Merge branch 'upstream-zstd'

# By zstd upstream
* upstream-zstd:
  zstd 2020-05-21 (b706286a)

1 files changed, 1453 insertions, 1465 deletions

diff --git a/Utilities/cmzstd/lib/compress/zstd_compress.c b/Utilities/cmzstd/lib/compress/zstd_compress.c
index c2c9d3b..3f963b1 100644
--- a/Utilities/cmzstd/lib/compress/zstd_compress.c
+++ b/Utilities/cmzstd/lib/compress/zstd_compress.c
@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
+ * Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
  * All rights reserved.
  *
  * This source code is licensed under both the BSD-style license (found in the
@@ -13,24 +13,34 @@
 ***************************************/
 #include <limits.h>         /* INT_MAX */
 #include <string.h>         /* memset */
-#include "cpu.h"
-#include "mem.h"
+#include "../common/cpu.h"
+#include "../common/mem.h"
 #include "hist.h"           /* HIST_countFast_wksp */
 #define FSE_STATIC_LINKING_ONLY   /* FSE_encodeSymbol */
-#include "fse.h"
+#include "../common/fse.h"
 #define HUF_STATIC_LINKING_ONLY
-#include "huf.h"
+#include "../common/huf.h"
 #include "zstd_compress_internal.h"
+#include "zstd_compress_sequences.h"
+#include "zstd_compress_literals.h"
 #include "zstd_fast.h"
 #include "zstd_double_fast.h"
 #include "zstd_lazy.h"
 #include "zstd_opt.h"
 #include "zstd_ldm.h"
+#include "zstd_compress_superblock.h"
 
 
 /*-*************************************
 *  Helper functions
 ***************************************/
+/* ZSTD_compressBound()
+ * Note that the result from this function is only compatible with the "normal"
+ * full-block strategy.
+ * When there are a lot of small blocks due to frequent flush in streaming mode
+ * the overhead of headers can make the compressed data to be larger than the
+ * return value of ZSTD_compressBound().
+ */
 size_t ZSTD_compressBound(size_t srcSize) {
     return ZSTD_COMPRESSBOUND(srcSize);
 }
@@ -40,15 +50,15 @@ size_t ZSTD_compressBound(size_t srcSize) {
 *  Context memory management
 ***************************************/
 struct ZSTD_CDict_s {
-    void* dictBuffer;
     const void* dictContent;
     size_t dictContentSize;
-    void* workspace;
-    size_t workspaceSize;
+    U32* entropyWorkspace; /* entropy workspace of HUF_WORKSPACE_SIZE bytes */
+    ZSTD_cwksp workspace;
     ZSTD_matchState_t matchState;
     ZSTD_compressedBlockState_t cBlockState;
     ZSTD_customMem customMem;
     U32 dictID;
+    int compressionLevel; /* 0 indicates that advanced API was used to select CDict params */
 };  /* typedef'd to ZSTD_CDict within "zstd.h" */
 
 ZSTD_CCtx* ZSTD_createCCtx(void)
@@ -80,46 +90,72 @@ ZSTD_CCtx* ZSTD_createCCtx_advanced(ZSTD_customMem customMem)
     }
 }
 
-ZSTD_CCtx* ZSTD_initStaticCCtx(void *workspace, size_t workspaceSize)
+ZSTD_CCtx* ZSTD_initStaticCCtx(void* workspace, size_t workspaceSize)
 {
-    ZSTD_CCtx* const cctx = (ZSTD_CCtx*) workspace;
+    ZSTD_cwksp ws;
+    ZSTD_CCtx* cctx;
     if (workspaceSize <= sizeof(ZSTD_CCtx)) return NULL;  /* minimum size */
     if ((size_t)workspace & 7) return NULL;  /* must be 8-aligned */
-    memset(workspace, 0, workspaceSize);   /* may be a bit generous, could memset be smaller ? */
+    ZSTD_cwksp_init(&ws, workspace, workspaceSize);
+
+    cctx = (ZSTD_CCtx*)ZSTD_cwksp_reserve_object(&ws, sizeof(ZSTD_CCtx));
+    if (cctx == NULL) return NULL;
+
+    memset(cctx, 0, sizeof(ZSTD_CCtx));
+    ZSTD_cwksp_move(&cctx->workspace, &ws);
     cctx->staticSize = workspaceSize;
-    cctx->workSpace = (void*)(cctx+1);
-    cctx->workSpaceSize = workspaceSize - sizeof(ZSTD_CCtx);
 
     /* statically sized space. entropyWorkspace never moves (but prev/next block swap places) */
-    if (cctx->workSpaceSize < HUF_WORKSPACE_SIZE + 2 * sizeof(ZSTD_compressedBlockState_t)) return NULL;
-    assert(((size_t)cctx->workSpace & (sizeof(void*)-1)) == 0);   /* ensure correct alignment */
-    cctx->blockState.prevCBlock = (ZSTD_compressedBlockState_t*)cctx->workSpace;
-    cctx->blockState.nextCBlock = cctx->blockState.prevCBlock + 1;
-    {
-        void* const ptr = cctx->blockState.nextCBlock + 1;
-        cctx->entropyWorkspace = (U32*)ptr;
-    }
+    if (!ZSTD_cwksp_check_available(&cctx->workspace, HUF_WORKSPACE_SIZE + 2 * sizeof(ZSTD_compressedBlockState_t))) return NULL;
+    cctx->blockState.prevCBlock = (ZSTD_compressedBlockState_t*)ZSTD_cwksp_reserve_object(&cctx->workspace, sizeof(ZSTD_compressedBlockState_t));
+    cctx->blockState.nextCBlock = (ZSTD_compressedBlockState_t*)ZSTD_cwksp_reserve_object(&cctx->workspace, sizeof(ZSTD_compressedBlockState_t));
+    cctx->entropyWorkspace = (U32*)ZSTD_cwksp_reserve_object(&cctx->workspace, HUF_WORKSPACE_SIZE);
     cctx->bmi2 = ZSTD_cpuid_bmi2(ZSTD_cpuid());
     return cctx;
 }
 
+/**
+ * Clears and frees all of the dictionaries in the CCtx.
+ */
+static void ZSTD_clearAllDicts(ZSTD_CCtx* cctx)
+{
+    ZSTD_free(cctx->localDict.dictBuffer, cctx->customMem);
+    ZSTD_freeCDict(cctx->localDict.cdict);
+    memset(&cctx->localDict, 0, sizeof(cctx->localDict));
+    memset(&cctx->prefixDict, 0, sizeof(cctx->prefixDict));
+    cctx->cdict = NULL;
+}
+
+static size_t ZSTD_sizeof_localDict(ZSTD_localDict dict)
+{
+    size_t const bufferSize = dict.dictBuffer != NULL ? dict.dictSize : 0;
+    size_t const cdictSize = ZSTD_sizeof_CDict(dict.cdict);
+    return bufferSize + cdictSize;
+}
+
 static void ZSTD_freeCCtxContent(ZSTD_CCtx* cctx)
 {
     assert(cctx != NULL);
     assert(cctx->staticSize == 0);
-    ZSTD_free(cctx->workSpace, cctx->customMem); cctx->workSpace = NULL;
-    ZSTD_freeCDict(cctx->cdictLocal); cctx->cdictLocal = NULL;
+    ZSTD_clearAllDicts(cctx);
 #ifdef ZSTD_MULTITHREAD
     ZSTDMT_freeCCtx(cctx->mtctx); cctx->mtctx = NULL;
 #endif
+    ZSTD_cwksp_free(&cctx->workspace, cctx->customMem);
 }
 
 size_t ZSTD_freeCCtx(ZSTD_CCtx* cctx)
 {
     if (cctx==NULL) return 0;   /* support free on NULL */
-    if (cctx->staticSize) return ERROR(memory_allocation);   /* not compatible with static CCtx */
-    ZSTD_freeCCtxContent(cctx);
-    ZSTD_free(cctx, cctx->customMem);
+    RETURN_ERROR_IF(cctx->staticSize, memory_allocation,
+                    "not compatible with static CCtx");
+    {
+        int cctxInWorkspace = ZSTD_cwksp_owns_buffer(&cctx->workspace, cctx);
+        ZSTD_freeCCtxContent(cctx);
+        if (!cctxInWorkspace) {
+            ZSTD_free(cctx, cctx->customMem);
+        }
+    }
     return 0;
 }
 
@@ -138,8 +174,10 @@ static size_t ZSTD_sizeof_mtctx(const ZSTD_CCtx* cctx)
 size_t ZSTD_sizeof_CCtx(const ZSTD_CCtx* cctx)
 {
     if (cctx==NULL) return 0;   /* support sizeof on NULL */
-    return sizeof(*cctx) + cctx->workSpaceSize
-           + ZSTD_sizeof_CDict(cctx->cdictLocal)
+    /* cctx may be in the workspace */
+    return (cctx->workspace.workspace == cctx ? 0 : sizeof(*cctx))
+           + ZSTD_cwksp_sizeof(&cctx->workspace)
+           + ZSTD_sizeof_localDict(cctx->localDict)
            + ZSTD_sizeof_mtctx(cctx);
 }
 
@@ -195,7 +233,7 @@ size_t ZSTD_CCtxParams_reset(ZSTD_CCtx_params* params)
 }
 
 size_t ZSTD_CCtxParams_init(ZSTD_CCtx_params* cctxParams, int compressionLevel) {
-    if (!cctxParams) { return ERROR(GENERIC); }
+    RETURN_ERROR_IF(!cctxParams, GENERIC, "NULL pointer!");
     memset(cctxParams, 0, sizeof(*cctxParams));
     cctxParams->compressionLevel = compressionLevel;
     cctxParams->fParams.contentSizeFlag = 1;
@@ -204,26 +242,26 @@ size_t ZSTD_CCtxParams_init(ZSTD_CCtx_params* cctxParams, int compressionLevel)
 
 size_t ZSTD_CCtxParams_init_advanced(ZSTD_CCtx_params* cctxParams, ZSTD_parameters params)
 {
-    if (!cctxParams) { return ERROR(GENERIC); }
-    CHECK_F( ZSTD_checkCParams(params.cParams) );
+    RETURN_ERROR_IF(!cctxParams, GENERIC, "NULL pointer!");
+    FORWARD_IF_ERROR( ZSTD_checkCParams(params.cParams) , "");
     memset(cctxParams, 0, sizeof(*cctxParams));
+    assert(!ZSTD_checkCParams(params.cParams));
     cctxParams->cParams = params.cParams;
     cctxParams->fParams = params.fParams;
     cctxParams->compressionLevel = ZSTD_CLEVEL_DEFAULT;   /* should not matter, as all cParams are presumed properly defined */
-    assert(!ZSTD_checkCParams(params.cParams));
     return 0;
 }
 
 /* ZSTD_assignParamsToCCtxParams() :
  * params is presumed valid at this stage */
 static ZSTD_CCtx_params ZSTD_assignParamsToCCtxParams(
-        ZSTD_CCtx_params cctxParams, ZSTD_parameters params)
+        const ZSTD_CCtx_params* cctxParams, const ZSTD_parameters* params)
 {
-    ZSTD_CCtx_params ret = cctxParams;
-    ret.cParams = params.cParams;
-    ret.fParams = params.fParams;
+    ZSTD_CCtx_params ret = *cctxParams;
+    assert(!ZSTD_checkCParams(params->cParams));
+    ret.cParams = params->cParams;
+    ret.fParams = params->fParams;
     ret.compressionLevel = ZSTD_CLEVEL_DEFAULT;   /* should not matter, as all cParams are presumed properly defined */
-    assert(!ZSTD_checkCParams(params.cParams));
     return ret;
 }
 
@@ -307,8 +345,13 @@ ZSTD_bounds ZSTD_cParam_getBounds(ZSTD_cParameter param)
         return bounds;
 
     case ZSTD_c_overlapLog:
+#ifdef ZSTD_MULTITHREAD
         bounds.lowerBound = ZSTD_OVERLAPLOG_MIN;
         bounds.upperBound = ZSTD_OVERLAPLOG_MAX;
+#else
+        bounds.lowerBound = 0;
+        bounds.upperBound = 0;
+#endif
         return bounds;
 
     case ZSTD_c_enableLongDistanceMatching:
@@ -356,32 +399,47 @@ ZSTD_bounds ZSTD_cParam_getBounds(ZSTD_cParameter param)
     case ZSTD_c_forceAttachDict:
         ZSTD_STATIC_ASSERT(ZSTD_dictDefaultAttach < ZSTD_dictForceCopy);
         bounds.lowerBound = ZSTD_dictDefaultAttach;
-        bounds.upperBound = ZSTD_dictForceCopy;       /* note : how to ensure at compile time that this is the highest value enum ? */
+        bounds.upperBound = ZSTD_dictForceLoad;       /* note : how to ensure at compile time that this is the highest value enum ? */
+        return bounds;
+
+    case ZSTD_c_literalCompressionMode:
+        ZSTD_STATIC_ASSERT(ZSTD_lcm_auto < ZSTD_lcm_huffman && ZSTD_lcm_huffman < ZSTD_lcm_uncompressed);
+        bounds.lowerBound = ZSTD_lcm_auto;
+        bounds.upperBound = ZSTD_lcm_uncompressed;
+        return bounds;
+
+    case ZSTD_c_targetCBlockSize:
+        bounds.lowerBound = ZSTD_TARGETCBLOCKSIZE_MIN;
+        bounds.upperBound = ZSTD_TARGETCBLOCKSIZE_MAX;
+        return bounds;
+
+    case ZSTD_c_srcSizeHint:
+        bounds.lowerBound = ZSTD_SRCSIZEHINT_MIN;
+        bounds.upperBound = ZSTD_SRCSIZEHINT_MAX;
         return bounds;
 
     default:
-        {   ZSTD_bounds const boundError = { ERROR(parameter_unsupported), 0, 0 };
-            return boundError;
-        }
+        bounds.error = ERROR(parameter_unsupported);
+        return bounds;
     }
 }
 
-/* ZSTD_cParam_withinBounds:
- * @return 1 if value is within cParam bounds,
- * 0 otherwise */
-static int ZSTD_cParam_withinBounds(ZSTD_cParameter cParam, int value)
+/* ZSTD_cParam_clampBounds:
+ * Clamps the value into the bounded range.
+ */
+static size_t ZSTD_cParam_clampBounds(ZSTD_cParameter cParam, int* value)
 {
     ZSTD_bounds const bounds = ZSTD_cParam_getBounds(cParam);
-    if (ZSTD_isError(bounds.error)) return 0;
-    if (value < bounds.lowerBound) return 0;
-    if (value > bounds.upperBound) return 0;
-    return 1;
+    if (ZSTD_isError(bounds.error)) return bounds.error;
+    if (*value < bounds.lowerBound) *value = bounds.lowerBound;
+    if (*value > bounds.upperBound) *value = bounds.upperBound;
+    return 0;
 }
 
-#define BOUNDCHECK(cParam, val) {                  \
-    if (!ZSTD_cParam_withinBounds(cParam,val)) {   \
-        return ERROR(parameter_outOfBound);        \
-}   }
+#define BOUNDCHECK(cParam, val) { \
+    RETURN_ERROR_IF(!ZSTD_cParam_withinBounds(cParam,val), \
+                    parameter_outOfBound, "Param out of bounds"); \
+}
 
 
 static int ZSTD_isUpdateAuthorized(ZSTD_cParameter param)
@@ -413,6 +471,9 @@ static int ZSTD_isUpdateAuthorized(ZSTD_cParameter param)
     case ZSTD_c_ldmBucketSizeLog:
     case ZSTD_c_ldmHashRateLog:
     case ZSTD_c_forceAttachDict:
+    case ZSTD_c_literalCompressionMode:
+    case ZSTD_c_targetCBlockSize:
+    case ZSTD_c_srcSizeHint:
     default:
         return 0;
     }
@@ -425,18 +486,17 @@ size_t ZSTD_CCtx_setParameter(ZSTD_CCtx* cctx, ZSTD_cParameter param, int value)
         if (ZSTD_isUpdateAuthorized(param)) {
             cctx->cParamsChanged = 1;
         } else {
-            return ERROR(stage_wrong);
+            RETURN_ERROR(stage_wrong, "can only set params in ctx init stage");
     }   }
 
     switch(param)
     {
-    case ZSTD_c_format :
-        return ZSTD_CCtxParam_setParameter(&cctx->requestedParams, param, value);
+    case ZSTD_c_nbWorkers:
+        RETURN_ERROR_IF((value!=0) && cctx->staticSize, parameter_unsupported,
+                        "MT not compatible with static alloc");
+        break;
 
     case ZSTD_c_compressionLevel:
-        if (cctx->cdict) return ERROR(stage_wrong);
-        return ZSTD_CCtxParam_setParameter(&cctx->requestedParams, param, value);
-
     case ZSTD_c_windowLog:
     case ZSTD_c_hashLog:
     case ZSTD_c_chainLog:
@@ -444,49 +504,34 @@ size_t ZSTD_CCtx_setParameter(ZSTD_CCtx* cctx, ZSTD_cParameter param, int value)
     case ZSTD_c_minMatch:
     case ZSTD_c_targetLength:
     case ZSTD_c_strategy:
-        if (cctx->cdict) return ERROR(stage_wrong);
-        return ZSTD_CCtxParam_setParameter(&cctx->requestedParams, param, value);
-
+    case ZSTD_c_ldmHashRateLog:
+    case ZSTD_c_format:
     case ZSTD_c_contentSizeFlag:
     case ZSTD_c_checksumFlag:
     case ZSTD_c_dictIDFlag:
-        return ZSTD_CCtxParam_setParameter(&cctx->requestedParams, param, value);
-
-    case ZSTD_c_forceMaxWindow :  /* Force back-references to remain < windowSize,
-                                   * even when referencing into Dictionary content.
-                                   * default : 0 when using a CDict, 1 when using a Prefix */
-        return ZSTD_CCtxParam_setParameter(&cctx->requestedParams, param, value);
-
+    case ZSTD_c_forceMaxWindow:
     case ZSTD_c_forceAttachDict:
-        return ZSTD_CCtxParam_setParameter(&cctx->requestedParams, param, value);
-
-    case ZSTD_c_nbWorkers:
-        if ((value!=0) && cctx->staticSize) {
-            return ERROR(parameter_unsupported);  /* MT not compatible with static alloc */
-        }
-        return ZSTD_CCtxParam_setParameter(&cctx->requestedParams, param, value);
-
+    case ZSTD_c_literalCompressionMode:
     case ZSTD_c_jobSize:
     case ZSTD_c_overlapLog:
     case ZSTD_c_rsyncable:
-        return ZSTD_CCtxParam_setParameter(&cctx->requestedParams, param, value);
-
     case ZSTD_c_enableLongDistanceMatching:
     case ZSTD_c_ldmHashLog:
     case ZSTD_c_ldmMinMatch:
     case ZSTD_c_ldmBucketSizeLog:
-    case ZSTD_c_ldmHashRateLog:
-        if (cctx->cdict) return ERROR(stage_wrong);
-        return ZSTD_CCtxParam_setParameter(&cctx->requestedParams, param, value);
+    case ZSTD_c_targetCBlockSize:
+    case ZSTD_c_srcSizeHint:
+        break;
 
-    default: return ERROR(parameter_unsupported);
+    default: RETURN_ERROR(parameter_unsupported, "unknown parameter");
     }
+    return ZSTD_CCtxParams_setParameter(&cctx->requestedParams, param, value);
 }
 
-size_t ZSTD_CCtxParam_setParameter(ZSTD_CCtx_params* CCtxParams,
-                                   ZSTD_cParameter param, int value)
+size_t ZSTD_CCtxParams_setParameter(ZSTD_CCtx_params* CCtxParams,
+                                    ZSTD_cParameter param, int value)
 {
-    DEBUGLOG(4, "ZSTD_CCtxParam_setParameter (%i, %i)", (int)param, value);
+    DEBUGLOG(4, "ZSTD_CCtxParams_setParameter (%i, %i)", (int)param, value);
     switch(param)
     {
     case ZSTD_c_format :
@@ -495,39 +540,37 @@ size_t ZSTD_CCtxParam_setParameter(ZSTD_CCtx_params* CCtxParams,
         return (size_t)CCtxParams->format;
 
     case ZSTD_c_compressionLevel : {
-        int cLevel = value;
-        if (cLevel > ZSTD_maxCLevel()) cLevel = ZSTD_maxCLevel();
-        if (cLevel < ZSTD_minCLevel()) cLevel = ZSTD_minCLevel();
-        if (cLevel) {  /* 0 : does not change current level */
-            CCtxParams->compressionLevel = cLevel;
+        FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(param, &value), "");
+        if (value) {  /* 0 : does not change current level */
+            CCtxParams->compressionLevel = value;
         }
-        if (CCtxParams->compressionLevel >= 0) return CCtxParams->compressionLevel;
+        if (CCtxParams->compressionLevel >= 0) return (size_t)CCtxParams->compressionLevel;
         return 0;  /* return type (size_t) cannot represent negative values */
     }
 
     case ZSTD_c_windowLog :
         if (value!=0)   /* 0 => use default */
             BOUNDCHECK(ZSTD_c_windowLog, value);
-        CCtxParams->cParams.windowLog = value;
+        CCtxParams->cParams.windowLog = (U32)value;
         return CCtxParams->cParams.windowLog;
 
     case ZSTD_c_hashLog :
         if (value!=0)   /* 0 => use default */
             BOUNDCHECK(ZSTD_c_hashLog, value);
-        CCtxParams->cParams.hashLog = value;
+        CCtxParams->cParams.hashLog = (U32)value;
         return CCtxParams->cParams.hashLog;
 
     case ZSTD_c_chainLog :
         if (value!=0)   /* 0 => use default */
             BOUNDCHECK(ZSTD_c_chainLog, value);
-        CCtxParams->cParams.chainLog = value;
+        CCtxParams->cParams.chainLog = (U32)value;
         return CCtxParams->cParams.chainLog;
 
     case ZSTD_c_searchLog :
         if (value!=0)   /* 0 => use default */
             BOUNDCHECK(ZSTD_c_searchLog, value);
-        CCtxParams->cParams.searchLog = value;
-        return value;
+        CCtxParams->cParams.searchLog = (U32)value;
+        return (size_t)value;
 
     case ZSTD_c_minMatch :
         if (value!=0)   /* 0 => use default */
@@ -573,33 +616,55 @@ size_t ZSTD_CCtxParam_setParameter(ZSTD_CCtx_params* CCtxParams,
         return CCtxParams->attachDictPref;
     }
 
+    case ZSTD_c_literalCompressionMode : {
+        const ZSTD_literalCompressionMode_e lcm = (ZSTD_literalCompressionMode_e)value;
+        BOUNDCHECK(ZSTD_c_literalCompressionMode, lcm);
+        CCtxParams->literalCompressionMode = lcm;
+        return CCtxParams->literalCompressionMode;
+    }
+
     case ZSTD_c_nbWorkers :
 #ifndef ZSTD_MULTITHREAD
-        if (value!=0) return ERROR(parameter_unsupported);
+        RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
         return 0;
 #else
-        return ZSTDMT_CCtxParam_setNbWorkers(CCtxParams, value);
+        FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(param, &value), "");
+        CCtxParams->nbWorkers = value;
+        return CCtxParams->nbWorkers;
 #endif
 
     case ZSTD_c_jobSize :
 #ifndef ZSTD_MULTITHREAD
-        return ERROR(parameter_unsupported);
+        RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
+        return 0;
 #else
-        return ZSTDMT_CCtxParam_setMTCtxParameter(CCtxParams, ZSTDMT_p_jobSize, value);
+        /* Adjust to the minimum non-default value. */
+        if (value != 0 && value < ZSTDMT_JOBSIZE_MIN)
+            value = ZSTDMT_JOBSIZE_MIN;
+        FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(param, &value), "");
+        assert(value >= 0);
+        CCtxParams->jobSize = value;
+        return CCtxParams->jobSize;
 #endif
 
     case ZSTD_c_overlapLog :
 #ifndef ZSTD_MULTITHREAD
-        return ERROR(parameter_unsupported);
+        RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
+        return 0;
 #else
-        return ZSTDMT_CCtxParam_setMTCtxParameter(CCtxParams, ZSTDMT_p_overlapLog, value);
+        FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(ZSTD_c_overlapLog, &value), "");
+        CCtxParams->overlapLog = value;
+        return CCtxParams->overlapLog;
 #endif
 
     case ZSTD_c_rsyncable :
 #ifndef ZSTD_MULTITHREAD
-        return ERROR(parameter_unsupported);
+        RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
+        return 0;
 #else
-        return ZSTDMT_CCtxParam_setMTCtxParameter(CCtxParams, ZSTDMT_p_rsyncable, value);
+        FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(ZSTD_c_overlapLog, &value), "");
+        CCtxParams->rsyncable = value;
+        return CCtxParams->rsyncable;
 #endif
 
     case ZSTD_c_enableLongDistanceMatching :
@@ -625,21 +690,33 @@ size_t ZSTD_CCtxParam_setParameter(ZSTD_CCtx_params* CCtxParams,
         return CCtxParams->ldmParams.bucketSizeLog;
 
     case ZSTD_c_ldmHashRateLog :
-        if (value > ZSTD_WINDOWLOG_MAX - ZSTD_HASHLOG_MIN)
-            return ERROR(parameter_outOfBound);
+        RETURN_ERROR_IF(value > ZSTD_WINDOWLOG_MAX - ZSTD_HASHLOG_MIN,
+                        parameter_outOfBound, "Param out of bounds!");
         CCtxParams->ldmParams.hashRateLog = value;
         return CCtxParams->ldmParams.hashRateLog;
 
-    default: return ERROR(parameter_unsupported);
+    case ZSTD_c_targetCBlockSize :
+        if (value!=0)   /* 0 ==> default */
+            BOUNDCHECK(ZSTD_c_targetCBlockSize, value);
+        CCtxParams->targetCBlockSize = value;
+        return CCtxParams->targetCBlockSize;
+
+    case ZSTD_c_srcSizeHint :
+        if (value!=0)    /* 0 ==> default */
+            BOUNDCHECK(ZSTD_c_srcSizeHint, value);
+        CCtxParams->srcSizeHint = value;
+        return CCtxParams->srcSizeHint;
+
+    default: RETURN_ERROR(parameter_unsupported, "unknown parameter");
     }
 }
 
 size_t ZSTD_CCtx_getParameter(ZSTD_CCtx* cctx, ZSTD_cParameter param, int* value)
 {
-    return ZSTD_CCtxParam_getParameter(&cctx->requestedParams, param, value);
+    return ZSTD_CCtxParams_getParameter(&cctx->requestedParams, param, value);
 }
 
-size_t ZSTD_CCtxParam_getParameter(
+size_t ZSTD_CCtxParams_getParameter(
         ZSTD_CCtx_params* CCtxParams, ZSTD_cParameter param, int* value)
 {
     switch(param)
@@ -651,13 +728,13 @@ size_t ZSTD_CCtxParam_getParameter(
         *value = CCtxParams->compressionLevel;
         break;
     case ZSTD_c_windowLog :
-        *value = CCtxParams->cParams.windowLog;
+        *value = (int)CCtxParams->cParams.windowLog;
         break;
     case ZSTD_c_hashLog :
-        *value = CCtxParams->cParams.hashLog;
+        *value = (int)CCtxParams->cParams.hashLog;
         break;
     case ZSTD_c_chainLog :
-        *value = CCtxParams->cParams.chainLog;
+        *value = (int)CCtxParams->cParams.chainLog;
         break;
     case ZSTD_c_searchLog :
         *value = CCtxParams->cParams.searchLog;
@@ -686,6 +763,9 @@ size_t ZSTD_CCtxParam_getParameter(
     case ZSTD_c_forceAttachDict :
         *value = CCtxParams->attachDictPref;
         break;
+    case ZSTD_c_literalCompressionMode :
+        *value = CCtxParams->literalCompressionMode;
+        break;
     case ZSTD_c_nbWorkers :
 #ifndef ZSTD_MULTITHREAD
         assert(CCtxParams->nbWorkers == 0);
@@ -694,7 +774,7 @@ size_t ZSTD_CCtxParam_getParameter(
         break;
     case ZSTD_c_jobSize :
 #ifndef ZSTD_MULTITHREAD
-        return ERROR(parameter_unsupported);
+        RETURN_ERROR(parameter_unsupported, "not compiled with multithreading");
 #else
         assert(CCtxParams->jobSize <= INT_MAX);
         *value = (int)CCtxParams->jobSize;
@@ -702,14 +782,14 @@ size_t ZSTD_CCtxParam_getParameter(
 #endif
     case ZSTD_c_overlapLog :
 #ifndef ZSTD_MULTITHREAD
-        return ERROR(parameter_unsupported);
+        RETURN_ERROR(parameter_unsupported, "not compiled with multithreading");
 #else
         *value = CCtxParams->overlapLog;
         break;
 #endif
     case ZSTD_c_rsyncable :
 #ifndef ZSTD_MULTITHREAD
-        return ERROR(parameter_unsupported);
+        RETURN_ERROR(parameter_unsupported, "not compiled with multithreading");
 #else
         *value = CCtxParams->rsyncable;
         break;
@@ -729,7 +809,13 @@ size_t ZSTD_CCtxParam_getParameter(
     case ZSTD_c_ldmHashRateLog :
         *value = CCtxParams->ldmParams.hashRateLog;
         break;
-    default: return ERROR(parameter_unsupported);
+    case ZSTD_c_targetCBlockSize :
+        *value = (int)CCtxParams->targetCBlockSize;
+        break;
+    case ZSTD_c_srcSizeHint :
+        *value = (int)CCtxParams->srcSizeHint;
+        break;
+    default: RETURN_ERROR(parameter_unsupported, "unknown parameter");
     }
     return 0;
 }
@@ -745,8 +831,11 @@ size_t ZSTD_CCtx_setParametersUsingCCtxParams(
         ZSTD_CCtx* cctx, const ZSTD_CCtx_params* params)
 {
     DEBUGLOG(4, "ZSTD_CCtx_setParametersUsingCCtxParams");
-    if (cctx->streamStage != zcss_init) return ERROR(stage_wrong);
-    if (cctx->cdict) return ERROR(stage_wrong);
+    RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
+                    "The context is in the wrong stage!");
+    RETURN_ERROR_IF(cctx->cdict, stage_wrong,
+                    "Can't override parameters with cdict attached (some must "
+                    "be inherited from the cdict).");
 
     cctx->requestedParams = *params;
     return 0;
@@ -755,33 +844,73 @@ size_t ZSTD_CCtx_setParametersUsingCCtxParams(
 ZSTDLIB_API size_t ZSTD_CCtx_setPledgedSrcSize(ZSTD_CCtx* cctx, unsigned long long pledgedSrcSize)
 {
     DEBUGLOG(4, "ZSTD_CCtx_setPledgedSrcSize to %u bytes", (U32)pledgedSrcSize);
-    if (cctx->streamStage != zcss_init) return ERROR(stage_wrong);
+    RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
+                    "Can't set pledgedSrcSize when not in init stage.");
     cctx->pledgedSrcSizePlusOne = pledgedSrcSize+1;
     return 0;
 }
 
+/**
+ * Initializes the local dict using the requested parameters.
+ * NOTE: This does not use the pledged src size, because it may be used for more
+ * than one compression.
+ */
+static size_t ZSTD_initLocalDict(ZSTD_CCtx* cctx)
+{
+    ZSTD_localDict* const dl = &cctx->localDict;
+    ZSTD_compressionParameters const cParams = ZSTD_getCParamsFromCCtxParams(
+            &cctx->requestedParams, ZSTD_CONTENTSIZE_UNKNOWN, dl->dictSize);
+    if (dl->dict == NULL) {
+        /* No local dictionary. */
+        assert(dl->dictBuffer == NULL);
+        assert(dl->cdict == NULL);
+        assert(dl->dictSize == 0);
+        return 0;
+    }
+    if (dl->cdict != NULL) {
+        assert(cctx->cdict == dl->cdict);
+        /* Local dictionary already initialized. */
+        return 0;
+    }
+    assert(dl->dictSize > 0);
+    assert(cctx->cdict == NULL);
+    assert(cctx->prefixDict.dict == NULL);
+
+    dl->cdict = ZSTD_createCDict_advanced(
+            dl->dict,
+            dl->dictSize,
+            ZSTD_dlm_byRef,
+            dl->dictContentType,
+            cParams,
+            cctx->customMem);
+    RETURN_ERROR_IF(!dl->cdict, memory_allocation, "ZSTD_createCDict_advanced failed");
+    cctx->cdict = dl->cdict;
+    return 0;
+}
+
 size_t ZSTD_CCtx_loadDictionary_advanced(
         ZSTD_CCtx* cctx, const void* dict, size_t dictSize,
         ZSTD_dictLoadMethod_e dictLoadMethod, ZSTD_dictContentType_e dictContentType)
 {
-    if (cctx->streamStage != zcss_init) return ERROR(stage_wrong);
-    if (cctx->staticSize) return ERROR(memory_allocation);  /* no malloc for static CCtx */
+    RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
+                    "Can't load a dictionary when ctx is not in init stage.");
+    RETURN_ERROR_IF(cctx->staticSize, memory_allocation,
+                    "no malloc for static CCtx");
     DEBUGLOG(4, "ZSTD_CCtx_loadDictionary_advanced (size: %u)", (U32)dictSize);
-    ZSTD_freeCDict(cctx->cdictLocal);  /* in case one already exists */
-    if (dict==NULL || dictSize==0) {   /* no dictionary mode */
-        cctx->cdictLocal = NULL;
-        cctx->cdict = NULL;
+    ZSTD_clearAllDicts(cctx);  /* in case one already exists */
+    if (dict == NULL || dictSize == 0)  /* no dictionary mode */
+        return 0;
+    if (dictLoadMethod == ZSTD_dlm_byRef) {
+        cctx->localDict.dict = dict;
     } else {
-        ZSTD_compressionParameters const cParams =
-                ZSTD_getCParamsFromCCtxParams(&cctx->requestedParams, cctx->pledgedSrcSizePlusOne-1, dictSize);
-        cctx->cdictLocal = ZSTD_createCDict_advanced(
-                                dict, dictSize,
-                                dictLoadMethod, dictContentType,
-                                cParams, cctx->customMem);
-        cctx->cdict = cctx->cdictLocal;
-        if (cctx->cdictLocal == NULL)
-            return ERROR(memory_allocation);
+        void* dictBuffer = ZSTD_malloc(dictSize, cctx->customMem);
+        RETURN_ERROR_IF(!dictBuffer, memory_allocation, "NULL pointer!");
+        memcpy(dictBuffer, dict, dictSize);
+        cctx->localDict.dictBuffer = dictBuffer;
+        cctx->localDict.dict = dictBuffer;
     }
+    cctx->localDict.dictSize = dictSize;
+    cctx->localDict.dictContentType = dictContentType;
     return 0;
 }
 
@@ -801,9 +930,11 @@ ZSTDLIB_API size_t ZSTD_CCtx_loadDictionary(ZSTD_CCtx* cctx, const void* dict, s
 
 size_t ZSTD_CCtx_refCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict)
 {
-    if (cctx->streamStage != zcss_init) return ERROR(stage_wrong);
+    RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
+                    "Can't ref a dict when ctx not in init stage.");
+    /* Free the existing local cdict (if any) to save memory. */
+    ZSTD_clearAllDicts(cctx);
     cctx->cdict = cdict;
-    memset(&cctx->prefixDict, 0, sizeof(cctx->prefixDict));  /* exclusive */
     return 0;
 }
 
@@ -815,11 +946,14 @@ size_t ZSTD_CCtx_refPrefix(ZSTD_CCtx* cctx, const void* prefix, size_t prefixSiz
 size_t ZSTD_CCtx_refPrefix_advanced(
         ZSTD_CCtx* cctx, const void* prefix, size_t prefixSize, ZSTD_dictContentType_e dictContentType)
 {
-    if (cctx->streamStage != zcss_init) return ERROR(stage_wrong);
-    cctx->cdict = NULL;   /* prefix discards any prior cdict */
-    cctx->prefixDict.dict = prefix;
-    cctx->prefixDict.dictSize = prefixSize;
-    cctx->prefixDict.dictContentType = dictContentType;
+    RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
+                    "Can't ref a prefix when ctx not in init stage.");
+    ZSTD_clearAllDicts(cctx);
+    if (prefix != NULL && prefixSize > 0) {
+        cctx->prefixDict.dict = prefix;
+        cctx->prefixDict.dictSize = prefixSize;
+        cctx->prefixDict.dictContentType = dictContentType;
+    }
     return 0;
 }
 
@@ -834,8 +968,9 @@ size_t ZSTD_CCtx_reset(ZSTD_CCtx* cctx, ZSTD_ResetDirective reset)
     }
     if ( (reset == ZSTD_reset_parameters)
       || (reset == ZSTD_reset_session_and_parameters) ) {
-        if (cctx->streamStage != zcss_init) return ERROR(stage_wrong);
-        cctx->cdict = NULL;
+        RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
+                        "Can't reset parameters only when not in init stage.");
+        ZSTD_clearAllDicts(cctx);
         return ZSTD_CCtxParams_reset(&cctx->requestedParams);
     }
     return 0;
@@ -847,12 +982,12 @@ size_t ZSTD_CCtx_reset(ZSTD_CCtx* cctx, ZSTD_ResetDirective reset)
     @return : 0, or an error code if one value is beyond authorized range */
 size_t ZSTD_checkCParams(ZSTD_compressionParameters cParams)
 {
-    BOUNDCHECK(ZSTD_c_windowLog, cParams.windowLog);
-    BOUNDCHECK(ZSTD_c_chainLog,  cParams.chainLog);
-    BOUNDCHECK(ZSTD_c_hashLog,   cParams.hashLog);
-    BOUNDCHECK(ZSTD_c_searchLog, cParams.searchLog);
-    BOUNDCHECK(ZSTD_c_minMatch,  cParams.minMatch);
-    BOUNDCHECK(ZSTD_c_targetLength,cParams.targetLength);
+    BOUNDCHECK(ZSTD_c_windowLog, (int)cParams.windowLog);
+    BOUNDCHECK(ZSTD_c_chainLog,  (int)cParams.chainLog);
+    BOUNDCHECK(ZSTD_c_hashLog,   (int)cParams.hashLog);
+    BOUNDCHECK(ZSTD_c_searchLog, (int)cParams.searchLog);
+    BOUNDCHECK(ZSTD_c_minMatch,  (int)cParams.minMatch);
+    BOUNDCHECK(ZSTD_c_targetLength,(int)cParams.targetLength);
     BOUNDCHECK(ZSTD_c_strategy,  cParams.strategy);
     return 0;
 }
@@ -868,7 +1003,7 @@ ZSTD_clampCParams(ZSTD_compressionParameters cParams)
         if ((int)val<bounds.lowerBound) val=(type)bounds.lowerBound;      \
         else if ((int)val>bounds.upperBound) val=(type)bounds.upperBound; \
     }
-#   define CLAMP(cParam, val) CLAMP_TYPE(cParam, val, int)
+#   define CLAMP(cParam, val) CLAMP_TYPE(cParam, val, unsigned)
     CLAMP(ZSTD_c_windowLog, cParams.windowLog);
     CLAMP(ZSTD_c_chainLog,  cParams.chainLog);
     CLAMP(ZSTD_c_hashLog,   cParams.hashLog);
@@ -881,17 +1016,18 @@ ZSTD_clampCParams(ZSTD_compressionParameters cParams)
 
 /** ZSTD_cycleLog() :
  *  condition for correct operation : hashLog > 1 */
-static U32 ZSTD_cycleLog(U32 hashLog, ZSTD_strategy strat)
+U32 ZSTD_cycleLog(U32 hashLog, ZSTD_strategy strat)
 {
     U32 const btScale = ((U32)strat >= (U32)ZSTD_btlazy2);
     return hashLog - btScale;
 }
 
 /** ZSTD_adjustCParams_internal() :
-    optimize `cPar` for a given input (`srcSize` and `dictSize`).
-    mostly downsizing to reduce memory consumption and initialization latency.
-    Both `srcSize` and `dictSize` are optional (use 0 if unknown).
-    Note : cPar is assumed validated. Use ZSTD_checkCParams() to ensure this condition. */
+ *  optimize `cPar` for a specified input (`srcSize` and `dictSize`).
+ *  mostly downsize to reduce memory consumption and initialization latency.
+ * `srcSize` can be ZSTD_CONTENTSIZE_UNKNOWN when not known.
+ *  note : `srcSize==0` means 0!
+ *  condition : cPar is presumed validated (can be checked using ZSTD_checkCParams()). */
 static ZSTD_compressionParameters
 ZSTD_adjustCParams_internal(ZSTD_compressionParameters cPar,
                             unsigned long long srcSize,
@@ -901,10 +1037,8 @@ ZSTD_adjustCParams_internal(ZSTD_compressionParameters cPar,
     static const U64 maxWindowResize = 1ULL << (ZSTD_WINDOWLOG_MAX-1);
     assert(ZSTD_checkCParams(cPar)==0);
 
-    if (dictSize && (srcSize+1<2) /* srcSize unknown */ )
-        srcSize = minSrcSize;  /* presumed small when there is a dictionary */
-    else if (srcSize == 0)
-        srcSize = ZSTD_CONTENTSIZE_UNKNOWN;  /* 0 == unknown : presumed large */
+    if (dictSize && srcSize == ZSTD_CONTENTSIZE_UNKNOWN)
+        srcSize = minSrcSize;
 
     /* resize windowLog if input is small enough, to use less memory */
     if ( (srcSize < maxWindowResize)
@@ -922,7 +1056,7 @@ ZSTD_adjustCParams_internal(ZSTD_compressionParameters cPar,
     }
 
     if (cPar.windowLog < ZSTD_WINDOWLOG_ABSOLUTEMIN)
-        cPar.windowLog = ZSTD_WINDOWLOG_ABSOLUTEMIN;  /* required for frame header */
+        cPar.windowLog = ZSTD_WINDOWLOG_ABSOLUTEMIN;  /* minimum wlog required for valid frame header */
 
     return cPar;
 }
@@ -932,14 +1066,22 @@ ZSTD_adjustCParams(ZSTD_compressionParameters cPar,
                    unsigned long long srcSize,
                    size_t dictSize)
 {
-    cPar = ZSTD_clampCParams(cPar);
+    cPar = ZSTD_clampCParams(cPar);   /* resulting cPar is necessarily valid (all parameters within range) */
+    if (srcSize == 0) srcSize = ZSTD_CONTENTSIZE_UNKNOWN;
     return ZSTD_adjustCParams_internal(cPar, srcSize, dictSize);
 }
 
+static ZSTD_compressionParameters ZSTD_getCParams_internal(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize);
+static ZSTD_parameters ZSTD_getParams_internal(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize);
+
 ZSTD_compressionParameters ZSTD_getCParamsFromCCtxParams(
         const ZSTD_CCtx_params* CCtxParams, U64 srcSizeHint, size_t dictSize)
 {
-    ZSTD_compressionParameters cParams = ZSTD_getCParams(CCtxParams->compressionLevel, srcSizeHint, dictSize);
+    ZSTD_compressionParameters cParams;
+    if (srcSizeHint == ZSTD_CONTENTSIZE_UNKNOWN && CCtxParams->srcSizeHint > 0) {
+      srcSizeHint = CCtxParams->srcSizeHint;
+    }
+    cParams = ZSTD_getCParams_internal(CCtxParams->compressionLevel, srcSizeHint, dictSize);
     if (CCtxParams->ldmParams.enableLdm) cParams.windowLog = ZSTD_LDM_DEFAULT_WINDOW_LOG;
     if (CCtxParams->cParams.windowLog) cParams.windowLog = CCtxParams->cParams.windowLog;
     if (CCtxParams->cParams.hashLog) cParams.hashLog = CCtxParams->cParams.hashLog;
@@ -949,6 +1091,7 @@ ZSTD_compressionParameters ZSTD_getCParamsFromCCtxParams(
     if (CCtxParams->cParams.targetLength) cParams.targetLength = CCtxParams->cParams.targetLength;
     if (CCtxParams->cParams.strategy) cParams.strategy = CCtxParams->cParams.strategy;
     assert(!ZSTD_checkCParams(cParams));
+    /* srcSizeHint == 0 means 0 */
     return ZSTD_adjustCParams_internal(cParams, srcSizeHint, dictSize);
 }
 
@@ -959,10 +1102,19 @@ ZSTD_sizeof_matchState(const ZSTD_compressionParameters* const cParams,
     size_t const chainSize = (cParams->strategy == ZSTD_fast) ? 0 : ((size_t)1 << cParams->chainLog);
     size_t const hSize = ((size_t)1) << cParams->hashLog;
     U32    const hashLog3 = (forCCtx && cParams->minMatch==3) ? MIN(ZSTD_HASHLOG3_MAX, cParams->windowLog) : 0;
-    size_t const h3Size = ((size_t)1) << hashLog3;
-    size_t const tableSpace = (chainSize + hSize + h3Size) * sizeof(U32);
-    size_t const optPotentialSpace = ((MaxML+1) + (MaxLL+1) + (MaxOff+1) + (1<<Litbits)) * sizeof(U32)
-                          + (ZSTD_OPT_NUM+1) * (sizeof(ZSTD_match_t)+sizeof(ZSTD_optimal_t));
+    size_t const h3Size = hashLog3 ? ((size_t)1) << hashLog3 : 0;
+    /* We don't use ZSTD_cwksp_alloc_size() here because the tables aren't
+     * surrounded by redzones in ASAN. */
+    size_t const tableSpace = chainSize * sizeof(U32)
+                            + hSize * sizeof(U32)
+                            + h3Size * sizeof(U32);
+    size_t const optPotentialSpace =
+        ZSTD_cwksp_alloc_size((MaxML+1) * sizeof(U32))
+      + ZSTD_cwksp_alloc_size((MaxLL+1) * sizeof(U32))
+      + ZSTD_cwksp_alloc_size((MaxOff+1) * sizeof(U32))
+      + ZSTD_cwksp_alloc_size((1<<Litbits) * sizeof(U32))
+      + ZSTD_cwksp_alloc_size((ZSTD_OPT_NUM+1) * sizeof(ZSTD_match_t))
+      + ZSTD_cwksp_alloc_size((ZSTD_OPT_NUM+1) * sizeof(ZSTD_optimal_t));
     size_t const optSpace = (forCCtx && (cParams->strategy >= ZSTD_btopt))
                                 ? optPotentialSpace
                                 : 0;
@@ -973,27 +1125,42 @@ ZSTD_sizeof_matchState(const ZSTD_compressionParameters* const cParams,
 
 size_t ZSTD_estimateCCtxSize_usingCCtxParams(const ZSTD_CCtx_params* params)
 {
-    /* Estimate CCtx size is supported for single-threaded compression only. */
-    if (params->nbWorkers > 0) { return ERROR(GENERIC); }
+    RETURN_ERROR_IF(params->nbWorkers > 0, GENERIC, "Estimate CCtx size is supported for single-threaded compression only.");
     {   ZSTD_compressionParameters const cParams =
-                ZSTD_getCParamsFromCCtxParams(params, 0, 0);
+                ZSTD_getCParamsFromCCtxParams(params, ZSTD_CONTENTSIZE_UNKNOWN, 0);
         size_t const blockSize = MIN(ZSTD_BLOCKSIZE_MAX, (size_t)1 << cParams.windowLog);
         U32    const divider = (cParams.minMatch==3) ? 3 : 4;
         size_t const maxNbSeq = blockSize / divider;
-        size_t const tokenSpace = WILDCOPY_OVERLENGTH + blockSize + 11*maxNbSeq;
-        size_t const entropySpace = HUF_WORKSPACE_SIZE;
-        size_t const blockStateSpace = 2 * sizeof(ZSTD_compressedBlockState_t);
+        size_t const tokenSpace = ZSTD_cwksp_alloc_size(WILDCOPY_OVERLENGTH + blockSize)
+                                + ZSTD_cwksp_alloc_size(maxNbSeq * sizeof(seqDef))
+                                + 3 * ZSTD_cwksp_alloc_size(maxNbSeq * sizeof(BYTE));
+        size_t const entropySpace = ZSTD_cwksp_alloc_size(HUF_WORKSPACE_SIZE);
+        size_t const blockStateSpace = 2 * ZSTD_cwksp_alloc_size(sizeof(ZSTD_compressedBlockState_t));
         size_t const matchStateSize = ZSTD_sizeof_matchState(&cParams, /* forCCtx */ 1);
 
         size_t const ldmSpace = ZSTD_ldm_getTableSize(params->ldmParams);
-        size_t const ldmSeqSpace = ZSTD_ldm_getMaxNbSeq(params->ldmParams, blockSize) * sizeof(rawSeq);
-
-        size_t const neededSpace = entropySpace + blockStateSpace + tokenSpace +
-                                   matchStateSize + ldmSpace + ldmSeqSpace;
-
-        DEBUGLOG(5, "sizeof(ZSTD_CCtx) : %u", (U32)sizeof(ZSTD_CCtx));
-        DEBUGLOG(5, "estimate workSpace : %u", (U32)neededSpace);
-        return sizeof(ZSTD_CCtx) + neededSpace;
+        size_t const ldmSeqSpace = ZSTD_cwksp_alloc_size(ZSTD_ldm_getMaxNbSeq(params->ldmParams, blockSize) * sizeof(rawSeq));
+
+        /* estimateCCtxSize is for one-shot compression. So no buffers should
+         * be needed. However, we still allocate two 0-sized buffers, which can
+         * take space under ASAN. */
+        size_t const bufferSpace = ZSTD_cwksp_alloc_size(0)
+                                 + ZSTD_cwksp_alloc_size(0);
+
+        size_t const cctxSpace = ZSTD_cwksp_alloc_size(sizeof(ZSTD_CCtx));
+
+        size_t const neededSpace =
+            cctxSpace +
+            entropySpace +
+            blockStateSpace +
+            ldmSpace +
+            ldmSeqSpace +
+            matchStateSize +
+            tokenSpace +
+            bufferSpace;
+
+        DEBUGLOG(5, "estimate workspace : %u", (U32)neededSpace);
+        return neededSpace;
     }
 }
 
@@ -1005,7 +1172,7 @@ size_t ZSTD_estimateCCtxSize_usingCParams(ZSTD_compressionParameters cParams)
 
 static size_t ZSTD_estimateCCtxSize_internal(int compressionLevel)
 {
-    ZSTD_compressionParameters const cParams = ZSTD_getCParams(compressionLevel, 0, 0);
+    ZSTD_compressionParameters const cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, 0);
     return ZSTD_estimateCCtxSize_usingCParams(cParams);
 }
 
@@ -1022,12 +1189,15 @@ size_t ZSTD_estimateCCtxSize(int compressionLevel)
 
 size_t ZSTD_estimateCStreamSize_usingCCtxParams(const ZSTD_CCtx_params* params)
 {
-    if (params->nbWorkers > 0) { return ERROR(GENERIC); }
-    {   size_t const CCtxSize = ZSTD_estimateCCtxSize_usingCCtxParams(params);
-        size_t const blockSize = MIN(ZSTD_BLOCKSIZE_MAX, (size_t)1 << params->cParams.windowLog);
-        size_t const inBuffSize = ((size_t)1 << params->cParams.windowLog) + blockSize;
+    RETURN_ERROR_IF(params->nbWorkers > 0, GENERIC, "Estimate CCtx size is supported for single-threaded compression only.");
+    {   ZSTD_compressionParameters const cParams =
+                ZSTD_getCParamsFromCCtxParams(params, ZSTD_CONTENTSIZE_UNKNOWN, 0);
+        size_t const CCtxSize = ZSTD_estimateCCtxSize_usingCCtxParams(params);
+        size_t const blockSize = MIN(ZSTD_BLOCKSIZE_MAX, (size_t)1 << cParams.windowLog);
+        size_t const inBuffSize = ((size_t)1 << cParams.windowLog) + blockSize;
         size_t const outBuffSize = ZSTD_compressBound(blockSize) + 1;
-        size_t const streamingSize = inBuffSize + outBuffSize;
+        size_t const streamingSize = ZSTD_cwksp_alloc_size(inBuffSize)
+                                   + ZSTD_cwksp_alloc_size(outBuffSize);
 
         return CCtxSize + streamingSize;
     }
@@ -1041,7 +1211,7 @@ size_t ZSTD_estimateCStreamSize_usingCParams(ZSTD_compressionParameters cParams)
 
 static size_t ZSTD_estimateCStreamSize_internal(int compressionLevel)
 {
-    ZSTD_compressionParameters const cParams = ZSTD_getCParams(compressionLevel, 0, 0);
+    ZSTD_compressionParameters const cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, 0);
     return ZSTD_estimateCStreamSize_usingCParams(cParams);
 }
 
@@ -1095,17 +1265,6 @@ size_t ZSTD_toFlushNow(ZSTD_CCtx* cctx)
     return 0;   /* over-simplification; could also check if context is currently running in streaming mode, and in which case, report how many bytes are left to be flushed within output buffer */
 }
 
-
-
-static U32 ZSTD_equivalentCParams(ZSTD_compressionParameters cParams1,
-                                  ZSTD_compressionParameters cParams2)
-{
-    return (cParams1.hashLog  == cParams2.hashLog)
-         & (cParams1.chainLog == cParams2.chainLog)
-         & (cParams1.strategy == cParams2.strategy)   /* opt parser space */
-         & ((cParams1.minMatch==3) == (cParams2.minMatch==3));  /* hashlog3 space */
-}
-
 static void ZSTD_assertEqualCParams(ZSTD_compressionParameters cParams1,
                                     ZSTD_compressionParameters cParams2)
 {
@@ -1120,72 +1279,7 @@ static void ZSTD_assertEqualCParams(ZSTD_compressionParameters cParams1,
     assert(cParams1.strategy     == cParams2.strategy);
 }
 
-/** The parameters are equivalent if ldm is not enabled in both sets or
- *  all the parameters are equivalent. */
-static U32 ZSTD_equivalentLdmParams(ldmParams_t ldmParams1,
-                                    ldmParams_t ldmParams2)
-{
-    return (!ldmParams1.enableLdm && !ldmParams2.enableLdm) ||
-           (ldmParams1.enableLdm == ldmParams2.enableLdm &&
-            ldmParams1.hashLog == ldmParams2.hashLog &&
-            ldmParams1.bucketSizeLog == ldmParams2.bucketSizeLog &&
-            ldmParams1.minMatchLength == ldmParams2.minMatchLength &&
-            ldmParams1.hashRateLog == ldmParams2.hashRateLog);
-}
-
-typedef enum { ZSTDb_not_buffered, ZSTDb_buffered } ZSTD_buffered_policy_e;
-
-/* ZSTD_sufficientBuff() :
- * check internal buffers exist for streaming if buffPol == ZSTDb_buffered .
- * Note : they are assumed to be correctly sized if ZSTD_equivalentCParams()==1 */
-static U32 ZSTD_sufficientBuff(size_t bufferSize1, size_t maxNbSeq1,
-                            size_t maxNbLit1,
-                            ZSTD_buffered_policy_e buffPol2,
-                            ZSTD_compressionParameters cParams2,
-                            U64 pledgedSrcSize)
-{
-    size_t const windowSize2 = MAX(1, (size_t)MIN(((U64)1 << cParams2.windowLog), pledgedSrcSize));
-    size_t const blockSize2 = MIN(ZSTD_BLOCKSIZE_MAX, windowSize2);
-    size_t const maxNbSeq2 = blockSize2 / ((cParams2.minMatch == 3) ? 3 : 4);
-    size_t const maxNbLit2 = blockSize2;
-    size_t const neededBufferSize2 = (buffPol2==ZSTDb_buffered) ? windowSize2 + blockSize2 : 0;
-    DEBUGLOG(4, "ZSTD_sufficientBuff: is neededBufferSize2=%u <= bufferSize1=%u",
-                (U32)neededBufferSize2, (U32)bufferSize1);
-    DEBUGLOG(4, "ZSTD_sufficientBuff: is maxNbSeq2=%u <= maxNbSeq1=%u",
-                (U32)maxNbSeq2, (U32)maxNbSeq1);
-    DEBUGLOG(4, "ZSTD_sufficientBuff: is maxNbLit2=%u <= maxNbLit1=%u",
-                (U32)maxNbLit2, (U32)maxNbLit1);
-    return (maxNbLit2 <= maxNbLit1)
-         & (maxNbSeq2 <= maxNbSeq1)
-         & (neededBufferSize2 <= bufferSize1);
-}
-
-/** Equivalence for resetCCtx purposes */
-static U32 ZSTD_equivalentParams(ZSTD_CCtx_params params1,
-                                 ZSTD_CCtx_params params2,
-                                 size_t buffSize1,
-                                 size_t maxNbSeq1, size_t maxNbLit1,
-                                 ZSTD_buffered_policy_e buffPol2,
-                                 U64 pledgedSrcSize)
-{
-    DEBUGLOG(4, "ZSTD_equivalentParams: pledgedSrcSize=%u", (U32)pledgedSrcSize);
-    if (!ZSTD_equivalentCParams(params1.cParams, params2.cParams)) {
-      DEBUGLOG(4, "ZSTD_equivalentCParams() == 0");
-      return 0;
-    }
-    if (!ZSTD_equivalentLdmParams(params1.ldmParams, params2.ldmParams)) {
-      DEBUGLOG(4, "ZSTD_equivalentLdmParams() == 0");
-      return 0;
-    }
-    if (!ZSTD_sufficientBuff(buffSize1, maxNbSeq1, maxNbLit1, buffPol2,
-                             params2.cParams, pledgedSrcSize)) {
-      DEBUGLOG(4, "ZSTD_sufficientBuff() == 0");
-      return 0;
-    }
-    return 1;
-}
-
-static void ZSTD_reset_compressedBlockState(ZSTD_compressedBlockState_t* bs)
+void ZSTD_reset_compressedBlockState(ZSTD_compressedBlockState_t* bs)
 {
     int i;
     for (i = 0; i < ZSTD_REP_NUM; ++i)
@@ -1197,132 +1291,143 @@ static void ZSTD_reset_compressedBlockState(ZSTD_compressedBlockState_t* bs)
 }
 
 /*! ZSTD_invalidateMatchState()
- * Invalidate all the matches in the match finder tables.
- * Requires nextSrc and base to be set (can be NULL).
+ *  Invalidate all the matches in the match finder tables.
+ *  Requires nextSrc and base to be set (can be NULL).
  */
 static void ZSTD_invalidateMatchState(ZSTD_matchState_t* ms)
 {
     ZSTD_window_clear(&ms->window);
 
     ms->nextToUpdate = ms->window.dictLimit;
-    ms->nextToUpdate3 = ms->window.dictLimit;
     ms->loadedDictEnd = 0;
     ms->opt.litLengthSum = 0;  /* force reset of btopt stats */
     ms->dictMatchState = NULL;
 }
 
-/*! ZSTD_continueCCtx() :
- *  reuse CCtx without reset (note : requires no dictionary) */
-static size_t ZSTD_continueCCtx(ZSTD_CCtx* cctx, ZSTD_CCtx_params params, U64 pledgedSrcSize)
-{
-    size_t const windowSize = MAX(1, (size_t)MIN(((U64)1 << params.cParams.windowLog), pledgedSrcSize));
-    size_t const blockSize = MIN(ZSTD_BLOCKSIZE_MAX, windowSize);
-    DEBUGLOG(4, "ZSTD_continueCCtx: re-use context in place");
+/**
+ * Indicates whether this compression proceeds directly from user-provided
+ * source buffer to user-provided destination buffer (ZSTDb_not_buffered), or
+ * whether the context needs to buffer the input/output (ZSTDb_buffered).
+ */
+typedef enum {
+    ZSTDb_not_buffered,
+    ZSTDb_buffered
+} ZSTD_buffered_policy_e;
 
-    cctx->blockSize = blockSize;   /* previous block size could be different even for same windowLog, due to pledgedSrcSize */
-    cctx->appliedParams = params;
-    cctx->blockState.matchState.cParams = params.cParams;
-    cctx->pledgedSrcSizePlusOne = pledgedSrcSize+1;
-    cctx->consumedSrcSize = 0;
-    cctx->producedCSize = 0;
-    if (pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN)
-        cctx->appliedParams.fParams.contentSizeFlag = 0;
-    DEBUGLOG(4, "pledged content size : %u ; flag : %u",
-        (U32)pledgedSrcSize, cctx->appliedParams.fParams.contentSizeFlag);
-    cctx->stage = ZSTDcs_init;
-    cctx->dictID = 0;
-    if (params.ldmParams.enableLdm)
-        ZSTD_window_clear(&cctx->ldmState.window);
-    ZSTD_referenceExternalSequences(cctx, NULL, 0);
-    ZSTD_invalidateMatchState(&cctx->blockState.matchState);
-    ZSTD_reset_compressedBlockState(cctx->blockState.prevCBlock);
-    XXH64_reset(&cctx->xxhState, 0);
-    return 0;
-}
+/**
+ * Controls, for this matchState reset, whether the tables need to be cleared /
+ * prepared for the coming compression (ZSTDcrp_makeClean), or whether the
+ * tables can be left unclean (ZSTDcrp_leaveDirty), because we know that a
+ * subsequent operation will overwrite the table space anyways (e.g., copying
+ * the matchState contents in from a CDict).
+ */
+typedef enum {
+    ZSTDcrp_makeClean,
+    ZSTDcrp_leaveDirty
+} ZSTD_compResetPolicy_e;
+
+/**
+ * Controls, for this matchState reset, whether indexing can continue where it
+ * left off (ZSTDirp_continue), or whether it needs to be restarted from zero
+ * (ZSTDirp_reset).
+ */
+typedef enum {
+    ZSTDirp_continue,
+    ZSTDirp_reset
+} ZSTD_indexResetPolicy_e;
 
-typedef enum { ZSTDcrp_continue, ZSTDcrp_noMemset } ZSTD_compResetPolicy_e;
+typedef enum {
+    ZSTD_resetTarget_CDict,
+    ZSTD_resetTarget_CCtx
+} ZSTD_resetTarget_e;
 
-static void*
+static size_t
 ZSTD_reset_matchState(ZSTD_matchState_t* ms,
-                      void* ptr,
+                      ZSTD_cwksp* ws,
                 const ZSTD_compressionParameters* cParams,
-                      ZSTD_compResetPolicy_e const crp, U32 const forCCtx)
+                const ZSTD_compResetPolicy_e crp,
+                const ZSTD_indexResetPolicy_e forceResetIndex,
+                const ZSTD_resetTarget_e forWho)
 {
     size_t const chainSize = (cParams->strategy == ZSTD_fast) ? 0 : ((size_t)1 << cParams->chainLog);
     size_t const hSize = ((size_t)1) << cParams->hashLog;
-    U32    const hashLog3 = (forCCtx && cParams->minMatch==3) ? MIN(ZSTD_HASHLOG3_MAX, cParams->windowLog) : 0;
-    size_t const h3Size = ((size_t)1) << hashLog3;
-    size_t const tableSpace = (chainSize + hSize + h3Size) * sizeof(U32);
+    U32    const hashLog3 = ((forWho == ZSTD_resetTarget_CCtx) && cParams->minMatch==3) ? MIN(ZSTD_HASHLOG3_MAX, cParams->windowLog) : 0;
+    size_t const h3Size = hashLog3 ? ((size_t)1) << hashLog3 : 0;
 
-    assert(((size_t)ptr & 3) == 0);
+    DEBUGLOG(4, "reset indices : %u", forceResetIndex == ZSTDirp_reset);
+    if (forceResetIndex == ZSTDirp_reset) {
+        ZSTD_window_init(&ms->window);
+        ZSTD_cwksp_mark_tables_dirty(ws);
+    }
 
     ms->hashLog3 = hashLog3;
-    memset(&ms->window, 0, sizeof(ms->window));
-    ms->window.dictLimit = 1;    /* start from 1, so that 1st position is valid */
-    ms->window.lowLimit = 1;     /* it ensures first and later CCtx usages compress the same */
-    ms->window.nextSrc = ms->window.base + 1;   /* see issue #1241 */
+
     ZSTD_invalidateMatchState(ms);
 
+    assert(!ZSTD_cwksp_reserve_failed(ws)); /* check that allocation hasn't already failed */
+
+    ZSTD_cwksp_clear_tables(ws);
+
+    DEBUGLOG(5, "reserving table space");
+    /* table Space */
+    ms->hashTable = (U32*)ZSTD_cwksp_reserve_table(ws, hSize * sizeof(U32));
+    ms->chainTable = (U32*)ZSTD_cwksp_reserve_table(ws, chainSize * sizeof(U32));
+    ms->hashTable3 = (U32*)ZSTD_cwksp_reserve_table(ws, h3Size * sizeof(U32));
+    RETURN_ERROR_IF(ZSTD_cwksp_reserve_failed(ws), memory_allocation,
+                    "failed a workspace allocation in ZSTD_reset_matchState");
+
+    DEBUGLOG(4, "reset table : %u", crp!=ZSTDcrp_leaveDirty);
+    if (crp!=ZSTDcrp_leaveDirty) {
+        /* reset tables only */
+        ZSTD_cwksp_clean_tables(ws);
+    }
+
     /* opt parser space */
-    if (forCCtx && (cParams->strategy >= ZSTD_btopt)) {
+    if ((forWho == ZSTD_resetTarget_CCtx) && (cParams->strategy >= ZSTD_btopt)) {
         DEBUGLOG(4, "reserving optimal parser space");
-        ms->opt.litFreq = (unsigned*)ptr;
-        ms->opt.litLengthFreq = ms->opt.litFreq + (1<<Litbits);
-        ms->opt.matchLengthFreq = ms->opt.litLengthFreq + (MaxLL+1);
-        ms->opt.offCodeFreq = ms->opt.matchLengthFreq + (MaxML+1);
-        ptr = ms->opt.offCodeFreq + (MaxOff+1);
-        ms->opt.matchTable = (ZSTD_match_t*)ptr;
-        ptr = ms->opt.matchTable + ZSTD_OPT_NUM+1;
-        ms->opt.priceTable = (ZSTD_optimal_t*)ptr;
-        ptr = ms->opt.priceTable + ZSTD_OPT_NUM+1;
+        ms->opt.litFreq = (unsigned*)ZSTD_cwksp_reserve_aligned(ws, (1<<Litbits) * sizeof(unsigned));
+        ms->opt.litLengthFreq = (unsigned*)ZSTD_cwksp_reserve_aligned(ws, (MaxLL+1) * sizeof(unsigned));
+        ms->opt.matchLengthFreq = (unsigned*)ZSTD_cwksp_reserve_aligned(ws, (MaxML+1) * sizeof(unsigned));
+        ms->opt.offCodeFreq = (unsigned*)ZSTD_cwksp_reserve_aligned(ws, (MaxOff+1) * sizeof(unsigned));
+        ms->opt.matchTable = (ZSTD_match_t*)ZSTD_cwksp_reserve_aligned(ws, (ZSTD_OPT_NUM+1) * sizeof(ZSTD_match_t));
+        ms->opt.priceTable = (ZSTD_optimal_t*)ZSTD_cwksp_reserve_aligned(ws, (ZSTD_OPT_NUM+1) * sizeof(ZSTD_optimal_t));
     }
 
-    /* table Space */
-    DEBUGLOG(4, "reset table : %u", crp!=ZSTDcrp_noMemset);
-    assert(((size_t)ptr & 3) == 0);  /* ensure ptr is properly aligned */
-    if (crp!=ZSTDcrp_noMemset) memset(ptr, 0, tableSpace);   /* reset tables only */
-    ms->hashTable = (U32*)(ptr);
-    ms->chainTable = ms->hashTable + hSize;
-    ms->hashTable3 = ms->chainTable + chainSize;
-    ptr = ms->hashTable3 + h3Size;
-
     ms->cParams = *cParams;
 
-    assert(((size_t)ptr & 3) == 0);
-    return ptr;
+    RETURN_ERROR_IF(ZSTD_cwksp_reserve_failed(ws), memory_allocation,
+                    "failed a workspace allocation in ZSTD_reset_matchState");
+
+    return 0;
 }
 
-#define ZSTD_WORKSPACETOOLARGE_FACTOR 3 /* define "workspace is too large" as this number of times larger than needed */
-#define ZSTD_WORKSPACETOOLARGE_MAXDURATION 128  /* when workspace is continuously too large
-                                         * during at least this number of times,
-                                         * context's memory usage is considered wasteful,
-                                         * because it's sized to handle a worst case scenario which rarely happens.
-                                         * In which case, resize it down to free some memory */
+/* ZSTD_indexTooCloseToMax() :
+ * minor optimization : prefer memset() rather than reduceIndex()
+ * which is measurably slow in some circumstances (reported for Visual Studio).
+ * Works when re-using a context for a lot of smallish inputs :
+ * if all inputs are smaller than ZSTD_INDEXOVERFLOW_MARGIN,
+ * memset() will be triggered before reduceIndex().
+ */
+#define ZSTD_INDEXOVERFLOW_MARGIN (16 MB)
+static int ZSTD_indexTooCloseToMax(ZSTD_window_t w)
+{
+    return (size_t)(w.nextSrc - w.base) > (ZSTD_CURRENT_MAX - ZSTD_INDEXOVERFLOW_MARGIN);
+}
 
 /*! ZSTD_resetCCtx_internal() :
     note : `params` are assumed fully validated at this stage */
 static size_t ZSTD_resetCCtx_internal(ZSTD_CCtx* zc,
                                       ZSTD_CCtx_params params,
-                                      U64 pledgedSrcSize,
+                                      U64 const pledgedSrcSize,
                                       ZSTD_compResetPolicy_e const crp,
                                       ZSTD_buffered_policy_e const zbuff)
 {
+    ZSTD_cwksp* const ws = &zc->workspace;
     DEBUGLOG(4, "ZSTD_resetCCtx_internal: pledgedSrcSize=%u, wlog=%u",
                 (U32)pledgedSrcSize, params.cParams.windowLog);
     assert(!ZSTD_isError(ZSTD_checkCParams(params.cParams)));
 
-    if (crp == ZSTDcrp_continue) {
-        if (ZSTD_equivalentParams(zc->appliedParams, params,
-                                  zc->inBuffSize,
-                                  zc->seqStore.maxNbSeq, zc->seqStore.maxNbLit,
-                                  zbuff, pledgedSrcSize)) {
-            DEBUGLOG(4, "ZSTD_equivalentParams()==1 -> continue mode (wLog1=%u, blockSize1=%zu)",
-                        zc->appliedParams.cParams.windowLog, zc->blockSize);
-            zc->workSpaceOversizedDuration += (zc->workSpaceOversizedDuration > 0);   /* if it was too large, it still is */
-            if (zc->workSpaceOversizedDuration <= ZSTD_WORKSPACETOOLARGE_MAXDURATION)
-                return ZSTD_continueCCtx(zc, params, pledgedSrcSize);
-    }   }
-    DEBUGLOG(4, "ZSTD_equivalentParams()==0 -> reset CCtx");
+    zc->isFirstBlock = 1;
 
     if (params.ldmParams.enableLdm) {
         /* Adjust long distance matching parameters */
@@ -1336,58 +1441,74 @@ static size_t ZSTD_resetCCtx_internal(ZSTD_CCtx* zc,
         size_t const blockSize = MIN(ZSTD_BLOCKSIZE_MAX, windowSize);
         U32    const divider = (params.cParams.minMatch==3) ? 3 : 4;
         size_t const maxNbSeq = blockSize / divider;
-        size_t const tokenSpace = WILDCOPY_OVERLENGTH + blockSize + 11*maxNbSeq;
+        size_t const tokenSpace = ZSTD_cwksp_alloc_size(WILDCOPY_OVERLENGTH + blockSize)
+                                + ZSTD_cwksp_alloc_size(maxNbSeq * sizeof(seqDef))
+                                + 3 * ZSTD_cwksp_alloc_size(maxNbSeq * sizeof(BYTE));
         size_t const buffOutSize = (zbuff==ZSTDb_buffered) ? ZSTD_compressBound(blockSize)+1 : 0;
         size_t const buffInSize = (zbuff==ZSTDb_buffered) ? windowSize + blockSize : 0;
         size_t const matchStateSize = ZSTD_sizeof_matchState(&params.cParams, /* forCCtx */ 1);
         size_t const maxNbLdmSeq = ZSTD_ldm_getMaxNbSeq(params.ldmParams, blockSize);
-        void* ptr;   /* used to partition workSpace */
 
-        /* Check if workSpace is large enough, alloc a new one if needed */
-        {   size_t const entropySpace = HUF_WORKSPACE_SIZE;
-            size_t const blockStateSpace = 2 * sizeof(ZSTD_compressedBlockState_t);
-            size_t const bufferSpace = buffInSize + buffOutSize;
+        ZSTD_indexResetPolicy_e needsIndexReset = zc->initialized ? ZSTDirp_continue : ZSTDirp_reset;
+
+        if (ZSTD_indexTooCloseToMax(zc->blockState.matchState.window)) {
+            needsIndexReset = ZSTDirp_reset;
+        }
+
+        if (!zc->staticSize) ZSTD_cwksp_bump_oversized_duration(ws, 0);
+
+        /* Check if workspace is large enough, alloc a new one if needed */
+        {   size_t const cctxSpace = zc->staticSize ? ZSTD_cwksp_alloc_size(sizeof(ZSTD_CCtx)) : 0;
+            size_t const entropySpace = ZSTD_cwksp_alloc_size(HUF_WORKSPACE_SIZE);
+            size_t const blockStateSpace = 2 * ZSTD_cwksp_alloc_size(sizeof(ZSTD_compressedBlockState_t));
+            size_t const bufferSpace = ZSTD_cwksp_alloc_size(buffInSize) + ZSTD_cwksp_alloc_size(buffOutSize);
             size_t const ldmSpace = ZSTD_ldm_getTableSize(params.ldmParams);
-            size_t const ldmSeqSpace = maxNbLdmSeq * sizeof(rawSeq);
+            size_t const ldmSeqSpace = ZSTD_cwksp_alloc_size(maxNbLdmSeq * sizeof(rawSeq));
 
-            size_t const neededSpace = entropySpace + blockStateSpace + ldmSpace +
-                                       ldmSeqSpace + matchStateSize + tokenSpace +
-                                       bufferSpace;
+            size_t const neededSpace =
+                cctxSpace +
+                entropySpace +
+                blockStateSpace +
+                ldmSpace +
+                ldmSeqSpace +
+                matchStateSize +
+                tokenSpace +
+                bufferSpace;
 
-            int const workSpaceTooSmall = zc->workSpaceSize < neededSpace;
-            int const workSpaceTooLarge = zc->workSpaceSize > ZSTD_WORKSPACETOOLARGE_FACTOR * neededSpace;
-            int const workSpaceWasteful = workSpaceTooLarge && (zc->workSpaceOversizedDuration > ZSTD_WORKSPACETOOLARGE_MAXDURATION);
-            zc->workSpaceOversizedDuration = workSpaceTooLarge ? zc->workSpaceOversizedDuration+1 : 0;
+            int const workspaceTooSmall = ZSTD_cwksp_sizeof(ws) < neededSpace;
+            int const workspaceWasteful = ZSTD_cwksp_check_wasteful(ws, neededSpace);
 
             DEBUGLOG(4, "Need %zuKB workspace, including %zuKB for match state, and %zuKB for buffers",
                         neededSpace>>10, matchStateSize>>10, bufferSpace>>10);
             DEBUGLOG(4, "windowSize: %zu - blockSize: %zu", windowSize, blockSize);
 
-            if (workSpaceTooSmall || workSpaceWasteful) {
-                DEBUGLOG(4, "Need to resize workSpaceSize from %zuKB to %zuKB",
-                            zc->workSpaceSize >> 10,
+            if (workspaceTooSmall || workspaceWasteful) {
+                DEBUGLOG(4, "Resize workspaceSize from %zuKB to %zuKB",
+                            ZSTD_cwksp_sizeof(ws) >> 10,
                             neededSpace >> 10);
-                /* static cctx : no resize, error out */
-                if (zc->staticSize) return ERROR(memory_allocation);
 
-                zc->workSpaceSize = 0;
-                ZSTD_free(zc->workSpace, zc->customMem);
-                zc->workSpace = ZSTD_malloc(neededSpace, zc->customMem);
-                if (zc->workSpace == NULL) return ERROR(memory_allocation);
-                zc->workSpaceSize = neededSpace;
-                zc->workSpaceOversizedDuration = 0;
+                RETURN_ERROR_IF(zc->staticSize, memory_allocation, "static cctx : no resize");
+
+                needsIndexReset = ZSTDirp_reset;
+
+                ZSTD_cwksp_free(ws, zc->customMem);
+                FORWARD_IF_ERROR(ZSTD_cwksp_create(ws, neededSpace, zc->customMem), "");
 
+                DEBUGLOG(5, "reserving object space");
                 /* Statically sized space.
                  * entropyWorkspace never moves,
                  * though prev/next block swap places */
-                assert(((size_t)zc->workSpace & 3) == 0);   /* ensure correct alignment */
-                assert(zc->workSpaceSize >= 2 * sizeof(ZSTD_compressedBlockState_t));
-                zc->blockState.prevCBlock = (ZSTD_compressedBlockState_t*)zc->workSpace;
-                zc->blockState.nextCBlock = zc->blockState.prevCBlock + 1;
-                ptr = zc->blockState.nextCBlock + 1;
-                zc->entropyWorkspace = (U32*)ptr;
+                assert(ZSTD_cwksp_check_available(ws, 2 * sizeof(ZSTD_compressedBlockState_t)));
+                zc->blockState.prevCBlock = (ZSTD_compressedBlockState_t*) ZSTD_cwksp_reserve_object(ws, sizeof(ZSTD_compressedBlockState_t));
+                RETURN_ERROR_IF(zc->blockState.prevCBlock == NULL, memory_allocation, "couldn't allocate prevCBlock");
+                zc->blockState.nextCBlock = (ZSTD_compressedBlockState_t*) ZSTD_cwksp_reserve_object(ws, sizeof(ZSTD_compressedBlockState_t));
+                RETURN_ERROR_IF(zc->blockState.nextCBlock == NULL, memory_allocation, "couldn't allocate nextCBlock");
+                zc->entropyWorkspace = (U32*) ZSTD_cwksp_reserve_object(ws, HUF_WORKSPACE_SIZE);
+                RETURN_ERROR_IF(zc->blockState.nextCBlock == NULL, memory_allocation, "couldn't allocate entropyWorkspace");
         }   }
 
+        ZSTD_cwksp_clear(ws);
+
         /* init params */
         zc->appliedParams = params;
         zc->blockState.matchState.cParams = params.cParams;
@@ -1406,57 +1527,60 @@ static size_t ZSTD_resetCCtx_internal(ZSTD_CCtx* zc,
 
         ZSTD_reset_compressedBlockState(zc->blockState.prevCBlock);
 
-        ptr = zc->entropyWorkspace + HUF_WORKSPACE_SIZE_U32;
-
-        /* ldm hash table */
-        /* initialize bucketOffsets table later for pointer alignment */
-        if (params.ldmParams.enableLdm) {
-            size_t const ldmHSize = ((size_t)1) << params.ldmParams.hashLog;
-            memset(ptr, 0, ldmHSize * sizeof(ldmEntry_t));
-            assert(((size_t)ptr & 3) == 0); /* ensure ptr is properly aligned */
-            zc->ldmState.hashTable = (ldmEntry_t*)ptr;
-            ptr = zc->ldmState.hashTable + ldmHSize;
-            zc->ldmSequences = (rawSeq*)ptr;
-            ptr = zc->ldmSequences + maxNbLdmSeq;
-            zc->maxNbLdmSequences = maxNbLdmSeq;
-
-            memset(&zc->ldmState.window, 0, sizeof(zc->ldmState.window));
-        }
-        assert(((size_t)ptr & 3) == 0); /* ensure ptr is properly aligned */
-
-        ptr = ZSTD_reset_matchState(&zc->blockState.matchState, ptr, &params.cParams, crp, /* forCCtx */ 1);
-
-        /* sequences storage */
-        zc->seqStore.maxNbSeq = maxNbSeq;
-        zc->seqStore.sequencesStart = (seqDef*)ptr;
-        ptr = zc->seqStore.sequencesStart + maxNbSeq;
-        zc->seqStore.llCode = (BYTE*) ptr;
-        zc->seqStore.mlCode = zc->seqStore.llCode + maxNbSeq;
-        zc->seqStore.ofCode = zc->seqStore.mlCode + maxNbSeq;
-        zc->seqStore.litStart = zc->seqStore.ofCode + maxNbSeq;
         /* ZSTD_wildcopy() is used to copy into the literals buffer,
          * so we have to oversize the buffer by WILDCOPY_OVERLENGTH bytes.
          */
+        zc->seqStore.litStart = ZSTD_cwksp_reserve_buffer(ws, blockSize + WILDCOPY_OVERLENGTH);
         zc->seqStore.maxNbLit = blockSize;
-        ptr = zc->seqStore.litStart + blockSize + WILDCOPY_OVERLENGTH;
+
+        /* buffers */
+        zc->inBuffSize = buffInSize;
+        zc->inBuff = (char*)ZSTD_cwksp_reserve_buffer(ws, buffInSize);
+        zc->outBuffSize = buffOutSize;
+        zc->outBuff = (char*)ZSTD_cwksp_reserve_buffer(ws, buffOutSize);
 
         /* ldm bucketOffsets table */
         if (params.ldmParams.enableLdm) {
+            /* TODO: avoid memset? */
             size_t const ldmBucketSize =
                   ((size_t)1) << (params.ldmParams.hashLog -
                                   params.ldmParams.bucketSizeLog);
-            memset(ptr, 0, ldmBucketSize);
-            zc->ldmState.bucketOffsets = (BYTE*)ptr;
-            ptr = zc->ldmState.bucketOffsets + ldmBucketSize;
-            ZSTD_window_clear(&zc->ldmState.window);
+            zc->ldmState.bucketOffsets = ZSTD_cwksp_reserve_buffer(ws, ldmBucketSize);
+            memset(zc->ldmState.bucketOffsets, 0, ldmBucketSize);
         }
+
+        /* sequences storage */
         ZSTD_referenceExternalSequences(zc, NULL, 0);
+        zc->seqStore.maxNbSeq = maxNbSeq;
+        zc->seqStore.llCode = ZSTD_cwksp_reserve_buffer(ws, maxNbSeq * sizeof(BYTE));
+        zc->seqStore.mlCode = ZSTD_cwksp_reserve_buffer(ws, maxNbSeq * sizeof(BYTE));
+        zc->seqStore.ofCode = ZSTD_cwksp_reserve_buffer(ws, maxNbSeq * sizeof(BYTE));
+        zc->seqStore.sequencesStart = (seqDef*)ZSTD_cwksp_reserve_aligned(ws, maxNbSeq * sizeof(seqDef));
+
+        FORWARD_IF_ERROR(ZSTD_reset_matchState(
+            &zc->blockState.matchState,
+            ws,
+            &params.cParams,
+            crp,
+            needsIndexReset,
+            ZSTD_resetTarget_CCtx), "");
 
-        /* buffers */
-        zc->inBuffSize = buffInSize;
-        zc->inBuff = (char*)ptr;
-        zc->outBuffSize = buffOutSize;
-        zc->outBuff = zc->inBuff + buffInSize;
+        /* ldm hash table */
+        if (params.ldmParams.enableLdm) {
+            /* TODO: avoid memset? */
+            size_t const ldmHSize = ((size_t)1) << params.ldmParams.hashLog;
+            zc->ldmState.hashTable = (ldmEntry_t*)ZSTD_cwksp_reserve_aligned(ws, ldmHSize * sizeof(ldmEntry_t));
+            memset(zc->ldmState.hashTable, 0, ldmHSize * sizeof(ldmEntry_t));
+            zc->ldmSequences = (rawSeq*)ZSTD_cwksp_reserve_aligned(ws, maxNbLdmSeq * sizeof(rawSeq));
+            zc->maxNbLdmSequences = maxNbLdmSeq;
+
+            ZSTD_window_init(&zc->ldmState.window);
+            ZSTD_window_clear(&zc->ldmState.window);
+            zc->ldmState.loadedDictEnd = 0;
+        }
+
+        DEBUGLOG(3, "wksp: finished allocating, %zd bytes remain available", ZSTD_cwksp_available_space(ws));
+        zc->initialized = 1;
 
         return 0;
     }
@@ -1490,40 +1614,39 @@ static const size_t attachDictSizeCutoffs[ZSTD_STRATEGY_MAX+1] = {
 };
 
 static int ZSTD_shouldAttachDict(const ZSTD_CDict* cdict,
-                                 ZSTD_CCtx_params params,
+                                 const ZSTD_CCtx_params* params,
                                  U64 pledgedSrcSize)
 {
     size_t cutoff = attachDictSizeCutoffs[cdict->matchState.cParams.strategy];
     return ( pledgedSrcSize <= cutoff
           || pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN
-          || params.attachDictPref == ZSTD_dictForceAttach )
-        && params.attachDictPref != ZSTD_dictForceCopy
-        && !params.forceWindow; /* dictMatchState isn't correctly
+          || params->attachDictPref == ZSTD_dictForceAttach )
+        && params->attachDictPref != ZSTD_dictForceCopy
+        && !params->forceWindow; /* dictMatchState isn't correctly
                                  * handled in _enforceMaxDist */
 }
 
-static size_t ZSTD_resetCCtx_byAttachingCDict(
-    ZSTD_CCtx* cctx,
-    const ZSTD_CDict* cdict,
-    ZSTD_CCtx_params params,
-    U64 pledgedSrcSize,
-    ZSTD_buffered_policy_e zbuff)
+static size_t
+ZSTD_resetCCtx_byAttachingCDict(ZSTD_CCtx* cctx,
+                        const ZSTD_CDict* cdict,
+                        ZSTD_CCtx_params params,
+                        U64 pledgedSrcSize,
+                        ZSTD_buffered_policy_e zbuff)
 {
-    {
-        const ZSTD_compressionParameters *cdict_cParams = &cdict->matchState.cParams;
+    {   const ZSTD_compressionParameters* const cdict_cParams = &cdict->matchState.cParams;
         unsigned const windowLog = params.cParams.windowLog;
         assert(windowLog != 0);
         /* Resize working context table params for input only, since the dict
          * has its own tables. */
+        /* pledgeSrcSize == 0 means 0! */
         params.cParams = ZSTD_adjustCParams_internal(*cdict_cParams, pledgedSrcSize, 0);
         params.cParams.windowLog = windowLog;
-        ZSTD_resetCCtx_internal(cctx, params, pledgedSrcSize,
-                                ZSTDcrp_continue, zbuff);
+        FORWARD_IF_ERROR(ZSTD_resetCCtx_internal(cctx, params, pledgedSrcSize,
+                                                 ZSTDcrp_makeClean, zbuff), "");
         assert(cctx->appliedParams.cParams.strategy == cdict_cParams->strategy);
     }
 
-    {
-        const U32 cdictEnd = (U32)( cdict->matchState.window.nextSrc
+    {   const U32 cdictEnd = (U32)( cdict->matchState.window.nextSrc
                                   - cdict->matchState.window.base);
         const U32 cdictLen = cdictEnd - cdict->matchState.window.dictLimit;
         if (cdictLen == 0) {
@@ -1540,9 +1663,9 @@ static size_t ZSTD_resetCCtx_byAttachingCDict(
                     cctx->blockState.matchState.window.base + cdictEnd;
                 ZSTD_window_clear(&cctx->blockState.matchState.window);
             }
+            /* loadedDictEnd is expressed within the referential of the active context */
             cctx->blockState.matchState.loadedDictEnd = cctx->blockState.matchState.window.dictLimit;
-        }
-    }
+    }   }
 
     cctx->dictID = cdict->dictID;
 
@@ -1567,36 +1690,41 @@ static size_t ZSTD_resetCCtx_byCopyingCDict(ZSTD_CCtx* cctx,
         /* Copy only compression parameters related to tables. */
         params.cParams = *cdict_cParams;
         params.cParams.windowLog = windowLog;
-        ZSTD_resetCCtx_internal(cctx, params, pledgedSrcSize,
-                                ZSTDcrp_noMemset, zbuff);
+        FORWARD_IF_ERROR(ZSTD_resetCCtx_internal(cctx, params, pledgedSrcSize,
+                                                 ZSTDcrp_leaveDirty, zbuff), "");
         assert(cctx->appliedParams.cParams.strategy == cdict_cParams->strategy);
         assert(cctx->appliedParams.cParams.hashLog == cdict_cParams->hashLog);
         assert(cctx->appliedParams.cParams.chainLog == cdict_cParams->chainLog);
     }
 
+    ZSTD_cwksp_mark_tables_dirty(&cctx->workspace);
+
     /* copy tables */
     {   size_t const chainSize = (cdict_cParams->strategy == ZSTD_fast) ? 0 : ((size_t)1 << cdict_cParams->chainLog);
         size_t const hSize =  (size_t)1 << cdict_cParams->hashLog;
-        size_t const tableSpace = (chainSize + hSize) * sizeof(U32);
-        assert((U32*)cctx->blockState.matchState.chainTable == (U32*)cctx->blockState.matchState.hashTable + hSize);  /* chainTable must follow hashTable */
-        assert((U32*)cctx->blockState.matchState.hashTable3 == (U32*)cctx->blockState.matchState.chainTable + chainSize);
-        assert((U32*)cdict->matchState.chainTable == (U32*)cdict->matchState.hashTable + hSize);  /* chainTable must follow hashTable */
-        assert((U32*)cdict->matchState.hashTable3 == (U32*)cdict->matchState.chainTable + chainSize);
-        memcpy(cctx->blockState.matchState.hashTable, cdict->matchState.hashTable, tableSpace);   /* presumes all tables follow each other */
+
+        memcpy(cctx->blockState.matchState.hashTable,
+               cdict->matchState.hashTable,
+               hSize * sizeof(U32));
+        memcpy(cctx->blockState.matchState.chainTable,
+               cdict->matchState.chainTable,
+               chainSize * sizeof(U32));
     }
 
     /* Zero the hashTable3, since the cdict never fills it */
-    {   size_t const h3Size = (size_t)1 << cctx->blockState.matchState.hashLog3;
+    {   int const h3log = cctx->blockState.matchState.hashLog3;
+        size_t const h3Size = h3log ? ((size_t)1 << h3log) : 0;
         assert(cdict->matchState.hashLog3 == 0);
         memset(cctx->blockState.matchState.hashTable3, 0, h3Size * sizeof(U32));
     }
 
+    ZSTD_cwksp_mark_tables_clean(&cctx->workspace);
+
     /* copy dictionary offsets */
     {   ZSTD_matchState_t const* srcMatchState = &cdict->matchState;
         ZSTD_matchState_t* dstMatchState = &cctx->blockState.matchState;
         dstMatchState->window       = srcMatchState->window;
         dstMatchState->nextToUpdate = srcMatchState->nextToUpdate;
-        dstMatchState->nextToUpdate3= srcMatchState->nextToUpdate3;
         dstMatchState->loadedDictEnd= srcMatchState->loadedDictEnd;
     }
 
@@ -1613,7 +1741,7 @@ static size_t ZSTD_resetCCtx_byCopyingCDict(ZSTD_CCtx* cctx,
  * in-place. We decide here which strategy to use. */
 static size_t ZSTD_resetCCtx_usingCDict(ZSTD_CCtx* cctx,
                             const ZSTD_CDict* cdict,
-                            ZSTD_CCtx_params params,
+                            const ZSTD_CCtx_params* params,
                             U64 pledgedSrcSize,
                             ZSTD_buffered_policy_e zbuff)
 {
@@ -1623,10 +1751,10 @@ static size_t ZSTD_resetCCtx_usingCDict(ZSTD_CCtx* cctx,
 
     if (ZSTD_shouldAttachDict(cdict, params, pledgedSrcSize)) {
         return ZSTD_resetCCtx_byAttachingCDict(
-            cctx, cdict, params, pledgedSrcSize, zbuff);
+            cctx, cdict, *params, pledgedSrcSize, zbuff);
     } else {
         return ZSTD_resetCCtx_byCopyingCDict(
-            cctx, cdict, params, pledgedSrcSize, zbuff);
+            cctx, cdict, *params, pledgedSrcSize, zbuff);
     }
 }
 
@@ -1644,7 +1772,8 @@ static size_t ZSTD_copyCCtx_internal(ZSTD_CCtx* dstCCtx,
                             ZSTD_buffered_policy_e zbuff)
 {
     DEBUGLOG(5, "ZSTD_copyCCtx_internal");
-    if (srcCCtx->stage!=ZSTDcs_init) return ERROR(stage_wrong);
+    RETURN_ERROR_IF(srcCCtx->stage!=ZSTDcs_init, stage_wrong,
+                    "Can't copy a ctx that's not in init stage.");
 
     memcpy(&dstCCtx->customMem, &srcCCtx->customMem, sizeof(ZSTD_customMem));
     {   ZSTD_CCtx_params params = dstCCtx->requestedParams;
@@ -1652,7 +1781,7 @@ static size_t ZSTD_copyCCtx_internal(ZSTD_CCtx* dstCCtx,
         params.cParams = srcCCtx->appliedParams.cParams;
         params.fParams = fParams;
         ZSTD_resetCCtx_internal(dstCCtx, params, pledgedSrcSize,
-                                ZSTDcrp_noMemset, zbuff);
+                                ZSTDcrp_leaveDirty, zbuff);
         assert(dstCCtx->appliedParams.cParams.windowLog == srcCCtx->appliedParams.cParams.windowLog);
         assert(dstCCtx->appliedParams.cParams.strategy == srcCCtx->appliedParams.cParams.strategy);
         assert(dstCCtx->appliedParams.cParams.hashLog == srcCCtx->appliedParams.cParams.hashLog);
@@ -1660,23 +1789,33 @@ static size_t ZSTD_copyCCtx_internal(ZSTD_CCtx* dstCCtx,
         assert(dstCCtx->blockState.matchState.hashLog3 == srcCCtx->blockState.matchState.hashLog3);
     }
 
+    ZSTD_cwksp_mark_tables_dirty(&dstCCtx->workspace);
+
     /* copy tables */
     {   size_t const chainSize = (srcCCtx->appliedParams.cParams.strategy == ZSTD_fast) ? 0 : ((size_t)1 << srcCCtx->appliedParams.cParams.chainLog);
         size_t const hSize =  (size_t)1 << srcCCtx->appliedParams.cParams.hashLog;
-        size_t const h3Size = (size_t)1 << srcCCtx->blockState.matchState.hashLog3;
-        size_t const tableSpace = (chainSize + hSize + h3Size) * sizeof(U32);
-        assert((U32*)dstCCtx->blockState.matchState.chainTable == (U32*)dstCCtx->blockState.matchState.hashTable + hSize);  /* chainTable must follow hashTable */
-        assert((U32*)dstCCtx->blockState.matchState.hashTable3 == (U32*)dstCCtx->blockState.matchState.chainTable + chainSize);
-        memcpy(dstCCtx->blockState.matchState.hashTable, srcCCtx->blockState.matchState.hashTable, tableSpace);   /* presumes all tables follow each other */
+        int const h3log = srcCCtx->blockState.matchState.hashLog3;
+        size_t const h3Size = h3log ? ((size_t)1 << h3log) : 0;
+
+        memcpy(dstCCtx->blockState.matchState.hashTable,
+               srcCCtx->blockState.matchState.hashTable,
+               hSize * sizeof(U32));
+        memcpy(dstCCtx->blockState.matchState.chainTable,
+               srcCCtx->blockState.matchState.chainTable,
+               chainSize * sizeof(U32));
+        memcpy(dstCCtx->blockState.matchState.hashTable3,
+               srcCCtx->blockState.matchState.hashTable3,
+               h3Size * sizeof(U32));
     }
 
+    ZSTD_cwksp_mark_tables_clean(&dstCCtx->workspace);
+
     /* copy dictionary offsets */
     {
         const ZSTD_matchState_t* srcMatchState = &srcCCtx->blockState.matchState;
         ZSTD_matchState_t* dstMatchState = &dstCCtx->blockState.matchState;
         dstMatchState->window       = srcMatchState->window;
         dstMatchState->nextToUpdate = srcMatchState->nextToUpdate;
-        dstMatchState->nextToUpdate3= srcMatchState->nextToUpdate3;
         dstMatchState->loadedDictEnd= srcMatchState->loadedDictEnd;
     }
     dstCCtx->dictID = srcCCtx->dictID;
@@ -1721,6 +1860,20 @@ ZSTD_reduceTable_internal (U32* const table, U32 const size, U32 const reducerVa
     int rowNb;
     assert((size & (ZSTD_ROWSIZE-1)) == 0);  /* multiple of ZSTD_ROWSIZE */
     assert(size < (1U<<31));   /* can be casted to int */
+
+#if defined (MEMORY_SANITIZER) && !defined (ZSTD_MSAN_DONT_POISON_WORKSPACE)
+    /* To validate that the table re-use logic is sound, and that we don't
+     * access table space that we haven't cleaned, we re-"poison" the table
+     * space every time we mark it dirty.
+     *
+     * This function however is intended to operate on those dirty tables and
+     * re-clean them. So when this function is used correctly, we can unpoison
+     * the memory it operated on. This introduces a blind spot though, since
+     * if we now try to operate on __actually__ poisoned memory, we will not
+     * detect that. */
+    __msan_unpoison(table, size * sizeof(U32));
+#endif
+
     for (rowNb=0 ; rowNb < nbRows ; rowNb++) {
         int column;
         for (column=0; column<ZSTD_ROWSIZE; column++) {
@@ -1746,16 +1899,15 @@ static void ZSTD_reduceTable_btlazy2(U32* const table, U32 const size, U32 const
 
 /*! ZSTD_reduceIndex() :
 *   rescale all indexes to avoid future overflow (indexes are U32) */
-static void ZSTD_reduceIndex (ZSTD_CCtx* zc, const U32 reducerValue)
+static void ZSTD_reduceIndex (ZSTD_matchState_t* ms, ZSTD_CCtx_params const* params, const U32 reducerValue)
 {
-    ZSTD_matchState_t* const ms = &zc->blockState.matchState;
-    {   U32 const hSize = (U32)1 << zc->appliedParams.cParams.hashLog;
+    {   U32 const hSize = (U32)1 << params->cParams.hashLog;
         ZSTD_reduceTable(ms->hashTable, hSize, reducerValue);
     }
 
-    if (zc->appliedParams.cParams.strategy != ZSTD_fast) {
-        U32 const chainSize = (U32)1 << zc->appliedParams.cParams.chainLog;
-        if (zc->appliedParams.cParams.strategy == ZSTD_btlazy2)
+    if (params->cParams.strategy != ZSTD_fast) {
+        U32 const chainSize = (U32)1 << params->cParams.chainLog;
+        if (params->cParams.strategy == ZSTD_btlazy2)
             ZSTD_reduceTable_btlazy2(ms->chainTable, chainSize, reducerValue);
         else
             ZSTD_reduceTable(ms->chainTable, chainSize, reducerValue);
@@ -1774,164 +1926,6 @@ static void ZSTD_reduceIndex (ZSTD_CCtx* zc, const U32 reducerValue)
 
 /* See doc/zstd_compression_format.md for detailed format description */
 
-static size_t ZSTD_noCompressBlock (void* dst, size_t dstCapacity, const void* src, size_t srcSize, U32 lastBlock)
-{
-    U32 const cBlockHeader24 = lastBlock + (((U32)bt_raw)<<1) + (U32)(srcSize << 3);
-    if (srcSize + ZSTD_blockHeaderSize > dstCapacity) return ERROR(dstSize_tooSmall);
-    MEM_writeLE24(dst, cBlockHeader24);
-    memcpy((BYTE*)dst + ZSTD_blockHeaderSize, src, srcSize);
-    return ZSTD_blockHeaderSize + srcSize;
-}
-
-static size_t ZSTD_noCompressLiterals (void* dst, size_t dstCapacity, const void* src, size_t srcSize)
-{
-    BYTE* const ostart = (BYTE* const)dst;
-    U32   const flSize = 1 + (srcSize>31) + (srcSize>4095);
-
-    if (srcSize + flSize > dstCapacity) return ERROR(dstSize_tooSmall);
-
-    switch(flSize)
-    {
-        case 1: /* 2 - 1 - 5 */
-            ostart[0] = (BYTE)((U32)set_basic + (srcSize<<3));
-            break;
-        case 2: /* 2 - 2 - 12 */
-            MEM_writeLE16(ostart, (U16)((U32)set_basic + (1<<2) + (srcSize<<4)));
-            break;
-        case 3: /* 2 - 2 - 20 */
-            MEM_writeLE32(ostart, (U32)((U32)set_basic + (3<<2) + (srcSize<<4)));
-            break;
-        default:   /* not necessary : flSize is {1,2,3} */
-            assert(0);
-    }
-
-    memcpy(ostart + flSize, src, srcSize);
-    return srcSize + flSize;
-}
-
-static size_t ZSTD_compressRleLiteralsBlock (void* dst, size_t dstCapacity, const void* src, size_t srcSize)
-{
-    BYTE* const ostart = (BYTE* const)dst;
-    U32   const flSize = 1 + (srcSize>31) + (srcSize>4095);
-
-    (void)dstCapacity;  /* dstCapacity already guaranteed to be >=4, hence large enough */
-
-    switch(flSize)
-    {
-        case 1: /* 2 - 1 - 5 */
-            ostart[0] = (BYTE)((U32)set_rle + (srcSize<<3));
-            break;
-        case 2: /* 2 - 2 - 12 */
-            MEM_writeLE16(ostart, (U16)((U32)set_rle + (1<<2) + (srcSize<<4)));
-            break;
-        case 3: /* 2 - 2 - 20 */
-            MEM_writeLE32(ostart, (U32)((U32)set_rle + (3<<2) + (srcSize<<4)));
-            break;
-        default:   /* not necessary : flSize is {1,2,3} */
-            assert(0);
-    }
-
-    ostart[flSize] = *(const BYTE*)src;
-    return flSize+1;
-}
-
-
-/* ZSTD_minGain() :
- * minimum compression required
- * to generate a compress block or a compressed literals section.
- * note : use same formula for both situations */
-static size_t ZSTD_minGain(size_t srcSize, ZSTD_strategy strat)
-{
-    U32 const minlog = (strat>=ZSTD_btultra) ? (U32)(strat) - 1 : 6;
-    ZSTD_STATIC_ASSERT(ZSTD_btultra == 8);
-    assert(ZSTD_cParam_withinBounds(ZSTD_c_strategy, strat));
-    return (srcSize >> minlog) + 2;
-}
-
-static size_t ZSTD_compressLiterals (ZSTD_hufCTables_t const* prevHuf,
-                                     ZSTD_hufCTables_t* nextHuf,
-                                     ZSTD_strategy strategy, int disableLiteralCompression,
-                                     void* dst, size_t dstCapacity,
-                               const void* src, size_t srcSize,
-                                     void* workspace, size_t wkspSize,
-                               const int bmi2)
-{
-    size_t const minGain = ZSTD_minGain(srcSize, strategy);
-    size_t const lhSize = 3 + (srcSize >= 1 KB) + (srcSize >= 16 KB);
-    BYTE*  const ostart = (BYTE*)dst;
-    U32 singleStream = srcSize < 256;
-    symbolEncodingType_e hType = set_compressed;
-    size_t cLitSize;
-
-    DEBUGLOG(5,"ZSTD_compressLiterals (disableLiteralCompression=%i)",
-                disableLiteralCompression);
-
-    /* Prepare nextEntropy assuming reusing the existing table */
-    memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
-
-    if (disableLiteralCompression)
-        return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);
-
-    /* small ? don't even attempt compression (speed opt) */
-#   define COMPRESS_LITERALS_SIZE_MIN 63
-    {   size_t const minLitSize = (prevHuf->repeatMode == HUF_repeat_valid) ? 6 : COMPRESS_LITERALS_SIZE_MIN;
-        if (srcSize <= minLitSize) return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);
-    }
-
-    if (dstCapacity < lhSize+1) return ERROR(dstSize_tooSmall);   /* not enough space for compression */
-    {   HUF_repeat repeat = prevHuf->repeatMode;
-        int const preferRepeat = strategy < ZSTD_lazy ? srcSize <= 1024 : 0;
-        if (repeat == HUF_repeat_valid && lhSize == 3) singleStream = 1;
-        cLitSize = singleStream ? HUF_compress1X_repeat(ostart+lhSize, dstCapacity-lhSize, src, srcSize, 255, 11,
-                                      workspace, wkspSize, (HUF_CElt*)nextHuf->CTable, &repeat, preferRepeat, bmi2)
-                                : HUF_compress4X_repeat(ostart+lhSize, dstCapacity-lhSize, src, srcSize, 255, 11,
-                                      workspace, wkspSize, (HUF_CElt*)nextHuf->CTable, &repeat, preferRepeat, bmi2);
-        if (repeat != HUF_repeat_none) {
-            /* reused the existing table */
-            hType = set_repeat;
-        }
-    }
-
-    if ((cLitSize==0) | (cLitSize >= srcSize - minGain) | ERR_isError(cLitSize)) {
-        memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
-        return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);
-    }
-    if (cLitSize==1) {
-        memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
-        return ZSTD_compressRleLiteralsBlock(dst, dstCapacity, src, srcSize);
-    }
-
-    if (hType == set_compressed) {
-        /* using a newly constructed table */
-        nextHuf->repeatMode = HUF_repeat_check;
-    }
-
-    /* Build header */
-    switch(lhSize)
-    {
-    case 3: /* 2 - 2 - 10 - 10 */
-        {   U32 const lhc = hType + ((!singleStream) << 2) + ((U32)srcSize<<4) + ((U32)cLitSize<<14);
-            MEM_writeLE24(ostart, lhc);
-            break;
-        }
-    case 4: /* 2 - 2 - 14 - 14 */
-        {   U32 const lhc = hType + (2 << 2) + ((U32)srcSize<<4) + ((U32)cLitSize<<18);
-            MEM_writeLE32(ostart, lhc);
-            break;
-        }
-    case 5: /* 2 - 2 - 18 - 18 */
-        {   U32 const lhc = hType + (3 << 2) + ((U32)srcSize<<4) + ((U32)cLitSize<<22);
-            MEM_writeLE32(ostart, lhc);
-            ostart[4] = (BYTE)(cLitSize >> 10);
-            break;
-        }
-    default:  /* not possible : lhSize is {3,4,5} */
-        assert(0);
-    }
-    return lhSize+cLitSize;
-}
-
-
 void ZSTD_seqToCodes(const seqStore_t* seqStorePtr)
 {
     const seqDef* const sequences = seqStorePtr->sequencesStart;
@@ -1954,418 +1948,14 @@ void ZSTD_seqToCodes(const seqStore_t* seqStorePtr)
         mlCodeTable[seqStorePtr->longLengthPos] = MaxML;
 }
 
-
-/**
- * -log2(x / 256) lookup table for x in [0, 256).
- * If x == 0: Return 0
- * Else: Return floor(-log2(x / 256) * 256)
- */
-static unsigned const kInverseProbabiltyLog256[256] = {
-    0,    2048, 1792, 1642, 1536, 1453, 1386, 1329, 1280, 1236, 1197, 1162,
-    1130, 1100, 1073, 1047, 1024, 1001, 980,  960,  941,  923,  906,  889,
-    874,  859,  844,  830,  817,  804,  791,  779,  768,  756,  745,  734,
-    724,  714,  704,  694,  685,  676,  667,  658,  650,  642,  633,  626,
-    618,  610,  603,  595,  588,  581,  574,  567,  561,  554,  548,  542,
-    535,  529,  523,  517,  512,  506,  500,  495,  489,  484,  478,  473,
-    468,  463,  458,  453,  448,  443,  438,  434,  429,  424,  420,  415,
-    411,  407,  402,  398,  394,  390,  386,  382,  377,  373,  370,  366,
-    362,  358,  354,  350,  347,  343,  339,  336,  332,  329,  325,  322,
-    318,  315,  311,  308,  305,  302,  298,  295,  292,  289,  286,  282,
-    279,  276,  273,  270,  267,  264,  261,  258,  256,  253,  250,  247,
-    244,  241,  239,  236,  233,  230,  228,  225,  222,  220,  217,  215,
-    212,  209,  207,  204,  202,  199,  197,  194,  192,  190,  187,  185,
-    182,  180,  178,  175,  173,  171,  168,  166,  164,  162,  159,  157,
-    155,  153,  151,  149,  146,  144,  142,  140,  138,  136,  134,  132,
-    130,  128,  126,  123,  121,  119,  117,  115,  114,  112,  110,  108,
-    106,  104,  102,  100,  98,   96,   94,   93,   91,   89,   87,   85,
-    83,   82,   80,   78,   76,   74,   73,   71,   69,   67,   66,   64,
-    62,   61,   59,   57,   55,   54,   52,   50,   49,   47,   46,   44,
-    42,   41,   39,   37,   36,   34,   33,   31,   30,   28,   26,   25,
-    23,   22,   20,   19,   17,   16,   14,   13,   11,   10,   8,    7,
-    5,    4,    2,    1,
-};
-
-
-/**
- * Returns the cost in bits of encoding the distribution described by count
- * using the entropy bound.
- */
-static size_t ZSTD_entropyCost(unsigned const* count, unsigned const max, size_t const total)
-{
-    unsigned cost = 0;
-    unsigned s;
-    for (s = 0; s <= max; ++s) {
-        unsigned norm = (unsigned)((256 * count[s]) / total);
-        if (count[s] != 0 && norm == 0)
-            norm = 1;
-        assert(count[s] < total);
-        cost += count[s] * kInverseProbabiltyLog256[norm];
-    }
-    return cost >> 8;
-}
-
-
-/**
- * Returns the cost in bits of encoding the distribution in count using the
- * table described by norm. The max symbol support by norm is assumed >= max.
- * norm must be valid for every symbol with non-zero probability in count.
- */
-static size_t ZSTD_crossEntropyCost(short const* norm, unsigned accuracyLog,
-                                    unsigned const* count, unsigned const max)
-{
-    unsigned const shift = 8 - accuracyLog;
-    size_t cost = 0;
-    unsigned s;
-    assert(accuracyLog <= 8);
-    for (s = 0; s <= max; ++s) {
-        unsigned const normAcc = norm[s] != -1 ? norm[s] : 1;
-        unsigned const norm256 = normAcc << shift;
-        assert(norm256 > 0);
-        assert(norm256 < 256);
-        cost += count[s] * kInverseProbabiltyLog256[norm256];
-    }
-    return cost >> 8;
-}
-
-
-static unsigned ZSTD_getFSEMaxSymbolValue(FSE_CTable const* ctable) {
-  void const* ptr = ctable;
-  U16 const* u16ptr = (U16 const*)ptr;
-  U32 const maxSymbolValue = MEM_read16(u16ptr + 1);
-  return maxSymbolValue;
-}
-
-
-/**
- * Returns the cost in bits of encoding the distribution in count using ctable.
- * Returns an error if ctable cannot represent all the symbols in count.
- */
-static size_t ZSTD_fseBitCost(
-    FSE_CTable const* ctable,
-    unsigned const* count,
-    unsigned const max)
-{
-    unsigned const kAccuracyLog = 8;
-    size_t cost = 0;
-    unsigned s;
-    FSE_CState_t cstate;
-    FSE_initCState(&cstate, ctable);
-    if (ZSTD_getFSEMaxSymbolValue(ctable) < max) {
-        DEBUGLOG(5, "Repeat FSE_CTable has maxSymbolValue %u < %u",
-                    ZSTD_getFSEMaxSymbolValue(ctable), max);
-        return ERROR(GENERIC);
-    }
-    for (s = 0; s <= max; ++s) {
-        unsigned const tableLog = cstate.stateLog;
-        unsigned const badCost = (tableLog + 1) << kAccuracyLog;
-        unsigned const bitCost = FSE_bitCost(cstate.symbolTT, tableLog, s, kAccuracyLog);
-        if (count[s] == 0)
-            continue;
-        if (bitCost >= badCost) {
-            DEBUGLOG(5, "Repeat FSE_CTable has Prob[%u] == 0", s);
-            return ERROR(GENERIC);
-        }
-        cost += count[s] * bitCost;
-    }
-    return cost >> kAccuracyLog;
-}
-
-/**
- * Returns the cost in bytes of encoding the normalized count header.
- * Returns an error if any of the helper functions return an error.
- */
-static size_t ZSTD_NCountCost(unsigned const* count, unsigned const max,
-                              size_t const nbSeq, unsigned const FSELog)
+/* ZSTD_useTargetCBlockSize():
+ * Returns if target compressed block size param is being used.
+ * If used, compression will do best effort to make a compressed block size to be around targetCBlockSize.
+ * Returns 1 if true, 0 otherwise. */
+static int ZSTD_useTargetCBlockSize(const ZSTD_CCtx_params* cctxParams)
 {
-    BYTE wksp[FSE_NCOUNTBOUND];
-    S16 norm[MaxSeq + 1];
-    const U32 tableLog = FSE_optimalTableLog(FSELog, nbSeq, max);
-    CHECK_F(FSE_normalizeCount(norm, tableLog, count, nbSeq, max));
-    return FSE_writeNCount(wksp, sizeof(wksp), norm, max, tableLog);
-}
-
-
-typedef enum {
-    ZSTD_defaultDisallowed = 0,
-    ZSTD_defaultAllowed = 1
-} ZSTD_defaultPolicy_e;
-
-MEM_STATIC symbolEncodingType_e
-ZSTD_selectEncodingType(
-        FSE_repeat* repeatMode, unsigned const* count, unsigned const max,
-        size_t const mostFrequent, size_t nbSeq, unsigned const FSELog,
-        FSE_CTable const* prevCTable,
-        short const* defaultNorm, U32 defaultNormLog,
-        ZSTD_defaultPolicy_e const isDefaultAllowed,
-        ZSTD_strategy const strategy)
-{
-    ZSTD_STATIC_ASSERT(ZSTD_defaultDisallowed == 0 && ZSTD_defaultAllowed != 0);
-    if (mostFrequent == nbSeq) {
-        *repeatMode = FSE_repeat_none;
-        if (isDefaultAllowed && nbSeq <= 2) {
-            /* Prefer set_basic over set_rle when there are 2 or less symbols,
-             * since RLE uses 1 byte, but set_basic uses 5-6 bits per symbol.
-             * If basic encoding isn't possible, always choose RLE.
-             */
-            DEBUGLOG(5, "Selected set_basic");
-            return set_basic;
-        }
-        DEBUGLOG(5, "Selected set_rle");
-        return set_rle;
-    }
-    if (strategy < ZSTD_lazy) {
-        if (isDefaultAllowed) {
-            size_t const staticFse_nbSeq_max = 1000;
-            size_t const mult = 10 - strategy;
-            size_t const baseLog = 3;
-            size_t const dynamicFse_nbSeq_min = (((size_t)1 << defaultNormLog) * mult) >> baseLog;  /* 28-36 for offset, 56-72 for lengths */
-            assert(defaultNormLog >= 5 && defaultNormLog <= 6);  /* xx_DEFAULTNORMLOG */
-            assert(mult <= 9 && mult >= 7);
-            if ( (*repeatMode == FSE_repeat_valid)
-              && (nbSeq < staticFse_nbSeq_max) ) {
-                DEBUGLOG(5, "Selected set_repeat");
-                return set_repeat;
-            }
-            if ( (nbSeq < dynamicFse_nbSeq_min)
-              || (mostFrequent < (nbSeq >> (defaultNormLog-1))) ) {
-                DEBUGLOG(5, "Selected set_basic");
-                /* The format allows default tables to be repeated, but it isn't useful.
-                 * When using simple heuristics to select encoding type, we don't want
-                 * to confuse these tables with dictionaries. When running more careful
-                 * analysis, we don't need to waste time checking both repeating tables
-                 * and default tables.
-                 */
-                *repeatMode = FSE_repeat_none;
-                return set_basic;
-            }
-        }
-    } else {
-        size_t const basicCost = isDefaultAllowed ? ZSTD_crossEntropyCost(defaultNorm, defaultNormLog, count, max) : ERROR(GENERIC);
-        size_t const repeatCost = *repeatMode != FSE_repeat_none ? ZSTD_fseBitCost(prevCTable, count, max) : ERROR(GENERIC);
-        size_t const NCountCost = ZSTD_NCountCost(count, max, nbSeq, FSELog);
-        size_t const compressedCost = (NCountCost << 3) + ZSTD_entropyCost(count, max, nbSeq);
-
-        if (isDefaultAllowed) {
-            assert(!ZSTD_isError(basicCost));
-            assert(!(*repeatMode == FSE_repeat_valid && ZSTD_isError(repeatCost)));
-        }
-        assert(!ZSTD_isError(NCountCost));
-        assert(compressedCost < ERROR(maxCode));
-        DEBUGLOG(5, "Estimated bit costs: basic=%u\trepeat=%u\tcompressed=%u",
-                    (unsigned)basicCost, (unsigned)repeatCost, (unsigned)compressedCost);
-        if (basicCost <= repeatCost && basicCost <= compressedCost) {
-            DEBUGLOG(5, "Selected set_basic");
-            assert(isDefaultAllowed);
-            *repeatMode = FSE_repeat_none;
-            return set_basic;
-        }
-        if (repeatCost <= compressedCost) {
-            DEBUGLOG(5, "Selected set_repeat");
-            assert(!ZSTD_isError(repeatCost));
-            return set_repeat;
-        }
-        assert(compressedCost < basicCost && compressedCost < repeatCost);
-    }
-    DEBUGLOG(5, "Selected set_compressed");
-    *repeatMode = FSE_repeat_check;
-    return set_compressed;
-}
-
-MEM_STATIC size_t
-ZSTD_buildCTable(void* dst, size_t dstCapacity,
-                FSE_CTable* nextCTable, U32 FSELog, symbolEncodingType_e type,
-                unsigned* count, U32 max,
-                const BYTE* codeTable, size_t nbSeq,
-                const S16* defaultNorm, U32 defaultNormLog, U32 defaultMax,
-                const FSE_CTable* prevCTable, size_t prevCTableSize,
-                void* workspace, size_t workspaceSize)
-{
-    BYTE* op = (BYTE*)dst;
-    const BYTE* const oend = op + dstCapacity;
-    DEBUGLOG(6, "ZSTD_buildCTable (dstCapacity=%u)", (unsigned)dstCapacity);
-
-    switch (type) {
-    case set_rle:
-        CHECK_F(FSE_buildCTable_rle(nextCTable, (BYTE)max));
-        if (dstCapacity==0) return ERROR(dstSize_tooSmall);
-        *op = codeTable[0];
-        return 1;
-    case set_repeat:
-        memcpy(nextCTable, prevCTable, prevCTableSize);
-        return 0;
-    case set_basic:
-        CHECK_F(FSE_buildCTable_wksp(nextCTable, defaultNorm, defaultMax, defaultNormLog, workspace, workspaceSize));  /* note : could be pre-calculated */
-        return 0;
-    case set_compressed: {
-        S16 norm[MaxSeq + 1];
-        size_t nbSeq_1 = nbSeq;
-        const U32 tableLog = FSE_optimalTableLog(FSELog, nbSeq, max);
-        if (count[codeTable[nbSeq-1]] > 1) {
-            count[codeTable[nbSeq-1]]--;
-            nbSeq_1--;
-        }
-        assert(nbSeq_1 > 1);
-        CHECK_F(FSE_normalizeCount(norm, tableLog, count, nbSeq_1, max));
-        {   size_t const NCountSize = FSE_writeNCount(op, oend - op, norm, max, tableLog);   /* overflow protected */
-            if (FSE_isError(NCountSize)) return NCountSize;
-            CHECK_F(FSE_buildCTable_wksp(nextCTable, norm, max, tableLog, workspace, workspaceSize));
-            return NCountSize;
-        }
-    }
-    default: return assert(0), ERROR(GENERIC);
-    }
-}
-
-FORCE_INLINE_TEMPLATE size_t
-ZSTD_encodeSequences_body(
-            void* dst, size_t dstCapacity,
-            FSE_CTable const* CTable_MatchLength, BYTE const* mlCodeTable,
-            FSE_CTable const* CTable_OffsetBits, BYTE const* ofCodeTable,
-            FSE_CTable const* CTable_LitLength, BYTE const* llCodeTable,
-            seqDef const* sequences, size_t nbSeq, int longOffsets)
-{
-    BIT_CStream_t blockStream;
-    FSE_CState_t  stateMatchLength;
-    FSE_CState_t  stateOffsetBits;
-    FSE_CState_t  stateLitLength;
-
-    CHECK_E(BIT_initCStream(&blockStream, dst, dstCapacity), dstSize_tooSmall); /* not enough space remaining */
-    DEBUGLOG(6, "available space for bitstream : %i  (dstCapacity=%u)",
-                (int)(blockStream.endPtr - blockStream.startPtr),
-                (unsigned)dstCapacity);
-
-    /* first symbols */
-    FSE_initCState2(&stateMatchLength, CTable_MatchLength, mlCodeTable[nbSeq-1]);
-    FSE_initCState2(&stateOffsetBits,  CTable_OffsetBits,  ofCodeTable[nbSeq-1]);
-    FSE_initCState2(&stateLitLength,   CTable_LitLength,   llCodeTable[nbSeq-1]);
-    BIT_addBits(&blockStream, sequences[nbSeq-1].litLength, LL_bits[llCodeTable[nbSeq-1]]);
-    if (MEM_32bits()) BIT_flushBits(&blockStream);
-    BIT_addBits(&blockStream, sequences[nbSeq-1].matchLength, ML_bits[mlCodeTable[nbSeq-1]]);
-    if (MEM_32bits()) BIT_flushBits(&blockStream);
-    if (longOffsets) {
-        U32 const ofBits = ofCodeTable[nbSeq-1];
-        int const extraBits = ofBits - MIN(ofBits, STREAM_ACCUMULATOR_MIN-1);
-        if (extraBits) {
-            BIT_addBits(&blockStream, sequences[nbSeq-1].offset, extraBits);
-            BIT_flushBits(&blockStream);
-        }
-        BIT_addBits(&blockStream, sequences[nbSeq-1].offset >> extraBits,
-                    ofBits - extraBits);
-    } else {
-        BIT_addBits(&blockStream, sequences[nbSeq-1].offset, ofCodeTable[nbSeq-1]);
-    }
-    BIT_flushBits(&blockStream);
-
-    {   size_t n;
-        for (n=nbSeq-2 ; n<nbSeq ; n--) {      /* intentional underflow */
-            BYTE const llCode = llCodeTable[n];
-            BYTE const ofCode = ofCodeTable[n];
-            BYTE const mlCode = mlCodeTable[n];
-            U32  const llBits = LL_bits[llCode];
-            U32  const ofBits = ofCode;
-            U32  const mlBits = ML_bits[mlCode];
-            DEBUGLOG(6, "encoding: litlen:%2u - matchlen:%2u - offCode:%7u",
-                        (unsigned)sequences[n].litLength,
-                        (unsigned)sequences[n].matchLength + MINMATCH,
-                        (unsigned)sequences[n].offset);
-                                                                            /* 32b*/  /* 64b*/
-                                                                            /* (7)*/  /* (7)*/
-            FSE_encodeSymbol(&blockStream, &stateOffsetBits, ofCode);       /* 15 */  /* 15 */
-            FSE_encodeSymbol(&blockStream, &stateMatchLength, mlCode);      /* 24 */  /* 24 */
-            if (MEM_32bits()) BIT_flushBits(&blockStream);                  /* (7)*/
-            FSE_encodeSymbol(&blockStream, &stateLitLength, llCode);        /* 16 */  /* 33 */
-            if (MEM_32bits() || (ofBits+mlBits+llBits >= 64-7-(LLFSELog+MLFSELog+OffFSELog)))
-                BIT_flushBits(&blockStream);                                /* (7)*/
-            BIT_addBits(&blockStream, sequences[n].litLength, llBits);
-            if (MEM_32bits() && ((llBits+mlBits)>24)) BIT_flushBits(&blockStream);
-            BIT_addBits(&blockStream, sequences[n].matchLength, mlBits);
-            if (MEM_32bits() || (ofBits+mlBits+llBits > 56)) BIT_flushBits(&blockStream);
-            if (longOffsets) {
-                int const extraBits = ofBits - MIN(ofBits, STREAM_ACCUMULATOR_MIN-1);
-                if (extraBits) {
-                    BIT_addBits(&blockStream, sequences[n].offset, extraBits);
-                    BIT_flushBits(&blockStream);                            /* (7)*/
-                }
-                BIT_addBits(&blockStream, sequences[n].offset >> extraBits,
-                            ofBits - extraBits);                            /* 31 */
-            } else {
-                BIT_addBits(&blockStream, sequences[n].offset, ofBits);     /* 31 */
-            }
-            BIT_flushBits(&blockStream);                                    /* (7)*/
-            DEBUGLOG(7, "remaining space : %i", (int)(blockStream.endPtr - blockStream.ptr));
-    }   }
-
-    DEBUGLOG(6, "ZSTD_encodeSequences: flushing ML state with %u bits", stateMatchLength.stateLog);
-    FSE_flushCState(&blockStream, &stateMatchLength);
-    DEBUGLOG(6, "ZSTD_encodeSequences: flushing Off state with %u bits", stateOffsetBits.stateLog);
-    FSE_flushCState(&blockStream, &stateOffsetBits);
-    DEBUGLOG(6, "ZSTD_encodeSequences: flushing LL state with %u bits", stateLitLength.stateLog);
-    FSE_flushCState(&blockStream, &stateLitLength);
-
-    {   size_t const streamSize = BIT_closeCStream(&blockStream);
-        if (streamSize==0) return ERROR(dstSize_tooSmall);   /* not enough space */
-        return streamSize;
-    }
-}
-
-static size_t
-ZSTD_encodeSequences_default(
-            void* dst, size_t dstCapacity,
-            FSE_CTable const* CTable_MatchLength, BYTE const* mlCodeTable,
-            FSE_CTable const* CTable_OffsetBits, BYTE const* ofCodeTable,
-            FSE_CTable const* CTable_LitLength, BYTE const* llCodeTable,
-            seqDef const* sequences, size_t nbSeq, int longOffsets)
-{
-    return ZSTD_encodeSequences_body(dst, dstCapacity,
-                                    CTable_MatchLength, mlCodeTable,
-                                    CTable_OffsetBits, ofCodeTable,
-                                    CTable_LitLength, llCodeTable,
-                                    sequences, nbSeq, longOffsets);
-}
-
-
-#if DYNAMIC_BMI2
-
-static TARGET_ATTRIBUTE("bmi2") size_t
-ZSTD_encodeSequences_bmi2(
-            void* dst, size_t dstCapacity,
-            FSE_CTable const* CTable_MatchLength, BYTE const* mlCodeTable,
-            FSE_CTable const* CTable_OffsetBits, BYTE const* ofCodeTable,
-            FSE_CTable const* CTable_LitLength, BYTE const* llCodeTable,
-            seqDef const* sequences, size_t nbSeq, int longOffsets)
-{
-    return ZSTD_encodeSequences_body(dst, dstCapacity,
-                                    CTable_MatchLength, mlCodeTable,
-                                    CTable_OffsetBits, ofCodeTable,
-                                    CTable_LitLength, llCodeTable,
-                                    sequences, nbSeq, longOffsets);
-}
-
-#endif
-
-static size_t ZSTD_encodeSequences(
-            void* dst, size_t dstCapacity,
-            FSE_CTable const* CTable_MatchLength, BYTE const* mlCodeTable,
-            FSE_CTable const* CTable_OffsetBits, BYTE const* ofCodeTable,
-            FSE_CTable const* CTable_LitLength, BYTE const* llCodeTable,
-            seqDef const* sequences, size_t nbSeq, int longOffsets, int bmi2)
-{
-    DEBUGLOG(5, "ZSTD_encodeSequences: dstCapacity = %u", (unsigned)dstCapacity);
-#if DYNAMIC_BMI2
-    if (bmi2) {
-        return ZSTD_encodeSequences_bmi2(dst, dstCapacity,
-                                         CTable_MatchLength, mlCodeTable,
-                                         CTable_OffsetBits, ofCodeTable,
-                                         CTable_LitLength, llCodeTable,
-                                         sequences, nbSeq, longOffsets);
-    }
-#endif
-    (void)bmi2;
-    return ZSTD_encodeSequences_default(dst, dstCapacity,
-                                        CTable_MatchLength, mlCodeTable,
-                                        CTable_OffsetBits, ofCodeTable,
-                                        CTable_LitLength, llCodeTable,
-                                        sequences, nbSeq, longOffsets);
+    DEBUGLOG(5, "ZSTD_useTargetCBlockSize (targetCBlockSize=%zu)", cctxParams->targetCBlockSize);
+    return (cctxParams->targetCBlockSize != 0);
 }
 
 /* ZSTD_compressSequences_internal():
@@ -2376,7 +1966,7 @@ ZSTD_compressSequences_internal(seqStore_t* seqStorePtr,
                                 ZSTD_entropyCTables_t* nextEntropy,
                           const ZSTD_CCtx_params* cctxParams,
                                 void* dst, size_t dstCapacity,
-                                void* workspace, size_t wkspSize,
+                                void* entropyWorkspace, size_t entropyWkspSize,
                           const int bmi2)
 {
     const int longOffsets = cctxParams->cParams.windowLog > STREAM_ACCUMULATOR_MIN;
@@ -2393,52 +1983,59 @@ ZSTD_compressSequences_internal(seqStore_t* seqStorePtr,
     BYTE* const ostart = (BYTE*)dst;
     BYTE* const oend = ostart + dstCapacity;
     BYTE* op = ostart;
-    size_t const nbSeq = seqStorePtr->sequences - seqStorePtr->sequencesStart;
+    size_t const nbSeq = (size_t)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
     BYTE* seqHead;
     BYTE* lastNCount = NULL;
 
+    DEBUGLOG(5, "ZSTD_compressSequences_internal (nbSeq=%zu)", nbSeq);
     ZSTD_STATIC_ASSERT(HUF_WORKSPACE_SIZE >= (1<<MAX(MLFSELog,LLFSELog)));
-    DEBUGLOG(5, "ZSTD_compressSequences_internal");
 
     /* Compress literals */
     {   const BYTE* const literals = seqStorePtr->litStart;
-        size_t const litSize = seqStorePtr->lit - literals;
-        int const disableLiteralCompression = (cctxParams->cParams.strategy == ZSTD_fast) && (cctxParams->cParams.targetLength > 0);
+        size_t const litSize = (size_t)(seqStorePtr->lit - literals);
         size_t const cSize = ZSTD_compressLiterals(
                                     &prevEntropy->huf, &nextEntropy->huf,
-                                    cctxParams->cParams.strategy, disableLiteralCompression,
+                                    cctxParams->cParams.strategy,
+                                    ZSTD_disableLiteralsCompression(cctxParams),
                                     op, dstCapacity,
                                     literals, litSize,
-                                    workspace, wkspSize,
+                                    entropyWorkspace, entropyWkspSize,
                                     bmi2);
-        if (ZSTD_isError(cSize))
-          return cSize;
+        FORWARD_IF_ERROR(cSize, "ZSTD_compressLiterals failed");
         assert(cSize <= dstCapacity);
         op += cSize;
     }
 
     /* Sequences Header */
-    if ((oend-op) < 3 /*max nbSeq Size*/ + 1 /*seqHead*/) return ERROR(dstSize_tooSmall);
-    if (nbSeq < 0x7F)
+    RETURN_ERROR_IF((oend-op) < 3 /*max nbSeq Size*/ + 1 /*seqHead*/,
+                    dstSize_tooSmall, "Can't fit seq hdr in output buf!");
+    if (nbSeq < 128) {
         *op++ = (BYTE)nbSeq;
-    else if (nbSeq < LONGNBSEQ)
-        op[0] = (BYTE)((nbSeq>>8) + 0x80), op[1] = (BYTE)nbSeq, op+=2;
-    else
-        op[0]=0xFF, MEM_writeLE16(op+1, (U16)(nbSeq - LONGNBSEQ)), op+=3;
+    } else if (nbSeq < LONGNBSEQ) {
+        op[0] = (BYTE)((nbSeq>>8) + 0x80);
+        op[1] = (BYTE)nbSeq;
+        op+=2;
+    } else {
+        op[0]=0xFF;
+        MEM_writeLE16(op+1, (U16)(nbSeq - LONGNBSEQ));
+        op+=3;
+    }
+    assert(op <= oend);
     if (nbSeq==0) {
         /* Copy the old tables over as if we repeated them */
         memcpy(&nextEntropy->fse, &prevEntropy->fse, sizeof(prevEntropy->fse));
-        return op - ostart;
+        return (size_t)(op - ostart);
     }
 
     /* seqHead : flags for FSE encoding type */
     seqHead = op++;
+    assert(op <= oend);
 
     /* convert length/distances into codes */
     ZSTD_seqToCodes(seqStorePtr);
     /* build CTable for Literal Lengths */
     {   unsigned max = MaxLL;
-        size_t const mostFrequent = HIST_countFast_wksp(count, &max, llCodeTable, nbSeq, workspace, wkspSize);   /* can't fail */
+        size_t const mostFrequent = HIST_countFast_wksp(count, &max, llCodeTable, nbSeq, entropyWorkspace, entropyWkspSize);   /* can't fail */
         DEBUGLOG(5, "Building LL table");
         nextEntropy->fse.litlength_repeatMode = prevEntropy->fse.litlength_repeatMode;
         LLtype = ZSTD_selectEncodingType(&nextEntropy->fse.litlength_repeatMode,
@@ -2448,18 +2045,24 @@ ZSTD_compressSequences_internal(seqStore_t* seqStorePtr,
                                         ZSTD_defaultAllowed, strategy);
         assert(set_basic < set_compressed && set_rle < set_compressed);
         assert(!(LLtype < set_compressed && nextEntropy->fse.litlength_repeatMode != FSE_repeat_none)); /* We don't copy tables */
-        {   size_t const countSize = ZSTD_buildCTable(op, oend - op, CTable_LitLength, LLFSELog, (symbolEncodingType_e)LLtype,
-                                                    count, max, llCodeTable, nbSeq, LL_defaultNorm, LL_defaultNormLog, MaxLL,
-                                                    prevEntropy->fse.litlengthCTable, sizeof(prevEntropy->fse.litlengthCTable),
-                                                    workspace, wkspSize);
-            if (ZSTD_isError(countSize)) return countSize;
+        {   size_t const countSize = ZSTD_buildCTable(
+                op, (size_t)(oend - op),
+                CTable_LitLength, LLFSELog, (symbolEncodingType_e)LLtype,
+                count, max, llCodeTable, nbSeq,
+                LL_defaultNorm, LL_defaultNormLog, MaxLL,
+                prevEntropy->fse.litlengthCTable,
+                sizeof(prevEntropy->fse.litlengthCTable),
+                entropyWorkspace, entropyWkspSize);
+            FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for LitLens failed");
             if (LLtype == set_compressed)
                 lastNCount = op;
             op += countSize;
+            assert(op <= oend);
     }   }
     /* build CTable for Offsets */
     {   unsigned max = MaxOff;
-        size_t const mostFrequent = HIST_countFast_wksp(count, &max, ofCodeTable, nbSeq, workspace, wkspSize);  /* can't fail */
+        size_t const mostFrequent = HIST_countFast_wksp(
+            count, &max, ofCodeTable, nbSeq, entropyWorkspace, entropyWkspSize);  /* can't fail */
         /* We can only use the basic table if max <= DefaultMaxOff, otherwise the offsets are too large */
         ZSTD_defaultPolicy_e const defaultPolicy = (max <= DefaultMaxOff) ? ZSTD_defaultAllowed : ZSTD_defaultDisallowed;
         DEBUGLOG(5, "Building OF table");
@@ -2470,18 +2073,24 @@ ZSTD_compressSequences_internal(seqStore_t* seqStorePtr,
                                         OF_defaultNorm, OF_defaultNormLog,
                                         defaultPolicy, strategy);
         assert(!(Offtype < set_compressed && nextEntropy->fse.offcode_repeatMode != FSE_repeat_none)); /* We don't copy tables */
-        {   size_t const countSize = ZSTD_buildCTable(op, oend - op, CTable_OffsetBits, OffFSELog, (symbolEncodingType_e)Offtype,
-                                                    count, max, ofCodeTable, nbSeq, OF_defaultNorm, OF_defaultNormLog, DefaultMaxOff,
-                                                    prevEntropy->fse.offcodeCTable, sizeof(prevEntropy->fse.offcodeCTable),
-                                                    workspace, wkspSize);
-            if (ZSTD_isError(countSize)) return countSize;
+        {   size_t const countSize = ZSTD_buildCTable(
+                op, (size_t)(oend - op),
+                CTable_OffsetBits, OffFSELog, (symbolEncodingType_e)Offtype,
+                count, max, ofCodeTable, nbSeq,
+                OF_defaultNorm, OF_defaultNormLog, DefaultMaxOff,
+                prevEntropy->fse.offcodeCTable,
+                sizeof(prevEntropy->fse.offcodeCTable),
+                entropyWorkspace, entropyWkspSize);
+            FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for Offsets failed");
             if (Offtype == set_compressed)
                 lastNCount = op;
             op += countSize;
+            assert(op <= oend);
     }   }
     /* build CTable for MatchLengths */
     {   unsigned max = MaxML;
-        size_t const mostFrequent = HIST_countFast_wksp(count, &max, mlCodeTable, nbSeq, workspace, wkspSize);   /* can't fail */
+        size_t const mostFrequent = HIST_countFast_wksp(
+            count, &max, mlCodeTable, nbSeq, entropyWorkspace, entropyWkspSize);   /* can't fail */
         DEBUGLOG(5, "Building ML table (remaining space : %i)", (int)(oend-op));
         nextEntropy->fse.matchlength_repeatMode = prevEntropy->fse.matchlength_repeatMode;
         MLtype = ZSTD_selectEncodingType(&nextEntropy->fse.matchlength_repeatMode,
@@ -2490,29 +2099,35 @@ ZSTD_compressSequences_internal(seqStore_t* seqStorePtr,
                                         ML_defaultNorm, ML_defaultNormLog,
                                         ZSTD_defaultAllowed, strategy);
         assert(!(MLtype < set_compressed && nextEntropy->fse.matchlength_repeatMode != FSE_repeat_none)); /* We don't copy tables */
-        {   size_t const countSize = ZSTD_buildCTable(op, oend - op, CTable_MatchLength, MLFSELog, (symbolEncodingType_e)MLtype,
-                                                    count, max, mlCodeTable, nbSeq, ML_defaultNorm, ML_defaultNormLog, MaxML,
-                                                    prevEntropy->fse.matchlengthCTable, sizeof(prevEntropy->fse.matchlengthCTable),
-                                                    workspace, wkspSize);
-            if (ZSTD_isError(countSize)) return countSize;
+        {   size_t const countSize = ZSTD_buildCTable(
+                op, (size_t)(oend - op),
+                CTable_MatchLength, MLFSELog, (symbolEncodingType_e)MLtype,
+                count, max, mlCodeTable, nbSeq,
+                ML_defaultNorm, ML_defaultNormLog, MaxML,
+                prevEntropy->fse.matchlengthCTable,
+                sizeof(prevEntropy->fse.matchlengthCTable),
+                entropyWorkspace, entropyWkspSize);
+            FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for MatchLengths failed");
             if (MLtype == set_compressed)
                 lastNCount = op;
             op += countSize;
+            assert(op <= oend);
     }   }
 
     *seqHead = (BYTE)((LLtype<<6) + (Offtype<<4) + (MLtype<<2));
 
     {   size_t const bitstreamSize = ZSTD_encodeSequences(
-                                        op, oend - op,
+                                        op, (size_t)(oend - op),
                                         CTable_MatchLength, mlCodeTable,
                                         CTable_OffsetBits, ofCodeTable,
                                         CTable_LitLength, llCodeTable,
                                         sequences, nbSeq,
                                         longOffsets, bmi2);
-        if (ZSTD_isError(bitstreamSize)) return bitstreamSize;
+        FORWARD_IF_ERROR(bitstreamSize, "ZSTD_encodeSequences failed");
         op += bitstreamSize;
+        assert(op <= oend);
         /* zstd versions <= 1.3.4 mistakenly report corruption when
-         * FSE_readNCount() recieves a buffer < 4 bytes.
+         * FSE_readNCount() receives a buffer < 4 bytes.
          * Fixed by https://github.com/facebook/zstd/pull/1146.
          * This can happen when the last set_compressed table present is 2
          * bytes and the bitstream is only one byte.
@@ -2529,7 +2144,7 @@ ZSTD_compressSequences_internal(seqStore_t* seqStorePtr,
     }
 
     DEBUGLOG(5, "compressed block size : %u", (unsigned)(op - ostart));
-    return op - ostart;
+    return (size_t)(op - ostart);
 }
 
 MEM_STATIC size_t
@@ -2539,20 +2154,20 @@ ZSTD_compressSequences(seqStore_t* seqStorePtr,
                        const ZSTD_CCtx_params* cctxParams,
                              void* dst, size_t dstCapacity,
                              size_t srcSize,
-                             void* workspace, size_t wkspSize,
+                             void* entropyWorkspace, size_t entropyWkspSize,
                              int bmi2)
 {
     size_t const cSize = ZSTD_compressSequences_internal(
                             seqStorePtr, prevEntropy, nextEntropy, cctxParams,
                             dst, dstCapacity,
-                            workspace, wkspSize, bmi2);
+                            entropyWorkspace, entropyWkspSize, bmi2);
     if (cSize == 0) return 0;
     /* When srcSize <= dstCapacity, there is enough space to write a raw uncompressed block.
      * Since we ran out of space, block must be not compressible, so fall back to raw uncompressed block.
      */
     if ((cSize == ERROR(dstSize_tooSmall)) & (srcSize <= dstCapacity))
         return 0;  /* block not compressed */
-    if (ZSTD_isError(cSize)) return cSize;
+    FORWARD_IF_ERROR(cSize, "ZSTD_compressSequences_internal failed");
 
     /* Check compressibility */
     {   size_t const maxCSize = srcSize - ZSTD_minGain(srcSize, cctxParams->cParams.strategy);
@@ -2622,27 +2237,24 @@ void ZSTD_resetSeqStore(seqStore_t* ssPtr)
     ssPtr->longLengthID = 0;
 }
 
-static size_t ZSTD_compressBlock_internal(ZSTD_CCtx* zc,
-                                        void* dst, size_t dstCapacity,
-                                        const void* src, size_t srcSize)
+typedef enum { ZSTDbss_compress, ZSTDbss_noCompress } ZSTD_buildSeqStore_e;
+
+static size_t ZSTD_buildSeqStore(ZSTD_CCtx* zc, const void* src, size_t srcSize)
 {
     ZSTD_matchState_t* const ms = &zc->blockState.matchState;
-    size_t cSize;
-    DEBUGLOG(5, "ZSTD_compressBlock_internal (dstCapacity=%u, dictLimit=%u, nextToUpdate=%u)",
-                (unsigned)dstCapacity, (unsigned)ms->window.dictLimit, (unsigned)ms->nextToUpdate);
+    DEBUGLOG(5, "ZSTD_buildSeqStore (srcSize=%zu)", srcSize);
     assert(srcSize <= ZSTD_BLOCKSIZE_MAX);
-
     /* Assert that we have correctly flushed the ctx params into the ms's copy */
     ZSTD_assertEqualCParams(zc->appliedParams.cParams, ms->cParams);
-
     if (srcSize < MIN_CBLOCK_SIZE+ZSTD_blockHeaderSize+1) {
         ZSTD_ldm_skipSequences(&zc->externSeqStore, srcSize, zc->appliedParams.cParams.minMatch);
-        cSize = 0;
-        goto out;  /* don't even attempt compression below a certain srcSize */
+        return ZSTDbss_noCompress; /* don't even attempt compression below a certain srcSize */
     }
     ZSTD_resetSeqStore(&(zc->seqStore));
-    ms->opt.symbolCosts = &zc->blockState.prevCBlock->entropy;   /* required for optimal parser to read stats from dictionary */
-
+    /* required for optimal parser to read stats from dictionary */
+    ms->opt.symbolCosts = &zc->blockState.prevCBlock->entropy;
+    /* tell the optimal parser how we expect to compress literals */
+    ms->opt.literalCompressionMode = zc->appliedParams.literalCompressionMode;
     /* a gap between an attached dict and the current window is not safe,
      * they must remain adjacent,
      * and when that stops being the case, the dict must be unset */
@@ -2679,9 +2291,9 @@ static size_t ZSTD_compressBlock_internal(ZSTD_CCtx* zc,
             ldmSeqStore.seq = zc->ldmSequences;
             ldmSeqStore.capacity = zc->maxNbLdmSequences;
             /* Updates ldmSeqStore.size */
-            CHECK_F(ZSTD_ldm_generateSequences(&zc->ldmState, &ldmSeqStore,
+            FORWARD_IF_ERROR(ZSTD_ldm_generateSequences(&zc->ldmState, &ldmSeqStore,
                                                &zc->appliedParams.ldmParams,
-                                               src, srcSize));
+                                               src, srcSize), "");
             /* Updates ldmSeqStore.pos */
             lastLLSize =
                 ZSTD_ldm_blockCompress(&ldmSeqStore,
@@ -2696,6 +2308,134 @@ static size_t ZSTD_compressBlock_internal(ZSTD_CCtx* zc,
         {   const BYTE* const lastLiterals = (const BYTE*)src + srcSize - lastLLSize;
             ZSTD_storeLastLiterals(&zc->seqStore, lastLiterals, lastLLSize);
     }   }
+    return ZSTDbss_compress;
+}
+
+static void ZSTD_copyBlockSequences(ZSTD_CCtx* zc)
+{
+    const seqStore_t* seqStore = ZSTD_getSeqStore(zc);
+    const seqDef* seqs = seqStore->sequencesStart;
+    size_t seqsSize = seqStore->sequences - seqs;
+
+    ZSTD_Sequence* outSeqs = &zc->seqCollector.seqStart[zc->seqCollector.seqIndex];
+    size_t i; size_t position; int repIdx;
+
+    assert(zc->seqCollector.seqIndex + 1 < zc->seqCollector.maxSequences);
+    for (i = 0, position = 0; i < seqsSize; ++i) {
+        outSeqs[i].offset = seqs[i].offset;
+        outSeqs[i].litLength = seqs[i].litLength;
+        outSeqs[i].matchLength = seqs[i].matchLength + MINMATCH;
+
+        if (i == seqStore->longLengthPos) {
+            if (seqStore->longLengthID == 1) {
+                outSeqs[i].litLength += 0x10000;
+            } else if (seqStore->longLengthID == 2) {
+                outSeqs[i].matchLength += 0x10000;
+            }
+        }
+
+        if (outSeqs[i].offset <= ZSTD_REP_NUM) {
+            outSeqs[i].rep = outSeqs[i].offset;
+            repIdx = (unsigned int)i - outSeqs[i].offset;
+
+            if (outSeqs[i].litLength == 0) {
+                if (outSeqs[i].offset < 3) {
+                    --repIdx;
+                } else {
+                    repIdx = (unsigned int)i - 1;
+                }
+                ++outSeqs[i].rep;
+            }
+            assert(repIdx >= -3);
+            outSeqs[i].offset = repIdx >= 0 ? outSeqs[repIdx].offset : repStartValue[-repIdx - 1];
+            if (outSeqs[i].rep == 4) {
+                --outSeqs[i].offset;
+            }
+        } else {
+            outSeqs[i].offset -= ZSTD_REP_NUM;
+        }
+
+        position += outSeqs[i].litLength;
+        outSeqs[i].matchPos = (unsigned int)position;
+        position += outSeqs[i].matchLength;
+    }
+    zc->seqCollector.seqIndex += seqsSize;
+}
+
+size_t ZSTD_getSequences(ZSTD_CCtx* zc, ZSTD_Sequence* outSeqs,
+    size_t outSeqsSize, const void* src, size_t srcSize)
+{
+    const size_t dstCapacity = ZSTD_compressBound(srcSize);
+    void* dst = ZSTD_malloc(dstCapacity, ZSTD_defaultCMem);
+    SeqCollector seqCollector;
+
+    RETURN_ERROR_IF(dst == NULL, memory_allocation, "NULL pointer!");
+
+    seqCollector.collectSequences = 1;
+    seqCollector.seqStart = outSeqs;
+    seqCollector.seqIndex = 0;
+    seqCollector.maxSequences = outSeqsSize;
+    zc->seqCollector = seqCollector;
+
+    ZSTD_compress2(zc, dst, dstCapacity, src, srcSize);
+    ZSTD_free(dst, ZSTD_defaultCMem);
+    return zc->seqCollector.seqIndex;
+}
+
+/* Returns true if the given block is a RLE block */
+static int ZSTD_isRLE(const BYTE *ip, size_t length) {
+    size_t i;
+    if (length < 2) return 1;
+    for (i = 1; i < length; ++i) {
+        if (ip[0] != ip[i]) return 0;
+    }
+    return 1;
+}
+
+/* Returns true if the given block may be RLE.
+ * This is just a heuristic based on the compressibility.
+ * It may return both false positives and false negatives.
+ */
+static int ZSTD_maybeRLE(seqStore_t const* seqStore)
+{
+    size_t const nbSeqs = (size_t)(seqStore->sequences - seqStore->sequencesStart);
+    size_t const nbLits = (size_t)(seqStore->lit - seqStore->litStart);
+
+    return nbSeqs < 4 && nbLits < 10;
+}
+
+static void ZSTD_confirmRepcodesAndEntropyTables(ZSTD_CCtx* zc)
+{
+    ZSTD_compressedBlockState_t* const tmp = zc->blockState.prevCBlock;
+    zc->blockState.prevCBlock = zc->blockState.nextCBlock;
+    zc->blockState.nextCBlock = tmp;
+}
+
+static size_t ZSTD_compressBlock_internal(ZSTD_CCtx* zc,
+                                        void* dst, size_t dstCapacity,
+                                        const void* src, size_t srcSize, U32 frame)
+{
+    /* This the upper bound for the length of an rle block.
+     * This isn't the actual upper bound. Finding the real threshold
+     * needs further investigation.
+     */
+    const U32 rleMaxLength = 25;
+    size_t cSize;
+    const BYTE* ip = (const BYTE*)src;
+    BYTE* op = (BYTE*)dst;
+    DEBUGLOG(5, "ZSTD_compressBlock_internal (dstCapacity=%u, dictLimit=%u, nextToUpdate=%u)",
+                (unsigned)dstCapacity, (unsigned)zc->blockState.matchState.window.dictLimit,
+                (unsigned)zc->blockState.matchState.nextToUpdate);
+
+    {   const size_t bss = ZSTD_buildSeqStore(zc, src, srcSize);
+        FORWARD_IF_ERROR(bss, "ZSTD_buildSeqStore failed");
+        if (bss == ZSTDbss_noCompress) { cSize = 0; goto out; }
+    }
+
+    if (zc->seqCollector.collectSequences) {
+        ZSTD_copyBlockSequences(zc);
+        return 0;
+    }
 
     /* encode sequences and literals */
     cSize = ZSTD_compressSequences(&zc->seqStore,
@@ -2706,12 +2446,22 @@ static size_t ZSTD_compressBlock_internal(ZSTD_CCtx* zc,
             zc->entropyWorkspace, HUF_WORKSPACE_SIZE /* statically allocated in resetCCtx */,
             zc->bmi2);
 
+    if (frame &&
+        /* We don't want to emit our first block as a RLE even if it qualifies because
+         * doing so will cause the decoder (cli only) to throw a "should consume all input error."
+         * This is only an issue for zstd <= v1.4.3
+         */
+        !zc->isFirstBlock &&
+        cSize < rleMaxLength &&
+        ZSTD_isRLE(ip, srcSize))
+    {
+        cSize = 1;
+        op[0] = ip[0];
+    }
+
 out:
-    if (!ZSTD_isError(cSize) && cSize != 0) {
-        /* confirm repcodes and entropy tables when emitting a compressed block */
-        ZSTD_compressedBlockState_t* const tmp = zc->blockState.prevCBlock;
-        zc->blockState.prevCBlock = zc->blockState.nextCBlock;
-        zc->blockState.nextCBlock = tmp;
+    if (!ZSTD_isError(cSize) && cSize > 1) {
+        ZSTD_confirmRepcodesAndEntropyTables(zc);
     }
     /* We check that dictionaries have offset codes available for the first
      * block. After the first block, the offcode table might not have large
@@ -2723,6 +2473,104 @@ out:
     return cSize;
 }
 
+static size_t ZSTD_compressBlock_targetCBlockSize_body(ZSTD_CCtx* zc,
+                               void* dst, size_t dstCapacity,
+                               const void* src, size_t srcSize,
+                               const size_t bss, U32 lastBlock)
+{
+    DEBUGLOG(6, "Attempting ZSTD_compressSuperBlock()");
+    if (bss == ZSTDbss_compress) {
+        if (/* We don't want to emit our first block as a RLE even if it qualifies because
+            * doing so will cause the decoder (cli only) to throw a "should consume all input error."
+            * This is only an issue for zstd <= v1.4.3
+            */
+            !zc->isFirstBlock &&
+            ZSTD_maybeRLE(&zc->seqStore) &&
+            ZSTD_isRLE((BYTE const*)src, srcSize))
+        {
+            return ZSTD_rleCompressBlock(dst, dstCapacity, *(BYTE const*)src, srcSize, lastBlock);
+        }
+        /* Attempt superblock compression.
+         *
+         * Note that compressed size of ZSTD_compressSuperBlock() is not bound by the
+         * standard ZSTD_compressBound(). This is a problem, because even if we have
+         * space now, taking an extra byte now could cause us to run out of space later
+         * and violate ZSTD_compressBound().
+         *
+         * Define blockBound(blockSize) = blockSize + ZSTD_blockHeaderSize.
+         *
+         * In order to respect ZSTD_compressBound() we must attempt to emit a raw
+         * uncompressed block in these cases:
+         *   * cSize == 0: Return code for an uncompressed block.
+         *   * cSize == dstSize_tooSmall: We may have expanded beyond blockBound(srcSize).
+         *     ZSTD_noCompressBlock() will return dstSize_tooSmall if we are really out of
+         *     output space.
+         *   * cSize >= blockBound(srcSize): We have expanded the block too much so
+         *     emit an uncompressed block.
+         */
+        {
+            size_t const cSize = ZSTD_compressSuperBlock(zc, dst, dstCapacity, src, srcSize, lastBlock);
+            if (cSize != ERROR(dstSize_tooSmall)) {
+                size_t const maxCSize = srcSize - ZSTD_minGain(srcSize, zc->appliedParams.cParams.strategy);
+                FORWARD_IF_ERROR(cSize, "ZSTD_compressSuperBlock failed");
+                if (cSize != 0 && cSize < maxCSize + ZSTD_blockHeaderSize) {
+                    ZSTD_confirmRepcodesAndEntropyTables(zc);
+                    return cSize;
+                }
+            }
+        }
+    }
+
+    DEBUGLOG(6, "Resorting to ZSTD_noCompressBlock()");
+    /* Superblock compression failed, attempt to emit a single no compress block.
+     * The decoder will be able to stream this block since it is uncompressed.
+     */
+    return ZSTD_noCompressBlock(dst, dstCapacity, src, srcSize, lastBlock);
+}
+
+static size_t ZSTD_compressBlock_targetCBlockSize(ZSTD_CCtx* zc,
+                               void* dst, size_t dstCapacity,
+                               const void* src, size_t srcSize,
+                               U32 lastBlock)
+{
+    size_t cSize = 0;
+    const size_t bss = ZSTD_buildSeqStore(zc, src, srcSize);
+    DEBUGLOG(5, "ZSTD_compressBlock_targetCBlockSize (dstCapacity=%u, dictLimit=%u, nextToUpdate=%u, srcSize=%zu)",
+                (unsigned)dstCapacity, (unsigned)zc->blockState.matchState.window.dictLimit, (unsigned)zc->blockState.matchState.nextToUpdate, srcSize);
+    FORWARD_IF_ERROR(bss, "ZSTD_buildSeqStore failed");
+
+    cSize = ZSTD_compressBlock_targetCBlockSize_body(zc, dst, dstCapacity, src, srcSize, bss, lastBlock);
+    FORWARD_IF_ERROR(cSize, "ZSTD_compressBlock_targetCBlockSize_body failed");
+
+    if (zc->blockState.prevCBlock->entropy.fse.offcode_repeatMode == FSE_repeat_valid)
+        zc->blockState.prevCBlock->entropy.fse.offcode_repeatMode = FSE_repeat_check;
+
+    return cSize;
+}
+
+static void ZSTD_overflowCorrectIfNeeded(ZSTD_matchState_t* ms,
+                                         ZSTD_cwksp* ws,
+                                         ZSTD_CCtx_params const* params,
+                                         void const* ip,
+                                         void const* iend)
+{
+    if (ZSTD_window_needOverflowCorrection(ms->window, iend)) {
+        U32 const maxDist = (U32)1 << params->cParams.windowLog;
+        U32 const cycleLog = ZSTD_cycleLog(params->cParams.chainLog, params->cParams.strategy);
+        U32 const correction = ZSTD_window_correctOverflow(&ms->window, cycleLog, maxDist, ip);
+        ZSTD_STATIC_ASSERT(ZSTD_CHAINLOG_MAX <= 30);
+        ZSTD_STATIC_ASSERT(ZSTD_WINDOWLOG_MAX_32 <= 30);
+        ZSTD_STATIC_ASSERT(ZSTD_WINDOWLOG_MAX <= 31);
+        ZSTD_cwksp_mark_tables_dirty(ws);
+        ZSTD_reduceIndex(ms, params, correction);
+        ZSTD_cwksp_mark_tables_clean(ws);
+        if (ms->nextToUpdate < correction) ms->nextToUpdate = 0;
+        else ms->nextToUpdate -= correction;
+        /* invalidate dictionaries on overflow correction */
+        ms->loadedDictEnd = 0;
+        ms->dictMatchState = NULL;
+    }
+}
 
 /*! ZSTD_compress_frameChunk() :
 *   Compress a chunk of data into one or multiple blocks.
@@ -2742,7 +2590,8 @@ static size_t ZSTD_compress_frameChunk (ZSTD_CCtx* cctx,
     BYTE* const ostart = (BYTE*)dst;
     BYTE* op = ostart;
     U32 const maxDist = (U32)1 << cctx->appliedParams.cParams.windowLog;
-    assert(cctx->appliedParams.cParams.windowLog <= 31);
+
+    assert(cctx->appliedParams.cParams.windowLog <= ZSTD_WINDOWLOG_MAX);
 
     DEBUGLOG(5, "ZSTD_compress_frameChunk (blockSize=%u)", (unsigned)blockSize);
     if (cctx->appliedParams.fParams.checksumFlag && srcSize)
@@ -2752,78 +2601,84 @@ static size_t ZSTD_compress_frameChunk (ZSTD_CCtx* cctx,
         ZSTD_matchState_t* const ms = &cctx->blockState.matchState;
         U32 const lastBlock = lastFrameChunk & (blockSize >= remaining);
 
-        if (dstCapacity < ZSTD_blockHeaderSize + MIN_CBLOCK_SIZE)
-            return ERROR(dstSize_tooSmall);   /* not enough space to store compressed block */
+        RETURN_ERROR_IF(dstCapacity < ZSTD_blockHeaderSize + MIN_CBLOCK_SIZE,
+                        dstSize_tooSmall,
+                        "not enough space to store compressed block");
         if (remaining < blockSize) blockSize = remaining;
 
-        if (ZSTD_window_needOverflowCorrection(ms->window, ip + blockSize)) {
-            U32 const cycleLog = ZSTD_cycleLog(cctx->appliedParams.cParams.chainLog, cctx->appliedParams.cParams.strategy);
-            U32 const correction = ZSTD_window_correctOverflow(&ms->window, cycleLog, maxDist, ip);
-            ZSTD_STATIC_ASSERT(ZSTD_CHAINLOG_MAX <= 30);
-            ZSTD_STATIC_ASSERT(ZSTD_WINDOWLOG_MAX_32 <= 30);
-            ZSTD_STATIC_ASSERT(ZSTD_WINDOWLOG_MAX <= 31);
-            ZSTD_reduceIndex(cctx, correction);
-            if (ms->nextToUpdate < correction) ms->nextToUpdate = 0;
-            else ms->nextToUpdate -= correction;
-            ms->loadedDictEnd = 0;
-            ms->dictMatchState = NULL;
-        }
-        ZSTD_window_enforceMaxDist(&ms->window, ip + blockSize, maxDist, &ms->loadedDictEnd, &ms->dictMatchState);
-        if (ms->nextToUpdate < ms->window.lowLimit) ms->nextToUpdate = ms->window.lowLimit;
+        ZSTD_overflowCorrectIfNeeded(
+            ms, &cctx->workspace, &cctx->appliedParams, ip, ip + blockSize);
+        ZSTD_checkDictValidity(&ms->window, ip + blockSize, maxDist, &ms->loadedDictEnd, &ms->dictMatchState);
 
-        {   size_t cSize = ZSTD_compressBlock_internal(cctx,
-                                op+ZSTD_blockHeaderSize, dstCapacity-ZSTD_blockHeaderSize,
-                                ip, blockSize);
-            if (ZSTD_isError(cSize)) return cSize;
+        /* Ensure hash/chain table insertion resumes no sooner than lowlimit */
+        if (ms->nextToUpdate < ms->window.lowLimit) ms->nextToUpdate = ms->window.lowLimit;
 
-            if (cSize == 0) {  /* block is not compressible */
-                cSize = ZSTD_noCompressBlock(op, dstCapacity, ip, blockSize, lastBlock);
-                if (ZSTD_isError(cSize)) return cSize;
+        {   size_t cSize;
+            if (ZSTD_useTargetCBlockSize(&cctx->appliedParams)) {
+                cSize = ZSTD_compressBlock_targetCBlockSize(cctx, op, dstCapacity, ip, blockSize, lastBlock);
+                FORWARD_IF_ERROR(cSize, "ZSTD_compressBlock_targetCBlockSize failed");
+                assert(cSize > 0);
+                assert(cSize <= blockSize + ZSTD_blockHeaderSize);
             } else {
-                U32 const cBlockHeader24 = lastBlock + (((U32)bt_compressed)<<1) + (U32)(cSize << 3);
-                MEM_writeLE24(op, cBlockHeader24);
-                cSize += ZSTD_blockHeaderSize;
+                cSize = ZSTD_compressBlock_internal(cctx,
+                                        op+ZSTD_blockHeaderSize, dstCapacity-ZSTD_blockHeaderSize,
+                                        ip, blockSize, 1 /* frame */);
+                FORWARD_IF_ERROR(cSize, "ZSTD_compressBlock_internal failed");
+
+                if (cSize == 0) {  /* block is not compressible */
+                    cSize = ZSTD_noCompressBlock(op, dstCapacity, ip, blockSize, lastBlock);
+                    FORWARD_IF_ERROR(cSize, "ZSTD_noCompressBlock failed");
+                } else {
+                    U32 const cBlockHeader = cSize == 1 ?
+                        lastBlock + (((U32)bt_rle)<<1) + (U32)(blockSize << 3) :
+                        lastBlock + (((U32)bt_compressed)<<1) + (U32)(cSize << 3);
+                    MEM_writeLE24(op, cBlockHeader);
+                    cSize += ZSTD_blockHeaderSize;
+                }
             }
 
+
             ip += blockSize;
             assert(remaining >= blockSize);
             remaining -= blockSize;
             op += cSize;
             assert(dstCapacity >= cSize);
             dstCapacity -= cSize;
+            cctx->isFirstBlock = 0;
             DEBUGLOG(5, "ZSTD_compress_frameChunk: adding a block of size %u",
                         (unsigned)cSize);
     }   }
 
     if (lastFrameChunk && (op>ostart)) cctx->stage = ZSTDcs_ending;
-    return op-ostart;
+    return (size_t)(op-ostart);
 }
 
 
 static size_t ZSTD_writeFrameHeader(void* dst, size_t dstCapacity,
-                                    ZSTD_CCtx_params params, U64 pledgedSrcSize, U32 dictID)
+                                    const ZSTD_CCtx_params* params, U64 pledgedSrcSize, U32 dictID)
 {   BYTE* const op = (BYTE*)dst;
     U32   const dictIDSizeCodeLength = (dictID>0) + (dictID>=256) + (dictID>=65536);   /* 0-3 */
-    U32   const dictIDSizeCode = params.fParams.noDictIDFlag ? 0 : dictIDSizeCodeLength;   /* 0-3 */
-    U32   const checksumFlag = params.fParams.checksumFlag>0;
-    U32   const windowSize = (U32)1 << params.cParams.windowLog;
-    U32   const singleSegment = params.fParams.contentSizeFlag && (windowSize >= pledgedSrcSize);
-    BYTE  const windowLogByte = (BYTE)((params.cParams.windowLog - ZSTD_WINDOWLOG_ABSOLUTEMIN) << 3);
-    U32   const fcsCode = params.fParams.contentSizeFlag ?
+    U32   const dictIDSizeCode = params->fParams.noDictIDFlag ? 0 : dictIDSizeCodeLength;   /* 0-3 */
+    U32   const checksumFlag = params->fParams.checksumFlag>0;
+    U32   const windowSize = (U32)1 << params->cParams.windowLog;
+    U32   const singleSegment = params->fParams.contentSizeFlag && (windowSize >= pledgedSrcSize);
+    BYTE  const windowLogByte = (BYTE)((params->cParams.windowLog - ZSTD_WINDOWLOG_ABSOLUTEMIN) << 3);
+    U32   const fcsCode = params->fParams.contentSizeFlag ?
                      (pledgedSrcSize>=256) + (pledgedSrcSize>=65536+256) + (pledgedSrcSize>=0xFFFFFFFFU) : 0;  /* 0-3 */
-    BYTE  const frameHeaderDecriptionByte = (BYTE)(dictIDSizeCode + (checksumFlag<<2) + (singleSegment<<5) + (fcsCode<<6) );
+    BYTE  const frameHeaderDescriptionByte = (BYTE)(dictIDSizeCode + (checksumFlag<<2) + (singleSegment<<5) + (fcsCode<<6) );
     size_t pos=0;
 
-    assert(!(params.fParams.contentSizeFlag && pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN));
-    if (dstCapacity < ZSTD_FRAMEHEADERSIZE_MAX) return ERROR(dstSize_tooSmall);
+    assert(!(params->fParams.contentSizeFlag && pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN));
+    RETURN_ERROR_IF(dstCapacity < ZSTD_FRAMEHEADERSIZE_MAX, dstSize_tooSmall,
+                    "dst buf is too small to fit worst-case frame header size.");
     DEBUGLOG(4, "ZSTD_writeFrameHeader : dictIDFlag : %u ; dictID : %u ; dictIDSizeCode : %u",
-                !params.fParams.noDictIDFlag, (unsigned)dictID, (unsigned)dictIDSizeCode);
+                !params->fParams.noDictIDFlag, (unsigned)dictID, (unsigned)dictIDSizeCode);
 
-    if (params.format == ZSTD_f_zstd1) {
+    if (params->format == ZSTD_f_zstd1) {
         MEM_writeLE32(dst, ZSTD_MAGICNUMBER);
         pos = 4;
     }
-    op[pos++] = frameHeaderDecriptionByte;
+    op[pos++] = frameHeaderDescriptionByte;
     if (!singleSegment) op[pos++] = windowLogByte;
     switch(dictIDSizeCode)
     {
@@ -2847,11 +2702,12 @@ static size_t ZSTD_writeFrameHeader(void* dst, size_t dstCapacity,
 /* ZSTD_writeLastEmptyBlock() :
  * output an empty Block with end-of-frame mark to complete a frame
  * @return : size of data written into `dst` (== ZSTD_blockHeaderSize (defined in zstd_internal.h))
- *           or an error code if `dstCapcity` is too small (<ZSTD_blockHeaderSize)
+ *           or an error code if `dstCapacity` is too small (<ZSTD_blockHeaderSize)
  */
 size_t ZSTD_writeLastEmptyBlock(void* dst, size_t dstCapacity)
 {
-    if (dstCapacity < ZSTD_blockHeaderSize) return ERROR(dstSize_tooSmall);
+    RETURN_ERROR_IF(dstCapacity < ZSTD_blockHeaderSize, dstSize_tooSmall,
+                    "dst buf is too small to write frame trailer empty block.");
     {   U32 const cBlockHeader24 = 1 /*lastBlock*/ + (((U32)bt_raw)<<1);  /* 0 size */
         MEM_writeLE24(dst, cBlockHeader24);
         return ZSTD_blockHeaderSize;
@@ -2860,10 +2716,11 @@ size_t ZSTD_writeLastEmptyBlock(void* dst, size_t dstCapacity)
 
 size_t ZSTD_referenceExternalSequences(ZSTD_CCtx* cctx, rawSeq* seq, size_t nbSeq)
 {
-    if (cctx->stage != ZSTDcs_init)
-        return ERROR(stage_wrong);
-    if (cctx->appliedParams.ldmParams.enableLdm)
-        return ERROR(parameter_unsupported);
+    RETURN_ERROR_IF(cctx->stage != ZSTDcs_init, stage_wrong,
+                    "wrong cctx stage");
+    RETURN_ERROR_IF(cctx->appliedParams.ldmParams.enableLdm,
+                    parameter_unsupported,
+                    "incompatible with ldm");
     cctx->externSeqStore.seq = seq;
     cctx->externSeqStore.size = nbSeq;
     cctx->externSeqStore.capacity = nbSeq;
@@ -2882,12 +2739,14 @@ static size_t ZSTD_compressContinue_internal (ZSTD_CCtx* cctx,
 
     DEBUGLOG(5, "ZSTD_compressContinue_internal, stage: %u, srcSize: %u",
                 cctx->stage, (unsigned)srcSize);
-    if (cctx->stage==ZSTDcs_created) return ERROR(stage_wrong);   /* missing init (ZSTD_compressBegin) */
+    RETURN_ERROR_IF(cctx->stage==ZSTDcs_created, stage_wrong,
+                    "missing init (ZSTD_compressBegin)");
 
     if (frame && (cctx->stage==ZSTDcs_init)) {
-        fhSize = ZSTD_writeFrameHeader(dst, dstCapacity, cctx->appliedParams,
+        fhSize = ZSTD_writeFrameHeader(dst, dstCapacity, &cctx->appliedParams,
                                        cctx->pledgedSrcSizePlusOne-1, cctx->dictID);
-        if (ZSTD_isError(fhSize)) return fhSize;
+        FORWARD_IF_ERROR(fhSize, "ZSTD_writeFrameHeader failed");
+        assert(fhSize <= dstCapacity);
         dstCapacity -= fhSize;
         dst = (char*)dst + fhSize;
         cctx->stage = ZSTDcs_ongoing;
@@ -2904,35 +2763,27 @@ static size_t ZSTD_compressContinue_internal (ZSTD_CCtx* cctx,
 
     if (!frame) {
         /* overflow check and correction for block mode */
-        if (ZSTD_window_needOverflowCorrection(ms->window, (const char*)src + srcSize)) {
-            U32 const cycleLog = ZSTD_cycleLog(cctx->appliedParams.cParams.chainLog, cctx->appliedParams.cParams.strategy);
-            U32 const correction = ZSTD_window_correctOverflow(&ms->window, cycleLog, 1 << cctx->appliedParams.cParams.windowLog, src);
-            ZSTD_STATIC_ASSERT(ZSTD_CHAINLOG_MAX <= 30);
-            ZSTD_STATIC_ASSERT(ZSTD_WINDOWLOG_MAX_32 <= 30);
-            ZSTD_STATIC_ASSERT(ZSTD_WINDOWLOG_MAX <= 31);
-            ZSTD_reduceIndex(cctx, correction);
-            if (ms->nextToUpdate < correction) ms->nextToUpdate = 0;
-            else ms->nextToUpdate -= correction;
-            ms->loadedDictEnd = 0;
-            ms->dictMatchState = NULL;
-        }
+        ZSTD_overflowCorrectIfNeeded(
+            ms, &cctx->workspace, &cctx->appliedParams,
+            src, (BYTE const*)src + srcSize);
     }
 
     DEBUGLOG(5, "ZSTD_compressContinue_internal (blockSize=%u)", (unsigned)cctx->blockSize);
     {   size_t const cSize = frame ?
                              ZSTD_compress_frameChunk (cctx, dst, dstCapacity, src, srcSize, lastFrameChunk) :
-                             ZSTD_compressBlock_internal (cctx, dst, dstCapacity, src, srcSize);
-        if (ZSTD_isError(cSize)) return cSize;
+                             ZSTD_compressBlock_internal (cctx, dst, dstCapacity, src, srcSize, 0 /* frame */);
+        FORWARD_IF_ERROR(cSize, "%s", frame ? "ZSTD_compress_frameChunk failed" : "ZSTD_compressBlock_internal failed");
         cctx->consumedSrcSize += srcSize;
         cctx->producedCSize += (cSize + fhSize);
         assert(!(cctx->appliedParams.fParams.contentSizeFlag && cctx->pledgedSrcSizePlusOne == 0));
         if (cctx->pledgedSrcSizePlusOne != 0) {  /* control src size */
             ZSTD_STATIC_ASSERT(ZSTD_CONTENTSIZE_UNKNOWN == (unsigned long long)-1);
-            if (cctx->consumedSrcSize+1 > cctx->pledgedSrcSizePlusOne) {
-                DEBUGLOG(4, "error : pledgedSrcSize = %u, while realSrcSize >= %u",
-                    (unsigned)cctx->pledgedSrcSizePlusOne-1, (unsigned)cctx->consumedSrcSize);
-                return ERROR(srcSize_wrong);
-            }
+            RETURN_ERROR_IF(
+                cctx->consumedSrcSize+1 > cctx->pledgedSrcSizePlusOne,
+                srcSize_wrong,
+                "error : pledgedSrcSize = %u, while realSrcSize >= %u",
+                (unsigned)cctx->pledgedSrcSizePlusOne-1,
+                (unsigned)cctx->consumedSrcSize);
         }
         return cSize + fhSize;
     }
@@ -2956,8 +2807,9 @@ size_t ZSTD_getBlockSize(const ZSTD_CCtx* cctx)
 
 size_t ZSTD_compressBlock(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const void* src, size_t srcSize)
 {
-    size_t const blockSizeMax = ZSTD_getBlockSize(cctx);
-    if (srcSize > blockSizeMax) return ERROR(srcSize_wrong);
+    DEBUGLOG(5, "ZSTD_compressBlock: srcSize = %u", (unsigned)srcSize);
+    { size_t const blockSizeMax = ZSTD_getBlockSize(cctx);
+      RETURN_ERROR_IF(srcSize > blockSizeMax, srcSize_wrong, "input is larger than a block"); }
 
     return ZSTD_compressContinue_internal(cctx, dst, dstCapacity, src, srcSize, 0 /* frame mode */, 0 /* last chunk */);
 }
@@ -2966,47 +2818,67 @@ size_t ZSTD_compressBlock(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity, const
  *  @return : 0, or an error code
  */
 static size_t ZSTD_loadDictionaryContent(ZSTD_matchState_t* ms,
+                                         ldmState_t* ls,
+                                         ZSTD_cwksp* ws,
                                          ZSTD_CCtx_params const* params,
                                          const void* src, size_t srcSize,
                                          ZSTD_dictTableLoadMethod_e dtlm)
 {
-    const BYTE* const ip = (const BYTE*) src;
+    const BYTE* ip = (const BYTE*) src;
     const BYTE* const iend = ip + srcSize;
 
     ZSTD_window_update(&ms->window, src, srcSize);
     ms->loadedDictEnd = params->forceWindow ? 0 : (U32)(iend - ms->window.base);
 
+    if (params->ldmParams.enableLdm && ls != NULL) {
+        ZSTD_window_update(&ls->window, src, srcSize);
+        ls->loadedDictEnd = params->forceWindow ? 0 : (U32)(iend - ls->window.base);
+    }
+
     /* Assert that we the ms params match the params we're being given */
     ZSTD_assertEqualCParams(params->cParams, ms->cParams);
 
     if (srcSize <= HASH_READ_SIZE) return 0;
 
-    switch(params->cParams.strategy)
-    {
-    case ZSTD_fast:
-        ZSTD_fillHashTable(ms, iend, dtlm);
-        break;
-    case ZSTD_dfast:
-        ZSTD_fillDoubleHashTable(ms, iend, dtlm);
-        break;
+    while (iend - ip > HASH_READ_SIZE) {
+        size_t const remaining = (size_t)(iend - ip);
+        size_t const chunk = MIN(remaining, ZSTD_CHUNKSIZE_MAX);
+        const BYTE* const ichunk = ip + chunk;
 
-    case ZSTD_greedy:
-    case ZSTD_lazy:
-    case ZSTD_lazy2:
-        if (srcSize >= HASH_READ_SIZE)
-            ZSTD_insertAndFindFirstIndex(ms, iend-HASH_READ_SIZE);
-        break;
+        ZSTD_overflowCorrectIfNeeded(ms, ws, params, ip, ichunk);
 
-    case ZSTD_btlazy2:   /* we want the dictionary table fully sorted */
-    case ZSTD_btopt:
-    case ZSTD_btultra:
-    case ZSTD_btultra2:
-        if (srcSize >= HASH_READ_SIZE)
-            ZSTD_updateTree(ms, iend-HASH_READ_SIZE, iend);
-        break;
+        if (params->ldmParams.enableLdm && ls != NULL)
+            ZSTD_ldm_fillHashTable(ls, (const BYTE*)src, (const BYTE*)src + srcSize, &params->ldmParams);
 
-    default:
-        assert(0);  /* not possible : not a valid strategy id */
+        switch(params->cParams.strategy)
+        {
+        case ZSTD_fast:
+            ZSTD_fillHashTable(ms, ichunk, dtlm);
+            break;
+        case ZSTD_dfast:
+            ZSTD_fillDoubleHashTable(ms, ichunk, dtlm);
+            break;
+
+        case ZSTD_greedy:
+        case ZSTD_lazy:
+        case ZSTD_lazy2:
+            if (chunk >= HASH_READ_SIZE)
+                ZSTD_insertAndFindFirstIndex(ms, ichunk-HASH_READ_SIZE);
+            break;
+
+        case ZSTD_btlazy2:   /* we want the dictionary table fully sorted */
+        case ZSTD_btopt:
+        case ZSTD_btultra:
+        case ZSTD_btultra2:
+            if (chunk >= HASH_READ_SIZE)
+                ZSTD_updateTree(ms, ichunk-HASH_READ_SIZE, ichunk);
+            break;
+
+        default:
+            assert(0);  /* not possible : not a valid strategy id */
+        }
+
+        ip = ichunk;
     }
 
     ms->nextToUpdate = (U32)(iend - ms->window.base);
@@ -3020,98 +2892,123 @@ static size_t ZSTD_loadDictionaryContent(ZSTD_matchState_t* ms,
    NOTE: This behavior is not standard and could be improved in the future. */
 static size_t ZSTD_checkDictNCount(short* normalizedCounter, unsigned dictMaxSymbolValue, unsigned maxSymbolValue) {
     U32 s;
-    if (dictMaxSymbolValue < maxSymbolValue) return ERROR(dictionary_corrupted);
+    RETURN_ERROR_IF(dictMaxSymbolValue < maxSymbolValue, dictionary_corrupted, "dict fse tables don't have all symbols");
     for (s = 0; s <= maxSymbolValue; ++s) {
-        if (normalizedCounter[s] == 0) return ERROR(dictionary_corrupted);
+        RETURN_ERROR_IF(normalizedCounter[s] == 0, dictionary_corrupted, "dict fse tables don't have all symbols");
     }
     return 0;
 }
 
-
-/* Dictionary format :
- * See :
- * https://github.com/facebook/zstd/blob/master/doc/zstd_compression_format.md#dictionary-format
- */
-/*! ZSTD_loadZstdDictionary() :
- * @return : dictID, or an error code
- *  assumptions : magic number supposed already checked
- *                dictSize supposed > 8
- */
-static size_t ZSTD_loadZstdDictionary(ZSTD_compressedBlockState_t* bs,
-                                      ZSTD_matchState_t* ms,
-                                      ZSTD_CCtx_params const* params,
-                                      const void* dict, size_t dictSize,
-                                      ZSTD_dictTableLoadMethod_e dtlm,
-                                      void* workspace)
+size_t ZSTD_loadCEntropy(ZSTD_compressedBlockState_t* bs, void* workspace,
+                         short* offcodeNCount, unsigned* offcodeMaxValue,
+                         const void* const dict, size_t dictSize)
 {
-    const BYTE* dictPtr = (const BYTE*)dict;
+    const BYTE* dictPtr = (const BYTE*)dict;    /* skip magic num and dict ID */
     const BYTE* const dictEnd = dictPtr + dictSize;
-    short offcodeNCount[MaxOff+1];
-    unsigned offcodeMaxValue = MaxOff;
-    size_t dictID;
+    dictPtr += 8;
+    bs->entropy.huf.repeatMode = HUF_repeat_check;
 
-    ZSTD_STATIC_ASSERT(HUF_WORKSPACE_SIZE >= (1<<MAX(MLFSELog,LLFSELog)));
-    assert(dictSize > 8);
-    assert(MEM_readLE32(dictPtr) == ZSTD_MAGIC_DICTIONARY);
+    {   unsigned maxSymbolValue = 255;
+        unsigned hasZeroWeights = 1;
+        size_t const hufHeaderSize = HUF_readCTable((HUF_CElt*)bs->entropy.huf.CTable, &maxSymbolValue, dictPtr,
+            dictEnd-dictPtr, &hasZeroWeights);
 
-    dictPtr += 4;   /* skip magic number */
-    dictID = params->fParams.noDictIDFlag ? 0 :  MEM_readLE32(dictPtr);
-    dictPtr += 4;
+        /* We only set the loaded table as valid if it contains all non-zero
+         * weights. Otherwise, we set it to check */
+        if (!hasZeroWeights)
+            bs->entropy.huf.repeatMode = HUF_repeat_valid;
 
-    {   unsigned maxSymbolValue = 255;
-        size_t const hufHeaderSize = HUF_readCTable((HUF_CElt*)bs->entropy.huf.CTable, &maxSymbolValue, dictPtr, dictEnd-dictPtr);
-        if (HUF_isError(hufHeaderSize)) return ERROR(dictionary_corrupted);
-        if (maxSymbolValue < 255) return ERROR(dictionary_corrupted);
+        RETURN_ERROR_IF(HUF_isError(hufHeaderSize), dictionary_corrupted, "");
+        RETURN_ERROR_IF(maxSymbolValue < 255, dictionary_corrupted, "");
         dictPtr += hufHeaderSize;
     }
 
     {   unsigned offcodeLog;
-        size_t const offcodeHeaderSize = FSE_readNCount(offcodeNCount, &offcodeMaxValue, &offcodeLog, dictPtr, dictEnd-dictPtr);
-        if (FSE_isError(offcodeHeaderSize)) return ERROR(dictionary_corrupted);
-        if (offcodeLog > OffFSELog) return ERROR(dictionary_corrupted);
+        size_t const offcodeHeaderSize = FSE_readNCount(offcodeNCount, offcodeMaxValue, &offcodeLog, dictPtr, dictEnd-dictPtr);
+        RETURN_ERROR_IF(FSE_isError(offcodeHeaderSize), dictionary_corrupted, "");
+        RETURN_ERROR_IF(offcodeLog > OffFSELog, dictionary_corrupted, "");
         /* Defer checking offcodeMaxValue because we need to know the size of the dictionary content */
         /* fill all offset symbols to avoid garbage at end of table */
-        CHECK_E( FSE_buildCTable_wksp(bs->entropy.fse.offcodeCTable,
-                                    offcodeNCount, MaxOff, offcodeLog,
-                                    workspace, HUF_WORKSPACE_SIZE),
-                 dictionary_corrupted);
+        RETURN_ERROR_IF(FSE_isError(FSE_buildCTable_wksp(
+                bs->entropy.fse.offcodeCTable,
+                offcodeNCount, MaxOff, offcodeLog,
+                workspace, HUF_WORKSPACE_SIZE)),
+            dictionary_corrupted, "");
         dictPtr += offcodeHeaderSize;
     }
 
     {   short matchlengthNCount[MaxML+1];
         unsigned matchlengthMaxValue = MaxML, matchlengthLog;
         size_t const matchlengthHeaderSize = FSE_readNCount(matchlengthNCount, &matchlengthMaxValue, &matchlengthLog, dictPtr, dictEnd-dictPtr);
-        if (FSE_isError(matchlengthHeaderSize)) return ERROR(dictionary_corrupted);
-        if (matchlengthLog > MLFSELog) return ERROR(dictionary_corrupted);
+        RETURN_ERROR_IF(FSE_isError(matchlengthHeaderSize), dictionary_corrupted, "");
+        RETURN_ERROR_IF(matchlengthLog > MLFSELog, dictionary_corrupted, "");
         /* Every match length code must have non-zero probability */
-        CHECK_F( ZSTD_checkDictNCount(matchlengthNCount, matchlengthMaxValue, MaxML));
-        CHECK_E( FSE_buildCTable_wksp(bs->entropy.fse.matchlengthCTable,
-                                    matchlengthNCount, matchlengthMaxValue, matchlengthLog,
-                                    workspace, HUF_WORKSPACE_SIZE),
-                 dictionary_corrupted);
+        FORWARD_IF_ERROR( ZSTD_checkDictNCount(matchlengthNCount, matchlengthMaxValue, MaxML), "");
+        RETURN_ERROR_IF(FSE_isError(FSE_buildCTable_wksp(
+                bs->entropy.fse.matchlengthCTable,
+                matchlengthNCount, matchlengthMaxValue, matchlengthLog,
+                workspace, HUF_WORKSPACE_SIZE)),
+            dictionary_corrupted, "");
         dictPtr += matchlengthHeaderSize;
     }
 
     {   short litlengthNCount[MaxLL+1];
         unsigned litlengthMaxValue = MaxLL, litlengthLog;
         size_t const litlengthHeaderSize = FSE_readNCount(litlengthNCount, &litlengthMaxValue, &litlengthLog, dictPtr, dictEnd-dictPtr);
-        if (FSE_isError(litlengthHeaderSize)) return ERROR(dictionary_corrupted);
-        if (litlengthLog > LLFSELog) return ERROR(dictionary_corrupted);
+        RETURN_ERROR_IF(FSE_isError(litlengthHeaderSize), dictionary_corrupted, "");
+        RETURN_ERROR_IF(litlengthLog > LLFSELog, dictionary_corrupted, "");
         /* Every literal length code must have non-zero probability */
-        CHECK_F( ZSTD_checkDictNCount(litlengthNCount, litlengthMaxValue, MaxLL));
-        CHECK_E( FSE_buildCTable_wksp(bs->entropy.fse.litlengthCTable,
-                                    litlengthNCount, litlengthMaxValue, litlengthLog,
-                                    workspace, HUF_WORKSPACE_SIZE),
-                 dictionary_corrupted);
+        FORWARD_IF_ERROR( ZSTD_checkDictNCount(litlengthNCount, litlengthMaxValue, MaxLL), "");
+        RETURN_ERROR_IF(FSE_isError(FSE_buildCTable_wksp(
+                bs->entropy.fse.litlengthCTable,
+                litlengthNCount, litlengthMaxValue, litlengthLog,
+                workspace, HUF_WORKSPACE_SIZE)),
+            dictionary_corrupted, "");
         dictPtr += litlengthHeaderSize;
     }
 
-    if (dictPtr+12 > dictEnd) return ERROR(dictionary_corrupted);
+    RETURN_ERROR_IF(dictPtr+12 > dictEnd, dictionary_corrupted, "");
     bs->rep[0] = MEM_readLE32(dictPtr+0);
     bs->rep[1] = MEM_readLE32(dictPtr+4);
     bs->rep[2] = MEM_readLE32(dictPtr+8);
     dictPtr += 12;
 
+    return dictPtr - (const BYTE*)dict;
+}
+
+/* Dictionary format :
+ * See :
+ * https://github.com/facebook/zstd/blob/master/doc/zstd_compression_format.md#dictionary-format
+ */
+/*! ZSTD_loadZstdDictionary() :
+ * @return : dictID, or an error code
+ *  assumptions : magic number supposed already checked
+ *                dictSize supposed >= 8
+ */
+static size_t ZSTD_loadZstdDictionary(ZSTD_compressedBlockState_t* bs,
+                                      ZSTD_matchState_t* ms,
+                                      ZSTD_cwksp* ws,
+                                      ZSTD_CCtx_params const* params,
+                                      const void* dict, size_t dictSize,
+                                      ZSTD_dictTableLoadMethod_e dtlm,
+                                      void* workspace)
+{
+    const BYTE* dictPtr = (const BYTE*)dict;
+    const BYTE* const dictEnd = dictPtr + dictSize;
+    short offcodeNCount[MaxOff+1];
+    unsigned offcodeMaxValue = MaxOff;
+    size_t dictID;
+    size_t eSize;
+
+    ZSTD_STATIC_ASSERT(HUF_WORKSPACE_SIZE >= (1<<MAX(MLFSELog,LLFSELog)));
+    assert(dictSize >= 8);
+    assert(MEM_readLE32(dictPtr) == ZSTD_MAGIC_DICTIONARY);
+
+    dictID = params->fParams.noDictIDFlag ? 0 :  MEM_readLE32(dictPtr + 4 /* skip magic number */ );
+    eSize = ZSTD_loadCEntropy(bs, workspace, offcodeNCount, &offcodeMaxValue, dict, dictSize);
+    FORWARD_IF_ERROR(eSize, "ZSTD_loadCEntropy failed");
+    dictPtr += eSize;
+
     {   size_t const dictContentSize = (size_t)(dictEnd - dictPtr);
         U32 offcodeMax = MaxOff;
         if (dictContentSize <= ((U32)-1) - 128 KB) {
@@ -3119,19 +3016,19 @@ static size_t ZSTD_loadZstdDictionary(ZSTD_compressedBlockState_t* bs,
             offcodeMax = ZSTD_highbit32(maxOffset); /* Calculate minimum offset code required to represent maxOffset */
         }
         /* All offset values <= dictContentSize + 128 KB must be representable */
-        CHECK_F (ZSTD_checkDictNCount(offcodeNCount, offcodeMaxValue, MIN(offcodeMax, MaxOff)));
+        FORWARD_IF_ERROR(ZSTD_checkDictNCount(offcodeNCount, offcodeMaxValue, MIN(offcodeMax, MaxOff)), "");
         /* All repCodes must be <= dictContentSize and != 0*/
         {   U32 u;
             for (u=0; u<3; u++) {
-                if (bs->rep[u] == 0) return ERROR(dictionary_corrupted);
-                if (bs->rep[u] > dictContentSize) return ERROR(dictionary_corrupted);
+                RETURN_ERROR_IF(bs->rep[u] == 0, dictionary_corrupted, "");
+                RETURN_ERROR_IF(bs->rep[u] > dictContentSize, dictionary_corrupted, "");
         }   }
 
-        bs->entropy.huf.repeatMode = HUF_repeat_valid;
         bs->entropy.fse.offcode_repeatMode = FSE_repeat_valid;
         bs->entropy.fse.matchlength_repeatMode = FSE_repeat_valid;
         bs->entropy.fse.litlength_repeatMode = FSE_repeat_valid;
-        CHECK_F(ZSTD_loadDictionaryContent(ms, params, dictPtr, dictContentSize, dtlm));
+        FORWARD_IF_ERROR(ZSTD_loadDictionaryContent(
+            ms, NULL, ws, params, dictPtr, dictContentSize, dtlm), "");
         return dictID;
     }
 }
@@ -3141,6 +3038,8 @@ static size_t ZSTD_loadZstdDictionary(ZSTD_compressedBlockState_t* bs,
 static size_t
 ZSTD_compress_insertDictionary(ZSTD_compressedBlockState_t* bs,
                                ZSTD_matchState_t* ms,
+                               ldmState_t* ls,
+                               ZSTD_cwksp* ws,
                          const ZSTD_CCtx_params* params,
                          const void* dict, size_t dictSize,
                                ZSTD_dictContentType_e dictContentType,
@@ -3148,28 +3047,35 @@ ZSTD_compress_insertDictionary(ZSTD_compressedBlockState_t* bs,
                                void* workspace)
 {
     DEBUGLOG(4, "ZSTD_compress_insertDictionary (dictSize=%u)", (U32)dictSize);
-    if ((dict==NULL) || (dictSize<=8)) return 0;
+    if ((dict==NULL) || (dictSize<8)) {
+        RETURN_ERROR_IF(dictContentType == ZSTD_dct_fullDict, dictionary_wrong, "");
+        return 0;
+    }
 
     ZSTD_reset_compressedBlockState(bs);
 
     /* dict restricted modes */
     if (dictContentType == ZSTD_dct_rawContent)
-        return ZSTD_loadDictionaryContent(ms, params, dict, dictSize, dtlm);
+        return ZSTD_loadDictionaryContent(ms, ls, ws, params, dict, dictSize, dtlm);
 
     if (MEM_readLE32(dict) != ZSTD_MAGIC_DICTIONARY) {
         if (dictContentType == ZSTD_dct_auto) {
             DEBUGLOG(4, "raw content dictionary detected");
-            return ZSTD_loadDictionaryContent(ms, params, dict, dictSize, dtlm);
+            return ZSTD_loadDictionaryContent(
+                ms, ls, ws, params, dict, dictSize, dtlm);
         }
-        if (dictContentType == ZSTD_dct_fullDict)
-            return ERROR(dictionary_wrong);
+        RETURN_ERROR_IF(dictContentType == ZSTD_dct_fullDict, dictionary_wrong, "");
         assert(0);   /* impossible */
     }
 
     /* dict as full zstd dictionary */
-    return ZSTD_loadZstdDictionary(bs, ms, params, dict, dictSize, dtlm, workspace);
+    return ZSTD_loadZstdDictionary(
+        bs, ms, ws, params, dict, dictSize, dtlm, workspace);
 }
 
+#define ZSTD_USE_CDICT_PARAMS_SRCSIZE_CUTOFF (128 KB)
+#define ZSTD_USE_CDICT_PARAMS_DICTSIZE_MULTIPLIER (6)
+
 /*! ZSTD_compressBegin_internal() :
  * @return : 0, or an error code */
 static size_t ZSTD_compressBegin_internal(ZSTD_CCtx* cctx,
@@ -3177,26 +3083,37 @@ static size_t ZSTD_compressBegin_internal(ZSTD_CCtx* cctx,
                                     ZSTD_dictContentType_e dictContentType,
                                     ZSTD_dictTableLoadMethod_e dtlm,
                                     const ZSTD_CDict* cdict,
-                                    ZSTD_CCtx_params params, U64 pledgedSrcSize,
+                                    const ZSTD_CCtx_params* params, U64 pledgedSrcSize,
                                     ZSTD_buffered_policy_e zbuff)
 {
-    DEBUGLOG(4, "ZSTD_compressBegin_internal: wlog=%u", params.cParams.windowLog);
+    DEBUGLOG(4, "ZSTD_compressBegin_internal: wlog=%u", params->cParams.windowLog);
     /* params are supposed to be fully validated at this point */
-    assert(!ZSTD_isError(ZSTD_checkCParams(params.cParams)));
+    assert(!ZSTD_isError(ZSTD_checkCParams(params->cParams)));
     assert(!((dict) && (cdict)));  /* either dict or cdict, not both */
-
-    if (cdict && cdict->dictContentSize>0) {
+    if ( (cdict)
+      && (cdict->dictContentSize > 0)
+      && ( pledgedSrcSize < ZSTD_USE_CDICT_PARAMS_SRCSIZE_CUTOFF
+        || pledgedSrcSize < cdict->dictContentSize * ZSTD_USE_CDICT_PARAMS_DICTSIZE_MULTIPLIER
+        || pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN
+        || cdict->compressionLevel == 0)
+      && (params->attachDictPref != ZSTD_dictForceLoad) ) {
         return ZSTD_resetCCtx_usingCDict(cctx, cdict, params, pledgedSrcSize, zbuff);
     }
 
-    CHECK_F( ZSTD_resetCCtx_internal(cctx, params, pledgedSrcSize,
-                                     ZSTDcrp_continue, zbuff) );
-    {
-        size_t const dictID = ZSTD_compress_insertDictionary(
-                cctx->blockState.prevCBlock, &cctx->blockState.matchState,
-                &params, dict, dictSize, dictContentType, dtlm, cctx->entropyWorkspace);
-        if (ZSTD_isError(dictID)) return dictID;
-        assert(dictID <= (size_t)(U32)-1);
+    FORWARD_IF_ERROR( ZSTD_resetCCtx_internal(cctx, *params, pledgedSrcSize,
+                                     ZSTDcrp_makeClean, zbuff) , "");
+    {   size_t const dictID = cdict ?
+                ZSTD_compress_insertDictionary(
+                        cctx->blockState.prevCBlock, &cctx->blockState.matchState,
+                        &cctx->ldmState, &cctx->workspace, &cctx->appliedParams, cdict->dictContent,
+                        cdict->dictContentSize, dictContentType, dtlm,
+                        cctx->entropyWorkspace)
+              : ZSTD_compress_insertDictionary(
+                        cctx->blockState.prevCBlock, &cctx->blockState.matchState,
+                        &cctx->ldmState, &cctx->workspace, &cctx->appliedParams, dict, dictSize,
+                        dictContentType, dtlm, cctx->entropyWorkspace);
+        FORWARD_IF_ERROR(dictID, "ZSTD_compress_insertDictionary failed");
+        assert(dictID <= UINT_MAX);
         cctx->dictID = (U32)dictID;
     }
     return 0;
@@ -3207,12 +3124,12 @@ size_t ZSTD_compressBegin_advanced_internal(ZSTD_CCtx* cctx,
                                     ZSTD_dictContentType_e dictContentType,
                                     ZSTD_dictTableLoadMethod_e dtlm,
                                     const ZSTD_CDict* cdict,
-                                    ZSTD_CCtx_params params,
+                                    const ZSTD_CCtx_params* params,
                                     unsigned long long pledgedSrcSize)
 {
-    DEBUGLOG(4, "ZSTD_compressBegin_advanced_internal: wlog=%u", params.cParams.windowLog);
+    DEBUGLOG(4, "ZSTD_compressBegin_advanced_internal: wlog=%u", params->cParams.windowLog);
     /* compression parameters verification and optimization */
-    CHECK_F( ZSTD_checkCParams(params.cParams) );
+    FORWARD_IF_ERROR( ZSTD_checkCParams(params->cParams) , "");
     return ZSTD_compressBegin_internal(cctx,
                                        dict, dictSize, dictContentType, dtlm,
                                        cdict,
@@ -3227,21 +3144,21 @@ size_t ZSTD_compressBegin_advanced(ZSTD_CCtx* cctx,
                                    ZSTD_parameters params, unsigned long long pledgedSrcSize)
 {
     ZSTD_CCtx_params const cctxParams =
-            ZSTD_assignParamsToCCtxParams(cctx->requestedParams, params);
+            ZSTD_assignParamsToCCtxParams(&cctx->requestedParams, &params);
     return ZSTD_compressBegin_advanced_internal(cctx,
                                             dict, dictSize, ZSTD_dct_auto, ZSTD_dtlm_fast,
                                             NULL /*cdict*/,
-                                            cctxParams, pledgedSrcSize);
+                                            &cctxParams, pledgedSrcSize);
 }
 
 size_t ZSTD_compressBegin_usingDict(ZSTD_CCtx* cctx, const void* dict, size_t dictSize, int compressionLevel)
 {
-    ZSTD_parameters const params = ZSTD_getParams(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, dictSize);
+    ZSTD_parameters const params = ZSTD_getParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, dictSize);
     ZSTD_CCtx_params const cctxParams =
-            ZSTD_assignParamsToCCtxParams(cctx->requestedParams, params);
+            ZSTD_assignParamsToCCtxParams(&cctx->requestedParams, &params);
     DEBUGLOG(4, "ZSTD_compressBegin_usingDict (dictSize=%u)", (unsigned)dictSize);
     return ZSTD_compressBegin_internal(cctx, dict, dictSize, ZSTD_dct_auto, ZSTD_dtlm_fast, NULL,
-                                       cctxParams, ZSTD_CONTENTSIZE_UNKNOWN, ZSTDb_not_buffered);
+                                       &cctxParams, ZSTD_CONTENTSIZE_UNKNOWN, ZSTDb_not_buffered);
 }
 
 size_t ZSTD_compressBegin(ZSTD_CCtx* cctx, int compressionLevel)
@@ -3260,12 +3177,12 @@ static size_t ZSTD_writeEpilogue(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity)
     size_t fhSize = 0;
 
     DEBUGLOG(4, "ZSTD_writeEpilogue");
-    if (cctx->stage == ZSTDcs_created) return ERROR(stage_wrong);  /* init missing */
+    RETURN_ERROR_IF(cctx->stage == ZSTDcs_created, stage_wrong, "init missing");
 
     /* special case : empty frame */
     if (cctx->stage == ZSTDcs_init) {
-        fhSize = ZSTD_writeFrameHeader(dst, dstCapacity, cctx->appliedParams, 0, 0);
-        if (ZSTD_isError(fhSize)) return fhSize;
+        fhSize = ZSTD_writeFrameHeader(dst, dstCapacity, &cctx->appliedParams, 0, 0);
+        FORWARD_IF_ERROR(fhSize, "ZSTD_writeFrameHeader failed");
         dstCapacity -= fhSize;
         op += fhSize;
         cctx->stage = ZSTDcs_ongoing;
@@ -3274,7 +3191,7 @@ static size_t ZSTD_writeEpilogue(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity)
     if (cctx->stage != ZSTDcs_ending) {
         /* write one last empty block, make it the "last" block */
         U32 const cBlockHeader24 = 1 /* last block */ + (((U32)bt_raw)<<1) + 0;
-        if (dstCapacity<4) return ERROR(dstSize_tooSmall);
+        RETURN_ERROR_IF(dstCapacity<4, dstSize_tooSmall, "no room for epilogue");
         MEM_writeLE32(op, cBlockHeader24);
         op += ZSTD_blockHeaderSize;
         dstCapacity -= ZSTD_blockHeaderSize;
@@ -3282,7 +3199,7 @@ static size_t ZSTD_writeEpilogue(ZSTD_CCtx* cctx, void* dst, size_t dstCapacity)
 
     if (cctx->appliedParams.fParams.checksumFlag) {
         U32 const checksum = (U32) XXH64_digest(&cctx->xxhState);
-        if (dstCapacity<4) return ERROR(dstSize_tooSmall);
+        RETURN_ERROR_IF(dstCapacity<4, dstSize_tooSmall, "no room for checksum");
         DEBUGLOG(4, "ZSTD_writeEpilogue: write checksum : %08X", (unsigned)checksum);
         MEM_writeLE32(op, checksum);
         op += 4;
@@ -3300,18 +3217,20 @@ size_t ZSTD_compressEnd (ZSTD_CCtx* cctx,
     size_t const cSize = ZSTD_compressContinue_internal(cctx,
                                 dst, dstCapacity, src, srcSize,
                                 1 /* frame mode */, 1 /* last chunk */);
-    if (ZSTD_isError(cSize)) return cSize;
+    FORWARD_IF_ERROR(cSize, "ZSTD_compressContinue_internal failed");
     endResult = ZSTD_writeEpilogue(cctx, (char*)dst + cSize, dstCapacity-cSize);
-    if (ZSTD_isError(endResult)) return endResult;
+    FORWARD_IF_ERROR(endResult, "ZSTD_writeEpilogue failed");
     assert(!(cctx->appliedParams.fParams.contentSizeFlag && cctx->pledgedSrcSizePlusOne == 0));
     if (cctx->pledgedSrcSizePlusOne != 0) {  /* control src size */
         ZSTD_STATIC_ASSERT(ZSTD_CONTENTSIZE_UNKNOWN == (unsigned long long)-1);
         DEBUGLOG(4, "end of frame : controlling src size");
-        if (cctx->pledgedSrcSizePlusOne != cctx->consumedSrcSize+1) {
-            DEBUGLOG(4, "error : pledgedSrcSize = %u, while realSrcSize = %u",
-                (unsigned)cctx->pledgedSrcSizePlusOne-1, (unsigned)cctx->consumedSrcSize);
-            return ERROR(srcSize_wrong);
-    }   }
+        RETURN_ERROR_IF(
+            cctx->pledgedSrcSizePlusOne != cctx->consumedSrcSize+1,
+            srcSize_wrong,
+             "error : pledgedSrcSize = %u, while realSrcSize = %u",
+            (unsigned)cctx->pledgedSrcSizePlusOne-1,
+            (unsigned)cctx->consumedSrcSize);
+    }
     return cSize + endResult;
 }
 
@@ -3320,16 +3239,16 @@ static size_t ZSTD_compress_internal (ZSTD_CCtx* cctx,
                                       void* dst, size_t dstCapacity,
                                 const void* src, size_t srcSize,
                                 const void* dict,size_t dictSize,
-                                      ZSTD_parameters params)
+                                const ZSTD_parameters* params)
 {
     ZSTD_CCtx_params const cctxParams =
-            ZSTD_assignParamsToCCtxParams(cctx->requestedParams, params);
+            ZSTD_assignParamsToCCtxParams(&cctx->requestedParams, params);
     DEBUGLOG(4, "ZSTD_compress_internal");
     return ZSTD_compress_advanced_internal(cctx,
                                            dst, dstCapacity,
                                            src, srcSize,
                                            dict, dictSize,
-                                           cctxParams);
+                                           &cctxParams);
 }
 
 size_t ZSTD_compress_advanced (ZSTD_CCtx* cctx,
@@ -3339,12 +3258,12 @@ size_t ZSTD_compress_advanced (ZSTD_CCtx* cctx,
                                ZSTD_parameters params)
 {
     DEBUGLOG(4, "ZSTD_compress_advanced");
-    CHECK_F(ZSTD_checkCParams(params.cParams));
+    FORWARD_IF_ERROR(ZSTD_checkCParams(params.cParams), "");
     return ZSTD_compress_internal(cctx,
                                   dst, dstCapacity,
                                   src, srcSize,
                                   dict, dictSize,
-                                  params);
+                                  &params);
 }
 
 /* Internal */
@@ -3353,12 +3272,12 @@ size_t ZSTD_compress_advanced_internal(
         void* dst, size_t dstCapacity,
         const void* src, size_t srcSize,
         const void* dict,size_t dictSize,
-        ZSTD_CCtx_params params)
+        const ZSTD_CCtx_params* params)
 {
     DEBUGLOG(4, "ZSTD_compress_advanced_internal (srcSize:%u)", (unsigned)srcSize);
-    CHECK_F( ZSTD_compressBegin_internal(cctx,
+    FORWARD_IF_ERROR( ZSTD_compressBegin_internal(cctx,
                          dict, dictSize, ZSTD_dct_auto, ZSTD_dtlm_fast, NULL,
-                         params, srcSize, ZSTDb_not_buffered) );
+                         params, srcSize, ZSTDb_not_buffered) , "");
     return ZSTD_compressEnd(cctx, dst, dstCapacity, src, srcSize);
 }
 
@@ -3368,10 +3287,11 @@ size_t ZSTD_compress_usingDict(ZSTD_CCtx* cctx,
                          const void* dict, size_t dictSize,
                                int compressionLevel)
 {
-    ZSTD_parameters const params = ZSTD_getParams(compressionLevel, srcSize + (!srcSize), dict ? dictSize : 0);
-    ZSTD_CCtx_params cctxParams = ZSTD_assignParamsToCCtxParams(cctx->requestedParams, params);
+    ZSTD_parameters const params = ZSTD_getParams_internal(compressionLevel, srcSize, dict ? dictSize : 0);
+    ZSTD_CCtx_params cctxParams = ZSTD_assignParamsToCCtxParams(&cctx->requestedParams, &params);
+    DEBUGLOG(4, "ZSTD_compress_usingDict (srcSize=%u)", (unsigned)srcSize);
     assert(params.fParams.contentSizeFlag == 1);
-    return ZSTD_compress_advanced_internal(cctx, dst, dstCapacity, src, srcSize, dict, dictSize, cctxParams);
+    return ZSTD_compress_advanced_internal(cctx, dst, dstCapacity, src, srcSize, dict, dictSize, &cctxParams);
 }
 
 size_t ZSTD_compressCCtx(ZSTD_CCtx* cctx,
@@ -3406,13 +3326,16 @@ size_t ZSTD_estimateCDictSize_advanced(
         ZSTD_dictLoadMethod_e dictLoadMethod)
 {
     DEBUGLOG(5, "sizeof(ZSTD_CDict) : %u", (unsigned)sizeof(ZSTD_CDict));
-    return sizeof(ZSTD_CDict) + HUF_WORKSPACE_SIZE + ZSTD_sizeof_matchState(&cParams, /* forCCtx */ 0)
-           + (dictLoadMethod == ZSTD_dlm_byRef ? 0 : dictSize);
+    return ZSTD_cwksp_alloc_size(sizeof(ZSTD_CDict))
+         + ZSTD_cwksp_alloc_size(HUF_WORKSPACE_SIZE)
+         + ZSTD_sizeof_matchState(&cParams, /* forCCtx */ 0)
+         + (dictLoadMethod == ZSTD_dlm_byRef ? 0
+            : ZSTD_cwksp_alloc_size(ZSTD_cwksp_align(dictSize, sizeof(void *))));
 }
 
 size_t ZSTD_estimateCDictSize(size_t dictSize, int compressionLevel)
 {
-    ZSTD_compressionParameters const cParams = ZSTD_getCParams(compressionLevel, 0, dictSize);
+    ZSTD_compressionParameters const cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, dictSize);
     return ZSTD_estimateCDictSize_advanced(dictSize, cParams, ZSTD_dlm_byCopy);
 }
 
@@ -3420,7 +3343,9 @@ size_t ZSTD_sizeof_CDict(const ZSTD_CDict* cdict)
 {
     if (cdict==NULL) return 0;   /* support sizeof on NULL */
     DEBUGLOG(5, "sizeof(*cdict) : %u", (unsigned)sizeof(*cdict));
-    return cdict->workspaceSize + (cdict->dictBuffer ? cdict->dictContentSize : 0) + sizeof(*cdict);
+    /* cdict may be in the workspace */
+    return (cdict->workspace.workspace == cdict ? 0 : sizeof(*cdict))
+        + ZSTD_cwksp_sizeof(&cdict->workspace);
 }
 
 static size_t ZSTD_initCDict_internal(
@@ -3434,28 +3359,29 @@ static size_t ZSTD_initCDict_internal(
     assert(!ZSTD_checkCParams(cParams));
     cdict->matchState.cParams = cParams;
     if ((dictLoadMethod == ZSTD_dlm_byRef) || (!dictBuffer) || (!dictSize)) {
-        cdict->dictBuffer = NULL;
         cdict->dictContent = dictBuffer;
     } else {
-        void* const internalBuffer = ZSTD_malloc(dictSize, cdict->customMem);
-        cdict->dictBuffer = internalBuffer;
+         void *internalBuffer = ZSTD_cwksp_reserve_object(&cdict->workspace, ZSTD_cwksp_align(dictSize, sizeof(void*)));
+        RETURN_ERROR_IF(!internalBuffer, memory_allocation, "NULL pointer!");
         cdict->dictContent = internalBuffer;
-        if (!internalBuffer) return ERROR(memory_allocation);
         memcpy(internalBuffer, dictBuffer, dictSize);
     }
     cdict->dictContentSize = dictSize;
 
+    cdict->entropyWorkspace = (U32*)ZSTD_cwksp_reserve_object(&cdict->workspace, HUF_WORKSPACE_SIZE);
+
+
     /* Reset the state to no dictionary */
     ZSTD_reset_compressedBlockState(&cdict->cBlockState);
-    {   void* const end = ZSTD_reset_matchState(
-                &cdict->matchState,
-                (U32*)cdict->workspace + HUF_WORKSPACE_SIZE_U32,
-                &cParams, ZSTDcrp_continue, /* forCCtx */ 0);
-        assert(end == (char*)cdict->workspace + cdict->workspaceSize);
-        (void)end;
-    }
+    FORWARD_IF_ERROR(ZSTD_reset_matchState(
+        &cdict->matchState,
+        &cdict->workspace,
+        &cParams,
+        ZSTDcrp_makeClean,
+        ZSTDirp_reset,
+        ZSTD_resetTarget_CDict), "");
     /* (Maybe) load the dictionary
-     * Skips loading the dictionary if it is <= 8 bytes.
+     * Skips loading the dictionary if it is < 8 bytes.
      */
     {   ZSTD_CCtx_params params;
         memset(&params, 0, sizeof(params));
@@ -3463,10 +3389,10 @@ static size_t ZSTD_initCDict_internal(
         params.fParams.contentSizeFlag = 1;
         params.cParams = cParams;
         {   size_t const dictID = ZSTD_compress_insertDictionary(
-                    &cdict->cBlockState, &cdict->matchState, &params,
-                    cdict->dictContent, cdict->dictContentSize,
-                    dictContentType, ZSTD_dtlm_full, cdict->workspace);
-            if (ZSTD_isError(dictID)) return dictID;
+                    &cdict->cBlockState, &cdict->matchState, NULL, &cdict->workspace,
+                    &params, cdict->dictContent, cdict->dictContentSize,
+                    dictContentType, ZSTD_dtlm_full, cdict->entropyWorkspace);
+            FORWARD_IF_ERROR(dictID, "ZSTD_compress_insertDictionary failed");
             assert(dictID <= (size_t)(U32)-1);
             cdict->dictID = (U32)dictID;
         }
@@ -3483,18 +3409,29 @@ ZSTD_CDict* ZSTD_createCDict_advanced(const void* dictBuffer, size_t dictSize,
     DEBUGLOG(3, "ZSTD_createCDict_advanced, mode %u", (unsigned)dictContentType);
     if (!customMem.customAlloc ^ !customMem.customFree) return NULL;
 
-    {   ZSTD_CDict* const cdict = (ZSTD_CDict*)ZSTD_malloc(sizeof(ZSTD_CDict), customMem);
-        size_t const workspaceSize = HUF_WORKSPACE_SIZE + ZSTD_sizeof_matchState(&cParams, /* forCCtx */ 0);
+    {   size_t const workspaceSize =
+            ZSTD_cwksp_alloc_size(sizeof(ZSTD_CDict)) +
+            ZSTD_cwksp_alloc_size(HUF_WORKSPACE_SIZE) +
+            ZSTD_sizeof_matchState(&cParams, /* forCCtx */ 0) +
+            (dictLoadMethod == ZSTD_dlm_byRef ? 0
+             : ZSTD_cwksp_alloc_size(ZSTD_cwksp_align(dictSize, sizeof(void*))));
         void* const workspace = ZSTD_malloc(workspaceSize, customMem);
+        ZSTD_cwksp ws;
+        ZSTD_CDict* cdict;
 
-        if (!cdict || !workspace) {
-            ZSTD_free(cdict, customMem);
+        if (!workspace) {
             ZSTD_free(workspace, customMem);
             return NULL;
         }
+
+        ZSTD_cwksp_init(&ws, workspace, workspaceSize);
+
+        cdict = (ZSTD_CDict*)ZSTD_cwksp_reserve_object(&ws, sizeof(ZSTD_CDict));
+        assert(cdict != NULL);
+        ZSTD_cwksp_move(&cdict->workspace, &ws);
         cdict->customMem = customMem;
-        cdict->workspace = workspace;
-        cdict->workspaceSize = workspaceSize;
+        cdict->compressionLevel = 0; /* signals advanced API usage */
+
         if (ZSTD_isError( ZSTD_initCDict_internal(cdict,
                                         dictBuffer, dictSize,
                                         dictLoadMethod, dictContentType,
@@ -3509,15 +3446,18 @@ ZSTD_CDict* ZSTD_createCDict_advanced(const void* dictBuffer, size_t dictSize,
 
 ZSTD_CDict* ZSTD_createCDict(const void* dict, size_t dictSize, int compressionLevel)
 {
-    ZSTD_compressionParameters cParams = ZSTD_getCParams(compressionLevel, 0, dictSize);
-    return ZSTD_createCDict_advanced(dict, dictSize,
-                                     ZSTD_dlm_byCopy, ZSTD_dct_auto,
-                                     cParams, ZSTD_defaultCMem);
+    ZSTD_compressionParameters cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, dictSize);
+    ZSTD_CDict* cdict = ZSTD_createCDict_advanced(dict, dictSize,
+                                                  ZSTD_dlm_byCopy, ZSTD_dct_auto,
+                                                  cParams, ZSTD_defaultCMem);
+    if (cdict)
+        cdict->compressionLevel = compressionLevel == 0 ? ZSTD_CLEVEL_DEFAULT : compressionLevel;
+    return cdict;
 }
 
 ZSTD_CDict* ZSTD_createCDict_byReference(const void* dict, size_t dictSize, int compressionLevel)
 {
-    ZSTD_compressionParameters cParams = ZSTD_getCParams(compressionLevel, 0, dictSize);
+    ZSTD_compressionParameters cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, dictSize);
     return ZSTD_createCDict_advanced(dict, dictSize,
                                      ZSTD_dlm_byRef, ZSTD_dct_auto,
                                      cParams, ZSTD_defaultCMem);
@@ -3527,9 +3467,11 @@ size_t ZSTD_freeCDict(ZSTD_CDict* cdict)
 {
     if (cdict==NULL) return 0;   /* support free on NULL */
     {   ZSTD_customMem const cMem = cdict->customMem;
-        ZSTD_free(cdict->workspace, cMem);
-        ZSTD_free(cdict->dictBuffer, cMem);
-        ZSTD_free(cdict, cMem);
+        int cdictInWorkspace = ZSTD_cwksp_owns_buffer(&cdict->workspace, cdict);
+        ZSTD_cwksp_free(&cdict->workspace, cMem);
+        if (!cdictInWorkspace) {
+            ZSTD_free(cdict, cMem);
+        }
         return 0;
     }
 }
@@ -3555,28 +3497,30 @@ const ZSTD_CDict* ZSTD_initStaticCDict(
                                  ZSTD_compressionParameters cParams)
 {
     size_t const matchStateSize = ZSTD_sizeof_matchState(&cParams, /* forCCtx */ 0);
-    size_t const neededSize = sizeof(ZSTD_CDict) + (dictLoadMethod == ZSTD_dlm_byRef ? 0 : dictSize)
-                            + HUF_WORKSPACE_SIZE + matchStateSize;
-    ZSTD_CDict* const cdict = (ZSTD_CDict*) workspace;
-    void* ptr;
+    size_t const neededSize = ZSTD_cwksp_alloc_size(sizeof(ZSTD_CDict))
+                            + (dictLoadMethod == ZSTD_dlm_byRef ? 0
+                               : ZSTD_cwksp_alloc_size(ZSTD_cwksp_align(dictSize, sizeof(void*))))
+                            + ZSTD_cwksp_alloc_size(HUF_WORKSPACE_SIZE)
+                            + matchStateSize;
+    ZSTD_CDict* cdict;
+
     if ((size_t)workspace & 7) return NULL;  /* 8-aligned */
+
+    {
+        ZSTD_cwksp ws;
+        ZSTD_cwksp_init(&ws, workspace, workspaceSize);
+        cdict = (ZSTD_CDict*)ZSTD_cwksp_reserve_object(&ws, sizeof(ZSTD_CDict));
+        if (cdict == NULL) return NULL;
+        ZSTD_cwksp_move(&cdict->workspace, &ws);
+    }
+
     DEBUGLOG(4, "(workspaceSize < neededSize) : (%u < %u) => %u",
         (unsigned)workspaceSize, (unsigned)neededSize, (unsigned)(workspaceSize < neededSize));
     if (workspaceSize < neededSize) return NULL;
 
-    if (dictLoadMethod == ZSTD_dlm_byCopy) {
-        memcpy(cdict+1, dict, dictSize);
-        dict = cdict+1;
-        ptr = (char*)workspace + sizeof(ZSTD_CDict) + dictSize;
-    } else {
-        ptr = cdict+1;
-    }
-    cdict->workspace = ptr;
-    cdict->workspaceSize = HUF_WORKSPACE_SIZE + matchStateSize;
-
     if (ZSTD_isError( ZSTD_initCDict_internal(cdict,
                                               dict, dictSize,
-                                              ZSTD_dlm_byRef, dictContentType,
+                                              dictLoadMethod, dictContentType,
                                               cParams) ))
         return NULL;
 
@@ -3596,9 +3540,17 @@ size_t ZSTD_compressBegin_usingCDict_advanced(
     ZSTD_frameParameters const fParams, unsigned long long const pledgedSrcSize)
 {
     DEBUGLOG(4, "ZSTD_compressBegin_usingCDict_advanced");
-    if (cdict==NULL) return ERROR(dictionary_wrong);
+    RETURN_ERROR_IF(cdict==NULL, dictionary_wrong, "NULL pointer!");
     {   ZSTD_CCtx_params params = cctx->requestedParams;
-        params.cParams = ZSTD_getCParamsFromCDict(cdict);
+        params.cParams = ( pledgedSrcSize < ZSTD_USE_CDICT_PARAMS_SRCSIZE_CUTOFF
+                        || pledgedSrcSize < cdict->dictContentSize * ZSTD_USE_CDICT_PARAMS_DICTSIZE_MULTIPLIER
+                        || pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN
+                        || cdict->compressionLevel == 0 )
+                      && (params.attachDictPref != ZSTD_dictForceLoad) ?
+                ZSTD_getCParamsFromCDict(cdict)
+              : ZSTD_getCParams(cdict->compressionLevel,
+                                pledgedSrcSize,
+                                cdict->dictContentSize);
         /* Increase window log to fit the entire dictionary and source if the
          * source size is known. Limit the increase to 19, which is the
          * window log for compression level 1 with the largest source size.
@@ -3612,7 +3564,7 @@ size_t ZSTD_compressBegin_usingCDict_advanced(
         return ZSTD_compressBegin_internal(cctx,
                                            NULL, 0, ZSTD_dct_auto, ZSTD_dtlm_fast,
                                            cdict,
-                                           params, pledgedSrcSize,
+                                           &params, pledgedSrcSize,
                                            ZSTDb_not_buffered);
     }
 }
@@ -3632,7 +3584,7 @@ size_t ZSTD_compress_usingCDict_advanced(ZSTD_CCtx* cctx,
                                 const void* src, size_t srcSize,
                                 const ZSTD_CDict* cdict, ZSTD_frameParameters fParams)
 {
-    CHECK_F (ZSTD_compressBegin_usingCDict_advanced(cctx, cdict, fParams, srcSize));   /* will check if cdict != NULL */
+    FORWARD_IF_ERROR(ZSTD_compressBegin_usingCDict_advanced(cctx, cdict, fParams, srcSize), "");   /* will check if cdict != NULL */
     return ZSTD_compressEnd(cctx, dst, dstCapacity, src, srcSize);
 }
 
@@ -3700,11 +3652,11 @@ static size_t ZSTD_resetCStream_internal(ZSTD_CStream* cctx,
     assert(!ZSTD_isError(ZSTD_checkCParams(params.cParams)));
     assert(!((dict) && (cdict)));  /* either dict or cdict, not both */
 
-    CHECK_F( ZSTD_compressBegin_internal(cctx,
+    FORWARD_IF_ERROR( ZSTD_compressBegin_internal(cctx,
                                          dict, dictSize, dictContentType, ZSTD_dtlm_fast,
                                          cdict,
-                                         params, pledgedSrcSize,
-                                         ZSTDb_buffered) );
+                                         &params, pledgedSrcSize,
+                                         ZSTDb_buffered) , "");
 
     cctx->inToCompress = 0;
     cctx->inBuffPos = 0;
@@ -3718,13 +3670,17 @@ static size_t ZSTD_resetCStream_internal(ZSTD_CStream* cctx,
 
 /* ZSTD_resetCStream():
  * pledgedSrcSize == 0 means "unknown" */
-size_t ZSTD_resetCStream(ZSTD_CStream* zcs, unsigned long long pledgedSrcSize)
+size_t ZSTD_resetCStream(ZSTD_CStream* zcs, unsigned long long pss)
 {
-    ZSTD_CCtx_params params = zcs->requestedParams;
+    /* temporary : 0 interpreted as "unknown" during transition period.
+     * Users willing to specify "unknown" **must** use ZSTD_CONTENTSIZE_UNKNOWN.
+     * 0 will be interpreted as "empty" in the future.
+     */
+    U64 const pledgedSrcSize = (pss==0) ? ZSTD_CONTENTSIZE_UNKNOWN : pss;
     DEBUGLOG(4, "ZSTD_resetCStream: pledgedSrcSize = %u", (unsigned)pledgedSrcSize);
-    if (pledgedSrcSize==0) pledgedSrcSize = ZSTD_CONTENTSIZE_UNKNOWN;
-    params.fParams.contentSizeFlag = 1;
-    return ZSTD_resetCStream_internal(zcs, NULL, 0, ZSTD_dct_auto, zcs->cdict, params, pledgedSrcSize);
+    FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , "");
+    return 0;
 }
 
 /*! ZSTD_initCStream_internal() :
@@ -3733,35 +3689,22 @@ size_t ZSTD_resetCStream(ZSTD_CStream* zcs, unsigned long long pledgedSrcSize)
  *  Assumption 2 : either dict, or cdict, is defined, not both */
 size_t ZSTD_initCStream_internal(ZSTD_CStream* zcs,
                     const void* dict, size_t dictSize, const ZSTD_CDict* cdict,
-                    ZSTD_CCtx_params params, unsigned long long pledgedSrcSize)
+                    const ZSTD_CCtx_params* params,
+                    unsigned long long pledgedSrcSize)
 {
     DEBUGLOG(4, "ZSTD_initCStream_internal");
-    params.cParams = ZSTD_getCParamsFromCCtxParams(&params, pledgedSrcSize, dictSize);
-    assert(!ZSTD_isError(ZSTD_checkCParams(params.cParams)));
+    FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , "");
+    assert(!ZSTD_isError(ZSTD_checkCParams(params->cParams)));
+    zcs->requestedParams = *params;
     assert(!((dict) && (cdict)));  /* either dict or cdict, not both */
-
-    if (dict && dictSize >= 8) {
-        DEBUGLOG(4, "loading dictionary of size %u", (unsigned)dictSize);
-        if (zcs->staticSize) {   /* static CCtx : never uses malloc */
-            /* incompatible with internal cdict creation */
-            return ERROR(memory_allocation);
-        }
-        ZSTD_freeCDict(zcs->cdictLocal);
-        zcs->cdictLocal = ZSTD_createCDict_advanced(dict, dictSize,
-                                            ZSTD_dlm_byCopy, ZSTD_dct_auto,
-                                            params.cParams, zcs->customMem);
-        zcs->cdict = zcs->cdictLocal;
-        if (zcs->cdictLocal == NULL) return ERROR(memory_allocation);
+    if (dict) {
+        FORWARD_IF_ERROR( ZSTD_CCtx_loadDictionary(zcs, dict, dictSize) , "");
     } else {
-        if (cdict) {
-            params.cParams = ZSTD_getCParamsFromCDict(cdict);  /* cParams are enforced from cdict; it includes windowLog */
-        }
-        ZSTD_freeCDict(zcs->cdictLocal);
-        zcs->cdictLocal = NULL;
-        zcs->cdict = cdict;
+        /* Dictionary is cleared if !cdict */
+        FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, cdict) , "");
     }
-
-    return ZSTD_resetCStream_internal(zcs, NULL, 0, ZSTD_dct_auto, zcs->cdict, params, pledgedSrcSize);
+    return 0;
 }
 
 /* ZSTD_initCStream_usingCDict_advanced() :
@@ -3772,58 +3715,76 @@ size_t ZSTD_initCStream_usingCDict_advanced(ZSTD_CStream* zcs,
                                             unsigned long long pledgedSrcSize)
 {
     DEBUGLOG(4, "ZSTD_initCStream_usingCDict_advanced");
-    if (!cdict) return ERROR(dictionary_wrong); /* cannot handle NULL cdict (does not know what to do) */
-    {   ZSTD_CCtx_params params = zcs->requestedParams;
-        params.cParams = ZSTD_getCParamsFromCDict(cdict);
-        params.fParams = fParams;
-        return ZSTD_initCStream_internal(zcs,
-                                NULL, 0, cdict,
-                                params, pledgedSrcSize);
-    }
+    FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , "");
+    zcs->requestedParams.fParams = fParams;
+    FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, cdict) , "");
+    return 0;
 }
 
 /* note : cdict must outlive compression session */
 size_t ZSTD_initCStream_usingCDict(ZSTD_CStream* zcs, const ZSTD_CDict* cdict)
 {
-    ZSTD_frameParameters const fParams = { 0 /* contentSizeFlag */, 0 /* checksum */, 0 /* hideDictID */ };
     DEBUGLOG(4, "ZSTD_initCStream_usingCDict");
-    return ZSTD_initCStream_usingCDict_advanced(zcs, cdict, fParams, ZSTD_CONTENTSIZE_UNKNOWN);  /* note : will check that cdict != NULL */
+    FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, cdict) , "");
+    return 0;
 }
 
 
 /* ZSTD_initCStream_advanced() :
  * pledgedSrcSize must be exact.
  * if srcSize is not known at init time, use value ZSTD_CONTENTSIZE_UNKNOWN.
- * dict is loaded with default parameters ZSTD_dm_auto and ZSTD_dlm_byCopy. */
+ * dict is loaded with default parameters ZSTD_dct_auto and ZSTD_dlm_byCopy. */
 size_t ZSTD_initCStream_advanced(ZSTD_CStream* zcs,
                                  const void* dict, size_t dictSize,
-                                 ZSTD_parameters params, unsigned long long pledgedSrcSize)
+                                 ZSTD_parameters params, unsigned long long pss)
 {
-    DEBUGLOG(4, "ZSTD_initCStream_advanced: pledgedSrcSize=%u, flag=%u",
-                (unsigned)pledgedSrcSize, params.fParams.contentSizeFlag);
-    CHECK_F( ZSTD_checkCParams(params.cParams) );
-    if ((pledgedSrcSize==0) && (params.fParams.contentSizeFlag==0)) pledgedSrcSize = ZSTD_CONTENTSIZE_UNKNOWN;  /* for compatibility with older programs relying on this behavior. Users should now specify ZSTD_CONTENTSIZE_UNKNOWN. This line will be removed in the future. */
-    zcs->requestedParams = ZSTD_assignParamsToCCtxParams(zcs->requestedParams, params);
-    return ZSTD_initCStream_internal(zcs, dict, dictSize, NULL /*cdict*/, zcs->requestedParams, pledgedSrcSize);
+    /* for compatibility with older programs relying on this behavior.
+     * Users should now specify ZSTD_CONTENTSIZE_UNKNOWN.
+     * This line will be removed in the future.
+     */
+    U64 const pledgedSrcSize = (pss==0 && params.fParams.contentSizeFlag==0) ? ZSTD_CONTENTSIZE_UNKNOWN : pss;
+    DEBUGLOG(4, "ZSTD_initCStream_advanced");
+    FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , "");
+    FORWARD_IF_ERROR( ZSTD_checkCParams(params.cParams) , "");
+    zcs->requestedParams = ZSTD_assignParamsToCCtxParams(&zcs->requestedParams, &params);
+    FORWARD_IF_ERROR( ZSTD_CCtx_loadDictionary(zcs, dict, dictSize) , "");
+    return 0;
 }
 
 size_t ZSTD_initCStream_usingDict(ZSTD_CStream* zcs, const void* dict, size_t dictSize, int compressionLevel)
 {
-    ZSTD_CCtxParams_init(&zcs->requestedParams, compressionLevel);
-    return ZSTD_initCStream_internal(zcs, dict, dictSize, NULL, zcs->requestedParams, ZSTD_CONTENTSIZE_UNKNOWN);
+    DEBUGLOG(4, "ZSTD_initCStream_usingDict");
+    FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_loadDictionary(zcs, dict, dictSize) , "");
+    return 0;
 }
 
 size_t ZSTD_initCStream_srcSize(ZSTD_CStream* zcs, int compressionLevel, unsigned long long pss)
 {
-    U64 const pledgedSrcSize = (pss==0) ? ZSTD_CONTENTSIZE_UNKNOWN : pss;  /* temporary : 0 interpreted as "unknown" during transition period. Users willing to specify "unknown" **must** use ZSTD_CONTENTSIZE_UNKNOWN. `0` will be interpreted as "empty" in the future */
-    ZSTD_CCtxParams_init(&zcs->requestedParams, compressionLevel);
-    return ZSTD_initCStream_internal(zcs, NULL, 0, NULL, zcs->requestedParams, pledgedSrcSize);
+    /* temporary : 0 interpreted as "unknown" during transition period.
+     * Users willing to specify "unknown" **must** use ZSTD_CONTENTSIZE_UNKNOWN.
+     * 0 will be interpreted as "empty" in the future.
+     */
+    U64 const pledgedSrcSize = (pss==0) ? ZSTD_CONTENTSIZE_UNKNOWN : pss;
+    DEBUGLOG(4, "ZSTD_initCStream_srcSize");
+    FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, NULL) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_setPledgedSrcSize(zcs, pledgedSrcSize) , "");
+    return 0;
 }
 
 size_t ZSTD_initCStream(ZSTD_CStream* zcs, int compressionLevel)
 {
     DEBUGLOG(4, "ZSTD_initCStream");
-    return ZSTD_initCStream_srcSize(zcs, compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN);
+    FORWARD_IF_ERROR( ZSTD_CCtx_reset(zcs, ZSTD_reset_session_only) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_refCDict(zcs, NULL) , "");
+    FORWARD_IF_ERROR( ZSTD_CCtx_setParameter(zcs, ZSTD_c_compressionLevel, compressionLevel) , "");
+    return 0;
 }
 
 /*======   Compression   ======*/
@@ -3835,29 +3796,21 @@ static size_t ZSTD_nextInputSizeHint(const ZSTD_CCtx* cctx)
     return hintInSize;
 }
 
-static size_t ZSTD_limitCopy(void* dst, size_t dstCapacity,
-                       const void* src, size_t srcSize)
-{
-    size_t const length = MIN(dstCapacity, srcSize);
-    if (length) memcpy(dst, src, length);
-    return length;
-}
-
 /** ZSTD_compressStream_generic():
  *  internal function for all *compressStream*() variants
  *  non-static, because can be called from zstdmt_compress.c
  * @return : hint size for next input */
-size_t ZSTD_compressStream_generic(ZSTD_CStream* zcs,
-                                   ZSTD_outBuffer* output,
-                                   ZSTD_inBuffer* input,
-                                   ZSTD_EndDirective const flushMode)
+static size_t ZSTD_compressStream_generic(ZSTD_CStream* zcs,
+                                          ZSTD_outBuffer* output,
+                                          ZSTD_inBuffer* input,
+                                          ZSTD_EndDirective const flushMode)
 {
     const char* const istart = (const char*)input->src;
-    const char* const iend = istart + input->size;
-    const char* ip = istart + input->pos;
+    const char* const iend = input->size != 0 ? istart + input->size : istart;
+    const char* ip = input->pos != 0 ? istart + input->pos : istart;
     char* const ostart = (char*)output->dst;
-    char* const oend = ostart + output->size;
-    char* op = ostart + output->pos;
+    char* const oend = output->size != 0 ? ostart + output->size : ostart;
+    char* op = output->pos != 0 ? ostart + output->pos : ostart;
     U32 someMoreWork = 1;
 
     /* check expectations */
@@ -3873,8 +3826,7 @@ size_t ZSTD_compressStream_generic(ZSTD_CStream* zcs,
         switch(zcs->streamStage)
         {
         case zcss_init:
-            /* call ZSTD_initCStream() first ! */
-            return ERROR(init_missing);
+            RETURN_ERROR(init_missing, "call ZSTD_initCStream() first!");
 
         case zcss_load:
             if ( (flushMode == ZSTD_e_end)
@@ -3884,7 +3836,7 @@ size_t ZSTD_compressStream_generic(ZSTD_CStream* zcs,
                 size_t const cSize = ZSTD_compressEnd(zcs,
                                                 op, oend-op, ip, iend-ip);
                 DEBUGLOG(4, "ZSTD_compressEnd : cSize=%u", (unsigned)cSize);
-                if (ZSTD_isError(cSize)) return cSize;
+                FORWARD_IF_ERROR(cSize, "ZSTD_compressEnd failed");
                 ip = iend;
                 op += cSize;
                 zcs->frameEnded = 1;
@@ -3897,7 +3849,8 @@ size_t ZSTD_compressStream_generic(ZSTD_CStream* zcs,
                                         zcs->inBuff + zcs->inBuffPos, toLoad,
                                         ip, iend-ip);
                 zcs->inBuffPos += loaded;
-                ip += loaded;
+                if (loaded != 0)
+                    ip += loaded;
                 if ( (flushMode == ZSTD_e_continue)
                   && (zcs->inBuffPos < zcs->inBuffTarget) ) {
                     /* not enough input to fill full block : stop here */
@@ -3925,7 +3878,7 @@ size_t ZSTD_compressStream_generic(ZSTD_CStream* zcs,
                                     zcs->inBuff + zcs->inToCompress, iSize) :
                         ZSTD_compressContinue(zcs, cDst, oSize,
                                     zcs->inBuff + zcs->inToCompress, iSize);
-                if (ZSTD_isError(cSize)) return cSize;
+                FORWARD_IF_ERROR(cSize, "%s", lastBlock ? "ZSTD_compressEnd failed" : "ZSTD_compressContinue failed");
                 zcs->frameEnded = lastBlock;
                 /* prepare next block */
                 zcs->inBuffTarget = zcs->inBuffPos + zcs->blockSize;
@@ -3953,11 +3906,12 @@ size_t ZSTD_compressStream_generic(ZSTD_CStream* zcs,
         case zcss_flush:
             DEBUGLOG(5, "flush stage");
             {   size_t const toFlush = zcs->outBuffContentSize - zcs->outBuffFlushedSize;
-                size_t const flushed = ZSTD_limitCopy(op, oend-op,
+                size_t const flushed = ZSTD_limitCopy(op, (size_t)(oend-op),
                             zcs->outBuff + zcs->outBuffFlushedSize, toFlush);
                 DEBUGLOG(5, "toFlush: %u into %u ==> flushed: %u",
                             (unsigned)toFlush, (unsigned)(oend-op), (unsigned)flushed);
-                op += flushed;
+                if (flushed)
+                    op += flushed;
                 zcs->outBuffFlushedSize += flushed;
                 if (toFlush!=flushed) {
                     /* flush not fully completed, presumably because dst is too small */
@@ -4001,7 +3955,7 @@ static size_t ZSTD_nextInputSizeHint_MTorST(const ZSTD_CCtx* cctx)
 
 size_t ZSTD_compressStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output, ZSTD_inBuffer* input)
 {
-    CHECK_F( ZSTD_compressStream2(zcs, output, input, ZSTD_e_continue) );
+    FORWARD_IF_ERROR( ZSTD_compressStream2(zcs, output, input, ZSTD_e_continue) , "");
     return ZSTD_nextInputSizeHint_MTorST(zcs);
 }
 
@@ -4013,14 +3967,15 @@ size_t ZSTD_compressStream2( ZSTD_CCtx* cctx,
 {
     DEBUGLOG(5, "ZSTD_compressStream2, endOp=%u ", (unsigned)endOp);
     /* check conditions */
-    if (output->pos > output->size) return ERROR(GENERIC);
-    if (input->pos  > input->size)  return ERROR(GENERIC);
+    RETURN_ERROR_IF(output->pos > output->size, GENERIC, "invalid buffer");
+    RETURN_ERROR_IF(input->pos  > input->size, GENERIC, "invalid buffer");
     assert(cctx!=NULL);
 
     /* transparent initialization stage */
     if (cctx->streamStage == zcss_init) {
         ZSTD_CCtx_params params = cctx->requestedParams;
         ZSTD_prefixDict const prefixDict = cctx->prefixDict;
+        FORWARD_IF_ERROR( ZSTD_initLocalDict(cctx) , ""); /* Init the local dict if present. */
         memset(&cctx->prefixDict, 0, sizeof(cctx->prefixDict));   /* single usage */
         assert(prefixDict.dict==NULL || cctx->cdict==NULL);    /* only one can be set */
         DEBUGLOG(4, "ZSTD_compressStream2 : transparent init stage");
@@ -4038,23 +3993,23 @@ size_t ZSTD_compressStream2( ZSTD_CCtx* cctx,
             if (cctx->mtctx == NULL) {
                 DEBUGLOG(4, "ZSTD_compressStream2: creating new mtctx for nbWorkers=%u",
                             params.nbWorkers);
-                cctx->mtctx = ZSTDMT_createCCtx_advanced(params.nbWorkers, cctx->customMem);
-                if (cctx->mtctx == NULL) return ERROR(memory_allocation);
+                cctx->mtctx = ZSTDMT_createCCtx_advanced((U32)params.nbWorkers, cctx->customMem);
+                RETURN_ERROR_IF(cctx->mtctx == NULL, memory_allocation, "NULL pointer!");
             }
             /* mt compression */
             DEBUGLOG(4, "call ZSTDMT_initCStream_internal as nbWorkers=%u", params.nbWorkers);
-            CHECK_F( ZSTDMT_initCStream_internal(
+            FORWARD_IF_ERROR( ZSTDMT_initCStream_internal(
                         cctx->mtctx,
-                        prefixDict.dict, prefixDict.dictSize, ZSTD_dct_rawContent,
-                        cctx->cdict, params, cctx->pledgedSrcSizePlusOne-1) );
+                        prefixDict.dict, prefixDict.dictSize, prefixDict.dictContentType,
+                        cctx->cdict, params, cctx->pledgedSrcSizePlusOne-1) , "");
             cctx->streamStage = zcss_load;
             cctx->appliedParams.nbWorkers = params.nbWorkers;
         } else
 #endif
-        {   CHECK_F( ZSTD_resetCStream_internal(cctx,
+        {   FORWARD_IF_ERROR( ZSTD_resetCStream_internal(cctx,
                             prefixDict.dict, prefixDict.dictSize, prefixDict.dictContentType,
                             cctx->cdict,
-                            params, cctx->pledgedSrcSizePlusOne-1) );
+                            params, cctx->pledgedSrcSizePlusOne-1) , "");
             assert(cctx->streamStage == zcss_load);
             assert(cctx->appliedParams.nbWorkers == 0);
     }   }
@@ -4063,20 +4018,30 @@ size_t ZSTD_compressStream2( ZSTD_CCtx* cctx,
     /* compression stage */
 #ifdef ZSTD_MULTITHREAD
     if (cctx->appliedParams.nbWorkers > 0) {
+        int const forceMaxProgress = (endOp == ZSTD_e_flush || endOp == ZSTD_e_end);
+        size_t flushMin;
+        assert(forceMaxProgress || endOp == ZSTD_e_continue /* Protection for a new flush type */);
         if (cctx->cParamsChanged) {
             ZSTDMT_updateCParams_whileCompressing(cctx->mtctx, &cctx->requestedParams);
             cctx->cParamsChanged = 0;
         }
-        {   size_t const flushMin = ZSTDMT_compressStream_generic(cctx->mtctx, output, input, endOp);
+        do {
+            flushMin = ZSTDMT_compressStream_generic(cctx->mtctx, output, input, endOp);
             if ( ZSTD_isError(flushMin)
               || (endOp == ZSTD_e_end && flushMin == 0) ) { /* compression completed */
                 ZSTD_CCtx_reset(cctx, ZSTD_reset_session_only);
             }
-            DEBUGLOG(5, "completed ZSTD_compressStream2 delegating to ZSTDMT_compressStream_generic");
-            return flushMin;
-    }   }
+            FORWARD_IF_ERROR(flushMin, "ZSTDMT_compressStream_generic failed");
+        } while (forceMaxProgress && flushMin != 0 && output->pos < output->size);
+        DEBUGLOG(5, "completed ZSTD_compressStream2 delegating to ZSTDMT_compressStream_generic");
+        /* Either we don't require maximum forward progress, we've finished the
+         * flush, or we are out of output space.
+         */
+        assert(!forceMaxProgress || flushMin == 0 || output->pos == output->size);
+        return flushMin;
+    }
 #endif
-    CHECK_F( ZSTD_compressStream_generic(cctx, output, input, endOp) );
+    FORWARD_IF_ERROR( ZSTD_compressStream_generic(cctx, output, input, endOp) , "");
     DEBUGLOG(5, "completed ZSTD_compressStream2");
     return cctx->outBuffContentSize - cctx->outBuffFlushedSize; /* remaining to flush */
 }
@@ -4100,6 +4065,7 @@ size_t ZSTD_compress2(ZSTD_CCtx* cctx,
                       void* dst, size_t dstCapacity,
                       const void* src, size_t srcSize)
 {
+    DEBUGLOG(4, "ZSTD_compress2 (srcSize=%u)", (unsigned)srcSize);
     ZSTD_CCtx_reset(cctx, ZSTD_reset_session_only);
     {   size_t oPos = 0;
         size_t iPos = 0;
@@ -4107,10 +4073,10 @@ size_t ZSTD_compress2(ZSTD_CCtx* cctx,
                                         dst, dstCapacity, &oPos,
                                         src, srcSize, &iPos,
                                         ZSTD_e_end);
-        if (ZSTD_isError(result)) return result;
+        FORWARD_IF_ERROR(result, "ZSTD_compressStream2_simpleArgs failed");
         if (result != 0) {  /* compression not completed, due to lack of output space */
             assert(oPos == dstCapacity);
-            return ERROR(dstSize_tooSmall);
+            RETURN_ERROR(dstSize_tooSmall, "");
         }
         assert(iPos == srcSize);   /* all input is expected consumed */
         return oPos;
@@ -4132,11 +4098,11 @@ size_t ZSTD_endStream(ZSTD_CStream* zcs, ZSTD_outBuffer* output)
 {
     ZSTD_inBuffer input = { NULL, 0, 0 };
     size_t const remainingToFlush = ZSTD_compressStream2(zcs, output, &input, ZSTD_e_end);
-    CHECK_F( remainingToFlush );
+    FORWARD_IF_ERROR( remainingToFlush , "ZSTD_compressStream2 failed");
     if (zcs->appliedParams.nbWorkers > 0) return remainingToFlush;   /* minimal estimation */
     /* single thread mode : attempt to calculate remaining to flush more precisely */
     {   size_t const lastBlockSize = zcs->frameEnded ? 0 : ZSTD_BLOCKHEADERSIZE;
-        size_t const checksumSize = zcs->frameEnded ? 0 : zcs->appliedParams.fParams.checksumFlag * 4;
+        size_t const checksumSize = (size_t)(zcs->frameEnded ? 0 : zcs->appliedParams.fParams.checksumFlag * 4);
         size_t const toFlush = remainingToFlush + lastBlockSize + checksumSize;
         DEBUGLOG(4, "ZSTD_endStream : remaining to flush : %u", (unsigned)toFlush);
         return toFlush;
@@ -4151,13 +4117,13 @@ int ZSTD_maxCLevel(void) { return ZSTD_MAX_CLEVEL; }
 int ZSTD_minCLevel(void) { return (int)-ZSTD_TARGETLENGTH_MAX; }
 
 static const ZSTD_compressionParameters ZSTD_defaultCParameters[4][ZSTD_MAX_CLEVEL+1] = {
-{   /* "default" - guarantees a monotonically increasing memory budget */
+{   /* "default" - for any srcSize > 256 KB */
     /* W,  C,  H,  S,  L, TL, strat */
     { 19, 12, 13,  1,  6,  1, ZSTD_fast    },  /* base for negative levels */
     { 19, 13, 14,  1,  7,  0, ZSTD_fast    },  /* level  1 */
     { 20, 15, 16,  1,  6,  0, ZSTD_fast    },  /* level  2 */
-    { 21, 16, 17,  1,  5,  1, ZSTD_dfast   },  /* level  3 */
-    { 21, 18, 18,  1,  5,  1, ZSTD_dfast   },  /* level  4 */
+    { 21, 16, 17,  1,  5,  0, ZSTD_dfast   },  /* level  3 */
+    { 21, 18, 18,  1,  5,  0, ZSTD_dfast   },  /* level  4 */
     { 21, 18, 19,  2,  5,  2, ZSTD_greedy  },  /* level  5 */
     { 21, 19, 19,  3,  5,  4, ZSTD_greedy  },  /* level  6 */
     { 21, 19, 19,  3,  5,  8, ZSTD_lazy    },  /* level  7 */
@@ -4181,8 +4147,8 @@ static const ZSTD_compressionParameters ZSTD_defaultCParameters[4][ZSTD_MAX_CLEV
     /* W,  C,  H,  S,  L,  T, strat */
     { 18, 12, 13,  1,  5,  1, ZSTD_fast    },  /* base for negative levels */
     { 18, 13, 14,  1,  6,  0, ZSTD_fast    },  /* level  1 */
-    { 18, 14, 14,  1,  5,  1, ZSTD_dfast   },  /* level  2 */
-    { 18, 16, 16,  1,  4,  1, ZSTD_dfast   },  /* level  3 */
+    { 18, 14, 14,  1,  5,  0, ZSTD_dfast   },  /* level  2 */
+    { 18, 16, 16,  1,  4,  0, ZSTD_dfast   },  /* level  3 */
     { 18, 16, 17,  2,  5,  2, ZSTD_greedy  },  /* level  4.*/
     { 18, 18, 18,  3,  5,  2, ZSTD_greedy  },  /* level  5.*/
     { 18, 18, 19,  3,  5,  4, ZSTD_lazy    },  /* level  6.*/
@@ -4208,8 +4174,8 @@ static const ZSTD_compressionParameters ZSTD_defaultCParameters[4][ZSTD_MAX_CLEV
     { 17, 12, 12,  1,  5,  1, ZSTD_fast    },  /* base for negative levels */
     { 17, 12, 13,  1,  6,  0, ZSTD_fast    },  /* level  1 */
     { 17, 13, 15,  1,  5,  0, ZSTD_fast    },  /* level  2 */
-    { 17, 15, 16,  2,  5,  1, ZSTD_dfast   },  /* level  3 */
-    { 17, 17, 17,  2,  4,  1, ZSTD_dfast   },  /* level  4 */
+    { 17, 15, 16,  2,  5,  0, ZSTD_dfast   },  /* level  3 */
+    { 17, 17, 17,  2,  4,  0, ZSTD_dfast   },  /* level  4 */
     { 17, 16, 17,  3,  4,  2, ZSTD_greedy  },  /* level  5 */
     { 17, 17, 17,  3,  4,  4, ZSTD_lazy    },  /* level  6 */
     { 17, 17, 17,  3,  4,  8, ZSTD_lazy2   },  /* level  7 */
@@ -4234,7 +4200,7 @@ static const ZSTD_compressionParameters ZSTD_defaultCParameters[4][ZSTD_MAX_CLEV
     { 14, 12, 13,  1,  5,  1, ZSTD_fast    },  /* base for negative levels */
     { 14, 14, 15,  1,  5,  0, ZSTD_fast    },  /* level  1 */
     { 14, 14, 15,  1,  4,  0, ZSTD_fast    },  /* level  2 */
-    { 14, 14, 15,  2,  4,  1, ZSTD_dfast   },  /* level  3 */
+    { 14, 14, 15,  2,  4,  0, ZSTD_dfast   },  /* level  3 */
     { 14, 14, 14,  4,  4,  2, ZSTD_greedy  },  /* level  4 */
     { 14, 14, 14,  3,  4,  4, ZSTD_lazy    },  /* level  5.*/
     { 14, 14, 14,  4,  4,  8, ZSTD_lazy2   },  /* level  6 */
@@ -4257,34 +4223,56 @@ static const ZSTD_compressionParameters ZSTD_defaultCParameters[4][ZSTD_MAX_CLEV
 },
 };
 
-/*! ZSTD_getCParams() :
-*  @return ZSTD_compressionParameters structure for a selected compression level, srcSize and dictSize.
-*   Size values are optional, provide 0 if not known or unused */
-ZSTD_compressionParameters ZSTD_getCParams(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize)
+/*! ZSTD_getCParams_internal() :
+ * @return ZSTD_compressionParameters structure for a selected compression level, srcSize and dictSize.
+ *  Note: srcSizeHint 0 means 0, use ZSTD_CONTENTSIZE_UNKNOWN for unknown.
+ *        Use dictSize == 0 for unknown or unused. */
+static ZSTD_compressionParameters ZSTD_getCParams_internal(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize)
 {
-    size_t const addedSize = srcSizeHint ? 0 : 500;
-    U64 const rSize = srcSizeHint+dictSize ? srcSizeHint+dictSize+addedSize : (U64)-1;
-    U32 const tableID = (rSize <= 256 KB) + (rSize <= 128 KB) + (rSize <= 16 KB);   /* intentional underflow for srcSizeHint == 0 */
+    int const unknown = srcSizeHint == ZSTD_CONTENTSIZE_UNKNOWN;
+    size_t const addedSize = unknown && dictSize > 0 ? 500 : 0;
+    U64 const rSize = unknown && dictSize == 0 ? ZSTD_CONTENTSIZE_UNKNOWN : srcSizeHint+dictSize+addedSize;
+    U32 const tableID = (rSize <= 256 KB) + (rSize <= 128 KB) + (rSize <= 16 KB);
     int row = compressionLevel;
-    DEBUGLOG(5, "ZSTD_getCParams (cLevel=%i)", compressionLevel);
+    DEBUGLOG(5, "ZSTD_getCParams_internal (cLevel=%i)", compressionLevel);
     if (compressionLevel == 0) row = ZSTD_CLEVEL_DEFAULT;   /* 0 == default */
     if (compressionLevel < 0) row = 0;   /* entry 0 is baseline for fast mode */
     if (compressionLevel > ZSTD_MAX_CLEVEL) row = ZSTD_MAX_CLEVEL;
     {   ZSTD_compressionParameters cp = ZSTD_defaultCParameters[tableID][row];
         if (compressionLevel < 0) cp.targetLength = (unsigned)(-compressionLevel);   /* acceleration factor */
+        /* refine parameters based on srcSize & dictSize */
         return ZSTD_adjustCParams_internal(cp, srcSizeHint, dictSize);
     }
 }
 
+/*! ZSTD_getCParams() :
+ * @return ZSTD_compressionParameters structure for a selected compression level, srcSize and dictSize.
+ *  Size values are optional, provide 0 if not known or unused */
+ZSTD_compressionParameters ZSTD_getCParams(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize)
+{
+    if (srcSizeHint == 0) srcSizeHint = ZSTD_CONTENTSIZE_UNKNOWN;
+    return ZSTD_getCParams_internal(compressionLevel, srcSizeHint, dictSize);
+}
+
 /*! ZSTD_getParams() :
-*   same as ZSTD_getCParams(), but @return a `ZSTD_parameters` object (instead of `ZSTD_compressionParameters`).
-*   All fields of `ZSTD_frameParameters` are set to default (0) */
-ZSTD_parameters ZSTD_getParams(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize) {
+ *  same idea as ZSTD_getCParams()
+ * @return a `ZSTD_parameters` structure (instead of `ZSTD_compressionParameters`).
+ *  Fields of `ZSTD_frameParameters` are set to default values */
+static ZSTD_parameters ZSTD_getParams_internal(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize) {
     ZSTD_parameters params;
-    ZSTD_compressionParameters const cParams = ZSTD_getCParams(compressionLevel, srcSizeHint, dictSize);
+    ZSTD_compressionParameters const cParams = ZSTD_getCParams_internal(compressionLevel, srcSizeHint, dictSize);
     DEBUGLOG(5, "ZSTD_getParams (cLevel=%i)", compressionLevel);
     memset(&params, 0, sizeof(params));
     params.cParams = cParams;
     params.fParams.contentSizeFlag = 1;
     return params;
 }
+
+/*! ZSTD_getParams() :
+ *  same idea as ZSTD_getCParams()
+ * @return a `ZSTD_parameters` structure (instead of `ZSTD_compressionParameters`).
+ *  Fields of `ZSTD_frameParameters` are set to default values */
+ZSTD_parameters ZSTD_getParams(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize) {
+    if (srcSizeHint == 0) srcSizeHint = ZSTD_CONTENTSIZE_UNKNOWN;
+    return ZSTD_getParams_internal(compressionLevel, srcSizeHint, dictSize);
+}