Merge pull request #529 from felixhandte/lz4f-fast-reset-for-streaming-only

LZ4F: Only Reset the LZ4_stream_t when Init'ing a Streaming Block

2 files changed, 33 insertions, 10 deletions

diff --git a/lib/lz4.h b/lib/lz4.h
index 410f480..7d13122 100644
--- a/lib/lz4.h
+++ b/lib/lz4.h
@@ -380,6 +380,15 @@ LZ4LIB_API int LZ4_decompress_fast_usingDict (const char* src, char* dst, int or
 #ifdef LZ4_STATIC_LINKING_ONLY
 
 /*! LZ4_resetStream_fast() :
+ *  Use this, like LZ4_resetStream(), to prepare a context for a new chain of
+ *  calls to a streaming API (e.g., LZ4_compress_fast_continue()).
+ *
+ *  Note:
+ *  Using this in advance of a non- streaming-compression function is redundant,
+ *  and potentially bad for performance, since they all perform their own custom
+ *  reset internally.
+ *
+ *  Differences from LZ4_resetStream():
  *  When an LZ4_stream_t is known to be in a internally coherent state,
  *  it can often be prepared for a new compression with almost no work, only
  *  sometimes falling back to the full, expensive reset that is always required
@@ -389,13 +398,17 @@ LZ4LIB_API int LZ4_decompress_fast_usingDict (const char* src, char* dst, int or
  *  LZ4_streams are guaranteed to be in a valid state when:
  *  - returned from LZ4_createStream()
  *  - reset by LZ4_resetStream()
- *  - memset(stream, 0, sizeof(LZ4_stream_t))
+ *  - memset(stream, 0, sizeof(LZ4_stream_t)), though this is discouraged
  *  - the stream was in a valid state and was reset by LZ4_resetStream_fast()
  *  - the stream was in a valid state and was then used in any compression call
  *    that returned success
  *  - the stream was in an indeterminate state and was used in a compression
- *    call that fully reset the state (LZ4_compress_fast_extState()) and that
- *    returned success
+ *    call that fully reset the state (e.g., LZ4_compress_fast_extState()) and
+ *    that returned success
+ *
+ *  When a stream isn't known to be in a valid state, it is not safe to pass to
+ *  any fastReset or streaming function. It must first be cleansed by the full
+ *  LZ4_resetStream().
  */
 LZ4LIB_API void LZ4_resetStream_fast (LZ4_stream_t* streamPtr);
 
@@ -406,8 +419,9 @@ LZ4LIB_API void LZ4_resetStream_fast (LZ4_stream_t* streamPtr);
  *  to call if the state buffer is known to be correctly initialized already
  *  (see above comment on LZ4_resetStream_fast() for a definition of "correctly
  *  initialized"). From a high level, the difference is that this function
- *  initializes the provided state with a call to LZ4_resetStream_fast() while
- *  LZ4_compress_fast_extState() starts with a call to LZ4_resetStream().
+ *  initializes the provided state with a call to something like
+ *  LZ4_resetStream_fast() while LZ4_compress_fast_extState() starts with a
+ *  call to LZ4_resetStream().
  */
 LZ4LIB_API int LZ4_compress_fast_extState_fastReset (void* state, const char* src, char* dst, int srcSize, int dstCapacity, int acceleration);
 
diff --git a/lib/lz4frame.c b/lib/lz4frame.c
index f57db24..aa7889b 100644
--- a/lib/lz4frame.c
+++ b/lib/lz4frame.c
@@ -551,9 +551,18 @@ LZ4F_errorCode_t LZ4F_freeCompressionContext(LZ4F_compressionContext_t LZ4F_comp
  */
 static void LZ4F_initStream(void* ctx,
                             const LZ4F_CDict* cdict,
-                            int level) {
+                            int level,
+                            LZ4F_blockMode_t blockMode) {
     if (level < LZ4HC_CLEVEL_MIN) {
-        LZ4_resetStream_fast((LZ4_stream_t *)ctx);
+        if (cdict != NULL || blockMode == LZ4F_blockLinked) {
+            /* In these cases, we will call LZ4_compress_fast_continue(),
+             * which needs an already reset context. Otherwise, we'll call a
+             * one-shot API. The non-continued APIs internally perform their own
+             * resets at the beginning of their calls, where they know what
+             * tableType they need the context to be in. So in that case this
+             * would be misguided / wasted work. */
+            LZ4_resetStream_fast((LZ4_stream_t*)ctx);
+        }
         LZ4_attach_dictionary((LZ4_stream_t *)ctx, cdict ? cdict->fastCtx : NULL);
     } else {
         LZ4_resetStreamHC_fast((LZ4_streamHC_t*)ctx, level);
@@ -631,7 +640,7 @@ size_t LZ4F_compressBegin_usingCDict(LZ4F_cctx* cctxPtr,
     cctxPtr->cdict = cdict;
     if (cctxPtr->prefs.frameInfo.blockMode == LZ4F_blockLinked) {
         /* frame init only for blockLinked : blockIndependent will be init at each block */
-        LZ4F_initStream(cctxPtr->lz4CtxPtr, cdict, cctxPtr->prefs.compressionLevel);
+        LZ4F_initStream(cctxPtr->lz4CtxPtr, cdict, cctxPtr->prefs.compressionLevel, LZ4F_blockLinked);
     }
     if (preferencesPtr->compressionLevel >= LZ4HC_CLEVEL_MIN) {
           LZ4_favorDecompressionSpeed((LZ4_streamHC_t*)cctxPtr->lz4CtxPtr, (int)preferencesPtr->favorDecSpeed);
@@ -729,7 +738,7 @@ static size_t LZ4F_makeBlock(void* dst, const void* src, size_t srcSize,
 static int LZ4F_compressBlock(void* ctx, const char* src, char* dst, int srcSize, int dstCapacity, int level, const LZ4F_CDict* cdict)
 {
     int const acceleration = (level < -1) ? -level : 1;
-    LZ4F_initStream(ctx, cdict, level);
+    LZ4F_initStream(ctx, cdict, level, LZ4F_blockIndependent);
     if (cdict) {
         return LZ4_compress_fast_continue((LZ4_stream_t*)ctx, src, dst, srcSize, dstCapacity, acceleration);
     } else {
@@ -746,7 +755,7 @@ static int LZ4F_compressBlock_continue(void* ctx, const char* src, char* dst, in
 
 static int LZ4F_compressBlockHC(void* ctx, const char* src, char* dst, int srcSize, int dstCapacity, int level, const LZ4F_CDict* cdict)
 {
-    LZ4F_initStream(ctx, cdict, level);
+    LZ4F_initStream(ctx, cdict, level, LZ4F_blockIndependent);
     if (cdict) {
         return LZ4_compress_HC_continue((LZ4_streamHC_t*)ctx, src, dst, srcSize, dstCapacity);
     }