krak/zstd
1
0
Fork 0
mirror of https://github.com/facebook/zstd.git synced 2025-07-16 04:04:28 +02:00
Code Issues Releases 1 Activity

merge conflict

Browse Source
This commit is contained in:
Sen Huang
2019-10-15 12:58:50 -04:00
parent 23dac23a49 8b6d96827c
commit a06b51879c
84 changed files with 2169 additions and 1009 deletions
Download Patch File Download Diff File Expand all files Collapse all files

823
lib/compress/zstd_compress.c
View File

File diff suppressed because it is too large Load Diff

62
lib/compress/zstd_compress_internal.h
View File

@ -19,6 +19,7 @@
* Dependencies
***************************************/
#include "zstd_internal.h"
#include "zstd_cwksp.h"
#ifdef ZSTD_MULTITHREAD
# include "zstdmt_compress.h"
#endif
@ -192,6 +193,13 @@ typedef struct {
size_t capacity; /* The capacity starting from `seq` pointer */
} rawSeqStore_t;
typedef struct {
int collectSequences;
ZSTD_Sequence* seqStart;
size_t seqIndex;
size_t maxSequences;
} SeqCollector;
struct ZSTD_CCtx_params_s {
ZSTD_format_e format;
ZSTD_compressionParameters cParams;
@ -231,9 +239,7 @@ struct ZSTD_CCtx_s {
ZSTD_CCtx_params appliedParams;
U32 dictID;
int workSpaceOversizedDuration;
void* workSpace;
size_t workSpaceSize;
ZSTD_cwksp workspace; /* manages buffer for dynamic allocations */
size_t blockSize;
unsigned long long pledgedSrcSizePlusOne; /* this way, 0 (default) == unknown */
unsigned long long consumedSrcSize;
@ -241,6 +247,7 @@ struct ZSTD_CCtx_s {
XXH64_state_t xxhState;
ZSTD_customMem customMem;
size_t staticSize;
SeqCollector seqCollector;
int isFirstBlock;
seqStore_t seqStore; /* sequences storage ptrs */
@ -341,26 +348,57 @@ MEM_STATIC size_t ZSTD_minGain(size_t srcSize, ZSTD_strategy strat)
return (srcSize >> minlog) + 2;
}
/*! ZSTD_safecopyLiterals() :
* memcpy() function that won't read beyond more than WILDCOPY_OVERLENGTH bytes past ilimit_w.
* Only called when the sequence ends past ilimit_w, so it only needs to be optimized for single
* large copies.
*/
static void ZSTD_safecopyLiterals(BYTE* op, BYTE const* ip, BYTE const* const iend, BYTE const* ilimit_w) {
assert(iend > ilimit_w);
if (ip <= ilimit_w) {
ZSTD_wildcopy(op, ip, ilimit_w - ip, ZSTD_no_overlap);
op += ilimit_w - ip;
ip = ilimit_w;
}
while (ip < iend) *op++ = *ip++;
}
/*! ZSTD_storeSeq() :
* Store a sequence (literal length, literals, offset code and match length code) into seqStore_t.
* `offsetCode` : distance to match + 3 (values 1-3 are repCodes).
* Store a sequence (litlen, litPtr, offCode and mlBase) into seqStore_t.
* `offCode` : distance to match + ZSTD_REP_MOVE (values <= ZSTD_REP_MOVE are repCodes).
* `mlBase` : matchLength - MINMATCH
* Allowed to overread literals up to litLimit.
*/
MEM_STATIC void ZSTD_storeSeq(seqStore_t* seqStorePtr, size_t litLength, const void* literals, U32 offsetCode, size_t mlBase)
HINT_INLINE UNUSED_ATTR
void ZSTD_storeSeq(seqStore_t* seqStorePtr, size_t litLength, const BYTE* literals, const BYTE* litLimit, U32 offCode, size_t mlBase)
{
BYTE const* const litLimit_w = litLimit - WILDCOPY_OVERLENGTH;
BYTE const* const litEnd = literals + litLength;
#if defined(DEBUGLEVEL) && (DEBUGLEVEL >= 6)
static const BYTE* g_start = NULL;
if (g_start==NULL) g_start = (const BYTE*)literals; /* note : index only works for compression within a single segment */
{ U32 const pos = (U32)((const BYTE*)literals - g_start);
DEBUGLOG(6, "Cpos%7u :%3u literals, match%4u bytes at offCode%7u",
pos, (U32)litLength, (U32)mlBase+MINMATCH, (U32)offsetCode);
pos, (U32)litLength, (U32)mlBase+MINMATCH, (U32)offCode);
}
#endif
assert((size_t)(seqStorePtr->sequences - seqStorePtr->sequencesStart) < seqStorePtr->maxNbSeq);
/* copy Literals */
assert(seqStorePtr->maxNbLit <= 128 KB);
assert(seqStorePtr->lit + litLength <= seqStorePtr->litStart + seqStorePtr->maxNbLit);
ZSTD_wildcopy(seqStorePtr->lit, literals, (ptrdiff_t)litLength, ZSTD_no_overlap);
assert(literals + litLength <= litLimit);
if (litEnd <= litLimit_w) {
/* Common case we can use wildcopy.
* First copy 16 bytes, because literals are likely short.
*/
assert(WILDCOPY_OVERLENGTH >= 16);
ZSTD_copy16(seqStorePtr->lit, literals);
if (litLength > 16) {
ZSTD_wildcopy(seqStorePtr->lit+16, literals+16, (ptrdiff_t)litLength-16, ZSTD_no_overlap);
}
} else {
ZSTD_safecopyLiterals(seqStorePtr->lit, literals, litEnd, litLimit_w);
}
seqStorePtr->lit += litLength;
/* literal Length */
@ -372,7 +410,7 @@ MEM_STATIC void ZSTD_storeSeq(seqStore_t* seqStorePtr, size_t litLength, const v
seqStorePtr->sequences[0].litLength = (U16)litLength;
/* match offset */
seqStorePtr->sequences[0].offset = offsetCode + 1;
seqStorePtr->sequences[0].offset = offCode + 1;
/* match Length */
if (mlBase>0xFFFF) {
@ -914,7 +952,7 @@ ZSTD_compressionParameters ZSTD_getCParamsFromCCtxParams(
size_t ZSTD_initCStream_internal(ZSTD_CStream* zcs,
const void* dict, size_t dictSize,
const ZSTD_CDict* cdict,
ZSTD_CCtx_params params, unsigned long long pledgedSrcSize);
const ZSTD_CCtx_params* params, unsigned long long pledgedSrcSize);
void ZSTD_resetSeqStore(seqStore_t* ssPtr);
@ -929,7 +967,7 @@ size_t ZSTD_compressBegin_advanced_internal(ZSTD_CCtx* cctx,
ZSTD_dictContentType_e dictContentType,
ZSTD_dictTableLoadMethod_e dtlm,
const ZSTD_CDict* cdict,
ZSTD_CCtx_params params,
const ZSTD_CCtx_params* params,
unsigned long long pledgedSrcSize);
/* ZSTD_compress_advanced_internal() :
@ -938,7 +976,7 @@ size_t ZSTD_compress_advanced_internal(ZSTD_CCtx* cctx,
void* dst, size_t dstCapacity,
const void* src, size_t srcSize,
const void* dict,size_t dictSize,
ZSTD_CCtx_params params);
const ZSTD_CCtx_params* params);
/* ZSTD_writeLastEmptyBlock() :

15
lib/compress/zstd_compress_literals.c
View File

@ -70,7 +70,7 @@ size_t ZSTD_compressLiterals (ZSTD_hufCTables_t const* prevHuf,
ZSTD_strategy strategy, int disableLiteralCompression,
void* dst, size_t dstCapacity,
const void* src, size_t srcSize,
void* workspace, size_t wkspSize,
void* entropyWorkspace, size_t entropyWorkspaceSize,
const int bmi2)
{
size_t const minGain = ZSTD_minGain(srcSize, strategy);
@ -99,10 +99,15 @@ size_t ZSTD_compressLiterals (ZSTD_hufCTables_t const* prevHuf,
{ HUF_repeat repeat = prevHuf->repeatMode;
int const preferRepeat = strategy < ZSTD_lazy ? srcSize <= 1024 : 0;
if (repeat == HUF_repeat_valid && lhSize == 3) singleStream = 1;
cLitSize = singleStream ? HUF_compress1X_repeat(ostart+lhSize, dstCapacity-lhSize, src, srcSize, 255, 11,
workspace, wkspSize, (HUF_CElt*)nextHuf->CTable, &repeat, preferRepeat, bmi2)
: HUF_compress4X_repeat(ostart+lhSize, dstCapacity-lhSize, src, srcSize, 255, 11,
workspace, wkspSize, (HUF_CElt*)nextHuf->CTable, &repeat, preferRepeat, bmi2);
cLitSize = singleStream ?
HUF_compress1X_repeat(
ostart+lhSize, dstCapacity-lhSize, src, srcSize,
255, 11, entropyWorkspace, entropyWorkspaceSize,
(HUF_CElt*)nextHuf->CTable, &repeat, preferRepeat, bmi2) :
HUF_compress4X_repeat(
ostart+lhSize, dstCapacity-lhSize, src, srcSize,
255, 11, entropyWorkspace, entropyWorkspaceSize,
(HUF_CElt*)nextHuf->CTable, &repeat, preferRepeat, bmi2);
if (repeat != HUF_repeat_none) {
/* reused the existing table */
hType = set_repeat;

2
lib/compress/zstd_compress_literals.h
View File

@ -23,7 +23,7 @@ size_t ZSTD_compressLiterals (ZSTD_hufCTables_t const* prevHuf,
ZSTD_strategy strategy, int disableLiteralCompression,
void* dst, size_t dstCapacity,
const void* src, size_t srcSize,
void* workspace, size_t wkspSize,
void* entropyWorkspace, size_t entropyWorkspaceSize,
const int bmi2);
#endif /* ZSTD_COMPRESS_LITERALS_H */

6
lib/compress/zstd_compress_sequences.c
View File

@ -222,7 +222,7 @@ ZSTD_buildCTable(void* dst, size_t dstCapacity,
const BYTE* codeTable, size_t nbSeq,
const S16* defaultNorm, U32 defaultNormLog, U32 defaultMax,
const FSE_CTable* prevCTable, size_t prevCTableSize,
void* workspace, size_t workspaceSize)
void* entropyWorkspace, size_t entropyWorkspaceSize)
{
BYTE* op = (BYTE*)dst;
const BYTE* const oend = op + dstCapacity;
@ -238,7 +238,7 @@ ZSTD_buildCTable(void* dst, size_t dstCapacity,
memcpy(nextCTable, prevCTable, prevCTableSize);
return 0;
case set_basic:
FORWARD_IF_ERROR(FSE_buildCTable_wksp(nextCTable, defaultNorm, defaultMax, defaultNormLog, workspace, workspaceSize)); /* note : could be pre-calculated */
FORWARD_IF_ERROR(FSE_buildCTable_wksp(nextCTable, defaultNorm, defaultMax, defaultNormLog, entropyWorkspace, entropyWorkspaceSize)); /* note : could be pre-calculated */
return 0;
case set_compressed: {
S16 norm[MaxSeq + 1];
@ -252,7 +252,7 @@ ZSTD_buildCTable(void* dst, size_t dstCapacity,
FORWARD_IF_ERROR(FSE_normalizeCount(norm, tableLog, count, nbSeq_1, max));
{ size_t const NCountSize = FSE_writeNCount(op, oend - op, norm, max, tableLog); /* overflow protected */
FORWARD_IF_ERROR(NCountSize);
FORWARD_IF_ERROR(FSE_buildCTable_wksp(nextCTable, norm, max, tableLog, workspace, workspaceSize));
FORWARD_IF_ERROR(FSE_buildCTable_wksp(nextCTable, norm, max, tableLog, entropyWorkspace, entropyWorkspaceSize));
return NCountSize;
}
}

2
lib/compress/zstd_compress_sequences.h
View File

@ -35,7 +35,7 @@ ZSTD_buildCTable(void* dst, size_t dstCapacity,
const BYTE* codeTable, size_t nbSeq,
const S16* defaultNorm, U32 defaultNormLog, U32 defaultMax,
const FSE_CTable* prevCTable, size_t prevCTableSize,
void* workspace, size_t workspaceSize);
void* entropyWorkspace, size_t entropyWorkspaceSize);
size_t ZSTD_encodeSequences(
void* dst, size_t dstCapacity,

449
lib/compress/zstd_cwksp.h Normal file
View File

@ -0,0 +1,449 @@
/*
* Copyright (c) 2016-present, Yann Collet, Facebook, Inc.
* 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.
*/
#ifndef ZSTD_CWKSP_H
#define ZSTD_CWKSP_H
/*-*************************************
* Dependencies
***************************************/
#include "zstd_internal.h"
#if defined (__cplusplus)
extern "C" {
#endif
/*-*************************************
* Constants
***************************************/
/* define "workspace is too large" as this number of times larger than needed */
#define ZSTD_WORKSPACETOOLARGE_FACTOR 3
/* when workspace is continuously too large
* during at least this number of times,
* context's memory usage is considered wasteful,
* because it's sized to handle a worst case scenario which rarely happens.
* In which case, resize it down to free some memory */
#define ZSTD_WORKSPACETOOLARGE_MAXDURATION 128
/*-*************************************
* Structures
***************************************/
typedef enum {
ZSTD_cwksp_alloc_objects,
ZSTD_cwksp_alloc_buffers,
ZSTD_cwksp_alloc_aligned
} ZSTD_cwksp_alloc_phase_e;
/**
* Zstd fits all its internal datastructures into a single continuous buffer,
* so that it only needs to perform a single OS allocation (or so that a buffer
* can be provided to it and it can perform no allocations at all). This buffer
* is called the workspace.
*
* Several optimizations complicate that process of allocating memory ranges
* from this workspace for each internal datastructure:
*
* - These different internal datastructures have different setup requirements:
*
* - The static objects need to be cleared once and can then be trivially
* reused for each compression.
*
* - Various buffers don't need to be initialized at all--they are always
* written into before they're read.
*
* - The matchstate tables have a unique requirement that they don't need
* their memory to be totally cleared, but they do need the memory to have
* some bound, i.e., a guarantee that all values in the memory they've been
* allocated is less than some maximum value (which is the starting value
* for the indices that they will then use for compression). When this
* guarantee is provided to them, they can use the memory without any setup
* work. When it can't, they have to clear the area.
*
* - These buffers also have different alignment requirements.
*
* - We would like to reuse the objects in the workspace for multiple
* compressions without having to perform any expensive reallocation or
* reinitialization work.
*
* - We would like to be able to efficiently reuse the workspace across
* multiple compressions **even when the compression parameters change** and
* we need to resize some of the objects (where possible).
*
* To attempt to manage this buffer, given these constraints, the ZSTD_cwksp
* abstraction was created. It works as follows:
*
* Workspace Layout:
*
* [ ... workspace ... ]
* [objects][tables ... ->] free space [<- ... aligned][<- ... buffers]
*
* The various objects that live in the workspace are divided into the
* following categories, and are allocated separately:
*
* - Static objects: this is optionally the enclosing ZSTD_CCtx or ZSTD_CDict,
* so that literally everything fits in a single buffer. Note: if present,
* this must be the first object in the workspace, since ZSTD_free{CCtx,
* CDict}() rely on a pointer comparison to see whether one or two frees are
* required.
*
* - Fixed size objects: these are fixed-size, fixed-count objects that are
* nonetheless "dynamically" allocated in the workspace so that we can
* control how they're initialized separately from the broader ZSTD_CCtx.
* Examples:
* - Entropy Workspace
* - 2 x ZSTD_compressedBlockState_t
* - CDict dictionary contents
*
* - Tables: these are any of several different datastructures (hash tables,
* chain tables, binary trees) that all respect a common format: they are
* uint32_t arrays, all of whose values are between 0 and (nextSrc - base).
* Their sizes depend on the cparams.
*
* - Aligned: these buffers are used for various purposes that require 4 byte
* alignment, but don't require any initialization before they're used.
*
* - Buffers: these buffers are used for various purposes that don't require
* any alignment or initialization before they're used. This means they can
* be moved around at no cost for a new compression.
*
* Allocating Memory:
*
* The various types of objects must be allocated in order, so they can be
* correctly packed into the workspace buffer. That order is:
*
* 1. Objects
* 2. Buffers
* 3. Aligned
* 4. Tables
*
* Attempts to reserve objects of different types out of order will fail.
*/
typedef struct {
void* workspace;
void* workspaceEnd;
void* objectEnd;
void* tableEnd;
void* tableValidEnd;
void* allocStart;
int allocFailed;
int workspaceOversizedDuration;
ZSTD_cwksp_alloc_phase_e phase;
} ZSTD_cwksp;
/*-*************************************
* Functions
***************************************/
MEM_STATIC size_t ZSTD_cwksp_available_space(ZSTD_cwksp* ws);
MEM_STATIC void ZSTD_cwksp_assert_internal_consistency(ZSTD_cwksp* ws) {
(void)ws;
assert(ws->workspace <= ws->objectEnd);
assert(ws->objectEnd <= ws->tableEnd);
assert(ws->objectEnd <= ws->tableValidEnd);
assert(ws->tableEnd <= ws->allocStart);
assert(ws->tableValidEnd <= ws->allocStart);
assert(ws->allocStart <= ws->workspaceEnd);
}
/**
* Align must be a power of 2.
*/
MEM_STATIC size_t ZSTD_cwksp_align(size_t size, size_t const align) {
size_t const mask = align - 1;
assert((align & mask) == 0);
return (size + mask) & ~mask;
}
MEM_STATIC void ZSTD_cwksp_internal_advance_phase(
ZSTD_cwksp* ws, ZSTD_cwksp_alloc_phase_e phase) {
assert(phase >= ws->phase);
if (phase > ws->phase) {
if (ws->phase < ZSTD_cwksp_alloc_buffers &&
phase >= ZSTD_cwksp_alloc_buffers) {
ws->tableValidEnd = ws->objectEnd;
}
if (ws->phase < ZSTD_cwksp_alloc_aligned &&
phase >= ZSTD_cwksp_alloc_aligned) {
/* If unaligned allocations down from a too-large top have left us
* unaligned, we need to realign our alloc ptr. Technically, this
* can consume space that is unaccounted for in the neededSpace
* calculation. However, I believe this can only happen when the
* workspace is too large, and specifically when it is too large
* by a larger margin than the space that will be consumed. */
/* TODO: cleaner, compiler warning friendly way to do this??? */
ws->allocStart = (BYTE*)ws->allocStart - ((size_t)ws->allocStart & (sizeof(U32)-1));
if (ws->allocStart < ws->tableValidEnd) {
ws->tableValidEnd = ws->allocStart;
}
}
ws->phase = phase;
}
}
/**
* Internal function. Do not use directly.
*/
MEM_STATIC void* ZSTD_cwksp_reserve_internal(
ZSTD_cwksp* ws, size_t bytes, ZSTD_cwksp_alloc_phase_e phase) {
void* alloc;
void* bottom = ws->tableEnd;
ZSTD_cwksp_internal_advance_phase(ws, phase);
alloc = (BYTE *)ws->allocStart - bytes;
DEBUGLOG(5, "cwksp: reserving %zd bytes, %zd bytes remaining",
bytes, ZSTD_cwksp_available_space(ws) - bytes);
ZSTD_cwksp_assert_internal_consistency(ws);
assert(alloc >= bottom);
if (alloc < bottom) {
DEBUGLOG(4, "cwksp: alloc failed!");
ws->allocFailed = 1;
return NULL;
}
if (alloc < ws->tableValidEnd) {
ws->tableValidEnd = alloc;
}
ws->allocStart = alloc;
return alloc;
}
/**
* Reserves and returns unaligned memory.
*/
MEM_STATIC BYTE* ZSTD_cwksp_reserve_buffer(ZSTD_cwksp* ws, size_t bytes) {
return (BYTE*)ZSTD_cwksp_reserve_internal(ws, bytes, ZSTD_cwksp_alloc_buffers);
}
/**
* Reserves and returns memory sized on and aligned on sizeof(unsigned).
*/
MEM_STATIC void* ZSTD_cwksp_reserve_aligned(ZSTD_cwksp* ws, size_t bytes) {
assert((bytes & (sizeof(U32)-1)) == 0);
return ZSTD_cwksp_reserve_internal(ws, ZSTD_cwksp_align(bytes, sizeof(U32)), ZSTD_cwksp_alloc_aligned);
}
/**
* Aligned on sizeof(unsigned). These buffers have the special property that
* their values remain constrained, allowing us to re-use them without
* memset()-ing them.
*/
MEM_STATIC void* ZSTD_cwksp_reserve_table(ZSTD_cwksp* ws, size_t bytes) {
const ZSTD_cwksp_alloc_phase_e phase = ZSTD_cwksp_alloc_aligned;
void* alloc = ws->tableEnd;
void* end = (BYTE *)alloc + bytes;
void* top = ws->allocStart;
DEBUGLOG(5, "cwksp: reserving table %zd bytes, %zd bytes remaining",
bytes, ZSTD_cwksp_available_space(ws) - bytes);
assert((bytes & (sizeof(U32)-1)) == 0);
ZSTD_cwksp_internal_advance_phase(ws, phase);
ZSTD_cwksp_assert_internal_consistency(ws);
assert(end <= top);
if (end > top) {
DEBUGLOG(4, "cwksp: table alloc failed!");
ws->allocFailed = 1;
return NULL;
}
ws->tableEnd = end;
return alloc;
}
/**
* Aligned on sizeof(void*).
*/
MEM_STATIC void* ZSTD_cwksp_reserve_object(ZSTD_cwksp* ws, size_t bytes) {
size_t roundedBytes = ZSTD_cwksp_align(bytes, sizeof(void*));
void* start = ws->objectEnd;
void* end = (BYTE*)start + roundedBytes;
DEBUGLOG(5,
"cwksp: reserving object %zd bytes (rounded to %zd), %zd bytes remaining",
bytes, roundedBytes, ZSTD_cwksp_available_space(ws) - roundedBytes);
assert(((size_t)start & (sizeof(void*)-1)) == 0);
assert((bytes & (sizeof(void*)-1)) == 0);
ZSTD_cwksp_assert_internal_consistency(ws);
/* we must be in the first phase, no advance is possible */
if (ws->phase != ZSTD_cwksp_alloc_objects || end > ws->workspaceEnd) {
DEBUGLOG(4, "cwksp: object alloc failed!");
ws->allocFailed = 1;
return NULL;
}
ws->objectEnd = end;
ws->tableEnd = end;
ws->tableValidEnd = end;
return start;
}
MEM_STATIC void ZSTD_cwksp_mark_tables_dirty(ZSTD_cwksp* ws) {
DEBUGLOG(4, "cwksp: ZSTD_cwksp_mark_tables_dirty");
#if defined (MEMORY_SANITIZER) && !defined (ZSTD_MSAN_DONT_POISON_WORKSPACE)
/* To validate that the table re-use logic is sound, and that we don't
* access table space that we haven't cleaned, we re-"poison" the table
* space every time we mark it dirty. */
{
size_t size = (BYTE*)ws->tableValidEnd - (BYTE*)ws->objectEnd;
assert(__msan_test_shadow(ws->objectEnd, size) == -1);
__msan_poison(ws->objectEnd, size);
}
#endif
assert(ws->tableValidEnd >= ws->objectEnd);
assert(ws->tableValidEnd <= ws->allocStart);
ws->tableValidEnd = ws->objectEnd;
ZSTD_cwksp_assert_internal_consistency(ws);
}
MEM_STATIC void ZSTD_cwksp_mark_tables_clean(ZSTD_cwksp* ws) {
DEBUGLOG(4, "cwksp: ZSTD_cwksp_mark_tables_clean");
assert(ws->tableValidEnd >= ws->objectEnd);
assert(ws->tableValidEnd <= ws->allocStart);
if (ws->tableValidEnd < ws->tableEnd) {
ws->tableValidEnd = ws->tableEnd;
}
ZSTD_cwksp_assert_internal_consistency(ws);
}
/**
* Zero the part of the allocated tables not already marked clean.
*/
MEM_STATIC void ZSTD_cwksp_clean_tables(ZSTD_cwksp* ws) {
DEBUGLOG(4, "cwksp: ZSTD_cwksp_clean_tables");
assert(ws->tableValidEnd >= ws->objectEnd);
assert(ws->tableValidEnd <= ws->allocStart);
if (ws->tableValidEnd < ws->tableEnd) {
memset(ws->tableValidEnd, 0, (BYTE*)ws->tableEnd - (BYTE*)ws->tableValidEnd);
}
ZSTD_cwksp_mark_tables_clean(ws);
}
/**
* Invalidates table allocations.
* All other allocations remain valid.
*/
MEM_STATIC void ZSTD_cwksp_clear_tables(ZSTD_cwksp* ws) {
DEBUGLOG(4, "cwksp: clearing tables!");
ws->tableEnd = ws->objectEnd;
ZSTD_cwksp_assert_internal_consistency(ws);
}
/**
* Invalidates all buffer, aligned, and table allocations.
* Object allocations remain valid.
*/
MEM_STATIC void ZSTD_cwksp_clear(ZSTD_cwksp* ws) {
DEBUGLOG(4, "cwksp: clearing!");
#if defined (MEMORY_SANITIZER) && !defined (ZSTD_MSAN_DONT_POISON_WORKSPACE)
/* To validate that the context re-use logic is sound, and that we don't
* access stuff that this compression hasn't initialized, we re-"poison"
* the workspace (or at least the non-static, non-table parts of it)
* every time we start a new compression. */
{
size_t size = (BYTE*)ws->workspaceEnd - (BYTE*)ws->tableValidEnd;
__msan_poison(ws->tableValidEnd, size);
}
#endif
ws->tableEnd = ws->objectEnd;
ws->allocStart = ws->workspaceEnd;
ws->allocFailed = 0;
if (ws->phase > ZSTD_cwksp_alloc_buffers) {
ws->phase = ZSTD_cwksp_alloc_buffers;
}
ZSTD_cwksp_assert_internal_consistency(ws);
}
/**
* The provided workspace takes ownership of the buffer [start, start+size).
* Any existing values in the workspace are ignored (the previously managed
* buffer, if present, must be separately freed).
*/
MEM_STATIC void ZSTD_cwksp_init(ZSTD_cwksp* ws, void* start, size_t size) {
DEBUGLOG(4, "cwksp: init'ing workspace with %zd bytes", size);
assert(((size_t)start & (sizeof(void*)-1)) == 0); /* ensure correct alignment */
ws->workspace = start;
ws->workspaceEnd = (BYTE*)start + size;
ws->objectEnd = ws->workspace;
ws->tableValidEnd = ws->objectEnd;
ws->phase = ZSTD_cwksp_alloc_objects;
ZSTD_cwksp_clear(ws);
ws->workspaceOversizedDuration = 0;
ZSTD_cwksp_assert_internal_consistency(ws);
}
MEM_STATIC size_t ZSTD_cwksp_create(ZSTD_cwksp* ws, size_t size, ZSTD_customMem customMem) {
void* workspace = ZSTD_malloc(size, customMem);
DEBUGLOG(4, "cwksp: creating new workspace with %zd bytes", size);
RETURN_ERROR_IF(workspace == NULL, memory_allocation);
ZSTD_cwksp_init(ws, workspace, size);
return 0;
}
MEM_STATIC void ZSTD_cwksp_free(ZSTD_cwksp* ws, ZSTD_customMem customMem) {
DEBUGLOG(4, "cwksp: freeing workspace");
ZSTD_free(ws->workspace, customMem);
memset(ws, 0, sizeof(ZSTD_cwksp));
}
/**
* Moves the management of a workspace from one cwksp to another. The src cwksp
* is left in an invalid state (src must be re-init()'ed before its used again).
*/
MEM_STATIC void ZSTD_cwksp_move(ZSTD_cwksp* dst, ZSTD_cwksp* src) {
*dst = *src;
memset(src, 0, sizeof(ZSTD_cwksp));
}
MEM_STATIC size_t ZSTD_cwksp_sizeof(const ZSTD_cwksp* ws) {
return (BYTE*)ws->workspaceEnd - (BYTE*)ws->workspace;
}
MEM_STATIC int ZSTD_cwksp_reserve_failed(const ZSTD_cwksp* ws) {
return ws->allocFailed;
}
/*-*************************************
* Functions Checking Free Space
***************************************/
MEM_STATIC size_t ZSTD_cwksp_available_space(ZSTD_cwksp* ws) {
return (size_t)((BYTE*)ws->allocStart - (BYTE*)ws->tableEnd);
}
MEM_STATIC int ZSTD_cwksp_check_available(ZSTD_cwksp* ws, size_t additionalNeededSpace) {
return ZSTD_cwksp_available_space(ws) >= additionalNeededSpace;
}
MEM_STATIC int ZSTD_cwksp_check_too_large(ZSTD_cwksp* ws, size_t additionalNeededSpace) {
return ZSTD_cwksp_check_available(
ws, additionalNeededSpace * ZSTD_WORKSPACETOOLARGE_FACTOR);
}
MEM_STATIC int ZSTD_cwksp_check_wasteful(ZSTD_cwksp* ws, size_t additionalNeededSpace) {
return ZSTD_cwksp_check_too_large(ws, additionalNeededSpace)
&& ws->workspaceOversizedDuration > ZSTD_WORKSPACETOOLARGE_MAXDURATION;
}
MEM_STATIC void ZSTD_cwksp_bump_oversized_duration(
ZSTD_cwksp* ws, size_t additionalNeededSpace) {
if (ZSTD_cwksp_check_too_large(ws, additionalNeededSpace)) {
ws->workspaceOversizedDuration++;
} else {
ws->workspaceOversizedDuration = 0;
}
}
#if defined (__cplusplus)
}
#endif
#endif /* ZSTD_CWKSP_H */

18
lib/compress/zstd_double_fast.c
View File

@ -148,7 +148,7 @@ size_t ZSTD_compressBlock_doubleFast_generic(
const BYTE* repMatchEnd = repIndex < prefixLowestIndex ? dictEnd : iend;
mLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixLowest) + 4;
ip++;
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, 0, mLength-MINMATCH);
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, mLength-MINMATCH);
goto _match_stored;
}
@ -157,7 +157,7 @@ size_t ZSTD_compressBlock_doubleFast_generic(
&& ((offset_1 > 0) & (MEM_read32(ip+1-offset_1) == MEM_read32(ip+1)))) {
mLength = ZSTD_count(ip+1+4, ip+1+4-offset_1, iend) + 4;
ip++;
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, 0, mLength-MINMATCH);
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, mLength-MINMATCH);
goto _match_stored;
}
@ -247,7 +247,7 @@ _match_found:
offset_2 = offset_1;
offset_1 = offset;
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
_match_stored:
/* match found */
@ -278,7 +278,7 @@ _match_stored:
const BYTE* const repEnd2 = repIndex2 < prefixLowestIndex ? dictEnd : iend;
size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixLowest) + 4;
U32 tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset; /* swap offset_2 <=> offset_1 */
ZSTD_storeSeq(seqStore, 0, anchor, 0, repLength2-MINMATCH);
ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, repLength2-MINMATCH);
hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = current2;
hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = current2;
ip += repLength2;
@ -297,7 +297,7 @@ _match_stored:
U32 const tmpOff = offset_2; offset_2 = offset_1; offset_1 = tmpOff; /* swap offset_2 <=> offset_1 */
hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = (U32)(ip-base);
hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = (U32)(ip-base);
ZSTD_storeSeq(seqStore, 0, anchor, 0, rLength-MINMATCH);
ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, rLength-MINMATCH);
ip += rLength;
anchor = ip;
continue; /* faster when present ... (?) */
@ -411,7 +411,7 @@ static size_t ZSTD_compressBlock_doubleFast_extDict_generic(
const BYTE* repMatchEnd = repIndex < prefixStartIndex ? dictEnd : iend;
mLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixStart) + 4;
ip++;
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, 0, mLength-MINMATCH);
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, mLength-MINMATCH);
} else {
if ((matchLongIndex > dictStartIndex) && (MEM_read64(matchLong) == MEM_read64(ip))) {
const BYTE* const matchEnd = matchLongIndex < prefixStartIndex ? dictEnd : iend;
@ -422,7 +422,7 @@ static size_t ZSTD_compressBlock_doubleFast_extDict_generic(
while (((ip>anchor) & (matchLong>lowMatchPtr)) && (ip[-1] == matchLong[-1])) { ip--; matchLong--; mLength++; } /* catch up */
offset_2 = offset_1;
offset_1 = offset;
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
} else if ((matchIndex > dictStartIndex) && (MEM_read32(match) == MEM_read32(ip))) {
size_t const h3 = ZSTD_hashPtr(ip+1, hBitsL, 8);
@ -447,7 +447,7 @@ static size_t ZSTD_compressBlock_doubleFast_extDict_generic(
}
offset_2 = offset_1;
offset_1 = offset;
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
} else {
ip += ((ip-anchor) >> kSearchStrength) + 1;
@ -479,7 +479,7 @@ static size_t ZSTD_compressBlock_doubleFast_extDict_generic(
const BYTE* const repEnd2 = repIndex2 < prefixStartIndex ? dictEnd : iend;
size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixStart) + 4;
U32 const tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset; /* swap offset_2 <=> offset_1 */
ZSTD_storeSeq(seqStore, 0, anchor, 0, repLength2-MINMATCH);
ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, repLength2-MINMATCH);
hashSmall[ZSTD_hashPtr(ip, hBitsS, mls)] = current2;
hashLong[ZSTD_hashPtr(ip, hBitsL, 8)] = current2;
ip += repLength2;

69
lib/compress/zstd_fast.c
View File

@ -8,7 +8,7 @@
* You may select, at your option, one of the above-listed licenses.
*/
#include "zstd_compress_internal.h"
#include "zstd_compress_internal.h" /* ZSTD_hashPtr, ZSTD_count, ZSTD_storeSeq */
#include "zstd_fast.h"
@ -43,8 +43,8 @@ void ZSTD_fillHashTable(ZSTD_matchState_t* ms,
}
FORCE_INLINE_TEMPLATE
size_t ZSTD_compressBlock_fast_generic(
FORCE_INLINE_TEMPLATE size_t
ZSTD_compressBlock_fast_generic(
ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
void const* src, size_t srcSize,
U32 const mls)
@ -74,8 +74,7 @@ size_t ZSTD_compressBlock_fast_generic(
DEBUGLOG(5, "ZSTD_compressBlock_fast_generic");
ip0 += (ip0 == prefixStart);
ip1 = ip0 + 1;
{
U32 const maxRep = (U32)(ip0 - prefixStart);
{ U32 const maxRep = (U32)(ip0 - prefixStart);
if (offset_2 > maxRep) offsetSaved = offset_2, offset_2 = 0;
if (offset_1 > maxRep) offsetSaved = offset_1, offset_1 = 0;
}
@ -118,8 +117,7 @@ size_t ZSTD_compressBlock_fast_generic(
match0 = match1;
goto _offset;
}
{
size_t const step = ((ip0-anchor) >> (kSearchStrength - 1)) + stepSize;
{ size_t const step = ((size_t)(ip0-anchor) >> (kSearchStrength - 1)) + stepSize;
assert(step >= 2);
ip0 += step;
ip1 += step;
@ -138,7 +136,7 @@ _offset: /* Requires: ip0, match0 */
_match: /* Requires: ip0, match0, offcode */
/* Count the forward length */
mLength += ZSTD_count(ip0+mLength+4, match0+mLength+4, iend) + 4;
ZSTD_storeSeq(seqStore, ip0-anchor, anchor, offcode, mLength-MINMATCH);
ZSTD_storeSeq(seqStore, (size_t)(ip0-anchor), anchor, iend, offcode, mLength-MINMATCH);
/* match found */
ip0 += mLength;
anchor = ip0;
@ -150,16 +148,15 @@ _match: /* Requires: ip0, match0, offcode */
hashTable[ZSTD_hashPtr(base+current0+2, hlog, mls)] = current0+2; /* here because current+2 could be > iend-8 */
hashTable[ZSTD_hashPtr(ip0-2, hlog, mls)] = (U32)(ip0-2-base);
while ( (ip0 <= ilimit)
&& ( (offset_2>0)
& (MEM_read32(ip0) == MEM_read32(ip0 - offset_2)) )) {
while ( ((ip0 <= ilimit) & (offset_2>0)) /* offset_2==0 means offset_2 is invalidated */
&& (MEM_read32(ip0) == MEM_read32(ip0 - offset_2)) ) {
/* store sequence */
size_t const rLength = ZSTD_count(ip0+4, ip0+4-offset_2, iend) + 4;
U32 const tmpOff = offset_2; offset_2 = offset_1; offset_1 = tmpOff; /* swap offset_2 <=> offset_1 */
{ U32 const tmpOff = offset_2; offset_2 = offset_1; offset_1 = tmpOff; } /* swap offset_2 <=> offset_1 */
hashTable[ZSTD_hashPtr(ip0, hlog, mls)] = (U32)(ip0-base);
ip0 += rLength;
ip1 = ip0 + 1;
ZSTD_storeSeq(seqStore, 0, anchor, 0, rLength-MINMATCH);
ZSTD_storeSeq(seqStore, 0 /*litLen*/, anchor, iend, 0 /*offCode*/, rLength-MINMATCH);
anchor = ip0;
continue; /* faster when present (confirmed on gcc-8) ... (?) */
}
@ -179,8 +176,7 @@ size_t ZSTD_compressBlock_fast(
ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
void const* src, size_t srcSize)
{
ZSTD_compressionParameters const* cParams = &ms->cParams;
U32 const mls = cParams->minMatch;
U32 const mls = ms->cParams.minMatch;
assert(ms->dictMatchState == NULL);
switch(mls)
{
@ -265,7 +261,7 @@ size_t ZSTD_compressBlock_fast_dictMatchState_generic(
const BYTE* const repMatchEnd = repIndex < prefixStartIndex ? dictEnd : iend;
mLength = ZSTD_count_2segments(ip+1+4, repMatch+4, iend, repMatchEnd, prefixStart) + 4;
ip++;
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, 0, mLength-MINMATCH);
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, mLength-MINMATCH);
} else if ( (matchIndex <= prefixStartIndex) ) {
size_t const dictHash = ZSTD_hashPtr(ip, dictHLog, mls);
U32 const dictMatchIndex = dictHashTable[dictHash];
@ -285,7 +281,7 @@ size_t ZSTD_compressBlock_fast_dictMatchState_generic(
} /* catch up */
offset_2 = offset_1;
offset_1 = offset;
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
}
} else if (MEM_read32(match) != MEM_read32(ip)) {
/* it's not a match, and we're not going to check the dictionary */
@ -300,7 +296,7 @@ size_t ZSTD_compressBlock_fast_dictMatchState_generic(
&& (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */
offset_2 = offset_1;
offset_1 = offset;
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
}
/* match found */
@ -325,7 +321,7 @@ size_t ZSTD_compressBlock_fast_dictMatchState_generic(
const BYTE* const repEnd2 = repIndex2 < prefixStartIndex ? dictEnd : iend;
size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixStart) + 4;
U32 tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset; /* swap offset_2 <=> offset_1 */
ZSTD_storeSeq(seqStore, 0, anchor, 0, repLength2-MINMATCH);
ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, repLength2-MINMATCH);
hashTable[ZSTD_hashPtr(ip, hlog, mls)] = current2;
ip += repLength2;
anchor = ip;
@ -348,8 +344,7 @@ size_t ZSTD_compressBlock_fast_dictMatchState(
ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
void const* src, size_t srcSize)
{
ZSTD_compressionParameters const* cParams = &ms->cParams;
U32 const mls = cParams->minMatch;
U32 const mls = ms->cParams.minMatch;
assert(ms->dictMatchState != NULL);
switch(mls)
{
@ -408,16 +403,17 @@ static size_t ZSTD_compressBlock_fast_extDict_generic(
const U32 repIndex = current + 1 - offset_1;
const BYTE* const repBase = repIndex < prefixStartIndex ? dictBase : base;
const BYTE* const repMatch = repBase + repIndex;
size_t mLength;
hashTable[h] = current; /* update hash table */
assert(offset_1 <= current +1); /* check repIndex */
if ( (((U32)((prefixStartIndex-1) - repIndex) >= 3) /* intentional underflow */ & (repIndex > dictStartIndex))
&& (MEM_read32(repMatch) == MEM_read32(ip+1)) ) {
const BYTE* const repMatchEnd = repIndex < prefixStartIndex ? dictEnd : iend;
mLength = ZSTD_count_2segments(ip+1 +4, repMatch +4, iend, repMatchEnd, prefixStart) + 4;
size_t const rLength = ZSTD_count_2segments(ip+1 +4, repMatch +4, iend, repMatchEnd, prefixStart) + 4;
ip++;
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, 0, mLength-MINMATCH);
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, 0, rLength-MINMATCH);
ip += rLength;
anchor = ip;
} else {
if ( (matchIndex < dictStartIndex) ||
(MEM_read32(match) != MEM_read32(ip)) ) {
@ -427,19 +423,15 @@ static size_t ZSTD_compressBlock_fast_extDict_generic(
}
{ const BYTE* const matchEnd = matchIndex < prefixStartIndex ? dictEnd : iend;
const BYTE* const lowMatchPtr = matchIndex < prefixStartIndex ? dictStart : prefixStart;
U32 offset;
mLength = ZSTD_count_2segments(ip+4, match+4, iend, matchEnd, prefixStart) + 4;
U32 const offset = current - matchIndex;
size_t mLength = ZSTD_count_2segments(ip+4, match+4, iend, matchEnd, prefixStart) + 4;
while (((ip>anchor) & (match>lowMatchPtr)) && (ip[-1] == match[-1])) { ip--; match--; mLength++; } /* catch up */
offset = current - matchIndex;
offset_2 = offset_1;
offset_1 = offset;
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
offset_2 = offset_1; offset_1 = offset; /* update offset history */
ZSTD_storeSeq(seqStore, (size_t)(ip-anchor), anchor, iend, offset + ZSTD_REP_MOVE, mLength-MINMATCH);
ip += mLength;
anchor = ip;
} }
/* found a match : store it */
ip += mLength;
anchor = ip;
if (ip <= ilimit) {
/* Fill Table */
hashTable[ZSTD_hashPtr(base+current+2, hlog, mls)] = current+2;
@ -448,13 +440,13 @@ static size_t ZSTD_compressBlock_fast_extDict_generic(
while (ip <= ilimit) {
U32 const current2 = (U32)(ip-base);
U32 const repIndex2 = current2 - offset_2;
const BYTE* repMatch2 = repIndex2 < prefixStartIndex ? dictBase + repIndex2 : base + repIndex2;
const BYTE* const repMatch2 = repIndex2 < prefixStartIndex ? dictBase + repIndex2 : base + repIndex2;
if ( (((U32)((prefixStartIndex-1) - repIndex2) >= 3) & (repIndex2 > dictStartIndex)) /* intentional overflow */
&& (MEM_read32(repMatch2) == MEM_read32(ip)) ) {
const BYTE* const repEnd2 = repIndex2 < prefixStartIndex ? dictEnd : iend;
size_t const repLength2 = ZSTD_count_2segments(ip+4, repMatch2+4, iend, repEnd2, prefixStart) + 4;
U32 const tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset; /* swap offset_2 <=> offset_1 */
ZSTD_storeSeq(seqStore, 0, anchor, 0, repLength2-MINMATCH);
{ U32 const tmpOffset = offset_2; offset_2 = offset_1; offset_1 = tmpOffset; } /* swap offset_2 <=> offset_1 */
ZSTD_storeSeq(seqStore, 0 /*litlen*/, anchor, iend, 0 /*offcode*/, repLength2-MINMATCH);
hashTable[ZSTD_hashPtr(ip, hlog, mls)] = current2;
ip += repLength2;
anchor = ip;
@ -476,8 +468,7 @@ size_t ZSTD_compressBlock_fast_extDict(
ZSTD_matchState_t* ms, seqStore_t* seqStore, U32 rep[ZSTD_REP_NUM],
void const* src, size_t srcSize)
{
ZSTD_compressionParameters const* cParams = &ms->cParams;
U32 const mls = cParams->minMatch;
U32 const mls = ms->cParams.minMatch;
switch(mls)
{
default: /* includes case 3 */

10
lib/compress/zstd_lazy.c
View File

@ -810,7 +810,7 @@ ZSTD_compressBlock_lazy_generic(
/* store sequence */
_storeSequence:
{ size_t const litLength = start - anchor;
ZSTD_storeSeq(seqStore, litLength, anchor, (U32)offset, matchLength-MINMATCH);
ZSTD_storeSeq(seqStore, litLength, anchor, iend, (U32)offset, matchLength-MINMATCH);
anchor = ip = start + matchLength;
}
@ -828,7 +828,7 @@ _storeSequence:
const BYTE* const repEnd2 = repIndex < prefixLowestIndex ? dictEnd : iend;
matchLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd2, prefixLowest) + 4;
offset = offset_2; offset_2 = offset_1; offset_1 = (U32)offset; /* swap offset_2 <=> offset_1 */
ZSTD_storeSeq(seqStore, 0, anchor, 0, matchLength-MINMATCH);
ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, matchLength-MINMATCH);
ip += matchLength;
anchor = ip;
continue;
@ -843,7 +843,7 @@ _storeSequence:
/* store sequence */
matchLength = ZSTD_count(ip+4, ip+4-offset_2, iend) + 4;
offset = offset_2; offset_2 = offset_1; offset_1 = (U32)offset; /* swap repcodes */
ZSTD_storeSeq(seqStore, 0, anchor, 0, matchLength-MINMATCH);
ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, matchLength-MINMATCH);
ip += matchLength;
anchor = ip;
continue; /* faster when present ... (?) */
@ -1051,7 +1051,7 @@ size_t ZSTD_compressBlock_lazy_extDict_generic(
/* store sequence */
_storeSequence:
{ size_t const litLength = start - anchor;
ZSTD_storeSeq(seqStore, litLength, anchor, (U32)offset, matchLength-MINMATCH);
ZSTD_storeSeq(seqStore, litLength, anchor, iend, (U32)offset, matchLength-MINMATCH);
anchor = ip = start + matchLength;
}
@ -1066,7 +1066,7 @@ _storeSequence:
const BYTE* const repEnd = repIndex < dictLimit ? dictEnd : iend;
matchLength = ZSTD_count_2segments(ip+4, repMatch+4, iend, repEnd, prefixStart) + 4;
offset = offset_2; offset_2 = offset_1; offset_1 = (U32)offset; /* swap offset history */
ZSTD_storeSeq(seqStore, 0, anchor, 0, matchLength-MINMATCH);
ZSTD_storeSeq(seqStore, 0, anchor, iend, 0, matchLength-MINMATCH);
ip += matchLength;
anchor = ip;
continue; /* faster when present ... (?) */

2
lib/compress/zstd_ldm.c
View File

@ -583,7 +583,7 @@ size_t ZSTD_ldm_blockCompress(rawSeqStore_t* rawSeqStore,
rep[i] = rep[i-1];
rep[0] = sequence.offset;
/* Store the sequence */
ZSTD_storeSeq(seqStore, newLitLength, ip - newLitLength,
ZSTD_storeSeq(seqStore, newLitLength, ip - newLitLength, iend,
sequence.offset + ZSTD_REP_MOVE,
sequence.matchLength - MINMATCH);
ip += sequence.matchLength;

2
lib/compress/zstd_opt.c
View File

@ -1098,7 +1098,7 @@ _shortestPath: /* cur, last_pos, best_mlen, best_off have to be set */
assert(anchor + llen <= iend);
ZSTD_updateStats(optStatePtr, llen, anchor, offCode, mlen);
ZSTD_storeSeq(seqStore, llen, anchor, offCode, mlen-MINMATCH);
ZSTD_storeSeq(seqStore, llen, anchor, iend, offCode, mlen-MINMATCH);
anchor += advance;
ip = anchor;
} }

46
lib/compress/zstdmt_compress.c
View File

@ -668,7 +668,7 @@ static void ZSTDMT_compressionJob(void* jobDescription)
/* init */
if (job->cdict) {
size_t const initError = ZSTD_compressBegin_advanced_internal(cctx, NULL, 0, ZSTD_dct_auto, ZSTD_dtlm_fast, job->cdict, jobParams, job->fullFrameSize);
size_t const initError = ZSTD_compressBegin_advanced_internal(cctx, NULL, 0, ZSTD_dct_auto, ZSTD_dtlm_fast, job->cdict, &jobParams, job->fullFrameSize);
assert(job->firstJob); /* only allowed for first job */
if (ZSTD_isError(initError)) JOB_ERROR(initError);
} else { /* srcStart points at reloaded section */
@ -680,7 +680,7 @@ static void ZSTDMT_compressionJob(void* jobDescription)
job->prefix.start, job->prefix.size, ZSTD_dct_rawContent, /* load dictionary in "content-only" mode (no header analysis) */
ZSTD_dtlm_fast,
NULL, /*cdict*/
jobParams, pledgedSrcSize);
&jobParams, pledgedSrcSize);
if (ZSTD_isError(initError)) JOB_ERROR(initError);
} }
@ -1028,9 +1028,9 @@ size_t ZSTDMT_getMTCtxParameter(ZSTDMT_CCtx* mtctx, ZSTDMT_parameter parameter,
/* Sets parameters relevant to the compression job,
* initializing others to default values. */
static ZSTD_CCtx_params ZSTDMT_initJobCCtxParams(ZSTD_CCtx_params const params)
static ZSTD_CCtx_params ZSTDMT_initJobCCtxParams(const ZSTD_CCtx_params* params)
{
ZSTD_CCtx_params jobParams = params;
ZSTD_CCtx_params jobParams = *params;
/* Clear parameters related to multithreading */
jobParams.forceWindow = 0;
jobParams.nbWorkers = 0;
@ -1151,16 +1151,16 @@ size_t ZSTDMT_toFlushNow(ZSTDMT_CCtx* mtctx)
/* ===== Multi-threaded compression ===== */
/* ------------------------------------------ */
static unsigned ZSTDMT_computeTargetJobLog(ZSTD_CCtx_params const params)
static unsigned ZSTDMT_computeTargetJobLog(const ZSTD_CCtx_params* params)
{
unsigned jobLog;
if (params.ldmParams.enableLdm) {
if (params->ldmParams.enableLdm) {
/* In Long Range Mode, the windowLog is typically oversized.
* In which case, it's preferable to determine the jobSize
* based on chainLog instead. */
jobLog = MAX(21, params.cParams.chainLog + 4);
jobLog = MAX(21, params->cParams.chainLog + 4);
} else {
jobLog = MAX(20, params.cParams.windowLog + 2);
jobLog = MAX(20, params->cParams.windowLog + 2);
}
return MIN(jobLog, (unsigned)ZSTDMT_JOBLOG_MAX);
}
@ -1193,27 +1193,27 @@ static int ZSTDMT_overlapLog(int ovlog, ZSTD_strategy strat)
return ovlog;
}
static size_t ZSTDMT_computeOverlapSize(ZSTD_CCtx_params const params)
static size_t ZSTDMT_computeOverlapSize(const ZSTD_CCtx_params* params)
{
int const overlapRLog = 9 - ZSTDMT_overlapLog(params.overlapLog, params.cParams.strategy);
int ovLog = (overlapRLog >= 8) ? 0 : (params.cParams.windowLog - overlapRLog);
int const overlapRLog = 9 - ZSTDMT_overlapLog(params->overlapLog, params->cParams.strategy);
int ovLog = (overlapRLog >= 8) ? 0 : (params->cParams.windowLog - overlapRLog);
assert(0 <= overlapRLog && overlapRLog <= 8);
if (params.ldmParams.enableLdm) {
if (params->ldmParams.enableLdm) {
/* In Long Range Mode, the windowLog is typically oversized.
* In which case, it's preferable to determine the jobSize
* based on chainLog instead.
* Then, ovLog becomes a fraction of the jobSize, rather than windowSize */
ovLog = MIN(params.cParams.windowLog, ZSTDMT_computeTargetJobLog(params) - 2)
ovLog = MIN(params->cParams.windowLog, ZSTDMT_computeTargetJobLog(params) - 2)
- overlapRLog;
}
assert(0 <= ovLog && ovLog <= ZSTD_WINDOWLOG_MAX);
DEBUGLOG(4, "overlapLog : %i", params.overlapLog);
DEBUGLOG(4, "overlapLog : %i", params->overlapLog);
DEBUGLOG(4, "overlap size : %i", 1 << ovLog);
return (ovLog==0) ? 0 : (size_t)1 << ovLog;
}
static unsigned
ZSTDMT_computeNbJobs(ZSTD_CCtx_params params, size_t srcSize, unsigned nbWorkers)
ZSTDMT_computeNbJobs(const ZSTD_CCtx_params* params, size_t srcSize, unsigned nbWorkers)
{
assert(nbWorkers>0);
{ size_t const jobSizeTarget = (size_t)1 << ZSTDMT_computeTargetJobLog(params);
@ -1236,9 +1236,9 @@ static size_t ZSTDMT_compress_advanced_internal(
const ZSTD_CDict* cdict,
ZSTD_CCtx_params params)
{
ZSTD_CCtx_params const jobParams = ZSTDMT_initJobCCtxParams(params);
size_t const overlapSize = ZSTDMT_computeOverlapSize(params);
unsigned const nbJobs = ZSTDMT_computeNbJobs(params, srcSize, params.nbWorkers);
ZSTD_CCtx_params const jobParams = ZSTDMT_initJobCCtxParams(&params);
size_t const overlapSize = ZSTDMT_computeOverlapSize(&params);
unsigned const nbJobs = ZSTDMT_computeNbJobs(&params, srcSize, params.nbWorkers);
size_t const proposedJobSize = (srcSize + (nbJobs-1)) / nbJobs;
size_t const avgJobSize = (((proposedJobSize-1) & 0x1FFFF) < 0x7FFF) ? proposedJobSize + 0xFFFF : proposedJobSize; /* avoid too small last block */
const char* const srcStart = (const char*)src;
@ -1256,7 +1256,7 @@ static size_t ZSTDMT_compress_advanced_internal(
ZSTD_CCtx* const cctx = mtctx->cctxPool->cctx[0];
DEBUGLOG(4, "ZSTDMT_compress_advanced_internal: fallback to single-thread mode");
if (cdict) return ZSTD_compress_usingCDict_advanced(cctx, dst, dstCapacity, src, srcSize, cdict, jobParams.fParams);
return ZSTD_compress_advanced_internal(cctx, dst, dstCapacity, src, srcSize, NULL, 0, jobParams);
return ZSTD_compress_advanced_internal(cctx, dst, dstCapacity, src, srcSize, NULL, 0, &jobParams);
}
assert(avgJobSize >= 256 KB); /* condition for ZSTD_compressBound(A) + ZSTD_compressBound(B) <= ZSTD_compressBound(A+B), required to compress directly into Dst (no additional buffer) */
@ -1404,12 +1404,12 @@ size_t ZSTDMT_initCStream_internal(
mtctx->singleBlockingThread = (pledgedSrcSize <= ZSTDMT_JOBSIZE_MIN); /* do not trigger multi-threading when srcSize is too small */
if (mtctx->singleBlockingThread) {
ZSTD_CCtx_params const singleThreadParams = ZSTDMT_initJobCCtxParams(params);
ZSTD_CCtx_params const singleThreadParams = ZSTDMT_initJobCCtxParams(&params);
DEBUGLOG(5, "ZSTDMT_initCStream_internal: switch to single blocking thread mode");
assert(singleThreadParams.nbWorkers == 0);
return ZSTD_initCStream_internal(mtctx->cctxPool->cctx[0],
dict, dictSize, cdict,
singleThreadParams, pledgedSrcSize);
&singleThreadParams, pledgedSrcSize);
}
DEBUGLOG(4, "ZSTDMT_initCStream_internal: %u workers", params.nbWorkers);
@ -1435,11 +1435,11 @@ size_t ZSTDMT_initCStream_internal(
mtctx->cdict = cdict;
}
mtctx->targetPrefixSize = ZSTDMT_computeOverlapSize(params);
mtctx->targetPrefixSize = ZSTDMT_computeOverlapSize(&params);
DEBUGLOG(4, "overlapLog=%i => %u KB", params.overlapLog, (U32)(mtctx->targetPrefixSize>>10));
mtctx->targetSectionSize = params.jobSize;
if (mtctx->targetSectionSize == 0) {
mtctx->targetSectionSize = 1ULL << ZSTDMT_computeTargetJobLog(params);
mtctx->targetSectionSize = 1ULL << ZSTDMT_computeTargetJobLog(&params);
}
assert(mtctx->targetSectionSize <= (size_t)ZSTDMT_JOBSIZE_MAX);