Home | History | Annotate | Line # | Download | only in decompress
      1 /*
      2  * Copyright (c) Meta Platforms, Inc. and affiliates.
      3  * All rights reserved.
      4  *
      5  * This source code is licensed under both the BSD-style license (found in the
      6  * LICENSE file in the root directory of this source tree) and the GPLv2 (found
      7  * in the COPYING file in the root directory of this source tree).
      8  * You may select, at your option, one of the above-listed licenses.
      9  */
     10 
     11 
     12 /* zstd_decompress_internal:
     13  * objects and definitions shared within lib/decompress modules */
     14 
     15  #ifndef ZSTD_DECOMPRESS_INTERNAL_H
     16  #define ZSTD_DECOMPRESS_INTERNAL_H
     17 
     18 
     19 /*-*******************************************************
     20  *  Dependencies
     21  *********************************************************/
     22 #include "../common/mem.h"             /* BYTE, U16, U32 */
     23 #include "../common/zstd_internal.h"   /* constants : MaxLL, MaxML, MaxOff, LLFSELog, etc. */
     24 
     25 
     26 
     27 /*-*******************************************************
     28  *  Constants
     29  *********************************************************/
     30 static UNUSED_ATTR const U32 LL_base[MaxLL+1] = {
     31                  0,    1,    2,     3,     4,     5,     6,      7,
     32                  8,    9,   10,    11,    12,    13,    14,     15,
     33                 16,   18,   20,    22,    24,    28,    32,     40,
     34                 48,   64, 0x80, 0x100, 0x200, 0x400, 0x800, 0x1000,
     35                 0x2000, 0x4000, 0x8000, 0x10000 };
     36 
     37 static UNUSED_ATTR const U32 OF_base[MaxOff+1] = {
     38                  0,        1,       1,       5,     0xD,     0x1D,     0x3D,     0x7D,
     39                  0xFD,   0x1FD,   0x3FD,   0x7FD,   0xFFD,   0x1FFD,   0x3FFD,   0x7FFD,
     40                  0xFFFD, 0x1FFFD, 0x3FFFD, 0x7FFFD, 0xFFFFD, 0x1FFFFD, 0x3FFFFD, 0x7FFFFD,
     41                  0xFFFFFD, 0x1FFFFFD, 0x3FFFFFD, 0x7FFFFFD, 0xFFFFFFD, 0x1FFFFFFD, 0x3FFFFFFD, 0x7FFFFFFD };
     42 
     43 static UNUSED_ATTR const U8 OF_bits[MaxOff+1] = {
     44                      0,  1,  2,  3,  4,  5,  6,  7,
     45                      8,  9, 10, 11, 12, 13, 14, 15,
     46                     16, 17, 18, 19, 20, 21, 22, 23,
     47                     24, 25, 26, 27, 28, 29, 30, 31 };
     48 
     49 static UNUSED_ATTR const U32 ML_base[MaxML+1] = {
     50                      3,  4,  5,    6,     7,     8,     9,    10,
     51                     11, 12, 13,   14,    15,    16,    17,    18,
     52                     19, 20, 21,   22,    23,    24,    25,    26,
     53                     27, 28, 29,   30,    31,    32,    33,    34,
     54                     35, 37, 39,   41,    43,    47,    51,    59,
     55                     67, 83, 99, 0x83, 0x103, 0x203, 0x403, 0x803,
     56                     0x1003, 0x2003, 0x4003, 0x8003, 0x10003 };
     57 
     58 
     59 /*-*******************************************************
     60  *  Decompression types
     61  *********************************************************/
     62  typedef struct {
     63      U32 fastMode;
     64      U32 tableLog;
     65  } ZSTD_seqSymbol_header;
     66 
     67  typedef struct {
     68      U16  nextState;
     69      BYTE nbAdditionalBits;
     70      BYTE nbBits;
     71      U32  baseValue;
     72  } ZSTD_seqSymbol;
     73 
     74  #define SEQSYMBOL_TABLE_SIZE(log)   (1 + (1 << (log)))
     75 
     76 #define ZSTD_BUILD_FSE_TABLE_WKSP_SIZE (sizeof(S16) * (MaxSeq + 1) + (1u << MaxFSELog) + sizeof(U64))
     77 #define ZSTD_BUILD_FSE_TABLE_WKSP_SIZE_U32 ((ZSTD_BUILD_FSE_TABLE_WKSP_SIZE + sizeof(U32) - 1) / sizeof(U32))
     78 #define ZSTD_HUFFDTABLE_CAPACITY_LOG 12
     79 
     80 typedef struct {
     81     ZSTD_seqSymbol LLTable[SEQSYMBOL_TABLE_SIZE(LLFSELog)];    /* Note : Space reserved for FSE Tables */
     82     ZSTD_seqSymbol OFTable[SEQSYMBOL_TABLE_SIZE(OffFSELog)];   /* is also used as temporary workspace while building hufTable during DDict creation */
     83     ZSTD_seqSymbol MLTable[SEQSYMBOL_TABLE_SIZE(MLFSELog)];    /* and therefore must be at least HUF_DECOMPRESS_WORKSPACE_SIZE large */
     84     HUF_DTable hufTable[HUF_DTABLE_SIZE(ZSTD_HUFFDTABLE_CAPACITY_LOG)];  /* can accommodate HUF_decompress4X */
     85     U32 rep[ZSTD_REP_NUM];
     86     U32 workspace[ZSTD_BUILD_FSE_TABLE_WKSP_SIZE_U32];
     87 } ZSTD_entropyDTables_t;
     88 
     89 typedef enum { ZSTDds_getFrameHeaderSize, ZSTDds_decodeFrameHeader,
     90                ZSTDds_decodeBlockHeader, ZSTDds_decompressBlock,
     91                ZSTDds_decompressLastBlock, ZSTDds_checkChecksum,
     92                ZSTDds_decodeSkippableHeader, ZSTDds_skipFrame } ZSTD_dStage;
     93 
     94 typedef enum { zdss_init=0, zdss_loadHeader,
     95                zdss_read, zdss_load, zdss_flush } ZSTD_dStreamStage;
     96 
     97 typedef enum {
     98     ZSTD_use_indefinitely = -1,  /* Use the dictionary indefinitely */
     99     ZSTD_dont_use = 0,           /* Do not use the dictionary (if one exists free it) */
    100     ZSTD_use_once = 1            /* Use the dictionary once and set to ZSTD_dont_use */
    101 } ZSTD_dictUses_e;
    102 
    103 /* Hashset for storing references to multiple ZSTD_DDict within ZSTD_DCtx */
    104 typedef struct {
    105     const ZSTD_DDict** ddictPtrTable;
    106     size_t ddictPtrTableSize;
    107     size_t ddictPtrCount;
    108 } ZSTD_DDictHashSet;
    109 
    110 #ifndef ZSTD_DECODER_INTERNAL_BUFFER
    111 #  define ZSTD_DECODER_INTERNAL_BUFFER  (1 << 16)
    112 #endif
    113 
    114 #define ZSTD_LBMIN 64
    115 #define ZSTD_LBMAX (128 << 10)
    116 
    117 /* extra buffer, compensates when dst is not large enough to store litBuffer */
    118 #define ZSTD_LITBUFFEREXTRASIZE  BOUNDED(ZSTD_LBMIN, ZSTD_DECODER_INTERNAL_BUFFER, ZSTD_LBMAX)
    119 
    120 typedef enum {
    121     ZSTD_not_in_dst = 0,  /* Stored entirely within litExtraBuffer */
    122     ZSTD_in_dst = 1,           /* Stored entirely within dst (in memory after current output write) */
    123     ZSTD_split = 2            /* Split between litExtraBuffer and dst */
    124 } ZSTD_litLocation_e;
    125 
    126 struct ZSTD_DCtx_s
    127 {
    128     const ZSTD_seqSymbol* LLTptr;
    129     const ZSTD_seqSymbol* MLTptr;
    130     const ZSTD_seqSymbol* OFTptr;
    131     const HUF_DTable* HUFptr;
    132     ZSTD_entropyDTables_t entropy;
    133     U32 workspace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32];   /* space needed when building huffman tables */
    134     const void* previousDstEnd;   /* detect continuity */
    135     const void* prefixStart;      /* start of current segment */
    136     const void* virtualStart;     /* virtual start of previous segment if it was just before current one */
    137     const void* dictEnd;          /* end of previous segment */
    138     size_t expected;
    139     ZSTD_frameHeader fParams;
    140     U64 processedCSize;
    141     U64 decodedSize;
    142     blockType_e bType;            /* used in ZSTD_decompressContinue(), store blockType between block header decoding and block decompression stages */
    143     ZSTD_dStage stage;
    144     U32 litEntropy;
    145     U32 fseEntropy;
    146     XXH64_state_t xxhState;
    147     size_t headerSize;
    148     ZSTD_format_e format;
    149     ZSTD_forceIgnoreChecksum_e forceIgnoreChecksum;   /* User specified: if == 1, will ignore checksums in compressed frame. Default == 0 */
    150     U32 validateChecksum;         /* if == 1, will validate checksum. Is == 1 if (fParams.checksumFlag == 1) and (forceIgnoreChecksum == 0). */
    151     const BYTE* litPtr;
    152     ZSTD_customMem customMem;
    153     size_t litSize;
    154     size_t rleSize;
    155     size_t staticSize;
    156     int isFrameDecompression;
    157 #if DYNAMIC_BMI2 != 0
    158     int bmi2;                     /* == 1 if the CPU supports BMI2 and 0 otherwise. CPU support is determined dynamically once per context lifetime. */
    159 #endif
    160 
    161     /* dictionary */
    162     ZSTD_DDict* ddictLocal;
    163     const ZSTD_DDict* ddict;     /* set by ZSTD_initDStream_usingDDict(), or ZSTD_DCtx_refDDict() */
    164     U32 dictID;
    165     int ddictIsCold;             /* if == 1 : dictionary is "new" for working context, and presumed "cold" (not in cpu cache) */
    166     ZSTD_dictUses_e dictUses;
    167     ZSTD_DDictHashSet* ddictSet;                    /* Hash set for multiple ddicts */
    168     ZSTD_refMultipleDDicts_e refMultipleDDicts;     /* User specified: if == 1, will allow references to multiple DDicts. Default == 0 (disabled) */
    169     int disableHufAsm;
    170     int maxBlockSizeParam;
    171 
    172     /* streaming */
    173     ZSTD_dStreamStage streamStage;
    174     char*  inBuff;
    175     size_t inBuffSize;
    176     size_t inPos;
    177     size_t maxWindowSize;
    178     char*  outBuff;
    179     size_t outBuffSize;
    180     size_t outStart;
    181     size_t outEnd;
    182     size_t lhSize;
    183 #if defined(ZSTD_LEGACY_SUPPORT) && (ZSTD_LEGACY_SUPPORT>=1)
    184     void* legacyContext;
    185     U32 previousLegacyVersion;
    186     U32 legacyVersion;
    187 #endif
    188     U32 hostageByte;
    189     int noForwardProgress;
    190     ZSTD_bufferMode_e outBufferMode;
    191     ZSTD_outBuffer expectedOutBuffer;
    192 
    193     /* workspace */
    194     BYTE* litBuffer;
    195     const BYTE* litBufferEnd;
    196     ZSTD_litLocation_e litBufferLocation;
    197     BYTE litExtraBuffer[ZSTD_LITBUFFEREXTRASIZE + WILDCOPY_OVERLENGTH]; /* literal buffer can be split between storage within dst and within this scratch buffer */
    198     BYTE headerBuffer[ZSTD_FRAMEHEADERSIZE_MAX];
    199 
    200     size_t oversizedDuration;
    201 
    202 #ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
    203     void const* dictContentBeginForFuzzing;
    204     void const* dictContentEndForFuzzing;
    205 #endif
    206 
    207     /* Tracing */
    208 #if ZSTD_TRACE
    209     ZSTD_TraceCtx traceCtx;
    210 #endif
    211 };  /* typedef'd to ZSTD_DCtx within "zstd.h" */
    212 
    213 MEM_STATIC int ZSTD_DCtx_get_bmi2(const struct ZSTD_DCtx_s *dctx) {
    214 #if DYNAMIC_BMI2 != 0
    215 	return dctx->bmi2;
    216 #else
    217     (void)dctx;
    218 	return 0;
    219 #endif
    220 }
    221 
    222 /*-*******************************************************
    223  *  Shared internal functions
    224  *********************************************************/
    225 
    226 /*! ZSTD_loadDEntropy() :
    227  *  dict : must point at beginning of a valid zstd dictionary.
    228  * @return : size of dictionary header (size of magic number + dict ID + entropy tables) */
    229 size_t ZSTD_loadDEntropy(ZSTD_entropyDTables_t* entropy,
    230                    const void* const dict, size_t const dictSize);
    231 
    232 /*! ZSTD_checkContinuity() :
    233  *  check if next `dst` follows previous position, where decompression ended.
    234  *  If yes, do nothing (continue on current segment).
    235  *  If not, classify previous segment as "external dictionary", and start a new segment.
    236  *  This function cannot fail. */
    237 void ZSTD_checkContinuity(ZSTD_DCtx* dctx, const void* dst, size_t dstSize);
    238 
    239 
    240 #endif /* ZSTD_DECOMPRESS_INTERNAL_H */
    241