add <sys/sysmacros.h> to avoid warning with glibc 2.25

Summary: https://github.com/facebook/rocksdb/issues/2152 Closes https://github.com/facebook/rocksdb/pull/2208 Differential Revision: D4945577 Pulled By: lightmark fbshipit-source-id: 4e679150f2c9443d3be0b6008b26b65fabbda75a
Disable warning as error
2026-11-06 11:02:36 -05:00 · 2022-02-07 15:13:01 -08:00 · 2019-11-05 11:09:30 -08:00 · 2019-10-31 14:10:18 -07:00 · 2019-10-31 13:56:50 -07:00 · 2019-10-31 13:51:10 -07:00
439 changed files with 29274 additions and 11538 deletions
@@ -1,4 +1,3 @@
-TARGETS
 make_config.mk

 *.a
@@ -71,3 +70,4 @@ db_logs/
 tp2/
 fbcode/
 fbcode
+buckifier/*.pyc
@@ -27,6 +27,8 @@ env:
  - JOB_NAME=java_test
  # Build ROCKSDB_LITE
  - JOB_NAME=lite_build
+  # Build examples
+  - JOB_NAME=examples

 matrix:
  exclude:
@@ -53,6 +55,7 @@ script:
  - 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
 notifications:
    email:
      - leveldb@fb.com
@@ -41,20 +41,21 @@ endif()

 list(APPEND CMAKE_MODULE_PATH "${CMAKE_SOURCE_DIR}/cmake/modules/")

-if(WIN32)
+if(MSVC)
  include(${CMAKE_CURRENT_SOURCE_DIR}/thirdparty.inc)
 else()
  option(WITH_JEMALLOC "build with JeMalloc" OFF)
-  if(WITH_JEMALLOC)
-    find_package(JeMalloc REQUIRED)
-    add_definitions(-DROCKSDB_JEMALLOC)
-    include_directories(${JEMALLOC_INCLUDE_DIR})
-  endif()
-  if(CMAKE_SYSTEM_NAME STREQUAL "FreeBSD")
+  if(CMAKE_SYSTEM_NAME MATCHES "FreeBSD")
    # FreeBSD has jemaloc as default malloc
-    add_definitions(-DROCKSDB_JEMALLOC)
+    # but it does not have all the jemalloc files in include/...
    set(WITH_JEMALLOC ON)
-  endif()
+  else()
+    if(WITH_JEMALLOC)
+      find_package(JeMalloc REQUIRED)
+      add_definitions(-DROCKSDB_JEMALLOC -DJEMALLOC_NO_DEMANGLE)
+      include_directories(${JEMALLOC_INCLUDE_DIR})
+    endif()
+  endif() 
  option(WITH_SNAPPY "build with SNAPPY" OFF)
  if(WITH_SNAPPY)
    find_package(snappy REQUIRED)
@@ -78,7 +79,7 @@ endif()

 find_package(Git)

-if (GIT_FOUND AND EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/.git")
+if(GIT_FOUND AND EXISTS "${CMAKE_CURRENT_SOURCE_DIR}/.git")
  if(WIN32)
    execute_process(COMMAND $ENV{COMSPEC} /C ${GIT_EXECUTABLE} -C ${CMAKE_CURRENT_SOURCE_DIR} rev-parse HEAD OUTPUT_VARIABLE GIT_SHA)
  else()
@@ -90,7 +91,7 @@ endif()

 string(REGEX REPLACE "[^0-9a-f]+" "" GIT_SHA "${GIT_SHA}")

-if (NOT WIN32)
+if(NOT WIN32)
  execute_process(COMMAND
      "./build_tools/version.sh" "full"
      WORKING_DIRECTORY ${PROJECT_SOURCE_DIR}
@@ -105,18 +106,31 @@ if (NOT WIN32)
  string(STRIP "${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()
+  endif()
+endif()
+
 set(BUILD_VERSION_CC ${CMAKE_BINARY_DIR}/build_version.cc)
 configure_file(util/build_version.cc.in ${BUILD_VERSION_CC} @ONLY)
 add_library(build_version OBJECT ${BUILD_VERSION_CC})
 target_include_directories(build_version PRIVATE
  ${CMAKE_CURRENT_SOURCE_DIR}/util)
-if(WIN32)
-  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /Zi /nologo  /EHsc /GS /Gd /GR /GF /fp:precise /Zc:wchar_t /Zc:forScope /errorReport:queue")
-
+if(MSVC)
+  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /Zi /nologo /EHsc /GS /Gd /GR /GF /fp:precise /Zc:wchar_t /Zc:forScope /errorReport:queue")
  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /FC /d2Zi+ /W3 /wd4127 /wd4800 /wd4996 /wd4351")
 else()
  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -W -Wextra -Wall")
  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wsign-compare -Wshadow -Wno-unused-parameter -Wno-unused-variable -Woverloaded-virtual -Wnon-virtual-dtor -Wno-missing-field-initializers")
+  if(MINGW)
+    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -Wno-format")
+  endif()
  set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -std=c++11")
  if(NOT CMAKE_BUILD_TYPE STREQUAL "Debug")
    set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fno-omit-frame-pointer")
@@ -178,7 +192,7 @@ else()
   set(OPTIMIZE_DEBUG ${OPTIMIZE_DEBUG_DEFAULT})
 endif()

-if(WIN32)
+if(MSVC)
  if((${OPTIMIZE_DEBUG} EQUAL 1))
    message(STATUS "Debug optimization is enabled")
    set(CMAKE_CXX_FLAGS_DEBUG "/Oxt /MDd")
@@ -226,6 +240,9 @@ elseif(CMAKE_SYSTEM_NAME MATCHES "Android")
  add_definitions(-DOS_ANDROID)
 elseif(CMAKE_SYSTEM_NAME MATCHES "Windows")
  add_definitions(-DWIN32 -DOS_WIN -D_MBCS -DWIN64 -DNOMINMAX)
+  if(MINGW)
+    add_definitions(-D_WIN32_WINNT=_WIN32_WINNT_VISTA)
+  endif()
 endif()

 if(NOT WIN32)
@@ -265,7 +282,9 @@ add_subdirectory(third-party/gtest-1.7.0/fused-src/gtest)
 # Main library source code

 set(SOURCES
-        db/auto_roll_logger.cc
+        cache/clock_cache.cc
+        cache/lru_cache.cc
+        cache/sharded_cache.cc
        db/builder.cc
        db/c.cc
        db/column_family.cc
@@ -274,19 +293,23 @@ set(SOURCES
        db/compaction_iterator.cc
        db/compaction_job.cc
        db/compaction_picker.cc
+        db/compaction_picker_universal.cc
        db/convenience.cc
-        db/dbformat.cc
        db/db_filesnapshot.cc
        db/db_impl.cc
+        db/db_impl_write.cc
+        db/db_impl_compaction_flush.cc
+        db/db_impl_files.cc
+        db/db_impl_open.cc
        db/db_impl_debug.cc
        db/db_impl_experimental.cc
        db/db_impl_readonly.cc
        db/db_info_dumper.cc
        db/db_iter.cc
+        db/dbformat.cc
        db/event_helpers.cc
-        db/external_sst_file_ingestion_job.cc
        db/experimental.cc
-        db/filename.cc
+        db/external_sst_file_ingestion_job.cc
        db/file_indexer.cc
        db/flush_job.cc
        db/flush_scheduler.cc
@@ -296,7 +319,6 @@ set(SOURCES
        db/log_writer.cc
        db/managed_iterator.cc
        db/memtable.cc
-        db/memtable_allocator.cc
        db/memtable_list.cc
        db/merge_helper.cc
        db/merge_operator.cc
@@ -314,11 +336,33 @@ set(SOURCES
        db/write_batch_base.cc
        db/write_controller.cc
        db/write_thread.cc
+        env/env.cc
+        env/env_chroot.cc
+        env/env_hdfs.cc
+        env/memenv.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
+        monitoring/histogram.cc
+        monitoring/histogram_windowing.cc
+        monitoring/instrumented_mutex.cc
+        monitoring/iostats_context.cc
+        monitoring/perf_context.cc
+        monitoring/perf_level.cc
+        monitoring/statistics.cc
+        monitoring/thread_status_impl.cc
+        monitoring/thread_status_updater.cc
+        monitoring/thread_status_util.cc
+        monitoring/thread_status_util_debug.cc
+        options/cf_options.cc
+        options/db_options.cc
+        options/options.cc
+        options/options_helper.cc
+        options/options_parser.cc
+        options/options_sanity_check.cc
        port/stack_trace.cc
        table/adaptive_table_factory.cc
        table/block.cc
@@ -339,64 +383,44 @@ set(SOURCES
        table/index_builder.cc
        table/iterator.cc
        table/merging_iterator.cc
-        table/sst_file_writer.cc
        table/meta_blocks.cc
        table/partitioned_filter_block.cc
+        table/persistent_cache_helper.cc
        table/plain_table_builder.cc
        table/plain_table_factory.cc
        table/plain_table_index.cc
        table/plain_table_key_coding.cc
        table/plain_table_reader.cc
-        table/persistent_cache_helper.cc
+        table/sst_file_writer.cc
        table/table_properties.cc
        table/two_level_iterator.cc
-        tools/sst_dump_tool.cc
        tools/db_bench_tool.cc
        tools/dump/db_dump_tool.cc
+        tools/ldb_cmd.cc
+        tools/ldb_tool.cc
+        tools/sst_dump_tool.cc
        util/arena.cc
+        util/auto_roll_logger.cc
        util/bloom.cc
-        util/cf_options.cc
-        util/clock_cache.cc
        util/coding.cc
        util/compaction_job_stats_impl.cc
        util/comparator.cc
        util/concurrent_arena.cc
        util/crc32c.cc
-        util/db_options.cc
        util/delete_scheduler.cc
        util/dynamic_bloom.cc
-        util/env.cc
-        util/env_chroot.cc
-        util/env_hdfs.cc
        util/event_logger.cc
-        util/file_util.cc
        util/file_reader_writer.cc
-        util/sst_file_manager_impl.cc
+        util/file_util.cc
+        util/filename.cc
        util/filter_policy.cc
        util/hash.cc
-        util/histogram.cc
-        util/histogram_windowing.cc
-        util/instrumented_mutex.cc
-        util/iostats_context.cc
-        
-        util/lru_cache.cc
-        tools/ldb_cmd.cc
-        tools/ldb_tool.cc
-        util/logging.cc
        util/log_buffer.cc
-        util/memenv.cc
        util/murmurhash.cc
-        util/options.cc
-        util/options_helper.cc
-        util/options_parser.cc
-        util/options_sanity_check.cc
-        util/perf_context.cc
-        util/perf_level.cc
        util/random.cc
        util/rate_limiter.cc
-        util/sharded_cache.cc
        util/slice.cc
-        util/statistics.cc
+        util/sst_file_manager_impl.cc
        util/status.cc
        util/status_message.cc
        util/string_util.cc
@@ -404,29 +428,35 @@ set(SOURCES
        util/testutil.cc
        util/thread_local.cc
        util/threadpool_imp.cc
-        util/thread_status_impl.cc
-        util/thread_status_updater.cc
-        util/thread_status_util.cc
-        util/thread_status_util_debug.cc
        util/transaction_test_util.cc
        util/xxhash.cc
        utilities/backupable/backupable_db.cc
        utilities/blob_db/blob_db.cc
+        utilities/blob_db/blob_db_impl.cc
+        utilities/blob_db/blob_db_options_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/checkpoint/checkpoint.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/document/document_db.cc
        utilities/document/json_document.cc
        utilities/document/json_document_builder.cc
        utilities/env_mirror.cc
+        utilities/env_timed.cc
        utilities/geodb/geodb_impl.cc
        utilities/leveldb_options/leveldb_options.cc
        utilities/lua/rocks_lua_compaction_filter.cc
        utilities/memory/memory_util.cc
+        utilities/merge_operators/max.cc
+        utilities/merge_operators/put.cc
        utilities/merge_operators/string_append/stringappend.cc
        utilities/merge_operators/string_append/stringappend2.cc
-        utilities/merge_operators/put.cc
-        utilities/merge_operators/max.cc
        utilities/merge_operators/uint64add.cc
        utilities/option_change_migration/option_change_migration.cc
        utilities/options/options_util.cc
@@ -439,20 +469,17 @@ set(SOURCES
        utilities/simulator_cache/sim_cache.cc
        utilities/spatialdb/spatial_db.cc
        utilities/table_properties_collectors/compact_on_deletion_collector.cc
-        utilities/transactions/optimistic_transaction_impl.cc
        utilities/transactions/optimistic_transaction_db_impl.cc
+        utilities/transactions/optimistic_transaction_impl.cc
        utilities/transactions/transaction_base.cc
-        utilities/transactions/transaction_impl.cc
        utilities/transactions/transaction_db_impl.cc
        utilities/transactions/transaction_db_mutex_impl.cc
+        utilities/transactions/transaction_impl.cc
        utilities/transactions/transaction_lock_mgr.cc
        utilities/transactions/transaction_util.cc
        utilities/ttl/db_ttl_impl.cc
        utilities/write_batch_with_index/write_batch_with_index.cc
        utilities/write_batch_with_index/write_batch_with_index_internal.cc
-        utilities/col_buf_encoder.cc
-        utilities/col_buf_decoder.cc
-        utilities/column_aware_encoding_util.cc
        $<TARGET_OBJECTS:build_version>)

 if(WIN32)
@@ -467,8 +494,8 @@ if(WIN32)
 else()
  list(APPEND SOURCES
    port/port_posix.cc
-    util/env_posix.cc
-    util/io_posix.cc)
+    env/env_posix.cc
+    env/io_posix.cc)
 endif()

 if(WIN32)
@@ -505,18 +532,18 @@ add_library(${ROCKSDB_STATIC_LIB} STATIC ${SOURCES})
 target_link_libraries(${ROCKSDB_STATIC_LIB}
  ${THIRDPARTY_LIBS} ${SYSTEM_LIBS})

-if(WIN32)
-  set_target_properties(${ROCKSDB_STATIC_LIB} PROPERTIES
-    COMPILE_FLAGS "/Fd${CMAKE_CFG_INTDIR}/${ROCKSDB_STATIC_LIB}.pdb")
-endif()
-
 if(WIN32)
  add_library(${ROCKSDB_IMPORT_LIB} SHARED ${SOURCES})
  target_link_libraries(${ROCKSDB_IMPORT_LIB}
    ${THIRDPARTY_LIBS} ${SYSTEM_LIBS})
  set_target_properties(${ROCKSDB_IMPORT_LIB} PROPERTIES
-    COMPILE_FLAGS "-DROCKSDB_DLL -DROCKSDB_LIBRARY_EXPORTS /Fd${CMAKE_CFG_INTDIR}/${ROCKSDB_IMPORT_LIB}.pdb")
-else()
+    COMPILE_DEFINITIONS "ROCKSDB_DLL;ROCKSDB_LIBRARY_EXPORTS")
+  if(MSVC)
+    set_target_properties(${ROCKSDB_STATIC_LIB} PROPERTIES
+      COMPILE_FLAGS "/Fd${CMAKE_CFG_INTDIR}/${ROCKSDB_STATIC_LIB}.pdb")
+    set_target_properties(${ROCKSDB_IMPORT_LIB} PROPERTIES
+      COMPILE_FLAGS "/Fd${CMAKE_CFG_INTDIR}/${ROCKSDB_IMPORT_LIB}.pdb")
+  endif()
 endif()

 option(WITH_JNI "build with JNI" OFF)
@@ -528,60 +555,61 @@ else()
 endif()

 set(TESTS
-        db/auto_roll_logger_test.cc
+        cache/cache_test.cc
+        cache/lru_cache_test.cc
        db/column_family_test.cc
        db/compact_files_test.cc
        db/compaction_iterator_test.cc
-        db/compaction_job_test.cc
        db/compaction_job_stats_test.cc
+        db/compaction_job_test.cc
        db/compaction_picker_test.cc
        db/comparator_db_test.cc
        db/corruption_test.cc
        db/cuckoo_table_db_test.cc
        db/db_basic_test.cc
+        db/db_block_cache_test.cc
+        db/db_bloom_filter_test.cc
        db/db_compaction_filter_test.cc
        db/db_compaction_test.cc
        db/db_dynamic_level_test.cc
+        db/db_flush_test.cc
        db/db_inplace_update_test.cc
+        db/db_io_failure_test.cc
        db/db_iter_test.cc
+        db/db_iterator_test.cc
        db/db_log_iter_test.cc
        db/db_memtable_test.cc
        db/db_merge_operator_test.cc
        db/db_options_test.cc
        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_block_cache_test.cc
-        db/db_bloom_filter_test.cc
-        db/db_flush_test.cc
-        db/db_iterator_test.cc
-        db/db_range_del_test.cc
-        db/db_sst_test.cc
-        db/external_sst_file_basic_test.cc
-        db/external_sst_file_test.cc
        db/db_universal_compaction_test.cc
        db/db_wal_test.cc
        db/dbformat_test.cc
        db/deletefile_test.cc
+        db/external_sst_file_basic_test.cc
+        db/external_sst_file_test.cc
        db/fault_injection_test.cc
        db/file_indexer_test.cc
        db/filename_test.cc
        db/flush_job_test.cc
-        db/inlineskiplist_test.cc
        db/listener_test.cc
        db/log_test.cc
        db/manual_compaction_test.cc
        db/memtable_list_test.cc
-        db/merge_test.cc
        db/merge_helper_test.cc
+        db/merge_test.cc
        db/options_file_test.cc
        db/perf_context_test.cc
        db/plain_table_db_test.cc
        db/prefix_test.cc
        db/repair_test.cc
-        db/skiplist_test.cc
        db/table_properties_collector_test.cc
        db/version_builder_test.cc
        db/version_edit_test.cc
@@ -590,7 +618,16 @@ set(TESTS
        db/write_batch_test.cc
        db/write_callback_test.cc
        db/write_controller_test.cc
-        db/db_io_failure_test.cc
+        env/env_basic_test.cc
+        env/env_test.cc
+        env/mock_env_test.cc
+        memtable/inlineskiplist_test.cc
+        memtable/skiplist_test.cc
+        monitoring/histogram_test.cc
+        monitoring/iostats_context_test.cc
+        monitoring/statistics_test.cc
+        options/options_settable_test.cc
+        options/options_test.cc
        table/block_based_filter_block_test.cc
        table/block_test.cc
        table/cuckoo_table_builder_test.cc
@@ -602,40 +639,33 @@ set(TESTS
        tools/reduce_levels_test.cc
        tools/sst_dump_test.cc
        util/arena_test.cc
+        util/auto_roll_logger_test.cc
        util/autovector_test.cc
        util/bloom_test.cc
-        util/cache_test.cc
        util/coding_test.cc
        util/crc32c_test.cc
        util/delete_scheduler_test.cc
        util/dynamic_bloom_test.cc
-        util/env_basic_test.cc
-        util/env_test.cc
        util/event_logger_test.cc
-        util/filelock_test.cc
        util/file_reader_writer_test.cc
+        util/filelock_test.cc
        util/heap_test.cc
-        util/histogram_test.cc
-        util/iostats_context_test.cc
-        util/lru_cache_test.cc
-        util/mock_env_test.cc
-        util/options_settable_test.cc
-        util/options_test.cc
        util/rate_limiter_test.cc
        util/slice_transform_test.cc
-        util/statistics_test.cc
+        util/timer_queue_test.cc
        util/thread_list_test.cc
        util/thread_local_test.cc
        utilities/backupable/backupable_db_test.cc
-        utilities/blob_db/blob_db_test.cc
        utilities/checkpoint/checkpoint_test.cc
+        utilities/column_aware_encoding_test.cc
        utilities/date_tiered/date_tiered_test.cc
        utilities/document/document_db_test.cc
        utilities/document/json_document_test.cc
-        utilities/object_registry_test.cc
        utilities/geodb/geodb_test.cc
+        utilities/lua/rocks_lua_test.cc
        utilities/memory/memory_test.cc
        utilities/merge_operators/string_append/stringappend_test.cc
+        utilities/object_registry_test.cc
        utilities/option_change_migration/option_change_migration_test.cc
        utilities/options/options_util_test.cc
        utilities/persistent_cache/hash_table_test.cc
@@ -647,18 +677,16 @@ set(TESTS
        utilities/transactions/transaction_test.cc
        utilities/ttl/ttl_test.cc
        utilities/write_batch_with_index/write_batch_with_index_test.cc
-        utilities/column_aware_encoding_test.cc
-        utilities/lua/rocks_lua_test.cc
 )
 if(WITH_LIBRADOS)
  list(APPEND TESTS utilities/env_librados_test.cc)
 endif()

 set(BENCHMARKS
+  cache/cache_bench.cc
+  memtable/memtablerep_bench.cc
  tools/db_bench.cc
  table/table_reader_bench.cc
-  util/cache_bench.cc
-  db/memtablerep_bench.cc
  utilities/column_aware_encoding_exp.cc
  utilities/persistent_cache/hash_table_bench.cc)
 add_library(testharness OBJECT util/testharness.cc)
@@ -673,17 +701,17 @@ 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/mock_env.cc
    util/fault_injection_test_env.cc
-    util/thread_status_updater_debug.cc
 )
 # test utilities are only build in debug
 enable_testing()
 add_custom_target(check COMMAND ${CMAKE_CTEST_COMMAND})
 set(TESTUTILLIB testutillib${ARTIFACT_SUFFIX})
 add_library(${TESTUTILLIB} STATIC ${TESTUTIL_SOURCE})
-if(WIN32)
+if(MSVC)
  set_target_properties(${TESTUTILLIB} PROPERTIES COMPILE_FLAGS "/Fd${CMAKE_CFG_INTDIR}/testutillib${ARTIFACT_SUFFIX}.pdb")
 endif()
 set_target_properties(${TESTUTILLIB}
@@ -731,7 +759,7 @@ endforeach(sourcefile ${C_TEST_EXES})
 add_subdirectory(tools)

 # Installation and packaging for Linux
-if (NOT WIN32)
+if(NOT WIN32)
 install(TARGETS ${ROCKSDB_STATIC_LIB} COMPONENT devel ARCHIVE DESTINATION lib64)
 install(TARGETS ${ROCKSDB_SHARED_LIB} COMPONENT runtime DESTINATION lib64)
 install(DIRECTORY "${PROJECT_SOURCE_DIR}/include/rocksdb/"
@@ -1,7 +1,78 @@
 # Rocksdb Change Log
-## 5.3.7
+## 5.4.12 (08/21/2017)
+* Statistics::Reset() to reset user stats.
+
+## 5.4.11 (08/17/2017)
+### Bug Fixes
+* Fix a bug in PosixWritableFile::Close()
+
+## 5.4.10 (08/12/2017)
+### Bug Fixes
+* Fix incorrect dropping of deletions during intra-L0 compaction.
+
+## 5.4.9 (07/20/2017)
 * Change license to Apache + GPLv2

+## 5.4.8 (07/13/2017)
+No change from 5.4.7.
+
+## 5.4.7 (06/28/2017)
+### Bug Fixes
+* Shouldn't ignore return value of fsync() in flush.
+
+## 5.4.6 (05/30/2017)
+### Bug Fixes
+* Fix db_bench with lite build.
+
+## 5.4.5 (05/19/2017)
+### Bug Fixes
+* Facility for cross-building RocksJava using Docker.
+* Build and link with ZStd when creating the static RocksJava build.
+* Fix snprintf's usage to be cross-platform.
+* Fix build error with blob DB.
+* Fix readamp test type inconsistency.
+
+## 5.4.4 (05/11/2017)
+### New Features
+* Added DB::CreateColumnFamilie() and DB::DropColumnFamilies() to bulk create/drop column families.
+
+### Bug Fixes
+* Fix a manual / auto compaction data race.
+* Fix CentOS 5 cross-building of RocksJava.
+
+## 5.4.3 (05/10/2017)
+### Bug Fixes
+* Fix memory alignment with logical sector size.
+* Fix alignment in ReadaheadRandomAccessFile.
+* Fix bias with read amplification stats (READ_AMP_ESTIMATE_USEFUL_BYTES and READ_AMP_TOTAL_READ_BYTES).
+
+## 5.4.2 (05/08/2017)
+### Bug Fixes
+* Fix huge fallocate calls fail and make XFS unhappy.
+
+## 5.4.1 (05/02/2017)
+### Bug Fxies
+* Fix WriteBatchWithIndex address use after scope error.
+* Fix WritableFile buffer size in direct IO.
+* Add prefetch to PosixRandomAccessFile in buffered io.
+* Fix PinnableSlice access invalid address when row cache is enabled.
+
+## 5.4.0 (04/11/2017)
+### Public API Change
+* 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.
+* Introduce level-based L0->L0 compactions to reduce file count, so write delays are incurred less often.
+* (Experimental) Partitioning filters which creates an index on the partitions. The feature can be enabled by setting partition_filters when using kFullFilter. Currently the feature also requires two-level indexing to be enabled. Number of partitions is the same as the number of partitions for indexes, which is controlled by metadata_block_size.
+* DB::ResetStats() to reset internal stats.
+
 ## 5.3.0 (03/08/2017)
 ### Public API Change
 * Remove disableDataSync option.
@@ -9,9 +80,6 @@
 * Remove option min_partial_merge_operands. Partial merge operands will always be merged in flush or compaction if there are more than one.
 * Remove option verify_checksums_in_compaction. Compaction will always verify checksum.

-### New Features
-* Memtable flush can be avoided during checkpoint creation if total log file size is smaller than a threshold specified by the user.
-
 ### Bug Fixes
 * Fix the bug that iterator may skip keys

@@ -1,6 +1,6 @@
 ## Compilation

-**Important**: If you plan to run RocksDB in production, don't compile using default 
+**Important**: If you plan to run RocksDB in production, don't compile using default
 `make` or `make all`. That will compile RocksDB in debug mode, which is much slower
 than release mode.

@@ -22,7 +22,8 @@ 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`
+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`.

 ## Dependencies

@@ -58,10 +59,14 @@ your make commands, like this: `PORTABLE=1 make static_lib`
    * Install gflags:

              git clone https://github.com/gflags/gflags.git
-              git checkout v2.0
              cd gflags
+              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.
+
    * Install snappy:

              wget https://github.com/google/snappy/releases/download/1.1.4/snappy-1.1.4.tar.gz
@@ -82,7 +87,7 @@ your make commands, like this: `PORTABLE=1 make static_lib`
    * Install zstandard:

             wget https://github.com/facebook/zstd/archive/v1.1.3.tar.gz
-             mv v1.1.3.tar.gz zstandard-1.1.3.tar.gz
+             mv v1.1.3.tar.gz zstd-1.1.3.tar.gz
             tar zxvf zstd-1.1.3.tar.gz
             cd zstd-1.1.3
             make && sudo make install
@@ -204,15 +204,16 @@ endif

 ifndef DISABLE_JEMALLOC
 	ifdef JEMALLOC
-		PLATFORM_CXXFLAGS += "-DROCKSDB_JEMALLOC"
-		PLATFORM_CCFLAGS +=  "-DROCKSDB_JEMALLOC"
+		PLATFORM_CXXFLAGS += -DROCKSDB_JEMALLOC -DJEMALLOC_NO_DEMANGLE
+		PLATFORM_CCFLAGS  += -DROCKSDB_JEMALLOC -DJEMALLOC_NO_DEMANGLE
 	endif
 	EXEC_LDFLAGS := $(JEMALLOC_LIB) $(EXEC_LDFLAGS)
 	PLATFORM_CXXFLAGS += $(JEMALLOC_INCLUDE)
 	PLATFORM_CCFLAGS += $(JEMALLOC_INCLUDE)
 endif

-export GTEST_THROW_ON_FAILURE=1 GTEST_HAS_EXCEPTIONS=1
+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)
@@ -333,6 +334,7 @@ TESTS = \
 	db_io_failure_test \
 	db_properties_test \
 	db_table_properties_test \
+	db_statistics_test \
 	autovector_test \
 	cleanable_test \
 	column_family_test \
@@ -349,6 +351,7 @@ TESTS = \
 	file_reader_writer_test \
 	block_based_filter_block_test \
 	full_filter_block_test \
+	partitioned_filter_block_test \
 	hash_table_test \
 	histogram_test \
 	log_test \
@@ -372,7 +375,6 @@ TESTS = \
 	ttl_test \
 	date_tiered_test \
 	backupable_db_test \
-	blob_db_test \
 	document_db_test \
 	json_document_test \
 	sim_cache_test \
@@ -393,6 +395,7 @@ 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 \
@@ -420,6 +423,7 @@ TESTS = \
 	lru_cache_test \
 	object_registry_test \
 	repair_test \
+	env_timed_test \

 PARALLEL_TEST = \
 	backupable_db_test \
@@ -886,20 +890,20 @@ $(TOOLS_LIBRARY): $(BENCH_LIB_SOURCES:.cc=.o) $(TOOL_LIB_SOURCES:.cc=.o) $(LIB_S
 	$(AM_V_AR)rm -f $@
 	$(AM_V_at)$(AR) $(ARFLAGS) $@ $^

-librocksdb_env_basic_test.a: util/env_basic_test.o $(LIBOBJECTS) $(TESTHARNESS)
+librocksdb_env_basic_test.a: env/env_basic_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_V_AR)rm -f $@
 	$(AM_V_at)$(AR) $(ARFLAGS) $@ $^

 db_bench: tools/db_bench.o $(BENCHTOOLOBJECTS)
 	$(AM_LINK)

-cache_bench: util/cache_bench.o $(LIBOBJECTS) $(TESTUTIL)
+cache_bench: cache/cache_bench.o $(LIBOBJECTS) $(TESTUTIL)
 	$(AM_LINK)

 persistent_cache_bench: utilities/persistent_cache/persistent_cache_bench.o $(LIBOBJECTS) $(TESTUTIL)
 	$(AM_LINK)

-memtablerep_bench: db/memtablerep_bench.o $(LIBOBJECTS) $(TESTUTIL)
+memtablerep_bench: memtable/memtablerep_bench.o $(LIBOBJECTS) $(TESTUTIL)
 	$(AM_LINK)

 db_stress: tools/db_stress.o $(LIBOBJECTS) $(TESTUTIL)
@@ -935,7 +939,7 @@ dynamic_bloom_test: util/dynamic_bloom_test.o $(LIBOBJECTS) $(TESTHARNESS)
 c_test: db/c_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-cache_test: util/cache_test.o $(LIBOBJECTS) $(TESTHARNESS)
+cache_test: cache/cache_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 coding_test: util/coding_test.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -953,7 +957,7 @@ redis_test: utilities/redis/redis_lists_test.o $(LIBOBJECTS) $(TESTHARNESS)
 hash_table_test: utilities/persistent_cache/hash_table_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-histogram_test: util/histogram_test.o $(LIBOBJECTS) $(TESTHARNESS)
+histogram_test: monitoring/histogram_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 thread_local_test: util/thread_local_test.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1019,6 +1023,9 @@ 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)
+
 external_sst_file_basic_test: db/external_sst_file_basic_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

@@ -1067,9 +1074,6 @@ 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)

@@ -1088,6 +1092,9 @@ spatial_db_test: utilities/spatialdb/spatial_db_test.o $(LIBOBJECTS) $(TESTHARNE
 env_mirror_test: utilities/env_mirror_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

+env_timed_test: utilities/env_timed_test.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 ifdef ROCKSDB_USE_LIBRADOS
 env_librados_test: utilities/env_librados_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_V_CCLD)$(CXX) $^ $(EXEC_LDFLAGS) -o $@ $(LDFLAGS) $(COVERAGEFLAGS)
@@ -1126,10 +1133,10 @@ wal_manager_test: db/wal_manager_test.o $(LIBOBJECTS) $(TESTHARNESS)
 dbformat_test: db/dbformat_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-env_basic_test: util/env_basic_test.o $(LIBOBJECTS) $(TESTHARNESS)
+env_basic_test: env/env_basic_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-env_test: util/env_test.o $(LIBOBJECTS) $(TESTHARNESS)
+env_test: env/env_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 fault_injection_test: db/fault_injection_test.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1153,6 +1160,9 @@ block_based_filter_block_test: table/block_based_filter_block_test.o $(LIBOBJECT
 full_filter_block_test: table/full_filter_block_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

+partitioned_filter_block_test: table/partitioned_filter_block_test.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(AM_LINK)
+
 log_test: db/log_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

@@ -1165,10 +1175,10 @@ table_test: table/table_test.o $(LIBOBJECTS) $(TESTHARNESS)
 block_test: table/block_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-inlineskiplist_test: db/inlineskiplist_test.o $(LIBOBJECTS) $(TESTHARNESS)
+inlineskiplist_test: memtable/inlineskiplist_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-skiplist_test: db/skiplist_test.o $(LIBOBJECTS) $(TESTHARNESS)
+skiplist_test: memtable/skiplist_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 version_edit_test: db/version_edit_test.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1243,10 +1253,10 @@ thread_list_test: util/thread_list_test.o $(LIBOBJECTS) $(TESTHARNESS)
 compact_files_test: db/compact_files_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-options_test: util/options_test.o $(LIBOBJECTS) $(TESTHARNESS)
+options_test: options/options_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-options_settable_test: util/options_settable_test.o $(LIBOBJECTS) $(TESTHARNESS)
+options_settable_test: options/options_settable_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 options_util_test: utilities/options/options_util_test.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1258,6 +1268,9 @@ 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)

@@ -1267,7 +1280,7 @@ column_aware_encoding_test: utilities/column_aware_encoding_test.o $(TESTHARNESS
 optimistic_transaction_test: utilities/transactions/optimistic_transaction_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-mock_env_test : util/mock_env_test.o $(LIBOBJECTS) $(TESTHARNESS)
+mock_env_test : env/mock_env_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 manual_compaction_test: db/manual_compaction_test.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1276,7 +1289,7 @@ manual_compaction_test: db/manual_compaction_test.o $(LIBOBJECTS) $(TESTHARNESS)
 filelock_test: util/filelock_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-auto_roll_logger_test: db/auto_roll_logger_test.o $(LIBOBJECTS) $(TESTHARNESS)
+auto_roll_logger_test: util/auto_roll_logger_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 memtable_list_test: db/memtable_list_test.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1306,16 +1319,16 @@ ldb_cmd_test: tools/ldb_cmd_test.o $(LIBOBJECTS) $(TESTHARNESS)
 ldb: tools/ldb.o $(LIBOBJECTS)
 	$(AM_LINK)

-iostats_context_test: util/iostats_context_test.o $(LIBOBJECTS) $(TESTHARNESS)
+iostats_context_test: monitoring/iostats_context_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_V_CCLD)$(CXX) $^ $(EXEC_LDFLAGS) -o $@ $(LDFLAGS)

 persistent_cache_test: utilities/persistent_cache/persistent_cache_test.o  db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-statistics_test: util/statistics_test.o $(LIBOBJECTS) $(TESTHARNESS)
+statistics_test: monitoring/statistics_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

-lru_cache_test: util/lru_cache_test.o $(LIBOBJECTS) $(TESTHARNESS)
+lru_cache_test: cache/lru_cache_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(AM_LINK)

 lua_test: utilities/lua/rocks_lua_test.o db/db_test_util.o $(LIBOBJECTS) $(TESTHARNESS)
@@ -1371,7 +1384,12 @@ ifeq ($(PLATFORM), OS_SOLARIS)
 else
 	ARCH := $(shell getconf LONG_BIT)
 endif
-ROCKSDBJNILIB = librocksdbjni-linux$(ARCH).so
+
+ifeq (,$(findstring ppc,$(MACHINE)))
+        ROCKSDBJNILIB = librocksdbjni-linux$(ARCH).so
+else
+        ROCKSDBJNILIB = librocksdbjni-linux-$(MACHINE).so
+endif
 ROCKSDB_JAR = rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-linux$(ARCH).jar
 ROCKSDB_JAR_ALL = rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH).jar
 ROCKSDB_JAVADOCS_JAR = rocksdbjni-$(ROCKSDB_MAJOR).$(ROCKSDB_MINOR).$(ROCKSDB_PATCH)-javadoc.jar
@@ -0,0 +1,477 @@
+REPO_PATH = "internal_repo_rocksdb/repo/"
+BUCK_BINS = "buck-out/gen/" + REPO_PATH
+TEST_RUNNER = REPO_PATH + "buckifier/rocks_test_runner.sh"
+rocksdb_compiler_flags = [
+  "-msse",
+  "-msse4.2",
+  "-fno-builtin-memcmp",
+  "-DROCKSDB_PLATFORM_POSIX",
+  "-DROCKSDB_LIB_IO_POSIX",
+  "-DROCKSDB_FALLOCATE_PRESENT",
+  "-DROCKSDB_MALLOC_USABLE_SIZE",
+  "-DOS_LINUX",
+  # Flags to enable libs we include
+  "-DSNAPPY",
+  "-DZLIB",
+  "-DBZIP2",
+  "-DLZ4",
+  "-DZSTD",
+  "-DGFLAGS=gflags",
+  "-DNUMA",
+  "-DTBB",
+  # Needed to compile in fbcode
+  "-Wno-expansion-to-defined",
+]
+
+rocksdb_external_deps = [
+  ('bzip2', None, 'bz2'),
+  ('snappy', None, "snappy"),
+  ('zlib', None, 'z'),
+  ('gflags', None, 'gflags'),
+  ('lz4', None, 'lz4'),
+  ('zstd', None),
+  ('tbb', None),
+  ("numa", "2.0.8", "numa"),
+  ("googletest", None, "gtest"),
+]
+
+rocksdb_preprocessor_flags = [
+  # Directories with files for #include
+  "-I" + REPO_PATH + "include/",
+  "-I" + REPO_PATH,
+]
+
+cpp_library(
+    name = "rocksdb_lib",
+    headers = AutoHeaders.RECURSIVE_GLOB,
+    srcs = [
+      "cache/clock_cache.cc",
+      "cache/lru_cache.cc",
+      "cache/sharded_cache.cc",
+      "db/builder.cc",
+      "db/c.cc",
+      "db/column_family.cc",
+      "db/compacted_db_impl.cc",
+      "db/compaction.cc",
+      "db/compaction_iterator.cc",
+      "db/compaction_job.cc",
+      "db/compaction_picker.cc",
+      "db/compaction_picker_universal.cc",
+      "db/convenience.cc",
+      "db/db_filesnapshot.cc",
+      "db/db_impl.cc",
+      "db/db_impl_write.cc",
+      "db/db_impl_compaction_flush.cc",
+      "db/db_impl_files.cc",
+      "db/db_impl_open.cc",
+      "db/db_impl_debug.cc",
+      "db/db_impl_experimental.cc",
+      "db/db_impl_readonly.cc",
+      "db/db_info_dumper.cc",
+      "db/db_iter.cc",
+      "db/dbformat.cc",
+      "db/event_helpers.cc",
+      "db/experimental.cc",
+      "db/external_sst_file_ingestion_job.cc",
+      "db/file_indexer.cc",
+      "db/flush_job.cc",
+      "db/flush_scheduler.cc",
+      "db/forward_iterator.cc",
+      "db/internal_stats.cc",
+      "db/log_reader.cc",
+      "db/log_writer.cc",
+      "db/managed_iterator.cc",
+      "db/memtable.cc",
+      "db/memtable_list.cc",
+      "db/merge_helper.cc",
+      "db/merge_operator.cc",
+      "db/range_del_aggregator.cc",
+      "db/repair.cc",
+      "db/snapshot_impl.cc",
+      "db/table_cache.cc",
+      "db/table_properties_collector.cc",
+      "db/transaction_log_impl.cc",
+      "db/version_builder.cc",
+      "db/version_edit.cc",
+      "db/version_set.cc",
+      "db/wal_manager.cc",
+      "db/write_batch.cc",
+      "db/write_batch_base.cc",
+      "db/write_controller.cc",
+      "db/write_thread.cc",
+      "env/env.cc",
+      "env/env_chroot.cc",
+      "env/env_hdfs.cc",
+      "env/env_posix.cc",
+      "env/io_posix.cc",
+      "env/memenv.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",
+      "monitoring/histogram.cc",
+      "monitoring/histogram_windowing.cc",
+      "monitoring/instrumented_mutex.cc",
+      "monitoring/iostats_context.cc",
+      "monitoring/perf_context.cc",
+      "monitoring/perf_level.cc",
+      "monitoring/statistics.cc",
+      "monitoring/thread_status_impl.cc",
+      "monitoring/thread_status_updater.cc",
+      "monitoring/thread_status_updater_debug.cc",
+      "monitoring/thread_status_util.cc",
+      "monitoring/thread_status_util_debug.cc",
+      "options/cf_options.cc",
+      "options/db_options.cc",
+      "options/options.cc",
+      "options/options_helper.cc",
+      "options/options_parser.cc",
+      "options/options_sanity_check.cc",
+      "port/port_posix.cc",
+      "port/stack_trace.cc",
+      "table/adaptive_table_factory.cc",
+      "table/block.cc",
+      "table/block_based_filter_block.cc",
+      "table/block_based_table_builder.cc",
+      "table/block_based_table_factory.cc",
+      "table/block_based_table_reader.cc",
+      "table/block_builder.cc",
+      "table/block_prefix_index.cc",
+      "table/bloom_block.cc",
+      "table/cuckoo_table_builder.cc",
+      "table/cuckoo_table_factory.cc",
+      "table/cuckoo_table_reader.cc",
+      "table/flush_block_policy.cc",
+      "table/format.cc",
+      "table/full_filter_block.cc",
+      "table/get_context.cc",
+      "table/index_builder.cc",
+      "table/iterator.cc",
+      "table/merging_iterator.cc",
+      "table/meta_blocks.cc",
+      "table/partitioned_filter_block.cc",
+      "table/persistent_cache_helper.cc",
+      "table/plain_table_builder.cc",
+      "table/plain_table_factory.cc",
+      "table/plain_table_index.cc",
+      "table/plain_table_key_coding.cc",
+      "table/plain_table_reader.cc",
+      "table/sst_file_writer.cc",
+      "table/table_properties.cc",
+      "table/two_level_iterator.cc",
+      "tools/dump/db_dump_tool.cc",
+      "util/arena.cc",
+      "util/auto_roll_logger.cc",
+      "util/bloom.cc",
+      "util/build_version.cc",
+      "util/coding.cc",
+      "util/compaction_job_stats_impl.cc",
+      "util/comparator.cc",
+      "util/concurrent_arena.cc",
+      "util/crc32c.cc",
+      "util/delete_scheduler.cc",
+      "util/dynamic_bloom.cc",
+      "util/event_logger.cc",
+      "util/file_reader_writer.cc",
+      "util/file_util.cc",
+      "util/filename.cc",
+      "util/filter_policy.cc",
+      "util/hash.cc",
+      "util/log_buffer.cc",
+      "util/murmurhash.cc",
+      "util/random.cc",
+      "util/rate_limiter.cc",
+      "util/slice.cc",
+      "util/sst_file_manager_impl.cc",
+      "util/status.cc",
+      "util/status_message.cc",
+      "util/string_util.cc",
+      "util/sync_point.cc",
+      "util/thread_local.cc",
+      "util/threadpool_imp.cc",
+      "util/transaction_test_util.cc",
+      "util/xxhash.cc",
+      "utilities/backupable/backupable_db.cc",
+      "utilities/blob_db/blob_db.cc",
+      "utilities/blob_db/blob_db_impl.cc",
+      "utilities/blob_db/blob_db_options_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/checkpoint/checkpoint.cc",
+      "utilities/compaction_filters/remove_emptyvalue_compactionfilter.cc",
+      "utilities/convenience/info_log_finder.cc",
+      "utilities/date_tiered/date_tiered_db_impl.cc",
+      "utilities/document/document_db.cc",
+      "utilities/document/json_document.cc",
+      "utilities/document/json_document_builder.cc",
+      "utilities/env_mirror.cc",
+      "utilities/env_timed.cc",
+      "utilities/geodb/geodb_impl.cc",
+      "utilities/leveldb_options/leveldb_options.cc",
+      "utilities/lua/rocks_lua_compaction_filter.cc",
+      "utilities/memory/memory_util.cc",
+      "utilities/merge_operators/max.cc",
+      "utilities/merge_operators/put.cc",
+      "utilities/merge_operators/string_append/stringappend.cc",
+      "utilities/merge_operators/string_append/stringappend2.cc",
+      "utilities/merge_operators/uint64add.cc",
+      "utilities/option_change_migration/option_change_migration.cc",
+      "utilities/options/options_util.cc",
+      "utilities/persistent_cache/block_cache_tier.cc",
+      "utilities/persistent_cache/block_cache_tier_file.cc",
+      "utilities/persistent_cache/block_cache_tier_metadata.cc",
+      "utilities/persistent_cache/persistent_cache_tier.cc",
+      "utilities/persistent_cache/volatile_tier_impl.cc",
+      "utilities/redis/redis_lists.cc",
+      "utilities/simulator_cache/sim_cache.cc",
+      "utilities/spatialdb/spatial_db.cc",
+      "utilities/table_properties_collectors/compact_on_deletion_collector.cc",
+      "utilities/transactions/optimistic_transaction_db_impl.cc",
+      "utilities/transactions/optimistic_transaction_impl.cc",
+      "utilities/transactions/transaction_base.cc",
+      "utilities/transactions/transaction_db_impl.cc",
+      "utilities/transactions/transaction_db_mutex_impl.cc",
+      "utilities/transactions/transaction_impl.cc",
+      "utilities/transactions/transaction_lock_mgr.cc",
+      "utilities/transactions/transaction_util.cc",
+      "utilities/ttl/db_ttl_impl.cc",
+      "utilities/write_batch_with_index/write_batch_with_index.cc",
+      "utilities/write_batch_with_index/write_batch_with_index_internal.cc",
+      "tools/ldb_cmd.cc",
+      "tools/ldb_tool.cc",
+      "tools/sst_dump_tool.cc",
+    ],
+    deps = [],
+    preprocessor_flags = rocksdb_preprocessor_flags,
+    compiler_flags = rocksdb_compiler_flags,
+    external_deps = rocksdb_external_deps,
+)
+
+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/col_buf_encoder.cc",
+      "utilities/col_buf_decoder.cc",
+      "utilities/column_aware_encoding_util.cc",
+    ],
+    deps = [":rocksdb_lib"],
+    preprocessor_flags = rocksdb_preprocessor_flags,
+    compiler_flags = rocksdb_compiler_flags,
+    external_deps = rocksdb_external_deps,
+)
+
+cpp_library(
+    name = "rocksdb_tools_lib",
+    headers = AutoHeaders.RECURSIVE_GLOB,
+    srcs = [
+      "tools/db_bench_tool.cc",
+      "util/testutil.cc",
+    ],
+    deps = [":rocksdb_lib"],
+    preprocessor_flags = rocksdb_preprocessor_flags,
+    compiler_flags = rocksdb_compiler_flags,
+    external_deps = rocksdb_external_deps,
+)
+
+cpp_library(
+    name = "env_basic_test_lib",
+    headers = AutoHeaders.RECURSIVE_GLOB,
+    srcs = ["env/env_basic_test.cc"],
+    deps = [":rocksdb_test_lib"],
+    preprocessor_flags = rocksdb_preprocessor_flags,
+    compiler_flags = rocksdb_compiler_flags,
+    external_deps = rocksdb_external_deps,
+)
+
+# [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'],
+ ['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'],
+ ['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',
+  'serial'],
+ ['hash_table_test',
+  'utilities/persistent_cache/hash_table_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',
+  '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',
+  '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',
+  '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'],
+ ['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'],
+ ['db_memtable_test', 'db/db_memtable_test.cc', 'serial'],
+ ['db_universal_compaction_test',
+  'db/db_universal_compaction_test.cc',
+  'parallel'],
+ ['histogram_test', 'monitoring/histogram_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_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'],
+ ['db_statistics_test', 'db/db_statistics_test.cc', 'serial']]
+
+
+# Generate a test rule for each entry in ROCKS_TESTS
+for test_cfg in ROCKS_TESTS:
+    test_name = test_cfg[0]
+    test_cc = test_cfg[1]
+    ttype = "gtest" if test_cfg[2] == "parallel" else "simple"
+    test_bin = test_name + "_bin"
+
+    cpp_binary (
+      name = test_bin,
+      srcs = [test_cc],
+      deps = [":rocksdb_test_lib"],
+      preprocessor_flags = rocksdb_preprocessor_flags,
+      compiler_flags = rocksdb_compiler_flags,
+      external_deps = rocksdb_external_deps,
+    )
+
+    custom_unittest(
+      name = test_name,
+      type = ttype,
+      deps = [":" + test_bin],
+      command = [TEST_RUNNER, BUCK_BINS + test_bin]
+    )
@@ -11,8 +11,5 @@ build:
  verbosity: minimal
 test:
 test_script:
- ps: build_tools\run_ci_db_test.ps1 -Run db_basic_test -Concurrency 8
- ps: build_tools\run_ci_db_test.ps1 -Run db_test2 -Concurrency 8
- ps: build_tools\run_ci_db_test.ps1 -Run db_test -Concurrency 8
- ps: build_tools\run_ci_db_test.ps1 -Run env_test,env_basic_test -Concurrency 2
+- ps: build_tools\run_ci_db_test.ps1 -SuiteRun db_basic_test,db_test2,db_test,env_basic_test,env_test -Concurrency 8

@@ -0,0 +1,172 @@
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+from __future__ import unicode_literals
+from targets_builder import TARGETSBuilder
+from optparse import OptionParser
+import os
+import fnmatch
+import sys
+import tempfile
+
+from util import ColorString
+import util
+
+# tests to export as libraries for inclusion in other projects
+_EXPORTED_TEST_LIBS = ["env_basic_test"]
+
+# Parse src.mk files as a Dictionary of
+# VAR_NAME => list of files
+def parse_src_mk(repo_path):
+    src_mk = repo_path + "/src.mk"
+    src_files = {}
+    for line in open(src_mk):
+        line = line.strip()
+        if len(line) == 0 or line[0] == '#':
+            continue
+        if '=' in line:
+            current_src = line.split('=')[0].strip()
+            src_files[current_src] = []
+        elif '.cc' in line:
+            src_path = line.split('.cc')[0].strip() + '.cc'
+            src_files[current_src].append(src_path)
+    return src_files
+
+
+# get all .cc / .c files
+def get_cc_files(repo_path):
+    cc_files = []
+    for root, dirnames, filenames in os.walk(repo_path):
+        root = root[(len(repo_path) + 1):]
+        if "java" in root:
+            # Skip java
+            continue
+        for filename in fnmatch.filter(filenames, '*.cc'):
+            cc_files.append(os.path.join(root, filename))
+        for filename in fnmatch.filter(filenames, '*.c'):
+            cc_files.append(os.path.join(root, filename))
+    return cc_files
+
+
+# Get tests from Makefile
+def get_tests(repo_path):
+    Makefile = repo_path + "/Makefile"
+
+    # Dictionary TEST_NAME => IS_PARALLEL
+    tests = {}
+
+    found_tests = False
+    for line in open(Makefile):
+        line = line.strip()
+        if line.startswith("TESTS ="):
+            found_tests = True
+        elif found_tests:
+            if line.endswith("\\"):
+                # remove the trailing \
+                line = line[:-1]
+                line = line.strip()
+                tests[line] = False
+            else:
+                # we consumed all the tests
+                break
+
+    found_parallel_tests = False
+    for line in open(Makefile):
+        line = line.strip()
+        if line.startswith("PARALLEL_TEST ="):
+            found_parallel_tests = True
+        elif found_parallel_tests:
+            if line.endswith("\\"):
+                # remove the trailing \
+                line = line[:-1]
+                line = line.strip()
+                tests[line] = True
+            else:
+                # we consumed all the parallel tests
+                break
+
+    return tests
+
+
+# Prepare TARGETS file for buck
+def generate_targets(repo_path):
+    print(ColorString.info("Generating TARGETS"))
+    # parsed src.mk file
+    src_mk = parse_src_mk(repo_path)
+    # get all .cc files
+    cc_files = get_cc_files(repo_path)
+    # get tests from Makefile
+    tests = get_tests(repo_path)
+
+    if src_mk is None or cc_files is None or tests is None:
+        return False
+
+    TARGETS = TARGETSBuilder("%s/TARGETS" % repo_path)
+    # rocksdb_lib
+    TARGETS.add_library(
+        "rocksdb_lib",
+        src_mk["LIB_SOURCES"] +
+        src_mk["TOOL_LIB_SOURCES"])
+    # rocksdb_test_lib
+    TARGETS.add_library(
+        "rocksdb_test_lib",
+        src_mk.get("MOCK_LIB_SOURCES", []) +
+        src_mk.get("TEST_LIB_SOURCES", []) +
+        src_mk.get("EXP_LIB_SOURCES", []),
+        [":rocksdb_lib"])
+    # rocksdb_tools_lib
+    TARGETS.add_library(
+        "rocksdb_tools_lib",
+        src_mk.get("BENCH_LIB_SOURCES", []) +
+        ["util/testutil.cc"],
+        [":rocksdb_lib"])
+
+    # test for every test we found in the Makefile
+    for test in 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))
+            continue
+        elif len(match_src) > 1:
+            print(ColorString.warning("Found more than one .cc for %s" % test))
+            print(match_src)
+            continue
+
+        assert(len(match_src) == 1)
+        is_parallel = tests[test]
+        TARGETS.register_test(test, match_src[0], is_parallel)
+
+        if test in _EXPORTED_TEST_LIBS:
+            test_library = "%s_lib" % test
+            TARGETS.add_library(test_library, match_src, [":rocksdb_test_lib"])
+    TARGETS.flush_tests()
+
+    print(ColorString.info("Generated TARGETS Summary:"))
+    print(ColorString.info("- %d libs" % TARGETS.total_lib))
+    print(ColorString.info("- %d binarys" % TARGETS.total_bin))
+    print(ColorString.info("- %d tests" % TARGETS.total_test))
+    return True
+
+
+def get_rocksdb_path():
+    # rocksdb = {script_dir}/..
+    script_dir = os.path.dirname(sys.argv[0])
+    script_dir = os.path.abspath(script_dir)
+    rocksdb_path = os.path.abspath(
+        os.path.join(script_dir, "../"))
+
+    return rocksdb_path
+
+def exit_with_error(msg):
+    print(ColorString.error(msg))
+    sys.exit(1)
+
+
+def main():
+    # Generate TARGETS file for buck
+    ok = generate_targets(get_rocksdb_path())
+    if not ok:
+        exit_with_error("Failed to generate TARGETS files")
+
+if __name__ == "__main__":
+    main()
@@ -0,0 +1,4 @@
+#!/bin/bash
+# Create a tmp directory for the test to use
+TEST_DIR=$(mktemp -d /dev/shm/fbcode_rocksdb_XXXXXXX)
+TEST_TMPDIR="$TEST_DIR" $@ && rm -rf "$TEST_DIR"
@@ -0,0 +1,65 @@
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+from __future__ import unicode_literals
+import targets_cfg
+import pprint
+
+# TODO(tec): replace this with PrettyPrinter
+def pretty_list(lst, indent=6):
+    if lst is None or len(lst) == 0:
+        return ""
+
+    if len(lst) == 1:
+        return "\"%s\"" % lst[0]
+
+    separator = "\",\n%s\"" % (" " * indent)
+    res = separator.join(lst)
+    res = "\n" + (" " * indent) + "\"" + res + "\",\n" + (" " * (indent - 2))
+    return res
+
+
+class TARGETSBuilder:
+    def __init__(self, path):
+        self.path = path
+        self.targets_file = open(path, 'w')
+        self.targets_file.write(targets_cfg.rocksdb_target_header)
+        self.total_lib = 0
+        self.total_bin = 0
+        self.total_test = 0
+        self.tests_cfg = []
+
+    def __del__(self):
+        self.targets_file.close()
+
+    def add_library(self, name, srcs, deps=None, headers=None):
+        if headers is None:
+            headers = "AutoHeaders.RECURSIVE_GLOB"
+        self.targets_file.write(targets_cfg.library_template % (
+            name,
+            headers,
+            pretty_list(srcs),
+            pretty_list(deps)))
+        self.total_lib = self.total_lib + 1
+
+    def add_binary(self, name, srcs, deps=None):
+        self.targets_file.write(targets_cfg.binary_template % (
+            name,
+            pretty_list(srcs),
+            pretty_list(deps)))
+        self.total_bin = self.total_bin + 1
+
+    def register_test(self, test_name, src, is_parallel):
+        exec_mode = "serial"
+        if is_parallel:
+            exec_mode = "parallel"
+        self.tests_cfg.append([test_name, str(src), str(exec_mode)])
+
+        self.total_test = self.total_test + 1
+
+    def flush_tests(self):
+        self.targets_file.write(targets_cfg.unittests_template % (
+            pprint.PrettyPrinter().pformat(self.tests_cfg)
+        ))
+
+        self.tests_cfg = []
@@ -0,0 +1,112 @@
+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/"
+BUCK_BINS = "buck-out/gen/" + REPO_PATH
+TEST_RUNNER = REPO_PATH + "buckifier/rocks_test_runner.sh"
+rocksdb_compiler_flags = [
+  "-msse",
+  "-msse4.2",
+  "-fno-builtin-memcmp",
+  "-DROCKSDB_PLATFORM_POSIX",
+  "-DROCKSDB_LIB_IO_POSIX",
+  "-DROCKSDB_FALLOCATE_PRESENT",
+  "-DROCKSDB_MALLOC_USABLE_SIZE",
+  "-DOS_LINUX",
+  # Flags to enable libs we include
+  "-DSNAPPY",
+  "-DZLIB",
+  "-DBZIP2",
+  "-DLZ4",
+  "-DZSTD",
+  "-DGFLAGS=gflags",
+  "-DNUMA",
+  "-DTBB",
+  # Needed to compile in fbcode
+  "-Wno-expansion-to-defined",
+]
+
+rocksdb_external_deps = [
+  ('bzip2', None, 'bz2'),
+  ('snappy', None, "snappy"),
+  ('zlib', None, 'z'),
+  ('gflags', None, 'gflags'),
+  ('lz4', None, 'lz4'),
+  ('zstd', None),
+  ('tbb', None),
+  ("numa", "2.0.8", "numa"),
+  ("googletest", None, "gtest"),
+]
+
+rocksdb_preprocessor_flags = [
+  # Directories with files for #include
+  "-I" + REPO_PATH + "include/",
+  "-I" + REPO_PATH,
+]
+"""
+
+
+library_template = """
+cpp_library(
+    name = "%s",
+    headers = %s,
+    srcs = [%s],
+    deps = [%s],
+    preprocessor_flags = rocksdb_preprocessor_flags,
+    compiler_flags = rocksdb_compiler_flags,
+    external_deps = rocksdb_external_deps,
+)
+"""
+
+binary_template = """
+cpp_binary(
+  name = "%s",
+  srcs = [%s],
+  deps = [%s],
+  preprocessor_flags = rocksdb_preprocessor_flags,
+  compiler_flags = rocksdb_compiler_flags,
+  external_deps = rocksdb_external_deps,
+)
+"""
+
+unittests_template = """
+# [test_name, test_src, test_type]
+ROCKS_TESTS = %s
+
+
+# Generate a test rule for each entry in ROCKS_TESTS
+for test_cfg in ROCKS_TESTS:
+    test_name = test_cfg[0]
+    test_cc = test_cfg[1]
+    ttype = "gtest" if test_cfg[2] == "parallel" else "simple"
+    test_bin = test_name + "_bin"
+
+    cpp_binary (
+      name = test_bin,
+      srcs = [test_cc],
+      deps = [":rocksdb_test_lib"],
+      preprocessor_flags = rocksdb_preprocessor_flags,
+      compiler_flags = rocksdb_compiler_flags,
+      external_deps = rocksdb_external_deps,
+    )
+
+    custom_unittest(
+      name = test_name,
+      type = ttype,
+      deps = [":" + test_bin],
+      command = [TEST_RUNNER, BUCK_BINS + test_bin]
+    )
+
+custom_unittest(
+    name = "make_rocksdbjavastatic",
+    type = "simple",
+    command = ["internal_repo_rocksdb/make_rocksdbjavastatic.sh"],
+)
+
+custom_unittest(
+    name = "make_rocksdb_lite_release",
+    type = "simple",
+    command = ["internal_repo_rocksdb/make_rocksdb_lite_release.sh"],
+)
+"""
@@ -0,0 +1,107 @@
+"""
+This module keeps commonly used components.
+"""
+from __future__ import absolute_import
+from __future__ import division
+from __future__ import print_function
+from __future__ import unicode_literals
+import subprocess
+import os
+import time
+
+class ColorString:
+    """ Generate colorful strings on terminal """
+    HEADER = '\033[95m'
+    BLUE = '\033[94m'
+    GREEN = '\033[92m'
+    WARNING = '\033[93m'
+    FAIL = '\033[91m'
+    ENDC = '\033[0m'
+
+    @staticmethod
+    def _make_color_str(text, color):
+        return "".join([color, text.encode('utf-8'), ColorString.ENDC])
+
+    @staticmethod
+    def ok(text):
+        if ColorString.is_disabled:
+            return text
+        return ColorString._make_color_str(text, ColorString.GREEN)
+
+    @staticmethod
+    def info(text):
+        if ColorString.is_disabled:
+            return text
+        return ColorString._make_color_str(text, ColorString.BLUE)
+
+    @staticmethod
+    def header(text):
+        if ColorString.is_disabled:
+            return text
+        return ColorString._make_color_str(text, ColorString.HEADER)
+
+    @staticmethod
+    def error(text):
+        if ColorString.is_disabled:
+            return text
+        return ColorString._make_color_str(text, ColorString.FAIL)
+
+    @staticmethod
+    def warning(text):
+        if ColorString.is_disabled:
+            return text
+        return ColorString._make_color_str(text, ColorString.WARNING)
+
+    is_disabled = False
+
+
+def run_shell_command(shell_cmd, cmd_dir=None):
+    """ Run a single shell command.
+        @returns a tuple of shell command return code, stdout, stderr """
+
+    if cmd_dir is not None and not os.path.exists(cmd_dir):
+        run_shell_command("mkdir -p %s" % cmd_dir)
+
+    start = time.time()
+    print("\t>>> Running: " + shell_cmd)
+    p = subprocess.Popen(shell_cmd,
+                         shell=True,
+                         stdout=subprocess.PIPE,
+                         stderr=subprocess.PIPE,
+                         cwd=cmd_dir)
+    stdout, stderr = p.communicate()
+    end = time.time()
+
+    # Report time if we spent more than 5 minutes executing a command
+    execution_time = end - start
+    if execution_time > (60 * 5):
+        mins = (execution_time / 60)
+        secs = (execution_time % 60)
+        print("\t>time spent: %d minutes %d seconds" % (mins, secs))
+
+
+    return p.returncode, stdout, stderr
+
+
+def run_shell_commands(shell_cmds, cmd_dir=None, verbose=False):
+    """ Execute a sequence of shell commands, which is equivalent to
+        running `cmd1 && cmd2 && cmd3`
+        @returns boolean indication if all commands succeeds.
+    """
+
+    if cmd_dir:
+        print("\t=== Set current working directory => %s" % cmd_dir)
+
+    for shell_cmd in shell_cmds:
+        ret_code, stdout, stderr = run_shell_command(shell_cmd, cmd_dir)
+        if stdout:
+            if verbose or ret_code != 0:
+                print(ColorString.info("stdout: \n"), stdout)
+        if stderr:
+            # contents in stderr is not necessarily to be error messages.
+            if verbose or ret_code != 0:
+                print(ColorString.error("stderr: \n"), stderr)
+        if ret_code != 0:
+            return False
+
+    return True
@@ -9,10 +9,12 @@ fi
 # Check clang-format-diff.py
 if ! which $CLANG_FORMAT_DIFF &> /dev/null
 then
-  echo "You didn't have clang-format-diff.py available in your computer!"
-  echo "You can download it by running: "
+  echo "You didn't have clang-format-diff.py and/or clang-format available in your computer!"
+  echo "You can download clang-format-diff.py by running: "
  echo "    curl --location http://goo.gl/iUW1u2 -o ${CLANG_FORMAT_DIFF}"
-  echo "and move ${CLANG_FORMAT_DIFF} to some directory within PATH=${PATH}"
+  echo "You can download clang-format by running: "
+  echo "    brew install clang-format"
+  echo "Then, move both files (i.e. ${CLANG_FORMAT_DIFF} and clang-format) to some directory within PATH=${PATH}"
  exit 128
 fi

@@ -81,6 +81,7 @@ PARALLELISM="$PARALLEL_J $PARALLEL_j"

 DEBUG="OPT=-g"
 SHM="TEST_TMPDIR=/dev/shm/rocksdb"
+NON_SHM="TMPD=/tmp/rocksdb_test_tmp"
 GCC_481="ROCKSDB_FBCODE_BUILD_WITH_481=1"
 ASAN="COMPILE_WITH_ASAN=1"
 CLANG="USE_CLANG=1"
@@ -191,7 +192,7 @@ UNIT_TEST_NON_SHM_COMMANDS="[
            {
                'name':'Build and test RocksDB debug version',
                'timeout': 86400,
-                'shell':'$DEBUG make $PARALLELISM check || $CONTRUN_NAME=non_shm_check $TASK_CREATION_TOOL',
+                'shell':'$NON_SHM $DEBUG make $PARALLELISM check || $CONTRUN_NAME=non_shm_check $TASK_CREATION_TOOL',
                'user':'root',
                $PARSER
            },
@@ -1,24 +1,50 @@
 # This script enables you running RocksDB tests by running
-# All the tests in parallel and utilizing all the cores
-# For db_test the script first lists and parses the tests
-# and then fires them up in parallel using async PS Job functionality
-# Run the script from the enlistment
+# All the tests concurrently and utilizing all the cores
 Param(
-  [switch]$EnableJE = $false,  # Use je executable
-  [switch]$EnableRerun = $false, # Rerun failed tests sequentially at the end
-  [string]$WorkFolder = "",  # Direct tests to use that folder
-  [int]$Limit = -1, # -1 means run all otherwise limit for testing purposes
-  [string]$Exclude = "", # Expect a comma separated list, no spaces
-  [string]$Run = "db_test",  # Run db_test|db_test2|tests|testname1,testname2...
+  [switch]$EnableJE = $false,  # Look for and use _je executable, append _je to listed exclusions
+  [switch]$RunAll = $false,    # Will attempt discover all *_test[_je].exe binaries and run all
+                               # of them as Google suites. I.e. It will run test cases concurrently
+                               # except those mentioned as $Run, those will run as individual test cases
+                               # And any execlued with $ExcludeExes or $ExcludeCases
+                               # It will also not run any individual test cases
+                               # excluded but $ExcludeCasese
+  [string]$SuiteRun = "",      # Split test suites in test cases and run in parallel, not compatible with $RunAll
+  [string]$Run = "",           # Run specified executables in parallel but do not split to test cases
+  [string]$ExcludeCases = "",  # Exclude test cases, expects a comma separated list, no spaces
+                               # Takes effect when $RunAll or $SuiteRun is specified. Must have full
+                               # Test cases name including a group and a parameter if any
+  [string]$ExcludeExes = "",   # Exclude exes from consideration, expects a comma separated list,
+                               # no spaces. Takes effect only when $RunAll is specified
+  [string]$WorkFolder = "",    # Direct tests to use that folder. SSD or Ram drive are better options.
   # Number of async tasks that would run concurrently. Recommend a number below 64.
   # However, CPU utlization really depends on the storage media. Recommend ram based disk.
   # a value of 1 will run everything serially
-  [int]$Concurrency = 16
+  [int]$Concurrency = 8,
+  [int]$Limit = -1 # -1 means do not limit for test purposes
 )

 # Folders and commands must be fullpath to run assuming
 # the current folder is at the root of the git enlistment
-Get-Date
+$StartDate = (Get-Date)
+$StartDate
+
+
+$DebugPreference = "Continue"
+
+# These tests are not google test suites and we should guard
+# Against running them as suites
+$RunOnly = New-Object System.Collections.Generic.HashSet[string]
+$RunOnly.Add("c_test") | Out-Null
+$RunOnly.Add("compact_on_deletion_collector_test") | Out-Null
+$RunOnly.Add("merge_test") | Out-Null
+$RunOnly.Add("stringappend_test") | Out-Null # Apparently incorrectly written
+$RunOnly.Add("backupable_db_test") | Out-Null # Disabled
+
+
+if($RunAll -and $SuiteRun -ne "") {
+    Write-Error "$RunAll and $SuiteRun are not compatible"
+    exit 1
+}

 # If running under Appveyor assume that root
 [string]$Appveyor = $Env:APPVEYOR_BUILD_FOLDER
@@ -47,40 +73,32 @@ if($WorkFolder -eq "") {
 }

 Write-Output "Root: $RootFolder, WorkFolder: $WorkFolder"
-
-# Use JEMALLOC executables
-if($Run -ceq "db_test" -or
-   $Run -ceq "db_test2" -or
-   $Run -ceq "db_basic_test") {
-
-   $file_name = $Run
-
-   if($EnableJE) {
-     $file_name += "_je"
-   }
-
-   $file_name += ".exe"
-
-   $db_test = -Join ($BinariesFolder, $file_name)
-
-   Write-Output "Binaries: $BinariesFolder db_test: $db_test"
-}
-
-
-#Exclusions that we do not want to run
-$ExcludeTests = New-Object System.Collections.Generic.HashSet[string]
-
-
-if($Exclude -ne "") {
-    Write-Host "Exclude: $Exclude"
-    $l = $Exclude -split ' '
-    ForEach($t in $l) { $ExcludeTests.Add($t) | Out-Null }
-}
+Write-Output "BinariesFolder: $BinariesFolder, LogFolder: $LogFolder"

 # Create test directories in the current folder
 md -Path $WorkFolder -ErrorAction Ignore | Out-Null
 md -Path $LogFolder -ErrorAction Ignore | Out-Null

+
+$ExcludeCasesSet = New-Object System.Collections.Generic.HashSet[string]
+if($ExcludeCases -ne "") {
+    Write-Host "ExcludeCases: $ExcludeCases"
+    $l = $ExcludeCases -split ' '
+    ForEach($t in $l) { 
+      $ExcludeCasesSet.Add($t) | Out-Null
+    }
+}
+
+$ExcludeExesSet = New-Object System.Collections.Generic.HashSet[string]
+if($ExcludeExes -ne "") {
+    Write-Host "ExcludeExe: $ExcludeExes"
+    $l = $ExcludeExes -split ' '
+    ForEach($t in $l) { 
+      $ExcludeExesSet.Add($t) | Out-Null
+    }
+}
+
+
 # Extract the names of its tests by running db_test with --gtest_list_tests.
 # This filter removes the "#"-introduced comments, and expands to
 # fully-qualified names by changing input like this:
@@ -98,28 +116,37 @@ md -Path $LogFolder -ErrorAction Ignore | Out-Null
 #   DBTest.WriteEmptyBatch
 #   MultiThreaded/MultiThreadedDBTest.MultiThreaded/0
 #   MultiThreaded/MultiThreadedDBTest.MultiThreaded/1
+#
 # Output into the parameter in a form TestName -> Log File Name
-function Normalize-DbTests($HashTable) {
+function ExtractTestCases([string]$GTestExe, $HashTable) {

    $Tests = @()
 # Run db_test to get a list of tests and store it into $a array
-    &$db_test --gtest_list_tests | tee -Variable Tests | Out-Null
+    &$GTestExe --gtest_list_tests | tee -Variable Tests | Out-Null

    # Current group
    $Group=""

    ForEach( $l in $Tests) {
-      # Trailing dot is a test group
-      if( $l -match "\.$") {
+
+      # Leading whitespace is fine
+      $l = $l -replace '^\s+',''
+      # but no whitespace any other place
+      if($l -match "\s+") {
+        continue
+      }
+      # Trailing dot is a test group but no whitespace
+      elseif ( $l -match "\.$" ) {
        $Group = $l
      }  else {
        # Otherwise it is a test name, remove leading space
-        $test = $l -replace '^\s+',''
+        $test = $l
        # remove trailing comment if any and create a log name
        $test = $test -replace '\s+\#.*',''
        $test = "$Group$test"

-        if($ExcludeTests.Contains($test)) {
+        if($ExcludeCasesSet.Contains($test)) {
+            Write-Warning "$test case is excluded"
            continue
        }

@@ -135,59 +162,130 @@ function Normalize-DbTests($HashTable) {

 # The function removes trailing .exe siffix if any,
 # creates a name for the log file
+# Then adds the test name if it was not excluded into
+# a HashTable in a form of test_name -> log_path
 function MakeAndAdd([string]$token, $HashTable) {
+
    $test_name = $token -replace '.exe$', ''
    $log_name =  -join ($test_name, ".log")
    $log_path = -join ($LogFolder, $log_name)
-    if(!$ExcludeTests.Contains($test_name)) {
-        $HashTable.Add($test_name, $log_path)
-    } else {
-        Write-Warning "Test $test_name is excluded"
-    }
+    $HashTable.Add($test_name, $log_path)
 }

-# The function scans build\Debug folder to discover
-# Test executables. It then populates a table with
-# Test executable name -> Log file
-function Discover-TestBinaries([string]$Pattern, $HashTable) {
+# This function takes a list of Suites to run
+# Lists all the test cases in each of the suite
+# and populates HashOfHashes
+# Ordered by suite(exe) @{ Exe = @{ TestCase = LogName }}
+function ProcessSuites($ListOfSuites, $HashOfHashes) {

-    $Exclusions = @("db_test*", "db_sanity_test*", "db_basic_test*")
+  $suite_list = $ListOfSuites
+  # Problem: if you run --gtest_list_tests on
+  # a non Google Test executable then it will start executing
+  # and we will get nowhere
+  ForEach($suite in $suite_list) {

-    $p = -join ($BinariesFolder, $pattern)
-
-    Write-Host "Path: $p"
-
-    dir -Path $p -Exclude $Exclusions | ForEach-Object {
-       MakeAndAdd -token ($_.Name) -HashTable $HashTable
+    if($RunOnly.Contains($suite)) {
+      Write-Warning "$suite is excluded from running as Google test suite"
+      continue
    }
-}

-$TestsToRun = [ordered]@{}
-
-if($Run -ceq "db_test" -or
-   $Run -ceq "db_test2" -or
-   $Run -ceq "db_basic_test") {
-    Normalize-DbTests -HashTable $TestsToRun
-} elseif($Run -ceq "tests") {
    if($EnableJE) {
-        $pattern = "*_test_je.exe"
+      $suite += "_je"
+    }
+
+    $Cases = [ordered]@{}
+    $Cases.Clear()
+    $suite_exe = -Join ($BinariesFolder, $suite)
+    ExtractTestCases -GTestExe $suite_exe -HashTable $Cases
+    if($Cases.Count -gt 0) {
+      $HashOfHashes.Add($suite, $Cases);
+    }
+  }
+
+  # Make logs and run
+  if($CasesToRun.Count -lt 1) {
+     Write-Error "Failed to extract tests from $SuiteRun"
+     exit 1
+  }
+
+}
+
+# This will contain all test executables to run
+
+# Hash table that contains all non suite
+# Test executable to run
+$TestExes = [ordered]@{}
+
+# Check for test exe that are not
+# Google Test Suites
+# Since this is explicitely mentioned it is not subject
+# for exclusions
+if($Run -ne "") {
+
+  $test_list = $Run -split ' '
+
+  ForEach($t in $test_list) {
+
+    if($EnableJE) {
+      $t += "_je"
+    }
+
+    MakeAndAdd -token $t -HashTable $TestExes
+  }
+
+  if($TestExes.Count -lt 1) {
+     Write-Error "Failed to extract tests from $Run"
+     exit 1
+  }
+}
+
+# Ordered by exe @{ Exe = @{ TestCase = LogName }}
+$CasesToRun = [ordered]@{}
+
+if($SuiteRun -ne "") {
+  $suite_list = $SuiteRun -split ' '
+  ProcessSuites -ListOfSuites $suite_list -HashOfHashes $CasesToRun
+}
+
+if($RunAll) {
+# Discover all the test binaries
+  if($EnableJE) {
+    $pattern = "*_test_je.exe"
+  } else {
+    $pattern = "*_test.exe"
+  }
+
+
+  $search_path = -join ($BinariesFolder, $pattern)
+  Write-Host "Binaries Search Path: $search_path"
+
+  $ListOfExe = @()
+  dir -Path $search_path | ForEach-Object {
+     $ListOfExe += ($_.Name)     
+  }
+
+  # Exclude those in RunOnly from running as suites
+  $ListOfSuites = @()
+  ForEach($e in $ListOfExe) {
+
+    $e = $e -replace '.exe$', ''
+    $bare_name = $e -replace '_je$', ''
+
+    if($ExcludeExesSet.Contains($bare_name)) {
+      Write-Warning "Test $e is excluded"
+      continue
+    }
+
+    if($RunOnly.Contains($e)) {
+      MakeAndAdd -token $e -HashTable $TestExes
    } else {
-        $pattern = "*_test.exe"
+      $ListOfSuites += $bare_name
    }
-    Discover-TestBinaries -Pattern $pattern -HashTable $TestsToRun
-} else {
+  }

-    $test_list = $Run -split ' '
-
-    ForEach($t in $test_list) {
-       MakeAndAdd -token $t -HashTable $TestsToRun
-    }
+  ProcessSuites -ListOfSuites $ListOfSuites -HashOfHashes $CasesToRun
 }

-$NumTestsToStart = $TestsToRun.Count
-if($Limit -ge 0 -and $NumTestsToStart -gt $Limit) {
-    $NumTestsToStart = $Limit
-}

 Write-Host "Attempting to start: $NumTestsToStart tests"

@@ -205,13 +303,13 @@ $InvokeTestAsync = {

 # Hash that contains tests to rerun if any failed
 # Those tests will be rerun sequentially
-$Rerun = [ordered]@{}
+# $Rerun = [ordered]@{}
 # Test limiting factor here
-$count = 0
+[int]$count = 0
 # Overall status
 [bool]$success = $true;

-function RunJobs($TestToLog, [int]$ConcurrencyVal, [bool]$AddForRerun)
+function RunJobs($Suites, $TestCmds, [int]$ConcurrencyVal)
 {
    # Array to wait for any of the running jobs
    $jobs = @()
@@ -220,38 +318,66 @@ function RunJobs($TestToLog, [int]$ConcurrencyVal, [bool]$AddForRerun)

    # Wait for all to finish and get the results
    while(($JobToLog.Count -gt 0) -or
-          ($TestToLog.Count -gt 0)) {
+          ($TestCmds.Count -gt 0) -or 
+           ($Suites.Count -gt 0)) {

        # Make sure we have maximum concurrent jobs running if anything
        # and the $Limit either not set or allows to proceed
        while(($JobToLog.Count -lt $ConcurrencyVal) -and
-              (($TestToLog.Count -gt 0) -and
+              ((($TestCmds.Count -gt 0) -or ($Suites.Count -gt 0)) -and
              (($Limit -lt 0) -or ($count -lt $Limit)))) {

+            # We always favore suites to run if available
+            [string]$exe_name = ""
+            [string]$log_path = ""
+            $Cases = @{}

-            # We only need the first key
-            foreach($key in $TestToLog.keys) {
-                $k = $key
+            if($Suites.Count -gt 0) {
+              # Will teh first one
+              ForEach($e in $Suites.Keys) {
+                $exe_name = $e
+                $Cases = $Suites[$e]
                break
-            }
+              }
+              [string]$test_case = ""
+              [string]$log_path = ""
+              ForEach($c in $Cases.Keys) {
+                 $test_case = $c
+                 $log_path = $Cases[$c]
+                 break
+              }

-            Write-Host "Starting $k"
-            $log_path = ($TestToLog.$k)
+              Write-Host "Starting $exe_name::$test_case"
+              [string]$Exe =  -Join ($BinariesFolder, $exe_name)
+              $job = Start-Job -Name "$exe_name::$test_case" -ArgumentList @($Exe,$test_case,$log_path) -ScriptBlock $InvokeTestCase
+              $JobToLog.Add($job, $log_path)
+
+              $Cases.Remove($test_case)
+              if($Cases.Count -lt 1) {
+                $Suites.Remove($exe_name)
+              }
+
+            } elseif ($TestCmds.Count -gt 0) {
+
+               ForEach($e in $TestCmds.Keys) {
+                 $exe_name = $e
+                 $log_path = $TestCmds[$e]
+                 break
+               }
+
+              [string]$Exe =  -Join ($BinariesFolder, $exe_name)
+              $job = Start-Job -Name $exe_name -ScriptBlock $InvokeTestAsync -ArgumentList @($Exe,$log_path)
+              $JobToLog.Add($job, $log_path)
+
+              $TestCmds.Remove($exe_name)

-            if($Run -ceq "db_test" -or
-               $Run -ceq "db_test2" -or
-               $Run -ceq "db_basic_test") {
-              $job = Start-Job -Name $k -ScriptBlock $InvokeTestCase -ArgumentList @($db_test,$k,$log_path)
            } else {
-              [string]$Exe =  -Join ($BinariesFolder, $k)
-               $job = Start-Job -Name $k -ScriptBlock $InvokeTestAsync -ArgumentList @($exe,$log_path)
+                Write-Error "In the job loop but nothing to run"
+                exit 1
            }

-            $JobToLog.Add($job, $log_path)
-            $TestToLog.Remove($k)
-
            ++$count
-        }
+        } # End of Job starting loop

        if($JobToLog.Count -lt 1) {
          break
@@ -296,9 +422,6 @@ function RunJobs($TestToLog, [int]$ConcurrencyVal, [bool]$AddForRerun)
                $success = $false;
                Write-Warning $message
                $log_content | Write-Warning
-                if($AddForRerun) {
-                    $Rerun.Add($completed.Name, $log)
-                }
            } else {
                Write-Host $message
            }
@@ -310,16 +433,14 @@ function RunJobs($TestToLog, [int]$ConcurrencyVal, [bool]$AddForRerun)
    }
 }

-RunJobs -TestToLog $TestsToRun -ConcurrencyVal $Concurrency -AddForRerun $EnableRerun
+RunJobs -Suites $CasesToRun -TestCmds $TestExes -ConcurrencyVal $Concurrency

-if($Rerun.Count -gt 0) {
-    Write-Host "Rerunning " ($Rerun.Count) " tests sequentially"
-    $success = $true
-    $count = 0
-    RunJobs -TestToLog $Rerun -ConcurrencyVal 1 -AddForRerun $false
-}
+$EndDate = (Get-Date)

-Get-Date
+New-TimeSpan -Start $StartDate -End $EndDate | 
+  ForEach-Object { 
+    "Elapsed time: {0:g}" -f $_
+  }


 if(!$success) {
@@ -327,7 +448,9 @@ if(!$success) {
 # So we simply exit
 #    Remove-Job -Job $jobs -Force
 # indicate failure using this exit code
-    exit 12345
+    exit 1
 }

+ exit 0
+
 
@@ -14,9 +14,9 @@
 #include <iostream>
 #include <string>
 #include <vector>
-#include "util/clock_cache.h"
+#include "cache/clock_cache.h"
+#include "cache/lru_cache.h"
 #include "util/coding.h"
-#include "util/lru_cache.h"
 #include "util/string_util.h"
 #include "util/testharness.h"

@@ -7,7 +7,7 @@
 // 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.

-#include "util/clock_cache.h"
+#include "cache/clock_cache.h"

 #ifndef SUPPORT_CLOCK_CACHE

@@ -30,10 +30,10 @@ std::shared_ptr<Cache> NewClockCache(size_t capacity, int num_shard_bits,

 #include "tbb/concurrent_hash_map.h"

+#include "cache/sharded_cache.h"
 #include "port/port.h"
 #include "util/autovector.h"
 #include "util/mutexlock.h"
-#include "util/sharded_cache.h"

 namespace rocksdb {

@@ -11,7 +11,7 @@
 #define __STDC_FORMAT_MACROS
 #endif

-#include "util/lru_cache.h"
+#include "cache/lru_cache.h"

 #include <assert.h>
 #include <stdio.h>
@@ -10,7 +10,7 @@

 #include <string>

-#include "util/sharded_cache.h"
+#include "cache/sharded_cache.h"

 #include "port/port.h"
 #include "util/autovector.h"
@@ -3,7 +3,7 @@
 //  COPYING file in the root directory) and Apache 2.0 License
 //  (found in the LICENSE.Apache file in the root directory).

-#include "util/lru_cache.h"
+#include "cache/lru_cache.h"

 #include <string>
 #include <vector>
@@ -11,7 +11,7 @@
 #define __STDC_FORMAT_MACROS
 #endif

-#include "util/sharded_cache.h"
+#include "cache/sharded_cache.h"

 #include <string>

@@ -1,22 +1,21 @@
 # - Find JeMalloc library
 # Find the native JeMalloc includes and library
-# This module defines
-#  JEMALLOC_INCLUDE_DIRS, where to find jemalloc.h, Set when
-#                        JEMALLOC_INCLUDE_DIR is found.
-#  JEMALLOC_LIBRARIES, libraries to link against to use JeMalloc.
-#  JEMALLOC_FOUND, If false, do not try to use JeMalloc.
 #
+# JEMALLOC_INCLUDE_DIR - where to find jemalloc.h, etc.
+# JEMALLOC_LIBRARIES - List of libraries when using jemalloc.
+# JEMALLOC_FOUND - True if jemalloc found.
+
 find_path(JEMALLOC_INCLUDE_DIR
-  jemalloc/jemalloc.h)
+  NAMES jemalloc/jemalloc.h
+  HINTS ${JEMALLOC_ROOT_DIR}/include)

 find_library(JEMALLOC_LIBRARIES
-  jemalloc)
+  NAMES jemalloc
+  HINTS ${JEMALLOC_ROOT_DIR}/lib)

 include(FindPackageHandleStandardArgs)
-find_package_handle_standard_args(JeMalloc DEFAULT_MSG
-    JEMALLOC_LIBRARIES JEMALLOC_INCLUDE_DIR)
+find_package_handle_standard_args(jemalloc DEFAULT_MSG JEMALLOC_LIBRARIES JEMALLOC_INCLUDE_DIR)

-MARK_AS_ADVANCED(
-  JEMALLOC_INCLUDE_DIR
+mark_as_advanced(
  JEMALLOC_LIBRARIES
-)
+  JEMALLOC_INCLUDE_DIR)
@@ -13,8 +13,6 @@ find_library(SNAPPY_LIBRARIES
  NAMES snappy
  HINTS ${SNAPPY_ROOT_DIR}/lib)

-# handle the QUIETLY and REQUIRED arguments and set UUID_FOUND to TRUE if
-# all listed variables are TRUE
 include(FindPackageHandleStandardArgs)
 find_package_handle_standard_args(snappy DEFAULT_MSG SNAPPY_LIBRARIES SNAPPY_INCLUDE_DIR)

@@ -16,11 +16,12 @@
 #include "db/compaction_iterator.h"
 #include "db/dbformat.h"
 #include "db/event_helpers.h"
-#include "db/filename.h"
 #include "db/internal_stats.h"
 #include "db/merge_helper.h"
 #include "db/table_cache.h"
 #include "db/version_edit.h"
+#include "monitoring/iostats_context_imp.h"
+#include "monitoring/thread_status_util.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
 #include "rocksdb/iterator.h"
@@ -29,9 +30,9 @@
 #include "table/block_based_table_builder.h"
 #include "table/internal_iterator.h"
 #include "util/file_reader_writer.h"
-#include "util/iostats_context_imp.h"
+#include "util/filename.h"
 #include "util/stop_watch.h"
-#include "util/thread_status_util.h"
+#include "util/sync_point.h"

 namespace rocksdb {

@@ -103,6 +104,10 @@ 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(
@@ -172,7 +177,7 @@ 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 +185,11 @@ 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,
@@ -10,6 +10,7 @@
 #include <utility>
 #include <vector>
 #include "db/table_properties_collector.h"
+#include "options/cf_options.h"
 #include "rocksdb/comparator.h"
 #include "rocksdb/env.h"
 #include "rocksdb/listener.h"
@@ -18,7 +19,6 @@
 #include "rocksdb/table_properties.h"
 #include "rocksdb/types.h"
 #include "table/scoped_arena_iterator.h"
-#include "util/cf_options.h"
 #include "util/event_logger.h"

 namespace rocksdb {
@@ -32,8 +32,10 @@
 #include "rocksdb/table.h"
 #include "rocksdb/rate_limiter.h"
 #include "rocksdb/utilities/backupable_db.h"
+#include "rocksdb/utilities/write_batch_with_index.h"
 #include "utilities/merge_operators.h"

+using rocksdb::BytewiseComparator;
 using rocksdb::Cache;
 using rocksdb::ColumnFamilyDescriptor;
 using rocksdb::ColumnFamilyHandle;
@@ -75,6 +77,7 @@ using rocksdb::SstFileWriter;
 using rocksdb::Status;
 using rocksdb::WritableFile;
 using rocksdb::WriteBatch;
+using rocksdb::WriteBatchWithIndex;
 using rocksdb::WriteOptions;
 using rocksdb::LiveFileMetaData;
 using rocksdb::BackupEngine;
@@ -95,6 +98,7 @@ struct rocksdb_backup_engine_info_t { std::vector<BackupInfo> rep; };
 struct rocksdb_restore_options_t { RestoreOptions rep; };
 struct rocksdb_iterator_t        { Iterator*         rep; };
 struct rocksdb_writebatch_t      { WriteBatch        rep; };
+struct rocksdb_writebatch_wi_t   { WriteBatchWithIndex* rep; };
 struct rocksdb_snapshot_t        { const Snapshot*   rep; };
 struct rocksdb_flushoptions_t    { FlushOptions      rep; };
 struct rocksdb_fifo_compaction_options_t { CompactionOptionsFIFO rep; };
@@ -1356,6 +1360,377 @@ void rocksdb_writebatch_rollback_to_save_point(rocksdb_writebatch_t* b,
  SaveError(errptr, b->rep.RollbackToSavePoint());
 }

+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);
+  return b;
+}
+
+void rocksdb_writebatch_wi_destroy(rocksdb_writebatch_wi_t* b) {
+  if (b->rep) {
+    delete b->rep;
+  }
+  delete b;
+}
+
+void rocksdb_writebatch_wi_clear(rocksdb_writebatch_wi_t* b) {
+  b->rep->Clear();
+}
+
+int rocksdb_writebatch_wi_count(rocksdb_writebatch_wi_t* b) {
+  return b->rep->GetWriteBatch()->Count();
+}
+
+void rocksdb_writebatch_wi_put(
+    rocksdb_writebatch_wi_t* b,
+    const char* key, size_t klen,
+    const char* val, size_t vlen) {
+  b->rep->Put(Slice(key, klen), Slice(val, vlen));
+}
+
+void rocksdb_writebatch_wi_put_cf(
+    rocksdb_writebatch_wi_t* b,
+    rocksdb_column_family_handle_t* column_family,
+    const char* key, size_t klen,
+    const char* val, size_t vlen) {
+  b->rep->Put(column_family->rep, Slice(key, klen), Slice(val, vlen));
+}
+
+void rocksdb_writebatch_wi_putv(
+    rocksdb_writebatch_wi_t* b,
+    int num_keys, const char* const* keys_list,
+    const size_t* keys_list_sizes,
+    int num_values, const char* const* values_list,
+    const size_t* values_list_sizes) {
+  std::vector<Slice> key_slices(num_keys);
+  for (int i = 0; i < num_keys; i++) {
+    key_slices[i] = Slice(keys_list[i], keys_list_sizes[i]);
+  }
+  std::vector<Slice> value_slices(num_values);
+  for (int i = 0; i < num_values; i++) {
+    value_slices[i] = Slice(values_list[i], values_list_sizes[i]);
+  }
+  b->rep->Put(SliceParts(key_slices.data(), num_keys),
+             SliceParts(value_slices.data(), num_values));
+}
+
+void rocksdb_writebatch_wi_putv_cf(
+    rocksdb_writebatch_wi_t* b,
+    rocksdb_column_family_handle_t* column_family,
+    int num_keys, const char* const* keys_list,
+    const size_t* keys_list_sizes,
+    int num_values, const char* const* values_list,
+    const size_t* values_list_sizes) {
+  std::vector<Slice> key_slices(num_keys);
+  for (int i = 0; i < num_keys; i++) {
+    key_slices[i] = Slice(keys_list[i], keys_list_sizes[i]);
+  }
+  std::vector<Slice> value_slices(num_values);
+  for (int i = 0; i < num_values; i++) {
+    value_slices[i] = Slice(values_list[i], values_list_sizes[i]);
+  }
+  b->rep->Put(column_family->rep, SliceParts(key_slices.data(), num_keys),
+             SliceParts(value_slices.data(), num_values));
+}
+
+void rocksdb_writebatch_wi_merge(
+    rocksdb_writebatch_wi_t* b,
+    const char* key, size_t klen,
+    const char* val, size_t vlen) {
+  b->rep->Merge(Slice(key, klen), Slice(val, vlen));
+}
+
+void rocksdb_writebatch_wi_merge_cf(
+    rocksdb_writebatch_wi_t* b,
+    rocksdb_column_family_handle_t* column_family,
+    const char* key, size_t klen,
+    const char* val, size_t vlen) {
+  b->rep->Merge(column_family->rep, Slice(key, klen), Slice(val, vlen));
+}
+
+void rocksdb_writebatch_wi_mergev(
+    rocksdb_writebatch_wi_t* b,
+    int num_keys, const char* const* keys_list,
+    const size_t* keys_list_sizes,
+    int num_values, const char* const* values_list,
+    const size_t* values_list_sizes) {
+  std::vector<Slice> key_slices(num_keys);
+  for (int i = 0; i < num_keys; i++) {
+    key_slices[i] = Slice(keys_list[i], keys_list_sizes[i]);
+  }
+  std::vector<Slice> value_slices(num_values);
+  for (int i = 0; i < num_values; i++) {
+    value_slices[i] = Slice(values_list[i], values_list_sizes[i]);
+  }
+  b->rep->Merge(SliceParts(key_slices.data(), num_keys),
+               SliceParts(value_slices.data(), num_values));
+}
+
+void rocksdb_writebatch_wi_mergev_cf(
+    rocksdb_writebatch_wi_t* b,
+    rocksdb_column_family_handle_t* column_family,
+    int num_keys, const char* const* keys_list,
+    const size_t* keys_list_sizes,
+    int num_values, const char* const* values_list,
+    const size_t* values_list_sizes) {
+  std::vector<Slice> key_slices(num_keys);
+  for (int i = 0; i < num_keys; i++) {
+    key_slices[i] = Slice(keys_list[i], keys_list_sizes[i]);
+  }
+  std::vector<Slice> value_slices(num_values);
+  for (int i = 0; i < num_values; i++) {
+    value_slices[i] = Slice(values_list[i], values_list_sizes[i]);
+  }
+  b->rep->Merge(column_family->rep, SliceParts(key_slices.data(), num_keys),
+               SliceParts(value_slices.data(), num_values));
+}
+
+void rocksdb_writebatch_wi_delete(
+    rocksdb_writebatch_wi_t* b,
+    const char* key, size_t klen) {
+  b->rep->Delete(Slice(key, klen));
+}
+
+void rocksdb_writebatch_wi_delete_cf(
+    rocksdb_writebatch_wi_t* b,
+    rocksdb_column_family_handle_t* column_family,
+    const char* key, size_t klen) {
+  b->rep->Delete(column_family->rep, Slice(key, klen));
+}
+
+void rocksdb_writebatch_wi_deletev(
+    rocksdb_writebatch_wi_t* b,
+    int num_keys, const char* const* keys_list,
+    const size_t* keys_list_sizes) {
+  std::vector<Slice> key_slices(num_keys);
+  for (int i = 0; i < num_keys; i++) {
+    key_slices[i] = Slice(keys_list[i], keys_list_sizes[i]);
+  }
+  b->rep->Delete(SliceParts(key_slices.data(), num_keys));
+}
+
+void rocksdb_writebatch_wi_deletev_cf(
+    rocksdb_writebatch_wi_t* b,
+    rocksdb_column_family_handle_t* column_family,
+    int num_keys, const char* const* keys_list,
+    const size_t* keys_list_sizes) {
+  std::vector<Slice> key_slices(num_keys);
+  for (int i = 0; i < num_keys; i++) {
+    key_slices[i] = Slice(keys_list[i], keys_list_sizes[i]);
+  }
+  b->rep->Delete(column_family->rep, SliceParts(key_slices.data(), num_keys));
+}
+
+void rocksdb_writebatch_wi_delete_range(rocksdb_writebatch_wi_t* b,
+                                     const char* start_key,
+                                     size_t start_key_len, const char* end_key,
+                                     size_t end_key_len) {
+  b->rep->DeleteRange(Slice(start_key, start_key_len),
+                     Slice(end_key, end_key_len));
+}
+
+void rocksdb_writebatch_wi_delete_range_cf(
+    rocksdb_writebatch_wi_t* b, rocksdb_column_family_handle_t* column_family,
+    const char* start_key, size_t start_key_len, const char* end_key,
+    size_t end_key_len) {
+  b->rep->DeleteRange(column_family->rep, Slice(start_key, start_key_len),
+                     Slice(end_key, end_key_len));
+}
+
+void rocksdb_writebatch_wi_delete_rangev(rocksdb_writebatch_wi_t* b, int num_keys,
+                                      const char* const* start_keys_list,
+                                      const size_t* start_keys_list_sizes,
+                                      const char* const* end_keys_list,
+                                      const size_t* end_keys_list_sizes) {
+  std::vector<Slice> start_key_slices(num_keys);
+  std::vector<Slice> end_key_slices(num_keys);
+  for (int i = 0; i < num_keys; i++) {
+    start_key_slices[i] = Slice(start_keys_list[i], start_keys_list_sizes[i]);
+    end_key_slices[i] = Slice(end_keys_list[i], end_keys_list_sizes[i]);
+  }
+  b->rep->DeleteRange(SliceParts(start_key_slices.data(), num_keys),
+                     SliceParts(end_key_slices.data(), num_keys));
+}
+
+void rocksdb_writebatch_wi_delete_rangev_cf(
+    rocksdb_writebatch_wi_t* b, rocksdb_column_family_handle_t* column_family,
+    int num_keys, const char* const* start_keys_list,
+    const size_t* start_keys_list_sizes, const char* const* end_keys_list,
+    const size_t* end_keys_list_sizes) {
+  std::vector<Slice> start_key_slices(num_keys);
+  std::vector<Slice> end_key_slices(num_keys);
+  for (int i = 0; i < num_keys; i++) {
+    start_key_slices[i] = Slice(start_keys_list[i], start_keys_list_sizes[i]);
+    end_key_slices[i] = Slice(end_keys_list[i], end_keys_list_sizes[i]);
+  }
+  b->rep->DeleteRange(column_family->rep,
+                     SliceParts(start_key_slices.data(), num_keys),
+                     SliceParts(end_key_slices.data(), num_keys));
+}
+
+void rocksdb_writebatch_wi_put_log_data(
+    rocksdb_writebatch_wi_t* b,
+    const char* blob, size_t len) {
+  b->rep->PutLogData(Slice(blob, len));
+}
+
+void rocksdb_writebatch_wi_iterate(
+    rocksdb_writebatch_wi_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;
+  handler.deleted_ = deleted;
+  b->rep->GetWriteBatch()->Iterate(&handler);
+}
+
+const char* rocksdb_writebatch_wi_data(rocksdb_writebatch_wi_t* b, size_t* size) {
+  WriteBatch* wb = b->rep->GetWriteBatch();
+  *size = wb->GetDataSize();
+  return wb->Data().c_str();
+}
+
+void rocksdb_writebatch_wi_set_save_point(rocksdb_writebatch_wi_t* b) {
+  b->rep->SetSavePoint();
+}
+
+void rocksdb_writebatch_wi_rollback_to_save_point(rocksdb_writebatch_wi_t* b,
+                                               char** errptr) {
+  SaveError(errptr, b->rep->RollbackToSavePoint());
+}
+
+rocksdb_iterator_t* rocksdb_writebatch_wi_create_iterator_with_base(
+    rocksdb_writebatch_wi_t* wbwi,
+    rocksdb_iterator_t* base_iterator) {
+  rocksdb_iterator_t* result = new rocksdb_iterator_t;
+  result->rep = wbwi->rep->NewIteratorWithBase(base_iterator->rep);
+  delete base_iterator;
+  return result;
+}
+
+rocksdb_iterator_t* rocksdb_writebatch_wi_create_iterator_with_base_cf(
+    rocksdb_writebatch_wi_t* wbwi,
+    rocksdb_iterator_t* base_iterator,
+    rocksdb_column_family_handle_t* column_family) {
+  rocksdb_iterator_t* result = new rocksdb_iterator_t;
+  result->rep = wbwi->rep->NewIteratorWithBase(column_family->rep, base_iterator->rep);
+  delete base_iterator;
+  return result;
+}
+
+char* rocksdb_writebatch_wi_get_from_batch(
+    rocksdb_writebatch_wi_t* wbwi,
+    const rocksdb_options_t* options,
+    const char* key, size_t keylen,
+    size_t* vallen,
+    char** errptr) {
+  char* result = nullptr;
+  std::string tmp;
+  Status s = wbwi->rep->GetFromBatch(options->rep, Slice(key, keylen), &tmp);
+  if (s.ok()) {
+    *vallen = tmp.size();
+    result = CopyString(tmp);
+  } else {
+    *vallen = 0;
+    if (!s.IsNotFound()) {
+      SaveError(errptr, s);
+    }
+  }
+  return result;
+}
+
+char* rocksdb_writebatch_wi_get_from_batch_cf(
+    rocksdb_writebatch_wi_t* wbwi,
+    const rocksdb_options_t* options,
+    rocksdb_column_family_handle_t* column_family,
+    const char* key, size_t keylen,
+    size_t* vallen,
+    char** errptr) {
+  char* result = nullptr;
+  std::string tmp;
+  Status s = wbwi->rep->GetFromBatch(column_family->rep, options->rep,
+      Slice(key, keylen), &tmp);
+  if (s.ok()) {
+    *vallen = tmp.size();
+    result = CopyString(tmp);
+  } else {
+    *vallen = 0;
+    if (!s.IsNotFound()) {
+      SaveError(errptr, s);
+    }
+  }
+  return result;
+}
+
+char* rocksdb_writebatch_wi_get_from_batch_and_db(
+    rocksdb_writebatch_wi_t* wbwi,
+    rocksdb_t* db,
+    const rocksdb_readoptions_t* options,
+    const char* key, size_t keylen,
+    size_t* vallen,
+    char** errptr) {
+  char* result = nullptr;
+  std::string tmp;
+  Status s = wbwi->rep->GetFromBatchAndDB(db->rep, options->rep, Slice(key, keylen), &tmp);
+  if (s.ok()) {
+    *vallen = tmp.size();
+    result = CopyString(tmp);
+  } else {
+    *vallen = 0;
+    if (!s.IsNotFound()) {
+      SaveError(errptr, s);
+    }
+  }
+  return result;
+}
+
+char* rocksdb_writebatch_wi_get_from_batch_and_db_cf(
+    rocksdb_writebatch_wi_t* wbwi,
+    rocksdb_t* db,
+    const rocksdb_readoptions_t* options,
+    rocksdb_column_family_handle_t* column_family,
+    const char* key, size_t keylen,
+    size_t* vallen,
+    char** errptr) {
+  char* result = nullptr;
+  std::string tmp;
+  Status s = wbwi->rep->GetFromBatchAndDB(db->rep, options->rep, column_family->rep,
+      Slice(key, keylen), &tmp);
+  if (s.ok()) {
+    *vallen = tmp.size();
+    result = CopyString(tmp);
+  } else {
+    *vallen = 0;
+    if (!s.IsNotFound()) {
+      SaveError(errptr, s);
+    }
+  }
+  return result;
+}
+
+void rocksdb_write_writebatch_wi(
+    rocksdb_t* db,
+    const rocksdb_writeoptions_t* options,
+    rocksdb_writebatch_wi_t* wbwi,
+    char** errptr) {
+  WriteBatch* wb = wbwi->rep->GetWriteBatch();
+  SaveError(errptr, db->rep->Write(options->rep, wb));
+}
+
 rocksdb_block_based_table_options_t*
 rocksdb_block_based_options_create() {
  return new rocksdb_block_based_table_options_t;
@@ -1746,9 +2121,9 @@ 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(
@@ -2397,8 +2772,7 @@ void rocksdb_envoptions_destroy(rocksdb_envoptions_t* opt) { delete opt; }
 rocksdb_sstfilewriter_t* rocksdb_sstfilewriter_create(
    const rocksdb_envoptions_t* env, const rocksdb_options_t* io_options) {
  rocksdb_sstfilewriter_t* writer = new rocksdb_sstfilewriter_t;
-  writer->rep =
-      new SstFileWriter(env->rep, io_options->rep, io_options->rep.comparator);
+  writer->rep = new SstFileWriter(env->rep, io_options->rep);
  return writer;
 }

@@ -2406,8 +2780,7 @@ rocksdb_sstfilewriter_t* rocksdb_sstfilewriter_create_with_comparator(
    const rocksdb_envoptions_t* env, const rocksdb_options_t* io_options,
    const rocksdb_comparator_t* comparator) {
  rocksdb_sstfilewriter_t* writer = new rocksdb_sstfilewriter_t;
-  writer->rep =
-      new SstFileWriter(env->rep, io_options->rep, io_options->rep.comparator);
+  writer->rep = new SstFileWriter(env->rep, io_options->rep);
  return writer;
 }

@@ -87,6 +87,16 @@ static void Free(char** ptr) {
  }
 }

+static void CheckValue(
+    char* err,
+    const char* expected,
+    char** actual,
+    size_t actual_length) {
+  CheckNoError(err);
+  CheckEqual(expected, *actual, actual_length);
+  Free(actual);
+}
+
 static void CheckGet(
    rocksdb_t* db,
    const rocksdb_readoptions_t* options,
@@ -608,6 +618,90 @@ int main(int argc, char** argv) {
    rocksdb_writebatch_destroy(wb2);
  }

+  StartPhase("writebatch_wi");
+  {
+    rocksdb_writebatch_wi_t* wbi = rocksdb_writebatch_wi_create(0, 1);
+    rocksdb_writebatch_wi_put(wbi, "foo", 3, "a", 1);
+    rocksdb_writebatch_wi_clear(wbi);
+    rocksdb_writebatch_wi_put(wbi, "bar", 3, "b", 1);
+    rocksdb_writebatch_wi_put(wbi, "box", 3, "c", 1);
+    rocksdb_writebatch_wi_delete(wbi, "bar", 3);
+    int count = rocksdb_writebatch_wi_count(wbi);
+    CheckCondition(count == 3);
+    size_t size;
+    char* value;
+    value = rocksdb_writebatch_wi_get_from_batch(wbi, options, "box", 3, &size, &err);
+    CheckValue(err, "c", &value, size);
+    value = rocksdb_writebatch_wi_get_from_batch(wbi, options, "bar", 3, &size, &err);
+    CheckValue(err, NULL, &value, size);
+    value = rocksdb_writebatch_wi_get_from_batch_and_db(wbi, db, roptions, "foo", 3, &size, &err);
+    CheckValue(err, "hello", &value, size);
+    value = rocksdb_writebatch_wi_get_from_batch_and_db(wbi, db, roptions, "box", 3, &size, &err);
+    CheckValue(err, "c", &value, size);
+    rocksdb_write_writebatch_wi(db, woptions, wbi, &err);
+    CheckNoError(err);
+    CheckGet(db, roptions, "foo", "hello");
+    CheckGet(db, roptions, "bar", NULL);
+    CheckGet(db, roptions, "box", "c");
+    int pos = 0;
+    rocksdb_writebatch_wi_iterate(wbi, &pos, CheckPut, CheckDel);
+    CheckCondition(pos == 3);
+    rocksdb_writebatch_wi_clear(wbi);
+    rocksdb_writebatch_wi_put(wbi, "bar", 3, "b", 1);
+    rocksdb_writebatch_wi_put(wbi, "bay", 3, "d", 1);
+    rocksdb_writebatch_wi_delete_range(wbi, "bar", 3, "bay", 3);
+    rocksdb_write_writebatch_wi(db, woptions, wbi, &err);
+    CheckNoError(err);
+    CheckGet(db, roptions, "bar", NULL);
+    CheckGet(db, roptions, "bay", "d");
+    rocksdb_writebatch_wi_clear(wbi);
+    const char* start_list[1] = {"bay"};
+    const size_t start_sizes[1] = {3};
+    const char* end_list[1] = {"baz"};
+    const size_t end_sizes[1] = {3};
+    rocksdb_writebatch_wi_delete_rangev(wbi, 1, start_list, start_sizes, end_list,
+                                     end_sizes);
+    rocksdb_write_writebatch_wi(db, woptions, wbi, &err);
+    CheckNoError(err);
+    CheckGet(db, roptions, "bay", NULL);
+    rocksdb_writebatch_wi_destroy(wbi);
+  }
+
+  StartPhase("writebatch_wi_vectors");
+  {
+    rocksdb_writebatch_wi_t* wb = rocksdb_writebatch_wi_create(0, 1);
+    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_wi_putv(wb, 2, k_list, k_sizes, 3, v_list, v_sizes);
+    rocksdb_write_writebatch_wi(db, woptions, wb, &err);
+    CheckNoError(err);
+    CheckGet(db, roptions, "zap", "xyz");
+    rocksdb_writebatch_wi_delete(wb, "zap", 3);
+    rocksdb_write_writebatch_wi(db, woptions, wb, &err);
+    CheckNoError(err);
+    CheckGet(db, roptions, "zap", NULL);
+    rocksdb_writebatch_wi_destroy(wb);
+  }
+
+  StartPhase("writebatch_wi_savepoint");
+  {
+    rocksdb_writebatch_wi_t* wb = rocksdb_writebatch_wi_create(0, 1);
+    rocksdb_writebatch_wi_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_wi_putv(wb, 2, k_list, k_sizes, 3, v_list, v_sizes);
+    rocksdb_writebatch_wi_rollback_to_save_point(wb, &err);
+    CheckNoError(err);
+    rocksdb_write_writebatch_wi(db, woptions, wb, &err);
+    CheckNoError(err);
+    CheckGet(db, roptions, "zap", NULL);
+    rocksdb_writebatch_wi_destroy(wb);
+  }
+
  StartPhase("iter");
  {
    rocksdb_iterator_t* iter = rocksdb_create_iterator(db, roptions);
@@ -634,6 +728,37 @@ int main(int argc, char** argv) {
    rocksdb_iter_destroy(iter);
  }

+  StartPhase("wbwi_iter");
+  {
+    rocksdb_iterator_t* base_iter = rocksdb_create_iterator(db, roptions);
+    rocksdb_writebatch_wi_t* wbi = rocksdb_writebatch_wi_create(0, 1);
+    rocksdb_writebatch_wi_put(wbi, "bar", 3, "b", 1);
+    rocksdb_writebatch_wi_delete(wbi, "foo", 3);
+    rocksdb_iterator_t* iter = rocksdb_writebatch_wi_create_iterator_with_base(wbi, base_iter);
+    CheckCondition(!rocksdb_iter_valid(iter));
+    rocksdb_iter_seek_to_first(iter);
+    CheckCondition(rocksdb_iter_valid(iter));
+    CheckIter(iter, "bar", "b");
+    rocksdb_iter_next(iter);
+    CheckIter(iter, "box", "c");
+    rocksdb_iter_prev(iter);
+    CheckIter(iter, "bar", "b");
+    rocksdb_iter_prev(iter);
+    CheckCondition(!rocksdb_iter_valid(iter));
+    rocksdb_iter_seek_to_last(iter);
+    CheckIter(iter, "box", "c");
+    rocksdb_iter_seek(iter, "b", 1);
+    CheckIter(iter, "bar", "b");
+    rocksdb_iter_seek_for_prev(iter, "c", 1);
+    CheckIter(iter, "box", "c");
+    rocksdb_iter_seek_for_prev(iter, "box", 3);
+    CheckIter(iter, "box", "c");
+    rocksdb_iter_get_error(iter, &err);
+    CheckNoError(err);
+    rocksdb_iter_destroy(iter);
+    rocksdb_writebatch_wi_destroy(wbi);
+  }
+
  StartPhase("multiget");
  {
    const char* keys[3] = { "box", "foo", "notfound" };
@@ -20,6 +20,7 @@
 #include <limits>

 #include "db/compaction_picker.h"
+#include "db/compaction_picker_universal.h"
 #include "db/db_impl.h"
 #include "db/internal_stats.h"
 #include "db/job_context.h"
@@ -27,11 +28,11 @@
 #include "db/version_set.h"
 #include "db/write_controller.h"
 #include "memtable/hash_skiplist_rep.h"
+#include "monitoring/thread_status_util.h"
+#include "options/options_helper.h"
 #include "table/block_based_table_factory.h"
 #include "util/autovector.h"
 #include "util/compression.h"
-#include "util/options_helper.h"
-#include "util/thread_status_util.h"

 namespace rocksdb {

@@ -205,8 +206,8 @@ ColumnFamilyOptions SanitizeOptions(const ImmutableDBOptions& db_options,
  }

  if (result.level0_file_num_compaction_trigger == 0) {
-    Warn(db_options.info_log.get(),
-         "level0_file_num_compaction_trigger cannot be 0");
+    ROCKS_LOG_WARN(db_options.info_log.get(),
+                   "level0_file_num_compaction_trigger cannot be 0");
    result.level0_file_num_compaction_trigger = 1;
  }

@@ -214,14 +215,14 @@ ColumnFamilyOptions SanitizeOptions(const ImmutableDBOptions& db_options,
          result.level0_slowdown_writes_trigger ||
      result.level0_slowdown_writes_trigger <
          result.level0_file_num_compaction_trigger) {
-    Warn(db_options.info_log.get(),
-         "This condition must be satisfied: "
-         "level0_stop_writes_trigger(%d) >= "
-         "level0_slowdown_writes_trigger(%d) >= "
-         "level0_file_num_compaction_trigger(%d)",
-         result.level0_stop_writes_trigger,
-         result.level0_slowdown_writes_trigger,
-         result.level0_file_num_compaction_trigger);
+    ROCKS_LOG_WARN(db_options.info_log.get(),
+                   "This condition must be satisfied: "
+                   "level0_stop_writes_trigger(%d) >= "
+                   "level0_slowdown_writes_trigger(%d) >= "
+                   "level0_file_num_compaction_trigger(%d)",
+                   result.level0_stop_writes_trigger,
+                   result.level0_slowdown_writes_trigger,
+                   result.level0_file_num_compaction_trigger);
    if (result.level0_slowdown_writes_trigger <
        result.level0_file_num_compaction_trigger) {
      result.level0_slowdown_writes_trigger =
@@ -231,14 +232,14 @@ ColumnFamilyOptions SanitizeOptions(const ImmutableDBOptions& db_options,
        result.level0_slowdown_writes_trigger) {
      result.level0_stop_writes_trigger = result.level0_slowdown_writes_trigger;
    }
-    Warn(db_options.info_log.get(),
-         "Adjust the value to "
-         "level0_stop_writes_trigger(%d)"
-         "level0_slowdown_writes_trigger(%d)"
-         "level0_file_num_compaction_trigger(%d)",
-         result.level0_stop_writes_trigger,
-         result.level0_slowdown_writes_trigger,
-         result.level0_file_num_compaction_trigger);
+    ROCKS_LOG_WARN(db_options.info_log.get(),
+                   "Adjust the value to "
+                   "level0_stop_writes_trigger(%d)"
+                   "level0_slowdown_writes_trigger(%d)"
+                   "level0_file_num_compaction_trigger(%d)",
+                   result.level0_stop_writes_trigger,
+                   result.level0_slowdown_writes_trigger,
+                   result.level0_file_num_compaction_trigger);
  }

  if (result.soft_pending_compaction_bytes_limit == 0) {
@@ -386,27 +387,27 @@ ColumnFamilyData::ColumnFamilyData(
    } else if (ioptions_.compaction_style == kCompactionStyleNone) {
      compaction_picker_.reset(new NullCompactionPicker(
          ioptions_, &internal_comparator_));
-      Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log,
-          "Column family %s does not use any background compaction. "
-          "Compactions can only be done via CompactFiles\n",
-          GetName().c_str());
+      ROCKS_LOG_WARN(ioptions_.info_log,
+                     "Column family %s does not use any background compaction. "
+                     "Compactions can only be done via CompactFiles\n",
+                     GetName().c_str());
 #endif  // !ROCKSDB_LITE
    } else {
-      Log(InfoLogLevel::ERROR_LEVEL, ioptions_.info_log,
-          "Unable to recognize the specified compaction style %d. "
-          "Column family %s will use kCompactionStyleLevel.\n",
-          ioptions_.compaction_style, GetName().c_str());
+      ROCKS_LOG_ERROR(ioptions_.info_log,
+                      "Unable to recognize the specified compaction style %d. "
+                      "Column family %s will use kCompactionStyleLevel.\n",
+                      ioptions_.compaction_style, GetName().c_str());
      compaction_picker_.reset(
          new LevelCompactionPicker(ioptions_, &internal_comparator_));
    }

    if (column_family_set_->NumberOfColumnFamilies() < 10) {
-      Log(InfoLogLevel::INFO_LEVEL, ioptions_.info_log,
-          "--------------- Options for column family [%s]:\n", name.c_str());
+      ROCKS_LOG_INFO(ioptions_.info_log,
+                     "--------------- Options for column family [%s]:\n",
+                     name.c_str());
      initial_cf_options_.Dump(ioptions_.info_log);
    } else {
-      Log(InfoLogLevel::INFO_LEVEL, ioptions_.info_log,
-          "\t(skipping printing options)\n");
+      ROCKS_LOG_INFO(ioptions_.info_log, "\t(skipping printing options)\n");
    }
  }

@@ -620,7 +621,8 @@ void ColumnFamilyData::RecalculateWriteStallConditions(
    if (imm()->NumNotFlushed() >= mutable_cf_options.max_write_buffer_number) {
      write_controller_token_ = write_controller->GetStopToken();
      internal_stats_->AddCFStats(InternalStats::MEMTABLE_COMPACTION, 1);
-      Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log,
+      ROCKS_LOG_WARN(
+          ioptions_.info_log,
          "[%s] Stopping writes because we have %d immutable memtables "
          "(waiting for flush), max_write_buffer_number is set to %d",
          name_.c_str(), imm()->NumNotFlushed(),
@@ -634,9 +636,9 @@ void ColumnFamilyData::RecalculateWriteStallConditions(
        internal_stats_->AddCFStats(
            InternalStats::LEVEL0_NUM_FILES_WITH_COMPACTION, 1);
      }
-      Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log,
-          "[%s] Stopping writes because we have %d level-0 files",
-          name_.c_str(), vstorage->l0_delay_trigger_count());
+      ROCKS_LOG_WARN(ioptions_.info_log,
+                     "[%s] Stopping writes because we have %d level-0 files",
+                     name_.c_str(), vstorage->l0_delay_trigger_count());
    } else if (!mutable_cf_options.disable_auto_compactions &&
               mutable_cf_options.hard_pending_compaction_bytes_limit > 0 &&
               compaction_needed_bytes >=
@@ -644,7 +646,8 @@ void ColumnFamilyData::RecalculateWriteStallConditions(
      write_controller_token_ = write_controller->GetStopToken();
      internal_stats_->AddCFStats(
          InternalStats::HARD_PENDING_COMPACTION_BYTES_LIMIT, 1);
-      Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log,
+      ROCKS_LOG_WARN(
+          ioptions_.info_log,
          "[%s] Stopping writes because of estimated pending compaction "
          "bytes %" PRIu64,
          name_.c_str(), compaction_needed_bytes);
@@ -656,7 +659,8 @@ void ColumnFamilyData::RecalculateWriteStallConditions(
                     prev_compaction_needed_bytes_, was_stopped,
                     mutable_cf_options.disable_auto_compactions);
      internal_stats_->AddCFStats(InternalStats::MEMTABLE_SLOWDOWN, 1);
-      Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log,
+      ROCKS_LOG_WARN(
+          ioptions_.info_log,
          "[%s] Stalling writes because we have %d immutable memtables "
          "(waiting for flush), max_write_buffer_number is set to %d "
          "rate %" PRIu64,
@@ -679,11 +683,11 @@ void ColumnFamilyData::RecalculateWriteStallConditions(
        internal_stats_->AddCFStats(
            InternalStats::LEVEL0_SLOWDOWN_WITH_COMPACTION, 1);
      }
-      Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log,
-          "[%s] Stalling writes because we have %d level-0 files "
-          "rate %" PRIu64,
-          name_.c_str(), vstorage->l0_delay_trigger_count(),
-          write_controller->delayed_write_rate());
+      ROCKS_LOG_WARN(ioptions_.info_log,
+                     "[%s] Stalling writes because we have %d level-0 files "
+                     "rate %" PRIu64,
+                     name_.c_str(), vstorage->l0_delay_trigger_count(),
+                     write_controller->delayed_write_rate());
    } else if (!mutable_cf_options.disable_auto_compactions &&
               mutable_cf_options.soft_pending_compaction_bytes_limit > 0 &&
               vstorage->estimated_compaction_needed_bytes() >=
@@ -705,7 +709,8 @@ void ColumnFamilyData::RecalculateWriteStallConditions(
                     mutable_cf_options.disable_auto_compactions);
      internal_stats_->AddCFStats(
          InternalStats::SOFT_PENDING_COMPACTION_BYTES_LIMIT, 1);
-      Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log,
+      ROCKS_LOG_WARN(
+          ioptions_.info_log,
          "[%s] Stalling writes because of estimated pending compaction "
          "bytes %" PRIu64 " rate %" PRIu64,
          name_.c_str(), vstorage->estimated_compaction_needed_bytes(),
@@ -717,7 +722,8 @@ void ColumnFamilyData::RecalculateWriteStallConditions(
              mutable_cf_options.level0_slowdown_writes_trigger)) {
        write_controller_token_ =
            write_controller->GetCompactionPressureToken();
-        Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log,
+        ROCKS_LOG_WARN(
+            ioptions_.info_log,
            "[%s] Increasing compaction threads because we have %d level-0 "
            "files ",
            name_.c_str(), vstorage->l0_delay_trigger_count());
@@ -730,7 +736,8 @@ void ColumnFamilyData::RecalculateWriteStallConditions(
        write_controller_token_ =
            write_controller->GetCompactionPressureToken();
        if (mutable_cf_options.soft_pending_compaction_bytes_limit > 0) {
-          Log(InfoLogLevel::WARN_LEVEL, ioptions_.info_log,
+          ROCKS_LOG_WARN(
+              ioptions_.info_log,
              "[%s] Increasing compaction threads because of estimated pending "
              "compaction "
              "bytes %" PRIu64,
@@ -19,11 +19,11 @@
 #include "db/table_properties_collector.h"
 #include "db/write_batch_internal.h"
 #include "db/write_controller.h"
+#include "options/cf_options.h"
 #include "rocksdb/compaction_job_stats.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
 #include "rocksdb/options.h"
-#include "util/cf_options.h"
 #include "util/thread_local.h"

 namespace rocksdb {
@@ -14,13 +14,13 @@

 #include "db/db_impl.h"
 #include "db/db_test_util.h"
+#include "options/options_parser.h"
 #include "port/port.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
 #include "rocksdb/iterator.h"
 #include "util/coding.h"
 #include "util/fault_injection_test_env.h"
-#include "util/options_parser.h"
 #include "util/string_util.h"
 #include "util/sync_point.h"
 #include "util/testharness.h"
@@ -675,6 +675,44 @@ 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
@@ -1730,7 +1768,7 @@ TEST_F(ColumnFamilyTest, SameCFManualAutomaticCompactionsLevel) {

  one.num_levels = 1;
  // trigger compaction if there are >= 4 files
-  one.level0_file_num_compaction_trigger = 4;
+  one.level0_file_num_compaction_trigger = 3;
  one.write_buffer_size = 120000;

  Reopen({default_cf, one});
@@ -42,8 +42,8 @@ size_t CompactedDBImpl::FindFile(const Slice& key) {
  return right;
 }

-Status CompactedDBImpl::Get(const ReadOptions& options,
-     ColumnFamilyHandle*, const Slice& key, std::string* value) {
+Status CompactedDBImpl::Get(const ReadOptions& options, ColumnFamilyHandle*,
+                            const Slice& key, PinnableSlice* value) {
  GetContext get_context(user_comparator_, nullptr, nullptr, nullptr,
                         GetContext::kNotFound, key, value, nullptr, nullptr,
                         nullptr, nullptr);
@@ -75,11 +75,14 @@ std::vector<Status> CompactedDBImpl::MultiGet(const ReadOptions& options,
  int idx = 0;
  for (auto* r : reader_list) {
    if (r != nullptr) {
+      PinnableSlice pinnable_val;
+      std::string& value = (*values)[idx];
      GetContext get_context(user_comparator_, nullptr, nullptr, nullptr,
-                             GetContext::kNotFound, keys[idx], &(*values)[idx],
+                             GetContext::kNotFound, keys[idx], &pinnable_val,
                             nullptr, nullptr, nullptr, nullptr);
      LookupKey lkey(keys[idx], kMaxSequenceNumber);
      r->Get(options, lkey.internal_key(), &get_context);
+      value.assign(pinnable_val.data(), pinnable_val.size());
      if (get_context.State() == GetContext::kFound) {
        statuses[idx] = Status::OK();
      }
@@ -151,8 +154,8 @@ Status CompactedDBImpl::Open(const Options& options,
  std::unique_ptr<CompactedDBImpl> db(new CompactedDBImpl(db_options, dbname));
  Status s = db->Init(options);
  if (s.ok()) {
-    Log(INFO_LEVEL, db->immutable_db_options_.info_log,
-        "Opened the db as fully compacted mode");
+    ROCKS_LOG_INFO(db->immutable_db_options_.info_log,
+                   "Opened the db as fully compacted mode");
    LogFlush(db->immutable_db_options_.info_log);
    *dbptr = db.release();
  }
@@ -23,7 +23,7 @@ class CompactedDBImpl : public DBImpl {
  using DB::Get;
  virtual Status Get(const ReadOptions& options,
                     ColumnFamilyHandle* column_family, const Slice& key,
-                     std::string* value) override;
+                     PinnableSlice* value) override;
  using DB::MultiGet;
  virtual std::vector<Status> MultiGet(
      const ReadOptions& options,
@@ -18,7 +18,7 @@

 #include "db/column_family.h"
 #include "rocksdb/compaction_filter.h"
-#include "util/logging.h"
+#include "util/string_util.h"
 #include "util/sync_point.h"

 namespace rocksdb {
@@ -287,6 +287,10 @@ bool Compaction::KeyNotExistsBeyondOutputLevel(
  if (cfd_->ioptions()->compaction_style == kCompactionStyleUniversal) {
    return bottommost_level_;
  }
+  if (cfd_->ioptions()->compaction_style == kCompactionStyleLevel &&
+      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++) {
@@ -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
@@ -450,7 +455,7 @@ bool Compaction::ShouldFormSubcompactions() const {
    return false;
  }
  if (cfd_->ioptions()->compaction_style == kCompactionStyleLevel) {
-    return start_level_ == 0 && !IsOutputLevelEmpty();
+    return start_level_ == 0 && output_level_ > 0 && !IsOutputLevelEmpty();
  } else if (cfd_->ioptions()->compaction_style == kCompactionStyleUniversal) {
    return number_levels_ > 1 && output_level_ > 0;
  } else {
@@ -9,9 +9,9 @@

 #pragma once
 #include "db/version_set.h"
+#include "options/cf_options.h"
 #include "util/arena.h"
 #include "util/autovector.h"
-#include "util/cf_options.h"

 namespace rocksdb {

@@ -8,12 +8,36 @@

 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,
@@ -21,7 +45,7 @@ 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,
@@ -30,6 +54,7 @@ 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),
@@ -41,7 +66,9 @@ CompactionIterator::CompactionIterator(
      range_del_agg_(range_del_agg),
      compaction_(std::move(compaction)),
      compaction_filter_(compaction_filter),
+      compaction_listener_(compaction_listener),
      shutting_down_(shutting_down),
+      ignore_snapshots_(false),
      merge_out_iter_(merge_helper_) {
  assert(compaction_filter_ == nullptr || compaction_ != nullptr);
  bottommost_level_ =
@@ -60,8 +87,10 @@ CompactionIterator::CompactionIterator(
    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;
  }
@@ -103,7 +132,7 @@ void CompactionIterator::Next() {
      assert(valid_key);
      // Keep current_key_ in sync.
      current_key_.UpdateInternalKey(ikey_.sequence, ikey_.type);
-      key_ = current_key_.GetKey();
+      key_ = current_key_.GetInternalKey();
      ikey_.user_key = current_key_.GetUserKey();
      valid_ = true;
    } else {
@@ -149,7 +178,7 @@ void CompactionIterator::NextFromInput() {
        status_ = Status::Corruption("Corrupted internal key not expected.");
        break;
      }
-      key_ = current_key_.SetKey(key_);
+      key_ = current_key_.SetInternalKey(key_);
      has_current_user_key_ = false;
      current_user_key_sequence_ = kMaxSequenceNumber;
      current_user_key_snapshot_ = 0;
@@ -179,13 +208,21 @@ void CompactionIterator::NextFromInput() {
        !cmp_->Equal(ikey_.user_key, current_user_key_)) {
      // First occurrence of this user key
      // Copy key for output
-      key_ = current_key_.SetKey(key_, &ikey_);
+      key_ = current_key_.SetInternalKey(key_, &ikey_);
      current_user_key_ = ikey_.user_key;
      has_current_user_key_ = true;
      has_outputted_key_ = false;
      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_ ||
@@ -233,13 +270,21 @@ 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
      // Need to have the compaction filter process multiple versions
      // if we have versions on both sides of a snapshot
      current_key_.UpdateInternalKey(ikey_.sequence, ikey_.type);
-      key_ = current_key_.GetKey();
+      key_ = current_key_.GetInternalKey();
      ikey_.user_key = current_key_.GetUserKey();
    }

@@ -392,7 +437,7 @@ 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
@@ -451,7 +496,7 @@ void CompactionIterator::NextFromInput() {
        assert(valid_key);
        // Keep current_key_ in sync.
        current_key_.UpdateInternalKey(ikey_.sequence, ikey_.type);
-        key_ = current_key_.GetKey();
+        key_ = current_key_.GetInternalKey();
        ikey_.user_key = current_key_.GetUserKey();
        valid_ = true;
      } else {
@@ -18,6 +18,8 @@

 namespace rocksdb {

+class CompactionEventListener;
+
 class CompactionIterator {
 public:
  // A wrapper around Compaction. Has a much smaller interface, only what
@@ -58,6 +60,7 @@ 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.
@@ -69,6 +72,7 @@ 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();
@@ -122,6 +126,7 @@ class CompactionIterator {
  RangeDelAggregator* range_del_agg_;
  std::unique_ptr<CompactionProxy> compaction_;
  const CompactionFilter* compaction_filter_;
+  CompactionEventListener* compaction_listener_;
  const std::atomic<bool>* shutting_down_;
  bool bottommost_level_;
  bool valid_ = false;
@@ -189,7 +189,7 @@ 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); }
@@ -28,7 +28,6 @@
 #include "db/db_iter.h"
 #include "db/dbformat.h"
 #include "db/event_helpers.h"
-#include "db/filename.h"
 #include "db/log_reader.h"
 #include "db/log_writer.h"
 #include "db/memtable.h"
@@ -36,6 +35,9 @@
 #include "db/merge_context.h"
 #include "db/merge_helper.h"
 #include "db/version_set.h"
+#include "monitoring/iostats_context_imp.h"
+#include "monitoring/perf_context_imp.h"
+#include "monitoring/thread_status_util.h"
 #include "port/likely.h"
 #include "port/port.h"
 #include "rocksdb/db.h"
@@ -49,16 +51,15 @@
 #include "table/table_builder.h"
 #include "util/coding.h"
 #include "util/file_reader_writer.h"
-#include "util/iostats_context_imp.h"
+#include "util/filename.h"
 #include "util/log_buffer.h"
 #include "util/logging.h"
-#include "util/sst_file_manager_impl.h"
 #include "util/mutexlock.h"
-#include "util/perf_context_imp.h"
+#include "util/random.h"
+#include "util/sst_file_manager_impl.h"
 #include "util/stop_watch.h"
 #include "util/string_util.h"
 #include "util/sync_point.h"
-#include "util/thread_status_util.h"

 namespace rocksdb {

@@ -600,7 +601,7 @@ Status CompactionJob::Install(const MutableCFOptions& mutable_cf_options) {
    write_amp = stats.bytes_written /
                static_cast<double>(stats.bytes_read_non_output_levels);
  }
-  LogToBuffer(
+  ROCKS_LOG_BUFFER(
      log_buffer_,
      "[%s] compacted to: %s, MB/sec: %.1f rd, %.1f wr, level %d, "
      "files in(%d, %d) out(%d) "
@@ -733,17 +734,28 @@ void CompactionJob::ProcessKeyValueCompaction(SubcompactionState* sub_compact) {
  if (start != nullptr) {
    IterKey start_iter;
    start_iter.SetInternalKey(*start, kMaxSequenceNumber, kValueTypeForSeek);
-    input->Seek(start_iter.GetKey());
+    input->Seek(start_iter.GetInternalKey());
  } else {
    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() &&
@@ -1071,6 +1083,11 @@ Status CompactionJob::FinishCompactionOutputFile(
  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,
@@ -1091,12 +1108,12 @@ Status CompactionJob::FinishCompactionOutputFile(
      tp = sub_compact->builder->GetTableProperties();
      sub_compact->current_output()->table_properties =
          std::make_shared<TableProperties>(tp);
-      Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log,
-          "[%s] [JOB %d] Generated table #%" PRIu64 ": %" PRIu64
-          " keys, %" PRIu64 " bytes%s",
-          cfd->GetName().c_str(), job_id_, output_number, current_entries,
-          current_bytes,
-          meta->marked_for_compaction ? " (need compaction)" : "");
+      ROCKS_LOG_INFO(db_options_.info_log,
+                     "[%s] [JOB %d] Generated table #%" PRIu64 ": %" PRIu64
+                     " keys, %" PRIu64 " bytes%s",
+                     cfd->GetName().c_str(), job_id_, output_number,
+                     current_entries, current_bytes,
+                     meta->marked_for_compaction ? " (need compaction)" : "");
    }
  }
  std::string fname = TableFileName(db_options_.db_paths, meta->fd.GetNumber(),
@@ -1142,17 +1159,16 @@ Status CompactionJob::InstallCompactionResults(
  if (!versions_->VerifyCompactionFileConsistency(compaction)) {
    Compaction::InputLevelSummaryBuffer inputs_summary;

-    Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log,
-        "[%s] [JOB %d] Compaction %s aborted",
-        compaction->column_family_data()->GetName().c_str(), job_id_,
-        compaction->InputLevelSummary(&inputs_summary));
+    ROCKS_LOG_ERROR(db_options_.info_log, "[%s] [JOB %d] Compaction %s aborted",
+                    compaction->column_family_data()->GetName().c_str(),
+                    job_id_, compaction->InputLevelSummary(&inputs_summary));
    return Status::Corruption("Compaction input files inconsistent");
  }

  {
    Compaction::InputLevelSummaryBuffer inputs_summary;
-    Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log,
-        "[%s] [JOB %d] Compacted %s => %" PRIu64 " bytes",
+    ROCKS_LOG_INFO(
+        db_options_.info_log, "[%s] [JOB %d] Compacted %s => %" PRIu64 " bytes",
        compaction->column_family_data()->GetName().c_str(), job_id_,
        compaction->InputLevelSummary(&inputs_summary), compact_->total_bytes);
  }
@@ -1198,9 +1214,14 @@ 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()) {
-    Log(InfoLogLevel::ERROR_LEVEL, db_options_.info_log,
+    ROCKS_LOG_ERROR(
+        db_options_.info_log,
        "[%s] [JOB %d] OpenCompactionOutputFiles for table #%" PRIu64
        " fails at NewWritableFile with status %s",
        sub_compact->compaction->column_family_data()->GetName().c_str(),
@@ -1376,14 +1397,14 @@ void CompactionJob::LogCompaction() {
  // we're not logging
  if (db_options_.info_log_level <= InfoLogLevel::INFO_LEVEL) {
    Compaction::InputLevelSummaryBuffer inputs_summary;
-    Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log,
-        "[%s] [JOB %d] Compacting %s, score %.2f", cfd->GetName().c_str(),
-        job_id_, compaction->InputLevelSummary(&inputs_summary),
-        compaction->score());
+    ROCKS_LOG_INFO(
+        db_options_.info_log, "[%s] [JOB %d] Compacting %s, score %.2f",
+        cfd->GetName().c_str(), job_id_,
+        compaction->InputLevelSummary(&inputs_summary), compaction->score());
    char scratch[2345];
    compaction->Summary(scratch, sizeof(scratch));
-    Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log,
-        "[%s] Compaction start summary: %s\n", cfd->GetName().c_str(), scratch);
+    ROCKS_LOG_INFO(db_options_.info_log, "[%s] Compaction start summary: %s\n",
+                   cfd->GetName().c_str(), scratch);
    // build event logger report
    auto stream = event_logger_->Log();
    stream << "job" << job_id_ << "event"
@@ -29,6 +29,7 @@
 #include "db/version_edit.h"
 #include "db/write_controller.h"
 #include "db/write_thread.h"
+#include "options/db_options.h"
 #include "port/port.h"
 #include "rocksdb/compaction_filter.h"
 #include "rocksdb/compaction_job_stats.h"
@@ -38,7 +39,6 @@
 #include "rocksdb/transaction_log.h"
 #include "table/scoped_arena_iterator.h"
 #include "util/autovector.h"
-#include "util/db_options.h"
 #include "util/event_logger.h"
 #include "util/stop_watch.h"
 #include "util/thread_local.h"
@@ -23,11 +23,13 @@

 #include "db/db_impl.h"
 #include "db/dbformat.h"
-#include "db/filename.h"
 #include "db/job_context.h"
 #include "db/version_set.h"
 #include "db/write_batch_internal.h"
+#include "env/mock_env.h"
 #include "memtable/hash_linklist_rep.h"
+#include "monitoring/statistics.h"
+#include "monitoring/thread_status_util.h"
 #include "port/stack_trace.h"
 #include "rocksdb/cache.h"
 #include "rocksdb/compaction_filter.h"
@@ -50,17 +52,15 @@
 #include "table/plain_table_factory.h"
 #include "table/scoped_arena_iterator.h"
 #include "util/compression.h"
+#include "util/filename.h"
 #include "util/hash.h"
 #include "util/logging.h"
-#include "util/mock_env.h"
 #include "util/mutexlock.h"
 #include "util/rate_limiter.h"
-#include "util/statistics.h"
 #include "util/string_util.h"
 #include "util/sync_point.h"
 #include "util/testharness.h"
 #include "util/testutil.h"
-#include "util/thread_status_util.h"
 #include "utilities/merge_operators.h"

 #if !defined(IOS_CROSS_COMPILE)
@@ -17,10 +17,10 @@

 #include "db/compaction.h"
 #include "db/version_set.h"
+#include "options/cf_options.h"
 #include "rocksdb/env.h"
 #include "rocksdb/options.h"
 #include "rocksdb/status.h"
-#include "util/cf_options.h"

 namespace rocksdb {

@@ -84,15 +84,15 @@ class CompactionPicker {
  void ReleaseCompactionFiles(Compaction* c, Status status);

  // Returns true if any one of the specified files are being compacted
-  bool FilesInCompaction(const std::vector<FileMetaData*>& files);
+  bool AreFilesInCompaction(const std::vector<FileMetaData*>& files);

  // Takes a list of CompactionInputFiles and returns a (manual) Compaction
  // object.
-  Compaction* FormCompaction(
-      const CompactionOptions& compact_options,
-      const std::vector<CompactionInputFiles>& input_files, int output_level,
-      VersionStorageInfo* vstorage, const MutableCFOptions& mutable_cf_options,
-      uint32_t output_path_id);
+  Compaction* CompactFiles(const CompactionOptions& compact_options,
+                           const std::vector<CompactionInputFiles>& input_files,
+                           int output_level, VersionStorageInfo* vstorage,
+                           const MutableCFOptions& mutable_cf_options,
+                           uint32_t output_path_id);

  // Converts a set of compaction input file numbers into
  // a list of CompactionInputFiles.
@@ -102,12 +102,6 @@ class CompactionPicker {
      const VersionStorageInfo* vstorage,
      const CompactionOptions& compact_options) const;

-  // Used in universal compaction when the enabled_trivial_move
-  // option is set. Checks whether there are any overlapping files
-  // in the input. Returns true if the input files are non
-  // overlapping.
-  bool IsInputNonOverlapping(Compaction* c);
-
  // Is there currently a compaction involving level 0 taking place
  bool IsLevel0CompactionInProgress() const {
    return !level0_compactions_in_progress_.empty();
@@ -138,7 +132,6 @@ class CompactionPicker {
  void GetRange(const std::vector<CompactionInputFiles>& inputs,
                InternalKey* smallest, InternalKey* largest) const;

- protected:
  int NumberLevels() const { return ioptions_.num_levels; }

  // Add more files to the inputs on "level" to make sure that
@@ -151,14 +144,14 @@ class CompactionPicker {
  // populated.
  //
  // Will return false if it is impossible to apply this compaction.
-  bool ExpandWhileOverlapping(const std::string& cf_name,
+  bool ExpandInputsToCleanCut(const std::string& cf_name,
                              VersionStorageInfo* vstorage,
                              CompactionInputFiles* inputs);

  // Returns true if any one of the parent files are being compacted
-  bool RangeInCompaction(VersionStorageInfo* vstorage,
-                         const InternalKey* smallest,
-                         const InternalKey* largest, int level, int* index);
+  bool IsRangeInCompaction(VersionStorageInfo* vstorage,
+                           const InternalKey* smallest,
+                           const InternalKey* largest, int level, int* index);

  // Returns true if the key range that `inputs` files cover overlap with the
  // key range of a currently running compaction.
@@ -177,6 +170,20 @@ class CompactionPicker {
                       const CompactionInputFiles& output_level_inputs,
                       std::vector<FileMetaData*>* grandparents);

+  // Register this compaction in the set of running compactions
+  void RegisterCompaction(Compaction* c);
+
+  // Remove this compaction from the set of running compactions
+  void UnregisterCompaction(Compaction* c);
+
+  std::set<Compaction*>* level0_compactions_in_progress() {
+    return &level0_compactions_in_progress_;
+  }
+  std::unordered_set<Compaction*>* compactions_in_progress() {
+    return &compactions_in_progress_;
+  }
+
+ protected:
  const ImmutableCFOptions& ioptions_;

 // A helper function to SanitizeCompactionInputFiles() that
@@ -187,12 +194,6 @@ class CompactionPicker {
      const ColumnFamilyMetaData& cf_meta, const int output_level) const;
 #endif  // ROCKSDB_LITE

-  // Register this compaction in the set of running compactions
-  void RegisterCompaction(Compaction* c);
-
-  // Remove this compaction from the set of running compactions
-  void UnregisterCompaction(Compaction* c);
-
  // Keeps track of all compactions that are running on Level0.
  // Protected by DB mutex
  std::set<Compaction*> level0_compactions_in_progress_;
@@ -216,101 +217,9 @@ class LevelCompactionPicker : public CompactionPicker {

  virtual bool NeedsCompaction(
      const VersionStorageInfo* vstorage) const override;
-
-  // Pick a path ID to place a newly generated file, with its level
-  static uint32_t GetPathId(const ImmutableCFOptions& ioptions,
-                            const MutableCFOptions& mutable_cf_options,
-                            int level);
-
- private:
-  // For the specfied level, pick a file that we want to compact.
-  // Returns false if there is no file to compact.
-  // If it returns true, inputs->files.size() will be exactly one.
-  // If level is 0 and there is already a compaction on that level, this
-  // function will return false.
-  bool PickCompactionBySize(VersionStorageInfo* vstorage, int level,
-                            int output_level, CompactionInputFiles* inputs,
-                            int* parent_index, int* base_index);
-
-  // If there is any file marked for compaction, put put it into inputs.
-  // This is still experimental. It will return meaningful results only if
-  // clients call experimental feature SuggestCompactRange()
-  void PickFilesMarkedForCompactionExperimental(const std::string& cf_name,
-                                                VersionStorageInfo* vstorage,
-                                                CompactionInputFiles* inputs,
-                                                int* level, int* output_level);
 };

 #ifndef ROCKSDB_LITE
-class UniversalCompactionPicker : public CompactionPicker {
- public:
-  UniversalCompactionPicker(const ImmutableCFOptions& ioptions,
-                            const InternalKeyComparator* icmp)
-      : CompactionPicker(ioptions, icmp) {}
-  virtual Compaction* PickCompaction(const std::string& cf_name,
-                                     const MutableCFOptions& mutable_cf_options,
-                                     VersionStorageInfo* vstorage,
-                                     LogBuffer* log_buffer) override;
-
-  virtual int MaxOutputLevel() const override { return NumberLevels() - 1; }
-
-  virtual bool NeedsCompaction(
-      const VersionStorageInfo* vstorage) const override;
-
- private:
-  struct SortedRun {
-    SortedRun(int _level, FileMetaData* _file, uint64_t _size,
-              uint64_t _compensated_file_size, bool _being_compacted)
-        : level(_level),
-          file(_file),
-          size(_size),
-          compensated_file_size(_compensated_file_size),
-          being_compacted(_being_compacted) {
-      assert(compensated_file_size > 0);
-      assert(level != 0 || file != nullptr);
-    }
-
-    void Dump(char* out_buf, size_t out_buf_size,
-              bool print_path = false) const;
-
-    // sorted_run_count is added into the string to print
-    void DumpSizeInfo(char* out_buf, size_t out_buf_size,
-                      size_t sorted_run_count) const;
-
-    int level;
-    // `file` Will be null for level > 0. For level = 0, the sorted run is
-    // for this file.
-    FileMetaData* file;
-    // For level > 0, `size` and `compensated_file_size` are sum of sizes all
-    // files in the level. `being_compacted` should be the same for all files
-    // in a non-zero level. Use the value here.
-    uint64_t size;
-    uint64_t compensated_file_size;
-    bool being_compacted;
-  };
-
-  // Pick Universal compaction to limit read amplification
-  Compaction* PickCompactionUniversalReadAmp(
-      const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
-      VersionStorageInfo* vstorage, double score, unsigned int ratio,
-      unsigned int num_files, const std::vector<SortedRun>& sorted_runs,
-      LogBuffer* log_buffer);
-
-  // Pick Universal compaction to limit space amplification.
-  Compaction* PickCompactionUniversalSizeAmp(
-      const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
-      VersionStorageInfo* vstorage, double score,
-      const std::vector<SortedRun>& sorted_runs, LogBuffer* log_buffer);
-
-  static std::vector<SortedRun> CalculateSortedRuns(
-      const VersionStorageInfo& vstorage, const ImmutableCFOptions& ioptions);
-
-  // Pick a path ID to place a newly generated file, with its estimated file
-  // size.
-  static uint32_t GetPathId(const ImmutableCFOptions& ioptions,
-                            uint64_t file_size);
-};
-
 class FIFOCompactionPicker : public CompactionPicker {
 public:
  FIFOCompactionPicker(const ImmutableCFOptions& ioptions,
@@ -3,11 +3,12 @@
 //  COPYING file in the root directory) and Apache 2.0 License
 //  (found in the LICENSE.Apache file in the root directory).

-#include "db/compaction.h"
 #include "db/compaction_picker.h"
 #include <limits>
 #include <string>
 #include <utility>
+#include "db/compaction.h"
+#include "db/compaction_picker_universal.h"

 #include "util/logging.h"
 #include "util/string_util.h"
@@ -197,7 +198,7 @@ TEST_F(CompactionPickerTest, LevelMaxScore) {
  mutable_cf_options_.target_file_size_base = 10000000;
  mutable_cf_options_.target_file_size_multiplier = 10;
  mutable_cf_options_.max_bytes_for_level_base = 10 * 1024 * 1024;
-  Add(0, 1U, "150", "200", 1000000000U);
+  Add(0, 1U, "150", "200", 1000000U);
  // Level 1 score 1.2
  Add(1, 66U, "150", "200", 6000000U);
  Add(1, 88U, "201", "300", 6000000U);
@@ -0,0 +1,735 @@
+//  Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+//
+// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file. See the AUTHORS file for names of contributors.
+
+#include "db/compaction_picker_universal.h"
+#ifndef ROCKSDB_LITE
+
+#ifndef __STDC_FORMAT_MACROS
+#define __STDC_FORMAT_MACROS
+#endif
+
+#include <inttypes.h>
+#include <limits>
+#include <queue>
+#include <string>
+#include <utility>
+#include "db/column_family.h"
+#include "monitoring/statistics.h"
+#include "util/filename.h"
+#include "util/log_buffer.h"
+#include "util/random.h"
+#include "util/string_util.h"
+#include "util/sync_point.h"
+
+namespace rocksdb {
+namespace {
+// Used in universal compaction when trivial move is enabled.
+// This structure is used for the construction of min heap
+// that contains the file meta data, the level of the file
+// and the index of the file in that level
+
+struct InputFileInfo {
+  InputFileInfo() : f(nullptr) {}
+
+  FileMetaData* f;
+  size_t level;
+  size_t index;
+};
+
+// Used in universal compaction when trivial move is enabled.
+// This comparator is used for the construction of min heap
+// based on the smallest key of the file.
+struct SmallestKeyHeapComparator {
+  explicit SmallestKeyHeapComparator(const Comparator* ucmp) { ucmp_ = ucmp; }
+
+  bool operator()(InputFileInfo i1, InputFileInfo i2) const {
+    return (ucmp_->Compare(i1.f->smallest.user_key(),
+                           i2.f->smallest.user_key()) > 0);
+  }
+
+ private:
+  const Comparator* ucmp_;
+};
+
+typedef std::priority_queue<InputFileInfo, std::vector<InputFileInfo>,
+                            SmallestKeyHeapComparator>
+    SmallestKeyHeap;
+
+// This function creates the heap that is used to find if the files are
+// overlapping during universal compaction when the allow_trivial_move
+// is set.
+SmallestKeyHeap create_level_heap(Compaction* c, const Comparator* ucmp) {
+  SmallestKeyHeap smallest_key_priority_q =
+      SmallestKeyHeap(SmallestKeyHeapComparator(ucmp));
+
+  InputFileInfo input_file;
+
+  for (size_t l = 0; l < c->num_input_levels(); l++) {
+    if (c->num_input_files(l) != 0) {
+      if (l == 0 && c->start_level() == 0) {
+        for (size_t i = 0; i < c->num_input_files(0); i++) {
+          input_file.f = c->input(0, i);
+          input_file.level = 0;
+          input_file.index = i;
+          smallest_key_priority_q.push(std::move(input_file));
+        }
+      } else {
+        input_file.f = c->input(l, 0);
+        input_file.level = l;
+        input_file.index = 0;
+        smallest_key_priority_q.push(std::move(input_file));
+      }
+    }
+  }
+  return smallest_key_priority_q;
+}
+
+#ifndef NDEBUG
+// smallest_seqno and largest_seqno are set iff. `files` is not empty.
+void GetSmallestLargestSeqno(const std::vector<FileMetaData*>& files,
+                             SequenceNumber* smallest_seqno,
+                             SequenceNumber* largest_seqno) {
+  bool is_first = true;
+  for (FileMetaData* f : files) {
+    assert(f->smallest_seqno <= f->largest_seqno);
+    if (is_first) {
+      is_first = false;
+      *smallest_seqno = f->smallest_seqno;
+      *largest_seqno = f->largest_seqno;
+    } else {
+      if (f->smallest_seqno < *smallest_seqno) {
+        *smallest_seqno = f->smallest_seqno;
+      }
+      if (f->largest_seqno > *largest_seqno) {
+        *largest_seqno = f->largest_seqno;
+      }
+    }
+  }
+}
+#endif
+}  // namespace
+
+// Algorithm that checks to see if there are any overlapping
+// files in the input
+bool UniversalCompactionPicker::IsInputFilesNonOverlapping(Compaction* c) {
+  auto comparator = icmp_->user_comparator();
+  int first_iter = 1;
+
+  InputFileInfo prev, curr, next;
+
+  SmallestKeyHeap smallest_key_priority_q =
+      create_level_heap(c, icmp_->user_comparator());
+
+  while (!smallest_key_priority_q.empty()) {
+    curr = smallest_key_priority_q.top();
+    smallest_key_priority_q.pop();
+
+    if (first_iter) {
+      prev = curr;
+      first_iter = 0;
+    } else {
+      if (comparator->Compare(prev.f->largest.user_key(),
+                              curr.f->smallest.user_key()) >= 0) {
+        // found overlapping files, return false
+        return false;
+      }
+      assert(comparator->Compare(curr.f->largest.user_key(),
+                                 prev.f->largest.user_key()) > 0);
+      prev = curr;
+    }
+
+    next.f = nullptr;
+
+    if (curr.level != 0 && curr.index < c->num_input_files(curr.level) - 1) {
+      next.f = c->input(curr.level, curr.index + 1);
+      next.level = curr.level;
+      next.index = curr.index + 1;
+    }
+
+    if (next.f) {
+      smallest_key_priority_q.push(std::move(next));
+    }
+  }
+  return true;
+}
+
+bool UniversalCompactionPicker::NeedsCompaction(
+    const VersionStorageInfo* vstorage) const {
+  const int kLevel0 = 0;
+  return vstorage->CompactionScore(kLevel0) >= 1;
+}
+
+void UniversalCompactionPicker::SortedRun::Dump(char* out_buf,
+                                                size_t out_buf_size,
+                                                bool print_path) const {
+  if (level == 0) {
+    assert(file != nullptr);
+    if (file->fd.GetPathId() == 0 || !print_path) {
+      snprintf(out_buf, out_buf_size, "file %" PRIu64, file->fd.GetNumber());
+    } else {
+      snprintf(out_buf, out_buf_size, "file %" PRIu64
+                                      "(path "
+                                      "%" PRIu32 ")",
+               file->fd.GetNumber(), file->fd.GetPathId());
+    }
+  } else {
+    snprintf(out_buf, out_buf_size, "level %d", level);
+  }
+}
+
+void UniversalCompactionPicker::SortedRun::DumpSizeInfo(
+    char* out_buf, size_t out_buf_size, size_t sorted_run_count) const {
+  if (level == 0) {
+    assert(file != nullptr);
+    snprintf(out_buf, out_buf_size,
+             "file %" PRIu64 "[%" ROCKSDB_PRIszt
+             "] "
+             "with size %" PRIu64 " (compensated size %" PRIu64 ")",
+             file->fd.GetNumber(), sorted_run_count, file->fd.GetFileSize(),
+             file->compensated_file_size);
+  } else {
+    snprintf(out_buf, out_buf_size,
+             "level %d[%" ROCKSDB_PRIszt
+             "] "
+             "with size %" PRIu64 " (compensated size %" PRIu64 ")",
+             level, sorted_run_count, size, compensated_file_size);
+  }
+}
+
+std::vector<UniversalCompactionPicker::SortedRun>
+UniversalCompactionPicker::CalculateSortedRuns(
+    const VersionStorageInfo& vstorage, const ImmutableCFOptions& ioptions) {
+  std::vector<UniversalCompactionPicker::SortedRun> ret;
+  for (FileMetaData* f : vstorage.LevelFiles(0)) {
+    ret.emplace_back(0, f, f->fd.GetFileSize(), f->compensated_file_size,
+                     f->being_compacted);
+  }
+  for (int level = 1; level < vstorage.num_levels(); level++) {
+    uint64_t total_compensated_size = 0U;
+    uint64_t total_size = 0U;
+    bool being_compacted = false;
+    bool is_first = true;
+    for (FileMetaData* f : vstorage.LevelFiles(level)) {
+      total_compensated_size += f->compensated_file_size;
+      total_size += f->fd.GetFileSize();
+      if (ioptions.compaction_options_universal.allow_trivial_move == true) {
+        if (f->being_compacted) {
+          being_compacted = f->being_compacted;
+        }
+      } else {
+        // Compaction always includes all files for a non-zero level, so for a
+        // non-zero level, all the files should share the same being_compacted
+        // value.
+        // This assumption is only valid when
+        // ioptions.compaction_options_universal.allow_trivial_move is false
+        assert(is_first || f->being_compacted == being_compacted);
+      }
+      if (is_first) {
+        being_compacted = f->being_compacted;
+        is_first = false;
+      }
+    }
+    if (total_compensated_size > 0) {
+      ret.emplace_back(level, nullptr, total_size, total_compensated_size,
+                       being_compacted);
+    }
+  }
+  return ret;
+}
+
+// Universal style of compaction. Pick files that are contiguous in
+// time-range to compact.
+//
+Compaction* UniversalCompactionPicker::PickCompaction(
+    const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
+    VersionStorageInfo* vstorage, LogBuffer* log_buffer) {
+  const int kLevel0 = 0;
+  double score = vstorage->CompactionScore(kLevel0);
+  std::vector<SortedRun> sorted_runs =
+      CalculateSortedRuns(*vstorage, ioptions_);
+
+  if (sorted_runs.size() == 0 ||
+      sorted_runs.size() <
+          (unsigned int)mutable_cf_options.level0_file_num_compaction_trigger) {
+    ROCKS_LOG_BUFFER(log_buffer, "[%s] Universal: nothing to do\n",
+                     cf_name.c_str());
+    TEST_SYNC_POINT_CALLBACK("UniversalCompactionPicker::PickCompaction:Return",
+                             nullptr);
+    return nullptr;
+  }
+  VersionStorageInfo::LevelSummaryStorage tmp;
+  ROCKS_LOG_BUFFER_MAX_SZ(
+      log_buffer, 3072,
+      "[%s] Universal: sorted runs files(%" ROCKSDB_PRIszt "): %s\n",
+      cf_name.c_str(), sorted_runs.size(), vstorage->LevelSummary(&tmp));
+
+  // Check for size amplification first.
+  Compaction* c;
+  if ((c = PickCompactionToReduceSizeAmp(cf_name, mutable_cf_options, vstorage,
+                                         score, sorted_runs, log_buffer)) !=
+      nullptr) {
+    ROCKS_LOG_BUFFER(log_buffer, "[%s] Universal: compacting for size amp\n",
+                     cf_name.c_str());
+  } else {
+    // Size amplification is within limits. Try reducing read
+    // amplification while maintaining file size ratios.
+    unsigned int ratio = ioptions_.compaction_options_universal.size_ratio;
+
+    if ((c = PickCompactionToReduceSortedRuns(
+             cf_name, mutable_cf_options, vstorage, score, ratio, UINT_MAX,
+             sorted_runs, log_buffer)) != nullptr) {
+      ROCKS_LOG_BUFFER(log_buffer,
+                       "[%s] Universal: compacting for size ratio\n",
+                       cf_name.c_str());
+    } else {
+      // Size amplification and file size ratios are within configured limits.
+      // If max read amplification is exceeding configured limits, then force
+      // compaction without looking at filesize ratios and try to reduce
+      // the number of files to fewer than level0_file_num_compaction_trigger.
+      // This is guaranteed by NeedsCompaction()
+      assert(sorted_runs.size() >=
+             static_cast<size_t>(
+                 mutable_cf_options.level0_file_num_compaction_trigger));
+      // Get the total number of sorted runs that are not being compacted
+      int num_sr_not_compacted = 0;
+      for (size_t i = 0; i < sorted_runs.size(); i++) {
+        if (sorted_runs[i].being_compacted == false) {
+          num_sr_not_compacted++;
+        }
+      }
+
+      // The number of sorted runs that are not being compacted is greater than
+      // the maximum allowed number of sorted runs
+      if (num_sr_not_compacted >
+          mutable_cf_options.level0_file_num_compaction_trigger) {
+        unsigned int num_files =
+            num_sr_not_compacted -
+            mutable_cf_options.level0_file_num_compaction_trigger + 1;
+        if ((c = PickCompactionToReduceSortedRuns(
+                 cf_name, mutable_cf_options, vstorage, score, UINT_MAX,
+                 num_files, sorted_runs, log_buffer)) != nullptr) {
+          ROCKS_LOG_BUFFER(log_buffer,
+                           "[%s] Universal: compacting for file num -- %u\n",
+                           cf_name.c_str(), num_files);
+        }
+      }
+    }
+  }
+  if (c == nullptr) {
+    TEST_SYNC_POINT_CALLBACK("UniversalCompactionPicker::PickCompaction:Return",
+                             nullptr);
+    return nullptr;
+  }
+
+  if (ioptions_.compaction_options_universal.allow_trivial_move == true) {
+    c->set_is_trivial_move(IsInputFilesNonOverlapping(c));
+  }
+
+// validate that all the chosen files of L0 are non overlapping in time
+#ifndef NDEBUG
+  SequenceNumber prev_smallest_seqno = 0U;
+  bool is_first = true;
+
+  size_t level_index = 0U;
+  if (c->start_level() == 0) {
+    for (auto f : *c->inputs(0)) {
+      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;
+    }
+    level_index = 1U;
+  }
+  for (; level_index < c->num_input_levels(); level_index++) {
+    if (c->num_input_files(level_index) != 0) {
+      SequenceNumber smallest_seqno = 0U;
+      SequenceNumber largest_seqno = 0U;
+      GetSmallestLargestSeqno(*(c->inputs(level_index)), &smallest_seqno,
+                              &largest_seqno);
+      if (is_first) {
+        is_first = false;
+      } else if (prev_smallest_seqno > 0) {
+        // A level is considered as the bottommost level if there are
+        // no files in higher levels or if files in higher levels do
+        // not overlap with the files being compacted. Sequence numbers
+        // of files in bottommost level can be set to 0 to help
+        // compression. As a result, the following assert may not hold
+        // if the prev_smallest_seqno is 0.
+        assert(prev_smallest_seqno > largest_seqno);
+      }
+      prev_smallest_seqno = smallest_seqno;
+    }
+  }
+#endif
+  // update statistics
+  MeasureTime(ioptions_.statistics, NUM_FILES_IN_SINGLE_COMPACTION,
+              c->inputs(0)->size());
+
+  RegisterCompaction(c);
+
+  TEST_SYNC_POINT_CALLBACK("UniversalCompactionPicker::PickCompaction:Return",
+                           c);
+  return c;
+}
+
+uint32_t UniversalCompactionPicker::GetPathId(
+    const ImmutableCFOptions& ioptions, uint64_t file_size) {
+  // Two conditions need to be satisfied:
+  // (1) the target path needs to be able to hold the file's size
+  // (2) Total size left in this and previous paths need to be not
+  //     smaller than expected future file size before this new file is
+  //     compacted, which is estimated based on size_ratio.
+  // For example, if now we are compacting files of size (1, 1, 2, 4, 8),
+  // we will make sure the target file, probably with size of 16, will be
+  // placed in a path so that eventually when new files are generated and
+  // compacted to (1, 1, 2, 4, 8, 16), all those files can be stored in or
+  // before the path we chose.
+  //
+  // TODO(sdong): now the case of multiple column families is not
+  // considered in this algorithm. So the target size can be violated in
+  // that case. We need to improve it.
+  uint64_t accumulated_size = 0;
+  uint64_t future_size =
+      file_size * (100 - ioptions.compaction_options_universal.size_ratio) /
+      100;
+  uint32_t p = 0;
+  assert(!ioptions.db_paths.empty());
+  for (; p < ioptions.db_paths.size() - 1; p++) {
+    uint64_t target_size = ioptions.db_paths[p].target_size;
+    if (target_size > file_size &&
+        accumulated_size + (target_size - file_size) > future_size) {
+      return p;
+    }
+    accumulated_size += target_size;
+  }
+  return p;
+}
+
+//
+// Consider compaction files based on their size differences with
+// the next file in time order.
+//
+Compaction* UniversalCompactionPicker::PickCompactionToReduceSortedRuns(
+    const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
+    VersionStorageInfo* vstorage, double score, unsigned int ratio,
+    unsigned int max_number_of_files_to_compact,
+    const std::vector<SortedRun>& sorted_runs, LogBuffer* log_buffer) {
+  unsigned int min_merge_width =
+      ioptions_.compaction_options_universal.min_merge_width;
+  unsigned int max_merge_width =
+      ioptions_.compaction_options_universal.max_merge_width;
+
+  const SortedRun* sr = nullptr;
+  bool done = false;
+  size_t start_index = 0;
+  unsigned int candidate_count = 0;
+
+  unsigned int max_files_to_compact =
+      std::min(max_merge_width, max_number_of_files_to_compact);
+  min_merge_width = std::max(min_merge_width, 2U);
+
+  // Caller checks the size before executing this function. This invariant is
+  // important because otherwise we may have a possible integer underflow when
+  // dealing with unsigned types.
+  assert(sorted_runs.size() > 0);
+
+  // Considers a candidate file only if it is smaller than the
+  // total size accumulated so far.
+  for (size_t loop = 0; loop < sorted_runs.size(); loop++) {
+    candidate_count = 0;
+
+    // Skip files that are already being compacted
+    for (sr = nullptr; loop < sorted_runs.size(); loop++) {
+      sr = &sorted_runs[loop];
+
+      if (!sr->being_compacted) {
+        candidate_count = 1;
+        break;
+      }
+      char file_num_buf[kFormatFileNumberBufSize];
+      sr->Dump(file_num_buf, sizeof(file_num_buf));
+      ROCKS_LOG_BUFFER(log_buffer,
+                       "[%s] Universal: %s"
+                       "[%d] being compacted, skipping",
+                       cf_name.c_str(), file_num_buf, loop);
+
+      sr = nullptr;
+    }
+
+    // This file is not being compacted. Consider it as the
+    // first candidate to be compacted.
+    uint64_t candidate_size = sr != nullptr ? sr->compensated_file_size : 0;
+    if (sr != nullptr) {
+      char file_num_buf[kFormatFileNumberBufSize];
+      sr->Dump(file_num_buf, sizeof(file_num_buf), true);
+      ROCKS_LOG_BUFFER(log_buffer, "[%s] Universal: Possible candidate %s[%d].",
+                       cf_name.c_str(), file_num_buf, loop);
+    }
+
+    // Check if the succeeding files need compaction.
+    for (size_t i = loop + 1;
+         candidate_count < max_files_to_compact && i < sorted_runs.size();
+         i++) {
+      const SortedRun* succeeding_sr = &sorted_runs[i];
+      if (succeeding_sr->being_compacted) {
+        break;
+      }
+      // Pick files if the total/last candidate file size (increased by the
+      // specified ratio) is still larger than the next candidate file.
+      // candidate_size is the total size of files picked so far with the
+      // default kCompactionStopStyleTotalSize; with
+      // kCompactionStopStyleSimilarSize, it's simply the size of the last
+      // picked file.
+      double sz = candidate_size * (100.0 + ratio) / 100.0;
+      if (sz < static_cast<double>(succeeding_sr->size)) {
+        break;
+      }
+      if (ioptions_.compaction_options_universal.stop_style ==
+          kCompactionStopStyleSimilarSize) {
+        // Similar-size stopping rule: also check the last picked file isn't
+        // far larger than the next candidate file.
+        sz = (succeeding_sr->size * (100.0 + ratio)) / 100.0;
+        if (sz < static_cast<double>(candidate_size)) {
+          // If the small file we've encountered begins a run of similar-size
+          // files, we'll pick them up on a future iteration of the outer
+          // loop. If it's some lonely straggler, it'll eventually get picked
+          // by the last-resort read amp strategy which disregards size ratios.
+          break;
+        }
+        candidate_size = succeeding_sr->compensated_file_size;
+      } else {  // default kCompactionStopStyleTotalSize
+        candidate_size += succeeding_sr->compensated_file_size;
+      }
+      candidate_count++;
+    }
+
+    // Found a series of consecutive files that need compaction.
+    if (candidate_count >= (unsigned int)min_merge_width) {
+      start_index = loop;
+      done = true;
+      break;
+    } else {
+      for (size_t i = loop;
+           i < loop + candidate_count && i < sorted_runs.size(); i++) {
+        const SortedRun* skipping_sr = &sorted_runs[i];
+        char file_num_buf[256];
+        skipping_sr->DumpSizeInfo(file_num_buf, sizeof(file_num_buf), loop);
+        ROCKS_LOG_BUFFER(log_buffer, "[%s] Universal: Skipping %s",
+                         cf_name.c_str(), file_num_buf);
+      }
+    }
+  }
+  if (!done || candidate_count <= 1) {
+    return nullptr;
+  }
+  size_t first_index_after = start_index + candidate_count;
+  // Compression is enabled if files compacted earlier already reached
+  // size ratio of compression.
+  bool enable_compression = true;
+  int ratio_to_compress =
+      ioptions_.compaction_options_universal.compression_size_percent;
+  if (ratio_to_compress >= 0) {
+    uint64_t total_size = 0;
+    for (auto& sorted_run : sorted_runs) {
+      total_size += sorted_run.compensated_file_size;
+    }
+
+    uint64_t older_file_size = 0;
+    for (size_t i = sorted_runs.size() - 1; i >= first_index_after; i--) {
+      older_file_size += sorted_runs[i].size;
+      if (older_file_size * 100L >= total_size * (long)ratio_to_compress) {
+        enable_compression = false;
+        break;
+      }
+    }
+  }
+
+  uint64_t estimated_total_size = 0;
+  for (unsigned int i = 0; i < first_index_after; i++) {
+    estimated_total_size += sorted_runs[i].size;
+  }
+  uint32_t path_id = GetPathId(ioptions_, estimated_total_size);
+  int start_level = sorted_runs[start_index].level;
+  int output_level;
+  if (first_index_after == sorted_runs.size()) {
+    output_level = vstorage->num_levels() - 1;
+  } else if (sorted_runs[first_index_after].level == 0) {
+    output_level = 0;
+  } else {
+    output_level = sorted_runs[first_index_after].level - 1;
+  }
+
+  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);
+  }
+  for (size_t i = start_index; i < first_index_after; i++) {
+    auto& picking_sr = sorted_runs[i];
+    if (picking_sr.level == 0) {
+      FileMetaData* picking_file = picking_sr.file;
+      inputs[0].files.push_back(picking_file);
+    } else {
+      auto& files = inputs[picking_sr.level - start_level].files;
+      for (auto* f : vstorage->LevelFiles(picking_sr.level)) {
+        files.push_back(f);
+      }
+    }
+    char file_num_buf[256];
+    picking_sr.DumpSizeInfo(file_num_buf, sizeof(file_num_buf), i);
+    ROCKS_LOG_BUFFER(log_buffer, "[%s] Universal: Picking %s", cf_name.c_str(),
+                     file_num_buf);
+  }
+
+  CompactionReason compaction_reason;
+  if (max_number_of_files_to_compact == UINT_MAX) {
+    compaction_reason = CompactionReason::kUniversalSortedRunNum;
+  } else {
+    compaction_reason = CompactionReason::kUniversalSizeRatio;
+  }
+  return new Compaction(
+      vstorage, ioptions_, mutable_cf_options, std::move(inputs), output_level,
+      mutable_cf_options.MaxFileSizeForLevel(output_level), LLONG_MAX, path_id,
+      GetCompressionType(ioptions_, vstorage, mutable_cf_options, start_level,
+                         1, enable_compression),
+      /* grandparents */ {}, /* is manual */ false, score,
+      false /* deletion_compaction */, compaction_reason);
+}
+
+// Look at overall size amplification. If size amplification
+// exceeeds the configured value, then do a compaction
+// of the candidate files all the way upto the earliest
+// base file (overrides configured values of file-size ratios,
+// min_merge_width and max_merge_width).
+//
+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
+  uint64_t ratio =
+      ioptions_.compaction_options_universal.max_size_amplification_percent;
+
+  unsigned int candidate_count = 0;
+  uint64_t candidate_size = 0;
+  size_t start_index = 0;
+  const SortedRun* sr = nullptr;
+
+  // Skip files that are already being compacted
+  for (size_t loop = 0; loop < sorted_runs.size() - 1; loop++) {
+    sr = &sorted_runs[loop];
+    if (!sr->being_compacted) {
+      start_index = loop;  // Consider this as the first candidate.
+      break;
+    }
+    char file_num_buf[kFormatFileNumberBufSize];
+    sr->Dump(file_num_buf, sizeof(file_num_buf), true);
+    ROCKS_LOG_BUFFER(log_buffer, "[%s] Universal: skipping %s[%d] compacted %s",
+                     cf_name.c_str(), file_num_buf, loop,
+                     " cannot be a candidate to reduce size amp.\n");
+    sr = nullptr;
+  }
+
+  if (sr == nullptr) {
+    return nullptr;  // no candidate files
+  }
+  {
+    char file_num_buf[kFormatFileNumberBufSize];
+    sr->Dump(file_num_buf, sizeof(file_num_buf), true);
+    ROCKS_LOG_BUFFER(
+        log_buffer,
+        "[%s] Universal: First candidate %s[%" ROCKSDB_PRIszt "] %s",
+        cf_name.c_str(), file_num_buf, start_index, " to reduce size amp.\n");
+  }
+
+  // keep adding up all the remaining files
+  for (size_t loop = start_index; loop < sorted_runs.size() - 1; loop++) {
+    sr = &sorted_runs[loop];
+    if (sr->being_compacted) {
+      char file_num_buf[kFormatFileNumberBufSize];
+      sr->Dump(file_num_buf, sizeof(file_num_buf), true);
+      ROCKS_LOG_BUFFER(
+          log_buffer, "[%s] Universal: Possible candidate %s[%d] %s",
+          cf_name.c_str(), file_num_buf, start_index,
+          " is already being compacted. No size amp reduction possible.\n");
+      return nullptr;
+    }
+    candidate_size += sr->compensated_file_size;
+    candidate_count++;
+  }
+  if (candidate_count == 0) {
+    return nullptr;
+  }
+
+  // size of earliest file
+  uint64_t earliest_file_size = sorted_runs.back().size;
+
+  // size amplification = percentage of additional size
+  if (candidate_size * 100 < ratio * earliest_file_size) {
+    ROCKS_LOG_BUFFER(
+        log_buffer,
+        "[%s] Universal: size amp not needed. newer-files-total-size %" PRIu64
+        " earliest-file-size %" PRIu64,
+        cf_name.c_str(), candidate_size, earliest_file_size);
+    return nullptr;
+  } else {
+    ROCKS_LOG_BUFFER(
+        log_buffer,
+        "[%s] Universal: size amp needed. newer-files-total-size %" PRIu64
+        " earliest-file-size %" PRIu64,
+        cf_name.c_str(), candidate_size, earliest_file_size);
+  }
+  assert(start_index < sorted_runs.size() - 1);
+
+  // Estimate total file size
+  uint64_t estimated_total_size = 0;
+  for (size_t loop = start_index; loop < sorted_runs.size(); loop++) {
+    estimated_total_size += sorted_runs[loop].size;
+  }
+  uint32_t path_id = GetPathId(ioptions_, estimated_total_size);
+  int start_level = sorted_runs[start_index].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);
+  }
+  // We always compact all the files, so always compress.
+  for (size_t loop = start_index; loop < sorted_runs.size(); loop++) {
+    auto& picking_sr = sorted_runs[loop];
+    if (picking_sr.level == 0) {
+      FileMetaData* f = picking_sr.file;
+      inputs[0].files.push_back(f);
+    } else {
+      auto& files = inputs[picking_sr.level - start_level].files;
+      for (auto* f : vstorage->LevelFiles(picking_sr.level)) {
+        files.push_back(f);
+      }
+    }
+    char file_num_buf[256];
+    picking_sr.DumpSizeInfo(file_num_buf, sizeof(file_num_buf), loop);
+    ROCKS_LOG_BUFFER(log_buffer, "[%s] Universal: size amp picking %s",
+                     cf_name.c_str(), file_num_buf);
+  }
+
+  return new Compaction(
+      vstorage, ioptions_, mutable_cf_options, std::move(inputs),
+      vstorage->num_levels() - 1,
+      mutable_cf_options.MaxFileSizeForLevel(vstorage->num_levels() - 1),
+      /* max_grandparent_overlap_bytes */ LLONG_MAX, path_id,
+      GetCompressionType(ioptions_, vstorage, mutable_cf_options,
+                         vstorage->num_levels() - 1, 1),
+      /* grandparents */ {}, /* is manual */ false, score,
+      false /* deletion_compaction */,
+      CompactionReason::kUniversalSizeAmplification);
+}
+}  // namespace rocksdb
+
+#endif  // !ROCKSDB_LITE
@@ -0,0 +1,91 @@
+//  Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+//
+// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file. See the AUTHORS file for names of contributors.
+
+#pragma once
+#ifndef ROCKSDB_LITE
+
+#include "db/compaction_picker.h"
+
+namespace rocksdb {
+class UniversalCompactionPicker : public CompactionPicker {
+ public:
+  UniversalCompactionPicker(const ImmutableCFOptions& ioptions,
+                            const InternalKeyComparator* icmp)
+      : CompactionPicker(ioptions, icmp) {}
+  virtual Compaction* PickCompaction(const std::string& cf_name,
+                                     const MutableCFOptions& mutable_cf_options,
+                                     VersionStorageInfo* vstorage,
+                                     LogBuffer* log_buffer) override;
+
+  virtual int MaxOutputLevel() const override { return NumberLevels() - 1; }
+
+  virtual bool NeedsCompaction(
+      const VersionStorageInfo* vstorage) const override;
+
+ private:
+  struct SortedRun {
+    SortedRun(int _level, FileMetaData* _file, uint64_t _size,
+              uint64_t _compensated_file_size, bool _being_compacted)
+        : level(_level),
+          file(_file),
+          size(_size),
+          compensated_file_size(_compensated_file_size),
+          being_compacted(_being_compacted) {
+      assert(compensated_file_size > 0);
+      assert(level != 0 || file != nullptr);
+    }
+
+    void Dump(char* out_buf, size_t out_buf_size,
+              bool print_path = false) const;
+
+    // sorted_run_count is added into the string to print
+    void DumpSizeInfo(char* out_buf, size_t out_buf_size,
+                      size_t sorted_run_count) const;
+
+    int level;
+    // `file` Will be null for level > 0. For level = 0, the sorted run is
+    // for this file.
+    FileMetaData* file;
+    // For level > 0, `size` and `compensated_file_size` are sum of sizes all
+    // files in the level. `being_compacted` should be the same for all files
+    // in a non-zero level. Use the value here.
+    uint64_t size;
+    uint64_t compensated_file_size;
+    bool being_compacted;
+  };
+
+  // Pick Universal compaction to limit read amplification
+  Compaction* PickCompactionToReduceSortedRuns(
+      const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
+      VersionStorageInfo* vstorage, double score, unsigned int ratio,
+      unsigned int num_files, const std::vector<SortedRun>& sorted_runs,
+      LogBuffer* log_buffer);
+
+  // Pick Universal compaction to limit space amplification.
+  Compaction* PickCompactionToReduceSizeAmp(
+      const std::string& cf_name, const MutableCFOptions& mutable_cf_options,
+      VersionStorageInfo* vstorage, double score,
+      const std::vector<SortedRun>& sorted_runs, LogBuffer* log_buffer);
+
+  // Used in universal compaction when the enabled_trivial_move
+  // option is set. Checks whether there are any overlapping files
+  // in the input. Returns true if the input files are non
+  // overlapping.
+  bool IsInputFilesNonOverlapping(Compaction* c);
+
+  static std::vector<SortedRun> CalculateSortedRuns(
+      const VersionStorageInfo& vstorage, const ImmutableCFOptions& ioptions);
+
+  // Pick a path ID to place a newly generated file, with its estimated file
+  // size.
+  static uint32_t GetPathId(const ImmutableCFOptions& ioptions,
+                            uint64_t file_size);
+};
+}  // namespace rocksdb
+#endif  // !ROCKSDB_LITE
@@ -11,20 +11,20 @@

 #include "rocksdb/db.h"

-#include <inttypes.h>
 #include <errno.h>
 #include <fcntl.h>
+#include <inttypes.h>
 #include <sys/stat.h>
 #include <sys/types.h>
+#include "db/db_impl.h"
+#include "db/log_format.h"
+#include "db/version_set.h"
 #include "rocksdb/cache.h"
 #include "rocksdb/env.h"
 #include "rocksdb/table.h"
 #include "rocksdb/write_batch.h"
-#include "db/db_impl.h"
-#include "db/filename.h"
-#include "db/log_format.h"
-#include "db/version_set.h"
-#include "util/logging.h"
+#include "util/filename.h"
+#include "util/string_util.h"
 #include "util/testharness.h"
 #include "util/testutil.h"

@@ -8,9 +8,10 @@
 #include "db/db_impl.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
-#include "table/meta_blocks.h"
 #include "table/cuckoo_table_factory.h"
 #include "table/cuckoo_table_reader.h"
+#include "table/meta_blocks.h"
+#include "util/string_util.h"
 #include "util/testharness.h"
 #include "util/testutil.h"

@@ -7,9 +7,9 @@
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file. See the AUTHORS file for names of contributors.
 #include <cstdlib>
+#include "cache/lru_cache.h"
 #include "db/db_test_util.h"
 #include "port/stack_trace.h"
-#include "util/lru_cache.h"

 namespace rocksdb {

@@ -20,17 +20,23 @@ class DBBloomFilterTest : public DBTestBase {
  DBBloomFilterTest() : DBTestBase("/db_bloom_filter_test") {}
 };

-class DBBloomFilterTestWithParam : public DBTestBase,
-                                   public testing::WithParamInterface<bool> {
+class DBBloomFilterTestWithParam
+    : public DBTestBase,
+      public testing::WithParamInterface<std::tuple<bool, bool>> {
+  //                             public testing::WithParamInterface<bool> {
 protected:
  bool use_block_based_filter_;
+  bool partition_filters_;

 public:
  DBBloomFilterTestWithParam() : DBTestBase("/db_bloom_filter_tests") {}

  ~DBBloomFilterTestWithParam() {}

-  void SetUp() override { use_block_based_filter_ = GetParam(); }
+  void SetUp() override {
+    use_block_based_filter_ = std::get<0>(GetParam());
+    partition_filters_ = std::get<1>(GetParam());
+  }
 };

 // KeyMayExist can lead to a few false positives, but not false negatives.
@@ -43,7 +49,17 @@ TEST_P(DBBloomFilterTestWithParam, KeyMayExist) {
    anon::OptionsOverride options_override;
    options_override.filter_policy.reset(
        NewBloomFilterPolicy(20, use_block_based_filter_));
+    options_override.partition_filters = partition_filters_;
+    options_override.metadata_block_size = 32;
    Options options = CurrentOptions(options_override);
+    if (partition_filters_ &&
+        static_cast<BlockBasedTableOptions*>(
+            options.table_factory->GetOptions())
+                ->index_type != BlockBasedTableOptions::kTwoLevelIndexSearch) {
+      // In the current implementation partitioned filters depend on partitioned
+      // indexes
+      continue;
+    }
    options.statistics = rocksdb::CreateDBStatistics();
    CreateAndReopenWithCF({"pikachu"}, options);

@@ -102,192 +118,204 @@ TEST_P(DBBloomFilterTestWithParam, KeyMayExist) {
 }

 TEST_F(DBBloomFilterTest, GetFilterByPrefixBloom) {
-  Options options = last_options_;
-  options.prefix_extractor.reset(NewFixedPrefixTransform(8));
-  options.statistics = rocksdb::CreateDBStatistics();
-  BlockBasedTableOptions bbto;
-  bbto.filter_policy.reset(NewBloomFilterPolicy(10, false));
-  bbto.whole_key_filtering = false;
-  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
-  DestroyAndReopen(options);
+  for (bool partition_filters : {true, false}) {
+    Options options = last_options_;
+    options.prefix_extractor.reset(NewFixedPrefixTransform(8));
+    options.statistics = rocksdb::CreateDBStatistics();
+    BlockBasedTableOptions bbto;
+    bbto.filter_policy.reset(NewBloomFilterPolicy(10, false));
+    if (partition_filters) {
+      bbto.partition_filters = true;
+      bbto.index_type = BlockBasedTableOptions::IndexType::kTwoLevelIndexSearch;
+    }
+    bbto.whole_key_filtering = false;
+    options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+    DestroyAndReopen(options);

-  WriteOptions wo;
-  ReadOptions ro;
-  FlushOptions fo;
-  fo.wait = true;
-  std::string value;
+    WriteOptions wo;
+    ReadOptions ro;
+    FlushOptions fo;
+    fo.wait = true;
+    std::string value;

-  ASSERT_OK(dbfull()->Put(wo, "barbarbar", "foo"));
-  ASSERT_OK(dbfull()->Put(wo, "barbarbar2", "foo2"));
-  ASSERT_OK(dbfull()->Put(wo, "foofoofoo", "bar"));
+    ASSERT_OK(dbfull()->Put(wo, "barbarbar", "foo"));
+    ASSERT_OK(dbfull()->Put(wo, "barbarbar2", "foo2"));
+    ASSERT_OK(dbfull()->Put(wo, "foofoofoo", "bar"));

-  dbfull()->Flush(fo);
+    dbfull()->Flush(fo);

-  ASSERT_EQ("foo", Get("barbarbar"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0);
-  ASSERT_EQ("foo2", Get("barbarbar2"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0);
-  ASSERT_EQ("NOT_FOUND", Get("barbarbar3"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0);
+    ASSERT_EQ("foo", Get("barbarbar"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0);
+    ASSERT_EQ("foo2", Get("barbarbar2"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0);
+    ASSERT_EQ("NOT_FOUND", Get("barbarbar3"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0);

-  ASSERT_EQ("NOT_FOUND", Get("barfoofoo"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
+    ASSERT_EQ("NOT_FOUND", Get("barfoofoo"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);

-  ASSERT_EQ("NOT_FOUND", Get("foobarbar"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 2);
+    ASSERT_EQ("NOT_FOUND", Get("foobarbar"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 2);

-  ro.total_order_seek = true;
-  ASSERT_TRUE(db_->Get(ro, "foobarbar", &value).IsNotFound());
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 2);
+    ro.total_order_seek = true;
+    ASSERT_TRUE(db_->Get(ro, "foobarbar", &value).IsNotFound());
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 2);
+  }
 }

 TEST_F(DBBloomFilterTest, WholeKeyFilterProp) {
-  Options options = last_options_;
-  options.prefix_extractor.reset(NewFixedPrefixTransform(3));
-  options.statistics = rocksdb::CreateDBStatistics();
+  for (bool partition_filters : {true, false}) {
+    Options options = last_options_;
+    options.prefix_extractor.reset(NewFixedPrefixTransform(3));
+    options.statistics = rocksdb::CreateDBStatistics();

-  BlockBasedTableOptions bbto;
-  bbto.filter_policy.reset(NewBloomFilterPolicy(10, false));
-  bbto.whole_key_filtering = false;
-  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
-  DestroyAndReopen(options);
+    BlockBasedTableOptions bbto;
+    bbto.filter_policy.reset(NewBloomFilterPolicy(10, false));
+    bbto.whole_key_filtering = false;
+    if (partition_filters) {
+      bbto.partition_filters = true;
+      bbto.index_type = BlockBasedTableOptions::IndexType::kTwoLevelIndexSearch;
+    }
+    options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+    DestroyAndReopen(options);

-  WriteOptions wo;
-  ReadOptions ro;
-  FlushOptions fo;
-  fo.wait = true;
-  std::string value;
+    WriteOptions wo;
+    ReadOptions ro;
+    FlushOptions fo;
+    fo.wait = true;
+    std::string value;

-  ASSERT_OK(dbfull()->Put(wo, "foobar", "foo"));
-  // Needs insert some keys to make sure files are not filtered out by key
-  // ranges.
-  ASSERT_OK(dbfull()->Put(wo, "aaa", ""));
-  ASSERT_OK(dbfull()->Put(wo, "zzz", ""));
-  dbfull()->Flush(fo);
+    ASSERT_OK(dbfull()->Put(wo, "foobar", "foo"));
+    // Needs insert some keys to make sure files are not filtered out by key
+    // ranges.
+    ASSERT_OK(dbfull()->Put(wo, "aaa", ""));
+    ASSERT_OK(dbfull()->Put(wo, "zzz", ""));
+    dbfull()->Flush(fo);

-  Reopen(options);
-  ASSERT_EQ("NOT_FOUND", Get("foo"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0);
-  ASSERT_EQ("NOT_FOUND", Get("bar"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
-  ASSERT_EQ("foo", Get("foobar"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
+    Reopen(options);
+    ASSERT_EQ("NOT_FOUND", Get("foo"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0);
+    ASSERT_EQ("NOT_FOUND", Get("bar"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
+    ASSERT_EQ("foo", Get("foobar"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);

-  // Reopen with whole key filtering enabled and prefix extractor
-  // NULL. Bloom filter should be off for both of whole key and
-  // prefix bloom.
-  bbto.whole_key_filtering = true;
-  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
-  options.prefix_extractor.reset();
-  Reopen(options);
+    // Reopen with whole key filtering enabled and prefix extractor
+    // NULL. Bloom filter should be off for both of whole key and
+    // prefix bloom.
+    bbto.whole_key_filtering = true;
+    options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+    options.prefix_extractor.reset();
+    Reopen(options);

-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
-  ASSERT_EQ("NOT_FOUND", Get("foo"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
-  ASSERT_EQ("NOT_FOUND", Get("bar"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
-  ASSERT_EQ("foo", Get("foobar"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
-  // Write DB with only full key filtering.
-  ASSERT_OK(dbfull()->Put(wo, "foobar", "foo"));
-  // Needs insert some keys to make sure files are not filtered out by key
-  // ranges.
-  ASSERT_OK(dbfull()->Put(wo, "aaa", ""));
-  ASSERT_OK(dbfull()->Put(wo, "zzz", ""));
-  db_->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
+    ASSERT_EQ("NOT_FOUND", Get("foo"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
+    ASSERT_EQ("NOT_FOUND", Get("bar"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
+    ASSERT_EQ("foo", Get("foobar"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
+    // Write DB with only full key filtering.
+    ASSERT_OK(dbfull()->Put(wo, "foobar", "foo"));
+    // Needs insert some keys to make sure files are not filtered out by key
+    // ranges.
+    ASSERT_OK(dbfull()->Put(wo, "aaa", ""));
+    ASSERT_OK(dbfull()->Put(wo, "zzz", ""));
+    db_->CompactRange(CompactRangeOptions(), nullptr, nullptr);

-  // Reopen with both of whole key off and prefix extractor enabled.
-  // Still no bloom filter should be used.
-  options.prefix_extractor.reset(NewFixedPrefixTransform(3));
-  bbto.whole_key_filtering = false;
-  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
-  Reopen(options);
+    // Reopen with both of whole key off and prefix extractor enabled.
+    // Still no bloom filter should be used.
+    options.prefix_extractor.reset(NewFixedPrefixTransform(3));
+    bbto.whole_key_filtering = false;
+    options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+    Reopen(options);

-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
-  ASSERT_EQ("NOT_FOUND", Get("foo"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
-  ASSERT_EQ("NOT_FOUND", Get("bar"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
-  ASSERT_EQ("foo", Get("foobar"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
+    ASSERT_EQ("NOT_FOUND", Get("foo"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
+    ASSERT_EQ("NOT_FOUND", Get("bar"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
+    ASSERT_EQ("foo", Get("foobar"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);

-  // Try to create a DB with mixed files:
-  ASSERT_OK(dbfull()->Put(wo, "foobar", "foo"));
-  // Needs insert some keys to make sure files are not filtered out by key
-  // ranges.
-  ASSERT_OK(dbfull()->Put(wo, "aaa", ""));
-  ASSERT_OK(dbfull()->Put(wo, "zzz", ""));
-  db_->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+    // Try to create a DB with mixed files:
+    ASSERT_OK(dbfull()->Put(wo, "foobar", "foo"));
+    // Needs insert some keys to make sure files are not filtered out by key
+    // ranges.
+    ASSERT_OK(dbfull()->Put(wo, "aaa", ""));
+    ASSERT_OK(dbfull()->Put(wo, "zzz", ""));
+    db_->CompactRange(CompactRangeOptions(), nullptr, nullptr);

-  options.prefix_extractor.reset();
-  bbto.whole_key_filtering = true;
-  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
-  Reopen(options);
+    options.prefix_extractor.reset();
+    bbto.whole_key_filtering = true;
+    options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+    Reopen(options);

-  // Try to create a DB with mixed files.
-  ASSERT_OK(dbfull()->Put(wo, "barfoo", "bar"));
-  // In this case needs insert some keys to make sure files are
-  // not filtered out by key ranges.
-  ASSERT_OK(dbfull()->Put(wo, "aaa", ""));
-  ASSERT_OK(dbfull()->Put(wo, "zzz", ""));
-  Flush();
+    // Try to create a DB with mixed files.
+    ASSERT_OK(dbfull()->Put(wo, "barfoo", "bar"));
+    // In this case needs insert some keys to make sure files are
+    // not filtered out by key ranges.
+    ASSERT_OK(dbfull()->Put(wo, "aaa", ""));
+    ASSERT_OK(dbfull()->Put(wo, "zzz", ""));
+    Flush();

-  // Now we have two files:
-  // File 1: An older file with prefix bloom.
-  // File 2: A newer file with whole bloom filter.
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
-  ASSERT_EQ("NOT_FOUND", Get("foo"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 2);
-  ASSERT_EQ("NOT_FOUND", Get("bar"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 3);
-  ASSERT_EQ("foo", Get("foobar"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 4);
-  ASSERT_EQ("bar", Get("barfoo"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 4);
+    // Now we have two files:
+    // File 1: An older file with prefix bloom.
+    // File 2: A newer file with whole bloom filter.
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 1);
+    ASSERT_EQ("NOT_FOUND", Get("foo"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 2);
+    ASSERT_EQ("NOT_FOUND", Get("bar"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 3);
+    ASSERT_EQ("foo", Get("foobar"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 4);
+    ASSERT_EQ("bar", Get("barfoo"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 4);

-  // Reopen with the same setting: only whole key is used
-  Reopen(options);
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 4);
-  ASSERT_EQ("NOT_FOUND", Get("foo"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 5);
-  ASSERT_EQ("NOT_FOUND", Get("bar"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 6);
-  ASSERT_EQ("foo", Get("foobar"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 7);
-  ASSERT_EQ("bar", Get("barfoo"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 7);
+    // Reopen with the same setting: only whole key is used
+    Reopen(options);
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 4);
+    ASSERT_EQ("NOT_FOUND", Get("foo"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 5);
+    ASSERT_EQ("NOT_FOUND", Get("bar"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 6);
+    ASSERT_EQ("foo", Get("foobar"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 7);
+    ASSERT_EQ("bar", Get("barfoo"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 7);

-  // Restart with both filters are allowed
-  options.prefix_extractor.reset(NewFixedPrefixTransform(3));
-  bbto.whole_key_filtering = true;
-  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
-  Reopen(options);
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 7);
-  // File 1 will has it filtered out.
-  // File 2 will not, as prefix `foo` exists in the file.
-  ASSERT_EQ("NOT_FOUND", Get("foo"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 8);
-  ASSERT_EQ("NOT_FOUND", Get("bar"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 10);
-  ASSERT_EQ("foo", Get("foobar"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 11);
-  ASSERT_EQ("bar", Get("barfoo"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 11);
+    // Restart with both filters are allowed
+    options.prefix_extractor.reset(NewFixedPrefixTransform(3));
+    bbto.whole_key_filtering = true;
+    options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+    Reopen(options);
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 7);
+    // File 1 will has it filtered out.
+    // File 2 will not, as prefix `foo` exists in the file.
+    ASSERT_EQ("NOT_FOUND", Get("foo"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 8);
+    ASSERT_EQ("NOT_FOUND", Get("bar"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 10);
+    ASSERT_EQ("foo", Get("foobar"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 11);
+    ASSERT_EQ("bar", Get("barfoo"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 11);

-  // Restart with only prefix bloom is allowed.
-  options.prefix_extractor.reset(NewFixedPrefixTransform(3));
-  bbto.whole_key_filtering = false;
-  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
-  Reopen(options);
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 11);
-  ASSERT_EQ("NOT_FOUND", Get("foo"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 11);
-  ASSERT_EQ("NOT_FOUND", Get("bar"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 12);
-  ASSERT_EQ("foo", Get("foobar"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 12);
-  ASSERT_EQ("bar", Get("barfoo"));
-  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 12);
+    // Restart with only prefix bloom is allowed.
+    options.prefix_extractor.reset(NewFixedPrefixTransform(3));
+    bbto.whole_key_filtering = false;
+    options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+    Reopen(options);
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 11);
+    ASSERT_EQ("NOT_FOUND", Get("foo"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 11);
+    ASSERT_EQ("NOT_FOUND", Get("bar"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 12);
+    ASSERT_EQ("foo", Get("foobar"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 12);
+    ASSERT_EQ("bar", Get("barfoo"));
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 12);
+  }
 }

 TEST_P(DBBloomFilterTestWithParam, BloomFilter) {
@@ -301,6 +329,12 @@ TEST_P(DBBloomFilterTestWithParam, BloomFilter) {
    table_options.no_block_cache = true;
    table_options.filter_policy.reset(
        NewBloomFilterPolicy(10, use_block_based_filter_));
+    table_options.partition_filters = partition_filters_;
+    if (partition_filters_) {
+      table_options.index_type =
+          BlockBasedTableOptions::IndexType::kTwoLevelIndexSearch;
+    }
+    table_options.metadata_block_size = 32;
    options.table_factory.reset(NewBlockBasedTableFactory(table_options));

    CreateAndReopenWithCF({"pikachu"}, options);
@@ -327,7 +361,13 @@ TEST_P(DBBloomFilterTestWithParam, BloomFilter) {
    int reads = env_->random_read_counter_.Read();
    fprintf(stderr, "%d present => %d reads\n", N, reads);
    ASSERT_GE(reads, N);
-    ASSERT_LE(reads, N + 2 * N / 100);
+    if (partition_filters_) {
+      // Without block cache, we read an extra partition filter per each
+      // level*read and a partition index per each read
+      ASSERT_LE(reads, 4 * N + 2 * N / 100);
+    } else {
+      ASSERT_LE(reads, N + 2 * N / 100);
+    }

    // Lookup present keys.  Should rarely read from either sstable.
    env_->random_read_counter_.Reset();
@@ -336,7 +376,13 @@ TEST_P(DBBloomFilterTestWithParam, BloomFilter) {
    }
    reads = env_->random_read_counter_.Read();
    fprintf(stderr, "%d missing => %d reads\n", N, reads);
-    ASSERT_LE(reads, 3 * N / 100);
+    if (partition_filters_) {
+      // With partitioned filter we read one extra filter per level per each
+      // missed read.
+      ASSERT_LE(reads, 2 * N + 3 * N / 100);
+    } else {
+      ASSERT_LE(reads, 3 * N / 100);
+    }

    env_->delay_sstable_sync_.store(false, std::memory_order_release);
    Close();
@@ -344,7 +390,9 @@ TEST_P(DBBloomFilterTestWithParam, BloomFilter) {
 }

 INSTANTIATE_TEST_CASE_P(DBBloomFilterTestWithParam, DBBloomFilterTestWithParam,
-                        ::testing::Bool());
+                        ::testing::Values(std::make_tuple(true, false),
+                                          std::make_tuple(false, true),
+                                          std::make_tuple(false, false)));

 TEST_F(DBBloomFilterTest, BloomFilterRate) {
  while (ChangeFilterOptions()) {
@@ -401,28 +449,13 @@ TEST_F(DBBloomFilterTest, BloomFilterCompatibility) {
    ASSERT_EQ(Key(i), Get(1, Key(i)));
  }
  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0);
-}

-TEST_F(DBBloomFilterTest, BloomFilterReverseCompatibility) {
-  Options options = CurrentOptions();
-  options.statistics = rocksdb::CreateDBStatistics();
-  BlockBasedTableOptions table_options;
+  // Check db with partitioned full filter
+  table_options.partition_filters = true;
+  table_options.index_type =
+      BlockBasedTableOptions::IndexType::kTwoLevelIndexSearch;
  table_options.filter_policy.reset(NewBloomFilterPolicy(10, false));
  options.table_factory.reset(NewBlockBasedTableFactory(table_options));
-
-  // Create with full filter
-  CreateAndReopenWithCF({"pikachu"}, options);
-
-  const int maxKey = 10000;
-  for (int i = 0; i < maxKey; i++) {
-    ASSERT_OK(Put(1, Key(i), Key(i)));
-  }
-  ASSERT_OK(Put(1, Key(maxKey + 55555), Key(maxKey + 55555)));
-  Flush(1);
-
-  // Check db with block_based filter
-  table_options.filter_policy.reset(NewBloomFilterPolicy(10, true));
-  options.table_factory.reset(NewBlockBasedTableFactory(table_options));
  ReopenWithColumnFamilies({"default", "pikachu"}, options);

  // Check if they can be found
@@ -432,6 +465,43 @@ TEST_F(DBBloomFilterTest, BloomFilterReverseCompatibility) {
  ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0);
 }

+TEST_F(DBBloomFilterTest, BloomFilterReverseCompatibility) {
+  for (bool partition_filters : {true, false}) {
+    Options options = CurrentOptions();
+    options.statistics = rocksdb::CreateDBStatistics();
+    BlockBasedTableOptions table_options;
+    if (partition_filters) {
+      table_options.partition_filters = true;
+      table_options.index_type =
+          BlockBasedTableOptions::IndexType::kTwoLevelIndexSearch;
+    }
+    table_options.filter_policy.reset(NewBloomFilterPolicy(10, false));
+    options.table_factory.reset(NewBlockBasedTableFactory(table_options));
+    DestroyAndReopen(options);
+
+    // Create with full filter
+    CreateAndReopenWithCF({"pikachu"}, options);
+
+    const int maxKey = 10000;
+    for (int i = 0; i < maxKey; i++) {
+      ASSERT_OK(Put(1, Key(i), Key(i)));
+    }
+    ASSERT_OK(Put(1, Key(maxKey + 55555), Key(maxKey + 55555)));
+    Flush(1);
+
+    // Check db with block_based filter
+    table_options.filter_policy.reset(NewBloomFilterPolicy(10, true));
+    options.table_factory.reset(NewBlockBasedTableFactory(table_options));
+    ReopenWithColumnFamilies({"default", "pikachu"}, options);
+
+    // Check if they can be found
+    for (int i = 0; i < maxKey; i++) {
+      ASSERT_EQ(Key(i), Get(1, Key(i)));
+    }
+    ASSERT_EQ(TestGetTickerCount(options, BLOOM_FILTER_USEFUL), 0);
+  }
+}
+
 namespace {
 // A wrapped bloom over default FilterPolicy
 class WrappedBloom : public FilterPolicy {
@@ -586,11 +656,12 @@ TEST_F(DBBloomFilterTest, PrefixExtractorBlockFilter) {
 #ifndef ROCKSDB_LITE
 class BloomStatsTestWithParam
    : public DBBloomFilterTest,
-      public testing::WithParamInterface<std::tuple<bool, bool>> {
+      public testing::WithParamInterface<std::tuple<bool, bool, bool>> {
 public:
  BloomStatsTestWithParam() {
    use_block_table_ = std::get<0>(GetParam());
    use_block_based_builder_ = std::get<1>(GetParam());
+    partition_filters_ = std::get<2>(GetParam());

    options_.create_if_missing = true;
    options_.prefix_extractor.reset(rocksdb::NewFixedPrefixTransform(4));
@@ -599,13 +670,21 @@ class BloomStatsTestWithParam
    if (use_block_table_) {
      BlockBasedTableOptions table_options;
      table_options.hash_index_allow_collision = false;
+      if (partition_filters_) {
+        assert(!use_block_based_builder_);
+        table_options.partition_filters = partition_filters_;
+        table_options.index_type =
+            BlockBasedTableOptions::IndexType::kTwoLevelIndexSearch;
+      }
      table_options.filter_policy.reset(
          NewBloomFilterPolicy(10, use_block_based_builder_));
      options_.table_factory.reset(NewBlockBasedTableFactory(table_options));
    } else {
+      assert(!partition_filters_);  // not supported in plain table
      PlainTableOptions table_options;
      options_.table_factory.reset(NewPlainTableFactory(table_options));
    }
+    options_.env = env_;

    perf_context.Reset();
    DestroyAndReopen(options_);
@@ -622,6 +701,7 @@ class BloomStatsTestWithParam

  bool use_block_table_;
  bool use_block_based_builder_;
+  bool partition_filters_;
  Options options_;
 };

@@ -716,25 +796,44 @@ TEST_P(BloomStatsTestWithParam, BloomStatsTestWithIter) {
  ASSERT_OK(iter->status());
  ASSERT_TRUE(iter->Valid());
  ASSERT_EQ(value1, iter->value().ToString());
-  ASSERT_EQ(1, perf_context.bloom_sst_hit_count);
+  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);
+  }

  iter->Seek(key3);
  ASSERT_OK(iter->status());
  ASSERT_TRUE(iter->Valid());
  ASSERT_EQ(value3, iter->value().ToString());
-  ASSERT_EQ(2, perf_context.bloom_sst_hit_count);
+  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);
+  }

  iter->Seek(key2);
  ASSERT_OK(iter->status());
-  ASSERT_TRUE(!iter->Valid());
-  ASSERT_EQ(1, perf_context.bloom_sst_miss_count);
-  ASSERT_EQ(2, perf_context.bloom_sst_hit_count);
+  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);
+  }
 }

 INSTANTIATE_TEST_CASE_P(BloomStatsTestWithParam, BloomStatsTestWithParam,
-                        ::testing::Values(std::make_tuple(true, true),
-                                          std::make_tuple(true, false),
-                                          std::make_tuple(false, false)));
+                        ::testing::Values(std::make_tuple(true, true, false),
+                                          std::make_tuple(true, false, false),
+                                          std::make_tuple(true, false, true),
+                                          std::make_tuple(false, false,
+                                                          false)));

 namespace {
 void PrefixScanInit(DBBloomFilterTest* dbtest) {
@@ -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 {
@@ -772,6 +778,41 @@ 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) {
@@ -1843,6 +1884,47 @@ 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 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_;
@@ -2492,12 +2574,164 @@ TEST_P(DBCompactionTestWithParam, ForceBottommostLevelCompaction) {
  rocksdb::SyncPoint::GetInstance()->DisableProcessing();
 }

+TEST_P(DBCompactionTestWithParam, IntraL0Compaction) {
+  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 2MB 1MB 1MB 1MB 1MB
+  // score:                     1.5 1.3 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.
+  //
+  // 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) {
+    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());
+  }
+  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 the 2MB file (not compacted) and 4MB file (output of L0->L0)
+  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][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:
@@ -75,6 +75,7 @@ TEST_F(DBTestDynamicLevel, DynamicLevelMaxBytesBase) {
      options.compression_per_level[0] = kNoCompression;
      options.compression_per_level[1] = kLZ4Compression;
      options.compression_per_level[2] = kSnappyCompression;
+      options.env = env_;

      DestroyAndReopen(options);

@@ -15,13 +15,13 @@
 #include <algorithm>
 #include <string>
 #include "db/db_impl.h"
-#include "db/filename.h"
 #include "db/job_context.h"
 #include "db/version_set.h"
 #include "port/port.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
 #include "util/file_util.h"
+#include "util/filename.h"
 #include "util/mutexlock.h"
 #include "util/sync_point.h"

@@ -31,12 +31,11 @@ Status DBImpl::DisableFileDeletions() {
  InstrumentedMutexLock l(&mutex_);
  ++disable_delete_obsolete_files_;
  if (disable_delete_obsolete_files_ == 1) {
-    Log(InfoLogLevel::INFO_LEVEL, immutable_db_options_.info_log,
-        "File Deletions Disabled");
+    ROCKS_LOG_INFO(immutable_db_options_.info_log, "File Deletions Disabled");
  } else {
-    Log(InfoLogLevel::WARN_LEVEL, immutable_db_options_.info_log,
-        "File Deletions Disabled, but already disabled. Counter: %d",
-        disable_delete_obsolete_files_);
+    ROCKS_LOG_WARN(immutable_db_options_.info_log,
+                   "File Deletions Disabled, but already disabled. Counter: %d",
+                   disable_delete_obsolete_files_);
  }
  return Status::OK();
 }
@@ -55,12 +54,12 @@ Status DBImpl::EnableFileDeletions(bool force) {
      --disable_delete_obsolete_files_;
    }
    if (disable_delete_obsolete_files_ == 0)  {
-      Log(InfoLogLevel::INFO_LEVEL, immutable_db_options_.info_log,
-          "File Deletions Enabled");
+      ROCKS_LOG_INFO(immutable_db_options_.info_log, "File Deletions Enabled");
      should_purge_files = true;
      FindObsoleteFiles(&job_context, true);
    } else {
-      Log(InfoLogLevel::WARN_LEVEL, immutable_db_options_.info_log,
+      ROCKS_LOG_WARN(
+          immutable_db_options_.info_log,
          "File Deletions Enable, but not really enabled. Counter: %d",
          disable_delete_obsolete_files_);
    }
@@ -106,8 +105,8 @@ Status DBImpl::GetLiveFiles(std::vector<std::string>& ret,

    if (!status.ok()) {
      mutex_.Unlock();
-      Log(InfoLogLevel::ERROR_LEVEL, immutable_db_options_.info_log,
-          "Cannot Flush data %s\n", status.ToString().c_str());
+      ROCKS_LOG_ERROR(immutable_db_options_.info_log, "Cannot Flush data %s\n",
+                      status.ToString().c_str());
      return status;
    }
  }
@@ -19,12 +19,19 @@ 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) {
  Options options;
  options.disable_auto_compactions = true;
  options.max_background_flushes = 2;
+  options.env = env_;
  Reopen(options);
  FlushOptions no_wait;
  no_wait.wait = false;
@@ -50,7 +57,7 @@ TEST_F(DBFlushTest, FlushWhileWritingManifest) {

 TEST_F(DBFlushTest, SyncFail) {
  std::unique_ptr<FaultInjectionTestEnv> fault_injection_env(
-      new FaultInjectionTestEnv(Env::Default()));
+      new FaultInjectionTestEnv(env_));
  Options options;
  options.disable_auto_compactions = true;
  options.env = fault_injection_env.get();
@@ -82,6 +89,33 @@ TEST_F(DBFlushTest, SyncFail) {
  Destroy(options);
 }

+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) {
@@ -34,18 +34,19 @@
 #include "db/write_controller.h"
 #include "db/write_thread.h"
 #include "memtable_list.h"
+#include "monitoring/instrumented_mutex.h"
+#include "options/db_options.h"
 #include "port/port.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
 #include "rocksdb/memtablerep.h"
+#include "rocksdb/status.h"
 #include "rocksdb/transaction_log.h"
 #include "rocksdb/write_buffer_manager.h"
 #include "table/scoped_arena_iterator.h"
 #include "util/autovector.h"
-#include "util/db_options.h"
 #include "util/event_logger.h"
 #include "util/hash.h"
-#include "util/instrumented_mutex.h"
 #include "util/stop_watch.h"
 #include "util/thread_local.h"

@@ -91,7 +92,7 @@ class DBImpl : public DB {
  using DB::Get;
  virtual Status Get(const ReadOptions& options,
                     ColumnFamilyHandle* column_family, const Slice& key,
-                     std::string* value) override;
+                     PinnableSlice* value) override;
  using DB::MultiGet;
  virtual std::vector<Status> MultiGet(
      const ReadOptions& options,
@@ -99,10 +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
@@ -191,7 +201,11 @@ class DBImpl : public DB {

  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;
@@ -314,7 +328,7 @@ class DBImpl : public DB {
                           ColumnFamilyHandle* column_family = nullptr,
                           bool disallow_trivial_move = false);

-  void TEST_MaybeFlushColumnFamilies();
+  void TEST_HandleWALFull();

  bool TEST_UnableToFlushOldestLog() {
    return unable_to_flush_oldest_log_;
@@ -451,6 +465,9 @@ 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() {
@@ -546,9 +563,10 @@ 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.
@@ -556,6 +574,10 @@ 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);
@@ -600,7 +622,19 @@ class DBImpl : public DB {
 #endif
  struct CompactionState;

-  struct WriteContext;
+  struct WriteContext {
+    autovector<SuperVersion*> superversions_to_free_;
+    autovector<MemTable*> memtables_to_free_;
+
+    ~WriteContext() {
+      for (auto& sv : superversions_to_free_) {
+        delete sv;
+      }
+      for (auto& m : memtables_to_free_) {
+        delete m;
+      }
+    }
+  };

  struct PurgeFileInfo;

@@ -615,6 +649,12 @@ 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
@@ -677,6 +717,20 @@ class DBImpl : public DB {
  // Wait for memtable flushed
  Status WaitForFlushMemTable(ColumnFamilyData* cfd);

+  // REQUIRES: mutex locked
+  Status HandleWALFull(WriteContext* write_context);
+
+  // REQUIRES: mutex locked
+  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 WriteToWAL(const autovector<WriteThread::Writer*>& write_group,
+                    log::Writer* log_writer, bool need_log_sync,
+                    bool need_log_dir_sync, SequenceNumber sequence);
+
 #ifndef ROCKSDB_LITE

  Status CompactFilesImpl(const CompactionOptions& compact_options,
@@ -740,12 +794,6 @@ class DBImpl : public DB {

  const Snapshot* GetSnapshotImpl(bool is_write_conflict_boundary);

-  // Persist RocksDB options under the single write thread
-  // REQUIRES: mutex locked
-  Status PersistOptions();
-
-  void MaybeFlushColumnFamilies();
-
  uint64_t GetMaxTotalWalSize() const;

  // table_cache_ provides its own synchronization
@@ -821,7 +869,7 @@ class DBImpl : public DB {
  std::deque<LogWriterNumber> logs_;
  // Signaled when getting_synced becomes false for some of the logs_.
  InstrumentedCondVar log_sync_cv_;
-  uint64_t total_log_size_;
+  std::atomic<uint64_t> total_log_size_;
  // only used for dynamically adjusting max_total_wal_size. it is a sum of
  // [write_buffer_size * max_write_buffer_number] over all column families
  uint64_t max_total_in_memory_state_;
@@ -1070,13 +1118,6 @@ 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
@@ -1106,7 +1147,7 @@ class DBImpl : public DB {
  // Function that Get and KeyMayExist call with no_io true or false
  // Note: 'value_found' from KeyMayExist propagates here
  Status GetImpl(const ReadOptions& options, ColumnFamilyHandle* column_family,
-                 const Slice& key, std::string* value,
+                 const Slice& key, PinnableSlice* value,
                 bool* value_found = nullptr);

  bool GetIntPropertyInternal(ColumnFamilyData* cfd,
@@ -1129,6 +1170,10 @@ extern Options SanitizeOptions(const std::string& db,

 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) {
@@ -10,7 +10,7 @@
 #ifndef NDEBUG

 #include "db/db_impl.h"
-#include "util/thread_status_updater.h"
+#include "monitoring/thread_status_updater.h"

 namespace rocksdb {

@@ -19,9 +19,10 @@ uint64_t DBImpl::TEST_GetLevel0TotalSize() {
  return default_cf_handle_->cfd()->current()->storage_info()->NumLevelBytes(0);
 }

-void DBImpl::TEST_MaybeFlushColumnFamilies() {
+void DBImpl::TEST_HandleWALFull() {
+  WriteContext write_context;
  InstrumentedMutexLock l(&mutex_);
-  MaybeFlushColumnFamilies();
+  HandleWALFull(&write_context);
 }

 int64_t DBImpl::TEST_MaxNextLevelOverlappingBytes(
@@ -61,8 +61,8 @@ Status DBImpl::PromoteL0(ColumnFamilyHandle* column_family, int target_level) {
  assert(column_family);

  if (target_level < 1) {
-    Log(InfoLogLevel::INFO_LEVEL, immutable_db_options_.info_log,
-        "PromoteL0 FAILED. Invalid target level %d\n", target_level);
+    ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                   "PromoteL0 FAILED. Invalid target level %d\n", target_level);
    return Status::InvalidArgument("Invalid target level");
  }

@@ -75,8 +75,9 @@ Status DBImpl::PromoteL0(ColumnFamilyHandle* column_family, int target_level) {
    const auto* vstorage = cfd->current()->storage_info();

    if (target_level >= vstorage->num_levels()) {
-      Log(InfoLogLevel::INFO_LEVEL, immutable_db_options_.info_log,
-          "PromoteL0 FAILED. Target level %d does not exist\n", target_level);
+      ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                     "PromoteL0 FAILED. Target level %d does not exist\n",
+                     target_level);
      job_context.Clean();
      return Status::InvalidArgument("Target level does not exist");
    }
@@ -94,9 +95,9 @@ Status DBImpl::PromoteL0(ColumnFamilyHandle* column_family, int target_level) {
    for (size_t i = 0; i < l0_files.size(); ++i) {
      auto f = l0_files[i];
      if (f->being_compacted) {
-        Log(InfoLogLevel::INFO_LEVEL, immutable_db_options_.info_log,
-            "PromoteL0 FAILED. File %" PRIu64 " being compacted\n",
-            f->fd.GetNumber());
+        ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                       "PromoteL0 FAILED. File %" PRIu64 " being compacted\n",
+                       f->fd.GetNumber());
        job_context.Clean();
        return Status::InvalidArgument("PromoteL0 called during L0 compaction");
      }
@@ -104,10 +105,10 @@ Status DBImpl::PromoteL0(ColumnFamilyHandle* column_family, int target_level) {
      if (i == 0) continue;
      auto prev_f = l0_files[i - 1];
      if (icmp->Compare(prev_f->largest, f->smallest) >= 0) {
-        Log(InfoLogLevel::INFO_LEVEL, immutable_db_options_.info_log,
-            "PromoteL0 FAILED. Files %" PRIu64 " and %" PRIu64
-            " have overlapping ranges\n",
-            prev_f->fd.GetNumber(), f->fd.GetNumber());
+        ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                       "PromoteL0 FAILED. Files %" PRIu64 " and %" PRIu64
+                       " have overlapping ranges\n",
+                       prev_f->fd.GetNumber(), f->fd.GetNumber());
        job_context.Clean();
        return Status::InvalidArgument("L0 has overlapping files");
      }
@@ -116,8 +117,8 @@ Status DBImpl::PromoteL0(ColumnFamilyHandle* column_family, int target_level) {
    // Check that all levels up to target_level are empty.
    for (int level = 1; level <= target_level; ++level) {
      if (vstorage->NumLevelFiles(level) > 0) {
-        Log(InfoLogLevel::INFO_LEVEL, immutable_db_options_.info_log,
-            "PromoteL0 FAILED. Level %d not empty\n", level);
+        ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                       "PromoteL0 FAILED. Level %d not empty\n", level);
        job_context.Clean();
        return Status::InvalidArgument(
            "All levels up to target_level "
@@ -0,0 +1,536 @@
+//  Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+//
+// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file. See the AUTHORS file for names of contributors.
+#include "db/db_impl.h"
+
+#ifndef __STDC_FORMAT_MACROS
+#define __STDC_FORMAT_MACROS
+#endif
+#include <inttypes.h>
+#include "db/event_helpers.h"
+#include "util/file_util.h"
+#include "util/sst_file_manager_impl.h"
+
+
+namespace rocksdb {
+uint64_t DBImpl::FindMinPrepLogReferencedByMemTable() {
+  if (!allow_2pc()) {
+    return 0;
+  }
+
+  uint64_t min_log = 0;
+
+  // we must look through the memtables for two phase transactions
+  // that have been committed but not yet flushed
+  for (auto loop_cfd : *versions_->GetColumnFamilySet()) {
+    if (loop_cfd->IsDropped()) {
+      continue;
+    }
+
+    auto log = loop_cfd->imm()->GetMinLogContainingPrepSection();
+
+    if (log > 0 && (min_log == 0 || log < min_log)) {
+      min_log = log;
+    }
+
+    log = loop_cfd->mem()->GetMinLogContainingPrepSection();
+
+    if (log > 0 && (min_log == 0 || log < min_log)) {
+      min_log = log;
+    }
+  }
+
+  return min_log;
+}
+
+void DBImpl::MarkLogAsHavingPrepSectionFlushed(uint64_t log) {
+  assert(log != 0);
+  std::lock_guard<std::mutex> lock(prep_heap_mutex_);
+  auto it = prepared_section_completed_.find(log);
+  assert(it != prepared_section_completed_.end());
+  it->second += 1;
+}
+
+void DBImpl::MarkLogAsContainingPrepSection(uint64_t log) {
+  assert(log != 0);
+  std::lock_guard<std::mutex> lock(prep_heap_mutex_);
+  min_log_with_prep_.push(log);
+  auto it = prepared_section_completed_.find(log);
+  if (it == prepared_section_completed_.end()) {
+    prepared_section_completed_[log] = 0;
+  }
+}
+
+uint64_t DBImpl::FindMinLogContainingOutstandingPrep() {
+
+  if (!allow_2pc()) {
+    return 0;
+  }
+
+  std::lock_guard<std::mutex> lock(prep_heap_mutex_);
+  uint64_t min_log = 0;
+
+  // first we look in the prepared heap where we keep
+  // track of transactions that have been prepared (written to WAL)
+  // but not yet committed.
+  while (!min_log_with_prep_.empty()) {
+    min_log = min_log_with_prep_.top();
+
+    auto it = prepared_section_completed_.find(min_log);
+
+    // value was marked as 'deleted' from heap
+    if (it != prepared_section_completed_.end() && it->second > 0) {
+      it->second -= 1;
+      min_log_with_prep_.pop();
+
+      // back to squere one...
+      min_log = 0;
+      continue;
+    } else {
+      // found a valid value
+      break;
+    }
+  }
+
+  return min_log;
+}
+
+uint64_t DBImpl::MinLogNumberToKeep() {
+  uint64_t log_number = versions_->MinLogNumber();
+
+  if (allow_2pc()) {
+    // if are 2pc we must consider logs containing prepared
+    // sections of outstanding transactions.
+    //
+    // We must check min logs with outstanding prep before we check
+    // logs referneces by memtables because a log referenced by the
+    // first data structure could transition to the second under us.
+    //
+    // TODO(horuff): iterating over all column families under db mutex.
+    // should find more optimial solution
+    auto min_log_in_prep_heap = FindMinLogContainingOutstandingPrep();
+
+    if (min_log_in_prep_heap != 0 && min_log_in_prep_heap < log_number) {
+      log_number = min_log_in_prep_heap;
+    }
+
+    auto min_log_refed_by_mem = FindMinPrepLogReferencedByMemTable();
+
+    if (min_log_refed_by_mem != 0 && min_log_refed_by_mem < log_number) {
+      log_number = min_log_refed_by_mem;
+    }
+  }
+  return log_number;
+}
+
+// * Returns the list of live files in 'sst_live'
+// If it's doing full scan:
+// * Returns the list of all files in the filesystem in
+// 'full_scan_candidate_files'.
+// Otherwise, gets obsolete files from VersionSet.
+// no_full_scan = true -- never do the full scan using GetChildren()
+// force = false -- don't force the full scan, except every
+//  mutable_db_options_.delete_obsolete_files_period_micros
+// force = true -- force the full scan
+void DBImpl::FindObsoleteFiles(JobContext* job_context, bool force,
+                               bool no_full_scan) {
+  mutex_.AssertHeld();
+
+  // if deletion is disabled, do nothing
+  if (disable_delete_obsolete_files_ > 0) {
+    return;
+  }
+
+  bool doing_the_full_scan = false;
+
+  // logic for figurint out if we're doing the full scan
+  if (no_full_scan) {
+    doing_the_full_scan = false;
+  } else if (force ||
+             mutable_db_options_.delete_obsolete_files_period_micros == 0) {
+    doing_the_full_scan = true;
+  } else {
+    const uint64_t now_micros = env_->NowMicros();
+    if ((delete_obsolete_files_last_run_ +
+         mutable_db_options_.delete_obsolete_files_period_micros) <
+        now_micros) {
+      doing_the_full_scan = true;
+      delete_obsolete_files_last_run_ = now_micros;
+    }
+  }
+
+  // don't delete files that might be currently written to from compaction
+  // threads
+  // Since job_context->min_pending_output is set, until file scan finishes,
+  // mutex_ cannot be released. Otherwise, we might see no min_pending_output
+  // here but later find newer generated unfinalized files while scannint.
+  if (!pending_outputs_.empty()) {
+    job_context->min_pending_output = *pending_outputs_.begin();
+  } else {
+    // delete all of them
+    job_context->min_pending_output = std::numeric_limits<uint64_t>::max();
+  }
+
+  // Get obsolete files.  This function will also update the list of
+  // pending files in VersionSet().
+  versions_->GetObsoleteFiles(&job_context->sst_delete_files,
+                              &job_context->manifest_delete_files,
+                              job_context->min_pending_output);
+
+  // store the current filenum, lognum, etc
+  job_context->manifest_file_number = versions_->manifest_file_number();
+  job_context->pending_manifest_file_number =
+      versions_->pending_manifest_file_number();
+  job_context->log_number = MinLogNumberToKeep();
+
+  job_context->prev_log_number = versions_->prev_log_number();
+
+  versions_->AddLiveFiles(&job_context->sst_live);
+  if (doing_the_full_scan) {
+    for (size_t path_id = 0; path_id < immutable_db_options_.db_paths.size();
+         path_id++) {
+      // set of all files in the directory. We'll exclude files that are still
+      // alive in the subsequent processings.
+      std::vector<std::string> files;
+      env_->GetChildren(immutable_db_options_.db_paths[path_id].path,
+                        &files);  // Ignore errors
+      for (std::string file : files) {
+        // TODO(icanadi) clean up this mess to avoid having one-off "/" prefixes
+        job_context->full_scan_candidate_files.emplace_back(
+            "/" + file, static_cast<uint32_t>(path_id));
+      }
+    }
+
+    // Add log files in wal_dir
+    if (immutable_db_options_.wal_dir != dbname_) {
+      std::vector<std::string> log_files;
+      env_->GetChildren(immutable_db_options_.wal_dir,
+                        &log_files);  // Ignore errors
+      for (std::string log_file : log_files) {
+        job_context->full_scan_candidate_files.emplace_back(log_file, 0);
+      }
+    }
+    // Add info log files in db_log_dir
+    if (!immutable_db_options_.db_log_dir.empty() &&
+        immutable_db_options_.db_log_dir != dbname_) {
+      std::vector<std::string> info_log_files;
+      // Ignore errors
+      env_->GetChildren(immutable_db_options_.db_log_dir, &info_log_files);
+      for (std::string log_file : info_log_files) {
+        job_context->full_scan_candidate_files.emplace_back(log_file, 0);
+      }
+    }
+  }
+
+  // logs_ is empty when called during recovery, in which case there can't yet
+  // be any tracked obsolete logs
+  if (!alive_log_files_.empty() && !logs_.empty()) {
+    uint64_t min_log_number = job_context->log_number;
+    size_t num_alive_log_files = alive_log_files_.size();
+    // find newly obsoleted log files
+    while (alive_log_files_.begin()->number < min_log_number) {
+      auto& earliest = *alive_log_files_.begin();
+      if (immutable_db_options_.recycle_log_file_num >
+          log_recycle_files.size()) {
+        ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                       "adding log %" PRIu64 " to recycle list\n",
+                       earliest.number);
+        log_recycle_files.push_back(earliest.number);
+      } else {
+        job_context->log_delete_files.push_back(earliest.number);
+      }
+      if (job_context->size_log_to_delete == 0) {
+        job_context->prev_total_log_size = total_log_size_;
+        job_context->num_alive_log_files = num_alive_log_files;
+      }
+      job_context->size_log_to_delete += earliest.size;
+      total_log_size_ -= earliest.size;
+      alive_log_files_.pop_front();
+      // Current log should always stay alive since it can't have
+      // number < MinLogNumber().
+      assert(alive_log_files_.size());
+    }
+    while (!logs_.empty() && logs_.front().number < min_log_number) {
+      auto& log = logs_.front();
+      if (log.getting_synced) {
+        log_sync_cv_.Wait();
+        // logs_ could have changed while we were waiting.
+        continue;
+      }
+      logs_to_free_.push_back(log.ReleaseWriter());
+      logs_.pop_front();
+    }
+    // Current log cannot be obsolete.
+    assert(!logs_.empty());
+  }
+
+  // We're just cleaning up for DB::Write().
+  assert(job_context->logs_to_free.empty());
+  job_context->logs_to_free = logs_to_free_;
+  job_context->log_recycle_files.assign(log_recycle_files.begin(),
+                                        log_recycle_files.end());
+  logs_to_free_.clear();
+}
+
+namespace {
+bool CompareCandidateFile(const JobContext::CandidateFileInfo& first,
+                          const JobContext::CandidateFileInfo& second) {
+  if (first.file_name > second.file_name) {
+    return true;
+  } else if (first.file_name < second.file_name) {
+    return false;
+  } else {
+    return (first.path_id > second.path_id);
+  }
+}
+};  // namespace
+
+// Delete obsolete files and log status and information of file deletion
+void DBImpl::DeleteObsoleteFileImpl(Status file_deletion_status, int job_id,
+                                    const std::string& fname, FileType type,
+                                    uint64_t number, uint32_t path_id) {
+  if (type == kTableFile) {
+    file_deletion_status =
+        DeleteSSTFile(&immutable_db_options_, fname, path_id);
+  } else {
+    file_deletion_status = env_->DeleteFile(fname);
+  }
+  if (file_deletion_status.ok()) {
+    ROCKS_LOG_DEBUG(immutable_db_options_.info_log,
+                    "[JOB %d] Delete %s type=%d #%" PRIu64 " -- %s\n", job_id,
+                    fname.c_str(), type, number,
+                    file_deletion_status.ToString().c_str());
+  } else if (env_->FileExists(fname).IsNotFound()) {
+    ROCKS_LOG_INFO(
+        immutable_db_options_.info_log,
+        "[JOB %d] Tried to delete a non-existing file %s type=%d #%" PRIu64
+        " -- %s\n",
+        job_id, fname.c_str(), type, number,
+        file_deletion_status.ToString().c_str());
+  } else {
+    ROCKS_LOG_ERROR(immutable_db_options_.info_log,
+                    "[JOB %d] Failed to delete %s type=%d #%" PRIu64 " -- %s\n",
+                    job_id, fname.c_str(), type, number,
+                    file_deletion_status.ToString().c_str());
+  }
+  if (type == kTableFile) {
+    EventHelpers::LogAndNotifyTableFileDeletion(
+        &event_logger_, job_id, number, fname, file_deletion_status, GetName(),
+        immutable_db_options_.listeners);
+  }
+}
+
+// Diffs the files listed in filenames and those that do not
+// belong to live files are posibly removed. Also, removes all the
+// files in sst_delete_files and log_delete_files.
+// It is not necessary to hold the mutex when invoking this method.
+void DBImpl::PurgeObsoleteFiles(const JobContext& state, bool schedule_only) {
+  // we'd better have sth to delete
+  assert(state.HaveSomethingToDelete());
+
+  // this checks if FindObsoleteFiles() was run before. If not, don't do
+  // PurgeObsoleteFiles(). If FindObsoleteFiles() was run, we need to also
+  // run PurgeObsoleteFiles(), even if disable_delete_obsolete_files_ is true
+  if (state.manifest_file_number == 0) {
+    return;
+  }
+
+  // Now, convert live list to an unordered map, WITHOUT mutex held;
+  // set is slow.
+  std::unordered_map<uint64_t, const FileDescriptor*> sst_live_map;
+  for (const FileDescriptor& fd : state.sst_live) {
+    sst_live_map[fd.GetNumber()] = &fd;
+  }
+  std::unordered_set<uint64_t> log_recycle_files_set(
+      state.log_recycle_files.begin(), state.log_recycle_files.end());
+
+  auto candidate_files = state.full_scan_candidate_files;
+  candidate_files.reserve(
+      candidate_files.size() + state.sst_delete_files.size() +
+      state.log_delete_files.size() + state.manifest_delete_files.size());
+  // We may ignore the dbname when generating the file names.
+  const char* kDumbDbName = "";
+  for (auto file : state.sst_delete_files) {
+    candidate_files.emplace_back(
+        MakeTableFileName(kDumbDbName, file->fd.GetNumber()),
+        file->fd.GetPathId());
+    delete file;
+  }
+
+  for (auto file_num : state.log_delete_files) {
+    if (file_num > 0) {
+      candidate_files.emplace_back(LogFileName(kDumbDbName, file_num), 0);
+    }
+  }
+  for (const auto& filename : state.manifest_delete_files) {
+    candidate_files.emplace_back(filename, 0);
+  }
+
+  // dedup state.candidate_files so we don't try to delete the same
+  // file twice
+  std::sort(candidate_files.begin(), candidate_files.end(),
+            CompareCandidateFile);
+  candidate_files.erase(
+      std::unique(candidate_files.begin(), candidate_files.end()),
+      candidate_files.end());
+
+  if (state.prev_total_log_size > 0) {
+    ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                   "[JOB %d] Try to delete WAL files size %" PRIu64
+                   ", prev total WAL file size %" PRIu64
+                   ", number of live WAL files %" ROCKSDB_PRIszt ".\n",
+                   state.job_id, state.size_log_to_delete,
+                   state.prev_total_log_size, state.num_alive_log_files);
+  }
+
+  std::vector<std::string> old_info_log_files;
+  InfoLogPrefix info_log_prefix(!immutable_db_options_.db_log_dir.empty(),
+                                dbname_);
+  for (const auto& candidate_file : candidate_files) {
+    std::string to_delete = candidate_file.file_name;
+    uint32_t path_id = candidate_file.path_id;
+    uint64_t number;
+    FileType type;
+    // Ignore file if we cannot recognize it.
+    if (!ParseFileName(to_delete, &number, info_log_prefix.prefix, &type)) {
+      continue;
+    }
+
+    bool keep = true;
+    switch (type) {
+      case kLogFile:
+        keep = ((number >= state.log_number) ||
+                (number == state.prev_log_number) ||
+                (log_recycle_files_set.find(number) !=
+                 log_recycle_files_set.end()));
+        break;
+      case kDescriptorFile:
+        // Keep my manifest file, and any newer incarnations'
+        // (can happen during manifest roll)
+        keep = (number >= state.manifest_file_number);
+        break;
+      case kTableFile:
+        // If the second condition is not there, this makes
+        // DontDeletePendingOutputs fail
+        keep = (sst_live_map.find(number) != sst_live_map.end()) ||
+               number >= state.min_pending_output;
+        break;
+      case kTempFile:
+        // Any temp files that are currently being written to must
+        // be recorded in pending_outputs_, which is inserted into "live".
+        // Also, SetCurrentFile creates a temp file when writing out new
+        // manifest, which is equal to state.pending_manifest_file_number. We
+        // should not delete that file
+        //
+        // TODO(yhchiang): carefully modify the third condition to safely
+        //                 remove the temp options files.
+        keep = (sst_live_map.find(number) != sst_live_map.end()) ||
+               (number == state.pending_manifest_file_number) ||
+               (to_delete.find(kOptionsFileNamePrefix) != std::string::npos);
+        break;
+      case kInfoLogFile:
+        keep = true;
+        if (number != 0) {
+          old_info_log_files.push_back(to_delete);
+        }
+        break;
+      case kCurrentFile:
+      case kDBLockFile:
+      case kIdentityFile:
+      case kMetaDatabase:
+      case kOptionsFile:
+      case kBlobFile:
+        keep = true;
+        break;
+    }
+
+    if (keep) {
+      continue;
+    }
+
+    std::string fname;
+    if (type == kTableFile) {
+      // evict from cache
+      TableCache::Evict(table_cache_.get(), number);
+      fname = TableFileName(immutable_db_options_.db_paths, number, path_id);
+    } else {
+      fname = ((type == kLogFile) ? immutable_db_options_.wal_dir : dbname_) +
+              "/" + to_delete;
+    }
+
+#ifndef ROCKSDB_LITE
+    if (type == kLogFile && (immutable_db_options_.wal_ttl_seconds > 0 ||
+                             immutable_db_options_.wal_size_limit_mb > 0)) {
+      wal_manager_.ArchiveWALFile(fname, number);
+      continue;
+    }
+#endif  // !ROCKSDB_LITE
+
+    Status file_deletion_status;
+    if (schedule_only) {
+      InstrumentedMutexLock guard_lock(&mutex_);
+      SchedulePendingPurge(fname, type, number, path_id, state.job_id);
+    } else {
+      DeleteObsoleteFileImpl(file_deletion_status, state.job_id, fname, type,
+                             number, path_id);
+    }
+  }
+
+  // Delete old info log files.
+  size_t old_info_log_file_count = old_info_log_files.size();
+  if (old_info_log_file_count != 0 &&
+      old_info_log_file_count >= immutable_db_options_.keep_log_file_num) {
+    std::sort(old_info_log_files.begin(), old_info_log_files.end());
+    size_t end =
+        old_info_log_file_count - immutable_db_options_.keep_log_file_num;
+    for (unsigned int i = 0; i <= end; i++) {
+      std::string& to_delete = old_info_log_files.at(i);
+      std::string full_path_to_delete =
+          (immutable_db_options_.db_log_dir.empty()
+               ? dbname_
+               : immutable_db_options_.db_log_dir) +
+          "/" + to_delete;
+      ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                     "[JOB %d] Delete info log file %s\n", state.job_id,
+                     full_path_to_delete.c_str());
+      Status s = env_->DeleteFile(full_path_to_delete);
+      if (!s.ok()) {
+        if (env_->FileExists(full_path_to_delete).IsNotFound()) {
+          ROCKS_LOG_INFO(
+              immutable_db_options_.info_log,
+              "[JOB %d] Tried to delete non-existing info log file %s FAILED "
+              "-- %s\n",
+              state.job_id, to_delete.c_str(), s.ToString().c_str());
+        } else {
+          ROCKS_LOG_ERROR(immutable_db_options_.info_log,
+                          "[JOB %d] Delete info log file %s FAILED -- %s\n",
+                          state.job_id, to_delete.c_str(),
+                          s.ToString().c_str());
+        }
+      }
+    }
+  }
+#ifndef ROCKSDB_LITE
+  wal_manager_.PurgeObsoleteWALFiles();
+#endif  // ROCKSDB_LITE
+  LogFlush(immutable_db_options_.info_log);
+}
+
+void DBImpl::DeleteObsoleteFiles() {
+  mutex_.AssertHeld();
+  JobContext job_context(next_job_id_.fetch_add(1));
+  FindObsoleteFiles(&job_context, true);
+
+  mutex_.Unlock();
+  if (job_context.HaveSomethingToDelete()) {
+    PurgeObsoleteFiles(job_context);
+  }
+  job_context.Clean();
+  mutex_.Lock();
+}
+}  // namespace rocksdb
@@ -11,7 +11,7 @@
 #include "db/db_iter.h"
 #include "db/merge_context.h"
 #include "db/range_del_aggregator.h"
-#include "util/perf_context_imp.h"
+#include "monitoring/perf_context_imp.h"

 namespace rocksdb {

@@ -20,8 +20,8 @@ namespace rocksdb {
 DBImplReadOnly::DBImplReadOnly(const DBOptions& db_options,
                               const std::string& dbname)
    : DBImpl(db_options, dbname) {
-  Log(INFO_LEVEL, immutable_db_options_.info_log,
-      "Opening the db in read only mode");
+  ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                 "Opening the db in read only mode");
  LogFlush(immutable_db_options_.info_log);
 }

@@ -31,7 +31,8 @@ DBImplReadOnly::~DBImplReadOnly() {
 // Implementations of the DB interface
 Status DBImplReadOnly::Get(const ReadOptions& read_options,
                           ColumnFamilyHandle* column_family, const Slice& key,
-                           std::string* value) {
+                           PinnableSlice* pinnable_val) {
+  assert(pinnable_val != nullptr);
  Status s;
  SequenceNumber snapshot = versions_->LastSequence();
  auto cfh = reinterpret_cast<ColumnFamilyHandleImpl*>(column_family);
@@ -40,12 +41,13 @@ Status DBImplReadOnly::Get(const ReadOptions& read_options,
  MergeContext merge_context;
  RangeDelAggregator range_del_agg(cfd->internal_comparator(), snapshot);
  LookupKey lkey(key, snapshot);
-  if (super_version->mem->Get(lkey, value, &s, &merge_context, &range_del_agg,
-                              read_options)) {
+  if (super_version->mem->Get(lkey, pinnable_val->GetSelf(), &s, &merge_context,
+                              &range_del_agg, read_options)) {
+    pinnable_val->PinSelf();
  } else {
    PERF_TIMER_GUARD(get_from_output_files_time);
-    super_version->current->Get(read_options, lkey, value, &s, &merge_context,
-                                &range_del_agg);
+    super_version->current->Get(read_options, lkey, pinnable_val, &s,
+                                &merge_context, &range_del_agg);
  }
  return s;
 }
@@ -57,7 +59,7 @@ Iterator* DBImplReadOnly::NewIterator(const ReadOptions& read_options,
  SuperVersion* super_version = cfd->GetSuperVersion()->Ref();
  SequenceNumber latest_snapshot = versions_->LastSequence();
  auto db_iter = NewArenaWrappedDbIterator(
-      env_, *cfd->ioptions(), cfd->user_comparator(),
+      env_, read_options, *cfd->ioptions(), cfd->user_comparator(),
      (read_options.snapshot != nullptr
           ? reinterpret_cast<const SnapshotImpl*>(read_options.snapshot)
                 ->number_
@@ -86,7 +88,7 @@ Status DBImplReadOnly::NewIterators(
    auto* cfd = reinterpret_cast<ColumnFamilyHandleImpl*>(cfh)->cfd();
    auto* sv = cfd->GetSuperVersion()->Ref();
    auto* db_iter = NewArenaWrappedDbIterator(
-        env_, *cfd->ioptions(), cfd->user_comparator(),
+        env_, read_options, *cfd->ioptions(), cfd->user_comparator(),
        (read_options.snapshot != nullptr
             ? reinterpret_cast<const SnapshotImpl*>(read_options.snapshot)
                   ->number_
@@ -22,7 +22,7 @@ class DBImplReadOnly : public DBImpl {
  using DB::Get;
  virtual Status Get(const ReadOptions& options,
                     ColumnFamilyHandle* column_family, const Slice& key,
-                     std::string* value) override;
+                     PinnableSlice* value) override;

  // TODO: Implement ReadOnly MultiGet?

@@ -0,0 +1,834 @@
+//  Copyright (c) 2011-present, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under both the GPLv2 (found in the
+//  COPYING file in the root directory) and Apache 2.0 License
+//  (found in the LICENSE.Apache file in the root directory).
+//
+// Copyright (c) 2011 The LevelDB Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file. See the AUTHORS file for names of contributors.
+#include "db/db_impl.h"
+
+#ifndef __STDC_FORMAT_MACROS
+#define __STDC_FORMAT_MACROS
+#endif
+#include <inttypes.h>
+#include "monitoring/perf_context_imp.h"
+#include "options/options_helper.h"
+#include "util/sync_point.h"
+
+namespace rocksdb {
+// Convenience methods
+Status DBImpl::Put(const WriteOptions& o, ColumnFamilyHandle* column_family,
+                   const Slice& key, const Slice& val) {
+  return DB::Put(o, column_family, key, val);
+}
+
+Status DBImpl::Merge(const WriteOptions& o, ColumnFamilyHandle* column_family,
+                     const Slice& key, const Slice& val) {
+  auto cfh = reinterpret_cast<ColumnFamilyHandleImpl*>(column_family);
+  if (!cfh->cfd()->ioptions()->merge_operator) {
+    return Status::NotSupported("Provide a merge_operator when opening DB");
+  } else {
+    return DB::Merge(o, column_family, key, val);
+  }
+}
+
+Status DBImpl::Delete(const WriteOptions& write_options,
+                      ColumnFamilyHandle* column_family, const Slice& key) {
+  return DB::Delete(write_options, column_family, key);
+}
+
+Status DBImpl::SingleDelete(const WriteOptions& write_options,
+                            ColumnFamilyHandle* column_family,
+                            const Slice& key) {
+  return DB::SingleDelete(write_options, column_family, key);
+}
+
+Status DBImpl::Write(const WriteOptions& write_options, WriteBatch* my_batch) {
+  return WriteImpl(write_options, my_batch, nullptr, nullptr);
+}
+
+#ifndef ROCKSDB_LITE
+Status DBImpl::WriteWithCallback(const WriteOptions& write_options,
+                                 WriteBatch* my_batch,
+                                 WriteCallback* callback) {
+  return WriteImpl(write_options, my_batch, callback, nullptr);
+}
+#endif  // ROCKSDB_LITE
+
+Status DBImpl::WriteImpl(const WriteOptions& write_options,
+                         WriteBatch* my_batch, WriteCallback* callback,
+                         uint64_t* log_used, uint64_t log_ref,
+                         bool disable_memtable) {
+  if (my_batch == nullptr) {
+    return Status::Corruption("Batch is nullptr!");
+  }
+
+  Status status;
+
+  PERF_TIMER_GUARD(write_pre_and_post_process_time);
+  WriteThread::Writer w(write_options, my_batch, callback, log_ref,
+                        disable_memtable);
+
+  if (!write_options.disableWAL) {
+    RecordTick(stats_, WRITE_WITH_WAL);
+  }
+
+  StopWatch write_sw(env_, immutable_db_options_.statistics.get(), DB_WRITE);
+
+  write_thread_.JoinBatchGroup(&w);
+  if (w.state == WriteThread::STATE_PARALLEL_FOLLOWER) {
+    // 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_,
+          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;
+      versions_->SetLastSequence(last_sequence);
+      write_thread_.EarlyExitParallelGroup(&w);
+    }
+    assert(w.state == WriteThread::STATE_COMPLETED);
+    // STATE_COMPLETED conditional below handles exit
+
+    status = w.FinalStatus();
+  }
+  if (w.state == WriteThread::STATE_COMPLETED) {
+    if (log_used != nullptr) {
+      *log_used = w.log_used;
+    }
+    // write is complete and leader has updated sequence
+    return w.FinalStatus();
+  }
+  // else we are the leader of the write batch group
+  assert(w.state == WriteThread::STATE_GROUP_LEADER);
+
+  // Once reaches this point, the current writer "w" will try to do its write
+  // job.  It may also pick up some of the remaining writers in the "writers_"
+  // 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;
+
+  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;
+
+  mutex_.Unlock();
+
+  // Add to log and apply to memtable.  We can release the lock
+  // during this phase since &w is currently responsible for logging
+  // and protects against concurrent loggers and concurrent writes
+  // into memtables
+
+  bool exit_completed_early = false;
+  last_batch_group_size_ =
+      write_thread_.EnterAsBatchGroupLeader(&w, &last_writer, &write_group);
+
+  if (status.ok()) {
+    // Rules for when we can update the memtable concurrently
+    // 1. supported by memtable
+    // 2. Puts are not okay if inplace_update_support
+    // 3. Merges are not okay
+    //
+    // Rules 1..2 are enforced by checking the options
+    // during startup (CheckConcurrentWritesSupported), so if
+    // options.allow_concurrent_memtable_write is true then they can be
+    // assumed to be true.  Rule 3 is checked for each batch.  We could
+    // 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;
+    int total_count = 0;
+    uint64_t total_byte_size = 0;
+    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));
+        }
+      }
+    }
+
+    uint64_t last_sequence = versions_->LastSequence();
+    const SequenceNumber current_sequence = last_sequence + 1;
+    last_sequence += total_count;
+
+    // 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.
+    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);
+    stats->AddDBStats(InternalStats::WRITE_DONE_BY_SELF, 1);
+    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);
+      RecordTick(stats_, WRITE_DONE_BY_OTHER, write_done_by_other);
+    }
+    MeasureTime(stats_, BYTES_PER_WRITE, total_byte_size);
+
+    if (write_options.disableWAL) {
+      has_unpersisted_data_.store(true, std::memory_order_relaxed);
+    }
+
+    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 (status.ok()) {
+      PERF_TIMER_GUARD(write_memtable_time);
+
+      if (!parallel) {
+        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);
+
+        // Each parallel follower is doing each own writes. The leader should
+        // also do its own.
+        if (w.ShouldWriteToMemtable()) {
+          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_,
+              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 (logs_getting_synced) {
+    mutex_.Lock();
+    MarkLogsSynced(logfile_number_, need_log_dir_sync, status);
+    mutex_.Unlock();
+  }
+
+  if (!exit_completed_early) {
+    write_thread_.ExitAsBatchGroupLeader(&w, last_writer, w.status);
+  }
+
+  return status;
+}
+
+Status DBImpl::PreprocessWrite(const WriteOptions& write_options,
+                               bool need_log_sync, bool* logs_getting_synced,
+                               WriteContext* write_context) {
+  mutex_.AssertHeld();
+  assert(write_context != nullptr && logs_getting_synced != nullptr);
+  Status status;
+
+  assert(!single_column_family_mode_ ||
+         versions_->GetColumnFamilySet()->NumberOfColumnFamilies() == 1);
+  if (UNLIKELY(status.ok() && !single_column_family_mode_ &&
+               total_log_size_ > GetMaxTotalWalSize())) {
+    status = HandleWALFull(write_context);
+  }
+
+  if (UNLIKELY(status.ok() && write_buffer_manager_->ShouldFlush())) {
+    // Before a new memtable is added in SwitchMemtable(),
+    // write_buffer_manager_->ShouldFlush() will keep returning true. If another
+    // thread is writing to another DB with the same write buffer, they may also
+    // be flushed. We may end up with flushing much more DBs than needed. It's
+    // suboptimal but still correct.
+    status = HandleWriteBufferFull(write_context);
+  }
+
+  if (UNLIKELY(status.ok() && !bg_error_.ok())) {
+    status = bg_error_;
+  }
+
+  if (UNLIKELY(status.ok() && !flush_scheduler_.Empty())) {
+    status = ScheduleFlushes(write_context);
+  }
+
+  if (UNLIKELY(status.ok() && (write_controller_.IsStopped() ||
+                               write_controller_.NeedsDelay()))) {
+    PERF_TIMER_GUARD(write_delay_time);
+    // We don't know size of curent batch so that we always use the size
+    // for previous one. It might create a fairness issue that expiration
+    // might happen for smaller writes but larger writes can go through.
+    // Can optimize it if it is an issue.
+    status = DelayWrite(last_batch_group_size_, write_options);
+  }
+
+  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
+    // finished
+    // Note: there does not seem to be a reason to wait for parallel sync at
+    // this early step but it is not important since parallel sync (SyncWAL) and
+    // need_log_sync are usually not used together.
+    while (logs_.front().getting_synced) {
+      log_sync_cv_.Wait();
+    }
+    for (auto& log : logs_) {
+      assert(!log.getting_synced);
+      // This is just to prevent the logs to be synced by a parallel SyncWAL
+      // call. We will do the actual syncing later after we will write to the
+      // WAL.
+      // Note: there does not seem to be a reason to set this early before we
+      // actually write to the WAL
+      log.getting_synced = true;
+    }
+    *logs_getting_synced = true;
+  }
+
+  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* 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()) {
+    // 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;
+  } 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_;
+    for (auto writer : write_group) {
+      if (writer->ShouldWriteToWAL()) {
+        WriteBatchInternal::Append(merged_batch, writer->batch,
+                                   /*WAL_only*/ true);
+        write_with_wal++;
+      }
+      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();
+
+  if (status.ok() && need_log_sync) {
+    StopWatch sw(env_, stats_, WAL_FILE_SYNC_MICROS);
+    // It's safe to access logs_ with unlocked mutex_ here because:
+    //  - we've set getting_synced=true for all logs,
+    //    so other threads won't pop from logs_ while we're here,
+    //  - only writer thread can push to logs_, and we're in
+    //    writer thread, so no one will push to logs_,
+    //  - as long as other threads don't modify it, it's safe to read
+    //    from std::deque from multiple threads concurrently.
+    for (auto& log : logs_) {
+      status = log.writer->file()->Sync(immutable_db_options_.use_fsync);
+      if (!status.ok()) {
+        break;
+      }
+    }
+    if (status.ok() && need_log_dir_sync) {
+      // We only sync WAL directory the first time WAL syncing is
+      // requested, so that in case users never turn on WAL sync,
+      // we can avoid the disk I/O in the write code path.
+      status = directories_.GetWalDir()->Fsync();
+    }
+  }
+
+  if (merged_batch == &tmp_batch_) {
+    tmp_batch_.Clear();
+  }
+  if (status.ok()) {
+    auto stats = default_cf_internal_stats_;
+    if (need_log_sync) {
+      stats->AddDBStats(InternalStats::WAL_FILE_SYNCED, 1);
+      RecordTick(stats_, WAL_FILE_SYNCED);
+    }
+    stats->AddDBStats(InternalStats::WAL_FILE_BYTES, log_size);
+    RecordTick(stats_, WAL_FILE_BYTES, log_size);
+    stats->AddDBStats(InternalStats::WRITE_WITH_WAL, write_with_wal);
+    RecordTick(stats_, WRITE_WITH_WAL, write_with_wal);
+  }
+  return status;
+}
+
+Status DBImpl::HandleWALFull(WriteContext* write_context) {
+  mutex_.AssertHeld();
+  assert(write_context != nullptr);
+  Status status;
+
+  if (alive_log_files_.begin()->getting_flushed) {
+    return status;
+  }
+
+  auto oldest_alive_log = alive_log_files_.begin()->number;
+  auto oldest_log_with_uncommited_prep = FindMinLogContainingOutstandingPrep();
+
+  if (allow_2pc() &&
+      oldest_log_with_uncommited_prep > 0 &&
+      oldest_log_with_uncommited_prep <= oldest_alive_log) {
+    if (unable_to_flush_oldest_log_) {
+        // we already attempted to flush all column families dependent on
+        // the oldest alive log but the log still contained uncommited transactions.
+        // the oldest alive log STILL contains uncommited transaction so there
+        // is still nothing that we can do.
+        return status;
+    } else {
+      ROCKS_LOG_WARN(
+          immutable_db_options_.info_log,
+          "Unable to release oldest log due to uncommited transaction");
+      unable_to_flush_oldest_log_ = true;
+    }
+  } else {
+    // we only mark this log as getting flushed if we have successfully
+    // flushed all data in this log. If this log contains outstanding prepared
+    // transactions then we cannot flush this log until those transactions are commited.
+    unable_to_flush_oldest_log_ = false;
+    alive_log_files_.begin()->getting_flushed = true;
+  }
+
+  ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                 "Flushing all column families with data in WAL number %" PRIu64
+                 ". Total log size is %" PRIu64
+                 " while max_total_wal_size is %" PRIu64,
+                 oldest_alive_log, total_log_size_.load(), GetMaxTotalWalSize());
+  // no need to refcount because drop is happening in write thread, so can't
+  // happen while we're in the write thread
+  for (auto cfd : *versions_->GetColumnFamilySet()) {
+    if (cfd->IsDropped()) {
+      continue;
+    }
+    if (cfd->OldestLogToKeep() <= oldest_alive_log) {
+      status = SwitchMemtable(cfd, write_context);
+      if (!status.ok()) {
+        break;
+      }
+      cfd->imm()->FlushRequested();
+      SchedulePendingFlush(cfd);
+    }
+  }
+  MaybeScheduleFlushOrCompaction();
+  return status;
+}
+
+Status DBImpl::HandleWriteBufferFull(WriteContext* write_context) {
+  mutex_.AssertHeld();
+  assert(write_context != nullptr);
+  Status status;
+
+  // Before a new memtable is added in SwitchMemtable(),
+  // write_buffer_manager_->ShouldFlush() will keep returning true. If another
+  // thread is writing to another DB with the same write buffer, they may also
+  // be flushed. We may end up with flushing much more DBs than needed. It's
+  // suboptimal but still correct.
+  ROCKS_LOG_INFO(
+      immutable_db_options_.info_log,
+      "Flushing column family with largest mem table size. Write buffer is "
+      "using %" PRIu64 " bytes out of a total of %" PRIu64 ".",
+      write_buffer_manager_->memory_usage(),
+      write_buffer_manager_->buffer_size());
+  // no need to refcount because drop is happening in write thread, so can't
+  // happen while we're in the write thread
+  ColumnFamilyData* cfd_picked = nullptr;
+  SequenceNumber seq_num_for_cf_picked = kMaxSequenceNumber;
+
+  for (auto cfd : *versions_->GetColumnFamilySet()) {
+    if (cfd->IsDropped()) {
+      continue;
+    }
+    if (!cfd->mem()->IsEmpty()) {
+      // We only consider active mem table, hoping immutable memtable is
+      // already in the process of flushing.
+      uint64_t seq = cfd->mem()->GetCreationSeq();
+      if (cfd_picked == nullptr || seq < seq_num_for_cf_picked) {
+        cfd_picked = cfd;
+        seq_num_for_cf_picked = seq;
+      }
+    }
+  }
+  if (cfd_picked != nullptr) {
+    status = SwitchMemtable(cfd_picked, write_context);
+    if (status.ok()) {
+      cfd_picked->imm()->FlushRequested();
+      SchedulePendingFlush(cfd_picked);
+      MaybeScheduleFlushOrCompaction();
+    }
+  }
+  return status;
+}
+
+uint64_t DBImpl::GetMaxTotalWalSize() const {
+  mutex_.AssertHeld();
+  return mutable_db_options_.max_total_wal_size == 0
+             ? 4 * max_total_in_memory_state_
+             : mutable_db_options_.max_total_wal_size;
+}
+
+// REQUIRES: mutex_ is held
+// REQUIRES: this thread is currently at the front of the writer queue
+Status DBImpl::DelayWrite(uint64_t num_bytes,
+                          const WriteOptions& write_options) {
+  uint64_t time_delayed = 0;
+  bool delayed = false;
+  {
+    StopWatch sw(env_, stats_, WRITE_STALL, &time_delayed);
+    uint64_t delay = write_controller_.GetDelay(env_, num_bytes);
+    if (delay > 0) {
+      if (write_options.no_slowdown) {
+        return Status::Incomplete();
+      }
+      TEST_SYNC_POINT("DBImpl::DelayWrite:Sleep");
+
+      mutex_.Unlock();
+      // We will delay the write until we have slept for delay ms or
+      // we don't need a delay anymore
+      const uint64_t kDelayInterval = 1000;
+      uint64_t stall_end = sw.start_time() + delay;
+      while (write_controller_.NeedsDelay()) {
+        if (env_->NowMicros() >= stall_end) {
+          // We already delayed this write `delay` microseconds
+          break;
+        }
+
+        delayed = true;
+        // Sleep for 0.001 seconds
+        env_->SleepForMicroseconds(kDelayInterval);
+      }
+      mutex_.Lock();
+    }
+
+    while (bg_error_.ok() && write_controller_.IsStopped()) {
+      if (write_options.no_slowdown) {
+        return Status::Incomplete();
+      }
+      delayed = true;
+      TEST_SYNC_POINT("DBImpl::DelayWrite:Wait");
+      bg_cv_.Wait();
+    }
+  }
+  assert(!delayed || !write_options.no_slowdown);
+  if (delayed) {
+    default_cf_internal_stats_->AddDBStats(InternalStats::WRITE_STALL_MICROS,
+                                           time_delayed);
+    RecordTick(stats_, STALL_MICROS, time_delayed);
+  }
+
+  return bg_error_;
+}
+
+Status DBImpl::ScheduleFlushes(WriteContext* context) {
+  ColumnFamilyData* cfd;
+  while ((cfd = flush_scheduler_.TakeNextColumnFamily()) != nullptr) {
+    auto status = SwitchMemtable(cfd, context);
+    if (cfd->Unref()) {
+      delete cfd;
+    }
+    if (!status.ok()) {
+      return status;
+    }
+  }
+  return Status::OK();
+}
+
+#ifndef ROCKSDB_LITE
+void DBImpl::NotifyOnMemTableSealed(ColumnFamilyData* cfd,
+                                    const MemTableInfo& mem_table_info) {
+  if (immutable_db_options_.listeners.size() == 0U) {
+    return;
+  }
+  if (shutting_down_.load(std::memory_order_acquire)) {
+    return;
+  }
+
+  for (auto listener : immutable_db_options_.listeners) {
+    listener->OnMemTableSealed(mem_table_info);
+  }
+}
+#endif  // ROCKSDB_LITE
+
+// REQUIRES: mutex_ is held
+// REQUIRES: this thread is currently at the front of the writer queue
+Status DBImpl::SwitchMemtable(ColumnFamilyData* cfd, WriteContext* context) {
+  mutex_.AssertHeld();
+  unique_ptr<WritableFile> lfile;
+  log::Writer* new_log = nullptr;
+  MemTable* new_mem = nullptr;
+
+  // 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);
+  bool creating_new_log = !log_empty_;
+  uint64_t recycle_log_number = 0;
+  if (creating_new_log && immutable_db_options_.recycle_log_file_num &&
+      !log_recycle_files.empty()) {
+    recycle_log_number = log_recycle_files.front();
+    log_recycle_files.pop_front();
+  }
+  uint64_t new_log_number =
+      creating_new_log ? versions_->NewFileNumber() : logfile_number_;
+  SuperVersion* new_superversion = nullptr;
+  const MutableCFOptions mutable_cf_options = *cfd->GetLatestMutableCFOptions();
+
+  // Set current_memtble_info for memtable sealed callback
+#ifndef ROCKSDB_LITE
+  MemTableInfo memtable_info;
+  memtable_info.cf_name = cfd->GetName();
+  memtable_info.first_seqno = cfd->mem()->GetFirstSequenceNumber();
+  memtable_info.earliest_seqno = cfd->mem()->GetEarliestSequenceNumber();
+  memtable_info.num_entries = cfd->mem()->num_entries();
+  memtable_info.num_deletes = cfd->mem()->num_deletes();
+#endif  // ROCKSDB_LITE
+  // Log this later after lock release. It may be outdated, e.g., if background
+  // flush happens before logging, but that should be ok.
+  int num_imm_unflushed = cfd->imm()->NumNotFlushed();
+  DBOptions db_options =
+      BuildDBOptions(immutable_db_options_, mutable_db_options_);
+  const auto preallocate_block_size =
+    GetWalPreallocateBlockSize(mutable_cf_options.write_buffer_size);
+  mutex_.Unlock();
+  Status s;
+  {
+    if (creating_new_log) {
+      EnvOptions opt_env_opt =
+          env_->OptimizeForLogWrite(env_options_, db_options);
+      if (recycle_log_number) {
+        ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                       "reusing log %" PRIu64 " from recycle list\n",
+                       recycle_log_number);
+        s = env_->ReuseWritableFile(
+            LogFileName(immutable_db_options_.wal_dir, new_log_number),
+            LogFileName(immutable_db_options_.wal_dir, recycle_log_number),
+            &lfile, opt_env_opt);
+      } else {
+        s = NewWritableFile(
+            env_, LogFileName(immutable_db_options_.wal_dir, new_log_number),
+            &lfile, opt_env_opt);
+      }
+      if (s.ok()) {
+        // Our final size should be less than write_buffer_size
+        // (compression, etc) but err on the side of caution.
+
+        // use preallocate_block_size instead
+        // of calling GetWalPreallocateBlockSize()
+        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);
+      }
+    }
+
+    if (s.ok()) {
+      SequenceNumber seq = versions_->LastSequence();
+      new_mem = cfd->ConstructNewMemtable(mutable_cf_options, seq);
+      new_superversion = new SuperVersion();
+    }
+
+#ifndef ROCKSDB_LITE
+    // PLEASE NOTE: We assume that there are no failable operations
+    // after lock is acquired below since we are already notifying
+    // client about mem table becoming immutable.
+    NotifyOnMemTableSealed(cfd, memtable_info);
+#endif //ROCKSDB_LITE
+  }
+  ROCKS_LOG_INFO(immutable_db_options_.info_log,
+                 "[%s] New memtable created with log file: #%" PRIu64
+                 ". Immutable memtables: %d.\n",
+                 cfd->GetName().c_str(), new_log_number, num_imm_unflushed);
+  mutex_.Lock();
+  if (!s.ok()) {
+    // how do we fail if we're not creating new log?
+    assert(creating_new_log);
+    assert(!new_mem);
+    assert(!new_log);
+    return s;
+  }
+  if (creating_new_log) {
+    logfile_number_ = new_log_number;
+    assert(new_log != nullptr);
+    log_empty_ = true;
+    log_dir_synced_ = false;
+    logs_.emplace_back(logfile_number_, new_log);
+    alive_log_files_.push_back(LogFileNumberSize(logfile_number_));
+  }
+  for (auto loop_cfd : *versions_->GetColumnFamilySet()) {
+    // all this is just optimization to delete logs that
+    // are no longer needed -- if CF is empty, that means it
+    // doesn't need that particular log to stay alive, so we just
+    // advance the log number. no need to persist this in the manifest
+    if (loop_cfd->mem()->GetFirstSequenceNumber() == 0 &&
+        loop_cfd->imm()->NumNotFlushed() == 0) {
+      if (creating_new_log) {
+        loop_cfd->SetLogNumber(logfile_number_);
+      }
+      loop_cfd->mem()->SetCreationSeq(versions_->LastSequence());
+    }
+  }
+
+  cfd->mem()->SetNextLogNumber(logfile_number_);
+  cfd->imm()->Add(cfd->mem(), &context->memtables_to_free_);
+  new_mem->Ref();
+  cfd->SetMemtable(new_mem);
+  context->superversions_to_free_.push_back(InstallSuperVersionAndScheduleWork(
+      cfd, new_superversion, mutable_cf_options));
+  return s;
+}
+
+size_t DBImpl::GetWalPreallocateBlockSize(uint64_t write_buffer_size) const {
+  mutex_.AssertHeld();
+  size_t bsize = write_buffer_size / 10 + write_buffer_size;
+  // Some users might set very high write_buffer_size and rely on
+  // max_total_wal_size or other parameters to control the WAL size.
+  if (mutable_db_options_.max_total_wal_size > 0) {
+    bsize = std::min<size_t>(bsize, mutable_db_options_.max_total_wal_size);
+  }
+  if (immutable_db_options_.db_write_buffer_size > 0) {
+    bsize = std::min<size_t>(bsize, immutable_db_options_.db_write_buffer_size);
+  }
+  if (immutable_db_options_.write_buffer_manager &&
+      immutable_db_options_.write_buffer_manager->enabled()) {
+    bsize = std::min<size_t>(
+        bsize, immutable_db_options_.write_buffer_manager->buffer_size());
+  }
+
+  return bsize;
+}
+
+// Default implementations of convenience methods that subclasses of DB
+// can call if they wish
+Status DB::Put(const WriteOptions& opt, ColumnFamilyHandle* column_family,
+               const Slice& key, const Slice& value) {
+  // Pre-allocate size of write batch conservatively.
+  // 8 bytes are taken by header, 4 bytes for count, 1 byte for type,
+  // and we allocate 11 extra bytes for key length, as well as value length.
+  WriteBatch batch(key.size() + value.size() + 24);
+  batch.Put(column_family, key, value);
+  return Write(opt, &batch);
+}
+
+Status DB::Delete(const WriteOptions& opt, ColumnFamilyHandle* column_family,
+                  const Slice& key) {
+  WriteBatch batch;
+  batch.Delete(column_family, key);
+  return Write(opt, &batch);
+}
+
+Status DB::SingleDelete(const WriteOptions& opt,
+                        ColumnFamilyHandle* column_family, const Slice& key) {
+  WriteBatch batch;
+  batch.SingleDelete(column_family, key);
+  return Write(opt, &batch);
+}
+
+Status DB::DeleteRange(const WriteOptions& opt,
+                       ColumnFamilyHandle* column_family,
+                       const Slice& begin_key, const Slice& end_key) {
+  WriteBatch batch;
+  batch.DeleteRange(column_family, begin_key, end_key);
+  return Write(opt, &batch);
+}
+
+Status DB::Merge(const WriteOptions& opt, ColumnFamilyHandle* column_family,
+                 const Slice& key, const Slice& value) {
+  WriteBatch batch;
+  batch.Merge(column_family, key, value);
+  return Write(opt, &batch);
+}
+}  // namespace rocksdb
@@ -15,8 +15,8 @@
 #include <algorithm>
 #include <vector>

-#include "db/filename.h"
 #include "rocksdb/env.h"
+#include "util/filename.h"

 namespace rocksdb {

@@ -6,7 +6,7 @@

 #include <string>

-#include "util/db_options.h"
+#include "options/db_options.h"

 namespace rocksdb {
 void DumpDBFileSummary(const ImmutableDBOptions& options,
@@ -14,10 +14,10 @@
 #include <limits>

 #include "db/dbformat.h"
-#include "db/filename.h"
 #include "db/merge_context.h"
 #include "db/merge_helper.h"
 #include "db/pinned_iterators_manager.h"
+#include "monitoring/perf_context_imp.h"
 #include "port/port.h"
 #include "rocksdb/env.h"
 #include "rocksdb/iterator.h"
@@ -25,9 +25,9 @@
 #include "rocksdb/options.h"
 #include "table/internal_iterator.h"
 #include "util/arena.h"
+#include "util/filename.h"
 #include "util/logging.h"
 #include "util/mutexlock.h"
-#include "util/perf_context_imp.h"
 #include "util/string_util.h"

 namespace rocksdb {
@@ -101,33 +101,32 @@ class DBIter: public Iterator {
    uint64_t bytes_read_;
  };

-  DBIter(Env* env, const ImmutableCFOptions& ioptions, const Comparator* cmp,
+  DBIter(Env* env, const ReadOptions& read_options,
+         const ImmutableCFOptions& cf_options, const Comparator* cmp,
         InternalIterator* iter, SequenceNumber s, bool arena_mode,
-         uint64_t max_sequential_skip_in_iterations, uint64_t version_number,
-         const Slice* iterate_upper_bound = nullptr,
-         bool prefix_same_as_start = false, bool pin_data = false,
-         bool total_order_seek = false)
+         uint64_t max_sequential_skip_in_iterations, uint64_t version_number)
      : arena_mode_(arena_mode),
        env_(env),
-        logger_(ioptions.info_log),
+        logger_(cf_options.info_log),
        user_comparator_(cmp),
-        merge_operator_(ioptions.merge_operator),
+        merge_operator_(cf_options.merge_operator),
        iter_(iter),
        sequence_(s),
        direction_(kForward),
        valid_(false),
        current_entry_is_merged_(false),
-        statistics_(ioptions.statistics),
+        statistics_(cf_options.statistics),
        version_number_(version_number),
-        iterate_upper_bound_(iterate_upper_bound),
-        prefix_same_as_start_(prefix_same_as_start),
-        pin_thru_lifetime_(pin_data),
-        total_order_seek_(total_order_seek),
-        range_del_agg_(ioptions.internal_comparator, s,
+        iterate_upper_bound_(read_options.iterate_upper_bound),
+        prefix_same_as_start_(read_options.prefix_same_as_start),
+        pin_thru_lifetime_(read_options.pin_data),
+        total_order_seek_(read_options.total_order_seek),
+        range_del_agg_(cf_options.internal_comparator, s,
                       true /* collapse_deletions */) {
    RecordTick(statistics_, NO_ITERATORS);
-    prefix_extractor_ = ioptions.prefix_extractor;
+    prefix_extractor_ = cf_options.prefix_extractor;
    max_skip_ = max_sequential_skip_in_iterations;
+    max_skippable_internal_keys_ = read_options.max_skippable_internal_keys;
    if (pin_thru_lifetime_) {
      pinned_iters_mgr_.StartPinning();
    }
@@ -160,7 +159,7 @@ class DBIter: public Iterator {
  virtual bool Valid() const override { return valid_; }
  virtual Slice key() const override {
    assert(valid_);
-    return saved_key_.GetKey();
+    return saved_key_.GetUserKey();
  }
  virtual Slice value() const override {
    assert(valid_);
@@ -224,6 +223,7 @@ class DBIter: public Iterator {
  void FindNextUserEntryInternal(bool skipping, bool prefix_check);
  bool ParseKey(ParsedInternalKey* key);
  void MergeValuesNewToOld();
+  bool TooManyInternalKeysSkipped(bool increment = true);

  // Temporarily pin the blocks that we encounter until ReleaseTempPinnedData()
  // is called
@@ -249,6 +249,10 @@ class DBIter: public Iterator {
    }
  }

+  inline void ResetInternalKeysSkippedCounter() {
+    num_internal_keys_skipped_ = 0;
+  }
+
  const SliceTransform* prefix_extractor_;
  bool arena_mode_;
  Env* const env_;
@@ -268,6 +272,8 @@ class DBIter: public Iterator {
  // for prefix seek mode to support prev()
  Statistics* statistics_;
  uint64_t max_skip_;
+  uint64_t max_skippable_internal_keys_;
+  uint64_t num_internal_keys_skipped_;
  uint64_t version_number_;
  const Slice* iterate_upper_bound_;
  IterKey prefix_start_buf_;
@@ -291,9 +297,8 @@ class DBIter: public Iterator {
 inline bool DBIter::ParseKey(ParsedInternalKey* ikey) {
  if (!ParseInternalKey(iter_->key(), ikey)) {
    status_ = Status::Corruption("corrupted internal key in DBIter");
-    Log(InfoLogLevel::ERROR_LEVEL,
-        logger_, "corrupted internal key in DBIter: %s",
-        iter_->key().ToString(true).c_str());
+    ROCKS_LOG_ERROR(logger_, "corrupted internal key in DBIter: %s",
+                    iter_->key().ToString(true).c_str());
    return false;
  } else {
    return true;
@@ -305,6 +310,7 @@ void DBIter::Next() {

  // Release temporarily pinned blocks from last operation
  ReleaseTempPinnedData();
+  ResetInternalKeysSkippedCounter();
  if (direction_ == kReverse) {
    ReverseToForward();
  } else if (iter_->Valid() && !current_entry_is_merged_) {
@@ -391,9 +397,14 @@ void DBIter::FindNextUserEntryInternal(bool skipping, bool prefix_check) {
      break;
    }

+    if (TooManyInternalKeysSkipped()) {
+      return;
+    }
+
    if (ikey.sequence <= sequence_) {
      if (skipping &&
-          user_comparator_->Compare(ikey.user_key, saved_key_.GetKey()) <= 0) {
+          user_comparator_->Compare(ikey.user_key, saved_key_.GetUserKey()) <=
+              0) {
        num_skipped++;  // skip this entry
        PERF_COUNTER_ADD(internal_key_skipped_count, 1);
      } else {
@@ -403,14 +414,14 @@ void DBIter::FindNextUserEntryInternal(bool skipping, bool prefix_check) {
          case kTypeSingleDeletion:
            // Arrange to skip all upcoming entries for this key since
            // they are hidden by this deletion.
-            saved_key_.SetKey(
+            saved_key_.SetUserKey(
                ikey.user_key,
                !iter_->IsKeyPinned() || !pin_thru_lifetime_ /* copy */);
            skipping = true;
            PERF_COUNTER_ADD(internal_delete_skipped_count, 1);
            break;
          case kTypeValue:
-            saved_key_.SetKey(
+            saved_key_.SetUserKey(
                ikey.user_key,
                !iter_->IsKeyPinned() || !pin_thru_lifetime_ /* copy */);
            if (range_del_agg_.ShouldDelete(
@@ -427,7 +438,7 @@ void DBIter::FindNextUserEntryInternal(bool skipping, bool prefix_check) {
            }
            break;
          case kTypeMerge:
-            saved_key_.SetKey(
+            saved_key_.SetUserKey(
                ikey.user_key,
                !iter_->IsKeyPinned() || !pin_thru_lifetime_ /* copy */);
            if (range_del_agg_.ShouldDelete(
@@ -458,12 +469,13 @@ void DBIter::FindNextUserEntryInternal(bool skipping, bool prefix_check) {

      // Here saved_key_ may contain some old key, or the default empty key, or
      // key assigned by some random other method. We don't care.
-      if (user_comparator_->Compare(ikey.user_key, saved_key_.GetKey()) <= 0) {
+      if (user_comparator_->Compare(ikey.user_key, saved_key_.GetUserKey()) <=
+          0) {
        num_skipped++;
      } else {
-        saved_key_.SetKey(
-                ikey.user_key,
-                !iter_->IsKeyPinned() || !pin_thru_lifetime_ /* copy */);
+        saved_key_.SetUserKey(
+            ikey.user_key,
+            !iter_->IsKeyPinned() || !pin_thru_lifetime_ /* copy */);
        skipping = false;
        num_skipped = 0;
      }
@@ -478,8 +490,8 @@ void DBIter::FindNextUserEntryInternal(bool skipping, bool prefix_check) {
        // We're looking for the next user-key but all we see are the same
        // user-key with decreasing sequence numbers. Fast forward to
        // sequence number 0 and type deletion (the smallest type).
-        AppendInternalKey(&last_key, ParsedInternalKey(saved_key_.GetKey(), 0,
-                                                       kTypeDeletion));
+        AppendInternalKey(&last_key, ParsedInternalKey(saved_key_.GetUserKey(),
+                                                       0, kTypeDeletion));
        // Don't set skipping = false because we may still see more user-keys
        // equal to saved_key_.
      } else {
@@ -489,7 +501,7 @@ void DBIter::FindNextUserEntryInternal(bool skipping, bool prefix_check) {
        // many times since our snapshot was taken, not the case when a lot of
        // different keys were inserted after our snapshot was taken.
        AppendInternalKey(&last_key,
-                          ParsedInternalKey(saved_key_.GetKey(), sequence_,
+                          ParsedInternalKey(saved_key_.GetUserKey(), sequence_,
                                            kValueTypeForSeek));
      }
      iter_->Seek(last_key);
@@ -509,8 +521,7 @@ void DBIter::FindNextUserEntryInternal(bool skipping, bool prefix_check) {
 //       iter_ points to the next entry (or invalid)
 void DBIter::MergeValuesNewToOld() {
  if (!merge_operator_) {
-    Log(InfoLogLevel::ERROR_LEVEL,
-        logger_, "Options::merge_operator is null.");
+    ROCKS_LOG_ERROR(logger_, "Options::merge_operator is null.");
    status_ = Status::InvalidArgument("merge_operator_ must be set.");
    valid_ = false;
    return;
@@ -531,7 +542,7 @@ void DBIter::MergeValuesNewToOld() {
      continue;
    }

-    if (!user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())) {
+    if (!user_comparator_->Equal(ikey.user_key, saved_key_.GetUserKey())) {
      // hit the next user key, stop right here
      break;
    } else if (kTypeDeletion == ikey.type || kTypeSingleDeletion == ikey.type ||
@@ -571,9 +582,10 @@ void DBIter::MergeValuesNewToOld() {
  // a deletion marker.
  // feed null as the existing value to the merge operator, such that
  // client can differentiate this scenario and do things accordingly.
-  s = MergeHelper::TimedFullMerge(merge_operator_, saved_key_.GetKey(), nullptr,
-                                  merge_context_.GetOperands(), &saved_value_,
-                                  logger_, statistics_, env_, &pinned_value_);
+  s = MergeHelper::TimedFullMerge(merge_operator_, saved_key_.GetUserKey(),
+                                  nullptr, merge_context_.GetOperands(),
+                                  &saved_value_, logger_, statistics_, env_,
+                                  &pinned_value_);
  if (!s.ok()) {
    status_ = s;
  }
@@ -582,6 +594,7 @@ void DBIter::MergeValuesNewToOld() {
 void DBIter::Prev() {
  assert(valid_);
  ReleaseTempPinnedData();
+  ResetInternalKeysSkippedCounter();
  if (direction_ == kForward) {
    ReverseToBackward();
  }
@@ -599,8 +612,8 @@ void DBIter::ReverseToForward() {
  if (prefix_extractor_ != nullptr && !total_order_seek_) {
    IterKey last_key;
    last_key.SetInternalKey(ParsedInternalKey(
-        saved_key_.GetKey(), kMaxSequenceNumber, kValueTypeForSeek));
-    iter_->Seek(last_key.GetKey());
+        saved_key_.GetUserKey(), kMaxSequenceNumber, kValueTypeForSeek));
+    iter_->Seek(last_key.GetInternalKey());
  }
  FindNextUserKey();
  direction_ = kForward;
@@ -613,9 +626,9 @@ void DBIter::ReverseToForward() {
 void DBIter::ReverseToBackward() {
  if (prefix_extractor_ != nullptr && !total_order_seek_) {
    IterKey last_key;
-    last_key.SetInternalKey(
-        ParsedInternalKey(saved_key_.GetKey(), 0, kValueTypeForSeekForPrev));
-    iter_->SeekForPrev(last_key.GetKey());
+    last_key.SetInternalKey(ParsedInternalKey(saved_key_.GetUserKey(), 0,
+                                              kValueTypeForSeekForPrev));
+    iter_->SeekForPrev(last_key.GetInternalKey());
  }
  if (current_entry_is_merged_) {
    // Not placed in the same key. Need to call Prev() until finding the
@@ -627,7 +640,8 @@ void DBIter::ReverseToBackward() {
    ParsedInternalKey ikey;
    FindParseableKey(&ikey, kReverse);
    while (iter_->Valid() &&
-           user_comparator_->Compare(ikey.user_key, saved_key_.GetKey()) > 0) {
+           user_comparator_->Compare(ikey.user_key, saved_key_.GetUserKey()) >
+               0) {
      if (ikey.sequence > sequence_) {
        PERF_COUNTER_ADD(internal_recent_skipped_count, 1);
      } else {
@@ -641,7 +655,8 @@ void DBIter::ReverseToBackward() {
  if (iter_->Valid()) {
    ParsedInternalKey ikey;
    assert(ParseKey(&ikey));
-    assert(user_comparator_->Compare(ikey.user_key, saved_key_.GetKey()) <= 0);
+    assert(user_comparator_->Compare(ikey.user_key, saved_key_.GetUserKey()) <=
+           0);
  }
 #endif

@@ -658,29 +673,36 @@ void DBIter::PrevInternal() {
  ParsedInternalKey ikey;

  while (iter_->Valid()) {
-    saved_key_.SetKey(ExtractUserKey(iter_->key()),
-                      !iter_->IsKeyPinned() || !pin_thru_lifetime_ /* copy */);
+    saved_key_.SetUserKey(
+        ExtractUserKey(iter_->key()),
+        !iter_->IsKeyPinned() || !pin_thru_lifetime_ /* copy */);
+
    if (FindValueForCurrentKey()) {
      valid_ = true;
      if (!iter_->Valid()) {
        return;
      }
      FindParseableKey(&ikey, kReverse);
-      if (user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())) {
+      if (user_comparator_->Equal(ikey.user_key, saved_key_.GetUserKey())) {
        FindPrevUserKey();
      }
      if (valid_ && prefix_extractor_ && prefix_same_as_start_ &&
-          prefix_extractor_->Transform(saved_key_.GetKey())
+          prefix_extractor_->Transform(saved_key_.GetUserKey())
                  .compare(prefix_start_key_) != 0) {
        valid_ = false;
      }
      return;
    }
+
+    if (TooManyInternalKeysSkipped(false)) {
+      return;
+    }
+
    if (!iter_->Valid()) {
      break;
    }
    FindParseableKey(&ikey, kReverse);
-    if (user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())) {
+    if (user_comparator_->Equal(ikey.user_key, saved_key_.GetUserKey())) {
      FindPrevUserKey();
    }
  }
@@ -710,7 +732,11 @@ bool DBIter::FindValueForCurrentKey() {
  TempPinData();
  size_t num_skipped = 0;
  while (iter_->Valid() && ikey.sequence <= sequence_ &&
-         user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())) {
+         user_comparator_->Equal(ikey.user_key, saved_key_.GetUserKey())) {
+    if (TooManyInternalKeysSkipped()) {
+      return false;
+    }
+
    // We iterate too much: let's use Seek() to avoid too much key comparisons
    if (num_skipped >= max_skip_) {
      return FindValueForCurrentKeyUsingSeek();
@@ -757,7 +783,7 @@ bool DBIter::FindValueForCurrentKey() {
    }

    PERF_COUNTER_ADD(internal_key_skipped_count, 1);
-    assert(user_comparator_->Equal(ikey.user_key, saved_key_.GetKey()));
+    assert(user_comparator_->Equal(ikey.user_key, saved_key_.GetUserKey()));
    iter_->Prev();
    ++num_skipped;
    FindParseableKey(&ikey, kReverse);
@@ -775,14 +801,14 @@ bool DBIter::FindValueForCurrentKey() {
      if (last_not_merge_type == kTypeDeletion ||
          last_not_merge_type == kTypeSingleDeletion ||
          last_not_merge_type == kTypeRangeDeletion) {
-        s = MergeHelper::TimedFullMerge(merge_operator_, saved_key_.GetKey(),
-                                        nullptr, merge_context_.GetOperands(),
-                                        &saved_value_, logger_, statistics_,
-                                        env_, &pinned_value_);
+        s = MergeHelper::TimedFullMerge(
+            merge_operator_, saved_key_.GetUserKey(), nullptr,
+            merge_context_.GetOperands(), &saved_value_, logger_, statistics_,
+            env_, &pinned_value_);
      } else {
        assert(last_not_merge_type == kTypeValue);
        s = MergeHelper::TimedFullMerge(
-            merge_operator_, saved_key_.GetKey(), &pinned_value_,
+            merge_operator_, saved_key_.GetUserKey(), &pinned_value_,
            merge_context_.GetOperands(), &saved_value_, logger_, statistics_,
            env_, &pinned_value_);
      }
@@ -808,8 +834,8 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
  // FindValueForCurrentKeyUsingSeek()
  assert(pinned_iters_mgr_.PinningEnabled());
  std::string last_key;
-  AppendInternalKey(&last_key, ParsedInternalKey(saved_key_.GetKey(), sequence_,
-                                                 kValueTypeForSeek));
+  AppendInternalKey(&last_key, ParsedInternalKey(saved_key_.GetUserKey(),
+                                                 sequence_, kValueTypeForSeek));
  iter_->Seek(last_key);
  RecordTick(statistics_, NUMBER_OF_RESEEKS_IN_ITERATION);

@@ -836,7 +862,7 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
  merge_context_.Clear();
  while (
      iter_->Valid() &&
-      user_comparator_->Equal(ikey.user_key, saved_key_.GetKey()) &&
+      user_comparator_->Equal(ikey.user_key, saved_key_.GetUserKey()) &&
      ikey.type == kTypeMerge &&
      !range_del_agg_.ShouldDelete(
          ikey, RangeDelAggregator::RangePositioningMode::kBackwardTraversal)) {
@@ -849,17 +875,17 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {

  Status s;
  if (!iter_->Valid() ||
-      !user_comparator_->Equal(ikey.user_key, saved_key_.GetKey()) ||
+      !user_comparator_->Equal(ikey.user_key, saved_key_.GetUserKey()) ||
      ikey.type == kTypeDeletion || ikey.type == kTypeSingleDeletion ||
      range_del_agg_.ShouldDelete(
          ikey, RangeDelAggregator::RangePositioningMode::kBackwardTraversal)) {
-    s = MergeHelper::TimedFullMerge(merge_operator_, saved_key_.GetKey(),
+    s = MergeHelper::TimedFullMerge(merge_operator_, saved_key_.GetUserKey(),
                                    nullptr, merge_context_.GetOperands(),
                                    &saved_value_, logger_, statistics_, env_,
                                    &pinned_value_);
    // Make iter_ valid and point to saved_key_
    if (!iter_->Valid() ||
-        !user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())) {
+        !user_comparator_->Equal(ikey.user_key, saved_key_.GetUserKey())) {
      iter_->Seek(last_key);
      RecordTick(statistics_, NUMBER_OF_RESEEKS_IN_ITERATION);
    }
@@ -871,9 +897,10 @@ bool DBIter::FindValueForCurrentKeyUsingSeek() {
  }

  const Slice& val = iter_->value();
-  s = MergeHelper::TimedFullMerge(merge_operator_, saved_key_.GetKey(), &val,
-                                  merge_context_.GetOperands(), &saved_value_,
-                                  logger_, statistics_, env_, &pinned_value_);
+  s = MergeHelper::TimedFullMerge(merge_operator_, saved_key_.GetUserKey(),
+                                  &val, merge_context_.GetOperands(),
+                                  &saved_value_, logger_, statistics_, env_,
+                                  &pinned_value_);
  valid_ = true;
  if (!s.ok()) {
    status_ = s;
@@ -892,7 +919,7 @@ void DBIter::FindNextUserKey() {
  ParsedInternalKey ikey;
  FindParseableKey(&ikey, kForward);
  while (iter_->Valid() &&
-         !user_comparator_->Equal(ikey.user_key, saved_key_.GetKey())) {
+         !user_comparator_->Equal(ikey.user_key, saved_key_.GetUserKey())) {
    iter_->Next();
    FindParseableKey(&ikey, kForward);
  }
@@ -907,16 +934,21 @@ void DBIter::FindPrevUserKey() {
  ParsedInternalKey ikey;
  FindParseableKey(&ikey, kReverse);
  int cmp;
-  while (iter_->Valid() && ((cmp = user_comparator_->Compare(
-                                 ikey.user_key, saved_key_.GetKey())) == 0 ||
-                            (cmp > 0 && ikey.sequence > sequence_))) {
+  while (iter_->Valid() &&
+         ((cmp = user_comparator_->Compare(ikey.user_key,
+                                           saved_key_.GetUserKey())) == 0 ||
+          (cmp > 0 && ikey.sequence > sequence_))) {
+    if (TooManyInternalKeysSkipped()) {
+      return;
+    }
+
    if (cmp == 0) {
      if (num_skipped >= max_skip_) {
        num_skipped = 0;
        IterKey last_key;
        last_key.SetInternalKey(ParsedInternalKey(
-            saved_key_.GetKey(), kMaxSequenceNumber, kValueTypeForSeek));
-        iter_->Seek(last_key.GetKey());
+            saved_key_.GetUserKey(), kMaxSequenceNumber, kValueTypeForSeek));
+        iter_->Seek(last_key.GetInternalKey());
        RecordTick(statistics_, NUMBER_OF_RESEEKS_IN_ITERATION);
      } else {
        ++num_skipped;
@@ -932,6 +964,18 @@ void DBIter::FindPrevUserKey() {
  }
 }

+bool DBIter::TooManyInternalKeysSkipped(bool increment) {
+  if ((max_skippable_internal_keys_ > 0) &&
+      (num_internal_keys_skipped_ > max_skippable_internal_keys_)) {
+    valid_ = false;
+    status_ = Status::Incomplete("Too many internal keys skipped.");
+    return true;
+  } else if (increment) {
+    num_internal_keys_skipped_++;
+  }
+  return false;
+}
+
 // Skip all unparseable keys
 void DBIter::FindParseableKey(ParsedInternalKey* ikey, Direction direction) {
  while (iter_->Valid() && !ParseKey(ikey)) {
@@ -946,12 +990,13 @@ void DBIter::FindParseableKey(ParsedInternalKey* ikey, Direction direction) {
 void DBIter::Seek(const Slice& target) {
  StopWatch sw(env_, statistics_, DB_SEEK);
  ReleaseTempPinnedData();
+  ResetInternalKeysSkippedCounter();
  saved_key_.Clear();
  saved_key_.SetInternalKey(target, sequence_);

  {
    PERF_TIMER_GUARD(seek_internal_seek_time);
-    iter_->Seek(saved_key_.GetKey());
+    iter_->Seek(saved_key_.GetInternalKey());
    range_del_agg_.InvalidateTombstoneMapPositions();
  }
  RecordTick(statistics_, NUMBER_DB_SEEK);
@@ -961,9 +1006,6 @@ void DBIter::Seek(const Slice& target) {
    }
    direction_ = kForward;
    ClearSavedValue();
-    // convert the InternalKey to UserKey in saved_key_ before
-    // passed to FindNextUserEntry
-    saved_key_.Reserve(saved_key_.Size() - 8);
    FindNextUserEntry(false /* not skipping */, prefix_same_as_start_);
    if (!valid_) {
      prefix_start_key_.clear();
@@ -979,14 +1021,15 @@ void DBIter::Seek(const Slice& target) {
  }

  if (valid_ && prefix_extractor_ && prefix_same_as_start_) {
-    prefix_start_buf_.SetKey(prefix_start_key_);
-    prefix_start_key_ = prefix_start_buf_.GetKey();
+    prefix_start_buf_.SetUserKey(prefix_start_key_);
+    prefix_start_key_ = prefix_start_buf_.GetUserKey();
  }
 }

 void DBIter::SeekForPrev(const Slice& target) {
  StopWatch sw(env_, statistics_, DB_SEEK);
  ReleaseTempPinnedData();
+  ResetInternalKeysSkippedCounter();
  saved_key_.Clear();
  // now saved_key is used to store internal key.
  saved_key_.SetInternalKey(target, 0 /* sequence_number */,
@@ -994,7 +1037,7 @@ void DBIter::SeekForPrev(const Slice& target) {

  {
    PERF_TIMER_GUARD(seek_internal_seek_time);
-    iter_->SeekForPrev(saved_key_.GetKey());
+    iter_->SeekForPrev(saved_key_.GetInternalKey());
    range_del_agg_.InvalidateTombstoneMapPositions();
  }

@@ -1019,8 +1062,8 @@ void DBIter::SeekForPrev(const Slice& target) {
    valid_ = false;
  }
  if (valid_ && prefix_extractor_ && prefix_same_as_start_) {
-    prefix_start_buf_.SetKey(prefix_start_key_);
-    prefix_start_key_ = prefix_start_buf_.GetKey();
+    prefix_start_buf_.SetUserKey(prefix_start_key_);
+    prefix_start_key_ = prefix_start_buf_.GetUserKey();
  }
 }

@@ -1032,6 +1075,7 @@ void DBIter::SeekToFirst() {
  }
  direction_ = kForward;
  ReleaseTempPinnedData();
+  ResetInternalKeysSkippedCounter();
  ClearSavedValue();

  {
@@ -1042,8 +1086,9 @@ void DBIter::SeekToFirst() {

  RecordTick(statistics_, NUMBER_DB_SEEK);
  if (iter_->Valid()) {
-    saved_key_.SetKey(ExtractUserKey(iter_->key()),
-                      !iter_->IsKeyPinned() || !pin_thru_lifetime_ /* copy */);
+    saved_key_.SetUserKey(
+        ExtractUserKey(iter_->key()),
+        !iter_->IsKeyPinned() || !pin_thru_lifetime_ /* copy */);
    FindNextUserEntry(false /* not skipping */, false /* no prefix check */);
    if (statistics_ != nullptr) {
      if (valid_) {
@@ -1055,8 +1100,9 @@ void DBIter::SeekToFirst() {
    valid_ = false;
  }
  if (valid_ && prefix_extractor_ && prefix_same_as_start_) {
-    prefix_start_buf_.SetKey(prefix_extractor_->Transform(saved_key_.GetKey()));
-    prefix_start_key_ = prefix_start_buf_.GetKey();
+    prefix_start_buf_.SetUserKey(
+        prefix_extractor_->Transform(saved_key_.GetUserKey()));
+    prefix_start_key_ = prefix_start_buf_.GetUserKey();
  }
 }

@@ -1068,6 +1114,7 @@ void DBIter::SeekToLast() {
  }
  direction_ = kReverse;
  ReleaseTempPinnedData();
+  ResetInternalKeysSkippedCounter();
  ClearSavedValue();

  {
@@ -1097,21 +1144,22 @@ void DBIter::SeekToLast() {
    }
  }
  if (valid_ && prefix_extractor_ && prefix_same_as_start_) {
-    prefix_start_buf_.SetKey(prefix_extractor_->Transform(saved_key_.GetKey()));
-    prefix_start_key_ = prefix_start_buf_.GetKey();
+    prefix_start_buf_.SetUserKey(
+        prefix_extractor_->Transform(saved_key_.GetUserKey()));
+    prefix_start_key_ = prefix_start_buf_.GetUserKey();
  }
 }

-Iterator* NewDBIterator(
-    Env* env, const ImmutableCFOptions& ioptions,
-    const Comparator* user_key_comparator, InternalIterator* internal_iter,
-    const SequenceNumber& sequence, uint64_t max_sequential_skip_in_iterations,
-    uint64_t version_number, const Slice* iterate_upper_bound,
-    bool prefix_same_as_start, bool pin_data, bool total_order_seek) {
+Iterator* NewDBIterator(Env* env, const ReadOptions& read_options,
+                        const ImmutableCFOptions& cf_options,
+                        const Comparator* user_key_comparator,
+                        InternalIterator* internal_iter,
+                        const SequenceNumber& sequence,
+                        uint64_t max_sequential_skip_in_iterations,
+                        uint64_t version_number) {
  DBIter* db_iter = new DBIter(
-      env, ioptions, user_key_comparator, internal_iter, sequence, false,
-      max_sequential_skip_in_iterations, version_number, iterate_upper_bound,
-      prefix_same_as_start, pin_data, total_order_seek);
+      env, read_options, cf_options, user_key_comparator, internal_iter,
+      sequence, false, max_sequential_skip_in_iterations, version_number);
  return db_iter;
 }

@@ -1151,18 +1199,16 @@ void ArenaWrappedDBIter::RegisterCleanup(CleanupFunction function, void* arg1,
 }

 ArenaWrappedDBIter* NewArenaWrappedDbIterator(
-    Env* env, const ImmutableCFOptions& ioptions,
-    const Comparator* user_key_comparator, const SequenceNumber& sequence,
-    uint64_t max_sequential_skip_in_iterations, uint64_t version_number,
-    const Slice* iterate_upper_bound, bool prefix_same_as_start, bool pin_data,
-    bool total_order_seek) {
+    Env* env, const ReadOptions& read_options,
+    const ImmutableCFOptions& cf_options, const Comparator* user_key_comparator,
+    const SequenceNumber& sequence, uint64_t max_sequential_skip_in_iterations,
+    uint64_t version_number) {
  ArenaWrappedDBIter* iter = new ArenaWrappedDBIter();
  Arena* arena = iter->GetArena();
  auto mem = arena->AllocateAligned(sizeof(DBIter));
-  DBIter* db_iter = new (mem) DBIter(
-      env, ioptions, user_key_comparator, nullptr, sequence, true,
-      max_sequential_skip_in_iterations, version_number, iterate_upper_bound,
-      prefix_same_as_start, pin_data, total_order_seek);
+  DBIter* db_iter = new (mem)
+      DBIter(env, read_options, cf_options, user_key_comparator, nullptr,
+             sequence, true, max_sequential_skip_in_iterations, version_number);

  iter->SetDBIter(db_iter);

@@ -12,11 +12,11 @@
 #include <string>
 #include "db/dbformat.h"
 #include "db/range_del_aggregator.h"
+#include "options/cf_options.h"
 #include "rocksdb/db.h"
 #include "rocksdb/iterator.h"
 #include "util/arena.h"
 #include "util/autovector.h"
-#include "util/cf_options.h"

 namespace rocksdb {

@@ -27,13 +27,13 @@ class InternalIterator;
 // Return a new iterator that converts internal keys (yielded by
 // "*internal_iter") that were live at the specified "sequence" number
 // into appropriate user keys.
-extern Iterator* NewDBIterator(
-    Env* env, const ImmutableCFOptions& options,
-    const Comparator* user_key_comparator, InternalIterator* internal_iter,
-    const SequenceNumber& sequence, uint64_t max_sequential_skip_in_iterations,
-    uint64_t version_number, const Slice* iterate_upper_bound = nullptr,
-    bool prefix_same_as_start = false, bool pin_data = false,
-    bool total_order_seek = false);
+extern Iterator* NewDBIterator(Env* env, const ReadOptions& read_options,
+                               const ImmutableCFOptions& cf_options,
+                               const Comparator* user_key_comparator,
+                               InternalIterator* internal_iter,
+                               const SequenceNumber& sequence,
+                               uint64_t max_sequential_skip_in_iterations,
+                               uint64_t version_number);

 // A wrapper iterator which wraps DB Iterator and the arena, with which the DB
 // iterator is supposed be allocated. This class is used as an entry point of
@@ -77,11 +77,9 @@ class ArenaWrappedDBIter : public Iterator {

 // Generate the arena wrapped iterator class.
 extern ArenaWrappedDBIter* NewArenaWrappedDbIterator(
-    Env* env, const ImmutableCFOptions& options,
-    const Comparator* user_key_comparator, const SequenceNumber& sequence,
-    uint64_t max_sequential_skip_in_iterations, uint64_t version_number,
-    const Slice* iterate_upper_bound = nullptr,
-    bool prefix_same_as_start = false, bool pin_data = false,
-    bool total_order_seek = false);
+    Env* env, const ReadOptions& read_options,
+    const ImmutableCFOptions& cf_options, const Comparator* user_key_comparator,
+    const SequenceNumber& sequence, uint64_t max_sequential_skip_in_iterations,
+    uint64_t version_number);

 }  // namespace rocksdb
@@ -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();
@@ -975,12 +1003,57 @@ TEST_F(DBIteratorTest, DBIteratorBoundTest) {
  }
 }

+TEST_F(DBIteratorTest, DBIteratorBoundOptimizationTest) {
+  int upper_bound_hits = 0;
+  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 options = CurrentOptions();
+  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) {
  Options options;
  options.create_if_missing = true;
  options.merge_operator = MergeOperators::CreatePutOperator();
+  options.env = env_;
  DestroyAndReopen(options);

  // write three entries with different keys using Merge()
@@ -123,6 +123,7 @@ TEST_F(DBMemTableTest, InsertWithHint) {
  options.memtable_factory.reset(new MockMemTableRepFactory());
  options.memtable_insert_with_hint_prefix_extractor.reset(
      new TestPrefixExtractor());
+  options.env = env_;
  Reopen(options);
  MockMemTableRep* rep =
      reinterpret_cast<MockMemTableRepFactory*>(options.memtable_factory.get())
@@ -44,6 +44,7 @@ TEST_F(DBMergeOperatorTest, MergeErrorOnRead) {
  Options options;
  options.create_if_missing = true;
  options.merge_operator.reset(new TestPutOperator());
+  options.env = env_;
  Reopen(options);
  ASSERT_OK(Merge("k1", "v1"));
  ASSERT_OK(Merge("k1", "corrupted"));
@@ -57,6 +58,7 @@ TEST_F(DBMergeOperatorTest, MergeErrorOnWrite) {
  options.create_if_missing = true;
  options.merge_operator.reset(new TestPutOperator());
  options.max_successive_merges = 3;
+  options.env = env_;
  Reopen(options);
  ASSERT_OK(Merge("k1", "v1"));
  ASSERT_OK(Merge("k1", "v2"));
@@ -71,6 +73,7 @@ TEST_F(DBMergeOperatorTest, MergeErrorOnIteration) {
  Options options;
  options.create_if_missing = true;
  options.merge_operator.reset(new TestPutOperator());
+  options.env = env_;

  DestroyAndReopen(options);
  ASSERT_OK(Merge("k1", "v1"));
@@ -89,6 +92,294 @@ TEST_F(DBMergeOperatorTest, MergeErrorOnIteration) {
  VerifyDBInternal({{"k1", "v1"}, {"k2", "corrupted"}, {"k2", "v2"}});
 }

+
+class MergeOperatorPinningTest : public DBMergeOperatorTest,
+                                 public testing::WithParamInterface<bool> {
+ public:
+  MergeOperatorPinningTest() { disable_block_cache_ = GetParam(); }
+
+  bool disable_block_cache_;
+};
+
+INSTANTIATE_TEST_CASE_P(MergeOperatorPinningTest, MergeOperatorPinningTest,
+                        ::testing::Bool());
+
+#ifndef ROCKSDB_LITE
+TEST_P(MergeOperatorPinningTest, OperandsMultiBlocks) {
+  Options options = CurrentOptions();
+  BlockBasedTableOptions table_options;
+  table_options.block_size = 1;  // every block will contain one entry
+  table_options.no_block_cache = disable_block_cache_;
+  options.table_factory.reset(NewBlockBasedTableFactory(table_options));
+  options.merge_operator = MergeOperators::CreateStringAppendTESTOperator();
+  options.level0_slowdown_writes_trigger = (1 << 30);
+  options.level0_stop_writes_trigger = (1 << 30);
+  options.disable_auto_compactions = true;
+  DestroyAndReopen(options);
+
+  const int kKeysPerFile = 10;
+  const int kOperandsPerKeyPerFile = 7;
+  const int kOperandSize = 100;
+  // Filse to write in L0 before compacting to lower level
+  const int kFilesPerLevel = 3;
+
+  Random rnd(301);
+  std::map<std::string, std::string> true_data;
+  int batch_num = 1;
+  int lvl_to_fill = 4;
+  int key_id = 0;
+  while (true) {
+    for (int j = 0; j < kKeysPerFile; j++) {
+      std::string key = Key(key_id % 35);
+      key_id++;
+      for (int k = 0; k < kOperandsPerKeyPerFile; k++) {
+        std::string val = RandomString(&rnd, kOperandSize);
+        ASSERT_OK(db_->Merge(WriteOptions(), key, val));
+        if (true_data[key].size() == 0) {
+          true_data[key] = val;
+        } else {
+          true_data[key] += "," + val;
+        }
+      }
+    }
+
+    if (lvl_to_fill == -1) {
+      // Keep last batch in memtable and stop
+      break;
+    }
+
+    ASSERT_OK(Flush());
+    if (batch_num % kFilesPerLevel == 0) {
+      if (lvl_to_fill != 0) {
+        MoveFilesToLevel(lvl_to_fill);
+      }
+      lvl_to_fill--;
+    }
+    batch_num++;
+  }
+
+  // 3 L0 files
+  // 1 L1 file
+  // 3 L2 files
+  // 1 L3 file
+  // 3 L4 Files
+  ASSERT_EQ(FilesPerLevel(), "3,1,3,1,3");
+
+  VerifyDBFromMap(true_data);
+}
+
+TEST_P(MergeOperatorPinningTest, Randomized) {
+  do {
+    Options options = CurrentOptions();
+    options.merge_operator = MergeOperators::CreateMaxOperator();
+    BlockBasedTableOptions table_options;
+    table_options.no_block_cache = disable_block_cache_;
+    options.table_factory.reset(NewBlockBasedTableFactory(table_options));
+    DestroyAndReopen(options);
+
+    Random rnd(301);
+    std::map<std::string, std::string> true_data;
+
+    const int kTotalMerges = 10000;
+    // Every key gets ~10 operands
+    const int kKeyRange = kTotalMerges / 10;
+    const int kOperandSize = 20;
+    const int kNumPutBefore = kKeyRange / 10;  // 10% value
+    const int kNumPutAfter = kKeyRange / 10;   // 10% overwrite
+    const int kNumDelete = kKeyRange / 10;     // 10% delete
+
+    // kNumPutBefore keys will have base values
+    for (int i = 0; i < kNumPutBefore; i++) {
+      std::string key = Key(rnd.Next() % kKeyRange);
+      std::string value = RandomString(&rnd, kOperandSize);
+      ASSERT_OK(db_->Put(WriteOptions(), key, value));
+
+      true_data[key] = value;
+    }
+
+    // Do kTotalMerges merges
+    for (int i = 0; i < kTotalMerges; i++) {
+      std::string key = Key(rnd.Next() % kKeyRange);
+      std::string value = RandomString(&rnd, kOperandSize);
+      ASSERT_OK(db_->Merge(WriteOptions(), key, value));
+
+      if (true_data[key] < value) {
+        true_data[key] = value;
+      }
+    }
+
+    // Overwrite random kNumPutAfter keys
+    for (int i = 0; i < kNumPutAfter; i++) {
+      std::string key = Key(rnd.Next() % kKeyRange);
+      std::string value = RandomString(&rnd, kOperandSize);
+      ASSERT_OK(db_->Put(WriteOptions(), key, value));
+
+      true_data[key] = value;
+    }
+
+    // Delete random kNumDelete keys
+    for (int i = 0; i < kNumDelete; i++) {
+      std::string key = Key(rnd.Next() % kKeyRange);
+      ASSERT_OK(db_->Delete(WriteOptions(), key));
+
+      true_data.erase(key);
+    }
+
+    VerifyDBFromMap(true_data);
+
+    // Skip HashCuckoo since it does not support merge operators
+  } while (ChangeOptions(kSkipMergePut | kSkipHashCuckoo));
+}
+
+class MergeOperatorHook : public MergeOperator {
+ public:
+  explicit MergeOperatorHook(std::shared_ptr<MergeOperator> _merge_op)
+      : merge_op_(_merge_op) {}
+
+  virtual bool FullMergeV2(const MergeOperationInput& merge_in,
+                           MergeOperationOutput* merge_out) const override {
+    before_merge_();
+    bool res = merge_op_->FullMergeV2(merge_in, merge_out);
+    after_merge_();
+    return res;
+  }
+
+  virtual const char* Name() const override { return merge_op_->Name(); }
+
+  std::shared_ptr<MergeOperator> merge_op_;
+  std::function<void()> before_merge_ = []() {};
+  std::function<void()> after_merge_ = []() {};
+};
+
+TEST_P(MergeOperatorPinningTest, EvictCacheBeforeMerge) {
+  Options options = CurrentOptions();
+
+  auto merge_hook =
+      std::make_shared<MergeOperatorHook>(MergeOperators::CreateMaxOperator());
+  options.merge_operator = merge_hook;
+  options.disable_auto_compactions = true;
+  options.level0_slowdown_writes_trigger = (1 << 30);
+  options.level0_stop_writes_trigger = (1 << 30);
+  options.max_open_files = 20;
+  BlockBasedTableOptions bbto;
+  bbto.no_block_cache = disable_block_cache_;
+  if (bbto.no_block_cache == false) {
+    bbto.block_cache = NewLRUCache(64 * 1024 * 1024);
+  } else {
+    bbto.block_cache = nullptr;
+  }
+  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+  DestroyAndReopen(options);
+
+  const int kNumOperands = 30;
+  const int kNumKeys = 1000;
+  const int kOperandSize = 100;
+  Random rnd(301);
+
+  // 1000 keys every key have 30 operands, every operand is in a different file
+  std::map<std::string, std::string> true_data;
+  for (int i = 0; i < kNumOperands; i++) {
+    for (int j = 0; j < kNumKeys; j++) {
+      std::string k = Key(j);
+      std::string v = RandomString(&rnd, kOperandSize);
+      ASSERT_OK(db_->Merge(WriteOptions(), k, v));
+
+      true_data[k] = std::max(true_data[k], v);
+    }
+    ASSERT_OK(Flush());
+  }
+
+  std::vector<uint64_t> file_numbers = ListTableFiles(env_, dbname_);
+  ASSERT_EQ(file_numbers.size(), kNumOperands);
+  int merge_cnt = 0;
+
+  // Code executed before merge operation
+  merge_hook->before_merge_ = [&]() {
+    // Evict all tables from cache before every merge operation
+    for (uint64_t num : file_numbers) {
+      TableCache::Evict(dbfull()->TEST_table_cache(), num);
+    }
+    // Decrease cache capacity to force all unrefed blocks to be evicted
+    if (bbto.block_cache) {
+      bbto.block_cache->SetCapacity(1);
+    }
+    merge_cnt++;
+  };
+
+  // Code executed after merge operation
+  merge_hook->after_merge_ = [&]() {
+    // Increase capacity again after doing the merge
+    if (bbto.block_cache) {
+      bbto.block_cache->SetCapacity(64 * 1024 * 1024);
+    }
+  };
+
+  size_t total_reads;
+  VerifyDBFromMap(true_data, &total_reads);
+  ASSERT_EQ(merge_cnt, total_reads);
+
+  db_->CompactRange(CompactRangeOptions(), nullptr, nullptr);
+
+  VerifyDBFromMap(true_data, &total_reads);
+}
+
+TEST_P(MergeOperatorPinningTest, TailingIterator) {
+  Options options = CurrentOptions();
+  options.merge_operator = MergeOperators::CreateMaxOperator();
+  BlockBasedTableOptions bbto;
+  bbto.no_block_cache = disable_block_cache_;
+  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+  DestroyAndReopen(options);
+
+  const int kNumOperands = 100;
+  const int kNumWrites = 100000;
+
+  std::function<void()> writer_func = [&]() {
+    int k = 0;
+    for (int i = 0; i < kNumWrites; i++) {
+      db_->Merge(WriteOptions(), Key(k), Key(k));
+
+      if (i && i % kNumOperands == 0) {
+        k++;
+      }
+      if (i && i % 127 == 0) {
+        ASSERT_OK(Flush());
+      }
+      if (i && i % 317 == 0) {
+        ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+      }
+    }
+  };
+
+  std::function<void()> reader_func = [&]() {
+    ReadOptions ro;
+    ro.tailing = true;
+    Iterator* iter = db_->NewIterator(ro);
+
+    iter->SeekToFirst();
+    for (int i = 0; i < (kNumWrites / kNumOperands); i++) {
+      while (!iter->Valid()) {
+        // wait for the key to be written
+        env_->SleepForMicroseconds(100);
+        iter->Seek(Key(i));
+      }
+      ASSERT_EQ(iter->key(), Key(i));
+      ASSERT_EQ(iter->value(), Key(i));
+
+      iter->Next();
+    }
+
+    delete iter;
+  };
+
+  rocksdb::port::Thread writer_thread(writer_func);
+  rocksdb::port::Thread reader_thread(reader_func);
+
+  writer_thread.join();
+  reader_thread.join();
+}
+#endif  // ROCKSDB_LITE
+
 }  // namespace rocksdb

 int main(int argc, char** argv) {
@@ -13,9 +13,9 @@
 #include "db/column_family.h"
 #include "db/db_impl.h"
 #include "db/db_test_util.h"
+#include "options/options_helper.h"
 #include "port/stack_trace.h"
 #include "rocksdb/convenience.h"
-#include "util/options_helper.h"
 #include "util/random.h"
 #include "util/sync_point.h"
 #include "util/testutil.h"
@@ -62,6 +62,7 @@ class DBOptionsTest : public DBTestBase {
  std::unordered_map<std::string, std::string> GetRandomizedMutableCFOptionsMap(
      Random* rnd) {
    Options options;
+    options.env = env_;
    ImmutableDBOptions db_options(options);
    test::RandomInitCFOptions(&options, rnd);
    auto sanitized_options = SanitizeOptions(db_options, options);
@@ -87,6 +88,7 @@ TEST_F(DBOptionsTest, GetLatestDBOptions) {
  // GetOptions should be able to get latest option changed by SetOptions.
  Options options;
  options.create_if_missing = true;
+  options.env = env_;
  Random rnd(228);
  Reopen(options);
  auto new_options = GetRandomizedMutableDBOptionsMap(&rnd);
@@ -98,6 +100,7 @@ TEST_F(DBOptionsTest, GetLatestCFOptions) {
  // GetOptions should be able to get latest option changed by SetOptions.
  Options options;
  options.create_if_missing = true;
+  options.env = env_;
  Random rnd(228);
  Reopen(options);
  CreateColumnFamilies({"foo"}, options);
@@ -118,6 +121,7 @@ TEST_F(DBOptionsTest, SetOptionsAndReopen) {
  ASSERT_OK(dbfull()->SetOptions(rand_opts));
  // Verify if DB can be reopen after setting options.
  Options options;
+  options.env = env_;
  ASSERT_OK(TryReopen(options));
 }

@@ -137,6 +141,7 @@ TEST_F(DBOptionsTest, EnableAutoCompactionAndTriggerStall) {
          std::numeric_limits<uint64_t>::max();
      options.soft_pending_compaction_bytes_limit =
          std::numeric_limits<uint64_t>::max();
+      options.env = env_;

      DestroyAndReopen(options);
      int i = 0;
@@ -228,6 +233,7 @@ TEST_F(DBOptionsTest, SetOptionsMayTriggerCompaction) {
  Options options;
  options.create_if_missing = true;
  options.level0_file_num_compaction_trigger = 1000;
+  options.env = env_;
  Reopen(options);
  for (int i = 0; i < 3; i++) {
    // Need to insert two keys to avoid trivial move.
@@ -247,6 +253,7 @@ TEST_F(DBOptionsTest, SetBackgroundCompactionThreads) {
  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"},
@@ -260,6 +267,7 @@ TEST_F(DBOptionsTest, AvoidFlushDuringShutdown) {
  Options options;
  options.create_if_missing = true;
  options.disable_auto_compactions = true;
+  options.env = env_;
  WriteOptions write_without_wal;
  write_without_wal.disableWAL = true;

@@ -282,6 +290,7 @@ TEST_F(DBOptionsTest, SetDelayedWriteRateOption) {
  Options options;
  options.create_if_missing = true;
  options.delayed_write_rate = 2 * 1024U * 1024U;
+  options.env = env_;
  Reopen(options);
  ASSERT_EQ(2 * 1024U * 1024U, dbfull()->TEST_write_controler().max_delayed_write_rate());

@@ -297,6 +306,7 @@ TEST_F(DBOptionsTest, MaxTotalWalSizeChange) {

  Options options;
  options.create_if_missing = true;
+  options.env = env_;
  CreateColumnFamilies({"1", "2", "3"}, options);
  ReopenWithColumnFamilies({"default", "1", "2", "3"}, options);

@@ -316,6 +326,22 @@ TEST_F(DBOptionsTest, MaxTotalWalSizeChange) {
  }
 }

+TEST_F(DBOptionsTest, SetStatsDumpPeriodSec) {
+  Options options;
+  options.create_if_missing = true;
+  options.stats_dump_period_sec = 5;
+  options.env = env_;
+  Reopen(options);
+  ASSERT_EQ(5, dbfull()->GetDBOptions().stats_dump_period_sec);
+
+  for (int i = 0; i < 20; i++) {
+    int num = rand() % 5000 + 1;
+    ASSERT_OK(dbfull()->SetDBOptions(
+        {{"stats_dump_period_sec", std::to_string(num)}}));
+    ASSERT_EQ(num, dbfull()->GetDBOptions().stats_dump_period_sec);
+  }
+}
+
 static void assert_candidate_files_empty(DBImpl* dbfull, const bool empty) {
  dbfull->TEST_LockMutex();
  JobContext job_context(0);
@@ -326,7 +352,7 @@ static void assert_candidate_files_empty(DBImpl* dbfull, const bool empty) {
 }

 TEST_F(DBOptionsTest, DeleteObsoleteFilesPeriodChange) {
-  SpecialEnv env(Env::Default());
+  SpecialEnv env(env_);
  env.time_elapse_only_sleep_ = true;
  Options options;
  options.env = &env;
@@ -342,7 +342,7 @@ TEST_F(DBPropertiesTest, ReadLatencyHistogramByLevel) {
  BlockBasedTableOptions table_options;
  table_options.no_block_cache = true;

-  DestroyAndReopen(options);
+  CreateAndReopenWithCF({"pikachu"}, options);
  int key_index = 0;
  Random rnd(301);
  for (int num = 0; num < 8; num++) {
@@ -359,25 +359,26 @@ TEST_F(DBPropertiesTest, ReadLatencyHistogramByLevel) {
  for (int key = 0; key < key_index; key++) {
    Get(Key(key));
  }
-  ASSERT_TRUE(dbfull()->GetProperty("rocksdb.dbstats", &prop));
+  ASSERT_TRUE(dbfull()->GetProperty("rocksdb.cfstats", &prop));
  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"));

  // Reopen and issue Get(). See thee latency tracked
-  Reopen(options);
+  ReopenWithColumnFamilies({"default", "pikachu"}, options);
  dbfull()->TEST_WaitForCompact();
  for (int key = 0; key < key_index; key++) {
    Get(Key(key));
  }
-  ASSERT_TRUE(dbfull()->GetProperty("rocksdb.dbstats", &prop));
+  ASSERT_TRUE(dbfull()->GetProperty(dbfull()->DefaultColumnFamily(),
+                                    "rocksdb.cf-file-histogram", &prop));
  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"));

  // Reopen and issue iterating. See thee latency tracked
-  Reopen(options);
-  ASSERT_TRUE(dbfull()->GetProperty("rocksdb.dbstats", &prop));
+  ReopenWithColumnFamilies({"default", "pikachu"}, options);
+  ASSERT_TRUE(dbfull()->GetProperty("rocksdb.cf-file-histogram", &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"));
@@ -386,25 +387,51 @@ TEST_F(DBPropertiesTest, ReadLatencyHistogramByLevel) {
    for (iter->Seek(Key(0)); iter->Valid(); iter->Next()) {
    }
  }
-  ASSERT_TRUE(dbfull()->GetProperty("rocksdb.dbstats", &prop));
+  ASSERT_TRUE(dbfull()->GetProperty("rocksdb.cf-file-histogram", &prop));
  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"));

+  // CF 1 should show no histogram.
+  ASSERT_TRUE(
+      dbfull()->GetProperty(handles_[1], "rocksdb.cf-file-histogram", &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"));
+  // put something and read it back , CF 1 should show histogram.
+  Put(1, "foo", "bar");
+  Flush(1);
+  dbfull()->TEST_WaitForCompact();
+  ASSERT_EQ("bar", Get(1, "foo"));
+
+  ASSERT_TRUE(
+      dbfull()->GetProperty(handles_[1], "rocksdb.cf-file-histogram", &prop));
+  ASSERT_NE(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"));
+
  // options.max_open_files preloads table readers.
  options.max_open_files = -1;
-  Reopen(options);
-  ASSERT_TRUE(dbfull()->GetProperty("rocksdb.dbstats", &prop));
+  ReopenWithColumnFamilies({"default", "pikachu"}, options);
+  ASSERT_TRUE(dbfull()->GetProperty(dbfull()->DefaultColumnFamily(),
+                                    "rocksdb.cf-file-histogram", &prop));
  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"));
  for (int key = 0; key < key_index; key++) {
    Get(Key(key));
  }
-  ASSERT_TRUE(dbfull()->GetProperty("rocksdb.dbstats", &prop));
+  ASSERT_TRUE(dbfull()->GetProperty("rocksdb.cfstats", &prop));
  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) {
@@ -1156,6 +1183,7 @@ TEST_F(DBPropertiesTest, TablePropertiesNeedCompactTest) {
  options.max_bytes_for_level_multiplier = 4;
  options.soft_pending_compaction_bytes_limit = 1024 * 1024;
  options.num_levels = 8;
+  options.env = env_;

  std::shared_ptr<TablePropertiesCollectorFactory> collector_factory =
      std::make_shared<CountingDeleteTabPropCollectorFactory>();
@@ -1227,6 +1255,7 @@ TEST_F(DBPropertiesTest, NeedCompactHintPersistentTest) {
  options.level0_slowdown_writes_trigger = 10;
  options.level0_stop_writes_trigger = 10;
  options.disable_auto_compactions = true;
+  options.env = env_;

  std::shared_ptr<TablePropertiesCollectorFactory> collector_factory =
      std::make_shared<CountingDeleteTabPropCollectorFactory>();
@@ -1271,6 +1300,18 @@ TEST_F(DBPropertiesTest, NeedCompactHintPersistentTest) {
    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

@@ -324,6 +324,7 @@ TEST_F(DBRangeDelTest, CompactionRemovesCoveredKeys) {
 TEST_F(DBRangeDelTest, ValidLevelSubcompactionBoundaries) {
  const int kNumPerFile = 100, kNumFiles = 4, kFileBytes = 100 << 10;
  Options options = CurrentOptions();
+  options.disable_auto_compactions = true;
  options.level0_file_num_compaction_trigger = kNumFiles;
  options.max_bytes_for_level_base = 2 * kFileBytes;
  options.max_subcompactions = 4;
@@ -361,7 +362,14 @@ TEST_F(DBRangeDelTest, ValidLevelSubcompactionBoundaries) {
        // new L1 files must be generated with non-overlapping key ranges even
        // though multiple subcompactions see the same ranges deleted, else an
        // assertion will fail.
+        //
+        // Only enable auto-compactions when we're ready; otherwise, the
+        // oversized L0 (relative to base_level) causes the compaction to run
+        // earlier.
+        ASSERT_OK(db_->EnableAutoCompaction({db_->DefaultColumnFamily()}));
        dbfull()->TEST_WaitForCompact();
+        ASSERT_OK(db_->SetOptions(db_->DefaultColumnFamily(),
+                                  {{"disable_auto_compactions", "true"}}));
        ASSERT_EQ(NumTableFilesAtLevel(0), 0);
        ASSERT_GT(NumTableFilesAtLevel(1), 0);
        ASSERT_GT(NumTableFilesAtLevel(2), 0);
@@ -210,7 +210,9 @@ TEST_F(DBSSTTest, DeleteObsoleteFilesPendingOutputs) {
  blocking_thread.WakeUp();
  blocking_thread.WaitUntilDone();
  dbfull()->TEST_WaitForFlushMemTable();
-  ASSERT_EQ("1,0,0,0,1", FilesPerLevel(0));
+  // File just flushed is too big for L0 and L1 so gets moved to L2.
+  dbfull()->TEST_WaitForCompact();
+  ASSERT_EQ("0,0,1,0,1", FilesPerLevel(0));

  metadata.clear();
  db_->GetLiveFilesMetaData(&metadata);
@@ -0,0 +1,149 @@
+//  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.
+
+#include <string>
+
+#include "db/db_test_util.h"
+#include "monitoring/thread_status_util.h"
+#include "port/stack_trace.h"
+#include "rocksdb/statistics.h"
+
+namespace rocksdb {
+
+class DBStatisticsTest : public DBTestBase {
+ public:
+  DBStatisticsTest() : DBTestBase("/db_statistics_test") {}
+};
+
+TEST_F(DBStatisticsTest, CompressionStatsTest) {
+  CompressionType type;
+
+  if (Snappy_Supported()) {
+    type = kSnappyCompression;
+    fprintf(stderr, "using snappy\n");
+  } else if (Zlib_Supported()) {
+    type = kZlibCompression;
+    fprintf(stderr, "using zlib\n");
+  } else if (BZip2_Supported()) {
+    type = kBZip2Compression;
+    fprintf(stderr, "using bzip2\n");
+  } else if (LZ4_Supported()) {
+    type = kLZ4Compression;
+    fprintf(stderr, "using lz4\n");
+  } else if (XPRESS_Supported()) {
+    type = kXpressCompression;
+    fprintf(stderr, "using xpress\n");
+  } else if (ZSTD_Supported()) {
+    type = kZSTD;
+    fprintf(stderr, "using ZSTD\n");
+  } else {
+    fprintf(stderr, "skipping test, compression disabled\n");
+    return;
+  }
+
+  Options options = CurrentOptions();
+  options.compression = type;
+  options.statistics = rocksdb::CreateDBStatistics();
+  options.statistics->stats_level_ = StatsLevel::kExceptTimeForMutex;
+  DestroyAndReopen(options);
+
+  int kNumKeysWritten = 100000;
+
+  // Check that compressions occur and are counted when compression is turned on
+  Random rnd(301);
+  for (int i = 0; i < kNumKeysWritten; ++i) {
+    // compressible string
+    ASSERT_OK(Put(Key(i), RandomString(&rnd, 128) + std::string(128, 'a')));
+  }
+  ASSERT_OK(Flush());
+  ASSERT_GT(options.statistics->getTickerCount(NUMBER_BLOCK_COMPRESSED), 0);
+
+  for (int i = 0; i < kNumKeysWritten; ++i) {
+    auto r = Get(Key(i));
+  }
+  ASSERT_GT(options.statistics->getTickerCount(NUMBER_BLOCK_DECOMPRESSED), 0);
+
+  options.compression = kNoCompression;
+  DestroyAndReopen(options);
+  uint64_t currentCompressions =
+            options.statistics->getTickerCount(NUMBER_BLOCK_COMPRESSED);
+  uint64_t currentDecompressions =
+            options.statistics->getTickerCount(NUMBER_BLOCK_DECOMPRESSED);
+
+  // Check that compressions do not occur when turned off
+  for (int i = 0; i < kNumKeysWritten; ++i) {
+    // compressible string
+    ASSERT_OK(Put(Key(i), RandomString(&rnd, 128) + std::string(128, 'a')));
+  }
+  ASSERT_OK(Flush());
+  ASSERT_EQ(options.statistics->getTickerCount(NUMBER_BLOCK_COMPRESSED)
+            - currentCompressions, 0);
+
+  for (int i = 0; i < kNumKeysWritten; ++i) {
+    auto r = Get(Key(i));
+  }
+  ASSERT_EQ(options.statistics->getTickerCount(NUMBER_BLOCK_DECOMPRESSED)
+            - currentDecompressions, 0);
+}
+
+TEST_F(DBStatisticsTest, MutexWaitStatsDisabledByDefault) {
+  Options options = CurrentOptions();
+  options.create_if_missing = true;
+  options.statistics = rocksdb::CreateDBStatistics();
+  CreateAndReopenWithCF({"pikachu"}, options);
+  const uint64_t kMutexWaitDelay = 100;
+  ThreadStatusUtil::TEST_SetStateDelay(ThreadStatus::STATE_MUTEX_WAIT,
+                                       kMutexWaitDelay);
+  ASSERT_OK(Put("hello", "rocksdb"));
+  ASSERT_EQ(TestGetTickerCount(options, DB_MUTEX_WAIT_MICROS), 0);
+  ThreadStatusUtil::TEST_SetStateDelay(ThreadStatus::STATE_MUTEX_WAIT, 0);
+}
+
+TEST_F(DBStatisticsTest, MutexWaitStats) {
+  Options options = CurrentOptions();
+  options.create_if_missing = true;
+  options.statistics = rocksdb::CreateDBStatistics();
+  options.statistics->stats_level_ = StatsLevel::kAll;
+  CreateAndReopenWithCF({"pikachu"}, options);
+  const uint64_t kMutexWaitDelay = 100;
+  ThreadStatusUtil::TEST_SetStateDelay(ThreadStatus::STATE_MUTEX_WAIT,
+                                       kMutexWaitDelay);
+  ASSERT_OK(Put("hello", "rocksdb"));
+  ASSERT_GE(TestGetTickerCount(options, DB_MUTEX_WAIT_MICROS), kMutexWaitDelay);
+  ThreadStatusUtil::TEST_SetStateDelay(ThreadStatus::STATE_MUTEX_WAIT, 0);
+}
+
+TEST_F(DBStatisticsTest, ResetStats) {
+  Options options = CurrentOptions();
+  options.create_if_missing = true;
+  options.statistics = rocksdb::CreateDBStatistics();
+  DestroyAndReopen(options);
+  for (int i = 0; i < 2; ++i) {
+    // pick arbitrary ticker and histogram. On first iteration they're zero
+    // because db is unused. On second iteration they're zero due to Reset().
+    ASSERT_EQ(0, TestGetTickerCount(options, NUMBER_KEYS_WRITTEN));
+    HistogramData histogram_data;
+    options.statistics->histogramData(DB_WRITE, &histogram_data);
+    ASSERT_EQ(0.0, histogram_data.max);
+
+    if (i == 0) {
+      // The Put() makes some of the ticker/histogram stats nonzero until we
+      // Reset().
+      ASSERT_OK(Put("hello", "rocksdb"));
+      ASSERT_EQ(1, TestGetTickerCount(options, NUMBER_KEYS_WRITTEN));
+      options.statistics->histogramData(DB_WRITE, &histogram_data);
+      ASSERT_GT(histogram_data.max, 0.0);
+      options.statistics->Reset();
+    }
+  }
+}
+
+}  // namespace rocksdb
+
+int main(int argc, char** argv) {
+  rocksdb::port::InstallStackTraceHandler();
+  ::testing::InitGoogleTest(&argc, argv);
+  return RUN_ALL_TESTS();
+}
@@ -147,6 +147,7 @@ TEST_F(DBTablePropertiesTest, GetPropertiesOfTablesInRange) {
  options.max_bytes_for_level_multiplier = 4;
  options.hard_pending_compaction_bytes_limit = 16 * 1024;
  options.num_levels = 8;
+  options.env = env_;

  DestroyAndReopen(options);

@@ -126,6 +126,7 @@ TEST_F(DBTestTailingIterator, TailingIteratorTrimSeekToNext) {
  options.write_buffer_size = k150KB;
  options.max_write_buffer_number = 3;
  options.min_write_buffer_number_to_merge = 2;
+  options.env = env_;
  CreateAndReopenWithCF({"pikachu"}, options);
  ReadOptions read_options;
  read_options.tailing = true;
@@ -298,7 +299,6 @@ TEST_F(DBTestTailingIterator, TailingIteratorPrefixSeek) {
  read_options.tailing = true;

  Options options = CurrentOptions();
-  options.env = env_;
  options.create_if_missing = true;
  options.disable_auto_compactions = true;
  options.prefix_extractor.reset(NewFixedPrefixTransform(2));
@@ -616,7 +616,6 @@ TEST_F(DBTestTailingIterator, ManagedTailingIteratorPrefixSeek) {
  read_options.managed = true;

  Options options = CurrentOptions();
-  options.env = env_;
  options.create_if_missing = true;
  options.disable_auto_compactions = true;
  options.prefix_extractor.reset(NewFixedPrefixTransform(2));
@@ -26,11 +26,12 @@
 #include "db/db_impl.h"
 #include "db/db_test_util.h"
 #include "db/dbformat.h"
-#include "db/filename.h"
 #include "db/job_context.h"
 #include "db/version_set.h"
 #include "db/write_batch_internal.h"
+#include "env/mock_env.h"
 #include "memtable/hash_linklist_rep.h"
+#include "monitoring/thread_status_util.h"
 #include "port/port.h"
 #include "port/stack_trace.h"
 #include "rocksdb/cache.h"
@@ -57,16 +58,15 @@
 #include "table/scoped_arena_iterator.h"
 #include "util/compression.h"
 #include "util/file_reader_writer.h"
+#include "util/filename.h"
 #include "util/hash.h"
 #include "util/logging.h"
-#include "util/mock_env.h"
 #include "util/mutexlock.h"
 #include "util/rate_limiter.h"
 #include "util/string_util.h"
 #include "util/sync_point.h"
 #include "util/testharness.h"
 #include "util/testutil.h"
-#include "util/thread_status_util.h"
 #include "utilities/merge_operators.h"

 namespace rocksdb {
@@ -2212,7 +2212,7 @@ class ModelDB : public DB {
  }
  using DB::Get;
  virtual Status Get(const ReadOptions& options, ColumnFamilyHandle* cf,
-                     const Slice& key, std::string* value) override {
+                     const Slice& key, PinnableSlice* value) override {
    return Status::NotSupported(key);
  }

@@ -2965,7 +2965,6 @@ TEST_F(DBTest, DynamicMemtableOptions) {
  const uint64_t k64KB = 1 << 16;
  const uint64_t k128KB = 1 << 17;
  const uint64_t k5KB = 5 * 1024;
-  const int kNumPutsBeforeWaitForFlush = 64;
  Options options;
  options.env = env_;
  options.create_if_missing = true;
@@ -2980,7 +2979,8 @@ TEST_F(DBTest, DynamicMemtableOptions) {
  options.level0_stop_writes_trigger = 1024;
  DestroyAndReopen(options);

-  auto gen_l0_kb = [this, kNumPutsBeforeWaitForFlush](int size) {
+  auto gen_l0_kb = [this](int size) {
+    const int kNumPutsBeforeWaitForFlush = 64;
    Random rnd(301);
    for (int i = 0; i < size; i++) {
      ASSERT_OK(Put(Key(i), RandomString(&rnd, 1024)));
@@ -4087,7 +4087,7 @@ TEST_F(DBTest, DynamicMiscOptions) {
  // sanity check
  ASSERT_OK(dbfull()->TEST_GetLatestMutableCFOptions(handles_[1],
                                                     &mutable_cf_options));
-  ASSERT_EQ(true, mutable_cf_options.report_bg_io_stats);
+  ASSERT_TRUE(mutable_cf_options.report_bg_io_stats);
  // Test compression
  // sanity check
  ASSERT_OK(dbfull()->SetOptions({{"compression", "kNoCompression"}}));
@@ -4104,7 +4104,7 @@ TEST_F(DBTest, DynamicMiscOptions) {
      dbfull()->SetOptions(handles_[1], {{"paranoid_file_checks", "true"}}));
  ASSERT_OK(dbfull()->TEST_GetLatestMutableCFOptions(handles_[1],
                                                     &mutable_cf_options));
-  ASSERT_EQ(true, mutable_cf_options.report_bg_io_stats);
+  ASSERT_TRUE(mutable_cf_options.report_bg_io_stats);
 }
 #endif  // ROCKSDB_LITE

@@ -4193,104 +4193,6 @@ TEST_F(DBTest, EncodeDecompressedBlockSizeTest) {
  }
 }

-TEST_F(DBTest, CompressionStatsTest) {
-  CompressionType type;
-
-  if (Snappy_Supported()) {
-    type = kSnappyCompression;
-    fprintf(stderr, "using snappy\n");
-  } else if (Zlib_Supported()) {
-    type = kZlibCompression;
-    fprintf(stderr, "using zlib\n");
-  } else if (BZip2_Supported()) {
-    type = kBZip2Compression;
-    fprintf(stderr, "using bzip2\n");
-  } else if (LZ4_Supported()) {
-    type = kLZ4Compression;
-    fprintf(stderr, "using lz4\n");
-  } else if (XPRESS_Supported()) {
-    type = kXpressCompression;
-    fprintf(stderr, "using xpress\n");
-  } else if (ZSTD_Supported()) {
-    type = kZSTD;
-    fprintf(stderr, "using ZSTD\n");
-  } else {
-    fprintf(stderr, "skipping test, compression disabled\n");
-    return;
-  }
-
-  Options options = CurrentOptions();
-  options.compression = type;
-  options.statistics = rocksdb::CreateDBStatistics();
-  options.statistics->stats_level_ = StatsLevel::kExceptTimeForMutex;
-  DestroyAndReopen(options);
-
-  int kNumKeysWritten = 100000;
-
-  // Check that compressions occur and are counted when compression is turned on
-  Random rnd(301);
-  for (int i = 0; i < kNumKeysWritten; ++i) {
-    // compressible string
-    ASSERT_OK(Put(Key(i), RandomString(&rnd, 128) + std::string(128, 'a')));
-  }
-  ASSERT_OK(Flush());
-  ASSERT_GT(options.statistics->getTickerCount(NUMBER_BLOCK_COMPRESSED), 0);
-
-  for (int i = 0; i < kNumKeysWritten; ++i) {
-    auto r = Get(Key(i));
-  }
-  ASSERT_GT(options.statistics->getTickerCount(NUMBER_BLOCK_DECOMPRESSED), 0);
-
-  options.compression = kNoCompression;
-  DestroyAndReopen(options);
-  uint64_t currentCompressions =
-            options.statistics->getTickerCount(NUMBER_BLOCK_COMPRESSED);
-  uint64_t currentDecompressions =
-            options.statistics->getTickerCount(NUMBER_BLOCK_DECOMPRESSED);
-
-  // Check that compressions do not occur when turned off
-  for (int i = 0; i < kNumKeysWritten; ++i) {
-    // compressible string
-    ASSERT_OK(Put(Key(i), RandomString(&rnd, 128) + std::string(128, 'a')));
-  }
-  ASSERT_OK(Flush());
-  ASSERT_EQ(options.statistics->getTickerCount(NUMBER_BLOCK_COMPRESSED)
-            - currentCompressions, 0);
-
-  for (int i = 0; i < kNumKeysWritten; ++i) {
-    auto r = Get(Key(i));
-  }
-  ASSERT_EQ(options.statistics->getTickerCount(NUMBER_BLOCK_DECOMPRESSED)
-            - currentDecompressions, 0);
-}
-
-TEST_F(DBTest, MutexWaitStatsDisabledByDefault) {
-  Options options = CurrentOptions();
-  options.create_if_missing = true;
-  options.statistics = rocksdb::CreateDBStatistics();
-  CreateAndReopenWithCF({"pikachu"}, options);
-  const uint64_t kMutexWaitDelay = 100;
-  ThreadStatusUtil::TEST_SetStateDelay(ThreadStatus::STATE_MUTEX_WAIT,
-                                       kMutexWaitDelay);
-  ASSERT_OK(Put("hello", "rocksdb"));
-  ASSERT_EQ(TestGetTickerCount(options, DB_MUTEX_WAIT_MICROS), 0);
-  ThreadStatusUtil::TEST_SetStateDelay(ThreadStatus::STATE_MUTEX_WAIT, 0);
-}
-
-TEST_F(DBTest, MutexWaitStats) {
-  Options options = CurrentOptions();
-  options.create_if_missing = true;
-  options.statistics = rocksdb::CreateDBStatistics();
-  options.statistics->stats_level_ = StatsLevel::kAll;
-  CreateAndReopenWithCF({"pikachu"}, options);
-  const uint64_t kMutexWaitDelay = 100;
-  ThreadStatusUtil::TEST_SetStateDelay(ThreadStatus::STATE_MUTEX_WAIT,
-                                       kMutexWaitDelay);
-  ASSERT_OK(Put("hello", "rocksdb"));
-  ASSERT_GE(TestGetTickerCount(options, DB_MUTEX_WAIT_MICROS), kMutexWaitDelay);
-  ThreadStatusUtil::TEST_SetStateDelay(ThreadStatus::STATE_MUTEX_WAIT, 0);
-}
-
 TEST_F(DBTest, CloseSpeedup) {
  Options options = CurrentOptions();
  options.compaction_style = kCompactionStyleLevel;
@@ -5214,7 +5116,7 @@ TEST_F(DBTest, PauseBackgroundWorkTest) {
    t.join();
  }
  // now it's done
-  ASSERT_EQ(true, done.load());
+  ASSERT_TRUE(done.load());
 }

 }  // namespace rocksdb
@@ -183,19 +183,32 @@ TEST_P(DBTestSharedWriteBufferAcrossCFs, SharedWriteBufferAcrossCFs) {
  options.write_buffer_size = 500000;  // this is never hit
  CreateAndReopenWithCF({"pikachu", "dobrynia", "nikitich"}, options);

-  // Trigger a flush on CF "nikitich"
-  ASSERT_OK(Put(0, Key(1), DummyString(1)));
-  ASSERT_OK(Put(1, Key(1), DummyString(1)));
-  ASSERT_OK(Put(3, Key(1), DummyString(90000)));
-  ASSERT_OK(Put(2, Key(2), DummyString(20000)));
-  ASSERT_OK(Put(2, Key(1), DummyString(1)));
+  WriteOptions wo;
+  wo.disableWAL = true;
+
+  // Create some data and flush "default" and "nikitich" so that they
+  // are newer CFs created.
+  ASSERT_OK(Put(3, Key(1), DummyString(1), wo));
+  Flush(3);
+  ASSERT_OK(Put(3, Key(1), DummyString(1), wo));
+  ASSERT_OK(Put(0, Key(1), DummyString(1), wo));
+  Flush(0);
+  ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"),
+            static_cast<uint64_t>(1));
+  ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "nikitich"),
+            static_cast<uint64_t>(1));
+
+  ASSERT_OK(Put(3, Key(1), DummyString(30000), wo));
+  ASSERT_OK(Put(0, Key(1), DummyString(60000), wo));
+  ASSERT_OK(Put(2, Key(1), DummyString(1), wo));
+  // No flush should trigger
  dbfull()->TEST_WaitForFlushMemTable(handles_[0]);
  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
  dbfull()->TEST_WaitForFlushMemTable(handles_[2]);
  dbfull()->TEST_WaitForFlushMemTable(handles_[3]);
  {
    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"),
-              static_cast<uint64_t>(0));
+              static_cast<uint64_t>(1));
    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "pikachu"),
              static_cast<uint64_t>(0));
    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "dobrynia"),
@@ -204,98 +217,84 @@ TEST_P(DBTestSharedWriteBufferAcrossCFs, SharedWriteBufferAcrossCFs) {
              static_cast<uint64_t>(1));
  }

-  // "dobrynia": 20KB
-  // Flush 'dobrynia'
-  ASSERT_OK(Put(3, Key(2), DummyString(40000)));
-  ASSERT_OK(Put(2, Key(2), DummyString(70000)));
-  ASSERT_OK(Put(0, Key(1), DummyString(1)));
-  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
-  dbfull()->TEST_WaitForFlushMemTable(handles_[2]);
-  dbfull()->TEST_WaitForFlushMemTable(handles_[3]);
-  {
-    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"),
-              static_cast<uint64_t>(0));
-    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "pikachu"),
-              static_cast<uint64_t>(0));
-    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "dobrynia"),
-              static_cast<uint64_t>(1));
-    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "nikitich"),
-              static_cast<uint64_t>(1));
-  }
-
-  // "nikitich" still has data of 80KB
-  // Inserting Data in "dobrynia" triggers "nikitich" flushing.
-  ASSERT_OK(Put(3, Key(2), DummyString(40000)));
-  ASSERT_OK(Put(2, Key(2), DummyString(40000)));
-  ASSERT_OK(Put(0, Key(1), DummyString(1)));
-  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
-  dbfull()->TEST_WaitForFlushMemTable(handles_[2]);
-  dbfull()->TEST_WaitForFlushMemTable(handles_[3]);
-  {
-    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"),
-              static_cast<uint64_t>(0));
-    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "pikachu"),
-              static_cast<uint64_t>(0));
-    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "dobrynia"),
-              static_cast<uint64_t>(1));
-    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "nikitich"),
-              static_cast<uint64_t>(2));
-  }
-
-  // "dobrynia" still has 40KB
-  ASSERT_OK(Put(1, Key(2), DummyString(20000)));
-  ASSERT_OK(Put(0, Key(1), DummyString(10000)));
-  ASSERT_OK(Put(0, Key(1), DummyString(1)));
+  // Trigger a flush. Flushing "nikitich".
+  ASSERT_OK(Put(3, Key(2), DummyString(30000), wo));
+  ASSERT_OK(Put(0, Key(1), DummyString(1), wo));
  dbfull()->TEST_WaitForFlushMemTable(handles_[0]);
  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
  dbfull()->TEST_WaitForFlushMemTable(handles_[2]);
  dbfull()->TEST_WaitForFlushMemTable(handles_[3]);
-  // This should triggers no flush
  {
    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"),
-              static_cast<uint64_t>(0));
+              static_cast<uint64_t>(1));
    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "pikachu"),
              static_cast<uint64_t>(0));
    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "dobrynia"),
-              static_cast<uint64_t>(1));
+              static_cast<uint64_t>(0));
    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "nikitich"),
              static_cast<uint64_t>(2));
  }

-  // "default": 10KB, "pikachu": 20KB, "dobrynia": 40KB
-  ASSERT_OK(Put(1, Key(2), DummyString(40000)));
-  ASSERT_OK(Put(0, Key(1), DummyString(1)));
+  // Without hitting the threshold, no flush should trigger.
+  ASSERT_OK(Put(2, Key(1), DummyString(30000), wo));
+  ASSERT_OK(Put(2, Key(1), DummyString(1), wo));
+  ASSERT_OK(Put(2, Key(1), DummyString(1), wo));
  dbfull()->TEST_WaitForFlushMemTable(handles_[0]);
  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
  dbfull()->TEST_WaitForFlushMemTable(handles_[2]);
  dbfull()->TEST_WaitForFlushMemTable(handles_[3]);
-  // This should triggers flush of "pikachu"
  {
    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"),
-              static_cast<uint64_t>(0));
+              static_cast<uint64_t>(1));
    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "pikachu"),
-              static_cast<uint64_t>(1));
+              static_cast<uint64_t>(0));
    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "dobrynia"),
-              static_cast<uint64_t>(1));
+              static_cast<uint64_t>(0));
    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "nikitich"),
              static_cast<uint64_t>(2));
  }

-  // "default": 10KB, "dobrynia": 40KB
-  // Some remaining writes so 'default', 'dobrynia' and 'nikitich' flush on
-  // closure.
-  ASSERT_OK(Put(3, Key(1), DummyString(1)));
-  ReopenWithColumnFamilies({"default", "pikachu", "dobrynia", "nikitich"},
-                           options);
+  // Hit the write buffer limit again. "default"
+  // will have been flushed.
+  ASSERT_OK(Put(2, Key(2), DummyString(10000), wo));
+  ASSERT_OK(Put(3, Key(1), DummyString(1), wo));
+  ASSERT_OK(Put(0, Key(1), DummyString(1), wo));
+  ASSERT_OK(Put(0, Key(1), DummyString(1), wo));
+  ASSERT_OK(Put(0, Key(1), DummyString(1), wo));
+  dbfull()->TEST_WaitForFlushMemTable(handles_[0]);
+  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
+  dbfull()->TEST_WaitForFlushMemTable(handles_[2]);
+  dbfull()->TEST_WaitForFlushMemTable(handles_[3]);
  {
    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"),
-              static_cast<uint64_t>(1));
-    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "pikachu"),
-              static_cast<uint64_t>(1));
-    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "dobrynia"),
              static_cast<uint64_t>(2));
+    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "pikachu"),
+              static_cast<uint64_t>(0));
+    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "dobrynia"),
+              static_cast<uint64_t>(0));
    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "nikitich"),
-              static_cast<uint64_t>(3));
+              static_cast<uint64_t>(2));
+  }
+
+  // Trigger another flush. This time "dobrynia". "pikachu" should not
+  // be flushed, althrough it was never flushed.
+  ASSERT_OK(Put(1, Key(1), DummyString(1), wo));
+  ASSERT_OK(Put(2, Key(1), DummyString(80000), wo));
+  ASSERT_OK(Put(1, Key(1), DummyString(1), wo));
+  ASSERT_OK(Put(2, Key(1), DummyString(1), wo));
+  dbfull()->TEST_WaitForFlushMemTable(handles_[0]);
+  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
+  dbfull()->TEST_WaitForFlushMemTable(handles_[2]);
+  dbfull()->TEST_WaitForFlushMemTable(handles_[3]);
+  {
+    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"),
+              static_cast<uint64_t>(2));
+    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "pikachu"),
+              static_cast<uint64_t>(0));
+    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "dobrynia"),
+              static_cast<uint64_t>(1));
+    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "nikitich"),
+              static_cast<uint64_t>(2));
  }
 }

@@ -314,23 +313,38 @@ TEST_F(DBTest2, SharedWriteBufferLimitAcrossDB) {
  ASSERT_OK(DB::Open(options, dbname2, &db2));

  WriteOptions wo;
+  wo.disableWAL = true;

  // Trigger a flush on cf2
-  ASSERT_OK(Put(0, Key(1), DummyString(1)));
-  ASSERT_OK(Put(1, Key(1), DummyString(1)));
-  ASSERT_OK(Put(2, Key(1), DummyString(90000)));
+  ASSERT_OK(Put(2, Key(1), DummyString(70000), wo));
+  ASSERT_OK(Put(0, Key(1), DummyString(20000), wo));

  // Insert to DB2
  ASSERT_OK(db2->Put(wo, Key(2), DummyString(20000)));

-  ASSERT_OK(Put(2, Key(1), DummyString(1)));
+  ASSERT_OK(Put(2, Key(1), DummyString(1), wo));
  dbfull()->TEST_WaitForFlushMemTable(handles_[0]);
  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
  dbfull()->TEST_WaitForFlushMemTable(handles_[2]);
  static_cast<DBImpl*>(db2)->TEST_WaitForFlushMemTable();
  {
-    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"),
+    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default") +
+                  GetNumberOfSstFilesForColumnFamily(db_, "cf1") +
+                  GetNumberOfSstFilesForColumnFamily(db_, "cf2"),
+              static_cast<uint64_t>(1));
+    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db2, "default"),
              static_cast<uint64_t>(0));
+  }
+
+  // Triggering to flush another CF in DB1
+  ASSERT_OK(db2->Put(wo, Key(2), DummyString(70000)));
+  ASSERT_OK(Put(2, Key(1), DummyString(1), wo));
+  dbfull()->TEST_WaitForFlushMemTable(handles_[0]);
+  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
+  dbfull()->TEST_WaitForFlushMemTable(handles_[2]);
+  {
+    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"),
+              static_cast<uint64_t>(1));
    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "cf1"),
              static_cast<uint64_t>(0));
    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "cf2"),
@@ -339,9 +353,8 @@ TEST_F(DBTest2, SharedWriteBufferLimitAcrossDB) {
              static_cast<uint64_t>(0));
  }

-  // db2: 20KB
-  ASSERT_OK(db2->Put(wo, Key(2), DummyString(40000)));
-  ASSERT_OK(db2->Put(wo, Key(3), DummyString(70000)));
+  // Triggering flush in DB2.
+  ASSERT_OK(db2->Put(wo, Key(3), DummyString(40000)));
  ASSERT_OK(db2->Put(wo, Key(1), DummyString(1)));
  dbfull()->TEST_WaitForFlushMemTable(handles_[0]);
  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
@@ -349,7 +362,7 @@ TEST_F(DBTest2, SharedWriteBufferLimitAcrossDB) {
  static_cast<DBImpl*>(db2)->TEST_WaitForFlushMemTable();
  {
    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"),
-              static_cast<uint64_t>(0));
+              static_cast<uint64_t>(1));
    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "cf1"),
              static_cast<uint64_t>(0));
    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "cf2"),
@@ -358,26 +371,6 @@ TEST_F(DBTest2, SharedWriteBufferLimitAcrossDB) {
              static_cast<uint64_t>(1));
  }

-  //
-  // Inserting Data in db2 and db_ triggers flushing in db_.
-  ASSERT_OK(db2->Put(wo, Key(3), DummyString(70000)));
-  ASSERT_OK(Put(2, Key(2), DummyString(45000)));
-  ASSERT_OK(Put(0, Key(1), DummyString(1)));
-  dbfull()->TEST_WaitForFlushMemTable(handles_[0]);
-  dbfull()->TEST_WaitForFlushMemTable(handles_[1]);
-  dbfull()->TEST_WaitForFlushMemTable(handles_[2]);
-  static_cast<DBImpl*>(db2)->TEST_WaitForFlushMemTable();
-  {
-    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "default"),
-              static_cast<uint64_t>(0));
-    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "cf1"),
-              static_cast<uint64_t>(0));
-    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db_, "cf2"),
-              static_cast<uint64_t>(2));
-    ASSERT_EQ(GetNumberOfSstFilesForColumnFamily(db2, "default"),
-              static_cast<uint64_t>(1));
-  }
-
  delete db2;
  ASSERT_OK(DestroyDB(dbname2, options));
 }
@@ -1228,7 +1221,7 @@ class PinL0IndexAndFilterBlocksTest : public DBTestBase,
  PinL0IndexAndFilterBlocksTest() : DBTestBase("/db_pin_l0_index_bloom_test") {}
  virtual void SetUp() override { infinite_max_files_ = GetParam(); }

-  void CreateTwoLevels(Options* options) {
+  void CreateTwoLevels(Options* options, bool close_afterwards) {
    if (infinite_max_files_) {
      options->max_open_files = -1;
    }
@@ -1256,6 +1249,9 @@ class PinL0IndexAndFilterBlocksTest : public DBTestBase,
    Put(1, "z2", "end2");
    ASSERT_OK(Flush(1));

+    if (close_afterwards) {
+      Close();  // This ensures that there is no ref to block cache entries
+    }
    table_options.block_cache->EraseUnRefEntries();
  }

@@ -1310,7 +1306,7 @@ TEST_P(PinL0IndexAndFilterBlocksTest,
 TEST_P(PinL0IndexAndFilterBlocksTest,
       MultiLevelIndexAndFilterBlocksCachedWithPinning) {
  Options options = CurrentOptions();
-  PinL0IndexAndFilterBlocksTest::CreateTwoLevels(&options);
+  PinL0IndexAndFilterBlocksTest::CreateTwoLevels(&options, false);
  // get base cache values
  uint64_t fm = TestGetTickerCount(options, BLOCK_CACHE_FILTER_MISS);
  uint64_t fh = TestGetTickerCount(options, BLOCK_CACHE_FILTER_HIT);
@@ -1339,7 +1335,10 @@ TEST_P(PinL0IndexAndFilterBlocksTest,

 TEST_P(PinL0IndexAndFilterBlocksTest, DisablePrefetchingNonL0IndexAndFilter) {
  Options options = CurrentOptions();
-  PinL0IndexAndFilterBlocksTest::CreateTwoLevels(&options);
+  // This ensures that db does not ref anything in the block cache, so
+  // EraseUnRefEntries could clear them up.
+  bool close_afterwards = true;
+  PinL0IndexAndFilterBlocksTest::CreateTwoLevels(&options, close_afterwards);

  // Get base cache values
  uint64_t fm = TestGetTickerCount(options, BLOCK_CACHE_FILTER_MISS);
@@ -1628,293 +1627,6 @@ TEST_F(DBTest2, SyncPointMarker) {
 }
 #endif

-class MergeOperatorPinningTest : public DBTest2,
-                                 public testing::WithParamInterface<bool> {
- public:
-  MergeOperatorPinningTest() { disable_block_cache_ = GetParam(); }
-
-  bool disable_block_cache_;
-};
-
-INSTANTIATE_TEST_CASE_P(MergeOperatorPinningTest, MergeOperatorPinningTest,
-                        ::testing::Bool());
-
-#ifndef ROCKSDB_LITE
-TEST_P(MergeOperatorPinningTest, OperandsMultiBlocks) {
-  Options options = CurrentOptions();
-  BlockBasedTableOptions table_options;
-  table_options.block_size = 1;  // every block will contain one entry
-  table_options.no_block_cache = disable_block_cache_;
-  options.table_factory.reset(NewBlockBasedTableFactory(table_options));
-  options.merge_operator = MergeOperators::CreateStringAppendTESTOperator();
-  options.level0_slowdown_writes_trigger = (1 << 30);
-  options.level0_stop_writes_trigger = (1 << 30);
-  options.disable_auto_compactions = true;
-  DestroyAndReopen(options);
-
-  const int kKeysPerFile = 10;
-  const int kOperandsPerKeyPerFile = 7;
-  const int kOperandSize = 100;
-  // Filse to write in L0 before compacting to lower level
-  const int kFilesPerLevel = 3;
-
-  Random rnd(301);
-  std::map<std::string, std::string> true_data;
-  int batch_num = 1;
-  int lvl_to_fill = 4;
-  int key_id = 0;
-  while (true) {
-    for (int j = 0; j < kKeysPerFile; j++) {
-      std::string key = Key(key_id % 35);
-      key_id++;
-      for (int k = 0; k < kOperandsPerKeyPerFile; k++) {
-        std::string val = RandomString(&rnd, kOperandSize);
-        ASSERT_OK(db_->Merge(WriteOptions(), key, val));
-        if (true_data[key].size() == 0) {
-          true_data[key] = val;
-        } else {
-          true_data[key] += "," + val;
-        }
-      }
-    }
-
-    if (lvl_to_fill == -1) {
-      // Keep last batch in memtable and stop
-      break;
-    }
-
-    ASSERT_OK(Flush());
-    if (batch_num % kFilesPerLevel == 0) {
-      if (lvl_to_fill != 0) {
-        MoveFilesToLevel(lvl_to_fill);
-      }
-      lvl_to_fill--;
-    }
-    batch_num++;
-  }
-
-  // 3 L0 files
-  // 1 L1 file
-  // 3 L2 files
-  // 1 L3 file
-  // 3 L4 Files
-  ASSERT_EQ(FilesPerLevel(), "3,1,3,1,3");
-
-  VerifyDBFromMap(true_data);
-}
-
-TEST_P(MergeOperatorPinningTest, Randomized) {
-  do {
-    Options options = CurrentOptions();
-    options.merge_operator = MergeOperators::CreateMaxOperator();
-    BlockBasedTableOptions table_options;
-    table_options.no_block_cache = disable_block_cache_;
-    options.table_factory.reset(NewBlockBasedTableFactory(table_options));
-    DestroyAndReopen(options);
-
-    Random rnd(301);
-    std::map<std::string, std::string> true_data;
-
-    const int kTotalMerges = 10000;
-    // Every key gets ~10 operands
-    const int kKeyRange = kTotalMerges / 10;
-    const int kOperandSize = 20;
-    const int kNumPutBefore = kKeyRange / 10;  // 10% value
-    const int kNumPutAfter = kKeyRange / 10;   // 10% overwrite
-    const int kNumDelete = kKeyRange / 10;     // 10% delete
-
-    // kNumPutBefore keys will have base values
-    for (int i = 0; i < kNumPutBefore; i++) {
-      std::string key = Key(rnd.Next() % kKeyRange);
-      std::string value = RandomString(&rnd, kOperandSize);
-      ASSERT_OK(db_->Put(WriteOptions(), key, value));
-
-      true_data[key] = value;
-    }
-
-    // Do kTotalMerges merges
-    for (int i = 0; i < kTotalMerges; i++) {
-      std::string key = Key(rnd.Next() % kKeyRange);
-      std::string value = RandomString(&rnd, kOperandSize);
-      ASSERT_OK(db_->Merge(WriteOptions(), key, value));
-
-      if (true_data[key] < value) {
-        true_data[key] = value;
-      }
-    }
-
-    // Overwrite random kNumPutAfter keys
-    for (int i = 0; i < kNumPutAfter; i++) {
-      std::string key = Key(rnd.Next() % kKeyRange);
-      std::string value = RandomString(&rnd, kOperandSize);
-      ASSERT_OK(db_->Put(WriteOptions(), key, value));
-
-      true_data[key] = value;
-    }
-
-    // Delete random kNumDelete keys
-    for (int i = 0; i < kNumDelete; i++) {
-      std::string key = Key(rnd.Next() % kKeyRange);
-      ASSERT_OK(db_->Delete(WriteOptions(), key));
-
-      true_data.erase(key);
-    }
-
-    VerifyDBFromMap(true_data);
-
-    // Skip HashCuckoo since it does not support merge operators
-  } while (ChangeOptions(kSkipMergePut | kSkipHashCuckoo));
-}
-
-class MergeOperatorHook : public MergeOperator {
- public:
-  explicit MergeOperatorHook(std::shared_ptr<MergeOperator> _merge_op)
-      : merge_op_(_merge_op) {}
-
-  virtual bool FullMergeV2(const MergeOperationInput& merge_in,
-                           MergeOperationOutput* merge_out) const override {
-    before_merge_();
-    bool res = merge_op_->FullMergeV2(merge_in, merge_out);
-    after_merge_();
-    return res;
-  }
-
-  virtual const char* Name() const override { return merge_op_->Name(); }
-
-  std::shared_ptr<MergeOperator> merge_op_;
-  std::function<void()> before_merge_ = []() {};
-  std::function<void()> after_merge_ = []() {};
-};
-
-TEST_P(MergeOperatorPinningTest, EvictCacheBeforeMerge) {
-  Options options = CurrentOptions();
-
-  auto merge_hook =
-      std::make_shared<MergeOperatorHook>(MergeOperators::CreateMaxOperator());
-  options.merge_operator = merge_hook;
-  options.disable_auto_compactions = true;
-  options.level0_slowdown_writes_trigger = (1 << 30);
-  options.level0_stop_writes_trigger = (1 << 30);
-  options.max_open_files = 20;
-  BlockBasedTableOptions bbto;
-  bbto.no_block_cache = disable_block_cache_;
-  if (bbto.no_block_cache == false) {
-    bbto.block_cache = NewLRUCache(64 * 1024 * 1024);
-  } else {
-    bbto.block_cache = nullptr;
-  }
-  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
-  DestroyAndReopen(options);
-
-  const int kNumOperands = 30;
-  const int kNumKeys = 1000;
-  const int kOperandSize = 100;
-  Random rnd(301);
-
-  // 1000 keys every key have 30 operands, every operand is in a different file
-  std::map<std::string, std::string> true_data;
-  for (int i = 0; i < kNumOperands; i++) {
-    for (int j = 0; j < kNumKeys; j++) {
-      std::string k = Key(j);
-      std::string v = RandomString(&rnd, kOperandSize);
-      ASSERT_OK(db_->Merge(WriteOptions(), k, v));
-
-      true_data[k] = std::max(true_data[k], v);
-    }
-    ASSERT_OK(Flush());
-  }
-
-  std::vector<uint64_t> file_numbers = ListTableFiles(env_, dbname_);
-  ASSERT_EQ(file_numbers.size(), kNumOperands);
-  int merge_cnt = 0;
-
-  // Code executed before merge operation
-  merge_hook->before_merge_ = [&]() {
-    // Evict all tables from cache before every merge operation
-    for (uint64_t num : file_numbers) {
-      TableCache::Evict(dbfull()->TEST_table_cache(), num);
-    }
-    // Decrease cache capacity to force all unrefed blocks to be evicted
-    if (bbto.block_cache) {
-      bbto.block_cache->SetCapacity(1);
-    }
-    merge_cnt++;
-  };
-
-  // Code executed after merge operation
-  merge_hook->after_merge_ = [&]() {
-    // Increase capacity again after doing the merge
-    if (bbto.block_cache) {
-      bbto.block_cache->SetCapacity(64 * 1024 * 1024);
-    }
-  };
-
-  size_t total_reads;
-  VerifyDBFromMap(true_data, &total_reads);
-  ASSERT_EQ(merge_cnt, total_reads);
-
-  db_->CompactRange(CompactRangeOptions(), nullptr, nullptr);
-
-  VerifyDBFromMap(true_data, &total_reads);
-}
-
-TEST_P(MergeOperatorPinningTest, TailingIterator) {
-  Options options = CurrentOptions();
-  options.merge_operator = MergeOperators::CreateMaxOperator();
-  BlockBasedTableOptions bbto;
-  bbto.no_block_cache = disable_block_cache_;
-  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
-  DestroyAndReopen(options);
-
-  const int kNumOperands = 100;
-  const int kNumWrites = 100000;
-
-  std::function<void()> writer_func = [&]() {
-    int k = 0;
-    for (int i = 0; i < kNumWrites; i++) {
-      db_->Merge(WriteOptions(), Key(k), Key(k));
-
-      if (i && i % kNumOperands == 0) {
-        k++;
-      }
-      if (i && i % 127 == 0) {
-        ASSERT_OK(Flush());
-      }
-      if (i && i % 317 == 0) {
-        ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
-      }
-    }
-  };
-
-  std::function<void()> reader_func = [&]() {
-    ReadOptions ro;
-    ro.tailing = true;
-    Iterator* iter = db_->NewIterator(ro);
-
-    iter->SeekToFirst();
-    for (int i = 0; i < (kNumWrites / kNumOperands); i++) {
-      while (!iter->Valid()) {
-        // wait for the key to be written
-        env_->SleepForMicroseconds(100);
-        iter->Seek(Key(i));
-      }
-      ASSERT_EQ(iter->key(), Key(i));
-      ASSERT_EQ(iter->value(), Key(i));
-
-      iter->Next();
-    }
-
-    delete iter;
-  };
-
-  rocksdb::port::Thread writer_thread(writer_func);
-  rocksdb::port::Thread reader_thread(reader_func);
-
-  writer_thread.join();
-  reader_thread.join();
-}
-#endif  // ROCKSDB_LITE
-
 size_t GetEncodedEntrySize(size_t key_size, size_t value_size) {
  std::string buffer;

@@ -1928,66 +1640,80 @@ size_t GetEncodedEntrySize(size_t key_size, size_t value_size) {
 TEST_F(DBTest2, ReadAmpBitmap) {
  Options options = CurrentOptions();
  BlockBasedTableOptions bbto;
-  // Disable delta encoding to make it easier to calculate read amplification
-  bbto.use_delta_encoding = false;
-  // Huge block cache to make it easier to calculate read amplification
-  bbto.block_cache = NewLRUCache(1024 * 1024 * 1024);
-  bbto.read_amp_bytes_per_bit = 16;
-  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
-  options.statistics = rocksdb::CreateDBStatistics();
-  DestroyAndReopen(options);
+  uint32_t bytes_per_bit[2] = {1, 16};
+  for (size_t k = 0; k < 2; k++) {
+    // Disable delta encoding to make it easier to calculate read amplification
+    bbto.use_delta_encoding = false;
+    // Huge block cache to make it easier to calculate read amplification
+    bbto.block_cache = NewLRUCache(1024 * 1024 * 1024);
+    bbto.read_amp_bytes_per_bit = bytes_per_bit[k];
+    options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+    options.statistics = rocksdb::CreateDBStatistics();
+    DestroyAndReopen(options);

-  const size_t kNumEntries = 10000;
+    const size_t kNumEntries = 10000;

-  Random rnd(301);
-  for (size_t i = 0; i < kNumEntries; i++) {
-    ASSERT_OK(Put(Key(static_cast<int>(i)), RandomString(&rnd, 100)));
-  }
-  ASSERT_OK(Flush());
+    Random rnd(301);
+    for (size_t i = 0; i < kNumEntries; i++) {
+      ASSERT_OK(Put(Key(static_cast<int>(i)), RandomString(&rnd, 100)));
+    }
+    ASSERT_OK(Flush());

-  Close();
-  Reopen(options);
+    Close();
+    Reopen(options);

-  // Read keys/values randomly and verify that reported read amp error
-  // is less than 2%
-  uint64_t total_useful_bytes = 0;
-  std::set<int> read_keys;
-  std::string value;
-  for (size_t i = 0; i < kNumEntries * 5; i++) {
-    int key_idx = rnd.Next() % kNumEntries;
-    std::string k = Key(key_idx);
-    ASSERT_OK(db_->Get(ReadOptions(), k, &value));
+    // Read keys/values randomly and verify that reported read amp error
+    // is less than 2%
+    uint64_t total_useful_bytes = 0;
+    std::set<int> read_keys;
+    std::string value;
+    for (size_t i = 0; i < kNumEntries * 5; i++) {
+      int key_idx = rnd.Next() % kNumEntries;
+      std::string key = Key(key_idx);
+      ASSERT_OK(db_->Get(ReadOptions(), key, &value));

-    if (read_keys.find(key_idx) == read_keys.end()) {
-      auto ik = InternalKey(k, 0, ValueType::kTypeValue);
-      total_useful_bytes += GetEncodedEntrySize(ik.size(), value.size());
-      read_keys.insert(key_idx);
+      if (read_keys.find(key_idx) == read_keys.end()) {
+        auto internal_key = InternalKey(key, 0, ValueType::kTypeValue);
+        total_useful_bytes +=
+            GetEncodedEntrySize(internal_key.size(), value.size());
+        read_keys.insert(key_idx);
+      }
+
+      double expected_read_amp =
+          static_cast<double>(total_useful_bytes) /
+          options.statistics->getTickerCount(READ_AMP_TOTAL_READ_BYTES);
+
+      double read_amp =
+          static_cast<double>(options.statistics->getTickerCount(
+              READ_AMP_ESTIMATE_USEFUL_BYTES)) /
+          options.statistics->getTickerCount(READ_AMP_TOTAL_READ_BYTES);
+
+      double error_pct = fabs(expected_read_amp - read_amp) * 100;
+      // Error between reported read amp and real read amp should be less than
+      // 2%
+      EXPECT_LE(error_pct, 2);
    }

-    double expected_read_amp =
-        static_cast<double>(total_useful_bytes) /
-        options.statistics->getTickerCount(READ_AMP_TOTAL_READ_BYTES);
+    // Make sure we read every thing in the DB (which is smaller than our cache)
+    Iterator* iter = db_->NewIterator(ReadOptions());
+    for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
+      ASSERT_EQ(iter->value().ToString(), Get(iter->key().ToString()));
+    }
+    delete iter;

-    double read_amp =
-        static_cast<double>(options.statistics->getTickerCount(
-            READ_AMP_ESTIMATE_USEFUL_BYTES)) /
-        options.statistics->getTickerCount(READ_AMP_TOTAL_READ_BYTES);
-
-    double error_pct = fabs(expected_read_amp - read_amp) * 100;
-    // Error between reported read amp and real read amp should be less than 2%
-    EXPECT_LE(error_pct, 2);
+    // Read amp is on average 100% since we read all what we loaded in memory
+    if (k == 0) {
+      ASSERT_EQ(
+          options.statistics->getTickerCount(READ_AMP_ESTIMATE_USEFUL_BYTES),
+          options.statistics->getTickerCount(READ_AMP_TOTAL_READ_BYTES));
+    } else {
+      ASSERT_NEAR(
+          options.statistics->getTickerCount(READ_AMP_ESTIMATE_USEFUL_BYTES) *
+              1.0f /
+              options.statistics->getTickerCount(READ_AMP_TOTAL_READ_BYTES),
+          1, .01);
+    }
  }
-
-  // Make sure we read every thing in the DB (which is smaller than our cache)
-  Iterator* iter = db_->NewIterator(ReadOptions());
-  for (iter->SeekToFirst(); iter->Valid(); iter->Next()) {
-    ASSERT_EQ(iter->value().ToString(), Get(iter->key().ToString()));
-  }
-  delete iter;
-
-  // Read amp is 100% since we read all what we loaded in memory
-  ASSERT_EQ(options.statistics->getTickerCount(READ_AMP_ESTIMATE_USEFUL_BYTES),
-            options.statistics->getTickerCount(READ_AMP_TOTAL_READ_BYTES));
 }

 TEST_F(DBTest2, ReadAmpBitmapLiveInCacheAfterDBClose) {
@@ -1997,82 +1723,93 @@ TEST_F(DBTest2, ReadAmpBitmapLiveInCacheAfterDBClose) {
    // the blocks again regardless of them being already in the cache
    return;
  }
+  uint32_t bytes_per_bit[2] = {1, 16};
+  for (size_t k = 0; k < 2; k++) {
+    std::shared_ptr<Cache> lru_cache = NewLRUCache(1024 * 1024 * 1024);
+    std::shared_ptr<Statistics> stats = rocksdb::CreateDBStatistics();

-  std::shared_ptr<Cache> lru_cache = NewLRUCache(1024 * 1024 * 1024);
-  std::shared_ptr<Statistics> stats = rocksdb::CreateDBStatistics();
+    Options options = CurrentOptions();
+    BlockBasedTableOptions bbto;
+    // Disable delta encoding to make it easier to calculate read amplification
+    bbto.use_delta_encoding = false;
+    // Huge block cache to make it easier to calculate read amplification
+    bbto.block_cache = lru_cache;
+    bbto.read_amp_bytes_per_bit = bytes_per_bit[k];
+    options.table_factory.reset(NewBlockBasedTableFactory(bbto));
+    options.statistics = stats;
+    DestroyAndReopen(options);

-  Options options = CurrentOptions();
-  BlockBasedTableOptions bbto;
-  // Disable delta encoding to make it easier to calculate read amplification
-  bbto.use_delta_encoding = false;
-  // Huge block cache to make it easier to calculate read amplification
-  bbto.block_cache = lru_cache;
-  bbto.read_amp_bytes_per_bit = 16;
-  options.table_factory.reset(NewBlockBasedTableFactory(bbto));
-  options.statistics = stats;
-  DestroyAndReopen(options);
+    const int kNumEntries = 10000;

-  const int kNumEntries = 10000;
+    Random rnd(301);
+    for (int i = 0; i < kNumEntries; i++) {
+      ASSERT_OK(Put(Key(i), RandomString(&rnd, 100)));
+    }
+    ASSERT_OK(Flush());

-  Random rnd(301);
-  for (int i = 0; i < kNumEntries; i++) {
-    ASSERT_OK(Put(Key(i), RandomString(&rnd, 100)));
-  }
-  ASSERT_OK(Flush());
+    Close();
+    Reopen(options);

-  Close();
-  Reopen(options);
+    uint64_t total_useful_bytes = 0;
+    std::set<int> read_keys;
+    std::string value;
+    // Iter1: Read half the DB, Read even keys
+    // Key(0), Key(2), Key(4), Key(6), Key(8), ...
+    for (int i = 0; i < kNumEntries; i += 2) {
+      std::string key = Key(i);
+      ASSERT_OK(db_->Get(ReadOptions(), key, &value));

-  uint64_t total_useful_bytes = 0;
-  std::set<int> read_keys;
-  std::string value;
-  // Iter1: Read half the DB, Read even keys
-  // Key(0), Key(2), Key(4), Key(6), Key(8), ...
-  for (int i = 0; i < kNumEntries; i += 2) {
-    std::string k = Key(i);
-    ASSERT_OK(db_->Get(ReadOptions(), k, &value));
+      if (read_keys.find(i) == read_keys.end()) {
+        auto internal_key = InternalKey(key, 0, ValueType::kTypeValue);
+        total_useful_bytes +=
+            GetEncodedEntrySize(internal_key.size(), value.size());
+        read_keys.insert(i);
+      }
+    }

-    if (read_keys.find(i) == read_keys.end()) {
-      auto ik = InternalKey(k, 0, ValueType::kTypeValue);
-      total_useful_bytes += GetEncodedEntrySize(ik.size(), value.size());
-      read_keys.insert(i);
+    size_t total_useful_bytes_iter1 =
+        options.statistics->getTickerCount(READ_AMP_ESTIMATE_USEFUL_BYTES);
+    size_t total_loaded_bytes_iter1 =
+        options.statistics->getTickerCount(READ_AMP_TOTAL_READ_BYTES);
+
+    Close();
+    std::shared_ptr<Statistics> new_statistics = rocksdb::CreateDBStatistics();
+    // Destroy old statistics obj that the blocks in lru_cache are pointing to
+    options.statistics.reset();
+    // Use the statistics object that we just created
+    options.statistics = new_statistics;
+    Reopen(options);
+
+    // Iter2: Read half the DB, Read odd keys
+    // Key(1), Key(3), Key(5), Key(7), Key(9), ...
+    for (int i = 1; i < kNumEntries; i += 2) {
+      std::string key = Key(i);
+      ASSERT_OK(db_->Get(ReadOptions(), key, &value));
+
+      if (read_keys.find(i) == read_keys.end()) {
+        auto internal_key = InternalKey(key, 0, ValueType::kTypeValue);
+        total_useful_bytes +=
+            GetEncodedEntrySize(internal_key.size(), value.size());
+        read_keys.insert(i);
+      }
+    }
+
+    size_t total_useful_bytes_iter2 =
+        options.statistics->getTickerCount(READ_AMP_ESTIMATE_USEFUL_BYTES);
+    size_t total_loaded_bytes_iter2 =
+        options.statistics->getTickerCount(READ_AMP_TOTAL_READ_BYTES);
+
+
+    // Read amp is on average 100% since we read all what we loaded in memory
+    if (k == 0) {
+      ASSERT_EQ(total_useful_bytes_iter1 + total_useful_bytes_iter2,
+                total_loaded_bytes_iter1 + total_loaded_bytes_iter2);
+    } else {
+      ASSERT_NEAR((total_useful_bytes_iter1 + total_useful_bytes_iter2) * 1.0f /
+                      (total_loaded_bytes_iter1 + total_loaded_bytes_iter2),
+                  1, .01);
    }
  }
-
-  size_t total_useful_bytes_iter1 =
-      options.statistics->getTickerCount(READ_AMP_ESTIMATE_USEFUL_BYTES);
-  size_t total_loaded_bytes_iter1 =
-      options.statistics->getTickerCount(READ_AMP_TOTAL_READ_BYTES);
-
-  Close();
-  std::shared_ptr<Statistics> new_statistics = rocksdb::CreateDBStatistics();
-  // Destroy old statistics obj that the blocks in lru_cache are pointing to
-  options.statistics.reset();
-  // Use the statistics object that we just created
-  options.statistics = new_statistics;
-  Reopen(options);
-
-  // Iter2: Read half the DB, Read odd keys
-  // Key(1), Key(3), Key(5), Key(7), Key(9), ...
-  for (int i = 1; i < kNumEntries; i += 2) {
-    std::string k = Key(i);
-    ASSERT_OK(db_->Get(ReadOptions(), k, &value));
-
-    if (read_keys.find(i) == read_keys.end()) {
-      auto ik = InternalKey(k, 0, ValueType::kTypeValue);
-      total_useful_bytes += GetEncodedEntrySize(ik.size(), value.size());
-      read_keys.insert(i);
-    }
-  }
-
-  size_t total_useful_bytes_iter2 =
-      options.statistics->getTickerCount(READ_AMP_ESTIMATE_USEFUL_BYTES);
-  size_t total_loaded_bytes_iter2 =
-      options.statistics->getTickerCount(READ_AMP_TOTAL_READ_BYTES);
-
-  // We reached read_amp of 100% because we read all the keys in the DB
-  ASSERT_EQ(total_useful_bytes_iter1 + total_useful_bytes_iter2,
-            total_loaded_bytes_iter1 + total_loaded_bytes_iter2);
 }

 #ifndef ROCKSDB_LITE
@@ -2283,7 +2020,8 @@ TEST_F(DBTest2, DirectIO) {
    return;
  }
  Options options = CurrentOptions();
-  options.use_direct_reads = options.use_direct_writes = true;
+  options.use_direct_reads = options.use_direct_io_for_flush_and_compaction =
+      true;
  options.allow_mmap_reads = options.allow_mmap_writes = false;
  DestroyAndReopen(options);

@@ -52,6 +52,7 @@ DBTestBase::DBTestBase(const std::string path)
  alternative_wal_dir_ = dbname_ + "/wal";
  alternative_db_log_dir_ = dbname_ + "/db_log_dir";
  auto options = CurrentOptions();
+  options.env = env_;
  auto delete_options = options;
  delete_options.wal_dir = alternative_wal_dir_;
  EXPECT_OK(DestroyDB(dbname_, delete_options));
@@ -72,6 +73,7 @@ DBTestBase::~DBTestBase() {
  options.db_paths.emplace_back(dbname_ + "_2", 0);
  options.db_paths.emplace_back(dbname_ + "_3", 0);
  options.db_paths.emplace_back(dbname_ + "_4", 0);
+  options.env = env_;

  if (getenv("KEEP_DB")) {
    printf("DB is still at %s\n", dbname_.c_str());
@@ -228,6 +230,8 @@ bool DBTestBase::ChangeFilterOptions() {
    option_config_ = kFilter;
  } else if (option_config_ == kFilter) {
    option_config_ = kFullFilterWithNewTableReaderForCompactions;
+  } else if (option_config_ == kFullFilterWithNewTableReaderForCompactions) {
+    option_config_ = kPartitionedFilterWithNewTableReaderForCompactions;
  } else {
    return false;
  }
@@ -323,6 +327,14 @@ Options DBTestBase::CurrentOptions(
      options.new_table_reader_for_compaction_inputs = true;
      options.compaction_readahead_size = 10 * 1024 * 1024;
      break;
+    case kPartitionedFilterWithNewTableReaderForCompactions:
+      table_options.filter_policy.reset(NewBloomFilterPolicy(10, false));
+      table_options.partition_filters = true;
+      table_options.index_type =
+          BlockBasedTableOptions::IndexType::kTwoLevelIndexSearch;
+      options.new_table_reader_for_compaction_inputs = true;
+      options.compaction_readahead_size = 10 * 1024 * 1024;
+      break;
    case kUncompressed:
      options.compression = kNoCompression;
      break;
@@ -424,6 +436,8 @@ Options DBTestBase::CurrentOptions(

  if (options_override.filter_policy) {
    table_options.filter_policy = options_override.filter_policy;
+    table_options.partition_filters = options_override.partition_filters;
+    table_options.metadata_block_size = options_override.metadata_block_size;
  }
  if (set_block_based_table_factory) {
    options.table_factory.reset(NewBlockBasedTableFactory(table_options));
@@ -14,9 +14,6 @@

 #include <fcntl.h>
 #include <inttypes.h>
-#ifndef OS_WIN
-#include <unistd.h>
-#endif

 #include <algorithm>
 #include <map>
@@ -29,7 +26,7 @@

 #include "db/db_impl.h"
 #include "db/dbformat.h"
-#include "db/filename.h"
+#include "env/mock_env.h"
 #include "memtable/hash_linklist_rep.h"
 #include "rocksdb/cache.h"
 #include "rocksdb/compaction_filter.h"
@@ -48,7 +45,7 @@
 #include "table/plain_table_factory.h"
 #include "table/scoped_arena_iterator.h"
 #include "util/compression.h"
-#include "util/mock_env.h"
+#include "util/filename.h"
 #include "util/mutexlock.h"

 #include "util/string_util.h"
@@ -112,6 +109,11 @@ class AtomicCounter {

 struct OptionsOverride {
  std::shared_ptr<const FilterPolicy> filter_policy = nullptr;
+  // These will be used only if filter_policy is set
+  bool partition_filters = false;
+  uint64_t metadata_block_size = 1024;
+  BlockBasedTableOptions::IndexType index_type =
+      BlockBasedTableOptions::IndexType::kBinarySearch;

  // Used as a bit mask of individual enums in which to skip an XF test point
  int skip_policy = 0;
@@ -275,6 +277,9 @@ class SpecialEnv : public EnvWrapper {
      bool use_direct_io() const override {
        return base_->use_direct_io();
      }
+      Status Allocate(uint64_t offset, uint64_t len) override {
+        return base_->Allocate(offset, len);
+      }
    };
    class ManifestFile : public WritableFile {
     public:
@@ -620,6 +625,7 @@ class DBTestBase : public testing::Test {
    kUniversalSubcompactions = 32,
    kBlockBasedTableWithIndexRestartInterval = 33,
    kBlockBasedTableWithPartitionedIndex = 34,
+    kPartitionedFilterWithNewTableReaderForCompactions = 35,
  };
  int option_config_;

@@ -8,10 +8,10 @@
 // found in the LICENSE file. See the AUTHORS file for names of contributors.

 #include "db/db_test_util.h"
+#include "options/options_helper.h"
 #include "port/port.h"
 #include "port/stack_trace.h"
 #include "util/fault_injection_test_env.h"
-#include "util/options_helper.h"
 #include "util/sync_point.h"

 namespace rocksdb {
@@ -14,9 +14,10 @@

 #include <inttypes.h>
 #include <stdio.h>
+#include "monitoring/perf_context_imp.h"
 #include "port/port.h"
 #include "util/coding.h"
-#include "util/perf_context_imp.h"
+#include "util/string_util.h"

 namespace rocksdb {

@@ -55,7 +56,7 @@ void AppendInternalKeyFooter(std::string* result, SequenceNumber s,

 std::string ParsedInternalKey::DebugString(bool hex) const {
  char buf[50];
-  snprintf(buf, sizeof(buf), "' @ %" PRIu64 ": %d", sequence,
+  snprintf(buf, sizeof(buf), "' seq:%" PRIu64 ", type:%d", sequence,
           static_cast<int>(type));
  std::string result = "'";
  result += user_key.ToString(hex);
@@ -128,7 +129,7 @@ void InternalKeyComparator::FindShortestSeparator(
  Slice user_limit = ExtractUserKey(limit);
  std::string tmp(user_start.data(), user_start.size());
  user_comparator_->FindShortestSeparator(&tmp, user_limit);
-  if (tmp.size() < user_start.size() &&
+  if (tmp.size() <= user_start.size() &&
      user_comparator_->Compare(user_start, tmp) < 0) {
    // User key has become shorter physically, but larger logically.
    // Tack on the earliest possible number to the shortened user key.
@@ -143,7 +144,7 @@ void InternalKeyComparator::FindShortSuccessor(std::string* key) const {
  Slice user_key = ExtractUserKey(*key);
  std::string tmp(user_key.data(), user_key.size());
  user_comparator_->FindShortSuccessor(&tmp);
-  if (tmp.size() < user_key.size() &&
+  if (tmp.size() <= user_key.size() &&
      user_comparator_->Compare(user_key, tmp) < 0) {
    // User key has become shorter physically, but larger logically.
    // Tack on the earliest possible number to the shortened user key.
@@ -303,15 +303,26 @@ inline LookupKey::~LookupKey() {
 class IterKey {
 public:
  IterKey()
-      : buf_(space_), buf_size_(sizeof(space_)), key_(buf_), key_size_(0) {}
+      : buf_(space_),
+        buf_size_(sizeof(space_)),
+        key_(buf_),
+        key_size_(0),
+        is_user_key_(true) {}

  ~IterKey() { ResetBuffer(); }

-  Slice GetKey() const { return Slice(key_, key_size_); }
+  Slice GetInternalKey() const {
+    assert(!IsUserKey());
+    return Slice(key_, key_size_);
+  }

  Slice GetUserKey() const {
-    assert(key_size_ >= 8);
-    return Slice(key_, key_size_ - 8);
+    if (IsUserKey()) {
+      return Slice(key_, key_size_);
+    } else {
+      assert(key_size_ >= 8);
+      return Slice(key_, key_size_ - 8);
+    }
  }

  size_t Size() const { return key_size_; }
@@ -349,27 +360,22 @@ class IterKey {
    key_size_ = total_size;
  }

-  Slice SetKey(const Slice& key, bool copy = true) {
-    size_t size = key.size();
-    if (copy) {
-      // Copy key to buf_
-      EnlargeBufferIfNeeded(size);
-      memcpy(buf_, key.data(), size);
-      key_ = buf_;
-    } else {
-      // Update key_ to point to external memory
-      key_ = key.data();
-    }
-    key_size_ = size;
-    return Slice(key_, key_size_);
+  Slice SetUserKey(const Slice& key, bool copy = true) {
+    is_user_key_ = true;
+    return SetKeyImpl(key, copy);
+  }
+
+  Slice SetInternalKey(const Slice& key, bool copy = true) {
+    is_user_key_ = false;
+    return SetKeyImpl(key, copy);
  }

  // Copies the content of key, updates the reference to the user key in ikey
  // and returns a Slice referencing the new copy.
-  Slice SetKey(const Slice& key, ParsedInternalKey* ikey) {
+  Slice SetInternalKey(const Slice& key, ParsedInternalKey* ikey) {
    size_t key_n = key.size();
    assert(key_n >= 8);
-    SetKey(key);
+    SetInternalKey(key);
    ikey->user_key = Slice(key_, key_n - 8);
    return Slice(key_, key_n);
  }
@@ -408,6 +414,7 @@ class IterKey {

    key_ = buf_;
    key_size_ = psize + usize + sizeof(uint64_t);
+    is_user_key_ = false;
  }

  void SetInternalKey(const Slice& user_key, SequenceNumber s,
@@ -436,14 +443,33 @@ class IterKey {
    char* ptr = EncodeVarint32(buf_, static_cast<uint32_t>(size));
    memcpy(ptr, key.data(), size);
    key_ = buf_;
+    is_user_key_ = true;
  }

+  bool IsUserKey() const { return is_user_key_; }
+
 private:
  char* buf_;
  size_t buf_size_;
  const char* key_;
  size_t key_size_;
  char space_[32];  // Avoid allocation for short keys
+  bool is_user_key_;
+
+  Slice SetKeyImpl(const Slice& key, bool copy) {
+    size_t size = key.size();
+    if (copy) {
+      // Copy key to buf_
+      EnlargeBufferIfNeeded(size);
+      memcpy(buf_, key.data(), size);
+      key_ = buf_;
+    } else {
+      // Update key_ to point to external memory
+      key_ = key.data();
+    }
+    key_size_ = size;
+    return Slice(key_, key_size_);
+  }

  void ResetBuffer() {
    if (buf_ != space_) {
@@ -139,38 +139,38 @@ TEST_F(FormatTest, IterKeyOperation) {
  const char p[] = "abcdefghijklmnopqrstuvwxyz";
  const char q[] = "0123456789";

-  ASSERT_EQ(std::string(k.GetKey().data(), k.GetKey().size()),
+  ASSERT_EQ(std::string(k.GetUserKey().data(), k.GetUserKey().size()),
            std::string(""));

  k.TrimAppend(0, p, 3);
-  ASSERT_EQ(std::string(k.GetKey().data(), k.GetKey().size()),
+  ASSERT_EQ(std::string(k.GetUserKey().data(), k.GetUserKey().size()),
            std::string("abc"));

  k.TrimAppend(1, p, 3);
-  ASSERT_EQ(std::string(k.GetKey().data(), k.GetKey().size()),
+  ASSERT_EQ(std::string(k.GetUserKey().data(), k.GetUserKey().size()),
            std::string("aabc"));

  k.TrimAppend(0, p, 26);
-  ASSERT_EQ(std::string(k.GetKey().data(), k.GetKey().size()),
+  ASSERT_EQ(std::string(k.GetUserKey().data(), k.GetUserKey().size()),
            std::string("abcdefghijklmnopqrstuvwxyz"));

  k.TrimAppend(26, q, 10);
-  ASSERT_EQ(std::string(k.GetKey().data(), k.GetKey().size()),
+  ASSERT_EQ(std::string(k.GetUserKey().data(), k.GetUserKey().size()),
            std::string("abcdefghijklmnopqrstuvwxyz0123456789"));

  k.TrimAppend(36, q, 1);
-  ASSERT_EQ(std::string(k.GetKey().data(), k.GetKey().size()),
+  ASSERT_EQ(std::string(k.GetUserKey().data(), k.GetUserKey().size()),
            std::string("abcdefghijklmnopqrstuvwxyz01234567890"));

  k.TrimAppend(26, q, 1);
-  ASSERT_EQ(std::string(k.GetKey().data(), k.GetKey().size()),
+  ASSERT_EQ(std::string(k.GetUserKey().data(), k.GetUserKey().size()),
            std::string("abcdefghijklmnopqrstuvwxyz0"));

  // Size going up, memory allocation is triggered
  k.TrimAppend(27, p, 26);
-  ASSERT_EQ(std::string(k.GetKey().data(), k.GetKey().size()),
+  ASSERT_EQ(std::string(k.GetUserKey().data(), k.GetUserKey().size()),
            std::string("abcdefghijklmnopqrstuvwxyz0"
-              "abcdefghijklmnopqrstuvwxyz"));
+                        "abcdefghijklmnopqrstuvwxyz"));
 }

 TEST_F(FormatTest, UpdateInternalKey) {
@@ -14,12 +14,12 @@
 #include <string>
 #include <vector>
 #include "db/db_impl.h"
-#include "db/filename.h"
 #include "db/version_set.h"
 #include "db/write_batch_internal.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
 #include "rocksdb/transaction_log.h"
+#include "util/filename.h"
 #include "util/string_util.h"
 #include "util/sync_point.h"
 #include "util/testharness.h"
@@ -41,8 +41,7 @@ class ExternalSSTFileBasicTest : public DBTestBase {
      const Options options, std::vector<int> keys, int file_id,
      std::map<std::string, std::string>* true_data) {
    std::string file_path = sst_files_dir_ + ToString(file_id);
-    SstFileWriter sst_file_writer(EnvOptions(), options, options.comparator,
-                                  nullptr);
+    SstFileWriter sst_file_writer(EnvOptions(), options);

    Status s = sst_file_writer.Open(file_path);
    if (!s.ok()) {
@@ -78,7 +77,7 @@ class ExternalSSTFileBasicTest : public DBTestBase {
 TEST_F(ExternalSSTFileBasicTest, Basic) {
  Options options = CurrentOptions();

-  SstFileWriter sst_file_writer(EnvOptions(), options, options.comparator);
+  SstFileWriter sst_file_writer(EnvOptions(), options);

  // Current file size should be 0 after sst_file_writer init and before open a
  // file.
@@ -121,7 +120,7 @@ TEST_F(ExternalSSTFileBasicTest, NoCopy) {
  Options options = CurrentOptions();
  const ImmutableCFOptions ioptions(options);

-  SstFileWriter sst_file_writer(EnvOptions(), options, options.comparator);
+  SstFileWriter sst_file_writer(EnvOptions(), options);

  // file1.sst (0 => 99)
  std::string file1 = sst_files_dir_ + "file1.sst";
@@ -286,8 +285,8 @@ TEST_F(ExternalSSTFileBasicTest, FadviseTrigger) {
  std::unique_ptr<SstFileWriter> sst_file_writer;

  std::string sst_file_path = sst_files_dir_ + "file_fadvise_disable.sst";
-  sst_file_writer.reset(new SstFileWriter(EnvOptions(), options,
-                                          options.comparator, nullptr, false));
+  sst_file_writer.reset(
+      new SstFileWriter(EnvOptions(), options, nullptr, false));
  ASSERT_OK(sst_file_writer->Open(sst_file_path));
  for (int i = 0; i < kNumKeys; i++) {
    ASSERT_OK(sst_file_writer->Add(Key(i), Key(i)));
@@ -298,8 +297,8 @@ TEST_F(ExternalSSTFileBasicTest, FadviseTrigger) {


  sst_file_path = sst_files_dir_ + "file_fadvise_enable.sst";
-  sst_file_writer.reset(new SstFileWriter(EnvOptions(), options,
-                                          options.comparator, nullptr, true));
+  sst_file_writer.reset(
+      new SstFileWriter(EnvOptions(), options, nullptr, true));
  ASSERT_OK(sst_file_writer->Open(sst_file_path));
  for (int i = 0; i < kNumKeys; i++) {
    ASSERT_OK(sst_file_writer->Add(Key(i), Key(i)));
@@ -120,9 +120,9 @@ Status ExternalSstFileIngestionJob::Prepare(
      }
      Status s = env_->DeleteFile(f.internal_file_path);
      if (!s.ok()) {
-        Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log,
-            "AddFile() clean up for file %s failed : %s",
-            f.internal_file_path.c_str(), s.ToString().c_str());
+        ROCKS_LOG_WARN(db_options_.info_log,
+                       "AddFile() clean up for file %s failed : %s",
+                       f.internal_file_path.c_str(), s.ToString().c_str());
      }
    }
  }
@@ -212,7 +212,8 @@ void ExternalSstFileIngestionJob::UpdateStats() {
    if (f.picked_level == 0) {
      total_l0_files += 1;
    }
-    Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log,
+    ROCKS_LOG_INFO(
+        db_options_.info_log,
        "[AddFile] External SST file %s was ingested in L%d with path %s "
        "(global_seqno=%" PRIu64 ")\n",
        f.external_file_path.c_str(), f.picked_level,
@@ -233,9 +234,9 @@ void ExternalSstFileIngestionJob::Cleanup(const Status& status) {
    for (IngestedFileInfo& f : files_to_ingest_) {
      Status s = env_->DeleteFile(f.internal_file_path);
      if (!s.ok()) {
-        Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log,
-            "AddFile() clean up for file %s failed : %s",
-            f.internal_file_path.c_str(), s.ToString().c_str());
+        ROCKS_LOG_WARN(db_options_.info_log,
+                       "AddFile() clean up for file %s failed : %s",
+                       f.internal_file_path.c_str(), s.ToString().c_str());
      }
    }
  } else if (status.ok() && ingestion_options_.move_files) {
@@ -243,7 +244,8 @@ void ExternalSstFileIngestionJob::Cleanup(const Status& status) {
    for (IngestedFileInfo& f : files_to_ingest_) {
      Status s = env_->DeleteFile(f.external_file_path);
      if (!s.ok()) {
-        Log(InfoLogLevel::WARN_LEVEL, db_options_.info_log,
+        ROCKS_LOG_WARN(
+            db_options_.info_log,
            "%s was added to DB successfully but failed to remove original "
            "file link : %s",
            f.external_file_path.c_str(), s.ToString().c_str());
@@ -12,11 +12,11 @@
 #include "db/dbformat.h"
 #include "db/internal_stats.h"
 #include "db/snapshot_impl.h"
+#include "options/db_options.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
 #include "rocksdb/sst_file_writer.h"
 #include "util/autovector.h"
-#include "util/db_options.h"

 namespace rocksdb {

@@ -54,8 +54,7 @@ class ExternalSSTFileTest : public DBTestBase {
      data.resize(uniq_iter - data.begin());
    }
    std::string file_path = sst_files_dir_ + ToString(file_id);
-    SstFileWriter sst_file_writer(EnvOptions(), options, options.comparator,
-                                  cfh);
+    SstFileWriter sst_file_writer(EnvOptions(), options, cfh);

    Status s = sst_file_writer.Open(file_path);
    if (!s.ok()) {
@@ -139,7 +138,7 @@ TEST_F(ExternalSSTFileTest, Basic) {
  do {
    Options options = CurrentOptions();

-    SstFileWriter sst_file_writer(EnvOptions(), options, options.comparator);
+    SstFileWriter sst_file_writer(EnvOptions(), options);

    // Current file size should be 0 after sst_file_writer init and before open a file.
    ASSERT_EQ(sst_file_writer.FileSize(), 0);
@@ -376,7 +375,7 @@ TEST_F(ExternalSSTFileTest, AddList) {
    options.table_properties_collector_factories.emplace_back(abc_collector);
    options.table_properties_collector_factories.emplace_back(xyz_collector);

-    SstFileWriter sst_file_writer(EnvOptions(), options, options.comparator);
+    SstFileWriter sst_file_writer(EnvOptions(), options);

    // file1.sst (0 => 99)
    std::string file1 = sst_files_dir_ + "file1.sst";
@@ -565,7 +564,7 @@ TEST_F(ExternalSSTFileTest, AddListAtomicity) {
  do {
    Options options = CurrentOptions();

-    SstFileWriter sst_file_writer(EnvOptions(), options, options.comparator);
+    SstFileWriter sst_file_writer(EnvOptions(), options);

    // files[0].sst (0 => 99)
    // files[1].sst (100 => 199)
@@ -608,7 +607,7 @@ TEST_F(ExternalSSTFileTest, AddListAtomicity) {
 // This situation may result in deleting the file while it's being added.
 TEST_F(ExternalSSTFileTest, PurgeObsoleteFilesBug) {
  Options options = CurrentOptions();
-  SstFileWriter sst_file_writer(EnvOptions(), options, options.comparator);
+  SstFileWriter sst_file_writer(EnvOptions(), options);

  // file1.sst (0 => 500)
  std::string sst_file_path = sst_files_dir_ + "file1.sst";
@@ -653,7 +652,7 @@ TEST_F(ExternalSSTFileTest, PurgeObsoleteFilesBug) {
 TEST_F(ExternalSSTFileTest, SkipSnapshot) {
  Options options = CurrentOptions();

-  SstFileWriter sst_file_writer(EnvOptions(), options, options.comparator);
+  SstFileWriter sst_file_writer(EnvOptions(), options);

  // file1.sst (0 => 99)
  std::string file1 = sst_files_dir_ + "file1.sst";
@@ -744,7 +743,7 @@ TEST_F(ExternalSSTFileTest, MultiThreaded) {
      int range_start = file_idx * keys_per_file;
      int range_end = range_start + keys_per_file;

-      SstFileWriter sst_file_writer(EnvOptions(), options, options.comparator);
+      SstFileWriter sst_file_writer(EnvOptions(), options);

      ASSERT_OK(sst_file_writer.Open(file_names[file_idx]));

@@ -842,7 +841,7 @@ TEST_F(ExternalSSTFileTest, OverlappingRanges) {
    Options options = CurrentOptions();
    DestroyAndReopen(options);

-    SstFileWriter sst_file_writer(EnvOptions(), options, options.comparator);
+    SstFileWriter sst_file_writer(EnvOptions(), options);

    printf("Option config = %d\n", option_config_);
    std::vector<std::pair<int, int>> key_ranges;
@@ -1252,7 +1251,7 @@ TEST_F(ExternalSSTFileTest, AddExternalSstFileWithCustomCompartor) {
  options.comparator = ReverseBytewiseComparator();
  DestroyAndReopen(options);

-  SstFileWriter sst_file_writer(EnvOptions(), options, options.comparator);
+  SstFileWriter sst_file_writer(EnvOptions(), options);

  // Generate files with these key ranges
  // {14  -> 0}
@@ -1354,7 +1353,7 @@ TEST_F(ExternalSSTFileTest, SstFileWriterNonSharedKeys) {
  Options options = CurrentOptions();
  DestroyAndReopen(options);
  std::string file_path = sst_files_dir_ + "/not_shared";
-  SstFileWriter sst_file_writer(EnvOptions(), options, options.comparator);
+  SstFileWriter sst_file_writer(EnvOptions(), options);

  std::string suffix(100, 'X');
  ASSERT_OK(sst_file_writer.Open(file_path));
@@ -1636,14 +1635,12 @@ TEST_F(ExternalSSTFileTest, DirtyExit) {
  std::unique_ptr<SstFileWriter> sst_file_writer;

  // Destruct SstFileWriter without calling Finish()
-  sst_file_writer.reset(
-      new SstFileWriter(EnvOptions(), options, options.comparator));
+  sst_file_writer.reset(new SstFileWriter(EnvOptions(), options));
  ASSERT_OK(sst_file_writer->Open(file_path));
  sst_file_writer.reset();

  // Destruct SstFileWriter with a failing Finish
-  sst_file_writer.reset(
-      new SstFileWriter(EnvOptions(), options, options.comparator));
+  sst_file_writer.reset(new SstFileWriter(EnvOptions(), options));
  ASSERT_OK(sst_file_writer->Open(file_path));
  ASSERT_NOK(sst_file_writer->Finish());
 }
@@ -1652,11 +1649,10 @@ TEST_F(ExternalSSTFileTest, FileWithCFInfo) {
  Options options = CurrentOptions();
  CreateAndReopenWithCF({"koko", "toto"}, options);

-  SstFileWriter sfw_default(EnvOptions(), options, options.comparator,
-                            handles_[0]);
-  SstFileWriter sfw_cf1(EnvOptions(), options, options.comparator, handles_[1]);
-  SstFileWriter sfw_cf2(EnvOptions(), options, options.comparator, handles_[2]);
-  SstFileWriter sfw_unknown(EnvOptions(), options, options.comparator);
+  SstFileWriter sfw_default(EnvOptions(), options, handles_[0]);
+  SstFileWriter sfw_cf1(EnvOptions(), options, handles_[1]);
+  SstFileWriter sfw_cf2(EnvOptions(), options, handles_[2]);
+  SstFileWriter sfw_unknown(EnvOptions(), options);

  // default_cf.sst
  const std::string cf_default_sst = sst_files_dir_ + "/default_cf.sst";
@@ -1774,7 +1770,7 @@ TEST_F(ExternalSSTFileTest, SnapshotInconsistencyBug) {

  // Overwrite all keys using IngestExternalFile
  std::string sst_file_path = sst_files_dir_ + "file1.sst";
-  SstFileWriter sst_file_writer(EnvOptions(), options, options.comparator);
+  SstFileWriter sst_file_writer(EnvOptions(), options);
  ASSERT_OK(sst_file_writer.Open(sst_file_path));
  for (int i = 0; i < kNumKeys; i++) {
    ASSERT_OK(sst_file_writer.Add(Key(i), Key(i) + "_V2"));
@@ -12,17 +12,17 @@
 // file data (or entire files) not protected by a "sync".

 #include "db/db_impl.h"
-#include "db/filename.h"
 #include "db/log_format.h"
 #include "db/version_set.h"
+#include "env/mock_env.h"
 #include "rocksdb/cache.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
 #include "rocksdb/table.h"
 #include "rocksdb/write_batch.h"
 #include "util/fault_injection_test_env.h"
+#include "util/filename.h"
 #include "util/logging.h"
-#include "util/mock_env.h"
 #include "util/mutexlock.h"
 #include "util/sync_point.h"
 #include "util/testharness.h"
@@ -7,7 +7,7 @@
 // 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.

-#include "db/filename.h"
+#include "util/filename.h"

 #include "db/dbformat.h"
 #include "port/port.h"
@@ -22,15 +22,17 @@
 #include "db/db_iter.h"
 #include "db/dbformat.h"
 #include "db/event_helpers.h"
-#include "db/filename.h"
 #include "db/log_reader.h"
 #include "db/log_writer.h"
-#include "db/memtable.h"
 #include "db/memtable_list.h"
 #include "db/merge_context.h"
 #include "db/version_set.h"
+#include "monitoring/iostats_context_imp.h"
+#include "monitoring/perf_context_imp.h"
+#include "monitoring/thread_status_util.h"
 #include "port/likely.h"
 #include "port/port.h"
+#include "db/memtable.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
 #include "rocksdb/statistics.h"
@@ -44,14 +46,12 @@
 #include "util/coding.h"
 #include "util/event_logger.h"
 #include "util/file_util.h"
-#include "util/iostats_context_imp.h"
+#include "util/filename.h"
 #include "util/log_buffer.h"
 #include "util/logging.h"
 #include "util/mutexlock.h"
-#include "util/perf_context_imp.h"
 #include "util/stop_watch.h"
 #include "util/sync_point.h"
-#include "util/thread_status_util.h"

 namespace rocksdb {

@@ -158,8 +158,8 @@ Status FlushJob::Run(FileMetaData* file_meta) {
  AutoThreadOperationStageUpdater stage_run(
      ThreadStatus::STAGE_FLUSH_RUN);
  if (mems_.empty()) {
-    LogToBuffer(log_buffer_, "[%s] Nothing in memtable to flush",
-                cfd_->GetName().c_str());
+    ROCKS_LOG_BUFFER(log_buffer_, "[%s] Nothing in memtable to flush",
+                     cfd_->GetName().c_str());
    return Status::OK();
  }

@@ -258,7 +258,8 @@ Status FlushJob::WriteLevel0Table() {
    uint64_t total_num_entries = 0, total_num_deletes = 0;
    size_t total_memory_usage = 0;
    for (MemTable* m : mems_) {
-      Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log,
+      ROCKS_LOG_INFO(
+          db_options_.info_log,
          "[%s] [JOB %d] Flushing memtable with next log file: %" PRIu64 "\n",
          cfd_->GetName().c_str(), job_context_->job_id, m->GetNextLogNumber());
      memtables.push_back(m->NewIterator(ro, &arena));
@@ -286,15 +287,18 @@ Status FlushJob::WriteLevel0Table() {
          &cfd_->internal_comparator(),
          range_del_iters.empty() ? nullptr : &range_del_iters[0],
          static_cast<int>(range_del_iters.size())));
-      Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log,
-          "[%s] [JOB %d] Level-0 flush table #%" PRIu64 ": started",
-          cfd_->GetName().c_str(), job_context_->job_id, meta_.fd.GetNumber());
+      ROCKS_LOG_INFO(db_options_.info_log,
+                     "[%s] [JOB %d] Level-0 flush table #%" PRIu64 ": started",
+                     cfd_->GetName().c_str(), job_context_->job_id,
+                     meta_.fd.GetNumber());

      TEST_SYNC_POINT_CALLBACK("FlushJob::WriteLevel0Table:output_compression",
                               &output_compression_);
+      EnvOptions optimized_env_options =
+          db_options_.env->OptimizeForCompactionTableWrite(env_options_, db_options_);
      s = BuildTable(
          dbname_, db_options_.env, *cfd_->ioptions(), mutable_cf_options_,
-          env_options_, cfd_->table_cache(), iter.get(),
+          optimized_env_options, cfd_->table_cache(), iter.get(),
          std::move(range_del_iter), &meta_, cfd_->internal_comparator(),
          cfd_->int_tbl_prop_collector_factories(), cfd_->GetID(),
          cfd_->GetName(), existing_snapshots_,
@@ -305,13 +309,14 @@ Status FlushJob::WriteLevel0Table() {
          Env::IO_HIGH, &table_properties_, 0 /* level */);
      LogFlush(db_options_.info_log);
    }
-    Log(InfoLogLevel::INFO_LEVEL, db_options_.info_log,
-        "[%s] [JOB %d] Level-0 flush table #%" PRIu64 ": %" PRIu64
-        " bytes %s"
-        "%s",
-        cfd_->GetName().c_str(), job_context_->job_id, meta_.fd.GetNumber(),
-        meta_.fd.GetFileSize(), s.ToString().c_str(),
-        meta_.marked_for_compaction ? " (needs compaction)" : "");
+    ROCKS_LOG_INFO(db_options_.info_log,
+                   "[%s] [JOB %d] Level-0 flush table #%" PRIu64 ": %" PRIu64
+                   " bytes %s"
+                   "%s",
+                   cfd_->GetName().c_str(), job_context_->job_id,
+                   meta_.fd.GetNumber(), meta_.fd.GetFileSize(),
+                   s.ToString().c_str(),
+                   meta_.marked_for_compaction ? " (needs compaction)" : "");

    if (output_file_directory_ != nullptr) {
      output_file_directory_->Fsync();
@@ -27,6 +27,8 @@
 #include "db/version_edit.h"
 #include "db/write_controller.h"
 #include "db/write_thread.h"
+#include "monitoring/instrumented_mutex.h"
+#include "options/db_options.h"
 #include "port/port.h"
 #include "rocksdb/db.h"
 #include "rocksdb/env.h"
@@ -34,9 +36,7 @@
 #include "rocksdb/transaction_log.h"
 #include "table/scoped_arena_iterator.h"
 #include "util/autovector.h"
-#include "util/db_options.h"
 #include "util/event_logger.h"
-#include "util/instrumented_mutex.h"
 #include "util/stop_watch.h"
 #include "util/thread_local.h"