Merge branch 'master' into columnfamilies

Conflicts:
	db/db_impl.cc

CONTRIBUTING.md

@@ -10,6 +10,10 @@ the CLA and we can cross-check with your GitHub username.
 
 Complete your CLA here: <https://developers.facebook.com/opensource/cla>
 
+If you don't have a Facebook account, we can send you a PDF that you can
+sign offline. Send us an e-mail or create a new github issue to
+request the CLA in PDF format.
+
 ## License
 
 By contributing to RocksDB, you agree that your contributions will be

INSTALL.md

@@ -71,7 +71,7 @@ libraries. You are on your own.
 `make clean; make` will compile librocksdb.a (RocskDB static library) and all
 the unit tests. You can run all unit tests with `make check`.
 
-For shared library builds, exec `make librocksdb.so` instead.
+For shared library builds, exec `make shared_lib` instead.
 
 If you followed the above steps and your compile or unit tests fail,
 please submit an issue: (https://github.com/facebook/rocksdb/issues)

Makefile

@@ -78,6 +78,7 @@ TESTS = \
 	redis_test \
 	reduce_levels_test \
 	plain_table_db_test \
+  prefix_test \
 	simple_table_db_test \
 	skiplist_test \
 	stringappend_test \
@@ -288,6 +289,9 @@ crc32c_test: util/crc32c_test.o $(LIBOBJECTS) $(TESTHARNESS)
 db_test: db/db_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(CXX) db/db_test.o $(LIBOBJECTS) $(TESTHARNESS) $(EXEC_LDFLAGS) -o $@ $(LDFLAGS) $(COVERAGEFLAGS)
 
+log_write_bench: util/log_write_bench.o $(LIBOBJECTS) $(TESTHARNESS)
+	$(CXX) util/log_write_bench.o $(LIBOBJECTS) $(TESTHARNESS) $(EXEC_LDFLAGS) -o $@ $(LDFLAGS) $(COVERAGEFLAGS) -pg
+
 plain_table_db_test: db/plain_table_db_test.o $(LIBOBJECTS) $(TESTHARNESS)
 	$(CXX) db/plain_table_db_test.o $(LIBOBJECTS) $(TESTHARNESS) $(EXEC_LDFLAGS) -o $@ $(LDFLAGS) $(COVERAGEFLAGS)
 

db/c.cc

@@ -17,11 +17,14 @@
 #include "rocksdb/env.h"
 #include "rocksdb/filter_policy.h"
 #include "rocksdb/iterator.h"
+#include "rocksdb/merge_operator.h"
 #include "rocksdb/options.h"
 #include "rocksdb/status.h"
 #include "rocksdb/write_batch.h"
 #include "rocksdb/memtablerep.h"
 #include "rocksdb/universal_compaction.h"
+#include "rocksdb/statistics.h"
+#include "rocksdb/slice_transform.h"
 
 using rocksdb::Cache;
 using rocksdb::Comparator;
@@ -30,8 +33,10 @@ using rocksdb::DB;
 using rocksdb::Env;
 using rocksdb::FileLock;
 using rocksdb::FilterPolicy;
+using rocksdb::FlushOptions;
 using rocksdb::Iterator;
 using rocksdb::Logger;
+using rocksdb::MergeOperator;
 using rocksdb::NewBloomFilterPolicy;
 using rocksdb::NewLRUCache;
 using rocksdb::Options;
@@ -40,6 +45,7 @@ using rocksdb::Range;
 using rocksdb::ReadOptions;
 using rocksdb::SequentialFile;
 using rocksdb::Slice;
+using rocksdb::SliceTransform;
 using rocksdb::Snapshot;
 using rocksdb::Status;
 using rocksdb::WritableFile;
@@ -54,6 +60,7 @@ struct rocksdb_t              { DB*               rep; };
 struct rocksdb_iterator_t        { Iterator*         rep; };
 struct rocksdb_writebatch_t      { WriteBatch        rep; };
 struct rocksdb_snapshot_t        { const Snapshot*   rep; };
+struct rocksdb_flushoptions_t    { FlushOptions      rep; };
 struct rocksdb_readoptions_t     { ReadOptions       rep; };
 struct rocksdb_writeoptions_t    { WriteOptions      rep; };
 struct rocksdb_options_t         { Options           rep; };
@@ -103,6 +110,9 @@ struct rocksdb_filterpolicy_t : public FilterPolicy {
       void*,
       const char* key, size_t length,
       const char* filter, size_t filter_length);
+  void (*delete_filter_)(
+      void*,
+      const char* filter, size_t filter_length);
 
   virtual ~rocksdb_filterpolicy_t() {
     (*destructor_)(state_);
@@ -122,8 +132,13 @@ struct rocksdb_filterpolicy_t : public FilterPolicy {
     size_t len;
     char* filter = (*create_)(state_, &key_pointers[0], &key_sizes[0], n, &len);
     dst->append(filter, len);
+
+    if (delete_filter_ != nullptr) {
+      (*delete_filter_)(state_, filter, len);
+    } else {
       free(filter);
     }
+  }
 
   virtual bool KeyMayMatch(const Slice& key, const Slice& filter) const {
     return (*key_match_)(state_, key.data(), key.size(),
@@ -131,16 +146,151 @@ struct rocksdb_filterpolicy_t : public FilterPolicy {
   }
 };
 
+struct rocksdb_mergeoperator_t : public MergeOperator {
+  void* state_;
+  void (*destructor_)(void*);
+  const char* (*name_)(void*);
+  char* (*full_merge_)(
+      void*,
+      const char* key, size_t key_length,
+      const char* existing_value, size_t existing_value_length,
+      const char* const* operands_list, const size_t* operands_list_length,
+      int num_operands,
+      unsigned char* success, size_t* new_value_length);
+  char* (*partial_merge_)(
+      void*,
+      const char* key, size_t key_length,
+      const char* left_operand, size_t left_operand_length,
+      const char* right_operand, size_t right_operand_length,
+      unsigned char* success, size_t* new_value_length);
+  void (*delete_value_)(
+      void*,
+      const char* value, size_t value_length);
+
+  virtual ~rocksdb_mergeoperator_t() {
+    (*destructor_)(state_);
+  }
+
+  virtual const char* Name() const {
+    return (*name_)(state_);
+  }
+
+  virtual bool FullMerge(
+      const Slice& key,
+      const Slice* existing_value,
+      const std::deque<std::string>& operand_list,
+      std::string* new_value,
+      Logger* logger) const {
+
+    size_t n = operand_list.size();
+    std::vector<const char*> operand_pointers(n);
+    std::vector<size_t> operand_sizes(n);
+    for (size_t i = 0; i < n; i++) {
+      Slice operand(operand_list[i]);
+      operand_pointers[i] = operand.data();
+      operand_sizes[i] = operand.size();
+    }
+
+    const char* existing_value_data = nullptr;
+    size_t existing_value_len = 0;
+    if (existing_value != nullptr) {
+      existing_value_data = existing_value->data();
+      existing_value_len = existing_value->size();
+    }
+
+    unsigned char success;
+    size_t new_value_len;
+    char* tmp_new_value = (*full_merge_)(
+        state_,
+        key.data(), key.size(),
+        existing_value_data, existing_value_len,
+        &operand_pointers[0], &operand_sizes[0], n,
+        &success, &new_value_len);
+    new_value->assign(tmp_new_value, new_value_len);
+
+    if (delete_value_ != nullptr) {
+      (*delete_value_)(state_, tmp_new_value, new_value_len);
+    } else {
+      free(tmp_new_value);
+    }
+
+    return success;
+  }
+
+  virtual bool PartialMerge(
+      const Slice& key,
+      const Slice& left_operand,
+      const Slice& right_operand,
+      std::string* new_value,
+      Logger* logger) const {
+
+    unsigned char success;
+    size_t new_value_len;
+    char* tmp_new_value = (*partial_merge_)(
+        state_,
+        key.data(), key.size(),
+        left_operand.data(), left_operand.size(),
+        right_operand.data(), right_operand.size(),
+        &success, &new_value_len);
+    new_value->assign(tmp_new_value, new_value_len);
+
+    if (delete_value_ != nullptr) {
+      (*delete_value_)(state_, tmp_new_value, new_value_len);
+    } else {
+      free(tmp_new_value);
+    }
+
+    return success;
+  }
+};
+
 struct rocksdb_env_t {
   Env* rep;
   bool is_default;
 };
 
+struct rocksdb_slicetransform_t : public SliceTransform {
+  void* state_;
+  void (*destructor_)(void*);
+  const char* (*name_)(void*);
+  char* (*transform_)(
+      void*,
+      const char* key, size_t length,
+      size_t* dst_length);
+  unsigned char (*in_domain_)(
+      void*,
+      const char* key, size_t length);
+  unsigned char (*in_range_)(
+      void*,
+      const char* key, size_t length);
+
+  virtual ~rocksdb_slicetransform_t() {
+    (*destructor_)(state_);
+  }
+
+  virtual const char* Name() const {
+    return (*name_)(state_);
+  }
+
+  virtual Slice Transform(const Slice& src) const {
+    size_t len;
+    char* dst = (*transform_)(state_, src.data(), src.size(), &len);
+    return Slice(dst, len);
+  }
+
+  virtual bool InDomain(const Slice& src) const {
+    return (*in_domain_)(state_, src.data(), src.size());
+  }
+
+  virtual bool InRange(const Slice& src) const {
+    return (*in_range_)(state_, src.data(), src.size());
+  }
+};
+
 struct rocksdb_universal_compaction_options_t {
   rocksdb::CompactionOptionsUniversal *rep;
 };
 
-
 static bool SaveError(char** errptr, const Status& s) {
   assert(errptr != NULL);
   if (s.ok()) {
@@ -197,6 +347,15 @@ void rocksdb_delete(
   SaveError(errptr, db->rep->Delete(options->rep, Slice(key, keylen)));
 }
 
+void rocksdb_merge(
+    rocksdb_t* db,
+    const rocksdb_writeoptions_t* options,
+    const char* key, size_t keylen,
+    const char* val, size_t vallen,
+    char** errptr) {
+  SaveError(errptr,
+            db->rep->Merge(options->rep, Slice(key, keylen), Slice(val, vallen)));
+}
 
 void rocksdb_write(
     rocksdb_t* db,
@@ -287,6 +446,26 @@ void rocksdb_compact_range(
       (limit_key ? (b = Slice(limit_key, limit_key_len), &b) : NULL));
 }
 
+void rocksdb_flush(
+    rocksdb_t* db,
+    const rocksdb_flushoptions_t* options,
+    char** errptr) {
+  SaveError(errptr, db->rep->Flush(options->rep));
+}
+
+void rocksdb_disable_file_deletions(
+    rocksdb_t* db,
+    char** errptr) {
+  SaveError(errptr, db->rep->DisableFileDeletions());
+}
+
+void rocksdb_enable_file_deletions(
+    rocksdb_t* db,
+    unsigned char force,
+    char** errptr) {
+  SaveError(errptr, db->rep->EnableFileDeletions(force));
+}
+
 void rocksdb_destroy_db(
     const rocksdb_options_t* options,
     const char* name,
@@ -365,6 +544,13 @@ void rocksdb_writebatch_put(
   b->rep.Put(Slice(key, klen), Slice(val, vlen));
 }
 
+void rocksdb_writebatch_merge(
+    rocksdb_writebatch_t* b,
+    const char* key, size_t klen,
+    const char* val, size_t vlen) {
+  b->rep.Merge(Slice(key, klen), Slice(val, vlen));
+}
+
 void rocksdb_writebatch_delete(
     rocksdb_writebatch_t* b,
     const char* key, size_t klen) {
@@ -409,6 +595,12 @@ void rocksdb_options_set_comparator(
   opt->rep.comparator = cmp;
 }
 
+void rocksdb_options_set_merge_operator(
+    rocksdb_options_t* opt,
+    rocksdb_mergeoperator_t* merge_operator) {
+  opt->rep.merge_operator = std::shared_ptr<MergeOperator>(merge_operator);
+}
+
 void rocksdb_options_set_filter_policy(
     rocksdb_options_t* opt,
     rocksdb_filterpolicy_t* policy) {
@@ -454,6 +646,12 @@ void rocksdb_options_set_cache(rocksdb_options_t* opt, rocksdb_cache_t* c) {
   }
 }
 
+void rocksdb_options_set_cache_compressed(rocksdb_options_t* opt, rocksdb_cache_t* c) {
+  if (c) {
+    opt->rep.block_cache_compressed = c->rep;
+  }
+}
+
 void rocksdb_options_set_block_size(rocksdb_options_t* opt, size_t s) {
   opt->rep.block_size = s;
 }
@@ -492,6 +690,10 @@ void rocksdb_options_set_max_grandparent_overlap_factor(
   opt->rep.max_grandparent_overlap_factor = n;
 }
 
+void rocksdb_options_enable_statistics(rocksdb_options_t* opt) {
+  opt->rep.statistics = rocksdb::CreateDBStatistics();
+}
+
 void rocksdb_options_set_num_levels(rocksdb_options_t* opt, int n) {
   opt->rep.num_levels = n;
 }
@@ -537,6 +739,16 @@ void rocksdb_options_set_compression_options(
   opt->rep.compression_opts.strategy = strategy;
 }
 
+void rocksdb_options_set_prefix_extractor(
+    rocksdb_options_t* opt, rocksdb_slicetransform_t* prefix_extractor) {
+  opt->rep.prefix_extractor = prefix_extractor;
+}
+
+void rocksdb_options_set_whole_key_filtering(
+    rocksdb_options_t* opt, unsigned char v) {
+  opt->rep.whole_key_filtering = v;
+}
+
 void rocksdb_options_set_disable_data_sync(
     rocksdb_options_t* opt, int disable_data_sync) {
   opt->rep.disableDataSync = disable_data_sync;
@@ -557,6 +769,11 @@ void rocksdb_options_set_db_log_dir(
   opt->rep.db_log_dir = db_log_dir;
 }
 
+void rocksdb_options_set_wal_dir(
+    rocksdb_options_t* opt, const char* v) {
+  opt->rep.wal_dir = v;
+}
+
 void rocksdb_options_set_WAL_ttl_seconds(rocksdb_options_t* opt, uint64_t ttl) {
   opt->rep.WAL_ttl_seconds = ttl;
 }
@@ -566,6 +783,76 @@ void rocksdb_options_set_WAL_size_limit_MB(
   opt->rep.WAL_size_limit_MB = limit;
 }
 
+void rocksdb_options_set_manifest_preallocation_size(
+    rocksdb_options_t* opt, size_t v) {
+  opt->rep.manifest_preallocation_size = v;
+}
+
+void rocksdb_options_set_purge_redundant_kvs_while_flush(
+    rocksdb_options_t* opt, unsigned char v) {
+  opt->rep.purge_redundant_kvs_while_flush = v;
+}
+
+void rocksdb_options_set_allow_os_buffer(
+    rocksdb_options_t* opt, unsigned char v) {
+  opt->rep.allow_os_buffer = v;
+}
+
+void rocksdb_options_set_allow_mmap_reads(
+    rocksdb_options_t* opt, unsigned char v) {
+  opt->rep.allow_mmap_reads = v;
+}
+
+void rocksdb_options_set_allow_mmap_writes(
+    rocksdb_options_t* opt, unsigned char v) {
+  opt->rep.allow_mmap_writes = v;
+}
+
+void rocksdb_options_set_is_fd_close_on_exec(
+    rocksdb_options_t* opt, unsigned char v) {
+  opt->rep.is_fd_close_on_exec = v;
+}
+
+void rocksdb_options_set_skip_log_error_on_recovery(
+    rocksdb_options_t* opt, unsigned char v) {
+  opt->rep.skip_log_error_on_recovery = v;
+}
+
+void rocksdb_options_set_stats_dump_period_sec(
+    rocksdb_options_t* opt, unsigned int v) {
+  opt->rep.stats_dump_period_sec = v;
+}
+
+void rocksdb_options_set_block_size_deviation(
+    rocksdb_options_t* opt, int v) {
+  opt->rep.block_size_deviation = v;
+}
+
+void rocksdb_options_set_advise_random_on_open(
+    rocksdb_options_t* opt, unsigned char v) {
+  opt->rep.advise_random_on_open = v;
+}
+
+void rocksdb_options_set_use_adaptive_mutex(
+    rocksdb_options_t* opt, unsigned char v) {
+  opt->rep.use_adaptive_mutex = v;
+}
+
+void rocksdb_options_set_bytes_per_sync(
+    rocksdb_options_t* opt, uint64_t v) {
+  opt->rep.bytes_per_sync = v;
+}
+
+void rocksdb_options_set_filter_deletes(
+    rocksdb_options_t* opt, unsigned char v) {
+  opt->rep.filter_deletes = v;
+}
+
+void rocksdb_options_set_max_sequential_skip_in_iterations(
+    rocksdb_options_t* opt, uint64_t v) {
+  opt->rep.max_sequential_skip_in_iterations = v;
+}
+
 void rocksdb_options_set_max_write_buffer_number(rocksdb_options_t* opt, int n) {
   opt->rep.max_write_buffer_number = n;
 }
@@ -582,6 +869,56 @@ void rocksdb_options_set_max_background_flushes(rocksdb_options_t* opt, int n) {
   opt->rep.max_background_flushes = n;
 }
 
+void rocksdb_options_set_max_log_file_size(rocksdb_options_t* opt, size_t v) {
+  opt->rep.max_log_file_size = v;
+}
+
+void rocksdb_options_set_log_file_time_to_roll(rocksdb_options_t* opt, size_t v) {
+  opt->rep.log_file_time_to_roll = v;
+}
+
+void rocksdb_options_set_keep_log_file_num(rocksdb_options_t* opt, size_t v) {
+  opt->rep.keep_log_file_num = v;
+}
+
+void rocksdb_options_set_soft_rate_limit(rocksdb_options_t* opt, double v) {
+  opt->rep.soft_rate_limit = v;
+}
+
+void rocksdb_options_set_hard_rate_limit(rocksdb_options_t* opt, double v) {
+  opt->rep.hard_rate_limit = v;
+}
+
+void rocksdb_options_set_rate_limit_delay_max_milliseconds(
+    rocksdb_options_t* opt, unsigned int v) {
+  opt->rep.rate_limit_delay_max_milliseconds = v;
+}
+
+void rocksdb_options_set_max_manifest_file_size(
+    rocksdb_options_t* opt, size_t v) {
+  opt->rep.max_manifest_file_size = v;
+}
+
+void rocksdb_options_set_no_block_cache(
+    rocksdb_options_t* opt, unsigned char v) {
+  opt->rep.no_block_cache = v;
+}
+
+void rocksdb_options_set_table_cache_numshardbits(
+    rocksdb_options_t* opt, int v) {
+  opt->rep.table_cache_numshardbits = v;
+}
+
+void rocksdb_options_set_table_cache_remove_scan_count_limit(
+    rocksdb_options_t* opt, int v) {
+  opt->rep.table_cache_remove_scan_count_limit = v;
+}
+
+void rocksdb_options_set_arena_block_size(
+    rocksdb_options_t* opt, size_t v) {
+  opt->rep.arena_block_size = v;
+}
+
 void rocksdb_options_set_disable_auto_compactions(rocksdb_options_t* opt, int disable) {
   opt->rep.disable_auto_compactions = disable;
 }
@@ -590,6 +927,11 @@ void rocksdb_options_set_disable_seek_compaction(rocksdb_options_t* opt, int dis
   opt->rep.disable_seek_compaction = disable;
 }
 
+void rocksdb_options_set_delete_obsolete_files_period_micros(
+    rocksdb_options_t* opt, uint64_t v) {
+  opt->rep.delete_obsolete_files_period_micros = v;
+}
+
 void rocksdb_options_set_source_compaction_factor(
     rocksdb_options_t* opt, int n) {
   opt->rep.expanded_compaction_factor = n;
@@ -607,6 +949,21 @@ void rocksdb_options_set_memtable_vector_rep(rocksdb_options_t *opt) {
   opt->rep.memtable_factory.reset(factory);
 }
 
+void rocksdb_options_set_memtable_prefix_bloom_bits(
+    rocksdb_options_t* opt, uint32_t v) {
+  opt->rep.memtable_prefix_bloom_bits = v;
+}
+
+void rocksdb_options_set_memtable_prefix_bloom_probes(
+    rocksdb_options_t* opt, uint32_t v) {
+  opt->rep.memtable_prefix_bloom_probes = v;
+}
+
+void rocksdb_options_set_max_successive_merges(
+    rocksdb_options_t* opt, size_t v) {
+  opt->rep.max_successive_merges = v;
+}
+
 void rocksdb_options_set_compaction_style(rocksdb_options_t *opt, int style) {
   opt->rep.compaction_style = static_cast<rocksdb::CompactionStyle>(style);
 }
@@ -617,38 +974,22 @@ void rocksdb_options_set_universal_compaction_options(rocksdb_options_t *opt, ro
 
 /*
 TODO:
-merge_operator
+DB::OpenForReadOnly
+DB::MultiGet
+DB::KeyMayExist
+DB::GetOptions
+DB::GetLiveFiles
+DB::GetSortedWalFiles
+DB::GetLatestSequenceNumber
+DB::GetUpdatesSince
+DB::DeleteFile
+DB::GetLiveFilesMetaData
+DB::GetDbIdentity
+DB::RunManualCompaction
+custom cache
 compaction_filter
-prefix_extractor
-whole_key_filtering
 max_bytes_for_level_multiplier_additional
-delete_obsolete_files_period_micros
-max_log_file_size
-log_file_time_to_roll
-keep_log_file_num
-soft_rate_limit
-hard_rate_limit
-rate_limit_delay_max_milliseconds
-max_manifest_file_size
-no_block_cache
-table_cache_numshardbits
-table_cache_remove_scan_count_limit
-arena_block_size
-manifest_preallocation_size
-purge_redundant_kvs_while_flush
-allow_os_buffer
-allow_mmap_reads
-allow_mmap_writes
-is_fd_close_on_exec
-skip_log_error_on_recovery
-stats_dump_period_sec
-block_size_deviation
-advise_random_on_open
 access_hint_on_compaction_start
-use_adaptive_mutex
-bytes_per_sync
-filter_deletes
-max_sequential_skip_in_iterations
 table_factory
 table_properties_collectors
 inplace_update_support
@@ -687,12 +1028,16 @@ rocksdb_filterpolicy_t* rocksdb_filterpolicy_create(
         void*,
         const char* key, size_t length,
         const char* filter, size_t filter_length),
+    void (*delete_filter)(
+        void*,
+        const char* filter, size_t filter_length),
     const char* (*name)(void*)) {
   rocksdb_filterpolicy_t* result = new rocksdb_filterpolicy_t;
   result->state_ = state;
   result->destructor_ = destructor;
   result->create_ = create_filter;
   result->key_match_ = key_may_match;
+  result->delete_filter_ = delete_filter;
   result->name_ = name;
   return result;
 }
@@ -720,10 +1065,45 @@ rocksdb_filterpolicy_t* rocksdb_filterpolicy_create_bloom(int bits_per_key) {
   Wrapper* wrapper = new Wrapper;
   wrapper->rep_ = NewBloomFilterPolicy(bits_per_key);
   wrapper->state_ = NULL;
+  wrapper->delete_filter_ = NULL;
   wrapper->destructor_ = &Wrapper::DoNothing;
   return wrapper;
 }
 
+rocksdb_mergeoperator_t* rocksdb_mergeoperator_create(
+    void* state,
+    void (*destructor)(void*),
+    char* (*full_merge)(
+        void*,
+        const char* key, size_t key_length,
+        const char* existing_value, size_t existing_value_length,
+        const char* const* operands_list, const size_t* operands_list_length,
+        int num_operands,
+        unsigned char* success, size_t* new_value_length),
+    char* (*partial_merge)(
+        void*,
+        const char* key, size_t key_length,
+        const char* left_operand, size_t left_operand_length,
+        const char* right_operand, size_t right_operand_length,
+        unsigned char* success, size_t* new_value_length),
+    void (*delete_value)(
+        void*,
+        const char* value, size_t value_length),
+    const char* (*name)(void*)) {
+  rocksdb_mergeoperator_t* result = new rocksdb_mergeoperator_t;
+  result->state_ = state;
+  result->destructor_ = destructor;
+  result->full_merge_ = full_merge;
+  result->partial_merge_ = partial_merge;
+  result->delete_value_ = delete_value;
+  result->name_ = name;
+  return result;
+}
+
+void rocksdb_mergeoperator_destroy(rocksdb_mergeoperator_t* merge_operator) {
+  delete merge_operator;
+}
+
 rocksdb_readoptions_t* rocksdb_readoptions_create() {
   return new rocksdb_readoptions_t;
 }
@@ -743,12 +1123,33 @@ void rocksdb_readoptions_set_fill_cache(
   opt->rep.fill_cache = v;
 }
 
+void rocksdb_readoptions_set_prefix_seek(
+    rocksdb_readoptions_t* opt, unsigned char v) {
+  opt->rep.prefix_seek = v;
+}
+
 void rocksdb_readoptions_set_snapshot(
     rocksdb_readoptions_t* opt,
     const rocksdb_snapshot_t* snap) {
   opt->rep.snapshot = (snap ? snap->rep : NULL);
 }
 
+void rocksdb_readoptions_set_prefix(
+    rocksdb_readoptions_t* opt, const char* key, size_t keylen) {
+  Slice prefix = Slice(key, keylen);
+  opt->rep.prefix = &prefix;
+}
+
+void rocksdb_readoptions_set_read_tier(
+    rocksdb_readoptions_t* opt, int v) {
+  opt->rep.read_tier = static_cast<rocksdb::ReadTier>(v);
+}
+
+void rocksdb_readoptions_set_tailing(
+    rocksdb_readoptions_t* opt, unsigned char v) {
+  opt->rep.tailing = v;
+}
+
 rocksdb_writeoptions_t* rocksdb_writeoptions_create() {
   return new rocksdb_writeoptions_t;
 }
@@ -767,6 +1168,19 @@ void rocksdb_writeoptions_disable_WAL(rocksdb_writeoptions_t* opt, int disable)
 }
 
 
+rocksdb_flushoptions_t* rocksdb_flushoptions_create() {
+  return new rocksdb_flushoptions_t;
+}
+
+void rocksdb_flushoptions_destroy(rocksdb_flushoptions_t* opt) {
+  delete opt;
+}
+
+void rocksdb_flushoptions_set_wait(
+    rocksdb_flushoptions_t* opt, unsigned char v) {
+  opt->rep.wait = v;
+}
+
 rocksdb_cache_t* rocksdb_cache_create_lru(size_t capacity) {
   rocksdb_cache_t* c = new rocksdb_cache_t;
   c->rep = NewLRUCache(capacity);
@@ -797,6 +1211,57 @@ void rocksdb_env_destroy(rocksdb_env_t* env) {
   delete env;
 }
 
+rocksdb_slicetransform_t* rocksdb_slicetransform_create(
+    void* state,
+    void (*destructor)(void*),
+    char* (*transform)(
+        void*,
+        const char* key, size_t length,
+        size_t* dst_length),
+    unsigned char (*in_domain)(
+        void*,
+        const char* key, size_t length),
+    unsigned char (*in_range)(
+        void*,
+        const char* key, size_t length),
+    const char* (*name)(void*)) {
+  rocksdb_slicetransform_t* result = new rocksdb_slicetransform_t;
+  result->state_ = state;
+  result->destructor_ = destructor;
+  result->transform_ = transform;
+  result->in_domain_ = in_domain;
+  result->in_range_ = in_range;
+  result->name_ = name;
+  return result;
+}
+
+void rocksdb_slicetransform_destroy(rocksdb_slicetransform_t* st) {
+  delete st;
+}
+
+rocksdb_slicetransform_t* rocksdb_slicetransform_create_fixed_prefix(size_t prefixLen) {
+  struct Wrapper : public rocksdb_slicetransform_t {
+    const SliceTransform* rep_;
+    ~Wrapper() { delete rep_; }
+    const char* Name() const { return rep_->Name(); }
+    Slice Transform(const Slice& src) const {
+      return rep_->Transform(src);
+    }
+    bool InDomain(const Slice& src) const {
+      return rep_->InDomain(src);
+    }
+    bool InRange(const Slice& src) const {
+      return rep_->InRange(src);
+    }
+    static void DoNothing(void*) { }
+  };
+  Wrapper* wrapper = new Wrapper;
+  wrapper->rep_ = rocksdb::NewFixedPrefixTransform(prefixLen);
+  wrapper->state_ = NULL;
+  wrapper->destructor_ = &Wrapper::DoNothing;
+  return wrapper;
+}
+
 rocksdb_universal_compaction_options_t* rocksdb_universal_compaction_options_create() {
   rocksdb_universal_compaction_options_t* result = new rocksdb_universal_compaction_options_t;
   result->rep = new rocksdb::CompactionOptionsUniversal;

db/c_test.c

@@ -154,6 +154,37 @@ unsigned char FilterKeyMatch(
   return fake_filter_result;
 }
 
+// Custom merge operator
+static void MergeOperatorDestroy(void* arg) { }
+static const char* MergeOperatorName(void* arg) {
+  return "TestMergeOperator";
+}
+static char* MergeOperatorFullMerge(
+    void* arg,
+    const char* key, size_t key_length,
+    const char* existing_value, size_t existing_value_length,
+    const char* const* operands_list, const size_t* operands_list_length,
+    int num_operands,
+    unsigned char* success, size_t* new_value_length) {
+  *new_value_length = 4;
+  *success = 1;
+  char* result = malloc(4);
+  memcpy(result, "fake", 4);
+  return result;
+}
+static char* MergeOperatorPartialMerge(
+    void* arg,
+    const char* key, size_t key_length,
+    const char* left_operand, size_t left_operand_length,
+    const char* right_operand, size_t right_operand_length,
+    unsigned char* success, size_t* new_value_length) {
+  *new_value_length = 4;
+  *success = 1;
+  char* result = malloc(4);
+  memcpy(result, "fake", 4);
+  return result;
+}
+
 int main(int argc, char** argv) {
   rocksdb_t* db;
   rocksdb_comparator_t* cmp;
@@ -342,7 +373,7 @@ int main(int argc, char** argv) {
     rocksdb_filterpolicy_t* policy;
     if (run == 0) {
       policy = rocksdb_filterpolicy_create(
-          NULL, FilterDestroy, FilterCreate, FilterKeyMatch, FilterName);
+          NULL, FilterDestroy, FilterCreate, FilterKeyMatch, NULL, FilterName);
     } else {
       policy = rocksdb_filterpolicy_create_bloom(10);
     }
@@ -376,6 +407,32 @@ int main(int argc, char** argv) {
     rocksdb_filterpolicy_destroy(policy);
   }
 
+  StartPhase("merge_operator");
+  {
+    rocksdb_mergeoperator_t* merge_operator;
+    merge_operator = rocksdb_mergeoperator_create(
+        NULL, MergeOperatorDestroy, MergeOperatorFullMerge,
+        MergeOperatorPartialMerge, NULL, MergeOperatorName);
+    // Create new database
+    rocksdb_close(db);
+    rocksdb_destroy_db(options, dbname, &err);
+    rocksdb_options_set_merge_operator(options, merge_operator);
+    db = rocksdb_open(options, dbname, &err);
+    CheckNoError(err);
+    rocksdb_put(db, woptions, "foo", 3, "foovalue", 8, &err);
+    CheckNoError(err);
+    CheckGet(db, roptions, "foo", "foovalue");
+    rocksdb_merge(db, woptions, "foo", 3, "barvalue", 8, &err);
+    CheckNoError(err);
+    CheckGet(db, roptions, "foo", "fake");
+
+    // Merge of a non-existing value
+    rocksdb_merge(db, woptions, "bar", 3, "barvalue", 8, &err);
+    CheckNoError(err);
+    CheckGet(db, roptions, "bar", "fake");
+
+  }
+
   StartPhase("cleanup");
   rocksdb_close(db);
   rocksdb_options_destroy(options);

db/db_impl.cc

@@ -3749,6 +3749,12 @@ Status DBImpl::DeleteFile(std::string name) {
   LogFlush(options_.info_log);
   // remove files outside the db-lock
   PurgeObsoleteFiles(deletion_state);
+  {
+    MutexLock l(&mutex_);
+    // schedule flush if file deletion means we freed the space for flushes to
+    // continue
+    MaybeScheduleFlushOrCompaction();
+  }
   return status;
 }
 
@@ -3864,7 +3870,7 @@ Status DB::Open(const DBOptions& db_options, const std::string& dbname,
                 std::vector<ColumnFamilyHandle*>* handles, DB** dbptr) {
   *dbptr = nullptr;
   handles->clear();
-  EnvOptions soptions;
+  EnvOptions soptions(db_options);
 
   size_t max_write_buffer_size = 0;
   for (auto cf : column_families) {

db/plain_table_db_test.cc

@@ -220,7 +220,6 @@ class TestPlainTableFactory : public PlainTableFactory {
                                  size_t index_sparseness = 16)
       : PlainTableFactory(user_key_len, user_key_len, hash_table_ratio,
                           hash_table_ratio),
-        user_key_len_(user_key_len),
         bloom_bits_per_key_(bloom_bits_per_key),
         hash_table_ratio_(hash_table_ratio),
         index_sparseness_(index_sparseness),
@@ -245,7 +244,6 @@ class TestPlainTableFactory : public PlainTableFactory {
   }
 
  private:
-  uint32_t user_key_len_;
   int bloom_bits_per_key_;
   double hash_table_ratio_;
   size_t index_sparseness_;

db/prefix_test.cc

@@ -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.
+
 #include <algorithm>
 #include <iostream>
 #include <vector>
@@ -6,6 +11,8 @@
 #include "rocksdb/comparator.h"
 #include "rocksdb/db.h"
 #include "rocksdb/perf_context.h"
+#include "rocksdb/slice_transform.h"
+#include "rocksdb/memtablerep.h"
 #include "util/histogram.h"
 #include "util/stop_watch.h"
 #include "util/testharness.h"
@@ -97,6 +104,36 @@ class TestKeyComparator : public Comparator {
 
 };
 
+void PutKey(DB* db, WriteOptions write_options, uint64_t prefix,
+            uint64_t suffix, const Slice& value) {
+  TestKey test_key(prefix, suffix);
+  Slice key = TestKeyToSlice(test_key);
+  ASSERT_OK(db->Put(write_options, key, value));
+}
+
+void SeekIterator(Iterator* iter, uint64_t prefix, uint64_t suffix) {
+  TestKey test_key(prefix, suffix);
+  Slice key = TestKeyToSlice(test_key);
+  iter->Seek(key);
+}
+
+const std::string kNotFoundResult = "NOT_FOUND";
+
+std::string Get(DB* db, const ReadOptions& read_options, uint64_t prefix,
+                uint64_t suffix) {
+  TestKey test_key(prefix, suffix);
+  Slice key = TestKeyToSlice(test_key);
+
+  std::string result;
+  Status s = db->Get(read_options, key, &result);
+  if (s.IsNotFound()) {
+    result = kNotFoundResult;
+  } else if (!s.ok()) {
+    result = s.ToString();
+  }
+  return result;
+}
+
 class PrefixTest {
  public:
   std::shared_ptr<DB> OpenDb() {
@@ -116,7 +153,11 @@ class PrefixTest {
     return std::shared_ptr<DB>(db);
   }
 
-  bool NextOptions() {
+  void FirstOption() {
+    option_config_ = kBegin;
+  }
+
+  bool NextOptions(int bucket_count) {
     // skip some options
     option_config_++;
     if (option_config_ < kEnd) {
@@ -124,15 +165,12 @@ class PrefixTest {
       options.prefix_extractor = prefix_extractor;
       switch(option_config_) {
         case kHashSkipList:
-          options.memtable_factory.reset(
+          options.memtable_factory.reset(NewHashSkipListRepFactory(
-              NewHashSkipListRepFactory(options.prefix_extractor,
+              options.prefix_extractor, bucket_count, FLAGS_skiplist_height));
-                                        FLAGS_bucket_count,
-                                        FLAGS_skiplist_height));
           return true;
         case kHashLinkList:
-          options.memtable_factory.reset(
+          options.memtable_factory.reset(NewHashLinkListRepFactory(
-              NewHashLinkListRepFactory(options.prefix_extractor,
+              options.prefix_extractor, bucket_count));
-                                        FLAGS_bucket_count));
           return true;
         default:
           return false;
@@ -158,8 +196,182 @@ class PrefixTest {
   Options options;
 };
 
+TEST(PrefixTest, TestResult) {
+  for (int num_buckets = 1; num_buckets <= 2; num_buckets++) {
+    FirstOption();
+    while (NextOptions(num_buckets)) {
+      std::cout << "*** Mem table: " << options.memtable_factory->Name()
+                << " number of buckets: " << num_buckets
+                << std::endl;
+      DestroyDB(kDbName, Options());
+      auto db = OpenDb();
+      WriteOptions write_options;
+      ReadOptions read_options;
+      read_options.prefix_seek = true;
+
+      // 1. Insert one row.
+      Slice v16("v16");
+      PutKey(db.get(), write_options, 1, 6, v16);
+      std::unique_ptr<Iterator> iter(db->NewIterator(read_options));
+      SeekIterator(iter.get(), 1, 6);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v16 == iter->value());
+      SeekIterator(iter.get(), 1, 5);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v16 == iter->value());
+      SeekIterator(iter.get(), 1, 5);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v16 == iter->value());
+      iter->Next();
+      ASSERT_TRUE(!iter->Valid());
+
+      SeekIterator(iter.get(), 2, 0);
+      ASSERT_TRUE(!iter->Valid());
+
+      ASSERT_EQ(v16.ToString(), Get(db.get(), read_options, 1, 6));
+      ASSERT_EQ(kNotFoundResult, Get(db.get(), read_options, 1, 5));
+      ASSERT_EQ(kNotFoundResult, Get(db.get(), read_options, 1, 7));
+      ASSERT_EQ(kNotFoundResult, Get(db.get(), read_options, 0, 6));
+      ASSERT_EQ(kNotFoundResult, Get(db.get(), read_options, 2, 6));
+
+      // 2. Insert an entry for the same prefix as the last entry in the bucket.
+      Slice v17("v17");
+      PutKey(db.get(), write_options, 1, 7, v17);
+      iter.reset(db->NewIterator(read_options));
+      SeekIterator(iter.get(), 1, 7);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v17 == iter->value());
+
+      SeekIterator(iter.get(), 1, 6);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v16 == iter->value());
+      iter->Next();
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v17 == iter->value());
+      iter->Next();
+      ASSERT_TRUE(!iter->Valid());
+
+      SeekIterator(iter.get(), 2, 0);
+      ASSERT_TRUE(!iter->Valid());
+
+      // 3. Insert an entry for the same prefix as the head of the bucket.
+      Slice v15("v15");
+      PutKey(db.get(), write_options, 1, 5, v15);
+      iter.reset(db->NewIterator(read_options));
+
+      SeekIterator(iter.get(), 1, 7);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v17 == iter->value());
+
+      SeekIterator(iter.get(), 1, 5);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v15 == iter->value());
+      iter->Next();
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v16 == iter->value());
+      iter->Next();
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v17 == iter->value());
+
+      SeekIterator(iter.get(), 1, 5);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v15 == iter->value());
+
+      ASSERT_EQ(v15.ToString(), Get(db.get(), read_options, 1, 5));
+      ASSERT_EQ(v16.ToString(), Get(db.get(), read_options, 1, 6));
+      ASSERT_EQ(v17.ToString(), Get(db.get(), read_options, 1, 7));
+
+      // 4. Insert an entry with a larger prefix
+      Slice v22("v22");
+      PutKey(db.get(), write_options, 2, 2, v22);
+      iter.reset(db->NewIterator(read_options));
+
+      SeekIterator(iter.get(), 2, 2);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v22 == iter->value());
+      SeekIterator(iter.get(), 2, 0);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v22 == iter->value());
+
+      SeekIterator(iter.get(), 1, 5);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v15 == iter->value());
+
+      SeekIterator(iter.get(), 1, 7);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v17 == iter->value());
+
+      // 5. Insert an entry with a smaller prefix
+      Slice v02("v02");
+      PutKey(db.get(), write_options, 0, 2, v02);
+      iter.reset(db->NewIterator(read_options));
+
+      SeekIterator(iter.get(), 0, 2);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v02 == iter->value());
+      SeekIterator(iter.get(), 0, 0);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v02 == iter->value());
+
+      SeekIterator(iter.get(), 2, 0);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v22 == iter->value());
+
+      SeekIterator(iter.get(), 1, 5);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v15 == iter->value());
+
+      SeekIterator(iter.get(), 1, 7);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v17 == iter->value());
+
+      // 6. Insert to the beginning and the end of the first prefix
+      Slice v13("v13");
+      Slice v18("v18");
+      PutKey(db.get(), write_options, 1, 3, v13);
+      PutKey(db.get(), write_options, 1, 8, v18);
+      iter.reset(db->NewIterator(read_options));
+      SeekIterator(iter.get(), 1, 7);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v17 == iter->value());
+
+      SeekIterator(iter.get(), 1, 3);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v13 == iter->value());
+      iter->Next();
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v15 == iter->value());
+      iter->Next();
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v16 == iter->value());
+      iter->Next();
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v17 == iter->value());
+      iter->Next();
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v18 == iter->value());
+
+      SeekIterator(iter.get(), 0, 0);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v02 == iter->value());
+
+      SeekIterator(iter.get(), 2, 0);
+      ASSERT_TRUE(iter->Valid());
+      ASSERT_TRUE(v22 == iter->value());
+
+      ASSERT_EQ(v22.ToString(), Get(db.get(), read_options, 2, 2));
+      ASSERT_EQ(v02.ToString(), Get(db.get(), read_options, 0, 2));
+      ASSERT_EQ(v13.ToString(), Get(db.get(), read_options, 1, 3));
+      ASSERT_EQ(v15.ToString(), Get(db.get(), read_options, 1, 5));
+      ASSERT_EQ(v16.ToString(), Get(db.get(), read_options, 1, 6));
+      ASSERT_EQ(v17.ToString(), Get(db.get(), read_options, 1, 7));
+      ASSERT_EQ(v18.ToString(), Get(db.get(), read_options, 1, 8));
+    }
+  }
+}
+
 TEST(PrefixTest, DynamicPrefixIterator) {
-  while (NextOptions()) {
+  while (NextOptions(FLAGS_bucket_count)) {
     std::cout << "*** Mem table: " << options.memtable_factory->Name()
         << std::endl;
     DestroyDB(kDbName, Options());
@@ -260,7 +472,7 @@ TEST(PrefixTest, DynamicPrefixIterator) {
 }
 
 TEST(PrefixTest, PrefixHash) {
-  while (NextOptions()) {
+  while (NextOptions(FLAGS_bucket_count)) {
     std::cout << "*** Mem table: " << options.memtable_factory->Name()
         << std::endl;
     DestroyDB(kDbName, Options());

include/rocksdb/c.h

@@ -60,12 +60,15 @@ typedef struct rocksdb_comparator_t    rocksdb_comparator_t;
 typedef struct rocksdb_env_t             rocksdb_env_t;
 typedef struct rocksdb_filelock_t        rocksdb_filelock_t;
 typedef struct rocksdb_filterpolicy_t    rocksdb_filterpolicy_t;
+typedef struct rocksdb_flushoptions_t    rocksdb_flushoptions_t;
 typedef struct rocksdb_iterator_t        rocksdb_iterator_t;
 typedef struct rocksdb_logger_t          rocksdb_logger_t;
+typedef struct rocksdb_mergeoperator_t   rocksdb_mergeoperator_t;
 typedef struct rocksdb_options_t         rocksdb_options_t;
 typedef struct rocksdb_randomfile_t      rocksdb_randomfile_t;
 typedef struct rocksdb_readoptions_t     rocksdb_readoptions_t;
 typedef struct rocksdb_seqfile_t         rocksdb_seqfile_t;
+typedef struct rocksdb_slicetransform_t  rocksdb_slicetransform_t;
 typedef struct rocksdb_snapshot_t        rocksdb_snapshot_t;
 typedef struct rocksdb_writablefile_t    rocksdb_writablefile_t;
 typedef struct rocksdb_writebatch_t      rocksdb_writebatch_t;
@@ -94,6 +97,13 @@ extern void rocksdb_delete(
     const char* key, size_t keylen,
     char** errptr);
 
+extern void rocksdb_merge(
+    rocksdb_t* db,
+    const rocksdb_writeoptions_t* options,
+    const char* key, size_t keylen,
+    const char* val, size_t vallen,
+    char** errptr);
+
 extern void rocksdb_write(
     rocksdb_t* db,
     const rocksdb_writeoptions_t* options,
@@ -138,6 +148,20 @@ extern void rocksdb_compact_range(
     const char* start_key, size_t start_key_len,
     const char* limit_key, size_t limit_key_len);
 
+extern void rocksdb_flush(
+    rocksdb_t* db,
+    const rocksdb_flushoptions_t* options,
+    char** errptr);
+
+extern void rocksdb_disable_file_deletions(
+    rocksdb_t* db,
+    char** errptr);
+
+extern void rocksdb_enable_file_deletions(
+    rocksdb_t* db,
+    unsigned char force,
+    char** errptr);
+
 /* Management operations */
 
 extern void rocksdb_destroy_db(
@@ -172,6 +196,10 @@ extern void rocksdb_writebatch_put(
     rocksdb_writebatch_t*,
     const char* key, size_t klen,
     const char* val, size_t vlen);
+extern void rocksdb_writebatch_merge(
+    rocksdb_writebatch_t*,
+    const char* key, size_t klen,
+    const char* val, size_t vlen);
 extern void rocksdb_writebatch_delete(
     rocksdb_writebatch_t*,
     const char* key, size_t klen);
@@ -188,6 +216,8 @@ extern void rocksdb_options_destroy(rocksdb_options_t*);
 extern void rocksdb_options_set_comparator(
     rocksdb_options_t*,
     rocksdb_comparator_t*);
+extern void rocksdb_options_set_merge_operator(rocksdb_options_t*,
+                                               rocksdb_mergeoperator_t*);
 extern void rocksdb_options_set_compression_per_level(
   rocksdb_options_t* opt,
   int* level_values,
@@ -206,10 +236,14 @@ extern void rocksdb_options_set_info_log(rocksdb_options_t*, rocksdb_logger_t*);
 extern void rocksdb_options_set_write_buffer_size(rocksdb_options_t*, size_t);
 extern void rocksdb_options_set_max_open_files(rocksdb_options_t*, int);
 extern void rocksdb_options_set_cache(rocksdb_options_t*, rocksdb_cache_t*);
+extern void rocksdb_options_set_cache_compressed(rocksdb_options_t*, rocksdb_cache_t*);
 extern void rocksdb_options_set_block_size(rocksdb_options_t*, size_t);
 extern void rocksdb_options_set_block_restart_interval(rocksdb_options_t*, int);
 extern void rocksdb_options_set_compression_options(
     rocksdb_options_t*, int, int, int);
+extern void rocksdb_options_set_whole_key_filtering(rocksdb_options_t*, unsigned char);
+extern void rocksdb_options_set_prefix_extractor(
+    rocksdb_options_t*, rocksdb_slicetransform_t*);
 extern void rocksdb_options_set_num_levels(rocksdb_options_t*, int);
 extern void rocksdb_options_set_level0_file_num_compaction_trigger(
     rocksdb_options_t*, int);
@@ -217,23 +251,97 @@ extern void rocksdb_options_set_level0_slowdown_writes_trigger(
     rocksdb_options_t*, int);
 extern void rocksdb_options_set_level0_stop_writes_trigger(
     rocksdb_options_t*, int);
+extern void rocksdb_options_set_max_mem_compaction_level(
+    rocksdb_options_t*, int);
 extern void rocksdb_options_set_target_file_size_base(
     rocksdb_options_t*, uint64_t);
 extern void rocksdb_options_set_target_file_size_multiplier(
     rocksdb_options_t*, int);
+extern void rocksdb_options_set_max_bytes_for_level_base(
+    rocksdb_options_t*, uint64_t);
+extern void rocksdb_options_set_max_bytes_for_level_multiplier(
+    rocksdb_options_t*, int);
+extern void rocksdb_options_set_expanded_compaction_factor(
+    rocksdb_options_t*, int);
+extern void rocksdb_options_set_max_grandparent_overlap_factor(
+    rocksdb_options_t*, int);
+extern void rocksdb_options_enable_statistics(rocksdb_options_t*);
+
 extern void rocksdb_options_set_max_write_buffer_number(rocksdb_options_t*, int);
 extern void rocksdb_options_set_min_write_buffer_number_to_merge(rocksdb_options_t*, int);
 extern void rocksdb_options_set_max_background_compactions(rocksdb_options_t*, int);
 extern void rocksdb_options_set_max_background_flushes(rocksdb_options_t*, int);
+extern void rocksdb_options_set_max_log_file_size(rocksdb_options_t*, size_t);
+extern void rocksdb_options_set_log_file_time_to_roll(rocksdb_options_t*, size_t);
+extern void rocksdb_options_set_keep_log_file_num(rocksdb_options_t*, size_t);
+extern void rocksdb_options_set_soft_rate_limit(rocksdb_options_t*, double);
+extern void rocksdb_options_set_hard_rate_limit(rocksdb_options_t*, double);
+extern void rocksdb_options_set_rate_limit_delay_max_milliseconds(
+    rocksdb_options_t*, unsigned int);
+extern void rocksdb_options_set_max_manifest_file_size(
+    rocksdb_options_t*, size_t);
+extern void rocksdb_options_set_no_block_cache(
+    rocksdb_options_t*, unsigned char);
+extern void rocksdb_options_set_table_cache_numshardbits(
+    rocksdb_options_t*, int);
+extern void rocksdb_options_set_table_cache_remove_scan_count_limit(
+    rocksdb_options_t*, int);
+extern void rocksdb_options_set_arena_block_size(
+    rocksdb_options_t*, size_t);
 extern void rocksdb_options_set_use_fsync(
     rocksdb_options_t*, int);
+extern void rocksdb_options_set_db_stats_log_interval(
+    rocksdb_options_t*, int);
+extern void rocksdb_options_set_db_log_dir(
+    rocksdb_options_t*, const char*);
+extern void rocksdb_options_set_wal_dir(
+    rocksdb_options_t*, const char*);
+extern void rocksdb_options_set_WAL_ttl_seconds(
+    rocksdb_options_t*, uint64_t);
+extern void rocksdb_options_set_WAL_size_limit_MB(
+    rocksdb_options_t*, uint64_t);
+extern void rocksdb_options_set_manifest_preallocation_size(
+    rocksdb_options_t*, size_t);
+extern void rocksdb_options_set_purge_redundant_kvs_while_flush(
+    rocksdb_options_t*, unsigned char);
+extern void rocksdb_options_set_allow_os_buffer(
+    rocksdb_options_t*, unsigned char);
+extern void rocksdb_options_set_allow_mmap_reads(
+    rocksdb_options_t*, unsigned char);
+extern void rocksdb_options_set_allow_mmap_writes(
+    rocksdb_options_t*, unsigned char);
+extern void rocksdb_options_set_is_fd_close_on_exec(
+    rocksdb_options_t*, unsigned char);
+extern void rocksdb_options_set_skip_log_error_on_recovery(
+    rocksdb_options_t*, unsigned char);
+extern void rocksdb_options_set_stats_dump_period_sec(
+    rocksdb_options_t*, unsigned int);
+extern void rocksdb_options_set_block_size_deviation(
+    rocksdb_options_t*, int);
+extern void rocksdb_options_set_advise_random_on_open(
+    rocksdb_options_t*, unsigned char);
+extern void rocksdb_options_set_use_adaptive_mutex(
+    rocksdb_options_t*, unsigned char);
+extern void rocksdb_options_set_bytes_per_sync(
+    rocksdb_options_t*, uint64_t);
+extern void rocksdb_options_set_filter_deletes(
+    rocksdb_options_t*, unsigned char);
+extern void rocksdb_options_set_max_sequential_skip_in_iterations(
+    rocksdb_options_t*, uint64_t);
 extern void rocksdb_options_set_disable_data_sync(rocksdb_options_t*, int);
 extern void rocksdb_options_set_disable_auto_compactions(rocksdb_options_t*, int);
 extern void rocksdb_options_set_disable_seek_compaction(rocksdb_options_t*, int);
+extern void rocksdb_options_set_delete_obsolete_files_period_micros(
+    rocksdb_options_t*, uint64_t);
 extern void rocksdb_options_set_source_compaction_factor(rocksdb_options_t*, int);
 extern void rocksdb_options_prepare_for_bulk_load(rocksdb_options_t*);
 extern void rocksdb_options_set_memtable_vector_rep(rocksdb_options_t*);
-
+extern void rocksdb_options_set_memtable_prefix_bloom_bits(
+    rocksdb_options_t*, uint32_t);
+extern void rocksdb_options_set_memtable_prefix_bloom_probes(
+    rocksdb_options_t*, uint32_t);
+extern void rocksdb_options_set_max_successive_merges(
+    rocksdb_options_t*, size_t);
 
 enum {
   rocksdb_no_compression = 0,
@@ -277,12 +385,39 @@ extern rocksdb_filterpolicy_t* rocksdb_filterpolicy_create(
         void*,
         const char* key, size_t length,
         const char* filter, size_t filter_length),
+    void (*delete_filter)(
+        void*,
+        const char* filter, size_t filter_length),
     const char* (*name)(void*));
 extern void rocksdb_filterpolicy_destroy(rocksdb_filterpolicy_t*);
 
 extern rocksdb_filterpolicy_t* rocksdb_filterpolicy_create_bloom(
     int bits_per_key);
 
+/* Merge Operator */
+
+extern rocksdb_mergeoperator_t* rocksdb_mergeoperator_create(
+    void* state,
+    void (*destructor)(void*),
+    char* (*full_merge)(
+        void*,
+        const char* key, size_t key_length,
+        const char* existing_value, size_t existing_value_length,
+        const char* const* operands_list, const size_t* operands_list_length,
+        int num_operands,
+        unsigned char* success, size_t* new_value_length),
+    char* (*partial_merge)(
+        void*,
+        const char* key, size_t key_length,
+        const char* left_operand, size_t left_operand_length,
+        const char* right_operand, size_t right_operand_length,
+        unsigned char* success, size_t* new_value_length),
+    void (*delete_value)(
+        void*,
+        const char* value, size_t value_length),
+    const char* (*name)(void*));
+extern void rocksdb_mergeoperator_destroy(rocksdb_mergeoperator_t*);
+
 /* Read options */
 
 extern rocksdb_readoptions_t* rocksdb_readoptions_create();
@@ -292,9 +427,17 @@ extern void rocksdb_readoptions_set_verify_checksums(
     unsigned char);
 extern void rocksdb_readoptions_set_fill_cache(
     rocksdb_readoptions_t*, unsigned char);
+extern void rocksdb_readoptions_set_prefix_seek(
+    rocksdb_readoptions_t*, unsigned char);
 extern void rocksdb_readoptions_set_snapshot(
     rocksdb_readoptions_t*,
     const rocksdb_snapshot_t*);
+extern void rocksdb_readoptions_set_prefix(
+    rocksdb_readoptions_t*, const char* key, size_t keylen);
+extern void rocksdb_readoptions_set_read_tier(
+    rocksdb_readoptions_t*, int);
+extern void rocksdb_readoptions_set_tailing(
+    rocksdb_readoptions_t*, unsigned char);
 
 /* Write options */
 
@@ -304,6 +447,13 @@ extern void rocksdb_writeoptions_set_sync(
     rocksdb_writeoptions_t*, unsigned char);
 extern void rocksdb_writeoptions_disable_WAL(rocksdb_writeoptions_t* opt, int disable);
 
+/* Flush options */
+
+extern rocksdb_flushoptions_t* rocksdb_flushoptions_create();
+extern void rocksdb_flushoptions_destroy(rocksdb_flushoptions_t*);
+extern void rocksdb_flushoptions_set_wait(
+    rocksdb_flushoptions_t*, unsigned char);
+
 /* Cache */
 
 extern rocksdb_cache_t* rocksdb_cache_create_lru(size_t capacity);
@@ -316,6 +466,25 @@ extern void rocksdb_env_set_background_threads(rocksdb_env_t* env, int n);
 extern void rocksdb_env_set_high_priority_background_threads(rocksdb_env_t* env, int n);
 extern void rocksdb_env_destroy(rocksdb_env_t*);
 
+/* SliceTransform */
+
+extern rocksdb_slicetransform_t* rocksdb_slicetransform_create(
+    void* state,
+    void (*destructor)(void*),
+    char* (*transform)(
+        void*,
+        const char* key, size_t length,
+        size_t* dst_length),
+    unsigned char (*in_domain)(
+        void*,
+        const char* key, size_t length),
+    unsigned char (*in_range)(
+        void*,
+        const char* key, size_t length),
+    const char* (*name)(void*));
+extern rocksdb_slicetransform_t* rocksdb_slicetransform_create_fixed_prefix(size_t);
+extern void rocksdb_slicetransform_destroy(rocksdb_slicetransform_t*);
+
 /* Universal Compaction options */
 
 enum {

linters/cpp_linter/ArcanistCpplintLinter.php

@@ -19,32 +19,8 @@ final class ArcanistCpplintLinter extends ArcanistLinter {
     return 'cpplint.py';
   }
 
-  public function getLintOptions() {
-    $config = $this->getEngine()->getConfigurationManager();
-    $options = $config->getConfigFromAnySource('lint.cpplint.options', '');
-
-    return $options;
-  }
-
   public function getLintPath() {
-    $config = $this->getEngine()->getConfigurationManager();
+    $bin = 'cpplint.py';
-    $prefix = $config->getConfigFromAnySource('lint.cpplint.prefix');
-    $bin = $config->getConfigFromAnySource('lint.cpplint.bin', 'cpplint.py');
-
-    if ($prefix !== null) {
-      if (!Filesystem::pathExists($prefix.'/'.$bin)) {
-        throw new ArcanistUsageException(
-          "Unable to find cpplint.py binary in a specified directory. Make ".
-          "sure that 'lint.cpplint.prefix' and 'lint.cpplint.bin' keys are ".
-          "set correctly. If you'd rather use a copy of cpplint installed ".
-          "globally, you can just remove these keys from your .arcconfig.");
-      }
-
-      $bin = csprintf("%s/%s", $prefix, $bin);
-
-      return $bin;
-    }
-
     // Search under current dir
     list($err) = exec_manual('which %s/%s', $this->linterDir(), $bin);
     if (!$err) {
@@ -57,7 +33,7 @@ final class ArcanistCpplintLinter extends ArcanistLinter {
       throw new ArcanistUsageException(
         "cpplint.py does not appear to be installed on this system. Install ".
         "it (e.g., with 'wget \"http://google-styleguide.googlecode.com/".
-        "svn/trunk/cpplint/cpplint.py\"') or configure 'lint.cpplint.prefix' ".
+        "svn/trunk/cpplint/cpplint.py\"') ".
         "in your .arcconfig to point to the directory where it resides. ".
         "Also don't forget to chmod a+x cpplint.py!");
     }
@@ -67,10 +43,9 @@ final class ArcanistCpplintLinter extends ArcanistLinter {
 
   public function lintPath($path) {
     $bin = $this->getLintPath();
-    $options = $this->getLintOptions();
     $path = $this->rocksdbDir().'/'.$path;
 
-    $f = new ExecFuture("%C %C $path", $bin, $options);
+    $f = new ExecFuture("%C $path", $bin);
 
     list($err, $stdout, $stderr) = $f->resolve();
 

linters/cpp_linter/cpplint.py

@@ -1420,6 +1420,9 @@ def CheckForHeaderGuard(filename, lines, error):
   endif = None
   endif_linenum = 0
   for linenum, line in enumerate(lines):
+    # Already been well guarded, no need for further checking.
+    if line.strip() == "#pragma once":
+        return
     linesplit = line.split()
     if len(linesplit) >= 2:
       # find the first occurrence of #ifndef and #define, save arg
@@ -3101,6 +3104,11 @@ def CheckBraces(filename, clean_lines, linenum, error):
                'LOCKS_EXCLUDED', 'INTERFACE_DEF')) or
           Search(r'\s+=\s*$', line_prefix)):
         match = None
+    # Whitelist lambda function definition which also requires a ";" after
+    # closing brace
+    if match:
+        if Match(r'^.*\[.*\]\s*(.*\)\s*)\{', line):
+            match = None
 
   else:
     # Try matching cases 2-3.

table/block_based_table_reader.cc

@@ -300,7 +300,7 @@ Status BlockBasedTable::Open(const Options& options, const EnvOptions& soptions,
       assert(index_block->compressionType() == kNoCompression);
       rep->index_block.reset(index_block);
 
-      // Set index block
+      // Set filter block
       if (rep->options.filter_policy) {
         std::string key = kFilterBlockPrefix;
         key.append(rep->options.filter_policy->Name());
@@ -681,8 +681,14 @@ Iterator* BlockBasedTable::BlockReader(void* arg,
 
 BlockBasedTable::CachableEntry<FilterBlockReader>
 BlockBasedTable::GetFilter(bool no_io) const {
-  if (!rep_->options.filter_policy || !rep_->options.block_cache) {
+  // filter pre-populated
-    return {rep_->filter.get(), nullptr};
+  if (rep_->filter != nullptr) {
+    return {rep_->filter.get(), nullptr /* cache handle */};
+  }
+
+  if (rep_->options.filter_policy == nullptr /* do not use filter at all */ ||
+      rep_->options.block_cache == nullptr /* no block cache at all */) {
+    return {nullptr /* filter */, nullptr /* cache handle */};
   }
 
   // Fetching from the cache
@@ -979,4 +985,12 @@ uint64_t BlockBasedTable::ApproximateOffsetOf(const Slice& key) {
   return result;
 }
 
+bool BlockBasedTable::TEST_filter_block_preloaded() const {
+  return rep_->filter != nullptr;
+}
+
+bool BlockBasedTable::TEST_index_block_preloaded() const {
+  return rep_->index_block != nullptr;
+}
+
 }  // namespace rocksdb

table/block_based_table_reader.h

@@ -90,6 +90,9 @@ class BlockBasedTable : public TableReader {
 
   ~BlockBasedTable();
 
+  bool TEST_filter_block_preloaded() const;
+  bool TEST_index_block_preloaded() const;
+
  private:
   template <class TValue>
   struct CachableEntry;

table/table_test.cc

@@ -29,6 +29,7 @@
 #include "rocksdb/memtablerep.h"
 #include "table/block.h"
 #include "table/meta_blocks.h"
+#include "table/block_based_table_reader.h"
 #include "table/block_based_table_builder.h"
 #include "table/block_based_table_factory.h"
 #include "table/block_based_table_reader.h"
@@ -961,6 +962,7 @@ class BlockBasedTableTest : public TableTest {};
 class PlainTableTest : public TableTest {};
 class TablePropertyTest {};
 
+/*
 // This test serves as the living tutorial for the prefix scan of user collected
 // properties.
 TEST(TablePropertyTest, PrefixScanTest) {
@@ -1122,19 +1124,37 @@ TEST(BlockBasedTableTest, NumBlockStat) {
   ASSERT_EQ(kvmap.size(),
             c.table_reader()->GetTableProperties()->num_data_blocks);
 }
+*/
 
-class BlockCacheProperties {
+// A simple tool that takes the snapshot of block cache statistics.
+class BlockCachePropertiesSnapshot {
  public:
-  explicit BlockCacheProperties(Statistics* statistics) {
+  explicit BlockCachePropertiesSnapshot(Statistics* statistics) {
     block_cache_miss = statistics->getTickerCount(BLOCK_CACHE_MISS);
     block_cache_hit = statistics->getTickerCount(BLOCK_CACHE_HIT);
     index_block_cache_miss = statistics->getTickerCount(BLOCK_CACHE_INDEX_MISS);
     index_block_cache_hit = statistics->getTickerCount(BLOCK_CACHE_INDEX_HIT);
     data_block_cache_miss = statistics->getTickerCount(BLOCK_CACHE_DATA_MISS);
     data_block_cache_hit = statistics->getTickerCount(BLOCK_CACHE_DATA_HIT);
+    filter_block_cache_miss =
+        statistics->getTickerCount(BLOCK_CACHE_FILTER_MISS);
+    filter_block_cache_hit = statistics->getTickerCount(BLOCK_CACHE_FILTER_HIT);
+  }
+
+  void AssertIndexBlockStat(int64_t index_block_cache_miss,
+                            int64_t index_block_cache_hit) {
+    ASSERT_EQ(index_block_cache_miss, this->index_block_cache_miss);
+    ASSERT_EQ(index_block_cache_hit, this->index_block_cache_hit);
+  }
+
+  void AssertFilterBlockStat(int64_t filter_block_cache_miss,
+                             int64_t filter_block_cache_hit) {
+    ASSERT_EQ(filter_block_cache_miss, this->filter_block_cache_miss);
+    ASSERT_EQ(filter_block_cache_hit, this->filter_block_cache_hit);
   }
 
   // Check if the fetched props matches the expected ones.
+  // TODO(kailiu) Use this only when you disabled filter policy!
   void AssertEqual(int64_t index_block_cache_miss,
                    int64_t index_block_cache_hit, int64_t data_block_cache_miss,
                    int64_t data_block_cache_hit) const {
@@ -1155,9 +1175,55 @@ class BlockCacheProperties {
   int64_t index_block_cache_hit = 0;
   int64_t data_block_cache_miss = 0;
   int64_t data_block_cache_hit = 0;
+  int64_t filter_block_cache_miss = 0;
+  int64_t filter_block_cache_hit = 0;
 };
 
-TEST(BlockBasedTableTest, BlockCacheTest) {
+// Make sure, by default, index/filter blocks were pre-loaded (meaning we won't
+// use block cache to store them).
+TEST(BlockBasedTableTest, BlockCacheDisabledTest) {
+  Options options;
+  options.create_if_missing = true;
+  options.statistics = CreateDBStatistics();
+  options.block_cache = NewLRUCache(1024);
+  std::unique_ptr<const FilterPolicy> filter_policy(NewBloomFilterPolicy(10));
+  options.filter_policy = filter_policy.get();
+  BlockBasedTableOptions table_options;
+  // Intentionally commented out: table_options.cache_index_and_filter_blocks =
+  // true;
+  options.table_factory.reset(new BlockBasedTableFactory(table_options));
+  std::vector<std::string> keys;
+  KVMap kvmap;
+
+  TableConstructor c(BytewiseComparator());
+  c.Add("key", "value");
+  c.Finish(options, GetPlainInternalComparator(options.comparator), &keys,
+           &kvmap);
+
+  // preloading filter/index blocks is enabled.
+  auto reader = dynamic_cast<BlockBasedTable*>(c.table_reader());
+  ASSERT_TRUE(reader->TEST_filter_block_preloaded());
+  ASSERT_TRUE(reader->TEST_index_block_preloaded());
+
+  {
+    // nothing happens in the beginning
+    BlockCachePropertiesSnapshot props(options.statistics.get());
+    props.AssertIndexBlockStat(0, 0);
+    props.AssertFilterBlockStat(0, 0);
+  }
+
+  {
+    // a hack that just to trigger BlockBasedTable::GetFilter.
+    reader->Get(ReadOptions(), "non-exist-key", nullptr, nullptr, nullptr);
+    BlockCachePropertiesSnapshot props(options.statistics.get());
+    props.AssertIndexBlockStat(0, 0);
+    props.AssertFilterBlockStat(0, 0);
+  }
+}
+
+// Due to the difficulities of the intersaction between statistics, this test
+// only tests the case when "index block is put to block cache"
+TEST(BlockBasedTableTest, FilterBlockInBlockCache) {
   // -- Table construction
   Options options;
   options.create_if_missing = true;
@@ -1175,6 +1241,10 @@ TEST(BlockBasedTableTest, BlockCacheTest) {
   c.Add("key", "value");
   c.Finish(options, GetPlainInternalComparator(options.comparator), &keys,
            &kvmap);
+  // preloading filter/index blocks is prohibited.
+  auto reader = dynamic_cast<BlockBasedTable*>(c.table_reader());
+  ASSERT_TRUE(!reader->TEST_filter_block_preloaded());
+  ASSERT_TRUE(!reader->TEST_index_block_preloaded());
 
   // -- PART 1: Open with regular block cache.
   // Since block_cache is disabled, no cache activities will be involved.
@@ -1182,7 +1252,7 @@ TEST(BlockBasedTableTest, BlockCacheTest) {
 
   // At first, no block will be accessed.
   {
-    BlockCacheProperties props(options.statistics.get());
+    BlockCachePropertiesSnapshot props(options.statistics.get());
     // index will be added to block cache.
     props.AssertEqual(1,  // index block miss
                       0, 0, 0);
@@ -1191,7 +1261,7 @@ TEST(BlockBasedTableTest, BlockCacheTest) {
   // Only index block will be accessed
   {
     iter.reset(c.NewIterator());
-    BlockCacheProperties props(options.statistics.get());
+    BlockCachePropertiesSnapshot props(options.statistics.get());
     // NOTE: to help better highlight the "detla" of each ticker, I use
     // <last_value> + <added_value> to indicate the increment of changed
     // value; other numbers remain the same.
@@ -1202,7 +1272,7 @@ TEST(BlockBasedTableTest, BlockCacheTest) {
   // Only data block will be accessed
   {
     iter->SeekToFirst();
-    BlockCacheProperties props(options.statistics.get());
+    BlockCachePropertiesSnapshot props(options.statistics.get());
     props.AssertEqual(1, 1, 0 + 1,  // data block miss
                       0);
   }
@@ -1211,7 +1281,7 @@ TEST(BlockBasedTableTest, BlockCacheTest) {
   {
     iter.reset(c.NewIterator());
     iter->SeekToFirst();
-    BlockCacheProperties props(options.statistics.get());
+    BlockCachePropertiesSnapshot props(options.statistics.get());
     props.AssertEqual(1, 1 + 1, /* index block hit */
                       1, 0 + 1 /* data block hit */);
   }
@@ -1227,7 +1297,7 @@ TEST(BlockBasedTableTest, BlockCacheTest) {
     iter.reset(c.NewIterator());
     iter->SeekToFirst();
     ASSERT_EQ("key", iter->key().ToString());
-    BlockCacheProperties props(options.statistics.get());
+    BlockCachePropertiesSnapshot props(options.statistics.get());
     // Nothing is affected at all
     props.AssertEqual(0, 0, 0, 0);
   }
@@ -1238,7 +1308,7 @@ TEST(BlockBasedTableTest, BlockCacheTest) {
   options.block_cache = NewLRUCache(1);
   c.Reopen(options);
   {
-    BlockCacheProperties props(options.statistics.get());
+    BlockCachePropertiesSnapshot props(options.statistics.get());
     props.AssertEqual(1,  // index block miss
                       0, 0, 0);
   }
@@ -1249,7 +1319,7 @@ TEST(BlockBasedTableTest, BlockCacheTest) {
     // It first cache index block then data block. But since the cache size
     // is only 1, index block will be purged after data block is inserted.
     iter.reset(c.NewIterator());
-    BlockCacheProperties props(options.statistics.get());
+    BlockCachePropertiesSnapshot props(options.statistics.get());
     props.AssertEqual(1 + 1,  // index block miss
                       0, 0,   // data block miss
                       0);
@@ -1259,7 +1329,7 @@ TEST(BlockBasedTableTest, BlockCacheTest) {
     // SeekToFirst() accesses data block. With similar reason, we expect data
     // block's cache miss.
     iter->SeekToFirst();
-    BlockCacheProperties props(options.statistics.get());
+    BlockCachePropertiesSnapshot props(options.statistics.get());
     props.AssertEqual(2, 0, 0 + 1,  // data block miss
                       0);
   }

util/log_write_bench.cc

@@ -0,0 +1,69 @@
+//  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 <gflags/gflags.h>
+
+#include "rocksdb/env.h"
+#include "util/histogram.h"
+#include "util/testharness.h"
+#include "util/testutil.h"
+
+// A simple benchmark to simulate transactional logs
+
+DEFINE_int32(num_records, 6000, "Size of each record.");
+DEFINE_int32(record_size, 249, "Size of each record.");
+DEFINE_int32(record_interval, 10000, "Interval between records (microSec)");
+DEFINE_int32(bytes_per_sync, 0, "bytes_per_sync parameter in EnvOptions");
+DEFINE_bool(enable_sync, false, "sync after each write.");
+
+namespace rocksdb {
+void RunBenchmark() {
+  std::string file_name = test::TmpDir() + "/log_write_bench.log";
+  Env* env = Env::Default();
+  EnvOptions env_options;
+  env_options.use_mmap_writes = false;
+  env_options.bytes_per_sync = FLAGS_bytes_per_sync;
+  unique_ptr<WritableFile> file;
+  env->NewWritableFile(file_name, &file, env_options);
+
+  std::string record;
+  record.assign('X', FLAGS_record_size);
+
+  HistogramImpl hist;
+
+  uint64_t start_time = env->NowMicros();
+  for (int i = 0; i < FLAGS_num_records; i++) {
+    uint64_t start_nanos = env->NowNanos();
+    file->Append(record);
+    file->Flush();
+    if (FLAGS_enable_sync) {
+      file->Sync();
+    }
+    hist.Add(env->NowNanos() - start_nanos);
+
+    if (i % 1000 == 1) {
+      fprintf(stderr, "Wrote %d records...\n", i);
+    }
+
+    int time_to_sleep =
+        (i + 1) * FLAGS_record_interval - (env->NowMicros() - start_time);
+    if (time_to_sleep > 0) {
+      env->SleepForMicroseconds(time_to_sleep);
+    }
+  }
+
+  fprintf(stderr, "Distribution of latency of append+flush: \n%s",
+          hist.ToString().c_str());
+}
+}  // namespace rocksdb
+
+int main(int argc, char** argv) {
+  google::SetUsageMessage(std::string("\nUSAGE:\n") + std::string(argv[0]) +
+                          " [OPTIONS]...");
+  google::ParseCommandLineFlags(&argc, &argv, true);
+
+  rocksdb::RunBenchmark();
+  return 0;
+}