table/block.h: change memset

Summary: In gcc-7 the following is an error identified by -Werror=class-memaccess In file included from ./table/get_context.h:14:0, from db/version_set.cc:43: ./table/block.h: In constructor ‘rocksdb::BlockReadAmpBitmap::BlockReadAmpBitmap(size_t, size_t, rocksdb::Statistics*)’: ./table/block.h:73:53: error: ‘void* memset(void*, int, size_t)’ clearing an object of type ‘struct std::atomic<unsigned int>’ with no trivial copy-assignment; use value-initialization instead [-Werror=class-memaccess] memset(bitmap_, 0, bitmap_size * kBytesPersEntry); ^ In file included from ./db/version_set.h:23:0, from db/version_set.cc:12: /toolchain/include/c++/8.0.0/atomic:684:12: note: ‘struct std::atomic<unsigned int>’ declared here struct atomic<unsigned int> : __atomic_base<unsigned int> ^~~~~~~~~~~~~~~~~~~~ As a solution the default initializer can be applied in list context. Signed-off-by: Daniel Black <daniel.black@au.ibm.com> Closes https://github.com/facebook/rocksdb/pull/2561 Differential Revision: D5398714 Pulled By: siying fbshipit-source-id: d883fb88ec7535eee60d551038fe91f14488be36
Fix GitHub issue #3716 : gcc-8 warnings
2026-11-06 11:02:36 -05:00 · 2019-10-30 14:00:55 -07:00 · 2019-10-30 13:54:49 -07:00 · 2019-10-30 13:54:49 -07:00 · 2019-10-30 13:54:49 -07:00 · 2017-11-28 12:25:25 -08:00
940 changed files with 28505 additions and 9114 deletions
@@ -31,6 +31,7 @@ build/
 ldb
 manifest_dump
 sst_dump
+blob_dump
 column_aware_encoding_exp
 util/build_version.cc
 build_tools/VALGRIND_LOGS/
@@ -41,6 +42,7 @@ package/
 .phutil_module_cache
 unity.a
 tags
+etags
 rocksdb_dump
 rocksdb_undump
 db_test2
@@ -59,7 +61,7 @@ rocksdb.h
 unity.cc
 java/crossbuild/.vagrant
 .vagrant/
-java/**.asc
+java/**/*.asc
 java/javadoc

 scan_build_report/
@@ -7,6 +7,7 @@ os:
 compiler:
  - clang
  - gcc
+osx_image: xcode8.3
 jdk:
  - oraclejdk7
 cache:
@@ -15,47 +16,55 @@ cache:

 addons:
   apt:
-      sources: ['ubuntu-toolchain-r-test', 'llvm-toolchain-precise-3.6']
-      packages: ['clang-3.6' , 'g++-6', 'zlib1g-dev', 'libbz2-dev', 'libsnappy-dev', 'curl', 'libgflags-dev']
+      packages: ['zlib1g-dev', 'libbz2-dev', 'libsnappy-dev', 'curl', 'libgflags-dev', 'mingw-w64']
 env:
-  - TEST_GROUP=platform_dependent1
-  - TEST_GROUP=platform_dependent2
-  - TEST_GROUP=platform_dependent3
-  - TEST_GROUP=1
-  - TEST_GROUP=2
+  - TEST_GROUP=platform_dependent # 16-18 minutes
+  - TEST_GROUP=1 # 33-35 minutes
+  - TEST_GROUP=2 # 30-32 minutes
  # Run java tests
-  - JOB_NAME=java_test
+  - JOB_NAME=java_test # 4-11 minutes
  # Build ROCKSDB_LITE
-  - JOB_NAME=lite_build
+  - JOB_NAME=lite_build # 3-4 minutes
  # Build examples
-  - JOB_NAME=examples
+  - JOB_NAME=examples # 5-7 minutes
+  - JOB_NAME=cmake # 3-5 minutes
+  - JOB_NAME=cmake-mingw # 3 minutes

 matrix:
  exclude:
-  - os: osx
-    compiler: gcc
  - os: osx
    env: TEST_GROUP=1
  - os: osx
    env: TEST_GROUP=2
+  - os : osx
+    env: JOB_NAME=cmake-mingw
+  - os : linux
+    compiler: clang
+  - os : osx
+    compiler: gcc
+
+# https://docs.travis-ci.com/user/caching/#ccache-cache
+install:
+  - if [ "${TRAVIS_OS_NAME}" == osx ]; then
+      brew install ccache;
+      PATH=$PATH:/usr/local/opt/ccache/libexec;
+    fi

 before_script:
-  - if [[ "${TRAVIS_OS_NAME}" == 'linux' && "${CXX}" == 'clang++' ]]; then CXX=clang++-3.6; fi
-  # test one linux g++ build with g++-6
-  - if [[ "${TRAVIS_OS_NAME}" == 'linux' && "${CXX}" == 'g++' && "${JOB_NAME}" == 'unittests' ]]; then CXX=g++-6; fi
-  # Limit the maximum number of open file descriptors to 8192
-  - ulimit -n 8192 || true
+  # Increase the maximum number of open file descriptors, since some tests use
+  # more FDs than the default limit.
+  - ulimit -n 8192

 script:
  - ${CXX} --version
-  - if [[ "${TEST_GROUP}" == 'platform_dependent1' ]]; then OPT=-DTRAVIS V=1 ROCKSDBTESTS_END=external_sst_file_basic_test make -j4 check_some; fi
-  - if [[ "${TEST_GROUP}" == 'platform_dependent2' ]]; then OPT=-DTRAVIS V=1 ROCKSDBTESTS_START=external_sst_file_basic_test ROCKSDBTESTS_END=checkpoint_test make -j4 check_some; fi
-  - if [[ "${TEST_GROUP}" == 'platform_dependent3' ]]; then OPT=-DTRAVIS V=1 ROCKSDBTESTS_START=checkpoint_test ROCKSDBTESTS_END=db_block_cache_test make -j4 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 [[ "${JOB_NAME}" == 'java_test' ]]; then OPT=-DTRAVIS V=1 make clean jclean rocksdbjava jtest; fi
-  - if [[ "${JOB_NAME}" == 'lite_build' ]]; then OPT="-DTRAVIS -DROCKSDB_LITE" V=1 make -j4 static_lib; fi
-  - if [[ "${JOB_NAME}" == 'examples' ]]; then OPT=-DTRAVIS V=1 make -j4 static_lib; cd examples; make -j4; fi
+  - if [ "${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 [ "${TEST_GROUP}" == '1' ]; then OPT=-DTRAVIS V=1 ROCKSDBTESTS_START=db_block_cache_test ROCKSDBTESTS_END=comparator_db_test make -j4 check_some; fi
+  - if [ "${TEST_GROUP}" == '2' ]; then OPT=-DTRAVIS V=1 ROCKSDBTESTS_START=comparator_db_test make -j4 check_some; fi
+  - if [ "${JOB_NAME}" == 'java_test' ]; then OPT=-DTRAVIS V=1 make clean jclean && make rocksdbjava jtest; fi
+  - if [ "${JOB_NAME}" == 'lite_build' ]; then OPT="-DTRAVIS -DROCKSDB_LITE" V=1 make -j4 static_lib; fi
+  - if [ "${JOB_NAME}" == 'examples' ]; then OPT=-DTRAVIS V=1 make -j4 static_lib; cd examples; make -j4; fi
+  - if [ "${JOB_NAME}" == 'cmake' ]; then mkdir build && cd build && cmake .. && make -j4 rocksdb; fi
+  - if [ "${JOB_NAME}" == 'cmake-mingw' ]; then mkdir build && cd build && cmake .. -DCMAKE_C_COMPILER=x86_64-w64-mingw32-gcc -DCMAKE_CXX_COMPILER=x86_64-w64-mingw32-g++ -DCMAKE_SYSTEM_NAME=Windows && make -j4 rocksdb; fi
 notifications:
    email:
      - leveldb@fb.com
@@ -55,7 +55,7 @@ else()
      add_definitions(-DROCKSDB_JEMALLOC -DJEMALLOC_NO_DEMANGLE)
      include_directories(${JEMALLOC_INCLUDE_DIR})
    endif()
-  endif() 
+  endif()
  option(WITH_SNAPPY "build with SNAPPY" OFF)
  if(WITH_SNAPPY)
    find_package(snappy REQUIRED)
@@ -66,8 +66,8 @@ else()
 endif()

 if(WIN32)
-  execute_process(COMMAND powershell -Command "Get-Date -format MM_dd_yyyy" OUTPUT_VARIABLE DATE)
-  execute_process(COMMAND powershell -Command "Get-Date -format HH:mm:ss" OUTPUT_VARIABLE TIME)
+  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} ${TIME}")
@@ -106,17 +106,58 @@ if(NOT WIN32)
  string(STRIP "${ROCKSDB_VERSION_MAJOR}" ROCKSDB_VERSION_MAJOR)
 endif()

-if(WIN32)
-  option(WITH_AVX2 "build with AVX2" ON)
-  if(WITH_AVX2)
-    if(MSVC)
-      set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /arch:AVX2")
-    else()
-      set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -mavx2")
-    endif()
+option(WITH_MD_LIBRARY "build with MD" ON)
+if(WIN32 AND MSVC)
+  if(WITH_MD_LIBRARY)
+    set(RUNTIME_LIBRARY "MD")
+  else()
+    set(RUNTIME_LIBRARY "MT")
  endif()
 endif()

+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 "${CMAKE_CXX_FLAGS} -msse4.2")
+  endif()
+else()
+  if(MSVC)
+    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /arch:AVX2")
+  else()
+    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -march=native")
+  endif()
+endif()
+
+set(CMAKE_REQUIRED_FLAGS ${CMAKE_CXX_FLAGS})
+include(CheckCXXSourceCompiles)
+CHECK_CXX_SOURCE_COMPILES("
+#include <cstdint>
+#include <nmmintrin.h>
+int main() {
+  volatile uint32_t x = _mm_crc32_u32(0, 0);
+}
+" HAVE_SSE42)
+if(HAVE_SSE42)
+  add_definitions(-DHAVE_SSE42)
+elseif(FORCE_SSE42)
+  message(FATAL_ERROR "FORCE_SSE42=ON but unable to compile with SSE4.2 enabled")
+endif()
+
+CHECK_CXX_SOURCE_COMPILES("
+#if defined(_MSC_VER) && !defined(__thread)
+#define __thread __declspec(thread)
+#endif
+int main() {
+  static __thread int tls;
+}
+" HAVE_THREAD_LOCAL)
+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 ${BUILD_VERSION_CC})
@@ -133,7 +174,8 @@ else()
  endif()
  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")
  if(NOT CMAKE_BUILD_TYPE STREQUAL "Debug")
-    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fno-omit-frame-pointer")
+    add_definitions(-DNDEBUG)
+    set(CMAKE_CXX_FLAGS "${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)
@@ -153,7 +195,6 @@ endif()

 option(WITH_ASAN "build with ASAN" OFF)
 if(WITH_ASAN)
-  add_definitions(-DROCKSDB_TSAN_RUN)
  set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -fsanitize=address")
  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fsanitize=address")
  set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -fsanitize=address")
@@ -165,7 +206,6 @@ endif()
 option(WITH_TSAN "build with TSAN" OFF)
 if(WITH_TSAN)
  set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -fsanitize=thread -pie")
-  add_definitions(-DROCKSDB_TSAN_RUN)
  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fsanitize=thread -fPIC")
  set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -fsanitize=thread -fPIC")
  if(WITH_JEMALLOC)
@@ -175,6 +215,7 @@ endif()

 option(WITH_UBSAN "build with UBSAN" OFF)
 if(WITH_UBSAN)
+  add_definitions(-DROCKSDB_UBSAN_RUN)
  set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} -fsanitize=undefined")
  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fsanitize=undefined")
  set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -fsanitize=undefined")
@@ -195,11 +236,11 @@ endif()
 if(MSVC)
  if((${OPTIMIZE_DEBUG} EQUAL 1))
    message(STATUS "Debug optimization is enabled")
-    set(CMAKE_CXX_FLAGS_DEBUG "/Oxt /MDd")
+    set(CMAKE_CXX_FLAGS_DEBUG "/Oxt /${RUNTIME_LIBRARY}d")
  else()
-    set(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} /Od /RTC1 /Gm /MDd")
+    set(CMAKE_CXX_FLAGS_DEBUG "${CMAKE_CXX_FLAGS_DEBUG} /Od /RTC1 /Gm /${RUNTIME_LIBRARY}d")
  endif()
-  set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} /Oxt /Zp8 /Gm- /Gy /MD")
+  set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} /Oxt /Zp8 /Gm- /Gy /${RUNTIME_LIBRARY}")

  set(CMAKE_SHARED_LINKER_FLAGS "${CMAKE_SHARED_LINKER_FLAGS} /DEBUG")
  set(CMAKE_EXE_LINKER_FLAGS "${CMAKE_EXE_LINKER_FLAGS} /DEBUG")
@@ -317,6 +358,7 @@ set(SOURCES
        db/internal_stats.cc
        db/log_reader.cc
        db/log_writer.cc
+        db/malloc_stats.cc
        db/managed_iterator.cc
        db/memtable.cc
        db/memtable_list.cc
@@ -338,14 +380,16 @@ set(SOURCES
        db/write_thread.cc
        env/env.cc
        env/env_chroot.cc
+        env/env_encryption.cc
        env/env_hdfs.cc
-        env/memenv.cc
+        env/mock_env.cc
+        memtable/alloc_tracker.cc
        memtable/hash_cuckoo_rep.cc
        memtable/hash_linklist_rep.cc
        memtable/hash_skiplist_rep.cc
-        memtable/memtable_allocator.cc
        memtable/skiplistrep.cc
        memtable/vectorrep.cc
+        memtable/write_buffer_manager.cc
        monitoring/histogram.cc
        monitoring/histogram_windowing.cc
        monitoring/instrumented_mutex.cc
@@ -432,12 +476,20 @@ set(SOURCES
        util/xxhash.cc
        utilities/backupable/backupable_db.cc
        utilities/blob_db/blob_db.cc
-        utilities/checkpoint/checkpoint.cc
+        utilities/blob_db/blob_db_impl.cc
+        utilities/blob_db/blob_dump_tool.cc
+        utilities/blob_db/blob_file.cc
+        utilities/blob_db/blob_log_reader.cc
+        utilities/blob_db/blob_log_writer.cc
+        utilities/blob_db/blob_log_format.cc
+        utilities/blob_db/ttl_extractor.cc
+        utilities/checkpoint/checkpoint_impl.cc
        utilities/col_buf_decoder.cc
        utilities/col_buf_encoder.cc
        utilities/column_aware_encoding_util.cc
        utilities/compaction_filters/remove_emptyvalue_compactionfilter.cc
        utilities/date_tiered/date_tiered_db_impl.cc
+        utilities/debug.cc
        utilities/document/document_db.cc
        utilities/document/json_document.cc
        utilities/document/json_document_builder.cc
@@ -449,6 +501,8 @@ 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
@@ -492,16 +546,14 @@ else()
    env/io_posix.cc)
 endif()

+set(ROCKSDB_STATIC_LIB rocksdb${ARTIFACT_SUFFIX})
+set(ROCKSDB_SHARED_LIB rocksdb-shared${ARTIFACT_SUFFIX})
+set(ROCKSDB_IMPORT_LIB ${ROCKSDB_SHARED_LIB})
 if(WIN32)
  set(SYSTEM_LIBS ${SYSTEM_LIBS} Shlwapi.lib Rpcrt4.lib)
-  set(ROCKSDB_STATIC_LIB rocksdblib${ARTIFACT_SUFFIX})
-  set(ROCKSDB_IMPORT_LIB rocksdb${ARTIFACT_SUFFIX})
  set(LIBS ${ROCKSDB_STATIC_LIB} ${THIRDPARTY_LIBS} ${SYSTEM_LIBS})
 else()
-  set(SYSTEM_LIBS ${CMAKE_THREAD_LIBS_INIT} rt)
-  set(ROCKSDB_STATIC_LIB rocksdb${ARTIFACT_SUFFIX})
-  set(ROCKSDB_SHARED_LIB rocksdb-shared)
-  set(ROCKSDB_IMPORT_LIB ${ROCKSDB_SHARED_LIB})
+  set(SYSTEM_LIBS ${CMAKE_THREAD_LIBS_INIT})
  set(LIBS ${ROCKSDB_SHARED_LIB} ${THIRDPARTY_LIBS} ${SYSTEM_LIBS})

  add_library(${ROCKSDB_SHARED_LIB} SHARED ${SOURCES})
@@ -578,12 +630,14 @@ set(TESTS
        db/db_properties_test.cc
        db/db_range_del_test.cc
        db/db_sst_test.cc
+        db/db_statistics_test.cc
        db/db_table_properties_test.cc
        db/db_tailing_iter_test.cc
        db/db_test.cc
        db/db_test2.cc
        db/db_universal_compaction_test.cc
        db/db_wal_test.cc
+        db/db_write_test.cc
        db/dbformat_test.cc
        db/deletefile_test.cc
        db/external_sst_file_basic_test.cc
@@ -616,6 +670,7 @@ set(TESTS
        env/mock_env_test.cc
        memtable/inlineskiplist_test.cc
        memtable/skiplist_test.cc
+        memtable/write_buffer_manager_test.cc
        monitoring/histogram_test.cc
        monitoring/iostats_context_test.cc
        monitoring/statistics_test.cc
@@ -642,9 +697,11 @@ set(TESTS
        util/event_logger_test.cc
        util/file_reader_writer_test.cc
        util/filelock_test.cc
+        util/hash_test.cc
        util/heap_test.cc
        util/rate_limiter_test.cc
        util/slice_transform_test.cc
+        util/timer_queue_test.cc
        util/thread_list_test.cc
        util/thread_local_test.cc
        utilities/backupable/backupable_db_test.cc
@@ -657,6 +714,10 @@ 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
@@ -694,10 +755,10 @@ endforeach(sourcefile ${BENCHMARKS})
 # and linked to tests. Add test only code that is not #ifdefed for Release here.
 set(TESTUTIL_SOURCE
    db/db_test_util.cc
-    env/mock_env.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()
@@ -12,8 +12,3 @@ Complete your CLA here: <https://code.facebook.com/cla>

 If you prefer to sign a paper copy, we can send you a PDF.  Send us an 
 e-mail or create a new github issue to request the CLA in PDF format.
-
-## License
-
-By contributing to RocksDB, you agree that your contributions will be
-licensed under the [BSD License](LICENSE).
@@ -0,0 +1,339 @@
+                    GNU GENERAL PUBLIC LICENSE
+                       Version 2, June 1991
+
+ Copyright (C) 1989, 1991 Free Software Foundation, Inc.,
+ 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ Everyone is permitted to copy and distribute verbatim copies
+ of this license document, but changing it is not allowed.
+
+                            Preamble
+
+  The licenses for most software are designed to take away your
+freedom to share and change it.  By contrast, the GNU General Public
+License is intended to guarantee your freedom to share and change free
+software--to make sure the software is free for all its users.  This
+General Public License applies to most of the Free Software
+Foundation's software and to any other program whose authors commit to
+using it.  (Some other Free Software Foundation software is covered by
+the GNU Lesser General Public License instead.)  You can apply it to
+your programs, too.
+
+  When we speak of free software, we are referring to freedom, not
+price.  Our General Public Licenses are designed to make sure that you
+have the freedom to distribute copies of free software (and charge for
+this service if you wish), that you receive source code or can get it
+if you want it, that you can change the software or use pieces of it
+in new free programs; and that you know you can do these things.
+
+  To protect your rights, we need to make restrictions that forbid
+anyone to deny you these rights or to ask you to surrender the rights.
+These restrictions translate to certain responsibilities for you if you
+distribute copies of the software, or if you modify it.
+
+  For example, if you distribute copies of such a program, whether
+gratis or for a fee, you must give the recipients all the rights that
+you have.  You must make sure that they, too, receive or can get the
+source code.  And you must show them these terms so they know their
+rights.
+
+  We protect your rights with two steps: (1) copyright the software, and
+(2) offer you this license which gives you legal permission to copy,
+distribute and/or modify the software.
+
+  Also, for each author's protection and ours, we want to make certain
+that everyone understands that there is no warranty for this free
+software.  If the software is modified by someone else and passed on, we
+want its recipients to know that what they have is not the original, so
+that any problems introduced by others will not reflect on the original
+authors' reputations.
+
+  Finally, any free program is threatened constantly by software
+patents.  We wish to avoid the danger that redistributors of a free
+program will individually obtain patent licenses, in effect making the
+program proprietary.  To prevent this, we have made it clear that any
+patent must be licensed for everyone's free use or not licensed at all.
+
+  The precise terms and conditions for copying, distribution and
+modification follow.
+
+                    GNU GENERAL PUBLIC LICENSE
+   TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
+
+  0. This License applies to any program or other work which contains
+a notice placed by the copyright holder saying it may be distributed
+under the terms of this General Public License.  The "Program", below,
+refers to any such program or work, and a "work based on the Program"
+means either the Program or any derivative work under copyright law:
+that is to say, a work containing the Program or a portion of it,
+either verbatim or with modifications and/or translated into another
+language.  (Hereinafter, translation is included without limitation in
+the term "modification".)  Each licensee is addressed as "you".
+
+Activities other than copying, distribution and modification are not
+covered by this License; they are outside its scope.  The act of
+running the Program is not restricted, and the output from the Program
+is covered only if its contents constitute a work based on the
+Program (independent of having been made by running the Program).
+Whether that is true depends on what the Program does.
+
+  1. You may copy and distribute verbatim copies of the Program's
+source code as you receive it, in any medium, provided that you
+conspicuously and appropriately publish on each copy an appropriate
+copyright notice and disclaimer of warranty; keep intact all the
+notices that refer to this License and to the absence of any warranty;
+and give any other recipients of the Program a copy of this License
+along with the Program.
+
+You may charge a fee for the physical act of transferring a copy, and
+you may at your option offer warranty protection in exchange for a fee.
+
+  2. You may modify your copy or copies of the Program or any portion
+of it, thus forming a work based on the Program, and copy and
+distribute such modifications or work under the terms of Section 1
+above, provided that you also meet all of these conditions:
+
+    a) You must cause the modified files to carry prominent notices
+    stating that you changed the files and the date of any change.
+
+    b) You must cause any work that you distribute or publish, that in
+    whole or in part contains or is derived from the Program or any
+    part thereof, to be licensed as a whole at no charge to all third
+    parties under the terms of this License.
+
+    c) If the modified program normally reads commands interactively
+    when run, you must cause it, when started running for such
+    interactive use in the most ordinary way, to print or display an
+    announcement including an appropriate copyright notice and a
+    notice that there is no warranty (or else, saying that you provide
+    a warranty) and that users may redistribute the program under
+    these conditions, and telling the user how to view a copy of this
+    License.  (Exception: if the Program itself is interactive but
+    does not normally print such an announcement, your work based on
+    the Program is not required to print an announcement.)
+
+These requirements apply to the modified work as a whole.  If
+identifiable sections of that work are not derived from the Program,
+and can be reasonably considered independent and separate works in
+themselves, then this License, and its terms, do not apply to those
+sections when you distribute them as separate works.  But when you
+distribute the same sections as part of a whole which is a work based
+on the Program, the distribution of the whole must be on the terms of
+this License, whose permissions for other licensees extend to the
+entire whole, and thus to each and every part regardless of who wrote it.
+
+Thus, it is not the intent of this section to claim rights or contest
+your rights to work written entirely by you; rather, the intent is to
+exercise the right to control the distribution of derivative or
+collective works based on the Program.
+
+In addition, mere aggregation of another work not based on the Program
+with the Program (or with a work based on the Program) on a volume of
+a storage or distribution medium does not bring the other work under
+the scope of this License.
+
+  3. You may copy and distribute the Program (or a work based on it,
+under Section 2) in object code or executable form under the terms of
+Sections 1 and 2 above provided that you also do one of the following:
+
+    a) Accompany it with the complete corresponding machine-readable
+    source code, which must be distributed under the terms of Sections
+    1 and 2 above on a medium customarily used for software interchange; or,
+
+    b) Accompany it with a written offer, valid for at least three
+    years, to give any third party, for a charge no more than your
+    cost of physically performing source distribution, a complete
+    machine-readable copy of the corresponding source code, to be
+    distributed under the terms of Sections 1 and 2 above on a medium
+    customarily used for software interchange; or,
+
+    c) Accompany it with the information you received as to the offer
+    to distribute corresponding source code.  (This alternative is
+    allowed only for noncommercial distribution and only if you
+    received the program in object code or executable form with such
+    an offer, in accord with Subsection b above.)
+
+The source code for a work means the preferred form of the work for
+making modifications to it.  For an executable work, complete source
+code means all the source code for all modules it contains, plus any
+associated interface definition files, plus the scripts used to
+control compilation and installation of the executable.  However, as a
+special exception, the source code distributed need not include
+anything that is normally distributed (in either source or binary
+form) with the major components (compiler, kernel, and so on) of the
+operating system on which the executable runs, unless that component
+itself accompanies the executable.
+
+If distribution of executable or object code is made by offering
+access to copy from a designated place, then offering equivalent
+access to copy the source code from the same place counts as
+distribution of the source code, even though third parties are not
+compelled to copy the source along with the object code.
+
+  4. You may not copy, modify, sublicense, or distribute the Program
+except as expressly provided under this License.  Any attempt
+otherwise to copy, modify, sublicense or distribute the Program is
+void, and will automatically terminate your rights under this License.
+However, parties who have received copies, or rights, from you under
+this License will not have their licenses terminated so long as such
+parties remain in full compliance.
+
+  5. You are not required to accept this License, since you have not
+signed it.  However, nothing else grants you permission to modify or
+distribute the Program or its derivative works.  These actions are
+prohibited by law if you do not accept this License.  Therefore, by
+modifying or distributing the Program (or any work based on the
+Program), you indicate your acceptance of this License to do so, and
+all its terms and conditions for copying, distributing or modifying
+the Program or works based on it.
+
+  6. Each time you redistribute the Program (or any work based on the
+Program), the recipient automatically receives a license from the
+original licensor to copy, distribute or modify the Program subject to
+these terms and conditions.  You may not impose any further
+restrictions on the recipients' exercise of the rights granted herein.
+You are not responsible for enforcing compliance by third parties to
+this License.
+
+  7. If, as a consequence of a court judgment or allegation of patent
+infringement or for any other reason (not limited to patent issues),
+conditions are imposed on you (whether by court order, agreement or
+otherwise) that contradict the conditions of this License, they do not
+excuse you from the conditions of this License.  If you cannot
+distribute so as to satisfy simultaneously your obligations under this
+License and any other pertinent obligations, then as a consequence you
+may not distribute the Program at all.  For example, if a patent
+license would not permit royalty-free redistribution of the Program by
+all those who receive copies directly or indirectly through you, then
+the only way you could satisfy both it and this License would be to
+refrain entirely from distribution of the Program.
+
+If any portion of this section is held invalid or unenforceable under
+any particular circumstance, the balance of the section is intended to
+apply and the section as a whole is intended to apply in other
+circumstances.
+
+It is not the purpose of this section to induce you to infringe any
+patents or other property right claims or to contest validity of any
+such claims; this section has the sole purpose of protecting the
+integrity of the free software distribution system, which is
+implemented by public license practices.  Many people have made
+generous contributions to the wide range of software distributed
+through that system in reliance on consistent application of that
+system; it is up to the author/donor to decide if he or she is willing
+to distribute software through any other system and a licensee cannot
+impose that choice.
+
+This section is intended to make thoroughly clear what is believed to
+be a consequence of the rest of this License.
+
+  8. If the distribution and/or use of the Program is restricted in
+certain countries either by patents or by copyrighted interfaces, the
+original copyright holder who places the Program under this License
+may add an explicit geographical distribution limitation excluding
+those countries, so that distribution is permitted only in or among
+countries not thus excluded.  In such case, this License incorporates
+the limitation as if written in the body of this License.
+
+  9. The Free Software Foundation may publish revised and/or new versions
+of the General Public License from time to time.  Such new versions will
+be similar in spirit to the present version, but may differ in detail to
+address new problems or concerns.
+
+Each version is given a distinguishing version number.  If the Program
+specifies a version number of this License which applies to it and "any
+later version", you have the option of following the terms and conditions
+either of that version or of any later version published by the Free
+Software Foundation.  If the Program does not specify a version number of
+this License, you may choose any version ever published by the Free Software
+Foundation.
+
+  10. If you wish to incorporate parts of the Program into other free
+programs whose distribution conditions are different, write to the author
+to ask for permission.  For software which is copyrighted by the Free
+Software Foundation, write to the Free Software Foundation; we sometimes
+make exceptions for this.  Our decision will be guided by the two goals
+of preserving the free status of all derivatives of our free software and
+of promoting the sharing and reuse of software generally.
+
+                            NO WARRANTY
+
+  11. BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
+FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW.  EXCEPT WHEN
+OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
+PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
+OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.  THE ENTIRE RISK AS
+TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU.  SHOULD THE
+PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
+REPAIR OR CORRECTION.
+
+  12. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
+WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
+REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
+INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
+OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
+TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
+YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
+PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
+POSSIBILITY OF SUCH DAMAGES.
+
+                     END OF TERMS AND CONDITIONS
+
+            How to Apply These Terms to Your New Programs
+
+  If you develop a new program, and you want it to be of the greatest
+possible use to the public, the best way to achieve this is to make it
+free software which everyone can redistribute and change under these terms.
+
+  To do so, attach the following notices to the program.  It is safest
+to attach them to the start of each source file to most effectively
+convey the exclusion of warranty; and each file should have at least
+the "copyright" line and a pointer to where the full notice is found.
+
+    <one line to give the program's name and a brief idea of what it does.>
+    Copyright (C) <year>  <name of author>
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    (at your option) any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+
+    You should have received a copy of the GNU General Public License along
+    with this program; if not, write to the Free Software Foundation, Inc.,
+    51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
+
+Also add information on how to contact you by electronic and paper mail.
+
+If the program is interactive, make it output a short notice like this
+when it starts in an interactive mode:
+
+    Gnomovision version 69, Copyright (C) year name of author
+    Gnomovision comes with ABSOLUTELY NO WARRANTY; for details type `show w'.
+    This is free software, and you are welcome to redistribute it
+    under certain conditions; type `show c' for details.
+
+The hypothetical commands `show w' and `show c' should show the appropriate
+parts of the General Public License.  Of course, the commands you use may
+be called something other than `show w' and `show c'; they could even be
+mouse-clicks or menu items--whatever suits your program.
+
+You should also get your employer (if you work as a programmer) or your
+school, if any, to sign a "copyright disclaimer" for the program, if
+necessary.  Here is a sample; alter the names:
+
+  Yoyodyne, Inc., hereby disclaims all copyright interest in the program
+  `Gnomovision' (which makes passes at compilers) written by James Hacker.
+
+  <signature of Ty Coon>, 1 April 1989
+  Ty Coon, President of Vice
+
+This General Public License does not permit incorporating your program into
+proprietary programs.  If your program is a subroutine library, you may
+consider it more useful to permit linking proprietary applications with the
+library.  If this is what you want to do, use the GNU Lesser General
+Public License instead of this License.
@@ -1,5 +1,8 @@
 # RocksDB default options change log
 ## Unreleased
+* delayed_write_rate takes the rate given by rate_limiter if not specified.
+
+## 5.2
 * Change the default of delayed slowdown value to 16MB/s and further increase the L0 stop condition to 36 files.

 ## 5.0 (11/17/2016)
@@ -1,13 +1,85 @@
 # Rocksdb Change Log
-## Unreleased
+## 5.7.5 (11/28/2017)
+### Bug Fixes
+* Fix IOError on WAL write doesn't propagate to write group follower
+
+## 5.7.4 (08/31/2017)
+No significant changes.
+
+## 5.7.3 (08/29/2017)
+### Bug Fixes
+* 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.
+
+### New Features
+* Measure estimated number of reads per file. The information can be accessed through DB::GetColumnFamilyMetaData or "rocksdb.sstables" DB property.
+* RateLimiter support for throttling background reads, or throttling the sum of background reads and writes. This can give more predictable I/O usage when compaction reads more data than it writes, e.g., due to lots of deletions.
+* [Experimental] FIFO compaction with TTL support. It can be enabled by setting CompactionOptionsFIFO.ttl > 0.
+* Introduce `EventListener::OnBackgroundError()` callback. Users can implement it to be notified of errors causing the DB to enter read-only mode, and optionally override them.
+* Partitioned Index/Filters exiting the experimental mode. To enable partitioned indexes set index_type to kTwoLevelIndexSearch and to further enable partitioned filters set partition_filters to true. To configure the partition size set metadata_block_size.
+
+
+### Bug Fixes
+* Fix discarding empty compaction output files when `DeleteRange()` is used together with subcompactions.
+
+## 5.6.0 (06/06/2017)
+### Public API Change
+* Scheduling flushes and compactions in the same thread pool is no longer supported by setting `max_background_flushes=0`. Instead, users can achieve this by configuring their high-pri thread pool to have zero threads.
+* Replace `Options::max_background_flushes`, `Options::max_background_compactions`, and `Options::base_background_compactions` all with `Options::max_background_jobs`, which automatically decides how many threads to allocate towards flush/compaction.
+* options.delayed_write_rate by default take the value of options.rate_limiter rate.
+* Replace global variable `IOStatsContext iostats_context` with `IOStatsContext* get_iostats_context()`; replace global variable `PerfContext perf_context` with `PerfContext* get_perf_context()`.
+
+### New Features
+* Change ticker/histogram statistics implementations to use core-local storage. This improves aggregation speed compared to our previous thread-local approach, particularly for applications with many threads.
+* Users can pass a cache object to write buffer manager, so that they can cap memory usage for memtable and block cache using one single limit.
+* Flush will be triggered when 7/8 of the limit introduced by write_buffer_manager or db_write_buffer_size is triggered, so that the hard threshold is hard to hit.
+* Introduce WriteOptions.low_pri. If it is true, low priority writes will be throttled if the compaction is behind.
+* `DB::IngestExternalFile()` now supports ingesting files into a database containing range deletions.
+
+### Bug Fixes
+* Shouldn't ignore return value of fsync() in flush.
+
+## 5.5.0 (05/17/2017)
+### New Features
+* FIFO compaction to support Intra L0 compaction too with CompactionOptionsFIFO.allow_compaction=true.
+* DB::ResetStats() to reset internal stats.
+* Statistics::Reset() to reset user stats.
+* ldb add option --try_load_options, which will open DB with its own option file.
+* Introduce WriteBatch::PopSavePoint to pop the most recent save point explicitly.
+* Support dynamically change `max_open_files` option via SetDBOptions()
+* Added DB::CreateColumnFamilie() and DB::DropColumnFamilies() to bulk create/drop column families.
+* Add debugging function `GetAllKeyVersions` to see internal versions of a range of keys.
+* Support file ingestion with universal compaction style
+* Support file ingestion behind with option `allow_ingest_behind`
+* New option enable_pipelined_write which may improve write throughput in case writing from multiple threads and WAL enabled.
+
+### Bug Fixes
+* Fix the bug that Direct I/O uses direct reads for non-SST file

 ## 5.4.0 (04/11/2017)
 ### Public API Change
+* random_access_max_buffer_size no longer has any effect
+* Removed Env::EnableReadAhead(), Env::ShouldForwardRawRequest()
 * Support dynamically change `stats_dump_period_sec` option via SetDBOptions().
 * Added ReadOptions::max_skippable_internal_keys to set a threshold to fail a request as incomplete when too many keys are being skipped when using iterators.
 * DB::Get in place of std::string accepts PinnableSlice, which avoids the extra memcpy of value to std::string in most of cases.
    * PinnableSlice releases the pinned resources that contain the value when it is destructed or when ::Reset() is called on it.
    * The old API that accepts std::string, although discouraged, is still supported.
+* Replace Options::use_direct_writes with Options::use_direct_io_for_flush_and_compaction. Read Direct IO wiki for details.
+* Added CompactionEventListener and EventListener::OnFlushBegin interfaces.

 ### New Features
 * Memtable flush can be avoided during checkpoint creation if total log file size is smaller than a threshold specified by the user.
@@ -239,7 +311,7 @@
 * Added a new way to report QPS from db_bench (check out --report_file and --report_interval_seconds)
 * Added a cache for individual rows. See DBOptions::row_cache for more info.
 * Several new features on EventListener (see include/rocksdb/listener.h):
- - OnCompationCompleted() now returns per-compaciton job statistics, defined in include/rocksdb/compaction_job_stats.h.
+ - OnCompationCompleted() now returns per-compaction job statistics, defined in include/rocksdb/compaction_job_stats.h.
 - Added OnTableFileCreated() and OnTableFileDeleted().
 * Add compaction_options_universal.enable_trivial_move to true, to allow trivial move while performing universal compaction. Trivial move will happen only when all the input files are non overlapping.

@@ -21,15 +21,18 @@ depend on gflags. You will need to have gflags installed to run `make all`. This
 use binaries compiled by `make all` in production.

 * By default the binary we produce is optimized for the platform you're compiling on
-(-march=native or the equivalent). If you want to build a portable binary, add 'PORTABLE=1' before
-your make commands, like this: `PORTABLE=1 make static_lib`. If you want to build a binary that
-makes use of SSE4, add 'USE_SSE=1' before your make commands, like this: `USE_SSE=1 make static_lib`.
+(`-march=native` or the equivalent). SSE4.2 will thus be enabled automatically if your
+CPU supports it. To print a warning if your CPU does not support SSE4.2, build with
+`USE_SSE=1 make static_lib` or, if using CMake, `cmake -DFORCE_SSE42=ON`. If you want
+to build a portable binary, add `PORTABLE=1` before your make commands, like this:
+`PORTABLE=1 make static_lib`.

 ## Dependencies

 * You can link RocksDB with following compression libraries:
  - [zlib](http://www.zlib.net/) - a library for data compression.
  - [bzip2](http://www.bzip.org/) - a library for data compression.
+  - [lz4](https://github.com/lz4/lz4) - a library for extremely fast data compression.
  - [snappy](http://google.github.io/snappy/) - a library for fast
      data compression.
  - [zstandard](http://www.zstd.net) - Fast real-time compression
@@ -51,6 +54,7 @@ makes use of SSE4, add 'USE_SSE=1' before your make commands, like this: `USE_SS
      `sudo apt-get install libsnappy-dev`.
    * Install zlib. Try: `sudo apt-get install zlib1g-dev`.
    * Install bzip2: `sudo apt-get install libbz2-dev`.
+    * Install lz4: `sudo apt-get install liblz4-dev`.
    * Install zstandard: `sudo apt-get install libzstd-dev`.

 * **Linux - CentOS / RHEL**
@@ -63,26 +67,29 @@ makes use of SSE4, add 'USE_SSE=1' before your make commands, like this: `USE_SS
              git checkout v2.0
              ./configure && make && sudo make install

-      **Notice**: Once installed, please add the include path for gflags to your CPATH env var and the
-      lib path to LIBRARY_PATH. If installed with default settings, the lib will be /usr/local/lib
-      and the include path will be /usr/local/include.
+      **Notice**: Once installed, please add the include path for gflags to your `CPATH` environment variable and the
+      lib path to `LIBRARY_PATH`. If installed with default settings, the include path will be `/usr/local/include`
+      and the lib path will be `/usr/local/lib`.

    * Install snappy:

-              wget https://github.com/google/snappy/releases/download/1.1.4/snappy-1.1.4.tar.gz
-              tar -xzvf snappy-1.1.4.tar.gz
-              cd snappy-1.1.4
-              ./configure && make && sudo make install
+              sudo yum install snappy snappy-devel

    * Install zlib:

-              sudo yum install zlib
-              sudo yum install zlib-devel
+              sudo yum install zlib zlib-devel

    * Install bzip2:

-              sudo yum install bzip2
-              sudo yum install bzip2-devel
+              sudo yum install bzip2 bzip2-devel
+
+    * Install lz4:
+
+              sudo yum install lz4-devel
+
+    * Install ASAN (optional for debugging):
+
+              sudo yum install libasan

    * Install zstandard:

@@ -106,3 +113,30 @@ makes use of SSE4, add 'USE_SSE=1' before your make commands, like this: `USE_SS
 * **Windows**:
  * For building with MS Visual Studio 13 you will need Update 4 installed.
  * Read and follow the instructions at CMakeLists.txt
+
+* **AIX 6.1**
+    * Install AIX Toolbox rpms with gcc
+    * Use these environment variables:
+  
+             export PORTABLE=1
+             export CC=gcc
+             export AR="ar -X64"
+             export EXTRA_ARFLAGS=-X64
+             export EXTRA_CFLAGS=-maix64
+             export EXTRA_CXXFLAGS=-maix64
+             export PLATFORM_LDFLAGS="-static-libstdc++ -static-libgcc"
+             export LIBPATH=/opt/freeware/lib
+             export JAVA_HOME=/usr/java8_64
+             export PATH=/opt/freeware/bin:$PATH
+  
+* **Solaris Sparc**
+    * Install GCC 4.8.2 and higher.
+    * Use these environment variables:
+
+             export CC=gcc
+             export EXTRA_CFLAGS=-m64
+             export EXTRA_CXXFLAGS=-m64
+             export EXTRA_LDFLAGS=-m64
+             export PORTABLE=1
+             export PLATFORM_LDFLAGS="-static-libstdc++ -static-libgcc"
+
@@ -0,0 +1,202 @@
+
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.
@@ -1,10 +1,4 @@
-BSD License
-
-For rocksdb software
-
-Copyright (c) 2011-present, Facebook, Inc.
-All rights reserved.
---------------------------------------------------------------------
+This contains code that is from LevelDB, and that code is under the following license:

 Copyright (c) 2011 The LevelDB Authors. All rights reserved.

@@ -14,7 +14,8 @@ CFLAGS += ${EXTRA_CFLAGS}
 CXXFLAGS += ${EXTRA_CXXFLAGS}
 LDFLAGS += $(EXTRA_LDFLAGS)
 MACHINE ?= $(shell uname -m)
-ARFLAGS = rs
+ARFLAGS = ${EXTRA_ARFLAGS} rs
+STRIPFLAGS = -S -x

 # Transform parallel LOG output into something more readable.
 perl_command = perl -n \
@@ -154,7 +155,9 @@ missing_make_config_paths := $(shell				\
 $(foreach path, $(missing_make_config_paths), \
 	$(warning Warning: $(path) dont exist))

-ifneq ($(PLATFORM), IOS)
+ifeq ($(PLATFORM), OS_AIX)
+# no debug info
+else ifneq ($(PLATFORM), IOS)
 CFLAGS += -g
 CXXFLAGS += -g
 else
@@ -162,6 +165,11 @@ else
 OPT += -DNDEBUG
 endif

+ifeq ($(PLATFORM), OS_AIX)
+ARFLAGS = -X64 rs
+STRIPFLAGS = -X64 -x
+endif
+
 ifeq ($(PLATFORM), OS_SOLARIS)
 	PLATFORM_CXXFLAGS += -D _GLIBCXX_USE_C99
 endif
@@ -184,9 +192,9 @@ endif
 # TSAN doesn't work well with jemalloc. If we're compiling with TSAN, we should use regular malloc.
 ifdef COMPILE_WITH_TSAN
 	DISABLE_JEMALLOC=1
-	EXEC_LDFLAGS += -fsanitize=thread -pie
-	PLATFORM_CCFLAGS += -fsanitize=thread -fPIC -DROCKSDB_TSAN_RUN
-	PLATFORM_CXXFLAGS += -fsanitize=thread -fPIC -DROCKSDB_TSAN_RUN
+	EXEC_LDFLAGS += -fsanitize=thread
+	PLATFORM_CCFLAGS += -fsanitize=thread -fPIC
+	PLATFORM_CXXFLAGS += -fsanitize=thread -fPIC
        # Turn off -pg when enabling TSAN testing, because that induces
        # a link failure.  TODO: find the root cause
 	PROFILING_FLAGS =
@@ -194,12 +202,17 @@ ifdef COMPILE_WITH_TSAN
 	LUA_PATH =
 endif

+# AIX doesn't work with -pg
+ifeq ($(PLATFORM), OS_AIX)
+	PROFILING_FLAGS =
+endif
+
 # USAN doesn't work well with jemalloc. If we're compiling with USAN, we should use regular malloc.
 ifdef COMPILE_WITH_UBSAN
 	DISABLE_JEMALLOC=1
 	EXEC_LDFLAGS += -fsanitize=undefined
-	PLATFORM_CCFLAGS += -fsanitize=undefined
-	PLATFORM_CXXFLAGS += -fsanitize=undefined
+	PLATFORM_CCFLAGS += -fsanitize=undefined -DROCKSDB_UBSAN_RUN
+	PLATFORM_CXXFLAGS += -fsanitize=undefined -DROCKSDB_UBSAN_RUN
 endif

 ifndef DISABLE_JEMALLOC
@@ -215,8 +228,14 @@ endif
 export GTEST_THROW_ON_FAILURE=1
 export GTEST_HAS_EXCEPTIONS=1
 GTEST_DIR = ./third-party/gtest-1.7.0/fused-src
-PLATFORM_CCFLAGS += -isystem $(GTEST_DIR)
-PLATFORM_CXXFLAGS += -isystem $(GTEST_DIR)
+# AIX: pre-defined system headers are surrounded by an extern "C" block
+ifeq ($(PLATFORM), OS_AIX)
+	PLATFORM_CCFLAGS += -I$(GTEST_DIR)
+	PLATFORM_CXXFLAGS += -I$(GTEST_DIR)
+else
+	PLATFORM_CCFLAGS += -isystem $(GTEST_DIR)
+	PLATFORM_CXXFLAGS += -isystem $(GTEST_DIR)
+endif

 # This (the first rule) must depend on "all".
 default: all
@@ -261,6 +280,9 @@ CXXFLAGS += $(WARNING_FLAGS) -I. -I./include $(PLATFORM_CXXFLAGS) $(OPT) -Woverl

 LDFLAGS += $(PLATFORM_LDFLAGS)

+# If NO_UPDATE_BUILD_VERSION is set we don't update util/build_version.cc, but
+# the file needs to already exist or else the build will fail
+ifndef NO_UPDATE_BUILD_VERSION
 date := $(shell date +%F)
 ifdef FORCE_GIT_SHA
 	git_sha := $(FORCE_GIT_SHA)
@@ -274,7 +296,6 @@ gen_build_version = sed -e s/@@GIT_SHA@@/$(git_sha)/ -e s/@@GIT_DATE_TIME@@/$(da
 # as a regular source file as part of the compilation process.
 # One can run "strings executable_filename | grep _build_" to find
 # the version of the source that we used to build the executable file.
-CLEAN_FILES += util/build_version.cc:
 FORCE:
 util/build_version.cc: FORCE
 	$(AM_V_GEN)rm -f $@-t
@@ -282,6 +303,7 @@ util/build_version.cc: FORCE
 	$(AM_V_at)if test -f $@; then					\
 	  cmp -s $@-t $@ && rm -f $@-t || mv -f $@-t $@;		\
 	else mv -f $@-t $@; fi
+endif

 LIBOBJECTS = $(LIB_SOURCES:.cc=.o)
 LIBOBJECTS += $(TOOL_LIB_SOURCES:.cc=.o)
@@ -301,6 +323,7 @@ EXPOBJECTS = $(EXP_LIB_SOURCES:.cc=.o) $(LIBOBJECTS) $(TESTUTIL)

 TESTS = \
 	db_basic_test \
+	db_encryption_test \
 	db_test2 \
 	external_sst_file_basic_test \
 	auto_roll_logger_test \
@@ -313,6 +336,7 @@ TESTS = \
 	inlineskiplist_test \
 	env_basic_test \
 	env_test \
+	hash_test \
 	thread_local_test \
 	rate_limiter_test \
 	perf_context_test \
@@ -339,7 +363,10 @@ TESTS = \
 	db_io_failure_test \
 	db_properties_test \
 	db_table_properties_test \
+	db_statistics_test \
+	db_write_test \
 	autovector_test \
+	blob_db_test \
 	cleanable_test \
 	column_family_test \
 	table_properties_collector_test \
@@ -375,11 +402,15 @@ TESTS = \
 	external_sst_file_test \
 	prefix_test \
 	skiplist_test \
+	write_buffer_manager_test \
 	stringappend_test \
+	cassandra_format_test \
+	cassandra_merge_test \
+	cassandra_row_merge_test \
+	cassandra_serialize_test \
 	ttl_test \
 	date_tiered_test \
 	backupable_db_test \
-	blob_db_test \
 	document_db_test \
 	json_document_test \
 	sim_cache_test \
@@ -400,6 +431,7 @@ TESTS = \
 	options_settable_test \
 	options_util_test \
 	event_logger_test \
+	timer_queue_test \
 	cuckoo_table_builder_test \
 	cuckoo_table_reader_test \
 	cuckoo_table_db_test \
@@ -462,7 +494,8 @@ TOOLS = \
 	ldb \
 	db_repl_stress \
 	rocksdb_dump \
-	rocksdb_undump
+	rocksdb_undump \
+	blob_dump \

 TEST_LIBS = \
 	librocksdb_env_basic_test.a
@@ -523,9 +556,14 @@ $(SHARED3): $(SHARED4)
 	ln -fs $(SHARED4) $(SHARED3)
 endif

-$(SHARED4):
-	$(CXX) $(PLATFORM_SHARED_LDFLAGS)$(SHARED3) $(CXXFLAGS) $(PLATFORM_SHARED_CFLAGS) $(LIB_SOURCES) $(TOOL_LIB_SOURCES) \
-		$(LDFLAGS) -o $@
+shared_libobjects = $(patsubst %,shared-objects/%,$(LIBOBJECTS))
+CLEAN_FILES += shared-objects
+
+$(shared_libobjects): shared-objects/%.o: %.cc
+	$(AM_V_CC)mkdir -p $(@D) && $(CXX) $(CXXFLAGS) $(PLATFORM_SHARED_CFLAGS) -c $< -o $@
+
+$(SHARED4): $(shared_libobjects)
+	$(CXX) $(PLATFORM_SHARED_LDFLAGS)$(SHARED3) $(CXXFLAGS) $(PLATFORM_SHARED_CFLAGS) $(shared_libobjects) $(LDFLAGS) -o $@

 endif  # PLATFORM_SHARED_EXT

@@ -566,7 +604,8 @@ ifneq (,$(filter check parallel_check,$(MAKECMDGOALS)),)
 # and create a randomly-named rocksdb.XXXX directory therein.
 # We'll use that directory in the "make check" rules.
 ifeq ($(TMPD),)
-TMPD := $(shell f=/dev/shm; test -k $$f || f=/tmp;			\
+TMPDIR := $(shell echo $${TMPDIR:-/tmp})
+TMPD := $(shell f=/dev/shm; test -k $$f || f=$(TMPDIR);     \
  perl -le 'use File::Temp "tempdir";'					\
    -e 'print tempdir("'$$f'/rocksdb.XXXX", CLEANUP => 0)')
 endif
@@ -721,10 +760,12 @@ check: all
 	      echo "===== Running $$t"; ./$$t || exit 1; done;          \
 	fi
 	rm -rf $(TMPD)
+ifneq ($(PLATFORM), OS_AIX)
 ifeq ($(filter -DROCKSDB_LITE,$(OPT)),)
 	python tools/ldb_test.py
 	sh tools/rocksdb_dump_test.sh
 endif
+endif

 # TODO add ldb_tests
 check_some: $(SUBSET)
@@ -870,12 +911,13 @@ clean:
 	rm -rf $(CLEAN_FILES) ios-x86 ios-arm scan_build_report
 	find . -name "*.[oda]" -exec rm -f {} \;
 	find . -type f -regex ".*\.\(\(gcda\)\|\(gcno\)\)" -exec rm {} \;
-	rm -rf bzip2* snappy* zlib* lz4*
+	rm -rf bzip2* snappy* zlib* lz4* zstd*
 	cd java; $(MAKE) clean

 tags:
 	ctags * -R
 	cscope -b `find . -name '*.cc'` `find . -name '*.h'` `find . -name '*.c'`
+	ctags -e -R -o etags *

 format:
 	build_tools/format-diff.sh
@@ -949,12 +991,27 @@ cache_test: cache/cache_test.o $(LIBOBJECTS) $(TESTHARNESS)
 coding_test: util/coding_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

+hash_test: util/hash_test.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 option_change_migration_test: utilities/option_change_migration/option_change_migration_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 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)
+	$(AM_LINK)
+
+cassandra_merge_test: utilities/merge_operators/cassandra/cassandra_merge_test.o utilities/merge_operators/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)
+	$(AM_LINK)
+
+cassandra_serialize_test: utilities/merge_operators/cassandra/cassandra_serialize_test.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 redis_test: utilities/redis/redis_lists_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

@@ -979,6 +1036,9 @@ slice_transform_test: util/slice_transform_test.o $(LIBOBJECTS) $(TESTHARNESS)
 db_basic_test: db/db_basic_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

+db_encryption_test: db/db_encryption_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 db_test: db/db_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

@@ -1027,6 +1087,12 @@ db_range_del_test: db/db_range_del_test.o db/db_test_util.o $(LIBOBJECTS) $(TEST
 db_sst_test: db/db_sst_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

+db_statistics_test: db/db_statistics_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
+db_write_test: db/db_write_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 external_sst_file_basic_test: db/external_sst_file_basic_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

@@ -1075,9 +1141,6 @@ prefix_test: db/prefix_test.o $(LIBOBJECTS) $(TESTHARNESS)
 backupable_db_test: utilities/backupable/backupable_db_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-blob_db_test: utilities/blob_db/blob_db_test.o $(LIBOBJECTS) $(TESTHARNESS)
-	$(AM_LINK)
-
 checkpoint_test: utilities/checkpoint/checkpoint_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

@@ -1185,6 +1248,9 @@ inlineskiplist_test: memtable/inlineskiplist_test.o $(LIBOBJECTS) $(TESTHARNESS)
 skiplist_test: memtable/skiplist_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

+write_buffer_manager_test: memtable/write_buffer_manager_test.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 version_edit_test: db/version_edit_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

@@ -1272,6 +1338,9 @@ db_bench_tool_test: tools/db_bench_tool_test.o $(BENCHTOOLOBJECTS) $(TESTHARNESS
 event_logger_test: util/event_logger_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

+timer_queue_test: util/timer_queue_test.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 sst_dump_test: tools/sst_dump_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

@@ -1308,6 +1377,9 @@ transaction_test: utilities/transactions/transaction_test.o $(LIBOBJECTS) $(TEST
 sst_dump: tools/sst_dump.o $(LIBOBJECTS)
 	$(AM_LINK)

+blob_dump: tools/blob_dump.o $(LIBOBJECTS)
+	$(AM_LINK)
+
 column_aware_encoding_exp: utilities/column_aware_encoding_exp.o $(EXPOBJECTS)
 	$(AM_LINK)

@@ -1338,6 +1410,9 @@ lua_test: utilities/lua/rocks_lua_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTH
 range_del_aggregator_test: db/range_del_aggregator_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

+blob_db_test: utilities/blob_db/blob_db_test.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 #-------------------------------------------------
 # make install related stuff
 INSTALL_PATH ?= /usr/local
@@ -1397,14 +1472,21 @@ ROCKSDB_JAVADOCS_JAR = rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PA
 ROCKSDB_SOURCES_JAR = rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-sources.jar
 SHA256_CMD = sha256sum

-ZLIB_VER = 1.2.11
-ZLIB_SHA256 = c3e5e9fdd5004dcb542feda5ee4f0ff0744628baf8ed2dd5d66f8ca1197cb1a1
-BZIP2_VER = 1.0.6
-BZIP2_SHA256 = a2848f34fcd5d6cf47def00461fcb528a0484d8edef8208d6d2e2909dc61d9cd
-SNAPPY_VER = 1.1.4
-SNAPPY_SHA256 = 134bfe122fd25599bb807bb8130e7ba6d9bdb851e0b16efcb83ac4f5d0b70057
-LZ4_VER = 1.7.5
-LZ4_SHA256 = 0190cacd63022ccb86f44fa5041dc6c3804407ad61550ca21c382827319e7e7e
+ZLIB_VER ?= 1.2.11
+ZLIB_SHA256 ?= c3e5e9fdd5004dcb542feda5ee4f0ff0744628baf8ed2dd5d66f8ca1197cb1a1
+ZLIB_DOWNLOAD_BASE ?= http://zlib.net
+BZIP2_VER ?= 1.0.6
+BZIP2_SHA256 ?= a2848f34fcd5d6cf47def00461fcb528a0484d8edef8208d6d2e2909dc61d9cd
+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_DOWNLOAD_BASE ?= https://github.com/lz4/lz4/archive
+ZSTD_VER ?= 1.2.0
+ZSTD_SHA256 ?= 4a7e4593a3638276ca7f2a09dc4f38e674d8317bbea51626393ca73fc047cbfb
+ZSTD_DOWNLOAD_BASE ?= https://github.com/facebook/zstd/archive

 ifeq ($(PLATFORM), OS_MACOSX)
 	ROCKSDBJNILIB = librocksdbjni-osx.jnilib
@@ -1427,47 +1509,53 @@ ifeq ($(PLATFORM), OS_SOLARIS)
 	JAVA_INCLUDE = -I$(JAVA_HOME)/include/ -I$(JAVA_HOME)/include/solaris
 	SHA256_CMD = digest -a sha256
 endif
+ifeq ($(PLATFORM), OS_AIX)
+	JAVA_INCLUDE = -I$(JAVA_HOME)/include/ -I$(JAVA_HOME)/include/aix
+	ROCKSDBJNILIB = librocksdbjni-aix.so
+	EXTRACT_SOURCES = gunzip < TAR_GZ | tar xvf -
+	SNAPPY_MAKE_TARGET = libsnappy.la
+endif

 libz.a:
 	-rm -rf zlib-$(ZLIB_VER)
-	curl -O -L http://zlib.net/zlib-$(ZLIB_VER).tar.gz
+	curl -O -L ${ZLIB_DOWNLOAD_BASE}/zlib-$(ZLIB_VER).tar.gz
 	ZLIB_SHA256_ACTUAL=`$(SHA256_CMD) zlib-$(ZLIB_VER).tar.gz | cut -d ' ' -f 1`; \
 	if [ "$(ZLIB_SHA256)" != "$$ZLIB_SHA256_ACTUAL" ]; then \
 		echo zlib-$(ZLIB_VER).tar.gz checksum mismatch, expected=\"$(ZLIB_SHA256)\" actual=\"$$ZLIB_SHA256_ACTUAL\"; \
 		exit 1; \
 	fi
 	tar xvzf zlib-$(ZLIB_VER).tar.gz
-	cd zlib-$(ZLIB_VER) && CFLAGS='-fPIC' ./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:
 	-rm -rf bzip2-$(BZIP2_VER)
-	curl -O -L http://www.bzip.org/$(BZIP2_VER)/bzip2-$(BZIP2_VER).tar.gz
+	curl -O -L ${BZIP2_DOWNLOAD_BASE}/$(BZIP2_VER)/bzip2-$(BZIP2_VER).tar.gz
 	BZIP2_SHA256_ACTUAL=`$(SHA256_CMD) bzip2-$(BZIP2_VER).tar.gz | cut -d ' ' -f 1`; \
 	if [ "$(BZIP2_SHA256)" != "$$BZIP2_SHA256_ACTUAL" ]; then \
 		echo bzip2-$(BZIP2_VER).tar.gz checksum mismatch, expected=\"$(BZIP2_SHA256)\" actual=\"$$BZIP2_SHA256_ACTUAL\"; \
 		exit 1; \
 	fi
 	tar xvzf bzip2-$(BZIP2_VER).tar.gz
-	cd bzip2-$(BZIP2_VER) && make CFLAGS='-fPIC -O2 -g -D_FILE_OFFSET_BITS=64'
+	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:
 	-rm -rf snappy-$(SNAPPY_VER)
-	curl -O -L https://github.com/google/snappy/releases/download/$(SNAPPY_VER)/snappy-$(SNAPPY_VER).tar.gz
+	curl -O -L ${SNAPPY_DOWNLOAD_BASE}/$(SNAPPY_VER)/snappy-$(SNAPPY_VER).tar.gz
 	SNAPPY_SHA256_ACTUAL=`$(SHA256_CMD) snappy-$(SNAPPY_VER).tar.gz | cut -d ' ' -f 1`; \
 	if [ "$(SNAPPY_SHA256)" != "$$SNAPPY_SHA256_ACTUAL" ]; then \
 		echo snappy-$(SNAPPY_VER).tar.gz checksum mismatch, expected=\"$(SNAPPY_SHA256)\" actual=\"$$SNAPPY_SHA256_ACTUAL\"; \
 		exit 1; \
 	fi
 	tar xvzf snappy-$(SNAPPY_VER).tar.gz
-	cd snappy-$(SNAPPY_VER) && ./configure --with-pic --enable-static
-	cd snappy-$(SNAPPY_VER) && make
+	cd snappy-$(SNAPPY_VER) && CFLAGS='${EXTRA_CFLAGS}' CXXFLAGS='${EXTRA_CXXFLAGS}' LDFLAGS='${EXTRA_LDFLAGS}' ./configure --with-pic --enable-static --disable-shared
+	cd snappy-$(SNAPPY_VER) && make ${SNAPPY_MAKE_TARGET}
 	cp snappy-$(SNAPPY_VER)/.libs/libsnappy.a .

 liblz4.a:
 	-rm -rf lz4-$(LZ4_VER)
-	curl -O -L https://github.com/lz4/lz4/archive/v$(LZ4_VER).tar.gz
+	curl -O -L ${LZ4_DOWNLOAD_BASE}/v$(LZ4_VER).tar.gz
 	mv v$(LZ4_VER).tar.gz lz4-$(LZ4_VER).tar.gz
 	LZ4_SHA256_ACTUAL=`$(SHA256_CMD) lz4-$(LZ4_VER).tar.gz | cut -d ' ' -f 1`; \
 	if [ "$(LZ4_SHA256)" != "$$LZ4_SHA256_ACTUAL" ]; then \
@@ -1475,19 +1563,32 @@ liblz4.a:
 		exit 1; \
 	fi
 	tar xvzf lz4-$(LZ4_VER).tar.gz
-	cd lz4-$(LZ4_VER)/lib && make CFLAGS='-fPIC' all
+	cd lz4-$(LZ4_VER)/lib && make CFLAGS='-fPIC -O2 ${EXTRA_CFLAGS}' all
 	cp lz4-$(LZ4_VER)/lib/liblz4.a .

+libzstd.a:
+	-rm -rf zstd-$(ZSTD_VER)
+	curl -O -L ${ZSTD_DOWNLOAD_BASE}/v$(ZSTD_VER).tar.gz
+	mv v$(ZSTD_VER).tar.gz zstd-$(ZSTD_VER).tar.gz
+	ZSTD_SHA256_ACTUAL=`$(SHA256_CMD) zstd-$(ZSTD_VER).tar.gz | cut -d ' ' -f 1`; \
+	if [ "$(ZSTD_SHA256)" != "$$ZSTD_SHA256_ACTUAL" ]; then \
+		echo zstd-$(ZSTD_VER).tar.gz checksum mismatch, expected=\"$(ZSTD_SHA256)\" actual=\"$$ZSTD_SHA256_ACTUAL\"; \
+		exit 1; \
+	fi
+	tar xvzf zstd-$(ZSTD_VER).tar.gz
+	cd zstd-$(ZSTD_VER)/lib && make CFLAGS='-fPIC -O2 ${EXTRA_CFLAGS}' all
+	cp zstd-$(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))
 CLEAN_FILES += jls

 ifneq ($(ROCKSDB_JAVA_NO_COMPRESSION), 1)
-JAVA_COMPRESSIONS = libz.a libbz2.a libsnappy.a liblz4.a
+JAVA_COMPRESSIONS = libz.a libbz2.a libsnappy.a liblz4.a libzstd.a
 endif

-JAVA_STATIC_FLAGS = -DZLIB -DBZIP2 -DSNAPPY -DLZ4
-JAVA_STATIC_INCLUDES = -I./zlib-$(ZLIB_VER) -I./bzip2-$(BZIP2_VER) -I./snappy-$(SNAPPY_VER) -I./lz4-$(LZ4_VER)/lib
+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_libobjects): jls/%.o: %.cc $(JAVA_COMPRESSIONS)
 	$(AM_V_CC)mkdir -p $(@D) && $(CXX) $(CXXFLAGS) $(JAVA_STATIC_FLAGS) $(JAVA_STATIC_INCLUDES) -fPIC -c $< -o $@ $(COVERAGEFLAGS)
@@ -1499,7 +1600,7 @@ rocksdbjavastatic: $(java_static_libobjects)
 	  -o ./java/target/$(ROCKSDBJNILIB) $(JNI_NATIVE_SOURCES) \
 	  $(java_static_libobjects) $(COVERAGEFLAGS) \
 	  $(JAVA_COMPRESSIONS) $(JAVA_STATIC_LDFLAGS)
-	cd java/target;strip -S -x $(ROCKSDBJNILIB)
+	cd java/target;strip $(STRIPFLAGS) $(ROCKSDBJNILIB)
 	cd java;jar -cf target/$(ROCKSDB_JAR) HISTORY*.md
 	cd java/target;jar -uf $(ROCKSDB_JAR) $(ROCKSDBJNILIB)
 	cd java/target/classes;jar -uf ../$(ROCKSDB_JAR) org/rocksdb/*.class org/rocksdb/util/*.class
@@ -1512,8 +1613,25 @@ rocksdbjavastaticrelease: 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

+rocksdbjavastaticreleasedocker: rocksdbjavastatic
+	DOCKER_LINUX_X64_CONTAINER=`docker ps -aqf name=rocksdb_linux_x64-be`; \
+	if [ -z "$$DOCKER_LINUX_X64_CONTAINER" ]; then \
+		docker container create --attach stdin --attach stdout --attach stderr --volume `pwd`:/rocksdb-host --name rocksdb_linux_x64-be evolvedbinary/rocksjava:centos6_x64-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh; \
+	fi
+	docker start -a rocksdb_linux_x64-be
+	DOCKER_LINUX_X86_CONTAINER=`docker ps -aqf name=rocksdb_linux_x86-be`; \
+	if [ -z "$$DOCKER_LINUX_X86_CONTAINER" ]; then \
+		docker container create --attach stdin --attach stdout --attach stderr --volume `pwd`:/rocksdb-host --name rocksdb_linux_x86-be evolvedbinary/rocksjava:centos6_x86-be /rocksdb-host/java/crossbuild/docker-build-linux-centos.sh; \
+	fi
+	docker start -a rocksdb_linux_x86-be
+	cd java;jar -cf target/$(ROCKSDB_JAR_ALL) HISTORY*.md
+	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
+
 rocksdbjavastaticpublish: rocksdbjavastaticrelease rocksdbjavastaticpublishcentral

+rocksdbjavastaticpublishdocker: rocksdbjavastaticreleasedocker rocksdbjavastaticpublishcentral
+
 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).$(ROCKSDB_PATCH)-javadoc.jar -Dclassifier=javadoc
 	mvn gpg:sign-and-deploy-file -Durl=https://oss.sonatype.org/service/local/staging/deploy/maven2/ -DrepositoryId=sonatype-nexus-staging -DpomFile=java/rocksjni.pom -Dfile=java/target/rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-sources.jar -Dclassifier=sources
@@ -1,33 +0,0 @@
-Additional Grant of Patent Rights Version 2
-
-"Software" means the RocksDB software distributed by Facebook, Inc.
-
-Facebook, Inc. ("Facebook") hereby grants to each recipient of the Software
-("you") a perpetual, worldwide, royalty-free, non-exclusive, irrevocable
-(subject to the termination provision below) license under any Necessary
-Claims, to make, have made, use, sell, offer to sell, import, and otherwise
-transfer the Software. For avoidance of doubt, no license is granted under
-Facebook’s rights in any patent claims that are infringed by (i) modifications
-to the Software made by you or any third party or (ii) the Software in
-combination with any software or other technology.
-
-The license granted hereunder will terminate, automatically and without notice,
-if you (or any of your subsidiaries, corporate affiliates or agents) initiate
-directly or indirectly, or take a direct financial interest in, any Patent
-Assertion: (i) against Facebook or any of its subsidiaries or corporate
-affiliates, (ii) against any party if such Patent Assertion arises in whole or
-in part from any software, technology, product or service of Facebook or any of
-its subsidiaries or corporate affiliates, or (iii) against any party relating
-to the Software. Notwithstanding the foregoing, if Facebook or any of its
-subsidiaries or corporate affiliates files a lawsuit alleging patent
-infringement against you in the first instance, and you respond by filing a
-patent infringement counterclaim in that lawsuit against that party that is
-unrelated to the Software, the license granted hereunder will not terminate
-under section (i) of this paragraph due to such counterclaim.
-
-A "Necessary Claim" is a claim of a patent owned by Facebook that is
-necessarily infringed by the Software standing alone.
-
-A "Patent Assertion" is any lawsuit or other action alleging direct, indirect,
-or contributory infringement or inducement to infringe any patent, including a
-cross-claim or counterclaim.
@@ -1,4 +1,8 @@
-REPO_PATH = "internal_repo_rocksdb/repo/"
+
+import os
+
+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 = [
@@ -9,6 +13,10 @@ rocksdb_compiler_flags = [
  "-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",
@@ -31,7 +39,7 @@ rocksdb_external_deps = [
  ('lz4', None, 'lz4'),
  ('zstd', None),
  ('tbb', None),
-  ("numa", "2.0.8", "numa"),
+  ("numa", None, "numa"),
  ("googletest", None, "gtest"),
 ]

@@ -80,6 +88,7 @@ cpp_library(
      "db/internal_stats.cc",
      "db/log_reader.cc",
      "db/log_writer.cc",
+      "db/malloc_stats.cc",
      "db/managed_iterator.cc",
      "db/memtable.cc",
      "db/memtable_list.cc",
@@ -101,16 +110,18 @@ cpp_library(
      "db/write_thread.cc",
      "env/env.cc",
      "env/env_chroot.cc",
+      "env/env_encryption.cc",
      "env/env_hdfs.cc",
      "env/env_posix.cc",
      "env/io_posix.cc",
-      "env/memenv.cc",
+      "env/mock_env.cc",
+      "memtable/alloc_tracker.cc",
      "memtable/hash_cuckoo_rep.cc",
      "memtable/hash_linklist_rep.cc",
      "memtable/hash_skiplist_rep.cc",
-      "memtable/memtable_allocator.cc",
      "memtable/skiplistrep.cc",
      "memtable/vectorrep.cc",
+      "memtable/write_buffer_manager.cc",
      "monitoring/histogram.cc",
      "monitoring/histogram_windowing.cc",
      "monitoring/instrumented_mutex.cc",
@@ -195,10 +206,17 @@ cpp_library(
      "util/xxhash.cc",
      "utilities/backupable/backupable_db.cc",
      "utilities/blob_db/blob_db.cc",
-      "utilities/checkpoint/checkpoint.cc",
+      "utilities/blob_db/blob_db_impl.cc",
+      "utilities/blob_db/blob_file.cc",
+      "utilities/blob_db/blob_log_reader.cc",
+      "utilities/blob_db/blob_log_writer.cc",
+      "utilities/blob_db/blob_log_format.cc",
+      "utilities/blob_db/ttl_extractor.cc",
+      "utilities/checkpoint/checkpoint_impl.cc",
      "utilities/compaction_filters/remove_emptyvalue_compactionfilter.cc",
      "utilities/convenience/info_log_finder.cc",
      "utilities/date_tiered/date_tiered_db_impl.cc",
+      "utilities/debug.cc",
      "utilities/document/document_db.cc",
      "utilities/document/json_document.cc",
      "utilities/document/json_document_builder.cc",
@@ -208,6 +226,8 @@ cpp_library(
      "utilities/leveldb_options/leveldb_options.cc",
      "utilities/lua/rocks_lua_compaction_filter.cc",
      "utilities/memory/memory_util.cc",
+      "utilities/merge_operators/cassandra/format.cc",
+      "utilities/merge_operators/cassandra/merge_operator.cc",
      "utilities/merge_operators/max.cc",
      "utilities/merge_operators/put.cc",
      "utilities/merge_operators/string_append/stringappend.cc",
@@ -238,6 +258,7 @@ cpp_library(
      "tools/ldb_cmd.cc",
      "tools/ldb_tool.cc",
      "tools/sst_dump_tool.cc",
+      "utilities/blob_db/blob_dump_tool.cc",
    ],
    deps = [],
    preprocessor_flags = rocksdb_preprocessor_flags,
@@ -249,12 +270,12 @@ cpp_library(
    name = "rocksdb_test_lib",
    headers = AutoHeaders.RECURSIVE_GLOB,
    srcs = [
-      "env/mock_env.cc",
      "table/mock_table.cc",
      "util/fault_injection_test_env.cc",
      "util/testharness.cc",
      "util/testutil.cc",
      "db/db_test_util.cc",
+      "utilities/merge_operators/cassandra/test_utils.cc",
      "utilities/col_buf_encoder.cc",
      "utilities/col_buf_decoder.cc",
      "utilities/column_aware_encoding_util.cc",
@@ -289,162 +310,182 @@ cpp_library(
 )

 # [test_name, test_src, test_type]
-ROCKS_TESTS = [['merger_test', 'table/merger_test.cc', 'serial'],
- ['cache_test', 'cache/cache_test.cc', 'serial'],
- ['options_file_test', 'db/options_file_test.cc', 'serial'],
- ['compaction_picker_test', 'db/compaction_picker_test.cc', 'serial'],
- ['corruption_test', 'db/corruption_test.cc', 'serial'],
- ['thread_list_test', 'util/thread_list_test.cc', 'serial'],
- ['table_properties_collector_test',
-  'db/table_properties_collector_test.cc',
-  'serial'],
- ['document_db_test', 'utilities/document/document_db_test.cc', 'serial'],
- ['event_logger_test', 'util/event_logger_test.cc', 'serial'],
- ['coding_test', 'util/coding_test.cc', 'serial'],
- ['statistics_test', 'monitoring/statistics_test.cc', 'serial'],
- ['options_settable_test', 'options/options_settable_test.cc', 'serial'],
- ['sst_dump_test', 'tools/sst_dump_test.cc', 'serial'],
- ['column_aware_encoding_test',
-  'utilities/column_aware_encoding_test.cc',
-  'serial'],
- ['db_iterator_test', 'db/db_iterator_test.cc', 'serial'],
- ['db_sst_test', 'db/db_sst_test.cc', 'parallel'],
- ['geodb_test', 'utilities/geodb/geodb_test.cc', 'serial'],
- ['listener_test', 'db/listener_test.cc', 'serial'],
- ['write_callback_test', 'db/write_callback_test.cc', 'serial'],
- ['version_set_test', 'db/version_set_test.cc', 'serial'],
- ['full_filter_block_test', 'table/full_filter_block_test.cc', 'serial'],
- ['cleanable_test', 'table/cleanable_test.cc', 'serial'],
- ['checkpoint_test', 'utilities/checkpoint/checkpoint_test.cc', 'serial'],
- ['compact_files_test', 'db/compact_files_test.cc', 'serial'],
- ['db_options_test', 'db/db_options_test.cc', 'serial'],
- ['object_registry_test', 'utilities/object_registry_test.cc', 'serial'],
- ['blob_db_test', 'utilities/blob_db/blob_db_test.cc', 'serial'],
+ROCKS_TESTS = [['arena_test', 'util/arena_test.cc', 'serial'],
 ['auto_roll_logger_test', 'util/auto_roll_logger_test.cc', 'serial'],
- ['dbformat_test', 'db/dbformat_test.cc', 'serial'],
- ['write_batch_with_index_test',
-  'utilities/write_batch_with_index/write_batch_with_index_test.cc',
-  'serial'],
- ['json_document_test', 'utilities/document/json_document_test.cc', 'serial'],
- ['file_reader_writer_test', 'util/file_reader_writer_test.cc', 'serial'],
- ['repair_test', 'db/repair_test.cc', 'serial'],
- ['persistent_cache_test',
-  'utilities/persistent_cache/persistent_cache_test.cc',
-  'parallel'],
- ['db_bloom_filter_test', 'db/db_bloom_filter_test.cc', 'serial'],
- ['external_sst_file_basic_test',
-  'db/external_sst_file_basic_test.cc',
-  'serial'],
- ['options_test', 'options/options_test.cc', 'serial'],
- ['perf_context_test', 'db/perf_context_test.cc', 'serial'],
- ['db_block_cache_test', 'db/db_block_cache_test.cc', 'serial'],
- ['heap_test', 'util/heap_test.cc', 'serial'],
- ['db_test2', 'db/db_test2.cc', 'serial'],
- ['filelock_test', 'util/filelock_test.cc', 'serial'],
- ['write_controller_test', 'db/write_controller_test.cc', 'serial'],
- ['compaction_iterator_test', 'db/compaction_iterator_test.cc', 'serial'],
- ['spatial_db_test', 'utilities/spatialdb/spatial_db_test.cc', 'serial'],
- ['c_test', 'db/c_test.c', 'serial'],
- ['range_del_aggregator_test', 'db/range_del_aggregator_test.cc', 'serial'],
- ['date_tiered_test', 'utilities/date_tiered/date_tiered_test.cc', 'serial'],
- ['ldb_cmd_test', 'tools/ldb_cmd_test.cc', 'serial'],
- ['db_test', 'db/db_test.cc', 'parallel'],
- ['block_based_filter_block_test',
-  'table/block_based_filter_block_test.cc',
-  'serial'],
- ['merge_test', 'db/merge_test.cc', 'serial'],
- ['bloom_test', 'util/bloom_test.cc', 'serial'],
- ['block_test', 'table/block_test.cc', 'serial'],
- ['cuckoo_table_builder_test', 'table/cuckoo_table_builder_test.cc', 'serial'],
+ ['autovector_test', 'util/autovector_test.cc', 'serial'],
 ['backupable_db_test',
  'utilities/backupable/backupable_db_test.cc',
  'parallel'],
- ['db_flush_test', 'db/db_flush_test.cc', 'serial'],
- ['filename_test', 'db/filename_test.cc', 'serial'],
- ['cuckoo_table_reader_test', 'table/cuckoo_table_reader_test.cc', 'serial'],
- ['slice_transform_test', 'util/slice_transform_test.cc', 'serial'],
- ['cuckoo_table_db_test', 'db/cuckoo_table_db_test.cc', 'serial'],
- ['inlineskiplist_test', 'memtable/inlineskiplist_test.cc', 'parallel'],
- ['optimistic_transaction_test',
-  'utilities/transactions/optimistic_transaction_test.cc',
+ ['blob_db_test', 'utilities/blob_db/blob_db_test.cc', 'serial'],
+ ['block_based_filter_block_test',
+  'table/block_based_filter_block_test.cc',
  'serial'],
- ['hash_table_test',
-  'utilities/persistent_cache/hash_table_test.cc',
+ ['block_test', 'table/block_test.cc', 'serial'],
+ ['bloom_test', 'util/bloom_test.cc', 'serial'],
+ ['c_test', 'db/c_test.c', 'serial'],
+ ['cache_test', 'cache/cache_test.cc', 'serial'],
+ ['cassandra_format_test',
+  'utilities/merge_operators/cassandra/cassandra_format_test.cc',
  'serial'],
- ['db_dynamic_level_test', 'db/db_dynamic_level_test.cc', 'serial'],
- ['option_change_migration_test',
-  'utilities/option_change_migration/option_change_migration_test.cc',
+ ['cassandra_merge_test',
+  'utilities/merge_operators/cassandra/cassandra_merge_test.cc',
  'serial'],
- ['db_inplace_update_test', 'db/db_inplace_update_test.cc', 'serial'],
- ['autovector_test', 'util/autovector_test.cc', 'serial'],
- ['db_iter_test', 'db/db_iter_test.cc', 'serial'],
- ['flush_job_test', 'db/flush_job_test.cc', 'serial'],
- ['wal_manager_test', 'db/wal_manager_test.cc', 'serial'],
- ['write_batch_test', 'db/write_batch_test.cc', 'serial'],
- ['crc32c_test', 'util/crc32c_test.cc', 'serial'],
- ['rate_limiter_test', 'util/rate_limiter_test.cc', 'serial'],
- ['external_sst_file_test', 'db/external_sst_file_test.cc', 'parallel'],
- ['compaction_job_test', 'db/compaction_job_test.cc', 'serial'],
- ['mock_env_test', 'env/mock_env_test.cc', 'serial'],
- ['db_table_properties_test', 'db/db_table_properties_test.cc', 'serial'],
- ['db_compaction_test', 'db/db_compaction_test.cc', 'parallel'],
- ['arena_test', 'util/arena_test.cc', 'serial'],
- ['stringappend_test',
-  'utilities/merge_operators/string_append/stringappend_test.cc',
+ ['cassandra_row_merge_test',
+  'utilities/merge_operators/cassandra/cassandra_row_merge_test.cc',
  'serial'],
- ['reduce_levels_test', 'tools/reduce_levels_test.cc', 'serial'],
- ['prefix_test', 'db/prefix_test.cc', 'serial'],
- ['ttl_test', 'utilities/ttl/ttl_test.cc', 'serial'],
- ['merge_helper_test', 'db/merge_helper_test.cc', 'serial'],
- ['file_indexer_test', 'db/file_indexer_test.cc', 'serial'],
- ['memory_test', 'utilities/memory/memory_test.cc', 'serial'],
- ['log_test', 'db/log_test.cc', 'serial'],
- ['env_timed_test', 'utilities/env_timed_test.cc', 'serial'],
- ['deletefile_test', 'db/deletefile_test.cc', 'serial'],
- ['partitioned_filter_block_test',
-  'table/partitioned_filter_block_test.cc',
+ ['cassandra_serialize_test',
+  'utilities/merge_operators/cassandra/cassandra_serialize_test.cc',
  'serial'],
- ['comparator_db_test', 'db/comparator_db_test.cc', 'serial'],
- ['compaction_job_stats_test', 'db/compaction_job_stats_test.cc', 'serial'],
- ['thread_local_test', 'util/thread_local_test.cc', 'serial'],
- ['version_builder_test', 'db/version_builder_test.cc', 'serial'],
- ['db_range_del_test', 'db/db_range_del_test.cc', 'serial'],
- ['table_test', 'table/table_test.cc', 'parallel'],
- ['db_tailing_iter_test', 'db/db_tailing_iter_test.cc', 'serial'],
- ['db_compaction_filter_test', 'db/db_compaction_filter_test.cc', 'parallel'],
- ['options_util_test', 'utilities/options/options_util_test.cc', 'serial'],
- ['dynamic_bloom_test', 'util/dynamic_bloom_test.cc', 'serial'],
- ['db_basic_test', 'db/db_basic_test.cc', 'serial'],
- ['db_merge_operator_test', 'db/db_merge_operator_test.cc', 'serial'],
- ['manual_compaction_test', 'db/manual_compaction_test.cc', 'parallel'],
- ['delete_scheduler_test', 'util/delete_scheduler_test.cc', 'serial'],
- ['transaction_test', 'utilities/transactions/transaction_test.cc', 'serial'],
- ['db_io_failure_test', 'db/db_io_failure_test.cc', 'serial'],
- ['db_log_iter_test', 'db/db_log_iter_test.cc', 'serial'],
+ ['checkpoint_test', 'utilities/checkpoint/checkpoint_test.cc', 'serial'],
+ ['cleanable_test', 'table/cleanable_test.cc', 'serial'],
+ ['coding_test', 'util/coding_test.cc', 'serial'],
+ ['column_aware_encoding_test',
+  'utilities/column_aware_encoding_test.cc',
+  'serial'],
+ ['column_family_test', 'db/column_family_test.cc', 'serial'],
+ ['compact_files_test', 'db/compact_files_test.cc', 'serial'],
 ['compact_on_deletion_collector_test',
  'utilities/table_properties_collectors/compact_on_deletion_collector_test.cc',
  'serial'],
- ['env_test', 'env/env_test.cc', 'serial'],
- ['db_wal_test', 'db/db_wal_test.cc', 'parallel'],
- ['sim_cache_test', 'utilities/simulator_cache/sim_cache_test.cc', 'serial'],
+ ['compaction_iterator_test', 'db/compaction_iterator_test.cc', 'serial'],
+ ['compaction_job_stats_test', 'db/compaction_job_stats_test.cc', 'serial'],
+ ['compaction_job_test', 'db/compaction_job_test.cc', 'serial'],
+ ['compaction_picker_test', 'db/compaction_picker_test.cc', 'serial'],
+ ['comparator_db_test', 'db/comparator_db_test.cc', 'serial'],
+ ['corruption_test', 'db/corruption_test.cc', 'serial'],
+ ['crc32c_test', 'util/crc32c_test.cc', 'serial'],
+ ['cuckoo_table_builder_test', 'table/cuckoo_table_builder_test.cc', 'serial'],
+ ['cuckoo_table_db_test', 'db/cuckoo_table_db_test.cc', 'serial'],
+ ['cuckoo_table_reader_test', 'table/cuckoo_table_reader_test.cc', 'serial'],
+ ['date_tiered_test', 'utilities/date_tiered/date_tiered_test.cc', 'serial'],
+ ['db_basic_test', 'db/db_basic_test.cc', 'serial'],
+ ['db_block_cache_test', 'db/db_block_cache_test.cc', 'serial'],
+ ['db_bloom_filter_test', 'db/db_bloom_filter_test.cc', 'serial'],
+ ['db_compaction_filter_test', 'db/db_compaction_filter_test.cc', 'parallel'],
+ ['db_compaction_test', 'db/db_compaction_test.cc', 'parallel'],
+ ['db_dynamic_level_test', 'db/db_dynamic_level_test.cc', 'serial'],
+ ['db_encryption_test', 'db/db_encryption_test.cc', 'serial'],
+ ['db_flush_test', 'db/db_flush_test.cc', 'serial'],
+ ['db_inplace_update_test', 'db/db_inplace_update_test.cc', 'serial'],
+ ['db_io_failure_test', 'db/db_io_failure_test.cc', 'serial'],
+ ['db_iter_test', 'db/db_iter_test.cc', 'serial'],
+ ['db_iterator_test', 'db/db_iterator_test.cc', 'serial'],
+ ['db_log_iter_test', 'db/db_log_iter_test.cc', 'serial'],
 ['db_memtable_test', 'db/db_memtable_test.cc', 'serial'],
+ ['db_merge_operator_test', 'db/db_merge_operator_test.cc', 'serial'],
+ ['db_options_test', 'db/db_options_test.cc', 'serial'],
+ ['db_properties_test', 'db/db_properties_test.cc', 'serial'],
+ ['db_range_del_test', 'db/db_range_del_test.cc', 'serial'],
+ ['db_sst_test', 'db/db_sst_test.cc', 'parallel'],
+ ['db_statistics_test', 'db/db_statistics_test.cc', 'serial'],
+ ['db_table_properties_test', 'db/db_table_properties_test.cc', 'serial'],
+ ['db_tailing_iter_test', 'db/db_tailing_iter_test.cc', 'serial'],
+ ['db_test', 'db/db_test.cc', 'parallel'],
+ ['db_test2', 'db/db_test2.cc', 'serial'],
 ['db_universal_compaction_test',
  'db/db_universal_compaction_test.cc',
  'parallel'],
+ ['db_wal_test', 'db/db_wal_test.cc', 'parallel'],
+ ['db_write_test', 'db/db_write_test.cc', 'serial'],
+ ['dbformat_test', 'db/dbformat_test.cc', 'serial'],
+ ['delete_scheduler_test', 'util/delete_scheduler_test.cc', 'serial'],
+ ['deletefile_test', 'db/deletefile_test.cc', 'serial'],
+ ['document_db_test', 'utilities/document/document_db_test.cc', 'serial'],
+ ['dynamic_bloom_test', 'util/dynamic_bloom_test.cc', 'serial'],
+ ['env_basic_test', 'env/env_basic_test.cc', 'serial'],
+ ['env_test', 'env/env_test.cc', 'serial'],
+ ['env_timed_test', 'utilities/env_timed_test.cc', 'serial'],
+ ['event_logger_test', 'util/event_logger_test.cc', 'serial'],
+ ['external_sst_file_basic_test',
+  'db/external_sst_file_basic_test.cc',
+  'serial'],
+ ['external_sst_file_test', 'db/external_sst_file_test.cc', 'parallel'],
+ ['fault_injection_test', 'db/fault_injection_test.cc', 'parallel'],
+ ['file_indexer_test', 'db/file_indexer_test.cc', 'serial'],
+ ['file_reader_writer_test', 'util/file_reader_writer_test.cc', 'serial'],
+ ['filelock_test', 'util/filelock_test.cc', 'serial'],
+ ['filename_test', 'db/filename_test.cc', 'serial'],
+ ['flush_job_test', 'db/flush_job_test.cc', 'serial'],
+ ['full_filter_block_test', 'table/full_filter_block_test.cc', 'serial'],
+ ['geodb_test', 'utilities/geodb/geodb_test.cc', 'serial'],
+ ['hash_table_test',
+  'utilities/persistent_cache/hash_table_test.cc',
+  'serial'],
+ ['hash_test', 'util/hash_test.cc', 'serial'],
+ ['heap_test', 'util/heap_test.cc', 'serial'],
 ['histogram_test', 'monitoring/histogram_test.cc', 'serial'],
+ ['inlineskiplist_test', 'memtable/inlineskiplist_test.cc', 'parallel'],
+ ['iostats_context_test', 'monitoring/iostats_context_test.cc', 'serial'],
+ ['json_document_test', 'utilities/document/json_document_test.cc', 'serial'],
+ ['ldb_cmd_test', 'tools/ldb_cmd_test.cc', 'serial'],
+ ['listener_test', 'db/listener_test.cc', 'serial'],
+ ['log_test', 'db/log_test.cc', 'serial'],
+ ['lru_cache_test', 'cache/lru_cache_test.cc', 'serial'],
+ ['manual_compaction_test', 'db/manual_compaction_test.cc', 'parallel'],
+ ['memory_test', 'utilities/memory/memory_test.cc', 'serial'],
+ ['memtable_list_test', 'db/memtable_list_test.cc', 'serial'],
+ ['merge_helper_test', 'db/merge_helper_test.cc', 'serial'],
+ ['merge_test', 'db/merge_test.cc', 'serial'],
+ ['merger_test', 'table/merger_test.cc', 'serial'],
+ ['mock_env_test', 'env/mock_env_test.cc', 'serial'],
+ ['object_registry_test', 'utilities/object_registry_test.cc', 'serial'],
+ ['optimistic_transaction_test',
+  'utilities/transactions/optimistic_transaction_test.cc',
+  'serial'],
+ ['option_change_migration_test',
+  'utilities/option_change_migration/option_change_migration_test.cc',
+  'serial'],
+ ['options_file_test', 'db/options_file_test.cc', 'serial'],
+ ['options_settable_test', 'options/options_settable_test.cc', 'serial'],
+ ['options_test', 'options/options_test.cc', 'serial'],
+ ['options_util_test', 'utilities/options/options_util_test.cc', 'serial'],
+ ['partitioned_filter_block_test',
+  'table/partitioned_filter_block_test.cc',
+  'serial'],
+ ['perf_context_test', 'db/perf_context_test.cc', 'serial'],
+ ['persistent_cache_test',
+  'utilities/persistent_cache/persistent_cache_test.cc',
+  'parallel'],
+ ['plain_table_db_test', 'db/plain_table_db_test.cc', 'serial'],
+ ['prefix_test', 'db/prefix_test.cc', 'serial'],
+ ['range_del_aggregator_test', 'db/range_del_aggregator_test.cc', 'serial'],
+ ['rate_limiter_test', 'util/rate_limiter_test.cc', 'serial'],
+ ['reduce_levels_test', 'tools/reduce_levels_test.cc', 'serial'],
+ ['repair_test', 'db/repair_test.cc', 'serial'],
+ ['sim_cache_test', 'utilities/simulator_cache/sim_cache_test.cc', 'serial'],
+ ['skiplist_test', 'memtable/skiplist_test.cc', 'serial'],
+ ['slice_transform_test', 'util/slice_transform_test.cc', 'serial'],
+ ['spatial_db_test', 'utilities/spatialdb/spatial_db_test.cc', 'serial'],
+ ['sst_dump_test', 'tools/sst_dump_test.cc', 'serial'],
+ ['statistics_test', 'monitoring/statistics_test.cc', 'serial'],
+ ['stringappend_test',
+  'utilities/merge_operators/string_append/stringappend_test.cc',
+  'serial'],
+ ['table_properties_collector_test',
+  'db/table_properties_collector_test.cc',
+  'serial'],
+ ['table_test', 'table/table_test.cc', 'parallel'],
+ ['thread_list_test', 'util/thread_list_test.cc', 'serial'],
+ ['thread_local_test', 'util/thread_local_test.cc', 'serial'],
+ ['timer_queue_test', 'util/timer_queue_test.cc', 'serial'],
+ ['transaction_test', 'utilities/transactions/transaction_test.cc', 'serial'],
+ ['ttl_test', 'utilities/ttl/ttl_test.cc', 'serial'],
 ['util_merge_operators_test',
  'utilities/util_merge_operators_test.cc',
  'serial'],
- ['fault_injection_test', 'db/fault_injection_test.cc', 'parallel'],
- ['env_basic_test', 'env/env_basic_test.cc', 'serial'],
- ['iostats_context_test', 'monitoring/iostats_context_test.cc', 'serial'],
- ['memtable_list_test', 'db/memtable_list_test.cc', 'serial'],
- ['column_family_test', 'db/column_family_test.cc', 'serial'],
- ['db_properties_test', 'db/db_properties_test.cc', 'serial'],
+ ['version_builder_test', 'db/version_builder_test.cc', 'serial'],
 ['version_edit_test', 'db/version_edit_test.cc', 'serial'],
- ['skiplist_test', 'memtable/skiplist_test.cc', 'serial'],
- ['lru_cache_test', 'cache/lru_cache_test.cc', 'serial'],
- ['plain_table_db_test', 'db/plain_table_db_test.cc', 'serial']]
+ ['version_set_test', 'db/version_set_test.cc', 'serial'],
+ ['wal_manager_test', 'db/wal_manager_test.cc', 'serial'],
+ ['write_batch_test', 'db/write_batch_test.cc', 'serial'],
+ ['write_batch_with_index_test',
+  'utilities/write_batch_with_index/write_batch_with_index_test.cc',
+  'serial'],
+ ['write_buffer_manager_test',
+  'memtable/write_buffer_manager_test.cc',
+  'serial'],
+ ['write_callback_test', 'db/write_callback_test.cc', 'serial'],
+ ['write_controller_test', 'db/write_controller_test.cc', 'serial']]


 # Generate a test rule for each entry in ROCKS_TESTS
@@ -469,3 +510,15 @@ for test_cfg in ROCKS_TESTS:
      deps = [":" + test_bin],
      command = [TEST_RUNNER, BUCK_BINS + test_bin]
    )
+
+custom_unittest(
+    name = "make_rocksdbjavastatic",
+    type = "simple",
+    command = ["internal_repo_rocksdb/make_rocksdbjavastatic.sh"],
+)
+
+custom_unittest(
+    name = "make_rocksdb_lite_release",
+    type = "simple",
+    command = ["internal_repo_rocksdb/make_rocksdb_lite_release.sh"],
+)
@@ -76,7 +76,7 @@ quasardb uses a heavily tuned RocksDB as its persistence layer.
 [Dgraph](https://github.com/dgraph-io/dgraph) is an open-source, scalable, distributed, low latency, high throughput Graph database .They use RocksDB to store state locally on a machine.

 ## Uber
-[Uber](http://eng.uber.com/cherami/) uses RocksDB as a durable and and scalable task queue.
+[Uber](http://eng.uber.com/cherami/) uses RocksDB as a durable and scalable task queue.

 ## 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
@@ -84,7 +84,7 @@ def get_tests(repo_path):
            else:
                # we consumed all the parallel tests
                break
-
+    
    return tests


@@ -122,7 +122,7 @@ def generate_targets(repo_path):
        [":rocksdb_lib"])

    # test for every test we found in the Makefile
-    for test in tests:
+    for test in sorted(tests):
        match_src = [src for src in cc_files if ("/%s.c" % test) in src]
        if len(match_src) == 0:
            print(ColorString.warning("Cannot find .cc file for %s" % test))
@@ -2,7 +2,11 @@ from __future__ import absolute_import
 from __future__ import division
 from __future__ import print_function
 from __future__ import unicode_literals
-rocksdb_target_header = """REPO_PATH = "internal_repo_rocksdb/repo/"
+rocksdb_target_header = """
+import os
+
+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 = [
@@ -13,6 +17,10 @@ rocksdb_compiler_flags = [
  "-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",
@@ -35,7 +43,7 @@ rocksdb_external_deps = [
  ('lz4', None, 'lz4'),
  ('zstd', None),
  ('tbb', None),
-  ("numa", "2.0.8", "numa"),
+  ("numa", None, "numa"),
  ("googletest", None, "gtest"),
 ]

@@ -51,11 +51,13 @@ 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 [ -z "$ROCKSDB_FBCODE_BUILD_WITH_481" ]; then
-      source "$PWD/build_tools/fbcode_config.sh"
-    else
+    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 "$ROCKSDB_FBCODE_BUILD_WITH_5xx" ]; then
+      source "$PWD/build_tools/fbcode_config.sh"
+    else
+      source "$PWD/build_tools/fbcode_config_platform007.sh"
    fi
 fi

@@ -128,10 +130,17 @@ case "$TARGET_OS" in
        ;;
    SunOS)
        PLATFORM=OS_SOLARIS
-        COMMON_FLAGS="$COMMON_FLAGS -fno-builtin-memcmp -D_REENTRANT -DOS_SOLARIS"
-        PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lpthread -lrt"
+        COMMON_FLAGS="$COMMON_FLAGS -fno-builtin-memcmp -D_REENTRANT -DOS_SOLARIS -m64"
+        PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lpthread -lrt -static-libstdc++ -static-libgcc -m64"
        # PORT_FILES=port/sunos/sunos_specific.cc
        ;;
+    AIX)
+        PLATFORM=OS_AIX
+        CC=gcc
+        COMMON_FLAGS="$COMMON_FLAGS -maix64 -pthread -fno-builtin-memcmp -D_REENTRANT -DOS_AIX -D__STDC_FORMAT_MACROS"
+        PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -pthread -lpthread -lrt -maix64 -static-libstdc++ -static-libgcc"
+        # PORT_FILES=port/aix/aix_specific.cc
+        ;;
    FreeBSD)
        PLATFORM=OS_FREEBSD
        COMMON_FLAGS="$COMMON_FLAGS -fno-builtin-memcmp -D_REENTRANT -DOS_FREEBSD"
@@ -382,6 +391,29 @@ EOF
    if [ "$?" = 0 ]; then
        PROFILING_FLAGS=-pg
    fi
+
+    # Test whether sync_file_range is supported for compatibility with an old glibc
+    $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
+      #include <fcntl.h>
+      int main() {
+        int fd = open("/dev/null", 0);
+        sync_file_range(fd, 0, 1024, SYNC_FILE_RANGE_WRITE);
+      }
+EOF
+    if [ "$?" = 0 ]; then
+        COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_RANGESYNC_PRESENT"
+    fi
+
+    # Test whether sched_getcpu is supported
+    $CXX $CFLAGS -x c++ - -o /dev/null 2>/dev/null  <<EOF
+      #include <sched.h>
+      int main() {
+        int cpuid = sched_getcpu();
+      }
+EOF
+    if [ "$?" = 0 ]; then
+        COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_SCHED_GETCPU_PRESENT"
+    fi
 fi

 # TODO(tec): Fix -Wshorten-64-to-32 errors on FreeBSD and enable the warning.
@@ -412,14 +444,8 @@ if test "$USE_HDFS"; then
  JAVA_LDFLAGS="$JAVA_LDFLAGS $HDFS_LDFLAGS"
 fi

-if [ "$TARGET_OS" = FreeBSD -a "$TARGET_ARCHITECTURE" = i386 ]; then
-  # Intel SSE instructions breaks compilation on FreeBSD i386
-  unset USE_SSE
-fi
-
 if test "$USE_SSE"; then
-  # if Intel SSE instruction set is supported, set USE_SSE=1
-  COMMON_FLAGS="$COMMON_FLAGS -msse -msse4.2 "
+  COMMON_FLAGS="$COMMON_FLAGS -msse4.2"
 elif test -z "$PORTABLE"; then
  if test -n "`echo $TARGET_ARCHITECTURE | grep ^ppc64`"; then
    # Tune for this POWER processor, treating '+' models as base models
@@ -427,11 +453,41 @@ elif test -z "$PORTABLE"; then
    COMMON_FLAGS="$COMMON_FLAGS -mcpu=$POWER -mtune=$POWER "
  elif test -n "`echo $TARGET_ARCHITECTURE | grep ^s390x`"; then
    COMMON_FLAGS="$COMMON_FLAGS -march=z10 "
-  else
+  elif [ "$TARGET_OS" != AIX ] && [ "$TARGET_OS" != SunOS ]; then
    COMMON_FLAGS="$COMMON_FLAGS -march=native "
  fi
 fi

+$CXX $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null <<EOF
+  #include <cstdint>
+  #include <nmmintrin.h>
+  int main() {
+    volatile uint32_t x = _mm_crc32_u32(0, 0);
+  }
+EOF
+if [ "$?" = 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"
+fi
+
+# iOS doesn't support thread-local storage, but this check would erroneously
+# succeed because the cross-compiler flags are added by the Makefile, not this
+# script.
+if [ "$PLATFORM" != IOS ]; then
+  $CXX $COMMON_FLAGS -x c++ - -o /dev/null 2>/dev/null <<EOF
+  #if defined(_MSC_VER) && !defined(__thread)
+  #define __thread __declspec(thread)
+  #endif
+  int main() {
+    static __thread int tls;
+  }
+EOF
+  if [ "$?" = 0 ]; then
+    COMMON_FLAGS="$COMMON_FLAGS -DROCKSDB_SUPPORT_THREAD_LOCAL"
+  fi
+fi
+
 PLATFORM_CCFLAGS="$PLATFORM_CCFLAGS $COMMON_FLAGS"
 PLATFORM_CXXFLAGS="$PLATFORM_CXXFLAGS $COMMON_FLAGS"

@@ -1,18 +1,18 @@
-GCC_BASE=/mnt/gvfs/third-party2/gcc/cf7d14c625ce30bae1a4661c2319c5a283e4dd22/4.9.x/centos6-native/108cf83
-CLANG_BASE=/mnt/gvfs/third-party2/llvm-fb/8598c375b0e94e1448182eb3df034704144a838d/stable/centos6-native/3f16ddd
-LIBGCC_BASE=/mnt/gvfs/third-party2/libgcc/d6e0a7da6faba45f5e5b1638f9edd7afc2f34e7d/4.9.x/gcc-4.9-glibc-2.20/024dbc3
-GLIBC_BASE=/mnt/gvfs/third-party2/glibc/d282e6e8f3d20f4e40a516834847bdc038e07973/2.20/gcc-4.9-glibc-2.20/500e281
-SNAPPY_BASE=/mnt/gvfs/third-party2/snappy/8c38a4c1e52b4c2cc8a9cdc31b9c947ed7dbfcb4/1.1.3/gcc-4.9-glibc-2.20/e9936bf
-ZLIB_BASE=/mnt/gvfs/third-party2/zlib/0882df3713c7a84f15abe368dc004581f20b39d7/1.2.8/gcc-5-glibc-2.23/9bc6787
+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/lz4/0e790b441e2d9acd68d51e1d2e028f88c6a79ddf/r131/gcc-5-glibc-2.23/9bc6787
-ZSTD_BASE=/mnt/gvfs/third-party2/zstd/9455f75ff7f4831dc9fda02a6a0f8c68922fad8f/1.0.0/gcc-5-glibc-2.23/9bc6787
-GFLAGS_BASE=/mnt/gvfs/third-party2/gflags/f001a51b2854957676d07306ef3abf67186b5c8b/2.1.1/gcc-4.8.1-glibc-2.17/c3f970a
-JEMALLOC_BASE=/mnt/gvfs/third-party2/jemalloc/fc8a13ca1fffa4d0765c716c5a0b49f0c107518f/master/gcc-5-glibc-2.23/1c32b4b
+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/ad576de2a1ea560c4d3434304f0fc4e079bede42/trunk/gcc-5-glibc-2.23/b1847cb
-TBB_BASE=/mnt/gvfs/third-party2/tbb/9d9a554877d0c5bef330fe818ab7178806dd316a/4.0_update2/gcc-4.9-glibc-2.20/e9936bf
-KERNEL_HEADERS_BASE=/mnt/gvfs/third-party2/kernel-headers/7c111ff27e0c466235163f00f280a9d617c3d2ec/4.0.9-36_fbk5_2933_gd092e3f/gcc-5-glibc-2.23/da39a3e
-BINUTILS_BASE=/mnt/gvfs/third-party2/binutils/b7fd454c4b10c6a81015d4524ed06cdeab558490/2.26/centos6-native/da39a3e
-VALGRIND_BASE=/mnt/gvfs/third-party2/valgrind/d7f4d4d86674a57668e3a96f76f0e17dd0eb8765/3.10.0/gcc-4.9-glibc-2.20/e9936bf
-LUA_BASE=/mnt/gvfs/third-party2/lua/61e4abf5813bbc39bc4f548757ccfcadde175a48/5.2.3/gcc-4.9-glibc-2.20/690f0d7
+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
@@ -0,0 +1,18 @@
+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
@@ -1,7 +1,7 @@
-# Copyright (c) 2016-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.
+#  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).

 '''Filter for error messages in test output:
    - Receives merged stdout/stderr from test on stdin
@@ -56,7 +56,7 @@ if test -z $PIC_BUILD; then
 else
  GFLAGS_LIBS=" $GFLAGS_BASE/lib/libgflags_pic.a"
 fi
-CFLAGS+=" -DGFLAGS=google"
+CFLAGS+=" -DGFLAGS=gflags"

 # location of jemalloc
 JEMALLOC_INCLUDE=" -I $JEMALLOC_BASE/include/"
@@ -116,8 +116,8 @@ else
  KERNEL_HEADERS_INCLUDE="$KERNEL_HEADERS_BASE/include"

  CFLAGS+=" -B$BINUTILS/gold -nostdinc -nostdlib"
-  CFLAGS+=" -isystem $LIBGCC_BASE/include/c++/4.9.x "
-  CFLAGS+=" -isystem $LIBGCC_BASE/include/c++/4.9.x/x86_64-facebook-linux "
+  CFLAGS+=" -isystem $LIBGCC_BASE/include/c++/5.x "
+  CFLAGS+=" -isystem $LIBGCC_BASE/include/c++/5.x/x86_64-facebook-linux "
  CFLAGS+=" -isystem $GLIBC_INCLUDE"
  CFLAGS+=" -isystem $LIBGCC_INCLUDE"
  CFLAGS+=" -isystem $CLANG_INCLUDE"
@@ -128,13 +128,14 @@ else
 fi

 CFLAGS+=" $DEPS_INCLUDE"
-CFLAGS+=" -DROCKSDB_PLATFORM_POSIX -DROCKSDB_LIB_IO_POSIX -DROCKSDB_FALLOCATE_PRESENT -DROCKSDB_MALLOC_USABLE_SIZE"
+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 $TBB_LIBS"
-EXEC_LDFLAGS+=" -Wl,--dynamic-linker,/usr/local/fbcode/gcc-4.9-glibc-2.20/lib/ld.so"
+EXEC_LDFLAGS+=" -B$BINUTILS/gold"
+EXEC_LDFLAGS+=" -Wl,--dynamic-linker,/usr/local/fbcode/gcc-5-glibc-2.23/lib/ld.so"
 EXEC_LDFLAGS+=" $LIBUNWIND"
-EXEC_LDFLAGS+=" -Wl,-rpath=/usr/local/fbcode/gcc-4.9-glibc-2.20/lib"
+EXEC_LDFLAGS+=" -Wl,-rpath=/usr/local/fbcode/gcc-5-glibc-2.23/lib"
 # required by libtbb
 EXEC_LDFLAGS+=" -ldl"

@@ -93,7 +93,7 @@ else
 fi

 CFLAGS+=" $DEPS_INCLUDE"
-CFLAGS+=" -DROCKSDB_PLATFORM_POSIX -DROCKSDB_LIB_IO_POSIX -DROCKSDB_FALLOCATE_PRESENT -DROCKSDB_MALLOC_USABLE_SIZE"
+CFLAGS+=" -DROCKSDB_PLATFORM_POSIX -DROCKSDB_LIB_IO_POSIX -DROCKSDB_FALLOCATE_PRESENT -DROCKSDB_MALLOC_USABLE_SIZE -DROCKSDB_RANGESYNC_PRESENT -DROCKSDB_SCHED_GETCPU_PRESENT -DROCKSDB_SUPPORT_THREAD_LOCAL -DHAVE_SSE42"
 CFLAGS+=" -DSNAPPY -DGFLAGS=google -DZLIB -DBZIP2 -DLZ4 -DZSTD -DNUMA -DTBB"
 CXXFLAGS+=" $CFLAGS"

@@ -0,0 +1,157 @@
+#!/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
@@ -17,7 +17,9 @@ function title() {
 }

 usage="Create new RocksDB version and prepare it for the release process\n"
-usage+="USAGE: ./make_new_version.sh <version>"
+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
@@ -27,6 +29,11 @@ fi

 ROCKSDB_VERSION=$1

+REMOTE="origin"
+if [[ $# > 1 ]]; then
+  REMOTE=$2
+fi
+
 GIT_BRANCH=`git rev-parse --abbrev-ref HEAD`
 echo $GIT_BRANCH

@@ -41,6 +48,6 @@ $GIT checkout -b $BRANCH

 # Setting up the proxy for remote repo access
 title "Pushing new branch to remote repo ..."
-git push origin --set-upstream $BRANCH
+git push $REMOTE --set-upstream $BRANCH

 title "Branch $BRANCH is pushed to github;"
@@ -99,6 +99,18 @@ function main() {

  make static_lib
  make install INSTALL_PATH=package
+
+  cd package
+
+  LIB_DIR=lib
+  if [[ -z "$ARCH" ]]; then
+      ARCH=$(getconf LONG_BIT)
+  fi
+  if [[ ("$FPM_OUTPUT" = "rpm") && ($ARCH -eq 64) ]]; then
+      mv lib lib64
+      LIB_DIR=lib64
+  fi
+
  fpm \
    -s dir \
    -t $FPM_OUTPUT \
@@ -110,7 +122,7 @@ function main() {
    --license BSD \
    --vendor Facebook \
    --description "RocksDB is an embeddable persistent key-value store for fast storage." \
-    package
+    include $LIB_DIR
 }

 main $@
@@ -85,8 +85,9 @@ NON_SHM="TMPD=/tmp/rocksdb_test_tmp"
 GCC_481="ROCKSDB_FBCODE_BUILD_WITH_481=1"
 ASAN="COMPILE_WITH_ASAN=1"
 CLANG="USE_CLANG=1"
-LITE="OPT=\"-DROCKSDB_LITE -g\""
-TSAN="COMPILE_WITH_TSAN=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"
 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"
@@ -130,35 +131,6 @@ DISABLE_COMMANDS="[
    }
 ]"

-#
-# RocksDB unit test in parallel
-# Currently we always have noise in our parallel runs. This job is to help
-# manage the noise
-#
-PARALLEL_UNIT_TEST_COMMANDS="[
-    {
-        'name':'Rocksdb Parallel Unit Test',
-        'oncall':'$ONCALL',
-        'steps': [
-            $CLEANUP_ENV,
-            {
-                'name':'Build and test RocksDB debug version',
-                'shell':'$DEBUG make -j$(nproc) all && $SHM make check > /dev/null 2>&1 || $CONTRUN_NAME=punit_check $TASK_CREATION_TOOL || cat t/log-*',
-                'user':'root',
-                $PARSER
-            },
-            $CLEANUP_ENV,
-            {
-                'name':'Build and test RocksDB debug version under gcc-4.8.1',
-                'shell':'$GCC_481 $DEBUG make -j$(nproc) all && $SHM make check > /dev/null 2>&1 || $CONTRUN_NAME=punit_check_gcc481 $TASK_CREATION_TOOL || cat t/log-*',
-                'user':'root',
-                $PARSER
-            },
-        ],
-        $REPORT
-    }
-]"
-
 #
 # RocksDB unit test
 #
@@ -776,9 +748,6 @@ JAVA_BUILD_TEST_COMMANDS="[


 case $1 in
-  punit)
-    echo $PARALLEL_UNIT_TEST_COMMANDS
-    ;;
  unit)
    echo $UNIT_TEST_COMMANDS
    ;;
@@ -276,7 +276,7 @@ if($RunAll) {
      continue
    }

-    if($RunOnly.Contains($e)) {
+    if($RunOnly.Contains($bare_name)) {
      MakeAndAdd -token $e -HashTable $TestExes
    } else {
      $ListOfSuites += $bare_name
@@ -333,7 +333,7 @@ function RunJobs($Suites, $TestCmds, [int]$ConcurrencyVal)
            $Cases = @{}

            if($Suites.Count -gt 0) {
-              # Will teh first one
+              # Will the first one
              ForEach($e in $Suites.Keys) {
                $exe_name = $e
                $Cases = $Suites[$e]
@@ -53,7 +53,46 @@ function get_lib_base()
 }

 ###########################################################
-#                   4.9.x dependencies                    #
+#                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                      #
 ###########################################################

 OUTPUT="$BASEDIR/dependencies.sh"
@@ -64,16 +103,16 @@ touch "$OUTPUT"
 echo "Writing dependencies to $OUTPUT"

 # Compilers locations
-GCC_BASE=`readlink -f $TP2_LATEST/gcc/4.9.x/centos6-native/*/`
+GCC_BASE=`readlink -f $TP2_LATEST/gcc/5.x/centos6-native/*/`
 CLANG_BASE=`readlink -f $TP2_LATEST/llvm-fb/stable/centos6-native/*/`

 log_variable GCC_BASE
 log_variable CLANG_BASE

 # Libraries locations
-get_lib_base libgcc     4.9.x
-get_lib_base glibc      2.20
-get_lib_base snappy     LATEST gcc-4.9-glibc-2.20
+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
@@ -82,12 +121,12 @@ 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-4.9-glibc-2.20
+get_lib_base tbb        4.0_update2 gcc-5-glibc-2.23

 get_lib_base kernel-headers LATEST 
 get_lib_base binutils   LATEST centos6-native 
-get_lib_base valgrind   3.10.0 gcc-4.9-glibc-2.20
-get_lib_base lua        5.2.3 gcc-4.9-glibc-2.20
+get_lib_base valgrind   3.10.0 gcc-5-glibc-2.23
+get_lib_base lua        5.2.3 gcc-5-glibc-2.23

 git diff $OUTPUT

@@ -1,7 +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 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).

 #ifndef __STDC_FORMAT_MACROS
 #define __STDC_FORMAT_MACROS
@@ -1,7 +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 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
@@ -475,6 +475,37 @@ void deleter(const Slice& key, void* value) {
 }
 }  // namespace

+TEST_P(CacheTest, ReleaseAndErase) {
+  std::shared_ptr<Cache> cache = NewCache(5, 0, false);
+  Cache::Handle* handle;
+  Status s = cache->Insert(EncodeKey(100), EncodeValue(100), 1,
+                           &CacheTest::Deleter, &handle);
+  ASSERT_TRUE(s.ok());
+  ASSERT_EQ(5U, cache->GetCapacity());
+  ASSERT_EQ(1U, cache->GetUsage());
+  ASSERT_EQ(0U, deleted_keys_.size());
+  auto erased = cache->Release(handle, true);
+  ASSERT_TRUE(erased);
+  // This tests that deleter has been called
+  ASSERT_EQ(1U, deleted_keys_.size());
+}
+
+TEST_P(CacheTest, ReleaseWithoutErase) {
+  std::shared_ptr<Cache> cache = NewCache(5, 0, false);
+  Cache::Handle* handle;
+  Status s = cache->Insert(EncodeKey(100), EncodeValue(100), 1,
+                           &CacheTest::Deleter, &handle);
+  ASSERT_TRUE(s.ok());
+  ASSERT_EQ(5U, cache->GetCapacity());
+  ASSERT_EQ(1U, cache->GetUsage());
+  ASSERT_EQ(0U, deleted_keys_.size());
+  auto erased = cache->Release(handle);
+  ASSERT_FALSE(erased);
+  // This tests that deleter is not called. When cache has free capacity it is
+  // not expected to immediately erase the released items.
+  ASSERT_EQ(0U, deleted_keys_.size());
+}
+
 TEST_P(CacheTest, SetCapacity) {
  // test1: increase capacity
  // lets create a cache with capacity 5,
@@ -1,7 +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 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
@@ -26,7 +26,6 @@ std::shared_ptr<Cache> NewClockCache(size_t capacity, int num_shard_bits,
 #include <assert.h>
 #include <atomic>
 #include <deque>
-#include <limits>

 #include "tbb/concurrent_hash_map.h"

@@ -252,8 +251,11 @@ class ClockCacheShard : public CacheShard {
  //
  // Not necessary to hold mutex_ before being called.
  virtual bool Ref(Cache::Handle* handle) override;
-  virtual void Release(Cache::Handle* handle) override;
+  virtual bool Release(Cache::Handle* handle,
+                       bool force_erase = false) override;
  virtual void Erase(const Slice& key, uint32_t hash) override;
+  bool EraseAndConfirm(const Slice& key, uint32_t hash,
+                       CleanupContext* context);
  virtual size_t GetUsage() const override;
  virtual size_t GetPinnedUsage() const override;
  virtual void EraseUnRefEntries() override;
@@ -274,13 +276,17 @@ class ClockCacheShard : public CacheShard {
  // Decrease reference count of the entry. If this decreases the count to 0,
  // recycle the entry. If set_usage is true, also set the usage bit.
  //
+  // returns true if a value is erased.
+  //
  // Not necessary to hold mutex_ before being called.
-  void Unref(CacheHandle* handle, bool set_usage, CleanupContext* context);
+  bool Unref(CacheHandle* handle, bool set_usage, CleanupContext* context);

  // Unset in-cache bit of the entry. Recycle the handle if necessary.
  //
+  // returns true if a value is erased.
+  //
  // Has to hold mutex_ before being called.
-  void UnsetInCache(CacheHandle* handle, CleanupContext* context);
+  bool UnsetInCache(CacheHandle* handle, CleanupContext* context);

  // Put the handle back to recycle_ list, and put the value associated with
  // it into to-be-deleted list. It doesn't cleanup the key as it might be
@@ -432,7 +438,7 @@ bool ClockCacheShard::Ref(Cache::Handle* h) {
  return false;
 }

-void ClockCacheShard::Unref(CacheHandle* handle, bool set_usage,
+bool ClockCacheShard::Unref(CacheHandle* handle, bool set_usage,
                            CleanupContext* context) {
  if (set_usage) {
    handle->flags.fetch_or(kUsageBit, std::memory_order_relaxed);
@@ -451,9 +457,10 @@ void ClockCacheShard::Unref(CacheHandle* handle, bool set_usage,
      RecycleHandle(handle, context);
    }
  }
+  return context->to_delete_value.size();
 }

-void ClockCacheShard::UnsetInCache(CacheHandle* handle,
+bool ClockCacheShard::UnsetInCache(CacheHandle* handle,
                                   CleanupContext* context) {
  mutex_.AssertHeld();
  // Use acquire-release semantics as previous operations on the cache entry
@@ -465,6 +472,7 @@ void ClockCacheShard::UnsetInCache(CacheHandle* handle,
  if (InCache(flags) && CountRefs(flags) == 0) {
    RecycleHandle(handle, context);
  }
+  return context->to_delete_value.size();
 }

 bool ClockCacheShard::TryEvict(CacheHandle* handle, CleanupContext* context) {
@@ -618,27 +626,36 @@ Cache::Handle* ClockCacheShard::Lookup(const Slice& key, uint32_t hash) {
  return reinterpret_cast<Cache::Handle*>(handle);
 }

-void ClockCacheShard::Release(Cache::Handle* h) {
+bool ClockCacheShard::Release(Cache::Handle* h, bool force_erase) {
  CleanupContext context;
  CacheHandle* handle = reinterpret_cast<CacheHandle*>(h);
-  Unref(handle, true, &context);
+  bool erased = Unref(handle, true, &context);
+  if (force_erase && !erased) {
+    erased = EraseAndConfirm(handle->key, handle->hash, &context);
+  }
  Cleanup(context);
+  return erased;
 }

 void ClockCacheShard::Erase(const Slice& key, uint32_t hash) {
  CleanupContext context;
-  {
-    MutexLock l(&mutex_);
-    HashTable::accessor accessor;
-    if (table_.find(accessor, CacheKey(key, hash))) {
-      CacheHandle* handle = accessor->second;
-      table_.erase(accessor);
-      UnsetInCache(handle, &context);
-    }
-  }
+  EraseAndConfirm(key, hash, &context);
  Cleanup(context);
 }

+bool ClockCacheShard::EraseAndConfirm(const Slice& key, uint32_t hash,
+                                      CleanupContext* context) {
+  MutexLock l(&mutex_);
+  HashTable::accessor accessor;
+  bool erased = false;
+  if (table_.find(accessor, CacheKey(key, hash))) {
+    CacheHandle* handle = accessor->second;
+    table_.erase(accessor);
+    erased = UnsetInCache(handle, context);
+  }
+  return erased;
+}
+
 void ClockCacheShard::EraseUnRefEntries() {
  CleanupContext context;
  {
@@ -1,7 +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 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
@@ -1,7 +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 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
@@ -273,9 +273,9 @@ void LRUCacheShard::SetHighPriorityPoolRatio(double high_pri_pool_ratio) {
  MaintainPoolSize();
 }

-void LRUCacheShard::Release(Cache::Handle* handle) {
+bool LRUCacheShard::Release(Cache::Handle* handle, bool force_erase) {
  if (handle == nullptr) {
-    return;
+    return false;
  }
  LRUHandle* e = reinterpret_cast<LRUHandle*>(handle);
  bool last_reference = false;
@@ -287,10 +287,10 @@ void LRUCacheShard::Release(Cache::Handle* handle) {
    }
    if (e->refs == 1 && e->InCache()) {
      // The item is still in cache, and nobody else holds a reference to it
-      if (usage_ > capacity_) {
+      if (usage_ > capacity_ || force_erase) {
        // the cache is full
        // The LRU list must be empty since the cache is full
-        assert(lru_.next == &lru_);
+        assert(!(usage_ > capacity_) || lru_.next == &lru_);
        // take this opportunity and remove the item
        table_.Remove(e->key(), e->hash);
        e->SetInCache(false);
@@ -308,6 +308,7 @@ void LRUCacheShard::Release(Cache::Handle* handle) {
  if (last_reference) {
    e->Free();
  }
+  return last_reference;
 }

 Status LRUCacheShard::Insert(const Slice& key, uint32_t hash, void* value,
@@ -1,7 +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 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
@@ -178,7 +178,8 @@ class LRUCacheShard : public CacheShard {
                        Cache::Priority priority) override;
  virtual Cache::Handle* Lookup(const Slice& key, uint32_t hash) override;
  virtual bool Ref(Cache::Handle* handle) override;
-  virtual void Release(Cache::Handle* handle) override;
+  virtual bool Release(Cache::Handle* handle,
+                       bool force_erase = false) override;
  virtual void Erase(const Slice& key, uint32_t hash) override;

  // Although in some platforms the update of size_t is atomic, to make sure
@@ -1,7 +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 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 "cache/lru_cache.h"

@@ -1,7 +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 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
@@ -63,9 +63,9 @@ bool ShardedCache::Ref(Handle* handle) {
  return GetShard(Shard(hash))->Ref(handle);
 }

-void ShardedCache::Release(Handle* handle) {
+bool ShardedCache::Release(Handle* handle, bool force_erase) {
  uint32_t hash = GetHash(handle);
-  GetShard(Shard(hash))->Release(handle);
+  return GetShard(Shard(hash))->Release(handle, force_erase);
 }

 void ShardedCache::Erase(const Slice& key) {
@@ -1,7 +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 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
@@ -30,7 +30,7 @@ class CacheShard {
                        Cache::Handle** handle, Cache::Priority priority) = 0;
  virtual Cache::Handle* Lookup(const Slice& key, uint32_t hash) = 0;
  virtual bool Ref(Cache::Handle* handle) = 0;
-  virtual void Release(Cache::Handle* handle) = 0;
+  virtual bool Release(Cache::Handle* handle, bool force_erase = false) = 0;
  virtual void Erase(const Slice& key, uint32_t hash) = 0;
  virtual void SetCapacity(size_t capacity) = 0;
  virtual void SetStrictCapacityLimit(bool strict_capacity_limit) = 0;
@@ -65,7 +65,7 @@ class ShardedCache : public Cache {
                        Handle** handle, Priority priority) override;
  virtual Handle* Lookup(const Slice& key, Statistics* stats) override;
  virtual bool Ref(Handle* handle) override;
-  virtual void Release(Handle* handle) override;
+  virtual bool Release(Handle* handle, bool force_erase = false) override;
  virtual void Erase(const Slice& key) override;
  virtual uint64_t NewId() override;
  virtual size_t GetCapacity() const override;
@@ -1,7 +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 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
@@ -32,6 +32,7 @@
 #include "util/file_reader_writer.h"
 #include "util/filename.h"
 #include "util/stop_watch.h"
+#include "util/sync_point.h"

 namespace rocksdb {

@@ -44,9 +45,9 @@ TableBuilder* NewTableBuilder(
        int_tbl_prop_collector_factories,
    uint32_t column_family_id, const std::string& column_family_name,
    WritableFileWriter* file, const CompressionType compression_type,
-    const CompressionOptions& compression_opts,
-    int level,
-    const std::string* compression_dict, const bool skip_filters) {
+    const CompressionOptions& compression_opts, int level,
+    const std::string* compression_dict, const bool skip_filters,
+    const uint64_t creation_time) {
  assert((column_family_id ==
          TablePropertiesCollectorFactory::Context::kUnknownColumnFamily) ==
         column_family_name.empty());
@@ -54,7 +55,7 @@ TableBuilder* NewTableBuilder(
      TableBuilderOptions(ioptions, internal_comparator,
                          int_tbl_prop_collector_factories, compression_type,
                          compression_opts, compression_dict, skip_filters,
-                          column_family_name, level),
+                          column_family_name, level, creation_time),
      column_family_id, file);
 }

@@ -73,7 +74,8 @@ Status BuildTable(
    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) {
+    TableProperties* table_properties, int level,
+    const uint64_t creation_time) {
  assert((column_family_id ==
          TablePropertiesCollectorFactory::Context::kUnknownColumnFamily) ==
         column_family_name.empty());
@@ -103,6 +105,10 @@ Status BuildTable(
    unique_ptr<WritableFileWriter> file_writer;
    {
      unique_ptr<WritableFile> file;
+#ifndef NDEBUG
+      bool use_direct_writes = env_options.use_direct_writes;
+      TEST_SYNC_POINT_CALLBACK("BuildTable:create_file", &use_direct_writes);
+#endif  // !NDEBUG
      s = NewWritableFile(env, fname, &file, env_options);
      if (!s.ok()) {
        EventHelpers::LogAndNotifyTableFileCreationFinished(
@@ -118,7 +124,8 @@ Status BuildTable(
      builder = NewTableBuilder(
          ioptions, internal_comparator, int_tbl_prop_collector_factories,
          column_family_id, column_family_name, file_writer.get(), compression,
-          compression_opts, level);
+          compression_opts, level, nullptr /* compression_dict */,
+          false /* skip_filters */, creation_time);
    }

    MergeHelper merge(env, internal_comparator.user_comparator(),
@@ -172,7 +179,7 @@ Status BuildTable(
    // Finish and check for file errors
    if (s.ok() && !empty) {
      StopWatch sw(env, ioptions.statistics, TABLE_SYNC_MICROS);
-      file_writer->Sync(ioptions.use_fsync);
+      s = file_writer->Sync(ioptions.use_fsync);
    }
    if (s.ok() && !empty) {
      s = file_writer->Close();
@@ -180,6 +187,11 @@ Status BuildTable(

    if (s.ok() && !empty) {
      // Verify that the table is usable
+      // We set for_compaction to false and don't OptimizeForCompactionTableRead
+      // here because this is a special case after we finish the table building
+      // No matter whether use_direct_io_for_flush_and_compaction is true,
+      // we will regrad this verification as user reads since the goal is
+      // to cache it here for further user reads
      std::unique_ptr<InternalIterator> it(table_cache->NewIterator(
          ReadOptions(), env_options, internal_comparator, meta->fd,
          nullptr /* range_del_agg */, nullptr,
@@ -1,7 +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 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.
@@ -48,10 +48,9 @@ TableBuilder* NewTableBuilder(
        int_tbl_prop_collector_factories,
    uint32_t column_family_id, const std::string& column_family_name,
    WritableFileWriter* file, const CompressionType compression_type,
-    const CompressionOptions& compression_opts,
-    int level,
+    const CompressionOptions& compression_opts, int level,
    const std::string* compression_dict = nullptr,
-    const bool skip_filters = false);
+    const bool skip_filters = false, const uint64_t creation_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
@@ -77,6 +76,7 @@ extern Status BuildTable(
    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);
+    TableProperties* table_properties = nullptr, int level = -1,
+    const uint64_t creation_time = 0);

 }  // namespace rocksdb
@@ -1,7 +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 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
@@ -34,6 +34,9 @@
 #include "rocksdb/utilities/backupable_db.h"
 #include "rocksdb/utilities/write_batch_with_index.h"
 #include "utilities/merge_operators.h"
+#include "rocksdb/utilities/transaction.h"
+#include "rocksdb/utilities/transaction_db.h"
+#include "rocksdb/utilities/checkpoint.h"

 using rocksdb::BytewiseComparator;
 using rocksdb::Cache;
@@ -87,6 +90,12 @@ using rocksdb::RestoreOptions;
 using rocksdb::CompactRangeOptions;
 using rocksdb::RateLimiter;
 using rocksdb::NewGenericRateLimiter;
+using rocksdb::PinnableSlice;
+using rocksdb::TransactionDBOptions;
+using rocksdb::TransactionDB;
+using rocksdb::TransactionOptions;
+using rocksdb::Transaction;
+using rocksdb::Checkpoint;

 using std::shared_ptr;

@@ -125,6 +134,24 @@ struct rocksdb_envoptions_t      { EnvOptions        rep; };
 struct rocksdb_ingestexternalfileoptions_t  { IngestExternalFileOptions rep; };
 struct rocksdb_sstfilewriter_t   { SstFileWriter*    rep; };
 struct rocksdb_ratelimiter_t     { RateLimiter*      rep; };
+struct rocksdb_pinnableslice_t {
+  PinnableSlice rep;
+};
+struct rocksdb_transactiondb_options_t {
+  TransactionDBOptions rep;
+};
+struct rocksdb_transactiondb_t {
+  TransactionDB* rep;
+};
+struct rocksdb_transaction_options_t {
+  TransactionOptions rep;
+};
+struct rocksdb_transaction_t {
+  Transaction* rep;
+};
+struct rocksdb_checkpoint_t {
+  Checkpoint* rep;
+};

 struct rocksdb_compactionfiltercontext_t {
  CompactionFilter::Context rep;
@@ -537,6 +564,29 @@ void rocksdb_backup_engine_close(rocksdb_backup_engine_t* be) {
  delete be;
 }

+rocksdb_checkpoint_t* rocksdb_checkpoint_object_create(rocksdb_t* db,
+                                                       char** errptr) {
+  Checkpoint* checkpoint;
+  if (SaveError(errptr, Checkpoint::Create(db->rep, &checkpoint))) {
+    return nullptr;
+  }
+  rocksdb_checkpoint_t* result = new rocksdb_checkpoint_t;
+  result->rep = checkpoint;
+  return result;
+}
+
+void rocksdb_checkpoint_create(rocksdb_checkpoint_t* checkpoint,
+                               const char* checkpoint_dir,
+                               uint64_t log_size_for_flush, char** errptr) {
+  SaveError(errptr, checkpoint->rep->CreateCheckpoint(
+                        std::string(checkpoint_dir), log_size_for_flush));
+}
+
+void rocksdb_checkpoint_object_destroy(rocksdb_checkpoint_t* checkpoint) {
+  delete checkpoint->rep;
+  delete checkpoint;
+}
+
 void rocksdb_close(rocksdb_t* db) {
  delete db->rep;
  delete db;
@@ -1322,23 +1372,24 @@ 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;
@@ -1360,6 +1411,10 @@ void rocksdb_writebatch_rollback_to_save_point(rocksdb_writebatch_t* b,
  SaveError(errptr, b->rep.RollbackToSavePoint());
 }

+void rocksdb_writebatch_pop_save_point(rocksdb_writebatch_t* b, char** errptr) {
+  SaveError(errptr, b->rep.PopSavePoint());
+}
+
 rocksdb_writebatch_wi_t* rocksdb_writebatch_wi_create(size_t reserved_bytes, unsigned char overwrite_key) {
  rocksdb_writebatch_wi_t* b = new rocksdb_writebatch_wi_t;
  b->rep = new WriteBatchWithIndex(BytewiseComparator(), reserved_bytes, overwrite_key);
@@ -1579,18 +1634,6 @@ 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;
@@ -1983,6 +2026,10 @@ void rocksdb_options_set_max_open_files(rocksdb_options_t* opt, int n) {
  opt->rep.max_open_files = n;
 }

+void rocksdb_options_set_max_file_opening_threads(rocksdb_options_t* opt, int n) {
+  opt->rep.max_file_opening_threads = n;
+}
+
 void rocksdb_options_set_max_total_wal_size(rocksdb_options_t* opt, uint64_t n) {
  opt->rep.max_total_wal_size = n;
 }
@@ -2121,9 +2168,9 @@ void rocksdb_options_set_use_direct_reads(rocksdb_options_t* opt,
  opt->rep.use_direct_reads = v;
 }

-void rocksdb_options_set_use_direct_writes(rocksdb_options_t* opt,
-                                           unsigned char v) {
-  opt->rep.use_direct_writes = v;
+void rocksdb_options_set_use_direct_io_for_flush_and_compaction(
+    rocksdb_options_t* opt, unsigned char v) {
+  opt->rep.use_direct_io_for_flush_and_compaction = v;
 }

 void rocksdb_options_set_allow_mmap_reads(
@@ -2792,7 +2839,26 @@ void rocksdb_sstfilewriter_open(rocksdb_sstfilewriter_t* writer,
 void rocksdb_sstfilewriter_add(rocksdb_sstfilewriter_t* writer, const char* key,
                               size_t keylen, const char* val, size_t vallen,
                               char** errptr) {
-  SaveError(errptr, writer->rep->Add(Slice(key, keylen), Slice(val, vallen)));
+  SaveError(errptr, writer->rep->Put(Slice(key, keylen), Slice(val, vallen)));
+}
+
+void rocksdb_sstfilewriter_put(rocksdb_sstfilewriter_t* writer, const char* key,
+                               size_t keylen, const char* val, size_t vallen,
+                               char** errptr) {
+  SaveError(errptr, writer->rep->Put(Slice(key, keylen), Slice(val, vallen)));
+}
+
+void rocksdb_sstfilewriter_merge(rocksdb_sstfilewriter_t* writer,
+                                 const char* key, size_t keylen,
+                                 const char* val, size_t vallen,
+                                 char** errptr) {
+  SaveError(errptr, writer->rep->Merge(Slice(key, keylen), Slice(val, vallen)));
+}
+
+void rocksdb_sstfilewriter_delete(rocksdb_sstfilewriter_t* writer,
+                                  const char* key, size_t keylen,
+                                  char** errptr) {
+  SaveError(errptr, writer->rep->Delete(Slice(key, keylen)));
 }

 void rocksdb_sstfilewriter_finish(rocksdb_sstfilewriter_t* writer,
@@ -2889,20 +2955,21 @@ 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;
@@ -2911,19 +2978,6 @@ 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;
@@ -3080,8 +3134,270 @@ void rocksdb_delete_file_in_range_cf(
          (limit_key ? (b = Slice(limit_key, limit_key_len), &b) : nullptr)));
 }

+rocksdb_transactiondb_options_t* rocksdb_transactiondb_options_create() {
+  return new rocksdb_transactiondb_options_t;
+}
+
+void rocksdb_transactiondb_options_destroy(rocksdb_transactiondb_options_t* opt){
+  delete opt;
+}
+
+void rocksdb_transactiondb_options_set_max_num_locks(
+    rocksdb_transactiondb_options_t* opt, int64_t max_num_locks) {
+  opt->rep.max_num_locks = max_num_locks;
+}
+
+void rocksdb_transactiondb_options_set_num_stripes(
+    rocksdb_transactiondb_options_t* opt, size_t num_stripes) {
+  opt->rep.num_stripes = num_stripes;
+}
+
+void rocksdb_transactiondb_options_set_transaction_lock_timeout(
+    rocksdb_transactiondb_options_t* opt, int64_t txn_lock_timeout) {
+  opt->rep.transaction_lock_timeout = txn_lock_timeout;
+}
+
+void rocksdb_transactiondb_options_set_default_lock_timeout(
+    rocksdb_transactiondb_options_t* opt, int64_t default_lock_timeout) {
+  opt->rep.default_lock_timeout = default_lock_timeout;
+}
+
+rocksdb_transaction_options_t* rocksdb_transaction_options_create() {
+  return new rocksdb_transaction_options_t;
+}
+
+void rocksdb_transaction_options_destroy(rocksdb_transaction_options_t* opt) {
+  delete opt;
+}
+
+void rocksdb_transaction_options_set_set_snapshot(
+    rocksdb_transaction_options_t* opt, unsigned char v) {
+  opt->rep.set_snapshot = v;
+}
+
+void rocksdb_transaction_options_set_deadlock_detect(
+    rocksdb_transaction_options_t* opt, unsigned char v) {
+  opt->rep.deadlock_detect = v;
+}
+
+void rocksdb_transaction_options_set_lock_timeout(
+    rocksdb_transaction_options_t* opt, int64_t lock_timeout) {
+  opt->rep.lock_timeout = lock_timeout;
+}
+
+void rocksdb_transaction_options_set_expiration(
+    rocksdb_transaction_options_t* opt, int64_t expiration) {
+  opt->rep.expiration = expiration;
+}
+
+void rocksdb_transaction_options_set_deadlock_detect_depth(
+    rocksdb_transaction_options_t* opt, int64_t depth) {
+  opt->rep.deadlock_detect_depth = depth;
+}
+
+void rocksdb_transaction_options_set_max_write_batch_size(
+    rocksdb_transaction_options_t* opt, size_t size) {
+  opt->rep.max_write_batch_size = size;
+}
+
+rocksdb_transactiondb_t* rocksdb_transactiondb_open(
+    const rocksdb_options_t* options,
+    const rocksdb_transactiondb_options_t* txn_db_options, const char* name,
+    char** errptr) {
+  TransactionDB* txn_db;
+  if (SaveError(errptr, TransactionDB::Open(options->rep, txn_db_options->rep,
+                                            std::string(name), &txn_db))) {
+    return nullptr;
+  }
+  rocksdb_transactiondb_t* result = new rocksdb_transactiondb_t;
+  result->rep = txn_db;
+  return result;
+}
+
+const rocksdb_snapshot_t* rocksdb_transactiondb_create_snapshot(
+    rocksdb_transactiondb_t* txn_db) {
+  rocksdb_snapshot_t* result = new rocksdb_snapshot_t;
+  result->rep = txn_db->rep->GetSnapshot();
+  return result;
+}
+
+void rocksdb_transactiondb_release_snapshot(
+    rocksdb_transactiondb_t* txn_db, const rocksdb_snapshot_t* snapshot) {
+  txn_db->rep->ReleaseSnapshot(snapshot->rep);
+  delete snapshot;
+}
+
+rocksdb_transaction_t* rocksdb_transaction_begin(
+    rocksdb_transactiondb_t* txn_db,
+    const rocksdb_writeoptions_t* write_options,
+    const rocksdb_transaction_options_t* txn_options,
+    rocksdb_transaction_t* old_txn) {
+  if (old_txn == nullptr) {
+    rocksdb_transaction_t* result = new rocksdb_transaction_t;
+    result->rep = txn_db->rep->BeginTransaction(write_options->rep,
+                                                txn_options->rep, nullptr);
+    return result;
+  }
+  old_txn->rep = txn_db->rep->BeginTransaction(write_options->rep,
+                                                txn_options->rep, old_txn->rep);
+  return old_txn;
+}
+
+void rocksdb_transaction_commit(rocksdb_transaction_t* txn, char** errptr) {
+  SaveError(errptr, txn->rep->Commit());
+}
+
+void rocksdb_transaction_rollback(rocksdb_transaction_t* txn, char** errptr) {
+  SaveError(errptr, txn->rep->Rollback());
+}
+
+void rocksdb_transaction_destroy(rocksdb_transaction_t* txn) {
+  delete txn->rep;
+  delete txn;
+}
+
+//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,
+                              char** errptr) {
+  char* result = nullptr;
+  std::string tmp;
+  Status s = txn->rep->Get(options->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 outside a transaction
+char* rocksdb_transactiondb_get(
+    rocksdb_transactiondb_t* txn_db,
+    const rocksdb_readoptions_t* options,
+    const char* key, size_t klen,
+    size_t* vlen,
+    char** errptr){
+  char* result = nullptr;
+  std::string tmp;
+  Status s = txn_db->rep->Get(options->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;
+}
+
+// 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,
+                             char** errptr) {
+  SaveError(errptr, txn->rep->Put(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,
+                               size_t vlen, char** errptr) {
+  SaveError(errptr,
+            txn_db->rep->Put(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)));
+}
+
+// Delete a key outside a transaction
+void rocksdb_transactiondb_delete(rocksdb_transactiondb_t* txn_db,
+                                  const rocksdb_writeoptions_t* options,
+                                  const char* key, size_t klen, char** errptr) {
+  SaveError(errptr, txn_db->rep->Delete(options->rep, Slice(key, klen)));
+}
+
+// Create an iterator inside a transaction
+rocksdb_iterator_t* rocksdb_transaction_create_iterator(
+    rocksdb_transaction_t* txn, const rocksdb_readoptions_t* options) {
+  rocksdb_iterator_t* result = new rocksdb_iterator_t;
+  result->rep = txn->rep->GetIterator(options->rep);
+  return result;
+}
+
+void rocksdb_transactiondb_close(rocksdb_transactiondb_t* txn_db) {
+  delete txn_db->rep;
+  delete txn_db;
+}
+
+rocksdb_checkpoint_t* rocksdb_transactiondb_checkpoint_object_create(
+    rocksdb_transactiondb_t* txn_db, char** errptr) {
+  Checkpoint* checkpoint;
+  if (SaveError(errptr, Checkpoint::Create(txn_db->rep, &checkpoint))) {
+    return nullptr;
+  }
+  rocksdb_checkpoint_t* result = new rocksdb_checkpoint_t;
+  result->rep = checkpoint;
+  return result;
+}
+
 void rocksdb_free(void* ptr) { free(ptr); }

+rocksdb_pinnableslice_t* rocksdb_get_pinned(
+    rocksdb_t* db, const rocksdb_readoptions_t* options, const char* key,
+    size_t keylen, char** errptr) {
+  rocksdb_pinnableslice_t* v = new (rocksdb_pinnableslice_t);
+  Status s = db->rep->Get(options->rep, db->rep->DefaultColumnFamily(),
+                          Slice(key, keylen), &v->rep);
+  if (!s.ok()) {
+    delete (v);
+    if (!s.IsNotFound()) {
+      SaveError(errptr, s);
+    }
+    return NULL;
+  }
+  return v;
+}
+
+rocksdb_pinnableslice_t* rocksdb_get_pinned_cf(
+    rocksdb_t* db, const rocksdb_readoptions_t* options,
+    rocksdb_column_family_handle_t* column_family, const char* key,
+    size_t keylen, char** errptr) {
+  rocksdb_pinnableslice_t* v = new (rocksdb_pinnableslice_t);
+  Status s = db->rep->Get(options->rep, column_family->rep, Slice(key, keylen),
+                          &v->rep);
+  if (!s.ok()) {
+    delete v;
+    if (!s.IsNotFound()) {
+      SaveError(errptr, s);
+    }
+    return NULL;
+  }
+  return v;
+}
+
+void rocksdb_pinnableslice_destroy(rocksdb_pinnableslice_t* v) { delete v; }
+
+const char* rocksdb_pinnableslice_value(const rocksdb_pinnableslice_t* v,
+                                        size_t* vlen) {
+  if (!v) {
+    *vlen = 0;
+    return NULL;
+  }
+
+  *vlen = v->rep.size();
+  return v->rep.data();
+}
 }  // end extern "C"

 #endif  // !ROCKSDB_LITE
@@ -20,7 +20,7 @@
 // Can not use port/port.h macros as this is a c file
 #ifdef OS_WIN

-#include <Windows.h>
+#include <windows.h>

 #define snprintf _snprintf

@@ -40,6 +40,7 @@ const char* phase = "";
 static char dbname[200];
 static char sstfilename[200];
 static char dbbackupname[200];
+static char dbcheckpointname[200];

 static void StartPhase(const char* name) {
  fprintf(stderr, "=== Test %s\n", name);
@@ -126,6 +127,32 @@ static void CheckGetCF(
  Free(&val);
 }

+static void CheckPinGet(rocksdb_t* db, const rocksdb_readoptions_t* options,
+                        const char* key, const char* expected) {
+  char* err = NULL;
+  size_t val_len;
+  const char* val;
+  rocksdb_pinnableslice_t* p;
+  p = rocksdb_get_pinned(db, options, key, strlen(key), &err);
+  CheckNoError(err);
+  val = rocksdb_pinnableslice_value(p, &val_len);
+  CheckEqual(expected, val, val_len);
+  rocksdb_pinnableslice_destroy(p);
+}
+
+static void CheckPinGetCF(rocksdb_t* db, const rocksdb_readoptions_t* options,
+                          rocksdb_column_family_handle_t* handle,
+                          const char* key, const char* expected) {
+  char* err = NULL;
+  size_t val_len;
+  const char* val;
+  rocksdb_pinnableslice_t* p;
+  p = rocksdb_get_pinned_cf(db, options, handle, key, strlen(key), &err);
+  CheckNoError(err);
+  val = rocksdb_pinnableslice_value(p, &val_len);
+  CheckEqual(expected, val, val_len);
+  rocksdb_pinnableslice_destroy(p);
+}

 static void CheckIter(rocksdb_iterator_t* iter,
                      const char* key, const char* val) {
@@ -292,6 +319,34 @@ static char* MergeOperatorPartialMerge(
  return result;
 }

+static void CheckTxnGet(
+        rocksdb_transaction_t* txn,
+        const rocksdb_readoptions_t* options,
+        const char* key,
+        const char* expected) {
+        char* err = NULL;
+        size_t val_len;
+        char* val;
+        val = rocksdb_transaction_get(txn, options, 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,
+        const char* key,
+        const char* expected) {
+        char* err = NULL;
+        size_t val_len;
+        char* val;
+        val = rocksdb_transactiondb_get(txn_db, options, 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;
@@ -303,6 +358,10 @@ int main(int argc, char** argv) {
  rocksdb_readoptions_t* roptions;
  rocksdb_writeoptions_t* woptions;
  rocksdb_ratelimiter_t* rate_limiter;
+  rocksdb_transactiondb_t* txn_db;
+  rocksdb_transactiondb_options_t* txn_db_options;
+  rocksdb_transaction_t* txn;
+  rocksdb_transaction_options_t* txn_options;
  char* err = NULL;
  int run = -1;

@@ -316,6 +375,11 @@ int main(int argc, char** argv) {
           GetTempDir(),
           ((int) geteuid()));

+  snprintf(dbcheckpointname, sizeof(dbcheckpointname),
+           "%s/rocksdb_c_test-%d-checkpoint",
+           GetTempDir(),
+           ((int) geteuid()));
+
  snprintf(sstfilename, sizeof(sstfilename),
           "%s/rocksdb_c_test-%d-sst",
           GetTempDir(),
@@ -431,6 +495,36 @@ int main(int argc, char** argv) {
    rocksdb_backup_engine_close(be);
  }

+  StartPhase("checkpoint");
+  {
+    rocksdb_destroy_db(options, dbcheckpointname, &err);
+    CheckNoError(err);
+
+    rocksdb_checkpoint_t* checkpoint = rocksdb_checkpoint_object_create(db, &err);
+    CheckNoError(err);
+
+    rocksdb_checkpoint_create(checkpoint, dbcheckpointname, 0, &err);
+    CheckNoError(err);
+
+    // start a new database from the checkpoint
+    rocksdb_close(db);
+    rocksdb_options_set_error_if_exists(options, 0);
+    db = rocksdb_open(options, dbcheckpointname, &err);
+    CheckNoError(err);
+
+    CheckGet(db, roptions, "foo", "hello");
+
+    rocksdb_checkpoint_object_destroy(checkpoint);
+
+    rocksdb_close(db);
+    rocksdb_destroy_db(options, dbcheckpointname, &err);
+    CheckNoError(err);
+
+    db = rocksdb_open(options, dbname, &err);
+    CheckNoError(err);
+    rocksdb_options_set_error_if_exists(options, 1);
+  }
+
  StartPhase("compactall");
  rocksdb_compact_range(db, NULL, 0, NULL, 0);
  CheckGet(db, roptions, "foo", "hello");
@@ -475,11 +569,11 @@ int main(int argc, char** argv) {
    unlink(sstfilename);
    rocksdb_sstfilewriter_open(writer, sstfilename, &err);
    CheckNoError(err);
-    rocksdb_sstfilewriter_add(writer, "sstk1", 5, "v1", 2, &err);
+    rocksdb_sstfilewriter_put(writer, "sstk1", 5, "v1", 2, &err);
    CheckNoError(err);
-    rocksdb_sstfilewriter_add(writer, "sstk2", 5, "v2", 2, &err);
+    rocksdb_sstfilewriter_put(writer, "sstk2", 5, "v2", 2, &err);
    CheckNoError(err);
-    rocksdb_sstfilewriter_add(writer, "sstk3", 5, "v3", 2, &err);
+    rocksdb_sstfilewriter_put(writer, "sstk3", 5, "v3", 2, &err);
    CheckNoError(err);
    rocksdb_sstfilewriter_finish(writer, &err);
    CheckNoError(err);
@@ -496,11 +590,11 @@ int main(int argc, char** argv) {
    unlink(sstfilename);
    rocksdb_sstfilewriter_open(writer, sstfilename, &err);
    CheckNoError(err);
-    rocksdb_sstfilewriter_add(writer, "sstk2", 5, "v4", 2, &err);
+    rocksdb_sstfilewriter_put(writer, "sstk2", 5, "v4", 2, &err);
    CheckNoError(err);
-    rocksdb_sstfilewriter_add(writer, "sstk22", 6, "v5", 2, &err);
+    rocksdb_sstfilewriter_put(writer, "sstk22", 6, "v5", 2, &err);
    CheckNoError(err);
-    rocksdb_sstfilewriter_add(writer, "sstk3", 5, "v6", 2, &err);
+    rocksdb_sstfilewriter_put(writer, "sstk3", 5, "v6", 2, &err);
    CheckNoError(err);
    rocksdb_sstfilewriter_finish(writer, &err);
    CheckNoError(err);
@@ -587,10 +681,13 @@ int main(int argc, char** argv) {
  {
    rocksdb_writebatch_t* wb = rocksdb_writebatch_create();
    rocksdb_writebatch_set_save_point(wb);
+    rocksdb_writebatch_set_save_point(wb);
    const char* k_list[2] = {"z", "ap"};
    const size_t k_sizes[2] = {1, 2};
    const char* v_list[3] = {"x", "y", "z"};
    const size_t v_sizes[3] = {1, 1, 1};
+    rocksdb_writebatch_pop_save_point(wb, &err);
+    CheckNoError(err);
    rocksdb_writebatch_putv(wb, 2, k_list, k_sizes, 3, v_list, v_sizes);
    rocksdb_writebatch_rollback_to_save_point(wb, &err);
    CheckNoError(err);
@@ -786,6 +883,13 @@ int main(int argc, char** argv) {
    }
  }

+  StartPhase("pin_get");
+  {
+    CheckPinGet(db, roptions, "box", "c");
+    CheckPinGet(db, roptions, "foo", "hello");
+    CheckPinGet(db, roptions, "notfound", NULL);
+  }
+
  StartPhase("approximate_sizes");
  {
    int i;
@@ -997,11 +1101,13 @@ int main(int argc, char** argv) {
    CheckNoError(err);

    CheckGetCF(db, roptions, handles[1], "foo", "hello");
+    CheckPinGetCF(db, roptions, handles[1], "foo", "hello");

    rocksdb_delete_cf(db, woptions, handles[1], "foo", 3, &err);
    CheckNoError(err);

    CheckGetCF(db, roptions, handles[1], "foo", NULL);
+    CheckPinGetCF(db, roptions, handles[1], "foo", NULL);

    rocksdb_writebatch_t* wb = rocksdb_writebatch_create();
    rocksdb_writebatch_put_cf(wb, handles[1], "baz", 3, "a", 1);
@@ -1014,6 +1120,9 @@ int main(int argc, char** argv) {
    CheckGetCF(db, roptions, handles[1], "baz", NULL);
    CheckGetCF(db, roptions, handles[1], "bar", NULL);
    CheckGetCF(db, roptions, handles[1], "box", "c");
+    CheckPinGetCF(db, roptions, handles[1], "baz", NULL);
+    CheckPinGetCF(db, roptions, handles[1], "bar", NULL);
+    CheckPinGetCF(db, roptions, handles[1], "box", "c");
    rocksdb_writebatch_destroy(wb);

    const char* keys[3] = { "box", "box", "barfooxx" };
@@ -1225,14 +1334,98 @@ int main(int argc, char** argv) {
    }
  }

-  // Simple sanity check that setting memtable rep works.
-  StartPhase("memtable_reps");
+  StartPhase("transactions");
  {
-    // Create database with vector memtable.
    rocksdb_close(db);
    rocksdb_destroy_db(options, dbname, &err);
    CheckNoError(err);

+    // open a TransactionDB
+    txn_db_options = rocksdb_transactiondb_options_create();
+    txn_options = rocksdb_transaction_options_create();
+    rocksdb_options_set_create_if_missing(options, 1);
+    txn_db = rocksdb_transactiondb_open(options, txn_db_options, dbname, &err);
+    CheckNoError(err);
+
+    // put outside a transaction
+    rocksdb_transactiondb_put(txn_db, woptions, "foo", 3, "hello", 5, &err);
+    CheckNoError(err);
+    CheckTxnDBGet(txn_db, roptions, "foo", "hello");
+
+    // delete from outside transaction
+    rocksdb_transactiondb_delete(txn_db, woptions, "foo", 3, &err);
+    CheckNoError(err);
+    CheckTxnDBGet(txn_db, roptions, "foo", NULL);
+
+    // begin a transaction
+    txn = rocksdb_transaction_begin(txn_db, woptions, txn_options, NULL);
+    // put
+    rocksdb_transaction_put(txn, "foo", 3, "hello", 5, &err);
+    CheckNoError(err);
+    CheckTxnGet(txn, roptions, "foo", "hello");
+    // delete
+    rocksdb_transaction_delete(txn, "foo", 3, &err);
+    CheckNoError(err);
+    CheckTxnGet(txn, roptions, "foo", NULL);
+
+    rocksdb_transaction_put(txn, "foo", 3, "hello", 5, &err);
+    CheckNoError(err);
+
+    // read from outside transaction, before commit
+    CheckTxnDBGet(txn_db, roptions, "foo", NULL);
+
+    // commit
+    rocksdb_transaction_commit(txn, &err);
+    CheckNoError(err);
+
+    // read from outside transaction, after commit
+    CheckTxnDBGet(txn_db, roptions, "foo", "hello");
+
+    // reuse old transaction
+    txn = rocksdb_transaction_begin(txn_db, woptions, txn_options, txn);
+
+    // snapshot
+    const rocksdb_snapshot_t* snapshot;
+    snapshot = rocksdb_transactiondb_create_snapshot(txn_db);
+    rocksdb_readoptions_set_snapshot(roptions, snapshot);
+  
+    rocksdb_transactiondb_put(txn_db, woptions, "foo", 3, "hey", 3,  &err);
+    CheckNoError(err);
+
+    CheckTxnDBGet(txn_db, roptions, "foo", "hello");
+    rocksdb_readoptions_set_snapshot(roptions, NULL);
+    rocksdb_transactiondb_release_snapshot(txn_db, snapshot);
+    CheckTxnDBGet(txn_db, roptions, "foo", "hey");
+
+    // iterate
+    rocksdb_transaction_put(txn, "bar", 3, "hi", 2, &err);
+    rocksdb_iterator_t* iter = rocksdb_transaction_create_iterator(txn, roptions);
+    CheckCondition(!rocksdb_iter_valid(iter));
+    rocksdb_iter_seek_to_first(iter);
+    CheckCondition(rocksdb_iter_valid(iter));
+    CheckIter(iter, "bar", "hi");
+    rocksdb_iter_get_error(iter, &err);
+    CheckNoError(err);
+    rocksdb_iter_destroy(iter);
+
+    // rollback
+    rocksdb_transaction_rollback(txn, &err);
+    CheckNoError(err);
+    CheckTxnDBGet(txn_db, roptions, "bar", NULL);
+
+    // close and destroy
+    rocksdb_transaction_destroy(txn);
+    rocksdb_transactiondb_close(txn_db);
+    rocksdb_destroy_db(options, dbname, &err);
+    CheckNoError(err);
+    rocksdb_transaction_options_destroy(txn_options);
+    rocksdb_transactiondb_options_destroy(txn_db_options);
+  }
+
+  // Simple sanity check that setting memtable rep works.
+  StartPhase("memtable_reps");
+  {
+    // Create database with vector memtable.
    rocksdb_options_set_memtable_vector_rep(options);
    db = rocksdb_open(options, dbname, &err);
    CheckNoError(err);
@@ -1,7 +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 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
@@ -170,6 +170,12 @@ ColumnFamilyOptions SanitizeOptions(const ImmutableDBOptions& db_options,
      result.num_levels < 2) {
    result.num_levels = 2;
  }
+
+  if (result.compaction_style == kCompactionStyleUniversal &&
+      db_options.allow_ingest_behind && result.num_levels < 3) {
+    result.num_levels = 3;
+  }
+
  if (result.max_write_buffer_number < 2) {
    result.max_write_buffer_number = 2;
  }
@@ -196,7 +202,6 @@ ColumnFamilyOptions SanitizeOptions(const ImmutableDBOptions& db_options,
    result.num_levels = 1;
    // since we delete level0 files in FIFO compaction when there are too many
    // of them, these options don't really mean anything
-    result.level0_file_num_compaction_trigger = std::numeric_limits<int>::max();
    result.level0_slowdown_writes_trigger = std::numeric_limits<int>::max();
    result.level0_stop_writes_trigger = std::numeric_limits<int>::max();
  }
@@ -342,13 +347,14 @@ ColumnFamilyData::ColumnFamilyData(
      dummy_versions_(_dummy_versions),
      current_(nullptr),
      refs_(0),
+      initialized_(false),
      dropped_(false),
      internal_comparator_(cf_options.comparator),
      initial_cf_options_(SanitizeOptions(db_options, cf_options)),
      ioptions_(db_options, initial_cf_options_),
      mutable_cf_options_(initial_cf_options_),
-      is_delete_range_supported_(strcmp(cf_options.table_factory->Name(),
-                                        BlockBasedTableFactory().Name()) == 0),
+      is_delete_range_supported_(
+          cf_options.table_factory->IsDeleteRangeSupported()),
      write_buffer_manager_(write_buffer_manager),
      mem_(nullptr),
      imm_(ioptions_.min_write_buffer_number_to_merge,
@@ -544,7 +550,7 @@ std::unique_ptr<WriteControllerToken> SetupDelay(
    // If DB just falled into the stop condition, we need to further reduce
    // the write rate to avoid the stop condition.
    if (penalize_stop) {
-      // Penalize the near stop or stop condition by more agressive slowdown.
+      // Penalize the near stop or stop condition by more aggressive slowdown.
      // This is to provide the long term slowdown increase signal.
      // The penalty is more than the reward of recovering to the normal
      // condition.
@@ -722,7 +728,7 @@ void ColumnFamilyData::RecalculateWriteStallConditions(
              mutable_cf_options.level0_slowdown_writes_trigger)) {
        write_controller_token_ =
            write_controller->GetCompactionPressureToken();
-        ROCKS_LOG_WARN(
+        ROCKS_LOG_INFO(
            ioptions_.info_log,
            "[%s] Increasing compaction threads because we have %d level-0 "
            "files ",
@@ -736,7 +742,7 @@ void ColumnFamilyData::RecalculateWriteStallConditions(
        write_controller_token_ =
            write_controller->GetCompactionPressureToken();
        if (mutable_cf_options.soft_pending_compaction_bytes_limit > 0) {
-          ROCKS_LOG_WARN(
+          ROCKS_LOG_INFO(
              ioptions_.info_log,
              "[%s] Increasing compaction threads because of estimated pending "
              "compaction "
@@ -753,6 +759,12 @@ void ColumnFamilyData::RecalculateWriteStallConditions(
        uint64_t write_rate = write_controller->delayed_write_rate();
        write_controller->set_delayed_write_rate(static_cast<uint64_t>(
            static_cast<double>(write_rate) * kDelayRecoverSlowdownRatio));
+        // Set the low pri limit to be 1/4 the delayed write rate.
+        // Note we don't reset this value even after delay condition is relased.
+        // Low-pri rate will continue to apply if there is a compaction
+        // pressure.
+        write_controller->low_pri_rate_limiter()->SetBytesPerSecond(write_rate /
+                                                                    4);
      }
    }
    prev_compaction_needed_bytes_ = compaction_needed_bytes;
@@ -777,9 +789,8 @@ uint64_t ColumnFamilyData::GetTotalSstFilesSize() const {

 MemTable* ColumnFamilyData::ConstructNewMemtable(
    const MutableCFOptions& mutable_cf_options, SequenceNumber earliest_seq) {
-  assert(current_ != nullptr);
  return new MemTable(internal_comparator_, ioptions_, mutable_cf_options,
-                      write_buffer_manager_, earliest_seq);
+                      write_buffer_manager_, earliest_seq, id_);
 }

 void ColumnFamilyData::CreateNewMemtable(
@@ -1,7 +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 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
@@ -333,6 +333,10 @@ class ColumnFamilyData {
  void RecalculateWriteStallConditions(
      const MutableCFOptions& mutable_cf_options);

+  void set_initialized() { initialized_.store(true); }
+
+  bool initialized() const { return initialized_.load(); }
+
 private:
  friend class ColumnFamilySet;
  ColumnFamilyData(uint32_t id, const std::string& name,
@@ -349,6 +353,7 @@ class ColumnFamilyData {
  Version* current_;         // == dummy_versions->prev_

  std::atomic<int> refs_;      // outstanding references to ColumnFamilyData
+  std::atomic<bool> initialized_;
  bool dropped_;               // true if client dropped it

  const InternalKeyComparator internal_comparator_;
@@ -1,8 +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 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
@@ -55,7 +54,7 @@ class EnvCounter : public EnvWrapper {
  }

 private:
-  int num_new_writable_file_;
+  std::atomic<int> num_new_writable_file_;
 };

 class ColumnFamilyTest : public testing::Test {
@@ -303,6 +302,7 @@ class ColumnFamilyTest : public testing::Test {
        ASSERT_OK(Put(cf, key, RandomString(&rnd_, key_value_size - 10)));
      }
    }
+    db_->FlushWAL(false);
  }

 #ifndef ROCKSDB_LITE  // TEST functions in DB are not supported in lite
@@ -521,6 +521,28 @@ TEST_F(ColumnFamilyTest, DontReuseColumnFamilyID) {
  }
 }

+TEST_F(ColumnFamilyTest, CreateCFRaceWithGetAggProperty) {
+  Open();
+
+  rocksdb::SyncPoint::GetInstance()->LoadDependency(
+      {{"DBImpl::WriteOptionsFile:1",
+        "ColumnFamilyTest.CreateCFRaceWithGetAggProperty:1"},
+       {"ColumnFamilyTest.CreateCFRaceWithGetAggProperty:2",
+        "DBImpl::WriteOptionsFile:2"}});
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+  rocksdb::port::Thread thread([&] { CreateColumnFamilies({"one"}); });
+
+  TEST_SYNC_POINT("ColumnFamilyTest.CreateCFRaceWithGetAggProperty:1");
+  uint64_t pv;
+  db_->GetAggregatedIntProperty(DB::Properties::kEstimateTableReadersMem, &pv);
+  TEST_SYNC_POINT("ColumnFamilyTest.CreateCFRaceWithGetAggProperty:2");
+
+  thread.join();
+
+  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+}
+
 class FlushEmptyCFTestWithParam : public ColumnFamilyTest,
                                  public testing::WithParamInterface<bool> {
 public:
@@ -558,6 +580,7 @@ TEST_P(FlushEmptyCFTestWithParam, FlushEmptyCFTest) {
  Flush(0);
  ASSERT_OK(Put(1, "bar", "v3"));  // seqID 4
  ASSERT_OK(Put(1, "foo", "v4"));  // seqID 5
+  db_->FlushWAL(false);

  // Preserve file system state up to here to simulate a crash condition.
  fault_env->SetFilesystemActive(false);
@@ -620,6 +643,7 @@ TEST_P(FlushEmptyCFTestWithParam, FlushEmptyCFTest2) {
  // Write to log file D
  ASSERT_OK(Put(1, "bar", "v4"));  // seqID 7
  ASSERT_OK(Put(1, "bar", "v5"));  // seqID 8
+  db_->FlushWAL(false);
  // Preserve file system state up to here to simulate a crash condition.
  fault_env->SetFilesystemActive(false);
  std::vector<std::string> names;
@@ -676,6 +700,44 @@ TEST_F(ColumnFamilyTest, AddDrop) {
              std::vector<std::string>({"default", "four", "three"}));
 }

+TEST_F(ColumnFamilyTest, BulkAddDrop) {
+  constexpr int kNumCF = 1000;
+  ColumnFamilyOptions cf_options;
+  WriteOptions write_options;
+  Open();
+  std::vector<std::string> cf_names;
+  std::vector<ColumnFamilyHandle*> cf_handles;
+  for (int i = 1; i <= kNumCF; i++) {
+    cf_names.push_back("cf1-" + ToString(i));
+  }
+  ASSERT_OK(db_->CreateColumnFamilies(cf_options, cf_names, &cf_handles));
+  for (int i = 1; i <= kNumCF; i++) {
+    ASSERT_OK(db_->Put(write_options, cf_handles[i - 1], "foo", "bar"));
+  }
+  ASSERT_OK(db_->DropColumnFamilies(cf_handles));
+  std::vector<ColumnFamilyDescriptor> cf_descriptors;
+  for (auto* handle : cf_handles) {
+    delete handle;
+  }
+  cf_handles.clear();
+  for (int i = 1; i <= kNumCF; i++) {
+    cf_descriptors.emplace_back("cf2-" + ToString(i), ColumnFamilyOptions());
+  }
+  ASSERT_OK(db_->CreateColumnFamilies(cf_descriptors, &cf_handles));
+  for (int i = 1; i <= kNumCF; i++) {
+    ASSERT_OK(db_->Put(write_options, cf_handles[i - 1], "foo", "bar"));
+  }
+  ASSERT_OK(db_->DropColumnFamilies(cf_handles));
+  for (auto* handle : cf_handles) {
+    delete handle;
+  }
+  Close();
+  std::vector<std::string> families;
+  ASSERT_OK(DB::ListColumnFamilies(db_options_, dbname_, &families));
+  std::sort(families.begin(), families.end());
+  ASSERT_TRUE(families == std::vector<std::string>({"default"}));
+}
+
 TEST_F(ColumnFamilyTest, DropTest) {
  // first iteration - dont reopen DB before dropping
  // second iteration - reopen DB before dropping
@@ -1342,7 +1404,6 @@ TEST_F(ColumnFamilyTest, AutomaticAndManualCompactions) {
  ColumnFamilyOptions default_cf, one, two;
  db_options_.max_open_files = 20;  // only 10 files in file cache
  db_options_.max_background_compactions = 3;
-  db_options_.base_background_compactions = 3;

  default_cf.compaction_style = kCompactionStyleLevel;
  default_cf.num_levels = 3;
@@ -1366,6 +1427,9 @@ TEST_F(ColumnFamilyTest, AutomaticAndManualCompactions) {
  two.write_buffer_size = 100000;

  Reopen({default_cf, one, two});
+  // make sure all background compaction jobs can be scheduled
+  auto stop_token =
+      dbfull()->TEST_write_controler().GetCompactionPressureToken();

  bool cf_1_1 = true;
  rocksdb::SyncPoint::GetInstance()->LoadDependency(
@@ -1425,7 +1489,6 @@ TEST_F(ColumnFamilyTest, AutomaticAndManualCompactions) {
    ASSERT_NE("NOT_FOUND", Get(1, *key_iter));
    key_iter++;
  }
-  Close();
 }

 TEST_F(ColumnFamilyTest, ManualAndAutomaticCompactions) {
@@ -1434,7 +1497,6 @@ TEST_F(ColumnFamilyTest, ManualAndAutomaticCompactions) {
  ColumnFamilyOptions default_cf, one, two;
  db_options_.max_open_files = 20;  // only 10 files in file cache
  db_options_.max_background_compactions = 3;
-  db_options_.base_background_compactions = 3;

  default_cf.compaction_style = kCompactionStyleLevel;
  default_cf.num_levels = 3;
@@ -1458,6 +1520,9 @@ TEST_F(ColumnFamilyTest, ManualAndAutomaticCompactions) {
  two.write_buffer_size = 100000;

  Reopen({default_cf, one, two});
+  // 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) {
@@ -1520,7 +1585,6 @@ TEST_F(ColumnFamilyTest, ManualAndAutomaticCompactions) {
    ASSERT_NE("NOT_FOUND", Get(1, *key_iter));
    key_iter++;
  }
-  Close();
 }

 TEST_F(ColumnFamilyTest, SameCFManualManualCompactions) {
@@ -1529,7 +1593,6 @@ TEST_F(ColumnFamilyTest, SameCFManualManualCompactions) {
  ColumnFamilyOptions default_cf, one;
  db_options_.max_open_files = 20;  // only 10 files in file cache
  db_options_.max_background_compactions = 3;
-  db_options_.base_background_compactions = 3;

  default_cf.compaction_style = kCompactionStyleLevel;
  default_cf.num_levels = 3;
@@ -1548,6 +1611,9 @@ TEST_F(ColumnFamilyTest, SameCFManualManualCompactions) {
  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) {
@@ -1618,7 +1684,6 @@ TEST_F(ColumnFamilyTest, SameCFManualManualCompactions) {
    ASSERT_NE("NOT_FOUND", Get(1, *key_iter));
    key_iter++;
  }
-  Close();
 }

 TEST_F(ColumnFamilyTest, SameCFManualAutomaticCompactions) {
@@ -1627,7 +1692,6 @@ TEST_F(ColumnFamilyTest, SameCFManualAutomaticCompactions) {
  ColumnFamilyOptions default_cf, one;
  db_options_.max_open_files = 20;  // only 10 files in file cache
  db_options_.max_background_compactions = 3;
-  db_options_.base_background_compactions = 3;

  default_cf.compaction_style = kCompactionStyleLevel;
  default_cf.num_levels = 3;
@@ -1646,6 +1710,9 @@ TEST_F(ColumnFamilyTest, SameCFManualAutomaticCompactions) {
  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) {
@@ -1707,7 +1774,6 @@ TEST_F(ColumnFamilyTest, SameCFManualAutomaticCompactions) {
    ASSERT_NE("NOT_FOUND", Get(1, *key_iter));
    key_iter++;
  }
-  Close();
 }

 TEST_F(ColumnFamilyTest, SameCFManualAutomaticCompactionsLevel) {
@@ -1716,7 +1782,6 @@ TEST_F(ColumnFamilyTest, SameCFManualAutomaticCompactionsLevel) {
  ColumnFamilyOptions default_cf, one;
  db_options_.max_open_files = 20;  // only 10 files in file cache
  db_options_.max_background_compactions = 3;
-  db_options_.base_background_compactions = 3;

  default_cf.compaction_style = kCompactionStyleLevel;
  default_cf.num_levels = 3;
@@ -1735,6 +1800,9 @@ TEST_F(ColumnFamilyTest, SameCFManualAutomaticCompactionsLevel) {
  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" -- level style
  for (int i = 0; i < one.level0_file_num_compaction_trigger - 2; ++i) {
@@ -1796,7 +1864,6 @@ TEST_F(ColumnFamilyTest, SameCFManualAutomaticCompactionsLevel) {
    ASSERT_NE("NOT_FOUND", Get(1, *key_iter));
    key_iter++;
  }
-  Close();
 }

 // This test checks for automatic getting a conflict if there is a
@@ -1812,7 +1879,6 @@ TEST_F(ColumnFamilyTest, SameCFManualAutomaticConflict) {
  ColumnFamilyOptions default_cf, one;
  db_options_.max_open_files = 20;  // only 10 files in file cache
  db_options_.max_background_compactions = 3;
-  db_options_.base_background_compactions = 3;

  default_cf.compaction_style = kCompactionStyleLevel;
  default_cf.num_levels = 3;
@@ -1831,6 +1897,9 @@ TEST_F(ColumnFamilyTest, SameCFManualAutomaticConflict) {
  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) {
@@ -1915,8 +1984,6 @@ TEST_F(ColumnFamilyTest, SameCFManualAutomaticConflict) {
    ASSERT_NE("NOT_FOUND", Get(1, *key_iter));
    key_iter++;
  }
-
-  Close();
 }

 // In this test, we generate enough files to trigger automatic compactions.
@@ -1931,7 +1998,6 @@ TEST_F(ColumnFamilyTest, SameCFAutomaticManualCompactions) {
  ColumnFamilyOptions default_cf, one;
  db_options_.max_open_files = 20;  // only 10 files in file cache
  db_options_.max_background_compactions = 3;
-  db_options_.base_background_compactions = 3;

  default_cf.compaction_style = kCompactionStyleLevel;
  default_cf.num_levels = 3;
@@ -1950,6 +2016,9 @@ TEST_F(ColumnFamilyTest, SameCFAutomaticManualCompactions) {
  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();

  bool cf_1_1 = true;
  bool cf_1_2 = true;
@@ -2004,8 +2073,6 @@ TEST_F(ColumnFamilyTest, SameCFAutomaticManualCompactions) {
    ASSERT_NE("NOT_FOUND", Get(1, *key_iter));
    key_iter++;
  }
-
-  Close();
 }
 #endif  // !ROCKSDB_LITE

@@ -2458,7 +2525,6 @@ TEST_F(ColumnFamilyTest, CreateAndDropRace) {
 TEST_F(ColumnFamilyTest, WriteStallSingleColumnFamily) {
  const uint64_t kBaseRate = 800000u;
  db_options_.delayed_write_rate = kBaseRate;
-  db_options_.base_background_compactions = 2;
  db_options_.max_background_compactions = 6;

  Open({"default"});
@@ -2639,7 +2705,6 @@ TEST_F(ColumnFamilyTest, WriteStallSingleColumnFamily) {
 }

 TEST_F(ColumnFamilyTest, CompactionSpeedupSingleColumnFamily) {
-  db_options_.base_background_compactions = 2;
  db_options_.max_background_compactions = 6;
  Open({"default"});
  ColumnFamilyData* cfd =
@@ -2659,7 +2724,7 @@ TEST_F(ColumnFamilyTest, CompactionSpeedupSingleColumnFamily) {

  vstorage->TEST_set_estimated_compaction_needed_bytes(40);
  cfd->RecalculateWriteStallConditions(mutable_cf_options);
-  ASSERT_EQ(2, dbfull()->TEST_BGCompactionsAllowed());
+  ASSERT_EQ(1, dbfull()->TEST_BGCompactionsAllowed());

  vstorage->TEST_set_estimated_compaction_needed_bytes(50);
  cfd->RecalculateWriteStallConditions(mutable_cf_options);
@@ -2671,11 +2736,11 @@ TEST_F(ColumnFamilyTest, CompactionSpeedupSingleColumnFamily) {

  vstorage->TEST_set_estimated_compaction_needed_bytes(45);
  cfd->RecalculateWriteStallConditions(mutable_cf_options);
-  ASSERT_EQ(2, dbfull()->TEST_BGCompactionsAllowed());
+  ASSERT_EQ(1, dbfull()->TEST_BGCompactionsAllowed());

  vstorage->set_l0_delay_trigger_count(7);
  cfd->RecalculateWriteStallConditions(mutable_cf_options);
-  ASSERT_EQ(2, dbfull()->TEST_BGCompactionsAllowed());
+  ASSERT_EQ(1, dbfull()->TEST_BGCompactionsAllowed());

  vstorage->set_l0_delay_trigger_count(9);
  cfd->RecalculateWriteStallConditions(mutable_cf_options);
@@ -2683,7 +2748,7 @@ TEST_F(ColumnFamilyTest, CompactionSpeedupSingleColumnFamily) {

  vstorage->set_l0_delay_trigger_count(6);
  cfd->RecalculateWriteStallConditions(mutable_cf_options);
-  ASSERT_EQ(2, dbfull()->TEST_BGCompactionsAllowed());
+  ASSERT_EQ(1, dbfull()->TEST_BGCompactionsAllowed());

  // Speed up threshold = min(4 * 2, 4 + (12 - 4)/4) = 6
  mutable_cf_options.level0_file_num_compaction_trigger = 4;
@@ -2692,7 +2757,7 @@ TEST_F(ColumnFamilyTest, CompactionSpeedupSingleColumnFamily) {

  vstorage->set_l0_delay_trigger_count(5);
  cfd->RecalculateWriteStallConditions(mutable_cf_options);
-  ASSERT_EQ(2, dbfull()->TEST_BGCompactionsAllowed());
+  ASSERT_EQ(1, dbfull()->TEST_BGCompactionsAllowed());

  vstorage->set_l0_delay_trigger_count(7);
  cfd->RecalculateWriteStallConditions(mutable_cf_options);
@@ -2700,7 +2765,7 @@ TEST_F(ColumnFamilyTest, CompactionSpeedupSingleColumnFamily) {

  vstorage->set_l0_delay_trigger_count(3);
  cfd->RecalculateWriteStallConditions(mutable_cf_options);
-  ASSERT_EQ(2, dbfull()->TEST_BGCompactionsAllowed());
+  ASSERT_EQ(1, dbfull()->TEST_BGCompactionsAllowed());
 }

 TEST_F(ColumnFamilyTest, WriteStallTwoColumnFamilies) {
@@ -2779,7 +2844,6 @@ TEST_F(ColumnFamilyTest, WriteStallTwoColumnFamilies) {
 }

 TEST_F(ColumnFamilyTest, CompactionSpeedupTwoColumnFamilies) {
-  db_options_.base_background_compactions = 2;
  db_options_.max_background_compactions = 6;
  column_family_options_.soft_pending_compaction_bytes_limit = 200;
  column_family_options_.hard_pending_compaction_bytes_limit = 2000;
@@ -2807,11 +2871,11 @@ TEST_F(ColumnFamilyTest, CompactionSpeedupTwoColumnFamilies) {

  vstorage->TEST_set_estimated_compaction_needed_bytes(40);
  cfd->RecalculateWriteStallConditions(mutable_cf_options);
-  ASSERT_EQ(2, dbfull()->TEST_BGCompactionsAllowed());
+  ASSERT_EQ(1, dbfull()->TEST_BGCompactionsAllowed());

  vstorage->TEST_set_estimated_compaction_needed_bytes(60);
  cfd1->RecalculateWriteStallConditions(mutable_cf_options);
-  ASSERT_EQ(2, dbfull()->TEST_BGCompactionsAllowed());
+  ASSERT_EQ(1, dbfull()->TEST_BGCompactionsAllowed());
  cfd->RecalculateWriteStallConditions(mutable_cf_options);
  ASSERT_EQ(6, dbfull()->TEST_BGCompactionsAllowed());

@@ -2829,7 +2893,7 @@ TEST_F(ColumnFamilyTest, CompactionSpeedupTwoColumnFamilies) {

  vstorage1->TEST_set_estimated_compaction_needed_bytes(3);
  cfd1->RecalculateWriteStallConditions(mutable_cf_options);
-  ASSERT_EQ(2, dbfull()->TEST_BGCompactionsAllowed());
+  ASSERT_EQ(1, dbfull()->TEST_BGCompactionsAllowed());

  vstorage->set_l0_delay_trigger_count(9);
  cfd->RecalculateWriteStallConditions(mutable_cf_options);
@@ -2841,7 +2905,7 @@ TEST_F(ColumnFamilyTest, CompactionSpeedupTwoColumnFamilies) {

  vstorage->set_l0_delay_trigger_count(0);
  cfd->RecalculateWriteStallConditions(mutable_cf_options);
-  ASSERT_EQ(2, dbfull()->TEST_BGCompactionsAllowed());
+  ASSERT_EQ(1, dbfull()->TEST_BGCompactionsAllowed());
 }

 #ifndef ROCKSDB_LITE
@@ -1,7 +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 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).

 #ifndef ROCKSDB_LITE

@@ -1,7 +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 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).

 #ifndef ROCKSDB_LITE
 #include "db/compacted_db_impl.h"
@@ -1,7 +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 licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).

 #pragma once
 #ifndef ROCKSDB_LITE
@@ -1,7 +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 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
@@ -283,28 +283,32 @@ bool Compaction::KeyNotExistsBeyondOutputLevel(
  assert(input_version_ != nullptr);
  assert(level_ptrs != nullptr);
  assert(level_ptrs->size() == static_cast<size_t>(number_levels_));
-  assert(cfd_->ioptions()->compaction_style != kCompactionStyleFIFO);
-  if (cfd_->ioptions()->compaction_style == kCompactionStyleUniversal) {
-    return bottommost_level_;
-  }
-  // 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++) {
-    const std::vector<FileMetaData*>& files = input_vstorage_->LevelFiles(lvl);
-    for (; level_ptrs->at(lvl) < files.size(); level_ptrs->at(lvl)++) {
-      auto* f = files[level_ptrs->at(lvl)];
-      if (user_cmp->Compare(user_key, f->largest.user_key()) <= 0) {
-        // We've advanced far enough
-        if (user_cmp->Compare(user_key, f->smallest.user_key()) >= 0) {
-          // Key falls in this file's range, so definitely
-          // exists beyond output level
-          return false;
+  if (cfd_->ioptions()->compaction_style == kCompactionStyleLevel) {
+    if (output_level_ == 0) {
+      return false;
+    }
+    // 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++) {
+      const std::vector<FileMetaData*>& files =
+          input_vstorage_->LevelFiles(lvl);
+      for (; level_ptrs->at(lvl) < files.size(); level_ptrs->at(lvl)++) {
+        auto* f = files[level_ptrs->at(lvl)];
+        if (user_cmp->Compare(user_key, f->largest.user_key()) <= 0) {
+          // We've advanced far enough
+          if (user_cmp->Compare(user_key, f->smallest.user_key()) >= 0) {
+            // Key falls in this file's range, so definitely
+            // exists beyond output level
+            return false;
+          }
+          break;
        }
-        break;
      }
    }
+    return true;
+  } else {
+    return bottommost_level_;
  }
-  return true;
 }

 // Mark (or clear) each file that is being compacted
@@ -413,14 +417,15 @@ void Compaction::Summary(char* output, int len) {
 uint64_t Compaction::OutputFilePreallocationSize() const {
  uint64_t preallocation_size = 0;

-  if (cfd_->ioptions()->compaction_style == kCompactionStyleLevel ||
-      output_level() > 0) {
+  if (max_output_file_size_ != port::kMaxUint64 &&
+      (cfd_->ioptions()->compaction_style == kCompactionStyleLevel ||
+       output_level() > 0)) {
    preallocation_size = max_output_file_size_;
  } else {
-    // output_level() == 0
-    assert(num_input_levels() > 0);
-    for (const auto& f : inputs_[0].files) {
-      preallocation_size += f->fd.GetFileSize();
+    for (const auto& level_files : inputs_) {
+      for (const auto& file : level_files.files) {
+        preallocation_size += file->fd.GetFileSize();
+      }
    }
  }
  // Over-estimate slightly so we don't end up just barely crossing
@@ -458,4 +463,17 @@ bool Compaction::ShouldFormSubcompactions() const {
  }
 }

+uint64_t Compaction::MaxInputFileCreationTime() const {
+  uint64_t max_creation_time = 0;
+  for (const auto& file : inputs_[0].files) {
+    if (file->fd.table_reader != nullptr &&
+        file->fd.table_reader->GetTableProperties() != nullptr) {
+      uint64_t creation_time =
+          file->fd.table_reader->GetTableProperties()->creation_time;
+      max_creation_time = std::max(max_creation_time, creation_time);
+    }
+  }
+  return max_creation_time;
+}
+
 }  // namespace rocksdb
@@ -1,7 +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 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
@@ -99,7 +99,8 @@ class Compaction {
  // input level.
  // REQUIREMENT: "compaction_input_level" must be >= 0 and
  //              < "input_levels()"
-  const std::vector<FileMetaData*>* inputs(size_t compaction_input_level) {
+  const std::vector<FileMetaData*>* inputs(
+      size_t compaction_input_level) const {
    assert(compaction_input_level < inputs_.size());
    return &inputs_[compaction_input_level].files;
  }
@@ -240,6 +241,8 @@ class Compaction {

  uint64_t max_compaction_bytes() const { return max_compaction_bytes_; }

+  uint64_t MaxInputFileCreationTime() const;
+
 private:
  // mark (or clear) all files that are being compacted
  void MarkFilesBeingCompacted(bool mark_as_compacted);
@@ -1,7 +1,7 @@
 //  Copyright (c) 2016-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 licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).

 #pragma once

@@ -1,21 +1,44 @@
-// 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.
 //  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 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/compaction_iterator.h"
+#include "rocksdb/listener.h"
 #include "table/internal_iterator.h"

 namespace rocksdb {

+#ifndef ROCKSDB_LITE
+CompactionEventListener::CompactionListenerValueType fromInternalValueType(
+    ValueType vt) {
+  switch (vt) {
+    case kTypeDeletion:
+      return CompactionEventListener::CompactionListenerValueType::kDelete;
+    case kTypeValue:
+      return CompactionEventListener::CompactionListenerValueType::kValue;
+    case kTypeMerge:
+      return CompactionEventListener::CompactionListenerValueType::
+          kMergeOperand;
+    case kTypeSingleDeletion:
+      return CompactionEventListener::CompactionListenerValueType::
+          kSingleDelete;
+    case kTypeRangeDeletion:
+      return CompactionEventListener::CompactionListenerValueType::kRangeDelete;
+    default:
+      assert(false);
+      return CompactionEventListener::CompactionListenerValueType::kInvalid;
+  }
+}
+#endif  // ROCKSDB_LITE
+
 CompactionIterator::CompactionIterator(
    InternalIterator* input, const Comparator* cmp, MergeHelper* merge_helper,
    SequenceNumber last_sequence, std::vector<SequenceNumber>* snapshots,
    SequenceNumber earliest_write_conflict_snapshot, 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)
    : CompactionIterator(
          input, cmp, merge_helper, last_sequence, snapshots,
@@ -23,7 +46,7 @@ CompactionIterator::CompactionIterator(
          range_del_agg,
          std::unique_ptr<CompactionProxy>(
              compaction ? new CompactionProxy(compaction) : nullptr),
-          compaction_filter, shutting_down) {}
+          compaction_filter, compaction_listener, shutting_down) {}

 CompactionIterator::CompactionIterator(
    InternalIterator* input, const Comparator* cmp, MergeHelper* merge_helper,
@@ -32,6 +55,7 @@ CompactionIterator::CompactionIterator(
    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)
    : input_(input),
      cmp_(cmp),
@@ -43,7 +67,11 @@ CompactionIterator::CompactionIterator(
      range_del_agg_(range_del_agg),
      compaction_(std::move(compaction)),
      compaction_filter_(compaction_filter),
+#ifndef ROCKSDB_LITE
+      compaction_listener_(compaction_listener),
+#endif  // ROCKSDB_LITE
      shutting_down_(shutting_down),
+      ignore_snapshots_(false),
      merge_out_iter_(merge_helper_) {
  assert(compaction_filter_ == nullptr || compaction_ != nullptr);
  bottommost_level_ =
@@ -55,15 +83,17 @@ CompactionIterator::CompactionIterator(
  if (snapshots_->size() == 0) {
    // optimize for fast path if there are no snapshots
    visible_at_tip_ = true;
-    earliest_snapshot_ = last_sequence;
+    earliest_snapshot_ = kMaxSequenceNumber;
    latest_snapshot_ = 0;
  } else {
    visible_at_tip_ = false;
    earliest_snapshot_ = snapshots_->at(0);
    latest_snapshot_ = snapshots_->back();
  }
-  if (compaction_filter_ != nullptr && compaction_filter_->IgnoreSnapshots()) {
-    ignore_snapshots_ = true;
+  if (compaction_filter_ != nullptr) {
+    if (compaction_filter_->IgnoreSnapshots()) {
+      ignore_snapshots_ = true;
+    }
  } else {
    ignore_snapshots_ = false;
  }
@@ -188,6 +218,14 @@ void CompactionIterator::NextFromInput() {
      current_user_key_sequence_ = kMaxSequenceNumber;
      current_user_key_snapshot_ = 0;

+#ifndef ROCKSDB_LITE
+      if (compaction_listener_) {
+        compaction_listener_->OnCompaction(compaction_->level(), ikey_.user_key,
+                                           fromInternalValueType(ikey_.type),
+                                           value_, ikey_.sequence, true);
+      }
+#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_ ||
@@ -235,6 +273,14 @@ void CompactionIterator::NextFromInput() {
        }
      }
    } else {
+#ifndef ROCKSDB_LITE
+      if (compaction_listener_) {
+        compaction_listener_->OnCompaction(compaction_->level(), ikey_.user_key,
+                                           fromInternalValueType(ikey_.type),
+                                           value_, ikey_.sequence, false);
+      }
+#endif  // ROCKSDB_LITE
+
      // Update the current key to reflect the new sequence number/type without
      // copying the user key.
      // TODO(rven): Compaction filter does not process keys in this path
@@ -394,7 +440,7 @@ void CompactionIterator::NextFromInput() {
      // is the same as the visibility of a previous instance of the
      // same key, then this kv is not visible in any snapshot.
      // Hidden by an newer entry for same user key
-      // TODO: why not > ?
+      // TODO(noetzli): why not > ?
      //
      // Note: Dropping this key will not affect TransactionDB write-conflict
      // checking since there has already been a record returned for this key
@@ -500,7 +546,8 @@ void CompactionIterator::PrepareOutput() {

  // This is safe for TransactionDB write-conflict checking since transactions
  // only care about sequence number larger than any active snapshots.
-  if (bottommost_level_ && valid_ && ikey_.sequence < earliest_snapshot_ &&
+  if ((compaction_ != nullptr && !compaction_->allow_ingest_behind()) &&
+      bottommost_level_ && valid_ && ikey_.sequence <= earliest_snapshot_ &&
      ikey_.type != kTypeMerge &&
      !cmp_->Equal(compaction_->GetLargestUserKey(), ikey_.user_key)) {
    assert(ikey_.type != kTypeDeletion && ikey_.type != kTypeSingleDeletion);
@@ -1,9 +1,7 @@
-// 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.
 //  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 licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
 #pragma once

 #include <algorithm>
@@ -16,10 +14,13 @@
 #include "db/merge_helper.h"
 #include "db/pinned_iterators_manager.h"
 #include "db/range_del_aggregator.h"
+#include "options/cf_options.h"
 #include "rocksdb/compaction_filter.h"

 namespace rocksdb {

+class CompactionEventListener;
+
 class CompactionIterator {
 public:
  // A wrapper around Compaction. Has a much smaller interface, only what
@@ -44,6 +45,9 @@ class CompactionIterator {
    virtual Slice GetLargestUserKey() const {
      return compaction_->GetLargestUserKey();
    }
+    virtual bool allow_ingest_behind() const {
+      return compaction_->immutable_cf_options()->allow_ingest_behind;
+    }

   protected:
    CompactionProxy() = default;
@@ -60,6 +64,7 @@ class CompactionIterator {
                     RangeDelAggregator* range_del_agg,
                     const Compaction* compaction = nullptr,
                     const CompactionFilter* compaction_filter = nullptr,
+                     CompactionEventListener* compaction_listener = nullptr,
                     const std::atomic<bool>* shutting_down = nullptr);

  // Constructor with custom CompactionProxy, used for tests.
@@ -71,6 +76,7 @@ class CompactionIterator {
                     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);

  ~CompactionIterator();
@@ -124,6 +130,9 @@ class CompactionIterator {
  RangeDelAggregator* range_del_agg_;
  std::unique_ptr<CompactionProxy> compaction_;
  const CompactionFilter* compaction_filter_;
+#ifndef ROCKSDB_LITE
+  CompactionEventListener* compaction_listener_;
+#endif  // ROCKSDB_LITE
  const std::atomic<bool>* shutting_down_;
  bool bottommost_level_;
  bool valid_ = false;
@@ -1,7 +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 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/compaction_iterator.h"

@@ -154,6 +154,7 @@ class FakeCompaction : public CompactionIterator::CompactionProxy {
  virtual Slice GetLargestUserKey() const {
    return "\xff\xff\xff\xff\xff\xff\xff\xff\xff";
  }
+  virtual bool allow_ingest_behind() const { return false; }

  bool key_not_exists_beyond_output_level = false;
 };
@@ -189,7 +190,7 @@ class CompactionIteratorTest : public testing::Test {
    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, &shutting_down_));
+        std::move(compaction), filter, nullptr, &shutting_down_));
  }

  void AddSnapshot(SequenceNumber snapshot) { snapshots_.push_back(snapshot); }
@@ -1,7 +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 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
@@ -739,12 +739,23 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
    input->SeekToFirst();
  }

+  // we allow only 1 compaction event listener. Used by blob storage
+  CompactionEventListener* comp_event_listener = nullptr;
+#ifndef ROCKSDB_LITE
+  for (auto& celitr : cfd->ioptions()->listeners) {
+    comp_event_listener = celitr->GetCompactionEventListener();
+    if (comp_event_listener != nullptr) {
+      break;
+    }
+  }
+#endif  // ROCKSDB_LITE
+
  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,
-      shutting_down_));
+      comp_event_listener, shutting_down_));
  auto c_iter = sub_compact->c_iter.get();
  c_iter->SeekToFirst();
  if (c_iter->Valid() &&
@@ -1012,7 +1023,6 @@ Status CompactionJob::FinishCompactionOutputFile(
  uint64_t output_number = sub_compact->current_output()->meta.fd.GetNumber();
  assert(output_number != 0);

-  TableProperties table_properties;
  // Check for iterator errors
  Status s = input_status;
  auto meta = &sub_compact->current_output()->meta;
@@ -1068,10 +1078,32 @@ Status CompactionJob::FinishCompactionOutputFile(
  }
  sub_compact->outfile.reset();

+  if (s.ok() && current_entries == 0) {
+    // If there is nothing to output, no necessary to generate a sst file.
+    // This happens when the output level is bottom level, at the same time
+    // the sub_compact output nothing.
+    std::string fname = TableFileName(
+        db_options_.db_paths, meta->fd.GetNumber(), meta->fd.GetPathId());
+    env_->DeleteFile(fname);
+
+    // Also need to remove the file from outputs, or it will be added to the
+    // VersionEdit.
+    assert(!sub_compact->outputs.empty());
+    sub_compact->outputs.pop_back();
+    sub_compact->builder.reset();
+    sub_compact->current_output_file_size = 0;
+    return s;
+  }
+
  ColumnFamilyData* cfd = sub_compact->compaction->column_family_data();
  TableProperties tp;
  if (s.ok() && current_entries > 0) {
    // Verify that the table is usable
+    // We set for_compaction to false and don't OptimizeForCompactionTableRead
+    // here because this is a special case after we finish the table building
+    // No matter whether use_direct_io_for_flush_and_compaction is true,
+    // we will regrad this verification as user reads since the goal is
+    // to cache it here for further user reads
    InternalIterator* iter = cfd->table_cache()->NewIterator(
        ReadOptions(), env_options_, cfd->internal_comparator(), meta->fd,
        nullptr /* range_del_agg */, nullptr,
@@ -1115,12 +1147,22 @@ Status CompactionJob::FinishCompactionOutputFile(
                            meta->fd.GetPathId());
    sfm->OnAddFile(fn);
    if (sfm->IsMaxAllowedSpaceReached()) {
+      // TODO(ajkr): should we return OK() if max space was reached by the final
+      // compaction output file (similarly to how flush works when full)?
+      s = Status::IOError("Max allowed space was reached");
+      TEST_SYNC_POINT(
+          "CompactionJob::FinishCompactionOutputFile:"
+          "MaxAllowedSpaceReached");
      InstrumentedMutexLock l(db_mutex_);
      if (db_bg_error_->ok()) {
-        s = Status::IOError("Max allowed space was reached");
-        *db_bg_error_ = s;
-        TEST_SYNC_POINT(
-            "CompactionJob::FinishCompactionOutputFile:MaxAllowedSpaceReached");
+        Status new_bg_error = s;
+        // may temporarily unlock and lock the mutex.
+        EventHelpers::NotifyOnBackgroundError(
+            cfd->ioptions()->listeners, BackgroundErrorReason::kCompaction,
+            &new_bg_error, db_mutex_);
+        if (!new_bg_error.ok()) {
+          *db_bg_error_ = new_bg_error;
+        }
      }
    }
  }
@@ -1198,7 +1240,11 @@ Status CompactionJob::OpenCompactionOutputFile(
 #endif  // !ROCKSDB_LITE
  // Make the output file
  unique_ptr<WritableFile> writable_file;
-  Status s = NewWritableFile(env_, fname, &writable_file, env_options_);
+  EnvOptions opt_env_opts =
+      env_->OptimizeForCompactionTableWrite(env_options_, db_options_);
+  TEST_SYNC_POINT_CALLBACK("CompactionJob::OpenCompactionOutputFile",
+                           &opt_env_opts.use_direct_writes);
+  Status s = NewWritableFile(env_, fname, &writable_file, opt_env_opts);
  if (!s.ok()) {
    ROCKS_LOG_ERROR(
        db_options_.info_log,
@@ -1231,14 +1277,22 @@ Status CompactionJob::OpenCompactionOutputFile(
  // data is going to be found
  bool skip_filters =
      cfd->ioptions()->optimize_filters_for_hits && bottommost_level_;
+
+  uint64_t output_file_creation_time =
+      sub_compact->compaction->MaxInputFileCreationTime();
+  if (output_file_creation_time == 0) {
+    int64_t _current_time = 0;
+    db_options_.env->GetCurrentTime(&_current_time);  // ignore error
+    output_file_creation_time = static_cast<uint64_t>(_current_time);
+  }
+
  sub_compact->builder.reset(NewTableBuilder(
      *cfd->ioptions(), cfd->internal_comparator(),
      cfd->int_tbl_prop_collector_factories(), cfd->GetID(), cfd->GetName(),
      sub_compact->outfile.get(), sub_compact->compaction->output_compression(),
      cfd->ioptions()->compression_opts,
-      sub_compact->compaction->output_level(),
-      &sub_compact->compression_dict,
-      skip_filters));
+      sub_compact->compaction->output_level(), &sub_compact->compression_dict,
+      skip_filters, output_file_creation_time));
  LogFlush(db_options_.info_log);
  return s;
 }
@@ -1,7 +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 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
@@ -1,7 +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 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
@@ -654,7 +654,6 @@ TEST_P(CompactionJobStatsTest, CompactionJobStatsTest) {
  Options options;
  options.listeners.emplace_back(stats_checker);
  options.create_if_missing = true;
-  options.max_background_flushes = 0;
  // just enough setting to hold off auto-compaction.
  options.level0_file_num_compaction_trigger = kTestScale + 1;
  options.num_levels = 3;
@@ -876,7 +875,6 @@ TEST_P(CompactionJobStatsTest, DeletionStatsTest) {
  Options options;
  options.listeners.emplace_back(stats_checker);
  options.create_if_missing = true;
-  options.max_background_flushes = 0;
  options.level0_file_num_compaction_trigger = kTestScale+1;
  options.num_levels = 3;
  options.compression = kNoCompression;
@@ -1,7 +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 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).

 #ifndef ROCKSDB_LITE

@@ -142,6 +142,7 @@ class CompactionJobTest : public testing::Test {
  }

  void SetLastSequence(const SequenceNumber sequence_number) {
+    versions_->SetLastToBeWrittenSequence(sequence_number + 1);
    versions_->SetLastSequence(sequence_number + 1);
  }

@@ -337,6 +338,24 @@ TEST_F(CompactionJobTest, SimpleDeletion) {
  RunCompaction({files}, expected_results);
 }

+TEST_F(CompactionJobTest, OutputNothing) {
+  NewDB();
+
+  auto file1 = mock::MakeMockFile({{KeyStr("a", 1U, kTypeValue), "val"}});
+
+  AddMockFile(file1);
+
+  auto file2 = mock::MakeMockFile({{KeyStr("a", 2U, kTypeDeletion), ""}});
+
+  AddMockFile(file2);
+
+  auto expected_results = mock::MakeMockFile();
+
+  SetLastSequence(4U);
+  auto files = cfd_->current()->storage_info()->LevelFiles(0);
+  RunCompaction({files}, expected_results);
+}
+
 TEST_F(CompactionJobTest, SimpleOverwrite) {
  NewDB();

@@ -1,7 +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 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
@@ -36,6 +36,39 @@ uint64_t TotalCompensatedFileSize(const std::vector<FileMetaData*>& files) {
  }
  return sum;
 }
+
+bool FindIntraL0Compaction(const std::vector<FileMetaData*>& level_files,
+                           size_t min_files_to_compact,
+                           uint64_t max_compact_bytes_per_del_file,
+                           CompactionInputFiles* comp_inputs) {
+  size_t compact_bytes = level_files[0]->fd.file_size;
+  size_t compact_bytes_per_del_file = port::kMaxSizet;
+  // compaction range will be [0, span_len).
+  size_t span_len;
+  // pull in files until the amount of compaction work per deleted file begins
+  // increasing.
+  size_t new_compact_bytes_per_del_file = 0;
+  for (span_len = 1; span_len < level_files.size(); ++span_len) {
+    compact_bytes += level_files[span_len]->fd.file_size;
+    new_compact_bytes_per_del_file = compact_bytes / span_len;
+    if (level_files[span_len]->being_compacted ||
+        new_compact_bytes_per_del_file > compact_bytes_per_del_file) {
+      break;
+    }
+    compact_bytes_per_del_file = new_compact_bytes_per_del_file;
+  }
+
+  if (span_len >= min_files_to_compact &&
+      new_compact_bytes_per_del_file < max_compact_bytes_per_del_file) {
+    assert(comp_inputs != nullptr);
+    comp_inputs->level = 0;
+    for (size_t i = 0; i < span_len; ++i) {
+      comp_inputs->files.push_back(level_files[i]);
+    }
+    return true;
+  }
+  return false;
+}
 }  // anonymous namespace

 // Determine compression type, based on user options, level of the output
@@ -494,7 +527,11 @@ Compaction* CompactionPicker::CompactRange(
    // files together to the last level.
    assert(vstorage->num_levels() > 1);
    // DBImpl::CompactRange() set output level to be the last level
-    assert(output_level == vstorage->num_levels() - 1);
+    if (ioptions_.allow_ingest_behind) {
+      assert(output_level == vstorage->num_levels() - 2);
+    } else {
+      assert(output_level == vstorage->num_levels() - 1);
+    }
    // DBImpl::RunManualCompaction will make full range for universal compaction
    assert(begin == nullptr);
    assert(end == nullptr);
@@ -1346,31 +1383,8 @@ bool LevelCompactionBuilder::PickIntraL0Compaction() {
    // resort to L0->L0 compaction yet.
    return false;
  }
-
-  size_t compact_bytes = level_files[0]->fd.file_size;
-  size_t compact_bytes_per_del_file = port::kMaxSizet;
-  // compaction range will be [0, span_len).
-  size_t span_len;
-  // pull in files until the amount of compaction work per deleted file begins
-  // increasing.
-  for (span_len = 1; span_len < level_files.size(); ++span_len) {
-    compact_bytes += level_files[span_len]->fd.file_size;
-    size_t new_compact_bytes_per_del_file = compact_bytes / span_len;
-    if (level_files[span_len]->being_compacted ||
-        new_compact_bytes_per_del_file > compact_bytes_per_del_file) {
-      break;
-    }
-    compact_bytes_per_del_file = new_compact_bytes_per_del_file;
-  }
-
-  if (span_len >= kMinFilesForIntraL0Compaction) {
-    start_level_inputs_.level = 0;
-    for (size_t i = 0; i < span_len; ++i) {
-      start_level_inputs_.files.push_back(level_files[i]);
-    }
-    return true;
-  }
-  return false;
+  return FindIntraL0Compaction(level_files, kMinFilesForIntraL0Compaction,
+                               port::kMaxUint64, &start_level_inputs_);
 }
 }  // namespace

@@ -1389,20 +1403,111 @@ bool FIFOCompactionPicker::NeedsCompaction(
  return vstorage->CompactionScore(kLevel0) >= 1;
 }

-Compaction* FIFOCompactionPicker::PickCompaction(
+namespace {
+uint64_t GetTotalFilesSize(
+    const std::vector<FileMetaData*>& files) {
+  uint64_t total_size = 0;
+  for (const auto& f : files) {
+    total_size += f->fd.file_size;
+  }
+  return total_size;
+}
+}  // anonymous namespace
+
+Compaction* FIFOCompactionPicker::PickTTLCompaction(
    const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
    VersionStorageInfo* vstorage, LogBuffer* log_buffer) {
-  assert(vstorage->num_levels() == 1);
+  assert(ioptions_.compaction_options_fifo.ttl > 0);
+
  const int kLevel0 = 0;
  const std::vector<FileMetaData*>& level_files = vstorage->LevelFiles(kLevel0);
-  uint64_t total_size = 0;
-  for (const auto& file : level_files) {
-    total_size += file->fd.file_size;
+  uint64_t total_size = GetTotalFilesSize(level_files);
+
+  int64_t _current_time;
+  auto status = ioptions_.env->GetCurrentTime(&_current_time);
+  if (!status.ok()) {
+    ROCKS_LOG_BUFFER(log_buffer,
+                     "[%s] FIFO compaction: Couldn't get current time: %s. "
+                     "Not doing compactions based on TTL. ",
+                     cf_name.c_str(), status.ToString().c_str());
+    return nullptr;
  }
+  const uint64_t current_time = static_cast<uint64_t>(_current_time);
+
+  std::vector<CompactionInputFiles> inputs;
+  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;
+      }
+      total_size -= f->compensated_file_size;
+      inputs[0].files.push_back(f);
+    }
+  }
+
+  // Return a nullptr and proceed to size-based FIFO compaction if:
+  // 1. there are no files older than ttl OR
+  // 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) {
+    return nullptr;
+  }
+
+  for (const auto& f : inputs[0].files) {
+    ROCKS_LOG_BUFFER(log_buffer,
+                     "[%s] FIFO compaction: picking file %" PRIu64
+                     " with creation time %" PRIu64 " for deletion",
+                     cf_name.c_str(), f->fd.GetNumber(),
+                     f->fd.table_reader->GetTableProperties()->creation_time);
+  }
+
+  Compaction* c = new Compaction(
+      vstorage, ioptions_, mutable_cf_options, std::move(inputs), 0, 0, 0, 0,
+      kNoCompression, {}, /* is manual */ false, vstorage->CompactionScore(0),
+      /* is deletion compaction */ true, CompactionReason::kFIFOTtl);
+  return c;
+}
+
+Compaction* FIFOCompactionPicker::PickSizeCompaction(
+    const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
+    VersionStorageInfo* vstorage, LogBuffer* log_buffer) {
+  const int kLevel0 = 0;
+  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 ||
      level_files.size() == 0) {
    // total size not exceeded
+    if (ioptions_.compaction_options_fifo.allow_compaction &&
+        level_files.size() > 0) {
+      CompactionInputFiles comp_inputs;
+      if (FindIntraL0Compaction(
+              level_files,
+              mutable_cf_options
+                  .level0_file_num_compaction_trigger /* min_files_to_compact */,
+              mutable_cf_options.write_buffer_size, &comp_inputs)) {
+        Compaction* c = new Compaction(
+            vstorage, ioptions_, mutable_cf_options, {comp_inputs}, 0,
+            16 * 1024 * 1024 /* output file size limit */,
+            0 /* max compaction bytes, not applicable */,
+            0 /* output path ID */, mutable_cf_options.compression, {},
+            /* is manual */ false, vstorage->CompactionScore(0),
+            /* is deletion compaction */ false,
+            CompactionReason::kFIFOReduceNumFiles);
+        return c;
+      }
+    }
+
    ROCKS_LOG_BUFFER(log_buffer,
                     "[%s] FIFO compaction: nothing to do. Total size %" PRIu64
                     ", max size %" PRIu64 "\n",
@@ -1423,24 +1528,41 @@ Compaction* FIFOCompactionPicker::PickCompaction(
  std::vector<CompactionInputFiles> inputs;
  inputs.emplace_back();
  inputs[0].level = 0;
-  // delete old files (FIFO)
+
  for (auto ritr = level_files.rbegin(); ritr != level_files.rend(); ++ritr) {
    auto f = *ritr;
    total_size -= f->compensated_file_size;
    inputs[0].files.push_back(f);
    char tmp_fsize[16];
    AppendHumanBytes(f->fd.GetFileSize(), tmp_fsize, sizeof(tmp_fsize));
-    ROCKS_LOG_BUFFER(log_buffer, "[%s] FIFO compaction: picking file %" PRIu64
-                                 " with size %s for deletion",
+    ROCKS_LOG_BUFFER(log_buffer,
+                     "[%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) {
      break;
    }
  }
+
  Compaction* c = new Compaction(
      vstorage, ioptions_, mutable_cf_options, std::move(inputs), 0, 0, 0, 0,
      kNoCompression, {}, /* is manual */ false, vstorage->CompactionScore(0),
      /* is deletion compaction */ true, CompactionReason::kFIFOMaxSize);
+  return c;
+}
+
+Compaction* FIFOCompactionPicker::PickCompaction(
+    const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
+    VersionStorageInfo* vstorage, LogBuffer* log_buffer) {
+  assert(vstorage->num_levels() == 1);
+
+  Compaction* c = nullptr;
+  if (ioptions_.compaction_options_fifo.ttl > 0) {
+    c = PickTTLCompaction(cf_name, mutable_cf_options, vstorage, log_buffer);
+  }
+  if (c == nullptr) {
+    c = PickSizeCompaction(cf_name, mutable_cf_options, vstorage, log_buffer);
+  }
  RegisterCompaction(c);
  return c;
 }
@@ -1,7 +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 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
@@ -242,6 +242,17 @@ class FIFOCompactionPicker : public CompactionPicker {

  virtual bool NeedsCompaction(
      const VersionStorageInfo* vstorage) const override;
+
+ private:
+  Compaction* PickTTLCompaction(const std::string& cf_name,
+                                const MutableCFOptions& mutable_cf_options,
+                                VersionStorageInfo* version,
+                                LogBuffer* log_buffer);
+
+  Compaction* PickSizeCompaction(const std::string& cf_name,
+                                 const MutableCFOptions& mutable_cf_options,
+                                 VersionStorageInfo* version,
+                                 LogBuffer* log_buffer);
 };

 class NullCompactionPicker : public CompactionPicker {
@@ -1,7 +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 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/compaction_picker.h"
 #include <limits>
@@ -400,6 +400,35 @@ TEST_F(CompactionPickerTest, NeedsCompactionUniversal) {
              vstorage_->CompactionScore(0) >= 1);
  }
 }
+
+TEST_F(CompactionPickerTest, CompactionUniversalIngestBehindReservedLevel) {
+  const uint64_t kFileSize = 100000;
+  NewVersionStorage(1, kCompactionStyleUniversal);
+  ioptions_.allow_ingest_behind = true;
+  ioptions_.num_levels = 3;
+  UniversalCompactionPicker universal_compaction_picker(ioptions_, &icmp_);
+  // must return false when there's no files.
+  ASSERT_EQ(universal_compaction_picker.NeedsCompaction(vstorage_.get()),
+            false);
+
+  NewVersionStorage(3, kCompactionStyleUniversal);
+
+  Add(0, 1U, "150", "200", kFileSize, 0, 500, 550);
+  Add(0, 2U, "201", "250", kFileSize, 0, 401, 450);
+  Add(0, 4U, "260", "300", kFileSize, 0, 260, 300);
+  Add(1, 5U, "100", "151", kFileSize, 0, 200, 251);
+  Add(1, 3U, "301", "350", kFileSize, 0, 101, 150);
+  Add(2, 6U, "120", "200", kFileSize, 0, 20, 100);
+
+  UpdateVersionStorageInfo();
+
+  std::unique_ptr<Compaction> compaction(
+      universal_compaction_picker.PickCompaction(
+          cf_name_, mutable_cf_options_, vstorage_.get(), &log_buffer_));
+
+  // output level should be the one above the bottom-most
+  ASSERT_EQ(1, compaction->output_level());
+}
 // Tests if the files can be trivially moved in multi level
 // universal compaction when allow_trivial_move option is set
 // In this test as the input files overlaps, they cannot
@@ -921,7 +950,7 @@ TEST_F(CompactionPickerTest, NotScheduleL1IfL0WithHigherPri3) {
 TEST_F(CompactionPickerTest, EstimateCompactionBytesNeeded1) {
  int num_levels = ioptions_.num_levels;
  ioptions_.level_compaction_dynamic_level_bytes = false;
-  mutable_cf_options_.level0_file_num_compaction_trigger = 3;
+  mutable_cf_options_.level0_file_num_compaction_trigger = 4;
  mutable_cf_options_.max_bytes_for_level_base = 1000;
  mutable_cf_options_.max_bytes_for_level_multiplier = 10;
  NewVersionStorage(num_levels, kCompactionStyleLevel);
@@ -1011,7 +1040,8 @@ TEST_F(CompactionPickerTest, EstimateCompactionBytesNeededDynamicLevel) {

  // Set Last level size 50000
  // num_levels - 1 target 5000
-  // num_levels - 2 is base level with taret 500
+  // num_levels - 2 is base level with target 1000 (rounded up to
+  // max_bytes_for_level_base).
  Add(num_levels - 1, 10U, "400", "500", 50000);

  Add(0, 1U, "150", "200", 200);
@@ -1020,16 +1050,16 @@ TEST_F(CompactionPickerTest, EstimateCompactionBytesNeededDynamicLevel) {
  Add(0, 5U, "150", "200", 200);
  Add(0, 6U, "150", "200", 200);
  // num_levels - 3 is over target by 100 + 1000
-  Add(num_levels - 3, 7U, "400", "500", 300);
-  Add(num_levels - 3, 8U, "600", "700", 300);
+  Add(num_levels - 3, 7U, "400", "500", 550);
+  Add(num_levels - 3, 8U, "600", "700", 550);
  // num_levels - 2 is over target by 1100 + 200
  Add(num_levels - 2, 9U, "150", "200", 5200);

  UpdateVersionStorageInfo();

-  // Merging to the second last level: (5200 / 1600 + 1) * 1100
+  // Merging to the second last level: (5200 / 2100 + 1) * 1100
  // Merging to the last level: (50000 / 6300 + 1) * 1300
-  ASSERT_EQ(1600u + 4675u + 11617u,
+  ASSERT_EQ(2100u + 3823u + 11617u,
            vstorage_->estimated_compaction_needed_bytes());
 }

@@ -1,7 +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 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
@@ -342,8 +342,6 @@ Compaction* UniversalCompactionPicker::PickCompaction(
      assert(f->smallest_seqno <= f->largest_seqno);
      if (is_first) {
        is_first = false;
-      } else {
-        assert(prev_smallest_seqno > f->largest_seqno);
      }
      prev_smallest_seqno = f->smallest_seqno;
    }
@@ -568,6 +566,13 @@ Compaction* UniversalCompactionPicker::PickCompactionToReduceSortedRuns(
    output_level = sorted_runs[first_index_after].level - 1;
  }

+  // last level is reserved for the files ingested behind
+  if (ioptions_.allow_ingest_behind &&
+      (output_level == vstorage->num_levels() - 1)) {
+    assert(output_level > 1);
+    output_level--;
+  }
+
  std::vector<CompactionInputFiles> inputs(vstorage->num_levels());
  for (size_t i = 0; i < inputs.size(); ++i) {
    inputs[i].level = start_level + static_cast<int>(i);
@@ -614,7 +619,7 @@ Compaction* UniversalCompactionPicker::PickCompactionToReduceSizeAmp(
    const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
    VersionStorageInfo* vstorage, double score,
    const std::vector<SortedRun>& sorted_runs, LogBuffer* log_buffer) {
-  // percentage flexibilty while reducing size amplification
+  // percentage flexibility while reducing size amplification
  uint64_t ratio =
      ioptions_.compaction_options_universal.max_size_amplification_percent;

@@ -719,13 +724,20 @@ Compaction* UniversalCompactionPicker::PickCompactionToReduceSizeAmp(
                     cf_name.c_str(), file_num_buf);
  }

+  // output files at the bottom most level, unless it's reserved
+  int output_level = vstorage->num_levels() - 1;
+  // last level is reserved for the files ingested behind
+  if (ioptions_.allow_ingest_behind) {
+    assert(output_level > 1);
+    output_level--;
+  }
+
  return new Compaction(
      vstorage, ioptions_, mutable_cf_options, std::move(inputs),
-      vstorage->num_levels() - 1,
-      mutable_cf_options.MaxFileSizeForLevel(vstorage->num_levels() - 1),
+      output_level, mutable_cf_options.MaxFileSizeForLevel(output_level),
      /* max_grandparent_overlap_bytes */ LLONG_MAX, path_id,
      GetCompressionType(ioptions_, vstorage, mutable_cf_options,
-                         vstorage->num_levels() - 1, 1),
+                         output_level, 1),
      /* grandparents */ {}, /* is manual */ false, score,
      false /* deletion_compaction */,
      CompactionReason::kUniversalSizeAmplification);
@@ -1,7 +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 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
@@ -1,9 +1,7 @@
-// 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.
 //  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 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 <map>
 #include <string>

@@ -1,11 +1,8 @@
 //  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 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) 2012 Facebook.
-// Use of this source code is governed by a BSD-style license that can be
-// found in the LICENSE file.

 #ifndef ROCKSDB_LITE

@@ -16,12 +13,12 @@
 namespace rocksdb {

 void CancelAllBackgroundWork(DB* db, bool wait) {
-  (dynamic_cast<DBImpl*>(db))->CancelAllBackgroundWork(wait);
+  (dynamic_cast<DBImpl*>(db->GetRootDB()))->CancelAllBackgroundWork(wait);
 }

 Status DeleteFilesInRange(DB* db, ColumnFamilyHandle* column_family,
                          const Slice* begin, const Slice* end) {
-  return (dynamic_cast<DBImpl*>(db))
+  return (dynamic_cast<DBImpl*>(db->GetRootDB()))
      ->DeleteFilesInRange(column_family, begin, end);
 }

@@ -1,7 +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 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
@@ -1,7 +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 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).

 #ifndef ROCKSDB_LITE

@@ -320,8 +320,14 @@ TEST_F(CuckooTableDBTest, AdaptiveTable) {
 }  // namespace rocksdb

 int main(int argc, char** argv) {
-  ::testing::InitGoogleTest(&argc, argv);
-  return RUN_ALL_TESTS();
+  if (rocksdb::port::kLittleEndian) {
+    ::testing::InitGoogleTest(&argc, argv);
+    return RUN_ALL_TESTS();
+  }
+  else {
+    fprintf(stderr, "SKIPPED as Cuckoo table doesn't support Big Endian\n");
+    return 0;
+  }
 }

 #else
@@ -1,7 +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 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
@@ -364,9 +364,9 @@ TEST_F(DBBasicTest, FLUSH) {
    ASSERT_OK(Flush(1));
    ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "bar", "v1"));

-    perf_context.Reset();
+    get_perf_context()->Reset();
    Get(1, "foo");
-    ASSERT_TRUE((int)perf_context.get_from_output_files_time > 0);
+    ASSERT_TRUE((int)get_perf_context()->get_from_output_files_time > 0);

    ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions());
    ASSERT_EQ("v1", Get(1, "foo"));
@@ -379,9 +379,9 @@ TEST_F(DBBasicTest, FLUSH) {

    ReopenWithColumnFamilies({"default", "pikachu"}, CurrentOptions());
    ASSERT_EQ("v2", Get(1, "bar"));
-    perf_context.Reset();
+    get_perf_context()->Reset();
    ASSERT_EQ("v2", Get(1, "foo"));
-    ASSERT_TRUE((int)perf_context.get_from_output_files_time > 0);
+    ASSERT_TRUE((int)get_perf_context()->get_from_output_files_time > 0);

    writeOpt.disableWAL = false;
    ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "bar", "v3"));
@@ -822,6 +822,9 @@ TEST_F(DBBasicTest, ChecksumTest) {
 // sense to run
 #ifndef OS_WIN
 TEST_F(DBBasicTest, MmapAndBufferOptions) {
+  if (!IsMemoryMappedAccessSupported()) {
+    return;
+  }
  Options options = CurrentOptions();

  options.use_direct_reads = true;
@@ -1,7 +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 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
@@ -1,7 +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 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
@@ -686,12 +686,12 @@ class BloomStatsTestWithParam
    }
    options_.env = env_;

-    perf_context.Reset();
+    get_perf_context()->Reset();
    DestroyAndReopen(options_);
  }

  ~BloomStatsTestWithParam() {
-    perf_context.Reset();
+    get_perf_context()->Reset();
    Destroy(options_);
  }

@@ -724,33 +724,33 @@ TEST_P(BloomStatsTestWithParam, BloomStatsTest) {

  // check memtable bloom stats
  ASSERT_EQ(value1, Get(key1));
-  ASSERT_EQ(1, perf_context.bloom_memtable_hit_count);
+  ASSERT_EQ(1, get_perf_context()->bloom_memtable_hit_count);
  ASSERT_EQ(value3, Get(key3));
-  ASSERT_EQ(2, perf_context.bloom_memtable_hit_count);
-  ASSERT_EQ(0, perf_context.bloom_memtable_miss_count);
+  ASSERT_EQ(2, get_perf_context()->bloom_memtable_hit_count);
+  ASSERT_EQ(0, get_perf_context()->bloom_memtable_miss_count);

  ASSERT_EQ("NOT_FOUND", Get(key2));
-  ASSERT_EQ(1, perf_context.bloom_memtable_miss_count);
-  ASSERT_EQ(2, perf_context.bloom_memtable_hit_count);
+  ASSERT_EQ(1, get_perf_context()->bloom_memtable_miss_count);
+  ASSERT_EQ(2, get_perf_context()->bloom_memtable_hit_count);

  // sanity checks
-  ASSERT_EQ(0, perf_context.bloom_sst_hit_count);
-  ASSERT_EQ(0, perf_context.bloom_sst_miss_count);
+  ASSERT_EQ(0, get_perf_context()->bloom_sst_hit_count);
+  ASSERT_EQ(0, get_perf_context()->bloom_sst_miss_count);

  Flush();

  // sanity checks
-  ASSERT_EQ(0, perf_context.bloom_sst_hit_count);
-  ASSERT_EQ(0, perf_context.bloom_sst_miss_count);
+  ASSERT_EQ(0, get_perf_context()->bloom_sst_hit_count);
+  ASSERT_EQ(0, get_perf_context()->bloom_sst_miss_count);

  // check SST bloom stats
  ASSERT_EQ(value1, Get(key1));
-  ASSERT_EQ(1, perf_context.bloom_sst_hit_count);
+  ASSERT_EQ(1, get_perf_context()->bloom_sst_hit_count);
  ASSERT_EQ(value3, Get(key3));
-  ASSERT_EQ(2, perf_context.bloom_sst_hit_count);
+  ASSERT_EQ(2, get_perf_context()->bloom_sst_hit_count);

  ASSERT_EQ("NOT_FOUND", Get(key2));
-  ASSERT_EQ(1, perf_context.bloom_sst_miss_count);
+  ASSERT_EQ(1, get_perf_context()->bloom_sst_miss_count);
 }

 // Same scenario as in BloomStatsTest but using an iterator
@@ -771,21 +771,21 @@ TEST_P(BloomStatsTestWithParam, BloomStatsTestWithIter) {
  ASSERT_OK(iter->status());
  ASSERT_TRUE(iter->Valid());
  ASSERT_EQ(value1, iter->value().ToString());
-  ASSERT_EQ(1, perf_context.bloom_memtable_hit_count);
-  ASSERT_EQ(0, perf_context.bloom_memtable_miss_count);
+  ASSERT_EQ(1, get_perf_context()->bloom_memtable_hit_count);
+  ASSERT_EQ(0, get_perf_context()->bloom_memtable_miss_count);

  iter->Seek(key3);
  ASSERT_OK(iter->status());
  ASSERT_TRUE(iter->Valid());
  ASSERT_EQ(value3, iter->value().ToString());
-  ASSERT_EQ(2, perf_context.bloom_memtable_hit_count);
-  ASSERT_EQ(0, perf_context.bloom_memtable_miss_count);
+  ASSERT_EQ(2, get_perf_context()->bloom_memtable_hit_count);
+  ASSERT_EQ(0, get_perf_context()->bloom_memtable_miss_count);

  iter->Seek(key2);
  ASSERT_OK(iter->status());
  ASSERT_TRUE(!iter->Valid());
-  ASSERT_EQ(1, perf_context.bloom_memtable_miss_count);
-  ASSERT_EQ(2, perf_context.bloom_memtable_hit_count);
+  ASSERT_EQ(1, get_perf_context()->bloom_memtable_miss_count);
+  ASSERT_EQ(2, get_perf_context()->bloom_memtable_hit_count);

  Flush();

@@ -796,36 +796,19 @@ TEST_P(BloomStatsTestWithParam, BloomStatsTestWithIter) {
  ASSERT_OK(iter->status());
  ASSERT_TRUE(iter->Valid());
  ASSERT_EQ(value1, iter->value().ToString());
-  if (partition_filters_) {
-    ASSERT_EQ(0, perf_context.bloom_sst_hit_count);   // no_io
-    ASSERT_EQ(0, perf_context.bloom_sst_miss_count);  // no_io
-  } else {
-    ASSERT_EQ(1, perf_context.bloom_sst_hit_count);
-  }
+  ASSERT_EQ(1, get_perf_context()->bloom_sst_hit_count);

  iter->Seek(key3);
  ASSERT_OK(iter->status());
  ASSERT_TRUE(iter->Valid());
  ASSERT_EQ(value3, iter->value().ToString());
-  if (partition_filters_) {
-    ASSERT_EQ(0, perf_context.bloom_sst_hit_count);   // no_io
-    ASSERT_EQ(0, perf_context.bloom_sst_miss_count);  // no_io
-  } else {
-    ASSERT_EQ(2, perf_context.bloom_sst_hit_count);
-  }
+  ASSERT_EQ(2, get_perf_context()->bloom_sst_hit_count);

  iter->Seek(key2);
  ASSERT_OK(iter->status());
-  if (partition_filters_) {
-    // iter is still valid since filter did not reject the key2
-    ASSERT_TRUE(iter->Valid());
-    ASSERT_EQ(0, perf_context.bloom_sst_hit_count);   // no_io
-    ASSERT_EQ(0, perf_context.bloom_sst_miss_count);  // no_io
-  } else {
-    ASSERT_TRUE(!iter->Valid());
-    ASSERT_EQ(1, perf_context.bloom_sst_miss_count);
-    ASSERT_EQ(2, perf_context.bloom_sst_hit_count);
-  }
+  ASSERT_TRUE(!iter->Valid());
+  ASSERT_EQ(1, get_perf_context()->bloom_sst_miss_count);
+  ASSERT_EQ(2, get_perf_context()->bloom_sst_hit_count);
 }

 INSTANTIATE_TEST_CASE_P(BloomStatsTestWithParam, BloomStatsTestWithParam,
@@ -1,7 +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 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
@@ -626,7 +626,7 @@ TEST_F(DBTestCompactionFilter, CompactionFilterContextManual) {
      iter->Next();
    }
    ASSERT_EQ(total, 700);
-    ASSERT_EQ(count, 2);
+    ASSERT_EQ(count, 1);
  }
 }
 #endif  // ROCKSDB_LITE
@@ -778,7 +778,7 @@ TEST_F(DBTestCompactionFilter, SkipUntil) {

  cfilter_skips = 0;
  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
-  // Numberof skips in tables: 2, 3, 3, 3.
+  // Number of skips in tables: 2, 3, 3, 3.
  ASSERT_EQ(11, cfilter_skips);

  for (int table = 0; table < 4; ++table) {
@@ -799,6 +799,43 @@ TEST_F(DBTestCompactionFilter, SkipUntil) {
  }
 }

+TEST_F(DBTestCompactionFilter, SkipUntilWithBloomFilter) {
+  BlockBasedTableOptions table_options;
+  table_options.whole_key_filtering = false;
+  table_options.filter_policy.reset(NewBloomFilterPolicy(100, false));
+
+  Options options = CurrentOptions();
+  options.table_factory.reset(NewBlockBasedTableFactory(table_options));
+  options.prefix_extractor.reset(NewCappedPrefixTransform(9));
+  options.compaction_filter_factory = std::make_shared<SkipEvenFilterFactory>();
+  options.disable_auto_compactions = true;
+  options.create_if_missing = true;
+  DestroyAndReopen(options);
+
+  Put("0000000010", "v10");
+  Put("0000000020", "v20"); // skipped
+  Put("0000000050", "v50");
+  Flush();
+
+  cfilter_skips = 0;
+  EXPECT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  EXPECT_EQ(1, cfilter_skips);
+
+  Status s;
+  std::string val;
+
+  s = db_->Get(ReadOptions(), "0000000010", &val);
+  ASSERT_OK(s);
+  EXPECT_EQ("v10", val);
+
+  s = db_->Get(ReadOptions(), "0000000020", &val);
+  EXPECT_TRUE(s.IsNotFound());
+
+  s = db_->Get(ReadOptions(), "0000000050", &val);
+  ASSERT_OK(s);
+  EXPECT_EQ("v50", val);
+}
+
 }  // namespace rocksdb

 int main(int argc, char** argv) {
@@ -1,7 +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 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
@@ -40,6 +40,12 @@ class DBCompactionTestWithParam
  bool exclusive_manual_compaction_;
 };

+class DBCompactionDirectIOTest : public DBCompactionTest,
+                                 public ::testing::WithParamInterface<bool> {
+ public:
+  DBCompactionDirectIOTest() : DBCompactionTest() {}
+};
+
 namespace {

 class FlushedFileCollector : public EventListener {
@@ -513,7 +519,6 @@ TEST_F(DBCompactionTest, BGCompactionsAllowed) {
  options.level0_file_num_compaction_trigger = 2;
  options.level0_slowdown_writes_trigger = 20;
  options.soft_pending_compaction_bytes_limit = 1 << 30;  // Infinitely large
-  options.base_background_compactions = 1;
  options.max_background_compactions = 3;
  options.memtable_factory.reset(new SpecialSkipListFactory(kNumKeysPerFile));

@@ -772,6 +777,41 @@ TEST_F(DBCompactionTest, ZeroSeqIdCompaction) {
  ASSERT_OK(Put("", ""));
 }

+TEST_F(DBCompactionTest, ManualCompactionUnknownOutputSize) {
+  // github issue #2249
+  Options options = CurrentOptions();
+  options.compaction_style = kCompactionStyleLevel;
+  options.level0_file_num_compaction_trigger = 3;
+  DestroyAndReopen(options);
+
+  // create two files in l1 that we can compact
+  for (int i = 0; i < 2; ++i) {
+    for (int j = 0; j < options.level0_file_num_compaction_trigger; j++) {
+      // make l0 files' ranges overlap to avoid trivial move
+      Put(std::to_string(2 * i), std::string(1, 'A'));
+      Put(std::to_string(2 * i + 1), std::string(1, 'A'));
+      Flush();
+      dbfull()->TEST_WaitForFlushMemTable();
+    }
+    dbfull()->TEST_WaitForCompact();
+    ASSERT_EQ(NumTableFilesAtLevel(0, 0), 0);
+    ASSERT_EQ(NumTableFilesAtLevel(1, 0), i + 1);
+  }
+
+  ColumnFamilyMetaData cf_meta;
+  dbfull()->GetColumnFamilyMetaData(dbfull()->DefaultColumnFamily(), &cf_meta);
+  ASSERT_EQ(2, cf_meta.levels[1].files.size());
+  std::vector<std::string> input_filenames;
+  for (const auto& sst_file : cf_meta.levels[1].files) {
+    input_filenames.push_back(sst_file.name);
+  }
+
+  // note CompactionOptions::output_file_size_limit is unset.
+  CompactionOptions compact_opt;
+  compact_opt.compression = kNoCompression;
+  dbfull()->CompactFiles(compact_opt, input_filenames, 1);
+}
+
 // Check that writes done during a memtable compaction are recovered
 // if the database is shutdown during the memtable compaction.
 TEST_F(DBCompactionTest, RecoverDuringMemtableCompaction) {
@@ -1152,7 +1192,8 @@ TEST_P(DBCompactionTestWithParam, ManualCompactionPartial) {
  }
 }

-TEST_F(DBCompactionTest, ManualPartialFill) {
+// Disable as the test is flaky.
+TEST_F(DBCompactionTest, DISABLED_ManualPartialFill) {
  int32_t trivial_move = 0;
  int32_t non_trivial_move = 0;
  rocksdb::SyncPoint::GetInstance()->SetCallBack(
@@ -1186,6 +1227,9 @@ TEST_F(DBCompactionTest, ManualPartialFill) {
  options.max_background_compactions = 3;

  DestroyAndReopen(options);
+  // make sure all background compaction jobs can be scheduled
+  auto stop_token =
+      dbfull()->TEST_write_controler().GetCompactionPressureToken();
  int32_t value_size = 10 * 1024;  // 10 KB

  // Add 2 non-overlapping files
@@ -1843,6 +1887,47 @@ TEST_F(DBCompactionTest, L0_CompactionBug_Issue44_b) {
  } while (ChangeCompactOptions());
 }

+TEST_F(DBCompactionTest, ManualAutoRace) {
+  CreateAndReopenWithCF({"pikachu"}, CurrentOptions());
+  rocksdb::SyncPoint::GetInstance()->LoadDependency(
+      {{"DBImpl::BGWorkCompaction", "DBCompactionTest::ManualAutoRace:1"},
+       {"DBImpl::RunManualCompaction:WaitScheduled",
+        "BackgroundCallCompaction:0"}});
+
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+  Put(1, "foo", "");
+  Put(1, "bar", "");
+  Flush(1);
+  Put(1, "foo", "");
+  Put(1, "bar", "");
+  // Generate four files in CF 0, which should trigger an auto compaction
+  Put("foo", "");
+  Put("bar", "");
+  Flush();
+  Put("foo", "");
+  Put("bar", "");
+  Flush();
+  Put("foo", "");
+  Put("bar", "");
+  Flush();
+  Put("foo", "");
+  Put("bar", "");
+  Flush();
+
+  // The auto compaction is scheduled but waited until here
+  TEST_SYNC_POINT("DBCompactionTest::ManualAutoRace:1");
+  // The auto compaction will wait until the manual compaction is registerd
+  // before processing so that it will be cancelled.
+  dbfull()->CompactRange(CompactRangeOptions(), handles_[1], nullptr, nullptr);
+  ASSERT_EQ("0,1", FilesPerLevel(1));
+
+  // Eventually the cancelled compaction will be rescheduled and executed.
+  dbfull()->TEST_WaitForCompact();
+  ASSERT_EQ("0,1", FilesPerLevel(0));
+  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+}
+
 TEST_P(DBCompactionTestWithParam, ManualCompaction) {
  Options options = CurrentOptions();
  options.max_subcompactions = max_subcompactions_;
@@ -1892,7 +1977,6 @@ TEST_P(DBCompactionTestWithParam, ManualCompaction) {

    if (iter == 0) {
      options = CurrentOptions();
-      options.max_background_flushes = 0;
      options.num_levels = 3;
      options.create_if_missing = true;
      options.statistics = rocksdb::CreateDBStatistics();
@@ -2338,15 +2422,12 @@ TEST_F(DBCompactionTest, SanitizeCompactionOptionsTest) {
  options.hard_pending_compaction_bytes_limit = 100;
  options.create_if_missing = true;
  DestroyAndReopen(options);
-  ASSERT_EQ(5, db_->GetOptions().base_background_compactions);
  ASSERT_EQ(100, db_->GetOptions().soft_pending_compaction_bytes_limit);

-  options.base_background_compactions = 4;
  options.max_background_compactions = 3;
  options.soft_pending_compaction_bytes_limit = 200;
  options.hard_pending_compaction_bytes_limit = 150;
  DestroyAndReopen(options);
-  ASSERT_EQ(3, db_->GetOptions().base_background_compactions);
  ASSERT_EQ(150, db_->GetOptions().soft_pending_compaction_bytes_limit);
 }

@@ -2521,13 +2602,11 @@ TEST_P(DBCompactionTestWithParam, IntraL0Compaction) {
  // Files 6-9 are the longest span of available files for which
  // work-per-deleted-file decreases (see "score" row above).
  for (int i = 0; i < 10; ++i) {
-    for (int j = 0; j < 2; ++j) {
-      ASSERT_OK(Put(Key(0), ""));  // prevents trivial move
-      if (i == 5) {
-        ASSERT_OK(Put(Key(i + 1), value + value));
-      } else {
-        ASSERT_OK(Put(Key(i + 1), value));
-      }
+    ASSERT_OK(Put(Key(0), ""));  // prevents trivial move
+    if (i == 5) {
+      ASSERT_OK(Put(Key(i + 1), value + value));
+    } else {
+      ASSERT_OK(Put(Key(i + 1), value));
    }
    ASSERT_OK(Flush());
  }
@@ -2542,16 +2621,116 @@ TEST_P(DBCompactionTestWithParam, IntraL0Compaction) {
  ASSERT_EQ(2, level_to_files[0].size());
  ASSERT_GT(level_to_files[1].size(), 0);
  for (int i = 0; i < 2; ++i) {
-    ASSERT_GE(level_to_files[0][0].fd.file_size, 1 << 21);
+    ASSERT_GE(level_to_files[0][i].fd.file_size, 1 << 21);
  }
 }

+TEST_P(DBCompactionTestWithParam, IntraL0CompactionDoesNotObsoleteDeletions) {
+  // regression test for issue #2722: L0->L0 compaction can resurrect deleted
+  // keys from older L0 files if L1+ files' key-ranges do not include the key.
+  Options options = CurrentOptions();
+  options.compression = kNoCompression;
+  options.level0_file_num_compaction_trigger = 5;
+  options.max_background_compactions = 2;
+  options.max_subcompactions = max_subcompactions_;
+  DestroyAndReopen(options);
+
+  const size_t kValueSize = 1 << 20;
+  Random rnd(301);
+  std::string value(RandomString(&rnd, kValueSize));
+
+  rocksdb::SyncPoint::GetInstance()->LoadDependency(
+      {{"LevelCompactionPicker::PickCompactionBySize:0",
+        "CompactionJob::Run():Start"}});
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+  // index:   0   1   2   3   4    5    6   7   8   9
+  // size:  1MB 1MB 1MB 1MB 1MB  1MB  1MB 1MB 1MB 1MB
+  // score:                     1.25 1.33 1.5 2.0 inf
+  //
+  // Files 0-4 will be included in an L0->L1 compaction.
+  //
+  // L0->L0 will be triggered since the sync points guarantee compaction to base
+  // level is still blocked when files 5-9 trigger another compaction. All files
+  // 5-9 are included in the L0->L0 due to work-per-deleted file decreasing.
+  //
+  // Put a key-value in files 0-4. Delete that key in files 5-9. Verify the
+  // L0->L0 preserves the deletion such that the key remains deleted.
+  for (int i = 0; i < 10; ++i) {
+    // key 0 serves both to prevent trivial move and as the key we want to
+    // verify is not resurrected by L0->L0 compaction.
+    if (i < 5) {
+      ASSERT_OK(Put(Key(0), ""));
+    } else {
+      ASSERT_OK(Delete(Key(0)));
+    }
+    ASSERT_OK(Put(Key(i + 1), value));
+    ASSERT_OK(Flush());
+  }
+  dbfull()->TEST_WaitForCompact();
+  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+
+  std::vector<std::vector<FileMetaData>> level_to_files;
+  dbfull()->TEST_GetFilesMetaData(dbfull()->DefaultColumnFamily(),
+                                  &level_to_files);
+  ASSERT_GE(level_to_files.size(), 2);  // at least L0 and L1
+  // L0 has a single output file from L0->L0
+  ASSERT_EQ(1, level_to_files[0].size());
+  ASSERT_GT(level_to_files[1].size(), 0);
+  ASSERT_GE(level_to_files[0][0].fd.file_size, 1 << 22);
+
+  ReadOptions roptions;
+  std::string result;
+  ASSERT_TRUE(db_->Get(roptions, Key(0), &result).IsNotFound());
+}
+
 INSTANTIATE_TEST_CASE_P(DBCompactionTestWithParam, DBCompactionTestWithParam,
                        ::testing::Values(std::make_tuple(1, true),
                                          std::make_tuple(1, false),
                                          std::make_tuple(4, true),
                                          std::make_tuple(4, false)));

+TEST_P(DBCompactionDirectIOTest, DirectIO) {
+  Options options = CurrentOptions();
+  Destroy(options);
+  options.create_if_missing = true;
+  options.disable_auto_compactions = true;
+  options.use_direct_io_for_flush_and_compaction = GetParam();
+  options.env = new MockEnv(Env::Default());
+  Reopen(options);
+  bool readahead = false;
+  SyncPoint::GetInstance()->SetCallBack(
+      "TableCache::NewIterator:for_compaction", [&](void* arg) {
+        bool* use_direct_reads = static_cast<bool*>(arg);
+        ASSERT_EQ(*use_direct_reads,
+                  options.use_direct_io_for_flush_and_compaction);
+      });
+  SyncPoint::GetInstance()->SetCallBack(
+      "CompactionJob::OpenCompactionOutputFile", [&](void* arg) {
+        bool* use_direct_writes = static_cast<bool*>(arg);
+        ASSERT_EQ(*use_direct_writes,
+                  options.use_direct_io_for_flush_and_compaction);
+      });
+  if (options.use_direct_io_for_flush_and_compaction) {
+    SyncPoint::GetInstance()->SetCallBack(
+        "SanitizeOptions:direct_io", [&](void* arg) {
+          readahead = true;
+        });
+  }
+  SyncPoint::GetInstance()->EnableProcessing();
+  CreateAndReopenWithCF({"pikachu"}, options);
+  MakeTables(3, "p", "q", 1);
+  ASSERT_EQ("1,1,1", FilesPerLevel(1));
+  Compact(1, "p1", "p9");
+  ASSERT_FALSE(readahead ^ options.use_direct_io_for_flush_and_compaction);
+  ASSERT_EQ("0,0,1", FilesPerLevel(1));
+  Destroy(options);
+  delete options.env;
+}
+
+INSTANTIATE_TEST_CASE_P(DBCompactionDirectIOTest, DBCompactionDirectIOTest,
+                        testing::Bool());
+
 class CompactionPriTest : public DBTestBase,
                          public testing::WithParamInterface<uint32_t> {
 public:
@@ -1,7 +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 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
@@ -57,7 +57,6 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBase) {
      Options options;
      options.env = env.get();
      options.create_if_missing = true;
-      options.db_write_buffer_size = 2048;
      options.write_buffer_size = 2048;
      options.max_write_buffer_number = 2;
      options.level0_file_num_compaction_trigger = 2;
@@ -127,7 +126,6 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBase2) {

  Options options = CurrentOptions();
  options.create_if_missing = true;
-  options.db_write_buffer_size = 204800;
  options.write_buffer_size = 20480;
  options.max_write_buffer_number = 2;
  options.level0_file_num_compaction_trigger = 2;
@@ -284,7 +282,6 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesCompactRange) {

  Options options = CurrentOptions();
  options.create_if_missing = true;
-  options.db_write_buffer_size = 2048;
  options.write_buffer_size = 2048;
  options.max_write_buffer_number = 2;
  options.level0_file_num_compaction_trigger = 2;
@@ -362,7 +359,6 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesCompactRange) {
 TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBaseInc) {
  Options options = CurrentOptions();
  options.create_if_missing = true;
-  options.db_write_buffer_size = 2048;
  options.write_buffer_size = 2048;
  options.max_write_buffer_number = 2;
  options.level0_file_num_compaction_trigger = 2;
@@ -375,6 +371,7 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBaseInc) {
  options.soft_rate_limit = 1.1;
  options.max_background_compactions = 2;
  options.num_levels = 5;
+  options.max_compaction_bytes = 100000000;

  DestroyAndReopen(options);

@@ -414,7 +411,6 @@ TEST_F(DBTestDynamicLevel, DISABLED_MigrateToDynamicLevelMaxBytesBase) {

  Options options;
  options.create_if_missing = true;
-  options.db_write_buffer_size = 2048;
  options.write_buffer_size = 2048;
  options.max_write_buffer_number = 8;
  options.level0_file_num_compaction_trigger = 4;
@@ -0,0 +1,96 @@
+//  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.
+//
+#include "db/db_test_util.h"
+#include "port/stack_trace.h"
+#include "rocksdb/perf_context.h"
+#if !defined(ROCKSDB_LITE)
+#include "util/sync_point.h"
+#endif
+#include <iostream>
+#include <string>
+
+namespace rocksdb {
+
+class DBEncryptionTest : public DBTestBase {
+ public:
+  DBEncryptionTest() : DBTestBase("/db_encryption_test") {}
+};
+
+#ifndef ROCKSDB_LITE
+
+TEST_F(DBEncryptionTest, CheckEncrypted) {
+  ASSERT_OK(Put("foo567", "v1.fetdq"));
+  ASSERT_OK(Put("bar123", "v2.dfgkjdfghsd"));
+  Close();
+
+  // Open all files and look for the values we've put in there.
+  // They should not be found if encrypted, otherwise
+  // they should be found.
+  std::vector<std::string> fileNames;
+  auto status = env_->GetChildren(dbname_, &fileNames);
+  ASSERT_OK(status);
+
+  auto defaultEnv = Env::Default();
+  int hits = 0;
+  for (auto it = fileNames.begin() ; it != fileNames.end(); ++it) {
+    if ((*it == "..") || (*it == ".")) {
+      continue;
+    }
+    auto filePath = dbname_ + "/" + *it;
+    unique_ptr<SequentialFile> seqFile;
+    auto envOptions = EnvOptions(CurrentOptions());
+    status = defaultEnv->NewSequentialFile(filePath, &seqFile, envOptions);
+    ASSERT_OK(status);
+
+    uint64_t fileSize;
+    status = defaultEnv->GetFileSize(filePath, &fileSize);
+    ASSERT_OK(status);
+
+    std::string scratch;
+    scratch.reserve(fileSize);
+    Slice data;
+    status = seqFile->Read(fileSize, &data, (char*)scratch.data());
+    ASSERT_OK(status);
+
+    if (data.ToString().find("foo567") != std::string::npos) {
+      hits++; 
+      //std::cout << "Hit in " << filePath << "\n";
+    }
+    if (data.ToString().find("v1.fetdq") != std::string::npos) {
+      hits++; 
+      //std::cout << "Hit in " << filePath << "\n";
+    }
+    if (data.ToString().find("bar123") != std::string::npos) {
+      hits++; 
+      //std::cout << "Hit in " << filePath << "\n";
+    }
+    if (data.ToString().find("v2.dfgkjdfghsd") != std::string::npos) {
+      hits++; 
+      //std::cout << "Hit in " << filePath << "\n";
+    }
+    if (data.ToString().find("dfgk") != std::string::npos) {
+      hits++; 
+      //std::cout << "Hit in " << filePath << "\n";
+    }
+  }
+  if (encrypted_env_) {
+    ASSERT_EQ(hits, 0);
+  } else {
+    ASSERT_GE(hits, 4);
+  }
+}
+
+#endif // ROCKSDB_LITE
+
+}  // namespace rocksdb
+
+int main(int argc, char** argv) {
+  rocksdb::port::InstallStackTraceHandler();
+  ::testing::InitGoogleTest(&argc, argv);
+  return RUN_ALL_TESTS();
+}
@@ -1,11 +1,8 @@
 //  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 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) 2012 Facebook.
-// Use of this source code is governed by a BSD-style license that can be
-// found in the LICENSE file.

 #ifndef ROCKSDB_LITE

@@ -1,7 +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 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
@@ -19,6 +19,12 @@ class DBFlushTest : public DBTestBase {
  DBFlushTest() : DBTestBase("/db_flush_test") {}
 };

+class DBFlushDirectIOTest : public DBFlushTest,
+                            public ::testing::WithParamInterface<bool> {
+ public:
+  DBFlushDirectIOTest() : DBFlushTest() {}
+};
+
 // We had issue when two background threads trying to flush at the same time,
 // only one of them get committed. The test verifies the issue is fixed.
 TEST_F(DBFlushTest, FlushWhileWritingManifest) {
@@ -83,6 +89,70 @@ TEST_F(DBFlushTest, SyncFail) {
  Destroy(options);
 }

+TEST_F(DBFlushTest, FlushInLowPriThreadPool) {
+  // Verify setting an empty high-pri (flush) thread pool causes flushes to be
+  // scheduled in the low-pri (compaction) thread pool.
+  Options options = CurrentOptions();
+  options.level0_file_num_compaction_trigger = 4;
+  options.memtable_factory.reset(new SpecialSkipListFactory(1));
+  Reopen(options);
+  env_->SetBackgroundThreads(0, Env::HIGH);
+
+  std::thread::id tid;
+  int num_flushes = 0, num_compactions = 0;
+  SyncPoint::GetInstance()->SetCallBack(
+      "DBImpl::BGWorkFlush", [&](void* arg) {
+        if (tid == std::thread::id()) {
+          tid = std::this_thread::get_id();
+        } else {
+          ASSERT_EQ(tid, std::this_thread::get_id());
+        }
+        ++num_flushes;
+      });
+  SyncPoint::GetInstance()->SetCallBack(
+      "DBImpl::BGWorkCompaction", [&](void* arg) {
+        ASSERT_EQ(tid, std::this_thread::get_id());
+        ++num_compactions;
+      });
+  SyncPoint::GetInstance()->EnableProcessing();
+
+  ASSERT_OK(Put("key", "val"));
+  for (int i = 0; i < 4; ++i) {
+    ASSERT_OK(Put("key", "val"));
+    dbfull()->TEST_WaitForFlushMemTable();
+  }
+  dbfull()->TEST_WaitForCompact();
+  ASSERT_EQ(4, num_flushes);
+  ASSERT_EQ(1, num_compactions);
+}
+
+TEST_P(DBFlushDirectIOTest, DirectIO) {
+  Options options;
+  options.create_if_missing = true;
+  options.disable_auto_compactions = true;
+  options.max_background_flushes = 2;
+  options.use_direct_io_for_flush_and_compaction = GetParam();
+  options.env = new MockEnv(Env::Default());
+  SyncPoint::GetInstance()->SetCallBack(
+      "BuildTable:create_file", [&](void* arg) {
+        bool* use_direct_writes = static_cast<bool*>(arg);
+        ASSERT_EQ(*use_direct_writes,
+                  options.use_direct_io_for_flush_and_compaction);
+      });
+
+  SyncPoint::GetInstance()->EnableProcessing();
+  Reopen(options);
+  ASSERT_OK(Put("foo", "v"));
+  FlushOptions flush_options;
+  flush_options.wait = true;
+  ASSERT_OK(dbfull()->Flush(flush_options));
+  Destroy(options);
+  delete options.env;
+}
+
+INSTANTIATE_TEST_CASE_P(DBFlushDirectIOTest, DBFlushDirectIOTest,
+                        testing::Bool());
+
 }  // namespace rocksdb

 int main(int argc, char** argv) {
@@ -1,7 +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 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
@@ -16,9 +16,6 @@
 #ifdef OS_SOLARIS
 #include <alloca.h>
 #endif
-#ifdef ROCKSDB_JEMALLOC
-#include "jemalloc/jemalloc.h"
-#endif

 #include <algorithm>
 #include <climits>
@@ -44,6 +41,7 @@
 #include "db/job_context.h"
 #include "db/log_reader.h"
 #include "db/log_writer.h"
+#include "db/malloc_stats.h"
 #include "db/managed_iterator.h"
 #include "db/memtable.h"
 #include "db/memtable_list.h"
@@ -134,6 +132,8 @@ void DumpSupportInfo(Logger* logger) {
  ROCKS_LOG_HEADER(logger, "Fast CRC32 supported: %d",
                   crc32c::IsFastCrc32Supported());
 }
+
+int64_t kDefaultLowPriThrottledRate = 2 * 1024 * 1024;
 } // namespace

 DBImpl::DBImpl(const DBOptions& options, const std::string& dbname)
@@ -157,11 +157,14 @@ DBImpl::DBImpl(const DBOptions& options, const std::string& dbname)
      max_total_in_memory_state_(0),
      is_snapshot_supported_(true),
      write_buffer_manager_(immutable_db_options_.write_buffer_manager.get()),
-      write_thread_(immutable_db_options_.enable_write_thread_adaptive_yield
-                        ? immutable_db_options_.write_thread_max_yield_usec
-                        : 0,
-                    immutable_db_options_.write_thread_slow_yield_usec),
+      write_thread_(immutable_db_options_),
+      nonmem_write_thread_(immutable_db_options_),
      write_controller_(mutable_db_options_.delayed_write_rate),
+      // Use delayed_write_rate as a base line to determine the initial
+      // low pri write rate limit. It may be adjusted later.
+      low_pri_write_rate_limiter_(NewGenericRateLimiter(std::min(
+          static_cast<int64_t>(mutable_db_options_.delayed_write_rate / 8),
+          kDefaultLowPriThrottledRate))),
      last_batch_group_size_(0),
      unscheduled_flushes_(0),
      unscheduled_compactions_(0),
@@ -185,14 +188,16 @@ DBImpl::DBImpl(const DBOptions& options, const std::string& dbname)
      bg_work_paused_(0),
      bg_compaction_paused_(0),
      refitting_level_(false),
-      opened_successfully_(false) {
+      opened_successfully_(false),
+      concurrent_prepare_(options.concurrent_prepare),
+      manual_wal_flush_(options.manual_wal_flush) {
  env_->GetAbsolutePath(dbname, &db_absolute_path_);

  // Reserve ten files or so for other uses and give the rest to TableCache.
  // Give a large number for setting of "infinite" open files.
-  const int table_cache_size = (immutable_db_options_.max_open_files == -1)
-                                   ? 4194304
-                                   : immutable_db_options_.max_open_files - 10;
+  const int table_cache_size = (mutable_db_options_.max_open_files == -1)
+                                   ? TableCache::kInfiniteCapacity
+                                   : mutable_db_options_.max_open_files - 10;
  table_cache_ = NewLRUCache(table_cache_size,
                             immutable_db_options_.table_cache_numshardbits);

@@ -220,7 +225,7 @@ void DBImpl::CancelAllBackgroundWork(bool wait) {
      has_unpersisted_data_.load(std::memory_order_relaxed) &&
      !mutable_db_options_.avoid_flush_during_shutdown) {
    for (auto cfd : *versions_->GetColumnFamilySet()) {
-      if (!cfd->IsDropped() && !cfd->mem()->IsEmpty()) {
+      if (!cfd->IsDropped() && cfd->initialized() && !cfd->mem()->IsEmpty()) {
        cfd->Ref();
        mutex_.Unlock();
        FlushMemTable(cfd, FlushOptions());
@@ -371,39 +376,6 @@ void DBImpl::PrintStatistics() {
  }
 }

-#ifndef ROCKSDB_LITE
-#ifdef ROCKSDB_JEMALLOC
-typedef struct {
-  char* cur;
-  char* end;
-} MallocStatus;
-
-static void GetJemallocStatus(void* mstat_arg, const char* status) {
-  MallocStatus* mstat = reinterpret_cast<MallocStatus*>(mstat_arg);
-  size_t status_len = status ? strlen(status) : 0;
-  size_t buf_size = (size_t)(mstat->end - mstat->cur);
-  if (!status_len || status_len > buf_size) {
-    return;
-  }
-
-  snprintf(mstat->cur, buf_size, "%s", status);
-  mstat->cur += status_len;
-}
-#endif  // ROCKSDB_JEMALLOC
-
-static void DumpMallocStats(std::string* stats) {
-#ifdef ROCKSDB_JEMALLOC
-  MallocStatus mstat;
-  const unsigned int kMallocStatusLen = 1000000;
-  std::unique_ptr<char[]> buf{new char[kMallocStatusLen + 1]};
-  mstat.cur = buf.get();
-  mstat.end = buf.get() + kMallocStatusLen;
-  je_malloc_stats_print(GetJemallocStatus, &mstat, "");
-  stats->append(buf.get());
-#endif  // ROCKSDB_JEMALLOC
-}
-#endif  // !ROCKSDB_LITE
-
 void DBImpl::MaybeDumpStats() {
  mutex_.Lock();
  unsigned int stats_dump_period_sec =
@@ -435,12 +407,17 @@ void DBImpl::MaybeDumpStats() {
      default_cf_internal_stats_->GetStringProperty(
          *db_property_info, DB::Properties::kDBStats, &stats);
      for (auto cfd : *versions_->GetColumnFamilySet()) {
-        cfd->internal_stats()->GetStringProperty(
-            *cf_property_info, DB::Properties::kCFStatsNoFileHistogram, &stats);
+        if (cfd->initialized()) {
+          cfd->internal_stats()->GetStringProperty(
+              *cf_property_info, DB::Properties::kCFStatsNoFileHistogram,
+              &stats);
+        }
      }
      for (auto cfd : *versions_->GetColumnFamilySet()) {
-        cfd->internal_stats()->GetStringProperty(
-            *cf_property_info, DB::Properties::kCFFileHistogram, &stats);
+        if (cfd->initialized()) {
+          cfd->internal_stats()->GetStringProperty(
+              *cf_property_info, DB::Properties::kCFFileHistogram, &stats);
+        }
      }
    }
    ROCKS_LOG_WARN(immutable_db_options_.info_log,
@@ -514,9 +491,8 @@ Status DBImpl::SetOptions(ColumnFamilyHandle* column_family,
          InstallSuperVersionAndScheduleWork(cfd, nullptr, new_options);
      delete old_sv;

-      write_thread_.EnterUnbatched(&w, &mutex_);
-      persist_options_status = WriteOptionsFile();
-      write_thread_.ExitUnbatched(&w);
+      persist_options_status = WriteOptionsFile(
+          false /*need_mutex_lock*/, true /*need_enter_write_thread*/);
    }
  }

@@ -532,14 +508,7 @@ Status DBImpl::SetOptions(ColumnFamilyHandle* column_family,
                   "[%s] SetOptions() succeeded", cfd->GetName().c_str());
    new_options.Dump(immutable_db_options_.info_log.get());
    if (!persist_options_status.ok()) {
-      if (immutable_db_options_.fail_if_options_file_error) {
-        s = Status::IOError(
-            "SetOptions() succeeded, but unable to persist options",
-            persist_options_status.ToString());
-      }
-      ROCKS_LOG_WARN(immutable_db_options_.info_log,
-                     "Unable to persist options in SetOptions() -- %s",
-                     persist_options_status.ToString().c_str());
+      s = persist_options_status;
    }
  } else {
    ROCKS_LOG_WARN(immutable_db_options_.info_log, "[%s] SetOptions() failed",
@@ -579,6 +548,9 @@ Status DBImpl::SetDBOptions(
      }

      write_controller_.set_max_delayed_write_rate(new_options.delayed_write_rate);
+      table_cache_.get()->SetCapacity(new_options.max_open_files == -1
+                                          ? TableCache::kInfiniteCapacity
+                                          : new_options.max_open_files - 10);

      mutable_db_options_ = new_options;

@@ -591,7 +563,8 @@ Status DBImpl::SetDBOptions(
                         purge_wal_status.ToString().c_str());
        }
      }
-      persist_options_status = WriteOptionsFile();
+      persist_options_status = WriteOptionsFile(
+          false /*need_mutex_lock*/, false /*need_enter_write_thread*/);
      write_thread_.ExitUnbatched(&w);
    }
  }
@@ -640,6 +613,26 @@ int DBImpl::FindMinimumEmptyLevelFitting(ColumnFamilyData* cfd,
  return minimum_level;
 }

+Status DBImpl::FlushWAL(bool sync) {
+  {
+    // We need to lock log_write_mutex_ since logs_ might change concurrently
+    InstrumentedMutexLock wl(&log_write_mutex_);
+    log::Writer* cur_log_writer = logs_.back().writer;
+    auto s = cur_log_writer->WriteBuffer();
+    if (!s.ok()) {
+      ROCKS_LOG_ERROR(immutable_db_options_.info_log, "WAL flush error %s",
+                      s.ToString().c_str());
+    }
+    if (!sync) {
+      ROCKS_LOG_DEBUG(immutable_db_options_.info_log, "FlushWAL sync=false");
+      return s;
+    }
+  }
+  // sync = true
+  ROCKS_LOG_DEBUG(immutable_db_options_.info_log, "FlushWAL sync=true");
+  return SyncWAL();
+}
+
 Status DBImpl::SyncWAL() {
  autovector<log::Writer*, 1> logs_to_sync;
  bool need_log_dir_sync;
@@ -678,6 +671,7 @@ Status DBImpl::SyncWAL() {
    need_log_dir_sync = !log_dir_synced_;
  }

+  TEST_SYNC_POINT("DBWALTest::SyncWALNotWaitWrite:1");
  RecordTick(stats_, WAL_FILE_SYNCED);
  Status status;
  for (log::Writer* log : logs_to_sync) {
@@ -689,6 +683,7 @@ Status DBImpl::SyncWAL() {
  if (status.ok() && need_log_dir_sync) {
    status = directories_.GetWalDir()->Fsync();
  }
+  TEST_SYNC_POINT("DBWALTest::SyncWALNotWaitWrite:2");

  TEST_SYNC_POINT("DBImpl::SyncWAL:BeforeMarkLogsSynced:1");
  {
@@ -1121,8 +1116,76 @@ std::vector<Status> DBImpl::MultiGet(
 }

 Status DBImpl::CreateColumnFamily(const ColumnFamilyOptions& cf_options,
-                                  const std::string& column_family_name,
+                                  const std::string& column_family,
                                  ColumnFamilyHandle** handle) {
+  assert(handle != nullptr);
+  Status s = CreateColumnFamilyImpl(cf_options, column_family, handle);
+  if (s.ok()) {
+    s = WriteOptionsFile(true /*need_mutex_lock*/,
+                         true /*need_enter_write_thread*/);
+  }
+  return s;
+}
+
+Status DBImpl::CreateColumnFamilies(
+    const ColumnFamilyOptions& cf_options,
+    const std::vector<std::string>& column_family_names,
+    std::vector<ColumnFamilyHandle*>* handles) {
+  assert(handles != nullptr);
+  handles->clear();
+  size_t num_cf = column_family_names.size();
+  Status s;
+  bool success_once = false;
+  for (size_t i = 0; i < num_cf; i++) {
+    ColumnFamilyHandle* handle;
+    s = CreateColumnFamilyImpl(cf_options, column_family_names[i], &handle);
+    if (!s.ok()) {
+      break;
+    }
+    handles->push_back(handle);
+    success_once = true;
+  }
+  if (success_once) {
+    Status persist_options_status = WriteOptionsFile(
+        true /*need_mutex_lock*/, true /*need_enter_write_thread*/);
+    if (s.ok() && !persist_options_status.ok()) {
+      s = persist_options_status;
+    }
+  }
+  return s;
+}
+
+Status DBImpl::CreateColumnFamilies(
+    const std::vector<ColumnFamilyDescriptor>& column_families,
+    std::vector<ColumnFamilyHandle*>* handles) {
+  assert(handles != nullptr);
+  handles->clear();
+  size_t num_cf = column_families.size();
+  Status s;
+  bool success_once = false;
+  for (size_t i = 0; i < num_cf; i++) {
+    ColumnFamilyHandle* handle;
+    s = CreateColumnFamilyImpl(column_families[i].options,
+                               column_families[i].name, &handle);
+    if (!s.ok()) {
+      break;
+    }
+    handles->push_back(handle);
+    success_once = true;
+  }
+  if (success_once) {
+    Status persist_options_status = WriteOptionsFile(
+        true /*need_mutex_lock*/, true /*need_enter_write_thread*/);
+    if (s.ok() && !persist_options_status.ok()) {
+      s = persist_options_status;
+    }
+  }
+  return s;
+}
+
+Status DBImpl::CreateColumnFamilyImpl(const ColumnFamilyOptions& cf_options,
+                                      const std::string& column_family_name,
+                                      ColumnFamilyHandle** handle) {
  Status s;
  Status persist_options_status;
  *handle = nullptr;
@@ -1159,12 +1222,6 @@ Status DBImpl::CreateColumnFamily(const ColumnFamilyOptions& cf_options,
      s = versions_->LogAndApply(nullptr, MutableCFOptions(cf_options), &edit,
                                 &mutex_, directories_.GetDbDir(), false,
                                 &cf_options);
-
-      if (s.ok()) {
-        // If the column family was created successfully, we then persist
-        // the updated RocksDB options under the same single write thread
-        persist_options_status = WriteOptionsFile();
-      }
      write_thread_.ExitUnbatched(&w);
    }
    if (s.ok()) {
@@ -1179,6 +1236,8 @@ Status DBImpl::CreateColumnFamily(const ColumnFamilyOptions& cf_options,
        is_snapshot_supported_ = false;
      }

+      cfd->set_initialized();
+
      *handle = new ColumnFamilyHandleImpl(cfd, this, &mutex_);
      ROCKS_LOG_INFO(immutable_db_options_.info_log,
                     "Created column family [%s] (ID %u)",
@@ -1194,22 +1253,42 @@ Status DBImpl::CreateColumnFamily(const ColumnFamilyOptions& cf_options,
  if (s.ok()) {
    NewThreadStatusCfInfo(
        reinterpret_cast<ColumnFamilyHandleImpl*>(*handle)->cfd());
-    if (!persist_options_status.ok()) {
-      if (immutable_db_options_.fail_if_options_file_error) {
-        s = Status::IOError(
-            "ColumnFamily has been created, but unable to persist"
-            "options in CreateColumnFamily()",
-            persist_options_status.ToString().c_str());
-      }
-      ROCKS_LOG_WARN(immutable_db_options_.info_log,
-                     "Unable to persist options in CreateColumnFamily() -- %s",
-                     persist_options_status.ToString().c_str());
-    }
  }
  return s;
 }

 Status DBImpl::DropColumnFamily(ColumnFamilyHandle* column_family) {
+  assert(column_family != nullptr);
+  Status s = DropColumnFamilyImpl(column_family);
+  if (s.ok()) {
+    s = WriteOptionsFile(true /*need_mutex_lock*/,
+                         true /*need_enter_write_thread*/);
+  }
+  return s;
+}
+
+Status DBImpl::DropColumnFamilies(
+    const std::vector<ColumnFamilyHandle*>& column_families) {
+  Status s;
+  bool success_once = false;
+  for (auto* handle : column_families) {
+    s = DropColumnFamilyImpl(handle);
+    if (!s.ok()) {
+      break;
+    }
+    success_once = true;
+  }
+  if (success_once) {
+    Status persist_options_status = WriteOptionsFile(
+        true /*need_mutex_lock*/, true /*need_enter_write_thread*/);
+    if (s.ok() && !persist_options_status.ok()) {
+      s = persist_options_status;
+    }
+  }
+  return s;
+}
+
+Status DBImpl::DropColumnFamilyImpl(ColumnFamilyHandle* column_family) {
  auto cfh = reinterpret_cast<ColumnFamilyHandleImpl*>(column_family);
  auto cfd = cfh->cfd();
  if (cfd->GetID() == 0) {
@@ -1223,7 +1302,6 @@ Status DBImpl::DropColumnFamily(ColumnFamilyHandle* column_family) {
  edit.SetColumnFamily(cfd->GetID());

  Status s;
-  Status options_persist_status;
  {
    InstrumentedMutexLock l(&mutex_);
    if (cfd->IsDropped()) {
@@ -1235,11 +1313,6 @@ Status DBImpl::DropColumnFamily(ColumnFamilyHandle* column_family) {
      write_thread_.EnterUnbatched(&w, &mutex_);
      s = versions_->LogAndApply(cfd, *cfd->GetLatestMutableCFOptions(),
                                 &edit, &mutex_);
-      if (s.ok()) {
-        // If the column family was dropped successfully, we then persist
-        // the updated RocksDB options under the same single write thread
-        options_persist_status = WriteOptionsFile();
-      }
      write_thread_.ExitUnbatched(&w);
    }

@@ -1268,18 +1341,6 @@ Status DBImpl::DropColumnFamily(ColumnFamilyHandle* column_family) {
                                  mutable_cf_options->max_write_buffer_number;
    ROCKS_LOG_INFO(immutable_db_options_.info_log,
                   "Dropped column family with id %u\n", cfd->GetID());
-
-    if (!options_persist_status.ok()) {
-      if (immutable_db_options_.fail_if_options_file_error) {
-        s = Status::IOError(
-            "ColumnFamily has been dropped, but unable to persist "
-            "options in DropColumnFamily()",
-            options_persist_status.ToString().c_str());
-      }
-      ROCKS_LOG_WARN(immutable_db_options_.info_log,
-                     "Unable to persist options in DropColumnFamily() -- %s",
-                     options_persist_status.ToString().c_str());
-    }
  } else {
    ROCKS_LOG_ERROR(immutable_db_options_.info_log,
                    "Dropping column family with id %u FAILED -- %s\n",
@@ -1515,6 +1576,14 @@ void DBImpl::ReleaseSnapshot(const Snapshot* s) {
  delete casted_s;
 }

+bool DBImpl::HasActiveSnapshotLaterThanSN(SequenceNumber sn) {
+  InstrumentedMutexLock l(&mutex_);
+  if (snapshots_.empty()) {
+    return false;
+  }
+  return (snapshots_.newest()->GetSequenceNumber() > sn);
+}
+
 #ifndef ROCKSDB_LITE
 Status DBImpl::GetPropertiesOfAllTables(ColumnFamilyHandle* column_family,
                                        TablePropertiesCollection* props) {
@@ -1666,6 +1735,18 @@ bool DBImpl::GetIntPropertyInternal(ColumnFamilyData* cfd,
  }
 }

+#ifndef ROCKSDB_LITE
+Status DBImpl::ResetStats() {
+  InstrumentedMutexLock l(&mutex_);
+  for (auto* cfd : *versions_->GetColumnFamilySet()) {
+    if (cfd->initialized()) {
+      cfd->internal_stats()->Clear();
+    }
+  }
+  return Status::OK();
+}
+#endif  // ROCKSDB_LITE
+
 bool DBImpl::GetAggregatedIntProperty(const Slice& property,
                                      uint64_t* aggregated_value) {
  const DBPropertyInfo* property_info = GetPropertyInfo(property);
@@ -1679,6 +1760,9 @@ bool DBImpl::GetAggregatedIntProperty(const Slice& property,
    InstrumentedMutexLock l(&mutex_);
    uint64_t value;
    for (auto* cfd : *versions_->GetColumnFamilySet()) {
+      if (!cfd->initialized()) {
+        continue;
+      }
      if (GetIntPropertyInternal(cfd, *property_info, true, &value)) {
        sum += value;
      } else {
@@ -1749,6 +1833,20 @@ ColumnFamilyHandle* DBImpl::GetColumnFamilyHandle(uint32_t column_family_id) {
  return cf_memtables->GetColumnFamilyHandle();
 }

+// REQUIRED: mutex is NOT held.
+ColumnFamilyHandle* DBImpl::GetColumnFamilyHandleUnlocked(
+    uint32_t column_family_id) {
+  ColumnFamilyMemTables* cf_memtables = column_family_memtables_.get();
+
+  InstrumentedMutexLock l(&mutex_);
+
+  if (!cf_memtables->Seek(column_family_id)) {
+    return nullptr;
+  }
+
+  return cf_memtables->GetColumnFamilyHandle();
+}
+
 void DBImpl::GetApproximateMemTableStats(ColumnFamilyHandle* column_family,
                                         const Range& range,
                                         uint64_t* const count,
@@ -2097,9 +2195,29 @@ Status DB::CreateColumnFamily(const ColumnFamilyOptions& cf_options,
                              ColumnFamilyHandle** handle) {
  return Status::NotSupported("");
 }
+
+Status DB::CreateColumnFamilies(
+    const ColumnFamilyOptions& cf_options,
+    const std::vector<std::string>& column_family_names,
+    std::vector<ColumnFamilyHandle*>* handles) {
+  return Status::NotSupported("");
+}
+
+Status DB::CreateColumnFamilies(
+    const std::vector<ColumnFamilyDescriptor>& column_families,
+    std::vector<ColumnFamilyHandle*>* handles) {
+  return Status::NotSupported("");
+}
+
 Status DB::DropColumnFamily(ColumnFamilyHandle* column_family) {
  return Status::NotSupported("");
 }
+
+Status DB::DropColumnFamilies(
+    const std::vector<ColumnFamilyHandle*>& column_families) {
+  return Status::NotSupported("");
+}
+
 Status DB::DestroyColumnFamilyHandle(ColumnFamilyHandle* column_family) {
  delete column_family;
  return Status::OK();
@@ -2206,9 +2324,18 @@ Status DestroyDB(const std::string& dbname, const Options& options) {
  return result;
 }

-Status DBImpl::WriteOptionsFile() {
+Status DBImpl::WriteOptionsFile(bool need_mutex_lock,
+                                bool need_enter_write_thread) {
 #ifndef ROCKSDB_LITE
-  mutex_.AssertHeld();
+  WriteThread::Writer w;
+  if (need_mutex_lock) {
+    mutex_.Lock();
+  } else {
+    mutex_.AssertHeld();
+  }
+  if (need_enter_write_thread) {
+    write_thread_.EnterUnbatched(&w, &mutex_);
+  }

  std::vector<std::string> cf_names;
  std::vector<ColumnFamilyOptions> cf_opts;
@@ -2228,6 +2355,9 @@ Status DBImpl::WriteOptionsFile() {
      BuildDBOptions(immutable_db_options_, mutable_db_options_);
  mutex_.Unlock();

+  TEST_SYNC_POINT("DBImpl::WriteOptionsFile:1");
+  TEST_SYNC_POINT("DBImpl::WriteOptionsFile:2");
+
  std::string file_name =
      TempOptionsFileName(GetName(), versions_->NewFileNumber());
  Status s =
@@ -2236,11 +2366,23 @@ Status DBImpl::WriteOptionsFile() {
  if (s.ok()) {
    s = RenameTempFileToOptionsFile(file_name);
  }
-  mutex_.Lock();
-  return s;
-#else
-  return Status::OK();
+  // restore lock
+  if (!need_mutex_lock) {
+    mutex_.Lock();
+  }
+  if (need_enter_write_thread) {
+    write_thread_.ExitUnbatched(&w);
+  }
+  if (!s.ok()) {
+    ROCKS_LOG_WARN(immutable_db_options_.info_log,
+                   "Unnable to persist options -- %s", s.ToString().c_str());
+    if (immutable_db_options_.fail_if_options_file_error) {
+      return Status::IOError("Unable to persist options.",
+                             s.ToString().c_str());
+    }
+  }
 #endif  // !ROCKSDB_LITE
+  return Status::OK();
 }

 #ifndef ROCKSDB_LITE
@@ -2351,7 +2493,7 @@ void DBImpl::EraseThreadStatusDbInfo() const {
 // A global method that can dump out the build version
 void DumpRocksDBBuildVersion(Logger * log) {
 #if !defined(IOS_CROSS_COMPILE)
-  // if we compile with Xcode, we don't run build_detect_vesion, so we don't
+  // if we compile with Xcode, we don't run build_detect_version, so we don't
  // generate util/build_version.cc
  ROCKS_LOG_HEADER(log, "RocksDB version: %d.%d.%d\n", ROCKSDB_MAJOR,
                   ROCKSDB_MINOR, ROCKSDB_PATCH);
@@ -2479,6 +2621,15 @@ Status DBImpl::IngestExternalFile(
  auto cfh = reinterpret_cast<ColumnFamilyHandleImpl*>(column_family);
  auto cfd = cfh->cfd();

+  // Ingest should immediately fail if ingest_behind is requested,
+  // but the DB doesn't support it.
+  if (ingestion_options.ingest_behind) {
+    if (!immutable_db_options_.allow_ingest_behind) {
+      return Status::InvalidArgument(
+        "Can't ingest_behind file in DB with allow_ingest_behind=false");
+    }
+  }
+
  ExternalSstFileIngestionJob ingestion_job(env_, versions_.get(), cfd,
                                            immutable_db_options_, env_options_,
                                            &snapshots_, ingestion_options);
@@ -2506,9 +2657,13 @@ Status DBImpl::IngestExternalFile(
    InstrumentedMutexLock l(&mutex_);
    TEST_SYNC_POINT("DBImpl::AddFile:MutexLock");

-    // Stop writes to the DB
+    // Stop writes to the DB by entering both write threads
    WriteThread::Writer w;
    write_thread_.EnterUnbatched(&w, &mutex_);
+    WriteThread::Writer nonmem_w;
+    if (concurrent_prepare_) {
+      nonmem_write_thread_.EnterUnbatched(&nonmem_w, &mutex_);
+    }

    num_running_ingest_file_++;

@@ -2522,7 +2677,8 @@ Status DBImpl::IngestExternalFile(
    if (status.ok()) {
      bool need_flush = false;
      status = ingestion_job.NeedsFlush(&need_flush);
-
+      TEST_SYNC_POINT_CALLBACK("DBImpl::IngestExternalFile:NeedFlush",
+                               &need_flush);
      if (status.ok() && need_flush) {
        mutex_.Unlock();
        status = FlushMemTable(cfd, FlushOptions(), true /* writes_stopped */);
@@ -2548,6 +2704,9 @@ Status DBImpl::IngestExternalFile(
    }

    // Resume writes to the DB
+    if (concurrent_prepare_) {
+      nonmem_write_thread_.ExitUnbatched(&nonmem_w);
+    }
    write_thread_.ExitUnbatched(&w);

    // Update stats
@@ -1,7 +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 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
@@ -100,10 +100,19 @@ class DBImpl : public DB {
      const std::vector<Slice>& keys,
      std::vector<std::string>* values) override;

-  virtual Status CreateColumnFamily(const ColumnFamilyOptions& options,
+  virtual Status CreateColumnFamily(const ColumnFamilyOptions& cf_options,
                                    const std::string& column_family,
                                    ColumnFamilyHandle** handle) override;
+  virtual Status CreateColumnFamilies(
+      const ColumnFamilyOptions& cf_options,
+      const std::vector<std::string>& column_family_names,
+      std::vector<ColumnFamilyHandle*>* handles) override;
+  virtual Status CreateColumnFamilies(
+      const std::vector<ColumnFamilyDescriptor>& column_families,
+      std::vector<ColumnFamilyHandle*>* handles) override;
  virtual Status DropColumnFamily(ColumnFamilyHandle* column_family) override;
+  virtual Status DropColumnFamilies(
+      const std::vector<ColumnFamilyHandle*>& column_families) override;

  // Returns false if key doesn't exist in the database and true if it may.
  // If value_found is not passed in as null, then return the value if found in
@@ -188,11 +197,16 @@ class DBImpl : public DB {
  using DB::Flush;
  virtual Status Flush(const FlushOptions& options,
                       ColumnFamilyHandle* column_family) override;
+  virtual Status FlushWAL(bool sync) override;
  virtual Status SyncWAL() override;

  virtual SequenceNumber GetLatestSequenceNumber() const override;

+  bool HasActiveSnapshotLaterThanSN(SequenceNumber sn);
+
 #ifndef ROCKSDB_LITE
+  using DB::ResetStats;
+  virtual Status ResetStats() override;
  virtual Status DisableFileDeletions() override;
  virtual Status EnableFileDeletions(bool force) override;
  virtual int IsFileDeletionsEnabled() const;
@@ -374,9 +388,9 @@ class DBImpl : public DB {
  Status TEST_GetAllImmutableCFOptions(
      std::unordered_map<std::string, const ImmutableCFOptions*>* iopts_map);

-  // Return the lastest MutableCFOptions of of a column family
+  // Return the lastest MutableCFOptions of a column family
  Status TEST_GetLatestMutableCFOptions(ColumnFamilyHandle* column_family,
-                                        MutableCFOptions* mutable_cf_opitons);
+                                        MutableCFOptions* mutable_cf_options);

  Cache* TEST_table_cache() { return table_cache_.get(); }

@@ -386,12 +400,21 @@ class DBImpl : public DB {
  uint64_t TEST_FindMinPrepLogReferencedByMemTable();

  int TEST_BGCompactionsAllowed() const;
+  int TEST_BGFlushesAllowed() const;

 #endif  // NDEBUG

-  // Return maximum background compaction allowed to be scheduled based on
-  // compaction status.
-  int BGCompactionsAllowed() const;
+  struct BGJobLimits {
+    int max_flushes;
+    int max_compactions;
+  };
+  // Returns maximum background flushes and compactions allowed to be scheduled
+  BGJobLimits GetBGJobLimits() const;
+  // Need a static version that can be called during SanitizeOptions().
+  static BGJobLimits GetBGJobLimits(int max_background_flushes,
+                                    int max_background_compactions,
+                                    int max_background_jobs,
+                                    bool parallelize_compactions);

  // move logs pending closing from job_context to the DB queue and
  // schedule a purge
@@ -452,6 +475,9 @@ class DBImpl : public DB {
  // mutex is released.
  ColumnFamilyHandle* GetColumnFamilyHandle(uint32_t column_family_id);

+  // Same as above, should called without mutex held and not on write thread.
+  ColumnFamilyHandle* GetColumnFamilyHandleUnlocked(uint32_t column_family_id);
+
  // Returns the number of currently running flushes.
  // REQUIREMENT: mutex_ must be held when calling this function.
  int num_running_flushes() {
@@ -547,9 +573,10 @@ class DBImpl : public DB {
                                        RangeDelAggregator* range_del_agg);

  // Except in DB::Open(), WriteOptionsFile can only be called when:
-  // 1. WriteThread::Writer::EnterUnbatched() is used.
-  // 2. db_mutex is held
-  Status WriteOptionsFile();
+  // Persist options to options file.
+  // If need_mutex_lock = false, the method will lock DB mutex.
+  // If need_enter_write_thread = false, the method will enter write thread.
+  Status WriteOptionsFile(bool need_mutex_lock, bool need_enter_write_thread);

  // The following two functions can only be called when:
  // 1. WriteThread::Writer::EnterUnbatched() is used.
@@ -557,6 +584,10 @@ class DBImpl : public DB {
  Status RenameTempFileToOptionsFile(const std::string& file_name);
  Status DeleteObsoleteOptionsFiles();

+  void NotifyOnFlushBegin(ColumnFamilyData* cfd, FileMetaData* file_meta,
+                          const MutableCFOptions& mutable_cf_options,
+                          int job_id, TableProperties prop);
+
  void NotifyOnFlushCompleted(ColumnFamilyData* cfd, FileMetaData* file_meta,
                              const MutableCFOptions& mutable_cf_options,
                              int job_id, TableProperties prop);
@@ -584,6 +615,15 @@ class DBImpl : public DB {
                   uint64_t* log_used = nullptr, uint64_t log_ref = 0,
                   bool disable_memtable = false);

+  Status PipelinedWriteImpl(const WriteOptions& options, WriteBatch* updates,
+                            WriteCallback* callback = nullptr,
+                            uint64_t* log_used = nullptr, uint64_t log_ref = 0,
+                            bool disable_memtable = false);
+
+  Status WriteImplWALOnly(const WriteOptions& options, WriteBatch* updates,
+                          WriteCallback* callback = nullptr,
+                          uint64_t* log_used = nullptr, uint64_t log_ref = 0);
+
  uint64_t FindMinLogContainingOutstandingPrep();
  uint64_t FindMinPrepLogReferencedByMemTable();

@@ -628,6 +668,12 @@ class DBImpl : public DB {

  const Status CreateArchivalDirectory();

+  Status CreateColumnFamilyImpl(const ColumnFamilyOptions& cf_options,
+                                const std::string& cf_name,
+                                ColumnFamilyHandle** handle);
+
+  Status DropColumnFamilyImpl(ColumnFamilyHandle* column_family);
+
  // Delete any unneeded files and stale in-memory entries.
  void DeleteObsoleteFiles();
  // Delete obsolete files and log status and information of file deletion
@@ -679,6 +725,9 @@ class DBImpl : public DB {
  //            `num_bytes` going through.
  Status DelayWrite(uint64_t num_bytes, const WriteOptions& write_options);

+  Status ThrottleLowPriWritesIfNeeded(const WriteOptions& write_options,
+                                      WriteBatch* my_batch);
+
  Status ScheduleFlushes(WriteContext* context);

  Status SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context);
@@ -697,12 +746,29 @@ class DBImpl : public DB {
  Status HandleWriteBufferFull(WriteContext* write_context);

  // REQUIRES: mutex locked
-  Status PreprocessWrite(const WriteOptions& write_options, bool need_log_sync,
-                         bool* logs_getting_syned, WriteContext* write_context);
+  Status PreprocessWrite(const WriteOptions& write_options, bool* need_log_sync,
+                         WriteContext* write_context);

-  Status WriteToWAL(const autovector<WriteThread::Writer*>& write_group,
-                    log::Writer* log_writer, bool need_log_sync,
-                    bool need_log_dir_sync, SequenceNumber sequence);
+  WriteBatch* MergeBatch(const WriteThread::WriteGroup& write_group,
+                         WriteBatch* tmp_batch, size_t* write_with_wal);
+
+  Status WriteToWAL(const WriteBatch& merged_batch, log::Writer* log_writer,
+                    uint64_t* log_used, uint64_t* log_size);
+
+  Status WriteToWAL(const WriteThread::WriteGroup& write_group,
+                    log::Writer* log_writer, uint64_t* log_used,
+                    bool need_log_sync, bool need_log_dir_sync,
+                    SequenceNumber sequence);
+
+  Status ConcurrentWriteToWAL(const WriteThread::WriteGroup& write_group,
+                              uint64_t* log_used, SequenceNumber* last_sequence,
+                              int total_count);
+
+  // Used by WriteImpl to update bg_error_ if paranoid check is enabled.
+  void WriteCallbackStatusCheck(const Status& status);
+
+  // Used by WriteImpl to update bg_error_ in case of memtable insert error.
+  void MemTableInsertStatusCheck(const Status& memtable_insert_status);

 #ifndef ROCKSDB_LITE

@@ -775,11 +841,16 @@ class DBImpl : public DB {
  // Lock over the persistent DB state.  Non-nullptr iff successfully acquired.
  FileLock* db_lock_;

-  // The mutex for options file related operations.
-  // NOTE: should never acquire options_file_mutex_ and mutex_ at the
-  //       same time.
-  InstrumentedMutex options_files_mutex_;
+  // In addition to mutex_, log_write_mutex_ protected writes to logs_ and
+  // logfile_number_. With concurrent_prepare 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
+  // WriteToWAL need different synchronization: log_empty_, alive_log_files_,
+  // logs_, logfile_number_. Refer to the definition of each variable below for
+  // more description.
  mutable InstrumentedMutex mutex_;

  std::atomic<bool> shutting_down_;
@@ -793,10 +864,20 @@ class DBImpl : public DB {
  // * whenever there is an error in background purge, flush or compaction
  // * whenever num_running_ingest_file_ 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,
+  // logfile_number_ is currently updated only in write_thread_, it can be read
+  // 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
  bool log_dir_synced_;
+  // Without concurrent_prepare, 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,
+  // 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_;
  ColumnFamilyHandleImpl* default_cf_handle_;
  InternalStats* default_cf_internal_stats_;
@@ -831,14 +912,26 @@ 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
+  // 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
+  // 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_;
  // Log files that aren't fully synced, and the current log file.
  // Synchronization:
-  //  - push_back() is done from write thread with locked mutex_,
-  //  - pop_front() is done from any thread with locked mutex_,
+  //  - push_back() is done from write_thread_ with locked mutex_ and
+  //  log_write_mutex_
+  //  - pop_front() is done from any thread with locked mutex_ and
+  //  log_write_mutex_
+  //  - reads are done with either locked mutex_ or log_write_mutex_
  //  - back() and items with getting_synced=true are not popped,
-  //  - it follows that write thread with unlocked mutex_ can safely access
-  //    back() and items with getting_synced=true.
+  //  - The same thread that sets getting_synced=true will reset it.
+  //  - it follows that the object referred by back() can be safely read from
+  //  the write_thread_ without using mutex
+  //  - it follows that the items with getting_synced=true can be safely read
+  //  from the same thread that has set getting_synced=true
  std::deque<LogWriterNumber> logs_;
  // Signaled when getting_synced becomes false for some of the logs_.
  InstrumentedCondVar log_sync_cv_;
@@ -887,13 +980,19 @@ class DBImpl : public DB {
  WriteBufferManager* write_buffer_manager_;

  WriteThread write_thread_;
-
  WriteBatch tmp_batch_;
+  // The write thread when the writers have no memtable write. This will be used
+  // in 2PC to batch the prepares separately from the serial commit.
+  WriteThread nonmem_write_thread_;

  WriteController write_controller_;

+  unique_ptr<RateLimiter> low_pri_write_rate_limiter_;
+
  // Size of the last batch group. In slowdown mode, next write needs to
  // sleep if it uses up the quota.
+  // Note: This is to protect memtable and compaction. If the batch only writes
+  // to the WAL its size need not to be included in this.
  uint64_t last_batch_group_size_;

  FlushScheduler flush_scheduler_;
@@ -1065,7 +1164,7 @@ class DBImpl : public DB {
  // Indicate DB was opened successfully
  bool opened_successfully_;

-  // minmum log number still containing prepared data.
+  // minimum log number still containing prepared data.
  // this is used by FindObsoleteFiles to determine which
  // flushed logs we must keep around because they still
  // contain prepared data which has not been flushed or rolled back
@@ -1078,7 +1177,7 @@ class DBImpl : public DB {
  // to prepared_section_completed_ which maps LOG -> instance_count
  // since a log could contain multiple prepared sections
  //
-  // when trying to determine the minmum log still active we first
+  // when trying to determine the minimum log still active we first
  // consult min_log_with_prep_. while that root value maps to
  // a value > 0 in prepared_section_completed_ we decrement the
  // instance_count for that log and pop the root value in
@@ -1091,13 +1190,6 @@ class DBImpl : public DB {
  DBImpl(const DBImpl&);
  void operator=(const DBImpl&);

-  // Return the earliest snapshot where seqno is visible.
-  // Store the snapshot right before that, if any, in prev_snapshot
-  inline SequenceNumber findEarliestVisibleSnapshot(
-    SequenceNumber in,
-    std::vector<SequenceNumber>& snapshots,
-    SequenceNumber* prev_snapshot);
-
  // 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
@@ -1143,6 +1235,11 @@ class DBImpl : public DB {
  bool MCOverlap(ManualCompaction* m, ManualCompaction* m1);

  size_t GetWalPreallocateBlockSize(uint64_t write_buffer_size) const;
+
+  // 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 manual_wal_flush_;
 };

 extern Options SanitizeOptions(const std::string& db,
@@ -1153,7 +1250,7 @@ extern DBOptions SanitizeOptions(const std::string& db, const DBOptions& src);
 extern CompressionType GetCompressionFlush(
    const ImmutableCFOptions& ioptions,
    const MutableCFOptions& mutable_cf_options);
- 
+
 // Fix user-supplied options to be reasonable
 template <class T, class V>
 static void ClipToRange(T* ptr, V minvalue, V maxvalue) {
@@ -1,7 +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 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
@@ -14,6 +14,7 @@
 #include <inttypes.h>

 #include "db/builder.h"
+#include "db/event_helpers.h"
 #include "monitoring/iostats_context_imp.h"
 #include "monitoring/perf_context_imp.h"
 #include "monitoring/thread_status_updater.h"
@@ -59,7 +60,14 @@ Status DBImpl::SyncClosedLogs(JobContext* job_context) {
    // "number < current_log_number".
    MarkLogsSynced(current_log_number - 1, true, s);
    if (!s.ok()) {
-      bg_error_ = s;
+      Status new_bg_error = s;
+      // may temporarily unlock and lock the mutex.
+      EventHelpers::NotifyOnBackgroundError(immutable_db_options_.listeners,
+                                            BackgroundErrorReason::kFlush,
+                                            &new_bg_error, &mutex_);
+      if (!new_bg_error.ok()) {
+        bg_error_ = new_bg_error;
+      }
      TEST_SYNC_POINT("DBImpl::SyncClosedLogs:Failed");
      return s;
    }
@@ -90,6 +98,12 @@ Status DBImpl::FlushMemTableToOutputFile(

  flush_job.PickMemTable();

+#ifndef ROCKSDB_LITE
+  // may temporarily unlock and lock the mutex.
+  NotifyOnFlushBegin(cfd, &file_meta, mutable_cf_options, job_context->job_id,
+                     flush_job.GetTableProperties());
+#endif  // ROCKSDB_LITE
+
  Status s;
  if (logfile_number_ > 0 &&
      versions_->GetColumnFamilySet()->NumberOfColumnFamilies() > 0) {
@@ -128,9 +142,16 @@ Status DBImpl::FlushMemTableToOutputFile(

  if (!s.ok() && !s.IsShutdownInProgress() &&
      immutable_db_options_.paranoid_checks && bg_error_.ok()) {
-    // if a bad error happened (not ShutdownInProgress) and paranoid_checks is
-    // true, mark DB read-only
-    bg_error_ = s;
+    Status new_bg_error = s;
+    // may temporarily unlock and lock the mutex.
+    EventHelpers::NotifyOnBackgroundError(immutable_db_options_.listeners,
+                                          BackgroundErrorReason::kFlush,
+                                          &new_bg_error, &mutex_);
+    if (!new_bg_error.ok()) {
+      // if a bad error happened (not ShutdownInProgress), paranoid_checks is
+      // true, and the error isn't handled by callback, mark DB read-only
+      bg_error_ = new_bg_error;
+    }
  }
  if (s.ok()) {
 #ifndef ROCKSDB_LITE
@@ -145,10 +166,17 @@ Status DBImpl::FlushMemTableToOutputFile(
          immutable_db_options_.db_paths[0].path, file_meta.fd.GetNumber());
      sfm->OnAddFile(file_path);
      if (sfm->IsMaxAllowedSpaceReached() && bg_error_.ok()) {
-        bg_error_ = Status::IOError("Max allowed space was reached");
+        Status new_bg_error = Status::IOError("Max allowed space was reached");
        TEST_SYNC_POINT_CALLBACK(
            "DBImpl::FlushMemTableToOutputFile:MaxAllowedSpaceReached",
-            &bg_error_);
+            &new_bg_error);
+        // may temporarily unlock and lock the mutex.
+        EventHelpers::NotifyOnBackgroundError(immutable_db_options_.listeners,
+                                              BackgroundErrorReason::kFlush,
+                                              &new_bg_error, &mutex_);
+        if (!new_bg_error.ok()) {
+          bg_error_ = new_bg_error;
+        }
      }
    }
 #endif  // ROCKSDB_LITE
@@ -156,6 +184,49 @@ Status DBImpl::FlushMemTableToOutputFile(
  return s;
 }

+void DBImpl::NotifyOnFlushBegin(ColumnFamilyData* cfd, FileMetaData* file_meta,
+                                const MutableCFOptions& mutable_cf_options,
+                                int job_id, TableProperties prop) {
+#ifndef ROCKSDB_LITE
+  if (immutable_db_options_.listeners.size() == 0U) {
+    return;
+  }
+  mutex_.AssertHeld();
+  if (shutting_down_.load(std::memory_order_acquire)) {
+    return;
+  }
+  bool triggered_writes_slowdown =
+      (cfd->current()->storage_info()->NumLevelFiles(0) >=
+       mutable_cf_options.level0_slowdown_writes_trigger);
+  bool triggered_writes_stop =
+      (cfd->current()->storage_info()->NumLevelFiles(0) >=
+       mutable_cf_options.level0_stop_writes_trigger);
+  // release lock while notifying events
+  mutex_.Unlock();
+  {
+    FlushJobInfo info;
+    info.cf_name = cfd->GetName();
+    // TODO(yhchiang): make db_paths dynamic in case flush does not
+    //                 go to L0 in the future.
+    info.file_path = MakeTableFileName(immutable_db_options_.db_paths[0].path,
+                                       file_meta->fd.GetNumber());
+    info.thread_id = env_->GetThreadID();
+    info.job_id = job_id;
+    info.triggered_writes_slowdown = triggered_writes_slowdown;
+    info.triggered_writes_stop = triggered_writes_stop;
+    info.smallest_seqno = file_meta->smallest_seqno;
+    info.largest_seqno = file_meta->largest_seqno;
+    info.table_properties = prop;
+    for (auto listener : immutable_db_options_.listeners) {
+      listener->OnFlushBegin(this, info);
+    }
+  }
+  mutex_.Lock();
+// no need to signal bg_cv_ as it will be signaled at the end of the
+// flush process.
+#endif  // ROCKSDB_LITE
+}
+
 void DBImpl::NotifyOnFlushCompleted(ColumnFamilyData* cfd,
                                    FileMetaData* file_meta,
                                    const MutableCFOptions& mutable_cf_options,
@@ -233,10 +304,14 @@ Status DBImpl::CompactRange(const CompactRangeOptions& options,
  if (cfd->ioptions()->compaction_style == kCompactionStyleUniversal &&
      cfd->NumberLevels() > 1) {
    // Always compact all files together.
-    s = RunManualCompaction(cfd, ColumnFamilyData::kCompactAllLevels,
-                            cfd->NumberLevels() - 1, options.target_path_id,
-                            begin, end, exclusive);
    final_output_level = cfd->NumberLevels() - 1;
+    // if bottom most level is reserved
+    if (immutable_db_options_.allow_ingest_behind) {
+      final_output_level--;
+    }
+    s = RunManualCompaction(cfd, ColumnFamilyData::kCompactAllLevels,
+                            final_output_level, options.target_path_id,
+                            begin, end, exclusive);
  } else {
    for (int level = 0; level <= max_level_with_files; level++) {
      int output_level;
@@ -512,7 +587,14 @@ Status DBImpl::CompactFilesImpl(
                   c->column_family_data()->GetName().c_str(),
                   job_context->job_id, status.ToString().c_str());
    if (immutable_db_options_.paranoid_checks && bg_error_.ok()) {
-      bg_error_ = status;
+      Status new_bg_error = status;
+      // may temporarily unlock and lock the mutex.
+      EventHelpers::NotifyOnBackgroundError(immutable_db_options_.listeners,
+                                            BackgroundErrorReason::kCompaction,
+                                            &new_bg_error, &mutex_);
+      if (!new_bg_error.ok()) {
+        bg_error_ = new_bg_error;
+      }
    }
  }

@@ -777,6 +859,7 @@ Status DBImpl::RunManualCompaction(ColumnFamilyData* cfd, int input_level,
  TEST_SYNC_POINT_CALLBACK("DBImpl::RunManualCompaction:NotScheduled", &mutex_);
  if (exclusive) {
    while (bg_compaction_scheduled_ > 0) {
+      TEST_SYNC_POINT("DBImpl::RunManualCompaction:WaitScheduled");
      ROCKS_LOG_INFO(
          immutable_db_options_.info_log,
          "[%s] Manual compaction waiting for all other scheduled background "
@@ -927,22 +1010,22 @@ void DBImpl::MaybeScheduleFlushOrCompaction() {
    // DB is being deleted; no more background compactions
    return;
  }
-
-  while (unscheduled_flushes_ > 0 &&
-         bg_flush_scheduled_ < immutable_db_options_.max_background_flushes) {
+  auto bg_job_limits = GetBGJobLimits();
+  bool is_flush_pool_empty =
+    env_->GetBackgroundThreads(Env::Priority::HIGH) == 0;
+  while (!is_flush_pool_empty && unscheduled_flushes_ > 0 &&
+         bg_flush_scheduled_ < bg_job_limits.max_flushes) {
    unscheduled_flushes_--;
    bg_flush_scheduled_++;
    env_->Schedule(&DBImpl::BGWorkFlush, this, Env::Priority::HIGH, this);
  }

-  auto bg_compactions_allowed = BGCompactionsAllowed();
-
-  // special case -- if max_background_flushes == 0, then schedule flush on a
-  // compaction thread
-  if (immutable_db_options_.max_background_flushes == 0) {
+  // special case -- if high-pri (flush) thread pool is empty, then schedule
+  // flushes in low-pri (compaction) thread pool.
+  if (is_flush_pool_empty) {
    while (unscheduled_flushes_ > 0 &&
           bg_flush_scheduled_ + bg_compaction_scheduled_ <
-               bg_compactions_allowed) {
+               bg_job_limits.max_flushes) {
      unscheduled_flushes_--;
      bg_flush_scheduled_++;
      env_->Schedule(&DBImpl::BGWorkFlush, this, Env::Priority::LOW, this);
@@ -960,7 +1043,7 @@ void DBImpl::MaybeScheduleFlushOrCompaction() {
    return;
  }

-  while (bg_compaction_scheduled_ < bg_compactions_allowed &&
+  while (bg_compaction_scheduled_ < bg_job_limits.max_compactions &&
         unscheduled_compactions_ > 0) {
    CompactionArg* ca = new CompactionArg;
    ca->db = this;
@@ -972,13 +1055,35 @@ void DBImpl::MaybeScheduleFlushOrCompaction() {
  }
 }

-int DBImpl::BGCompactionsAllowed() const {
+DBImpl::BGJobLimits DBImpl::GetBGJobLimits() const {
  mutex_.AssertHeld();
-  if (write_controller_.NeedSpeedupCompaction()) {
-    return mutable_db_options_.max_background_compactions;
+  return GetBGJobLimits(immutable_db_options_.max_background_flushes,
+                        mutable_db_options_.max_background_compactions,
+                        mutable_db_options_.max_background_jobs,
+                        write_controller_.NeedSpeedupCompaction());
+}
+
+DBImpl::BGJobLimits DBImpl::GetBGJobLimits(int max_background_flushes,
+                                           int max_background_compactions,
+                                           int max_background_jobs,
+                                           bool parallelize_compactions) {
+  BGJobLimits res;
+  if (max_background_flushes == -1 && max_background_compactions == -1) {
+    // for our first stab implementing max_background_jobs, simply allocate a
+    // quarter of the threads to flushes.
+    res.max_flushes = std::max(1, max_background_jobs / 4);
+    res.max_compactions = std::max(1, max_background_jobs - res.max_flushes);
  } else {
-    return mutable_db_options_.base_background_compactions;
+    // compatibility code in case users haven't migrated to max_background_jobs,
+    // which automatically computes flush/compaction limits
+    res.max_flushes = std::max(1, max_background_flushes);
+    res.max_compactions = std::max(1, max_background_compactions);
  }
+  if (!parallelize_compactions) {
+    // throttle background compactions until we deem necessary
+    res.max_compactions = 1;
+  }
+  return res;
 }

 void DBImpl::AddToCompactionQueue(ColumnFamilyData* cfd) {
@@ -1100,13 +1205,15 @@ Status DBImpl::BackgroundFlush(bool* made_progress, JobContext* job_context,
  if (cfd != nullptr) {
    const MutableCFOptions mutable_cf_options =
        *cfd->GetLatestMutableCFOptions();
+    auto bg_job_limits = GetBGJobLimits();
    ROCKS_LOG_BUFFER(
        log_buffer,
        "Calling FlushMemTableToOutputFile with column "
-        "family [%s], flush slots available %d, compaction slots allowed %d, "
-        "compaction slots scheduled %d",
-        cfd->GetName().c_str(), immutable_db_options_.max_background_flushes,
-        bg_flush_scheduled_, BGCompactionsAllowed() - bg_compaction_scheduled_);
+        "family [%s], flush slots available %d, compaction slots available %d, "
+        "flush slots scheduled %d, compaction slots scheduled %d",
+        cfd->GetName().c_str(), bg_job_limits.max_flushes,
+        bg_job_limits.max_compactions, bg_flush_scheduled_,
+        bg_compaction_scheduled_);
    status = FlushMemTableToOutputFile(cfd, mutable_cf_options, made_progress,
                                       job_context, log_buffer);
    if (cfd->Unref()) {
@@ -1119,7 +1226,6 @@ Status DBImpl::BackgroundFlush(bool* made_progress, JobContext* job_context,
 void DBImpl::BackgroundCallFlush() {
  bool made_progress = false;
  JobContext job_context(next_job_id_.fetch_add(1), true);
-  assert(bg_flush_scheduled_);

  TEST_SYNC_POINT("DBImpl::BackgroundCallFlush:start");

@@ -1127,6 +1233,7 @@ void DBImpl::BackgroundCallFlush() {
                       immutable_db_options_.info_log.get());
  {
    InstrumentedMutexLock l(&mutex_);
+    assert(bg_flush_scheduled_);
    num_running_flushes_++;

    auto pending_outputs_inserted_elem =
@@ -1343,6 +1450,16 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
               : m->manual_end->DebugString().c_str()));
    }
  } else if (!compaction_queue_.empty()) {
+    if (HaveManualCompaction(compaction_queue_.front())) {
+      // Can't compact right now, but try again later
+      TEST_SYNC_POINT("DBImpl::BackgroundCompaction()::Conflict");
+
+      // Stay in the compaction queue.
+      unscheduled_compactions_++;
+
+      return Status::OK();
+    }
+
    // cfd is referenced here
    auto cfd = PopFirstFromCompactionQueue();
    // We unreference here because the following code will take a Ref() on
@@ -1357,12 +1474,6 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
      return Status::OK();
    }

-    if (HaveManualCompaction(cfd)) {
-      // Can't compact right now, but try again later
-      TEST_SYNC_POINT("DBImpl::BackgroundCompaction()::Conflict");
-      return Status::OK();
-    }
-
    // Pick up latest mutable CF Options and use it throughout the
    // compaction job
    // Compaction makes a copy of the latest MutableCFOptions. It should be used
@@ -1541,7 +1652,14 @@ Status DBImpl::BackgroundCompaction(bool* made_progress,
    ROCKS_LOG_WARN(immutable_db_options_.info_log, "Compaction error: %s",
                   status.ToString().c_str());
    if (immutable_db_options_.paranoid_checks && bg_error_.ok()) {
-      bg_error_ = status;
+      Status new_bg_error = status;
+      // may temporarily unlock and lock the mutex.
+      EventHelpers::NotifyOnBackgroundError(immutable_db_options_.listeners,
+                                            BackgroundErrorReason::kCompaction,
+                                            &new_bg_error, &mutex_);
+      if (!new_bg_error.ok()) {
+        bg_error_ = new_bg_error;
+      }
    }
  }

@@ -1646,7 +1764,7 @@ bool DBImpl::HaveManualCompaction(ColumnFamilyData* cfd) {
    }
    if ((cfd == (*it)->cfd) && (!((*it)->in_progress || (*it)->done))) {
      // Allow automatic compaction if manual compaction is
-      // is in progress
+      // in progress
      return true;
    }
    it++;
@@ -1,7 +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 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
@@ -186,7 +186,12 @@ Status DBImpl::TEST_GetLatestMutableCFOptions(

 int DBImpl::TEST_BGCompactionsAllowed() const {
  InstrumentedMutexLock l(&mutex_);
-  return BGCompactionsAllowed();
+  return GetBGJobLimits().max_compactions;
+}
+
+int DBImpl::TEST_BGFlushesAllowed() const {
+  InstrumentedMutexLock l(&mutex_);
+  return GetBGJobLimits().max_flushes;
 }

 }  // namespace rocksdb
@@ -1,7 +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 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
@@ -1,7 +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 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
@@ -250,7 +250,13 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
      }
      job_context->size_log_to_delete += earliest.size;
      total_log_size_ -= earliest.size;
+      if (concurrent_prepare_) {
+        log_write_mutex_.Lock();
+      }
      alive_log_files_.pop_front();
+      if (concurrent_prepare_) {
+        log_write_mutex_.Unlock();
+      }
      // Current log should always stay alive since it can't have
      // number < MinLogNumber().
      assert(alive_log_files_.size());
@@ -263,7 +269,10 @@ void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
        continue;
      }
      logs_to_free_.push_back(log.ReleaseWriter());
-      logs_.pop_front();
+      {
+        InstrumentedMutexLock wl(&log_write_mutex_);
+        logs_.pop_front();
+      }
    }
    // Current log cannot be obsolete.
    assert(!logs_.empty());
@@ -444,6 +453,7 @@ void DBImpl::PurgeObsoleteFiles(const JobContext& state, bool schedule_only) {
      case kIdentityFile:
      case kMetaDatabase:
      case kOptionsFile:
+      case kBlobFile:
        keep = true;
        break;
    }
@@ -1,7 +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 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
@@ -16,6 +16,8 @@
 #include "db/builder.h"
 #include "options/options_helper.h"
 #include "rocksdb/wal_filter.h"
+#include "table/block_based_table_factory.h"
+#include "util/rate_limiter.h"
 #include "util/sst_file_manager_impl.h"
 #include "util/sync_point.h"

@@ -36,7 +38,7 @@ DBOptions SanitizeOptions(const std::string& dbname, const DBOptions& src) {
  if (result.max_open_files != -1) {
    int max_max_open_files = port::GetMaxOpenFiles();
    if (max_max_open_files == -1) {
-      max_max_open_files = 1000000;
+      max_max_open_files = 0x400000;
    }
    ClipToRange(&result.max_open_files, 20, max_max_open_files);
  }
@@ -53,15 +55,13 @@ DBOptions SanitizeOptions(const std::string& dbname, const DBOptions& src) {
    result.write_buffer_manager.reset(
        new WriteBufferManager(result.db_write_buffer_size));
  }
-  if (result.base_background_compactions == -1) {
-    result.base_background_compactions = result.max_background_compactions;
-  }
-  if (result.base_background_compactions > result.max_background_compactions) {
-    result.base_background_compactions = result.max_background_compactions;
-  }
-  result.env->IncBackgroundThreadsIfNeeded(src.max_background_compactions,
+  auto bg_job_limits = DBImpl::GetBGJobLimits(result.max_background_flushes,
+                                              result.max_background_compactions,
+                                              result.max_background_jobs,
+                                              true /* parallelize_compactions */);
+  result.env->IncBackgroundThreadsIfNeeded(bg_job_limits.max_compactions,
                                           Env::Priority::LOW);
-  result.env->IncBackgroundThreadsIfNeeded(src.max_background_flushes,
+  result.env->IncBackgroundThreadsIfNeeded(bg_job_limits.max_flushes,
                                           Env::Priority::HIGH);

  if (result.rate_limiter.get() != nullptr) {
@@ -70,6 +70,15 @@ DBOptions SanitizeOptions(const std::string& dbname, const DBOptions& src) {
    }
  }

+  if (result.delayed_write_rate == 0) {
+    if (result.rate_limiter.get() != nullptr) {
+      result.delayed_write_rate = result.rate_limiter->GetBytesPerSecond();
+    }
+    if (result.delayed_write_rate == 0) {
+      result.delayed_write_rate = 16 * 1024 * 1024;
+    }
+  }
+
  if (result.WAL_ttl_seconds > 0 || result.WAL_size_limit_MB > 0) {
    result.recycle_log_file_num = false;
  }
@@ -97,11 +106,14 @@ DBOptions SanitizeOptions(const std::string& dbname, const DBOptions& src) {
    result.db_paths.emplace_back(dbname, std::numeric_limits<uint64_t>::max());
  }

-  if (result.use_direct_reads && result.compaction_readahead_size == 0) {
+  if (result.use_direct_io_for_flush_and_compaction &&
+      result.compaction_readahead_size == 0) {
+    TEST_SYNC_POINT_CALLBACK("SanitizeOptions:direct_io", nullptr);
    result.compaction_readahead_size = 1024 * 1024 * 2;
  }

-  if (result.compaction_readahead_size > 0) {
+  if (result.compaction_readahead_size > 0 ||
+      result.use_direct_io_for_flush_and_compaction) {
    result.new_table_reader_for_compaction_inputs = true;
  }

@@ -151,6 +163,19 @@ static Status ValidateOptions(
            "universal and level compaction styles. ");
      }
    }
+    if (cfd.options.compaction_options_fifo.ttl > 0) {
+      if (db_options.max_open_files != -1) {
+        return Status::NotSupported(
+            "FIFO Compaction with TTL is only supported when files are always "
+            "kept open (set max_open_files = -1). ");
+      }
+      if (cfd.options.table_factory->Name() !=
+          BlockBasedTableFactory().Name()) {
+        return Status::NotSupported(
+            "FIFO Compaction with TTL is only supported in "
+            "Block-Based Table format. ");
+      }
+    }
  }

  if (db_options.db_paths.size() > 4) {
@@ -165,10 +190,12 @@ static Status ValidateOptions(
        "then direct I/O reads (use_direct_reads) must be disabled. ");
  }

-  if (db_options.allow_mmap_writes && db_options.use_direct_writes) {
+  if (db_options.allow_mmap_writes &&
+      db_options.use_direct_io_for_flush_and_compaction) {
    return Status::NotSupported(
        "If memory mapped writes (allow_mmap_writes) are enabled "
-        "then direct I/O writes (use_direct_writes) must be disabled. ");
+        "then direct I/O writes (use_direct_io_for_flush_and_compaction) must "
+        "be disabled. ");
  }

  if (db_options.keep_log_file_num == 0) {
@@ -493,7 +520,8 @@ Status DBImpl::RecoverLogFiles(const std::vector<uint64_t>& log_numbers,
    unique_ptr<SequentialFileReader> file_reader;
    {
      unique_ptr<SequentialFile> file;
-      status = env_->NewSequentialFile(fname, &file, env_options_);
+      status = env_->NewSequentialFile(fname, &file,
+                                       env_->OptimizeForLogRead(env_options_));
      if (!status.ok()) {
        MaybeIgnoreError(&status);
        if (!status.ok()) {
@@ -709,6 +737,7 @@ 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_->SetLastSequence(last_sequence);
    }
  }
@@ -780,9 +809,15 @@ 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_) {
+      log_write_mutex_.Lock();
+    }
    for (auto log_number : log_numbers) {
      alive_log_files_.push_back(LogFileNumberSize(log_number));
    }
+    if (concurrent_prepare_) {
+      log_write_mutex_.Unlock();
+    }
  }

  event_logger_.Log() << "job" << job_id << "event"
@@ -816,6 +851,11 @@ Status DBImpl::WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd,
        *cfd->GetLatestMutableCFOptions();
    bool paranoid_file_checks =
        cfd->GetLatestMutableCFOptions()->paranoid_file_checks;
+
+    int64_t _current_time = 0;
+    env_->GetCurrentTime(&_current_time);  // ignore error
+    const uint64_t current_time = static_cast<uint64_t>(_current_time);
+
    {
      mutex_.Unlock();

@@ -823,9 +863,11 @@ Status DBImpl::WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd,
      std::vector<SequenceNumber> snapshot_seqs =
          snapshots_.GetAll(&earliest_write_conflict_snapshot);

+      EnvOptions optimized_env_options =
+          env_->OptimizeForCompactionTableWrite(env_options_, immutable_db_options_);
      s = BuildTable(
-          dbname_, env_, *cfd->ioptions(), mutable_cf_options, env_options_,
-          cfd->table_cache(), iter.get(),
+          dbname_, env_, *cfd->ioptions(), mutable_cf_options,
+          optimized_env_options, 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(),
@@ -833,7 +875,8 @@ Status DBImpl::WriteLevel0TableForRecovery(int job_id, ColumnFamilyData* cfd,
          GetCompressionFlush(*cfd->ioptions(), mutable_cf_options),
          cfd->ioptions()->compression_opts, paranoid_file_checks,
          cfd->internal_stats(), TableFileCreationReason::kRecovery,
-          &event_logger_, job_id);
+          &event_logger_, job_id, Env::IO_HIGH, nullptr /* table_properties */,
+          -1 /* level */, current_time);
      LogFlush(immutable_db_options_.info_log);
      ROCKS_LOG_DEBUG(immutable_db_options_.info_log,
                      "[%s] [WriteLevel0TableForRecovery]"
@@ -882,7 +925,7 @@ Status DB::Open(const Options& options, const std::string& dbname, DB** dbptr) {
  }
  return s;
 }
-  
+
 Status DB::Open(const DBOptions& db_options, const std::string& dbname,
                const std::vector<ColumnFamilyDescriptor>& column_families,
                std::vector<ColumnFamilyHandle*>* handles, DB** dbptr) {
@@ -944,14 +987,17 @@ Status DB::Open(const DBOptions& db_options, const std::string& dbname,
    if (s.ok()) {
      lfile->SetPreallocationBlockSize(
          impl->GetWalPreallocateBlockSize(max_write_buffer_size));
-      impl->logfile_number_ = new_log_number;
-      unique_ptr<WritableFileWriter> file_writer(
-          new WritableFileWriter(std::move(lfile), opt_env_options));
-      impl->logs_.emplace_back(
-          new_log_number,
-          new log::Writer(
-              std::move(file_writer), new_log_number,
-              impl->immutable_db_options_.recycle_log_file_num > 0));
+      {
+        InstrumentedMutexLock wl(&impl->log_write_mutex_);
+        impl->logfile_number_ = new_log_number;
+        unique_ptr<WritableFileWriter> file_writer(
+            new WritableFileWriter(std::move(lfile), opt_env_options));
+        impl->logs_.emplace_back(
+            new_log_number,
+            new log::Writer(
+                std::move(file_writer), new_log_number,
+                impl->immutable_db_options_.recycle_log_file_num > 0));
+      }

      // set column family handles
      for (auto cf : column_families) {
@@ -985,8 +1031,14 @@ Status DB::Open(const DBOptions& db_options, const std::string& dbname,
        delete impl->InstallSuperVersionAndScheduleWork(
            cfd, nullptr, *cfd->GetLatestMutableCFOptions());
      }
+      if (impl->concurrent_prepare_) {
+        impl->log_write_mutex_.Lock();
+      }
      impl->alive_log_files_.push_back(
          DBImpl::LogFileNumberSize(impl->logfile_number_));
+      if (impl->concurrent_prepare_) {
+        impl->log_write_mutex_.Unlock();
+      }
      impl->DeleteObsoleteFiles();
      s = impl->directories_.GetDbDir()->Fsync();
    }
@@ -1025,7 +1077,8 @@ Status DB::Open(const DBOptions& db_options, const std::string& dbname,
  if (s.ok()) {
    // Persist RocksDB Options before scheduling the compaction.
    // The WriteOptionsFile() will release and lock the mutex internally.
-    persist_options_status = impl->WriteOptionsFile();
+    persist_options_status = impl->WriteOptionsFile(
+        false /*need_mutex_lock*/, false /*need_enter_write_thread*/);

    *dbptr = impl;
    impl->opened_successfully_ = true;
@@ -1058,14 +1111,9 @@ Status DB::Open(const DBOptions& db_options, const std::string& dbname,
    ROCKS_LOG_INFO(impl->immutable_db_options_.info_log, "DB pointer %p", impl);
    LogFlush(impl->immutable_db_options_.info_log);
    if (!persist_options_status.ok()) {
-      if (db_options.fail_if_options_file_error) {
-        s = Status::IOError(
-            "DB::Open() failed --- Unable to persist Options file",
-            persist_options_status.ToString());
-      }
-      ROCKS_LOG_WARN(impl->immutable_db_options_.info_log,
-                     "Unable to persist options in DB::Open() -- %s",
-                     persist_options_status.ToString().c_str());
+      s = Status::IOError(
+          "DB::Open() failed --- Unable to persist Options file",
+          persist_options_status.ToString());
    }
  }
  if (!s.ok()) {
@@ -1,8 +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 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_impl_readonly.h"

@@ -1,7 +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 licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).

 #pragma once

@@ -1,7 +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 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
@@ -12,6 +12,7 @@
 #define __STDC_FORMAT_MACROS
 #endif
 #include <inttypes.h>
+#include "db/event_helpers.h"
 #include "monitoring/perf_context_imp.h"
 #include "options/options_helper.h"
 #include "util/sync_point.h"
@@ -63,8 +64,28 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
  if (my_batch == nullptr) {
    return Status::Corruption("Batch is nullptr!");
  }
+  if (concurrent_prepare_ && immutable_db_options_.enable_pipelined_write) {
+    return Status::NotSupported(
+        "pipelined_writes is not compatible with concurrent prepares");
+  }

  Status status;
+  if (write_options.low_pri) {
+    status = ThrottleLowPriWritesIfNeeded(write_options, my_batch);
+    if (!status.ok()) {
+      return status;
+    }
+  }
+
+  if (concurrent_prepare_ && disable_memtable) {
+    return WriteImplWALOnly(write_options, my_batch, callback, log_used,
+                            log_ref);
+  }
+
+  if (immutable_db_options_.enable_pipelined_write) {
+    return PipelinedWriteImpl(write_options, my_batch, callback, log_used,
+                              log_ref, disable_memtable);
+  }

  PERF_TIMER_GUARD(write_pre_and_post_process_time);
  WriteThread::Writer w(write_options, my_batch, callback, log_ref,
@@ -77,25 +98,25 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
  StopWatch write_sw(env_, immutable_db_options_.statistics.get(), DB_WRITE);

  write_thread_.JoinBatchGroup(&w);
-  if (w.state == WriteThread::STATE_PARALLEL_FOLLOWER) {
+  if (w.state == WriteThread::STATE_PARALLEL_MEMTABLE_WRITER) {
    // we are a non-leader in a parallel group
    PERF_TIMER_GUARD(write_memtable_time);

    if (w.ShouldWriteToMemtable()) {
      ColumnFamilyMemTablesImpl column_family_memtables(
          versions_->GetColumnFamilySet());
-      WriteBatchInternal::SetSequence(w.batch, w.sequence);
      w.status = WriteBatchInternal::InsertInto(
-          &w, &column_family_memtables, &flush_scheduler_,
+          &w, w.sequence, &column_family_memtables, &flush_scheduler_,
          write_options.ignore_missing_column_families, 0 /*log_number*/, this,
          true /*concurrent_memtable_writes*/);
    }

-    if (write_thread_.CompleteParallelWorker(&w)) {
-      // we're responsible for early exit
-      auto last_sequence = w.parallel_group->last_sequence;
+    if (write_thread_.CompleteParallelMemTableWriter(&w)) {
+      // we're responsible for exit batch group
+      auto last_sequence = w.write_group->last_sequence;
      versions_->SetLastSequence(last_sequence);
-      write_thread_.EarlyExitParallelGroup(&w);
+      MemTableInsertStatusCheck(w.status);
+      write_thread_.ExitAsBatchGroupFollower(&w);
    }
    assert(w.state == WriteThread::STATE_COMPLETED);
    // STATE_COMPLETED conditional below handles exit
@@ -117,17 +138,24 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
  // when it finds suitable, and finish them in the same write batch.
  // This is how a write job could be done by the other writer.
  WriteContext write_context;
-  WriteThread::Writer* last_writer = &w;  // Dummy intial value
-  autovector<WriteThread::Writer*> write_group;
-  bool logs_getting_synced = false;
+  WriteThread::WriteGroup write_group;
+  bool in_parallel_group = false;
+  uint64_t last_sequence = kMaxSequenceNumber;
+  if (!concurrent_prepare_) {
+    last_sequence = versions_->LastSequence();
+  }

  mutex_.Lock();

  bool need_log_sync = !write_options.disableWAL && write_options.sync;
  bool need_log_dir_sync = need_log_sync && !log_dir_synced_;
-  status = PreprocessWrite(write_options, need_log_sync, &logs_getting_synced,
-                           &write_context);
-  log::Writer* cur_log_writer = logs_.back().writer;
+  if (!concurrent_prepare_ || !disable_memtable) {
+    // With concurrent writes we do preprocess only in the write thread that
+    // also does write to memtable to avoid sync issue on shared data structure
+    // with the other thread
+    status = PreprocessWrite(write_options, &need_log_sync, &write_context);
+  }
+  log::Writer* log_writer = logs_.back().writer;

  mutex_.Unlock();

@@ -136,9 +164,8 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
  // and protects against concurrent loggers and concurrent writes
  // into memtables

-  bool exit_completed_early = false;
  last_batch_group_size_ =
-      write_thread_.EnterAsBatchGroupLeader(&w, &last_writer, &write_group);
+      write_thread_.EnterAsBatchGroupLeader(&w, &write_group);

  if (status.ok()) {
    // Rules for when we can update the memtable concurrently
@@ -153,43 +180,39 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
    // relax rules 2 if we could prevent write batches from referring
    // more than once to a particular key.
    bool parallel = immutable_db_options_.allow_concurrent_memtable_write &&
-                    write_group.size() > 1;
+                    write_group.size > 1;
    int total_count = 0;
    uint64_t total_byte_size = 0;
-    for (auto writer : write_group) {
+    for (auto* writer : write_group) {
      if (writer->CheckCallback(this)) {
        if (writer->ShouldWriteToMemtable()) {
          total_count += WriteBatchInternal::Count(writer->batch);
          parallel = parallel && !writer->batch->HasMerge();
        }

-        if (writer->ShouldWriteToWAL()) {
-          total_byte_size = WriteBatchInternal::AppendedByteSize(
-              total_byte_size, WriteBatchInternal::ByteSize(writer->batch));
-        }
+        total_byte_size = WriteBatchInternal::AppendedByteSize(
+            total_byte_size, WriteBatchInternal::ByteSize(writer->batch));
      }
    }

-    uint64_t last_sequence = versions_->LastSequence();
-    const SequenceNumber current_sequence = last_sequence + 1;
-    last_sequence += total_count;
-
+    const bool concurrent_update = concurrent_prepare_;
    // Update stats while we are an exclusive group leader, so we know
    // that nobody else can be writing to these particular stats.
    // We're optimistic, updating the stats before we successfully
-    // commit.  That lets us release our leader status early in
-    // some cases.
+    // commit.  That lets us release our leader status early.
    auto stats = default_cf_internal_stats_;
-    stats->AddDBStats(InternalStats::NUMBER_KEYS_WRITTEN, total_count);
+    stats->AddDBStats(InternalStats::NUMBER_KEYS_WRITTEN, total_count,
+                      concurrent_update);
    RecordTick(stats_, NUMBER_KEYS_WRITTEN, total_count);
-    stats->AddDBStats(InternalStats::BYTES_WRITTEN, total_byte_size);
+    stats->AddDBStats(InternalStats::BYTES_WRITTEN, total_byte_size,
+                      concurrent_update);
    RecordTick(stats_, BYTES_WRITTEN, total_byte_size);
-    stats->AddDBStats(InternalStats::WRITE_DONE_BY_SELF, 1);
+    stats->AddDBStats(InternalStats::WRITE_DONE_BY_SELF, 1, concurrent_update);
    RecordTick(stats_, WRITE_DONE_BY_SELF);
-    auto write_done_by_other = write_group.size() - 1;
+    auto write_done_by_other = write_group.size - 1;
    if (write_done_by_other > 0) {
-      stats->AddDBStats(InternalStats::WRITE_DONE_BY_OTHER,
-                        write_done_by_other);
+      stats->AddDBStats(InternalStats::WRITE_DONE_BY_OTHER, write_done_by_other,
+                        concurrent_update);
      RecordTick(stats_, WRITE_DONE_BY_OTHER, write_done_by_other);
    }
    MeasureTime(stats_, BYTES_PER_WRITE, total_byte_size);
@@ -200,39 +223,49 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,

    PERF_TIMER_STOP(write_pre_and_post_process_time);

-    if (status.ok() && !write_options.disableWAL) {
-      PERF_TIMER_GUARD(write_wal_time);
-      status = WriteToWAL(write_group, cur_log_writer, need_log_sync,
-                          need_log_dir_sync, current_sequence);
-      if (log_used != nullptr) {
-        *log_used = logfile_number_;
+    if (!concurrent_prepare_) {
+      if (status.ok() && !write_options.disableWAL) {
+        PERF_TIMER_GUARD(write_wal_time);
+        status = WriteToWAL(write_group, log_writer, log_used, need_log_sync,
+                            need_log_dir_sync, last_sequence + 1);
+      }
+    } else {
+      if (status.ok() && !write_options.disableWAL) {
+        PERF_TIMER_GUARD(write_wal_time);
+        // LastToBeWrittenSequence is increased inside WriteToWAL under
+        // wal_write_mutex_ to ensure ordered events in WAL
+        status = ConcurrentWriteToWAL(write_group, log_used, &last_sequence,
+                                      total_count);
+      } else {
+        // Otherwise we inc seq number for memtable writes
+        last_sequence = versions_->FetchAddLastToBeWrittenSequence(total_count);
      }
    }
+    assert(last_sequence != kMaxSequenceNumber);
+    const SequenceNumber current_sequence = last_sequence + 1;
+    last_sequence += total_count;

    if (status.ok()) {
      PERF_TIMER_GUARD(write_memtable_time);

      if (!parallel) {
-        status = WriteBatchInternal::InsertInto(
+        w.status = WriteBatchInternal::InsertInto(
            write_group, current_sequence, column_family_memtables_.get(),
            &flush_scheduler_, write_options.ignore_missing_column_families,
            0 /*recovery_log_number*/, this);
-
-        if (status.ok()) {
-          // There were no write failures. Set leader's status
-          // in case the write callback returned a non-ok status.
-          status = w.FinalStatus();
-        }
-
      } else {
-        WriteThread::ParallelGroup pg;
-        pg.leader = &w;
-        pg.last_writer = last_writer;
-        pg.last_sequence = last_sequence;
-        pg.early_exit_allowed = !need_log_sync;
-        pg.running.store(static_cast<uint32_t>(write_group.size()),
-                         std::memory_order_relaxed);
-        write_thread_.LaunchParallelFollowers(&pg, current_sequence);
+        SequenceNumber next_sequence = current_sequence;
+        for (auto* writer : write_group) {
+          if (writer->ShouldWriteToMemtable()) {
+            writer->sequence = next_sequence;
+            next_sequence += WriteBatchInternal::Count(writer->batch);
+          }
+        }
+        write_group.last_sequence = last_sequence;
+        write_group.running.store(static_cast<uint32_t>(write_group.size),
+                                  std::memory_order_relaxed);
+        write_thread_.LaunchParallelMemTableWriters(&write_group);
+        in_parallel_group = true;

        // Each parallel follower is doing each own writes. The leader should
        // also do its own.
@@ -240,73 +273,309 @@ Status DBImpl::WriteImpl(const WriteOptions& write_options,
          ColumnFamilyMemTablesImpl column_family_memtables(
              versions_->GetColumnFamilySet());
          assert(w.sequence == current_sequence);
-          WriteBatchInternal::SetSequence(w.batch, w.sequence);
          w.status = WriteBatchInternal::InsertInto(
-              &w, &column_family_memtables, &flush_scheduler_,
+              &w, w.sequence, &column_family_memtables, &flush_scheduler_,
              write_options.ignore_missing_column_families, 0 /*log_number*/,
              this, true /*concurrent_memtable_writes*/);
        }
-
-        // CompleteParallelWorker returns true if this thread should
-        // handle exit, false means somebody else did
-        exit_completed_early = !write_thread_.CompleteParallelWorker(&w);
-        status = w.FinalStatus();
-      }
-
-      if (!exit_completed_early && w.status.ok()) {
-        versions_->SetLastSequence(last_sequence);
-        if (!need_log_sync) {
-          write_thread_.ExitAsBatchGroupLeader(&w, last_writer, w.status);
-          exit_completed_early = true;
-        }
-      }
-
-      // A non-OK status here indicates that the state implied by the
-      // WAL has diverged from the in-memory state.  This could be
-      // because of a corrupt write_batch (very bad), or because the
-      // client specified an invalid column family and didn't specify
-      // ignore_missing_column_families.
-      //
-      // Is setting bg_error_ enough here?  This will at least stop
-      // compaction and fail any further writes.
-      if (!status.ok() && !w.CallbackFailed()) {
-        mutex_.Lock();
-        if (bg_error_.ok()) {
-          bg_error_ = status;  // stop compaction & fail any further writes
-        }
-        mutex_.Unlock();
      }
    }
  }
  PERF_TIMER_START(write_pre_and_post_process_time);

-  if (immutable_db_options_.paranoid_checks && !status.ok() &&
-      !w.CallbackFailed() && !status.IsBusy() && !status.IsIncomplete()) {
-    mutex_.Lock();
-    if (bg_error_.ok()) {
-      bg_error_ = status;  // stop compaction & fail any further writes
-    }
-    mutex_.Unlock();
+  if (!w.CallbackFailed()) {
+    WriteCallbackStatusCheck(status);
  }

-  if (logs_getting_synced) {
+  if (need_log_sync) {
    mutex_.Lock();
    MarkLogsSynced(logfile_number_, need_log_dir_sync, status);
    mutex_.Unlock();
+    // Requesting sync with concurrent_prepare_ is expected to be very rare. We
+    // hance provide a simple implementation that is not necessarily efficient.
+    if (concurrent_prepare_) {
+      if (manual_wal_flush_) {
+        status = FlushWAL(true);
+      } else {
+        status = SyncWAL();
+      }
+    }
  }

-  if (!exit_completed_early) {
-    write_thread_.ExitAsBatchGroupLeader(&w, last_writer, w.status);
+  bool should_exit_batch_group = true;
+  if (in_parallel_group) {
+    // CompleteParallelWorker returns true if this thread should
+    // handle exit, false means somebody else did
+    should_exit_batch_group = write_thread_.CompleteParallelMemTableWriter(&w);
+  }
+  if (should_exit_batch_group) {
+    if (status.ok()) {
+      versions_->SetLastSequence(last_sequence);
+    }
+    MemTableInsertStatusCheck(w.status);
+    write_thread_.ExitAsBatchGroupLeader(write_group, status);
  }

+  if (status.ok()) {
+    status = w.FinalStatus();
+  }
  return status;
 }

+Status DBImpl::PipelinedWriteImpl(const WriteOptions& write_options,
+                                  WriteBatch* my_batch, WriteCallback* callback,
+                                  uint64_t* log_used, uint64_t log_ref,
+                                  bool disable_memtable) {
+  PERF_TIMER_GUARD(write_pre_and_post_process_time);
+  StopWatch write_sw(env_, immutable_db_options_.statistics.get(), DB_WRITE);
+
+  WriteContext write_context;
+
+  WriteThread::Writer w(write_options, my_batch, callback, log_ref,
+                        disable_memtable);
+  write_thread_.JoinBatchGroup(&w);
+  if (w.state == WriteThread::STATE_GROUP_LEADER) {
+    WriteThread::WriteGroup wal_write_group;
+    if (w.callback && !w.callback->AllowWriteBatching()) {
+      write_thread_.WaitForMemTableWriters();
+    }
+    mutex_.Lock();
+    bool need_log_sync = !write_options.disableWAL && write_options.sync;
+    bool need_log_dir_sync = need_log_sync && !log_dir_synced_;
+    w.status = PreprocessWrite(write_options, &need_log_sync, &write_context);
+    log::Writer* log_writer = logs_.back().writer;
+    mutex_.Unlock();
+
+    // This can set non-OK status if callback fail.
+    last_batch_group_size_ =
+        write_thread_.EnterAsBatchGroupLeader(&w, &wal_write_group);
+    const SequenceNumber current_sequence =
+        write_thread_.UpdateLastSequence(versions_->LastSequence()) + 1;
+    size_t total_count = 0;
+    size_t total_byte_size = 0;
+
+    if (w.status.ok()) {
+      SequenceNumber next_sequence = current_sequence;
+      for (auto writer : wal_write_group) {
+        if (writer->CheckCallback(this)) {
+          if (writer->ShouldWriteToMemtable()) {
+            writer->sequence = next_sequence;
+            size_t count = WriteBatchInternal::Count(writer->batch);
+            next_sequence += count;
+            total_count += count;
+          }
+          total_byte_size = WriteBatchInternal::AppendedByteSize(
+              total_byte_size, WriteBatchInternal::ByteSize(writer->batch));
+        }
+      }
+      if (w.disable_wal) {
+        has_unpersisted_data_.store(true, std::memory_order_relaxed);
+      }
+      write_thread_.UpdateLastSequence(current_sequence + total_count - 1);
+    }
+
+    auto stats = default_cf_internal_stats_;
+    stats->AddDBStats(InternalStats::NUMBER_KEYS_WRITTEN, total_count);
+    RecordTick(stats_, NUMBER_KEYS_WRITTEN, total_count);
+    stats->AddDBStats(InternalStats::BYTES_WRITTEN, total_byte_size);
+    RecordTick(stats_, BYTES_WRITTEN, total_byte_size);
+
+    PERF_TIMER_STOP(write_pre_and_post_process_time);
+
+    if (w.ShouldWriteToWAL()) {
+      PERF_TIMER_GUARD(write_wal_time);
+      stats->AddDBStats(InternalStats::WRITE_DONE_BY_SELF, 1);
+      RecordTick(stats_, WRITE_DONE_BY_SELF, 1);
+      if (wal_write_group.size > 1) {
+        stats->AddDBStats(InternalStats::WRITE_DONE_BY_OTHER,
+                          wal_write_group.size - 1);
+        RecordTick(stats_, WRITE_DONE_BY_OTHER, wal_write_group.size - 1);
+      }
+      w.status = WriteToWAL(wal_write_group, log_writer, log_used,
+                            need_log_sync, need_log_dir_sync, current_sequence);
+    }
+
+    if (!w.CallbackFailed()) {
+      WriteCallbackStatusCheck(w.status);
+    }
+
+    if (need_log_sync) {
+      mutex_.Lock();
+      MarkLogsSynced(logfile_number_, need_log_dir_sync, w.status);
+      mutex_.Unlock();
+    }
+
+    write_thread_.ExitAsBatchGroupLeader(wal_write_group, w.status);
+  }
+
+  WriteThread::WriteGroup memtable_write_group;
+  if (w.state == WriteThread::STATE_MEMTABLE_WRITER_LEADER) {
+    PERF_TIMER_GUARD(write_memtable_time);
+    assert(w.status.ok());
+    write_thread_.EnterAsMemTableWriter(&w, &memtable_write_group);
+    if (memtable_write_group.size > 1 &&
+        immutable_db_options_.allow_concurrent_memtable_write) {
+      write_thread_.LaunchParallelMemTableWriters(&memtable_write_group);
+    } else {
+      memtable_write_group.status = WriteBatchInternal::InsertInto(
+          memtable_write_group, w.sequence, column_family_memtables_.get(),
+          &flush_scheduler_, write_options.ignore_missing_column_families,
+          0 /*log_number*/, this);
+      versions_->SetLastSequence(memtable_write_group.last_sequence);
+      write_thread_.ExitAsMemTableWriter(&w, memtable_write_group);
+    }
+  }
+
+  if (w.state == WriteThread::STATE_PARALLEL_MEMTABLE_WRITER) {
+    assert(w.ShouldWriteToMemtable());
+    ColumnFamilyMemTablesImpl column_family_memtables(
+        versions_->GetColumnFamilySet());
+    w.status = WriteBatchInternal::InsertInto(
+        &w, w.sequence, &column_family_memtables, &flush_scheduler_,
+        write_options.ignore_missing_column_families, 0 /*log_number*/, this,
+        true /*concurrent_memtable_writes*/);
+    if (write_thread_.CompleteParallelMemTableWriter(&w)) {
+      MemTableInsertStatusCheck(w.status);
+      versions_->SetLastSequence(w.write_group->last_sequence);
+      write_thread_.ExitAsMemTableWriter(&w, *w.write_group);
+    }
+  }
+
+  assert(w.state == WriteThread::STATE_COMPLETED);
+  return w.FinalStatus();
+}
+
+Status DBImpl::WriteImplWALOnly(const WriteOptions& write_options,
+                                WriteBatch* my_batch, WriteCallback* callback,
+                                uint64_t* log_used, uint64_t log_ref) {
+  Status status;
+  PERF_TIMER_GUARD(write_pre_and_post_process_time);
+  WriteThread::Writer w(write_options, my_batch, callback, log_ref,
+                        true /* disable_memtable */);
+  if (write_options.disableWAL) {
+    return status;
+  }
+  RecordTick(stats_, WRITE_WITH_WAL);
+  StopWatch write_sw(env_, immutable_db_options_.statistics.get(), DB_WRITE);
+
+  nonmem_write_thread_.JoinBatchGroup(&w);
+  assert(w.state != WriteThread::STATE_PARALLEL_MEMTABLE_WRITER);
+  if (w.state == WriteThread::STATE_COMPLETED) {
+    if (log_used != nullptr) {
+      *log_used = w.log_used;
+    }
+    return w.FinalStatus();
+  }
+  // else we are the leader of the write batch group
+  assert(w.state == WriteThread::STATE_GROUP_LEADER);
+  WriteContext write_context;
+  WriteThread::WriteGroup write_group;
+  uint64_t last_sequence;
+  nonmem_write_thread_.EnterAsBatchGroupLeader(&w, &write_group);
+  // Note: no need to update last_batch_group_size_ here since the batch writes
+  // to WAL only
+
+  uint64_t total_byte_size = 0;
+  for (auto* writer : write_group) {
+    if (writer->CheckCallback(this)) {
+      total_byte_size = WriteBatchInternal::AppendedByteSize(
+          total_byte_size, WriteBatchInternal::ByteSize(writer->batch));
+    }
+  }
+
+  const bool concurrent_update = true;
+  // Update stats while we are an exclusive group leader, so we know
+  // that nobody else can be writing to these particular stats.
+  // We're optimistic, updating the stats before we successfully
+  // commit.  That lets us release our leader status early.
+  auto stats = default_cf_internal_stats_;
+  stats->AddDBStats(InternalStats::BYTES_WRITTEN, total_byte_size,
+                    concurrent_update);
+  RecordTick(stats_, BYTES_WRITTEN, total_byte_size);
+  stats->AddDBStats(InternalStats::WRITE_DONE_BY_SELF, 1, concurrent_update);
+  RecordTick(stats_, WRITE_DONE_BY_SELF);
+  auto write_done_by_other = write_group.size - 1;
+  if (write_done_by_other > 0) {
+    stats->AddDBStats(InternalStats::WRITE_DONE_BY_OTHER, write_done_by_other,
+                      concurrent_update);
+    RecordTick(stats_, WRITE_DONE_BY_OTHER, write_done_by_other);
+  }
+  MeasureTime(stats_, BYTES_PER_WRITE, total_byte_size);
+
+  PERF_TIMER_STOP(write_pre_and_post_process_time);
+
+  PERF_TIMER_GUARD(write_wal_time);
+  // LastToBeWrittenSequence is increased inside WriteToWAL under
+  // wal_write_mutex_ to ensure ordered events in WAL
+  status = ConcurrentWriteToWAL(write_group, log_used, &last_sequence,
+                                0 /*total_count*/);
+  if (status.ok() && write_options.sync) {
+    // Requesting sync with concurrent_prepare_ is expected to be very rare. We
+    // hance provide a simple implementation that is not necessarily efficient.
+    if (manual_wal_flush_) {
+      status = FlushWAL(true);
+    } else {
+      status = SyncWAL();
+    }
+  }
+  PERF_TIMER_START(write_pre_and_post_process_time);
+
+  if (!w.CallbackFailed()) {
+    WriteCallbackStatusCheck(status);
+  }
+  nonmem_write_thread_.ExitAsBatchGroupLeader(write_group, status);
+  if (status.ok()) {
+    status = w.FinalStatus();
+  }
+  return status;
+}
+
+void DBImpl::WriteCallbackStatusCheck(const Status& status) {
+  // Is setting bg_error_ enough here?  This will at least stop
+  // compaction and fail any further writes.
+  if (immutable_db_options_.paranoid_checks && !status.ok() &&
+      !status.IsBusy() && !status.IsIncomplete()) {
+    mutex_.Lock();
+    if (bg_error_.ok()) {
+      Status new_bg_error = status;
+      // may temporarily unlock and lock the mutex.
+      EventHelpers::NotifyOnBackgroundError(immutable_db_options_.listeners,
+                                            BackgroundErrorReason::kWriteCallback,
+                                            &new_bg_error, &mutex_);
+      if (!new_bg_error.ok()) {
+        bg_error_ = new_bg_error;  // stop compaction & fail any further writes
+      }
+    }
+    mutex_.Unlock();
+  }
+}
+
+void DBImpl::MemTableInsertStatusCheck(const Status& status) {
+  // A non-OK status here indicates that the state implied by the
+  // WAL has diverged from the in-memory state.  This could be
+  // because of a corrupt write_batch (very bad), or because the
+  // client specified an invalid column family and didn't specify
+  // ignore_missing_column_families.
+  if (!status.ok()) {
+    mutex_.Lock();
+    assert(bg_error_.ok());
+    Status new_bg_error = status;
+    // may temporarily unlock and lock the mutex.
+    EventHelpers::NotifyOnBackgroundError(immutable_db_options_.listeners,
+                                          BackgroundErrorReason::kMemTable,
+                                          &new_bg_error, &mutex_);
+    if (!new_bg_error.ok()) {
+      bg_error_ = new_bg_error;  // stop compaction & fail any further writes
+    }
+    mutex_.Unlock();
+  }
+}
+
 Status DBImpl::PreprocessWrite(const WriteOptions& write_options,
-                               bool need_log_sync, bool* logs_getting_synced,
+                               bool* need_log_sync,
                               WriteContext* write_context) {
  mutex_.AssertHeld();
-  assert(write_context != nullptr && logs_getting_synced != nullptr);
+  assert(write_context != nullptr && need_log_sync != nullptr);
  Status status;

  assert(!single_column_family_mode_ ||
@@ -326,7 +595,7 @@ Status DBImpl::PreprocessWrite(const WriteOptions& write_options,
  }

  if (UNLIKELY(status.ok() && !bg_error_.ok())) {
-    status = bg_error_;
+    return bg_error_;
  }

  if (UNLIKELY(status.ok() && !flush_scheduler_.Empty())) {
@@ -343,7 +612,7 @@ Status DBImpl::PreprocessWrite(const WriteOptions& write_options,
    status = DelayWrite(last_batch_group_size_, write_options);
  }

-  if (status.ok() && need_log_sync) {
+  if (status.ok() && *need_log_sync) {
    // Wait until the parallel syncs are finished. Any sync process has to sync
    // the front log too so it is enough to check the status of front()
    // We do a while loop since log_sync_cv_ is signalled when any sync is
@@ -363,50 +632,84 @@ Status DBImpl::PreprocessWrite(const WriteOptions& write_options,
      // actually write to the WAL
      log.getting_synced = true;
    }
-    *logs_getting_synced = true;
+  } else {
+    *need_log_sync = false;
  }

  return status;
 }

-Status DBImpl::WriteToWAL(const autovector<WriteThread::Writer*>& write_group,
-                          log::Writer* log_writer, bool need_log_sync,
-                          bool need_log_dir_sync, SequenceNumber sequence) {
-  Status status;
-
+WriteBatch* DBImpl::MergeBatch(const WriteThread::WriteGroup& write_group,
+                               WriteBatch* tmp_batch, size_t* write_with_wal) {
+  assert(write_with_wal != nullptr);
+  assert(tmp_batch != nullptr);
  WriteBatch* merged_batch = nullptr;
-  size_t write_with_wal = 0;
-  if (write_group.size() == 1 && write_group[0]->ShouldWriteToWAL() &&
-      write_group[0]->batch->GetWalTerminationPoint().is_cleared()) {
+  *write_with_wal = 0;
+  auto* leader = write_group.leader;
+  if (write_group.size == 1 && leader->ShouldWriteToWAL() &&
+      leader->batch->GetWalTerminationPoint().is_cleared()) {
    // we simply write the first WriteBatch to WAL if the group only
    // contains one batch, that batch should be written to the WAL,
    // and the batch is not wanting to be truncated
-    merged_batch = write_group[0]->batch;
-    write_group[0]->log_used = logfile_number_;
-    write_with_wal = 1;
+    merged_batch = leader->batch;
+    *write_with_wal = 1;
  } else {
    // WAL needs all of the batches flattened into a single batch.
    // We could avoid copying here with an iov-like AddRecord
    // interface
-    merged_batch = &tmp_batch_;
+    merged_batch = tmp_batch;
    for (auto writer : write_group) {
      if (writer->ShouldWriteToWAL()) {
        WriteBatchInternal::Append(merged_batch, writer->batch,
                                   /*WAL_only*/ true);
-        write_with_wal++;
+        (*write_with_wal)++;
      }
+    }
+  }
+  return merged_batch;
+}
+
+// When concurrent_prepare_ is disabled, this function is called from the only
+// write thread. Otherwise this must be called holding log_write_mutex_.
+Status DBImpl::WriteToWAL(const WriteBatch& merged_batch,
+                          log::Writer* log_writer, uint64_t* log_used,
+                          uint64_t* log_size) {
+  assert(log_size != nullptr);
+  Slice log_entry = WriteBatchInternal::Contents(&merged_batch);
+  *log_size = log_entry.size();
+  Status status = log_writer->AddRecord(log_entry);
+  if (log_used != nullptr) {
+    *log_used = logfile_number_;
+  }
+  total_log_size_ += log_entry.size();
+  // TODO(myabandeh): it might be unsafe to access alive_log_files_.back() here
+  // since alive_log_files_ might be modified concurrently
+  alive_log_files_.back().AddSize(log_entry.size());
+  log_empty_ = false;
+  return status;
+}
+
+Status DBImpl::WriteToWAL(const WriteThread::WriteGroup& write_group,
+                          log::Writer* log_writer, uint64_t* log_used,
+                          bool need_log_sync, bool need_log_dir_sync,
+                          SequenceNumber sequence) {
+  Status status;
+
+  size_t write_with_wal = 0;
+  WriteBatch* merged_batch =
+      MergeBatch(write_group, &tmp_batch_, &write_with_wal);
+  if (merged_batch == write_group.leader->batch) {
+    write_group.leader->log_used = logfile_number_;
+  } else if (write_with_wal > 1) {
+    for (auto writer : write_group) {
      writer->log_used = logfile_number_;
    }
  }

  WriteBatchInternal::SetSequence(merged_batch, sequence);

-  Slice log_entry = WriteBatchInternal::Contents(merged_batch);
-  status = log_writer->AddRecord(log_entry);
-  total_log_size_ += log_entry.size();
-  alive_log_files_.back().AddSize(log_entry.size());
-  log_empty_ = false;
-  uint64_t log_size = log_entry.size();
+  uint64_t log_size;
+  status = WriteToWAL(*merged_batch, log_writer, log_used, &log_size);

  if (status.ok() && need_log_sync) {
    StopWatch sw(env_, stats_, WAL_FILE_SYNC_MICROS);
@@ -448,6 +751,48 @@ Status DBImpl::WriteToWAL(const autovector<WriteThread::Writer*>& write_group,
  return status;
 }

+Status DBImpl::ConcurrentWriteToWAL(const WriteThread::WriteGroup& write_group,
+                                    uint64_t* log_used,
+                                    SequenceNumber* last_sequence,
+                                    int total_count) {
+  Status status;
+
+  WriteBatch tmp_batch;
+  size_t write_with_wal = 0;
+  WriteBatch* merged_batch =
+      MergeBatch(write_group, &tmp_batch, &write_with_wal);
+
+  // We need to lock log_write_mutex_ since logs_ and alive_log_files might be
+  // pushed back concurrently
+  log_write_mutex_.Lock();
+  if (merged_batch == write_group.leader->batch) {
+    write_group.leader->log_used = logfile_number_;
+  } else if (write_with_wal > 1) {
+    for (auto writer : write_group) {
+      writer->log_used = logfile_number_;
+    }
+  }
+  *last_sequence = versions_->FetchAddLastToBeWrittenSequence(total_count);
+  auto sequence = *last_sequence + 1;
+  WriteBatchInternal::SetSequence(merged_batch, sequence);
+
+  log::Writer* log_writer = logs_.back().writer;
+  uint64_t log_size;
+  status = WriteToWAL(*merged_batch, log_writer, log_used, &log_size);
+  log_write_mutex_.Unlock();
+
+  if (status.ok()) {
+    const bool concurrent = true;
+    auto stats = default_cf_internal_stats_;
+    stats->AddDBStats(InternalStats::WAL_FILE_BYTES, log_size, concurrent);
+    RecordTick(stats_, WAL_FILE_BYTES, log_size);
+    stats->AddDBStats(InternalStats::WRITE_WITH_WAL, write_with_wal,
+                      concurrent);
+    RecordTick(stats_, WRITE_WITH_WAL, write_with_wal);
+  }
+  return status;
+}
+
 Status DBImpl::HandleWALFull(WriteContext* write_context) {
  mutex_.AssertHeld();
  assert(write_context != nullptr);
@@ -612,6 +957,35 @@ Status DBImpl::DelayWrite(uint64_t num_bytes,
  return bg_error_;
 }

+Status DBImpl::ThrottleLowPriWritesIfNeeded(const WriteOptions& write_options,
+                                            WriteBatch* my_batch) {
+  assert(write_options.low_pri);
+  // This is called outside the DB mutex. Although it is safe to make the call,
+  // the consistency condition is not guaranteed to hold. It's OK to live with
+  // it in this case.
+  // If we need to speed compaction, it means the compaction is left behind
+  // and we start to limit low pri writes to a limit.
+  if (write_controller_.NeedSpeedupCompaction()) {
+    if (allow_2pc() && (my_batch->HasCommit() || my_batch->HasRollback())) {
+      // For 2PC, we only rate limit prepare, not commit.
+      return Status::OK();
+    }
+    if (write_options.no_slowdown) {
+      return Status::Incomplete();
+    } else {
+      assert(my_batch != nullptr);
+      // Rate limit those writes. The reason that we don't completely wait
+      // is that in case the write is heavy, low pri writes may never have
+      // a chance to run. Now we guarantee we are still slowly making
+      // progress.
+      write_controller_.low_pri_rate_limiter()->Request(
+          my_batch->GetDataSize(), Env::IO_HIGH, nullptr /* stats */,
+          RateLimiter::OpType::kWrite);
+    }
+  }
+  return Status::OK();
+}
+
 Status DBImpl::ScheduleFlushes(WriteContext* context) {
  ColumnFamilyData* cfd;
  while ((cfd = flush_scheduler_.TakeNextColumnFamily()) != nullptr) {
@@ -646,14 +1020,33 @@ void DBImpl::NotifyOnMemTableSealed(ColumnFamilyData* cfd,
 // REQUIRES: this thread is currently at the front of the writer queue
 Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
  mutex_.AssertHeld();
+  WriteThread::Writer nonmem_w;
+  if (concurrent_prepare_) {
+    // SwitchMemtable is a rare event. To simply the reasoning, we make sure
+    // that there is no concurrent thread writing to WAL.
+    nonmem_write_thread_.EnterUnbatched(&nonmem_w, &mutex_);
+  }
+
  unique_ptr<WritableFile> lfile;
  log::Writer* new_log = nullptr;
  MemTable* new_mem = nullptr;

+  // In case of pipelined write is enabled, wait for all pending memtable
+  // writers.
+  if (immutable_db_options_.enable_pipelined_write) {
+    write_thread_.WaitForMemTableWriters();
+  }
+
  // Attempt to switch to a new memtable and trigger flush of old.
  // Do this without holding the dbmutex lock.
  assert(versions_->prev_log_number() == 0);
+  if (concurrent_prepare_) {
+    log_write_mutex_.Lock();
+  }
  bool creating_new_log = !log_empty_;
+  if (concurrent_prepare_) {
+    log_write_mutex_.Unlock();
+  }
  uint64_t recycle_log_number = 0;
  if (creating_new_log && immutable_db_options_.recycle_log_file_num &&
      !log_recycle_files.empty()) {
@@ -709,9 +1102,9 @@ Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
        lfile->SetPreallocationBlockSize(preallocate_block_size);
        unique_ptr<WritableFileWriter> file_writer(
            new WritableFileWriter(std::move(lfile), opt_env_opt));
-        new_log =
-            new log::Writer(std::move(file_writer), new_log_number,
-                            immutable_db_options_.recycle_log_file_num > 0);
+        new_log = new log::Writer(
+            std::move(file_writer), new_log_number,
+            immutable_db_options_.recycle_log_file_num > 0, manual_wal_flush_);
      }
    }

@@ -738,15 +1131,25 @@ Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
    assert(creating_new_log);
    assert(!new_mem);
    assert(!new_log);
+    if (concurrent_prepare_) {
+      nonmem_write_thread_.ExitUnbatched(&nonmem_w);
+    }
    return s;
  }
  if (creating_new_log) {
+    log_write_mutex_.Lock();
    logfile_number_ = new_log_number;
    assert(new_log != nullptr);
    log_empty_ = true;
    log_dir_synced_ = false;
+    if (!logs_.empty()) {
+      // Alway flush the buffer of the last log before switching to a new one
+      log::Writer* cur_log_writer = logs_.back().writer;
+      cur_log_writer->WriteBuffer();
+    }
    logs_.emplace_back(logfile_number_, new_log);
    alive_log_files_.push_back(LogFileNumberSize(logfile_number_));
+    log_write_mutex_.Unlock();
  }
  for (auto loop_cfd : *versions_->GetColumnFamilySet()) {
    // all this is just optimization to delete logs that
@@ -768,6 +1171,9 @@ Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
  cfd->SetMemtable(new_mem);
  context->superversions_to_free_.push_back(InstallSuperVersionAndScheduleWork(
      cfd, new_superversion, mutable_cf_options));
+  if (concurrent_prepare_) {
+    nonmem_write_thread_.ExitUnbatched(&nonmem_w);
+  }
  return s;
 }

@@ -1,7 +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 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).

 #ifndef __STDC_FORMAT_MACROS
 #define __STDC_FORMAT_MACROS
@@ -1,7 +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 licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
 #pragma once

 #include <string>
@@ -1,7 +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 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
@@ -1,7 +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 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
@@ -250,7 +250,315 @@ TEST_F(DBIOFailureTest, PutFailsParanoid) {
  // the next put should NOT fail
  ASSERT_TRUE(s.ok());
 }
+#if !(defined NDEBUG) || !defined(OS_WIN)
+TEST_F(DBIOFailureTest, FlushSstRangeSyncError) {
+  Options options = CurrentOptions();
+  options.env = env_;
+  options.create_if_missing = true;
+  options.error_if_exists = false;
+  options.paranoid_checks = true;
+  options.write_buffer_size = 256 * 1024 * 1024;
+  options.writable_file_max_buffer_size = 128 * 1024;
+  options.bytes_per_sync = 128 * 1024;
+  options.level0_file_num_compaction_trigger = 4;
+  options.memtable_factory.reset(new SpecialSkipListFactory(10));
+  BlockBasedTableOptions table_options;
+  table_options.filter_policy.reset(NewBloomFilterPolicy(10));
+  options.table_factory.reset(NewBlockBasedTableFactory(table_options));

+  DestroyAndReopen(options);
+  CreateAndReopenWithCF({"pikachu"}, options);
+  Status s;
+
+  std::atomic<int> range_sync_called(0);
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "SpecialEnv::SStableFile::RangeSync", [&](void* arg) {
+        if (range_sync_called.fetch_add(1) == 0) {
+          Status* st = static_cast<Status*>(arg);
+          *st = Status::IOError("range sync dummy error");
+        }
+      });
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+  Random rnd(301);
+  std::string rnd_str =
+      RandomString(&rnd, static_cast<int>(options.bytes_per_sync / 2));
+  std::string rnd_str_512kb = RandomString(&rnd, 512 * 1024);
+
+  ASSERT_OK(Put(1, "foo", "bar"));
+  // First 1MB doesn't get range synced
+  ASSERT_OK(Put(1, "foo0_0", rnd_str_512kb));
+  ASSERT_OK(Put(1, "foo0_1", rnd_str_512kb));
+  ASSERT_OK(Put(1, "foo1_1", rnd_str));
+  ASSERT_OK(Put(1, "foo1_2", rnd_str));
+  ASSERT_OK(Put(1, "foo1_3", rnd_str));
+  ASSERT_OK(Put(1, "foo2", "bar"));
+  ASSERT_OK(Put(1, "foo3_1", rnd_str));
+  ASSERT_OK(Put(1, "foo3_2", rnd_str));
+  ASSERT_OK(Put(1, "foo3_3", rnd_str));
+  ASSERT_OK(Put(1, "foo4", "bar"));
+  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
+
+  // Following writes should fail as flush failed.
+  ASSERT_NOK(Put(1, "foo2", "bar3"));
+  ASSERT_EQ("bar", Get(1, "foo"));
+
+  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  ASSERT_GE(1, range_sync_called.load());
+
+  ReopenWithColumnFamilies({"default", "pikachu"}, options);
+  ASSERT_EQ("bar", Get(1, "foo"));
+}
+
+TEST_F(DBIOFailureTest, CompactSstRangeSyncError) {
+  Options options = CurrentOptions();
+  options.env = env_;
+  options.create_if_missing = true;
+  options.error_if_exists = false;
+  options.paranoid_checks = true;
+  options.write_buffer_size = 256 * 1024 * 1024;
+  options.writable_file_max_buffer_size = 128 * 1024;
+  options.bytes_per_sync = 128 * 1024;
+  options.level0_file_num_compaction_trigger = 2;
+  options.target_file_size_base = 256 * 1024 * 1024;
+  options.disable_auto_compactions = true;
+  BlockBasedTableOptions table_options;
+  table_options.filter_policy.reset(NewBloomFilterPolicy(10));
+  options.table_factory.reset(NewBlockBasedTableFactory(table_options));
+  DestroyAndReopen(options);
+  CreateAndReopenWithCF({"pikachu"}, options);
+  Status s;
+
+  Random rnd(301);
+  std::string rnd_str =
+      RandomString(&rnd, static_cast<int>(options.bytes_per_sync / 2));
+  std::string rnd_str_512kb = RandomString(&rnd, 512 * 1024);
+
+  ASSERT_OK(Put(1, "foo", "bar"));
+  // First 1MB doesn't get range synced
+  ASSERT_OK(Put(1, "foo0_0", rnd_str_512kb));
+  ASSERT_OK(Put(1, "foo0_1", rnd_str_512kb));
+  ASSERT_OK(Put(1, "foo1_1", rnd_str));
+  ASSERT_OK(Put(1, "foo1_2", rnd_str));
+  ASSERT_OK(Put(1, "foo1_3", rnd_str));
+  Flush(1);
+  ASSERT_OK(Put(1, "foo", "bar"));
+  ASSERT_OK(Put(1, "foo3_1", rnd_str));
+  ASSERT_OK(Put(1, "foo3_2", rnd_str));
+  ASSERT_OK(Put(1, "foo3_3", rnd_str));
+  ASSERT_OK(Put(1, "foo4", "bar"));
+  Flush(1);
+  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
+
+  std::atomic<int> range_sync_called(0);
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "SpecialEnv::SStableFile::RangeSync", [&](void* arg) {
+        if (range_sync_called.fetch_add(1) == 0) {
+          Status* st = static_cast<Status*>(arg);
+          *st = Status::IOError("range sync dummy error");
+        }
+      });
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+  ASSERT_OK(dbfull()->SetOptions(handles_[1],
+                                 {
+                                     {"disable_auto_compactions", "false"},
+                                 }));
+  dbfull()->TEST_WaitForCompact();
+
+  // Following writes should fail as flush failed.
+  ASSERT_NOK(Put(1, "foo2", "bar3"));
+  ASSERT_EQ("bar", Get(1, "foo"));
+
+  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+  ASSERT_GE(1, range_sync_called.load());
+
+  ReopenWithColumnFamilies({"default", "pikachu"}, options);
+  ASSERT_EQ("bar", Get(1, "foo"));
+}
+
+TEST_F(DBIOFailureTest, FlushSstCloseError) {
+  Options options = CurrentOptions();
+  options.env = env_;
+  options.create_if_missing = true;
+  options.error_if_exists = false;
+  options.paranoid_checks = true;
+  options.level0_file_num_compaction_trigger = 4;
+  options.memtable_factory.reset(new SpecialSkipListFactory(2));
+
+  DestroyAndReopen(options);
+  CreateAndReopenWithCF({"pikachu"}, options);
+  Status s;
+  std::atomic<int> close_called(0);
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "SpecialEnv::SStableFile::Close", [&](void* arg) {
+        if (close_called.fetch_add(1) == 0) {
+          Status* st = static_cast<Status*>(arg);
+          *st = Status::IOError("close dummy error");
+        }
+      });
+
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+  ASSERT_OK(Put(1, "foo", "bar"));
+  ASSERT_OK(Put(1, "foo1", "bar1"));
+  ASSERT_OK(Put(1, "foo", "bar2"));
+  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
+
+  // Following writes should fail as flush failed.
+  ASSERT_NOK(Put(1, "foo2", "bar3"));
+  ASSERT_EQ("bar2", Get(1, "foo"));
+  ASSERT_EQ("bar1", Get(1, "foo1"));
+
+  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+
+  ReopenWithColumnFamilies({"default", "pikachu"}, options);
+  ASSERT_EQ("bar2", Get(1, "foo"));
+  ASSERT_EQ("bar1", Get(1, "foo1"));
+}
+
+TEST_F(DBIOFailureTest, CompactionSstCloseError) {
+  Options options = CurrentOptions();
+  options.env = env_;
+  options.create_if_missing = true;
+  options.error_if_exists = false;
+  options.paranoid_checks = true;
+  options.level0_file_num_compaction_trigger = 2;
+  options.disable_auto_compactions = true;
+
+  DestroyAndReopen(options);
+  CreateAndReopenWithCF({"pikachu"}, options);
+  Status s;
+
+  ASSERT_OK(Put(1, "foo", "bar"));
+  ASSERT_OK(Put(1, "foo2", "bar"));
+  Flush(1);
+  ASSERT_OK(Put(1, "foo", "bar2"));
+  ASSERT_OK(Put(1, "foo2", "bar"));
+  Flush(1);
+  ASSERT_OK(Put(1, "foo", "bar3"));
+  ASSERT_OK(Put(1, "foo2", "bar"));
+  Flush(1);
+  dbfull()->TEST_WaitForCompact();
+
+  std::atomic<int> close_called(0);
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "SpecialEnv::SStableFile::Close", [&](void* arg) {
+        if (close_called.fetch_add(1) == 0) {
+          Status* st = static_cast<Status*>(arg);
+          *st = Status::IOError("close dummy error");
+        }
+      });
+
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ASSERT_OK(dbfull()->SetOptions(handles_[1],
+                                 {
+                                     {"disable_auto_compactions", "false"},
+                                 }));
+  dbfull()->TEST_WaitForCompact();
+
+  // Following writes should fail as compaction failed.
+  ASSERT_NOK(Put(1, "foo2", "bar3"));
+  ASSERT_EQ("bar3", Get(1, "foo"));
+
+  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+
+  ReopenWithColumnFamilies({"default", "pikachu"}, options);
+  ASSERT_EQ("bar3", Get(1, "foo"));
+}
+
+TEST_F(DBIOFailureTest, FlushSstSyncError) {
+  Options options = CurrentOptions();
+  options.env = env_;
+  options.create_if_missing = true;
+  options.error_if_exists = false;
+  options.paranoid_checks = true;
+  options.use_fsync = false;
+  options.level0_file_num_compaction_trigger = 4;
+  options.memtable_factory.reset(new SpecialSkipListFactory(2));
+
+  DestroyAndReopen(options);
+  CreateAndReopenWithCF({"pikachu"}, options);
+  Status s;
+  std::atomic<int> sync_called(0);
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "SpecialEnv::SStableFile::Sync", [&](void* arg) {
+        if (sync_called.fetch_add(1) == 0) {
+          Status* st = static_cast<Status*>(arg);
+          *st = Status::IOError("sync dummy error");
+        }
+      });
+
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+
+  ASSERT_OK(Put(1, "foo", "bar"));
+  ASSERT_OK(Put(1, "foo1", "bar1"));
+  ASSERT_OK(Put(1, "foo", "bar2"));
+  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
+
+  // Following writes should fail as flush failed.
+  ASSERT_NOK(Put(1, "foo2", "bar3"));
+  ASSERT_EQ("bar2", Get(1, "foo"));
+  ASSERT_EQ("bar1", Get(1, "foo1"));
+
+  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+
+  ReopenWithColumnFamilies({"default", "pikachu"}, options);
+  ASSERT_EQ("bar2", Get(1, "foo"));
+  ASSERT_EQ("bar1", Get(1, "foo1"));
+}
+
+TEST_F(DBIOFailureTest, CompactionSstSyncError) {
+  Options options = CurrentOptions();
+  options.env = env_;
+  options.create_if_missing = true;
+  options.error_if_exists = false;
+  options.paranoid_checks = true;
+  options.level0_file_num_compaction_trigger = 2;
+  options.disable_auto_compactions = true;
+  options.use_fsync = false;
+
+  DestroyAndReopen(options);
+  CreateAndReopenWithCF({"pikachu"}, options);
+  Status s;
+
+  ASSERT_OK(Put(1, "foo", "bar"));
+  ASSERT_OK(Put(1, "foo2", "bar"));
+  Flush(1);
+  ASSERT_OK(Put(1, "foo", "bar2"));
+  ASSERT_OK(Put(1, "foo2", "bar"));
+  Flush(1);
+  ASSERT_OK(Put(1, "foo", "bar3"));
+  ASSERT_OK(Put(1, "foo2", "bar"));
+  Flush(1);
+  dbfull()->TEST_WaitForCompact();
+
+  std::atomic<int> sync_called(0);
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "SpecialEnv::SStableFile::Sync", [&](void* arg) {
+        if (sync_called.fetch_add(1) == 0) {
+          Status* st = static_cast<Status*>(arg);
+          *st = Status::IOError("close dummy error");
+        }
+      });
+
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  ASSERT_OK(dbfull()->SetOptions(handles_[1],
+                                 {
+                                     {"disable_auto_compactions", "false"},
+                                 }));
+  dbfull()->TEST_WaitForCompact();
+
+  // Following writes should fail as compaction failed.
+  ASSERT_NOK(Put(1, "foo2", "bar3"));
+  ASSERT_EQ("bar3", Get(1, "foo"));
+
+  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
+
+  ReopenWithColumnFamilies({"default", "pikachu"}, options);
+  ASSERT_EQ("bar3", Get(1, "foo"));
+}
+#endif  // !(defined NDEBUG) || !defined(OS_WIN)
 }  // namespace rocksdb

 int main(int argc, char** argv) {
@@ -1,7 +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 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
@@ -560,7 +560,7 @@ void DBIter::MergeValuesNewToOld() {
      const Slice val = iter_->value();
      s = MergeHelper::TimedFullMerge(
          merge_operator_, ikey.user_key, &val, merge_context_.GetOperands(),
-          &saved_value_, logger_, statistics_, env_, &pinned_value_);
+          &saved_value_, logger_, statistics_, env_, &pinned_value_, true);
      if (!s.ok()) {
        status_ = s;
      }
@@ -585,7 +585,7 @@ void DBIter::MergeValuesNewToOld() {
  s = MergeHelper::TimedFullMerge(merge_operator_, saved_key_.GetUserKey(),
                                  nullptr, merge_context_.GetOperands(),
                                  &saved_value_, logger_, statistics_, env_,
-                                  &pinned_value_);
+                                  &pinned_value_, true);
  if (!s.ok()) {
    status_ = s;
  }
@@ -642,6 +642,7 @@ void DBIter::ReverseToBackward() {
    while (iter_->Valid() &&
           user_comparator_->Compare(ikey.user_key, saved_key_.GetUserKey()) >
               0) {
+      assert(ikey.sequence != kMaxSequenceNumber);
      if (ikey.sequence > sequence_) {
        PERF_COUNTER_ADD(internal_recent_skipped_count, 1);
      } else {
@@ -804,13 +805,13 @@ bool DBIter::FindValueForCurrentKey() {
        s = MergeHelper::TimedFullMerge(
            merge_operator_, saved_key_.GetUserKey(), nullptr,
            merge_context_.GetOperands(), &saved_value_, logger_, statistics_,
-            env_, &pinned_value_);
+            env_, &pinned_value_, true);
      } else {
        assert(last_not_merge_type == kTypeValue);
        s = MergeHelper::TimedFullMerge(
            merge_operator_, saved_key_.GetUserKey(), &pinned_value_,
            merge_context_.GetOperands(), &saved_value_, logger_, statistics_,
-            env_, &pinned_value_);
+            env_, &pinned_value_, true);
      }
      break;
    case kTypeValue:
@@ -882,7 +883,7 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
    s = MergeHelper::TimedFullMerge(merge_operator_, saved_key_.GetUserKey(),
                                    nullptr, merge_context_.GetOperands(),
                                    &saved_value_, logger_, statistics_, env_,
-                                    &pinned_value_);
+                                    &pinned_value_, true);
    // Make iter_ valid and point to saved_key_
    if (!iter_->Valid() ||
        !user_comparator_->Equal(ikey.user_key, saved_key_.GetUserKey())) {
@@ -900,7 +901,7 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
  s = MergeHelper::TimedFullMerge(merge_operator_, saved_key_.GetUserKey(),
                                  &val, merge_context_.GetOperands(),
                                  &saved_value_, logger_, statistics_, env_,
-                                  &pinned_value_);
+                                  &pinned_value_, true);
  valid_ = true;
  if (!s.ok()) {
    status_ = s;
@@ -954,6 +955,7 @@ void DBIter::FindPrevUserKey() {
        ++num_skipped;
      }
    }
+    assert(ikey.sequence != kMaxSequenceNumber);
    if (ikey.sequence > sequence_) {
      PERF_COUNTER_ADD(internal_recent_skipped_count, 1);
    } else {
@@ -1,7 +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 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
@@ -1,7 +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 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 <string>
 #include <vector>
@@ -352,11 +352,11 @@ TEST_F(DBIteratorTest, DBIteratorPrevNext) {
    SetPerfLevel(kEnableCount);
    ASSERT_TRUE(GetPerfLevel() == kEnableCount);

-    perf_context.Reset();
+    get_perf_context()->Reset();
    db_iter->SeekToLast();

    ASSERT_TRUE(db_iter->Valid());
-    ASSERT_EQ(static_cast<int>(perf_context.internal_key_skipped_count), 7);
+    ASSERT_EQ(static_cast<int>(get_perf_context()->internal_key_skipped_count), 7);
    ASSERT_EQ(db_iter->key().ToString(), "b");

    SetPerfLevel(kDisable);
@@ -471,11 +471,11 @@ TEST_F(DBIteratorTest, DBIteratorPrevNext) {
    SetPerfLevel(kEnableCount);
    ASSERT_TRUE(GetPerfLevel() == kEnableCount);

-    perf_context.Reset();
+    get_perf_context()->Reset();
    db_iter->SeekToLast();

    ASSERT_TRUE(db_iter->Valid());
-    ASSERT_EQ(static_cast<int>(perf_context.internal_delete_skipped_count), 1);
+    ASSERT_EQ(static_cast<int>(get_perf_context()->internal_delete_skipped_count), 1);
    ASSERT_EQ(db_iter->key().ToString(), "b");

    SetPerfLevel(kDisable);
@@ -1,7 +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 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
@@ -22,6 +22,34 @@ class DBIteratorTest : public DBTestBase {
  DBIteratorTest() : DBTestBase("/db_iterator_test") {}
 };

+class FlushBlockEveryKeyPolicy : public FlushBlockPolicy {
+ public:
+  virtual bool Update(const Slice& key, const Slice& value) override {
+    if (!start_) {
+      start_ = true;
+      return false;
+    }
+    return true;
+  }
+ private:
+  bool start_ = false;
+};
+
+class FlushBlockEveryKeyPolicyFactory : public FlushBlockPolicyFactory {
+ public:
+  explicit FlushBlockEveryKeyPolicyFactory() {}
+
+  const char* Name() const override {
+    return "FlushBlockEveryKeyPolicyFactory";
+  }
+
+  FlushBlockPolicy* NewFlushBlockPolicy(
+    const BlockBasedTableOptions& table_options,
+    const BlockBuilder& data_block_builder) const override {
+    return new FlushBlockEveryKeyPolicy;
+  }
+};
+
 TEST_F(DBIteratorTest, IteratorProperty) {
  // The test needs to be changed if kPersistedTier is supported in iterator.
  Options options = CurrentOptions();
@@ -944,11 +972,11 @@ TEST_F(DBIteratorTest, DBIteratorBoundTest) {
    ASSERT_TRUE(iter->Valid());
    ASSERT_EQ(iter->key().compare(("b1")), 0);

-    perf_context.Reset();
+    get_perf_context()->Reset();
    iter->Next();

    ASSERT_TRUE(iter->Valid());
-    ASSERT_EQ(static_cast<int>(perf_context.internal_delete_skipped_count), 2);
+    ASSERT_EQ(static_cast<int>(get_perf_context()->internal_delete_skipped_count), 2);

    // now testing with iterate_bound
    Slice prefix("c");
@@ -956,7 +984,7 @@ TEST_F(DBIteratorTest, DBIteratorBoundTest) {

    iter.reset(db_->NewIterator(ro));

-    perf_context.Reset();
+    get_perf_context()->Reset();

    iter->Seek("b");
    ASSERT_TRUE(iter->Valid());
@@ -971,10 +999,54 @@ TEST_F(DBIteratorTest, DBIteratorBoundTest) {
    // even though the key is deleted
    // hence internal_delete_skipped_count should be 0
    ASSERT_TRUE(!iter->Valid());
-    ASSERT_EQ(static_cast<int>(perf_context.internal_delete_skipped_count), 0);
+    ASSERT_EQ(static_cast<int>(get_perf_context()->internal_delete_skipped_count), 0);
  }
 }

+TEST_F(DBIteratorTest, DBIteratorBoundOptimizationTest) {
+  int upper_bound_hits = 0;
+  Options options = CurrentOptions();
+  rocksdb::SyncPoint::GetInstance()->SetCallBack(
+      "BlockBasedTable::BlockEntryIteratorState::KeyReachedUpperBound",
+      [&upper_bound_hits](void* arg) {
+        assert(arg != nullptr);
+        upper_bound_hits += (*static_cast<bool*>(arg) ? 1 : 0);
+      });
+  rocksdb::SyncPoint::GetInstance()->EnableProcessing();
+  options.env = env_;
+  options.create_if_missing = true;
+  options.prefix_extractor = nullptr;
+  BlockBasedTableOptions table_options;
+  table_options.flush_block_policy_factory =
+    std::make_shared<FlushBlockEveryKeyPolicyFactory>();
+  options.table_factory.reset(NewBlockBasedTableFactory(table_options));
+
+  DestroyAndReopen(options);
+  ASSERT_OK(Put("foo1", "bar1"));
+  ASSERT_OK(Put("foo2", "bar2"));
+  ASSERT_OK(Put("foo4", "bar4"));
+  ASSERT_OK(Flush());
+
+  Slice ub("foo3");
+  ReadOptions ro;
+  ro.iterate_upper_bound = &ub;
+
+  std::unique_ptr<Iterator> iter(db_->NewIterator(ro));
+
+  iter->Seek("foo");
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_EQ(iter->key().compare(Slice("foo1")), 0);
+  ASSERT_EQ(upper_bound_hits, 0);
+
+  iter->Next();
+  ASSERT_TRUE(iter->Valid());
+  ASSERT_EQ(iter->key().compare(Slice("foo2")), 0);
+  ASSERT_EQ(upper_bound_hits, 0);
+
+  iter->Next();
+  ASSERT_FALSE(iter->Valid());
+  ASSERT_EQ(upper_bound_hits, 1);
+}
 // TODO(3.13): fix the issue of Seek() + Prev() which might not necessary
 //             return the biggest key which is smaller than the seek key.
 TEST_F(DBIteratorTest, PrevAfterAndNextAfterMerge) {
@@ -1814,7 +1886,7 @@ TEST_F(DBIteratorTest, DBIteratorSkipRecentDuplicatesTest) {
 #endif

  // Seek iterator to a smaller key.
-  perf_context.Reset();
+  get_perf_context()->Reset();
  iter->Seek("a");
  ASSERT_TRUE(iter->Valid());
  EXPECT_EQ("b", iter->key().ToString());
@@ -1822,17 +1894,17 @@ TEST_F(DBIteratorTest, DBIteratorSkipRecentDuplicatesTest) {

  // Check that the seek didn't do too much work.
  // Checks are not tight, just make sure that everything is well below 100.
-  EXPECT_LT(perf_context.internal_key_skipped_count, 4);
-  EXPECT_LT(perf_context.internal_recent_skipped_count, 8);
-  EXPECT_LT(perf_context.seek_on_memtable_count, 10);
-  EXPECT_LT(perf_context.next_on_memtable_count, 10);
-  EXPECT_LT(perf_context.prev_on_memtable_count, 10);
+  EXPECT_LT(get_perf_context()->internal_key_skipped_count, 4);
+  EXPECT_LT(get_perf_context()->internal_recent_skipped_count, 8);
+  EXPECT_LT(get_perf_context()->seek_on_memtable_count, 10);
+  EXPECT_LT(get_perf_context()->next_on_memtable_count, 10);
+  EXPECT_LT(get_perf_context()->prev_on_memtable_count, 10);

  // Check that iterator did something like what we expect.
-  EXPECT_EQ(perf_context.internal_delete_skipped_count, 0);
-  EXPECT_EQ(perf_context.internal_merge_count, 0);
-  EXPECT_GE(perf_context.internal_recent_skipped_count, 2);
-  EXPECT_GE(perf_context.seek_on_memtable_count, 2);
+  EXPECT_EQ(get_perf_context()->internal_delete_skipped_count, 0);
+  EXPECT_EQ(get_perf_context()->internal_merge_count, 0);
+  EXPECT_GE(get_perf_context()->internal_recent_skipped_count, 2);
+  EXPECT_GE(get_perf_context()->seek_on_memtable_count, 2);
  EXPECT_EQ(1, options.statistics->getTickerCount(
                 NUMBER_OF_RESEEKS_IN_ITERATION));
 }
@@ -1,7 +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 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
@@ -116,6 +116,7 @@ TEST_F(DBTestXactLogIterator, TransactionLogIteratorRace) {
      dbfull()->Flush(FlushOptions());
      Put("key4", DummyString(1024));
      ASSERT_EQ(dbfull()->GetLatestSequenceNumber(), 4U);
+      dbfull()->FlushWAL(false);

      {
        auto iter = OpenTransactionLogIter(0);
@@ -132,6 +133,7 @@ TEST_F(DBTestXactLogIterator, TransactionLogIteratorRace) {

      // "key5" would be written in a new memtable and log
      Put("key5", DummyString(1024));
+      dbfull()->FlushWAL(false);
      {
        // this iter would miss "key4" if not fixed
        auto iter = OpenTransactionLogIter(0);
@@ -181,6 +183,7 @@ TEST_F(DBTestXactLogIterator, TransactionLogIteratorCorruptedLog) {
      Put("key"+ToString(i), DummyString(10));
    }
    dbfull()->Flush(FlushOptions());
+    dbfull()->FlushWAL(false);
    // Corrupt this log to create a gap
    rocksdb::VectorLogPtr wal_files;
    ASSERT_OK(dbfull()->GetSortedWalFiles(wal_files));
@@ -194,6 +197,7 @@ TEST_F(DBTestXactLogIterator, TransactionLogIteratorCorruptedLog) {

    // Insert a new entry to a new log file
    Put("key1025", DummyString(10));
+    dbfull()->FlushWAL(false);
    // Try to read from the beginning. Should stop before the gap and read less
    // than 1025 entries
    auto iter = OpenTransactionLogIter(0);
@@ -1,7 +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 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 <memory>
 #include <string>
@@ -21,7 +21,7 @@ class DBMemTableTest : public DBTestBase {

 class MockMemTableRep : public MemTableRep {
 public:
-  explicit MockMemTableRep(MemTableAllocator* allocator, MemTableRep* rep)
+  explicit MockMemTableRep(Allocator* allocator, MemTableRep* rep)
      : MemTableRep(allocator), rep_(rep), num_insert_with_hint_(0) {}

  virtual KeyHandle Allocate(const size_t len, char** buf) override {
@@ -72,7 +72,7 @@ class MockMemTableRep : public MemTableRep {
 class MockMemTableRepFactory : public MemTableRepFactory {
 public:
  virtual MemTableRep* CreateMemTableRep(const MemTableRep::KeyComparator& cmp,
-                                         MemTableAllocator* allocator,
+                                         Allocator* allocator,
                                         const SliceTransform* transform,
                                         Logger* logger) override {
    SkipListFactory factory;
@@ -82,14 +82,27 @@ class MockMemTableRepFactory : public MemTableRepFactory {
    return mock_rep_;
  }

+  virtual MemTableRep* CreateMemTableRep(const MemTableRep::KeyComparator& cmp,
+                                         Allocator* allocator,
+                                         const SliceTransform* transform,
+                                         Logger* logger,
+                                         uint32_t column_family_id) override {
+    last_column_family_id_ = column_family_id;
+    return CreateMemTableRep(cmp, allocator, transform, logger);
+  }
+
  virtual const char* Name() const override { return "MockMemTableRepFactory"; }

  MockMemTableRep* rep() { return mock_rep_; }

  bool IsInsertConcurrentlySupported() const override { return false; }

+  uint32_t GetLastColumnFamilyId() { return last_column_family_id_; }
+
 private:
  MockMemTableRep* mock_rep_;
+  // workaround since there's no port::kMaxUint32 yet.
+  uint32_t last_column_family_id_ = static_cast<uint32_t>(-1);
 };

 class TestPrefixExtractor : public SliceTransform {
@@ -155,6 +168,24 @@ TEST_F(DBMemTableTest, InsertWithHint) {
  ASSERT_EQ("vvv", Get("whitelisted"));
 }

+TEST_F(DBMemTableTest, ColumnFamilyId) {
+  // Verifies MemTableRepFactory is told the right column family id.
+  Options options;
+  options.allow_concurrent_memtable_write = false;
+  options.create_if_missing = true;
+  options.memtable_factory.reset(new MockMemTableRepFactory());
+  DestroyAndReopen(options);
+  CreateAndReopenWithCF({"pikachu"}, options);
+
+  for (int cf = 0; cf < 2; ++cf) {
+    ASSERT_OK(Put(cf, "key", "val"));
+    ASSERT_OK(Flush(cf));
+    ASSERT_EQ(
+        cf, static_cast<MockMemTableRepFactory*>(options.memtable_factory.get())
+                ->GetLastColumnFamilyId());
+  }
+}
+
 }  // namespace rocksdb

 int main(int argc, char** argv) {
@@ -1,7 +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 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 <string>
 #include <vector>

@@ -18,28 +18,6 @@ class DBMergeOperatorTest : public DBTestBase {
  DBMergeOperatorTest() : DBTestBase("/db_merge_operator_test") {}
 };

-// A test merge operator mimics put but also fails if one of merge operands is
-// "corrupted".
-class TestPutOperator : public MergeOperator {
- public:
-  virtual bool FullMergeV2(const MergeOperationInput& merge_in,
-                           MergeOperationOutput* merge_out) const override {
-    if (merge_in.existing_value != nullptr &&
-        *(merge_in.existing_value) == "corrupted") {
-      return false;
-    }
-    for (auto value : merge_in.operand_list) {
-      if (value == "corrupted") {
-        return false;
-      }
-    }
-    merge_out->existing_operand = merge_in.operand_list.back();
-    return true;
-  }
-
-  virtual const char* Name() const override { return "TestPutOperator"; }
-};
-
 TEST_F(DBMergeOperatorTest, MergeErrorOnRead) {
  Options options;
  options.create_if_missing = true;
@@ -1,7 +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 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
@@ -15,7 +15,9 @@
 #include "db/db_test_util.h"
 #include "options/options_helper.h"
 #include "port/stack_trace.h"
+#include "rocksdb/cache.h"
 #include "rocksdb/convenience.h"
+#include "rocksdb/rate_limiter.h"
 #include "util/random.h"
 #include "util/sync_point.h"
 #include "util/testutil.h"
@@ -251,18 +253,44 @@ TEST_F(DBOptionsTest, SetOptionsMayTriggerCompaction) {
 TEST_F(DBOptionsTest, SetBackgroundCompactionThreads) {
  Options options;
  options.create_if_missing = true;
-  options.base_background_compactions = 1;  // default value
  options.max_background_compactions = 1;   // default value
  options.env = env_;
  Reopen(options);
  ASSERT_EQ(1, dbfull()->TEST_BGCompactionsAllowed());
-  ASSERT_OK(dbfull()->SetDBOptions({{"base_background_compactions", "2"},
-                                    {"max_background_compactions", "3"}}));
-  ASSERT_EQ(2, dbfull()->TEST_BGCompactionsAllowed());
+  ASSERT_OK(dbfull()->SetDBOptions({{"max_background_compactions", "3"}}));
+  ASSERT_EQ(1, dbfull()->TEST_BGCompactionsAllowed());
  auto stop_token = dbfull()->TEST_write_controler().GetStopToken();
  ASSERT_EQ(3, dbfull()->TEST_BGCompactionsAllowed());
 }

+TEST_F(DBOptionsTest, SetBackgroundJobs) {
+  Options options;
+  options.create_if_missing = true;
+  options.max_background_jobs = 8;
+  options.env = env_;
+  Reopen(options);
+
+  for (int i = 0; i < 2; ++i) {
+    if (i > 0) {
+      options.max_background_jobs = 12;
+      ASSERT_OK(dbfull()->SetDBOptions(
+          {{"max_background_jobs",
+            std::to_string(options.max_background_jobs)}}));
+    }
+
+    ASSERT_EQ(options.max_background_jobs / 4,
+              dbfull()->TEST_BGFlushesAllowed());
+    ASSERT_EQ(1, dbfull()->TEST_BGCompactionsAllowed());
+
+    auto stop_token = dbfull()->TEST_write_controler().GetStopToken();
+
+    ASSERT_EQ(options.max_background_jobs / 4,
+              dbfull()->TEST_BGFlushesAllowed());
+    ASSERT_EQ(3 * options.max_background_jobs / 4,
+              dbfull()->TEST_BGCompactionsAllowed());
+  }
+}
+
 TEST_F(DBOptionsTest, AvoidFlushDuringShutdown) {
  Options options;
  options.create_if_missing = true;
@@ -383,6 +411,36 @@ TEST_F(DBOptionsTest, DeleteObsoleteFilesPeriodChange) {
  Close();
 }

+TEST_F(DBOptionsTest, MaxOpenFilesChange) {
+  SpecialEnv env(env_);
+  Options options;
+  options.env = CurrentOptions().env;
+  options.max_open_files = -1;
+
+  Reopen(options);
+
+  Cache* tc = dbfull()->TEST_table_cache();
+
+  ASSERT_EQ(-1, dbfull()->GetDBOptions().max_open_files);
+  ASSERT_LT(2000, tc->GetCapacity());
+  ASSERT_OK(dbfull()->SetDBOptions({{"max_open_files", "1024"}}));
+  ASSERT_EQ(1024, dbfull()->GetDBOptions().max_open_files);
+  // examine the table cache (actual size should be 1014)
+  ASSERT_GT(1500, tc->GetCapacity());
+  Close();
+}
+
+TEST_F(DBOptionsTest, SanitizeDelayedWriteRate) {
+  Options options;
+  options.delayed_write_rate = 0;
+  Reopen(options);
+  ASSERT_EQ(16 * 1024 * 1024, dbfull()->GetDBOptions().delayed_write_rate);
+
+  options.rate_limiter.reset(NewGenericRateLimiter(31 * 1024 * 1024));
+  Reopen(options);
+  ASSERT_EQ(31 * 1024 * 1024, dbfull()->GetDBOptions().delayed_write_rate);
+}
+
 #endif  // ROCKSDB_LITE

 }  // namespace rocksdb
@@ -1,7 +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 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
@@ -425,6 +425,13 @@ TEST_F(DBPropertiesTest, ReadLatencyHistogramByLevel) {
  ASSERT_NE(std::string::npos, prop.find("** Level 0 read latency histogram"));
  ASSERT_NE(std::string::npos, prop.find("** Level 1 read latency histogram"));
  ASSERT_EQ(std::string::npos, prop.find("** Level 2 read latency histogram"));
+
+  // Clear internal stats
+  dbfull()->ResetStats();
+  ASSERT_TRUE(dbfull()->GetProperty("rocksdb.cfstats", &prop));
+  ASSERT_EQ(std::string::npos, prop.find("** Level 0 read latency histogram"));
+  ASSERT_EQ(std::string::npos, prop.find("** Level 1 read latency histogram"));
+  ASSERT_EQ(std::string::npos, prop.find("** Level 2 read latency histogram"));
 }

 TEST_F(DBPropertiesTest, AggregatedTablePropertiesAtLevel) {
@@ -526,9 +533,9 @@ TEST_F(DBPropertiesTest, NumImmutableMemTable) {
    ASSERT_TRUE(dbfull()->GetProperty(
        handles_[1], "rocksdb.num-entries-active-mem-table", &num));
    ASSERT_EQ(num, "1");
-    perf_context.Reset();
+    get_perf_context()->Reset();
    Get(1, "k1");
-    ASSERT_EQ(1, static_cast<int>(perf_context.get_from_memtable_count));
+    ASSERT_EQ(1, static_cast<int>(get_perf_context()->get_from_memtable_count));

    ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "k2", big_value));
    ASSERT_TRUE(dbfull()->GetProperty(handles_[1],
@@ -541,12 +548,12 @@ TEST_F(DBPropertiesTest, NumImmutableMemTable) {
        handles_[1], "rocksdb.num-entries-imm-mem-tables", &num));
    ASSERT_EQ(num, "1");

-    perf_context.Reset();
+    get_perf_context()->Reset();
    Get(1, "k1");
-    ASSERT_EQ(2, static_cast<int>(perf_context.get_from_memtable_count));
-    perf_context.Reset();
+    ASSERT_EQ(2, static_cast<int>(get_perf_context()->get_from_memtable_count));
+    get_perf_context()->Reset();
    Get(1, "k2");
-    ASSERT_EQ(1, static_cast<int>(perf_context.get_from_memtable_count));
+    ASSERT_EQ(1, static_cast<int>(get_perf_context()->get_from_memtable_count));

    ASSERT_OK(dbfull()->Put(writeOpt, handles_[1], "k3", big_value));
    ASSERT_TRUE(dbfull()->GetProperty(
@@ -560,15 +567,15 @@ TEST_F(DBPropertiesTest, NumImmutableMemTable) {
    ASSERT_TRUE(dbfull()->GetProperty(
        handles_[1], "rocksdb.num-entries-imm-mem-tables", &num));
    ASSERT_EQ(num, "2");
-    perf_context.Reset();
+    get_perf_context()->Reset();
    Get(1, "k2");
-    ASSERT_EQ(2, static_cast<int>(perf_context.get_from_memtable_count));
-    perf_context.Reset();
+    ASSERT_EQ(2, static_cast<int>(get_perf_context()->get_from_memtable_count));
+    get_perf_context()->Reset();
    Get(1, "k3");
-    ASSERT_EQ(1, static_cast<int>(perf_context.get_from_memtable_count));
-    perf_context.Reset();
+    ASSERT_EQ(1, static_cast<int>(get_perf_context()->get_from_memtable_count));
+    get_perf_context()->Reset();
    Get(1, "k1");
-    ASSERT_EQ(3, static_cast<int>(perf_context.get_from_memtable_count));
+    ASSERT_EQ(3, static_cast<int>(get_perf_context()->get_from_memtable_count));

    ASSERT_OK(Flush(1));
    ASSERT_TRUE(dbfull()->GetProperty(handles_[1],
@@ -663,7 +670,7 @@ TEST_F(DBPropertiesTest, DISABLED_GetProperty) {
  ASSERT_EQ(num, "0");
  ASSERT_TRUE(dbfull()->GetProperty("rocksdb.estimate-num-keys", &num));
  ASSERT_EQ(num, "1");
-  perf_context.Reset();
+  get_perf_context()->Reset();

  ASSERT_OK(dbfull()->Put(writeOpt, "k2", big_value));
  ASSERT_TRUE(dbfull()->GetProperty("rocksdb.num-immutable-mem-table", &num));
@@ -906,7 +913,7 @@ TEST_F(DBPropertiesTest, EstimatePendingCompBytes) {
  Flush();
  ASSERT_TRUE(dbfull()->GetIntProperty(
      "rocksdb.estimate-pending-compaction-bytes", &int_num));
-  ASSERT_EQ(int_num, 0U);
+  ASSERT_GT(int_num, 0U);

  ASSERT_OK(dbfull()->Put(writeOpt, "k3", big_value));
  Flush();
@@ -933,7 +940,6 @@ TEST_F(DBPropertiesTest, EstimateCompressionRatio) {
  Options options = CurrentOptions();
  options.compression_per_level = {kNoCompression, kSnappyCompression};
  options.disable_auto_compactions = true;
-  options.max_background_flushes = 0;
  options.num_levels = 2;
  Reopen(options);

@@ -1061,7 +1067,6 @@ class CountingDeleteTabPropCollectorFactory
 TEST_F(DBPropertiesTest, GetUserDefinedTableProperties) {
  Options options = CurrentOptions();
  options.level0_file_num_compaction_trigger = (1 << 30);
-  options.max_background_flushes = 0;
  options.table_properties_collector_factories.resize(1);
  std::shared_ptr<CountingUserTblPropCollectorFactory> collector_factory =
      std::make_shared<CountingUserTblPropCollectorFactory>(0);
@@ -1102,7 +1107,6 @@ TEST_F(DBPropertiesTest, GetUserDefinedTableProperties) {
 TEST_F(DBPropertiesTest, UserDefinedTablePropertiesContext) {
  Options options = CurrentOptions();
  options.level0_file_num_compaction_trigger = 3;
-  options.max_background_flushes = 0;
  options.table_properties_collector_factories.resize(1);
  std::shared_ptr<CountingUserTblPropCollectorFactory> collector_factory =
      std::make_shared<CountingUserTblPropCollectorFactory>(1);
@@ -1224,7 +1228,7 @@ TEST_F(DBPropertiesTest, TablePropertiesNeedCompactTest) {

  {
    SetPerfLevel(kEnableCount);
-    perf_context.Reset();
+    get_perf_context()->Reset();
    int c = 0;
    std::unique_ptr<Iterator> iter(db_->NewIterator(ReadOptions()));
    iter->Seek(Key(kMaxKey - 100));
@@ -1232,8 +1236,8 @@ TEST_F(DBPropertiesTest, TablePropertiesNeedCompactTest) {
      iter->Next();
    }
    ASSERT_EQ(c, 0);
-    ASSERT_LT(perf_context.internal_delete_skipped_count, 30u);
-    ASSERT_LT(perf_context.internal_key_skipped_count, 30u);
+    ASSERT_LT(get_perf_context()->internal_delete_skipped_count, 30u);
+    ASSERT_LT(get_perf_context()->internal_key_skipped_count, 30u);
    SetPerfLevel(kDisable);
  }
 }
@@ -1280,19 +1284,31 @@ TEST_F(DBPropertiesTest, NeedCompactHintPersistentTest) {
  ASSERT_EQ(NumTableFilesAtLevel(0), 0);
  {
    SetPerfLevel(kEnableCount);
-    perf_context.Reset();
+    get_perf_context()->Reset();
    int c = 0;
    std::unique_ptr<Iterator> iter(db_->NewIterator(ReadOptions()));
    for (iter->Seek(Key(0)); iter->Valid(); iter->Next()) {
      c++;
    }
    ASSERT_EQ(c, 2);
-    ASSERT_EQ(perf_context.internal_delete_skipped_count, 0);
+    ASSERT_EQ(get_perf_context()->internal_delete_skipped_count, 0);
    // We iterate every key twice. Is it a bug?
-    ASSERT_LE(perf_context.internal_key_skipped_count, 2);
+    ASSERT_LE(get_perf_context()->internal_key_skipped_count, 2);
    SetPerfLevel(kDisable);
  }
 }
+
+TEST_F(DBPropertiesTest, EstimateNumKeysUnderflow) {
+  Options options;
+  Reopen(options);
+  Put("foo", "bar");
+  Delete("foo");
+  Delete("foo");
+  uint64_t num_keys = 0;
+  ASSERT_TRUE(dbfull()->GetIntProperty("rocksdb.estimate-num-keys", &num_keys));
+  ASSERT_EQ(0, num_keys);
+}
+
 #endif  // ROCKSDB_LITE
 }  // namespace rocksdb

@@ -1,7 +1,7 @@
 //  Copyright (c) 2016-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 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"
@@ -27,6 +27,9 @@ class DBRangeDelTest : public DBTestBase {
 // ROCKSDB_LITE
 #ifndef ROCKSDB_LITE
 TEST_F(DBRangeDelTest, NonBlockBasedTableNotSupported) {
+  if (!IsMemoryMappedAccessSupported()) {
+    return;
+  }
  Options opts = CurrentOptions();
  opts.table_factory.reset(new PlainTableFactory());
  opts.prefix_extractor.reset(NewNoopTransform());
@@ -789,6 +792,7 @@ TEST_F(DBRangeDelTest, IteratorIgnoresRangeDeletions) {
  db_->ReleaseSnapshot(snapshot);
 }

+#ifndef ROCKSDB_UBSAN_RUN
 TEST_F(DBRangeDelTest, TailingIteratorRangeTombstoneUnsupported) {
  db_->Put(WriteOptions(), "key", "val");
  // snapshot prevents key from being deleted during flush
@@ -815,6 +819,115 @@ TEST_F(DBRangeDelTest, TailingIteratorRangeTombstoneUnsupported) {
  }
  db_->ReleaseSnapshot(snapshot);
 }
+
+#endif  // !ROCKSDB_UBSAN_RUN
+
+TEST_F(DBRangeDelTest, SubcompactionHasEmptyDedicatedRangeDelFile) {
+  const int kNumFiles = 2, kNumKeysPerFile = 4;
+  Options options = CurrentOptions();
+  options.compression = kNoCompression;
+  options.disable_auto_compactions = true;
+  options.level0_file_num_compaction_trigger = kNumFiles;
+  options.max_subcompactions = 2;
+  options.num_levels = 2;
+  options.target_file_size_base = 4096;
+  Reopen(options);
+
+  // need a L1 file for subcompaction to be triggered
+  ASSERT_OK(
+      db_->Put(WriteOptions(), db_->DefaultColumnFamily(), Key(0), "val"));
+  ASSERT_OK(db_->Flush(FlushOptions()));
+  MoveFilesToLevel(1);
+
+  // put enough keys to fill up the first subcompaction, and later range-delete
+  // them so that the first subcompaction outputs no key-values. In that case
+  // it'll consider making an SST file dedicated to range deletions.
+  for (int i = 0; i < kNumKeysPerFile; ++i) {
+    ASSERT_OK(db_->Put(WriteOptions(), db_->DefaultColumnFamily(), Key(i),
+                       std::string(1024, 'a')));
+  }
+  ASSERT_OK(db_->Flush(FlushOptions()));
+  ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), Key(0),
+                             Key(kNumKeysPerFile)));
+
+  // the above range tombstone can be dropped, so that one alone won't cause a
+  // dedicated file to be opened. We can make one protected by snapshot that
+  // must be considered. Make its range outside the first subcompaction's range
+  // to exercise the tricky part of the code.
+  const Snapshot* snapshot = db_->GetSnapshot();
+  ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(),
+                             Key(kNumKeysPerFile + 1),
+                             Key(kNumKeysPerFile + 2)));
+  ASSERT_OK(db_->Flush(FlushOptions()));
+
+  ASSERT_EQ(kNumFiles, NumTableFilesAtLevel(0));
+  ASSERT_EQ(1, NumTableFilesAtLevel(1));
+
+  db_->EnableAutoCompaction({db_->DefaultColumnFamily()});
+  dbfull()->TEST_WaitForCompact();
+  db_->ReleaseSnapshot(snapshot);
+}
+
+TEST_F(DBRangeDelTest, MemtableBloomFilter) {
+  // regression test for #2743. the range delete tombstones in memtable should
+  // be added even when Get() skips searching due to its prefix bloom filter
+  const int kMemtableSize = 1 << 20;              // 1MB
+  const int kMemtablePrefixFilterSize = 1 << 13;  // 8KB
+  const int kNumKeys = 1000;
+  const int kPrefixLen = 8;
+  Options options = CurrentOptions();
+  options.memtable_prefix_bloom_size_ratio =
+      static_cast<double>(kMemtablePrefixFilterSize) / kMemtableSize;
+  options.prefix_extractor.reset(rocksdb::NewFixedPrefixTransform(kPrefixLen));
+  options.write_buffer_size = kMemtableSize;
+  Reopen(options);
+
+  for (int i = 0; i < kNumKeys; ++i) {
+    ASSERT_OK(Put(Key(i), "val"));
+  }
+  Flush();
+  ASSERT_OK(db_->DeleteRange(WriteOptions(), db_->DefaultColumnFamily(), Key(0),
+                             Key(kNumKeys)));
+  for (int i = 0; i < kNumKeys; ++i) {
+    std::string value;
+    ASSERT_TRUE(db_->Get(ReadOptions(), Key(i), &value).IsNotFound());
+  }
+}
+
+TEST_F(DBRangeDelTest, UnorderedTombstones) {
+  // Regression test for #2752. Range delete tombstones between
+  // different snapshot stripes are not stored in order, so the first
+  // tombstone of each snapshot stripe should be checked as a smallest
+  // candidate.
+  Options options = CurrentOptions();
+  DestroyAndReopen(options);
+
+  auto cf = db_->DefaultColumnFamily();
+
+  ASSERT_OK(db_->Put(WriteOptions(), cf, "a", "a"));
+  ASSERT_OK(db_->Flush(FlushOptions(), cf));
+  ASSERT_EQ(1, NumTableFilesAtLevel(0));
+  ASSERT_OK(dbfull()->TEST_CompactRange(0, nullptr, nullptr));
+  ASSERT_EQ(1, NumTableFilesAtLevel(1));
+
+  ASSERT_OK(db_->DeleteRange(WriteOptions(), cf, "b", "c"));
+  // Hold a snapshot to separate these two delete ranges.
+  auto snapshot = db_->GetSnapshot();
+  ASSERT_OK(db_->DeleteRange(WriteOptions(), cf, "a", "b"));
+  ASSERT_OK(db_->Flush(FlushOptions(), cf));
+  db_->ReleaseSnapshot(snapshot);
+
+  std::vector<std::vector<FileMetaData>> files;
+  dbfull()->TEST_GetFilesMetaData(cf, &files);
+  ASSERT_EQ(1, files[0].size());
+  ASSERT_EQ("a", files[0][0].smallest.user_key());
+  ASSERT_EQ("c", files[0][0].largest.user_key());
+
+  std::string v;
+  auto s = db_->Get(ReadOptions(), "a", &v);
+  ASSERT_TRUE(s.IsNotFound());
+}
+
 #endif  // ROCKSDB_LITE

 }  // namespace rocksdb
@@ -1,7 +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 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
@@ -297,7 +297,7 @@ TEST_F(DBSSTTest, RateLimitedDelete) {
  std::vector<uint64_t> penalties;
  rocksdb::SyncPoint::GetInstance()->SetCallBack(
      "DeleteScheduler::BackgroundEmptyTrash:Wait",
-      [&](void* arg) { penalties.push_back(*(static_cast<int*>(arg))); });
+      [&](void* arg) { penalties.push_back(*(static_cast<uint64_t*>(arg))); });
  rocksdb::SyncPoint::GetInstance()->SetCallBack(
      "InstrumentedCondVar::TimedWaitInternal", [&](void* arg) {
        // Turn timed wait into a simulated sleep
@@ -335,6 +335,7 @@ TEST_F(DBSSTTest, RateLimitedDelete) {
  ASSERT_OK(s);
  options.sst_file_manager->SetDeleteRateBytesPerSecond(rate_bytes_per_sec);
  auto sfm = static_cast<SstFileManagerImpl*>(options.sst_file_manager.get());
+  sfm->delete_scheduler()->TEST_SetMaxTrashDBRatio(1.1);

  ASSERT_OK(TryReopen(options));
  // Create 4 files in L0
@@ -400,6 +401,7 @@ TEST_F(DBSSTTest, DeleteSchedulerMultipleDBPaths) {
      env_, nullptr, trash_dir, rate_bytes_per_sec, false, &s));
  ASSERT_OK(s);
  auto sfm = static_cast<SstFileManagerImpl*>(options.sst_file_manager.get());
+  sfm->delete_scheduler()->TEST_SetMaxTrashDBRatio(1.1);

  DestroyAndReopen(options);

@@ -454,9 +456,14 @@ TEST_F(DBSSTTest, DestroyDBWithRateLimitedDelete) {
      [&](void* arg) { bg_delete_file++; });
  rocksdb::SyncPoint::GetInstance()->EnableProcessing();

+  Status s;
  Options options = CurrentOptions();
  options.disable_auto_compactions = true;
  options.env = env_;
+  std::string trash_dir = test::TmpDir(env_) + "/trash";
+  options.sst_file_manager.reset(
+      NewSstFileManager(env_, nullptr, trash_dir, 0, false, &s));
+  ASSERT_OK(s);
  DestroyAndReopen(options);

  // Create 4 files in L0
@@ -469,15 +476,12 @@ TEST_F(DBSSTTest, DestroyDBWithRateLimitedDelete) {

  // Close DB and destroy it using DeleteScheduler
  Close();
-  std::string trash_dir = test::TmpDir(env_) + "/trash";
-  int64_t rate_bytes_per_sec = 1024 * 1024;  // 1 Mb / Sec
-  Status s;
-  options.sst_file_manager.reset(NewSstFileManager(
-      env_, nullptr, trash_dir, rate_bytes_per_sec, false, &s));
-  ASSERT_OK(s);
-  ASSERT_OK(DestroyDB(dbname_, options));

  auto sfm = static_cast<SstFileManagerImpl*>(options.sst_file_manager.get());
+
+  sfm->SetDeleteRateBytesPerSecond(1024 * 1024);
+  sfm->delete_scheduler()->TEST_SetMaxTrashDBRatio(1.1);
+  ASSERT_OK(DestroyDB(dbname_, options));
  sfm->WaitForEmptyTrash();
  // We have deleted the 4 sst files in the delete_scheduler
  ASSERT_EQ(bg_delete_file, 4);