mikronova/openzl-sys-rs
openzl/deps/zstd/lib/compress/zstd_compress.c
/*
* Copyright (c) Meta Platforms, Inc. and affiliates.
* All rights reserved.
*
* This source code is licensed under both the BSD-style license (found in the
* LICENSE file in the root directory of this source tree) and the GPLv2 (found
* in the COPYING file in the root directory of this source tree).
* You may select, at your option, one of the above-listed licenses.
*/
/*-*************************************
* Dependencies
***************************************/
#include "../common/allocations.h" /* ZSTD_customMalloc, ZSTD_customCalloc, ZSTD_customFree */
#include "../common/zstd_deps.h" /* INT_MAX, ZSTD_memset, ZSTD_memcpy */
#include "../common/mem.h"
#include "../common/error_private.h"
#include "hist.h" /* HIST_countFast_wksp */
#define FSE_STATIC_LINKING_ONLY /* FSE_encodeSymbol */
#include "../common/fse.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"
#include "../common/bits.h" /* ZSTD_highbit32, ZSTD_rotateRight_U64 */
/* ***************************************************************
* Tuning parameters
*****************************************************************/
/*!
* COMPRESS_HEAPMODE :
* Select how default decompression function ZSTD_compress() allocates its context,
* on stack (0, default), or into heap (1).
* Note that functions with explicit context such as ZSTD_compressCCtx() are unaffected.
*/
#ifndef ZSTD_COMPRESS_HEAPMODE
# define ZSTD_COMPRESS_HEAPMODE 0
#endif
/*!
* ZSTD_HASHLOG3_MAX :
* Maximum size of the hash table dedicated to find 3-bytes matches,
* in log format, aka 17 => 1 << 17 == 128Ki positions.
* This structure is only used in zstd_opt.
* Since allocation is centralized for all strategies, it has to be known here.
* The actual (selected) size of the hash table is then stored in ZSTD_MatchState_t.hashLog3,
* so that zstd_opt.c doesn't need to know about this constant.
*/
#ifndef ZSTD_HASHLOG3_MAX
# define ZSTD_HASHLOG3_MAX 17
#endif
/*-*************************************
* Helper functions
***************************************/
/* ZSTD_compressBound()
* Note that the result from this function is only valid for
* the one-pass compression functions.
* When employing the streaming mode,
* if flushes are frequently altering the size of blocks,
* the overhead from block headers can make the compressed data larger
* than the return value of ZSTD_compressBound().
*/
size_t ZSTD_compressBound(size_t srcSize) {
size_t const r = ZSTD_COMPRESSBOUND(srcSize);
if (r==0) return ERROR(srcSize_wrong);
return r;
}
/*-*************************************
* Context memory management
***************************************/
struct ZSTD_CDict_s {
const void* dictContent;
size_t dictContentSize;
ZSTD_dictContentType_e dictContentType; /* The dictContentType the CDict was created with */
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 */
ZSTD_ParamSwitch_e useRowMatchFinder; /* Indicates whether the CDict was created with params that would use
* row-based matchfinder. Unless the cdict is reloaded, we will use
* the same greedy/lazy matchfinder at compression time.
*/
}; /* typedef'd to ZSTD_CDict within "zstd.h" */
ZSTD_CCtx* ZSTD_createCCtx(void)
{
return ZSTD_createCCtx_advanced(ZSTD_defaultCMem);
}
static void ZSTD_initCCtx(ZSTD_CCtx* cctx, ZSTD_customMem memManager)
{
assert(cctx != NULL);
ZSTD_memset(cctx, 0, sizeof(*cctx));
cctx->customMem = memManager;
cctx->bmi2 = ZSTD_cpuSupportsBmi2();
{ size_t const err = ZSTD_CCtx_reset(cctx, ZSTD_reset_parameters);
assert(!ZSTD_isError(err));
(void)err;
}
}
ZSTD_CCtx* ZSTD_createCCtx_advanced(ZSTD_customMem customMem)
{
ZSTD_STATIC_ASSERT(zcss_init==0);
ZSTD_STATIC_ASSERT(ZSTD_CONTENTSIZE_UNKNOWN==(0ULL - 1));
if ((!customMem.customAlloc) ^ (!customMem.customFree)) return NULL;
{ ZSTD_CCtx* const cctx = (ZSTD_CCtx*)ZSTD_customMalloc(sizeof(ZSTD_CCtx), customMem);
if (!cctx) return NULL;
ZSTD_initCCtx(cctx, customMem);
return cctx;
}
}
ZSTD_CCtx* ZSTD_initStaticCCtx(void* workspace, size_t workspaceSize)
{
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 */
ZSTD_cwksp_init(&ws, workspace, workspaceSize, ZSTD_cwksp_static_alloc);
cctx = (ZSTD_CCtx*)ZSTD_cwksp_reserve_object(&ws, sizeof(ZSTD_CCtx));
if (cctx == NULL) return NULL;
ZSTD_memset(cctx, 0, sizeof(ZSTD_CCtx));
ZSTD_cwksp_move(&cctx->workspace, &ws);
cctx->staticSize = workspaceSize;
/* statically sized space. tmpWorkspace never moves (but prev/next block swap places) */
if (!ZSTD_cwksp_check_available(&cctx->workspace, TMP_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->tmpWorkspace = ZSTD_cwksp_reserve_object(&cctx->workspace, TMP_WORKSPACE_SIZE);
cctx->tmpWkspSize = TMP_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_customFree(cctx->localDict.dictBuffer, cctx->customMem);
ZSTD_freeCDict(cctx->localDict.cdict);
ZSTD_memset(&cctx->localDict, 0, sizeof(cctx->localDict));
ZSTD_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_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)
{
DEBUGLOG(3, "ZSTD_freeCCtx (address: %p)", (void*)cctx);
if (cctx==NULL) return 0; /* support free on NULL */
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_customFree(cctx, cctx->customMem);
}
return 0;
}
static size_t ZSTD_sizeof_mtctx(const ZSTD_CCtx* cctx)
{
#ifdef ZSTD_MULTITHREAD
return ZSTDMT_sizeof_CCtx(cctx->mtctx);
#else
(void)cctx;
return 0;
#endif
}
size_t ZSTD_sizeof_CCtx(const ZSTD_CCtx* cctx)
{
if (cctx==NULL) return 0; /* support sizeof on NULL */
/* 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);
}
size_t ZSTD_sizeof_CStream(const ZSTD_CStream* zcs)
{
return ZSTD_sizeof_CCtx(zcs); /* same object */
}
/* private API call, for dictBuilder only */
const SeqStore_t* ZSTD_getSeqStore(const ZSTD_CCtx* ctx) { return &(ctx->seqStore); }
/* Returns true if the strategy supports using a row based matchfinder */
static int ZSTD_rowMatchFinderSupported(const ZSTD_strategy strategy) {
return (strategy >= ZSTD_greedy && strategy <= ZSTD_lazy2);
}
/* Returns true if the strategy and useRowMatchFinder mode indicate that we will use the row based matchfinder
* for this compression.
*/
static int ZSTD_rowMatchFinderUsed(const ZSTD_strategy strategy, const ZSTD_ParamSwitch_e mode) {
assert(mode != ZSTD_ps_auto);
return ZSTD_rowMatchFinderSupported(strategy) && (mode == ZSTD_ps_enable);
}
/* Returns row matchfinder usage given an initial mode and cParams */
static ZSTD_ParamSwitch_e ZSTD_resolveRowMatchFinderMode(ZSTD_ParamSwitch_e mode,
const ZSTD_compressionParameters* const cParams) {
if (mode != ZSTD_ps_auto) return mode; /* if requested enabled, but no SIMD, we still will use row matchfinder */
mode = ZSTD_ps_disable;
if (!ZSTD_rowMatchFinderSupported(cParams->strategy)) return mode;
if (cParams->windowLog > 14) mode = ZSTD_ps_enable;
return mode;
}
/* Returns block splitter usage (generally speaking, when using slower/stronger compression modes) */
static ZSTD_ParamSwitch_e ZSTD_resolveBlockSplitterMode(ZSTD_ParamSwitch_e mode,
const ZSTD_compressionParameters* const cParams) {
if (mode != ZSTD_ps_auto) return mode;
return (cParams->strategy >= ZSTD_btopt && cParams->windowLog >= 17) ? ZSTD_ps_enable : ZSTD_ps_disable;
}
/* Returns 1 if the arguments indicate that we should allocate a chainTable, 0 otherwise */
static int ZSTD_allocateChainTable(const ZSTD_strategy strategy,
const ZSTD_ParamSwitch_e useRowMatchFinder,
const U32 forDDSDict) {
assert(useRowMatchFinder != ZSTD_ps_auto);
/* We always should allocate a chaintable if we are allocating a matchstate for a DDS dictionary matchstate.
* We do not allocate a chaintable if we are using ZSTD_fast, or are using the row-based matchfinder.
*/
return forDDSDict || ((strategy != ZSTD_fast) && !ZSTD_rowMatchFinderUsed(strategy, useRowMatchFinder));
}
/* Returns ZSTD_ps_enable if compression parameters are such that we should
* enable long distance matching (wlog >= 27, strategy >= btopt).
* Returns ZSTD_ps_disable otherwise.
*/
static ZSTD_ParamSwitch_e ZSTD_resolveEnableLdm(ZSTD_ParamSwitch_e mode,
const ZSTD_compressionParameters* const cParams) {
if (mode != ZSTD_ps_auto) return mode;
return (cParams->strategy >= ZSTD_btopt && cParams->windowLog >= 27) ? ZSTD_ps_enable : ZSTD_ps_disable;
}
static int ZSTD_resolveExternalSequenceValidation(int mode) {
return mode;
}
/* Resolves maxBlockSize to the default if no value is present. */
static size_t ZSTD_resolveMaxBlockSize(size_t maxBlockSize) {
if (maxBlockSize == 0) {
return ZSTD_BLOCKSIZE_MAX;
} else {
return maxBlockSize;
}
}
static ZSTD_ParamSwitch_e ZSTD_resolveExternalRepcodeSearch(ZSTD_ParamSwitch_e value, int cLevel) {
if (value != ZSTD_ps_auto) return value;
if (cLevel < 10) {
return ZSTD_ps_disable;
} else {
return ZSTD_ps_enable;
}
}
/* Returns 1 if compression parameters are such that CDict hashtable and chaintable indices are tagged.
* If so, the tags need to be removed in ZSTD_resetCCtx_byCopyingCDict. */
static int ZSTD_CDictIndicesAreTagged(const ZSTD_compressionParameters* const cParams) {
return cParams->strategy == ZSTD_fast || cParams->strategy == ZSTD_dfast;
}
static ZSTD_CCtx_params ZSTD_makeCCtxParamsFromCParams(
ZSTD_compressionParameters cParams)
{
ZSTD_CCtx_params cctxParams;
/* should not matter, as all cParams are presumed properly defined */
ZSTD_CCtxParams_init(&cctxParams, ZSTD_CLEVEL_DEFAULT);
cctxParams.cParams = cParams;
/* Adjust advanced params according to cParams */
cctxParams.ldmParams.enableLdm = ZSTD_resolveEnableLdm(cctxParams.ldmParams.enableLdm, &cParams);
if (cctxParams.ldmParams.enableLdm == ZSTD_ps_enable) {
ZSTD_ldm_adjustParameters(&cctxParams.ldmParams, &cParams);
assert(cctxParams.ldmParams.hashLog >= cctxParams.ldmParams.bucketSizeLog);
assert(cctxParams.ldmParams.hashRateLog < 32);
}
cctxParams.postBlockSplitter = ZSTD_resolveBlockSplitterMode(cctxParams.postBlockSplitter, &cParams);
cctxParams.useRowMatchFinder = ZSTD_resolveRowMatchFinderMode(cctxParams.useRowMatchFinder, &cParams);
cctxParams.validateSequences = ZSTD_resolveExternalSequenceValidation(cctxParams.validateSequences);
cctxParams.maxBlockSize = ZSTD_resolveMaxBlockSize(cctxParams.maxBlockSize);
cctxParams.searchForExternalRepcodes = ZSTD_resolveExternalRepcodeSearch(cctxParams.searchForExternalRepcodes,
cctxParams.compressionLevel);
assert(!ZSTD_checkCParams(cParams));
return cctxParams;
}
static ZSTD_CCtx_params* ZSTD_createCCtxParams_advanced(
ZSTD_customMem customMem)
{
ZSTD_CCtx_params* params;
if ((!customMem.customAlloc) ^ (!customMem.customFree)) return NULL;
params = (ZSTD_CCtx_params*)ZSTD_customCalloc(
sizeof(ZSTD_CCtx_params), customMem);
if (!params) { return NULL; }
ZSTD_CCtxParams_init(params, ZSTD_CLEVEL_DEFAULT);
params->customMem = customMem;
return params;
}
ZSTD_CCtx_params* ZSTD_createCCtxParams(void)
{
return ZSTD_createCCtxParams_advanced(ZSTD_defaultCMem);
}
size_t ZSTD_freeCCtxParams(ZSTD_CCtx_params* params)
{
if (params == NULL) { return 0; }
ZSTD_customFree(params, params->customMem);
return 0;
}
size_t ZSTD_CCtxParams_reset(ZSTD_CCtx_params* params)
{
return ZSTD_CCtxParams_init(params, ZSTD_CLEVEL_DEFAULT);
}
size_t ZSTD_CCtxParams_init(ZSTD_CCtx_params* cctxParams, int compressionLevel) {
RETURN_ERROR_IF(!cctxParams, GENERIC, "NULL pointer!");
ZSTD_memset(cctxParams, 0, sizeof(*cctxParams));
cctxParams->compressionLevel = compressionLevel;
cctxParams->fParams.contentSizeFlag = 1;
return 0;
}
#define ZSTD_NO_CLEVEL 0
/**
* Initializes `cctxParams` from `params` and `compressionLevel`.
* @param compressionLevel If params are derived from a compression level then that compression level, otherwise ZSTD_NO_CLEVEL.
*/
static void
ZSTD_CCtxParams_init_internal(ZSTD_CCtx_params* cctxParams,
const ZSTD_parameters* params,
int compressionLevel)
{
assert(!ZSTD_checkCParams(params->cParams));
ZSTD_memset(cctxParams, 0, sizeof(*cctxParams));
cctxParams->cParams = params->cParams;
cctxParams->fParams = params->fParams;
/* Should not matter, as all cParams are presumed properly defined.
* But, set it for tracing anyway.
*/
cctxParams->compressionLevel = compressionLevel;
cctxParams->useRowMatchFinder = ZSTD_resolveRowMatchFinderMode(cctxParams->useRowMatchFinder, ¶ms->cParams);
cctxParams->postBlockSplitter = ZSTD_resolveBlockSplitterMode(cctxParams->postBlockSplitter, ¶ms->cParams);
cctxParams->ldmParams.enableLdm = ZSTD_resolveEnableLdm(cctxParams->ldmParams.enableLdm, ¶ms->cParams);
cctxParams->validateSequences = ZSTD_resolveExternalSequenceValidation(cctxParams->validateSequences);
cctxParams->maxBlockSize = ZSTD_resolveMaxBlockSize(cctxParams->maxBlockSize);
cctxParams->searchForExternalRepcodes = ZSTD_resolveExternalRepcodeSearch(cctxParams->searchForExternalRepcodes, compressionLevel);
DEBUGLOG(4, "ZSTD_CCtxParams_init_internal: useRowMatchFinder=%d, useBlockSplitter=%d ldm=%d",
cctxParams->useRowMatchFinder, cctxParams->postBlockSplitter, cctxParams->ldmParams.enableLdm);
}
size_t ZSTD_CCtxParams_init_advanced(ZSTD_CCtx_params* cctxParams, ZSTD_parameters params)
{
RETURN_ERROR_IF(!cctxParams, GENERIC, "NULL pointer!");
FORWARD_IF_ERROR( ZSTD_checkCParams(params.cParams) , "");
ZSTD_CCtxParams_init_internal(cctxParams, ¶ms, ZSTD_NO_CLEVEL);
return 0;
}
/**
* Sets cctxParams' cParams and fParams from params, but otherwise leaves them alone.
* @param params Validated zstd parameters.
*/
static void ZSTD_CCtxParams_setZstdParams(
ZSTD_CCtx_params* cctxParams, const ZSTD_parameters* params)
{
assert(!ZSTD_checkCParams(params->cParams));
cctxParams->cParams = params->cParams;
cctxParams->fParams = params->fParams;
/* Should not matter, as all cParams are presumed properly defined.
* But, set it for tracing anyway.
*/
cctxParams->compressionLevel = ZSTD_NO_CLEVEL;
}
ZSTD_bounds ZSTD_cParam_getBounds(ZSTD_cParameter param)
{
ZSTD_bounds bounds = { 0, 0, 0 };
switch(param)
{
case ZSTD_c_compressionLevel:
bounds.lowerBound = ZSTD_minCLevel();
bounds.upperBound = ZSTD_maxCLevel();
return bounds;
case ZSTD_c_windowLog:
bounds.lowerBound = ZSTD_WINDOWLOG_MIN;
bounds.upperBound = ZSTD_WINDOWLOG_MAX;
return bounds;
case ZSTD_c_hashLog:
bounds.lowerBound = ZSTD_HASHLOG_MIN;
bounds.upperBound = ZSTD_HASHLOG_MAX;
return bounds;
case ZSTD_c_chainLog:
bounds.lowerBound = ZSTD_CHAINLOG_MIN;
bounds.upperBound = ZSTD_CHAINLOG_MAX;
return bounds;
case ZSTD_c_searchLog:
bounds.lowerBound = ZSTD_SEARCHLOG_MIN;
bounds.upperBound = ZSTD_SEARCHLOG_MAX;
return bounds;
case ZSTD_c_minMatch:
bounds.lowerBound = ZSTD_MINMATCH_MIN;
bounds.upperBound = ZSTD_MINMATCH_MAX;
return bounds;
case ZSTD_c_targetLength:
bounds.lowerBound = ZSTD_TARGETLENGTH_MIN;
bounds.upperBound = ZSTD_TARGETLENGTH_MAX;
return bounds;
case ZSTD_c_strategy:
bounds.lowerBound = ZSTD_STRATEGY_MIN;
bounds.upperBound = ZSTD_STRATEGY_MAX;
return bounds;
case ZSTD_c_contentSizeFlag:
bounds.lowerBound = 0;
bounds.upperBound = 1;
return bounds;
case ZSTD_c_checksumFlag:
bounds.lowerBound = 0;
bounds.upperBound = 1;
return bounds;
case ZSTD_c_dictIDFlag:
bounds.lowerBound = 0;
bounds.upperBound = 1;
return bounds;
case ZSTD_c_nbWorkers:
bounds.lowerBound = 0;
#ifdef ZSTD_MULTITHREAD
bounds.upperBound = ZSTDMT_NBWORKERS_MAX;
#else
bounds.upperBound = 0;
#endif
return bounds;
case ZSTD_c_jobSize:
bounds.lowerBound = 0;
#ifdef ZSTD_MULTITHREAD
bounds.upperBound = ZSTDMT_JOBSIZE_MAX;
#else
bounds.upperBound = 0;
#endif
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_enableDedicatedDictSearch:
bounds.lowerBound = 0;
bounds.upperBound = 1;
return bounds;
case ZSTD_c_enableLongDistanceMatching:
bounds.lowerBound = (int)ZSTD_ps_auto;
bounds.upperBound = (int)ZSTD_ps_disable;
return bounds;
case ZSTD_c_ldmHashLog:
bounds.lowerBound = ZSTD_LDM_HASHLOG_MIN;
bounds.upperBound = ZSTD_LDM_HASHLOG_MAX;
return bounds;
case ZSTD_c_ldmMinMatch:
bounds.lowerBound = ZSTD_LDM_MINMATCH_MIN;
bounds.upperBound = ZSTD_LDM_MINMATCH_MAX;
return bounds;
case ZSTD_c_ldmBucketSizeLog:
bounds.lowerBound = ZSTD_LDM_BUCKETSIZELOG_MIN;
bounds.upperBound = ZSTD_LDM_BUCKETSIZELOG_MAX;
return bounds;
case ZSTD_c_ldmHashRateLog:
bounds.lowerBound = ZSTD_LDM_HASHRATELOG_MIN;
bounds.upperBound = ZSTD_LDM_HASHRATELOG_MAX;
return bounds;
/* experimental parameters */
case ZSTD_c_rsyncable:
bounds.lowerBound = 0;
bounds.upperBound = 1;
return bounds;
case ZSTD_c_forceMaxWindow :
bounds.lowerBound = 0;
bounds.upperBound = 1;
return bounds;
case ZSTD_c_format:
ZSTD_STATIC_ASSERT(ZSTD_f_zstd1 < ZSTD_f_zstd1_magicless);
bounds.lowerBound = ZSTD_f_zstd1;
bounds.upperBound = ZSTD_f_zstd1_magicless; /* note : how to ensure at compile time that this is the highest value enum ? */
return bounds;
case ZSTD_c_forceAttachDict:
ZSTD_STATIC_ASSERT(ZSTD_dictDefaultAttach < ZSTD_dictForceLoad);
bounds.lowerBound = ZSTD_dictDefaultAttach;
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_ps_auto < ZSTD_ps_enable && ZSTD_ps_enable < ZSTD_ps_disable);
bounds.lowerBound = (int)ZSTD_ps_auto;
bounds.upperBound = (int)ZSTD_ps_disable;
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;
case ZSTD_c_stableInBuffer:
case ZSTD_c_stableOutBuffer:
bounds.lowerBound = (int)ZSTD_bm_buffered;
bounds.upperBound = (int)ZSTD_bm_stable;
return bounds;
case ZSTD_c_blockDelimiters:
bounds.lowerBound = (int)ZSTD_sf_noBlockDelimiters;
bounds.upperBound = (int)ZSTD_sf_explicitBlockDelimiters;
return bounds;
case ZSTD_c_validateSequences:
bounds.lowerBound = 0;
bounds.upperBound = 1;
return bounds;
case ZSTD_c_splitAfterSequences:
bounds.lowerBound = (int)ZSTD_ps_auto;
bounds.upperBound = (int)ZSTD_ps_disable;
return bounds;
case ZSTD_c_blockSplitterLevel:
bounds.lowerBound = 0;
bounds.upperBound = ZSTD_BLOCKSPLITTER_LEVEL_MAX;
return bounds;
case ZSTD_c_useRowMatchFinder:
bounds.lowerBound = (int)ZSTD_ps_auto;
bounds.upperBound = (int)ZSTD_ps_disable;
return bounds;
case ZSTD_c_deterministicRefPrefix:
bounds.lowerBound = 0;
bounds.upperBound = 1;
return bounds;
case ZSTD_c_prefetchCDictTables:
bounds.lowerBound = (int)ZSTD_ps_auto;
bounds.upperBound = (int)ZSTD_ps_disable;
return bounds;
case ZSTD_c_enableSeqProducerFallback:
bounds.lowerBound = 0;
bounds.upperBound = 1;
return bounds;
case ZSTD_c_maxBlockSize:
bounds.lowerBound = ZSTD_BLOCKSIZE_MAX_MIN;
bounds.upperBound = ZSTD_BLOCKSIZE_MAX;
return bounds;
case ZSTD_c_repcodeResolution:
bounds.lowerBound = (int)ZSTD_ps_auto;
bounds.upperBound = (int)ZSTD_ps_disable;
return bounds;
default:
bounds.error = ERROR(parameter_unsupported);
return bounds;
}
}
/* 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 bounds.error;
if (*value < bounds.lowerBound) *value = bounds.lowerBound;
if (*value > bounds.upperBound) *value = bounds.upperBound;
return 0;
}
#define BOUNDCHECK(cParam, val) \
do { \
RETURN_ERROR_IF(!ZSTD_cParam_withinBounds(cParam,val), \
parameter_outOfBound, "Param out of bounds"); \
} while (0)
static int ZSTD_isUpdateAuthorized(ZSTD_cParameter param)
{
switch(param)
{
case ZSTD_c_compressionLevel:
case ZSTD_c_hashLog:
case ZSTD_c_chainLog:
case ZSTD_c_searchLog:
case ZSTD_c_minMatch:
case ZSTD_c_targetLength:
case ZSTD_c_strategy:
case ZSTD_c_blockSplitterLevel:
return 1;
case ZSTD_c_format:
case ZSTD_c_windowLog:
case ZSTD_c_contentSizeFlag:
case ZSTD_c_checksumFlag:
case ZSTD_c_dictIDFlag:
case ZSTD_c_forceMaxWindow :
case ZSTD_c_nbWorkers:
case ZSTD_c_jobSize:
case ZSTD_c_overlapLog:
case ZSTD_c_rsyncable:
case ZSTD_c_enableDedicatedDictSearch:
case ZSTD_c_enableLongDistanceMatching:
case ZSTD_c_ldmHashLog:
case ZSTD_c_ldmMinMatch:
case ZSTD_c_ldmBucketSizeLog:
case ZSTD_c_ldmHashRateLog:
case ZSTD_c_forceAttachDict:
case ZSTD_c_literalCompressionMode:
case ZSTD_c_targetCBlockSize:
case ZSTD_c_srcSizeHint:
case ZSTD_c_stableInBuffer:
case ZSTD_c_stableOutBuffer:
case ZSTD_c_blockDelimiters:
case ZSTD_c_validateSequences:
case ZSTD_c_splitAfterSequences:
case ZSTD_c_useRowMatchFinder:
case ZSTD_c_deterministicRefPrefix:
case ZSTD_c_prefetchCDictTables:
case ZSTD_c_enableSeqProducerFallback:
case ZSTD_c_maxBlockSize:
case ZSTD_c_repcodeResolution:
default:
return 0;
}
}
size_t ZSTD_CCtx_setParameter(ZSTD_CCtx* cctx, ZSTD_cParameter param, int value)
{
DEBUGLOG(4, "ZSTD_CCtx_setParameter (%i, %i)", (int)param, value);
if (cctx->streamStage != zcss_init) {
if (ZSTD_isUpdateAuthorized(param)) {
cctx->cParamsChanged = 1;
} else {
RETURN_ERROR(stage_wrong, "can only set params in cctx init stage");
} }
switch(param)
{
case ZSTD_c_nbWorkers:
RETURN_ERROR_IF((value!=0) && cctx->staticSize, parameter_unsupported,
"MT not compatible with static alloc");
break;
case ZSTD_c_compressionLevel:
case ZSTD_c_windowLog:
case ZSTD_c_hashLog:
case ZSTD_c_chainLog:
case ZSTD_c_searchLog:
case ZSTD_c_minMatch:
case ZSTD_c_targetLength:
case ZSTD_c_strategy:
case ZSTD_c_ldmHashRateLog:
case ZSTD_c_format:
case ZSTD_c_contentSizeFlag:
case ZSTD_c_checksumFlag:
case ZSTD_c_dictIDFlag:
case ZSTD_c_forceMaxWindow:
case ZSTD_c_forceAttachDict:
case ZSTD_c_literalCompressionMode:
case ZSTD_c_jobSize:
case ZSTD_c_overlapLog:
case ZSTD_c_rsyncable:
case ZSTD_c_enableDedicatedDictSearch:
case ZSTD_c_enableLongDistanceMatching:
case ZSTD_c_ldmHashLog:
case ZSTD_c_ldmMinMatch:
case ZSTD_c_ldmBucketSizeLog:
case ZSTD_c_targetCBlockSize:
case ZSTD_c_srcSizeHint:
case ZSTD_c_stableInBuffer:
case ZSTD_c_stableOutBuffer:
case ZSTD_c_blockDelimiters:
case ZSTD_c_validateSequences:
case ZSTD_c_splitAfterSequences:
case ZSTD_c_blockSplitterLevel:
case ZSTD_c_useRowMatchFinder:
case ZSTD_c_deterministicRefPrefix:
case ZSTD_c_prefetchCDictTables:
case ZSTD_c_enableSeqProducerFallback:
case ZSTD_c_maxBlockSize:
case ZSTD_c_repcodeResolution:
break;
default: RETURN_ERROR(parameter_unsupported, "unknown parameter");
}
return ZSTD_CCtxParams_setParameter(&cctx->requestedParams, param, value);
}
size_t ZSTD_CCtxParams_setParameter(ZSTD_CCtx_params* CCtxParams,
ZSTD_cParameter param, int value)
{
DEBUGLOG(4, "ZSTD_CCtxParams_setParameter (%i, %i)", (int)param, value);
switch(param)
{
case ZSTD_c_format :
BOUNDCHECK(ZSTD_c_format, value);
CCtxParams->format = (ZSTD_format_e)value;
return (size_t)CCtxParams->format;
case ZSTD_c_compressionLevel : {
FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(param, &value), "");
if (value == 0)
CCtxParams->compressionLevel = ZSTD_CLEVEL_DEFAULT; /* 0 == default */
else
CCtxParams->compressionLevel = value;
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 = (U32)value;
return CCtxParams->cParams.windowLog;
case ZSTD_c_hashLog :
if (value!=0) /* 0 => use default */
BOUNDCHECK(ZSTD_c_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 = (U32)value;
return CCtxParams->cParams.chainLog;
case ZSTD_c_searchLog :
if (value!=0) /* 0 => use default */
BOUNDCHECK(ZSTD_c_searchLog, value);
CCtxParams->cParams.searchLog = (U32)value;
return (size_t)value;
case ZSTD_c_minMatch :
if (value!=0) /* 0 => use default */
BOUNDCHECK(ZSTD_c_minMatch, value);
CCtxParams->cParams.minMatch = (U32)value;
return CCtxParams->cParams.minMatch;
case ZSTD_c_targetLength :
BOUNDCHECK(ZSTD_c_targetLength, value);
CCtxParams->cParams.targetLength = (U32)value;
return CCtxParams->cParams.targetLength;
case ZSTD_c_strategy :
if (value!=0) /* 0 => use default */
BOUNDCHECK(ZSTD_c_strategy, value);
CCtxParams->cParams.strategy = (ZSTD_strategy)value;
return (size_t)CCtxParams->cParams.strategy;
case ZSTD_c_contentSizeFlag :
/* Content size written in frame header _when known_ (default:1) */
DEBUGLOG(4, "set content size flag = %u", (value!=0));
CCtxParams->fParams.contentSizeFlag = value != 0;
return (size_t)CCtxParams->fParams.contentSizeFlag;
case ZSTD_c_checksumFlag :
/* A 32-bits content checksum will be calculated and written at end of frame (default:0) */
CCtxParams->fParams.checksumFlag = value != 0;
return (size_t)CCtxParams->fParams.checksumFlag;
case ZSTD_c_dictIDFlag : /* When applicable, dictionary's dictID is provided in frame header (default:1) */
DEBUGLOG(4, "set dictIDFlag = %u", (value!=0));
CCtxParams->fParams.noDictIDFlag = !value;
return !CCtxParams->fParams.noDictIDFlag;
case ZSTD_c_forceMaxWindow :
CCtxParams->forceWindow = (value != 0);
return (size_t)CCtxParams->forceWindow;
case ZSTD_c_forceAttachDict : {
const ZSTD_dictAttachPref_e pref = (ZSTD_dictAttachPref_e)value;
BOUNDCHECK(ZSTD_c_forceAttachDict, (int)pref);
CCtxParams->attachDictPref = pref;
return CCtxParams->attachDictPref;
}
case ZSTD_c_literalCompressionMode : {
const ZSTD_ParamSwitch_e lcm = (ZSTD_ParamSwitch_e)value;
BOUNDCHECK(ZSTD_c_literalCompressionMode, (int)lcm);
CCtxParams->literalCompressionMode = lcm;
return CCtxParams->literalCompressionMode;
}
case ZSTD_c_nbWorkers :
#ifndef ZSTD_MULTITHREAD
RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
return 0;
#else
FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(param, &value), "");
CCtxParams->nbWorkers = value;
return (size_t)(CCtxParams->nbWorkers);
#endif
case ZSTD_c_jobSize :
#ifndef ZSTD_MULTITHREAD
RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
return 0;
#else
/* 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 = (size_t)value;
return CCtxParams->jobSize;
#endif
case ZSTD_c_overlapLog :
#ifndef ZSTD_MULTITHREAD
RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
return 0;
#else
FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(ZSTD_c_overlapLog, &value), "");
CCtxParams->overlapLog = value;
return (size_t)CCtxParams->overlapLog;
#endif
case ZSTD_c_rsyncable :
#ifndef ZSTD_MULTITHREAD
RETURN_ERROR_IF(value!=0, parameter_unsupported, "not compiled with multithreading");
return 0;
#else
FORWARD_IF_ERROR(ZSTD_cParam_clampBounds(ZSTD_c_overlapLog, &value), "");
CCtxParams->rsyncable = value;
return (size_t)CCtxParams->rsyncable;
#endif
case ZSTD_c_enableDedicatedDictSearch :
CCtxParams->enableDedicatedDictSearch = (value!=0);
return (size_t)CCtxParams->enableDedicatedDictSearch;
case ZSTD_c_enableLongDistanceMatching :
BOUNDCHECK(ZSTD_c_enableLongDistanceMatching, value);
CCtxParams->ldmParams.enableLdm = (ZSTD_ParamSwitch_e)value;
return CCtxParams->ldmParams.enableLdm;
case ZSTD_c_ldmHashLog :
if (value!=0) /* 0 ==> auto */
BOUNDCHECK(ZSTD_c_ldmHashLog, value);
CCtxParams->ldmParams.hashLog = (U32)value;
return CCtxParams->ldmParams.hashLog;
case ZSTD_c_ldmMinMatch :
if (value!=0) /* 0 ==> default */
BOUNDCHECK(ZSTD_c_ldmMinMatch, value);
CCtxParams->ldmParams.minMatchLength = (U32)value;
return CCtxParams->ldmParams.minMatchLength;
case ZSTD_c_ldmBucketSizeLog :
if (value!=0) /* 0 ==> default */
BOUNDCHECK(ZSTD_c_ldmBucketSizeLog, value);
CCtxParams->ldmParams.bucketSizeLog = (U32)value;
return CCtxParams->ldmParams.bucketSizeLog;
case ZSTD_c_ldmHashRateLog :
if (value!=0) /* 0 ==> default */
BOUNDCHECK(ZSTD_c_ldmHashRateLog, value);
CCtxParams->ldmParams.hashRateLog = (U32)value;
return CCtxParams->ldmParams.hashRateLog;
case ZSTD_c_targetCBlockSize :
if (value!=0) { /* 0 ==> default */
value = MAX(value, ZSTD_TARGETCBLOCKSIZE_MIN);
BOUNDCHECK(ZSTD_c_targetCBlockSize, value);
}
CCtxParams->targetCBlockSize = (U32)value;
return CCtxParams->targetCBlockSize;
case ZSTD_c_srcSizeHint :
if (value!=0) /* 0 ==> default */
BOUNDCHECK(ZSTD_c_srcSizeHint, value);
CCtxParams->srcSizeHint = value;
return (size_t)CCtxParams->srcSizeHint;
case ZSTD_c_stableInBuffer:
BOUNDCHECK(ZSTD_c_stableInBuffer, value);
CCtxParams->inBufferMode = (ZSTD_bufferMode_e)value;
return CCtxParams->inBufferMode;
case ZSTD_c_stableOutBuffer:
BOUNDCHECK(ZSTD_c_stableOutBuffer, value);
CCtxParams->outBufferMode = (ZSTD_bufferMode_e)value;
return CCtxParams->outBufferMode;
case ZSTD_c_blockDelimiters:
BOUNDCHECK(ZSTD_c_blockDelimiters, value);
CCtxParams->blockDelimiters = (ZSTD_SequenceFormat_e)value;
return CCtxParams->blockDelimiters;
case ZSTD_c_validateSequences:
BOUNDCHECK(ZSTD_c_validateSequences, value);
CCtxParams->validateSequences = value;
return (size_t)CCtxParams->validateSequences;
case ZSTD_c_splitAfterSequences:
BOUNDCHECK(ZSTD_c_splitAfterSequences, value);
CCtxParams->postBlockSplitter = (ZSTD_ParamSwitch_e)value;
return CCtxParams->postBlockSplitter;
case ZSTD_c_blockSplitterLevel:
BOUNDCHECK(ZSTD_c_blockSplitterLevel, value);
CCtxParams->preBlockSplitter_level = value;
return (size_t)CCtxParams->preBlockSplitter_level;
case ZSTD_c_useRowMatchFinder:
BOUNDCHECK(ZSTD_c_useRowMatchFinder, value);
CCtxParams->useRowMatchFinder = (ZSTD_ParamSwitch_e)value;
return CCtxParams->useRowMatchFinder;
case ZSTD_c_deterministicRefPrefix:
BOUNDCHECK(ZSTD_c_deterministicRefPrefix, value);
CCtxParams->deterministicRefPrefix = !!value;
return (size_t)CCtxParams->deterministicRefPrefix;
case ZSTD_c_prefetchCDictTables:
BOUNDCHECK(ZSTD_c_prefetchCDictTables, value);
CCtxParams->prefetchCDictTables = (ZSTD_ParamSwitch_e)value;
return CCtxParams->prefetchCDictTables;
case ZSTD_c_enableSeqProducerFallback:
BOUNDCHECK(ZSTD_c_enableSeqProducerFallback, value);
CCtxParams->enableMatchFinderFallback = value;
return (size_t)CCtxParams->enableMatchFinderFallback;
case ZSTD_c_maxBlockSize:
if (value!=0) /* 0 ==> default */
BOUNDCHECK(ZSTD_c_maxBlockSize, value);
assert(value>=0);
CCtxParams->maxBlockSize = (size_t)value;
return CCtxParams->maxBlockSize;
case ZSTD_c_repcodeResolution:
BOUNDCHECK(ZSTD_c_repcodeResolution, value);
CCtxParams->searchForExternalRepcodes = (ZSTD_ParamSwitch_e)value;
return CCtxParams->searchForExternalRepcodes;
default: RETURN_ERROR(parameter_unsupported, "unknown parameter");
}
}
size_t ZSTD_CCtx_getParameter(ZSTD_CCtx const* cctx, ZSTD_cParameter param, int* value)
{
return ZSTD_CCtxParams_getParameter(&cctx->requestedParams, param, value);
}
size_t ZSTD_CCtxParams_getParameter(
ZSTD_CCtx_params const* CCtxParams, ZSTD_cParameter param, int* value)
{
switch(param)
{
case ZSTD_c_format :
*value = (int)CCtxParams->format;
break;
case ZSTD_c_compressionLevel :
*value = CCtxParams->compressionLevel;
break;
case ZSTD_c_windowLog :
*value = (int)CCtxParams->cParams.windowLog;
break;
case ZSTD_c_hashLog :
*value = (int)CCtxParams->cParams.hashLog;
break;
case ZSTD_c_chainLog :
*value = (int)CCtxParams->cParams.chainLog;
break;
case ZSTD_c_searchLog :
*value = (int)CCtxParams->cParams.searchLog;
break;
case ZSTD_c_minMatch :
*value = (int)CCtxParams->cParams.minMatch;
break;
case ZSTD_c_targetLength :
*value = (int)CCtxParams->cParams.targetLength;
break;
case ZSTD_c_strategy :
*value = (int)CCtxParams->cParams.strategy;
break;
case ZSTD_c_contentSizeFlag :
*value = CCtxParams->fParams.contentSizeFlag;
break;
case ZSTD_c_checksumFlag :
*value = CCtxParams->fParams.checksumFlag;
break;
case ZSTD_c_dictIDFlag :
*value = !CCtxParams->fParams.noDictIDFlag;
break;
case ZSTD_c_forceMaxWindow :
*value = CCtxParams->forceWindow;
break;
case ZSTD_c_forceAttachDict :
*value = (int)CCtxParams->attachDictPref;
break;
case ZSTD_c_literalCompressionMode :
*value = (int)CCtxParams->literalCompressionMode;
break;
case ZSTD_c_nbWorkers :
#ifndef ZSTD_MULTITHREAD
assert(CCtxParams->nbWorkers == 0);
#endif
*value = CCtxParams->nbWorkers;
break;
case ZSTD_c_jobSize :
#ifndef ZSTD_MULTITHREAD
RETURN_ERROR(parameter_unsupported, "not compiled with multithreading");
#else
assert(CCtxParams->jobSize <= INT_MAX);
*value = (int)CCtxParams->jobSize;
break;
#endif
case ZSTD_c_overlapLog :
#ifndef ZSTD_MULTITHREAD
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, "not compiled with multithreading");
#else
*value = CCtxParams->rsyncable;
break;
#endif
case ZSTD_c_enableDedicatedDictSearch :
*value = CCtxParams->enableDedicatedDictSearch;
break;
case ZSTD_c_enableLongDistanceMatching :
*value = (int)CCtxParams->ldmParams.enableLdm;
break;
case ZSTD_c_ldmHashLog :
*value = (int)CCtxParams->ldmParams.hashLog;
break;
case ZSTD_c_ldmMinMatch :
*value = (int)CCtxParams->ldmParams.minMatchLength;
break;
case ZSTD_c_ldmBucketSizeLog :
*value = (int)CCtxParams->ldmParams.bucketSizeLog;
break;
case ZSTD_c_ldmHashRateLog :
*value = (int)CCtxParams->ldmParams.hashRateLog;
break;
case ZSTD_c_targetCBlockSize :
*value = (int)CCtxParams->targetCBlockSize;
break;
case ZSTD_c_srcSizeHint :
*value = (int)CCtxParams->srcSizeHint;
break;
case ZSTD_c_stableInBuffer :
*value = (int)CCtxParams->inBufferMode;
break;
case ZSTD_c_stableOutBuffer :
*value = (int)CCtxParams->outBufferMode;
break;
case ZSTD_c_blockDelimiters :
*value = (int)CCtxParams->blockDelimiters;
break;
case ZSTD_c_validateSequences :
*value = (int)CCtxParams->validateSequences;
break;
case ZSTD_c_splitAfterSequences :
*value = (int)CCtxParams->postBlockSplitter;
break;
case ZSTD_c_blockSplitterLevel :
*value = CCtxParams->preBlockSplitter_level;
break;
case ZSTD_c_useRowMatchFinder :
*value = (int)CCtxParams->useRowMatchFinder;
break;
case ZSTD_c_deterministicRefPrefix:
*value = (int)CCtxParams->deterministicRefPrefix;
break;
case ZSTD_c_prefetchCDictTables:
*value = (int)CCtxParams->prefetchCDictTables;
break;
case ZSTD_c_enableSeqProducerFallback:
*value = CCtxParams->enableMatchFinderFallback;
break;
case ZSTD_c_maxBlockSize:
*value = (int)CCtxParams->maxBlockSize;
break;
case ZSTD_c_repcodeResolution:
*value = (int)CCtxParams->searchForExternalRepcodes;
break;
default: RETURN_ERROR(parameter_unsupported, "unknown parameter");
}
return 0;
}
/** ZSTD_CCtx_setParametersUsingCCtxParams() :
* just applies `params` into `cctx`
* no action is performed, parameters are merely stored.
* If ZSTDMT is enabled, parameters are pushed to cctx->mtctx.
* This is possible even if a compression is ongoing.
* In which case, new parameters will be applied on the fly, starting with next compression job.
*/
size_t ZSTD_CCtx_setParametersUsingCCtxParams(
ZSTD_CCtx* cctx, const ZSTD_CCtx_params* params)
{
DEBUGLOG(4, "ZSTD_CCtx_setParametersUsingCCtxParams");
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;
}
size_t ZSTD_CCtx_setCParams(ZSTD_CCtx* cctx, ZSTD_compressionParameters cparams)
{
ZSTD_STATIC_ASSERT(sizeof(cparams) == 7 * 4 /* all params are listed below */);
DEBUGLOG(4, "ZSTD_CCtx_setCParams");
/* only update if all parameters are valid */
FORWARD_IF_ERROR(ZSTD_checkCParams(cparams), "");
FORWARD_IF_ERROR(ZSTD_CCtx_setParameter(cctx, ZSTD_c_windowLog, (int)cparams.windowLog), "");
FORWARD_IF_ERROR(ZSTD_CCtx_setParameter(cctx, ZSTD_c_chainLog, (int)cparams.chainLog), "");
FORWARD_IF_ERROR(ZSTD_CCtx_setParameter(cctx, ZSTD_c_hashLog, (int)cparams.hashLog), "");
FORWARD_IF_ERROR(ZSTD_CCtx_setParameter(cctx, ZSTD_c_searchLog, (int)cparams.searchLog), "");
FORWARD_IF_ERROR(ZSTD_CCtx_setParameter(cctx, ZSTD_c_minMatch, (int)cparams.minMatch), "");
FORWARD_IF_ERROR(ZSTD_CCtx_setParameter(cctx, ZSTD_c_targetLength, (int)cparams.targetLength), "");
FORWARD_IF_ERROR(ZSTD_CCtx_setParameter(cctx, ZSTD_c_strategy, (int)cparams.strategy), "");
return 0;
}
size_t ZSTD_CCtx_setFParams(ZSTD_CCtx* cctx, ZSTD_frameParameters fparams)
{
ZSTD_STATIC_ASSERT(sizeof(fparams) == 3 * 4 /* all params are listed below */);
DEBUGLOG(4, "ZSTD_CCtx_setFParams");
FORWARD_IF_ERROR(ZSTD_CCtx_setParameter(cctx, ZSTD_c_contentSizeFlag, fparams.contentSizeFlag != 0), "");
FORWARD_IF_ERROR(ZSTD_CCtx_setParameter(cctx, ZSTD_c_checksumFlag, fparams.checksumFlag != 0), "");
FORWARD_IF_ERROR(ZSTD_CCtx_setParameter(cctx, ZSTD_c_dictIDFlag, fparams.noDictIDFlag == 0), "");
return 0;
}
size_t ZSTD_CCtx_setParams(ZSTD_CCtx* cctx, ZSTD_parameters params)
{
DEBUGLOG(4, "ZSTD_CCtx_setParams");
/* First check cParams, because we want to update all or none. */
FORWARD_IF_ERROR(ZSTD_checkCParams(params.cParams), "");
/* Next set fParams, because this could fail if the cctx isn't in init stage. */
FORWARD_IF_ERROR(ZSTD_CCtx_setFParams(cctx, params.fParams), "");
/* Finally set cParams, which should succeed. */
FORWARD_IF_ERROR(ZSTD_CCtx_setCParams(cctx, params.cParams), "");
return 0;
}
size_t ZSTD_CCtx_setPledgedSrcSize(ZSTD_CCtx* cctx, unsigned long long pledgedSrcSize)
{
DEBUGLOG(4, "ZSTD_CCtx_setPledgedSrcSize to %llu bytes", pledgedSrcSize);
RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
"Can't set pledgedSrcSize when not in init stage.");
cctx->pledgedSrcSizePlusOne = pledgedSrcSize+1;
return 0;
}
static ZSTD_compressionParameters ZSTD_dedicatedDictSearch_getCParams(
int const compressionLevel,
size_t const dictSize);
static int ZSTD_dedicatedDictSearch_isSupported(
const ZSTD_compressionParameters* cParams);
static void ZSTD_dedicatedDictSearch_revertCParams(
ZSTD_compressionParameters* cParams);
/**
* Initializes the local dictionary using requested parameters.
* NOTE: Initialization does not employ the pledged src size,
* because the dictionary may be used for multiple compressions.
*/
static size_t ZSTD_initLocalDict(ZSTD_CCtx* cctx)
{
ZSTD_localDict* const dl = &cctx->localDict;
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) {
/* Local dictionary already initialized. */
assert(cctx->cdict == dl->cdict);
return 0;
}
assert(dl->dictSize > 0);
assert(cctx->cdict == NULL);
assert(cctx->prefixDict.dict == NULL);
dl->cdict = ZSTD_createCDict_advanced2(
dl->dict,
dl->dictSize,
ZSTD_dlm_byRef,
dl->dictContentType,
&cctx->requestedParams,
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)
{
DEBUGLOG(4, "ZSTD_CCtx_loadDictionary_advanced (size: %u)", (U32)dictSize);
RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
"Can't load a dictionary when cctx is not in init stage.");
ZSTD_clearAllDicts(cctx); /* erase any previously set dictionary */
if (dict == NULL || dictSize == 0) /* no dictionary */
return 0;
if (dictLoadMethod == ZSTD_dlm_byRef) {
cctx->localDict.dict = dict;
} else {
/* copy dictionary content inside CCtx to own its lifetime */
void* dictBuffer;
RETURN_ERROR_IF(cctx->staticSize, memory_allocation,
"static CCtx can't allocate for an internal copy of dictionary");
dictBuffer = ZSTD_customMalloc(dictSize, cctx->customMem);
RETURN_ERROR_IF(dictBuffer==NULL, memory_allocation,
"allocation failed for dictionary content");
ZSTD_memcpy(dictBuffer, dict, dictSize);
cctx->localDict.dictBuffer = dictBuffer; /* owned ptr to free */
cctx->localDict.dict = dictBuffer; /* read-only reference */
}
cctx->localDict.dictSize = dictSize;
cctx->localDict.dictContentType = dictContentType;
return 0;
}
size_t ZSTD_CCtx_loadDictionary_byReference(
ZSTD_CCtx* cctx, const void* dict, size_t dictSize)
{
return ZSTD_CCtx_loadDictionary_advanced(
cctx, dict, dictSize, ZSTD_dlm_byRef, ZSTD_dct_auto);
}
size_t ZSTD_CCtx_loadDictionary(ZSTD_CCtx* cctx, const void* dict, size_t dictSize)
{
return ZSTD_CCtx_loadDictionary_advanced(
cctx, dict, dictSize, ZSTD_dlm_byCopy, ZSTD_dct_auto);
}
size_t ZSTD_CCtx_refCDict(ZSTD_CCtx* cctx, const ZSTD_CDict* cdict)
{
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;
return 0;
}
size_t ZSTD_CCtx_refThreadPool(ZSTD_CCtx* cctx, ZSTD_threadPool* pool)
{
RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
"Can't ref a pool when ctx not in init stage.");
cctx->pool = pool;
return 0;
}
size_t ZSTD_CCtx_refPrefix(ZSTD_CCtx* cctx, const void* prefix, size_t prefixSize)
{
return ZSTD_CCtx_refPrefix_advanced(cctx, prefix, prefixSize, ZSTD_dct_rawContent);
}
size_t ZSTD_CCtx_refPrefix_advanced(
ZSTD_CCtx* cctx, const void* prefix, size_t prefixSize, ZSTD_dictContentType_e 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;
}
/*! ZSTD_CCtx_reset() :
* Also dumps dictionary */
size_t ZSTD_CCtx_reset(ZSTD_CCtx* cctx, ZSTD_ResetDirective reset)
{
if ( (reset == ZSTD_reset_session_only)
|| (reset == ZSTD_reset_session_and_parameters) ) {
cctx->streamStage = zcss_init;
cctx->pledgedSrcSizePlusOne = 0;
}
if ( (reset == ZSTD_reset_parameters)
|| (reset == ZSTD_reset_session_and_parameters) ) {
RETURN_ERROR_IF(cctx->streamStage != zcss_init, stage_wrong,
"Reset parameters is only possible during init stage.");
ZSTD_clearAllDicts(cctx);
return ZSTD_CCtxParams_reset(&cctx->requestedParams);
}
return 0;
}
/** ZSTD_checkCParams() :
control CParam values remain within authorized range.
@return : 0, or an error code if one value is beyond authorized range */
size_t ZSTD_checkCParams(ZSTD_compressionParameters cParams)
{
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, (int)cParams.strategy);
return 0;
}
/** ZSTD_clampCParams() :
* make CParam values within valid range.
* @return : valid CParams */
static ZSTD_compressionParameters
ZSTD_clampCParams(ZSTD_compressionParameters cParams)
{
# define CLAMP_TYPE(cParam, val, type) \
do { \
ZSTD_bounds const bounds = ZSTD_cParam_getBounds(cParam); \
if ((int)val<bounds.lowerBound) val=(type)bounds.lowerBound; \
else if ((int)val>bounds.upperBound) val=(type)bounds.upperBound; \
} while (0)
# 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);
CLAMP(ZSTD_c_searchLog, cParams.searchLog);
CLAMP(ZSTD_c_minMatch, cParams.minMatch);
CLAMP(ZSTD_c_targetLength,cParams.targetLength);
CLAMP_TYPE(ZSTD_c_strategy,cParams.strategy, ZSTD_strategy);
return cParams;
}
/** ZSTD_cycleLog() :
* condition for correct operation : hashLog > 1 */
U32 ZSTD_cycleLog(U32 hashLog, ZSTD_strategy strat)
{
U32 const btScale = ((U32)strat >= (U32)ZSTD_btlazy2);
return hashLog - btScale;
}
/** ZSTD_dictAndWindowLog() :
* Returns an adjusted window log that is large enough to fit the source and the dictionary.
* The zstd format says that the entire dictionary is valid if one byte of the dictionary
* is within the window. So the hashLog and chainLog should be large enough to reference both
* the dictionary and the window. So we must use this adjusted dictAndWindowLog when downsizing
* the hashLog and windowLog.
* NOTE: srcSize must not be ZSTD_CONTENTSIZE_UNKNOWN.
*/
static U32 ZSTD_dictAndWindowLog(U32 windowLog, U64 srcSize, U64 dictSize)
{
const U64 maxWindowSize = 1ULL << ZSTD_WINDOWLOG_MAX;
/* No dictionary ==> No change */
if (dictSize == 0) {
return windowLog;
}
assert(windowLog <= ZSTD_WINDOWLOG_MAX);
assert(srcSize != ZSTD_CONTENTSIZE_UNKNOWN); /* Handled in ZSTD_adjustCParams_internal() */
{
U64 const windowSize = 1ULL << windowLog;
U64 const dictAndWindowSize = dictSize + windowSize;
/* If the window size is already large enough to fit both the source and the dictionary
* then just use the window size. Otherwise adjust so that it fits the dictionary and
* the window.
*/
if (windowSize >= dictSize + srcSize) {
return windowLog; /* Window size large enough already */
} else if (dictAndWindowSize >= maxWindowSize) {
return ZSTD_WINDOWLOG_MAX; /* Larger than max window log */
} else {
return ZSTD_highbit32((U32)dictAndWindowSize - 1) + 1;
}
}
}
/** ZSTD_adjustCParams_internal() :
* 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.
* `mode` is the mode for parameter adjustment. See docs for `ZSTD_CParamMode_e`.
* 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,
size_t dictSize,
ZSTD_CParamMode_e mode,
ZSTD_ParamSwitch_e useRowMatchFinder)
{
const U64 minSrcSize = 513; /* (1<<9) + 1 */
const U64 maxWindowResize = 1ULL << (ZSTD_WINDOWLOG_MAX-1);
assert(ZSTD_checkCParams(cPar)==0);
/* Cascade the selected strategy down to the next-highest one built into
* this binary. */
#ifdef ZSTD_EXCLUDE_BTULTRA_BLOCK_COMPRESSOR
if (cPar.strategy == ZSTD_btultra2) {
cPar.strategy = ZSTD_btultra;
}
if (cPar.strategy == ZSTD_btultra) {
cPar.strategy = ZSTD_btopt;
}
#endif
#ifdef ZSTD_EXCLUDE_BTOPT_BLOCK_COMPRESSOR
if (cPar.strategy == ZSTD_btopt) {
cPar.strategy = ZSTD_btlazy2;
}
#endif
#ifdef ZSTD_EXCLUDE_BTLAZY2_BLOCK_COMPRESSOR
if (cPar.strategy == ZSTD_btlazy2) {
cPar.strategy = ZSTD_lazy2;
}
#endif
#ifdef ZSTD_EXCLUDE_LAZY2_BLOCK_COMPRESSOR
if (cPar.strategy == ZSTD_lazy2) {
cPar.strategy = ZSTD_lazy;
}
#endif
#ifdef ZSTD_EXCLUDE_LAZY_BLOCK_COMPRESSOR
if (cPar.strategy == ZSTD_lazy) {
cPar.strategy = ZSTD_greedy;
}
#endif
#ifdef ZSTD_EXCLUDE_GREEDY_BLOCK_COMPRESSOR
if (cPar.strategy == ZSTD_greedy) {
cPar.strategy = ZSTD_dfast;
}
#endif
#ifdef ZSTD_EXCLUDE_DFAST_BLOCK_COMPRESSOR
if (cPar.strategy == ZSTD_dfast) {
cPar.strategy = ZSTD_fast;
cPar.targetLength = 0;
}
#endif
switch (mode) {
case ZSTD_cpm_unknown:
case ZSTD_cpm_noAttachDict:
/* If we don't know the source size, don't make any
* assumptions about it. We will already have selected
* smaller parameters if a dictionary is in use.
*/
break;
case ZSTD_cpm_createCDict:
/* Assume a small source size when creating a dictionary
* with an unknown source size.
*/
if (dictSize && srcSize == ZSTD_CONTENTSIZE_UNKNOWN)
srcSize = minSrcSize;
break;
case ZSTD_cpm_attachDict:
/* Dictionary has its own dedicated parameters which have
* already been selected. We are selecting parameters
* for only the source.
*/
dictSize = 0;
break;
default:
assert(0);
break;
}
/* resize windowLog if input is small enough, to use less memory */
if ( (srcSize <= maxWindowResize)
&& (dictSize <= maxWindowResize) ) {
U32 const tSize = (U32)(srcSize + dictSize);
static U32 const hashSizeMin = 1 << ZSTD_HASHLOG_MIN;
U32 const srcLog = (tSize < hashSizeMin) ? ZSTD_HASHLOG_MIN :
ZSTD_highbit32(tSize-1) + 1;
if (cPar.windowLog > srcLog) cPar.windowLog = srcLog;
}
if (srcSize != ZSTD_CONTENTSIZE_UNKNOWN) {
U32 const dictAndWindowLog = ZSTD_dictAndWindowLog(cPar.windowLog, (U64)srcSize, (U64)dictSize);
U32 const cycleLog = ZSTD_cycleLog(cPar.chainLog, cPar.strategy);
if (cPar.hashLog > dictAndWindowLog+1) cPar.hashLog = dictAndWindowLog+1;
if (cycleLog > dictAndWindowLog)
cPar.chainLog -= (cycleLog - dictAndWindowLog);
}
if (cPar.windowLog < ZSTD_WINDOWLOG_ABSOLUTEMIN)
cPar.windowLog = ZSTD_WINDOWLOG_ABSOLUTEMIN; /* minimum wlog required for valid frame header */
/* We can't use more than 32 bits of hash in total, so that means that we require:
* (hashLog + 8) <= 32 && (chainLog + 8) <= 32
*/
if (mode == ZSTD_cpm_createCDict && ZSTD_CDictIndicesAreTagged(&cPar)) {
U32 const maxShortCacheHashLog = 32 - ZSTD_SHORT_CACHE_TAG_BITS;
if (cPar.hashLog > maxShortCacheHashLog) {
cPar.hashLog = maxShortCacheHashLog;
}
if (cPar.chainLog > maxShortCacheHashLog) {
cPar.chainLog = maxShortCacheHashLog;
}
}
/* At this point, we aren't 100% sure if we are using the row match finder.
* Unless it is explicitly disabled, conservatively assume that it is enabled.
* In this case it will only be disabled for small sources, so shrinking the
* hash log a little bit shouldn't result in any ratio loss.
*/
if (useRowMatchFinder == ZSTD_ps_auto)
useRowMatchFinder = ZSTD_ps_enable;
/* We can't hash more than 32-bits in total. So that means that we require:
* (hashLog - rowLog + 8) <= 32
*/
if (ZSTD_rowMatchFinderUsed(cPar.strategy, useRowMatchFinder)) {
/* Switch to 32-entry rows if searchLog is 5 (or more) */
U32 const rowLog = BOUNDED(4, cPar.searchLog, 6);
U32 const maxRowHashLog = 32 - ZSTD_ROW_HASH_TAG_BITS;
U32 const maxHashLog = maxRowHashLog + rowLog;
assert(cPar.hashLog >= rowLog);
if (cPar.hashLog > maxHashLog) {
cPar.hashLog = maxHashLog;
}
}
return cPar;
}
ZSTD_compressionParameters
ZSTD_adjustCParams(ZSTD_compressionParameters cPar,
unsigned long long srcSize,
size_t dictSize)
{
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, ZSTD_cpm_unknown, ZSTD_ps_auto);
}
static ZSTD_compressionParameters ZSTD_getCParams_internal(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize, ZSTD_CParamMode_e mode);
static ZSTD_parameters ZSTD_getParams_internal(int compressionLevel, unsigned long long srcSizeHint, size_t dictSize, ZSTD_CParamMode_e mode);
static void ZSTD_overrideCParams(
ZSTD_compressionParameters* cParams,
const ZSTD_compressionParameters* overrides)
{
if (overrides->windowLog) cParams->windowLog = overrides->windowLog;
if (overrides->hashLog) cParams->hashLog = overrides->hashLog;
if (overrides->chainLog) cParams->chainLog = overrides->chainLog;
if (overrides->searchLog) cParams->searchLog = overrides->searchLog;
if (overrides->minMatch) cParams->minMatch = overrides->minMatch;
if (overrides->targetLength) cParams->targetLength = overrides->targetLength;
if (overrides->strategy) cParams->strategy = overrides->strategy;
}
ZSTD_compressionParameters ZSTD_getCParamsFromCCtxParams(
const ZSTD_CCtx_params* CCtxParams, U64 srcSizeHint, size_t dictSize, ZSTD_CParamMode_e mode)
{
ZSTD_compressionParameters cParams;
if (srcSizeHint == ZSTD_CONTENTSIZE_UNKNOWN && CCtxParams->srcSizeHint > 0) {
assert(CCtxParams->srcSizeHint>=0);
srcSizeHint = (U64)CCtxParams->srcSizeHint;
}
cParams = ZSTD_getCParams_internal(CCtxParams->compressionLevel, srcSizeHint, dictSize, mode);
if (CCtxParams->ldmParams.enableLdm == ZSTD_ps_enable) cParams.windowLog = ZSTD_LDM_DEFAULT_WINDOW_LOG;
ZSTD_overrideCParams(&cParams, &CCtxParams->cParams);
assert(!ZSTD_checkCParams(cParams));
/* srcSizeHint == 0 means 0 */
return ZSTD_adjustCParams_internal(cParams, srcSizeHint, dictSize, mode, CCtxParams->useRowMatchFinder);
}
static size_t
ZSTD_sizeof_matchState(const ZSTD_compressionParameters* const cParams,
const ZSTD_ParamSwitch_e useRowMatchFinder,
const int enableDedicatedDictSearch,
const U32 forCCtx)
{
/* chain table size should be 0 for fast or row-hash strategies */
size_t const chainSize = ZSTD_allocateChainTable(cParams->strategy, useRowMatchFinder, enableDedicatedDictSearch && !forCCtx)
? ((size_t)1 << cParams->chainLog)
: 0;
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 = 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_aligned64_alloc_size((MaxML+1) * sizeof(U32))
+ ZSTD_cwksp_aligned64_alloc_size((MaxLL+1) * sizeof(U32))
+ ZSTD_cwksp_aligned64_alloc_size((MaxOff+1) * sizeof(U32))
+ ZSTD_cwksp_aligned64_alloc_size((1<<Litbits) * sizeof(U32))
+ ZSTD_cwksp_aligned64_alloc_size(ZSTD_OPT_SIZE * sizeof(ZSTD_match_t))
+ ZSTD_cwksp_aligned64_alloc_size(ZSTD_OPT_SIZE * sizeof(ZSTD_optimal_t));
size_t const lazyAdditionalSpace = ZSTD_rowMatchFinderUsed(cParams->strategy, useRowMatchFinder)
? ZSTD_cwksp_aligned64_alloc_size(hSize)
: 0;
size_t const optSpace = (forCCtx && (cParams->strategy >= ZSTD_btopt))
? optPotentialSpace
: 0;
size_t const slackSpace = ZSTD_cwksp_slack_space_required();
/* tables are guaranteed to be sized in multiples of 64 bytes (or 16 uint32_t) */
ZSTD_STATIC_ASSERT(ZSTD_HASHLOG_MIN >= 4 && ZSTD_WINDOWLOG_MIN >= 4 && ZSTD_CHAINLOG_MIN >= 4);
assert(useRowMatchFinder != ZSTD_ps_auto);
DEBUGLOG(4, "chainSize: %u - hSize: %u - h3Size: %u",
(U32)chainSize, (U32)hSize, (U32)h3Size);
return tableSpace + optSpace + slackSpace + lazyAdditionalSpace;
}
/* Helper function for calculating memory requirements.
* Gives a tighter bound than ZSTD_sequenceBound() by taking minMatch into account. */
static size_t ZSTD_maxNbSeq(size_t blockSize, unsigned minMatch, int useSequenceProducer) {
U32 const divider = (minMatch==3 || useSequenceProducer) ? 3 : 4;
return blockSize / divider;
}
static size_t ZSTD_estimateCCtxSize_usingCCtxParams_internal(
const ZSTD_compressionParameters* cParams,
const ldmParams_t* ldmParams,
const int isStatic,
const ZSTD_ParamSwitch_e useRowMatchFinder,
const size_t buffInSize,
const size_t buffOutSize,
const U64 pledgedSrcSize,
int useSequenceProducer,
size_t maxBlockSize)
{
size_t const windowSize = (size_t) BOUNDED(1ULL, 1ULL << cParams->windowLog, pledgedSrcSize);
size_t const blockSize = MIN(ZSTD_resolveMaxBlockSize(maxBlockSize), windowSize);
size_t const maxNbSeq = ZSTD_maxNbSeq(blockSize, cParams->minMatch, useSequenceProducer);
size_t const tokenSpace = ZSTD_cwksp_alloc_size(WILDCOPY_OVERLENGTH + blockSize)
+ ZSTD_cwksp_aligned64_alloc_size(maxNbSeq * sizeof(SeqDef))
+ 3 * ZSTD_cwksp_alloc_size(maxNbSeq * sizeof(BYTE));
size_t const tmpWorkSpace = ZSTD_cwksp_alloc_size(TMP_WORKSPACE_SIZE);
size_t const blockStateSpace = 2 * ZSTD_cwksp_alloc_size(sizeof(ZSTD_compressedBlockState_t));
size_t const matchStateSize = ZSTD_sizeof_matchState(cParams, useRowMatchFinder, /* enableDedicatedDictSearch */ 0, /* forCCtx */ 1);
size_t const ldmSpace = ZSTD_ldm_getTableSize(*ldmParams);
size_t const maxNbLdmSeq = ZSTD_ldm_getMaxNbSeq(*ldmParams, blockSize);
size_t const ldmSeqSpace = ldmParams->enableLdm == ZSTD_ps_enable ?
ZSTD_cwksp_aligned64_alloc_size(maxNbLdmSeq * sizeof(rawSeq)) : 0;
size_t const bufferSpace = ZSTD_cwksp_alloc_size(buffInSize)
+ ZSTD_cwksp_alloc_size(buffOutSize);
size_t const cctxSpace = isStatic ? ZSTD_cwksp_alloc_size(sizeof(ZSTD_CCtx)) : 0;
size_t const maxNbExternalSeq = ZSTD_sequenceBound(blockSize);
size_t const externalSeqSpace = useSequenceProducer
? ZSTD_cwksp_aligned64_alloc_size(maxNbExternalSeq * sizeof(ZSTD_Sequence))
: 0;
size_t const neededSpace =
cctxSpace +
tmpWorkSpace +
blockStateSpace +
ldmSpace +
ldmSeqSpace +
matchStateSize +
tokenSpace +
bufferSpace +
externalSeqSpace;
DEBUGLOG(5, "estimate workspace : %u", (U32)neededSpace);
return neededSpace;
}
size_t ZSTD_estimateCCtxSize_usingCCtxParams(const ZSTD_CCtx_params* params)
{
ZSTD_compressionParameters const cParams =
ZSTD_getCParamsFromCCtxParams(params, ZSTD_CONTENTSIZE_UNKNOWN, 0, ZSTD_cpm_noAttachDict);
ZSTD_ParamSwitch_e const useRowMatchFinder = ZSTD_resolveRowMatchFinderMode(params->useRowMatchFinder,
&cParams);
RETURN_ERROR_IF(params->nbWorkers > 0, GENERIC, "Estimate CCtx size is supported for single-threaded compression only.");
/* 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. */
return ZSTD_estimateCCtxSize_usingCCtxParams_internal(
&cParams, ¶ms->ldmParams, 1, useRowMatchFinder, 0, 0, ZSTD_CONTENTSIZE_UNKNOWN, ZSTD_hasExtSeqProd(params), params->maxBlockSize);
}
size_t ZSTD_estimateCCtxSize_usingCParams(ZSTD_compressionParameters cParams)
{
ZSTD_CCtx_params initialParams = ZSTD_makeCCtxParamsFromCParams(cParams);
if (ZSTD_rowMatchFinderSupported(cParams.strategy)) {
/* Pick bigger of not using and using row-based matchfinder for greedy and lazy strategies */
size_t noRowCCtxSize;
size_t rowCCtxSize;
initialParams.useRowMatchFinder = ZSTD_ps_disable;
noRowCCtxSize = ZSTD_estimateCCtxSize_usingCCtxParams(&initialParams);
initialParams.useRowMatchFinder = ZSTD_ps_enable;
rowCCtxSize = ZSTD_estimateCCtxSize_usingCCtxParams(&initialParams);
return MAX(noRowCCtxSize, rowCCtxSize);
} else {
return ZSTD_estimateCCtxSize_usingCCtxParams(&initialParams);
}
}
static size_t ZSTD_estimateCCtxSize_internal(int compressionLevel)
{
int tier = 0;
size_t largestSize = 0;
static const unsigned long long srcSizeTiers[4] = {16 KB, 128 KB, 256 KB, ZSTD_CONTENTSIZE_UNKNOWN};
for (; tier < 4; ++tier) {
/* Choose the set of cParams for a given level across all srcSizes that give the largest cctxSize */
ZSTD_compressionParameters const cParams = ZSTD_getCParams_internal(compressionLevel, srcSizeTiers[tier], 0, ZSTD_cpm_noAttachDict);
largestSize = MAX(ZSTD_estimateCCtxSize_usingCParams(cParams), largestSize);
}
return largestSize;
}
size_t ZSTD_estimateCCtxSize(int compressionLevel)
{
int level;
size_t memBudget = 0;
for (level=MIN(compressionLevel, 1); level<=compressionLevel; level++) {
/* Ensure monotonically increasing memory usage as compression level increases */
size_t const newMB = ZSTD_estimateCCtxSize_internal(level);
if (newMB > memBudget) memBudget = newMB;
}
return memBudget;
}
size_t ZSTD_estimateCStreamSize_usingCCtxParams(const ZSTD_CCtx_params* params)
{
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, ZSTD_cpm_noAttachDict);
size_t const blockSize = MIN(ZSTD_resolveMaxBlockSize(params->maxBlockSize), (size_t)1 << cParams.windowLog);
size_t const inBuffSize = (params->inBufferMode == ZSTD_bm_buffered)
? ((size_t)1 << cParams.windowLog) + blockSize
: 0;
size_t const outBuffSize = (params->outBufferMode == ZSTD_bm_buffered)
? ZSTD_compressBound(blockSize) + 1
: 0;
ZSTD_ParamSwitch_e const useRowMatchFinder = ZSTD_resolveRowMatchFinderMode(params->useRowMatchFinder, ¶ms->cParams);
return ZSTD_estimateCCtxSize_usingCCtxParams_internal(
&cParams, ¶ms->ldmParams, 1, useRowMatchFinder, inBuffSize, outBuffSize,
ZSTD_CONTENTSIZE_UNKNOWN, ZSTD_hasExtSeqProd(params), params->maxBlockSize);
}
}
size_t ZSTD_estimateCStreamSize_usingCParams(ZSTD_compressionParameters cParams)
{
ZSTD_CCtx_params initialParams = ZSTD_makeCCtxParamsFromCParams(cParams);
if (ZSTD_rowMatchFinderSupported(cParams.strategy)) {
/* Pick bigger of not using and using row-based matchfinder for greedy and lazy strategies */
size_t noRowCCtxSize;
size_t rowCCtxSize;
initialParams.useRowMatchFinder = ZSTD_ps_disable;
noRowCCtxSize = ZSTD_estimateCStreamSize_usingCCtxParams(&initialParams);
initialParams.useRowMatchFinder = ZSTD_ps_enable;
rowCCtxSize = ZSTD_estimateCStreamSize_usingCCtxParams(&initialParams);
return MAX(noRowCCtxSize, rowCCtxSize);
} else {
return ZSTD_estimateCStreamSize_usingCCtxParams(&initialParams);
}
}
static size_t ZSTD_estimateCStreamSize_internal(int compressionLevel)
{
ZSTD_compressionParameters const cParams = ZSTD_getCParams_internal(compressionLevel, ZSTD_CONTENTSIZE_UNKNOWN, 0, ZSTD_cpm_noAttachDict);
return ZSTD_estimateCStreamSize_usingCParams(cParams);
}
size_t ZSTD_estimateCStreamSize(int compressionLevel)
{
int level;
size_t memBudget = 0;
for (level=MIN(compressionLevel, 1); level<=compressionLevel; level++) {
size_t const newMB = ZSTD_estimateCStreamSize_internal(level);
if (newMB > memBudget) memBudget = newMB;
}
return memBudget;
}
/* ZSTD_getFrameProgression():
* tells how much data has been consumed (input) and produced (output) for current frame.
* able to count progression inside worker threads (non-blocking mode).
*/
ZSTD_frameProgression ZSTD_getFrameProgression(const ZSTD_CCtx* cctx)
{
#ifdef ZSTD_MULTITHREAD
if (cctx->appliedParams.nbWorkers > 0) {
return ZSTDMT_getFrameProgression(cctx->mtctx);
}
#endif
{ ZSTD_frameProgression fp;
size_t const buffered = (cctx->inBuff == NULL) ? 0 :
cctx->inBuffPos - cctx->inToCompress;
if (buffered) assert(cctx->inBuffPos >= cctx->inToCompress);
assert(buffered <= ZSTD_BLOCKSIZE_MAX);
fp.ingested = cctx->consumedSrcSize + buffered;
fp.consumed = cctx->consumedSrcSize;
fp.produced = cctx->producedCSize;
fp.flushed = cctx->producedCSize; /* simplified; some data might still be left within streaming output buffer */
fp.currentJobID = 0;
fp.nbActiveWorkers = 0;
return fp;
} }
/*! ZSTD_toFlushNow()
* Only useful for multithreading scenarios currently (nbWorkers >= 1).
*/
size_t ZSTD_toFlushNow(ZSTD_CCtx* cctx)
{
#ifdef ZSTD_MULTITHREAD
if (cctx->appliedParams.nbWorkers > 0) {
return ZSTDMT_toFlushNow(cctx->mtctx);
}
#endif
(void)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 void ZSTD_assertEqualCParams(ZSTD_compressionParameters cParams1,
ZSTD_compressionParameters cParams2)
{
(void)cParams1;
(void)cParams2;
assert(cParams1.windowLog == cParams2.windowLog);
assert(cParams1.chainLog == cParams2.chainLog);
assert(cParams1.hashLog == cParams2.hashLog);
assert(cParams1.searchLog == cParams2.searchLog);
assert(cParams1.minMatch == cParams2.minMatch);
assert(cParams1.targetLength == cParams2.targetLength);
assert(cParams1.strategy == cParams2.strategy);
}
void ZSTD_reset_compressedBlockState(ZSTD_compressedBlockState_t* bs)
{
int i;
for (i = 0; i < ZSTD_REP_NUM; ++i)
bs->rep[i] = repStartValue[i];
bs->entropy.huf.repeatMode = HUF_repeat_none;
bs->entropy.fse.offcode_repeatMode = FSE_repeat_none;
bs->entropy.fse.matchlength_repeatMode = FSE_repeat_none;
bs->entropy.fse.litlength_repeatMode = FSE_repeat_none;
}
/*! ZSTD_invalidateMatchState()
* 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->loadedDictEnd = 0;
ms->opt.litLengthSum = 0; /* force reset of btopt stats */
ms->dictMatchState = NULL;
}
/**
* 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 {
ZSTD_resetTarget_CDict,
ZSTD_resetTarget_CCtx
} ZSTD_resetTarget_e;
/* Mixes bits in a 64 bits in a value, based on XXH3_rrmxmx */
static U64 ZSTD_bitmix(U64 val, U64 len) {
val ^= ZSTD_rotateRight_U64(val, 49) ^ ZSTD_rotateRight_U64(val, 24);
val *= 0x9FB21C651E98DF25ULL;
val ^= (val >> 35) + len ;
val *= 0x9FB21C651E98DF25ULL;
return val ^ (val >> 28);
}
/* Mixes in the hashSalt and hashSaltEntropy to create a new hashSalt */
static void ZSTD_advanceHashSalt(ZSTD_MatchState_t* ms) {
ms->hashSalt = ZSTD_bitmix(ms->hashSalt, 8) ^ ZSTD_bitmix((U64) ms->hashSaltEntropy, 4);
}
static size_t
ZSTD_reset_matchState(ZSTD_MatchState_t* ms,
ZSTD_cwksp* ws,
const ZSTD_compressionParameters* cParams,
const ZSTD_ParamSwitch_e useRowMatchFinder,
const ZSTD_compResetPolicy_e crp,
const ZSTD_indexResetPolicy_e forceResetIndex,
const ZSTD_resetTarget_e forWho)
{
/* disable chain table allocation for fast or row-based strategies */
size_t const chainSize = ZSTD_allocateChainTable(cParams->strategy, useRowMatchFinder,
ms->dedicatedDictSearch && (forWho == ZSTD_resetTarget_CDict))
? ((size_t)1 << cParams->chainLog)
: 0;
size_t const hSize = ((size_t)1) << cParams->hashLog;
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;
DEBUGLOG(4, "reset indices : %u", forceResetIndex == ZSTDirp_reset);
assert(useRowMatchFinder != ZSTD_ps_auto);
if (forceResetIndex == ZSTDirp_reset) {
ZSTD_window_init(&ms->window);
ZSTD_cwksp_mark_tables_dirty(ws);
}
ms->hashLog3 = hashLog3;
ms->lazySkipping = 0;
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);
}
if (ZSTD_rowMatchFinderUsed(cParams->strategy, useRowMatchFinder)) {
/* Row match finder needs an additional table of hashes ("tags") */
size_t const tagTableSize = hSize;
/* We want to generate a new salt in case we reset a Cctx, but we always want to use
* 0 when we reset a Cdict */
if(forWho == ZSTD_resetTarget_CCtx) {
ms->tagTable = (BYTE*) ZSTD_cwksp_reserve_aligned_init_once(ws, tagTableSize);
ZSTD_advanceHashSalt(ms);
} else {
/* When we are not salting we want to always memset the memory */
ms->tagTable = (BYTE*) ZSTD_cwksp_reserve_aligned64(ws, tagTableSize);
ZSTD_memset(ms->tagTable, 0, tagTableSize);
ms->hashSalt = 0;
}
{ /* Switch to 32-entry rows if searchLog is 5 (or more) */
U32 const rowLog = BOUNDED(4, cParams->searchLog, 6);
assert(cParams->hashLog >= rowLog);
ms->rowHashLog = cParams->hashLog - rowLog;
}
}
/* opt parser space */
if ((forWho == ZSTD_resetTarget_CCtx) && (cParams->strategy >= ZSTD_btopt)) {
DEBUGLOG(4, "reserving optimal parser space");
ms->opt.litFreq = (unsigned*)ZSTD_cwksp_reserve_aligned64(ws, (1<<Litbits) * sizeof(unsigned));
ms->opt.litLengthFreq = (unsigned*)ZSTD_cwksp_reserve_aligned64(ws, (MaxLL+1) * sizeof(unsigned));
ms->opt.matchLengthFreq = (unsigned*)ZSTD_cwksp_reserve_aligned64(ws, (MaxML+1) * sizeof(unsigned));
ms->opt.offCodeFreq = (unsigned*)ZSTD_cwksp_reserve_aligned64(ws, (MaxOff+1) * sizeof(unsigned));
ms->opt.matchTable = (ZSTD_match_t*)ZSTD_cwksp_reserve_aligned64(ws, ZSTD_OPT_SIZE * sizeof(ZSTD_match_t));
ms->opt.priceTable = (ZSTD_optimal_t*)ZSTD_cwksp_reserve_aligned64(ws, ZSTD_OPT_SIZE * sizeof(ZSTD_optimal_t));
}
ms->cParams = *cParams;
RETURN_ERROR_IF(ZSTD_cwksp_reserve_failed(ws), memory_allocation,
"failed a workspace allocation in ZSTD_reset_matchState");
return 0;
}
/* 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_dictTooBig():
* When dictionaries are larger than ZSTD_CHUNKSIZE_MAX they can't be loaded in
* one go generically. So we ensure that in that case we reset the tables to zero,
* so that we can load as much of the dictionary as possible.
*/
static int ZSTD_dictTooBig(size_t const loadedDictSize)
{
return loadedDictSize > ZSTD_CHUNKSIZE_MAX;
}
/*! ZSTD_resetCCtx_internal() :
* @param loadedDictSize The size of the dictionary to be loaded
* into the context, if any. If no dictionary is used, or the
* dictionary is being attached / copied, then pass 0.
* note : `params` are assumed fully validated at this stage.
*/
static size_t ZSTD_resetCCtx_internal(ZSTD_CCtx* zc,
ZSTD_CCtx_params const* params,
U64 const pledgedSrcSize,
size_t const loadedDictSize,
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, useRowMatchFinder=%d useBlockSplitter=%d",
(U32)pledgedSrcSize, params->cParams.windowLog, (int)params->useRowMatchFinder, (int)params->postBlockSplitter);
assert(!ZSTD_isError(ZSTD_checkCParams(params->cParams)));
zc->isFirstBlock = 1;
/* Set applied params early so we can modify them for LDM,
* and point params at the applied params.
*/
zc->appliedParams = *params;
params = &zc->appliedParams;
assert(params->useRowMatchFinder != ZSTD_ps_auto);
assert(params->postBlockSplitter != ZSTD_ps_auto);
assert(params->ldmParams.enableLdm != ZSTD_ps_auto);
assert(params->maxBlockSize != 0);
if (params->ldmParams.enableLdm == ZSTD_ps_enable) {
/* Adjust long distance matching parameters */
ZSTD_ldm_adjustParameters(&zc->appliedParams.ldmParams, ¶ms->cParams);
assert(params->ldmParams.hashLog >= params->ldmParams.bucketSizeLog);
assert(params->ldmParams.hashRateLog < 32);
}
{ size_t const windowSize = MAX(1, (size_t)MIN(((U64)1 << params->cParams.windowLog), pledgedSrcSize));
size_t const blockSize = MIN(params->maxBlockSize, windowSize);
size_t const maxNbSeq = ZSTD_maxNbSeq(blockSize, params->cParams.minMatch, ZSTD_hasExtSeqProd(params));
size_t const buffOutSize = (zbuff == ZSTDb_buffered && params->outBufferMode == ZSTD_bm_buffered)
? ZSTD_compressBound(blockSize) + 1
: 0;
size_t const buffInSize = (zbuff == ZSTDb_buffered && params->inBufferMode == ZSTD_bm_buffered)
? windowSize + blockSize
: 0;
size_t const maxNbLdmSeq = ZSTD_ldm_getMaxNbSeq(params->ldmParams, blockSize);
int const indexTooClose = ZSTD_indexTooCloseToMax(zc->blockState.matchState.window);
int const dictTooBig = ZSTD_dictTooBig(loadedDictSize);
ZSTD_indexResetPolicy_e needsIndexReset =
(indexTooClose || dictTooBig || !zc->initialized) ? ZSTDirp_reset : ZSTDirp_continue;
size_t const neededSpace =
ZSTD_estimateCCtxSize_usingCCtxParams_internal(
¶ms->cParams, ¶ms->ldmParams, zc->staticSize != 0, params->useRowMatchFinder,
buffInSize, buffOutSize, pledgedSrcSize, ZSTD_hasExtSeqProd(params), params->maxBlockSize);
FORWARD_IF_ERROR(neededSpace, "cctx size estimate failed!");
if (!zc->staticSize) ZSTD_cwksp_bump_oversized_duration(ws, 0);
{ /* Check if workspace is large enough, alloc a new one if needed */
int const workspaceTooSmall = ZSTD_cwksp_sizeof(ws) < neededSpace;
int const workspaceWasteful = ZSTD_cwksp_check_wasteful(ws, neededSpace);
int resizeWorkspace = workspaceTooSmall || workspaceWasteful;
DEBUGLOG(4, "Need %zu B workspace", neededSpace);
DEBUGLOG(4, "windowSize: %zu - blockSize: %zu", windowSize, blockSize);
if (resizeWorkspace) {
DEBUGLOG(4, "Resize workspaceSize from %zuKB to %zuKB",
ZSTD_cwksp_sizeof(ws) >> 10,
neededSpace >> 10);
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.
* tmpWorkspace never moves,
* though prev/next block swap places */
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->tmpWorkspace = ZSTD_cwksp_reserve_object(ws, TMP_WORKSPACE_SIZE);
RETURN_ERROR_IF(zc->tmpWorkspace == NULL, memory_allocation, "couldn't allocate tmpWorkspace");
zc->tmpWkspSize = TMP_WORKSPACE_SIZE;
} }
ZSTD_cwksp_clear(ws);
/* init params */
zc->blockState.matchState.cParams = params->cParams;
zc->blockState.matchState.prefetchCDictTables = params->prefetchCDictTables == ZSTD_ps_enable;
zc->pledgedSrcSizePlusOne = pledgedSrcSize+1;
zc->consumedSrcSize = 0;
zc->producedCSize = 0;
if (pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN)
zc->appliedParams.fParams.contentSizeFlag = 0;
DEBUGLOG(4, "pledged content size : %u ; flag : %u",
(unsigned)pledgedSrcSize, zc->appliedParams.fParams.contentSizeFlag);
zc->blockSizeMax = blockSize;
XXH64_reset(&zc->xxhState, 0);
zc->stage = ZSTDcs_init;
zc->dictID = 0;
zc->dictContentSize = 0;
ZSTD_reset_compressedBlockState(zc->blockState.prevCBlock);
FORWARD_IF_ERROR(ZSTD_reset_matchState(
&zc->blockState.matchState,
ws,
¶ms->cParams,
params->useRowMatchFinder,
crp,
needsIndexReset,
ZSTD_resetTarget_CCtx), "");
zc->seqStore.sequencesStart = (SeqDef*)ZSTD_cwksp_reserve_aligned64(ws, maxNbSeq * sizeof(SeqDef));
/* ldm hash table */
if (params->ldmParams.enableLdm == ZSTD_ps_enable) {
/* TODO: avoid memset? */
size_t const ldmHSize = ((size_t)1) << params->ldmParams.hashLog;
zc->ldmState.hashTable = (ldmEntry_t*)ZSTD_cwksp_reserve_aligned64(ws, ldmHSize * sizeof(ldmEntry_t));
ZSTD_memset(zc->ldmState.hashTable, 0, ldmHSize * sizeof(ldmEntry_t));
zc->ldmSequences = (rawSeq*)ZSTD_cwksp_reserve_aligned64(ws, maxNbLdmSeq * sizeof(rawSeq));
zc->maxNbLdmSequences = maxNbLdmSeq;
ZSTD_window_init(&zc->ldmState.window);
zc->ldmState.loadedDictEnd = 0;
}
/* reserve space for block-level external sequences */
if (ZSTD_hasExtSeqProd(params)) {
size_t const maxNbExternalSeq = ZSTD_sequenceBound(blockSize);
zc->extSeqBufCapacity = maxNbExternalSeq;
zc->extSeqBuf =
(ZSTD_Sequence*)ZSTD_cwksp_reserve_aligned64(ws, maxNbExternalSeq * sizeof(ZSTD_Sequence));
}
/* buffers */
/* 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;
zc->bufferedPolicy = zbuff;
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 == ZSTD_ps_enable) {
/* TODO: avoid memset? */
size_t const numBuckets =
((size_t)1) << (params->ldmParams.hashLog -
params->ldmParams.bucketSizeLog);
zc->ldmState.bucketOffsets = ZSTD_cwksp_reserve_buffer(ws, numBuckets);
ZSTD_memset(zc->ldmState.bucketOffsets, 0, numBuckets);
}
/* 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));
DEBUGLOG(3, "wksp: finished allocating, %zd bytes remain available", ZSTD_cwksp_available_space(ws));
assert(ZSTD_cwksp_estimated_space_within_bounds(ws, neededSpace));
zc->initialized = 1;
return 0;
}
}
/* ZSTD_invalidateRepCodes() :
* ensures next compression will not use repcodes from previous block.
* Note : only works with regular variant;
* do not use with extDict variant ! */
void ZSTD_invalidateRepCodes(ZSTD_CCtx* cctx) {
int i;
for (i=0; i<ZSTD_REP_NUM; i++) cctx->blockState.prevCBlock->rep[i] = 0;
assert(!ZSTD_window_hasExtDict(cctx->blockState.matchState.window));
}
/* These are the approximate sizes for each strategy past which copying the
* dictionary tables into the working context is faster than using them
* in-place.
*/
static const size_t attachDictSizeCutoffs[ZSTD_STRATEGY_MAX+1] = {
8 KB, /* unused */
8 KB, /* ZSTD_fast */
16 KB, /* ZSTD_dfast */
32 KB, /* ZSTD_greedy */
32 KB, /* ZSTD_lazy */
32 KB, /* ZSTD_lazy2 */
32 KB, /* ZSTD_btlazy2 */
32 KB, /* ZSTD_btopt */
8 KB, /* ZSTD_btultra */
8 KB /* ZSTD_btultra2 */
};
static int ZSTD_shouldAttachDict(const ZSTD_CDict* cdict,
const ZSTD_CCtx_params* params,
U64 pledgedSrcSize)
{
size_t cutoff = attachDictSizeCutoffs[cdict->matchState.cParams.strategy];
int const dedicatedDictSearch = cdict->matchState.dedicatedDictSearch;
return dedicatedDictSearch
|| ( ( pledgedSrcSize <= cutoff
|| pledgedSrcSize == ZSTD_CONTENTSIZE_UNKNOWN
|| 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)
{
DEBUGLOG(4, "ZSTD_resetCCtx_byAttachingCDict() pledgedSrcSize=%llu",
(unsigned long long)pledgedSrcSize);
{
ZSTD_compressionParameters adjusted_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. */
/* pledgedSrcSize == 0 means 0! */
if (cdict->matchState.dedicatedDictSearch) {
ZSTD_dedicatedDictSearch_revertCParams(&adjusted_cdict_cParams);
}
params.cParams = ZSTD_adjustCParams_internal(adjusted_cdict_cParams, pledgedSrcSize,
cdict->dictContentSize, ZSTD_cpm_attachDict,
params.useRowMatchFinder);
params.cParams.windowLog = windowLog;
params.useRowMatchFinder = cdict->useRowMatchFinder; /* cdict overrides */
FORWARD_IF_ERROR(ZSTD_resetCCtx_internal(cctx, ¶ms, pledgedSrcSize,
/* loadedDictSize */ 0,
ZSTDcrp_makeClean, zbuff), "");
assert(cctx->appliedParams.cParams.strategy == adjusted_cdict_cParams.strategy);
}
{ const U32 cdictEnd = (U32)( cdict->matchState.window.nextSrc
- cdict->matchState.window.base);
const U32 cdictLen = cdictEnd - cdict->matchState.window.dictLimit;
if (cdictLen == 0) {
/* don't even attach dictionaries with no contents */
DEBUGLOG(4, "skipping attaching empty dictionary");
} else {
DEBUGLOG(4, "attaching dictionary into context");
cctx->blockState.matchState.dictMatchState = &cdict->matchState;
/* prep working match state so dict matches never have negative indices
* when they are translated to the working context's index space. */
if (cctx->blockState.matchState.window.dictLimit < cdictEnd) {
cctx->blockState.matchState.window.nextSrc =
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;
cctx->dictContentSize = cdict->dictContentSize;
/* copy block state */
ZSTD_memcpy(cctx->blockState.prevCBlock, &cdict->cBlockState, sizeof(cdict->cBlockState));
return 0;
}
static void ZSTD_copyCDictTableIntoCCtx(U32* dst, U32 const* src, size_t tableSize,
ZSTD_compressionParameters const* cParams) {
if (ZSTD_CDictIndicesAreTagged(cParams)){
/* Remove tags from the CDict table if they are present.
* See docs on "short cache" in zstd_compress_internal.h for context. */
size_t i;
for (i = 0; i < tableSize; i++) {
U32 const taggedIndex = src[i];
U32 const index = taggedIndex >> ZSTD_SHORT_CACHE_TAG_BITS;
dst[i] = index;
}
} else {
ZSTD_memcpy(dst, src, tableSize * sizeof(U32));
}
}
static size_t ZSTD_resetCCtx_byCopyingCDict(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;
assert(!cdict->matchState.dedicatedDictSearch);
DEBUGLOG(4, "ZSTD_resetCCtx_byCopyingCDict() pledgedSrcSize=%llu",
(unsigned long long)pledgedSrcSize);
{ unsigned const windowLog = params.cParams.windowLog;
assert(windowLog != 0);
/* Copy only compression parameters related to tables. */
params.cParams = *cdict_cParams;
params.cParams.windowLog = windowLog;
params.useRowMatchFinder = cdict->useRowMatchFinder;
FORWARD_IF_ERROR(ZSTD_resetCCtx_internal(cctx, ¶ms, pledgedSrcSize,
/* loadedDictSize */ 0,
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);
assert(params.useRowMatchFinder != ZSTD_ps_auto);
/* copy tables */
{ size_t const chainSize = ZSTD_allocateChainTable(cdict_cParams->strategy, cdict->useRowMatchFinder, 0 /* DDS guaranteed disabled */)
? ((size_t)1 << cdict_cParams->chainLog)
: 0;
size_t const hSize = (size_t)1 << cdict_cParams->hashLog;
ZSTD_copyCDictTableIntoCCtx(cctx->blockState.matchState.hashTable,
cdict->matchState.hashTable,
hSize, cdict_cParams);
/* Do not copy cdict's chainTable if cctx has parameters such that it would not use chainTable */
if (ZSTD_allocateChainTable(cctx->appliedParams.cParams.strategy, cctx->appliedParams.useRowMatchFinder, 0 /* forDDSDict */)) {
ZSTD_copyCDictTableIntoCCtx(cctx->blockState.matchState.chainTable,
cdict->matchState.chainTable,
chainSize, cdict_cParams);
}
/* copy tag table */
if (ZSTD_rowMatchFinderUsed(cdict_cParams->strategy, cdict->useRowMatchFinder)) {
size_t const tagTableSize = hSize;
ZSTD_memcpy(cctx->blockState.matchState.tagTable,
cdict->matchState.tagTable,
tagTableSize);
cctx->blockState.matchState.hashSalt = cdict->matchState.hashSalt;
}
}
/* Zero the hashTable3, since the cdict never fills it */
assert(cctx->blockState.matchState.hashLog3 <= 31);
{ U32 const h3log = cctx->blockState.matchState.hashLog3;
size_t const h3Size = h3log ? ((size_t)1 << h3log) : 0;
assert(cdict->matchState.hashLog3 == 0);
ZSTD_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->loadedDictEnd= srcMatchState->loadedDictEnd;
}
cctx->dictID = cdict->dictID;
cctx->dictContentSize = cdict->dictContentSize;
/* copy block state */
ZSTD_memcpy(cctx->blockState.prevCBlock, &cdict->cBlockState, sizeof(cdict->cBlockState));
return 0;
}
/* We have a choice between copying the dictionary context into the working
* context, or referencing the dictionary context from the working context
* in-place. We decide here which strategy to use. */
static size_t ZSTD_resetCCtx_usingCDict(ZSTD_CCtx* cctx,
const ZSTD_CDict* cdict,
const ZSTD_CCtx_params* params,
U64 pledgedSrcSize,
ZSTD_buffered_policy_e zbuff)
{
DEBUGLOG(4, "ZSTD_resetCCtx_usingCDict (pledgedSrcSize=%u)",
(unsigned)pledgedSrcSize);
if (ZSTD_shouldAttachDict(cdict, params, pledgedSrcSize)) {
return ZSTD_resetCCtx_byAttachingCDict(
cctx, cdict, *params, pledgedSrcSize, zbuff);
} else {
return ZSTD_resetCCtx_byCopyingCDict(
cctx, cdict, *params, pledgedSrcSize, zbuff);
}
}
/*! ZSTD_copyCCtx_internal() :
* Duplicate an existing context `srcCCtx` into another one `dstCCtx`.
* Only works during stage ZSTDcs_init (i.e. after creation, but before first call to ZSTD_compressContinue()).
* The "context", in this case, refers to the hash and chain tables,
* entropy tables, and dictionary references.
* `windowLog` value is enforced if != 0, otherwise value is copied from srcCCtx.
* @return : 0, or an error code */
static size_t ZSTD_copyCCtx_internal(ZSTD_CCtx* dstCCtx,
const ZSTD_CCtx* srcCCtx,
ZSTD_frameParameters fParams,
U64 pledgedSrcSize,
ZSTD_buffered_policy_e zbuff)
{
RETURN_ERROR_IF(srcCCtx->stage!=ZSTDcs_init, stage_wrong,
"Can't copy a ctx that's not in init stage.");
DEBUGLOG(5, "ZSTD_copyCCtx_internal");
ZSTD_memcpy(&dstCCtx->customMem, &srcCCtx->customMem, sizeof(ZSTD_customMem));
{ ZSTD_CCtx_params params = dstCCtx->requestedParams;
/* Copy only compression parameters related to tables. */
params.cParams = srcCCtx->appliedParams.cParams;
assert(srcCCtx->appliedParams.useRowMatchFinder != ZSTD_ps_auto);
assert(srcCCtx->appliedParams.postBlockSplitter != ZSTD_ps_auto);
assert(srcCCtx->appliedParams.ldmParams.enableLdm != ZSTD_ps_auto);
params.useRowMatchFinder = srcCCtx->appliedParams.useRowMatchFinder;
params.postBlockSplitter = srcCCtx->appliedParams.postBlockSplitter;
params.ldmParams = srcCCtx->appliedParams.ldmParams;
params.fParams = fParams;
params.maxBlockSize = srcCCtx->appliedParams.maxBlockSize;
ZSTD_resetCCtx_internal(dstCCtx, ¶ms, pledgedSrcSize,
/* loadedDictSize */ 0,
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);
assert(dstCCtx->appliedParams.cParams.chainLog == srcCCtx->appliedParams.cParams.chainLog);
assert(dstCCtx->blockState.matchState.hashLog3 == srcCCtx->blockState.matchState.hashLog3);
}
ZSTD_cwksp_mark_tables_dirty(&dstCCtx->workspace);
/* copy tables */
{ size_t const chainSize = ZSTD_allocateChainTable(srcCCtx->appliedParams.cParams.strategy,
srcCCtx->appliedParams.useRowMatchFinder,
0 /* forDDSDict */)
? ((size_t)1 << srcCCtx->appliedParams.cParams.chainLog)
: 0;
size_t const hSize = (size_t)1 << srcCCtx->appliedParams.cParams.hashLog;
U32 const h3log = srcCCtx->blockState.matchState.hashLog3;
size_t const h3Size = h3log ? ((size_t)1 << h3log) : 0;
ZSTD_memcpy(dstCCtx->blockState.matchState.hashTable,
srcCCtx->blockState.matchState.hashTable,
hSize * sizeof(U32));
ZSTD_memcpy(dstCCtx->blockState.matchState.chainTable,
srcCCtx->blockState.matchState.chainTable,
chainSize * sizeof(U32));
ZSTD_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->loadedDictEnd= srcMatchState->loadedDictEnd;
}
dstCCtx->dictID = srcCCtx->dictID;
dstCCtx->dictContentSize = srcCCtx->dictContentSize;
/* copy block state */
ZSTD_memcpy(dstCCtx->blockState.prevCBlock, srcCCtx->blockState.prevCBlock, sizeof(*srcCCtx->blockState.prevCBlock));
return 0;
}
/*! ZSTD_copyCCtx() :
* Duplicate an existing context `srcCCtx` into another one `dstCCtx`.
* Only works during stage ZSTDcs_init (i.e. after creation, but before first call to ZSTD_compressContinue()).
* pledgedSrcSize==0 means "unknown".
* @return : 0, or an error code */
size_t ZSTD_copyCCtx(ZSTD_CCtx* dstCCtx, const ZSTD_CCtx* srcCCtx, unsigned long long pledgedSrcSize)
{
ZSTD_frameParameters fParams = { 1 /*content*/, 0 /*checksum*/, 0 /*noDictID*/ };
ZSTD_buffered_policy_e const zbuff = srcCCtx->bufferedPolicy;
ZSTD_STATIC_ASSERT((U32)ZSTDb_buffered==1);
if (pledgedSrcSize==0) pledgedSrcSize = ZSTD_CONTENTSIZE_UNKNOWN;
fParams.contentSizeFlag = (pledgedSrcSize != ZSTD_CONTENTSIZE_UNKNOWN);
return ZSTD_copyCCtx_internal(dstCCtx, srcCCtx,
fParams, pledgedSrcSize,
zbuff);
}
#define ZSTD_ROWSIZE 16
/*! ZSTD_reduceTable() :
* reduce table indexes by `reducerValue`, or squash to zero.
* PreserveMark preserves "unsorted mark" for btlazy2 strategy.
* It must be set to a clear 0/1 value, to remove branch during inlining.
* Presume table size is a multiple of ZSTD_ROWSIZE
* to help auto-vectorization */
FORCE_INLINE_TEMPLATE void
ZSTD_reduceTable_internal (U32* const table, U32 const size, U32 const reducerValue, int const preserveMark)
{
int const nbRows = (int)size / ZSTD_ROWSIZE;
int cellNb = 0;
int rowNb;
/* Protect special index values < ZSTD_WINDOW_START_INDEX. */
U32 const reducerThreshold = reducerValue + ZSTD_WINDOW_START_INDEX;
assert((size & (ZSTD_ROWSIZE-1)) == 0); /* multiple of ZSTD_ROWSIZE */
assert(size < (1U<<31)); /* can be cast to int */
#if ZSTD_MEMORY_SANITIZER && !defined (ZSTD_MSAN_DONT_POISON_WORKSPACE)
/* To validate that the table reuse 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++) {
U32 newVal;
if (preserveMark && table[cellNb] == ZSTD_DUBT_UNSORTED_MARK) {
/* This write is pointless, but is required(?) for the compiler
* to auto-vectorize the loop. */
newVal = ZSTD_DUBT_UNSORTED_MARK;
} else if (table[cellNb] < reducerThreshold) {
newVal = 0;
} else {
newVal = table[cellNb] - reducerValue;
}
table[cellNb] = newVal;
cellNb++;
} }
}
static void ZSTD_reduceTable(U32* const table, U32 const size, U32 const reducerValue)
{
ZSTD_reduceTable_internal(table, size, reducerValue, 0);
}
static void ZSTD_reduceTable_btlazy2(U32* const table, U32 const size, U32 const reducerValue)
{
ZSTD_reduceTable_internal(table, size, reducerValue, 1);
}
/*! ZSTD_reduceIndex() :
* rescale all indexes to avoid future overflow (indexes are U32) */
static void ZSTD_reduceIndex (ZSTD_MatchState_t* ms, ZSTD_CCtx_params const* params, const U32 reducerValue)
{
{ U32 const hSize = (U32)1 << params->cParams.hashLog;
ZSTD_reduceTable(ms->hashTable, hSize, reducerValue);
}
if (ZSTD_allocateChainTable(params->cParams.strategy, params->useRowMatchFinder, (U32)ms->dedicatedDictSearch)) {
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);
}
if (ms->hashLog3) {
U32 const h3Size = (U32)1 << ms->hashLog3;
ZSTD_reduceTable(ms->hashTable3, h3Size, reducerValue);
}
}
/*-*******************************************************
* Block entropic compression
*********************************************************/
/* See doc/zstd_compression_format.md for detailed format description */
int ZSTD_seqToCodes(const SeqStore_t* seqStorePtr)
{
const SeqDef* const sequences = seqStorePtr->sequencesStart;
BYTE* const llCodeTable = seqStorePtr->llCode;
BYTE* const ofCodeTable = seqStorePtr->ofCode;
BYTE* const mlCodeTable = seqStorePtr->mlCode;
U32 const nbSeq = (U32)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
U32 u;
int longOffsets = 0;
assert(nbSeq <= seqStorePtr->maxNbSeq);
for (u=0; u<nbSeq; u++) {
U32 const llv = sequences[u].litLength;
U32 const ofCode = ZSTD_highbit32(sequences[u].offBase);
U32 const mlv = sequences[u].mlBase;
llCodeTable[u] = (BYTE)ZSTD_LLcode(llv);
ofCodeTable[u] = (BYTE)ofCode;
mlCodeTable[u] = (BYTE)ZSTD_MLcode(mlv);
assert(!(MEM_64bits() && ofCode >= STREAM_ACCUMULATOR_MIN));
if (MEM_32bits() && ofCode >= STREAM_ACCUMULATOR_MIN)
longOffsets = 1;
}
if (seqStorePtr->longLengthType==ZSTD_llt_literalLength)
llCodeTable[seqStorePtr->longLengthPos] = MaxLL;
if (seqStorePtr->longLengthType==ZSTD_llt_matchLength)
mlCodeTable[seqStorePtr->longLengthPos] = MaxML;
return longOffsets;
}
/* 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)
{
DEBUGLOG(5, "ZSTD_useTargetCBlockSize (targetCBlockSize=%zu)", cctxParams->targetCBlockSize);
return (cctxParams->targetCBlockSize != 0);
}
/* ZSTD_blockSplitterEnabled():
* Returns if block splitting param is being used
* If used, compression will do best effort to split a block in order to improve compression ratio.
* At the time this function is called, the parameter must be finalized.
* Returns 1 if true, 0 otherwise. */
static int ZSTD_blockSplitterEnabled(ZSTD_CCtx_params* cctxParams)
{
DEBUGLOG(5, "ZSTD_blockSplitterEnabled (postBlockSplitter=%d)", cctxParams->postBlockSplitter);
assert(cctxParams->postBlockSplitter != ZSTD_ps_auto);
return (cctxParams->postBlockSplitter == ZSTD_ps_enable);
}
/* Type returned by ZSTD_buildSequencesStatistics containing finalized symbol encoding types
* and size of the sequences statistics
*/
typedef struct {
U32 LLtype;
U32 Offtype;
U32 MLtype;
size_t size;
size_t lastCountSize; /* Accounts for bug in 1.3.4. More detail in ZSTD_entropyCompressSeqStore_internal() */
int longOffsets;
} ZSTD_symbolEncodingTypeStats_t;
/* ZSTD_buildSequencesStatistics():
* Returns a ZSTD_symbolEncodingTypeStats_t, or a zstd error code in the `size` field.
* Modifies `nextEntropy` to have the appropriate values as a side effect.
* nbSeq must be greater than 0.
*
* entropyWkspSize must be of size at least ENTROPY_WORKSPACE_SIZE - (MaxSeq + 1)*sizeof(U32)
*/
static ZSTD_symbolEncodingTypeStats_t
ZSTD_buildSequencesStatistics(
const SeqStore_t* seqStorePtr, size_t nbSeq,
const ZSTD_fseCTables_t* prevEntropy, ZSTD_fseCTables_t* nextEntropy,
BYTE* dst, const BYTE* const dstEnd,
ZSTD_strategy strategy, unsigned* countWorkspace,
void* entropyWorkspace, size_t entropyWkspSize)
{
BYTE* const ostart = dst;
const BYTE* const oend = dstEnd;
BYTE* op = ostart;
FSE_CTable* CTable_LitLength = nextEntropy->litlengthCTable;
FSE_CTable* CTable_OffsetBits = nextEntropy->offcodeCTable;
FSE_CTable* CTable_MatchLength = nextEntropy->matchlengthCTable;
const BYTE* const ofCodeTable = seqStorePtr->ofCode;
const BYTE* const llCodeTable = seqStorePtr->llCode;
const BYTE* const mlCodeTable = seqStorePtr->mlCode;
ZSTD_symbolEncodingTypeStats_t stats;
stats.lastCountSize = 0;
/* convert length/distances into codes */
stats.longOffsets = ZSTD_seqToCodes(seqStorePtr);
assert(op <= oend);
assert(nbSeq != 0); /* ZSTD_selectEncodingType() divides by nbSeq */
/* build CTable for Literal Lengths */
{ unsigned max = MaxLL;
size_t const mostFrequent = HIST_countFast_wksp(countWorkspace, &max, llCodeTable, nbSeq, entropyWorkspace, entropyWkspSize); /* can't fail */
DEBUGLOG(5, "Building LL table");
nextEntropy->litlength_repeatMode = prevEntropy->litlength_repeatMode;
stats.LLtype = ZSTD_selectEncodingType(&nextEntropy->litlength_repeatMode,
countWorkspace, max, mostFrequent, nbSeq,
LLFSELog, prevEntropy->litlengthCTable,
LL_defaultNorm, LL_defaultNormLog,
ZSTD_defaultAllowed, strategy);
assert(set_basic < set_compressed && set_rle < set_compressed);
assert(!(stats.LLtype < set_compressed && nextEntropy->litlength_repeatMode != FSE_repeat_none)); /* We don't copy tables */
{ size_t const countSize = ZSTD_buildCTable(
op, (size_t)(oend - op),
CTable_LitLength, LLFSELog, (SymbolEncodingType_e)stats.LLtype,
countWorkspace, max, llCodeTable, nbSeq,
LL_defaultNorm, LL_defaultNormLog, MaxLL,
prevEntropy->litlengthCTable,
sizeof(prevEntropy->litlengthCTable),
entropyWorkspace, entropyWkspSize);
if (ZSTD_isError(countSize)) {
DEBUGLOG(3, "ZSTD_buildCTable for LitLens failed");
stats.size = countSize;
return stats;
}
if (stats.LLtype == set_compressed)
stats.lastCountSize = countSize;
op += countSize;
assert(op <= oend);
} }
/* build CTable for Offsets */
{ unsigned max = MaxOff;
size_t const mostFrequent = HIST_countFast_wksp(
countWorkspace, &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");
nextEntropy->offcode_repeatMode = prevEntropy->offcode_repeatMode;
stats.Offtype = ZSTD_selectEncodingType(&nextEntropy->offcode_repeatMode,
countWorkspace, max, mostFrequent, nbSeq,
OffFSELog, prevEntropy->offcodeCTable,
OF_defaultNorm, OF_defaultNormLog,
defaultPolicy, strategy);
assert(!(stats.Offtype < set_compressed && nextEntropy->offcode_repeatMode != FSE_repeat_none)); /* We don't copy tables */
{ size_t const countSize = ZSTD_buildCTable(
op, (size_t)(oend - op),
CTable_OffsetBits, OffFSELog, (SymbolEncodingType_e)stats.Offtype,
countWorkspace, max, ofCodeTable, nbSeq,
OF_defaultNorm, OF_defaultNormLog, DefaultMaxOff,
prevEntropy->offcodeCTable,
sizeof(prevEntropy->offcodeCTable),
entropyWorkspace, entropyWkspSize);
if (ZSTD_isError(countSize)) {
DEBUGLOG(3, "ZSTD_buildCTable for Offsets failed");
stats.size = countSize;
return stats;
}
if (stats.Offtype == set_compressed)
stats.lastCountSize = countSize;
op += countSize;
assert(op <= oend);
} }
/* build CTable for MatchLengths */
{ unsigned max = MaxML;
size_t const mostFrequent = HIST_countFast_wksp(
countWorkspace, &max, mlCodeTable, nbSeq, entropyWorkspace, entropyWkspSize); /* can't fail */
DEBUGLOG(5, "Building ML table (remaining space : %i)", (int)(oend-op));
nextEntropy->matchlength_repeatMode = prevEntropy->matchlength_repeatMode;
stats.MLtype = ZSTD_selectEncodingType(&nextEntropy->matchlength_repeatMode,
countWorkspace, max, mostFrequent, nbSeq,
MLFSELog, prevEntropy->matchlengthCTable,
ML_defaultNorm, ML_defaultNormLog,
ZSTD_defaultAllowed, strategy);
assert(!(stats.MLtype < set_compressed && nextEntropy->matchlength_repeatMode != FSE_repeat_none)); /* We don't copy tables */
{ size_t const countSize = ZSTD_buildCTable(
op, (size_t)(oend - op),
CTable_MatchLength, MLFSELog, (SymbolEncodingType_e)stats.MLtype,
countWorkspace, max, mlCodeTable, nbSeq,
ML_defaultNorm, ML_defaultNormLog, MaxML,
prevEntropy->matchlengthCTable,
sizeof(prevEntropy->matchlengthCTable),
entropyWorkspace, entropyWkspSize);
if (ZSTD_isError(countSize)) {
DEBUGLOG(3, "ZSTD_buildCTable for MatchLengths failed");
stats.size = countSize;
return stats;
}
if (stats.MLtype == set_compressed)
stats.lastCountSize = countSize;
op += countSize;
assert(op <= oend);
} }
stats.size = (size_t)(op-ostart);
return stats;
}
/* ZSTD_entropyCompressSeqStore_internal():
* compresses both literals and sequences
* Returns compressed size of block, or a zstd error.
*/
#define SUSPECT_UNCOMPRESSIBLE_LITERAL_RATIO 20
MEM_STATIC size_t
ZSTD_entropyCompressSeqStore_internal(
void* dst, size_t dstCapacity,
const void* literals, size_t litSize,
const SeqStore_t* seqStorePtr,
const ZSTD_entropyCTables_t* prevEntropy,
ZSTD_entropyCTables_t* nextEntropy,
const ZSTD_CCtx_params* cctxParams,
void* entropyWorkspace, size_t entropyWkspSize,
const int bmi2)
{
ZSTD_strategy const strategy = cctxParams->cParams.strategy;
unsigned* count = (unsigned*)entropyWorkspace;
FSE_CTable* CTable_LitLength = nextEntropy->fse.litlengthCTable;
FSE_CTable* CTable_OffsetBits = nextEntropy->fse.offcodeCTable;
FSE_CTable* CTable_MatchLength = nextEntropy->fse.matchlengthCTable;
const SeqDef* const sequences = seqStorePtr->sequencesStart;
const size_t nbSeq = (size_t)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
const BYTE* const ofCodeTable = seqStorePtr->ofCode;
const BYTE* const llCodeTable = seqStorePtr->llCode;
const BYTE* const mlCodeTable = seqStorePtr->mlCode;
BYTE* const ostart = (BYTE*)dst;
BYTE* const oend = ostart + dstCapacity;
BYTE* op = ostart;
size_t lastCountSize;
int longOffsets = 0;
entropyWorkspace = count + (MaxSeq + 1);
entropyWkspSize -= (MaxSeq + 1) * sizeof(*count);
DEBUGLOG(5, "ZSTD_entropyCompressSeqStore_internal (nbSeq=%zu, dstCapacity=%zu)", nbSeq, dstCapacity);
ZSTD_STATIC_ASSERT(HUF_WORKSPACE_SIZE >= (1<<MAX(MLFSELog,LLFSELog)));
assert(entropyWkspSize >= HUF_WORKSPACE_SIZE);
/* Compress literals */
{ size_t const numSequences = (size_t)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
/* Base suspicion of uncompressibility on ratio of literals to sequences */
int const suspectUncompressible = (numSequences == 0) || (litSize / numSequences >= SUSPECT_UNCOMPRESSIBLE_LITERAL_RATIO);
size_t const cSize = ZSTD_compressLiterals(
op, dstCapacity,
literals, litSize,
entropyWorkspace, entropyWkspSize,
&prevEntropy->huf, &nextEntropy->huf,
cctxParams->cParams.strategy,
ZSTD_literalsCompressionIsDisabled(cctxParams),
suspectUncompressible, bmi2);
FORWARD_IF_ERROR(cSize, "ZSTD_compressLiterals failed");
assert(cSize <= dstCapacity);
op += cSize;
}
/* Sequences Header */
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;
}
assert(op <= oend);
if (nbSeq==0) {
/* Copy the old tables over as if we repeated them */
ZSTD_memcpy(&nextEntropy->fse, &prevEntropy->fse, sizeof(prevEntropy->fse));
return (size_t)(op - ostart);
}
{ BYTE* const seqHead = op++;
/* build stats for sequences */
const ZSTD_symbolEncodingTypeStats_t stats =
ZSTD_buildSequencesStatistics(seqStorePtr, nbSeq,
&prevEntropy->fse, &nextEntropy->fse,
op, oend,
strategy, count,
entropyWorkspace, entropyWkspSize);
FORWARD_IF_ERROR(stats.size, "ZSTD_buildSequencesStatistics failed!");
*seqHead = (BYTE)((stats.LLtype<<6) + (stats.Offtype<<4) + (stats.MLtype<<2));
lastCountSize = stats.lastCountSize;
op += stats.size;
longOffsets = stats.longOffsets;
}
{ size_t const bitstreamSize = ZSTD_encodeSequences(
op, (size_t)(oend - op),
CTable_MatchLength, mlCodeTable,
CTable_OffsetBits, ofCodeTable,
CTable_LitLength, llCodeTable,
sequences, nbSeq,
longOffsets, bmi2);
FORWARD_IF_ERROR(bitstreamSize, "ZSTD_encodeSequences failed");
op += bitstreamSize;
assert(op <= oend);
/* zstd versions <= 1.3.4 mistakenly report corruption when
* 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.
* In this exceedingly rare case, we will simply emit an uncompressed
* block, since it isn't worth optimizing.
*/
if (lastCountSize && (lastCountSize + bitstreamSize) < 4) {
/* lastCountSize >= 2 && bitstreamSize > 0 ==> lastCountSize == 3 */
assert(lastCountSize + bitstreamSize == 3);
DEBUGLOG(5, "Avoiding bug in zstd decoder in versions <= 1.3.4 by "
"emitting an uncompressed block.");
return 0;
}
}
DEBUGLOG(5, "compressed block size : %u", (unsigned)(op - ostart));
return (size_t)(op - ostart);
}
static size_t
ZSTD_entropyCompressSeqStore_wExtLitBuffer(
void* dst, size_t dstCapacity,
const void* literals, size_t litSize,
size_t blockSize,
const SeqStore_t* seqStorePtr,
const ZSTD_entropyCTables_t* prevEntropy,
ZSTD_entropyCTables_t* nextEntropy,
const ZSTD_CCtx_params* cctxParams,
void* entropyWorkspace, size_t entropyWkspSize,
int bmi2)
{
size_t const cSize = ZSTD_entropyCompressSeqStore_internal(
dst, dstCapacity,
literals, litSize,
seqStorePtr, prevEntropy, nextEntropy, cctxParams,
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)) & (blockSize <= dstCapacity)) {
DEBUGLOG(4, "not enough dstCapacity (%zu) for ZSTD_entropyCompressSeqStore_internal()=> do not compress block", dstCapacity);
return 0; /* block not compressed */
}
FORWARD_IF_ERROR(cSize, "ZSTD_entropyCompressSeqStore_internal failed");
/* Check compressibility */
{ size_t const maxCSize = blockSize - ZSTD_minGain(blockSize, cctxParams->cParams.strategy);
if (cSize >= maxCSize) return 0; /* block not compressed */
}
DEBUGLOG(5, "ZSTD_entropyCompressSeqStore() cSize: %zu", cSize);
/* libzstd decoder before > v1.5.4 is not compatible with compressed blocks of size ZSTD_BLOCKSIZE_MAX exactly.
* This restriction is indirectly already fulfilled by respecting ZSTD_minGain() condition above.
*/
assert(cSize < ZSTD_BLOCKSIZE_MAX);
return cSize;
}
static size_t
ZSTD_entropyCompressSeqStore(
const SeqStore_t* seqStorePtr,
const ZSTD_entropyCTables_t* prevEntropy,
ZSTD_entropyCTables_t* nextEntropy,
const ZSTD_CCtx_params* cctxParams,
void* dst, size_t dstCapacity,
size_t srcSize,
void* entropyWorkspace, size_t entropyWkspSize,
int bmi2)
{
return ZSTD_entropyCompressSeqStore_wExtLitBuffer(
dst, dstCapacity,
seqStorePtr->litStart, (size_t)(seqStorePtr->lit - seqStorePtr->litStart),
srcSize,
seqStorePtr,
prevEntropy, nextEntropy,
cctxParams,
entropyWorkspace, entropyWkspSize,
bmi2);
}
/* ZSTD_selectBlockCompressor() :
* Not static, but internal use only (used by long distance matcher)
* assumption : strat is a valid strategy */
ZSTD_BlockCompressor_f ZSTD_selectBlockCompressor(ZSTD_strategy strat, ZSTD_ParamSwitch_e useRowMatchFinder, ZSTD_dictMode_e dictMode)
{
static const ZSTD_BlockCompressor_f blockCompressor[4][ZSTD_STRATEGY_MAX+1] = {
{ ZSTD_compressBlock_fast /* default for 0 */,
ZSTD_compressBlock_fast,
ZSTD_COMPRESSBLOCK_DOUBLEFAST,
ZSTD_COMPRESSBLOCK_GREEDY,
ZSTD_COMPRESSBLOCK_LAZY,
ZSTD_COMPRESSBLOCK_LAZY2,
ZSTD_COMPRESSBLOCK_BTLAZY2,
ZSTD_COMPRESSBLOCK_BTOPT,
ZSTD_COMPRESSBLOCK_BTULTRA,
ZSTD_COMPRESSBLOCK_BTULTRA2
},
{ ZSTD_compressBlock_fast_extDict /* default for 0 */,
ZSTD_compressBlock_fast_extDict,
ZSTD_COMPRESSBLOCK_DOUBLEFAST_EXTDICT,
ZSTD_COMPRESSBLOCK_GREEDY_EXTDICT,
ZSTD_COMPRESSBLOCK_LAZY_EXTDICT,
ZSTD_COMPRESSBLOCK_LAZY2_EXTDICT,
ZSTD_COMPRESSBLOCK_BTLAZY2_EXTDICT,
ZSTD_COMPRESSBLOCK_BTOPT_EXTDICT,
ZSTD_COMPRESSBLOCK_BTULTRA_EXTDICT,
ZSTD_COMPRESSBLOCK_BTULTRA_EXTDICT
},
{ ZSTD_compressBlock_fast_dictMatchState /* default for 0 */,
ZSTD_compressBlock_fast_dictMatchState,
ZSTD_COMPRESSBLOCK_DOUBLEFAST_DICTMATCHSTATE,
ZSTD_COMPRESSBLOCK_GREEDY_DICTMATCHSTATE,
ZSTD_COMPRESSBLOCK_LAZY_DICTMATCHSTATE,
ZSTD_COMPRESSBLOCK_LAZY2_DICTMATCHSTATE,
ZSTD_COMPRESSBLOCK_BTLAZY2_DICTMATCHSTATE,
ZSTD_COMPRESSBLOCK_BTOPT_DICTMATCHSTATE,
ZSTD_COMPRESSBLOCK_BTULTRA_DICTMATCHSTATE,
ZSTD_COMPRESSBLOCK_BTULTRA_DICTMATCHSTATE
},
{ NULL /* default for 0 */,
NULL,
NULL,
ZSTD_COMPRESSBLOCK_GREEDY_DEDICATEDDICTSEARCH,
ZSTD_COMPRESSBLOCK_LAZY_DEDICATEDDICTSEARCH,
ZSTD_COMPRESSBLOCK_LAZY2_DEDICATEDDICTSEARCH,
NULL,
NULL,
NULL,
NULL }
};
ZSTD_BlockCompressor_f selectedCompressor;
ZSTD_STATIC_ASSERT((unsigned)ZSTD_fast == 1);
assert(ZSTD_cParam_withinBounds(ZSTD_c_strategy, (int)strat));
DEBUGLOG(5, "Selected block compressor: dictMode=%d strat=%d rowMatchfinder=%d", (int)dictMode, (int)strat, (int)useRowMatchFinder);
if (ZSTD_rowMatchFinderUsed(strat, useRowMatchFinder)) {
static const ZSTD_BlockCompressor_f rowBasedBlockCompressors[4][3] = {
{
ZSTD_COMPRESSBLOCK_GREEDY_ROW,
ZSTD_COMPRESSBLOCK_LAZY_ROW,
ZSTD_COMPRESSBLOCK_LAZY2_ROW
},
{
ZSTD_COMPRESSBLOCK_GREEDY_EXTDICT_ROW,
ZSTD_COMPRESSBLOCK_LAZY_EXTDICT_ROW,
ZSTD_COMPRESSBLOCK_LAZY2_EXTDICT_ROW
},
{
ZSTD_COMPRESSBLOCK_GREEDY_DICTMATCHSTATE_ROW,
ZSTD_COMPRESSBLOCK_LAZY_DICTMATCHSTATE_ROW,
ZSTD_COMPRESSBLOCK_LAZY2_DICTMATCHSTATE_ROW
},
{
ZSTD_COMPRESSBLOCK_GREEDY_DEDICATEDDICTSEARCH_ROW,
ZSTD_COMPRESSBLOCK_LAZY_DEDICATEDDICTSEARCH_ROW,
ZSTD_COMPRESSBLOCK_LAZY2_DEDICATEDDICTSEARCH_ROW
}
};
DEBUGLOG(5, "Selecting a row-based matchfinder");
assert(useRowMatchFinder != ZSTD_ps_auto);
selectedCompressor = rowBasedBlockCompressors[(int)dictMode][(int)strat - (int)ZSTD_greedy];
} else {
selectedCompressor = blockCompressor[(int)dictMode][(int)strat];
}
assert(selectedCompressor != NULL);
return selectedCompressor;
}
static void ZSTD_storeLastLiterals(SeqStore_t* seqStorePtr,
const BYTE* anchor, size_t lastLLSize)
{
ZSTD_memcpy(seqStorePtr->lit, anchor, lastLLSize);
seqStorePtr->lit += lastLLSize;
}
void ZSTD_resetSeqStore(SeqStore_t* ssPtr)
{
ssPtr->lit = ssPtr->litStart;
ssPtr->sequences = ssPtr->sequencesStart;
ssPtr->longLengthType = ZSTD_llt_none;
}
/* ZSTD_postProcessSequenceProducerResult() :
* Validates and post-processes sequences obtained through the external matchfinder API:
* - Checks whether nbExternalSeqs represents an error condition.
* - Appends a block delimiter to outSeqs if one is not already present.
* See zstd.h for context regarding block delimiters.
* Returns the number of sequences after post-processing, or an error code. */
static size_t ZSTD_postProcessSequenceProducerResult(
ZSTD_Sequence* outSeqs, size_t nbExternalSeqs, size_t outSeqsCapacity, size_t srcSize
) {
RETURN_ERROR_IF(
nbExternalSeqs > outSeqsCapacity,
sequenceProducer_failed,
"External sequence producer returned error code %lu",
(unsigned long)nbExternalSeqs
);
RETURN_ERROR_IF(
nbExternalSeqs == 0 && srcSize > 0,
sequenceProducer_failed,
"Got zero sequences from external sequence producer for a non-empty src buffer!"
);
if (srcSize == 0) {
ZSTD_memset(&outSeqs[0], 0, sizeof(ZSTD_Sequence));
return 1;
}
{
ZSTD_Sequence const lastSeq = outSeqs[nbExternalSeqs - 1];
/* We can return early if lastSeq is already a block delimiter. */
if (lastSeq.offset == 0 && lastSeq.matchLength == 0) {
return nbExternalSeqs;
}
/* This error condition is only possible if the external matchfinder
* produced an invalid parse, by definition of ZSTD_sequenceBound(). */
RETURN_ERROR_IF(
nbExternalSeqs == outSeqsCapacity,
sequenceProducer_failed,
"nbExternalSeqs == outSeqsCapacity but lastSeq is not a block delimiter!"
);
/* lastSeq is not a block delimiter, so we need to append one. */
ZSTD_memset(&outSeqs[nbExternalSeqs], 0, sizeof(ZSTD_Sequence));
return nbExternalSeqs + 1;
}
}
/* ZSTD_fastSequenceLengthSum() :
* Returns sum(litLen) + sum(matchLen) + lastLits for *seqBuf*.
* Similar to another function in zstd_compress.c (determine_blockSize),
* except it doesn't check for a block delimiter to end summation.
* Removing the early exit allows the compiler to auto-vectorize (https://godbolt.org/z/cY1cajz9P).
* This function can be deleted and replaced by determine_blockSize after we resolve issue #3456. */
static size_t ZSTD_fastSequenceLengthSum(ZSTD_Sequence const* seqBuf, size_t seqBufSize) {
size_t matchLenSum, litLenSum, i;
matchLenSum = 0;
litLenSum = 0;
for (i = 0; i < seqBufSize; i++) {
litLenSum += seqBuf[i].litLength;
matchLenSum += seqBuf[i].matchLength;
}
return litLenSum + matchLenSum;
}
/**
* Function to validate sequences produced by a block compressor.
*/
static void ZSTD_validateSeqStore(const SeqStore_t* seqStore, const ZSTD_compressionParameters* cParams)
{
#if DEBUGLEVEL >= 1
const SeqDef* seq = seqStore->sequencesStart;
const SeqDef* const seqEnd = seqStore->sequences;
size_t const matchLenLowerBound = cParams->minMatch == 3 ? 3 : 4;
for (; seq < seqEnd; ++seq) {
const ZSTD_SequenceLength seqLength = ZSTD_getSequenceLength(seqStore, seq);
assert(seqLength.matchLength >= matchLenLowerBound);
(void)seqLength;
(void)matchLenLowerBound;
}
#else
(void)seqStore;
(void)cParams;
#endif
}
static size_t
ZSTD_transferSequences_wBlockDelim(ZSTD_CCtx* cctx,
ZSTD_SequencePosition* seqPos,
const ZSTD_Sequence* const inSeqs, size_t inSeqsSize,
const void* src, size_t blockSize,
ZSTD_ParamSwitch_e externalRepSearch);
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;
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);
/* TODO: See 3090. We reduced MIN_CBLOCK_SIZE from 3 to 2 so to compensate we are adding
* additional 1. We need to revisit and change this logic to be more consistent */
if (srcSize < MIN_CBLOCK_SIZE+ZSTD_blockHeaderSize+1+1) {
if (zc->appliedParams.cParams.strategy >= ZSTD_btopt) {
ZSTD_ldm_skipRawSeqStoreBytes(&zc->externSeqStore, srcSize);
} else {
ZSTD_ldm_skipSequences(&zc->externSeqStore, srcSize, zc->appliedParams.cParams.minMatch);
}
return ZSTDbss_noCompress; /* don't even attempt compression below a certain srcSize */
}
ZSTD_resetSeqStore(&(zc->seqStore));
/* 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 */
assert(ms->dictMatchState == NULL || ms->loadedDictEnd == ms->window.dictLimit);
/* limited update after a very long match */
{ const BYTE* const base = ms->window.base;
const BYTE* const istart = (const BYTE*)src;
const U32 curr = (U32)(istart-base);
if (sizeof(ptrdiff_t)==8) assert(istart - base < (ptrdiff_t)(U32)(-1)); /* ensure no overflow */
if (curr > ms->nextToUpdate + 384)
ms->nextToUpdate = curr - MIN(192, (U32)(curr - ms->nextToUpdate - 384));
}
/* select and store sequences */
{ ZSTD_dictMode_e const dictMode = ZSTD_matchState_dictMode(ms);
size_t lastLLSize;
{ int i;
for (i = 0; i < ZSTD_REP_NUM; ++i)
zc->blockState.nextCBlock->rep[i] = zc->blockState.prevCBlock->rep[i];
}
if (zc->externSeqStore.pos < zc->externSeqStore.size) {
assert(zc->appliedParams.ldmParams.enableLdm == ZSTD_ps_disable);
/* External matchfinder + LDM is technically possible, just not implemented yet.
* We need to revisit soon and implement it. */
RETURN_ERROR_IF(
ZSTD_hasExtSeqProd(&zc->appliedParams),
parameter_combination_unsupported,
"Long-distance matching with external sequence producer enabled is not currently supported."
);
/* Updates ldmSeqStore.pos */
lastLLSize =
ZSTD_ldm_blockCompress(&zc->externSeqStore,
ms, &zc->seqStore,
zc->blockState.nextCBlock->rep,
zc->appliedParams.useRowMatchFinder,
src, srcSize);
assert(zc->externSeqStore.pos <= zc->externSeqStore.size);
} else if (zc->appliedParams.ldmParams.enableLdm == ZSTD_ps_enable) {
RawSeqStore_t ldmSeqStore = kNullRawSeqStore;
/* External matchfinder + LDM is technically possible, just not implemented yet.
* We need to revisit soon and implement it. */
RETURN_ERROR_IF(
ZSTD_hasExtSeqProd(&zc->appliedParams),
parameter_combination_unsupported,
"Long-distance matching with external sequence producer enabled is not currently supported."
);
ldmSeqStore.seq = zc->ldmSequences;
ldmSeqStore.capacity = zc->maxNbLdmSequences;
/* Updates ldmSeqStore.size */
FORWARD_IF_ERROR(ZSTD_ldm_generateSequences(&zc->ldmState, &ldmSeqStore,
&zc->appliedParams.ldmParams,
src, srcSize), "");
/* Updates ldmSeqStore.pos */
lastLLSize =
ZSTD_ldm_blockCompress(&ldmSeqStore,
ms, &zc->seqStore,
zc->blockState.nextCBlock->rep,
zc->appliedParams.useRowMatchFinder,
src, srcSize);
assert(ldmSeqStore.pos == ldmSeqStore.size);
} else if (ZSTD_hasExtSeqProd(&zc->appliedParams)) {
assert(
zc->extSeqBufCapacity >= ZSTD_sequenceBound(srcSize)
);
assert(zc->appliedParams.extSeqProdFunc != NULL);
{ U32 const windowSize = (U32)1 << zc->appliedParams.cParams.windowLog;
size_t const nbExternalSeqs = (zc->appliedParams.extSeqProdFunc)(
zc->appliedParams.extSeqProdState,
zc->extSeqBuf,
zc->extSeqBufCapacity,
src, srcSize,
NULL, 0, /* dict and dictSize, currently not supported */
zc->appliedParams.compressionLevel,
windowSize
);
size_t const nbPostProcessedSeqs = ZSTD_postProcessSequenceProducerResult(
zc->extSeqBuf,
nbExternalSeqs,
zc->extSeqBufCapacity,
srcSize
);
/* Return early if there is no error, since we don't need to worry about last literals */
if (!ZSTD_isError(nbPostProcessedSeqs)) {
ZSTD_SequencePosition seqPos = {0,0,0};
size_t const seqLenSum = ZSTD_fastSequenceLengthSum(zc->extSeqBuf, nbPostProcessedSeqs);
RETURN_ERROR_IF(seqLenSum > srcSize, externalSequences_invalid, "External sequences imply too large a block!");
FORWARD_IF_ERROR(
ZSTD_transferSequences_wBlockDelim(
zc, &seqPos,
zc->extSeqBuf, nbPostProcessedSeqs,
src, srcSize,
zc->appliedParams.searchForExternalRepcodes
),
"Failed to copy external sequences to seqStore!"
);
ms->ldmSeqStore = NULL;
DEBUGLOG(5, "Copied %lu sequences from external sequence producer to internal seqStore.", (unsigned long)nbExternalSeqs);
return ZSTDbss_compress;
}
/* Propagate the error if fallback is disabled */
if (!zc->appliedParams.enableMatchFinderFallback) {
return nbPostProcessedSeqs;
}
/* Fallback to software matchfinder */
{ ZSTD_BlockCompressor_f const blockCompressor =
ZSTD_selectBlockCompressor(
zc->appliedParams.cParams.strategy,
zc->appliedParams.useRowMatchFinder,
dictMode);
ms->ldmSeqStore = NULL;
DEBUGLOG(
5,
"External sequence producer returned error code %lu. Falling back to internal parser.",
(unsigned long)nbExternalSeqs
);
lastLLSize = blockCompressor(ms, &zc->seqStore, zc->blockState.nextCBlock->rep, src, srcSize);
} }
} else { /* not long range mode and no external matchfinder */
ZSTD_BlockCompressor_f const blockCompressor = ZSTD_selectBlockCompressor(
zc->appliedParams.cParams.strategy,
zc->appliedParams.useRowMatchFinder,
dictMode);
ms->ldmSeqStore = NULL;
lastLLSize = blockCompressor(ms, &zc->seqStore, zc->blockState.nextCBlock->rep, src, srcSize);
}
{ const BYTE* const lastLiterals = (const BYTE*)src + srcSize - lastLLSize;
ZSTD_storeLastLiterals(&zc->seqStore, lastLiterals, lastLLSize);
} }
ZSTD_validateSeqStore(&zc->seqStore, &zc->appliedParams.cParams);
return ZSTDbss_compress;
}
static size_t ZSTD_copyBlockSequences(SeqCollector* seqCollector, const SeqStore_t* seqStore, const U32 prevRepcodes[ZSTD_REP_NUM])
{
const SeqDef* inSeqs = seqStore->sequencesStart;
const size_t nbInSequences = (size_t)(seqStore->sequences - inSeqs);
const size_t nbInLiterals = (size_t)(seqStore->lit - seqStore->litStart);
ZSTD_Sequence* outSeqs = seqCollector->seqIndex == 0 ? seqCollector->seqStart : seqCollector->seqStart + seqCollector->seqIndex;
const size_t nbOutSequences = nbInSequences + 1;
size_t nbOutLiterals = 0;
Repcodes_t repcodes;
size_t i;
/* Bounds check that we have enough space for every input sequence
* and the block delimiter
*/
assert(seqCollector->seqIndex <= seqCollector->maxSequences);
RETURN_ERROR_IF(
nbOutSequences > (size_t)(seqCollector->maxSequences - seqCollector->seqIndex),
dstSize_tooSmall,
"Not enough space to copy sequences");
ZSTD_memcpy(&repcodes, prevRepcodes, sizeof(repcodes));
for (i = 0; i < nbInSequences; ++i) {
U32 rawOffset;
outSeqs[i].litLength = inSeqs[i].litLength;
outSeqs[i].matchLength = inSeqs[i].mlBase + MINMATCH;
outSeqs[i].rep = 0;
/* Handle the possible single length >= 64K
* There can only be one because we add MINMATCH to every match length,
* and blocks are at most 128K.
*/
if (i == seqStore->longLengthPos) {
if (seqStore->longLengthType == ZSTD_llt_literalLength) {
outSeqs[i].litLength += 0x10000;
} else if (seqStore->longLengthType == ZSTD_llt_matchLength) {
outSeqs[i].matchLength += 0x10000;
}
}
/* Determine the raw offset given the offBase, which may be a repcode. */
if (OFFBASE_IS_REPCODE(inSeqs[i].offBase)) {
const U32 repcode = OFFBASE_TO_REPCODE(inSeqs[i].offBase);
assert(repcode > 0);
outSeqs[i].rep = repcode;
if (outSeqs[i].litLength != 0) {
rawOffset = repcodes.rep[repcode - 1];
} else {
if (repcode == 3) {
assert(repcodes.rep[0] > 1);
rawOffset = repcodes.rep[0] - 1;
} else {
rawOffset = repcodes.rep[repcode];
}
}
} else {
rawOffset = OFFBASE_TO_OFFSET(inSeqs[i].offBase);
}
outSeqs[i].offset = rawOffset;
/* Update repcode history for the sequence */
ZSTD_updateRep(repcodes.rep,
inSeqs[i].offBase,
inSeqs[i].litLength == 0);
nbOutLiterals += outSeqs[i].litLength;
}
/* Insert last literals (if any exist) in the block as a sequence with ml == off == 0.
* If there are no last literals, then we'll emit (of: 0, ml: 0, ll: 0), which is a marker
* for the block boundary, according to the API.
*/
assert(nbInLiterals >= nbOutLiterals);
{
const size_t lastLLSize = nbInLiterals - nbOutLiterals;
outSeqs[nbInSequences].litLength = (U32)lastLLSize;
outSeqs[nbInSequences].matchLength = 0;
outSeqs[nbInSequences].offset = 0;
assert(nbOutSequences == nbInSequences + 1);
}
seqCollector->seqIndex += nbOutSequences;
assert(seqCollector->seqIndex <= seqCollector->maxSequences);
return 0;
}
size_t ZSTD_sequenceBound(size_t srcSize) {
const size_t maxNbSeq = (srcSize / ZSTD_MINMATCH_MIN) + 1;
const size_t maxNbDelims = (srcSize / ZSTD_BLOCKSIZE_MAX_MIN) + 1;
return maxNbSeq + maxNbDelims;
}
size_t ZSTD_generateSequences(ZSTD_CCtx* zc, ZSTD_Sequence* outSeqs,
size_t outSeqsSize, const void* src, size_t srcSize)
{
const size_t dstCapacity = ZSTD_compressBound(srcSize);
void* dst; /* Make C90 happy. */
SeqCollector seqCollector;
{
int targetCBlockSize;
FORWARD_IF_ERROR(ZSTD_CCtx_getParameter(zc, ZSTD_c_targetCBlockSize, &targetCBlockSize), "");
RETURN_ERROR_IF(targetCBlockSize != 0, parameter_unsupported, "targetCBlockSize != 0");
}
{
int nbWorkers;
FORWARD_IF_ERROR(ZSTD_CCtx_getParameter(zc, ZSTD_c_nbWorkers, &nbWorkers), "");
RETURN_ERROR_IF(nbWorkers != 0, parameter_unsupported, "nbWorkers != 0");
}
dst = ZSTD_customMalloc(dstCapacity, ZSTD_defaultCMem);
RETURN_ERROR_IF(dst == NULL, memory_allocation, "NULL pointer!");
seqCollector.collectSequences = 1;
seqCollector.seqStart = outSeqs;
seqCollector.seqIndex = 0;
seqCollector.maxSequences = outSeqsSize;
zc->seqCollector = seqCollector;
{
const size_t ret = ZSTD_compress2(zc, dst, dstCapacity, src, srcSize);
ZSTD_customFree(dst, ZSTD_defaultCMem);
FORWARD_IF_ERROR(ret, "ZSTD_compress2 failed");
}
assert(zc->seqCollector.seqIndex <= ZSTD_sequenceBound(srcSize));
return zc->seqCollector.seqIndex;
}
size_t ZSTD_mergeBlockDelimiters(ZSTD_Sequence* sequences, size_t seqsSize) {
size_t in = 0;
size_t out = 0;
for (; in < seqsSize; ++in) {
if (sequences[in].offset == 0 && sequences[in].matchLength == 0) {
if (in != seqsSize - 1) {
sequences[in+1].litLength += sequences[in].litLength;
}
} else {
sequences[out] = sequences[in];
++out;
}
}
return out;
}
/* Unrolled loop to read four size_ts of input at a time. Returns 1 if is RLE, 0 if not. */
static int ZSTD_isRLE(const BYTE* src, size_t length) {
const BYTE* ip = src;
const BYTE value = ip[0];
const size_t valueST = (size_t)((U64)value * 0x0101010101010101ULL);
const size_t unrollSize = sizeof(size_t) * 4;
const size_t unrollMask = unrollSize - 1;
const size_t prefixLength = length & unrollMask;
size_t i;
if (length == 1) return 1;
/* Check if prefix is RLE first before using unrolled loop */
if (prefixLength && ZSTD_count(ip+1, ip, ip+prefixLength) != prefixLength-1) {
return 0;
}
for (i = prefixLength; i != length; i += unrollSize) {
size_t u;
for (u = 0; u < unrollSize; u += sizeof(size_t)) {
if (MEM_readST(ip + i + u) != valueST) {
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_blockState_confirmRepcodesAndEntropyTables(ZSTD_blockState_t* const bs)
{
ZSTD_compressedBlockState_t* const tmp = bs->prevCBlock;
bs->prevCBlock = bs->nextCBlock;
bs->nextCBlock = tmp;
}
/* Writes the block header */
static void
writeBlockHeader(void* op, size_t cSize, size_t blockSize, U32 lastBlock)
{
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);
DEBUGLOG(5, "writeBlockHeader: cSize: %zu blockSize: %zu lastBlock: %u", cSize, blockSize, lastBlock);
}
/** ZSTD_buildBlockEntropyStats_literals() :
* Builds entropy for the literals.
* Stores literals block type (raw, rle, compressed, repeat) and
* huffman description table to hufMetadata.
* Requires ENTROPY_WORKSPACE_SIZE workspace
* @return : size of huffman description table, or an error code
*/
static size_t
ZSTD_buildBlockEntropyStats_literals(void* const src, size_t srcSize,
const ZSTD_hufCTables_t* prevHuf,
ZSTD_hufCTables_t* nextHuf,
ZSTD_hufCTablesMetadata_t* hufMetadata,
const int literalsCompressionIsDisabled,
void* workspace, size_t wkspSize,
int hufFlags)
{
BYTE* const wkspStart = (BYTE*)workspace;
BYTE* const wkspEnd = wkspStart + wkspSize;
BYTE* const countWkspStart = wkspStart;
unsigned* const countWksp = (unsigned*)workspace;
const size_t countWkspSize = (HUF_SYMBOLVALUE_MAX + 1) * sizeof(unsigned);
BYTE* const nodeWksp = countWkspStart + countWkspSize;
const size_t nodeWkspSize = (size_t)(wkspEnd - nodeWksp);
unsigned maxSymbolValue = HUF_SYMBOLVALUE_MAX;
unsigned huffLog = LitHufLog;
HUF_repeat repeat = prevHuf->repeatMode;
DEBUGLOG(5, "ZSTD_buildBlockEntropyStats_literals (srcSize=%zu)", srcSize);
/* Prepare nextEntropy assuming reusing the existing table */
ZSTD_memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
if (literalsCompressionIsDisabled) {
DEBUGLOG(5, "set_basic - disabled");
hufMetadata->hType = set_basic;
return 0;
}
/* small ? don't even attempt compression (speed opt) */
#ifndef COMPRESS_LITERALS_SIZE_MIN
# define COMPRESS_LITERALS_SIZE_MIN 63 /* heuristic */
#endif
{ size_t const minLitSize = (prevHuf->repeatMode == HUF_repeat_valid) ? 6 : COMPRESS_LITERALS_SIZE_MIN;
if (srcSize <= minLitSize) {
DEBUGLOG(5, "set_basic - too small");
hufMetadata->hType = set_basic;
return 0;
} }
/* Scan input and build symbol stats */
{ size_t const largest =
HIST_count_wksp (countWksp, &maxSymbolValue,
(const BYTE*)src, srcSize,
workspace, wkspSize);
FORWARD_IF_ERROR(largest, "HIST_count_wksp failed");
if (largest == srcSize) {
/* only one literal symbol */
DEBUGLOG(5, "set_rle");
hufMetadata->hType = set_rle;
return 0;
}
if (largest <= (srcSize >> 7)+4) {
/* heuristic: likely not compressible */
DEBUGLOG(5, "set_basic - no gain");
hufMetadata->hType = set_basic;
return 0;
} }
/* Validate the previous Huffman table */
if (repeat == HUF_repeat_check
&& !HUF_validateCTable((HUF_CElt const*)prevHuf->CTable, countWksp, maxSymbolValue)) {
repeat = HUF_repeat_none;
}
/* Build Huffman Tree */
ZSTD_memset(nextHuf->CTable, 0, sizeof(nextHuf->CTable));
huffLog = HUF_optimalTableLog(huffLog, srcSize, maxSymbolValue, nodeWksp, nodeWkspSize, nextHuf->CTable, countWksp, hufFlags);
assert(huffLog <= LitHufLog);
{ size_t const maxBits = HUF_buildCTable_wksp((HUF_CElt*)nextHuf->CTable, countWksp,
maxSymbolValue, huffLog,
nodeWksp, nodeWkspSize);
FORWARD_IF_ERROR(maxBits, "HUF_buildCTable_wksp");
huffLog = (U32)maxBits;
}
{ /* Build and write the CTable */
size_t const newCSize = HUF_estimateCompressedSize(
(HUF_CElt*)nextHuf->CTable, countWksp, maxSymbolValue);
size_t const hSize = HUF_writeCTable_wksp(
hufMetadata->hufDesBuffer, sizeof(hufMetadata->hufDesBuffer),
(HUF_CElt*)nextHuf->CTable, maxSymbolValue, huffLog,
nodeWksp, nodeWkspSize);
/* Check against repeating the previous CTable */
if (repeat != HUF_repeat_none) {
size_t const oldCSize = HUF_estimateCompressedSize(
(HUF_CElt const*)prevHuf->CTable, countWksp, maxSymbolValue);
if (oldCSize < srcSize && (oldCSize <= hSize + newCSize || hSize + 12 >= srcSize)) {
DEBUGLOG(5, "set_repeat - smaller");
ZSTD_memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
hufMetadata->hType = set_repeat;
return 0;
} }
if (newCSize + hSize >= srcSize) {
DEBUGLOG(5, "set_basic - no gains");
ZSTD_memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
hufMetadata->hType = set_basic;
return 0;
}
DEBUGLOG(5, "set_compressed (hSize=%u)", (U32)hSize);
hufMetadata->hType = set_compressed;
nextHuf->repeatMode = HUF_repeat_check;
return hSize;
}
}
/* ZSTD_buildDummySequencesStatistics():
* Returns a ZSTD_symbolEncodingTypeStats_t with all encoding types as set_basic,
* and updates nextEntropy to the appropriate repeatMode.
*/
static ZSTD_symbolEncodingTypeStats_t
ZSTD_buildDummySequencesStatistics(ZSTD_fseCTables_t* nextEntropy)
{
ZSTD_symbolEncodingTypeStats_t stats = {set_basic, set_basic, set_basic, 0, 0, 0};
nextEntropy->litlength_repeatMode = FSE_repeat_none;
nextEntropy->offcode_repeatMode = FSE_repeat_none;
nextEntropy->matchlength_repeatMode = FSE_repeat_none;
return stats;
}
/** ZSTD_buildBlockEntropyStats_sequences() :
* Builds entropy for the sequences.
* Stores symbol compression modes and fse table to fseMetadata.
* Requires ENTROPY_WORKSPACE_SIZE wksp.
* @return : size of fse tables or error code */
static size_t
ZSTD_buildBlockEntropyStats_sequences(
const SeqStore_t* seqStorePtr,
const ZSTD_fseCTables_t* prevEntropy,
ZSTD_fseCTables_t* nextEntropy,
const ZSTD_CCtx_params* cctxParams,
ZSTD_fseCTablesMetadata_t* fseMetadata,
void* workspace, size_t wkspSize)
{
ZSTD_strategy const strategy = cctxParams->cParams.strategy;
size_t const nbSeq = (size_t)(seqStorePtr->sequences - seqStorePtr->sequencesStart);
BYTE* const ostart = fseMetadata->fseTablesBuffer;
BYTE* const oend = ostart + sizeof(fseMetadata->fseTablesBuffer);
BYTE* op = ostart;
unsigned* countWorkspace = (unsigned*)workspace;
unsigned* entropyWorkspace = countWorkspace + (MaxSeq + 1);
size_t entropyWorkspaceSize = wkspSize - (MaxSeq + 1) * sizeof(*countWorkspace);
ZSTD_symbolEncodingTypeStats_t stats;
DEBUGLOG(5, "ZSTD_buildBlockEntropyStats_sequences (nbSeq=%zu)", nbSeq);
stats = nbSeq != 0 ? ZSTD_buildSequencesStatistics(seqStorePtr, nbSeq,
prevEntropy, nextEntropy, op, oend,
strategy, countWorkspace,
entropyWorkspace, entropyWorkspaceSize)
: ZSTD_buildDummySequencesStatistics(nextEntropy);
FORWARD_IF_ERROR(stats.size, "ZSTD_buildSequencesStatistics failed!");
fseMetadata->llType = (SymbolEncodingType_e) stats.LLtype;
fseMetadata->ofType = (SymbolEncodingType_e) stats.Offtype;
fseMetadata->mlType = (SymbolEncodingType_e) stats.MLtype;
fseMetadata->lastCountSize = stats.lastCountSize;
return stats.size;
}
/** ZSTD_buildBlockEntropyStats() :
* Builds entropy for the block.
* Requires workspace size ENTROPY_WORKSPACE_SIZE
* @return : 0 on success, or an error code
* Note : also employed in superblock
*/
size_t ZSTD_buildBlockEntropyStats(
const SeqStore_t* seqStorePtr,
const ZSTD_entropyCTables_t* prevEntropy,
ZSTD_entropyCTables_t* nextEntropy,
const ZSTD_CCtx_params* cctxParams,
ZSTD_entropyCTablesMetadata_t* entropyMetadata,
void* workspace, size_t wkspSize)
{
size_t const litSize = (size_t)(seqStorePtr->lit - seqStorePtr->litStart);
int const huf_useOptDepth = (cctxParams->cParams.strategy >= HUF_OPTIMAL_DEPTH_THRESHOLD);
int const hufFlags = huf_useOptDepth ? HUF_flags_optimalDepth : 0;
entropyMetadata->hufMetadata.hufDesSize =
ZSTD_buildBlockEntropyStats_literals(seqStorePtr->litStart, litSize,
&prevEntropy->huf, &nextEntropy->huf,
&entropyMetadata->hufMetadata,
ZSTD_literalsCompressionIsDisabled(cctxParams),
workspace, wkspSize, hufFlags);
FORWARD_IF_ERROR(entropyMetadata->hufMetadata.hufDesSize, "ZSTD_buildBlockEntropyStats_literals failed");
entropyMetadata->fseMetadata.fseTablesSize =
ZSTD_buildBlockEntropyStats_sequences(seqStorePtr,
&prevEntropy->fse, &nextEntropy->fse,
cctxParams,
&entropyMetadata->fseMetadata,
workspace, wkspSize);
FORWARD_IF_ERROR(entropyMetadata->fseMetadata.fseTablesSize, "ZSTD_buildBlockEntropyStats_sequences failed");
return 0;
}
/* Returns the size estimate for the literals section (header + content) of a block */
static size_t
ZSTD_estimateBlockSize_literal(const BYTE* literals, size_t litSize,
const ZSTD_hufCTables_t* huf,
const ZSTD_hufCTablesMetadata_t* hufMetadata,
void* workspace, size_t wkspSize,
int writeEntropy)
{
unsigned* const countWksp = (unsigned*)workspace;
unsigned maxSymbolValue = HUF_SYMBOLVALUE_MAX;
size_t literalSectionHeaderSize = 3 + (litSize >= 1 KB) + (litSize >= 16 KB);
U32 singleStream = litSize < 256;
if (hufMetadata->hType == set_basic) return litSize;
else if (hufMetadata->hType == set_rle) return 1;
else if (hufMetadata->hType == set_compressed || hufMetadata->hType == set_repeat) {
size_t const largest = HIST_count_wksp (countWksp, &maxSymbolValue, (const BYTE*)literals, litSize, workspace, wkspSize);
if (ZSTD_isError(largest)) return litSize;
{ size_t cLitSizeEstimate = HUF_estimateCompressedSize((const HUF_CElt*)huf->CTable, countWksp, maxSymbolValue);
if (writeEntropy) cLitSizeEstimate += hufMetadata->hufDesSize;
if (!singleStream) cLitSizeEstimate += 6; /* multi-stream huffman uses 6-byte jump table */
return cLitSizeEstimate + literalSectionHeaderSize;
} }
assert(0); /* impossible */
return 0;
}
/* Returns the size estimate for the FSE-compressed symbols (of, ml, ll) of a block */
static size_t
ZSTD_estimateBlockSize_symbolType(SymbolEncodingType_e type,
const BYTE* codeTable, size_t nbSeq, unsigned maxCode,
const FSE_CTable* fseCTable,
const U8* additionalBits,
short const* defaultNorm, U32 defaultNormLog, U32 defaultMax,
void* workspace, size_t wkspSize)
{
unsigned* const countWksp = (unsigned*)workspace;
const BYTE* ctp = codeTable;
const BYTE* const ctStart = ctp;
const BYTE* const ctEnd = ctStart + nbSeq;
size_t cSymbolTypeSizeEstimateInBits = 0;
unsigned max = maxCode;
HIST_countFast_wksp(countWksp, &max, codeTable, nbSeq, workspace, wkspSize); /* can't fail */
if (type == set_basic) {
/* We selected this encoding type, so it must be valid. */
assert(max <= defaultMax);
(void)defaultMax;
cSymbolTypeSizeEstimateInBits = ZSTD_crossEntropyCost(defaultNorm, defaultNormLog, countWksp, max);
} else if (type == set_rle) {
cSymbolTypeSizeEstimateInBits = 0;
} else if (type == set_compressed || type == set_repeat) {
cSymbolTypeSizeEstimateInBits = ZSTD_fseBitCost(fseCTable, countWksp, max);
}
if (ZSTD_isError(cSymbolTypeSizeEstimateInBits)) {
return nbSeq * 10;
}
while (ctp < ctEnd) {
if (additionalBits) cSymbolTypeSizeEstimateInBits += additionalBits[*ctp];
else cSymbolTypeSizeEstimateInBits += *ctp; /* for offset, offset code is also the number of additional bits */
ctp++;
}
return cSymbolTypeSizeEstimateInBits >> 3;
}
/* Returns the size estimate for the sequences section (header + content) of a block */
static size_t
ZSTD_estimateBlockSize_sequences(const BYTE* ofCodeTable,
const BYTE* llCodeTable,
const BYTE* mlCodeTable,
size_t nbSeq,
const ZSTD_fseCTables_t* fseTables,
const ZSTD_fseCTablesMetadata_t* fseMetadata,
void* workspace, size_t wkspSize,
int writeEntropy)
{
size_t sequencesSectionHeaderSize = 1 /* seqHead */ + 1 /* min seqSize size */ + (nbSeq >= 128) + (nbSeq >= LONGNBSEQ);
size_t cSeqSizeEstimate = 0;
cSeqSizeEstimate += ZSTD_estimateBlockSize_symbolType(fseMetadata->ofType, ofCodeTable, nbSeq, MaxOff,
fseTables->offcodeCTable, NULL,
OF_defaultNorm, OF_defaultNormLog, DefaultMaxOff,
workspace, wkspSize);
cSeqSizeEstimate += ZSTD_estimateBlockSize_symbolType(fseMetadata->llType, llCodeTable, nbSeq, MaxLL,
fseTables->litlengthCTable, LL_bits,
LL_defaultNorm, LL_defaultNormLog, MaxLL,
workspace, wkspSize);
cSeqSizeEstimate += ZSTD_estimateBlockSize_symbolType(fseMetadata->mlType, mlCodeTable, nbSeq, MaxML,
fseTables->matchlengthCTable, ML_bits,
ML_defaultNorm, ML_defaultNormLog, MaxML,
workspace, wkspSize);
if (writeEntropy) cSeqSizeEstimate += fseMetadata->fseTablesSize;
return cSeqSizeEstimate + sequencesSectionHeaderSize;
}
/* Returns the size estimate for a given stream of literals, of, ll, ml */
static size_t
ZSTD_estimateBlockSize(const BYTE* literals, size_t litSize,
const BYTE* ofCodeTable,
const BYTE* llCodeTable,
const BYTE* mlCodeTable,
size_t nbSeq,
const ZSTD_entropyCTables_t* entropy,
const ZSTD_entropyCTablesMetadata_t* entropyMetadata,
void* workspace, size_t wkspSize,
int writeLitEntropy, int writeSeqEntropy)
{
size_t const literalsSize = ZSTD_estimateBlockSize_literal(literals, litSize,
&entropy->huf, &entropyMetadata->hufMetadata,
workspace, wkspSize, writeLitEntropy);
size_t const seqSize = ZSTD_estimateBlockSize_sequences(ofCodeTable, llCodeTable, mlCodeTable,
nbSeq, &entropy->fse, &entropyMetadata->fseMetadata,
workspace, wkspSize, writeSeqEntropy);
return seqSize + literalsSize + ZSTD_blockHeaderSize;
}
/* Builds entropy statistics and uses them for blocksize estimation.
*
* @return: estimated compressed size of the seqStore, or a zstd error.
*/
static size_t
ZSTD_buildEntropyStatisticsAndEstimateSubBlockSize(SeqStore_t* seqStore, ZSTD_CCtx* zc)
{
ZSTD_entropyCTablesMetadata_t* const entropyMetadata = &zc->blockSplitCtx.entropyMetadata;
DEBUGLOG(6, "ZSTD_buildEntropyStatisticsAndEstimateSubBlockSize()");
FORWARD_IF_ERROR(ZSTD_buildBlockEntropyStats(seqStore,
&zc->blockState.prevCBlock->entropy,
&zc->blockState.nextCBlock->entropy,
&zc->appliedParams,
entropyMetadata,
zc->tmpWorkspace, zc->tmpWkspSize), "");
return ZSTD_estimateBlockSize(
seqStore->litStart, (size_t)(seqStore->lit - seqStore->litStart),
seqStore->ofCode, seqStore->llCode, seqStore->mlCode,
(size_t)(seqStore->sequences - seqStore->sequencesStart),
&zc->blockState.nextCBlock->entropy,
entropyMetadata,
zc->tmpWorkspace, zc->tmpWkspSize,
(int)(entropyMetadata->hufMetadata.hType == set_compressed), 1);
}
/* Returns literals bytes represented in a seqStore */
static size_t ZSTD_countSeqStoreLiteralsBytes(const SeqStore_t* const seqStore)
{
size_t literalsBytes = 0;
size_t const nbSeqs = (size_t)(seqStore->sequences - seqStore->sequencesStart);
size_t i;
for (i = 0; i < nbSeqs; ++i) {
SeqDef const seq = seqStore->sequencesStart[i];
literalsBytes += seq.litLength;
if (i == seqStore->longLengthPos && seqStore->longLengthType == ZSTD_llt_literalLength) {
literalsBytes += 0x10000;
} }
return literalsBytes;
}
/* Returns match bytes represented in a seqStore */
static size_t ZSTD_countSeqStoreMatchBytes(const SeqStore_t* const seqStore)
{
size_t matchBytes = 0;
size_t const nbSeqs = (size_t)(seqStore->sequences - seqStore->sequencesStart);
size_t i;
for (i = 0; i < nbSeqs; ++i) {
SeqDef seq = seqStore->sequencesStart[i];
matchBytes += seq.mlBase + MINMATCH;
if (i == seqStore->longLengthPos && seqStore->longLengthType == ZSTD_llt_matchLength) {
matchBytes += 0x10000;
} }
return matchBytes;
}
/* Derives the seqStore that is a chunk of the originalSeqStore from [startIdx, endIdx).
* Stores the result in resultSeqStore.
*/
The file goes on; the console shows all of it.