm.S)
-endif(HAVE_POWER8)
+endif(CMAKE_SYSTEM_PROCESSOR MATCHES "^(powerpc|ppc)64")
 
 if(HAS_ARMV8_CRC)
   list(APPEND SOURCES
@@ -753,6 +792,7 @@ else()
   list(APPEND SOURCES
     port/port_posix.cc
     env/env_posix.cc
+    env/fs_posix.cc
     env/io_posix.cc)
 endif()
 
@@ -767,7 +807,8 @@ endif()
 
 set(ROCKSDB_STATIC_LIB rocksdb${ARTIFACT_SUFFIX})
 set(ROCKSDB_SHARED_LIB rocksdb-shared${ARTIFACT_SUFFIX})
-set(ROCKSDB_IMPORT_LIB ${ROCKSDB_SHARED_LIB})
+
+option(ROCKSDB_BUILD_SHARED "Build shared versions of the RocksDB libraries" ON)
 
 option(WITH_LIBRADOS "Build with librados" OFF)
 if(WITH_LIBRADOS)
@@ -778,40 +819,44 @@ endif()
 
 if(WIN32)
   set(SYSTEM_LIBS ${SYSTEM_LIBS} shlwapi.lib rpcrt4.lib)
-  set(LIBS ${ROCKSDB_STATIC_LIB} ${THIRDPARTY_LIBS} ${SYSTEM_LIBS})
 else()
   set(SYSTEM_LIBS ${CMAKE_THREAD_LIBS_INIT})
-  set(LIBS ${ROCKSDB_SHARED_LIB} ${THIRDPARTY_LIBS} ${SYSTEM_LIBS})
-
-  add_library(${ROCKSDB_SHARED_LIB} SHARED ${SOURCES})
-  target_link_libraries(${ROCKSDB_SHARED_LIB}
-    ${THIRDPARTY_LIBS} ${SYSTEM_LIBS})
-  set_target_properties(${ROCKSDB_SHARED_LIB} PROPERTIES
-                        LINKER_LANGUAGE CXX
-                        VERSION ${rocksdb_VERSION}
-                        SOVERSION ${rocksdb_VERSION_MAJOR}
-                        CXX_STANDARD 11
-                        OUTPUT_NAME "rocksdb")
 endif()
 
 add_library(${ROCKSDB_STATIC_LIB} STATIC ${SOURCES})
 target_link_libraries(${ROCKSDB_STATIC_LIB}
   ${THIRDPARTY_LIBS} ${SYSTEM_LIBS})
 
-if(WIN32)
-  add_library(${ROCKSDB_IMPORT_LIB} SHARED ${SOURCES})
-  target_link_libraries(${ROCKSDB_IMPORT_LIB}
+if(ROCKSDB_BUILD_SHARED)
+  add_library(${ROCKSDB_SHARED_LIB} SHARED ${SOURCES})
+  target_link_libraries(${ROCKSDB_SHARED_LIB}
     ${THIRDPARTY_LIBS} ${SYSTEM_LIBS})
-  set_target_properties(${ROCKSDB_IMPORT_LIB} PROPERTIES
-    COMPILE_DEFINITIONS "ROCKSDB_DLL;ROCKSDB_LIBRARY_EXPORTS")
-  if(MSVC)
-    set_target_properties(${ROCKSDB_STATIC_LIB} PROPERTIES
-      COMPILE_FLAGS "/Fd${CMAKE_CFG_INTDIR}/${ROCKSDB_STATIC_LIB}.pdb")
-    set_target_properties(${ROCKSDB_IMPORT_LIB} PROPERTIES
-      COMPILE_FLAGS "/Fd${CMAKE_CFG_INTDIR}/${ROCKSDB_IMPORT_LIB}.pdb")
+
+  if(WIN32)
+    set_target_properties(${ROCKSDB_SHARED_LIB} PROPERTIES
+      COMPILE_DEFINITIONS "ROCKSDB_DLL;ROCKSDB_LIBRARY_EXPORTS")
+    if(MSVC)
+      set_target_properties(${ROCKSDB_STATIC_LIB} PROPERTIES
+        COMPILE_FLAGS "/Fd${CMAKE_CFG_INTDIR}/${ROCKSDB_STATIC_LIB}.pdb")
+      set_target_properties(${ROCKSDB_SHARED_LIB} PROPERTIES
+        COMPILE_FLAGS "/Fd${CMAKE_CFG_INTDIR}/${ROCKSDB_SHARED_LIB}.pdb")
+    endif()
+  else()
+    set_target_properties(${ROCKSDB_SHARED_LIB} PROPERTIES
+                          LINKER_LANGUAGE CXX
+                          VERSION ${rocksdb_VERSION}
+                          SOVERSION ${rocksdb_VERSION_MAJOR}
+                          CXX_STANDARD 11
+                          OUTPUT_NAME "rocksdb")
   endif()
 endif()
 
+if(ROCKSDB_BUILD_SHARED AND NOT WIN32)
+  set(ROCKSDB_LIB ${ROCKSDB_SHARED_LIB})
+else()
+  set(ROCKSDB_LIB ${ROCKSDB_STATIC_LIB})
+endif()
+
 option(WITH_JNI "build with JNI" OFF)
 if(WITH_JNI OR JNI)
   message(STATUS "JNI library is enabled")
@@ -858,15 +903,17 @@ if(NOT WIN32 OR ROCKSDB_INSTALL_ON_WINDOWS)
     INCLUDES DESTINATION "${CMAKE_INSTALL_INCLUDEDIR}"
   )
 
-  install(
-    TARGETS ${ROCKSDB_SHARED_LIB}
-    EXPORT RocksDBTargets
-    COMPONENT runtime
-    ARCHIVE DESTINATION "${CMAKE_INSTALL_LIBDIR}"
-    RUNTIME DESTINATION "${CMAKE_INSTALL_BINDIR}"
-    LIBRARY DESTINATION "${CMAKE_INSTALL_LIBDIR}"
-    INCLUDES DESTINATION "${CMAKE_INSTALL_INCLUDEDIR}"
-  )
+  if(ROCKSDB_BUILD_SHARED)
+    install(
+      TARGETS ${ROCKSDB_SHARED_LIB}
+      EXPORT RocksDBTargets
+      COMPONENT runtime
+      ARCHIVE DESTINATION "${CMAKE_INSTALL_LIBDIR}"
+      RUNTIME DESTINATION "${CMAKE_INSTALL_BINDIR}"
+      LIBRARY DESTINATION "${CMAKE_INSTALL_LIBDIR}"
+      INCLUDES DESTINATION "${CMAKE_INSTALL_INCLUDEDIR}"
+    )
+  endif()
 
   install(
     EXPORT RocksDBTargets
@@ -884,12 +931,21 @@ if(NOT WIN32 OR ROCKSDB_INSTALL_ON_WINDOWS)
   )
 endif()
 
-option(WITH_TESTS "build with tests" ON)
+# Tests are excluded from Release builds
+CMAKE_DEPENDENT_OPTION(WITH_TESTS "build with tests" ON
+  "CMAKE_BUILD_TYPE STREQUAL Debug" OFF)
 if(WITH_TESTS)
   add_subdirectory(third-party/gtest-1.8.1/fused-src/gtest)
+  add_library(testharness STATIC
+  test_util/testharness.cc)
+  target_link_libraries(testharness gtest)
+
   set(TESTS
         cache/cache_test.cc
         cache/lru_cache_test.cc
+        db/blob/blob_file_addition_test.cc
+        db/blob/blob_file_garbage_test.cc
+        db/blob/db_blob_index_test.cc
         db/column_family_test.cc
         db/compact_files_test.cc
         db/compaction/compaction_job_stats_test.cc
@@ -900,7 +956,7 @@ if(WITH_TESTS)
         db/corruption_test.cc
         db/cuckoo_table_db_test.cc
         db/db_basic_test.cc
-        db/db_blob_index_test.cc
+        db/db_with_timestamp_basic_test.cc
         db/db_block_cache_test.cc
         db/db_bloom_filter_test.cc
         db/db_compaction_filter_test.cc
@@ -926,12 +982,13 @@ if(WITH_TESTS)
         db/db_tailing_iter_test.cc
         db/db_test.cc
         db/db_test2.cc
+        db/db_logical_block_size_cache_test.cc
         db/db_universal_compaction_test.cc
         db/db_wal_test.cc
         db/db_write_test.cc
         db/dbformat_test.cc
         db/deletefile_test.cc
-        db/error_handler_test.cc
+        db/error_handler_fs_test.cc
         db/obsolete_files_test.cc
         db/external_sst_file_basic_test.cc
         db/external_sst_file_test.cc
@@ -962,6 +1019,7 @@ if(WITH_TESTS)
         db/write_controller_test.cc
         env/env_basic_test.cc
         env/env_test.cc
+        env/io_posix_test.cc
         env/mock_env_test.cc
         file/delete_scheduler_test.cc
         logging/auto_roll_logger_test.cc
@@ -997,13 +1055,16 @@ if(WITH_TESTS)
         util/bloom_test.cc
         util/coding_test.cc
         util/crc32c_test.cc
+        util/defer_test.cc
         util/dynamic_bloom_test.cc
         util/file_reader_writer_test.cc
         util/filelock_test.cc
         util/hash_test.cc
         util/heap_test.cc
+        util/random_test.cc
         util/rate_limiter_test.cc
         util/repeatable_thread_test.cc
+        util/slice_test.cc
         util/slice_transform_test.cc
         util/timer_queue_test.cc
         util/thread_list_test.cc
@@ -1040,37 +1101,19 @@ if(WITH_TESTS)
     list(APPEND TESTS third-party/folly/folly/synchronization/test/DistributedMutexTest.cpp)
   endif()
 
-  set(BENCHMARKS
-    cache/cache_bench.cc
-    memtable/memtablerep_bench.cc
-    db/range_del_aggregator_bench.cc
-    tools/db_bench.cc
-    table/table_reader_bench.cc
-    util/filter_bench.cc
-    utilities/persistent_cache/hash_table_bench.cc)
-  add_library(testharness OBJECT test_util/testharness.cc)
-  foreach(sourcefile ${BENCHMARKS})
-    get_filename_component(exename ${sourcefile} NAME_WE)
-    add_executable(${exename}${ARTIFACT_SUFFIX} ${sourcefile}
-      $<TARGET_OBJECTS:testharness>)
-    target_link_libraries(${exename}${ARTIFACT_SUFFIX} gtest ${LIBS})
-  endforeach(sourcefile ${BENCHMARKS})
-
-  # For test util library that is build only in DEBUG mode
-  # and linked to tests. Add test only code that is not #ifdefed for Release here.
   set(TESTUTIL_SOURCE
       db/db_test_util.cc
       monitoring/thread_status_updater_debug.cc
       table/mock_table.cc
       test_util/fault_injection_test_env.cc
+      test_util/fault_injection_test_fs.cc
       utilities/cassandra/test_utils.cc
   )
-  # test utilities are only build in debug
   enable_testing()
   add_custom_target(check COMMAND ${CMAKE_CTEST_COMMAND})
   set(TESTUTILLIB testutillib${ARTIFACT_SUFFIX})
   add_library(${TESTUTILLIB} STATIC ${TESTUTIL_SOURCE})
-  target_link_libraries(${TESTUTILLIB} ${LIBS})
+  target_link_libraries(${TESTUTILLIB} ${ROCKSDB_LIB})
   if(MSVC)
     set_target_properties(${TESTUTILLIB} PROPERTIES COMPILE_FLAGS "/Fd${CMAKE_CFG_INTDIR}/testutillib${ARTIFACT_SUFFIX}.pdb")
   endif()
@@ -1080,46 +1123,92 @@ if(WITH_TESTS)
         EXCLUDE_FROM_DEFAULT_BUILD_RELWITHDEBINFO 1
         )
 
-  # Tests are excluded from Release builds
-  set(TEST_EXES ${TESTS})
-
-  foreach(sourcefile ${TEST_EXES})
+  foreach(sourcefile ${TESTS})
       get_filename_component(exename ${sourcefile} NAME_WE)
-      add_executable(${CMAKE_PROJECT_NAME}_${exename}${ARTIFACT_SUFFIX} ${sourcefile}
-        $<TARGET_OBJECTS:testharness>)
+      add_executable(${CMAKE_PROJECT_NAME}_${exename}${ARTIFACT_SUFFIX} ${sourcefile})
       set_target_properties(${CMAKE_PROJECT_NAME}_${exename}${ARTIFACT_SUFFIX}
         PROPERTIES EXCLUDE_FROM_DEFAULT_BUILD_RELEASE 1
         EXCLUDE_FROM_DEFAULT_BUILD_MINRELEASE 1
         EXCLUDE_FROM_DEFAULT_BUILD_RELWITHDEBINFO 1
         OUTPUT_NAME ${exename}${ARTIFACT_SUFFIX}
         )
-      target_link_libraries(${CMAKE_PROJECT_NAME}_${exename}${ARTIFACT_SUFFIX} testutillib${ARTIFACT_SUFFIX} gtest ${LIBS})
+      target_link_libraries(${CMAKE_PROJECT_NAME}_${exename}${ARTIFACT_SUFFIX} testutillib${ARTIFACT_SUFFIX} testharness gtest ${ROCKSDB_LIB})
       if(NOT "${exename}" MATCHES "db_sanity_test")
         add_test(NAME ${exename} COMMAND ${exename}${ARTIFACT_SUFFIX})
         add_dependencies(check ${CMAKE_PROJECT_NAME}_${exename}${ARTIFACT_SUFFIX})
       endif()
-  endforeach(sourcefile ${TEST_EXES})
+  endforeach(sourcefile ${TESTS})
 
-  # C executables must link to a shared object
-  set(C_TESTS db/c_test.c)
-  set(C_TEST_EXES ${C_TESTS})
+  if(WIN32)
+    # C executables must link to a shared object
+    if(ROCKSDB_BUILD_SHARED)
+      set(ROCKSDB_LIB_FOR_C ${ROCKSDB_SHARED_LIB})
+    else()
+      set(ROCKSDB_LIB_FOR_C OFF)
+    endif()
+  else()
+    set(ROCKSDB_LIB_FOR_C ${ROCKSDB_LIB})
+  endif()
 
-  foreach(sourcefile ${C_TEST_EXES})
-      string(REPLACE ".c" "" exename ${sourcefile})
-      string(REGEX REPLACE "^((.+)/)+" "" exename ${exename})
-      add_executable(${exename}${ARTIFACT_SUFFIX} ${sourcefile})
-      set_target_properties(${exename}${ARTIFACT_SUFFIX}
-        PROPERTIES EXCLUDE_FROM_DEFAULT_BUILD_RELEASE 1
-        EXCLUDE_FROM_DEFAULT_BUILD_MINRELEASE 1
-        EXCLUDE_FROM_DEFAULT_BUILD_RELWITHDEBINFO 1
-        )
-      target_link_libraries(${exename}${ARTIFACT_SUFFIX} ${ROCKSDB_IMPORT_LIB} testutillib${ARTIFACT_SUFFIX})
-      add_test(NAME ${exename} COMMAND ${exename}${ARTIFACT_SUFFIX})
-      add_dependencies(check ${exename}${ARTIFACT_SUFFIX})
-  endforeach(sourcefile ${C_TEST_EXES})
+  if(ROCKSDB_LIB_FOR_C)
+    set(C_TESTS db/c_test.c)
+    # C executables must link to a shared object
+    add_executable(c_test db/c_test.c)
+    target_link_libraries(c_test ${ROCKSDB_SHARED_LIB} testharness)
+    add_test(NAME c_test COMMAND c_test${ARTIFACT_SUFFIX})
+    add_dependencies(check c_test)
+  endif()
 endif()
 
+option(WITH_BENCHMARK_TOOLS "build with benchmarks" ON)
+if(WITH_BENCHMARK_TOOLS)
+  add_executable(db_bench
+    tools/db_bench.cc
+    tools/db_bench_tool.cc)
+  target_link_libraries(db_bench
+    ${ROCKSDB_LIB})
+
+  add_executable(cache_bench
+    cache/cache_bench.cc)
+  target_link_libraries(cache_bench
+    ${ROCKSDB_LIB})
+
+  add_executable(memtablerep_bench
+    memtable/memtablerep_bench.cc)
+  target_link_libraries(memtablerep_bench
+    ${ROCKSDB_LIB})
+
+  add_executable(range_del_aggregator_bench
+    db/range_del_aggregator_bench.cc)
+  target_link_libraries(range_del_aggregator_bench
+    ${ROCKSDB_LIB})
+
+  add_executable(table_reader_bench
+    table/table_reader_bench.cc)
+  target_link_libraries(table_reader_bench
+    ${ROCKSDB_LIB} testharness)
+
+  add_executable(filter_bench
+    util/filter_bench.cc)
+  target_link_libraries(filter_bench
+    ${ROCKSDB_LIB})
+
+  add_executable(hash_table_bench
+    utilities/persistent_cache/hash_table_bench.cc)
+  target_link_libraries(hash_table_bench
+    ${ROCKSDB_LIB})
+endif()
+
+option(WITH_CORE_TOOLS "build with ldb and sst_dump" ON)
 option(WITH_TOOLS "build with tools" ON)
-if(WITH_TOOLS)
+if(WITH_CORE_TOOLS OR WITH_TOOLS)
   add_subdirectory(tools)
+  add_custom_target(core_tools
+    DEPENDS ${core_tool_deps})
+endif()
+
+if(WITH_TOOLS)
+  add_subdirectory(db_stress_tool)
+  add_custom_target(tools
+    DEPENDS ${tool_deps})
 endif()

HISTORY.md

@@ -1,10 +1,91 @@
 # Rocksdb Change Log
+## Unreleased
+### Public API Change
+* Fix spelling so that API now has correctly spelled transaction state name `COMMITTED`, while the old misspelled `COMMITED` is still available as an alias.
+
+### Bug Fixes
+* Fix a bug where range tombstone blocks in ingested files were cached incorrectly during ingestion. If range tombstones were read from those incorrectly cached blocks, the keys they covered would be exposed.
+* Fix a data race that might cause crash when calling DB::GetCreationTimeOfOldestFile() by a small chance. The bug was introduced in 6.6 Release.
+
+### Performance Improvements
+* In CompactRange, for levels starting from 0, if the level does not have any file with any key falling in the specified range, the level is skipped. So instead of always compacting from level 0, the compaction starts from the first level with keys in the specified range until the last such level.
+
+### New Features
+* Basic support for user timestamp in iterator. Seek/SeekToFirst/Next and lower/upper bounds are supported. Reverse iteration is not supported. Merge is not considered.
+* When file lock failure when the lock is held by the current process, return acquiring time and thread ID in the error message.
+
+## 6.8.0 (02/24/2020)
+### Java API Changes
+* Major breaking changes to Java comparators, toward standardizing on ByteBuffer for performant, locale-neutral operations on keys (#6252).
+* Added overloads of common API methods using direct ByteBuffers for keys and values (#2283).
+
+### Bug Fixes
+* Fix incorrect results while block-based table uses kHashSearch, together with Prev()/SeekForPrev().
+* Fix a bug that prevents opening a DB after two consecutive crash with TransactionDB, where the first crash recovers from a corrupted WAL with kPointInTimeRecovery but the second cannot.
+* Fixed issue #6316 that can cause a corruption of the MANIFEST file in the middle when writing to it fails due to no disk space.
+* Add DBOptions::skip_checking_sst_file_sizes_on_db_open. It disables potentially expensive checking of all sst file sizes in DB::Open().
+* BlobDB now ignores trivially moved files when updating the mapping between blob files and SSTs. This should mitigate issue #6338 where out of order flush/compaction notifications could trigger an assertion with the earlier code.
+* Batched MultiGet() ignores IO errors while reading data blocks, causing it to potentially continue looking for a key and returning stale results.
+* `WriteBatchWithIndex::DeleteRange` returns `Status::NotSupported`. Previously it returned success even though reads on the batch did not account for range tombstones. The corresponding language bindings now cannot be used. In C, that includes `rocksdb_writebatch_wi_delete_range`, `rocksdb_writebatch_wi_delete_range_cf`, `rocksdb_writebatch_wi_delete_rangev`, and `rocksdb_writebatch_wi_delete_rangev_cf`. In Java, that includes `WriteBatchWithIndex::deleteRange`.
+* Assign new MANIFEST file number when caller tries to create a new MANIFEST by calling LogAndApply(..., new_descriptor_log=true). This bug can cause MANIFEST being overwritten during recovery if options.write_dbid_to_manifest = true and there are WAL file(s).
+
+### Performance Improvements
+* Perfom readahead when reading from option files. Inside DB, options.log_readahead_size will be used as the readahead size. In other cases, a default 512KB is used.
+
+### Public API Change
+* The BlobDB garbage collector now emits the statistics `BLOB_DB_GC_NUM_FILES` (number of blob files obsoleted during GC), `BLOB_DB_GC_NUM_NEW_FILES` (number of new blob files generated during GC), `BLOB_DB_GC_FAILURES` (number of failed GC passes), `BLOB_DB_GC_NUM_KEYS_RELOCATED` (number of blobs relocated during GC), and `BLOB_DB_GC_BYTES_RELOCATED` (total size of blobs relocated during GC). On the other hand, the following statistics, which are not relevant for the new GC implementation, are now deprecated: `BLOB_DB_GC_NUM_KEYS_OVERWRITTEN`, `BLOB_DB_GC_NUM_KEYS_EXPIRED`, `BLOB_DB_GC_BYTES_OVERWRITTEN`, `BLOB_DB_GC_BYTES_EXPIRED`, and `BLOB_DB_GC_MICROS`.
+* Disable recycle_log_file_num when an inconsistent recovery modes are requested: kPointInTimeRecovery and kAbsoluteConsistency
+
+### New Features
+* Added the checksum for each SST file generated by Flush or Compaction. Added sst_file_checksum_func to Options such that user can plugin their own SST file checksum function via override the FileChecksumFunc class. If user does not set the sst_file_checksum_func, SST file checksum calculation will not be enabled. The checksum information inlcuding uint32_t checksum value and a checksum function name (string). The checksum information is stored in FileMetadata in version store and also logged to MANIFEST. A new tool is added to LDB such that user can dump out a list of file checksum information from MANIFEST (stored in an unordered_map).
+* `db_bench` now supports `value_size_distribution_type`, `value_size_min`, `value_size_max` options for generating random variable sized value. Added `blob_db_compression_type` option for BlobDB to enable blob compression.
+* Replace RocksDB namespace "rocksdb" with flag "ROCKSDB_NAMESPACE" which if is not defined, defined as "rocksdb" in header file rocksdb_namespace.h.
+
+## 6.7.0 (01/21/2020)
+### Public API Change
+* Added a rocksdb::FileSystem class in include/rocksdb/file_system.h to encapsulate file creation/read/write operations, and an option DBOptions::file_system to allow a user to pass in an instance of rocksdb::FileSystem. If its a non-null value, this will take precendence over DBOptions::env for file operations. A new API rocksdb::FileSystem::Default() returns a platform default object. The DBOptions::env option and Env::Default() API will continue to be used for threading and other OS related functions, and where DBOptions::file_system is not specified, for file operations. For storage developers who are accustomed to rocksdb::Env, the interface in rocksdb::FileSystem is new and will probably undergo some changes as more storage systems are ported to it from rocksdb::Env. As of now, no env other than Posix has been ported to the new interface.
+* A new rocksdb::NewSstFileManager() API that allows the caller to pass in separate Env and FileSystem objects.
+* Changed Java API for RocksDB.keyMayExist functions to use Holder<byte[]> instead of StringBuilder, so that retrieved values need not decode to Strings.
+* A new `OptimisticTransactionDBOptions` Option that allows users to configure occ validation policy. The default policy changes from kValidateSerial to kValidateParallel to reduce mutex contention.
+
+### Bug Fixes
+* Fix a bug that can cause unnecessary bg thread to be scheduled(#6104).
+* Fix crash caused by concurrent CF iterations and drops(#6147).
+* Fix a race condition for cfd->log_number_ between manifest switch and memtable switch (PR 6249) when number of column families is greater than 1.
+* Fix a bug on fractional cascading index when multiple files at the same level contain the same smallest user key, and those user keys are for merge operands. In this case, Get() the exact key may miss some merge operands.
+* Delcare kHashSearch index type feature-incompatible with index_block_restart_interval larger than 1.
+* Fixed an issue where the thread pools were not resized upon setting `max_background_jobs` dynamically through the `SetDBOptions` interface.
+* Fix a bug that can cause write threads to hang when a slowdown/stall happens and there is a mix of writers with WriteOptions::no_slowdown set/unset.
+* Fixed an issue where an incorrect "number of input records" value was used to compute the "records dropped" statistics for compactions.
+* Fix a regression bug that causes segfault when hash is used, max_open_files != -1 and total order seek is used and switched back.
+
+### New Features
+* It is now possible to enable periodic compactions for the base DB when using BlobDB.
+* BlobDB now garbage collects non-TTL blobs when `enable_garbage_collection` is set to `true` in `BlobDBOptions`. Garbage collection is performed during compaction: any valid blobs located in the oldest N files (where N is the number of non-TTL blob files multiplied by the value of `BlobDBOptions::garbage_collection_cutoff`) encountered during compaction get relocated to new blob files, and old blob files are dropped once they are no longer needed. Note: we recommend enabling periodic compactions for the base DB when using this feature to deal with the case when some old blob files are kept alive by SSTs that otherwise do not get picked for compaction.
+* `db_bench` now supports the `garbage_collection_cutoff` option for BlobDB.
+* Introduce ReadOptions.auto_prefix_mode. When set to true, iterator will return the same result as total order seek, but may choose to use prefix seek internally based on seek key and iterator upper bound.
+* MultiGet() can use IO Uring to parallelize read from the same SST file. This featuer is by default disabled. It can be enabled with environment variable ROCKSDB_USE_IO_URING.
+
+## 6.6.2 (01/13/2020)
+### Bug Fixes
+* Fixed a bug where non-L0 compaction input files were not considered to compute the `creation_time` of new compaction outputs.
+
+## 6.6.1 (01/02/2020)
+### Bug Fixes
+* Fix a bug in WriteBatchWithIndex::MultiGetFromBatchAndDB, which is called by Transaction::MultiGet, that causes due to stale pointer access when the number of keys is > 32
+* Fixed two performance issues related to memtable history trimming. First, a new SuperVersion is now created only if some memtables were actually trimmed. Second, trimming is only scheduled if there is at least one flushed memtable that is kept in memory for the purposes of transaction conflict checking.
+* BlobDB no longer updates the SST to blob file mapping upon failed compactions.
+* Fix a bug in which a snapshot read through an iterator could be affected by a DeleteRange after the snapshot (#6062).
+* Fixed a bug where BlobDB was comparing the `ColumnFamilyHandle` pointers themselves instead of only the column family IDs when checking whether an API call uses the default column family or not.
+* Delete superversions in BackgroundCallPurge.
+* Fix use-after-free and double-deleting files in BackgroundCallPurge().
+
 ## 6.6.0 (11/25/2019)
 ### Bug Fixes
-* Fix data corruption casued by output of intra-L0 compaction on ingested file not being placed in correct order in L0.
+* Fix data corruption caused by output of intra-L0 compaction on ingested file not being placed in correct order in L0.
 * Fix a data race between Version::GetColumnFamilyMetaData() and Compaction::MarkFilesBeingCompacted() for access to being_compacted (#6056). The current fix acquires the db mutex during Version::GetColumnFamilyMetaData(), which may cause regression.
 * Fix a bug in DBIter that is_blob_ state isn't updated when iterating backward using seek.
-* Fix a bug when format_version=3, partitioned fitlers, and prefix search are used in conjunction. The bug could result into Seek::(prefix) returning NotFound for an existing prefix.
+* Fix a bug when format_version=3, partitioned filters, and prefix search are used in conjunction. The bug could result into Seek::(prefix) returning NotFound for an existing prefix.
 * Revert the feature "Merging iterator to avoid child iterator reseek for some cases (#5286)" since it might cause strong results when reseek happens with a different iterator upper bound.
 * Fix a bug causing a crash during ingest external file when background compaction cause severe error (file not found).
 * Fix a bug when partitioned filters and prefix search are used in conjunction, ::SeekForPrev could return invalid for an existing prefix. ::SeekForPrev might be called by the user, or internally on ::Prev, or within ::Seek if the return value involves Delete or a Merge operand.
@@ -17,12 +98,13 @@
 * Universal compaction to support options.periodic_compaction_seconds. A full compaction will be triggered if any file is over the threshold.
 * `GetLiveFilesMetaData` and `GetColumnFamilyMetaData` now expose the file number of SST files as well as the oldest blob file referenced by each SST.
 * A batched MultiGet API (DB::MultiGet()) that supports retrieving keys from multiple column families.
-* Full and partitioned filters in the block-based table use an improved Bloom filter implementation, enabled with format_version 5 (or above) because previous releases cannot read this filter. This replacement is faster and more accurate, especially for high bits per key or millions of keys in a single (full) filter. For example, the new Bloom filter has the same false postive rate at 9.55 bits per key as the old one at 10 bits per key, and a lower false positive rate at 16 bits per key than the old one at 100 bits per key.
+* Full and partitioned filters in the block-based table use an improved Bloom filter implementation, enabled with format_version 5 (or above) because previous releases cannot read this filter. This replacement is faster and more accurate, especially for high bits per key or millions of keys in a single (full) filter. For example, the new Bloom filter has the same false positive rate at 9.55 bits per key as the old one at 10 bits per key, and a lower false positive rate at 16 bits per key than the old one at 100 bits per key.
 * Added AVX2 instructions to USE_SSE builds to accelerate the new Bloom filter and XXH3-based hash function on compatible x86_64 platforms (Haswell and later, ~2014).
-* Support options.ttl with options.max_open_files = -1. File's oldest ancester time will be written to manifest. If it is availalbe, this information will be used instead of creation_time in table properties.
+* Support options.ttl or options.periodic_compaction_seconds with options.max_open_files = -1. File's oldest ancester time and file creation time will be written to manifest. If it is availalbe, this information will be used instead of creation_time and file_creation_time in table properties.
 * Setting options.ttl for universal compaction now has the same meaning as setting periodic_compaction_seconds.
+* SstFileMetaData also returns file creation time and oldest ancester time.
 * The `sst_dump` command line tool `recompress` command now displays how many blocks were compressed and how many were not, in particular how many were not compressed because the compression ratio was not met (12.5% threshold for GoodCompressionRatio), as seen in the `number.block.not_compressed` counter stat since version 6.0.0.
-* The block cache usage is now takes into account the overhead of metadata per each entry. This results into more accurate managment of memory. A side-effect of this feature is that less items are fit into the block cache of the same size, which would result to higher cache miss rates. This can be remedied by increasing the block cache size or passing kDontChargeCacheMetadata to its constuctor to restore the old behavior.
+* The block cache usage is now takes into account the overhead of metadata per each entry. This results into more accurate management of memory. A side-effect of this feature is that less items are fit into the block cache of the same size, which would result to higher cache miss rates. This can be remedied by increasing the block cache size or passing kDontChargeCacheMetadata to its constuctor to restore the old behavior.
 * When using BlobDB, a mapping is maintained and persisted in the MANIFEST between each SST file and the oldest non-TTL blob file it references.
 * `db_bench` now supports and by default issues non-TTL Puts to BlobDB. TTL Puts can be enabled by specifying a non-zero value for the `blob_db_max_ttl_range` command line parameter explicitly.
 * `sst_dump` now supports printing BlobDB blob indexes in a human-readable format. This can be enabled by specifying the `decode_blob_index` flag on the command line.
@@ -45,7 +127,7 @@
 
 ### Default Option Changes
 * Changed the default value of periodic_compaction_seconds to `UINT64_MAX - 1` which allows RocksDB to auto-tune periodic compaction scheduling. When using the default value, periodic compactions are now auto-enabled if a compaction filter is used. A value of `0` will turn off the feature completely.
-* Changed the default value of ttl to `UINT64_MAX - 1` which allows RocksDB to auto-tune ttl value. When using the default value, TTL will be auto-enabled to 30 days, when the feature is supported. To revert the old behavior, you can explictly set it to 0.
+* Changed the default value of ttl to `UINT64_MAX - 1` which allows RocksDB to auto-tune ttl value. When using the default value, TTL will be auto-enabled to 30 days, when the feature is supported. To revert the old behavior, you can explicitly set it to 0.
 
 ### Performance Improvements
 * For 64-bit hashing, RocksDB is standardizing on a slightly modified preview version of XXH3. This function is now used for many non-persisted hashes, along with fastrange64() in place of the modulus operator, and some benchmarks show a slight improvement.
@@ -86,7 +168,7 @@
 
 ## 6.4.0 (7/30/2019)
 ### Default Option Change
-* LRUCacheOptions.high_pri_pool_ratio is set to 0.5 (previously 0.0) by default, which means that by default midpoint insertion is enabled. The same change is made for the default value of high_pri_pool_ratio argument in NewLRUCache(). When block cache is not explictly created, the small block cache created by BlockBasedTable will still has this option to be 0.0.
+* LRUCacheOptions.high_pri_pool_ratio is set to 0.5 (previously 0.0) by default, which means that by default midpoint insertion is enabled. The same change is made for the default value of high_pri_pool_ratio argument in NewLRUCache(). When block cache is not explicitly created, the small block cache created by BlockBasedTable will still has this option to be 0.0.
 * Change BlockBasedTableOptions.cache_index_and_filter_blocks_with_high_priority's default value from false to true.
 
 ### Public API Change
@@ -110,7 +192,7 @@
 * Support loading custom objects in unit tests. In the affected unit tests, RocksDB will create custom Env objects based on environment variable TEST_ENV_URI. Users need to make sure custom object types are properly registered. For example, a static library should expose a `RegisterCustomObjects` function. By linking the unit test binary with the static library, the unit test can execute this function.
 
 ### Performance Improvements
-* Reduce iterator key comparision for upper/lower bound check.
+* Reduce iterator key comparison for upper/lower bound check.
 * Improve performance of row_cache: make reads with newer snapshots than data in an SST file share the same cache key, except in some transaction cases.
 * The compression dictionary is no longer copied to a new object upon retrieval.
 
@@ -206,7 +288,7 @@
 * Introduce two more stats levels, kExceptHistogramOrTimers and kExceptTimers.
 * Added a feature to perform data-block sampling for compressibility, and report stats to user.
 * Add support for trace filtering.
-* Add DBOptions.avoid_unnecessary_blocking_io. If true, we avoid file deletion when destorying ColumnFamilyHandle and Iterator. Instead, a job is scheduled to delete the files in background.
+* Add DBOptions.avoid_unnecessary_blocking_io. If true, we avoid file deletion when destroying ColumnFamilyHandle and Iterator. Instead, a job is scheduled to delete the files in background.
 
 ### Public API Change
 * Remove bundled fbson library.
@@ -585,7 +667,7 @@ if set to something > 0 user will see 2 changes in iterators behavior 1) only ke
 
 ## 5.2.0 (02/08/2017)
 ### Public API Change
-* NewLRUCache() will determine number of shard bits automatically based on capacity, if the user doesn't pass one. This also impacts the default block cache when the user doesn't explict provide one.
+* NewLRUCache() will determine number of shard bits automatically based on capacity, if the user doesn't pass one. This also impacts the default block cache when the user doesn't explicit provide one.
 * Change the default of delayed slowdown value to 16MB/s and further increase the L0 stop condition to 36 files.
 * Options::use_direct_writes and Options::use_direct_reads are now ready to use.
 * (Experimental) Two-level indexing that partition the index and creates a 2nd level index on the partitions. The feature can be enabled by setting kTwoLevelIndexSearch as IndexType and configuring index_per_partition.
@@ -656,7 +738,7 @@ if set to something > 0 user will see 2 changes in iterators behavior 1) only ke
 
 ### New Features
 * A tool to migrate DB after options change. See include/rocksdb/utilities/option_change_migration.h.
-* Add ReadOptions.background_purge_on_iterator_cleanup. If true, we avoid file deletion when destorying iterators.
+* Add ReadOptions.background_purge_on_iterator_cleanup. If true, we avoid file deletion when destroying iterators.
 
 ## 4.10.0 (7/5/2016)
 ### Public API Change

Makefile

@@ -207,6 +207,7 @@ AM_LINK = $(AM_V_CCLD)$(CXX) $^ $(EXEC_LDFLAGS) -o $@ $(LDFLAGS) $(COVERAGEFLAGS
 dummy := $(shell (export ROCKSDB_ROOT="$(CURDIR)"; export PORTABLE="$(PORTABLE)"; "$(CURDIR)/build_tools/build_detect_platform" "$(CURDIR)/make_config.mk"))
 # this file is generated by the previous line to set build flags and sources
 include make_config.mk
+export JAVAC_ARGS
 CLEAN_FILES += make_config.mk
 
 missing_make_config_paths := $(shell				\
@@ -216,7 +217,7 @@ missing_make_config_paths := $(shell				\
 	done | sort | uniq)
 
 $(foreach path, $(missing_make_config_paths), \
-	$(warning Warning: $(path) dont exist))
+	$(warning Warning: $(path) does not exist))
 
 ifeq ($(PLATFORM), OS_AIX)
 # no debug info
@@ -445,6 +446,7 @@ EXPOBJECTS = $(LIBOBJECTS) $(TESTUTIL)
 
 TESTS = \
 	db_basic_test \
+	db_with_timestamp_basic_test \
 	db_encryption_test \
 	db_test2 \
 	external_sst_file_basic_test \
@@ -459,7 +461,9 @@ TESTS = \
 	env_basic_test \
 	env_test \
 	env_logger_test \
+	io_posix_test \
 	hash_test \
+	random_test \
 	thread_local_test \
 	rate_limiter_test \
 	perf_context_test \
@@ -467,6 +471,7 @@ TESTS = \
 	db_wal_test \
 	db_block_cache_test \
 	db_test \
+	db_logical_block_size_cache_test \
 	db_blob_index_test \
 	db_iter_test \
 	db_iter_stress_test \
@@ -491,7 +496,7 @@ TESTS = \
 	db_table_properties_test \
 	db_statistics_test \
 	db_write_test \
-	error_handler_test \
+	error_handler_fs_test \
 	autovector_test \
 	blob_db_test \
 	cleanable_test \
@@ -502,6 +507,7 @@ TESTS = \
 	data_block_hash_index_test \
 	cache_test \
 	corruption_test \
+	slice_test \
 	slice_transform_test \
 	dbformat_test \
 	fault_injection_test \
@@ -594,6 +600,9 @@ TESTS = \
 	db_secondary_test \
 	block_cache_tracer_test \
 	block_cache_trace_analyzer_test \
+	defer_test \
+	blob_file_addition_test \
+	blob_file_garbage_test \
 
 ifeq ($(USE_FOLLY_DISTRIBUTED_MUTEX),1)
 	TESTS += folly_synchronization_distributed_mutex_test
@@ -743,7 +752,8 @@ endif  # PLATFORM_SHARED_EXT
 	release tags tags0 valgrind_check whitebox_crash_test format static_lib shared_lib all \
 	dbg rocksdbjavastatic rocksdbjava install install-static install-shared uninstall \
 	analyze tools tools_lib \
-	blackbox_crash_test_with_atomic_flush whitebox_crash_test_with_atomic_flush
+	blackbox_crash_test_with_atomic_flush whitebox_crash_test_with_atomic_flush  \
+	blackbox_crash_test_with_txn whitebox_crash_test_with_txn
 
 
 all: $(LIBRARY) $(BENCHMARKS) tools tools_lib test_libs $(TESTS)
@@ -933,10 +943,11 @@ check: all
 	    $(MAKE) T="$$t" TMPD=$(TMPD) check_0;                       \
 	else                                                            \
 	    for t in $(TESTS); do                                       \
-	      echo "===== Running $$t"; ./$$t || exit 1; done;          \
+	      echo "===== Running $$t (`date`)"; ./$$t || exit 1; done;          \
 	fi
 	rm -rf $(TMPD)
 ifneq ($(PLATFORM), OS_AIX)
+	python tools/check_all_python.py
 ifeq ($(filter -DROCKSDB_LITE,$(OPT)),)
 	python tools/ldb_test.py
 	sh tools/rocksdb_dump_test.sh
@@ -945,7 +956,7 @@ endif
 
 # TODO add ldb_tests
 check_some: $(SUBSET)
-	for t in $(SUBSET); do echo "===== Running $$t"; ./$$t || exit 1; done
+	for t in $(SUBSET); do echo "===== Running $$t (`date`)"; ./$$t || exit 1; done
 
 .PHONY: ldb_tests
 ldb_tests: ldb
@@ -955,6 +966,8 @@ crash_test: whitebox_crash_test blackbox_crash_test
 
 crash_test_with_atomic_flush: whitebox_crash_test_with_atomic_flush blackbox_crash_test_with_atomic_flush
 
+crash_test_with_txn: whitebox_crash_test_with_txn blackbox_crash_test_with_txn
+
 blackbox_crash_test: db_stress
 	python -u tools/db_crashtest.py --simple blackbox $(CRASH_TEST_EXT_ARGS)
 	python -u tools/db_crashtest.py blackbox $(CRASH_TEST_EXT_ARGS)
@@ -962,6 +975,9 @@ blackbox_crash_test: db_stress
 blackbox_crash_test_with_atomic_flush: db_stress
 	python -u tools/db_crashtest.py --cf_consistency blackbox $(CRASH_TEST_EXT_ARGS)
 
+blackbox_crash_test_with_txn: db_stress
+	python -u tools/db_crashtest.py --txn blackbox $(CRASH_TEST_EXT_ARGS)
+
 ifeq ($(CRASH_TEST_KILL_ODD),)
   CRASH_TEST_KILL_ODD=888887
 endif
@@ -976,6 +992,10 @@ whitebox_crash_test_with_atomic_flush: db_stress
 	python -u tools/db_crashtest.py --cf_consistency whitebox  --random_kill_odd \
       $(CRASH_TEST_KILL_ODD) $(CRASH_TEST_EXT_ARGS)
 
+whitebox_crash_test_with_txn: db_stress
+	python -u tools/db_crashtest.py --txn whitebox --random_kill_odd \
+      $(CRASH_TEST_KILL_ODD) $(CRASH_TEST_EXT_ARGS)
+
 asan_check:
 	$(MAKE) clean
 	COMPILE_WITH_ASAN=1 $(MAKE) check -j32
@@ -991,6 +1011,11 @@ asan_crash_test_with_atomic_flush:
 	COMPILE_WITH_ASAN=1 $(MAKE) crash_test_with_atomic_flush
 	$(MAKE) clean
 
+asan_crash_test_with_txn:
+	$(MAKE) clean
+	COMPILE_WITH_ASAN=1 $(MAKE) crash_test_with_txn
+	$(MAKE) clean
+
 ubsan_check:
 	$(MAKE) clean
 	COMPILE_WITH_UBSAN=1 $(MAKE) check -j32
@@ -1006,6 +1031,11 @@ ubsan_crash_test_with_atomic_flush:
 	COMPILE_WITH_UBSAN=1 $(MAKE) crash_test_with_atomic_flush
 	$(MAKE) clean
 
+ubsan_crash_test_with_txn:
+	$(MAKE) clean
+	COMPILE_WITH_UBSAN=1 $(MAKE) crash_test_with_txn
+	$(MAKE) clean
+
 valgrind_test:
 	ROCKSDB_VALGRIND_RUN=1 DISABLE_JEMALLOC=1 $(MAKE) valgrind_check
 
@@ -1032,7 +1062,7 @@ ifneq ($(PAR_TEST),)
 parloop:
 	ret_bad=0;							\
 	for t in $(PAR_TEST); do		\
-		echo "===== Running $$t in parallel $(NUM_PAR)";\
+		echo "===== Running $$t in parallel $(NUM_PAR) (`date`)";\
 		if [ $(db_test) -eq 1 ]; then \
 			seq $(J) | v="$$t" build_tools/gnu_parallel --gnu --plain 's=$(TMPD)/rdb-{};  export TEST_TMPDIR=$$s;' \
 				'timeout 2m ./db_test --gtest_filter=$$v >> $$s/log-{} 2>1'; \
@@ -1075,6 +1105,9 @@ parallel_check: $(TESTS)
 		TMPD=$(TMPD) J=$(J) db_test=0 parloop;
 
 analyze: clean
+	USE_CLANG=1 $(MAKE) analyze_incremental
+
+analyze_incremental:
 	$(CLANG_SCAN_BUILD) --use-analyzer=$(CLANG_ANALYZER) \
 		--use-c++=$(CXX) --use-cc=$(CC) --status-bugs \
 		-o $(CURDIR)/scan_build_report \
@@ -1103,16 +1136,21 @@ unity_test: db/db_test.o db/db_test_util.o $(TESTHARNESS) $(TOOLLIBOBJECTS) unit
 rocksdb.h rocksdb.cc: build_tools/amalgamate.py Makefile $(LIB_SOURCES) unity.cc
 	build_tools/amalgamate.py -I. -i./include unity.cc -x include/rocksdb/c.h -H rocksdb.h -o rocksdb.cc
 
-clean: clean-ext-libraries-all clean-rocks
+clean: clean-ext-libraries-all clean-rocks clean-rocksjava
 
-clean-not-downloaded: clean-ext-libraries-bin clean-rocks
+clean-not-downloaded: clean-ext-libraries-bin clean-rocks clean-not-downloaded-rocksjava
 
 clean-rocks:
 	rm -f $(BENCHMARKS) $(TOOLS) $(TESTS) $(PARALLEL_TEST) $(LIBRARY) $(SHARED)
 	rm -rf $(CLEAN_FILES) ios-x86 ios-arm scan_build_report
 	$(FIND) . -name "*.[oda]" -exec rm -f {} \;
 	$(FIND) . -type f -regex ".*\.\(\(gcda\)\|\(gcno\)\)" -exec rm {} \;
-	cd java; $(MAKE) clean
+
+clean-rocksjava:
+	cd java && $(MAKE) clean
+
+clean-not-downloaded-rocksjava:
+	cd java && $(MAKE) clean-not-downloaded
 
 clean-ext-libraries-all:
 	rm -rf bzip2* snappy* zlib* lz4* zstd*
@@ -1183,7 +1221,7 @@ memtablerep_bench: memtable/memtablerep_bench.o $(LIBOBJECTS) $(TESTUTIL)
 filter_bench: util/filter_bench.o $(LIBOBJECTS) $(TESTUTIL)
 	$(AM_LINK)
 
-db_stress: tools/db_stress.o $(STRESSTOOLOBJECTS)
+db_stress: db_stress_tool/db_stress.o $(STRESSTOOLOBJECTS)
 	$(AM_LINK)
 
 write_stress: tools/write_stress.o $(LIBOBJECTS) $(TESTUTIL)
@@ -1225,6 +1263,9 @@ coding_test: util/coding_test.o $(LIBOBJECTS) $(TESTHARNESS)
 hash_test: util/hash_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
+random_test: util/random_test.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 option_change_migration_test: utilities/option_change_migration/option_change_migration_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
@@ -1258,12 +1299,18 @@ corruption_test: db/corruption_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARN
 crc32c_test: util/crc32c_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
+slice_test: util/slice_test.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 slice_transform_test: util/slice_transform_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
 db_basic_test: db/db_basic_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
+db_with_timestamp_basic_test: db/db_with_timestamp_basic_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 db_encryption_test: db/db_encryption_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
@@ -1273,7 +1320,10 @@ db_test: db/db_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 db_test2: db/db_test2.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
-db_blob_index_test: db/db_blob_index_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
+db_logical_block_size_cache_test: db/db_logical_block_size_cache_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
+db_blob_index_test: db/blob/db_blob_index_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
 db_block_cache_test: db/db_block_cache_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1327,7 +1377,7 @@ db_statistics_test: db/db_statistics_test.o db/db_test_util.o $(LIBOBJECTS) $(TE
 db_write_test: db/db_write_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
-error_handler_test: db/error_handler_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
+error_handler_fs_test: db/error_handler_fs_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
 external_sst_file_basic_test: db/external_sst_file_basic_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1440,6 +1490,9 @@ env_basic_test: env/env_basic_test.o $(LIBOBJECTS) $(TESTHARNESS)
 env_test: env/env_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
+io_posix_test: env/io_posix_test.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 fault_injection_test: db/fault_injection_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
@@ -1677,6 +1730,15 @@ block_cache_tracer_test: trace_replay/block_cache_tracer_test.o trace_replay/blo
 block_cache_trace_analyzer_test: tools/block_cache_analyzer/block_cache_trace_analyzer_test.o tools/block_cache_analyzer/block_cache_trace_analyzer.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
+defer_test: util/defer_test.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
+blob_file_addition_test: db/blob/blob_file_addition_test.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
+blob_file_garbage_test: db/blob/blob_file_garbage_test.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 #-------------------------------------------------
 # make install related stuff
 INSTALL_PATH ?= /usr/local
@@ -1759,8 +1821,8 @@ ZLIB_DOWNLOAD_BASE ?= http://zlib.net
 BZIP2_VER ?= 1.0.6
 BZIP2_SHA256 ?= a2848f34fcd5d6cf47def00461fcb528a0484d8edef8208d6d2e2909dc61d9cd
 BZIP2_DOWNLOAD_BASE ?= https://downloads.sourceforge.net/project/bzip2
-SNAPPY_VER ?= 1.1.7
-SNAPPY_SHA256 ?= 3dfa02e873ff51a11ee02b9ca391807f0c8ea0529a4924afa645fbf97163f9d4
+SNAPPY_VER ?= 1.1.8
+SNAPPY_SHA256 ?= 16b677f07832a612b0836178db7f374e414f94657c138e6993cbfc5dcc58651f
 SNAPPY_DOWNLOAD_BASE ?= https://github.com/google/snappy/archive
 LZ4_VER ?= 1.9.2
 LZ4_SHA256 ?= 658ba6191fa44c92280d4aa2c271b0f4fbc0e34d249578dd05e50e76d0e5efcc
@@ -1806,7 +1868,7 @@ endif
 libz.a:
 	-rm -rf zlib-$(ZLIB_VER)
 ifeq (,$(wildcard ./zlib-$(ZLIB_VER).tar.gz))
-	curl --output zlib-$(ZLIB_VER).tar.gz -L ${ZLIB_DOWNLOAD_BASE}/zlib-$(ZLIB_VER).tar.gz
+	curl --fail --output zlib-$(ZLIB_VER).tar.gz --location ${ZLIB_DOWNLOAD_BASE}/zlib-$(ZLIB_VER).tar.gz
 endif
 	ZLIB_SHA256_ACTUAL=`$(SHA256_CMD) zlib-$(ZLIB_VER).tar.gz | cut -d ' ' -f 1`; \
 	if [ "$(ZLIB_SHA256)" != "$$ZLIB_SHA256_ACTUAL" ]; then \
@@ -1820,7 +1882,7 @@ endif
 libbz2.a:
 	-rm -rf bzip2-$(BZIP2_VER)
 ifeq (,$(wildcard ./bzip2-$(BZIP2_VER).tar.gz))
-	curl --output bzip2-$(BZIP2_VER).tar.gz -L ${CURL_SSL_OPTS} ${BZIP2_DOWNLOAD_BASE}/bzip2-$(BZIP2_VER).tar.gz
+	curl --fail --output bzip2-$(BZIP2_VER).tar.gz --location ${CURL_SSL_OPTS} ${BZIP2_DOWNLOAD_BASE}/bzip2-$(BZIP2_VER).tar.gz
 endif
 	BZIP2_SHA256_ACTUAL=`$(SHA256_CMD) bzip2-$(BZIP2_VER).tar.gz | cut -d ' ' -f 1`; \
 	if [ "$(BZIP2_SHA256)" != "$$BZIP2_SHA256_ACTUAL" ]; then \
@@ -1834,7 +1896,7 @@ endif
 libsnappy.a:
 	-rm -rf snappy-$(SNAPPY_VER)
 ifeq (,$(wildcard ./snappy-$(SNAPPY_VER).tar.gz))
-	curl --output snappy-$(SNAPPY_VER).tar.gz -L ${CURL_SSL_OPTS} ${SNAPPY_DOWNLOAD_BASE}/$(SNAPPY_VER).tar.gz
+	curl --fail --output snappy-$(SNAPPY_VER).tar.gz --location ${CURL_SSL_OPTS} ${SNAPPY_DOWNLOAD_BASE}/$(SNAPPY_VER).tar.gz
 endif
 	SNAPPY_SHA256_ACTUAL=`$(SHA256_CMD) snappy-$(SNAPPY_VER).tar.gz | cut -d ' ' -f 1`; \
 	if [ "$(SNAPPY_SHA256)" != "$$SNAPPY_SHA256_ACTUAL" ]; then \
@@ -1849,7 +1911,7 @@ endif
 liblz4.a:
 	-rm -rf lz4-$(LZ4_VER)
 ifeq (,$(wildcard ./lz4-$(LZ4_VER).tar.gz))
-	curl --output lz4-$(LZ4_VER).tar.gz -L ${CURL_SSL_OPTS} ${LZ4_DOWNLOAD_BASE}/v$(LZ4_VER).tar.gz
+	curl --fail --output lz4-$(LZ4_VER).tar.gz --location ${CURL_SSL_OPTS} ${LZ4_DOWNLOAD_BASE}/v$(LZ4_VER).tar.gz
 endif
 	LZ4_SHA256_ACTUAL=`$(SHA256_CMD) lz4-$(LZ4_VER).tar.gz | cut -d ' ' -f 1`; \
 	if [ "$(LZ4_SHA256)" != "$$LZ4_SHA256_ACTUAL" ]; then \
@@ -1863,7 +1925,7 @@ endif
 libzstd.a:
 	-rm -rf zstd-$(ZSTD_VER)
 ifeq (,$(wildcard ./zstd-$(ZSTD_VER).tar.gz))
-	curl --output zstd-$(ZSTD_VER).tar.gz -L ${CURL_SSL_OPTS} ${ZSTD_DOWNLOAD_BASE}/v$(ZSTD_VER).tar.gz
+	curl --fail --output zstd-$(ZSTD_VER).tar.gz --location ${CURL_SSL_OPTS} ${ZSTD_DOWNLOAD_BASE}/v$(ZSTD_VER).tar.gz
 endif
 	ZSTD_SHA256_ACTUAL=`$(SHA256_CMD) zstd-$(ZSTD_VER).tar.gz | cut -d ' ' -f 1`; \
 	if [ "$(ZSTD_SHA256)" != "$$ZSTD_SHA256_ACTUAL" ]; then \
@@ -1933,35 +1995,35 @@ rocksdbjavastaticreleasedocker: rocksdbjavastatic rocksdbjavastaticdockerx86 roc
 
 rocksdbjavastaticdockerx86:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_x86-be --attach stdin --attach stdout --attach stderr --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:centos6_x86-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
+	docker run --rm --name rocksdb_linux_x86-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:centos6_x86-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
 
 rocksdbjavastaticdockerx86_64:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_x64-be --attach stdin --attach stdout --attach stderr --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:centos6_x64-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
+	docker run --rm --name rocksdb_linux_x64-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:centos6_x64-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
 
 rocksdbjavastaticdockerppc64le:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_ppc64le-be --attach stdin --attach stdout --attach stderr --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:centos7_ppc64le-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
+	docker run --rm --name rocksdb_linux_ppc64le-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:centos7_ppc64le-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
 
 rocksdbjavastaticdockerarm64v8:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_arm64v8-be --attach stdin --attach stdout --attach stderr --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:centos7_arm64v8-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
+	docker run --rm --name rocksdb_linux_arm64v8-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:centos7_arm64v8-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
 
 rocksdbjavastaticdockerx86musl:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_x86-musl-be --attach stdin --attach stdout --attach stderr --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:alpine3_x86-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
+	docker run --rm --name rocksdb_linux_x86-musl-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:alpine3_x86-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
 
 rocksdbjavastaticdockerx86_64musl:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_x64-musl-be --attach stdin --attach stdout --attach stderr --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:alpine3_x64-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
+	docker run --rm --name rocksdb_linux_x64-musl-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:alpine3_x64-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
 
 rocksdbjavastaticdockerppc64lemusl:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_ppc64le-musl-be --attach stdin --attach stdout --attach stderr --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:alpine3_ppc64le-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
+	docker run --rm --name rocksdb_linux_ppc64le-musl-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:alpine3_ppc64le-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
 
 rocksdbjavastaticdockerarm64v8musl:
 	mkdir -p java/target
-	docker run --rm --name rocksdb_linux_arm64v8-musl-be --attach stdin --attach stdout --attach stderr --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:alpine3_arm64v8-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
+	docker run --rm --name rocksdb_linux_arm64v8-musl-be --attach stdin --attach stdout --attach stderr --volume $(HOME)/.m2:/root/.m2:ro --volume `pwd`:/rocksdb-host:ro --volume /rocksdb-local-build --volume `pwd`/java/target:/rocksdb-java-target --env DEBUG_LEVEL=$(DEBUG_LEVEL) evolvedbinary/rocksjava:alpine3_arm64v8-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh
 
 rocksdbjavastaticpublish: rocksdbjavastaticrelease rocksdbjavastaticpublishcentral
 
@@ -2027,6 +2089,7 @@ jtest_run:
 
 jtest: rocksdbjava
 	cd java;$(MAKE) sample;$(MAKE) test;
+	python tools/check_all_python.py # TODO peterd: find a better place for this check in CI targets
 
 jdb_bench:
 	cd java;$(MAKE) db_bench;
@@ -2074,6 +2137,9 @@ endif
 .cc.o:
 	$(AM_V_CC)$(CXX) $(CXXFLAGS) -c $< -o $@ $(COVERAGEFLAGS)
 
+.cpp.o:
+	$(AM_V_CC)$(CXX) $(CXXFLAGS) -c $< -o $@ $(COVERAGEFLAGS)
+
 .c.o:
 	$(AM_V_CC)$(CC) $(CFLAGS) -c $< -o $@
 endif
@@ -2084,6 +2150,10 @@ endif
 all_sources = $(LIB_SOURCES) $(MAIN_SOURCES) $(MOCK_LIB_SOURCES) $(TOOL_LIB_SOURCES) $(BENCH_LIB_SOURCES) $(TEST_LIB_SOURCES) $(ANALYZER_LIB_SOURCES) $(STRESS_LIB_SOURCES)
 DEPFILES = $(all_sources:.cc=.cc.d)
 
+ifeq ($(USE_FOLLY_DISTRIBUTED_MUTEX),1)
+  DEPFILES += $(FOLLY_SOURCES:.cpp=.cpp.d)
+endif
+
 # Add proper dependency support so changing a .h file forces a .cc file to
 # rebuild.
 
@@ -2093,6 +2163,10 @@ DEPFILES = $(all_sources:.cc=.cc.d)
 	@$(CXX) $(CXXFLAGS) $(PLATFORM_SHARED_CFLAGS) \
 	  -MM -MT'$@' -MT'$(<:.cc=.o)' "$<" -o '$@'
 
+%.cpp.d: %.cpp
+	@$(CXX) $(CXXFLAGS) $(PLATFORM_SHARED_CFLAGS) \
+	  -MM -MT'$@' -MT'$(<:.cpp=.o)' "$<" -o '$@'
+
 ifeq ($(HAVE_POWER8),1)
 DEPFILES_C = $(LIB_SOURCES_C:.c=.c.d)
 DEPFILES_ASM = $(LIB_SOURCES_ASM:.S=.S.d)

TARGETS

@@ -32,7 +32,7 @@ ROCKSDB_EXTERNAL_DEPS = [
 ROCKSDB_OS_DEPS = [
     (
         "linux",
-        ["third-party//numa:numa"],
+        ["third-party//numa:numa", "third-party//liburing:uring"],
     ),
 ]
 
@@ -46,7 +46,9 @@ ROCKSDB_OS_PREPROCESSOR_FLAGS = [
             "-DROCKSDB_PTHREAD_ADAPTIVE_MUTEX",
             "-DROCKSDB_RANGESYNC_PRESENT",
             "-DROCKSDB_SCHED_GETCPU_PRESENT",
+            "-DROCKSDB_IOURING_PRESENT",
             "-DHAVE_SSE42",
+            "-DLIBURING",
             "-DNUMA",
         ],
     ),
@@ -114,6 +116,8 @@ cpp_library(
         "cache/lru_cache.cc",
         "cache/sharded_cache.cc",
         "db/arena_wrapped_db_iter.cc",
+        "db/blob/blob_file_addition.cc",
+        "db/blob/blob_file_garbage.cc",
         "db/builder.cc",
         "db/c.cc",
         "db/column_family.cc",
@@ -178,6 +182,8 @@ cpp_library(
         "env/env_encryption.cc",
         "env/env_hdfs.cc",
         "env/env_posix.cc",
+        "env/file_system.cc",
+        "env/fs_posix.cc",
         "env/io_posix.cc",
         "env/mock_env.cc",
         "file/delete_scheduler.cc",
@@ -225,12 +231,15 @@ cpp_library(
         "port/port_posix.cc",
         "port/stack_trace.cc",
         "table/adaptive/adaptive_table_factory.cc",
+        "table/block_based/binary_search_index_reader.cc",
         "table/block_based/block.cc",
         "table/block_based/block_based_filter_block.cc",
         "table/block_based/block_based_table_builder.cc",
         "table/block_based/block_based_table_factory.cc",
+        "table/block_based/block_based_table_iterator.cc",
         "table/block_based/block_based_table_reader.cc",
         "table/block_based/block_builder.cc",
+        "table/block_based/block_prefetcher.cc",
         "table/block_based/block_prefix_index.cc",
         "table/block_based/data_block_footer.cc",
         "table/block_based/data_block_hash_index.cc",
@@ -238,9 +247,14 @@ cpp_library(
         "table/block_based/filter_policy.cc",
         "table/block_based/flush_block_policy.cc",
         "table/block_based/full_filter_block.cc",
+        "table/block_based/hash_index_reader.cc",
         "table/block_based/index_builder.cc",
+        "table/block_based/index_reader_common.cc",
         "table/block_based/parsed_full_filter_block.cc",
         "table/block_based/partitioned_filter_block.cc",
+        "table/block_based/partitioned_index_iterator.cc",
+        "table/block_based/partitioned_index_reader.cc",
+        "table/block_based/reader_common.cc",
         "table/block_based/uncompression_dict_reader.cc",
         "table/block_fetcher.cc",
         "table/cuckoo/cuckoo_table_builder.cc",
@@ -279,6 +293,7 @@ cpp_library(
         "util/concurrent_task_limiter_impl.cc",
         "util/crc32c.cc",
         "util/dynamic_bloom.cc",
+        "util/file_checksum_helper.cc",
         "util/hash.cc",
         "util/murmurhash.cc",
         "util/random.cc",
@@ -362,6 +377,7 @@ cpp_library(
         "db/db_test_util.cc",
         "table/mock_table.cc",
         "test_util/fault_injection_test_env.cc",
+        "test_util/fault_injection_test_fs.cc",
         "test_util/testharness.cc",
         "test_util/testutil.cc",
         "tools/block_cache_analyzer/block_cache_trace_analyzer.cc",
@@ -399,9 +415,17 @@ cpp_library(
 cpp_library(
     name = "rocksdb_stress_lib",
     srcs = [
+        "db_stress_tool/batched_ops_stress.cc",
+        "db_stress_tool/cf_consistency_stress.cc",
+        "db_stress_tool/db_stress_common.cc",
+        "db_stress_tool/db_stress_driver.cc",
+        "db_stress_tool/db_stress_gflags.cc",
+        "db_stress_tool/db_stress_shared_state.cc",
+        "db_stress_tool/db_stress_test_base.cc",
+        "db_stress_tool/db_stress_tool.cc",
+        "db_stress_tool/no_batched_ops_stress.cc",
         "test_util/testutil.cc",
         "tools/block_cache_analyzer/block_cache_trace_analyzer.cc",
-        "tools/db_stress_tool.cc",
         "tools/trace_analyzer_tool.cc",
     ],
     auto_headers = AutoHeaders.RECURSIVE_GLOB,
@@ -464,6 +488,20 @@ ROCKS_TESTS = [
         [],
         [],
     ],
+    [
+        "blob_file_addition_test",
+        "db/blob/blob_file_addition_test.cc",
+        "serial",
+        [],
+        [],
+    ],
+    [
+        "blob_file_garbage_test",
+        "db/blob/blob_file_garbage_test.cc",
+        "serial",
+        [],
+        [],
+    ],
     [
         "block_based_filter_block_test",
         "table/block_based/block_based_filter_block_test.cc",
@@ -676,7 +714,7 @@ ROCKS_TESTS = [
     ],
     [
         "db_blob_index_test",
-        "db/db_blob_index_test.cc",
+        "db/blob/db_blob_index_test.cc",
         "serial",
         [],
         [],
@@ -772,6 +810,13 @@ ROCKS_TESTS = [
         [],
         [],
     ],
+    [
+        "db_logical_block_size_cache_test",
+        "db/db_logical_block_size_cache_test.cc",
+        "serial",
+        [],
+        [],
+    ],
     [
         "db_memtable_test",
         "db/db_memtable_test.cc",
@@ -877,6 +922,13 @@ ROCKS_TESTS = [
         [],
         [],
     ],
+    [
+        "db_with_timestamp_basic_test",
+        "db/db_with_timestamp_basic_test.cc",
+        "serial",
+        [],
+        [],
+    ],
     [
         "db_write_test",
         "db/db_write_test.cc",
@@ -891,6 +943,13 @@ ROCKS_TESTS = [
         [],
         [],
     ],
+    [
+        "defer_test",
+        "util/defer_test.cc",
+        "serial",
+        [],
+        [],
+    ],
     [
         "delete_scheduler_test",
         "file/delete_scheduler_test.cc",
@@ -941,8 +1000,8 @@ ROCKS_TESTS = [
         [],
     ],
     [
-        "error_handler_test",
-        "db/error_handler_test.cc",
+        "error_handler_fs_test",
+        "db/error_handler_fs_test.cc",
         "serial",
         [],
         [],
@@ -1059,6 +1118,13 @@ ROCKS_TESTS = [
         [],
         [],
     ],
+    [
+        "io_posix_test",
+        "env/io_posix_test.cc",
+        "serial",
+        [],
+        [],
+    ],
     [
         "iostats_context_test",
         "monitoring/iostats_context_test.cc",
@@ -1234,6 +1300,13 @@ ROCKS_TESTS = [
         [],
         [],
     ],
+    [
+        "random_test",
+        "util/random_test.cc",
+        "serial",
+        [],
+        [],
+    ],
     [
         "range_del_aggregator_test",
         "db/range_del_aggregator_test.cc",
@@ -1290,6 +1363,13 @@ ROCKS_TESTS = [
         [],
         [],
     ],
+    [
+        "slice_test",
+        "util/slice_test.cc",
+        "serial",
+        [],
+        [],
+    ],
     [
         "slice_transform_test",
         "util/slice_transform_test.cc",

USERS.md

@@ -102,3 +102,7 @@ LzLabs is using RocksDB as a storage engine in their multi-database distributed
  
 ## Crux
 [Crux](https://github.com/juxt/crux) is a document database that uses RocksDB for local [EAV](https://en.wikipedia.org/wiki/Entity%E2%80%93attribute%E2%80%93value_model) index storage to enable point-in-time bitemporal Datalog queries. The "unbundled" architecture uses Kafka to provide horizontal scalability.
+
+## Nebula Graph
+
+[Nebula Graph](https://github.com/vesoft-inc/nebula) is a distributed, scalable, lightning-fast, open source graph database capable of hosting super large scale graphs with dozens of billions of vertices (nodes) and trillions of edges, with milliseconds of latency.

appveyor.yml

@@ -17,41 +17,49 @@ environment:
   ZSTD_INCLUDE: $(ZSTD_HOME)\lib;$(ZSTD_HOME)\lib\dictBuilder
   ZSTD_LIB_DEBUG: $(ZSTD_HOME)\build\VS2010\bin\x64_Debug\libzstd_static.lib
   ZSTD_LIB_RELEASE: $(ZSTD_HOME)\build\VS2010\bin\x64_Release\libzstd_static.lib
+  matrix:
+    - APPVEYOR_BUILD_WORKER_IMAGE: Visual Studio 2015
+      CMAKE_GENERATOR: Visual Studio 14 Win64
+      DEV_ENV: C:\Program Files (x86)\Microsoft Visual Studio 14.0\Common7\IDE\devenv.com
+    - APPVEYOR_BUILD_WORKER_IMAGE: Visual Studio 2017
+      CMAKE_GENERATOR: Visual Studio 15 Win64
+      DEV_ENV: C:\Program Files (x86)\Microsoft Visual Studio\2017\Community\Common7\IDE\devenv.com
 
 install:
   - md %THIRDPARTY_HOME%
   - echo "Building Snappy dependency..."
   - cd %THIRDPARTY_HOME%
-  - curl -fsSL -o snappy-1.1.7.zip https://github.com/google/snappy/archive/1.1.7.zip
+  - curl --fail --silent --show-error --output snappy-1.1.7.zip --location https://github.com/google/snappy/archive/1.1.7.zip
   - unzip snappy-1.1.7.zip
   - cd snappy-1.1.7
   - mkdir build
   - cd build
-  - cmake -DCMAKE_GENERATOR_PLATFORM=x64 ..
+  - cmake -G "%CMAKE_GENERATOR%" ..
   - msbuild Snappy.sln /p:Configuration=Debug /p:Platform=x64
   - msbuild Snappy.sln /p:Configuration=Release /p:Platform=x64
   - echo "Building LZ4 dependency..."
   - cd %THIRDPARTY_HOME%
-  - curl -fsSL -o lz4-1.8.3.zip https://github.com/lz4/lz4/archive/v1.8.3.zip
+  - curl --fail --silent --show-error --output lz4-1.8.3.zip --location https://github.com/lz4/lz4/archive/v1.8.3.zip
   - unzip lz4-1.8.3.zip
   - cd lz4-1.8.3\visual\VS2010
-  - ps: $CMD="C:\Program Files (x86)\Microsoft Visual Studio\2017\Community\Common7\IDE\devenv.com"; & $CMD lz4.sln /upgrade
+  - ps: $CMD="$Env:DEV_ENV"; & $CMD lz4.sln /upgrade
   - msbuild lz4.sln /p:Configuration=Debug /p:Platform=x64
   - msbuild lz4.sln /p:Configuration=Release /p:Platform=x64
   - echo "Building ZStd dependency..."
   - cd %THIRDPARTY_HOME%
-  - curl -fsSL -o zstd-1.4.0.zip https://github.com/facebook/zstd/archive/v1.4.0.zip
+  - curl --fail --silent --show-error --output zstd-1.4.0.zip --location https://github.com/facebook/zstd/archive/v1.4.0.zip
   - unzip zstd-1.4.0.zip
   - cd zstd-1.4.0\build\VS2010
-  - ps: $CMD="C:\Program Files (x86)\Microsoft Visual Studio\2017\Community\Common7\IDE\devenv.com"; & $CMD zstd.sln /upgrade
+  - ps: $CMD="$Env:DEV_ENV"; & $CMD zstd.sln /upgrade
   - msbuild zstd.sln /p:Configuration=Debug /p:Platform=x64
   - msbuild zstd.sln /p:Configuration=Release /p:Platform=x64
 
 before_build:
   - md %APPVEYOR_BUILD_FOLDER%\build
   - cd %APPVEYOR_BUILD_FOLDER%\build
-  - cmake -G "Visual Studio 15 Win64" -DOPTDBG=1 -DPORTABLE=1 -DSNAPPY=1 -DLZ4=1 -DZSTD=1 -DXPRESS=1 -DJNI=1 ..
+  - cmake -G "%CMAKE_GENERATOR%" -DCMAKE_BUILD_TYPE=Debug -DOPTDBG=1 -DPORTABLE=1 -DSNAPPY=1 -DLZ4=1 -DZSTD=1 -DXPRESS=1 -DJNI=1 ..
   - cd ..
+
 build:
   project: build\rocksdb.sln
   parallel: true
@@ -60,7 +68,7 @@ build:
 test:
 
 test_script:
-  - ps: build_tools\run_ci_db_test.ps1 -SuiteRun db_basic_test,db_test2,db_test,env_basic_test,env_test,db_merge_operand_test -Concurrency 8
+  - ps: build_tools\run_ci_db_test.ps1 -SuiteRun db_basic_test,db_with_timestamp_basic_test,db_test2,db_test,env_basic_test,env_test,db_merge_operand_test -Concurrency 8
 
 on_failure:
   - cmd: 7z a build-failed.zip %APPVEYOR_BUILD_FOLDER%\build\ && appveyor PushArtifact build-failed.zip

buckifier/targets_cfg.py

@@ -38,7 +38,7 @@ ROCKSDB_EXTERNAL_DEPS = [
 ROCKSDB_OS_DEPS = [
     (
         "linux",
-        ["third-party//numa:numa"],
+        ["third-party//numa:numa", "third-party//liburing:uring"],
     ),
 ]
 
@@ -52,7 +52,9 @@ ROCKSDB_OS_PREPROCESSOR_FLAGS = [
             "-DROCKSDB_PTHREAD_ADAPTIVE_MUTEX",
             "-DROCKSDB_RANGESYNC_PRESENT",
             "-DROCKSDB_SCHED_GETCPU_PRESENT",
+            "-DROCKSDB_IOURING_PRESENT",
             "-DHAVE_SSE42",
+            "-DLIBURING",
             "-DNUMA",
         ],
     ),

build_tools/build_detect_platform

@@ -9,6 +9,7 @@
 #   PLATFORM_LDFLAGS            Linker flags
 #   JAVA_LDFLAGS                Linker flags for RocksDBJava
 #   JAVA_STATIC_LDFLAGS         Linker flags for RocksDBJava static build
+#   JAVAC_ARGS                  Arguments for javac
 #   PLATFORM_SHARED_EXT         Extension for shared libraries
 #   PLATFORM_SHARED_LDFLAGS     Flags for building shared library
 #   PLATFORM_SHARED_CFLAGS      Flags for compiling objects for shared library
@@ -150,6 +151,21 @@ case "$TARGET_OS" in
             PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -latomic"
         fi
         PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lpthread -lrt"
+        if test $ROCKSDB_USE_IO_URING; then
+            # check for liburing
+            $CXX $CFLAGS -x c++ - -luring -o /dev/null 2>/dev/null  <<EOF
+              #include <liburing.h>
+              int main() {
+                struct io_uring ring;
+                io_uring_queue_init(1, &ring, 0);
+                return 0;
+              }
+EOF
+            if [ "$?" = 0 ]; then
+                PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -luring"
+                COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_IOURING_PRESENT"
+            fi
+        fi
         if test -z "$USE_FOLLY_DISTRIBUTED_MUTEX"; then
           USE_FOLLY_DISTRIBUTED_MUTEX=1
         fi
@@ -224,6 +240,7 @@ esac
 PLATFORM_CXXFLAGS="$PLATFORM_CXXFLAGS ${CXXFLAGS}"
 JAVA_LDFLAGS="$PLATFORM_LDFLAGS"
 JAVA_STATIC_LDFLAGS="$PLATFORM_LDFLAGS"
+JAVAC_ARGS="-source 7"
 
 if [ "$CROSS_COMPILE" = "true" -o "$FBCODE_BUILD" = "true" ]; then
     # Cross-compiling; do not try any compilation tests.
@@ -493,6 +510,20 @@ EOF
         fi
     fi
 
+    if ! test $ROCKSDB_DISABLE_AUXV_GETAUXVAL; then
+        # Test whether getauxval is supported
+        $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
+          #include <sys/auxv.h>
+          int main() {
+            uint64_t auxv = getauxval(AT_HWCAP);
+            (void)auxv;
+          }
+EOF
+        if [ "$?" = 0 ]; then
+            COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_AUXV_GETAUXVAL_PRESENT"
+        fi
+    fi
+
     if ! test $ROCKSDB_DISABLE_ALIGNED_NEW; then
         # Test whether c++17 aligned-new is supported
         $CXX $PLATFORM_CXXFLAGS -faligned-new -x c++ - -o /dev/null 2>/dev/null <<EOF
@@ -681,6 +712,7 @@ echo "PLATFORM=$PLATFORM" >> "$OUTPUT"
 echo "PLATFORM_LDFLAGS=$PLATFORM_LDFLAGS" >> "$OUTPUT"
 echo "JAVA_LDFLAGS=$JAVA_LDFLAGS" >> "$OUTPUT"
 echo "JAVA_STATIC_LDFLAGS=$JAVA_STATIC_LDFLAGS" >> "$OUTPUT"
+echo "JAVAC_ARGS=$JAVAC_ARGS" >> "$OUTPUT"
 echo "VALGRIND_VER=$VALGRIND_VER" >> "$OUTPUT"
 echo "PLATFORM_CCFLAGS=$PLATFORM_CCFLAGS" >> "$OUTPUT"
 echo "PLATFORM_CXXFLAGS=$PLATFORM_CXXFLAGS" >> "$OUTPUT"

build_tools/dependencies_platform007.sh

@@ -13,6 +13,7 @@ JEMALLOC_BASE=/mnt/gvfs/third-party2/jemalloc/c26f08f47ac35fc31da2633b7da92d6b86
 NUMA_BASE=/mnt/gvfs/third-party2/numa/3f3fb57a5ccc5fd21c66416c0b83e0aa76a05376/2.0.11/platform007/ca4da3d
 LIBUNWIND_BASE=/mnt/gvfs/third-party2/libunwind/40c73d874898b386a71847f1b99115d93822d11f/1.4/platform007/6f3e0a9
 TBB_BASE=/mnt/gvfs/third-party2/tbb/4ce8e8dba77cdbd81b75d6f0c32fd7a1b76a11ec/2018_U5/platform007/ca4da3d
+LIBURING_BASE=/mnt/gvfs/third-party2/liburing/79427253fd0d42677255aacfe6d13bfe63f752eb/20190828/platform007/ca4da3d
 KERNEL_HEADERS_BASE=/mnt/gvfs/third-party2/kernel-headers/fb251ecd2f5ae16f8671f7014c246e52a748fe0b/fb/platform007/da39a3e
 BINUTILS_BASE=/mnt/gvfs/third-party2/binutils/ab9f09bba370e7066cafd4eb59752db93f2e8312/2.29.1/platform007/15a3614
 VALGRIND_BASE=/mnt/gvfs/third-party2/valgrind/d42d152a15636529b0861ec493927200ebebca8e/3.15.0/platform007/ca4da3d

build_tools/error_filter.py

@@ -133,13 +133,16 @@ _TEST_NAME_TO_PARSERS = {
     'lite_test': [CompilerErrorParser, GTestErrorParser],
     'stress_crash': [CompilerErrorParser, DbCrashErrorParser],
     'stress_crash_with_atomic_flush': [CompilerErrorParser, DbCrashErrorParser],
+    'stress_crash_with_txn': [CompilerErrorParser, DbCrashErrorParser],
     'write_stress': [CompilerErrorParser, WriteStressErrorParser],
     'asan': [CompilerErrorParser, GTestErrorParser, AsanErrorParser],
     'asan_crash': [CompilerErrorParser, AsanErrorParser, DbCrashErrorParser],
     'asan_crash_with_atomic_flush': [CompilerErrorParser, AsanErrorParser, DbCrashErrorParser],
+    'asan_crash_with_txn': [CompilerErrorParser, AsanErrorParser, DbCrashErrorParser],
     'ubsan': [CompilerErrorParser, GTestErrorParser, UbsanErrorParser],
     'ubsan_crash': [CompilerErrorParser, UbsanErrorParser, DbCrashErrorParser],
     'ubsan_crash_with_atomic_flush': [CompilerErrorParser, UbsanErrorParser, DbCrashErrorParser],
+    'ubsan_crash_with_txn': [CompilerErrorParser, UbsanErrorParser, DbCrashErrorParser],
     'valgrind': [CompilerErrorParser, GTestErrorParser, ValgrindErrorParser],
     'tsan': [CompilerErrorParser, GTestErrorParser, TsanErrorParser],
     'format_compatible': [CompilerErrorParser, CompatErrorParser],

build_tools/fbcode_config_platform007.sh

@@ -86,6 +86,15 @@ else
 fi
 CFLAGS+=" -DTBB"
 
+# location of LIBURING
+LIBURING_INCLUDE=" -isystem $LIBURING_BASE/include/"
+if test -z $PIC_BUILD; then
+  LIBURING_LIBS="$LIBURING_BASE/lib/liburing.a"
+else
+  LIBURING_LIBS="$LIBURING_BASE/lib/liburing_pic.a"
+fi
+CFLAGS+=" -DLIBURING"
+
 test "$USE_SSE" || USE_SSE=1
 export USE_SSE
 test "$PORTABLE" || PORTABLE=1
@@ -94,7 +103,7 @@ export PORTABLE
 BINUTILS="$BINUTILS_BASE/bin"
 AR="$BINUTILS/ar"
 
-DEPS_INCLUDE="$SNAPPY_INCLUDE $ZLIB_INCLUDE $BZIP_INCLUDE $LZ4_INCLUDE $ZSTD_INCLUDE $GFLAGS_INCLUDE $NUMA_INCLUDE $TBB_INCLUDE"
+DEPS_INCLUDE="$SNAPPY_INCLUDE $ZLIB_INCLUDE $BZIP_INCLUDE $LZ4_INCLUDE $ZSTD_INCLUDE $GFLAGS_INCLUDE $NUMA_INCLUDE $TBB_INCLUDE $LIBURING_INCLUDE"
 
 STDLIBS="-L $GCC_BASE/lib64"
 
@@ -135,10 +144,10 @@ else
 fi
 
 CFLAGS+=" $DEPS_INCLUDE"
-CFLAGS+=" -DROCKSDB_PLATFORM_POSIX -DROCKSDB_LIB_IO_POSIX -DROCKSDB_FALLOCATE_PRESENT -DROCKSDB_MALLOC_USABLE_SIZE -DROCKSDB_RANGESYNC_PRESENT -DROCKSDB_SCHED_GETCPU_PRESENT -DROCKSDB_SUPPORT_THREAD_LOCAL -DHAVE_SSE42"
+CFLAGS+=" -DROCKSDB_PLATFORM_POSIX -DROCKSDB_LIB_IO_POSIX -DROCKSDB_FALLOCATE_PRESENT -DROCKSDB_MALLOC_USABLE_SIZE -DROCKSDB_RANGESYNC_PRESENT -DROCKSDB_SCHED_GETCPU_PRESENT -DROCKSDB_SUPPORT_THREAD_LOCAL -DHAVE_SSE42 -DROCKSDB_IOURING_PRESENT"
 CXXFLAGS+=" $CFLAGS"
 
-EXEC_LDFLAGS=" $SNAPPY_LIBS $ZLIB_LIBS $BZIP_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS $NUMA_LIB $TBB_LIBS"
+EXEC_LDFLAGS=" $SNAPPY_LIBS $ZLIB_LIBS $BZIP_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS $NUMA_LIB $TBB_LIBS $LIBURING_LIBS"
 EXEC_LDFLAGS+=" -B$BINUTILS/gold"
 EXEC_LDFLAGS+=" -Wl,--dynamic-linker,/usr/local/fbcode/platform007/lib/ld.so"
 EXEC_LDFLAGS+=" $LIBUNWIND"
@@ -148,7 +157,7 @@ EXEC_LDFLAGS+=" -ldl"
 
 PLATFORM_LDFLAGS="$LIBGCC_LIBS $GLIBC_LIBS $STDLIBS -lgcc -lstdc++"
 
-EXEC_LDFLAGS_SHARED="$SNAPPY_LIBS $ZLIB_LIBS $BZIP_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS $TBB_LIBS"
+EXEC_LDFLAGS_SHARED="$SNAPPY_LIBS $ZLIB_LIBS $BZIP_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS $TBB_LIBS $LIBURING_LIBS"
 
 VALGRIND_VER="$VALGRIND_BASE/bin/"
 

build_tools/format-diff.sh

@@ -13,9 +13,14 @@ then
   echo "You didn't have clang-format-diff.py and/or clang-format available in your computer!"
   echo "You can download clang-format-diff.py by running: "
   echo "    curl --location http://goo.gl/iUW1u2 -o ${CLANG_FORMAT_DIFF}"
-  echo "You can download clang-format by running: "
+  echo "You can download clang-format by running:"
   echo "    brew install clang-format"
+  echo "  Or"
+  echo "    apt install clang-format"
+  echo "  This might work too:"
+  echo "    yum install git-clang-format"
   echo "Then, move both files (i.e. ${CLANG_FORMAT_DIFF} and clang-format) to some directory within PATH=${PATH}"
+  echo "and make sure ${CLANG_FORMAT_DIFF} is executable."
   exit 128
 fi
 

build_tools/gnu_parallel

@@ -5801,7 +5801,7 @@ sub workdir {
 		    . "-" . $self->seq();
 	    } else {
 		$workdir = $opt::workdir;
-		# Rsync treats /./ special. We dont want that
+		# Rsync treats /./ special. We don't want that
 		$workdir =~ s:/\./:/:g; # Remove /./
 		$workdir =~ s:/+$::; # Remove ending / if any
 		$workdir =~ s:^\./::g; # Remove starting ./ if any

build_tools/precommit_checker.py

@@ -1,4 +1,4 @@
-#!/usr/local/fbcode/gcc-4.9-glibc-2.20-fb/bin/python2.7
+#!/usr/bin/env python2.7
 # Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
 
 from __future__ import absolute_import

build_tools/regression_build_test.sh

@@ -378,7 +378,7 @@ function send_to_ods {
     echo >&2 "ERROR: Key $key doesn't have a value."
     return
   fi
-  curl -s "https://www.intern.facebook.com/intern/agent/ods_set.php?entity=rocksdb_build$git_br&key=$key&value=$value" \
+  curl --silent "https://www.intern.facebook.com/intern/agent/ods_set.php?entity=rocksdb_build$git_br&key=$key&value=$value" \
     --connect-timeout 60
 }
 

build_tools/rocksdb-lego-determinator

@@ -414,7 +414,8 @@ STRESS_CRASH_TEST_COMMANDS="[
                 'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 crash_test || $CONTRUN_NAME=crash_test $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
-            }
+            },
+            $UPLOAD_DB_DIR,
         ],
         $REPORT
     }
@@ -450,6 +451,36 @@ STRESS_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS="[
     }
 ]"
 
+#
+# RocksDB stress/crash test with txn
+#
+STRESS_CRASH_TEST_WITH_TXN_COMMANDS="[
+    {
+        'name':'Rocksdb Stress and Crash Test with txn',
+        'oncall':'$ONCALL',
+        'executeLocal': 'true',
+        'timeout': 86400,
+        'steps': [
+            $CLEANUP_ENV,
+            {
+                'name':'Build and run RocksDB debug stress tests',
+                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 db_stress || $CONTRUN_NAME=db_stress $TASK_CREATION_TOOL',
+                'user':'root',
+                $PARSER
+            },
+            {
+                'name':'Build and run RocksDB debug crash tests with txn',
+                'timeout': 86400,
+                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 crash_test_with_txn || $CONTRUN_NAME=crash_test_with_txn $TASK_CREATION_TOOL',
+                'user':'root',
+                $PARSER
+            },
+            $UPLOAD_DB_DIR,
+        ],
+        $REPORT
+    }
+]"
+
 # RocksDB write stress test.
 # We run on disk device on purpose (i.e. no $SHM)
 # because we want to add some randomness to fsync commands
@@ -512,6 +543,7 @@ ASAN_CRASH_TEST_COMMANDS="[
                 'user':'root',
                 $PARSER
             },
+            $UPLOAD_DB_DIR,
         ],
         $REPORT
     }
@@ -541,6 +573,30 @@ ASAN_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS="[
     }
 ]"
 
+#
+# RocksDB crash testing with txn under address sanitizer
+#
+ASAN_CRASH_TEST_WITH_TXN_COMMANDS="[
+    {
+        'name':'Rocksdb crash test with txn under ASAN',
+        'oncall':'$ONCALL',
+        'executeLocal': 'true',
+        'timeout': 86400,
+        'steps': [
+            $CLEANUP_ENV,
+            {
+                'name':'Build and run RocksDB debug asan_crash_test_with_txn',
+                'timeout': 86400,
+                'shell':'$SHM $DEBUG $NON_TSAN_CRASH make J=1 asan_crash_test_with_txn || $CONTRUN_NAME=asan_crash_test_with_txn $TASK_CREATION_TOOL',
+                'user':'root',
+                $PARSER
+            },
+            $UPLOAD_DB_DIR,
+        ],
+        $REPORT
+    }
+]"
+
 #
 # RocksDB test under undefined behavior sanitizer
 #
@@ -580,6 +636,7 @@ UBSAN_CRASH_TEST_COMMANDS="[
                 'user':'root',
                 $PARSER
             },
+            $UPLOAD_DB_DIR,
         ],
         $REPORT
     }
@@ -609,6 +666,30 @@ UBSAN_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS="[
     }
 ]"
 
+#
+# RocksDB crash testing with txn under undefined behavior sanitizer
+#
+UBSAN_CRASH_TEST_WITH_TXN_COMMANDS="[
+    {
+        'name':'Rocksdb crash test with txn under UBSAN',
+        'oncall':'$ONCALL',
+        'executeLocal': 'true',
+        'timeout': 86400,
+        'steps': [
+            $CLEANUP_ENV,
+            {
+                'name':'Build and run RocksDB debug ubsan_crash_test_with_txn',
+                'timeout': 86400,
+                'shell':'$SHM $DEBUG $NON_TSAN_CRASH $CLANG make J=1 ubsan_crash_test_with_txn || $CONTRUN_NAME=ubsan_crash_test_with_txn $TASK_CREATION_TOOL',
+                'user':'root',
+                $PARSER
+            },
+            $UPLOAD_DB_DIR,
+        ],
+        $REPORT
+    }
+]"
+
 #
 # RocksDB unit test under valgrind
 #
@@ -673,6 +754,7 @@ TSAN_CRASH_TEST_COMMANDS="[
                 'user':'root',
                 $PARSER
             },
+            $UPLOAD_DB_DIR,
         ],
         $REPORT
     }
@@ -702,6 +784,30 @@ TSAN_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS="[
     }
 ]"
 
+#
+# RocksDB crash test with txn under TSAN
+#
+TSAN_CRASH_TEST_WITH_TXN_COMMANDS="[
+    {
+        'name':'Rocksdb Crash Test with txn under TSAN',
+        'oncall':'$ONCALL',
+        'executeLocal': 'true',
+        'timeout': 86400,
+        'steps': [
+            $CLEANUP_ENV,
+            {
+                'name':'Compile and run',
+                'timeout': 86400,
+                'shell':'set -o pipefail && $SHM $DEBUG $TSAN $TSAN_CRASH CRASH_TEST_KILL_ODD=1887 make J=1 crash_test_with_txn || $CONTRUN_NAME=tsan_crash_test_with_txn $TASK_CREATION_TOOL',
+                'user':'root',
+                $PARSER
+            },
+            $UPLOAD_DB_DIR,
+        ],
+        $REPORT
+    }
+]"
+
 #
 # RocksDB format compatible
 #
@@ -778,7 +884,7 @@ run_regression()
 
   # parameters: $1 -- key, $2 -- value
   function send_size_to_ods {
-    curl -s "https://www.intern.facebook.com/intern/agent/ods_set.php?entity=rocksdb_build&key=rocksdb.build_size.$1&value=$2" \
+    curl --silent "https://www.intern.facebook.com/intern/agent/ods_set.php?entity=rocksdb_build&key=rocksdb.build_size.$1&value=$2" \
       --connect-timeout 60
   }
 
@@ -889,6 +995,9 @@ case $1 in
   stress_crash_with_atomic_flush)
     echo $STRESS_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS
     ;;
+  stress_crash_with_txn)
+    echo $STRESS_CRASH_TEST_WITH_TXN_COMMANDS
+    ;;
   write_stress)
     echo $WRITE_STRESS_COMMANDS
     ;;
@@ -901,6 +1010,9 @@ case $1 in
   asan_crash_with_atomic_flush)
     echo $ASAN_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS
     ;;
+  asan_crash_with_txn)
+    echo $ASAN_CRASH_TEST_WITH_TXN_COMMANDS
+    ;;
   ubsan)
     echo $UBSAN_TEST_COMMANDS
     ;;
@@ -910,6 +1022,9 @@ case $1 in
   ubsan_crash_with_atomic_flush)
     echo $UBSAN_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS
     ;;
+  ubsan_crash_with_txn)
+    echo $UBSAN_CRASH_TEST_WITH_TXN_COMMANDS
+    ;;
   valgrind)
     echo $VALGRIND_TEST_COMMANDS
     ;;
@@ -922,6 +1037,9 @@ case $1 in
   tsan_crash_with_atomic_flush)
     echo $TSAN_CRASH_TEST_WITH_ATOMIC_FLUSH_COMMANDS
     ;;
+  tsan_crash_with_txn)
+    echo $TSAN_CRASH_TEST_WITH_TXN_COMMANDS
+    ;;
   format_compatible)
     echo $FORMAT_COMPATIBLE_COMMANDS
     ;;
@@ -945,5 +1063,6 @@ case $1 in
     ;;
   *)
     echo "Invalid determinator command"
+    exit 1
     ;;
 esac

build_tools/update_dependencies.sh

@@ -92,6 +92,7 @@ get_lib_base jemalloc   LATEST  platform007
 get_lib_base numa       LATEST  platform007
 get_lib_base libunwind  LATEST  platform007
 get_lib_base tbb        LATEST  platform007
+get_lib_base liburing   LATEST  platform007
 
 get_lib_base kernel-headers fb platform007
 get_lib_base binutils   LATEST centos7-native

cache/cache_bench.cc

@@ -45,7 +45,7 @@ DEFINE_int32(erase_percent, 10,
 
 DEFINE_bool(use_clock_cache, false, "");
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class CacheBench;
 namespace {
@@ -154,7 +154,7 @@ class CacheBench {
   }
 
   bool Run() {
-    rocksdb::Env* env = rocksdb::Env::Default();
+    ROCKSDB_NAMESPACE::Env* env = ROCKSDB_NAMESPACE::Env::Default();
 
     PrintEnv();
     SharedState shared(this);
@@ -257,7 +257,7 @@ class CacheBench {
     printf("----------------------------\n");
   }
 };
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
   ParseCommandLineFlags(&argc, &argv, true);
@@ -267,7 +267,7 @@ int main(int argc, char** argv) {
     exit(1);
   }
 
-  rocksdb::CacheBench bench;
+  ROCKSDB_NAMESPACE::CacheBench bench;
   if (FLAGS_populate_cache) {
     bench.PopulateCache();
   }

cache/cache_test.cc

@@ -20,7 +20,7 @@
 #include "util/coding.h"
 #include "util/string_util.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 // Conversions between numeric keys/values and the types expected by Cache.
 static std::string EncodeKey(int k) {
@@ -765,7 +765,7 @@ INSTANTIATE_TEST_CASE_P(CacheTestInstance, CacheTest, testing::Values(kLRU));
 #endif  // SUPPORT_CLOCK_CACHE
 INSTANTIATE_TEST_CASE_P(CacheTestInstance, LRUCacheTest, testing::Values(kLRU));
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
   ::testing::InitGoogleTest(&argc, argv);

cache/clock_cache.cc

@@ -11,7 +11,7 @@
 
 #ifndef SUPPORT_CLOCK_CACHE
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 std::shared_ptr<Cache> NewClockCache(
     size_t /*capacity*/, int /*num_shard_bits*/, bool /*strict_capacity_limit*/,
@@ -20,7 +20,7 @@ std::shared_ptr<Cache> NewClockCache(
   return nullptr;
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 #else
 
@@ -41,7 +41,7 @@ std::shared_ptr<Cache> NewClockCache(
 #include "util/autovector.h"
 #include "util/mutexlock.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 namespace {
 
@@ -756,6 +756,6 @@ std::shared_ptr<Cache> NewClockCache(
       capacity, num_shard_bits, strict_capacity_limit, metadata_charge_policy);
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 #endif  // SUPPORT_CLOCK_CACHE

cache/lru_cache.cc

@@ -16,7 +16,7 @@
 
 #include "util/mutexlock.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 LRUHandleTable::LRUHandleTable() : list_(nullptr), length_(0), elems_(0) {
   Resize();
@@ -571,4 +571,4 @@ std::shared_ptr<Cache> NewLRUCache(
       std::move(memory_allocator), use_adaptive_mutex, metadata_charge_policy);
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

cache/lru_cache.h

@@ -16,7 +16,7 @@
 #include "port/port.h"
 #include "util/autovector.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 // LRU cache implementation. This class is not thread-safe.
 
@@ -133,7 +133,6 @@ struct LRUHandle {
   // Caclculate the memory usage by metadata
   inline size_t CalcTotalCharge(
       CacheMetadataChargePolicy metadata_charge_policy) {
-    assert(key_length);
     size_t meta_charge = 0;
     if (metadata_charge_policy == kFullChargeCacheMetadata) {
 #ifdef ROCKSDB_MALLOC_USABLE_SIZE
@@ -337,4 +336,4 @@ class LRUCache
   int num_shards_ = 0;
 };
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

cache/lru_cache_test.cc

@@ -10,7 +10,7 @@
 #include "port/port.h"
 #include "test_util/testharness.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class LRUCacheTest : public testing::Test {
  public:
@@ -190,7 +190,7 @@ TEST_F(LRUCacheTest, EntriesWithPriority) {
   ValidateLRUList({"e", "f", "g", "Z", "d"}, 2);
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
   ::testing::InitGoogleTest(&argc, argv);

cache/sharded_cache.cc

@@ -13,7 +13,7 @@
 
 #include "util/mutexlock.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 ShardedCache::ShardedCache(size_t capacity, int num_shard_bits,
                            bool strict_capacity_limit,
@@ -159,4 +159,4 @@ int GetDefaultCacheShardBits(size_t capacity) {
   return num_shard_bits;
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

cache/sharded_cache.h

@@ -16,7 +16,7 @@
 #include "rocksdb/cache.h"
 #include "util/hash.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 // Single cache shard interface.
 class CacheShard {
@@ -108,4 +108,4 @@ class ShardedCache : public Cache {
 
 extern int GetDefaultCacheShardBits(size_t capacity);
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

cmake/modules/Findgflags.cmake

@@ -0,0 +1,29 @@
+# - Find gflags library
+# Find the gflags includes and library
+#
+# gflags_INCLUDE_DIR - where to find gflags.h.
+# gflags_LIBRARIES - List of libraries when using gflags.
+# gflags_FOUND - True if gflags found.
+
+find_path(GFLAGS_INCLUDE_DIR
+  NAMES gflags/gflags.h)
+
+find_library(GFLAGS_LIBRARIES
+  NAMES gflags)
+
+include(FindPackageHandleStandardArgs)
+find_package_handle_standard_args(gflags
+  DEFAULT_MSG GFLAGS_LIBRARIES GFLAGS_INCLUDE_DIR)
+
+mark_as_advanced(
+  GFLAGS_LIBRARIES
+  GFLAGS_INCLUDE_DIR)
+
+if(gflags_FOUND AND NOT (TARGET gflags::gflags))
+  add_library(gflags::gflags UNKNOWN IMPORTED)
+  set_target_properties(gflags::gflags
+    PROPERTIES
+      IMPORTED_LOCATION ${GFLAGS_LIBRARIES}
+      INTERFACE_INCLUDE_DIRECTORIES ${GFLAGS_INCLUDE_DIR}
+      IMPORTED_LINK_INTERFACE_LANGUAGES "CXX")
+endif()

db/arena_wrapped_db_iter.cc

@@ -16,7 +16,7 @@
 #include "table/iterator_wrapper.h"
 #include "util/user_comparator_wrapper.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 Status ArenaWrappedDBIter::GetProperty(std::string prop_name,
                                        std::string* prop) {
@@ -64,7 +64,7 @@ Status ArenaWrappedDBIter::Refresh() {
     arena_.~Arena();
     new (&arena_) Arena();
 
-    SuperVersion* sv = cfd_->GetReferencedSuperVersion(db_impl_->mutex());
+    SuperVersion* sv = cfd_->GetReferencedSuperVersion(db_impl_);
     if (read_callback_) {
       read_callback_->Refresh(latest_seq);
     }
@@ -103,4 +103,4 @@ ArenaWrappedDBIter* NewArenaWrappedDbIterator(
   return iter;
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/arena_wrapped_db_iter.h

@@ -20,7 +20,7 @@
 #include "rocksdb/iterator.h"
 #include "util/autovector.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class Arena;
 
@@ -45,26 +45,29 @@ class ArenaWrappedDBIter : public Iterator {
   // Set the internal iterator wrapped inside the DB Iterator. Usually it is
   // a merging iterator.
   virtual void SetIterUnderDBIter(InternalIterator* iter) {
-    static_cast<DBIter*>(db_iter_)->SetIter(iter);
+    db_iter_->SetIter(iter);
   }
 
-  virtual bool Valid() const override { return db_iter_->Valid(); }
-  virtual void SeekToFirst() override { db_iter_->SeekToFirst(); }
-  virtual void SeekToLast() override { db_iter_->SeekToLast(); }
-  virtual void Seek(const Slice& target) override { db_iter_->Seek(target); }
-  virtual void SeekForPrev(const Slice& target) override {
+  bool Valid() const override { return db_iter_->Valid(); }
+  void SeekToFirst() override { db_iter_->SeekToFirst(); }
+  void SeekToLast() override { db_iter_->SeekToLast(); }
+  // 'target' does not contain timestamp, even if user timestamp feature is
+  // enabled.
+  void Seek(const Slice& target) override { db_iter_->Seek(target); }
+  void SeekForPrev(const Slice& target) override {
     db_iter_->SeekForPrev(target);
   }
-  virtual void Next() override { db_iter_->Next(); }
-  virtual void Prev() override { db_iter_->Prev(); }
-  virtual Slice key() const override { return db_iter_->key(); }
-  virtual Slice value() const override { return db_iter_->value(); }
-  virtual Status status() const override { return db_iter_->status(); }
+  void Next() override { db_iter_->Next(); }
+  void Prev() override { db_iter_->Prev(); }
+  Slice key() const override { return db_iter_->key(); }
+  Slice value() const override { return db_iter_->value(); }
+  Status status() const override { return db_iter_->status(); }
+  Slice timestamp() const override { return db_iter_->timestamp(); }
   bool IsBlob() const { return db_iter_->IsBlob(); }
 
-  virtual Status GetProperty(std::string prop_name, std::string* prop) override;
+  Status GetProperty(std::string prop_name, std::string* prop) override;
 
-  virtual Status Refresh() override;
+  Status Refresh() override;
 
   void Init(Env* env, const ReadOptions& read_options,
             const ImmutableCFOptions& cf_options,
@@ -109,4 +112,4 @@ extern ArenaWrappedDBIter* NewArenaWrappedDbIterator(
     ReadCallback* read_callback, DBImpl* db_impl = nullptr,
     ColumnFamilyData* cfd = nullptr, bool allow_blob = false,
     bool allow_refresh = true);
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/blob/blob_constants.h

@@ -0,0 +1,16 @@
+//  Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#pragma once
+
+#include <cstdint>
+
+#include "rocksdb/rocksdb_namespace.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+constexpr uint64_t kInvalidBlobFileNumber = 0;
+
+}  // namespace ROCKSDB_NAMESPACE

db/blob/blob_file_addition.cc

@@ -0,0 +1,154 @@
+//  Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#include "db/blob/blob_file_addition.h"
+
+#include <ostream>
+#include <sstream>
+
+#include "logging/event_logger.h"
+#include "rocksdb/slice.h"
+#include "rocksdb/status.h"
+#include "test_util/sync_point.h"
+#include "util/coding.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+// Tags for custom fields. Note that these get persisted in the manifest,
+// so existing tags should not be modified.
+enum BlobFileAddition::CustomFieldTags : uint32_t {
+  kEndMarker,
+
+  // Add forward compatible fields here
+
+  /////////////////////////////////////////////////////////////////////
+
+  kForwardIncompatibleMask = 1 << 6,
+
+  // Add forward incompatible fields here
+};
+
+void BlobFileAddition::EncodeTo(std::string* output) const {
+  PutVarint64(output, blob_file_number_);
+  PutVarint64(output, total_blob_count_);
+  PutVarint64(output, total_blob_bytes_);
+  PutLengthPrefixedSlice(output, checksum_method_);
+  PutLengthPrefixedSlice(output, checksum_value_);
+
+  // Encode any custom fields here. The format to use is a Varint32 tag (see
+  // CustomFieldTags above) followed by a length prefixed slice. Unknown custom
+  // fields will be ignored during decoding unless they're in the forward
+  // incompatible range.
+
+  TEST_SYNC_POINT_CALLBACK("BlobFileAddition::EncodeTo::CustomFields", output);
+
+  PutVarint32(output, kEndMarker);
+}
+
+Status BlobFileAddition::DecodeFrom(Slice* input) {
+  constexpr char class_name[] = "BlobFileAddition";
+
+  if (!GetVarint64(input, &blob_file_number_)) {
+    return Status::Corruption(class_name, "Error decoding blob file number");
+  }
+
+  if (!GetVarint64(input, &total_blob_count_)) {
+    return Status::Corruption(class_name, "Error decoding total blob count");
+  }
+
+  if (!GetVarint64(input, &total_blob_bytes_)) {
+    return Status::Corruption(class_name, "Error decoding total blob bytes");
+  }
+
+  Slice checksum_method;
+  if (!GetLengthPrefixedSlice(input, &checksum_method)) {
+    return Status::Corruption(class_name, "Error decoding checksum method");
+  }
+  checksum_method_ = checksum_method.ToString();
+
+  Slice checksum_value;
+  if (!GetLengthPrefixedSlice(input, &checksum_value)) {
+    return Status::Corruption(class_name, "Error decoding checksum value");
+  }
+  checksum_value_ = checksum_value.ToString();
+
+  while (true) {
+    uint32_t custom_field_tag = 0;
+    if (!GetVarint32(input, &custom_field_tag)) {
+      return Status::Corruption(class_name, "Error decoding custom field tag");
+    }
+
+    if (custom_field_tag == kEndMarker) {
+      break;
+    }
+
+    if (custom_field_tag & kForwardIncompatibleMask) {
+      return Status::Corruption(
+          class_name, "Forward incompatible custom field encountered");
+    }
+
+    Slice custom_field_value;
+    if (!GetLengthPrefixedSlice(input, &custom_field_value)) {
+      return Status::Corruption(class_name,
+                                "Error decoding custom field value");
+    }
+  }
+
+  return Status::OK();
+}
+
+std::string BlobFileAddition::DebugString() const {
+  std::ostringstream oss;
+
+  oss << *this;
+
+  return oss.str();
+}
+
+std::string BlobFileAddition::DebugJSON() const {
+  JSONWriter jw;
+
+  jw << *this;
+
+  jw.EndObject();
+
+  return jw.Get();
+}
+
+bool operator==(const BlobFileAddition& lhs, const BlobFileAddition& rhs) {
+  return lhs.GetBlobFileNumber() == rhs.GetBlobFileNumber() &&
+         lhs.GetTotalBlobCount() == rhs.GetTotalBlobCount() &&
+         lhs.GetTotalBlobBytes() == rhs.GetTotalBlobBytes() &&
+         lhs.GetChecksumMethod() == rhs.GetChecksumMethod() &&
+         lhs.GetChecksumValue() == rhs.GetChecksumValue();
+}
+
+bool operator!=(const BlobFileAddition& lhs, const BlobFileAddition& rhs) {
+  return !(lhs == rhs);
+}
+
+std::ostream& operator<<(std::ostream& os,
+                         const BlobFileAddition& blob_file_addition) {
+  os << "blob_file_number: " << blob_file_addition.GetBlobFileNumber()
+     << " total_blob_count: " << blob_file_addition.GetTotalBlobCount()
+     << " total_blob_bytes: " << blob_file_addition.GetTotalBlobBytes()
+     << " checksum_method: " << blob_file_addition.GetChecksumMethod()
+     << " checksum_value: " << blob_file_addition.GetChecksumValue();
+
+  return os;
+}
+
+JSONWriter& operator<<(JSONWriter& jw,
+                       const BlobFileAddition& blob_file_addition) {
+  jw << "BlobFileNumber" << blob_file_addition.GetBlobFileNumber()
+     << "TotalBlobCount" << blob_file_addition.GetTotalBlobCount()
+     << "TotalBlobBytes" << blob_file_addition.GetTotalBlobBytes()
+     << "ChecksumMethod" << blob_file_addition.GetChecksumMethod()
+     << "ChecksumValue" << blob_file_addition.GetChecksumValue();
+
+  return jw;
+}
+
+}  // namespace ROCKSDB_NAMESPACE

db/blob/blob_file_addition.h

@@ -0,0 +1,67 @@
+//  Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#pragma once
+
+#include <cassert>
+#include <cstdint>
+#include <iosfwd>
+#include <string>
+
+#include "db/blob/blob_constants.h"
+#include "rocksdb/rocksdb_namespace.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+class JSONWriter;
+class Slice;
+class Status;
+
+class BlobFileAddition {
+ public:
+  BlobFileAddition() = default;
+
+  BlobFileAddition(uint64_t blob_file_number, uint64_t total_blob_count,
+                   uint64_t total_blob_bytes, std::string checksum_method,
+                   std::string checksum_value)
+      : blob_file_number_(blob_file_number),
+        total_blob_count_(total_blob_count),
+        total_blob_bytes_(total_blob_bytes),
+        checksum_method_(std::move(checksum_method)),
+        checksum_value_(std::move(checksum_value)) {
+    assert(checksum_method_.empty() == checksum_value_.empty());
+  }
+
+  uint64_t GetBlobFileNumber() const { return blob_file_number_; }
+  uint64_t GetTotalBlobCount() const { return total_blob_count_; }
+  uint64_t GetTotalBlobBytes() const { return total_blob_bytes_; }
+  const std::string& GetChecksumMethod() const { return checksum_method_; }
+  const std::string& GetChecksumValue() const { return checksum_value_; }
+
+  void EncodeTo(std::string* output) const;
+  Status DecodeFrom(Slice* input);
+
+  std::string DebugString() const;
+  std::string DebugJSON() const;
+
+ private:
+  enum CustomFieldTags : uint32_t;
+
+  uint64_t blob_file_number_ = kInvalidBlobFileNumber;
+  uint64_t total_blob_count_ = 0;
+  uint64_t total_blob_bytes_ = 0;
+  std::string checksum_method_;
+  std::string checksum_value_;
+};
+
+bool operator==(const BlobFileAddition& lhs, const BlobFileAddition& rhs);
+bool operator!=(const BlobFileAddition& lhs, const BlobFileAddition& rhs);
+
+std::ostream& operator<<(std::ostream& os,
+                         const BlobFileAddition& blob_file_addition);
+JSONWriter& operator<<(JSONWriter& jw,
+                       const BlobFileAddition& blob_file_addition);
+
+}  // namespace ROCKSDB_NAMESPACE

db/blob/blob_file_addition_test.cc

@@ -0,0 +1,206 @@
+//  Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#include "db/blob/blob_file_addition.h"
+
+#include <cstdint>
+#include <cstring>
+#include <string>
+
+#include "test_util/sync_point.h"
+#include "test_util/testharness.h"
+#include "util/coding.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+class BlobFileAdditionTest : public testing::Test {
+ public:
+  static void TestEncodeDecode(const BlobFileAddition& blob_file_addition) {
+    std::string encoded;
+    blob_file_addition.EncodeTo(&encoded);
+
+    BlobFileAddition decoded;
+    Slice input(encoded);
+    ASSERT_OK(decoded.DecodeFrom(&input));
+
+    ASSERT_EQ(blob_file_addition, decoded);
+  }
+};
+
+TEST_F(BlobFileAdditionTest, Empty) {
+  BlobFileAddition blob_file_addition;
+
+  ASSERT_EQ(blob_file_addition.GetBlobFileNumber(), kInvalidBlobFileNumber);
+  ASSERT_EQ(blob_file_addition.GetTotalBlobCount(), 0);
+  ASSERT_EQ(blob_file_addition.GetTotalBlobBytes(), 0);
+  ASSERT_TRUE(blob_file_addition.GetChecksumMethod().empty());
+  ASSERT_TRUE(blob_file_addition.GetChecksumValue().empty());
+
+  TestEncodeDecode(blob_file_addition);
+}
+
+TEST_F(BlobFileAdditionTest, NonEmpty) {
+  constexpr uint64_t blob_file_number = 123;
+  constexpr uint64_t total_blob_count = 2;
+  constexpr uint64_t total_blob_bytes = 123456;
+  const std::string checksum_method("SHA1");
+  const std::string checksum_value("bdb7f34a59dfa1592ce7f52e99f98c570c525cbd");
+
+  BlobFileAddition blob_file_addition(blob_file_number, total_blob_count,
+                                      total_blob_bytes, checksum_method,
+                                      checksum_value);
+
+  ASSERT_EQ(blob_file_addition.GetBlobFileNumber(), blob_file_number);
+  ASSERT_EQ(blob_file_addition.GetTotalBlobCount(), total_blob_count);
+  ASSERT_EQ(blob_file_addition.GetTotalBlobBytes(), total_blob_bytes);
+  ASSERT_EQ(blob_file_addition.GetChecksumMethod(), checksum_method);
+  ASSERT_EQ(blob_file_addition.GetChecksumValue(), checksum_value);
+
+  TestEncodeDecode(blob_file_addition);
+}
+
+TEST_F(BlobFileAdditionTest, DecodeErrors) {
+  std::string str;
+  Slice slice(str);
+
+  BlobFileAddition blob_file_addition;
+
+  {
+    const Status s = blob_file_addition.DecodeFrom(&slice);
+    ASSERT_TRUE(s.IsCorruption());
+    ASSERT_TRUE(std::strstr(s.getState(), "blob file number"));
+  }
+
+  constexpr uint64_t blob_file_number = 123;
+  PutVarint64(&str, blob_file_number);
+  slice = str;
+
+  {
+    const Status s = blob_file_addition.DecodeFrom(&slice);
+    ASSERT_TRUE(s.IsCorruption());
+    ASSERT_TRUE(std::strstr(s.getState(), "total blob count"));
+  }
+
+  constexpr uint64_t total_blob_count = 4567;
+  PutVarint64(&str, total_blob_count);
+  slice = str;
+
+  {
+    const Status s = blob_file_addition.DecodeFrom(&slice);
+    ASSERT_TRUE(s.IsCorruption());
+    ASSERT_TRUE(std::strstr(s.getState(), "total blob bytes"));
+  }
+
+  constexpr uint64_t total_blob_bytes = 12345678;
+  PutVarint64(&str, total_blob_bytes);
+  slice = str;
+
+  {
+    const Status s = blob_file_addition.DecodeFrom(&slice);
+    ASSERT_TRUE(s.IsCorruption());
+    ASSERT_TRUE(std::strstr(s.getState(), "checksum method"));
+  }
+
+  constexpr char checksum_method[] = "SHA1";
+  PutLengthPrefixedSlice(&str, checksum_method);
+  slice = str;
+
+  {
+    const Status s = blob_file_addition.DecodeFrom(&slice);
+    ASSERT_TRUE(s.IsCorruption());
+    ASSERT_TRUE(std::strstr(s.getState(), "checksum value"));
+  }
+
+  constexpr char checksum_value[] = "bdb7f34a59dfa1592ce7f52e99f98c570c525cbd";
+  PutLengthPrefixedSlice(&str, checksum_value);
+  slice = str;
+
+  {
+    const Status s = blob_file_addition.DecodeFrom(&slice);
+    ASSERT_TRUE(s.IsCorruption());
+    ASSERT_TRUE(std::strstr(s.getState(), "custom field tag"));
+  }
+
+  constexpr uint32_t custom_tag = 2;
+  PutVarint32(&str, custom_tag);
+  slice = str;
+
+  {
+    const Status s = blob_file_addition.DecodeFrom(&slice);
+    ASSERT_TRUE(s.IsCorruption());
+    ASSERT_TRUE(std::strstr(s.getState(), "custom field value"));
+  }
+}
+
+TEST_F(BlobFileAdditionTest, ForwardCompatibleCustomField) {
+  SyncPoint::GetInstance()->SetCallBack(
+      "BlobFileAddition::EncodeTo::CustomFields", [&](void* arg) {
+        std::string* output = static_cast<std::string*>(arg);
+
+        constexpr uint32_t forward_compatible_tag = 2;
+        PutVarint32(output, forward_compatible_tag);
+
+        PutLengthPrefixedSlice(output, "deadbeef");
+      });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  constexpr uint64_t blob_file_number = 678;
+  constexpr uint64_t total_blob_count = 9999;
+  constexpr uint64_t total_blob_bytes = 100000000;
+  const std::string checksum_method("CRC32");
+  const std::string checksum_value("3d87ff57");
+
+  BlobFileAddition blob_file_addition(blob_file_number, total_blob_count,
+                                      total_blob_bytes, checksum_method,
+                                      checksum_value);
+
+  TestEncodeDecode(blob_file_addition);
+
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+}
+
+TEST_F(BlobFileAdditionTest, ForwardIncompatibleCustomField) {
+  SyncPoint::GetInstance()->SetCallBack(
+      "BlobFileAddition::EncodeTo::CustomFields", [&](void* arg) {
+        std::string* output = static_cast<std::string*>(arg);
+
+        constexpr uint32_t forward_incompatible_tag = (1 << 6) + 1;
+        PutVarint32(output, forward_incompatible_tag);
+
+        PutLengthPrefixedSlice(output, "foobar");
+      });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  constexpr uint64_t blob_file_number = 456;
+  constexpr uint64_t total_blob_count = 100;
+  constexpr uint64_t total_blob_bytes = 2000000;
+  const std::string checksum_method("CRC32B");
+  const std::string checksum_value("6dbdf23a");
+
+  BlobFileAddition blob_file_addition(blob_file_number, total_blob_count,
+                                      total_blob_bytes, checksum_method,
+                                      checksum_value);
+
+  std::string encoded;
+  blob_file_addition.EncodeTo(&encoded);
+
+  BlobFileAddition decoded_blob_file_addition;
+  Slice input(encoded);
+  const Status s = decoded_blob_file_addition.DecodeFrom(&input);
+
+  ASSERT_TRUE(s.IsCorruption());
+  ASSERT_TRUE(std::strstr(s.getState(), "Forward incompatible"));
+
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+}
+
+}  // namespace ROCKSDB_NAMESPACE
+
+int main(int argc, char** argv) {
+  ::testing::InitGoogleTest(&argc, argv);
+  return RUN_ALL_TESTS();
+}

db/blob/blob_file_garbage.cc

@@ -0,0 +1,134 @@
+//  Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#include "db/blob/blob_file_garbage.h"
+
+#include <ostream>
+#include <sstream>
+
+#include "logging/event_logger.h"
+#include "rocksdb/slice.h"
+#include "rocksdb/status.h"
+#include "test_util/sync_point.h"
+#include "util/coding.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+// Tags for custom fields. Note that these get persisted in the manifest,
+// so existing tags should not be modified.
+enum BlobFileGarbage::CustomFieldTags : uint32_t {
+  kEndMarker,
+
+  // Add forward compatible fields here
+
+  /////////////////////////////////////////////////////////////////////
+
+  kForwardIncompatibleMask = 1 << 6,
+
+  // Add forward incompatible fields here
+};
+
+void BlobFileGarbage::EncodeTo(std::string* output) const {
+  PutVarint64(output, blob_file_number_);
+  PutVarint64(output, garbage_blob_count_);
+  PutVarint64(output, garbage_blob_bytes_);
+
+  // Encode any custom fields here. The format to use is a Varint32 tag (see
+  // CustomFieldTags above) followed by a length prefixed slice. Unknown custom
+  // fields will be ignored during decoding unless they're in the forward
+  // incompatible range.
+
+  TEST_SYNC_POINT_CALLBACK("BlobFileGarbage::EncodeTo::CustomFields", output);
+
+  PutVarint32(output, kEndMarker);
+}
+
+Status BlobFileGarbage::DecodeFrom(Slice* input) {
+  constexpr char class_name[] = "BlobFileGarbage";
+
+  if (!GetVarint64(input, &blob_file_number_)) {
+    return Status::Corruption(class_name, "Error decoding blob file number");
+  }
+
+  if (!GetVarint64(input, &garbage_blob_count_)) {
+    return Status::Corruption(class_name, "Error decoding garbage blob count");
+  }
+
+  if (!GetVarint64(input, &garbage_blob_bytes_)) {
+    return Status::Corruption(class_name, "Error decoding garbage blob bytes");
+  }
+
+  while (true) {
+    uint32_t custom_field_tag = 0;
+    if (!GetVarint32(input, &custom_field_tag)) {
+      return Status::Corruption(class_name, "Error decoding custom field tag");
+    }
+
+    if (custom_field_tag == kEndMarker) {
+      break;
+    }
+
+    if (custom_field_tag & kForwardIncompatibleMask) {
+      return Status::Corruption(
+          class_name, "Forward incompatible custom field encountered");
+    }
+
+    Slice custom_field_value;
+    if (!GetLengthPrefixedSlice(input, &custom_field_value)) {
+      return Status::Corruption(class_name,
+                                "Error decoding custom field value");
+    }
+  }
+
+  return Status::OK();
+}
+
+std::string BlobFileGarbage::DebugString() const {
+  std::ostringstream oss;
+
+  oss << *this;
+
+  return oss.str();
+}
+
+std::string BlobFileGarbage::DebugJSON() const {
+  JSONWriter jw;
+
+  jw << *this;
+
+  jw.EndObject();
+
+  return jw.Get();
+}
+
+bool operator==(const BlobFileGarbage& lhs, const BlobFileGarbage& rhs) {
+  return lhs.GetBlobFileNumber() == rhs.GetBlobFileNumber() &&
+         lhs.GetGarbageBlobCount() == rhs.GetGarbageBlobCount() &&
+         lhs.GetGarbageBlobBytes() == rhs.GetGarbageBlobBytes();
+}
+
+bool operator!=(const BlobFileGarbage& lhs, const BlobFileGarbage& rhs) {
+  return !(lhs == rhs);
+}
+
+std::ostream& operator<<(std::ostream& os,
+                         const BlobFileGarbage& blob_file_garbage) {
+  os << "blob_file_number: " << blob_file_garbage.GetBlobFileNumber()
+     << " garbage_blob_count: " << blob_file_garbage.GetGarbageBlobCount()
+     << " garbage_blob_bytes: " << blob_file_garbage.GetGarbageBlobBytes();
+
+  return os;
+}
+
+JSONWriter& operator<<(JSONWriter& jw,
+                       const BlobFileGarbage& blob_file_garbage) {
+  jw << "BlobFileNumber" << blob_file_garbage.GetBlobFileNumber()
+     << "GarbageBlobCount" << blob_file_garbage.GetGarbageBlobCount()
+     << "GarbageBlobBytes" << blob_file_garbage.GetGarbageBlobBytes();
+
+  return jw;
+}
+
+}  // namespace ROCKSDB_NAMESPACE

db/blob/blob_file_garbage.h

@@ -0,0 +1,57 @@
+//  Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#pragma once
+
+#include <cstdint>
+#include <iosfwd>
+#include <string>
+
+#include "db/blob/blob_constants.h"
+#include "rocksdb/rocksdb_namespace.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+class JSONWriter;
+class Slice;
+class Status;
+
+class BlobFileGarbage {
+ public:
+  BlobFileGarbage() = default;
+
+  BlobFileGarbage(uint64_t blob_file_number, uint64_t garbage_blob_count,
+                  uint64_t garbage_blob_bytes)
+      : blob_file_number_(blob_file_number),
+        garbage_blob_count_(garbage_blob_count),
+        garbage_blob_bytes_(garbage_blob_bytes) {}
+
+  uint64_t GetBlobFileNumber() const { return blob_file_number_; }
+  uint64_t GetGarbageBlobCount() const { return garbage_blob_count_; }
+  uint64_t GetGarbageBlobBytes() const { return garbage_blob_bytes_; }
+
+  void EncodeTo(std::string* output) const;
+  Status DecodeFrom(Slice* input);
+
+  std::string DebugString() const;
+  std::string DebugJSON() const;
+
+ private:
+  enum CustomFieldTags : uint32_t;
+
+  uint64_t blob_file_number_ = kInvalidBlobFileNumber;
+  uint64_t garbage_blob_count_ = 0;
+  uint64_t garbage_blob_bytes_ = 0;
+};
+
+bool operator==(const BlobFileGarbage& lhs, const BlobFileGarbage& rhs);
+bool operator!=(const BlobFileGarbage& lhs, const BlobFileGarbage& rhs);
+
+std::ostream& operator<<(std::ostream& os,
+                         const BlobFileGarbage& blob_file_garbage);
+JSONWriter& operator<<(JSONWriter& jw,
+                       const BlobFileGarbage& blob_file_garbage);
+
+}  // namespace ROCKSDB_NAMESPACE

db/blob/blob_file_garbage_test.cc

@@ -0,0 +1,173 @@
+//  Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+
+#include "db/blob/blob_file_garbage.h"
+
+#include <cstdint>
+#include <cstring>
+#include <string>
+
+#include "test_util/sync_point.h"
+#include "test_util/testharness.h"
+#include "util/coding.h"
+
+namespace ROCKSDB_NAMESPACE {
+
+class BlobFileGarbageTest : public testing::Test {
+ public:
+  static void TestEncodeDecode(const BlobFileGarbage& blob_file_garbage) {
+    std::string encoded;
+    blob_file_garbage.EncodeTo(&encoded);
+
+    BlobFileGarbage decoded;
+    Slice input(encoded);
+    ASSERT_OK(decoded.DecodeFrom(&input));
+
+    ASSERT_EQ(blob_file_garbage, decoded);
+  }
+};
+
+TEST_F(BlobFileGarbageTest, Empty) {
+  BlobFileGarbage blob_file_garbage;
+
+  ASSERT_EQ(blob_file_garbage.GetBlobFileNumber(), kInvalidBlobFileNumber);
+  ASSERT_EQ(blob_file_garbage.GetGarbageBlobCount(), 0);
+  ASSERT_EQ(blob_file_garbage.GetGarbageBlobBytes(), 0);
+
+  TestEncodeDecode(blob_file_garbage);
+}
+
+TEST_F(BlobFileGarbageTest, NonEmpty) {
+  constexpr uint64_t blob_file_number = 123;
+  constexpr uint64_t garbage_blob_count = 1;
+  constexpr uint64_t garbage_blob_bytes = 9876;
+
+  BlobFileGarbage blob_file_garbage(blob_file_number, garbage_blob_count,
+                                    garbage_blob_bytes);
+
+  ASSERT_EQ(blob_file_garbage.GetBlobFileNumber(), blob_file_number);
+  ASSERT_EQ(blob_file_garbage.GetGarbageBlobCount(), garbage_blob_count);
+  ASSERT_EQ(blob_file_garbage.GetGarbageBlobBytes(), garbage_blob_bytes);
+
+  TestEncodeDecode(blob_file_garbage);
+}
+
+TEST_F(BlobFileGarbageTest, DecodeErrors) {
+  std::string str;
+  Slice slice(str);
+
+  BlobFileGarbage blob_file_garbage;
+
+  {
+    const Status s = blob_file_garbage.DecodeFrom(&slice);
+    ASSERT_TRUE(s.IsCorruption());
+    ASSERT_TRUE(std::strstr(s.getState(), "blob file number"));
+  }
+
+  constexpr uint64_t blob_file_number = 123;
+  PutVarint64(&str, blob_file_number);
+  slice = str;
+
+  {
+    const Status s = blob_file_garbage.DecodeFrom(&slice);
+    ASSERT_TRUE(s.IsCorruption());
+    ASSERT_TRUE(std::strstr(s.getState(), "garbage blob count"));
+  }
+
+  constexpr uint64_t garbage_blob_count = 4567;
+  PutVarint64(&str, garbage_blob_count);
+  slice = str;
+
+  {
+    const Status s = blob_file_garbage.DecodeFrom(&slice);
+    ASSERT_TRUE(s.IsCorruption());
+    ASSERT_TRUE(std::strstr(s.getState(), "garbage blob bytes"));
+  }
+
+  constexpr uint64_t garbage_blob_bytes = 12345678;
+  PutVarint64(&str, garbage_blob_bytes);
+  slice = str;
+
+  {
+    const Status s = blob_file_garbage.DecodeFrom(&slice);
+    ASSERT_TRUE(s.IsCorruption());
+    ASSERT_TRUE(std::strstr(s.getState(), "custom field tag"));
+  }
+
+  constexpr uint32_t custom_tag = 2;
+  PutVarint32(&str, custom_tag);
+  slice = str;
+
+  {
+    const Status s = blob_file_garbage.DecodeFrom(&slice);
+    ASSERT_TRUE(s.IsCorruption());
+    ASSERT_TRUE(std::strstr(s.getState(), "custom field value"));
+  }
+}
+
+TEST_F(BlobFileGarbageTest, ForwardCompatibleCustomField) {
+  SyncPoint::GetInstance()->SetCallBack(
+      "BlobFileGarbage::EncodeTo::CustomFields", [&](void* arg) {
+        std::string* output = static_cast<std::string*>(arg);
+
+        constexpr uint32_t forward_compatible_tag = 2;
+        PutVarint32(output, forward_compatible_tag);
+
+        PutLengthPrefixedSlice(output, "deadbeef");
+      });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  constexpr uint64_t blob_file_number = 678;
+  constexpr uint64_t garbage_blob_count = 9999;
+  constexpr uint64_t garbage_blob_bytes = 100000000;
+
+  BlobFileGarbage blob_file_garbage(blob_file_number, garbage_blob_count,
+                                    garbage_blob_bytes);
+
+  TestEncodeDecode(blob_file_garbage);
+
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+}
+
+TEST_F(BlobFileGarbageTest, ForwardIncompatibleCustomField) {
+  SyncPoint::GetInstance()->SetCallBack(
+      "BlobFileGarbage::EncodeTo::CustomFields", [&](void* arg) {
+        std::string* output = static_cast<std::string*>(arg);
+
+        constexpr uint32_t forward_incompatible_tag = (1 << 6) + 1;
+        PutVarint32(output, forward_incompatible_tag);
+
+        PutLengthPrefixedSlice(output, "foobar");
+      });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  constexpr uint64_t blob_file_number = 456;
+  constexpr uint64_t garbage_blob_count = 100;
+  constexpr uint64_t garbage_blob_bytes = 2000000;
+
+  BlobFileGarbage blob_file_garbage(blob_file_number, garbage_blob_count,
+                                    garbage_blob_bytes);
+
+  std::string encoded;
+  blob_file_garbage.EncodeTo(&encoded);
+
+  BlobFileGarbage decoded_blob_file_addition;
+  Slice input(encoded);
+  const Status s = decoded_blob_file_addition.DecodeFrom(&input);
+
+  ASSERT_TRUE(s.IsCorruption());
+  ASSERT_TRUE(std::strstr(s.getState(), "Forward incompatible"));
+
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+}
+
+}  // namespace ROCKSDB_NAMESPACE
+
+int main(int argc, char** argv) {
+  ::testing::InitGoogleTest(&argc, argv);
+  return RUN_ALL_TESTS();
+}

db/blob/blob_index.h

@@ -12,7 +12,7 @@
 #include "util/coding.h"
 #include "util/string_util.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 // BlobIndex is a pointer to the blob and metadata of the blob. The index is
 // stored in base DB as ValueType::kTypeBlobIndex.
@@ -111,7 +111,7 @@ class BlobIndex {
     return Status::OK();
   }
 
-  std::string DebugString(bool output_hex) {
+  std::string DebugString(bool output_hex) const {
     std::ostringstream oss;
 
     if (IsInlined()) {
@@ -175,5 +175,5 @@ class BlobIndex {
   CompressionType compression_ = kNoCompression;
 };
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 #endif  // ROCKSDB_LITE

db/blob/db_blob_index_test.cc

@@ -23,7 +23,7 @@
 #include "util/string_util.h"
 #include "utilities/merge_operators.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 // kTypeBlobIndex is a value type used by BlobDB only. The base rocksdb
 // should accept the value type on write, and report not supported value
@@ -427,10 +427,10 @@ TEST_F(DBBlobIndexTest, Iterate) {
   }
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
-  rocksdb::port::InstallStackTraceHandler();
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
   ::testing::InitGoogleTest(&argc, argv);
   return RUN_ALL_TESTS();
 }

db/builder.cc

@@ -37,7 +37,7 @@
 #include "test_util/sync_point.h"
 #include "util/stop_watch.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class TableFactory;
 
@@ -66,8 +66,9 @@ TableBuilder* NewTableBuilder(
 }
 
 Status BuildTable(
-    const std::string& dbname, Env* env, const ImmutableCFOptions& ioptions,
-    const MutableCFOptions& mutable_cf_options, const EnvOptions& env_options,
+    const std::string& dbname, Env* env, FileSystem* fs,
+    const ImmutableCFOptions& ioptions,
+    const MutableCFOptions& mutable_cf_options, const FileOptions& file_options,
     TableCache* table_cache, InternalIterator* iter,
     std::vector<std::unique_ptr<FragmentedRangeTombstoneIterator>>
         range_del_iters,
@@ -115,12 +116,12 @@ Status BuildTable(
     compression_opts_for_flush.max_dict_bytes = 0;
     compression_opts_for_flush.zstd_max_train_bytes = 0;
     {
-      std::unique_ptr<WritableFile> file;
+      std::unique_ptr<FSWritableFile> file;
 #ifndef NDEBUG
-      bool use_direct_writes = env_options.use_direct_writes;
+      bool use_direct_writes = file_options.use_direct_writes;
       TEST_SYNC_POINT_CALLBACK("BuildTable:create_file", &use_direct_writes);
 #endif  // !NDEBUG
-      s = NewWritableFile(env, fname, &file, env_options);
+      s = NewWritableFile(fs, fname, &file, file_options);
       if (!s.ok()) {
         EventHelpers::LogAndNotifyTableFileCreationFinished(
             event_logger, ioptions.listeners, dbname, column_family_name, fname,
@@ -130,9 +131,10 @@ Status BuildTable(
       file->SetIOPriority(io_priority);
       file->SetWriteLifeTimeHint(write_hint);
 
-      file_writer.reset(
-          new WritableFileWriter(std::move(file), fname, env_options, env,
-                                 ioptions.statistics, ioptions.listeners));
+      file_writer.reset(new WritableFileWriter(
+          std::move(file), fname, file_options, env, ioptions.statistics,
+          ioptions.listeners, ioptions.sst_file_checksum_func));
+
       builder = NewTableBuilder(
           ioptions, mutable_cf_options, internal_comparator,
           int_tbl_prop_collector_factories, column_family_id,
@@ -198,6 +200,9 @@ Status BuildTable(
       if (table_properties) {
         *table_properties = tp;
       }
+      // Add the checksum information to file metadata.
+      meta->file_checksum = builder->GetFileChecksum();
+      meta->file_checksum_func_name = builder->GetFileChecksumFuncName();
     }
     delete builder;
 
@@ -218,7 +223,7 @@ Status BuildTable(
       // we will regrad this verification as user reads since the goal is
       // to cache it here for further user reads
       std::unique_ptr<InternalIterator> it(table_cache->NewIterator(
-          ReadOptions(), env_options, internal_comparator, *meta,
+          ReadOptions(), file_options, internal_comparator, *meta,
           nullptr /* range_del_agg */,
           mutable_cf_options.prefix_extractor.get(), nullptr,
           (internal_stats == nullptr) ? nullptr
@@ -241,7 +246,7 @@ Status BuildTable(
   }
 
   if (!s.ok() || meta->fd.GetFileSize() == 0) {
-    env->DeleteFile(fname);
+    fs->DeleteFile(fname, IOOptions(), nullptr);
   }
 
   if (meta->fd.GetFileSize() == 0) {
@@ -255,4 +260,4 @@ Status BuildTable(
   return s;
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/builder.h

@@ -22,7 +22,7 @@
 #include "rocksdb/types.h"
 #include "table/scoped_arena_iterator.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 struct Options;
 struct FileMetaData;
@@ -62,8 +62,9 @@ TableBuilder* NewTableBuilder(
 // @param column_family_name Name of the column family that is also identified
 //    by column_family_id, or empty string if unknown.
 extern Status BuildTable(
-    const std::string& dbname, Env* env, const ImmutableCFOptions& options,
-    const MutableCFOptions& mutable_cf_options, const EnvOptions& env_options,
+    const std::string& dbname, Env* env, FileSystem* fs,
+    const ImmutableCFOptions& options,
+    const MutableCFOptions& mutable_cf_options, const FileOptions& file_options,
     TableCache* table_cache, InternalIterator* iter,
     std::vector<std::unique_ptr<FragmentedRangeTombstoneIterator>>
         range_del_iters,
@@ -84,4 +85,4 @@ extern Status BuildTable(
     Env::WriteLifeTimeHint write_hint = Env::WLTH_NOT_SET,
     const uint64_t file_creation_time = 0);
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/c.cc

@@ -46,74 +46,74 @@
 #include <unordered_set>
 #include <map>
 
-using rocksdb::BytewiseComparator;
-using rocksdb::Cache;
-using rocksdb::ColumnFamilyDescriptor;
-using rocksdb::ColumnFamilyHandle;
-using rocksdb::ColumnFamilyOptions;
-using rocksdb::CompactionFilter;
-using rocksdb::CompactionFilterFactory;
-using rocksdb::CompactionFilterContext;
-using rocksdb::CompactionOptionsFIFO;
-using rocksdb::Comparator;
-using rocksdb::CompressionType;
-using rocksdb::WALRecoveryMode;
-using rocksdb::DB;
-using rocksdb::DBOptions;
-using rocksdb::DbPath;
-using rocksdb::Env;
-using rocksdb::EnvOptions;
-using rocksdb::InfoLogLevel;
-using rocksdb::FileLock;
-using rocksdb::FilterPolicy;
-using rocksdb::FlushOptions;
-using rocksdb::IngestExternalFileOptions;
-using rocksdb::Iterator;
-using rocksdb::Logger;
-using rocksdb::MergeOperator;
-using rocksdb::MergeOperators;
-using rocksdb::NewBloomFilterPolicy;
-using rocksdb::NewLRUCache;
-using rocksdb::Options;
-using rocksdb::BlockBasedTableOptions;
-using rocksdb::CuckooTableOptions;
-using rocksdb::RandomAccessFile;
-using rocksdb::Range;
-using rocksdb::ReadOptions;
-using rocksdb::SequentialFile;
-using rocksdb::Slice;
-using rocksdb::SliceParts;
-using rocksdb::SliceTransform;
-using rocksdb::Snapshot;
-using rocksdb::SstFileWriter;
-using rocksdb::Status;
-using rocksdb::WritableFile;
-using rocksdb::WriteBatch;
-using rocksdb::WriteBatchWithIndex;
-using rocksdb::WriteOptions;
-using rocksdb::LiveFileMetaData;
-using rocksdb::BackupEngine;
-using rocksdb::BackupableDBOptions;
-using rocksdb::BackupInfo;
-using rocksdb::BackupID;
-using rocksdb::RestoreOptions;
-using rocksdb::CompactRangeOptions;
-using rocksdb::BottommostLevelCompaction;
-using rocksdb::RateLimiter;
-using rocksdb::NewGenericRateLimiter;
-using rocksdb::PinnableSlice;
-using rocksdb::TransactionDBOptions;
-using rocksdb::TransactionDB;
-using rocksdb::TransactionOptions;
-using rocksdb::OptimisticTransactionDB;
-using rocksdb::OptimisticTransactionOptions;
-using rocksdb::Transaction;
-using rocksdb::Checkpoint;
-using rocksdb::TransactionLogIterator;
-using rocksdb::BatchResult;
-using rocksdb::PerfLevel;
-using rocksdb::PerfContext;
-using rocksdb::MemoryUtil;
+using ROCKSDB_NAMESPACE::BackupableDBOptions;
+using ROCKSDB_NAMESPACE::BackupEngine;
+using ROCKSDB_NAMESPACE::BackupID;
+using ROCKSDB_NAMESPACE::BackupInfo;
+using ROCKSDB_NAMESPACE::BatchResult;
+using ROCKSDB_NAMESPACE::BlockBasedTableOptions;
+using ROCKSDB_NAMESPACE::BottommostLevelCompaction;
+using ROCKSDB_NAMESPACE::BytewiseComparator;
+using ROCKSDB_NAMESPACE::Cache;
+using ROCKSDB_NAMESPACE::Checkpoint;
+using ROCKSDB_NAMESPACE::ColumnFamilyDescriptor;
+using ROCKSDB_NAMESPACE::ColumnFamilyHandle;
+using ROCKSDB_NAMESPACE::ColumnFamilyOptions;
+using ROCKSDB_NAMESPACE::CompactionFilter;
+using ROCKSDB_NAMESPACE::CompactionFilterContext;
+using ROCKSDB_NAMESPACE::CompactionFilterFactory;
+using ROCKSDB_NAMESPACE::CompactionOptionsFIFO;
+using ROCKSDB_NAMESPACE::CompactRangeOptions;
+using ROCKSDB_NAMESPACE::Comparator;
+using ROCKSDB_NAMESPACE::CompressionType;
+using ROCKSDB_NAMESPACE::CuckooTableOptions;
+using ROCKSDB_NAMESPACE::DB;
+using ROCKSDB_NAMESPACE::DBOptions;
+using ROCKSDB_NAMESPACE::DbPath;
+using ROCKSDB_NAMESPACE::Env;
+using ROCKSDB_NAMESPACE::EnvOptions;
+using ROCKSDB_NAMESPACE::FileLock;
+using ROCKSDB_NAMESPACE::FilterPolicy;
+using ROCKSDB_NAMESPACE::FlushOptions;
+using ROCKSDB_NAMESPACE::InfoLogLevel;
+using ROCKSDB_NAMESPACE::IngestExternalFileOptions;
+using ROCKSDB_NAMESPACE::Iterator;
+using ROCKSDB_NAMESPACE::LiveFileMetaData;
+using ROCKSDB_NAMESPACE::Logger;
+using ROCKSDB_NAMESPACE::MemoryUtil;
+using ROCKSDB_NAMESPACE::MergeOperator;
+using ROCKSDB_NAMESPACE::MergeOperators;
+using ROCKSDB_NAMESPACE::NewBloomFilterPolicy;
+using ROCKSDB_NAMESPACE::NewGenericRateLimiter;
+using ROCKSDB_NAMESPACE::NewLRUCache;
+using ROCKSDB_NAMESPACE::OptimisticTransactionDB;
+using ROCKSDB_NAMESPACE::OptimisticTransactionOptions;
+using ROCKSDB_NAMESPACE::Options;
+using ROCKSDB_NAMESPACE::PerfContext;
+using ROCKSDB_NAMESPACE::PerfLevel;
+using ROCKSDB_NAMESPACE::PinnableSlice;
+using ROCKSDB_NAMESPACE::RandomAccessFile;
+using ROCKSDB_NAMESPACE::Range;
+using ROCKSDB_NAMESPACE::RateLimiter;
+using ROCKSDB_NAMESPACE::ReadOptions;
+using ROCKSDB_NAMESPACE::RestoreOptions;
+using ROCKSDB_NAMESPACE::SequentialFile;
+using ROCKSDB_NAMESPACE::Slice;
+using ROCKSDB_NAMESPACE::SliceParts;
+using ROCKSDB_NAMESPACE::SliceTransform;
+using ROCKSDB_NAMESPACE::Snapshot;
+using ROCKSDB_NAMESPACE::SstFileWriter;
+using ROCKSDB_NAMESPACE::Status;
+using ROCKSDB_NAMESPACE::Transaction;
+using ROCKSDB_NAMESPACE::TransactionDB;
+using ROCKSDB_NAMESPACE::TransactionDBOptions;
+using ROCKSDB_NAMESPACE::TransactionLogIterator;
+using ROCKSDB_NAMESPACE::TransactionOptions;
+using ROCKSDB_NAMESPACE::WALRecoveryMode;
+using ROCKSDB_NAMESPACE::WritableFile;
+using ROCKSDB_NAMESPACE::WriteBatch;
+using ROCKSDB_NAMESPACE::WriteBatchWithIndex;
+using ROCKSDB_NAMESPACE::WriteOptions;
 
 using std::shared_ptr;
 using std::vector;
@@ -452,7 +452,7 @@ struct rocksdb_slicetransform_t : public SliceTransform {
 };
 
 struct rocksdb_universal_compaction_options_t {
-  rocksdb::CompactionOptionsUniversal *rep;
+  ROCKSDB_NAMESPACE::CompactionOptionsUniversal* rep;
 };
 
 static bool SaveError(char** errptr, const Status& s) {
@@ -494,8 +494,9 @@ rocksdb_t* rocksdb_open_with_ttl(
     const char* name,
     int ttl,
     char** errptr) {
-  rocksdb::DBWithTTL* db;
-  if (SaveError(errptr, rocksdb::DBWithTTL::Open(options->rep, std::string(name), &db, ttl))) {
+  ROCKSDB_NAMESPACE::DBWithTTL* db;
+  if (SaveError(errptr, ROCKSDB_NAMESPACE::DBWithTTL::Open(
+                            options->rep, std::string(name), &db, ttl))) {
     return nullptr;
   }
   rocksdb_t* result = new rocksdb_t;
@@ -664,17 +665,15 @@ void rocksdb_close(rocksdb_t* db) {
 }
 
 void rocksdb_options_set_uint64add_merge_operator(rocksdb_options_t* opt) {
-  opt->rep.merge_operator = rocksdb::MergeOperators::CreateUInt64AddOperator();
+  opt->rep.merge_operator =
+      ROCKSDB_NAMESPACE::MergeOperators::CreateUInt64AddOperator();
 }
 
 rocksdb_t* rocksdb_open_column_families(
-    const rocksdb_options_t* db_options,
-    const char* name,
-    int num_column_families,
-    const char** column_family_names,
-    const rocksdb_options_t** column_family_options,
-    rocksdb_column_family_handle_t** column_family_handles,
-    char** errptr) {
+    const rocksdb_options_t* db_options, const char* name,
+    int num_column_families, const char* const* column_family_names,
+    const rocksdb_options_t* const* column_family_options,
+    rocksdb_column_family_handle_t** column_family_handles, char** errptr) {
   std::vector<ColumnFamilyDescriptor> column_families;
   for (int i = 0; i < num_column_families; i++) {
     column_families.push_back(ColumnFamilyDescriptor(
@@ -700,14 +699,11 @@ rocksdb_t* rocksdb_open_column_families(
 }
 
 rocksdb_t* rocksdb_open_for_read_only_column_families(
-    const rocksdb_options_t* db_options,
-    const char* name,
-    int num_column_families,
-    const char** column_family_names,
-    const rocksdb_options_t** column_family_options,
+    const rocksdb_options_t* db_options, const char* name,
+    int num_column_families, const char* const* column_family_names,
+    const rocksdb_options_t* const* column_family_options,
     rocksdb_column_family_handle_t** column_family_handles,
-    unsigned char error_if_log_file_exist,
-    char** errptr) {
+    unsigned char error_if_log_file_exist, char** errptr) {
   std::vector<ColumnFamilyDescriptor> column_families;
   for (int i = 0; i < num_column_families; i++) {
     column_families.push_back(ColumnFamilyDescriptor(
@@ -735,8 +731,8 @@ rocksdb_t* rocksdb_open_for_read_only_column_families(
 rocksdb_t* rocksdb_open_as_secondary_column_families(
     const rocksdb_options_t* db_options, const char* name,
     const char* secondary_path, int num_column_families,
-    const char** column_family_names,
-    const rocksdb_options_t** column_family_options,
+    const char* const* column_family_names,
+    const rocksdb_options_t* const* column_family_options,
     rocksdb_column_family_handle_t** column_family_handles, char** errptr) {
   std::vector<ColumnFamilyDescriptor> column_families;
   for (int i = 0; i != num_column_families; ++i) {
@@ -852,6 +848,17 @@ void rocksdb_delete_cf(
         Slice(key, keylen)));
 }
 
+void rocksdb_delete_range_cf(rocksdb_t* db,
+                             const rocksdb_writeoptions_t* options,
+                             rocksdb_column_family_handle_t* column_family,
+                             const char* start_key, size_t start_key_len,
+                             const char* end_key, size_t end_key_len,
+                             char** errptr) {
+  SaveError(errptr, db->rep->DeleteRange(options->rep, column_family->rep,
+                                         Slice(start_key, start_key_len),
+                                         Slice(end_key, end_key_len)));
+}
+
 void rocksdb_merge(
     rocksdb_t* db,
     const rocksdb_writeoptions_t* options,
@@ -2032,6 +2039,17 @@ void rocksdb_block_based_options_set_index_type(
   options->rep.index_type = static_cast<BlockBasedTableOptions::IndexType>(v);
 }
 
+void rocksdb_block_based_options_set_data_block_index_type(
+    rocksdb_block_based_table_options_t* options, int v) {
+  options->rep.data_block_index_type =
+          static_cast<BlockBasedTableOptions::DataBlockIndexType>(v);
+}
+
+void rocksdb_block_based_options_set_data_block_hash_ratio(
+    rocksdb_block_based_table_options_t* options, double v) {
+  options->rep.data_block_hash_table_util_ratio = v;
+}
+
 void rocksdb_block_based_options_set_hash_index_allow_collision(
     rocksdb_block_based_table_options_t* options, unsigned char v) {
   options->rep.hash_index_allow_collision = v;
@@ -2062,7 +2080,7 @@ void rocksdb_options_set_block_based_table_factory(
     rocksdb_block_based_table_options_t* table_options) {
   if (table_options) {
     opt->rep.table_factory.reset(
-        rocksdb::NewBlockBasedTableFactory(table_options->rep));
+        ROCKSDB_NAMESPACE::NewBlockBasedTableFactory(table_options->rep));
   }
 }
 
@@ -2106,7 +2124,7 @@ void rocksdb_options_set_cuckoo_table_factory(
     rocksdb_cuckoo_table_options_t* table_options) {
   if (table_options) {
     opt->rep.table_factory.reset(
-        rocksdb::NewCuckooTableFactory(table_options->rep));
+        ROCKSDB_NAMESPACE::NewCuckooTableFactory(table_options->rep));
   }
 }
 
@@ -2296,7 +2314,7 @@ void rocksdb_options_set_max_bytes_for_level_multiplier_additional(
 }
 
 void rocksdb_options_enable_statistics(rocksdb_options_t* opt) {
-  opt->rep.statistics = rocksdb::CreateDBStatistics();
+  opt->rep.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
 }
 
 void rocksdb_options_set_skip_stats_update_on_db_open(rocksdb_options_t* opt,
@@ -2304,6 +2322,11 @@ void rocksdb_options_set_skip_stats_update_on_db_open(rocksdb_options_t* opt,
   opt->rep.skip_stats_update_on_db_open = val;
 }
 
+void rocksdb_options_set_skip_checking_sst_file_sizes_on_db_open(
+    rocksdb_options_t* opt, unsigned char val) {
+  opt->rep.skip_checking_sst_file_sizes_on_db_open = val;
+}
+
 void rocksdb_options_set_num_levels(rocksdb_options_t* opt, int n) {
   opt->rep.num_levels = n;
 }
@@ -2447,16 +2470,20 @@ void rocksdb_options_set_access_hint_on_compaction_start(
     rocksdb_options_t* opt, int v) {
   switch(v) {
     case 0:
-      opt->rep.access_hint_on_compaction_start = rocksdb::Options::NONE;
+      opt->rep.access_hint_on_compaction_start =
+          ROCKSDB_NAMESPACE::Options::NONE;
       break;
     case 1:
-      opt->rep.access_hint_on_compaction_start = rocksdb::Options::NORMAL;
+      opt->rep.access_hint_on_compaction_start =
+          ROCKSDB_NAMESPACE::Options::NORMAL;
       break;
     case 2:
-      opt->rep.access_hint_on_compaction_start = rocksdb::Options::SEQUENTIAL;
+      opt->rep.access_hint_on_compaction_start =
+          ROCKSDB_NAMESPACE::Options::SEQUENTIAL;
       break;
     case 3:
-      opt->rep.access_hint_on_compaction_start = rocksdb::Options::WILLNEED;
+      opt->rep.access_hint_on_compaction_start =
+          ROCKSDB_NAMESPACE::Options::WILLNEED;
       break;
   }
 }
@@ -2622,7 +2649,7 @@ void rocksdb_options_prepare_for_bulk_load(rocksdb_options_t* opt) {
 }
 
 void rocksdb_options_set_memtable_vector_rep(rocksdb_options_t *opt) {
-  opt->rep.memtable_factory.reset(new rocksdb::VectorRepFactory);
+  opt->rep.memtable_factory.reset(new ROCKSDB_NAMESPACE::VectorRepFactory);
 }
 
 void rocksdb_options_set_memtable_prefix_bloom_size_ratio(
@@ -2638,26 +2665,29 @@ void rocksdb_options_set_memtable_huge_page_size(rocksdb_options_t* opt,
 void rocksdb_options_set_hash_skip_list_rep(
     rocksdb_options_t *opt, size_t bucket_count,
     int32_t skiplist_height, int32_t skiplist_branching_factor) {
-  rocksdb::MemTableRepFactory* factory = rocksdb::NewHashSkipListRepFactory(
-      bucket_count, skiplist_height, skiplist_branching_factor);
+  ROCKSDB_NAMESPACE::MemTableRepFactory* factory =
+      ROCKSDB_NAMESPACE::NewHashSkipListRepFactory(
+          bucket_count, skiplist_height, skiplist_branching_factor);
   opt->rep.memtable_factory.reset(factory);
 }
 
 void rocksdb_options_set_hash_link_list_rep(
     rocksdb_options_t *opt, size_t bucket_count) {
-  opt->rep.memtable_factory.reset(rocksdb::NewHashLinkListRepFactory(bucket_count));
+  opt->rep.memtable_factory.reset(
+      ROCKSDB_NAMESPACE::NewHashLinkListRepFactory(bucket_count));
 }
 
 void rocksdb_options_set_plain_table_factory(
     rocksdb_options_t *opt, uint32_t user_key_len, int bloom_bits_per_key,
     double hash_table_ratio, size_t index_sparseness) {
-  rocksdb::PlainTableOptions options;
+  ROCKSDB_NAMESPACE::PlainTableOptions options;
   options.user_key_len = user_key_len;
   options.bloom_bits_per_key = bloom_bits_per_key;
   options.hash_table_ratio = hash_table_ratio;
   options.index_sparseness = index_sparseness;
 
-  rocksdb::TableFactory* factory = rocksdb::NewPlainTableFactory(options);
+  ROCKSDB_NAMESPACE::TableFactory* factory =
+      ROCKSDB_NAMESPACE::NewPlainTableFactory(options);
   opt->rep.table_factory.reset(factory);
 }
 
@@ -2687,7 +2717,8 @@ void rocksdb_options_set_report_bg_io_stats(
 }
 
 void rocksdb_options_set_compaction_style(rocksdb_options_t *opt, int style) {
-  opt->rep.compaction_style = static_cast<rocksdb::CompactionStyle>(style);
+  opt->rep.compaction_style =
+      static_cast<ROCKSDB_NAMESPACE::CompactionStyle>(style);
 }
 
 void rocksdb_options_set_universal_compaction_options(rocksdb_options_t *opt, rocksdb_universal_compaction_options_t *uco) {
@@ -2701,7 +2732,7 @@ void rocksdb_options_set_fifo_compaction_options(
 }
 
 char *rocksdb_options_statistics_get_string(rocksdb_options_t *opt) {
-  rocksdb::Statistics *statistics = opt->rep.statistics.get();
+  ROCKSDB_NAMESPACE::Statistics* statistics = opt->rep.statistics.get();
   if (statistics) {
     return strdup(statistics->ToString().c_str());
   }
@@ -2714,6 +2745,11 @@ void rocksdb_options_set_ratelimiter(rocksdb_options_t *opt, rocksdb_ratelimiter
   }
 }
 
+void rocksdb_options_set_atomic_flush(rocksdb_options_t* opt,
+                                      unsigned char atomic_flush) {
+  opt->rep.atomic_flush = atomic_flush;
+}
+
 rocksdb_ratelimiter_t* rocksdb_ratelimiter_create(
     int64_t rate_bytes_per_sec,
     int64_t refill_period_us,
@@ -2729,6 +2765,12 @@ void rocksdb_ratelimiter_destroy(rocksdb_ratelimiter_t *limiter) {
   delete limiter;
 }
 
+void rocksdb_options_set_row_cache(rocksdb_options_t* opt, rocksdb_cache_t* cache) {
+  if(cache) {
+    opt->rep.row_cache = cache->rep;
+  }
+}
+
 void rocksdb_set_perf_level(int v) {
   PerfLevel level = static_cast<PerfLevel>(v);
   SetPerfLevel(level);
@@ -2736,7 +2778,7 @@ void rocksdb_set_perf_level(int v) {
 
 rocksdb_perfcontext_t* rocksdb_perfcontext_create() {
   rocksdb_perfcontext_t* context = new rocksdb_perfcontext_t;
-  context->rep = rocksdb::get_perf_context();
+  context->rep = ROCKSDB_NAMESPACE::get_perf_context();
   return context;
 }
 
@@ -3037,7 +3079,17 @@ rocksdb_filterpolicy_t* rocksdb_filterpolicy_create_bloom_format(int bits_per_ke
     bool KeyMayMatch(const Slice& key, const Slice& filter) const override {
       return rep_->KeyMayMatch(key, filter);
     }
-    static void DoNothing(void*) { }
+    // No need to override GetFilterBitsBuilder if this one is overridden
+    ROCKSDB_NAMESPACE::FilterBitsBuilder* GetBuilderWithContext(
+        const ROCKSDB_NAMESPACE::FilterBuildingContext& context)
+        const override {
+      return rep_->GetBuilderWithContext(context);
+    }
+    ROCKSDB_NAMESPACE::FilterBitsReader* GetFilterBitsReader(
+        const Slice& contents) const override {
+      return rep_->GetFilterBitsReader(contents);
+    }
+    static void DoNothing(void*) {}
   };
   Wrapper* wrapper = new Wrapper;
   wrapper->rep_ = NewBloomFilterPolicy(bits_per_key, original_format);
@@ -3135,7 +3187,7 @@ void rocksdb_readoptions_set_iterate_lower_bound(
 
 void rocksdb_readoptions_set_read_tier(
     rocksdb_readoptions_t* opt, int v) {
-  opt->rep.read_tier = static_cast<rocksdb::ReadTier>(v);
+  opt->rep.read_tier = static_cast<ROCKSDB_NAMESPACE::ReadTier>(v);
 }
 
 void rocksdb_readoptions_set_tailing(
@@ -3307,7 +3359,7 @@ rocksdb_env_t* rocksdb_create_default_env() {
 
 rocksdb_env_t* rocksdb_create_mem_env() {
   rocksdb_env_t* result = new rocksdb_env_t;
-  result->rep = rocksdb::NewMemEnv(Env::Default());
+  result->rep = ROCKSDB_NAMESPACE::NewMemEnv(Env::Default());
   result->is_default = false;
   return result;
 }
@@ -3522,7 +3574,7 @@ struct Wrapper : public rocksdb_slicetransform_t {
 
 rocksdb_slicetransform_t* rocksdb_slicetransform_create_fixed_prefix(size_t prefixLen) {
   Wrapper* wrapper = new Wrapper;
-  wrapper->rep_ = rocksdb::NewFixedPrefixTransform(prefixLen);
+  wrapper->rep_ = ROCKSDB_NAMESPACE::NewFixedPrefixTransform(prefixLen);
   wrapper->state_ = nullptr;
   wrapper->destructor_ = &Wrapper::DoNothing;
   return wrapper;
@@ -3530,7 +3582,7 @@ rocksdb_slicetransform_t* rocksdb_slicetransform_create_fixed_prefix(size_t pref
 
 rocksdb_slicetransform_t* rocksdb_slicetransform_create_noop() {
   Wrapper* wrapper = new Wrapper;
-  wrapper->rep_ = rocksdb::NewNoopTransform();
+  wrapper->rep_ = ROCKSDB_NAMESPACE::NewNoopTransform();
   wrapper->state_ = nullptr;
   wrapper->destructor_ = &Wrapper::DoNothing;
   return wrapper;
@@ -3538,7 +3590,7 @@ rocksdb_slicetransform_t* rocksdb_slicetransform_create_noop() {
 
 rocksdb_universal_compaction_options_t* rocksdb_universal_compaction_options_create() {
   rocksdb_universal_compaction_options_t* result = new rocksdb_universal_compaction_options_t;
-  result->rep = new rocksdb::CompactionOptionsUniversal;
+  result->rep = new ROCKSDB_NAMESPACE::CompactionOptionsUniversal;
   return result;
 }
 
@@ -3569,7 +3621,8 @@ void rocksdb_universal_compaction_options_set_compression_size_percent(
 
 void rocksdb_universal_compaction_options_set_stop_style(
   rocksdb_universal_compaction_options_t* uco, int style) {
-  uco->rep->stop_style = static_cast<rocksdb::CompactionStopStyle>(style);
+  uco->rep->stop_style =
+      static_cast<ROCKSDB_NAMESPACE::CompactionStopStyle>(style);
 }
 
 void rocksdb_universal_compaction_options_destroy(
@@ -3599,7 +3652,7 @@ void rocksdb_options_set_min_level_to_compress(rocksdb_options_t* opt, int level
     assert(level <= opt->rep.num_levels);
     opt->rep.compression_per_level.resize(opt->rep.num_levels);
     for (int i = 0; i < level; i++) {
-      opt->rep.compression_per_level[i] = rocksdb::kNoCompression;
+      opt->rep.compression_per_level[i] = ROCKSDB_NAMESPACE::kNoCompression;
     }
     for (int i = level; i < opt->rep.num_levels; i++) {
       opt->rep.compression_per_level[i] = opt->rep.compression;
@@ -3806,8 +3859,8 @@ rocksdb_transactiondb_t* rocksdb_transactiondb_open(
 rocksdb_transactiondb_t* rocksdb_transactiondb_open_column_families(
     const rocksdb_options_t* options,
     const rocksdb_transactiondb_options_t* txn_db_options, const char* name,
-    int num_column_families, const char** column_family_names,
-    const rocksdb_options_t** column_family_options,
+    int num_column_families, const char* const* column_family_names,
+    const rocksdb_options_t* const* column_family_options,
     rocksdb_column_family_handle_t** column_family_handles, char** errptr) {
   std::vector<ColumnFamilyDescriptor> column_families;
   for (int i = 0; i < num_column_families; i++) {
@@ -4184,8 +4237,8 @@ rocksdb_optimistictransactiondb_t* rocksdb_optimistictransactiondb_open(
 rocksdb_optimistictransactiondb_t*
 rocksdb_optimistictransactiondb_open_column_families(
     const rocksdb_options_t* db_options, const char* name,
-    int num_column_families, const char** column_family_names,
-    const rocksdb_options_t** column_family_options,
+    int num_column_families, const char* const* column_family_names,
+    const rocksdb_options_t* const* column_family_options,
     rocksdb_column_family_handle_t** column_family_handles, char** errptr) {
   std::vector<ColumnFamilyDescriptor> column_families;
   for (int i = 0; i < num_column_families; i++) {
@@ -4301,7 +4354,8 @@ const char* rocksdb_pinnableslice_value(const rocksdb_pinnableslice_t* v,
   return v->rep.data();
 }
 
-// container to keep databases and caches in order to use rocksdb::MemoryUtil
+// container to keep databases and caches in order to use
+// ROCKSDB_NAMESPACE::MemoryUtil
 struct rocksdb_memory_consumers_t {
   std::vector<rocksdb_t*> dbs;
   std::unordered_set<rocksdb_cache_t*> caches;
@@ -4329,7 +4383,7 @@ void rocksdb_memory_consumers_destroy(rocksdb_memory_consumers_t* consumers) {
   delete consumers;
 }
 
-// contains memory usage statistics provided by rocksdb::MemoryUtil
+// contains memory usage statistics provided by ROCKSDB_NAMESPACE::MemoryUtil
 struct rocksdb_memory_usage_t {
   uint64_t mem_table_total;
   uint64_t mem_table_unflushed;
@@ -4351,7 +4405,7 @@ rocksdb_memory_usage_t* rocksdb_approximate_memory_usage_create(
     cache_set.insert(const_cast<const Cache*>(cache->rep.get()));
   }
 
-  std::map<rocksdb::MemoryUtil::UsageType, uint64_t> usage_by_type;
+  std::map<ROCKSDB_NAMESPACE::MemoryUtil::UsageType, uint64_t> usage_by_type;
 
   auto status = MemoryUtil::GetApproximateMemoryUsageByType(dbs, cache_set,
                                                             &usage_by_type);

db/c_test.c

@@ -487,8 +487,11 @@ int main(int argc, char** argv) {
   rocksdb_options_set_paranoid_checks(options, 1);
   rocksdb_options_set_max_open_files(options, 10);
   rocksdb_options_set_base_background_compactions(options, 1);
+
   table_options = rocksdb_block_based_options_create();
   rocksdb_block_based_options_set_block_cache(table_options, cache);
+  rocksdb_block_based_options_set_data_block_index_type(table_options, 1);
+  rocksdb_block_based_options_set_data_block_hash_ratio(table_options, 0.75);
   rocksdb_options_set_block_based_table_factory(options, table_options);
 
   rocksdb_options_set_compression(options, rocksdb_no_compression);
@@ -832,23 +835,6 @@ int main(int argc, char** argv) {
     rocksdb_writebatch_wi_iterate(wbi, &pos, CheckPut, CheckDel);
     CheckCondition(pos == 3);
     rocksdb_writebatch_wi_clear(wbi);
-    rocksdb_writebatch_wi_put(wbi, "bar", 3, "b", 1);
-    rocksdb_writebatch_wi_put(wbi, "bay", 3, "d", 1);
-    rocksdb_writebatch_wi_delete_range(wbi, "bar", 3, "bay", 3);
-    rocksdb_write_writebatch_wi(db, woptions, wbi, &err);
-    CheckNoError(err);
-    CheckGet(db, roptions, "bar", NULL);
-    CheckGet(db, roptions, "bay", "d");
-    rocksdb_writebatch_wi_clear(wbi);
-    const char* start_list[1] = {"bay"};
-    const size_t start_sizes[1] = {3};
-    const char* end_list[1] = {"baz"};
-    const size_t end_sizes[1] = {3};
-    rocksdb_writebatch_wi_delete_rangev(wbi, 1, start_list, start_sizes, end_list,
-                                     end_sizes);
-    rocksdb_write_writebatch_wi(db, woptions, wbi, &err);
-    CheckNoError(err);
-    CheckGet(db, roptions, "bay", NULL);
     rocksdb_writebatch_wi_destroy(wbi);
   }
 
@@ -1048,17 +1034,20 @@ int main(int argc, char** argv) {
   }
 
   StartPhase("filter");
-  for (run = 0; run < 2; run++) {
-    // First run uses custom filter, second run uses bloom filter
+  for (run = 0; run <= 2; run++) {
+    // First run uses custom filter
+    // Second run uses old block-based bloom filter
+    // Third run uses full bloom filter
     CheckNoError(err);
     rocksdb_filterpolicy_t* policy;
     if (run == 0) {
-      policy = rocksdb_filterpolicy_create(
-          NULL, FilterDestroy, FilterCreate, FilterKeyMatch, NULL, FilterName);
+      policy = rocksdb_filterpolicy_create(NULL, FilterDestroy, FilterCreate,
+                                           FilterKeyMatch, NULL, FilterName);
+    } else if (run == 1) {
+      policy = rocksdb_filterpolicy_create_bloom(8);
     } else {
-      policy = rocksdb_filterpolicy_create_bloom(10);
+      policy = rocksdb_filterpolicy_create_bloom_full(8);
     }
-
     rocksdb_block_based_options_set_filter_policy(table_options, policy);
 
     // Create new database
@@ -1071,12 +1060,24 @@ int main(int argc, char** argv) {
     CheckNoError(err);
     rocksdb_put(db, woptions, "bar", 3, "barvalue", 8, &err);
     CheckNoError(err);
+
+    {
+      // Add enough keys to get just one reasonably populated Bloom filter
+      const int keys_to_add = 1500;
+      int i;
+      char keybuf[100];
+      for (i = 0; i < keys_to_add; i++) {
+        snprintf(keybuf, sizeof(keybuf), "yes%020d", i);
+        rocksdb_put(db, woptions, keybuf, strlen(keybuf), "val", 3, &err);
+        CheckNoError(err);
+      }
+    }
     rocksdb_compact_range(db, NULL, 0, NULL, 0);
 
     fake_filter_result = 1;
     CheckGet(db, roptions, "foo", "foovalue");
     CheckGet(db, roptions, "bar", "barvalue");
-    if (phase == 0) {
+    if (run == 0) {
       // Must not find value when custom filter returns false
       fake_filter_result = 0;
       CheckGet(db, roptions, "foo", NULL);
@@ -1086,6 +1087,43 @@ int main(int argc, char** argv) {
       CheckGet(db, roptions, "foo", "foovalue");
       CheckGet(db, roptions, "bar", "barvalue");
     }
+
+    {
+      // Query some keys not added to identify Bloom filter implementation
+      // from false positive queries, using perfcontext to detect Bloom
+      // filter behavior
+      rocksdb_perfcontext_t* perf = rocksdb_perfcontext_create();
+      rocksdb_perfcontext_reset(perf);
+
+      const int keys_to_query = 10000;
+      int i;
+      char keybuf[100];
+      for (i = 0; i < keys_to_query; i++) {
+        fake_filter_result = i % 2;
+        snprintf(keybuf, sizeof(keybuf), "no%020d", i);
+        CheckGet(db, roptions, keybuf, NULL);
+      }
+
+      const int hits =
+          (int)rocksdb_perfcontext_metric(perf, rocksdb_bloom_sst_hit_count);
+      if (run == 0) {
+        // Due to half true, half false with fake filter result
+        CheckCondition(hits == keys_to_query / 2);
+      } else if (run == 1) {
+        // Essentially a fingerprint of the block-based Bloom schema
+        CheckCondition(hits == 241);
+      } else {
+        // Essentially a fingerprint of the full Bloom schema(s),
+        // format_version < 5, which vary for three different CACHE_LINE_SIZEs
+        CheckCondition(hits == 224 || hits == 180 || hits == 125);
+      }
+      CheckCondition(
+          (keys_to_query - hits) ==
+          (int)rocksdb_perfcontext_metric(perf, rocksdb_bloom_sst_miss_count));
+
+      rocksdb_perfcontext_destroy(perf);
+    }
+
     // Reset the policy
     rocksdb_block_based_options_set_filter_policy(table_options, NULL);
     rocksdb_options_set_block_based_table_factory(options, table_options);
@@ -1189,6 +1227,15 @@ int main(int argc, char** argv) {
     rocksdb_put_cf(db, woptions, handles[1], "foo", 3, "hello", 5, &err);
     CheckNoError(err);
 
+    rocksdb_put_cf(db, woptions, handles[1], "foobar1", 7, "hello1", 6, &err);
+    CheckNoError(err);
+    rocksdb_put_cf(db, woptions, handles[1], "foobar2", 7, "hello2", 6, &err);
+    CheckNoError(err);
+    rocksdb_put_cf(db, woptions, handles[1], "foobar3", 7, "hello3", 6, &err);
+    CheckNoError(err);
+    rocksdb_put_cf(db, woptions, handles[1], "foobar4", 7, "hello4", 6, &err);
+    CheckNoError(err);
+
     rocksdb_flushoptions_t *flush_options = rocksdb_flushoptions_create();
     rocksdb_flushoptions_set_wait(flush_options, 1);
     rocksdb_flush_cf(db, flush_options, handles[1], &err);
@@ -1201,6 +1248,10 @@ int main(int argc, char** argv) {
     rocksdb_delete_cf(db, woptions, handles[1], "foo", 3, &err);
     CheckNoError(err);
 
+    rocksdb_delete_range_cf(db, woptions, handles[1], "foobar2", 7, "foobar4",
+                            7, &err);
+    CheckNoError(err);
+
     CheckGetCF(db, roptions, handles[1], "foo", NULL);
     CheckPinGetCF(db, roptions, handles[1], "foo", NULL);
 
@@ -1253,7 +1304,7 @@ int main(int argc, char** argv) {
     for (i = 0; rocksdb_iter_valid(iter) != 0; rocksdb_iter_next(iter)) {
       i++;
     }
-    CheckCondition(i == 1);
+    CheckCondition(i == 3);
     rocksdb_iter_get_error(iter, &err);
     CheckNoError(err);
     rocksdb_iter_destroy(iter);
@@ -1277,7 +1328,7 @@ int main(int argc, char** argv) {
     for (i = 0; rocksdb_iter_valid(iter) != 0; rocksdb_iter_next(iter)) {
       i++;
     }
-    CheckCondition(i == 1);
+    CheckCondition(i == 3);
     rocksdb_iter_get_error(iter, &err);
     CheckNoError(err);
     rocksdb_iter_destroy(iter);

db/column_family.cc

@@ -36,7 +36,7 @@
 #include "util/autovector.h"
 #include "util/compression.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 ColumnFamilyHandleImpl::ColumnFamilyHandleImpl(
     ColumnFamilyData* column_family_data, DBImpl* db, InstrumentedMutex* mutex)
@@ -60,11 +60,8 @@ ColumnFamilyHandleImpl::~ColumnFamilyHandleImpl() {
     ColumnFamilyOptions initial_cf_options_copy = cfd_->initial_cf_options();
     JobContext job_context(0);
     mutex_->Lock();
-    if (cfd_->Unref()) {
-      bool dropped = cfd_->IsDropped();
-
-      delete cfd_;
-
+    bool dropped = cfd_->IsDropped();
+    if (cfd_->UnrefAndTryDelete()) {
       if (dropped) {
         db_->FindObsoleteFiles(&job_context, false, true);
       }
@@ -439,13 +436,18 @@ void SuperVersion::Cleanup() {
     to_delete.push_back(m);
   }
   current->Unref();
+  if (cfd->Unref()) {
+    delete cfd;
+  }
 }
 
-void SuperVersion::Init(MemTable* new_mem, MemTableListVersion* new_imm,
-                        Version* new_current) {
+void SuperVersion::Init(ColumnFamilyData* new_cfd, MemTable* new_mem,
+                        MemTableListVersion* new_imm, Version* new_current) {
+  cfd = new_cfd;
   mem = new_mem;
   imm = new_imm;
   current = new_current;
+  cfd->Ref();
   mem->Ref();
   imm->Ref();
   current->Ref();
@@ -469,11 +471,22 @@ void SuperVersionUnrefHandle(void* ptr) {
 }
 }  // anonymous namespace
 
+std::vector<std::string> ColumnFamilyData::GetDbPaths() const {
+  std::vector<std::string> paths;
+  paths.reserve(ioptions_.cf_paths.size());
+  for (const DbPath& db_path : ioptions_.cf_paths) {
+    paths.emplace_back(db_path.path);
+  }
+  return paths;
+}
+
+const uint32_t ColumnFamilyData::kDummyColumnFamilyDataId = port::kMaxUint32;
+
 ColumnFamilyData::ColumnFamilyData(
     uint32_t id, const std::string& name, Version* _dummy_versions,
     Cache* _table_cache, WriteBufferManager* write_buffer_manager,
     const ColumnFamilyOptions& cf_options, const ImmutableDBOptions& db_options,
-    const EnvOptions& env_options, ColumnFamilySet* column_family_set,
+    const FileOptions& file_options, ColumnFamilySet* column_family_set,
     BlockCacheTracer* const block_cache_tracer)
     : id_(id),
       name_(name),
@@ -505,7 +518,23 @@ ColumnFamilyData::ColumnFamilyData(
       queued_for_compaction_(false),
       prev_compaction_needed_bytes_(0),
       allow_2pc_(db_options.allow_2pc),
-      last_memtable_id_(0) {
+      last_memtable_id_(0),
+      db_paths_registered_(false) {
+  if (id_ != kDummyColumnFamilyDataId) {
+    // TODO(cc): RegisterDbPaths can be expensive, considering moving it
+    // outside of this constructor which might be called with db mutex held.
+    // TODO(cc): considering using ioptions_.fs, currently some tests rely on
+    // EnvWrapper, that's the main reason why we use env here.
+    Status s = ioptions_.env->RegisterDbPaths(GetDbPaths());
+    if (s.ok()) {
+      db_paths_registered_ = true;
+    } else {
+      ROCKS_LOG_ERROR(
+          ioptions_.info_log,
+          "Failed to register data paths of column family (id: %d, name: %s)",
+          id_, name_.c_str());
+    }
+  }
   Ref();
 
   // Convert user defined table properties collector factories to internal ones.
@@ -515,7 +544,7 @@ ColumnFamilyData::ColumnFamilyData(
   if (_dummy_versions != nullptr) {
     internal_stats_.reset(
         new InternalStats(ioptions_.num_levels, db_options.env, this));
-    table_cache_.reset(new TableCache(ioptions_, env_options, _table_cache,
+    table_cache_.reset(new TableCache(ioptions_, file_options, _table_cache,
                                       block_cache_tracer));
     if (ioptions_.compaction_style == kCompactionStyleLevel) {
       compaction_picker_.reset(
@@ -581,21 +610,7 @@ ColumnFamilyData::~ColumnFamilyData() {
   // compaction_queue_ and we destroyed it
   assert(!queued_for_flush_);
   assert(!queued_for_compaction_);
-
-  if (super_version_ != nullptr) {
-    // Release SuperVersion reference kept in ThreadLocalPtr.
-    // This must be done outside of mutex_ since unref handler can lock mutex.
-    super_version_->db_mutex->Unlock();
-    local_sv_.reset();
-    super_version_->db_mutex->Lock();
-
-    bool is_last_reference __attribute__((__unused__));
-    is_last_reference = super_version_->Unref();
-    assert(is_last_reference);
-    super_version_->Cleanup();
-    delete super_version_;
-    super_version_ = nullptr;
-  }
+  assert(super_version_ == nullptr);
 
   if (dummy_versions_ != nullptr) {
     // List must be empty
@@ -613,6 +628,48 @@ ColumnFamilyData::~ColumnFamilyData() {
   for (MemTable* m : to_delete) {
     delete m;
   }
+
+  if (db_paths_registered_) {
+    // TODO(cc): considering using ioptions_.fs, currently some tests rely on
+    // EnvWrapper, that's the main reason why we use env here.
+    Status s = ioptions_.env->UnregisterDbPaths(GetDbPaths());
+    if (!s.ok()) {
+      ROCKS_LOG_ERROR(
+          ioptions_.info_log,
+          "Failed to unregister data paths of column family (id: %d, name: %s)",
+          id_, name_.c_str());
+    }
+  }
+}
+
+bool ColumnFamilyData::UnrefAndTryDelete() {
+  int old_refs = refs_.fetch_sub(1);
+  assert(old_refs > 0);
+
+  if (old_refs == 1) {
+    assert(super_version_ == nullptr);
+    delete this;
+    return true;
+  }
+
+  if (old_refs == 2 && super_version_ != nullptr) {
+    // Only the super_version_ holds me
+    SuperVersion* sv = super_version_;
+    super_version_ = nullptr;
+    // Release SuperVersion reference kept in ThreadLocalPtr.
+    // This must be done outside of mutex_ since unref handler can lock mutex.
+    sv->db_mutex->Unlock();
+    local_sv_.reset();
+    sv->db_mutex->Lock();
+
+    if (sv->Unref()) {
+      // May delete this ColumnFamilyData after calling Cleanup()
+      sv->Cleanup();
+      delete sv;
+      return true;
+    }
+  }
+  return false;
 }
 
 void ColumnFamilyData::SetDropped() {
@@ -949,8 +1006,8 @@ WriteStallCondition ColumnFamilyData::RecalculateWriteStallConditions(
   return write_stall_condition;
 }
 
-const EnvOptions* ColumnFamilyData::soptions() const {
-  return &(column_family_set_->env_options_);
+const FileOptions* ColumnFamilyData::soptions() const {
+  return &(column_family_set_->file_options_);
 }
 
 void ColumnFamilyData::SetCurrent(Version* current_version) {
@@ -1080,9 +1137,8 @@ Compaction* ColumnFamilyData::CompactRange(
   return result;
 }
 
-SuperVersion* ColumnFamilyData::GetReferencedSuperVersion(
-    InstrumentedMutex* db_mutex) {
-  SuperVersion* sv = GetThreadLocalSuperVersion(db_mutex);
+SuperVersion* ColumnFamilyData::GetReferencedSuperVersion(DBImpl* db) {
+  SuperVersion* sv = GetThreadLocalSuperVersion(db);
   sv->Ref();
   if (!ReturnThreadLocalSuperVersion(sv)) {
     // This Unref() corresponds to the Ref() in GetThreadLocalSuperVersion()
@@ -1094,8 +1150,7 @@ SuperVersion* ColumnFamilyData::GetReferencedSuperVersion(
   return sv;
 }
 
-SuperVersion* ColumnFamilyData::GetThreadLocalSuperVersion(
-    InstrumentedMutex* db_mutex) {
+SuperVersion* ColumnFamilyData::GetThreadLocalSuperVersion(DBImpl* db) {
   // The SuperVersion is cached in thread local storage to avoid acquiring
   // mutex when SuperVersion does not change since the last use. When a new
   // SuperVersion is installed, the compaction or flush thread cleans up
@@ -1122,16 +1177,21 @@ SuperVersion* ColumnFamilyData::GetThreadLocalSuperVersion(
 
     if (sv && sv->Unref()) {
       RecordTick(ioptions_.statistics, NUMBER_SUPERVERSION_CLEANUPS);
-      db_mutex->Lock();
+      db->mutex()->Lock();
       // NOTE: underlying resources held by superversion (sst files) might
       // not be released until the next background job.
       sv->Cleanup();
-      sv_to_delete = sv;
+      if (db->immutable_db_options().avoid_unnecessary_blocking_io) {
+        db->AddSuperVersionsToFreeQueue(sv);
+        db->SchedulePurge();
+      } else {
+        sv_to_delete = sv;
+      }
     } else {
-      db_mutex->Lock();
+      db->mutex()->Lock();
     }
     sv = super_version_->Ref();
-    db_mutex->Unlock();
+    db->mutex()->Unlock();
 
     delete sv_to_delete;
   }
@@ -1169,7 +1229,7 @@ void ColumnFamilyData::InstallSuperVersion(
   SuperVersion* new_superversion = sv_context->new_superversion.release();
   new_superversion->db_mutex = db_mutex;
   new_superversion->mutable_cf_options = mutable_cf_options;
-  new_superversion->Init(mem_, imm_.current(), current_);
+  new_superversion->Init(this, mem_, imm_.current(), current_);
   SuperVersion* old_superversion = super_version_;
   super_version_ = new_superversion;
   ++super_version_number_;
@@ -1226,6 +1286,11 @@ Status ColumnFamilyData::ValidateOptions(
   if (s.ok() && db_options.allow_concurrent_memtable_write) {
     s = CheckConcurrentWritesSupported(cf_options);
   }
+  if (s.ok() && db_options.unordered_write &&
+      cf_options.max_successive_merges != 0) {
+    s = Status::InvalidArgument(
+        "max_successive_merges > 0 is incompatible with unordered_write");
+  }
   if (s.ok()) {
     s = CheckCFPathsSupported(db_options, cf_options);
   }
@@ -1285,28 +1350,41 @@ Env::WriteLifeTimeHint ColumnFamilyData::CalculateSSTWriteHint(int level) {
   // L1: medium, L2: long, ...
   if (level - base_level >= 2) {
     return Env::WLTH_EXTREME;
+  } else if (level < base_level) {
+    // There is no restriction which prevents level passed in to be smaller
+    // than base_level.
+    return Env::WLTH_MEDIUM;
   }
   return static_cast<Env::WriteLifeTimeHint>(level - base_level +
                             static_cast<int>(Env::WLTH_MEDIUM));
 }
 
-Status ColumnFamilyData::AddDirectories() {
+Status ColumnFamilyData::AddDirectories(
+    std::map<std::string, std::shared_ptr<FSDirectory>>* created_dirs) {
   Status s;
+  assert(created_dirs != nullptr);
   assert(data_dirs_.empty());
   for (auto& p : ioptions_.cf_paths) {
-    std::unique_ptr<Directory> path_directory;
-    s = DBImpl::CreateAndNewDirectory(ioptions_.env, p.path, &path_directory);
-    if (!s.ok()) {
-      return s;
+    auto existing_dir = created_dirs->find(p.path);
+
+    if (existing_dir == created_dirs->end()) {
+      std::unique_ptr<FSDirectory> path_directory;
+      s = DBImpl::CreateAndNewDirectory(ioptions_.fs, p.path, &path_directory);
+      if (!s.ok()) {
+        return s;
+      }
+      assert(path_directory != nullptr);
+      data_dirs_.emplace_back(path_directory.release());
+      (*created_dirs)[p.path] = data_dirs_.back();
+    } else {
+      data_dirs_.emplace_back(existing_dir->second);
     }
-    assert(path_directory != nullptr);
-    data_dirs_.emplace_back(path_directory.release());
   }
   assert(data_dirs_.size() == ioptions_.cf_paths.size());
   return s;
 }
 
-Directory* ColumnFamilyData::GetDataDir(size_t path_id) const {
+FSDirectory* ColumnFamilyData::GetDataDir(size_t path_id) const {
   if (data_dirs_.empty()) {
     return nullptr;
   }
@@ -1317,19 +1395,20 @@ Directory* ColumnFamilyData::GetDataDir(size_t path_id) const {
 
 ColumnFamilySet::ColumnFamilySet(const std::string& dbname,
                                  const ImmutableDBOptions* db_options,
-                                 const EnvOptions& env_options,
+                                 const FileOptions& file_options,
                                  Cache* table_cache,
                                  WriteBufferManager* write_buffer_manager,
                                  WriteController* write_controller,
                                  BlockCacheTracer* const block_cache_tracer)
     : max_column_family_(0),
       dummy_cfd_(new ColumnFamilyData(
-          0, "", nullptr, nullptr, nullptr, ColumnFamilyOptions(), *db_options,
-          env_options, nullptr, block_cache_tracer)),
+          ColumnFamilyData::kDummyColumnFamilyDataId, "", nullptr, nullptr,
+          nullptr, ColumnFamilyOptions(), *db_options, file_options, nullptr,
+          block_cache_tracer)),
       default_cfd_cache_(nullptr),
       db_name_(dbname),
       db_options_(db_options),
-      env_options_(env_options),
+      file_options_(file_options),
       table_cache_(table_cache),
       write_buffer_manager_(write_buffer_manager),
       write_controller_(write_controller),
@@ -1344,14 +1423,12 @@ ColumnFamilySet::~ColumnFamilySet() {
     // cfd destructor will delete itself from column_family_data_
     auto cfd = column_family_data_.begin()->second;
     bool last_ref __attribute__((__unused__));
-    last_ref = cfd->Unref();
+    last_ref = cfd->UnrefAndTryDelete();
     assert(last_ref);
-    delete cfd;
   }
   bool dummy_last_ref __attribute__((__unused__));
-  dummy_last_ref = dummy_cfd_->Unref();
+  dummy_last_ref = dummy_cfd_->UnrefAndTryDelete();
   assert(dummy_last_ref);
-  delete dummy_cfd_;
 }
 
 ColumnFamilyData* ColumnFamilySet::GetDefault() const {
@@ -1401,7 +1478,7 @@ ColumnFamilyData* ColumnFamilySet::CreateColumnFamily(
   assert(column_families_.find(name) == column_families_.end());
   ColumnFamilyData* new_cfd = new ColumnFamilyData(
       id, name, dummy_versions, table_cache_, write_buffer_manager_, options,
-      *db_options_, env_options_, this, block_cache_tracer_);
+      *db_options_, file_options_, this, block_cache_tracer_);
   column_families_.insert({name, id});
   column_family_data_.insert({id, new_cfd});
   max_column_family_ = std::max(max_column_family_, id);
@@ -1483,4 +1560,4 @@ const Comparator* GetColumnFamilyUserComparator(
   return nullptr;
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/column_family.h

@@ -27,7 +27,7 @@
 #include "trace_replay/block_cache_tracer.h"
 #include "util/thread_local.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class Version;
 class VersionSet;
@@ -198,6 +198,7 @@ class ColumnFamilyHandleInternal : public ColumnFamilyHandleImpl {
 struct SuperVersion {
   // Accessing members of this class is not thread-safe and requires external
   // synchronization (ie db mutex held or on write thread).
+  ColumnFamilyData* cfd;
   MemTable* mem;
   MemTableListVersion* imm;
   Version* current;
@@ -221,8 +222,8 @@ struct SuperVersion {
   // that needs to be deleted in to_delete vector. Unrefing those
   // objects needs to be done in the mutex
   void Cleanup();
-  void Init(MemTable* new_mem, MemTableListVersion* new_imm,
-            Version* new_current);
+  void Init(ColumnFamilyData* new_cfd, MemTable* new_mem,
+            MemTableListVersion* new_imm, Version* new_current);
 
   // The value of dummy is not actually used. kSVInUse takes its address as a
   // mark in the thread local storage to indicate the SuperVersion is in use
@@ -288,6 +289,11 @@ class ColumnFamilyData {
     return old_refs == 1;
   }
 
+  // UnrefAndTryDelete() decreases the reference count and do free if needed,
+  // return true if this is freed else false, UnrefAndTryDelete() can only
+  // be called while holding a DB mutex, or during single-threaded recovery.
+  bool UnrefAndTryDelete();
+
   // SetDropped() can only be called under following conditions:
   // 1) Holding a DB mutex,
   // 2) from single-threaded write thread, AND
@@ -318,7 +324,7 @@ class ColumnFamilyData {
   }
   FlushReason GetFlushReason() const { return flush_reason_; }
   // thread-safe
-  const EnvOptions* soptions() const;
+  const FileOptions* soptions() const;
   const ImmutableCFOptions* ioptions() const { return &ioptions_; }
   // REQUIRES: DB mutex held
   // This returns the MutableCFOptions used by current SuperVersion
@@ -430,11 +436,11 @@ class ColumnFamilyData {
   SuperVersion* GetSuperVersion() { return super_version_; }
   // thread-safe
   // Return a already referenced SuperVersion to be used safely.
-  SuperVersion* GetReferencedSuperVersion(InstrumentedMutex* db_mutex);
+  SuperVersion* GetReferencedSuperVersion(DBImpl* db);
   // thread-safe
   // Get SuperVersion stored in thread local storage. If it does not exist,
   // get a reference from a current SuperVersion.
-  SuperVersion* GetThreadLocalSuperVersion(InstrumentedMutex* db_mutex);
+  SuperVersion* GetThreadLocalSuperVersion(DBImpl* db);
   // Try to return SuperVersion back to thread local storage. Retrun true on
   // success and false on failure. It fails when the thread local storage
   // contains anything other than SuperVersion::kSVInUse flag.
@@ -491,23 +497,29 @@ class ColumnFamilyData {
 
   Env::WriteLifeTimeHint CalculateSSTWriteHint(int level);
 
-  Status AddDirectories();
+  // created_dirs remembers directory created, so that we don't need to call
+  // the same data creation operation again.
+  Status AddDirectories(
+      std::map<std::string, std::shared_ptr<FSDirectory>>* created_dirs);
 
-  Directory* GetDataDir(size_t path_id) const;
+  FSDirectory* GetDataDir(size_t path_id) const;
 
   ThreadLocalPtr* TEST_GetLocalSV() { return local_sv_.get(); }
 
  private:
   friend class ColumnFamilySet;
+  static const uint32_t kDummyColumnFamilyDataId;
   ColumnFamilyData(uint32_t id, const std::string& name,
                    Version* dummy_versions, Cache* table_cache,
                    WriteBufferManager* write_buffer_manager,
                    const ColumnFamilyOptions& options,
                    const ImmutableDBOptions& db_options,
-                   const EnvOptions& env_options,
+                   const FileOptions& file_options,
                    ColumnFamilySet* column_family_set,
                    BlockCacheTracer* const block_cache_tracer);
 
+  std::vector<std::string> GetDbPaths() const;
+
   uint32_t id_;
   const std::string name_;
   Version* dummy_versions_;  // Head of circular doubly-linked list of versions.
@@ -583,7 +595,9 @@ class ColumnFamilyData {
   std::atomic<uint64_t> last_memtable_id_;
 
   // Directories corresponding to cf_paths.
-  std::vector<std::unique_ptr<Directory>> data_dirs_;
+  std::vector<std::shared_ptr<FSDirectory>> data_dirs_;
+
+  bool db_paths_registered_;
 };
 
 // ColumnFamilySet has interesting thread-safety requirements
@@ -632,7 +646,7 @@ class ColumnFamilySet {
 
   ColumnFamilySet(const std::string& dbname,
                   const ImmutableDBOptions* db_options,
-                  const EnvOptions& env_options, Cache* table_cache,
+                  const FileOptions& file_options, Cache* table_cache,
                   WriteBufferManager* write_buffer_manager,
                   WriteController* write_controller,
                   BlockCacheTracer* const block_cache_tracer);
@@ -690,7 +704,7 @@ class ColumnFamilySet {
 
   const std::string db_name_;
   const ImmutableDBOptions* const db_options_;
-  const EnvOptions env_options_;
+  const FileOptions file_options_;
   Cache* table_cache_;
   WriteBufferManager* write_buffer_manager_;
   WriteController* write_controller_;
@@ -745,4 +759,4 @@ extern uint32_t GetColumnFamilyID(ColumnFamilyHandle* column_family);
 extern const Comparator* GetColumnFamilyUserComparator(
     ColumnFamilyHandle* column_family);
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/column_family_test.cc

@@ -30,7 +30,7 @@
 #include "util/string_util.h"
 #include "utilities/merge_operators.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 static const int kValueSize = 1000;
 
@@ -67,9 +67,9 @@ class ColumnFamilyTestBase : public testing::Test {
 #ifndef ROCKSDB_LITE
     const char* test_env_uri = getenv("TEST_ENV_URI");
     if (test_env_uri) {
-      Status s = ObjectRegistry::NewInstance()->NewSharedObject(test_env_uri,
-                                                                &env_guard_);
-      base_env = env_guard_.get();
+      Env* test_env = nullptr;
+      Status s = Env::LoadEnv(test_env_uri, &test_env, &env_guard_);
+      base_env = test_env;
       EXPECT_OK(s);
       EXPECT_NE(Env::Default(), base_env);
     }
@@ -91,7 +91,7 @@ class ColumnFamilyTestBase : public testing::Test {
       column_families.push_back(cfdescriptor);
     }
     Close();
-    rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
     Destroy(column_families);
     delete env_;
   }
@@ -595,14 +595,15 @@ TEST_P(ColumnFamilyTest, DontReuseColumnFamilyID) {
 TEST_P(ColumnFamilyTest, CreateCFRaceWithGetAggProperty) {
   Open();
 
-  rocksdb::SyncPoint::GetInstance()->LoadDependency(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency(
       {{"DBImpl::WriteOptionsFile:1",
         "ColumnFamilyTest.CreateCFRaceWithGetAggProperty:1"},
        {"ColumnFamilyTest.CreateCFRaceWithGetAggProperty:2",
         "DBImpl::WriteOptionsFile:2"}});
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
-  rocksdb::port::Thread thread([&] { CreateColumnFamilies({"one"}); });
+  ROCKSDB_NAMESPACE::port::Thread thread(
+      [&] { CreateColumnFamilies({"one"}); });
 
   TEST_SYNC_POINT("ColumnFamilyTest.CreateCFRaceWithGetAggProperty:1");
   uint64_t pv;
@@ -611,7 +612,7 @@ TEST_P(ColumnFamilyTest, CreateCFRaceWithGetAggProperty) {
 
   thread.join();
 
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
 }
 #endif  // !ROCKSDB_LITE
 
@@ -820,7 +821,7 @@ TEST_P(ColumnFamilyTest, BulkAddDrop) {
 }
 
 TEST_P(ColumnFamilyTest, DropTest) {
-  // first iteration - dont reopen DB before dropping
+  // first iteration - don't reopen DB before dropping
   // second iteration - reopen DB before dropping
   for (int iter = 0; iter < 2; ++iter) {
     Open({"default"});
@@ -1253,13 +1254,14 @@ TEST_P(ColumnFamilyTest, DifferentWriteBufferSizes) {
 // #endif  // !ROCKSDB_LITE
 
 class TestComparator : public Comparator {
-  int Compare(const rocksdb::Slice& /*a*/,
-              const rocksdb::Slice& /*b*/) const override {
+  int Compare(const ROCKSDB_NAMESPACE::Slice& /*a*/,
+              const ROCKSDB_NAMESPACE::Slice& /*b*/) const override {
     return 0;
   }
   const char* Name() const override { return "Test"; }
-  void FindShortestSeparator(std::string* /*start*/,
-                             const rocksdb::Slice& /*limit*/) const override {}
+  void FindShortestSeparator(
+      std::string* /*start*/,
+      const ROCKSDB_NAMESPACE::Slice& /*limit*/) const override {}
   void FindShortSuccessor(std::string* /*key*/) const override {}
 };
 
@@ -1427,11 +1429,11 @@ TEST_P(ColumnFamilyTest, MultipleManualCompactions) {
     AssertFilesPerLevel(ToString(i + 1), 1);
   }
   bool cf_1_1 = true;
-  rocksdb::SyncPoint::GetInstance()->LoadDependency(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency(
       {{"ColumnFamilyTest::MultiManual:4", "ColumnFamilyTest::MultiManual:1"},
        {"ColumnFamilyTest::MultiManual:2", "ColumnFamilyTest::MultiManual:5"},
        {"ColumnFamilyTest::MultiManual:2", "ColumnFamilyTest::MultiManual:3"}});
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* /*arg*/) {
         if (cf_1_1) {
           TEST_SYNC_POINT("ColumnFamilyTest::MultiManual:4");
@@ -1440,7 +1442,7 @@ TEST_P(ColumnFamilyTest, MultipleManualCompactions) {
         }
       });
 
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
   std::vector<port::Thread> threads;
   threads.emplace_back([&] {
     CompactRangeOptions compact_options;
@@ -1483,8 +1485,8 @@ TEST_P(ColumnFamilyTest, MultipleManualCompactions) {
     ASSERT_NE("NOT_FOUND", Get(1, *key_iter));
     key_iter++;
   }
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
-  rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
   Close();
 }
 
@@ -1523,11 +1525,11 @@ TEST_P(ColumnFamilyTest, AutomaticAndManualCompactions) {
       dbfull()->TEST_write_controler().GetCompactionPressureToken();
 
   bool cf_1_1 = true;
-  rocksdb::SyncPoint::GetInstance()->LoadDependency(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency(
       {{"ColumnFamilyTest::AutoManual:4", "ColumnFamilyTest::AutoManual:1"},
        {"ColumnFamilyTest::AutoManual:2", "ColumnFamilyTest::AutoManual:5"},
        {"ColumnFamilyTest::AutoManual:2", "ColumnFamilyTest::AutoManual:3"}});
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* /*arg*/) {
         if (cf_1_1) {
           cf_1_1 = false;
@@ -1535,7 +1537,7 @@ TEST_P(ColumnFamilyTest, AutomaticAndManualCompactions) {
           TEST_SYNC_POINT("ColumnFamilyTest::AutoManual:3");
         }
       });
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
   // SETUP column family "one" -- universal style
   for (int i = 0; i < one.level0_file_num_compaction_trigger; ++i) {
     PutRandomData(1, 10, 12000, true);
@@ -1553,7 +1555,7 @@ TEST_P(ColumnFamilyTest, AutomaticAndManualCompactions) {
     WaitForFlush(2);
     AssertFilesPerLevel(ToString(i + 1), 2);
   }
-  rocksdb::port::Thread threads([&] {
+  ROCKSDB_NAMESPACE::port::Thread threads([&] {
     CompactRangeOptions compact_options;
     compact_options.exclusive_manual_compaction = false;
     ASSERT_OK(
@@ -1580,8 +1582,8 @@ TEST_P(ColumnFamilyTest, AutomaticAndManualCompactions) {
     ASSERT_NE("NOT_FOUND", Get(1, *key_iter));
     key_iter++;
   }
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
-  rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
 }
 
 TEST_P(ColumnFamilyTest, ManualAndAutomaticCompactions) {
@@ -1627,11 +1629,11 @@ TEST_P(ColumnFamilyTest, ManualAndAutomaticCompactions) {
   }
   bool cf_1_1 = true;
   bool cf_1_2 = true;
-  rocksdb::SyncPoint::GetInstance()->LoadDependency(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency(
       {{"ColumnFamilyTest::ManualAuto:4", "ColumnFamilyTest::ManualAuto:1"},
        {"ColumnFamilyTest::ManualAuto:5", "ColumnFamilyTest::ManualAuto:2"},
        {"ColumnFamilyTest::ManualAuto:2", "ColumnFamilyTest::ManualAuto:3"}});
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* /*arg*/) {
         if (cf_1_1) {
           TEST_SYNC_POINT("ColumnFamilyTest::ManualAuto:4");
@@ -1643,8 +1645,8 @@ TEST_P(ColumnFamilyTest, ManualAndAutomaticCompactions) {
         }
       });
 
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
-  rocksdb::port::Thread threads([&] {
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::port::Thread threads([&] {
     CompactRangeOptions compact_options;
     compact_options.exclusive_manual_compaction = false;
     ASSERT_OK(
@@ -1679,8 +1681,8 @@ TEST_P(ColumnFamilyTest, ManualAndAutomaticCompactions) {
     ASSERT_NE("NOT_FOUND", Get(1, *key_iter));
     key_iter++;
   }
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
-  rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
 }
 
 TEST_P(ColumnFamilyTest, SameCFManualManualCompactions) {
@@ -1721,13 +1723,13 @@ TEST_P(ColumnFamilyTest, SameCFManualManualCompactions) {
   }
   bool cf_1_1 = true;
   bool cf_1_2 = true;
-  rocksdb::SyncPoint::GetInstance()->LoadDependency(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency(
       {{"ColumnFamilyTest::ManualManual:4", "ColumnFamilyTest::ManualManual:2"},
        {"ColumnFamilyTest::ManualManual:4", "ColumnFamilyTest::ManualManual:5"},
        {"ColumnFamilyTest::ManualManual:1", "ColumnFamilyTest::ManualManual:2"},
        {"ColumnFamilyTest::ManualManual:1",
         "ColumnFamilyTest::ManualManual:3"}});
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* /*arg*/) {
         if (cf_1_1) {
           TEST_SYNC_POINT("ColumnFamilyTest::ManualManual:4");
@@ -1739,8 +1741,8 @@ TEST_P(ColumnFamilyTest, SameCFManualManualCompactions) {
         }
       });
 
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
-  rocksdb::port::Thread threads([&] {
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::port::Thread threads([&] {
     CompactRangeOptions compact_options;
     compact_options.exclusive_manual_compaction = true;
     ASSERT_OK(
@@ -1760,7 +1762,7 @@ TEST_P(ColumnFamilyTest, SameCFManualManualCompactions) {
                         1);
   }
 
-  rocksdb::port::Thread threads1([&] {
+  ROCKSDB_NAMESPACE::port::Thread threads1([&] {
     CompactRangeOptions compact_options;
     compact_options.exclusive_manual_compaction = false;
     ASSERT_OK(
@@ -1781,8 +1783,8 @@ TEST_P(ColumnFamilyTest, SameCFManualManualCompactions) {
     ASSERT_NE("NOT_FOUND", Get(1, *key_iter));
     key_iter++;
   }
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
-  rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
 }
 
 TEST_P(ColumnFamilyTest, SameCFManualAutomaticCompactions) {
@@ -1823,12 +1825,12 @@ TEST_P(ColumnFamilyTest, SameCFManualAutomaticCompactions) {
   }
   bool cf_1_1 = true;
   bool cf_1_2 = true;
-  rocksdb::SyncPoint::GetInstance()->LoadDependency(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency(
       {{"ColumnFamilyTest::ManualAuto:4", "ColumnFamilyTest::ManualAuto:2"},
        {"ColumnFamilyTest::ManualAuto:4", "ColumnFamilyTest::ManualAuto:5"},
        {"ColumnFamilyTest::ManualAuto:1", "ColumnFamilyTest::ManualAuto:2"},
        {"ColumnFamilyTest::ManualAuto:1", "ColumnFamilyTest::ManualAuto:3"}});
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* /*arg*/) {
         if (cf_1_1) {
           TEST_SYNC_POINT("ColumnFamilyTest::ManualAuto:4");
@@ -1840,8 +1842,8 @@ TEST_P(ColumnFamilyTest, SameCFManualAutomaticCompactions) {
         }
       });
 
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
-  rocksdb::port::Thread threads([&] {
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::port::Thread threads([&] {
     CompactRangeOptions compact_options;
     compact_options.exclusive_manual_compaction = false;
     ASSERT_OK(
@@ -1874,8 +1876,8 @@ TEST_P(ColumnFamilyTest, SameCFManualAutomaticCompactions) {
     ASSERT_NE("NOT_FOUND", Get(1, *key_iter));
     key_iter++;
   }
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
-  rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
 }
 
 TEST_P(ColumnFamilyTest, SameCFManualAutomaticCompactionsLevel) {
@@ -1916,14 +1918,14 @@ TEST_P(ColumnFamilyTest, SameCFManualAutomaticCompactionsLevel) {
   }
   bool cf_1_1 = true;
   bool cf_1_2 = true;
-  rocksdb::SyncPoint::GetInstance()->LoadDependency(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency(
       {{"ColumnFamilyTest::ManualAuto:4", "ColumnFamilyTest::ManualAuto:2"},
        {"ColumnFamilyTest::ManualAuto:4", "ColumnFamilyTest::ManualAuto:5"},
        {"ColumnFamilyTest::ManualAuto:3", "ColumnFamilyTest::ManualAuto:2"},
        {"LevelCompactionPicker::PickCompactionBySize:0",
         "ColumnFamilyTest::ManualAuto:3"},
        {"ColumnFamilyTest::ManualAuto:1", "ColumnFamilyTest::ManualAuto:3"}});
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* /*arg*/) {
         if (cf_1_1) {
           TEST_SYNC_POINT("ColumnFamilyTest::ManualAuto:4");
@@ -1935,8 +1937,8 @@ TEST_P(ColumnFamilyTest, SameCFManualAutomaticCompactionsLevel) {
         }
       });
 
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
-  rocksdb::port::Thread threads([&] {
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::port::Thread threads([&] {
     CompactRangeOptions compact_options;
     compact_options.exclusive_manual_compaction = false;
     ASSERT_OK(
@@ -1967,8 +1969,8 @@ TEST_P(ColumnFamilyTest, SameCFManualAutomaticCompactionsLevel) {
     ASSERT_NE("NOT_FOUND", Get(1, *key_iter));
     key_iter++;
   }
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
-  rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
 }
 
 // In this test, we generate enough files to trigger automatic compactions.
@@ -2008,12 +2010,12 @@ TEST_P(ColumnFamilyTest, SameCFAutomaticManualCompactions) {
 
   bool cf_1_1 = true;
   bool cf_1_2 = true;
-  rocksdb::SyncPoint::GetInstance()->LoadDependency(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency(
       {{"ColumnFamilyTest::AutoManual:4", "ColumnFamilyTest::AutoManual:2"},
        {"ColumnFamilyTest::AutoManual:4", "ColumnFamilyTest::AutoManual:5"},
        {"CompactionPicker::CompactRange:Conflict",
         "ColumnFamilyTest::AutoManual:3"}});
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* /*arg*/) {
         if (cf_1_1) {
           TEST_SYNC_POINT("ColumnFamilyTest::AutoManual:4");
@@ -2025,7 +2027,7 @@ TEST_P(ColumnFamilyTest, SameCFAutomaticManualCompactions) {
         }
       });
 
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   // SETUP column family "one" -- universal style
   for (int i = 0; i < one.level0_file_num_compaction_trigger; ++i) {
@@ -2059,8 +2061,8 @@ TEST_P(ColumnFamilyTest, SameCFAutomaticManualCompactions) {
     ASSERT_NE("NOT_FOUND", Get(1, *key_iter));
     key_iter++;
   }
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
-  rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
 }
 #endif  // !ROCKSDB_LITE
 
@@ -2365,6 +2367,45 @@ TEST_P(ColumnFamilyTest, ReadDroppedColumnFamily) {
   }
 }
 
+TEST_P(ColumnFamilyTest, LiveIteratorWithDroppedColumnFamily) {
+  db_options_.create_missing_column_families = true;
+  db_options_.max_open_files = 20;
+  // delete obsolete files always
+  db_options_.delete_obsolete_files_period_micros = 0;
+  Open({"default", "one", "two"});
+  ColumnFamilyOptions options;
+  options.level0_file_num_compaction_trigger = 100;
+  options.level0_slowdown_writes_trigger = 200;
+  options.level0_stop_writes_trigger = 200;
+  options.write_buffer_size = 100000;  // small write buffer size
+  Reopen({options, options, options});
+
+  // 1MB should create ~10 files for each CF
+  int kKeysNum = 10000;
+  PutRandomData(1, kKeysNum, 100);
+
+  {
+    std::unique_ptr<Iterator> iterator(
+        db_->NewIterator(ReadOptions(), handles_[1]));
+    iterator->SeekToFirst();
+
+    DropColumnFamilies({1});
+
+    // Make sure iterator created can still be used.
+    int count = 0;
+    for (; iterator->Valid(); iterator->Next()) {
+      ASSERT_OK(iterator->status());
+      ++count;
+    }
+    ASSERT_OK(iterator->status());
+    ASSERT_EQ(count, kKeysNum);
+  }
+
+  Reopen();
+  Close();
+  Destroy();
+}
+
 TEST_P(ColumnFamilyTest, FlushAndDropRaceCondition) {
   db_options_.create_missing_column_families = true;
   Open({"default", "one"});
@@ -2376,7 +2417,7 @@ TEST_P(ColumnFamilyTest, FlushAndDropRaceCondition) {
   options.write_buffer_size = 100000;  // small write buffer size
   Reopen({options, options});
 
-  rocksdb::SyncPoint::GetInstance()->LoadDependency(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency(
       {{"VersionSet::LogAndApply::ColumnFamilyDrop:0",
         "FlushJob::WriteLevel0Table"},
        {"VersionSet::LogAndApply::ColumnFamilyDrop:1",
@@ -2384,7 +2425,7 @@ TEST_P(ColumnFamilyTest, FlushAndDropRaceCondition) {
        {"FlushJob::InstallResults",
         "VersionSet::LogAndApply::ColumnFamilyDrop:2"}});
 
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
   test::SleepingBackgroundTask sleeping_task;
 
   env_->Schedule(&test::SleepingBackgroundTask::DoSleepTask, &sleeping_task,
@@ -2463,21 +2504,21 @@ TEST_P(ColumnFamilyTest, CreateAndDropRace) {
 
   auto main_thread_id = std::this_thread::get_id();
 
-  rocksdb::SyncPoint::GetInstance()->SetCallBack("PersistRocksDBOptions:start",
-                                                 [&](void* /*arg*/) {
-    auto current_thread_id = std::this_thread::get_id();
-    // If it's the main thread hitting this sync-point, then it
-    // will be blocked until some other thread update the test_stage.
-    if (main_thread_id == current_thread_id) {
-      test_stage = kMainThreadStartPersistingOptionsFile;
-      while (test_stage < kChildThreadFinishDroppingColumnFamily &&
-             !ordered_by_writethread) {
-        Env::Default()->SleepForMicroseconds(100);
-      }
-    }
-  });
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
+      "PersistRocksDBOptions:start", [&](void* /*arg*/) {
+        auto current_thread_id = std::this_thread::get_id();
+        // If it's the main thread hitting this sync-point, then it
+        // will be blocked until some other thread update the test_stage.
+        if (main_thread_id == current_thread_id) {
+          test_stage = kMainThreadStartPersistingOptionsFile;
+          while (test_stage < kChildThreadFinishDroppingColumnFamily &&
+                 !ordered_by_writethread) {
+            Env::Default()->SleepForMicroseconds(100);
+          }
+        }
+      });
 
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "WriteThread::EnterUnbatched:Wait", [&](void* /*arg*/) {
         // This means a thread doing DropColumnFamily() is waiting for
         // other thread to finish persisting options.
@@ -2489,12 +2530,12 @@ TEST_P(ColumnFamilyTest, CreateAndDropRace) {
   Open({"default", "one", "two", "three"},
        {cf_opts[0], cf_opts[1], cf_opts[2], cf_opts[3]});
 
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   // Start a thread that will drop the first column family
   // and its comparator
-  rocksdb::port::Thread drop_cf_thread(DropSingleColumnFamily, this, 1,
-                                       &comparators);
+  ROCKSDB_NAMESPACE::port::Thread drop_cf_thread(DropSingleColumnFamily, this,
+                                                 1, &comparators);
 
   DropColumnFamilies({2});
 
@@ -2507,8 +2548,8 @@ TEST_P(ColumnFamilyTest, CreateAndDropRace) {
     }
   }
 
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
-  rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
 }
 #endif  // !ROCKSDB_LITE
 
@@ -2943,10 +2984,10 @@ TEST_P(ColumnFamilyTest, FlushCloseWALFiles) {
   ASSERT_OK(Put(0, "fodor", "mirko"));
   ASSERT_OK(Put(1, "fodor", "mirko"));
 
-  rocksdb::SyncPoint::GetInstance()->LoadDependency({
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency({
       {"DBImpl::BGWorkFlush:done", "FlushCloseWALFiles:0"},
   });
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   // Block flush jobs from running
   test::SleepingBackgroundTask sleeping_task;
@@ -2962,7 +3003,7 @@ TEST_P(ColumnFamilyTest, FlushCloseWALFiles) {
   sleeping_task.WakeUp();
   sleeping_task.WaitUntilDone();
   TEST_SYNC_POINT("FlushCloseWALFiles:0");
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
   ASSERT_EQ(1, env.num_open_wal_file_.load());
 
   Reopen();
@@ -3040,14 +3081,14 @@ TEST_P(ColumnFamilyTest, IteratorCloseWALFile2) {
   ASSERT_OK(Put(0, "fodor", "mirko"));
   ASSERT_OK(Put(1, "fodor", "mirko"));
 
-  rocksdb::SyncPoint::GetInstance()->LoadDependency({
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency({
       {"ColumnFamilyTest::IteratorCloseWALFile2:0",
        "DBImpl::BGWorkPurge:start"},
       {"ColumnFamilyTest::IteratorCloseWALFile2:2",
        "DBImpl::BackgroundCallFlush:start"},
       {"DBImpl::BGWorkPurge:end", "ColumnFamilyTest::IteratorCloseWALFile2:1"},
   });
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   WriteOptions wo;
   wo.sync = true;
@@ -3065,7 +3106,7 @@ TEST_P(ColumnFamilyTest, IteratorCloseWALFile2) {
   TEST_SYNC_POINT("ColumnFamilyTest::IteratorCloseWALFile2:2");
   WaitForFlush(1);
   ASSERT_EQ(1, env.num_open_wal_file_.load());
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
 
   Reopen();
   ASSERT_EQ("mirko", Get(0, "fodor"));
@@ -3109,14 +3150,14 @@ TEST_P(ColumnFamilyTest, ForwardIteratorCloseWALFile) {
   ASSERT_OK(Put(0, "fodor", "mirko"));
   ASSERT_OK(Put(1, "fodor", "mirko"));
 
-  rocksdb::SyncPoint::GetInstance()->LoadDependency({
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency({
       {"ColumnFamilyTest::IteratorCloseWALFile2:0",
        "DBImpl::BGWorkPurge:start"},
       {"ColumnFamilyTest::IteratorCloseWALFile2:2",
        "DBImpl::BackgroundCallFlush:start"},
       {"DBImpl::BGWorkPurge:end", "ColumnFamilyTest::IteratorCloseWALFile2:1"},
   });
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   WriteOptions wo;
   wo.sync = true;
@@ -3140,7 +3181,7 @@ TEST_P(ColumnFamilyTest, ForwardIteratorCloseWALFile) {
   ASSERT_EQ(1, env.delete_count_.load());
 
   delete it;
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
 
   Reopen();
   ASSERT_EQ("mirko", Get(0, "fodor"));
@@ -3160,15 +3201,15 @@ TEST_P(ColumnFamilyTest, LogSyncConflictFlush) {
   Put(0, "", "");
   Put(1, "foo", "bar");
 
-  rocksdb::SyncPoint::GetInstance()->LoadDependency(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency(
       {{"DBImpl::SyncWAL:BeforeMarkLogsSynced:1",
         "ColumnFamilyTest::LogSyncConflictFlush:1"},
        {"ColumnFamilyTest::LogSyncConflictFlush:2",
         "DBImpl::SyncWAL:BeforeMarkLogsSynced:2"}});
 
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
-  rocksdb::port::Thread thread([&] { db_->SyncWAL(); });
+  ROCKSDB_NAMESPACE::port::Thread thread([&] { db_->SyncWAL(); });
 
   TEST_SYNC_POINT("ColumnFamilyTest::LogSyncConflictFlush:1");
   Flush(1);
@@ -3179,7 +3220,7 @@ TEST_P(ColumnFamilyTest, LogSyncConflictFlush) {
 
   thread.join();
 
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
   Close();
 }
 #endif
@@ -3328,7 +3369,7 @@ TEST_P(ColumnFamilyTest, MultipleCFPathsTest) {
   }
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 #ifdef ROCKSDB_UNITTESTS_WITH_CUSTOM_OBJECTS_FROM_STATIC_LIBS
 extern "C" {
@@ -3339,7 +3380,7 @@ void RegisterCustomObjects(int /*argc*/, char** /*argv*/) {}
 #endif  // !ROCKSDB_UNITTESTS_WITH_CUSTOM_OBJECTS_FROM_STATIC_LIBS
 
 int main(int argc, char** argv) {
-  rocksdb::port::InstallStackTraceHandler();
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
   ::testing::InitGoogleTest(&argc, argv);
   RegisterCustomObjects(argc, argv);
   return RUN_ALL_TESTS();

db/compact_files_test.cc

@@ -18,7 +18,7 @@
 #include "test_util/testharness.h"
 #include "util/string_util.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class CompactFilesTest : public testing::Test {
  public:
@@ -81,11 +81,11 @@ TEST_F(CompactFilesTest, L0ConflictsFiles) {
   assert(s.ok());
   assert(db);
 
-  rocksdb::SyncPoint::GetInstance()->LoadDependency({
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency({
       {"CompactFilesImpl:0", "BackgroundCallCompaction:0"},
       {"BackgroundCallCompaction:1", "CompactFilesImpl:1"},
   });
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   // create couple files
   // Background compaction starts and waits in BackgroundCallCompaction:0
@@ -95,7 +95,7 @@ TEST_F(CompactFilesTest, L0ConflictsFiles) {
     db->Flush(FlushOptions());
   }
 
-  rocksdb::ColumnFamilyMetaData meta;
+  ROCKSDB_NAMESPACE::ColumnFamilyMetaData meta;
   db->GetColumnFamilyMetaData(&meta);
   std::string file1;
   for (auto& file : meta.levels[0].files) {
@@ -108,12 +108,12 @@ TEST_F(CompactFilesTest, L0ConflictsFiles) {
       // The background compaction then notices that there is an L0 compaction
       // already in progress and doesn't do an L0 compaction
       // Once the background compaction finishes, the compact files finishes
-      ASSERT_OK(
-          db->CompactFiles(rocksdb::CompactionOptions(), {file1, file2}, 0));
+      ASSERT_OK(db->CompactFiles(ROCKSDB_NAMESPACE::CompactionOptions(),
+                                 {file1, file2}, 0));
       break;
     }
   }
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
   delete db;
 }
 
@@ -224,14 +224,14 @@ TEST_F(CompactFilesTest, CapturingPendingFiles) {
   auto l0_files = collector->GetFlushedFiles();
   EXPECT_EQ(5, l0_files.size());
 
-  rocksdb::SyncPoint::GetInstance()->LoadDependency({
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency({
       {"CompactFilesImpl:2", "CompactFilesTest.CapturingPendingFiles:0"},
       {"CompactFilesTest.CapturingPendingFiles:1", "CompactFilesImpl:3"},
   });
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   // Start compacting files.
-  rocksdb::port::Thread compaction_thread(
+  ROCKSDB_NAMESPACE::port::Thread compaction_thread(
       [&] { EXPECT_OK(db->CompactFiles(CompactionOptions(), l0_files, 1)); });
 
   // In the meantime flush another file.
@@ -242,7 +242,7 @@ TEST_F(CompactFilesTest, CapturingPendingFiles) {
 
   compaction_thread.join();
 
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
 
   delete db;
 
@@ -296,12 +296,12 @@ TEST_F(CompactFilesTest, CompactionFilterWithGetSv) {
   db->Flush(FlushOptions());
 
   // Compact all L0 files using CompactFiles
-  rocksdb::ColumnFamilyMetaData meta;
+  ROCKSDB_NAMESPACE::ColumnFamilyMetaData meta;
   db->GetColumnFamilyMetaData(&meta);
   for (auto& file : meta.levels[0].files) {
     std::string fname = file.db_path + "/" + file.name;
     ASSERT_OK(
-        db->CompactFiles(rocksdb::CompactionOptions(), {fname}, 0));
+        db->CompactFiles(ROCKSDB_NAMESPACE::CompactionOptions(), {fname}, 0));
   }
 
 
@@ -347,7 +347,7 @@ TEST_F(CompactFilesTest, SentinelCompressionType) {
     compaction_opts.compression = CompressionType::kDisableCompressionOption;
     ASSERT_OK(db->CompactFiles(compaction_opts, l0_files, 1));
 
-    rocksdb::TablePropertiesCollection all_tables_props;
+    ROCKSDB_NAMESPACE::TablePropertiesCollection all_tables_props;
     ASSERT_OK(db->GetPropertiesOfAllTables(&all_tables_props));
     for (const auto& name_and_table_props : all_tables_props) {
       ASSERT_EQ(CompressionTypeToString(CompressionType::kZlibCompression),
@@ -402,7 +402,7 @@ TEST_F(CompactFilesTest, GetCompactionJobInfo) {
   delete db;
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
   ::testing::InitGoogleTest(&argc, argv);

db/compacted_db_impl.cc

@@ -9,7 +9,7 @@
 #include "db/version_set.h"
 #include "table/get_context.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 extern void MarkKeyMayExist(void* arg);
 extern bool SaveValue(void* arg, const ParsedInternalKey& parsed_key,
@@ -37,7 +37,7 @@ Status CompactedDBImpl::Get(const ReadOptions& options, ColumnFamilyHandle*,
                             const Slice& key, PinnableSlice* value) {
   GetContext get_context(user_comparator_, nullptr, nullptr, nullptr,
                          GetContext::kNotFound, key, value, nullptr, nullptr,
-                         true, nullptr, nullptr);
+                         nullptr, true, nullptr, nullptr);
   LookupKey lkey(key, kMaxSequenceNumber);
   files_.files[FindFile(key)].fd.table_reader->Get(options, lkey.internal_key(),
                                                    &get_context, nullptr);
@@ -70,7 +70,7 @@ std::vector<Status> CompactedDBImpl::MultiGet(const ReadOptions& options,
       std::string& value = (*values)[idx];
       GetContext get_context(user_comparator_, nullptr, nullptr, nullptr,
                              GetContext::kNotFound, keys[idx], &pinnable_val,
-                             nullptr, nullptr, true, nullptr, nullptr);
+                             nullptr, nullptr, nullptr, true, nullptr, nullptr);
       LookupKey lkey(keys[idx], kMaxSequenceNumber);
       r->Get(options, lkey.internal_key(), &get_context, nullptr);
       value.assign(pinnable_val.data(), pinnable_val.size());
@@ -156,5 +156,5 @@ Status CompactedDBImpl::Open(const Options& options,
   return s;
 }
 
-}   // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 #endif  // ROCKSDB_LITE

db/compacted_db_impl.h

@@ -9,7 +9,7 @@
 #include <vector>
 #include "db/db_impl/db_impl.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class CompactedDBImpl : public DBImpl {
  public:
@@ -109,5 +109,5 @@ class CompactedDBImpl : public DBImpl {
   const Comparator* user_comparator_;
   LevelFilesBrief files_;
 };
-}
+}  // namespace ROCKSDB_NAMESPACE
 #endif  // ROCKSDB_LITE

db/compaction/compaction.cc

@@ -16,7 +16,7 @@
 #include "test_util/sync_point.h"
 #include "util/string_util.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 const uint64_t kRangeTombstoneSentinel =
     PackSequenceAndType(kMaxSequenceNumber, kTypeRangeDeletion);
@@ -275,9 +275,7 @@ Compaction::~Compaction() {
     input_version_->Unref();
   }
   if (cfd_ != nullptr) {
-    if (cfd_->Unref()) {
-      delete cfd_;
-    }
+    cfd_->UnrefAndTryDelete();
   }
 }
 
@@ -547,11 +545,13 @@ bool Compaction::ShouldFormSubcompactions() const {
 
 uint64_t Compaction::MinInputFileOldestAncesterTime() const {
   uint64_t min_oldest_ancester_time = port::kMaxUint64;
-  for (const auto& file : inputs_[0].files) {
-    uint64_t oldest_ancester_time = file->TryGetOldestAncesterTime();
-    if (oldest_ancester_time != 0) {
-      min_oldest_ancester_time =
-          std::min(min_oldest_ancester_time, oldest_ancester_time);
+  for (const auto& level_files : inputs_) {
+    for (const auto& file : level_files.files) {
+      uint64_t oldest_ancester_time = file->TryGetOldestAncesterTime();
+      if (oldest_ancester_time != 0) {
+        min_oldest_ancester_time =
+            std::min(min_oldest_ancester_time, oldest_ancester_time);
+      }
     }
   }
   return min_oldest_ancester_time;
@@ -561,4 +561,4 @@ int Compaction::GetInputBaseLevel() const {
   return input_vstorage_->base_level();
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/compaction/compaction.h

@@ -13,7 +13,7 @@
 #include "options/cf_options.h"
 #include "util/autovector.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 // The file contains class Compaction, as well as some helper functions
 // and data structures used by the class.
 
@@ -381,4 +381,4 @@ class Compaction {
 // Return sum of sizes of all files in `files`.
 extern uint64_t TotalFileSize(const std::vector<FileMetaData*>& files);
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/compaction/compaction_iteration_stats.h

@@ -5,6 +5,8 @@
 
 #pragma once
 
+#include "rocksdb/rocksdb_namespace.h"
+
 struct CompactionIterationStats {
   // Compaction statistics
 

db/compaction/compaction_iterator.cc

@@ -28,7 +28,7 @@
   ((seq) <= earliest_snapshot_ && \
    (snapshot_checker_ == nullptr || LIKELY(IsInEarliestSnapshot(seq))))
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 CompactionIterator::CompactionIterator(
     InternalIterator* input, const Comparator* cmp, MergeHelper* merge_helper,
@@ -639,28 +639,48 @@ void CompactionIterator::NextFromInput() {
 }
 
 void CompactionIterator::PrepareOutput() {
-  // Zeroing out the sequence number leads to better compression.
-  // If this is the bottommost level (no files in lower levels)
-  // and the earliest snapshot is larger than this seqno
-  // and the userkey differs from the last userkey in compaction
-  // then we can squash the seqno to zero.
-  //
-  // This is safe for TransactionDB write-conflict checking since transactions
-  // only care about sequence number larger than any active snapshots.
-  //
-  // Can we do the same for levels above bottom level as long as
-  // KeyNotExistsBeyondOutputLevel() return true?
-  if ((compaction_ != nullptr && !compaction_->allow_ingest_behind()) &&
-      ikeyNotNeededForIncrementalSnapshot() && bottommost_level_ && valid_ &&
-      IN_EARLIEST_SNAPSHOT(ikey_.sequence) && ikey_.type != kTypeMerge) {
-    assert(ikey_.type != kTypeDeletion && ikey_.type != kTypeSingleDeletion);
-    if (ikey_.type == kTypeDeletion || ikey_.type == kTypeSingleDeletion) {
-      ROCKS_LOG_FATAL(info_log_,
-                      "Unexpected key type %d for seq-zero optimization",
-                      ikey_.type);
+  if (valid_) {
+    if (compaction_filter_ && ikey_.type == kTypeBlobIndex) {
+      const auto blob_decision = compaction_filter_->PrepareBlobOutput(
+          user_key(), value_, &compaction_filter_value_);
+
+      if (blob_decision == CompactionFilter::BlobDecision::kCorruption) {
+        status_ = Status::Corruption(
+            "Corrupted blob reference encountered during GC");
+        valid_ = false;
+      } else if (blob_decision == CompactionFilter::BlobDecision::kIOError) {
+        status_ = Status::IOError("Could not relocate blob during GC");
+        valid_ = false;
+      } else if (blob_decision ==
+                 CompactionFilter::BlobDecision::kChangeValue) {
+        value_ = compaction_filter_value_;
+      }
+    }
+
+    // Zeroing out the sequence number leads to better compression.
+    // If this is the bottommost level (no files in lower levels)
+    // and the earliest snapshot is larger than this seqno
+    // and the userkey differs from the last userkey in compaction
+    // then we can squash the seqno to zero.
+    //
+    // This is safe for TransactionDB write-conflict checking since transactions
+    // only care about sequence number larger than any active snapshots.
+    //
+    // Can we do the same for levels above bottom level as long as
+    // KeyNotExistsBeyondOutputLevel() return true?
+    if (valid_ && compaction_ != nullptr &&
+        !compaction_->allow_ingest_behind() &&
+        ikeyNotNeededForIncrementalSnapshot() && bottommost_level_ &&
+        IN_EARLIEST_SNAPSHOT(ikey_.sequence) && ikey_.type != kTypeMerge) {
+      assert(ikey_.type != kTypeDeletion && ikey_.type != kTypeSingleDeletion);
+      if (ikey_.type == kTypeDeletion || ikey_.type == kTypeSingleDeletion) {
+        ROCKS_LOG_FATAL(info_log_,
+                        "Unexpected key type %d for seq-zero optimization",
+                        ikey_.type);
+      }
+      ikey_.sequence = 0;
+      current_key_.UpdateInternalKey(0, ikey_.type);
     }
-    ikey_.sequence = 0;
-    current_key_.UpdateInternalKey(0, ikey_.type);
   }
 }
 
@@ -751,4 +771,4 @@ bool CompactionIterator::IsInEarliestSnapshot(SequenceNumber sequence) {
   return in_snapshot == SnapshotCheckerResult::kInSnapshot;
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/compaction/compaction_iterator.h

@@ -19,7 +19,7 @@
 #include "options/cf_options.h"
 #include "rocksdb/compaction_filter.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class CompactionIterator {
  public:
@@ -237,4 +237,4 @@ class CompactionIterator {
            manual_compaction_paused_->load(std::memory_order_relaxed);
   }
 };
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/compaction/compaction_iterator_test.cc

@@ -14,7 +14,7 @@
 #include "util/string_util.h"
 #include "utilities/merge_operators.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 // Expects no merging attempts.
 class NoMergingMergeOp : public MergeOperator {
@@ -494,7 +494,7 @@ TEST_P(CompactionIteratorTest, ShuttingDownInFilter) {
   compaction_proxy_->key_not_exists_beyond_output_level = true;
 
   std::atomic<bool> seek_done{false};
-  rocksdb::port::Thread compaction_thread([&] {
+  ROCKSDB_NAMESPACE::port::Thread compaction_thread([&] {
     c_iter_->SeekToFirst();
     EXPECT_FALSE(c_iter_->Valid());
     EXPECT_TRUE(c_iter_->status().IsShutdownInProgress());
@@ -531,7 +531,7 @@ TEST_P(CompactionIteratorTest, ShuttingDownInMerge) {
   compaction_proxy_->key_not_exists_beyond_output_level = true;
 
   std::atomic<bool> seek_done{false};
-  rocksdb::port::Thread compaction_thread([&] {
+  ROCKSDB_NAMESPACE::port::Thread compaction_thread([&] {
     c_iter_->SeekToFirst();
     ASSERT_FALSE(c_iter_->Valid());
     ASSERT_TRUE(c_iter_->status().IsShutdownInProgress());
@@ -968,7 +968,7 @@ TEST_F(CompactionIteratorWithSnapshotCheckerTest, CompactionFilter_FullMerge) {
       compaction_filter.get());
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
   ::testing::InitGoogleTest(&argc, argv);

db/compaction/compaction_job.cc

@@ -59,7 +59,7 @@
 #include "util/stop_watch.h"
 #include "util/string_util.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 const char* GetCompactionReasonString(CompactionReason compaction_reason) {
   switch (compaction_reason) {
@@ -144,7 +144,6 @@ struct CompactionJob::SubcompactionState {
 
   // State during the subcompaction
   uint64_t total_bytes;
-  uint64_t num_input_records;
   uint64_t num_output_records;
   CompactionJobStats compaction_job_stats;
   uint64_t approx_size;
@@ -165,7 +164,6 @@ struct CompactionJob::SubcompactionState {
         builder(nullptr),
         current_output_file_size(0),
         total_bytes(0),
-        num_input_records(0),
         num_output_records(0),
         approx_size(size),
         grandparent_index(0),
@@ -186,7 +184,6 @@ struct CompactionJob::SubcompactionState {
     builder = std::move(o.builder);
     current_output_file_size = std::move(o.current_output_file_size);
     total_bytes = std::move(o.total_bytes);
-    num_input_records = std::move(o.num_input_records);
     num_output_records = std::move(o.num_output_records);
     compaction_job_stats = std::move(o.compaction_job_stats);
     approx_size = std::move(o.approx_size);
@@ -245,13 +242,11 @@ struct CompactionJob::CompactionState {
   Status status;
 
   uint64_t total_bytes;
-  uint64_t num_input_records;
   uint64_t num_output_records;
 
   explicit CompactionState(Compaction* c)
       : compaction(c),
         total_bytes(0),
-        num_input_records(0),
         num_output_records(0) {}
 
   size_t NumOutputFiles() {
@@ -289,7 +284,6 @@ struct CompactionJob::CompactionState {
 void CompactionJob::AggregateStatistics() {
   for (SubcompactionState& sc : compact_->sub_compact_states) {
     compact_->total_bytes += sc.total_bytes;
-    compact_->num_input_records += sc.num_input_records;
     compact_->num_output_records += sc.num_output_records;
   }
   if (compaction_job_stats_) {
@@ -301,10 +295,10 @@ void CompactionJob::AggregateStatistics() {
 
 CompactionJob::CompactionJob(
     int job_id, Compaction* compaction, const ImmutableDBOptions& db_options,
-    const EnvOptions env_options, VersionSet* versions,
+    const FileOptions& file_options, VersionSet* versions,
     const std::atomic<bool>* shutting_down,
     const SequenceNumber preserve_deletes_seqnum, LogBuffer* log_buffer,
-    Directory* db_directory, Directory* output_directory, Statistics* stats,
+    FSDirectory* db_directory, FSDirectory* output_directory, Statistics* stats,
     InstrumentedMutex* db_mutex, ErrorHandler* db_error_handler,
     std::vector<SequenceNumber> existing_snapshots,
     SequenceNumber earliest_write_conflict_snapshot,
@@ -318,10 +312,11 @@ CompactionJob::CompactionJob(
       compaction_stats_(compaction->compaction_reason(), 1),
       dbname_(dbname),
       db_options_(db_options),
-      env_options_(env_options),
+      file_options_(file_options),
       env_(db_options.env),
-      env_options_for_read_(
-          env_->OptimizeForCompactionTableRead(env_options, db_options_)),
+      fs_(db_options.fs.get()),
+      file_options_for_read_(
+          fs_->OptimizeForCompactionTableRead(file_options, db_options_)),
       versions_(versions),
       shutting_down_(shutting_down),
       manual_compaction_paused_(manual_compaction_paused),
@@ -619,7 +614,7 @@ Status CompactionJob::Run() {
   }
 
   if (status.ok() && output_directory_) {
-    status = output_directory_->Fsync();
+    status = output_directory_->Fsync(IOOptions(), nullptr);
   }
 
   if (status.ok()) {
@@ -647,7 +642,7 @@ Status CompactionJob::Run() {
         // we will regard this verification as user reads since the goal is
         // to cache it here for further user reads
         InternalIterator* iter = cfd->table_cache()->NewIterator(
-            ReadOptions(), env_options_, cfd->internal_comparator(),
+            ReadOptions(), file_options_, cfd->internal_comparator(),
             *files_meta[file_idx], /*range_del_agg=*/nullptr, prefix_extractor,
             /*table_reader_ptr=*/nullptr,
             cfd->internal_stats()->GetFileReadHist(
@@ -769,12 +764,12 @@ Status CompactionJob::Install(const MutableCFOptions& mutable_cf_options) {
   auto stream = event_logger_->LogToBuffer(log_buffer_);
   stream << "job" << job_id_ << "event"
          << "compaction_finished"
-         << "compaction_time_micros" << compaction_stats_.micros
-         << "compaction_time_cpu_micros" << compaction_stats_.cpu_micros
-         << "output_level" << compact_->compaction->output_level()
-         << "num_output_files" << compact_->NumOutputFiles()
-         << "total_output_size" << compact_->total_bytes << "num_input_records"
-         << compact_->num_input_records << "num_output_records"
+         << "compaction_time_micros" << stats.micros
+         << "compaction_time_cpu_micros" << stats.cpu_micros << "output_level"
+         << compact_->compaction->output_level() << "num_output_files"
+         << compact_->NumOutputFiles() << "total_output_size"
+         << compact_->total_bytes << "num_input_records"
+         << stats.num_input_records << "num_output_records"
          << compact_->num_output_records << "num_subcompactions"
          << compact_->sub_compact_states.size() << "output_compression"
          << CompressionTypeToString(compact_->compaction->output_compression());
@@ -836,7 +831,7 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
   // Although the v2 aggregator is what the level iterator(s) know about,
   // the AddTombstones calls will be propagated down to the v1 aggregator.
   std::unique_ptr<InternalIterator> input(versions_->MakeInputIterator(
-      sub_compact->compaction, &range_del_agg, env_options_for_read_));
+      sub_compact->compaction, &range_del_agg, file_options_for_read_));
 
   AutoThreadOperationStageUpdater stage_updater(
       ThreadStatus::STAGE_COMPACTION_PROCESS_KV);
@@ -990,7 +985,6 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
     }
   }
 
-  sub_compact->num_input_records = c_iter_stats.num_input_records;
   sub_compact->compaction_job_stats.num_input_deletion_records =
       c_iter_stats.num_input_deletion_records;
   sub_compact->compaction_job_stats.num_corrupt_keys =
@@ -1302,6 +1296,11 @@ Status CompactionJob::FinishCompactionOutputFile(
   }
   const uint64_t current_bytes = sub_compact->builder->FileSize();
   if (s.ok()) {
+    // Add the checksum information to file metadata.
+    meta->file_checksum = sub_compact->builder->GetFileChecksum();
+    meta->file_checksum_func_name =
+        sub_compact->builder->GetFileChecksumFuncName();
+
     meta->fd.file_size = current_bytes;
   }
   sub_compact->current_output()->finished = true;
@@ -1457,13 +1456,13 @@ Status CompactionJob::OpenCompactionOutputFile(
       TableFileCreationReason::kCompaction);
 #endif  // !ROCKSDB_LITE
   // Make the output file
-  std::unique_ptr<WritableFile> writable_file;
+  std::unique_ptr<FSWritableFile> writable_file;
 #ifndef NDEBUG
-  bool syncpoint_arg = env_options_.use_direct_writes;
+  bool syncpoint_arg = file_options_.use_direct_writes;
   TEST_SYNC_POINT_CALLBACK("CompactionJob::OpenCompactionOutputFile",
                            &syncpoint_arg);
 #endif
-  Status s = NewWritableFile(env_, fname, &writable_file, env_options_);
+  Status s = NewWritableFile(fs_, fname, &writable_file, file_options_);
   if (!s.ok()) {
     ROCKS_LOG_ERROR(
         db_options_.info_log,
@@ -1501,19 +1500,21 @@ Status CompactionJob::OpenCompactionOutputFile(
     out.meta.fd = FileDescriptor(file_number,
                                  sub_compact->compaction->output_path_id(), 0);
     out.meta.oldest_ancester_time = oldest_ancester_time;
+    out.meta.file_creation_time = current_time;
     out.finished = false;
     sub_compact->outputs.push_back(out);
   }
 
-  writable_file->SetIOPriority(Env::IO_LOW);
+  writable_file->SetIOPriority(Env::IOPriority::IO_LOW);
   writable_file->SetWriteLifeTimeHint(write_hint_);
   writable_file->SetPreallocationBlockSize(static_cast<size_t>(
       sub_compact->compaction->OutputFilePreallocationSize()));
   const auto& listeners =
       sub_compact->compaction->immutable_cf_options()->listeners;
   sub_compact->outfile.reset(
-      new WritableFileWriter(std::move(writable_file), fname, env_options_,
-                             env_, db_options_.statistics.get(), listeners));
+      new WritableFileWriter(std::move(writable_file), fname, file_options_,
+                             env_, db_options_.statistics.get(), listeners,
+                             db_options_.sst_file_checksum_func.get()));
 
   // If the Column family flag is to only optimize filters for hits,
   // we can skip creating filters if this is the bottommost_level where
@@ -1587,6 +1588,8 @@ void CompactionJob::UpdateCompactionStats() {
     }
   }
 
+  uint64_t num_output_records = 0;
+
   for (const auto& sub_compact : compact_->sub_compact_states) {
     size_t num_output_files = sub_compact.outputs.size();
     if (sub_compact.builder != nullptr) {
@@ -1596,13 +1599,16 @@ void CompactionJob::UpdateCompactionStats() {
     }
     compaction_stats_.num_output_files += static_cast<int>(num_output_files);
 
+    num_output_records += sub_compact.num_output_records;
+
     for (const auto& out : sub_compact.outputs) {
       compaction_stats_.bytes_written += out.meta.fd.file_size;
     }
-    if (sub_compact.num_input_records > sub_compact.num_output_records) {
-      compaction_stats_.num_dropped_records +=
-          sub_compact.num_input_records - sub_compact.num_output_records;
-    }
+  }
+
+  if (compaction_stats_.num_input_records > num_output_records) {
+    compaction_stats_.num_dropped_records =
+        compaction_stats_.num_input_records - num_output_records;
   }
 }
 
@@ -1630,7 +1636,7 @@ void CompactionJob::UpdateCompactionJobStats(
     // input information
     compaction_job_stats_->total_input_bytes =
         stats.bytes_read_non_output_levels + stats.bytes_read_output_level;
-    compaction_job_stats_->num_input_records = compact_->num_input_records;
+    compaction_job_stats_->num_input_records = stats.num_input_records;
     compaction_job_stats_->num_input_files =
         stats.num_input_files_in_non_output_levels +
         stats.num_input_files_in_output_level;
@@ -1691,4 +1697,4 @@ void CompactionJob::LogCompaction() {
   }
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/compaction/compaction_job.h

@@ -44,7 +44,7 @@
 #include "util/stop_watch.h"
 #include "util/thread_local.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class Arena;
 class ErrorHandler;
@@ -64,11 +64,11 @@ class CompactionJob {
  public:
   CompactionJob(int job_id, Compaction* compaction,
                 const ImmutableDBOptions& db_options,
-                const EnvOptions env_options, VersionSet* versions,
+                const FileOptions& file_options, VersionSet* versions,
                 const std::atomic<bool>* shutting_down,
                 const SequenceNumber preserve_deletes_seqnum,
-                LogBuffer* log_buffer, Directory* db_directory,
-                Directory* output_directory, Statistics* stats,
+                LogBuffer* log_buffer, FSDirectory* db_directory,
+                FSDirectory* output_directory, Statistics* stats,
                 InstrumentedMutex* db_mutex, ErrorHandler* db_error_handler,
                 std::vector<SequenceNumber> existing_snapshots,
                 SequenceNumber earliest_write_conflict_snapshot,
@@ -150,18 +150,19 @@ class CompactionJob {
   // DBImpl state
   const std::string& dbname_;
   const ImmutableDBOptions& db_options_;
-  const EnvOptions env_options_;
+  const FileOptions file_options_;
 
   Env* env_;
+  FileSystem* fs_;
   // env_option optimized for compaction table reads
-  EnvOptions env_options_for_read_;
+  FileOptions file_options_for_read_;
   VersionSet* versions_;
   const std::atomic<bool>* shutting_down_;
   const std::atomic<bool>* manual_compaction_paused_;
   const SequenceNumber preserve_deletes_seqnum_;
   LogBuffer* log_buffer_;
-  Directory* db_directory_;
-  Directory* output_directory_;
+  FSDirectory* db_directory_;
+  FSDirectory* output_directory_;
   Statistics* stats_;
   InstrumentedMutex* db_mutex_;
   ErrorHandler* db_error_handler_;
@@ -194,4 +195,4 @@ class CompactionJob {
   Env::Priority thread_pri_;
 };
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/compaction/compaction_job_stats_test.cc

@@ -61,7 +61,7 @@
 
 #if !defined(IOS_CROSS_COMPILE)
 #ifndef ROCKSDB_LITE
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 static std::string RandomString(Random* rnd, int len, double ratio) {
   std::string r;
@@ -110,9 +110,9 @@ class CompactionJobStatsTest : public testing::Test,
   }
 
   ~CompactionJobStatsTest() override {
-    rocksdb::SyncPoint::GetInstance()->DisableProcessing();
-    rocksdb::SyncPoint::GetInstance()->LoadDependency({});
-    rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks();
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency({});
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
     Close();
     Options options;
     options.db_paths.emplace_back(dbname_, 0);
@@ -801,7 +801,7 @@ TEST_P(CompactionJobStatsTest, CompactionJobStatsTest) {
 
     stats_checker->set_verify_next_comp_io_stats(true);
     std::atomic<bool> first_prepare_write(true);
-    rocksdb::SyncPoint::GetInstance()->SetCallBack(
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
         "WritableFileWriter::Append:BeforePrepareWrite", [&](void* /*arg*/) {
           if (first_prepare_write.load()) {
             options.env->SleepForMicroseconds(3);
@@ -810,7 +810,7 @@ TEST_P(CompactionJobStatsTest, CompactionJobStatsTest) {
         });
 
     std::atomic<bool> first_flush(true);
-    rocksdb::SyncPoint::GetInstance()->SetCallBack(
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
         "WritableFileWriter::Flush:BeforeAppend", [&](void* /*arg*/) {
           if (first_flush.load()) {
             options.env->SleepForMicroseconds(3);
@@ -819,7 +819,7 @@ TEST_P(CompactionJobStatsTest, CompactionJobStatsTest) {
         });
 
     std::atomic<bool> first_sync(true);
-    rocksdb::SyncPoint::GetInstance()->SetCallBack(
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
         "WritableFileWriter::SyncInternal:0", [&](void* /*arg*/) {
           if (first_sync.load()) {
             options.env->SleepForMicroseconds(3);
@@ -828,14 +828,14 @@ TEST_P(CompactionJobStatsTest, CompactionJobStatsTest) {
         });
 
     std::atomic<bool> first_range_sync(true);
-    rocksdb::SyncPoint::GetInstance()->SetCallBack(
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
         "WritableFileWriter::RangeSync:0", [&](void* /*arg*/) {
           if (first_range_sync.load()) {
             options.env->SleepForMicroseconds(3);
             first_range_sync.store(false);
           }
         });
-    rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
     Compact(1, smallest_key, largest_key);
 
@@ -844,7 +844,7 @@ TEST_P(CompactionJobStatsTest, CompactionJobStatsTest) {
     ASSERT_TRUE(!first_flush.load());
     ASSERT_TRUE(!first_sync.load());
     ASSERT_TRUE(!first_range_sync.load());
-    rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+    ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
   }
   ASSERT_EQ(stats_checker->NumberOfUnverifiedStats(), 0U);
 }
@@ -1019,10 +1019,10 @@ TEST_P(CompactionJobStatsTest, UniversalCompactionTest) {
 
 INSTANTIATE_TEST_CASE_P(CompactionJobStatsTest, CompactionJobStatsTest,
                         ::testing::Values(1, 4));
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
-  rocksdb::port::InstallStackTraceHandler();
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
   ::testing::InitGoogleTest(&argc, argv);
   return RUN_ALL_TESTS();
 }

db/compaction/compaction_job_test.cc

@@ -12,7 +12,7 @@
 #include <string>
 #include <tuple>
 
-#include "db/blob_index.h"
+#include "db/blob/blob_index.h"
 #include "db/column_family.h"
 #include "db/compaction/compaction_job.h"
 #include "db/db_impl/db_impl.h"
@@ -29,7 +29,7 @@
 #include "util/string_util.h"
 #include "utilities/merge_operators.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 namespace {
 
@@ -72,6 +72,7 @@ class CompactionJobTest : public testing::Test {
  public:
   CompactionJobTest()
       : env_(Env::Default()),
+        fs_(std::make_shared<LegacyFileSystemWrapper>(env_)),
         dbname_(test::PerThreadDBPath("compaction_job_test")),
         db_options_(),
         mutable_cf_options_(cf_options_),
@@ -86,6 +87,8 @@ class CompactionJobTest : public testing::Test {
         mock_table_factory_(new mock::MockTableFactory()),
         error_handler_(nullptr, db_options_, &mutex_) {
     EXPECT_OK(env_->CreateDirIfMissing(dbname_));
+    db_options_.env = env_;
+    db_options_.fs = fs_;
     db_options_.db_paths.emplace_back(dbname_,
                                       std::numeric_limits<uint64_t>::max());
   }
@@ -184,7 +187,8 @@ class CompactionJobTest : public testing::Test {
     VersionEdit edit;
     edit.AddFile(level, file_number, 0, 10, smallest_key, largest_key,
                  smallest_seqno, largest_seqno, false, oldest_blob_file_number,
-                 kUnknownOldestAncesterTime);
+                 kUnknownOldestAncesterTime, kUnknownFileCreationTime,
+                 kUnknownFileChecksum, kUnknownFileChecksumFuncName);
 
     mutex_.Lock();
     versions_->LogAndApply(versions_->GetColumnFamilySet()->GetDefault(),
@@ -267,8 +271,8 @@ class CompactionJobTest : public testing::Test {
     Status s = env_->NewWritableFile(
         manifest, &file, env_->OptimizeForManifestWrite(env_options_));
     ASSERT_OK(s);
-    std::unique_ptr<WritableFileWriter> file_writer(
-        new WritableFileWriter(std::move(file), manifest, env_options_));
+    std::unique_ptr<WritableFileWriter> file_writer(new WritableFileWriter(
+        NewLegacyWritableFileWrapper(std::move(file)), manifest, env_options_));
     {
       log::Writer log(std::move(file_writer), 0, false);
       std::string record;
@@ -360,6 +364,7 @@ class CompactionJobTest : public testing::Test {
   }
 
   Env* env_;
+  std::shared_ptr<FileSystem> fs_;
   std::string dbname_;
   EnvOptions env_options_;
   ImmutableDBOptions db_options_;
@@ -1058,7 +1063,7 @@ TEST_F(CompactionJobTest, OldestBlobFileNumber) {
                 /* expected_oldest_blob_file_number */ 19);
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
   ::testing::InitGoogleTest(&argc, argv);

db/compaction/compaction_picker.cc

@@ -23,7 +23,7 @@
 #include "util/random.h"
 #include "util/string_util.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 namespace {
 uint64_t TotalCompensatedFileSize(const std::vector<FileMetaData*>& files) {
@@ -1128,4 +1128,4 @@ bool CompactionPicker::GetOverlappingL0Files(
   return true;
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/compaction/compaction_picker.h

@@ -22,7 +22,7 @@
 #include "rocksdb/options.h"
 #include "rocksdb/status.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 // The file contains an abstract class CompactionPicker, and its two
 // sub-classes LevelCompactionPicker and NullCompactionPicker, as
@@ -310,4 +310,4 @@ CompressionOptions GetCompressionOptions(const ImmutableCFOptions& ioptions,
                                          int level,
                                          const bool enable_compression = true);
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/compaction/compaction_picker_fifo.cc

@@ -17,7 +17,7 @@
 #include "logging/log_buffer.h"
 #include "util/string_util.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 namespace {
 uint64_t GetTotalFilesSize(const std::vector<FileMetaData*>& files) {
   uint64_t total_size = 0;
@@ -70,18 +70,14 @@ Compaction* FIFOCompactionPicker::PickTTLCompaction(
   // avoid underflow
   if (current_time > mutable_cf_options.ttl) {
     for (auto ritr = level_files.rbegin(); ritr != level_files.rend(); ++ritr) {
-      auto f = *ritr;
-      if (f->fd.table_reader != nullptr &&
-          f->fd.table_reader->GetTableProperties() != nullptr) {
-        auto creation_time =
-            f->fd.table_reader->GetTableProperties()->creation_time;
-        if (creation_time == 0 ||
-            creation_time >= (current_time - mutable_cf_options.ttl)) {
-          break;
-        }
-        total_size -= f->compensated_file_size;
-        inputs[0].files.push_back(f);
+      FileMetaData* f = *ritr;
+      uint64_t creation_time = f->TryGetFileCreationTime();
+      if (creation_time == kUnknownFileCreationTime ||
+          creation_time >= (current_time - mutable_cf_options.ttl)) {
+        break;
       }
+      total_size -= f->compensated_file_size;
+      inputs[0].files.push_back(f);
     }
   }
 
@@ -100,7 +96,7 @@ Compaction* FIFOCompactionPicker::PickTTLCompaction(
                      "[%s] FIFO compaction: picking file %" PRIu64
                      " with creation time %" PRIu64 " for deletion",
                      cf_name.c_str(), f->fd.GetNumber(),
-                     f->fd.table_reader->GetTableProperties()->creation_time);
+                     f->TryGetFileCreationTime());
   }
 
   Compaction* c = new Compaction(
@@ -238,5 +234,5 @@ Compaction* FIFOCompactionPicker::CompactRange(
   return c;
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 #endif  // !ROCKSDB_LITE

db/compaction/compaction_picker_fifo.h

@@ -12,7 +12,7 @@
 
 #include "db/compaction/compaction_picker.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 class FIFOCompactionPicker : public CompactionPicker {
  public:
   FIFOCompactionPicker(const ImmutableCFOptions& ioptions,
@@ -49,5 +49,5 @@ class FIFOCompactionPicker : public CompactionPicker {
                                  VersionStorageInfo* version,
                                  LogBuffer* log_buffer);
 };
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 #endif  // !ROCKSDB_LITE

db/compaction/compaction_picker_level.cc

@@ -15,7 +15,7 @@
 #include "logging/log_buffer.h"
 #include "test_util/sync_point.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 bool LevelCompactionPicker::NeedsCompaction(
     const VersionStorageInfo* vstorage) const {
@@ -555,4 +555,4 @@ Compaction* LevelCompactionPicker::PickCompaction(
                                  log_buffer, mutable_cf_options, ioptions_);
   return builder.PickCompaction();
 }
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/compaction/compaction_picker_level.h

@@ -11,7 +11,7 @@
 
 #include "db/compaction/compaction_picker.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 // Picking compactions for leveled compaction. See wiki page
 // https://github.com/facebook/rocksdb/wiki/Leveled-Compaction
 // for description of Leveled compaction.
@@ -29,4 +29,4 @@ class LevelCompactionPicker : public CompactionPicker {
       const VersionStorageInfo* vstorage) const override;
 };
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/compaction/compaction_picker_test.cc

@@ -17,7 +17,7 @@
 #include "test_util/testutil.h"
 #include "util/string_util.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class CountingLogger : public Logger {
  public:
@@ -95,7 +95,8 @@ class CompactionPickerTest : public testing::Test {
         InternalKey(smallest, smallest_seq, kTypeValue),
         InternalKey(largest, largest_seq, kTypeValue), smallest_seq,
         largest_seq, /* marked_for_compact */ false, kInvalidBlobFileNumber,
-        kUnknownOldestAncesterTime);
+        kUnknownOldestAncesterTime, kUnknownFileCreationTime,
+        kUnknownFileChecksum, kUnknownFileChecksumFuncName);
     f->compensated_file_size =
         (compensated_file_size != 0) ? compensated_file_size : file_size;
     vstorage_->AddFile(level, f);
@@ -1732,7 +1733,7 @@ TEST_F(CompactionPickerTest, IntraL0ForEarliestSeqno) {
   ASSERT_EQ(0, compaction->output_level());
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
   ::testing::InitGoogleTest(&argc, argv);

db/compaction/compaction_picker_universal.cc

@@ -23,7 +23,7 @@
 #include "util/random.h"
 #include "util/string_util.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 namespace {
 // A helper class that form universal compactions. The class is used by
 // UniversalCompactionPicker::PickCompaction().
@@ -247,7 +247,8 @@ bool UniversalCompactionBuilder::IsInputFilesNonOverlapping(Compaction* c) {
 
     next.f = nullptr;
 
-    if (curr.level != 0 && curr.index < c->num_input_files(curr.level) - 1) {
+    if (c->level(curr.level) != 0 &&
+        curr.index < c->num_input_files(curr.level) - 1) {
       next.f = c->input(curr.level, curr.index + 1);
       next.level = curr.level;
       next.index = curr.index + 1;
@@ -1099,6 +1100,6 @@ Compaction* UniversalCompactionBuilder::PickPeriodicCompaction() {
 
   return c;
 }
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 #endif  // !ROCKSDB_LITE

db/compaction/compaction_picker_universal.h

@@ -12,7 +12,7 @@
 
 #include "db/compaction/compaction_picker.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 class UniversalCompactionPicker : public CompactionPicker {
  public:
   UniversalCompactionPicker(const ImmutableCFOptions& ioptions,
@@ -27,5 +27,5 @@ class UniversalCompactionPicker : public CompactionPicker {
   virtual bool NeedsCompaction(
       const VersionStorageInfo* vstorage) const override;
 };
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 #endif  // !ROCKSDB_LITE

db/comparator_db_test.cc

@@ -18,10 +18,10 @@
 
 using std::unique_ptr;
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 namespace {
 
-static const Comparator* comparator;
+static const Comparator* kTestComparator = nullptr;
 
 class KVIter : public Iterator {
  public:
@@ -74,7 +74,7 @@ void AssertItersEqual(Iterator* iter1, Iterator* iter2) {
 void DoRandomIteraratorTest(DB* db, std::vector<std::string> source_strings,
                             Random* rnd, int num_writes, int num_iter_ops,
                             int num_trigger_flush) {
-  stl_wrappers::KVMap map((stl_wrappers::LessOfComparator(comparator)));
+  stl_wrappers::KVMap map((stl_wrappers::LessOfComparator(kTestComparator)));
 
   for (int i = 0; i < num_writes; i++) {
     if (num_trigger_flush > 0 && i != 0 && i % num_trigger_flush == 0) {
@@ -263,19 +263,19 @@ class ComparatorDBTest
 
  public:
   ComparatorDBTest() : env_(Env::Default()), db_(nullptr) {
-    comparator = BytewiseComparator();
+    kTestComparator = BytewiseComparator();
     dbname_ = test::PerThreadDBPath("comparator_db_test");
     BlockBasedTableOptions toptions;
     toptions.format_version = GetParam();
     last_options_.table_factory.reset(
-        rocksdb::NewBlockBasedTableFactory(toptions));
+        ROCKSDB_NAMESPACE::NewBlockBasedTableFactory(toptions));
     EXPECT_OK(DestroyDB(dbname_, last_options_));
   }
 
   ~ComparatorDBTest() override {
     delete db_;
     EXPECT_OK(DestroyDB(dbname_, last_options_));
-    comparator = BytewiseComparator();
+    kTestComparator = BytewiseComparator();
   }
 
   DB* GetDB() { return db_; }
@@ -286,7 +286,7 @@ class ComparatorDBTest
     } else {
       comparator_guard.reset();
     }
-    comparator = cmp;
+    kTestComparator = cmp;
     last_options_.comparator = cmp;
   }
 
@@ -334,7 +334,7 @@ TEST_P(ComparatorDBTest, SimpleSuffixReverseComparator) {
 
   for (int rnd_seed = 301; rnd_seed < 316; rnd_seed++) {
     Options* opt = GetOptions();
-    opt->comparator = comparator;
+    opt->comparator = kTestComparator;
     DestroyAndReopen();
     Random rnd(rnd_seed);
 
@@ -360,7 +360,7 @@ TEST_P(ComparatorDBTest, Uint64Comparator) {
 
   for (int rnd_seed = 301; rnd_seed < 316; rnd_seed++) {
     Options* opt = GetOptions();
-    opt->comparator = comparator;
+    opt->comparator = kTestComparator;
     DestroyAndReopen();
     Random rnd(rnd_seed);
     Random64 rnd64(rnd_seed);
@@ -384,7 +384,7 @@ TEST_P(ComparatorDBTest, DoubleComparator) {
 
   for (int rnd_seed = 301; rnd_seed < 316; rnd_seed++) {
     Options* opt = GetOptions();
-    opt->comparator = comparator;
+    opt->comparator = kTestComparator;
     DestroyAndReopen();
     Random rnd(rnd_seed);
 
@@ -409,7 +409,7 @@ TEST_P(ComparatorDBTest, HashComparator) {
 
   for (int rnd_seed = 301; rnd_seed < 316; rnd_seed++) {
     Options* opt = GetOptions();
-    opt->comparator = comparator;
+    opt->comparator = kTestComparator;
     DestroyAndReopen();
     Random rnd(rnd_seed);
 
@@ -428,7 +428,7 @@ TEST_P(ComparatorDBTest, TwoStrComparator) {
 
   for (int rnd_seed = 301; rnd_seed < 316; rnd_seed++) {
     Options* opt = GetOptions();
-    opt->comparator = comparator;
+    opt->comparator = kTestComparator;
     DestroyAndReopen();
     Random rnd(rnd_seed);
 
@@ -652,7 +652,7 @@ TEST_P(ComparatorDBTest, SeparatorSuccessorRandomizeTest) {
   }
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
   ::testing::InitGoogleTest(&argc, argv);

db/convenience.cc

@@ -11,7 +11,7 @@
 #include "db/db_impl/db_impl.h"
 #include "util/cast_util.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 void CancelAllBackgroundWork(DB* db, bool wait) {
   (static_cast_with_check<DBImpl, DB>(db->GetRootDB()))
@@ -41,14 +41,16 @@ Status VerifySstFileChecksum(const Options& options,
                              const EnvOptions& env_options,
                              const ReadOptions& read_options,
                              const std::string& file_path) {
-  std::unique_ptr<RandomAccessFile> file;
+  std::unique_ptr<FSRandomAccessFile> file;
   uint64_t file_size;
   InternalKeyComparator internal_comparator(options.comparator);
   ImmutableCFOptions ioptions(options);
 
-  Status s = ioptions.env->NewRandomAccessFile(file_path, &file, env_options);
+  Status s = ioptions.fs->NewRandomAccessFile(file_path,
+                                              FileOptions(env_options),
+                                              &file, nullptr);
   if (s.ok()) {
-    s = ioptions.env->GetFileSize(file_path, &file_size);
+    s = ioptions.fs->GetFileSize(file_path, IOOptions(), &file_size, nullptr);
   } else {
     return s;
   }
@@ -70,6 +72,6 @@ Status VerifySstFileChecksum(const Options& options,
   return s;
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 #endif  // ROCKSDB_LITE

db/corruption_test.cc

@@ -20,6 +20,7 @@
 #include "db/db_test_util.h"
 #include "db/log_format.h"
 #include "db/version_set.h"
+#include "env/composite_env_wrapper.h"
 #include "file/filename.h"
 #include "rocksdb/cache.h"
 #include "rocksdb/convenience.h"
@@ -32,7 +33,7 @@
 #include "test_util/testutil.h"
 #include "util/string_util.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 static const int kValueSize = 1000;
 
@@ -97,7 +98,7 @@ class CorruptionTest : public testing::Test {
   void RepairDB() {
     delete db_;
     db_ = nullptr;
-    ASSERT_OK(::rocksdb::RepairDB(dbname_, options_));
+    ASSERT_OK(::ROCKSDB_NAMESPACE::RepairDB(dbname_, options_));
   }
 
   void Build(int n, int flush_every = 0) {
@@ -189,6 +190,7 @@ class CorruptionTest : public testing::Test {
     ASSERT_TRUE(s.ok()) << s.ToString();
     Options options;
     EnvOptions env_options;
+    options.file_system.reset(new LegacyFileSystemWrapper(options.env));
     ASSERT_NOK(VerifySstFileChecksum(options, env_options, fname));
   }
 
@@ -391,12 +393,16 @@ TEST_F(CorruptionTest, TableFileIndexData) {
 
   // corrupt an index block of an entire file
   Corrupt(kTableFile, -2000, 500);
-  Reopen();
+  options.paranoid_checks = false;
+  Reopen(&options);
   dbi = reinterpret_cast<DBImpl*>(db_);
   // one full file may be readable, since only one was corrupted
   // the other file should be fully non-readable, since index was corrupted
   Check(0, 5000);
   ASSERT_NOK(dbi->VerifyChecksum());
+
+  // In paranoid mode, the db cannot be opened due to the corrupted file.
+  ASSERT_TRUE(TryReopen().IsCorruption());
 }
 
 TEST_F(CorruptionTest, MissingDescriptor) {
@@ -539,7 +545,8 @@ TEST_F(CorruptionTest, RangeDeletionCorrupted) {
   std::unique_ptr<RandomAccessFile> file;
   ASSERT_OK(options_.env->NewRandomAccessFile(filename, &file, EnvOptions()));
   std::unique_ptr<RandomAccessFileReader> file_reader(
-      new RandomAccessFileReader(std::move(file), filename));
+      new RandomAccessFileReader(NewLegacyRandomAccessFileWrapper(file),
+                                 filename));
 
   uint64_t file_size;
   ASSERT_OK(options_.env->GetFileSize(filename, &file_size));
@@ -551,13 +558,7 @@ TEST_F(CorruptionTest, RangeDeletionCorrupted) {
 
   ASSERT_OK(TryReopen());
   CorruptFile(filename, static_cast<int>(range_del_handle.offset()), 1);
-  // The test case does not fail on TryReopen because failure to preload table
-  // handlers is not considered critical.
-  ASSERT_OK(TryReopen());
-  std::string val;
-  // However, it does fail on any read involving that file since that file
-  // cannot be opened with a corrupt range deletion meta-block.
-  ASSERT_TRUE(db_->Get(ReadOptions(), "a", &val).IsCorruption());
+  ASSERT_TRUE(TryReopen().IsCorruption());
 }
 
 TEST_F(CorruptionTest, FileSystemStateCorrupted) {
@@ -582,18 +583,19 @@ TEST_F(CorruptionTest, FileSystemStateCorrupted) {
       env_.NewWritableFile(filename, &file, EnvOptions());
       file->Append(Slice("corrupted sst"));
       file.reset();
+      Status x = TryReopen(&options);
+      ASSERT_TRUE(x.IsCorruption());
     } else {  // delete the file
       env_.DeleteFile(filename);
+      Status x = TryReopen(&options);
+      ASSERT_TRUE(x.IsPathNotFound());
     }
 
-    Status x = TryReopen(&options);
-    ASSERT_TRUE(x.IsCorruption());
     DestroyDB(dbname_, options_);
-    Reopen(&options);
   }
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
   ::testing::InitGoogleTest(&argc, argv);

db/cuckoo_table_db_test.cc

@@ -15,7 +15,7 @@
 #include "test_util/testutil.h"
 #include "util/string_util.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class CuckooTableDBTest : public testing::Test {
  private:
@@ -298,17 +298,25 @@ TEST_F(CuckooTableDBTest, AdaptiveTable) {
   dbfull()->TEST_FlushMemTable();
 
   // Write some keys using plain table.
+  std::shared_ptr<TableFactory> block_based_factory(
+      NewBlockBasedTableFactory());
+  std::shared_ptr<TableFactory> plain_table_factory(
+      NewPlainTableFactory());
+  std::shared_ptr<TableFactory> cuckoo_table_factory(
+      NewCuckooTableFactory());
   options.create_if_missing = false;
-  options.table_factory.reset(NewPlainTableFactory());
+  options.table_factory.reset(NewAdaptiveTableFactory(
+    plain_table_factory, block_based_factory, plain_table_factory,
+    cuckoo_table_factory));
   Reopen(&options);
   ASSERT_OK(Put("key4", "v4"));
   ASSERT_OK(Put("key1", "v5"));
   dbfull()->TEST_FlushMemTable();
 
   // Write some keys using block based table.
-  std::shared_ptr<TableFactory> block_based_factory(
-      NewBlockBasedTableFactory());
-  options.table_factory.reset(NewAdaptiveTableFactory(block_based_factory));
+  options.table_factory.reset(NewAdaptiveTableFactory(
+    block_based_factory, block_based_factory, plain_table_factory,
+    cuckoo_table_factory));
   Reopen(&options);
   ASSERT_OK(Put("key5", "v6"));
   ASSERT_OK(Put("key2", "v7"));
@@ -320,14 +328,13 @@ TEST_F(CuckooTableDBTest, AdaptiveTable) {
   ASSERT_EQ("v4", Get("key4"));
   ASSERT_EQ("v6", Get("key5"));
 }
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
-  if (rocksdb::port::kLittleEndian) {
+  if (ROCKSDB_NAMESPACE::port::kLittleEndian) {
     ::testing::InitGoogleTest(&argc, argv);
     return RUN_ALL_TESTS();
-  }
-  else {
+  } else {
     fprintf(stderr, "SKIPPED as Cuckoo table doesn't support Big Endian\n");
     return 0;
   }

db/db_basic_test.cc

@@ -6,7 +6,6 @@
 // Copyright (c) 2011 The LevelDB Authors. All rights reserved.
 // Use of this source 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 <iostream>
 #include "db/db_test_util.h"
 #include "port/stack_trace.h"
 #include "rocksdb/perf_context.h"
@@ -18,7 +17,7 @@
 #include "test_util/sync_point.h"
 #endif
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class DBBasicTest : public DBTestBase {
  public:
@@ -28,8 +27,8 @@ class DBBasicTest : public DBTestBase {
 TEST_F(DBBasicTest, OpenWhenOpen) {
   Options options = CurrentOptions();
   options.env = env_;
-  rocksdb::DB* db2 = nullptr;
-  rocksdb::Status s = DB::Open(options, dbname_, &db2);
+  ROCKSDB_NAMESPACE::DB* db2 = nullptr;
+  ROCKSDB_NAMESPACE::Status s = DB::Open(options, dbname_, &db2);
 
   ASSERT_EQ(Status::Code::kIOError, s.code());
   ASSERT_EQ(Status::SubCode::kNone, s.subcode());
@@ -325,7 +324,12 @@ TEST_F(DBBasicTest, CheckLock) {
     ASSERT_OK(TryReopen(options));
 
     // second open should fail
-    ASSERT_TRUE(!(DB::Open(options, dbname_, &localdb)).ok());
+    Status s = DB::Open(options, dbname_, &localdb);
+    ASSERT_NOK(s);
+#ifdef OS_LINUX
+    ASSERT_TRUE(s.ToString().find("lock hold by current process") !=
+                std::string::npos);
+#endif  // OS_LINUX
   } while (ChangeCompactOptions());
 }
 
@@ -393,7 +397,7 @@ TEST_F(DBBasicTest, FlushEmptyColumnFamily) {
   sleeping_task_low.WaitUntilDone();
 }
 
-TEST_F(DBBasicTest, FLUSH) {
+TEST_F(DBBasicTest, Flush) {
   do {
     CreateAndReopenWithCF({"pikachu"}, CurrentOptions());
     WriteOptions writeOpt = WriteOptions();
@@ -525,6 +529,7 @@ TEST_F(DBBasicTest, Snapshot) {
     ASSERT_EQ(1U, GetNumSnapshots());
     uint64_t time_snap1 = GetTimeOldestSnapshots();
     ASSERT_GT(time_snap1, 0U);
+    ASSERT_EQ(GetSequenceOldestSnapshots(), s1->GetSequenceNumber());
     Put(0, "foo", "0v2");
     Put(1, "foo", "1v2");
 
@@ -533,6 +538,7 @@ TEST_F(DBBasicTest, Snapshot) {
     const Snapshot* s2 = db_->GetSnapshot();
     ASSERT_EQ(2U, GetNumSnapshots());
     ASSERT_EQ(time_snap1, GetTimeOldestSnapshots());
+    ASSERT_EQ(GetSequenceOldestSnapshots(), s1->GetSequenceNumber());
     Put(0, "foo", "0v3");
     Put(1, "foo", "1v3");
 
@@ -540,6 +546,7 @@ TEST_F(DBBasicTest, Snapshot) {
       ManagedSnapshot s3(db_);
       ASSERT_EQ(3U, GetNumSnapshots());
       ASSERT_EQ(time_snap1, GetTimeOldestSnapshots());
+      ASSERT_EQ(GetSequenceOldestSnapshots(), s1->GetSequenceNumber());
 
       Put(0, "foo", "0v4");
       Put(1, "foo", "1v4");
@@ -555,6 +562,7 @@ TEST_F(DBBasicTest, Snapshot) {
 
     ASSERT_EQ(2U, GetNumSnapshots());
     ASSERT_EQ(time_snap1, GetTimeOldestSnapshots());
+    ASSERT_EQ(GetSequenceOldestSnapshots(), s1->GetSequenceNumber());
     ASSERT_EQ("0v1", Get(0, "foo", s1));
     ASSERT_EQ("1v1", Get(1, "foo", s1));
     ASSERT_EQ("0v2", Get(0, "foo", s2));
@@ -569,9 +577,11 @@ TEST_F(DBBasicTest, Snapshot) {
     ASSERT_EQ("1v4", Get(1, "foo"));
     ASSERT_EQ(1U, GetNumSnapshots());
     ASSERT_LT(time_snap1, GetTimeOldestSnapshots());
+    ASSERT_EQ(GetSequenceOldestSnapshots(), s2->GetSequenceNumber());
 
     db_->ReleaseSnapshot(s2);
     ASSERT_EQ(0U, GetNumSnapshots());
+    ASSERT_EQ(GetSequenceOldestSnapshots(), 0);
     ASSERT_EQ("0v4", Get(0, "foo"));
     ASSERT_EQ("1v4", Get(1, "foo"));
   } while (ChangeOptions());
@@ -1043,8 +1053,8 @@ TEST_P(DBMultiGetTestWithParam, MultiGetMultiCF) {
   }
 
   int get_sv_count = 0;
-  rocksdb::DBImpl* db = reinterpret_cast<DBImpl*>(db_);
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::DBImpl* db = reinterpret_cast<DBImpl*>(db_);
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "DBImpl::MultiGet::AfterRefSV", [&](void* /*arg*/) {
         if (++get_sv_count == 2) {
           // After MultiGet refs a couple of CFs, flush all CFs so MultiGet
@@ -1068,7 +1078,7 @@ TEST_P(DBMultiGetTestWithParam, MultiGetMultiCF) {
           }
         }
       });
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   std::vector<int> cfs;
   std::vector<std::string> keys;
@@ -1134,12 +1144,12 @@ TEST_P(DBMultiGetTestWithParam, MultiGetMultiCFMutex) {
   int get_sv_count = 0;
   int retries = 0;
   bool last_try = false;
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "DBImpl::MultiGet::LastTry", [&](void* /*arg*/) {
         last_try = true;
-        rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+        ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
       });
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "DBImpl::MultiGet::AfterRefSV", [&](void* /*arg*/) {
         if (last_try) {
           return;
@@ -1155,7 +1165,7 @@ TEST_P(DBMultiGetTestWithParam, MultiGetMultiCFMutex) {
           }
         }
       });
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   std::vector<int> cfs;
   std::vector<std::string> keys;
@@ -1193,8 +1203,8 @@ TEST_P(DBMultiGetTestWithParam, MultiGetMultiCFSnapshot) {
   }
 
   int get_sv_count = 0;
-  rocksdb::DBImpl* db = reinterpret_cast<DBImpl*>(db_);
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::DBImpl* db = reinterpret_cast<DBImpl*>(db_);
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "DBImpl::MultiGet::AfterRefSV", [&](void* /*arg*/) {
         if (++get_sv_count == 2) {
           for (int i = 0; i < 8; ++i) {
@@ -1214,7 +1224,7 @@ TEST_P(DBMultiGetTestWithParam, MultiGetMultiCFSnapshot) {
           }
         }
       });
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   std::vector<int> cfs;
   std::vector<std::string> keys;
@@ -1400,6 +1410,91 @@ TEST_F(DBBasicTest, MultiGetBatchedMultiLevel) {
   }
 }
 
+TEST_F(DBBasicTest, MultiGetBatchedMultiLevelMerge) {
+  Options options = CurrentOptions();
+  options.disable_auto_compactions = true;
+  options.merge_operator = MergeOperators::CreateStringAppendOperator();
+  BlockBasedTableOptions bbto;
+  bbto.filter_policy.reset(NewBloomFilterPolicy(10, false));
+  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+  Reopen(options);
+  int num_keys = 0;
+
+  for (int i = 0; i < 128; ++i) {
+    ASSERT_OK(Put("key_" + std::to_string(i), "val_l2_" + std::to_string(i)));
+    num_keys++;
+    if (num_keys == 8) {
+      Flush();
+      num_keys = 0;
+    }
+  }
+  if (num_keys > 0) {
+    Flush();
+    num_keys = 0;
+  }
+  MoveFilesToLevel(2);
+
+  for (int i = 0; i < 128; i += 3) {
+    ASSERT_OK(Merge("key_" + std::to_string(i), "val_l1_" + std::to_string(i)));
+    num_keys++;
+    if (num_keys == 8) {
+      Flush();
+      num_keys = 0;
+    }
+  }
+  if (num_keys > 0) {
+    Flush();
+    num_keys = 0;
+  }
+  MoveFilesToLevel(1);
+
+  for (int i = 0; i < 128; i += 5) {
+    ASSERT_OK(Merge("key_" + std::to_string(i), "val_l0_" + std::to_string(i)));
+    num_keys++;
+    if (num_keys == 8) {
+      Flush();
+      num_keys = 0;
+    }
+  }
+  if (num_keys > 0) {
+    Flush();
+    num_keys = 0;
+  }
+  ASSERT_EQ(0, num_keys);
+
+  for (int i = 0; i < 128; i += 9) {
+    ASSERT_OK(Merge("key_" + std::to_string(i), "val_mem_" + std::to_string(i)));
+  }
+
+  std::vector<std::string> keys;
+  std::vector<std::string> values;
+
+  for (int i = 32; i < 80; ++i) {
+    keys.push_back("key_" + std::to_string(i));
+  }
+
+  values = MultiGet(keys, nullptr);
+  ASSERT_EQ(values.size(), keys.size());
+  for (unsigned int j = 0; j < 48; ++j) {
+    int key = j + 32;
+    std::string value;
+    value.append("val_l2_" + std::to_string(key));
+    if (key % 3 == 0) {
+      value.append(",");
+      value.append("val_l1_" + std::to_string(key));
+    }
+    if (key % 5 == 0) {
+      value.append(",");
+      value.append("val_l0_" + std::to_string(key));
+    }
+    if (key % 9 == 0) {
+      value.append(",");
+      value.append("val_mem_" + std::to_string(key));
+    }
+    ASSERT_EQ(values[j], value);
+  }
+}
+
 // Test class for batched MultiGet with prefix extractor
 // Param bool - If true, use partitioned filters
 //              If false, use full filter block
@@ -1468,7 +1563,7 @@ TEST_P(DBMultiGetRowCacheTest, MultiGetBatched) {
   do {
     option_config_ = kRowCache;
     Options options = CurrentOptions();
-    options.statistics = rocksdb::CreateDBStatistics();
+    options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
     CreateAndReopenWithCF({"pikachu"}, options);
     SetPerfLevel(kEnableCount);
     ASSERT_OK(Put(1, "k1", "v1"));
@@ -1574,13 +1669,13 @@ TEST_F(DBBasicTest, GetAllKeyVersions) {
     ASSERT_OK(Delete(std::to_string(i)));
   }
   std::vector<KeyVersion> key_versions;
-  ASSERT_OK(rocksdb::GetAllKeyVersions(db_, Slice(), Slice(),
-                                       std::numeric_limits<size_t>::max(),
-                                       &key_versions));
+  ASSERT_OK(ROCKSDB_NAMESPACE::GetAllKeyVersions(
+      db_, Slice(), Slice(), std::numeric_limits<size_t>::max(),
+      &key_versions));
   ASSERT_EQ(kNumInserts + kNumDeletes + kNumUpdates, key_versions.size());
-  ASSERT_OK(rocksdb::GetAllKeyVersions(db_, handles_[0], Slice(), Slice(),
-                                       std::numeric_limits<size_t>::max(),
-                                       &key_versions));
+  ASSERT_OK(ROCKSDB_NAMESPACE::GetAllKeyVersions(
+      db_, handles_[0], Slice(), Slice(), std::numeric_limits<size_t>::max(),
+      &key_versions));
   ASSERT_EQ(kNumInserts + kNumDeletes + kNumUpdates, key_versions.size());
 
   // Check non-default column family
@@ -1593,9 +1688,9 @@ TEST_F(DBBasicTest, GetAllKeyVersions) {
   for (size_t i = 0; i != kNumDeletes - 1; ++i) {
     ASSERT_OK(Delete(1, std::to_string(i)));
   }
-  ASSERT_OK(rocksdb::GetAllKeyVersions(db_, handles_[1], Slice(), Slice(),
-                                       std::numeric_limits<size_t>::max(),
-                                       &key_versions));
+  ASSERT_OK(ROCKSDB_NAMESPACE::GetAllKeyVersions(
+      db_, handles_[1], Slice(), Slice(), std::numeric_limits<size_t>::max(),
+      &key_versions));
   ASSERT_EQ(kNumInserts + kNumDeletes + kNumUpdates - 3, key_versions.size());
 }
 #endif  // !ROCKSDB_LITE
@@ -1706,6 +1801,14 @@ class DBBasicTestWithParallelIO
       assert(Put(Key(i), values_[i]) == Status::OK());
     }
     Flush();
+
+    for (int i = 0; i < 100; ++i) {
+      // block cannot gain space by compression
+      uncompressable_values_.emplace_back(RandomString(&rnd, 256) + '\0');
+      std::string tmp_key = "a" + Key(i);
+      assert(Put(tmp_key, uncompressable_values_[i]) == Status::OK());
+    }
+    Flush();
   }
 
   bool CheckValue(int i, const std::string& value) {
@@ -1715,6 +1818,13 @@ class DBBasicTestWithParallelIO
     return false;
   }
 
+  bool CheckUncompressableValue(int i, const std::string& value) {
+    if (uncompressable_values_[i].compare(value) == 0) {
+      return true;
+    }
+    return false;
+  }
+
   int num_lookups() { return uncompressed_cache_->num_lookups(); }
   int num_found() { return uncompressed_cache_->num_found(); }
   int num_inserts() { return uncompressed_cache_->num_inserts(); }
@@ -1864,9 +1974,12 @@ class DBBasicTestWithParallelIO
   std::shared_ptr<MyBlockCache> uncompressed_cache_;
   bool compression_enabled_;
   std::vector<std::string> values_;
+  std::vector<std::string> uncompressable_values_;
   bool fill_cache_;
 };
 
+// TODO: fails on CircleCI's Windows env
+#ifndef OS_WIN
 TEST_P(DBBasicTestWithParallelIO, MultiGet) {
   std::vector<std::string> key_data(10);
   std::vector<Slice> keys;
@@ -1928,8 +2041,151 @@ TEST_P(DBBasicTestWithParallelIO, MultiGet) {
     ASSERT_OK(statuses[i]);
     ASSERT_TRUE(CheckValue(key_ints[i], values[i].ToString()));
   }
-  expected_reads += (read_from_cache ? 2 : 4);
+  if (compression_enabled() && !has_compressed_cache()) {
+    expected_reads += (read_from_cache ? 2 : 3);
+  } else {
+    expected_reads += (read_from_cache ? 2 : 4);
+  }
   ASSERT_EQ(env_->random_read_counter_.Read(), expected_reads);
+
+  keys.resize(10);
+  statuses.resize(10);
+  std::vector<int> key_uncmp{1, 2, 15, 16, 55, 81, 82, 83, 84, 85};
+  for (size_t i = 0; i < key_uncmp.size(); ++i) {
+    key_data[i] = "a" + Key(key_uncmp[i]);
+    keys[i] = Slice(key_data[i]);
+    statuses[i] = Status::OK();
+    values[i].Reset();
+  }
+  dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
+                     keys.data(), values.data(), statuses.data(), true);
+  for (size_t i = 0; i < key_uncmp.size(); ++i) {
+    ASSERT_OK(statuses[i]);
+    ASSERT_TRUE(CheckUncompressableValue(key_uncmp[i], values[i].ToString()));
+  }
+  if (compression_enabled() && !has_compressed_cache()) {
+    expected_reads += (read_from_cache ? 3 : 3);
+  } else {
+    expected_reads += (read_from_cache ? 4 : 4);
+  }
+  ASSERT_EQ(env_->random_read_counter_.Read(), expected_reads);
+
+  keys.resize(5);
+  statuses.resize(5);
+  std::vector<int> key_tr{1, 2, 15, 16, 55};
+  for (size_t i = 0; i < key_tr.size(); ++i) {
+    key_data[i] = "a" + Key(key_tr[i]);
+    keys[i] = Slice(key_data[i]);
+    statuses[i] = Status::OK();
+    values[i].Reset();
+  }
+  dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
+                     keys.data(), values.data(), statuses.data(), true);
+  for (size_t i = 0; i < key_tr.size(); ++i) {
+    ASSERT_OK(statuses[i]);
+    ASSERT_TRUE(CheckUncompressableValue(key_tr[i], values[i].ToString()));
+  }
+  if (compression_enabled() && !has_compressed_cache()) {
+    expected_reads += (read_from_cache ? 0 : 2);
+    ASSERT_EQ(env_->random_read_counter_.Read(), expected_reads);
+  } else {
+    if (has_uncompressed_cache()) {
+      expected_reads += (read_from_cache ? 0 : 3);
+      ASSERT_EQ(env_->random_read_counter_.Read(), expected_reads);
+    } else {
+      // A rare case, even we enable the block compression but some of data
+      // blocks are not compressed due to content. If user only enable the
+      // compressed cache, the uncompressed blocks will not tbe cached, and
+      // block reads will be triggered. The number of reads is related to
+      // the compression algorithm.
+      ASSERT_TRUE(env_->random_read_counter_.Read() >= expected_reads);
+    }
+  }
+}
+#endif // OS_WIN
+
+TEST_P(DBBasicTestWithParallelIO, MultiGetWithChecksumMismatch) {
+  std::vector<std::string> key_data(10);
+  std::vector<Slice> keys;
+  // We cannot resize a PinnableSlice vector, so just set initial size to
+  // largest we think we will need
+  std::vector<PinnableSlice> values(10);
+  std::vector<Status> statuses;
+  int read_count = 0;
+  ReadOptions ro;
+  ro.fill_cache = fill_cache();
+
+  SyncPoint::GetInstance()->SetCallBack(
+      "RetrieveMultipleBlocks:VerifyChecksum", [&](void *status) {
+      Status* s = static_cast<Status*>(status);
+      read_count++;
+      if (read_count == 2) {
+        *s = Status::Corruption();
+      }
+    });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  // Warm up the cache first
+  key_data.emplace_back(Key(0));
+  keys.emplace_back(Slice(key_data.back()));
+  key_data.emplace_back(Key(50));
+  keys.emplace_back(Slice(key_data.back()));
+  statuses.resize(keys.size());
+
+  dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
+                     keys.data(), values.data(), statuses.data(), true);
+  ASSERT_TRUE(CheckValue(0, values[0].ToString()));
+  //ASSERT_TRUE(CheckValue(50, values[1].ToString()));
+  ASSERT_EQ(statuses[0], Status::OK());
+  ASSERT_EQ(statuses[1], Status::Corruption());
+
+  SyncPoint::GetInstance()->DisableProcessing();
+}
+
+TEST_P(DBBasicTestWithParallelIO, MultiGetWithMissingFile) {
+  std::vector<std::string> key_data(10);
+  std::vector<Slice> keys;
+  // We cannot resize a PinnableSlice vector, so just set initial size to
+  // largest we think we will need
+  std::vector<PinnableSlice> values(10);
+  std::vector<Status> statuses;
+  ReadOptions ro;
+  ro.fill_cache = fill_cache();
+
+  SyncPoint::GetInstance()->SetCallBack(
+      "TableCache::MultiGet:FindTable", [&](void *status) {
+      Status* s = static_cast<Status*>(status);
+      *s = Status::IOError();
+    });
+  // DB open will create table readers unless we reduce the table cache
+  // capacity.
+  // SanitizeOptions will set max_open_files to minimum of 20. Table cache
+  // is allocated with max_open_files - 10 as capacity. So override
+  // max_open_files to 11 so table cache capacity will become 1. This will
+  // prevent file open during DB open and force the file to be opened
+  // during MultiGet
+  SyncPoint::GetInstance()->SetCallBack(
+      "SanitizeOptions::AfterChangeMaxOpenFiles", [&](void *arg) {
+      int* max_open_files = (int*)arg;
+      *max_open_files = 11;
+    });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  Reopen(CurrentOptions());
+
+  // Warm up the cache first
+  key_data.emplace_back(Key(0));
+  keys.emplace_back(Slice(key_data.back()));
+  key_data.emplace_back(Key(50));
+  keys.emplace_back(Slice(key_data.back()));
+  statuses.resize(keys.size());
+
+  dbfull()->MultiGet(ro, dbfull()->DefaultColumnFamily(), keys.size(),
+                     keys.data(), values.data(), statuses.data(), true);
+  ASSERT_EQ(statuses[0], Status::IOError());
+  ASSERT_EQ(statuses[1], Status::IOError());
+
+  SyncPoint::GetInstance()->DisableProcessing();
 }
 
 INSTANTIATE_TEST_CASE_P(
@@ -1942,157 +2198,8 @@ INSTANTIATE_TEST_CASE_P(
     ::testing::Combine(::testing::Bool(), ::testing::Bool(),
                        ::testing::Bool(), ::testing::Bool()));
 
-class DBBasicTestWithTimestampWithParam
-    : public DBTestBase,
-      public testing::WithParamInterface<bool> {
- public:
-  DBBasicTestWithTimestampWithParam()
-      : DBTestBase("/db_basic_test_with_timestamp") {}
 
- protected:
-  class TestComparator : public Comparator {
-   private:
-    const Comparator* cmp_without_ts_;
-
-   public:
-    explicit TestComparator(size_t ts_sz)
-        : Comparator(ts_sz), cmp_without_ts_(nullptr) {
-      cmp_without_ts_ = BytewiseComparator();
-    }
-
-    const char* Name() const override { return "TestComparator"; }
-
-    void FindShortSuccessor(std::string*) const override {}
-
-    void FindShortestSeparator(std::string*, const Slice&) const override {}
-
-    int Compare(const Slice& a, const Slice& b) const override {
-      int r = CompareWithoutTimestamp(a, b);
-      if (r != 0 || 0 == timestamp_size()) {
-        return r;
-      }
-      return CompareTimestamp(
-          Slice(a.data() + a.size() - timestamp_size(), timestamp_size()),
-          Slice(b.data() + b.size() - timestamp_size(), timestamp_size()));
-    }
-
-    int CompareWithoutTimestamp(const Slice& a, const Slice& b) const override {
-      assert(a.size() >= timestamp_size());
-      assert(b.size() >= timestamp_size());
-      Slice k1 = StripTimestampFromUserKey(a, timestamp_size());
-      Slice k2 = StripTimestampFromUserKey(b, timestamp_size());
-
-      return cmp_without_ts_->Compare(k1, k2);
-    }
-
-    int CompareTimestamp(const Slice& ts1, const Slice& ts2) const override {
-      if (!ts1.data() && !ts2.data()) {
-        return 0;
-      } else if (ts1.data() && !ts2.data()) {
-        return 1;
-      } else if (!ts1.data() && ts2.data()) {
-        return -1;
-      }
-      assert(ts1.size() == ts2.size());
-      uint64_t low1 = 0;
-      uint64_t low2 = 0;
-      uint64_t high1 = 0;
-      uint64_t high2 = 0;
-      auto* ptr1 = const_cast<Slice*>(&ts1);
-      auto* ptr2 = const_cast<Slice*>(&ts2);
-      if (!GetFixed64(ptr1, &low1) || !GetFixed64(ptr1, &high1) ||
-          !GetFixed64(ptr2, &low2) || !GetFixed64(ptr2, &high2)) {
-        assert(false);
-      }
-      if (high1 < high2) {
-        return 1;
-      } else if (high1 > high2) {
-        return -1;
-      }
-      if (low1 < low2) {
-        return 1;
-      } else if (low1 > low2) {
-        return -1;
-      }
-      return 0;
-    }
-  };
-
-  Slice EncodeTimestamp(uint64_t low, uint64_t high, std::string* ts) {
-    assert(nullptr != ts);
-    ts->clear();
-    PutFixed64(ts, low);
-    PutFixed64(ts, high);
-    assert(ts->size() == sizeof(low) + sizeof(high));
-    return Slice(*ts);
-  }
-};
-
-TEST_P(DBBasicTestWithTimestampWithParam, PutAndGet) {
-  const int kNumKeysPerFile = 8192;
-  const size_t kNumTimestamps = 6;
-  bool memtable_only = GetParam();
-  Options options = CurrentOptions();
-  options.create_if_missing = true;
-  options.env = env_;
-  options.memtable_factory.reset(new SpecialSkipListFactory(kNumKeysPerFile));
-  std::string tmp;
-  size_t ts_sz = EncodeTimestamp(0, 0, &tmp).size();
-  TestComparator test_cmp(ts_sz);
-  options.comparator = &test_cmp;
-  BlockBasedTableOptions bbto;
-  bbto.filter_policy.reset(NewBloomFilterPolicy(
-      10 /*bits_per_key*/, false /*use_block_based_builder*/));
-  bbto.whole_key_filtering = true;
-  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
-  DestroyAndReopen(options);
-  CreateAndReopenWithCF({"pikachu"}, options);
-  size_t num_cfs = handles_.size();
-  ASSERT_EQ(2, num_cfs);
-  std::vector<std::string> write_ts_strs(kNumTimestamps);
-  std::vector<std::string> read_ts_strs(kNumTimestamps);
-  std::vector<Slice> write_ts_list;
-  std::vector<Slice> read_ts_list;
-
-  for (size_t i = 0; i != kNumTimestamps; ++i) {
-    write_ts_list.emplace_back(EncodeTimestamp(i * 2, 0, &write_ts_strs[i]));
-    read_ts_list.emplace_back(EncodeTimestamp(1 + i * 2, 0, &read_ts_strs[i]));
-    const Slice& write_ts = write_ts_list.back();
-    WriteOptions wopts;
-    wopts.timestamp = &write_ts;
-    for (int cf = 0; cf != static_cast<int>(num_cfs); ++cf) {
-      for (size_t j = 0; j != (kNumKeysPerFile - 1) / kNumTimestamps; ++j) {
-        ASSERT_OK(Put(cf, "key" + std::to_string(j),
-                      "value_" + std::to_string(j) + "_" + std::to_string(i),
-                      wopts));
-      }
-      if (!memtable_only) {
-        ASSERT_OK(Flush(cf));
-      }
-    }
-  }
-  const auto& verify_db_func = [&]() {
-    for (size_t i = 0; i != kNumTimestamps; ++i) {
-      ReadOptions ropts;
-      ropts.timestamp = &read_ts_list[i];
-      for (int cf = 0; cf != static_cast<int>(num_cfs); ++cf) {
-        ColumnFamilyHandle* cfh = handles_[cf];
-        for (size_t j = 0; j != (kNumKeysPerFile - 1) / kNumTimestamps; ++j) {
-          std::string value;
-          ASSERT_OK(db_->Get(ropts, cfh, "key" + std::to_string(j), &value));
-          ASSERT_EQ("value_" + std::to_string(j) + "_" + std::to_string(i),
-                    value);
-        }
-      }
-    }
-  };
-  verify_db_func();
-}
-
-INSTANTIATE_TEST_CASE_P(Timestamp, DBBasicTestWithTimestampWithParam,
-                        ::testing::Bool());
-
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 #ifdef ROCKSDB_UNITTESTS_WITH_CUSTOM_OBJECTS_FROM_STATIC_LIBS
 extern "C" {
@@ -2103,7 +2210,7 @@ void RegisterCustomObjects(int /*argc*/, char** /*argv*/) {}
 #endif  // !ROCKSDB_UNITTESTS_WITH_CUSTOM_OBJECTS_FROM_STATIC_LIBS
 
 int main(int argc, char** argv) {
-  rocksdb::port::InstallStackTraceHandler();
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
   ::testing::InitGoogleTest(&argc, argv);
   RegisterCustomObjects(argc, argv);
   return RUN_ALL_TESTS();

db/db_block_cache_test.cc

@@ -10,8 +10,9 @@
 #include "cache/lru_cache.h"
 #include "db/db_test_util.h"
 #include "port/stack_trace.h"
+#include "util/compression.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class DBBlockCacheTest : public DBTestBase {
  private:
@@ -45,7 +46,7 @@ class DBBlockCacheTest : public DBTestBase {
     options.create_if_missing = true;
     options.avoid_flush_during_recovery = false;
     // options.compression = kNoCompression;
-    options.statistics = rocksdb::CreateDBStatistics();
+    options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
     options.table_factory.reset(new BlockBasedTableFactory(table_options));
     return options;
   }
@@ -291,7 +292,7 @@ TEST_F(DBBlockCacheTest, TestWithCompressedBlockCache) {
 TEST_F(DBBlockCacheTest, IndexAndFilterBlocksOfNewTableAddedToCache) {
   Options options = CurrentOptions();
   options.create_if_missing = true;
-  options.statistics = rocksdb::CreateDBStatistics();
+  options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
   BlockBasedTableOptions table_options;
   table_options.cache_index_and_filter_blocks = true;
   table_options.filter_policy.reset(NewBloomFilterPolicy(20));
@@ -377,7 +378,7 @@ TEST_F(DBBlockCacheTest, FillCacheAndIterateDB) {
 TEST_F(DBBlockCacheTest, IndexAndFilterBlocksStats) {
   Options options = CurrentOptions();
   options.create_if_missing = true;
-  options.statistics = rocksdb::CreateDBStatistics();
+  options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
   BlockBasedTableOptions table_options;
   table_options.cache_index_and_filter_blocks = true;
   LRUCacheOptions co;
@@ -463,7 +464,7 @@ TEST_F(DBBlockCacheTest, IndexAndFilterBlocksCachePriority) {
   for (auto priority : {Cache::Priority::LOW, Cache::Priority::HIGH}) {
     Options options = CurrentOptions();
     options.create_if_missing = true;
-    options.statistics = rocksdb::CreateDBStatistics();
+    options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
     BlockBasedTableOptions table_options;
     table_options.cache_index_and_filter_blocks = true;
     table_options.block_cache.reset(new MockCache());
@@ -519,7 +520,7 @@ TEST_F(DBBlockCacheTest, IndexAndFilterBlocksCachePriority) {
 TEST_F(DBBlockCacheTest, ParanoidFileChecks) {
   Options options = CurrentOptions();
   options.create_if_missing = true;
-  options.statistics = rocksdb::CreateDBStatistics();
+  options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
   options.level0_file_num_compaction_trigger = 2;
   options.paranoid_file_checks = true;
   BlockBasedTableOptions table_options;
@@ -575,7 +576,7 @@ TEST_F(DBBlockCacheTest, CompressedCache) {
   for (int iter = 0; iter < 4; iter++) {
     Options options = CurrentOptions();
     options.write_buffer_size = 64 * 1024;  // small write buffer
-    options.statistics = rocksdb::CreateDBStatistics();
+    options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
 
     BlockBasedTableOptions table_options;
     switch (iter) {
@@ -685,16 +686,18 @@ TEST_F(DBBlockCacheTest, CacheCompressionDict) {
 
   // Try all the available libraries that support dictionary compression
   std::vector<CompressionType> compression_types;
-#ifdef ZLIB
-  compression_types.push_back(kZlibCompression);
-#endif  // ZLIB
-#if LZ4_VERSION_NUMBER >= 10400
-  compression_types.push_back(kLZ4Compression);
-  compression_types.push_back(kLZ4HCCompression);
-#endif  // LZ4_VERSION_NUMBER >= 10400
-#if ZSTD_VERSION_NUMBER >= 500
-  compression_types.push_back(kZSTD);
-#endif  // ZSTD_VERSION_NUMBER >= 500
+  if (Zlib_Supported()) {
+    compression_types.push_back(kZlibCompression);
+  }
+  if (LZ4_Supported()) {
+    compression_types.push_back(kLZ4Compression);
+    compression_types.push_back(kLZ4HCCompression);
+  }
+  if (ZSTD_Supported()) {
+    compression_types.push_back(kZSTD);
+  } else if (ZSTDNotFinal_Supported()) {
+    compression_types.push_back(kZSTDNotFinalCompression);
+  }
   Random rnd(301);
   for (auto compression_type : compression_types) {
     Options options = CurrentOptions();
@@ -702,7 +705,7 @@ TEST_F(DBBlockCacheTest, CacheCompressionDict) {
     options.compression_opts.max_dict_bytes = 4096;
     options.create_if_missing = true;
     options.num_levels = 2;
-    options.statistics = rocksdb::CreateDBStatistics();
+    options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
     options.target_file_size_base = kNumEntriesPerFile * kNumBytesPerEntry;
     BlockBasedTableOptions table_options;
     table_options.cache_index_and_filter_blocks = true;
@@ -749,10 +752,10 @@ TEST_F(DBBlockCacheTest, CacheCompressionDict) {
 
 #endif  // ROCKSDB_LITE
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
-  rocksdb::port::InstallStackTraceHandler();
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
   ::testing::InitGoogleTest(&argc, argv);
   return RUN_ALL_TESTS();
 }

db/db_bloom_filter_test.cc

@@ -12,16 +12,10 @@
 #include "rocksdb/perf_context.h"
 #include "table/block_based/filter_policy_internal.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 namespace {
 using BFP = BloomFilterPolicy;
-
-namespace BFP2 {
-// Extends BFP::Mode with option to use Plain table
-using PseudoMode = int;
-static constexpr PseudoMode kPlainTable = -1;
-}  // namespace BFP2
 }  // namespace
 
 // DB tests related to bloom filter.
@@ -94,7 +88,7 @@ TEST_P(DBBloomFilterTestDefFormatVersion, KeyMayExist) {
       // indexes
       continue;
     }
-    options.statistics = rocksdb::CreateDBStatistics();
+    options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
     CreateAndReopenWithCF({"pikachu"}, options);
 
     ASSERT_TRUE(!db_->KeyMayExist(ropts, handles_[1], "a", &value));
@@ -156,7 +150,7 @@ TEST_F(DBBloomFilterTest, GetFilterByPrefixBloomCustomPrefixExtractor) {
     Options options = last_options_;
     options.prefix_extractor =
         std::make_shared<SliceTransformLimitedDomainGeneric>();
-    options.statistics = rocksdb::CreateDBStatistics();
+    options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
     get_perf_context()->EnablePerLevelPerfContext();
     BlockBasedTableOptions bbto;
     bbto.filter_policy.reset(NewBloomFilterPolicy(10, false));
@@ -222,7 +216,7 @@ TEST_F(DBBloomFilterTest, GetFilterByPrefixBloom) {
   for (bool partition_filters : {true, false}) {
     Options options = last_options_;
     options.prefix_extractor.reset(NewFixedPrefixTransform(8));
-    options.statistics = rocksdb::CreateDBStatistics();
+    options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
     get_perf_context()->EnablePerLevelPerfContext();
     BlockBasedTableOptions bbto;
     bbto.filter_policy.reset(NewBloomFilterPolicy(10, false));
@@ -273,7 +267,7 @@ TEST_F(DBBloomFilterTest, WholeKeyFilterProp) {
   for (bool partition_filters : {true, false}) {
     Options options = last_options_;
     options.prefix_extractor.reset(NewFixedPrefixTransform(3));
-    options.statistics = rocksdb::CreateDBStatistics();
+    options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
     get_perf_context()->EnablePerLevelPerfContext();
 
     BlockBasedTableOptions bbto;
@@ -537,7 +531,7 @@ INSTANTIATE_TEST_CASE_P(
 TEST_F(DBBloomFilterTest, BloomFilterRate) {
   while (ChangeFilterOptions()) {
     Options options = CurrentOptions();
-    options.statistics = rocksdb::CreateDBStatistics();
+    options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
     get_perf_context()->EnablePerLevelPerfContext();
     CreateAndReopenWithCF({"pikachu"}, options);
 
@@ -569,7 +563,7 @@ TEST_F(DBBloomFilterTest, BloomFilterRate) {
 
 TEST_F(DBBloomFilterTest, BloomFilterCompatibility) {
   Options options = CurrentOptions();
-  options.statistics = rocksdb::CreateDBStatistics();
+  options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
   BlockBasedTableOptions table_options;
   table_options.filter_policy.reset(NewBloomFilterPolicy(10, true));
   options.table_factory.reset(NewBlockBasedTableFactory(table_options));
@@ -613,7 +607,7 @@ TEST_F(DBBloomFilterTest, BloomFilterCompatibility) {
 TEST_F(DBBloomFilterTest, BloomFilterReverseCompatibility) {
   for (bool partition_filters : {true, false}) {
     Options options = CurrentOptions();
-    options.statistics = rocksdb::CreateDBStatistics();
+    options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
     BlockBasedTableOptions table_options;
     if (partition_filters) {
       table_options.partition_filters = true;
@@ -660,17 +654,18 @@ class TestingWrappedBlockBasedFilterPolicy : public FilterPolicy {
     return "TestingWrappedBlockBasedFilterPolicy";
   }
 
-  void CreateFilter(const rocksdb::Slice* keys, int n,
+  void CreateFilter(const ROCKSDB_NAMESPACE::Slice* keys, int n,
                     std::string* dst) const override {
-    std::unique_ptr<rocksdb::Slice[]> user_keys(new rocksdb::Slice[n]);
+    std::unique_ptr<ROCKSDB_NAMESPACE::Slice[]> user_keys(
+        new ROCKSDB_NAMESPACE::Slice[n]);
     for (int i = 0; i < n; ++i) {
       user_keys[i] = convertKey(keys[i]);
     }
     return filter_->CreateFilter(user_keys.get(), n, dst);
   }
 
-  bool KeyMayMatch(const rocksdb::Slice& key,
-                   const rocksdb::Slice& filter) const override {
+  bool KeyMayMatch(const ROCKSDB_NAMESPACE::Slice& key,
+                   const ROCKSDB_NAMESPACE::Slice& filter) const override {
     counter_++;
     return filter_->KeyMayMatch(convertKey(key), filter);
   }
@@ -681,13 +676,16 @@ class TestingWrappedBlockBasedFilterPolicy : public FilterPolicy {
   const FilterPolicy* filter_;
   mutable uint32_t counter_;
 
-  rocksdb::Slice convertKey(const rocksdb::Slice& key) const { return key; }
+  ROCKSDB_NAMESPACE::Slice convertKey(
+      const ROCKSDB_NAMESPACE::Slice& key) const {
+    return key;
+  }
 };
 }  // namespace
 
 TEST_F(DBBloomFilterTest, WrappedBlockBasedFilterPolicy) {
   Options options = CurrentOptions();
-  options.statistics = rocksdb::CreateDBStatistics();
+  options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
 
   BlockBasedTableOptions table_options;
   TestingWrappedBlockBasedFilterPolicy* policy =
@@ -803,7 +801,7 @@ class TestingContextCustomFilterPolicy
 TEST_F(DBBloomFilterTest, ContextCustomFilterPolicy) {
   for (bool fifo : {true, false}) {
     Options options = CurrentOptions();
-    options.statistics = rocksdb::CreateDBStatistics();
+    options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
     options.compaction_style =
         fifo ? kCompactionStyleFIFO : kCompactionStyleLevel;
 
@@ -902,7 +900,7 @@ class SliceTransformLimitedDomain : public SliceTransform {
 TEST_F(DBBloomFilterTest, PrefixExtractorFullFilter) {
   BlockBasedTableOptions bbto;
   // Full Filter Block
-  bbto.filter_policy.reset(rocksdb::NewBloomFilterPolicy(10, false));
+  bbto.filter_policy.reset(ROCKSDB_NAMESPACE::NewBloomFilterPolicy(10, false));
   bbto.whole_key_filtering = false;
 
   Options options = CurrentOptions();
@@ -931,7 +929,7 @@ TEST_F(DBBloomFilterTest, PrefixExtractorFullFilter) {
 TEST_F(DBBloomFilterTest, PrefixExtractorBlockFilter) {
   BlockBasedTableOptions bbto;
   // Block Filter Block
-  bbto.filter_policy.reset(rocksdb::NewBloomFilterPolicy(10, true));
+  bbto.filter_policy.reset(ROCKSDB_NAMESPACE::NewBloomFilterPolicy(10, true));
 
   Options options = CurrentOptions();
   options.prefix_extractor = std::make_shared<SliceTransformLimitedDomain>();
@@ -970,7 +968,8 @@ TEST_F(DBBloomFilterTest, MemtableWholeKeyBloomFilter) {
   Options options = CurrentOptions();
   options.memtable_prefix_bloom_size_ratio =
       static_cast<double>(kMemtablePrefixFilterSize) / kMemtableSize;
-  options.prefix_extractor.reset(rocksdb::NewFixedPrefixTransform(kPrefixLen));
+  options.prefix_extractor.reset(
+      ROCKSDB_NAMESPACE::NewFixedPrefixTransform(kPrefixLen));
   options.write_buffer_size = kMemtableSize;
   options.memtable_whole_key_filtering = false;
   Reopen(options);
@@ -1031,6 +1030,14 @@ TEST_F(DBBloomFilterTest, MemtablePrefixBloomOutOfDomain) {
 }
 
 #ifndef ROCKSDB_LITE
+namespace {
+namespace BFP2 {
+// Extends BFP::Mode with option to use Plain table
+using PseudoMode = int;
+static constexpr PseudoMode kPlainTable = -1;
+}  // namespace BFP2
+}  // namespace
+
 class BloomStatsTestWithParam
     : public DBBloomFilterTest,
       public testing::WithParamInterface<std::tuple<BFP2::PseudoMode, bool>> {
@@ -1040,7 +1047,8 @@ class BloomStatsTestWithParam
     partition_filters_ = std::get<1>(GetParam());
 
     options_.create_if_missing = true;
-    options_.prefix_extractor.reset(rocksdb::NewFixedPrefixTransform(4));
+    options_.prefix_extractor.reset(
+        ROCKSDB_NAMESPACE::NewFixedPrefixTransform(4));
     options_.memtable_prefix_bloom_size_ratio =
         8.0 * 1024.0 / static_cast<double>(options_.write_buffer_size);
     if (bfp_impl_ == BFP2::kPlainTable) {
@@ -1321,7 +1329,7 @@ TEST_F(DBBloomFilterTest, OptimizeFiltersForHits) {
   bbto.whole_key_filtering = true;
   options.table_factory.reset(NewBlockBasedTableFactory(bbto));
   options.optimize_filters_for_hits = true;
-  options.statistics = rocksdb::CreateDBStatistics();
+  options.statistics = ROCKSDB_NAMESPACE::CreateDBStatistics();
   get_perf_context()->Reset();
   get_perf_context()->EnablePerLevelPerfContext();
   CreateAndReopenWithCF({"mypikachu"}, options);
@@ -1447,13 +1455,13 @@ TEST_F(DBBloomFilterTest, OptimizeFiltersForHits) {
 
   int32_t trivial_move = 0;
   int32_t non_trivial_move = 0;
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "DBImpl::BackgroundCompaction:TrivialMove",
       [&](void* /*arg*/) { trivial_move++; });
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "DBImpl::BackgroundCompaction:NonTrivial",
       [&](void* /*arg*/) { non_trivial_move++; });
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   CompactRangeOptions compact_options;
   compact_options.bottommost_level_compaction =
@@ -1893,10 +1901,10 @@ TEST_F(DBBloomFilterTest, DynamicBloomFilterOptions) {
 
 #endif  // ROCKSDB_LITE
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
-  rocksdb::port::InstallStackTraceHandler();
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
   ::testing::InitGoogleTest(&argc, argv);
   return RUN_ALL_TESTS();
 }

db/db_compaction_filter_test.cc

@@ -10,7 +10,7 @@
 #include "db/db_test_util.h"
 #include "port/stack_trace.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 static int cfilter_count = 0;
 static int cfilter_skips = 0;
@@ -47,8 +47,7 @@ class DBTestCompactionFilterWithCompactParam
 
 #ifndef ROCKSDB_VALGRIND_RUN
 INSTANTIATE_TEST_CASE_P(
-    DBTestCompactionFilterWithCompactOption,
-    DBTestCompactionFilterWithCompactParam,
+    CompactionFilterWithOption, DBTestCompactionFilterWithCompactParam,
     ::testing::Values(DBTestBase::OptionConfig::kDefault,
                       DBTestBase::OptionConfig::kUniversalCompaction,
                       DBTestBase::OptionConfig::kUniversalCompactionMultiLevel,
@@ -56,7 +55,7 @@ INSTANTIATE_TEST_CASE_P(
                       DBTestBase::OptionConfig::kUniversalSubcompactions));
 #else
 // Run fewer cases in valgrind
-INSTANTIATE_TEST_CASE_P(DBTestCompactionFilterWithCompactOption,
+INSTANTIATE_TEST_CASE_P(CompactionFilterWithOption,
                         DBTestCompactionFilterWithCompactParam,
                         ::testing::Values(DBTestBase::OptionConfig::kDefault));
 #endif  // ROCKSDB_VALGRIND_RUN
@@ -864,10 +863,10 @@ TEST_F(DBTestCompactionFilter, IgnoreSnapshotsFalse) {
   delete options.compaction_filter;
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
-  rocksdb::port::InstallStackTraceHandler();
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
   ::testing::InitGoogleTest(&argc, argv);
   return RUN_ALL_TESTS();
 }

db/db_dynamic_level_test.cc

@@ -16,7 +16,7 @@
 #include "port/port.h"
 #include "port/stack_trace.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 class DBTestDynamicLevel : public DBTestBase {
  public:
   DBTestDynamicLevel() : DBTestBase("/db_dynamic_level_test") {}
@@ -221,10 +221,10 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBase2) {
 
   // Make sure that the compaction starts before the last bit of data is
   // flushed, so that the base level isn't raised to L1.
-  rocksdb::SyncPoint::GetInstance()->LoadDependency({
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency({
       {"CompactionJob::Run():Start", "DynamicLevelMaxBytesBase2:0"},
   });
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   ASSERT_OK(dbfull()->SetOptions({
       {"disable_auto_compactions", "false"},
@@ -235,20 +235,20 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBase2) {
   dbfull()->TEST_WaitForCompact();
   ASSERT_TRUE(db_->GetIntProperty("rocksdb.base-level", &int_prop));
   ASSERT_EQ(2U, int_prop);
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
-  rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
 
   // Write more data until the base level changes to L1. There will be
   // a manual compaction going on at the same time.
-  rocksdb::SyncPoint::GetInstance()->LoadDependency({
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->LoadDependency({
       {"CompactionJob::Run():Start", "DynamicLevelMaxBytesBase2:1"},
       {"DynamicLevelMaxBytesBase2:2", "CompactionJob::Run():End"},
       {"DynamicLevelMaxBytesBase2:compact_range_finish",
        "FlushJob::WriteLevel0Table"},
   });
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
-  rocksdb::port::Thread thread([this] {
+  ROCKSDB_NAMESPACE::port::Thread thread([this] {
     TEST_SYNC_POINT("DynamicLevelMaxBytesBase2:compact_range_start");
     ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
     TEST_SYNC_POINT("DynamicLevelMaxBytesBase2:compact_range_finish");
@@ -265,8 +265,8 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBase2) {
 
   thread.join();
 
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
-  rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
 
   ASSERT_TRUE(db_->GetIntProperty("rocksdb.base-level", &int_prop));
   ASSERT_EQ(1U, int_prop);
@@ -329,16 +329,16 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesCompactRange) {
   ASSERT_TRUE(db_->GetProperty("rocksdb.num-files-at-level2", &str_prop));
   ASSERT_EQ("0", str_prop);
 
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
-  rocksdb::SyncPoint::GetInstance()->ClearAllCallBacks();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->ClearAllCallBacks();
 
   std::set<int> output_levels;
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "CompactionPicker::CompactRange:Return", [&](void* arg) {
         Compaction* compaction = reinterpret_cast<Compaction*>(arg);
         output_levels.insert(compaction->output_level());
       });
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr);
   ASSERT_EQ(output_levels.size(), 2);
@@ -373,10 +373,10 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBaseInc) {
   DestroyAndReopen(options);
 
   int non_trivial = 0;
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "DBImpl::BackgroundCompaction:NonTrivial",
       [&](void* /*arg*/) { non_trivial++; });
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   Random rnd(301);
   const int total_keys = 3000;
@@ -388,7 +388,7 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBaseInc) {
   }
   Flush();
   dbfull()->TEST_WaitForCompact();
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
 
   ASSERT_EQ(non_trivial, 0);
 
@@ -457,7 +457,7 @@ TEST_F(DBTestDynamicLevel, DISABLED_MigrateToDynamicLevelMaxBytesBase) {
   compaction_finished = false;
   // Issue manual compaction in one thread and still verify DB state
   // in main thread.
-  rocksdb::port::Thread t([&]() {
+  ROCKSDB_NAMESPACE::port::Thread t([&]() {
     CompactRangeOptions compact_options;
     compact_options.change_level = true;
     compact_options.target_level = options.num_levels - 1;
@@ -488,13 +488,13 @@ TEST_F(DBTestDynamicLevel, DISABLED_MigrateToDynamicLevelMaxBytesBase) {
   ASSERT_EQ(NumTableFilesAtLevel(1), 0);
   ASSERT_EQ(NumTableFilesAtLevel(2), 0);
 }
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 #endif  // !defined(ROCKSDB_LITE)
 
 int main(int argc, char** argv) {
 #if !defined(ROCKSDB_LITE)
-  rocksdb::port::InstallStackTraceHandler();
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
   ::testing::InitGoogleTest(&argc, argv);
   return RUN_ALL_TESTS();
 #else

db/db_encryption_test.cc

@@ -12,7 +12,7 @@
 #include <iostream>
 #include <string>
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class DBEncryptionTest : public DBTestBase {
  public:
@@ -113,10 +113,10 @@ TEST_F(DBEncryptionTest, ReadEmptyFile) {
 
 #endif // ROCKSDB_LITE
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
-  rocksdb::port::InstallStackTraceHandler();
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
   ::testing::InitGoogleTest(&argc, argv);
   return RUN_ALL_TESTS();
 }

db/db_filesnapshot.cc

@@ -21,7 +21,7 @@
 #include "test_util/sync_point.h"
 #include "util/mutexlock.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 Status DBImpl::DisableFileDeletions() {
   InstrumentedMutexLock l(&mutex_);
@@ -102,7 +102,7 @@ Status DBImpl::GetLiveFiles(std::vector<std::string>& ret,
         TEST_SYNC_POINT("DBImpl::GetLiveFiles:1");
         TEST_SYNC_POINT("DBImpl::GetLiveFiles:2");
         mutex_.Lock();
-        cfd->Unref();
+        cfd->UnrefAndTryDelete();
         if (!status.ok()) {
           break;
         }
@@ -172,6 +172,6 @@ Status DBImpl::GetCurrentWalFile(std::unique_ptr<LogFile>* current_log_file) {
 
   return wal_manager_.GetLiveWalFile(current_logfile_number, current_log_file);
 }
-}
+}  // namespace ROCKSDB_NAMESPACE
 
 #endif  // ROCKSDB_LITE

db/db_flush_test.cc

@@ -18,7 +18,7 @@
 #include "util/cast_util.h"
 #include "util/mutexlock.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class DBFlushTest : public DBTestBase {
  public:
@@ -212,6 +212,30 @@ TEST_F(DBFlushTest, ManualFlushWithMinWriteBufferNumberToMerge) {
   t.join();
 }
 
+TEST_F(DBFlushTest, ScheduleOnlyOneBgThread) {
+  Options options = CurrentOptions();
+  Reopen(options);
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+  int called = 0;
+  SyncPoint::GetInstance()->SetCallBack(
+      "DBImpl::MaybeScheduleFlushOrCompaction:AfterSchedule:0", [&](void* arg) {
+        ASSERT_NE(nullptr, arg);
+        auto unscheduled_flushes = *reinterpret_cast<int*>(arg);
+        ASSERT_EQ(0, unscheduled_flushes);
+        ++called;
+      });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  ASSERT_OK(Put("a", "foo"));
+  FlushOptions flush_opts;
+  ASSERT_OK(dbfull()->Flush(flush_opts));
+  ASSERT_EQ(1, called);
+
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+}
+
 TEST_P(DBFlushDirectIOTest, DirectIO) {
   Options options;
   options.create_if_missing = true;
@@ -717,15 +741,44 @@ TEST_P(DBAtomicFlushTest, CFDropRaceWithWaitForFlushMemTables) {
   SyncPoint::GetInstance()->DisableProcessing();
 }
 
+TEST_P(DBAtomicFlushTest, RollbackAfterFailToInstallResults) {
+  bool atomic_flush = GetParam();
+  if (!atomic_flush) {
+    return;
+  }
+  auto fault_injection_env = std::make_shared<FaultInjectionTestEnv>(env_);
+  Options options = CurrentOptions();
+  options.env = fault_injection_env.get();
+  options.create_if_missing = true;
+  options.atomic_flush = atomic_flush;
+  CreateAndReopenWithCF({"pikachu"}, options);
+  ASSERT_EQ(2, handles_.size());
+  for (size_t cf = 0; cf < handles_.size(); ++cf) {
+    ASSERT_OK(Put(static_cast<int>(cf), "a", "value"));
+  }
+  SyncPoint::GetInstance()->DisableProcessing();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+  SyncPoint::GetInstance()->SetCallBack(
+      "VersionSet::ProcessManifestWrites:BeforeWriteLastVersionEdit:0",
+      [&](void* /*arg*/) { fault_injection_env->SetFilesystemActive(false); });
+  SyncPoint::GetInstance()->EnableProcessing();
+  FlushOptions flush_opts;
+  Status s = db_->Flush(flush_opts, handles_);
+  ASSERT_NOK(s);
+  fault_injection_env->SetFilesystemActive(true);
+  Close();
+  SyncPoint::GetInstance()->ClearAllCallBacks();
+}
+
 INSTANTIATE_TEST_CASE_P(DBFlushDirectIOTest, DBFlushDirectIOTest,
                         testing::Bool());
 
 INSTANTIATE_TEST_CASE_P(DBAtomicFlushTest, DBAtomicFlushTest, testing::Bool());
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
-  rocksdb::port::InstallStackTraceHandler();
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
   ::testing::InitGoogleTest(&argc, argv);
   return RUN_ALL_TESTS();
 }

db/db_impl/db_impl.cc

@@ -21,7 +21,6 @@
 #include <stdexcept>
 #include <string>
 #include <unordered_map>
-#include <unordered_set>
 #include <utility>
 #include <vector>
 
@@ -52,6 +51,7 @@
 #include "db/version_set.h"
 #include "db/write_batch_internal.h"
 #include "db/write_callback.h"
+#include "env/composite_env_wrapper.h"
 #include "file/file_util.h"
 #include "file/filename.h"
 #include "file/random_access_file_reader.h"
@@ -101,7 +101,8 @@
 #include "util/stop_watch.h"
 #include "util/string_util.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
+
 const std::string kDefaultColumnFamilyName("default");
 const std::string kPersistentStatsColumnFamilyName(
     "___rocksdb_stats_history___");
@@ -141,16 +142,15 @@ void DumpSupportInfo(Logger* logger) {
   ROCKS_LOG_HEADER(logger, "Fast CRC32 supported: %s",
                    crc32c::IsFastCrc32Supported().c_str());
 }
-
-int64_t kDefaultLowPriThrottledRate = 2 * 1024 * 1024;
 }  // namespace
 
 DBImpl::DBImpl(const DBOptions& options, const std::string& dbname,
                const bool seq_per_batch, const bool batch_per_txn)
-    : env_(options.env),
-      dbname_(dbname),
+    : dbname_(dbname),
       own_info_log_(options.info_log == nullptr),
       initial_db_options_(SanitizeOptions(dbname, options)),
+      env_(initial_db_options_.env),
+      fs_(initial_db_options_.file_system),
       immutable_db_options_(initial_db_options_),
       mutable_db_options_(initial_db_options_),
       stats_(immutable_db_options_.statistics.get()),
@@ -158,9 +158,9 @@ DBImpl::DBImpl(const DBOptions& options, const std::string& dbname,
              immutable_db_options_.use_adaptive_mutex),
       default_cf_handle_(nullptr),
       max_total_in_memory_state_(0),
-      env_options_(BuildDBOptions(immutable_db_options_, mutable_db_options_)),
-      env_options_for_compaction_(env_->OptimizeForCompactionTableWrite(
-          env_options_, immutable_db_options_)),
+      file_options_(BuildDBOptions(immutable_db_options_, mutable_db_options_)),
+      file_options_for_compaction_(fs_->OptimizeForCompactionTableWrite(
+          file_options_, immutable_db_options_)),
       seq_per_batch_(seq_per_batch),
       batch_per_txn_(batch_per_txn),
       db_lock_(nullptr),
@@ -178,11 +178,6 @@ DBImpl::DBImpl(const DBOptions& options, const std::string& dbname,
       write_thread_(immutable_db_options_),
       nonmem_write_thread_(immutable_db_options_),
       write_controller_(mutable_db_options_.delayed_write_rate),
-      // Use delayed_write_rate as a base line to determine the initial
-      // low pri write rate limit. It may be adjusted later.
-      low_pri_write_rate_limiter_(NewGenericRateLimiter(std::min(
-          static_cast<int64_t>(mutable_db_options_.delayed_write_rate / 8),
-          kDefaultLowPriThrottledRate))),
       last_batch_group_size_(0),
       unscheduled_flushes_(0),
       unscheduled_compactions_(0),
@@ -201,7 +196,7 @@ DBImpl::DBImpl(const DBOptions& options, const std::string& dbname,
       unable_to_release_oldest_log_(false),
       num_running_ingest_file_(0),
 #ifndef ROCKSDB_LITE
-      wal_manager_(immutable_db_options_, env_options_, seq_per_batch),
+      wal_manager_(immutable_db_options_, file_options_, seq_per_batch),
 #endif  // ROCKSDB_LITE
       event_logger_(immutable_db_options_.info_log.get()),
       bg_work_paused_(0),
@@ -247,7 +242,7 @@ DBImpl::DBImpl(const DBOptions& options, const std::string& dbname,
   co.metadata_charge_policy = kDontChargeCacheMetadata;
   table_cache_ = NewLRUCache(co);
 
-  versions_.reset(new VersionSet(dbname_, &immutable_db_options_, env_options_,
+  versions_.reset(new VersionSet(dbname_, &immutable_db_options_, file_options_,
                                  table_cache_.get(), write_buffer_manager_,
                                  &write_controller_, &block_cache_tracer_));
   column_family_memtables_.reset(
@@ -337,7 +332,7 @@ Status DBImpl::ResumeImpl() {
         mutex_.Unlock();
         s = FlushMemTable(cfd, flush_opts, FlushReason::kErrorRecovery);
         mutex_.Lock();
-        cfd->Unref();
+        cfd->UnrefAndTryDelete();
         if (!s.ok()) {
           break;
         }
@@ -425,7 +420,7 @@ void DBImpl::CancelAllBackgroundWork(bool wait) {
           mutex_.Unlock();
           FlushMemTable(cfd, FlushOptions(), FlushReason::kShutDown);
           mutex_.Lock();
-          cfd->Unref();
+          cfd->UnrefAndTryDelete();
         }
       }
     }
@@ -482,17 +477,12 @@ Status DBImpl::CloseHelper() {
   while (!flush_queue_.empty()) {
     const FlushRequest& flush_req = PopFirstFromFlushQueue();
     for (const auto& iter : flush_req) {
-      ColumnFamilyData* cfd = iter.first;
-      if (cfd->Unref()) {
-        delete cfd;
-      }
+      iter.first->UnrefAndTryDelete();
     }
   }
   while (!compaction_queue_.empty()) {
     auto cfd = PopFirstFromCompactionQueue();
-    if (cfd->Unref()) {
-      delete cfd;
-    }
+    cfd->UnrefAndTryDelete();
   }
 
   if (default_cf_handle_ != nullptr || persist_stats_cf_handle_ != nullptr) {
@@ -599,6 +589,7 @@ Status DBImpl::CloseHelper() {
       ret = s;
     }
   }
+
   if (ret.IsAborted()) {
     // Reserve IsAborted() error for those where users didn't release
     // certain resource and they can release them and come back and
@@ -652,7 +643,7 @@ void DBImpl::StartTimedTasks() {
     stats_dump_period_sec = mutable_db_options_.stats_dump_period_sec;
     if (stats_dump_period_sec > 0) {
       if (!thread_dump_stats_) {
-        thread_dump_stats_.reset(new rocksdb::RepeatableThread(
+        thread_dump_stats_.reset(new ROCKSDB_NAMESPACE::RepeatableThread(
             [this]() { DBImpl::DumpStats(); }, "dump_st", env_,
             static_cast<uint64_t>(stats_dump_period_sec) * kMicrosInSecond));
       }
@@ -660,7 +651,7 @@ void DBImpl::StartTimedTasks() {
     stats_persist_period_sec = mutable_db_options_.stats_persist_period_sec;
     if (stats_persist_period_sec > 0) {
       if (!thread_persist_stats_) {
-        thread_persist_stats_.reset(new rocksdb::RepeatableThread(
+        thread_persist_stats_.reset(new ROCKSDB_NAMESPACE::RepeatableThread(
             [this]() { DBImpl::PersistStats(); }, "pst_st", env_,
             static_cast<uint64_t>(stats_persist_period_sec) * kMicrosInSecond));
       }
@@ -879,7 +870,7 @@ Status DBImpl::TablesRangeTombstoneSummary(ColumnFamilyHandle* column_family,
           column_family);
   ColumnFamilyData* cfd = cfh->cfd();
 
-  SuperVersion* super_version = cfd->GetReferencedSuperVersion(&mutex_);
+  SuperVersion* super_version = cfd->GetReferencedSuperVersion(this);
   Version* version = super_version->current;
 
   Status s =
@@ -895,13 +886,12 @@ void DBImpl::ScheduleBgLogWriterClose(JobContext* job_context) {
       AddToLogsToFreeQueue(l);
     }
     job_context->logs_to_free.clear();
-    SchedulePurge();
   }
 }
 
-Directory* DBImpl::GetDataDir(ColumnFamilyData* cfd, size_t path_id) const {
+FSDirectory* DBImpl::GetDataDir(ColumnFamilyData* cfd, size_t path_id) const {
   assert(cfd);
-  Directory* ret_dir = cfd->GetDataDir(path_id);
+  FSDirectory* ret_dir = cfd->GetDataDir(path_id);
   if (ret_dir == nullptr) {
     return directories_.GetDataDir(path_id);
   }
@@ -1014,12 +1004,36 @@ Status DBImpl::SetDBOptions(
       }
     }
     if (s.ok()) {
-      if (new_options.max_background_compactions >
-          mutable_db_options_.max_background_compactions) {
-        env_->IncBackgroundThreadsIfNeeded(
-            new_options.max_background_compactions, Env::Priority::LOW);
+      const BGJobLimits current_bg_job_limits =
+          GetBGJobLimits(immutable_db_options_.max_background_flushes,
+                         mutable_db_options_.max_background_compactions,
+                         mutable_db_options_.max_background_jobs,
+                         /* parallelize_compactions */ true);
+      const BGJobLimits new_bg_job_limits = GetBGJobLimits(
+          immutable_db_options_.max_background_flushes,
+          new_options.max_background_compactions,
+          new_options.max_background_jobs, /* parallelize_compactions */ true);
+
+      const bool max_flushes_increased =
+          new_bg_job_limits.max_flushes > current_bg_job_limits.max_flushes;
+      const bool max_compactions_increased =
+          new_bg_job_limits.max_compactions >
+          current_bg_job_limits.max_compactions;
+
+      if (max_flushes_increased || max_compactions_increased) {
+        if (max_flushes_increased) {
+          env_->IncBackgroundThreadsIfNeeded(new_bg_job_limits.max_flushes,
+                                             Env::Priority::HIGH);
+        }
+
+        if (max_compactions_increased) {
+          env_->IncBackgroundThreadsIfNeeded(new_bg_job_limits.max_compactions,
+                                             Env::Priority::LOW);
+        }
+
         MaybeScheduleFlushOrCompaction();
       }
+
       if (new_options.stats_dump_period_sec !=
           mutable_db_options_.stats_dump_period_sec) {
         if (thread_dump_stats_) {
@@ -1028,7 +1042,7 @@ Status DBImpl::SetDBOptions(
           mutex_.Lock();
         }
         if (new_options.stats_dump_period_sec > 0) {
-          thread_dump_stats_.reset(new rocksdb::RepeatableThread(
+          thread_dump_stats_.reset(new ROCKSDB_NAMESPACE::RepeatableThread(
               [this]() { DBImpl::DumpStats(); }, "dump_st", env_,
               static_cast<uint64_t>(new_options.stats_dump_period_sec) *
                   kMicrosInSecond));
@@ -1044,7 +1058,7 @@ Status DBImpl::SetDBOptions(
           mutex_.Lock();
         }
         if (new_options.stats_persist_period_sec > 0) {
-          thread_persist_stats_.reset(new rocksdb::RepeatableThread(
+          thread_persist_stats_.reset(new ROCKSDB_NAMESPACE::RepeatableThread(
               [this]() { DBImpl::PersistStats(); }, "pst_st", env_,
               static_cast<uint64_t>(new_options.stats_persist_period_sec) *
                   kMicrosInSecond));
@@ -1060,14 +1074,14 @@ Status DBImpl::SetDBOptions(
       wal_changed = mutable_db_options_.wal_bytes_per_sync !=
                     new_options.wal_bytes_per_sync;
       mutable_db_options_ = new_options;
-      env_options_for_compaction_ = EnvOptions(new_db_options);
-      env_options_for_compaction_ = env_->OptimizeForCompactionTableWrite(
-          env_options_for_compaction_, immutable_db_options_);
-      versions_->ChangeEnvOptions(mutable_db_options_);
+      file_options_for_compaction_ = FileOptions(new_db_options);
+      file_options_for_compaction_ = fs_->OptimizeForCompactionTableWrite(
+          file_options_for_compaction_, immutable_db_options_);
+      versions_->ChangeFileOptions(mutable_db_options_);
       //TODO(xiez): clarify why apply optimize for read to write options
-      env_options_for_compaction_ = env_->OptimizeForCompactionTableRead(
-          env_options_for_compaction_, immutable_db_options_);
-      env_options_for_compaction_.compaction_readahead_size =
+      file_options_for_compaction_ = fs_->OptimizeForCompactionTableRead(
+          file_options_for_compaction_, immutable_db_options_);
+      file_options_for_compaction_.compaction_readahead_size =
           mutable_db_options_.compaction_readahead_size;
       WriteThread::Writer w;
       write_thread_.EnterUnbatched(&w, &mutex_);
@@ -1209,7 +1223,7 @@ Status DBImpl::SyncWAL() {
     }
   }
   if (status.ok() && need_log_dir_sync) {
-    status = directories_.GetWalDir()->Fsync();
+    status = directories_.GetWalDir()->Fsync(IOOptions(), nullptr);
   }
   TEST_SYNC_POINT("DBWALTest::SyncWALNotWaitWrite:2");
 
@@ -1322,19 +1336,34 @@ void DBImpl::BackgroundCallPurge() {
     delete log_writer;
     mutex_.Lock();
   }
-  for (const auto& file : purge_files_) {
-    const PurgeFileInfo& purge_file = file.second;
+  while (!superversions_to_free_queue_.empty()) {
+    assert(!superversions_to_free_queue_.empty());
+    SuperVersion* sv = superversions_to_free_queue_.front();
+    superversions_to_free_queue_.pop_front();
+    mutex_.Unlock();
+    delete sv;
+    mutex_.Lock();
+  }
+
+  // Can't use iterator to go over purge_files_ because inside the loop we're
+  // unlocking the mutex that protects purge_files_.
+  while (!purge_files_.empty()) {
+    auto it = purge_files_.begin();
+    // Need to make a copy of the PurgeFilesInfo before unlocking the mutex.
+    PurgeFileInfo purge_file = it->second;
+
     const std::string& fname = purge_file.fname;
     const std::string& dir_to_sync = purge_file.dir_to_sync;
     FileType type = purge_file.type;
     uint64_t number = purge_file.number;
     int job_id = purge_file.job_id;
 
+    purge_files_.erase(it);
+
     mutex_.Unlock();
     DeleteObsoleteFileImpl(job_id, fname, dir_to_sync, type, number);
     mutex_.Lock();
   }
-  purge_files_.clear();
 
   bg_purge_scheduled_--;
 
@@ -1374,10 +1403,14 @@ static void CleanupIteratorState(void* arg1, void* /*arg2*/) {
     state->db->FindObsoleteFiles(&job_context, false, true);
     if (state->background_purge) {
       state->db->ScheduleBgLogWriterClose(&job_context);
+      state->db->AddSuperVersionsToFreeQueue(state->super_version);
+      state->db->SchedulePurge();
     }
     state->mu->Unlock();
 
-    delete state->super_version;
+    if (!state->background_purge) {
+      delete state->super_version;
+    }
     if (job_context.HaveSomethingToDelete()) {
       if (state->background_purge) {
         // PurgeObsoleteFiles here does not delete files. Instead, it adds the
@@ -1434,7 +1467,7 @@ InternalIterator* DBImpl::NewInternalIterator(const ReadOptions& read_options,
   if (s.ok()) {
     // Collect iterators for files in L0 - Ln
     if (read_options.read_tier != kMemtableTier) {
-      super_version->current->AddIterators(read_options, env_options_,
+      super_version->current->AddIterators(read_options, file_options_,
                                            &merge_iter_builder, range_del_agg);
     }
     internal_iter = merge_iter_builder.Finish();
@@ -1462,14 +1495,22 @@ ColumnFamilyHandle* DBImpl::PersistentStatsColumnFamily() const {
 Status DBImpl::Get(const ReadOptions& read_options,
                    ColumnFamilyHandle* column_family, const Slice& key,
                    PinnableSlice* value) {
+  return Get(read_options, column_family, key, value, /*timestamp=*/nullptr);
+}
+
+Status DBImpl::Get(const ReadOptions& read_options,
+                   ColumnFamilyHandle* column_family, const Slice& key,
+                   PinnableSlice* value, std::string* timestamp) {
   GetImplOptions get_impl_options;
   get_impl_options.column_family = column_family;
   get_impl_options.value = value;
-  return GetImpl(read_options, key, get_impl_options);
+  get_impl_options.timestamp = timestamp;
+  Status s = GetImpl(read_options, key, get_impl_options);
+  return s;
 }
 
 Status DBImpl::GetImpl(const ReadOptions& read_options, const Slice& key,
-                       GetImplOptions get_impl_options) {
+                       GetImplOptions& get_impl_options) {
   assert(get_impl_options.value != nullptr ||
          get_impl_options.merge_operands != nullptr);
   PERF_CPU_TIMER_GUARD(get_cpu_nanos, env_);
@@ -1549,10 +1590,14 @@ Status DBImpl::GetImpl(const ReadOptions& read_options, const Slice& key,
   bool skip_memtable = (read_options.read_tier == kPersistedTier &&
                         has_unpersisted_data_.load(std::memory_order_relaxed));
   bool done = false;
+  const Comparator* comparator =
+      get_impl_options.column_family->GetComparator();
+  size_t ts_sz = comparator->timestamp_size();
+  std::string* timestamp = ts_sz > 0 ? get_impl_options.timestamp : nullptr;
   if (!skip_memtable) {
     // Get value associated with key
     if (get_impl_options.get_value) {
-      if (sv->mem->Get(lkey, get_impl_options.value->GetSelf(), &s,
+      if (sv->mem->Get(lkey, get_impl_options.value->GetSelf(), timestamp, &s,
                        &merge_context, &max_covering_tombstone_seq,
                        read_options, get_impl_options.callback,
                        get_impl_options.is_blob_index)) {
@@ -1560,9 +1605,10 @@ Status DBImpl::GetImpl(const ReadOptions& read_options, const Slice& key,
         get_impl_options.value->PinSelf();
         RecordTick(stats_, MEMTABLE_HIT);
       } else if ((s.ok() || s.IsMergeInProgress()) &&
-                 sv->imm->Get(lkey, get_impl_options.value->GetSelf(), &s,
-                              &merge_context, &max_covering_tombstone_seq,
-                              read_options, get_impl_options.callback,
+                 sv->imm->Get(lkey, get_impl_options.value->GetSelf(),
+                              timestamp, &s, &merge_context,
+                              &max_covering_tombstone_seq, read_options,
+                              get_impl_options.callback,
                               get_impl_options.is_blob_index)) {
         done = true;
         get_impl_options.value->PinSelf();
@@ -1571,9 +1617,9 @@ Status DBImpl::GetImpl(const ReadOptions& read_options, const Slice& key,
     } else {
       // Get Merge Operands associated with key, Merge Operands should not be
       // merged and raw values should be returned to the user.
-      if (sv->mem->Get(lkey, nullptr, &s, &merge_context,
-                       &max_covering_tombstone_seq, read_options, nullptr,
-                       nullptr, false)) {
+      if (sv->mem->Get(lkey, /*value*/ nullptr, /*timestamp=*/nullptr, &s,
+                       &merge_context, &max_covering_tombstone_seq,
+                       read_options, nullptr, nullptr, false)) {
         done = true;
         RecordTick(stats_, MEMTABLE_HIT);
       } else if ((s.ok() || s.IsMergeInProgress()) &&
@@ -1592,8 +1638,8 @@ Status DBImpl::GetImpl(const ReadOptions& read_options, const Slice& key,
   if (!done) {
     PERF_TIMER_GUARD(get_from_output_files_time);
     sv->current->Get(
-        read_options, lkey, get_impl_options.value, &s, &merge_context,
-        &max_covering_tombstone_seq,
+        read_options, lkey, get_impl_options.value, timestamp, &s,
+        &merge_context, &max_covering_tombstone_seq,
         get_impl_options.get_value ? get_impl_options.value_found : nullptr,
         nullptr, nullptr,
         get_impl_options.get_value ? get_impl_options.callback : nullptr,
@@ -1704,13 +1750,14 @@ std::vector<Status> DBImpl::MultiGet(
          has_unpersisted_data_.load(std::memory_order_relaxed));
     bool done = false;
     if (!skip_memtable) {
-      if (super_version->mem->Get(lkey, value, &s, &merge_context,
-                                  &max_covering_tombstone_seq, read_options)) {
+      if (super_version->mem->Get(lkey, value, /*timestamp=*/nullptr, &s,
+                                  &merge_context, &max_covering_tombstone_seq,
+                                  read_options)) {
         done = true;
         RecordTick(stats_, MEMTABLE_HIT);
-      } else if (super_version->imm->Get(lkey, value, &s, &merge_context,
-                                         &max_covering_tombstone_seq,
-                                         read_options)) {
+      } else if (super_version->imm->Get(
+                     lkey, value, nullptr, &s, &merge_context,
+                     &max_covering_tombstone_seq, read_options)) {
         done = true;
         RecordTick(stats_, MEMTABLE_HIT);
       }
@@ -1718,8 +1765,9 @@ std::vector<Status> DBImpl::MultiGet(
     if (!done) {
       PinnableSlice pinnable_val;
       PERF_TIMER_GUARD(get_from_output_files_time);
-      super_version->current->Get(read_options, lkey, &pinnable_val, &s,
-                                  &merge_context, &max_covering_tombstone_seq);
+      super_version->current->Get(read_options, lkey, &pinnable_val,
+                                  /*timestamp=*/nullptr, &s, &merge_context,
+                                  &max_covering_tombstone_seq);
       value->assign(pinnable_val.data(), pinnable_val.size());
       RecordTick(stats_, MEMTABLE_MISS);
     }
@@ -2267,7 +2315,8 @@ Status DBImpl::CreateColumnFamilyImpl(const ColumnFamilyOptions& cf_options,
       auto* cfd =
           versions_->GetColumnFamilySet()->GetColumnFamily(column_family_name);
       assert(cfd != nullptr);
-      s = cfd->AddDirectories();
+      std::map<std::string, std::shared_ptr<FSDirectory>> dummy_created_dirs;
+      s = cfd->AddDirectories(&dummy_created_dirs);
     }
     if (s.ok()) {
       single_column_family_mode_ = false;
@@ -2401,7 +2450,8 @@ Status DBImpl::DropColumnFamilyImpl(ColumnFamilyHandle* column_family) {
 
 bool DBImpl::KeyMayExist(const ReadOptions& read_options,
                          ColumnFamilyHandle* column_family, const Slice& key,
-                         std::string* value, bool* value_found) {
+                         std::string* value, std::string* timestamp,
+                         bool* value_found) {
   assert(value != nullptr);
   if (value_found != nullptr) {
     // falsify later if key-may-exist but can't fetch value
@@ -2414,6 +2464,7 @@ bool DBImpl::KeyMayExist(const ReadOptions& read_options,
   get_impl_options.column_family = column_family;
   get_impl_options.value = &pinnable_val;
   get_impl_options.value_found = value_found;
+  get_impl_options.timestamp = timestamp;
   auto s = GetImpl(roptions, key, get_impl_options);
   value->assign(pinnable_val.data(), pinnable_val.size());
 
@@ -2452,7 +2503,7 @@ Iterator* DBImpl::NewIterator(const ReadOptions& read_options,
     result = nullptr;
 
 #else
-    SuperVersion* sv = cfd->GetReferencedSuperVersion(&mutex_);
+    SuperVersion* sv = cfd->GetReferencedSuperVersion(this);
     auto iter = new ForwardIterator(this, read_options, cfd, sv);
     result = NewDBIterator(
         env_, read_options, *cfd->ioptions(), sv->mutable_cf_options,
@@ -2478,7 +2529,7 @@ ArenaWrappedDBIter* DBImpl::NewIteratorImpl(const ReadOptions& read_options,
                                             ReadCallback* read_callback,
                                             bool allow_blob,
                                             bool allow_refresh) {
-  SuperVersion* sv = cfd->GetReferencedSuperVersion(&mutex_);
+  SuperVersion* sv = cfd->GetReferencedSuperVersion(this);
 
   // Try to generate a DB iterator tree in continuous memory area to be
   // cache friendly. Here is an example of result:
@@ -2526,7 +2577,7 @@ ArenaWrappedDBIter* DBImpl::NewIteratorImpl(const ReadOptions& read_options,
       env_, read_options, *cfd->ioptions(), sv->mutable_cf_options, snapshot,
       sv->mutable_cf_options.max_sequential_skip_in_iterations,
       sv->version_number, read_callback, this, cfd, allow_blob,
-      ((read_options.snapshot != nullptr) ? false : allow_refresh));
+      read_options.snapshot != nullptr ? false : allow_refresh);
 
   InternalIterator* internal_iter =
       NewInternalIterator(read_options, cfd, sv, db_iter->GetArena(),
@@ -2557,7 +2608,7 @@ Status DBImpl::NewIterators(
 #else
     for (auto cfh : column_families) {
       auto cfd = reinterpret_cast<ColumnFamilyHandleImpl*>(cfh)->cfd();
-      SuperVersion* sv = cfd->GetReferencedSuperVersion(&mutex_);
+      SuperVersion* sv = cfd->GetReferencedSuperVersion(this);
       auto iter = new ForwardIterator(this, read_options, cfd, sv);
       iterators->push_back(NewDBIterator(
           env_, read_options, *cfd->ioptions(), sv->mutable_cf_options,
@@ -2729,6 +2780,15 @@ const std::string& DBImpl::GetName() const { return dbname_; }
 
 Env* DBImpl::GetEnv() const { return env_; }
 
+FileSystem* DB::GetFileSystem() const {
+  static LegacyFileSystemWrapper fs_wrap(GetEnv());
+  return &fs_wrap;
+}
+
+FileSystem* DBImpl::GetFileSystem() const {
+  return immutable_db_options_.fs.get();
+}
+
 Options DBImpl::GetOptions(ColumnFamilyHandle* column_family) const {
   InstrumentedMutexLock l(&mutex_);
   auto cfh = reinterpret_cast<ColumnFamilyHandleImpl*>(column_family);
@@ -2884,7 +2944,7 @@ bool DBImpl::GetAggregatedIntProperty(const Slice& property,
 
 SuperVersion* DBImpl::GetAndRefSuperVersion(ColumnFamilyData* cfd) {
   // TODO(ljin): consider using GetReferencedSuperVersion() directly
-  return cfd->GetThreadLocalSuperVersion(&mutex_);
+  return cfd->GetThreadLocalSuperVersion(this);
 }
 
 // REQUIRED: this function should only be called on the write thread or if the
@@ -2902,11 +2962,19 @@ SuperVersion* DBImpl::GetAndRefSuperVersion(uint32_t column_family_id) {
 void DBImpl::CleanupSuperVersion(SuperVersion* sv) {
   // Release SuperVersion
   if (sv->Unref()) {
+    bool defer_purge =
+            immutable_db_options().avoid_unnecessary_blocking_io;
     {
       InstrumentedMutexLock l(&mutex_);
       sv->Cleanup();
+      if (defer_purge) {
+        AddSuperVersionsToFreeQueue(sv);
+        SchedulePurge();
+      }
+    }
+    if (!defer_purge) {
+      delete sv;
     }
-    delete sv;
     RecordTick(stats_, NUMBER_SUPERVERSION_CLEANUPS);
   }
   RecordTick(stats_, NUMBER_SUPERVERSION_RELEASES);
@@ -3266,27 +3334,67 @@ Status DBImpl::CheckConsistency() {
   TEST_SYNC_POINT("DBImpl::CheckConsistency:AfterGetLiveFilesMetaData");
 
   std::string corruption_messages;
-  for (const auto& md : metadata) {
-    // md.name has a leading "/".
-    std::string file_path = md.db_path + md.name;
 
-    uint64_t fsize = 0;
-    TEST_SYNC_POINT("DBImpl::CheckConsistency:BeforeGetFileSize");
-    Status s = env_->GetFileSize(file_path, &fsize);
-    if (!s.ok() &&
-        env_->GetFileSize(Rocks2LevelTableFileName(file_path), &fsize).ok()) {
-      s = Status::OK();
+  if (immutable_db_options_.skip_checking_sst_file_sizes_on_db_open) {
+    // Instead of calling GetFileSize() for each expected file, call
+    // GetChildren() for the DB directory and check that all expected files
+    // are listed, without checking their sizes.
+    // Since sst files might be in different directories, do it for each
+    // directory separately.
+    std::map<std::string, std::vector<std::string>> files_by_directory;
+    for (const auto& md : metadata) {
+      // md.name has a leading "/". Remove it.
+      std::string fname = md.name;
+      if (!fname.empty() && fname[0] == '/') {
+        fname = fname.substr(1);
+      }
+      files_by_directory[md.db_path].push_back(fname);
     }
-    if (!s.ok()) {
-      corruption_messages +=
-          "Can't access " + md.name + ": " + s.ToString() + "\n";
-    } else if (fsize != md.size) {
-      corruption_messages += "Sst file size mismatch: " + file_path +
-                             ". Size recorded in manifest " +
-                             ToString(md.size) + ", actual size " +
-                             ToString(fsize) + "\n";
+    for (const auto& dir_files : files_by_directory) {
+      std::string directory = dir_files.first;
+      std::vector<std::string> existing_files;
+      Status s = env_->GetChildren(directory, &existing_files);
+      if (!s.ok()) {
+        corruption_messages +=
+            "Can't list files in " + directory + ": " + s.ToString() + "\n";
+        continue;
+      }
+      std::sort(existing_files.begin(), existing_files.end());
+
+      for (const std::string& fname : dir_files.second) {
+        if (!std::binary_search(existing_files.begin(), existing_files.end(),
+                                fname) &&
+            !std::binary_search(existing_files.begin(), existing_files.end(),
+                                Rocks2LevelTableFileName(fname))) {
+          corruption_messages +=
+              "Missing sst file " + fname + " in " + directory + "\n";
+        }
+      }
+    }
+  } else {
+    for (const auto& md : metadata) {
+      // md.name has a leading "/".
+      std::string file_path = md.db_path + md.name;
+
+      uint64_t fsize = 0;
+      TEST_SYNC_POINT("DBImpl::CheckConsistency:BeforeGetFileSize");
+      Status s = env_->GetFileSize(file_path, &fsize);
+      if (!s.ok() &&
+          env_->GetFileSize(Rocks2LevelTableFileName(file_path), &fsize).ok()) {
+        s = Status::OK();
+      }
+      if (!s.ok()) {
+        corruption_messages +=
+            "Can't access " + md.name + ": " + s.ToString() + "\n";
+      } else if (fsize != md.size) {
+        corruption_messages += "Sst file size mismatch: " + file_path +
+                               ". Size recorded in manifest " +
+                               ToString(md.size) + ", actual size " +
+                               ToString(fsize) + "\n";
+      }
     }
   }
+
   if (corruption_messages.size() == 0) {
     return Status::OK();
   } else {
@@ -3301,32 +3409,13 @@ Status DBImpl::GetDbIdentity(std::string& identity) const {
 
 Status DBImpl::GetDbIdentityFromIdentityFile(std::string* identity) const {
   std::string idfilename = IdentityFileName(dbname_);
-  const EnvOptions soptions;
-  std::unique_ptr<SequentialFileReader> id_file_reader;
-  Status s;
-  {
-    std::unique_ptr<SequentialFile> idfile;
-    s = env_->NewSequentialFile(idfilename, &idfile, soptions);
-    if (!s.ok()) {
-      return s;
-    }
-    id_file_reader.reset(
-        new SequentialFileReader(std::move(idfile), idfilename));
+  const FileOptions soptions;
+
+  Status s = ReadFileToString(fs_.get(), idfilename, identity);
+  if (!s.ok()) {
+    return s;
   }
 
-  uint64_t file_size;
-  s = env_->GetFileSize(idfilename, &file_size);
-  if (!s.ok()) {
-    return s;
-  }
-  char* buffer =
-      reinterpret_cast<char*>(alloca(static_cast<size_t>(file_size)));
-  Slice id;
-  s = id_file_reader->Read(static_cast<size_t>(file_size), &id, buffer);
-  if (!s.ok()) {
-    return s;
-  }
-  identity->assign(id.ToString());
   // If last character is '\n' remove it from identity
   if (identity->size() > 0 && identity->back() == '\n') {
     identity->pop_back();
@@ -3389,7 +3478,12 @@ Status DBImpl::Close() {
 Status DB::ListColumnFamilies(const DBOptions& db_options,
                               const std::string& name,
                               std::vector<std::string>* column_families) {
-  return VersionSet::ListColumnFamilies(column_families, name, db_options.env);
+  FileSystem* fs = db_options.file_system.get();
+  LegacyFileSystemWrapper legacy_fs(db_options.env);
+  if (!fs) {
+    fs = &legacy_fs;
+  }
+  return VersionSet::ListColumnFamilies(column_families, name, fs);
 }
 
 Snapshot::~Snapshot() {}
@@ -3433,24 +3527,15 @@ Status DestroyDB(const std::string& dbname, const Options& options,
       }
     }
 
-    std::vector<std::string> paths;
-
-    for (const auto& path : options.db_paths) {
-      paths.emplace_back(path.path);
+    std::set<std::string> paths;
+    for (const DbPath& db_path : options.db_paths) {
+      paths.insert(db_path.path);
     }
-    for (const auto& cf : column_families) {
-      for (const auto& path : cf.options.cf_paths) {
-        paths.emplace_back(path.path);
+    for (const ColumnFamilyDescriptor& cf : column_families) {
+      for (const DbPath& cf_path : cf.options.cf_paths) {
+        paths.insert(cf_path.path);
       }
     }
-
-    // Remove duplicate paths.
-    // Note that we compare only the actual paths but not path ids.
-    // This reason is that same path can appear at different path_ids
-    // for different column families.
-    std::sort(paths.begin(), paths.end());
-    paths.erase(std::unique(paths.begin(), paths.end()), paths.end());
-
     for (const auto& path : paths) {
       if (env->GetChildren(path, &filenames).ok()) {
         for (const auto& fname : filenames) {
@@ -3513,6 +3598,11 @@ Status DestroyDB(const std::string& dbname, const Options& options,
 
     env->UnlockFile(lock);  // Ignore error since state is already gone
     env->DeleteFile(lockname);
+
+    // sst_file_manager holds a ref to the logger. Make sure the logger is
+    // gone before trying to remove the directory.
+    soptions.sst_file_manager.reset();
+
     env->DeleteDir(dbname);  // Ignore error in case dir contains other files
   }
   return result;
@@ -3554,8 +3644,8 @@ Status DBImpl::WriteOptionsFile(bool need_mutex_lock,
 
   std::string file_name =
       TempOptionsFileName(GetName(), versions_->NewFileNumber());
-  Status s =
-      PersistRocksDBOptions(db_options, cf_names, cf_opts, file_name, GetEnv());
+  Status s = PersistRocksDBOptions(db_options, cf_names, cf_opts, file_name,
+                                   GetFileSystem());
 
   if (s.ok()) {
     s = RenameTempFileToOptionsFile(file_name);
@@ -3736,8 +3826,9 @@ Status DBImpl::GetLatestSequenceForKey(SuperVersion* sv, const Slice& key,
   *found_record_for_key = false;
 
   // Check if there is a record for this key in the latest memtable
-  sv->mem->Get(lkey, nullptr, &s, &merge_context, &max_covering_tombstone_seq,
-               seq, read_options, nullptr /*read_callback*/, is_blob_index);
+  sv->mem->Get(lkey, nullptr, nullptr, &s, &merge_context,
+               &max_covering_tombstone_seq, seq, read_options,
+               nullptr /*read_callback*/, is_blob_index);
 
   if (!(s.ok() || s.IsNotFound() || s.IsMergeInProgress())) {
     // unexpected error reading memtable.
@@ -3762,8 +3853,9 @@ Status DBImpl::GetLatestSequenceForKey(SuperVersion* sv, const Slice& key,
   }
 
   // Check if there is a record for this key in the immutable memtables
-  sv->imm->Get(lkey, nullptr, &s, &merge_context, &max_covering_tombstone_seq,
-               seq, read_options, nullptr /*read_callback*/, is_blob_index);
+  sv->imm->Get(lkey, nullptr, nullptr, &s, &merge_context,
+               &max_covering_tombstone_seq, seq, read_options,
+               nullptr /*read_callback*/, is_blob_index);
 
   if (!(s.ok() || s.IsNotFound() || s.IsMergeInProgress())) {
     // unexpected error reading memtable.
@@ -3788,7 +3880,7 @@ Status DBImpl::GetLatestSequenceForKey(SuperVersion* sv, const Slice& key,
   }
 
   // Check if there is a record for this key in the immutable memtables
-  sv->imm->GetFromHistory(lkey, nullptr, &s, &merge_context,
+  sv->imm->GetFromHistory(lkey, nullptr, nullptr, &s, &merge_context,
                           &max_covering_tombstone_seq, seq, read_options,
                           is_blob_index);
 
@@ -3816,7 +3908,7 @@ Status DBImpl::GetLatestSequenceForKey(SuperVersion* sv, const Slice& key,
   // SST files if cache_only=true?
   if (!cache_only) {
     // Check tables
-    sv->current->Get(read_options, lkey, nullptr, &s, &merge_context,
+    sv->current->Get(read_options, lkey, nullptr, nullptr, &s, &merge_context,
                      &max_covering_tombstone_seq, nullptr /* value_found */,
                      found_record_for_key, seq, nullptr /*read_callback*/,
                      is_blob_index);
@@ -3899,7 +3991,7 @@ Status DBImpl::IngestExternalFiles(
   for (const auto& arg : args) {
     auto* cfd = static_cast<ColumnFamilyHandleImpl*>(arg.column_family)->cfd();
     ingestion_jobs.emplace_back(
-        env_, versions_.get(), cfd, immutable_db_options_, env_options_,
+        env_, versions_.get(), cfd, immutable_db_options_, file_options_,
         &snapshots_, arg.options, &directories_, &event_logger_);
   }
   std::vector<std::pair<bool, Status>> exec_results;
@@ -3912,7 +4004,7 @@ Status DBImpl::IngestExternalFiles(
     start_file_number += args[i - 1].external_files.size();
     auto* cfd =
         static_cast<ColumnFamilyHandleImpl*>(args[i].column_family)->cfd();
-    SuperVersion* super_version = cfd->GetReferencedSuperVersion(&mutex_);
+    SuperVersion* super_version = cfd->GetReferencedSuperVersion(this);
     exec_results[i].second = ingestion_jobs[i].Prepare(
         args[i].external_files, start_file_number, super_version);
     exec_results[i].first = true;
@@ -3923,7 +4015,7 @@ Status DBImpl::IngestExternalFiles(
   {
     auto* cfd =
         static_cast<ColumnFamilyHandleImpl*>(args[0].column_family)->cfd();
-    SuperVersion* super_version = cfd->GetReferencedSuperVersion(&mutex_);
+    SuperVersion* super_version = cfd->GetReferencedSuperVersion(this);
     exec_results[0].second = ingestion_jobs[0].Prepare(
         args[0].external_files, next_file_number, super_version);
     exec_results[0].first = true;
@@ -3965,6 +4057,13 @@ Status DBImpl::IngestExternalFiles(
       nonmem_write_thread_.EnterUnbatched(&nonmem_w, &mutex_);
     }
 
+    // When unordered_write is enabled, the keys are writing to memtable in an
+    // unordered way. If the ingestion job checks memtable key range before the
+    // key landing in memtable, the ingestion job may skip the necessary
+    // memtable flush.
+    // So wait here to ensure there is no pending write to memtable.
+    WaitForPendingWrites();
+
     num_running_ingest_file_ += static_cast<int>(num_cfs);
     TEST_SYNC_POINT("DBImpl::IngestExternalFile:AfterIncIngestFileCounter");
 
@@ -4156,7 +4255,7 @@ Status DBImpl::CreateColumnFamilyWithImport(
   auto cfh = reinterpret_cast<ColumnFamilyHandleImpl*>(*handle);
   auto cfd = cfh->cfd();
   ImportColumnFamilyJob import_job(env_, versions_.get(), cfd,
-                                   immutable_db_options_, env_options_,
+                                   immutable_db_options_, file_options_,
                                    import_options, metadata.files);
 
   SuperVersionContext dummy_sv_ctx(/* create_superversion */ true);
@@ -4192,7 +4291,7 @@ Status DBImpl::CreateColumnFamilyWithImport(
   dummy_sv_ctx.Clean();
 
   if (status.ok()) {
-    SuperVersion* sv = cfd->GetReferencedSuperVersion(&mutex_);
+    SuperVersion* sv = cfd->GetReferencedSuperVersion(this);
     status = import_job.Prepare(next_file_number, sv);
     CleanupSuperVersion(sv);
   }
@@ -4269,7 +4368,7 @@ Status DBImpl::VerifyChecksum(const ReadOptions& read_options) {
   }
   std::vector<SuperVersion*> sv_list;
   for (auto cfd : cfd_list) {
-    sv_list.push_back(cfd->GetReferencedSuperVersion(&mutex_));
+    sv_list.push_back(cfd->GetReferencedSuperVersion(this));
   }
   for (auto& sv : sv_list) {
     VersionStorageInfo* vstorage = sv->current->storage_info();
@@ -4286,24 +4385,33 @@ Status DBImpl::VerifyChecksum(const ReadOptions& read_options) {
         const auto& fd = vstorage->LevelFilesBrief(i).files[j].fd;
         std::string fname = TableFileName(cfd->ioptions()->cf_paths,
                                           fd.GetNumber(), fd.GetPathId());
-        s = rocksdb::VerifySstFileChecksum(opts, env_options_, read_options,
-                                           fname);
+        s = ROCKSDB_NAMESPACE::VerifySstFileChecksum(opts, file_options_,
+                                                     read_options, fname);
       }
     }
     if (!s.ok()) {
       break;
     }
   }
+  bool defer_purge =
+          immutable_db_options().avoid_unnecessary_blocking_io;
   {
     InstrumentedMutexLock l(&mutex_);
     for (auto sv : sv_list) {
       if (sv && sv->Unref()) {
         sv->Cleanup();
-        delete sv;
+        if (defer_purge) {
+          AddSuperVersionsToFreeQueue(sv);
+        } else {
+          delete sv;
+        }
       }
     }
+    if (defer_purge) {
+      SchedulePurge();
+    }
     for (auto cfd : cfd_list) {
-      cfd->Unref();
+      cfd->UnrefAndTryDelete();
     }
   }
   return s;
@@ -4423,13 +4531,21 @@ Status DBImpl::GetCreationTimeOfOldestFile(uint64_t* creation_time) {
   if (mutable_db_options_.max_open_files == -1) {
     uint64_t oldest_time = port::kMaxUint64;
     for (auto cfd : *versions_->GetColumnFamilySet()) {
-      uint64_t ctime;
-      cfd->current()->GetCreationTimeOfOldestFile(&ctime);
-      if (ctime < oldest_time) {
-        oldest_time = ctime;
-      }
-      if (oldest_time == 0) {
-        break;
+      if (!cfd->IsDropped()) {
+        uint64_t ctime;
+        {
+          SuperVersion* sv = GetAndRefSuperVersion(cfd);
+          Version* version = sv->current;
+          version->GetCreationTimeOfOldestFile(&ctime);
+          ReturnAndCleanupSuperVersion(cfd, sv);
+        }
+
+        if (ctime < oldest_time) {
+          oldest_time = ctime;
+        }
+        if (oldest_time == 0) {
+          break;
+        }
       }
     }
     *creation_time = oldest_time;
@@ -4440,4 +4556,4 @@ Status DBImpl::GetCreationTimeOfOldestFile(uint64_t* creation_time) {
 }
 #endif  // ROCKSDB_LITE
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/db_impl/db_impl.h

@@ -62,7 +62,7 @@
 #include "util/stop_watch.h"
 #include "util/thread_local.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class Arena;
 class ArenaWrappedDBIter;
@@ -82,13 +82,13 @@ struct MemTableInfo;
 // Class to maintain directories for all database paths other than main one.
 class Directories {
  public:
-  Status SetDirectories(Env* env, const std::string& dbname,
-                        const std::string& wal_dir,
-                        const std::vector<DbPath>& data_paths);
+  IOStatus SetDirectories(FileSystem* fs, const std::string& dbname,
+                          const std::string& wal_dir,
+                          const std::vector<DbPath>& data_paths);
 
-  Directory* GetDataDir(size_t path_id) const {
+  FSDirectory* GetDataDir(size_t path_id) const {
     assert(path_id < data_dirs_.size());
-    Directory* ret_dir = data_dirs_[path_id].get();
+    FSDirectory* ret_dir = data_dirs_[path_id].get();
     if (ret_dir == nullptr) {
       // Should use db_dir_
       return db_dir_.get();
@@ -96,19 +96,19 @@ class Directories {
     return ret_dir;
   }
 
-  Directory* GetWalDir() {
+  FSDirectory* GetWalDir() {
     if (wal_dir_) {
       return wal_dir_.get();
     }
     return db_dir_.get();
   }
 
-  Directory* GetDbDir() { return db_dir_.get(); }
+  FSDirectory* GetDbDir() { return db_dir_.get(); }
 
  private:
-  std::unique_ptr<Directory> db_dir_;
-  std::vector<std::unique_ptr<Directory>> data_dirs_;
-  std::unique_ptr<Directory> wal_dir_;
+  std::unique_ptr<FSDirectory> db_dir_;
+  std::vector<std::unique_ptr<FSDirectory>> data_dirs_;
+  std::unique_ptr<FSDirectory> wal_dir_;
 };
 
 // While DB is the public interface of RocksDB, and DBImpl is the actual
@@ -163,6 +163,9 @@ class DBImpl : public DB {
   virtual Status Get(const ReadOptions& options,
                      ColumnFamilyHandle* column_family, const Slice& key,
                      PinnableSlice* value) override;
+  virtual Status Get(const ReadOptions& options,
+                     ColumnFamilyHandle* column_family, const Slice& key,
+                     PinnableSlice* value, std::string* timestamp) override;
 
   using DB::GetMergeOperands;
   Status GetMergeOperands(const ReadOptions& options,
@@ -230,7 +233,7 @@ class DBImpl : public DB {
   using DB::KeyMayExist;
   virtual bool KeyMayExist(const ReadOptions& options,
                            ColumnFamilyHandle* column_family, const Slice& key,
-                           std::string* value,
+                           std::string* value, std::string* timestamp,
                            bool* value_found = nullptr) override;
 
   using DB::NewIterator;
@@ -306,6 +309,7 @@ class DBImpl : public DB {
       ColumnFamilyHandle* column_family) override;
   virtual const std::string& GetName() const override;
   virtual Env* GetEnv() const override;
+  virtual FileSystem* GetFileSystem() const override;
   using DB::GetOptions;
   virtual Options GetOptions(ColumnFamilyHandle* column_family) const override;
   using DB::GetDBOptions;
@@ -432,6 +436,7 @@ class DBImpl : public DB {
   struct GetImplOptions {
     ColumnFamilyHandle* column_family = nullptr;
     PinnableSlice* value = nullptr;
+    std::string* timestamp = nullptr;
     bool* value_found = nullptr;
     ReadCallback* callback = nullptr;
     bool* is_blob_index = nullptr;
@@ -454,7 +459,7 @@ class DBImpl : public DB {
   // If get_impl_options.get_value = false get merge operands associated with
   // get_impl_options.key via get_impl_options.merge_operands
   Status GetImpl(const ReadOptions& options, const Slice& key,
-                 GetImplOptions get_impl_options);
+                 GetImplOptions& get_impl_options);
 
   ArenaWrappedDBIter* NewIteratorImpl(const ReadOptions& options,
                                       ColumnFamilyData* cfd,
@@ -798,6 +803,10 @@ class DBImpl : public DB {
     logs_to_free_queue_.push_back(log_writer);
   }
 
+  void AddSuperVersionsToFreeQueue(SuperVersion* sv) {
+    superversions_to_free_queue_.push_back(sv);
+  }
+
   void SetSnapshotChecker(SnapshotChecker* snapshot_checker);
 
   // Fill JobContext with snapshot information needed by flush and compaction.
@@ -821,9 +830,9 @@ class DBImpl : public DB {
                      std::vector<ColumnFamilyHandle*>* handles, DB** dbptr,
                      const bool seq_per_batch, const bool batch_per_txn);
 
-
-  static Status CreateAndNewDirectory(Env* env, const std::string& dirname,
-                                      std::unique_ptr<Directory>* directory);
+  static IOStatus CreateAndNewDirectory(
+      FileSystem* fs, const std::string& dirname,
+      std::unique_ptr<FSDirectory>* directory);
 
   // find stats map from stats_history_ with smallest timestamp in
   // the range of [start_time, end_time)
@@ -941,13 +950,14 @@ class DBImpl : public DB {
 #endif  // NDEBUG
 
  protected:
-  Env* const env_;
   const std::string dbname_;
   std::string db_id_;
   std::unique_ptr<VersionSet> versions_;
   // Flag to check whether we allocated and own the info log file
   bool own_info_log_;
   const DBOptions initial_db_options_;
+  Env* const env_;
+  std::shared_ptr<FileSystem> fs_;
   const ImmutableDBOptions immutable_db_options_;
   MutableDBOptions mutable_db_options_;
   Statistics* stats_;
@@ -975,10 +985,10 @@ class DBImpl : public DB {
   bool single_column_family_mode_;
 
   // The options to access storage files
-  const EnvOptions env_options_;
+  const FileOptions file_options_;
 
   // Additonal options for compaction and flush
-  EnvOptions env_options_for_compaction_;
+  FileOptions file_options_for_compaction_;
 
   std::unique_ptr<ColumnFamilyMemTablesImpl> column_family_memtables_;
 
@@ -1104,10 +1114,15 @@ 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.
+  // recovered_seq is set to less than kMaxSequenceNumber if the log's tail is
+  // skipped.
   virtual Status Recover(
       const std::vector<ColumnFamilyDescriptor>& column_families,
       bool read_only = false, bool error_if_log_file_exist = false,
-      bool error_if_data_exists_in_logs = false);
+      bool error_if_data_exists_in_logs = false,
+      uint64_t* recovered_seq = nullptr);
+
+  virtual bool OwnTablesAndLogs() const { return true; }
 
  private:
   friend class DB;
@@ -1193,7 +1208,7 @@ class DBImpl : public DB {
   };
 
   // PurgeFileInfo is a structure to hold information of files to be deleted in
-  // purge_queue_
+  // purge_files_
   struct PurgeFileInfo {
     std::string fname;
     std::string dir_to_sync;
@@ -1347,8 +1362,10 @@ class DBImpl : public DB {
       JobContext* job_context, LogBuffer* log_buffer, Env::Priority thread_pri);
 
   // REQUIRES: log_numbers are sorted in ascending order
+  // corrupted_log_found is set to true if we recover from a corrupted log file.
   Status RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
-                         SequenceNumber* next_sequence, bool read_only);
+                         SequenceNumber* next_sequence, bool read_only,
+                         bool* corrupted_log_found);
 
   // The following two methods are used to flush a memtable to
   // storage. The first one is used at database RecoveryTime (when the
@@ -1371,6 +1388,7 @@ class DBImpl : public DB {
   Status ThrottleLowPriWritesIfNeeded(const WriteOptions& write_options,
                                       WriteBatch* my_batch);
 
+  // REQUIRES: mutex locked and in write thread.
   Status ScheduleFlushes(WriteContext* context);
 
   void MaybeFlushStatsCF(autovector<ColumnFamilyData*>* cfds);
@@ -1414,6 +1432,7 @@ class DBImpl : public DB {
 
   inline void WaitForPendingWrites() {
     mutex_.AssertHeld();
+    TEST_SYNC_POINT("DBImpl::WaitForPendingWrites:BeforeBlock");
     // In case of pipelined write is enabled, wait for all pending memtable
     // writers.
     if (immutable_db_options_.enable_pipelined_write) {
@@ -1441,10 +1460,10 @@ class DBImpl : public DB {
   // REQUIRES: mutex locked and in write thread.
   void AssignAtomicFlushSeq(const autovector<ColumnFamilyData*>& cfds);
 
-  // REQUIRES: mutex locked
+  // REQUIRES: mutex locked and in write thread.
   Status SwitchWAL(WriteContext* write_context);
 
-  // REQUIRES: mutex locked
+  // REQUIRES: mutex locked and in write thread.
   Status HandleWriteBufferFull(WriteContext* write_context);
 
   // REQUIRES: mutex locked
@@ -1584,7 +1603,7 @@ class DBImpl : public DB {
 
   uint64_t GetMaxTotalWalSize() const;
 
-  Directory* GetDataDir(ColumnFamilyData* cfd, size_t path_id) const;
+  FSDirectory* GetDataDir(ColumnFamilyData* cfd, size_t path_id) const;
 
   Status CloseHelper();
 
@@ -1829,8 +1848,6 @@ class DBImpl : public DB {
 
   WriteController write_controller_;
 
-  std::unique_ptr<RateLimiter> low_pri_write_rate_limiter_;
-
   // Size of the last batch group. In slowdown mode, next write needs to
   // sleep if it uses up the quota.
   // Note: This is to protect memtable and compaction. If the batch only writes
@@ -1888,6 +1905,7 @@ class DBImpl : public DB {
 
   // A queue to store log writers to close
   std::deque<log::Writer*> logs_to_free_queue_;
+  std::deque<SuperVersion*> superversions_to_free_queue_;
   int unscheduled_flushes_;
   int unscheduled_compactions_;
 
@@ -2002,13 +2020,13 @@ class DBImpl : public DB {
 
   // handle for scheduling stats dumping at fixed intervals
   // REQUIRES: mutex locked
-  std::unique_ptr<rocksdb::RepeatableThread> thread_dump_stats_;
+  std::unique_ptr<ROCKSDB_NAMESPACE::RepeatableThread> thread_dump_stats_;
 
   // handle for scheduling stats snapshoting at fixed intervals
   // REQUIRES: mutex locked
-  std::unique_ptr<rocksdb::RepeatableThread> thread_persist_stats_;
+  std::unique_ptr<ROCKSDB_NAMESPACE::RepeatableThread> thread_persist_stats_;
 
-  // When set, we use a separate queue for writes that dont write to memtable.
+  // When set, we use a separate queue for writes that don't write to memtable.
   // In 2PC these are the writes at Prepare phase.
   const bool two_write_queues_;
   const bool manual_wal_flush_;
@@ -2091,4 +2109,4 @@ static void ClipToRange(T* ptr, V minvalue, V maxvalue) {
   if (static_cast<V>(*ptr) < minvalue) *ptr = minvalue;
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/db_impl/db_impl_compaction_flush.cc

@@ -22,7 +22,7 @@
 #include "util/cast_util.h"
 #include "util/concurrent_task_limiter_impl.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 bool DBImpl::EnoughRoomForCompaction(
     ColumnFamilyData* cfd, const std::vector<CompactionInputFiles>& inputs,
@@ -117,7 +117,7 @@ Status DBImpl::SyncClosedLogs(JobContext* job_context) {
       }
     }
     if (s.ok()) {
-      s = directories_.GetWalDir()->Fsync();
+      s = directories_.GetWalDir()->Fsync(IOOptions(), nullptr);
     }
 
     mutex_.Lock();
@@ -148,7 +148,7 @@ Status DBImpl::FlushMemTableToOutputFile(
 
   FlushJob flush_job(
       dbname_, cfd, immutable_db_options_, mutable_cf_options,
-      nullptr /* memtable_id */, env_options_for_compaction_, versions_.get(),
+      nullptr /* memtable_id */, file_options_for_compaction_, versions_.get(),
       &mutex_, &shutting_down_, snapshot_seqs, earliest_write_conflict_snapshot,
       snapshot_checker, job_context, log_buffer, directories_.GetDbDir(),
       GetDataDir(cfd, 0U),
@@ -301,7 +301,7 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
   GetSnapshotContext(job_context, &snapshot_seqs,
                      &earliest_write_conflict_snapshot, &snapshot_checker);
 
-  autovector<Directory*> distinct_output_dirs;
+  autovector<FSDirectory*> distinct_output_dirs;
   autovector<std::string> distinct_output_dir_paths;
   std::vector<std::unique_ptr<FlushJob>> jobs;
   std::vector<MutableCFOptions> all_mutable_cf_options;
@@ -309,7 +309,7 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
   all_mutable_cf_options.reserve(num_cfs);
   for (int i = 0; i < num_cfs; ++i) {
     auto cfd = cfds[i];
-    Directory* data_dir = GetDataDir(cfd, 0U);
+    FSDirectory* data_dir = GetDataDir(cfd, 0U);
     const std::string& curr_path = cfd->ioptions()->cf_paths[0].path;
 
     // Add to distinct output directories if eligible. Use linear search. Since
@@ -332,7 +332,7 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
     const uint64_t* max_memtable_id = &(bg_flush_args[i].max_memtable_id_);
     jobs.emplace_back(new FlushJob(
         dbname_, cfd, immutable_db_options_, mutable_cf_options,
-        max_memtable_id, env_options_for_compaction_, versions_.get(), &mutex_,
+        max_memtable_id, file_options_for_compaction_, versions_.get(), &mutex_,
         &shutting_down_, snapshot_seqs, earliest_write_conflict_snapshot,
         snapshot_checker, job_context, log_buffer, directories_.GetDbDir(),
         data_dir, GetCompressionFlush(*cfd->ioptions(), mutable_cf_options),
@@ -409,17 +409,34 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
     s = Status::OK();
   }
 
-  if (s.ok() || s.IsShutdownInProgress() || s.IsColumnFamilyDropped()) {
+  if (s.ok() || s.IsShutdownInProgress()) {
     // Sync on all distinct output directories.
     for (auto dir : distinct_output_dirs) {
       if (dir != nullptr) {
-        Status error_status = dir->Fsync();
+        Status error_status = dir->Fsync(IOOptions(), nullptr);
         if (!error_status.ok()) {
           s = error_status;
           break;
         }
       }
     }
+  } else {
+    // Need to undo atomic flush if something went wrong, i.e. s is not OK and
+    // it is not because of CF drop.
+    // Have to cancel the flush jobs that have NOT executed because we need to
+    // unref the versions.
+    for (int i = 0; i != num_cfs; ++i) {
+      if (!exec_status[i].first) {
+        jobs[i]->Cancel();
+      }
+    }
+    for (int i = 0; i != num_cfs; ++i) {
+      if (exec_status[i].first && exec_status[i].second.ok()) {
+        auto& mems = jobs[i]->GetMemTables();
+        cfds[i]->imm()->RollbackMemtableFlush(mems,
+                                              file_meta[i].fd.GetNumber());
+      }
+    }
   }
 
   if (s.ok()) {
@@ -526,23 +543,7 @@ Status DBImpl::AtomicFlushMemTablesToOutputFiles(
 #endif  // ROCKSDB_LITE
   }
 
-  // Need to undo atomic flush if something went wrong, i.e. s is not OK and
-  // it is not because of CF drop.
-  if (!s.ok() && !s.IsColumnFamilyDropped()) {
-    // Have to cancel the flush jobs that have NOT executed because we need to
-    // unref the versions.
-    for (int i = 0; i != num_cfs; ++i) {
-      if (!exec_status[i].first) {
-        jobs[i]->Cancel();
-      }
-    }
-    for (int i = 0; i != num_cfs; ++i) {
-      if (exec_status[i].first && exec_status[i].second.ok()) {
-        auto& mems = jobs[i]->GetMemTables();
-        cfds[i]->imm()->RollbackMemtableFlush(mems,
-                                              file_meta[i].fd.GetNumber());
-      }
-    }
+  if (!s.ok() && !s.IsShutdownInProgress()) {
     Status new_bg_error = s;
     error_handler_.SetBGError(new_bg_error, BackgroundErrorReason::kFlush);
   }
@@ -651,15 +652,13 @@ Status DBImpl::CompactRange(const CompactRangeOptions& options,
     return Status::InvalidArgument("Invalid target path ID");
   }
 
-  bool exclusive = options.exclusive_manual_compaction;
-
   bool flush_needed = true;
   if (begin != nullptr && end != nullptr) {
     // TODO(ajkr): We could also optimize away the flush in certain cases where
     // one/both sides of the interval are unbounded. But it requires more
     // changes to RangesOverlapWithMemtables.
     Range range(*begin, *end);
-    SuperVersion* super_version = cfd->GetReferencedSuperVersion(&mutex_);
+    SuperVersion* super_version = cfd->GetReferencedSuperVersion(this);
     cfd->RangesOverlapWithMemtables({range}, super_version, &flush_needed);
     CleanupSuperVersion(super_version);
   }
@@ -685,25 +684,9 @@ Status DBImpl::CompactRange(const CompactRangeOptions& options,
     }
   }
 
-  int max_level_with_files = 0;
-  // max_file_num_to_ignore can be used to filter out newly created SST files,
-  // useful for bottom level compaction in a manual compaction
-  uint64_t max_file_num_to_ignore = port::kMaxUint64;
-  uint64_t next_file_number = port::kMaxUint64;
-  {
-    InstrumentedMutexLock l(&mutex_);
-    Version* base = cfd->current();
-    for (int level = 1; level < base->storage_info()->num_non_empty_levels();
-         level++) {
-      if (base->storage_info()->OverlapInLevel(level, begin, end)) {
-        max_level_with_files = level;
-      }
-    }
-    next_file_number = versions_->current_next_file_number();
-  }
-
-  int final_output_level = 0;
-
+  constexpr int kInvalidLevel = -1;
+  int final_output_level = kInvalidLevel;
+  bool exclusive = options.exclusive_manual_compaction;
   if (cfd->ioptions()->compaction_style == kCompactionStyleUniversal &&
       cfd->NumberLevels() > 1) {
     // Always compact all files together.
@@ -714,58 +697,98 @@ Status DBImpl::CompactRange(const CompactRangeOptions& options,
     }
     s = RunManualCompaction(cfd, ColumnFamilyData::kCompactAllLevels,
                             final_output_level, options, begin, end, exclusive,
-                            false, max_file_num_to_ignore);
+                            false, port::kMaxUint64);
   } else {
-    for (int level = 0; level <= max_level_with_files; level++) {
+    int first_overlapped_level = kInvalidLevel;
+    int max_overlapped_level = kInvalidLevel;
+    {
+      SuperVersion* super_version = cfd->GetReferencedSuperVersion(this);
+      Version* current_version = super_version->current;
+      ReadOptions ro;
+      ro.total_order_seek = true;
+      bool overlap;
+      for (int level = 0;
+           level < current_version->storage_info()->num_non_empty_levels();
+           level++) {
+        overlap = true;
+        if (begin != nullptr && end != nullptr) {
+          Status status = current_version->OverlapWithLevelIterator(
+              ro, file_options_, *begin, *end, level, &overlap);
+          if (!status.ok()) {
+            overlap = current_version->storage_info()->OverlapInLevel(
+                level, begin, end);
+          }
+        } else {
+          overlap = current_version->storage_info()->OverlapInLevel(level,
+                                                                    begin, end);
+        }
+        if (overlap) {
+          if (first_overlapped_level == kInvalidLevel) {
+            first_overlapped_level = level;
+          }
+          max_overlapped_level = level;
+        }
+      }
+      CleanupSuperVersion(super_version);
+    }
+    if (s.ok() && first_overlapped_level != kInvalidLevel) {
+      // max_file_num_to_ignore can be used to filter out newly created SST
+      // files, useful for bottom level compaction in a manual compaction
+      uint64_t max_file_num_to_ignore = port::kMaxUint64;
+      uint64_t next_file_number = versions_->current_next_file_number();
+      final_output_level = max_overlapped_level;
       int output_level;
-      // in case the compaction is universal or if we're compacting the
-      // bottom-most level, the output level will be the same as input one.
-      // level 0 can never be the bottommost level (i.e. if all files are in
-      // level 0, we will compact to level 1)
-      if (cfd->ioptions()->compaction_style == kCompactionStyleUniversal ||
-          cfd->ioptions()->compaction_style == kCompactionStyleFIFO) {
-        output_level = level;
-      } else if (level == max_level_with_files && level > 0) {
-        if (options.bottommost_level_compaction ==
-            BottommostLevelCompaction::kSkip) {
-          // Skip bottommost level compaction
-          continue;
-        } else if (options.bottommost_level_compaction ==
-                       BottommostLevelCompaction::kIfHaveCompactionFilter &&
-                   cfd->ioptions()->compaction_filter == nullptr &&
-                   cfd->ioptions()->compaction_filter_factory == nullptr) {
-          // Skip bottommost level compaction since we don't have a compaction
-          // filter
-          continue;
+      for (int level = first_overlapped_level; level <= max_overlapped_level;
+           level++) {
+        // in case the compaction is universal or if we're compacting the
+        // bottom-most level, the output level will be the same as input one.
+        // level 0 can never be the bottommost level (i.e. if all files are in
+        // level 0, we will compact to level 1)
+        if (cfd->ioptions()->compaction_style == kCompactionStyleUniversal ||
+            cfd->ioptions()->compaction_style == kCompactionStyleFIFO) {
+          output_level = level;
+        } else if (level == max_overlapped_level && level > 0) {
+          if (options.bottommost_level_compaction ==
+              BottommostLevelCompaction::kSkip) {
+            // Skip bottommost level compaction
+            continue;
+          } else if (options.bottommost_level_compaction ==
+                         BottommostLevelCompaction::kIfHaveCompactionFilter &&
+                     cfd->ioptions()->compaction_filter == nullptr &&
+                     cfd->ioptions()->compaction_filter_factory == nullptr) {
+            // Skip bottommost level compaction since we don't have a compaction
+            // filter
+            continue;
+          }
+          output_level = level;
+          // update max_file_num_to_ignore only for bottom level compaction
+          // because data in newly compacted files in middle levels may still
+          // need to be pushed down
+          max_file_num_to_ignore = next_file_number;
+        } else {
+          output_level = level + 1;
+          if (cfd->ioptions()->compaction_style == kCompactionStyleLevel &&
+              cfd->ioptions()->level_compaction_dynamic_level_bytes &&
+              level == 0) {
+            output_level = ColumnFamilyData::kCompactToBaseLevel;
+          }
         }
-        output_level = level;
-        // update max_file_num_to_ignore only for bottom level compaction
-        // because data in newly compacted files in middle levels may still need
-        // to be pushed down
-        max_file_num_to_ignore = next_file_number;
-      } else {
-        output_level = level + 1;
-        if (cfd->ioptions()->compaction_style == kCompactionStyleLevel &&
-            cfd->ioptions()->level_compaction_dynamic_level_bytes &&
-            level == 0) {
-          output_level = ColumnFamilyData::kCompactToBaseLevel;
+        s = RunManualCompaction(cfd, level, output_level, options, begin, end,
+                                exclusive, false, max_file_num_to_ignore);
+        if (!s.ok()) {
+          break;
         }
+        if (output_level == ColumnFamilyData::kCompactToBaseLevel) {
+          final_output_level = cfd->NumberLevels() - 1;
+        } else if (output_level > final_output_level) {
+          final_output_level = output_level;
+        }
+        TEST_SYNC_POINT("DBImpl::RunManualCompaction()::1");
+        TEST_SYNC_POINT("DBImpl::RunManualCompaction()::2");
       }
-      s = RunManualCompaction(cfd, level, output_level, options, begin, end,
-                              exclusive, false, max_file_num_to_ignore);
-      if (!s.ok()) {
-        break;
-      }
-      if (output_level == ColumnFamilyData::kCompactToBaseLevel) {
-        final_output_level = cfd->NumberLevels() - 1;
-      } else if (output_level > final_output_level) {
-        final_output_level = output_level;
-      }
-      TEST_SYNC_POINT("DBImpl::RunManualCompaction()::1");
-      TEST_SYNC_POINT("DBImpl::RunManualCompaction()::2");
     }
   }
-  if (!s.ok()) {
+  if (!s.ok() || final_output_level == kInvalidLevel) {
     LogFlush(immutable_db_options_.info_log);
     return s;
   }
@@ -968,7 +991,7 @@ Status DBImpl::CompactFilesImpl(
   CompactionJobStats compaction_job_stats;
   CompactionJob compaction_job(
       job_context->job_id, c.get(), immutable_db_options_,
-      env_options_for_compaction_, versions_.get(), &shutting_down_,
+      file_options_for_compaction_, versions_.get(), &shutting_down_,
       preserve_deletes_seqnum_.load(), log_buffer, directories_.GetDbDir(),
       GetDataDir(c->column_family_data(), c->output_path_id()), stats_, &mutex_,
       &error_handler_, snapshot_seqs, earliest_write_conflict_snapshot,
@@ -1036,7 +1059,7 @@ Status DBImpl::CompactFilesImpl(
   }
 
   if (output_file_names != nullptr) {
-    for (const auto newf : c->edit()->GetNewFiles()) {
+    for (const auto& newf : c->edit()->GetNewFiles()) {
       (*output_file_names)
           .push_back(TableFileName(c->immutable_cf_options()->cf_paths,
                                    newf.second.fd.GetNumber(),
@@ -1136,7 +1159,7 @@ void DBImpl::NotifyOnCompactionBegin(ColumnFamilyData* cfd, Compaction* c,
         }
       }
     }
-    for (const auto newf : c->edit()->GetNewFiles()) {
+    for (const auto& newf : c->edit()->GetNewFiles()) {
       const FileMetaData& meta = newf.second;
       const FileDescriptor& desc = meta.fd;
       const uint64_t file_number = desc.GetNumber();
@@ -1257,7 +1280,8 @@ Status DBImpl::ReFitLevel(ColumnFamilyData* cfd, int level, int target_level) {
                    f->fd.GetFileSize(), f->smallest, f->largest,
                    f->fd.smallest_seqno, f->fd.largest_seqno,
                    f->marked_for_compaction, f->oldest_blob_file_number,
-                   f->oldest_ancester_time);
+                   f->oldest_ancester_time, f->file_creation_time,
+                   f->file_checksum, f->file_checksum_func_name);
     }
     ROCKS_LOG_DEBUG(immutable_db_options_.info_log,
                     "[%s] Apply version edit:\n%s", cfd->GetName().c_str(),
@@ -1540,6 +1564,7 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
         nonmem_write_thread_.EnterUnbatched(&nonmem_w, &mutex_);
       }
     }
+    WaitForPendingWrites();
 
     if (!cfd->mem()->IsEmpty() || !cached_recoverable_state_empty_.load()) {
       s = SwitchMemtable(cfd, &context);
@@ -1617,19 +1642,16 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
     }
     s = WaitForFlushMemTables(cfds, flush_memtable_ids,
                               (flush_reason == FlushReason::kErrorRecovery));
+    InstrumentedMutexLock lock_guard(&mutex_);
     for (auto* tmp_cfd : cfds) {
-      if (tmp_cfd->Unref()) {
-        // Only one thread can reach here.
-        InstrumentedMutexLock lock_guard(&mutex_);
-        delete tmp_cfd;
-      }
+      tmp_cfd->UnrefAndTryDelete();
     }
   }
-  TEST_SYNC_POINT("FlushMemTableFinished");
+  TEST_SYNC_POINT("DBImpl::FlushMemTable:FlushMemTableFinished");
   return s;
 }
 
-// Flush all elments in 'column_family_datas'
+// Flush all elements in 'column_family_datas'
 // and atomically record the result to the MANIFEST.
 Status DBImpl::AtomicFlushMemTables(
     const autovector<ColumnFamilyData*>& column_family_datas,
@@ -1665,6 +1687,7 @@ Status DBImpl::AtomicFlushMemTables(
         nonmem_write_thread_.EnterUnbatched(&nonmem_w, &mutex_);
       }
     }
+    WaitForPendingWrites();
 
     for (auto cfd : column_family_datas) {
       if (cfd->IsDropped()) {
@@ -1681,7 +1704,7 @@ Status DBImpl::AtomicFlushMemTables(
       }
       cfd->Ref();
       s = SwitchMemtable(cfd, &context);
-      cfd->Unref();
+      cfd->UnrefAndTryDelete();
       if (!s.ok()) {
         break;
       }
@@ -1721,12 +1744,9 @@ Status DBImpl::AtomicFlushMemTables(
     }
     s = WaitForFlushMemTables(cfds, flush_memtable_ids,
                               (flush_reason == FlushReason::kErrorRecovery));
+    InstrumentedMutexLock lock_guard(&mutex_);
     for (auto* cfd : cfds) {
-      if (cfd->Unref()) {
-        // Only one thread can reach here.
-        InstrumentedMutexLock lock_guard(&mutex_);
-        delete cfd;
-      }
+      cfd->UnrefAndTryDelete();
     }
   }
   return s;
@@ -1918,6 +1938,10 @@ void DBImpl::MaybeScheduleFlushOrCompaction() {
     fta->thread_pri_ = Env::Priority::HIGH;
     env_->Schedule(&DBImpl::BGWorkFlush, fta, Env::Priority::HIGH, this,
                    &DBImpl::UnscheduleFlushCallback);
+    --unscheduled_flushes_;
+    TEST_SYNC_POINT_CALLBACK(
+        "DBImpl::MaybeScheduleFlushOrCompaction:AfterSchedule:0",
+        &unscheduled_flushes_);
   }
 
   // special case -- if high-pri (flush) thread pool is empty, then schedule
@@ -1932,6 +1956,7 @@ void DBImpl::MaybeScheduleFlushOrCompaction() {
       fta->thread_pri_ = Env::Priority::LOW;
       env_->Schedule(&DBImpl::BGWorkFlush, fta, Env::Priority::LOW, this,
                      &DBImpl::UnscheduleFlushCallback);
+      --unscheduled_flushes_;
     }
   }
 
@@ -2015,8 +2040,6 @@ ColumnFamilyData* DBImpl::PopFirstFromCompactionQueue() {
 DBImpl::FlushRequest DBImpl::PopFirstFromFlushQueue() {
   assert(!flush_queue_.empty());
   FlushRequest flush_req = flush_queue_.front();
-  assert(unscheduled_flushes_ >= static_cast<int>(flush_req.size()));
-  unscheduled_flushes_ -= static_cast<int>(flush_req.size());
   flush_queue_.pop_front();
   // TODO: need to unset flush reason?
   return flush_req;
@@ -2058,7 +2081,7 @@ void DBImpl::SchedulePendingFlush(const FlushRequest& flush_req,
     cfd->Ref();
     cfd->SetFlushReason(flush_reason);
   }
-  unscheduled_flushes_ += static_cast<int>(flush_req.size());
+  ++unscheduled_flushes_;
   flush_queue_.push_back(flush_req);
 }
 
@@ -2204,16 +2227,13 @@ Status DBImpl::BackgroundFlush(bool* made_progress, JobContext* job_context,
     *reason = bg_flush_args[0].cfd_->GetFlushReason();
     for (auto& arg : bg_flush_args) {
       ColumnFamilyData* cfd = arg.cfd_;
-      if (cfd->Unref()) {
-        delete cfd;
+      if (cfd->UnrefAndTryDelete()) {
         arg.cfd_ = nullptr;
       }
     }
   }
   for (auto cfd : column_families_not_to_flush) {
-    if (cfd->Unref()) {
-      delete cfd;
-    }
+    cfd->UnrefAndTryDelete();
   }
   return status;
 }
@@ -2542,10 +2562,9 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
     // reference).
     // This will all happen under a mutex so we don't have to be afraid of
     // somebody else deleting it.
-    if (cfd->Unref()) {
+    if (cfd->UnrefAndTryDelete()) {
       // This was the last reference of the column family, so no need to
       // compact.
-      delete cfd;
       return Status::OK();
     }
 
@@ -2672,7 +2691,9 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
                            f->fd.GetPathId(), f->fd.GetFileSize(), f->smallest,
                            f->largest, f->fd.smallest_seqno,
                            f->fd.largest_seqno, f->marked_for_compaction,
-                           f->oldest_blob_file_number, f->oldest_ancester_time);
+                           f->oldest_blob_file_number, f->oldest_ancester_time,
+                           f->file_creation_time, f->file_checksum,
+                           f->file_checksum_func_name);
 
         ROCKS_LOG_BUFFER(
             log_buffer,
@@ -2751,7 +2772,7 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
     assert(is_snapshot_supported_ || snapshots_.empty());
     CompactionJob compaction_job(
         job_context->job_id, c.get(), immutable_db_options_,
-        env_options_for_compaction_, versions_.get(), &shutting_down_,
+        file_options_for_compaction_, versions_.get(), &shutting_down_,
         preserve_deletes_seqnum_.load(), log_buffer, directories_.GetDbDir(),
         GetDataDir(c->column_family_data(), c->output_path_id()), stats_,
         &mutex_, &error_handler_, snapshot_seqs,
@@ -3114,4 +3135,4 @@ void DBImpl::GetSnapshotContext(
   }
   *snapshot_seqs = snapshots_.GetAll(earliest_write_conflict_snapshot);
 }
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/db_impl/db_impl_debug.cc

@@ -15,7 +15,7 @@
 #include "monitoring/thread_status_updater.h"
 #include "util/cast_util.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 uint64_t DBImpl::TEST_GetLevel0TotalSize() {
   InstrumentedMutexLock l(&mutex_);
   return default_cf_handle_->cfd()->current()->storage_info()->NumLevelBytes(0);
@@ -24,7 +24,9 @@ uint64_t DBImpl::TEST_GetLevel0TotalSize() {
 void DBImpl::TEST_SwitchWAL() {
   WriteContext write_context;
   InstrumentedMutexLock l(&mutex_);
+  void* writer = TEST_BeginWrite();
   SwitchWAL(&write_context);
+  TEST_EndWrite(writer);
 }
 
 bool DBImpl::TEST_WALBufferIsEmpty(bool lock) {
@@ -106,15 +108,18 @@ Status DBImpl::TEST_SwitchMemtable(ColumnFamilyData* cfd) {
     cfd = default_cf_handle_->cfd();
   }
 
+  Status s;
+  void* writer = TEST_BeginWrite();
   if (two_write_queues_) {
     WriteThread::Writer nonmem_w;
     nonmem_write_thread_.EnterUnbatched(&nonmem_w, &mutex_);
-    Status s = SwitchMemtable(cfd, &write_context);
+    s = SwitchMemtable(cfd, &write_context);
     nonmem_write_thread_.ExitUnbatched(&nonmem_w);
-    return s;
   } else {
-    return SwitchMemtable(cfd, &write_context);
+    s = SwitchMemtable(cfd, &write_context);
   }
+  TEST_EndWrite(writer);
+  return s;
 }
 
 Status DBImpl::TEST_FlushMemTable(bool wait, bool allow_write_stall,
@@ -285,5 +290,5 @@ bool DBImpl::TEST_IsPersistentStatsEnabled() const {
 size_t DBImpl::TEST_EstimateInMemoryStatsHistorySize() const {
   return EstimateInMemoryStatsHistorySize();
 }
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 #endif  // NDEBUG

db/db_impl/db_impl_experimental.cc

@@ -17,7 +17,7 @@
 #include "db/version_set.h"
 #include "rocksdb/status.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 #ifndef ROCKSDB_LITE
 Status DBImpl::SuggestCompactRange(ColumnFamilyHandle* column_family,
@@ -129,7 +129,8 @@ Status DBImpl::PromoteL0(ColumnFamilyHandle* column_family, int target_level) {
                    f->fd.GetFileSize(), f->smallest, f->largest,
                    f->fd.smallest_seqno, f->fd.largest_seqno,
                    f->marked_for_compaction, f->oldest_blob_file_number,
-                   f->oldest_ancester_time);
+                   f->oldest_ancester_time, f->file_creation_time,
+                   f->file_checksum, f->file_checksum_func_name);
     }
 
     status = versions_->LogAndApply(cfd, *cfd->GetLatestMutableCFOptions(),
@@ -147,4 +148,4 @@ Status DBImpl::PromoteL0(ColumnFamilyHandle* column_family, int target_level) {
 }
 #endif  // ROCKSDB_LITE
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/db_impl/db_impl_files.cc

@@ -17,7 +17,7 @@
 #include "file/sst_file_manager_impl.h"
 #include "util/autovector.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 uint64_t DBImpl::MinLogNumberToKeep() {
   if (allow_2pc()) {
@@ -385,6 +385,7 @@ void DBImpl::PurgeObsoleteFiles(JobContext& state, bool schedule_only) {
     w->Close();
   }
 
+  bool own_files = OwnTablesAndLogs();
   std::unordered_set<uint64_t> files_to_del;
   for (const auto& candidate_file : candidate_files) {
     const std::string& to_delete = candidate_file.file_name;
@@ -484,6 +485,12 @@ void DBImpl::PurgeObsoleteFiles(JobContext& state, bool schedule_only) {
     }
 #endif  // !ROCKSDB_LITE
 
+    // If I do not own these files, e.g. secondary instance with max_open_files
+    // = -1, then no need to delete or schedule delete these files since they
+    // will be removed by their owner, e.g. the primary instance.
+    if (!own_files) {
+      continue;
+    }
     Status file_deletion_status;
     if (schedule_only) {
       InstrumentedMutexLock guard_lock(&mutex_);
@@ -495,7 +502,6 @@ void DBImpl::PurgeObsoleteFiles(JobContext& state, bool schedule_only) {
 
   {
     // After purging obsolete files, remove them from files_grabbed_for_purge_.
-    // Use a temporary vector to perform bulk deletion via swap.
     InstrumentedMutexLock guard_lock(&mutex_);
     autovector<uint64_t> to_be_removed;
     for (auto fn : files_grabbed_for_purge_) {
@@ -612,9 +618,9 @@ uint64_t PrecomputeMinLogNumberToKeep(
   // family being flushed (`cfd_to_flush`).
   uint64_t cf_min_log_number_to_keep = 0;
   for (auto& e : edit_list) {
-    if (e->has_log_number()) {
+    if (e->HasLogNumber()) {
       cf_min_log_number_to_keep =
-          std::max(cf_min_log_number_to_keep, e->log_number());
+          std::max(cf_min_log_number_to_keep, e->GetLogNumber());
     }
   }
   if (cf_min_log_number_to_keep == 0) {
@@ -658,4 +664,4 @@ uint64_t PrecomputeMinLogNumberToKeep(
   return min_log_number_to_keep;
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/db_impl/db_impl_open.cc

@@ -12,6 +12,7 @@
 
 #include "db/builder.h"
 #include "db/error_handler.h"
+#include "env/composite_env_wrapper.h"
 #include "file/read_write_util.h"
 #include "file/sst_file_manager_impl.h"
 #include "file/writable_file_writer.h"
@@ -22,7 +23,7 @@
 #include "test_util/sync_point.h"
 #include "util/rate_limiter.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 Options SanitizeOptions(const std::string& dbname, const Options& src) {
   auto db_options = SanitizeOptions(dbname, DBOptions(src));
   ImmutableDBOptions immutable_db_options(db_options);
@@ -34,6 +35,18 @@ Options SanitizeOptions(const std::string& dbname, const Options& src) {
 DBOptions SanitizeOptions(const std::string& dbname, const DBOptions& src) {
   DBOptions result(src);
 
+  if (result.file_system == nullptr) {
+    if (result.env == Env::Default()) {
+      result.file_system = FileSystem::Default();
+    } else {
+      result.file_system.reset(new LegacyFileSystemWrapper(result.env));
+    }
+  } else {
+    if (result.env == nullptr) {
+      result.env = Env::Default();
+    }
+  }
+
   // result.max_open_files means an "infinite" open files.
   if (result.max_open_files != -1) {
     int max_max_open_files = port::GetMaxOpenFiles();
@@ -87,12 +100,11 @@ DBOptions SanitizeOptions(const std::string& dbname, const DBOptions& src) {
   if (result.recycle_log_file_num &&
       (result.wal_recovery_mode == WALRecoveryMode::kPointInTimeRecovery ||
        result.wal_recovery_mode == WALRecoveryMode::kAbsoluteConsistency)) {
-    // kPointInTimeRecovery is indistinguishable from
-    // kTolerateCorruptedTailRecords in recycle mode since we define
-    // the "end" of the log as the first corrupt record we encounter.
+    // kPointInTimeRecovery is inconsistent with recycle log file feature since
+    // we define the "end" of the log as the first corrupt record we encounter.
     // kAbsoluteConsistency doesn't make sense because even a clean
     // shutdown leaves old junk at the end of the log file.
-    result.wal_recovery_mode = WALRecoveryMode::kTolerateCorruptedTailRecords;
+    result.recycle_log_file_num = 0;
   }
 
   if (result.wal_dir.empty()) {
@@ -160,6 +172,13 @@ DBOptions SanitizeOptions(const std::string& dbname, const DBOptions& src) {
     result.sst_file_manager = sst_file_manager;
   }
 #endif
+
+  if (!result.paranoid_checks) {
+    result.skip_checking_sst_file_sizes_on_db_open = true;
+    ROCKS_LOG_INFO(result.info_log,
+                   "file size check will be skipped during open.");
+  }
+
   return result;
 }
 
@@ -254,16 +273,16 @@ Status DBImpl::NewDB() {
   ROCKS_LOG_INFO(immutable_db_options_.info_log, "Creating manifest 1 \n");
   const std::string manifest = DescriptorFileName(dbname_, 1);
   {
-    std::unique_ptr<WritableFile> file;
-    EnvOptions env_options = env_->OptimizeForManifestWrite(env_options_);
-    s = NewWritableFile(env_, manifest, &file, env_options);
+    std::unique_ptr<FSWritableFile> file;
+    FileOptions file_options = fs_->OptimizeForManifestWrite(file_options_);
+    s = NewWritableFile(fs_.get(), manifest, &file, file_options);
     if (!s.ok()) {
       return s;
     }
     file->SetPreallocationBlockSize(
         immutable_db_options_.manifest_preallocation_size);
     std::unique_ptr<WritableFileWriter> file_writer(new WritableFileWriter(
-        std::move(file), manifest, env_options, env_, nullptr /* stats */,
+        std::move(file), manifest, file_options, env_, nullptr /* stats */,
         immutable_db_options_.listeners));
     log::Writer log(std::move(file_writer), 0, false);
     std::string record;
@@ -277,13 +296,14 @@ Status DBImpl::NewDB() {
     // Make "CURRENT" file that points to the new manifest file.
     s = SetCurrentFile(env_, dbname_, 1, directories_.GetDbDir());
   } else {
-    env_->DeleteFile(manifest);
+    fs_->DeleteFile(manifest, IOOptions(), nullptr);
   }
   return s;
 }
 
-Status DBImpl::CreateAndNewDirectory(Env* env, const std::string& dirname,
-                                     std::unique_ptr<Directory>* directory) {
+IOStatus DBImpl::CreateAndNewDirectory(
+    FileSystem* fs, const std::string& dirname,
+    std::unique_ptr<FSDirectory>* directory) {
   // We call CreateDirIfMissing() as the directory may already exist (if we
   // are reopening a DB), when this happens we don't want creating the
   // directory to cause an error. However, we need to check if creating the
@@ -291,24 +311,24 @@ Status DBImpl::CreateAndNewDirectory(Env* env, const std::string& dirname,
   // file not existing. One real-world example of this occurring is if
   // env->CreateDirIfMissing() doesn't create intermediate directories, e.g.
   // when dbname_ is "dir/db" but when "dir" doesn't exist.
-  Status s = env->CreateDirIfMissing(dirname);
-  if (!s.ok()) {
-    return s;
+  IOStatus io_s = fs->CreateDirIfMissing(dirname, IOOptions(), nullptr);
+  if (!io_s.ok()) {
+    return io_s;
   }
-  return env->NewDirectory(dirname, directory);
+  return fs->NewDirectory(dirname, IOOptions(), directory, nullptr);
 }
 
-Status Directories::SetDirectories(Env* env, const std::string& dbname,
-                                   const std::string& wal_dir,
-                                   const std::vector<DbPath>& data_paths) {
-  Status s = DBImpl::CreateAndNewDirectory(env, dbname, &db_dir_);
-  if (!s.ok()) {
-    return s;
+IOStatus Directories::SetDirectories(FileSystem* fs, const std::string& dbname,
+                                     const std::string& wal_dir,
+                                     const std::vector<DbPath>& data_paths) {
+  IOStatus io_s = DBImpl::CreateAndNewDirectory(fs, dbname, &db_dir_);
+  if (!io_s.ok()) {
+    return io_s;
   }
   if (!wal_dir.empty() && dbname != wal_dir) {
-    s = DBImpl::CreateAndNewDirectory(env, wal_dir, &wal_dir_);
-    if (!s.ok()) {
-      return s;
+    io_s = DBImpl::CreateAndNewDirectory(fs, wal_dir, &wal_dir_);
+    if (!io_s.ok()) {
+      return io_s;
     }
   }
 
@@ -318,27 +338,28 @@ Status Directories::SetDirectories(Env* env, const std::string& dbname,
     if (db_path == dbname) {
       data_dirs_.emplace_back(nullptr);
     } else {
-      std::unique_ptr<Directory> path_directory;
-      s = DBImpl::CreateAndNewDirectory(env, db_path, &path_directory);
-      if (!s.ok()) {
-        return s;
+      std::unique_ptr<FSDirectory> path_directory;
+      io_s = DBImpl::CreateAndNewDirectory(fs, db_path, &path_directory);
+      if (!io_s.ok()) {
+        return io_s;
       }
       data_dirs_.emplace_back(path_directory.release());
     }
   }
   assert(data_dirs_.size() == data_paths.size());
-  return Status::OK();
+  return IOStatus::OK();
 }
 
 Status DBImpl::Recover(
     const std::vector<ColumnFamilyDescriptor>& column_families, bool read_only,
-    bool error_if_log_file_exist, bool error_if_data_exists_in_logs) {
+    bool error_if_log_file_exist, bool error_if_data_exists_in_logs,
+    uint64_t* recovered_seq) {
   mutex_.AssertHeld();
 
   bool is_new_db = false;
   assert(db_lock_ == nullptr);
   if (!read_only) {
-    Status s = directories_.SetDirectories(env_, dbname_,
+    Status s = directories_.SetDirectories(fs_.get(), dbname_,
                                            immutable_db_options_.wal_dir,
                                            immutable_db_options_.db_paths);
     if (!s.ok()) {
@@ -350,12 +371,10 @@ Status DBImpl::Recover(
       return s;
     }
 
-    s = env_->FileExists(CurrentFileName(dbname_));
+    std::string current_fname = CurrentFileName(dbname_);
+    s = env_->FileExists(current_fname);
     if (s.IsNotFound()) {
       if (immutable_db_options_.create_if_missing) {
-        // Has to be called only after Identity File creation is successful
-        // because DB ID is stored in Manifest if
-        // immutable_db_options_.write_dbid_to_manifest = true
         s = NewDB();
         is_new_db = true;
         if (!s.ok()) {
@@ -363,7 +382,7 @@ Status DBImpl::Recover(
         }
       } else {
         return Status::InvalidArgument(
-            dbname_, "does not exist (create_if_missing is false)");
+            current_fname, "does not exist (create_if_missing is false)");
       }
     } else if (s.ok()) {
       if (immutable_db_options_.error_if_exists) {
@@ -375,20 +394,20 @@ Status DBImpl::Recover(
       assert(s.IsIOError());
       return s;
     }
-    // Verify compatibility of env_options_ and filesystem
+    // Verify compatibility of file_options_ and filesystem
     {
-      std::unique_ptr<RandomAccessFile> idfile;
-      EnvOptions customized_env(env_options_);
-      customized_env.use_direct_reads |=
+      std::unique_ptr<FSRandomAccessFile> idfile;
+      FileOptions customized_fs(file_options_);
+      customized_fs.use_direct_reads |=
           immutable_db_options_.use_direct_io_for_flush_and_compaction;
-      s = env_->NewRandomAccessFile(CurrentFileName(dbname_), &idfile,
-                                    customized_env);
+      s = fs_->NewRandomAccessFile(current_fname, customized_fs, &idfile,
+                                   nullptr);
       if (!s.ok()) {
         std::string error_str = s.ToString();
         // Check if unsupported Direct I/O is the root cause
-        customized_env.use_direct_reads = false;
-        s = env_->NewRandomAccessFile(CurrentFileName(dbname_), &idfile,
-                                      customized_env);
+        customized_fs.use_direct_reads = false;
+        s = fs_->NewRandomAccessFile(current_fname, customized_fs, &idfile,
+                                     nullptr);
         if (s.ok()) {
           return Status::InvalidArgument(
               "Direct I/O is not supported by the specified DB.");
@@ -408,7 +427,7 @@ Status DBImpl::Recover(
   // the very first time.
   if (db_id_.empty()) {
     // Check for the IDENTITY file and create it if not there.
-    s = env_->FileExists(IdentityFileName(dbname_));
+    s = fs_->FileExists(IdentityFileName(dbname_), IOOptions(), nullptr);
     // Typically Identity file is created in NewDB() and for some reason if
     // it is no longer available then at this point DB ID is not in Identity
     // file or Manifest.
@@ -421,27 +440,28 @@ Status DBImpl::Recover(
       assert(s.IsIOError());
       return s;
     }
-    GetDbIdentityFromIdentityFile(&db_id_);
-    if (immutable_db_options_.write_dbid_to_manifest) {
+    s = GetDbIdentityFromIdentityFile(&db_id_);
+    if (immutable_db_options_.write_dbid_to_manifest && s.ok()) {
       VersionEdit edit;
       edit.SetDBId(db_id_);
       Options options;
       MutableCFOptions mutable_cf_options(options);
       versions_->db_id_ = db_id_;
-      versions_->LogAndApply(versions_->GetColumnFamilySet()->GetDefault(),
+      s = versions_->LogAndApply(versions_->GetColumnFamilySet()->GetDefault(),
                              mutable_cf_options, &edit, &mutex_, nullptr,
                              false);
     }
   } else {
-    SetIdentityFile(env_, dbname_, db_id_);
+    s = SetIdentityFile(env_, dbname_, db_id_);
   }
 
   if (immutable_db_options_.paranoid_checks && s.ok()) {
     s = CheckConsistency();
   }
   if (s.ok() && !read_only) {
+    std::map<std::string, std::shared_ptr<FSDirectory>> created_dirs;
     for (auto cfd : *versions_->GetColumnFamilySet()) {
-      s = cfd->AddDirectories();
+      s = cfd->AddDirectories(&created_dirs);
       if (!s.ok()) {
         return s;
       }
@@ -452,16 +472,16 @@ Status DBImpl::Recover(
     s = InitPersistStatsColumnFamily();
   }
 
-  // Initial max_total_in_memory_state_ before recovery logs. Log recovery
-  // may check this value to decide whether to flush.
-  max_total_in_memory_state_ = 0;
-  for (auto cfd : *versions_->GetColumnFamilySet()) {
-    auto* mutable_cf_options = cfd->GetLatestMutableCFOptions();
-    max_total_in_memory_state_ += mutable_cf_options->write_buffer_size *
-                                  mutable_cf_options->max_write_buffer_number;
-  }
-
   if (s.ok()) {
+    // Initial max_total_in_memory_state_ before recovery logs. Log recovery
+    // may check this value to decide whether to flush.
+    max_total_in_memory_state_ = 0;
+    for (auto cfd : *versions_->GetColumnFamilySet()) {
+      auto* mutable_cf_options = cfd->GetLatestMutableCFOptions();
+      max_total_in_memory_state_ += mutable_cf_options->write_buffer_size *
+                                    mutable_cf_options->max_write_buffer_number;
+    }
+
     SequenceNumber next_sequence(kMaxSequenceNumber);
     default_cf_handle_ = new ColumnFamilyHandleImpl(
         versions_->GetColumnFamilySet()->GetDefault(), this, &mutex_);
@@ -527,7 +547,12 @@ Status DBImpl::Recover(
     if (!logs.empty()) {
       // Recover in the order in which the logs were generated
       std::sort(logs.begin(), logs.end());
-      s = RecoverLogFiles(logs, &next_sequence, read_only);
+      bool corrupted_log_found = false;
+      s = RecoverLogFiles(logs, &next_sequence, read_only,
+                          &corrupted_log_found);
+      if (corrupted_log_found && recovered_seq != nullptr) {
+        *recovered_seq = next_sequence;
+      }
       if (!s.ok()) {
         // Clear memtables if recovery failed
         for (auto cfd : *versions_->GetColumnFamilySet()) {
@@ -657,7 +682,8 @@ Status DBImpl::InitPersistStatsColumnFamily() {
 
 // REQUIRES: log_numbers are sorted in ascending order
 Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
-                               SequenceNumber* next_sequence, bool read_only) {
+                               SequenceNumber* next_sequence, bool read_only,
+                               bool* corrupted_log_found) {
   struct LogReporter : public log::Reader::Reporter {
     Env* env;
     Logger* info_log;
@@ -748,9 +774,10 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
 
     std::unique_ptr<SequentialFileReader> file_reader;
     {
-      std::unique_ptr<SequentialFile> file;
-      status = env_->NewSequentialFile(fname, &file,
-                                       env_->OptimizeForLogRead(env_options_));
+      std::unique_ptr<FSSequentialFile> file;
+      status = fs_->NewSequentialFile(fname,
+                                      fs_->OptimizeForLogRead(file_options_),
+                                      &file, nullptr);
       if (!status.ok()) {
         MaybeIgnoreError(&status);
         if (!status.ok()) {
@@ -920,7 +947,7 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
         ColumnFamilyData* cfd;
 
         while ((cfd = flush_scheduler_.TakeNextColumnFamily()) != nullptr) {
-          cfd->Unref();
+          cfd->UnrefAndTryDelete();
           // If this asserts, it means that InsertInto failed in
           // filtering updates to already-flushed column families
           assert(cfd->GetLogNumber() <= log_number);
@@ -958,6 +985,9 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
         status = Status::OK();
         stop_replay_for_corruption = true;
         corrupted_log_number = log_number;
+        if (corrupted_log_found != nullptr) {
+          *corrupted_log_found = true;
+        }
         ROCKS_LOG_INFO(immutable_db_options_.info_log,
                        "Point in time recovered to log #%" PRIu64
                        " seq #%" PRIu64,
@@ -986,6 +1016,9 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
   // the corrupted log number, which means CF contains data beyond the point of
   // corruption. This could during PIT recovery when the WAL is corrupted and
   // some (but not all) CFs are flushed
+  // Exclude the PIT case where no log is dropped after the corruption point.
+  // This is to cover the case for empty logs after corrupted log, in which we
+  // don't reset stop_replay_for_corruption.
   if (stop_replay_for_corruption == true &&
       (immutable_db_options_.wal_recovery_mode ==
            WALRecoveryMode::kPointInTimeRecovery ||
@@ -1119,17 +1152,18 @@ Status DBImpl::RestoreAliveLogFiles(const std::vector<uint64_t>& log_numbers) {
     // preallocated space are not needed anymore. It is likely only the last
     // log has such preallocated space, so we only truncate for the last log.
     if (log_number == log_numbers.back()) {
-      std::unique_ptr<WritableFile> last_log;
-      Status truncate_status = env_->ReopenWritableFile(
-          fname, &last_log,
-          env_->OptimizeForLogWrite(
-              env_options_,
-              BuildDBOptions(immutable_db_options_, mutable_db_options_)));
+      std::unique_ptr<FSWritableFile> last_log;
+      Status truncate_status = fs_->ReopenWritableFile(
+          fname,
+          fs_->OptimizeForLogWrite(
+              file_options_,
+              BuildDBOptions(immutable_db_options_, mutable_db_options_)),
+          &last_log, nullptr);
       if (truncate_status.ok()) {
-        truncate_status = last_log->Truncate(log.size);
+        truncate_status = last_log->Truncate(log.size, IOOptions(), nullptr);
       }
       if (truncate_status.ok()) {
-        truncate_status = last_log->Close();
+        truncate_status = last_log->Close(IOOptions(), nullptr);
       }
       // Not a critical error if fail to truncate.
       if (!truncate_status.ok()) {
@@ -1196,8 +1230,8 @@ Status DBImpl::WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd,
         range_del_iters.emplace_back(range_del_iter);
       }
       s = BuildTable(
-          dbname_, env_, *cfd->ioptions(), mutable_cf_options,
-          env_options_for_compaction_, cfd->table_cache(), iter.get(),
+          dbname_, env_, fs_.get(), *cfd->ioptions(), mutable_cf_options,
+          file_options_for_compaction_, cfd->table_cache(), iter.get(),
           std::move(range_del_iters), &meta, cfd->internal_comparator(),
           cfd->int_tbl_prop_collector_factories(), cfd->GetID(), cfd->GetName(),
           snapshot_seqs, earliest_write_conflict_snapshot, snapshot_checker,
@@ -1226,7 +1260,8 @@ Status DBImpl::WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd,
                   meta.fd.GetFileSize(), meta.smallest, meta.largest,
                   meta.fd.smallest_seqno, meta.fd.largest_seqno,
                   meta.marked_for_compaction, meta.oldest_blob_file_number,
-                  meta.oldest_ancester_time);
+                  meta.oldest_ancester_time, meta.file_creation_time,
+                  meta.file_checksum, meta.file_checksum_func_name);
   }
 
   InternalStats::CompactionStats stats(CompactionReason::kFlush, 1);
@@ -1280,12 +1315,12 @@ Status DB::Open(const DBOptions& db_options, const std::string& dbname,
 Status DBImpl::CreateWAL(uint64_t log_file_num, uint64_t recycle_log_number,
                          size_t preallocate_block_size, log::Writer** new_log) {
   Status s;
-  std::unique_ptr<WritableFile> lfile;
+  std::unique_ptr<FSWritableFile> lfile;
 
   DBOptions db_options =
       BuildDBOptions(immutable_db_options_, mutable_db_options_);
-  EnvOptions opt_env_options =
-      env_->OptimizeForLogWrite(env_options_, db_options);
+  FileOptions opt_file_options =
+      fs_->OptimizeForLogWrite(file_options_, db_options);
   std::string log_fname =
       LogFileName(immutable_db_options_.wal_dir, log_file_num);
 
@@ -1295,10 +1330,12 @@ Status DBImpl::CreateWAL(uint64_t log_file_num, uint64_t recycle_log_number,
                    recycle_log_number);
     std::string old_log_fname =
         LogFileName(immutable_db_options_.wal_dir, recycle_log_number);
-    s = env_->ReuseWritableFile(log_fname, old_log_fname, &lfile,
-                                opt_env_options);
+    TEST_SYNC_POINT("DBImpl::CreateWAL:BeforeReuseWritableFile1");
+    TEST_SYNC_POINT("DBImpl::CreateWAL:BeforeReuseWritableFile2");
+    s = fs_->ReuseWritableFile(log_fname, old_log_fname, opt_file_options,
+                               &lfile, /*dbg=*/nullptr);
   } else {
-    s = NewWritableFile(env_, log_fname, &lfile, opt_env_options);
+    s = NewWritableFile(fs_.get(), log_fname, &lfile, opt_file_options);
   }
 
   if (s.ok()) {
@@ -1307,7 +1344,7 @@ Status DBImpl::CreateWAL(uint64_t log_file_num, uint64_t recycle_log_number,
 
     const auto& listeners = immutable_db_options_.listeners;
     std::unique_ptr<WritableFileWriter> file_writer(
-        new WritableFileWriter(std::move(lfile), log_fname, opt_env_options,
+        new WritableFileWriter(std::move(lfile), log_fname, opt_file_options,
                                env_, nullptr /* stats */, listeners));
     *new_log = new log::Writer(std::move(file_writer), log_file_num,
                                immutable_db_options_.recycle_log_file_num > 0,
@@ -1364,13 +1401,9 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
       impl->error_handler_.EnableAutoRecovery();
     }
   }
-
-  if (!s.ok()) {
-    delete impl;
-    return s;
+  if (s.ok()) {
+    s = impl->CreateArchivalDirectory();
   }
-
-  s = impl->CreateArchivalDirectory();
   if (!s.ok()) {
     delete impl;
     return s;
@@ -1380,7 +1413,8 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
 
   impl->mutex_.Lock();
   // Handles create_if_missing, error_if_exists
-  s = impl->Recover(column_families);
+  uint64_t recovered_seq(kMaxSequenceNumber);
+  s = impl->Recover(column_families, false, false, false, &recovered_seq);
   if (s.ok()) {
     uint64_t new_log_number = impl->versions_->NewFileNumber();
     log::Writer* new_log = nullptr;
@@ -1438,8 +1472,35 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
       if (impl->two_write_queues_) {
         impl->log_write_mutex_.Unlock();
       }
+
       impl->DeleteObsoleteFiles();
-      s = impl->directories_.GetDbDir()->Fsync();
+      s = impl->directories_.GetDbDir()->Fsync(IOOptions(), nullptr);
+    }
+    if (s.ok()) {
+      // In WritePrepared there could be gap in sequence numbers. This breaks
+      // the trick we use in kPointInTimeRecovery which assumes the first seq in
+      // the log right after the corrupted log is one larger than the last seq
+      // we read from the logs. To let this trick keep working, we add a dummy
+      // entry with the expected sequence to the first log right after recovery.
+      // In non-WritePrepared case also the new log after recovery could be
+      // empty, and thus missing the consecutive seq hint to distinguish
+      // middle-log corruption to corrupted-log-remained-after-recovery. This
+      // case also will be addressed by a dummy write.
+      if (recovered_seq != kMaxSequenceNumber) {
+        WriteBatch empty_batch;
+        WriteBatchInternal::SetSequence(&empty_batch, recovered_seq);
+        WriteOptions write_options;
+        uint64_t log_used, log_size;
+        log::Writer* log_writer = impl->logs_.back().writer;
+        s = impl->WriteToWAL(empty_batch, log_writer, &log_used, &log_size);
+        if (s.ok()) {
+          // Need to fsync, otherwise it might get lost after a power reset.
+          s = impl->FlushWAL(false);
+          if (s.ok()) {
+            s = log_writer->file()->Sync(impl->immutable_db_options_.use_fsync);
+          }
+        }
+      }
     }
   }
   if (s.ok() && impl->immutable_db_options_.persist_stats_to_disk) {
@@ -1496,6 +1557,27 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
   if (s.ok() && sfm) {
     // Notify SstFileManager about all sst files that already exist in
     // db_paths[0] and cf_paths[0] when the DB is opened.
+
+    // SstFileManagerImpl needs to know sizes of the files. For files whose size
+    // we already know (sst files that appear in manifest - typically that's the
+    // vast majority of all files), we'll pass the size to SstFileManager.
+    // For all other files SstFileManager will query the size from filesystem.
+
+    std::vector<LiveFileMetaData> metadata;
+
+    impl->mutex_.Lock();
+    impl->versions_->GetLiveFilesMetaData(&metadata);
+    impl->mutex_.Unlock();
+
+    std::unordered_map<std::string, uint64_t> known_file_sizes;
+    for (const auto& md : metadata) {
+      std::string name = md.name;
+      if (!name.empty() && name[0] == '/') {
+        name = name.substr(1);
+      }
+      known_file_sizes[name] = md.size;
+    }
+
     std::vector<std::string> paths;
     paths.emplace_back(impl->immutable_db_options_.db_paths[0].path);
     for (auto& cf : column_families) {
@@ -1515,7 +1597,14 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
         std::string file_path = path + "/" + file_name;
         if (ParseFileName(file_name, &file_number, &file_type) &&
             file_type == kTableFile) {
-          sfm->OnAddFile(file_path);
+          if (known_file_sizes.count(file_name)) {
+            // We're assuming that each sst file name exists in at most one of
+            // the paths.
+            sfm->OnAddFile(file_path, known_file_sizes.at(file_name),
+                           /* compaction */ false);
+          } else {
+            sfm->OnAddFile(file_path);
+          }
         }
       }
     }
@@ -1556,4 +1645,4 @@ Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
   }
   return s;
 }
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/db_impl/db_impl_readonly.cc

@@ -12,7 +12,7 @@
 #include "db/merge_context.h"
 #include "monitoring/perf_context_imp.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 #ifndef ROCKSDB_LITE
 
@@ -48,14 +48,16 @@ Status DBImplReadOnly::Get(const ReadOptions& read_options,
   SequenceNumber max_covering_tombstone_seq = 0;
   LookupKey lkey(key, snapshot);
   PERF_TIMER_STOP(get_snapshot_time);
-  if (super_version->mem->Get(lkey, pinnable_val->GetSelf(), &s, &merge_context,
+  if (super_version->mem->Get(lkey, pinnable_val->GetSelf(),
+                              /*timestamp=*/nullptr, &s, &merge_context,
                               &max_covering_tombstone_seq, read_options)) {
     pinnable_val->PinSelf();
     RecordTick(stats_, MEMTABLE_HIT);
   } else {
     PERF_TIMER_GUARD(get_from_output_files_time);
-    super_version->current->Get(read_options, lkey, pinnable_val, &s,
-                                &merge_context, &max_covering_tombstone_seq);
+    super_version->current->Get(read_options, lkey, pinnable_val,
+                                /*timestamp=*/nullptr, &s, &merge_context,
+                                &max_covering_tombstone_seq);
     RecordTick(stats_, MEMTABLE_MISS);
   }
   RecordTick(stats_, NUMBER_KEYS_READ);
@@ -218,4 +220,4 @@ Status DB::OpenForReadOnly(
 }
 #endif  // !ROCKSDB_LITE
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/db_impl/db_impl_readonly.h

@@ -11,7 +11,7 @@
 #include <vector>
 #include "db/db_impl/db_impl.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class DBImplReadOnly : public DBImpl {
  public:
@@ -132,6 +132,6 @@ class DBImplReadOnly : public DBImpl {
  private:
   friend class DB;
 };
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 #endif  // !ROCKSDB_LITE

db/db_impl/db_impl_secondary.cc

@@ -11,8 +11,9 @@
 #include "db/merge_context.h"
 #include "logging/auto_roll_logger.h"
 #include "monitoring/perf_context_imp.h"
+#include "util/cast_util.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 #ifndef ROCKSDB_LITE
 DBImplSecondary::DBImplSecondary(const DBOptions& db_options,
@@ -28,7 +29,7 @@ DBImplSecondary::~DBImplSecondary() {}
 Status DBImplSecondary::Recover(
     const std::vector<ColumnFamilyDescriptor>& column_families,
     bool /*readonly*/, bool /*error_if_log_file_exist*/,
-    bool /*error_if_data_exists_in_logs*/) {
+    bool /*error_if_data_exists_in_logs*/, uint64_t*) {
   mutex_.AssertHeld();
 
   JobContext job_context(0);
@@ -143,9 +144,10 @@ Status DBImplSecondary::MaybeInitLogReader(
 
     std::unique_ptr<SequentialFileReader> file_reader;
     {
-      std::unique_ptr<SequentialFile> file;
-      Status status = env_->NewSequentialFile(
-          fname, &file, env_->OptimizeForLogRead(env_options_));
+      std::unique_ptr<FSSequentialFile> file;
+      Status status = fs_->NewSequentialFile(
+          fname, fs_->OptimizeForLogRead(file_options_), &file,
+          nullptr);
       if (!status.ok()) {
         *log_reader = nullptr;
         return status;
@@ -338,15 +340,16 @@ Status DBImplSecondary::GetImpl(const ReadOptions& read_options,
   PERF_TIMER_STOP(get_snapshot_time);
 
   bool done = false;
-  if (super_version->mem->Get(lkey, pinnable_val->GetSelf(), &s, &merge_context,
+  if (super_version->mem->Get(lkey, pinnable_val->GetSelf(),
+                              /*timestamp=*/nullptr, &s, &merge_context,
                               &max_covering_tombstone_seq, read_options)) {
     done = true;
     pinnable_val->PinSelf();
     RecordTick(stats_, MEMTABLE_HIT);
   } else if ((s.ok() || s.IsMergeInProgress()) &&
              super_version->imm->Get(
-                 lkey, pinnable_val->GetSelf(), &s, &merge_context,
-                 &max_covering_tombstone_seq, read_options)) {
+                 lkey, pinnable_val->GetSelf(), /*timestamp=*/nullptr, &s,
+                 &merge_context, &max_covering_tombstone_seq, read_options)) {
     done = true;
     pinnable_val->PinSelf();
     RecordTick(stats_, MEMTABLE_HIT);
@@ -357,8 +360,9 @@ Status DBImplSecondary::GetImpl(const ReadOptions& read_options,
   }
   if (!done) {
     PERF_TIMER_GUARD(get_from_output_files_time);
-    super_version->current->Get(read_options, lkey, pinnable_val, &s,
-                                &merge_context, &max_covering_tombstone_seq);
+    super_version->current->Get(read_options, lkey, pinnable_val,
+                                /*timestamp=*/nullptr, &s, &merge_context,
+                                &max_covering_tombstone_seq);
     RecordTick(stats_, MEMTABLE_MISS);
   }
   {
@@ -405,7 +409,7 @@ ArenaWrappedDBIter* DBImplSecondary::NewIteratorImpl(
     const ReadOptions& read_options, ColumnFamilyData* cfd,
     SequenceNumber snapshot, ReadCallback* read_callback) {
   assert(nullptr != cfd);
-  SuperVersion* super_version = cfd->GetReferencedSuperVersion(&mutex_);
+  SuperVersion* super_version = cfd->GetReferencedSuperVersion(this);
   auto db_iter = NewArenaWrappedDbIterator(
       env_, read_options, *cfd->ioptions(), super_version->mutable_cf_options,
       snapshot,
@@ -466,6 +470,11 @@ Status DBImplSecondary::CheckConsistency() {
   // approach and just proceed.
   TEST_SYNC_POINT_CALLBACK(
       "DBImplSecondary::CheckConsistency:AfterFirstAttempt", &s);
+
+  if (immutable_db_options_.skip_checking_sst_file_sizes_on_db_open) {
+    return Status::OK();
+  }
+
   std::vector<LiveFileMetaData> metadata;
   versions_->GetLiveFilesMetaData(&metadata);
 
@@ -497,45 +506,61 @@ Status DBImplSecondary::TryCatchUpWithPrimary() {
   // read the manifest and apply new changes to the secondary instance
   std::unordered_set<ColumnFamilyData*> cfds_changed;
   JobContext job_context(0, true /*create_superversion*/);
-  InstrumentedMutexLock lock_guard(&mutex_);
-  s = static_cast<ReactiveVersionSet*>(versions_.get())
-          ->ReadAndApply(&mutex_, &manifest_reader_, &cfds_changed);
+  {
+    InstrumentedMutexLock lock_guard(&mutex_);
+    s = static_cast_with_check<ReactiveVersionSet>(versions_.get())
+            ->ReadAndApply(&mutex_, &manifest_reader_, &cfds_changed);
 
-  ROCKS_LOG_INFO(immutable_db_options_.info_log, "Last sequence is %" PRIu64,
-                 static_cast<uint64_t>(versions_->LastSequence()));
-  for (ColumnFamilyData* cfd : cfds_changed) {
-    if (cfd->IsDropped()) {
-      ROCKS_LOG_DEBUG(immutable_db_options_.info_log, "[%s] is dropped\n",
-                      cfd->GetName().c_str());
-      continue;
+    ROCKS_LOG_INFO(immutable_db_options_.info_log, "Last sequence is %" PRIu64,
+                   static_cast<uint64_t>(versions_->LastSequence()));
+    for (ColumnFamilyData* cfd : cfds_changed) {
+      if (cfd->IsDropped()) {
+        ROCKS_LOG_DEBUG(immutable_db_options_.info_log, "[%s] is dropped\n",
+                        cfd->GetName().c_str());
+        continue;
+      }
+      VersionStorageInfo::LevelSummaryStorage tmp;
+      ROCKS_LOG_DEBUG(immutable_db_options_.info_log,
+                      "[%s] Level summary: %s\n", cfd->GetName().c_str(),
+                      cfd->current()->storage_info()->LevelSummary(&tmp));
     }
-    VersionStorageInfo::LevelSummaryStorage tmp;
-    ROCKS_LOG_DEBUG(immutable_db_options_.info_log, "[%s] Level summary: %s\n",
-                    cfd->GetName().c_str(),
-                    cfd->current()->storage_info()->LevelSummary(&tmp));
-  }
 
-  // list wal_dir to discover new WALs and apply new changes to the secondary
-  // instance
-  if (s.ok()) {
-    s = FindAndRecoverLogFiles(&cfds_changed, &job_context);
-  }
-  if (s.IsPathNotFound()) {
-    ROCKS_LOG_INFO(immutable_db_options_.info_log,
-                   "Secondary tries to read WAL, but WAL file(s) have already "
-                   "been purged by primary.");
-    s = Status::OK();
-  }
-  if (s.ok()) {
-    for (auto cfd : cfds_changed) {
-      cfd->imm()->RemoveOldMemTables(cfd->GetLogNumber(),
-                                     &job_context.memtables_to_free);
-      auto& sv_context = job_context.superversion_contexts.back();
-      cfd->InstallSuperVersion(&sv_context, &mutex_);
-      sv_context.NewSuperVersion();
+    // list wal_dir to discover new WALs and apply new changes to the secondary
+    // instance
+    if (s.ok()) {
+      s = FindAndRecoverLogFiles(&cfds_changed, &job_context);
+    }
+    if (s.IsPathNotFound()) {
+      ROCKS_LOG_INFO(
+          immutable_db_options_.info_log,
+          "Secondary tries to read WAL, but WAL file(s) have already "
+          "been purged by primary.");
+      s = Status::OK();
+    }
+    if (s.ok()) {
+      for (auto cfd : cfds_changed) {
+        cfd->imm()->RemoveOldMemTables(cfd->GetLogNumber(),
+                                       &job_context.memtables_to_free);
+        auto& sv_context = job_context.superversion_contexts.back();
+        cfd->InstallSuperVersion(&sv_context, &mutex_);
+        sv_context.NewSuperVersion();
+      }
     }
-    job_context.Clean();
   }
+  job_context.Clean();
+
+  // Cleanup unused, obsolete files.
+  JobContext purge_files_job_context(0);
+  {
+    InstrumentedMutexLock lock_guard(&mutex_);
+    // Currently, secondary instance does not own the database files, thus it
+    // is unnecessary for the secondary to force full scan.
+    FindObsoleteFiles(&purge_files_job_context, /*force=*/false);
+  }
+  if (purge_files_job_context.HaveSomethingToDelete()) {
+    PurgeObsoleteFiles(purge_files_job_context);
+  }
+  purge_files_job_context.Clean();
   return s;
 }
 
@@ -583,7 +608,7 @@ Status DB::OpenAsSecondary(
   handles->clear();
   DBImplSecondary* impl = new DBImplSecondary(tmp_opts, dbname);
   impl->versions_.reset(new ReactiveVersionSet(
-      dbname, &impl->immutable_db_options_, impl->env_options_,
+      dbname, &impl->immutable_db_options_, impl->file_options_,
       impl->table_cache_.get(), impl->write_buffer_manager_,
       &impl->write_controller_));
   impl->column_family_memtables_.reset(
@@ -645,4 +670,4 @@ Status DB::OpenAsSecondary(
 }
 #endif  // !ROCKSDB_LITE
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/db_impl/db_impl_secondary.h

@@ -11,7 +11,7 @@
 #include <vector>
 #include "db/db_impl/db_impl.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 // A wrapper class to hold log reader, log reporter, log status.
 class LogReaderContainer {
@@ -78,7 +78,8 @@ class DBImplSecondary : public DBImpl {
   // and log_readers_ to facilitate future operations.
   Status Recover(const std::vector<ColumnFamilyDescriptor>& column_families,
                  bool read_only, bool error_if_log_file_exist,
-                 bool error_if_data_exists_in_logs) override;
+                 bool error_if_data_exists_in_logs,
+                 uint64_t* = nullptr) override;
 
   // Implementations of the DB interface
   using DB::Get;
@@ -172,6 +173,24 @@ class DBImplSecondary : public DBImpl {
     return Status::NotSupported("Not supported operation in secondary mode.");
   }
 
+  using DBImpl::SetDBOptions;
+  Status SetDBOptions(const std::unordered_map<std::string, std::string>&
+                      /*options_map*/) override {
+    // Currently not supported because changing certain options may cause
+    // flush/compaction.
+    return Status::NotSupported("Not supported operation in secondary mode.");
+  }
+
+  using DBImpl::SetOptions;
+  Status SetOptions(
+      ColumnFamilyHandle* /*cfd*/,
+      const std::unordered_map<std::string, std::string>& /*options_map*/)
+      override {
+    // Currently not supported because changing certain options may cause
+    // flush/compaction and/or write to MANIFEST.
+    return Status::NotSupported("Not supported operation in secondary mode.");
+  }
+
   using DBImpl::SyncWAL;
   Status SyncWAL() override {
     return Status::NotSupported("Not supported operation in secondary mode.");
@@ -269,6 +288,14 @@ class DBImplSecondary : public DBImpl {
     return s;
   }
 
+  bool OwnTablesAndLogs() const override {
+    // Currently, the secondary instance does not own the database files. It
+    // simply opens the files of the primary instance and tracks their file
+    // descriptors until they become obsolete. In the future, the secondary may
+    // create links to database files. OwnTablesAndLogs will return true then.
+    return false;
+  }
+
  private:
   friend class DB;
 
@@ -301,6 +328,6 @@ class DBImplSecondary : public DBImpl {
   std::unordered_map<ColumnFamilyData*, uint64_t> cfd_to_current_log_;
 };
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 #endif  // !ROCKSDB_LITE

db/db_impl/db_impl_write.cc

@@ -15,7 +15,7 @@
 #include "options/options_helper.h"
 #include "test_util/sync_point.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 // Convenience methods
 Status DBImpl::Put(const WriteOptions& o, ColumnFamilyHandle* column_family,
                    const Slice& key, const Slice& val) {
@@ -131,6 +131,7 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
                               log_used, log_ref, &seq, sub_batch_cnt,
                               pre_release_callback, kDoAssignOrder,
                               kDoPublishLastSeq, disable_memtable);
+    TEST_SYNC_POINT("DBImpl::WriteImpl:UnorderedWriteAfterWriteWAL");
     if (!status.ok()) {
       return status;
     }
@@ -138,6 +139,7 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
       *seq_used = seq;
     }
     if (!disable_memtable) {
+      TEST_SYNC_POINT("DBImpl::WriteImpl:BeforeUnorderedWriteMemtable");
       status = UnorderedWriteMemtable(write_options, my_batch, callback,
                                       log_ref, seq, sub_batch_cnt);
     }
@@ -1039,7 +1041,7 @@ Status DBImpl::WriteToWAL(const WriteThread::WriteGroup& write_group,
       // We only sync WAL directory the first time WAL syncing is
       // requested, so that in case users never turn on WAL sync,
       // we can avoid the disk I/O in the write code path.
-      status = directories_.GetWalDir()->Fsync();
+      status = directories_.GetWalDir()->Fsync(IOOptions(), nullptr);
     }
   }
 
@@ -1254,7 +1256,7 @@ Status DBImpl::SwitchWAL(WriteContext* write_context) {
   for (const auto cfd : cfds) {
     cfd->Ref();
     status = SwitchMemtable(cfd, write_context);
-    cfd->Unref();
+    cfd->UnrefAndTryDelete();
     if (!status.ok()) {
       break;
     }
@@ -1290,7 +1292,7 @@ Status DBImpl::HandleWriteBufferFull(WriteContext* write_context) {
   // suboptimal but still correct.
   ROCKS_LOG_INFO(
       immutable_db_options_.info_log,
-      "Flushing column family with largest mem table size. Write buffer is "
+      "Flushing column family with oldest memtable entry. Write buffer is "
       "using %" ROCKSDB_PRIszt " bytes out of a total of %" ROCKSDB_PRIszt ".",
       write_buffer_manager_->memory_usage(),
       write_buffer_manager_->buffer_size());
@@ -1333,7 +1335,7 @@ Status DBImpl::HandleWriteBufferFull(WriteContext* write_context) {
     }
     cfd->Ref();
     status = SwitchMemtable(cfd, write_context);
-    cfd->Unref();
+    cfd->UnrefAndTryDelete();
     if (!status.ok()) {
       break;
     }
@@ -1456,7 +1458,7 @@ Status DBImpl::ThrottleLowPriWritesIfNeeded(const WriteOptions& write_options,
       return Status::OK();
     }
     if (write_options.no_slowdown) {
-      return Status::Incomplete();
+      return Status::Incomplete("Low priority write stall");
     } else {
       assert(my_batch != nullptr);
       // Rate limit those writes. The reason that we don't completely wait
@@ -1516,15 +1518,16 @@ Status DBImpl::TrimMemtableHistory(WriteContext* context) {
   for (auto& cfd : cfds) {
     autovector<MemTable*> to_delete;
     cfd->imm()->TrimHistory(&to_delete, cfd->mem()->ApproximateMemoryUsage());
-    for (auto m : to_delete) {
-      delete m;
+    if (!to_delete.empty()) {
+      for (auto m : to_delete) {
+        delete m;
+      }
+      context->superversion_context.NewSuperVersion();
+      assert(context->superversion_context.new_superversion.get() != nullptr);
+      cfd->InstallSuperVersion(&context->superversion_context, &mutex_);
     }
-    context->superversion_context.NewSuperVersion();
-    assert(context->superversion_context.new_superversion.get() != nullptr);
-    cfd->InstallSuperVersion(&context->superversion_context, &mutex_);
 
-    if (cfd->Unref()) {
-      delete cfd;
+    if (cfd->UnrefAndTryDelete()) {
       cfd = nullptr;
     }
   }
@@ -1556,8 +1559,7 @@ Status DBImpl::ScheduleFlushes(WriteContext* context) {
     if (!cfd->mem()->IsEmpty()) {
       status = SwitchMemtable(cfd, context);
     }
-    if (cfd->Unref()) {
-      delete cfd;
+    if (cfd->UnrefAndTryDelete()) {
       cfd = nullptr;
     }
     if (!status.ok()) {
@@ -1629,7 +1631,6 @@ Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
   if (creating_new_log && immutable_db_options_.recycle_log_file_num &&
       !log_recycle_files_.empty()) {
     recycle_log_number = log_recycle_files_.front();
-    log_recycle_files_.pop_front();
   }
   uint64_t new_log_number =
       creating_new_log ? versions_->NewFileNumber() : logfile_number_;
@@ -1666,6 +1667,14 @@ Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
                  ". Immutable memtables: %d.\n",
                  cfd->GetName().c_str(), new_log_number, num_imm_unflushed);
   mutex_.Lock();
+  if (recycle_log_number != 0) {
+    // Since renaming the file is done outside DB mutex, we need to ensure
+    // concurrent full purges don't delete the file while we're recycling it.
+    // To achieve that we hold the old log number in the recyclable list until
+    // after it has been renamed.
+    assert(log_recycle_files_.front() == recycle_log_number);
+    log_recycle_files_.pop_front();
+  }
   if (s.ok() && creating_new_log) {
     log_write_mutex_.Lock();
     assert(new_log != nullptr);
@@ -1827,4 +1836,4 @@ Status DB::Merge(const WriteOptions& opt, ColumnFamilyHandle* column_family,
   }
   return Write(opt, &batch);
 }
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/db_impl/db_secondary_test.cc

@@ -13,7 +13,7 @@
 #include "test_util/fault_injection_test_env.h"
 #include "test_util/sync_point.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 #ifndef ROCKSDB_LITE
 class DBSecondaryTest : public DBTestBase {
@@ -195,6 +195,90 @@ TEST_F(DBSecondaryTest, OpenAsSecondary) {
   verify_db_func("new_foo_value", "new_bar_value");
 }
 
+namespace {
+class TraceFileEnv : public EnvWrapper {
+ public:
+  explicit TraceFileEnv(Env* _target) : EnvWrapper(_target) {}
+  Status NewRandomAccessFile(const std::string& f,
+                             std::unique_ptr<RandomAccessFile>* r,
+                             const EnvOptions& env_options) override {
+    class TracedRandomAccessFile : public RandomAccessFile {
+     public:
+      TracedRandomAccessFile(std::unique_ptr<RandomAccessFile>&& target,
+                             std::atomic<int>& counter)
+          : target_(std::move(target)), files_closed_(counter) {}
+      ~TracedRandomAccessFile() override {
+        files_closed_.fetch_add(1, std::memory_order_relaxed);
+      }
+      Status Read(uint64_t offset, size_t n, Slice* result,
+                  char* scratch) const override {
+        return target_->Read(offset, n, result, scratch);
+      }
+
+     private:
+      std::unique_ptr<RandomAccessFile> target_;
+      std::atomic<int>& files_closed_;
+    };
+    Status s = target()->NewRandomAccessFile(f, r, env_options);
+    if (s.ok()) {
+      r->reset(new TracedRandomAccessFile(std::move(*r), files_closed_));
+    }
+    return s;
+  }
+
+  int files_closed() const {
+    return files_closed_.load(std::memory_order_relaxed);
+  }
+
+ private:
+  std::atomic<int> files_closed_{0};
+};
+}  // namespace
+
+TEST_F(DBSecondaryTest, SecondaryCloseFiles) {
+  Options options;
+  options.env = env_;
+  options.max_open_files = 1;
+  options.disable_auto_compactions = true;
+  Reopen(options);
+  Options options1;
+  std::unique_ptr<Env> traced_env(new TraceFileEnv(env_));
+  options1.env = traced_env.get();
+  OpenSecondary(options1);
+
+  static const auto verify_db = [&]() {
+    std::unique_ptr<Iterator> iter1(dbfull()->NewIterator(ReadOptions()));
+    std::unique_ptr<Iterator> iter2(db_secondary_->NewIterator(ReadOptions()));
+    for (iter1->SeekToFirst(), iter2->SeekToFirst();
+         iter1->Valid() && iter2->Valid(); iter1->Next(), iter2->Next()) {
+      ASSERT_EQ(iter1->key(), iter2->key());
+      ASSERT_EQ(iter1->value(), iter2->value());
+    }
+    ASSERT_FALSE(iter1->Valid());
+    ASSERT_FALSE(iter2->Valid());
+  };
+
+  ASSERT_OK(Put("a", "value"));
+  ASSERT_OK(Put("c", "value"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());
+  verify_db();
+
+  ASSERT_OK(Put("b", "value"));
+  ASSERT_OK(Put("d", "value"));
+  ASSERT_OK(Flush());
+  ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());
+  verify_db();
+
+  ASSERT_OK(dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  ASSERT_OK(db_secondary_->TryCatchUpWithPrimary());
+  ASSERT_EQ(2, static_cast<TraceFileEnv*>(traced_env.get())->files_closed());
+
+  Status s = db_secondary_->SetDBOptions({{"max_open_files", "-1"}});
+  ASSERT_TRUE(s.IsNotSupported());
+  CloseSecondary();
+}
+
 TEST_F(DBSecondaryTest, OpenAsSecondaryWALTailing) {
   Options options;
   options.env = env_;
@@ -776,10 +860,10 @@ TEST_F(DBSecondaryTest, CheckConsistencyWhenOpen) {
 }
 #endif  //! ROCKSDB_LITE
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
-  rocksdb::port::InstallStackTraceHandler();
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
   ::testing::InitGoogleTest(&argc, argv);
   return RUN_ALL_TESTS();
 }

db/db_info_dumper.cc

@@ -14,7 +14,7 @@
 #include "file/filename.h"
 #include "rocksdb/env.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 void DumpDBFileSummary(const ImmutableDBOptions& options,
                        const std::string& dbname) {
@@ -120,4 +120,4 @@ void DumpDBFileSummary(const ImmutableDBOptions& options,
   Header(options.info_log, "Write Ahead Log file in %s: %s\n",
          options.wal_dir.c_str(), wal_info.c_str());
 }
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/db_info_dumper.h

@@ -8,7 +8,7 @@
 
 #include "options/db_options.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 void DumpDBFileSummary(const ImmutableDBOptions& options,
                        const std::string& dbname);
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE

db/db_inplace_update_test.cc

@@ -9,7 +9,7 @@
 #include "db/db_test_util.h"
 #include "port/stack_trace.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class DBTestInPlaceUpdate : public DBTestBase {
  public:
@@ -73,7 +73,7 @@ TEST_F(DBTestInPlaceUpdate, InPlaceUpdateCallbackSmallerSize) {
     options.env = env_;
     options.write_buffer_size = 100000;
     options.inplace_callback =
-      rocksdb::DBTestInPlaceUpdate::updateInPlaceSmallerSize;
+        ROCKSDB_NAMESPACE::DBTestInPlaceUpdate::updateInPlaceSmallerSize;
     options.allow_concurrent_memtable_write = false;
     Reopen(options);
     CreateAndReopenWithCF({"pikachu"}, options);
@@ -102,7 +102,7 @@ TEST_F(DBTestInPlaceUpdate, InPlaceUpdateCallbackSmallerVarintSize) {
     options.env = env_;
     options.write_buffer_size = 100000;
     options.inplace_callback =
-      rocksdb::DBTestInPlaceUpdate::updateInPlaceSmallerVarintSize;
+        ROCKSDB_NAMESPACE::DBTestInPlaceUpdate::updateInPlaceSmallerVarintSize;
     options.allow_concurrent_memtable_write = false;
     Reopen(options);
     CreateAndReopenWithCF({"pikachu"}, options);
@@ -131,7 +131,7 @@ TEST_F(DBTestInPlaceUpdate, InPlaceUpdateCallbackLargeNewValue) {
     options.env = env_;
     options.write_buffer_size = 100000;
     options.inplace_callback =
-      rocksdb::DBTestInPlaceUpdate::updateInPlaceLargerSize;
+        ROCKSDB_NAMESPACE::DBTestInPlaceUpdate::updateInPlaceLargerSize;
     options.allow_concurrent_memtable_write = false;
     Reopen(options);
     CreateAndReopenWithCF({"pikachu"}, options);
@@ -158,7 +158,7 @@ TEST_F(DBTestInPlaceUpdate, InPlaceUpdateCallbackNoAction) {
     options.env = env_;
     options.write_buffer_size = 100000;
     options.inplace_callback =
-        rocksdb::DBTestInPlaceUpdate::updateInPlaceNoAction;
+        ROCKSDB_NAMESPACE::DBTestInPlaceUpdate::updateInPlaceNoAction;
     options.allow_concurrent_memtable_write = false;
     Reopen(options);
     CreateAndReopenWithCF({"pikachu"}, options);
@@ -168,10 +168,10 @@ TEST_F(DBTestInPlaceUpdate, InPlaceUpdateCallbackNoAction) {
     ASSERT_EQ(Get(1, "key"), "NOT_FOUND");
   } while (ChangeCompactOptions());
 }
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
-  rocksdb::port::InstallStackTraceHandler();
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
   ::testing::InitGoogleTest(&argc, argv);
   return RUN_ALL_TESTS();
 }

db/db_io_failure_test.cc

@@ -10,7 +10,7 @@
 #include "db/db_test_util.h"
 #include "port/stack_trace.h"
 
-namespace rocksdb {
+namespace ROCKSDB_NAMESPACE {
 
 class DBIOFailureTest : public DBTestBase {
  public:
@@ -270,14 +270,14 @@ TEST_F(DBIOFailureTest, FlushSstRangeSyncError) {
   Status s;
 
   std::atomic<int> range_sync_called(0);
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "SpecialEnv::SStableFile::RangeSync", [&](void* arg) {
         if (range_sync_called.fetch_add(1) == 0) {
           Status* st = static_cast<Status*>(arg);
           *st = Status::IOError("range sync dummy error");
         }
       });
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   Random rnd(301);
   std::string rnd_str =
@@ -302,7 +302,7 @@ TEST_F(DBIOFailureTest, FlushSstRangeSyncError) {
   ASSERT_NOK(Put(1, "foo2", "bar3"));
   ASSERT_EQ("bar", Get(1, "foo"));
 
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
   ASSERT_GE(1, range_sync_called.load());
 
   ReopenWithColumnFamilies({"default", "pikachu"}, options);
@@ -350,14 +350,14 @@ TEST_F(DBIOFailureTest, CompactSstRangeSyncError) {
   dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
 
   std::atomic<int> range_sync_called(0);
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "SpecialEnv::SStableFile::RangeSync", [&](void* arg) {
         if (range_sync_called.fetch_add(1) == 0) {
           Status* st = static_cast<Status*>(arg);
           *st = Status::IOError("range sync dummy error");
         }
       });
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   ASSERT_OK(dbfull()->SetOptions(handles_[1],
                                  {
@@ -369,7 +369,7 @@ TEST_F(DBIOFailureTest, CompactSstRangeSyncError) {
   ASSERT_NOK(Put(1, "foo2", "bar3"));
   ASSERT_EQ("bar", Get(1, "foo"));
 
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
   ASSERT_GE(1, range_sync_called.load());
 
   ReopenWithColumnFamilies({"default", "pikachu"}, options);
@@ -389,7 +389,7 @@ TEST_F(DBIOFailureTest, FlushSstCloseError) {
   CreateAndReopenWithCF({"pikachu"}, options);
   Status s;
   std::atomic<int> close_called(0);
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "SpecialEnv::SStableFile::Close", [&](void* arg) {
         if (close_called.fetch_add(1) == 0) {
           Status* st = static_cast<Status*>(arg);
@@ -397,7 +397,7 @@ TEST_F(DBIOFailureTest, FlushSstCloseError) {
         }
       });
 
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   ASSERT_OK(Put(1, "foo", "bar"));
   ASSERT_OK(Put(1, "foo1", "bar1"));
@@ -409,7 +409,7 @@ TEST_F(DBIOFailureTest, FlushSstCloseError) {
   ASSERT_EQ("bar2", Get(1, "foo"));
   ASSERT_EQ("bar1", Get(1, "foo1"));
 
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
 
   ReopenWithColumnFamilies({"default", "pikachu"}, options);
   ASSERT_EQ("bar2", Get(1, "foo"));
@@ -441,7 +441,7 @@ TEST_F(DBIOFailureTest, CompactionSstCloseError) {
   dbfull()->TEST_WaitForCompact();
 
   std::atomic<int> close_called(0);
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "SpecialEnv::SStableFile::Close", [&](void* arg) {
         if (close_called.fetch_add(1) == 0) {
           Status* st = static_cast<Status*>(arg);
@@ -449,7 +449,7 @@ TEST_F(DBIOFailureTest, CompactionSstCloseError) {
         }
       });
 
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
   ASSERT_OK(dbfull()->SetOptions(handles_[1],
                                  {
                                      {"disable_auto_compactions", "false"},
@@ -460,7 +460,7 @@ TEST_F(DBIOFailureTest, CompactionSstCloseError) {
   ASSERT_NOK(Put(1, "foo2", "bar3"));
   ASSERT_EQ("bar3", Get(1, "foo"));
 
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
 
   ReopenWithColumnFamilies({"default", "pikachu"}, options);
   ASSERT_EQ("bar3", Get(1, "foo"));
@@ -480,7 +480,7 @@ TEST_F(DBIOFailureTest, FlushSstSyncError) {
   CreateAndReopenWithCF({"pikachu"}, options);
   Status s;
   std::atomic<int> sync_called(0);
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "SpecialEnv::SStableFile::Sync", [&](void* arg) {
         if (sync_called.fetch_add(1) == 0) {
           Status* st = static_cast<Status*>(arg);
@@ -488,7 +488,7 @@ TEST_F(DBIOFailureTest, FlushSstSyncError) {
         }
       });
 
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
 
   ASSERT_OK(Put(1, "foo", "bar"));
   ASSERT_OK(Put(1, "foo1", "bar1"));
@@ -500,7 +500,7 @@ TEST_F(DBIOFailureTest, FlushSstSyncError) {
   ASSERT_EQ("bar2", Get(1, "foo"));
   ASSERT_EQ("bar1", Get(1, "foo1"));
 
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
 
   ReopenWithColumnFamilies({"default", "pikachu"}, options);
   ASSERT_EQ("bar2", Get(1, "foo"));
@@ -533,7 +533,7 @@ TEST_F(DBIOFailureTest, CompactionSstSyncError) {
   dbfull()->TEST_WaitForCompact();
 
   std::atomic<int> sync_called(0);
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->SetCallBack(
       "SpecialEnv::SStableFile::Sync", [&](void* arg) {
         if (sync_called.fetch_add(1) == 0) {
           Status* st = static_cast<Status*>(arg);
@@ -541,7 +541,7 @@ TEST_F(DBIOFailureTest, CompactionSstSyncError) {
         }
       });
 
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->EnableProcessing();
   ASSERT_OK(dbfull()->SetOptions(handles_[1],
                                  {
                                      {"disable_auto_compactions", "false"},
@@ -552,17 +552,17 @@ TEST_F(DBIOFailureTest, CompactionSstSyncError) {
   ASSERT_NOK(Put(1, "foo2", "bar3"));
   ASSERT_EQ("bar3", Get(1, "foo"));
 
-  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  ROCKSDB_NAMESPACE::SyncPoint::GetInstance()->DisableProcessing();
 
   ReopenWithColumnFamilies({"default", "pikachu"}, options);
   ASSERT_EQ("bar3", Get(1, "foo"));
 }
 #endif  // !(defined NDEBUG) || !defined(OS_WIN)
 #endif  // ROCKSDB_LITE
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE
 
 int main(int argc, char** argv) {
-  rocksdb::port::InstallStackTraceHandler();
+  ROCKSDB_NAMESPACE::port::InstallStackTraceHandler();
   ::testing::InitGoogleTest(&argc, argv);
   return RUN_ALL_TESTS();
 }

db/db_iter.cc

@@ -31,20 +31,7 @@
 #include "util/string_util.h"
 #include "util/user_comparator_wrapper.h"
 
-namespace rocksdb {
-
-#if 0
-static void DumpInternalIter(Iterator* iter) {
-  for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
-    ParsedInternalKey k;
-    if (!ParseInternalKey(iter->key(), &k)) {
-      fprintf(stderr, "Corrupt '%s'\n", EscapeString(iter->key()).c_str());
-    } else {
-      fprintf(stderr, "@ '%s'\n", k.DebugString().c_str());
-    }
-  }
-}
-#endif
+namespace ROCKSDB_NAMESPACE {
 
 DBIter::DBIter(Env* _env, const ReadOptions& read_options,
                const ImmutableCFOptions& cf_options,
@@ -62,6 +49,8 @@ DBIter::DBIter(Env* _env, const ReadOptions& read_options,
       read_callback_(read_callback),
       sequence_(s),
       statistics_(cf_options.statistics),
+      max_skip_(max_sequential_skip_in_iterations),
+      max_skippable_internal_keys_(read_options.max_skippable_internal_keys),
       num_internal_keys_skipped_(0),
       iterate_lower_bound_(read_options.iterate_lower_bound),
       iterate_upper_bound_(read_options.iterate_upper_bound),
@@ -73,23 +62,26 @@ DBIter::DBIter(Env* _env, const ReadOptions& read_options,
                                 ? read_options.prefix_same_as_start
                                 : false),
       pin_thru_lifetime_(read_options.pin_data),
-      total_order_seek_(read_options.total_order_seek),
+      expect_total_order_inner_iter_(prefix_extractor_ == nullptr ||
+                                     read_options.total_order_seek ||
+                                     read_options.auto_prefix_mode),
       allow_blob_(allow_blob),
       is_blob_(false),
       arena_mode_(arena_mode),
       range_del_agg_(&cf_options.internal_comparator, s),
       db_impl_(db_impl),
       cfd_(cfd),
-      start_seqnum_(read_options.iter_start_seqnum) {
+      start_seqnum_(read_options.iter_start_seqnum),
+      timestamp_ub_(read_options.timestamp),
+      timestamp_size_(timestamp_ub_ ? timestamp_ub_->size() : 0) {
   RecordTick(statistics_, NO_ITERATOR_CREATED);
-  max_skip_ = max_sequential_skip_in_iterations;
-  max_skippable_internal_keys_ = read_options.max_skippable_internal_keys;
   if (pin_thru_lifetime_) {
     pinned_iters_mgr_.StartPinning();
   }
   if (iter_.iter()) {
     iter_.iter()->SetPinnedItersMgr(&pinned_iters_mgr_);
   }
+  assert(timestamp_size_ == user_comparator_.timestamp_size());
 }
 
 Status DBIter::GetProperty(std::string prop_name, std::string* prop) {
@@ -181,7 +173,7 @@ void DBIter::Next() {
 //       a delete marker or a sequence number higher than sequence_
 //       saved_key_ MUST have a proper user_key before calling this function
 //
-// The prefix parameter, if not null, indicates that we need to iterator
+// The prefix parameter, if not null, indicates that we need to iterate
 // within the prefix, and the iterator needs to be made invalid, if no
 // more entry for the prefix can be found.
 bool DBIter::FindNextUserEntry(bool skipping_saved_key, const Slice* prefix) {
@@ -202,13 +194,14 @@ bool DBIter::FindNextUserEntryInternal(bool skipping_saved_key,
   // or equal to saved_key_. We could skip these entries either because
   // sequence numbers were too high or because skipping_saved_key = true.
   // What saved_key_ contains throughout this method:
-  //  - if skipping_saved_key        : saved_key_ contains the key that we need
-  //  to skip,
-  //                         and we haven't seen any keys greater than that,
-  //  - if num_skipped > 0 : saved_key_ contains the key that we have skipped
-  //                         num_skipped times, and we haven't seen any keys
-  //                         greater than that,
-  //  - none of the above  : saved_key_ can contain anything, it doesn't matter.
+  //  - if skipping_saved_key : saved_key_ contains the key that we need
+  //                            to skip, and we haven't seen any keys greater
+  //                            than that,
+  //  - if num_skipped > 0    : saved_key_ contains the key that we have skipped
+  //                            num_skipped times, and we haven't seen any keys
+  //                            greater than that,
+  //  - none of the above     : saved_key_ can contain anything, it doesn't
+  //                            matter.
   uint64_t num_skipped = 0;
   // For write unprepared, the target sequence number in reseek could be larger
   // than the snapshot, and thus needs to be skipped again. This could result in
@@ -230,9 +223,13 @@ bool DBIter::FindNextUserEntryInternal(bool skipping_saved_key,
     is_key_seqnum_zero_ = (ikey_.sequence == 0);
 
     assert(iterate_upper_bound_ == nullptr || iter_.MayBeOutOfUpperBound() ||
-           user_comparator_.Compare(ikey_.user_key, *iterate_upper_bound_) < 0);
+           user_comparator_.CompareWithoutTimestamp(
+               ikey_.user_key, /*a_has_ts=*/true, *iterate_upper_bound_,
+               /*b_has_ts=*/false) < 0);
     if (iterate_upper_bound_ != nullptr && iter_.MayBeOutOfUpperBound() &&
-        user_comparator_.Compare(ikey_.user_key, *iterate_upper_bound_) >= 0) {
+        user_comparator_.CompareWithoutTimestamp(
+            ikey_.user_key, /*a_has_ts=*/true, *iterate_upper_bound_,
+            /*b_has_ts=*/false) >= 0) {
       break;
     }
 
@@ -247,20 +244,25 @@ bool DBIter::FindNextUserEntryInternal(bool skipping_saved_key,
       return false;
     }
 
-    if (IsVisible(ikey_.sequence)) {
+    assert(ikey_.user_key.size() >= timestamp_size_);
+    Slice ts;
+    if (timestamp_size_ > 0) {
+      ts = ExtractTimestampFromUserKey(ikey_.user_key, timestamp_size_);
+    }
+    if (IsVisible(ikey_.sequence, ts)) {
       // If the previous entry is of seqnum 0, the current entry will not
       // possibly be skipped. This condition can potentially be relaxed to
       // prev_key.seq <= ikey_.sequence. We are cautious because it will be more
       // prone to bugs causing the same user key with the same sequence number.
       if (!is_prev_key_seqnum_zero && skipping_saved_key &&
-          user_comparator_.Compare(ikey_.user_key, saved_key_.GetUserKey()) <=
-              0) {
+          user_comparator_.CompareWithoutTimestamp(
+              ikey_.user_key, saved_key_.GetUserKey()) <= 0) {
         num_skipped++;  // skip this entry
         PERF_COUNTER_ADD(internal_key_skipped_count, 1);
       } else {
         assert(!skipping_saved_key ||
-               user_comparator_.Compare(ikey_.user_key,
-                                        saved_key_.GetUserKey()) > 0);
+               user_comparator_.CompareWithoutTimestamp(
+                   ikey_.user_key, saved_key_.GetUserKey()) > 0);
         num_skipped = 0;
         reseek_done = false;
         switch (ikey_.type) {
@@ -290,9 +292,7 @@ bool DBIter::FindNextUserEntryInternal(bool skipping_saved_key,
             if (start_seqnum_ > 0) {
               // we are taking incremental snapshot here
               // incremental snapshots aren't supported on DB with range deletes
-              assert(!(
-                (ikey_.type == kTypeBlobIndex) && (start_seqnum_ > 0)
-              ));
+              assert(ikey_.type != kTypeBlobIndex);
               if (ikey_.sequence >= start_seqnum_) {
                 saved_key_.SetInternalKey(ikey_);
                 valid_ = true;
@@ -323,7 +323,7 @@ bool DBIter::FindNextUserEntryInternal(bool skipping_saved_key,
                   ROCKS_LOG_ERROR(logger_, "Encounter unexpected blob index.");
                   status_ = Status::NotSupported(
                       "Encounter unexpected blob index. Please open DB with "
-                      "rocksdb::blob_db::BlobDB instead.");
+                      "ROCKSDB_NAMESPACE::blob_db::BlobDB instead.");
                   valid_ = false;
                   return false;
                 }
@@ -368,9 +368,9 @@ bool DBIter::FindNextUserEntryInternal(bool skipping_saved_key,
       // This key was inserted after our snapshot was taken.
       // If this happens too many times in a row for the same user key, we want
       // to seek to the target sequence number.
-      int cmp =
-          user_comparator_.Compare(ikey_.user_key, saved_key_.GetUserKey());
-      if (cmp == 0 || (skipping_saved_key && cmp <= 0)) {
+      int cmp = user_comparator_.CompareWithoutTimestamp(
+          ikey_.user_key, saved_key_.GetUserKey());
+      if (cmp == 0 || (skipping_saved_key && cmp < 0)) {
         num_skipped++;
       } else {
         saved_key_.SetUserKey(
@@ -389,7 +389,7 @@ bool DBIter::FindNextUserEntryInternal(bool skipping_saved_key,
     // reseek previously.
     //
     // TODO(lth): If we reseek to sequence number greater than ikey_.sequence,
-    // than it does not make sense to reseek as we would actually land further
+    // then it does not make sense to reseek as we would actually land further
     // away from the desired key. There is opportunity for optimization here.
     if (num_skipped > max_skip_ && !reseek_done) {
       is_key_seqnum_zero_ = false;
@@ -400,8 +400,17 @@ bool DBIter::FindNextUserEntryInternal(bool skipping_saved_key,
         // We're looking for the next user-key but all we see are the same
         // user-key with decreasing sequence numbers. Fast forward to
         // sequence number 0 and type deletion (the smallest type).
-        AppendInternalKey(&last_key, ParsedInternalKey(saved_key_.GetUserKey(),
-                                                       0, kTypeDeletion));
+        if (timestamp_size_ == 0) {
+          AppendInternalKey(
+              &last_key,
+              ParsedInternalKey(saved_key_.GetUserKey(), 0, kTypeDeletion));
+        } else {
+          std::string min_ts(timestamp_size_, static_cast<char>(0));
+          AppendInternalKeyWithDifferentTimestamp(
+              &last_key,
+              ParsedInternalKey(saved_key_.GetUserKey(), 0, kTypeDeletion),
+              min_ts);
+        }
         // Don't set skipping_saved_key = false because we may still see more
         // user-keys equal to saved_key_.
       } else {
@@ -410,9 +419,17 @@ bool DBIter::FindNextUserEntryInternal(bool skipping_saved_key,
         // Note that this only covers a case when a higher key was overwritten
         // many times since our snapshot was taken, not the case when a lot of
         // different keys were inserted after our snapshot was taken.
-        AppendInternalKey(&last_key,
-                          ParsedInternalKey(saved_key_.GetUserKey(), sequence_,
-                                            kValueTypeForSeek));
+        if (timestamp_size_ == 0) {
+          AppendInternalKey(
+              &last_key, ParsedInternalKey(saved_key_.GetUserKey(), sequence_,
+                                           kValueTypeForSeek));
+        } else {
+          AppendInternalKeyWithDifferentTimestamp(
+              &last_key,
+              ParsedInternalKey(saved_key_.GetUserKey(), sequence_,
+                                kValueTypeForSeek),
+              *timestamp_ub_);
+        }
       }
       iter_.Seek(last_key);
       RecordTick(statistics_, NUMBER_OF_RESEEKS_IN_ITERATION);
@@ -495,7 +512,7 @@ bool DBIter::MergeValuesNewToOld() {
         ROCKS_LOG_ERROR(logger_, "Encounter unexpected blob index.");
         status_ = Status::NotSupported(
             "Encounter unexpected blob index. Please open DB with "
-            "rocksdb::blob_db::BlobDB instead.");
+            "ROCKSDB_NAMESPACE::blob_db::BlobDB instead.");
       } else {
         status_ =
             Status::NotSupported("Blob DB does not support merge operator.");
@@ -531,6 +548,13 @@ bool DBIter::MergeValuesNewToOld() {
 }
 
 void DBIter::Prev() {
+  if (timestamp_size_ > 0) {
+    valid_ = false;
+    status_ = Status::NotSupported(
+        "SeekToLast/SeekForPrev/Prev currently not supported with timestamp.");
+    return;
+  }
+
   assert(valid_);
   assert(status_.ok());
 
@@ -567,7 +591,7 @@ bool DBIter::ReverseToForward() {
   // When moving backwards, iter_ is positioned on _previous_ key, which may
   // not exist or may have different prefix than the current key().
   // If that's the case, seek iter_ to current key.
-  if ((prefix_extractor_ != nullptr && !total_order_seek_) || !iter_.Valid()) {
+  if (!expect_total_order_inner_iter() || !iter_.Valid()) {
     IterKey last_key;
     last_key.SetInternalKey(ParsedInternalKey(
         saved_key_.GetUserKey(), kMaxSequenceNumber, kValueTypeForSeek));
@@ -603,15 +627,14 @@ bool DBIter::ReverseToBackward() {
   // key, which may not exist or may have prefix different from current.
   // If that's the case, seek to saved_key_.
   if (current_entry_is_merged_ &&
-      ((prefix_extractor_ != nullptr && !total_order_seek_) ||
-       !iter_.Valid())) {
+      (!expect_total_order_inner_iter() || !iter_.Valid())) {
     IterKey last_key;
     // Using kMaxSequenceNumber and kValueTypeForSeek
     // (not kValueTypeForSeekForPrev) to seek to a key strictly smaller
     // than saved_key_.
     last_key.SetInternalKey(ParsedInternalKey(
         saved_key_.GetUserKey(), kMaxSequenceNumber, kValueTypeForSeek));
-    if (prefix_extractor_ != nullptr && !total_order_seek_) {
+    if (!expect_total_order_inner_iter()) {
       iter_.SeekForPrev(last_key.GetInternalKey());
     } else {
       // Some iterators may not support SeekForPrev(), so we avoid using it
@@ -710,7 +733,13 @@ bool DBIter::FindValueForCurrentKey() {
       return false;
     }
 
-    if (!IsVisible(ikey.sequence) ||
+    assert(ikey.user_key.size() >= timestamp_size_);
+    Slice ts;
+    if (timestamp_size_ > 0) {
+      ts = Slice(ikey.user_key.data() + ikey.user_key.size() - timestamp_size_,
+                 timestamp_size_);
+    }
+    if (!IsVisible(ikey.sequence, ts) ||
         !user_comparator_.Equal(ikey.user_key, saved_key_.GetUserKey())) {
       break;
     }
@@ -797,7 +826,7 @@ bool DBIter::FindValueForCurrentKey() {
           ROCKS_LOG_ERROR(logger_, "Encounter unexpected blob index.");
           status_ = Status::NotSupported(
               "Encounter unexpected blob index. Please open DB with "
-              "rocksdb::blob_db::BlobDB instead.");
+              "ROCKSDB_NAMESPACE::blob_db::BlobDB instead.");
         } else {
           status_ =
               Status::NotSupported("Blob DB does not support merge operator.");
@@ -820,7 +849,7 @@ bool DBIter::FindValueForCurrentKey() {
         ROCKS_LOG_ERROR(logger_, "Encounter unexpected blob index.");
         status_ = Status::NotSupported(
             "Encounter unexpected blob index. Please open DB with "
-            "rocksdb::blob_db::BlobDB instead.");
+            "ROCKSDB_NAMESPACE::blob_db::BlobDB instead.");
         valid_ = false;
         return false;
       }
@@ -866,6 +895,13 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
     if (!ParseKey(&ikey)) {
       return false;
     }
+    assert(ikey.user_key.size() >= timestamp_size_);
+    Slice ts;
+    if (timestamp_size_ > 0) {
+      ts = Slice(ikey.user_key.data() + ikey.user_key.size() - timestamp_size_,
+                 timestamp_size_);
+    }
+
     if (!user_comparator_.Equal(ikey.user_key, saved_key_.GetUserKey())) {
       // No visible values for this key, even though FindValueForCurrentKey()
       // has seen some. This is possible if we're using a tailing iterator, and
@@ -874,7 +910,7 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
       return true;
     }
 
-    if (IsVisible(ikey.sequence)) {
+    if (IsVisible(ikey.sequence, ts)) {
       break;
     }
 
@@ -891,7 +927,7 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
     ROCKS_LOG_ERROR(logger_, "Encounter unexpected blob index.");
     status_ = Status::NotSupported(
         "Encounter unexpected blob index. Please open DB with "
-        "rocksdb::blob_db::BlobDB instead.");
+        "ROCKSDB_NAMESPACE::blob_db::BlobDB instead.");
     valid_ = false;
     return false;
   }
@@ -953,7 +989,7 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
         ROCKS_LOG_ERROR(logger_, "Encounter unexpected blob index.");
         status_ = Status::NotSupported(
             "Encounter unexpected blob index. Please open DB with "
-            "rocksdb::blob_db::BlobDB instead.");
+            "ROCKSDB_NAMESPACE::blob_db::BlobDB instead.");
       } else {
         status_ =
             Status::NotSupported("Blob DB does not support merge operator.");
@@ -978,8 +1014,8 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
   // Make sure we leave iter_ in a good state. If it's valid and we don't care
   // about prefixes, that's already good enough. Otherwise it needs to be
   // seeked to the current key.
-  if ((prefix_extractor_ != nullptr && !total_order_seek_) || !iter_.Valid()) {
-    if (prefix_extractor_ != nullptr && !total_order_seek_) {
+  if (!expect_total_order_inner_iter() || !iter_.Valid()) {
+    if (!expect_total_order_inner_iter()) {
       iter_.SeekForPrev(last_key);
     } else {
       iter_.Seek(last_key);
@@ -1014,7 +1050,13 @@ bool DBIter::FindUserKeyBeforeSavedKey() {
     }
 
     assert(ikey.sequence != kMaxSequenceNumber);
-    if (!IsVisible(ikey.sequence)) {
+    assert(ikey.user_key.size() >= timestamp_size_);
+    Slice ts;
+    if (timestamp_size_ > 0) {
+      ts = Slice(ikey.user_key.data() + ikey.user_key.size() - timestamp_size_,
+                 timestamp_size_);
+    }
+    if (!IsVisible(ikey.sequence, ts)) {
       PERF_COUNTER_ADD(internal_recent_skipped_count, 1);
     } else {
       PERF_COUNTER_ADD(internal_key_skipped_count, 1);
@@ -1059,10 +1101,18 @@ bool DBIter::TooManyInternalKeysSkipped(bool increment) {
   return false;
 }
 
-bool DBIter::IsVisible(SequenceNumber sequence) {
+bool DBIter::IsVisible(SequenceNumber sequence, const Slice& ts) {
+  // Remember that comparator orders preceding timestamp as larger.
+  int cmp_ts = timestamp_ub_ != nullptr
+                   ? user_comparator_.CompareTimestamp(ts, *timestamp_ub_)
+                   : 0;
+  if (cmp_ts > 0) {
+    return false;
+  }
   if (read_callback_ == nullptr) {
     return sequence <= sequence_;
   } else {
+    // TODO(yanqin): support timestamp in read_callback_.
     return read_callback_->IsVisible(sequence);
   }
 }
@@ -1071,14 +1121,16 @@ void DBIter::SetSavedKeyToSeekTarget(const Slice& target) {
   is_key_seqnum_zero_ = false;
   SequenceNumber seq = sequence_;
   saved_key_.Clear();
-  saved_key_.SetInternalKey(target, seq);
+  saved_key_.SetInternalKey(target, seq, kValueTypeForSeek, timestamp_ub_);
 
   if (iterate_lower_bound_ != nullptr &&
-      user_comparator_.Compare(saved_key_.GetUserKey(), *iterate_lower_bound_) <
-          0) {
+      user_comparator_.CompareWithoutTimestamp(
+          saved_key_.GetUserKey(), /*a_has_ts=*/true, *iterate_lower_bound_,
+          /*b_has_ts=*/false) < 0) {
     // Seek key is smaller than the lower bound.
     saved_key_.Clear();
-    saved_key_.SetInternalKey(*iterate_lower_bound_, seq);
+    saved_key_.SetInternalKey(*iterate_lower_bound_, seq, kValueTypeForSeek,
+                              timestamp_ub_);
   }
 }
 
@@ -1129,18 +1181,16 @@ void DBIter::Seek(const Slice& target) {
 
   // Now the inner iterator is placed to the target position. From there,
   // we need to find out the next key that is visible to the user.
-  //
   ClearSavedValue();
   if (prefix_same_as_start_) {
     // The case where the iterator needs to be invalidated if it has exausted
     // keys within the same prefix of the seek key.
     assert(prefix_extractor_ != nullptr);
-    Slice target_prefix;
-    target_prefix = prefix_extractor_->Transform(target);
+    Slice target_prefix = prefix_extractor_->Transform(target);
     FindNextUserEntry(false /* not skipping saved_key */,
                       &target_prefix /* prefix */);
     if (valid_) {
-      // Remember the prefix of the seek key for the future Prev() call to
+      // Remember the prefix of the seek key for the future Next() call to
       // check.
       prefix_.SetUserKey(target_prefix);
     }
@@ -1170,6 +1220,13 @@ void DBIter::SeekForPrev(const Slice& target) {
   }
 #endif  // ROCKSDB_LITE
 
+  if (timestamp_size_ > 0) {
+    valid_ = false;
+    status_ = Status::NotSupported(
+        "SeekToLast/SeekForPrev/Prev currently not supported with timestamp.");
+    return;
+  }
+
   status_ = Status::OK();
   ReleaseTempPinnedData();
   ResetInternalKeysSkippedCounter();
@@ -1196,8 +1253,7 @@ void DBIter::SeekForPrev(const Slice& target) {
     // The case where the iterator needs to be invalidated if it has exausted
     // keys within the same prefix of the seek key.
     assert(prefix_extractor_ != nullptr);
-    Slice target_prefix;
-    target_prefix = prefix_extractor_->Transform(target);
+    Slice target_prefix = prefix_extractor_->Transform(target);
     PrevInternal(&target_prefix);
     if (valid_) {
       // Remember the prefix of the seek key for the future Prev() call to
@@ -1224,7 +1280,7 @@ void DBIter::SeekToFirst() {
   PERF_CPU_TIMER_GUARD(iter_seek_cpu_nanos, env_);
   // Don't use iter_::Seek() if we set a prefix extractor
   // because prefix seek will be used.
-  if (prefix_extractor_ != nullptr && !total_order_seek_) {
+  if (!expect_total_order_inner_iter()) {
     max_skip_ = std::numeric_limits<uint64_t>::max();
   }
   status_ = Status::OK();
@@ -1264,6 +1320,13 @@ void DBIter::SeekToFirst() {
 }
 
 void DBIter::SeekToLast() {
+  if (timestamp_size_ > 0) {
+    valid_ = false;
+    status_ = Status::NotSupported(
+        "SeekToLast/SeekForPrev/Prev currently not supported with timestamp.");
+    return;
+  }
+
   if (iterate_upper_bound_ != nullptr) {
     // Seek to last key strictly less than ReadOptions.iterate_upper_bound.
     SeekForPrev(*iterate_upper_bound_);
@@ -1277,7 +1340,7 @@ void DBIter::SeekToLast() {
   PERF_CPU_TIMER_GUARD(iter_seek_cpu_nanos, env_);
   // Don't use iter_::Seek() if we set a prefix extractor
   // because prefix seek will be used.
-  if (prefix_extractor_ != nullptr && !total_order_seek_) {
+  if (!expect_total_order_inner_iter()) {
     max_skip_ = std::numeric_limits<uint64_t>::max();
   }
   status_ = Status::OK();
@@ -1323,4 +1386,4 @@ Iterator* NewDBIterator(Env* env, const ReadOptions& read_options,
   return db_iter;
 }
 
-}  // namespace rocksdb
+}  // namespace ROCKSDB_NAMESPACE