Merge pull request #83 from basho/pevm-fast-fold

Improve fold speed for large files filled with small objects.

Author: evanmcc

.gitignore

@@ -4,3 +4,5 @@ ebin
 priv/*.so
 *.o
 *.beam
+*~
+#*#

c_src/bitcask_nifs.c

@@ -215,6 +215,7 @@ ERL_NIF_TERM bitcask_nifs_file_pread(ErlNifEnv* env, int argc, const ERL_NIF_TER
 ERL_NIF_TERM bitcask_nifs_file_pwrite(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]);
 ERL_NIF_TERM bitcask_nifs_file_read(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]);
 ERL_NIF_TERM bitcask_nifs_file_write(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]);
+ERL_NIF_TERM bitcask_nifs_file_position(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]);
 ERL_NIF_TERM bitcask_nifs_file_seekbof(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]);
 
 ERL_NIF_TERM errno_atom(ErlNifEnv* env, int error);
@@ -257,6 +258,7 @@ static ErlNifFunc nif_funcs[] =
     {"file_pwrite_int", 3, bitcask_nifs_file_pwrite},
     {"file_read_int",   2, bitcask_nifs_file_read},
     {"file_write_int",  2, bitcask_nifs_file_write},
+    {"file_position_int",  2, bitcask_nifs_file_position},
     {"file_seekbof_int", 1, bitcask_nifs_file_seekbof}
 };
 
@@ -1601,6 +1603,33 @@ ERL_NIF_TERM bitcask_nifs_file_write(ErlNifEnv* env, int argc, const ERL_NIF_TER
     }
 }
 
+ERL_NIF_TERM bitcask_nifs_file_position(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
+{
+    bitcask_file_handle* handle;
+    unsigned long offset_ul;
+
+    if (enif_get_resource(env, argv[0], bitcask_file_RESOURCE, (void**)&handle) &&
+        enif_get_ulong(env, argv[1], &offset_ul))
+    {
+
+        off_t offset = offset_ul;
+        off_t new_offset = lseek(handle->fd, offset, SEEK_SET);
+        if (new_offset != -1)
+        {
+            return enif_make_tuple2(env, ATOM_OK, enif_make_ulong(env, new_offset));
+        }
+        else
+        {
+            /* Write failed altogether */
+            return enif_make_tuple2(env, ATOM_ERROR, errno_atom(env, errno));
+        }
+    }
+    else
+    {
+        return enif_make_badarg(env);
+    }
+}
+
 ERL_NIF_TERM bitcask_nifs_file_seekbof(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
 {
     bitcask_file_handle* handle;

include/bitcask.hrl

@@ -37,3 +37,8 @@
 -define(MAXKEYSIZE, 2#1111111111111111).
 -define(MAXVALSIZE, 2#11111111111111111111111111111111).
 -define(MAXOFFSET, 16#ffffffffffffffff). % max 64-bit unsigned
+
+%% for hintfile validation
+-define(CHUNK_SIZE, 65535).
+-define(MIN_CHUNK_SIZE, 1024).
+-define(MAX_CHUNK_SIZE, 134217728).

src/bitcask.erl

@@ -315,21 +315,27 @@ fold_keys(Ref, Fun, Acc0, MaxAge, MaxPut) ->
 
 %% @doc fold over all K/V pairs in a bitcask datastore.
 %% Fun is expected to take F(K,V,Acc0) -> Acc
--spec fold(reference(), fun((binary(), binary(), any()) -> any()), any()) -> any() | {error, any()}.
-fold(Ref, Fun, Acc0) ->
+-spec fold(reference() | record(), 
+           fun((binary(), binary(), any()) -> any()), 
+           any()) -> any() | {error, any()}.
+fold(Ref, Fun, Acc0) when is_reference(Ref)->
     State = get_state(Ref),
+    fold(State, Fun, Acc0);
+fold(State, Fun, Acc0) ->
     MaxAge = get_opt(max_fold_age, State#bc_state.opts) * 1000, % convert from ms to us
     MaxPuts = get_opt(max_fold_puts, State#bc_state.opts),
-    fold(Ref, Fun, Acc0, MaxAge, MaxPuts).
+    fold(State, Fun, Acc0, MaxAge, MaxPuts).
 
 %% @doc fold over all K/V pairs in a bitcask datastore specifying max age/updates of
 %% the frozen keystore.
 %% Fun is expected to take F(K,V,Acc0) -> Acc
--spec fold(reference(), fun((binary(), binary(), any()) -> any()), any(),
-          non_neg_integer() | undefined, non_neg_integer() | undefined) -> 
+-spec fold(reference() | record(), fun((binary(), binary(), any()) -> any()), any(),
+           non_neg_integer() | undefined, non_neg_integer() | undefined) -> 
                   any() | {error, any()}.
-fold(Ref, Fun, Acc0, MaxAge, MaxPut) ->
+fold(Ref, Fun, Acc0, MaxAge, MaxPut) when is_reference(Ref)->
     State = get_state(Ref),
+    fold(State, Fun, Acc0, MaxAge, MaxPut);
+fold(State, Fun, Acc0, MaxAge, MaxPut) ->
     FrozenFun = 
         fun() ->
                 case open_fold_files(State#bc_state.dirname, 3) of
@@ -853,13 +859,15 @@ scan_key_files([Filename | Rest], KeyDir, Acc, CloseFile, EnoentOK) ->
     %% Restrictive pattern matching below is intentional
     case bitcask_fileops:open_file(Filename) of
         {ok, File} ->
+            FileTstamp = bitcask_fileops:file_tstamp(File),
             F = fun(K, Tstamp, {Offset, TotalSz}, _) ->
                         bitcask_nifs:keydir_put(KeyDir,
                                                 K,
-                                                bitcask_fileops:file_tstamp(File),
+                                                FileTstamp,
                                                 TotalSz,
                                                 Offset,
-                                                Tstamp)
+                                                Tstamp,
+                                                false)
                 end,
             bitcask_fileops:fold_keys(File, F, undefined, recovery),
             if CloseFile == true ->

src/bitcask_file.erl

@@ -64,6 +64,9 @@ file_read(Pid, Size) ->
 file_write(Pid, Bytes) ->
     file_request(Pid, {file_write, Bytes}).
 
+file_position(Pid, Position) ->
+    file_request(Pid, {file_position, Position}).
+
 file_seekbof(Pid) ->
     file_request(Pid, file_seekbof).
 
@@ -106,11 +109,11 @@ handle_call({file_open, Owner, Filename, Opts}, _From, State) ->
     IsReadOnly = proplists:get_bool(readonly, Opts),
     Mode = case {IsReadOnly, IsCreate} of
                {true, _} ->
-                   [read, raw, binary, read_ahead];
+                   [read, raw, binary];
                {_, false} ->
-                   [read, write, raw, binary, read_ahead];
+                   [read, write, raw, binary];
                {_, true} ->
-                   [read, write, exclusive, raw, binary, read_ahead]
+                   [read, write, exclusive, raw, binary]
            end,
     [warn("Bitcask file option '~p' not supported~n", [Opt])
      || Opt <- [o_sync],
@@ -146,6 +149,10 @@ handle_call({file_write, Bytes}, From, State=#state{fd=Fd}) ->
     check_owner(From, State),
     Reply = file:write(Fd, Bytes),
     {reply, Reply, State};
+handle_call({file_position, Position}, From, State=#state{fd=Fd}) ->
+    check_owner(From, State),
+    Reply = file:position(Fd, Position),
+    {reply, Reply, State};
 handle_call(file_seekbof, From, State=#state{fd=Fd}) ->
     check_owner(From, State),
     {ok, _} = file:position(Fd, bof),