@20602303. Default file permission is now 755.

git-svn-id: https://leveldb.googlecode.com/svn/trunk@20 62dab493-f737-651d-591e-8d6aee1b9529

TODO

@@ -1,11 +1,3 @@
-Before adding to chrome
------------------------
-- multi-threaded test/benchmark
-- Allow missing crc32c in Table format?
-
-Maybe afterwards
-----------------
-
 ss
 - Stats
 

db/corruption_test.cc

@@ -8,6 +8,7 @@
 #include <fcntl.h>
 #include <sys/stat.h>
 #include <sys/types.h>
+#include "leveldb/cache.h"
 #include "leveldb/env.h"
 #include "leveldb/table.h"
 #include "leveldb/write_batch.h"
@@ -28,10 +29,12 @@ class CorruptionTest {
   test::ErrorEnv env_;
   Random rnd_;
   std::string dbname_;
+  Cache* tiny_cache_;
   Options options_;
   DB* db_;
 
   CorruptionTest() : rnd_(test::RandomSeed()) {
+    tiny_cache_ = NewLRUCache(100);
     options_.env = &env_;
     dbname_ = test::TmpDir() + "/db_test";
     DestroyDB(dbname_, options_);
@@ -45,6 +48,7 @@ class CorruptionTest {
   ~CorruptionTest() {
      delete db_;
      DestroyDB(dbname_, Options());
+     delete tiny_cache_;
   }
 
   Status TryReopen(Options* options = NULL) {
@@ -52,6 +56,7 @@ class CorruptionTest {
     db_ = NULL;
     Options opt = (options ? *options : options_);
     opt.env = &env_;
+    opt.block_cache = tiny_cache_;
     return DB::Open(opt, dbname_, &db_);
   }
 
@@ -160,12 +165,15 @@ class CorruptionTest {
     ASSERT_TRUE(s.ok()) << s.ToString();
   }
 
-  uint64_t Property(const std::string& name) {
-    uint64_t result;
-    if (!db_->GetProperty(name, &result)) {
-      result = ~static_cast<uint64_t>(0);
+  int Property(const std::string& name) {
+    std::string property;
+    int result;
+    if (db_->GetProperty(name, &property) &&
+        sscanf(property.c_str(), "%d", &result) == 1) {
+      return result;
+    } else {
+      return -1;
     }
-    return result;
   }
 
   // Return the ith key
@@ -235,7 +243,7 @@ TEST(CorruptionTest, TableFileIndexData) {
   dbi->TEST_CompactRange(0, "", "~");
   dbi->TEST_CompactRange(1, "", "~");
 
-  Corrupt(kTableFile, -1000, 500);
+  Corrupt(kTableFile, -2000, 500);
   Reopen();
   Check(5000, 9999);
 }
@@ -327,6 +335,7 @@ TEST(CorruptionTest, CompactionInputError) {
 TEST(CorruptionTest, CompactionInputErrorParanoid) {
   Options options;
   options.paranoid_checks = true;
+  options.write_buffer_size = 1048576;
   Reopen(&options);
 
   Build(10);

db/db_bench.cc

@@ -31,11 +31,8 @@
 //      sha1          -- repeated SHA1 computation over 4K of data
 //   Meta operations:
 //      compact     -- Compact the entire DB
+//      stats       -- Print DB stats
 //      heapprofile -- Dump a heap profile (if supported by this port)
-//      sync        -- switch to synchronous writes (not the default)
-//      nosync      -- switch to asynchronous writes (the default)
-//      tenth       -- divide N by 10 (i.e., following benchmarks are smaller)
-//      normal      -- reset N back to its normal value (1000000)
 static const char* FLAGS_benchmarks =
     "fillseq,"
     "fillsync,"
@@ -51,7 +48,9 @@ static const char* FLAGS_benchmarks =
     "readreverse,"
     "fill100K,"
     "crc32c,"
-    "sha1"
+    "sha1,"
+    "snappycomp,"
+    "snappyuncomp,"
     ;
 
 // Number of key/values to place in database
@@ -68,7 +67,12 @@ static double FLAGS_compression_ratio = 0.5;
 static bool FLAGS_histogram = false;
 
 // Number of bytes to buffer in memtable before compacting
-static int FLAGS_write_buffer_size = 1 << 20;
+// (initialized to default value by "main")
+static int FLAGS_write_buffer_size = 0;
+
+// Number of bytes to use as a cache of uncompressed data.
+// Negative means use default settings.
+static int FLAGS_cache_size = -1;
 
 namespace leveldb {
 
@@ -129,6 +133,7 @@ class Benchmark {
   double last_op_finish_;
   int64_t bytes_;
   std::string message_;
+  std::string post_message_;
   Histogram hist_;
   RandomGenerator gen_;
   Random rand_;
@@ -146,7 +151,8 @@ class Benchmark {
             static_cast<int>(FLAGS_value_size * FLAGS_compression_ratio + 0.5));
     fprintf(stdout, "Entries:    %d\n", num_);
     fprintf(stdout, "RawSize:    %.1f MB (estimated)\n",
-            (((kKeySize + FLAGS_value_size) * num_) / 1048576.0));
+            ((static_cast<int64_t>(kKeySize + FLAGS_value_size) * num_)
+             / 1048576.0));
     fprintf(stdout, "FileSize:   %.1f MB (estimated)\n",
             (((kKeySize + FLAGS_value_size * FLAGS_compression_ratio) * num_)
              / 1048576.0));
@@ -164,6 +170,15 @@ class Benchmark {
     fprintf(stdout,
             "WARNING: Assertions are enabled; benchmarks unnecessarily slow\n");
 #endif
+
+    // See if snappy is working by attempting to compress a compressible string
+    const char text[] = "yyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyyy";
+    std::string compressed;
+    if (!port::Snappy_Compress(text, sizeof(text), &compressed)) {
+      fprintf(stdout, "WARNING: Snappy compression is not enabled\n");
+    } else if (compressed.size() >= sizeof(text)) {
+      fprintf(stdout, "WARNING: Snappy compression is not effective\n");
+    }
   }
 
   void PrintEnvironment() {
@@ -225,15 +240,13 @@ class Benchmark {
 
     done_++;
     if (done_ >= next_report_) {
-      if (next_report_ < 1000) {
-        next_report_ += 100;
-      } else if (next_report_ < 10000) {
-        next_report_ += 1000;
-      } else if (next_report_ < 100000) {
-        next_report_ += 10000;
-      } else {
-        next_report_ += 100000;
-      }
+      if      (next_report_ < 1000)   next_report_ += 100;
+      else if (next_report_ < 5000)   next_report_ += 500;
+      else if (next_report_ < 10000)  next_report_ += 1000;
+      else if (next_report_ < 50000)  next_report_ += 5000;
+      else if (next_report_ < 100000) next_report_ += 10000;
+      else if (next_report_ < 500000) next_report_ += 50000;
+      else                            next_report_ += 100000;
       fprintf(stderr, "... finished %d ops%30s\r", done_, "");
       fflush(stderr);
     }
@@ -248,7 +261,7 @@ class Benchmark {
 
     if (bytes_ > 0) {
       char rate[100];
-      snprintf(rate, sizeof(rate), "%5.1f MB/s",
+      snprintf(rate, sizeof(rate), "%6.1f MB/s",
                (bytes_ / 1048576.0) / (finish - start_));
       if (!message_.empty()) {
         message_  = std::string(rate) + " " + message_;
@@ -266,6 +279,11 @@ class Benchmark {
       fprintf(stdout, "Microseconds per op:\n%s\n", hist_.ToString().c_str());
     }
     fflush(stdout);
+
+    if (!post_message_.empty()) {
+      fprintf(stdout, "\n%s\n", post_message_.c_str());
+      post_message_.clear();
+    }
   }
 
  public:
@@ -278,12 +296,13 @@ class Benchmark {
     EXISTING
   };
 
-  Benchmark() : cache_(NewLRUCache(200<<20)),
-                db_(NULL),
-                num_(FLAGS_num),
-                heap_counter_(0),
-                bytes_(0),
-                rand_(301) {
+  Benchmark()
+  : cache_(FLAGS_cache_size >= 0 ? NewLRUCache(FLAGS_cache_size) : NULL),
+    db_(NULL),
+    num_(FLAGS_num),
+    heap_counter_(0),
+    bytes_(0),
+    rand_(301) {
     std::vector<std::string> files;
     Env::Default()->GetChildren("/tmp/dbbench", &files);
     for (int i = 0; i < files.size(); i++) {
@@ -318,36 +337,54 @@ class Benchmark {
       Start();
 
       WriteOptions write_options;
-      write_options.sync = false;
+      bool known = true;
       if (name == Slice("fillseq")) {
-        Write(write_options, SEQUENTIAL, FRESH, num_, FLAGS_value_size);
+        Write(write_options, SEQUENTIAL, FRESH, num_, FLAGS_value_size, 1);
+      } else if (name == Slice("fillbatch")) {
+        Write(write_options, SEQUENTIAL, FRESH, num_, FLAGS_value_size, 1000);
       } else if (name == Slice("fillrandom")) {
-        Write(write_options, RANDOM, FRESH, num_, FLAGS_value_size);
+        Write(write_options, RANDOM, FRESH, num_, FLAGS_value_size, 1);
       } else if (name == Slice("overwrite")) {
-        Write(write_options, RANDOM, EXISTING, num_, FLAGS_value_size);
+        Write(write_options, RANDOM, EXISTING, num_, FLAGS_value_size, 1);
       } else if (name == Slice("fillsync")) {
         write_options.sync = true;
-        Write(write_options, RANDOM, FRESH, num_ / 100, FLAGS_value_size);
+        Write(write_options, RANDOM, FRESH, num_ / 100, FLAGS_value_size, 1);
       } else if (name == Slice("fill100K")) {
-        Write(write_options, RANDOM, FRESH, num_ / 1000, 100 * 1000);
+        Write(write_options, RANDOM, FRESH, num_ / 1000, 100 * 1000, 1);
       } else if (name == Slice("readseq")) {
         ReadSequential();
       } else if (name == Slice("readreverse")) {
         ReadReverse();
       } else if (name == Slice("readrandom")) {
         ReadRandom();
+      } else if (name == Slice("readrandomsmall")) {
+        int n = num_;
+        num_ /= 1000;
+        ReadRandom();
+        num_ = n;
       } else if (name == Slice("compact")) {
         Compact();
       } else if (name == Slice("crc32c")) {
         Crc32c(4096, "(4K per op)");
       } else if (name == Slice("sha1")) {
         SHA1(4096, "(4K per op)");
+      } else if (name == Slice("snappycomp")) {
+        SnappyCompress();
+      } else if (name == Slice("snappyuncomp")) {
+        SnappyUncompress();
       } else if (name == Slice("heapprofile")) {
         HeapProfile();
+      } else if (name == Slice("stats")) {
+        PrintStats();
       } else {
-        fprintf(stderr, "unknown benchmark '%s'\n", name.ToString().c_str());
+        known = false;
+        if (name != Slice()) {  // No error message for empty name
+          fprintf(stderr, "unknown benchmark '%s'\n", name.ToString().c_str());
+        }
+      }
+      if (known) {
+        Stop(name);
       }
-      Stop(name);
     }
   }
 
@@ -387,11 +424,54 @@ class Benchmark {
     message_ = label;
   }
 
+  void SnappyCompress() {
+    Slice input = gen_.Generate(Options().block_size);
+    int64_t bytes = 0;
+    int64_t produced = 0;
+    bool ok = true;
+    std::string compressed;
+    while (ok && bytes < 1024 * 1048576) {  // Compress 1G
+      ok = port::Snappy_Compress(input.data(), input.size(), &compressed);
+      produced += compressed.size();
+      bytes += input.size();
+      FinishedSingleOp();
+    }
+
+    if (!ok) {
+      message_ = "(snappy failure)";
+    } else {
+      char buf[100];
+      snprintf(buf, sizeof(buf), "(output: %.1f%%)",
+               (produced * 100.0) / bytes);
+      message_ = buf;
+      bytes_ = bytes;
+    }
+  }
+
+  void SnappyUncompress() {
+    Slice input = gen_.Generate(Options().block_size);
+    std::string compressed;
+    bool ok = port::Snappy_Compress(input.data(), input.size(), &compressed);
+    int64_t bytes = 0;
+    std::string uncompressed;
+    while (ok && bytes < 1024 * 1048576) {  // Compress 1G
+      ok =  port::Snappy_Uncompress(compressed.data(), compressed.size(),
+                                    &uncompressed);
+      bytes += uncompressed.size();
+      FinishedSingleOp();
+    }
+
+    if (!ok) {
+      message_ = "(snappy failure)";
+    } else {
+      bytes_ = bytes;
+    }
+  }
+
   void Open() {
     assert(db_ == NULL);
     Options options;
     options.create_if_missing = true;
-    options.max_open_files = 10000;
     options.block_cache = cache_;
     options.write_buffer_size = FLAGS_write_buffer_size;
     Status s = DB::Open(options, "/tmp/dbbench", &db_);
@@ -402,7 +482,7 @@ class Benchmark {
   }
 
   void Write(const WriteOptions& options, Order order, DBState state,
-             int num_entries, int value_size) {
+             int num_entries, int value_size, int entries_per_batch) {
     if (state == FRESH) {
       delete db_;
       db_ = NULL;
@@ -420,19 +500,21 @@ class Benchmark {
     WriteBatch batch;
     Status s;
     std::string val;
-    for (int i = 0; i < num_entries; i++) {
-      const int k = (order == SEQUENTIAL) ? i : (rand_.Next() % FLAGS_num);
-      char key[100];
-      snprintf(key, sizeof(key), "%016d", k);
+    for (int i = 0; i < num_entries; i += entries_per_batch) {
       batch.Clear();
-      batch.Put(key, gen_.Generate(value_size));
+      for (int j = 0; j < entries_per_batch; j++) {
+        const int k = (order == SEQUENTIAL) ? i+j : (rand_.Next() % FLAGS_num);
+        char key[100];
+        snprintf(key, sizeof(key), "%016d", k);
+        batch.Put(key, gen_.Generate(value_size));
+        bytes_ += value_size + strlen(key);
+        FinishedSingleOp();
+      }
       s = db_->Write(options, &batch);
-      bytes_ += value_size + strlen(key);
       if (!s.ok()) {
         fprintf(stderr, "put error: %s\n", s.ToString().c_str());
         exit(1);
       }
-      FinishedSingleOp();
     }
   }
 
@@ -475,10 +557,10 @@ class Benchmark {
     dbi->TEST_CompactMemTable();
     int max_level_with_files = 1;
     for (int level = 1; level < config::kNumLevels; level++) {
-      uint64_t v;
+      std::string property;
       char name[100];
       snprintf(name, sizeof(name), "leveldb.num-files-at-level%d", level);
-      if (db_->GetProperty(name, &v) && v > 0) {
+      if (db_->GetProperty(name, &property) && atoi(property.c_str()) > 0) {
         max_level_with_files = level;
       }
     }
@@ -487,6 +569,15 @@ class Benchmark {
     }
   }
 
+  void PrintStats() {
+    std::string stats;
+    if (!db_->GetProperty("leveldb.stats", &stats)) {
+      message_ = "(failed)";
+    } else {
+      post_message_ = stats;
+    }
+  }
+
   static void WriteToFile(void* arg, const char* buf, int n) {
     reinterpret_cast<WritableFile*>(arg)->Append(Slice(buf, n));
   }
@@ -512,6 +603,7 @@ class Benchmark {
 }
 
 int main(int argc, char** argv) {
+  FLAGS_write_buffer_size = leveldb::Options().write_buffer_size;
   for (int i = 1; i < argc; i++) {
     double d;
     int n;
@@ -529,7 +621,9 @@ 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 {
+    } else if (sscanf(argv[i], "--cache_size=%d%c", &n, &junk) == 1) {
+      FLAGS_cache_size = n;
+    } else {
       fprintf(stderr, "Invalid flag '%s'\n", argv[i]);
       exit(1);
     }

db/db_impl.cc

@@ -104,6 +104,9 @@ Options SanitizeOptions(const std::string& dbname,
       result.info_log = new NullWritableFile;
     }
   }
+  if (result.block_cache == NULL) {
+    result.block_cache = NewLRUCache(8 << 20);
+  }
   return result;
 }
 
@@ -112,18 +115,20 @@ DBImpl::DBImpl(const Options& options, const std::string& dbname)
       internal_comparator_(options.comparator),
       options_(SanitizeOptions(dbname, &internal_comparator_, options)),
       owns_info_log_(options_.info_log != options.info_log),
+      owns_cache_(options_.block_cache != options.block_cache),
       dbname_(dbname),
       db_lock_(NULL),
       shutting_down_(NULL),
       bg_cv_(&mutex_),
       compacting_cv_(&mutex_),
-      last_sequence_(0),
       mem_(new MemTable(internal_comparator_)),
+      imm_(NULL),
       logfile_(NULL),
       log_(NULL),
-      log_number_(0),
       bg_compaction_scheduled_(false),
       compacting_(false) {
+  has_imm_.Release_Store(NULL);
+
   // Reserve ten files or so for other uses and give the rest to TableCache.
   const int table_cache_size = options.max_open_files - 10;
   table_cache_ = new TableCache(dbname_, &options_, table_cache_size);
@@ -149,6 +154,7 @@ DBImpl::~DBImpl() {
 
   delete versions_;
   delete mem_;
+  delete imm_;
   delete log_;
   delete logfile_;
   delete table_cache_;
@@ -156,15 +162,15 @@ DBImpl::~DBImpl() {
   if (owns_info_log_) {
     delete options_.info_log;
   }
+  if (owns_cache_) {
+    delete options_.block_cache;
+  }
 }
 
 Status DBImpl::NewDB() {
-  assert(log_number_ == 0);
-  assert(last_sequence_ == 0);
-
   VersionEdit new_db;
   new_db.SetComparatorName(user_comparator()->Name());
-  new_db.SetLogNumber(log_number_);
+  new_db.SetLogNumber(0);
   new_db.SetNextFile(2);
   new_db.SetLastSequence(0);
 
@@ -193,15 +199,6 @@ Status DBImpl::NewDB() {
   return s;
 }
 
-Status DBImpl::Install(VersionEdit* edit,
-                       uint64_t new_log_number,
-                       MemTable* cleanup_mem) {
-  mutex_.AssertHeld();
-  edit->SetLogNumber(new_log_number);
-  edit->SetLastSequence(last_sequence_);
-  return versions_->LogAndApply(edit, cleanup_mem);
-}
-
 void DBImpl::MaybeIgnoreError(Status* s) const {
   if (s->ok() || options_.paranoid_checks) {
     // No change needed
@@ -216,7 +213,7 @@ void DBImpl::DeleteObsoleteFiles() {
   std::set<uint64_t> live = pending_outputs_;
   versions_->AddLiveFiles(&live);
 
-  versions_->CleanupLargeValueRefs(live, log_number_);
+  versions_->CleanupLargeValueRefs(live);
 
   std::vector<std::string> filenames;
   env_->GetChildren(dbname_, &filenames); // Ignoring errors on purpose
@@ -228,7 +225,8 @@ void DBImpl::DeleteObsoleteFiles() {
       bool keep = true;
       switch (type) {
         case kLogFile:
-          keep = (number == log_number_);
+          keep = ((number == versions_->LogNumber()) ||
+                  (number == versions_->PrevLogNumber()));
           break;
         case kDescriptorFile:
           // Keep my manifest file, and any newer incarnations'
@@ -296,16 +294,20 @@ Status DBImpl::Recover(VersionEdit* edit) {
     }
   }
 
-  s = versions_->Recover(&log_number_, &last_sequence_);
+  s = versions_->Recover();
   if (s.ok()) {
-    // Recover from the log file named in the descriptor
+    // Recover from the log files named in the descriptor
     SequenceNumber max_sequence(0);
-    if (log_number_ != 0) {  // log_number_ == 0 indicates initial empty state
-      s = RecoverLogFile(log_number_, edit, &max_sequence);
+    if (versions_->PrevLogNumber() != 0) { // log#==0 means no prev log
+      s = RecoverLogFile(versions_->PrevLogNumber(), edit, &max_sequence);
+    }
+    if (s.ok() && versions_->LogNumber() != 0) { // log#==0 for initial state
+      s = RecoverLogFile(versions_->LogNumber(), edit, &max_sequence);
     }
     if (s.ok()) {
-      last_sequence_ =
-          last_sequence_ > max_sequence ? last_sequence_ : max_sequence;
+      if (versions_->LastSequence() < max_sequence) {
+        versions_->SetLastSequence(max_sequence);
+      }
     }
   }
 
@@ -407,56 +409,58 @@ Status DBImpl::RecoverLogFile(uint64_t log_number,
 
 Status DBImpl::WriteLevel0Table(MemTable* mem, VersionEdit* edit) {
   mutex_.AssertHeld();
+  const uint64_t start_micros = env_->NowMicros();
   FileMetaData meta;
   meta.number = versions_->NewFileNumber();
   pending_outputs_.insert(meta.number);
   Iterator* iter = mem->NewIterator();
   Log(env_, options_.info_log, "Level-0 table #%llu: started",
       (unsigned long long) meta.number);
-  Status s = BuildTable(dbname_, env_, options_, table_cache_,
-                        iter, &meta, edit);
+
+  Status s;
+  {
+    mutex_.Unlock();
+    s = BuildTable(dbname_, env_, options_, table_cache_, iter, &meta, edit);
+    mutex_.Lock();
+  }
+
   Log(env_, options_.info_log, "Level-0 table #%llu: %lld bytes %s",
       (unsigned long long) meta.number,
       (unsigned long long) meta.file_size,
       s.ToString().c_str());
   delete iter;
   pending_outputs_.erase(meta.number);
+
+  CompactionStats stats;
+  stats.micros = env_->NowMicros() - start_micros;
+  stats.bytes_written = meta.file_size;
+  stats_[0].Add(stats);
   return s;
 }
 
 Status DBImpl::CompactMemTable() {
   mutex_.AssertHeld();
-
-  WritableFile* lfile = NULL;
-  uint64_t new_log_number = versions_->NewFileNumber();
-
-  VersionEdit edit;
+  assert(imm_ != NULL);
+  assert(compacting_);
 
   // Save the contents of the memtable as a new Table
-  Status s = WriteLevel0Table(mem_, &edit);
-  if (s.ok()) {
-    s = env_->NewWritableFile(LogFileName(dbname_, new_log_number), &lfile);
-  }
+  VersionEdit edit;
+  Status s = WriteLevel0Table(imm_, &edit);
 
-  // Save a new descriptor with the new table and log number.
+  // Replace immutable memtable with the generated Table
   if (s.ok()) {
-    s = Install(&edit, new_log_number, mem_);
+    edit.SetPrevLogNumber(0);
+    s = versions_->LogAndApply(&edit, imm_);
   }
 
   if (s.ok()) {
     // Commit to the new state
-    mem_ = new MemTable(internal_comparator_);
-    delete log_;
-    delete logfile_;
-    logfile_ = lfile;
-    log_ = new log::Writer(lfile);
-    log_number_ = new_log_number;
+    imm_ = NULL;
+    has_imm_.Release_Store(NULL);
     DeleteObsoleteFiles();
-    MaybeScheduleCompaction();
-  } else {
-    delete lfile;
-    env_->DeleteFile(LogFileName(dbname_, new_log_number));
   }
+
+  compacting_cv_.SignalAll();  // Wake up waiter even if there was an error
   return s;
 }
 
@@ -485,7 +489,17 @@ void DBImpl::TEST_CompactRange(
 
 Status DBImpl::TEST_CompactMemTable() {
   MutexLock l(&mutex_);
-  return CompactMemTable();
+  Status s = MakeRoomForWrite(true /* force compaction */);
+  if (s.ok()) {
+    // Wait until the compaction completes
+    while (imm_ != NULL && bg_error_.ok()) {
+      compacting_cv_.Wait();
+    }
+    if (imm_ != NULL) {
+      s = bg_error_;
+    }
+  }
+  return s;
 }
 
 void DBImpl::MaybeScheduleCompaction() {
@@ -496,7 +510,7 @@ void DBImpl::MaybeScheduleCompaction() {
     // Some other thread is running a compaction.  Do not conflict with it.
   } else if (shutting_down_.Acquire_Load()) {
     // DB is being deleted; no more background compactions
-  } else if (!versions_->NeedsCompaction()) {
+  } else if (imm_ == NULL && !versions_->NeedsCompaction()) {
     // No work to be done
   } else {
     bg_compaction_scheduled_ = true;
@@ -525,6 +539,16 @@ void DBImpl::BackgroundCall() {
 
 void DBImpl::BackgroundCompaction() {
   mutex_.AssertHeld();
+  assert(!compacting_);
+
+  if (imm_ != NULL) {
+    compacting_ = true;
+    CompactMemTable();
+    compacting_ = false;
+    compacting_cv_.SignalAll();
+    return;
+  }
+
   Compaction* c = versions_->PickCompaction();
   if (c == NULL) {
     // Nothing to do
@@ -539,7 +563,7 @@ void DBImpl::BackgroundCompaction() {
     c->edit()->DeleteFile(c->level(), f->number);
     c->edit()->AddFile(c->level() + 1, f->number, f->file_size,
                        f->smallest, f->largest);
-    status = Install(c->edit(), log_number_, NULL);
+    status = versions_->LogAndApply(c->edit(), NULL);
     Log(env_, options_.info_log, "Moved #%lld to level-%d %lld bytes %s\n",
         static_cast<unsigned long long>(f->number),
         c->level() + 1,
@@ -680,7 +704,7 @@ Status DBImpl::InstallCompactionResults(CompactionState* compact) {
   }
   compact->outputs.clear();
 
-  Status s = Install(compact->compaction->edit(), log_number_, NULL);
+  Status s = versions_->LogAndApply(compact->compaction->edit(), NULL);
   if (s.ok()) {
     compact->compaction->ReleaseInputs();
     DeleteObsoleteFiles();
@@ -694,6 +718,9 @@ Status DBImpl::InstallCompactionResults(CompactionState* compact) {
 }
 
 Status DBImpl::DoCompactionWork(CompactionState* compact) {
+  const uint64_t start_micros = env_->NowMicros();
+  int64_t imm_micros = 0;  // Micros spent doing imm_ compactions
+
   Log(env_, options_.info_log,  "Compacting %d@%d + %d@%d files",
       compact->compaction->num_input_files(0),
       compact->compaction->level(),
@@ -704,7 +731,7 @@ Status DBImpl::DoCompactionWork(CompactionState* compact) {
   assert(compact->builder == NULL);
   assert(compact->outfile == NULL);
   if (snapshots_.empty()) {
-    compact->smallest_snapshot = last_sequence_;
+    compact->smallest_snapshot = versions_->LastSequence();
   } else {
     compact->smallest_snapshot = snapshots_.oldest()->number_;
   }
@@ -721,6 +748,18 @@ Status DBImpl::DoCompactionWork(CompactionState* compact) {
   bool has_current_user_key = false;
   SequenceNumber last_sequence_for_key = kMaxSequenceNumber;
   for (; input->Valid() && !shutting_down_.Acquire_Load(); ) {
+    // Prioritize immutable compaction work
+    if (has_imm_.NoBarrier_Load() != NULL) {
+      const uint64_t imm_start = env_->NowMicros();
+      mutex_.Lock();
+      if (imm_ != NULL) {
+        CompactMemTable();
+        compacting_cv_.SignalAll();  // Wakeup MakeRoomForWrite() if necessary
+      }
+      mutex_.Unlock();
+      imm_micros += (env_->NowMicros() - imm_start);
+    }
+
     Slice key = input->key();
     InternalKey tmp_internal_key;
     tmp_internal_key.DecodeFrom(key);
@@ -835,7 +874,19 @@ Status DBImpl::DoCompactionWork(CompactionState* compact) {
   delete input;
   input = NULL;
 
+  CompactionStats stats;
+  stats.micros = env_->NowMicros() - start_micros - imm_micros;
+  for (int which = 0; which < 2; which++) {
+    for (int i = 0; i < compact->compaction->num_input_files(which); i++) {
+      stats.bytes_read += compact->compaction->input(which, i)->file_size;
+    }
+  }
+  for (int i = 0; i < compact->outputs.size(); i++) {
+    stats.bytes_written += compact->outputs[i].file_size;
+  }
+
   mutex_.Lock();
+  stats_[compact->compaction->level() + 1].Add(stats);
 
   if (status.ok()) {
     status = InstallCompactionResults(compact);
@@ -848,11 +899,14 @@ Status DBImpl::DoCompactionWork(CompactionState* compact) {
 Iterator* DBImpl::NewInternalIterator(const ReadOptions& options,
                                       SequenceNumber* latest_snapshot) {
   mutex_.Lock();
-  *latest_snapshot = last_sequence_;
+  *latest_snapshot = versions_->LastSequence();
 
   // Collect together all needed child iterators
   std::vector<Iterator*> list;
   list.push_back(mem_->NewIterator());
+  if (imm_ != NULL) {
+    list.push_back(imm_->NewIterator());
+  }
   versions_->current()->AddIterators(options, &list);
   Iterator* internal_iter =
       NewMergingIterator(&internal_comparator_, &list[0], list.size());
@@ -912,7 +966,7 @@ void DBImpl::Unref(void* arg1, void* arg2) {
 
 const Snapshot* DBImpl::GetSnapshot() {
   MutexLock l(&mutex_);
-  return snapshots_.New(last_sequence_);
+  return snapshots_.New(versions_->LastSequence());
 }
 
 void DBImpl::ReleaseSnapshot(const Snapshot* s) {
@@ -935,17 +989,16 @@ Status DBImpl::Write(const WriteOptions& options, WriteBatch* updates) {
   WriteBatch* final = NULL;
   {
     MutexLock l(&mutex_);
-    if (!bg_error_.ok()) {
-      status = bg_error_;
-    } else if (mem_->ApproximateMemoryUsage() > options_.write_buffer_size) {
-      status = CompactMemTable();
-    }
+    status = MakeRoomForWrite(false);  // May temporarily release lock and wait
+
+    uint64_t last_sequence = versions_->LastSequence();
     if (status.ok()) {
-      status = HandleLargeValues(last_sequence_ + 1, updates, &final);
+      status = HandleLargeValues(last_sequence + 1, updates, &final);
     }
     if (status.ok()) {
-      WriteBatchInternal::SetSequence(final, last_sequence_ + 1);
-      last_sequence_ += WriteBatchInternal::Count(final);
+      WriteBatchInternal::SetSequence(final, last_sequence + 1);
+      last_sequence += WriteBatchInternal::Count(final);
+      versions_->SetLastSequence(last_sequence);
 
       // Add to log and apply to memtable
       status = log_->AddRecord(WriteBatchInternal::Contents(final));
@@ -959,7 +1012,7 @@ Status DBImpl::Write(const WriteOptions& options, WriteBatch* updates) {
 
     if (options.post_write_snapshot != NULL) {
       *options.post_write_snapshot =
-          status.ok() ? snapshots_.New(last_sequence_) : NULL;
+          status.ok() ? snapshots_.New(last_sequence) : NULL;
     }
   }
   if (final != updates) {
@@ -969,6 +1022,54 @@ Status DBImpl::Write(const WriteOptions& options, WriteBatch* updates) {
   return status;
 }
 
+Status DBImpl::MakeRoomForWrite(bool force) {
+  mutex_.AssertHeld();
+  Status s;
+  while (true) {
+    if (!bg_error_.ok()) {
+      // Yield previous error
+      s = bg_error_;
+      break;
+    } else if (!force &&
+               (mem_->ApproximateMemoryUsage() <= options_.write_buffer_size)) {
+      // There is room in current memtable
+      break;
+    } else if (imm_ != NULL) {
+      // We have filled up the current memtable, but the previous
+      // one is still being compacted, so we wait.
+      compacting_cv_.Wait();
+    } else {
+      // Attempt to switch to a new memtable and trigger compaction of old
+      assert(versions_->PrevLogNumber() == 0);
+      uint64_t new_log_number = versions_->NewFileNumber();
+      WritableFile* lfile = NULL;
+      s = env_->NewWritableFile(LogFileName(dbname_, new_log_number), &lfile);
+      if (!s.ok()) {
+        break;
+      }
+      VersionEdit edit;
+      edit.SetPrevLogNumber(versions_->LogNumber());
+      edit.SetLogNumber(new_log_number);
+      s = versions_->LogAndApply(&edit, NULL);
+      if (!s.ok()) {
+        delete lfile;
+        env_->DeleteFile(LogFileName(dbname_, new_log_number));
+        break;
+      }
+      delete log_;
+      delete logfile_;
+      logfile_ = lfile;
+      log_ = new log::Writer(lfile);
+      imm_ = mem_;
+      has_imm_.Release_Store(imm_);
+      mem_ = new MemTable(internal_comparator_);
+      force = false;   // Do not force another compaction if have room
+      MaybeScheduleCompaction();
+    }
+  }
+  return s;
+}
+
 bool DBImpl::HasLargeValues(const WriteBatch& batch) const {
   if (WriteBatchInternal::ByteSize(&batch) >= options_.large_value_threshold) {
     for (WriteBatchInternal::Iterator it(batch); !it.Done(); it.Next()) {
@@ -1033,9 +1134,10 @@ Status DBImpl::HandleLargeValues(SequenceNumber assigned_seq,
             MaybeCompressLargeValue(
                 it.value(), &file_bytes, &scratch, &large_ref);
             InternalKey ikey(it.key(), seq, kTypeLargeValueRef);
-            if (versions_->RegisterLargeValueRef(large_ref, log_number_,ikey)) {
+            if (versions_->RegisterLargeValueRef(
+                    large_ref, versions_->LogNumber(), ikey)) {
               // TODO(opt): avoid holding the lock here (but be careful about
-              // another thread doing a Write and changing log_number_ or
+              // another thread doing a Write and switching logs or
               // having us get a different "assigned_seq" value).
 
               uint64_t tmp_number = versions_->NewFileNumber();
@@ -1086,7 +1188,9 @@ Status DBImpl::HandleLargeValues(SequenceNumber assigned_seq,
   return Status::OK();
 }
 
-bool DBImpl::GetProperty(const Slice& property, uint64_t* value) {
+bool DBImpl::GetProperty(const Slice& property, std::string* value) {
+  value->clear();
+
   MutexLock l(&mutex_);
   Slice in = property;
   Slice prefix("leveldb.");
@@ -1100,10 +1204,37 @@ bool DBImpl::GetProperty(const Slice& property, uint64_t* value) {
     if (!ok || level < 0 || level >= config::kNumLevels) {
       return false;
     } else {
-      *value = versions_->NumLevelFiles(level);
+      char buf[100];
+      snprintf(buf, sizeof(buf), "%d", versions_->NumLevelFiles(level));
+      *value = buf;
       return true;
     }
+  } else if (in == "stats") {
+    char buf[200];
+    snprintf(buf, sizeof(buf),
+             "                               Compactions\n"
+             "Level  Files Size(MB) Time(sec) Read(MB) Write(MB)\n"
+             "--------------------------------------------------\n"
+             );
+    value->append(buf);
+    for (int level = 0; level < config::kNumLevels; level++) {
+      int files = versions_->NumLevelFiles(level);
+      if (stats_[level].micros > 0 || files > 0) {
+        snprintf(
+            buf, sizeof(buf),
+            "%3d %8d %8.0f %9.0f %8.0f %9.0f\n",
+            level,
+            files,
+            versions_->NumLevelBytes(level) / 1048576.0,
+            stats_[level].micros / 1e6,
+            stats_[level].bytes_read / 1048576.0,
+            stats_[level].bytes_written / 1048576.0);
+        value->append(buf);
+      }
+    }
+    return true;
   }
+
   return false;
 }
 
@@ -1158,14 +1289,15 @@ Status DB::Open(const Options& options, const std::string& dbname,
   VersionEdit edit;
   Status s = impl->Recover(&edit); // Handles create_if_missing, error_if_exists
   if (s.ok()) {
-    impl->log_number_ = impl->versions_->NewFileNumber();
+    uint64_t new_log_number = impl->versions_->NewFileNumber();
     WritableFile* lfile;
-    s = options.env->NewWritableFile(LogFileName(dbname, impl->log_number_),
+    s = options.env->NewWritableFile(LogFileName(dbname, new_log_number),
                                      &lfile);
     if (s.ok()) {
+      edit.SetLogNumber(new_log_number);
       impl->logfile_ = lfile;
       impl->log_ = new log::Writer(lfile);
-      s = impl->Install(&edit, impl->log_number_, NULL);
+      s = impl->versions_->LogAndApply(&edit, NULL);
     }
     if (s.ok()) {
       impl->DeleteObsoleteFiles();

db/db_impl.h

@@ -36,7 +36,7 @@ class DBImpl : public DB {
   virtual Iterator* NewIterator(const ReadOptions&);
   virtual const Snapshot* GetSnapshot();
   virtual void ReleaseSnapshot(const Snapshot* snapshot);
-  virtual bool GetProperty(const Slice& property, uint64_t* value);
+  virtual bool GetProperty(const Slice& property, std::string* value);
   virtual void GetApproximateSizes(const Range* range, int n, uint64_t* sizes);
 
   // Extra methods (for testing) that are not in the public DB interface
@@ -72,14 +72,6 @@ class DBImpl : public DB {
   // be made to the descriptor are added to *edit.
   Status Recover(VersionEdit* edit);
 
-  // Apply the specified updates and save the resulting descriptor to
-  // persistent storage.  If cleanup_mem is non-NULL, arrange to
-  // delete it when all existing snapshots have gone away iff Install()
-  // returns OK.
-  Status Install(VersionEdit* edit,
-                 uint64_t new_log_number,
-                 MemTable* cleanup_mem);
-
   void MaybeIgnoreError(Status* s) const;
 
   // Delete any unneeded files and stale in-memory entries.
@@ -99,6 +91,7 @@ class DBImpl : public DB {
 
   Status WriteLevel0Table(MemTable* mem, VersionEdit* edit);
 
+  Status MakeRoomForWrite(bool force /* compact even if there is room? */);
   bool HasLargeValues(const WriteBatch& batch) const;
 
   // Process data in "*updates" and return a status.  "assigned_seq"
@@ -141,6 +134,7 @@ class DBImpl : public DB {
   const InternalKeyComparator internal_comparator_;
   const Options options_;  // options_.comparator == &internal_comparator_
   bool owns_info_log_;
+  bool owns_cache_;
   const std::string dbname_;
 
   // table_cache_ provides its own synchronization
@@ -152,13 +146,13 @@ class DBImpl : public DB {
   // State below is protected by mutex_
   port::Mutex mutex_;
   port::AtomicPointer shutting_down_;
-  port::CondVar bg_cv_;         // Signalled when !bg_compaction_scheduled_
+  port::CondVar bg_cv_;          // Signalled when !bg_compaction_scheduled_
   port::CondVar compacting_cv_;  // Signalled when !compacting_
-  SequenceNumber last_sequence_;
   MemTable* mem_;
+  MemTable* imm_;                // Memtable being compacted
+  port::AtomicPointer has_imm_;  // So bg thread can detect non-NULL imm_
   WritableFile* logfile_;
   log::Writer* log_;
-  uint64_t log_number_;
   SnapshotList snapshots_;
 
   // Set of table files to protect from deletion because they are
@@ -176,6 +170,23 @@ class DBImpl : public DB {
   // Have we encountered a background error in paranoid mode?
   Status bg_error_;
 
+  // Per level compaction stats.  stats_[level] stores the stats for
+  // compactions that produced data for the specified "level".
+  struct CompactionStats {
+    int64_t micros;
+    int64_t bytes_read;
+    int64_t bytes_written;
+
+    CompactionStats() : micros(0), bytes_read(0), bytes_written(0) { }
+
+    void Add(const CompactionStats& c) {
+      this->micros += c.micros;
+      this->bytes_read += c.bytes_read;
+      this->bytes_written += c.bytes_written;
+    }
+  };
+  CompactionStats stats_[config::kNumLevels];
+
   // No copying allowed
   DBImpl(const DBImpl&);
   void operator=(const DBImpl&);

db/db_test.cc

@@ -72,19 +72,11 @@ class DBTest {
   }
 
   Status Put(const std::string& k, const std::string& v) {
-    WriteOptions options;
-    options.sync = false;
-    WriteBatch batch;
-    batch.Put(k, v);
-    return db_->Write(options, &batch);
+    return db_->Put(WriteOptions(), k, v);
   }
 
   Status Delete(const std::string& k) {
-    WriteOptions options;
-    options.sync = false;
-    WriteBatch batch;
-    batch.Delete(k);
-    return db_->Write(options, &batch);
+    return db_->Delete(WriteOptions(), k);
   }
 
   std::string Get(const std::string& k, const Snapshot* snapshot = NULL) {
@@ -147,11 +139,11 @@ class DBTest {
   }
 
   int NumTableFilesAtLevel(int level) {
-    uint64_t val;
+    std::string property;
     ASSERT_TRUE(
         db_->GetProperty("leveldb.num-files-at-level" + NumberToString(level),
-                         &val));
-    return val;
+                         &property));
+    return atoi(property.c_str());
   }
 
   uint64_t Size(const Slice& start, const Slice& limit) {
@@ -185,10 +177,7 @@ class DBTest {
     dbfull()->TEST_CompactMemTable();
     int max_level_with_files = 1;
     for (int level = 1; level < config::kNumLevels; level++) {
-      uint64_t v;
-      char name[100];
-      snprintf(name, sizeof(name), "leveldb.num-files-at-level%d", level);
-      if (dbfull()->GetProperty(name, &v) && v > 0) {
+      if (NumTableFilesAtLevel(level) > 0) {
         max_level_with_files = level;
       }
     }
@@ -459,7 +448,7 @@ TEST(DBTest, MinorCompactionsHappen) {
   options.write_buffer_size = 10000;
   Reopen(&options);
 
-  const int N = 100;
+  const int N = 500;
 
   int starting_num_tables = NumTableFilesAtLevel(0);
   for (int i = 0; i < N; i++) {
@@ -1047,7 +1036,7 @@ class ModelDB: public DB {
     return Status::OK();
   }
 
-  virtual bool GetProperty(const Slice& property, uint64_t* value) {
+  virtual bool GetProperty(const Slice& property, std::string* value) {
     return false;
   }
   virtual void GetApproximateSizes(const Range* r, int n, uint64_t* sizes) {

db/dbformat.h

@@ -15,6 +15,12 @@
 
 namespace leveldb {
 
+// Grouping of constants.  We may want to make some of these
+// parameters set via options.
+namespace config {
+static const int kNumLevels = 7;
+}
+
 class InternalKey;
 
 // Value types encoded as the last component of internal keys.

db/version_edit.cc

@@ -20,15 +20,18 @@ enum Tag {
   kDeletedFile          = 6,
   kNewFile              = 7,
   kLargeValueRef        = 8,
+  kPrevLogNumber        = 9,
 };
 
 void VersionEdit::Clear() {
   comparator_.clear();
   log_number_ = 0;
+  prev_log_number_ = 0;
   last_sequence_ = 0;
   next_file_number_ = 0;
   has_comparator_ = false;
   has_log_number_ = false;
+  has_prev_log_number_ = false;
   has_next_file_number_ = false;
   has_last_sequence_ = false;
   deleted_files_.clear();
@@ -45,6 +48,10 @@ void VersionEdit::EncodeTo(std::string* dst) const {
     PutVarint32(dst, kLogNumber);
     PutVarint64(dst, log_number_);
   }
+  if (has_prev_log_number_) {
+    PutVarint32(dst, kPrevLogNumber);
+    PutVarint64(dst, prev_log_number_);
+  }
   if (has_next_file_number_) {
     PutVarint32(dst, kNextFileNumber);
     PutVarint64(dst, next_file_number_);
@@ -142,6 +149,14 @@ Status VersionEdit::DecodeFrom(const Slice& src) {
         }
         break;
 
+      case kPrevLogNumber:
+        if (GetVarint64(&input, &prev_log_number_)) {
+          has_prev_log_number_ = true;
+        } else {
+          msg = "previous log number";
+        }
+        break;
+
       case kNextFileNumber:
         if (GetVarint64(&input, &next_file_number_)) {
           has_next_file_number_ = true;
@@ -228,6 +243,10 @@ std::string VersionEdit::DebugString() const {
     r.append("\n  LogNumber: ");
     AppendNumberTo(&r, log_number_);
   }
+  if (has_prev_log_number_) {
+    r.append("\n  PrevLogNumber: ");
+    AppendNumberTo(&r, prev_log_number_);
+  }
   if (has_next_file_number_) {
     r.append("\n  NextFile: ");
     AppendNumberTo(&r, next_file_number_);

db/version_edit.h

@@ -39,6 +39,10 @@ class VersionEdit {
     has_log_number_ = true;
     log_number_ = num;
   }
+  void SetPrevLogNumber(uint64_t num) {
+    has_prev_log_number_ = true;
+    prev_log_number_ = num;
+  }
   void SetNextFile(uint64_t num) {
     has_next_file_number_ = true;
     next_file_number_ = num;
@@ -95,10 +99,12 @@ class VersionEdit {
 
   std::string comparator_;
   uint64_t log_number_;
+  uint64_t prev_log_number_;
   uint64_t next_file_number_;
   SequenceNumber last_sequence_;
   bool has_comparator_;
   bool has_log_number_;
+  bool has_prev_log_number_;
   bool has_next_file_number_;
   bool has_last_sequence_;
 

db/version_set.cc

@@ -27,16 +27,14 @@ static const int kTargetFileSize = 2 * 1048576;
 static const int64_t kMaxGrandParentOverlapBytes = 10 * kTargetFileSize;
 
 static double MaxBytesForLevel(int level) {
-  if (level == 0) {
-    return 4 * 1048576.0;
-  } else {
-    double result = 10 * 1048576.0;
-    while (level > 1) {
-      result *= 10;
-      level--;
-    }
-    return result;
+  // Note: the result for level zero is not really used since we set
+  // the level-0 compaction threshold based on number of files.
+  double result = 10 * 1048576.0;  // Result for both level-0 and level-1
+  while (level > 1) {
+    result *= 10;
+    level--;
   }
+  return result;
 }
 
 static uint64_t MaxFileSizeForLevel(int level) {
@@ -327,6 +325,9 @@ VersionSet::VersionSet(const std::string& dbname,
       icmp_(*cmp),
       next_file_number_(2),
       manifest_file_number_(0),  // Filled by Recover()
+      last_sequence_(0),
+      log_number_(0),
+      prev_log_number_(0),
       descriptor_file_(NULL),
       descriptor_log_(NULL),
       current_(new Version(this)),
@@ -345,7 +346,19 @@ VersionSet::~VersionSet() {
 }
 
 Status VersionSet::LogAndApply(VersionEdit* edit, MemTable* cleanup_mem) {
+  if (edit->has_log_number_) {
+    assert(edit->log_number_ >= log_number_);
+    assert(edit->log_number_ < next_file_number_);
+  } else {
+    edit->SetLogNumber(log_number_);
+  }
+
+  if (!edit->has_prev_log_number_) {
+    edit->SetPrevLogNumber(prev_log_number_);
+  }
+
   edit->SetNextFile(next_file_number_);
+  edit->SetLastSequence(last_sequence_);
 
   Version* v = new Version(this);
   {
@@ -372,7 +385,7 @@ Status VersionSet::LogAndApply(VersionEdit* edit, MemTable* cleanup_mem) {
     }
   }
 
-  // Write new record to log file
+  // Write new record to MANIFEST log
   if (s.ok()) {
     std::string record;
     edit->EncodeTo(&record);
@@ -396,6 +409,8 @@ Status VersionSet::LogAndApply(VersionEdit* edit, MemTable* cleanup_mem) {
     v->next_ = NULL;
     current_->next_ = v;
     current_ = v;
+    log_number_ = edit->log_number_;
+    prev_log_number_ = edit->prev_log_number_;
   } else {
     delete v;
     if (!new_manifest_file.empty()) {
@@ -406,13 +421,11 @@ Status VersionSet::LogAndApply(VersionEdit* edit, MemTable* cleanup_mem) {
       env_->DeleteFile(new_manifest_file);
     }
   }
-  //Log(env_, options_->info_log, "State\n%s", current_->DebugString().c_str());
 
   return s;
 }
 
-Status VersionSet::Recover(uint64_t* log_number,
-                           SequenceNumber* last_sequence) {
+Status VersionSet::Recover() {
   struct LogReporter : public log::Reader::Reporter {
     Status* status;
     virtual void Corruption(size_t bytes, const Status& s) {
@@ -439,9 +452,13 @@ Status VersionSet::Recover(uint64_t* log_number,
   }
 
   bool have_log_number = false;
+  bool have_prev_log_number = false;
   bool have_next_file = false;
   bool have_last_sequence = false;
   uint64_t next_file = 0;
+  uint64_t last_sequence = 0;
+  uint64_t log_number = 0;
+  uint64_t prev_log_number = 0;
   Builder builder(this, current_);
 
   {
@@ -467,17 +484,22 @@ Status VersionSet::Recover(uint64_t* log_number,
       }
 
       if (edit.has_log_number_) {
-        *log_number = edit.log_number_;
+        log_number = edit.log_number_;
         have_log_number = true;
       }
 
+      if (edit.has_prev_log_number_) {
+        prev_log_number = edit.prev_log_number_;
+        have_prev_log_number = true;
+      }
+
       if (edit.has_next_file_number_) {
         next_file = edit.next_file_number_;
         have_next_file = true;
       }
 
       if (edit.has_last_sequence_) {
-        *last_sequence = edit.last_sequence_;
+        last_sequence = edit.last_sequence_;
         have_last_sequence = true;
       }
     }
@@ -493,6 +515,10 @@ Status VersionSet::Recover(uint64_t* log_number,
     } else if (!have_last_sequence) {
       s = Status::Corruption("no last-sequence-number entry in descriptor");
     }
+
+    if (!have_prev_log_number) {
+      prev_log_number = 0;
+    }
   }
 
   if (s.ok()) {
@@ -508,12 +534,23 @@ Status VersionSet::Recover(uint64_t* log_number,
       current_ = v;
       manifest_file_number_ = next_file;
       next_file_number_ = next_file + 1;
+      last_sequence_ = last_sequence;
+      log_number_ = log_number;
+      prev_log_number_ = prev_log_number;
     }
   }
 
   return s;
 }
 
+static int64_t TotalFileSize(const std::vector<FileMetaData*>& files) {
+  int64_t sum = 0;
+  for (int i = 0; i < files.size(); i++) {
+    sum += files[i]->file_size;
+  }
+  return sum;
+}
+
 Status VersionSet::Finalize(Version* v) {
   // Precomputed best level for next compaction
   int best_level = -1;
@@ -523,23 +560,24 @@ Status VersionSet::Finalize(Version* v) {
   for (int level = 0; s.ok() && level < config::kNumLevels-1; level++) {
     s = SortLevel(v, level);
 
-    // Compute the ratio of current size to size limit.
-    uint64_t level_bytes = 0;
-    for (int i = 0; i < v->files_[level].size(); i++) {
-      level_bytes += v->files_[level][i]->file_size;
-    }
-    double score = static_cast<double>(level_bytes) / MaxBytesForLevel(level);
-
+    double score;
     if (level == 0) {
-      // Level-0 file sizes are going to be often much smaller than
-      // MaxBytesForLevel(0) since we do not account for compression
-      // when producing a level-0 file; and too many level-0 files
-      // increase merging costs.  So use a file-count limit for
-      // level-0 in addition to the byte-count limit.
-      double count_score = v->files_[level].size() / 4.0;
-      if (count_score > score) {
-        score = count_score;
-      }
+      // We treat level-0 specially by bounding the number of files
+      // instead of number of bytes for two reasons:
+      //
+      // (1) With larger write-buffer sizes, it is nice not to do too
+      // many level-0 compactions.
+      //
+      // (2) The files in level-0 are merged on every read and
+      // therefore we wish to avoid too many files when the individual
+      // file size is small (perhaps because of a small write-buffer
+      // setting, or very high compression ratios, or lots of
+      // overwrites/deletions).
+      score = v->files_[level].size() / 4.0;
+    } else {
+      // Compute the ratio of current size to size limit.
+      const uint64_t level_bytes = TotalFileSize(v->files_[level]);
+      score = static_cast<double>(level_bytes) / MaxBytesForLevel(level);
     }
 
     if (score > best_score) {
@@ -696,8 +734,7 @@ bool VersionSet::RegisterLargeValueRef(const LargeValueRef& large_ref,
   return is_first;
 }
 
-void VersionSet::CleanupLargeValueRefs(const std::set<uint64_t>& live_tables,
-                                       uint64_t log_file_num) {
+void VersionSet::CleanupLargeValueRefs(const std::set<uint64_t>& live_tables) {
   for (LargeValueMap::iterator it = large_value_refs_.begin();
        it != large_value_refs_.end();
        ) {
@@ -705,7 +742,8 @@ void VersionSet::CleanupLargeValueRefs(const std::set<uint64_t>& live_tables,
     for (LargeReferencesSet::iterator ref_it = refs->begin();
          ref_it != refs->end();
          ) {
-      if (ref_it->first != log_file_num &&              // Not in log file
+      if (ref_it->first != log_number_ &&               // Not in log file
+          ref_it->first != prev_log_number_ &&          // Not in prev log
           live_tables.count(ref_it->first) == 0) {      // Not in a live table
         // No longer live: erase
         LargeReferencesSet::iterator to_erase = ref_it;
@@ -762,12 +800,10 @@ void VersionSet::AddLiveFiles(std::set<uint64_t>* live) {
   }
 }
 
-static int64_t TotalFileSize(const std::vector<FileMetaData*>& files) {
-  int64_t sum = 0;
-  for (int i = 0; i < files.size(); i++) {
-    sum += files[i]->file_size;
-  }
-  return sum;
+int64_t VersionSet::NumLevelBytes(int level) const {
+  assert(level >= 0);
+  assert(level < config::kNumLevels);
+  return TotalFileSize(current_->files_[level]);
 }
 
 int64_t VersionSet::MaxNextLevelOverlappingBytes() {

db/version_set.h

@@ -24,12 +24,6 @@
 
 namespace leveldb {
 
-// Grouping of constants.  We may want to make some of these
-// parameters set via options.
-namespace config {
-static const int kNumLevels = 7;
-}
-
 namespace log { class Writer; }
 
 class Compaction;
@@ -107,7 +101,7 @@ class VersionSet {
   Status LogAndApply(VersionEdit* edit, MemTable* cleanup_mem);
 
   // Recover the last saved descriptor from persistent storage.
-  Status Recover(uint64_t* log_number, SequenceNumber* last_sequence);
+  Status Recover();
 
   // Save current contents to *log
   Status WriteSnapshot(log::Writer* log);
@@ -124,6 +118,25 @@ class VersionSet {
   // Return the number of Table files at the specified level.
   int NumLevelFiles(int level) const;
 
+  // Return the combined file size of all files at the specified level.
+  int64_t NumLevelBytes(int level) const;
+
+  // Return the last sequence number.
+  uint64_t LastSequence() const { return last_sequence_; }
+
+  // Set the last sequence number to s.
+  void SetLastSequence(uint64_t s) {
+    assert(s >= last_sequence_);
+    last_sequence_ = s;
+  }
+
+  // Return the current log file number.
+  uint64_t LogNumber() const { return log_number_; }
+
+  // Return the log file number for the log file that is currently
+  // being compacted, or zero if there is no such log file.
+  uint64_t PrevLogNumber() const { return prev_log_number_; }
+
   // Pick level and inputs for a new compaction.
   // Returns NULL if there is no compaction to be done.
   // Otherwise returns a pointer to a heap-allocated object that
@@ -168,9 +181,8 @@ class VersionSet {
 
   // Cleanup the large value reference state by eliminating any
   // references from files that are not includes in either "live_tables"
-  // or "log_file".
-  void CleanupLargeValueRefs(const std::set<uint64_t>& live_tables,
-                             uint64_t log_file_num);
+  // or the current log.
+  void CleanupLargeValueRefs(const std::set<uint64_t>& live_tables);
 
   // Returns true if a large value with the given reference is live.
   bool LargeValueIsLive(const LargeValueRef& large_ref);
@@ -213,6 +225,9 @@ class VersionSet {
   const InternalKeyComparator icmp_;
   uint64_t next_file_number_;
   uint64_t manifest_file_number_;
+  uint64_t last_sequence_;
+  uint64_t log_number_;
+  uint64_t prev_log_number_;  // 0 or backing store for memtable being compacted
 
   // Opened lazily
   WritableFile* descriptor_file_;

doc/index.html

@@ -63,15 +63,12 @@ Example:
 The database provides <code>Put</code>, <code>Delete</code>, and <code>Get</code> methods to
 modify/query the database.  For example, the following code
 moves the value stored under key1 to key2.
-<p>
 <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);
 </pre>
-See <a href="#async">important performance note</a> below for how to
-speed up writes significantly.
 
 <h1>Atomic Updates</h1>
 <p>
@@ -100,6 +97,47 @@ we do not end up erroneously dropping the value entirely.
 Apart from its atomicity benefits, <code>WriteBatch</code> may also be used to
 speed up bulk updates by placing lots of individual mutations into the
 same batch.
+
+<h1>Synchronous Writes</h1>
+By default, each write to <code>leveldb</code> is asynchronous: it
+returns after pushing the write from the process into the operating
+system.  The transfer from operating system memory to the underlying
+persistent storage happens asynchronously.  The <code>sync</code> flag
+can be turned on for a particular write to make the write operation
+not return until the data being written has been pushed all the way to
+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;
+  write_options.sync = true;
+  db-&gt;Put(write_options, ...);
+</pre>
+Asynchronous writes are often more than a thousand times as fast as
+synchronous writes.  The downside of asynchronous writes is that a
+crash of the machine may cause the last few updates to be lost.  Note
+that a crash of just the writing process (i.e., not a reboot) will not
+cause any loss since even when <code>sync</code> is false, an update
+is pushed from the process memory into the operating system before it
+is considered done.
+
+<p>
+Asynchronous writes can often be used safely.  For example, when
+loading a large amount of data into the database you can handle lost
+updates by restarting the bulk load after a crash.  A hybrid scheme is
+also possible where every Nth write is synchronous, and in the event
+of a crash, the bulk load is restarted just after the last synchronous
+write finished by the previous run.  (The synchronous write can update
+a marker that describes where to restart on a crash.)
+
+<p>
+<code>WriteBatch</code> provides an alternative to asynchronous writes.
+Multiple updates may be placed in the same <code>WriteBatch</code> and
+applied together using a synchronous write (i.e.,
+<code>write_options.sync</code> is set to true).  The extra cost of
+the synchronous write will be amortized across all of the writes in
+the batch.
+
 <p>
 <h1>Concurrency</h1>
 <p>
@@ -289,48 +327,12 @@ version numbers found in the keys to decide how to interpret them.
 Performance can be tuned by changing the default values of the
 types defined in <code>leveldb/include/options.h</code>.
 
-<p>
-<h2><a name="async">Asynchronous Writes</a></h2>
-
-By default, each write to <code>leveldb</code> is synchronous: it does
-not return until the write has been pushed from memory to persistent
-storage.  (On Posix systems, this is implemented by calling either
-<code>fdatasync(...)</code> or <code>msync(..., MS_SYNC)</code>.)
-<strong>Synchronous writes may be very slow and the synchrony can be
-optionally disabled</strong>:
-<pre>
-  leveldb::WriteOptions write_options;
-  write_options.sync = false;
-  db-&gt;Put(write_options, ...);
-</pre>
-Asynchronous writes are often more than a hundred times as fast as
-synchronous writes.  The downside of asynchronous writes is that a
-crash of the machine may cause the last few updates to be lost.  Note
-that a crash of just the writing process (i.e., not a reboot) will not
-cause any loss since even when <code>sync</code> is false, an update
-is pushed from the process memory into the operating system before it
-is considered done.
-
-<p>
-Asynchronous writes can be particularly beneficial when loading a
-large amount of data into the database since you can mitigate the
-problem of lost updates by restarting the bulk load.  A hybrid scheme
-is also possible where every Nth write is synchronous, and in the
-event of a crash, the bulk load is restarted just after the last
-synchronous write finished by the previous run.
-
-<p>
-<code>WriteBatch</code> provides an alternative to asynchronous writes.
-Multiple updates may be placed in the same <code>WriteBatch</code> and
-applied together using a synchronous write.  The extra cost of the
-synchronous write will be amortized across all of the writes in the batch.
-
 <p>
 <h2>Block size</h2>
 <p>
 <code>leveldb</code> groups adjacent keys together into the same block and such a
 block is the unit of transfer to and from persistent storage.  The
-default block size is approximately 8192 uncompressed bytes.
+default block size is approximately 4096 uncompressed bytes.
 Applications that mostly do bulk scans over the contents of the
 database may wish to increase this size.  Applications that do a lot
 of point reads of small values may wish to switch to a smaller block

include/leveldb/db.h

@@ -13,7 +13,7 @@
 namespace leveldb {
 
 static const int kMajorVersion = 1;
-static const int kMinorVersion = 0;
+static const int kMinorVersion = 1;
 
 struct Options;
 struct ReadOptions;
@@ -49,7 +49,7 @@ class DB {
 
   // Set the database entry for "key" to "value".  Returns OK on success,
   // and a non-OK status on error.
-  // Note: consider setting options.sync = false.
+  // Note: consider setting options.sync = true.
   virtual Status Put(const WriteOptions& options,
                      const Slice& key,
                      const Slice& value) = 0;
@@ -57,12 +57,12 @@ class DB {
   // Remove the database entry (if any) for "key".  Returns OK on
   // success, and a non-OK status on error.  It is not an error if "key"
   // did not exist in the database.
-  // Note: consider setting options.sync = false.
+  // Note: consider setting options.sync = true.
   virtual Status Delete(const WriteOptions& options, const Slice& key) = 0;
 
   // Apply the specified updates to the database.
   // Returns OK on success, non-OK on failure.
-  // Note: consider setting options.sync = false.
+  // Note: consider setting options.sync = true.
   virtual Status Write(const WriteOptions& options, WriteBatch* updates) = 0;
 
   // If the database contains an entry for "key" store the
@@ -103,7 +103,9 @@ class DB {
   //
   //  "leveldb.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").
-  virtual bool GetProperty(const Slice& property, uint64_t* value) = 0;
+  //  "leveldb.stats" - returns a multi-line string that describes statistics
+  //     about the internal operation of the DB.
+  virtual bool GetProperty(const Slice& property, std::string* value) = 0;
 
   // For each i in [0,n-1], store in "sizes[i]", the approximate
   // file system space used by keys in "[range[i].start .. range[i].limit)".

include/leveldb/options.h

@@ -69,15 +69,14 @@ struct Options {
   // -------------------
   // Parameters that affect performance
 
-  // Amount of data to build up in memory before converting to an
-  // on-disk file.
+  // Amount of data to build up in memory (backed by an unsorted log
+  // on disk) before converting to a sorted on-disk file.
   //
-  // Some DB operations may encounter a delay proportional to the size
-  // of this parameter.  Therefore we recommend against increasing
-  // this parameter unless you are willing to live with an occasional
-  // slow operation in exchange for faster bulk loading throughput.
+  // Larger values increase performance, especially during bulk loads.
+  // Up to two write buffers may be held in memory at the same time,
+  // so you may wish to adjust this parameter to control memory usage.
   //
-  // Default: 1MB
+  // Default: 4MB
   size_t write_buffer_size;
 
   // Number of open files that can be used by the DB.  You may need to
@@ -100,7 +99,8 @@ struct Options {
   // Control over blocks (user data is stored in a set of blocks, and
   // a block is the unit of reading from disk).
 
-  // Use the specified cache for blocks (if non-NULL).
+  // If non-NULL, use the specified cache for blocks.
+  // If NULL, leveldb will automatically create and use an 8MB internal cache.
   // Default: NULL
   Cache* block_cache;
 
@@ -109,7 +109,7 @@ struct Options {
   // actual size of the unit read from disk may be smaller if
   // compression is enabled.  This parameter can be changed dynamically.
   //
-  // Default: 8K
+  // Default: 4K
   int block_size;
 
   // Number of keys between restart points for delta encoding of keys.
@@ -177,7 +177,12 @@ struct WriteOptions {
   // crashes (i.e., the machine does not reboot), no writes will be
   // lost even if sync==false.
   //
-  // Default: true
+  // In other words, a DB write with sync==false has similar
+  // crash semantics as the "write()" system call.  A DB write
+  // with sync==true has similar crash semantics to a "write()"
+  // system call followed by "fsync()".
+  //
+  // Default: false
   bool sync;
 
   // If "post_write_snapshot" is non-NULL, and the write succeeds,
@@ -193,7 +198,7 @@ struct WriteOptions {
   const Snapshot** post_write_snapshot;
 
   WriteOptions()
-      : sync(true),
+      : sync(false),
         post_write_snapshot(NULL) {
   }
 };

port/port_android.cc

@@ -24,7 +24,6 @@ int fdatasync(int fd) {
 }
 }
 
-// TODO(gabor): This is copied from port_posix.cc - not sure if I should do this?
 namespace leveldb {
 namespace port {
 

port/port_android.h

@@ -15,6 +15,20 @@
 #include <string>
 #include <cctype>
 
+// Collapse the plethora of ARM flavors available to an easier to manage set
+// Defs reference is at https://wiki.edubuntu.org/ARM/Thumb2PortingHowto
+#if defined(__ARM_ARCH_6__) || \
+    defined(__ARM_ARCH_6J__) || \
+    defined(__ARM_ARCH_6K__) || \
+    defined(__ARM_ARCH_6Z__) || \
+    defined(__ARM_ARCH_6T2__) || \
+    defined(__ARM_ARCH_6ZK__) || \
+    defined(__ARM_ARCH_7__) || \
+    defined(__ARM_ARCH_7R__) || \
+    defined(__ARM_ARCH_7A__)
+#define ARMV6_OR_7 1
+#endif
+
 extern "C" {
   size_t fread_unlocked(void *a, size_t b, size_t c, FILE *d);
   size_t fwrite_unlocked(const void *a, size_t b, size_t c, FILE *d);
@@ -61,28 +75,50 @@ class CondVar {
   pthread_cond_t cv_;
 };
 
+#ifndef ARMV6_OR_7
+// On ARM chipsets <V6, 0xffff0fa0 is the hard coded address of a 
+// memory barrier function provided by the kernel.
+typedef void (*LinuxKernelMemoryBarrierFunc)(void);
+LinuxKernelMemoryBarrierFunc pLinuxKernelMemoryBarrier ATTRIBUTE_WEAK =
+    (LinuxKernelMemoryBarrierFunc) 0xffff0fa0;
+#endif
+
 // Storage for a lock-free pointer
 class AtomicPointer {
  private:
-  std::atomic<void*> rep_;
+  void* rep_;
+
+  inline void MemoryBarrier() const {
+    // TODO(gabor): This only works on Android instruction sets >= V6
+#ifdef ARMV6_OR_7
+    __asm__ __volatile__("dmb" : : : "memory");
+#else
+    pLinuxKernelMemoryBarrier();
+#endif
+  }
+
  public:
   AtomicPointer() { }
   explicit AtomicPointer(void* v) : rep_(v) { }
   inline void* Acquire_Load() const {
-    return rep_.load(std::memory_order_acquire);
+    void* r = rep_;
+    MemoryBarrier();
+    return r;
   }
   inline void Release_Store(void* v) {
-    rep_.store(v, std::memory_order_release);
+    MemoryBarrier();
+    rep_ = v;
   }
   inline void* NoBarrier_Load() const {
-    return rep_.load(std::memory_order_relaxed);
+    void* r = rep_;
+    return r;
   }
   inline void NoBarrier_Store(void* v) {
-    rep_.store(v, std::memory_order_relaxed);
+    rep_ = v;
   }
 };
 
-// TODO(gabor): Implement actual compress
+// TODO(gabor): Implement compress
 inline bool Snappy_Compress(
     const char* input,
     size_t input_length,
@@ -90,7 +126,7 @@ inline bool Snappy_Compress(
   return false;
 }
 
-// TODO(gabor): Implement actual uncompress
+// TODO(gabor): Implement uncompress
 inline bool Snappy_Uncompress(
     const char* input_data,
     size_t input_length,

table/table_test.cc

@@ -725,10 +725,10 @@ TEST(Harness, RandomizedLongDB) {
   Test(&rnd);
 
   // We must have created enough data to force merging
-  uint64_t l0_files, l1_files;
+  std::string l0_files, l1_files;
   ASSERT_TRUE(db()->GetProperty("leveldb.num-files-at-level0", &l0_files));
   ASSERT_TRUE(db()->GetProperty("leveldb.num-files-at-level1", &l1_files));
-  ASSERT_GT(l0_files + l1_files, 0);
+  ASSERT_GT(atoi(l0_files.c_str()) + atoi(l1_files.c_str()), 0);
 
 }