Julius Werner | 027b728 | 2015-10-06 20:03:53 -0700 | [diff] [blame] | 1 | /* |
| 2 | LZ4 - Fast LZ compression algorithm |
| 3 | Copyright (C) 2011-2015, Yann Collet. |
| 4 | |
| 5 | SPDX-License-Identifier: BSD-2-Clause |
| 6 | |
| 7 | You can contact the author at : |
| 8 | - LZ4 source repository : https://github.com/Cyan4973/lz4 |
| 9 | - LZ4 public forum : https://groups.google.com/forum/#!forum/lz4c |
| 10 | */ |
| 11 | |
| 12 | |
| 13 | /************************************** |
| 14 | * Reading and writing into memory |
| 15 | **************************************/ |
| 16 | |
| 17 | /* customized version of memcpy, which may overwrite up to 7 bytes beyond dstEnd */ |
| 18 | static void LZ4_wildCopy(void* dstPtr, const void* srcPtr, void* dstEnd) |
| 19 | { |
| 20 | BYTE* d = (BYTE*)dstPtr; |
| 21 | const BYTE* s = (const BYTE*)srcPtr; |
| 22 | BYTE* e = (BYTE*)dstEnd; |
| 23 | do { LZ4_copy8(d,s); d+=8; s+=8; } while (d<e); |
| 24 | } |
| 25 | |
| 26 | |
| 27 | /************************************** |
| 28 | * Common Constants |
| 29 | **************************************/ |
| 30 | #define MINMATCH 4 |
| 31 | |
| 32 | #define COPYLENGTH 8 |
| 33 | #define LASTLITERALS 5 |
| 34 | #define MFLIMIT (COPYLENGTH+MINMATCH) |
| 35 | static const int LZ4_minLength = (MFLIMIT+1); |
| 36 | |
| 37 | #define KB *(1 <<10) |
| 38 | #define MB *(1 <<20) |
| 39 | #define GB *(1U<<30) |
| 40 | |
| 41 | #define MAXD_LOG 16 |
| 42 | #define MAX_DISTANCE ((1 << MAXD_LOG) - 1) |
| 43 | |
| 44 | #define ML_BITS 4 |
| 45 | #define ML_MASK ((1U<<ML_BITS)-1) |
| 46 | #define RUN_BITS (8-ML_BITS) |
| 47 | #define RUN_MASK ((1U<<RUN_BITS)-1) |
| 48 | |
| 49 | |
| 50 | /************************************** |
| 51 | * Local Structures and types |
| 52 | **************************************/ |
| 53 | typedef enum { noDict = 0, withPrefix64k, usingExtDict } dict_directive; |
| 54 | typedef enum { endOnOutputSize = 0, endOnInputSize = 1 } endCondition_directive; |
| 55 | typedef enum { full = 0, partial = 1 } earlyEnd_directive; |
| 56 | |
| 57 | |
| 58 | |
| 59 | /******************************* |
| 60 | * Decompression functions |
| 61 | *******************************/ |
| 62 | /* |
| 63 | * This generic decompression function cover all use cases. |
| 64 | * It shall be instantiated several times, using different sets of directives |
| 65 | * Note that it is essential this generic function is really inlined, |
| 66 | * in order to remove useless branches during compilation optimization. |
| 67 | */ |
| 68 | FORCE_INLINE int LZ4_decompress_generic( |
| 69 | const char* const source, |
| 70 | char* const dest, |
| 71 | int inputSize, |
| 72 | int outputSize, /* If endOnInput==endOnInputSize, this value is the max size of Output Buffer. */ |
| 73 | |
| 74 | int endOnInput, /* endOnOutputSize, endOnInputSize */ |
| 75 | int partialDecoding, /* full, partial */ |
| 76 | int targetOutputSize, /* only used if partialDecoding==partial */ |
| 77 | int dict, /* noDict, withPrefix64k, usingExtDict */ |
| 78 | const BYTE* const lowPrefix, /* == dest if dict == noDict */ |
| 79 | const BYTE* const dictStart, /* only if dict==usingExtDict */ |
| 80 | const size_t dictSize /* note : = 0 if noDict */ |
| 81 | ) |
| 82 | { |
| 83 | /* Local Variables */ |
| 84 | const BYTE* ip = (const BYTE*) source; |
| 85 | const BYTE* const iend = ip + inputSize; |
| 86 | |
| 87 | BYTE* op = (BYTE*) dest; |
| 88 | BYTE* const oend = op + outputSize; |
| 89 | BYTE* cpy; |
| 90 | BYTE* oexit = op + targetOutputSize; |
| 91 | const BYTE* const lowLimit = lowPrefix - dictSize; |
| 92 | |
| 93 | const BYTE* const dictEnd = (const BYTE*)dictStart + dictSize; |
| 94 | const size_t dec32table[] = {4, 1, 2, 1, 4, 4, 4, 4}; |
| 95 | const size_t dec64table[] = {0, 0, 0, (size_t)-1, 0, 1, 2, 3}; |
| 96 | |
| 97 | const int safeDecode = (endOnInput==endOnInputSize); |
| 98 | const int checkOffset = ((safeDecode) && (dictSize < (int)(64 KB))); |
| 99 | |
| 100 | |
| 101 | /* Special cases */ |
| 102 | if ((partialDecoding) && (oexit> oend-MFLIMIT)) oexit = oend-MFLIMIT; /* targetOutputSize too high => decode everything */ |
| 103 | if ((endOnInput) && (unlikely(outputSize==0))) return ((inputSize==1) && (*ip==0)) ? 0 : -1; /* Empty output buffer */ |
| 104 | if ((!endOnInput) && (unlikely(outputSize==0))) return (*ip==0?1:-1); |
| 105 | |
| 106 | |
| 107 | /* Main Loop */ |
| 108 | while (1) |
| 109 | { |
| 110 | unsigned token; |
| 111 | size_t length; |
| 112 | const BYTE* match; |
| 113 | |
| 114 | /* get literal length */ |
| 115 | token = *ip++; |
| 116 | if ((length=(token>>ML_BITS)) == RUN_MASK) |
| 117 | { |
| 118 | unsigned s; |
| 119 | do |
| 120 | { |
| 121 | s = *ip++; |
| 122 | length += s; |
| 123 | } |
| 124 | while (likely((endOnInput)?ip<iend-RUN_MASK:1) && (s==255)); |
| 125 | if ((safeDecode) && unlikely((size_t)(op+length)<(size_t)(op))) goto _output_error; /* overflow detection */ |
| 126 | if ((safeDecode) && unlikely((size_t)(ip+length)<(size_t)(ip))) goto _output_error; /* overflow detection */ |
| 127 | } |
| 128 | |
| 129 | /* copy literals */ |
| 130 | cpy = op+length; |
| 131 | if (((endOnInput) && ((cpy>(partialDecoding?oexit:oend-MFLIMIT)) || (ip+length>iend-(2+1+LASTLITERALS))) ) |
| 132 | || ((!endOnInput) && (cpy>oend-COPYLENGTH))) |
| 133 | { |
| 134 | if (partialDecoding) |
| 135 | { |
| 136 | if (cpy > oend) goto _output_error; /* Error : write attempt beyond end of output buffer */ |
| 137 | if ((endOnInput) && (ip+length > iend)) goto _output_error; /* Error : read attempt beyond end of input buffer */ |
| 138 | } |
| 139 | else |
| 140 | { |
| 141 | if ((!endOnInput) && (cpy != oend)) goto _output_error; /* Error : block decoding must stop exactly there */ |
| 142 | if ((endOnInput) && ((ip+length != iend) || (cpy > oend))) goto _output_error; /* Error : input must be consumed */ |
| 143 | } |
| 144 | memcpy(op, ip, length); |
| 145 | ip += length; |
| 146 | op += length; |
| 147 | break; /* Necessarily EOF, due to parsing restrictions */ |
| 148 | } |
| 149 | LZ4_wildCopy(op, ip, cpy); |
| 150 | ip += length; op = cpy; |
| 151 | |
| 152 | /* get offset */ |
| 153 | match = cpy - LZ4_readLE16(ip); ip+=2; |
| 154 | if ((checkOffset) && (unlikely(match < lowLimit))) goto _output_error; /* Error : offset outside destination buffer */ |
| 155 | |
| 156 | /* get matchlength */ |
| 157 | length = token & ML_MASK; |
| 158 | if (length == ML_MASK) |
| 159 | { |
| 160 | unsigned s; |
| 161 | do |
| 162 | { |
| 163 | if ((endOnInput) && (ip > iend-LASTLITERALS)) goto _output_error; |
| 164 | s = *ip++; |
| 165 | length += s; |
| 166 | } while (s==255); |
| 167 | if ((safeDecode) && unlikely((size_t)(op+length)<(size_t)op)) goto _output_error; /* overflow detection */ |
| 168 | } |
| 169 | length += MINMATCH; |
| 170 | |
| 171 | /* check external dictionary */ |
| 172 | if ((dict==usingExtDict) && (match < lowPrefix)) |
| 173 | { |
| 174 | if (unlikely(op+length > oend-LASTLITERALS)) goto _output_error; /* doesn't respect parsing restriction */ |
| 175 | |
| 176 | if (length <= (size_t)(lowPrefix-match)) |
| 177 | { |
| 178 | /* match can be copied as a single segment from external dictionary */ |
| 179 | match = dictEnd - (lowPrefix-match); |
| 180 | memmove(op, match, length); op += length; |
| 181 | } |
| 182 | else |
| 183 | { |
| 184 | /* match encompass external dictionary and current segment */ |
| 185 | size_t copySize = (size_t)(lowPrefix-match); |
| 186 | memcpy(op, dictEnd - copySize, copySize); |
| 187 | op += copySize; |
| 188 | copySize = length - copySize; |
| 189 | if (copySize > (size_t)(op-lowPrefix)) /* overlap within current segment */ |
| 190 | { |
| 191 | BYTE* const endOfMatch = op + copySize; |
| 192 | const BYTE* copyFrom = lowPrefix; |
| 193 | while (op < endOfMatch) *op++ = *copyFrom++; |
| 194 | } |
| 195 | else |
| 196 | { |
| 197 | memcpy(op, lowPrefix, copySize); |
| 198 | op += copySize; |
| 199 | } |
| 200 | } |
| 201 | continue; |
| 202 | } |
| 203 | |
| 204 | /* copy repeated sequence */ |
| 205 | cpy = op + length; |
| 206 | if (unlikely((op-match)<8)) |
| 207 | { |
| 208 | const size_t dec64 = dec64table[op-match]; |
| 209 | op[0] = match[0]; |
| 210 | op[1] = match[1]; |
| 211 | op[2] = match[2]; |
| 212 | op[3] = match[3]; |
| 213 | match += dec32table[op-match]; |
| 214 | LZ4_copy4(op+4, match); |
| 215 | op += 8; match -= dec64; |
| 216 | } else { LZ4_copy8(op, match); op+=8; match+=8; } |
| 217 | |
| 218 | if (unlikely(cpy>oend-12)) |
| 219 | { |
| 220 | if (cpy > oend-LASTLITERALS) goto _output_error; /* Error : last LASTLITERALS bytes must be literals */ |
| 221 | if (op < oend-8) |
| 222 | { |
| 223 | LZ4_wildCopy(op, match, oend-8); |
| 224 | match += (oend-8) - op; |
| 225 | op = oend-8; |
| 226 | } |
| 227 | while (op<cpy) *op++ = *match++; |
| 228 | } |
| 229 | else |
| 230 | LZ4_wildCopy(op, match, cpy); |
| 231 | op=cpy; /* correction */ |
| 232 | } |
| 233 | |
| 234 | /* end of decoding */ |
| 235 | if (endOnInput) |
| 236 | return (int) (((char*)op)-dest); /* Nb of output bytes decoded */ |
| 237 | else |
| 238 | return (int) (((const char*)ip)-source); /* Nb of input bytes read */ |
| 239 | |
| 240 | /* Overflow error detected */ |
| 241 | _output_error: |
| 242 | return (int) (-(((const char*)ip)-source))-1; |
| 243 | } |