1*a28cd43dSSascha Wildner /*
2*a28cd43dSSascha Wildner  * Copyright (c) 2016-2020, Yann Collet, Facebook, Inc.
3*a28cd43dSSascha Wildner  * All rights reserved.
4*a28cd43dSSascha Wildner  *
5*a28cd43dSSascha Wildner  * This source code is licensed under both the BSD-style license (found in the
6*a28cd43dSSascha Wildner  * LICENSE file in the root directory of this source tree) and the GPLv2 (found
7*a28cd43dSSascha Wildner  * in the COPYING file in the root directory of this source tree).
8*a28cd43dSSascha Wildner  * You may select, at your option, one of the above-listed licenses.
9*a28cd43dSSascha Wildner  */
10*a28cd43dSSascha Wildner 
11*a28cd43dSSascha Wildner 
12*a28cd43dSSascha Wildner /* zstd_decompress_internal:
13*a28cd43dSSascha Wildner  * objects and definitions shared within lib/decompress modules */
14*a28cd43dSSascha Wildner 
15*a28cd43dSSascha Wildner  #ifndef ZSTD_DECOMPRESS_INTERNAL_H
16*a28cd43dSSascha Wildner  #define ZSTD_DECOMPRESS_INTERNAL_H
17*a28cd43dSSascha Wildner 
18*a28cd43dSSascha Wildner 
19*a28cd43dSSascha Wildner /*-*******************************************************
20*a28cd43dSSascha Wildner  *  Dependencies
21*a28cd43dSSascha Wildner  *********************************************************/
22*a28cd43dSSascha Wildner #include "../common/mem.h"             /* BYTE, U16, U32 */
23*a28cd43dSSascha Wildner #include "../common/zstd_internal.h"   /* ZSTD_seqSymbol */
24*a28cd43dSSascha Wildner 
25*a28cd43dSSascha Wildner 
26*a28cd43dSSascha Wildner 
27*a28cd43dSSascha Wildner /*-*******************************************************
28*a28cd43dSSascha Wildner  *  Constants
29*a28cd43dSSascha Wildner  *********************************************************/
30*a28cd43dSSascha Wildner static UNUSED_ATTR const U32 LL_base[MaxLL+1] = {
31*a28cd43dSSascha Wildner                  0,    1,    2,     3,     4,     5,     6,      7,
32*a28cd43dSSascha Wildner                  8,    9,   10,    11,    12,    13,    14,     15,
33*a28cd43dSSascha Wildner                 16,   18,   20,    22,    24,    28,    32,     40,
34*a28cd43dSSascha Wildner                 48,   64, 0x80, 0x100, 0x200, 0x400, 0x800, 0x1000,
35*a28cd43dSSascha Wildner                 0x2000, 0x4000, 0x8000, 0x10000 };
36*a28cd43dSSascha Wildner 
37*a28cd43dSSascha Wildner static UNUSED_ATTR const U32 OF_base[MaxOff+1] = {
38*a28cd43dSSascha Wildner                  0,        1,       1,       5,     0xD,     0x1D,     0x3D,     0x7D,
39*a28cd43dSSascha Wildner                  0xFD,   0x1FD,   0x3FD,   0x7FD,   0xFFD,   0x1FFD,   0x3FFD,   0x7FFD,
40*a28cd43dSSascha Wildner                  0xFFFD, 0x1FFFD, 0x3FFFD, 0x7FFFD, 0xFFFFD, 0x1FFFFD, 0x3FFFFD, 0x7FFFFD,
41*a28cd43dSSascha Wildner                  0xFFFFFD, 0x1FFFFFD, 0x3FFFFFD, 0x7FFFFFD, 0xFFFFFFD, 0x1FFFFFFD, 0x3FFFFFFD, 0x7FFFFFFD };
42*a28cd43dSSascha Wildner 
43*a28cd43dSSascha Wildner static UNUSED_ATTR const U32 OF_bits[MaxOff+1] = {
44*a28cd43dSSascha Wildner                      0,  1,  2,  3,  4,  5,  6,  7,
45*a28cd43dSSascha Wildner                      8,  9, 10, 11, 12, 13, 14, 15,
46*a28cd43dSSascha Wildner                     16, 17, 18, 19, 20, 21, 22, 23,
47*a28cd43dSSascha Wildner                     24, 25, 26, 27, 28, 29, 30, 31 };
48*a28cd43dSSascha Wildner 
49*a28cd43dSSascha Wildner static UNUSED_ATTR const U32 ML_base[MaxML+1] = {
50*a28cd43dSSascha Wildner                      3,  4,  5,    6,     7,     8,     9,    10,
51*a28cd43dSSascha Wildner                     11, 12, 13,   14,    15,    16,    17,    18,
52*a28cd43dSSascha Wildner                     19, 20, 21,   22,    23,    24,    25,    26,
53*a28cd43dSSascha Wildner                     27, 28, 29,   30,    31,    32,    33,    34,
54*a28cd43dSSascha Wildner                     35, 37, 39,   41,    43,    47,    51,    59,
55*a28cd43dSSascha Wildner                     67, 83, 99, 0x83, 0x103, 0x203, 0x403, 0x803,
56*a28cd43dSSascha Wildner                     0x1003, 0x2003, 0x4003, 0x8003, 0x10003 };
57*a28cd43dSSascha Wildner 
58*a28cd43dSSascha Wildner 
59*a28cd43dSSascha Wildner /*-*******************************************************
60*a28cd43dSSascha Wildner  *  Decompression types
61*a28cd43dSSascha Wildner  *********************************************************/
62*a28cd43dSSascha Wildner  typedef struct {
63*a28cd43dSSascha Wildner      U32 fastMode;
64*a28cd43dSSascha Wildner      U32 tableLog;
65*a28cd43dSSascha Wildner  } ZSTD_seqSymbol_header;
66*a28cd43dSSascha Wildner 
67*a28cd43dSSascha Wildner  typedef struct {
68*a28cd43dSSascha Wildner      U16  nextState;
69*a28cd43dSSascha Wildner      BYTE nbAdditionalBits;
70*a28cd43dSSascha Wildner      BYTE nbBits;
71*a28cd43dSSascha Wildner      U32  baseValue;
72*a28cd43dSSascha Wildner  } ZSTD_seqSymbol;
73*a28cd43dSSascha Wildner 
74*a28cd43dSSascha Wildner  #define SEQSYMBOL_TABLE_SIZE(log)   (1 + (1 << (log)))
75*a28cd43dSSascha Wildner 
76*a28cd43dSSascha Wildner #define ZSTD_BUILD_FSE_TABLE_WKSP_SIZE (sizeof(S16) * (MaxSeq + 1) + (1u << MaxFSELog) + sizeof(U64))
77*a28cd43dSSascha Wildner #define ZSTD_BUILD_FSE_TABLE_WKSP_SIZE_U32 ((ZSTD_BUILD_FSE_TABLE_WKSP_SIZE + sizeof(U32) - 1) / sizeof(U32))
78*a28cd43dSSascha Wildner 
79*a28cd43dSSascha Wildner typedef struct {
80*a28cd43dSSascha Wildner     ZSTD_seqSymbol LLTable[SEQSYMBOL_TABLE_SIZE(LLFSELog)];    /* Note : Space reserved for FSE Tables */
81*a28cd43dSSascha Wildner     ZSTD_seqSymbol OFTable[SEQSYMBOL_TABLE_SIZE(OffFSELog)];   /* is also used as temporary workspace while building hufTable during DDict creation */
82*a28cd43dSSascha Wildner     ZSTD_seqSymbol MLTable[SEQSYMBOL_TABLE_SIZE(MLFSELog)];    /* and therefore must be at least HUF_DECOMPRESS_WORKSPACE_SIZE large */
83*a28cd43dSSascha Wildner     HUF_DTable hufTable[HUF_DTABLE_SIZE(HufLog)];  /* can accommodate HUF_decompress4X */
84*a28cd43dSSascha Wildner     U32 rep[ZSTD_REP_NUM];
85*a28cd43dSSascha Wildner     U32 workspace[ZSTD_BUILD_FSE_TABLE_WKSP_SIZE_U32];
86*a28cd43dSSascha Wildner } ZSTD_entropyDTables_t;
87*a28cd43dSSascha Wildner 
88*a28cd43dSSascha Wildner typedef enum { ZSTDds_getFrameHeaderSize, ZSTDds_decodeFrameHeader,
89*a28cd43dSSascha Wildner                ZSTDds_decodeBlockHeader, ZSTDds_decompressBlock,
90*a28cd43dSSascha Wildner                ZSTDds_decompressLastBlock, ZSTDds_checkChecksum,
91*a28cd43dSSascha Wildner                ZSTDds_decodeSkippableHeader, ZSTDds_skipFrame } ZSTD_dStage;
92*a28cd43dSSascha Wildner 
93*a28cd43dSSascha Wildner typedef enum { zdss_init=0, zdss_loadHeader,
94*a28cd43dSSascha Wildner                zdss_read, zdss_load, zdss_flush } ZSTD_dStreamStage;
95*a28cd43dSSascha Wildner 
96*a28cd43dSSascha Wildner typedef enum {
97*a28cd43dSSascha Wildner     ZSTD_use_indefinitely = -1,  /* Use the dictionary indefinitely */
98*a28cd43dSSascha Wildner     ZSTD_dont_use = 0,           /* Do not use the dictionary (if one exists free it) */
99*a28cd43dSSascha Wildner     ZSTD_use_once = 1            /* Use the dictionary once and set to ZSTD_dont_use */
100*a28cd43dSSascha Wildner } ZSTD_dictUses_e;
101*a28cd43dSSascha Wildner 
102*a28cd43dSSascha Wildner struct ZSTD_DCtx_s
103*a28cd43dSSascha Wildner {
104*a28cd43dSSascha Wildner     const ZSTD_seqSymbol* LLTptr;
105*a28cd43dSSascha Wildner     const ZSTD_seqSymbol* MLTptr;
106*a28cd43dSSascha Wildner     const ZSTD_seqSymbol* OFTptr;
107*a28cd43dSSascha Wildner     const HUF_DTable* HUFptr;
108*a28cd43dSSascha Wildner     ZSTD_entropyDTables_t entropy;
109*a28cd43dSSascha Wildner     U32 workspace[HUF_DECOMPRESS_WORKSPACE_SIZE_U32];   /* space needed when building huffman tables */
110*a28cd43dSSascha Wildner     const void* previousDstEnd;   /* detect continuity */
111*a28cd43dSSascha Wildner     const void* prefixStart;      /* start of current segment */
112*a28cd43dSSascha Wildner     const void* virtualStart;     /* virtual start of previous segment if it was just before current one */
113*a28cd43dSSascha Wildner     const void* dictEnd;          /* end of previous segment */
114*a28cd43dSSascha Wildner     size_t expected;
115*a28cd43dSSascha Wildner     ZSTD_frameHeader fParams;
116*a28cd43dSSascha Wildner     U64 decodedSize;
117*a28cd43dSSascha Wildner     blockType_e bType;            /* used in ZSTD_decompressContinue(), store blockType between block header decoding and block decompression stages */
118*a28cd43dSSascha Wildner     ZSTD_dStage stage;
119*a28cd43dSSascha Wildner     U32 litEntropy;
120*a28cd43dSSascha Wildner     U32 fseEntropy;
121*a28cd43dSSascha Wildner     XXH64_state_t xxhState;
122*a28cd43dSSascha Wildner     size_t headerSize;
123*a28cd43dSSascha Wildner     ZSTD_format_e format;
124*a28cd43dSSascha Wildner     ZSTD_forceIgnoreChecksum_e forceIgnoreChecksum;   /* User specified: if == 1, will ignore checksums in compressed frame. Default == 0 */
125*a28cd43dSSascha Wildner     U32 validateChecksum;         /* if == 1, will validate checksum. Is == 1 if (fParams.checksumFlag == 1) and (forceIgnoreChecksum == 0). */
126*a28cd43dSSascha Wildner     const BYTE* litPtr;
127*a28cd43dSSascha Wildner     ZSTD_customMem customMem;
128*a28cd43dSSascha Wildner     size_t litSize;
129*a28cd43dSSascha Wildner     size_t rleSize;
130*a28cd43dSSascha Wildner     size_t staticSize;
131*a28cd43dSSascha Wildner     int bmi2;                     /* == 1 if the CPU supports BMI2 and 0 otherwise. CPU support is determined dynamically once per context lifetime. */
132*a28cd43dSSascha Wildner 
133*a28cd43dSSascha Wildner     /* dictionary */
134*a28cd43dSSascha Wildner     ZSTD_DDict* ddictLocal;
135*a28cd43dSSascha Wildner     const ZSTD_DDict* ddict;     /* set by ZSTD_initDStream_usingDDict(), or ZSTD_DCtx_refDDict() */
136*a28cd43dSSascha Wildner     U32 dictID;
137*a28cd43dSSascha Wildner     int ddictIsCold;             /* if == 1 : dictionary is "new" for working context, and presumed "cold" (not in cpu cache) */
138*a28cd43dSSascha Wildner     ZSTD_dictUses_e dictUses;
139*a28cd43dSSascha Wildner 
140*a28cd43dSSascha Wildner     /* streaming */
141*a28cd43dSSascha Wildner     ZSTD_dStreamStage streamStage;
142*a28cd43dSSascha Wildner     char*  inBuff;
143*a28cd43dSSascha Wildner     size_t inBuffSize;
144*a28cd43dSSascha Wildner     size_t inPos;
145*a28cd43dSSascha Wildner     size_t maxWindowSize;
146*a28cd43dSSascha Wildner     char*  outBuff;
147*a28cd43dSSascha Wildner     size_t outBuffSize;
148*a28cd43dSSascha Wildner     size_t outStart;
149*a28cd43dSSascha Wildner     size_t outEnd;
150*a28cd43dSSascha Wildner     size_t lhSize;
151*a28cd43dSSascha Wildner     void* legacyContext;
152*a28cd43dSSascha Wildner     U32 previousLegacyVersion;
153*a28cd43dSSascha Wildner     U32 legacyVersion;
154*a28cd43dSSascha Wildner     U32 hostageByte;
155*a28cd43dSSascha Wildner     int noForwardProgress;
156*a28cd43dSSascha Wildner     ZSTD_bufferMode_e outBufferMode;
157*a28cd43dSSascha Wildner     ZSTD_outBuffer expectedOutBuffer;
158*a28cd43dSSascha Wildner 
159*a28cd43dSSascha Wildner     /* workspace */
160*a28cd43dSSascha Wildner     BYTE litBuffer[ZSTD_BLOCKSIZE_MAX + WILDCOPY_OVERLENGTH];
161*a28cd43dSSascha Wildner     BYTE headerBuffer[ZSTD_FRAMEHEADERSIZE_MAX];
162*a28cd43dSSascha Wildner 
163*a28cd43dSSascha Wildner     size_t oversizedDuration;
164*a28cd43dSSascha Wildner 
165*a28cd43dSSascha Wildner #ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION
166*a28cd43dSSascha Wildner     void const* dictContentBeginForFuzzing;
167*a28cd43dSSascha Wildner     void const* dictContentEndForFuzzing;
168*a28cd43dSSascha Wildner #endif
169*a28cd43dSSascha Wildner };  /* typedef'd to ZSTD_DCtx within "zstd.h" */
170*a28cd43dSSascha Wildner 
171*a28cd43dSSascha Wildner 
172*a28cd43dSSascha Wildner /*-*******************************************************
173*a28cd43dSSascha Wildner  *  Shared internal functions
174*a28cd43dSSascha Wildner  *********************************************************/
175*a28cd43dSSascha Wildner 
176*a28cd43dSSascha Wildner /*! ZSTD_loadDEntropy() :
177*a28cd43dSSascha Wildner  *  dict : must point at beginning of a valid zstd dictionary.
178*a28cd43dSSascha Wildner  * @return : size of dictionary header (size of magic number + dict ID + entropy tables) */
179*a28cd43dSSascha Wildner size_t ZSTD_loadDEntropy(ZSTD_entropyDTables_t* entropy,
180*a28cd43dSSascha Wildner                    const void* const dict, size_t const dictSize);
181*a28cd43dSSascha Wildner 
182*a28cd43dSSascha Wildner /*! ZSTD_checkContinuity() :
183*a28cd43dSSascha Wildner  *  check if next `dst` follows previous position, where decompression ended.
184*a28cd43dSSascha Wildner  *  If yes, do nothing (continue on current segment).
185*a28cd43dSSascha Wildner  *  If not, classify previous segment as "external dictionary", and start a new segment.
186*a28cd43dSSascha Wildner  *  This function cannot fail. */
187*a28cd43dSSascha Wildner void ZSTD_checkContinuity(ZSTD_DCtx* dctx, const void* dst);
188*a28cd43dSSascha Wildner 
189*a28cd43dSSascha Wildner 
190*a28cd43dSSascha Wildner #endif /* ZSTD_DECOMPRESS_INTERNAL_H */
191