src/liblzma/common/index_decoder.c

   1 ///////////////////////////////////////////////////////////////////////////////
   2 //
   3 /// \file       index_decoder.c
   4 /// \brief      Decodes the Index field
   5 //
   6 //  Author:     Lasse Collin
   7 //
   8 //  This file has been put into the public domain.
   9 //  You can do whatever you want with this file.
  10 //
  11 ///////////////////////////////////////////////////////////////////////////////
  12
  13 #include "index.h"
  14 #include "check.h"
  15
  16
  17 struct lzma_coder_s {
  18         enum {
  19                 SEQ_INDICATOR,
  20                 SEQ_COUNT,
  21                 SEQ_MEMUSAGE,
  22                 SEQ_UNPADDED,
  23                 SEQ_UNCOMPRESSED,
  24                 SEQ_PADDING_INIT,
  25                 SEQ_PADDING,
  26                 SEQ_CRC32,
  27         } sequence;
  28
  29         /// Memory usage limit
  30         uint64_t memlimit;
  31
  32         /// Target Index
  33         lzma_index *index;
  34
  35         /// Pointer give by the application, which is set after
  36         /// successful decoding.
  37         lzma_index **index_ptr;
  38
  39         /// Number of Records left to decode.
  40         lzma_vli count;
  41
  42         /// The most recent Unpadded Size field
  43         lzma_vli unpadded_size;
  44
  45         /// The most recent Uncompressed Size field
  46         lzma_vli uncompressed_size;
  47
  48         /// Position in integers
  49         size_t pos;
  50
  51         /// CRC32 of the List of Records field
  52         uint32_t crc32;
  53 };
  54
  55
  56 static lzma_ret
  57 index_decode(lzma_coder *coder, lzma_allocator *allocator,
  58                 const uint8_t *restrict in, size_t *restrict in_pos,
  59                 size_t in_size, uint8_t *restrict out lzma_attribute((unused)),
  60                 size_t *restrict out_pos lzma_attribute((unused)),
  61                 size_t out_size lzma_attribute((unused)),
  62                 lzma_action action lzma_attribute((unused)))
  63 {
  64         // Similar optimization as in index_encoder.c
  65         const size_t in_start = *in_pos;
  66         lzma_ret ret = LZMA_OK;
  67
  68         while (*in_pos < in_size)
  69         switch (coder->sequence) {
  70         case SEQ_INDICATOR:
  71                 // Return LZMA_DATA_ERROR instead of e.g. LZMA_PROG_ERROR or
  72                 // LZMA_FORMAT_ERROR, because a typical usage case for Index
  73                 // decoder is when parsing the Stream backwards. If seeking
  74                 // backward from the Stream Footer gives us something that
  75                 // doesn't begin with Index Indicator, the file is considered
  76                 // corrupt, not "programming error" or "unrecognized file
  77                 // format". One could argue that the application should
  78                 // verify the Index Indicator before trying to decode the
  79                 // Index, but well, I suppose it is simpler this way.
  80                 if (in[(*in_pos)++] != 0x00)
  81                         return LZMA_DATA_ERROR;
  82
  83                 coder->sequence = SEQ_COUNT;
  84                 break;
  85
  86         case SEQ_COUNT:
  87                 ret = lzma_vli_decode(&coder->count, &coder->pos,
  88                                 in, in_pos, in_size);
  89                 if (ret != LZMA_STREAM_END)
  90                         goto out;
  91
  92                 coder->pos = 0;
  93                 coder->sequence = SEQ_MEMUSAGE;
  94
  95         // Fall through
  96
  97         case SEQ_MEMUSAGE:
  98                 if (lzma_index_memusage(1, coder->count) > coder->memlimit) {
  99                         ret = LZMA_MEMLIMIT_ERROR;
 100                         goto out;
 101                 }
 102
 103                 // Tell the Index handling code how many Records this
 104                 // Index has to allow it to allocate memory more efficiently.
 105                 lzma_index_prealloc(coder->index, coder->count);
 106
 107                 ret = LZMA_OK;
 108                 coder->sequence = coder->count == 0
 109                                 ? SEQ_PADDING_INIT : SEQ_UNPADDED;
 110                 break;
 111
 112         case SEQ_UNPADDED:
 113         case SEQ_UNCOMPRESSED: {
 114                 lzma_vli *size = coder->sequence == SEQ_UNPADDED
 115                                 ? &coder->unpadded_size
 116                                 : &coder->uncompressed_size;
 117
 118                 ret = lzma_vli_decode(size, &coder->pos,
 119                                 in, in_pos, in_size);
 120                 if (ret != LZMA_STREAM_END)
 121                         goto out;
 122
 123                 ret = LZMA_OK;
 124                 coder->pos = 0;
 125
 126                 if (coder->sequence == SEQ_UNPADDED) {
 127                         // Validate that encoded Unpadded Size isn't too small
 128                         // or too big.
 129                         if (coder->unpadded_size < UNPADDED_SIZE_MIN
 130                                         || coder->unpadded_size
 131                                                 > UNPADDED_SIZE_MAX)
 132                                 return LZMA_DATA_ERROR;
 133
 134                         coder->sequence = SEQ_UNCOMPRESSED;
 135                 } else {
 136                         // Add the decoded Record to the Index.
 137                         return_if_error(lzma_index_append(
 138                                         coder->index, allocator,
 139                                         coder->unpadded_size,
 140                                         coder->uncompressed_size));
 141
 142                         // Check if this was the last Record.
 143                         coder->sequence = --coder->count == 0
 144                                         ? SEQ_PADDING_INIT
 145                                         : SEQ_UNPADDED;
 146                 }
 147
 148                 break;
 149         }
 150
 151         case SEQ_PADDING_INIT:
 152                 coder->pos = lzma_index_padding_size(coder->index);
 153                 coder->sequence = SEQ_PADDING;
 154
 155         // Fall through
 156
 157         case SEQ_PADDING:
 158                 if (coder->pos > 0) {
 159                         --coder->pos;
 160                         if (in[(*in_pos)++] != 0x00)
 161                                 return LZMA_DATA_ERROR;
 162
 163                         break;
 164                 }
 165
 166                 // Finish the CRC32 calculation.
 167                 coder->crc32 = lzma_crc32(in + in_start,
 168                                 *in_pos - in_start, coder->crc32);
 169
 170                 coder->sequence = SEQ_CRC32;
 171
 172         // Fall through
 173
 174         case SEQ_CRC32:
 175                 do {
 176                         if (*in_pos == in_size)
 177                                 return LZMA_OK;
 178
 179                         if (((coder->crc32 >> (coder->pos * 8)) & 0xFF)
 180                                         != in[(*in_pos)++])
 181                                 return LZMA_DATA_ERROR;
 182
 183                 } while (++coder->pos < 4);
 184
 185                 // Decoding was successful, now we can let the application
 186                 // see the decoded Index.
 187                 *coder->index_ptr = coder->index;
 188
 189                 // Make index NULL so we don't free it unintentionally.
 190                 coder->index = NULL;
 191
 192                 return LZMA_STREAM_END;
 193
 194         default:
 195                 assert(0);
 196                 return LZMA_PROG_ERROR;
 197         }
 198
 199 out:
 200         // Update the CRC32,
 201         coder->crc32 = lzma_crc32(in + in_start,
 202                         *in_pos - in_start, coder->crc32);
 203
 204         return ret;
 205 }
 206
 207
 208 static void
 209 index_decoder_end(lzma_coder *coder, lzma_allocator *allocator)
 210 {
 211         lzma_index_end(coder->index, allocator);
 212         lzma_free(coder, allocator);
 213         return;
 214 }
 215
 216
 217 static lzma_ret
 218 index_decoder_memconfig(lzma_coder *coder, uint64_t *memusage,
 219                 uint64_t *old_memlimit, uint64_t new_memlimit)
 220 {
 221         *memusage = lzma_index_memusage(1, coder->count);
 222
 223         if (new_memlimit != 0 && new_memlimit < *memusage)
 224                 return LZMA_MEMLIMIT_ERROR;
 225
 226         *old_memlimit = coder->memlimit;
 227         coder->memlimit = new_memlimit;
 228
 229         return LZMA_OK;
 230 }
 231
 232
 233 static lzma_ret
 234 index_decoder_reset(lzma_coder *coder, lzma_allocator *allocator,
 235                 lzma_index **i, uint64_t memlimit)
 236 {
 237         // Remember the pointer given by the application. We will set it
 238         // to point to the decoded Index only if decoding is successful.
 239         // Before that, keep it NULL so that applications can always safely
 240         // pass it to lzma_index_end() no matter did decoding succeed or not.
 241         coder->index_ptr = i;
 242         *i = NULL;
 243
 244         // We always allocate a new lzma_index.
 245         coder->index = lzma_index_init(allocator);
 246         if (coder->index == NULL)
 247                 return LZMA_MEM_ERROR;
 248
 249         // Initialize the rest.
 250         coder->sequence = SEQ_INDICATOR;
 251         coder->memlimit = memlimit;
 252         coder->count = 0; // Needs to be initialized due to _memconfig().
 253         coder->pos = 0;
 254         coder->crc32 = 0;
 255
 256         return LZMA_OK;
 257 }
 258
 259
 260 static lzma_ret
 261 index_decoder_init(lzma_next_coder *next, lzma_allocator *allocator,
 262                 lzma_index **i, uint64_t memlimit)
 263 {
 264         lzma_next_coder_init(&index_decoder_init, next, allocator);
 265
 266         if (i == NULL || memlimit == 0)
 267                 return LZMA_PROG_ERROR;
 268
 269         if (next->coder == NULL) {
 270                 next->coder = lzma_alloc(sizeof(lzma_coder), allocator);
 271                 if (next->coder == NULL)
 272                         return LZMA_MEM_ERROR;
 273
 274                 next->code = &index_decode;
 275                 next->end = &index_decoder_end;
 276                 next->memconfig = &index_decoder_memconfig;
 277                 next->coder->index = NULL;
 278         } else {
 279                 lzma_index_end(next->coder->index, allocator);
 280         }
 281
 282         return index_decoder_reset(next->coder, allocator, i, memlimit);
 283 }
 284
 285
 286 extern LZMA_API(lzma_ret)
 287 lzma_index_decoder(lzma_stream *strm, lzma_index **i, uint64_t memlimit)
 288 {
 289         lzma_next_strm_init(index_decoder_init, strm, i, memlimit);
 290
 291         strm->internal->supported_actions[LZMA_RUN] = true;
 292
 293         return LZMA_OK;
 294 }
 295
 296
 297 extern LZMA_API(lzma_ret)
 298 lzma_index_buffer_decode(
 299                 lzma_index **i, uint64_t *memlimit, lzma_allocator *allocator,
 300                 const uint8_t *in, size_t *in_pos, size_t in_size)
 301 {
 302         // Sanity checks
 303         if (i == NULL || memlimit == NULL
 304                         || in == NULL || in_pos == NULL || *in_pos > in_size)
 305                 return LZMA_PROG_ERROR;
 306
 307         // Initialize the decoder.
 308         lzma_coder coder;
 309         return_if_error(index_decoder_reset(&coder, allocator, i, *memlimit));
 310
 311         // Store the input start position so that we can restore it in case
 312         // of an error.
 313         const size_t in_start = *in_pos;
 314
 315         // Do the actual decoding.
 316         lzma_ret ret = index_decode(&coder, allocator, in, in_pos, in_size,
 317                         NULL, NULL, 0, LZMA_RUN);
 318
 319         if (ret == LZMA_STREAM_END) {
 320                 ret = LZMA_OK;
 321         } else {
 322                 // Something went wrong, free the Index structure and restore
 323                 // the input position.
 324                 lzma_index_end(coder.index, allocator);
 325                 *in_pos = in_start;
 326
 327                 if (ret == LZMA_OK) {
 328                         // The input is truncated or otherwise corrupt.
 329                         // Use LZMA_DATA_ERROR instead of LZMA_BUF_ERROR
 330                         // like lzma_vli_decode() does in single-call mode.
 331                         ret = LZMA_DATA_ERROR;
 332
 333                 } else if (ret == LZMA_MEMLIMIT_ERROR) {
 334                         // Tell the caller how much memory would have
 335                         // been needed.
 336                         *memlimit = lzma_index_memusage(1, coder.count);
 337                 }
 338         }
 339
 340         return ret;
 341 }