Dbid feature

Summary: Create a new type of file on startup if it doesn't already exist called DBID. This will store a unique number generated from boost library's uuid header file. The use-case is to identify the case of a db losing all its data and coming back up either empty or from an image(backup/live replica's recovery) the key point to note is that DBID is not stored in a backup or db snapshot It's preferable to use Boost for uuid because: 1) A non-standard way of generating uuid is not good 2) /proc/sys/kernel/random/uuid generates a uuid but only on linux environments and the solution would not be clean 3) c++ doesn't have any direct way to get a uuid 4) Boost is a very good library that was already having linkage in rocksdb from third-party Note: I had to update the TOOLCHAIN_REV in build files to get latest verison of boost from third-party as the older version had a bug. I had to put Wno-uninitialized in Makefile because boost-1.51 has an unitialized variable and rocksdb would not comiple otherwise. Latet open-source for boost is 1.54 but is not there in third-party. I have notified the concerned people in fbcode about it. @kailiu : While releasing to third-party, an additional dependency will need to be created for boost in TARGETS file. I can help identify. Test Plan: Expand db_test to test 2 cases 1) Restarting db with Id file present - verify that no change to Id 2)Restarting db with Id file deleted - verify that a different Id is there after reopen Also run make all check Reviewers: dhruba, haobo, kailiu, sdong Reviewed By: dhruba CC: leveldb Differential Revision: https://reviews.facebook.net/D13587
Disallow transaction log iterator to skip sequences
2026-07-07 22:55:23 +08:00 · 2013-10-22 12:23:34 -07:00 · 2013-10-22 11:45:35 -07:00 · 2013-10-21 10:23:56 -07:00 · 2013-10-20 22:02:05 -07:00 · 2013-10-17 22:15:57 -07:00
548 changed files with 8709 additions and 108433 deletions
@@ -21,3 +21,4 @@ util/build_version.cc
 build_tools/VALGRIND_LOGS/
 coverage/COVERAGE_REPORT
 util/build_version.cc.tmp
+.gdbhistory
@@ -1,3 +1,11 @@
+BSD License
+
+For rocksdb software
+
+Copyright (c) 2013, Facebook, Inc.
+All rights reserved.
+---------------------------------------------------------------------
+
 Copyright (c) 2011 The LevelDB Authors. All rights reserved.

 Redistribution and use in source and binary forms, with or without
@@ -14,7 +14,7 @@ OPT += -O2 -fno-omit-frame-pointer -momit-leaf-frame-pointer
 #-----------------------------------------------

 # detect what platform we're building on
-$(shell (cd build_tools/; ROCKSDB_ROOT=.. ./build_detect_platform ../build_config.mk))
+$(shell (export ROCKSDB_ROOT=$(CURDIR); $(CURDIR)/build_tools/build_detect_platform $(CURDIR)/build_config.mk))
 # this file is generated by the previous line to set build flags and sources
 include build_config.mk

@@ -48,6 +48,7 @@ TESTS = \
 	db_test \
 	dbformat_test \
 	env_test \
+	blob_store_test \
 	filelock_test \
 	filename_test \
 	filter_block_test \
@@ -64,7 +65,8 @@ TESTS = \
 	ttl_test \
 	version_edit_test \
 	version_set_test \
-	write_batch_test
+	write_batch_test\
+	deletefile_test

 TOOLS = \
        sst_dump \
@@ -121,7 +123,7 @@ release:

 coverage:
 	$(MAKE) clean
-	COVERAGEFLAGS="-fprofile-arcs -ftest-coverage" $(MAKE) all check
+	COVERAGEFLAGS="-fprofile-arcs -ftest-coverage" LDFLAGS+="-lgcov" $(MAKE) all check
 	(cd coverage; ./coverage_test.sh)
 	# Delete intermediate files
 	find . -type f -regex ".*\.\(\(gcda\)\|\(gcno\)\)" | xargs --no-run-if-empty rm
@@ -155,7 +157,7 @@ valgrind_check: all $(PROGRAMS) $(TESTS)
 	done

 clean:
-	-rm -f $(PROGRAMS) $(BENCHMARKS) $(LIBRARY) $(SHARED) $(MEMENVLIBRARY) $(THRIFTSERVER) build_config.mk
+	-rm -f $(PROGRAMS) $(BENCHMARKS) $(LIBRARY) $(SHARED) $(MEMENVLIBRARY) build_config.mk
 	-rm -rf ios-x86/* ios-arm/*
 	-find . -name "*.[od]" | xargs --no-run-if-empty rm
 	-find . -type f -regex ".*\.\(\(gcda\)\|\(gcno\)\)" | xargs --no-run-if-empty rm
@@ -203,6 +205,9 @@ cache_test: util/cache_test.o $(LIBOBJECTS) $(TESTHARNESS)
 coding_test: util/coding_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(CXX) util/coding_test.o $(LIBOBJECTS) $(TESTHARNESS) $(EXEC_LDFLAGS) -o $@ $(LDFLAGS) $(COVERAGEFLAGS)

+blob_store_test: util/blob_store_test.o $(LIBOBJECTS) $(TESTHARNESS) $(TESTUTIL)
+	$(CXX) util/blob_store_test.o $(LIBOBJECTS) $(TESTHARNESS) $(EXEC_LDFLAGS) -o$@ $(LDFLAGS) $(COVERAGEFLAGS)
+
 stringappend_test: utilities/merge_operators/string_append/stringappend_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(CXX) utilities/merge_operators/string_append/stringappend_test.o $(LIBOBJECTS) $(TESTHARNESS) $(EXEC_LDFLAGS) -o $@ $(LDFLAGS) $(COVERAGEFLAGS)

@@ -266,6 +271,9 @@ write_batch_test: db/write_batch_test.o $(LIBOBJECTS) $(TESTHARNESS)
 merge_test: db/merge_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(CXX) db/merge_test.o $(LIBOBJECTS) $(TESTHARNESS) $(EXEC_LDFLAGS) -o $@ $(LDFLAGS) $(COVERAGEFLAGS)

+deletefile_test: db/deletefile_test.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(CXX) db/deletefile_test.o $(LIBOBJECTS) $(TESTHARNESS) $(EXEC_LDFLAGS) -o $@ $(LDFLAGS)
+
 $(MEMENVLIBRARY) : $(MEMENVOBJECTS)
 	rm -f $@
 	$(AR) -rs $@ $(MEMENVOBJECTS)
@@ -0,0 +1,23 @@
+Additional Grant of Patent Rights
+
+“Software” means the rocksdb software distributed by Facebook, Inc.
+
+Facebook hereby grants you a perpetual, worldwide, royalty-free,
+non-exclusive, irrevocable (subject to the termination provision below)
+license under any rights in any patent claims owned by Facebook, to make,
+have made, use, sell, offer to sell, import, and otherwise transfer the
+Software. For avoidance of doubt, no license is granted under Facebook’s
+rights in any patent claims that are infringed by (i) modifications to the
+Software made by you or a third party, or (ii) the Software in combination
+with any software or other technology provided by you or a third party.
+
+The license granted hereunder will terminate, automatically and without
+notice, for anyone that makes any claim (including by filing any lawsuit,
+assertion or other action) alleging (a) direct, indirect, or contributory
+infringement or inducement to infringe any patent: (i) by Facebook or any
+of its subsidiaries or affiliates, whether or not such claim is related
+to the Software, (ii) by any party if such claim arises in whole or in
+part from any software, product or service of Facebook or any of its
+subsidiaries or affiliates, whether or not such claim is related to the
+Software, or (iii) by any party relating to the Software; or (b) that
+any right in any patent claim of Facebook is invalid or unenforceable.
@@ -1,6 +1,6 @@
 rocksdb: A persistent key-value store for flash storage
 Authors: * The Facebook Database Engineering Team
-         * Build on earlier work on leveldb by Sanjay Ghemawat 
+         * Build on earlier work on leveldb by Sanjay Ghemawat
           (sanjay@google.com) and Jeff Dean (jeff@google.com)

 This code is a library that forms the core building block for a fast
@@ -19,64 +19,64 @@ persistent key/value store.
 See doc/index.html for more explanation.
 See doc/impl.html for a brief overview of the implementation.

-The public interface is in include/*.h.  Callers should not include or
+The public interface is in include/*.  Callers should not include or
 rely on the details of any other header files in this package.  Those
 internal APIs may be changed without warning.

 Guide to header files:

-include/db.h
+include/rocksdb/db.h
    Main interface to the DB: Start here

-include/options.h
+include/rocksdb/options.h
    Control over the behavior of an entire database, and also
    control over the behavior of individual reads and writes.

-include/comparator.h
+include/rocksdb/comparator.h
    Abstraction for user-specified comparison function.  If you want
    just bytewise comparison of keys, you can use the default comparator,
    but clients can write their own comparator implementations if they
    want custom ordering (e.g. to handle different character
    encodings, etc.)

-include/iterator.h
+include/rocksdb/iterator.h
    Interface for iterating over data. You can get an iterator
    from a DB object.

-include/write_batch.h
+include/rocksdb/write_batch.h
    Interface for atomically applying multiple updates to a database.

-include/slice.h
+include/rocksdb/slice.h
    A simple module for maintaining a pointer and a length into some
    other byte array.

-include/status.h
+include/rocksdb/status.h
    Status is returned from many of the public interfaces and is used
    to report success and various kinds of errors.

-include/env.h
+include/rocksdb/env.h
    Abstraction of the OS environment.  A posix implementation of
    this interface is in util/env_posix.cc

-include/table_builder.h
+include/rocksdb/table_builder.h
    Lower-level modules that most clients probably won't use directly

-include/cache.h
+include/rocksdb/cache.h
    An API for the block cache.

-include/compaction_filter.h
+include/rocksdb/compaction_filter.h
    An API for a application filter invoked on every compaction.

-include/filter_policy.h
+include/rocksdb/filter_policy.h
    An API for configuring a bloom filter.

-include/memtablerep.h
+include/rocksdb/memtablerep.h
    An API for implementing a memtable.

-include/statistics.h
+include/rocksdb/statistics.h
    An API to retrieve various database statistics.

-include/transaction_log_iterator.h
+include/rocksdb/transaction_log.h
    An API to retrieve transaction logs from a database.


@@ -1,3 +1,3 @@
 * Detailed instructions on how to compile using fbcode and jemalloc

-* Latest release is 2.1.fb
+* Latest release is 2.4.fb
@@ -31,9 +31,9 @@ fi
 # Default to fbcode gcc on internal fb machines
 if [ -d /mnt/gvfs/third-party -a -z "$CXX" ]; then
    if [ -z "$USE_CLANG" ]; then
-        source ./fbcode.gcc471.sh
+        source $PWD/build_tools/fbcode.gcc471.sh
    else
-        source ./fbcode.clang31.sh
+        source $PWD/build_tools/fbcode.clang31.sh
    fi
 fi

@@ -65,7 +65,7 @@ PLATFORM_SHARED_CFLAGS="-fPIC"
 PLATFORM_SHARED_VERSIONED=true

 # generic port files (working on all platform by #ifdef) go directly in /port
-GENERIC_PORT_FILES=`cd $ROCKSDB_ROOT; find port -name '*.cc' | tr "\n" " "`
+GENERIC_PORT_FILES=`find $ROCKSDB_ROOT/port -name '*.cc' | tr "\n" " "`

 # On GCC, we pick libc's memcmp over GCC's memcmp via -fno-builtin-memcmp
 case "$TARGET_OS" in
@@ -82,7 +82,7 @@ case "$TARGET_OS" in
        if [ -z "$USE_CLANG" ]; then
            COMMON_FLAGS="$COMMON_FLAGS -fno-builtin-memcmp"
        fi
-        PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lpthread"
+        PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS -lpthread -lrt"
        # PORT_FILES=port/linux/linux_specific.cc
        ;;
    SunOS)
@@ -127,21 +127,13 @@ case "$TARGET_OS" in
        exit 1
 esac

-./build_detect_version
+$PWD/build_tools/build_detect_version

 # We want to make a list of all cc files within util, db, table, and helpers
 # except for the test and benchmark files. By default, find will output a list
 # of all files matching either rule, so we need to append -print to make the
 # prune take effect.
 DIRS="util db table utilities"
-if test "$USE_THRIFT"; then
-  DIRS="$DIRS thrift/server_utils.cpp thrift/gen-cpp "
-  THRIFTSERVER=leveldb_server
-fi
-
-if test "$USE_SCRIBE"; then
-  DIRS="$DIRS scribe "
-fi

 set -f # temporarily disable globbing so that our patterns arent expanded
 PRUNE_TEST="-name *test*.cc -prune"
@@ -227,14 +219,6 @@ if test "$USE_HDFS"; then
  PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS $HDFS_LDFLAGS"
 fi

-# shall we build thrift server or scribe logger
-if test "$USE_THRIFT" ||  test "$USE_SCRIBE" ; then
-    THRIFT_CCFLAGS=" -I./thrift -I./thrift/gen-cpp -I./thrift/lib/cpp -I/usr/include -std=gnu++0x"
-    THRIFT_LDFLAGS=" -lexample -lserver -lthrift_base -ltransport -lthrift_exception -lutil -L./thrift/libs "
-    COMMON_FLAGS="$COMMON_FLAGS $THRIFT_CCFLAGS"
-    PLATFORM_LDFLAGS="$PLATFORM_LDFLAGS $THRIFT_LDFLAGS"
-fi
-
 # if Intel SSE instruction set is supported, set USE_SSE=" -msse -msse4.2 "
 COMMON_FLAGS="$COMMON_FLAGS $USE_SSE"

@@ -254,5 +238,4 @@ echo "PLATFORM_SHARED_CFLAGS=$PLATFORM_SHARED_CFLAGS" >> $OUTPUT
 echo "PLATFORM_SHARED_EXT=$PLATFORM_SHARED_EXT" >> $OUTPUT
 echo "PLATFORM_SHARED_LDFLAGS=$PLATFORM_SHARED_LDFLAGS" >> $OUTPUT
 echo "PLATFORM_SHARED_VERSIONED=$PLATFORM_SHARED_VERSIONED" >> $OUTPUT
-echo "THRIFTSERVER=$THRIFTSERVER" >> $OUTPUT
 echo "EXEC_LDFLAGS=$EXEC_LDFLAGS" >> $OUTPUT
@@ -19,22 +19,22 @@ if [ "$?" = 0 ]; then
  awk '
  BEGIN {
    print "#include \"build_version.h\"\n"
-  } 
-  { print "const char* leveldb_build_git_sha = \"leveldb_build_git_sha:" $0"\";" }
+  }
+  { print "const char* rocksdb_build_git_sha = \"rocksdb_build_git_sha:" $0"\";" }
  ' >  ${VFILE}
 else
  echo "git not found" |
  awk '
    BEGIN {
-      print "#include \"build_version.h\"" 
-    } 
-    { print "const char* leveldb_build_git_sha = \"leveldb_build_git_sha:git not found\";" }
+      print "#include \"build_version.h\""
+    }
+    { print "const char* rocksdb_build_git_sha = \"rocksdb_build_git_sha:git not found\";" }
    ' > ${VFILE}
 fi

-echo "const char* leveldb_build_git_datetime = \"leveldb_build_git_datetime:$(date)\";" >> ${VFILE}
-echo "const char* leveldb_build_compile_date = __DATE__;" >> ${VFILE}
-echo "const char* leveldb_build_compile_time = __TIME__;" >> ${VFILE}
+echo "const char* rocksdb_build_git_datetime = \"rocksdb_build_git_datetime:$(date)\";" >> ${VFILE}
+echo "const char* rocksdb_build_compile_date = __DATE__;" >> ${VFILE}
+echo "const char* rocksdb_build_compile_time = __TIME__;" >> ${VFILE}

 OUTFILE=$ROCKSDB_ROOT/util/build_version.cc
 if [ ! -e $OUTFILE ] || ! cmp -s $VFILE $OUTFILE; then
@@ -1,86 +0,0 @@
-#!/bin/bash -e
-function print_usage {
-  echo "Usage: $0 [build | release | bump_version version ]"
-}
-
-function check_env {
-  if [ -z $LEVELDB_HOME ] ; then
-    echo "Expect LEVELDB_HOME to be SET"
-    exit 1
-  fi
-  if [ -z $JAVA_HOME ] ; then
-    export JAVA_HOME=/usr/local/jdk-6u14-64/
-    echo "JAVA_HOME not set. Assuming JAVA_HOME=$JAVA_HOME"
-  fi
-  if [ -z $LEVELDBJNI_HOME ] ; then
-    export LEVELDBJNI_HOME=$LEVELDB_HOME/java/leveldbjni/
-    echo "LEVELDBJNI_HOME not set. Assuming LEVELDBJNI_HOME=$LEVELDBJNI_HOME"
-  fi
-  if [ -z $SNAPPY_HOME ] ; then
-    export SNAPPY_HOME="/home/dhruba/snappy-1.0.5"
-    echo "SNAPPY_HOME not set. Assuming SNAPPY_HOME=$SNAPPY_HOME"
-  fi
-  if [ -z $LEVELDB_PATCH ] ; then 
-    LEVELDB_PATCH=$LEVELDB_HOME/java/leveldbjni/db.h.patch
-    echo "LEVELDB_PATCH not set. Assuming LEVELDB_PATCH=$LEVELDB_PATCH"
-  fi
-}
-
-function build {
-  cd $LEVELDB_HOME
-  git apply $LEVELDB_PATCH
-  make clean libleveldb.a
-  cd $LEVELDB_HOME/java/leveldb/leveldb-api
-  mvn clean package
-  cd $LEVELDBJNI_HOME
-  mvn clean install -P linux64
-  cd $LEVELDB_HOME
-  git checkout $LEVELDB_HOME/include/leveldb/db.h
-}
-
-function release {
-  cd $LEVELDB_HOME
-  git apply $LEVELDB_PATCH
-  make clean libleveldb.a
-  cd $LEVELDB_HOME/java/leveldb/leveldb-api
-  mvn clean package
-  cd $LEVELDBJNI_HOME
-  mvn deploy -P linux64 -DskipTests
-  cd $LEVELDB_HOME
-  git checkout $LEVELDB_HOME/include/leveldb/db.h
-}
-
-CMD=$1
-if [ -z $CMD ]; then
-  print_usage
-  exit 1
-fi
-
-case "$CMD" in 
-  build) 
-    check_env
-    build
-    ;;
-  bump_version)
-    if [ -z $2 ] ; then
-      echo "bump_version requires a version.no parameter at the end"
-      exit 1
-    fi 
-    check_env
-    VERSION=$2
-    pushd $LEVELDBJNI_HOME
-    mvn versions:set -DnewVersion="$VERSION"
-    popd
-    echo "VERSION SET TO $VERSION"
-    ;;
-  release)
-    check_env
-    pushd $LEVELDBJNI_HOME
-    release
-    popd
-    ;;
-  *)
-    print_usage
-    exit 1
-    ;;
-esac
@@ -4,21 +4,16 @@
 # fbcode settings.  It uses the latest g++ compiler and also
 # uses jemalloc

-TOOLCHAIN_REV=f365dbeae46a30414a2874a6f45e73e10f1caf7d
+TOOLCHAIN_REV=fbe3b095a4cc4a3713730050d182b7b4a80c342f
 TOOLCHAIN_EXECUTABLES="/mnt/gvfs/third-party/$TOOLCHAIN_REV/centos5.2-native"
 TOOLCHAIN_LIB_BASE="/mnt/gvfs/third-party/$TOOLCHAIN_REV/gcc-4.7.1-glibc-2.14.1"
-TOOL_JEMALLOC=jemalloc-3.0.0/2f45f3a
+TOOL_JEMALLOC=jemalloc-3.3.1/9202ce3
 GLIBC_RUNTIME_PATH=/usr/local/fbcode/gcc-4.7.1-glibc-2.14.1

 # location of snappy headers and libraries
 SNAPPY_INCLUDE=" -I $TOOLCHAIN_LIB_BASE/snappy/snappy-1.0.3/7518bbe/include"
 SNAPPY_LIBS=" $TOOLCHAIN_LIB_BASE/snappy/snappy-1.0.3/7518bbe/lib/libsnappy.a"

-# location of boost headers and libraries
-THRIFT_INCLUDE=" -I $TOOLCHAIN_LIB_BASE/boost/boost-1.48.0/bef9365/include -std=gnu++0x"
-THRIFT_INCLUDE+=" -I./thrift -I./thrift/gen-cpp -I./thrift/lib/cpp"
-THRIFT_LIBS=" -L $TOOLCHAIN_LIB_BASE/boost/boost-1.48.0/bef9365/lib"
-
 # location of libevent
 LIBEVENT_INCLUDE=" -I $TOOLCHAIN_LIB_BASE/libevent/libevent-1.4.14b/91ddd43/include"
 LIBEVENT_LIBS=" -L $TOOLCHAIN_LIB_BASE/libevent/libevent-1.4.14b/91ddd43/lib"
@@ -27,7 +22,7 @@ LIBEVENT_LIBS=" -L $TOOLCHAIN_LIB_BASE/libevent/libevent-1.4.14b/91ddd43/lib"
 export USE_SSE=" -msse -msse4.2 "

 CC="$TOOLCHAIN_EXECUTABLES/clang/clang-3.1/6ca8a59/bin/clang $CLANG_INCLUDES"
-CXX="$TOOLCHAIN_EXECUTABLES/clang/clang-3.1/6ca8a59/bin/clang++ $CLANG_INCLUDES $JINCLUDE $SNAPPY_INCLUDE $THRIFT_INCLUDE $LIBEVENT_INCLUDE"
+CXX="$TOOLCHAIN_EXECUTABLES/clang/clang-3.1/6ca8a59/bin/clang++ $CLANG_INCLUDES $JINCLUDE $SNAPPY_INCLUDE $LIBEVENT_INCLUDE"
 AR=$TOOLCHAIN_EXECUTABLES/binutils/binutils-2.21.1/da39a3e/bin/ar
 RANLIB=$TOOLCHAIN_EXECUTABLES/binutils/binutils-2.21.1/da39a3e/bin/ranlib

@@ -37,8 +32,8 @@ CXXFLAGS="$CFLAGS -nostdinc++ -std=gnu++0x"
 CFLAGS+=" -I $TOOLCHAIN_LIB_BASE/jemalloc/$TOOL_JEMALLOC/include -DHAVE_JEMALLOC"

 EXEC_LDFLAGS=" -Wl,--whole-archive $TOOLCHAIN_LIB_BASE/jemalloc/$TOOL_JEMALLOC/lib/libjemalloc.a"
-EXEC_LDFLAGS+=" -Wl,--no-whole-archive $TOOLCHAIN_LIB_BASE/libunwind/libunwind-1.0.1/91ddd43/lib/libunwind.a"
-EXEC_LDFLAGS+=" $HDFSLIB $SNAPPY_LIBS $THRIFT_LIBS $LIBEVENT_LIBS"
+EXEC_LDFLAGS+=" -Wl,--no-whole-archive $TOOLCHAIN_LIB_BASE/libunwind/libunwind-1.0.1/350336c/lib/libunwind.a"
+EXEC_LDFLAGS+=" $HDFSLIB $SNAPPY_LIBS $LIBEVENT_LIBS"
 EXEC_LDFLAGS+=" -Wl,--dynamic-linker,$GLIBC_RUNTIME_PATH/lib/ld-linux-x86-64.so.2"
 EXEC_LDFLAGS+=" -B$TOOLCHAIN_EXECUTABLES/binutils/binutils-2.21.1/da39a3e/bin"

@@ -4,10 +4,10 @@
 # fbcode settings.  It uses the latest g++ compiler and also
 # uses jemalloc

-TOOLCHAIN_REV=83eb773e262fa705eaebbf3de40db71d53fabf2e
+TOOLCHAIN_REV=fbe3b095a4cc4a3713730050d182b7b4a80c342f
 TOOLCHAIN_EXECUTABLES="/mnt/gvfs/third-party/$TOOLCHAIN_REV/centos5.2-native"
 TOOLCHAIN_LIB_BASE="/mnt/gvfs/third-party/$TOOLCHAIN_REV/gcc-4.7.1-glibc-2.14.1"
-TOOL_JEMALLOC=jemalloc-3.3.1/2f45f3a
+TOOL_JEMALLOC=jemalloc-3.3.1/9202ce3

 # location of libhdfs libraries
 if test "$USE_HDFS"; then
@@ -27,11 +27,6 @@ SNAPPY_LIBS=" $TOOLCHAIN_LIB_BASE/snappy/snappy-1.0.3/7518bbe/lib/libsnappy.a"
 ZLIB_INCLUDE=" -I $TOOLCHAIN_LIB_BASE/zlib/zlib-1.2.5/91ddd43/include"
 ZLIB_LIBS=" $TOOLCHAIN_LIB_BASE/zlib/zlib-1.2.5/91ddd43/lib/libz.a"

-# location of boost headers and libraries
-THRIFT_INCLUDE=" -I $TOOLCHAIN_LIB_BASE/boost/boost-1.48.0/2a0840d/include"
-THRIFT_INCLUDE+=" -I./thrift -I./thrift/gen-cpp -I./thrift/lib/cpp"
-THRIFT_LIBS=" -L $TOOLCHAIN_LIB_BASE/boost/boost-1.48.0/2a0840d/lib"
-
 # location of libevent
 LIBEVENT_INCLUDE=" -I $TOOLCHAIN_LIB_BASE/libevent/libevent-1.4.14b/91ddd43/include"
 LIBEVENT_LIBS=" -L $TOOLCHAIN_LIB_BASE/libevent/libevent-1.4.14b/91ddd43/lib"
@@ -40,7 +35,7 @@ LIBEVENT_LIBS=" -L $TOOLCHAIN_LIB_BASE/libevent/libevent-1.4.14b/91ddd43/lib"
 export USE_SSE=" -msse -msse4.2 "

 CC="$TOOLCHAIN_EXECUTABLES/gcc/gcc-4.7.1-glibc-2.14.1/bin/gcc"
-CXX="$TOOLCHAIN_EXECUTABLES/gcc/gcc-4.7.1-glibc-2.14.1/bin/g++ $JINCLUDE $SNAPPY_INCLUDE $ZLIB_INCLUDE $THRIFT_INCLUDE $LIBEVENT_INCLUDE"
+CXX="$TOOLCHAIN_EXECUTABLES/gcc/gcc-4.7.1-glibc-2.14.1/bin/g++ $JINCLUDE $SNAPPY_INCLUDE $ZLIB_INCLUDE $LIBEVENT_INCLUDE"
 AR=$TOOLCHAIN_EXECUTABLES/binutils/binutils-2.21.1/da39a3e/bin/ar
 RANLIB=$TOOLCHAIN_EXECUTABLES/binutils/binutils-2.21.1/da39a3e/bin/ranlib

@@ -48,8 +43,8 @@ CFLAGS="-B$TOOLCHAIN_EXECUTABLES/binutils/binutils-2.21.1/bin/gold -m64 -mtune=g
 CFLAGS+=" -I $TOOLCHAIN_LIB_BASE/jemalloc/$TOOL_JEMALLOC/include -DHAVE_JEMALLOC"

 EXEC_LDFLAGS=" -Wl,--whole-archive $TOOLCHAIN_LIB_BASE/jemalloc/$TOOL_JEMALLOC/lib/libjemalloc.a"
-EXEC_LDFLAGS+=" -Wl,--no-whole-archive $TOOLCHAIN_LIB_BASE/libunwind/libunwind-1.0.1/91ddd43/lib/libunwind.a"
-EXEC_LDFLAGS+=" $HDFSLIB $SNAPPY_LIBS $ZLIB_LIBS $THRIFT_LIBS $LIBEVENT_LIBS"
+EXEC_LDFLAGS+=" -Wl,--no-whole-archive $TOOLCHAIN_LIB_BASE/libunwind/libunwind-1.0.1/350336c/lib/libunwind.a"
+EXEC_LDFLAGS+=" $HDFSLIB $SNAPPY_LIBS $ZLIB_LIBS $LIBEVENT_LIBS"

 PLATFORM_LDFLAGS="-L$TOOLCHAIN_LIB_BASE/libgcc/libgcc-4.7.1/afc21dc/lib -L$TOOLCHAIN_LIB_BASE/glibc/glibc-2.14.1/99df8fc/lib"

@@ -55,12 +55,19 @@ then
  exit 0
 fi

+LCOV_VERSION=$(lcov -v | grep 1.1 || true)
+if [ $LCOV_VERSION ]
+then
+  echo "Not supported lcov version. Expect lcov 1.1."
+  exit 0
+fi
+
 (cd $ROOT; lcov --no-external \
     --capture  \
     --directory $PWD \
     --gcov-tool $GCOV \
-     --output-file $COVERAGE_DIR/coverage.info &>/dev/null)
+     --output-file $COVERAGE_DIR/coverage.info)

-genhtml $COVERAGE_DIR/coverage.info -o $COVERAGE_DIR &>/dev/null
+genhtml $COVERAGE_DIR/coverage.info -o $COVERAGE_DIR

 echo "HTML Coverage report is generated in $COVERAGE_DIR"
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -9,12 +14,12 @@
 #include "db/merge_helper.h"
 #include "db/table_cache.h"
 #include "db/version_edit.h"
-#include "leveldb/db.h"
-#include "leveldb/env.h"
-#include "leveldb/iterator.h"
+#include "rocksdb/db.h"
+#include "rocksdb/env.h"
+#include "rocksdb/iterator.h"
 #include "util/stop_watch.h"

-namespace leveldb {
+namespace rocksdb {

 Status BuildTable(const std::string& dbname,
                  Env* env,
@@ -25,7 +30,8 @@ Status BuildTable(const std::string& dbname,
                  FileMetaData* meta,
                  const Comparator* user_comparator,
                  const SequenceNumber newest_snapshot,
-                  const SequenceNumber earliest_seqno_in_memtable) {
+                  const SequenceNumber earliest_seqno_in_memtable,
+                  const bool enable_compression) {
  Status s;
  meta->file_size = 0;
  meta->smallest_seqno = meta->largest_seqno = 0;
@@ -46,7 +52,8 @@ Status BuildTable(const std::string& dbname,
    if (!s.ok()) {
      return s;
    }
-    TableBuilder* builder = new TableBuilder(options, file.get(), 0);
+    TableBuilder* builder = new TableBuilder(options, file.get(), 0,
+                                             enable_compression);

    // the first key is the smallest key
    Slice key = iter->key();
@@ -204,4 +211,4 @@ Status BuildTable(const std::string& dbname,
  return s;
 }

-}  // namespace leveldb
+}  // namespace rocksdb
@@ -1,15 +1,16 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
 // Copyright (c) 2011 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
+#include "rocksdb/comparator.h"
+#include "rocksdb/status.h"
+#include "rocksdb/types.h"

-#ifndef STORAGE_LEVELDB_DB_BUILDER_H_
-#define STORAGE_LEVELDB_DB_BUILDER_H_
-
-#include "leveldb/comparator.h"
-#include "leveldb/status.h"
-#include "leveldb/types.h"
-
-namespace leveldb {
+namespace rocksdb {

 struct Options;
 struct FileMetaData;
@@ -34,8 +35,7 @@ extern Status BuildTable(const std::string& dbname,
                         FileMetaData* meta,
                         const Comparator* user_comparator,
                         const SequenceNumber newest_snapshot,
-                         const SequenceNumber earliest_seqno_in_memtable);
+                         const SequenceNumber earliest_seqno_in_memtable,
+                         const bool enable_compression);

-}  // namespace leveldb
-
-#endif  // STORAGE_LEVELDB_DB_BUILDER_H_
+}  // namespace rocksdb
@@ -1,43 +1,48 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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 "leveldb/c.h"
+#include "rocksdb/c.h"

 #include <stdlib.h>
 #include <unistd.h>
-#include "leveldb/cache.h"
-#include "leveldb/comparator.h"
-#include "leveldb/db.h"
-#include "leveldb/env.h"
-#include "leveldb/filter_policy.h"
-#include "leveldb/iterator.h"
-#include "leveldb/options.h"
-#include "leveldb/status.h"
-#include "leveldb/write_batch.h"
+#include "rocksdb/cache.h"
+#include "rocksdb/comparator.h"
+#include "rocksdb/db.h"
+#include "rocksdb/env.h"
+#include "rocksdb/filter_policy.h"
+#include "rocksdb/iterator.h"
+#include "rocksdb/options.h"
+#include "rocksdb/status.h"
+#include "rocksdb/write_batch.h"

-using leveldb::Cache;
-using leveldb::Comparator;
-using leveldb::CompressionType;
-using leveldb::DB;
-using leveldb::Env;
-using leveldb::FileLock;
-using leveldb::FilterPolicy;
-using leveldb::Iterator;
-using leveldb::Logger;
-using leveldb::NewBloomFilterPolicy;
-using leveldb::NewLRUCache;
-using leveldb::Options;
-using leveldb::RandomAccessFile;
-using leveldb::Range;
-using leveldb::ReadOptions;
-using leveldb::SequentialFile;
-using leveldb::Slice;
-using leveldb::Snapshot;
-using leveldb::Status;
-using leveldb::WritableFile;
-using leveldb::WriteBatch;
-using leveldb::WriteOptions;
+using rocksdb::Cache;
+using rocksdb::Comparator;
+using rocksdb::CompressionType;
+using rocksdb::DB;
+using rocksdb::Env;
+using rocksdb::FileLock;
+using rocksdb::FilterPolicy;
+using rocksdb::Iterator;
+using rocksdb::Logger;
+using rocksdb::NewBloomFilterPolicy;
+using rocksdb::NewLRUCache;
+using rocksdb::Options;
+using rocksdb::RandomAccessFile;
+using rocksdb::Range;
+using rocksdb::ReadOptions;
+using rocksdb::SequentialFile;
+using rocksdb::Slice;
+using rocksdb::Snapshot;
+using rocksdb::Status;
+using rocksdb::WritableFile;
+using rocksdb::WriteBatch;
+using rocksdb::WriteOptions;

 using std::shared_ptr;

@@ -2,7 +2,7 @@
   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 "leveldb/c.h"
+#include "rocksdb/c.h"

 #include <stddef.h>
 #include <stdio.h>
@@ -296,7 +296,7 @@ int main(int argc, char** argv) {
  {
    char* prop = leveldb_property_value(db, "nosuchprop");
    CheckCondition(prop == NULL);
-    prop = leveldb_property_value(db, "leveldb.stats");
+    prop = leveldb_property_value(db, "rocksdb.stats");
    CheckCondition(prop != NULL);
    Free(&prop);
  }
@@ -1,16 +1,21 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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 "leveldb/db.h"
+#include "rocksdb/db.h"

 #include <errno.h>
 #include <fcntl.h>
 #include <sys/stat.h>
 #include <sys/types.h>
-#include "leveldb/cache.h"
-#include "leveldb/env.h"
-#include "leveldb/write_batch.h"
+#include "rocksdb/cache.h"
+#include "rocksdb/env.h"
+#include "rocksdb/write_batch.h"
 #include "db/db_impl.h"
 #include "db/filename.h"
 #include "db/log_format.h"
@@ -20,7 +25,7 @@
 #include "util/testharness.h"
 #include "util/testutil.h"

-namespace leveldb {
+namespace rocksdb {

 static const int kValueSize = 1000;

@@ -68,7 +73,7 @@ class CorruptionTest {
  void RepairDB() {
    delete db_;
    db_ = nullptr;
-    ASSERT_OK(::leveldb::RepairDB(dbname_, options_));
+    ASSERT_OK(::rocksdb::RepairDB(dbname_, options_));
  }

  void Build(int n) {
@@ -230,7 +235,7 @@ TEST(CorruptionTest, NewFileErrorDuringWrite) {
 TEST(CorruptionTest, TableFile) {
  Build(100);
  DBImpl* dbi = reinterpret_cast<DBImpl*>(db_);
-  dbi->TEST_CompactMemTable();
+  dbi->TEST_FlushMemTable();
  dbi->TEST_CompactRange(0, nullptr, nullptr);
  dbi->TEST_CompactRange(1, nullptr, nullptr);

@@ -241,7 +246,7 @@ TEST(CorruptionTest, TableFile) {
 TEST(CorruptionTest, TableFileIndexData) {
  Build(10000);  // Enough to build multiple Tables
  DBImpl* dbi = reinterpret_cast<DBImpl*>(db_);
-  dbi->TEST_CompactMemTable();
+  dbi->TEST_FlushMemTable();

  Corrupt(kTableFile, -2000, 500);
  Reopen();
@@ -279,7 +284,7 @@ TEST(CorruptionTest, SequenceNumberRecovery) {
 TEST(CorruptionTest, CorruptedDescriptor) {
  ASSERT_OK(db_->Put(WriteOptions(), "foo", "hello"));
  DBImpl* dbi = reinterpret_cast<DBImpl*>(db_);
-  dbi->TEST_CompactMemTable();
+  dbi->TEST_FlushMemTable();
  dbi->TEST_CompactRange(0, nullptr, nullptr);

  Corrupt(kDescriptorFile, 0, 1000);
@@ -296,9 +301,9 @@ TEST(CorruptionTest, CorruptedDescriptor) {
 TEST(CorruptionTest, CompactionInputError) {
  Build(10);
  DBImpl* dbi = reinterpret_cast<DBImpl*>(db_);
-  dbi->TEST_CompactMemTable();
+  dbi->TEST_FlushMemTable();
  const int last = dbi->MaxMemCompactionLevel();
-  ASSERT_EQ(1, Property("leveldb.num-files-at-level" + NumberToString(last)));
+  ASSERT_EQ(1, Property("rocksdb.num-files-at-level" + NumberToString(last)));

  Corrupt(kTableFile, 100, 1);
  Check(9, 9);
@@ -319,12 +324,13 @@ TEST(CorruptionTest, CompactionInputErrorParanoid) {
  for (int level = 1; level < dbi->NumberLevels(); level++) {
    dbi->Put(WriteOptions(), "", "begin");
    dbi->Put(WriteOptions(), "~", "end");
-    dbi->TEST_CompactMemTable();
+    dbi->TEST_FlushMemTable();
  }

  Build(10);
-  dbi->TEST_CompactMemTable();
-  ASSERT_EQ(1, Property("leveldb.num-files-at-level0"));
+  dbi->TEST_FlushMemTable();
+  dbi->TEST_WaitForCompact();
+  ASSERT_EQ(1, Property("rocksdb.num-files-at-level0"));

  Corrupt(kTableFile, 100, 1);
  Check(9, 9);
@@ -341,7 +347,7 @@ TEST(CorruptionTest, CompactionInputErrorParanoid) {
 TEST(CorruptionTest, UnrelatedKeys) {
  Build(10);
  DBImpl* dbi = reinterpret_cast<DBImpl*>(db_);
-  dbi->TEST_CompactMemTable();
+  dbi->TEST_FlushMemTable();
  Corrupt(kTableFile, 100, 1);

  std::string tmp1, tmp2;
@@ -349,13 +355,13 @@ TEST(CorruptionTest, UnrelatedKeys) {
  std::string v;
  ASSERT_OK(db_->Get(ReadOptions(), Key(1000, &tmp1), &v));
  ASSERT_EQ(Value(1000, &tmp2).ToString(), v);
-  dbi->TEST_CompactMemTable();
+  dbi->TEST_FlushMemTable();
  ASSERT_OK(db_->Get(ReadOptions(), Key(1000, &tmp1), &v));
  ASSERT_EQ(Value(1000, &tmp2).ToString(), v);
 }

-}  // namespace leveldb
+}  // namespace rocksdb

 int main(int argc, char** argv) {
-  return leveldb::test::RunAllTests();
+  return rocksdb::test::RunAllTests();
 }
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2012 Facebook.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
@@ -8,12 +13,12 @@
 #include "db/db_impl.h"
 #include "db/filename.h"
 #include "db/version_set.h"
-#include "leveldb/db.h"
-#include "leveldb/env.h"
+#include "rocksdb/db.h"
+#include "rocksdb/env.h"
 #include "port/port.h"
 #include "util/mutexlock.h"

-namespace leveldb {
+namespace rocksdb {

 Status DBImpl::DisableFileDeletions() {
  MutexLock l(&mutex_);
@@ -30,16 +35,19 @@ Status DBImpl::EnableFileDeletions() {
 }

 Status DBImpl::GetLiveFiles(std::vector<std::string>& ret,
-                            uint64_t* manifest_file_size) {
+                            uint64_t* manifest_file_size,
+                            bool flush_memtable) {

  *manifest_file_size = 0;

-  // flush all dirty data to disk.
-  Status status =  Flush(FlushOptions());
-  if (!status.ok()) {
-    Log(options_.info_log, "Cannot Flush data %s\n",
-        status.ToString().c_str());
-    return status;
+  if (flush_memtable) {
+    // flush all dirty data to disk.
+    Status status =  Flush(FlushOptions());
+    if (!status.ok()) {
+      Log(options_.info_log, "Cannot Flush data %s\n",
+          status.ToString().c_str());
+      return status;
+    }
  }

  MutexLock l(&mutex_);
@@ -48,18 +56,17 @@ Status DBImpl::GetLiveFiles(std::vector<std::string>& ret,
  std::set<uint64_t> live;
  versions_->AddLiveFilesCurrentVersion(&live);

-  ret.resize(live.size() + 2); //*.sst + CURRENT + MANIFEST
+  ret.clear();
+  ret.reserve(live.size() + 2); //*.sst + CURRENT + MANIFEST

  // create names of the live files. The names are not absolute
  // paths, instead they are relative to dbname_;
-  std::set<uint64_t>::iterator it = live.begin();
-  for (unsigned int i = 0; i < live.size(); i++, it++) {
-    ret[i] = TableFileName("", *it);
+  for (auto live_file : live) {
+    ret.push_back(TableFileName("", live_file));
  }

-  ret[live.size()] = CurrentFileName("");
-  ret[live.size()+1] = DescriptorFileName("",
-                                          versions_->ManifestFileNumber());
+  ret.push_back(CurrentFileName(""));
+  ret.push_back(DescriptorFileName("", versions_->ManifestFileNumber()));

  // find length of manifest file while holding the mutex lock
  *manifest_file_size = versions_->ManifestFileSize();
@@ -71,9 +78,9 @@ Status DBImpl::GetSortedWalFiles(VectorLogPtr& files) {
  // First get sorted files in archive dir, then append sorted files from main
  // dir to maintain sorted order

-  //  list wal files in archive dir.
+  // list wal files in archive dir.
  Status s;
-  std::string archivedir = ArchivalDirectory(dbname_);
+  std::string archivedir = ArchivalDirectory(options_.wal_dir);
  if (env_->FileExists(archivedir)) {
    s = AppendSortedWalsOfType(archivedir, files, kArchivedLogFile);
    if (!s.ok()) {
@@ -81,26 +88,25 @@ Status DBImpl::GetSortedWalFiles(VectorLogPtr& files) {
    }
  }
  // list wal files in main db dir.
-  s = AppendSortedWalsOfType(dbname_, files, kAliveLogFile);
-  if (!s.ok()) {
-    return s;
-  }
-  return s;
+  return AppendSortedWalsOfType(options_.wal_dir, files, kAliveLogFile);
 }

 Status DBImpl::DeleteWalFiles(const VectorLogPtr& files) {
  Status s;
-  std::string archivedir = ArchivalDirectory(dbname_);
+  std::string archivedir = ArchivalDirectory(options_.wal_dir);
  std::string files_not_deleted;
  for (const auto& wal : files) {
-    /* Try deleting in archive dir. If fails, try deleting in main db dir.
-     * This is efficient because all except for very few wal files will be in
-     * archive. Checking for WalType is not much helpful because alive wal could
-       be archived now.
+    /* Try deleting in the dir that pathname points to for the logfile.
+       This may fail if we try to delete a log file which was live when captured
+       but is archived now. Try deleting it from archive also
     */
-    if (!env_->DeleteFile(archivedir + "/" + wal->Filename()).ok() &&
-        !env_->DeleteFile(dbname_ + "/" + wal->Filename()).ok()) {
-      files_not_deleted.append(wal->Filename());
+    Status st = env_->DeleteFile(options_.wal_dir + "/" + wal->PathName());
+    if (!st.ok()) {
+      if (wal->Type() == kAliveLogFile &&
+          env_->DeleteFile(LogFileName(archivedir, wal->LogNumber())).ok()) {
+        continue;
+      }
+      files_not_deleted.append(wal->PathName());
    }
  }
  if (!files_not_deleted.empty()) {
@@ -1,10 +1,12 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
-
-#ifndef STORAGE_LEVELDB_DB_DB_IMPL_H_
-#define STORAGE_LEVELDB_DB_DB_IMPL_H_
-
+#pragma once
 #include <atomic>
 #include <deque>
 #include <set>
@@ -12,19 +14,15 @@
 #include "db/dbformat.h"
 #include "db/log_writer.h"
 #include "db/snapshot.h"
-#include "leveldb/db.h"
-#include "leveldb/env.h"
-#include "leveldb/memtablerep.h"
-#include "leveldb/transaction_log.h"
+#include "rocksdb/db.h"
+#include "rocksdb/env.h"
+#include "rocksdb/memtablerep.h"
+#include "rocksdb/transaction_log.h"
 #include "port/port.h"
 #include "util/stats_logger.h"
 #include "memtablelist.h"

-#ifdef USE_SCRIBE
-#include "scribe/scribe_logger.h"
-#endif
-
-namespace leveldb {
+namespace rocksdb {

 class MemTable;
 class TableCache;
@@ -64,7 +62,7 @@ class DBImpl : public DB {
  virtual bool GetProperty(const Slice& property, std::string* value);
  virtual void GetApproximateSizes(const Range* range, int n, uint64_t* sizes);
  virtual void CompactRange(const Slice* begin, const Slice* end,
-                            bool reduce_level = false);
+                            bool reduce_level = false, int target_level = -1);
  virtual int NumberLevels();
  virtual int MaxMemCompactionLevel();
  virtual int Level0StopWriteTrigger();
@@ -72,23 +70,28 @@ class DBImpl : public DB {
  virtual Status DisableFileDeletions();
  virtual Status EnableFileDeletions();
  virtual Status GetLiveFiles(std::vector<std::string>&,
-                              uint64_t* manifest_file_size);
+                              uint64_t* manifest_file_size,
+                              bool flush_memtable = true);
  virtual Status GetSortedWalFiles(VectorLogPtr& files);
  virtual Status DeleteWalFiles(const VectorLogPtr& files);
  virtual SequenceNumber GetLatestSequenceNumber();
  virtual Status GetUpdatesSince(SequenceNumber seq_number,
                                 unique_ptr<TransactionLogIterator>* iter);
+  virtual Status DeleteFile(std::string name);
+
+  virtual void GetLiveFilesMetaData(
+    std::vector<LiveFileMetaData> *metadata);

  // Extra methods (for testing) that are not in the public DB interface

  // Compact any files in the named level that overlap [*begin, *end]
  void TEST_CompactRange(int level, const Slice* begin, const Slice* end);

-  // Force current memtable contents to be compacted.
-  Status TEST_CompactMemTable();
+  // Force current memtable contents to be flushed.
+  Status TEST_FlushMemTable();

  // Wait for memtable compaction
-  Status TEST_WaitForCompactMemTable();
+  Status TEST_WaitForFlushMemTable();

  // Wait for any compaction
  Status TEST_WaitForCompact();
@@ -111,6 +114,9 @@ class DBImpl : public DB {
  // Trigger's a background call for testing.
  void TEST_PurgeObsoleteteWAL();

+  // get total level0 file size. Only for testing.
+  uint64_t TEST_GetLevel0TotalSize() { return versions_->NumLevelBytes(0);}
+
 protected:
  Env* const env_;
  const std::string dbname_;
@@ -150,9 +156,9 @@ class DBImpl : public DB {
  // Delete any unneeded files and stale in-memory entries.
  void DeleteObsoleteFiles();

-  // Compact the in-memory write buffer to disk.  Switches to a new
+  // Flush the in-memory write buffer to storage.  Switches to a new
  // log-file/memtable and writes a new descriptor iff successful.
-  Status CompactMemTable(bool* madeProgress = nullptr);
+  Status FlushMemTableToOutputFile(bool* madeProgress = nullptr);

  Status RecoverLogFile(uint64_t log_number,
                        VersionEdit* edit,
@@ -175,17 +181,20 @@ class DBImpl : public DB {
  // Force current memtable contents to be flushed.
  Status FlushMemTable(const FlushOptions& options);

-  // Wait for memtable compaction
-  Status WaitForCompactMemTable();
+  // Wait for memtable flushed
+  Status WaitForFlushMemTable();

  void MaybeScheduleLogDBDeployStats();
  static void BGLogDBDeployStats(void* db);
  void LogDBDeployStats();

-  void MaybeScheduleCompaction();
-  static void BGWork(void* db);
-  void BackgroundCall();
+  void MaybeScheduleFlushOrCompaction();
+  static void BGWorkCompaction(void* db);
+  static void BGWorkFlush(void* db);
+  void BackgroundCallCompaction();
+  void BackgroundCallFlush();
  Status BackgroundCompaction(bool* madeProgress,DeletionState& deletion_state);
+  Status BackgroundFlush(bool* madeProgress);
  void CleanupCompaction(CompactionState* compact);
  Status DoCompactionWork(CompactionState* compact);

@@ -208,6 +217,8 @@ class DBImpl : public DB {
  // Removes the file listed in files_to_evict from the table_cache
  void EvictObsoleteFiles(DeletionState& deletion_state);

+  Status DeleteLogFile(uint64_t number);
+
  void PurgeObsoleteWALFiles();

  Status AppendSortedWalsOfType(const std::string& path,
@@ -230,16 +241,17 @@ class DBImpl : public DB {

  void PrintStatistics();

-  // dump leveldb.stats to LOG
+  // dump rocksdb.stats to LOG
  void MaybeDumpStats();

  // Return the minimum empty level that could hold the total data in the
  // input level. Return the input level, if such level could not be found.
  int FindMinimumEmptyLevelFitting(int level);

-  // Move the files in the input level to the minimum level that could hold
-  // the data set.
-  void ReFitLevel(int level);
+  // Move the files in the input level to the target level.
+  // If target_level < 0, automatically calculate the minimum level that could
+  // hold the data set.
+  void ReFitLevel(int level, int target_level = -1);

  // Constant after construction
  const InternalFilterPolicy internal_filter_policy_;
@@ -276,6 +288,9 @@ class DBImpl : public DB {
  // count how many background compaction been scheduled or is running?
  int bg_compaction_scheduled_;

+  // number of background memtable flush jobs, submitted to the HIGH pool
+  int bg_flush_scheduled_;
+
  // Has a background stats log thread scheduled?
  bool bg_logstats_scheduled_;

@@ -420,7 +435,6 @@ class DBImpl : public DB {
  Status GetImpl(const ReadOptions& options,
                 const Slice& key,
                 std::string* value,
-                 const bool no_io = false,
                 bool* value_found = nullptr);
 };

@@ -431,6 +445,4 @@ extern Options SanitizeOptions(const std::string& db,
                               const InternalFilterPolicy* ipolicy,
                               const Options& src);

-}  // namespace leveldb
-
-#endif  // STORAGE_LEVELDB_DB_DB_IMPL_H_
+}  // namespace rocksdb
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2012 Facebook. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
@@ -21,10 +26,10 @@
 #include "db/table_cache.h"
 #include "db/version_set.h"
 #include "db/write_batch_internal.h"
-#include "leveldb/db.h"
-#include "leveldb/env.h"
-#include "leveldb/status.h"
-#include "leveldb/table_builder.h"
+#include "rocksdb/db.h"
+#include "rocksdb/env.h"
+#include "rocksdb/status.h"
+#include "rocksdb/table_builder.h"
 #include "port/port.h"
 #include "table/block.h"
 #include "table/merger.h"
@@ -34,7 +39,7 @@
 #include "util/logging.h"
 #include "util/build_version.h"

-namespace leveldb {
+namespace rocksdb {

 DBImplReadOnly::DBImplReadOnly(const Options& options,
    const std::string& dbname)
@@ -1,10 +1,13 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2012 Facebook. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.

-#ifndef STORAGE_LEVELDB_DB_DB_IMPL_READONLY_H_
-#define STORAGE_LEVELDB_DB_DB_IMPL_READONLY_H_
-
+#pragma once
 #include "db/db_impl.h"

 #include <deque>
@@ -12,16 +15,12 @@
 #include "db/dbformat.h"
 #include "db/log_writer.h"
 #include "db/snapshot.h"
-#include "leveldb/db.h"
-#include "leveldb/env.h"
+#include "rocksdb/db.h"
+#include "rocksdb/env.h"
 #include "port/port.h"
 #include "util/stats_logger.h"

-#ifdef USE_SCRIBE
-#include "scribe/scribe_logger.h"
-#endif
-
-namespace leveldb {
+namespace rocksdb {

 class DBImplReadOnly : public DBImpl {
 public:
@@ -51,7 +50,7 @@ public:
   return Status::NotSupported("Not supported operation in read only mode.");
 }
 virtual void CompactRange(const Slice* begin, const Slice* end,
-                           bool reduce_level = false) {
+                           bool reduce_level = false, int target_level = -1) {
 }
 virtual Status DisableFileDeletions() {
   return Status::NotSupported("Not supported operation in read only mode.");
@@ -60,7 +59,8 @@ public:
   return Status::NotSupported("Not supported operation in read only mode.");
 }
 virtual Status GetLiveFiles(std::vector<std::string>&,
-                             uint64_t* manifest_file_size) {
+                             uint64_t* manifest_file_size,
+                             bool flush_memtable = true) {
   return Status::NotSupported("Not supported operation in read only mode.");
 }
 virtual Status Flush(const FlushOptions& options) {
@@ -76,5 +76,3 @@ private:
 };

 }
-
-#endif
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -8,15 +13,16 @@

 #include "db/filename.h"
 #include "db/dbformat.h"
-#include "leveldb/env.h"
-#include "leveldb/options.h"
-#include "leveldb/iterator.h"
-#include "leveldb/merge_operator.h"
+#include "rocksdb/env.h"
+#include "rocksdb/options.h"
+#include "rocksdb/iterator.h"
+#include "rocksdb/merge_operator.h"
 #include "port/port.h"
 #include "util/logging.h"
 #include "util/mutexlock.h"
+#include "util/perf_context_imp.h"

-namespace leveldb {
+namespace rocksdb {

 #if 0
 static void DumpInternalIter(Iterator* iter) {
@@ -65,6 +71,7 @@ class DBIter: public Iterator {
        current_entry_is_merged_(false),
        statistics_(options.statistics) {
    RecordTick(statistics_, NO_ITERATORS, 1);
+    max_skip_ = options.max_sequential_skip_in_iterations;
  }
  virtual ~DBIter() {
    RecordTick(statistics_, NO_ITERATORS, -1);
@@ -129,6 +136,7 @@ class DBIter: public Iterator {
  bool valid_;
  bool current_entry_is_merged_;
  std::shared_ptr<Statistics> statistics_;
+  uint64_t max_skip_;

  // No copying allowed
  DBIter(const DBIter&);
@@ -188,12 +196,14 @@ void DBIter::FindNextUserEntry(bool skipping) {
  assert(iter_->Valid());
  assert(direction_ == kForward);
  current_entry_is_merged_ = false;
+  uint64_t num_skipped = 0;
  do {
    ParsedInternalKey ikey;
    if (ParseKey(&ikey) && ikey.sequence <= sequence_) {
      if (skipping &&
          user_comparator_->Compare(ikey.user_key, saved_key_) <= 0) {
-        // skip this entry
+        num_skipped++; // skip this entry
+        BumpPerfCount(&perf_context.internal_key_skipped_count);
      } else {
        skipping = false;
        switch (ikey.type) {
@@ -202,6 +212,8 @@ void DBIter::FindNextUserEntry(bool skipping) {
            // they are hidden by this deletion.
            SaveKey(ikey.user_key, &saved_key_);
            skipping = true;
+            num_skipped = 0;
+            BumpPerfCount(&perf_context.internal_delete_skipped_count);
            break;
          case kTypeValue:
            valid_ = true;
@@ -220,7 +232,20 @@ void DBIter::FindNextUserEntry(bool skipping) {
        }
      }
    }
-    iter_->Next();
+    // If we have sequentially iterated via numerous keys and still not
+    // found the next user-key, then it is better to seek so that we can
+    // avoid too many key comparisons. We seek to the last occurence of
+    // our current key by looking for sequence number 0.
+    if (skipping && num_skipped > max_skip_) {
+      num_skipped = 0;
+      std::string last_key;
+      AppendInternalKey(&last_key,
+        ParsedInternalKey(Slice(saved_key_), 0, kValueTypeForSeek));
+      iter_->Seek(last_key);
+      RecordTick(statistics_, NUMBER_OF_RESEEKS_IN_ITERATION);
+    } else {
+      iter_->Next();
+    }
  } while (iter_->Valid());
  valid_ = false;
 }
@@ -342,11 +367,13 @@ void DBIter::Prev() {

 void DBIter::FindPrevUserEntry() {
  assert(direction_ == kReverse);
+  uint64_t num_skipped = 0;

  ValueType value_type = kTypeDeletion;
  if (iter_->Valid()) {
    do {
      ParsedInternalKey ikey;
+      bool saved_key_cleared = false;
      if (ParseKey(&ikey) && ikey.sequence <= sequence_) {
        if ((value_type != kTypeDeletion) &&
            user_comparator_->Compare(ikey.user_key, saved_key_) < 0) {
@@ -357,6 +384,7 @@ void DBIter::FindPrevUserEntry() {
        if (value_type == kTypeDeletion) {
          saved_key_.clear();
          ClearSavedValue();
+          saved_key_cleared = true;
        } else {
          Slice raw_value = iter_->value();
          if (saved_value_.capacity() > raw_value.size() + 1048576) {
@@ -367,7 +395,22 @@ void DBIter::FindPrevUserEntry() {
          saved_value_.assign(raw_value.data(), raw_value.size());
        }
      }
-      iter_->Prev();
+      num_skipped++;
+      // If we have sequentially iterated via numerous keys and still not
+      // found the prev user-key, then it is better to seek so that we can
+      // avoid too many key comparisons. We seek to the first occurence of
+      // our current key by looking for max sequence number.
+      if (!saved_key_cleared && num_skipped > max_skip_) {
+        num_skipped = 0;
+        std::string last_key;
+        AppendInternalKey(&last_key,
+          ParsedInternalKey(Slice(saved_key_), kMaxSequenceNumber,
+                            kValueTypeForSeek));
+        iter_->Seek(last_key);
+        RecordTick(statistics_, NUMBER_OF_RESEEKS_IN_ITERATION);
+      } else {
+        iter_->Prev();
+      }
    } while (iter_->Valid());
  }

@@ -435,4 +478,4 @@ Iterator* NewDBIterator(
                    internal_iter, sequence);
 }

-}  // namespace leveldb
+}  // namespace rocksdb
@@ -1,15 +1,18 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.

-#ifndef STORAGE_LEVELDB_DB_DB_ITER_H_
-#define STORAGE_LEVELDB_DB_DB_ITER_H_
-
+#pragma once
 #include <stdint.h>
-#include "leveldb/db.h"
+#include "rocksdb/db.h"
 #include "db/dbformat.h"

-namespace leveldb {
+namespace rocksdb {

 // Return a new iterator that converts internal keys (yielded by
 // "*internal_iter") that were live at the specified "sequence" number
@@ -22,6 +25,4 @@ extern Iterator* NewDBIterator(
    Iterator* internal_iter,
    const SequenceNumber& sequence);

-}  // namespace leveldb
-
-#endif  // STORAGE_LEVELDB_DB_DB_ITER_H_
+}  // namespace rocksdb
@@ -1,22 +1,25 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.

-#ifndef LEVELDB_STORAGE_DB_DB_STATISTICS_H_
-#define LEVELDB_STORAGE_DB_DB_STATISTICS_H_
-
+#pragma once
 #include <cassert>
 #include <stdlib.h>
 #include <vector>
 #include <memory>

-#include "leveldb/statistics.h"
+#include "rocksdb/statistics.h"
 #include "util/histogram.h"
 #include "port/port.h"
 #include "util/mutexlock.h"


-namespace leveldb {
+namespace rocksdb {

 class DBStatistics: public Statistics {
 public:
@@ -59,8 +62,4 @@ std::shared_ptr<Statistics> CreateDBStatistics() {
  return std::make_shared<DBStatistics>();
 }

-} // namespace leveldb
-
-#endif // LEVELDB_STORAGE_DB_DB_STATISTICS_H_
-
-
+} // namespace rocksdb
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -7,12 +12,12 @@
 #include <stdint.h>
 #include <stdio.h>
 #include "db/version_set.h"
-#include "leveldb/db.h"
-#include "leveldb/env.h"
+#include "rocksdb/db.h"
+#include "rocksdb/env.h"
 #include "port/port.h"
 #include "util/mutexlock.h"

-namespace leveldb {
+namespace rocksdb {

 void DBImpl::MaybeScheduleLogDBDeployStats() {

@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -6,9 +11,9 @@
 #include "db/dbformat.h"
 #include "port/port.h"
 #include "util/coding.h"
-#include "include/leveldb/perf_context.h"
+#include "util/perf_context_imp.h"

-namespace leveldb {
+namespace rocksdb {

 static uint64_t PackSequenceAndType(uint64_t seq, ValueType t) {
  assert(seq <= kMaxSequenceNumber);
@@ -54,7 +59,7 @@ int InternalKeyComparator::Compare(const Slice& akey, const Slice& bkey) const {
  //    decreasing sequence number
  //    decreasing type (though sequence# should be enough to disambiguate)
  int r = user_comparator_->Compare(ExtractUserKey(akey), ExtractUserKey(bkey));
-  perf_context.user_key_comparison_count++;
+  BumpPerfCount(&perf_context.user_key_comparison_count);
  if (r == 0) {
    const uint64_t anum = DecodeFixed64(akey.data() + akey.size() - 8);
    const uint64_t bnum = DecodeFixed64(bkey.data() + bkey.size() - 8);
@@ -139,4 +144,4 @@ LookupKey::LookupKey(const Slice& user_key, SequenceNumber s) {
  end_ = dst;
 }

-}  // namespace leveldb
+}  // namespace rocksdb
@@ -1,21 +1,24 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.

-#ifndef STORAGE_LEVELDB_DB_FORMAT_H_
-#define STORAGE_LEVELDB_DB_FORMAT_H_
-
+#pragma once
 #include <stdio.h>
-#include "leveldb/comparator.h"
-#include "leveldb/db.h"
-#include "leveldb/filter_policy.h"
-#include "leveldb/slice.h"
-#include "leveldb/table_builder.h"
-#include "leveldb/types.h"
+#include "rocksdb/comparator.h"
+#include "rocksdb/db.h"
+#include "rocksdb/filter_policy.h"
+#include "rocksdb/slice.h"
+#include "rocksdb/table_builder.h"
+#include "rocksdb/types.h"
 #include "util/coding.h"
 #include "util/logging.h"

-namespace leveldb {
+namespace rocksdb {

 class InternalKey;

@@ -90,7 +93,7 @@ class InternalKeyComparator : public Comparator {
  std::string name_;
 public:
  explicit InternalKeyComparator(const Comparator* c) : user_comparator_(c),
-    name_("leveldb.InternalKeyComparator:" +
+    name_("rocksdb.InternalKeyComparator:" +
          std::string(user_comparator_->Name())) {
  }

@@ -223,6 +226,4 @@ inline LookupKey::~LookupKey() {
  if (start_ != space_) delete[] start_;
 }

-}  // namespace leveldb
-
-#endif  // STORAGE_LEVELDB_DB_FORMAT_H_
+}  // namespace rocksdb
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -6,7 +11,7 @@
 #include "util/logging.h"
 #include "util/testharness.h"

-namespace leveldb {
+namespace rocksdb {

 static std::string IKey(const std::string& user_key,
                        uint64_t seq,
@@ -105,8 +110,8 @@ TEST(FormatTest, InternalKeyShortestSuccessor) {
            ShortSuccessor(IKey("\xff\xff", 100, kTypeValue)));
 }

-}  // namespace leveldb
+}  // namespace rocksdb

 int main(int argc, char** argv) {
-  return leveldb::test::RunAllTests();
+  return rocksdb::test::RunAllTests();
 }
@@ -0,0 +1,192 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
+// Copyright (c) 2011 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 "rocksdb/db.h"
+#include "db/db_impl.h"
+#include "db/filename.h"
+#include "db/version_set.h"
+#include "db/write_batch_internal.h"
+#include "util/testharness.h"
+#include "util/testutil.h"
+#include "rocksdb/env.h"
+#include <vector>
+#include <stdlib.h>
+#include <map>
+#include <string>
+
+namespace rocksdb {
+
+class DeleteFileTest {
+ public:
+  std::string dbname_;
+  Options options_;
+  DB* db_;
+  Env* env_;
+  int numlevels_;
+
+  DeleteFileTest() {
+    db_ = nullptr;
+    env_ = Env::Default();
+    options_.write_buffer_size = 1024*1024*1000;
+    options_.target_file_size_base = 1024*1024*1000;
+    options_.max_bytes_for_level_base = 1024*1024*1000;
+    dbname_ = test::TmpDir() + "/deletefile_test";
+    DestroyDB(dbname_, options_);
+    numlevels_ = 7;
+    ASSERT_OK(ReopenDB(true));
+  }
+
+  Status ReopenDB(bool create) {
+    delete db_;
+    if (create) {
+      DestroyDB(dbname_, options_);
+    }
+    db_ = nullptr;
+    options_.create_if_missing = create;
+    return DB::Open(options_, dbname_, &db_);
+  }
+
+  void CloseDB() {
+    delete db_;
+  }
+
+  void AddKeys(int numkeys, int startkey = 0) {
+    WriteOptions options;
+    options.sync = false;
+    ReadOptions roptions;
+    for (int i = startkey; i < (numkeys + startkey) ; i++) {
+      std::string temp = std::to_string(i);
+      Slice key(temp);
+      Slice value(temp);
+      ASSERT_OK(db_->Put(options, key, value));
+    }
+  }
+
+  int numKeysInLevels(
+    std::vector<LiveFileMetaData> &metadata,
+    std::vector<int> *keysperlevel = nullptr) {
+
+    if (keysperlevel != nullptr) {
+      keysperlevel->resize(numlevels_);
+    }
+
+    int numKeys = 0;
+    for (size_t i = 0; i < metadata.size(); i++) {
+      int startkey = atoi(metadata[i].smallestkey.c_str());
+      int endkey = atoi(metadata[i].largestkey.c_str());
+      int numkeysinfile = (endkey - startkey + 1);
+      numKeys += numkeysinfile;
+      if (keysperlevel != nullptr) {
+        (*keysperlevel)[(int)metadata[i].level] += numkeysinfile;
+      }
+      fprintf(stderr, "level %d name %s smallest %s largest %s\n",
+              metadata[i].level, metadata[i].name.c_str(),
+              metadata[i].smallestkey.c_str(),
+              metadata[i].largestkey.c_str());
+    }
+    return numKeys;
+  }
+
+  void CreateTwoLevels() {
+    AddKeys(50000, 10000);
+    DBImpl* dbi = reinterpret_cast<DBImpl*>(db_);
+    ASSERT_OK(dbi->TEST_FlushMemTable());
+    ASSERT_OK(dbi->TEST_WaitForFlushMemTable());
+
+    AddKeys(50000, 10000);
+    ASSERT_OK(dbi->TEST_FlushMemTable());
+    ASSERT_OK(dbi->TEST_WaitForFlushMemTable());
+  }
+
+};
+
+TEST(DeleteFileTest, AddKeysAndQueryLevels) {
+  CreateTwoLevels();
+  std::vector<LiveFileMetaData> metadata;
+  std::vector<int> keysinlevel;
+  db_->GetLiveFilesMetaData(&metadata);
+
+  std::string level1file = "";
+  int level1keycount = 0;
+  std::string level2file = "";
+  int level2keycount = 0;
+  int level1index = 0;
+  int level2index = 1;
+
+  ASSERT_EQ((int)metadata.size(), 2);
+  if (metadata[0].level == 2) {
+    level1index = 1;
+    level2index = 0;
+  }
+
+  level1file = metadata[level1index].name;
+  int startkey = atoi(metadata[level1index].smallestkey.c_str());
+  int endkey = atoi(metadata[level1index].largestkey.c_str());
+  level1keycount = (endkey - startkey + 1);
+  level2file = metadata[level2index].name;
+  startkey = atoi(metadata[level2index].smallestkey.c_str());
+  endkey = atoi(metadata[level2index].largestkey.c_str());
+  level2keycount = (endkey - startkey + 1);
+
+  // COntrolled setup. Levels 1 and 2 should both have 50K files.
+  // This is a little fragile as it depends on the current
+  // compaction heuristics.
+  ASSERT_EQ(level1keycount, 50000);
+  ASSERT_EQ(level2keycount, 50000);
+
+  Status status = db_->DeleteFile("0.sst");
+  ASSERT_TRUE(status.IsInvalidArgument());
+
+  // intermediate level files cannot be deleted.
+  status = db_->DeleteFile(level1file);
+  ASSERT_TRUE(status.IsInvalidArgument());
+
+  // Lowest level file deletion should succeed.
+  ASSERT_OK(db_->DeleteFile(level2file));
+
+  CloseDB();
+}
+
+
+TEST(DeleteFileTest, DeleteFileWithIterator) {
+  CreateTwoLevels();
+  ReadOptions options;
+  Iterator* it = db_->NewIterator(options);
+  std::vector<LiveFileMetaData> metadata;
+  db_->GetLiveFilesMetaData(&metadata);
+
+  std::string level2file = "";
+
+  ASSERT_EQ((int)metadata.size(), 2);
+  if (metadata[0].level == 1) {
+    level2file = metadata[1].name;
+  } else {
+    level2file = metadata[0].name;
+  }
+
+  Status status = db_->DeleteFile(level2file);
+  fprintf(stdout, "Deletion status %s: %s\n",
+          level2file.c_str(), status.ToString().c_str());
+  ASSERT_TRUE(status.ok());
+  it->SeekToFirst();
+  int numKeysIterated = 0;
+  while(it->Valid()) {
+    numKeysIterated++;
+    it->Next();
+  }
+  ASSERT_EQ(numKeysIterated, 50000);
+  delete it;
+  CloseDB();
+}
+} //namespace rocksdb
+
+int main(int argc, char** argv) {
+  return rocksdb::test::RunAllTests();
+}
+
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -7,10 +12,10 @@
 #include <ctype.h>
 #include <stdio.h>
 #include "db/dbformat.h"
-#include "leveldb/env.h"
+#include "rocksdb/env.h"
 #include "util/logging.h"

-namespace leveldb {
+namespace rocksdb {

 // Given a path, flatten the path name by replacing all chars not in
 // {[0-9,a-z,A-Z,-,_,.]} with _. And append '\0' at the end.
@@ -46,13 +51,10 @@ extern Status WriteStringToFileSync(Env* env, const Slice& data,
 static std::string MakeFileName(const std::string& name, uint64_t number,
                                const char* suffix) {
  char buf[100];
-  snprintf(buf, sizeof(buf), "%06llu.%s",
+  snprintf(buf, sizeof(buf), "/%06llu.%s",
           static_cast<unsigned long long>(number),
           suffix);
-  if (name.empty()) {
-    return buf;
-  }
-  return name + "/" + buf;
+  return name + buf;
 }

 std::string LogFileName(const std::string& name, uint64_t number) {
@@ -60,12 +62,12 @@ std::string LogFileName(const std::string& name, uint64_t number) {
  return MakeFileName(name, number, "log");
 }

-std::string ArchivalDirectory(const std::string& dbname) {
-  return dbname + "/" + ARCHIVAL_DIR;
+std::string ArchivalDirectory(const std::string& dir) {
+  return dir + "/" + ARCHIVAL_DIR;
 }
 std::string ArchivedLogFileName(const std::string& name, uint64_t number) {
  assert(number > 0);
-  return MakeFileName(name + "/archive", number, "log");
+  return MakeFileName(name + "/" + ARCHIVAL_DIR, number, "log");
 }

 std::string TableFileName(const std::string& name, uint64_t number) {
@@ -125,7 +127,12 @@ std::string MetaDatabaseName(const std::string& dbname, uint64_t number) {
  return dbname + buf;
 }

+std::string IdentityFileName(const std::string& dbname) {
+  return dbname + "/IDENTITY";
+}
+
 // Owned filenames have the form:
+//    dbname/IDENTITY
 //    dbname/CURRENT
 //    dbname/LOCK
 //    dbname/LOG
@@ -133,11 +140,18 @@ std::string MetaDatabaseName(const std::string& dbname, uint64_t number) {
 //    dbname/MANIFEST-[0-9]+
 //    dbname/[0-9]+.(log|sst)
 //    dbname/METADB-[0-9]+
+//    Disregards / at the beginning
 bool ParseFileName(const std::string& fname,
                   uint64_t* number,
                   FileType* type) {
  Slice rest(fname);
-  if (rest == "CURRENT") {
+  if (fname.length() > 1 && fname[0] == '/') {
+    rest.remove_prefix(1);
+  }
+  if (rest == "IDENTITY") {
+    *number = 0;
+    *type = kIdentityFile;
+  } else if (rest == "CURRENT") {
    *number = 0;
    *type = kCurrentFile;
  } else if (rest == "LOCK") {
@@ -217,4 +231,18 @@ Status SetCurrentFile(Env* env, const std::string& dbname,
  return s;
 }

-}  // namespace leveldb
+Status SetIdentityFile(Env* env, const std::string& dbname) {
+  std::string id = env->GenerateUniqueId();
+  assert(!id.empty());
+  std::string tmp = TempFileName(dbname, id.size());
+  Status s = WriteStringToFileSync(env, id, tmp);
+  if (s.ok()) {
+    s = env->RenameFile(tmp, IdentityFileName(dbname));
+  }
+  if (!s.ok()) {
+    env->DeleteFile(tmp);
+  }
+  return s;
+}
+
+}  // namespace rocksdb
@@ -1,19 +1,22 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
 //
 // File names used by DB code

-#ifndef STORAGE_LEVELDB_DB_FILENAME_H_
-#define STORAGE_LEVELDB_DB_FILENAME_H_
-
+#pragma once
 #include <stdint.h>
 #include <string>
-#include "leveldb/slice.h"
-#include "leveldb/status.h"
+#include "rocksdb/slice.h"
+#include "rocksdb/status.h"
 #include "port/port.h"

-namespace leveldb {
+namespace rocksdb {

 class Env;

@@ -25,7 +28,8 @@ enum FileType {
  kCurrentFile,
  kTempFile,
  kInfoLogFile,  // Either the current one, or an old one
-  kMetaDatabase
+  kMetaDatabase,
+  kIdentityFile
 };

 // Return the name of the log file with the specified number
@@ -79,7 +83,12 @@ extern std::string OldInfoLogFileName(const std::string& dbname, uint64_t ts,
 extern std::string MetaDatabaseName(const std::string& dbname,
                                    uint64_t number);

-// If filename is a leveldb file, store the type of the file in *type.
+// Return the name of the Identity file which stores a unique number for the db
+// that will get regenerated if the db loses all its data and is recreated fresh
+// either from a backup-image or empty
+extern std::string IdentityFileName(const std::string& dbname);
+
+// If filename is a rocksdb file, store the type of the file in *type.
 // The number encoded in the filename is stored in *number.  If the
 // filename was successfully parsed, returns true.  Else return false.
 extern bool ParseFileName(const std::string& filename,
@@ -91,7 +100,7 @@ extern bool ParseFileName(const std::string& filename,
 extern Status SetCurrentFile(Env* env, const std::string& dbname,
                             uint64_t descriptor_number);

+// Make the IDENTITY file for the db
+extern Status SetIdentityFile(Env* env, const std::string& dbname);

-}  // namespace leveldb
-
-#endif  // STORAGE_LEVELDB_DB_FILENAME_H_
+}  // namespace rocksdb
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -9,7 +14,7 @@
 #include "util/logging.h"
 #include "util/testharness.h"

-namespace leveldb {
+namespace rocksdb {

 class FileNameTest { };

@@ -128,8 +133,8 @@ TEST(FileNameTest, Construction) {
  ASSERT_EQ(kMetaDatabase, type);
 }

-}  // namespace leveldb
+}  // namespace rocksdb

 int main(int argc, char** argv) {
-  return leveldb::test::RunAllTests();
+  return rocksdb::test::RunAllTests();
 }
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -5,10 +10,8 @@
 // Log format information shared by reader and writer.
 // See ../doc/log_format.txt for more detail.

-#ifndef STORAGE_LEVELDB_DB_LOG_FORMAT_H_
-#define STORAGE_LEVELDB_DB_LOG_FORMAT_H_
-
-namespace leveldb {
+#pragma once
+namespace rocksdb {
 namespace log {

 enum RecordType {
@@ -30,6 +33,4 @@ static const unsigned int kBlockSize = 32768;
 static const int kHeaderSize = 4 + 1 + 2;

 }  // namespace log
-}  // namespace leveldb
-
-#endif  // STORAGE_LEVELDB_DB_LOG_FORMAT_H_
+}  // namespace rocksdb
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -5,11 +10,11 @@
 #include "db/log_reader.h"

 #include <stdio.h>
-#include "leveldb/env.h"
+#include "rocksdb/env.h"
 #include "util/coding.h"
 #include "util/crc32c.h"

-namespace leveldb {
+namespace rocksdb {
 namespace log {

 Reader::Reporter::~Reporter() {
@@ -256,4 +261,4 @@ unsigned int Reader::ReadPhysicalRecord(Slice* result) {
 }

 }  // namespace log
-}  // namespace leveldb
+}  // namespace rocksdb
@@ -1,18 +1,21 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.

-#ifndef STORAGE_LEVELDB_DB_LOG_READER_H_
-#define STORAGE_LEVELDB_DB_LOG_READER_H_
-
+#pragma once
 #include <memory>
 #include <stdint.h>

 #include "db/log_format.h"
-#include "leveldb/slice.h"
-#include "leveldb/status.h"
+#include "rocksdb/slice.h"
+#include "rocksdb/status.h"

-namespace leveldb {
+namespace rocksdb {

 class SequentialFile;
 using std::unique_ptr;
@@ -118,6 +121,4 @@ class Reader {
 };

 }  // namespace log
-}  // namespace leveldb
-
-#endif  // STORAGE_LEVELDB_DB_LOG_READER_H_
+}  // namespace rocksdb
@@ -1,16 +1,21 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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/log_reader.h"
 #include "db/log_writer.h"
-#include "leveldb/env.h"
+#include "rocksdb/env.h"
 #include "util/coding.h"
 #include "util/crc32c.h"
 #include "util/random.h"
 #include "util/testharness.h"

-namespace leveldb {
+namespace rocksdb {
 namespace log {

 // Construct a string of the specified length made out of the supplied
@@ -516,8 +521,8 @@ TEST(LogTest, ReadPastEnd) {
 }

 }  // namespace log
-}  // namespace leveldb
+}  // namespace rocksdb

 int main(int argc, char** argv) {
-  return leveldb::test::RunAllTests();
+  return rocksdb::test::RunAllTests();
 }
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -5,11 +10,11 @@
 #include "db/log_writer.h"

 #include <stdint.h>
-#include "leveldb/env.h"
+#include "rocksdb/env.h"
 #include "util/coding.h"
 #include "util/crc32c.h"

-namespace leveldb {
+namespace rocksdb {
 namespace log {

 Writer::Writer(unique_ptr<WritableFile>&& dest)
@@ -100,4 +105,4 @@ Status Writer::EmitPhysicalRecord(RecordType t, const char* ptr, size_t n) {
 }

 }  // namespace log
-}  // namespace leveldb
+}  // namespace rocksdb
@@ -1,17 +1,20 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.

-#ifndef STORAGE_LEVELDB_DB_LOG_WRITER_H_
-#define STORAGE_LEVELDB_DB_LOG_WRITER_H_
-
+#pragma once
 #include <memory>
 #include <stdint.h>
 #include "db/log_format.h"
-#include "leveldb/slice.h"
-#include "leveldb/status.h"
+#include "rocksdb/slice.h"
+#include "rocksdb/status.h"

-namespace leveldb {
+namespace rocksdb {

 class WritableFile;

@@ -49,6 +52,4 @@ class Writer {
 };

 }  // namespace log
-}  // namespace leveldb
-
-#endif  // STORAGE_LEVELDB_DB_LOG_WRITER_H_
+}  // namespace rocksdb
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -7,20 +12,14 @@
 #include <memory>

 #include "db/dbformat.h"
-#include "leveldb/comparator.h"
-#include "leveldb/env.h"
-#include "leveldb/iterator.h"
-#include "leveldb/merge_operator.h"
+#include "rocksdb/comparator.h"
+#include "rocksdb/env.h"
+#include "rocksdb/iterator.h"
+#include "rocksdb/merge_operator.h"
 #include "util/coding.h"
+#include "util/murmurhash.h"

-namespace leveldb {
-
-static Slice GetLengthPrefixedSlice(const char* data) {
-  uint32_t len;
-  const char* p = data;
-  p = GetVarint32Ptr(p, p + 5, &len);  // +5: we assume "p" is not corrupted
-  return Slice(p, len);
-}
+namespace rocksdb {

 MemTable::MemTable(const InternalKeyComparator& cmp,
                   std::shared_ptr<MemTableRepFactory> table_factory,
@@ -35,6 +34,7 @@ MemTable::MemTable(const InternalKeyComparator& cmp,
      file_number_(0),
      edit_(numlevel),
      first_seqno_(0),
+      mem_next_logfile_number_(0),
      mem_logfile_number_(0) { }

 MemTable::~MemTable() {
@@ -42,7 +42,8 @@ MemTable::~MemTable() {
 }

 size_t MemTable::ApproximateMemoryUsage() {
-  return arena_impl_.ApproximateMemoryUsage();
+  return arena_impl_.ApproximateMemoryUsage() +
+    table_->ApproximateMemoryUsage();
 }

 int MemTable::KeyComparator::operator()(const char* aptr, const char* bptr)
@@ -53,6 +54,11 @@ int MemTable::KeyComparator::operator()(const char* aptr, const char* bptr)
  return comparator.Compare(a, b);
 }

+Slice MemTableRep::UserKey(const char* key) const {
+  Slice slice = GetLengthPrefixedSlice(key);
+  return Slice(slice.data(), slice.size() - 8);
+}
+
 // Encode a suitable internal key target for "target" and return it.
 // Uses *scratch as scratch space, and the returned pointer will point
 // into this scratch space.
@@ -68,6 +74,9 @@ class MemTableIterator: public Iterator {
  explicit MemTableIterator(MemTableRep* table)
    : iter_(table->GetIterator()) { }

+  MemTableIterator(MemTableRep* table, const Slice* prefix)
+    : iter_(table->GetPrefixIterator(*prefix)) { }
+
  virtual bool Valid() const { return iter_->Valid(); }
  virtual void Seek(const Slice& k) { iter_->Seek(EncodeKey(&tmp_, k)); }
  virtual void SeekToFirst() { iter_->SeekToFirst(); }
@@ -93,8 +102,12 @@ class MemTableIterator: public Iterator {
  void operator=(const MemTableIterator&);
 };

-Iterator* MemTable::NewIterator() {
-  return new MemTableIterator(table_.get());
+Iterator* MemTable::NewIterator(const Slice* prefix) {
+  if (prefix) {
+    return new MemTableIterator(table_.get(), prefix);
+  } else {
+    return new MemTableIterator(table_.get());
+  }
 }

 void MemTable::Add(SequenceNumber s, ValueType type,
@@ -132,7 +145,8 @@ void MemTable::Add(SequenceNumber s, ValueType type,
 bool MemTable::Get(const LookupKey& key, std::string* value, Status* s,
                   std::deque<std::string>* operands, const Options& options) {
  Slice memkey = key.memtable_key();
-  std::shared_ptr<MemTableRep::Iterator> iter(table_.get()->GetIterator());
+  std::shared_ptr<MemTableRep::Iterator> iter(
+    table_->GetIterator(key.user_key()));
  iter->Seek(memkey.data());

  // It is the caller's responsibility to allocate/delete operands list
@@ -229,4 +243,4 @@ bool MemTable::Get(const LookupKey& key, std::string* value, Status* s,
  return false;
 }

-}  // namespace leveldb
+}  // namespace rocksdb
@@ -1,21 +1,24 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.

-#ifndef STORAGE_LEVELDB_DB_MEMTABLE_H_
-#define STORAGE_LEVELDB_DB_MEMTABLE_H_
-
+#pragma once
 #include <string>
 #include <memory>
 #include <deque>
 #include "db/dbformat.h"
 #include "db/skiplist.h"
 #include "db/version_set.h"
-#include "leveldb/db.h"
-#include "leveldb/memtablerep.h"
+#include "rocksdb/db.h"
+#include "rocksdb/memtablerep.h"
 #include "util/arena_impl.h"

-namespace leveldb {
+namespace rocksdb {

 class Mutex;
 class MemTableIterator;
@@ -61,7 +64,11 @@ class MemTable {
  // while the returned iterator is live.  The keys returned by this
  // iterator are internal keys encoded by AppendInternalKey in the
  // db/dbformat.{h,cc} module.
-  Iterator* NewIterator();
+  //
+  // If a prefix is supplied, it is passed to the underlying MemTableRep as a
+  // hint that the iterator only need to support access to keys with that
+  // prefix.
+  Iterator* NewIterator(const Slice* prefix = nullptr);

  // Add an entry into memtable that maps key to value at the
  // specified sequence number and with the specified type.
@@ -88,6 +95,14 @@ class MemTable {
  // into the memtable
  SequenceNumber GetFirstSequenceNumber() { return first_seqno_; }

+  // Returns the next active logfile number when this memtable is about to
+  // be flushed to storage
+  uint64_t GetNextLogNumber() { return mem_next_logfile_number_; }
+
+  // Sets the next active logfile number when this memtable is about to
+  // be flushed to storage
+  void SetNextLogNumber(uint64_t num) { mem_next_logfile_number_ = num; }
+
  // Returns the logfile number that can be safely deleted when this
  // memstore is flushed to storage
  uint64_t GetLogNumber() { return mem_logfile_number_; }
@@ -96,6 +111,9 @@ class MemTable {
  // memstore is flushed to storage
  void SetLogNumber(uint64_t num) { mem_logfile_number_ = num; }

+  // Notify the underlying storage that no more items will be added
+  void MarkImmutable() { table_->MarkReadOnly(); }
+
 private:
  ~MemTable();  // Private since only Unref() should be used to delete it
  friend class MemTableIterator;
@@ -120,6 +138,10 @@ class MemTable {
  SequenceNumber first_seqno_;

  // The log files earlier than this number can be deleted.
+  uint64_t mem_next_logfile_number_;
+
+  // The log file that backs this memtable (to be deleted when
+  // memtable flush is done)
  uint64_t mem_logfile_number_;

  // No copying allowed
@@ -127,6 +149,4 @@ class MemTable {
  void operator=(const MemTable&);
 };

-}  // namespace leveldb
-
-#endif  // STORAGE_LEVELDB_DB_MEMTABLE_H_
+}  // namespace rocksdb
@@ -1,16 +1,18 @@
-// Copyright (c) 2012 Facebook.
-
-
+//  Copyright (c) 2013, 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 "db/memtablelist.h"

 #include <string>
-#include "leveldb/db.h"
+#include "rocksdb/db.h"
 #include "db/memtable.h"
-#include "leveldb/env.h"
-#include "leveldb/iterator.h"
+#include "rocksdb/env.h"
+#include "rocksdb/iterator.h"
 #include "util/coding.h"

-namespace leveldb {
+namespace rocksdb {

 class InternalKeyComparator;
 class Mutex;
@@ -42,7 +44,8 @@ int MemTableList::size() {
 // Returns true if there is at least one memtable on which flush has
 // not yet started.
 bool MemTableList::IsFlushPending(int min_write_buffer_number_to_merge) {
-  if (num_flush_not_started_ >= min_write_buffer_number_to_merge) {
+  if ((flush_requested_ && num_flush_not_started_ >= 1) ||
+      (num_flush_not_started_ >= min_write_buffer_number_to_merge)) {
    assert(imm_flush_needed.NoBarrier_Load() != nullptr);
    return true;
  }
@@ -63,6 +66,7 @@ void MemTableList::PickMemtablesToFlush(std::vector<MemTable*>* ret) {
      ret->push_back(m);
    }
  }
+  flush_requested_ = false; // start-flush request is complete
 }

 // Record a successful flush in the manifest file
@@ -172,6 +176,7 @@ void MemTableList::Add(MemTable* m) {
  assert(size_ >= num_flush_not_started_);
  size_++;
  memlist_.push_front(m);
+  m->MarkImmutable();
  num_flush_not_started_++;
  if (num_flush_not_started_ == 1) {
    imm_flush_needed.Release_Store((void *)1);
@@ -207,4 +212,4 @@ void MemTableList::GetMemTables(std::vector<MemTable*>* output) {
  }
 }

-}  // namespace leveldb
+}  // namespace rocksdb
@@ -1,17 +1,19 @@
-// Copyright (c) 2012 Facebook.
-
-#ifndef STORAGE_LEVELDB_DB_MEMTABLELIST_H_
-#define STORAGE_LEVELDB_DB_MEMTABLELIST_H_
+//  Copyright (c) 2013, 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.
+//

+#pragma once
 #include <string>
 #include <list>
 #include <deque>
-#include "leveldb/db.h"
+#include "rocksdb/db.h"
 #include "db/dbformat.h"
 #include "db/skiplist.h"
 #include "memtable.h"

-namespace leveldb {
+namespace rocksdb {

 class InternalKeyComparator;
 class Mutex;
@@ -29,7 +31,8 @@ class MemTableList {
 public:
  // A list of memtables.
  MemTableList() : size_(0), num_flush_not_started_(0),
-    commit_in_progress_(false) {
+    commit_in_progress_(false),
+    flush_requested_(false) {
    imm_flush_needed.Release_Store(nullptr);
  }
  ~MemTableList() {};
@@ -76,6 +79,9 @@ class MemTableList {
  // Returns the list of underlying memtables.
  void GetMemTables(std::vector<MemTable*>* list);

+  // Request a flush of all existing memtables to storage
+  void FlushRequested() { flush_requested_ = true; }
+
  // Copying allowed
  // MemTableList(const MemTableList&);
  // void operator=(const MemTableList&);
@@ -90,8 +96,9 @@ class MemTableList {
  // committing in progress
  bool commit_in_progress_;

+  // Requested a flush of all memtables to storage
+  bool flush_requested_;
+
 };

-}  // namespace leveldb
-
-#endif  // STORAGE_LEVELDB_DB_MEMTABLELIST_H_
+}  // namespace rocksdb
@@ -1,12 +1,17 @@
+//  Copyright (c) 2013, 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 "merge_helper.h"
 #include "db/dbformat.h"
-#include "leveldb/comparator.h"
-#include "leveldb/db.h"
-#include "leveldb/merge_operator.h"
+#include "rocksdb/comparator.h"
+#include "rocksdb/db.h"
+#include "rocksdb/merge_operator.h"
 #include <string>
 #include <stdio.h>

-namespace leveldb {
+namespace rocksdb {

 // PRE:  iter points to the first merge type entry
 // POST: iter points to the first entry beyond the merge process (or the end)
@@ -189,4 +194,4 @@ void MergeHelper::MergeUntil(Iterator* iter, SequenceNumber stop_before,
  }
 }

-} // namespace leveldb
+} // namespace rocksdb
@@ -1,13 +1,18 @@
+//  Copyright (c) 2013, 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.
+//
 #ifndef MERGE_HELPER_H
 #define MERGE_HELPER_H

 #include "db/dbformat.h"
-#include "leveldb/slice.h"
-#include "leveldb/statistics.h"
+#include "rocksdb/slice.h"
+#include "rocksdb/statistics.h"
 #include <string>
 #include <deque>

-namespace leveldb {
+namespace rocksdb {

 class Comparator;
 class Iterator;
@@ -93,6 +98,6 @@ class MergeHelper {
  bool success_;
 };

-} // namespace leveldb
+} // namespace rocksdb

 #endif
@@ -1,13 +1,15 @@
+//  Copyright (c) 2013, 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.
+//
 /**
 * Back-end implementation details specific to the Merge Operator.
- *
- * @author Deon Nicholas (dnicholas@fb.com)
- * Copyright 2013 Facebook
 */

-#include "leveldb/merge_operator.h"
+#include "rocksdb/merge_operator.h"

-namespace leveldb {
+namespace rocksdb {

 // Given a "real" merge from the library, call the user's
 // associative merge function one-by-one on each of the operands.
@@ -48,4 +50,4 @@ bool AssociativeMergeOperator::PartialMerge(
  return Merge(key, &left_operand, right_operand, new_value, logger);
 }

-} // namespace leveldb
+} // namespace rocksdb
@@ -1,12 +1,17 @@
+//  Copyright (c) 2013, 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 <assert.h>
 #include <memory>
 #include <iostream>

-#include "leveldb/cache.h"
-#include "leveldb/comparator.h"
-#include "leveldb/db.h"
-#include "leveldb/env.h"
-#include "leveldb/merge_operator.h"
+#include "rocksdb/cache.h"
+#include "rocksdb/comparator.h"
+#include "rocksdb/db.h"
+#include "rocksdb/env.h"
+#include "rocksdb/merge_operator.h"
 #include "db/dbformat.h"
 #include "db/db_impl.h"
 #include "utilities/merge_operators.h"
@@ -14,7 +19,7 @@
 #include "utilities/utility_db.h"

 using namespace std;
-using namespace leveldb;
+using namespace rocksdb;


 std::shared_ptr<DB> OpenDb(const string& dbname, const bool ttl = false) {
@@ -85,7 +90,7 @@ class Counters {
    }
  }

-  // mapped to a leveldb Delete
+  // mapped to a rocksdb Delete
  bool remove(const string& key) {
    auto s = db_->Delete(delete_option_, key);

@@ -97,7 +102,7 @@ class Counters {
    }
  }

-  // mapped to a leveldb Get
+  // mapped to a rocksdb Get
  bool get(const string& key, uint64_t *value) {
    string str;
    auto s = db_->Get(get_option_, key, &str);
@@ -159,7 +164,7 @@ class MergeBasedCounters : public Counters {
        merge_option_() {
  }

-  // mapped to a leveldb Merge operation
+  // mapped to a rocksdb Merge operation
  virtual bool add(const string& key, uint64_t value) override {
    char encoded[sizeof(uint64_t)];
    EncodeFixed64(encoded, value);
@@ -1,12 +0,0 @@
-#include "include/leveldb/perf_context.h"
-
-
-namespace leveldb {
-
-void PerfContext::Reset() {
-  user_key_comparison_count = 0;
-}
-
-__thread PerfContext perf_context;
-
-}
@@ -1,20 +1,57 @@
+//  Copyright (c) 2013, 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 <algorithm>
 #include <iostream>
+#include <vector>
+#include "/usr/include/valgrind/callgrind.h"

-#include "leveldb/db.h"
-#include "leveldb/perf_context.h"
+#include "rocksdb/db.h"
+#include "rocksdb/perf_context.h"
+#include "util/histogram.h"
+#include "util/stop_watch.h"
 #include "util/testharness.h"


-namespace leveldb {
+bool FLAGS_random_key = false;
+bool FLAGS_use_set_based_memetable = false;
+int FLAGS_total_keys = 100;
+int FLAGS_write_buffer_size = 1000000000;
+int FLAGS_max_write_buffer_number = 8;
+int FLAGS_min_write_buffer_number_to_merge = 7;

 // Path to the database on file system
-const std::string kDbName = test::TmpDir() + "/perf_context_test";
+const std::string kDbName = rocksdb::test::TmpDir() + "/perf_context_test";
+
+void SeekToFirst(rocksdb::Iterator* iter) {
+//  std::cout << "Press a key to continue:";
+//  std::string s;
+//  std::cin >> s;
+  iter->SeekToFirst();
+//  std::cout << "Press a key to continue:";
+//  std::string s2;
+//  std::cin >> s2;
+}
+
+namespace rocksdb {

 std::shared_ptr<DB> OpenDb() {
    DB* db;
    Options options;
    options.create_if_missing = true;
-    options.write_buffer_size = 1000000000;     // give it a big memtable
+    options.write_buffer_size = FLAGS_write_buffer_size;
+    options.max_write_buffer_number = FLAGS_max_write_buffer_number;
+    options.min_write_buffer_number_to_merge =
+      FLAGS_min_write_buffer_number_to_merge;
+
+    if (FLAGS_use_set_based_memetable) {
+      auto prefix_extractor = rocksdb::NewFixedPrefixTransform(0);
+      options.memtable_factory =
+        std::make_shared<rocksdb::PrefixHashRepFactory>(prefix_extractor);
+    }
+
    Status s = DB::Open(options, kDbName,  &db);
    ASSERT_OK(s);
    return std::shared_ptr<DB>(db);
@@ -22,10 +59,125 @@ std::shared_ptr<DB> OpenDb() {

 class PerfContextTest { };

-int kTotalKeys = 100;
+TEST(PerfContextTest, SeekIntoDeletion) {
+  DestroyDB(kDbName, Options());
+  auto db = OpenDb();
+  WriteOptions write_options;
+  ReadOptions read_options;

-TEST(PerfContextTest, KeyComparisonCount) {
+  for (int i = 0; i < FLAGS_total_keys; ++i) {
+    std::string key = "k" + std::to_string(i);
+    std::string value = "v" + std::to_string(i);

+    db->Put(write_options, key, value);
+  }
+
+  for (int i = 0; i < FLAGS_total_keys -1 ; ++i) {
+    std::string key = "k" + std::to_string(i);
+    db->Delete(write_options, key);
+  }
+
+  HistogramImpl hist_get;
+  HistogramImpl hist_get_time;
+  for (int i = 0; i < FLAGS_total_keys - 1; ++i) {
+    std::string key = "k" + std::to_string(i);
+    std::string value;
+
+    perf_context.Reset();
+    StopWatchNano timer(Env::Default(), true);
+    auto status = db->Get(read_options, key, &value);
+    auto elapsed_nanos = timer.ElapsedNanos();
+    ASSERT_TRUE(status.IsNotFound());
+    hist_get.Add(perf_context.user_key_comparison_count);
+    hist_get_time.Add(elapsed_nanos);
+  }
+
+  std::cout << "Get uesr key comparison: \n" << hist_get.ToString()
+            << "Get time: \n" << hist_get_time.ToString();
+
+  HistogramImpl hist_seek_to_first;
+  std::unique_ptr<Iterator> iter(db->NewIterator(read_options));
+
+  perf_context.Reset();
+  StopWatchNano timer(Env::Default(), true);
+  //CALLGRIND_ZERO_STATS;
+  SeekToFirst(iter.get());
+  //iter->SeekToFirst();
+  //CALLGRIND_DUMP_STATS;
+  hist_seek_to_first.Add(perf_context.user_key_comparison_count);
+  auto elapsed_nanos = timer.ElapsedNanos();
+
+  std::cout << "SeekToFirst uesr key comparison: \n" << hist_seek_to_first.ToString()
+            << "ikey skipped: " << perf_context.internal_key_skipped_count << "\n"
+            << "idelete skipped: " << perf_context.internal_delete_skipped_count << "\n"
+            << "elapsed: " << elapsed_nanos << "\n";
+
+  HistogramImpl hist_seek;
+  for (int i = 0; i < FLAGS_total_keys; ++i) {
+    std::unique_ptr<Iterator> iter(db->NewIterator(read_options));
+    std::string key = "k" + std::to_string(i);
+
+    perf_context.Reset();
+    StopWatchNano timer(Env::Default(), true);
+    iter->Seek(key);
+    auto elapsed_nanos = timer.ElapsedNanos();
+    hist_seek.Add(perf_context.user_key_comparison_count);
+    std::cout << "seek cmp: " << perf_context.user_key_comparison_count
+              << " ikey skipped " << perf_context.internal_key_skipped_count
+              << " idelete skipped " << perf_context.internal_delete_skipped_count
+              << " elapsed: " << elapsed_nanos << "ns\n";
+
+    perf_context.Reset();
+    ASSERT_TRUE(iter->Valid());
+    StopWatchNano timer2(Env::Default(), true);
+    iter->Next();
+    auto elapsed_nanos2 = timer2.ElapsedNanos();
+    std::cout << "next cmp: " << perf_context.user_key_comparison_count
+              << "elapsed: " << elapsed_nanos2 << "ns\n";
+  }
+
+  std::cout << "Seek uesr key comparison: \n" << hist_seek.ToString();
+}
+
+TEST(PerfContextTest, StopWatchNanoOverhead) {
+  // profile the timer cost by itself!
+  const int kTotalIterations = 1000000;
+  std::vector<uint64_t> timings(kTotalIterations);
+
+  StopWatchNano timer(Env::Default(), true);
+  for (auto& timing : timings) {
+    timing = timer.ElapsedNanos(true /* reset */);
+  }
+
+  HistogramImpl histogram;
+  for (const auto timing : timings) {
+    histogram.Add(timing);
+  }
+
+  std::cout << histogram.ToString();
+}
+
+TEST(PerfContextTest, StopWatchOverhead) {
+  // profile the timer cost by itself!
+  const int kTotalIterations = 1000000;
+  std::vector<uint64_t> timings(kTotalIterations);
+
+  StopWatch timer(Env::Default());
+  for (auto& timing : timings) {
+    timing = timer.ElapsedMicros();
+  }
+
+  HistogramImpl histogram;
+  uint64_t prev_timing = 0;
+  for (const auto timing : timings) {
+    histogram.Add(timing - prev_timing);
+    prev_timing = timing;
+  }
+
+  std::cout << histogram.ToString();
+}
+
+void ProfileKeyComparison() {
  DestroyDB(kDbName, Options());    // Start this test with a fresh DB

  auto db = OpenDb();
@@ -33,49 +185,141 @@ TEST(PerfContextTest, KeyComparisonCount) {
  WriteOptions write_options;
  ReadOptions read_options;

-  uint64_t total_user_key_comparison_get = 0;
-  uint64_t total_user_key_comparison_put = 0;
-  uint64_t max_user_key_comparison_get = 0;
+  HistogramImpl hist_put;
+  HistogramImpl hist_get;

-  std::cout << "Inserting " << kTotalKeys << " key/value pairs\n...\n";
+  std::cout << "Inserting " << FLAGS_total_keys << " key/value pairs\n...\n";

-  for (int i = 0; i < kTotalKeys; ++i) {
+  std::vector<int> keys;
+  for (int i = 0; i < FLAGS_total_keys; ++i) {
+    keys.push_back(i);
+  }
+
+  if (FLAGS_random_key) {
+    std::random_shuffle(keys.begin(), keys.end());
+  }
+
+  for (const int i : keys) {
    std::string key = "k" + std::to_string(i);
    std::string value = "v" + std::to_string(i);

    perf_context.Reset();
    db->Put(write_options, key, value);
-    total_user_key_comparison_put += perf_context.user_key_comparison_count;
+    hist_put.Add(perf_context.user_key_comparison_count);

    perf_context.Reset();
    db->Get(read_options, key, &value);
-    total_user_key_comparison_get += perf_context.user_key_comparison_count;
-    max_user_key_comparison_get =
-      std::max(max_user_key_comparison_get,
-               perf_context.user_key_comparison_count);
+    hist_get.Add(perf_context.user_key_comparison_count);
  }

-  std::cout << "total user key comparison get: "
-            << total_user_key_comparison_get << "\n"
-            << "total user key comparison put: "
-            << total_user_key_comparison_put << "\n"
-            << "max user key comparison get: "
-            << max_user_key_comparison_get << "\n"
-            << "avg user key comparison get:"
-            << total_user_key_comparison_get/kTotalKeys << "\n";
+  std::cout << "Put uesr key comparison: \n" << hist_put.ToString()
+            << "Get uesr key comparison: \n" << hist_get.ToString();

 }

+TEST(PerfContextTest, KeyComparisonCount) {
+  SetPerfLevel(kEnableCount);
+  ProfileKeyComparison();

+  SetPerfLevel(kDisable);
+  ProfileKeyComparison();
+
+  SetPerfLevel(kEnableTime);
+  ProfileKeyComparison();
 }

+// make perf_context_test
+// export ROCKSDB_TESTS=PerfContextTest.SeekKeyComparison
+// For one memtable:
+// ./perf_context_test --write_buffer_size=500000 --total_keys=10000
+// For two memtables:
+// ./perf_context_test --write_buffer_size=250000 --total_keys=10000
+// Specify --random_key=1 to shuffle the key before insertion
+// Results show that, for sequential insertion, worst-case Seek Key comparison
+// is close to the total number of keys (linear), when there is only one
+// memtable. When there are two memtables, even the avg Seek Key comparison
+// starts to become linear to the input size.
+
+TEST(PerfContextTest, SeekKeyComparison) {
+  DestroyDB(kDbName, Options());
+  auto db = OpenDb();
+  WriteOptions write_options;
+  ReadOptions read_options;
+
+  std::cout << "Inserting " << FLAGS_total_keys << " key/value pairs\n...\n";
+
+  std::vector<int> keys;
+  for (int i = 0; i < FLAGS_total_keys; ++i) {
+    keys.push_back(i);
+  }
+
+  if (FLAGS_random_key) {
+    std::random_shuffle(keys.begin(), keys.end());
+  }
+
+  for (const int i : keys) {
+    std::string key = "k" + std::to_string(i);
+    std::string value = "v" + std::to_string(i);
+
+    db->Put(write_options, key, value);
+  }
+
+  HistogramImpl hist_seek;
+  HistogramImpl hist_next;
+
+  for (int i = 0; i < FLAGS_total_keys; ++i) {
+    std::string key = "k" + std::to_string(i);
+    std::string value = "v" + std::to_string(i);
+
+    std::unique_ptr<Iterator> iter(db->NewIterator(read_options));
+    perf_context.Reset();
+    iter->Seek(key);
+    ASSERT_TRUE(iter->Valid());
+    ASSERT_EQ(iter->value().ToString(), value);
+    hist_seek.Add(perf_context.user_key_comparison_count);
+  }
+
+  std::unique_ptr<Iterator> iter(db->NewIterator(read_options));
+  for (iter->SeekToFirst(); iter->Valid();) {
+    perf_context.Reset();
+    iter->Next();
+    hist_next.Add(perf_context.user_key_comparison_count);
+  }
+
+  std::cout << "Seek:\n" << hist_seek.ToString()
+            << "Next:\n" << hist_next.ToString();
+}
+
+}

 int main(int argc, char** argv) {

-  if (argc > 1) {
-    leveldb::kTotalKeys = std::stoi(argv[1]);
+  for (int i = 1; i < argc; i++) {
+    int n;
+    char junk;
+
+    if (sscanf(argv[i], "--write_buffer_size=%d%c", &n, &junk) == 1) {
+      FLAGS_write_buffer_size = n;
+    }
+
+    if (sscanf(argv[i], "--total_keys=%d%c", &n, &junk) == 1) {
+      FLAGS_total_keys = n;
+    }
+
+    if (sscanf(argv[i], "--random_key=%d%c", &n, &junk) == 1 &&
+        (n == 0 || n == 1)) {
+      FLAGS_random_key = n;
+    }
+
+    if (sscanf(argv[i], "--use_set_based_memetable=%d%c", &n, &junk) == 1 &&
+        (n == 0 || n == 1)) {
+      FLAGS_use_set_based_memetable = n;
+    }
+
  }

-  leveldb::test::RunAllTests();
+  std::cout << kDbName << "\n";
+
+  rocksdb::test::RunAllTests();
  return 0;
 }
@@ -1,6 +1,7 @@
-// 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.
+//  Copyright (c) 2013, 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.
 //
 // Wrap an underlying iterator, but exclude any results not starting
 // with a given prefix.  Seeking to keys not beginning with the prefix
@@ -9,12 +10,10 @@
 // underlying iterator would if there happened to be no non-matching
 // keys in the dataset.

-#ifndef STORAGE_LEVELDB_DB_PREFIX_FILTER_ITERATOR_H_
-#define STORAGE_LEVELDB_DB_PREFIX_FILTER_ITERATOR_H_
+#pragma once
+#include "rocksdb/iterator.h"

-#include "leveldb/iterator.h"
-
-namespace leveldb {
+namespace rocksdb {

 class PrefixFilterIterator : public Iterator {
 private:
@@ -71,6 +70,4 @@ class PrefixFilterIterator : public Iterator {
  }
 };

-}  // namespace leveldb
-
-#endif
+}  // namespace rocksdb
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -34,11 +39,11 @@
 #include "db/table_cache.h"
 #include "db/version_edit.h"
 #include "db/write_batch_internal.h"
-#include "leveldb/comparator.h"
-#include "leveldb/db.h"
-#include "leveldb/env.h"
+#include "rocksdb/comparator.h"
+#include "rocksdb/db.h"
+#include "rocksdb/env.h"

-namespace leveldb {
+namespace rocksdb {

 namespace {

@@ -74,7 +79,7 @@ class Repairer {
        bytes += tables_[i].meta.file_size;
      }
      Log(options_.info_log,
-          "**** Repaired leveldb %s; "
+          "**** Repaired rocksdb %s; "
          "recovered %d files; %llu bytes. "
          "Some data may have been lost. "
          "****",
@@ -221,7 +226,7 @@ class Repairer {
    Iterator* iter = mem->NewIterator();
    status = BuildTable(dbname_, env_, options_, storage_options_,
                        table_cache_, iter, &meta,
-                        icmp_.user_comparator(), 0, 0);
+                        icmp_.user_comparator(), 0, 0, true);
    delete iter;
    mem->Unref();
    mem = nullptr;
@@ -383,4 +388,4 @@ Status RepairDB(const std::string& dbname, const Options& options) {
  return repairer.Run();
 }

-}  // namespace leveldb
+}  // namespace rocksdb
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -25,15 +30,13 @@
 // ... prev vs. next pointer ordering ...
 //

-#ifndef STORAGE_LEVELDB_DB_SKIPLIST_H_
-#define STORAGE_LEVELDB_DB_SKIPLIST_H_
-
+#pragma once
 #include <assert.h>
 #include <stdlib.h>
 #include "port/port.h"
 #include "util/random.h"

-namespace leveldb {
+namespace rocksdb {

 template<typename Key, class Comparator>
 class SkipList {
@@ -107,6 +110,7 @@ class SkipList {

  // Used for optimizing sequential insert patterns
  Node* prev_[kMaxHeight];
+  int   prev_height_;

  inline int GetMaxHeight() const {
    return static_cast<int>(
@@ -267,6 +271,10 @@ typename SkipList<Key,Comparator>::Node* SkipList<Key,Comparator>::FindGreaterOr
    Node* x = prev[0];
    Node* next = x->Next(0);
    if ((x == head_) || KeyIsAfterNode(key, x)) {
+      // Adjust all relevant insertion points to the previous entry
+      for (int i = 1; i < prev_height_; i++) {
+        prev[i] = x;
+      }
      return next;
    }
  }
@@ -275,6 +283,9 @@ typename SkipList<Key,Comparator>::Node* SkipList<Key,Comparator>::FindGreaterOr
  int level = GetMaxHeight() - 1;
  while (true) {
    Node* next = x->Next(level);
+    // Make sure the lists are sorted.
+    // If x points to head_ or next points nullptr, it is trivially satisfied.
+    assert((x == head_) || (next == nullptr) || KeyIsAfterNode(next->key, x));
    if (KeyIsAfterNode(key, next)) {
      // Keep searching in this list
      x = next;
@@ -337,6 +348,7 @@ SkipList<Key,Comparator>::SkipList(Comparator cmp, Arena* arena)
      arena_(arena),
      head_(NewNode(0 /* any key will do */, kMaxHeight)),
      max_height_(reinterpret_cast<void*>(1)),
+      prev_height_(1),
      rnd_(0xdeadbeef) {
  for (int i = 0; i < kMaxHeight; i++) {
    head_->SetNext(i, nullptr);
@@ -378,6 +390,7 @@ void SkipList<Key,Comparator>::Insert(const Key& key) {
    prev_[i]->SetNext(i, x);
  }
  prev_[0] = x;
+  prev_height_ = height;
 }

 template<typename Key, class Comparator>
@@ -390,6 +403,4 @@ bool SkipList<Key,Comparator>::Contains(const Key& key) const {
  }
 }

-}  // namespace leveldb
-
-#endif // STORAGE_LEVELDB_DB_SKIPLIST_H_
+}  // namespace rocksdb
@@ -1,16 +1,21 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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/skiplist.h"
 #include <set>
-#include "leveldb/env.h"
+#include "rocksdb/env.h"
 #include "util/arena_impl.h"
 #include "util/hash.h"
 #include "util/random.h"
 #include "util/testharness.h"

-namespace leveldb {
+namespace rocksdb {

 typedef uint64_t Key;

@@ -371,8 +376,8 @@ TEST(SkipTest, Concurrent3) { RunConcurrent(3); }
 TEST(SkipTest, Concurrent4) { RunConcurrent(4); }
 TEST(SkipTest, Concurrent5) { RunConcurrent(5); }

-}  // namespace leveldb
+}  // namespace rocksdb

 int main(int argc, char** argv) {
-  return leveldb::test::RunAllTests();
+  return rocksdb::test::RunAllTests();
 }
@@ -1,13 +1,16 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.

-#ifndef STORAGE_LEVELDB_DB_SNAPSHOT_H_
-#define STORAGE_LEVELDB_DB_SNAPSHOT_H_
+#pragma once
+#include "rocksdb/db.h"

-#include "leveldb/db.h"
-
-namespace leveldb {
+namespace rocksdb {

 class SnapshotList;

@@ -80,6 +83,4 @@ class SnapshotList {
  SnapshotImpl list_;
 };

-}  // namespace leveldb
-
-#endif  // STORAGE_LEVELDB_DB_SNAPSHOT_H_
+}  // namespace rocksdb
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -6,12 +11,12 @@

 #include "db/filename.h"

-#include "leveldb/statistics.h"
+#include "rocksdb/statistics.h"
 #include "table/table.h"
 #include "util/coding.h"
 #include "util/stop_watch.h"

-namespace leveldb {
+namespace rocksdb {

 static void DeleteEntry(const Slice& key, void* value) {
  Table* table = reinterpret_cast<Table*>(value);
@@ -32,7 +37,10 @@ TableCache::TableCache(const std::string& dbname,
      dbname_(dbname),
      options_(options),
      storage_options_(storage_options),
-      cache_(NewLRUCache(entries, options->table_cache_numshardbits)) {}
+      cache_(
+        NewLRUCache(entries, options->table_cache_numshardbits,
+                    options->table_cache_remove_scan_count_limit)) {
+}

 TableCache::~TableCache() {
 }
@@ -48,7 +56,7 @@ Status TableCache::FindTable(const EnvOptions& toptions,
  *handle = cache_->Lookup(key);
  if (*handle == nullptr) {
    if (no_io) { // Dont do IO and return a not-found status
-      return Status::NotFound("Table not found in table_cache, no_io is set");
+      return Status::Incomplete("Table not found in table_cache, no_io is set");
    }
    if (table_io != nullptr) {
      *table_io = true;    // we had to do IO from storage
@@ -90,7 +98,8 @@ Iterator* TableCache::NewIterator(const ReadOptions& options,
  }

  Cache::Handle* handle = nullptr;
-  Status s = FindTable(toptions, file_number, file_size, &handle);
+  Status s = FindTable(toptions, file_number, file_size, &handle,
+                       nullptr, options.read_tier == kBlockCacheTier);
  if (!s.ok()) {
    return NewErrorIterator(s);
  }
@@ -117,17 +126,17 @@ Status TableCache::Get(const ReadOptions& options,
                       void* arg,
                       bool (*saver)(void*, const Slice&, const Slice&, bool),
                       bool* table_io,
-                       void (*mark_key_may_exist)(void*),
-                       const bool no_io) {
+                       void (*mark_key_may_exist)(void*)) {
  Cache::Handle* handle = nullptr;
  Status s = FindTable(storage_options_, file_number, file_size,
-                       &handle, table_io, no_io);
+                       &handle, table_io,
+                       options.read_tier == kBlockCacheTier);
  if (s.ok()) {
    Table* t =
      reinterpret_cast<Table*>(cache_->Value(handle));
-    s = t->InternalGet(options, k, arg, saver, mark_key_may_exist, no_io);
+    s = t->InternalGet(options, k, arg, saver, mark_key_may_exist);
    cache_->Release(handle);
-  } else if (no_io && s.IsNotFound()) {
+  } else if (options.read_tier && s.IsIncomplete()) {
    // Couldnt find Table in cache but treat as kFound if no_io set
    (*mark_key_may_exist)(arg);
    return Status::OK();
@@ -135,10 +144,28 @@ Status TableCache::Get(const ReadOptions& options,
  return s;
 }

+bool TableCache::PrefixMayMatch(const ReadOptions& options,
+                                uint64_t file_number,
+                                uint64_t file_size,
+                                const Slice& internal_prefix,
+                                bool* table_io) {
+  Cache::Handle* handle = nullptr;
+  Status s = FindTable(storage_options_, file_number,
+                       file_size, &handle, table_io);
+  bool may_match = true;
+  if (s.ok()) {
+    Table* t =
+      reinterpret_cast<Table*>(cache_->Value(handle));
+    may_match = t->PrefixMayMatch(internal_prefix);
+    cache_->Release(handle);
+  }
+  return may_match;
+}
+
 void TableCache::Evict(uint64_t file_number) {
  char buf[sizeof(file_number)];
  EncodeFixed64(buf, file_number);
  cache_->Erase(Slice(buf, sizeof(buf)));
 }

-}  // namespace leveldb
+}  // namespace rocksdb
@@ -1,21 +1,24 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
 //
 // Thread-safe (provides internal synchronization)

-#ifndef STORAGE_LEVELDB_DB_TABLE_CACHE_H_
-#define STORAGE_LEVELDB_DB_TABLE_CACHE_H_
-
+#pragma once
 #include <string>
 #include <stdint.h>
 #include "db/dbformat.h"
-#include "leveldb/env.h"
-#include "leveldb/cache.h"
+#include "rocksdb/env.h"
+#include "rocksdb/cache.h"
 #include "port/port.h"
 #include "table/table.h"

-namespace leveldb {
+namespace rocksdb {

 class Env;

@@ -49,8 +52,13 @@ class TableCache {
             void* arg,
             bool (*handle_result)(void*, const Slice&, const Slice&, bool),
             bool* table_io,
-             void (*mark_key_may_exist)(void*) = nullptr,
-             const bool no_io = false);
+             void (*mark_key_may_exist)(void*) = nullptr);
+
+  // Determine whether the table may contain the specified prefix.  If
+  // the table index of blooms are not in memory, this may cause an I/O
+  bool PrefixMayMatch(const ReadOptions& options, uint64_t file_number,
+                      uint64_t file_size, const Slice& internal_prefix,
+                      bool* table_io);

  // Evict any entry for the specified file number
  void Evict(uint64_t file_number);
@@ -67,6 +75,4 @@ class TableCache {
                   const bool no_io = false);
 };

-}  // namespace leveldb
-
-#endif  // STORAGE_LEVELDB_DB_TABLE_CACHE_H_
+}  // namespace rocksdb
@@ -1,16 +1,21 @@
+//  Copyright (c) 2013, 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 "db/transaction_log_impl.h"
 #include "db/write_batch_internal.h"

-namespace leveldb {
+namespace rocksdb {

 TransactionLogIteratorImpl::TransactionLogIteratorImpl(
-                           const std::string& dbname,
+                           const std::string& dir,
                           const Options* options,
                           const EnvOptions& soptions,
                           const SequenceNumber seq,
                           std::unique_ptr<VectorLogPtr> files,
                           SequenceNumber const * const lastFlushedSequence) :
-    dbname_(dbname),
+    dir_(dir),
    options_(options),
    soptions_(soptions),
    startingSequenceNumber_(seq),
@@ -18,31 +23,34 @@ TransactionLogIteratorImpl::TransactionLogIteratorImpl(
    started_(false),
    isValid_(false),
    currentFileIndex_(0),
+    currentBatchSeq_(0),
+    currentBatchCount_(0),
    lastFlushedSequence_(lastFlushedSequence) {
  assert(startingSequenceNumber_ <= *lastFlushedSequence_);
-  assert(files_.get() != nullptr);
+  assert(files_ != nullptr);

  reporter_.env = options_->env;
  reporter_.info_log = options_->info_log.get();
+  SeekToStartSequence(); // Seek till starting sequence
 }

 Status TransactionLogIteratorImpl::OpenLogFile(
-  const LogFile* logFile,
-  unique_ptr<SequentialFile>* file) {
+    const LogFile* logFile,
+    unique_ptr<SequentialFile>* file) {
  Env* env = options_->env;
  if (logFile->Type() == kArchivedLogFile) {
-    std::string fname = ArchivedLogFileName(dbname_, logFile->LogNumber());
+    std::string fname = ArchivedLogFileName(dir_, logFile->LogNumber());
    return env->NewSequentialFile(fname, file, soptions_);
  } else {
-    std::string fname = LogFileName(dbname_, logFile->LogNumber());
+    std::string fname = LogFileName(dir_, logFile->LogNumber());
    Status status = env->NewSequentialFile(fname, file, soptions_);
    if (!status.ok()) {
      //  If cannot open file in DB directory.
      //  Try the archive dir, as it could have moved in the meanwhile.
-      fname = ArchivedLogFileName(dbname_, logFile->LogNumber());
+      fname = ArchivedLogFileName(dir_, logFile->LogNumber());
      status = env->NewSequentialFile(fname, file, soptions_);
      if (!status.ok()) {
-        return Status::IOError(" Requested file not present in the dir");
+        return Status::IOError("Requested file not present in the dir");
      }
    }
    return status;
@@ -52,7 +60,7 @@ Status TransactionLogIteratorImpl::OpenLogFile(
 BatchResult TransactionLogIteratorImpl::GetBatch()  {
  assert(isValid_);  //  cannot call in a non valid state.
  BatchResult result;
-  result.sequence = currentSequence_;
+  result.sequence = currentBatchSeq_;
  result.writeBatchPtr = std::move(currentBatch_);
  return result;
 }
@@ -65,92 +73,155 @@ bool TransactionLogIteratorImpl::Valid() {
  return started_ && isValid_;
 }

-void TransactionLogIteratorImpl::Next() {
-  LogFile* currentLogFile = files_.get()->at(currentFileIndex_).get();
+bool TransactionLogIteratorImpl::RestrictedRead(
+    Slice* record,
+    std::string* scratch) {
+  // Don't read if no more complete entries to read from logs
+  if (currentBatchSeq_ >= *lastFlushedSequence_) {
+    return false;
+  }
+  return currentLogReader_->ReadRecord(record, scratch);
+}

-//  First seek to the given seqNo. in the current file.
+void TransactionLogIteratorImpl::SeekToStartSequence(
+    uint64_t startFileIndex,
+    bool strict) {
  std::string scratch;
  Slice record;
-  if (!started_) {
-    started_ = true;  // this piece only runs onced.
-    isValid_ = false;
-    if (startingSequenceNumber_ > *lastFlushedSequence_) {
-      currentStatus_ = Status::IOError("Looking for a sequence, "
-                                        "which is not flushed yet.");
-      return;
+  started_ = false;
+  isValid_ = false;
+  if (startingSequenceNumber_ > *lastFlushedSequence_) {
+    currentStatus_ = Status::IOError("Looking for a sequence, "
+                                     "which is not flushed yet.");
+    return;
+  }
+  if (files_->size() <= startFileIndex) {
+    return;
+  }
+  Status s = OpenLogReader(files_->at(startFileIndex).get());
+  if (!s.ok()) {
+    currentStatus_ = s;
+    return;
+  }
+  while (RestrictedRead(&record, &scratch)) {
+    if (record.size() < 12) {
+      reporter_.Corruption(
+        record.size(), Status::Corruption("very small log record"));
+      continue;
    }
-    Status s = OpenLogReader(currentLogFile);
-    if (!s.ok()) {
-      currentStatus_ = s;
+    UpdateCurrentWriteBatch(record);
+    if (currentBatchSeq_ + currentBatchCount_ - 1 >=
+        startingSequenceNumber_) {
+      if (strict && currentBatchSeq_ != startingSequenceNumber_) {
+        currentStatus_ = Status::Corruption("Gap in sequence number. Could not "
+                                            "seek to required sequence number");
+        reporter_.Info(currentStatus_.ToString().c_str());
+        return;
+      } else if (strict) {
+        reporter_.Info("Could seek required sequence number. Iterator will "
+                       "continue.");
+      }
+      isValid_ = true;
+      started_ = true; // set started_ as we could seek till starting sequence
+      return;
+    } else {
      isValid_ = false;
-      return;
-    }
-    while (currentLogReader_->ReadRecord(&record, &scratch)) {
-      if (record.size() < 12) {
-        reporter_.Corruption(
-          record.size(), Status::Corruption("log record too small"));
-        continue;
-      }
-      UpdateCurrentWriteBatch(record);
-      if (currentSequence_ >= startingSequenceNumber_) {
-        assert(currentSequence_ <= *lastFlushedSequence_);
-        isValid_ = true;
-        break;
-      } else {
-        isValid_ = false;
-      }
-    }
-    if (isValid_) {
-      // Done for this iteration
-      return;
    }
  }
-  bool openNextFile = true;
-  while(openNextFile) {
+  // Could not find start sequence in first file. Normally this must be the
+  // only file. Otherwise log the error and let the iterator return next entry
+  if (files_->size() != 1) {
+    currentStatus_ = Status::Corruption("Start sequence was not found, "
+                                        "skipping to the next available");
+    reporter_.Corruption(0, currentStatus_);
+    started_ = true; // Let Next find next available entry
+    Next();
+  }
+}
+
+void TransactionLogIteratorImpl::Next() {
+  std::string scratch;
+  Slice record;
+  isValid_ = false;
+  if (!started_) {  // Runs every time until we can seek to the start sequence
+    return SeekToStartSequence();
+  }
+  while(true) {
    assert(currentLogReader_);
-    if (currentSequence_ < *lastFlushedSequence_) {
+    if (currentBatchSeq_ < *lastFlushedSequence_) {
      if (currentLogReader_->IsEOF()) {
        currentLogReader_->UnmarkEOF();
      }
      while (currentLogReader_->ReadRecord(&record, &scratch)) {
        if (record.size() < 12) {
          reporter_.Corruption(
-            record.size(), Status::Corruption("log record too small"));
+            record.size(), Status::Corruption("very small log record"));
          continue;
        } else {
-          UpdateCurrentWriteBatch(record);
-          openNextFile = false;
-          break;
+          return UpdateCurrentWriteBatch(record);
        }
      }
    }

-    if (openNextFile) {
-      if (currentFileIndex_ < files_.get()->size() - 1) {
-        ++currentFileIndex_;
-        Status status =OpenLogReader(files_.get()->at(currentFileIndex_).get());
-        if (!status.ok()) {
-          isValid_ = false;
-          currentStatus_ = status;
-          return;
-        }
-      } else {
+    // Open the next file
+    if (currentFileIndex_ < files_->size() - 1) {
+      ++currentFileIndex_;
+      Status status =OpenLogReader(files_->at(currentFileIndex_).get());
+      if (!status.ok()) {
        isValid_ = false;
-        openNextFile = false;
-        if (currentSequence_ == *lastFlushedSequence_) {
-          currentStatus_ = Status::OK();
-        } else {
-          currentStatus_ = Status::IOError(" NO MORE DATA LEFT");
-        }
+        currentStatus_ = status;
+        return;
      }
+    } else {
+      isValid_ = false;
+      if (currentBatchSeq_ == *lastFlushedSequence_) {
+        currentStatus_ = Status::OK();
+      } else {
+        currentStatus_ = Status::IOError(" NO MORE DATA LEFT");
+      }
+      return;
    }
  }
 }

+bool TransactionLogIteratorImpl::IsBatchContinuous(
+    const WriteBatch* batch,
+    const SequenceNumber expectedSeq) {
+  assert(batch);
+  SequenceNumber batchSeq = WriteBatchInternal::Sequence(batch);
+  if (started_ && batchSeq != expectedSeq) {
+    char buf[200];
+    snprintf(buf, sizeof(buf),
+             "Discontinuity in log records. Got seq=%lu, Expected seq=%lu, "
+             "Last flushed seq=%lu. Log iterator will seek the correct batch.",
+             batchSeq, expectedSeq, *lastFlushedSequence_);
+    reporter_.Info(buf);
+    return false;
+  }
+  return true;
+}
+
 void TransactionLogIteratorImpl::UpdateCurrentWriteBatch(const Slice& record) {
  WriteBatch* batch = new WriteBatch();
  WriteBatchInternal::SetContents(batch, record);
-  currentSequence_ = WriteBatchInternal::Sequence(batch);
+
+  SequenceNumber expectedSeq = currentBatchSeq_ + currentBatchCount_;
+  if (!IsBatchContinuous(batch, expectedSeq)) {
+    // Seek to the batch having expected sequence number
+    if (expectedSeq < files_->at(currentFileIndex_)->StartSequence()) {
+      // Expected batch must lie in the previous log file
+      currentFileIndex_--;
+      currentFileIndex_ = (currentFileIndex_ >= 0) ? currentFileIndex_ : 0;
+    }
+    startingSequenceNumber_ = expectedSeq;
+    return SeekToStartSequence(currentFileIndex_, true);
+  }
+
+  currentBatchSeq_ = WriteBatchInternal::Sequence(batch);
+  currentBatchCount_ = WriteBatchInternal::Count(batch);
+  // currentBatchSeq_ can only change here
+  assert(currentBatchSeq_ <= *lastFlushedSequence_);
+
  currentBatch_.reset(batch);
  isValid_ = true;
  currentStatus_ = Status::OK();
@@ -168,4 +239,4 @@ Status TransactionLogIteratorImpl::OpenLogReader(const LogFile* logFile) {
  );
  return Status::OK();
 }
-}  //  namespace leveldb
+}  //  namespace rocksdb
@@ -1,17 +1,19 @@
-// Copyright 2008-present Facebook. All Rights Reserved.
-#ifndef STORAGE_LEVELDB_INCLUDE_WRITES_ITERATOR_IMPL_H_
-#define STORAGE_LEVELDB_INCLUDE_WRITES_ITERATOR_IMPL_H_
-
+//  Copyright (c) 2013, 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.
+//
+#pragma once
 #include <vector>

-#include "leveldb/env.h"
-#include "leveldb/options.h"
-#include "leveldb/types.h"
-#include "leveldb/transaction_log.h"
+#include "rocksdb/env.h"
+#include "rocksdb/options.h"
+#include "rocksdb/types.h"
+#include "rocksdb/transaction_log.h"
 #include "db/log_reader.h"
 #include "db/filename.h"

-namespace leveldb {
+namespace rocksdb {

 struct LogReporter : public log::Reader::Reporter {
  Env* env;
@@ -19,6 +21,9 @@ struct LogReporter : public log::Reader::Reporter {
  virtual void Corruption(size_t bytes, const Status& s) {
    Log(info_log, "dropping %zu bytes; %s", bytes, s.ToString().c_str());
  }
+  virtual void Info(const char* s) {
+    Log(info_log, "%s", s);
+  }
 };

 class LogFileImpl : public LogFile {
@@ -31,7 +36,12 @@ class LogFileImpl : public LogFile {
    sizeFileBytes_(sizeBytes) {
  }

-  std::string Filename() const { return LogFileName("", logNumber_); }
+  std::string PathName() const {
+    if (type_ == kArchivedLogFile) {
+      return ArchivedLogFileName("", logNumber_);
+    }
+    return LogFileName("", logNumber_);
+  }

  uint64_t LogNumber() const { return logNumber_; }

@@ -55,7 +65,7 @@ class LogFileImpl : public LogFile {

 class TransactionLogIteratorImpl : public TransactionLogIterator {
 public:
-  TransactionLogIteratorImpl(const std::string& dbname,
+  TransactionLogIteratorImpl(const std::string& dir,
                             const Options* options,
                             const EnvOptions& soptions,
                             const SequenceNumber seqNum,
@@ -71,10 +81,10 @@ class TransactionLogIteratorImpl : public TransactionLogIterator {
  virtual BatchResult GetBatch();

 private:
-  const std::string& dbname_;
+  const std::string& dir_;
  const Options* options_;
  const EnvOptions& soptions_;
-  const SequenceNumber startingSequenceNumber_;
+  SequenceNumber startingSequenceNumber_;
  std::unique_ptr<VectorLogPtr> files_;
  bool started_;
  bool isValid_;  // not valid when it starts of.
@@ -84,15 +94,19 @@ class TransactionLogIteratorImpl : public TransactionLogIterator {
  unique_ptr<log::Reader> currentLogReader_;
  Status OpenLogFile(const LogFile* logFile, unique_ptr<SequentialFile>* file);
  LogReporter reporter_;
+  SequenceNumber currentBatchSeq_; // sequence number at start of current batch
+  uint64_t currentBatchCount_; // count in current batch
  SequenceNumber const * const lastFlushedSequence_;
-  // represents the sequence number being read currently.
-  SequenceNumber currentSequence_;

+  // Reads from transaction log only if the writebatch record has been written
+  bool RestrictedRead(Slice* record, std::string* scratch);
+  // Seeks to startingSequenceNumber reading from startFileIndex in files_.
+  // If strict is set,then must get a batch starting with startingSequenceNumber
+  void SeekToStartSequence(uint64_t startFileIndex = 0, bool strict = false);
+  // Check if batch is continuous starting from expectedSeq, else return false
+  bool IsBatchContinuous(const WriteBatch* batch, SequenceNumber expectedSeq);
+  // Update current batch if a continuous batch is found, else return false
  void UpdateCurrentWriteBatch(const Slice& record);
  Status OpenLogReader(const LogFile* file);
 };
-
-
-
-}  //  namespace leveldb
-#endif  //  STORAGE_LEVELDB_INCLUDE_WRITES_ITERATOR_IMPL_H_
+}  //  namespace rocksdb
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -7,7 +12,7 @@
 #include "db/version_set.h"
 #include "util/coding.h"

-namespace leveldb {
+namespace rocksdb {

 // Tag numbers for serialized VersionEdit.  These numbers are written to
 // disk and should not be changed.
@@ -293,4 +298,4 @@ std::string VersionEdit::DebugString(bool hex_key) const {
  return r;
 }

-}  // namespace leveldb
+}  // namespace rocksdb
@@ -1,16 +1,19 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.

-#ifndef STORAGE_LEVELDB_DB_VERSION_EDIT_H_
-#define STORAGE_LEVELDB_DB_VERSION_EDIT_H_
-
+#pragma once
 #include <set>
 #include <utility>
 #include <vector>
 #include "db/dbformat.h"

-namespace leveldb {
+namespace rocksdb {

 class VersionSet;

@@ -122,6 +125,4 @@ class VersionEdit {
  std::vector< std::pair<int, FileMetaData> > new_files_;
 };

-}  // namespace leveldb
-
-#endif  // STORAGE_LEVELDB_DB_VERSION_EDIT_H_
+}  // namespace rocksdb
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -5,7 +10,7 @@
 #include "db/version_edit.h"
 #include "util/testharness.h"

-namespace leveldb {
+namespace rocksdb {

 static void TestEncodeDecode(const VersionEdit& edit) {
  std::string encoded, encoded2;
@@ -41,8 +46,8 @@ TEST(VersionEditTest, EncodeDecode) {
  TestEncodeDecode(edit);
 }

-}  // namespace leveldb
+}  // namespace rocksdb

 int main(int argc, char** argv) {
-  return leveldb::test::RunAllTests();
+  return rocksdb::test::RunAllTests();
 }
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -12,16 +17,16 @@
 #include "db/log_writer.h"
 #include "db/memtable.h"
 #include "db/table_cache.h"
-#include "leveldb/env.h"
-#include "leveldb/merge_operator.h"
-#include "leveldb/table_builder.h"
+#include "rocksdb/env.h"
+#include "rocksdb/merge_operator.h"
+#include "rocksdb/table_builder.h"
 #include "table/merger.h"
 #include "table/two_level_iterator.h"
 #include "util/coding.h"
 #include "util/logging.h"
 #include "util/stop_watch.h"

-namespace leveldb {
+namespace rocksdb {

 static uint64_t TotalFileSize(const std::vector<FileMetaData*>& files) {
  uint64_t sum = 0;
@@ -189,7 +194,14 @@ static Iterator* GetFileIterator(void* arg,
    return NewErrorIterator(
        Status::Corruption("FileReader invoked with unexpected value"));
  } else {
-    return cache->NewIterator(options,
+    ReadOptions options_copy;
+    if (options.prefix) {
+      // suppress prefix filtering since we have already checked the
+      // filters once at this point
+      options_copy = options;
+      options_copy.prefix = nullptr;
+    }
+    return cache->NewIterator(options.prefix ? options_copy : options,
                              soptions,
                              DecodeFixed64(file_value.data()),
                              DecodeFixed64(file_value.data() + 8),
@@ -198,22 +210,55 @@ static Iterator* GetFileIterator(void* arg,
  }
 }

+bool Version::PrefixMayMatch(const ReadOptions& options,
+                             const EnvOptions& soptions,
+                             const Slice& internal_prefix,
+                             Iterator* level_iter) const {
+  bool may_match = true;
+  level_iter->Seek(internal_prefix);
+  if (!level_iter->Valid()) {
+    // we're past end of level
+    may_match = false;
+  } else if (ExtractUserKey(level_iter->key()).starts_with(
+                                             ExtractUserKey(internal_prefix))) {
+    // TODO(tylerharter): do we need this case?  Or are we guaranteed
+    // key() will always be the biggest value for this SST?
+    may_match = true;
+  } else {
+    may_match = vset_->table_cache_->PrefixMayMatch(
+                           options,
+                           DecodeFixed64(level_iter->value().data()),
+                           DecodeFixed64(level_iter->value().data() + 8),
+                           internal_prefix, nullptr);
+  }
+  return may_match;
+}
+
 Iterator* Version::NewConcatenatingIterator(const ReadOptions& options,
                                            const EnvOptions& soptions,
                                            int level) const {
-  return NewTwoLevelIterator(
-      new LevelFileNumIterator(vset_->icmp_, &files_[level]),
-      &GetFileIterator, vset_->table_cache_, options, soptions);
+  Iterator* level_iter = new LevelFileNumIterator(vset_->icmp_, &files_[level]);
+  if (options.prefix) {
+    InternalKey internal_prefix(*options.prefix, 0, kTypeValue);
+    if (!PrefixMayMatch(options, soptions,
+                        internal_prefix.Encode(), level_iter)) {
+      delete level_iter;
+      // nothing in this level can match the prefix
+      return NewEmptyIterator();
+    }
+  }
+  return NewTwoLevelIterator(level_iter, &GetFileIterator,
+                             vset_->table_cache_, options, soptions);
 }

 void Version::AddIterators(const ReadOptions& options,
                           const EnvOptions& soptions,
                           std::vector<Iterator*>* iters) {
  // Merge all level zero files together since they may overlap
-  for (size_t i = 0; i < files_[0].size(); i++) {
+  for (const FileMetaData* file : files_[0]) {
    iters->push_back(
        vset_->table_cache_->NewIterator(
-            options, soptions, files_[0][i]->number, files_[0][i]->file_size));
+            options, soptions, file->number, file->file_size));
  }

  // For levels > 0, we can use a concatenating iterator that sequentially
@@ -357,6 +402,7 @@ static bool NewestFirstBySeqNo(FileMetaData* a, FileMetaData* b) {

 Version::Version(VersionSet* vset, uint64_t version_number)
    : vset_(vset), next_(this), prev_(this), refs_(0),
+      files_(new std::vector<FileMetaData*>[vset->NumberLevels()]),
      files_by_size_(vset->NumberLevels()),
      next_file_to_compact_by_size_(vset->NumberLevels()),
      file_to_compact_(nullptr),
@@ -365,7 +411,6 @@ Version::Version(VersionSet* vset, uint64_t version_number)
      compaction_level_(vset->NumberLevels()),
      offset_manifest_file_(0),
      version_number_(version_number) {
-  files_ = new std::vector<FileMetaData*>[vset->NumberLevels()];
 }

 void Version::Get(const ReadOptions& options,
@@ -375,7 +420,6 @@ void Version::Get(const ReadOptions& options,
                  std::deque<std::string>* operands,
                  GetStats* stats,
                  const Options& db_options,
-                  const bool no_io,
                  bool* value_found) {
  Slice ikey = k.internal_key();
  Slice user_key = k.user_key();
@@ -385,9 +429,6 @@ void Version::Get(const ReadOptions& options,
  auto logger = db_options.info_log;

  assert(status->ok() || status->IsMergeInProgress());
-  if (no_io) {
-    assert(status->ok());
-  }
  Saver saver;
  saver.state = status->ok()? kNotFound : kMerge;
  saver.ucmp = ucmp;
@@ -476,7 +517,7 @@ void Version::Get(const ReadOptions& options,
      bool tableIO = false;
      *status = vset_->table_cache_->Get(options, f->number, f->file_size,
                                         ikey, &saver, SaveValue, &tableIO,
-                                         MarkKeyMayExist, no_io);
+                                         MarkKeyMayExist);
      // TODO: examine the behavior for corrupted key
      if (!status->ok()) {
        return;
@@ -1394,9 +1435,9 @@ Status VersionSet::Recover() {
      if (s.ok()) {
        if (edit.has_comparator_ &&
            edit.comparator_ != icmp_.user_comparator()->Name()) {
-          s = Status::InvalidArgument(
-              edit.comparator_ + "does not match existing comparator ",
-              icmp_.user_comparator()->Name());
+          s = Status::InvalidArgument(icmp_.user_comparator()->Name(),
+                                      "does not match existing comparator " +
+                                      edit.comparator_);
        }
      }

@@ -1512,9 +1553,9 @@ Status VersionSet::DumpManifest(Options& options, std::string& dscname,
      if (s.ok()) {
        if (edit.has_comparator_ &&
            edit.comparator_ != icmp_.user_comparator()->Name()) {
-          s = Status::InvalidArgument(
-              edit.comparator_ + "does not match existing comparator ",
-              icmp_.user_comparator()->Name());
+          s = Status::InvalidArgument(icmp_.user_comparator()->Name(),
+                                      "does not match existing comparator " +
+                                      edit.comparator_);
        }
      }

@@ -1609,7 +1650,12 @@ void VersionSet::Finalize(Version* v,
  double max_score = 0;
  int max_score_level = 0;

-  for (int level = 0; level < NumberLevels()-1; level++) {
+  int num_levels_to_check =
+      (options_->compaction_style != kCompactionStyleUniversal) ?
+          NumberLevels() - 1 : 1;
+
+  for (int level = 0; level < num_levels_to_check; level++) {
+
    double score;
    if (level == 0) {
      // We treat level-0 specially by bounding the number of files
@@ -2123,17 +2169,236 @@ void VersionSet::SizeBeingCompacted(std::vector<uint64_t>& sizes) {
  }
 }

-Compaction* VersionSet::PickCompactionUniversal(int level, double score) {
+//
+// 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* VersionSet::PickCompactionUniversalSizeAmp(
+    int level, double score) {
  assert (level == 0);

-  // percentage flexibilty while comparing file sizes
-  uint64_t ratio = options_->compaction_options_universal.size_ratio;
+  // percentage flexibilty while reducing size amplification
+  uint64_t ratio = options_->compaction_options_universal.
+                     max_size_amplification_percent;
+
+  // The files are sorted from newest first to oldest last.
+  std::vector<int>& file_by_time = current_->files_by_size_[level];
+  assert(file_by_time.size() == current_->files_[level].size());
+
+  unsigned int candidate_count = 0;
+  uint64_t candidate_size = 0;
+  unsigned int start_index = 0;
+  FileMetaData* f = nullptr;
+
+  // Skip files that are already being compacted
+  for (unsigned int loop = 0; loop < file_by_time.size() - 1; loop++) {
+    int index = file_by_time[loop];
+    f = current_->files_[level][index];
+    if (!f->being_compacted) {
+      start_index = loop;         // Consider this as the first candidate.
+      break;
+    }
+    Log(options_->info_log, "Universal: skipping file %ld[%d] compacted %s",
+        f->number, loop, " cannot be a candidate to reduce size amp.\n");
+    f = nullptr;
+  }
+  if (f == nullptr) {
+    return nullptr;             // no candidate files
+  }
+
+  Log(options_->info_log, "Universal: First candidate file %ld[%d] %s",
+      f->number, start_index, " to reduce size amp.\n");
+
+  // keep adding up all the remaining files
+  for (unsigned int loop = start_index; loop < file_by_time.size() - 1;
+       loop++) {
+    int index = file_by_time[loop];
+    f = current_->files_[level][index];
+    if (f->being_compacted) {
+      Log(options_->info_log,
+          "Universal: Possible candidate file %ld[%d] %s.", f->number, loop,
+          " is already being compacted. No size amp reduction possible.\n");
+      return nullptr;
+    }
+    candidate_size += f->file_size;
+    candidate_count++;
+  }
+  if (candidate_count == 0) {
+    return nullptr;
+  }
+
+  // size of earliest file
+  int index = file_by_time[file_by_time.size() - 1];
+  uint64_t earliest_file_size = current_->files_[level][index]->file_size;
+
+  // size amplification = percentage of additional size
+  if (candidate_size * 100 < ratio * earliest_file_size) {
+    Log(options_->info_log,
+        "Universal: size amp not needed. newer-files-total-size %ld "
+        "earliest-file-size %ld",
+        candidate_size, earliest_file_size);
+    return nullptr;
+  } else {
+    Log(options_->info_log,
+        "Universal: size amp needed. newer-files-total-size %ld "
+        "earliest-file-size %ld",
+        candidate_size, earliest_file_size);
+  }
+  assert(start_index >= 0 && start_index < file_by_time.size() - 1);
+
+  // create a compaction request
+  // We always compact all the files, so always compress.
+  Compaction* c = new Compaction(level, level, MaxFileSizeForLevel(level),
+                                 LLONG_MAX, NumberLevels(), false,
+                                 true);
+  c->score_ = score;
+  for (unsigned int loop = start_index; loop < file_by_time.size(); loop++) {
+    int index = file_by_time[loop];
+    f = current_->files_[level][index];
+    c->inputs_[0].push_back(f);
+    Log(options_->info_log,
+        "Universal: size amp picking file %ld[%d] with size %ld",
+        f->number, index, f->file_size);
+  }
+  return c;
+}
+
+//
+// Consider compaction files based on their size differences with
+// the next file in time order.
+//
+Compaction* VersionSet::PickCompactionUniversalReadAmp(
+    int level, double score, unsigned int ratio,
+    unsigned int max_number_of_files_to_compact) {
+
  unsigned int min_merge_width =
    options_->compaction_options_universal.min_merge_width;
  unsigned int max_merge_width =
    options_->compaction_options_universal.max_merge_width;

-  if ((current_->files_[level].size() <=
+  // The files are sorted from newest first to oldest last.
+  std::vector<int>& file_by_time = current_->files_by_size_[level];
+  FileMetaData* f = nullptr;
+  bool done = false;
+  int start_index = 0;
+  unsigned int candidate_count;
+  assert(file_by_time.size() == current_->files_[level].size());
+
+  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);
+
+  // Considers a candidate file only if it is smaller than the
+  // total size accumulated so far.
+  for (unsigned int loop = 0; loop < file_by_time.size(); loop++) {
+
+    candidate_count = 0;
+
+    // Skip files that are already being compacted
+    for (f = nullptr; loop < file_by_time.size(); loop++) {
+      int index = file_by_time[loop];
+      f = current_->files_[level][index];
+
+      if (!f->being_compacted) {
+        candidate_count = 1;
+        break;
+      }
+      Log(options_->info_log,
+          "Universal: file %ld[%d] being compacted, skipping",
+          f->number, loop);
+      f = nullptr;
+    }
+
+    // This file is not being compacted. Consider it as the
+    // first candidate to be compacted.
+    uint64_t candidate_size =  f != nullptr? f->file_size : 0;
+    if (f != nullptr) {
+      Log(options_->info_log, "Universal: Possible candidate file %ld[%d].",
+          f->number, loop);
+    }
+
+    // Check if the suceeding files need compaction.
+    for (unsigned int i = loop+1;
+         candidate_count < max_files_to_compact && i < file_by_time.size();
+         i++) {
+      int index = file_by_time[i];
+      FileMetaData* f = current_->files_[level][index];
+      if (f->being_compacted) {
+        break;
+      }
+      // pick files if the total candidate file size (increased by the
+      // specified ratio) is still larger than the next candidate file.
+      uint64_t sz = (candidate_size * (100L + ratio)) /100;
+      if (sz < f->file_size) {
+        break;
+      }
+      candidate_count++;
+      candidate_size += f->file_size;
+    }
+
+    // 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 (unsigned int i = loop;
+           i < loop + candidate_count && i < file_by_time.size(); i++) {
+       int index = file_by_time[i];
+       FileMetaData* f = current_->files_[level][index];
+       Log(options_->info_log,
+           "Universal: Skipping file %ld[%d] with size %ld %d\n",
+           f->number, i, f->file_size, f->being_compacted);
+      }
+    }
+  }
+  if (!done || candidate_count <= 1) {
+    return nullptr;
+  }
+  unsigned int 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 =
+      options_->compaction_options_universal.compression_size_percent;
+  if (ratio_to_compress >= 0) {
+    uint64_t total_size = TotalFileSize(current_->files_[level]);
+    uint64_t older_file_size = 0;
+    for (unsigned int i = file_by_time.size() - 1; i >= first_index_after;
+        i--) {
+      older_file_size += current_->files_[level][file_by_time[i]]->file_size;
+      if (older_file_size * 100L >= total_size * (long) ratio_to_compress) {
+        enable_compression = false;
+        break;
+      }
+    }
+  }
+  Compaction* c = new Compaction(level, level, MaxFileSizeForLevel(level),
+                                 LLONG_MAX, NumberLevels(), false,
+                                 enable_compression);
+  c->score_ = score;
+
+  for (unsigned int i = start_index; i < first_index_after; i++) {
+    int index = file_by_time[i];
+    FileMetaData* f = current_->files_[level][index];
+    c->inputs_[0].push_back(f);
+    Log(options_->info_log, "Universal: Picking file %ld[%d] with size %ld\n",
+        f->number, i, f->file_size);
+  }
+  return c;
+}
+
+//
+// Universal style of compaction. Pick files that are contiguous in
+// time-range to compact.
+//
+Compaction* VersionSet::PickCompactionUniversal(int level, double score) {
+  assert (level == 0);
+
+  if ((current_->files_[level].size() <
      (unsigned int)options_->level0_file_num_compaction_trigger)) {
    Log(options_->info_log, "Universal: nothing to do\n");
    return nullptr;
@@ -2143,127 +2408,29 @@ Compaction* VersionSet::PickCompactionUniversal(int level, double score) {
      current_->files_[level].size(),
      LevelFileSummary(&tmp, 0));

-  Compaction* c = nullptr;
-  c = new Compaction(level, level, MaxFileSizeForLevel(level),
-                     LLONG_MAX, NumberLevels());
-  c->score_ = score;
+  // Check for size amplification first.
+  Compaction* c = PickCompactionUniversalSizeAmp(level, score);
+  if (c == nullptr) {

-  // The files are sorted from newest first to oldest last.
-  std::vector<int>& file_by_time = current_->files_by_size_[level];
-  FileMetaData* f = nullptr;
-  bool done = false;
-  assert(file_by_time.size() == current_->files_[level].size());
+    // Size amplification is within limits. Try reducing read
+    // amplification while maintaining file size ratios.
+    unsigned int ratio = options_->compaction_options_universal.size_ratio;
+    c = PickCompactionUniversalReadAmp(level, score, ratio, UINT_MAX);

-  unsigned int max_files_to_compact = std::min(max_merge_width, UINT_MAX);
-
-  // Make two pass. The first pass considers a candidate file
-  // only if it is smaller than the total size accumulated so far.
-  // The second pass does not look at the slope of the
-  // file-size  curve to decide what to pick for compaction.
-  for (int iter = 0; !done && iter < 2; iter++) {
-
-    for (unsigned int loop = 0; loop < file_by_time.size(); ) {
-
-      // Skip files that are already being compacted
-      for (f = nullptr; loop < file_by_time.size(); loop++) {
-        int index = file_by_time[loop];
-        f = current_->files_[level][index];
-
-        if (!f->being_compacted) {
-          break;
-        }
-        Log(options_->info_log, "Universal: file %ld[%d] being compacted, skipping",
-            f->number, loop);
-        f = nullptr;
-      }
-
-      // This file is not being compacted. Consider it as the
-      // first candidate to be compacted.
-      unsigned int candidate_count = 1;
-      uint64_t candidate_size =  f != nullptr? f->file_size : 0;
-      if (f != nullptr) {
-        Log(options_->info_log, "Universal: Possible candidate file %ld[%d] %s.",
-            f->number, loop, iter == 0? "" : "forced ");
-      }
-
-      // Check if the suceeding files need compaction.
-      for (unsigned int i = loop+1;
-           candidate_count < max_files_to_compact && i < file_by_time.size();
-           i++) {
-        int index = file_by_time[i];
-        FileMetaData* f = current_->files_[level][index];
-        if (f->being_compacted) {
-          break;
-        }
-        // If this is the first iteration, then we pick files if the
-        // total candidate file size (increased by the specified ratio)
-        // is still larger than the next candidate file.
-        if (iter == 0) {
-          uint64_t sz = (candidate_size * (100 + ratio)) /100;
-          if (sz < f->file_size) {
-            break;
-          }
-        }
-        candidate_count++;
-        candidate_size += f->file_size;
-      }
-
-      // Found a series of consecutive files that need compaction.
-      if (candidate_count >= (unsigned int)min_merge_width) {
-        for (unsigned int i = loop; i < loop + candidate_count; i++) {
-          int index = file_by_time[i];
-          FileMetaData* f = current_->files_[level][index];
-          c->inputs_[0].push_back(f);
-          Log(options_->info_log, "Universal: Picking file %ld[%d] with size %ld %s",
-              f->number, i, f->file_size,
-              (iter == 0 ? "" : "forced"));
-        }
-        done = true;
-        break;
-      } else {
-        for (unsigned int i = loop;
-             i < loop + candidate_count && i < file_by_time.size(); i++) {
-         int index = file_by_time[i];
-         FileMetaData* f = current_->files_[level][index];
-         Log(options_->info_log, "Universal: Skipping file %ld[%d] with size %ld %d %s",
-             f->number, i, f->file_size, f->being_compacted,
-              (iter == 0 ? "" : "forced"));
-        }
-      }
-      loop += candidate_count;
-    }
-    assert(done || c->inputs_[0].size() == 0);
-
-    // If we are unable to find a normal compaction run and we are still
-    // above the compaction threshold, iterate again to pick compaction
-    // candidates, this time without considering their size differences.
-    if (!done) {
-      int files_not_in_compaction = 0;
-      for (unsigned int i = 0; i < current_->files_[level].size(); i++) {
-        f = current_->files_[level][i];
-        if (!f->being_compacted) {
-          files_not_in_compaction++;
-        }
-      }
-      int expected_num_files = files_not_in_compaction +
-                compactions_in_progress_[level].size();
-      if (expected_num_files <=
-          options_->level0_file_num_compaction_trigger + 1) {
-        done = true;     // nothing more to do
-      } else {
-        max_files_to_compact = std::min((int)max_merge_width,
-          expected_num_files - options_->level0_file_num_compaction_trigger);
-        Log(options_->info_log, "Universal: second loop with maxfiles %d",
-            max_files_to_compact);
-      }
+    // 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.
+    if (c == nullptr) {
+      unsigned int num_files = current_->files_[level].size() -
+                               options_->level0_file_num_compaction_trigger;
+      c = PickCompactionUniversalReadAmp(level, score, UINT_MAX, num_files);
    }
  }
-  if (c->inputs_[0].size() <= 1) {
-    Log(options_->info_log, "Universal: only %ld files, nothing to do.\n",
-        c->inputs_[0].size());
-    delete c;
+  if (c == nullptr) {
    return nullptr;
  }
+  assert(c->inputs_[0].size() > 1);

  // validate that all the chosen files are non overlapping in time
  FileMetaData* newerfile __attribute__((unused)) = nullptr;
@@ -2275,6 +2442,9 @@ Compaction* VersionSet::PickCompactionUniversal(int level, double score) {
    newerfile = f;
  }

+  // The files are sorted from newest first to oldest last.
+  std::vector<int>& file_by_time = current_->files_by_size_[level];
+
  // Is the earliest file part of this compaction?
  int last_index = file_by_time[file_by_time.size()-1];
  FileMetaData* last_file = current_->files_[level][last_index];
@@ -2637,6 +2807,41 @@ void VersionSet::SetupOtherInputs(Compaction* c) {
  c->edit_->SetCompactPointer(level, largest);
 }

+Status VersionSet::GetMetadataForFile(
+    uint64_t number,
+    int *filelevel,
+    FileMetaData *meta) {
+  for (int level = 0; level < NumberLevels(); level++) {
+    const std::vector<FileMetaData*>& files = current_->files_[level];
+    for (size_t i = 0; i < files.size(); i++) {
+      if (files[i]->number == number) {
+        *meta = *files[i];
+        *filelevel = level;
+        return Status::OK();
+      }
+    }
+  }
+  return Status::NotFound("File not present in any level");
+}
+
+void VersionSet::GetLiveFilesMetaData(
+    std::vector<LiveFileMetaData> * metadata) {
+  for (int level = 0; level < NumberLevels(); level++) {
+    const std::vector<FileMetaData*>& files = current_->files_[level];
+    for (size_t i = 0; i < files.size(); i++) {
+      LiveFileMetaData filemetadata;
+      filemetadata.name = TableFileName("", files[i]->number);
+      filemetadata.level = level;
+      filemetadata.size = files[i]->file_size;
+      filemetadata.smallestkey = files[i]->smallest.user_key().ToString();
+      filemetadata.largestkey = files[i]->largest.user_key().ToString();
+      filemetadata.smallest_seqno = files[i]->smallest_seqno;
+      filemetadata.largest_seqno = files[i]->largest_seqno;
+      metadata->push_back(filemetadata);
+    }
+  }
+}
+
 Compaction* VersionSet::CompactRange(
    int level,
    const InternalKey* begin,
@@ -2700,7 +2905,7 @@ Compaction* VersionSet::CompactRange(

 Compaction::Compaction(int level, int out_level, uint64_t target_file_size,
  uint64_t max_grandparent_overlap_bytes, int number_levels,
-  bool seek_compaction)
+  bool seek_compaction, bool enable_compression)
    : level_(level),
      out_level_(out_level),
      max_output_file_size_(target_file_size),
@@ -2708,6 +2913,7 @@ Compaction::Compaction(int level, int out_level, uint64_t target_file_size,
      input_version_(nullptr),
      number_levels_(number_levels),
      seek_compaction_(seek_compaction),
+      enable_compression_(enable_compression),
      grandparent_index_(0),
      seen_key_(false),
      overlapped_bytes_(0),
@@ -2881,4 +3087,4 @@ void Compaction::Summary(char* output, int len) {
      level_low_summary, level_up_summary);
 }

-}  // namespace leveldb
+}  // namespace rocksdb
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -12,9 +17,7 @@
 // Version,VersionSet are thread-compatible, but require external
 // synchronization on all accesses.

-#ifndef STORAGE_LEVELDB_DB_VERSION_SET_H_
-#define STORAGE_LEVELDB_DB_VERSION_SET_H_
-
+#pragma once
 #include <map>
 #include <memory>
 #include <set>
@@ -25,18 +28,16 @@
 #include "port/port.h"
 #include "db/table_cache.h"

-namespace leveldb {
+namespace rocksdb {

 namespace log { class Writer; }

 class Compaction;
 class Iterator;
 class MemTable;
-class TableBuilder;
 class TableCache;
 class Version;
 class VersionSet;
-class WritableFile;

 // Return the smallest index i such that files[i]->largest >= key.
 // Return files.size() if there is no such file.
@@ -76,7 +77,7 @@ class Version {
  };
  void Get(const ReadOptions&, const LookupKey& key, std::string* val,
           Status* status, std::deque<std::string>* operands, GetStats* stats,
-           const Options& db_option, const bool no_io = false,
+           const Options& db_option,
           bool* value_found = nullptr);

  // Adds "stats" into the current state.  Returns true if a new
@@ -152,6 +153,8 @@ class Version {
  Iterator* NewConcatenatingIterator(const ReadOptions&,
                                     const EnvOptions& soptions,
                                     int level) const;
+  bool PrefixMayMatch(const ReadOptions& options, const EnvOptions& soptions,
+                      const Slice& internal_prefix, Iterator* level_iter) const;

  VersionSet* vset_;            // VersionSet to which this Version belongs
  Version* next_;               // Next version in linked list
@@ -314,7 +317,17 @@ class VersionSet {
  // Returns true iff some level needs a compaction because it has
  // exceeded its target size.
  bool NeedsSizeCompaction() const {
-    for (int i = 0; i < NumberLevels()-1; i++) {
+    // In universal compaction case, this check doesn't really
+    // check the compaction condition, but checks num of files threshold
+    // only. We are not going to miss any compaction opportunity
+    // but it's likely that more compactions are scheduled but
+    // ending up with nothing to do. We can improve it later.
+    // TODO: improve this function to be accurate for universal
+    //       compactions.
+    int num_levels_to_check =
+        (options_->compaction_style != kCompactionStyleUniversal) ?
+            NumberLevels() - 1 : 1;
+    for (int i = 0; i < num_levels_to_check; i++) {
      if (current_->compaction_score_[i] >= 1) {
        return true;
      }
@@ -381,6 +394,13 @@ class VersionSet {
  // Pick files to compact in Universal mode
  Compaction* PickCompactionUniversal(int level, double score);

+  // Pick Universal compaction to limit read amplification
+  Compaction* PickCompactionUniversalReadAmp(int level, double score,
+                unsigned int ratio, unsigned int num_files);
+
+  // Pick Universal compaction to limit space amplification.
+  Compaction* PickCompactionUniversalSizeAmp(int level, double score);
+
  // Free up the files that were participated in a compaction
  void ReleaseCompactionFiles(Compaction* c, Status status);

@@ -405,6 +425,12 @@ class VersionSet {

  double MaxBytesForLevel(int level);

+  Status GetMetadataForFile(
+    uint64_t number, int *filelevel, FileMetaData *metadata);
+
+  void GetLiveFilesMetaData(
+    std::vector<LiveFileMetaData> *metadata);
+
 private:
  class Builder;
  struct ManifestWriter;
@@ -532,6 +558,9 @@ class Compaction {
  // Maximum size of files to build during this compaction.
  uint64_t MaxOutputFileSize() const { return max_output_file_size_; }

+  // Whether compression will be enabled for compaction outputs
+  bool enable_compression() const { return enable_compression_; }
+
  // Is this a trivial compaction that can be implemented by just
  // moving a single input file to the next level (no merging or splitting)
  bool IsTrivialMove() const;
@@ -566,7 +595,7 @@ class Compaction {

  explicit Compaction(int level, int out_level, uint64_t target_file_size,
    uint64_t max_grandparent_overlap_bytes, int number_levels,
-    bool seek_compaction = false);
+    bool seek_compaction = false, bool enable_compression = true);

  int level_;
  int out_level_; // levels to which output files are stored
@@ -577,6 +606,7 @@ class Compaction {
  int number_levels_;

  bool seek_compaction_;
+  bool enable_compression_;

  // Each compaction reads inputs from "level_" and "level_+1"
  std::vector<FileMetaData*> inputs_[2];      // The two sets of inputs
@@ -612,6 +642,4 @@ class Compaction {
  void ResetNextCompactionIndex();
 };

-}  // namespace leveldb
-
-#endif  // STORAGE_LEVELDB_DB_VERSION_SET_H_
+}  // namespace rocksdb
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2012 Facebook. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
@@ -10,7 +15,7 @@
 #include "db/log_writer.h"
 #include "util/logging.h"

-namespace leveldb {
+namespace rocksdb {

 Status VersionSet::ReduceNumberOfLevels(int new_levels, port::Mutex* mu) {

@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -7,7 +12,7 @@
 #include "util/testharness.h"
 #include "util/testutil.h"

-namespace leveldb {
+namespace rocksdb {

 class FindFileTest {
 public:
@@ -172,8 +177,8 @@ TEST(FindFileTest, OverlappingFiles) {
  ASSERT_TRUE(Overlaps("600", "700"));
 }

-}  // namespace leveldb
+}  // namespace rocksdb

 int main(int argc, char** argv) {
-  return leveldb::test::RunAllTests();
+  return rocksdb::test::RunAllTests();
 }
@@ -1,3 +1,8 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.
@@ -14,10 +19,10 @@
 //    len: varint32
 //    data: uint8[len]

-#include "leveldb/write_batch.h"
+#include "rocksdb/write_batch.h"

-#include "leveldb/options.h"
-#include "leveldb/statistics.h"
+#include "rocksdb/options.h"
+#include "rocksdb/statistics.h"
 #include "db/dbformat.h"
 #include "db/db_impl.h"
 #include "db/memtable.h"
@@ -26,7 +31,7 @@
 #include "util/coding.h"
 #include <stdexcept>

-namespace leveldb {
+namespace rocksdb {

 // WriteBatch header has an 8-byte sequence number followed by a 4-byte count.
 static const size_t kHeader = 12;
@@ -48,6 +53,10 @@ void WriteBatch::Handler::LogData(const Slice& blob) {
  // them.
 }

+bool WriteBatch::Handler::Continue() {
+  return true;
+}
+
 void WriteBatch::Clear() {
  rep_.clear();
  rep_.resize(kHeader);
@@ -66,7 +75,7 @@ Status WriteBatch::Iterate(Handler* handler) const {
  input.remove_prefix(kHeader);
  Slice key, value, blob;
  int found = 0;
-  while (!input.empty()) {
+  while (!input.empty() && handler->Continue()) {
    char tag = input[0];
    input.remove_prefix(1);
    switch (tag) {
@@ -223,4 +232,4 @@ void WriteBatchInternal::Append(WriteBatch* dst, const WriteBatch* src) {
  dst->rep_.append(src->rep_.data() + kHeader, src->rep_.size() - kHeader);
 }

-}  // namespace leveldb
+}  // namespace rocksdb
@@ -1,16 +1,19 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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.

-#ifndef STORAGE_LEVELDB_DB_WRITE_BATCH_INTERNAL_H_
-#define STORAGE_LEVELDB_DB_WRITE_BATCH_INTERNAL_H_
+#pragma once
+#include "rocksdb/types.h"
+#include "rocksdb/write_batch.h"
+#include "rocksdb/db.h"
+#include "rocksdb/options.h"

-#include "leveldb/types.h"
-#include "leveldb/write_batch.h"
-#include "leveldb/db.h"
-#include "leveldb/options.h"
-
-namespace leveldb {
+namespace rocksdb {

 class MemTable;

@@ -51,7 +54,4 @@ class WriteBatchInternal {
  static void Append(WriteBatch* dst, const WriteBatch* src);
 };

-}  // namespace leveldb
-
-
-#endif  // STORAGE_LEVELDB_DB_WRITE_BATCH_INTERNAL_H_
+}  // namespace rocksdb
@@ -1,18 +1,23 @@
+//  Copyright (c) 2013, Facebook, Inc.  All rights reserved.
+//  This source code is licensed under the BSD-style license found in the
+//  LICENSE file in the root directory of this source tree. An additional grant
+//  of patent rights can be found in the PATENTS file in the same directory.
+//
 // Copyright (c) 2011 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 "leveldb/db.h"
+#include "rocksdb/db.h"

 #include <memory>
-#include "db/skiplistrep.h"
 #include "db/memtable.h"
 #include "db/write_batch_internal.h"
-#include "leveldb/env.h"
+#include "rocksdb/env.h"
+#include "rocksdb/memtablerep.h"
 #include "util/logging.h"
 #include "util/testharness.h"

-namespace leveldb {
+namespace rocksdb {

 static std::string PrintContents(WriteBatch* b) {
  InternalKeyComparator cmp(BytewiseComparator());
@@ -134,6 +139,24 @@ TEST(WriteBatchTest, Append) {
  ASSERT_EQ(4, b1.Count());
 }

+namespace {
+  struct TestHandler : public WriteBatch::Handler {
+    std::string seen;
+    virtual void Put(const Slice& key, const Slice& value) {
+      seen += "Put(" + key.ToString() + ", " + value.ToString() + ")";
+    }
+    virtual void Merge(const Slice& key, const Slice& value) {
+      seen += "Merge(" + key.ToString() + ", " + value.ToString() + ")";
+    }
+    virtual void LogData(const Slice& blob) {
+      seen += "LogData(" + blob.ToString() + ")";
+    }
+    virtual void Delete(const Slice& key) {
+      seen += "Delete(" + key.ToString() + ")";
+    }
+  };
+}
+
 TEST(WriteBatchTest, Blob) {
  WriteBatch batch;
  batch.Put(Slice("k1"), Slice("v1"));
@@ -151,21 +174,7 @@ TEST(WriteBatchTest, Blob) {
            "Put(k3, v3)@2",
            PrintContents(&batch));

-  struct Handler : public WriteBatch::Handler {
-    std::string seen;
-    virtual void Put(const Slice& key, const Slice& value) {
-      seen += "Put(" + key.ToString() + ", " + value.ToString() + ")";
-    }
-    virtual void Merge(const Slice& key, const Slice& value) {
-      seen += "Merge(" + key.ToString() + ", " + value.ToString() + ")";
-    }
-    virtual void LogData(const Slice& blob) {
-      seen += "LogData(" + blob.ToString() + ")";
-    }
-    virtual void Delete(const Slice& key) {
-      seen += "Delete(" + key.ToString() + ")";
-    }
-  } handler;
+  TestHandler handler;
  batch.Iterate(&handler);
  ASSERT_EQ(
            "Put(k1, v1)"
@@ -178,8 +187,47 @@ TEST(WriteBatchTest, Blob) {
            handler.seen);
 }

-}  // namespace leveldb
+TEST(WriteBatchTest, Continue) {
+  WriteBatch batch;
+
+  struct Handler : public TestHandler {
+    int num_seen = 0;
+    virtual void Put(const Slice& key, const Slice& value) {
+      ++num_seen;
+      TestHandler::Put(key, value);
+    }
+    virtual void Merge(const Slice& key, const Slice& value) {
+      ++num_seen;
+      TestHandler::Merge(key, value);
+    }
+    virtual void LogData(const Slice& blob) {
+      ++num_seen;
+      TestHandler::LogData(blob);
+    }
+    virtual void Delete(const Slice& key) {
+      ++num_seen;
+      TestHandler::Delete(key);
+    }
+    virtual bool Continue() override {
+      return num_seen < 3;
+    }
+  } handler;
+
+  batch.Put(Slice("k1"), Slice("v1"));
+  batch.PutLogData(Slice("blob1"));
+  batch.Delete(Slice("k1"));
+  batch.PutLogData(Slice("blob2"));
+  batch.Merge(Slice("foo"), Slice("bar"));
+  batch.Iterate(&handler);
+  ASSERT_EQ(
+            "Put(k1, v1)"
+            "LogData(blob1)"
+            "Delete(k1)",
+            handler.seen);
+}
+
+}  // namespace rocksdb

 int main(int argc, char** argv) {
-  return leveldb::test::RunAllTests();
+  return rocksdb::test::RunAllTests();
 }
@@ -111,7 +111,7 @@ static void WalCheckpoint(sqlite3* db_) {
  }
 }

-namespace leveldb {
+namespace rocksdb {

 // Helper for quickly generating random data.
 namespace {
@@ -664,7 +664,7 @@ class Benchmark {

 };

-}  // namespace leveldb
+}  // namespace rocksdb

 int main(int argc, char** argv) {
  std::string default_db_path;
@@ -672,7 +672,7 @@ int main(int argc, char** argv) {
    double d;
    int n;
    char junk;
-    if (leveldb::Slice(argv[i]).starts_with("--benchmarks=")) {
+    if (rocksdb::Slice(argv[i]).starts_with("--benchmarks=")) {
      FLAGS_benchmarks = argv[i] + strlen("--benchmarks=");
    } else if (sscanf(argv[i], "--histogram=%d%c", &n, &junk) == 1 &&
               (n == 0 || n == 1)) {
@@ -688,7 +688,7 @@ int main(int argc, char** argv) {
      FLAGS_reads = n;
    } else if (sscanf(argv[i], "--value_size=%d%c", &n, &junk) == 1) {
      FLAGS_value_size = n;
-    } else if (leveldb::Slice(argv[i]) == leveldb::Slice("--no_transaction")) {
+    } else if (rocksdb::Slice(argv[i]) == rocksdb::Slice("--no_transaction")) {
      FLAGS_transaction = false;
    } else if (sscanf(argv[i], "--page_size=%d%c", &n, &junk) == 1) {
      FLAGS_page_size = n;
@@ -707,12 +707,12 @@ int main(int argc, char** argv) {

  // Choose a location for the test database if none given with --db=<path>
  if (FLAGS_db == NULL) {
-      leveldb::Env::Default()->GetTestDirectory(&default_db_path);
+      rocksdb::Env::Default()->GetTestDirectory(&default_db_path);
      default_db_path += "/dbbench";
      FLAGS_db = default_db_path.c_str();
  }

-  leveldb::Benchmark benchmark;
+  rocksdb::Benchmark benchmark;
  benchmark.Run();
  return 0;
 }
@@ -80,7 +80,7 @@ static void DBSynchronize(kyotocabinet::TreeDB* db_)
  }
 }

-namespace leveldb {
+namespace rocksdb {

 // Helper for quickly generating random data.
 namespace {
@@ -479,7 +479,7 @@ class Benchmark {
  }
 };

-}  // namespace leveldb
+}  // namespace rocksdb

 int main(int argc, char** argv) {
  std::string default_db_path;
@@ -487,7 +487,7 @@ int main(int argc, char** argv) {
    double d;
    int n;
    char junk;
-    if (leveldb::Slice(argv[i]).starts_with("--benchmarks=")) {
+    if (rocksdb::Slice(argv[i]).starts_with("--benchmarks=")) {
      FLAGS_benchmarks = argv[i] + strlen("--benchmarks=");
    } else if (sscanf(argv[i], "--compression_ratio=%lf%c", &d, &junk) == 1) {
      FLAGS_compression_ratio = d;
@@ -517,12 +517,12 @@ int main(int argc, char** argv) {

  // Choose a location for the test database if none given with --db=<path>
  if (FLAGS_db == NULL) {
-      leveldb::Env::Default()->GetTestDirectory(&default_db_path);
+      rocksdb::Env::Default()->GetTestDirectory(&default_db_path);
      default_db_path += "/dbbench";
      FLAGS_db = default_db_path.c_str();
  }

-  leveldb::Benchmark benchmark;
+  rocksdb::Benchmark benchmark;
  benchmark.Run();
  return 0;
 }
@@ -25,27 +25,27 @@ creating it if necessary:
  #include &lt;assert&gt;
  #include "leveldb/db.h"

-  leveldb::DB* db;
-  leveldb::Options options;
+  rocksdb::DB* db;
+  rocksdb::Options options;
  options.create_if_missing = true;
-  leveldb::Status status = leveldb::DB::Open(options, "/tmp/testdb", &amp;db);
+  rocksdb::Status status = rocksdb::DB::Open(options, "/tmp/testdb", &amp;db);
  assert(status.ok());
  ...
 </pre>
 If you want to raise an error if the database already exists, add
-the following line before the <code>leveldb::DB::Open</code> call:
+the following line before the <code>rocksdb::DB::Open</code> call:
 <pre>
  options.error_if_exists = true;
 </pre>
 <h1>Status</h1>
 <p>
-You may have noticed the <code>leveldb::Status</code> type above.  Values of this
+You may have noticed the <code>rocksdb::Status</code> type above.  Values of this
 type are returned by most functions in <code>leveldb</code> that may encounter an
 error.  You can check if such a result is ok, and also print an
 associated error message:
 <p>
 <pre>
-   leveldb::Status s = ...;
+   rocksdb::Status s = ...;
   if (!s.ok()) cerr &lt;&lt; s.ToString() &lt;&lt; endl;
 </pre>
 <h1>Closing A Database</h1>
@@ -65,9 +65,9 @@ modify/query the database.  For example, the following code
 moves the value stored under key1 to key2.
 <pre>
  std::string value;
-  leveldb::Status s = db-&gt;Get(leveldb::ReadOptions(), key1, &amp;value);
-  if (s.ok()) s = db-&gt;Put(leveldb::WriteOptions(), key2, value);
-  if (s.ok()) s = db-&gt;Delete(leveldb::WriteOptions(), key1);
+  rocksdb::Status s = db-&gt;Get(rocksdb::ReadOptions(), key1, &amp;value);
+  if (s.ok()) s = db-&gt;Put(rocksdb::WriteOptions(), key2, value);
+  if (s.ok()) s = db-&gt;Delete(rocksdb::WriteOptions(), key1);
 </pre>

 <h1>Atomic Updates</h1>
@@ -81,12 +81,12 @@ atomically apply a set of updates:
  #include "leveldb/write_batch.h"
  ...
  std::string value;
-  leveldb::Status s = db-&gt;Get(leveldb::ReadOptions(), key1, &amp;value);
+  rocksdb::Status s = db-&gt;Get(rocksdb::ReadOptions(), key1, &amp;value);
  if (s.ok()) {
-    leveldb::WriteBatch batch;
+    rocksdb::WriteBatch batch;
    batch.Delete(key1);
    batch.Put(key2, value);
-    s = db-&gt;Write(leveldb::WriteOptions(), &amp;batch);
+    s = db-&gt;Write(rocksdb::WriteOptions(), &amp;batch);
  }
 </pre>
 The <code>WriteBatch</code> holds a sequence of edits to be made to the database,
@@ -109,7 +109,7 @@ persistent storage.  (On Posix systems, this is implemented by calling
 either <code>fsync(...)</code> or <code>fdatasync(...)</code> or
 <code>msync(..., MS_SYNC)</code> before the write operation returns.)
 <pre>
-  leveldb::WriteOptions write_options;
+  rocksdb::WriteOptions write_options;
  write_options.sync = true;
  db-&gt;Put(write_options, ...);
 </pre>
@@ -144,7 +144,7 @@ the batch.
 A database may only be opened by one process at a time.
 The <code>leveldb</code> implementation acquires a lock from the
 operating system to prevent misuse.  Within a single process, the
-same <code>leveldb::DB</code> object may be safely shared by multiple
+same <code>rocksdb::DB</code> object may be safely shared by multiple
 concurrent threads.  I.e., different threads may write into or fetch
 iterators or call <code>Get</code> on the same database without any
 external synchronization (the leveldb implementation will
@@ -160,7 +160,7 @@ The following example demonstrates how to print all key,value pairs
 in a database.
 <p>
 <pre>
-  leveldb::Iterator* it = db-&gt;NewIterator(leveldb::ReadOptions());
+  rocksdb::Iterator* it = db-&gt;NewIterator(rocksdb::ReadOptions());
  for (it-&gt;SeekToFirst(); it-&gt;Valid(); it-&gt;Next()) {
    cout &lt;&lt; it-&gt;key().ToString() &lt;&lt; ": "  &lt;&lt; it-&gt;value().ToString() &lt;&lt; endl;
  }
@@ -196,10 +196,10 @@ implicit snapshot of the current state.
 Snapshots are created by the DB::GetSnapshot() method:
 <p>
 <pre>
-  leveldb::ReadOptions options;
+  rocksdb::ReadOptions options;
  options.snapshot = db-&gt;GetSnapshot();
  ... apply some updates to db ...
-  leveldb::Iterator* iter = db-&gt;NewIterator(options);
+  rocksdb::Iterator* iter = db-&gt;NewIterator(options);
  ... read using iter to view the state when the snapshot was created ...
  delete iter;
  db-&gt;ReleaseSnapshot(options.snapshot);
@@ -211,7 +211,7 @@ support reading as of that snapshot.
 <h1>Slice</h1>
 <p>
 The return value of the <code>it->key()</code> and <code>it->value()</code> calls above
-are instances of the <code>leveldb::Slice</code> type.  <code>Slice</code> is a simple
+are instances of the <code>rocksdb::Slice</code> type.  <code>Slice</code> is a simple
 structure that contains a length and a pointer to an external byte
 array.  Returning a <code>Slice</code> is a cheaper alternative to returning a
 <code>std::string</code> since we do not need to copy potentially large keys and
@@ -223,10 +223,10 @@ C++ strings and null-terminated C-style strings can be easily converted
 to a Slice:
 <p>
 <pre>
-   leveldb::Slice s1 = "hello";
+   rocksdb::Slice s1 = "hello";

   std::string str("world");
-   leveldb::Slice s2 = str;
+   rocksdb::Slice s2 = str;
 </pre>
 A Slice can be easily converted back to a C++ string:
 <pre>
@@ -238,7 +238,7 @@ the external byte array into which the Slice points remains live while
 the Slice is in use.  For example, the following is buggy:
 <p>
 <pre>
-   leveldb::Slice slice;
+   rocksdb::Slice slice;
   if (...) {
     std::string str = ...;
     slice = str;
@@ -255,16 +255,16 @@ which orders bytes lexicographically.  You can however supply a custom
 comparator when opening a database.  For example, suppose each
 database key consists of two numbers and we should sort by the first
 number, breaking ties by the second number.  First, define a proper
-subclass of <code>leveldb::Comparator</code> that expresses these rules:
+subclass of <code>rocksdb::Comparator</code> that expresses these rules:
 <p>
 <pre>
-  class TwoPartComparator : public leveldb::Comparator {
+  class TwoPartComparator : public rocksdb::Comparator {
   public:
    // Three-way comparison function:
    //   if a &lt; b: negative result
    //   if a &gt; b: positive result
    //   else: zero result
-    int Compare(const leveldb::Slice&amp; a, const leveldb::Slice&amp; b) const {
+    int Compare(const rocksdb::Slice&amp; a, const rocksdb::Slice&amp; b) const {
      int a1, a2, b1, b2;
      ParseKey(a, &amp;a1, &amp;a2);
      ParseKey(b, &amp;b1, &amp;b2);
@@ -277,7 +277,7 @@ subclass of <code>leveldb::Comparator</code> that expresses these rules:

    // Ignore the following methods for now:
    const char* Name() const { return "TwoPartComparator"; }
-    void FindShortestSeparator(std::string*, const leveldb::Slice&amp;) const { }
+    void FindShortestSeparator(std::string*, const rocksdb::Slice&amp;) const { }
    void FindShortSuccessor(std::string*) const { }
  };
 </pre>
@@ -285,18 +285,18 @@ Now create a database using this custom comparator:
 <p>
 <pre>
  TwoPartComparator cmp;
-  leveldb::DB* db;
-  leveldb::Options options;
+  rocksdb::DB* db;
+  rocksdb::Options options;
  options.create_if_missing = true;
  options.comparator = &amp;cmp;
-  leveldb::Status status = leveldb::DB::Open(options, "/tmp/testdb", &amp;db);
+  rocksdb::Status status = rocksdb::DB::Open(options, "/tmp/testdb", &amp;db);
  ...
 </pre>
 <h2>Backwards compatibility</h2>
 <p>
 The result of the comparator's <code>Name</code> method is attached to the
 database when it is created, and is checked on every subsequent
-database open.  If the name changes, the <code>leveldb::DB::Open</code> call will
+database open.  If the name changes, the <code>rocksdb::DB::Open</code> call will
 fail.  Therefore, change the name if and only if the new key format
 and comparison function are incompatible with existing databases, and
 it is ok to discard the contents of all existing databases.
@@ -339,9 +339,9 @@ compression entirely, but should only do so if benchmarks show a
 performance improvement:
 <p>
 <pre>
-  leveldb::Options options;
-  options.compression = leveldb::kNoCompression;
-  ... leveldb::DB::Open(options, name, ...) ....
+  rocksdb::Options options;
+  options.compression = rocksdb::kNoCompression;
+  ... rocksdb::DB::Open(options, name, ...) ....
 </pre>
 <h2>Cache</h2>
 <p>
@@ -353,10 +353,10 @@ uncompressed block contents.
 <pre>
  #include "leveldb/cache.h"

-  leveldb::Options options;
-  options.cache = leveldb::NewLRUCache(100 * 1048576);  // 100MB cache
-  leveldb::DB* db;
-  leveldb::DB::Open(options, name, &db);
+  rocksdb::Options options;
+  options.cache = rocksdb::NewLRUCache(100 * 1048576);  // 100MB cache
+  rocksdb::DB* db;
+  rocksdb::DB::Open(options, name, &db);
  ... use the db ...
  delete db
  delete options.cache;
@@ -373,9 +373,9 @@ displacing most of the cached contents.  A per-iterator option can be
 used to achieve this:
 <p>
 <pre>
-  leveldb::ReadOptions options;
+  rocksdb::ReadOptions options;
  options.fill_cache = false;
-  leveldb::Iterator* it = db-&gt;NewIterator(options);
+  rocksdb::Iterator* it = db-&gt;NewIterator(options);
  for (it-&gt;SeekToFirst(); it-&gt;Valid(); it-&gt;Next()) {
    ...
  }
@@ -407,10 +407,10 @@ a single <code>Get()</code> call may involve multiple reads from disk.
 The optional <code>FilterPolicy</code> mechanism can be used to reduce
 the number of disk reads substantially.
 <pre>
-   leveldb::Options options;
+   rocksdb::Options options;
   options.filter_policy = NewBloomFilter(10);
-   leveldb::DB* db;
-   leveldb::DB::Open(options, "/tmp/testdb", &amp;db);
+   rocksdb::DB* db;
+   rocksdb::DB::Open(options, "/tmp/testdb", &amp;db);
   ... use the database ...
   delete db;
   delete options.filter_policy;
@@ -434,7 +434,7 @@ consider a comparator that ignores trailing spaces when comparing keys.
 Instead, the application should provide a custom filter policy that
 also ignores trailing spaces.  For example:
 <pre>
-  class CustomFilterPolicy : public leveldb::FilterPolicy {
+  class CustomFilterPolicy : public rocksdb::FilterPolicy {
   private:
    FilterPolicy* builtin_policy_;
   public:
@@ -484,7 +484,7 @@ checksums are verified:
  parts of its persistent storage have been corrupted.
 <p>
  If a database is corrupted (perhaps it cannot be opened when
-  paranoid checking is turned on), the <code>leveldb::RepairDB</code> function
+  paranoid checking is turned on), the <code>rocksdb::RepairDB</code> function
  may be used to recover as much of the data as possible
 <p>
 </ul>
@@ -494,11 +494,11 @@ The <code>GetApproximateSizes</code> method can used to get the approximate
 number of bytes of file system space used by one or more key ranges.
 <p>
 <pre>
-   leveldb::Range ranges[2];
-   ranges[0] = leveldb::Range("a", "c");
-   ranges[1] = leveldb::Range("x", "z");
+   rocksdb::Range ranges[2];
+   ranges[0] = rocksdb::Range("a", "c");
+   ranges[1] = rocksdb::Range("x", "z");
   uint64_t sizes[2];
-   leveldb::Status s = db-&gt;GetApproximateSizes(ranges, 2, sizes);
+   rocksdb::Status s = db-&gt;GetApproximateSizes(ranges, 2, sizes);
 </pre>
 The preceding call will set <code>sizes[0]</code> to the approximate number of
 bytes of file system space used by the key range <code>[a..c)</code> and
@@ -508,21 +508,21 @@ bytes of file system space used by the key range <code>[a..c)</code> and
 <h1>Environment</h1>
 <p>
 All file operations (and other operating system calls) issued by the
-<code>leveldb</code> implementation are routed through a <code>leveldb::Env</code> object.
+<code>leveldb</code> implementation are routed through a <code>rocksdb::Env</code> object.
 Sophisticated clients may wish to provide their own <code>Env</code>
 implementation to get better control.  For example, an application may
 introduce artificial delays in the file IO paths to limit the impact
 of <code>leveldb</code> on other activities in the system.
 <p>
 <pre>
-  class SlowEnv : public leveldb::Env {
+  class SlowEnv : public rocksdb::Env {
    .. implementation of the Env interface ...
  };

  SlowEnv env;
-  leveldb::Options options;
+  rocksdb::Options options;
  options.env = &amp;env;
-  Status s = leveldb::DB::Open(options, ...);
+  Status s = rocksdb::DB::Open(options, ...);
 </pre>
 <h1>Porting</h1>
 <p>
@@ -531,7 +531,7 @@ specific implementations of the types/methods/functions exported by
 <code>leveldb/port/port.h</code>.  See <code>leveldb/port/port_example.h</code> for more
 details.
 <p>
-In addition, the new platform may need a new default <code>leveldb::Env</code>
+In addition, the new platform may need a new default <code>rocksdb::Env</code>
 implementation.  See <code>leveldb/util/env_posix.h</code> for an example.

 <h1>Other Information</h1>
@@ -1,4 +1,4 @@
-This directory contains the hdfs extensions needed to make leveldb store
+This directory contains the hdfs extensions needed to make rocksdb store
 files in HDFS.

 The hdfs.h file is copied from the Apache Hadoop 1.0 source code. 
@@ -7,12 +7,12 @@ It defines the libhdfs library
 data in HDFS.  The libhdfs.a is copied from the Apache Hadoop 1.0 build. 
 It implements the API defined in hdfs.h. If your hadoop cluster is running
 a different hadoop release, then install these two files manually from your
-hadoop distribution and then recompile leveldb.
+hadoop distribution and then recompile rocksdb.

-The env_hdfs.h file defines the leveldb objects that are needed to talk to an
+The env_hdfs.h file defines the rocksdb objects that are needed to talk to an
 underlying filesystem. 

-If you want to compile leveldb with hdfs support, please set the following
+If you want to compile rocksdb with hdfs support, please set the following
 enviroment variables appropriately:
   USE_HDFS=1
   JAVA_HOME=/usr/local/jdk-6u22-64
@@ -1,37 +1,22 @@
-/**
- * Licensed to the Apache Software Foundation (ASF) under one
- * or more contributor license agreements.  See the NOTICE file
- * distributed with this work for additional information
- * regarding copyright ownership.  The ASF licenses this file
- * to you under the Apache License, Version 2.0 (the
- * "License"); you may not use this file except in compliance
- * with the License.  You may obtain a copy of the License at
- *
- *     http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-// Copyright (c) 2012 Facebook. All rights reserved.
-
-#ifndef LEVELDB_HDFS_FILE_H
-#define LEVELDB_HDFS_FILE_H
+//  Copyright (c) 2013, 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.
+//

+#pragma once
 #include <algorithm>
 #include <stdio.h>
 #include <sys/time.h>
 #include <time.h>
 #include <iostream>
-#include "leveldb/env.h"
-#include "leveldb/status.h"
+#include "rocksdb/env.h"
+#include "rocksdb/status.h"

 #ifdef USE_HDFS
 #include "hdfs/hdfs.h"

-namespace leveldb {
+namespace rocksdb {

 static const std::string kProto = "hdfs://";
 static const std::string pathsep = "/";
@@ -55,7 +40,7 @@ private:
 };

 //
-// The HDFS environment for leveldb. This class overrides all the
+// The HDFS environment for rocksdb. This class overrides all the
 // file/dir access methods and delegates the thread-mgmt methods to the
 // default posix environment.
 //
@@ -81,6 +66,10 @@ class HdfsEnv : public Env {
  virtual Status NewWritableFile(const std::string& fname,
                                 WritableFile** result);

+  virtual Status NewRandomRWFile(const std::string& fname,
+                                 unique_ptr<RandomRWFile>* result,
+                                 const EnvOptions& options);
+
  virtual bool FileExists(const std::string& fname);

  virtual Status GetChildren(const std::string& path,
@@ -107,8 +96,9 @@ class HdfsEnv : public Env {

  virtual Status NewLogger(const std::string& fname, Logger** result);

-  virtual void Schedule( void (*function)(void* arg), void* arg) {
-    posixEnv->Schedule(function, arg);
+  virtual void Schedule(void (*function)(void* arg), void* arg,
+                        Priority pri = LOW) {
+    posixEnv->Schedule(function, arg, pri);
  }

  virtual void StartThread(void (*function)(void* arg), void* arg) {
@@ -140,8 +130,8 @@ class HdfsEnv : public Env {
    return posixEnv->GetAbsolutePath(db_path, output_path);
  }

-  virtual void SetBackgroundThreads(int number) {
-    posixEnv->SetBackgroundThreads(number);
+  virtual void SetBackgroundThreads(int number, Priority pri = LOW) {
+    posixEnv->SetBackgroundThreads(number, pri);
  }

  virtual std::string TimeToString(uint64_t number) {
@@ -213,12 +203,12 @@ class HdfsEnv : public Env {
  }
 };

-}  // namespace leveldb
+}  // namespace rocksdb

 #else // USE_HDFS


-namespace leveldb {
+namespace rocksdb {

 static const Status notsup;

@@ -226,7 +216,7 @@ class HdfsEnv : public Env {

 public:
  HdfsEnv(const std::string& fsname) {
-    fprintf(stderr, "You have not build leveldb with HDFS support\n");
+    fprintf(stderr, "You have not build rocksdb with HDFS support\n");
    fprintf(stderr, "Please see hdfs/README for details\n");
    throw new std::exception();
  }
@@ -250,6 +240,12 @@ class HdfsEnv : public Env {
    return notsup;
  }

+  virtual Status NewRandomRWFile(const std::string& fname,
+                                 unique_ptr<RandomRWFile>* result,
+                                 const EnvOptions& options) {
+    return notsup;
+  }
+
  virtual bool FileExists(const std::string& fname){return false;}

  virtual Status GetChildren(const std::string& path,
@@ -279,7 +275,8 @@ class HdfsEnv : public Env {
  virtual Status NewLogger(const std::string& fname,
                           shared_ptr<Logger>* result){return notsup;}

-  virtual void Schedule( void (*function)(void* arg), void* arg) {}
+  virtual void Schedule(void (*function)(void* arg), void* arg,
+                        Priority pri = LOW) {}

  virtual void StartThread(void (*function)(void* arg), void* arg) {}

@@ -296,12 +293,10 @@ class HdfsEnv : public Env {
  virtual Status GetAbsolutePath(const std::string& db_path,
      std::string* outputpath) {return notsup;}

-  virtual void SetBackgroundThreads(int number) {}
+  virtual void SetBackgroundThreads(int number, Priority pri = LOW) {}

  virtual std::string TimeToString(uint64_t number) { return "";}
 };
 }

 #endif // USE_HDFS
-
-#endif // LEVELDB_HDFS_FILE_H
@@ -1,21 +1,8 @@
-/**
- * Licensed to the Apache Software Foundation (ASF) under one
- * or more contributor license agreements.  See the NOTICE file
- * distributed with this work for additional information
- * regarding copyright ownership.  The ASF licenses this file
- * to you under the Apache License, Version 2.0 (the
- * "License"); you may not use this file except in compliance
- * with the License.  You may obtain a copy of the License at
- *
- *     http://www.apache.org/licenses/LICENSE-2.0
- *
- * Unless required by applicable law or agreed to in writing, software
- * distributed under the License is distributed on an "AS IS" BASIS,
- * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- * See the License for the specific language governing permissions and
- * limitations under the License.
- */
-
+//  Copyright (c) 2013, 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.
+//
 #ifndef LIBHDFS_HDFS_H
 #define LIBHDFS_HDFS_H

@@ -4,8 +4,8 @@

 #include "helpers/memenv/memenv.h"

-#include "leveldb/env.h"
-#include "leveldb/status.h"
+#include "rocksdb/env.h"
+#include "rocksdb/status.h"
 #include "port/port.h"
 #include "util/mutexlock.h"
 #include <map>
@@ -13,7 +13,7 @@
 #include <string>
 #include <vector>

-namespace leveldb {
+namespace rocksdb {

 namespace {

@@ -383,4 +383,4 @@ Env* NewMemEnv(Env* base_env) {
  return new InMemoryEnv(base_env);
 }

-}  // namespace leveldb
+}  // namespace rocksdb
@@ -2,10 +2,9 @@
 // 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.

-#ifndef STORAGE_LEVELDB_HELPERS_MEMENV_MEMENV_H_
-#define STORAGE_LEVELDB_HELPERS_MEMENV_MEMENV_H_
-
-namespace leveldb {
+#ifndef STORAGE_ROCKSDB_HELPERS_MEMENV_MEMENV_H_
+#define STORAGE_ROCKSDB_HELPERS_MEMENV_MEMENV_H_
+namespace rocksdb {

 class Env;

@@ -15,6 +14,6 @@ class Env;
 // *base_env must remain live while the result is in use.
 Env* NewMemEnv(Env* base_env);

-}  // namespace leveldb
+}  // namespace rocksdb

-#endif  // STORAGE_LEVELDB_HELPERS_MEMENV_MEMENV_H_
+#endif  // STORAGE_ROCKSDB_HELPERS_MEMENV_MEMENV_H_
@@ -5,14 +5,14 @@
 #include "helpers/memenv/memenv.h"

 #include "db/db_impl.h"
-#include "leveldb/db.h"
-#include "leveldb/env.h"
+#include "rocksdb/db.h"
+#include "rocksdb/env.h"
 #include "util/testharness.h"
 #include <memory>
 #include <string>
 #include <vector>

-namespace leveldb {
+namespace rocksdb {

 class MemEnvTest {
 public:
@@ -215,7 +215,7 @@ TEST(MemEnvTest, DBTest) {
  delete iterator;

  DBImpl* dbi = reinterpret_cast<DBImpl*>(db);
-  ASSERT_OK(dbi->TEST_CompactMemTable());
+  ASSERT_OK(dbi->TEST_FlushMemTable());

  for (size_t i = 0; i < 3; ++i) {
    std::string res;
@@ -226,8 +226,8 @@ TEST(MemEnvTest, DBTest) {
  delete db;
 }

-}  // namespace leveldb
+}  // namespace rocksdb

 int main(int argc, char** argv) {
-  return leveldb::test::RunAllTests();
+  return rocksdb::test::RunAllTests();
 }
@@ -1,14 +0,0 @@
-// Copyright 2008-present Facebook. All Rights Reserved.
-#ifndef STORAGE_LEVELDB_INCLUDE_LDB_TOOL_H
-#define STORAGE_LEVELDB_INCLUDE_LDB_TOOL_H
-#include "leveldb/options.h"
-
-namespace leveldb {
-
-class LDBTool {
- public:
-  void Run(int argc, char** argv, Options = Options());
-};
-
-} // namespace leveldb
-#endif // STORAGE_LEVELDB_INCLUDE_LDB_TOOL_H
@@ -1,88 +0,0 @@
-// This file contains the interface that must be implemented by any collection
-// to be used as the backing store for a MemTable. Such a collection must
-// satisfy the following properties:
-//  (1) It does not store duplicate items.
-//  (2) It uses MemTableRep::KeyComparator to compare items for iteration and
-//     equality.
-//  (3) It can be accessed concurrently by multiple readers but need not support
-//      concurrent writes.
-//  (4) Items are never deleted.
-// The liberal use of assertions is encouraged to enforce (1).
-
-#ifndef STORAGE_LEVELDB_DB_TABLE_H_
-#define STORAGE_LEVELDB_DB_TABLE_H_
-
-#include <memory>
-#include "leveldb/arena.h"
-
-namespace leveldb {
-
-class MemTableRep {
- public:
-  // KeyComparator(a, b) returns a negative value if a is less than b, 0 if they
-  // are equal, and a positive value if b is greater than a
-  class KeyComparator {
-  public:
-    virtual int operator()(const char* a, const char* b) const = 0;
-    virtual ~KeyComparator() { }
-  };
-
-  // Insert key into the collection. (The caller will pack key and value into a
-  // single buffer and pass that in as the parameter to Insert)
-  // REQUIRES: nothing that compares equal to key is currently in the
-  // collection.
-  virtual void Insert(const char* key) = 0;
-
-  // Returns true iff an entry that compares equal to key is in the collection.
-  virtual bool Contains(const char* key) const = 0;
-
-  virtual ~MemTableRep() { }
-
-  // Iteration over the contents of a skip collection
-  class Iterator {
-   public:
-    // Initialize an iterator over the specified collection.
-    // The returned iterator is not valid.
-    // explicit Iterator(const MemTableRep* collection);
-    virtual ~Iterator() { };
-
-    // Returns true iff the iterator is positioned at a valid node.
-    virtual bool Valid() const = 0;
-
-    // Returns the key at the current position.
-    // REQUIRES: Valid()
-    virtual const char* key() const = 0;
-
-    // Advances to the next position.
-    // REQUIRES: Valid()
-    virtual void Next() = 0;
-
-    // Advances to the previous position.
-    // REQUIRES: Valid()
-    virtual void Prev() = 0;
-
-    // Advance to the first entry with a key >= target
-    virtual void Seek(const char* target) = 0;
-
-    // Position at the first entry in collection.
-    // Final state of iterator is Valid() iff collection is not empty.
-    virtual void SeekToFirst() = 0;
-
-    // Position at the last entry in collection.
-    // Final state of iterator is Valid() iff collection is not empty.
-    virtual void SeekToLast() = 0;
-  };
-
-  virtual std::shared_ptr<Iterator> GetIterator() = 0;
-};
-
-class MemTableRepFactory {
- public:
-  virtual ~MemTableRepFactory() { };
-  virtual std::shared_ptr<MemTableRep> CreateMemTableRep(
-    MemTableRep::KeyComparator&, Arena* arena) = 0;
-};
-
-}
-
-#endif // STORAGE_LEVELDB_DB_TABLE_H_
@@ -1,24 +0,0 @@
-#ifndef STORAGE_LEVELDB_INCLUDE_PERF_CONTEXT_H
-#define STORAGE_LEVELDB_INCLUDE_PERF_CONTEXT_H
-
-#include <stdint.h>
-
-namespace leveldb {
-
-// A thread local context for gathering performance counter efficiently
-// and transparently.
-
-struct PerfContext {
-
-  void Reset(); // reset all performance counters to zero
-
-
-  uint64_t user_key_comparison_count; // total number of user key comparisons
-};
-
-extern __thread PerfContext perf_context;
-
-}
-
-
-#endif
@@ -1,14 +0,0 @@
-#ifndef STORAGE_LEVELDB_INCLUDE_TYPES_H_
-#define STORAGE_LEVELDB_INCLUDE_TYPES_H_
-
-#include <stdint.h>
-
-namespace leveldb {
-
-// Define all public custom types here.
-
-// Represents a sequence number in a WAL file.
-typedef uint64_t SequenceNumber;
-
-}  //  namespace leveldb
-#endif //  STORAGE_LEVELDB_INCLUDE_TYPES_H_
@@ -1,57 +0,0 @@
-// 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.
-
-#ifndef STORAGE_ROCKSDB_UNIVERSAL_COMPACTION_OPTIONS_H
-#define STORAGE_ROCKSDB_UNIVERSAL_COMPACTION_OPTIONS_H
-
-#include <stddef.h>
-#include <string>
-#include <memory>
-#include <vector>
-#include <stdint.h>
-#include <climits>
-#include "leveldb/slice.h"
-#include "leveldb/statistics.h"
-
-namespace leveldb {
-
-//
-// Algorithm used to make a compaction request stop picking new files
-// into a single compaction run
-//
-enum CompactionStopStyle {
-  kCompactionStopStyleSimilarSize, // pick files of similar size
-  kCompactionStopStyleTotalSize    // total size of picked files > next file
-};
-
-class CompactionOptionsUniversal {
- public:
-
-  // Percentage flexibilty while comparing file size. If the candidate file(s)
-  // size is 1% smaller than the next file's size, then include next file into
-  // this candidate set. // Default: 1
-  unsigned int size_ratio;
-
-  // The minimum number of files in a single compaction run. Default: 2
-  unsigned int min_merge_width;
-
-  // The maximum number of files in a single compaction run. Default: INT_MAX
-  unsigned int max_merge_width;
-
-  // The algorithm used to stop picking files into a single compaction run
-  // Default: kCompactionStopStyleTotalSize
-  CompactionStopStyle stop_style;
-
-  // Default set of parameters
-  CompactionOptionsUniversal() :
-    size_ratio(1),
-    min_merge_width(2),
-    max_merge_width(UINT_MAX),
-    stop_style(kCompactionStopStyleTotalSize) {
-  }
-};
-
-}  // namespace leveldb
-
-#endif  // STORAGE_ROCKSDB_UNIVERSAL_COMPACTION_OPTIONS_H
@@ -5,10 +5,13 @@
 // Arena class defines memory allocation methods. It's used by memtable and
 // skiplist.

-#ifndef STORAGE_LEVELDB_INCLUDE_ARENA_H_
-#define STORAGE_LEVELDB_INCLUDE_ARENA_H_
+#ifndef STORAGE_ROCKSDB_INCLUDE_ARENA_H_
+#define STORAGE_ROCKSDB_INCLUDE_ARENA_H_

-namespace leveldb {
+#include <limits>
+#include <memory>
+
+namespace rocksdb {

 class Arena {
 public:
@@ -33,6 +36,6 @@ class Arena {
  void operator=(const Arena&);
 };

-}  // namespace leveldb
+}  // namespace rocksdb

-#endif  // STORAGE_LEVELDB_INCLUDE_ARENA_H_
+#endif  // STORAGE_ROCKSDB_INCLUDE_ARENA_H_
@@ -37,8 +37,8 @@
  (5) All of the pointer arguments must be non-NULL.
 */

-#ifndef STORAGE_LEVELDB_INCLUDE_C_H_
-#define STORAGE_LEVELDB_INCLUDE_C_H_
+#ifndef STORAGE_ROCKSDB_INCLUDE_C_H_
+#define STORAGE_ROCKSDB_INCLUDE_C_H_

 #ifdef __cplusplus
 extern "C" {
@@ -278,4 +278,4 @@ extern void leveldb_env_destroy(leveldb_env_t*);
 }  /* end extern "C" */
 #endif

-#endif  /* STORAGE_LEVELDB_INCLUDE_C_H_ */
+#endif  /* STORAGE_ROCKSDB_INCLUDE_C_H_ */
@@ -15,23 +15,35 @@
 // they want something more sophisticated (like scan-resistance, a
 // custom eviction policy, variable cache sizing, etc.)

-#ifndef STORAGE_LEVELDB_INCLUDE_CACHE_H_
-#define STORAGE_LEVELDB_INCLUDE_CACHE_H_
+#ifndef STORAGE_ROCKSDB_INCLUDE_CACHE_H_
+#define STORAGE_ROCKSDB_INCLUDE_CACHE_H_

 #include <memory>
 #include <stdint.h>
-#include "leveldb/slice.h"
+#include "rocksdb/slice.h"

-namespace leveldb {
+namespace rocksdb {

 using std::shared_ptr;

 class Cache;

-// Create a new cache with a fixed size capacity.  This implementation
-// of Cache uses a least-recently-used eviction policy.
+// Create a new cache with a fixed size capacity. The cache is sharded
+// to 2^numShardBits shards, by hash of the key. The total capacity
+// is divided and evenly assigned to each shard. Inside each shard,
+// the eviction is done in two passes: first try to free spaces by
+// evicting entries that are among the most least used removeScanCountLimit
+// entries and do not have reference other than by the cache itself, in
+// the least-used order. If not enough space is freed, further free the
+// entries in least used order.
+//
+// The functions without parameter numShardBits and/or removeScanCountLimit
+// use default values. removeScanCountLimit's default value is 0, which
+// means a strict LRU order inside each shard.
 extern shared_ptr<Cache> NewLRUCache(size_t capacity);
 extern shared_ptr<Cache> NewLRUCache(size_t capacity, int numShardBits);
+extern shared_ptr<Cache> NewLRUCache(size_t capacity, int numShardBits,
+                                     int removeScanCountLimit);

 class Cache {
 public:
@@ -101,6 +113,6 @@ class Cache {
  void operator=(const Cache&);
 };

-}  // namespace leveldb
+}  // namespace rocksdb

-#endif  // STORAGE_LEVELDB_UTIL_CACHE_H_
+#endif  // STORAGE_ROCKSDB_UTIL_CACHE_H_
@@ -2,12 +2,12 @@
 // 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.

-#ifndef STORAGE_LEVELDB_INCLUDE_COMPACTION_FILTER_H_
-#define STORAGE_LEVELDB_INCLUDE_COMPACTION_FILTER_H_
+#ifndef STORAGE_ROCKSDB_INCLUDE_COMPACTION_FILTER_H_
+#define STORAGE_ROCKSDB_INCLUDE_COMPACTION_FILTER_H_

 #include <string>

-namespace leveldb {
+namespace rocksdb {

 class Slice;

@@ -66,6 +66,6 @@ class DefaultCompactionFilterFactory : public CompactionFilterFactory {
    }
 };

-}  // namespace leveldb
+}  // namespace rocksdb

-#endif  // STORAGE_LEVELDB_INCLUDE_COMPACTION_FILTER_H_
+#endif  // STORAGE_ROCKSDB_INCLUDE_COMPACTION_FILTER_H_
@@ -2,18 +2,18 @@
 // 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.

-#ifndef STORAGE_LEVELDB_INCLUDE_COMPARATOR_H_
-#define STORAGE_LEVELDB_INCLUDE_COMPARATOR_H_
+#ifndef STORAGE_ROCKSDB_INCLUDE_COMPARATOR_H_
+#define STORAGE_ROCKSDB_INCLUDE_COMPARATOR_H_

 #include <string>

-namespace leveldb {
+namespace rocksdb {

 class Slice;

 // A Comparator object provides a total order across slices that are
 // used as keys in an sstable or a database.  A Comparator implementation
-// must be thread-safe since leveldb may invoke its methods concurrently
+// must be thread-safe since rocksdb may invoke its methods concurrently
 // from multiple threads.
 class Comparator {
 public:
@@ -33,7 +33,7 @@ class Comparator {
  // the comparator implementation changes in a way that will cause
  // the relative ordering of any two keys to change.
  //
-  // Names starting with "leveldb." are reserved and should not be used
+  // Names starting with "rocksdb." are reserved and should not be used
  // by any clients of this package.
  virtual const char* Name() const = 0;

@@ -58,6 +58,6 @@ class Comparator {
 // must not be deleted.
 extern const Comparator* BytewiseComparator();

-}  // namespace leveldb
+}  // namespace rocksdb

-#endif  // STORAGE_LEVELDB_INCLUDE_COMPARATOR_H_
+#endif  // STORAGE_ROCKSDB_INCLUDE_COMPARATOR_H_
@@ -2,19 +2,19 @@
 // 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.

-#ifndef STORAGE_LEVELDB_INCLUDE_DB_H_
-#define STORAGE_LEVELDB_INCLUDE_DB_H_
+#ifndef STORAGE_ROCKSDB_INCLUDE_DB_H_
+#define STORAGE_ROCKSDB_INCLUDE_DB_H_

 #include <stdint.h>
 #include <stdio.h>
 #include <memory>
 #include <vector>
-#include "leveldb/iterator.h"
-#include "leveldb/options.h"
-#include "leveldb/types.h"
-#include "leveldb/transaction_log.h"
+#include "rocksdb/iterator.h"
+#include "rocksdb/options.h"
+#include "rocksdb/types.h"
+#include "rocksdb/transaction_log.h"

-namespace leveldb {
+namespace rocksdb {

 using std::unique_ptr;

@@ -28,6 +28,17 @@ struct WriteOptions;
 struct FlushOptions;
 class WriteBatch;

+// Metadata associated with each SST file.
+struct LiveFileMetaData {
+  std::string name;        // Name of the file
+  int level;               // Level at which this file resides.
+  size_t size;             // File size in bytes.
+  std::string smallestkey; // Smallest user defined key in the file.
+  std::string largestkey;  // Largest user defined key in the file.
+  SequenceNumber smallest_seqno; // smallest seqno in file
+  SequenceNumber largest_seqno;  // largest seqno in file
+};
+
 // Abstract handle to particular state of a DB.
 // A Snapshot is an immutable object and can therefore be safely
 // accessed from multiple threads without any external synchronization.
@@ -164,11 +175,11 @@ class DB {
  //
  // Valid property names include:
  //
-  //  "leveldb.num-files-at-level<N>" - return the number of files at level <N>,
+  //  "rocksdb.num-files-at-level<N>" - return the number of files at level <N>,
  //     where <N> is an ASCII representation of a level number (e.g. "0").
-  //  "leveldb.stats" - returns a multi-line string that describes statistics
+  //  "rocksdb.stats" - returns a multi-line string that describes statistics
  //     about the internal operation of the DB.
-  //  "leveldb.sstables" - returns a multi-line string that describes all
+  //  "rocksdb.sstables" - returns a multi-line string that describes all
  //     of the sstables that make up the db contents.
  virtual bool GetProperty(const Slice& property, std::string* value) = 0;

@@ -198,9 +209,10 @@ class DB {
  // after compaction is reduced, that level might not be appropriate for
  // hosting all the files. In this case, client could set reduce_level
  // to true, to move the files back to the minimum level capable of holding
-  // the data set.
+  // the data set or a given level (specified by non-negative target_level).
  virtual void CompactRange(const Slice* begin, const Slice* end,
-                            bool reduce_level = false) = 0;
+                            bool reduce_level = false,
+                            int target_level = -1) = 0;

  // Number of levels used for this DB.
  virtual int NumberLevels() = 0;
@@ -223,21 +235,30 @@ class DB {
  // Allow compactions to delete obselete files.
  virtual Status EnableFileDeletions() = 0;

+  // GetLiveFiles followed by GetSortedWalFiles can generate a lossless backup
+
+  // THIS METHOD IS DEPRECATED. Use the GetTableMetaData to get more
+  // detailed information on the live files.
  // Retrieve the list of all files in the database. The files are
-  // relative to the dbname and are not absolute paths. This list
-  // can be used to generate a backup. The valid size of the manifest
-  // file is returned in manifest_file_size. The manifest file is
-  // an ever growing file, but only the portion specified
-  // by manifest_file_size is valid for this snapshot.
+  // relative to the dbname and are not absolute paths. The valid size of the
+  // manifest file is returned in manifest_file_size. The manifest file is an
+  // ever growing file, but only the portion specified by manifest_file_size is
+  // valid for this snapshot.
+  // Setting flush_memtable to true does Flush before recording the live files.
+  // Setting flush_memtable to false is useful when we don't want to wait for
+  // flush which may have to wait for compaction to complete taking an
+  // indeterminate time. But this will have to use GetSortedWalFiles after
+  // GetLiveFiles to compensate for memtables missed in this snapshot due to the
+  // absence of Flush, by WAL files to recover the database consistently later
  virtual Status GetLiveFiles(std::vector<std::string>&,
-                              uint64_t* manifest_file_size) = 0;
+                              uint64_t* manifest_file_size,
+                              bool flush_memtable = true) = 0;

  // Retrieve the sorted list of all wal files with earliest file first
  virtual Status GetSortedWalFiles(VectorLogPtr& files) = 0;

  // Delete wal files in files. These can be either live or archived.
  // Returns Status::OK if all files could be deleted, otherwise Status::IOError
-  // which contains information about files that could not be deleted.
  virtual Status DeleteWalFiles(const VectorLogPtr& files) = 0;

  // The sequence number of the most recent transaction.
@@ -256,6 +277,19 @@ class DB {
  // an update is read.
  virtual Status GetUpdatesSince(SequenceNumber seq_number,
                                 unique_ptr<TransactionLogIterator>* iter) = 0;
+
+  // Delete the file name from the db directory and update the internal
+  // state to reflect that.
+  virtual Status DeleteFile(std::string name) {
+    return Status::OK();
+  }
+
+  // Returns a list of all table files with their level, start key
+  // and end key
+  virtual void GetLiveFilesMetaData(
+    std::vector<LiveFileMetaData> *metadata) {
+  }
+
 private:
  // No copying allowed
  DB(const DB&);
@@ -272,6 +306,6 @@ Status DestroyDB(const std::string& name, const Options& options);
 // on a database that contains important information.
 Status RepairDB(const std::string& dbname, const Options& options);

-}  // namespace leveldb
+}  // namespace rocksdb

-#endif  // STORAGE_LEVELDB_INCLUDE_DB_H_
+#endif  // STORAGE_ROCKSDB_INCLUDE_DB_H_
@@ -2,7 +2,7 @@
 // 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.
 //
-// An Env is an interface used by the leveldb implementation to access
+// An Env is an interface used by the rocksdb implementation to access
 // operating system functionality like the filesystem etc.  Callers
 // may wish to provide a custom Env object when opening a database to
 // get fine gain control; e.g., to rate limit file system operations.
@@ -10,17 +10,17 @@
 // All Env implementations are safe for concurrent access from
 // multiple threads without any external synchronization.

-#ifndef STORAGE_LEVELDB_INCLUDE_ENV_H_
-#define STORAGE_LEVELDB_INCLUDE_ENV_H_
+#ifndef STORAGE_ROCKSDB_INCLUDE_ENV_H_
+#define STORAGE_ROCKSDB_INCLUDE_ENV_H_

 #include <cstdarg>
 #include <string>
 #include <memory>
 #include <vector>
 #include <stdint.h>
-#include "leveldb/status.h"
+#include "rocksdb/status.h"

-namespace leveldb {
+namespace rocksdb {

 class FileLock;
 class Logger;
@@ -28,6 +28,7 @@ class RandomAccessFile;
 class SequentialFile;
 class Slice;
 class WritableFile;
+class RandomRWFile;
 class Options;

 using std::unique_ptr;
@@ -41,7 +42,7 @@ struct EnvOptions {
  EnvOptions();

  // construct from Options
-  EnvOptions(const Options& options);
+  explicit EnvOptions(const Options& options);

  // If true, then allow caching of data in environment buffers
  bool use_os_buffer;
@@ -72,7 +73,7 @@ class Env {
  // system.  Sophisticated users may wish to provide their own Env
  // implementation instead of relying on this default environment.
  //
-  // The result of Default() belongs to leveldb and must never be deleted.
+  // The result of Default() belongs to rocksdb and must never be deleted.
  static Env* Default();

  // Create a brand new sequentially-readable file with the specified name.
@@ -109,6 +110,15 @@ class Env {
                                 unique_ptr<WritableFile>* result,
                                 const EnvOptions& options) = 0;

+  // Create an object that both reads and writes to a file on
+  // specified offsets (random access). If file already exists,
+  // does not overwrite it. On success, stores a pointer to the
+  // new file in *result and returns OK. On failure stores nullptr
+  // in *result and returns non-OK.
+  virtual Status NewRandomRWFile(const std::string& fname,
+                                 unique_ptr<RandomRWFile>* result,
+                                 const EnvOptions& options) = 0;
+
  // Returns true iff the named file exists.
  virtual bool FileExists(const std::string& fname) = 0;

@@ -162,15 +172,20 @@ class Env {
  // REQUIRES: lock has not already been unlocked.
  virtual Status UnlockFile(FileLock* lock) = 0;

-  // Arrange to run "(*function)(arg)" once in a background thread.
-  //
+  enum Priority { LOW, HIGH, TOTAL };
+
+  // Arrange to run "(*function)(arg)" once in a background thread, in
+  // the thread pool specified by pri. By default, jobs go to the 'LOW'
+  // priority thread pool.
+
  // "function" may run in an unspecified thread.  Multiple functions
  // added to the same Env may run concurrently in different threads.
  // I.e., the caller may not assume that background work items are
  // serialized.
  virtual void Schedule(
      void (*function)(void* arg),
-      void* arg) = 0;
+      void* arg,
+      Priority pri = LOW) = 0;

  // Start a new thread, invoking "function(arg)" within the new thread.
  // When "function(arg)" returns, the thread will be destroyed.
@@ -190,6 +205,13 @@ class Env {
  // useful for computing deltas of time.
  virtual uint64_t NowMicros() = 0;

+  // Returns the number of nano-seconds since some fixed point in time. Only
+  // useful for computing deltas of time in one run.
+  // Default implementation simply relies on NowMicros
+  virtual uint64_t NowNanos() {
+    return NowMicros() * 1000;
+  }
+
  // Sleep/delay the thread for the perscribed number of micro-seconds.
  virtual void SleepForMicroseconds(int micros) = 0;

@@ -203,13 +225,17 @@ class Env {
  virtual Status GetAbsolutePath(const std::string& db_path,
      std::string* output_path) = 0;

-  // The number of background worker threads for this environment.
-  // default: 1
-  virtual void SetBackgroundThreads(int number) = 0;
+  // The number of background worker threads of a specific thread pool
+  // for this environment. 'LOW' is the default pool.
+  // default number: 1
+  virtual void SetBackgroundThreads(int number, Priority pri = LOW) = 0;

  // Converts seconds-since-Jan-01-1970 to a printable string
  virtual std::string TimeToString(uint64_t time) = 0;

+  // Generates a unique id that can be used to identify a db
+  virtual std::string GenerateUniqueId();
+
 private:
  // No copying allowed
  Env(const Env&);
@@ -240,6 +266,13 @@ class SequentialFile {
  //
  // REQUIRES: External synchronization
  virtual Status Skip(uint64_t n) = 0;
+
+  // Remove any kind of caching of data from the offset to offset+length
+  // of this file. If the length is 0, then it refers to the end of file.
+  // If the system is not caching the file contents, then this is a noop.
+  virtual Status InvalidateCache(size_t offset, size_t length) {
+    return Status::NotSupported("InvalidateCache not supported.");
+  }
 };

 // A file abstraction for randomly reading the contents of a file.
@@ -285,6 +318,12 @@ class RandomAccessFile {

  virtual void Hint(AccessPattern pattern) {}

+  // Remove any kind of caching of data from the offset to offset+length
+  // of this file. If the length is 0, then it refers to the end of file.
+  // If the system is not caching the file contents, then this is a noop.
+  virtual Status InvalidateCache(size_t offset, size_t length) {
+    return Status::NotSupported("InvalidateCache not supported.");
+  }
 };

 // A file abstraction for sequential writing.  The implementation
@@ -303,7 +342,7 @@ class WritableFile {

  /*
   * Sync data and/or metadata as well.
-   * By default, sync only metadata.
+   * By default, sync only data.
   * Override this method for environments where we need to sync
   * metadata as well.
   */
@@ -334,6 +373,14 @@ class WritableFile {
    *block_size = preallocation_block_size_;
  }

+  // Remove any kind of caching of data from the offset to offset+length
+  // of this file. If the length is 0, then it refers to the end of file.
+  // If the system is not caching the file contents, then this is a noop.
+  // This call has no effect on dirty pages in the cache.
+  virtual Status InvalidateCache(size_t offset, size_t length) {
+    return Status::NotSupported("InvalidateCache not supported.");
+  }
+
 protected:
  // PrepareWrite performs any necessary preparation for a write
  // before the write actually occurs.  This allows for pre-allocation
@@ -384,6 +431,55 @@ class WritableFile {
  void operator=(const WritableFile&);
 };

+// A file abstraction for random reading and writing.
+class RandomRWFile {
+ public:
+  RandomRWFile() {}
+  virtual ~RandomRWFile() {}
+
+  // Write data from Slice data to file starting from offset
+  // Returns IOError on failure, but does not guarantee
+  // atomicity of a write.  Returns OK status on success.
+  //
+  // Safe for concurrent use.
+  virtual Status Write(uint64_t offset, const Slice& data) = 0;
+  // Read up to "n" bytes from the file starting at "offset".
+  // "scratch[0..n-1]" may be written by this routine.  Sets "*result"
+  // to the data that was read (including if fewer than "n" bytes were
+  // successfully read).  May set "*result" to point at data in
+  // "scratch[0..n-1]", so "scratch[0..n-1]" must be live when
+  // "*result" is used.  If an error was encountered, returns a non-OK
+  // status.
+  //
+  // Safe for concurrent use by multiple threads.
+  virtual Status Read(uint64_t offset, size_t n, Slice* result,
+                      char* scratch) const = 0;
+  virtual Status Close() = 0; // closes the file
+  virtual Status Sync() = 0; // sync data
+
+  /*
+   * Sync data and/or metadata as well.
+   * By default, sync only data.
+   * Override this method for environments where we need to sync
+   * metadata as well.
+   */
+  virtual Status Fsync() {
+    return Sync();
+  }
+
+  /*
+   * Pre-allocate space for a file.
+   */
+  virtual Status Allocate(off_t offset, off_t len) {
+    return Status::OK();
+  }
+
+ private:
+  // No copying allowed
+  RandomRWFile(const RandomRWFile&);
+  void operator=(const RandomRWFile&);
+};
+
 // An interface for writing log messages.
 class Logger {
 public:
@@ -463,6 +559,10 @@ class EnvWrapper : public Env {
                         const EnvOptions& options) {
    return target_->NewWritableFile(f, r, options);
  }
+  Status NewRandomRWFile(const std::string& f, unique_ptr<RandomRWFile>* r,
+                         const EnvOptions& options) {
+    return target_->NewRandomRWFile(f, r, options);
+  }
  bool FileExists(const std::string& f) { return target_->FileExists(f); }
  Status GetChildren(const std::string& dir, std::vector<std::string>* r) {
    return target_->GetChildren(dir, r);
@@ -489,8 +589,8 @@ class EnvWrapper : public Env {
    return target_->LockFile(f, l);
  }
  Status UnlockFile(FileLock* l) { return target_->UnlockFile(l); }
-  void Schedule(void (*f)(void*), void* a) {
-    return target_->Schedule(f, a);
+  void Schedule(void (*f)(void*), void* a, Priority pri) {
+    return target_->Schedule(f, a, pri);
  }
  void StartThread(void (*f)(void*), void* a) {
    return target_->StartThread(f, a);
@@ -518,8 +618,8 @@ class EnvWrapper : public Env {
      std::string* output_path) {
    return target_->GetAbsolutePath(db_path, output_path);
  }
-  void SetBackgroundThreads(int num) {
-    return target_->SetBackgroundThreads(num);
+  void SetBackgroundThreads(int num, Priority pri) {
+    return target_->SetBackgroundThreads(num, pri);
  }
  std::string TimeToString(uint64_t time) {
    return target_->TimeToString(time);
@@ -529,6 +629,6 @@ class EnvWrapper : public Env {
  Env* target_;
 };

-}  // namespace leveldb
+}  // namespace rocksdb

-#endif  // STORAGE_LEVELDB_INCLUDE_ENV_H_
+#endif  // STORAGE_ROCKSDB_INCLUDE_ENV_H_
@@ -4,8 +4,8 @@
 //
 // A database can be configured with a custom FilterPolicy object.
 // This object is responsible for creating a small filter from a set
-// of keys.  These filters are stored in leveldb and are consulted
-// automatically by leveldb to decide whether or not to read some
+// of keys.  These filters are stored in rocksdb and are consulted
+// automatically by rocksdb to decide whether or not to read some
 // information from disk. In many cases, a filter can cut down the
 // number of disk seeks form a handful to a single disk seek per
 // DB::Get() call.
@@ -13,12 +13,12 @@
 // Most people will want to use the builtin bloom filter support (see
 // NewBloomFilterPolicy() below).

-#ifndef STORAGE_LEVELDB_INCLUDE_FILTER_POLICY_H_
-#define STORAGE_LEVELDB_INCLUDE_FILTER_POLICY_H_
+#ifndef STORAGE_ROCKSDB_INCLUDE_FILTER_POLICY_H_
+#define STORAGE_ROCKSDB_INCLUDE_FILTER_POLICY_H_

 #include <string>

-namespace leveldb {
+namespace rocksdb {

 class Slice;

@@ -67,4 +67,4 @@ extern const FilterPolicy* NewBloomFilterPolicy(int bits_per_key);

 }

-#endif  // STORAGE_LEVELDB_INCLUDE_FILTER_POLICY_H_
+#endif  // STORAGE_ROCKSDB_INCLUDE_FILTER_POLICY_H_
@@ -12,13 +12,13 @@
 // non-const method, all threads accessing the same Iterator must use
 // external synchronization.

-#ifndef STORAGE_LEVELDB_INCLUDE_ITERATOR_H_
-#define STORAGE_LEVELDB_INCLUDE_ITERATOR_H_
+#ifndef STORAGE_ROCKSDB_INCLUDE_ITERATOR_H_
+#define STORAGE_ROCKSDB_INCLUDE_ITERATOR_H_

-#include "leveldb/slice.h"
-#include "leveldb/status.h"
+#include "rocksdb/slice.h"
+#include "rocksdb/status.h"

-namespace leveldb {
+namespace rocksdb {

 class Iterator {
 public:
@@ -65,6 +65,8 @@ class Iterator {
  virtual Slice value() const = 0;

  // If an error has occurred, return it.  Else return an ok status.
+  // If non-blocking IO is requested and this operation cannot be
+  // satisfied without doing some IO, then this returns Status::Incomplete().
  virtual Status status() const = 0;

  // Clients are allowed to register function/arg1/arg2 triples that
@@ -95,6 +97,6 @@ extern Iterator* NewEmptyIterator();
 // Return an empty iterator with the specified status.
 extern Iterator* NewErrorIterator(const Status& status);

-}  // namespace leveldb
+}  // namespace rocksdb

-#endif  // STORAGE_LEVELDB_INCLUDE_ITERATOR_H_
+#endif  // STORAGE_ROCKSDB_INCLUDE_ITERATOR_H_
@@ -0,0 +1,15 @@
+// Copyright 2008-present Facebook. All Rights Reserved.
+#ifndef STORAGE_ROCKSDB_INCLUDE_LDB_TOOL_H
+#define STORAGE_ROCKSDB_INCLUDE_LDB_TOOL_H
+#include "rocksdb/options.h"
+
+namespace rocksdb {
+
+class LDBTool {
+ public:
+  void Run(int argc, char** argv, Options = Options());
+};
+
+} // namespace rocksdb
+
+#endif // STORAGE_ROCKSDB_INCLUDE_LDB_TOOL_H
@@ -0,0 +1,259 @@
+// This file contains the interface that must be implemented by any collection
+// to be used as the backing store for a MemTable. Such a collection must
+// satisfy the following properties:
+//  (1) It does not store duplicate items.
+//  (2) It uses MemTableRep::KeyComparator to compare items for iteration and
+//     equality.
+//  (3) It can be accessed concurrently by multiple readers and can support
+//     during reads. However, it needn't support multiple concurrent writes.
+//  (4) Items are never deleted.
+// The liberal use of assertions is encouraged to enforce (1).
+//
+// The factory will be passed an Arena object when a new MemTableRep is
+// requested. The API for this object is in rocksdb/arena.h.
+//
+// Users can implement their own memtable representations. We include four
+// types built in:
+//  - SkipListRep: This is the default; it is backed by a skip list.
+//  - TransformRep: This is backed by an std::unordered_map<Slice,
+// std::set>. On construction, they are given a SliceTransform object. This
+// object is applied to the user key of stored items which indexes into the
+// unordered map to yield a set containing all records that share the same user
+// key under the transform function.
+//  - UnsortedRep: A subclass of TransformRep where the transform function is
+// the identity function. Optimized for point lookups.
+//  - PrefixHashRep: A subclass of TransformRep where the transform function is
+// a fixed-size prefix extractor. If you use PrefixHashRepFactory, the transform
+// must be identical to options.prefix_extractor, otherwise it will be discarded
+// and the default will be used. It is optimized for ranged scans over a
+// prefix.
+//  - VectorRep: This is backed by an unordered std::vector. On iteration, the
+// vector is sorted. It is intelligent about sorting; once the MarkReadOnly()
+// has been called, the vector will only be sorted once. It is optimized for
+// random-write-heavy workloads.
+//
+// The last four implementations are designed for situations in which
+// iteration over the entire collection is rare since doing so requires all the
+// keys to be copied into a sorted data structure.
+
+#ifndef STORAGE_ROCKSDB_DB_MEMTABLEREP_H_
+#define STORAGE_ROCKSDB_DB_MEMTABLEREP_H_
+
+#include <memory>
+#include "rocksdb/arena.h"
+#include "rocksdb/slice.h"
+#include "rocksdb/slice_transform.h"
+
+namespace rocksdb {
+
+class MemTableRep {
+ public:
+  // KeyComparator provides a means to compare keys, which are internal keys
+  // concatenated with values.
+  class KeyComparator {
+   public:
+    // Compare a and b. Return a negative value if a is less than b, 0 if they
+    // are equal, and a positive value if a is greater than b
+    virtual int operator()(const char* a, const char* b) const = 0;
+
+    virtual ~KeyComparator() { }
+  };
+
+  // Insert key into the collection. (The caller will pack key and value into a
+  // single buffer and pass that in as the parameter to Insert)
+  // REQUIRES: nothing that compares equal to key is currently in the
+  // collection.
+  virtual void Insert(const char* key) = 0;
+
+  // Returns true iff an entry that compares equal to key is in the collection.
+  virtual bool Contains(const char* key) const = 0;
+
+  // Notify this table rep that it will no longer be added to. By default, does
+  // nothing.
+  virtual void MarkReadOnly() { }
+
+  // Report an approximation of how much memory has been used other than memory
+  // that was allocated through the arena.
+  virtual size_t ApproximateMemoryUsage() = 0;
+
+  virtual ~MemTableRep() { }
+
+  // Iteration over the contents of a skip collection
+  class Iterator {
+   public:
+    // Initialize an iterator over the specified collection.
+    // The returned iterator is not valid.
+    // explicit Iterator(const MemTableRep* collection);
+    virtual ~Iterator() { };
+
+    // Returns true iff the iterator is positioned at a valid node.
+    virtual bool Valid() const = 0;
+
+    // Returns the key at the current position.
+    // REQUIRES: Valid()
+    virtual const char* key() const = 0;
+
+    // Advances to the next position.
+    // REQUIRES: Valid()
+    virtual void Next() = 0;
+
+    // Advances to the previous position.
+    // REQUIRES: Valid()
+    virtual void Prev() = 0;
+
+    // Advance to the first entry with a key >= target
+    virtual void Seek(const char* target) = 0;
+
+    // Position at the first entry in collection.
+    // Final state of iterator is Valid() iff collection is not empty.
+    virtual void SeekToFirst() = 0;
+
+    // Position at the last entry in collection.
+    // Final state of iterator is Valid() iff collection is not empty.
+    virtual void SeekToLast() = 0;
+  };
+
+  // Return an iterator over the keys in this representation.
+  virtual std::shared_ptr<Iterator> GetIterator() = 0;
+
+  // Return an iterator over at least the keys with the specified user key. The
+  // iterator may also allow access to other keys, but doesn't have to. Default:
+  // GetIterator().
+  virtual std::shared_ptr<Iterator> GetIterator(const Slice& user_key) {
+    return GetIterator();
+  }
+
+  // Return an iterator over at least the keys with the specified prefix. The
+  // iterator may also allow access to other keys, but doesn't have to. Default:
+  // GetIterator().
+  virtual std::shared_ptr<Iterator> GetPrefixIterator(const Slice& prefix) {
+    return GetIterator();
+  }
+
+ protected:
+  // When *key is an internal key concatenated with the value, returns the
+  // user key.
+  virtual Slice UserKey(const char* key) const;
+};
+
+// This is the base class for all factories that are used by RocksDB to create
+// new MemTableRep objects
+class MemTableRepFactory {
+ public:
+  virtual ~MemTableRepFactory() { };
+  virtual std::shared_ptr<MemTableRep> CreateMemTableRep(
+    MemTableRep::KeyComparator&, Arena*) = 0;
+  virtual const char* Name() const = 0;
+};
+
+// This creates MemTableReps that are backed by an std::vector. On iteration,
+// the vector is sorted. This is useful for workloads where iteration is very
+// rare and writes are generally not issued after reads begin.
+//
+// Parameters:
+//   count: Passed to the constructor of the underlying std::vector of each
+//     VectorRep. On initialization, the underlying array will be at least count
+//     bytes reserved for usage.
+class VectorRepFactory : public MemTableRepFactory {
+  const size_t count_;
+public:
+  explicit VectorRepFactory(size_t count = 0) : count_(count) { }
+  virtual std::shared_ptr<MemTableRep> CreateMemTableRep(
+    MemTableRep::KeyComparator&, Arena*) override;
+  virtual const char* Name() const override {
+    return "VectorRepFactory";
+  }
+};
+
+// This uses a skip list to store keys. It is the default.
+class SkipListFactory : public MemTableRepFactory {
+public:
+  virtual std::shared_ptr<MemTableRep> CreateMemTableRep(
+    MemTableRep::KeyComparator&, Arena*) override;
+  virtual const char* Name() const override {
+    return "SkipListFactory";
+  }
+};
+
+// TransformReps are backed by an unordered map of buffers to buckets. When
+// looking up a key, the user key is extracted and a user-supplied transform
+// function (see rocksdb/slice_transform.h) is applied to get the key into the
+// unordered map. This allows the user to bin user keys based on arbitrary
+// criteria. Two example implementations are UnsortedRepFactory and
+// PrefixHashRepFactory.
+//
+// Iteration over the entire collection is implemented by dumping all the keys
+// into an std::set. Thus, these data structures are best used when iteration
+// over the entire collection is rare.
+//
+// Parameters:
+//   transform: The SliceTransform to bucket user keys on. TransformRepFactory
+//     owns the pointer.
+//   bucket_count: Passed to the constructor of the underlying
+//     std::unordered_map of each TransformRep. On initialization, the
+//     underlying array will be at least bucket_count size.
+//   num_locks: Number of read-write locks to have for the rep. Each bucket is
+//     hashed onto a read-write lock which controls access to that lock. More
+//     locks means finer-grained concurrency but more memory overhead.
+class TransformRepFactory : public MemTableRepFactory {
+ public:
+  explicit TransformRepFactory(const SliceTransform* transform,
+    size_t bucket_count, size_t num_locks = 1000)
+    : transform_(transform),
+      bucket_count_(bucket_count),
+      num_locks_(num_locks) { }
+
+  virtual ~TransformRepFactory() { delete transform_; }
+
+  virtual std::shared_ptr<MemTableRep> CreateMemTableRep(
+    MemTableRep::KeyComparator&, Arena*) override;
+
+  virtual const char* Name() const override {
+    return "TransformRepFactory";
+  }
+
+  const SliceTransform* GetTransform() { return transform_; }
+
+ protected:
+  const SliceTransform* transform_;
+  const size_t bucket_count_;
+  const size_t num_locks_;
+};
+
+// UnsortedReps bin user keys based on an identity function transform -- that
+// is, transform(key) = key. This optimizes for point look-ups.
+//
+// Parameters: See TransformRepFactory.
+class UnsortedRepFactory : public TransformRepFactory {
+public:
+  explicit UnsortedRepFactory(size_t bucket_count = 0, size_t num_locks = 1000)
+    : TransformRepFactory(NewNoopTransform(),
+                          bucket_count,
+                          num_locks) { }
+  virtual const char* Name() const override {
+    return "UnsortedRepFactory";
+  }
+};
+
+// PrefixHashReps bin user keys based on a fixed-size prefix. This optimizes for
+// short ranged scans over a given prefix.
+//
+// Parameters: See TransformRepFactory.
+class PrefixHashRepFactory : public TransformRepFactory {
+public:
+  explicit PrefixHashRepFactory(const SliceTransform* prefix_extractor,
+    size_t bucket_count = 0, size_t num_locks = 1000)
+    : TransformRepFactory(prefix_extractor, bucket_count, num_locks)
+    { }
+
+  virtual std::shared_ptr<MemTableRep> CreateMemTableRep(
+    MemTableRep::KeyComparator&, Arena*) override;
+
+  virtual const char* Name() const override {
+    return "PrefixHashRepFactory";
+  }
+};
+
+}
+
+#endif // STORAGE_ROCKSDB_DB_MEMTABLEREP_H_
@@ -2,14 +2,14 @@
 // 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.

-#ifndef STORAGE_LEVELDB_INCLUDE_MERGE_OPERATOR_H_
-#define STORAGE_LEVELDB_INCLUDE_MERGE_OPERATOR_H_
+#ifndef STORAGE_ROCKSDB_INCLUDE_MERGE_OPERATOR_H_
+#define STORAGE_ROCKSDB_INCLUDE_MERGE_OPERATOR_H_

 #include <string>
 #include <deque>
-#include "leveldb/slice.h"  // TODO: Remove this when migration is done;
+#include "rocksdb/slice.h"

-namespace leveldb {
+namespace rocksdb {

 class Slice;
 class Logger;
@@ -142,6 +142,6 @@ class AssociativeMergeOperator : public MergeOperator {
                            Logger* logger) const override;
 };

-}  // namespace leveldb
+}  // namespace rocksdb

-#endif  // STORAGE_LEVELDB_INCLUDE_MERGE_OPERATOR_H_
+#endif  // STORAGE_ROCKSDB_INCLUDE_MERGE_OPERATOR_H_
@@ -2,22 +2,21 @@
 // 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.

-#ifndef STORAGE_LEVELDB_INCLUDE_OPTIONS_H_
-#define STORAGE_LEVELDB_INCLUDE_OPTIONS_H_
+#ifndef STORAGE_ROCKSDB_INCLUDE_OPTIONS_H_
+#define STORAGE_ROCKSDB_INCLUDE_OPTIONS_H_

 #include <stddef.h>
 #include <string>
 #include <memory>
 #include <vector>
 #include <stdint.h>
-#include "leveldb/slice.h"
-#include "leveldb/statistics.h"
-#include "leveldb/universal_compaction.h"
-#include "leveldb/memtablerep.h"
+#include "rocksdb/slice.h"
+#include "rocksdb/statistics.h"
+#include "rocksdb/universal_compaction.h"
+#include "rocksdb/memtablerep.h"
+#include "rocksdb/slice_transform.h"

-#include "leveldb/slice_transform.h"
-
-namespace leveldb {
+namespace rocksdb {

 class Cache;
 class Comparator;
@@ -166,7 +165,7 @@ struct Options {
  // a block is the unit of reading from disk).

  // If non-NULL use the specified cache for blocks.
-  // If NULL, leveldb will automatically create and use an 8MB internal cache.
+  // If NULL, rocksdb will automatically create and use an 8MB internal cache.
  // Default: nullptr
  shared_ptr<Cache> block_cache;

@@ -352,13 +351,21 @@ struct Options {
  // Default value is 1800 (half an hour).
  int db_stats_log_interval;

-  // This specifies the log dir.
+  // This specifies the info LOG dir.
  // If it is empty, the log files will be in the same dir as data.
  // If it is non empty, the log files will be in the specified dir,
  // and the db data dir's absolute path will be used as the log file
  // name's prefix.
  std::string db_log_dir;

+  // This specifies the absolute dir path for write-ahead logs (WAL).
+  // If it is empty, the log files will be in the same dir as data,
+  //   dbname is used as the data dir by default
+  // If it is non empty, the log files will be in kept the specified dir.
+  // When destroying the db,
+  //   all log files in wal_dir and the dir itself is deleted
+  std::string wal_dir;
+
  // Disable compaction triggered by seek.
  // With bloomfilter and fast storage, a miss on one level
  // is very cheap if the file handle is cached in table cache
@@ -370,10 +377,23 @@ struct Options {
  // every compaction run.
  uint64_t delete_obsolete_files_period_micros;

-  // Maximum number of concurrent background compactions.
+  // Maximum number of concurrent background jobs, submitted to
+  // the default LOW priority thread pool
  // Default: 1
  int max_background_compactions;

+  // Maximum number of concurrent background memtable flush jobs, submitted to
+  // the HIGH priority thread pool.
+  // By default, all background jobs (major compaction and memtable flush) go
+  // to the LOW priority pool. If this option is set to a positive number,
+  // memtable flush jobs will be submitted to the HIGH priority pool.
+  // It is important when the same Env is shared by multiple db instances.
+  // Without a separate pool, long running major compaction jobs could
+  // potentially block memtable flush jobs of other db instances, leading to
+  // unnecessary Put stalls.
+  // Default: 0
+  int max_background_flushes;
+
  // Specify the maximal size of the info log file. If the log file
  // is larger than `max_log_file_size`, a new info log file will
  // be created.
@@ -422,6 +442,15 @@ struct Options {
  // Number of shards used for table cache.
  int table_cache_numshardbits;

+  // During data eviction of table's LRU cache, it would be inefficient
+  // to strictly follow LRU because this piece of memory will not really
+  // be released unless its refcount falls to zero. Instead, make two
+  // passes: the first pass will release items with refcount = 1,
+  // and if not enough space releases after scanning the number of
+  // elements specified by this parameter, we will remove items in LRU
+  // order.
+  int table_cache_remove_scan_count_limit;
+
  // size of one block in arena memory allocation.
  // If <= 0, a proper value is automatically calculated (usually 1/10 of
  // writer_buffer_size).
@@ -449,7 +478,7 @@ struct Options {

  // The number of seconds a WAL(write ahead log) should be kept after it has
  // been marked as Not Live. If the value is set. The WAL files are moved to
-  // the archive direcotory and deleted after the given TTL.
+  // the archive directory and deleted after the given TTL.
  // If set to 0, WAL files are deleted as soon as they are not required by
  // the database.
  // If set to std::numeric_limits<uint64_t>::max() the WAL files will never be
@@ -485,7 +514,7 @@ struct Options {
  // Default: false
  bool skip_log_error_on_recovery;

-  // if not zero, dump leveldb.stats to LOG every stats_dump_period_sec
+  // if not zero, dump rocksdb.stats to LOG every stats_dump_period_sec
  // Default: 3600 (1 hour)
  unsigned int stats_dump_period_sec;

@@ -533,6 +562,13 @@ struct Options {
  // Default: false
  bool filter_deletes;

+  // An iteration->Next() sequentially skips over keys with the same
+  // user-key unless this option is set. This number specifies the number
+  // of keys (with the same userkey) that will be sequentially
+  // skipped before a reseek is issued.
+  // Default: 8
+  uint64_t max_sequential_skip_in_iterations;
+
  // This is a factory that provides MemTableRep objects.
  // Default: a factory that provides a skip-list-based implementation of
  // MemTableRep.
@@ -542,6 +578,23 @@ struct Options {
  // an application to modify/delete a key-value during background compaction.
  // Default: a factory that doesn't provide any object
  std::shared_ptr<CompactionFilterFactory> compaction_filter_factory;
+
+  // Remove the log file immediately after the corresponding memtable is flushed
+  // to data file.
+  // Default: true
+  bool purge_log_after_memtable_flush;
+};
+
+//
+// An application can issue a read request (via Get/Iterators) and specify
+// if that read should process data that ALREADY resides on a specified cache
+// level. For example, if an application specifies kBlockCacheTier then the
+// Get call will process data that is already processed in the memtable or
+// the block cache. It will not page in data from the OS cache or data that
+// resides in storage.
+enum ReadTier {
+  kReadAllTier    = 0x0, // data in memtable, block cache, OS cache or storage
+  kBlockCacheTier = 0x1  // data in memtable or block cache
 };

 // Options that control read operations
@@ -576,15 +629,23 @@ struct ReadOptions {
  // Default: nullptr
  const Slice* prefix;

+  // Specify if this read request should process data that ALREADY
+  // resides on a particular cache. If the required data is not
+  // found at the specified cache, then Status::Incomplete is returned.
+  // Default: kReadAllTier
+  ReadTier read_tier;
+
  ReadOptions()
      : verify_checksums(false),
        fill_cache(true),
        snapshot(nullptr),
-        prefix(nullptr) {
+        prefix(nullptr),
+        read_tier(kReadAllTier) {
  }
  ReadOptions(bool cksum, bool cache) :
              verify_checksums(cksum), fill_cache(cache),
-              snapshot(nullptr), prefix(nullptr) {
+              snapshot(nullptr), prefix(nullptr),
+              read_tier(kReadAllTier) {
  }
 };

@@ -629,6 +690,6 @@ struct FlushOptions {
  }
 };

-}  // namespace leveldb
+}  // namespace rocksdb

-#endif  // STORAGE_LEVELDB_INCLUDE_OPTIONS_H_
+#endif  // STORAGE_ROCKSDB_INCLUDE_OPTIONS_H_
--- a/Show More
+++ b/Show More