qcow2-refcount.c 29.5 KB
Newer Older
K
Kevin Wolf 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33
/*
 * Block driver for the QCOW version 2 format
 *
 * Copyright (c) 2004-2006 Fabrice Bellard
 *
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 * THE SOFTWARE.
 */

#include "qemu-common.h"
#include "block_int.h"
#include "block/qcow2.h"

static int64_t alloc_clusters_noref(BlockDriverState *bs, int64_t size);
static int update_refcount(BlockDriverState *bs,
                            int64_t offset, int64_t length,
                            int addend);

34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53

static int cache_refcount_updates = 0;

static int write_refcount_block(BDRVQcowState *s)
{
    size_t size = s->cluster_size;

    if (s->refcount_block_cache_offset == 0) {
        return 0;
    }

    if (bdrv_pwrite(s->hd, s->refcount_block_cache_offset,
            s->refcount_block_cache, size) != size)
    {
        return -EIO;
    }

    return 0;
}

K
Kevin Wolf 已提交
54 55 56
/*********************************************************/
/* refcount handling */

K
Kevin Wolf 已提交
57
int qcow2_refcount_init(BlockDriverState *bs)
K
Kevin Wolf 已提交
58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77
{
    BDRVQcowState *s = bs->opaque;
    int ret, refcount_table_size2, i;

    s->refcount_block_cache = qemu_malloc(s->cluster_size);
    refcount_table_size2 = s->refcount_table_size * sizeof(uint64_t);
    s->refcount_table = qemu_malloc(refcount_table_size2);
    if (s->refcount_table_size > 0) {
        ret = bdrv_pread(s->hd, s->refcount_table_offset,
                         s->refcount_table, refcount_table_size2);
        if (ret != refcount_table_size2)
            goto fail;
        for(i = 0; i < s->refcount_table_size; i++)
            be64_to_cpus(&s->refcount_table[i]);
    }
    return 0;
 fail:
    return -ENOMEM;
}

K
Kevin Wolf 已提交
78
void qcow2_refcount_close(BlockDriverState *bs)
K
Kevin Wolf 已提交
79 80 81 82 83 84 85 86 87 88 89 90
{
    BDRVQcowState *s = bs->opaque;
    qemu_free(s->refcount_block_cache);
    qemu_free(s->refcount_table);
}


static int load_refcount_block(BlockDriverState *bs,
                               int64_t refcount_block_offset)
{
    BDRVQcowState *s = bs->opaque;
    int ret;
91 92 93 94 95

    if (cache_refcount_updates) {
        write_refcount_block(s);
    }

K
Kevin Wolf 已提交
96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170
    ret = bdrv_pread(s->hd, refcount_block_offset, s->refcount_block_cache,
                     s->cluster_size);
    if (ret != s->cluster_size)
        return -EIO;
    s->refcount_block_cache_offset = refcount_block_offset;
    return 0;
}

static int get_refcount(BlockDriverState *bs, int64_t cluster_index)
{
    BDRVQcowState *s = bs->opaque;
    int refcount_table_index, block_index;
    int64_t refcount_block_offset;

    refcount_table_index = cluster_index >> (s->cluster_bits - REFCOUNT_SHIFT);
    if (refcount_table_index >= s->refcount_table_size)
        return 0;
    refcount_block_offset = s->refcount_table[refcount_table_index];
    if (!refcount_block_offset)
        return 0;
    if (refcount_block_offset != s->refcount_block_cache_offset) {
        /* better than nothing: return allocated if read error */
        if (load_refcount_block(bs, refcount_block_offset) < 0)
            return 1;
    }
    block_index = cluster_index &
        ((1 << (s->cluster_bits - REFCOUNT_SHIFT)) - 1);
    return be16_to_cpu(s->refcount_block_cache[block_index]);
}

static int grow_refcount_table(BlockDriverState *bs, int min_size)
{
    BDRVQcowState *s = bs->opaque;
    int new_table_size, new_table_size2, refcount_table_clusters, i, ret;
    uint64_t *new_table;
    int64_t table_offset;
    uint8_t data[12];
    int old_table_size;
    int64_t old_table_offset;

    if (min_size <= s->refcount_table_size)
        return 0;
    /* compute new table size */
    refcount_table_clusters = s->refcount_table_size >> (s->cluster_bits - 3);
    for(;;) {
        if (refcount_table_clusters == 0) {
            refcount_table_clusters = 1;
        } else {
            refcount_table_clusters = (refcount_table_clusters * 3 + 1) / 2;
        }
        new_table_size = refcount_table_clusters << (s->cluster_bits - 3);
        if (min_size <= new_table_size)
            break;
    }
#ifdef DEBUG_ALLOC2
    printf("grow_refcount_table from %d to %d\n",
           s->refcount_table_size,
           new_table_size);
#endif
    new_table_size2 = new_table_size * sizeof(uint64_t);
    new_table = qemu_mallocz(new_table_size2);
    memcpy(new_table, s->refcount_table,
           s->refcount_table_size * sizeof(uint64_t));
    for(i = 0; i < s->refcount_table_size; i++)
        cpu_to_be64s(&new_table[i]);
    /* Note: we cannot update the refcount now to avoid recursion */
    table_offset = alloc_clusters_noref(bs, new_table_size2);
    ret = bdrv_pwrite(s->hd, table_offset, new_table, new_table_size2);
    if (ret != new_table_size2)
        goto fail;
    for(i = 0; i < s->refcount_table_size; i++)
        be64_to_cpus(&new_table[i]);

    cpu_to_be64w((uint64_t*)data, table_offset);
    cpu_to_be32w((uint32_t*)(data + 8), refcount_table_clusters);
171 172 173
    ret = bdrv_pwrite(s->hd, offsetof(QCowHeader, refcount_table_offset),
                    data, sizeof(data));
    if (ret != sizeof(data)) {
K
Kevin Wolf 已提交
174
        goto fail;
175 176
    }

K
Kevin Wolf 已提交
177 178 179 180 181 182 183 184
    qemu_free(s->refcount_table);
    old_table_offset = s->refcount_table_offset;
    old_table_size = s->refcount_table_size;
    s->refcount_table = new_table;
    s->refcount_table_size = new_table_size;
    s->refcount_table_offset = table_offset;

    update_refcount(bs, table_offset, new_table_size2, 1);
K
Kevin Wolf 已提交
185
    qcow2_free_clusters(bs, old_table_offset, old_table_size * sizeof(uint64_t));
K
Kevin Wolf 已提交
186 187 188
    return 0;
 fail:
    qemu_free(new_table);
189
    return ret < 0 ? ret : -EIO;
K
Kevin Wolf 已提交
190 191 192 193 194 195 196
}


static int64_t alloc_refcount_block(BlockDriverState *bs, int64_t cluster_index)
{
    BDRVQcowState *s = bs->opaque;
    int64_t offset, refcount_block_offset;
197 198
    unsigned int refcount_table_index;
    int ret;
K
Kevin Wolf 已提交
199
    uint64_t data64;
200
    int cache = cache_refcount_updates;
K
Kevin Wolf 已提交
201 202 203 204 205 206 207 208 209 210 211 212

    /* Find L1 index and grow refcount table if needed */
    refcount_table_index = cluster_index >> (s->cluster_bits - REFCOUNT_SHIFT);
    if (refcount_table_index >= s->refcount_table_size) {
        ret = grow_refcount_table(bs, refcount_table_index + 1);
        if (ret < 0)
            return ret;
    }

    /* Load or allocate the refcount block */
    refcount_block_offset = s->refcount_table[refcount_table_index];
    if (!refcount_block_offset) {
213 214 215 216
        if (cache_refcount_updates) {
            write_refcount_block(s);
            cache_refcount_updates = 0;
        }
K
Kevin Wolf 已提交
217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234
        /* create a new refcount block */
        /* Note: we cannot update the refcount now to avoid recursion */
        offset = alloc_clusters_noref(bs, s->cluster_size);
        memset(s->refcount_block_cache, 0, s->cluster_size);
        ret = bdrv_pwrite(s->hd, offset, s->refcount_block_cache, s->cluster_size);
        if (ret != s->cluster_size)
            return -EINVAL;
        s->refcount_table[refcount_table_index] = offset;
        data64 = cpu_to_be64(offset);
        ret = bdrv_pwrite(s->hd, s->refcount_table_offset +
                          refcount_table_index * sizeof(uint64_t),
                          &data64, sizeof(data64));
        if (ret != sizeof(data64))
            return -EINVAL;

        refcount_block_offset = offset;
        s->refcount_block_cache_offset = offset;
        update_refcount(bs, offset, s->cluster_size, 1);
235
        cache_refcount_updates = cache;
K
Kevin Wolf 已提交
236 237 238 239 240 241 242 243 244 245
    } else {
        if (refcount_block_offset != s->refcount_block_cache_offset) {
            if (load_refcount_block(bs, refcount_block_offset) < 0)
                return -EIO;
        }
    }

    return refcount_block_offset;
}

246 247 248 249 250 251
#define REFCOUNTS_PER_SECTOR (512 >> REFCOUNT_SHIFT)
static int write_refcount_block_entries(BDRVQcowState *s,
    int64_t refcount_block_offset, int first_index, int last_index)
{
    size_t size;

252 253 254 255
    if (cache_refcount_updates) {
        return 0;
    }

256 257 258 259 260 261 262 263 264 265 266 267 268 269 270
    first_index &= ~(REFCOUNTS_PER_SECTOR - 1);
    last_index = (last_index + REFCOUNTS_PER_SECTOR)
        & ~(REFCOUNTS_PER_SECTOR - 1);

    size = (last_index - first_index) << REFCOUNT_SHIFT;
    if (bdrv_pwrite(s->hd,
        refcount_block_offset + (first_index << REFCOUNT_SHIFT),
        &s->refcount_block_cache[first_index], size) != size)
    {
        return -EIO;
    }

    return 0;
}

K
Kevin Wolf 已提交
271 272 273 274 275 276 277 278 279 280
/* XXX: cache several refcount block clusters ? */
static int update_refcount(BlockDriverState *bs,
                            int64_t offset, int64_t length,
                            int addend)
{
    BDRVQcowState *s = bs->opaque;
    int64_t start, last, cluster_offset;
    int64_t refcount_block_offset = 0;
    int64_t table_index = -1, old_table_index;
    int first_index = -1, last_index = -1;
281
    int ret;
K
Kevin Wolf 已提交
282 283

#ifdef DEBUG_ALLOC2
284
    printf("update_refcount: offset=%" PRId64 " size=%" PRId64 " addend=%d\n",
K
Kevin Wolf 已提交
285 286
           offset, length, addend);
#endif
K
Kevin Wolf 已提交
287
    if (length < 0) {
K
Kevin Wolf 已提交
288
        return -EINVAL;
K
Kevin Wolf 已提交
289 290 291 292
    } else if (length == 0) {
        return 0;
    }

K
Kevin Wolf 已提交
293 294 295 296 297 298 299
    start = offset & ~(s->cluster_size - 1);
    last = (offset + length - 1) & ~(s->cluster_size - 1);
    for(cluster_offset = start; cluster_offset <= last;
        cluster_offset += s->cluster_size)
    {
        int block_index, refcount;
        int64_t cluster_index = cluster_offset >> s->cluster_bits;
300
        int64_t new_block;
K
Kevin Wolf 已提交
301 302 303 304 305

        /* Only write refcount block to disk when we are done with it */
        old_table_index = table_index;
        table_index = cluster_index >> (s->cluster_bits - REFCOUNT_SHIFT);
        if ((old_table_index >= 0) && (table_index != old_table_index)) {
306 307 308

            if (write_refcount_block_entries(s, refcount_block_offset,
                first_index, last_index) < 0)
K
Kevin Wolf 已提交
309 310 311 312 313 314 315 316 317
            {
                return -EIO;
            }

            first_index = -1;
            last_index = -1;
        }

        /* Load the refcount block and allocate it if needed */
318 319 320 321
        new_block = alloc_refcount_block(bs, cluster_index);
        if (new_block < 0) {
            ret = new_block;
            goto fail;
K
Kevin Wolf 已提交
322
        }
323
        refcount_block_offset = new_block;
K
Kevin Wolf 已提交
324 325 326 327 328 329 330 331 332 333 334 335 336

        /* we can update the count and save it */
        block_index = cluster_index &
            ((1 << (s->cluster_bits - REFCOUNT_SHIFT)) - 1);
        if (first_index == -1 || block_index < first_index) {
            first_index = block_index;
        }
        if (block_index > last_index) {
            last_index = block_index;
        }

        refcount = be16_to_cpu(s->refcount_block_cache[block_index]);
        refcount += addend;
337 338 339 340
        if (refcount < 0 || refcount > 0xffff) {
            ret = -EINVAL;
            goto fail;
        }
K
Kevin Wolf 已提交
341 342 343 344 345 346
        if (refcount == 0 && cluster_index < s->free_cluster_index) {
            s->free_cluster_index = cluster_index;
        }
        s->refcount_block_cache[block_index] = cpu_to_be16(refcount);
    }

347 348 349
    ret = 0;
fail:

K
Kevin Wolf 已提交
350 351
    /* Write last changed block to disk */
    if (refcount_block_offset != 0) {
352 353
        if (write_refcount_block_entries(s, refcount_block_offset,
            first_index, last_index) < 0)
K
Kevin Wolf 已提交
354
        {
355
            return ret < 0 ? ret : -EIO;
K
Kevin Wolf 已提交
356 357 358
        }
    }

359 360 361 362 363 364 365 366 367 368
    /*
     * Try do undo any updates if an error is returned (This may succeed in
     * some cases like ENOSPC for allocating a new refcount block)
     */
    if (ret < 0) {
        int dummy;
        dummy = update_refcount(bs, offset, cluster_offset - offset, -addend);
    }

    return ret;
K
Kevin Wolf 已提交
369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407
}

/* addend must be 1 or -1 */
static int update_cluster_refcount(BlockDriverState *bs,
                                   int64_t cluster_index,
                                   int addend)
{
    BDRVQcowState *s = bs->opaque;
    int ret;

    ret = update_refcount(bs, cluster_index << s->cluster_bits, 1, addend);
    if (ret < 0) {
        return ret;
    }

    return get_refcount(bs, cluster_index);
}



/*********************************************************/
/* cluster allocation functions */



/* return < 0 if error */
static int64_t alloc_clusters_noref(BlockDriverState *bs, int64_t size)
{
    BDRVQcowState *s = bs->opaque;
    int i, nb_clusters;

    nb_clusters = size_to_clusters(s, size);
retry:
    for(i = 0; i < nb_clusters; i++) {
        int64_t i = s->free_cluster_index++;
        if (get_refcount(bs, i) != 0)
            goto retry;
    }
#ifdef DEBUG_ALLOC2
408
    printf("alloc_clusters: size=%" PRId64 " -> %" PRId64 "\n",
K
Kevin Wolf 已提交
409 410 411 412 413 414
            size,
            (s->free_cluster_index - nb_clusters) << s->cluster_bits);
#endif
    return (s->free_cluster_index - nb_clusters) << s->cluster_bits;
}

K
Kevin Wolf 已提交
415
int64_t qcow2_alloc_clusters(BlockDriverState *bs, int64_t size)
K
Kevin Wolf 已提交
416 417 418 419 420 421 422 423 424 425
{
    int64_t offset;

    offset = alloc_clusters_noref(bs, size);
    update_refcount(bs, offset, size, 1);
    return offset;
}

/* only used to allocate compressed sectors. We try to allocate
   contiguous sectors. size must be <= cluster_size */
K
Kevin Wolf 已提交
426
int64_t qcow2_alloc_bytes(BlockDriverState *bs, int size)
K
Kevin Wolf 已提交
427 428 429 430 431 432 433
{
    BDRVQcowState *s = bs->opaque;
    int64_t offset, cluster_offset;
    int free_in_cluster;

    assert(size > 0 && size <= s->cluster_size);
    if (s->free_byte_offset == 0) {
K
Kevin Wolf 已提交
434
        s->free_byte_offset = qcow2_alloc_clusters(bs, s->cluster_size);
K
Kevin Wolf 已提交
435 436 437 438 439 440 441 442 443 444 445 446 447 448
    }
 redo:
    free_in_cluster = s->cluster_size -
        (s->free_byte_offset & (s->cluster_size - 1));
    if (size <= free_in_cluster) {
        /* enough space in current cluster */
        offset = s->free_byte_offset;
        s->free_byte_offset += size;
        free_in_cluster -= size;
        if (free_in_cluster == 0)
            s->free_byte_offset = 0;
        if ((offset & (s->cluster_size - 1)) != 0)
            update_cluster_refcount(bs, offset >> s->cluster_bits, 1);
    } else {
K
Kevin Wolf 已提交
449
        offset = qcow2_alloc_clusters(bs, s->cluster_size);
K
Kevin Wolf 已提交
450 451 452 453 454 455 456 457 458 459 460 461 462 463
        cluster_offset = s->free_byte_offset & ~(s->cluster_size - 1);
        if ((cluster_offset + s->cluster_size) == offset) {
            /* we are lucky: contiguous data */
            offset = s->free_byte_offset;
            update_cluster_refcount(bs, offset >> s->cluster_bits, 1);
            s->free_byte_offset += size;
        } else {
            s->free_byte_offset = offset;
            goto redo;
        }
    }
    return offset;
}

K
Kevin Wolf 已提交
464
void qcow2_free_clusters(BlockDriverState *bs,
K
Kevin Wolf 已提交
465 466 467 468 469
                          int64_t offset, int64_t size)
{
    update_refcount(bs, offset, size, -1);
}

470 471 472 473 474 475 476
/*
 * free_any_clusters
 *
 * free clusters according to its type: compressed or not
 *
 */

K
Kevin Wolf 已提交
477
void qcow2_free_any_clusters(BlockDriverState *bs,
478 479 480 481 482 483 484 485 486 487
    uint64_t cluster_offset, int nb_clusters)
{
    BDRVQcowState *s = bs->opaque;

    /* free the cluster */

    if (cluster_offset & QCOW_OFLAG_COMPRESSED) {
        int nb_csectors;
        nb_csectors = ((cluster_offset >> s->csize_shift) &
                       s->csize_mask) + 1;
K
Kevin Wolf 已提交
488 489 490
        qcow2_free_clusters(bs,
            (cluster_offset & s->cluster_offset_mask) & ~511,
            nb_csectors * 512);
491 492 493
        return;
    }

K
Kevin Wolf 已提交
494
    qcow2_free_clusters(bs, cluster_offset, nb_clusters << s->cluster_bits);
495 496 497 498

    return;
}

K
Kevin Wolf 已提交
499 500 501 502 503 504 505


/*********************************************************/
/* snapshots and image creation */



K
Kevin Wolf 已提交
506 507
void qcow2_create_refcount_update(QCowCreateState *s, int64_t offset,
    int64_t size)
K
Kevin Wolf 已提交
508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524
{
    int refcount;
    int64_t start, last, cluster_offset;
    uint16_t *p;

    start = offset & ~(s->cluster_size - 1);
    last = (offset + size - 1)  & ~(s->cluster_size - 1);
    for(cluster_offset = start; cluster_offset <= last;
        cluster_offset += s->cluster_size) {
        p = &s->refcount_block[cluster_offset >> s->cluster_bits];
        refcount = be16_to_cpu(*p);
        refcount++;
        *p = cpu_to_be16(refcount);
    }
}

/* update the refcounts of snapshots and the copied flag */
K
Kevin Wolf 已提交
525 526
int qcow2_update_snapshot_refcount(BlockDriverState *bs,
    int64_t l1_table_offset, int l1_size, int addend)
K
Kevin Wolf 已提交
527 528 529 530 531 532
{
    BDRVQcowState *s = bs->opaque;
    uint64_t *l1_table, *l2_table, l2_offset, offset, l1_size2, l1_allocated;
    int64_t old_offset, old_l2_offset;
    int l2_size, i, j, l1_modified, l2_modified, nb_csectors, refcount;

K
Kevin Wolf 已提交
533
    qcow2_l2_cache_reset(bs);
534
    cache_refcount_updates = 1;
K
Kevin Wolf 已提交
535 536 537 538 539

    l2_table = NULL;
    l1_table = NULL;
    l1_size2 = l1_size * sizeof(uint64_t);
    if (l1_table_offset != s->l1_table_offset) {
540 541 542 543 544
        if (l1_size2 != 0) {
            l1_table = qemu_mallocz(align_offset(l1_size2, 512));
        } else {
            l1_table = NULL;
        }
K
Kevin Wolf 已提交
545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629
        l1_allocated = 1;
        if (bdrv_pread(s->hd, l1_table_offset,
                       l1_table, l1_size2) != l1_size2)
            goto fail;
        for(i = 0;i < l1_size; i++)
            be64_to_cpus(&l1_table[i]);
    } else {
        assert(l1_size == s->l1_size);
        l1_table = s->l1_table;
        l1_allocated = 0;
    }

    l2_size = s->l2_size * sizeof(uint64_t);
    l2_table = qemu_malloc(l2_size);
    l1_modified = 0;
    for(i = 0; i < l1_size; i++) {
        l2_offset = l1_table[i];
        if (l2_offset) {
            old_l2_offset = l2_offset;
            l2_offset &= ~QCOW_OFLAG_COPIED;
            l2_modified = 0;
            if (bdrv_pread(s->hd, l2_offset, l2_table, l2_size) != l2_size)
                goto fail;
            for(j = 0; j < s->l2_size; j++) {
                offset = be64_to_cpu(l2_table[j]);
                if (offset != 0) {
                    old_offset = offset;
                    offset &= ~QCOW_OFLAG_COPIED;
                    if (offset & QCOW_OFLAG_COMPRESSED) {
                        nb_csectors = ((offset >> s->csize_shift) &
                                       s->csize_mask) + 1;
                        if (addend != 0)
                            update_refcount(bs, (offset & s->cluster_offset_mask) & ~511,
                                            nb_csectors * 512, addend);
                        /* compressed clusters are never modified */
                        refcount = 2;
                    } else {
                        if (addend != 0) {
                            refcount = update_cluster_refcount(bs, offset >> s->cluster_bits, addend);
                        } else {
                            refcount = get_refcount(bs, offset >> s->cluster_bits);
                        }
                    }

                    if (refcount == 1) {
                        offset |= QCOW_OFLAG_COPIED;
                    }
                    if (offset != old_offset) {
                        l2_table[j] = cpu_to_be64(offset);
                        l2_modified = 1;
                    }
                }
            }
            if (l2_modified) {
                if (bdrv_pwrite(s->hd,
                                l2_offset, l2_table, l2_size) != l2_size)
                    goto fail;
            }

            if (addend != 0) {
                refcount = update_cluster_refcount(bs, l2_offset >> s->cluster_bits, addend);
            } else {
                refcount = get_refcount(bs, l2_offset >> s->cluster_bits);
            }
            if (refcount == 1) {
                l2_offset |= QCOW_OFLAG_COPIED;
            }
            if (l2_offset != old_l2_offset) {
                l1_table[i] = l2_offset;
                l1_modified = 1;
            }
        }
    }
    if (l1_modified) {
        for(i = 0; i < l1_size; i++)
            cpu_to_be64s(&l1_table[i]);
        if (bdrv_pwrite(s->hd, l1_table_offset, l1_table,
                        l1_size2) != l1_size2)
            goto fail;
        for(i = 0; i < l1_size; i++)
            be64_to_cpus(&l1_table[i]);
    }
    if (l1_allocated)
        qemu_free(l1_table);
    qemu_free(l2_table);
630 631
    cache_refcount_updates = 0;
    write_refcount_block(s);
K
Kevin Wolf 已提交
632 633 634 635 636
    return 0;
 fail:
    if (l1_allocated)
        qemu_free(l1_table);
    qemu_free(l2_table);
637 638
    cache_refcount_updates = 0;
    write_refcount_block(s);
K
Kevin Wolf 已提交
639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799
    return -EIO;
}




/*********************************************************/
/* refcount checking functions */



/*
 * Increases the refcount for a range of clusters in a given refcount table.
 * This is used to construct a temporary refcount table out of L1 and L2 tables
 * which can be compared the the refcount table saved in the image.
 *
 * Returns the number of errors in the image that were found
 */
static int inc_refcounts(BlockDriverState *bs,
                          uint16_t *refcount_table,
                          int refcount_table_size,
                          int64_t offset, int64_t size)
{
    BDRVQcowState *s = bs->opaque;
    int64_t start, last, cluster_offset;
    int k;
    int errors = 0;

    if (size <= 0)
        return 0;

    start = offset & ~(s->cluster_size - 1);
    last = (offset + size - 1) & ~(s->cluster_size - 1);
    for(cluster_offset = start; cluster_offset <= last;
        cluster_offset += s->cluster_size) {
        k = cluster_offset >> s->cluster_bits;
        if (k < 0 || k >= refcount_table_size) {
            fprintf(stderr, "ERROR: invalid cluster offset=0x%" PRIx64 "\n",
                cluster_offset);
            errors++;
        } else {
            if (++refcount_table[k] == 0) {
                fprintf(stderr, "ERROR: overflow cluster offset=0x%" PRIx64
                    "\n", cluster_offset);
                errors++;
            }
        }
    }

    return errors;
}

/*
 * Increases the refcount in the given refcount table for the all clusters
 * referenced in the L2 table. While doing so, performs some checks on L2
 * entries.
 *
 * Returns the number of errors found by the checks or -errno if an internal
 * error occurred.
 */
static int check_refcounts_l2(BlockDriverState *bs,
    uint16_t *refcount_table, int refcount_table_size, int64_t l2_offset,
    int check_copied)
{
    BDRVQcowState *s = bs->opaque;
    uint64_t *l2_table, offset;
    int i, l2_size, nb_csectors, refcount;
    int errors = 0;

    /* Read L2 table from disk */
    l2_size = s->l2_size * sizeof(uint64_t);
    l2_table = qemu_malloc(l2_size);

    if (bdrv_pread(s->hd, l2_offset, l2_table, l2_size) != l2_size)
        goto fail;

    /* Do the actual checks */
    for(i = 0; i < s->l2_size; i++) {
        offset = be64_to_cpu(l2_table[i]);
        if (offset != 0) {
            if (offset & QCOW_OFLAG_COMPRESSED) {
                /* Compressed clusters don't have QCOW_OFLAG_COPIED */
                if (offset & QCOW_OFLAG_COPIED) {
                    fprintf(stderr, "ERROR: cluster %" PRId64 ": "
                        "copied flag must never be set for compressed "
                        "clusters\n", offset >> s->cluster_bits);
                    offset &= ~QCOW_OFLAG_COPIED;
                    errors++;
                }

                /* Mark cluster as used */
                nb_csectors = ((offset >> s->csize_shift) &
                               s->csize_mask) + 1;
                offset &= s->cluster_offset_mask;
                errors += inc_refcounts(bs, refcount_table,
                              refcount_table_size,
                              offset & ~511, nb_csectors * 512);
            } else {
                /* QCOW_OFLAG_COPIED must be set iff refcount == 1 */
                if (check_copied) {
                    uint64_t entry = offset;
                    offset &= ~QCOW_OFLAG_COPIED;
                    refcount = get_refcount(bs, offset >> s->cluster_bits);
                    if ((refcount == 1) != ((entry & QCOW_OFLAG_COPIED) != 0)) {
                        fprintf(stderr, "ERROR OFLAG_COPIED: offset=%"
                            PRIx64 " refcount=%d\n", entry, refcount);
                        errors++;
                    }
                }

                /* Mark cluster as used */
                offset &= ~QCOW_OFLAG_COPIED;
                errors += inc_refcounts(bs, refcount_table,
                              refcount_table_size,
                              offset, s->cluster_size);

                /* Correct offsets are cluster aligned */
                if (offset & (s->cluster_size - 1)) {
                    fprintf(stderr, "ERROR offset=%" PRIx64 ": Cluster is not "
                        "properly aligned; L2 entry corrupted.\n", offset);
                    errors++;
                }
            }
        }
    }

    qemu_free(l2_table);
    return errors;

fail:
    fprintf(stderr, "ERROR: I/O error in check_refcounts_l1\n");
    qemu_free(l2_table);
    return -EIO;
}

/*
 * Increases the refcount for the L1 table, its L2 tables and all referenced
 * clusters in the given refcount table. While doing so, performs some checks
 * on L1 and L2 entries.
 *
 * Returns the number of errors found by the checks or -errno if an internal
 * error occurred.
 */
static int check_refcounts_l1(BlockDriverState *bs,
                              uint16_t *refcount_table,
                              int refcount_table_size,
                              int64_t l1_table_offset, int l1_size,
                              int check_copied)
{
    BDRVQcowState *s = bs->opaque;
    uint64_t *l1_table, l2_offset, l1_size2;
    int i, refcount, ret;
    int errors = 0;

    l1_size2 = l1_size * sizeof(uint64_t);

    /* Mark L1 table as used */
    errors += inc_refcounts(bs, refcount_table, refcount_table_size,
                  l1_table_offset, l1_size2);

    /* Read L1 table entries from disk */
800 801 802 803 804 805 806 807 808 809
    if (l1_size2 == 0) {
        l1_table = NULL;
    } else {
        l1_table = qemu_malloc(l1_size2);
        if (bdrv_pread(s->hd, l1_table_offset,
                       l1_table, l1_size2) != l1_size2)
            goto fail;
        for(i = 0;i < l1_size; i++)
            be64_to_cpus(&l1_table[i]);
    }
K
Kevin Wolf 已提交
810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863

    /* Do the actual checks */
    for(i = 0; i < l1_size; i++) {
        l2_offset = l1_table[i];
        if (l2_offset) {
            /* QCOW_OFLAG_COPIED must be set iff refcount == 1 */
            if (check_copied) {
                refcount = get_refcount(bs, (l2_offset & ~QCOW_OFLAG_COPIED)
                    >> s->cluster_bits);
                if ((refcount == 1) != ((l2_offset & QCOW_OFLAG_COPIED) != 0)) {
                    fprintf(stderr, "ERROR OFLAG_COPIED: l2_offset=%" PRIx64
                        " refcount=%d\n", l2_offset, refcount);
                    errors++;
                }
            }

            /* Mark L2 table as used */
            l2_offset &= ~QCOW_OFLAG_COPIED;
            errors += inc_refcounts(bs, refcount_table,
                          refcount_table_size,
                          l2_offset,
                          s->cluster_size);

            /* L2 tables are cluster aligned */
            if (l2_offset & (s->cluster_size - 1)) {
                fprintf(stderr, "ERROR l2_offset=%" PRIx64 ": Table is not "
                    "cluster aligned; L1 entry corrupted\n", l2_offset);
                errors++;
            }

            /* Process and check L2 entries */
            ret = check_refcounts_l2(bs, refcount_table, refcount_table_size,
                l2_offset, check_copied);
            if (ret < 0) {
                goto fail;
            }
            errors += ret;
        }
    }
    qemu_free(l1_table);
    return errors;

fail:
    fprintf(stderr, "ERROR: I/O error in check_refcounts_l1\n");
    qemu_free(l1_table);
    return -EIO;
}

/*
 * Checks an image for refcount consistency.
 *
 * Returns 0 if no errors are found, the number of errors in case the image is
 * detected as corrupted, and -errno when an internal error occured.
 */
K
Kevin Wolf 已提交
864
int qcow2_check_refcounts(BlockDriverState *bs)
K
Kevin Wolf 已提交
865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926
{
    BDRVQcowState *s = bs->opaque;
    int64_t size;
    int nb_clusters, refcount1, refcount2, i;
    QCowSnapshot *sn;
    uint16_t *refcount_table;
    int ret, errors = 0;

    size = bdrv_getlength(s->hd);
    nb_clusters = size_to_clusters(s, size);
    refcount_table = qemu_mallocz(nb_clusters * sizeof(uint16_t));

    /* header */
    errors += inc_refcounts(bs, refcount_table, nb_clusters,
                  0, s->cluster_size);

    /* current L1 table */
    ret = check_refcounts_l1(bs, refcount_table, nb_clusters,
                       s->l1_table_offset, s->l1_size, 1);
    if (ret < 0) {
        return ret;
    }
    errors += ret;

    /* snapshots */
    for(i = 0; i < s->nb_snapshots; i++) {
        sn = s->snapshots + i;
        check_refcounts_l1(bs, refcount_table, nb_clusters,
                           sn->l1_table_offset, sn->l1_size, 0);
    }
    errors += inc_refcounts(bs, refcount_table, nb_clusters,
                  s->snapshots_offset, s->snapshots_size);

    /* refcount data */
    errors += inc_refcounts(bs, refcount_table, nb_clusters,
                  s->refcount_table_offset,
                  s->refcount_table_size * sizeof(uint64_t));
    for(i = 0; i < s->refcount_table_size; i++) {
        int64_t offset;
        offset = s->refcount_table[i];
        if (offset != 0) {
            errors += inc_refcounts(bs, refcount_table, nb_clusters,
                          offset, s->cluster_size);
        }
    }

    /* compare ref counts */
    for(i = 0; i < nb_clusters; i++) {
        refcount1 = get_refcount(bs, i);
        refcount2 = refcount_table[i];
        if (refcount1 != refcount2) {
            fprintf(stderr, "ERROR cluster %d refcount=%d reference=%d\n",
                   i, refcount1, refcount2);
            errors++;
        }
    }

    qemu_free(refcount_table);

    return errors;
}