2aa14b1ab2
Updates the kernel's zstd library to v1.5.2, the latest zstd release. The upstream tag it is updated to is `v1.5.2-kernel`, which contains several cherry-picked commits on top of the v1.5.2 release which are required for the kernel update. I will create this tag once the PR is ready to merge, until then reference the temporary upstream branch `v1.5.2-kernel-cherrypicks`. I plan to submit this patch as part of the v6.2 merge window. I've done basic build testing & testing on x86-64, i386, and aarch64. I'm merging these patches into my `zstd-next` branch, which is pulled into `linux-next` for further testing. I've benchmarked BtrFS with zstd compression on a x86-64 machine, and saw these results. Decompression speed is a small win across the board. The lower compression levels 1-4 see both compression speed and compression ratio wins. The higher compression levels see a small compression speed loss and about neutral ratio. I expect the lower compression levels to be used much more heavily than the high compression levels, so this should be a net win. Level CTime DTime Ratio 1 -2.95% -1.1% -0.7% 3 -3.5% -1.2% -0.5% 5 +3.7% -1.0% +0.0% 7 +3.2% -0.9% +0.0% 9 -4.3% -0.8% +0.1% Signed-off-by: Nick Terrell <terrelln@fb.com>
160 lines
6.2 KiB
C
160 lines
6.2 KiB
C
/*
|
|
* Copyright (c) Yann Collet, Facebook, Inc.
|
|
* All rights reserved.
|
|
*
|
|
* This source code is licensed under both the BSD-style license (found in the
|
|
* LICENSE file in the root directory of this source tree) and the GPLv2 (found
|
|
* in the COPYING file in the root directory of this source tree).
|
|
* You may select, at your option, one of the above-listed licenses.
|
|
*/
|
|
|
|
/*-*************************************
|
|
* Dependencies
|
|
***************************************/
|
|
#include "zstd_compress_literals.h"
|
|
|
|
size_t ZSTD_noCompressLiterals (void* dst, size_t dstCapacity, const void* src, size_t srcSize)
|
|
{
|
|
BYTE* const ostart = (BYTE*)dst;
|
|
U32 const flSize = 1 + (srcSize>31) + (srcSize>4095);
|
|
|
|
RETURN_ERROR_IF(srcSize + flSize > dstCapacity, dstSize_tooSmall, "");
|
|
|
|
switch(flSize)
|
|
{
|
|
case 1: /* 2 - 1 - 5 */
|
|
ostart[0] = (BYTE)((U32)set_basic + (srcSize<<3));
|
|
break;
|
|
case 2: /* 2 - 2 - 12 */
|
|
MEM_writeLE16(ostart, (U16)((U32)set_basic + (1<<2) + (srcSize<<4)));
|
|
break;
|
|
case 3: /* 2 - 2 - 20 */
|
|
MEM_writeLE32(ostart, (U32)((U32)set_basic + (3<<2) + (srcSize<<4)));
|
|
break;
|
|
default: /* not necessary : flSize is {1,2,3} */
|
|
assert(0);
|
|
}
|
|
|
|
ZSTD_memcpy(ostart + flSize, src, srcSize);
|
|
DEBUGLOG(5, "Raw literals: %u -> %u", (U32)srcSize, (U32)(srcSize + flSize));
|
|
return srcSize + flSize;
|
|
}
|
|
|
|
size_t ZSTD_compressRleLiteralsBlock (void* dst, size_t dstCapacity, const void* src, size_t srcSize)
|
|
{
|
|
BYTE* const ostart = (BYTE*)dst;
|
|
U32 const flSize = 1 + (srcSize>31) + (srcSize>4095);
|
|
|
|
(void)dstCapacity; /* dstCapacity already guaranteed to be >=4, hence large enough */
|
|
|
|
switch(flSize)
|
|
{
|
|
case 1: /* 2 - 1 - 5 */
|
|
ostart[0] = (BYTE)((U32)set_rle + (srcSize<<3));
|
|
break;
|
|
case 2: /* 2 - 2 - 12 */
|
|
MEM_writeLE16(ostart, (U16)((U32)set_rle + (1<<2) + (srcSize<<4)));
|
|
break;
|
|
case 3: /* 2 - 2 - 20 */
|
|
MEM_writeLE32(ostart, (U32)((U32)set_rle + (3<<2) + (srcSize<<4)));
|
|
break;
|
|
default: /* not necessary : flSize is {1,2,3} */
|
|
assert(0);
|
|
}
|
|
|
|
ostart[flSize] = *(const BYTE*)src;
|
|
DEBUGLOG(5, "RLE literals: %u -> %u", (U32)srcSize, (U32)flSize + 1);
|
|
return flSize+1;
|
|
}
|
|
|
|
size_t ZSTD_compressLiterals (ZSTD_hufCTables_t const* prevHuf,
|
|
ZSTD_hufCTables_t* nextHuf,
|
|
ZSTD_strategy strategy, int disableLiteralCompression,
|
|
void* dst, size_t dstCapacity,
|
|
const void* src, size_t srcSize,
|
|
void* entropyWorkspace, size_t entropyWorkspaceSize,
|
|
const int bmi2,
|
|
unsigned suspectUncompressible)
|
|
{
|
|
size_t const minGain = ZSTD_minGain(srcSize, strategy);
|
|
size_t const lhSize = 3 + (srcSize >= 1 KB) + (srcSize >= 16 KB);
|
|
BYTE* const ostart = (BYTE*)dst;
|
|
U32 singleStream = srcSize < 256;
|
|
symbolEncodingType_e hType = set_compressed;
|
|
size_t cLitSize;
|
|
|
|
DEBUGLOG(5,"ZSTD_compressLiterals (disableLiteralCompression=%i srcSize=%u)",
|
|
disableLiteralCompression, (U32)srcSize);
|
|
|
|
/* Prepare nextEntropy assuming reusing the existing table */
|
|
ZSTD_memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
|
|
|
|
if (disableLiteralCompression)
|
|
return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);
|
|
|
|
/* small ? don't even attempt compression (speed opt) */
|
|
# define COMPRESS_LITERALS_SIZE_MIN 63
|
|
{ size_t const minLitSize = (prevHuf->repeatMode == HUF_repeat_valid) ? 6 : COMPRESS_LITERALS_SIZE_MIN;
|
|
if (srcSize <= minLitSize) return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);
|
|
}
|
|
|
|
RETURN_ERROR_IF(dstCapacity < lhSize+1, dstSize_tooSmall, "not enough space for compression");
|
|
{ HUF_repeat repeat = prevHuf->repeatMode;
|
|
int const preferRepeat = strategy < ZSTD_lazy ? srcSize <= 1024 : 0;
|
|
if (repeat == HUF_repeat_valid && lhSize == 3) singleStream = 1;
|
|
cLitSize = singleStream ?
|
|
HUF_compress1X_repeat(
|
|
ostart+lhSize, dstCapacity-lhSize, src, srcSize,
|
|
HUF_SYMBOLVALUE_MAX, HUF_TABLELOG_DEFAULT, entropyWorkspace, entropyWorkspaceSize,
|
|
(HUF_CElt*)nextHuf->CTable, &repeat, preferRepeat, bmi2, suspectUncompressible) :
|
|
HUF_compress4X_repeat(
|
|
ostart+lhSize, dstCapacity-lhSize, src, srcSize,
|
|
HUF_SYMBOLVALUE_MAX, HUF_TABLELOG_DEFAULT, entropyWorkspace, entropyWorkspaceSize,
|
|
(HUF_CElt*)nextHuf->CTable, &repeat, preferRepeat, bmi2, suspectUncompressible);
|
|
if (repeat != HUF_repeat_none) {
|
|
/* reused the existing table */
|
|
DEBUGLOG(5, "Reusing previous huffman table");
|
|
hType = set_repeat;
|
|
}
|
|
}
|
|
|
|
if ((cLitSize==0) || (cLitSize >= srcSize - minGain) || ERR_isError(cLitSize)) {
|
|
ZSTD_memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
|
|
return ZSTD_noCompressLiterals(dst, dstCapacity, src, srcSize);
|
|
}
|
|
if (cLitSize==1) {
|
|
ZSTD_memcpy(nextHuf, prevHuf, sizeof(*prevHuf));
|
|
return ZSTD_compressRleLiteralsBlock(dst, dstCapacity, src, srcSize);
|
|
}
|
|
|
|
if (hType == set_compressed) {
|
|
/* using a newly constructed table */
|
|
nextHuf->repeatMode = HUF_repeat_check;
|
|
}
|
|
|
|
/* Build header */
|
|
switch(lhSize)
|
|
{
|
|
case 3: /* 2 - 2 - 10 - 10 */
|
|
{ U32 const lhc = hType + ((!singleStream) << 2) + ((U32)srcSize<<4) + ((U32)cLitSize<<14);
|
|
MEM_writeLE24(ostart, lhc);
|
|
break;
|
|
}
|
|
case 4: /* 2 - 2 - 14 - 14 */
|
|
{ U32 const lhc = hType + (2 << 2) + ((U32)srcSize<<4) + ((U32)cLitSize<<18);
|
|
MEM_writeLE32(ostart, lhc);
|
|
break;
|
|
}
|
|
case 5: /* 2 - 2 - 18 - 18 */
|
|
{ U32 const lhc = hType + (3 << 2) + ((U32)srcSize<<4) + ((U32)cLitSize<<22);
|
|
MEM_writeLE32(ostart, lhc);
|
|
ostart[4] = (BYTE)(cLitSize >> 10);
|
|
break;
|
|
}
|
|
default: /* not possible : lhSize is {3,4,5} */
|
|
assert(0);
|
|
}
|
|
DEBUGLOG(5, "Compressed literals: %u -> %u", (U32)srcSize, (U32)(lhSize+cLitSize));
|
|
return lhSize+cLitSize;
|
|
}
|