Squashed 'src/leveldb/' changes from a31c8aa..196962f

196962f Add AcceleratedCRC32C to port_win.h
1bdf1c3 Merge upstream LevelDB v1.20
d31721e Merge #17: Fixed file sharing errors
fecd449 Fixed file sharing error in Win32Env::GetFileSize(), Win32SequentialFile::_Init(), Win32RandomAccessFile::_Init() Fixed error checking in Win32SequentialFile::_Init()
5b7510f Merge #14: Merge upstream LevelDB 1.19
0d969fd Merge #16: [LevelDB] Do no crash if filesystem can't fsync
c8c029b [LevelDB] Do no crash if filesystem can't fsync
a53934a Increase leveldb version to 1.20.
f3f1397 Separate Env tests from PosixEnv tests.
eb4f097 leveldb: Fix compilation warnings in port_posix_sse.cc on x86 (32-bit).
d0883b6 Fixed path to doc file: index.md.
7fa2094 Convert documentation to markdown.
ea175e2 Implement support for Intel crc32 instruction (SSE 4.2)
95cd743 Including <limits> for std::numeric_limits.
646c358 Limit the number of read-only files the POSIX Env will have open.
d40bc3f Merge #13: Typo
ebbd772 Typo
a2fb086 Add option for max file size. The currend hard-coded value of 2M is inefficient in colossus.

git-subtree-dir: src/leveldb
git-subtree-split: 196962f

Author: sipa

Makefile

@@ -44,6 +44,7 @@ TESTS = \
 	util/cache_test \
 	util/coding_test \
 	util/crc32c_test \
+	util/env_posix_test \
 	util/env_test \
 	util/hash_test
 
@@ -121,7 +122,7 @@ SHARED_MEMENVLIB = $(SHARED_OUTDIR)/libmemenv.a
 else
 # Update db.h if you change these.
 SHARED_VERSION_MAJOR = 1
-SHARED_VERSION_MINOR = 19
+SHARED_VERSION_MINOR = 20
 SHARED_LIB1 = libleveldb.$(PLATFORM_SHARED_EXT)
 SHARED_LIB2 = $(SHARED_LIB1).$(SHARED_VERSION_MAJOR)
 SHARED_LIB3 = $(SHARED_LIB1).$(SHARED_VERSION_MAJOR).$(SHARED_VERSION_MINOR)
@@ -337,6 +338,9 @@ $(STATIC_OUTDIR)/db_test:db/db_test.cc $(STATIC_LIBOBJECTS) $(TESTHARNESS)
 $(STATIC_OUTDIR)/dbformat_test:db/dbformat_test.cc $(STATIC_LIBOBJECTS) $(TESTHARNESS)
 	$(CXX) $(LDFLAGS) $(CXXFLAGS) db/dbformat_test.cc $(STATIC_LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
 
+$(STATIC_OUTDIR)/env_posix_test:util/env_posix_test.cc $(STATIC_LIBOBJECTS) $(TESTHARNESS)
+	$(CXX) $(LDFLAGS) $(CXXFLAGS) util/env_posix_test.cc $(STATIC_LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
+
 $(STATIC_OUTDIR)/env_test:util/env_test.cc $(STATIC_LIBOBJECTS) $(TESTHARNESS)
 	$(CXX) $(LDFLAGS) $(CXXFLAGS) util/env_test.cc $(STATIC_LIBOBJECTS) $(TESTHARNESS) -o $@ $(LIBS)
 
@@ -412,3 +416,9 @@ $(SHARED_OUTDIR)/%.o: %.cc
 
 $(SHARED_OUTDIR)/%.o: %.c
 	$(CC) $(CFLAGS) $(PLATFORM_SHARED_CFLAGS) -c $< -o $@
+
+$(STATIC_OUTDIR)/port/port_posix_sse.o: port/port_posix_sse.cc
+	$(CXX) $(CXXFLAGS) $(PLATFORM_SSEFLAGS) -c $< -o $@
+
+$(SHARED_OUTDIR)/port/port_posix_sse.o: port/port_posix_sse.cc
+	$(CXX) $(CXXFLAGS) $(PLATFORM_SHARED_CFLAGS) $(PLATFORM_SSEFLAGS) -c $< -o $@

README.md

@@ -16,7 +16,7 @@ Authors: Sanjay Ghemawat ([email protected]) and Jeff Dean ([email protected])
   * External activity (file system operations etc.) is relayed through a virtual interface so users can customize the operating system interactions.
 
 # Documentation
-  [LevelDB library documentation](https://rawgit.com/google/leveldb/master/doc/index.html) is online and bundled with the source code.
+  [LevelDB library documentation](https://github.com/google/leveldb/blob/master/doc/index.md) is online and bundled with the source code.
 
 
 # Limitations
@@ -113,29 +113,30 @@ by the one or two disk seeks needed to fetch the data from disk.
 Write performance will be mostly unaffected by whether or not the
 working set fits in memory.
 
-    readrandom   :      16.677 micros/op;  (approximately 60,000 reads per second)
-    readseq      :       0.476 micros/op;  232.3 MB/s
-    readreverse  :       0.724 micros/op;  152.9 MB/s
+    readrandom  : 16.677 micros/op;  (approximately 60,000 reads per second)
+    readseq     :  0.476 micros/op;  232.3 MB/s
+    readreverse :  0.724 micros/op;  152.9 MB/s
 
 LevelDB compacts its underlying storage data in the background to
 improve read performance.  The results listed above were done
 immediately after a lot of random writes.  The results after
 compactions (which are usually triggered automatically) are better.
 
-    readrandom   :      11.602 micros/op;  (approximately 85,000 reads per second)
-    readseq      :       0.423 micros/op;  261.8 MB/s
-    readreverse  :       0.663 micros/op;  166.9 MB/s
+    readrandom  : 11.602 micros/op;  (approximately 85,000 reads per second)
+    readseq     :  0.423 micros/op;  261.8 MB/s
+    readreverse :  0.663 micros/op;  166.9 MB/s
 
 Some of the high cost of reads comes from repeated decompression of blocks
 read from disk.  If we supply enough cache to the leveldb so it can hold the
 uncompressed blocks in memory, the read performance improves again:
 
-    readrandom   :       9.775 micros/op;  (approximately 100,000 reads per second before compaction)
-    readrandom   :       5.215 micros/op;  (approximately 190,000 reads per second after compaction)
+    readrandom  : 9.775 micros/op;  (approximately 100,000 reads per second before compaction)
+    readrandom  : 5.215 micros/op;  (approximately 190,000 reads per second after compaction)
 
 ## Repository contents
 
-See doc/index.html for more explanation. See doc/impl.html for a brief overview of the implementation.
+See [doc/index.md](doc/index.md) for more explanation. See
+[doc/impl.md](doc/impl.md) for a brief overview of the implementation.
 
 The public interface is in include/*.h.  Callers should not include or
 rely on the details of any other header files in this package.  Those
@@ -148,7 +149,7 @@ Guide to header files:
 * **include/options.h**: Control over the behavior of an entire database,
 and also control over the behavior of individual reads and writes.
 
-* **include/comparator.h**: Abstraction for user-specified comparison function. 
+* **include/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.)
@@ -165,7 +166,7 @@ length into some other byte array.
 * **include/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/env.h**:
 Abstraction of the OS environment.  A posix implementation of this interface is
 in util/env_posix.cc
 

build_detect_platform

@@ -63,6 +63,7 @@ PLATFORM_SHARED_EXT="so"
 PLATFORM_SHARED_LDFLAGS="-shared -Wl,-soname -Wl,"
 PLATFORM_SHARED_CFLAGS="-fPIC"
 PLATFORM_SHARED_VERSIONED=true
+PLATFORM_SSEFLAGS=
 
 MEMCMP_FLAG=
 if [ "$CXX" = "g++" ]; then
@@ -77,6 +78,7 @@ case "$TARGET_OS" in
         COMMON_FLAGS="$MEMCMP_FLAG -lpthread -DOS_LINUX -DCYGWIN"
         PLATFORM_LDFLAGS="-lpthread"
         PORT_FILE=port/port_posix.cc
+        PORT_SSE_FILE=port/port_posix_sse.cc
         ;;
     Darwin)
         PLATFORM=OS_MACOSX
@@ -85,24 +87,28 @@ case "$TARGET_OS" in
         [ -z "$INSTALL_PATH" ] && INSTALL_PATH=`pwd`
         PLATFORM_SHARED_LDFLAGS="-dynamiclib -install_name $INSTALL_PATH/"
         PORT_FILE=port/port_posix.cc
+        PORT_SSE_FILE=port/port_posix_sse.cc
         ;;
     Linux)
         PLATFORM=OS_LINUX
         COMMON_FLAGS="$MEMCMP_FLAG -pthread -DOS_LINUX"
         PLATFORM_LDFLAGS="-pthread"
         PORT_FILE=port/port_posix.cc
+        PORT_SSE_FILE=port/port_posix_sse.cc
         ;;
     SunOS)
         PLATFORM=OS_SOLARIS
         COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_SOLARIS"
         PLATFORM_LIBS="-lpthread -lrt"
         PORT_FILE=port/port_posix.cc
+        PORT_SSE_FILE=port/port_posix_sse.cc
         ;;
     FreeBSD)
         PLATFORM=OS_FREEBSD
         COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_FREEBSD"
         PLATFORM_LIBS="-lpthread"
         PORT_FILE=port/port_posix.cc
+        PORT_SSE_FILE=port/port_posix_sse.cc
         ;;
     GNU/kFreeBSD)
         PLATFORM=OS_KFREEBSD
@@ -115,31 +121,36 @@ case "$TARGET_OS" in
         COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_NETBSD"
         PLATFORM_LIBS="-lpthread -lgcc_s"
         PORT_FILE=port/port_posix.cc
+        PORT_SSE_FILE=port/port_posix_sse.cc
         ;;
     OpenBSD)
         PLATFORM=OS_OPENBSD
         COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_OPENBSD"
         PLATFORM_LDFLAGS="-pthread"
         PORT_FILE=port/port_posix.cc
+        PORT_SSE_FILE=port/port_posix_sse.cc
         ;;
     DragonFly)
         PLATFORM=OS_DRAGONFLYBSD
         COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_DRAGONFLYBSD"
         PLATFORM_LIBS="-lpthread"
         PORT_FILE=port/port_posix.cc
+        PORT_SSE_FILE=port/port_posix_sse.cc
         ;;
     OS_ANDROID_CROSSCOMPILE)
         PLATFORM=OS_ANDROID
         COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_ANDROID -DLEVELDB_PLATFORM_POSIX"
         PLATFORM_LDFLAGS=""  # All pthread features are in the Android C library
         PORT_FILE=port/port_posix.cc
+        PORT_SSE_FILE=port/port_posix_sse.cc
         CROSS_COMPILE=true
         ;;
     HP-UX)
         PLATFORM=OS_HPUX
         COMMON_FLAGS="$MEMCMP_FLAG -D_REENTRANT -DOS_HPUX"
         PLATFORM_LDFLAGS="-pthread"
         PORT_FILE=port/port_posix.cc
+        PORT_SSE_FILE=port/port_posix_sse.cc
         # man ld: +h internal_name
         PLATFORM_SHARED_LDFLAGS="-shared -Wl,+h -Wl,"
         ;;
@@ -148,6 +159,7 @@ case "$TARGET_OS" in
         COMMON_FLAGS="$MEMCMP_FLAG -DOS_MACOSX"
         [ -z "$INSTALL_PATH" ] && INSTALL_PATH=`pwd`
         PORT_FILE=port/port_posix.cc
+        PORT_SSE_FILE=port/port_posix_sse.cc
         PLATFORM_SHARED_EXT=
         PLATFORM_SHARED_LDFLAGS=
         PLATFORM_SHARED_CFLAGS=
@@ -182,7 +194,7 @@ set +f # re-enable globbing
 
 # The sources consist of the portable files, plus the platform-specific port
 # file.
-echo "SOURCES=$PORTABLE_FILES $PORT_FILE" >> $OUTPUT
+echo "SOURCES=$PORTABLE_FILES $PORT_FILE $PORT_SSE_FILE" >> $OUTPUT
 echo "MEMENV_SOURCES=helpers/memenv/memenv.cc" >> $OUTPUT
 
 if [ "$CROSS_COMPILE" = "true" ]; then
@@ -213,6 +225,21 @@ EOF
     fi
 
     rm -f $CXXOUTPUT 2>/dev/null
+
+    # Test if gcc SSE 4.2 is supported
+    $CXX $CXXFLAGS -x c++ - -o $CXXOUTPUT -msse4.2 2>/dev/null  <<EOF
+      int main() {}
+EOF
+    if [ "$?" = 0 ]; then
+        PLATFORM_SSEFLAGS="-msse4.2"
+    fi
+
+    rm -f $CXXOUTPUT 2>/dev/null
+fi
+
+# Use the SSE 4.2 CRC32C intrinsics iff runtime checks indicate compiler supports them.
+if [ -n "$PLATFORM_SSEFLAGS" ]; then
+    PLATFORM_SSEFLAGS="$PLATFORM_SSEFLAGS -DLEVELDB_PLATFORM_POSIX_SSE"
 fi
 
 PLATFORM_CCFLAGS="$PLATFORM_CCFLAGS $COMMON_FLAGS"
@@ -225,6 +252,7 @@ echo "PLATFORM_LDFLAGS=$PLATFORM_LDFLAGS" >> $OUTPUT
 echo "PLATFORM_LIBS=$PLATFORM_LIBS" >> $OUTPUT
 echo "PLATFORM_CCFLAGS=$PLATFORM_CCFLAGS" >> $OUTPUT
 echo "PLATFORM_CXXFLAGS=$PLATFORM_CXXFLAGS" >> $OUTPUT
+echo "PLATFORM_SSEFLAGS=$PLATFORM_SSEFLAGS" >> $OUTPUT
 echo "PLATFORM_SHARED_CFLAGS=$PLATFORM_SHARED_CFLAGS" >> $OUTPUT
 echo "PLATFORM_SHARED_EXT=$PLATFORM_SHARED_EXT" >> $OUTPUT
 echo "PLATFORM_SHARED_LDFLAGS=$PLATFORM_SHARED_LDFLAGS" >> $OUTPUT

db/db_bench.cc

@@ -84,6 +84,14 @@ static bool FLAGS_histogram = false;
 // (initialized to default value by "main")
 static int FLAGS_write_buffer_size = 0;
 
+// Number of bytes written to each file.
+// (initialized to default value by "main")
+static int FLAGS_max_file_size = 0;
+
+// Approximate size of user data packed per block (before compression.
+// (initialized to default value by "main")
+static int FLAGS_block_size = 0;
+
 // Number of bytes to use as a cache of uncompressed data.
 // Negative means use default settings.
 static int FLAGS_cache_size = -1;
@@ -109,6 +117,7 @@ static const char* FLAGS_db = NULL;
 namespace leveldb {
 
 namespace {
+leveldb::Env* g_env = NULL;
 
 // Helper for quickly generating random data.
 class RandomGenerator {
@@ -186,7 +195,7 @@ class Stats {
     done_ = 0;
     bytes_ = 0;
     seconds_ = 0;
-    start_ = Env::Default()->NowMicros();
+    start_ = g_env->NowMicros();
     finish_ = start_;
     message_.clear();
   }
@@ -204,7 +213,7 @@ class Stats {
   }
 
   void Stop() {
-    finish_ = Env::Default()->NowMicros();
+    finish_ = g_env->NowMicros();
     seconds_ = (finish_ - start_) * 1e-6;
   }
 
@@ -214,7 +223,7 @@ class Stats {
 
   void FinishedSingleOp() {
     if (FLAGS_histogram) {
-      double now = Env::Default()->NowMicros();
+      double now = g_env->NowMicros();
       double micros = now - last_op_finish_;
       hist_.Add(micros);
       if (micros > 20000) {
@@ -404,10 +413,10 @@ class Benchmark {
     reads_(FLAGS_reads < 0 ? FLAGS_num : FLAGS_reads),
     heap_counter_(0) {
     std::vector<std::string> files;
-    Env::Default()->GetChildren(FLAGS_db, &files);
+    g_env->GetChildren(FLAGS_db, &files);
     for (size_t i = 0; i < files.size(); i++) {
       if (Slice(files[i]).starts_with("heap-")) {
-        Env::Default()->DeleteFile(std::string(FLAGS_db) + "/" + files[i]);
+        g_env->DeleteFile(std::string(FLAGS_db) + "/" + files[i]);
       }
     }
     if (!FLAGS_use_existing_db) {
@@ -589,7 +598,7 @@ class Benchmark {
       arg[i].shared = &shared;
       arg[i].thread = new ThreadState(i);
       arg[i].thread->shared = &shared;
-      Env::Default()->StartThread(ThreadBody, &arg[i]);
+      g_env->StartThread(ThreadBody, &arg[i]);
     }
 
     shared.mu.Lock();
@@ -700,9 +709,12 @@ class Benchmark {
   void Open() {
     assert(db_ == NULL);
     Options options;
+    options.env = g_env;
     options.create_if_missing = !FLAGS_use_existing_db;
     options.block_cache = cache_;
     options.write_buffer_size = FLAGS_write_buffer_size;
+    options.max_file_size = FLAGS_max_file_size;
+    options.block_size = FLAGS_block_size;
     options.max_open_files = FLAGS_open_files;
     options.filter_policy = filter_policy_;
     options.reuse_logs = FLAGS_reuse_logs;
@@ -925,7 +937,7 @@ class Benchmark {
     char fname[100];
     snprintf(fname, sizeof(fname), "%s/heap-%04d", FLAGS_db, ++heap_counter_);
     WritableFile* file;
-    Status s = Env::Default()->NewWritableFile(fname, &file);
+    Status s = g_env->NewWritableFile(fname, &file);
     if (!s.ok()) {
       fprintf(stderr, "%s\n", s.ToString().c_str());
       return;
@@ -934,7 +946,7 @@ class Benchmark {
     delete file;
     if (!ok) {
       fprintf(stderr, "heap profiling not supported\n");
-      Env::Default()->DeleteFile(fname);
+      g_env->DeleteFile(fname);
     }
   }
 };
@@ -943,6 +955,8 @@ class Benchmark {
 
 int main(int argc, char** argv) {
   FLAGS_write_buffer_size = leveldb::Options().write_buffer_size;
+  FLAGS_max_file_size = leveldb::Options().max_file_size;
+  FLAGS_block_size = leveldb::Options().block_size;
   FLAGS_open_files = leveldb::Options().max_open_files;
   std::string default_db_path;
 
@@ -973,6 +987,10 @@ int main(int argc, char** argv) {
       FLAGS_value_size = n;
     } else if (sscanf(argv[i], "--write_buffer_size=%d%c", &n, &junk) == 1) {
       FLAGS_write_buffer_size = n;
+    } else if (sscanf(argv[i], "--max_file_size=%d%c", &n, &junk) == 1) {
+      FLAGS_max_file_size = n;
+    } else if (sscanf(argv[i], "--block_size=%d%c", &n, &junk) == 1) {
+      FLAGS_block_size = n;
     } else if (sscanf(argv[i], "--cache_size=%d%c", &n, &junk) == 1) {
       FLAGS_cache_size = n;
     } else if (sscanf(argv[i], "--bloom_bits=%d%c", &n, &junk) == 1) {
@@ -987,9 +1005,11 @@ int main(int argc, char** argv) {
     }
   }
 
+  leveldb::g_env = leveldb::Env::Default();
+
   // Choose a location for the test database if none given with --db=<path>
   if (FLAGS_db == NULL) {
-      leveldb::Env::Default()->GetTestDirectory(&default_db_path);
+      leveldb::g_env->GetTestDirectory(&default_db_path);
       default_db_path += "/dbbench";
       FLAGS_db = default_db_path.c_str();
   }

db/db_impl.cc

@@ -96,6 +96,7 @@ Options SanitizeOptions(const std::string& dbname,
   result.filter_policy = (src.filter_policy != NULL) ? ipolicy : NULL;
   ClipToRange(&result.max_open_files,    64 + kNumNonTableCacheFiles, 50000);
   ClipToRange(&result.write_buffer_size, 64<<10,                      1<<30);
+  ClipToRange(&result.max_file_size,     1<<20,                       1<<30);
   ClipToRange(&result.block_size,        1<<10,                       4<<20);
   if (result.info_log == NULL) {
     // Open a log file in the same directory as the db

db/log_format.h

@@ -3,7 +3,7 @@
 // found in the LICENSE file. See the AUTHORS file for names of contributors.
 //
 // Log format information shared by reader and writer.
-// See ../doc/log_format.txt for more detail.
+// See ../doc/log_format.md for more detail.
 
 #ifndef STORAGE_LEVELDB_DB_LOG_FORMAT_H_
 #define STORAGE_LEVELDB_DB_LOG_FORMAT_H_