[3320] | 1 | /*
|
---|
| 2 | * .xz Stream decoder
|
---|
| 3 | *
|
---|
| 4 | * Author: Lasse Collin <lasse.collin@tukaani.org>
|
---|
| 5 | *
|
---|
| 6 | * This file has been put into the public domain.
|
---|
| 7 | * You can do whatever you want with this file.
|
---|
| 8 | */
|
---|
| 9 |
|
---|
| 10 | #include "xz_private.h"
|
---|
| 11 | #include "xz_stream.h"
|
---|
| 12 |
|
---|
| 13 | /* Hash used to validate the Index field */
|
---|
| 14 | struct xz_dec_hash {
|
---|
| 15 | vli_type unpadded;
|
---|
| 16 | vli_type uncompressed;
|
---|
| 17 | uint32_t crc32;
|
---|
| 18 | };
|
---|
| 19 |
|
---|
| 20 | struct xz_dec {
|
---|
| 21 | /* Position in dec_main() */
|
---|
| 22 | enum {
|
---|
| 23 | SEQ_STREAM_HEADER,
|
---|
| 24 | SEQ_BLOCK_START,
|
---|
| 25 | SEQ_BLOCK_HEADER,
|
---|
| 26 | SEQ_BLOCK_UNCOMPRESS,
|
---|
| 27 | SEQ_BLOCK_PADDING,
|
---|
| 28 | SEQ_BLOCK_CHECK,
|
---|
| 29 | SEQ_INDEX,
|
---|
| 30 | SEQ_INDEX_PADDING,
|
---|
| 31 | SEQ_INDEX_CRC32,
|
---|
| 32 | SEQ_STREAM_FOOTER
|
---|
| 33 | } sequence;
|
---|
| 34 |
|
---|
| 35 | /* Position in variable-length integers and Check fields */
|
---|
| 36 | uint32_t pos;
|
---|
| 37 |
|
---|
| 38 | /* Variable-length integer decoded by dec_vli() */
|
---|
| 39 | vli_type vli;
|
---|
| 40 |
|
---|
| 41 | /* Saved in_pos and out_pos */
|
---|
| 42 | size_t in_start;
|
---|
| 43 | size_t out_start;
|
---|
| 44 |
|
---|
| 45 | /* CRC32 value in Block or Index */
|
---|
| 46 | uint32_t crc32;
|
---|
| 47 |
|
---|
| 48 | /* Type of the integrity check calculated from uncompressed data */
|
---|
| 49 | enum xz_check check_type;
|
---|
| 50 |
|
---|
| 51 | /* Operation mode */
|
---|
| 52 | enum xz_mode mode;
|
---|
| 53 |
|
---|
| 54 | /*
|
---|
| 55 | * True if the next call to xz_dec_run() is allowed to return
|
---|
| 56 | * XZ_BUF_ERROR.
|
---|
| 57 | */
|
---|
| 58 | bool allow_buf_error;
|
---|
| 59 |
|
---|
| 60 | /* Information stored in Block Header */
|
---|
| 61 | struct {
|
---|
| 62 | /*
|
---|
| 63 | * Value stored in the Compressed Size field, or
|
---|
| 64 | * VLI_UNKNOWN if Compressed Size is not present.
|
---|
| 65 | */
|
---|
| 66 | vli_type compressed;
|
---|
| 67 |
|
---|
| 68 | /*
|
---|
| 69 | * Value stored in the Uncompressed Size field, or
|
---|
| 70 | * VLI_UNKNOWN if Uncompressed Size is not present.
|
---|
| 71 | */
|
---|
| 72 | vli_type uncompressed;
|
---|
| 73 |
|
---|
| 74 | /* Size of the Block Header field */
|
---|
| 75 | uint32_t size;
|
---|
| 76 | } block_header;
|
---|
| 77 |
|
---|
| 78 | /* Information collected when decoding Blocks */
|
---|
| 79 | struct {
|
---|
| 80 | /* Observed compressed size of the current Block */
|
---|
| 81 | vli_type compressed;
|
---|
| 82 |
|
---|
| 83 | /* Observed uncompressed size of the current Block */
|
---|
| 84 | vli_type uncompressed;
|
---|
| 85 |
|
---|
| 86 | /* Number of Blocks decoded so far */
|
---|
| 87 | vli_type count;
|
---|
| 88 |
|
---|
| 89 | /*
|
---|
| 90 | * Hash calculated from the Block sizes. This is used to
|
---|
| 91 | * validate the Index field.
|
---|
| 92 | */
|
---|
| 93 | struct xz_dec_hash hash;
|
---|
| 94 | } block;
|
---|
| 95 |
|
---|
| 96 | /* Variables needed when verifying the Index field */
|
---|
| 97 | struct {
|
---|
| 98 | /* Position in dec_index() */
|
---|
| 99 | enum {
|
---|
| 100 | SEQ_INDEX_COUNT,
|
---|
| 101 | SEQ_INDEX_UNPADDED,
|
---|
| 102 | SEQ_INDEX_UNCOMPRESSED
|
---|
| 103 | } sequence;
|
---|
| 104 |
|
---|
| 105 | /* Size of the Index in bytes */
|
---|
| 106 | vli_type size;
|
---|
| 107 |
|
---|
| 108 | /* Number of Records (matches block.count in valid files) */
|
---|
| 109 | vli_type count;
|
---|
| 110 |
|
---|
| 111 | /*
|
---|
| 112 | * Hash calculated from the Records (matches block.hash in
|
---|
| 113 | * valid files).
|
---|
| 114 | */
|
---|
| 115 | struct xz_dec_hash hash;
|
---|
| 116 | } index;
|
---|
| 117 |
|
---|
| 118 | /*
|
---|
| 119 | * Temporary buffer needed to hold Stream Header, Block Header,
|
---|
| 120 | * and Stream Footer. The Block Header is the biggest (1 KiB)
|
---|
| 121 | * so we reserve space according to that. buf[] has to be aligned
|
---|
| 122 | * to a multiple of four bytes; the size_t variables before it
|
---|
| 123 | * should guarantee this.
|
---|
| 124 | */
|
---|
| 125 | struct {
|
---|
| 126 | size_t pos;
|
---|
| 127 | size_t size;
|
---|
| 128 | uint8_t buf[1024];
|
---|
| 129 | } temp;
|
---|
| 130 |
|
---|
| 131 | struct xz_dec_lzma2 *lzma2;
|
---|
| 132 |
|
---|
| 133 | #ifdef XZ_DEC_BCJ
|
---|
| 134 | struct xz_dec_bcj *bcj;
|
---|
| 135 | bool bcj_active;
|
---|
| 136 | #endif
|
---|
| 137 | };
|
---|
| 138 |
|
---|
| 139 | #ifdef XZ_DEC_ANY_CHECK
|
---|
| 140 | /* Sizes of the Check field with different Check IDs */
|
---|
| 141 | static const uint8_t check_sizes[16] = {
|
---|
| 142 | 0,
|
---|
| 143 | 4, 4, 4,
|
---|
| 144 | 8, 8, 8,
|
---|
| 145 | 16, 16, 16,
|
---|
| 146 | 32, 32, 32,
|
---|
| 147 | 64, 64, 64
|
---|
| 148 | };
|
---|
| 149 | #endif
|
---|
| 150 |
|
---|
| 151 | /*
|
---|
| 152 | * Fill s->temp by copying data starting from b->in[b->in_pos]. Caller
|
---|
| 153 | * must have set s->temp.pos to indicate how much data we are supposed
|
---|
| 154 | * to copy into s->temp.buf. Return true once s->temp.pos has reached
|
---|
| 155 | * s->temp.size.
|
---|
| 156 | */
|
---|
| 157 | static bool XZ_FUNC fill_temp(struct xz_dec *s, struct xz_buf *b)
|
---|
| 158 | {
|
---|
| 159 | size_t copy_size = min_t(size_t,
|
---|
| 160 | b->in_size - b->in_pos, s->temp.size - s->temp.pos);
|
---|
| 161 |
|
---|
| 162 | memcpy(s->temp.buf + s->temp.pos, b->in + b->in_pos, copy_size);
|
---|
| 163 | b->in_pos += copy_size;
|
---|
| 164 | s->temp.pos += copy_size;
|
---|
| 165 |
|
---|
| 166 | if (s->temp.pos == s->temp.size) {
|
---|
| 167 | s->temp.pos = 0;
|
---|
| 168 | return true;
|
---|
| 169 | }
|
---|
| 170 |
|
---|
| 171 | return false;
|
---|
| 172 | }
|
---|
| 173 |
|
---|
| 174 | /* Decode a variable-length integer (little-endian base-128 encoding) */
|
---|
| 175 | static enum xz_ret XZ_FUNC dec_vli(struct xz_dec *s,
|
---|
| 176 | const uint8_t *in, size_t *in_pos, size_t in_size)
|
---|
| 177 | {
|
---|
| 178 | uint8_t byte;
|
---|
| 179 |
|
---|
| 180 | if (s->pos == 0)
|
---|
| 181 | s->vli = 0;
|
---|
| 182 |
|
---|
| 183 | while (*in_pos < in_size) {
|
---|
| 184 | byte = in[*in_pos];
|
---|
| 185 | ++*in_pos;
|
---|
| 186 |
|
---|
| 187 | s->vli |= (vli_type)(byte & 0x7F) << s->pos;
|
---|
| 188 |
|
---|
| 189 | if ((byte & 0x80) == 0) {
|
---|
| 190 | /* Don't allow non-minimal encodings. */
|
---|
| 191 | if (byte == 0 && s->pos != 0)
|
---|
| 192 | return XZ_DATA_ERROR;
|
---|
| 193 |
|
---|
| 194 | s->pos = 0;
|
---|
| 195 | return XZ_STREAM_END;
|
---|
| 196 | }
|
---|
| 197 |
|
---|
| 198 | s->pos += 7;
|
---|
| 199 | if (s->pos == 7 * VLI_BYTES_MAX)
|
---|
| 200 | return XZ_DATA_ERROR;
|
---|
| 201 | }
|
---|
| 202 |
|
---|
| 203 | return XZ_OK;
|
---|
| 204 | }
|
---|
| 205 |
|
---|
| 206 | /*
|
---|
| 207 | * Decode the Compressed Data field from a Block. Update and validate
|
---|
| 208 | * the observed compressed and uncompressed sizes of the Block so that
|
---|
| 209 | * they don't exceed the values possibly stored in the Block Header
|
---|
| 210 | * (validation assumes that no integer overflow occurs, since vli_type
|
---|
| 211 | * is normally uint64_t). Update the CRC32 if presence of the CRC32
|
---|
| 212 | * field was indicated in Stream Header.
|
---|
| 213 | *
|
---|
| 214 | * Once the decoding is finished, validate that the observed sizes match
|
---|
| 215 | * the sizes possibly stored in the Block Header. Update the hash and
|
---|
| 216 | * Block count, which are later used to validate the Index field.
|
---|
| 217 | */
|
---|
| 218 | static enum xz_ret XZ_FUNC dec_block(struct xz_dec *s, struct xz_buf *b)
|
---|
| 219 | {
|
---|
| 220 | enum xz_ret ret;
|
---|
| 221 |
|
---|
| 222 | s->in_start = b->in_pos;
|
---|
| 223 | s->out_start = b->out_pos;
|
---|
| 224 |
|
---|
| 225 | #ifdef XZ_DEC_BCJ
|
---|
| 226 | if (s->bcj_active)
|
---|
| 227 | ret = xz_dec_bcj_run(s->bcj, s->lzma2, b);
|
---|
| 228 | else
|
---|
| 229 | #endif
|
---|
| 230 | ret = xz_dec_lzma2_run(s->lzma2, b);
|
---|
| 231 |
|
---|
| 232 | s->block.compressed += b->in_pos - s->in_start;
|
---|
| 233 | s->block.uncompressed += b->out_pos - s->out_start;
|
---|
| 234 |
|
---|
| 235 | /*
|
---|
| 236 | * There is no need to separately check for VLI_UNKNOWN, since
|
---|
| 237 | * the observed sizes are always smaller than VLI_UNKNOWN.
|
---|
| 238 | */
|
---|
| 239 | if (s->block.compressed > s->block_header.compressed
|
---|
| 240 | || s->block.uncompressed
|
---|
| 241 | > s->block_header.uncompressed)
|
---|
| 242 | return XZ_DATA_ERROR;
|
---|
| 243 |
|
---|
| 244 | if (s->check_type == XZ_CHECK_CRC32)
|
---|
| 245 | s->crc32 = xz_crc32(b->out + s->out_start,
|
---|
| 246 | b->out_pos - s->out_start, s->crc32);
|
---|
| 247 |
|
---|
| 248 | if (ret == XZ_STREAM_END) {
|
---|
| 249 | if (s->block_header.compressed != VLI_UNKNOWN
|
---|
| 250 | && s->block_header.compressed
|
---|
| 251 | != s->block.compressed)
|
---|
| 252 | return XZ_DATA_ERROR;
|
---|
| 253 |
|
---|
| 254 | if (s->block_header.uncompressed != VLI_UNKNOWN
|
---|
| 255 | && s->block_header.uncompressed
|
---|
| 256 | != s->block.uncompressed)
|
---|
| 257 | return XZ_DATA_ERROR;
|
---|
| 258 |
|
---|
| 259 | s->block.hash.unpadded += s->block_header.size
|
---|
| 260 | + s->block.compressed;
|
---|
| 261 |
|
---|
| 262 | #ifdef XZ_DEC_ANY_CHECK
|
---|
| 263 | s->block.hash.unpadded += check_sizes[s->check_type];
|
---|
| 264 | #else
|
---|
| 265 | if (s->check_type == XZ_CHECK_CRC32)
|
---|
| 266 | s->block.hash.unpadded += 4;
|
---|
| 267 | #endif
|
---|
| 268 |
|
---|
| 269 | s->block.hash.uncompressed += s->block.uncompressed;
|
---|
| 270 | s->block.hash.crc32 = xz_crc32(
|
---|
| 271 | (const uint8_t *)&s->block.hash,
|
---|
| 272 | sizeof(s->block.hash), s->block.hash.crc32);
|
---|
| 273 |
|
---|
| 274 | ++s->block.count;
|
---|
| 275 | }
|
---|
| 276 |
|
---|
| 277 | return ret;
|
---|
| 278 | }
|
---|
| 279 |
|
---|
| 280 | /* Update the Index size and the CRC32 value. */
|
---|
| 281 | static void XZ_FUNC index_update(struct xz_dec *s, const struct xz_buf *b)
|
---|
| 282 | {
|
---|
| 283 | size_t in_used = b->in_pos - s->in_start;
|
---|
| 284 | s->index.size += in_used;
|
---|
| 285 | s->crc32 = xz_crc32(b->in + s->in_start, in_used, s->crc32);
|
---|
| 286 | }
|
---|
| 287 |
|
---|
| 288 | /*
|
---|
| 289 | * Decode the Number of Records, Unpadded Size, and Uncompressed Size
|
---|
| 290 | * fields from the Index field. That is, Index Padding and CRC32 are not
|
---|
| 291 | * decoded by this function.
|
---|
| 292 | *
|
---|
| 293 | * This can return XZ_OK (more input needed), XZ_STREAM_END (everything
|
---|
| 294 | * successfully decoded), or XZ_DATA_ERROR (input is corrupt).
|
---|
| 295 | */
|
---|
| 296 | static enum xz_ret XZ_FUNC dec_index(struct xz_dec *s, struct xz_buf *b)
|
---|
| 297 | {
|
---|
| 298 | enum xz_ret ret;
|
---|
| 299 |
|
---|
| 300 | do {
|
---|
| 301 | ret = dec_vli(s, b->in, &b->in_pos, b->in_size);
|
---|
| 302 | if (ret != XZ_STREAM_END) {
|
---|
| 303 | index_update(s, b);
|
---|
| 304 | return ret;
|
---|
| 305 | }
|
---|
| 306 |
|
---|
| 307 | switch (s->index.sequence) {
|
---|
| 308 | case SEQ_INDEX_COUNT:
|
---|
| 309 | s->index.count = s->vli;
|
---|
| 310 |
|
---|
| 311 | /*
|
---|
| 312 | * Validate that the Number of Records field
|
---|
| 313 | * indicates the same number of Records as
|
---|
| 314 | * there were Blocks in the Stream.
|
---|
| 315 | */
|
---|
| 316 | if (s->index.count != s->block.count)
|
---|
| 317 | return XZ_DATA_ERROR;
|
---|
| 318 |
|
---|
| 319 | s->index.sequence = SEQ_INDEX_UNPADDED;
|
---|
| 320 | break;
|
---|
| 321 |
|
---|
| 322 | case SEQ_INDEX_UNPADDED:
|
---|
| 323 | s->index.hash.unpadded += s->vli;
|
---|
| 324 | s->index.sequence = SEQ_INDEX_UNCOMPRESSED;
|
---|
| 325 | break;
|
---|
| 326 |
|
---|
| 327 | case SEQ_INDEX_UNCOMPRESSED:
|
---|
| 328 | s->index.hash.uncompressed += s->vli;
|
---|
| 329 | s->index.hash.crc32 = xz_crc32(
|
---|
| 330 | (const uint8_t *)&s->index.hash,
|
---|
| 331 | sizeof(s->index.hash),
|
---|
| 332 | s->index.hash.crc32);
|
---|
| 333 | --s->index.count;
|
---|
| 334 | s->index.sequence = SEQ_INDEX_UNPADDED;
|
---|
| 335 | break;
|
---|
| 336 | }
|
---|
| 337 | } while (s->index.count > 0);
|
---|
| 338 |
|
---|
| 339 | return XZ_STREAM_END;
|
---|
| 340 | }
|
---|
| 341 |
|
---|
| 342 | /*
|
---|
| 343 | * Validate that the next four input bytes match the value of s->crc32.
|
---|
| 344 | * s->pos must be zero when starting to validate the first byte.
|
---|
| 345 | */
|
---|
| 346 | static enum xz_ret XZ_FUNC crc32_validate(struct xz_dec *s, struct xz_buf *b)
|
---|
| 347 | {
|
---|
| 348 | do {
|
---|
| 349 | if (b->in_pos == b->in_size)
|
---|
| 350 | return XZ_OK;
|
---|
| 351 |
|
---|
| 352 | if (((s->crc32 >> s->pos) & 0xFF) != b->in[b->in_pos++])
|
---|
| 353 | return XZ_DATA_ERROR;
|
---|
| 354 |
|
---|
| 355 | s->pos += 8;
|
---|
| 356 |
|
---|
| 357 | } while (s->pos < 32);
|
---|
| 358 |
|
---|
| 359 | s->crc32 = 0;
|
---|
| 360 | s->pos = 0;
|
---|
| 361 |
|
---|
| 362 | return XZ_STREAM_END;
|
---|
| 363 | }
|
---|
| 364 |
|
---|
| 365 | #ifdef XZ_DEC_ANY_CHECK
|
---|
| 366 | /*
|
---|
| 367 | * Skip over the Check field when the Check ID is not supported.
|
---|
| 368 | * Returns true once the whole Check field has been skipped over.
|
---|
| 369 | */
|
---|
| 370 | static bool XZ_FUNC check_skip(struct xz_dec *s, struct xz_buf *b)
|
---|
| 371 | {
|
---|
| 372 | while (s->pos < check_sizes[s->check_type]) {
|
---|
| 373 | if (b->in_pos == b->in_size)
|
---|
| 374 | return false;
|
---|
| 375 |
|
---|
| 376 | ++b->in_pos;
|
---|
| 377 | ++s->pos;
|
---|
| 378 | }
|
---|
| 379 |
|
---|
| 380 | s->pos = 0;
|
---|
| 381 |
|
---|
| 382 | return true;
|
---|
| 383 | }
|
---|
| 384 | #endif
|
---|
| 385 |
|
---|
| 386 | /* Decode the Stream Header field (the first 12 bytes of the .xz Stream). */
|
---|
| 387 | static enum xz_ret XZ_FUNC dec_stream_header(struct xz_dec *s)
|
---|
| 388 | {
|
---|
| 389 | if (!memeq(s->temp.buf, HEADER_MAGIC, HEADER_MAGIC_SIZE))
|
---|
| 390 | return XZ_FORMAT_ERROR;
|
---|
| 391 |
|
---|
| 392 | if (xz_crc32(s->temp.buf + HEADER_MAGIC_SIZE, 2, 0)
|
---|
| 393 | != get_le32(s->temp.buf + HEADER_MAGIC_SIZE + 2))
|
---|
| 394 | return XZ_DATA_ERROR;
|
---|
| 395 |
|
---|
| 396 | if (s->temp.buf[HEADER_MAGIC_SIZE] != 0)
|
---|
| 397 | return XZ_OPTIONS_ERROR;
|
---|
| 398 |
|
---|
| 399 | /*
|
---|
| 400 | * Of integrity checks, we support only none (Check ID = 0) and
|
---|
| 401 | * CRC32 (Check ID = 1). However, if XZ_DEC_ANY_CHECK is defined,
|
---|
| 402 | * we will accept other check types too, but then the check won't
|
---|
| 403 | * be verified and a warning (XZ_UNSUPPORTED_CHECK) will be given.
|
---|
| 404 | */
|
---|
| 405 | s->check_type = s->temp.buf[HEADER_MAGIC_SIZE + 1];
|
---|
| 406 |
|
---|
| 407 | #ifdef XZ_DEC_ANY_CHECK
|
---|
| 408 | if (s->check_type > XZ_CHECK_MAX)
|
---|
| 409 | return XZ_OPTIONS_ERROR;
|
---|
| 410 |
|
---|
| 411 | if (s->check_type > XZ_CHECK_CRC32)
|
---|
| 412 | return XZ_UNSUPPORTED_CHECK;
|
---|
| 413 | #else
|
---|
| 414 | if (s->check_type > XZ_CHECK_CRC32)
|
---|
| 415 | return XZ_OPTIONS_ERROR;
|
---|
| 416 | #endif
|
---|
| 417 |
|
---|
| 418 | return XZ_OK;
|
---|
| 419 | }
|
---|
| 420 |
|
---|
| 421 | /* Decode the Stream Footer field (the last 12 bytes of the .xz Stream) */
|
---|
| 422 | static enum xz_ret XZ_FUNC dec_stream_footer(struct xz_dec *s)
|
---|
| 423 | {
|
---|
| 424 | if (!memeq(s->temp.buf + 10, FOOTER_MAGIC, FOOTER_MAGIC_SIZE))
|
---|
| 425 | return XZ_DATA_ERROR;
|
---|
| 426 |
|
---|
| 427 | if (xz_crc32(s->temp.buf + 4, 6, 0) != get_le32(s->temp.buf))
|
---|
| 428 | return XZ_DATA_ERROR;
|
---|
| 429 |
|
---|
| 430 | /*
|
---|
| 431 | * Validate Backward Size. Note that we never added the size of the
|
---|
| 432 | * Index CRC32 field to s->index.size, thus we use s->index.size / 4
|
---|
| 433 | * instead of s->index.size / 4 - 1.
|
---|
| 434 | */
|
---|
| 435 | if ((s->index.size >> 2) != get_le32(s->temp.buf + 4))
|
---|
| 436 | return XZ_DATA_ERROR;
|
---|
| 437 |
|
---|
| 438 | if (s->temp.buf[8] != 0 || s->temp.buf[9] != s->check_type)
|
---|
| 439 | return XZ_DATA_ERROR;
|
---|
| 440 |
|
---|
| 441 | /*
|
---|
| 442 | * Use XZ_STREAM_END instead of XZ_OK to be more convenient
|
---|
| 443 | * for the caller.
|
---|
| 444 | */
|
---|
| 445 | return XZ_STREAM_END;
|
---|
| 446 | }
|
---|
| 447 |
|
---|
| 448 | /* Decode the Block Header and initialize the filter chain. */
|
---|
| 449 | static enum xz_ret XZ_FUNC dec_block_header(struct xz_dec *s)
|
---|
| 450 | {
|
---|
| 451 | enum xz_ret ret;
|
---|
| 452 |
|
---|
| 453 | /*
|
---|
| 454 | * Validate the CRC32. We know that the temp buffer is at least
|
---|
| 455 | * eight bytes so this is safe.
|
---|
| 456 | */
|
---|
| 457 | s->temp.size -= 4;
|
---|
| 458 | if (xz_crc32(s->temp.buf, s->temp.size, 0)
|
---|
| 459 | != get_le32(s->temp.buf + s->temp.size))
|
---|
| 460 | return XZ_DATA_ERROR;
|
---|
| 461 |
|
---|
| 462 | s->temp.pos = 2;
|
---|
| 463 |
|
---|
| 464 | /*
|
---|
| 465 | * Catch unsupported Block Flags. We support only one or two filters
|
---|
| 466 | * in the chain, so we catch that with the same test.
|
---|
| 467 | */
|
---|
| 468 | #ifdef XZ_DEC_BCJ
|
---|
| 469 | if (s->temp.buf[1] & 0x3E)
|
---|
| 470 | #else
|
---|
| 471 | if (s->temp.buf[1] & 0x3F)
|
---|
| 472 | #endif
|
---|
| 473 | return XZ_OPTIONS_ERROR;
|
---|
| 474 |
|
---|
| 475 | /* Compressed Size */
|
---|
| 476 | if (s->temp.buf[1] & 0x40) {
|
---|
| 477 | if (dec_vli(s, s->temp.buf, &s->temp.pos, s->temp.size)
|
---|
| 478 | != XZ_STREAM_END)
|
---|
| 479 | return XZ_DATA_ERROR;
|
---|
| 480 |
|
---|
| 481 | s->block_header.compressed = s->vli;
|
---|
| 482 | } else {
|
---|
| 483 | s->block_header.compressed = VLI_UNKNOWN;
|
---|
| 484 | }
|
---|
| 485 |
|
---|
| 486 | /* Uncompressed Size */
|
---|
| 487 | if (s->temp.buf[1] & 0x80) {
|
---|
| 488 | if (dec_vli(s, s->temp.buf, &s->temp.pos, s->temp.size)
|
---|
| 489 | != XZ_STREAM_END)
|
---|
| 490 | return XZ_DATA_ERROR;
|
---|
| 491 |
|
---|
| 492 | s->block_header.uncompressed = s->vli;
|
---|
| 493 | } else {
|
---|
| 494 | s->block_header.uncompressed = VLI_UNKNOWN;
|
---|
| 495 | }
|
---|
| 496 |
|
---|
| 497 | #ifdef XZ_DEC_BCJ
|
---|
| 498 | /* If there are two filters, the first one must be a BCJ filter. */
|
---|
| 499 | s->bcj_active = s->temp.buf[1] & 0x01;
|
---|
| 500 | if (s->bcj_active) {
|
---|
| 501 | if (s->temp.size - s->temp.pos < 2)
|
---|
| 502 | return XZ_OPTIONS_ERROR;
|
---|
| 503 |
|
---|
| 504 | ret = xz_dec_bcj_reset(s->bcj, s->temp.buf[s->temp.pos++]);
|
---|
| 505 | if (ret != XZ_OK)
|
---|
| 506 | return ret;
|
---|
| 507 |
|
---|
| 508 | /*
|
---|
| 509 | * We don't support custom start offset,
|
---|
| 510 | * so Size of Properties must be zero.
|
---|
| 511 | */
|
---|
| 512 | if (s->temp.buf[s->temp.pos++] != 0x00)
|
---|
| 513 | return XZ_OPTIONS_ERROR;
|
---|
| 514 | }
|
---|
| 515 | #endif
|
---|
| 516 |
|
---|
| 517 | /* Valid Filter Flags always take at least two bytes. */
|
---|
| 518 | if (s->temp.size - s->temp.pos < 2)
|
---|
| 519 | return XZ_DATA_ERROR;
|
---|
| 520 |
|
---|
| 521 | /* Filter ID = LZMA2 */
|
---|
| 522 | if (s->temp.buf[s->temp.pos++] != 0x21)
|
---|
| 523 | return XZ_OPTIONS_ERROR;
|
---|
| 524 |
|
---|
| 525 | /* Size of Properties = 1-byte Filter Properties */
|
---|
| 526 | if (s->temp.buf[s->temp.pos++] != 0x01)
|
---|
| 527 | return XZ_OPTIONS_ERROR;
|
---|
| 528 |
|
---|
| 529 | /* Filter Properties contains LZMA2 dictionary size. */
|
---|
| 530 | if (s->temp.size - s->temp.pos < 1)
|
---|
| 531 | return XZ_DATA_ERROR;
|
---|
| 532 |
|
---|
| 533 | ret = xz_dec_lzma2_reset(s->lzma2, s->temp.buf[s->temp.pos++]);
|
---|
| 534 | if (ret != XZ_OK)
|
---|
| 535 | return ret;
|
---|
| 536 |
|
---|
| 537 | /* The rest must be Header Padding. */
|
---|
| 538 | while (s->temp.pos < s->temp.size)
|
---|
| 539 | if (s->temp.buf[s->temp.pos++] != 0x00)
|
---|
| 540 | return XZ_OPTIONS_ERROR;
|
---|
| 541 |
|
---|
| 542 | s->temp.pos = 0;
|
---|
| 543 | s->block.compressed = 0;
|
---|
| 544 | s->block.uncompressed = 0;
|
---|
| 545 |
|
---|
| 546 | return XZ_OK;
|
---|
| 547 | }
|
---|
| 548 |
|
---|
| 549 | static enum xz_ret XZ_FUNC dec_main(struct xz_dec *s, struct xz_buf *b)
|
---|
| 550 | {
|
---|
| 551 | enum xz_ret ret;
|
---|
| 552 |
|
---|
| 553 | /*
|
---|
| 554 | * Store the start position for the case when we are in the middle
|
---|
| 555 | * of the Index field.
|
---|
| 556 | */
|
---|
| 557 | s->in_start = b->in_pos;
|
---|
| 558 |
|
---|
| 559 | while (true) {
|
---|
| 560 | switch (s->sequence) {
|
---|
| 561 | case SEQ_STREAM_HEADER:
|
---|
| 562 | /*
|
---|
| 563 | * Stream Header is copied to s->temp, and then
|
---|
| 564 | * decoded from there. This way if the caller
|
---|
| 565 | * gives us only little input at a time, we can
|
---|
| 566 | * still keep the Stream Header decoding code
|
---|
| 567 | * simple. Similar approach is used in many places
|
---|
| 568 | * in this file.
|
---|
| 569 | */
|
---|
| 570 | if (!fill_temp(s, b))
|
---|
| 571 | return XZ_OK;
|
---|
| 572 |
|
---|
| 573 | /*
|
---|
| 574 | * If dec_stream_header() returns
|
---|
| 575 | * XZ_UNSUPPORTED_CHECK, it is still possible
|
---|
| 576 | * to continue decoding if working in multi-call
|
---|
| 577 | * mode. Thus, update s->sequence before calling
|
---|
| 578 | * dec_stream_header().
|
---|
| 579 | */
|
---|
| 580 | s->sequence = SEQ_BLOCK_START;
|
---|
| 581 |
|
---|
| 582 | ret = dec_stream_header(s);
|
---|
| 583 | if (ret != XZ_OK)
|
---|
| 584 | return ret;
|
---|
| 585 |
|
---|
| 586 | case SEQ_BLOCK_START:
|
---|
| 587 | /* We need one byte of input to continue. */
|
---|
| 588 | if (b->in_pos == b->in_size)
|
---|
| 589 | return XZ_OK;
|
---|
| 590 |
|
---|
| 591 | /* See if this is the beginning of the Index field. */
|
---|
| 592 | if (b->in[b->in_pos] == 0) {
|
---|
| 593 | s->in_start = b->in_pos++;
|
---|
| 594 | s->sequence = SEQ_INDEX;
|
---|
| 595 | break;
|
---|
| 596 | }
|
---|
| 597 |
|
---|
| 598 | /*
|
---|
| 599 | * Calculate the size of the Block Header and
|
---|
| 600 | * prepare to decode it.
|
---|
| 601 | */
|
---|
| 602 | s->block_header.size
|
---|
| 603 | = ((uint32_t)b->in[b->in_pos] + 1) * 4;
|
---|
| 604 |
|
---|
| 605 | s->temp.size = s->block_header.size;
|
---|
| 606 | s->temp.pos = 0;
|
---|
| 607 | s->sequence = SEQ_BLOCK_HEADER;
|
---|
| 608 |
|
---|
| 609 | case SEQ_BLOCK_HEADER:
|
---|
| 610 | if (!fill_temp(s, b))
|
---|
| 611 | return XZ_OK;
|
---|
| 612 |
|
---|
| 613 | ret = dec_block_header(s);
|
---|
| 614 | if (ret != XZ_OK)
|
---|
| 615 | return ret;
|
---|
| 616 |
|
---|
| 617 | s->sequence = SEQ_BLOCK_UNCOMPRESS;
|
---|
| 618 |
|
---|
| 619 | case SEQ_BLOCK_UNCOMPRESS:
|
---|
| 620 | ret = dec_block(s, b);
|
---|
| 621 | if (ret != XZ_STREAM_END)
|
---|
| 622 | return ret;
|
---|
| 623 |
|
---|
| 624 | s->sequence = SEQ_BLOCK_PADDING;
|
---|
| 625 |
|
---|
| 626 | case SEQ_BLOCK_PADDING:
|
---|
| 627 | /*
|
---|
| 628 | * Size of Compressed Data + Block Padding
|
---|
| 629 | * must be a multiple of four. We don't need
|
---|
| 630 | * s->block.compressed for anything else
|
---|
| 631 | * anymore, so we use it here to test the size
|
---|
| 632 | * of the Block Padding field.
|
---|
| 633 | */
|
---|
| 634 | while (s->block.compressed & 3) {
|
---|
| 635 | if (b->in_pos == b->in_size)
|
---|
| 636 | return XZ_OK;
|
---|
| 637 |
|
---|
| 638 | if (b->in[b->in_pos++] != 0)
|
---|
| 639 | return XZ_DATA_ERROR;
|
---|
| 640 |
|
---|
| 641 | ++s->block.compressed;
|
---|
| 642 | }
|
---|
| 643 |
|
---|
| 644 | s->sequence = SEQ_BLOCK_CHECK;
|
---|
| 645 |
|
---|
| 646 | case SEQ_BLOCK_CHECK:
|
---|
| 647 | if (s->check_type == XZ_CHECK_CRC32) {
|
---|
| 648 | ret = crc32_validate(s, b);
|
---|
| 649 | if (ret != XZ_STREAM_END)
|
---|
| 650 | return ret;
|
---|
| 651 | }
|
---|
| 652 | #ifdef XZ_DEC_ANY_CHECK
|
---|
| 653 | else if (!check_skip(s, b)) {
|
---|
| 654 | return XZ_OK;
|
---|
| 655 | }
|
---|
| 656 | #endif
|
---|
| 657 |
|
---|
| 658 | s->sequence = SEQ_BLOCK_START;
|
---|
| 659 | break;
|
---|
| 660 |
|
---|
| 661 | case SEQ_INDEX:
|
---|
| 662 | ret = dec_index(s, b);
|
---|
| 663 | if (ret != XZ_STREAM_END)
|
---|
| 664 | return ret;
|
---|
| 665 |
|
---|
| 666 | s->sequence = SEQ_INDEX_PADDING;
|
---|
| 667 |
|
---|
| 668 | case SEQ_INDEX_PADDING:
|
---|
| 669 | while ((s->index.size + (b->in_pos - s->in_start))
|
---|
| 670 | & 3) {
|
---|
| 671 | if (b->in_pos == b->in_size) {
|
---|
| 672 | index_update(s, b);
|
---|
| 673 | return XZ_OK;
|
---|
| 674 | }
|
---|
| 675 |
|
---|
| 676 | if (b->in[b->in_pos++] != 0)
|
---|
| 677 | return XZ_DATA_ERROR;
|
---|
| 678 | }
|
---|
| 679 |
|
---|
| 680 | /* Finish the CRC32 value and Index size. */
|
---|
| 681 | index_update(s, b);
|
---|
| 682 |
|
---|
| 683 | /* Compare the hashes to validate the Index field. */
|
---|
| 684 | if (!memeq(&s->block.hash, &s->index.hash,
|
---|
| 685 | sizeof(s->block.hash)))
|
---|
| 686 | return XZ_DATA_ERROR;
|
---|
| 687 |
|
---|
| 688 | s->sequence = SEQ_INDEX_CRC32;
|
---|
| 689 |
|
---|
| 690 | case SEQ_INDEX_CRC32:
|
---|
| 691 | ret = crc32_validate(s, b);
|
---|
| 692 | if (ret != XZ_STREAM_END)
|
---|
| 693 | return ret;
|
---|
| 694 |
|
---|
| 695 | s->temp.size = STREAM_HEADER_SIZE;
|
---|
| 696 | s->sequence = SEQ_STREAM_FOOTER;
|
---|
| 697 |
|
---|
| 698 | case SEQ_STREAM_FOOTER:
|
---|
| 699 | if (!fill_temp(s, b))
|
---|
| 700 | return XZ_OK;
|
---|
| 701 |
|
---|
| 702 | return dec_stream_footer(s);
|
---|
| 703 | }
|
---|
| 704 | }
|
---|
| 705 |
|
---|
| 706 | /* Never reached */
|
---|
| 707 | }
|
---|
| 708 |
|
---|
| 709 | /*
|
---|
| 710 | * xz_dec_run() is a wrapper for dec_main() to handle some special cases in
|
---|
| 711 | * multi-call and single-call decoding.
|
---|
| 712 | *
|
---|
| 713 | * In multi-call mode, we must return XZ_BUF_ERROR when it seems clear that we
|
---|
| 714 | * are not going to make any progress anymore. This is to prevent the caller
|
---|
| 715 | * from calling us infinitely when the input file is truncated or otherwise
|
---|
| 716 | * corrupt. Since zlib-style API allows that the caller fills the input buffer
|
---|
| 717 | * only when the decoder doesn't produce any new output, we have to be careful
|
---|
| 718 | * to avoid returning XZ_BUF_ERROR too easily: XZ_BUF_ERROR is returned only
|
---|
| 719 | * after the second consecutive call to xz_dec_run() that makes no progress.
|
---|
| 720 | *
|
---|
| 721 | * In single-call mode, if we couldn't decode everything and no error
|
---|
| 722 | * occurred, either the input is truncated or the output buffer is too small.
|
---|
| 723 | * Since we know that the last input byte never produces any output, we know
|
---|
| 724 | * that if all the input was consumed and decoding wasn't finished, the file
|
---|
| 725 | * must be corrupt. Otherwise the output buffer has to be too small or the
|
---|
| 726 | * file is corrupt in a way that decoding it produces too big output.
|
---|
| 727 | *
|
---|
| 728 | * If single-call decoding fails, we reset b->in_pos and b->out_pos back to
|
---|
| 729 | * their original values. This is because with some filter chains there won't
|
---|
| 730 | * be any valid uncompressed data in the output buffer unless the decoding
|
---|
| 731 | * actually succeeds (that's the price to pay of using the output buffer as
|
---|
| 732 | * the workspace).
|
---|
| 733 | */
|
---|
| 734 | XZ_EXTERN enum xz_ret XZ_FUNC xz_dec_run(struct xz_dec *s, struct xz_buf *b)
|
---|
| 735 | {
|
---|
| 736 | size_t in_start;
|
---|
| 737 | size_t out_start;
|
---|
| 738 | enum xz_ret ret;
|
---|
| 739 |
|
---|
| 740 | if (DEC_IS_SINGLE(s->mode))
|
---|
| 741 | xz_dec_reset(s);
|
---|
| 742 |
|
---|
| 743 | in_start = b->in_pos;
|
---|
| 744 | out_start = b->out_pos;
|
---|
| 745 | ret = dec_main(s, b);
|
---|
| 746 |
|
---|
| 747 | if (DEC_IS_SINGLE(s->mode)) {
|
---|
| 748 | if (ret == XZ_OK)
|
---|
| 749 | ret = b->in_pos == b->in_size
|
---|
| 750 | ? XZ_DATA_ERROR : XZ_BUF_ERROR;
|
---|
| 751 |
|
---|
| 752 | if (ret != XZ_STREAM_END) {
|
---|
| 753 | b->in_pos = in_start;
|
---|
| 754 | b->out_pos = out_start;
|
---|
| 755 | }
|
---|
| 756 |
|
---|
| 757 | } else if (ret == XZ_OK && in_start == b->in_pos
|
---|
| 758 | && out_start == b->out_pos) {
|
---|
| 759 | if (s->allow_buf_error)
|
---|
| 760 | ret = XZ_BUF_ERROR;
|
---|
| 761 |
|
---|
| 762 | s->allow_buf_error = true;
|
---|
| 763 | } else {
|
---|
| 764 | s->allow_buf_error = false;
|
---|
| 765 | }
|
---|
| 766 |
|
---|
| 767 | return ret;
|
---|
| 768 | }
|
---|
| 769 |
|
---|
| 770 | XZ_EXTERN struct xz_dec * XZ_FUNC xz_dec_init(
|
---|
| 771 | enum xz_mode mode, uint32_t dict_max)
|
---|
| 772 | {
|
---|
| 773 | struct xz_dec *s = kmalloc(sizeof(*s), GFP_KERNEL);
|
---|
| 774 | if (s == NULL)
|
---|
| 775 | return NULL;
|
---|
| 776 |
|
---|
| 777 | s->mode = mode;
|
---|
| 778 |
|
---|
| 779 | #ifdef XZ_DEC_BCJ
|
---|
| 780 | s->bcj = xz_dec_bcj_create(DEC_IS_SINGLE(mode));
|
---|
| 781 | if (s->bcj == NULL)
|
---|
| 782 | goto error_bcj;
|
---|
| 783 | #endif
|
---|
| 784 |
|
---|
| 785 | s->lzma2 = xz_dec_lzma2_create(mode, dict_max);
|
---|
| 786 | if (s->lzma2 == NULL)
|
---|
| 787 | goto error_lzma2;
|
---|
| 788 |
|
---|
| 789 | xz_dec_reset(s);
|
---|
| 790 | return s;
|
---|
| 791 |
|
---|
| 792 | error_lzma2:
|
---|
| 793 | #ifdef XZ_DEC_BCJ
|
---|
| 794 | xz_dec_bcj_end(s->bcj);
|
---|
| 795 | error_bcj:
|
---|
| 796 | #endif
|
---|
| 797 | kfree(s);
|
---|
| 798 | return NULL;
|
---|
| 799 | }
|
---|
| 800 |
|
---|
| 801 | XZ_EXTERN void XZ_FUNC xz_dec_reset(struct xz_dec *s)
|
---|
| 802 | {
|
---|
| 803 | s->sequence = SEQ_STREAM_HEADER;
|
---|
| 804 | s->allow_buf_error = false;
|
---|
| 805 | s->pos = 0;
|
---|
| 806 | s->crc32 = 0;
|
---|
| 807 | memzero(&s->block, sizeof(s->block));
|
---|
| 808 | memzero(&s->index, sizeof(s->index));
|
---|
| 809 | s->temp.pos = 0;
|
---|
| 810 | s->temp.size = STREAM_HEADER_SIZE;
|
---|
| 811 | }
|
---|
| 812 |
|
---|
| 813 | XZ_EXTERN void XZ_FUNC xz_dec_end(struct xz_dec *s)
|
---|
| 814 | {
|
---|
| 815 | if (s != NULL) {
|
---|
| 816 | xz_dec_lzma2_end(s->lzma2);
|
---|
| 817 | #ifdef XZ_DEC_BCJ
|
---|
| 818 | xz_dec_bcj_end(s->bcj);
|
---|
| 819 | #endif
|
---|
| 820 | kfree(s);
|
---|
| 821 | }
|
---|
| 822 | }
|
---|