Disable warning as error

This reverts commit 61720850f6.

HISTORY.md

@@ -1,14 +1,26 @@
 # Rocksdb Change Log
 ## Unreleased
+
+## 5.2.0 (02/08/2017)
 ### Public API Change
+* NewLRUCache() will determine number of shard bits automatically based on capacity, if the user doesn't pass one. This also impacts the default block cache when the user doesn't explict provide one.
+* Change the default of delayed slowdown value to 16MB/s and further increase the L0 stop condition to 36 files.
+* (Experimental) Two-level indexing that partition the index and creates a 2nd level index on the partitions. The feature can be enabled by setting kTwoLevelIndexSearch as IndexType and configuring index_per_partition.
+
+### New Features
+* Added new overloaded function GetApproximateSizes that allows to specify if memtable stats should be computed only without computing SST files' stats approximations.
+* Added new function GetApproximateMemTableStats that approximates both number of records and size of memtables.
+
+### Bug Fixes
+* RangeSync() should work if ROCKSDB_FALLOCATE_PRESENT is not set
+* Fix wrong results in a data race case in Get()
+* Some fixes related to 2PC.
+* Fix several bugs in Direct I/O supports.
+
+## 5.1.0 (01/13/2017)
 * Support dynamically change `delete_obsolete_files_period_micros` option via SetDBOptions().
 * Added EventListener::OnExternalFileIngested which will be called when IngestExternalFile() add a file successfully.
 * BackupEngine::Open and BackupEngineReadOnly::Open now always return error statuses matching those of the backup Env.
-* Added new overloaded function GetApproximateSizes that allows to specify if memtable stats should be computed only without computing SST files' stats approximations.
-* NewLRUCache() will determine number of shard bits automatically based on capacity, if the user doesn't pass one. This also impacts the default block cache when the user doesn't explict provide one.
-* Change the default of delayed slowdown value to 16MB/s and further increase the L0 stop condition to 36 files.
-* Added new function GetApproximateMemTableStats that approximates both number of records and size of memtables.
-* (Experimental) Two-level indexing that partition the index and creates a 2nd level index on the partitions. The feature can be enabled by setting kTwoLevelIndexSearch as IndexType and configuring index_per_partition.
 
 ### Bug Fixes
 * Fix the bug that if 2PC is enabled, checkpoints may loss some recent transactions.

Makefile

@@ -223,11 +223,6 @@ default: all
 WARNING_FLAGS = -W -Wextra -Wall -Wsign-compare -Wshadow \
   -Wno-unused-parameter
 
-ifndef DISABLE_WARNING_AS_ERROR
-	WARNING_FLAGS += -Werror
-endif
-
-
 ifdef LUA_PATH
 
 ifndef LUA_INCLUDE

build_tools/build_detect_platform

@@ -51,12 +51,7 @@ if [ -z "$ROCKSDB_NO_FBCODE" -a -d /mnt/gvfs/third-party ]; then
     FBCODE_BUILD="true"
     # If we're compiling with TSAN we need pic build
     PIC_BUILD=$COMPILE_WITH_TSAN
-    if [ -z "$ROCKSDB_FBCODE_BUILD_WITH_481" ]; then
     source "$PWD/build_tools/fbcode_config.sh"
-    else
-      # we need this to build with MySQL. Don't use for other purposes.
-      source "$PWD/build_tools/fbcode_config4.8.1.sh"
-    fi
 fi
 
 # Delete existing output, if it exists

build_tools/dependencies.sh

@@ -1,18 +1,19 @@
-GCC_BASE=/mnt/gvfs/third-party2/gcc/cf7d14c625ce30bae1a4661c2319c5a283e4dd22/4.9.x/centos6-native/108cf83
-CLANG_BASE=/mnt/gvfs/third-party2/llvm-fb/8598c375b0e94e1448182eb3df034704144a838d/stable/centos6-native/3f16ddd
-LIBGCC_BASE=/mnt/gvfs/third-party2/libgcc/d6e0a7da6faba45f5e5b1638f9edd7afc2f34e7d/4.9.x/gcc-4.9-glibc-2.20/024dbc3
-GLIBC_BASE=/mnt/gvfs/third-party2/glibc/d282e6e8f3d20f4e40a516834847bdc038e07973/2.20/gcc-4.9-glibc-2.20/500e281
-SNAPPY_BASE=/mnt/gvfs/third-party2/snappy/8c38a4c1e52b4c2cc8a9cdc31b9c947ed7dbfcb4/1.1.3/gcc-4.9-glibc-2.20/e9936bf
-ZLIB_BASE=/mnt/gvfs/third-party2/zlib/0882df3713c7a84f15abe368dc004581f20b39d7/1.2.8/gcc-5-glibc-2.23/9bc6787
-BZIP2_BASE=/mnt/gvfs/third-party2/bzip2/740325875f6729f42d28deaa2147b0854f3a347e/1.0.6/gcc-5-glibc-2.23/9bc6787
-LZ4_BASE=/mnt/gvfs/third-party2/lz4/0e790b441e2d9acd68d51e1d2e028f88c6a79ddf/r131/gcc-5-glibc-2.23/9bc6787
-ZSTD_BASE=/mnt/gvfs/third-party2/zstd/9455f75ff7f4831dc9fda02a6a0f8c68922fad8f/1.0.0/gcc-5-glibc-2.23/9bc6787
-GFLAGS_BASE=/mnt/gvfs/third-party2/gflags/f001a51b2854957676d07306ef3abf67186b5c8b/2.1.1/gcc-4.8.1-glibc-2.17/c3f970a
-JEMALLOC_BASE=/mnt/gvfs/third-party2/jemalloc/fc8a13ca1fffa4d0765c716c5a0b49f0c107518f/master/gcc-5-glibc-2.23/1c32b4b
-NUMA_BASE=/mnt/gvfs/third-party2/numa/17c514c4d102a25ca15f4558be564eeed76f4b6a/2.0.8/gcc-5-glibc-2.23/9bc6787
-LIBUNWIND_BASE=/mnt/gvfs/third-party2/libunwind/ad576de2a1ea560c4d3434304f0fc4e079bede42/trunk/gcc-5-glibc-2.23/b1847cb
-TBB_BASE=/mnt/gvfs/third-party2/tbb/9d9a554877d0c5bef330fe818ab7178806dd316a/4.0_update2/gcc-4.9-glibc-2.20/e9936bf
-KERNEL_HEADERS_BASE=/mnt/gvfs/third-party2/kernel-headers/7c111ff27e0c466235163f00f280a9d617c3d2ec/4.0.9-36_fbk5_2933_gd092e3f/gcc-5-glibc-2.23/da39a3e
-BINUTILS_BASE=/mnt/gvfs/third-party2/binutils/b7fd454c4b10c6a81015d4524ed06cdeab558490/2.26/centos6-native/da39a3e
-VALGRIND_BASE=/mnt/gvfs/third-party2/valgrind/d7f4d4d86674a57668e3a96f76f0e17dd0eb8765/3.10.0/gcc-4.9-glibc-2.20/e9936bf
-LUA_BASE=/mnt/gvfs/third-party2/lua/61e4abf5813bbc39bc4f548757ccfcadde175a48/5.2.3/gcc-4.9-glibc-2.20/690f0d7
+# Copyright (c) Facebook, Inc. and its affiliates. All Rights Reserved.
+GCC_BASE=/mnt/gvfs/third-party2/gcc/7331085db891a2ef4a88a48a751d834e8d68f4cb/7.x/centos7-native/b2ef2b6
+CLANG_BASE=/mnt/gvfs/third-party2/llvm-fb/963d9aeda70cc4779885b1277484fe7544a04e3e/9.0.0/platform007/9e92d53/
+LIBGCC_BASE=/mnt/gvfs/third-party2/libgcc/6ace84e956873d53638c738b6f65f3f469cca74c/7.x/platform007/5620abc
+GLIBC_BASE=/mnt/gvfs/third-party2/glibc/192b0f42d63dcf6210d6ceae387b49af049e6e0c/2.26/platform007/f259413
+SNAPPY_BASE=/mnt/gvfs/third-party2/snappy/7f9bdaada18f59bc27ec2b0871eb8a6144343aef/1.1.3/platform007/ca4da3d
+ZLIB_BASE=/mnt/gvfs/third-party2/zlib/2d9f0b9a4274cc21f61272a9e89bdb859bce8f1f/1.2.8/platform007/ca4da3d
+BZIP2_BASE=/mnt/gvfs/third-party2/bzip2/dc49a21c5fceec6456a7a28a94dcd16690af1337/1.0.6/platform007/ca4da3d
+LZ4_BASE=/mnt/gvfs/third-party2/lz4/0f607f8fc442ea7d6b876931b1898bb573d5e5da/1.9.1/platform007/ca4da3d
+ZSTD_BASE=/mnt/gvfs/third-party2/zstd/ca22bc441a4eb709e9e0b1f9fec9750fed7b31c5/1.4.x/platform007/15a3614
+GFLAGS_BASE=/mnt/gvfs/third-party2/gflags/0b9929d2588991c65a57168bf88aff2db87c5d48/2.2.0/platform007/ca4da3d
+JEMALLOC_BASE=/mnt/gvfs/third-party2/jemalloc/c26f08f47ac35fc31da2633b7da92d6b863246eb/master/platform007/c26c002
+NUMA_BASE=/mnt/gvfs/third-party2/numa/3f3fb57a5ccc5fd21c66416c0b83e0aa76a05376/2.0.11/platform007/ca4da3d
+LIBUNWIND_BASE=/mnt/gvfs/third-party2/libunwind/40c73d874898b386a71847f1b99115d93822d11f/1.4/platform007/6f3e0a9
+TBB_BASE=/mnt/gvfs/third-party2/tbb/4ce8e8dba77cdbd81b75d6f0c32fd7a1b76a11ec/2018_U5/platform007/ca4da3d
+KERNEL_HEADERS_BASE=/mnt/gvfs/third-party2/kernel-headers/fb251ecd2f5ae16f8671f7014c246e52a748fe0b/fb/platform007/da39a3e
+BINUTILS_BASE=/mnt/gvfs/third-party2/binutils/ab9f09bba370e7066cafd4eb59752db93f2e8312/2.29.1/platform007/15a3614
+VALGRIND_BASE=/mnt/gvfs/third-party2/valgrind/d42d152a15636529b0861ec493927200ebebca8e/3.15.0/platform007/ca4da3d
+LUA_BASE=/mnt/gvfs/third-party2/lua/f0cd714433206d5139df61659eb7b28b1dea6683/5.3.4/platform007/5007832

build_tools/fbcode_config.sh

@@ -13,7 +13,7 @@ source "$BASEDIR/dependencies.sh"
 CFLAGS=""
 
 # libgcc
-LIBGCC_INCLUDE="$LIBGCC_BASE/include"
+LIBGCC_INCLUDE="$LIBGCC_BASE/include/c++/7.3.0"
 LIBGCC_LIBS=" -L $LIBGCC_BASE/lib"
 
 # glibc
@@ -43,12 +43,16 @@ if test -z $PIC_BUILD; then
   LZ4_INCLUDE=" -I $LZ4_BASE/include/"
   LZ4_LIBS=" $LZ4_BASE/lib/liblz4.a"
   CFLAGS+=" -DLZ4"
-
-  ZSTD_INCLUDE=" -I $ZSTD_BASE/include/"
-  ZSTD_LIBS=" $ZSTD_BASE/lib/libzstd.a"
-  CFLAGS+=" -DZSTD"
 fi
 
+ZSTD_INCLUDE=" -I $ZSTD_BASE/include/"
+if test -z $PIC_BUILD; then
+  ZSTD_LIBS=" $ZSTD_BASE/lib/libzstd.a"
+else
+  ZSTD_LIBS=" $ZSTD_BASE/lib/libzstd_pic.a"
+fi
+CFLAGS+=" -DZSTD"
+
 # location of gflags headers and libraries
 GFLAGS_INCLUDE=" -I $GFLAGS_BASE/include/"
 if test -z $PIC_BUILD; then
@@ -56,7 +60,7 @@ if test -z $PIC_BUILD; then
 else
   GFLAGS_LIBS=" $GFLAGS_BASE/lib/libgflags_pic.a"
 fi
-CFLAGS+=" -DGFLAGS=google"
+CFLAGS+=" -DGFLAGS=gflags"
 
 # location of jemalloc
 JEMALLOC_INCLUDE=" -I $JEMALLOC_BASE/include/"
@@ -104,8 +108,8 @@ if [ -z "$USE_CLANG" ]; then
   CXX="$GCC_BASE/bin/g++"
 
   CFLAGS+=" -B$BINUTILS/gold"
-  CFLAGS+=" -isystem $GLIBC_INCLUDE"
   CFLAGS+=" -isystem $LIBGCC_INCLUDE"
+  CFLAGS+=" -isystem $GLIBC_INCLUDE"
   JEMALLOC=1
 else
   # clang
@@ -116,8 +120,8 @@ else
   KERNEL_HEADERS_INCLUDE="$KERNEL_HEADERS_BASE/include"
 
   CFLAGS+=" -B$BINUTILS/gold -nostdinc -nostdlib"
-  CFLAGS+=" -isystem $LIBGCC_BASE/include/c++/4.9.x "
-  CFLAGS+=" -isystem $LIBGCC_BASE/include/c++/4.9.x/x86_64-facebook-linux "
+  CFLAGS+=" -isystem $LIBGCC_BASE/include/c++/7.x "
+  CFLAGS+=" -isystem $LIBGCC_BASE/include/c++/7.x/x86_64-facebook-linux "
   CFLAGS+=" -isystem $GLIBC_INCLUDE"
   CFLAGS+=" -isystem $LIBGCC_INCLUDE"
   CFLAGS+=" -isystem $CLANG_INCLUDE"
@@ -128,13 +132,14 @@ else
 fi
 
 CFLAGS+=" $DEPS_INCLUDE"
-CFLAGS+=" -DROCKSDB_PLATFORM_POSIX -DROCKSDB_LIB_IO_POSIX -DROCKSDB_FALLOCATE_PRESENT -DROCKSDB_MALLOC_USABLE_SIZE"
+CFLAGS+=" -DROCKSDB_PLATFORM_POSIX -DROCKSDB_LIB_IO_POSIX -DROCKSDB_FALLOCATE_PRESENT -DROCKSDB_MALLOC_USABLE_SIZE -DROCKSDB_RANGESYNC_PRESENT -DROCKSDB_SCHED_GETCPU_PRESENT -DROCKSDB_SUPPORT_THREAD_LOCAL -DHAVE_SSE42"
 CXXFLAGS+=" $CFLAGS"
 
 EXEC_LDFLAGS=" $SNAPPY_LIBS $ZLIB_LIBS $BZIP_LIBS $LZ4_LIBS $ZSTD_LIBS $GFLAGS_LIBS $NUMA_LIB $TBB_LIBS"
-EXEC_LDFLAGS+=" -Wl,--dynamic-linker,/usr/local/fbcode/gcc-4.9-glibc-2.20/lib/ld.so"
+EXEC_LDFLAGS+=" -B$BINUTILS/gold"
+EXEC_LDFLAGS+=" -Wl,--dynamic-linker,/usr/local/fbcode/platform007/lib/ld.so"
 EXEC_LDFLAGS+=" $LIBUNWIND"
-EXEC_LDFLAGS+=" -Wl,-rpath=/usr/local/fbcode/gcc-4.9-glibc-2.20/lib"
+EXEC_LDFLAGS+=" -Wl,-rpath=/usr/local/fbcode/platform007/lib"
 # required by libtbb
 EXEC_LDFLAGS+=" -ldl"
 
@@ -144,12 +149,4 @@ EXEC_LDFLAGS_SHARED="$SNAPPY_LIBS $ZLIB_LIBS $BZIP_LIBS $LZ4_LIBS $ZSTD_LIBS $GF
 
 VALGRIND_VER="$VALGRIND_BASE/bin/"
 
-LUA_PATH="$LUA_BASE"
-
-if test -z $PIC_BUILD; then
-  LUA_LIB=" $LUA_PATH/lib/liblua.a"
-else
-  LUA_LIB=" $LUA_PATH/lib/liblua_pic.a"
-fi
-
-export CC CXX AR CFLAGS CXXFLAGS EXEC_LDFLAGS EXEC_LDFLAGS_SHARED VALGRIND_VER JEMALLOC_LIB JEMALLOC_INCLUDE CLANG_ANALYZER CLANG_SCAN_BUILD LUA_PATH LUA_LIB
+export CC CXX AR CFLAGS CXXFLAGS EXEC_LDFLAGS EXEC_LDFLAGS_SHARED VALGRIND_VER JEMALLOC_LIB JEMALLOC_INCLUDE CLANG_ANALYZER CLANG_SCAN_BUILD 

db/db_iter.cc

@@ -341,6 +341,7 @@ void DBIter::Next() {
 //
 // NOTE: In between, saved_key_ can point to a user key that has
 //       a delete marker or a sequence number higher than sequence_
+//       saved_key_ MUST have a proper user_key before calling this function
 //
 // The prefix_check parameter controls whether we check the iterated
 // keys against the prefix of the seeked key. Set to false when
@@ -946,7 +947,6 @@ void DBIter::Seek(const Slice& target) {
   StopWatch sw(env_, statistics_, DB_SEEK);
   ReleaseTempPinnedData();
   saved_key_.Clear();
-  // now saved_key is used to store internal key.
   saved_key_.SetInternalKey(target, sequence_);
 
   {
@@ -961,6 +961,9 @@ 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();
@@ -1039,6 +1042,8 @@ void DBIter::SeekToFirst() {
 
   RecordTick(statistics_, NUMBER_DB_SEEK);
   if (iter_->Valid()) {
+    saved_key_.SetKey(ExtractUserKey(iter_->key()),
+                      !iter_->IsKeyPinned() || !pin_thru_lifetime_ /* copy */);
     FindNextUserEntry(false /* not skipping */, false /* no prefix check */);
     if (statistics_ != nullptr) {
       if (valid_) {

db/db_iter_test.cc

@@ -1897,6 +1897,66 @@ TEST_F(DBIteratorTest, DBIterator12) {
   ASSERT_FALSE(db_iter->Valid());
 }
 
+TEST_F(DBIteratorTest, DBIterator13) {
+  Options options;
+  options.merge_operator = nullptr;
+
+  std::string key;
+  key.resize(9);
+  key.assign(9, static_cast<char>(0));
+  key[0] = 'b';
+
+  TestIterator* internal_iter = new TestIterator(BytewiseComparator());
+  internal_iter->AddPut(key, "0");
+  internal_iter->AddPut(key, "1");
+  internal_iter->AddPut(key, "2");
+  internal_iter->AddPut(key, "3");
+  internal_iter->AddPut(key, "4");
+  internal_iter->AddPut(key, "5");
+  internal_iter->AddPut(key, "6");
+  internal_iter->AddPut(key, "7");
+  internal_iter->AddPut(key, "8");
+  internal_iter->Finish();
+
+  std::unique_ptr<Iterator> db_iter(
+      NewDBIterator(env_, ImmutableCFOptions(options), BytewiseComparator(),
+                    internal_iter, 2, 3, 0));
+  db_iter->Seek("b");
+  ASSERT_TRUE(db_iter->Valid());
+  ASSERT_EQ(db_iter->key().ToString(), key);
+  ASSERT_EQ(db_iter->value().ToString(), "2");
+}
+
+TEST_F(DBIteratorTest, DBIterator14) {
+  Options options;
+  options.merge_operator = nullptr;
+
+  std::string key("b");
+  TestIterator* internal_iter = new TestIterator(BytewiseComparator());
+  internal_iter->AddPut("b", "0");
+  internal_iter->AddPut("b", "1");
+  internal_iter->AddPut("b", "2");
+  internal_iter->AddPut("b", "3");
+  internal_iter->AddPut("a", "4");
+  internal_iter->AddPut("a", "5");
+  internal_iter->AddPut("a", "6");
+  internal_iter->AddPut("c", "7");
+  internal_iter->AddPut("c", "8");
+  internal_iter->AddPut("c", "9");
+  internal_iter->Finish();
+
+  std::unique_ptr<Iterator> db_iter(
+      NewDBIterator(env_, ImmutableCFOptions(options), BytewiseComparator(),
+                    internal_iter, 4, 1, 0));
+  db_iter->Seek("b");
+  ASSERT_TRUE(db_iter->Valid());
+  ASSERT_EQ(db_iter->key().ToString(), "b");
+  ASSERT_EQ(db_iter->value().ToString(), "3");
+  db_iter->SeekToFirst();
+  ASSERT_EQ(db_iter->key().ToString(), "a");
+  ASSERT_EQ(db_iter->value().ToString(), "4");
+}
+
 class DBIterWithMergeIterTest : public testing::Test {
  public:
   DBIterWithMergeIterTest()

db/db_test2.cc

@@ -2277,6 +2277,27 @@ TEST_F(DBTest2, GetRaceFlush2) {
   t1.join();
   rocksdb::SyncPoint::GetInstance()->DisableProcessing();
 }
+
+TEST_F(DBTest2, DirectIO) {
+  if (!IsDirectIOSupported()) {
+    return;
+  }
+  Options options = CurrentOptions();
+  options.use_direct_reads = options.use_direct_writes = true;
+  options.allow_mmap_reads = options.allow_mmap_writes = false;
+  DestroyAndReopen(options);
+
+  ASSERT_OK(Put(Key(0), "a"));
+  ASSERT_OK(Put(Key(5), "a"));
+  ASSERT_OK(Flush());
+
+  ASSERT_OK(Put(Key(10), "a"));
+  ASSERT_OK(Put(Key(15), "a"));
+  ASSERT_OK(Flush());
+
+  ASSERT_OK(db_->CompactRange(CompactRangeOptions(), nullptr, nullptr));
+  Reopen(options);
+}
 }  // namespace rocksdb
 
 int main(int argc, char** argv) {

db/db_test_util.h

@@ -231,6 +231,20 @@ class SpecialEnv : public EnvWrapper {
           return base_->Append(data);
         }
       }
+      Status PositionedAppend(const Slice& data, uint64_t offset) override {
+        if (env_->table_write_callback_) {
+          (*env_->table_write_callback_)();
+        }
+        if (env_->drop_writes_.load(std::memory_order_acquire)) {
+          // Drop writes on the floor
+          return Status::OK();
+        } else if (env_->no_space_.load(std::memory_order_acquire)) {
+          return Status::NoSpace("No space left on device");
+        } else {
+          env_->bytes_written_ += data.size();
+          return base_->PositionedAppend(data, offset);
+        }
+      }
       Status Truncate(uint64_t size) override { return base_->Truncate(size); }
       Status Close() override {
 // SyncPoint is not supported in Released Windows Mode.
@@ -257,6 +271,9 @@ class SpecialEnv : public EnvWrapper {
       Env::IOPriority GetIOPriority() override {
         return base_->GetIOPriority();
       }
+      bool use_direct_io() const override {
+        return base_->use_direct_io();
+      }
     };
     class ManifestFile : public WritableFile {
      public:
@@ -358,7 +375,14 @@ class SpecialEnv : public EnvWrapper {
       return Status::IOError("simulated write error");
     }
 
-    Status s = target()->NewWritableFile(f, r, soptions);
+    EnvOptions optimized = soptions;
+    if (strstr(f.c_str(), "MANIFEST") != nullptr ||
+        strstr(f.c_str(), "log") != nullptr) {
+      optimized.use_mmap_writes = false;
+      optimized.use_direct_writes = false;
+    }
+
+    Status s = target()->NewWritableFile(f, r, optimized);
     if (s.ok()) {
       if (strstr(f.c_str(), ".sst") != nullptr) {
         r->reset(new SSTableFile(this, std::move(*r)));

include/rocksdb/version.h

@@ -5,8 +5,8 @@
 #pragma once
 
 #define ROCKSDB_MAJOR 5
-#define ROCKSDB_MINOR 0
-#define ROCKSDB_PATCH 0
+#define ROCKSDB_MINOR 2
+#define ROCKSDB_PATCH 1
 
 // Do not use these. We made the mistake of declaring macros starting with
 // double underscore. Now we have to live with our choice. We'll deprecate these

util/file_reader_writer.cc

@@ -468,13 +468,15 @@ class ReadaheadRandomAccessFile : public RandomAccessFile {
   ReadaheadRandomAccessFile(std::unique_ptr<RandomAccessFile>&& file,
                             size_t readahead_size)
       : file_(std::move(file)),
-        readahead_size_(readahead_size),
+        alignment_(file_->GetRequiredBufferAlignment()),
+        readahead_size_(Roundup(readahead_size, alignment_)),
         forward_calls_(file_->ShouldForwardRawRequest()),
         buffer_(),
         buffer_offset_(0),
         buffer_len_(0) {
     if (!forward_calls_) {
-      buffer_.reset(new char[readahead_size_]);
+      buffer_.Alignment(alignment_);
+      buffer_.AllocateNewBuffer(readahead_size_ + alignment_);
     } else if (readahead_size_ > 0) {
       file_->EnableReadAhead();
     }
@@ -500,31 +502,45 @@ class ReadaheadRandomAccessFile : public RandomAccessFile {
 
     std::unique_lock<std::mutex> lk(lock_);
 
-    size_t copied = 0;
-    // if offset between [buffer_offset_, buffer_offset_ + buffer_len>
-    if (offset >= buffer_offset_ && offset < buffer_len_ + buffer_offset_) {
-      uint64_t offset_in_buffer = offset - buffer_offset_;
-      copied = std::min(buffer_len_ - static_cast<size_t>(offset_in_buffer), n);
-      memcpy(scratch, buffer_.get() + offset_in_buffer, copied);
-      if (copied == n) {
-        // fully cached
-        *result = Slice(scratch, n);
+    size_t cached_len = 0;
+    // Check if there is a cache hit, means that [offset, offset + n) is either
+    // complitely or partially in the buffer
+    // If it's completely cached, including end of file case when offset + n is
+    // greater than EOF, return
+    if (TryReadFromCache_(offset, n, &cached_len, scratch) &&
+        (cached_len == n ||
+         // End of file
+         buffer_len_ < readahead_size_ + alignment_)) {
+      *result = Slice(scratch, cached_len);
       return Status::OK();
     }
-    }
+    size_t advanced_offset = offset + cached_len;
+    // In the case of cache hit advanced_offset is already aligned, means that
+    // chunk_offset equals to advanced_offset
+    size_t chunk_offset = TruncateToPageBoundary(alignment_, advanced_offset);
     Slice readahead_result;
-    Status s = file_->Read(offset + copied, readahead_size_, &readahead_result,
-      buffer_.get());
+    Status s = file_->Read(chunk_offset, readahead_size_ + alignment_,
+                           &readahead_result, buffer_.BufferStart());
     if (!s.ok()) {
       return s;
     }
+    // In the case of cache miss, i.e. when cached_len equals 0, an offset can
+    // exceed the file end position, so the following check is required
+    if (advanced_offset < chunk_offset + readahead_result.size()) {
+      // In the case of cache miss, the first chunk_padding bytes in buffer_ are
+      // stored for alignment only and must be skipped
+      size_t chunk_padding = advanced_offset - chunk_offset;
+      auto remaining_len =
+          std::min(readahead_result.size() - chunk_padding, n - cached_len);
+      memcpy(scratch + cached_len, readahead_result.data() + chunk_padding,
+             remaining_len);
+      *result = Slice(scratch, cached_len + remaining_len);
+    } else {
+      *result = Slice(scratch, cached_len);
+    }
 
-    auto left_to_copy = std::min(readahead_result.size(), n - copied);
-    memcpy(scratch + copied, readahead_result.data(), left_to_copy);
-    *result = Slice(scratch, copied + left_to_copy);
-
-    if (readahead_result.data() == buffer_.get()) {
-      buffer_offset_ = offset + copied;
+    if (readahead_result.data() == buffer_.BufferStart()) {
+      buffer_offset_ = chunk_offset;
       buffer_len_ = readahead_result.size();
     } else {
       buffer_len_ = 0;
@@ -543,13 +559,31 @@ class ReadaheadRandomAccessFile : public RandomAccessFile {
     return file_->InvalidateCache(offset, length);
   }
 
+  virtual bool use_direct_io() const override {
+    return file_->use_direct_io();
+  }
+
  private:
+  bool TryReadFromCache_(uint64_t offset, size_t n, size_t* cached_len,
+                         char* scratch) const {
+    if (offset < buffer_offset_ || offset >= buffer_offset_ + buffer_len_) {
+      *cached_len = 0;
+      return false;
+    }
+    uint64_t offset_in_buffer = offset - buffer_offset_;
+    *cached_len =
+        std::min(buffer_len_ - static_cast<size_t>(offset_in_buffer), n);
+    memcpy(scratch, buffer_.BufferStart() + offset_in_buffer, *cached_len);
+    return true;
+  }
+
   std::unique_ptr<RandomAccessFile> file_;
+  const size_t alignment_;
   size_t               readahead_size_;
   const bool           forward_calls_;
 
   mutable std::mutex   lock_;
-  mutable std::unique_ptr<char[]> buffer_;
+  mutable AlignedBuffer buffer_;
   mutable uint64_t     buffer_offset_;
   mutable size_t       buffer_len_;
 };

util/file_reader_writer_test.cc

@@ -3,10 +3,12 @@
 //  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 <vector>
 #include "util/file_reader_writer.h"
+#include <algorithm>
+#include <vector>
 #include "util/random.h"
 #include "util/testharness.h"
+#include "util/testutil.h"
 
 namespace rocksdb {
 
@@ -125,6 +127,108 @@ TEST_F(WritableFileWriterTest, AppendStatusReturn) {
   ASSERT_NOK(writer->Append(std::string(2 * kMb, 'b')));
 }
 
+class ReadaheadRandomAccessFileTest
+    : public testing::Test,
+      public testing::WithParamInterface<size_t> {
+ public:
+  static std::vector<size_t> GetReadaheadSizeList() {
+    return {1lu << 12, 1lu << 16};
+  }
+  virtual void SetUp() override {
+    readahead_size_ = GetParam();
+    scratch_.reset(new char[2 * readahead_size_]);
+    ResetSourceStr();
+  }
+  ReadaheadRandomAccessFileTest() : control_contents_() {}
+  std::string Read(uint64_t offset, size_t n) {
+    Slice result;
+    test_read_holder_->Read(offset, n, &result, scratch_.get());
+    return std::string(result.data(), result.size());
+  }
+  void ResetSourceStr(const std::string& str = "") {
+    auto write_holder = std::unique_ptr<WritableFileWriter>(
+        test::GetWritableFileWriter(new test::StringSink(&control_contents_)));
+    write_holder->Append(Slice(str));
+    write_holder->Flush();
+    auto read_holder = std::unique_ptr<RandomAccessFile>(
+        new test::StringSource(control_contents_));
+    test_read_holder_ =
+        NewReadaheadRandomAccessFile(std::move(read_holder), readahead_size_);
+  }
+  size_t GetReadaheadSize() const { return readahead_size_; }
+
+ private:
+  size_t readahead_size_;
+  Slice control_contents_;
+  std::unique_ptr<RandomAccessFile> test_read_holder_;
+  std::unique_ptr<char[]> scratch_;
+};
+
+TEST_P(ReadaheadRandomAccessFileTest, EmptySourceStrTest) {
+  ASSERT_EQ("", Read(0, 1));
+  ASSERT_EQ("", Read(0, 0));
+  ASSERT_EQ("", Read(13, 13));
+}
+
+TEST_P(ReadaheadRandomAccessFileTest, SourceStrLenLessThanReadaheadSizeTest) {
+  std::string str = "abcdefghijklmnopqrs";
+  ResetSourceStr(str);
+  ASSERT_EQ(str.substr(3, 4), Read(3, 4));
+  ASSERT_EQ(str.substr(0, 3), Read(0, 3));
+  ASSERT_EQ(str, Read(0, str.size()));
+  ASSERT_EQ(str.substr(7, std::min(static_cast<int>(str.size()) - 7, 30)),
+            Read(7, 30));
+  ASSERT_EQ("", Read(100, 100));
+}
+
+TEST_P(ReadaheadRandomAccessFileTest,
+       SourceStrLenCanBeGreaterThanReadaheadSizeTest) {
+  Random rng(42);
+  for (int k = 0; k < 100; ++k) {
+    size_t strLen = k * GetReadaheadSize() +
+                    rng.Uniform(static_cast<int>(GetReadaheadSize()));
+    std::string str =
+        test::RandomHumanReadableString(&rng, static_cast<int>(strLen));
+    ResetSourceStr(str);
+    for (int test = 1; test <= 100; ++test) {
+      size_t offset = rng.Uniform(static_cast<int>(strLen));
+      size_t n = rng.Uniform(static_cast<int>(GetReadaheadSize()));
+      ASSERT_EQ(str.substr(offset, std::min(n, str.size() - offset)),
+                Read(offset, n));
+    }
+  }
+}
+
+TEST_P(ReadaheadRandomAccessFileTest, NExceedReadaheadTest) {
+  Random rng(7);
+  size_t strLen = 4 * GetReadaheadSize() +
+                  rng.Uniform(static_cast<int>(GetReadaheadSize()));
+  std::string str =
+      test::RandomHumanReadableString(&rng, static_cast<int>(strLen));
+  ResetSourceStr(str);
+  for (int test = 1; test <= 100; ++test) {
+    size_t offset = rng.Uniform(static_cast<int>(strLen));
+    size_t n =
+        GetReadaheadSize() + rng.Uniform(static_cast<int>(GetReadaheadSize()));
+    ASSERT_EQ(str.substr(offset, std::min(n, str.size() - offset)),
+              Read(offset, n));
+  }
+}
+
+INSTANTIATE_TEST_CASE_P(
+    EmptySourceStrTest, ReadaheadRandomAccessFileTest,
+    ::testing::ValuesIn(ReadaheadRandomAccessFileTest::GetReadaheadSizeList()));
+INSTANTIATE_TEST_CASE_P(
+    SourceStrLenLessThanReadaheadSizeTest, ReadaheadRandomAccessFileTest,
+    ::testing::ValuesIn(ReadaheadRandomAccessFileTest::GetReadaheadSizeList()));
+INSTANTIATE_TEST_CASE_P(
+    SourceStrLenCanBeGreaterThanReadaheadSizeTest,
+    ReadaheadRandomAccessFileTest,
+    ::testing::ValuesIn(ReadaheadRandomAccessFileTest::GetReadaheadSizeList()));
+INSTANTIATE_TEST_CASE_P(
+    NExceedReadaheadTest, ReadaheadRandomAccessFileTest,
+    ::testing::ValuesIn(ReadaheadRandomAccessFileTest::GetReadaheadSizeList()));
+
 }  // namespace rocksdb
 
 int main(int argc, char** argv) {

util/io_posix.cc

@@ -656,6 +656,17 @@ Status PosixWritableFile::PositionedAppend(const Slice& data, uint64_t offset) {
   return Status::OK();
 }
 
+Status PosixWritableFile::Truncate(uint64_t size) {
+  Status s;
+  int r = ftruncate(fd_, size);
+  if (r < 0) {
+    s = IOError(filename_, errno);
+  } else {
+    filesize_ = size;
+  }
+  return s;
+}
+
 Status PosixWritableFile::Close() {
   Status s;
 

util/io_posix.h

@@ -93,9 +93,9 @@ class PosixWritableFile : public WritableFile {
                              const EnvOptions& options);
   virtual ~PosixWritableFile();
 
-  // Means Close() will properly take care of truncate
-  // and it does not need any additional information
-  virtual Status Truncate(uint64_t size) override { return Status::OK(); }
+  // Need to implement this so the file is truncated correctly
+  // with direct I/O
+  virtual Status Truncate(uint64_t size) override;
   virtual Status Close() override;
   virtual Status Append(const Slice& data) override;
   virtual Status PositionedAppend(const Slice& data, uint64_t offset) override;

util/options_helper.cc

@@ -536,6 +536,10 @@ bool ParseOptionHelper(char* opt_address, const OptionType& opt_type,
       return ParseEnum<CompactionStyle>(
           compaction_style_string_map, value,
           reinterpret_cast<CompactionStyle*>(opt_address));
+    case OptionType::kCompactionPri:
+      return ParseEnum<CompactionPri>(
+          compaction_pri_string_map, value,
+          reinterpret_cast<CompactionPri*>(opt_address));
     case OptionType::kCompressionType:
       return ParseEnum<CompressionType>(
           compression_type_string_map, value,
@@ -617,6 +621,10 @@ bool SerializeSingleOptionHelper(const char* opt_address,
       return SerializeEnum<CompactionStyle>(
           compaction_style_string_map,
           *(reinterpret_cast<const CompactionStyle*>(opt_address)), value);
+    case OptionType::kCompactionPri:
+      return SerializeEnum<CompactionPri>(
+          compaction_pri_string_map,
+          *(reinterpret_cast<const CompactionPri*>(opt_address)), value);
     case OptionType::kCompressionType:
       return SerializeEnum<CompressionType>(
           compression_type_string_map,

util/options_helper.h

@@ -106,6 +106,7 @@ enum class OptionType {
   kString,
   kDouble,
   kCompactionStyle,
+  kCompactionPri,
   kSliceTransform,
   kCompressionType,
   kVectorCompressionType,
@@ -573,8 +574,10 @@ static std::unordered_map<std::string, OptionTypeInfo> cf_options_type_info = {
       OptionType::kMergeOperator, OptionVerificationType::kByName, false, 0}},
     {"compaction_style",
      {offsetof(struct ColumnFamilyOptions, compaction_style),
-      OptionType::kCompactionStyle, OptionVerificationType::kNormal, false,
-      0}}};
+      OptionType::kCompactionStyle, OptionVerificationType::kNormal, false, 0}},
+    {"compaction_pri",
+     {offsetof(struct ColumnFamilyOptions, compaction_pri),
+      OptionType::kCompactionPri, OptionVerificationType::kNormal, false, 0}}};
 
 static std::unordered_map<std::string, OptionTypeInfo>
     block_based_table_type_info = {
@@ -706,6 +709,13 @@ static std::unordered_map<std::string, CompactionStyle>
         {"kCompactionStyleFIFO", kCompactionStyleFIFO},
         {"kCompactionStyleNone", kCompactionStyleNone}};
 
+static std::unordered_map<std::string, CompactionPri>
+    compaction_pri_string_map = {
+        {"kByCompensatedSize", kByCompensatedSize},
+        {"kOldestLargestSeqFirst", kOldestLargestSeqFirst},
+        {"kOldestSmallestSeqFirst", kOldestSmallestSeqFirst},
+        {"kMinOverlappingRatio", kMinOverlappingRatio}};
+
 static std::unordered_map<std::string,
                           WALRecoveryMode> wal_recovery_mode_string_map = {
     {"kTolerateCorruptedTailRecords",

util/options_parser.cc

@@ -542,6 +542,9 @@ bool AreEqualOptions(
     case OptionType::kCompactionStyle:
       return (*reinterpret_cast<const CompactionStyle*>(offset1) ==
               *reinterpret_cast<const CompactionStyle*>(offset2));
+    case OptionType::kCompactionPri:
+      return (*reinterpret_cast<const CompactionPri*>(offset1) ==
+              *reinterpret_cast<const CompactionPri*>(offset2));
     case OptionType::kCompressionType:
       return (*reinterpret_cast<const CompressionType*>(offset1) ==
               *reinterpret_cast<const CompressionType*>(offset2));

util/options_settable_test.cc

@@ -373,7 +373,6 @@ TEST_F(OptionsSettableTest, ColumnFamilyOptionsAllFieldsSettable) {
   // Following options are not settable through
   // GetColumnFamilyOptionsFromString():
   options->rate_limit_delay_max_milliseconds = 33;
-  options->compaction_pri = CompactionPri::kOldestSmallestSeqFirst;
   options->compaction_options_universal = CompactionOptionsUniversal();
   options->compression_opts = CompressionOptions();
   options->hard_rate_limit = 0;
@@ -433,6 +432,7 @@ TEST_F(OptionsSettableTest, ColumnFamilyOptionsAllFieldsSettable) {
       "level_compaction_dynamic_level_bytes=false;"
       "inplace_update_support=false;"
       "compaction_style=kCompactionStyleFIFO;"
+      "compaction_pri=kMinOverlappingRatio;"
       "purge_redundant_kvs_while_flush=true;"
       "hard_pending_compaction_bytes_limit=0;"
       "disable_auto_compactions=false;"

util/options_test.cc

@@ -77,6 +77,7 @@ TEST_F(OptionsTest, GetOptionsFromMapTest) {
       {"arena_block_size", "22"},
       {"disable_auto_compactions", "true"},
       {"compaction_style", "kCompactionStyleLevel"},
+      {"compaction_pri", "kOldestSmallestSeqFirst"},
       {"verify_checksums_in_compaction", "false"},
       {"compaction_options_fifo", "23"},
       {"max_sequential_skip_in_iterations", "24"},
@@ -176,6 +177,7 @@ TEST_F(OptionsTest, GetOptionsFromMapTest) {
   ASSERT_EQ(new_cf_opt.disable_auto_compactions, true);
   ASSERT_EQ(new_cf_opt.compaction_style, kCompactionStyleLevel);
   ASSERT_EQ(new_cf_opt.verify_checksums_in_compaction, false);
+  ASSERT_EQ(new_cf_opt.compaction_pri, kOldestSmallestSeqFirst);
   ASSERT_EQ(new_cf_opt.compaction_options_fifo.max_table_files_size,
             static_cast<uint64_t>(23));
   ASSERT_EQ(new_cf_opt.max_sequential_skip_in_iterations,