From 4b9b25d943ff95c7421cab261333fc29852fe3b1 Mon Sep 17 00:00:00 2001 From: Brandon Maier Date: Thu, 12 Jan 2023 10:27:45 -0600 Subject: lib: zstd: update to latest Linux zstd 1.5.2 Update the zstd implementation to match Linux zstd 1.5.2 from commit 2aa14b1ab2. This was motivated by running into decompression corruption issues when trying to uncompress files compressed with newer versions of zstd. zstd users also claim significantly improved decompression times with newer zstd versions which is a side benefit. Original zstd code was copied from Linux commit 2aa14b1ab2 which is a custom-built implementation based on zstd 1.3.1. Linux switched to an implementation that is a copy of the upstream zstd code in Linux commit e0c1b49f5b, this results in a large code diff. However this should make future updates easier along with other benefits[1]. This commit is a straight mirror of the Linux zstd code, except to: - update a few #include that do not translate cleanly - linux/swab.h -> asm/byteorder.h - linux/limits.h -> linux/kernel.h - linux/module.h -> linux/compat.h - remove assert() from debug.h so it doesn't conflict with u-boot's assert() - strip out the compressor code as was done in the previous u-boot zstd - update existing zstd users to the new Linux zstd API - change the #define for MEM_STATIC to use INLINE_KEYWORD for codesize - add a new KConfig option that sets zstd build options to minify code based on zstd's ZSTD_LIB_MINIFY[2]. These changes were tested by booting a zstd 1.5.2 compressed kernel inside a FIT. And the squashfs changes by loading a file from zstd compressed squashfs with sqfsload. buildman was used to compile test other boards and check for binary bloat, as follows: > $ buildman -b zstd2 --boards dh_imx6,m53menlo,mvebu_espressobin-88f3720,sandbox,sandbox64,stm32mp15_dhcom_basic,stm32mp15_dhcor_basic,turris_mox,turris_omnia -sS > Summary of 6 commits for 9 boards (8 threads, 1 job per thread) > 01: Merge branch '2023-01-10-platform-updates' > arm: w+ m53menlo dh_imx6 > 02: lib: zstd: update to latest Linux zstd 1.5.2 > aarch64: (for 2/2 boards) all -3186.0 rodata +920.0 text -4106.0 > arm: (for 5/5 boards) all +1254.4 rodata +940.0 text +314.4 > sandbox: (for 2/2 boards) all -4452.0 data -16.0 rodata +640.0 text -5076.0 [1] https://github.com/torvalds/linux/commit/e0c1b49f5b674cca7b10549c53b3791d0bbc90a8 [2] https://github.com/facebook/zstd/blob/f302ad8811643c428c4e3498e28f53a0578020d3/lib/libzstd.mk#L31 Signed-off-by: Brandon Maier [trini: Set ret to -EINVAL for the error of "failed to detect compressed" to fix warning, drop ZSTD_SRCSIZEHINT_MAX for non-Linux host tool builds] Signed-off-by: Tom Rini --- lib/zstd/common/zstd_internal.h | 443 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 443 insertions(+) create mode 100644 lib/zstd/common/zstd_internal.h (limited to 'lib/zstd/common/zstd_internal.h') diff --git a/lib/zstd/common/zstd_internal.h b/lib/zstd/common/zstd_internal.h new file mode 100644 index 0000000000..93305d9b41 --- /dev/null +++ b/lib/zstd/common/zstd_internal.h @@ -0,0 +1,443 @@ +/* + * 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. + */ + +#ifndef ZSTD_CCOMMON_H_MODULE +#define ZSTD_CCOMMON_H_MODULE + +/* this module contains definitions which must be identical + * across compression, decompression and dictBuilder. + * It also contains a few functions useful to at least 2 of them + * and which benefit from being inlined */ + +/*-************************************* +* Dependencies +***************************************/ +#include "compiler.h" +#include "cpu.h" +#include "mem.h" +#include "debug.h" /* assert, DEBUGLOG, RAWLOG, g_debuglevel */ +#include "error_private.h" +#define ZSTD_STATIC_LINKING_ONLY +#include +#define FSE_STATIC_LINKING_ONLY +#include "fse.h" +#define HUF_STATIC_LINKING_ONLY +#include "huf.h" +#include /* XXH_reset, update, digest */ +#define ZSTD_TRACE 0 + + +/* ---- static assert (debug) --- */ +#define ZSTD_STATIC_ASSERT(c) DEBUG_STATIC_ASSERT(c) +#define ZSTD_isError ERR_isError /* for inlining */ +#define FSE_isError ERR_isError +#define HUF_isError ERR_isError + + +/*-************************************* +* shared macros +***************************************/ +#undef MIN +#undef MAX +#define MIN(a,b) ((a)<(b) ? (a) : (b)) +#define MAX(a,b) ((a)>(b) ? (a) : (b)) +#define BOUNDED(min,val,max) (MAX(min,MIN(val,max))) + + +/*-************************************* +* Common constants +***************************************/ +#define ZSTD_OPT_NUM (1<<12) + +#define ZSTD_REP_NUM 3 /* number of repcodes */ +static UNUSED_ATTR const U32 repStartValue[ZSTD_REP_NUM] = { 1, 4, 8 }; + +#define KB *(1 <<10) +#define MB *(1 <<20) +#define GB *(1U<<30) + +#define BIT7 128 +#define BIT6 64 +#define BIT5 32 +#define BIT4 16 +#define BIT1 2 +#define BIT0 1 + +#define ZSTD_WINDOWLOG_ABSOLUTEMIN 10 +static UNUSED_ATTR const size_t ZSTD_fcs_fieldSize[4] = { 0, 2, 4, 8 }; +static UNUSED_ATTR const size_t ZSTD_did_fieldSize[4] = { 0, 1, 2, 4 }; + +#define ZSTD_FRAMEIDSIZE 4 /* magic number size */ + +#define ZSTD_BLOCKHEADERSIZE 3 /* C standard doesn't allow `static const` variable to be init using another `static const` variable */ +static UNUSED_ATTR const size_t ZSTD_blockHeaderSize = ZSTD_BLOCKHEADERSIZE; +typedef enum { bt_raw, bt_rle, bt_compressed, bt_reserved } blockType_e; + +#define ZSTD_FRAMECHECKSUMSIZE 4 + +#define MIN_SEQUENCES_SIZE 1 /* nbSeq==0 */ +#define MIN_CBLOCK_SIZE (1 /*litCSize*/ + 1 /* RLE or RAW */ + MIN_SEQUENCES_SIZE /* nbSeq==0 */) /* for a non-null block */ + +#define HufLog 12 +typedef enum { set_basic, set_rle, set_compressed, set_repeat } symbolEncodingType_e; + +#define LONGNBSEQ 0x7F00 + +#define MINMATCH 3 + +#define Litbits 8 +#define MaxLit ((1<= WILDCOPY_VECLEN || diff <= -WILDCOPY_VECLEN); + /* Separate out the first COPY16() call because the copy length is + * almost certain to be short, so the branches have different + * probabilities. Since it is almost certain to be short, only do + * one COPY16() in the first call. Then, do two calls per loop since + * at that point it is more likely to have a high trip count. + */ +#ifdef __aarch64__ + do { + COPY16(op, ip); + } + while (op < oend); +#else + ZSTD_copy16(op, ip); + if (16 >= length) return; + op += 16; + ip += 16; + do { + COPY16(op, ip); + COPY16(op, ip); + } + while (op < oend); +#endif + } +} + +MEM_STATIC size_t ZSTD_limitCopy(void* dst, size_t dstCapacity, const void* src, size_t srcSize) +{ + size_t const length = MIN(dstCapacity, srcSize); + if (length > 0) { + ZSTD_memcpy(dst, src, length); + } + return length; +} + +/* define "workspace is too large" as this number of times larger than needed */ +#define ZSTD_WORKSPACETOOLARGE_FACTOR 3 + +/* when workspace is continuously too large + * during at least this number of times, + * context's memory usage is considered wasteful, + * because it's sized to handle a worst case scenario which rarely happens. + * In which case, resize it down to free some memory */ +#define ZSTD_WORKSPACETOOLARGE_MAXDURATION 128 + +/* Controls whether the input/output buffer is buffered or stable. */ +typedef enum { + ZSTD_bm_buffered = 0, /* Buffer the input/output */ + ZSTD_bm_stable = 1 /* ZSTD_inBuffer/ZSTD_outBuffer is stable */ +} ZSTD_bufferMode_e; + + +/*-******************************************* +* Private declarations +*********************************************/ +typedef struct seqDef_s { + U32 offBase; /* offBase == Offset + ZSTD_REP_NUM, or repcode 1,2,3 */ + U16 litLength; + U16 mlBase; /* mlBase == matchLength - MINMATCH */ +} seqDef; + +/* Controls whether seqStore has a single "long" litLength or matchLength. See seqStore_t. */ +typedef enum { + ZSTD_llt_none = 0, /* no longLengthType */ + ZSTD_llt_literalLength = 1, /* represents a long literal */ + ZSTD_llt_matchLength = 2 /* represents a long match */ +} ZSTD_longLengthType_e; + +typedef struct { + seqDef* sequencesStart; + seqDef* sequences; /* ptr to end of sequences */ + BYTE* litStart; + BYTE* lit; /* ptr to end of literals */ + BYTE* llCode; + BYTE* mlCode; + BYTE* ofCode; + size_t maxNbSeq; + size_t maxNbLit; + + /* longLengthPos and longLengthType to allow us to represent either a single litLength or matchLength + * in the seqStore that has a value larger than U16 (if it exists). To do so, we increment + * the existing value of the litLength or matchLength by 0x10000. + */ + ZSTD_longLengthType_e longLengthType; + U32 longLengthPos; /* Index of the sequence to apply long length modification to */ +} seqStore_t; + +typedef struct { + U32 litLength; + U32 matchLength; +} ZSTD_sequenceLength; + +/* + * Returns the ZSTD_sequenceLength for the given sequences. It handles the decoding of long sequences + * indicated by longLengthPos and longLengthType, and adds MINMATCH back to matchLength. + */ +MEM_STATIC ZSTD_sequenceLength ZSTD_getSequenceLength(seqStore_t const* seqStore, seqDef const* seq) +{ + ZSTD_sequenceLength seqLen; + seqLen.litLength = seq->litLength; + seqLen.matchLength = seq->mlBase + MINMATCH; + if (seqStore->longLengthPos == (U32)(seq - seqStore->sequencesStart)) { + if (seqStore->longLengthType == ZSTD_llt_literalLength) { + seqLen.litLength += 0xFFFF; + } + if (seqStore->longLengthType == ZSTD_llt_matchLength) { + seqLen.matchLength += 0xFFFF; + } + } + return seqLen; +} + +/* + * Contains the compressed frame size and an upper-bound for the decompressed frame size. + * Note: before using `compressedSize`, check for errors using ZSTD_isError(). + * similarly, before using `decompressedBound`, check for errors using: + * `decompressedBound != ZSTD_CONTENTSIZE_ERROR` + */ +typedef struct { + size_t compressedSize; + unsigned long long decompressedBound; +} ZSTD_frameSizeInfo; /* decompress & legacy */ + +const seqStore_t* ZSTD_getSeqStore(const ZSTD_CCtx* ctx); /* compress & dictBuilder */ +void ZSTD_seqToCodes(const seqStore_t* seqStorePtr); /* compress, dictBuilder, decodeCorpus (shouldn't get its definition from here) */ + +/* custom memory allocation functions */ +void* ZSTD_customMalloc(size_t size, ZSTD_customMem customMem); +void* ZSTD_customCalloc(size_t size, ZSTD_customMem customMem); +void ZSTD_customFree(void* ptr, ZSTD_customMem customMem); + + +MEM_STATIC U32 ZSTD_highbit32(U32 val) /* compress, dictBuilder, decodeCorpus */ +{ + assert(val != 0); + { +# if (__GNUC__ >= 3) /* GCC Intrinsic */ + return __builtin_clz (val) ^ 31; +# else /* Software version */ + static const U32 DeBruijnClz[32] = { 0, 9, 1, 10, 13, 21, 2, 29, 11, 14, 16, 18, 22, 25, 3, 30, 8, 12, 20, 28, 15, 17, 24, 7, 19, 27, 23, 6, 26, 5, 4, 31 }; + U32 v = val; + v |= v >> 1; + v |= v >> 2; + v |= v >> 4; + v |= v >> 8; + v |= v >> 16; + return DeBruijnClz[(v * 0x07C4ACDDU) >> 27]; +# endif + } +} + +/* + * Counts the number of trailing zeros of a `size_t`. + * Most compilers should support CTZ as a builtin. A backup + * implementation is provided if the builtin isn't supported, but + * it may not be terribly efficient. + */ +MEM_STATIC unsigned ZSTD_countTrailingZeros(size_t val) +{ + if (MEM_64bits()) { +# if (__GNUC__ >= 4) + return __builtin_ctzll((U64)val); +# else + static const int DeBruijnBytePos[64] = { 0, 1, 2, 7, 3, 13, 8, 19, + 4, 25, 14, 28, 9, 34, 20, 56, + 5, 17, 26, 54, 15, 41, 29, 43, + 10, 31, 38, 35, 21, 45, 49, 57, + 63, 6, 12, 18, 24, 27, 33, 55, + 16, 53, 40, 42, 30, 37, 44, 48, + 62, 11, 23, 32, 52, 39, 36, 47, + 61, 22, 51, 46, 60, 50, 59, 58 }; + return DeBruijnBytePos[((U64)((val & -(long long)val) * 0x0218A392CDABBD3FULL)) >> 58]; +# endif + } else { /* 32 bits */ +# if (__GNUC__ >= 3) + return __builtin_ctz((U32)val); +# else + static const int DeBruijnBytePos[32] = { 0, 1, 28, 2, 29, 14, 24, 3, + 30, 22, 20, 15, 25, 17, 4, 8, + 31, 27, 13, 23, 21, 19, 16, 7, + 26, 12, 18, 6, 11, 5, 10, 9 }; + return DeBruijnBytePos[((U32)((val & -(S32)val) * 0x077CB531U)) >> 27]; +# endif + } +} + + +/* ZSTD_invalidateRepCodes() : + * ensures next compression will not use repcodes from previous block. + * Note : only works with regular variant; + * do not use with extDict variant ! */ +void ZSTD_invalidateRepCodes(ZSTD_CCtx* cctx); /* zstdmt, adaptive_compression (shouldn't get this definition from here) */ + + +typedef struct { + blockType_e blockType; + U32 lastBlock; + U32 origSize; +} blockProperties_t; /* declared here for decompress and fullbench */ + +/*! ZSTD_getcBlockSize() : + * Provides the size of compressed block from block header `src` */ +/* Used by: decompress, fullbench (does not get its definition from here) */ +size_t ZSTD_getcBlockSize(const void* src, size_t srcSize, + blockProperties_t* bpPtr); + +/*! ZSTD_decodeSeqHeaders() : + * decode sequence header from src */ +/* Used by: decompress, fullbench (does not get its definition from here) */ +size_t ZSTD_decodeSeqHeaders(ZSTD_DCtx* dctx, int* nbSeqPtr, + const void* src, size_t srcSize); + +/* + * @returns true iff the CPU supports dynamic BMI2 dispatch. + */ +MEM_STATIC int ZSTD_cpuSupportsBmi2(void) +{ + ZSTD_cpuid_t cpuid = ZSTD_cpuid(); + return ZSTD_cpuid_bmi1(cpuid) && ZSTD_cpuid_bmi2(cpuid); +} + + +#endif /* ZSTD_CCOMMON_H_MODULE */ -- cgit v1.2.3