zstd/lib/compress/zstd_compress_superblock.c

850 lines
42 KiB
C
Raw Normal View History

2019-11-05 20:51:25 +00:00
/*
* Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
2019-11-05 20:51:25 +00:00
* 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 "zstd_compress_superblock.h"
#include "../common/zstd_internal.h" /* ZSTD_getSequenceLength */
#include "hist.h" /* HIST_countFast_wksp */
2019-11-05 20:51:25 +00:00
#include "zstd_compress_internal.h"
#include "zstd_compress_sequences.h"
#include "zstd_compress_literals.h"
/*-*************************************
* Superblock entropy buffer structs
***************************************/
/** ZSTD_hufCTablesMetadata_t :
* Stores Literals Block Type for a super-block in hType, and
* huffman tree description in hufDesBuffer.
* hufDesSize refers to the size of huffman tree description in bytes.
* This metadata is populated in ZSTD_buildSuperBlockEntropy_literal() */
typedef struct {
symbolEncodingType_e hType;
BYTE hufDesBuffer[500]; /* TODO give name to this value */
2019-11-05 20:51:25 +00:00
size_t hufDesSize;
} ZSTD_hufCTablesMetadata_t;
/** ZSTD_fseCTablesMetadata_t :
* Stores symbol compression modes for a super-block in {ll, ol, ml}Type, and
* fse tables in fseTablesBuffer.
* fseTablesSize refers to the size of fse tables in bytes.
* This metadata is populated in ZSTD_buildSuperBlockEntropy_sequences() */
typedef struct {
symbolEncodingType_e llType;
symbolEncodingType_e ofType;
symbolEncodingType_e mlType;
BYTE fseTablesBuffer[500]; /* TODO give name to this value */
2019-11-05 20:51:25 +00:00
size_t fseTablesSize;
size_t lastCountSize; /* This is to account for bug in 1.3.4. More detail in ZSTD_compressSubBlock_sequences() */
2019-11-05 20:51:25 +00:00
} ZSTD_fseCTablesMetadata_t;
typedef struct {
ZSTD_hufCTablesMetadata_t hufMetadata;
ZSTD_fseCTablesMetadata_t fseMetadata;
} ZSTD_entropyCTablesMetadata_t;
/** ZSTD_buildSuperBlockEntropy_literal() :
* Builds entropy for the super-block literals.
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
* Stores literals block type (raw, rle, compressed, repeat) and
2019-11-05 20:51:25 +00:00
* huffman description table to hufMetadata.
* @return : size of huffman description table or error code */
static size_t ZSTD_buildSuperBlockEntropy_literal(void* const src, size_t srcSize,
const ZSTD_hufCTables_t* prevHuf,
ZSTD_hufCTables_t* nextHuf,
ZSTD_hufCTablesMetadata_t* hufMetadata,
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
const int disableLiteralsCompression,
2019-11-05 20:51:25 +00:00
void* workspace, size_t wkspSize)
{
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 = wkspEnd-nodeWksp;
unsigned maxSymbolValue = 255;
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
unsigned huffLog = HUF_TABLELOG_DEFAULT;
HUF_repeat repeat = prevHuf->repeatMode;
2019-11-05 20:51:25 +00:00
DEBUGLOG(5, "ZSTD_buildSuperBlockEntropy_literal (srcSize=%zu)", srcSize);
/* Prepare nextEntropy assuming reusing the existing table */
ZSTD_memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
2019-11-05 20:51:25 +00:00
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
if (disableLiteralsCompression) {
DEBUGLOG(5, "set_basic - disabled");
hufMetadata->hType = set_basic;
return 0;
}
2019-11-05 20:51:25 +00:00
/* small ? don't even attempt compression (speed opt) */
# define COMPRESS_LITERALS_SIZE_MIN 63
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
{ 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;
}
2019-11-05 20:51:25 +00:00
}
/* 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");
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
if (largest == srcSize) {
DEBUGLOG(5, "set_rle");
hufMetadata->hType = set_rle;
return 0;
}
if (largest <= (srcSize >> 7)+4) {
DEBUGLOG(5, "set_basic - no gain");
hufMetadata->hType = set_basic;
return 0;
}
2019-11-05 20:51:25 +00:00
}
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
/* Validate the previous Huffman table */
if (repeat == HUF_repeat_check && !HUF_validateCTable((HUF_CElt const*)prevHuf->CTable, countWksp, maxSymbolValue)) {
repeat = HUF_repeat_none;
}
2019-11-05 20:51:25 +00:00
/* Build Huffman Tree */
ZSTD_memset(nextHuf->CTable, 0, sizeof(nextHuf->CTable));
2019-11-05 20:51:25 +00:00
huffLog = HUF_optimalTableLog(huffLog, srcSize, maxSymbolValue);
{ size_t const maxBits = HUF_buildCTable_wksp((HUF_CElt*)nextHuf->CTable, countWksp,
maxSymbolValue, huffLog,
nodeWksp, nodeWkspSize);
FORWARD_IF_ERROR(maxBits, "HUF_buildCTable_wksp");
2019-11-05 20:51:25 +00:00
huffLog = (U32)maxBits;
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
{ /* Build and write the CTable */
size_t const newCSize = HUF_estimateCompressedSize(
(HUF_CElt*)nextHuf->CTable, countWksp, maxSymbolValue);
size_t const hSize = HUF_writeCTable(
hufMetadata->hufDesBuffer, sizeof(hufMetadata->hufDesBuffer),
(HUF_CElt*)nextHuf->CTable, maxSymbolValue, huffLog);
/* 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));
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
hufMetadata->hType = set_repeat;
return 0;
}
}
if (newCSize + hSize >= srcSize) {
DEBUGLOG(5, "set_basic - no gains");
ZSTD_memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
hufMetadata->hType = set_basic;
return 0;
}
DEBUGLOG(5, "set_compressed (hSize=%u)", (U32)hSize);
2019-11-05 20:51:25 +00:00
hufMetadata->hType = set_compressed;
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
nextHuf->repeatMode = HUF_repeat_check;
2019-11-05 20:51:25 +00:00
return hSize;
}
}
}
/** ZSTD_buildSuperBlockEntropy_sequences() :
* Builds entropy for the super-block sequences.
* Stores symbol compression modes and fse table to fseMetadata.
* @return : size of fse tables or error code */
static size_t ZSTD_buildSuperBlockEntropy_sequences(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)
{
BYTE* const wkspStart = (BYTE*)workspace;
BYTE* const wkspEnd = wkspStart + wkspSize;
BYTE* const countWkspStart = wkspStart;
unsigned* const countWksp = (unsigned*)workspace;
const size_t countWkspSize = (MaxSeq + 1) * sizeof(unsigned);
BYTE* const cTableWksp = countWkspStart + countWkspSize;
const size_t cTableWkspSize = wkspEnd-cTableWksp;
ZSTD_strategy const strategy = cctxParams->cParams.strategy;
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;
size_t const nbSeq = seqStorePtr->sequences - seqStorePtr->sequencesStart;
BYTE* const ostart = fseMetadata->fseTablesBuffer;
BYTE* const oend = ostart + sizeof(fseMetadata->fseTablesBuffer);
BYTE* op = ostart;
assert(cTableWkspSize >= (1 << MaxFSELog) * sizeof(FSE_FUNCTION_TYPE));
DEBUGLOG(5, "ZSTD_buildSuperBlockEntropy_sequences (nbSeq=%zu)", nbSeq);
ZSTD_memset(workspace, 0, wkspSize);
2019-11-05 20:51:25 +00:00
fseMetadata->lastCountSize = 0;
/* convert length/distances into codes */
ZSTD_seqToCodes(seqStorePtr);
/* build CTable for Literal Lengths */
{ U32 LLtype;
unsigned max = MaxLL;
size_t const mostFrequent = HIST_countFast_wksp(countWksp, &max, llCodeTable, nbSeq, workspace, wkspSize); /* can't fail */
DEBUGLOG(5, "Building LL table");
nextEntropy->litlength_repeatMode = prevEntropy->litlength_repeatMode;
LLtype = ZSTD_selectEncodingType(&nextEntropy->litlength_repeatMode,
countWksp, max, mostFrequent, nbSeq,
LLFSELog, prevEntropy->litlengthCTable,
LL_defaultNorm, LL_defaultNormLog,
ZSTD_defaultAllowed, strategy);
assert(set_basic < set_compressed && set_rle < set_compressed);
assert(!(LLtype < set_compressed && nextEntropy->litlength_repeatMode != FSE_repeat_none)); /* We don't copy tables */
{ size_t const countSize = ZSTD_buildCTable(op, oend - op, CTable_LitLength, LLFSELog, (symbolEncodingType_e)LLtype,
countWksp, max, llCodeTable, nbSeq, LL_defaultNorm, LL_defaultNormLog, MaxLL,
prevEntropy->litlengthCTable, sizeof(prevEntropy->litlengthCTable),
cTableWksp, cTableWkspSize);
FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for LitLens failed");
2019-11-05 20:51:25 +00:00
if (LLtype == set_compressed)
fseMetadata->lastCountSize = countSize;
op += countSize;
fseMetadata->llType = (symbolEncodingType_e) LLtype;
} }
/* build CTable for Offsets */
{ U32 Offtype;
unsigned max = MaxOff;
size_t const mostFrequent = HIST_countFast_wksp(countWksp, &max, ofCodeTable, nbSeq, workspace, wkspSize); /* 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;
Offtype = ZSTD_selectEncodingType(&nextEntropy->offcode_repeatMode,
countWksp, max, mostFrequent, nbSeq,
OffFSELog, prevEntropy->offcodeCTable,
OF_defaultNorm, OF_defaultNormLog,
defaultPolicy, strategy);
assert(!(Offtype < set_compressed && nextEntropy->offcode_repeatMode != FSE_repeat_none)); /* We don't copy tables */
{ size_t const countSize = ZSTD_buildCTable(op, oend - op, CTable_OffsetBits, OffFSELog, (symbolEncodingType_e)Offtype,
countWksp, max, ofCodeTable, nbSeq, OF_defaultNorm, OF_defaultNormLog, DefaultMaxOff,
prevEntropy->offcodeCTable, sizeof(prevEntropy->offcodeCTable),
cTableWksp, cTableWkspSize);
FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for Offsets failed");
2019-11-05 20:51:25 +00:00
if (Offtype == set_compressed)
fseMetadata->lastCountSize = countSize;
op += countSize;
fseMetadata->ofType = (symbolEncodingType_e) Offtype;
} }
/* build CTable for MatchLengths */
{ U32 MLtype;
unsigned max = MaxML;
size_t const mostFrequent = HIST_countFast_wksp(countWksp, &max, mlCodeTable, nbSeq, workspace, wkspSize); /* can't fail */
DEBUGLOG(5, "Building ML table (remaining space : %i)", (int)(oend-op));
nextEntropy->matchlength_repeatMode = prevEntropy->matchlength_repeatMode;
MLtype = ZSTD_selectEncodingType(&nextEntropy->matchlength_repeatMode,
countWksp, max, mostFrequent, nbSeq,
MLFSELog, prevEntropy->matchlengthCTable,
ML_defaultNorm, ML_defaultNormLog,
ZSTD_defaultAllowed, strategy);
assert(!(MLtype < set_compressed && nextEntropy->matchlength_repeatMode != FSE_repeat_none)); /* We don't copy tables */
{ size_t const countSize = ZSTD_buildCTable(op, oend - op, CTable_MatchLength, MLFSELog, (symbolEncodingType_e)MLtype,
countWksp, max, mlCodeTable, nbSeq, ML_defaultNorm, ML_defaultNormLog, MaxML,
prevEntropy->matchlengthCTable, sizeof(prevEntropy->matchlengthCTable),
cTableWksp, cTableWkspSize);
FORWARD_IF_ERROR(countSize, "ZSTD_buildCTable for MatchLengths failed");
2019-11-05 20:51:25 +00:00
if (MLtype == set_compressed)
fseMetadata->lastCountSize = countSize;
op += countSize;
fseMetadata->mlType = (symbolEncodingType_e) MLtype;
} }
assert((size_t) (op-ostart) <= sizeof(fseMetadata->fseTablesBuffer));
return op-ostart;
}
/** ZSTD_buildSuperBlockEntropy() :
* Builds entropy for the super-block.
* @return : 0 on success or error code */
static size_t
ZSTD_buildSuperBlockEntropy(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 = seqStorePtr->lit - seqStorePtr->litStart;
DEBUGLOG(5, "ZSTD_buildSuperBlockEntropy");
entropyMetadata->hufMetadata.hufDesSize =
ZSTD_buildSuperBlockEntropy_literal(seqStorePtr->litStart, litSize,
&prevEntropy->huf, &nextEntropy->huf,
&entropyMetadata->hufMetadata,
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
ZSTD_disableLiteralsCompression(cctxParams),
2019-11-05 20:51:25 +00:00
workspace, wkspSize);
FORWARD_IF_ERROR(entropyMetadata->hufMetadata.hufDesSize, "ZSTD_buildSuperBlockEntropy_literal failed");
2019-11-05 20:51:25 +00:00
entropyMetadata->fseMetadata.fseTablesSize =
ZSTD_buildSuperBlockEntropy_sequences(seqStorePtr,
&prevEntropy->fse, &nextEntropy->fse,
cctxParams,
&entropyMetadata->fseMetadata,
workspace, wkspSize);
FORWARD_IF_ERROR(entropyMetadata->fseMetadata.fseTablesSize, "ZSTD_buildSuperBlockEntropy_sequences failed");
2019-11-05 20:51:25 +00:00
return 0;
}
/** ZSTD_compressSubBlock_literal() :
* Compresses literals section for a sub-block.
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
* When we have to write the Huffman table we will sometimes choose a header
* size larger than necessary. This is because we have to pick the header size
* before we know the table size + compressed size, so we have a bound on the
* table size. If we guessed incorrectly, we fall back to uncompressed literals.
*
* We write the header when writeEntropy=1 and set entropyWrriten=1 when we succeeded
* in writing the header, otherwise it is set to 0.
*
2019-11-05 20:51:25 +00:00
* hufMetadata->hType has literals block type info.
* If it is set_basic, all sub-blocks literals section will be Raw_Literals_Block.
* If it is set_rle, all sub-blocks literals section will be RLE_Literals_Block.
* If it is set_compressed, first sub-block's literals section will be Compressed_Literals_Block
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
* If it is set_compressed, first sub-block's literals section will be Treeless_Literals_Block
2019-11-05 20:51:25 +00:00
* and the following sub-blocks' literals sections will be Treeless_Literals_Block.
* @return : compressed size of literals section of a sub-block
* Or 0 if it unable to compress.
* Or error code */
static size_t ZSTD_compressSubBlock_literal(const HUF_CElt* hufTable,
const ZSTD_hufCTablesMetadata_t* hufMetadata,
const BYTE* literals, size_t litSize,
void* dst, size_t dstSize,
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
const int bmi2, int writeEntropy, int* entropyWritten)
2019-11-05 20:51:25 +00:00
{
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
size_t const header = writeEntropy ? 200 : 0;
size_t const lhSize = 3 + (litSize >= (1 KB - header)) + (litSize >= (16 KB - header));
2019-11-05 20:51:25 +00:00
BYTE* const ostart = (BYTE*)dst;
BYTE* const oend = ostart + dstSize;
BYTE* op = ostart + lhSize;
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
U32 const singleStream = lhSize == 3;
symbolEncodingType_e hType = writeEntropy ? hufMetadata->hType : set_repeat;
2019-11-05 20:51:25 +00:00
size_t cLitSize = 0;
(void)bmi2; /* TODO bmi2... */
2019-11-05 20:51:25 +00:00
DEBUGLOG(5, "ZSTD_compressSubBlock_literal (litSize=%zu, lhSize=%zu, writeEntropy=%d)", litSize, lhSize, writeEntropy);
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
*entropyWritten = 0;
2019-11-05 20:51:25 +00:00
if (litSize == 0 || hufMetadata->hType == set_basic) {
DEBUGLOG(5, "ZSTD_compressSubBlock_literal using raw literal");
return ZSTD_noCompressLiterals(dst, dstSize, literals, litSize);
} else if (hufMetadata->hType == set_rle) {
DEBUGLOG(5, "ZSTD_compressSubBlock_literal using rle literal");
return ZSTD_compressRleLiteralsBlock(dst, dstSize, literals, litSize);
}
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
assert(litSize > 0);
assert(hufMetadata->hType == set_compressed || hufMetadata->hType == set_repeat);
if (writeEntropy && hufMetadata->hType == set_compressed) {
ZSTD_memcpy(op, hufMetadata->hufDesBuffer, hufMetadata->hufDesSize);
2019-11-05 20:51:25 +00:00
op += hufMetadata->hufDesSize;
cLitSize += hufMetadata->hufDesSize;
DEBUGLOG(5, "ZSTD_compressSubBlock_literal (hSize=%zu)", hufMetadata->hufDesSize);
}
/* TODO bmi2 */
2019-11-05 20:51:25 +00:00
{ const size_t cSize = singleStream ? HUF_compress1X_usingCTable(op, oend-op, literals, litSize, hufTable)
: HUF_compress4X_usingCTable(op, oend-op, literals, litSize, hufTable);
op += cSize;
cLitSize += cSize;
if (cSize == 0 || ERR_isError(cSize)) {
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
DEBUGLOG(5, "Failed to write entropy tables %s", ZSTD_getErrorName(cSize));
return 0;
}
/* If we expand and we aren't writing a header then emit uncompressed */
if (!writeEntropy && cLitSize >= litSize) {
DEBUGLOG(5, "ZSTD_compressSubBlock_literal using raw literal because uncompressible");
return ZSTD_noCompressLiterals(dst, dstSize, literals, litSize);
2019-11-05 20:51:25 +00:00
}
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
/* If we are writing headers then allow expansion that doesn't change our header size. */
if (lhSize < (size_t)(3 + (cLitSize >= 1 KB) + (cLitSize >= 16 KB))) {
assert(cLitSize > litSize);
DEBUGLOG(5, "Literals expanded beyond allowed header size");
return ZSTD_noCompressLiterals(dst, dstSize, literals, litSize);
2019-11-05 20:51:25 +00:00
}
DEBUGLOG(5, "ZSTD_compressSubBlock_literal (cSize=%zu)", cSize);
}
/* Build header */
switch(lhSize)
{
case 3: /* 2 - 2 - 10 - 10 */
{ U32 const lhc = hType + ((!singleStream) << 2) + ((U32)litSize<<4) + ((U32)cLitSize<<14);
MEM_writeLE24(ostart, lhc);
break;
}
case 4: /* 2 - 2 - 14 - 14 */
{ U32 const lhc = hType + (2 << 2) + ((U32)litSize<<4) + ((U32)cLitSize<<18);
MEM_writeLE32(ostart, lhc);
break;
}
case 5: /* 2 - 2 - 18 - 18 */
{ U32 const lhc = hType + (3 << 2) + ((U32)litSize<<4) + ((U32)cLitSize<<22);
MEM_writeLE32(ostart, lhc);
ostart[4] = (BYTE)(cLitSize >> 10);
break;
}
default: /* not possible : lhSize is {3,4,5} */
assert(0);
}
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
*entropyWritten = 1;
DEBUGLOG(5, "Compressed literals: %u -> %u", (U32)litSize, (U32)(op-ostart));
2019-11-05 20:51:25 +00:00
return op-ostart;
}
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
static size_t ZSTD_seqDecompressedSize(seqStore_t const* seqStore, const seqDef* sequences, size_t nbSeq, size_t litSize, int lastSequence) {
2019-11-05 20:51:25 +00:00
const seqDef* const sstart = sequences;
const seqDef* const send = sequences + nbSeq;
const seqDef* sp = sstart;
size_t matchLengthSum = 0;
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
size_t litLengthSum = 0;
2019-11-05 20:51:25 +00:00
while (send-sp > 0) {
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
ZSTD_sequenceLength const seqLen = ZSTD_getSequenceLength(seqStore, sp);
litLengthSum += seqLen.litLength;
matchLengthSum += seqLen.matchLength;
sp++;
}
assert(litLengthSum <= litSize);
if (!lastSequence) {
assert(litLengthSum == litSize);
2019-11-05 20:51:25 +00:00
}
return matchLengthSum + litSize;
}
/** ZSTD_compressSubBlock_sequences() :
* Compresses sequences section for a sub-block.
* fseMetadata->llType, fseMetadata->ofType, and fseMetadata->mlType have
* symbol compression modes for the super-block.
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
* The first successfully compressed block will have these in its header.
* We set entropyWritten=1 when we succeed in compressing the sequences.
* The following sub-blocks will always have repeat mode.
2019-11-05 20:51:25 +00:00
* @return : compressed size of sequences section of a sub-block
* Or 0 if it is unable to compress
* Or error code. */
static size_t ZSTD_compressSubBlock_sequences(const ZSTD_fseCTables_t* fseTables,
const ZSTD_fseCTablesMetadata_t* fseMetadata,
const seqDef* sequences, size_t nbSeq,
const BYTE* llCode, const BYTE* mlCode, const BYTE* ofCode,
const ZSTD_CCtx_params* cctxParams,
void* dst, size_t dstCapacity,
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
const int bmi2, int writeEntropy, int* entropyWritten)
2019-11-05 20:51:25 +00:00
{
const int longOffsets = cctxParams->cParams.windowLog > STREAM_ACCUMULATOR_MIN;
BYTE* const ostart = (BYTE*)dst;
BYTE* const oend = ostart + dstCapacity;
BYTE* op = ostart;
BYTE* seqHead;
DEBUGLOG(5, "ZSTD_compressSubBlock_sequences (nbSeq=%zu, writeEntropy=%d, longOffsets=%d)", nbSeq, writeEntropy, longOffsets);
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
*entropyWritten = 0;
2019-11-05 20:51:25 +00:00
/* Sequences Header */
RETURN_ERROR_IF((oend-op) < 3 /*max nbSeq Size*/ + 1 /*seqHead*/,
dstSize_tooSmall, "");
2019-11-05 20:51:25 +00:00
if (nbSeq < 0x7F)
*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;
if (nbSeq==0) {
return op - ostart;
}
/* seqHead : flags for FSE encoding type */
seqHead = op++;
DEBUGLOG(5, "ZSTD_compressSubBlock_sequences (seqHeadSize=%u)", (unsigned)(op-ostart));
if (writeEntropy) {
const U32 LLtype = fseMetadata->llType;
const U32 Offtype = fseMetadata->ofType;
const U32 MLtype = fseMetadata->mlType;
DEBUGLOG(5, "ZSTD_compressSubBlock_sequences (fseTablesSize=%zu)", fseMetadata->fseTablesSize);
*seqHead = (BYTE)((LLtype<<6) + (Offtype<<4) + (MLtype<<2));
ZSTD_memcpy(op, fseMetadata->fseTablesBuffer, fseMetadata->fseTablesSize);
2019-11-05 20:51:25 +00:00
op += fseMetadata->fseTablesSize;
} else {
const U32 repeat = set_repeat;
*seqHead = (BYTE)((repeat<<6) + (repeat<<4) + (repeat<<2));
}
{ size_t const bitstreamSize = ZSTD_encodeSequences(
op, oend - op,
fseTables->matchlengthCTable, mlCode,
fseTables->offcodeCTable, ofCode,
fseTables->litlengthCTable, llCode,
sequences, nbSeq,
longOffsets, bmi2);
FORWARD_IF_ERROR(bitstreamSize, "ZSTD_encodeSequences failed");
2019-11-05 20:51:25 +00:00
op += bitstreamSize;
/* 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.
*/
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
#ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
2019-11-05 20:51:25 +00:00
if (writeEntropy && fseMetadata->lastCountSize && fseMetadata->lastCountSize + bitstreamSize < 4) {
/* NCountSize >= 2 && bitstreamSize > 0 ==> lastCountSize == 3 */
assert(fseMetadata->lastCountSize + bitstreamSize == 3);
DEBUGLOG(5, "Avoiding bug in zstd decoder in versions <= 1.3.4 by "
"emitting an uncompressed block.");
return 0;
}
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
#endif
2019-11-05 20:51:25 +00:00
DEBUGLOG(5, "ZSTD_compressSubBlock_sequences (bitstreamSize=%zu)", bitstreamSize);
}
/* zstd versions <= 1.4.0 mistakenly report error when
* sequences section body size is less than 3 bytes.
* Fixed by https://github.com/facebook/zstd/pull/1664.
* This can happen when the previous sequences section block is compressed
* with rle mode and the current block's sequences section is compressed
* with repeat mode where sequences section body size can be 1 byte.
*/
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
#ifndef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
2019-11-05 20:51:25 +00:00
if (op-seqHead < 4) {
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
DEBUGLOG(5, "Avoiding bug in zstd decoder in versions <= 1.4.0 by emitting "
"an uncompressed block when sequences are < 4 bytes");
2019-11-05 20:51:25 +00:00
return 0;
}
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
#endif
2019-11-05 20:51:25 +00:00
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
*entropyWritten = 1;
2019-11-05 20:51:25 +00:00
return op - ostart;
}
/** ZSTD_compressSubBlock() :
* Compresses a single sub-block.
* @return : compressed size of the sub-block
* Or 0 if it failed to compress. */
static size_t ZSTD_compressSubBlock(const ZSTD_entropyCTables_t* entropy,
const ZSTD_entropyCTablesMetadata_t* entropyMetadata,
const seqDef* sequences, size_t nbSeq,
const BYTE* literals, size_t litSize,
const BYTE* llCode, const BYTE* mlCode, const BYTE* ofCode,
const ZSTD_CCtx_params* cctxParams,
void* dst, size_t dstCapacity,
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
const int bmi2,
int writeLitEntropy, int writeSeqEntropy,
int* litEntropyWritten, int* seqEntropyWritten,
U32 lastBlock)
2019-11-05 20:51:25 +00:00
{
BYTE* const ostart = (BYTE*)dst;
BYTE* const oend = ostart + dstCapacity;
BYTE* op = ostart + ZSTD_blockHeaderSize;
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
DEBUGLOG(5, "ZSTD_compressSubBlock (litSize=%zu, nbSeq=%zu, writeLitEntropy=%d, writeSeqEntropy=%d, lastBlock=%d)",
litSize, nbSeq, writeLitEntropy, writeSeqEntropy, lastBlock);
2019-11-05 20:51:25 +00:00
{ size_t cLitSize = ZSTD_compressSubBlock_literal((const HUF_CElt*)entropy->huf.CTable,
&entropyMetadata->hufMetadata, literals, litSize,
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
op, oend-op, bmi2, writeLitEntropy, litEntropyWritten);
FORWARD_IF_ERROR(cLitSize, "ZSTD_compressSubBlock_literal failed");
2019-11-05 20:51:25 +00:00
if (cLitSize == 0) return 0;
op += cLitSize;
}
{ size_t cSeqSize = ZSTD_compressSubBlock_sequences(&entropy->fse,
&entropyMetadata->fseMetadata,
sequences, nbSeq,
llCode, mlCode, ofCode,
cctxParams,
op, oend-op,
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
bmi2, writeSeqEntropy, seqEntropyWritten);
FORWARD_IF_ERROR(cSeqSize, "ZSTD_compressSubBlock_sequences failed");
2019-11-05 20:51:25 +00:00
if (cSeqSize == 0) return 0;
op += cSeqSize;
}
/* Write block header */
{ size_t cSize = (op-ostart)-ZSTD_blockHeaderSize;
U32 const cBlockHeader24 = lastBlock + (((U32)bt_compressed)<<1) + (U32)(cSize << 3);
MEM_writeLE24(ostart, cBlockHeader24);
}
return op-ostart;
}
static size_t ZSTD_estimateSubBlockSize_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 = 255;
size_t literalSectionHeaderSize = 3; /* Use hard coded size of 3 bytes */
if (hufMetadata->hType == set_basic) return litSize;
else if (hufMetadata->hType == set_rle) return 1;
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
else if (hufMetadata->hType == set_compressed || hufMetadata->hType == set_repeat) {
2019-11-05 20:51:25 +00:00
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;
return cLitSizeEstimate + literalSectionHeaderSize;
} }
assert(0); /* impossible */
return 0;
}
static size_t ZSTD_estimateSubBlockSize_symbolType(symbolEncodingType_e type,
const BYTE* codeTable, unsigned maxCode,
size_t nbSeq, const FSE_CTable* fseCTable,
const U32* additionalBits,
short const* defaultNorm, U32 defaultNormLog, U32 defaultMax,
2019-11-05 20:51:25 +00:00
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);
cSymbolTypeSizeEstimateInBits = max <= defaultMax
? ZSTD_crossEntropyCost(defaultNorm, defaultNormLog, countWksp, max)
: ERROR(GENERIC);
2019-11-05 20:51:25 +00:00
} 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 / 8;
}
static size_t ZSTD_estimateSubBlockSize_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 = 3; /* Use hard coded size of 3 bytes */
size_t cSeqSizeEstimate = 0;
cSeqSizeEstimate += ZSTD_estimateSubBlockSize_symbolType(fseMetadata->ofType, ofCodeTable, MaxOff,
nbSeq, fseTables->offcodeCTable, NULL,
OF_defaultNorm, OF_defaultNormLog, DefaultMaxOff,
2019-11-05 20:51:25 +00:00
workspace, wkspSize);
cSeqSizeEstimate += ZSTD_estimateSubBlockSize_symbolType(fseMetadata->llType, llCodeTable, MaxLL,
nbSeq, fseTables->litlengthCTable, LL_bits,
LL_defaultNorm, LL_defaultNormLog, MaxLL,
2019-11-05 20:51:25 +00:00
workspace, wkspSize);
cSeqSizeEstimate += ZSTD_estimateSubBlockSize_symbolType(fseMetadata->mlType, mlCodeTable, MaxML,
nbSeq, fseTables->matchlengthCTable, ML_bits,
ML_defaultNorm, ML_defaultNormLog, MaxML,
2019-11-05 20:51:25 +00:00
workspace, wkspSize);
if (writeEntropy) cSeqSizeEstimate += fseMetadata->fseTablesSize;
return cSeqSizeEstimate + sequencesSectionHeaderSize;
}
static size_t ZSTD_estimateSubBlockSize(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,
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
int writeLitEntropy, int writeSeqEntropy) {
2019-11-05 20:51:25 +00:00
size_t cSizeEstimate = 0;
cSizeEstimate += ZSTD_estimateSubBlockSize_literal(literals, litSize,
&entropy->huf, &entropyMetadata->hufMetadata,
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
workspace, wkspSize, writeLitEntropy);
2019-11-05 20:51:25 +00:00
cSizeEstimate += ZSTD_estimateSubBlockSize_sequences(ofCodeTable, llCodeTable, mlCodeTable,
nbSeq, &entropy->fse, &entropyMetadata->fseMetadata,
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
workspace, wkspSize, writeSeqEntropy);
2019-11-05 20:51:25 +00:00
return cSizeEstimate + ZSTD_blockHeaderSize;
}
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
static int ZSTD_needSequenceEntropyTables(ZSTD_fseCTablesMetadata_t const* fseMetadata)
{
if (fseMetadata->llType == set_compressed || fseMetadata->llType == set_rle)
return 1;
if (fseMetadata->mlType == set_compressed || fseMetadata->mlType == set_rle)
return 1;
if (fseMetadata->ofType == set_compressed || fseMetadata->ofType == set_rle)
return 1;
return 0;
}
2019-11-05 20:51:25 +00:00
/** ZSTD_compressSubBlock_multi() :
* Breaks super-block into multiple sub-blocks and compresses them.
* Entropy will be written to the first block.
* The following blocks will use repeat mode to compress.
* All sub-blocks are compressed blocks (no raw or rle blocks).
* @return : compressed size of the super block (which is multiple ZSTD blocks)
* Or 0 if it failed to compress. */
static size_t ZSTD_compressSubBlock_multi(const seqStore_t* seqStorePtr,
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
const ZSTD_compressedBlockState_t* prevCBlock,
ZSTD_compressedBlockState_t* nextCBlock,
2019-11-05 20:51:25 +00:00
const ZSTD_entropyCTablesMetadata_t* entropyMetadata,
const ZSTD_CCtx_params* cctxParams,
void* dst, size_t dstCapacity,
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
const void* src, size_t srcSize,
2019-11-05 20:51:25 +00:00
const int bmi2, U32 lastBlock,
void* workspace, size_t wkspSize)
{
const seqDef* const sstart = seqStorePtr->sequencesStart;
const seqDef* const send = seqStorePtr->sequences;
const seqDef* sp = sstart;
const BYTE* const lstart = seqStorePtr->litStart;
const BYTE* const lend = seqStorePtr->lit;
const BYTE* lp = lstart;
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
BYTE const* ip = (BYTE const*)src;
BYTE const* const iend = ip + srcSize;
2019-11-05 20:51:25 +00:00
BYTE* const ostart = (BYTE*)dst;
BYTE* const oend = ostart + dstCapacity;
BYTE* op = ostart;
const BYTE* llCodePtr = seqStorePtr->llCode;
const BYTE* mlCodePtr = seqStorePtr->mlCode;
const BYTE* ofCodePtr = seqStorePtr->ofCode;
size_t targetCBlockSize = cctxParams->targetCBlockSize;
size_t litSize, seqCount;
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
int writeLitEntropy = entropyMetadata->hufMetadata.hType == set_compressed;
int writeSeqEntropy = 1;
int lastSequence = 0;
2019-11-05 20:51:25 +00:00
DEBUGLOG(5, "ZSTD_compressSubBlock_multi (litSize=%u, nbSeq=%u)",
(unsigned)(lend-lp), (unsigned)(send-sstart));
litSize = 0;
seqCount = 0;
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
do {
size_t cBlockSizeEstimate = 0;
if (sstart == send) {
lastSequence = 1;
} else {
const seqDef* const sequence = sp + seqCount;
lastSequence = sequence == send - 1;
litSize += ZSTD_getSequenceLength(seqStorePtr, sequence).litLength;
seqCount++;
}
if (lastSequence) {
assert(lp <= lend);
assert(litSize <= (size_t)(lend - lp));
litSize = (size_t)(lend - lp);
}
2019-11-05 20:54:05 +00:00
/* I think there is an optimization opportunity here.
* Calling ZSTD_estimateSubBlockSize for every sequence can be wasteful
* since it recalculates estimate from scratch.
* For example, it would recount literal distribution and symbol codes everytime.
2019-11-05 20:54:05 +00:00
*/
2019-11-05 20:51:25 +00:00
cBlockSizeEstimate = ZSTD_estimateSubBlockSize(lp, litSize, ofCodePtr, llCodePtr, mlCodePtr, seqCount,
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
&nextCBlock->entropy, entropyMetadata,
workspace, wkspSize, writeLitEntropy, writeSeqEntropy);
2019-11-05 20:51:25 +00:00
if (cBlockSizeEstimate > targetCBlockSize || lastSequence) {
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
int litEntropyWritten = 0;
int seqEntropyWritten = 0;
const size_t decompressedSize = ZSTD_seqDecompressedSize(seqStorePtr, sp, seqCount, litSize, lastSequence);
const size_t cSize = ZSTD_compressSubBlock(&nextCBlock->entropy, entropyMetadata,
2019-11-05 20:51:25 +00:00
sp, seqCount,
lp, litSize,
llCodePtr, mlCodePtr, ofCodePtr,
cctxParams,
op, oend-op,
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
bmi2, writeLitEntropy, writeSeqEntropy,
&litEntropyWritten, &seqEntropyWritten,
lastBlock && lastSequence);
FORWARD_IF_ERROR(cSize, "ZSTD_compressSubBlock failed");
2019-11-05 20:51:25 +00:00
if (cSize > 0 && cSize < decompressedSize) {
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
DEBUGLOG(5, "Committed the sub-block");
assert(ip + decompressedSize <= iend);
ip += decompressedSize;
2019-11-05 20:51:25 +00:00
sp += seqCount;
lp += litSize;
op += cSize;
llCodePtr += seqCount;
mlCodePtr += seqCount;
ofCodePtr += seqCount;
litSize = 0;
seqCount = 0;
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
/* Entropy only needs to be written once */
if (litEntropyWritten) {
writeLitEntropy = 0;
}
if (seqEntropyWritten) {
writeSeqEntropy = 0;
}
2019-11-05 20:51:25 +00:00
}
}
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
} while (!lastSequence);
if (writeLitEntropy) {
DEBUGLOG(5, "ZSTD_compressSubBlock_multi has literal entropy tables unwritten");
ZSTD_memcpy(&nextCBlock->entropy.huf, &prevCBlock->entropy.huf, sizeof(prevCBlock->entropy.huf));
2019-11-05 20:51:25 +00:00
}
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
if (writeSeqEntropy && ZSTD_needSequenceEntropyTables(&entropyMetadata->fseMetadata)) {
/* If we haven't written our entropy tables, then we've violated our contract and
* must emit an uncompressed block.
*/
DEBUGLOG(5, "ZSTD_compressSubBlock_multi has sequence entropy tables unwritten");
2019-11-05 20:51:25 +00:00
return 0;
}
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
if (ip < iend) {
size_t const cSize = ZSTD_noCompressBlock(op, oend - op, ip, iend - ip, lastBlock);
DEBUGLOG(5, "ZSTD_compressSubBlock_multi last sub-block uncompressed, %zu bytes", (size_t)(iend - ip));
FORWARD_IF_ERROR(cSize, "ZSTD_noCompressBlock failed");
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
assert(cSize != 0);
op += cSize;
/* We have to regenerate the repcodes because we've skipped some sequences */
if (sp < send) {
seqDef const* seq;
repcodes_t rep;
ZSTD_memcpy(&rep, prevCBlock->rep, sizeof(rep));
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
for (seq = sstart; seq < sp; ++seq) {
rep = ZSTD_updateRep(rep.rep, seq->offset - 1, ZSTD_getSequenceLength(seqStorePtr, seq).litLength == 0);
}
ZSTD_memcpy(nextCBlock->rep, &rep, sizeof(rep));
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
}
}
2019-11-05 20:51:25 +00:00
DEBUGLOG(5, "ZSTD_compressSubBlock_multi compressed");
return op-ostart;
}
size_t ZSTD_compressSuperBlock(ZSTD_CCtx* zc,
void* dst, size_t dstCapacity,
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
void const* src, size_t srcSize,
2019-11-05 20:51:25 +00:00
unsigned lastBlock) {
ZSTD_entropyCTablesMetadata_t entropyMetadata;
FORWARD_IF_ERROR(ZSTD_buildSuperBlockEntropy(&zc->seqStore,
&zc->blockState.prevCBlock->entropy,
&zc->blockState.nextCBlock->entropy,
&zc->appliedParams,
&entropyMetadata,
zc->entropyWorkspace, HUF_WORKSPACE_SIZE /* statically allocated in resetCCtx */), "");
2019-11-05 20:51:25 +00:00
return ZSTD_compressSubBlock_multi(&zc->seqStore,
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
zc->blockState.prevCBlock,
zc->blockState.nextCBlock,
2019-11-05 20:51:25 +00:00
&entropyMetadata,
&zc->appliedParams,
dst, dstCapacity,
Fix superblock mode (#2100) Fixes: Enable RLE blocks for superblock mode Fix the limitation that the literals block must shrink. Instead, when we're within 200 bytes of the next header byte size, we will just use the next one up. That way we should (almost?) always have space for the table. Remove the limitation that the first sub-block MUST have compressed literals and be compressed. Now one sub-block MUST be compressed (otherwise we fall back to raw block which is okay, since that is streamable). If no block has compressed literals that is okay, we will fix up the next Huffman table. Handle the case where the last sub-block is uncompressed (maybe it is very small). Before it would skip superblock in this case, now we allow the last sub-block to be uncompressed. To do this we need to regenerate the correct repcodes. Respect disableLiteralsCompression in superblock mode Fix superblock mode to handle a block consisting of only compressed literals Fix a off by 1 error in superblock mode that disabled it whenever there were last literals Fix superblock mode with long literals/matches (> 0xFFFF) Allow superblock mode to repeat Huffman tables Respect ZSTD_minGain(). Tests: Simple check for the condition in #2096. When the simple_round_trip fuzzer enables superblock mode, it checks that the compressed size isn't expanded too much. Remaining limitations: O(targetCBlockSize^2) because we recompute statistics every sequence Unable to split literals of length > targetCBlockSize into multiple sequences Refuses to generate sub-blocks that don't shrink the compressed data, so we could end up with large sub-blocks. We should emit those sections as uncompressed blocks instead. ... Fixes #2096
2020-05-01 23:11:47 +00:00
src, srcSize,
2019-11-05 20:51:25 +00:00
zc->bmi2, lastBlock,
zc->entropyWorkspace, HUF_WORKSPACE_SIZE /* statically allocated in resetCCtx */);
}