1 /* 2 * Block driver for the QCOW version 2 format 3 * 4 * Copyright (c) 2004-2006 Fabrice Bellard 5 * 6 * Permission is hereby granted, free of charge, to any person obtaining a copy 7 * of this software and associated documentation files (the "Software"), to deal 8 * in the Software without restriction, including without limitation the rights 9 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell 10 * copies of the Software, and to permit persons to whom the Software is 11 * furnished to do so, subject to the following conditions: 12 * 13 * The above copyright notice and this permission notice shall be included in 14 * all copies or substantial portions of the Software. 15 * 16 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 17 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 18 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL 19 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 20 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, 21 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN 22 * THE SOFTWARE. 23 */ 24 25 #include "qemu-common.h" 26 #include "block/block_int.h" 27 #include "block/qcow2.h" 28 #include "qemu/range.h" 29 30 static int64_t alloc_clusters_noref(BlockDriverState *bs, uint64_t size); 31 static int QEMU_WARN_UNUSED_RESULT update_refcount(BlockDriverState *bs, 32 int64_t offset, int64_t length, uint64_t addend, 33 bool decrease, enum qcow2_discard_type type); 34 35 static uint64_t get_refcount_ro0(const void *refcount_array, uint64_t index); 36 static uint64_t get_refcount_ro1(const void *refcount_array, uint64_t index); 37 static uint64_t get_refcount_ro2(const void *refcount_array, uint64_t index); 38 static uint64_t get_refcount_ro3(const void *refcount_array, uint64_t index); 39 static uint64_t get_refcount_ro4(const void *refcount_array, uint64_t index); 40 static uint64_t get_refcount_ro5(const void *refcount_array, uint64_t index); 41 static uint64_t get_refcount_ro6(const void *refcount_array, uint64_t index); 42 43 static void set_refcount_ro0(void *refcount_array, uint64_t index, 44 uint64_t value); 45 static void set_refcount_ro1(void *refcount_array, uint64_t index, 46 uint64_t value); 47 static void set_refcount_ro2(void *refcount_array, uint64_t index, 48 uint64_t value); 49 static void set_refcount_ro3(void *refcount_array, uint64_t index, 50 uint64_t value); 51 static void set_refcount_ro4(void *refcount_array, uint64_t index, 52 uint64_t value); 53 static void set_refcount_ro5(void *refcount_array, uint64_t index, 54 uint64_t value); 55 static void set_refcount_ro6(void *refcount_array, uint64_t index, 56 uint64_t value); 57 58 59 static Qcow2GetRefcountFunc *const get_refcount_funcs[] = { 60 &get_refcount_ro0, 61 &get_refcount_ro1, 62 &get_refcount_ro2, 63 &get_refcount_ro3, 64 &get_refcount_ro4, 65 &get_refcount_ro5, 66 &get_refcount_ro6 67 }; 68 69 static Qcow2SetRefcountFunc *const set_refcount_funcs[] = { 70 &set_refcount_ro0, 71 &set_refcount_ro1, 72 &set_refcount_ro2, 73 &set_refcount_ro3, 74 &set_refcount_ro4, 75 &set_refcount_ro5, 76 &set_refcount_ro6 77 }; 78 79 80 /*********************************************************/ 81 /* refcount handling */ 82 83 int qcow2_refcount_init(BlockDriverState *bs) 84 { 85 BDRVQcow2State *s = bs->opaque; 86 unsigned int refcount_table_size2, i; 87 int ret; 88 89 assert(s->refcount_order >= 0 && s->refcount_order <= 6); 90 91 s->get_refcount = get_refcount_funcs[s->refcount_order]; 92 s->set_refcount = set_refcount_funcs[s->refcount_order]; 93 94 assert(s->refcount_table_size <= INT_MAX / sizeof(uint64_t)); 95 refcount_table_size2 = s->refcount_table_size * sizeof(uint64_t); 96 s->refcount_table = g_try_malloc(refcount_table_size2); 97 98 if (s->refcount_table_size > 0) { 99 if (s->refcount_table == NULL) { 100 ret = -ENOMEM; 101 goto fail; 102 } 103 BLKDBG_EVENT(bs->file, BLKDBG_REFTABLE_LOAD); 104 ret = bdrv_pread(bs->file->bs, s->refcount_table_offset, 105 s->refcount_table, refcount_table_size2); 106 if (ret < 0) { 107 goto fail; 108 } 109 for(i = 0; i < s->refcount_table_size; i++) 110 be64_to_cpus(&s->refcount_table[i]); 111 } 112 return 0; 113 fail: 114 return ret; 115 } 116 117 void qcow2_refcount_close(BlockDriverState *bs) 118 { 119 BDRVQcow2State *s = bs->opaque; 120 g_free(s->refcount_table); 121 } 122 123 124 static uint64_t get_refcount_ro0(const void *refcount_array, uint64_t index) 125 { 126 return (((const uint8_t *)refcount_array)[index / 8] >> (index % 8)) & 0x1; 127 } 128 129 static void set_refcount_ro0(void *refcount_array, uint64_t index, 130 uint64_t value) 131 { 132 assert(!(value >> 1)); 133 ((uint8_t *)refcount_array)[index / 8] &= ~(0x1 << (index % 8)); 134 ((uint8_t *)refcount_array)[index / 8] |= value << (index % 8); 135 } 136 137 static uint64_t get_refcount_ro1(const void *refcount_array, uint64_t index) 138 { 139 return (((const uint8_t *)refcount_array)[index / 4] >> (2 * (index % 4))) 140 & 0x3; 141 } 142 143 static void set_refcount_ro1(void *refcount_array, uint64_t index, 144 uint64_t value) 145 { 146 assert(!(value >> 2)); 147 ((uint8_t *)refcount_array)[index / 4] &= ~(0x3 << (2 * (index % 4))); 148 ((uint8_t *)refcount_array)[index / 4] |= value << (2 * (index % 4)); 149 } 150 151 static uint64_t get_refcount_ro2(const void *refcount_array, uint64_t index) 152 { 153 return (((const uint8_t *)refcount_array)[index / 2] >> (4 * (index % 2))) 154 & 0xf; 155 } 156 157 static void set_refcount_ro2(void *refcount_array, uint64_t index, 158 uint64_t value) 159 { 160 assert(!(value >> 4)); 161 ((uint8_t *)refcount_array)[index / 2] &= ~(0xf << (4 * (index % 2))); 162 ((uint8_t *)refcount_array)[index / 2] |= value << (4 * (index % 2)); 163 } 164 165 static uint64_t get_refcount_ro3(const void *refcount_array, uint64_t index) 166 { 167 return ((const uint8_t *)refcount_array)[index]; 168 } 169 170 static void set_refcount_ro3(void *refcount_array, uint64_t index, 171 uint64_t value) 172 { 173 assert(!(value >> 8)); 174 ((uint8_t *)refcount_array)[index] = value; 175 } 176 177 static uint64_t get_refcount_ro4(const void *refcount_array, uint64_t index) 178 { 179 return be16_to_cpu(((const uint16_t *)refcount_array)[index]); 180 } 181 182 static void set_refcount_ro4(void *refcount_array, uint64_t index, 183 uint64_t value) 184 { 185 assert(!(value >> 16)); 186 ((uint16_t *)refcount_array)[index] = cpu_to_be16(value); 187 } 188 189 static uint64_t get_refcount_ro5(const void *refcount_array, uint64_t index) 190 { 191 return be32_to_cpu(((const uint32_t *)refcount_array)[index]); 192 } 193 194 static void set_refcount_ro5(void *refcount_array, uint64_t index, 195 uint64_t value) 196 { 197 assert(!(value >> 32)); 198 ((uint32_t *)refcount_array)[index] = cpu_to_be32(value); 199 } 200 201 static uint64_t get_refcount_ro6(const void *refcount_array, uint64_t index) 202 { 203 return be64_to_cpu(((const uint64_t *)refcount_array)[index]); 204 } 205 206 static void set_refcount_ro6(void *refcount_array, uint64_t index, 207 uint64_t value) 208 { 209 ((uint64_t *)refcount_array)[index] = cpu_to_be64(value); 210 } 211 212 213 static int load_refcount_block(BlockDriverState *bs, 214 int64_t refcount_block_offset, 215 void **refcount_block) 216 { 217 BDRVQcow2State *s = bs->opaque; 218 int ret; 219 220 BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_LOAD); 221 ret = qcow2_cache_get(bs, s->refcount_block_cache, refcount_block_offset, 222 refcount_block); 223 224 return ret; 225 } 226 227 /* 228 * Retrieves the refcount of the cluster given by its index and stores it in 229 * *refcount. Returns 0 on success and -errno on failure. 230 */ 231 int qcow2_get_refcount(BlockDriverState *bs, int64_t cluster_index, 232 uint64_t *refcount) 233 { 234 BDRVQcow2State *s = bs->opaque; 235 uint64_t refcount_table_index, block_index; 236 int64_t refcount_block_offset; 237 int ret; 238 void *refcount_block; 239 240 refcount_table_index = cluster_index >> s->refcount_block_bits; 241 if (refcount_table_index >= s->refcount_table_size) { 242 *refcount = 0; 243 return 0; 244 } 245 refcount_block_offset = 246 s->refcount_table[refcount_table_index] & REFT_OFFSET_MASK; 247 if (!refcount_block_offset) { 248 *refcount = 0; 249 return 0; 250 } 251 252 if (offset_into_cluster(s, refcount_block_offset)) { 253 qcow2_signal_corruption(bs, true, -1, -1, "Refblock offset %#" PRIx64 254 " unaligned (reftable index: %#" PRIx64 ")", 255 refcount_block_offset, refcount_table_index); 256 return -EIO; 257 } 258 259 ret = qcow2_cache_get(bs, s->refcount_block_cache, refcount_block_offset, 260 &refcount_block); 261 if (ret < 0) { 262 return ret; 263 } 264 265 block_index = cluster_index & (s->refcount_block_size - 1); 266 *refcount = s->get_refcount(refcount_block, block_index); 267 268 qcow2_cache_put(bs, s->refcount_block_cache, &refcount_block); 269 270 return 0; 271 } 272 273 /* 274 * Rounds the refcount table size up to avoid growing the table for each single 275 * refcount block that is allocated. 276 */ 277 static unsigned int next_refcount_table_size(BDRVQcow2State *s, 278 unsigned int min_size) 279 { 280 unsigned int min_clusters = (min_size >> (s->cluster_bits - 3)) + 1; 281 unsigned int refcount_table_clusters = 282 MAX(1, s->refcount_table_size >> (s->cluster_bits - 3)); 283 284 while (min_clusters > refcount_table_clusters) { 285 refcount_table_clusters = (refcount_table_clusters * 3 + 1) / 2; 286 } 287 288 return refcount_table_clusters << (s->cluster_bits - 3); 289 } 290 291 292 /* Checks if two offsets are described by the same refcount block */ 293 static int in_same_refcount_block(BDRVQcow2State *s, uint64_t offset_a, 294 uint64_t offset_b) 295 { 296 uint64_t block_a = offset_a >> (s->cluster_bits + s->refcount_block_bits); 297 uint64_t block_b = offset_b >> (s->cluster_bits + s->refcount_block_bits); 298 299 return (block_a == block_b); 300 } 301 302 /* 303 * Loads a refcount block. If it doesn't exist yet, it is allocated first 304 * (including growing the refcount table if needed). 305 * 306 * Returns 0 on success or -errno in error case 307 */ 308 static int alloc_refcount_block(BlockDriverState *bs, 309 int64_t cluster_index, void **refcount_block) 310 { 311 BDRVQcow2State *s = bs->opaque; 312 unsigned int refcount_table_index; 313 int ret; 314 315 BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC); 316 317 /* Find the refcount block for the given cluster */ 318 refcount_table_index = cluster_index >> s->refcount_block_bits; 319 320 if (refcount_table_index < s->refcount_table_size) { 321 322 uint64_t refcount_block_offset = 323 s->refcount_table[refcount_table_index] & REFT_OFFSET_MASK; 324 325 /* If it's already there, we're done */ 326 if (refcount_block_offset) { 327 if (offset_into_cluster(s, refcount_block_offset)) { 328 qcow2_signal_corruption(bs, true, -1, -1, "Refblock offset %#" 329 PRIx64 " unaligned (reftable index: " 330 "%#x)", refcount_block_offset, 331 refcount_table_index); 332 return -EIO; 333 } 334 335 return load_refcount_block(bs, refcount_block_offset, 336 refcount_block); 337 } 338 } 339 340 /* 341 * If we came here, we need to allocate something. Something is at least 342 * a cluster for the new refcount block. It may also include a new refcount 343 * table if the old refcount table is too small. 344 * 345 * Note that allocating clusters here needs some special care: 346 * 347 * - We can't use the normal qcow2_alloc_clusters(), it would try to 348 * increase the refcount and very likely we would end up with an endless 349 * recursion. Instead we must place the refcount blocks in a way that 350 * they can describe them themselves. 351 * 352 * - We need to consider that at this point we are inside update_refcounts 353 * and potentially doing an initial refcount increase. This means that 354 * some clusters have already been allocated by the caller, but their 355 * refcount isn't accurate yet. If we allocate clusters for metadata, we 356 * need to return -EAGAIN to signal the caller that it needs to restart 357 * the search for free clusters. 358 * 359 * - alloc_clusters_noref and qcow2_free_clusters may load a different 360 * refcount block into the cache 361 */ 362 363 *refcount_block = NULL; 364 365 /* We write to the refcount table, so we might depend on L2 tables */ 366 ret = qcow2_cache_flush(bs, s->l2_table_cache); 367 if (ret < 0) { 368 return ret; 369 } 370 371 /* Allocate the refcount block itself and mark it as used */ 372 int64_t new_block = alloc_clusters_noref(bs, s->cluster_size); 373 if (new_block < 0) { 374 return new_block; 375 } 376 377 #ifdef DEBUG_ALLOC2 378 fprintf(stderr, "qcow2: Allocate refcount block %d for %" PRIx64 379 " at %" PRIx64 "\n", 380 refcount_table_index, cluster_index << s->cluster_bits, new_block); 381 #endif 382 383 if (in_same_refcount_block(s, new_block, cluster_index << s->cluster_bits)) { 384 /* Zero the new refcount block before updating it */ 385 ret = qcow2_cache_get_empty(bs, s->refcount_block_cache, new_block, 386 refcount_block); 387 if (ret < 0) { 388 goto fail_block; 389 } 390 391 memset(*refcount_block, 0, s->cluster_size); 392 393 /* The block describes itself, need to update the cache */ 394 int block_index = (new_block >> s->cluster_bits) & 395 (s->refcount_block_size - 1); 396 s->set_refcount(*refcount_block, block_index, 1); 397 } else { 398 /* Described somewhere else. This can recurse at most twice before we 399 * arrive at a block that describes itself. */ 400 ret = update_refcount(bs, new_block, s->cluster_size, 1, false, 401 QCOW2_DISCARD_NEVER); 402 if (ret < 0) { 403 goto fail_block; 404 } 405 406 ret = qcow2_cache_flush(bs, s->refcount_block_cache); 407 if (ret < 0) { 408 goto fail_block; 409 } 410 411 /* Initialize the new refcount block only after updating its refcount, 412 * update_refcount uses the refcount cache itself */ 413 ret = qcow2_cache_get_empty(bs, s->refcount_block_cache, new_block, 414 refcount_block); 415 if (ret < 0) { 416 goto fail_block; 417 } 418 419 memset(*refcount_block, 0, s->cluster_size); 420 } 421 422 /* Now the new refcount block needs to be written to disk */ 423 BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_WRITE); 424 qcow2_cache_entry_mark_dirty(bs, s->refcount_block_cache, *refcount_block); 425 ret = qcow2_cache_flush(bs, s->refcount_block_cache); 426 if (ret < 0) { 427 goto fail_block; 428 } 429 430 /* If the refcount table is big enough, just hook the block up there */ 431 if (refcount_table_index < s->refcount_table_size) { 432 uint64_t data64 = cpu_to_be64(new_block); 433 BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_HOOKUP); 434 ret = bdrv_pwrite_sync(bs->file->bs, 435 s->refcount_table_offset + refcount_table_index * sizeof(uint64_t), 436 &data64, sizeof(data64)); 437 if (ret < 0) { 438 goto fail_block; 439 } 440 441 s->refcount_table[refcount_table_index] = new_block; 442 443 /* The new refcount block may be where the caller intended to put its 444 * data, so let it restart the search. */ 445 return -EAGAIN; 446 } 447 448 qcow2_cache_put(bs, s->refcount_block_cache, refcount_block); 449 450 /* 451 * If we come here, we need to grow the refcount table. Again, a new 452 * refcount table needs some space and we can't simply allocate to avoid 453 * endless recursion. 454 * 455 * Therefore let's grab new refcount blocks at the end of the image, which 456 * will describe themselves and the new refcount table. This way we can 457 * reference them only in the new table and do the switch to the new 458 * refcount table at once without producing an inconsistent state in 459 * between. 460 */ 461 BLKDBG_EVENT(bs->file, BLKDBG_REFTABLE_GROW); 462 463 /* Calculate the number of refcount blocks needed so far; this will be the 464 * basis for calculating the index of the first cluster used for the 465 * self-describing refcount structures which we are about to create. 466 * 467 * Because we reached this point, there cannot be any refcount entries for 468 * cluster_index or higher indices yet. However, because new_block has been 469 * allocated to describe that cluster (and it will assume this role later 470 * on), we cannot use that index; also, new_block may actually have a higher 471 * cluster index than cluster_index, so it needs to be taken into account 472 * here (and 1 needs to be added to its value because that cluster is used). 473 */ 474 uint64_t blocks_used = DIV_ROUND_UP(MAX(cluster_index + 1, 475 (new_block >> s->cluster_bits) + 1), 476 s->refcount_block_size); 477 478 if (blocks_used > QCOW_MAX_REFTABLE_SIZE / sizeof(uint64_t)) { 479 return -EFBIG; 480 } 481 482 /* And now we need at least one block more for the new metadata */ 483 uint64_t table_size = next_refcount_table_size(s, blocks_used + 1); 484 uint64_t last_table_size; 485 uint64_t blocks_clusters; 486 do { 487 uint64_t table_clusters = 488 size_to_clusters(s, table_size * sizeof(uint64_t)); 489 blocks_clusters = 1 + 490 ((table_clusters + s->refcount_block_size - 1) 491 / s->refcount_block_size); 492 uint64_t meta_clusters = table_clusters + blocks_clusters; 493 494 last_table_size = table_size; 495 table_size = next_refcount_table_size(s, blocks_used + 496 ((meta_clusters + s->refcount_block_size - 1) 497 / s->refcount_block_size)); 498 499 } while (last_table_size != table_size); 500 501 #ifdef DEBUG_ALLOC2 502 fprintf(stderr, "qcow2: Grow refcount table %" PRId32 " => %" PRId64 "\n", 503 s->refcount_table_size, table_size); 504 #endif 505 506 /* Create the new refcount table and blocks */ 507 uint64_t meta_offset = (blocks_used * s->refcount_block_size) * 508 s->cluster_size; 509 uint64_t table_offset = meta_offset + blocks_clusters * s->cluster_size; 510 uint64_t *new_table = g_try_new0(uint64_t, table_size); 511 void *new_blocks = g_try_malloc0(blocks_clusters * s->cluster_size); 512 513 assert(table_size > 0 && blocks_clusters > 0); 514 if (new_table == NULL || new_blocks == NULL) { 515 ret = -ENOMEM; 516 goto fail_table; 517 } 518 519 /* Fill the new refcount table */ 520 memcpy(new_table, s->refcount_table, 521 s->refcount_table_size * sizeof(uint64_t)); 522 new_table[refcount_table_index] = new_block; 523 524 int i; 525 for (i = 0; i < blocks_clusters; i++) { 526 new_table[blocks_used + i] = meta_offset + (i * s->cluster_size); 527 } 528 529 /* Fill the refcount blocks */ 530 uint64_t table_clusters = size_to_clusters(s, table_size * sizeof(uint64_t)); 531 int block = 0; 532 for (i = 0; i < table_clusters + blocks_clusters; i++) { 533 s->set_refcount(new_blocks, block++, 1); 534 } 535 536 /* Write refcount blocks to disk */ 537 BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_WRITE_BLOCKS); 538 ret = bdrv_pwrite_sync(bs->file->bs, meta_offset, new_blocks, 539 blocks_clusters * s->cluster_size); 540 g_free(new_blocks); 541 new_blocks = NULL; 542 if (ret < 0) { 543 goto fail_table; 544 } 545 546 /* Write refcount table to disk */ 547 for(i = 0; i < table_size; i++) { 548 cpu_to_be64s(&new_table[i]); 549 } 550 551 BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_WRITE_TABLE); 552 ret = bdrv_pwrite_sync(bs->file->bs, table_offset, new_table, 553 table_size * sizeof(uint64_t)); 554 if (ret < 0) { 555 goto fail_table; 556 } 557 558 for(i = 0; i < table_size; i++) { 559 be64_to_cpus(&new_table[i]); 560 } 561 562 /* Hook up the new refcount table in the qcow2 header */ 563 struct QEMU_PACKED { 564 uint64_t d64; 565 uint32_t d32; 566 } data; 567 cpu_to_be64w(&data.d64, table_offset); 568 cpu_to_be32w(&data.d32, table_clusters); 569 BLKDBG_EVENT(bs->file, BLKDBG_REFBLOCK_ALLOC_SWITCH_TABLE); 570 ret = bdrv_pwrite_sync(bs->file->bs, 571 offsetof(QCowHeader, refcount_table_offset), 572 &data, sizeof(data)); 573 if (ret < 0) { 574 goto fail_table; 575 } 576 577 /* And switch it in memory */ 578 uint64_t old_table_offset = s->refcount_table_offset; 579 uint64_t old_table_size = s->refcount_table_size; 580 581 g_free(s->refcount_table); 582 s->refcount_table = new_table; 583 s->refcount_table_size = table_size; 584 s->refcount_table_offset = table_offset; 585 586 /* Free old table. */ 587 qcow2_free_clusters(bs, old_table_offset, old_table_size * sizeof(uint64_t), 588 QCOW2_DISCARD_OTHER); 589 590 ret = load_refcount_block(bs, new_block, refcount_block); 591 if (ret < 0) { 592 return ret; 593 } 594 595 /* If we were trying to do the initial refcount update for some cluster 596 * allocation, we might have used the same clusters to store newly 597 * allocated metadata. Make the caller search some new space. */ 598 return -EAGAIN; 599 600 fail_table: 601 g_free(new_blocks); 602 g_free(new_table); 603 fail_block: 604 if (*refcount_block != NULL) { 605 qcow2_cache_put(bs, s->refcount_block_cache, refcount_block); 606 } 607 return ret; 608 } 609 610 void qcow2_process_discards(BlockDriverState *bs, int ret) 611 { 612 BDRVQcow2State *s = bs->opaque; 613 Qcow2DiscardRegion *d, *next; 614 615 QTAILQ_FOREACH_SAFE(d, &s->discards, next, next) { 616 QTAILQ_REMOVE(&s->discards, d, next); 617 618 /* Discard is optional, ignore the return value */ 619 if (ret >= 0) { 620 bdrv_discard(bs->file->bs, 621 d->offset >> BDRV_SECTOR_BITS, 622 d->bytes >> BDRV_SECTOR_BITS); 623 } 624 625 g_free(d); 626 } 627 } 628 629 static void update_refcount_discard(BlockDriverState *bs, 630 uint64_t offset, uint64_t length) 631 { 632 BDRVQcow2State *s = bs->opaque; 633 Qcow2DiscardRegion *d, *p, *next; 634 635 QTAILQ_FOREACH(d, &s->discards, next) { 636 uint64_t new_start = MIN(offset, d->offset); 637 uint64_t new_end = MAX(offset + length, d->offset + d->bytes); 638 639 if (new_end - new_start <= length + d->bytes) { 640 /* There can't be any overlap, areas ending up here have no 641 * references any more and therefore shouldn't get freed another 642 * time. */ 643 assert(d->bytes + length == new_end - new_start); 644 d->offset = new_start; 645 d->bytes = new_end - new_start; 646 goto found; 647 } 648 } 649 650 d = g_malloc(sizeof(*d)); 651 *d = (Qcow2DiscardRegion) { 652 .bs = bs, 653 .offset = offset, 654 .bytes = length, 655 }; 656 QTAILQ_INSERT_TAIL(&s->discards, d, next); 657 658 found: 659 /* Merge discard requests if they are adjacent now */ 660 QTAILQ_FOREACH_SAFE(p, &s->discards, next, next) { 661 if (p == d 662 || p->offset > d->offset + d->bytes 663 || d->offset > p->offset + p->bytes) 664 { 665 continue; 666 } 667 668 /* Still no overlap possible */ 669 assert(p->offset == d->offset + d->bytes 670 || d->offset == p->offset + p->bytes); 671 672 QTAILQ_REMOVE(&s->discards, p, next); 673 d->offset = MIN(d->offset, p->offset); 674 d->bytes += p->bytes; 675 g_free(p); 676 } 677 } 678 679 /* XXX: cache several refcount block clusters ? */ 680 /* @addend is the absolute value of the addend; if @decrease is set, @addend 681 * will be subtracted from the current refcount, otherwise it will be added */ 682 static int QEMU_WARN_UNUSED_RESULT update_refcount(BlockDriverState *bs, 683 int64_t offset, 684 int64_t length, 685 uint64_t addend, 686 bool decrease, 687 enum qcow2_discard_type type) 688 { 689 BDRVQcow2State *s = bs->opaque; 690 int64_t start, last, cluster_offset; 691 void *refcount_block = NULL; 692 int64_t old_table_index = -1; 693 int ret; 694 695 #ifdef DEBUG_ALLOC2 696 fprintf(stderr, "update_refcount: offset=%" PRId64 " size=%" PRId64 697 " addend=%s%" PRIu64 "\n", offset, length, decrease ? "-" : "", 698 addend); 699 #endif 700 if (length < 0) { 701 return -EINVAL; 702 } else if (length == 0) { 703 return 0; 704 } 705 706 if (decrease) { 707 qcow2_cache_set_dependency(bs, s->refcount_block_cache, 708 s->l2_table_cache); 709 } 710 711 start = start_of_cluster(s, offset); 712 last = start_of_cluster(s, offset + length - 1); 713 for(cluster_offset = start; cluster_offset <= last; 714 cluster_offset += s->cluster_size) 715 { 716 int block_index; 717 uint64_t refcount; 718 int64_t cluster_index = cluster_offset >> s->cluster_bits; 719 int64_t table_index = cluster_index >> s->refcount_block_bits; 720 721 /* Load the refcount block and allocate it if needed */ 722 if (table_index != old_table_index) { 723 if (refcount_block) { 724 qcow2_cache_put(bs, s->refcount_block_cache, &refcount_block); 725 } 726 ret = alloc_refcount_block(bs, cluster_index, &refcount_block); 727 if (ret < 0) { 728 goto fail; 729 } 730 } 731 old_table_index = table_index; 732 733 qcow2_cache_entry_mark_dirty(bs, s->refcount_block_cache, 734 refcount_block); 735 736 /* we can update the count and save it */ 737 block_index = cluster_index & (s->refcount_block_size - 1); 738 739 refcount = s->get_refcount(refcount_block, block_index); 740 if (decrease ? (refcount - addend > refcount) 741 : (refcount + addend < refcount || 742 refcount + addend > s->refcount_max)) 743 { 744 ret = -EINVAL; 745 goto fail; 746 } 747 if (decrease) { 748 refcount -= addend; 749 } else { 750 refcount += addend; 751 } 752 if (refcount == 0 && cluster_index < s->free_cluster_index) { 753 s->free_cluster_index = cluster_index; 754 } 755 s->set_refcount(refcount_block, block_index, refcount); 756 757 if (refcount == 0 && s->discard_passthrough[type]) { 758 update_refcount_discard(bs, cluster_offset, s->cluster_size); 759 } 760 } 761 762 ret = 0; 763 fail: 764 if (!s->cache_discards) { 765 qcow2_process_discards(bs, ret); 766 } 767 768 /* Write last changed block to disk */ 769 if (refcount_block) { 770 qcow2_cache_put(bs, s->refcount_block_cache, &refcount_block); 771 } 772 773 /* 774 * Try do undo any updates if an error is returned (This may succeed in 775 * some cases like ENOSPC for allocating a new refcount block) 776 */ 777 if (ret < 0) { 778 int dummy; 779 dummy = update_refcount(bs, offset, cluster_offset - offset, addend, 780 !decrease, QCOW2_DISCARD_NEVER); 781 (void)dummy; 782 } 783 784 return ret; 785 } 786 787 /* 788 * Increases or decreases the refcount of a given cluster. 789 * 790 * @addend is the absolute value of the addend; if @decrease is set, @addend 791 * will be subtracted from the current refcount, otherwise it will be added. 792 * 793 * On success 0 is returned; on failure -errno is returned. 794 */ 795 int qcow2_update_cluster_refcount(BlockDriverState *bs, 796 int64_t cluster_index, 797 uint64_t addend, bool decrease, 798 enum qcow2_discard_type type) 799 { 800 BDRVQcow2State *s = bs->opaque; 801 int ret; 802 803 ret = update_refcount(bs, cluster_index << s->cluster_bits, 1, addend, 804 decrease, type); 805 if (ret < 0) { 806 return ret; 807 } 808 809 return 0; 810 } 811 812 813 814 /*********************************************************/ 815 /* cluster allocation functions */ 816 817 818 819 /* return < 0 if error */ 820 static int64_t alloc_clusters_noref(BlockDriverState *bs, uint64_t size) 821 { 822 BDRVQcow2State *s = bs->opaque; 823 uint64_t i, nb_clusters, refcount; 824 int ret; 825 826 /* We can't allocate clusters if they may still be queued for discard. */ 827 if (s->cache_discards) { 828 qcow2_process_discards(bs, 0); 829 } 830 831 nb_clusters = size_to_clusters(s, size); 832 retry: 833 for(i = 0; i < nb_clusters; i++) { 834 uint64_t next_cluster_index = s->free_cluster_index++; 835 ret = qcow2_get_refcount(bs, next_cluster_index, &refcount); 836 837 if (ret < 0) { 838 return ret; 839 } else if (refcount != 0) { 840 goto retry; 841 } 842 } 843 844 /* Make sure that all offsets in the "allocated" range are representable 845 * in an int64_t */ 846 if (s->free_cluster_index > 0 && 847 s->free_cluster_index - 1 > (INT64_MAX >> s->cluster_bits)) 848 { 849 return -EFBIG; 850 } 851 852 #ifdef DEBUG_ALLOC2 853 fprintf(stderr, "alloc_clusters: size=%" PRId64 " -> %" PRId64 "\n", 854 size, 855 (s->free_cluster_index - nb_clusters) << s->cluster_bits); 856 #endif 857 return (s->free_cluster_index - nb_clusters) << s->cluster_bits; 858 } 859 860 int64_t qcow2_alloc_clusters(BlockDriverState *bs, uint64_t size) 861 { 862 int64_t offset; 863 int ret; 864 865 BLKDBG_EVENT(bs->file, BLKDBG_CLUSTER_ALLOC); 866 do { 867 offset = alloc_clusters_noref(bs, size); 868 if (offset < 0) { 869 return offset; 870 } 871 872 ret = update_refcount(bs, offset, size, 1, false, QCOW2_DISCARD_NEVER); 873 } while (ret == -EAGAIN); 874 875 if (ret < 0) { 876 return ret; 877 } 878 879 return offset; 880 } 881 882 int64_t qcow2_alloc_clusters_at(BlockDriverState *bs, uint64_t offset, 883 int64_t nb_clusters) 884 { 885 BDRVQcow2State *s = bs->opaque; 886 uint64_t cluster_index, refcount; 887 uint64_t i; 888 int ret; 889 890 assert(nb_clusters >= 0); 891 if (nb_clusters == 0) { 892 return 0; 893 } 894 895 do { 896 /* Check how many clusters there are free */ 897 cluster_index = offset >> s->cluster_bits; 898 for(i = 0; i < nb_clusters; i++) { 899 ret = qcow2_get_refcount(bs, cluster_index++, &refcount); 900 if (ret < 0) { 901 return ret; 902 } else if (refcount != 0) { 903 break; 904 } 905 } 906 907 /* And then allocate them */ 908 ret = update_refcount(bs, offset, i << s->cluster_bits, 1, false, 909 QCOW2_DISCARD_NEVER); 910 } while (ret == -EAGAIN); 911 912 if (ret < 0) { 913 return ret; 914 } 915 916 return i; 917 } 918 919 /* only used to allocate compressed sectors. We try to allocate 920 contiguous sectors. size must be <= cluster_size */ 921 int64_t qcow2_alloc_bytes(BlockDriverState *bs, int size) 922 { 923 BDRVQcow2State *s = bs->opaque; 924 int64_t offset; 925 size_t free_in_cluster; 926 int ret; 927 928 BLKDBG_EVENT(bs->file, BLKDBG_CLUSTER_ALLOC_BYTES); 929 assert(size > 0 && size <= s->cluster_size); 930 assert(!s->free_byte_offset || offset_into_cluster(s, s->free_byte_offset)); 931 932 offset = s->free_byte_offset; 933 934 if (offset) { 935 uint64_t refcount; 936 ret = qcow2_get_refcount(bs, offset >> s->cluster_bits, &refcount); 937 if (ret < 0) { 938 return ret; 939 } 940 941 if (refcount == s->refcount_max) { 942 offset = 0; 943 } 944 } 945 946 free_in_cluster = s->cluster_size - offset_into_cluster(s, offset); 947 do { 948 if (!offset || free_in_cluster < size) { 949 int64_t new_cluster = alloc_clusters_noref(bs, s->cluster_size); 950 if (new_cluster < 0) { 951 return new_cluster; 952 } 953 954 if (!offset || ROUND_UP(offset, s->cluster_size) != new_cluster) { 955 offset = new_cluster; 956 free_in_cluster = s->cluster_size; 957 } else { 958 free_in_cluster += s->cluster_size; 959 } 960 } 961 962 assert(offset); 963 ret = update_refcount(bs, offset, size, 1, false, QCOW2_DISCARD_NEVER); 964 if (ret < 0) { 965 offset = 0; 966 } 967 } while (ret == -EAGAIN); 968 if (ret < 0) { 969 return ret; 970 } 971 972 /* The cluster refcount was incremented; refcount blocks must be flushed 973 * before the caller's L2 table updates. */ 974 qcow2_cache_set_dependency(bs, s->l2_table_cache, s->refcount_block_cache); 975 976 s->free_byte_offset = offset + size; 977 if (!offset_into_cluster(s, s->free_byte_offset)) { 978 s->free_byte_offset = 0; 979 } 980 981 return offset; 982 } 983 984 void qcow2_free_clusters(BlockDriverState *bs, 985 int64_t offset, int64_t size, 986 enum qcow2_discard_type type) 987 { 988 int ret; 989 990 BLKDBG_EVENT(bs->file, BLKDBG_CLUSTER_FREE); 991 ret = update_refcount(bs, offset, size, 1, true, type); 992 if (ret < 0) { 993 fprintf(stderr, "qcow2_free_clusters failed: %s\n", strerror(-ret)); 994 /* TODO Remember the clusters to free them later and avoid leaking */ 995 } 996 } 997 998 /* 999 * Free a cluster using its L2 entry (handles clusters of all types, e.g. 1000 * normal cluster, compressed cluster, etc.) 1001 */ 1002 void qcow2_free_any_clusters(BlockDriverState *bs, uint64_t l2_entry, 1003 int nb_clusters, enum qcow2_discard_type type) 1004 { 1005 BDRVQcow2State *s = bs->opaque; 1006 1007 switch (qcow2_get_cluster_type(l2_entry)) { 1008 case QCOW2_CLUSTER_COMPRESSED: 1009 { 1010 int nb_csectors; 1011 nb_csectors = ((l2_entry >> s->csize_shift) & 1012 s->csize_mask) + 1; 1013 qcow2_free_clusters(bs, 1014 (l2_entry & s->cluster_offset_mask) & ~511, 1015 nb_csectors * 512, type); 1016 } 1017 break; 1018 case QCOW2_CLUSTER_NORMAL: 1019 case QCOW2_CLUSTER_ZERO: 1020 if (l2_entry & L2E_OFFSET_MASK) { 1021 if (offset_into_cluster(s, l2_entry & L2E_OFFSET_MASK)) { 1022 qcow2_signal_corruption(bs, false, -1, -1, 1023 "Cannot free unaligned cluster %#llx", 1024 l2_entry & L2E_OFFSET_MASK); 1025 } else { 1026 qcow2_free_clusters(bs, l2_entry & L2E_OFFSET_MASK, 1027 nb_clusters << s->cluster_bits, type); 1028 } 1029 } 1030 break; 1031 case QCOW2_CLUSTER_UNALLOCATED: 1032 break; 1033 default: 1034 abort(); 1035 } 1036 } 1037 1038 1039 1040 /*********************************************************/ 1041 /* snapshots and image creation */ 1042 1043 1044 1045 /* update the refcounts of snapshots and the copied flag */ 1046 int qcow2_update_snapshot_refcount(BlockDriverState *bs, 1047 int64_t l1_table_offset, int l1_size, int addend) 1048 { 1049 BDRVQcow2State *s = bs->opaque; 1050 uint64_t *l1_table, *l2_table, l2_offset, offset, l1_size2, refcount; 1051 bool l1_allocated = false; 1052 int64_t old_offset, old_l2_offset; 1053 int i, j, l1_modified = 0, nb_csectors; 1054 int ret; 1055 1056 assert(addend >= -1 && addend <= 1); 1057 1058 l2_table = NULL; 1059 l1_table = NULL; 1060 l1_size2 = l1_size * sizeof(uint64_t); 1061 1062 s->cache_discards = true; 1063 1064 /* WARNING: qcow2_snapshot_goto relies on this function not using the 1065 * l1_table_offset when it is the current s->l1_table_offset! Be careful 1066 * when changing this! */ 1067 if (l1_table_offset != s->l1_table_offset) { 1068 l1_table = g_try_malloc0(align_offset(l1_size2, 512)); 1069 if (l1_size2 && l1_table == NULL) { 1070 ret = -ENOMEM; 1071 goto fail; 1072 } 1073 l1_allocated = true; 1074 1075 ret = bdrv_pread(bs->file->bs, l1_table_offset, l1_table, l1_size2); 1076 if (ret < 0) { 1077 goto fail; 1078 } 1079 1080 for(i = 0;i < l1_size; i++) 1081 be64_to_cpus(&l1_table[i]); 1082 } else { 1083 assert(l1_size == s->l1_size); 1084 l1_table = s->l1_table; 1085 l1_allocated = false; 1086 } 1087 1088 for(i = 0; i < l1_size; i++) { 1089 l2_offset = l1_table[i]; 1090 if (l2_offset) { 1091 old_l2_offset = l2_offset; 1092 l2_offset &= L1E_OFFSET_MASK; 1093 1094 if (offset_into_cluster(s, l2_offset)) { 1095 qcow2_signal_corruption(bs, true, -1, -1, "L2 table offset %#" 1096 PRIx64 " unaligned (L1 index: %#x)", 1097 l2_offset, i); 1098 ret = -EIO; 1099 goto fail; 1100 } 1101 1102 ret = qcow2_cache_get(bs, s->l2_table_cache, l2_offset, 1103 (void**) &l2_table); 1104 if (ret < 0) { 1105 goto fail; 1106 } 1107 1108 for(j = 0; j < s->l2_size; j++) { 1109 uint64_t cluster_index; 1110 1111 offset = be64_to_cpu(l2_table[j]); 1112 old_offset = offset; 1113 offset &= ~QCOW_OFLAG_COPIED; 1114 1115 switch (qcow2_get_cluster_type(offset)) { 1116 case QCOW2_CLUSTER_COMPRESSED: 1117 nb_csectors = ((offset >> s->csize_shift) & 1118 s->csize_mask) + 1; 1119 if (addend != 0) { 1120 ret = update_refcount(bs, 1121 (offset & s->cluster_offset_mask) & ~511, 1122 nb_csectors * 512, abs(addend), addend < 0, 1123 QCOW2_DISCARD_SNAPSHOT); 1124 if (ret < 0) { 1125 goto fail; 1126 } 1127 } 1128 /* compressed clusters are never modified */ 1129 refcount = 2; 1130 break; 1131 1132 case QCOW2_CLUSTER_NORMAL: 1133 case QCOW2_CLUSTER_ZERO: 1134 if (offset_into_cluster(s, offset & L2E_OFFSET_MASK)) { 1135 qcow2_signal_corruption(bs, true, -1, -1, "Data " 1136 "cluster offset %#llx " 1137 "unaligned (L2 offset: %#" 1138 PRIx64 ", L2 index: %#x)", 1139 offset & L2E_OFFSET_MASK, 1140 l2_offset, j); 1141 ret = -EIO; 1142 goto fail; 1143 } 1144 1145 cluster_index = (offset & L2E_OFFSET_MASK) >> s->cluster_bits; 1146 if (!cluster_index) { 1147 /* unallocated */ 1148 refcount = 0; 1149 break; 1150 } 1151 if (addend != 0) { 1152 ret = qcow2_update_cluster_refcount(bs, 1153 cluster_index, abs(addend), addend < 0, 1154 QCOW2_DISCARD_SNAPSHOT); 1155 if (ret < 0) { 1156 goto fail; 1157 } 1158 } 1159 1160 ret = qcow2_get_refcount(bs, cluster_index, &refcount); 1161 if (ret < 0) { 1162 goto fail; 1163 } 1164 break; 1165 1166 case QCOW2_CLUSTER_UNALLOCATED: 1167 refcount = 0; 1168 break; 1169 1170 default: 1171 abort(); 1172 } 1173 1174 if (refcount == 1) { 1175 offset |= QCOW_OFLAG_COPIED; 1176 } 1177 if (offset != old_offset) { 1178 if (addend > 0) { 1179 qcow2_cache_set_dependency(bs, s->l2_table_cache, 1180 s->refcount_block_cache); 1181 } 1182 l2_table[j] = cpu_to_be64(offset); 1183 qcow2_cache_entry_mark_dirty(bs, s->l2_table_cache, 1184 l2_table); 1185 } 1186 } 1187 1188 qcow2_cache_put(bs, s->l2_table_cache, (void **) &l2_table); 1189 1190 if (addend != 0) { 1191 ret = qcow2_update_cluster_refcount(bs, l2_offset >> 1192 s->cluster_bits, 1193 abs(addend), addend < 0, 1194 QCOW2_DISCARD_SNAPSHOT); 1195 if (ret < 0) { 1196 goto fail; 1197 } 1198 } 1199 ret = qcow2_get_refcount(bs, l2_offset >> s->cluster_bits, 1200 &refcount); 1201 if (ret < 0) { 1202 goto fail; 1203 } else if (refcount == 1) { 1204 l2_offset |= QCOW_OFLAG_COPIED; 1205 } 1206 if (l2_offset != old_l2_offset) { 1207 l1_table[i] = l2_offset; 1208 l1_modified = 1; 1209 } 1210 } 1211 } 1212 1213 ret = bdrv_flush(bs); 1214 fail: 1215 if (l2_table) { 1216 qcow2_cache_put(bs, s->l2_table_cache, (void**) &l2_table); 1217 } 1218 1219 s->cache_discards = false; 1220 qcow2_process_discards(bs, ret); 1221 1222 /* Update L1 only if it isn't deleted anyway (addend = -1) */ 1223 if (ret == 0 && addend >= 0 && l1_modified) { 1224 for (i = 0; i < l1_size; i++) { 1225 cpu_to_be64s(&l1_table[i]); 1226 } 1227 1228 ret = bdrv_pwrite_sync(bs->file->bs, l1_table_offset, 1229 l1_table, l1_size2); 1230 1231 for (i = 0; i < l1_size; i++) { 1232 be64_to_cpus(&l1_table[i]); 1233 } 1234 } 1235 if (l1_allocated) 1236 g_free(l1_table); 1237 return ret; 1238 } 1239 1240 1241 1242 1243 /*********************************************************/ 1244 /* refcount checking functions */ 1245 1246 1247 static uint64_t refcount_array_byte_size(BDRVQcow2State *s, uint64_t entries) 1248 { 1249 /* This assertion holds because there is no way we can address more than 1250 * 2^(64 - 9) clusters at once (with cluster size 512 = 2^9, and because 1251 * offsets have to be representable in bytes); due to every cluster 1252 * corresponding to one refcount entry, we are well below that limit */ 1253 assert(entries < (UINT64_C(1) << (64 - 9))); 1254 1255 /* Thanks to the assertion this will not overflow, because 1256 * s->refcount_order < 7. 1257 * (note: x << s->refcount_order == x * s->refcount_bits) */ 1258 return DIV_ROUND_UP(entries << s->refcount_order, 8); 1259 } 1260 1261 /** 1262 * Reallocates *array so that it can hold new_size entries. *size must contain 1263 * the current number of entries in *array. If the reallocation fails, *array 1264 * and *size will not be modified and -errno will be returned. If the 1265 * reallocation is successful, *array will be set to the new buffer, *size 1266 * will be set to new_size and 0 will be returned. The size of the reallocated 1267 * refcount array buffer will be aligned to a cluster boundary, and the newly 1268 * allocated area will be zeroed. 1269 */ 1270 static int realloc_refcount_array(BDRVQcow2State *s, void **array, 1271 int64_t *size, int64_t new_size) 1272 { 1273 int64_t old_byte_size, new_byte_size; 1274 void *new_ptr; 1275 1276 /* Round to clusters so the array can be directly written to disk */ 1277 old_byte_size = size_to_clusters(s, refcount_array_byte_size(s, *size)) 1278 * s->cluster_size; 1279 new_byte_size = size_to_clusters(s, refcount_array_byte_size(s, new_size)) 1280 * s->cluster_size; 1281 1282 if (new_byte_size == old_byte_size) { 1283 *size = new_size; 1284 return 0; 1285 } 1286 1287 assert(new_byte_size > 0); 1288 1289 if (new_byte_size > SIZE_MAX) { 1290 return -ENOMEM; 1291 } 1292 1293 new_ptr = g_try_realloc(*array, new_byte_size); 1294 if (!new_ptr) { 1295 return -ENOMEM; 1296 } 1297 1298 if (new_byte_size > old_byte_size) { 1299 memset((char *)new_ptr + old_byte_size, 0, 1300 new_byte_size - old_byte_size); 1301 } 1302 1303 *array = new_ptr; 1304 *size = new_size; 1305 1306 return 0; 1307 } 1308 1309 /* 1310 * Increases the refcount for a range of clusters in a given refcount table. 1311 * This is used to construct a temporary refcount table out of L1 and L2 tables 1312 * which can be compared to the refcount table saved in the image. 1313 * 1314 * Modifies the number of errors in res. 1315 */ 1316 static int inc_refcounts(BlockDriverState *bs, 1317 BdrvCheckResult *res, 1318 void **refcount_table, 1319 int64_t *refcount_table_size, 1320 int64_t offset, int64_t size) 1321 { 1322 BDRVQcow2State *s = bs->opaque; 1323 uint64_t start, last, cluster_offset, k, refcount; 1324 int ret; 1325 1326 if (size <= 0) { 1327 return 0; 1328 } 1329 1330 start = start_of_cluster(s, offset); 1331 last = start_of_cluster(s, offset + size - 1); 1332 for(cluster_offset = start; cluster_offset <= last; 1333 cluster_offset += s->cluster_size) { 1334 k = cluster_offset >> s->cluster_bits; 1335 if (k >= *refcount_table_size) { 1336 ret = realloc_refcount_array(s, refcount_table, 1337 refcount_table_size, k + 1); 1338 if (ret < 0) { 1339 res->check_errors++; 1340 return ret; 1341 } 1342 } 1343 1344 refcount = s->get_refcount(*refcount_table, k); 1345 if (refcount == s->refcount_max) { 1346 fprintf(stderr, "ERROR: overflow cluster offset=0x%" PRIx64 1347 "\n", cluster_offset); 1348 res->corruptions++; 1349 continue; 1350 } 1351 s->set_refcount(*refcount_table, k, refcount + 1); 1352 } 1353 1354 return 0; 1355 } 1356 1357 /* Flags for check_refcounts_l1() and check_refcounts_l2() */ 1358 enum { 1359 CHECK_FRAG_INFO = 0x2, /* update BlockFragInfo counters */ 1360 }; 1361 1362 /* 1363 * Increases the refcount in the given refcount table for the all clusters 1364 * referenced in the L2 table. While doing so, performs some checks on L2 1365 * entries. 1366 * 1367 * Returns the number of errors found by the checks or -errno if an internal 1368 * error occurred. 1369 */ 1370 static int check_refcounts_l2(BlockDriverState *bs, BdrvCheckResult *res, 1371 void **refcount_table, 1372 int64_t *refcount_table_size, int64_t l2_offset, 1373 int flags) 1374 { 1375 BDRVQcow2State *s = bs->opaque; 1376 uint64_t *l2_table, l2_entry; 1377 uint64_t next_contiguous_offset = 0; 1378 int i, l2_size, nb_csectors, ret; 1379 1380 /* Read L2 table from disk */ 1381 l2_size = s->l2_size * sizeof(uint64_t); 1382 l2_table = g_malloc(l2_size); 1383 1384 ret = bdrv_pread(bs->file->bs, l2_offset, l2_table, l2_size); 1385 if (ret < 0) { 1386 fprintf(stderr, "ERROR: I/O error in check_refcounts_l2\n"); 1387 res->check_errors++; 1388 goto fail; 1389 } 1390 1391 /* Do the actual checks */ 1392 for(i = 0; i < s->l2_size; i++) { 1393 l2_entry = be64_to_cpu(l2_table[i]); 1394 1395 switch (qcow2_get_cluster_type(l2_entry)) { 1396 case QCOW2_CLUSTER_COMPRESSED: 1397 /* Compressed clusters don't have QCOW_OFLAG_COPIED */ 1398 if (l2_entry & QCOW_OFLAG_COPIED) { 1399 fprintf(stderr, "ERROR: cluster %" PRId64 ": " 1400 "copied flag must never be set for compressed " 1401 "clusters\n", l2_entry >> s->cluster_bits); 1402 l2_entry &= ~QCOW_OFLAG_COPIED; 1403 res->corruptions++; 1404 } 1405 1406 /* Mark cluster as used */ 1407 nb_csectors = ((l2_entry >> s->csize_shift) & 1408 s->csize_mask) + 1; 1409 l2_entry &= s->cluster_offset_mask; 1410 ret = inc_refcounts(bs, res, refcount_table, refcount_table_size, 1411 l2_entry & ~511, nb_csectors * 512); 1412 if (ret < 0) { 1413 goto fail; 1414 } 1415 1416 if (flags & CHECK_FRAG_INFO) { 1417 res->bfi.allocated_clusters++; 1418 res->bfi.compressed_clusters++; 1419 1420 /* Compressed clusters are fragmented by nature. Since they 1421 * take up sub-sector space but we only have sector granularity 1422 * I/O we need to re-read the same sectors even for adjacent 1423 * compressed clusters. 1424 */ 1425 res->bfi.fragmented_clusters++; 1426 } 1427 break; 1428 1429 case QCOW2_CLUSTER_ZERO: 1430 if ((l2_entry & L2E_OFFSET_MASK) == 0) { 1431 break; 1432 } 1433 /* fall through */ 1434 1435 case QCOW2_CLUSTER_NORMAL: 1436 { 1437 uint64_t offset = l2_entry & L2E_OFFSET_MASK; 1438 1439 if (flags & CHECK_FRAG_INFO) { 1440 res->bfi.allocated_clusters++; 1441 if (next_contiguous_offset && 1442 offset != next_contiguous_offset) { 1443 res->bfi.fragmented_clusters++; 1444 } 1445 next_contiguous_offset = offset + s->cluster_size; 1446 } 1447 1448 /* Mark cluster as used */ 1449 ret = inc_refcounts(bs, res, refcount_table, refcount_table_size, 1450 offset, s->cluster_size); 1451 if (ret < 0) { 1452 goto fail; 1453 } 1454 1455 /* Correct offsets are cluster aligned */ 1456 if (offset_into_cluster(s, offset)) { 1457 fprintf(stderr, "ERROR offset=%" PRIx64 ": Cluster is not " 1458 "properly aligned; L2 entry corrupted.\n", offset); 1459 res->corruptions++; 1460 } 1461 break; 1462 } 1463 1464 case QCOW2_CLUSTER_UNALLOCATED: 1465 break; 1466 1467 default: 1468 abort(); 1469 } 1470 } 1471 1472 g_free(l2_table); 1473 return 0; 1474 1475 fail: 1476 g_free(l2_table); 1477 return ret; 1478 } 1479 1480 /* 1481 * Increases the refcount for the L1 table, its L2 tables and all referenced 1482 * clusters in the given refcount table. While doing so, performs some checks 1483 * on L1 and L2 entries. 1484 * 1485 * Returns the number of errors found by the checks or -errno if an internal 1486 * error occurred. 1487 */ 1488 static int check_refcounts_l1(BlockDriverState *bs, 1489 BdrvCheckResult *res, 1490 void **refcount_table, 1491 int64_t *refcount_table_size, 1492 int64_t l1_table_offset, int l1_size, 1493 int flags) 1494 { 1495 BDRVQcow2State *s = bs->opaque; 1496 uint64_t *l1_table = NULL, l2_offset, l1_size2; 1497 int i, ret; 1498 1499 l1_size2 = l1_size * sizeof(uint64_t); 1500 1501 /* Mark L1 table as used */ 1502 ret = inc_refcounts(bs, res, refcount_table, refcount_table_size, 1503 l1_table_offset, l1_size2); 1504 if (ret < 0) { 1505 goto fail; 1506 } 1507 1508 /* Read L1 table entries from disk */ 1509 if (l1_size2 > 0) { 1510 l1_table = g_try_malloc(l1_size2); 1511 if (l1_table == NULL) { 1512 ret = -ENOMEM; 1513 res->check_errors++; 1514 goto fail; 1515 } 1516 ret = bdrv_pread(bs->file->bs, l1_table_offset, l1_table, l1_size2); 1517 if (ret < 0) { 1518 fprintf(stderr, "ERROR: I/O error in check_refcounts_l1\n"); 1519 res->check_errors++; 1520 goto fail; 1521 } 1522 for(i = 0;i < l1_size; i++) 1523 be64_to_cpus(&l1_table[i]); 1524 } 1525 1526 /* Do the actual checks */ 1527 for(i = 0; i < l1_size; i++) { 1528 l2_offset = l1_table[i]; 1529 if (l2_offset) { 1530 /* Mark L2 table as used */ 1531 l2_offset &= L1E_OFFSET_MASK; 1532 ret = inc_refcounts(bs, res, refcount_table, refcount_table_size, 1533 l2_offset, s->cluster_size); 1534 if (ret < 0) { 1535 goto fail; 1536 } 1537 1538 /* L2 tables are cluster aligned */ 1539 if (offset_into_cluster(s, l2_offset)) { 1540 fprintf(stderr, "ERROR l2_offset=%" PRIx64 ": Table is not " 1541 "cluster aligned; L1 entry corrupted\n", l2_offset); 1542 res->corruptions++; 1543 } 1544 1545 /* Process and check L2 entries */ 1546 ret = check_refcounts_l2(bs, res, refcount_table, 1547 refcount_table_size, l2_offset, flags); 1548 if (ret < 0) { 1549 goto fail; 1550 } 1551 } 1552 } 1553 g_free(l1_table); 1554 return 0; 1555 1556 fail: 1557 g_free(l1_table); 1558 return ret; 1559 } 1560 1561 /* 1562 * Checks the OFLAG_COPIED flag for all L1 and L2 entries. 1563 * 1564 * This function does not print an error message nor does it increment 1565 * check_errors if qcow2_get_refcount fails (this is because such an error will 1566 * have been already detected and sufficiently signaled by the calling function 1567 * (qcow2_check_refcounts) by the time this function is called). 1568 */ 1569 static int check_oflag_copied(BlockDriverState *bs, BdrvCheckResult *res, 1570 BdrvCheckMode fix) 1571 { 1572 BDRVQcow2State *s = bs->opaque; 1573 uint64_t *l2_table = qemu_blockalign(bs, s->cluster_size); 1574 int ret; 1575 uint64_t refcount; 1576 int i, j; 1577 1578 for (i = 0; i < s->l1_size; i++) { 1579 uint64_t l1_entry = s->l1_table[i]; 1580 uint64_t l2_offset = l1_entry & L1E_OFFSET_MASK; 1581 bool l2_dirty = false; 1582 1583 if (!l2_offset) { 1584 continue; 1585 } 1586 1587 ret = qcow2_get_refcount(bs, l2_offset >> s->cluster_bits, 1588 &refcount); 1589 if (ret < 0) { 1590 /* don't print message nor increment check_errors */ 1591 continue; 1592 } 1593 if ((refcount == 1) != ((l1_entry & QCOW_OFLAG_COPIED) != 0)) { 1594 fprintf(stderr, "%s OFLAG_COPIED L2 cluster: l1_index=%d " 1595 "l1_entry=%" PRIx64 " refcount=%" PRIu64 "\n", 1596 fix & BDRV_FIX_ERRORS ? "Repairing" : 1597 "ERROR", 1598 i, l1_entry, refcount); 1599 if (fix & BDRV_FIX_ERRORS) { 1600 s->l1_table[i] = refcount == 1 1601 ? l1_entry | QCOW_OFLAG_COPIED 1602 : l1_entry & ~QCOW_OFLAG_COPIED; 1603 ret = qcow2_write_l1_entry(bs, i); 1604 if (ret < 0) { 1605 res->check_errors++; 1606 goto fail; 1607 } 1608 res->corruptions_fixed++; 1609 } else { 1610 res->corruptions++; 1611 } 1612 } 1613 1614 ret = bdrv_pread(bs->file->bs, l2_offset, l2_table, 1615 s->l2_size * sizeof(uint64_t)); 1616 if (ret < 0) { 1617 fprintf(stderr, "ERROR: Could not read L2 table: %s\n", 1618 strerror(-ret)); 1619 res->check_errors++; 1620 goto fail; 1621 } 1622 1623 for (j = 0; j < s->l2_size; j++) { 1624 uint64_t l2_entry = be64_to_cpu(l2_table[j]); 1625 uint64_t data_offset = l2_entry & L2E_OFFSET_MASK; 1626 int cluster_type = qcow2_get_cluster_type(l2_entry); 1627 1628 if ((cluster_type == QCOW2_CLUSTER_NORMAL) || 1629 ((cluster_type == QCOW2_CLUSTER_ZERO) && (data_offset != 0))) { 1630 ret = qcow2_get_refcount(bs, 1631 data_offset >> s->cluster_bits, 1632 &refcount); 1633 if (ret < 0) { 1634 /* don't print message nor increment check_errors */ 1635 continue; 1636 } 1637 if ((refcount == 1) != ((l2_entry & QCOW_OFLAG_COPIED) != 0)) { 1638 fprintf(stderr, "%s OFLAG_COPIED data cluster: " 1639 "l2_entry=%" PRIx64 " refcount=%" PRIu64 "\n", 1640 fix & BDRV_FIX_ERRORS ? "Repairing" : 1641 "ERROR", 1642 l2_entry, refcount); 1643 if (fix & BDRV_FIX_ERRORS) { 1644 l2_table[j] = cpu_to_be64(refcount == 1 1645 ? l2_entry | QCOW_OFLAG_COPIED 1646 : l2_entry & ~QCOW_OFLAG_COPIED); 1647 l2_dirty = true; 1648 res->corruptions_fixed++; 1649 } else { 1650 res->corruptions++; 1651 } 1652 } 1653 } 1654 } 1655 1656 if (l2_dirty) { 1657 ret = qcow2_pre_write_overlap_check(bs, QCOW2_OL_ACTIVE_L2, 1658 l2_offset, s->cluster_size); 1659 if (ret < 0) { 1660 fprintf(stderr, "ERROR: Could not write L2 table; metadata " 1661 "overlap check failed: %s\n", strerror(-ret)); 1662 res->check_errors++; 1663 goto fail; 1664 } 1665 1666 ret = bdrv_pwrite(bs->file->bs, l2_offset, l2_table, 1667 s->cluster_size); 1668 if (ret < 0) { 1669 fprintf(stderr, "ERROR: Could not write L2 table: %s\n", 1670 strerror(-ret)); 1671 res->check_errors++; 1672 goto fail; 1673 } 1674 } 1675 } 1676 1677 ret = 0; 1678 1679 fail: 1680 qemu_vfree(l2_table); 1681 return ret; 1682 } 1683 1684 /* 1685 * Checks consistency of refblocks and accounts for each refblock in 1686 * *refcount_table. 1687 */ 1688 static int check_refblocks(BlockDriverState *bs, BdrvCheckResult *res, 1689 BdrvCheckMode fix, bool *rebuild, 1690 void **refcount_table, int64_t *nb_clusters) 1691 { 1692 BDRVQcow2State *s = bs->opaque; 1693 int64_t i, size; 1694 int ret; 1695 1696 for(i = 0; i < s->refcount_table_size; i++) { 1697 uint64_t offset, cluster; 1698 offset = s->refcount_table[i]; 1699 cluster = offset >> s->cluster_bits; 1700 1701 /* Refcount blocks are cluster aligned */ 1702 if (offset_into_cluster(s, offset)) { 1703 fprintf(stderr, "ERROR refcount block %" PRId64 " is not " 1704 "cluster aligned; refcount table entry corrupted\n", i); 1705 res->corruptions++; 1706 *rebuild = true; 1707 continue; 1708 } 1709 1710 if (cluster >= *nb_clusters) { 1711 fprintf(stderr, "%s refcount block %" PRId64 " is outside image\n", 1712 fix & BDRV_FIX_ERRORS ? "Repairing" : "ERROR", i); 1713 1714 if (fix & BDRV_FIX_ERRORS) { 1715 int64_t new_nb_clusters; 1716 1717 if (offset > INT64_MAX - s->cluster_size) { 1718 ret = -EINVAL; 1719 goto resize_fail; 1720 } 1721 1722 ret = bdrv_truncate(bs->file->bs, offset + s->cluster_size); 1723 if (ret < 0) { 1724 goto resize_fail; 1725 } 1726 size = bdrv_getlength(bs->file->bs); 1727 if (size < 0) { 1728 ret = size; 1729 goto resize_fail; 1730 } 1731 1732 new_nb_clusters = size_to_clusters(s, size); 1733 assert(new_nb_clusters >= *nb_clusters); 1734 1735 ret = realloc_refcount_array(s, refcount_table, 1736 nb_clusters, new_nb_clusters); 1737 if (ret < 0) { 1738 res->check_errors++; 1739 return ret; 1740 } 1741 1742 if (cluster >= *nb_clusters) { 1743 ret = -EINVAL; 1744 goto resize_fail; 1745 } 1746 1747 res->corruptions_fixed++; 1748 ret = inc_refcounts(bs, res, refcount_table, nb_clusters, 1749 offset, s->cluster_size); 1750 if (ret < 0) { 1751 return ret; 1752 } 1753 /* No need to check whether the refcount is now greater than 1: 1754 * This area was just allocated and zeroed, so it can only be 1755 * exactly 1 after inc_refcounts() */ 1756 continue; 1757 1758 resize_fail: 1759 res->corruptions++; 1760 *rebuild = true; 1761 fprintf(stderr, "ERROR could not resize image: %s\n", 1762 strerror(-ret)); 1763 } else { 1764 res->corruptions++; 1765 } 1766 continue; 1767 } 1768 1769 if (offset != 0) { 1770 ret = inc_refcounts(bs, res, refcount_table, nb_clusters, 1771 offset, s->cluster_size); 1772 if (ret < 0) { 1773 return ret; 1774 } 1775 if (s->get_refcount(*refcount_table, cluster) != 1) { 1776 fprintf(stderr, "ERROR refcount block %" PRId64 1777 " refcount=%" PRIu64 "\n", i, 1778 s->get_refcount(*refcount_table, cluster)); 1779 res->corruptions++; 1780 *rebuild = true; 1781 } 1782 } 1783 } 1784 1785 return 0; 1786 } 1787 1788 /* 1789 * Calculates an in-memory refcount table. 1790 */ 1791 static int calculate_refcounts(BlockDriverState *bs, BdrvCheckResult *res, 1792 BdrvCheckMode fix, bool *rebuild, 1793 void **refcount_table, int64_t *nb_clusters) 1794 { 1795 BDRVQcow2State *s = bs->opaque; 1796 int64_t i; 1797 QCowSnapshot *sn; 1798 int ret; 1799 1800 if (!*refcount_table) { 1801 int64_t old_size = 0; 1802 ret = realloc_refcount_array(s, refcount_table, 1803 &old_size, *nb_clusters); 1804 if (ret < 0) { 1805 res->check_errors++; 1806 return ret; 1807 } 1808 } 1809 1810 /* header */ 1811 ret = inc_refcounts(bs, res, refcount_table, nb_clusters, 1812 0, s->cluster_size); 1813 if (ret < 0) { 1814 return ret; 1815 } 1816 1817 /* current L1 table */ 1818 ret = check_refcounts_l1(bs, res, refcount_table, nb_clusters, 1819 s->l1_table_offset, s->l1_size, CHECK_FRAG_INFO); 1820 if (ret < 0) { 1821 return ret; 1822 } 1823 1824 /* snapshots */ 1825 for (i = 0; i < s->nb_snapshots; i++) { 1826 sn = s->snapshots + i; 1827 ret = check_refcounts_l1(bs, res, refcount_table, nb_clusters, 1828 sn->l1_table_offset, sn->l1_size, 0); 1829 if (ret < 0) { 1830 return ret; 1831 } 1832 } 1833 ret = inc_refcounts(bs, res, refcount_table, nb_clusters, 1834 s->snapshots_offset, s->snapshots_size); 1835 if (ret < 0) { 1836 return ret; 1837 } 1838 1839 /* refcount data */ 1840 ret = inc_refcounts(bs, res, refcount_table, nb_clusters, 1841 s->refcount_table_offset, 1842 s->refcount_table_size * sizeof(uint64_t)); 1843 if (ret < 0) { 1844 return ret; 1845 } 1846 1847 return check_refblocks(bs, res, fix, rebuild, refcount_table, nb_clusters); 1848 } 1849 1850 /* 1851 * Compares the actual reference count for each cluster in the image against the 1852 * refcount as reported by the refcount structures on-disk. 1853 */ 1854 static void compare_refcounts(BlockDriverState *bs, BdrvCheckResult *res, 1855 BdrvCheckMode fix, bool *rebuild, 1856 int64_t *highest_cluster, 1857 void *refcount_table, int64_t nb_clusters) 1858 { 1859 BDRVQcow2State *s = bs->opaque; 1860 int64_t i; 1861 uint64_t refcount1, refcount2; 1862 int ret; 1863 1864 for (i = 0, *highest_cluster = 0; i < nb_clusters; i++) { 1865 ret = qcow2_get_refcount(bs, i, &refcount1); 1866 if (ret < 0) { 1867 fprintf(stderr, "Can't get refcount for cluster %" PRId64 ": %s\n", 1868 i, strerror(-ret)); 1869 res->check_errors++; 1870 continue; 1871 } 1872 1873 refcount2 = s->get_refcount(refcount_table, i); 1874 1875 if (refcount1 > 0 || refcount2 > 0) { 1876 *highest_cluster = i; 1877 } 1878 1879 if (refcount1 != refcount2) { 1880 /* Check if we're allowed to fix the mismatch */ 1881 int *num_fixed = NULL; 1882 if (refcount1 == 0) { 1883 *rebuild = true; 1884 } else if (refcount1 > refcount2 && (fix & BDRV_FIX_LEAKS)) { 1885 num_fixed = &res->leaks_fixed; 1886 } else if (refcount1 < refcount2 && (fix & BDRV_FIX_ERRORS)) { 1887 num_fixed = &res->corruptions_fixed; 1888 } 1889 1890 fprintf(stderr, "%s cluster %" PRId64 " refcount=%" PRIu64 1891 " reference=%" PRIu64 "\n", 1892 num_fixed != NULL ? "Repairing" : 1893 refcount1 < refcount2 ? "ERROR" : 1894 "Leaked", 1895 i, refcount1, refcount2); 1896 1897 if (num_fixed) { 1898 ret = update_refcount(bs, i << s->cluster_bits, 1, 1899 refcount_diff(refcount1, refcount2), 1900 refcount1 > refcount2, 1901 QCOW2_DISCARD_ALWAYS); 1902 if (ret >= 0) { 1903 (*num_fixed)++; 1904 continue; 1905 } 1906 } 1907 1908 /* And if we couldn't, print an error */ 1909 if (refcount1 < refcount2) { 1910 res->corruptions++; 1911 } else { 1912 res->leaks++; 1913 } 1914 } 1915 } 1916 } 1917 1918 /* 1919 * Allocates clusters using an in-memory refcount table (IMRT) in contrast to 1920 * the on-disk refcount structures. 1921 * 1922 * On input, *first_free_cluster tells where to start looking, and need not 1923 * actually be a free cluster; the returned offset will not be before that 1924 * cluster. On output, *first_free_cluster points to the first gap found, even 1925 * if that gap was too small to be used as the returned offset. 1926 * 1927 * Note that *first_free_cluster is a cluster index whereas the return value is 1928 * an offset. 1929 */ 1930 static int64_t alloc_clusters_imrt(BlockDriverState *bs, 1931 int cluster_count, 1932 void **refcount_table, 1933 int64_t *imrt_nb_clusters, 1934 int64_t *first_free_cluster) 1935 { 1936 BDRVQcow2State *s = bs->opaque; 1937 int64_t cluster = *first_free_cluster, i; 1938 bool first_gap = true; 1939 int contiguous_free_clusters; 1940 int ret; 1941 1942 /* Starting at *first_free_cluster, find a range of at least cluster_count 1943 * continuously free clusters */ 1944 for (contiguous_free_clusters = 0; 1945 cluster < *imrt_nb_clusters && 1946 contiguous_free_clusters < cluster_count; 1947 cluster++) 1948 { 1949 if (!s->get_refcount(*refcount_table, cluster)) { 1950 contiguous_free_clusters++; 1951 if (first_gap) { 1952 /* If this is the first free cluster found, update 1953 * *first_free_cluster accordingly */ 1954 *first_free_cluster = cluster; 1955 first_gap = false; 1956 } 1957 } else if (contiguous_free_clusters) { 1958 contiguous_free_clusters = 0; 1959 } 1960 } 1961 1962 /* If contiguous_free_clusters is greater than zero, it contains the number 1963 * of continuously free clusters until the current cluster; the first free 1964 * cluster in the current "gap" is therefore 1965 * cluster - contiguous_free_clusters */ 1966 1967 /* If no such range could be found, grow the in-memory refcount table 1968 * accordingly to append free clusters at the end of the image */ 1969 if (contiguous_free_clusters < cluster_count) { 1970 /* contiguous_free_clusters clusters are already empty at the image end; 1971 * we need cluster_count clusters; therefore, we have to allocate 1972 * cluster_count - contiguous_free_clusters new clusters at the end of 1973 * the image (which is the current value of cluster; note that cluster 1974 * may exceed old_imrt_nb_clusters if *first_free_cluster pointed beyond 1975 * the image end) */ 1976 ret = realloc_refcount_array(s, refcount_table, imrt_nb_clusters, 1977 cluster + cluster_count 1978 - contiguous_free_clusters); 1979 if (ret < 0) { 1980 return ret; 1981 } 1982 } 1983 1984 /* Go back to the first free cluster */ 1985 cluster -= contiguous_free_clusters; 1986 for (i = 0; i < cluster_count; i++) { 1987 s->set_refcount(*refcount_table, cluster + i, 1); 1988 } 1989 1990 return cluster << s->cluster_bits; 1991 } 1992 1993 /* 1994 * Creates a new refcount structure based solely on the in-memory information 1995 * given through *refcount_table. All necessary allocations will be reflected 1996 * in that array. 1997 * 1998 * On success, the old refcount structure is leaked (it will be covered by the 1999 * new refcount structure). 2000 */ 2001 static int rebuild_refcount_structure(BlockDriverState *bs, 2002 BdrvCheckResult *res, 2003 void **refcount_table, 2004 int64_t *nb_clusters) 2005 { 2006 BDRVQcow2State *s = bs->opaque; 2007 int64_t first_free_cluster = 0, reftable_offset = -1, cluster = 0; 2008 int64_t refblock_offset, refblock_start, refblock_index; 2009 uint32_t reftable_size = 0; 2010 uint64_t *on_disk_reftable = NULL; 2011 void *on_disk_refblock; 2012 int ret = 0; 2013 struct { 2014 uint64_t reftable_offset; 2015 uint32_t reftable_clusters; 2016 } QEMU_PACKED reftable_offset_and_clusters; 2017 2018 qcow2_cache_empty(bs, s->refcount_block_cache); 2019 2020 write_refblocks: 2021 for (; cluster < *nb_clusters; cluster++) { 2022 if (!s->get_refcount(*refcount_table, cluster)) { 2023 continue; 2024 } 2025 2026 refblock_index = cluster >> s->refcount_block_bits; 2027 refblock_start = refblock_index << s->refcount_block_bits; 2028 2029 /* Don't allocate a cluster in a refblock already written to disk */ 2030 if (first_free_cluster < refblock_start) { 2031 first_free_cluster = refblock_start; 2032 } 2033 refblock_offset = alloc_clusters_imrt(bs, 1, refcount_table, 2034 nb_clusters, &first_free_cluster); 2035 if (refblock_offset < 0) { 2036 fprintf(stderr, "ERROR allocating refblock: %s\n", 2037 strerror(-refblock_offset)); 2038 res->check_errors++; 2039 ret = refblock_offset; 2040 goto fail; 2041 } 2042 2043 if (reftable_size <= refblock_index) { 2044 uint32_t old_reftable_size = reftable_size; 2045 uint64_t *new_on_disk_reftable; 2046 2047 reftable_size = ROUND_UP((refblock_index + 1) * sizeof(uint64_t), 2048 s->cluster_size) / sizeof(uint64_t); 2049 new_on_disk_reftable = g_try_realloc(on_disk_reftable, 2050 reftable_size * 2051 sizeof(uint64_t)); 2052 if (!new_on_disk_reftable) { 2053 res->check_errors++; 2054 ret = -ENOMEM; 2055 goto fail; 2056 } 2057 on_disk_reftable = new_on_disk_reftable; 2058 2059 memset(on_disk_reftable + old_reftable_size, 0, 2060 (reftable_size - old_reftable_size) * sizeof(uint64_t)); 2061 2062 /* The offset we have for the reftable is now no longer valid; 2063 * this will leak that range, but we can easily fix that by running 2064 * a leak-fixing check after this rebuild operation */ 2065 reftable_offset = -1; 2066 } 2067 on_disk_reftable[refblock_index] = refblock_offset; 2068 2069 /* If this is apparently the last refblock (for now), try to squeeze the 2070 * reftable in */ 2071 if (refblock_index == (*nb_clusters - 1) >> s->refcount_block_bits && 2072 reftable_offset < 0) 2073 { 2074 uint64_t reftable_clusters = size_to_clusters(s, reftable_size * 2075 sizeof(uint64_t)); 2076 reftable_offset = alloc_clusters_imrt(bs, reftable_clusters, 2077 refcount_table, nb_clusters, 2078 &first_free_cluster); 2079 if (reftable_offset < 0) { 2080 fprintf(stderr, "ERROR allocating reftable: %s\n", 2081 strerror(-reftable_offset)); 2082 res->check_errors++; 2083 ret = reftable_offset; 2084 goto fail; 2085 } 2086 } 2087 2088 ret = qcow2_pre_write_overlap_check(bs, 0, refblock_offset, 2089 s->cluster_size); 2090 if (ret < 0) { 2091 fprintf(stderr, "ERROR writing refblock: %s\n", strerror(-ret)); 2092 goto fail; 2093 } 2094 2095 /* The size of *refcount_table is always cluster-aligned, therefore the 2096 * write operation will not overflow */ 2097 on_disk_refblock = (void *)((char *) *refcount_table + 2098 refblock_index * s->cluster_size); 2099 2100 ret = bdrv_write(bs->file->bs, refblock_offset / BDRV_SECTOR_SIZE, 2101 on_disk_refblock, s->cluster_sectors); 2102 if (ret < 0) { 2103 fprintf(stderr, "ERROR writing refblock: %s\n", strerror(-ret)); 2104 goto fail; 2105 } 2106 2107 /* Go to the end of this refblock */ 2108 cluster = refblock_start + s->refcount_block_size - 1; 2109 } 2110 2111 if (reftable_offset < 0) { 2112 uint64_t post_refblock_start, reftable_clusters; 2113 2114 post_refblock_start = ROUND_UP(*nb_clusters, s->refcount_block_size); 2115 reftable_clusters = size_to_clusters(s, 2116 reftable_size * sizeof(uint64_t)); 2117 /* Not pretty but simple */ 2118 if (first_free_cluster < post_refblock_start) { 2119 first_free_cluster = post_refblock_start; 2120 } 2121 reftable_offset = alloc_clusters_imrt(bs, reftable_clusters, 2122 refcount_table, nb_clusters, 2123 &first_free_cluster); 2124 if (reftable_offset < 0) { 2125 fprintf(stderr, "ERROR allocating reftable: %s\n", 2126 strerror(-reftable_offset)); 2127 res->check_errors++; 2128 ret = reftable_offset; 2129 goto fail; 2130 } 2131 2132 goto write_refblocks; 2133 } 2134 2135 assert(on_disk_reftable); 2136 2137 for (refblock_index = 0; refblock_index < reftable_size; refblock_index++) { 2138 cpu_to_be64s(&on_disk_reftable[refblock_index]); 2139 } 2140 2141 ret = qcow2_pre_write_overlap_check(bs, 0, reftable_offset, 2142 reftable_size * sizeof(uint64_t)); 2143 if (ret < 0) { 2144 fprintf(stderr, "ERROR writing reftable: %s\n", strerror(-ret)); 2145 goto fail; 2146 } 2147 2148 assert(reftable_size < INT_MAX / sizeof(uint64_t)); 2149 ret = bdrv_pwrite(bs->file->bs, reftable_offset, on_disk_reftable, 2150 reftable_size * sizeof(uint64_t)); 2151 if (ret < 0) { 2152 fprintf(stderr, "ERROR writing reftable: %s\n", strerror(-ret)); 2153 goto fail; 2154 } 2155 2156 /* Enter new reftable into the image header */ 2157 cpu_to_be64w(&reftable_offset_and_clusters.reftable_offset, 2158 reftable_offset); 2159 cpu_to_be32w(&reftable_offset_and_clusters.reftable_clusters, 2160 size_to_clusters(s, reftable_size * sizeof(uint64_t))); 2161 ret = bdrv_pwrite_sync(bs->file->bs, offsetof(QCowHeader, 2162 refcount_table_offset), 2163 &reftable_offset_and_clusters, 2164 sizeof(reftable_offset_and_clusters)); 2165 if (ret < 0) { 2166 fprintf(stderr, "ERROR setting reftable: %s\n", strerror(-ret)); 2167 goto fail; 2168 } 2169 2170 for (refblock_index = 0; refblock_index < reftable_size; refblock_index++) { 2171 be64_to_cpus(&on_disk_reftable[refblock_index]); 2172 } 2173 s->refcount_table = on_disk_reftable; 2174 s->refcount_table_offset = reftable_offset; 2175 s->refcount_table_size = reftable_size; 2176 2177 return 0; 2178 2179 fail: 2180 g_free(on_disk_reftable); 2181 return ret; 2182 } 2183 2184 /* 2185 * Checks an image for refcount consistency. 2186 * 2187 * Returns 0 if no errors are found, the number of errors in case the image is 2188 * detected as corrupted, and -errno when an internal error occurred. 2189 */ 2190 int qcow2_check_refcounts(BlockDriverState *bs, BdrvCheckResult *res, 2191 BdrvCheckMode fix) 2192 { 2193 BDRVQcow2State *s = bs->opaque; 2194 BdrvCheckResult pre_compare_res; 2195 int64_t size, highest_cluster, nb_clusters; 2196 void *refcount_table = NULL; 2197 bool rebuild = false; 2198 int ret; 2199 2200 size = bdrv_getlength(bs->file->bs); 2201 if (size < 0) { 2202 res->check_errors++; 2203 return size; 2204 } 2205 2206 nb_clusters = size_to_clusters(s, size); 2207 if (nb_clusters > INT_MAX) { 2208 res->check_errors++; 2209 return -EFBIG; 2210 } 2211 2212 res->bfi.total_clusters = 2213 size_to_clusters(s, bs->total_sectors * BDRV_SECTOR_SIZE); 2214 2215 ret = calculate_refcounts(bs, res, fix, &rebuild, &refcount_table, 2216 &nb_clusters); 2217 if (ret < 0) { 2218 goto fail; 2219 } 2220 2221 /* In case we don't need to rebuild the refcount structure (but want to fix 2222 * something), this function is immediately called again, in which case the 2223 * result should be ignored */ 2224 pre_compare_res = *res; 2225 compare_refcounts(bs, res, 0, &rebuild, &highest_cluster, refcount_table, 2226 nb_clusters); 2227 2228 if (rebuild && (fix & BDRV_FIX_ERRORS)) { 2229 BdrvCheckResult old_res = *res; 2230 int fresh_leaks = 0; 2231 2232 fprintf(stderr, "Rebuilding refcount structure\n"); 2233 ret = rebuild_refcount_structure(bs, res, &refcount_table, 2234 &nb_clusters); 2235 if (ret < 0) { 2236 goto fail; 2237 } 2238 2239 res->corruptions = 0; 2240 res->leaks = 0; 2241 2242 /* Because the old reftable has been exchanged for a new one the 2243 * references have to be recalculated */ 2244 rebuild = false; 2245 memset(refcount_table, 0, refcount_array_byte_size(s, nb_clusters)); 2246 ret = calculate_refcounts(bs, res, 0, &rebuild, &refcount_table, 2247 &nb_clusters); 2248 if (ret < 0) { 2249 goto fail; 2250 } 2251 2252 if (fix & BDRV_FIX_LEAKS) { 2253 /* The old refcount structures are now leaked, fix it; the result 2254 * can be ignored, aside from leaks which were introduced by 2255 * rebuild_refcount_structure() that could not be fixed */ 2256 BdrvCheckResult saved_res = *res; 2257 *res = (BdrvCheckResult){ 0 }; 2258 2259 compare_refcounts(bs, res, BDRV_FIX_LEAKS, &rebuild, 2260 &highest_cluster, refcount_table, nb_clusters); 2261 if (rebuild) { 2262 fprintf(stderr, "ERROR rebuilt refcount structure is still " 2263 "broken\n"); 2264 } 2265 2266 /* Any leaks accounted for here were introduced by 2267 * rebuild_refcount_structure() because that function has created a 2268 * new refcount structure from scratch */ 2269 fresh_leaks = res->leaks; 2270 *res = saved_res; 2271 } 2272 2273 if (res->corruptions < old_res.corruptions) { 2274 res->corruptions_fixed += old_res.corruptions - res->corruptions; 2275 } 2276 if (res->leaks < old_res.leaks) { 2277 res->leaks_fixed += old_res.leaks - res->leaks; 2278 } 2279 res->leaks += fresh_leaks; 2280 } else if (fix) { 2281 if (rebuild) { 2282 fprintf(stderr, "ERROR need to rebuild refcount structures\n"); 2283 res->check_errors++; 2284 ret = -EIO; 2285 goto fail; 2286 } 2287 2288 if (res->leaks || res->corruptions) { 2289 *res = pre_compare_res; 2290 compare_refcounts(bs, res, fix, &rebuild, &highest_cluster, 2291 refcount_table, nb_clusters); 2292 } 2293 } 2294 2295 /* check OFLAG_COPIED */ 2296 ret = check_oflag_copied(bs, res, fix); 2297 if (ret < 0) { 2298 goto fail; 2299 } 2300 2301 res->image_end_offset = (highest_cluster + 1) * s->cluster_size; 2302 ret = 0; 2303 2304 fail: 2305 g_free(refcount_table); 2306 2307 return ret; 2308 } 2309 2310 #define overlaps_with(ofs, sz) \ 2311 ranges_overlap(offset, size, ofs, sz) 2312 2313 /* 2314 * Checks if the given offset into the image file is actually free to use by 2315 * looking for overlaps with important metadata sections (L1/L2 tables etc.), 2316 * i.e. a sanity check without relying on the refcount tables. 2317 * 2318 * The ign parameter specifies what checks not to perform (being a bitmask of 2319 * QCow2MetadataOverlap values), i.e., what sections to ignore. 2320 * 2321 * Returns: 2322 * - 0 if writing to this offset will not affect the mentioned metadata 2323 * - a positive QCow2MetadataOverlap value indicating one overlapping section 2324 * - a negative value (-errno) indicating an error while performing a check, 2325 * e.g. when bdrv_read failed on QCOW2_OL_INACTIVE_L2 2326 */ 2327 int qcow2_check_metadata_overlap(BlockDriverState *bs, int ign, int64_t offset, 2328 int64_t size) 2329 { 2330 BDRVQcow2State *s = bs->opaque; 2331 int chk = s->overlap_check & ~ign; 2332 int i, j; 2333 2334 if (!size) { 2335 return 0; 2336 } 2337 2338 if (chk & QCOW2_OL_MAIN_HEADER) { 2339 if (offset < s->cluster_size) { 2340 return QCOW2_OL_MAIN_HEADER; 2341 } 2342 } 2343 2344 /* align range to test to cluster boundaries */ 2345 size = align_offset(offset_into_cluster(s, offset) + size, s->cluster_size); 2346 offset = start_of_cluster(s, offset); 2347 2348 if ((chk & QCOW2_OL_ACTIVE_L1) && s->l1_size) { 2349 if (overlaps_with(s->l1_table_offset, s->l1_size * sizeof(uint64_t))) { 2350 return QCOW2_OL_ACTIVE_L1; 2351 } 2352 } 2353 2354 if ((chk & QCOW2_OL_REFCOUNT_TABLE) && s->refcount_table_size) { 2355 if (overlaps_with(s->refcount_table_offset, 2356 s->refcount_table_size * sizeof(uint64_t))) { 2357 return QCOW2_OL_REFCOUNT_TABLE; 2358 } 2359 } 2360 2361 if ((chk & QCOW2_OL_SNAPSHOT_TABLE) && s->snapshots_size) { 2362 if (overlaps_with(s->snapshots_offset, s->snapshots_size)) { 2363 return QCOW2_OL_SNAPSHOT_TABLE; 2364 } 2365 } 2366 2367 if ((chk & QCOW2_OL_INACTIVE_L1) && s->snapshots) { 2368 for (i = 0; i < s->nb_snapshots; i++) { 2369 if (s->snapshots[i].l1_size && 2370 overlaps_with(s->snapshots[i].l1_table_offset, 2371 s->snapshots[i].l1_size * sizeof(uint64_t))) { 2372 return QCOW2_OL_INACTIVE_L1; 2373 } 2374 } 2375 } 2376 2377 if ((chk & QCOW2_OL_ACTIVE_L2) && s->l1_table) { 2378 for (i = 0; i < s->l1_size; i++) { 2379 if ((s->l1_table[i] & L1E_OFFSET_MASK) && 2380 overlaps_with(s->l1_table[i] & L1E_OFFSET_MASK, 2381 s->cluster_size)) { 2382 return QCOW2_OL_ACTIVE_L2; 2383 } 2384 } 2385 } 2386 2387 if ((chk & QCOW2_OL_REFCOUNT_BLOCK) && s->refcount_table) { 2388 for (i = 0; i < s->refcount_table_size; i++) { 2389 if ((s->refcount_table[i] & REFT_OFFSET_MASK) && 2390 overlaps_with(s->refcount_table[i] & REFT_OFFSET_MASK, 2391 s->cluster_size)) { 2392 return QCOW2_OL_REFCOUNT_BLOCK; 2393 } 2394 } 2395 } 2396 2397 if ((chk & QCOW2_OL_INACTIVE_L2) && s->snapshots) { 2398 for (i = 0; i < s->nb_snapshots; i++) { 2399 uint64_t l1_ofs = s->snapshots[i].l1_table_offset; 2400 uint32_t l1_sz = s->snapshots[i].l1_size; 2401 uint64_t l1_sz2 = l1_sz * sizeof(uint64_t); 2402 uint64_t *l1 = g_try_malloc(l1_sz2); 2403 int ret; 2404 2405 if (l1_sz2 && l1 == NULL) { 2406 return -ENOMEM; 2407 } 2408 2409 ret = bdrv_pread(bs->file->bs, l1_ofs, l1, l1_sz2); 2410 if (ret < 0) { 2411 g_free(l1); 2412 return ret; 2413 } 2414 2415 for (j = 0; j < l1_sz; j++) { 2416 uint64_t l2_ofs = be64_to_cpu(l1[j]) & L1E_OFFSET_MASK; 2417 if (l2_ofs && overlaps_with(l2_ofs, s->cluster_size)) { 2418 g_free(l1); 2419 return QCOW2_OL_INACTIVE_L2; 2420 } 2421 } 2422 2423 g_free(l1); 2424 } 2425 } 2426 2427 return 0; 2428 } 2429 2430 static const char *metadata_ol_names[] = { 2431 [QCOW2_OL_MAIN_HEADER_BITNR] = "qcow2_header", 2432 [QCOW2_OL_ACTIVE_L1_BITNR] = "active L1 table", 2433 [QCOW2_OL_ACTIVE_L2_BITNR] = "active L2 table", 2434 [QCOW2_OL_REFCOUNT_TABLE_BITNR] = "refcount table", 2435 [QCOW2_OL_REFCOUNT_BLOCK_BITNR] = "refcount block", 2436 [QCOW2_OL_SNAPSHOT_TABLE_BITNR] = "snapshot table", 2437 [QCOW2_OL_INACTIVE_L1_BITNR] = "inactive L1 table", 2438 [QCOW2_OL_INACTIVE_L2_BITNR] = "inactive L2 table", 2439 }; 2440 2441 /* 2442 * First performs a check for metadata overlaps (through 2443 * qcow2_check_metadata_overlap); if that fails with a negative value (error 2444 * while performing a check), that value is returned. If an impending overlap 2445 * is detected, the BDS will be made unusable, the qcow2 file marked corrupt 2446 * and -EIO returned. 2447 * 2448 * Returns 0 if there were neither overlaps nor errors while checking for 2449 * overlaps; or a negative value (-errno) on error. 2450 */ 2451 int qcow2_pre_write_overlap_check(BlockDriverState *bs, int ign, int64_t offset, 2452 int64_t size) 2453 { 2454 int ret = qcow2_check_metadata_overlap(bs, ign, offset, size); 2455 2456 if (ret < 0) { 2457 return ret; 2458 } else if (ret > 0) { 2459 int metadata_ol_bitnr = ctz32(ret); 2460 assert(metadata_ol_bitnr < QCOW2_OL_MAX_BITNR); 2461 2462 qcow2_signal_corruption(bs, true, offset, size, "Preventing invalid " 2463 "write on metadata (overlaps with %s)", 2464 metadata_ol_names[metadata_ol_bitnr]); 2465 return -EIO; 2466 } 2467 2468 return 0; 2469 } 2470