StackableDB optionally take shared ownership of the underlying DB

Summary:
This is a pre-cleaning up before a major block based table iterator refactoring. BlockBasedTable::NewDataBlockIterator() will always return BlockIter. This simplifies the logic and code and enable further refactoring and optimization.
Closes https://github.com/facebook/rocksdb/pull/3398

Differential Revision: D6780165

Pulled By: siying

fbshipit-source-id: 273f7dc896724f682c0118fb69a359d9cc4418b4

.deprecated_arcconfig

@@ -1,17 +0,0 @@
-{
-  "project_id" : "rocksdb",
-  "conduit_uri" : "https://phabricator.fb.com/api/",
-  "copyright_holder" : "Facebook",
-  "load" : [
-    "arcanist_util"
-  ],
-  "lint.engine" : "FacebookFbcodeLintEngine",
-  "lint.engine.single.linter" : "FbcodeCppLinter",
-  "unit.engine" : "FacebookFbcodeUnitTestEngine",
-  "arcanist_configuration" : "FacebookArcanistConfiguration",
-  "base" : "git:HEAD^, hg:.^",
-  "git.default-relative-commit" : "HEAD^",
-  "git:arc.feature.start.default" : "origin/master",
-  "arc.feature.start.default" : "master",
-  "history.immutable" : false
-}

.gitignore

@@ -39,7 +39,6 @@ coverage/COVERAGE_REPORT
 .gdbhistory
 .gdb_history
 package/
-.phutil_module_cache
 unity.a
 tags
 etags

.travis.yml

@@ -21,6 +21,7 @@ env:
   - TEST_GROUP=platform_dependent # 16-18 minutes
   - TEST_GROUP=1 # 33-35 minutes
   - TEST_GROUP=2 # 30-32 minutes
+  - TEST_GROUP=3 # ? minutes - under development
   # Run java tests
   - JOB_NAME=java_test # 4-11 minutes
   # Build ROCKSDB_LITE
@@ -36,6 +37,8 @@ matrix:
     env: TEST_GROUP=1
   - os: osx
     env: TEST_GROUP=2
+  - os: osx
+    env: TEST_GROUP=3
   - os : osx
     env: JOB_NAME=cmake-mingw
   - os : linux
@@ -49,6 +52,9 @@ install:
       brew install ccache;
       PATH=$PATH:/usr/local/opt/ccache/libexec;
     fi
+  - if [[ "${JOB_NAME}" == cmake* ]] && [ "${TRAVIS_OS_NAME}" == linux ]; then
+      mkdir cmake-dist && curl -sfSL https://cmake.org/files/v3.8/cmake-3.8.1-Linux-x86_64.tar.gz | tar --strip-components=1 -C cmake-dist -xz && export PATH=$PWD/cmake-dist/bin:$PATH;
+    fi
 
 before_script:
   # Increase the maximum number of open file descriptors, since some tests use
@@ -57,14 +63,16 @@ before_script:
 
 script:
   - ${CXX} --version
-  - if [ "${TEST_GROUP}" == 'platform_dependent' ]; then OPT=-DTRAVIS V=1 make -j4 all; OPT=-DTRAVIS V=1 ROCKSDBTESTS_END=db_block_cache_test make -j4 check_some; fi
+  - if [ `command -v ccache` ]; then ccache -C; fi
+  - if [ "${TEST_GROUP}" == 'platform_dependent' ]; then OPT=-DTRAVIS V=1 ROCKSDBTESTS_END=db_block_cache_test make -j4 all_but_some_tests check_some; fi
   - if [ "${TEST_GROUP}" == '1' ]; then OPT=-DTRAVIS V=1 ROCKSDBTESTS_START=db_block_cache_test ROCKSDBTESTS_END=comparator_db_test make -j4 check_some; fi
-  - if [ "${TEST_GROUP}" == '2' ]; then OPT=-DTRAVIS V=1 ROCKSDBTESTS_START=comparator_db_test make -j4 check_some; fi
+  - if [ "${TEST_GROUP}" == '2' ]; then OPT=-DTRAVIS V=1 ROCKSDBTESTS_START=comparator_db_test ROCKSDBTESTS_END=write_prepared_transaction_test make -j4 check_some; fi
+  - if [ "${TEST_GROUP}" == '3' ]; then OPT=-DTRAVIS V=1 ROCKSDBTESTS_START=write_prepared_transaction_test make -j4 check_some; fi
   - if [ "${JOB_NAME}" == 'java_test' ]; then OPT=-DTRAVIS V=1 make clean jclean && make rocksdbjava jtest; fi
-  - if [ "${JOB_NAME}" == 'lite_build' ]; then OPT="-DTRAVIS -DROCKSDB_LITE" V=1 make -j4 static_lib; fi
+  - if [ "${JOB_NAME}" == 'lite_build' ]; then OPT="-DTRAVIS -DROCKSDB_LITE" V=1 make -j4 static_lib tools; fi
   - if [ "${JOB_NAME}" == 'examples' ]; then OPT=-DTRAVIS V=1 make -j4 static_lib; cd examples; make -j4; fi
-  - if [ "${JOB_NAME}" == 'cmake' ]; then mkdir build && cd build && cmake .. && make -j4 rocksdb; fi
-  - if [ "${JOB_NAME}" == 'cmake-mingw' ]; then mkdir build && cd build && cmake .. -DCMAKE_C_COMPILER=x86_64-w64-mingw32-gcc -DCMAKE_CXX_COMPILER=x86_64-w64-mingw32-g++ -DCMAKE_SYSTEM_NAME=Windows && make -j4 rocksdb; fi
+  - if [ "${JOB_NAME}" == 'cmake' ]; then mkdir build && cd build && cmake -DJNI=1 .. && make -j4 rocksdb rocksdbjni; fi
+  - if [ "${JOB_NAME}" == 'cmake-mingw' ]; then 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; fi
 notifications:
     email:
       - leveldb@fb.com

AUTHORS

@@ -9,3 +9,4 @@ Sanjay Ghemawat <sanjay@google.com>
 # Partial list of contributors:
 Kevin Regan <kevin.d.regan@gmail.com>
 Johan Bilien <jobi@litl.com>
+Matthew Von-Maszewski <https://github.com/matthewvon> (Basho Technologies)

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 14 Win64" -DGFLAGS=1 -DSNAPPY=1 -DJEMALLOC=1 -DJNI=1 ..
+#        sample command: cmake -G "Visual Studio 15 Win64" -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 ot use all avail cores)
 #        msbuild rocksdb.sln
@@ -32,19 +32,29 @@
 # 3. cmake ..
 # 4. make -j
 
-cmake_minimum_required(VERSION 2.6)
+cmake_minimum_required(VERSION 2.8.12)
 project(rocksdb)
+enable_language(CXX)
+enable_language(C)
+enable_language(ASM)
 
 if(POLICY CMP0042)
   cmake_policy(SET CMP0042 NEW)
 endif()
 
-list(APPEND CMAKE_MODULE_PATH "${CMAKE_SOURCE_DIR}/cmake/modules/")
+list(APPEND CMAKE_MODULE_PATH "${CMAKE_CURRENT_LIST_DIR}/cmake/modules/")
 
+option(WITH_JEMALLOC "build with JeMalloc" OFF)
+option(WITH_SNAPPY "build with SNAPPY" OFF)
+option(WITH_LZ4 "build with lz4" OFF)
+option(WITH_ZLIB "build with zlib" 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(${CMAKE_CURRENT_SOURCE_DIR}/thirdparty.inc)
 else()
-  option(WITH_JEMALLOC "build with JeMalloc" OFF)
   if(CMAKE_SYSTEM_NAME MATCHES "FreeBSD")
     # FreeBSD has jemaloc as default malloc
     # but it does not have all the jemalloc files in include/...
@@ -56,26 +66,57 @@ else()
       include_directories(${JEMALLOC_INCLUDE_DIR})
     endif()
   endif()
-  option(WITH_SNAPPY "build with SNAPPY" OFF)
+  
+  # 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 ${gflags_LIBRARIES})
+    endif()
+  endif()
+
   if(WITH_SNAPPY)
     find_package(snappy REQUIRED)
     add_definitions(-DSNAPPY)
     include_directories(${SNAPPY_INCLUDE_DIR})
     list(APPEND THIRDPARTY_LIBS ${SNAPPY_LIBRARIES})
   endif()
+
+  if(WITH_ZLIB)
+    find_package(zlib REQUIRED)
+    add_definitions(-DZLIB)
+    include_directories(${ZLIB_INCLUDE_DIR})
+    list(APPEND THIRDPARTY_LIBS ${ZLIB_LIBRARIES})
+  endif()
+
+  option(WITH_BZ2 "build with bzip2" OFF)
+  if(WITH_BZ2)
+    find_package(bzip2 REQUIRED)
+    add_definitions(-DBZIP2)
+    include_directories(${BZIP2_INCLUDE_DIR})
+    list(APPEND THIRDPARTY_LIBS ${BZIP2_LIBRARIES})
+  endif()
+
+  if(WITH_LZ4)
+    find_package(lz4 REQUIRED)
+    add_definitions(-DLZ4)
+    include_directories(${LZ4_INCLUDE_DIR})
+    list(APPEND THIRDPARTY_LIBS ${LZ4_LIBRARIES})
+  endif()
+
+  option(WITH_ZSTD "build with zstd" OFF)
+  if(WITH_ZSTD)
+    find_package(zstd REQUIRED)
+    add_definitions(-DZSTD)
+    include_directories(${ZSTD_INCLUDE_DIR})
+    list(APPEND THIRDPARTY_LIBS ${ZSTD_LIBRARIES})
+  endif()
 endif()
 
-if(WIN32)
-  execute_process(COMMAND powershell -noprofile -Command "Get-Date -format MM_dd_yyyy" OUTPUT_VARIABLE DATE)
-  execute_process(COMMAND powershell -noprofile -Command "Get-Date -format HH:mm:ss" OUTPUT_VARIABLE TIME)
-  string(REGEX REPLACE "(..)_(..)_..(..).*" "\\1/\\2/\\3" DATE "${DATE}")
-  string(REGEX REPLACE "(..):(.....).*" " \\1:\\2" TIME "${TIME}")
-  set(GIT_DATE_TIME "${DATE} ${TIME}")
-else()
-  execute_process(COMMAND date "+%Y/%m/%d %H:%M:%S" OUTPUT_VARIABLE DATETIME)
-  string(REGEX REPLACE "\n" "" DATETIME ${DATETIME})
-  set(GIT_DATE_TIME "${DATETIME}")
-endif()
+string(TIMESTAMP GIT_DATE_TIME "%Y/%m/%d %H:%M:%S" UTC)
 
 find_package(Git)
 
@@ -91,20 +132,19 @@ endif()
 
 string(REGEX REPLACE "[^0-9a-f]+" "" GIT_SHA "${GIT_SHA}")
 
-if(NOT WIN32)
-  execute_process(COMMAND
-      "./build_tools/version.sh" "full"
-      WORKING_DIRECTORY ${PROJECT_SOURCE_DIR}
-      OUTPUT_VARIABLE ROCKSDB_VERSION
-  )
-  string(STRIP "${ROCKSDB_VERSION}" ROCKSDB_VERSION)
-  execute_process(COMMAND
-      "./build_tools/version.sh" "major"
-      WORKING_DIRECTORY ${PROJECT_SOURCE_DIR}
-      OUTPUT_VARIABLE ROCKSDB_VERSION_MAJOR
-  )
-  string(STRIP "${ROCKSDB_VERSION_MAJOR}" ROCKSDB_VERSION_MAJOR)
-endif()
+set(SH_CMD "sh")
+execute_process(COMMAND
+  ${SH_CMD} -c "build_tools/version.sh full"
+  WORKING_DIRECTORY ${PROJECT_SOURCE_DIR}
+  OUTPUT_VARIABLE ROCKSDB_VERSION
+)
+string(STRIP "${ROCKSDB_VERSION}" ROCKSDB_VERSION)
+execute_process(COMMAND
+  ${SH_CMD} -c "build_tools/version.sh major"
+  WORKING_DIRECTORY ${PROJECT_SOURCE_DIR}
+  OUTPUT_VARIABLE ROCKSDB_VERSION_MAJOR
+)
+string(STRIP "${ROCKSDB_VERSION_MAJOR}" ROCKSDB_VERSION_MAJOR)
 
 option(WITH_MD_LIBRARY "build with MD" ON)
 if(WIN32 AND MSVC)
@@ -115,31 +155,78 @@ if(WIN32 AND MSVC)
   endif()
 endif()
 
+set(BUILD_VERSION_CC ${CMAKE_BINARY_DIR}/build_version.cc)
+configure_file(util/build_version.cc.in ${BUILD_VERSION_CC} @ONLY)
+add_library(build_version OBJECT ${BUILD_VERSION_CC})
+target_include_directories(build_version PRIVATE
+  ${CMAKE_CURRENT_SOURCE_DIR}/util)
+if(MSVC)
+  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /Zi /nologo /EHsc /GS /Gd /GR /GF /fp:precise /Zc:wchar_t /Zc:forScope /errorReport:queue")
+  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /FC /d2Zi+ /W4 /wd4127 /wd4800 /wd4996 /wd4351 /wd4100 /wd4204 /wd4324")
+else()
+  set(CMAKE_CXX_FLAGS "${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")
+  endif()
+  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")
+  if(NOT CMAKE_BUILD_TYPE STREQUAL "Debug")
+    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -O2 -fno-omit-frame-pointer")
+    include(CheckCXXCompilerFlag)
+    CHECK_CXX_COMPILER_FLAG("-momit-leaf-frame-pointer" HAVE_OMIT_LEAF_FRAME_POINTER)
+    if(HAVE_OMIT_LEAF_FRAME_POINTER)
+      set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -momit-leaf-frame-pointer")
+    endif()
+  endif()
+endif()
+
+include(CheckCCompilerFlag)
+if(CMAKE_SYSTEM_PROCESSOR MATCHES "ppc64le")
+  CHECK_C_COMPILER_FLAG("-maltivec" HAS_ALTIVEC)
+  if(HAS_ALTIVEC)
+    message(STATUS " HAS_ALTIVEC yes")
+    set(CMAKE_C_FLAGS "${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 "${CMAKE_CXX_FLAGS} -mcpu=power8")
+  endif(HAS_ALTIVEC)
+endif(CMAKE_SYSTEM_PROCESSOR MATCHES "ppc64le")
+
 option(PORTABLE "build a portable binary" OFF)
 option(FORCE_SSE42 "force building with SSE4.2, even when PORTABLE=ON" OFF)
 if(PORTABLE)
   # MSVC does not need a separate compiler flag to enable SSE4.2; if nmmintrin.h
   # is available, it is available by default.
   if(FORCE_SSE42 AND NOT MSVC)
-    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -msse4.2")
+    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -msse4.2 -mpclmul")
   endif()
 else()
   if(MSVC)
     set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /arch:AVX2")
   else()
-    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=native")
+    if(NOT HAVE_POWER8)
+      set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=native")
+    endif()
   endif()
 endif()
 
-set(CMAKE_REQUIRED_FLAGS ${CMAKE_CXX_FLAGS})
 include(CheckCXXSourceCompiles)
+if(NOT MSVC)
+  set(CMAKE_REQUIRED_FLAGS "-msse4.2 -mpclmul")
+endif()
 CHECK_CXX_SOURCE_COMPILES("
 #include <cstdint>
 #include <nmmintrin.h>
+#include <wmmintrin.h>
 int main() {
   volatile uint32_t x = _mm_crc32_u32(0, 0);
+  const auto a = _mm_set_epi64x(0, 0);
+  const auto b = _mm_set_epi64x(0, 0);
+  const auto c = _mm_clmulepi64_si128(a, b, 0x00);
+  auto d = _mm_cvtsi128_si64(c);
 }
 " HAVE_SSE42)
+unset(CMAKE_REQUIRED_FLAGS)
 if(HAVE_SSE42)
   add_definitions(-DHAVE_SSE42)
 elseif(FORCE_SSE42)
@@ -158,32 +245,6 @@ if(HAVE_THREAD_LOCAL)
   add_definitions(-DROCKSDB_SUPPORT_THREAD_LOCAL)
 endif()
 
-set(BUILD_VERSION_CC ${CMAKE_BINARY_DIR}/build_version.cc)
-configure_file(util/build_version.cc.in ${BUILD_VERSION_CC} @ONLY)
-add_library(build_version OBJECT ${BUILD_VERSION_CC})
-target_include_directories(build_version PRIVATE
-  ${CMAKE_CURRENT_SOURCE_DIR}/util)
-if(MSVC)
-  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /Zi /nologo /EHsc /GS /Gd /GR /GF /fp:precise /Zc:wchar_t /Zc:forScope /errorReport:queue")
-  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /FC /d2Zi+ /W3 /wd4127 /wd4800 /wd4996 /wd4351")
-else()
-  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -W -Wextra -Wall")
-  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wsign-compare -Wshadow -Wno-unused-parameter -Wno-unused-variable -Woverloaded-virtual -Wnon-virtual-dtor -Wno-missing-field-initializers")
-  if(MINGW)
-    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wno-format")
-  endif()
-  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")
-  if(NOT CMAKE_BUILD_TYPE STREQUAL "Debug")
-    add_definitions(-DNDEBUG)
-    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -O2 -fno-omit-frame-pointer")
-    include(CheckCXXCompilerFlag)
-    CHECK_CXX_COMPILER_FLAG("-momit-leaf-frame-pointer" HAVE_OMIT_LEAF_FRAME_POINTER)
-    if(HAVE_OMIT_LEAF_FRAME_POINTER)
-      set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -momit-leaf-frame-pointer")
-    endif()
-  endif()
-endif()
-
 option(FAIL_ON_WARNINGS "Treat compile warnings as errors" ON)
 if(FAIL_ON_WARNINGS)
   if(MSVC)
@@ -224,6 +285,20 @@ if(WITH_UBSAN)
   endif()
 endif()
 
+find_package(NUMA)
+if(NUMA_FOUND)
+  add_definitions(-DNUMA)
+  include_directories(${NUMA_INCLUDE_DIR})
+  list(APPEND THIRDPARTY_LIBS ${NUMA_LIBRARIES})
+endif()
+
+find_package(TBB)
+if(TBB_FOUND)
+  add_definitions(-DTBB)
+  include_directories(${TBB_INCLUDE_DIR})
+  list(APPEND THIRDPARTY_LIBS ${TBB_LIBRARIES})
+endif()
+
 # Used to run CI build and tests so we can run faster
 set(OPTIMIZE_DEBUG_DEFAULT 0)        # Debug build is unoptimized by default use -DOPTDBG=1 to optimize
 
@@ -291,10 +366,8 @@ if(NOT WIN32)
 endif()
 
 option(WITH_FALLOCATE "build with fallocate" ON)
-
 if(WITH_FALLOCATE)
-  include(CheckCSourceCompiles)
-  CHECK_C_SOURCE_COMPILES("
+  CHECK_CXX_SOURCE_COMPILES("
 #include <fcntl.h>
 #include <linux/falloc.h>
 int main() {
@@ -307,12 +380,38 @@ int main() {
   endif()
 endif()
 
-include(CheckFunctionExists)
-CHECK_FUNCTION_EXISTS(malloc_usable_size HAVE_MALLOC_USABLE_SIZE)
+CHECK_CXX_SOURCE_COMPILES("
+#include <fcntl.h>
+int main() {
+  int fd = open(\"/dev/null\", 0);
+  sync_file_range(fd, 0, 1024, SYNC_FILE_RANGE_WRITE);
+}
+" HAVE_SYNC_FILE_RANGE_WRITE)
+if(HAVE_SYNC_FILE_RANGE_WRITE)
+  add_definitions(-DROCKSDB_RANGESYNC_PRESENT)
+endif()
+
+CHECK_CXX_SOURCE_COMPILES("
+#include <pthread.h>
+int main() {
+  (void) PTHREAD_MUTEX_ADAPTIVE_NP;
+}
+" HAVE_PTHREAD_MUTEX_ADAPTIVE_NP)
+if(HAVE_PTHREAD_MUTEX_ADAPTIVE_NP)
+  add_definitions(-DROCKSDB_PTHREAD_ADAPTIVE_MUTEX)
+endif()
+
+include(CheckCXXSymbolExists)
+check_cxx_symbol_exists(malloc_usable_size malloc.h HAVE_MALLOC_USABLE_SIZE)
 if(HAVE_MALLOC_USABLE_SIZE)
   add_definitions(-DROCKSDB_MALLOC_USABLE_SIZE)
 endif()
 
+check_cxx_symbol_exists(sched_getcpu sched.h HAVE_SCHED_GETCPU)
+if(HAVE_SCHED_GETCPU)
+  add_definitions(-DROCKSDB_SCHED_GETCPU_PRESENT)
+endif()
+
 include_directories(${PROJECT_SOURCE_DIR})
 include_directories(${PROJECT_SOURCE_DIR}/include)
 include_directories(SYSTEM ${PROJECT_SOURCE_DIR}/third-party/gtest-1.7.0/fused-src)
@@ -415,6 +514,7 @@ set(SOURCES
         table/block_based_table_factory.cc
         table/block_based_table_reader.cc
         table/block_builder.cc
+        table/block_fetcher.cc
         table/block_prefix_index.cc
         table/bloom_block.cc
         table/cuckoo_table_builder.cc
@@ -483,6 +583,9 @@ set(SOURCES
         utilities/blob_db/blob_log_writer.cc
         utilities/blob_db/blob_log_format.cc
         utilities/blob_db/ttl_extractor.cc
+        utilities/cassandra/cassandra_compaction_filter.cc
+        utilities/cassandra/format.cc
+        utilities/cassandra/merge_operator.cc
         utilities/checkpoint/checkpoint_impl.cc
         utilities/col_buf_decoder.cc
         utilities/col_buf_encoder.cc
@@ -501,8 +604,6 @@ set(SOURCES
         utilities/memory/memory_util.cc
         utilities/merge_operators/max.cc
         utilities/merge_operators/put.cc
-        utilities/merge_operators/cassandra/format.cc
-        utilities/merge_operators/cassandra/merge_operator.cc
         utilities/merge_operators/string_append/stringappend.cc
         utilities/merge_operators/string_append/stringappend2.cc
         utilities/merge_operators/uint64add.cc
@@ -518,18 +619,35 @@ set(SOURCES
         utilities/spatialdb/spatial_db.cc
         utilities/table_properties_collectors/compact_on_deletion_collector.cc
         utilities/transactions/optimistic_transaction_db_impl.cc
-        utilities/transactions/optimistic_transaction_impl.cc
+        utilities/transactions/optimistic_transaction.cc
+        utilities/transactions/pessimistic_transaction.cc
+        utilities/transactions/pessimistic_transaction_db.cc
+        utilities/transactions/snapshot_checker.cc
         utilities/transactions/transaction_base.cc
-        utilities/transactions/transaction_db_impl.cc
         utilities/transactions/transaction_db_mutex_impl.cc
-        utilities/transactions/transaction_impl.cc
         utilities/transactions/transaction_lock_mgr.cc
         utilities/transactions/transaction_util.cc
+        utilities/transactions/write_prepared_txn.cc
+        utilities/transactions/write_prepared_txn_db.cc
         utilities/ttl/db_ttl_impl.cc
         utilities/write_batch_with_index/write_batch_with_index.cc
         utilities/write_batch_with_index/write_batch_with_index_internal.cc
         $<TARGET_OBJECTS:build_version>)
 
+if(HAVE_SSE42 AND NOT FORCE_SSE42)
+  if(NOT MSVC)
+set_source_files_properties(
+  util/crc32c.cc
+      PROPERTIES COMPILE_FLAGS "-msse4.2 -mpclmul")
+  endif()
+endif()
+
+if(HAVE_POWER8)
+  list(APPEND SOURCES
+    util/crc32c_ppc.c
+    util/crc32c_ppc_asm.S)
+endif(HAVE_POWER8)
+
 if(WIN32)
   list(APPEND SOURCES
     port/win/io_win.cc
@@ -539,6 +657,12 @@ if(WIN32)
     port/win/win_logger.cc
     port/win/win_thread.cc
     port/win/xpress_win.cc)
+
+if(WITH_JEMALLOC)
+  list(APPEND SOURCES
+    port/win/win_jemalloc.cc)
+endif()
+
 else()
   list(APPEND SOURCES
     port/port_posix.cc
@@ -600,7 +724,72 @@ else()
   message(STATUS "JNI library is disabled")
 endif()
 
-set(TESTS
+# Installation and packaging
+if(WIN32)
+  option(ROCKSDB_INSTALL_ON_WINDOWS "Enable install target on Windows" OFF)
+endif()
+if(NOT WIN32 OR ROCKSDB_INSTALL_ON_WINDOWS)
+  if(CMAKE_INSTALL_PREFIX_INITIALIZED_TO_DEFAULT)
+    if(${CMAKE_SYSTEM_NAME} STREQUAL "Linux")
+      # Change default installation prefix on Linux to /usr
+      set(CMAKE_INSTALL_PREFIX /usr CACHE PATH "Install path prefix, prepended onto install directories." FORCE)
+    endif()
+  endif()
+
+  include(GNUInstallDirs)
+  include(CMakePackageConfigHelpers)
+
+  set(package_config_destination ${CMAKE_INSTALL_LIBDIR}/cmake/rocksdb)
+
+  configure_package_config_file(
+    ${CMAKE_CURRENT_LIST_DIR}/cmake/RocksDBConfig.cmake.in RocksDBConfig.cmake
+    INSTALL_DESTINATION ${package_config_destination}
+  )
+
+  write_basic_package_version_file(
+    RocksDBConfigVersion.cmake
+    VERSION ${ROCKSDB_VERSION}
+    COMPATIBILITY SameMajorVersion
+  )
+
+  install(DIRECTORY include/rocksdb COMPONENT devel DESTINATION "${CMAKE_INSTALL_INCLUDEDIR}")
+
+  install(
+    TARGETS ${ROCKSDB_STATIC_LIB}
+    EXPORT RocksDBTargets
+    COMPONENT devel
+    ARCHIVE DESTINATION "${CMAKE_INSTALL_LIBDIR}"
+    INCLUDES DESTINATION "${CMAKE_INSTALL_INCLUDEDIR}"
+  )
+
+  install(
+    TARGETS ${ROCKSDB_SHARED_LIB}
+    EXPORT RocksDBTargets
+    COMPONENT runtime
+    RUNTIME DESTINATION "${CMAKE_INSTALL_BINDIR}"
+    LIBRARY DESTINATION "${CMAKE_INSTALL_LIBDIR}"
+    INCLUDES DESTINATION "${CMAKE_INSTALL_INCLUDEDIR}"
+  )
+
+  install(
+    EXPORT RocksDBTargets
+    COMPONENT devel
+    DESTINATION ${package_config_destination}
+    NAMESPACE RocksDB::
+  )
+
+  install(
+    FILES
+    ${CMAKE_CURRENT_BINARY_DIR}/RocksDBConfig.cmake
+    ${CMAKE_CURRENT_BINARY_DIR}/RocksDBConfigVersion.cmake
+    COMPONENT devel
+    DESTINATION ${package_config_destination}
+  )
+endif()
+
+option(WITH_TESTS "build with tests" ON)
+if(WITH_TESTS)
+  set(TESTS
         cache/cache_test.cc
         cache/lru_cache_test.cc
         db/column_family_test.cc
@@ -613,6 +802,7 @@ set(TESTS
         db/corruption_test.cc
         db/cuckoo_table_db_test.cc
         db/db_basic_test.cc
+        db/db_blob_index_test.cc
         db/db_block_cache_test.cc
         db/db_bloom_filter_test.cc
         db/db_compaction_filter_test.cc
@@ -678,6 +868,7 @@ set(TESTS
         options/options_test.cc
         table/block_based_filter_block_test.cc
         table/block_test.cc
+        table/cleanable_test.cc
         table/cuckoo_table_builder_test.cc
         table/cuckoo_table_reader_test.cc
         table/full_filter_block_test.cc
@@ -706,6 +897,10 @@ set(TESTS
         util/thread_local_test.cc
         utilities/backupable/backupable_db_test.cc
         utilities/blob_db/blob_db_test.cc
+        utilities/cassandra/cassandra_functional_test.cc
+        utilities/cassandra/cassandra_format_test.cc
+        utilities/cassandra/cassandra_row_merge_test.cc
+        utilities/cassandra/cassandra_serialize_test.cc
         utilities/checkpoint/checkpoint_test.cc
         utilities/column_aware_encoding_test.cc
         utilities/date_tiered/date_tiered_test.cc
@@ -714,10 +909,6 @@ set(TESTS
         utilities/geodb/geodb_test.cc
         utilities/lua/rocks_lua_test.cc
         utilities/memory/memory_test.cc
-        utilities/merge_operators/cassandra/cassandra_merge_test.cc
-        utilities/merge_operators/cassandra/cassandra_format_test.cc
-        utilities/merge_operators/cassandra/cassandra_row_merge_test.cc
-        utilities/merge_operators/cassandra/cassandra_serialize_test.cc
         utilities/merge_operators/string_append/stringappend_test.cc
         utilities/object_registry_test.cc
         utilities/option_change_migration/option_change_migration_test.cc
@@ -726,98 +917,96 @@ set(TESTS
         utilities/persistent_cache/persistent_cache_test.cc
         utilities/redis/redis_lists_test.cc
         utilities/spatialdb/spatial_db_test.cc
+        utilities/simulator_cache/sim_cache_test.cc
         utilities/table_properties_collectors/compact_on_deletion_collector_test.cc
         utilities/transactions/optimistic_transaction_test.cc
         utilities/transactions/transaction_test.cc
+        utilities/transactions/write_prepared_transaction_test.cc
         utilities/ttl/ttl_test.cc
         utilities/write_batch_with_index/write_batch_with_index_test.cc
-)
-if(WITH_LIBRADOS)
-  list(APPEND TESTS utilities/env_librados_test.cc)
-endif()
+  )
+  if(WITH_LIBRADOS)
+    list(APPEND TESTS utilities/env_librados_test.cc)
+  endif()
 
-set(BENCHMARKS
-  cache/cache_bench.cc
-  memtable/memtablerep_bench.cc
-  tools/db_bench.cc
-  table/table_reader_bench.cc
-  utilities/column_aware_encoding_exp.cc
-  utilities/persistent_cache/hash_table_bench.cc)
-add_library(testharness OBJECT 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
-    util/fault_injection_test_env.cc
-    utilities/merge_operators/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 ${TESTUTIL_SOURCE})
-if(MSVC)
-  set_target_properties(${TESTUTILLIB} PROPERTIES COMPILE_FLAGS "/Fd${CMAKE_CFG_INTDIR}/testutillib${ARTIFACT_SUFFIX}.pdb")
-endif()
-set_target_properties(${TESTUTILLIB}
-      PROPERTIES EXCLUDE_FROM_DEFAULT_BUILD_RELEASE 1
-      EXCLUDE_FROM_DEFAULT_BUILD_MINRELEASE 1
-      EXCLUDE_FROM_DEFAULT_BUILD_RELWITHDEBINFO 1
-      )
-
-# Tests are excluded from Release builds
-set(TEST_EXES ${TESTS})
-
-foreach(sourcefile ${TEST_EXES})
+  set(BENCHMARKS
+    cache/cache_bench.cc
+    memtable/memtablerep_bench.cc
+    tools/db_bench.cc
+    table/table_reader_bench.cc
+    utilities/column_aware_encoding_exp.cc
+    utilities/persistent_cache/hash_table_bench.cc)
+  add_library(testharness OBJECT util/testharness.cc)
+  foreach(sourcefile ${BENCHMARKS})
     get_filename_component(exename ${sourcefile} NAME_WE)
     add_executable(${exename}${ARTIFACT_SUFFIX} ${sourcefile}
       $<TARGET_OBJECTS:testharness>)
-    set_target_properties(${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}${ARTIFACT_SUFFIX} testutillib${ARTIFACT_SUFFIX} gtest ${LIBS})
-    if(NOT "${exename}" MATCHES "db_sanity_test")
+    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
+      util/fault_injection_test_env.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 ${TESTUTIL_SOURCE})
+  if(MSVC)
+    set_target_properties(${TESTUTILLIB} PROPERTIES COMPILE_FLAGS "/Fd${CMAKE_CFG_INTDIR}/testutillib${ARTIFACT_SUFFIX}.pdb")
+  endif()
+  set_target_properties(${TESTUTILLIB}
+        PROPERTIES EXCLUDE_FROM_DEFAULT_BUILD_RELEASE 1
+        EXCLUDE_FROM_DEFAULT_BUILD_MINRELEASE 1
+        EXCLUDE_FROM_DEFAULT_BUILD_RELWITHDEBINFO 1
+        )
+
+  # Tests are excluded from Release builds
+  set(TEST_EXES ${TESTS})
+
+  foreach(sourcefile ${TEST_EXES})
+      get_filename_component(exename ${sourcefile} NAME_WE)
+      add_executable(${CMAKE_PROJECT_NAME}_${exename}${ARTIFACT_SUFFIX} ${sourcefile}
+        $<TARGET_OBJECTS:testharness>)
+      set_target_properties(${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 ${LIBS})
+      if(NOT "${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})
+
+  # C executables must link to a shared object
+  set(C_TESTS db/c_test.c)
+  set(C_TEST_EXES ${C_TESTS})
+
+  foreach(sourcefile ${C_TEST_EXES})
+      string(REPLACE ".c" "" exename ${sourcefile})
+      string(REGEX REPLACE "^((.+)/)+" "" exename ${exename})
+      add_executable(${exename}${ARTIFACT_SUFFIX} ${sourcefile})
+      set_target_properties(${exename}${ARTIFACT_SUFFIX}
+        PROPERTIES EXCLUDE_FROM_DEFAULT_BUILD_RELEASE 1
+        EXCLUDE_FROM_DEFAULT_BUILD_MINRELEASE 1
+        EXCLUDE_FROM_DEFAULT_BUILD_RELWITHDEBINFO 1
+        )
+      target_link_libraries(${exename}${ARTIFACT_SUFFIX} ${ROCKSDB_IMPORT_LIB} testutillib${ARTIFACT_SUFFIX})
       add_test(NAME ${exename} COMMAND ${exename}${ARTIFACT_SUFFIX})
       add_dependencies(check ${exename}${ARTIFACT_SUFFIX})
-    endif()
-endforeach(sourcefile ${TEST_EXES})
-
-# C executables must link to a shared object
-set(C_TESTS db/c_test.c)
-set(C_TEST_EXES ${C_TESTS})
-
-foreach(sourcefile ${C_TEST_EXES})
-    string(REPLACE ".c" "" exename ${sourcefile})
-    string(REGEX REPLACE "^((.+)/)+" "" exename ${exename})
-    add_executable(${exename}${ARTIFACT_SUFFIX} ${sourcefile})
-    set_target_properties(${exename}${ARTIFACT_SUFFIX}
-      PROPERTIES EXCLUDE_FROM_DEFAULT_BUILD_RELEASE 1
-      EXCLUDE_FROM_DEFAULT_BUILD_MINRELEASE 1
-      EXCLUDE_FROM_DEFAULT_BUILD_RELWITHDEBINFO 1
-      )
-    target_link_libraries(${exename}${ARTIFACT_SUFFIX} ${ROCKSDB_IMPORT_LIB} testutillib${ARTIFACT_SUFFIX})
-    add_test(NAME ${exename} COMMAND ${exename}${ARTIFACT_SUFFIX})
-    add_dependencies(check ${exename}${ARTIFACT_SUFFIX})
-endforeach(sourcefile ${C_TEST_EXES})
-add_subdirectory(tools)
-
-# Installation and packaging for Linux
-if(NOT WIN32)
-install(TARGETS ${ROCKSDB_STATIC_LIB} COMPONENT devel ARCHIVE DESTINATION lib64)
-install(TARGETS ${ROCKSDB_SHARED_LIB} COMPONENT runtime DESTINATION lib64)
-install(DIRECTORY "${PROJECT_SOURCE_DIR}/include/rocksdb/"
-        COMPONENT devel
-        DESTINATION include/rocksdb)
-set(CMAKE_INSTALL_PREFIX /usr)
+  endforeach(sourcefile ${C_TEST_EXES})
+endif()
+
+option(WITH_TOOLS "build with tools" ON)
+if(WITH_TOOLS)
+  add_subdirectory(tools)
 endif()

CODE_OF_CONDUCT.md

@@ -0,0 +1,3 @@
+# Code of Conduct
+
+Facebook has adopted a Code of Conduct that we expect project participants to adhere to. Please [read the full text](https://code.facebook.com/codeofconduct) so that you can understand what actions will and will not be tolerated.

CONTRIBUTING.md

@@ -1,5 +1,8 @@
 # Contributing to RocksDB
 
+## Code of Conduct
+The code of conduct is described in [`CODE_OF_CONDUCT.md`](CODE_OF_CONDUCT.md)
+
 ## Contributor License Agreement ("CLA")
 
 In order to accept your pull request, we need you to submit a CLA. You

HISTORY.md

@@ -1,25 +1,72 @@
 # Rocksdb Change Log
-## 5.7.5 (11/28/2017)
+## Unreleased
+### Public API Change
+* Iterator::SeekForPrev is now a pure virtual method. This is to prevent user who implement the Iterator interface fail to implement SeekForPrev by mistake.
+### Bug Fixes
+* Fix `DisableFileDeletions()` followed by `GetSortedWalFiles()` to not return obsolete WAL files that `PurgeObsoleteFiles()` is going to delete.
+* Fix DB::Flush() keep waiting after flush finish under certain condition.
+
+## 5.10.0 (12/11/2017)
+### Public API Change
+* When running `make` with environment variable `USE_SSE` set and `PORTABLE` unset, will use all machine features available locally. Previously this combination only compiled SSE-related features.
+
+### New Features
+* Provide lifetime hints when writing files on Linux. This reduces hardware write-amp on storage devices supporting multiple streams.
+* Add a DB stat, `NUMBER_ITER_SKIP`, which returns how many internal keys were skipped during iterations (e.g., due to being tombstones or duplicate versions of a key).
+* Add PerfContext counters, `key_lock_wait_count` and `key_lock_wait_time`, which measure the number of times transactions wait on key locks and total amount of time waiting.
+
 ### Bug Fixes
 * Fix IOError on WAL write doesn't propagate to write group follower
+* Make iterator invalid on merge error.
+* Fix performance issue in `IngestExternalFile()` affecting databases with large number of SST files.
+* Fix possible corruption to LSM structure when `DeleteFilesInRange()` deletes a subset of files spanned by a `DeleteRange()` marker.
 
-## 5.7.4 (08/31/2017)
-No significant changes.
+## 5.9.0 (11/1/2017)
+### Public API Change
+* `BackupableDBOptions::max_valid_backups_to_open == 0` now means no backups will be opened during BackupEngine initialization. Previously this condition disabled limiting backups opened.
+* `DBOptions::preserve_deletes` is a new option that allows one to specify that DB should not drop tombstones for regular deletes if they have sequence number larger than what was set by the new API call `DB::SetPreserveDeletesSequenceNumber(SequenceNumber seqnum)`. Disabled by default.
+* API call `DB::SetPreserveDeletesSequenceNumber(SequenceNumber seqnum)` was added, users who wish to preserve deletes are expected to periodically call this function to advance the cutoff seqnum (all deletes made before this seqnum can be dropped by DB). It's user responsibility to figure out how to advance the seqnum in the way so the tombstones are kept for the desired period of time, yet are eventually processed in time and don't eat up too much space.
+* `ReadOptions::iter_start_seqnum` was added; if set to something > 0 user will see 2 changes in iterators behavior 1) only keys written with sequence larger than this parameter would be returned and 2) the `Slice` returned by iter->key() now points to the the memory that keep User-oriented representation of the internal key, rather than user key. New struct `FullKey` was added to represent internal keys, along with a new helper function `ParseFullKey(const Slice& internal_key, FullKey* result);`.
+* Deprecate trash_dir param in NewSstFileManager, right now we will rename deleted files to <name>.trash instead of moving them to trash directory
+* Allow setting a custom trash/DB size ratio limit in the SstFileManager, after which files that are to be scheduled for deletion are deleted immediately, regardless of any delete ratelimit.
+* Return an error on write if write_options.sync = true and write_options.disableWAL = true to warn user of inconsistent options. Previously we will not write to WAL and not respecting the sync options in this case.
+
+### New Features
+* CRC32C is now using the 3-way pipelined SSE algorithm `crc32c_3way` on supported platforms to improve performance. The system will choose to use this algorithm on supported platforms automatically whenever possible. If PCLMULQDQ is not supported it will fall back to the old Fast_CRC32 algorithm.
+* `DBOptions::writable_file_max_buffer_size` can now be changed dynamically.
+* `DBOptions::bytes_per_sync`, `DBOptions::compaction_readahead_size`, and `DBOptions::wal_bytes_per_sync` can now be changed dynamically, `DBOptions::wal_bytes_per_sync` will flush all memtables and switch to a new WAL file.
+* Support dynamic adjustment of rate limit according to demand for background I/O. It can be enabled by passing `true` to the `auto_tuned` parameter in `NewGenericRateLimiter()`. The value passed as `rate_bytes_per_sec` will still be respected as an upper-bound.
+* Support dynamically changing `ColumnFamilyOptions::compaction_options_fifo`.
+* Introduce `EventListener::OnStallConditionsChanged()` callback. Users can implement it to be notified when user writes are stalled, stopped, or resumed.
+* Add a new db property "rocksdb.estimate-oldest-key-time" to return oldest data timestamp. The property is available only for FIFO compaction with compaction_options_fifo.allow_compaction = false.
+* Upon snapshot release, recompact bottommost files containing deleted/overwritten keys that previously could not be dropped due to the snapshot. This alleviates space-amp caused by long-held snapshots.
+* Support lower bound on iterators specified via `ReadOptions::iterate_lower_bound`.
+* Support for differential snapshots (via iterator emitting the sequence of key-values representing the difference between DB state at two different sequence numbers). Supports preserving and emitting puts and regular deletes, doesn't support SingleDeletes, MergeOperator, Blobs and Range Deletes.
 
-## 5.7.3 (08/29/2017)
 ### Bug Fixes
+* Fix a potential data inconsistency issue during point-in-time recovery. `DB:Open()` will abort if column family inconsistency is found during PIT recovery.
+* Fix possible metadata corruption in databases using `DeleteRange()`.
+
+## 5.8.0 (08/30/2017)
+### Public API Change
+* Users of `Statistics::getHistogramString()` will see fewer histogram buckets and different bucket endpoints.
+* `Slice::compare` and BytewiseComparator `Compare` no longer accept `Slice`s containing nullptr.
+* `Transaction::Get` and `Transaction::GetForUpdate` variants with `PinnableSlice` added.
+
+### New Features
+* Add Iterator::Refresh(), which allows users to update the iterator state so that they can avoid some initialization costs of recreating iterators.
+* Replace dynamic_cast<> (except unit test) so people can choose to build with RTTI off. With make, release mode is by default built with -fno-rtti and debug mode is built without it. Users can override it by setting USE_RTTI=0 or 1.
+* Universal compactions including the bottom level can be executed in a dedicated thread pool. This alleviates head-of-line blocking in the compaction queue, which cause write stalling, particularly in multi-instance use cases. Users can enable this feature via `Env::SetBackgroundThreads(N, Env::Priority::BOTTOM)`, where `N > 0`.
+* Allow merge operator to be called even with a single merge operand during compactions, by appropriately overriding `MergeOperator::AllowSingleOperand`.
+* Add `DB::VerifyChecksum()`, which verifies the checksums in all SST files in a running DB.
+* Block-based table support for disabling checksums by setting `BlockBasedTableOptions::checksum = kNoChecksum`.
+
+### Bug Fixes
+* Fix wrong latencies in `rocksdb.db.get.micros`, `rocksdb.db.write.micros`, and `rocksdb.sst.read.micros`.
+* Fix incorrect dropping of deletions during intra-L0 compaction.
 * Fix transient reappearance of keys covered by range deletions when memtable prefix bloom filter is enabled.
 * Fix potentially wrong file smallest key when range deletions separated by snapshot are written together.
 
-## 5.7.2 (08/15/2017)
-### Bug Fixes
-* Fix incorrect dropping of deletions issue with FIFO compaction.
-* Fix LITE build compiler error with missing abort().
-
-## 5.7.1 (08/13/2017)
-### Bug Fixes
-* Fix incorrect dropping of deletions during intra-L0 compaction.
-
 ## 5.7.0 (07/13/2017)
 ### Public API Change
 * DB property "rocksdb.sstables" now prints keys in hex form.

INSTALL.md

@@ -107,12 +107,49 @@ to build a portable binary, add `PORTABLE=1` before your make commands, like thi
             * run `brew tap homebrew/versions; brew install gcc48 --use-llvm` to install gcc 4.8 (or higher).
     * run `brew install rocksdb`
 
+* **FreeBSD** (11.01):
+
+    * You can either install RocksDB from the Ports system using `cd /usr/ports/databases/rocksdb && make install`, or you can follow the details below to install dependencies and compile from source code:
+
+    * Install the dependencies for RocksDB:
+
+        export BATCH=YES
+        cd /usr/ports/devel/gmake && make install
+        cd /usr/ports/devel/gflags && make install
+
+        cd /usr/ports/archivers/snappy && make install
+        cd /usr/ports/archivers/bzip2 && make install
+        cd /usr/ports/archivers/liblz4 && make install
+        cd /usr/ports/archivesrs/zstd && make install
+
+        cd /usr/ports/devel/git && make install
+
+
+    * Install the dependencies for RocksJava (optional):
+
+        export BATCH=yes
+        cd /usr/ports/java/openjdk7 && make install
+
+    * Build RocksDB from source:
+        cd ~
+        git clone https://github.com/facebook/rocksdb.git
+        cd rocksdb
+        gmake static_lib
+
+    * Build RocksJava from source (optional):
+        cd rocksdb
+        export JAVA_HOME=/usr/local/openjdk7
+        gmake rocksdbjava
+
+
 * **iOS**:
   * Run: `TARGET_OS=IOS make static_lib`. When building the project which uses rocksdb iOS library, make sure to define two important pre-processing macros: `ROCKSDB_LITE` and `IOS_CROSS_COMPILE`.
 
 * **Windows**:
   * For building with MS Visual Studio 13 you will need Update 4 installed.
   * Read and follow the instructions at CMakeLists.txt
+  * Or install via [vcpkg](https://github.com/microsoft/vcpkg) 
+       * run `vcpkg install rocksdb:x64-windows`
 
 * **AIX 6.1**
     * Install AIX Toolbox rpms with gcc

LANGUAGE-BINDINGS.md

@@ -9,5 +9,9 @@ This is the list of all known third-party language bindings for RocksDB. If some
 * Haskell - https://hackage.haskell.org/package/rocksdb-haskell
 * PHP - https://github.com/Photonios/rocksdb-php
 * C# - https://github.com/warrenfalk/rocksdb-sharp
-* Rust - https://github.com/spacejam/rust-rocksdb
+* Rust
+    * https://github.com/spacejam/rust-rocksdb
+    * https://github.com/bh1xuw/rust-rocks
 * D programming language - https://github.com/b1naryth1ef/rocksdb
+* Erlang - https://gitlab.com/barrel-db/erlang-rocksdb
+* Elixir - https://github.com/urbint/rox

Makefile

@@ -96,12 +96,34 @@ OPT += -momit-leaf-frame-pointer
 endif
 endif
 
-# if we're compiling for release, compile without debug code (-DNDEBUG) and
-# don't treat warnings as errors
+ifeq (,$(shell $(CXX) -fsyntax-only -maltivec -xc /dev/null 2>&1))
+CXXFLAGS += -DHAS_ALTIVEC
+CFLAGS += -DHAS_ALTIVEC
+HAS_ALTIVEC=1
+endif
+
+ifeq (,$(shell $(CXX) -fsyntax-only -mcpu=power8 -xc /dev/null 2>&1))
+CXXFLAGS += -DHAVE_POWER8
+CFLAGS +=  -DHAVE_POWER8
+HAVE_POWER8=1
+endif
+
+# if we're compiling for release, compile without debug code (-DNDEBUG)
 ifeq ($(DEBUG_LEVEL),0)
 OPT += -DNDEBUG
-DISABLE_WARNING_AS_ERROR=1
+
+ifneq ($(USE_RTTI), 1)
+	CXXFLAGS += -fno-rtti
 else
+	CXXFLAGS += -DROCKSDB_USE_RTTI
+endif
+else
+ifneq ($(USE_RTTI), 0)
+	CXXFLAGS += -DROCKSDB_USE_RTTI
+else
+	CXXFLAGS += -fno-rtti
+endif
+
 $(warning Warning: Compiling in debug mode. Don't use the resulting binary in production)
 endif
 
@@ -215,11 +237,20 @@ ifdef COMPILE_WITH_UBSAN
 	PLATFORM_CXXFLAGS += -fsanitize=undefined -DROCKSDB_UBSAN_RUN
 endif
 
+ifdef ROCKSDB_VALGRIND_RUN
+	PLATFORM_CCFLAGS += -DROCKSDB_VALGRIND_RUN
+	PLATFORM_CXXFLAGS += -DROCKSDB_VALGRIND_RUN
+endif
+
 ifndef DISABLE_JEMALLOC
 	ifdef JEMALLOC
 		PLATFORM_CXXFLAGS += -DROCKSDB_JEMALLOC -DJEMALLOC_NO_DEMANGLE
 		PLATFORM_CCFLAGS  += -DROCKSDB_JEMALLOC -DJEMALLOC_NO_DEMANGLE
 	endif
+	ifdef WITH_JEMALLOC_FLAG
+		PLATFORM_LDFLAGS += -ljemalloc
+		JAVA_LDFLAGS += -ljemalloc
+	endif
 	EXEC_LDFLAGS := $(JEMALLOC_LIB) $(EXEC_LDFLAGS)
 	PLATFORM_CXXFLAGS += $(JEMALLOC_INCLUDE)
 	PLATFORM_CCFLAGS += $(JEMALLOC_INCLUDE)
@@ -274,6 +305,9 @@ LDFLAGS += $(LUA_LIB)
 
 endif
 
+ifeq ($(NO_THREEWAY_CRC32C), 1)
+	CXXFLAGS += -DNO_THREEWAY_CRC32C
+endif
 
 CFLAGS += $(WARNING_FLAGS) -I. -I./include $(PLATFORM_CCFLAGS) $(OPT)
 CXXFLAGS += $(WARNING_FLAGS) -I. -I./include $(PLATFORM_CXXFLAGS) $(OPT) -Woverloaded-virtual -Wnon-virtual-dtor -Wno-missing-field-initializers
@@ -306,6 +340,14 @@ util/build_version.cc: FORCE
 endif
 
 LIBOBJECTS = $(LIB_SOURCES:.cc=.o)
+ifeq ($(HAVE_POWER8),1)
+LIB_CC_OBJECTS = $(LIB_SOURCES:.cc=.o)
+LIBOBJECTS += $(LIB_SOURCES_C:.c=.o)
+LIBOBJECTS += $(LIB_SOURCES_ASM:.S=.o)
+else
+LIB_CC_OBJECTS = $(LIB_SOURCES:.cc=.o)
+endif
+
 LIBOBJECTS += $(TOOL_LIB_SOURCES:.cc=.o)
 MOCKOBJECTS = $(MOCK_LIB_SOURCES:.cc=.o)
 
@@ -344,6 +386,7 @@ TESTS = \
 	db_wal_test \
 	db_block_cache_test \
 	db_test \
+	db_blob_index_test \
 	db_bloom_filter_test \
 	db_iter_test \
 	db_log_iter_test \
@@ -405,7 +448,7 @@ TESTS = \
 	write_buffer_manager_test \
 	stringappend_test \
 	cassandra_format_test \
-	cassandra_merge_test \
+	cassandra_functional_test \
 	cassandra_row_merge_test \
 	cassandra_serialize_test \
 	ttl_test \
@@ -460,11 +503,13 @@ TESTS = \
 	object_registry_test \
 	repair_test \
 	env_timed_test \
+	write_prepared_transaction_test \
 
 PARALLEL_TEST = \
 	backupable_db_test \
 	db_compaction_filter_test \
 	db_compaction_test \
+	db_merge_operator_test \
 	db_sst_test \
 	db_test \
 	db_universal_compaction_test \
@@ -475,7 +520,8 @@ PARALLEL_TEST = \
 	manual_compaction_test \
 	persistent_cache_test \
 	table_test \
-	transaction_test
+	transaction_test \
+	write_prepared_transaction_test \
 
 SUBSET := $(TESTS)
 ifdef ROCKSDBTESTS_START
@@ -555,14 +601,35 @@ $(SHARED2): $(SHARED4)
 $(SHARED3): $(SHARED4)
 	ln -fs $(SHARED4) $(SHARED3)
 endif
-
+ifeq ($(HAVE_POWER8),1)
+SHARED_C_OBJECTS = $(LIB_SOURCES_C:.c=.o)
+SHARED_ASM_OBJECTS = $(LIB_SOURCES_ASM:.S=.o)
+SHARED_C_LIBOBJECTS = $(patsubst %.o,shared-objects/%.o,$(SHARED_C_OBJECTS))
+SHARED_ASM_LIBOBJECTS = $(patsubst %.o,shared-objects/%.o,$(SHARED_ASM_OBJECTS))
+shared_libobjects = $(patsubst %,shared-objects/%,$(LIB_CC_OBJECTS))
+else
 shared_libobjects = $(patsubst %,shared-objects/%,$(LIBOBJECTS))
-CLEAN_FILES += shared-objects
+endif
 
+CLEAN_FILES += shared-objects
+shared_all_libobjects = $(shared_libobjects)
+
+ifeq ($(HAVE_POWER8),1)
+shared-ppc-objects = $(SHARED_C_LIBOBJECTS) $(SHARED_ASM_LIBOBJECTS)
+
+shared-objects/util/crc32c_ppc.o: util/crc32c_ppc.c
+	$(AM_V_CC)$(CC) $(CFLAGS) -c $< -o $@
+
+shared-objects/util/crc32c_ppc_asm.o: util/crc32c_ppc_asm.S
+	$(AM_V_CC)$(CC) $(CFLAGS) -c $< -o $@
+endif
 $(shared_libobjects): shared-objects/%.o: %.cc
 	$(AM_V_CC)mkdir -p $(@D) && $(CXX) $(CXXFLAGS) $(PLATFORM_SHARED_CFLAGS) -c $< -o $@
 
-$(SHARED4): $(shared_libobjects)
+ifeq ($(HAVE_POWER8),1)
+shared_all_libobjects = $(shared_libobjects) $(shared-ppc-objects)
+endif
+$(SHARED4): $(shared_all_libobjects)
 	$(CXX) $(PLATFORM_SHARED_LDFLAGS)$(SHARED3) $(CXXFLAGS) $(PLATFORM_SHARED_CFLAGS) $(shared_libobjects) $(LDFLAGS) -o $@
 
 endif  # PLATFORM_SHARED_EXT
@@ -575,6 +642,8 @@ endif  # PLATFORM_SHARED_EXT
 
 all: $(LIBRARY) $(BENCHMARKS) tools tools_lib test_libs $(TESTS)
 
+all_but_some_tests: $(LIBRARY) $(BENCHMARKS) tools tools_lib test_libs $(SUBSET)
+
 static_lib: $(LIBRARY)
 
 shared_lib: $(SHARED)
@@ -682,7 +751,7 @@ gen_parallel_tests:
 # 107.816 PASS t/DBTest.EncodeDecompressedBlockSizeTest
 #
 slow_test_regexp = \
-	^t/run-table_test-HarnessTest.Randomized$$|^t/run-db_test-.*(?:FileCreationRandomFailure|EncodeDecompressedBlockSizeTest)$$|^.*RecoverFromCorruptedWALWithoutFlush$$
+	^.*SnapshotConcurrentAccessTest.*$$|^t/run-table_test-HarnessTest.Randomized$$|^t/run-db_test-.*(?:FileCreationRandomFailure|EncodeDecompressedBlockSizeTest)$$|^.*RecoverFromCorruptedWALWithoutFlush$$
 prioritize_long_running_tests =						\
   perl -pe 's,($(slow_test_regexp)),100 $$1,'				\
     | sort -k1,1gr							\
@@ -778,8 +847,8 @@ ldb_tests: ldb
 crash_test: whitebox_crash_test blackbox_crash_test
 
 blackbox_crash_test: db_stress
-	python -u tools/db_crashtest.py --simple blackbox
-	python -u tools/db_crashtest.py blackbox
+	python -u tools/db_crashtest.py --simple blackbox $(CRASH_TEST_EXT_ARGS)
+	python -u tools/db_crashtest.py blackbox $(CRASH_TEST_EXT_ARGS)
 
 ifeq ($(CRASH_TEST_KILL_ODD),)
   CRASH_TEST_KILL_ODD=888887
@@ -787,9 +856,9 @@ endif
 
 whitebox_crash_test: db_stress
 	python -u tools/db_crashtest.py --simple whitebox --random_kill_odd \
-      $(CRASH_TEST_KILL_ODD)
+      $(CRASH_TEST_KILL_ODD) $(CRASH_TEST_EXT_ARGS)
 	python -u tools/db_crashtest.py whitebox  --random_kill_odd \
-      $(CRASH_TEST_KILL_ODD)
+      $(CRASH_TEST_KILL_ODD) $(CRASH_TEST_EXT_ARGS)
 
 asan_check:
 	$(MAKE) clean
@@ -812,7 +881,7 @@ ubsan_crash_test:
 	$(MAKE) clean
 
 valgrind_test:
-	DISABLE_JEMALLOC=1 $(MAKE) valgrind_check
+	ROCKSDB_VALGRIND_RUN=1 DISABLE_JEMALLOC=1 $(MAKE) valgrind_check
 
 valgrind_check: $(TESTS)
 	$(MAKE) DRIVER="$(VALGRIND_VER) $(VALGRIND_OPTS)" gen_parallel_tests
@@ -1000,16 +1069,16 @@ option_change_migration_test: utilities/option_change_migration/option_change_mi
 stringappend_test: utilities/merge_operators/string_append/stringappend_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
-cassandra_format_test: utilities/merge_operators/cassandra/cassandra_format_test.o $(LIBOBJECTS) $(TESTHARNESS)
+cassandra_format_test: utilities/cassandra/cassandra_format_test.o utilities/cassandra/test_utils.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
-cassandra_merge_test: utilities/merge_operators/cassandra/cassandra_merge_test.o utilities/merge_operators/cassandra/test_utils.o $(LIBOBJECTS) $(TESTHARNESS)
+cassandra_functional_test: utilities/cassandra/cassandra_functional_test.o utilities/cassandra/test_utils.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
-cassandra_row_merge_test: utilities/merge_operators/cassandra/cassandra_row_merge_test.o utilities/merge_operators/cassandra/test_utils.o $(LIBOBJECTS) $(TESTHARNESS)
+cassandra_row_merge_test: utilities/cassandra/cassandra_row_merge_test.o utilities/cassandra/test_utils.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
-cassandra_serialize_test: utilities/merge_operators/cassandra/cassandra_serialize_test.o $(LIBOBJECTS) $(TESTHARNESS)
+cassandra_serialize_test: utilities/cassandra/cassandra_serialize_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
 redis_test: utilities/redis/redis_lists_test.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1045,6 +1114,9 @@ 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)
+	$(AM_LINK)
+
 db_block_cache_test: db/db_block_cache_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
@@ -1209,7 +1281,7 @@ env_test: env/env_test.o $(LIBOBJECTS) $(TESTHARNESS)
 fault_injection_test: db/fault_injection_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
-rate_limiter_test: util/rate_limiter_test.o $(LIBOBJECTS) $(TESTHARNESS)
+rate_limiter_test: util/rate_limiter_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
 delete_scheduler_test: util/delete_scheduler_test.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1374,6 +1446,9 @@ heap_test: util/heap_test.o $(GTEST)
 transaction_test: utilities/transactions/transaction_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)
 
+write_prepared_transaction_test: utilities/transactions/write_prepared_transaction_test.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 sst_dump: tools/sst_dump.o $(LIBOBJECTS)
 	$(AM_LINK)
 
@@ -1481,11 +1556,11 @@ BZIP2_DOWNLOAD_BASE ?= http://www.bzip.org
 SNAPPY_VER ?= 1.1.4
 SNAPPY_SHA256 ?= 134bfe122fd25599bb807bb8130e7ba6d9bdb851e0b16efcb83ac4f5d0b70057
 SNAPPY_DOWNLOAD_BASE ?= https://github.com/google/snappy/releases/download
-LZ4_VER ?= 1.7.5
-LZ4_SHA256 ?= 0190cacd63022ccb86f44fa5041dc6c3804407ad61550ca21c382827319e7e7e
+LZ4_VER ?= 1.8.0
+LZ4_SHA256 ?= 2ca482ea7a9bb103603108b5a7510b7592b90158c151ff50a28f1ca8389fccf6
 LZ4_DOWNLOAD_BASE ?= https://github.com/lz4/lz4/archive
-ZSTD_VER ?= 1.2.0
-ZSTD_SHA256 ?= 4a7e4593a3638276ca7f2a09dc4f38e674d8317bbea51626393ca73fc047cbfb
+ZSTD_VER ?= 1.3.3
+ZSTD_SHA256 ?= a77c47153ee7de02626c5b2a097005786b71688be61e9fb81806a011f90b297b
 ZSTD_DOWNLOAD_BASE ?= https://github.com/facebook/zstd/archive
 
 ifeq ($(PLATFORM), OS_MACOSX)
@@ -1499,7 +1574,7 @@ else
 endif
 endif
 ifeq ($(PLATFORM), OS_FREEBSD)
-	JAVA_INCLUDE += -I$(JAVA_HOME)/include/freebsd
+	JAVA_INCLUDE = -I$(JAVA_HOME)/include -I$(JAVA_HOME)/include/freebsd
 	ROCKSDBJNILIB = librocksdbjni-freebsd$(ARCH).so
 	ROCKSDB_JAR = rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-freebsd$(ARCH).jar
 endif
@@ -1525,7 +1600,7 @@ libz.a:
 		exit 1; \
 	fi
 	tar xvzf zlib-$(ZLIB_VER).tar.gz
-	cd zlib-$(ZLIB_VER) && CFLAGS='-fPIC ${EXTRA_CFLAGS}' LDFLAGS='${EXTRA_LDFLAGS}' ./configure --static && make
+	cd zlib-$(ZLIB_VER) && CFLAGS='-fPIC ${EXTRA_CFLAGS}' LDFLAGS='${EXTRA_LDFLAGS}' ./configure --static && $(MAKE)
 	cp zlib-$(ZLIB_VER)/libz.a .
 
 libbz2.a:
@@ -1537,7 +1612,7 @@ libbz2.a:
 		exit 1; \
 	fi
 	tar xvzf bzip2-$(BZIP2_VER).tar.gz
-	cd bzip2-$(BZIP2_VER) && make CFLAGS='-fPIC -O2 -g -D_FILE_OFFSET_BITS=64 ${EXTRA_CFLAGS}' AR='ar ${EXTRA_ARFLAGS}'
+	cd bzip2-$(BZIP2_VER) && $(MAKE) CFLAGS='-fPIC -O2 -g -D_FILE_OFFSET_BITS=64 ${EXTRA_CFLAGS}' AR='ar ${EXTRA_ARFLAGS}'
 	cp bzip2-$(BZIP2_VER)/libbz2.a .
 
 libsnappy.a:
@@ -1550,7 +1625,7 @@ libsnappy.a:
 	fi
 	tar xvzf snappy-$(SNAPPY_VER).tar.gz
 	cd snappy-$(SNAPPY_VER) && CFLAGS='${EXTRA_CFLAGS}' CXXFLAGS='${EXTRA_CXXFLAGS}' LDFLAGS='${EXTRA_LDFLAGS}' ./configure --with-pic --enable-static --disable-shared
-	cd snappy-$(SNAPPY_VER) && make ${SNAPPY_MAKE_TARGET}
+	cd snappy-$(SNAPPY_VER) && $(MAKE) ${SNAPPY_MAKE_TARGET}
 	cp snappy-$(SNAPPY_VER)/.libs/libsnappy.a .
 
 liblz4.a:
@@ -1563,7 +1638,7 @@ liblz4.a:
 		exit 1; \
 	fi
 	tar xvzf lz4-$(LZ4_VER).tar.gz
-	cd lz4-$(LZ4_VER)/lib && make CFLAGS='-fPIC -O2 ${EXTRA_CFLAGS}' all
+	cd lz4-$(LZ4_VER)/lib && $(MAKE) CFLAGS='-fPIC -O2 ${EXTRA_CFLAGS}' all
 	cp lz4-$(LZ4_VER)/lib/liblz4.a .
 
 libzstd.a:
@@ -1576,29 +1651,45 @@ libzstd.a:
 		exit 1; \
 	fi
 	tar xvzf zstd-$(ZSTD_VER).tar.gz
-	cd zstd-$(ZSTD_VER)/lib && make CFLAGS='-fPIC -O2 ${EXTRA_CFLAGS}' all
+	cd zstd-$(ZSTD_VER)/lib && DESTDIR=. PREFIX= $(MAKE) CFLAGS='-fPIC -O2 ${EXTRA_CFLAGS}' install
 	cp zstd-$(ZSTD_VER)/lib/libzstd.a .
 
 # A version of each $(LIBOBJECTS) compiled with -fPIC and a fixed set of static compression libraries
-java_static_libobjects = $(patsubst %,jls/%,$(LIBOBJECTS))
+java_static_libobjects = $(patsubst %,jls/%,$(LIB_CC_OBJECTS))
 CLEAN_FILES += jls
+java_static_all_libobjects = $(java_static_libobjects)
 
 ifneq ($(ROCKSDB_JAVA_NO_COMPRESSION), 1)
 JAVA_COMPRESSIONS = libz.a libbz2.a libsnappy.a liblz4.a libzstd.a
 endif
 
 JAVA_STATIC_FLAGS = -DZLIB -DBZIP2 -DSNAPPY -DLZ4 -DZSTD
-JAVA_STATIC_INCLUDES = -I./zlib-$(ZLIB_VER) -I./bzip2-$(BZIP2_VER) -I./snappy-$(SNAPPY_VER) -I./lz4-$(LZ4_VER)/lib -I./zstd-$(ZSTD_VER)/lib
+JAVA_STATIC_INCLUDES = -I./zlib-$(ZLIB_VER) -I./bzip2-$(BZIP2_VER) -I./snappy-$(SNAPPY_VER) -I./lz4-$(LZ4_VER)/lib -I./zstd-$(ZSTD_VER)/lib/include
+
+ifeq ($(HAVE_POWER8),1)
+JAVA_STATIC_C_LIBOBJECTS = $(patsubst %.c.o,jls/%.c.o,$(LIB_SOURCES_C:.c=.o))
+JAVA_STATIC_ASM_LIBOBJECTS = $(patsubst %.S.o,jls/%.S.o,$(LIB_SOURCES_ASM:.S=.o))
+
+java_static_ppc_libobjects = $(JAVA_STATIC_C_LIBOBJECTS) $(JAVA_STATIC_ASM_LIBOBJECTS)
+
+jls/util/crc32c_ppc.o: util/crc32c_ppc.c
+	$(AM_V_CC)$(CC) $(CFLAGS) $(JAVA_STATIC_FLAGS) $(JAVA_STATIC_INCLUDES) -c $< -o $@
+
+jls/util/crc32c_ppc_asm.o: util/crc32c_ppc_asm.S
+	$(AM_V_CC)$(CC) $(CFLAGS) $(JAVA_STATIC_FLAGS) $(JAVA_STATIC_INCLUDES) -c $< -o $@
+
+java_static_all_libobjects += $(java_static_ppc_libobjects)
+endif
 
 $(java_static_libobjects): jls/%.o: %.cc $(JAVA_COMPRESSIONS)
 	$(AM_V_CC)mkdir -p $(@D) && $(CXX) $(CXXFLAGS) $(JAVA_STATIC_FLAGS) $(JAVA_STATIC_INCLUDES) -fPIC -c $< -o $@ $(COVERAGEFLAGS)
 
-rocksdbjavastatic: $(java_static_libobjects)
+rocksdbjavastatic: $(java_static_all_libobjects)
 	cd java;$(MAKE) javalib;
 	rm -f ./java/target/$(ROCKSDBJNILIB)
 	$(CXX) $(CXXFLAGS) -I./java/. $(JAVA_INCLUDE) -shared -fPIC \
 	  -o ./java/target/$(ROCKSDBJNILIB) $(JNI_NATIVE_SOURCES) \
-	  $(java_static_libobjects) $(COVERAGEFLAGS) \
+	  $(java_static_all_libobjects) $(COVERAGEFLAGS) \
 	  $(JAVA_COMPRESSIONS) $(JAVA_STATIC_LDFLAGS)
 	cd java/target;strip $(STRIPFLAGS) $(ROCKSDBJNILIB)
 	cd java;jar -cf target/$(ROCKSDB_JAR) HISTORY*.md
@@ -1628,6 +1719,14 @@ rocksdbjavastaticreleasedocker: rocksdbjavastatic
 	cd java/target;jar -uf $(ROCKSDB_JAR_ALL) librocksdbjni-*.so librocksdbjni-*.jnilib
 	cd java/target/classes;jar -uf ../$(ROCKSDB_JAR_ALL) org/rocksdb/*.class org/rocksdb/util/*.class
 
+rocksdbjavastaticdockerppc64le:
+	mkdir -p java/target
+	DOCKER_LINUX_PPC64LE_CONTAINER=`docker ps -aqf name=rocksdb_linux_ppc64le-be`; \
+	if [ -z "$$DOCKER_LINUX_PPC64LE_CONTAINER" ]; then \
+		docker container create --attach stdin --attach stdout --attach stderr --volume `pwd`:/rocksdb-host --name rocksdb_linux_ppc64le-be evolvedbinary/rocksjava:centos7_ppc64le-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh; \
+	fi
+	docker start -a rocksdb_linux_ppc64le-be
+
 rocksdbjavastaticpublish: rocksdbjavastaticrelease rocksdbjavastaticpublishcentral
 
 rocksdbjavastaticpublishdocker: rocksdbjavastaticreleasedocker rocksdbjavastaticpublishcentral
@@ -1642,13 +1741,36 @@ rocksdbjavastaticpublishcentral:
 	mvn gpg:sign-and-deploy-file -Durl=https://oss.sonatype.org/service/local/staging/deploy/maven2/ -DrepositoryId=sonatype-nexus-staging -DpomFile=java/rocksjni.pom -Dfile=java/target/rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH).jar
 
 # A version of each $(LIBOBJECTS) compiled with -fPIC
-java_libobjects = $(patsubst %,jl/%,$(LIBOBJECTS))
+ifeq ($(HAVE_POWER8),1)
+JAVA_CC_OBJECTS = $(SHARED_CC_OBJECTS)
+JAVA_C_OBJECTS = $(SHARED_C_OBJECTS)
+JAVA_ASM_OBJECTS = $(SHARED_ASM_OBJECTS)
+
+JAVA_C_LIBOBJECTS = $(patsubst %.c.o,jl/%.c.o,$(JAVA_C_OBJECTS))
+JAVA_ASM_LIBOBJECTS = $(patsubst %.S.o,jl/%.S.o,$(JAVA_ASM_OBJECTS))
+endif
+
+java_libobjects = $(patsubst %,jl/%,$(LIB_CC_OBJECTS))
 CLEAN_FILES += jl
+java_all_libobjects = $(java_libobjects)
+
+ifeq ($(HAVE_POWER8),1)
+java_ppc_libobjects = $(JAVA_C_LIBOBJECTS) $(JAVA_ASM_LIBOBJECTS)
+
+jl/crc32c_ppc.o: util/crc32c_ppc.c
+	$(AM_V_CC)$(CC) $(CFLAGS) -c $< -o $@
+
+jl/crc32c_ppc_asm.o: util/crc32c_ppc_asm.S
+	$(AM_V_CC)$(CC) $(CFLAGS) -c $< -o $@
+java_all_libobjects += $(java_ppc_libobjects)
+endif
 
 $(java_libobjects): jl/%.o: %.cc
 	$(AM_V_CC)mkdir -p $(@D) && $(CXX) $(CXXFLAGS) -fPIC -c $< -o $@ $(COVERAGEFLAGS)
 
-rocksdbjava: $(java_libobjects)
+
+
+rocksdbjava: $(java_all_libobjects)
 	$(AM_V_GEN)cd java;$(MAKE) javalib;
 	$(AM_V_at)rm -f ./java/target/$(ROCKSDBJNILIB)
 	$(AM_V_at)$(CXX) $(CXXFLAGS) -I./java/. $(JAVA_INCLUDE) -shared -fPIC -o ./java/target/$(ROCKSDBJNILIB) $(JNI_NATIVE_SOURCES) $(java_libobjects) $(JAVA_LDFLAGS) $(COVERAGEFLAGS)
@@ -1703,30 +1825,54 @@ IOSVERSION=$(shell defaults read $(PLATFORMSROOT)/iPhoneOS.platform/version CFBu
 	lipo ios-x86/$@ ios-arm/$@ -create -output $@
 
 else
+ifeq ($(HAVE_POWER8),1)
+util/crc32c_ppc.o: util/crc32c_ppc.c
+	$(AM_V_CC)$(CC) $(CFLAGS) -c $< -o $@
+
+util/crc32c_ppc_asm.o: util/crc32c_ppc_asm.S
+	$(AM_V_CC)$(CC) $(CFLAGS) -c $< -o $@
+endif
 .cc.o:
 	$(AM_V_CC)$(CXX) $(CXXFLAGS) -c $< -o $@ $(COVERAGEFLAGS)
 
 .c.o:
 	$(AM_V_CC)$(CC) $(CFLAGS) -c $< -o $@
 endif
-
 # ---------------------------------------------------------------------------
 #  	Source files dependencies detection
 # ---------------------------------------------------------------------------
 
 all_sources = $(LIB_SOURCES) $(MAIN_SOURCES) $(MOCK_LIB_SOURCES) $(TOOL_LIB_SOURCES) $(BENCH_LIB_SOURCES) $(TEST_LIB_SOURCES) $(EXP_LIB_SOURCES)
-DEPFILES = $(all_sources:.cc=.d)
+DEPFILES = $(all_sources:.cc=.cc.d)
 
 # Add proper dependency support so changing a .h file forces a .cc file to
 # rebuild.
 
 # The .d file indicates .cc file's dependencies on .h files. We generate such
 # dependency by g++'s -MM option, whose output is a make dependency rule.
-$(DEPFILES): %.d: %.cc
+%.cc.d: %.cc
 	@$(CXX) $(CXXFLAGS) $(PLATFORM_SHARED_CFLAGS) \
 	  -MM -MT'$@' -MT'$(<:.cc=.o)' "$<" -o '$@'
 
+ifeq ($(HAVE_POWER8),1)
+DEPFILES_C = $(LIB_SOURCES_C:.c=.c.d)
+DEPFILES_ASM = $(LIB_SOURCES_ASM:.S=.S.d)
+
+%.c.d: %.c
+	@$(CXX) $(CXXFLAGS) $(PLATFORM_SHARED_CFLAGS) \
+	  -MM -MT'$@' -MT'$(<:.c=.o)' "$<" -o '$@'
+
+%.S.d: %.S
+	@$(CXX) $(CXXFLAGS) $(PLATFORM_SHARED_CFLAGS) \
+	  -MM -MT'$@' -MT'$(<:.S=.o)' "$<" -o '$@'
+
+$(DEPFILES_C): %.c.d
+
+$(DEPFILES_ASM): %.S.d
+depend: $(DEPFILES) $(DEPFILES_C) $(DEPFILES_ASM)
+else
 depend: $(DEPFILES)
+endif
 
 # if the make goal is either "clean" or "format", we shouldn't
 # try to import the *.d files.

README.md

@@ -1,7 +1,8 @@
 ## RocksDB: A Persistent Key-Value Store for Flash and RAM Storage
 
-[![Build Status](https://travis-ci.org/facebook/rocksdb.svg?branch=master)](https://travis-ci.org/facebook/rocksdb)
-[![Build status](https://ci.appveyor.com/api/projects/status/fbgfu0so3afcno78/branch/master?svg=true)](https://ci.appveyor.com/project/Facebook/rocksdb/branch/master)
+[![Linux/Mac Build Status](https://travis-ci.org/facebook/rocksdb.svg?branch=master)](https://travis-ci.org/facebook/rocksdb)
+[![Windows Build status](https://ci.appveyor.com/api/projects/status/fbgfu0so3afcno78/branch/master?svg=true)](https://ci.appveyor.com/project/Facebook/rocksdb/branch/master)
+[![PPC64le Build Status](http://140.211.168.68:8080/buildStatus/icon?job=Rocksdb)](http://140.211.168.68:8080/job/Rocksdb)
 
 
 RocksDB is developed and maintained by Facebook Database Engineering Team.

USERS.md

@@ -5,12 +5,14 @@ At Facebook, we use RocksDB as storage engines in multiple data management servi
 
 1. MyRocks -- https://github.com/MySQLOnRocksDB/mysql-5.6
 2. MongoRocks -- https://github.com/mongodb-partners/mongo-rocks
-3. ZippyDB --  Facebook's distributed key-value store with Paxos-style replication, built on top of RocksDB.[*] https://www.youtube.com/watch?v=DfiN7pG0D0khtt
-4. Laser -- Laser is a high query throughput, low (millisecond) latency, key-value storage service built on top of RocksDB.[*]
+3. ZippyDB --  Facebook's distributed key-value store with Paxos-style replication, built on top of RocksDB.[1] https://www.youtube.com/watch?v=DfiN7pG0D0khtt
+4. Laser -- Laser is a high query throughput, low (millisecond) latency, key-value storage service built on top of RocksDB.[1]
 4. Dragon -- a distributed graph query engine. https://code.facebook.com/posts/1737605303120405/dragon-a-distributed-graph-query-engine/
-5. Stylus -- a low-level stream processing framework writtenin C++.[*]
+5. Stylus -- a low-level stream processing framework writtenin C++.[1]
+6. LogDevice -- a distributed data store for logs [2]
 
-[*] https://research.facebook.com/publications/realtime-data-processing-at-facebook/
+[1] https://research.facebook.com/publications/realtime-data-processing-at-facebook/
+[2] https://code.facebook.com/posts/357056558062811/logdevice-a-distributed-data-store-for-logs/
 
 ## LinkedIn
 Two different use cases at Linkedin are using RocksDB as a storage engine:
@@ -80,3 +82,9 @@ quasardb uses a heavily tuned RocksDB as its persistence layer.
 
 ## 360 Pika
 [360](http://www.360.cn/) [Pika](https://github.com/Qihoo360/pika) is a nosql compatible with redis. With the huge amount of data stored, redis may suffer for a capacity bottleneck, and pika was born for solving it. It has widely been widely used in many company
+
+## LzLabs
+LzLabs is using RocksDB as a storage engine in their multi-database distributed framework to store application configuration and user data.
+
+## ProfaneDB
+[ProfaneDB](https://profanedb.gitlab.io/) is a database for Protocol Buffers, and uses RocksDB for storage. It is accessible via gRPC, and the schema is defined using directly `.proto` files.

appveyor.yml

@@ -3,12 +3,12 @@ image: Visual Studio 2015
 before_build:
 - md %APPVEYOR_BUILD_FOLDER%\build
 - cd %APPVEYOR_BUILD_FOLDER%\build
-- cmake -G "Visual Studio 14 2015 Win64" -DOPTDBG=1 -DXPRESS=1 ..
+- cmake -G "Visual Studio 14 Win64" -DOPTDBG=1 -DWITH_XPRESS=1 -DPORTABLE=1 ..
 - cd ..
 build:
   project: build\rocksdb.sln
   parallel: true
-  verbosity: minimal
+  verbosity: normal
 test:
 test_script:
 - ps: build_tools\run_ci_db_test.ps1 -SuiteRun db_basic_test,db_test2,db_test,env_basic_test,env_test -Concurrency 8

arcanist_util/__phutil_library_init__.php

@@ -1,3 +0,0 @@
-<?php
-
-phutil_register_library('arcanist_util', __FILE__);

arcanist_util/__phutil_library_map__.php

@@ -1,71 +0,0 @@
-<?php
-
-/**
- * This file is automatically generated. Use 'arc liberate' to rebuild it.
- * @generated
- * @phutil-library-version 2
- */
-
-if (class_exists('ArcanistWorkflow')) {
-  phutil_register_library_map(array(
-    '__library_version__' => 2,
-    'class' =>
-    array(
-      'ArcanistCpplintLinter' => 'cpp_linter/ArcanistCpplintLinter.php',
-      'BaseDirectoryScopedFormatLinter' => 'cpp_linter/BaseDirectoryScopedFormatLinter.php',
-      'FacebookArcanistConfiguration' => 'config/FacebookArcanistConfiguration.php',
-      'FacebookFbcodeLintEngine' => 'lint_engine/FacebookFbcodeLintEngine.php',
-      'FacebookFbcodeUnitTestEngine' => 'unit_engine/FacebookFbcodeUnitTestEngine.php',
-      'FacebookHowtoevenLintEngine' => 'lint_engine/FacebookHowtoevenLintEngine.php',
-      'FacebookHowtoevenLinter' => 'cpp_linter/FacebookHowtoevenLinter.php',
-      'FbcodeClangFormatLinter' => 'cpp_linter/FbcodeClangFormatLinter.php',
-      'FbcodeCppLinter' => 'cpp_linter/FbcodeCppLinter.php',
-    ),
-    'function' =>
-    array(
-    ),
-    'xmap' =>
-    array(
-      'ArcanistCpplintLinter' => 'ArcanistLinter',
-      'BaseDirectoryScopedFormatLinter' => 'ArcanistLinter',
-      'FacebookArcanistConfiguration' => 'ArcanistConfiguration',
-      'FacebookFbcodeLintEngine' => 'ArcanistLintEngine',
-      'FacebookFbcodeUnitTestEngine' => 'ArcanistBaseUnitTestEngine',
-      'FacebookHowtoevenLintEngine' => 'ArcanistLintEngine',
-      'FacebookHowtoevenLinter' => 'ArcanistLinter',
-      'FbcodeClangFormatLinter' => 'BaseDirectoryScopedFormatLinter',
-      'FbcodeCppLinter' => 'ArcanistLinter',
-    ),
-  ));
-} else {
-  phutil_register_library_map(array(
-    '__library_version__' => 2,
-    'class' =>
-    array(
-      'ArcanistCpplintLinter' => 'cpp_linter/ArcanistCpplintLinter.php',
-      'BaseDirectoryScopedFormatLinter' => 'cpp_linter/BaseDirectoryScopedFormatLinter.php',
-      'FacebookArcanistConfiguration' => 'config/FacebookOldArcanistConfiguration.php',
-      'FacebookFbcodeLintEngine' => 'lint_engine/FacebookFbcodeLintEngine.php',
-      'FacebookFbcodeUnitTestEngine' => 'unit_engine/FacebookOldFbcodeUnitTestEngine.php',
-      'FacebookHowtoevenLintEngine' => 'lint_engine/FacebookHowtoevenLintEngine.php',
-      'FacebookHowtoevenLinter' => 'cpp_linter/FacebookHowtoevenLinter.php',
-      'FbcodeClangFormatLinter' => 'cpp_linter/FbcodeClangFormatLinter.php',
-      'FbcodeCppLinter' => 'cpp_linter/FbcodeCppLinter.php',
-    ),
-    'function' =>
-    array(
-    ),
-    'xmap' =>
-    array(
-      'ArcanistCpplintLinter' => 'ArcanistLinter',
-      'BaseDirectoryScopedFormatLinter' => 'ArcanistLinter',
-      'FacebookArcanistConfiguration' => 'ArcanistConfiguration',
-      'FacebookFbcodeLintEngine' => 'ArcanistLintEngine',
-      'FacebookFbcodeUnitTestEngine' => 'ArcanistBaseUnitTestEngine',
-      'FacebookHowtoevenLintEngine' => 'ArcanistLintEngine',
-      'FacebookHowtoevenLinter' => 'ArcanistLinter',
-      'FbcodeClangFormatLinter' => 'BaseDirectoryScopedFormatLinter',
-      'FbcodeCppLinter' => 'ArcanistLinter',
-    ),
-  ));
-}

arcanist_util/config/FacebookArcanistConfiguration.php

@@ -1,43 +0,0 @@
-<?php
-// Copyright 2004-present Facebook. All Rights Reserved.
-// This source code is licensed under the BSD-style license found in the
-// LICENSE file in the root directory of this source tree. An additional grant
-// of patent rights can be found in the PATENTS file in the same directory.
-
-require('RocksDBCommonHelper.php');
-
-define("DIFF_COMMAND", "diff");
-
-class FacebookArcanistConfiguration extends ArcanistConfiguration {
-  public function getCustomArgumentsForCommand($command) {
-    if ($command == "land") {
-      return array(
-          'async' => array('help' => 'Just to make tools happy'));
-    }
-    return array();
-  }
-
-  public function didRunWorkflow($command,
-                                 ArcanistWorkflow $workflow,
-                                 $error_code) {
-    // Default options don't terminate on failure, but that's what we want. In
-    // the current case we use assertions intentionally as "terminate on failure
-    // invariants".
-    assert_options(ASSERT_BAIL, true);
-
-    assert($workflow);
-    assert(strlen($command) > 0);
-
-    if ($command == DIFF_COMMAND && !$workflow->isRawDiffSource()) {
-      $diffID = $workflow->getDiffId();
-
-      // When submitting a diff this code path gets executed multiple times in
-      // a row. We only care about the case when ID for the diff is provided
-      // because that's what we need to apply the diff and trigger the tests.
-      if (strlen($diffID) > 0) {
-        assert(is_numeric($diffID));
-        startTestsInSandcastle(true /* $applyDiff */, $workflow, $diffID);
-      }
-    }
-  }
-}

arcanist_util/config/FacebookOldArcanistConfiguration.php

@@ -1,43 +0,0 @@
-<?php
-// Copyright 2004-present Facebook. All Rights Reserved.
-// This source code is licensed under the BSD-style license found in the
-// LICENSE file in the root directory of this source tree. An additional grant
-// of patent rights can be found in the PATENTS file in the same directory.
-
-require('RocksDBCommonHelper.php');
-
-define("DIFF_COMMAND", "diff");
-
-class FacebookArcanistConfiguration extends ArcanistConfiguration {
-  public function getCustomArgumentsForCommand($command) {
-    if ($command == "land") {
-      return array(
-          'async' => array('help' => 'Just to make tools happy'));
-    }
-    return array();
-  }
-
-  public function didRunWorkflow($command,
-                                 ArcanistBaseWorkflow $workflow,
-                                 $error_code) {
-    // Default options don't terminate on failure, but that's what we want. In
-    // the current case we use assertions intentionally as "terminate on failure
-    // invariants".
-    assert_options(ASSERT_BAIL, true);
-
-    assert($workflow);
-    assert(strlen($command) > 0);
-
-    if ($command == DIFF_COMMAND && !$workflow->isRawDiffSource()) {
-      $diffID = $workflow->getDiffId();
-
-      // When submitting a diff this code path gets executed multiple times in
-      // a row. We only care about the case when ID for the diff is provided
-      // because that's what we need to apply the diff and trigger the tests.
-      if (strlen($diffID) > 0) {
-        assert(is_numeric($diffID));
-        startTestsInSandcastle(true /* $applyDiff */, $workflow, $diffID);
-      }
-    }
-  }
-}

arcanist_util/cpp_linter/ArcanistCpplintLinter.php

@@ -1,88 +0,0 @@
-<?php
-
-/**
- * Uses google's cpplint.py to check code. RocksDB team forked this file from
- * phabricator's /src/lint/linter/ArcanistCpplintLinter.php, and customized it
- * for its own use.
- *
- * You can get it here:
- * http://google-styleguide.googlecode.com/svn/trunk/cpplint/cpplint.py
- * @group linter
- */
-final class ArcanistCpplintLinter extends ArcanistLinter {
-
-  public function willLintPaths(array $paths) {
-    return;
-  }
-
-  public function getLinterName() {
-    return 'cpplint.py';
-  }
-
-  public function getLintPath() {
-    $bin = 'cpplint.py';
-    // Search under current dir
-    list($err) = exec_manual('which %s/%s', $this->linterDir(), $bin);
-    if (!$err) {
-      return $this->linterDir().'/'.$bin;
-    }
-
-    // Look for globally installed cpplint.py
-    list($err) = exec_manual('which %s', $bin);
-    if ($err) {
-      throw new ArcanistUsageException(
-        "cpplint.py does not appear to be installed on this system. Install ".
-        "it (e.g., with 'wget \"http://google-styleguide.googlecode.com/".
-        "svn/trunk/cpplint/cpplint.py\"') ".
-        "in your .arcconfig to point to the directory where it resides. ".
-        "Also don't forget to chmod a+x cpplint.py!");
-    }
-
-    return $bin;
-  }
-
-  public function lintPath($path) {
-    $bin = $this->getLintPath();
-    $path = $this->rocksdbDir().'/'.$path;
-
-    $f = new ExecFuture("%C $path", $bin);
-
-    list($err, $stdout, $stderr) = $f->resolve();
-
-    if ($err === 2) {
-      throw new Exception("cpplint failed to run correctly:\n".$stderr);
-    }
-
-    $lines = explode("\n", $stderr);
-    $messages = array();
-    foreach ($lines as $line) {
-      $line = trim($line);
-      $matches = null;
-      $regex = '/^[^:]+:(\d+):\s*(.*)\s*\[(.*)\] \[(\d+)\]$/';
-      if (!preg_match($regex, $line, $matches)) {
-        continue;
-      }
-      foreach ($matches as $key => $match) {
-        $matches[$key] = trim($match);
-      }
-      $message = new ArcanistLintMessage();
-      $message->setPath($path);
-      $message->setLine($matches[1]);
-      $message->setCode($matches[3]);
-      $message->setName($matches[3]);
-      $message->setDescription($matches[2]);
-      $message->setSeverity(ArcanistLintSeverity::SEVERITY_WARNING);
-      $this->addLintMessage($message);
-    }
-  }
-
-  // The path of this linter
-  private function linterDir() {
-    return dirname(__FILE__);
-  }
-
-  // TODO(kaili) a quick and dirty way to figure out rocksdb's root dir.
-  private function rocksdbDir() {
-    return $this->linterDir()."/../..";
-  }
-}

arcanist_util/cpp_linter/BaseDirectoryScopedFormatLinter.php

@@ -1,74 +0,0 @@
-<?php
-// Copyright 2004-present Facebook. All Rights Reserved.
-// This source code is licensed under the BSD-style license found in the
-// LICENSE file in the root directory of this source tree. An additional grant
-// of patent rights can be found in the PATENTS file in the same directory.
-
-abstract class BaseDirectoryScopedFormatLinter extends ArcanistLinter {
-
-  const LINT_FORMATTING = 1;
-
-  private $changedLines = array();
-  private $rawLintOutput = array();
-
-  abstract protected function getPathsToLint();
-
-  protected function shouldLintPath($path) {
-    foreach ($this->getPathsToLint() as $p) {
-      // check if $path starts with $p
-      if (strncmp($path, $p, strlen($p)) === 0) {
-        return true;
-      }
-    }
-    return false;
-  }
-
-  // API to tell this linter which lines were changed
-  final public function setPathChangedLines($path, $changed) {
-    $this->changedLines[$path] = $changed;
-  }
-
-  final public function willLintPaths(array $paths) {
-    $futures = array();
-    foreach ($paths as $path) {
-      if (!$this->shouldLintPath($path)) {
-        continue;
-      }
-
-      $changed = $this->changedLines[$path];
-      if (!isset($changed)) {
-        // do not run linter if there are no changes
-        continue;
-      }
-
-      $futures[$path] = $this->getFormatFuture($path, $changed);
-    }
-
-    foreach (id(new FutureIterator($futures))->limit(8) as $p => $f) {
-      $this->rawLintOutput[$p] = $f->resolvex();
-    }
-  }
-
-  abstract protected function getFormatFuture($path, array $changed);
-  abstract protected function getLintMessage($diff);
-
-  final public function lintPath($path) {
-    if (!isset($this->rawLintOutput[$path])) {
-      return;
-    }
-
-    list($new_content) = $this->rawLintOutput[$path];
-    $old_content = $this->getData($path);
-
-    if ($new_content != $old_content) {
-      $diff = ArcanistDiffUtils::renderDifferences($old_content, $new_content);
-      $this->raiseLintAtOffset(
-        0,
-        self::LINT_FORMATTING,
-        $this->getLintMessage($diff),
-        $old_content,
-        $new_content);
-    }
-  }
-
-}

arcanist_util/cpp_linter/FacebookHowtoevenLinter.php

@@ -1,223 +0,0 @@
-<?php
-// Copyright 2015-present Facebook. All Rights Reserved.
-// This source code is licensed under the BSD-style license found in the
-// LICENSE file in the root directory of this source tree. An additional grant
-// of patent rights can be found in the PATENTS file in the same directory.
-
-final class FacebookHowtoevenLinter extends ArcanistLinter {
-
-  const VERSION = 'fd9192f324c36d28136d14380f0b552a1385b59b';
-
-  private $parsedTargets = array();
-
-  public function getLinterName() {
-    return 'Howtoeven';
-  }
-
-  protected function getSeverity($code) {
-    $severities = array(
-      ArcanistLintSeverity::SEVERITY_DISABLED,
-      ArcanistLintSeverity::SEVERITY_ADVICE,
-      ArcanistLintSeverity::SEVERITY_WARNING,
-      ArcanistLintSeverity::SEVERITY_ERROR,
-    );
-    return idx($severities, $code, ArcanistLintSeverity::SEVERITY_WARNING);
-  }
-
-  public function willLintPaths(array $paths) {
-    // Cleanup previous runs.
-    $this->localExecx("rm -rf _build/_lint");
-
-    // Build compilation database.
-    $lintable_paths = $this->getLintablePaths($paths);
-    $interesting_paths = $this->getInterestingPaths($lintable_paths);
-
-    if (!$lintable_paths) {
-      return;
-    }
-
-    // Run lint.
-    try {
-      $this->localExecx(
-        "%C %C -p _build/dev/ %Ls",
-        $this->getBinaryPath(),
-        $this->getFilteredIssues(),
-        $lintable_paths);
-    } catch (CommandException $exception) {
-      PhutilConsole::getConsole()->writeErr($exception->getMessage());
-    }
-
-    // Load results.
-    $result = id(
-      new SQLite3(
-        $this->getProjectRoot().'/_build/_lint/lint.db',
-        SQLITE3_OPEN_READONLY))
-      ->query("SELECT * FROM raised_issues");
-
-    while ($issue = $result->fetchArray(SQLITE3_ASSOC)) {
-      // Skip issues not part of the linted file.
-      if (in_array($issue['file'], $interesting_paths)) {
-        $this->addLintMessage(id(new ArcanistLintMessage())
-          ->setPath($issue['file'])
-          ->setLine($issue['line'])
-          ->setChar($issue['column'])
-          ->setCode('Howtoeven')
-          ->setSeverity($this->getSeverity($issue['severity']))
-          ->setName('Hte-'.$issue['name'])
-          ->setDescription(
-            sprintf(
-              "%s\n\n%s",
-              ($issue['message']) ? $issue['message'] : $issue['description'],
-              $issue['explanation']))
-          ->setOriginalText(idx($issue, 'original', ''))
-          ->setReplacementText(idx($issue, 'replacement', '')));
-      }
-    }
-  }
-
-  public function lintPath($path) {
-  }
-
-  /**
-   * Get the paths that we know how to lint.
-   *
-   * The strategy is to first look whether there's an existing compilation
-   * database and use that if it's exhaustive. We generate our own only if
-   * necessary.
-   */
-  private function getLintablePaths($paths) {
-    // Replace headers with existing sources.
-    for ($i = 0; $i < count($paths); $i++) {
-      if (preg_match("/\.h$/", $paths[$i])) {
-        $header = preg_replace("/\.h$/", ".cpp", $paths[$i]);
-        if (file_exists($header)) {
-          $paths[$i] = $header;
-        }
-      }
-    }
-
-    // Check if database exists and is exhaustive.
-    $available_paths = $this->getAvailablePaths();
-    $lintable_paths = array_intersect($paths, $available_paths);
-    if ($paths === $lintable_paths) {
-      return $lintable_paths;
-    }
-
-    // Generate our own database.
-    $targets = $this->getTargetsFor($paths);
-    if (!$targets) {
-      PhutilConsole::getConsole()->writeErr(
-        "No build targets found for %s\n",
-        implode(', ', $paths));
-      return array();
-    }
-
-    $this->localExecx("./tools/build/bin/fbconfig.par -r %Ls", $targets);
-    $this->localExecx("./tools/build/bin/fbmake.par gen_cdb");
-
-    $available_paths = $this->getAvailablePaths();
-    $lintable_paths = array_intersect($paths, $available_paths);
-    if ($paths != $lintable_paths) {
-      PhutilConsole::getConsole()->writeErr(
-        "Can't lint %s\n",
-        implode(', ', array_diff($paths, $available_paths)));
-    }
-
-    // Return what we know how to lint.
-    return $lintable_paths;
-  }
-
-  /**
-   * Get the available paths in the current compilation database.
-   */
-  private function getAvailablePaths() {
-    $database_path = $this->getProjectRoot()
-      .'/_build/dev/compile_commands.json';
-    if (!file_exists($database_path)) {
-      return array();
-    }
-
-    $entries = json_decode(file_get_contents($database_path), true);
-    $paths = array();
-    foreach ($entries as $entry) {
-      $paths[] = $entry['file'];
-    }
-    return $paths;
-  }
-
-  /**
-   * Search for the targets directories for the given files.
-   */
-  private static function getTargetsFor($paths) {
-    $targets = array();
-    foreach ($paths as $path) {
-      while (($path = dirname($path)) !== '.') {
-        if (in_array('TARGETS', scandir($path))) {
-          $contents = file_get_contents($path.'/TARGETS');
-          if (strpos($contents, 'cpp_binary') !== false) {
-            $targets[] = $path;
-            break;
-          }
-        }
-      }
-    }
-    return array_unique($targets);
-  }
-
-  /**
-   * The paths that we actually want to report on.
-   */
-  private function getInterestingPaths($paths) {
-    $headers = array();
-    foreach ($paths as $path) {
-      $headers[] = preg_replace("/\.cpp$/", ".h", $path);
-    }
-    return array_merge($paths, $headers);
-  }
-
-  /**
-   * The path where the binary is located. Will return the current dewey binary
-   * unless the `HOWTOEVEN_BUILD` environment variable is set.
-   */
-  private function getBinaryPath() {
-    $path = sprintf(
-      "/mnt/dewey/fbcode/.commits/%s/builds/howtoeven/client",
-      self::VERSION);
-
-    $build = getenv('HOWTOEVEN_BUILD');
-    if ($build) {
-      $path = sprintf(
-        "./_build/%s/tools/howtoeven/client",
-        $build);
-      if (!file_exists($path)) {
-        PhutilConsole::getConsole()->writeErr(">> %s does not exist\n", $path);
-        exit(1);
-      }
-    }
-
-    return $path;
-  }
-
-  /**
-   * Execute the command in the root directory.
-   */
-  private function localExecx($command /* , ... */) {
-    $arguments = func_get_args();
-    return newv('ExecFuture', $arguments)
-      ->setCWD($this->getProjectRoot())
-      ->resolvex();
-  }
-
-  /**
-   * The root of the project.
-   */
-  private function getProjectRoot() {
-    return $this->getEngine()->getWorkingCopy()->getProjectRoot();
-  }
-
-  private function getFilteredIssues() {
-    $issues = getenv('HOWTOEVEN_ISSUES');
-    return ($issues) ? csprintf('-issues %s', $issues) : '';
-  }
-
-}

arcanist_util/cpp_linter/FbcodeClangFormatLinter.php

@@ -1,58 +0,0 @@
-<?php
-// Copyright 2004-present Facebook. All Rights Reserved.
-// This source code is licensed under the BSD-style license found in the
-// LICENSE file in the root directory of this source tree. An additional grant
-// of patent rights can be found in the PATENTS file in the same directory.
-
-final class FbcodeClangFormatLinter extends BaseDirectoryScopedFormatLinter {
-
-  const LINT_FORMATTING = 1;
-  const CLANG_FORMAT_BINARY = '/mnt/vol/engshare/admin/scripts/clang-format';
-
-  protected function getPathsToLint() {
-    return array('');
-  }
-
-  public function getLinterName() {
-    return 'CLANG_FORMAT';
-  }
-
-  public function getLintSeverityMap() {
-    return array(
-      self::LINT_FORMATTING => ArcanistLintSeverity::SEVERITY_ADVICE,
-    );
-  }
-
-  public function getLintNameMap() {
-    return array(
-      self::LINT_FORMATTING => pht('Changes are not clang-formatted'),
-    );
-  }
-
-  protected function getFormatFuture($path, array $changed) {
-    $args = "";
-    foreach ($changed as $key => $value) {
-      $args .= " --lines=$key:$key";
-    }
-
-    $binary = self::CLANG_FORMAT_BINARY;
-    if (!file_exists($binary)) {
-      // trust the $PATH
-      $binary = "clang-format";
-    }
-
-    return new ExecFuture(
-      "%s %s $args",
-      $binary,
-      $this->getEngine()->getFilePathOnDisk($path));
-  }
-
-  protected function getLintMessage($diff) {
-    $link_to_clang_format =
-      "[[ http://fburl.com/clang-format | clang-format ]]";
-    return <<<LINT_MSG
-Changes in this file were not formatted using $link_to_clang_format.
-Please run build_tools/format-diff.sh or `make format`
-LINT_MSG;
-  }
-}

arcanist_util/cpp_linter/FbcodeCppLinter.php

@@ -1,126 +0,0 @@
-<?php
-// Copyright 2004-present Facebook.  All rights reserved.
-
-class FbcodeCppLinter extends ArcanistLinter {
-  const FLINT      = "/home/engshare/tools/flint";
-  const LINT_ERROR   = 1;
-  const LINT_WARNING = 2;
-  const LINT_ADVICE  = 3;
-  const C_FLAG = "--c_mode=true";
-
-  private $rawLintOutput = array();
-
-  public function willLintPaths(array $paths) {
-    if (!file_exists(self::FLINT)) {
-      return;
-    }
-    $futures = array();
-    foreach ($paths as $p) {
-      $lpath = $this->getEngine()->getFilePathOnDisk($p);
-      $lpath_file = file($lpath);
-      if (preg_match('/\.(c)$/', $lpath) ||
-          preg_match('/-\*-.*Mode: C[; ].*-\*-/', $lpath_file[0]) ||
-          preg_match('/vim(:.*)*:\s*(set\s+)?filetype=c\s*:/', $lpath_file[0])
-          ) {
-        $futures[$p] = new ExecFuture("%s %s %s 2>&1",
-                           self::FLINT, self::C_FLAG,
-                           $this->getEngine()->getFilePathOnDisk($p));
-      } else {
-        $futures[$p] = new ExecFuture("%s %s 2>&1",
-          self::FLINT, $this->getEngine()->getFilePathOnDisk($p));
-      }
-    }
-
-    foreach (Futures($futures)->limit(8) as $p => $f) {
-      $this->rawLintOutput[$p] = $f->resolvex();
-    }
-
-    return;
-  }
-
-  public function getLinterName() {
-    return "FBCPP";
-  }
-
-  public function lintPath($path) {
-    $this->runCppLint($path);
-  }
-
-  private function runCppLint($path) {
-    $msgs = $this->getCppLintOutput($path);
-    foreach ($msgs as $m) {
-      $this->raiseLintAtLine($m['line'], 0, $m['severity'], $m['msg']);
-    }
-  }
-
-  private function adviseOnEachPattern(
-    $path,
-    $regex,
-    $message,
-    $lint_type = self::LINT_ADVICE,
-    $match_idx = 0) {
-      $file_data = $this->getData($path);
-      $matches = array();
-      if (!preg_match_all($regex, $file_data, $matches, PREG_OFFSET_CAPTURE)) {
-        return;
-      }
-
-      foreach ($matches[$match_idx] as $match) {
-        list($match_str, $offset) = $match;
-        $this->raiseLintAtOffset($offset, $lint_type, $message, $match_str);
-      }
-  }
-
-  public function getLintSeverityMap() {
-    return array(
-      self::LINT_WARNING => ArcanistLintSeverity::SEVERITY_WARNING,
-      self::LINT_ADVICE  => ArcanistLintSeverity::SEVERITY_ADVICE,
-      self::LINT_ERROR   => ArcanistLintSeverity::SEVERITY_ERROR
-    );
-  }
-
-  public function getLintNameMap() {
-    return array(
-      self::LINT_ADVICE   => "CppLint Advice",
-      self::LINT_WARNING  => "CppLint Warning",
-      self::LINT_ERROR    => "CppLint Error"
-    );
-  }
-
-  private function getCppLintOutput($path) {
-     if (!array_key_exists($path, $this->rawLintOutput)) {
-       return array();
-     }
-    list($output) = $this->rawLintOutput[$path];
-
-    $msgs = array();
-    $current = null;
-    $matches = array();
-    foreach (explode("\n", $output) as $line) {
-      if (preg_match('/.*?:(\d+):(.*)/', $line, $matches)) {
-        if ($current) {
-          $msgs[] = $current;
-        }
-        $line = $matches[1];
-        $text = $matches[2];
-        if (preg_match('/.*Warning.*/', $text)) {
-          $sev = self::LINT_WARNING;
-        } else if (preg_match('/.*Advice.*/', $text)) {
-          $sev = self::LINT_ADVICE;
-        } else {
-          $sev = self::LINT_ERROR;
-        }
-        $current = array('line'     => $line,
-                         'msg'      => $text,
-                         'severity' => $sev);
-      } else if ($current) {
-        $current['msg'] .= ' ' . $line;
-      }
-    }
-    if ($current) {
-      $msgs[] = $current;
-    }
-
-    return $msgs;
-  }
-}

arcanist_util/lint_engine/FacebookFbcodeLintEngine.php

@@ -1,138 +0,0 @@
-<?php
-// Copyright 2004-present Facebook.  All rights reserved.
-// This source code is licensed under the BSD-style license found in the
-// LICENSE file in the root directory of this source tree. An additional grant
-// of patent rights can be found in the PATENTS file in the same directory.
-
-class FacebookFbcodeLintEngine extends ArcanistLintEngine {
-
-  public function buildLinters() {
-    $linters = array();
-    $paths = $this->getPaths();
-
-    // Remove all deleted files, which are not checked by the
-    // following linters.
-    foreach ($paths as $key => $path) {
-      if (!Filesystem::pathExists($this->getFilePathOnDisk($path))) {
-        unset($paths[$key]);
-      }
-    }
-
-    $generated_linter = new ArcanistGeneratedLinter();
-    $linters[] = $generated_linter;
-
-    $nolint_linter = new ArcanistNoLintLinter();
-    $linters[] = $nolint_linter;
-
-    $text_linter = new ArcanistTextLinter();
-    $text_linter->setCustomSeverityMap(array(
-      ArcanistTextLinter::LINT_LINE_WRAP
-        => ArcanistLintSeverity::SEVERITY_ADVICE,
-    ));
-    $linters[] = $text_linter;
-
-    $java_text_linter = new ArcanistTextLinter();
-    $java_text_linter->setMaxLineLength(100);
-    $java_text_linter->setCustomSeverityMap(array(
-      ArcanistTextLinter::LINT_LINE_WRAP
-        => ArcanistLintSeverity::SEVERITY_ADVICE,
-    ));
-    $linters[] = $java_text_linter;
-
-    $python_linter = new ArcanistPEP8Linter();
-    $linters[] = $python_linter;
-
-    $cpp_linters = array();
-    $cpp_linters[] = $linters[] = new ArcanistCpplintLinter();
-    $cpp_linters[] = $linters[] = new FbcodeCppLinter();
-
-    $clang_format_linter = new FbcodeClangFormatLinter();
-    $linters[] = $clang_format_linter;
-
-    $spelling_linter = new ArcanistSpellingLinter();
-    $linters[] = $spelling_linter;
-
-    foreach ($paths as $path) {
-      $is_text = false;
-
-      $text_extensions = (
-        '/\.('.
-        'cpp|cxx|c|cc|h|hpp|hxx|tcc|'.
-        'py|rb|hs|pl|pm|tw|'.
-        'php|phpt|css|js|'.
-        'java|'.
-        'thrift|'.
-        'lua|'.
-        'siv|'.
-        'txt'.
-        ')$/'
-      );
-      if (preg_match($text_extensions, $path)) {
-        $is_text = true;
-      }
-      if ($is_text) {
-        $nolint_linter->addPath($path);
-
-        $generated_linter->addPath($path);
-        $generated_linter->addData($path, $this->loadData($path));
-
-        if (preg_match('/\.java$/', $path)) {
-          $java_text_linter->addPath($path);
-          $java_text_linter->addData($path, $this->loadData($path));
-        } else {
-          $text_linter->addPath($path);
-          $text_linter->addData($path, $this->loadData($path));
-        }
-
-        $spelling_linter->addPath($path);
-        $spelling_linter->addData($path, $this->loadData($path));
-      }
-      if (preg_match('/\.(cpp|c|cc|cxx|h|hh|hpp|hxx|tcc)$/', $path)
-          && !preg_match('/third-party/', $path)) {
-        foreach ($cpp_linters as &$linter) {
-          $linter->addPath($path);
-          $linter->addData($path, $this->loadData($path));
-        }
-
-        $clang_format_linter->addPath($path);
-        $clang_format_linter->addData($path, $this->loadData($path));
-        $clang_format_linter->setPathChangedLines(
-          $path, $this->getPathChangedLines($path));
-      }
-
-      // Match *.py and contbuild config files
-      if (preg_match('/(\.(py|tw|smcprops)|^contbuild\/configs\/[^\/]*)$/',
-                    $path)) {
-        $space_count = 4;
-        $real_path = $this->getFilePathOnDisk($path);
-        $dir = dirname($real_path);
-        do {
-          if (file_exists($dir.'/.python2space')) {
-            $space_count = 2;
-            break;
-          }
-          $dir = dirname($dir);
-        } while ($dir != '/' && $dir != '.');
-
-        $cur_path_linter = $python_linter;
-        $cur_path_linter->addPath($path);
-        $cur_path_linter->addData($path, $this->loadData($path));
-
-        if (preg_match('/\.tw$/', $path)) {
-          $cur_path_linter->setCustomSeverityMap(array(
-            'E251' => ArcanistLintSeverity::SEVERITY_DISABLED,
-          ));
-        }
-      }
-    }
-
-    $name_linter = new ArcanistFilenameLinter();
-    $linters[] = $name_linter;
-    foreach ($paths as $path) {
-      $name_linter->addPath($path);
-    }
-
-    return $linters;
-  }
-
-}

arcanist_util/lint_engine/FacebookHowtoevenLintEngine.php

@@ -1,27 +0,0 @@
-<?php
-// Copyright 2015-present Facebook. All Rights Reserved.
-// This source code is licensed under the BSD-style license found in the
-// LICENSE file in the root directory of this source tree. An additional grant
-// of patent rights can be found in the PATENTS file in the same directory.
-
-final class FacebookHowtoevenLintEngine extends ArcanistLintEngine {
-
-  public function buildLinters() {
-    $paths = array();
-
-    foreach ($this->getPaths() as $path) {
-      // Don't try to lint deleted files or changed directories.
-      if (!Filesystem::pathExists($path) || is_dir($path)) {
-        continue;
-      }
-
-      if (preg_match('/\.(cpp|c|cc|cxx|h|hh|hpp|hxx|tcc)$/', $path)) {
-        $paths[] = $path;
-      }
-    }
-
-    $howtoeven = new FacebookHowtoevenLinter();
-    $howtoeven->setPaths($paths);
-    return array($howtoeven);
-  }
-}

arcanist_util/unit_engine/FacebookFbcodeUnitTestEngine.php

@@ -1,17 +0,0 @@
-<?php
-// Copyright 2004-present Facebook. All Rights Reserved.
-// This source code is licensed under the BSD-style license found in the
-// LICENSE file in the root directory of this source tree. An additional grant
-// of patent rights can be found in the PATENTS file in the same directory.
-
-class FacebookFbcodeUnitTestEngine extends ArcanistUnitTestEngine {
-
-  public function run() {
-      // For a call to `arc call-conduit differential.updateunitresults` to
-      // succeed we need at least one entry here.
-      $result = new ArcanistUnitTestResult();
-      $result->setName("dummy_placeholder_entry");
-      $result->setResult(ArcanistUnitTestResult::RESULT_PASS);
-      return array($result);
-  }
-}

arcanist_util/unit_engine/FacebookOldFbcodeUnitTestEngine.php

@@ -1,17 +0,0 @@
-<?php
-// Copyright 2004-present Facebook. All Rights Reserved.
-// This source code is licensed under the BSD-style license found in the
-// LICENSE file in the root directory of this source tree. An additional grant
-// of patent rights can be found in the PATENTS file in the same directory.
-
-class FacebookFbcodeUnitTestEngine extends ArcanistBaseUnitTestEngine {
-
-  public function run() {
-      // For a call to `arc call-conduit differential.updateunitresults` to
-      // succeed we need at least one entry here.
-      $result = new ArcanistUnitTestResult();
-      $result->setName("dummy_placeholder_entry");
-      $result->setResult(ArcanistUnitTestResult::RESULT_PASS);
-      return array($result);
-  }
-}

buckifier/rocks_test_runner.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 # Create a tmp directory for the test to use
 TEST_DIR=$(mktemp -d /dev/shm/fbcode_rocksdb_XXXXXXX)
 TEST_TMPDIR="$TEST_DIR" $@ && rm -rf "$TEST_DIR"

buckifier/targets_builder.py

@@ -3,19 +3,17 @@ from __future__ import division
 from __future__ import print_function
 from __future__ import unicode_literals
 import targets_cfg
-import pprint
 
-# TODO(tec): replace this with PrettyPrinter
-def pretty_list(lst, indent=6):
+def pretty_list(lst, indent=8):
     if lst is None or len(lst) == 0:
         return ""
 
     if len(lst) == 1:
         return "\"%s\"" % lst[0]
-
+    
     separator = "\",\n%s\"" % (" " * indent)
-    res = separator.join(lst)
-    res = "\n" + (" " * indent) + "\"" + res + "\",\n" + (" " * (indent - 2))
+    res = separator.join(sorted(lst))
+    res = "\n" + (" " * indent) + "\"" + res + "\",\n" + (" " * (indent - 4))
     return res
 
 
@@ -27,7 +25,7 @@ class TARGETSBuilder:
         self.total_lib = 0
         self.total_bin = 0
         self.total_test = 0
-        self.tests_cfg = []
+        self.tests_cfg = ""
 
     def __del__(self):
         self.targets_file.close()
@@ -37,8 +35,8 @@ class TARGETSBuilder:
             headers = "AutoHeaders.RECURSIVE_GLOB"
         self.targets_file.write(targets_cfg.library_template % (
             name,
-            headers,
             pretty_list(srcs),
+            headers,
             pretty_list(deps)))
         self.total_lib = self.total_lib + 1
 
@@ -53,13 +51,13 @@ class TARGETSBuilder:
         exec_mode = "serial"
         if is_parallel:
             exec_mode = "parallel"
-        self.tests_cfg.append([test_name, str(src), str(exec_mode)])
+        self.tests_cfg += targets_cfg.test_cfg_template % (
+            test_name,
+            str(src),
+            str(exec_mode))
 
         self.total_test = self.total_test + 1
 
     def flush_tests(self):
-        self.targets_file.write(targets_cfg.unittests_template % (
-            pprint.PrettyPrinter().pformat(self.tests_cfg)
-        ))
-
-        self.tests_cfg = []
+        self.targets_file.write(targets_cfg.unittests_template % self.tests_cfg)
+        self.tests_cfg = ""

buckifier/targets_cfg.py

@@ -2,119 +2,141 @@ from __future__ import absolute_import
 from __future__ import division
 from __future__ import print_function
 from __future__ import unicode_literals
-rocksdb_target_header = """
-import os
+rocksdb_target_header = """REPO_PATH = package_name() + "/"
 
-TARGETS_PATH = os.path.dirname(__file__)
-REPO_PATH = TARGETS_PATH[(TARGETS_PATH.find('fbcode/') + len('fbcode/')):] + "/"
 BUCK_BINS = "buck-out/gen/" + REPO_PATH
+
 TEST_RUNNER = REPO_PATH + "buckifier/rocks_test_runner.sh"
+
 rocksdb_compiler_flags = [
-  "-msse",
-  "-msse4.2",
-  "-fno-builtin-memcmp",
-  "-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",
-  "-DOS_LINUX",
-  # Flags to enable libs we include
-  "-DSNAPPY",
-  "-DZLIB",
-  "-DBZIP2",
-  "-DLZ4",
-  "-DZSTD",
-  "-DGFLAGS=gflags",
-  "-DNUMA",
-  "-DTBB",
-  # Needed to compile in fbcode
-  "-Wno-expansion-to-defined",
+    "-fno-builtin-memcmp",
+    "-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",
+    "-DOS_LINUX",
+    # Flags to enable libs we include
+    "-DSNAPPY",
+    "-DZLIB",
+    "-DBZIP2",
+    "-DLZ4",
+    "-DZSTD",
+    "-DGFLAGS=gflags",
+    "-DNUMA",
+    "-DTBB",
+    # Needed to compile in fbcode
+    "-Wno-expansion-to-defined",
 ]
 
 rocksdb_external_deps = [
-  ('bzip2', None, 'bz2'),
-  ('snappy', None, "snappy"),
-  ('zlib', None, 'z'),
-  ('gflags', None, 'gflags'),
-  ('lz4', None, 'lz4'),
-  ('zstd', None),
-  ('tbb', None),
-  ("numa", None, "numa"),
-  ("googletest", None, "gtest"),
+    ("bzip2", None, "bz2"),
+    ("snappy", None, "snappy"),
+    ("zlib", None, "z"),
+    ("gflags", None, "gflags"),
+    ("lz4", None, "lz4"),
+    ("zstd", None),
+    ("tbb", None),
+    ("numa", None, "numa"),
+    ("googletest", None, "gtest"),
 ]
 
 rocksdb_preprocessor_flags = [
-  # Directories with files for #include
-  "-I" + REPO_PATH + "include/",
-  "-I" + REPO_PATH,
+    # Directories with files for #include
+    "-I" + REPO_PATH + "include/",
+    "-I" + REPO_PATH,
 ]
+
+rocksdb_arch_preprocessor_flags = {
+    "x86_64": ["-DHAVE_SSE42"],
+}
+
+build_mode = read_config("fbcode", "build_mode")
+
+is_opt_mode = build_mode.startswith("opt")
+
+# -DNDEBUG is added by default in opt mode in fbcode. But adding it twice
+# doesn't harm and avoid forgetting to add it.
+if is_opt_mode:
+    rocksdb_compiler_flags.append("-DNDEBUG")
 """
 
 
 library_template = """
 cpp_library(
     name = "%s",
-    headers = %s,
     srcs = [%s],
-    deps = [%s],
-    preprocessor_flags = rocksdb_preprocessor_flags,
+    headers = %s,
+    arch_preprocessor_flags = rocksdb_arch_preprocessor_flags,
     compiler_flags = rocksdb_compiler_flags,
+    preprocessor_flags = rocksdb_preprocessor_flags,
+    deps = [%s],
     external_deps = rocksdb_external_deps,
 )
 """
 
 binary_template = """
 cpp_binary(
-  name = "%s",
-  srcs = [%s],
-  deps = [%s],
-  preprocessor_flags = rocksdb_preprocessor_flags,
-  compiler_flags = rocksdb_compiler_flags,
-  external_deps = rocksdb_external_deps,
+    name = "%s",
+    srcs = [%s],
+    arch_preprocessor_flags = rocksdb_arch_preprocessor_flags,
+    compiler_flags = rocksdb_compiler_flags,
+    preprocessor_flags = rocksdb_preprocessor_flags,
+    deps = [%s],
+    external_deps = rocksdb_external_deps,
 )
 """
 
+test_cfg_template = """    [
+        "%s",
+        "%s",
+        "%s",
+    ],
+"""
+
 unittests_template = """
 # [test_name, test_src, test_type]
-ROCKS_TESTS = %s
-
+ROCKS_TESTS = [
+%s]
 
 # Generate a test rule for each entry in ROCKS_TESTS
-for test_cfg in ROCKS_TESTS:
-    test_name = test_cfg[0]
-    test_cc = test_cfg[1]
-    ttype = "gtest" if test_cfg[2] == "parallel" else "simple"
-    test_bin = test_name + "_bin"
+# Do not build the tests in opt mode, since SyncPoint and other test code
+# will not be included.
+if not is_opt_mode:
+    for test_cfg in ROCKS_TESTS:
+        test_name = test_cfg[0]
+        test_cc = test_cfg[1]
+        ttype = "gtest" if test_cfg[2] == "parallel" else "simple"
+        test_bin = test_name + "_bin"
 
-    cpp_binary (
-      name = test_bin,
-      srcs = [test_cc],
-      deps = [":rocksdb_test_lib"],
-      preprocessor_flags = rocksdb_preprocessor_flags,
-      compiler_flags = rocksdb_compiler_flags,
-      external_deps = rocksdb_external_deps,
-    )
+        cpp_binary (
+          name = test_bin,
+          srcs = [test_cc],
+          deps = [":rocksdb_test_lib"],
+          preprocessor_flags = rocksdb_preprocessor_flags,
+          arch_preprocessor_flags = rocksdb_arch_preprocessor_flags,
+          compiler_flags = rocksdb_compiler_flags,
+          external_deps = rocksdb_external_deps,
+        )
 
-    custom_unittest(
-      name = test_name,
-      type = ttype,
-      deps = [":" + test_bin],
-      command = [TEST_RUNNER, BUCK_BINS + test_bin]
-    )
+        custom_unittest(
+          name = test_name,
+          type = ttype,
+          deps = [":" + test_bin],
+          command = [TEST_RUNNER, BUCK_BINS + test_bin]
+        )
 
 custom_unittest(
     name = "make_rocksdbjavastatic",
-    type = "simple",
     command = ["internal_repo_rocksdb/make_rocksdbjavastatic.sh"],
+    type = "simple",
 )
 
 custom_unittest(
     name = "make_rocksdb_lite_release",
-    type = "simple",
     command = ["internal_repo_rocksdb/make_rocksdb_lite_release.sh"],
+    type = "simple",
 )
 """

build_tools/RocksDBCommonHelper.php

@@ -1,8 +1,8 @@
 <?php
 // Copyright 2004-present Facebook. All Rights Reserved.
-// This source code is licensed under the BSD-style license found in the
-// LICENSE file in the root directory of this source tree. An additional grant
-// of patent rights can be found in the PATENTS file in the same directory.
+//  This 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).
 
 // Name of the environment variables which need to be set by the entity which
 // triggers continuous runs so that code at the end of the file gets executed
@@ -21,11 +21,17 @@ function postURL($diffID, $url) {
   assert(is_numeric($diffID));
   assert(strlen($url) > 0);
 
-  $cmd = 'echo \'{"diff_id": ' . $diffID . ', '
-         . '"name":"click here for sandcastle tests for D' . $diffID . '", '
-         . '"link":"' . $url . '"}\' | '
-         . 'arc call-conduit '
-         . 'differential.updateunitresults';
+  $cmd_args = array(
+    'diff_id' => (int)$diffID,
+    'name' => sprintf(
+      'click here for sandcastle tests for D%d',
+      (int)$diffID
+    ),
+    'link' => $url
+  );
+  $cmd = 'echo ' . escapeshellarg(json_encode($cmd_args))
+         . ' | arc call-conduit differential.updateunitresults';
+
   shell_exec($cmd);
 }
 
@@ -35,11 +41,15 @@ function buildUpdateTestStatusCmd($diffID, $test, $status) {
   assert(strlen($test) > 0);
   assert(strlen($status) > 0);
 
-  $cmd = 'echo \'{"diff_id": ' . $diffID . ', '
-         . '"name":"' . $test . '", '
-         . '"result":"' . $status . '"}\' | '
-         . 'arc call-conduit '
-         . 'differential.updateunitresults';
+  $cmd_args = array(
+    'diff_id' => (int)$diffID,
+    'name' => $test,
+    'result' => $status
+  );
+
+  $cmd = 'echo ' . escapeshellarg(json_encode($cmd_args))
+         . ' | arc call-conduit differential.updateunitresults';
+
   return $cmd;
 }
 
@@ -68,7 +78,7 @@ function getSteps($applyDiff, $diffID, $username, $test) {
     // and authenticate using that in Sandcastle.
     $setup = array(
       "name" => "Setup arcrc",
-      "shell" => "echo " . $arcrc_content . " | base64 --decode"
+      "shell" => "echo " . escapeshellarg($arcrc_content) . " | base64 --decode"
                  . " | gzip -d > ~/.arcrc",
       "user" => "root"
     );
@@ -94,7 +104,17 @@ function getSteps($applyDiff, $diffID, $username, $test) {
     "user" => "root"
   );
 
+  // This fixes "FATAL: ThreadSanitizer can not mmap the shadow memory"
+  // Source:
+  // https://github.com/google/sanitizers/wiki/ThreadSanitizerCppManual#FAQ
+  $fix_kernel_issue = array(
+    "name" => "Fix kernel issue with tsan",
+    "shell" => "echo 2 >/proc/sys/kernel/randomize_va_space",
+    "user" => "root"
+  );
+
   $steps[] = $fix_git_ignore;
+  $steps[] = $fix_kernel_issue;
 
   // This will be the command used to execute particular type of tests.
   $cmd = "";
@@ -104,7 +124,7 @@ function getSteps($applyDiff, $diffID, $username, $test) {
     $patch = array(
       "name" => "Patch " . $diffID,
       "shell" => "arc --arcrc-file ~/.arcrc "
-                  . "patch --nocommit --diff " . $diffID,
+                  . "patch --nocommit --diff " . escapeshellarg($diffID),
       "user" => "root"
     );
 
@@ -115,8 +135,8 @@ function getSteps($applyDiff, $diffID, $username, $test) {
   }
 
   // Run the actual command.
-  $cmd = $cmd . "J=$(nproc) ./build_tools/precommit_checker.py " . $test
-           . "; exit_code=$?; ";
+  $cmd = $cmd . "J=$(nproc) ./build_tools/precommit_checker.py " .
+           escapeshellarg($test) . "; exit_code=$?; ";
 
   if ($applyDiff) {
     $cmd = $cmd . "([[ \$exit_code -eq 0 ]] &&"
@@ -149,7 +169,7 @@ function getSteps($applyDiff, $diffID, $username, $test) {
     "name" => "Run " . $test,
     "shell" => $cmd,
     "user" => "root",
-    "parser" => "python build_tools/error_filter.py " . $test,
+    "parser" => "python build_tools/error_filter.py " . escapeshellarg($test),
   );
 
   $steps[] = $run_test;
@@ -197,7 +217,7 @@ function getSandcastleConfig() {
     if (file_exists(PRIMARY_TOKEN_FILE)) {
       $cmd = 'cat ' . PRIMARY_TOKEN_FILE;
     } else {
-      $cmd = 'cat ' . $cwd_token_file;
+      $cmd = 'cat ' . escapeshellarg($cwd_token_file);
     }
 
     assert(strlen($cmd) > 0);
@@ -241,7 +261,7 @@ function getSandcastleConfig() {
   // List of tests we want to run in Sandcastle.
   $tests = array("unit", "unit_non_shm", "unit_481", "clang_unit", "tsan",
                  "asan", "lite_test", "valgrind", "release", "release_481",
-                 "clang_release", "punit", "clang_analyze", "code_cov",
+                 "clang_release", "clang_analyze", "code_cov",
                  "java_build", "no_compression", "unity", "ubsan");
 
   $send_email_template = array(
@@ -321,9 +341,11 @@ function getSandcastleConfig() {
   $app = $sandcastle_config[0];
   $token = $sandcastle_config[1];
 
-  $cmd = 'curl -s -k -F app=' . $app . ' '
-          . '-F token=' . $token . ' -F job=\'' . json_encode($job)
-          .'\' "' . $url . '"';
+  $cmd = 'curl -s -k '
+          . ' -F app=' . escapeshellarg($app)
+          . ' -F token=' . escapeshellarg($token)
+          . ' -F job=' . escapeshellarg(json_encode($job))
+          .' ' . escapeshellarg($url);
 
   $output = shell_exec($cmd);
   assert(strlen($output) > 0);

build_tools/build_detect_platform

@@ -51,13 +51,11 @@ if [ -z "$ROCKSDB_NO_FBCODE" -a -d /mnt/gvfs/third-party ]; then
     FBCODE_BUILD="true"
     # If we're compiling with TSAN we need pic build
     PIC_BUILD=$COMPILE_WITH_TSAN
-    if [ -n "$ROCKSDB_FBCODE_BUILD_WITH_481" ]; then
-      # we need this to build with MySQL. Don't use for other purposes.
-      source "$PWD/build_tools/fbcode_config4.8.1.sh"
-    elif [ -n "$ROCKSDB_FBCODE_BUILD_WITH_5xx" ]; then
+    if [ -z "$ROCKSDB_FBCODE_BUILD_WITH_481" ]; then
       source "$PWD/build_tools/fbcode_config.sh"
     else
-      source "$PWD/build_tools/fbcode_config_platform007.sh"
+      # we need this to build with MySQL. Don't use for other purposes.
+      source "$PWD/build_tools/fbcode_config4.8.1.sh"
     fi
 fi
 
@@ -143,6 +141,7 @@ case "$TARGET_OS" in
         ;;
     FreeBSD)
         PLATFORM=OS_FREEBSD
+        CXX=clang++
         COMMON_FLAGS="$COMMON_FLAGS -fno-builtin-memcmp -D_REENTRANT -DOS_FREEBSD"
         PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lpthread"
         # PORT_FILES=port/freebsd/freebsd_specific.cc
@@ -212,117 +211,137 @@ EOF
         fi
     fi
 
-    # Test whether Snappy library is installed
-    # http://code.google.com/p/snappy/
-    $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
-      #include <snappy.h>
-      int main() {}
+    if ! test $ROCKSDB_DISABLE_SNAPPY; then
+        # Test whether Snappy library is installed
+        # http://code.google.com/p/snappy/
+        $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
+          #include <snappy.h>
+          int main() {}
 EOF
-    if [ "$?" = 0 ]; then
-        COMMON_FLAGS="$COMMON_FLAGS -DSNAPPY"
-        PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lsnappy"
-        JAVA_LDFLAGS="$JAVA_LDFLAGS -lsnappy"
+        if [ "$?" = 0 ]; then
+            COMMON_FLAGS="$COMMON_FLAGS -DSNAPPY"
+            PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lsnappy"
+            JAVA_LDFLAGS="$JAVA_LDFLAGS -lsnappy"
+        fi
     fi
 
-    # Test whether gflags library is installed
-    # http://gflags.github.io/gflags/
-    # check if the namespace is gflags
-    $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null << EOF
-      #include <gflags/gflags.h>
-      using namespace gflags;
-      int main() {}
+    if ! test $ROCKSDB_DISABLE_GFLAGS; then
+        # Test whether gflags library is installed
+        # http://gflags.github.io/gflags/
+        # check if the namespace is gflags
+        $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null << EOF
+          #include <gflags/gflags.h>
+          int main() {}
 EOF
-    if [ "$?" = 0 ]; then
-        COMMON_FLAGS="$COMMON_FLAGS -DGFLAGS=gflags"
-        PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lgflags"
-    else
-      # check if namespace is google
-      $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null << EOF
-        #include <gflags/gflags.h>
-        using namespace google;
-        int main() {}
+        if [ "$?" = 0 ]; then
+            COMMON_FLAGS="$COMMON_FLAGS -DGFLAGS=1"
+            PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lgflags"
+        else
+          # check if namespace is google
+          $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null << EOF
+            #include <gflags/gflags.h>
+            using namespace google;
+            int main() {}
 EOF
-      if [ "$?" = 0 ]; then
-          COMMON_FLAGS="$COMMON_FLAGS -DGFLAGS=google"
-          PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lgflags"
-      fi
+          if [ "$?" = 0 ]; then
+              COMMON_FLAGS="$COMMON_FLAGS -DGFLAGS=google"
+              PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lgflags"
+          fi
+        fi
     fi
 
-    # Test whether zlib library is installed
-    $CXX $CFLAGS $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
-      #include <zlib.h>
-      int main() {}
+    if ! test $ROCKSDB_DISABLE_ZLIB; then
+        # Test whether zlib library is installed
+        $CXX $CFLAGS $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
+          #include <zlib.h>
+          int main() {}
 EOF
-    if [ "$?" = 0 ]; then
-        COMMON_FLAGS="$COMMON_FLAGS -DZLIB"
-        PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lz"
-        JAVA_LDFLAGS="$JAVA_LDFLAGS -lz"
+        if [ "$?" = 0 ]; then
+            COMMON_FLAGS="$COMMON_FLAGS -DZLIB"
+            PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lz"
+            JAVA_LDFLAGS="$JAVA_LDFLAGS -lz"
+        fi
     fi
 
-    # Test whether bzip library is installed
-    $CXX $CFLAGS $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
-      #include <bzlib.h>
-      int main() {}
+    if ! test $ROCKSDB_DISABLE_BZIP; then
+        # Test whether bzip library is installed
+        $CXX $CFLAGS $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
+          #include <bzlib.h>
+          int main() {}
 EOF
-    if [ "$?" = 0 ]; then
-        COMMON_FLAGS="$COMMON_FLAGS -DBZIP2"
-        PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lbz2"
-        JAVA_LDFLAGS="$JAVA_LDFLAGS -lbz2"
+        if [ "$?" = 0 ]; then
+            COMMON_FLAGS="$COMMON_FLAGS -DBZIP2"
+            PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lbz2"
+            JAVA_LDFLAGS="$JAVA_LDFLAGS -lbz2"
+        fi
     fi
 
-    # Test whether lz4 library is installed
-    $CXX $CFLAGS $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
-      #include <lz4.h>
-      #include <lz4hc.h>
-      int main() {}
+    if ! test $ROCKSDB_DISABLE_LZ4; then
+        # Test whether lz4 library is installed
+        $CXX $CFLAGS $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
+          #include <lz4.h>
+          #include <lz4hc.h>
+          int main() {}
 EOF
-    if [ "$?" = 0 ]; then
-        COMMON_FLAGS="$COMMON_FLAGS -DLZ4"
-        PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -llz4"
-        JAVA_LDFLAGS="$JAVA_LDFLAGS -llz4"
+        if [ "$?" = 0 ]; then
+            COMMON_FLAGS="$COMMON_FLAGS -DLZ4"
+            PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -llz4"
+            JAVA_LDFLAGS="$JAVA_LDFLAGS -llz4"
+        fi
     fi
 
-    # Test whether zstd library is installed
-    $CXX $CFLAGS $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
-      #include <zstd.h>
-      int main() {}
+    if ! test $ROCKSDB_DISABLE_ZSTD; then
+        # Test whether zstd library is installed
+        $CXX $CFLAGS $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
+          #include <zstd.h>
+          int main() {}
 EOF
-    if [ "$?" = 0 ]; then
-        COMMON_FLAGS="$COMMON_FLAGS -DZSTD"
-        PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lzstd"
-        JAVA_LDFLAGS="$JAVA_LDFLAGS -lzstd"
+        if [ "$?" = 0 ]; then
+            COMMON_FLAGS="$COMMON_FLAGS -DZSTD"
+            PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lzstd"
+            JAVA_LDFLAGS="$JAVA_LDFLAGS -lzstd"
+        fi
     fi
 
-    # Test whether numa is available
-    $CXX $CFLAGS -x c++ - -o /dev/null -lnuma 2>/dev/null  <<EOF
-      #include <numa.h>
-      #include <numaif.h>
-      int main() {}
+    if ! test $ROCKSDB_DISABLE_NUMA; then
+        # Test whether numa is available
+        $CXX $CFLAGS -x c++ - -o /dev/null -lnuma 2>/dev/null  <<EOF
+          #include <numa.h>
+          #include <numaif.h>
+          int main() {}
 EOF
-    if [ "$?" = 0 ]; then
-        COMMON_FLAGS="$COMMON_FLAGS -DNUMA"
-        PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lnuma"
-        JAVA_LDFLAGS="$JAVA_LDFLAGS -lnuma"
+        if [ "$?" = 0 ]; then
+            COMMON_FLAGS="$COMMON_FLAGS -DNUMA"
+            PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lnuma"
+            JAVA_LDFLAGS="$JAVA_LDFLAGS -lnuma"
+        fi
     fi
 
-    # Test whether tbb is available
-    $CXX $CFLAGS $LDFLAGS -x c++ - -o /dev/null -ltbb 2>/dev/null  <<EOF
-      #include <tbb/tbb.h>
-      int main() {}
+    if ! test $ROCKSDB_DISABLE_TBB; then
+        # Test whether tbb is available
+        $CXX $CFLAGS $LDFLAGS -x c++ - -o /dev/null -ltbb 2>/dev/null  <<EOF
+          #include <tbb/tbb.h>
+          int main() {}
 EOF
-    if [ "$?" = 0 ]; then
-        COMMON_FLAGS="$COMMON_FLAGS -DTBB"
-        PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -ltbb"
-        JAVA_LDFLAGS="$JAVA_LDFLAGS -ltbb"
+        if [ "$?" = 0 ]; then
+            COMMON_FLAGS="$COMMON_FLAGS -DTBB"
+            PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -ltbb"
+            JAVA_LDFLAGS="$JAVA_LDFLAGS -ltbb"
+        fi
     fi
 
-    # Test whether jemalloc is available
-    if echo 'int main() {}' | $CXX $CFLAGS -x c++ - -o /dev/null -ljemalloc \
-      2>/dev/null; then
-        PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -ljemalloc"
-        JAVA_LDFLAGS="$JAVA_LDFLAGS -ljemalloc"
-        JEMALLOC=1
-    else
+    if ! test $ROCKSDB_DISABLE_JEMALLOC; then
+        # Test whether jemalloc is available
+        if echo 'int main() {}' | $CXX $CFLAGS -x c++ - -o /dev/null -ljemalloc \
+          2>/dev/null; then
+            # This will enable some preprocessor identifiers in the Makefile
+            JEMALLOC=1
+            # JEMALLOC can be enabled either using the flag (like here) or by
+            # providing direct link to the jemalloc library
+            WITH_JEMALLOC_FLAG=1
+        fi
+    fi
+    if ! test $JEMALLOC && ! test $ROCKSDB_DISABLE_TCMALLOC; then
         # jemalloc is not available. Let's try tcmalloc
         if echo 'int main() {}' | $CXX $CFLAGS -x c++ - -o /dev/null \
           -ltcmalloc 2>/dev/null; then
@@ -331,88 +350,100 @@ EOF
         fi
     fi
 
-    # Test whether malloc_usable_size is available
-    $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
-      #include <malloc.h>
-      int main() {
-        size_t res = malloc_usable_size(0);
-        return 0;
-      }
+    if ! test $ROCKSDB_DISABLE_MALLOC_USABLE_SIZE; then
+        # Test whether malloc_usable_size is available
+        $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
+          #include <malloc.h>
+          int main() {
+            size_t res = malloc_usable_size(0);
+            return 0;
+          }
 EOF
-    if [ "$?" = 0 ]; then
-        COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_MALLOC_USABLE_SIZE"
+        if [ "$?" = 0 ]; then
+            COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_MALLOC_USABLE_SIZE"
+        fi
     fi
 
-    # Test whether PTHREAD_MUTEX_ADAPTIVE_NP mutex type is available
-    $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
-      #include <pthread.h>
-      int main() {
-        int x = PTHREAD_MUTEX_ADAPTIVE_NP;
-        return 0;
-      }
+    if ! test $ROCKSDB_DISABLE_PTHREAD_MUTEX_ADAPTIVE_NP; then
+        # Test whether PTHREAD_MUTEX_ADAPTIVE_NP mutex type is available
+        $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
+          #include <pthread.h>
+          int main() {
+            int x = PTHREAD_MUTEX_ADAPTIVE_NP;
+            return 0;
+          }
 EOF
-    if [ "$?" = 0 ]; then
-        COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_PTHREAD_ADAPTIVE_MUTEX"
+        if [ "$?" = 0 ]; then
+            COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_PTHREAD_ADAPTIVE_MUTEX"
+        fi
     fi
 
-    # Test whether backtrace is available
-    $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
-      #include <execinfo.h>>
-      int main() {
-        void* frames[1];
-        backtrace_symbols(frames, backtrace(frames, 1));
-        return 0;
-      }
-EOF
-    if [ "$?" = 0 ]; then
-        COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_BACKTRACE"
-    else
-        # Test whether execinfo library is installed
-        $CXX $CFLAGS -lexecinfo -x c++ - -o /dev/null 2>/dev/null  <<EOF
-          #include <execinfo.h>
+    if ! test $ROCKSDB_DISABLE_BACKTRACE; then
+        # Test whether backtrace is available
+        $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
+          #include <execinfo.h>>
           int main() {
             void* frames[1];
             backtrace_symbols(frames, backtrace(frames, 1));
+            return 0;
           }
 EOF
         if [ "$?" = 0 ]; then
             COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_BACKTRACE"
-            PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lexecinfo"
-            JAVA_LDFLAGS="$JAVA_LDFLAGS -lexecinfo"
+        else
+            # Test whether execinfo library is installed
+            $CXX $CFLAGS -lexecinfo -x c++ - -o /dev/null 2>/dev/null  <<EOF
+              #include <execinfo.h>
+              int main() {
+                void* frames[1];
+                backtrace_symbols(frames, backtrace(frames, 1));
+              }
+EOF
+            if [ "$?" = 0 ]; then
+                COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_BACKTRACE"
+                PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lexecinfo"
+                JAVA_LDFLAGS="$JAVA_LDFLAGS -lexecinfo"
+            fi
         fi
     fi
 
-    # Test if -pg is supported
-    $CXX $CFLAGS -pg -x c++ - -o /dev/null 2>/dev/null  <<EOF
-      int main() {
-        return 0;
-      }
+    if ! test $ROCKSDB_DISABLE_PG; then
+        # Test if -pg is supported
+        $CXX $CFLAGS -pg -x c++ - -o /dev/null 2>/dev/null  <<EOF
+          int main() {
+            return 0;
+          }
 EOF
-    if [ "$?" = 0 ]; then
-        PROFILING_FLAGS=-pg
+        if [ "$?" = 0 ]; then
+            PROFILING_FLAGS=-pg
+        fi
     fi
 
-    # Test whether sync_file_range is supported for compatibility with an old glibc
-    $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
-      #include <fcntl.h>
-      int main() {
-        int fd = open("/dev/null", 0);
-        sync_file_range(fd, 0, 1024, SYNC_FILE_RANGE_WRITE);
-      }
+    if ! test $ROCKSDB_DISABLE_SYNC_FILE_RANGE; then
+        # Test whether sync_file_range is supported for compatibility with an old glibc
+        $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
+          #include <fcntl.h>
+          int main() {
+            int fd = open("/dev/null", 0);
+            sync_file_range(fd, 0, 1024, SYNC_FILE_RANGE_WRITE);
+          }
 EOF
-    if [ "$?" = 0 ]; then
-        COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_RANGESYNC_PRESENT"
+        if [ "$?" = 0 ]; then
+            COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_RANGESYNC_PRESENT"
+        fi
     fi
 
-    # Test whether sched_getcpu is supported
-    $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
-      #include <sched.h>
-      int main() {
-        int cpuid = sched_getcpu();
-      }
+    if ! test $ROCKSDB_DISABLE_SCHED_GETCPU; then
+        # Test whether sched_getcpu is supported
+        $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
+          #include <sched.h>
+          int main() {
+            int cpuid = sched_getcpu();
+          }
 EOF
-    if [ "$?" = 0 ]; then
-        COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_SCHED_GETCPU_PRESENT"
+        if [ "$?" = 0 ]; then
+            COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_SCHED_GETCPU_PRESENT"
+        fi
     fi
 fi
 
@@ -444,21 +475,26 @@ if test "$USE_HDFS"; then
   JAVA_LDFLAGS="$JAVA_LDFLAGS $HDFS_LDFLAGS"
 fi
 
-if test "$USE_SSE"; then
-  COMMON_FLAGS="$COMMON_FLAGS -msse4.2"
-elif test -z "$PORTABLE"; then
+if test -z "$PORTABLE"; then
   if test -n "`echo $TARGET_ARCHITECTURE | grep ^ppc64`"; then
     # Tune for this POWER processor, treating '+' models as base models
     POWER=`LD_SHOW_AUXV=1 /bin/true | grep AT_PLATFORM | grep -E -o power[0-9]+`
     COMMON_FLAGS="$COMMON_FLAGS -mcpu=$POWER -mtune=$POWER "
   elif test -n "`echo $TARGET_ARCHITECTURE | grep ^s390x`"; then
     COMMON_FLAGS="$COMMON_FLAGS -march=z10 "
+  elif test -n "`echo $TARGET_ARCHITECTURE | grep ^arm`"; then
+    # TODO: Handle this with approprite options.
+    COMMON_FLAGS="$COMMON_FLAGS"
   elif [ "$TARGET_OS" != AIX ] && [ "$TARGET_OS" != SunOS ]; then
     COMMON_FLAGS="$COMMON_FLAGS -march=native "
+  elif test "$USE_SSE"; then
+    COMMON_FLAGS="$COMMON_FLAGS -msse4.2 -mpclmul"
   fi
+elif test "$USE_SSE"; then
+  COMMON_FLAGS="$COMMON_FLAGS -msse4.2 -mpclmul"
 fi
 
-$CXX $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null <<EOF
+$CXX $PLATFORM_CXXFLAGS $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null <<EOF
   #include <cstdint>
   #include <nmmintrin.h>
   int main() {
@@ -469,6 +505,24 @@ if [ "$?" = 0 ]; then
   COMMON_FLAGS="$COMMON_FLAGS -DHAVE_SSE42"
 elif test "$USE_SSE"; then
   echo "warning: USE_SSE specified but compiler could not use SSE intrinsics, disabling"
+  exit 1
+fi
+
+$CXX $PLATFORM_CXXFLAGS $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null <<EOF
+  #include <cstdint>
+  #include <wmmintrin.h>
+  int main() {
+    const auto a = _mm_set_epi64x(0, 0);
+    const auto b = _mm_set_epi64x(0, 0);
+    const auto c = _mm_clmulepi64_si128(a, b, 0x00);
+    auto d = _mm_cvtsi128_si64(c);
+  }
+EOF
+if [ "$?" = 0 ]; then
+  COMMON_FLAGS="$COMMON_FLAGS -DHAVE_PCLMUL"
+elif test "$USE_SSE"; then
+  echo "warning: USE_SSE specified but compiler could not use PCLMUL intrinsics, disabling"
+  exit 1
 fi
 
 # iOS doesn't support thread-local storage, but this check would erroneously
@@ -519,7 +573,14 @@ echo "ROCKSDB_PATCH=$ROCKSDB_PATCH" >> "$OUTPUT"
 echo "CLANG_SCAN_BUILD=$CLANG_SCAN_BUILD" >> "$OUTPUT"
 echo "CLANG_ANALYZER=$CLANG_ANALYZER" >> "$OUTPUT"
 echo "PROFILING_FLAGS=$PROFILING_FLAGS" >> "$OUTPUT"
+# This will enable some related identifiers for the preprocessor
 if test -n "$JEMALLOC"; then
   echo "JEMALLOC=1" >> "$OUTPUT"
 fi
+# Indicates that jemalloc should be enabled using -ljemalloc flag
+# The alternative is to porvide a direct link to the library via JEMALLOC_LIB
+# and JEMALLOC_INCLUDE
+if test -n "$WITH_JEMALLOC_FLAG"; then
+  echo "WITH_JEMALLOC_FLAG=$WITH_JEMALLOC_FLAG" >> "$OUTPUT"
+fi
 echo "LUA_PATH=$LUA_PATH" >> "$OUTPUT"

build_tools/cont_integration.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 #
 # Copyright (c) 2016, Facebook. All rights reserved.
 #
@@ -67,7 +67,7 @@ function update_repo_status {
 #
 
 # Path to the determinator from the root of the RocksDB repo.
-CONTRUN_DETERMINATOR=./arcanist_util/config/RocksDBCommonHelper.php
+CONTRUN_DETERMINATOR=./build_tools/RocksDBCommonHelper.php
 
 # Value of the previous commit.
 PREV_COMMIT=

build_tools/dependencies.sh

@@ -1,18 +1,18 @@
-GCC_BASE=/mnt/gvfs/third-party2/gcc/2928bb3ed95bf64f5b388ee88c30dc74710c3b35/5.x/centos6-native/f4950a1
-CLANG_BASE=/mnt/gvfs/third-party2/llvm-fb/a5fea028cb7ba43498976e1f8054b0b2e790c295/stable/centos6-native/6aaf4de
-LIBGCC_BASE=/mnt/gvfs/third-party2/libgcc/7a9099f6587ee4378c0b1fa32bb8934019d30ca4/5.x/gcc-5-glibc-2.23/339d858
-GLIBC_BASE=/mnt/gvfs/third-party2/glibc/3b7c6469854dfc7832a1c3cc5b86919a84e5f865/2.23/gcc-5-glibc-2.23/ca1d1c0
-SNAPPY_BASE=/mnt/gvfs/third-party2/snappy/8c38a4c1e52b4c2cc8a9cdc31b9c947ed7dbfcb4/1.1.3/gcc-5-glibc-2.23/9bc6787
-ZLIB_BASE=/mnt/gvfs/third-party2/zlib/d7861abe6f0e27ab98c9303b95a662f0e4cdedb5/1.2.8/gcc-5-glibc-2.23/9bc6787
-BZIP2_BASE=/mnt/gvfs/third-party2/bzip2/740325875f6729f42d28deaa2147b0854f3a347e/1.0.6/gcc-5-glibc-2.23/9bc6787
-LZ4_BASE=/mnt/gvfs/third-party2/lz4/0815d59804160c96caac5f27ca004f51af893dc6/r131/gcc-5-glibc-2.23/9bc6787
-ZSTD_BASE=/mnt/gvfs/third-party2/zstd/c15a4f5f619a2930478d01e2e34dc1e0652b0873/1.1.4/gcc-5-glibc-2.23/03859b5
-GFLAGS_BASE=/mnt/gvfs/third-party2/gflags/f905a5e1032fb30c05db3d3752319857388c0c49/2.2.0/gcc-5-glibc-2.23/9bc6787
-JEMALLOC_BASE=/mnt/gvfs/third-party2/jemalloc/8d60633d822a2a55849c73db24e74a25e52b71db/master/gcc-5-glibc-2.23/1c32b4b
-NUMA_BASE=/mnt/gvfs/third-party2/numa/17c514c4d102a25ca15f4558be564eeed76f4b6a/2.0.8/gcc-5-glibc-2.23/9bc6787
-LIBUNWIND_BASE=/mnt/gvfs/third-party2/libunwind/8db74270cd6d0212ac92d69e7fc7beefe617d772/trunk/gcc-5-glibc-2.23/b1847cb
-TBB_BASE=/mnt/gvfs/third-party2/tbb/9d9a554877d0c5bef330fe818ab7178806dd316a/4.0_update2/gcc-5-glibc-2.23/9bc6787
-KERNEL_HEADERS_BASE=/mnt/gvfs/third-party2/kernel-headers/90c9734afc5579c9d1db529fa788d09f97763b85/4.0.9-36_fbk5_2933_gd092e3f/gcc-5-glibc-2.23/da39a3e
-BINUTILS_BASE=/mnt/gvfs/third-party2/binutils/9e829389ef61b92c62de8748c80169aaf25ce1f0/2.26.1/centos6-native/da39a3e
-VALGRIND_BASE=/mnt/gvfs/third-party2/valgrind/d7f4d4d86674a57668e3a96f76f0e17dd0eb8765/3.11.0/gcc-5-glibc-2.23/9bc6787
-LUA_BASE=/mnt/gvfs/third-party2/lua/61e4abf5813bbc39bc4f548757ccfcadde175a48/5.2.3/gcc-5-glibc-2.23/65372bd
+GCC_BASE=/mnt/gvfs/third-party2/gcc/8219ec1bcedf8ad9da05e121e193364de2cc4f61/5.x/centos6-native/c447969
+CLANG_BASE=/mnt/gvfs/third-party2/llvm-fb/64d8d58e3d84f8bde7a029763d4f5baf39d0d5b9/stable/centos6-native/6aaf4de
+LIBGCC_BASE=/mnt/gvfs/third-party2/libgcc/ba9be983c81de7299b59fe71950c664a84dcb5f8/5.x/gcc-5-glibc-2.23/339d858
+GLIBC_BASE=/mnt/gvfs/third-party2/glibc/f20197cf3d4bd50339c9777aaa0b2ccadad9e2cb/2.23/gcc-5-glibc-2.23/ca1d1c0
+SNAPPY_BASE=/mnt/gvfs/third-party2/snappy/6427ce8c7496e4ab06c2da81543b94c0de8be3d0/1.1.3/gcc-5-glibc-2.23/9bc6787
+ZLIB_BASE=/mnt/gvfs/third-party2/zlib/8f1e8b867d26efef93eac2fabbdb2e1d512665d7/1.2.8/gcc-5-glibc-2.23/9bc6787
+BZIP2_BASE=/mnt/gvfs/third-party2/bzip2/70471c0571559fe0af7db6d7e8860b93a7eadfe1/1.0.6/gcc-5-glibc-2.23/9bc6787
+LZ4_BASE=/mnt/gvfs/third-party2/lz4/453c89d6f0e68cdf1c151c769197fabedad9cac8/r131/gcc-5-glibc-2.23/9bc6787
+ZSTD_BASE=/mnt/gvfs/third-party2/zstd/00a40fa5f8bd2cd0622f2e868552793aef37ccf4/1.3.0/gcc-5-glibc-2.23/03859b5
+GFLAGS_BASE=/mnt/gvfs/third-party2/gflags/47eef08f9acb77de982fbda6047c26d330739538/2.2.0/gcc-5-glibc-2.23/9bc6787
+JEMALLOC_BASE=/mnt/gvfs/third-party2/jemalloc/4414ddc78df8008b35cc4adac23590ad29148584/master/gcc-5-glibc-2.23/d506c82
+NUMA_BASE=/mnt/gvfs/third-party2/numa/9d7ae2693d05d62f9a579cb21e6b717cf257a75d/2.0.11/gcc-5-glibc-2.23/9bc6787
+LIBUNWIND_BASE=/mnt/gvfs/third-party2/libunwind/2b2dd58e3a52ccf2c1d827def59e5f740de0ad15/1.2/gcc-5-glibc-2.23/b443de1
+TBB_BASE=/mnt/gvfs/third-party2/tbb/379addf7ab2468a2b4293b47456cfcd1c9cb318d/4.3/gcc-5-glibc-2.23/9bc6787
+KERNEL_HEADERS_BASE=/mnt/gvfs/third-party2/kernel-headers/3f68f5fe65a85b7c2d3e66852268fbd1efdb3151/4.0.9-36_fbk5_2933_gd092e3f/gcc-5-glibc-2.23/da39a3e
+BINUTILS_BASE=/mnt/gvfs/third-party2/binutils/b9fab0aec99d9c36408e810b2677e91c12807afd/2.28/centos6-native/da39a3e
+VALGRIND_BASE=/mnt/gvfs/third-party2/valgrind/423431d61786b20bcc3bde8972901130cb29e6b3/3.11.0/gcc-5-glibc-2.23/9bc6787
+LUA_BASE=/mnt/gvfs/third-party2/lua/3b0bb3bd9a0f690a069c479fcc0f7424fc7456d2/5.2.3/gcc-5-glibc-2.23/65372bd

build_tools/dependencies_platform007.sh

@@ -1,18 +0,0 @@
-GCC_BASE=/mnt/gvfs/third-party2/gcc/6e8e715624fd15256a7970073387793dfcf79b46/7.x/centos7-native/b2ef2b6
-CLANG_BASE=/mnt/gvfs/third-party2/llvm-fb/ef37e1faa1c29782abfac1ae65a291b9b7966f6d/stable/centos7-native/c9f9104
-LIBGCC_BASE=/mnt/gvfs/third-party2/libgcc/c67031f0f739ac61575a061518d6ef5038f99f90/7.x/platform007/5620abc
-GLIBC_BASE=/mnt/gvfs/third-party2/glibc/60d6f124a78798b73944f5ba87c2306ae3460153/2.26/platform007/f259413
-SNAPPY_BASE=/mnt/gvfs/third-party2/snappy/7f9bdaada18f59bc27ec2b0871eb8a6144343aef/1.1.3/platform007/ca4da3d
-ZLIB_BASE=/mnt/gvfs/third-party2/zlib/22c2d65676fb7c23cfa797c4f6937f38b026f3cf/1.2.8/platform007/ca4da3d
-BZIP2_BASE=/mnt/gvfs/third-party2/bzip2/dc49a21c5fceec6456a7a28a94dcd16690af1337/1.0.6/platform007/ca4da3d
-LZ4_BASE=/mnt/gvfs/third-party2/lz4/907b498203d297947f3bb70b9466f47e100f1873/r131/platform007/ca4da3d
-ZSTD_BASE=/mnt/gvfs/third-party2/zstd/3ee276cbacfad3074e3f07bf826ac47f06970f4e/1.3.5/platform007/15a3614
-GFLAGS_BASE=/mnt/gvfs/third-party2/gflags/0b9929d2588991c65a57168bf88aff2db87c5d48/2.2.0/platform007/ca4da3d
-JEMALLOC_BASE=/mnt/gvfs/third-party2/jemalloc/9c910d36d6235cc40e8ff559358f1833452300ca/master/platform007/5b0f53e
-NUMA_BASE=/mnt/gvfs/third-party2/numa/9cbf2460284c669ed19c3ccb200a71f7dd7e53c7/2.0.11/platform007/ca4da3d
-LIBUNWIND_BASE=/mnt/gvfs/third-party2/libunwind/bf3d7497fe4e6d007354f0adffa16ce3003f8338/1.3/platform007/6f3e0a9
-TBB_BASE=/mnt/gvfs/third-party2/tbb/ff4e0b093534704d8abab678a4fd7f5ea7b094c7/2018_U5/platform007/ca4da3d
-KERNEL_HEADERS_BASE=/mnt/gvfs/third-party2/kernel-headers/b5c4a61a5c483ba24722005ae07895971a2ac707/fb/platform007/da39a3e
-BINUTILS_BASE=/mnt/gvfs/third-party2/binutils/92ff90349e2f43ea0a8246d8b1cf17b6869013e3/2.29.1/centos7-native/da39a3e
-VALGRIND_BASE=/mnt/gvfs/third-party2/valgrind/f3f697a28122e6bcd513273dd9c1ff23852fc59f/3.13.0/platform007/ca4da3d
-LUA_BASE=/mnt/gvfs/third-party2/lua/f0cd714433206d5139df61659eb7b28b1dea6683/5.3.4/platform007/5007832

build_tools/dockerbuild.sh

@@ -1,2 +1,2 @@
-#!/bin/bash
+#!/usr/bin/env bash
 docker run -v $PWD:/rocks -w /rocks buildpack-deps make

build_tools/fbcode_config.sh

@@ -83,6 +83,7 @@ CFLAGS+=" -DTBB"
 
 # use Intel SSE support for checksum calculations
 export USE_SSE=1
+export PORTABLE=1
 
 BINUTILS="$BINUTILS_BASE/bin"
 AR="$BINUTILS/ar"

build_tools/fbcode_config4.8.1.sh

@@ -54,6 +54,7 @@ TBB_LIBS="$TBB_BASE/lib/libtbb.a"
 
 # use Intel SSE support for checksum calculations
 export USE_SSE=1
+export PORTABLE=1
 
 BINUTILS="$BINUTILS_BASE/bin"
 AR="$BINUTILS/ar"

build_tools/fbcode_config_platform007.sh

@@ -1,157 +0,0 @@
-#!/bin/sh
-#
-# Set environment variables so that we can compile rocksdb using
-# fbcode settings.  It uses the latest g++ and clang compilers and also
-# uses jemalloc
-# Environment variables that change the behavior of this script:
-# PIC_BUILD -- if true, it will only take pic versions of libraries from fbcode. libraries that don't have pic variant will not be included
-
-
-BASEDIR=`dirname $BASH_SOURCE`
-source "$BASEDIR/dependencies_platform007.sh"
-
-CFLAGS=""
-
-# libgcc
-LIBGCC_INCLUDE="$LIBGCC_BASE/include/c++/7.3.0"
-LIBGCC_LIBS=" -L $LIBGCC_BASE/lib"
-
-# glibc
-GLIBC_INCLUDE="$GLIBC_BASE/include"
-GLIBC_LIBS=" -L $GLIBC_BASE/lib"
-
-# snappy
-SNAPPY_INCLUDE=" -I $SNAPPY_BASE/include/"
-if test -z $PIC_BUILD; then
-  SNAPPY_LIBS=" $SNAPPY_BASE/lib/libsnappy.a"
-else
-  SNAPPY_LIBS=" $SNAPPY_BASE/lib/libsnappy_pic.a"
-fi
-CFLAGS+=" -DSNAPPY"
-
-if test -z $PIC_BUILD; then
-  # location of zlib headers and libraries
-  ZLIB_INCLUDE=" -I $ZLIB_BASE/include/"
-  ZLIB_LIBS=" $ZLIB_BASE/lib/libz.a"
-  CFLAGS+=" -DZLIB"
-
-  # location of bzip headers and libraries
-  BZIP_INCLUDE=" -I $BZIP2_BASE/include/"
-  BZIP_LIBS=" $BZIP2_BASE/lib/libbz2.a"
-  CFLAGS+=" -DBZIP2"
-
-  LZ4_INCLUDE=" -I $LZ4_BASE/include/"
-  LZ4_LIBS=" $LZ4_BASE/lib/liblz4.a"
-  CFLAGS+=" -DLZ4"
-fi
-
-ZSTD_INCLUDE=" -I $ZSTD_BASE/include/"
-if test -z $PIC_BUILD; then
-  ZSTD_LIBS=" $ZSTD_BASE/lib/libzstd.a"
-else
-  ZSTD_LIBS=" $ZSTD_BASE/lib/libzstd_pic.a"
-fi
-CFLAGS+=" -DZSTD"
-
-# location of gflags headers and libraries
-GFLAGS_INCLUDE=" -I $GFLAGS_BASE/include/"
-if test -z $PIC_BUILD; then
-  GFLAGS_LIBS=" $GFLAGS_BASE/lib/libgflags.a"
-else
-  GFLAGS_LIBS=" $GFLAGS_BASE/lib/libgflags_pic.a"
-fi
-CFLAGS+=" -DGFLAGS=gflags"
-
-# location of jemalloc
-JEMALLOC_INCLUDE=" -I $JEMALLOC_BASE/include/"
-JEMALLOC_LIB=" $JEMALLOC_BASE/lib/libjemalloc.a"
-
-if test -z $PIC_BUILD; then
-  # location of numa
-  NUMA_INCLUDE=" -I $NUMA_BASE/include/"
-  NUMA_LIB=" $NUMA_BASE/lib/libnuma.a"
-  CFLAGS+=" -DNUMA"
-
-  # location of libunwind
-  LIBUNWIND="$LIBUNWIND_BASE/lib/libunwind.a"
-fi
-
-# location of TBB
-TBB_INCLUDE=" -isystem $TBB_BASE/include/"
-if test -z $PIC_BUILD; then
-  TBB_LIBS="$TBB_BASE/lib/libtbb.a"
-else
-  TBB_LIBS="$TBB_BASE/lib/libtbb_pic.a"
-fi
-CFLAGS+=" -DTBB"
-
-# use Intel SSE support for checksum calculations
-export USE_SSE=1
-export PORTABLE=1
-
-BINUTILS="$BINUTILS_BASE/bin"
-AR="$BINUTILS/ar"
-
-DEPS_INCLUDE="$SNAPPY_INCLUDE $ZLIB_INCLUDE $BZIP_INCLUDE $LZ4_INCLUDE $ZSTD_INCLUDE $GFLAGS_INCLUDE $NUMA_INCLUDE $TBB_INCLUDE"
-
-STDLIBS="-L $GCC_BASE/lib64"
-
-CLANG_BIN="$CLANG_BASE/bin"
-CLANG_LIB="$CLANG_BASE/lib"
-CLANG_SRC="$CLANG_BASE/../../src"
-
-CLANG_ANALYZER="$CLANG_BIN/clang++"
-CLANG_SCAN_BUILD="$CLANG_SRC/llvm/tools/clang/tools/scan-build/bin/scan-build"
-
-if [ -z "$USE_CLANG" ]; then
-  # gcc
-  CC="$GCC_BASE/bin/gcc"
-  CXX="$GCC_BASE/bin/g++"
-
-  CFLAGS+=" -B$BINUTILS/gold"
-  CFLAGS+=" -isystem $LIBGCC_INCLUDE"
-  CFLAGS+=" -isystem $GLIBC_INCLUDE"
-  JEMALLOC=1
-else
-  # clang
-  CLANG_INCLUDE="$CLANG_LIB/clang/stable/include"
-  CC="$CLANG_BIN/clang"
-  CXX="$CLANG_BIN/clang++"
-
-  KERNEL_HEADERS_INCLUDE="$KERNEL_HEADERS_BASE/include"
-
-  CFLAGS+=" -B$BINUTILS/gold -nostdinc -nostdlib"
-  CFLAGS+=" -isystem $LIBGCC_BASE/include/c++/7.x "
-  CFLAGS+=" -isystem $LIBGCC_BASE/include/c++/7.x/x86_64-facebook-linux "
-  CFLAGS+=" -isystem $GLIBC_INCLUDE"
-  CFLAGS+=" -isystem $LIBGCC_INCLUDE"
-  CFLAGS+=" -isystem $CLANG_INCLUDE"
-  CFLAGS+=" -isystem $KERNEL_HEADERS_INCLUDE/linux "
-  CFLAGS+=" -isystem $KERNEL_HEADERS_INCLUDE "
-  CFLAGS+=" -Wno-expansion-to-defined "
-  CXXFLAGS="-nostdinc++"
-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"
-CXXFLAGS+=" $CFLAGS"
-
-EXEC_LDFLAGS=" $SNAPPY_LIBS $ZLIB_LIBS $BZIP_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS $NUMA_LIB $TBB_LIBS"
-EXEC_LDFLAGS+=" -B$BINUTILS/gold"
-EXEC_LDFLAGS+=" -Wl,--dynamic-linker,/usr/local/fbcode/platform007/lib/ld.so"
-EXEC_LDFLAGS+=" $LIBUNWIND"
-EXEC_LDFLAGS+=" -Wl,-rpath=/usr/local/fbcode/platform007/lib"
-# required by libtbb
-EXEC_LDFLAGS+=" -ldl"
-
-PLATFORM_LDFLAGS="$LIBGCC_LIBS $GLIBC_LIBS $STDLIBS -lgcc -lstdc++"
-
-EXEC_LDFLAGS_SHARED="$SNAPPY_LIBS $ZLIB_LIBS $BZIP_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS $TBB_LIBS"
-
-VALGRIND_VER="$VALGRIND_BASE/bin/"
-
-# lua not supported because it's on track for deprecation, I think
-LUA_PATH=
-LUA_LIB=
-
-export CC CXX AR CFLAGS CXXFLAGS EXEC_LDFLAGS EXEC_LDFLAGS_SHARED VALGRIND_VER JEMALLOC_LIB JEMALLOC_INCLUDE CLANG_ANALYZER CLANG_SCAN_BUILD LUA_PATH LUA_LIB

build_tools/format-diff.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 # If clang_format_diff.py command is not specfied, we assume we are able to
 # access directly without any path.
 if [ -z $CLANG_FORMAT_DIFF ]

build_tools/make_new_version.sh

@@ -1,53 +0,0 @@
-#!/bin/bash
-#  Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
-#  This source code is licensed under the BSD-style license found in the
-#  LICENSE file in the root directory of this source tree. An additional grant
-#  of patent rights can be found in the PATENTS file in the same directory.
-
-set -e
-if [ -z "$GIT" ]
-then
-  GIT="git"
-fi
-
-# Print out the colored progress info so that it can be brainlessly
-# distinguished by users.
-function title() {
-  echo -e "\033[1;32m$*\033[0m"
-}
-
-usage="Create new RocksDB version and prepare it for the release process\n"
-usage+="USAGE: ./make_new_version.sh <version> [<remote>]\n"
-usage+="  version: specify a version without '.fb' suffix (e.g. 5.4).\n"
-usage+="  remote: name of the remote to push the branch to (default: origin)."
-
-# -- Pre-check
-if [[ $# < 1 ]]; then
-  echo -e $usage
-  exit 1
-fi
-
-ROCKSDB_VERSION=$1
-
-REMOTE="origin"
-if [[ $# > 1 ]]; then
-  REMOTE=$2
-fi
-
-GIT_BRANCH=`git rev-parse --abbrev-ref HEAD`
-echo $GIT_BRANCH
-
-if [ $GIT_BRANCH != "master" ]; then
-  echo "Error: Current branch is '$GIT_BRANCH', Please switch to master branch."
-  exit 1
-fi
-
-title "Adding new tag for this release ..."
-BRANCH="$ROCKSDB_VERSION.fb"
-$GIT checkout -b $BRANCH
-
-# Setting up the proxy for remote repo access
-title "Pushing new branch to remote repo ..."
-git push $REMOTE --set-upstream $BRANCH
-
-title "Branch $BRANCH is pushed to github;"

build_tools/regression_build_test.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 
 set -e
 

build_tools/rocksdb-lego-determinator

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 # This script is executed by Sandcastle
 # to determine next steps to run
 
@@ -85,9 +85,8 @@ NON_SHM="TMPD=/tmp/rocksdb_test_tmp"
 GCC_481="ROCKSDB_FBCODE_BUILD_WITH_481=1"
 ASAN="COMPILE_WITH_ASAN=1"
 CLANG="USE_CLANG=1"
-# in gcc-5 there are known problems with TSAN like https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71090.
-# using platform007 gives us gcc-8 or higher which has that bug fixed.
-TSAN="ROCKSDB_FBCODE_BUILD_WITH_PLATFORM007=1 COMPILE_WITH_TSAN=1"
+LITE="OPT=\"-DROCKSDB_LITE -g\""
+TSAN="COMPILE_WITH_TSAN=1"
 UBSAN="COMPILE_WITH_UBSAN=1"
 DISABLE_JEMALLOC="DISABLE_JEMALLOC=1"
 HTTP_PROXY="https_proxy=http://fwdproxy.29.prn1:8080 http_proxy=http://fwdproxy.29.prn1:8080 ftp_proxy=http://fwdproxy.29.prn1:8080"
@@ -344,27 +343,7 @@ LITE_BUILD_COMMANDS="[
             $CLEANUP_ENV,
             {
                 'name':'Build RocksDB debug version',
-                'shell':'$LITE make J=1 static_lib || $CONTRUN_NAME=lite_static_lib $TASK_CREATION_TOOL',
-                'user':'root',
-                $PARSER
-            },
-        ],
-        $REPORT
-    }
-]"
-
-#
-# RocksDB lite tests
-#
-LITE_UNIT_TEST_COMMANDS="[
-    {
-        'name':'Rocksdb Lite Unit Test',
-        'oncall':'$ONCALL',
-        'steps': [
-            $CLEANUP_ENV,
-            {
-                'name':'Build RocksDB debug version',
-                'shell':'$SHM $LITE make J=1 check || $CONTRUN_NAME=lite_check $TASK_CREATION_TOOL',
+                'shell':'$LITE make J=1 all check || $CONTRUN_NAME=lite $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -565,7 +544,7 @@ TSAN_CRASH_TEST_COMMANDS="[
             {
                 'name':'Compile and run',
                 'timeout': 86400,
-                'shell':'set -o pipefail && $SHM $DEBUG $TSAN CRASH_TEST_KILL_ODD=1887 make J=1 crash_test || $CONTRUN_NAME=tsan_crash_test $TASK_CREATION_TOOL',
+                'shell':'set -o pipefail && $SHM $DEBUG $TSAN CRASH_TEST_KILL_ODD=1887 CRASH_TEST_EXT_ARGS=--log2_keys_per_lock=22  make J=1 crash_test || $CONTRUN_NAME=tsan_crash_test $TASK_CREATION_TOOL',
                 'user':'root',
                 $PARSER
             },
@@ -584,39 +563,7 @@ run_format_compatible()
   rm -rf /dev/shm/rocksdb
   mkdir /dev/shm/rocksdb
 
-  echo '
-  if [ -e "build_tools/build_detect_platform" ]
-  then
-    sed "s/tcmalloc/nothingnothingnothing/g" build_tools/build_detect_platform > $TEST_TMPDIR/temp_build_file
-    rm -rf build_tools/build_detect_platform
-    cp $TEST_TMPDIR/temp_build_file build_tools/build_detect_platform
-    chmod +x build_tools/build_detect_platform
-  fi
-
-  if [ -e "build_detect_platform" ]
-  then
-    sed "s/tcmalloc/nothingnothingnothing/g" build_detect_platform > $TEST_TMPDIR/temp_build_file
-    rm -rf build_detect_platform
-    cp $TEST_TMPDIR/temp_build_file build_detect_platform
-    chmod +x build_detect_platform
-  fi
-
-  make ldb -j32
-
-  if [ -e "build_detect_platform" ]
-  then
-    git checkout -- build_detect_platform
-  fi
-
-  if [ -e "build_tools/build_detect_platform" ]
-  then
-    git checkout -- build_tools/build_detect_platform
-  fi
-  ' > temp_build_ldb.sh
-
-  sed "s/make ldb -j32/source temp_build_ldb.sh/g" tools/check_format_compatible.sh > tools/temp_check_format_compatible.sh
-  chmod +x tools/temp_check_format_compatible.sh
-  tools/temp_check_format_compatible.sh
+  tools/check_format_compatible.sh
 }
 
 FORMAT_COMPATIBLE_COMMANDS="[
@@ -781,9 +728,6 @@ case $1 in
   lite)
     echo $LITE_BUILD_COMMANDS
     ;;
-  lite_test)
-    echo $LITE_UNIT_TEST_COMMANDS
-    ;;
   stress_crash)
     echo $STRESS_CRASH_TEST_COMMANDS
     ;;

build_tools/run_ci_db_test.ps1

@@ -1,13 +1,16 @@
 # This script enables you running RocksDB tests by running
 # All the tests concurrently and utilizing all the cores
 Param(
-  [switch]$EnableJE = $false,  # Look for and use _je executable, append _je to listed exclusions
+  [switch]$EnableJE = $false,  # Look for and use test executable, append _je to listed exclusions
   [switch]$RunAll = $false,    # Will attempt discover all *_test[_je].exe binaries and run all
                                # of them as Google suites. I.e. It will run test cases concurrently
                                # except those mentioned as $Run, those will run as individual test cases
                                # And any execlued with $ExcludeExes or $ExcludeCases
                                # It will also not run any individual test cases
                                # excluded but $ExcludeCasese
+  [switch]$RunAllExe = $false, # Look for and use test exdcutables, append _je to exclusions automatically
+                               # It will attempt to run them in parallel w/o breaking them up on individual
+                               # test cases. Those listed with $ExcludeExes will be excluded
   [string]$SuiteRun = "",      # Split test suites in test cases and run in parallel, not compatible with $RunAll
   [string]$Run = "",           # Run specified executables in parallel but do not split to test cases
   [string]$ExcludeCases = "",  # Exclude test cases, expects a comma separated list, no spaces
@@ -39,13 +42,18 @@ $RunOnly.Add("compact_on_deletion_collector_test") | Out-Null
 $RunOnly.Add("merge_test") | Out-Null
 $RunOnly.Add("stringappend_test") | Out-Null # Apparently incorrectly written
 $RunOnly.Add("backupable_db_test") | Out-Null # Disabled
-
+$RunOnly.Add("timer_queue_test") | Out-Null # Not a gtest
 
 if($RunAll -and $SuiteRun -ne "") {
     Write-Error "$RunAll and $SuiteRun are not compatible"
     exit 1
 }
 
+if($RunAllExe -and $Run -ne "") {
+    Write-Error "$RunAllExe and $Run are not compatible"
+    exit 1
+}
+
 # If running under Appveyor assume that root
 [string]$Appveyor = $Env:APPVEYOR_BUILD_FOLDER
 if($Appveyor -ne "") {
@@ -131,12 +139,8 @@ function ExtractTestCases([string]$GTestExe, $HashTable) {
 
       # Leading whitespace is fine
       $l = $l -replace '^\s+',''
-      # but no whitespace any other place
-      if($l -match "\s+") {
-        continue
-      }
       # Trailing dot is a test group but no whitespace
-      elseif ( $l -match "\.$" ) {
+      if ($l -match "\.$" -and $l -notmatch "\s+") {
         $Group = $l
       }  else {
         # Otherwise it is a test name, remove leading space
@@ -223,13 +227,11 @@ $TestExes = [ordered]@{}
 if($Run -ne "") {
 
   $test_list = $Run -split ' '
-
   ForEach($t in $test_list) {
 
     if($EnableJE) {
       $t += "_je"
     }
-
     MakeAndAdd -token $t -HashTable $TestExes
   }
 
@@ -237,6 +239,38 @@ if($Run -ne "") {
      Write-Error "Failed to extract tests from $Run"
      exit 1
   }
+} elseif($RunAllExe) {
+  # Discover all the test binaries
+  if($EnableJE) {
+    $pattern = "*_test_je.exe"
+  } else {
+    $pattern = "*_test.exe"
+  }
+
+  $search_path = -join ($BinariesFolder, $pattern)
+  Write-Host "Binaries Search Path: $search_path"
+
+  $DiscoveredExe = @()
+  dir -Path $search_path | ForEach-Object {
+     $DiscoveredExe += ($_.Name)     
+  }
+
+  # Remove exclusions
+  ForEach($e in $DiscoveredExe) {
+    $e = $e -replace '.exe$', ''
+    $bare_name = $e -replace '_je$', ''
+
+    if($ExcludeExesSet.Contains($bare_name)) {
+      Write-Warning "Test $e is excluded"
+      continue
+    }
+    MakeAndAdd -token $e -HashTable $TestExes
+  }
+
+  if($TestExes.Count -lt 1) {
+     Write-Error "Failed to discover test executables"
+     exit 1
+  }
 }
 
 # Ordered by exe @{ Exe = @{ TestCase = LogName }}
@@ -245,9 +279,7 @@ $CasesToRun = [ordered]@{}
 if($SuiteRun -ne "") {
   $suite_list = $SuiteRun -split ' '
   ProcessSuites -ListOfSuites $suite_list -HashOfHashes $CasesToRun
-}
-
-if($RunAll) {
+} elseif ($RunAll) {
 # Discover all the test binaries
   if($EnableJE) {
     $pattern = "*_test_je.exe"
@@ -255,7 +287,6 @@ if($RunAll) {
     $pattern = "*_test.exe"
   }
 
-
   $search_path = -join ($BinariesFolder, $pattern)
   Write-Host "Binaries Search Path: $search_path"
 
@@ -287,8 +318,6 @@ if($RunAll) {
 }
 
 
-Write-Host "Attempting to start: $NumTestsToStart tests"
-
 # Invoke a test with a filter and redirect all output
 $InvokeTestCase = {
     param($exe, $test, $log);
@@ -365,6 +394,7 @@ function RunJobs($Suites, $TestCmds, [int]$ConcurrencyVal)
                  break
                }
 
+              Write-Host "Starting $exe_name"
               [string]$Exe =  -Join ($BinariesFolder, $exe_name)
               $job = Start-Job -Name $exe_name -ScriptBlock $InvokeTestAsync -ArgumentList @($Exe,$log_path)
               $JobToLog.Add($job, $log_path)

build_tools/update_dependencies.sh

@@ -38,6 +38,7 @@ function get_lib_base()
     # platform is not provided, use latest gcc
     result=`ls -dr1v $result/gcc-*[^fb]/ | head -n1`
   else
+    echo $lib_platform
     result="$result/$lib_platform/"
   fi
   
@@ -52,45 +53,6 @@ function get_lib_base()
   log_variable $__res_var
 }
 
-###########################################################
-#                platform007 dependencies                 #
-###########################################################
-
-OUTPUT="$BASEDIR/dependencies_platform007.sh"
-
-rm -f "$OUTPUT"
-touch "$OUTPUT"
-
-echo "Writing dependencies to $OUTPUT"
-
-# Compilers locations
-GCC_BASE=`readlink -f $TP2_LATEST/gcc/7.x/centos7-native/*/`
-CLANG_BASE=`readlink -f $TP2_LATEST/llvm-fb/stable/centos7-native/*/`
-
-log_variable GCC_BASE
-log_variable CLANG_BASE
-
-# Libraries locations
-get_lib_base libgcc     7.x     platform007
-get_lib_base glibc      2.26    platform007
-get_lib_base snappy     LATEST  platform007
-get_lib_base zlib       LATEST  platform007
-get_lib_base bzip2      LATEST  platform007
-get_lib_base lz4        LATEST  platform007
-get_lib_base zstd       LATEST  platform007
-get_lib_base gflags     LATEST  platform007
-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 kernel-headers fb platform007
-get_lib_base binutils   LATEST centos7-native
-get_lib_base valgrind   LATEST platform007
-get_lib_base lua        5.3.4  platform007
-
-git diff $OUTPUT
-
 ###########################################################
 #                   5.x dependencies                      #
 ###########################################################
@@ -110,22 +72,22 @@ log_variable GCC_BASE
 log_variable CLANG_BASE
 
 # Libraries locations
-get_lib_base libgcc     5.x
-get_lib_base glibc      2.23
-get_lib_base snappy     LATEST gcc-5-glibc-2.23
-get_lib_base zlib       LATEST
-get_lib_base bzip2      LATEST
-get_lib_base lz4        LATEST
-get_lib_base zstd       LATEST
-get_lib_base gflags     LATEST
-get_lib_base jemalloc   LATEST
-get_lib_base numa       LATEST
-get_lib_base libunwind  LATEST
-get_lib_base tbb        4.0_update2 gcc-5-glibc-2.23
+get_lib_base libgcc     5.x     gcc-5-glibc-2.23
+get_lib_base glibc      2.23    gcc-5-glibc-2.23
+get_lib_base snappy     LATEST  gcc-5-glibc-2.23
+get_lib_base zlib       LATEST  gcc-5-glibc-2.23
+get_lib_base bzip2      LATEST  gcc-5-glibc-2.23
+get_lib_base lz4        LATEST  gcc-5-glibc-2.23
+get_lib_base zstd       LATEST  gcc-5-glibc-2.23
+get_lib_base gflags     LATEST  gcc-5-glibc-2.23
+get_lib_base jemalloc   LATEST  gcc-5-glibc-2.23
+get_lib_base numa       LATEST  gcc-5-glibc-2.23
+get_lib_base libunwind  LATEST  gcc-5-glibc-2.23
+get_lib_base tbb        LATEST  gcc-5-glibc-2.23
 
-get_lib_base kernel-headers LATEST 
+get_lib_base kernel-headers 4.0.9-36_fbk5_2933_gd092e3f gcc-5-glibc-2.23
 get_lib_base binutils   LATEST centos6-native 
-get_lib_base valgrind   3.10.0 gcc-5-glibc-2.23
+get_lib_base valgrind   LATEST gcc-5-glibc-2.23
 get_lib_base lua        5.2.3 gcc-5-glibc-2.23
 
 git diff $OUTPUT

cache/cache_bench.cc

@@ -17,16 +17,16 @@ int main() {
 #include <inttypes.h>
 #include <sys/types.h>
 #include <stdio.h>
-#include <gflags/gflags.h>
 
-#include "rocksdb/db.h"
-#include "rocksdb/cache.h"
-#include "rocksdb/env.h"
 #include "port/port.h"
+#include "rocksdb/cache.h"
+#include "rocksdb/db.h"
+#include "rocksdb/env.h"
+#include "util/gflags_compat.h"
 #include "util/mutexlock.h"
 #include "util/random.h"
 
-using GFLAGS::ParseCommandLineFlags;
+using GFLAGS_NAMESPACE::ParseCommandLineFlags;
 
 static const uint32_t KB = 1024;
 

cache/clock_cache.cc

@@ -27,6 +27,11 @@ std::shared_ptr<Cache> NewClockCache(size_t capacity, int num_shard_bits,
 #include <atomic>
 #include <deque>
 
+// "tbb/concurrent_hash_map.h" requires RTTI if exception is enabled.
+// Disable it so users can chooose to disable RTTI.
+#ifndef ROCKSDB_USE_RTTI
+#define TBB_USE_EXCEPTIONS 0
+#endif
 #include "tbb/concurrent_hash_map.h"
 
 #include "cache/sharded_cache.h"

cache/lru_cache.cc

@@ -22,7 +22,7 @@
 
 namespace rocksdb {
 
-LRUHandleTable::LRUHandleTable() : length_(0), elems_(0), list_(nullptr) {
+LRUHandleTable::LRUHandleTable() : list_(nullptr), length_(0), elems_(0) {
   Resize();
 }
 
@@ -100,7 +100,9 @@ void LRUHandleTable::Resize() {
 }
 
 LRUCacheShard::LRUCacheShard()
-    : usage_(0), lru_usage_(0), high_pri_pool_usage_(0) {
+    : capacity_(0), high_pri_pool_usage_(0), strict_capacity_limit_(false),
+      high_pri_pool_ratio_(0), high_pri_pool_capacity_(0), usage_(0),
+      lru_usage_(0) {
   // Make empty circular linked list
   lru_.next = &lru_;
   lru_.prev = &lru_;
@@ -157,6 +159,21 @@ void LRUCacheShard::TEST_GetLRUList(LRUHandle** lru, LRUHandle** lru_low_pri) {
   *lru_low_pri = lru_low_pri_;
 }
 
+size_t LRUCacheShard::TEST_GetLRUSize() {
+  LRUHandle* lru_handle = lru_.next;
+  size_t lru_size = 0;
+  while (lru_handle != &lru_) {
+    lru_size++;
+    lru_handle = lru_handle->next;
+  }
+  return lru_size;
+}
+
+double LRUCacheShard::GetHighPriPoolRatio() {
+  MutexLock l(&mutex_);
+  return high_pri_pool_ratio_;
+}
+
 void LRUCacheShard::LRU_Remove(LRUHandle* e) {
   assert(e->next != nullptr);
   assert(e->prev != nullptr);
@@ -223,6 +240,22 @@ void LRUCacheShard::EvictFromLRU(size_t charge,
   }
 }
 
+void* LRUCacheShard::operator new(size_t size) {
+  return port::cacheline_aligned_alloc(size);
+}
+
+void* LRUCacheShard::operator new[](size_t size) {
+  return port::cacheline_aligned_alloc(size);
+}
+
+void LRUCacheShard::operator delete(void *memblock) {
+  port::cacheline_aligned_free(memblock);
+}
+
+void LRUCacheShard::operator delete[](void* memblock) {
+  port::cacheline_aligned_free(memblock);
+}
+
 void LRUCacheShard::SetCapacity(size_t capacity) {
   autovector<LRUHandle*> last_reference_list;
   {
@@ -327,6 +360,7 @@ Status LRUCacheShard::Insert(const Slice& key, uint32_t hash, void* value,
   e->deleter = deleter;
   e->charge = charge;
   e->key_length = key.size();
+  e->flags = 0;
   e->hash = hash;
   e->refs = (handle == nullptr
                  ? 1
@@ -438,11 +472,11 @@ std::string LRUCacheShard::GetPrintableOptions() const {
 LRUCache::LRUCache(size_t capacity, int num_shard_bits,
                    bool strict_capacity_limit, double high_pri_pool_ratio)
     : ShardedCache(capacity, num_shard_bits, strict_capacity_limit) {
-  int num_shards = 1 << num_shard_bits;
-  shards_ = new LRUCacheShard[num_shards];
+  num_shards_ = 1 << num_shard_bits;
+  shards_ = new LRUCacheShard[num_shards_];
   SetCapacity(capacity);
   SetStrictCapacityLimit(strict_capacity_limit);
-  for (int i = 0; i < num_shards; i++) {
+  for (int i = 0; i < num_shards_; i++) {
     shards_[i].SetHighPriorityPoolRatio(high_pri_pool_ratio);
   }
 }
@@ -469,7 +503,34 @@ uint32_t LRUCache::GetHash(Handle* handle) const {
   return reinterpret_cast<const LRUHandle*>(handle)->hash;
 }
 
-void LRUCache::DisownData() { shards_ = nullptr; }
+void LRUCache::DisownData() {
+// Do not drop data if compile with ASAN to suppress leak warning.
+#ifndef __SANITIZE_ADDRESS__
+  shards_ = nullptr;
+#endif  // !__SANITIZE_ADDRESS__
+}
+
+size_t LRUCache::TEST_GetLRUSize() {
+  size_t lru_size_of_all_shards = 0;
+  for (int i = 0; i < num_shards_; i++) {
+    lru_size_of_all_shards += shards_[i].TEST_GetLRUSize();
+  }
+  return lru_size_of_all_shards;
+}
+
+double LRUCache::GetHighPriPoolRatio() {
+  double result = 0.0;
+  if (num_shards_ > 0) {
+    result = shards_[0].GetHighPriPoolRatio();
+  }
+  return result;
+}
+
+std::shared_ptr<Cache> NewLRUCache(const LRUCacheOptions& cache_opts) {
+  return NewLRUCache(cache_opts.capacity, cache_opts.num_shard_bits,
+                     cache_opts.strict_capacity_limit,
+                     cache_opts.high_pri_pool_ratio);
+}
 
 std::shared_ptr<Cache> NewLRUCache(size_t capacity, int num_shard_bits,
                                    bool strict_capacity_limit,

cache/lru_cache.h

@@ -148,13 +148,13 @@ class LRUHandleTable {
 
   // The table consists of an array of buckets where each bucket is
   // a linked list of cache entries that hash into the bucket.
+  LRUHandle** list_;
   uint32_t length_;
   uint32_t elems_;
-  LRUHandle** list_;
 };
 
 // A single shard of sharded cache.
-class LRUCacheShard : public CacheShard {
+class ALIGN_AS(CACHE_LINE_SIZE) LRUCacheShard : public CacheShard {
  public:
   LRUCacheShard();
   virtual ~LRUCacheShard();
@@ -198,6 +198,22 @@ class LRUCacheShard : public CacheShard {
 
   void TEST_GetLRUList(LRUHandle** lru, LRUHandle** lru_low_pri);
 
+  //  Retrieves number of elements in LRU, for unit test purpose only
+  //  not threadsafe
+  size_t TEST_GetLRUSize();
+
+  //  Retrives high pri pool ratio
+  double GetHighPriPoolRatio();
+
+  // Overloading to aligned it to cache line size
+  void* operator new(size_t);
+
+  void* operator new[](size_t);
+
+  void operator delete(void *);
+
+  void operator delete[](void*);
+
  private:
   void LRU_Remove(LRUHandle* e);
   void LRU_Insert(LRUHandle* e);
@@ -219,12 +235,6 @@ class LRUCacheShard : public CacheShard {
   // Initialized before use.
   size_t capacity_;
 
-  // Memory size for entries residing in the cache
-  size_t usage_;
-
-  // Memory size for entries residing only in the LRU list
-  size_t lru_usage_;
-
   // Memory size for entries in high-pri pool.
   size_t high_pri_pool_usage_;
 
@@ -238,11 +248,6 @@ class LRUCacheShard : public CacheShard {
   // Remember the value to avoid recomputing each time.
   double high_pri_pool_capacity_;
 
-  // mutex_ protects the following state.
-  // We don't count mutex_ as the cache's internal state so semantically we
-  // don't mind mutex_ invoking the non-const actions.
-  mutable port::Mutex mutex_;
-
   // Dummy head of LRU list.
   // lru.prev is newest entry, lru.next is oldest entry.
   // LRU contains items which can be evicted, ie reference only by cache
@@ -251,7 +256,29 @@ class LRUCacheShard : public CacheShard {
   // Pointer to head of low-pri pool in LRU list.
   LRUHandle* lru_low_pri_;
 
+  // ------------^^^^^^^^^^^^^-----------
+  // Not frequently modified data members
+  // ------------------------------------
+  //
+  // We separate data members that are updated frequently from the ones that
+  // are not frequently updated so that they don't share the same cache line
+  // which will lead into false cache sharing
+  //
+  // ------------------------------------
+  // Frequently modified data members
+  // ------------vvvvvvvvvvvvv-----------
   LRUHandleTable table_;
+
+  // Memory size for entries residing in the cache
+  size_t usage_;
+
+  // Memory size for entries residing only in the LRU list
+  size_t lru_usage_;
+
+  // mutex_ protects the following state.
+  // We don't count mutex_ as the cache's internal state so semantically we
+  // don't mind mutex_ invoking the non-const actions.
+  mutable port::Mutex mutex_;
 };
 
 class LRUCache : public ShardedCache {
@@ -267,8 +294,14 @@ class LRUCache : public ShardedCache {
   virtual uint32_t GetHash(Handle* handle) const override;
   virtual void DisownData() override;
 
+  //  Retrieves number of elements in LRU, for unit test purpose only
+  size_t TEST_GetLRUSize();
+  //  Retrives high pri pool ratio
+  double GetHighPriPoolRatio();
+
  private:
   LRUCacheShard* shards_;
+  int num_shards_ = 0;
 };
 
 }  // namespace rocksdb

cache/lru_cache_test.cc

@@ -17,7 +17,16 @@ class LRUCacheTest : public testing::Test {
   ~LRUCacheTest() {}
 
   void NewCache(size_t capacity, double high_pri_pool_ratio = 0.0) {
-    cache_.reset(new LRUCacheShard());
+    cache_.reset(
+#if defined(_MSC_VER)
+#pragma warning(push)
+#pragma warning(disable: 4316) // We've validated the alignment with the new operators
+#endif
+      new LRUCacheShard()
+#if defined(_MSC_VER)
+#pragma warning(pop)
+#endif
+    );
     cache_->SetCapacity(capacity);
     cache_->SetStrictCapacityLimit(false);
     cache_->SetHighPriorityPoolRatio(high_pri_pool_ratio);

cmake/RocksDBConfig.cmake.in

@@ -0,0 +1,3 @@
+@PACKAGE_INIT@
+include("${CMAKE_CURRENT_LIST_DIR}/RocksDBTargets.cmake")
+check_required_components(RocksDB)

cmake/modules/FindNUMA.cmake

@@ -0,0 +1,21 @@
+# - Find NUMA
+# Find the NUMA library and includes
+#
+# NUMA_INCLUDE_DIR - where to find numa.h, etc.
+# NUMA_LIBRARIES - List of libraries when using NUMA.
+# NUMA_FOUND - True if NUMA found.
+
+find_path(NUMA_INCLUDE_DIR
+  NAMES numa.h numaif.h
+  HINTS ${NUMA_ROOT_DIR}/include)
+
+find_library(NUMA_LIBRARIES
+  NAMES numa
+  HINTS ${NUMA_ROOT_DIR}/lib)
+
+include(FindPackageHandleStandardArgs)
+find_package_handle_standard_args(NUMA DEFAULT_MSG NUMA_LIBRARIES NUMA_INCLUDE_DIR)
+
+mark_as_advanced(
+  NUMA_LIBRARIES
+  NUMA_INCLUDE_DIR)

cmake/modules/FindTBB.cmake

@@ -0,0 +1,21 @@
+# - Find TBB
+# Find the Thread Building Blocks library and includes
+#
+# TBB_INCLUDE_DIR - where to find tbb.h, etc.
+# TBB_LIBRARIES - List of libraries when using TBB.
+# TBB_FOUND - True if TBB found.
+
+find_path(TBB_INCLUDE_DIR
+NAMES tbb/tbb.h
+HINTS ${TBB_ROOT_DIR}/include)
+
+find_library(TBB_LIBRARIES
+NAMES tbb
+HINTS ${TBB_ROOT_DIR}/lib)
+
+include(FindPackageHandleStandardArgs)
+find_package_handle_standard_args(TBB DEFAULT_MSG TBB_LIBRARIES TBB_INCLUDE_DIR)
+
+mark_as_advanced(
+TBB_LIBRARIES
+TBB_INCLUDE_DIR)

cmake/modules/Findbzip2.cmake

@@ -0,0 +1,21 @@
+# - Find Bzip2
+# Find the bzip2 compression library and includes
+#
+# BZIP2_INCLUDE_DIR - where to find bzlib.h, etc.
+# BZIP2_LIBRARIES - List of libraries when using bzip2.
+# BZIP2_FOUND - True if bzip2 found.
+
+find_path(BZIP2_INCLUDE_DIR
+  NAMES bzlib.h
+  HINTS ${BZIP2_ROOT_DIR}/include)
+
+find_library(BZIP2_LIBRARIES
+  NAMES bz2
+  HINTS ${BZIP2_ROOT_DIR}/lib)
+
+include(FindPackageHandleStandardArgs)
+find_package_handle_standard_args(bzip2 DEFAULT_MSG BZIP2_LIBRARIES BZIP2_INCLUDE_DIR)
+
+mark_as_advanced(
+  BZIP2_LIBRARIES
+  BZIP2_INCLUDE_DIR)

cmake/modules/Findlz4.cmake

@@ -0,0 +1,21 @@
+# - Find Lz4
+# Find the lz4 compression library and includes
+#
+# LZ4_INCLUDE_DIR - where to find lz4.h, etc.
+# LZ4_LIBRARIES - List of libraries when using lz4.
+# LZ4_FOUND - True if lz4 found.
+
+find_path(LZ4_INCLUDE_DIR
+  NAMES lz4.h
+  HINTS ${LZ4_ROOT_DIR}/include)
+
+find_library(LZ4_LIBRARIES
+  NAMES lz4
+  HINTS ${LZ4_ROOT_DIR}/lib)
+
+include(FindPackageHandleStandardArgs)
+find_package_handle_standard_args(lz4 DEFAULT_MSG LZ4_LIBRARIES LZ4_INCLUDE_DIR)
+
+mark_as_advanced(
+  LZ4_LIBRARIES
+  LZ4_INCLUDE_DIR)

cmake/modules/Findzlib.cmake

@@ -0,0 +1,21 @@
+# - Find zlib
+# Find the zlib compression library and includes
+#
+# ZLIB_INCLUDE_DIR - where to find zlib.h, etc.
+# ZLIB_LIBRARIES - List of libraries when using zlib.
+# ZLIB_FOUND - True if zlib found.
+
+find_path(ZLIB_INCLUDE_DIR
+  NAMES zlib.h
+  HINTS ${ZLIB_ROOT_DIR}/include)
+
+find_library(ZLIB_LIBRARIES
+  NAMES z
+  HINTS ${ZLIB_ROOT_DIR}/lib)
+
+include(FindPackageHandleStandardArgs)
+find_package_handle_standard_args(zlib DEFAULT_MSG ZLIB_LIBRARIES ZLIB_INCLUDE_DIR)
+
+mark_as_advanced(
+  ZLIB_LIBRARIES
+  ZLIB_INCLUDE_DIR)

cmake/modules/Findzstd.cmake

@@ -0,0 +1,21 @@
+# - Find zstd
+# Find the zstd compression library and includes
+#
+# ZSTD_INCLUDE_DIR - where to find zstd.h, etc.
+# ZSTD_LIBRARIES - List of libraries when using zstd.
+# ZSTD_FOUND - True if zstd found.
+
+find_path(ZSTD_INCLUDE_DIR
+  NAMES zstd.h
+  HINTS ${ZSTD_ROOT_DIR}/include)
+
+find_library(ZSTD_LIBRARIES
+  NAMES zstd
+  HINTS ${ZSTD_ROOT_DIR}/lib)
+
+include(FindPackageHandleStandardArgs)
+find_package_handle_standard_args(zstd DEFAULT_MSG ZSTD_LIBRARIES ZSTD_INCLUDE_DIR)
+
+mark_as_advanced(
+  ZSTD_LIBRARIES
+  ZSTD_INCLUDE_DIR)

coverage/coverage_test.sh

@@ -1,4 +1,4 @@
-#!/bin/bash
+#!/usr/bin/env bash
 
 # Exit on error.
 set -e

db/builder.cc

@@ -47,15 +47,15 @@ TableBuilder* NewTableBuilder(
     WritableFileWriter* file, const CompressionType compression_type,
     const CompressionOptions& compression_opts, int level,
     const std::string* compression_dict, const bool skip_filters,
-    const uint64_t creation_time) {
+    const uint64_t creation_time, const uint64_t oldest_key_time) {
   assert((column_family_id ==
           TablePropertiesCollectorFactory::Context::kUnknownColumnFamily) ==
          column_family_name.empty());
   return ioptions.table_factory->NewTableBuilder(
-      TableBuilderOptions(ioptions, internal_comparator,
-                          int_tbl_prop_collector_factories, compression_type,
-                          compression_opts, compression_dict, skip_filters,
-                          column_family_name, level, creation_time),
+      TableBuilderOptions(
+          ioptions, internal_comparator, int_tbl_prop_collector_factories,
+          compression_type, compression_opts, compression_dict, skip_filters,
+          column_family_name, level, creation_time, oldest_key_time),
       column_family_id, file);
 }
 
@@ -70,12 +70,12 @@ Status BuildTable(
     uint32_t column_family_id, const std::string& column_family_name,
     std::vector<SequenceNumber> snapshots,
     SequenceNumber earliest_write_conflict_snapshot,
-    const CompressionType compression,
+    SnapshotChecker* snapshot_checker, const CompressionType compression,
     const CompressionOptions& compression_opts, bool paranoid_file_checks,
     InternalStats* internal_stats, TableFileCreationReason reason,
     EventLogger* event_logger, int job_id, const Env::IOPriority io_priority,
-    TableProperties* table_properties, int level,
-    const uint64_t creation_time) {
+    TableProperties* table_properties, int level, const uint64_t creation_time,
+    const uint64_t oldest_key_time, Env::WriteLifeTimeHint write_hint) {
   assert((column_family_id ==
           TablePropertiesCollectorFactory::Context::kUnknownColumnFamily) ==
          column_family_name.empty());
@@ -117,15 +117,15 @@ Status BuildTable(
         return s;
       }
       file->SetIOPriority(io_priority);
+      file->SetWriteLifeTimeHint(write_hint);
 
       file_writer.reset(new WritableFileWriter(std::move(file), env_options,
                                                ioptions.statistics));
-
       builder = NewTableBuilder(
           ioptions, internal_comparator, int_tbl_prop_collector_factories,
           column_family_id, column_family_name, file_writer.get(), compression,
           compression_opts, level, nullptr /* compression_dict */,
-          false /* skip_filters */, creation_time);
+          false /* skip_filters */, creation_time, oldest_key_time);
     }
 
     MergeHelper merge(env, internal_comparator.user_comparator(),
@@ -135,7 +135,7 @@ Status BuildTable(
 
     CompactionIterator c_iter(
         iter, internal_comparator.user_comparator(), &merge, kMaxSequenceNumber,
-        &snapshots, earliest_write_conflict_snapshot, env,
+        &snapshots, earliest_write_conflict_snapshot, snapshot_checker, env,
         true /* internal key corruption is not ok */, range_del_agg.get());
     c_iter.SeekToFirst();
     for (; c_iter.Valid(); c_iter.Next()) {

db/builder.h

@@ -29,6 +29,7 @@ struct FileMetaData;
 class Env;
 struct EnvOptions;
 class Iterator;
+class SnapshotChecker;
 class TableCache;
 class VersionEdit;
 class TableBuilder;
@@ -50,7 +51,8 @@ TableBuilder* NewTableBuilder(
     WritableFileWriter* file, const CompressionType compression_type,
     const CompressionOptions& compression_opts, int level,
     const std::string* compression_dict = nullptr,
-    const bool skip_filters = false, const uint64_t creation_time = 0);
+    const bool skip_filters = false, const uint64_t creation_time = 0,
+    const uint64_t oldest_key_time = 0);
 
 // Build a Table file from the contents of *iter.  The generated file
 // will be named according to number specified in meta. On success, the rest of
@@ -71,12 +73,13 @@ extern Status BuildTable(
     uint32_t column_family_id, const std::string& column_family_name,
     std::vector<SequenceNumber> snapshots,
     SequenceNumber earliest_write_conflict_snapshot,
-    const CompressionType compression,
+    SnapshotChecker* snapshot_checker, const CompressionType compression,
     const CompressionOptions& compression_opts, bool paranoid_file_checks,
     InternalStats* internal_stats, TableFileCreationReason reason,
     EventLogger* event_logger = nullptr, int job_id = 0,
     const Env::IOPriority io_priority = Env::IO_HIGH,
     TableProperties* table_properties = nullptr, int level = -1,
-    const uint64_t creation_time = 0);
+    const uint64_t creation_time = 0, const uint64_t oldest_key_time = 0,
+    Env::WriteLifeTimeHint write_hint = Env::WLTH_NOT_SET);
 
 }  // namespace rocksdb

db/c.cc

@@ -21,22 +21,23 @@
 #include "rocksdb/env.h"
 #include "rocksdb/filter_policy.h"
 #include "rocksdb/iterator.h"
+#include "rocksdb/memtablerep.h"
 #include "rocksdb/merge_operator.h"
 #include "rocksdb/options.h"
-#include "rocksdb/status.h"
-#include "rocksdb/write_batch.h"
-#include "rocksdb/memtablerep.h"
-#include "rocksdb/universal_compaction.h"
-#include "rocksdb/statistics.h"
-#include "rocksdb/slice_transform.h"
-#include "rocksdb/table.h"
 #include "rocksdb/rate_limiter.h"
+#include "rocksdb/slice_transform.h"
+#include "rocksdb/statistics.h"
+#include "rocksdb/status.h"
+#include "rocksdb/table.h"
+#include "rocksdb/universal_compaction.h"
 #include "rocksdb/utilities/backupable_db.h"
-#include "rocksdb/utilities/write_batch_with_index.h"
-#include "utilities/merge_operators.h"
+#include "rocksdb/utilities/checkpoint.h"
+#include "rocksdb/utilities/optimistic_transaction_db.h"
 #include "rocksdb/utilities/transaction.h"
 #include "rocksdb/utilities/transaction_db.h"
-#include "rocksdb/utilities/checkpoint.h"
+#include "rocksdb/utilities/write_batch_with_index.h"
+#include "rocksdb/write_batch.h"
+#include "utilities/merge_operators.h"
 
 using rocksdb::BytewiseComparator;
 using rocksdb::Cache;
@@ -52,6 +53,7 @@ using rocksdb::CompressionType;
 using rocksdb::WALRecoveryMode;
 using rocksdb::DB;
 using rocksdb::DBOptions;
+using rocksdb::DbPath;
 using rocksdb::Env;
 using rocksdb::EnvOptions;
 using rocksdb::InfoLogLevel;
@@ -94,6 +96,8 @@ using rocksdb::PinnableSlice;
 using rocksdb::TransactionDBOptions;
 using rocksdb::TransactionDB;
 using rocksdb::TransactionOptions;
+using rocksdb::OptimisticTransactionDB;
+using rocksdb::OptimisticTransactionOptions;
 using rocksdb::Transaction;
 using rocksdb::Checkpoint;
 
@@ -113,7 +117,9 @@ struct rocksdb_flushoptions_t    { FlushOptions      rep; };
 struct rocksdb_fifo_compaction_options_t { CompactionOptionsFIFO rep; };
 struct rocksdb_readoptions_t {
    ReadOptions rep;
-   Slice upper_bound; // stack variable to set pointer to in ReadOptions
+   // stack variables to set pointers to in ReadOptions
+   Slice upper_bound;
+   Slice lower_bound;
 };
 struct rocksdb_writeoptions_t    { WriteOptions      rep; };
 struct rocksdb_options_t         { Options           rep; };
@@ -152,6 +158,12 @@ struct rocksdb_transaction_t {
 struct rocksdb_checkpoint_t {
   Checkpoint* rep;
 };
+struct rocksdb_optimistictransactiondb_t {
+  OptimisticTransactionDB* rep;
+};
+struct rocksdb_optimistictransaction_options_t {
+  OptimisticTransactionOptions rep;
+};
 
 struct rocksdb_compactionfiltercontext_t {
   CompactionFilter::Context rep;
@@ -382,6 +394,10 @@ struct rocksdb_mergeoperator_t : public MergeOperator {
   }
 };
 
+struct rocksdb_dbpath_t {
+  DbPath rep;
+};
+
 struct rocksdb_env_t {
   Env* rep;
   bool is_default;
@@ -1372,24 +1388,23 @@ void rocksdb_writebatch_put_log_data(
   b->rep.PutLogData(Slice(blob, len));
 }
 
-class H : public WriteBatch::Handler {
- public:
-  void* state_;
-  void (*put_)(void*, const char* k, size_t klen, const char* v, size_t vlen);
-  void (*deleted_)(void*, const char* k, size_t klen);
-  virtual void Put(const Slice& key, const Slice& value) override {
-    (*put_)(state_, key.data(), key.size(), value.data(), value.size());
-  }
-  virtual void Delete(const Slice& key) override {
-    (*deleted_)(state_, key.data(), key.size());
-  }
-};
-
 void rocksdb_writebatch_iterate(
     rocksdb_writebatch_t* b,
     void* state,
     void (*put)(void*, const char* k, size_t klen, const char* v, size_t vlen),
     void (*deleted)(void*, const char* k, size_t klen)) {
+  class H : public WriteBatch::Handler {
+   public:
+    void* state_;
+    void (*put_)(void*, const char* k, size_t klen, const char* v, size_t vlen);
+    void (*deleted_)(void*, const char* k, size_t klen);
+    virtual void Put(const Slice& key, const Slice& value) override {
+      (*put_)(state_, key.data(), key.size(), value.data(), value.size());
+    }
+    virtual void Delete(const Slice& key) override {
+      (*deleted_)(state_, key.data(), key.size());
+    }
+  };
   H handler;
   handler.state_ = state;
   handler.put_ = put;
@@ -1634,6 +1649,18 @@ void rocksdb_writebatch_wi_iterate(
     void* state,
     void (*put)(void*, const char* k, size_t klen, const char* v, size_t vlen),
     void (*deleted)(void*, const char* k, size_t klen)) {
+  class H : public WriteBatch::Handler {
+   public:
+    void* state_;
+    void (*put_)(void*, const char* k, size_t klen, const char* v, size_t vlen);
+    void (*deleted_)(void*, const char* k, size_t klen);
+    virtual void Put(const Slice& key, const Slice& value) override {
+      (*put_)(state_, key.data(), key.size(), value.data(), value.size());
+    }
+    virtual void Delete(const Slice& key) override {
+      (*deleted_)(state_, key.data(), key.size());
+    }
+  };
   H handler;
   handler.state_ = state;
   handler.put_ = put;
@@ -1799,6 +1826,26 @@ void rocksdb_block_based_options_set_block_restart_interval(
   options->rep.block_restart_interval = block_restart_interval;
 }
 
+void rocksdb_block_based_options_set_index_block_restart_interval(
+    rocksdb_block_based_table_options_t* options, int index_block_restart_interval) {
+  options->rep.index_block_restart_interval = index_block_restart_interval;
+}
+
+void rocksdb_block_based_options_set_metadata_block_size(
+    rocksdb_block_based_table_options_t* options, uint64_t metadata_block_size) {
+  options->rep.metadata_block_size = metadata_block_size;
+}
+
+void rocksdb_block_based_options_set_partition_filters(
+    rocksdb_block_based_table_options_t* options, unsigned char partition_filters) {
+  options->rep.partition_filters = partition_filters;
+}
+
+void rocksdb_block_based_options_set_use_delta_encoding(
+    rocksdb_block_based_table_options_t* options, unsigned char use_delta_encoding) {
+  options->rep.use_delta_encoding = use_delta_encoding;
+}
+
 void rocksdb_block_based_options_set_filter_policy(
     rocksdb_block_based_table_options_t* options,
     rocksdb_filterpolicy_t* filter_policy) {
@@ -1852,6 +1899,11 @@ void rocksdb_block_based_options_set_cache_index_and_filter_blocks(
   options->rep.cache_index_and_filter_blocks = v;
 }
 
+void rocksdb_block_based_options_set_cache_index_and_filter_blocks_with_high_priority(
+    rocksdb_block_based_table_options_t* options, unsigned char v) {
+  options->rep.cache_index_and_filter_blocks_with_high_priority = v;
+}
+
 void rocksdb_block_based_options_set_pin_l0_filter_and_index_blocks_in_cache(
     rocksdb_block_based_table_options_t* options, unsigned char v) {
   options->rep.pin_l0_filter_and_index_blocks_in_cache = v;
@@ -1866,7 +1918,6 @@ void rocksdb_options_set_block_based_table_factory(
   }
 }
 
-
 rocksdb_cuckoo_table_options_t*
 rocksdb_cuckoo_options_create() {
   return new rocksdb_cuckoo_table_options_t;
@@ -1948,6 +1999,11 @@ void rocksdb_options_optimize_universal_style_compaction(
   opt->rep.OptimizeUniversalStyleCompaction(memtable_memory_budget);
 }
 
+void rocksdb_options_set_allow_ingest_behind(
+    rocksdb_options_t* opt, unsigned char v) {
+  opt->rep.allow_ingest_behind = v;
+}
+
 void rocksdb_options_set_compaction_filter(
     rocksdb_options_t* opt,
     rocksdb_compactionfilter_t* filter) {
@@ -1998,6 +2054,16 @@ void rocksdb_options_set_paranoid_checks(
   opt->rep.paranoid_checks = v;
 }
 
+void rocksdb_options_set_db_paths(rocksdb_options_t* opt,
+                                  const rocksdb_dbpath_t** dbpath_values,
+                                  size_t num_paths) {
+  std::vector<DbPath> db_paths(num_paths);
+  for (size_t i = 0; i < num_paths; ++i) {
+    db_paths[i] = dbpath_values[i]->rep;
+  }
+  opt->rep.db_paths = db_paths;
+}
+
 void rocksdb_options_set_env(rocksdb_options_t* opt, rocksdb_env_t* env) {
   opt->rep.env = (env ? env->rep : nullptr);
 }
@@ -2076,6 +2142,11 @@ void rocksdb_options_enable_statistics(rocksdb_options_t* opt) {
   opt->rep.statistics = rocksdb::CreateDBStatistics();
 }
 
+void rocksdb_options_set_skip_stats_update_on_db_open(rocksdb_options_t* opt,
+                                                      unsigned char val) {
+  opt->rep.skip_stats_update_on_db_open = val;
+}
+
 void rocksdb_options_set_num_levels(rocksdb_options_t* opt, int n) {
   opt->rep.num_levels = n;
 }
@@ -2226,11 +2297,21 @@ void rocksdb_options_set_use_adaptive_mutex(
   opt->rep.use_adaptive_mutex = v;
 }
 
+void rocksdb_options_set_wal_bytes_per_sync(
+    rocksdb_options_t* opt, uint64_t v) {
+  opt->rep.wal_bytes_per_sync = v;
+}
+
 void rocksdb_options_set_bytes_per_sync(
     rocksdb_options_t* opt, uint64_t v) {
   opt->rep.bytes_per_sync = v;
 }
 
+void rocksdb_options_set_writable_file_max_buffer_size(rocksdb_options_t* opt,
+                                                       uint64_t v) {
+  opt->rep.writable_file_max_buffer_size = v;
+}
+
 void rocksdb_options_set_allow_concurrent_memtable_write(rocksdb_options_t* opt,
                                                          unsigned char v) {
   opt->rep.allow_concurrent_memtable_write = v;
@@ -2678,6 +2759,18 @@ void rocksdb_readoptions_set_iterate_upper_bound(
   }
 }
 
+void rocksdb_readoptions_set_iterate_lower_bound(
+    rocksdb_readoptions_t *opt,
+    const char* key, size_t keylen) {
+  if (key == nullptr) {
+    opt->lower_bound = Slice();
+    opt->rep.iterate_lower_bound = nullptr;
+  } else {
+    opt->lower_bound = Slice(key, keylen);
+    opt->rep.iterate_lower_bound = &opt->lower_bound;
+  }
+}
+
 void rocksdb_readoptions_set_read_tier(
     rocksdb_readoptions_t* opt, int v) {
   opt->rep.read_tier = static_cast<rocksdb::ReadTier>(v);
@@ -2688,11 +2781,21 @@ void rocksdb_readoptions_set_tailing(
   opt->rep.tailing = v;
 }
 
+void rocksdb_readoptions_set_managed(
+    rocksdb_readoptions_t* opt, unsigned char v) {
+  opt->rep.managed = v;
+}
+
 void rocksdb_readoptions_set_readahead_size(
     rocksdb_readoptions_t* opt, size_t v) {
   opt->rep.readahead_size = v;
 }
 
+void rocksdb_readoptions_set_prefix_same_as_start(
+    rocksdb_readoptions_t* opt, unsigned char v) {
+  opt->rep.prefix_same_as_start = v;
+}
+
 void rocksdb_readoptions_set_pin_data(rocksdb_readoptions_t* opt,
                                       unsigned char v) {
   opt->rep.pin_data = v;
@@ -2703,6 +2806,22 @@ void rocksdb_readoptions_set_total_order_seek(rocksdb_readoptions_t* opt,
   opt->rep.total_order_seek = v;
 }
 
+void rocksdb_readoptions_set_max_skippable_internal_keys(
+    rocksdb_readoptions_t* opt,
+    uint64_t v) {
+  opt->rep.max_skippable_internal_keys = v;
+}
+
+void rocksdb_readoptions_set_background_purge_on_iterator_cleanup(
+    rocksdb_readoptions_t* opt, unsigned char v) {
+  opt->rep.background_purge_on_iterator_cleanup = v;
+}
+
+void rocksdb_readoptions_set_ignore_range_deletions(
+    rocksdb_readoptions_t* opt, unsigned char v) {
+  opt->rep.ignore_range_deletions = v;
+}
+
 rocksdb_writeoptions_t* rocksdb_writeoptions_create() {
   return new rocksdb_writeoptions_t;
 }
@@ -2720,6 +2839,24 @@ void rocksdb_writeoptions_disable_WAL(rocksdb_writeoptions_t* opt, int disable)
   opt->rep.disableWAL = disable;
 }
 
+void rocksdb_writeoptions_set_ignore_missing_column_families(
+    rocksdb_writeoptions_t* opt,
+    unsigned char v) {
+  opt->rep.ignore_missing_column_families = v;
+}
+
+void rocksdb_writeoptions_set_no_slowdown(
+    rocksdb_writeoptions_t* opt,
+    unsigned char v) {
+  opt->rep.no_slowdown = v;
+}
+
+void rocksdb_writeoptions_set_low_pri(
+    rocksdb_writeoptions_t* opt,
+    unsigned char v) {
+  opt->rep.low_pri = v;
+}
+
 rocksdb_compactoptions_t* rocksdb_compactoptions_create() {
   return new rocksdb_compactoptions_t;
 }
@@ -2778,6 +2915,17 @@ size_t rocksdb_cache_get_pinned_usage(rocksdb_cache_t* cache) {
   return cache->rep->GetPinnedUsage();
 }
 
+rocksdb_dbpath_t* rocksdb_dbpath_create(const char* path, uint64_t target_size) {
+  rocksdb_dbpath_t* result = new rocksdb_dbpath_t;
+  result->rep.path = std::string(path);
+  result->rep.target_size = target_size;
+  return result;
+}
+
+void rocksdb_dbpath_destroy(rocksdb_dbpath_t* dbpath) {
+  delete dbpath;
+}
+
 rocksdb_env_t* rocksdb_create_default_env() {
   rocksdb_env_t* result = new rocksdb_env_t;
   result->rep = Env::Default();
@@ -2901,6 +3049,12 @@ void rocksdb_ingestexternalfileoptions_set_allow_blocking_flush(
   opt->rep.allow_blocking_flush = allow_blocking_flush;
 }
 
+void rocksdb_ingestexternalfileoptions_set_ingest_behind(
+    rocksdb_ingestexternalfileoptions_t* opt,
+    unsigned char ingest_behind) {
+  opt->rep.ingest_behind = ingest_behind;
+}
+
 void rocksdb_ingestexternalfileoptions_destroy(
     rocksdb_ingestexternalfileoptions_t* opt) {
   delete opt;
@@ -2955,21 +3109,20 @@ void rocksdb_slicetransform_destroy(rocksdb_slicetransform_t* st) {
   delete st;
 }
 
-struct Wrapper : public rocksdb_slicetransform_t {
-  const SliceTransform* rep_;
-  ~Wrapper() { delete rep_; }
-  const char* Name() const override { return rep_->Name(); }
-  Slice Transform(const Slice& src) const override {
-    return rep_->Transform(src);
-  }
-  bool InDomain(const Slice& src) const override {
-    return rep_->InDomain(src);
-  }
-  bool InRange(const Slice& src) const override { return rep_->InRange(src); }
-  static void DoNothing(void*) { }
-};
-
 rocksdb_slicetransform_t* rocksdb_slicetransform_create_fixed_prefix(size_t prefixLen) {
+  struct Wrapper : public rocksdb_slicetransform_t {
+    const SliceTransform* rep_;
+    ~Wrapper() { delete rep_; }
+    const char* Name() const override { return rep_->Name(); }
+    Slice Transform(const Slice& src) const override {
+      return rep_->Transform(src);
+    }
+    bool InDomain(const Slice& src) const override {
+      return rep_->InDomain(src);
+    }
+    bool InRange(const Slice& src) const override { return rep_->InRange(src); }
+    static void DoNothing(void*) { }
+  };
   Wrapper* wrapper = new Wrapper;
   wrapper->rep_ = rocksdb::NewFixedPrefixTransform(prefixLen);
   wrapper->state_ = nullptr;
@@ -2978,6 +3131,19 @@ rocksdb_slicetransform_t* rocksdb_slicetransform_create_fixed_prefix(size_t pref
 }
 
 rocksdb_slicetransform_t* rocksdb_slicetransform_create_noop() {
+  struct Wrapper : public rocksdb_slicetransform_t {
+    const SliceTransform* rep_;
+    ~Wrapper() { delete rep_; }
+    const char* Name() const override { return rep_->Name(); }
+    Slice Transform(const Slice& src) const override {
+      return rep_->Transform(src);
+    }
+    bool InDomain(const Slice& src) const override {
+      return rep_->InDomain(src);
+    }
+    bool InRange(const Slice& src) const override { return rep_->InRange(src); }
+    static void DoNothing(void*) { }
+  };
   Wrapper* wrapper = new Wrapper;
   wrapper->rep_ = rocksdb::NewNoopTransform();
   wrapper->state_ = nullptr;
@@ -3200,6 +3366,32 @@ void rocksdb_transaction_options_set_max_write_batch_size(
   opt->rep.max_write_batch_size = size;
 }
 
+rocksdb_optimistictransaction_options_t*
+rocksdb_optimistictransaction_options_create() {
+  return new rocksdb_optimistictransaction_options_t;
+}
+
+void rocksdb_optimistictransaction_options_destroy(
+    rocksdb_optimistictransaction_options_t* opt) {
+  delete opt;
+}
+
+void rocksdb_optimistictransaction_options_set_set_snapshot(
+    rocksdb_optimistictransaction_options_t* opt, unsigned char v) {
+  opt->rep.set_snapshot = v;
+}
+
+rocksdb_column_family_handle_t* rocksdb_transactiondb_create_column_family(
+    rocksdb_transactiondb_t* txn_db,
+    const rocksdb_options_t* column_family_options,
+    const char* column_family_name, char** errptr) {
+  rocksdb_column_family_handle_t* handle = new rocksdb_column_family_handle_t;
+  SaveError(errptr, txn_db->rep->CreateColumnFamily(
+                        ColumnFamilyOptions(column_family_options->rep),
+                        std::string(column_family_name), &(handle->rep)));
+  return handle;
+}
+
 rocksdb_transactiondb_t* rocksdb_transactiondb_open(
     const rocksdb_options_t* options,
     const rocksdb_transactiondb_options_t* txn_db_options, const char* name,
@@ -3251,12 +3443,27 @@ void rocksdb_transaction_rollback(rocksdb_transaction_t* txn, char** errptr) {
   SaveError(errptr, txn->rep->Rollback());
 }
 
+void rocksdb_transaction_set_savepoint(rocksdb_transaction_t* txn) {
+  txn->rep->SetSavePoint();
+}
+
+void rocksdb_transaction_rollback_to_savepoint(rocksdb_transaction_t* txn, char** errptr) {
+  SaveError(errptr, txn->rep->RollbackToSavePoint());
+}
+
 void rocksdb_transaction_destroy(rocksdb_transaction_t* txn) {
   delete txn->rep;
   delete txn;
 }
 
-//Read a key inside a transaction
+const rocksdb_snapshot_t* rocksdb_transaction_get_snapshot(
+    rocksdb_transaction_t* txn) {
+  rocksdb_snapshot_t* result = new rocksdb_snapshot_t;
+  result->rep = txn->rep->GetSnapshot();
+  return result;
+}
+
+// Read a key inside a transaction
 char* rocksdb_transaction_get(rocksdb_transaction_t* txn,
                               const rocksdb_readoptions_t* options,
                               const char* key, size_t klen, size_t* vlen,
@@ -3276,6 +3483,49 @@ char* rocksdb_transaction_get(rocksdb_transaction_t* txn,
   return result;
 }
 
+char* rocksdb_transaction_get_cf(rocksdb_transaction_t* txn,
+                                 const rocksdb_readoptions_t* options,
+                                 rocksdb_column_family_handle_t* column_family,
+                                 const char* key, size_t klen, size_t* vlen,
+                                 char** errptr) {
+  char* result = nullptr;
+  std::string tmp;
+  Status s =
+      txn->rep->Get(options->rep, column_family->rep, Slice(key, klen), &tmp);
+  if (s.ok()) {
+    *vlen = tmp.size();
+    result = CopyString(tmp);
+  } else {
+    *vlen = 0;
+    if (!s.IsNotFound()) {
+      SaveError(errptr, s);
+    }
+  }
+  return result;
+}
+
+// Read a key inside a transaction
+char* rocksdb_transaction_get_for_update(rocksdb_transaction_t* txn,
+                                         const rocksdb_readoptions_t* options,
+                                         const char* key, size_t klen,
+                                         size_t* vlen, unsigned char exclusive,
+                                         char** errptr) {
+  char* result = nullptr;
+  std::string tmp;
+  Status s =
+      txn->rep->GetForUpdate(options->rep, Slice(key, klen), &tmp, exclusive);
+  if (s.ok()) {
+    *vlen = tmp.size();
+    result = CopyString(tmp);
+  } else {
+    *vlen = 0;
+    if (!s.IsNotFound()) {
+      SaveError(errptr, s);
+    }
+  }
+  return result;
+}
+
 // Read a key outside a transaction
 char* rocksdb_transactiondb_get(
     rocksdb_transactiondb_t* txn_db,
@@ -3298,6 +3548,26 @@ char* rocksdb_transactiondb_get(
   return result;
 }
 
+char* rocksdb_transactiondb_get_cf(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_readoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, size_t* vallen, char** errptr) {
+  char* result = nullptr;
+  std::string tmp;
+  Status s = txn_db->rep->Get(options->rep, column_family->rep,
+                              Slice(key, keylen), &tmp);
+  if (s.ok()) {
+    *vallen = tmp.size();
+    result = CopyString(tmp);
+  } else {
+    *vallen = 0;
+    if (!s.IsNotFound()) {
+      SaveError(errptr, s);
+    }
+  }
+  return result;
+}
+
 // Put a key inside a transaction
 void rocksdb_transaction_put(rocksdb_transaction_t* txn, const char* key,
                              size_t klen, const char* val, size_t vlen,
@@ -3305,7 +3575,15 @@ void rocksdb_transaction_put(rocksdb_transaction_t* txn, const char* key,
   SaveError(errptr, txn->rep->Put(Slice(key, klen), Slice(val, vlen)));
 }
 
-//Put a key outside a transaction
+void rocksdb_transaction_put_cf(rocksdb_transaction_t* txn,
+                                rocksdb_column_family_handle_t* column_family,
+                                const char* key, size_t klen, const char* val,
+                                size_t vlen, char** errptr) {
+  SaveError(errptr, txn->rep->Put(column_family->rep, Slice(key, klen),
+                                  Slice(val, vlen)));
+}
+
+// Put a key outside a transaction
 void rocksdb_transactiondb_put(rocksdb_transactiondb_t* txn_db,
                                const rocksdb_writeoptions_t* options,
                                const char* key, size_t klen, const char* val,
@@ -3314,12 +3592,53 @@ void rocksdb_transactiondb_put(rocksdb_transactiondb_t* txn_db,
             txn_db->rep->Put(options->rep, Slice(key, klen), Slice(val, vlen)));
 }
 
+void rocksdb_transactiondb_put_cf(rocksdb_transactiondb_t* txn_db,
+                                  const rocksdb_writeoptions_t* options,
+                                  rocksdb_column_family_handle_t* column_family,
+                                  const char* key, size_t keylen,
+                                  const char* val, size_t vallen,
+                                  char** errptr) {
+  SaveError(errptr, txn_db->rep->Put(options->rep, column_family->rep,
+                                     Slice(key, keylen), Slice(val, vallen)));
+}
+
+// Write batch into transaction db
+void rocksdb_transactiondb_write(
+        rocksdb_transactiondb_t* db,
+        const rocksdb_writeoptions_t* options,
+        rocksdb_writebatch_t* batch,
+        char** errptr) {
+  SaveError(errptr, db->rep->Write(options->rep, &batch->rep));
+}
+
+// Merge a key inside a transaction
+void rocksdb_transaction_merge(rocksdb_transaction_t* txn, const char* key,
+                               size_t klen, const char* val, size_t vlen,
+                               char** errptr) {
+  SaveError(errptr, txn->rep->Merge(Slice(key, klen), Slice(val, vlen)));
+}
+
+// Merge a key outside a transaction
+void rocksdb_transactiondb_merge(rocksdb_transactiondb_t* txn_db,
+                                 const rocksdb_writeoptions_t* options,
+                                 const char* key, size_t klen, const char* val,
+                                 size_t vlen, char** errptr) {
+  SaveError(errptr, txn_db->rep->Merge(options->rep, Slice(key, klen),
+                                       Slice(val, vlen)));
+}
+
 // Delete a key inside a transaction
 void rocksdb_transaction_delete(rocksdb_transaction_t* txn, const char* key,
                                 size_t klen, char** errptr) {
   SaveError(errptr, txn->rep->Delete(Slice(key, klen)));
 }
 
+void rocksdb_transaction_delete_cf(
+    rocksdb_transaction_t* txn, rocksdb_column_family_handle_t* column_family,
+    const char* key, size_t klen, char** errptr) {
+  SaveError(errptr, txn->rep->Delete(column_family->rep, Slice(key, klen)));
+}
+
 // Delete a key outside a transaction
 void rocksdb_transactiondb_delete(rocksdb_transactiondb_t* txn_db,
                                   const rocksdb_writeoptions_t* options,
@@ -3327,6 +3646,14 @@ void rocksdb_transactiondb_delete(rocksdb_transactiondb_t* txn_db,
   SaveError(errptr, txn_db->rep->Delete(options->rep, Slice(key, klen)));
 }
 
+void rocksdb_transactiondb_delete_cf(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_writeoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, char** errptr) {
+  SaveError(errptr, txn_db->rep->Delete(options->rep, column_family->rep,
+                                        Slice(key, keylen)));
+}
+
 // Create an iterator inside a transaction
 rocksdb_iterator_t* rocksdb_transaction_create_iterator(
     rocksdb_transaction_t* txn, const rocksdb_readoptions_t* options) {
@@ -3335,6 +3662,23 @@ rocksdb_iterator_t* rocksdb_transaction_create_iterator(
   return result;
 }
 
+// Create an iterator inside a transaction with column family
+rocksdb_iterator_t* rocksdb_transaction_create_iterator_cf(
+    rocksdb_transaction_t* txn, const rocksdb_readoptions_t* options,
+    rocksdb_column_family_handle_t* column_family) {
+  rocksdb_iterator_t* result = new rocksdb_iterator_t;
+  result->rep = txn->rep->GetIterator(options->rep, column_family->rep);
+  return result;
+}
+
+// Create an iterator outside a transaction
+rocksdb_iterator_t* rocksdb_transactiondb_create_iterator(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_readoptions_t* options) {
+  rocksdb_iterator_t* result = new rocksdb_iterator_t;
+  result->rep = txn_db->rep->NewIterator(options->rep);
+  return result;
+}
+
 void rocksdb_transactiondb_close(rocksdb_transactiondb_t* txn_db) {
   delete txn_db->rep;
   delete txn_db;
@@ -3351,6 +3695,91 @@ rocksdb_checkpoint_t* rocksdb_transactiondb_checkpoint_object_create(
   return result;
 }
 
+rocksdb_optimistictransactiondb_t* rocksdb_optimistictransactiondb_open(
+    const rocksdb_options_t* options, const char* name, char** errptr) {
+  OptimisticTransactionDB* otxn_db;
+  if (SaveError(errptr, OptimisticTransactionDB::Open(
+                            options->rep, std::string(name), &otxn_db))) {
+    return nullptr;
+  }
+  rocksdb_optimistictransactiondb_t* result =
+      new rocksdb_optimistictransactiondb_t;
+  result->rep = otxn_db;
+  return result;
+}
+
+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,
+    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(
+        std::string(column_family_names[i]),
+        ColumnFamilyOptions(column_family_options[i]->rep)));
+  }
+
+  OptimisticTransactionDB* otxn_db;
+  std::vector<ColumnFamilyHandle*> handles;
+  if (SaveError(errptr, OptimisticTransactionDB::Open(
+                            DBOptions(db_options->rep), std::string(name),
+                            column_families, &handles, &otxn_db))) {
+    return nullptr;
+  }
+
+  for (size_t i = 0; i < handles.size(); i++) {
+    rocksdb_column_family_handle_t* c_handle =
+        new rocksdb_column_family_handle_t;
+    c_handle->rep = handles[i];
+    column_family_handles[i] = c_handle;
+  }
+  rocksdb_optimistictransactiondb_t* result =
+      new rocksdb_optimistictransactiondb_t;
+  result->rep = otxn_db;
+  return result;
+}
+
+rocksdb_t* rocksdb_optimistictransactiondb_get_base_db(
+    rocksdb_optimistictransactiondb_t* otxn_db) {
+  DB* base_db = otxn_db->rep->GetBaseDB();
+
+  if (base_db != nullptr) {
+    rocksdb_t* result = new rocksdb_t;
+    result->rep = base_db;
+    return result;
+  }
+
+  return nullptr;
+}
+
+void rocksdb_optimistictransactiondb_close_base_db(rocksdb_t* base_db) {
+  delete base_db;
+}
+
+rocksdb_transaction_t* rocksdb_optimistictransaction_begin(
+    rocksdb_optimistictransactiondb_t* otxn_db,
+    const rocksdb_writeoptions_t* write_options,
+    const rocksdb_optimistictransaction_options_t* otxn_options,
+    rocksdb_transaction_t* old_txn) {
+  if (old_txn == nullptr) {
+    rocksdb_transaction_t* result = new rocksdb_transaction_t;
+    result->rep = otxn_db->rep->BeginTransaction(write_options->rep,
+                                                 otxn_options->rep, nullptr);
+    return result;
+  }
+  old_txn->rep = otxn_db->rep->BeginTransaction(
+      write_options->rep, otxn_options->rep, old_txn->rep);
+  return old_txn;
+}
+
+void rocksdb_optimistictransactiondb_close(
+    rocksdb_optimistictransactiondb_t* otxn_db) {
+  delete otxn_db->rep;
+  delete otxn_db;
+}
+
 void rocksdb_free(void* ptr) { free(ptr); }
 
 rocksdb_pinnableslice_t* rocksdb_get_pinned(

db/c_test.c

@@ -41,6 +41,7 @@ static char dbname[200];
 static char sstfilename[200];
 static char dbbackupname[200];
 static char dbcheckpointname[200];
+static char dbpathname[200];
 
 static void StartPhase(const char* name) {
   fprintf(stderr, "=== Test %s\n", name);
@@ -333,6 +334,20 @@ static void CheckTxnGet(
         Free(&val);
 }
 
+static void CheckTxnGetCF(rocksdb_transaction_t* txn,
+                          const rocksdb_readoptions_t* options,
+                          rocksdb_column_family_handle_t* column_family,
+                          const char* key, const char* expected) {
+  char* err = NULL;
+  size_t val_len;
+  char* val;
+  val = rocksdb_transaction_get_cf(txn, options, column_family, key,
+                                   strlen(key), &val_len, &err);
+  CheckNoError(err);
+  CheckEqual(expected, val, val_len);
+  Free(&val);
+}
+
 static void CheckTxnDBGet(
         rocksdb_transactiondb_t* txn_db,
         const rocksdb_readoptions_t* options,
@@ -347,10 +362,25 @@ static void CheckTxnDBGet(
         Free(&val);
 }
 
+static void CheckTxnDBGetCF(rocksdb_transactiondb_t* txn_db,
+                            const rocksdb_readoptions_t* options,
+                            rocksdb_column_family_handle_t* column_family,
+                            const char* key, const char* expected) {
+  char* err = NULL;
+  size_t val_len;
+  char* val;
+  val = rocksdb_transactiondb_get_cf(txn_db, options, column_family, key,
+                                     strlen(key), &val_len, &err);
+  CheckNoError(err);
+  CheckEqual(expected, val, val_len);
+  Free(&val);
+}
+
 int main(int argc, char** argv) {
   rocksdb_t* db;
   rocksdb_comparator_t* cmp;
   rocksdb_cache_t* cache;
+  rocksdb_dbpath_t *dbpath;
   rocksdb_env_t* env;
   rocksdb_options_t* options;
   rocksdb_compactoptions_t* coptions;
@@ -362,6 +392,8 @@ int main(int argc, char** argv) {
   rocksdb_transactiondb_options_t* txn_db_options;
   rocksdb_transaction_t* txn;
   rocksdb_transaction_options_t* txn_options;
+  rocksdb_optimistictransactiondb_t* otxn_db;
+  rocksdb_optimistictransaction_options_t* otxn_options;
   char* err = NULL;
   int run = -1;
 
@@ -385,8 +417,14 @@ int main(int argc, char** argv) {
            GetTempDir(),
            ((int)geteuid()));
 
+  snprintf(dbpathname, sizeof(dbpathname),
+           "%s/rocksdb_c_test-%d-dbpath",
+           GetTempDir(),
+           ((int) geteuid()));
+
   StartPhase("create_objects");
   cmp = rocksdb_comparator_create(NULL, CmpDestroy, CmpCompare, CmpName);
+  dbpath = rocksdb_dbpath_create(dbpathname, 1024 * 1024);
   env = rocksdb_create_default_env();
   cache = rocksdb_cache_create_lru(100000);
 
@@ -1357,6 +1395,18 @@ int main(int argc, char** argv) {
     CheckNoError(err);
     CheckTxnDBGet(txn_db, roptions, "foo", NULL);
 
+    // write batch into TransactionDB
+    rocksdb_writebatch_t* wb = rocksdb_writebatch_create();
+    rocksdb_writebatch_put(wb, "foo", 3, "a", 1);
+    rocksdb_writebatch_clear(wb);
+    rocksdb_writebatch_put(wb, "bar", 3, "b", 1);
+    rocksdb_writebatch_put(wb, "box", 3, "c", 1);
+    rocksdb_writebatch_delete(wb, "bar", 3);
+    rocksdb_transactiondb_write(txn_db, woptions, wb, &err);
+    rocksdb_writebatch_destroy(wb);
+    CheckTxnDBGet(txn_db, roptions, "box", "c");
+    CheckNoError(err);
+
     // begin a transaction
     txn = rocksdb_transaction_begin(txn_db, woptions, txn_options, NULL);
     // put
@@ -1413,6 +1463,42 @@ int main(int argc, char** argv) {
     CheckNoError(err);
     CheckTxnDBGet(txn_db, roptions, "bar", NULL);
 
+    // save point
+    rocksdb_transaction_put(txn, "foo1", 4, "hi1", 3, &err);
+    rocksdb_transaction_set_savepoint(txn);
+    CheckTxnGet(txn, roptions, "foo1", "hi1");
+    rocksdb_transaction_put(txn, "foo2", 4, "hi2", 3, &err);
+    CheckTxnGet(txn, roptions, "foo2", "hi2");
+
+    // rollback to savepoint
+    rocksdb_transaction_rollback_to_savepoint(txn, &err);
+    CheckNoError(err);
+    CheckTxnGet(txn, roptions, "foo2", NULL);
+    CheckTxnGet(txn, roptions, "foo1", "hi1");
+    CheckTxnDBGet(txn_db, roptions, "foo1", NULL);
+    CheckTxnDBGet(txn_db, roptions, "foo2", NULL);
+    rocksdb_transaction_commit(txn, &err);
+    CheckNoError(err);
+    CheckTxnDBGet(txn_db, roptions, "foo1", "hi1");
+    CheckTxnDBGet(txn_db, roptions, "foo2", NULL);
+
+    // Column families.
+    rocksdb_column_family_handle_t* cfh;
+    cfh = rocksdb_transactiondb_create_column_family(txn_db, options,
+                                                     "txn_db_cf", &err);
+    CheckNoError(err);
+
+    rocksdb_transactiondb_put_cf(txn_db, woptions, cfh, "cf_foo", 6, "cf_hello",
+                                 8, &err);
+    CheckNoError(err);
+    CheckTxnDBGetCF(txn_db, roptions, cfh, "cf_foo", "cf_hello");
+
+    rocksdb_transactiondb_delete_cf(txn_db, woptions, cfh, "cf_foo", 6, &err);
+    CheckNoError(err);
+    CheckTxnDBGetCF(txn_db, roptions, cfh, "cf_foo", NULL);
+
+    rocksdb_column_family_handle_destroy(cfh);
+
     // close and destroy
     rocksdb_transaction_destroy(txn);
     rocksdb_transactiondb_close(txn_db);
@@ -1422,6 +1508,105 @@ int main(int argc, char** argv) {
     rocksdb_transactiondb_options_destroy(txn_db_options);
   }
 
+  StartPhase("optimistic_transactions");
+  {
+    rocksdb_options_t* db_options = rocksdb_options_create();
+    rocksdb_options_set_create_if_missing(db_options, 1);
+    rocksdb_options_set_allow_concurrent_memtable_write(db_options, 1);
+    otxn_db = rocksdb_optimistictransactiondb_open(db_options, dbname, &err);
+    otxn_options = rocksdb_optimistictransaction_options_create();
+    rocksdb_transaction_t* txn1 = rocksdb_optimistictransaction_begin(
+        otxn_db, woptions, otxn_options, NULL);
+    rocksdb_transaction_t* txn2 = rocksdb_optimistictransaction_begin(
+        otxn_db, woptions, otxn_options, NULL);
+    rocksdb_transaction_put(txn1, "key", 3, "value", 5, &err);
+    CheckNoError(err);
+    rocksdb_transaction_put(txn2, "key1", 4, "value1", 6, &err);
+    CheckNoError(err);
+    CheckTxnGet(txn1, roptions, "key", "value");
+    rocksdb_transaction_commit(txn1, &err);
+    CheckNoError(err);
+    rocksdb_transaction_commit(txn2, &err);
+    CheckNoError(err);
+    rocksdb_transaction_destroy(txn1);
+    rocksdb_transaction_destroy(txn2);
+
+    // Check column family
+    db = rocksdb_optimistictransactiondb_get_base_db(otxn_db);
+    rocksdb_put(db, woptions, "key", 3, "value", 5, &err);
+    CheckNoError(err);
+    rocksdb_column_family_handle_t *cfh1, *cfh2;
+    cfh1 = rocksdb_create_column_family(db, db_options, "txn_db_cf1", &err);
+    cfh2 = rocksdb_create_column_family(db, db_options, "txn_db_cf2", &err);
+    txn = rocksdb_optimistictransaction_begin(otxn_db, woptions, otxn_options,
+                                              NULL);
+    rocksdb_transaction_put_cf(txn, cfh1, "key_cf1", 7, "val_cf1", 7, &err);
+    CheckNoError(err);
+    rocksdb_transaction_put_cf(txn, cfh2, "key_cf2", 7, "val_cf2", 7, &err);
+    CheckNoError(err);
+    rocksdb_transaction_commit(txn, &err);
+    CheckNoError(err);
+    txn = rocksdb_optimistictransaction_begin(otxn_db, woptions, otxn_options,
+                                              txn);
+    CheckGetCF(db, roptions, cfh1, "key_cf1", "val_cf1");
+    CheckTxnGetCF(txn, roptions, cfh1, "key_cf1", "val_cf1");
+
+    // Check iterator with column family
+    rocksdb_transaction_put_cf(txn, cfh1, "key1_cf", 7, "val1_cf", 7, &err);
+    CheckNoError(err);
+    rocksdb_iterator_t* iter =
+        rocksdb_transaction_create_iterator_cf(txn, roptions, cfh1);
+    CheckCondition(!rocksdb_iter_valid(iter));
+    rocksdb_iter_seek_to_first(iter);
+    CheckCondition(rocksdb_iter_valid(iter));
+    CheckIter(iter, "key1_cf", "val1_cf");
+    rocksdb_iter_get_error(iter, &err);
+    CheckNoError(err);
+    rocksdb_iter_destroy(iter);
+
+    rocksdb_transaction_destroy(txn);
+    rocksdb_column_family_handle_destroy(cfh1);
+    rocksdb_column_family_handle_destroy(cfh2);
+    rocksdb_optimistictransactiondb_close_base_db(db);
+    rocksdb_optimistictransactiondb_close(otxn_db);
+
+    // Check open optimistic transaction db with column families
+    size_t cf_len;
+    char** column_fams =
+        rocksdb_list_column_families(db_options, dbname, &cf_len, &err);
+    CheckNoError(err);
+    CheckEqual("default", column_fams[0], 7);
+    CheckEqual("txn_db_cf1", column_fams[1], 10);
+    CheckEqual("txn_db_cf2", column_fams[2], 10);
+    CheckCondition(cf_len == 3);
+    rocksdb_list_column_families_destroy(column_fams, cf_len);
+
+    const char* cf_names[3] = {"default", "txn_db_cf1", "txn_db_cf2"};
+    rocksdb_options_t* cf_options = rocksdb_options_create();
+    const rocksdb_options_t* cf_opts[3] = {cf_options, cf_options, cf_options};
+
+    rocksdb_options_set_error_if_exists(cf_options, 0);
+    rocksdb_column_family_handle_t* cf_handles[3];
+    otxn_db = rocksdb_optimistictransactiondb_open_column_families(
+        db_options, dbname, 3, cf_names, cf_opts, cf_handles, &err);
+    CheckNoError(err);
+    rocksdb_transaction_t* txn_cf = rocksdb_optimistictransaction_begin(
+        otxn_db, woptions, otxn_options, NULL);
+    CheckTxnGetCF(txn_cf, roptions, cf_handles[0], "key", "value");
+    CheckTxnGetCF(txn_cf, roptions, cf_handles[1], "key_cf1", "val_cf1");
+    CheckTxnGetCF(txn_cf, roptions, cf_handles[2], "key_cf2", "val_cf2");
+    rocksdb_transaction_destroy(txn_cf);
+    rocksdb_options_destroy(cf_options);
+    rocksdb_column_family_handle_destroy(cf_handles[0]);
+    rocksdb_column_family_handle_destroy(cf_handles[1]);
+    rocksdb_column_family_handle_destroy(cf_handles[2]);
+    rocksdb_optimistictransactiondb_close(otxn_db);
+    rocksdb_destroy_db(db_options, dbname, &err);
+    rocksdb_options_destroy(db_options);
+    rocksdb_optimistictransaction_options_destroy(otxn_options);
+    CheckNoError(err);
+  }
+
   // Simple sanity check that setting memtable rep works.
   StartPhase("memtable_reps");
   {
@@ -1440,6 +1625,18 @@ int main(int argc, char** argv) {
     CheckNoError(err);
   }
 
+  // Simple sanity check that options setting db_paths work.
+  StartPhase("open_db_paths");
+  {
+    rocksdb_close(db);
+    rocksdb_destroy_db(options, dbname, &err);
+
+    const rocksdb_dbpath_t* paths[1] = {dbpath};
+    rocksdb_options_set_db_paths(options, paths, 1);
+    db = rocksdb_open(options, dbname, &err);
+    CheckNoError(err);
+  }
+  
   StartPhase("cleanup");
   rocksdb_close(db);
   rocksdb_options_destroy(options);
@@ -1449,6 +1646,7 @@ int main(int argc, char** argv) {
   rocksdb_compactoptions_destroy(coptions);
   rocksdb_cache_destroy(cache);
   rocksdb_comparator_destroy(cmp);
+  rocksdb_dbpath_destroy(dbpath);
   rocksdb_env_destroy(env);
 
   fprintf(stderr, "PASS\n");

db/column_family.cc

@@ -123,6 +123,22 @@ Status CheckCompressionSupported(const ColumnFamilyOptions& cf_options) {
           " is not linked with the binary.");
     }
   }
+  if (cf_options.compression_opts.zstd_max_train_bytes > 0) {
+    if (!CompressionTypeSupported(CompressionType::kZSTD)) {
+      // Dictionary trainer is available since v0.6.1, but ZSTD was marked
+      // stable only since v0.8.0. For now we enable the feature in stable
+      // versions only.
+      return Status::InvalidArgument(
+          "zstd dictionary trainer cannot be used because " +
+          CompressionTypeToString(CompressionType::kZSTD) +
+          " is not linked with the binary.");
+    }
+    if (cf_options.compression_opts.max_dict_bytes == 0) {
+      return Status::InvalidArgument(
+          "The dictionary size limit (`CompressionOptions::max_dict_bytes`) "
+          "should be nonzero if we're using zstd's dictionary generator.");
+    }
+  }
   return Status::OK();
 }
 
@@ -369,7 +385,8 @@ ColumnFamilyData::ColumnFamilyData(
       pending_flush_(false),
       pending_compaction_(false),
       prev_compaction_needed_bytes_(0),
-      allow_2pc_(db_options.allow_2pc) {
+      allow_2pc_(db_options.allow_2pc),
+      last_memtable_id_(0) {
   Ref();
 
   // Convert user defined table properties collector factories to internal ones.
@@ -463,7 +480,8 @@ ColumnFamilyData::~ColumnFamilyData() {
   if (dummy_versions_ != nullptr) {
     // List must be empty
     assert(dummy_versions_->TEST_Next() == dummy_versions_);
-    bool deleted __attribute__((unused)) = dummy_versions_->Unref();
+    bool deleted __attribute__((unused));
+    deleted = dummy_versions_->Unref();
     assert(deleted);
   }
 
@@ -613,8 +631,9 @@ int GetL0ThresholdSpeedupCompaction(int level0_file_num_compaction_trigger,
 }
 }  // namespace
 
-void ColumnFamilyData::RecalculateWriteStallConditions(
+WriteStallCondition ColumnFamilyData::RecalculateWriteStallConditions(
       const MutableCFOptions& mutable_cf_options) {
+  auto write_stall_condition = WriteStallCondition::kNormal;
   if (current_ != nullptr) {
     auto* vstorage = current_->storage_info();
     auto write_controller = column_family_set_->write_controller_;
@@ -626,7 +645,8 @@ void ColumnFamilyData::RecalculateWriteStallConditions(
 
     if (imm()->NumNotFlushed() >= mutable_cf_options.max_write_buffer_number) {
       write_controller_token_ = write_controller->GetStopToken();
-      internal_stats_->AddCFStats(InternalStats::MEMTABLE_COMPACTION, 1);
+      internal_stats_->AddCFStats(InternalStats::MEMTABLE_LIMIT_STOPS, 1);
+      write_stall_condition = WriteStallCondition::kStopped;
       ROCKS_LOG_WARN(
           ioptions_.info_log,
           "[%s] Stopping writes because we have %d immutable memtables "
@@ -637,10 +657,11 @@ void ColumnFamilyData::RecalculateWriteStallConditions(
                vstorage->l0_delay_trigger_count() >=
                    mutable_cf_options.level0_stop_writes_trigger) {
       write_controller_token_ = write_controller->GetStopToken();
-      internal_stats_->AddCFStats(InternalStats::LEVEL0_NUM_FILES_TOTAL, 1);
+      internal_stats_->AddCFStats(InternalStats::L0_FILE_COUNT_LIMIT_STOPS, 1);
+      write_stall_condition = WriteStallCondition::kStopped;
       if (compaction_picker_->IsLevel0CompactionInProgress()) {
         internal_stats_->AddCFStats(
-            InternalStats::LEVEL0_NUM_FILES_WITH_COMPACTION, 1);
+            InternalStats::LOCKED_L0_FILE_COUNT_LIMIT_STOPS, 1);
       }
       ROCKS_LOG_WARN(ioptions_.info_log,
                      "[%s] Stopping writes because we have %d level-0 files",
@@ -651,7 +672,8 @@ void ColumnFamilyData::RecalculateWriteStallConditions(
                    mutable_cf_options.hard_pending_compaction_bytes_limit) {
       write_controller_token_ = write_controller->GetStopToken();
       internal_stats_->AddCFStats(
-          InternalStats::HARD_PENDING_COMPACTION_BYTES_LIMIT, 1);
+          InternalStats::PENDING_COMPACTION_BYTES_LIMIT_STOPS, 1);
+      write_stall_condition = WriteStallCondition::kStopped;
       ROCKS_LOG_WARN(
           ioptions_.info_log,
           "[%s] Stopping writes because of estimated pending compaction "
@@ -664,7 +686,8 @@ void ColumnFamilyData::RecalculateWriteStallConditions(
           SetupDelay(write_controller, compaction_needed_bytes,
                      prev_compaction_needed_bytes_, was_stopped,
                      mutable_cf_options.disable_auto_compactions);
-      internal_stats_->AddCFStats(InternalStats::MEMTABLE_SLOWDOWN, 1);
+      internal_stats_->AddCFStats(InternalStats::MEMTABLE_LIMIT_SLOWDOWNS, 1);
+      write_stall_condition = WriteStallCondition::kDelayed;
       ROCKS_LOG_WARN(
           ioptions_.info_log,
           "[%s] Stalling writes because we have %d immutable memtables "
@@ -684,10 +707,12 @@ void ColumnFamilyData::RecalculateWriteStallConditions(
           SetupDelay(write_controller, compaction_needed_bytes,
                      prev_compaction_needed_bytes_, was_stopped || near_stop,
                      mutable_cf_options.disable_auto_compactions);
-      internal_stats_->AddCFStats(InternalStats::LEVEL0_SLOWDOWN_TOTAL, 1);
+      internal_stats_->AddCFStats(InternalStats::L0_FILE_COUNT_LIMIT_SLOWDOWNS,
+                                  1);
+      write_stall_condition = WriteStallCondition::kDelayed;
       if (compaction_picker_->IsLevel0CompactionInProgress()) {
         internal_stats_->AddCFStats(
-            InternalStats::LEVEL0_SLOWDOWN_WITH_COMPACTION, 1);
+            InternalStats::LOCKED_L0_FILE_COUNT_LIMIT_SLOWDOWNS, 1);
       }
       ROCKS_LOG_WARN(ioptions_.info_log,
                      "[%s] Stalling writes because we have %d level-0 files "
@@ -714,7 +739,8 @@ void ColumnFamilyData::RecalculateWriteStallConditions(
                      prev_compaction_needed_bytes_, was_stopped || near_stop,
                      mutable_cf_options.disable_auto_compactions);
       internal_stats_->AddCFStats(
-          InternalStats::SOFT_PENDING_COMPACTION_BYTES_LIMIT, 1);
+          InternalStats::PENDING_COMPACTION_BYTES_LIMIT_SLOWDOWNS, 1);
+      write_stall_condition = WriteStallCondition::kDelayed;
       ROCKS_LOG_WARN(
           ioptions_.info_log,
           "[%s] Stalling writes because of estimated pending compaction "
@@ -769,6 +795,7 @@ void ColumnFamilyData::RecalculateWriteStallConditions(
     }
     prev_compaction_needed_bytes_ = compaction_needed_bytes;
   }
+  return write_stall_condition;
 }
 
 const EnvOptions* ColumnFamilyData::soptions() const {
@@ -845,6 +872,10 @@ SuperVersion* ColumnFamilyData::GetReferencedSuperVersion(
   sv = GetThreadLocalSuperVersion(db_mutex);
   sv->Ref();
   if (!ReturnThreadLocalSuperVersion(sv)) {
+    // This Unref() corresponds to the Ref() in GetThreadLocalSuperVersion()
+    // when the thread-local pointer was populated. So, the Ref() earlier in
+    // this function still prevents the returned SuperVersion* from being
+    // deleted out from under the caller.
     sv->Unref();
   }
   return sv;
@@ -914,15 +945,16 @@ bool ColumnFamilyData::ReturnThreadLocalSuperVersion(SuperVersion* sv) {
   return false;
 }
 
-SuperVersion* ColumnFamilyData::InstallSuperVersion(
-    SuperVersion* new_superversion, InstrumentedMutex* db_mutex) {
+void ColumnFamilyData::InstallSuperVersion(
+    SuperVersionContext* sv_context, InstrumentedMutex* db_mutex) {
   db_mutex->AssertHeld();
-  return InstallSuperVersion(new_superversion, db_mutex, mutable_cf_options_);
+  return InstallSuperVersion(sv_context, db_mutex, mutable_cf_options_);
 }
 
-SuperVersion* ColumnFamilyData::InstallSuperVersion(
-    SuperVersion* new_superversion, InstrumentedMutex* db_mutex,
+void ColumnFamilyData::InstallSuperVersion(
+    SuperVersionContext* sv_context, InstrumentedMutex* db_mutex,
     const MutableCFOptions& mutable_cf_options) {
+  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_);
@@ -930,16 +962,28 @@ SuperVersion* ColumnFamilyData::InstallSuperVersion(
   super_version_ = new_superversion;
   ++super_version_number_;
   super_version_->version_number = super_version_number_;
+  super_version_->write_stall_condition =
+      RecalculateWriteStallConditions(mutable_cf_options);
+
+  if (old_superversion != nullptr) {
+    if (old_superversion->mutable_cf_options.write_buffer_size !=
+        mutable_cf_options.write_buffer_size) {
+      mem_->UpdateWriteBufferSize(mutable_cf_options.write_buffer_size);
+    }
+    if (old_superversion->write_stall_condition !=
+        new_superversion->write_stall_condition) {
+      sv_context->PushWriteStallNotification(
+          old_superversion->write_stall_condition,
+          new_superversion->write_stall_condition, GetName(), ioptions());
+    }
+    if (old_superversion->Unref()) {
+      old_superversion->Cleanup();
+      sv_context->superversions_to_free.push_back(old_superversion);
+    }
+  }
+
   // Reset SuperVersions cached in thread local storage
   ResetThreadLocalSuperVersions();
-
-  RecalculateWriteStallConditions(mutable_cf_options);
-
-  if (old_superversion != nullptr && old_superversion->Unref()) {
-    old_superversion->Cleanup();
-    return old_superversion;  // will let caller delete outside of mutex
-  }
-  return nullptr;
 }
 
 void ColumnFamilyData::ResetThreadLocalSuperVersions() {
@@ -972,6 +1016,24 @@ Status ColumnFamilyData::SetOptions(
 }
 #endif  // ROCKSDB_LITE
 
+// REQUIRES: DB mutex held
+Env::WriteLifeTimeHint ColumnFamilyData::CalculateSSTWriteHint(int level) {
+  if (initial_cf_options_.compaction_style != kCompactionStyleLevel) {
+    return Env::WLTH_NOT_SET;
+  }
+  if (level == 0) {
+    return Env::WLTH_MEDIUM;
+  }
+  int base_level = current_->storage_info()->base_level();
+
+  // L1: medium, L2: long, ...
+  if (level - base_level >= 2) {
+    return Env::WLTH_EXTREME;
+  }
+  return static_cast<Env::WriteLifeTimeHint>(level - base_level +
+                            static_cast<int>(Env::WLTH_MEDIUM));
+}
+
 ColumnFamilySet::ColumnFamilySet(const std::string& dbname,
                                  const ImmutableDBOptions* db_options,
                                  const EnvOptions& env_options,
@@ -998,10 +1060,12 @@ ColumnFamilySet::~ColumnFamilySet() {
   while (column_family_data_.size() > 0) {
     // cfd destructor will delete itself from column_family_data_
     auto cfd = column_family_data_.begin()->second;
-    cfd->Unref();
+    bool last_ref __attribute__((__unused__)) = cfd->Unref();
+    assert(last_ref);
     delete cfd;
   }
-  dummy_cfd_->Unref();
+  bool dummy_last_ref __attribute__((__unused__)) = dummy_cfd_->Unref();
+  assert(dummy_last_ref);
   delete dummy_cfd_;
 }
 

db/column_family.h

@@ -41,6 +41,7 @@ class DBImpl;
 class LogBuffer;
 class InstrumentedMutex;
 class InstrumentedMutexLock;
+struct SuperVersionContext;
 
 extern const double kIncSlowdownRatio;
 
@@ -76,7 +77,7 @@ class ColumnFamilyHandleImpl : public ColumnFamilyHandle {
 class ColumnFamilyHandleInternal : public ColumnFamilyHandleImpl {
  public:
   ColumnFamilyHandleInternal()
-      : ColumnFamilyHandleImpl(nullptr, nullptr, nullptr) {}
+      : ColumnFamilyHandleImpl(nullptr, nullptr, nullptr), internal_cfd_(nullptr) {}
 
   void SetCFD(ColumnFamilyData* _cfd) { internal_cfd_ = _cfd; }
   virtual ColumnFamilyData* cfd() const override { return internal_cfd_; }
@@ -95,6 +96,7 @@ struct SuperVersion {
   MutableCFOptions mutable_cf_options;
   // Version number of the current SuperVersion
   uint64_t version_number;
+  WriteStallCondition write_stall_condition;
 
   InstrumentedMutex* db_mutex;
 
@@ -192,7 +194,7 @@ class ColumnFamilyData {
   // *) delete all memory associated with that column family
   // *) delete all the files associated with that column family
   void SetDropped();
-  bool IsDropped() const { return dropped_; }
+  bool IsDropped() const { return dropped_.load(std::memory_order_relaxed); }
 
   // thread-safe
   int NumberLevels() const { return ioptions_.num_levels; }
@@ -237,7 +239,11 @@ class ColumnFamilyData {
   void SetCurrent(Version* _current);
   uint64_t GetNumLiveVersions() const;  // REQUIRE: DB mutex held
   uint64_t GetTotalSstFilesSize() const;  // REQUIRE: DB mutex held
-  void SetMemtable(MemTable* new_mem) { mem_ = new_mem; }
+  void SetMemtable(MemTable* new_mem) {
+    uint64_t memtable_id = last_memtable_id_.fetch_add(1) + 1;
+    new_mem->SetID(memtable_id);
+    mem_ = new_mem;
+  }
 
   // calculate the oldest log needed for the durability of this column family
   uint64_t OldestLogToKeep();
@@ -264,7 +270,7 @@ class ColumnFamilyData {
                                   int level) const;
 
   // A flag to tell a manual compaction is to compact all levels together
-  // instad of for specific level.
+  // instead of a specific level.
   static const int kCompactAllLevels;
   // A flag to tell a manual compaction's output is base level.
   static const int kCompactToBaseLevel;
@@ -311,11 +317,11 @@ class ColumnFamilyData {
   // As argument takes a pointer to allocated SuperVersion to enable
   // the clients to allocate SuperVersion outside of mutex.
   // IMPORTANT: Only call this from DBImpl::InstallSuperVersion()
-  SuperVersion* InstallSuperVersion(SuperVersion* new_superversion,
-                                    InstrumentedMutex* db_mutex,
-                                    const MutableCFOptions& mutable_cf_options);
-  SuperVersion* InstallSuperVersion(SuperVersion* new_superversion,
-                                    InstrumentedMutex* db_mutex);
+  void InstallSuperVersion(SuperVersionContext* sv_context,
+                           InstrumentedMutex* db_mutex,
+                           const MutableCFOptions& mutable_cf_options);
+  void InstallSuperVersion(SuperVersionContext* sv_context,
+                           InstrumentedMutex* db_mutex);
 
   void ResetThreadLocalSuperVersions();
 
@@ -330,13 +336,15 @@ class ColumnFamilyData {
   // recalculation of compaction score. These values are used in
   // DBImpl::MakeRoomForWrite function to decide, if it need to make
   // a write stall
-  void RecalculateWriteStallConditions(
+  WriteStallCondition RecalculateWriteStallConditions(
       const MutableCFOptions& mutable_cf_options);
 
   void set_initialized() { initialized_.store(true); }
 
   bool initialized() const { return initialized_.load(); }
 
+  Env::WriteLifeTimeHint CalculateSSTWriteHint(int level);
+
  private:
   friend class ColumnFamilySet;
   ColumnFamilyData(uint32_t id, const std::string& name,
@@ -354,7 +362,7 @@ class ColumnFamilyData {
 
   std::atomic<int> refs_;      // outstanding references to ColumnFamilyData
   std::atomic<bool> initialized_;
-  bool dropped_;               // true if client dropped it
+  std::atomic<bool> dropped_;  // true if client dropped it
 
   const InternalKeyComparator internal_comparator_;
   std::vector<std::unique_ptr<IntTblPropCollectorFactory>>
@@ -415,6 +423,9 @@ class ColumnFamilyData {
 
   // if the database was opened with 2pc enabled
   bool allow_2pc_;
+
+  // Memtable id to track flush.
+  std::atomic<uint64_t> last_memtable_id_;
 };
 
 // ColumnFamilySet has interesting thread-safety requirements

db/column_family_test.cc

@@ -521,6 +521,7 @@ TEST_F(ColumnFamilyTest, DontReuseColumnFamilyID) {
   }
 }
 
+#ifndef ROCKSDB_LITE
 TEST_F(ColumnFamilyTest, CreateCFRaceWithGetAggProperty) {
   Open();
 
@@ -542,6 +543,7 @@ TEST_F(ColumnFamilyTest, CreateCFRaceWithGetAggProperty) {
 
   rocksdb::SyncPoint::GetInstance()->DisableProcessing();
 }
+#endif  // !ROCKSDB_LITE
 
 class FlushEmptyCFTestWithParam : public ColumnFamilyTest,
                                   public testing::WithParamInterface<bool> {
@@ -931,62 +933,62 @@ TEST_F(ColumnFamilyTest, FlushTest) {
 TEST_F(ColumnFamilyTest, LogDeletionTest) {
   db_options_.max_total_wal_size = std::numeric_limits<uint64_t>::max();
   column_family_options_.arena_block_size = 4 * 1024;
-  column_family_options_.write_buffer_size = 100000;  // 100KB
+  column_family_options_.write_buffer_size = 128000;  // 128KB
   Open();
   CreateColumnFamilies({"one", "two", "three", "four"});
   // Each bracket is one log file. if number is in (), it means
   // we don't need it anymore (it's been flushed)
   // []
   AssertCountLiveLogFiles(0);
-  PutRandomData(0, 1, 100);
+  PutRandomData(0, 1, 128);
   // [0]
-  PutRandomData(1, 1, 100);
+  PutRandomData(1, 1, 128);
   // [0, 1]
-  PutRandomData(1, 1000, 100);
+  PutRandomData(1, 1000, 128);
   WaitForFlush(1);
   // [0, (1)] [1]
   AssertCountLiveLogFiles(2);
-  PutRandomData(0, 1, 100);
+  PutRandomData(0, 1, 128);
   // [0, (1)] [0, 1]
   AssertCountLiveLogFiles(2);
-  PutRandomData(2, 1, 100);
+  PutRandomData(2, 1, 128);
   // [0, (1)] [0, 1, 2]
-  PutRandomData(2, 1000, 100);
+  PutRandomData(2, 1000, 128);
   WaitForFlush(2);
   // [0, (1)] [0, 1, (2)] [2]
   AssertCountLiveLogFiles(3);
-  PutRandomData(2, 1000, 100);
+  PutRandomData(2, 1000, 128);
   WaitForFlush(2);
   // [0, (1)] [0, 1, (2)] [(2)] [2]
   AssertCountLiveLogFiles(4);
-  PutRandomData(3, 1, 100);
+  PutRandomData(3, 1, 128);
   // [0, (1)] [0, 1, (2)] [(2)] [2, 3]
-  PutRandomData(1, 1, 100);
+  PutRandomData(1, 1, 128);
   // [0, (1)] [0, 1, (2)] [(2)] [1, 2, 3]
   AssertCountLiveLogFiles(4);
-  PutRandomData(1, 1000, 100);
+  PutRandomData(1, 1000, 128);
   WaitForFlush(1);
   // [0, (1)] [0, (1), (2)] [(2)] [(1), 2, 3] [1]
   AssertCountLiveLogFiles(5);
-  PutRandomData(0, 1000, 100);
+  PutRandomData(0, 1000, 128);
   WaitForFlush(0);
   // [(0), (1)] [(0), (1), (2)] [(2)] [(1), 2, 3] [1, (0)] [0]
   // delete obsolete logs -->
   // [(1), 2, 3] [1, (0)] [0]
   AssertCountLiveLogFiles(3);
-  PutRandomData(0, 1000, 100);
+  PutRandomData(0, 1000, 128);
   WaitForFlush(0);
   // [(1), 2, 3] [1, (0)], [(0)] [0]
   AssertCountLiveLogFiles(4);
-  PutRandomData(1, 1000, 100);
+  PutRandomData(1, 1000, 128);
   WaitForFlush(1);
   // [(1), 2, 3] [(1), (0)] [(0)] [0, (1)] [1]
   AssertCountLiveLogFiles(5);
-  PutRandomData(2, 1000, 100);
+  PutRandomData(2, 1000, 128);
   WaitForFlush(2);
   // [(1), (2), 3] [(1), (0)] [(0)] [0, (1)] [1, (2)], [2]
   AssertCountLiveLogFiles(6);
-  PutRandomData(3, 1000, 100);
+  PutRandomData(3, 1000, 128);
   WaitForFlush(3);
   // [(1), (2), (3)] [(1), (0)] [(0)] [0, (1)] [1, (2)], [2, (3)] [3]
   // delete obsolete logs -->
@@ -1866,126 +1868,6 @@ TEST_F(ColumnFamilyTest, SameCFManualAutomaticCompactionsLevel) {
   }
 }
 
-// This test checks for automatic getting a conflict if there is a
-// manual which has not yet been scheduled.
-// The manual compaction waits in NotScheduled
-// We generate more files and then trigger an automatic compaction
-// This will wait because there is an unscheduled manual compaction.
-// Once the conflict is hit, the manual compaction starts and ends
-// Then another automatic will start and end.
-TEST_F(ColumnFamilyTest, SameCFManualAutomaticConflict) {
-  Open();
-  CreateColumnFamilies({"one"});
-  ColumnFamilyOptions default_cf, one;
-  db_options_.max_open_files = 20;  // only 10 files in file cache
-  db_options_.max_background_compactions = 3;
-
-  default_cf.compaction_style = kCompactionStyleLevel;
-  default_cf.num_levels = 3;
-  default_cf.write_buffer_size = 64 << 10;  // 64KB
-  default_cf.target_file_size_base = 30 << 10;
-  default_cf.max_compaction_bytes = default_cf.target_file_size_base * 1100;
-  BlockBasedTableOptions table_options;
-  table_options.no_block_cache = true;
-  default_cf.table_factory.reset(NewBlockBasedTableFactory(table_options));
-
-  one.compaction_style = kCompactionStyleUniversal;
-
-  one.num_levels = 1;
-  // trigger compaction if there are >= 4 files
-  one.level0_file_num_compaction_trigger = 4;
-  one.write_buffer_size = 120000;
-
-  Reopen({default_cf, one});
-  // make sure all background compaction jobs can be scheduled
-  auto stop_token =
-      dbfull()->TEST_write_controler().GetCompactionPressureToken();
-
-  // SETUP column family "one" -- universal style
-  for (int i = 0; i < one.level0_file_num_compaction_trigger - 2; ++i) {
-    PutRandomData(1, 10, 12000, true);
-    PutRandomData(1, 1, 10, true);
-    WaitForFlush(1);
-    AssertFilesPerLevel(ToString(i + 1), 1);
-  }
-  bool cf_1_1 = true;
-  bool cf_1_2 = true;
-  rocksdb::SyncPoint::GetInstance()->LoadDependency(
-      {{"DBImpl::BackgroundCompaction()::Conflict",
-        "ColumnFamilyTest::ManualAutoCon:7"},
-       {"ColumnFamilyTest::ManualAutoCon:9",
-        "ColumnFamilyTest::ManualAutoCon:8"},
-       {"ColumnFamilyTest::ManualAutoCon:2",
-        "ColumnFamilyTest::ManualAutoCon:6"},
-       {"ColumnFamilyTest::ManualAutoCon:4",
-        "ColumnFamilyTest::ManualAutoCon:5"},
-       {"ColumnFamilyTest::ManualAutoCon:1",
-        "ColumnFamilyTest::ManualAutoCon:2"},
-       {"ColumnFamilyTest::ManualAutoCon:1",
-        "ColumnFamilyTest::ManualAutoCon:3"}});
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
-      "DBImpl::BackgroundCompaction:NonTrivial:AfterRun", [&](void* arg) {
-        if (cf_1_1) {
-          TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:4");
-          cf_1_1 = false;
-          TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:3");
-        } else if (cf_1_2) {
-          cf_1_2 = false;
-          TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:2");
-        }
-      });
-  rocksdb::SyncPoint::GetInstance()->SetCallBack(
-      "DBImpl::RunManualCompaction:NotScheduled", [&](void* arg) {
-        InstrumentedMutex* mutex = static_cast<InstrumentedMutex*>(arg);
-        mutex->Unlock();
-        TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:9");
-        TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:7");
-        mutex->Lock();
-      });
-
-  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
-  rocksdb::port::Thread threads([&] {
-    CompactRangeOptions compact_options;
-    compact_options.exclusive_manual_compaction = false;
-    ASSERT_OK(
-        db_->CompactRange(compact_options, handles_[1], nullptr, nullptr));
-    TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:6");
-  });
-
-  TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:8");
-  WaitForFlush(1);
-
-  // Add more L0 files and force automatic compaction
-  for (int i = 0; i < one.level0_file_num_compaction_trigger; ++i) {
-    PutRandomData(1, 10, 12000, true);
-    PutRandomData(1, 1, 10, true);
-    WaitForFlush(1);
-    AssertFilesPerLevel(ToString(one.level0_file_num_compaction_trigger + i),
-                        1);
-  }
-
-  TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:5");
-  // Add more L0 files and force automatic compaction
-  for (int i = 0; i < one.level0_file_num_compaction_trigger; ++i) {
-    PutRandomData(1, 10, 12000, true);
-    PutRandomData(1, 1, 10, true);
-    WaitForFlush(1);
-  }
-  TEST_SYNC_POINT("ColumnFamilyTest::ManualAutoCon:1");
-
-  threads.join();
-  WaitForCompaction();
-  // VERIFY compaction "one"
-  ASSERT_LE(NumTableFilesAtLevel(0, 1), 3);
-
-  // Compare against saved keys
-  std::set<std::string>::iterator key_iter = keys_.begin();
-  while (key_iter != keys_.end()) {
-    ASSERT_NE("NOT_FOUND", Get(1, *key_iter));
-    key_iter++;
-  }
-}
-
 // In this test, we generate enough files to trigger automatic compactions.
 // The automatic compaction waits in NonTrivial:AfterRun
 // We generate more files and then trigger an automatic compaction

db/compacted_db_impl.cc

@@ -17,7 +17,8 @@ extern bool SaveValue(void* arg, const ParsedInternalKey& parsed_key,
 
 CompactedDBImpl::CompactedDBImpl(
   const DBOptions& options, const std::string& dbname)
-  : DBImpl(options, dbname) {
+  : DBImpl(options, dbname), cfd_(nullptr), version_(nullptr),
+    user_comparator_(nullptr) {
 }
 
 CompactedDBImpl::~CompactedDBImpl() {
@@ -93,6 +94,7 @@ std::vector<Status> CompactedDBImpl::MultiGet(const ReadOptions& options,
 }
 
 Status CompactedDBImpl::Init(const Options& options) {
+  SuperVersionContext sv_context(/* create_superversion */ true);
   mutex_.Lock();
   ColumnFamilyDescriptor cf(kDefaultColumnFamilyName,
                             ColumnFamilyOptions(options));
@@ -100,9 +102,10 @@ Status CompactedDBImpl::Init(const Options& options) {
   if (s.ok()) {
     cfd_ = reinterpret_cast<ColumnFamilyHandleImpl*>(
               DefaultColumnFamily())->cfd();
-    delete cfd_->InstallSuperVersion(new SuperVersion(), &mutex_);
+    cfd_->InstallSuperVersion(&sv_context, &mutex_);
   }
   mutex_.Unlock();
+  sv_context.Clean();
   if (!s.ok()) {
     return s;
   }

db/compaction.cc

@@ -86,35 +86,23 @@ void Compaction::GetBoundaryKeys(
 bool Compaction::IsBottommostLevel(
     int output_level, VersionStorageInfo* vstorage,
     const std::vector<CompactionInputFiles>& inputs) {
-  if (inputs[0].level == 0 &&
-      inputs[0].files.back() != vstorage->LevelFiles(0).back()) {
-    return false;
+  int output_l0_idx;
+  if (output_level == 0) {
+    output_l0_idx = 0;
+    for (const auto* file : vstorage->LevelFiles(0)) {
+      if (inputs[0].files.back() == file) {
+        break;
+      }
+      ++output_l0_idx;
+    }
+    assert(static_cast<size_t>(output_l0_idx) < vstorage->LevelFiles(0).size());
+  } else {
+    output_l0_idx = -1;
   }
-
   Slice smallest_key, largest_key;
   GetBoundaryKeys(vstorage, inputs, &smallest_key, &largest_key);
-
-  // Checks whether there are files living beyond the output_level.
-  // If lower levels have files, it checks for overlap between files
-  // if the compaction process and those files.
-  // Bottomlevel optimizations can be made if there are no files in
-  // lower levels or if there is no overlap with the files in
-  // the lower levels.
-  for (int i = output_level + 1; i < vstorage->num_levels(); i++) {
-    // It is not the bottommost level if there are files in higher
-    // levels when the output level is 0 or if there are files in
-    // higher levels which overlap with files to be compacted.
-    // output_level == 0 means that we want it to be considered
-    // s the bottommost level only if the last file on the level
-    // is a part of the files to be compacted - this is verified by
-    // the first if condition in this function
-    if (vstorage->NumLevelFiles(i) > 0 &&
-        (output_level == 0 ||
-         vstorage->OverlapInLevel(i, &smallest_key, &largest_key))) {
-      return false;
-    }
-  }
-  return true;
+  return !vstorage->RangeMightExistAfterSortedRun(smallest_key, largest_key,
+                                                  output_level, output_l0_idx);
 }
 
 // test function to validate the functionality of IsBottommostLevel()
@@ -169,6 +157,7 @@ Compaction::Compaction(VersionStorageInfo* vstorage,
       bottommost_level_(IsBottommostLevel(output_level_, vstorage, inputs_)),
       is_full_compaction_(IsFullCompaction(vstorage, inputs_)),
       is_manual_compaction_(_manual_compaction),
+      is_trivial_move_(false),
       compaction_reason_(_compaction_reason) {
   MarkFilesBeingCompacted(true);
   if (is_manual_compaction_) {
@@ -240,7 +229,7 @@ bool Compaction::IsTrivialMove() const {
 
   // Used in universal compaction, where trivial move can be done if the
   // input files are non overlapping
-  if ((immutable_cf_options_.compaction_options_universal.allow_trivial_move) &&
+  if ((mutable_cf_options_.compaction_options_universal.allow_trivial_move) &&
       (output_level_ != 0)) {
     return is_trivial_move_;
   }
@@ -283,10 +272,10 @@ bool Compaction::KeyNotExistsBeyondOutputLevel(
   assert(input_version_ != nullptr);
   assert(level_ptrs != nullptr);
   assert(level_ptrs->size() == static_cast<size_t>(number_levels_));
-  if (cfd_->ioptions()->compaction_style == kCompactionStyleLevel) {
-    if (output_level_ == 0) {
-      return false;
-    }
+  if (bottommost_level_) {
+    return true;
+  } else if (output_level_ != 0 &&
+             cfd_->ioptions()->compaction_style == kCompactionStyleLevel) {
     // Maybe use binary search to find right entry instead of linear search?
     const Comparator* user_cmp = cfd_->user_comparator();
     for (int lvl = output_level_ + 1; lvl < number_levels_; lvl++) {
@@ -306,9 +295,8 @@ bool Compaction::KeyNotExistsBeyondOutputLevel(
       }
     }
     return true;
-  } else {
-    return bottommost_level_;
   }
+  return false;
 }
 
 // Mark (or clear) each file that is being compacted
@@ -417,20 +405,23 @@ void Compaction::Summary(char* output, int len) {
 uint64_t Compaction::OutputFilePreallocationSize() const {
   uint64_t preallocation_size = 0;
 
-  if (max_output_file_size_ != port::kMaxUint64 &&
-      (cfd_->ioptions()->compaction_style == kCompactionStyleLevel ||
-       output_level() > 0)) {
-    preallocation_size = max_output_file_size_;
-  } else {
-    for (const auto& level_files : inputs_) {
-      for (const auto& file : level_files.files) {
-        preallocation_size += file->fd.GetFileSize();
-      }
+  for (const auto& level_files : inputs_) {
+    for (const auto& file : level_files.files) {
+      preallocation_size += file->fd.GetFileSize();
     }
   }
+
+  if (max_output_file_size_ != port::kMaxUint64 &&
+      (immutable_cf_options_.compaction_style == kCompactionStyleLevel ||
+       output_level() > 0)) {
+    preallocation_size = std::min(max_output_file_size_, preallocation_size);
+  }
+
   // Over-estimate slightly so we don't end up just barely crossing
   // the threshold
-  return preallocation_size + (preallocation_size / 10);
+  // No point to prellocate more than 1GB.
+  return std::min(uint64_t{1073741824},
+                  preallocation_size + (preallocation_size / 10));
 }
 
 std::unique_ptr<CompactionFilter> Compaction::CreateCompactionFilter() const {

db/compaction_iteration_stats.h

@@ -16,6 +16,8 @@ struct CompactionIterationStats {
   int64_t num_record_drop_obsolete = 0;
   int64_t num_record_drop_range_del = 0;
   int64_t num_range_del_drop_obsolete = 0;
+  // Deletions obsoleted before bottom level due to file gap optimization.
+  int64_t num_optimized_del_drop_obsolete = 0;
   uint64_t total_filter_time = 0;
 
   // Input statistics

db/compaction_iterator.cc

@@ -4,6 +4,9 @@
 //  (found in the LICENSE.Apache file in the root directory).
 
 #include "db/compaction_iterator.h"
+
+#include "db/snapshot_checker.h"
+#include "port/likely.h"
 #include "rocksdb/listener.h"
 #include "table/internal_iterator.h"
 
@@ -25,6 +28,8 @@ CompactionEventListener::CompactionListenerValueType fromInternalValueType(
           kSingleDelete;
     case kTypeRangeDeletion:
       return CompactionEventListener::CompactionListenerValueType::kRangeDelete;
+    case kTypeBlobIndex:
+      return CompactionEventListener::CompactionListenerValueType::kBlobIndex;
     default:
       assert(false);
       return CompactionEventListener::CompactionListenerValueType::kInvalid;
@@ -35,33 +40,40 @@ CompactionEventListener::CompactionListenerValueType fromInternalValueType(
 CompactionIterator::CompactionIterator(
     InternalIterator* input, const Comparator* cmp, MergeHelper* merge_helper,
     SequenceNumber last_sequence, std::vector<SequenceNumber>* snapshots,
-    SequenceNumber earliest_write_conflict_snapshot, Env* env,
+    SequenceNumber earliest_write_conflict_snapshot,
+    const SnapshotChecker* snapshot_checker, Env* env,
     bool expect_valid_internal_key, RangeDelAggregator* range_del_agg,
     const Compaction* compaction, const CompactionFilter* compaction_filter,
     CompactionEventListener* compaction_listener,
-    const std::atomic<bool>* shutting_down)
+    const std::atomic<bool>* shutting_down,
+    const SequenceNumber preserve_deletes_seqnum)
     : CompactionIterator(
           input, cmp, merge_helper, last_sequence, snapshots,
-          earliest_write_conflict_snapshot, env, expect_valid_internal_key,
-          range_del_agg,
+          earliest_write_conflict_snapshot, snapshot_checker, env,
+          expect_valid_internal_key, range_del_agg,
           std::unique_ptr<CompactionProxy>(
               compaction ? new CompactionProxy(compaction) : nullptr),
-          compaction_filter, compaction_listener, shutting_down) {}
+          compaction_filter, compaction_listener, shutting_down,
+          preserve_deletes_seqnum) {}
 
 CompactionIterator::CompactionIterator(
     InternalIterator* input, const Comparator* cmp, MergeHelper* merge_helper,
     SequenceNumber last_sequence, std::vector<SequenceNumber>* snapshots,
-    SequenceNumber earliest_write_conflict_snapshot, Env* env,
+    SequenceNumber earliest_write_conflict_snapshot,
+    const SnapshotChecker* snapshot_checker, Env* env,
     bool expect_valid_internal_key, RangeDelAggregator* range_del_agg,
     std::unique_ptr<CompactionProxy> compaction,
     const CompactionFilter* compaction_filter,
     CompactionEventListener* compaction_listener,
-    const std::atomic<bool>* shutting_down)
+    const std::atomic<bool>* shutting_down,
+    const SequenceNumber preserve_deletes_seqnum
+  )
     : input_(input),
       cmp_(cmp),
       merge_helper_(merge_helper),
       snapshots_(snapshots),
       earliest_write_conflict_snapshot_(earliest_write_conflict_snapshot),
+      snapshot_checker_(snapshot_checker),
       env_(env),
       expect_valid_internal_key_(expect_valid_internal_key),
       range_del_agg_(range_del_agg),
@@ -71,8 +83,12 @@ CompactionIterator::CompactionIterator(
       compaction_listener_(compaction_listener),
 #endif  // ROCKSDB_LITE
       shutting_down_(shutting_down),
+      preserve_deletes_seqnum_(preserve_deletes_seqnum),
       ignore_snapshots_(false),
-      merge_out_iter_(merge_helper_) {
+      current_user_key_sequence_(0),
+      current_user_key_snapshot_(0),
+      merge_out_iter_(merge_helper_),
+      current_key_committed_(false) {
   assert(compaction_filter_ == nullptr || compaction_ != nullptr);
   bottommost_level_ =
       compaction_ == nullptr ? false : compaction_->bottommost_level();
@@ -111,6 +127,7 @@ void CompactionIterator::ResetRecordCounts() {
   iter_stats_.num_record_drop_obsolete = 0;
   iter_stats_.num_record_drop_range_del = 0;
   iter_stats_.num_range_del_drop_obsolete = 0;
+  iter_stats_.num_optimized_del_drop_obsolete = 0;
 }
 
 void CompactionIterator::SeekToFirst() {
@@ -128,8 +145,8 @@ void CompactionIterator::Next() {
     if (merge_out_iter_.Valid()) {
       key_ = merge_out_iter_.key();
       value_ = merge_out_iter_.value();
-      bool valid_key __attribute__((__unused__)) =
-          ParseInternalKey(key_, &ikey_);
+      bool valid_key __attribute__((__unused__));
+      valid_key =  ParseInternalKey(key_, &ikey_);
       // MergeUntil stops when it encounters a corrupt key and does not
       // include them in the result, so we expect the keys here to be valid.
       assert(valid_key);
@@ -163,6 +180,61 @@ void CompactionIterator::Next() {
   PrepareOutput();
 }
 
+void CompactionIterator::InvokeFilterIfNeeded(bool* need_skip,
+                                              Slice* skip_until) {
+  if (compaction_filter_ != nullptr &&
+      (ikey_.type == kTypeValue || ikey_.type == kTypeBlobIndex) &&
+      (visible_at_tip_ || ignore_snapshots_ ||
+       ikey_.sequence > latest_snapshot_ ||
+       (snapshot_checker_ != nullptr &&
+        UNLIKELY(!snapshot_checker_->IsInSnapshot(ikey_.sequence,
+                                                  latest_snapshot_))))) {
+    // If the user has specified a compaction filter and the sequence
+    // number is greater than any external snapshot, then invoke the
+    // filter. If the return value of the compaction filter is true,
+    // replace the entry with a deletion marker.
+    CompactionFilter::Decision filter;
+    compaction_filter_value_.clear();
+    compaction_filter_skip_until_.Clear();
+    CompactionFilter::ValueType value_type =
+        ikey_.type == kTypeValue ? CompactionFilter::ValueType::kValue
+                                 : CompactionFilter::ValueType::kBlobIndex;
+    {
+      StopWatchNano timer(env_, true);
+      filter = compaction_filter_->FilterV2(
+          compaction_->level(), ikey_.user_key, value_type, value_,
+          &compaction_filter_value_, compaction_filter_skip_until_.rep());
+      iter_stats_.total_filter_time +=
+          env_ != nullptr ? timer.ElapsedNanos() : 0;
+    }
+
+    if (filter == CompactionFilter::Decision::kRemoveAndSkipUntil &&
+        cmp_->Compare(*compaction_filter_skip_until_.rep(), ikey_.user_key) <=
+            0) {
+      // Can't skip to a key smaller than the current one.
+      // Keep the key as per FilterV2 documentation.
+      filter = CompactionFilter::Decision::kKeep;
+    }
+
+    if (filter == CompactionFilter::Decision::kRemove) {
+      // convert the current key to a delete; key_ is pointing into
+      // current_key_ at this point, so updating current_key_ updates key()
+      ikey_.type = kTypeDeletion;
+      current_key_.UpdateInternalKey(ikey_.sequence, kTypeDeletion);
+      // no value associated with delete
+      value_.clear();
+      iter_stats_.num_record_drop_user++;
+    } else if (filter == CompactionFilter::Decision::kChangeValue) {
+      value_ = compaction_filter_value_;
+    } else if (filter == CompactionFilter::Decision::kRemoveAndSkipUntil) {
+      *need_skip = true;
+      compaction_filter_skip_until_.ConvertFromUserKey(kMaxSequenceNumber,
+                                                       kValueTypeForSeek);
+      *skip_until = compaction_filter_skip_until_.Encode();
+    }
+  }
+}
+
 void CompactionIterator::NextFromInput() {
   at_next_ = false;
   valid_ = false;
@@ -217,6 +289,9 @@ void CompactionIterator::NextFromInput() {
       has_outputted_key_ = false;
       current_user_key_sequence_ = kMaxSequenceNumber;
       current_user_key_snapshot_ = 0;
+      current_key_committed_ =
+          (snapshot_checker_ == nullptr ||
+           snapshot_checker_->IsInSnapshot(ikey_.sequence, kMaxSequenceNumber));
 
 #ifndef ROCKSDB_LITE
       if (compaction_listener_) {
@@ -224,53 +299,12 @@ void CompactionIterator::NextFromInput() {
                                            fromInternalValueType(ikey_.type),
                                            value_, ikey_.sequence, true);
       }
-#endif  // ROCKSDB_LITE
+#endif  // !ROCKSDB_LITE
 
-      // apply the compaction filter to the first occurrence of the user key
-      if (compaction_filter_ != nullptr && ikey_.type == kTypeValue &&
-          (visible_at_tip_ || ikey_.sequence > latest_snapshot_ ||
-           ignore_snapshots_)) {
-        // If the user has specified a compaction filter and the sequence
-        // number is greater than any external snapshot, then invoke the
-        // filter. If the return value of the compaction filter is true,
-        // replace the entry with a deletion marker.
-        CompactionFilter::Decision filter;
-        compaction_filter_value_.clear();
-        compaction_filter_skip_until_.Clear();
-        {
-          StopWatchNano timer(env_, true);
-          filter = compaction_filter_->FilterV2(
-              compaction_->level(), ikey_.user_key,
-              CompactionFilter::ValueType::kValue, value_,
-              &compaction_filter_value_, compaction_filter_skip_until_.rep());
-          iter_stats_.total_filter_time +=
-              env_ != nullptr ? timer.ElapsedNanos() : 0;
-        }
-
-        if (filter == CompactionFilter::Decision::kRemoveAndSkipUntil &&
-            cmp_->Compare(*compaction_filter_skip_until_.rep(),
-                          ikey_.user_key) <= 0) {
-          // Can't skip to a key smaller than the current one.
-          // Keep the key as per FilterV2 documentation.
-          filter = CompactionFilter::Decision::kKeep;
-        }
-
-        if (filter == CompactionFilter::Decision::kRemove) {
-          // convert the current key to a delete; key_ is pointing into
-          // current_key_ at this point, so updating current_key_ updates key()
-          ikey_.type = kTypeDeletion;
-          current_key_.UpdateInternalKey(ikey_.sequence, kTypeDeletion);
-          // no value associated with delete
-          value_.clear();
-          iter_stats_.num_record_drop_user++;
-        } else if (filter == CompactionFilter::Decision::kChangeValue) {
-          value_ = compaction_filter_value_;
-        } else if (filter == CompactionFilter::Decision::kRemoveAndSkipUntil) {
-          need_skip = true;
-          compaction_filter_skip_until_.ConvertFromUserKey(kMaxSequenceNumber,
-                                                           kValueTypeForSeek);
-          skip_until = compaction_filter_skip_until_.Encode();
-        }
+      // Apply the compaction filter to the first committed version of the user
+      // key.
+      if (current_key_committed_) {
+        InvokeFilterIfNeeded(&need_skip, &skip_until);
       }
     } else {
 #ifndef ROCKSDB_LITE
@@ -289,13 +323,33 @@ void CompactionIterator::NextFromInput() {
       current_key_.UpdateInternalKey(ikey_.sequence, ikey_.type);
       key_ = current_key_.GetInternalKey();
       ikey_.user_key = current_key_.GetUserKey();
+
+      // Note that newer version of a key is ordered before older versions. If a
+      // newer version of a key is committed, so as the older version. No need
+      // to query snapshot_checker_ in that case.
+      if (UNLIKELY(!current_key_committed_)) {
+        assert(snapshot_checker_ != nullptr);
+        current_key_committed_ =
+            snapshot_checker_->IsInSnapshot(ikey_.sequence, kMaxSequenceNumber);
+        // Apply the compaction filter to the first committed version of the
+        // user key.
+        if (current_key_committed_) {
+          InvokeFilterIfNeeded(&need_skip, &skip_until);
+        }
+      }
+    }
+
+    if (UNLIKELY(!current_key_committed_)) {
+      assert(snapshot_checker_ != nullptr);
+      valid_ = true;
+      break;
     }
 
     // If there are no snapshots, then this kv affect visibility at tip.
     // Otherwise, search though all existing snapshots to find the earliest
     // snapshot that is affected by this kv.
-    SequenceNumber last_sequence __attribute__((__unused__)) =
-        current_user_key_sequence_;
+    SequenceNumber last_sequence __attribute__((__unused__));
+    last_sequence = current_user_key_sequence_;
     current_user_key_sequence_ = ikey_.sequence;
     SequenceNumber last_snapshot = current_user_key_snapshot_;
     SequenceNumber prev_snapshot = 0;  // 0 means no previous snapshot
@@ -360,7 +414,10 @@ void CompactionIterator::NextFromInput() {
           cmp_->Equal(ikey_.user_key, next_ikey.user_key)) {
         // Check whether the next key belongs to the same snapshot as the
         // SingleDelete.
-        if (prev_snapshot == 0 || next_ikey.sequence > prev_snapshot) {
+        if (prev_snapshot == 0 || next_ikey.sequence > prev_snapshot ||
+            (snapshot_checker_ != nullptr &&
+             UNLIKELY(!snapshot_checker_->IsInSnapshot(next_ikey.sequence,
+                                                       prev_snapshot)))) {
           if (next_ikey.type == kTypeSingleDeletion) {
             // We encountered two SingleDeletes in a row.  This could be due to
             // unexpected user input.
@@ -371,8 +428,12 @@ void CompactionIterator::NextFromInput() {
             // input_->Next().
             ++iter_stats_.num_record_drop_obsolete;
             ++iter_stats_.num_single_del_mismatch;
-          } else if ((ikey_.sequence <= earliest_write_conflict_snapshot_) ||
-                     has_outputted_key_) {
+          } else if (has_outputted_key_ ||
+                     (ikey_.sequence <= earliest_write_conflict_snapshot_ &&
+                      (snapshot_checker_ == nullptr ||
+                       LIKELY(snapshot_checker_->IsInSnapshot(
+                           ikey_.sequence,
+                           earliest_write_conflict_snapshot_))))) {
             // Found a matching value, we can drop the single delete and the
             // value.  It is safe to drop both records since we've already
             // outputted a key in this snapshot, or there is no earlier
@@ -420,12 +481,18 @@ void CompactionIterator::NextFromInput() {
         // comparison, so the value of has_current_user_key does not matter.
         has_current_user_key_ = false;
         if (compaction_ != nullptr && ikey_.sequence <= earliest_snapshot_ &&
+            (snapshot_checker_ == nullptr ||
+             LIKELY(snapshot_checker_->IsInSnapshot(ikey_.sequence,
+                                                    earliest_snapshot_))) &&
             compaction_->KeyNotExistsBeyondOutputLevel(ikey_.user_key,
                                                        &level_ptrs_)) {
           // Key doesn't exist outside of this range.
           // Can compact out this SingleDelete.
           ++iter_stats_.num_record_drop_obsolete;
           ++iter_stats_.num_single_del_fallthru;
+          if (!bottommost_level_) {
+            ++iter_stats_.num_optimized_del_drop_obsolete;
+          }
         } else {
           // Output SingleDelete
           valid_ = true;
@@ -450,6 +517,10 @@ void CompactionIterator::NextFromInput() {
       input_->Next();
     } else if (compaction_ != nullptr && ikey_.type == kTypeDeletion &&
                ikey_.sequence <= earliest_snapshot_ &&
+               (snapshot_checker_ == nullptr ||
+                LIKELY(snapshot_checker_->IsInSnapshot(ikey_.sequence,
+                                                       earliest_snapshot_))) &&
+               ikeyNotNeededForIncrementalSnapshot() &&
                compaction_->KeyNotExistsBeyondOutputLevel(ikey_.user_key,
                                                           &level_ptrs_)) {
       // TODO(noetzli): This is the only place where we use compaction_
@@ -467,6 +538,9 @@ void CompactionIterator::NextFromInput() {
       // Note:  Dropping this Delete will not affect TransactionDB
       // write-conflict checking since it is earlier than any snapshot.
       ++iter_stats_.num_record_drop_obsolete;
+      if (!bottommost_level_) {
+        ++iter_stats_.num_optimized_del_drop_obsolete;
+      }
       input_->Next();
     } else if (ikey_.type == kTypeMerge) {
       if (!merge_helper_->HasOperator()) {
@@ -480,6 +554,10 @@ void CompactionIterator::NextFromInput() {
       // have hit (A)
       // We encapsulate the merge related state machine in a different
       // object to minimize change to the existing flow.
+      // In case snapshot_checker is present, we can probably merge further
+      // beyond prev_snapshot, since there could be more keys with sequence
+      // smaller than prev_snapshot, but reported by snapshot_checker as not
+      // visible by prev_snapshot. But it will make the logic more complicated.
       Status s = merge_helper_->MergeUntil(input_, range_del_agg_,
                                            prev_snapshot, bottommost_level_);
       merge_out_iter_.SeekToFirst();
@@ -492,8 +570,8 @@ void CompactionIterator::NextFromInput() {
         //       These will be correctly set below.
         key_ = merge_out_iter_.key();
         value_ = merge_out_iter_.value();
-        bool valid_key __attribute__((__unused__)) =
-            ParseInternalKey(key_, &ikey_);
+        bool valid_key __attribute__((__unused__));
+        valid_key = ParseInternalKey(key_, &ikey_);
         // MergeUntil stops when it encounters a corrupt key and does not
         // include them in the result, so we expect the keys here to valid.
         assert(valid_key);
@@ -546,8 +624,12 @@ void CompactionIterator::PrepareOutput() {
 
   // This is safe for TransactionDB write-conflict checking since transactions
   // only care about sequence number larger than any active snapshots.
-  if ((compaction_ != nullptr && !compaction_->allow_ingest_behind()) &&
+  if ((compaction_ != nullptr &&
+      !compaction_->allow_ingest_behind()) &&
+      ikeyNotNeededForIncrementalSnapshot() &&
       bottommost_level_ && valid_ && ikey_.sequence <= earliest_snapshot_ &&
+      (snapshot_checker_ == nullptr || LIKELY(snapshot_checker_->IsInSnapshot(
+        ikey_.sequence, earliest_snapshot_))) &&
       ikey_.type != kTypeMerge &&
       !cmp_->Equal(compaction_->GetLargestUserKey(), ikey_.user_key)) {
     assert(ikey_.type != kTypeDeletion && ikey_.type != kTypeSingleDeletion);
@@ -559,10 +641,11 @@ void CompactionIterator::PrepareOutput() {
 inline SequenceNumber CompactionIterator::findEarliestVisibleSnapshot(
     SequenceNumber in, SequenceNumber* prev_snapshot) {
   assert(snapshots_->size());
-  SequenceNumber prev __attribute__((__unused__)) = kMaxSequenceNumber;
+  SequenceNumber prev = kMaxSequenceNumber;
   for (const auto cur : *snapshots_) {
     assert(prev == kMaxSequenceNumber || prev <= cur);
-    if (cur >= in) {
+    if (cur >= in && (snapshot_checker_ == nullptr ||
+                      snapshot_checker_->IsInSnapshot(in, cur))) {
       *prev_snapshot = prev == kMaxSequenceNumber ? 0 : prev;
       return cur;
     }
@@ -573,4 +656,11 @@ inline SequenceNumber CompactionIterator::findEarliestVisibleSnapshot(
   return kMaxSequenceNumber;
 }
 
+// used in 2 places - prevents deletion markers to be dropped if they may be
+// needed and disables seqnum zero-out in PrepareOutput for recent keys.
+inline bool CompactionIterator::ikeyNotNeededForIncrementalSnapshot() {
+  return (!compaction_->preserve_deletes()) ||
+         (ikey_.sequence < preserve_deletes_seqnum_);
+}
+
 }  // namespace rocksdb

db/compaction_iterator.h

@@ -14,6 +14,7 @@
 #include "db/merge_helper.h"
 #include "db/pinned_iterators_manager.h"
 #include "db/range_del_aggregator.h"
+#include "db/snapshot_checker.h"
 #include "options/cf_options.h"
 #include "rocksdb/compaction_filter.h"
 
@@ -48,6 +49,9 @@ class CompactionIterator {
     virtual bool allow_ingest_behind() const {
       return compaction_->immutable_cf_options()->allow_ingest_behind;
     }
+    virtual bool preserve_deletes() const {
+      return compaction_->immutable_cf_options()->preserve_deletes;
+    }
 
    protected:
     CompactionProxy() = default;
@@ -59,25 +63,29 @@ class CompactionIterator {
   CompactionIterator(InternalIterator* input, const Comparator* cmp,
                      MergeHelper* merge_helper, SequenceNumber last_sequence,
                      std::vector<SequenceNumber>* snapshots,
-                     SequenceNumber earliest_write_conflict_snapshot, Env* env,
+                     SequenceNumber earliest_write_conflict_snapshot,
+                     const SnapshotChecker* snapshot_checker, Env* env,
                      bool expect_valid_internal_key,
                      RangeDelAggregator* range_del_agg,
                      const Compaction* compaction = nullptr,
                      const CompactionFilter* compaction_filter = nullptr,
                      CompactionEventListener* compaction_listener = nullptr,
-                     const std::atomic<bool>* shutting_down = nullptr);
+                     const std::atomic<bool>* shutting_down = nullptr,
+                     const SequenceNumber preserve_deletes_seqnum = 0);
 
   // Constructor with custom CompactionProxy, used for tests.
   CompactionIterator(InternalIterator* input, const Comparator* cmp,
                      MergeHelper* merge_helper, SequenceNumber last_sequence,
                      std::vector<SequenceNumber>* snapshots,
-                     SequenceNumber earliest_write_conflict_snapshot, Env* env,
+                     SequenceNumber earliest_write_conflict_snapshot,
+                     const SnapshotChecker* snapshot_checker, Env* env,
                      bool expect_valid_internal_key,
                      RangeDelAggregator* range_del_agg,
                      std::unique_ptr<CompactionProxy> compaction,
                      const CompactionFilter* compaction_filter = nullptr,
                      CompactionEventListener* compaction_listener = nullptr,
-                     const std::atomic<bool>* shutting_down = nullptr);
+                     const std::atomic<bool>* shutting_down = nullptr,
+                     const SequenceNumber preserve_deletes_seqnum = 0);
 
   ~CompactionIterator();
 
@@ -111,6 +119,9 @@ class CompactionIterator {
   // compression.
   void PrepareOutput();
 
+  // Invoke compaction filter if needed.
+  void InvokeFilterIfNeeded(bool* need_skip, Slice* skip_until);
+
   // Given a sequence number, return the sequence number of the
   // earliest snapshot that this sequence number is visible in.
   // The snapshots themselves are arranged in ascending order of
@@ -120,11 +131,17 @@ class CompactionIterator {
   inline SequenceNumber findEarliestVisibleSnapshot(
       SequenceNumber in, SequenceNumber* prev_snapshot);
 
+  // Checks whether the currently seen ikey_ is needed for
+  // incremental (differential) snapshot and hence can't be dropped
+  // or seqnum be zero-ed out even if all other conditions for it are met.
+  inline bool ikeyNotNeededForIncrementalSnapshot();
+
   InternalIterator* input_;
   const Comparator* cmp_;
   MergeHelper* merge_helper_;
   const std::vector<SequenceNumber>* snapshots_;
   const SequenceNumber earliest_write_conflict_snapshot_;
+  const SnapshotChecker* const snapshot_checker_;
   Env* env_;
   bool expect_valid_internal_key_;
   RangeDelAggregator* range_del_agg_;
@@ -132,8 +149,9 @@ class CompactionIterator {
   const CompactionFilter* compaction_filter_;
 #ifndef ROCKSDB_LITE
   CompactionEventListener* compaction_listener_;
-#endif  // ROCKSDB_LITE
+#endif  // !ROCKSDB_LITE
   const std::atomic<bool>* shutting_down_;
+  const SequenceNumber preserve_deletes_seqnum_;
   bool bottommost_level_;
   bool valid_ = false;
   bool visible_at_tip_;
@@ -189,6 +207,10 @@ class CompactionIterator {
   std::vector<size_t> level_ptrs_;
   CompactionIterationStats iter_stats_;
 
+  // Used to avoid purging uncommitted values. The application can specify
+  // uncommitted values by providing a SnapshotChecker object.
+  bool current_key_committed_;
+
   bool IsShuttingDown() {
     // This is a best-effort facility, so memory_order_relaxed is sufficient.
     return shutting_down_ && shutting_down_->load(std::memory_order_relaxed);

db/compaction_iterator_test.cc

@@ -156,6 +156,8 @@ class FakeCompaction : public CompactionIterator::CompactionProxy {
   }
   virtual bool allow_ingest_behind() const { return false; }
 
+  virtual bool preserve_deletes() const {return false; }
+
   bool key_not_exists_beyond_output_level = false;
 };
 
@@ -181,6 +183,8 @@ class CompactionIteratorTest : public testing::Test {
       compaction_proxy_ = new FakeCompaction();
       compaction.reset(compaction_proxy_);
     }
+    // TODO(yiwu) add a mock snapshot checker and add test for it.
+    SnapshotChecker* snapshot_checker = nullptr;
 
     merge_helper_.reset(new MergeHelper(Env::Default(), cmp_, merge_op, filter,
                                         nullptr, false, 0, 0, nullptr,
@@ -189,8 +193,9 @@ class CompactionIteratorTest : public testing::Test {
     iter_->SeekToFirst();
     c_iter_.reset(new CompactionIterator(
         iter_.get(), cmp_, merge_helper_.get(), last_sequence, &snapshots_,
-        kMaxSequenceNumber, Env::Default(), false, range_del_agg_.get(),
-        std::move(compaction), filter, nullptr, &shutting_down_));
+        kMaxSequenceNumber, snapshot_checker, Env::Default(), false,
+        range_del_agg_.get(), std::move(compaction), filter, nullptr,
+        &shutting_down_));
   }
 
   void AddSnapshot(SequenceNumber snapshot) { snapshots_.push_back(snapshot); }
@@ -455,6 +460,111 @@ TEST_F(CompactionIteratorTest, ShuttingDownInMerge) {
   EXPECT_EQ(2, filter.last_seen.load());
 }
 
+TEST_F(CompactionIteratorTest, SingleMergeOperand) {
+  class Filter : public CompactionFilter {
+    virtual Decision FilterV2(int level, const Slice& key, ValueType t,
+                              const Slice& existing_value,
+                              std::string* new_value,
+                              std::string* skip_until) const override {
+      std::string k = key.ToString();
+      std::string v = existing_value.ToString();
+
+      // See InitIterators() call below for the sequence of keys and their
+      // filtering decisions. Here we closely assert that compaction filter is
+      // called with the expected keys and only them, and with the right values.
+      if (k == "a") {
+        EXPECT_EQ(ValueType::kMergeOperand, t);
+        EXPECT_EQ("av1", v);
+        return Decision::kKeep;
+      } else if (k == "b") {
+        EXPECT_EQ(ValueType::kMergeOperand, t);
+        return Decision::kKeep;
+      } else if (k == "c") {
+        return Decision::kKeep;
+      }
+
+      ADD_FAILURE();
+      return Decision::kKeep;
+    }
+
+    const char* Name() const override {
+      return "CompactionIteratorTest.SingleMergeOperand::Filter";
+    }
+  };
+
+  class SingleMergeOp : public MergeOperator {
+   public:
+    bool FullMergeV2(const MergeOperationInput& merge_in,
+                     MergeOperationOutput* merge_out) const override {
+      // See InitIterators() call below for why "c" is the only key for which
+      // FullMergeV2 should be called.
+      EXPECT_EQ("c", merge_in.key.ToString());
+
+      std::string temp_value;
+      if (merge_in.existing_value != nullptr) {
+        temp_value = merge_in.existing_value->ToString();
+      }
+
+      for (auto& operand : merge_in.operand_list) {
+        temp_value.append(operand.ToString());
+      }
+      merge_out->new_value = temp_value;
+
+      return true;
+    }
+
+    bool PartialMergeMulti(const Slice& key,
+                           const std::deque<Slice>& operand_list,
+                           std::string* new_value,
+                           Logger* logger) const override {
+      std::string string_key = key.ToString();
+      EXPECT_TRUE(string_key == "a" || string_key == "b");
+
+      if (string_key == "a") {
+        EXPECT_EQ(1, operand_list.size());
+      } else if (string_key == "b") {
+        EXPECT_EQ(2, operand_list.size());
+      }
+
+      std::string temp_value;
+      for (auto& operand : operand_list) {
+        temp_value.append(operand.ToString());
+      }
+      swap(temp_value, *new_value);
+
+      return true;
+    }
+
+    const char* Name() const override {
+      return "CompactionIteratorTest SingleMergeOp";
+    }
+
+    bool AllowSingleOperand() const override { return true; }
+  };
+
+  SingleMergeOp merge_op;
+  Filter filter;
+  InitIterators(
+      // a should invoke PartialMergeMulti with a single merge operand.
+      {test::KeyStr("a", 50, kTypeMerge),
+       // b should invoke PartialMergeMulti with two operands.
+       test::KeyStr("b", 70, kTypeMerge), test::KeyStr("b", 60, kTypeMerge),
+       // c should invoke FullMerge due to kTypeValue at the beginning.
+       test::KeyStr("c", 90, kTypeMerge), test::KeyStr("c", 80, kTypeValue)},
+      {"av1", "bv2", "bv1", "cv2", "cv1"}, {}, {}, kMaxSequenceNumber,
+      &merge_op, &filter);
+
+  c_iter_->SeekToFirst();
+  ASSERT_TRUE(c_iter_->Valid());
+  ASSERT_EQ(test::KeyStr("a", 50, kTypeMerge), c_iter_->key().ToString());
+  ASSERT_EQ("av1", c_iter_->value().ToString());
+  c_iter_->Next();
+  ASSERT_TRUE(c_iter_->Valid());
+  ASSERT_EQ("bv1bv2", c_iter_->value().ToString());
+  c_iter_->Next();
+  ASSERT_EQ("cv1cv2", c_iter_->value().ToString());
+}
+
 }  // namespace rocksdb
 
 int main(int argc, char** argv) {

db/compaction_job.cc

@@ -38,7 +38,6 @@
 #include "monitoring/iostats_context_imp.h"
 #include "monitoring/perf_context_imp.h"
 #include "monitoring/thread_status_util.h"
-#include "port/likely.h"
 #include "port/port.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
@@ -264,15 +263,16 @@ void CompactionJob::AggregateStatistics() {
 
 CompactionJob::CompactionJob(
     int job_id, Compaction* compaction, const ImmutableDBOptions& db_options,
-    const EnvOptions& env_options, VersionSet* versions,
-    const std::atomic<bool>* shutting_down, LogBuffer* log_buffer,
+    const EnvOptions env_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,
     InstrumentedMutex* db_mutex, Status* db_bg_error,
     std::vector<SequenceNumber> existing_snapshots,
     SequenceNumber earliest_write_conflict_snapshot,
-    std::shared_ptr<Cache> table_cache, EventLogger* event_logger,
-    bool paranoid_file_checks, bool measure_io_stats, const std::string& dbname,
-    CompactionJobStats* compaction_job_stats)
+    const SnapshotChecker* snapshot_checker, std::shared_ptr<Cache> table_cache,
+    EventLogger* event_logger, bool paranoid_file_checks, bool measure_io_stats,
+    const std::string& dbname, CompactionJobStats* compaction_job_stats)
     : job_id_(job_id),
       compact_(new CompactionState(compaction)),
       compaction_job_stats_(compaction_job_stats),
@@ -281,8 +281,11 @@ CompactionJob::CompactionJob(
       db_options_(db_options),
       env_options_(env_options),
       env_(db_options.env),
+      env_optiosn_for_read_(
+          env_->OptimizeForCompactionTableRead(env_options, db_options_)),
       versions_(versions),
       shutting_down_(shutting_down),
+      preserve_deletes_seqnum_(preserve_deletes_seqnum),
       log_buffer_(log_buffer),
       db_directory_(db_directory),
       output_directory_(output_directory),
@@ -291,10 +294,13 @@ CompactionJob::CompactionJob(
       db_bg_error_(db_bg_error),
       existing_snapshots_(std::move(existing_snapshots)),
       earliest_write_conflict_snapshot_(earliest_write_conflict_snapshot),
+      snapshot_checker_(snapshot_checker),
       table_cache_(std::move(table_cache)),
       event_logger_(event_logger),
+      bottommost_level_(false),
       paranoid_file_checks_(paranoid_file_checks),
-      measure_io_stats_(measure_io_stats) {
+      measure_io_stats_(measure_io_stats),
+      write_hint_(Env::WLTH_NOT_SET) {
   assert(log_buffer_ != nullptr);
   const auto* cfd = compact_->compaction->column_family_data();
   ThreadStatusUtil::SetColumnFamily(cfd, cfd->ioptions()->env,
@@ -365,6 +371,8 @@ void CompactionJob::Prepare() {
   assert(c->column_family_data()->current()->storage_info()
       ->NumLevelFiles(compact_->compaction->level()) > 0);
 
+  write_hint_ = c->column_family_data()->CalculateSSTWriteHint(
+    c->output_level());
   // Is this compaction producing files at the bottommost level?
   bottommost_level_ = c->bottommost_level();
 
@@ -594,6 +602,9 @@ Status CompactionJob::Install(const MutableCFOptions& mutable_cf_options) {
 
   double read_write_amp = 0.0;
   double write_amp = 0.0;
+  double bytes_read_per_sec = 0;
+  double bytes_written_per_sec = 0;
+
   if (stats.bytes_read_non_output_levels > 0) {
     read_write_amp = (stats.bytes_written + stats.bytes_read_output_level +
                       stats.bytes_read_non_output_levels) /
@@ -601,37 +612,46 @@ Status CompactionJob::Install(const MutableCFOptions& mutable_cf_options) {
     write_amp = stats.bytes_written /
                 static_cast<double>(stats.bytes_read_non_output_levels);
   }
+  if (stats.micros > 0) {
+    bytes_read_per_sec =
+        (stats.bytes_read_non_output_levels + stats.bytes_read_output_level) /
+        static_cast<double>(stats.micros);
+    bytes_written_per_sec =
+        stats.bytes_written / static_cast<double>(stats.micros);
+  }
+
   ROCKS_LOG_BUFFER(
       log_buffer_,
       "[%s] compacted to: %s, MB/sec: %.1f rd, %.1f wr, level %d, "
       "files in(%d, %d) out(%d) "
       "MB in(%.1f, %.1f) out(%.1f), read-write-amplify(%.1f) "
-      "write-amplify(%.1f) %s, records in: %d, records dropped: %d\n",
-      cfd->GetName().c_str(), vstorage->LevelSummary(&tmp),
-      (stats.bytes_read_non_output_levels + stats.bytes_read_output_level) /
-          static_cast<double>(stats.micros),
-      stats.bytes_written / static_cast<double>(stats.micros),
-      compact_->compaction->output_level(),
+      "write-amplify(%.1f) %s, records in: %" PRIu64
+      ", records dropped: %" PRIu64 " output_compression: %s\n",
+      cfd->GetName().c_str(), vstorage->LevelSummary(&tmp), bytes_read_per_sec,
+      bytes_written_per_sec, compact_->compaction->output_level(),
       stats.num_input_files_in_non_output_levels,
       stats.num_input_files_in_output_level, stats.num_output_files,
       stats.bytes_read_non_output_levels / 1048576.0,
       stats.bytes_read_output_level / 1048576.0,
       stats.bytes_written / 1048576.0, read_write_amp, write_amp,
       status.ToString().c_str(), stats.num_input_records,
-      stats.num_dropped_records);
+      stats.num_dropped_records,
+      CompressionTypeToString(compact_->compaction->output_compression())
+          .c_str());
 
   UpdateCompactionJobStats(stats);
 
   auto stream = event_logger_->LogToBuffer(log_buffer_);
-  stream << "job" << job_id_
-         << "event" << "compaction_finished"
+  stream << "job" << job_id_ << "event"
+         << "compaction_finished"
          << "compaction_time_micros" << compaction_stats_.micros
          << "output_level" << compact_->compaction->output_level()
          << "num_output_files" << compact_->NumOutputFiles()
-         << "total_output_size" << compact_->total_bytes
-         << "num_input_records" << compact_->num_input_records
-         << "num_output_records" << compact_->num_output_records
-         << "num_subcompactions" << compact_->sub_compact_states.size();
+         << "total_output_size" << compact_->total_bytes << "num_input_records"
+         << compact_->num_input_records << "num_output_records"
+         << compact_->num_output_records << "num_subcompactions"
+         << compact_->sub_compact_states.size() << "output_compression"
+         << CompressionTypeToString(compact_->compaction->output_compression());
 
   if (compaction_job_stats_ != nullptr) {
     stream << "num_single_delete_mismatches"
@@ -666,7 +686,7 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
   std::unique_ptr<RangeDelAggregator> range_del_agg(
       new RangeDelAggregator(cfd->internal_comparator(), existing_snapshots_));
   std::unique_ptr<InternalIterator> input(versions_->MakeInputIterator(
-      sub_compact->compaction, range_del_agg.get()));
+      sub_compact->compaction, range_del_agg.get(), env_optiosn_for_read_));
 
   AutoThreadOperationStageUpdater stage_updater(
       ThreadStatus::STAGE_COMPACTION_PROCESS_KV);
@@ -691,15 +711,18 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
       sub_compact->compaction->mutable_cf_options();
 
   // To build compression dictionary, we sample the first output file, assuming
-  // it'll reach the maximum length, and then use the dictionary for compressing
-  // subsequent output files. The dictionary may be less than max_dict_bytes if
-  // the first output file's length is less than the maximum.
+  // it'll reach the maximum length. We optionally pass these samples through
+  // zstd's dictionary trainer, or just use them directly. Then, the dictionary
+  // is used for compressing subsequent output files in the same subcompaction.
+  const bool kUseZstdTrainer =
+      cfd->ioptions()->compression_opts.zstd_max_train_bytes > 0;
+  const size_t kSampleBytes =
+      kUseZstdTrainer ? cfd->ioptions()->compression_opts.zstd_max_train_bytes
+                      : cfd->ioptions()->compression_opts.max_dict_bytes;
   const int kSampleLenShift = 6;  // 2^6 = 64-byte samples
   std::set<size_t> sample_begin_offsets;
-  if (bottommost_level_ &&
-      cfd->ioptions()->compression_opts.max_dict_bytes > 0) {
-    const size_t kMaxSamples =
-        cfd->ioptions()->compression_opts.max_dict_bytes >> kSampleLenShift;
+  if (bottommost_level_ && kSampleBytes > 0) {
+    const size_t kMaxSamples = kSampleBytes >> kSampleLenShift;
     const size_t kOutFileLen = mutable_cf_options->MaxFileSizeForLevel(
         compact_->compaction->output_level());
     if (kOutFileLen != port::kMaxSizet) {
@@ -753,9 +776,10 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
   Status status;
   sub_compact->c_iter.reset(new CompactionIterator(
       input.get(), cfd->user_comparator(), &merge, versions_->LastSequence(),
-      &existing_snapshots_, earliest_write_conflict_snapshot_, env_, false,
-      range_del_agg.get(), sub_compact->compaction, compaction_filter,
-      comp_event_listener, shutting_down_));
+      &existing_snapshots_, earliest_write_conflict_snapshot_,
+      snapshot_checker_, env_, false, range_del_agg.get(),
+      sub_compact->compaction, compaction_filter, comp_event_listener,
+      shutting_down_, preserve_deletes_seqnum_));
   auto c_iter = sub_compact->c_iter.get();
   c_iter->SeekToFirst();
   if (c_iter->Valid() &&
@@ -768,11 +792,11 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
   }
   const auto& c_iter_stats = c_iter->iter_stats();
   auto sample_begin_offset_iter = sample_begin_offsets.cbegin();
-  // data_begin_offset and compression_dict are only valid while generating
+  // data_begin_offset and dict_sample_data are only valid while generating
   // dictionary from the first output file.
   size_t data_begin_offset = 0;
-  std::string compression_dict;
-  compression_dict.reserve(cfd->ioptions()->compression_opts.max_dict_bytes);
+  std::string dict_sample_data;
+  dict_sample_data.reserve(kSampleBytes);
 
   while (status.ok() && !cfd->IsDropped() && c_iter->Valid()) {
     // Invariant: c_iter.status() is guaranteed to be OK if c_iter->Valid()
@@ -844,7 +868,7 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
             data_elmt_copy_len =
                 data_end_offset - (data_begin_offset + data_elmt_copy_offset);
           }
-          compression_dict.append(&data_elmt.data()[data_elmt_copy_offset],
+          dict_sample_data.append(&data_elmt.data()[data_elmt_copy_offset],
                                   data_elmt_copy_len);
           if (sample_end_offset > data_end_offset) {
             // Didn't finish sample. Try to finish it with the next data_elmt.
@@ -899,9 +923,15 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
       RecordDroppedKeys(range_del_out_stats,
                         &sub_compact->compaction_job_stats);
       if (sub_compact->outputs.size() == 1) {
-        // Use dictionary from first output file for compression of subsequent
-        // files.
-        sub_compact->compression_dict = std::move(compression_dict);
+        // Use samples from first output file to create dictionary for
+        // compression of subsequent files.
+        if (kUseZstdTrainer) {
+          sub_compact->compression_dict = ZSTD_TrainDictionary(
+              dict_sample_data, kSampleLenShift,
+              cfd->ioptions()->compression_opts.max_dict_bytes);
+        } else {
+          sub_compact->compression_dict = std::move(dict_sample_data);
+        }
       }
     }
   }
@@ -1006,6 +1036,10 @@ void CompactionJob::RecordDroppedKeys(
     RecordTick(stats_, COMPACTION_RANGE_DEL_DROP_OBSOLETE,
                c_iter_stats.num_range_del_drop_obsolete);
   }
+  if (c_iter_stats.num_optimized_del_drop_obsolete > 0) {
+    RecordTick(stats_, COMPACTION_OPTIMIZED_DEL_DROP_OBSOLETE,
+               c_iter_stats.num_optimized_del_drop_obsolete);
+  }
 }
 
 Status CompactionJob::FinishCompactionOutputFile(
@@ -1028,6 +1062,7 @@ Status CompactionJob::FinishCompactionOutputFile(
   auto meta = &sub_compact->current_output()->meta;
   if (s.ok()) {
     Slice lower_bound_guard, upper_bound_guard;
+    std::string smallest_user_key;
     const Slice *lower_bound, *upper_bound;
     if (sub_compact->outputs.size() == 1) {
       // For the first output table, include range tombstones before the min key
@@ -1037,7 +1072,8 @@ Status CompactionJob::FinishCompactionOutputFile(
       // For subsequent output tables, only include range tombstones from min
       // key onwards since the previous file was extended to contain range
       // tombstones falling before min key.
-      lower_bound_guard = meta->smallest.user_key();
+      smallest_user_key = meta->smallest.user_key().ToString(false /*hex*/);
+      lower_bound_guard = Slice(smallest_user_key);
       lower_bound = &lower_bound_guard;
     } else {
       lower_bound = nullptr;
@@ -1055,16 +1091,18 @@ Status CompactionJob::FinishCompactionOutputFile(
     range_del_agg->AddToBuilder(sub_compact->builder.get(), lower_bound,
                                 upper_bound, meta, range_del_out_stats,
                                 bottommost_level_);
+    meta->marked_for_compaction = sub_compact->builder->NeedCompact();
   }
   const uint64_t current_entries = sub_compact->builder->NumEntries();
-  meta->marked_for_compaction = sub_compact->builder->NeedCompact();
   if (s.ok()) {
     s = sub_compact->builder->Finish();
   } else {
     sub_compact->builder->Abandon();
   }
   const uint64_t current_bytes = sub_compact->builder->FileSize();
-  meta->fd.file_size = current_bytes;
+  if (s.ok()) {
+    meta->fd.file_size = current_bytes;
+  }
   sub_compact->current_output()->finished = true;
   sub_compact->total_bytes += current_bytes;
 
@@ -1109,7 +1147,8 @@ Status CompactionJob::FinishCompactionOutputFile(
         nullptr /* range_del_agg */, nullptr,
         cfd->internal_stats()->GetFileReadHist(
             compact_->compaction->output_level()),
-        false);
+        false, nullptr /* arena */, false /* skip_filters */,
+        compact_->compaction->output_level());
     s = iter->status();
 
     if (s.ok() && paranoid_file_checks_) {
@@ -1132,17 +1171,24 @@ Status CompactionJob::FinishCompactionOutputFile(
                      meta->marked_for_compaction ? " (need compaction)" : "");
     }
   }
-  std::string fname = TableFileName(db_options_.db_paths, meta->fd.GetNumber(),
-                                    meta->fd.GetPathId());
+  std::string fname;
+  FileDescriptor output_fd;
+  if (meta != nullptr) {
+    fname = TableFileName(db_options_.db_paths, meta->fd.GetNumber(),
+                          meta->fd.GetPathId());
+    output_fd = meta->fd;
+  } else {
+    fname = "(nil)";
+  }
   EventHelpers::LogAndNotifyTableFileCreationFinished(
       event_logger_, cfd->ioptions()->listeners, dbname_, cfd->GetName(), fname,
-      job_id_, meta->fd, tp, TableFileCreationReason::kCompaction, s);
+      job_id_, output_fd, tp, TableFileCreationReason::kCompaction, s);
 
 #ifndef ROCKSDB_LITE
   // Report new file to SstFileManagerImpl
   auto sfm =
       static_cast<SstFileManagerImpl*>(db_options_.sst_file_manager.get());
-  if (sfm && meta->fd.GetPathId() == 0) {
+  if (sfm && meta != nullptr && meta->fd.GetPathId() == 0) {
     auto fn = TableFileName(cfd->ioptions()->db_paths, meta->fd.GetNumber(),
                             meta->fd.GetPathId());
     sfm->OnAddFile(fn);
@@ -1240,11 +1286,12 @@ Status CompactionJob::OpenCompactionOutputFile(
 #endif  // !ROCKSDB_LITE
   // Make the output file
   unique_ptr<WritableFile> writable_file;
-  EnvOptions opt_env_opts =
-      env_->OptimizeForCompactionTableWrite(env_options_, db_options_);
+#ifndef NDEBUG
+  bool syncpoint_arg = env_options_.use_direct_writes;
   TEST_SYNC_POINT_CALLBACK("CompactionJob::OpenCompactionOutputFile",
-                           &opt_env_opts.use_direct_writes);
-  Status s = NewWritableFile(env_, fname, &writable_file, opt_env_opts);
+                           &syncpoint_arg);
+#endif
+  Status s = NewWritableFile(env_, fname, &writable_file, env_options_);
   if (!s.ok()) {
     ROCKS_LOG_ERROR(
         db_options_.info_log,
@@ -1267,6 +1314,7 @@ Status CompactionJob::OpenCompactionOutputFile(
 
   sub_compact->outputs.push_back(out);
   writable_file->SetIOPriority(Env::IO_LOW);
+  writable_file->SetWriteLifeTimeHint(write_hint_);
   writable_file->SetPreallocationBlockSize(static_cast<size_t>(
       sub_compact->compaction->OutputFilePreallocationSize()));
   sub_compact->outfile.reset(new WritableFileWriter(
@@ -1282,7 +1330,15 @@ Status CompactionJob::OpenCompactionOutputFile(
       sub_compact->compaction->MaxInputFileCreationTime();
   if (output_file_creation_time == 0) {
     int64_t _current_time = 0;
-    db_options_.env->GetCurrentTime(&_current_time);  // ignore error
+    auto status = db_options_.env->GetCurrentTime(&_current_time);
+    // Safe to proceed even if GetCurrentTime fails. So, log and proceed.
+    if (!status.ok()) {
+      ROCKS_LOG_WARN(
+          db_options_.info_log,
+          "Failed to get current time to populate creation_time property. "
+          "Status: %s",
+          status.ToString().c_str());
+    }
     output_file_creation_time = static_cast<uint64_t>(_current_time);
   }
 

db/compaction_job.h

@@ -45,24 +45,28 @@
 
 namespace rocksdb {
 
+class Arena;
 class MemTable;
+class SnapshotChecker;
 class TableCache;
 class Version;
 class VersionEdit;
 class VersionSet;
-class Arena;
 
 class CompactionJob {
  public:
   CompactionJob(int job_id, Compaction* compaction,
                 const ImmutableDBOptions& db_options,
-                const EnvOptions& env_options, VersionSet* versions,
-                const std::atomic<bool>* shutting_down, LogBuffer* log_buffer,
+                const EnvOptions env_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, InstrumentedMutex* db_mutex,
                 Status* db_bg_error,
                 std::vector<SequenceNumber> existing_snapshots,
                 SequenceNumber earliest_write_conflict_snapshot,
+                const SnapshotChecker* snapshot_checker,
                 std::shared_ptr<Cache> table_cache, EventLogger* event_logger,
                 bool paranoid_file_checks, bool measure_io_stats,
                 const std::string& dbname,
@@ -127,11 +131,14 @@ class CompactionJob {
   // DBImpl state
   const std::string& dbname_;
   const ImmutableDBOptions& db_options_;
-  const EnvOptions& env_options_;
+  const EnvOptions env_options_;
 
   Env* env_;
+  // env_option optimized for compaction table reads
+  EnvOptions env_optiosn_for_read_;
   VersionSet* versions_;
   const std::atomic<bool>* shutting_down_;
+  const SequenceNumber preserve_deletes_seqnum_;
   LogBuffer* log_buffer_;
   Directory* db_directory_;
   Directory* output_directory_;
@@ -149,6 +156,8 @@ class CompactionJob {
   // should make sure not to remove evidence that a write occurred.
   SequenceNumber earliest_write_conflict_snapshot_;
 
+  const SnapshotChecker* const snapshot_checker_;
+
   std::shared_ptr<Cache> table_cache_;
 
   EventLogger* event_logger_;
@@ -160,6 +169,7 @@ class CompactionJob {
   std::vector<Slice> boundaries_;
   // Stores the approx size of keys covered in the range of each subcompaction
   std::vector<uint64_t> sizes_;
+  Env::WriteLifeTimeHint write_hint_;
 };
 
 }  // namespace rocksdb

db/compaction_job_test.cc

@@ -76,6 +76,7 @@ class CompactionJobTest : public testing::Test {
                                  table_cache_.get(), &write_buffer_manager_,
                                  &write_controller_)),
         shutting_down_(false),
+        preserve_deletes_seqnum_(0),
         mock_table_factory_(new mock::MockTableFactory()) {
     EXPECT_OK(env_->CreateDirIfMissing(dbname_));
     db_options_.db_paths.emplace_back(dbname_,
@@ -142,7 +143,8 @@ class CompactionJobTest : public testing::Test {
   }
 
   void SetLastSequence(const SequenceNumber sequence_number) {
-    versions_->SetLastToBeWrittenSequence(sequence_number + 1);
+    versions_->SetLastAllocatedSequence(sequence_number + 1);
+    versions_->SetLastPublishedSequence(sequence_number + 1);
     versions_->SetLastSequence(sequence_number + 1);
   }
 
@@ -250,12 +252,15 @@ class CompactionJobTest : public testing::Test {
     LogBuffer log_buffer(InfoLogLevel::INFO_LEVEL, db_options_.info_log.get());
     mutex_.Lock();
     EventLogger event_logger(db_options_.info_log.get());
-    CompactionJob compaction_job(
-        0, &compaction, db_options_, env_options_, versions_.get(),
-        &shutting_down_, &log_buffer, nullptr, nullptr, nullptr, &mutex_,
-        &bg_error_, snapshots, earliest_write_conflict_snapshot, table_cache_,
-        &event_logger, false, false, dbname_, &compaction_job_stats_);
-
+    // TODO(yiwu) add a mock snapshot checker and add test for it.
+    SnapshotChecker* snapshot_checker = nullptr;
+    CompactionJob compaction_job(0, &compaction, db_options_, env_options_,
+                                 versions_.get(), &shutting_down_,
+                                 preserve_deletes_seqnum_, &log_buffer,
+                                 nullptr, nullptr, nullptr, &mutex_, &bg_error_,
+                                 snapshots, earliest_write_conflict_snapshot,
+                                 snapshot_checker, table_cache_, &event_logger,
+                                 false, false, dbname_, &compaction_job_stats_);
     VerifyInitializationOfCompactionJobStats(compaction_job_stats_);
 
     compaction_job.Prepare();
@@ -291,6 +296,7 @@ class CompactionJobTest : public testing::Test {
   std::unique_ptr<VersionSet> versions_;
   InstrumentedMutex mutex_;
   std::atomic<bool> shutting_down_;
+  SequenceNumber preserve_deletes_seqnum_;
   std::shared_ptr<mock::MockTableFactory> mock_table_factory_;
   CompactionJobStats compaction_job_stats_;
   ColumnFamilyData* cfd_;

db/compaction_picker.cc

@@ -89,7 +89,7 @@ CompressionType GetCompressionType(const ImmutableCFOptions& ioptions,
   // If bottommost_compression is set and we are compacting to the
   // bottommost level then we should use it.
   if (ioptions.bottommost_compression != kDisableCompressionOption &&
-      level > base_level && level >= (vstorage->num_non_empty_levels() - 1)) {
+      level >= (vstorage->num_non_empty_levels() - 1)) {
     return ioptions.bottommost_compression;
   }
   // If the user has specified a different compression level for each level,
@@ -234,11 +234,6 @@ bool CompactionPicker::ExpandInputsToCleanCut(const std::string& cf_name,
   // If, after the expansion, there are files that are already under
   // compaction, then we must drop/cancel this compaction.
   if (AreFilesInCompaction(inputs->files)) {
-    ROCKS_LOG_WARN(
-        ioptions_.info_log,
-        "[%s] ExpandWhileOverlapping() failure because some of the necessary"
-        " compaction input files are currently being compacted.",
-        cf_name.c_str());
     return false;
   }
   return true;
@@ -297,25 +292,16 @@ Compaction* CompactionPicker::CompactFiles(
     VersionStorageInfo* vstorage, const MutableCFOptions& mutable_cf_options,
     uint32_t output_path_id) {
   assert(input_files.size());
+  // This compaction output should not overlap with a running compaction as
+  // `SanitizeCompactionInputFiles` should've checked earlier and db mutex
+  // shouldn't have been released since.
+  assert(!FilesRangeOverlapWithCompaction(input_files, output_level));
 
-  // TODO(rven ): we might be able to run concurrent level 0 compaction
-  // if the key ranges of the two compactions do not overlap, but for now
-  // we do not allow it.
-  if ((input_files[0].level == 0) && !level0_compactions_in_progress_.empty()) {
-    return nullptr;
-  }
-  // This compaction output could overlap with a running compaction
-  if (FilesRangeOverlapWithCompaction(input_files, output_level)) {
-    return nullptr;
-  }
   auto c =
       new Compaction(vstorage, ioptions_, mutable_cf_options, input_files,
                      output_level, compact_options.output_file_size_limit,
                      mutable_cf_options.max_compaction_bytes, output_path_id,
                      compact_options.compression, /* grandparents */ {}, true);
-
-  // If it's level 0 compaction, make sure we don't execute any other level 0
-  // compactions in parallel
   RegisterCompaction(c);
   return c;
 }
@@ -401,7 +387,10 @@ bool CompactionPicker::SetupOtherInputs(
   assert(output_level_inputs->empty());
   const int input_level = inputs->level;
   const int output_level = output_level_inputs->level;
-  assert(input_level != output_level);
+  if (input_level == output_level) {
+    // no possibility of conflict
+    return true;
+  }
 
   // For now, we only support merging two levels, start level and output level.
   // We need to assert other levels are empty.
@@ -735,10 +724,6 @@ Status CompactionPicker::SanitizeCompactionInputFilesForAllLevels(
   auto& levels = cf_meta.levels;
   auto comparator = icmp_->user_comparator();
 
-  // TODO(yhchiang): If there is any input files of L1 or up and there
-  // is at least one L0 files. All L0 files older than the L0 file needs
-  // to be included. Otherwise, it is a false conditoin
-
   // TODO(yhchiang): add is_adjustable to CompactionOptions
 
   // the smallest and largest key of the current compaction input
@@ -799,6 +784,8 @@ Status CompactionPicker::SanitizeCompactionInputFilesForAllLevels(
         }
         last_included++;
       }
+    } else if (output_level > 0) {
+      last_included = static_cast<int>(current_files.size() - 1);
     }
 
     // include all files between the first and the last compaction input files.
@@ -954,6 +941,9 @@ void CompactionPicker::UnregisterCompaction(Compaction* c) {
 
 bool LevelCompactionPicker::NeedsCompaction(
     const VersionStorageInfo* vstorage) const {
+  if (!vstorage->BottommostFilesMarkedForCompaction().empty()) {
+    return true;
+  }
   if (!vstorage->FilesMarkedForCompaction().empty()) {
     return true;
   }
@@ -1106,11 +1096,7 @@ void LevelCompactionBuilder::SetupInitialFiles() {
       }
       output_level_ =
           (start_level_ == 0) ? vstorage_->base_level() : start_level_ + 1;
-      if (PickFileToCompact() &&
-          compaction_picker_->ExpandInputsToCleanCut(cf_name_, vstorage_,
-                                                     &start_level_inputs_) &&
-          !compaction_picker_->FilesRangeOverlapWithCompaction(
-              {start_level_inputs_}, output_level_)) {
+      if (PickFileToCompact()) {
         // found the compaction!
         if (start_level_ == 0) {
           // L0 score = `num L0 files` / `level0_file_num_compaction_trigger`
@@ -1148,7 +1134,28 @@ void LevelCompactionBuilder::SetupInitialFiles() {
     is_manual_ = true;
     parent_index_ = base_index_ = -1;
     PickFilesMarkedForCompaction();
-    if (!start_level_inputs_.empty()) {
+    if (start_level_inputs_.empty()) {
+      size_t i;
+      for (i = 0; i < vstorage_->BottommostFilesMarkedForCompaction().size();
+           ++i) {
+        auto& level_and_file =
+            vstorage_->BottommostFilesMarkedForCompaction()[i];
+        assert(!level_and_file.second->being_compacted);
+        start_level_inputs_.level = output_level_ = start_level_ =
+            level_and_file.first;
+        start_level_inputs_.files = {level_and_file.second};
+        if (compaction_picker_->ExpandInputsToCleanCut(cf_name_, vstorage_,
+                                                       &start_level_inputs_)) {
+          break;
+        }
+      }
+      if (i == vstorage_->BottommostFilesMarkedForCompaction().size()) {
+        start_level_inputs_.clear();
+      } else {
+        assert(!start_level_inputs_.empty());
+        compaction_reason_ = CompactionReason::kBottommostFiles;
+      }
+    } else {
       compaction_reason_ = CompactionReason::kFilesMarkedForCompaction;
     }
   }
@@ -1288,6 +1295,8 @@ uint32_t LevelCompactionBuilder::GetPathId(
   uint64_t level_size;
   int cur_level = 0;
 
+  // max_bytes_for_level_base denotes L1 size.
+  // We estimate L0 size to be the same as L1.
   level_size = mutable_cf_options.max_bytes_for_level_base;
 
   // Last path is the fallback
@@ -1298,8 +1307,21 @@ uint32_t LevelCompactionBuilder::GetPathId(
         return p;
       } else {
         current_path_size -= level_size;
-        level_size = static_cast<uint64_t>(
-            level_size * mutable_cf_options.max_bytes_for_level_multiplier);
+        if (cur_level > 0) {
+          if (ioptions.level_compaction_dynamic_level_bytes) {
+            // Currently, level_compaction_dynamic_level_bytes is ignored when
+            // multiple db paths are specified. https://github.com/facebook/
+            // rocksdb/blob/master/db/column_family.cc.
+            // Still, adding this check to avoid accidentally using
+            // max_bytes_for_level_multiplier_additional
+            level_size = static_cast<uint64_t>(
+                level_size * mutable_cf_options.max_bytes_for_level_multiplier);
+          } else {
+            level_size = static_cast<uint64_t>(
+                level_size * mutable_cf_options.max_bytes_for_level_multiplier *
+                mutable_cf_options.MaxBytesMultiplerAdditional(cur_level));
+          }
+        }
         cur_level++;
         continue;
       }
@@ -1332,12 +1354,10 @@ bool LevelCompactionBuilder::PickFileToCompact() {
   const std::vector<FileMetaData*>& level_files =
       vstorage_->LevelFiles(start_level_);
 
-  // record the first file that is not yet compacted
-  int nextIndex = -1;
-
-  for (unsigned int i = vstorage_->NextCompactionIndex(start_level_);
-       i < file_size.size(); i++) {
-    int index = file_size[i];
+  unsigned int cmp_idx;
+  for (cmp_idx = vstorage_->NextCompactionIndex(start_level_);
+       cmp_idx < file_size.size(); cmp_idx++) {
+    int index = file_size[cmp_idx];
     auto* f = level_files[index];
 
     // do not pick a file to compact if it is being compacted
@@ -1346,27 +1366,42 @@ bool LevelCompactionBuilder::PickFileToCompact() {
       continue;
     }
 
-    // remember the startIndex for the next call to PickCompaction
-    if (nextIndex == -1) {
-      nextIndex = i;
-    }
-
-    // Do not pick this file if its parents at level+1 are being compacted.
-    // Maybe we can avoid redoing this work in SetupOtherInputs
-    parent_index_ = -1;
-    if (compaction_picker_->IsRangeInCompaction(vstorage_, &f->smallest,
-                                                &f->largest, output_level_,
-                                                &parent_index_)) {
-      continue;
-    }
     start_level_inputs_.files.push_back(f);
     start_level_inputs_.level = start_level_;
+    if (!compaction_picker_->ExpandInputsToCleanCut(cf_name_, vstorage_,
+                                                    &start_level_inputs_) ||
+        compaction_picker_->FilesRangeOverlapWithCompaction(
+            {start_level_inputs_}, output_level_)) {
+      // A locked (pending compaction) input-level file was pulled in due to
+      // user-key overlap.
+      start_level_inputs_.clear();
+      continue;
+    }
+
+    // Now that input level is fully expanded, we check whether any output files
+    // are locked due to pending compaction.
+    //
+    // Note we rely on ExpandInputsToCleanCut() to tell us whether any output-
+    // level files are locked, not just the extra ones pulled in for user-key
+    // overlap.
+    InternalKey smallest, largest;
+    compaction_picker_->GetRange(start_level_inputs_, &smallest, &largest);
+    CompactionInputFiles output_level_inputs;
+    output_level_inputs.level = output_level_;
+    vstorage_->GetOverlappingInputs(output_level_, &smallest, &largest,
+                                    &output_level_inputs.files);
+    if (!output_level_inputs.empty() &&
+        !compaction_picker_->ExpandInputsToCleanCut(cf_name_, vstorage_,
+                                                    &output_level_inputs)) {
+      start_level_inputs_.clear();
+      continue;
+    }
     base_index_ = index;
     break;
   }
 
   // store where to start the iteration in the next call to PickCompaction
-  vstorage_->SetNextCompactionIndex(start_level_, nextIndex);
+  vstorage_->SetNextCompactionIndex(start_level_, cmp_idx);
 
   return start_level_inputs_.size() > 0;
 }
@@ -1417,7 +1452,7 @@ uint64_t GetTotalFilesSize(
 Compaction* FIFOCompactionPicker::PickTTLCompaction(
     const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
     VersionStorageInfo* vstorage, LogBuffer* log_buffer) {
-  assert(ioptions_.compaction_options_fifo.ttl > 0);
+  assert(mutable_cf_options.compaction_options_fifo.ttl > 0);
 
   const int kLevel0 = 0;
   const std::vector<FileMetaData*>& level_files = vstorage->LevelFiles(kLevel0);
@@ -1438,19 +1473,22 @@ Compaction* FIFOCompactionPicker::PickTTLCompaction(
   inputs.emplace_back();
   inputs[0].level = 0;
 
-  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 - ioptions_.compaction_options_fifo.ttl)) {
-        break;
+  // avoid underflow
+  if (current_time > mutable_cf_options.compaction_options_fifo.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.compaction_options_fifo.ttl)) {
+          break;
+        }
+        total_size -= f->compensated_file_size;
+        inputs[0].files.push_back(f);
       }
-      total_size -= f->compensated_file_size;
-      inputs[0].files.push_back(f);
     }
   }
 
@@ -1459,7 +1497,8 @@ Compaction* FIFOCompactionPicker::PickTTLCompaction(
   // 2. there are a few files older than ttl, but deleting them will not bring
   //    the total size to be less than max_table_files_size threshold.
   if (inputs[0].files.empty() ||
-      total_size > ioptions_.compaction_options_fifo.max_table_files_size) {
+      total_size >
+          mutable_cf_options.compaction_options_fifo.max_table_files_size) {
     return nullptr;
   }
 
@@ -1485,10 +1524,11 @@ Compaction* FIFOCompactionPicker::PickSizeCompaction(
   const std::vector<FileMetaData*>& level_files = vstorage->LevelFiles(kLevel0);
   uint64_t total_size = GetTotalFilesSize(level_files);
 
-  if (total_size <= ioptions_.compaction_options_fifo.max_table_files_size ||
+  if (total_size <=
+          mutable_cf_options.compaction_options_fifo.max_table_files_size ||
       level_files.size() == 0) {
     // total size not exceeded
-    if (ioptions_.compaction_options_fifo.allow_compaction &&
+    if (mutable_cf_options.compaction_options_fifo.allow_compaction &&
         level_files.size() > 0) {
       CompactionInputFiles comp_inputs;
       if (FindIntraL0Compaction(
@@ -1508,11 +1548,12 @@ Compaction* FIFOCompactionPicker::PickSizeCompaction(
       }
     }
 
-    ROCKS_LOG_BUFFER(log_buffer,
-                     "[%s] FIFO compaction: nothing to do. Total size %" PRIu64
-                     ", max size %" PRIu64 "\n",
-                     cf_name.c_str(), total_size,
-                     ioptions_.compaction_options_fifo.max_table_files_size);
+    ROCKS_LOG_BUFFER(
+        log_buffer,
+        "[%s] FIFO compaction: nothing to do. Total size %" PRIu64
+        ", max size %" PRIu64 "\n",
+        cf_name.c_str(), total_size,
+        mutable_cf_options.compaction_options_fifo.max_table_files_size);
     return nullptr;
   }
 
@@ -1539,7 +1580,8 @@ Compaction* FIFOCompactionPicker::PickSizeCompaction(
                      "[%s] FIFO compaction: picking file %" PRIu64
                      " with size %s for deletion",
                      cf_name.c_str(), f->fd.GetNumber(), tmp_fsize);
-    if (total_size <= ioptions_.compaction_options_fifo.max_table_files_size) {
+    if (total_size <=
+        mutable_cf_options.compaction_options_fifo.max_table_files_size) {
       break;
     }
   }
@@ -1557,7 +1599,7 @@ Compaction* FIFOCompactionPicker::PickCompaction(
   assert(vstorage->num_levels() == 1);
 
   Compaction* c = nullptr;
-  if (ioptions_.compaction_options_fifo.ttl > 0) {
+  if (mutable_cf_options.compaction_options_fifo.ttl > 0) {
     c = PickTTLCompaction(cf_name, mutable_cf_options, vstorage, log_buffer);
   }
   if (c == nullptr) {

db/compaction_picker.h

@@ -88,6 +88,10 @@ class CompactionPicker {
 
   // Takes a list of CompactionInputFiles and returns a (manual) Compaction
   // object.
+  //
+  // Caller must provide a set of input files that has been passed through
+  // `SanitizeCompactionInputFiles` earlier. The lock should not be released
+  // between that call and this one.
   Compaction* CompactFiles(const CompactionOptions& compact_options,
                            const std::vector<CompactionInputFiles>& input_files,
                            int output_level, VersionStorageInfo* vstorage,

db/compaction_picker_test.cc

@@ -175,6 +175,8 @@ TEST_F(CompactionPickerTest, Level1Trigger) {
 }
 
 TEST_F(CompactionPickerTest, Level1Trigger2) {
+  mutable_cf_options_.target_file_size_base = 10000000000;
+  mutable_cf_options_.RefreshDerivedOptions(ioptions_);
   NewVersionStorage(6, kCompactionStyleLevel);
   Add(1, 66U, "150", "200", 1000000001U);
   Add(1, 88U, "201", "300", 1000000000U);
@@ -191,13 +193,14 @@ TEST_F(CompactionPickerTest, Level1Trigger2) {
   ASSERT_EQ(66U, compaction->input(0, 0)->fd.GetNumber());
   ASSERT_EQ(6U, compaction->input(1, 0)->fd.GetNumber());
   ASSERT_EQ(7U, compaction->input(1, 1)->fd.GetNumber());
+  ASSERT_EQ(uint64_t{1073741824}, compaction->OutputFilePreallocationSize());
 }
 
 TEST_F(CompactionPickerTest, LevelMaxScore) {
   NewVersionStorage(6, kCompactionStyleLevel);
   mutable_cf_options_.target_file_size_base = 10000000;
-  mutable_cf_options_.target_file_size_multiplier = 10;
   mutable_cf_options_.max_bytes_for_level_base = 10 * 1024 * 1024;
+  mutable_cf_options_.RefreshDerivedOptions(ioptions_);
   Add(0, 1U, "150", "200", 1000000U);
   // Level 1 score 1.2
   Add(1, 66U, "150", "200", 6000000U);
@@ -218,6 +221,9 @@ TEST_F(CompactionPickerTest, LevelMaxScore) {
   ASSERT_TRUE(compaction.get() != nullptr);
   ASSERT_EQ(1U, compaction->num_input_files(0));
   ASSERT_EQ(7U, compaction->input(0, 0)->fd.GetNumber());
+  ASSERT_EQ(mutable_cf_options_.target_file_size_base +
+                mutable_cf_options_.target_file_size_base / 10,
+            compaction->OutputFilePreallocationSize());
 }
 
 TEST_F(CompactionPickerTest, NeedsCompactionLevel) {
@@ -437,7 +443,7 @@ TEST_F(CompactionPickerTest, CompactionUniversalIngestBehindReservedLevel) {
 TEST_F(CompactionPickerTest, CannotTrivialMoveUniversal) {
   const uint64_t kFileSize = 100000;
 
-  ioptions_.compaction_options_universal.allow_trivial_move = true;
+  mutable_cf_options_.compaction_options_universal.allow_trivial_move = true;
   NewVersionStorage(1, kCompactionStyleUniversal);
   UniversalCompactionPicker universal_compaction_picker(ioptions_, &icmp_);
   // must return false when there's no files.
@@ -468,7 +474,7 @@ TEST_F(CompactionPickerTest, CannotTrivialMoveUniversal) {
 TEST_F(CompactionPickerTest, AllowsTrivialMoveUniversal) {
   const uint64_t kFileSize = 100000;
 
-  ioptions_.compaction_options_universal.allow_trivial_move = true;
+  mutable_cf_options_.compaction_options_universal.allow_trivial_move = true;
   UniversalCompactionPicker universal_compaction_picker(ioptions_, &icmp_);
 
   NewVersionStorage(3, kCompactionStyleUniversal);
@@ -496,7 +502,7 @@ TEST_F(CompactionPickerTest, NeedsCompactionFIFO) {
   const uint64_t kMaxSize = kFileSize * kFileCount / 2;
 
   fifo_options_.max_table_files_size = kMaxSize;
-  ioptions_.compaction_options_fifo = fifo_options_;
+  mutable_cf_options_.compaction_options_fifo = fifo_options_;
   FIFOCompactionPicker fifo_compaction_picker(ioptions_, &icmp_);
   UpdateVersionStorageInfo();
   // must return false when there's no files.
@@ -512,7 +518,7 @@ TEST_F(CompactionPickerTest, NeedsCompactionFIFO) {
         kFileSize, 0, i * 100, i * 100 + 99);
     current_size += kFileSize;
     UpdateVersionStorageInfo();
-    ASSERT_EQ(level_compaction_picker.NeedsCompaction(vstorage_.get()),
+    ASSERT_EQ(fifo_compaction_picker.NeedsCompaction(vstorage_.get()),
               vstorage_->CompactionScore(0) >= 1);
   }
 }
@@ -521,9 +527,10 @@ TEST_F(CompactionPickerTest, NeedsCompactionFIFO) {
 TEST_F(CompactionPickerTest, CompactionPriMinOverlapping1) {
   NewVersionStorage(6, kCompactionStyleLevel);
   ioptions_.compaction_pri = kMinOverlappingRatio;
-  mutable_cf_options_.target_file_size_base = 10000000;
+  mutable_cf_options_.target_file_size_base = 100000000000;
   mutable_cf_options_.target_file_size_multiplier = 10;
   mutable_cf_options_.max_bytes_for_level_base = 10 * 1024 * 1024;
+  mutable_cf_options_.RefreshDerivedOptions(ioptions_);
 
   Add(2, 6U, "150", "179", 50000000U);
   Add(2, 7U, "180", "220", 50000000U);
@@ -543,6 +550,8 @@ TEST_F(CompactionPickerTest, CompactionPriMinOverlapping1) {
   ASSERT_EQ(1U, compaction->num_input_files(0));
   // Pick file 8 because it overlaps with 0 files on level 3.
   ASSERT_EQ(8U, compaction->input(0, 0)->fd.GetNumber());
+  // Compaction input size * 1.1
+  ASSERT_GE(uint64_t{55000000}, compaction->OutputFilePreallocationSize());
 }
 
 TEST_F(CompactionPickerTest, CompactionPriMinOverlapping2) {
@@ -852,6 +861,80 @@ TEST_F(CompactionPickerTest, OverlappingUserKeys9) {
   ASSERT_EQ(8U, compaction->input(1, 1)->fd.GetNumber());
 }
 
+TEST_F(CompactionPickerTest, OverlappingUserKeys10) {
+  // Locked file encountered when pulling in extra input-level files with same
+  // user keys. Verify we pick the next-best file from the same input level.
+  NewVersionStorage(6, kCompactionStyleLevel);
+  mutable_cf_options_.max_compaction_bytes = 100000000000u;
+
+  // file_number 2U is largest and thus first choice. But it overlaps with
+  // file_number 1U which is being compacted. So instead we pick the next-
+  // biggest file, 3U, which is eligible for compaction.
+  Add(1 /* level */, 1U /* file_number */, "100" /* smallest */,
+      "150" /* largest */, 1U /* file_size */);
+  file_map_[1U].first->being_compacted = true;
+  Add(1 /* level */, 2U /* file_number */, "150" /* smallest */,
+      "200" /* largest */, 1000000000U /* file_size */, 0 /* smallest_seq */,
+      0 /* largest_seq */);
+  Add(1 /* level */, 3U /* file_number */, "201" /* smallest */,
+      "250" /* largest */, 900000000U /* file_size */);
+  Add(2 /* level */, 4U /* file_number */, "100" /* smallest */,
+      "150" /* largest */, 1U /* file_size */);
+  Add(2 /* level */, 5U /* file_number */, "151" /* smallest */,
+      "200" /* largest */, 1U /* file_size */);
+  Add(2 /* level */, 6U /* file_number */, "201" /* smallest */,
+      "250" /* largest */, 1U /* file_size */);
+
+  UpdateVersionStorageInfo();
+
+  std::unique_ptr<Compaction> compaction(level_compaction_picker.PickCompaction(
+      cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_));
+  ASSERT_TRUE(compaction.get() != nullptr);
+  ASSERT_EQ(2U, compaction->num_input_levels());
+  ASSERT_EQ(1U, compaction->num_input_files(0));
+  ASSERT_EQ(1U, compaction->num_input_files(1));
+  ASSERT_EQ(3U, compaction->input(0, 0)->fd.GetNumber());
+  ASSERT_EQ(6U, compaction->input(1, 0)->fd.GetNumber());
+}
+
+TEST_F(CompactionPickerTest, OverlappingUserKeys11) {
+  // Locked file encountered when pulling in extra output-level files with same
+  // user keys. Expected to skip that compaction and pick the next-best choice.
+  NewVersionStorage(6, kCompactionStyleLevel);
+  mutable_cf_options_.max_compaction_bytes = 100000000000u;
+
+  // score(L1) = 3.7
+  // score(L2) = 1.85
+  // There is no eligible file in L1 to compact since both candidates pull in
+  // file_number 5U, which overlaps with a file pending compaction (6U). The
+  // first eligible compaction is from L2->L3.
+  Add(1 /* level */, 2U /* file_number */, "151" /* smallest */,
+      "200" /* largest */, 1000000000U /* file_size */);
+  Add(1 /* level */, 3U /* file_number */, "201" /* smallest */,
+      "250" /* largest */, 1U /* file_size */);
+  Add(2 /* level */, 4U /* file_number */, "100" /* smallest */,
+      "149" /* largest */, 5000000000U /* file_size */);
+  Add(2 /* level */, 5U /* file_number */, "150" /* smallest */,
+      "201" /* largest */, 1U /* file_size */);
+  Add(2 /* level */, 6U /* file_number */, "201" /* smallest */,
+      "249" /* largest */, 1U /* file_size */, 0 /* smallest_seq */,
+      0 /* largest_seq */);
+  file_map_[6U].first->being_compacted = true;
+  Add(3 /* level */, 7U /* file_number */, "100" /* smallest */,
+      "149" /* largest */, 1U /* file_size */);
+
+  UpdateVersionStorageInfo();
+
+  std::unique_ptr<Compaction> compaction(level_compaction_picker.PickCompaction(
+      cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_));
+  ASSERT_TRUE(compaction.get() != nullptr);
+  ASSERT_EQ(2U, compaction->num_input_levels());
+  ASSERT_EQ(1U, compaction->num_input_files(0));
+  ASSERT_EQ(1U, compaction->num_input_files(1));
+  ASSERT_EQ(4U, compaction->input(0, 0)->fd.GetNumber());
+  ASSERT_EQ(7U, compaction->input(1, 0)->fd.GetNumber());
+}
+
 TEST_F(CompactionPickerTest, NotScheduleL1IfL0WithHigherPri1) {
   NewVersionStorage(6, kCompactionStyleLevel);
   mutable_cf_options_.level0_file_num_compaction_trigger = 2;
@@ -1316,6 +1399,49 @@ TEST_F(CompactionPickerTest, IsTrivialMoveOff) {
   ASSERT_FALSE(compaction->IsTrivialMove());
 }
 
+TEST_F(CompactionPickerTest, CacheNextCompactionIndex) {
+  NewVersionStorage(6, kCompactionStyleLevel);
+  mutable_cf_options_.max_compaction_bytes = 100000000000u;
+
+  Add(1 /* level */, 1U /* file_number */, "100" /* smallest */,
+      "149" /* largest */, 1000000000U /* file_size */);
+  file_map_[1U].first->being_compacted = true;
+  Add(1 /* level */, 2U /* file_number */, "150" /* smallest */,
+      "199" /* largest */, 900000000U /* file_size */);
+  Add(1 /* level */, 3U /* file_number */, "200" /* smallest */,
+      "249" /* largest */, 800000000U /* file_size */);
+  Add(1 /* level */, 4U /* file_number */, "250" /* smallest */,
+      "299" /* largest */, 700000000U /* file_size */);
+  Add(2 /* level */, 5U /* file_number */, "150" /* smallest */,
+      "199" /* largest */, 1U /* file_size */);
+  file_map_[5U].first->being_compacted = true;
+
+  UpdateVersionStorageInfo();
+
+  std::unique_ptr<Compaction> compaction(level_compaction_picker.PickCompaction(
+      cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_));
+  ASSERT_TRUE(compaction.get() != nullptr);
+  ASSERT_EQ(1U, compaction->num_input_levels());
+  ASSERT_EQ(1U, compaction->num_input_files(0));
+  ASSERT_EQ(0U, compaction->num_input_files(1));
+  ASSERT_EQ(3U, compaction->input(0, 0)->fd.GetNumber());
+  ASSERT_EQ(2, vstorage_->NextCompactionIndex(1 /* level */));
+
+  compaction.reset(level_compaction_picker.PickCompaction(
+      cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_));
+  ASSERT_TRUE(compaction.get() != nullptr);
+  ASSERT_EQ(1U, compaction->num_input_levels());
+  ASSERT_EQ(1U, compaction->num_input_files(0));
+  ASSERT_EQ(0U, compaction->num_input_files(1));
+  ASSERT_EQ(4U, compaction->input(0, 0)->fd.GetNumber());
+  ASSERT_EQ(3, vstorage_->NextCompactionIndex(1 /* level */));
+
+  compaction.reset(level_compaction_picker.PickCompaction(
+      cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_));
+  ASSERT_TRUE(compaction.get() == nullptr);
+  ASSERT_EQ(4, vstorage_->NextCompactionIndex(1 /* level */));
+}
+
 }  // namespace rocksdb
 
 int main(int argc, char** argv) {

db/compaction_picker_universal.cc

@@ -35,7 +35,7 @@ namespace {
 // and the index of the file in that level
 
 struct InputFileInfo {
-  InputFileInfo() : f(nullptr) {}
+  InputFileInfo() : f(nullptr), level(0), index(0) {}
 
   FileMetaData* f;
   size_t level;
@@ -204,7 +204,8 @@ void UniversalCompactionPicker::SortedRun::DumpSizeInfo(
 
 std::vector<UniversalCompactionPicker::SortedRun>
 UniversalCompactionPicker::CalculateSortedRuns(
-    const VersionStorageInfo& vstorage, const ImmutableCFOptions& ioptions) {
+    const VersionStorageInfo& vstorage, const ImmutableCFOptions& ioptions,
+    const MutableCFOptions& mutable_cf_options) {
   std::vector<UniversalCompactionPicker::SortedRun> ret;
   for (FileMetaData* f : vstorage.LevelFiles(0)) {
     ret.emplace_back(0, f, f->fd.GetFileSize(), f->compensated_file_size,
@@ -218,7 +219,8 @@ UniversalCompactionPicker::CalculateSortedRuns(
     for (FileMetaData* f : vstorage.LevelFiles(level)) {
       total_compensated_size += f->compensated_file_size;
       total_size += f->fd.GetFileSize();
-      if (ioptions.compaction_options_universal.allow_trivial_move == true) {
+      if (mutable_cf_options.compaction_options_universal.allow_trivial_move ==
+          true) {
         if (f->being_compacted) {
           being_compacted = f->being_compacted;
         }
@@ -227,7 +229,8 @@ UniversalCompactionPicker::CalculateSortedRuns(
         // non-zero level, all the files should share the same being_compacted
         // value.
         // This assumption is only valid when
-        // ioptions.compaction_options_universal.allow_trivial_move is false
+        // mutable_cf_options.compaction_options_universal.allow_trivial_move is
+        // false
         assert(is_first || f->being_compacted == being_compacted);
       }
       if (is_first) {
@@ -245,14 +248,13 @@ UniversalCompactionPicker::CalculateSortedRuns(
 
 // Universal style of compaction. Pick files that are contiguous in
 // time-range to compact.
-//
 Compaction* UniversalCompactionPicker::PickCompaction(
     const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
     VersionStorageInfo* vstorage, LogBuffer* log_buffer) {
   const int kLevel0 = 0;
   double score = vstorage->CompactionScore(kLevel0);
   std::vector<SortedRun> sorted_runs =
-      CalculateSortedRuns(*vstorage, ioptions_);
+      CalculateSortedRuns(*vstorage, ioptions_, mutable_cf_options);
 
   if (sorted_runs.size() == 0 ||
       sorted_runs.size() <
@@ -279,7 +281,8 @@ Compaction* UniversalCompactionPicker::PickCompaction(
   } else {
     // Size amplification is within limits. Try reducing read
     // amplification while maintaining file size ratios.
-    unsigned int ratio = ioptions_.compaction_options_universal.size_ratio;
+    unsigned int ratio =
+        mutable_cf_options.compaction_options_universal.size_ratio;
 
     if ((c = PickCompactionToReduceSortedRuns(
              cf_name, mutable_cf_options, vstorage, score, ratio, UINT_MAX,
@@ -327,7 +330,8 @@ Compaction* UniversalCompactionPicker::PickCompaction(
     return nullptr;
   }
 
-  if (ioptions_.compaction_options_universal.allow_trivial_move == true) {
+  if (mutable_cf_options.compaction_options_universal.allow_trivial_move ==
+      true) {
     c->set_is_trivial_move(IsInputFilesNonOverlapping(c));
   }
 
@@ -373,6 +377,7 @@ Compaction* UniversalCompactionPicker::PickCompaction(
               c->inputs(0)->size());
 
   RegisterCompaction(c);
+  vstorage->ComputeCompactionScore(ioptions_, mutable_cf_options);
 
   TEST_SYNC_POINT_CALLBACK("UniversalCompactionPicker::PickCompaction:Return",
                            c);
@@ -380,7 +385,8 @@ Compaction* UniversalCompactionPicker::PickCompaction(
 }
 
 uint32_t UniversalCompactionPicker::GetPathId(
-    const ImmutableCFOptions& ioptions, uint64_t file_size) {
+    const ImmutableCFOptions& ioptions,
+    const MutableCFOptions& mutable_cf_options, uint64_t file_size) {
   // Two conditions need to be satisfied:
   // (1) the target path needs to be able to hold the file's size
   // (2) Total size left in this and previous paths need to be not
@@ -397,8 +403,8 @@ uint32_t UniversalCompactionPicker::GetPathId(
   // that case. We need to improve it.
   uint64_t accumulated_size = 0;
   uint64_t future_size =
-      file_size * (100 - ioptions.compaction_options_universal.size_ratio) /
-      100;
+      file_size *
+      (100 - mutable_cf_options.compaction_options_universal.size_ratio) / 100;
   uint32_t p = 0;
   assert(!ioptions.db_paths.empty());
   for (; p < ioptions.db_paths.size() - 1; p++) {
@@ -422,9 +428,9 @@ Compaction* UniversalCompactionPicker::PickCompactionToReduceSortedRuns(
     unsigned int max_number_of_files_to_compact,
     const std::vector<SortedRun>& sorted_runs, LogBuffer* log_buffer) {
   unsigned int min_merge_width =
-      ioptions_.compaction_options_universal.min_merge_width;
+      mutable_cf_options.compaction_options_universal.min_merge_width;
   unsigned int max_merge_width =
-      ioptions_.compaction_options_universal.max_merge_width;
+      mutable_cf_options.compaction_options_universal.max_merge_width;
 
   const SortedRun* sr = nullptr;
   bool done = false;
@@ -491,7 +497,7 @@ Compaction* UniversalCompactionPicker::PickCompactionToReduceSortedRuns(
       if (sz < static_cast<double>(succeeding_sr->size)) {
         break;
       }
-      if (ioptions_.compaction_options_universal.stop_style ==
+      if (mutable_cf_options.compaction_options_universal.stop_style ==
           kCompactionStopStyleSimilarSize) {
         // Similar-size stopping rule: also check the last picked file isn't
         // far larger than the next candidate file.
@@ -534,7 +540,7 @@ Compaction* UniversalCompactionPicker::PickCompactionToReduceSortedRuns(
   // size ratio of compression.
   bool enable_compression = true;
   int ratio_to_compress =
-      ioptions_.compaction_options_universal.compression_size_percent;
+      mutable_cf_options.compaction_options_universal.compression_size_percent;
   if (ratio_to_compress >= 0) {
     uint64_t total_size = 0;
     for (auto& sorted_run : sorted_runs) {
@@ -555,7 +561,8 @@ Compaction* UniversalCompactionPicker::PickCompactionToReduceSortedRuns(
   for (unsigned int i = 0; i < first_index_after; i++) {
     estimated_total_size += sorted_runs[i].size;
   }
-  uint32_t path_id = GetPathId(ioptions_, estimated_total_size);
+  uint32_t path_id =
+      GetPathId(ioptions_, mutable_cf_options, estimated_total_size);
   int start_level = sorted_runs[start_index].level;
   int output_level;
   if (first_index_after == sorted_runs.size()) {
@@ -620,8 +627,8 @@ Compaction* UniversalCompactionPicker::PickCompactionToReduceSizeAmp(
     VersionStorageInfo* vstorage, double score,
     const std::vector<SortedRun>& sorted_runs, LogBuffer* log_buffer) {
   // percentage flexibility while reducing size amplification
-  uint64_t ratio =
-      ioptions_.compaction_options_universal.max_size_amplification_percent;
+  uint64_t ratio = mutable_cf_options.compaction_options_universal
+                       .max_size_amplification_percent;
 
   unsigned int candidate_count = 0;
   uint64_t candidate_size = 0;
@@ -699,7 +706,8 @@ Compaction* UniversalCompactionPicker::PickCompactionToReduceSizeAmp(
   for (size_t loop = start_index; loop < sorted_runs.size(); loop++) {
     estimated_total_size += sorted_runs[loop].size;
   }
-  uint32_t path_id = GetPathId(ioptions_, estimated_total_size);
+  uint32_t path_id =
+      GetPathId(ioptions_, mutable_cf_options, estimated_total_size);
   int start_level = sorted_runs[start_index].level;
 
   std::vector<CompactionInputFiles> inputs(vstorage->num_levels());

db/compaction_picker_universal.h

@@ -80,11 +80,13 @@ class UniversalCompactionPicker : public CompactionPicker {
   bool IsInputFilesNonOverlapping(Compaction* c);
 
   static std::vector<SortedRun> CalculateSortedRuns(
-      const VersionStorageInfo& vstorage, const ImmutableCFOptions& ioptions);
+      const VersionStorageInfo& vstorage, const ImmutableCFOptions& ioptions,
+      const MutableCFOptions& mutable_cf_options);
 
   // Pick a path ID to place a newly generated file, with its estimated file
   // size.
   static uint32_t GetPathId(const ImmutableCFOptions& ioptions,
+                            const MutableCFOptions& mutable_cf_options,
                             uint64_t file_size);
 };
 }  // namespace rocksdb

db/convenience.cc

@@ -9,19 +9,50 @@
 #include "rocksdb/convenience.h"
 
 #include "db/db_impl.h"
+#include "util/cast_util.h"
 
 namespace rocksdb {
 
 void CancelAllBackgroundWork(DB* db, bool wait) {
-  (dynamic_cast<DBImpl*>(db->GetRootDB()))->CancelAllBackgroundWork(wait);
+  (static_cast_with_check<DBImpl, DB>(db->GetRootDB()))
+      ->CancelAllBackgroundWork(wait);
 }
 
 Status DeleteFilesInRange(DB* db, ColumnFamilyHandle* column_family,
                           const Slice* begin, const Slice* end) {
-  return (dynamic_cast<DBImpl*>(db->GetRootDB()))
+  return (static_cast_with_check<DBImpl, DB>(db->GetRootDB()))
       ->DeleteFilesInRange(column_family, begin, end);
 }
 
+Status VerifySstFileChecksum(const Options& options,
+                             const EnvOptions& env_options,
+                             const std::string& file_path) {
+  unique_ptr<RandomAccessFile> file;
+  uint64_t file_size;
+  InternalKeyComparator internal_comparator(options.comparator);
+  ImmutableCFOptions ioptions(options);
+
+  Status s = ioptions.env->NewRandomAccessFile(file_path, &file, env_options);
+  if (s.ok()) {
+    s = ioptions.env->GetFileSize(file_path, &file_size);
+  } else {
+    return s;
+  }
+  unique_ptr<TableReader> table_reader;
+  std::unique_ptr<RandomAccessFileReader> file_reader(
+      new RandomAccessFileReader(std::move(file), file_path));
+  s = ioptions.table_factory->NewTableReader(
+      TableReaderOptions(ioptions, env_options, internal_comparator,
+                         false /* skip_filters */, -1 /* level */),
+      std::move(file_reader), file_size, &table_reader,
+      false /* prefetch_index_and_filter_in_cache */);
+  if (!s.ok()) {
+    return s;
+  }
+  s = table_reader->VerifyChecksum();
+  return s;
+}
+
 }  // namespace rocksdb
 
 #endif  // ROCKSDB_LITE

db/corruption_test.cc

@@ -20,6 +20,7 @@
 #include "db/log_format.h"
 #include "db/version_set.h"
 #include "rocksdb/cache.h"
+#include "rocksdb/convenience.h"
 #include "rocksdb/env.h"
 #include "rocksdb/table.h"
 #include "rocksdb/write_batch.h"
@@ -152,7 +153,7 @@ class CorruptionTest : public testing::Test {
     struct stat sbuf;
     if (stat(fname.c_str(), &sbuf) != 0) {
       const char* msg = strerror(errno);
-      ASSERT_TRUE(false) << fname << ": " << msg;
+      FAIL() << fname << ": " << msg;
     }
 
     if (offset < 0) {
@@ -179,6 +180,9 @@ class CorruptionTest : public testing::Test {
     }
     s = WriteStringToFile(Env::Default(), contents, fname);
     ASSERT_TRUE(s.ok()) << s.ToString();
+    Options options;
+    EnvOptions env_options;
+    ASSERT_NOK(VerifySstFileChecksum(options, env_options, fname));
   }
 
   void Corrupt(FileType filetype, int offset, int bytes_to_corrupt) {
@@ -213,7 +217,7 @@ class CorruptionTest : public testing::Test {
         return;
       }
     }
-    ASSERT_TRUE(false) << "no file found at level";
+    FAIL() << "no file found at level";
   }
 
 
@@ -312,6 +316,7 @@ TEST_F(CorruptionTest, TableFile) {
 
   Corrupt(kTableFile, 100, 1);
   Check(99, 99);
+  ASSERT_NOK(dbi->VerifyChecksum());
 }
 
 TEST_F(CorruptionTest, TableFileIndexData) {
@@ -327,9 +332,11 @@ TEST_F(CorruptionTest, TableFileIndexData) {
   // corrupt an index block of an entire file
   Corrupt(kTableFile, -2000, 500);
   Reopen();
+  dbi = reinterpret_cast<DBImpl*>(db_);
   // one full file should be readable, since only one was corrupted
   // the other file should be fully non-readable, since index was corrupted
   Check(5000, 5000);
+  ASSERT_NOK(dbi->VerifyChecksum());
 }
 
 TEST_F(CorruptionTest, MissingDescriptor) {
@@ -389,10 +396,12 @@ TEST_F(CorruptionTest, CompactionInputError) {
 
   Corrupt(kTableFile, 100, 1);
   Check(9, 9);
+  ASSERT_NOK(dbi->VerifyChecksum());
 
   // Force compactions by writing lots of values
   Build(10000);
   Check(10000, 10000);
+  ASSERT_NOK(dbi->VerifyChecksum());
 }
 
 TEST_F(CorruptionTest, CompactionInputErrorParanoid) {
@@ -424,6 +433,7 @@ TEST_F(CorruptionTest, CompactionInputErrorParanoid) {
 
   CorruptTableFileAtLevel(0, 100, 1);
   Check(9, 9);
+  ASSERT_NOK(dbi->VerifyChecksum());
 
   // Write must eventually fail because of corrupted table
   Status s;
@@ -445,6 +455,7 @@ TEST_F(CorruptionTest, UnrelatedKeys) {
   DBImpl* dbi = reinterpret_cast<DBImpl*>(db_);
   dbi->TEST_FlushMemTable();
   Corrupt(kTableFile, 100, 1);
+  ASSERT_NOK(dbi->VerifyChecksum());
 
   std::string tmp1, tmp2;
   ASSERT_OK(db_->Put(WriteOptions(), Key(1000, &tmp1), Value(1000, &tmp2)));

db/cuckoo_table_db_test.cc

@@ -241,7 +241,7 @@ TEST_F(CuckooTableDBTest, CompactionIntoMultipleFiles) {
 
   // Write 28 values, each 10016 B ~ 10KB
   for (int idx = 0; idx < 28; ++idx) {
-    ASSERT_OK(Put(Key(idx), std::string(10000, 'a' + idx)));
+    ASSERT_OK(Put(Key(idx), std::string(10000, 'a' + char(idx))));
   }
   dbfull()->TEST_WaitForFlushMemTable();
   ASSERT_EQ("1", FilesPerLevel());
@@ -250,7 +250,7 @@ TEST_F(CuckooTableDBTest, CompactionIntoMultipleFiles) {
                               true /* disallow trivial move */);
   ASSERT_EQ("0,2", FilesPerLevel());
   for (int idx = 0; idx < 28; ++idx) {
-    ASSERT_EQ(std::string(10000, 'a' + idx), Get(Key(idx)));
+    ASSERT_EQ(std::string(10000, 'a' + char(idx)), Get(Key(idx)));
   }
 }
 
@@ -271,14 +271,14 @@ TEST_F(CuckooTableDBTest, SameKeyInsertedInTwoDifferentFilesAndCompacted) {
 
   // Generate one more file in level-0, and should trigger level-0 compaction
   for (int idx = 0; idx < 11; ++idx) {
-    ASSERT_OK(Put(Key(idx), std::string(10000, 'a' + idx)));
+    ASSERT_OK(Put(Key(idx), std::string(10000, 'a' + char(idx))));
   }
   dbfull()->TEST_WaitForFlushMemTable();
   dbfull()->TEST_CompactRange(0, nullptr, nullptr);
 
   ASSERT_EQ("0,1", FilesPerLevel());
   for (int idx = 0; idx < 11; ++idx) {
-    ASSERT_EQ(std::string(10000, 'a' + idx), Get(Key(idx)));
+    ASSERT_EQ(std::string(10000, 'a' + char(idx)), Get(Key(idx)));
   }
 }
 

db/db_basic_test.cc

@@ -41,7 +41,7 @@ TEST_F(DBBasicTest, ReadOnlyDB) {
   Close();
 
   auto options = CurrentOptions();
-  assert(options.env = env_);
+  assert(options.env == env_);
   ASSERT_OK(ReadOnlyReopen(options));
   ASSERT_EQ("v3", Get("foo"));
   ASSERT_EQ("v2", Get("bar"));
@@ -175,6 +175,8 @@ TEST_F(DBBasicTest, LevelLimitReopen) {
   int i = 0;
   while (NumTableFilesAtLevel(2, 1) == 0) {
     ASSERT_OK(Put(1, Key(i++), value));
+    dbfull()->TEST_WaitForFlushMemTable();
+    dbfull()->TEST_WaitForCompact();
   }
 
   options.num_levels = 1;
@@ -358,7 +360,6 @@ TEST_F(DBBasicTest, FLUSH) {
     WriteOptions writeOpt = WriteOptions();
     writeOpt.disableWAL = true;
     SetPerfLevel(kEnableTime);
-    ;
     ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "foo", "v1"));
     // this will now also flush the last 2 writes
     ASSERT_OK(Flush(1));
@@ -367,6 +368,7 @@ TEST_F(DBBasicTest, FLUSH) {
     get_perf_context()->Reset();
     Get(1, "foo");
     ASSERT_TRUE((int)get_perf_context()->get_from_output_files_time > 0);
+    ASSERT_EQ(2, (int)get_perf_context()->get_read_bytes);
 
     ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions());
     ASSERT_EQ("v1", Get(1, "foo"));
@@ -568,19 +570,19 @@ TEST_F(DBBasicTest, CompactBetweenSnapshots) {
 
 TEST_F(DBBasicTest, DBOpen_Options) {
   Options options = CurrentOptions();
-  std::string dbname = test::TmpDir(env_) + "/db_options_test";
-  ASSERT_OK(DestroyDB(dbname, options));
+  Close();
+  Destroy(options);
 
   // Does not exist, and create_if_missing == false: error
   DB* db = nullptr;
   options.create_if_missing = false;
-  Status s = DB::Open(options, dbname, &db);
+  Status s = DB::Open(options, dbname_, &db);
   ASSERT_TRUE(strstr(s.ToString().c_str(), "does not exist") != nullptr);
   ASSERT_TRUE(db == nullptr);
 
   // Does not exist, and create_if_missing == true: OK
   options.create_if_missing = true;
-  s = DB::Open(options, dbname, &db);
+  s = DB::Open(options, dbname_, &db);
   ASSERT_OK(s);
   ASSERT_TRUE(db != nullptr);
 
@@ -590,14 +592,14 @@ TEST_F(DBBasicTest, DBOpen_Options) {
   // Does exist, and error_if_exists == true: error
   options.create_if_missing = false;
   options.error_if_exists = true;
-  s = DB::Open(options, dbname, &db);
+  s = DB::Open(options, dbname_, &db);
   ASSERT_TRUE(strstr(s.ToString().c_str(), "exists") != nullptr);
   ASSERT_TRUE(db == nullptr);
 
   // Does exist, and error_if_exists == false: OK
   options.create_if_missing = true;
   options.error_if_exists = false;
-  s = DB::Open(options, dbname, &db);
+  s = DB::Open(options, dbname_, &db);
   ASSERT_OK(s);
   ASSERT_TRUE(db != nullptr);
 
@@ -723,6 +725,7 @@ TEST_F(DBBasicTest, FlushOneColumnFamily) {
 TEST_F(DBBasicTest, MultiGetSimple) {
   do {
     CreateAndReopenWithCF({"pikachu"}, CurrentOptions());
+    SetPerfLevel(kEnableCount);
     ASSERT_OK(Put(1, "k1", "v1"));
     ASSERT_OK(Put(1, "k2", "v2"));
     ASSERT_OK(Put(1, "k3", "v3"));
@@ -736,12 +739,15 @@ TEST_F(DBBasicTest, MultiGetSimple) {
     std::vector<std::string> values(20, "Temporary data to be overwritten");
     std::vector<ColumnFamilyHandle*> cfs(keys.size(), handles_[1]);
 
+    get_perf_context()->Reset();
     std::vector<Status> s = db_->MultiGet(ReadOptions(), cfs, keys, &values);
     ASSERT_EQ(values.size(), keys.size());
     ASSERT_EQ(values[0], "v1");
     ASSERT_EQ(values[1], "v2");
     ASSERT_EQ(values[2], "v3");
     ASSERT_EQ(values[4], "v5");
+    // four kv pairs * two bytes per value
+    ASSERT_EQ(8, (int)get_perf_context()->multiget_read_bytes);
 
     ASSERT_OK(s[0]);
     ASSERT_OK(s[1]);
@@ -749,6 +755,7 @@ TEST_F(DBBasicTest, MultiGetSimple) {
     ASSERT_TRUE(s[3].IsNotFound());
     ASSERT_OK(s[4]);
     ASSERT_TRUE(s[5].IsNotFound());
+    SetPerfLevel(kDisable);
   } while (ChangeCompactOptions());
 }
 
@@ -785,36 +792,30 @@ TEST_F(DBBasicTest, MultiGetEmpty) {
 TEST_F(DBBasicTest, ChecksumTest) {
   BlockBasedTableOptions table_options;
   Options options = CurrentOptions();
+  // change when new checksum type added
+  int max_checksum = static_cast<int>(kxxHash);
+  const int kNumPerFile = 2;
 
-  table_options.checksum = kCRC32c;
-  options.table_factory.reset(NewBlockBasedTableFactory(table_options));
-  Reopen(options);
-  ASSERT_OK(Put("a", "b"));
-  ASSERT_OK(Put("c", "d"));
-  ASSERT_OK(Flush());  // table with crc checksum
+  // generate one table with each type of checksum
+  for (int i = 0; i <= max_checksum; ++i) {
+    table_options.checksum = static_cast<ChecksumType>(i);
+    options.table_factory.reset(NewBlockBasedTableFactory(table_options));
+    Reopen(options);
+    for (int j = 0; j < kNumPerFile; ++j) {
+      ASSERT_OK(Put(Key(i * kNumPerFile + j), Key(i * kNumPerFile + j)));
+    }
+    ASSERT_OK(Flush());
+  }
 
-  table_options.checksum = kxxHash;
-  options.table_factory.reset(NewBlockBasedTableFactory(table_options));
-  Reopen(options);
-  ASSERT_OK(Put("e", "f"));
-  ASSERT_OK(Put("g", "h"));
-  ASSERT_OK(Flush());  // table with xxhash checksum
-
-  table_options.checksum = kCRC32c;
-  options.table_factory.reset(NewBlockBasedTableFactory(table_options));
-  Reopen(options);
-  ASSERT_EQ("b", Get("a"));
-  ASSERT_EQ("d", Get("c"));
-  ASSERT_EQ("f", Get("e"));
-  ASSERT_EQ("h", Get("g"));
-
-  table_options.checksum = kCRC32c;
-  options.table_factory.reset(NewBlockBasedTableFactory(table_options));
-  Reopen(options);
-  ASSERT_EQ("b", Get("a"));
-  ASSERT_EQ("d", Get("c"));
-  ASSERT_EQ("f", Get("e"));
-  ASSERT_EQ("h", Get("g"));
+  // verify data with each type of checksum
+  for (int i = 0; i <= kxxHash; ++i) {
+    table_options.checksum = static_cast<ChecksumType>(i);
+    options.table_factory.reset(NewBlockBasedTableFactory(table_options));
+    Reopen(options);
+    for (int j = 0; j < (max_checksum + 1) * kNumPerFile; ++j) {
+      ASSERT_EQ(Key(j), Get(Key(j)));
+    }
+  }
 }
 
 // On Windows you can have either memory mapped file or a file
@@ -846,6 +847,57 @@ TEST_F(DBBasicTest, MmapAndBufferOptions) {
 }
 #endif
 
+class TestEnv : public EnvWrapper {
+  public:
+    explicit TestEnv(Env* base) : EnvWrapper(base) { };
+
+    class TestLogger : public Logger {
+      public:
+        using Logger::Logv;
+        virtual void Logv(const char *format, va_list ap) override { };
+      private:
+        virtual Status CloseImpl() override {
+          return Status::NotSupported();
+        }
+    };
+
+    virtual Status NewLogger(const std::string& fname,
+                             shared_ptr<Logger>* result) {
+      result->reset(new TestLogger());
+      return Status::OK();
+    }
+};
+
+TEST_F(DBBasicTest, DBClose) {
+  Options options = GetDefaultOptions();
+  std::string dbname = test::TmpDir(env_) + "/db_close_test";
+  ASSERT_OK(DestroyDB(dbname, options));
+
+  DB* db = nullptr;
+  options.create_if_missing = true;
+  options.env = new TestEnv(Env::Default());
+  Status s = DB::Open(options, dbname, &db);
+  ASSERT_OK(s);
+  ASSERT_TRUE(db != nullptr);
+
+  s = db->Close();
+  ASSERT_EQ(s, Status::NotSupported());
+
+  delete db;
+
+  // Provide our own logger and ensure DB::Close() does not close it
+  options.info_log.reset(new TestEnv::TestLogger());
+  options.create_if_missing = false;
+  s = DB::Open(options, dbname, &db);
+  ASSERT_OK(s);
+  ASSERT_TRUE(db != nullptr);
+
+  s = db->Close();
+  ASSERT_EQ(s, Status::OK());
+  delete db;
+  delete options.env;
+}
+
 }  // namespace rocksdb
 
 int main(int argc, char** argv) {

db/db_blob_index_test.cc

@@ -0,0 +1,410 @@
+//  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).
+//
+// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
+// Use of this source 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 <functional>
+#include <string>
+#include <utility>
+#include <vector>
+
+#include "db/column_family.h"
+#include "db/db_iter.h"
+#include "db/db_test_util.h"
+#include "db/dbformat.h"
+#include "db/write_batch_internal.h"
+#include "port/port.h"
+#include "port/stack_trace.h"
+#include "util/string_util.h"
+#include "utilities/merge_operators.h"
+
+namespace rocksdb {
+
+// kTypeBlobIndex is a value type used by BlobDB only. The base rocksdb
+// should accept the value type on write, and report not supported value
+// for reads, unless caller request for it explicitly. The base rocksdb
+// doesn't understand format of actual blob index (the value).
+class DBBlobIndexTest : public DBTestBase {
+ public:
+  enum Tier {
+    kMemtable = 0,
+    kImmutableMemtables = 1,
+    kL0SstFile = 2,
+    kLnSstFile = 3,
+  };
+  const std::vector<Tier> kAllTiers = {Tier::kMemtable,
+                                       Tier::kImmutableMemtables,
+                                       Tier::kL0SstFile, Tier::kLnSstFile};
+
+  DBBlobIndexTest() : DBTestBase("/db_blob_index_test") {}
+
+  ColumnFamilyHandle* cfh() { return dbfull()->DefaultColumnFamily(); }
+
+  ColumnFamilyData* cfd() {
+    return reinterpret_cast<ColumnFamilyHandleImpl*>(cfh())->cfd();
+  }
+
+  Status PutBlobIndex(WriteBatch* batch, const Slice& key,
+                      const Slice& blob_index) {
+    return WriteBatchInternal::PutBlobIndex(batch, cfd()->GetID(), key,
+                                            blob_index);
+  }
+
+  Status Write(WriteBatch* batch) {
+    return dbfull()->Write(WriteOptions(), batch);
+  }
+
+  std::string GetImpl(const Slice& key, bool* is_blob_index = nullptr,
+                      const Snapshot* snapshot = nullptr) {
+    ReadOptions read_options;
+    read_options.snapshot = snapshot;
+    PinnableSlice value;
+    auto s = dbfull()->GetImpl(read_options, cfh(), key, &value,
+                               nullptr /*value_found*/, nullptr /*callback*/,
+                               is_blob_index);
+    if (s.IsNotFound()) {
+      return "NOT_FOUND";
+    }
+    if (s.IsNotSupported()) {
+      return "NOT_SUPPORTED";
+    }
+    if (!s.ok()) {
+      return s.ToString();
+    }
+    return value.ToString();
+  }
+
+  std::string GetBlobIndex(const Slice& key,
+                           const Snapshot* snapshot = nullptr) {
+    bool is_blob_index = false;
+    std::string value = GetImpl(key, &is_blob_index, snapshot);
+    if (!is_blob_index) {
+      return "NOT_BLOB";
+    }
+    return value;
+  }
+
+  ArenaWrappedDBIter* GetBlobIterator() {
+    return dbfull()->NewIteratorImpl(
+        ReadOptions(), cfd(), dbfull()->GetLatestSequenceNumber(),
+        nullptr /*read_callback*/, true /*allow_blob*/);
+  }
+
+  Options GetTestOptions() {
+    Options options;
+    options.create_if_missing = true;
+    options.num_levels = 2;
+    options.disable_auto_compactions = true;
+    // Disable auto flushes.
+    options.max_write_buffer_number = 10;
+    options.min_write_buffer_number_to_merge = 10;
+    options.merge_operator = MergeOperators::CreateStringAppendOperator();
+    return options;
+  }
+
+  void MoveDataTo(Tier tier) {
+    switch (tier) {
+      case Tier::kMemtable:
+        break;
+      case Tier::kImmutableMemtables:
+        ASSERT_OK(dbfull()->TEST_SwitchMemtable());
+        break;
+      case Tier::kL0SstFile:
+        ASSERT_OK(Flush());
+        break;
+      case Tier::kLnSstFile:
+        ASSERT_OK(Flush());
+        ASSERT_OK(Put("a", "dummy"));
+        ASSERT_OK(Put("z", "dummy"));
+        ASSERT_OK(Flush());
+        ASSERT_OK(
+            dbfull()->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+#ifndef ROCKSDB_LITE
+        ASSERT_EQ("0,1", FilesPerLevel());
+#endif  // !ROCKSDB_LITE
+        break;
+    }
+  }
+};
+
+// Should be able to write kTypeBlobIndex to memtables and SST files.
+TEST_F(DBBlobIndexTest, Write) {
+  for (auto tier : kAllTiers) {
+    DestroyAndReopen(GetTestOptions());
+    for (int i = 1; i <= 5; i++) {
+      std::string index = ToString(i);
+      WriteBatch batch;
+      ASSERT_OK(PutBlobIndex(&batch, "key" + index, "blob" + index));
+      ASSERT_OK(Write(&batch));
+    }
+    MoveDataTo(tier);
+    for (int i = 1; i <= 5; i++) {
+      std::string index = ToString(i);
+      ASSERT_EQ("blob" + index, GetBlobIndex("key" + index));
+    }
+  }
+}
+
+// Get should be able to return blob index if is_blob_index is provided,
+// otherwise return Status::NotSupported status.
+TEST_F(DBBlobIndexTest, Get) {
+  for (auto tier : kAllTiers) {
+    DestroyAndReopen(GetTestOptions());
+    WriteBatch batch;
+    ASSERT_OK(batch.Put("key", "value"));
+    ASSERT_OK(PutBlobIndex(&batch, "blob_key", "blob_index"));
+    ASSERT_OK(Write(&batch));
+    MoveDataTo(tier);
+    // Verify normal value
+    bool is_blob_index = false;
+    PinnableSlice value;
+    ASSERT_EQ("value", Get("key"));
+    ASSERT_EQ("value", GetImpl("key"));
+    ASSERT_EQ("value", GetImpl("key", &is_blob_index));
+    ASSERT_FALSE(is_blob_index);
+    // Verify blob index
+    ASSERT_TRUE(Get("blob_key", &value).IsNotSupported());
+    ASSERT_EQ("NOT_SUPPORTED", GetImpl("blob_key"));
+    ASSERT_EQ("blob_index", GetImpl("blob_key", &is_blob_index));
+    ASSERT_TRUE(is_blob_index);
+  }
+}
+
+// Get should NOT return Status::NotSupported if blob index is updated with
+// a normal value.
+TEST_F(DBBlobIndexTest, Updated) {
+  for (auto tier : kAllTiers) {
+    DestroyAndReopen(GetTestOptions());
+    WriteBatch batch;
+    for (int i = 0; i < 10; i++) {
+      ASSERT_OK(PutBlobIndex(&batch, "key" + ToString(i), "blob_index"));
+    }
+    ASSERT_OK(Write(&batch));
+    // Avoid blob values from being purged.
+    const Snapshot* snapshot = dbfull()->GetSnapshot();
+    ASSERT_OK(Put("key1", "new_value"));
+    ASSERT_OK(Merge("key2", "a"));
+    ASSERT_OK(Merge("key2", "b"));
+    ASSERT_OK(Merge("key2", "c"));
+    ASSERT_OK(Delete("key3"));
+    ASSERT_OK(SingleDelete("key4"));
+    ASSERT_OK(Delete("key5"));
+    ASSERT_OK(Merge("key5", "a"));
+    ASSERT_OK(Merge("key5", "b"));
+    ASSERT_OK(Merge("key5", "c"));
+    ASSERT_OK(dbfull()->DeleteRange(WriteOptions(), cfh(), "key6", "key9"));
+    MoveDataTo(tier);
+    for (int i = 0; i < 10; i++) {
+      ASSERT_EQ("blob_index", GetBlobIndex("key" + ToString(i), snapshot));
+    }
+    ASSERT_EQ("new_value", Get("key1"));
+    ASSERT_EQ("NOT_SUPPORTED", GetImpl("key2"));
+    ASSERT_EQ("NOT_FOUND", Get("key3"));
+    ASSERT_EQ("NOT_FOUND", Get("key4"));
+    ASSERT_EQ("a,b,c", GetImpl("key5"));
+    for (int i = 6; i < 9; i++) {
+      ASSERT_EQ("NOT_FOUND", Get("key" + ToString(i)));
+    }
+    ASSERT_EQ("blob_index", GetBlobIndex("key9"));
+    dbfull()->ReleaseSnapshot(snapshot);
+  }
+}
+
+// Iterator should get blob value if allow_blob flag is set,
+// otherwise return Status::NotSupported status.
+TEST_F(DBBlobIndexTest, Iterate) {
+  const std::vector<std::vector<ValueType>> data = {
+      /*00*/ {kTypeValue},
+      /*01*/ {kTypeBlobIndex},
+      /*02*/ {kTypeValue},
+      /*03*/ {kTypeBlobIndex, kTypeValue},
+      /*04*/ {kTypeValue},
+      /*05*/ {kTypeValue, kTypeBlobIndex},
+      /*06*/ {kTypeValue},
+      /*07*/ {kTypeDeletion, kTypeBlobIndex},
+      /*08*/ {kTypeValue},
+      /*09*/ {kTypeSingleDeletion, kTypeBlobIndex},
+      /*10*/ {kTypeValue},
+      /*11*/ {kTypeMerge, kTypeMerge, kTypeMerge, kTypeBlobIndex},
+      /*12*/ {kTypeValue},
+      /*13*/
+      {kTypeMerge, kTypeMerge, kTypeMerge, kTypeDeletion, kTypeBlobIndex},
+      /*14*/ {kTypeValue},
+      /*15*/ {kTypeBlobIndex},
+      /*16*/ {kTypeValue},
+  };
+
+  auto get_key = [](int index) {
+    char buf[20];
+    snprintf(buf, sizeof(buf), "%02d", index);
+    return "key" + std::string(buf);
+  };
+
+  auto get_value = [&](int index, int version) {
+    return get_key(index) + "_value" + ToString(version);
+  };
+
+  auto check_iterator = [&](Iterator* iterator, Status::Code expected_status,
+                            const Slice& expected_value) {
+    ASSERT_EQ(expected_status, iterator->status().code());
+    if (expected_status == Status::kOk) {
+      ASSERT_TRUE(iterator->Valid());
+      ASSERT_EQ(expected_value, iterator->value());
+    } else {
+      ASSERT_FALSE(iterator->Valid());
+    }
+  };
+
+  auto create_normal_iterator = [&]() -> Iterator* {
+    return dbfull()->NewIterator(ReadOptions());
+  };
+
+  auto create_blob_iterator = [&]() -> Iterator* { return GetBlobIterator(); };
+
+  auto check_is_blob = [&](bool is_blob) {
+    return [is_blob](Iterator* iterator) {
+      ASSERT_EQ(is_blob,
+                reinterpret_cast<ArenaWrappedDBIter*>(iterator)->IsBlob());
+    };
+  };
+
+  auto verify = [&](int index, Status::Code expected_status,
+                    const Slice& forward_value, const Slice& backward_value,
+                    std::function<Iterator*()> create_iterator,
+                    std::function<void(Iterator*)> extra_check = nullptr) {
+    // Seek
+    auto* iterator = create_iterator();
+    ASSERT_OK(iterator->Refresh());
+    iterator->Seek(get_key(index));
+    check_iterator(iterator, expected_status, forward_value);
+    if (extra_check) {
+      extra_check(iterator);
+    }
+    delete iterator;
+
+    // Next
+    iterator = create_iterator();
+    ASSERT_OK(iterator->Refresh());
+    iterator->Seek(get_key(index - 1));
+    ASSERT_TRUE(iterator->Valid());
+    iterator->Next();
+    check_iterator(iterator, expected_status, forward_value);
+    if (extra_check) {
+      extra_check(iterator);
+    }
+    delete iterator;
+
+    // SeekForPrev
+    iterator = create_iterator();
+    ASSERT_OK(iterator->Refresh());
+    iterator->SeekForPrev(get_key(index));
+    check_iterator(iterator, expected_status, backward_value);
+    if (extra_check) {
+      extra_check(iterator);
+    }
+    delete iterator;
+
+    // Prev
+    iterator = create_iterator();
+    iterator->Seek(get_key(index + 1));
+    ASSERT_TRUE(iterator->Valid());
+    iterator->Prev();
+    check_iterator(iterator, expected_status, backward_value);
+    if (extra_check) {
+      extra_check(iterator);
+    }
+    delete iterator;
+  };
+
+  for (auto tier : {Tier::kMemtable} /*kAllTiers*/) {
+    // Avoid values from being purged.
+    std::vector<const Snapshot*> snapshots;
+    DestroyAndReopen(GetTestOptions());
+
+    // fill data
+    for (int i = 0; i < static_cast<int>(data.size()); i++) {
+      for (int j = static_cast<int>(data[i].size()) - 1; j >= 0; j--) {
+        std::string key = get_key(i);
+        std::string value = get_value(i, j);
+        WriteBatch batch;
+        switch (data[i][j]) {
+          case kTypeValue:
+            ASSERT_OK(Put(key, value));
+            break;
+          case kTypeDeletion:
+            ASSERT_OK(Delete(key));
+            break;
+          case kTypeSingleDeletion:
+            ASSERT_OK(SingleDelete(key));
+            break;
+          case kTypeMerge:
+            ASSERT_OK(Merge(key, value));
+            break;
+          case kTypeBlobIndex:
+            ASSERT_OK(PutBlobIndex(&batch, key, value));
+            ASSERT_OK(Write(&batch));
+            break;
+          default:
+            assert(false);
+        };
+      }
+      snapshots.push_back(dbfull()->GetSnapshot());
+    }
+    ASSERT_OK(
+        dbfull()->DeleteRange(WriteOptions(), cfh(), get_key(15), get_key(16)));
+    snapshots.push_back(dbfull()->GetSnapshot());
+    MoveDataTo(tier);
+
+    // Normal iterator
+    verify(1, Status::kNotSupported, "", "", create_normal_iterator);
+    verify(3, Status::kNotSupported, "", "", create_normal_iterator);
+    verify(5, Status::kOk, get_value(5, 0), get_value(5, 0),
+           create_normal_iterator);
+    verify(7, Status::kOk, get_value(8, 0), get_value(6, 0),
+           create_normal_iterator);
+    verify(9, Status::kOk, get_value(10, 0), get_value(8, 0),
+           create_normal_iterator);
+    verify(11, Status::kNotSupported, "", "", create_normal_iterator);
+    verify(13, Status::kOk,
+           get_value(13, 2) + "," + get_value(13, 1) + "," + get_value(13, 0),
+           get_value(13, 2) + "," + get_value(13, 1) + "," + get_value(13, 0),
+           create_normal_iterator);
+    verify(15, Status::kOk, get_value(16, 0), get_value(14, 0),
+           create_normal_iterator);
+
+    // Iterator with blob support
+    verify(1, Status::kOk, get_value(1, 0), get_value(1, 0),
+           create_blob_iterator, check_is_blob(true));
+    verify(3, Status::kOk, get_value(3, 0), get_value(3, 0),
+           create_blob_iterator, check_is_blob(true));
+    verify(5, Status::kOk, get_value(5, 0), get_value(5, 0),
+           create_blob_iterator, check_is_blob(false));
+    verify(7, Status::kOk, get_value(8, 0), get_value(6, 0),
+           create_blob_iterator, check_is_blob(false));
+    verify(9, Status::kOk, get_value(10, 0), get_value(8, 0),
+           create_blob_iterator, check_is_blob(false));
+    verify(11, Status::kNotSupported, "", "", create_blob_iterator);
+    verify(13, Status::kOk,
+           get_value(13, 2) + "," + get_value(13, 1) + "," + get_value(13, 0),
+           get_value(13, 2) + "," + get_value(13, 1) + "," + get_value(13, 0),
+           create_blob_iterator, check_is_blob(false));
+    verify(15, Status::kOk, get_value(16, 0), get_value(14, 0),
+           create_blob_iterator, check_is_blob(false));
+
+    for (auto* snapshot : snapshots) {
+      dbfull()->ReleaseSnapshot(snapshot);
+    }
+  }
+}
+
+}  // namespace rocksdb
+
+int main(int argc, char** argv) {
+  rocksdb::port::InstallStackTraceHandler();
+  ::testing::InitGoogleTest(&argc, argv);
+  return RUN_ALL_TESTS();
+}

db/db_block_cache_test.cc

@@ -497,7 +497,7 @@ TEST_F(DBBlockCacheTest, CompressedCache) {
         options.compression = kNoCompression;
         break;
       default:
-        ASSERT_TRUE(false);
+        FAIL();
     }
     CreateAndReopenWithCF({"pikachu"}, options);
     // default column family doesn't have block cache
@@ -560,7 +560,7 @@ TEST_F(DBBlockCacheTest, CompressedCache) {
         ASSERT_EQ(TestGetTickerCount(options, BLOCK_CACHE_COMPRESSED_HIT), 0);
         break;
       default:
-        ASSERT_TRUE(false);
+        FAIL();
     }
 
     options.create_if_missing = true;

db/db_compaction_filter_test.cc

@@ -24,6 +24,36 @@ class DBTestCompactionFilter : public DBTestBase {
   DBTestCompactionFilter() : DBTestBase("/db_compaction_filter_test") {}
 };
 
+// Param variant of DBTestBase::ChangeCompactOptions
+class DBTestCompactionFilterWithCompactParam
+    : public DBTestCompactionFilter,
+      public ::testing::WithParamInterface<DBTestBase::OptionConfig> {
+ public:
+  DBTestCompactionFilterWithCompactParam() : DBTestCompactionFilter() {
+    option_config_ = GetParam();
+    Destroy(last_options_);
+    auto options = CurrentOptions();
+    if (option_config_ == kDefault || option_config_ == kUniversalCompaction ||
+        option_config_ == kUniversalCompactionMultiLevel) {
+      options.create_if_missing = true;
+    }
+    if (option_config_ == kLevelSubcompactions ||
+        option_config_ == kUniversalSubcompactions) {
+      assert(options.max_subcompactions > 1);
+    }
+    TryReopen(options);
+  }
+};
+
+INSTANTIATE_TEST_CASE_P(
+    DBTestCompactionFilterWithCompactOption,
+    DBTestCompactionFilterWithCompactParam,
+    ::testing::Values(DBTestBase::OptionConfig::kDefault,
+                      DBTestBase::OptionConfig::kUniversalCompaction,
+                      DBTestBase::OptionConfig::kUniversalCompactionMultiLevel,
+                      DBTestBase::OptionConfig::kLevelSubcompactions,
+                      DBTestBase::OptionConfig::kUniversalSubcompactions));
+
 class KeepFilter : public CompactionFilter {
  public:
   virtual bool Filter(int level, const Slice& key, const Slice& value,
@@ -308,7 +338,6 @@ TEST_F(DBTestCompactionFilter, CompactionFilter) {
     ASSERT_OK(iter->status());
     while (iter->Valid()) {
       ParsedInternalKey ikey(Slice(), 0, kTypeValue);
-      ikey.sequence = -1;
       ASSERT_EQ(ParseInternalKey(iter->key(), &ikey), true);
       total++;
       if (ikey.sequence != 0) {
@@ -440,65 +469,63 @@ TEST_F(DBTestCompactionFilter, CompactionFilterDeletesAll) {
 }
 #endif  // ROCKSDB_LITE
 
-TEST_F(DBTestCompactionFilter, CompactionFilterWithValueChange) {
-  do {
-    Options options = CurrentOptions();
-    options.num_levels = 3;
-    options.compaction_filter_factory =
-      std::make_shared<ChangeFilterFactory>();
-    CreateAndReopenWithCF({"pikachu"}, options);
+TEST_P(DBTestCompactionFilterWithCompactParam,
+       CompactionFilterWithValueChange) {
+  Options options = CurrentOptions();
+  options.num_levels = 3;
+  options.compaction_filter_factory = std::make_shared<ChangeFilterFactory>();
+  CreateAndReopenWithCF({"pikachu"}, options);
 
-    // Write 100K+1 keys, these are written to a few files
-    // in L0. We do this so that the current snapshot points
-    // to the 100001 key.The compaction filter is  not invoked
-    // on keys that are visible via a snapshot because we
-    // anyways cannot delete it.
-    const std::string value(10, 'x');
-    for (int i = 0; i < 100001; i++) {
-      char key[100];
-      snprintf(key, sizeof(key), "B%010d", i);
-      Put(1, key, value);
-    }
+  // Write 100K+1 keys, these are written to a few files
+  // in L0. We do this so that the current snapshot points
+  // to the 100001 key.The compaction filter is  not invoked
+  // on keys that are visible via a snapshot because we
+  // anyways cannot delete it.
+  const std::string value(10, 'x');
+  for (int i = 0; i < 100001; i++) {
+    char key[100];
+    snprintf(key, sizeof(key), "B%010d", i);
+    Put(1, key, value);
+  }
 
-    // push all files to  lower levels
-    ASSERT_OK(Flush(1));
-    if (option_config_ != kUniversalCompactionMultiLevel &&
-        option_config_ != kUniversalSubcompactions) {
-      dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]);
-      dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]);
-    } else {
-      dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr,
-                             nullptr);
-    }
+  // push all files to  lower levels
+  ASSERT_OK(Flush(1));
+  if (option_config_ != kUniversalCompactionMultiLevel &&
+      option_config_ != kUniversalSubcompactions) {
+    dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]);
+    dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]);
+  } else {
+    dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr,
+                           nullptr);
+  }
 
-    // re-write all data again
-    for (int i = 0; i < 100001; i++) {
-      char key[100];
-      snprintf(key, sizeof(key), "B%010d", i);
-      Put(1, key, value);
-    }
+  // re-write all data again
+  for (int i = 0; i < 100001; i++) {
+    char key[100];
+    snprintf(key, sizeof(key), "B%010d", i);
+    Put(1, key, value);
+  }
 
-    // push all files to  lower levels. This should
-    // invoke the compaction filter for all 100000 keys.
-    ASSERT_OK(Flush(1));
-    if (option_config_ != kUniversalCompactionMultiLevel &&
-        option_config_ != kUniversalSubcompactions) {
-      dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]);
-      dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]);
-    } else {
-      dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr,
-                             nullptr);
-    }
+  // push all files to  lower levels. This should
+  // invoke the compaction filter for all 100000 keys.
+  ASSERT_OK(Flush(1));
+  if (option_config_ != kUniversalCompactionMultiLevel &&
+      option_config_ != kUniversalSubcompactions) {
+    dbfull()->TEST_CompactRange(0, nullptr, nullptr, handles_[1]);
+    dbfull()->TEST_CompactRange(1, nullptr, nullptr, handles_[1]);
+  } else {
+    dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr,
+                           nullptr);
+  }
 
-    // verify that all keys now have the new value that
-    // was set by the compaction process.
-    for (int i = 0; i < 100001; i++) {
-      char key[100];
-      snprintf(key, sizeof(key), "B%010d", i);
-      std::string newvalue = Get(1, key);
-      ASSERT_EQ(newvalue.compare(NEW_VALUE), 0);
-    }
-  } while (ChangeCompactOptions());
+  // verify that all keys now have the new value that
+  // was set by the compaction process.
+  for (int i = 0; i < 100001; i++) {
+    char key[100];
+    snprintf(key, sizeof(key), "B%010d", i);
+    std::string newvalue = Get(1, key);
+    ASSERT_EQ(newvalue.compare(NEW_VALUE), 0);
+  }
 }
 
 TEST_F(DBTestCompactionFilter, CompactionFilterWithMergeOperator) {
@@ -617,7 +644,6 @@ TEST_F(DBTestCompactionFilter, CompactionFilterContextManual) {
     ASSERT_OK(iter->status());
     while (iter->Valid()) {
       ParsedInternalKey ikey(Slice(), 0, kTypeValue);
-      ikey.sequence = -1;
       ASSERT_EQ(ParseInternalKey(iter->key(), &ikey), true);
       total++;
       if (ikey.sequence != 0) {

db/db_compaction_test.cc

@@ -218,6 +218,84 @@ TEST_P(DBCompactionTestWithParam, CompactionDeletionTrigger) {
   }
 }
 
+TEST_P(DBCompactionTestWithParam, CompactionsPreserveDeletes) {
+  //  For each options type we test following
+  //  - Enable preserve_deletes
+  //  - write bunch of keys and deletes
+  //  - Set start_seqnum to the beginning; compact; check that keys are present
+  //  - rewind start_seqnum way forward; compact; check that keys are gone
+
+  for (int tid = 0; tid < 3; ++tid) {
+    Options options = DeletionTriggerOptions(CurrentOptions());
+    options.max_subcompactions = max_subcompactions_;
+    options.preserve_deletes=true;
+    options.num_levels = 2;
+
+    if (tid == 1) {
+      options.skip_stats_update_on_db_open = true;
+    } else if (tid == 2) {
+      // third pass with universal compaction
+      options.compaction_style = kCompactionStyleUniversal;
+    }
+
+    DestroyAndReopen(options);
+    Random rnd(301);
+    // highlight the default; all deletes should be preserved
+    SetPreserveDeletesSequenceNumber(0);
+
+    const int kTestSize = kCDTKeysPerBuffer;
+    std::vector<std::string> values;
+    for (int k = 0; k < kTestSize; ++k) {
+      values.push_back(RandomString(&rnd, kCDTValueSize));
+      ASSERT_OK(Put(Key(k), values[k]));
+    }
+
+    for (int k = 0; k < kTestSize; ++k) {
+      ASSERT_OK(Delete(Key(k)));
+    }
+    // to ensure we tackle all tombstones
+    CompactRangeOptions cro;
+    cro.change_level = true;
+    cro.target_level = 2;
+    cro.bottommost_level_compaction = BottommostLevelCompaction::kForce;
+
+    dbfull()->TEST_WaitForFlushMemTable();
+    dbfull()->CompactRange(cro, nullptr, nullptr);
+
+    // check that normal user iterator doesn't see anything
+    Iterator* db_iter = dbfull()->NewIterator(ReadOptions());
+    int i = 0;
+    for (db_iter->SeekToFirst(); db_iter->Valid(); db_iter->Next()) {
+      i++;
+    }
+    ASSERT_EQ(i, 0);
+    delete db_iter;
+
+    // check that iterator that sees internal keys sees tombstones
+    ReadOptions ro;
+    ro.iter_start_seqnum=1;
+    db_iter = dbfull()->NewIterator(ro);
+    i = 0;
+    for (db_iter->SeekToFirst(); db_iter->Valid(); db_iter->Next()) {
+      i++;
+    }
+    ASSERT_EQ(i, 4);
+    delete db_iter;
+
+    // now all deletes should be gone
+    SetPreserveDeletesSequenceNumber(100000000);
+    dbfull()->CompactRange(cro, nullptr, nullptr);
+
+    db_iter = dbfull()->NewIterator(ro);
+    i = 0;
+    for (db_iter->SeekToFirst(); db_iter->Valid(); db_iter->Next()) {
+      i++;
+    }
+    ASSERT_EQ(i, 0);
+    delete db_iter;
+  }
+}
+
 TEST_F(DBCompactionTest, SkipStatsUpdateTest) {
   // This test verify UpdateAccumulatedStats is not on
   // if options.skip_stats_update_on_db_open = true
@@ -1439,6 +1517,60 @@ TEST_F(DBCompactionTest, DeleteFileRange) {
   ASSERT_GT(old_num_files, new_num_files);
 }
 
+TEST_F(DBCompactionTest, DeleteFileRangeFileEndpointsOverlapBug) {
+  // regression test for #2833: groups of files whose user-keys overlap at the
+  // endpoints could be split by `DeleteFilesInRange`. This caused old data to
+  // reappear, either because a new version of the key was removed, or a range
+  // deletion was partially dropped. It could also cause non-overlapping
+  // invariant to be violated if the files dropped by DeleteFilesInRange were
+  // a subset of files that a range deletion spans.
+  const int kNumL0Files = 2;
+  const int kValSize = 8 << 10;  // 8KB
+  Options options = CurrentOptions();
+  options.level0_file_num_compaction_trigger = kNumL0Files;
+  options.target_file_size_base = 1 << 10;  // 1KB
+  DestroyAndReopen(options);
+
+  // The snapshot prevents key 1 from having its old version dropped. The low
+  // `target_file_size_base` ensures two keys will be in each output file.
+  const Snapshot* snapshot = nullptr;
+  Random rnd(301);
+  // The value indicates which flush the key belonged to, which is enough
+  // for us to determine the keys' relative ages. After L0 flushes finish,
+  // files look like:
+  //
+  // File 0: 0 -> vals[0], 1 -> vals[0]
+  // File 1:               1 -> vals[1], 2 -> vals[1]
+  //
+  // Then L0->L1 compaction happens, which outputs keys as follows:
+  //
+  // File 0: 0 -> vals[0], 1 -> vals[1]
+  // File 1:               1 -> vals[0], 2 -> vals[1]
+  //
+  // DeleteFilesInRange shouldn't be allowed to drop just file 0, as that
+  // would cause `1 -> vals[0]` (an older key) to reappear.
+  std::string vals[kNumL0Files];
+  for (int i = 0; i < kNumL0Files; ++i) {
+    vals[i] = RandomString(&rnd, kValSize);
+    Put(Key(i), vals[i]);
+    Put(Key(i + 1), vals[i]);
+    Flush();
+    if (i == 0) {
+      snapshot = db_->GetSnapshot();
+    }
+  }
+  dbfull()->TEST_WaitForCompact();
+
+  // Verify `DeleteFilesInRange` can't drop only file 0 which would cause
+  // "1 -> vals[0]" to reappear.
+  std::string begin_str = Key(0), end_str = Key(1);
+  Slice begin = begin_str, end = end_str;
+  ASSERT_OK(DeleteFilesInRange(db_, db_->DefaultColumnFamily(), &begin, &end));
+  ASSERT_EQ(vals[1], Get(Key(1)));
+
+  db_->ReleaseSnapshot(snapshot);
+}
+
 TEST_P(DBCompactionTestWithParam, TrivialMoveToLastLevelWithFiles) {
   int32_t trivial_move = 0;
   int32_t non_trivial_move = 0;
@@ -2684,6 +2816,179 @@ TEST_P(DBCompactionTestWithParam, IntraL0CompactionDoesNotObsoleteDeletions) {
   ASSERT_TRUE(db_->Get(roptions, Key(0), &result).IsNotFound());
 }
 
+TEST_F(DBCompactionTest, OptimizedDeletionObsoleting) {
+  // Deletions can be dropped when compacted to non-last level if they fall
+  // outside the lower-level files' key-ranges.
+  const int kNumL0Files = 4;
+  Options options = CurrentOptions();
+  options.level0_file_num_compaction_trigger = kNumL0Files;
+  options.statistics = rocksdb::CreateDBStatistics();
+  DestroyAndReopen(options);
+
+  // put key 1 and 3 in separate L1, L2 files.
+  // So key 0, 2, and 4+ fall outside these levels' key-ranges.
+  for (int level = 2; level >= 1; --level) {
+    for (int i = 0; i < 2; ++i) {
+      Put(Key(2 * i + 1), "val");
+      Flush();
+    }
+    MoveFilesToLevel(level);
+    ASSERT_EQ(2, NumTableFilesAtLevel(level));
+  }
+
+  // Delete keys in range [1, 4]. These L0 files will be compacted with L1:
+  // - Tombstones for keys 2 and 4 can be dropped early.
+  // - Tombstones for keys 1 and 3 must be kept due to L2 files' key-ranges.
+  for (int i = 0; i < kNumL0Files; ++i) {
+    Put(Key(0), "val");  // sentinel to prevent trivial move
+    Delete(Key(i + 1));
+    Flush();
+  }
+  dbfull()->TEST_WaitForCompact();
+
+  for (int i = 0; i < kNumL0Files; ++i) {
+    std::string value;
+    ASSERT_TRUE(db_->Get(ReadOptions(), Key(i + 1), &value).IsNotFound());
+  }
+  ASSERT_EQ(2, options.statistics->getTickerCount(
+                   COMPACTION_OPTIMIZED_DEL_DROP_OBSOLETE));
+  ASSERT_EQ(2,
+            options.statistics->getTickerCount(COMPACTION_KEY_DROP_OBSOLETE));
+}
+
+TEST_F(DBCompactionTest, CompactFilesPendingL0Bug) {
+  // https://www.facebook.com/groups/rocksdb.dev/permalink/1389452781153232/
+  // CompactFiles() had a bug where it failed to pick a compaction when an L0
+  // compaction existed, but marked it as scheduled anyways. It'd never be
+  // unmarked as scheduled, so future compactions or DB close could hang.
+  const int kNumL0Files = 5;
+  Options options = CurrentOptions();
+  options.level0_file_num_compaction_trigger = kNumL0Files - 1;
+  options.max_background_compactions = 2;
+  DestroyAndReopen(options);
+
+  rocksdb::SyncPoint::GetInstance()->LoadDependency(
+      {{"LevelCompactionPicker::PickCompaction:Return",
+        "DBCompactionTest::CompactFilesPendingL0Bug:Picked"},
+       {"DBCompactionTest::CompactFilesPendingL0Bug:ManualCompacted",
+        "DBImpl::BackgroundCompaction:NonTrivial:AfterRun"}});
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+  auto schedule_multi_compaction_token =
+      dbfull()->TEST_write_controler().GetCompactionPressureToken();
+
+  // Files 0-3 will be included in an L0->L1 compaction.
+  //
+  // File 4 will be included in a call to CompactFiles() while the first
+  // compaction is running.
+  for (int i = 0; i < kNumL0Files - 1; ++i) {
+    ASSERT_OK(Put(Key(0), "val"));  // sentinel to prevent trivial move
+    ASSERT_OK(Put(Key(i + 1), "val"));
+    ASSERT_OK(Flush());
+  }
+  TEST_SYNC_POINT("DBCompactionTest::CompactFilesPendingL0Bug:Picked");
+  // file 4 flushed after 0-3 picked
+  ASSERT_OK(Put(Key(kNumL0Files), "val"));
+  ASSERT_OK(Flush());
+
+  // previously DB close would hang forever as this situation caused scheduled
+  // compactions count to never decrement to zero.
+  ColumnFamilyMetaData cf_meta;
+  dbfull()->GetColumnFamilyMetaData(dbfull()->DefaultColumnFamily(), &cf_meta);
+  ASSERT_EQ(kNumL0Files, cf_meta.levels[0].files.size());
+  std::vector<std::string> input_filenames;
+  input_filenames.push_back(cf_meta.levels[0].files.front().name);
+  ASSERT_OK(dbfull()
+                  ->CompactFiles(CompactionOptions(), input_filenames,
+                                 0 /* output_level */));
+  TEST_SYNC_POINT("DBCompactionTest::CompactFilesPendingL0Bug:ManualCompacted");
+  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+}
+
+TEST_F(DBCompactionTest, CompactFilesOverlapInL0Bug) {
+  // Regression test for bug of not pulling in L0 files that overlap the user-
+  // specified input files in time- and key-ranges.
+  Put(Key(0), "old_val");
+  Flush();
+  Put(Key(0), "new_val");
+  Flush();
+
+  ColumnFamilyMetaData cf_meta;
+  dbfull()->GetColumnFamilyMetaData(dbfull()->DefaultColumnFamily(), &cf_meta);
+  ASSERT_GE(cf_meta.levels.size(), 2);
+  ASSERT_EQ(2, cf_meta.levels[0].files.size());
+
+  // Compacting {new L0 file, L1 file} should pull in the old L0 file since it
+  // overlaps in key-range and time-range.
+  std::vector<std::string> input_filenames;
+  input_filenames.push_back(cf_meta.levels[0].files.front().name);
+  ASSERT_OK(dbfull()->CompactFiles(CompactionOptions(), input_filenames,
+                                   1 /* output_level */));
+  ASSERT_EQ("new_val", Get(Key(0)));
+}
+
+TEST_F(DBCompactionTest, CompactBottomLevelFilesWithDeletions) {
+  // bottom-level files may contain deletions due to snapshots protecting the
+  // deleted keys. Once the snapshot is released, we should see files with many
+  // such deletions undergo single-file compactions.
+  const int kNumKeysPerFile = 1024;
+  const int kNumLevelFiles = 4;
+  const int kValueSize = 128;
+  Options options = CurrentOptions();
+  options.compression = kNoCompression;
+  options.level0_file_num_compaction_trigger = kNumLevelFiles;
+  // inflate it a bit to account for key/metadata overhead
+  options.target_file_size_base = 120 * kNumKeysPerFile * kValueSize / 100;
+  Reopen(options);
+
+  Random rnd(301);
+  const Snapshot* snapshot = nullptr;
+  for (int i = 0; i < kNumLevelFiles; ++i) {
+    for (int j = 0; j < kNumKeysPerFile; ++j) {
+      ASSERT_OK(
+          Put(Key(i * kNumKeysPerFile + j), RandomString(&rnd, kValueSize)));
+    }
+    if (i == kNumLevelFiles - 1) {
+      snapshot = db_->GetSnapshot();
+      // delete every other key after grabbing a snapshot, so these deletions
+      // and the keys they cover can't be dropped until after the snapshot is
+      // released.
+      for (int j = 0; j < kNumLevelFiles * kNumKeysPerFile; j += 2) {
+        ASSERT_OK(Delete(Key(j)));
+      }
+    }
+    Flush();
+    if (i < kNumLevelFiles - 1) {
+      ASSERT_EQ(i + 1, NumTableFilesAtLevel(0));
+    }
+  }
+  dbfull()->TEST_WaitForCompact();
+  ASSERT_EQ(kNumLevelFiles, NumTableFilesAtLevel(1));
+
+  std::vector<LiveFileMetaData> pre_release_metadata, post_release_metadata;
+  db_->GetLiveFilesMetaData(&pre_release_metadata);
+  // just need to bump seqnum so ReleaseSnapshot knows the newest key in the SST
+  // files does not need to be preserved in case of a future snapshot.
+  ASSERT_OK(Put(Key(0), "val"));
+  // release snapshot and wait for compactions to finish. Single-file
+  // compactions should be triggered, which reduce the size of each bottom-level
+  // file without changing file count.
+  db_->ReleaseSnapshot(snapshot);
+  dbfull()->TEST_WaitForCompact();
+  db_->GetLiveFilesMetaData(&post_release_metadata);
+  ASSERT_EQ(pre_release_metadata.size(), post_release_metadata.size());
+
+  for (size_t i = 0; i < pre_release_metadata.size(); ++i) {
+    const auto& pre_file = pre_release_metadata[i];
+    const auto& post_file = post_release_metadata[i];
+    ASSERT_EQ(1, pre_file.level);
+    ASSERT_EQ(1, post_file.level);
+    // each file is smaller than it was before as it was rewritten without
+    // deletion markers/deleted keys.
+    ASSERT_LT(post_file.size, pre_file.size);
+  }
+}
+
 INSTANTIATE_TEST_CASE_P(DBCompactionTestWithParam, DBCompactionTestWithParam,
                         ::testing::Values(std::make_tuple(1, true),
                                           std::make_tuple(1, false),

db/db_encryption_test.cc

@@ -1,9 +1,7 @@
 //  Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
-//  This source code is licensed under the BSD-style license found in the
-//  LICENSE file in the root directory of this source tree. An additional grant
-//  of patent rights can be found in the PATENTS file in the same directory.
-//  This source code is also licensed under the GPLv2 license found in the
-//  COPYING file in the root directory of this source tree.
+//  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/db_test_util.h"
 #include "port/stack_trace.h"

db/db_filesnapshot.cc

@@ -57,6 +57,7 @@ Status DBImpl::EnableFileDeletions(bool force) {
       ROCKS_LOG_INFO(immutable_db_options_.info_log, "File Deletions Enabled");
       should_purge_files = true;
       FindObsoleteFiles(&job_context, true);
+      bg_cv_.SignalAll();
     } else {
       ROCKS_LOG_WARN(
           immutable_db_options_.info_log,
@@ -141,6 +142,18 @@ Status DBImpl::GetLiveFiles(std::vector<std::string>& ret,
 }
 
 Status DBImpl::GetSortedWalFiles(VectorLogPtr& files) {
+  {
+    // If caller disabled deletions, this function should return files that are
+    // guaranteed not to be deleted until deletions are re-enabled. We need to
+    // wait for pending purges to finish since WalManager doesn't know which
+    // files are going to be purged. Additional purges won't be scheduled as
+    // long as deletions are disabled (so the below loop must terminate).
+    InstrumentedMutexLock l(&mutex_);
+    while (disable_delete_obsolete_files_ > 0 &&
+           pending_purge_obsolete_files_ > 0) {
+      bg_cv_.Wait();
+    }
+  }
   return wal_manager_.GetSortedWalFiles(files);
 }
 

db/db_flush_test.cc

@@ -126,6 +126,41 @@ TEST_F(DBFlushTest, FlushInLowPriThreadPool) {
   ASSERT_EQ(1, num_compactions);
 }
 
+TEST_F(DBFlushTest, ManualFlushWithMinWriteBufferNumberToMerge) {
+  Options options = CurrentOptions();
+  options.write_buffer_size = 100;
+  options.max_write_buffer_number = 4;
+  options.min_write_buffer_number_to_merge = 3;
+  Reopen(options);
+
+  SyncPoint::GetInstance()->LoadDependency(
+      {{"DBImpl::BGWorkFlush",
+        "DBFlushTest::ManualFlushWithMinWriteBufferNumberToMerge:1"},
+       {"DBFlushTest::ManualFlushWithMinWriteBufferNumberToMerge:2",
+        "DBImpl::FlushMemTableToOutputFile:BeforeInstallSV"}});
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  ASSERT_OK(Put("key1", "value1"));
+
+  port::Thread t([&]() {
+    // The call wait for flush to finish, i.e. with flush_options.wait = true.
+    ASSERT_OK(Flush());
+  });
+
+  // Wait for flush start.
+  TEST_SYNC_POINT("DBFlushTest::ManualFlushWithMinWriteBufferNumberToMerge:1");
+  // Insert a second memtable before the manual flush finish.
+  // At the end of the manual flush job, it will check if further flush
+  // is needed, but it will not trigger flush of the second memtable because
+  // min_write_buffer_number_to_merge is not reached.
+  ASSERT_OK(Put("key2", "value2"));
+  ASSERT_OK(dbfull()->TEST_SwitchMemtable());
+  TEST_SYNC_POINT("DBFlushTest::ManualFlushWithMinWriteBufferNumberToMerge:2");
+
+  // Manual flush should return, without waiting for flush indefinitely.
+  t.join();
+}
+
 TEST_P(DBFlushDirectIOTest, DirectIO) {
   Options options;
   options.create_if_missing = true;

db/db_impl.h

@@ -28,6 +28,9 @@
 #include "db/flush_scheduler.h"
 #include "db/internal_stats.h"
 #include "db/log_writer.h"
+#include "db/pre_release_callback.h"
+#include "db/read_callback.h"
+#include "db/snapshot_checker.h"
 #include "db/snapshot_impl.h"
 #include "db/version_edit.h"
 #include "db/wal_manager.h"
@@ -52,12 +55,13 @@
 
 namespace rocksdb {
 
+class Arena;
+class ArenaWrappedDBIter;
 class MemTable;
 class TableCache;
 class Version;
 class VersionEdit;
 class VersionSet;
-class Arena;
 class WriteCallback;
 struct JobContext;
 struct ExternalSstFileInfo;
@@ -65,7 +69,8 @@ struct MemTableInfo;
 
 class DBImpl : public DB {
  public:
-  DBImpl(const DBOptions& options, const std::string& dbname);
+  DBImpl(const DBOptions& options, const std::string& dbname,
+         const bool seq_per_batch = false);
   virtual ~DBImpl();
 
   // Implementations of the DB interface
@@ -93,6 +98,14 @@ class DBImpl : public DB {
   virtual Status Get(const ReadOptions& options,
                      ColumnFamilyHandle* column_family, const Slice& key,
                      PinnableSlice* value) override;
+
+  // Function that Get and KeyMayExist call with no_io true or false
+  // Note: 'value_found' from KeyMayExist propagates here
+  Status GetImpl(const ReadOptions& options, ColumnFamilyHandle* column_family,
+                 const Slice& key, PinnableSlice* value,
+                 bool* value_found = nullptr, ReadCallback* callback = nullptr,
+                 bool* is_blob_index = nullptr);
+
   using DB::MultiGet;
   virtual std::vector<Status> MultiGet(
       const ReadOptions& options,
@@ -123,6 +136,7 @@ class DBImpl : public DB {
                            ColumnFamilyHandle* column_family, const Slice& key,
                            std::string* value,
                            bool* value_found = nullptr) override;
+
   using DB::NewIterator;
   virtual Iterator* NewIterator(const ReadOptions& options,
                                 ColumnFamilyHandle* column_family) override;
@@ -130,15 +144,22 @@ class DBImpl : public DB {
       const ReadOptions& options,
       const std::vector<ColumnFamilyHandle*>& column_families,
       std::vector<Iterator*>* iterators) override;
+  ArenaWrappedDBIter* NewIteratorImpl(const ReadOptions& options,
+                                      ColumnFamilyData* cfd,
+                                      SequenceNumber snapshot,
+                                      ReadCallback* read_callback,
+                                      bool allow_blob = false,
+                                      bool allow_refresh = true);
+
   virtual const Snapshot* GetSnapshot() override;
   virtual void ReleaseSnapshot(const Snapshot* snapshot) override;
   using DB::GetProperty;
   virtual bool GetProperty(ColumnFamilyHandle* column_family,
                            const Slice& property, std::string* value) override;
   using DB::GetMapProperty;
-  virtual bool GetMapProperty(ColumnFamilyHandle* column_family,
-                              const Slice& property,
-                              std::map<std::string, double>* value) override;
+  virtual bool GetMapProperty(
+      ColumnFamilyHandle* column_family, const Slice& property,
+      std::map<std::string, std::string>* value) override;
   using DB::GetIntProperty;
   virtual bool GetIntProperty(ColumnFamilyHandle* column_family,
                               const Slice& property, uint64_t* value) override;
@@ -201,8 +222,13 @@ class DBImpl : public DB {
   virtual Status SyncWAL() override;
 
   virtual SequenceNumber GetLatestSequenceNumber() const override;
+  virtual void SetLastPublishedSequence(SequenceNumber seq);
+  // Returns LastSequence in last_seq_same_as_publish_seq_
+  // mode and LastAllocatedSequence otherwise. This is useful when visiblility
+  // depends also on data written to the WAL but not to the memtable.
+  SequenceNumber TEST_GetLastVisibleSequence() const;
 
-  bool HasActiveSnapshotLaterThanSN(SequenceNumber sn);
+  virtual bool SetPreserveDeletesSequenceNumber(SequenceNumber seqnum) override;
 
 #ifndef ROCKSDB_LITE
   using DB::ResetStats;
@@ -235,11 +261,11 @@ class DBImpl : public DB {
       ColumnFamilyHandle* column_family,
       ColumnFamilyMetaData* metadata) override;
 
-  // experimental API
   Status SuggestCompactRange(ColumnFamilyHandle* column_family,
-                             const Slice* begin, const Slice* end);
+                             const Slice* begin, const Slice* end) override;
 
-  Status PromoteL0(ColumnFamilyHandle* column_family, int target_level);
+  Status PromoteL0(ColumnFamilyHandle* column_family,
+                   int target_level) override;
 
   // Similar to Write() but will call the callback once on the single write
   // thread to determine whether it is safe to perform the write.
@@ -285,7 +311,8 @@ class DBImpl : public DB {
   // TODO(andrewkr): this API need to be aware of range deletion operations
   Status GetLatestSequenceForKey(SuperVersion* sv, const Slice& key,
                                  bool cache_only, SequenceNumber* seq,
-                                 bool* found_record_for_key);
+                                 bool* found_record_for_key,
+                                 bool* is_blob_index = nullptr);
 
   using DB::IngestExternalFile;
   virtual Status IngestExternalFile(
@@ -293,6 +320,8 @@ class DBImpl : public DB {
       const std::vector<std::string>& external_files,
       const IngestExternalFileOptions& ingestion_options) override;
 
+  virtual Status VerifyChecksum() override;
+
 #endif  // ROCKSDB_LITE
 
   // Similar to GetSnapshot(), but also lets the db know that this snapshot
@@ -329,7 +358,7 @@ class DBImpl : public DB {
                            ColumnFamilyHandle* column_family = nullptr,
                            bool disallow_trivial_move = false);
 
-  void TEST_HandleWALFull();
+  void TEST_SwitchWAL();
 
   bool TEST_UnableToFlushOldestLog() {
     return unable_to_flush_oldest_log_;
@@ -339,6 +368,8 @@ class DBImpl : public DB {
     return alive_log_files_.begin()->getting_flushed;
   }
 
+  Status TEST_SwitchMemtable(ColumnFamilyData* cfd = nullptr);
+
   // Force current memtable contents to be flushed.
   Status TEST_FlushMemTable(bool wait = true,
                             ColumnFamilyHandle* cfh = nullptr);
@@ -357,6 +388,9 @@ class DBImpl : public DB {
   // Return the current manifest file no.
   uint64_t TEST_Current_Manifest_FileNo();
 
+  // Returns the number that'll be assigned to the next file that's created.
+  uint64_t TEST_Current_Next_FileNo();
+
   // get total level0 file size. Only for testing.
   uint64_t TEST_GetLevel0TotalSize();
 
@@ -434,6 +468,8 @@ class DBImpl : public DB {
   // belong to live files are posibly removed. Also, removes all the
   // files in sst_delete_files and log_delete_files.
   // It is not necessary to hold the mutex when invoking this method.
+  // If FindObsoleteFiles() was run, we need to also run
+  // PurgeObsoleteFiles(), even if disable_delete_obsolete_files_ is true
   void PurgeObsoleteFiles(const JobContext& background_contet,
                           bool schedule_only = false);
 
@@ -460,6 +496,9 @@ class DBImpl : public DB {
   // mutex is held.
   SuperVersion* GetAndRefSuperVersion(uint32_t column_family_id);
 
+  // Un-reference the super version and clean it up if it is the last reference.
+  void CleanupSuperVersion(SuperVersion* sv);
+
   // Un-reference the super version and return it to thread local cache if
   // needed. If it is the last reference of the super version. Clean it up
   // after un-referencing it.
@@ -494,6 +533,12 @@ class DBImpl : public DB {
 
   const WriteController& write_controller() { return write_controller_; }
 
+  InternalIterator* NewInternalIterator(const ReadOptions&,
+                                        ColumnFamilyData* cfd,
+                                        SuperVersion* super_version,
+                                        Arena* arena,
+                                        RangeDelAggregator* range_del_agg);
+
   // hollow transactions shell used for recovery.
   // these will then be passed to TransactionDB so that
   // locks can be reacquired before writing can resume.
@@ -501,9 +546,11 @@ class DBImpl : public DB {
     uint64_t log_number_;
     std::string name_;
     WriteBatch* batch_;
+    // The seq number of the first key in the batch
+    SequenceNumber seq_;
     explicit RecoveredTransaction(const uint64_t log, const std::string& name,
-                                  WriteBatch* batch)
-        : log_number_(log), name_(name), batch_(batch) {}
+                                  WriteBatch* batch, SequenceNumber seq)
+        : log_number_(log), name_(name), batch_(batch), seq_(seq) {}
 
     ~RecoveredTransaction() { delete batch_; }
   };
@@ -525,8 +572,9 @@ class DBImpl : public DB {
   }
 
   void InsertRecoveredTransaction(const uint64_t log, const std::string& name,
-                                  WriteBatch* batch) {
-    recovered_transactions_[name] = new RecoveredTransaction(log, name, batch);
+                                  WriteBatch* batch, SequenceNumber seq) {
+    recovered_transactions_[name] =
+        new RecoveredTransaction(log, name, batch, seq);
     MarkLogAsContainingPrepSection(log);
   }
 
@@ -553,12 +601,26 @@ class DBImpl : public DB {
     logs_to_free_queue_.push_back(log_writer);
   }
 
+  void SetSnapshotChecker(SnapshotChecker* snapshot_checker);
+
+  InstrumentedMutex* mutex() { return &mutex_; }
+
   Status NewDB();
 
+  // This is to be used only by internal rocksdb classes.
+  static Status Open(const DBOptions& db_options, const std::string& name,
+                     const std::vector<ColumnFamilyDescriptor>& column_families,
+                     std::vector<ColumnFamilyHandle*>* handles, DB** dbptr,
+                     const bool seq_per_batch);
+
+  virtual Status Close() override;
+
  protected:
   Env* const env_;
   const std::string dbname_;
   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_;
   const ImmutableDBOptions immutable_db_options_;
   MutableDBOptions mutable_db_options_;
@@ -566,12 +628,6 @@ class DBImpl : public DB {
   std::unordered_map<std::string, RecoveredTransaction*>
       recovered_transactions_;
 
-  InternalIterator* NewInternalIterator(const ReadOptions&,
-                                        ColumnFamilyData* cfd,
-                                        SuperVersion* super_version,
-                                        Arena* arena,
-                                        RangeDelAggregator* range_del_agg);
-
   // Except in DB::Open(), WriteOptionsFile can only be called when:
   // Persist options to options file.
   // If need_mutex_lock = false, the method will lock DB mutex.
@@ -610,51 +666,73 @@ class DBImpl : public DB {
 
   void EraseThreadStatusDbInfo() const;
 
+  // If disable_memtable is set the application logic must guarantee that the
+  // batch will still be skipped from memtable during the recovery. In
+  // WriteCommitted it is guarnateed since disable_memtable is used for prepare
+  // batch which will be written to memtable later during the commit, and in
+  // WritePrepared it is guaranteed since it will be used only for WAL markers
+  // which will never be written to memtable.
   Status WriteImpl(const WriteOptions& options, WriteBatch* updates,
                    WriteCallback* callback = nullptr,
                    uint64_t* log_used = nullptr, uint64_t log_ref = 0,
-                   bool disable_memtable = false);
+                   bool disable_memtable = false, uint64_t* seq_used = nullptr,
+                   PreReleaseCallback* pre_release_callback = nullptr);
 
   Status PipelinedWriteImpl(const WriteOptions& options, WriteBatch* updates,
                             WriteCallback* callback = nullptr,
                             uint64_t* log_used = nullptr, uint64_t log_ref = 0,
-                            bool disable_memtable = false);
+                            bool disable_memtable = false,
+                            uint64_t* seq_used = nullptr);
 
   Status WriteImplWALOnly(const WriteOptions& options, WriteBatch* updates,
                           WriteCallback* callback = nullptr,
-                          uint64_t* log_used = nullptr, uint64_t log_ref = 0);
+                          uint64_t* log_used = nullptr, uint64_t log_ref = 0,
+                          uint64_t* seq_used = nullptr,
+                          PreReleaseCallback* pre_release_callback = nullptr);
 
   uint64_t FindMinLogContainingOutstandingPrep();
   uint64_t FindMinPrepLogReferencedByMemTable();
+  // write cached_recoverable_state_ to memtable if it is not empty
+  // The writer must be the leader in write_thread_ and holding mutex_
+  Status WriteRecoverableState();
 
  private:
   friend class DB;
   friend class InternalStats;
-  friend class TransactionImpl;
+  friend class PessimisticTransaction;
+  friend class WriteCommittedTxn;
+  friend class WritePreparedTxn;
+  friend class WritePreparedTxnDB;
+  friend class WriteBatchWithIndex;
 #ifndef ROCKSDB_LITE
   friend class ForwardIterator;
 #endif
   friend struct SuperVersion;
   friend class CompactedDBImpl;
 #ifndef NDEBUG
+  friend class DBTest2_ReadCallbackTest_Test;
+  friend class WriteCallbackTest_WriteWithCallbackTest_Test;
   friend class XFTransactionWriteHandler;
+  friend class DBBlobIndexTest;
 #endif
   struct CompactionState;
 
   struct WriteContext {
-    autovector<SuperVersion*> superversions_to_free_;
+    SuperVersionContext superversion_context;
     autovector<MemTable*> memtables_to_free_;
 
+    explicit WriteContext(bool create_superversion = false)
+        : superversion_context(create_superversion) {}
+
     ~WriteContext() {
-      for (auto& sv : superversions_to_free_) {
-        delete sv;
-      }
+      superversion_context.Clean();
       for (auto& m : memtables_to_free_) {
         delete m;
       }
     }
   };
 
+  struct PrepickedCompaction;
   struct PurgeFileInfo;
 
   // Recover the descriptor from persistent storage.  May do a significant
@@ -677,9 +755,8 @@ class DBImpl : public DB {
   // Delete any unneeded files and stale in-memory entries.
   void DeleteObsoleteFiles();
   // Delete obsolete files and log status and information of file deletion
-  void DeleteObsoleteFileImpl(Status file_deletion_status, int job_id,
-                              const std::string& fname, FileType type,
-                              uint64_t number, uint32_t path_id);
+  void DeleteObsoleteFileImpl(int job_id, const std::string& fname,
+                              FileType type, uint64_t number, uint32_t path_id);
 
   // Background process needs to call
   //     auto x = CaptureCurrentFileNumberInPendingOutputs()
@@ -736,11 +813,15 @@ class DBImpl : public DB {
   Status FlushMemTable(ColumnFamilyData* cfd, const FlushOptions& options,
                        bool writes_stopped = false);
 
-  // Wait for memtable flushed
-  Status WaitForFlushMemTable(ColumnFamilyData* cfd);
+  // Wait for memtable flushed.
+  // If flush_memtable_id is non-null, wait until the memtable with the ID
+  // gets flush. Otherwise, wait until the column family don't have any
+  // memtable pending flush.
+  Status WaitForFlushMemTable(ColumnFamilyData* cfd,
+                              const uint64_t* flush_memtable_id = nullptr);
 
   // REQUIRES: mutex locked
-  Status HandleWALFull(WriteContext* write_context);
+  Status SwitchWAL(WriteContext* write_context);
 
   // REQUIRES: mutex locked
   Status HandleWriteBufferFull(WriteContext* write_context);
@@ -750,7 +831,8 @@ class DBImpl : public DB {
                          WriteContext* write_context);
 
   WriteBatch* MergeBatch(const WriteThread::WriteGroup& write_group,
-                         WriteBatch* tmp_batch, size_t* write_with_wal);
+                         WriteBatch* tmp_batch, size_t* write_with_wal,
+                         WriteBatch** to_be_cached_state);
 
   Status WriteToWAL(const WriteBatch& merged_batch, log::Writer* log_writer,
                     uint64_t* log_used, uint64_t* log_size);
@@ -762,7 +844,7 @@ class DBImpl : public DB {
 
   Status ConcurrentWriteToWAL(const WriteThread::WriteGroup& write_group,
                               uint64_t* log_used, SequenceNumber* last_sequence,
-                              int total_count);
+                              size_t seq_inc);
 
   // Used by WriteImpl to update bg_error_ if paranoid check is enabled.
   void WriteCallbackStatusCheck(const Status& status);
@@ -796,14 +878,19 @@ class DBImpl : public DB {
   void SchedulePendingPurge(std::string fname, FileType type, uint64_t number,
                             uint32_t path_id, int job_id);
   static void BGWorkCompaction(void* arg);
+  // Runs a pre-chosen universal compaction involving bottom level in a
+  // separate, bottom-pri thread pool.
+  static void BGWorkBottomCompaction(void* arg);
   static void BGWorkFlush(void* db);
   static void BGWorkPurge(void* arg);
   static void UnscheduleCallback(void* arg);
-  void BackgroundCallCompaction(void* arg);
+  void BackgroundCallCompaction(PrepickedCompaction* prepicked_compaction,
+                                Env::Priority bg_thread_pri);
   void BackgroundCallFlush();
   void BackgroundCallPurge();
   Status BackgroundCompaction(bool* madeProgress, JobContext* job_context,
-                              LogBuffer* log_buffer, void* m = 0);
+                              LogBuffer* log_buffer,
+                              PrepickedCompaction* prepicked_compaction);
   Status BackgroundFlush(bool* madeProgress, JobContext* job_context,
                          LogBuffer* log_buffer);
 
@@ -835,6 +922,9 @@ class DBImpl : public DB {
 
   uint64_t GetMaxTotalWalSize() const;
 
+  // Actual implementation of Close()
+  virtual Status CloseImpl();
+
   // table_cache_ provides its own synchronization
   std::shared_ptr<Cache> table_cache_;
 
@@ -842,12 +932,12 @@ class DBImpl : public DB {
   FileLock* db_lock_;
 
   // In addition to mutex_, log_write_mutex_ protected writes to logs_ and
-  // logfile_number_. With concurrent_prepare it also protects alive_log_files_,
+  // logfile_number_. With two_write_queues it also protects alive_log_files_,
   // and log_empty_. Refer to the definition of each variable below for more
   // details.
   InstrumentedMutex log_write_mutex_;
   // State below is protected by mutex_
-  // With concurrent_prepare enabled, some of the variables that accessed during
+  // With two_write_queues enabled, some of the variables that accessed during
   // WriteToWAL need different synchronization: log_empty_, alive_log_files_,
   // logs_, logfile_number_. Refer to the definition of each variable below for
   // more description.
@@ -863,6 +953,8 @@ class DBImpl : public DB {
   // (i.e. whenever a flush is done, even if it didn't make any progress)
   // * whenever there is an error in background purge, flush or compaction
   // * whenever num_running_ingest_file_ goes to 0.
+  // * whenever pending_purge_obsolete_files_ goes to 0.
+  // * whenever disable_delete_obsolete_files_ goes to 0.
   InstrumentedCondVar bg_cv_;
   // Writes are protected by locking both mutex_ and log_write_mutex_, and reads
   // must be under either mutex_ or log_write_mutex_. Since after ::Open,
@@ -870,12 +962,12 @@ class DBImpl : public DB {
   // from the same write_thread_ without any locks.
   uint64_t logfile_number_;
   std::deque<uint64_t>
-      log_recycle_files;  // a list of log files that we can recycle
+      log_recycle_files_;  // a list of log files that we can recycle
   bool log_dir_synced_;
-  // Without concurrent_prepare, read and writes to log_empty_ are protected by
+  // Without two_write_queues, read and writes to log_empty_ are protected by
   // mutex_. Since it is currently updated/read only in write_thread_, it can be
   // accessed from the same write_thread_ without any locks. With
-  // concurrent_prepare writes, where it can be updated in different threads,
+  // two_write_queues writes, where it can be updated in different threads,
   // read and writes are protected by log_write_mutex_ instead. This is to avoid
   // expesnive mutex_ lock during WAL write, which update log_empty_.
   bool log_empty_;
@@ -912,10 +1004,10 @@ class DBImpl : public DB {
     // true for some prefix of logs_
     bool getting_synced = false;
   };
-  // Without concurrent_prepare, read and writes to alive_log_files_ are
+  // Without two_write_queues, read and writes to alive_log_files_ are
   // protected by mutex_. However since back() is never popped, and push_back()
   // is done only from write_thread_, the same thread can access the item
-  // reffered by back() without mutex_. With concurrent_prepare_, writes
+  // reffered by back() without mutex_. With two_write_queues_, writes
   // are protected by locking both mutex_ and log_write_mutex_, and reads must
   // be under either mutex_ or log_write_mutex_.
   std::deque<LogFileNumberSize> alive_log_files_;
@@ -935,6 +1027,15 @@ class DBImpl : public DB {
   std::deque<LogWriterNumber> logs_;
   // Signaled when getting_synced becomes false for some of the logs_.
   InstrumentedCondVar log_sync_cv_;
+  // This is the app-level state that is written to the WAL but will be used
+  // only during recovery. Using this feature enables not writing the state to
+  // memtable on normal writes and hence improving the throughput. Each new
+  // write of the state will replace the previous state entirely even if the
+  // keys in the two consecuitive states do not overlap.
+  // It is protected by log_write_mutex_ when two_write_queues_ is enabled.
+  // Otherwise only the heaad of write_thread_ can access it.
+  WriteBatch cached_recoverable_state_;
+  std::atomic<bool> cached_recoverable_state_empty_ = {true};
   std::atomic<uint64_t> total_log_size_;
   // only used for dynamically adjusting max_total_wal_size. it is a sum of
   // [write_buffer_size * max_write_buffer_number] over all column families
@@ -1056,6 +1157,10 @@ class DBImpl : public DB {
   int unscheduled_flushes_;
   int unscheduled_compactions_;
 
+  // count how many background compactions are running or have been scheduled in
+  // the BOTTOM pool
+  int bg_bottom_compaction_scheduled_;
+
   // count how many background compactions are running or have been scheduled
   int bg_compaction_scheduled_;
 
@@ -1072,7 +1177,7 @@ class DBImpl : public DB {
   int bg_purge_scheduled_;
 
   // Information for a manual compaction
-  struct ManualCompaction {
+  struct ManualCompactionState {
     ColumnFamilyData* cfd;
     int input_level;
     int output_level;
@@ -1088,13 +1193,21 @@ class DBImpl : public DB {
     InternalKey* manual_end;      // how far we are compacting
     InternalKey tmp_storage;      // Used to keep track of compaction progress
     InternalKey tmp_storage1;     // Used to keep track of compaction progress
-    Compaction* compaction;
   };
-  std::deque<ManualCompaction*> manual_compaction_dequeue_;
+  struct PrepickedCompaction {
+    // background compaction takes ownership of `compaction`.
+    Compaction* compaction;
+    // caller retains ownership of `manual_compaction_state` as it is reused
+    // across background compactions.
+    ManualCompactionState* manual_compaction_state;  // nullptr if non-manual
+  };
+  std::deque<ManualCompactionState*> manual_compaction_dequeue_;
 
   struct CompactionArg {
+    // caller retains ownership of `db`.
     DBImpl* db;
-    ManualCompaction* m;
+    // background compaction takes ownership of `prepicked_compaction`.
+    PrepickedCompaction* prepicked_compaction;
   };
 
   // Have we encountered a background error in paranoid mode?
@@ -1108,6 +1221,10 @@ class DBImpl : public DB {
   // without any synchronization
   int disable_delete_obsolete_files_;
 
+  // Number of times FindObsoleteFiles has found deletable files and the
+  // corresponding call to PurgeObsoleteFiles has not yet finished.
+  int pending_purge_obsolete_files_;
+
   // last time when DeleteObsoleteFiles with full scan was executed. Originaly
   // initialized with startup time.
   uint64_t delete_obsolete_files_last_run_;
@@ -1141,6 +1258,9 @@ class DBImpl : public DB {
   // The options to access storage files
   const EnvOptions env_options_;
 
+  // Additonal options for compaction and flush
+  EnvOptions env_options_for_compaction_;
+
   // Number of running IngestExternalFile() calls.
   // REQUIRES: mutex held
   int num_running_ingest_file_;
@@ -1186,23 +1306,23 @@ class DBImpl : public DB {
   std::unordered_map<uint64_t, uint64_t> prepared_section_completed_;
   std::mutex prep_heap_mutex_;
 
+  // Callback for compaction to check if a key is visible to a snapshot.
+  // REQUIRES: mutex held
+  std::unique_ptr<SnapshotChecker> snapshot_checker_;
+
   // No copying allowed
   DBImpl(const DBImpl&);
   void operator=(const DBImpl&);
 
   // Background threads call this function, which is just a wrapper around
   // the InstallSuperVersion() function. Background threads carry
-  // job_context which can have new_superversion already
+  // sv_context which can have new_superversion already
   // allocated.
-  void InstallSuperVersionAndScheduleWorkWrapper(
-      ColumnFamilyData* cfd, JobContext* job_context,
-      const MutableCFOptions& mutable_cf_options);
-
   // All ColumnFamily state changes go through this function. Here we analyze
   // the new state and we schedule background work if we detect that the new
   // state needs flush or compaction.
-  SuperVersion* InstallSuperVersionAndScheduleWork(
-      ColumnFamilyData* cfd, SuperVersion* new_sv,
+  void InstallSuperVersionAndScheduleWork(
+      ColumnFamilyData* cfd, SuperVersionContext* sv_context,
       const MutableCFOptions& mutable_cf_options);
 
 #ifndef ROCKSDB_LITE
@@ -1216,30 +1336,51 @@ class DBImpl : public DB {
 
 #endif  // ROCKSDB_LITE
 
-  // Function that Get and KeyMayExist call with no_io true or false
-  // Note: 'value_found' from KeyMayExist propagates here
-  Status GetImpl(const ReadOptions& options, ColumnFamilyHandle* column_family,
-                 const Slice& key, PinnableSlice* value,
-                 bool* value_found = nullptr);
-
   bool GetIntPropertyInternal(ColumnFamilyData* cfd,
                               const DBPropertyInfo& property_info,
                               bool is_locked, uint64_t* value);
 
   bool HasPendingManualCompaction();
   bool HasExclusiveManualCompaction();
-  void AddManualCompaction(ManualCompaction* m);
-  void RemoveManualCompaction(ManualCompaction* m);
-  bool ShouldntRunManualCompaction(ManualCompaction* m);
+  void AddManualCompaction(ManualCompactionState* m);
+  void RemoveManualCompaction(ManualCompactionState* m);
+  bool ShouldntRunManualCompaction(ManualCompactionState* m);
   bool HaveManualCompaction(ColumnFamilyData* cfd);
-  bool MCOverlap(ManualCompaction* m, ManualCompaction* m1);
+  bool MCOverlap(ManualCompactionState* m, ManualCompactionState* m1);
 
   size_t GetWalPreallocateBlockSize(uint64_t write_buffer_size) const;
+  Env::WriteLifeTimeHint CalculateWALWriteHint() {
+    return Env::WLTH_SHORT;
+  }
 
   // When set, we use a seprate queue for writes that dont write to memtable. In
   // 2PC these are the writes at Prepare phase.
-  const bool concurrent_prepare_;
+  const bool two_write_queues_;
   const bool manual_wal_flush_;
+  // Increase the sequence number after writing each batch, whether memtable is
+  // disabled for that or not. Otherwise the sequence number is increased after
+  // writing each key into memtable. This implies that when disable_memtable is
+  // set, the seq is not increased at all.
+  //
+  // Default: false
+  const bool seq_per_batch_;
+  // LastSequence also indicates last published sequence visibile to the
+  // readers. Otherwise LastPublishedSequence should be used.
+  const bool last_seq_same_as_publish_seq_;
+  // It indicates that a customized gc algorithm must be used for
+  // flush/compaction and if it is not provided vis SnapshotChecker, we should
+  // disable gc to be safe.
+  const bool use_custom_gc_;
+
+  // Clients must periodically call SetPreserveDeletesSequenceNumber()
+  // to advance this seqnum. Default value is 0 which means ALL deletes are
+  // preserved. Note that this has no effect if DBOptions.preserve_deletes
+  // is set to false.
+  std::atomic<SequenceNumber> preserve_deletes_seqnum_;
+  const bool preserve_deletes_;
+
+  // Flag to check whether Close() has been called on this DB
+  bool closed_;
 };
 
 extern Options SanitizeOptions(const std::string& db,

db/db_impl_compaction_flush.cc

@@ -86,11 +86,16 @@ Status DBImpl::FlushMemTableToOutputFile(
   std::vector<SequenceNumber> snapshot_seqs =
       snapshots_.GetAll(&earliest_write_conflict_snapshot);
 
+  auto snapshot_checker = snapshot_checker_.get();
+  if (use_custom_gc_ && snapshot_checker == nullptr) {
+    snapshot_checker = DisableGCSnapshotChecker::Instance();
+  }
   FlushJob flush_job(
-      dbname_, cfd, immutable_db_options_, mutable_cf_options, env_options_,
-      versions_.get(), &mutex_, &shutting_down_, snapshot_seqs,
-      earliest_write_conflict_snapshot, job_context, log_buffer,
-      directories_.GetDbDir(), directories_.GetDataDir(0U),
+      dbname_, cfd, immutable_db_options_, mutable_cf_options,
+      env_options_for_compaction_, versions_.get(), &mutex_, &shutting_down_,
+      snapshot_seqs, earliest_write_conflict_snapshot, snapshot_checker,
+      job_context, log_buffer, directories_.GetDbDir(),
+      directories_.GetDataDir(0U),
       GetCompressionFlush(*cfd->ioptions(), mutable_cf_options), stats_,
       &event_logger_, mutable_cf_options.report_bg_io_stats);
 
@@ -129,8 +134,9 @@ Status DBImpl::FlushMemTableToOutputFile(
   }
 
   if (s.ok()) {
-    InstallSuperVersionAndScheduleWorkWrapper(cfd, job_context,
-                                              mutable_cf_options);
+    TEST_SYNC_POINT("DBImpl::FlushMemTableToOutputFile:BeforeInstallSV");
+    InstallSuperVersionAndScheduleWork(cfd, &job_context->superversion_context,
+                                       mutable_cf_options);
     if (made_progress) {
       *made_progress = 1;
     }
@@ -436,7 +442,8 @@ Status DBImpl::CompactFiles(
   }  // release the mutex
 
   // delete unnecessary files if any, this is done outside the mutex
-  if (job_context.HaveSomethingToDelete() || !log_buffer.IsEmpty()) {
+  if (job_context.HaveSomethingToClean() ||
+      job_context.HaveSomethingToDelete() || !log_buffer.IsEmpty()) {
     // Have to flush the info logs before bg_compaction_scheduled_--
     // because if bg_flush_scheduled_ becomes 0 and the lock is
     // released, the deconstructor of DB can kick in and destroy all the
@@ -515,9 +522,10 @@ Status DBImpl::CompactFilesImpl(
   c.reset(cfd->compaction_picker()->CompactFiles(
       compact_options, input_files, output_level, version->storage_info(),
       *cfd->GetLatestMutableCFOptions(), output_path_id));
-  if (!c) {
-    return Status::Aborted("Another Level 0 compaction is running");
-  }
+  // we already sanitized the set of input files and checked for conflicts
+  // without releasing the lock, so we're guaranteed a compaction can be formed.
+  assert(c != nullptr);
+
   c->SetInputVersion(version);
   // deletion compaction currently not allowed in CompactFiles.
   assert(!c->deletion_compaction());
@@ -529,13 +537,19 @@ Status DBImpl::CompactFilesImpl(
   auto pending_outputs_inserted_elem =
       CaptureCurrentFileNumberInPendingOutputs();
 
+  auto snapshot_checker = snapshot_checker_.get();
+  if (use_custom_gc_ && snapshot_checker == nullptr) {
+    snapshot_checker = DisableGCSnapshotChecker::Instance();
+  }
   assert(is_snapshot_supported_ || snapshots_.empty());
   CompactionJob compaction_job(
-      job_context->job_id, c.get(), immutable_db_options_, env_options_,
-      versions_.get(), &shutting_down_, log_buffer, directories_.GetDbDir(),
+      job_context->job_id, c.get(), immutable_db_options_,
+      env_options_for_compaction_, versions_.get(), &shutting_down_,
+      preserve_deletes_seqnum_.load(), log_buffer, directories_.GetDbDir(),
       directories_.GetDataDir(c->output_path_id()), stats_, &mutex_, &bg_error_,
-      snapshot_seqs, earliest_write_conflict_snapshot, table_cache_,
-      &event_logger_, c->mutable_cf_options()->paranoid_file_checks,
+      snapshot_seqs, earliest_write_conflict_snapshot, snapshot_checker,
+      table_cache_, &event_logger_,
+      c->mutable_cf_options()->paranoid_file_checks,
       c->mutable_cf_options()->report_bg_io_stats, dbname_,
       nullptr);  // Here we pass a nullptr for CompactionJobStats because
                  // CompactFiles does not trigger OnCompactionCompleted(),
@@ -570,8 +584,9 @@ Status DBImpl::CompactFilesImpl(
 
   Status status = compaction_job.Install(*c->mutable_cf_options());
   if (status.ok()) {
-    InstallSuperVersionAndScheduleWorkWrapper(
-        c->column_family_data(), job_context, *c->mutable_cf_options());
+    InstallSuperVersionAndScheduleWork(
+        c->column_family_data(), &job_context->superversion_context,
+       *c->mutable_cf_options());
   }
   c->ReleaseCompactionFiles(s);
 
@@ -612,7 +627,8 @@ Status DBImpl::CompactFilesImpl(
 Status DBImpl::PauseBackgroundWork() {
   InstrumentedMutexLock guard_lock(&mutex_);
   bg_compaction_paused_++;
-  while (bg_compaction_scheduled_ > 0 || bg_flush_scheduled_ > 0) {
+  while (bg_bottom_compaction_scheduled_ > 0 || bg_compaction_scheduled_ > 0 ||
+         bg_flush_scheduled_ > 0) {
     bg_cv_.Wait();
   }
   bg_work_paused_++;
@@ -648,6 +664,8 @@ void DBImpl::NotifyOnCompactionCompleted(
   if (shutting_down_.load(std::memory_order_acquire)) {
     return;
   }
+  Version* current = cfd->current();
+  current->Ref();
   // release lock while notifying events
   mutex_.Unlock();
   TEST_SYNC_POINT("DBImpl::NotifyOnCompactionCompleted::UnlockMutex");
@@ -670,7 +688,7 @@ void DBImpl::NotifyOnCompactionCompleted(
         info.input_files.push_back(fn);
         if (info.table_properties.count(fn) == 0) {
           std::shared_ptr<const TableProperties> tp;
-          auto s = cfd->current()->GetTableProperties(&tp, fmd, &fn);
+          auto s = current->GetTableProperties(&tp, fmd, &fn);
           if (s.ok()) {
             info.table_properties[fn] = tp;
           }
@@ -687,6 +705,7 @@ void DBImpl::NotifyOnCompactionCompleted(
     }
   }
   mutex_.Lock();
+  current->Unref();
   // no need to signal bg_cv_ as it will be signaled at the end of the
   // flush process.
 #endif  // ROCKSDB_LITE
@@ -700,8 +719,7 @@ Status DBImpl::ReFitLevel(ColumnFamilyData* cfd, int level, int target_level) {
     return Status::InvalidArgument("Target level exceeds number of levels");
   }
 
-  std::unique_ptr<SuperVersion> superversion_to_free;
-  std::unique_ptr<SuperVersion> new_superversion(new SuperVersion());
+  SuperVersionContext sv_context(/* create_superversion */ true);
 
   Status status;
 
@@ -756,8 +774,7 @@ Status DBImpl::ReFitLevel(ColumnFamilyData* cfd, int level, int target_level) {
 
     status = versions_->LogAndApply(cfd, mutable_cf_options, &edit, &mutex_,
                                     directories_.GetDbDir());
-    superversion_to_free.reset(InstallSuperVersionAndScheduleWork(
-        cfd, new_superversion.release(), mutable_cf_options));
+    InstallSuperVersionAndScheduleWork(cfd, &sv_context, mutable_cf_options);
 
     ROCKS_LOG_DEBUG(immutable_db_options_.info_log, "[%s] LogAndApply: %s\n",
                     cfd->GetName().c_str(), status.ToString().data());
@@ -769,6 +786,7 @@ Status DBImpl::ReFitLevel(ColumnFamilyData* cfd, int level, int target_level) {
     }
   }
 
+  sv_context.Clean();
   refitting_level_ = false;
 
   return status;
@@ -793,7 +811,13 @@ int DBImpl::Level0StopWriteTrigger(ColumnFamilyHandle* column_family) {
 Status DBImpl::Flush(const FlushOptions& flush_options,
                      ColumnFamilyHandle* column_family) {
   auto cfh = reinterpret_cast<ColumnFamilyHandleImpl*>(column_family);
-  return FlushMemTable(cfh->cfd(), flush_options);
+  ROCKS_LOG_INFO(immutable_db_options_.info_log, "[%s] Manual flush start.",
+                 cfh->GetName().c_str());
+  Status s = FlushMemTable(cfh->cfd(), flush_options);
+  ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                 "[%s] Manual flush finished, status: %s\n",
+                 cfh->GetName().c_str(), s.ToString().c_str());
+  return s;
 }
 
 Status DBImpl::RunManualCompaction(ColumnFamilyData* cfd, int input_level,
@@ -808,7 +832,7 @@ Status DBImpl::RunManualCompaction(ColumnFamilyData* cfd, int input_level,
 
   bool scheduled = false;
   bool manual_conflict = false;
-  ManualCompaction manual;
+  ManualCompactionState manual;
   manual.cfd = cfd;
   manual.input_level = input_level;
   manual.output_level = output_level;
@@ -825,7 +849,7 @@ Status DBImpl::RunManualCompaction(ColumnFamilyData* cfd, int input_level,
       cfd->ioptions()->compaction_style == kCompactionStyleFIFO) {
     manual.begin = nullptr;
   } else {
-    begin_storage.SetMaxPossibleForUserKey(*begin);
+    begin_storage.SetMinPossibleForUserKey(*begin);
     manual.begin = &begin_storage;
   }
   if (end == nullptr ||
@@ -833,7 +857,7 @@ Status DBImpl::RunManualCompaction(ColumnFamilyData* cfd, int input_level,
       cfd->ioptions()->compaction_style == kCompactionStyleFIFO) {
     manual.end = nullptr;
   } else {
-    end_storage.SetMinPossibleForUserKey(*end);
+    end_storage.SetMaxPossibleForUserKey(*end);
     manual.end = &end_storage;
   }
 
@@ -858,7 +882,8 @@ Status DBImpl::RunManualCompaction(ColumnFamilyData* cfd, int input_level,
   AddManualCompaction(&manual);
   TEST_SYNC_POINT_CALLBACK("DBImpl::RunManualCompaction:NotScheduled", &mutex_);
   if (exclusive) {
-    while (bg_compaction_scheduled_ > 0) {
+    while (bg_bottom_compaction_scheduled_ > 0 ||
+           bg_compaction_scheduled_ > 0) {
       TEST_SYNC_POINT("DBImpl::RunManualCompaction:WaitScheduled");
       ROCKS_LOG_INFO(
           immutable_db_options_.info_log,
@@ -878,15 +903,15 @@ Status DBImpl::RunManualCompaction(ColumnFamilyData* cfd, int input_level,
   while (!manual.done) {
     assert(HasPendingManualCompaction());
     manual_conflict = false;
+    Compaction* compaction = nullptr;
     if (ShouldntRunManualCompaction(&manual) || (manual.in_progress == true) ||
         scheduled ||
-        ((manual.manual_end = &manual.tmp_storage1)&&(
-             (manual.compaction = manual.cfd->CompactRange(
+        (((manual.manual_end = &manual.tmp_storage1) != nullptr) &&
+             ((compaction = manual.cfd->CompactRange(
                   *manual.cfd->GetLatestMutableCFOptions(), manual.input_level,
                   manual.output_level, manual.output_path_id, manual.begin,
-                  manual.end, &manual.manual_end, &manual_conflict)) ==
-             nullptr) &&
-         manual_conflict)) {
+                  manual.end, &manual.manual_end, &manual_conflict)) == nullptr &&
+         manual_conflict))) {
       // exclusive manual compactions should not see a conflict during
       // CompactRange
       assert(!exclusive || !manual_conflict);
@@ -898,14 +923,16 @@ Status DBImpl::RunManualCompaction(ColumnFamilyData* cfd, int input_level,
         manual.incomplete = false;
       }
     } else if (!scheduled) {
-      if (manual.compaction == nullptr) {
+      if (compaction == nullptr) {
         manual.done = true;
         bg_cv_.SignalAll();
         continue;
       }
       ca = new CompactionArg;
       ca->db = this;
-      ca->m = &manual;
+      ca->prepicked_compaction = new PrepickedCompaction;
+      ca->prepicked_compaction->manual_compaction_state = &manual;
+      ca->prepicked_compaction->compaction = compaction;
       manual.incomplete = false;
       bg_compaction_scheduled_++;
       env_->Schedule(&DBImpl::BGWorkCompaction, ca, Env::Priority::LOW, this,
@@ -925,11 +952,13 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
                              const FlushOptions& flush_options,
                              bool writes_stopped) {
   Status s;
+  uint64_t flush_memtable_id = 0;
   {
     WriteContext context;
     InstrumentedMutexLock guard_lock(&mutex_);
 
-    if (cfd->imm()->NumNotFlushed() == 0 && cfd->mem()->IsEmpty()) {
+    if (cfd->imm()->NumNotFlushed() == 0 && cfd->mem()->IsEmpty() &&
+        cached_recoverable_state_empty_.load()) {
       // Nothing to flush
       return Status::OK();
     }
@@ -939,9 +968,9 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
       write_thread_.EnterUnbatched(&w, &mutex_);
     }
 
-    // SwitchMemtable() will release and reacquire mutex
-    // during execution
+    // SwitchMemtable() will release and reacquire mutex during execution
     s = SwitchMemtable(cfd, &context);
+    flush_memtable_id = cfd->imm()->GetLatestMemTableID();
 
     if (!writes_stopped) {
       write_thread_.ExitUnbatched(&w);
@@ -956,16 +985,19 @@ Status DBImpl::FlushMemTable(ColumnFamilyData* cfd,
 
   if (s.ok() && flush_options.wait) {
     // Wait until the compaction completes
-    s = WaitForFlushMemTable(cfd);
+    s = WaitForFlushMemTable(cfd, &flush_memtable_id);
   }
   return s;
 }
 
-Status DBImpl::WaitForFlushMemTable(ColumnFamilyData* cfd) {
+Status DBImpl::WaitForFlushMemTable(ColumnFamilyData* cfd,
+                                    const uint64_t* flush_memtable_id) {
   Status s;
   // Wait until the compaction completes
   InstrumentedMutexLock l(&mutex_);
-  while (cfd->imm()->NumNotFlushed() > 0 && bg_error_.ok()) {
+  while (cfd->imm()->NumNotFlushed() > 0 && bg_error_.ok() &&
+         (flush_memtable_id == nullptr ||
+          cfd->imm()->GetEarliestMemTableID() <= *flush_memtable_id)) {
     if (shutting_down_.load(std::memory_order_acquire)) {
       return Status::ShutdownInProgress();
     }
@@ -1040,6 +1072,7 @@ void DBImpl::MaybeScheduleFlushOrCompaction() {
   if (HasExclusiveManualCompaction()) {
     // only manual compactions are allowed to run. don't schedule automatic
     // compactions
+    TEST_SYNC_POINT("DBImpl::MaybeScheduleFlushOrCompaction:Conflict");
     return;
   }
 
@@ -1047,7 +1080,7 @@ void DBImpl::MaybeScheduleFlushOrCompaction() {
          unscheduled_compactions_ > 0) {
     CompactionArg* ca = new CompactionArg;
     ca->db = this;
-    ca->m = nullptr;
+    ca->prepicked_compaction = nullptr;
     bg_compaction_scheduled_++;
     unscheduled_compactions_--;
     env_->Schedule(&DBImpl::BGWorkCompaction, ca, Env::Priority::LOW, this,
@@ -1152,7 +1185,23 @@ void DBImpl::BGWorkCompaction(void* arg) {
   delete reinterpret_cast<CompactionArg*>(arg);
   IOSTATS_SET_THREAD_POOL_ID(Env::Priority::LOW);
   TEST_SYNC_POINT("DBImpl::BGWorkCompaction");
-  reinterpret_cast<DBImpl*>(ca.db)->BackgroundCallCompaction(ca.m);
+  auto prepicked_compaction =
+      static_cast<PrepickedCompaction*>(ca.prepicked_compaction);
+  reinterpret_cast<DBImpl*>(ca.db)->BackgroundCallCompaction(
+      prepicked_compaction, Env::Priority::LOW);
+  delete prepicked_compaction;
+}
+
+void DBImpl::BGWorkBottomCompaction(void* arg) {
+  CompactionArg ca = *(static_cast<CompactionArg*>(arg));
+  delete static_cast<CompactionArg*>(arg);
+  IOSTATS_SET_THREAD_POOL_ID(Env::Priority::BOTTOM);
+  TEST_SYNC_POINT("DBImpl::BGWorkBottomCompaction");
+  auto* prepicked_compaction = ca.prepicked_compaction;
+  assert(prepicked_compaction && prepicked_compaction->compaction &&
+         !prepicked_compaction->manual_compaction_state);
+  ca.db->BackgroundCallCompaction(prepicked_compaction, Env::Priority::BOTTOM);
+  delete prepicked_compaction;
 }
 
 void DBImpl::BGWorkPurge(void* db) {
@@ -1165,8 +1214,11 @@ void DBImpl::BGWorkPurge(void* db) {
 void DBImpl::UnscheduleCallback(void* arg) {
   CompactionArg ca = *(reinterpret_cast<CompactionArg*>(arg));
   delete reinterpret_cast<CompactionArg*>(arg);
-  if ((ca.m != nullptr) && (ca.m->compaction != nullptr)) {
-    delete ca.m->compaction;
+  if (ca.prepicked_compaction != nullptr) {
+    if (ca.prepicked_compaction->compaction != nullptr) {
+      delete ca.prepicked_compaction->compaction;
+    }
+    delete ca.prepicked_compaction;
   }
   TEST_SYNC_POINT("DBImpl::UnscheduleCallback");
 }
@@ -1265,8 +1317,10 @@ void DBImpl::BackgroundCallFlush() {
     // created. Thus, we force full scan in FindObsoleteFiles()
     FindObsoleteFiles(&job_context, !s.ok() && !s.IsShutdownInProgress());
     // delete unnecessary files if any, this is done outside the mutex
-    if (job_context.HaveSomethingToDelete() || !log_buffer.IsEmpty()) {
+    if (job_context.HaveSomethingToClean() ||
+        job_context.HaveSomethingToDelete() || !log_buffer.IsEmpty()) {
       mutex_.Unlock();
+      TEST_SYNC_POINT("DBImpl::BackgroundCallFlush:FilesFound");
       // Have to flush the info logs before bg_flush_scheduled_--
       // because if bg_flush_scheduled_ becomes 0 and the lock is
       // released, the deconstructor of DB can kick in and destroy all the
@@ -1293,9 +1347,9 @@ void DBImpl::BackgroundCallFlush() {
   }
 }
 
-void DBImpl::BackgroundCallCompaction(void* arg) {
+void DBImpl::BackgroundCallCompaction(PrepickedCompaction* prepicked_compaction,
+                                      Env::Priority bg_thread_pri) {
   bool made_progress = false;
-  ManualCompaction* m = reinterpret_cast<ManualCompaction*>(arg);
   JobContext job_context(next_job_id_.fetch_add(1), true);
   TEST_SYNC_POINT("BackgroundCallCompaction:0");
   MaybeDumpStats();
@@ -1313,9 +1367,11 @@ void DBImpl::BackgroundCallCompaction(void* arg) {
     auto pending_outputs_inserted_elem =
         CaptureCurrentFileNumberInPendingOutputs();
 
-    assert(bg_compaction_scheduled_);
-    Status s =
-        BackgroundCompaction(&made_progress, &job_context, &log_buffer, m);
+    assert((bg_thread_pri == Env::Priority::BOTTOM &&
+            bg_bottom_compaction_scheduled_) ||
+           (bg_thread_pri == Env::Priority::LOW && bg_compaction_scheduled_));
+    Status s = BackgroundCompaction(&made_progress, &job_context, &log_buffer,
+                                    prepicked_compaction);
     TEST_SYNC_POINT("BackgroundCallCompaction:1");
     if (!s.ok() && !s.IsShutdownInProgress()) {
       // Wait a little bit before retrying background compaction in
@@ -1344,7 +1400,8 @@ void DBImpl::BackgroundCallCompaction(void* arg) {
     FindObsoleteFiles(&job_context, !s.ok() && !s.IsShutdownInProgress());
 
     // delete unnecessary files if any, this is done outside the mutex
-    if (job_context.HaveSomethingToDelete() || !log_buffer.IsEmpty()) {
+    if (job_context.HaveSomethingToClean() ||
+        job_context.HaveSomethingToDelete() || !log_buffer.IsEmpty()) {
       mutex_.Unlock();
       // Have to flush the info logs before bg_compaction_scheduled_--
       // because if bg_flush_scheduled_ becomes 0 and the lock is
@@ -1361,17 +1418,24 @@ void DBImpl::BackgroundCallCompaction(void* arg) {
 
     assert(num_running_compactions_ > 0);
     num_running_compactions_--;
-    bg_compaction_scheduled_--;
+    if (bg_thread_pri == Env::Priority::LOW) {
+      bg_compaction_scheduled_--;
+    } else {
+      assert(bg_thread_pri == Env::Priority::BOTTOM);
+      bg_bottom_compaction_scheduled_--;
+    }
 
     versions_->GetColumnFamilySet()->FreeDeadColumnFamilies();
 
     // See if there's more work to be done
     MaybeScheduleFlushOrCompaction();
-    if (made_progress || bg_compaction_scheduled_ == 0 ||
+    if (made_progress ||
+        (bg_compaction_scheduled_ == 0 &&
+         bg_bottom_compaction_scheduled_ == 0) ||
         HasPendingManualCompaction()) {
       // signal if
       // * made_progress -- need to wakeup DelayWrite
-      // * bg_compaction_scheduled_ == 0 -- need to wakeup ~DBImpl
+      // * bg_{bottom,}_compaction_scheduled_ == 0 -- need to wakeup ~DBImpl
       // * HasPendingManualCompaction -- need to wakeup RunManualCompaction
       // If none of this is true, there is no need to signal since nobody is
       // waiting for it
@@ -1386,14 +1450,23 @@ void DBImpl::BackgroundCallCompaction(void* arg) {
 
 Status DBImpl::BackgroundCompaction(bool* made_progress,
                                     JobContext* job_context,
-                                    LogBuffer* log_buffer, void* arg) {
-  ManualCompaction* manual_compaction =
-      reinterpret_cast<ManualCompaction*>(arg);
+                                    LogBuffer* log_buffer,
+                                    PrepickedCompaction* prepicked_compaction) {
+  ManualCompactionState* manual_compaction =
+      prepicked_compaction == nullptr
+          ? nullptr
+          : prepicked_compaction->manual_compaction_state;
   *made_progress = false;
   mutex_.AssertHeld();
   TEST_SYNC_POINT("DBImpl::BackgroundCompaction:Start");
 
   bool is_manual = (manual_compaction != nullptr);
+  unique_ptr<Compaction> c;
+  if (prepicked_compaction != nullptr &&
+      prepicked_compaction->compaction != nullptr) {
+    c.reset(prepicked_compaction->compaction);
+  }
+  bool is_prepicked = is_manual || c;
 
   // (manual_compaction->in_progress == false);
   bool trivial_move_disallowed =
@@ -1410,7 +1483,6 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
       manual_compaction->status = status;
       manual_compaction->done = true;
       manual_compaction->in_progress = false;
-      delete manual_compaction->compaction;
       manual_compaction = nullptr;
     }
     return status;
@@ -1421,13 +1493,11 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
     manual_compaction->in_progress = true;
   }
 
-  unique_ptr<Compaction> c;
   // InternalKey manual_end_storage;
   // InternalKey* manual_end = &manual_end_storage;
   if (is_manual) {
-    ManualCompaction* m = manual_compaction;
+    ManualCompactionState* m = manual_compaction;
     assert(m->in_progress);
-    c.reset(std::move(m->compaction));
     if (!c) {
       m->done = true;
       m->manual_end = nullptr;
@@ -1449,8 +1519,8 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
                ? "(end)"
                : m->manual_end->DebugString().c_str()));
     }
-  } else if (!compaction_queue_.empty()) {
-    if (HaveManualCompaction(compaction_queue_.front())) {
+  } else if (!is_prepicked && !compaction_queue_.empty()) {
+    if (HasExclusiveManualCompaction()) {
       // Can't compact right now, but try again later
       TEST_SYNC_POINT("DBImpl::BackgroundCompaction()::Conflict");
 
@@ -1532,8 +1602,9 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
     status = versions_->LogAndApply(c->column_family_data(),
                                     *c->mutable_cf_options(), c->edit(),
                                     &mutex_, directories_.GetDbDir());
-    InstallSuperVersionAndScheduleWorkWrapper(
-        c->column_family_data(), job_context, *c->mutable_cf_options());
+    InstallSuperVersionAndScheduleWork(
+        c->column_family_data(), &job_context->superversion_context,
+        *c->mutable_cf_options());
     ROCKS_LOG_BUFFER(log_buffer, "[%s] Deleted %d files\n",
                      c->column_family_data()->GetName().c_str(),
                      c->num_input_files(0));
@@ -1578,8 +1649,9 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
                                     *c->mutable_cf_options(), c->edit(),
                                     &mutex_, directories_.GetDbDir());
     // Use latest MutableCFOptions
-    InstallSuperVersionAndScheduleWorkWrapper(
-        c->column_family_data(), job_context, *c->mutable_cf_options());
+    InstallSuperVersionAndScheduleWork(
+        c->column_family_data(), &job_context->superversion_context,
+        *c->mutable_cf_options());
 
     VersionStorageInfo::LevelSummaryStorage tmp;
     c->column_family_data()->internal_stats()->IncBytesMoved(c->output_level(),
@@ -1601,22 +1673,49 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
 
     // Clear Instrument
     ThreadStatusUtil::ResetThreadStatus();
+  } else if (c->column_family_data()->ioptions()->compaction_style ==
+                 kCompactionStyleUniversal &&
+             !is_prepicked && c->output_level() > 0 &&
+             c->output_level() ==
+                 c->column_family_data()
+                     ->current()
+                     ->storage_info()
+                     ->MaxOutputLevel(
+                         immutable_db_options_.allow_ingest_behind) &&
+             env_->GetBackgroundThreads(Env::Priority::BOTTOM) > 0) {
+    // Forward universal compactions involving last level to the bottom pool
+    // if it exists, such that long-running compactions can't block short-
+    // lived ones, like L0->L0s.
+    TEST_SYNC_POINT("DBImpl::BackgroundCompaction:ForwardToBottomPriPool");
+    CompactionArg* ca = new CompactionArg;
+    ca->db = this;
+    ca->prepicked_compaction = new PrepickedCompaction;
+    ca->prepicked_compaction->compaction = c.release();
+    ca->prepicked_compaction->manual_compaction_state = nullptr;
+    ++bg_bottom_compaction_scheduled_;
+    env_->Schedule(&DBImpl::BGWorkBottomCompaction, ca, Env::Priority::BOTTOM,
+                   this, &DBImpl::UnscheduleCallback);
   } else {
-    int output_level  __attribute__((unused)) = c->output_level();
+    int output_level  __attribute__((unused));
+    output_level = c->output_level();
     TEST_SYNC_POINT_CALLBACK("DBImpl::BackgroundCompaction:NonTrivial",
                              &output_level);
-
     SequenceNumber earliest_write_conflict_snapshot;
     std::vector<SequenceNumber> snapshot_seqs =
         snapshots_.GetAll(&earliest_write_conflict_snapshot);
 
+    auto snapshot_checker = snapshot_checker_.get();
+    if (use_custom_gc_ && snapshot_checker == nullptr) {
+      snapshot_checker = DisableGCSnapshotChecker::Instance();
+    }
     assert(is_snapshot_supported_ || snapshots_.empty());
     CompactionJob compaction_job(
-        job_context->job_id, c.get(), immutable_db_options_, env_options_,
-        versions_.get(), &shutting_down_, log_buffer, directories_.GetDbDir(),
+        job_context->job_id, c.get(), immutable_db_options_,
+        env_options_for_compaction_, versions_.get(), &shutting_down_,
+        preserve_deletes_seqnum_.load(), log_buffer, directories_.GetDbDir(),
         directories_.GetDataDir(c->output_path_id()), stats_, &mutex_,
         &bg_error_, snapshot_seqs, earliest_write_conflict_snapshot,
-        table_cache_, &event_logger_,
+        snapshot_checker, table_cache_, &event_logger_,
         c->mutable_cf_options()->paranoid_file_checks,
         c->mutable_cf_options()->report_bg_io_stats, dbname_,
         &compaction_job_stats);
@@ -1629,8 +1728,9 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
 
     status = compaction_job.Install(*c->mutable_cf_options());
     if (status.ok()) {
-      InstallSuperVersionAndScheduleWorkWrapper(
-          c->column_family_data(), job_context, *c->mutable_cf_options());
+      InstallSuperVersionAndScheduleWork(
+          c->column_family_data(), &job_context->superversion_context,
+          *c->mutable_cf_options());
     }
     *made_progress = true;
   }
@@ -1664,7 +1764,7 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
   }
 
   if (is_manual) {
-    ManualCompaction* m = manual_compaction;
+    ManualCompactionState* m = manual_compaction;
     if (!status.ok()) {
       m->status = status;
       m->done = true;
@@ -1707,13 +1807,13 @@ bool DBImpl::HasPendingManualCompaction() {
   return (!manual_compaction_dequeue_.empty());
 }
 
-void DBImpl::AddManualCompaction(DBImpl::ManualCompaction* m) {
+void DBImpl::AddManualCompaction(DBImpl::ManualCompactionState* m) {
   manual_compaction_dequeue_.push_back(m);
 }
 
-void DBImpl::RemoveManualCompaction(DBImpl::ManualCompaction* m) {
+void DBImpl::RemoveManualCompaction(DBImpl::ManualCompactionState* m) {
   // Remove from queue
-  std::deque<ManualCompaction*>::iterator it =
+  std::deque<ManualCompactionState*>::iterator it =
       manual_compaction_dequeue_.begin();
   while (it != manual_compaction_dequeue_.end()) {
     if (m == (*it)) {
@@ -1726,16 +1826,17 @@ void DBImpl::RemoveManualCompaction(DBImpl::ManualCompaction* m) {
   return;
 }
 
-bool DBImpl::ShouldntRunManualCompaction(ManualCompaction* m) {
+bool DBImpl::ShouldntRunManualCompaction(ManualCompactionState* m) {
   if (num_running_ingest_file_ > 0) {
     // We need to wait for other IngestExternalFile() calls to finish
     // before running a manual compaction.
     return true;
   }
   if (m->exclusive) {
-    return (bg_compaction_scheduled_ > 0);
+    return (bg_bottom_compaction_scheduled_ > 0 ||
+            bg_compaction_scheduled_ > 0);
   }
-  std::deque<ManualCompaction*>::iterator it =
+  std::deque<ManualCompactionState*>::iterator it =
       manual_compaction_dequeue_.begin();
   bool seen = false;
   while (it != manual_compaction_dequeue_.end()) {
@@ -1756,7 +1857,7 @@ bool DBImpl::ShouldntRunManualCompaction(ManualCompaction* m) {
 
 bool DBImpl::HaveManualCompaction(ColumnFamilyData* cfd) {
   // Remove from priority queue
-  std::deque<ManualCompaction*>::iterator it =
+  std::deque<ManualCompactionState*>::iterator it =
       manual_compaction_dequeue_.begin();
   while (it != manual_compaction_dequeue_.end()) {
     if ((*it)->exclusive) {
@@ -1774,7 +1875,7 @@ bool DBImpl::HaveManualCompaction(ColumnFamilyData* cfd) {
 
 bool DBImpl::HasExclusiveManualCompaction() {
   // Remove from priority queue
-  std::deque<ManualCompaction*>::iterator it =
+  std::deque<ManualCompactionState*>::iterator it =
       manual_compaction_dequeue_.begin();
   while (it != manual_compaction_dequeue_.end()) {
     if ((*it)->exclusive) {
@@ -1785,7 +1886,7 @@ bool DBImpl::HasExclusiveManualCompaction() {
   return false;
 }
 
-bool DBImpl::MCOverlap(ManualCompaction* m, ManualCompaction* m1) {
+bool DBImpl::MCOverlap(ManualCompactionState* m, ManualCompactionState* m1) {
   if ((m->exclusive) || (m1->exclusive)) {
     return true;
   }
@@ -1795,30 +1896,21 @@ bool DBImpl::MCOverlap(ManualCompaction* m, ManualCompaction* m1) {
   return true;
 }
 
-// JobContext gets created and destructed outside of the lock --
+// SuperVersionContext gets created and destructed outside of the lock --
 // we
 // use this convinently to:
 // * malloc one SuperVersion() outside of the lock -- new_superversion
 // * delete SuperVersion()s outside of the lock -- superversions_to_free
 //
 // However, if InstallSuperVersionAndScheduleWork() gets called twice with the
-// same job_context, we can't reuse the SuperVersion() that got
+// same sv_context, we can't reuse the SuperVersion() that got
 // malloced because
 // first call already used it. In that rare case, we take a hit and create a
 // new SuperVersion() inside of the mutex. We do similar thing
 // for superversion_to_free
-void DBImpl::InstallSuperVersionAndScheduleWorkWrapper(
-    ColumnFamilyData* cfd, JobContext* job_context,
-    const MutableCFOptions& mutable_cf_options) {
-  mutex_.AssertHeld();
-  SuperVersion* old_superversion = InstallSuperVersionAndScheduleWork(
-      cfd, job_context->new_superversion, mutable_cf_options);
-  job_context->new_superversion = nullptr;
-  job_context->superversions_to_free.push_back(old_superversion);
-}
 
-SuperVersion* DBImpl::InstallSuperVersionAndScheduleWork(
-    ColumnFamilyData* cfd, SuperVersion* new_sv,
+void DBImpl::InstallSuperVersionAndScheduleWork(
+    ColumnFamilyData* cfd, SuperVersionContext* sv_context,
     const MutableCFOptions& mutable_cf_options) {
   mutex_.AssertHeld();
 
@@ -1830,8 +1922,11 @@ SuperVersion* DBImpl::InstallSuperVersionAndScheduleWork(
                         old_sv->mutable_cf_options.max_write_buffer_number;
   }
 
-  auto* old = cfd->InstallSuperVersion(
-      new_sv ? new_sv : new SuperVersion(), &mutex_, mutable_cf_options);
+  if (sv_context->new_superversion == nullptr) {
+    sv_context->NewSuperVersion();
+  }
+  cfd->InstallSuperVersion(sv_context, &mutex_,
+                           mutable_cf_options);
 
   // Whenever we install new SuperVersion, we might need to issue new flushes or
   // compactions.
@@ -1844,6 +1939,14 @@ SuperVersion* DBImpl::InstallSuperVersionAndScheduleWork(
       max_total_in_memory_state_ - old_memtable_size +
       mutable_cf_options.write_buffer_size *
       mutable_cf_options.max_write_buffer_number;
-  return old;
+}
+
+void DBImpl::SetSnapshotChecker(SnapshotChecker* snapshot_checker) {
+  InstrumentedMutexLock l(&mutex_);
+  // snapshot_checker_ should only set once. If we need to set it multiple
+  // times, we need to make sure the old one is not deleted while it is still
+  // using by a compaction job.
+  assert(!snapshot_checker_);
+  snapshot_checker_.reset(snapshot_checker);
 }
 }  // namespace rocksdb

db/db_impl_debug.cc

@@ -19,10 +19,10 @@ uint64_t DBImpl::TEST_GetLevel0TotalSize() {
   return default_cf_handle_->cfd()->current()->storage_info()->NumLevelBytes(0);
 }
 
-void DBImpl::TEST_HandleWALFull() {
+void DBImpl::TEST_SwitchWAL() {
   WriteContext write_context;
   InstrumentedMutexLock l(&mutex_);
-  HandleWALFull(&write_context);
+  SwitchWAL(&write_context);
 }
 
 int64_t DBImpl::TEST_MaxNextLevelOverlappingBytes(
@@ -60,6 +60,10 @@ uint64_t DBImpl::TEST_Current_Manifest_FileNo() {
   return versions_->manifest_file_number();
 }
 
+uint64_t DBImpl::TEST_Current_Next_FileNo() {
+  return versions_->current_next_file_number();
+}
+
 Status DBImpl::TEST_CompactRange(int level, const Slice* begin,
                                  const Slice* end,
                                  ColumnFamilyHandle* column_family,
@@ -80,6 +84,15 @@ Status DBImpl::TEST_CompactRange(int level, const Slice* begin,
                              disallow_trivial_move);
 }
 
+Status DBImpl::TEST_SwitchMemtable(ColumnFamilyData* cfd) {
+  WriteContext write_context;
+  InstrumentedMutexLock l(&mutex_);
+  if (cfd == nullptr) {
+    cfd = default_cf_handle_->cfd();
+  }
+  return SwitchMemtable(cfd, &write_context);
+}
+
 Status DBImpl::TEST_FlushMemTable(bool wait, ColumnFamilyHandle* cfh) {
   FlushOptions fo;
   fo.wait = wait;
@@ -112,7 +125,9 @@ Status DBImpl::TEST_WaitForCompact() {
   // OR flush to finish.
 
   InstrumentedMutexLock l(&mutex_);
-  while ((bg_compaction_scheduled_ || bg_flush_scheduled_) && bg_error_.ok()) {
+  while ((bg_bottom_compaction_scheduled_ || bg_compaction_scheduled_ ||
+          bg_flush_scheduled_) &&
+         bg_error_.ok()) {
     bg_cv_.Wait();
   }
   return bg_error_;
@@ -194,5 +209,13 @@ int DBImpl::TEST_BGFlushesAllowed() const {
   return GetBGJobLimits().max_flushes;
 }
 
+SequenceNumber DBImpl::TEST_GetLastVisibleSequence() const {
+  if (last_seq_same_as_publish_seq_) {
+    return versions_->LastSequence();
+  } else {
+    return versions_->LastAllocatedSequence();
+  }
+}
+
 }  // namespace rocksdb
 #endif  // NDEBUG

db/db_impl_experimental.cc

@@ -30,10 +30,10 @@ Status DBImpl::SuggestCompactRange(ColumnFamilyHandle* column_family,
   auto cfd = cfh->cfd();
   InternalKey start_key, end_key;
   if (begin != nullptr) {
-    start_key.SetMaxPossibleForUserKey(*begin);
+    start_key.SetMinPossibleForUserKey(*begin);
   }
   if (end != nullptr) {
-    end_key.SetMinPossibleForUserKey(*end);
+    end_key.SetMaxPossibleForUserKey(*end);
   }
   {
     InstrumentedMutexLock l(&mutex_);
@@ -138,8 +138,9 @@ Status DBImpl::PromoteL0(ColumnFamilyHandle* column_family, int target_level) {
     status = versions_->LogAndApply(cfd, *cfd->GetLatestMutableCFOptions(),
                                     &edit, &mutex_, directories_.GetDbDir());
     if (status.ok()) {
-      InstallSuperVersionAndScheduleWorkWrapper(
-          cfd, &job_context, *cfd->GetLatestMutableCFOptions());
+      InstallSuperVersionAndScheduleWork(
+          cfd, &job_context.superversion_context,
+          *cfd->GetLatestMutableCFOptions());
     }
   }  // lock released here
   LogFlush(immutable_db_options_.info_log);

db/db_impl_files.cc

@@ -48,6 +48,7 @@ uint64_t DBImpl::FindMinPrepLogReferencedByMemTable() {
   return min_log;
 }
 
+// TODO(myabandeh): Avoid using locks
 void DBImpl::MarkLogAsHavingPrepSectionFlushed(uint64_t log) {
   assert(log != 0);
   std::lock_guard<std::mutex> lock(prep_heap_mutex_);
@@ -56,6 +57,7 @@ void DBImpl::MarkLogAsHavingPrepSectionFlushed(uint64_t log) {
   it->second += 1;
 }
 
+// TODO(myabandeh): Avoid using locks
 void DBImpl::MarkLogAsContainingPrepSection(uint64_t log) {
   assert(log != 0);
   std::lock_guard<std::mutex> lock(prep_heap_mutex_);
@@ -236,11 +238,11 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
     while (alive_log_files_.begin()->number < min_log_number) {
       auto& earliest = *alive_log_files_.begin();
       if (immutable_db_options_.recycle_log_file_num >
-          log_recycle_files.size()) {
+          log_recycle_files_.size()) {
         ROCKS_LOG_INFO(immutable_db_options_.info_log,
                        "adding log %" PRIu64 " to recycle list\n",
                        earliest.number);
-        log_recycle_files.push_back(earliest.number);
+        log_recycle_files_.push_back(earliest.number);
       } else {
         job_context->log_delete_files.push_back(earliest.number);
       }
@@ -250,11 +252,11 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
       }
       job_context->size_log_to_delete += earliest.size;
       total_log_size_ -= earliest.size;
-      if (concurrent_prepare_) {
+      if (two_write_queues_) {
         log_write_mutex_.Lock();
       }
       alive_log_files_.pop_front();
-      if (concurrent_prepare_) {
+      if (two_write_queues_) {
         log_write_mutex_.Unlock();
       }
       // Current log should always stay alive since it can't have
@@ -281,8 +283,11 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
   // We're just cleaning up for DB::Write().
   assert(job_context->logs_to_free.empty());
   job_context->logs_to_free = logs_to_free_;
-  job_context->log_recycle_files.assign(log_recycle_files.begin(),
-                                        log_recycle_files.end());
+  job_context->log_recycle_files.assign(log_recycle_files_.begin(),
+                                        log_recycle_files_.end());
+  if (job_context->HaveSomethingToDelete()) {
+    ++pending_purge_obsolete_files_;
+  }
   logs_to_free_.clear();
 }
 
@@ -300,9 +305,10 @@ bool CompareCandidateFile(const JobContext::CandidateFileInfo& first,
 };  // namespace
 
 // Delete obsolete files and log status and information of file deletion
-void DBImpl::DeleteObsoleteFileImpl(Status file_deletion_status, int job_id,
-                                    const std::string& fname, FileType type,
-                                    uint64_t number, uint32_t path_id) {
+void DBImpl::DeleteObsoleteFileImpl(int job_id, const std::string& fname,
+                                    FileType type, uint64_t number,
+                                    uint32_t path_id) {
+  Status file_deletion_status;
   if (type == kTableFile) {
     file_deletion_status =
         DeleteSSTFile(&immutable_db_options_, fname, path_id);
@@ -339,15 +345,12 @@ void DBImpl::DeleteObsoleteFileImpl(Status file_deletion_status, int job_id,
 // files in sst_delete_files and log_delete_files.
 // It is not necessary to hold the mutex when invoking this method.
 void DBImpl::PurgeObsoleteFiles(const JobContext& state, bool schedule_only) {
+  TEST_SYNC_POINT("DBImpl::PurgeObsoleteFiles:Begin");
   // we'd better have sth to delete
   assert(state.HaveSomethingToDelete());
 
-  // this checks if FindObsoleteFiles() was run before. If not, don't do
-  // PurgeObsoleteFiles(). If FindObsoleteFiles() was run, we need to also
-  // run PurgeObsoleteFiles(), even if disable_delete_obsolete_files_ is true
-  if (state.manifest_file_number == 0) {
-    return;
-  }
+  // FindObsoleteFiles() should've populated this so nonzero
+  assert(state.manifest_file_number != 0);
 
   // Now, convert live list to an unordered map, WITHOUT mutex held;
   // set is slow.
@@ -368,6 +371,9 @@ void DBImpl::PurgeObsoleteFiles(const JobContext& state, bool schedule_only) {
     candidate_files.emplace_back(
         MakeTableFileName(kDumbDbName, file->fd.GetNumber()),
         file->fd.GetPathId());
+    if (file->table_reader_handle) {
+      table_cache_->Release(file->table_reader_handle);
+    }
     delete file;
   }
 
@@ -485,8 +491,7 @@ void DBImpl::PurgeObsoleteFiles(const JobContext& state, bool schedule_only) {
       InstrumentedMutexLock guard_lock(&mutex_);
       SchedulePendingPurge(fname, type, number, path_id, state.job_id);
     } else {
-      DeleteObsoleteFileImpl(file_deletion_status, state.job_id, fname, type,
-                             number, path_id);
+      DeleteObsoleteFileImpl(state.job_id, fname, type, number, path_id);
     }
   }
 
@@ -528,6 +533,13 @@ void DBImpl::PurgeObsoleteFiles(const JobContext& state, bool schedule_only) {
   wal_manager_.PurgeObsoleteWALFiles();
 #endif  // ROCKSDB_LITE
   LogFlush(immutable_db_options_.info_log);
+  InstrumentedMutexLock l(&mutex_);
+  --pending_purge_obsolete_files_;
+  assert(pending_purge_obsolete_files_ >= 0);
+  if (pending_purge_obsolete_files_ == 0) {
+    bg_cv_.SignalAll();
+  }
+  TEST_SYNC_POINT("DBImpl::PurgeObsoleteFiles:End");
 }
 
 void DBImpl::DeleteObsoleteFiles() {

db/db_impl_open.cc

@@ -124,6 +124,17 @@ DBOptions SanitizeOptions(const std::string& dbname, const DBOptions& src) {
     result.avoid_flush_during_recovery = false;
   }
 
+#ifndef ROCKSDB_LITE
+  // When the DB is stopped, it's possible that there are some .trash files that
+  // were not deleted yet, when we open the DB we will find these .trash files
+  // and schedule them to be deleted (or delete immediately if SstFileManager
+  // was not used)
+  auto sfm = static_cast<SstFileManagerImpl*>(result.sst_file_manager.get());
+  for (size_t i = 0; i < result.db_paths.size(); i++) {
+    DeleteScheduler::CleanupDirectory(result.env, sfm, result.db_paths[i].path);
+  }
+#endif
+
   return result;
 }
 
@@ -493,11 +504,12 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
   bool stop_replay_by_wal_filter = false;
   bool stop_replay_for_corruption = false;
   bool flushed = false;
+  uint64_t corrupted_log_number = kMaxSequenceNumber;
   for (auto log_number : log_numbers) {
     // The previous incarnation may not have written any MANIFEST
     // records after allocating this log number.  So we manually
     // update the file number allocation counter in VersionSet.
-    versions_->MarkFileNumberUsedDuringRecovery(log_number);
+    versions_->MarkFileNumberUsed(log_number);
     // Open the log file
     std::string fname = LogFileName(immutable_db_options_.wal_dir, log_number);
 
@@ -674,7 +686,7 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
       status = WriteBatchInternal::InsertInto(
           &batch, column_family_memtables_.get(), &flush_scheduler_, true,
           log_number, this, false /* concurrent_memtable_writes */,
-          next_sequence, &has_valid_writes);
+          next_sequence, &has_valid_writes, seq_per_batch_);
       MaybeIgnoreError(&status);
       if (!status.ok()) {
         // We are treating this as a failure while reading since we read valid
@@ -711,6 +723,12 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
     }
 
     if (!status.ok()) {
+      if (status.IsNotSupported()) {
+        // We should not treat NotSupported as corruption. It is rather a clear
+        // sign that we are processing a WAL that is produced by an incompatible
+        // version of the code.
+        return status;
+      }
       if (immutable_db_options_.wal_recovery_mode ==
           WALRecoveryMode::kSkipAnyCorruptedRecords) {
         // We should ignore all errors unconditionally
@@ -720,6 +738,7 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
         // We should ignore the error but not continue replaying
         status = Status::OK();
         stop_replay_for_corruption = true;
+        corrupted_log_number = log_number;
         ROCKS_LOG_INFO(immutable_db_options_.info_log,
                        "Point in time recovered to log #%" PRIu64
                        " seq #%" PRIu64,
@@ -737,10 +756,30 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
     auto last_sequence = *next_sequence - 1;
     if ((*next_sequence != kMaxSequenceNumber) &&
         (versions_->LastSequence() <= last_sequence)) {
-      versions_->SetLastToBeWrittenSequence(last_sequence);
+      versions_->SetLastAllocatedSequence(last_sequence);
+      versions_->SetLastPublishedSequence(last_sequence);
       versions_->SetLastSequence(last_sequence);
     }
   }
+  // Compare the corrupted log number to all columnfamily's current log number.
+  // Abort Open() if any column family's log number is greater than
+  // 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
+  if (stop_replay_for_corruption == true &&
+      (immutable_db_options_.wal_recovery_mode ==
+           WALRecoveryMode::kPointInTimeRecovery ||
+       immutable_db_options_.wal_recovery_mode ==
+           WALRecoveryMode::kTolerateCorruptedTailRecords)) {
+    for (auto cfd : *versions_->GetColumnFamilySet()) {
+      if (cfd->GetLogNumber() > corrupted_log_number) {
+        ROCKS_LOG_ERROR(immutable_db_options_.info_log,
+                        "Column family inconsistency: SST file contains data"
+                        " beyond the point of corruption.");
+        return Status::Corruption("SST file is ahead of WALs");
+      }
+    }
+  }
 
   // True if there's any data in the WALs; if not, we can skip re-processing
   // them later
@@ -796,7 +835,7 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
       // not actually used. that is because VersionSet assumes
       // VersionSet::next_file_number_ always to be strictly greater than any
       // log number
-      versions_->MarkFileNumberUsedDuringRecovery(max_log_number + 1);
+      versions_->MarkFileNumberUsed(max_log_number + 1);
       status = versions_->LogAndApply(
           cfd, *cfd->GetLatestMutableCFOptions(), edit, &mutex_);
       if (!status.ok()) {
@@ -809,13 +848,13 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
   if (data_seen && !flushed) {
     // Mark these as alive so they'll be considered for deletion later by
     // FindObsoleteFiles()
-    if (concurrent_prepare_) {
+    if (two_write_queues_) {
       log_write_mutex_.Lock();
     }
     for (auto log_number : log_numbers) {
       alive_log_files_.push_back(LogFileNumberSize(log_number));
     }
-    if (concurrent_prepare_) {
+    if (two_write_queues_) {
       log_write_mutex_.Unlock();
     }
   }
@@ -857,26 +896,28 @@ Status DBImpl::WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd,
     const uint64_t current_time = static_cast<uint64_t>(_current_time);
 
     {
+      auto write_hint = cfd->CalculateSSTWriteHint(0);
       mutex_.Unlock();
 
       SequenceNumber earliest_write_conflict_snapshot;
       std::vector<SequenceNumber> snapshot_seqs =
           snapshots_.GetAll(&earliest_write_conflict_snapshot);
-
-      EnvOptions optimized_env_options =
-          env_->OptimizeForCompactionTableWrite(env_options_, immutable_db_options_);
+      auto snapshot_checker = snapshot_checker_.get();
+      if (use_custom_gc_ && snapshot_checker == nullptr) {
+        snapshot_checker = DisableGCSnapshotChecker::Instance();
+      }
       s = BuildTable(
           dbname_, env_, *cfd->ioptions(), mutable_cf_options,
-          optimized_env_options, cfd->table_cache(), iter.get(),
+          env_options_for_compaction_, cfd->table_cache(), iter.get(),
           std::unique_ptr<InternalIterator>(mem->NewRangeTombstoneIterator(ro)),
           &meta, cfd->internal_comparator(),
           cfd->int_tbl_prop_collector_factories(), cfd->GetID(), cfd->GetName(),
-          snapshot_seqs, earliest_write_conflict_snapshot,
+          snapshot_seqs, earliest_write_conflict_snapshot, snapshot_checker,
           GetCompressionFlush(*cfd->ioptions(), mutable_cf_options),
           cfd->ioptions()->compression_opts, paranoid_file_checks,
           cfd->internal_stats(), TableFileCreationReason::kRecovery,
           &event_logger_, job_id, Env::IO_HIGH, nullptr /* table_properties */,
-          -1 /* level */, current_time);
+          -1 /* level */, current_time, write_hint);
       LogFlush(immutable_db_options_.info_log);
       ROCKS_LOG_DEBUG(immutable_db_options_.info_log,
                       "[%s] [WriteLevel0TableForRecovery]"
@@ -929,6 +970,15 @@ Status DB::Open(const Options& options, const std::string& dbname, DB** dbptr) {
 Status DB::Open(const DBOptions& db_options, const std::string& dbname,
                 const std::vector<ColumnFamilyDescriptor>& column_families,
                 std::vector<ColumnFamilyHandle*>* handles, DB** dbptr) {
+  const bool seq_per_batch = true;
+  return DBImpl::Open(db_options, dbname, column_families, handles, dbptr,
+                      !seq_per_batch);
+}
+
+Status DBImpl::Open(const DBOptions& db_options, const std::string& dbname,
+                    const std::vector<ColumnFamilyDescriptor>& column_families,
+                    std::vector<ColumnFamilyHandle*>* handles, DB** dbptr,
+                    const bool seq_per_batch) {
   Status s = SanitizeOptionsByTable(db_options, column_families);
   if (!s.ok()) {
     return s;
@@ -948,7 +998,7 @@ Status DB::Open(const DBOptions& db_options, const std::string& dbname,
         std::max(max_write_buffer_size, cf.options.write_buffer_size);
   }
 
-  DBImpl* impl = new DBImpl(db_options, dbname);
+  DBImpl* impl = new DBImpl(db_options, dbname, seq_per_batch);
   s = impl->env_->CreateDirIfMissing(impl->immutable_db_options_.wal_dir);
   if (s.ok()) {
     for (auto db_path : impl->immutable_db_options_.db_paths) {
@@ -970,6 +1020,7 @@ Status DB::Open(const DBOptions& db_options, const std::string& dbname,
     return s;
   }
   impl->mutex_.Lock();
+  auto write_hint = impl->CalculateWALWriteHint();
   // Handles create_if_missing, error_if_exists
   s = impl->Recover(column_families);
   if (s.ok()) {
@@ -985,6 +1036,7 @@ Status DB::Open(const DBOptions& db_options, const std::string& dbname,
         LogFileName(impl->immutable_db_options_.wal_dir, new_log_number),
         &lfile, opt_env_options);
     if (s.ok()) {
+      lfile->SetWriteLifeTimeHint(write_hint);
       lfile->SetPreallocationBlockSize(
           impl->GetWalPreallocateBlockSize(max_write_buffer_size));
       {
@@ -1027,16 +1079,18 @@ Status DB::Open(const DBOptions& db_options, const std::string& dbname,
       }
     }
     if (s.ok()) {
+      SuperVersionContext sv_context(/* create_superversion */ true);
       for (auto cfd : *impl->versions_->GetColumnFamilySet()) {
-        delete impl->InstallSuperVersionAndScheduleWork(
-            cfd, nullptr, *cfd->GetLatestMutableCFOptions());
+        impl->InstallSuperVersionAndScheduleWork(
+            cfd, &sv_context, *cfd->GetLatestMutableCFOptions());
       }
-      if (impl->concurrent_prepare_) {
+      sv_context.Clean();
+      if (impl->two_write_queues_) {
         impl->log_write_mutex_.Lock();
       }
       impl->alive_log_files_.push_back(
           DBImpl::LogFileNumberSize(impl->logfile_number_));
-      if (impl->concurrent_prepare_) {
+      if (impl->two_write_queues_) {
         impl->log_write_mutex_.Unlock();
       }
       impl->DeleteObsoleteFiles();

db/db_impl_readonly.cc

@@ -57,14 +57,15 @@ Iterator* DBImplReadOnly::NewIterator(const ReadOptions& read_options,
   auto cfd = cfh->cfd();
   SuperVersion* super_version = cfd->GetSuperVersion()->Ref();
   SequenceNumber latest_snapshot = versions_->LastSequence();
+  ReadCallback* read_callback = nullptr;  // No read callback provided.
   auto db_iter = NewArenaWrappedDbIterator(
-      env_, read_options, *cfd->ioptions(), cfd->user_comparator(),
+      env_, read_options, *cfd->ioptions(),
       (read_options.snapshot != nullptr
            ? reinterpret_cast<const SnapshotImpl*>(read_options.snapshot)
                  ->number_
            : latest_snapshot),
       super_version->mutable_cf_options.max_sequential_skip_in_iterations,
-      super_version->version_number);
+      super_version->version_number, read_callback);
   auto internal_iter =
       NewInternalIterator(read_options, cfd, super_version, db_iter->GetArena(),
                           db_iter->GetRangeDelAggregator());
@@ -76,6 +77,7 @@ Status DBImplReadOnly::NewIterators(
     const ReadOptions& read_options,
     const std::vector<ColumnFamilyHandle*>& column_families,
     std::vector<Iterator*>* iterators) {
+  ReadCallback* read_callback = nullptr;  // No read callback provided.
   if (iterators == nullptr) {
     return Status::InvalidArgument("iterators not allowed to be nullptr");
   }
@@ -87,13 +89,13 @@ Status DBImplReadOnly::NewIterators(
     auto* cfd = reinterpret_cast<ColumnFamilyHandleImpl*>(cfh)->cfd();
     auto* sv = cfd->GetSuperVersion()->Ref();
     auto* db_iter = NewArenaWrappedDbIterator(
-        env_, read_options, *cfd->ioptions(), cfd->user_comparator(),
+        env_, read_options, *cfd->ioptions(),
         (read_options.snapshot != nullptr
              ? reinterpret_cast<const SnapshotImpl*>(read_options.snapshot)
                    ->number_
              : latest_snapshot),
         sv->mutable_cf_options.max_sequential_skip_in_iterations,
-        sv->version_number);
+        sv->version_number, read_callback);
     auto* internal_iter =
         NewInternalIterator(read_options, cfd, sv, db_iter->GetArena(),
                             db_iter->GetRangeDelAggregator());
@@ -140,6 +142,7 @@ Status DB::OpenForReadOnly(
   *dbptr = nullptr;
   handles->clear();
 
+  SuperVersionContext sv_context(/* create_superversion */ true);
   DBImplReadOnly* impl = new DBImplReadOnly(db_options, dbname);
   impl->mutex_.Lock();
   Status s = impl->Recover(column_families, true /* read only */,
@@ -158,10 +161,12 @@ Status DB::OpenForReadOnly(
   }
   if (s.ok()) {
     for (auto cfd : *impl->versions_->GetColumnFamilySet()) {
-      delete cfd->InstallSuperVersion(new SuperVersion(), &impl->mutex_);
+      sv_context.NewSuperVersion();
+      cfd->InstallSuperVersion(&sv_context, &impl->mutex_);
     }
   }
   impl->mutex_.Unlock();
+  sv_context.Clean();
   if (s.ok()) {
     *dbptr = impl;
     for (auto* h : *handles) {