qcow2.c 30.3 KB
Newer Older
B
bellard 已提交
1 2
/*
 * Block driver for the QCOW version 2 format
3
 *
B
bellard 已提交
4
 * Copyright (c) 2004-2006 Fabrice Bellard
5
 *
B
bellard 已提交
6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
 * 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.
 */
P
pbrook 已提交
24
#include "qemu-common.h"
B
bellard 已提交
25
#include "block_int.h"
26
#include "module.h"
B
bellard 已提交
27 28
#include <zlib.h>
#include "aes.h"
K
Kevin Wolf 已提交
29
#include "block/qcow2.h"
B
bellard 已提交
30 31 32 33 34 35 36 37

/*
  Differences with QCOW:

  - Support for multiple incremental snapshots.
  - Memory management by reference counts.
  - Clusters which have a reference count of one have the bit
    QCOW_OFLAG_COPIED to optimize write performance.
38
  - Size of compressed clusters is stored in sectors to reduce bit usage
B
bellard 已提交
39 40
    in the cluster offsets.
  - Support for storing additional data (such as the VM state) in the
41
    snapshots.
B
bellard 已提交
42 43 44 45 46
  - If a backing store is used, the cluster size is not constrained
    (could be backported to QCOW).
  - L2 tables have always a size of one cluster.
*/

47 48 49 50 51 52

typedef struct {
    uint32_t magic;
    uint32_t len;
} QCowExtension;
#define  QCOW_EXT_MAGIC_END 0
53
#define  QCOW_EXT_MAGIC_BACKING_FORMAT 0xE2792ACA
54 55


B
bellard 已提交
56 57 58 59

static int qcow_probe(const uint8_t *buf, int buf_size, const char *filename)
{
    const QCowHeader *cow_header = (const void *)buf;
60

B
bellard 已提交
61 62
    if (buf_size >= sizeof(QCowHeader) &&
        be32_to_cpu(cow_header->magic) == QCOW_MAGIC &&
63
        be32_to_cpu(cow_header->version) == QCOW_VERSION)
B
bellard 已提交
64 65 66 67 68
        return 100;
    else
        return 0;
}

69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98

/* 
 * read qcow2 extension and fill bs
 * start reading from start_offset
 * finish reading upon magic of value 0 or when end_offset reached
 * unknown magic is skipped (future extension this version knows nothing about)
 * return 0 upon success, non-0 otherwise
 */
static int qcow_read_extensions(BlockDriverState *bs, uint64_t start_offset,
                                uint64_t end_offset)
{
    BDRVQcowState *s = bs->opaque;
    QCowExtension ext;
    uint64_t offset;

#ifdef DEBUG_EXT
    printf("qcow_read_extensions: start=%ld end=%ld\n", start_offset, end_offset);
#endif
    offset = start_offset;
    while (offset < end_offset) {

#ifdef DEBUG_EXT
        /* Sanity check */
        if (offset > s->cluster_size)
            printf("qcow_handle_extension: suspicious offset %lu\n", offset);

        printf("attemting to read extended header in offset %lu\n", offset);
#endif

        if (bdrv_pread(s->hd, offset, &ext, sizeof(ext)) != sizeof(ext)) {
99 100
            fprintf(stderr, "qcow_handle_extension: ERROR: pread fail from offset %llu\n",
                    (unsigned long long)offset);
101 102 103 104 105 106 107 108 109 110 111
            return 1;
        }
        be32_to_cpus(&ext.magic);
        be32_to_cpus(&ext.len);
        offset += sizeof(ext);
#ifdef DEBUG_EXT
        printf("ext.magic = 0x%x\n", ext.magic);
#endif
        switch (ext.magic) {
        case QCOW_EXT_MAGIC_END:
            return 0;
112 113 114 115

        case QCOW_EXT_MAGIC_BACKING_FORMAT:
            if (ext.len >= sizeof(bs->backing_format)) {
                fprintf(stderr, "ERROR: ext_backing_format: len=%u too large"
116
                        " (>=%zu)\n",
117 118 119 120 121 122 123 124 125 126 127 128 129
                        ext.len, sizeof(bs->backing_format));
                return 2;
            }
            if (bdrv_pread(s->hd, offset , bs->backing_format,
                           ext.len) != ext.len)
                return 3;
            bs->backing_format[ext.len] = '\0';
#ifdef DEBUG_EXT
            printf("Qcow2: Got format extension %s\n", bs->backing_format);
#endif
            offset += ((ext.len + 7) & ~7);
            break;

130 131 132 133 134 135 136 137 138 139 140
        default:
            /* unknown magic -- just skip it */
            offset += ((ext.len + 7) & ~7);
            break;
        }
    }

    return 0;
}


B
bellard 已提交
141 142 143 144 145
static int qcow_open(BlockDriverState *bs, const char *filename, int flags)
{
    BDRVQcowState *s = bs->opaque;
    int len, i, shift, ret;
    QCowHeader header;
146
    uint64_t ext_end;
B
bellard 已提交
147

148 149 150 151 152 153 154 155
    /* Performance is terrible right now with cache=writethrough due mainly
     * to reference count updates.  If the user does not explicitly specify
     * a caching type, force to writeback caching.
     */
    if ((flags & BDRV_O_CACHE_DEF)) {
        flags |= BDRV_O_CACHE_WB;
        flags &= ~BDRV_O_CACHE_DEF;
    }
156
    ret = bdrv_file_open(&s->hd, filename, flags);
B
bellard 已提交
157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173
    if (ret < 0)
        return ret;
    if (bdrv_pread(s->hd, 0, &header, sizeof(header)) != sizeof(header))
        goto fail;
    be32_to_cpus(&header.magic);
    be32_to_cpus(&header.version);
    be64_to_cpus(&header.backing_file_offset);
    be32_to_cpus(&header.backing_file_size);
    be64_to_cpus(&header.size);
    be32_to_cpus(&header.cluster_bits);
    be32_to_cpus(&header.crypt_method);
    be64_to_cpus(&header.l1_table_offset);
    be32_to_cpus(&header.l1_size);
    be64_to_cpus(&header.refcount_table_offset);
    be32_to_cpus(&header.refcount_table_clusters);
    be64_to_cpus(&header.snapshots_offset);
    be32_to_cpus(&header.nb_snapshots);
174

B
bellard 已提交
175 176
    if (header.magic != QCOW_MAGIC || header.version != QCOW_VERSION)
        goto fail;
177
    if (header.size <= 1 ||
178 179
        header.cluster_bits < MIN_CLUSTER_BITS ||
        header.cluster_bits > MAX_CLUSTER_BITS)
B
bellard 已提交
180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195
        goto fail;
    if (header.crypt_method > QCOW_CRYPT_AES)
        goto fail;
    s->crypt_method_header = header.crypt_method;
    if (s->crypt_method_header)
        bs->encrypted = 1;
    s->cluster_bits = header.cluster_bits;
    s->cluster_size = 1 << s->cluster_bits;
    s->cluster_sectors = 1 << (s->cluster_bits - 9);
    s->l2_bits = s->cluster_bits - 3; /* L2 is always one cluster */
    s->l2_size = 1 << s->l2_bits;
    bs->total_sectors = header.size / 512;
    s->csize_shift = (62 - (s->cluster_bits - 8));
    s->csize_mask = (1 << (s->cluster_bits - 8)) - 1;
    s->cluster_offset_mask = (1LL << s->csize_shift) - 1;
    s->refcount_table_offset = header.refcount_table_offset;
196
    s->refcount_table_size =
B
bellard 已提交
197 198 199 200 201 202 203 204 205 206 207 208 209 210 211
        header.refcount_table_clusters << (s->cluster_bits - 3);

    s->snapshots_offset = header.snapshots_offset;
    s->nb_snapshots = header.nb_snapshots;

    /* read the level 1 table */
    s->l1_size = header.l1_size;
    shift = s->cluster_bits + s->l2_bits;
    s->l1_vm_state_index = (header.size + (1LL << shift) - 1) >> shift;
    /* the L1 table must contain at least enough entries to put
       header.size bytes */
    if (s->l1_size < s->l1_vm_state_index)
        goto fail;
    s->l1_table_offset = header.l1_table_offset;
    s->l1_table = qemu_malloc(s->l1_size * sizeof(uint64_t));
212
    if (bdrv_pread(s->hd, s->l1_table_offset, s->l1_table, s->l1_size * sizeof(uint64_t)) !=
B
bellard 已提交
213 214 215 216 217 218 219 220 221
        s->l1_size * sizeof(uint64_t))
        goto fail;
    for(i = 0;i < s->l1_size; i++) {
        be64_to_cpus(&s->l1_table[i]);
    }
    /* alloc L2 cache */
    s->l2_cache = qemu_malloc(s->l2_size * L2_CACHE_SIZE * sizeof(uint64_t));
    s->cluster_cache = qemu_malloc(s->cluster_size);
    /* one more sector for decompressed data alignment */
222 223
    s->cluster_data = qemu_malloc(QCOW_MAX_CRYPT_CLUSTERS * s->cluster_size
                                  + 512);
B
bellard 已提交
224
    s->cluster_cache_offset = -1;
225

K
Kevin Wolf 已提交
226
    if (qcow2_refcount_init(bs) < 0)
B
bellard 已提交
227 228
        goto fail;

229 230 231 232 233 234 235 236
    /* read qcow2 extensions */
    if (header.backing_file_offset)
        ext_end = header.backing_file_offset;
    else
        ext_end = s->cluster_size;
    if (qcow_read_extensions(bs, sizeof(header), ext_end))
        goto fail;

B
bellard 已提交
237 238 239 240 241 242 243 244 245
    /* read the backing file name */
    if (header.backing_file_offset != 0) {
        len = header.backing_file_size;
        if (len > 1023)
            len = 1023;
        if (bdrv_pread(s->hd, header.backing_file_offset, bs->backing_file, len) != len)
            goto fail;
        bs->backing_file[len] = '\0';
    }
K
Kevin Wolf 已提交
246
    if (qcow2_read_snapshots(bs) < 0)
B
bellard 已提交
247 248 249
        goto fail;

#ifdef DEBUG_ALLOC
250
    qcow2_check_refcounts(bs);
B
bellard 已提交
251 252 253 254
#endif
    return 0;

 fail:
K
Kevin Wolf 已提交
255 256
    qcow2_free_snapshots(bs);
    qcow2_refcount_close(bs);
B
bellard 已提交
257 258 259 260 261 262 263 264 265 266 267 268 269
    qemu_free(s->l1_table);
    qemu_free(s->l2_cache);
    qemu_free(s->cluster_cache);
    qemu_free(s->cluster_data);
    bdrv_delete(s->hd);
    return -1;
}

static int qcow_set_key(BlockDriverState *bs, const char *key)
{
    BDRVQcowState *s = bs->opaque;
    uint8_t keybuf[16];
    int len, i;
270

B
bellard 已提交
271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306
    memset(keybuf, 0, 16);
    len = strlen(key);
    if (len > 16)
        len = 16;
    /* XXX: we could compress the chars to 7 bits to increase
       entropy */
    for(i = 0;i < len;i++) {
        keybuf[i] = key[i];
    }
    s->crypt_method = s->crypt_method_header;

    if (AES_set_encrypt_key(keybuf, 128, &s->aes_encrypt_key) != 0)
        return -1;
    if (AES_set_decrypt_key(keybuf, 128, &s->aes_decrypt_key) != 0)
        return -1;
#if 0
    /* test */
    {
        uint8_t in[16];
        uint8_t out[16];
        uint8_t tmp[16];
        for(i=0;i<16;i++)
            in[i] = i;
        AES_encrypt(in, tmp, &s->aes_encrypt_key);
        AES_decrypt(tmp, out, &s->aes_decrypt_key);
        for(i = 0; i < 16; i++)
            printf(" %02x", tmp[i]);
        printf("\n");
        for(i = 0; i < 16; i++)
            printf(" %02x", out[i]);
        printf("\n");
    }
#endif
    return 0;
}

307
static int qcow_is_allocated(BlockDriverState *bs, int64_t sector_num,
B
bellard 已提交
308 309 310 311
                             int nb_sectors, int *pnum)
{
    uint64_t cluster_offset;

312
    *pnum = nb_sectors;
K
Kevin Wolf 已提交
313
    cluster_offset = qcow2_get_cluster_offset(bs, sector_num << 9, pnum);
314

B
bellard 已提交
315 316 317
    return (cluster_offset != 0);
}

B
bellard 已提交
318
/* handle reading after the end of the backing file */
K
Kevin Wolf 已提交
319
int qcow2_backing_read1(BlockDriverState *bs,
320
                  int64_t sector_num, uint8_t *buf, int nb_sectors)
B
bellard 已提交
321 322 323 324 325 326 327 328 329 330 331 332
{
    int n1;
    if ((sector_num + nb_sectors) <= bs->total_sectors)
        return nb_sectors;
    if (sector_num >= bs->total_sectors)
        n1 = 0;
    else
        n1 = bs->total_sectors - sector_num;
    memset(buf + n1 * 512, 0, 512 * (nb_sectors - n1));
    return n1;
}

333 334
typedef struct QCowAIOCB {
    BlockDriverAIOCB common;
B
bellard 已提交
335
    int64_t sector_num;
336
    QEMUIOVector *qiov;
B
bellard 已提交
337
    uint8_t *buf;
338
    void *orig_buf;
B
bellard 已提交
339 340 341
    int nb_sectors;
    int n;
    uint64_t cluster_offset;
342
    uint8_t *cluster_data;
B
bellard 已提交
343
    BlockDriverAIOCB *hd_aiocb;
344 345
    struct iovec hd_iov;
    QEMUIOVector hd_qiov;
346
    QEMUBH *bh;
347
    QCowL2Meta l2meta;
B
bellard 已提交
348 349
} QCowAIOCB;

350 351 352 353 354 355 356 357 358 359 360 361 362
static void qcow_aio_cancel(BlockDriverAIOCB *blockacb)
{
    QCowAIOCB *acb = (QCowAIOCB *)blockacb;
    if (acb->hd_aiocb)
        bdrv_aio_cancel(acb->hd_aiocb);
    qemu_aio_release(acb);
}

static AIOPool qcow_aio_pool = {
    .aiocb_size         = sizeof(QCowAIOCB),
    .cancel             = qcow_aio_cancel,
};

363 364 365 366 367 368 369 370 371
static void qcow_aio_read_cb(void *opaque, int ret);
static void qcow_aio_read_bh(void *opaque)
{
    QCowAIOCB *acb = opaque;
    qemu_bh_delete(acb->bh);
    acb->bh = NULL;
    qcow_aio_read_cb(opaque, 0);
}

372 373 374 375 376 377 378 379 380 381 382 383 384 385
static int qcow_schedule_bh(QEMUBHFunc *cb, QCowAIOCB *acb)
{
    if (acb->bh)
        return -EIO;

    acb->bh = qemu_bh_new(cb, acb);
    if (!acb->bh)
        return -EIO;

    qemu_bh_schedule(acb->bh);

    return 0;
}

B
bellard 已提交
386 387
static void qcow_aio_read_cb(void *opaque, int ret)
{
388 389
    QCowAIOCB *acb = opaque;
    BlockDriverState *bs = acb->common.bs;
B
bellard 已提交
390
    BDRVQcowState *s = bs->opaque;
B
bellard 已提交
391
    int index_in_cluster, n1;
B
bellard 已提交
392

393
    acb->hd_aiocb = NULL;
394 395
    if (ret < 0)
        goto done;
B
bellard 已提交
396 397

    /* post process the read buffer */
398
    if (!acb->cluster_offset) {
B
bellard 已提交
399
        /* nothing to do */
400
    } else if (acb->cluster_offset & QCOW_OFLAG_COMPRESSED) {
B
bellard 已提交
401 402 403
        /* nothing to do */
    } else {
        if (s->crypt_method) {
K
Kevin Wolf 已提交
404
            qcow2_encrypt_sectors(s, acb->sector_num, acb->buf, acb->buf,
405
                            acb->n, 0,
B
bellard 已提交
406 407 408 409
                            &s->aes_decrypt_key);
        }
    }

410 411 412
    acb->nb_sectors -= acb->n;
    acb->sector_num += acb->n;
    acb->buf += acb->n * 512;
B
bellard 已提交
413

414
    if (acb->nb_sectors == 0) {
B
bellard 已提交
415
        /* request completed */
416 417
        ret = 0;
        goto done;
B
bellard 已提交
418
    }
419

B
bellard 已提交
420
    /* prepare next AIO request */
421
    acb->n = acb->nb_sectors;
K
Kevin Wolf 已提交
422 423
    acb->cluster_offset =
        qcow2_get_cluster_offset(bs, acb->sector_num << 9, &acb->n);
424 425 426
    index_in_cluster = acb->sector_num & (s->cluster_sectors - 1);

    if (!acb->cluster_offset) {
B
bellard 已提交
427 428
        if (bs->backing_hd) {
            /* read from the base image */
K
Kevin Wolf 已提交
429
            n1 = qcow2_backing_read1(bs->backing_hd, acb->sector_num,
430
                               acb->buf, acb->n);
B
bellard 已提交
431
            if (n1 > 0) {
432
                acb->hd_iov.iov_base = (void *)acb->buf;
433 434 435 436 437
                acb->hd_iov.iov_len = acb->n * 512;
                qemu_iovec_init_external(&acb->hd_qiov, &acb->hd_iov, 1);
                acb->hd_aiocb = bdrv_aio_readv(bs->backing_hd, acb->sector_num,
                                    &acb->hd_qiov, acb->n,
				    qcow_aio_read_cb, acb);
438
                if (acb->hd_aiocb == NULL)
439
                    goto done;
B
bellard 已提交
440
            } else {
441 442
                ret = qcow_schedule_bh(qcow_aio_read_bh, acb);
                if (ret < 0)
443
                    goto done;
B
bellard 已提交
444
            }
B
bellard 已提交
445 446
        } else {
            /* Note: in this case, no need to wait */
447
            memset(acb->buf, 0, 512 * acb->n);
448 449
            ret = qcow_schedule_bh(qcow_aio_read_bh, acb);
            if (ret < 0)
450
                goto done;
B
bellard 已提交
451
        }
452
    } else if (acb->cluster_offset & QCOW_OFLAG_COMPRESSED) {
B
bellard 已提交
453
        /* add AIO support for compressed blocks ? */
K
Kevin Wolf 已提交
454
        if (qcow2_decompress_cluster(s, acb->cluster_offset) < 0)
455
            goto done;
456
        memcpy(acb->buf,
457
               s->cluster_cache + index_in_cluster * 512, 512 * acb->n);
458 459
        ret = qcow_schedule_bh(qcow_aio_read_bh, acb);
        if (ret < 0)
460
            goto done;
B
bellard 已提交
461
    } else {
462
        if ((acb->cluster_offset & 511) != 0) {
B
bellard 已提交
463
            ret = -EIO;
464
            goto done;
B
bellard 已提交
465
        }
466

467
        acb->hd_iov.iov_base = (void *)acb->buf;
468 469 470
        acb->hd_iov.iov_len = acb->n * 512;
        qemu_iovec_init_external(&acb->hd_qiov, &acb->hd_iov, 1);
        acb->hd_aiocb = bdrv_aio_readv(s->hd,
471
                            (acb->cluster_offset >> 9) + index_in_cluster,
472
                            &acb->hd_qiov, acb->n, qcow_aio_read_cb, acb);
473
        if (acb->hd_aiocb == NULL)
474 475 476 477 478 479 480 481
            goto done;
    }

    return;
done:
    if (acb->qiov->niov > 1) {
        qemu_iovec_from_buffer(acb->qiov, acb->orig_buf, acb->qiov->size);
        qemu_vfree(acb->orig_buf);
B
bellard 已提交
482
    }
483 484
    acb->common.cb(acb->common.opaque, ret);
    qemu_aio_release(acb);
B
bellard 已提交
485 486
}

487
static QCowAIOCB *qcow_aio_setup(BlockDriverState *bs,
488 489
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
        BlockDriverCompletionFunc *cb, void *opaque, int is_write)
B
bellard 已提交
490
{
491 492
    QCowAIOCB *acb;

493
    acb = qemu_aio_get(&qcow_aio_pool, bs, cb, opaque);
494 495 496 497
    if (!acb)
        return NULL;
    acb->hd_aiocb = NULL;
    acb->sector_num = sector_num;
498 499
    acb->qiov = qiov;
    if (qiov->niov > 1) {
500
        acb->buf = acb->orig_buf = qemu_blockalign(bs, qiov->size);
501 502
        if (is_write)
            qemu_iovec_to_buffer(qiov, acb->buf);
503 504 505
    } else {
        acb->buf = (uint8_t *)qiov->iov->iov_base;
    }
506 507 508
    acb->nb_sectors = nb_sectors;
    acb->n = 0;
    acb->cluster_offset = 0;
509
    acb->l2meta.nb_clusters = 0;
510 511 512
    return acb;
}

513 514
static BlockDriverAIOCB *qcow_aio_readv(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
515 516 517 518
        BlockDriverCompletionFunc *cb, void *opaque)
{
    QCowAIOCB *acb;

519
    acb = qcow_aio_setup(bs, sector_num, qiov, nb_sectors, cb, opaque, 0);
520 521
    if (!acb)
        return NULL;
B
bellard 已提交
522 523

    qcow_aio_read_cb(acb, 0);
524
    return &acb->common;
B
bellard 已提交
525 526 527 528
}

static void qcow_aio_write_cb(void *opaque, int ret)
{
529 530
    QCowAIOCB *acb = opaque;
    BlockDriverState *bs = acb->common.bs;
B
bellard 已提交
531 532 533
    BDRVQcowState *s = bs->opaque;
    int index_in_cluster;
    const uint8_t *src_buf;
534
    int n_end;
535 536 537

    acb->hd_aiocb = NULL;

538 539
    if (ret < 0)
        goto done;
B
bellard 已提交
540

K
Kevin Wolf 已提交
541 542
    if (qcow2_alloc_cluster_link_l2(bs, acb->cluster_offset, &acb->l2meta) < 0) {
        qcow2_free_any_clusters(bs, acb->cluster_offset, acb->l2meta.nb_clusters);
543
        goto done;
544 545
    }

546 547 548
    acb->nb_sectors -= acb->n;
    acb->sector_num += acb->n;
    acb->buf += acb->n * 512;
B
bellard 已提交
549

550
    if (acb->nb_sectors == 0) {
B
bellard 已提交
551
        /* request completed */
552 553
        ret = 0;
        goto done;
B
bellard 已提交
554
    }
555

556
    index_in_cluster = acb->sector_num & (s->cluster_sectors - 1);
557 558 559 560 561
    n_end = index_in_cluster + acb->nb_sectors;
    if (s->crypt_method &&
        n_end > QCOW_MAX_CRYPT_CLUSTERS * s->cluster_sectors)
        n_end = QCOW_MAX_CRYPT_CLUSTERS * s->cluster_sectors;

K
Kevin Wolf 已提交
562
    acb->cluster_offset = qcow2_alloc_cluster_offset(bs, acb->sector_num << 9,
563
                                          index_in_cluster,
564 565
                                          n_end, &acb->n, &acb->l2meta);
    if (!acb->cluster_offset || (acb->cluster_offset & 511) != 0) {
B
bellard 已提交
566
        ret = -EIO;
567
        goto done;
B
bellard 已提交
568 569
    }
    if (s->crypt_method) {
570
        if (!acb->cluster_data) {
571 572
            acb->cluster_data = qemu_mallocz(QCOW_MAX_CRYPT_CLUSTERS *
                                             s->cluster_size);
B
bellard 已提交
573
        }
K
Kevin Wolf 已提交
574
        qcow2_encrypt_sectors(s, acb->sector_num, acb->cluster_data, acb->buf,
575 576
                        acb->n, 1, &s->aes_encrypt_key);
        src_buf = acb->cluster_data;
B
bellard 已提交
577
    } else {
578
        src_buf = acb->buf;
B
bellard 已提交
579
    }
580 581 582 583 584 585 586
    acb->hd_iov.iov_base = (void *)src_buf;
    acb->hd_iov.iov_len = acb->n * 512;
    qemu_iovec_init_external(&acb->hd_qiov, &acb->hd_iov, 1);
    acb->hd_aiocb = bdrv_aio_writev(s->hd,
                                    (acb->cluster_offset >> 9) + index_in_cluster,
                                    &acb->hd_qiov, acb->n,
                                    qcow_aio_write_cb, acb);
587
    if (acb->hd_aiocb == NULL)
588 589 590 591 592 593 594 595 596
        goto done;

    return;

done:
    if (acb->qiov->niov > 1)
        qemu_vfree(acb->orig_buf);
    acb->common.cb(acb->common.opaque, ret);
    qemu_aio_release(acb);
B
bellard 已提交
597 598
}

599 600
static BlockDriverAIOCB *qcow_aio_writev(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
601
        BlockDriverCompletionFunc *cb, void *opaque)
B
bellard 已提交
602 603
{
    BDRVQcowState *s = bs->opaque;
604
    QCowAIOCB *acb;
605

B
bellard 已提交
606 607
    s->cluster_cache_offset = -1; /* disable compressed cache */

608
    acb = qcow_aio_setup(bs, sector_num, qiov, nb_sectors, cb, opaque, 1);
609 610
    if (!acb)
        return NULL;
611

B
bellard 已提交
612
    qcow_aio_write_cb(acb, 0);
613
    return &acb->common;
B
bellard 已提交
614 615 616 617 618 619 620 621 622
}

static void qcow_close(BlockDriverState *bs)
{
    BDRVQcowState *s = bs->opaque;
    qemu_free(s->l1_table);
    qemu_free(s->l2_cache);
    qemu_free(s->cluster_cache);
    qemu_free(s->cluster_data);
K
Kevin Wolf 已提交
623
    qcow2_refcount_close(bs);
B
bellard 已提交
624 625 626
    bdrv_delete(s->hd);
}

627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647
static int get_bits_from_size(size_t size)
{
    int res = 0;

    if (size == 0) {
        return -1;
    }

    while (size != 1) {
        /* Not a power of two */
        if (size & 1) {
            return -1;
        }

        size >>= 1;
        res++;
    }

    return res;
}

648 649
static int qcow_create2(const char *filename, int64_t total_size,
                        const char *backing_file, const char *backing_format,
650
                        int flags, size_t cluster_size)
B
bellard 已提交
651
{
652

B
bellard 已提交
653
    int fd, header_size, backing_filename_len, l1_size, i, shift, l2_bits;
654
    int ref_clusters, backing_format_len = 0;
B
bellard 已提交
655 656 657
    QCowHeader header;
    uint64_t tmp, offset;
    QCowCreateState s1, *s = &s1;
658 659
    QCowExtension ext_bf = {0, 0};

660

B
bellard 已提交
661 662 663 664 665 666 667 668 669 670 671 672
    memset(s, 0, sizeof(*s));

    fd = open(filename, O_WRONLY | O_CREAT | O_TRUNC | O_BINARY, 0644);
    if (fd < 0)
        return -1;
    memset(&header, 0, sizeof(header));
    header.magic = cpu_to_be32(QCOW_MAGIC);
    header.version = cpu_to_be32(QCOW_VERSION);
    header.size = cpu_to_be64(total_size * 512);
    header_size = sizeof(header);
    backing_filename_len = 0;
    if (backing_file) {
673 674 675 676 677 678
        if (backing_format) {
            ext_bf.magic = QCOW_EXT_MAGIC_BACKING_FORMAT;
            backing_format_len = strlen(backing_format);
            ext_bf.len = (backing_format_len + 7) & ~7;
            header_size += ((sizeof(ext_bf) + ext_bf.len + 7) & ~7);
        }
B
bellard 已提交
679 680 681 682 683
        header.backing_file_offset = cpu_to_be64(header_size);
        backing_filename_len = strlen(backing_file);
        header.backing_file_size = cpu_to_be32(backing_filename_len);
        header_size += backing_filename_len;
    }
684 685 686 687 688 689 690 691 692 693 694 695

    /* Cluster size */
    s->cluster_bits = get_bits_from_size(cluster_size);
    if (s->cluster_bits < MIN_CLUSTER_BITS ||
        s->cluster_bits > MAX_CLUSTER_BITS)
    {
        fprintf(stderr, "Cluster size must be a power of two between "
            "%d and %dk\n",
            1 << MIN_CLUSTER_BITS,
            1 << (MAX_CLUSTER_BITS - 10));
        return -EINVAL;
    }
B
bellard 已提交
696
    s->cluster_size = 1 << s->cluster_bits;
697

B
bellard 已提交
698 699
    header.cluster_bits = cpu_to_be32(s->cluster_bits);
    header_size = (header_size + 7) & ~7;
700
    if (flags & BLOCK_FLAG_ENCRYPT) {
B
bellard 已提交
701 702 703 704 705 706 707 708 709 710 711
        header.crypt_method = cpu_to_be32(QCOW_CRYPT_AES);
    } else {
        header.crypt_method = cpu_to_be32(QCOW_CRYPT_NONE);
    }
    l2_bits = s->cluster_bits - 3;
    shift = s->cluster_bits + l2_bits;
    l1_size = (((total_size * 512) + (1LL << shift) - 1) >> shift);
    offset = align_offset(header_size, s->cluster_size);
    s->l1_table_offset = offset;
    header.l1_table_offset = cpu_to_be64(s->l1_table_offset);
    header.l1_size = cpu_to_be32(l1_size);
B
bellard 已提交
712
    offset += align_offset(l1_size * sizeof(uint64_t), s->cluster_size);
B
bellard 已提交
713 714

    s->refcount_table = qemu_mallocz(s->cluster_size);
715

B
bellard 已提交
716 717 718 719 720
    s->refcount_table_offset = offset;
    header.refcount_table_offset = cpu_to_be64(offset);
    header.refcount_table_clusters = cpu_to_be32(1);
    offset += s->cluster_size;
    s->refcount_block_offset = offset;
721 722 723 724 725 726 727 728 729 730

    /* count how many refcount blocks needed */
    tmp = offset >> s->cluster_bits;
    ref_clusters = (tmp >> (s->cluster_bits - REFCOUNT_SHIFT)) + 1;
    for (i=0; i < ref_clusters; i++) {
        s->refcount_table[i] = cpu_to_be64(offset);
        offset += s->cluster_size;
    }

    s->refcount_block = qemu_mallocz(ref_clusters * s->cluster_size);
B
bellard 已提交
731 732

    /* update refcounts */
K
Kevin Wolf 已提交
733 734 735 736 737 738
    qcow2_create_refcount_update(s, 0, header_size);
    qcow2_create_refcount_update(s, s->l1_table_offset,
        l1_size * sizeof(uint64_t));
    qcow2_create_refcount_update(s, s->refcount_table_offset, s->cluster_size);
    qcow2_create_refcount_update(s, s->refcount_block_offset,
        ref_clusters * s->cluster_size);
739

B
bellard 已提交
740 741 742
    /* write all the data */
    write(fd, &header, sizeof(header));
    if (backing_file) {
743 744 745 746 747 748 749 750 751 752 753 754 755
        if (backing_format_len) {
            char zero[16];
            int d = ext_bf.len - backing_format_len;

            memset(zero, 0, sizeof(zero));
            cpu_to_be32s(&ext_bf.magic);
            cpu_to_be32s(&ext_bf.len);
            write(fd, &ext_bf, sizeof(ext_bf));
            write(fd, backing_format, backing_format_len);
            if (d>0) {
                write(fd, zero, d);
            }
        }
B
bellard 已提交
756 757 758 759 760 761 762 763 764
        write(fd, backing_file, backing_filename_len);
    }
    lseek(fd, s->l1_table_offset, SEEK_SET);
    tmp = 0;
    for(i = 0;i < l1_size; i++) {
        write(fd, &tmp, sizeof(tmp));
    }
    lseek(fd, s->refcount_table_offset, SEEK_SET);
    write(fd, s->refcount_table, s->cluster_size);
765

B
bellard 已提交
766
    lseek(fd, s->refcount_block_offset, SEEK_SET);
767
    write(fd, s->refcount_block, ref_clusters * s->cluster_size);
B
bellard 已提交
768 769 770 771 772 773 774

    qemu_free(s->refcount_table);
    qemu_free(s->refcount_block);
    close(fd);
    return 0;
}

775 776 777 778 779 780
static int qcow_create(const char *filename, QEMUOptionParameter *options)
{
    const char *backing_file = NULL;
    const char *backing_fmt = NULL;
    uint64_t sectors = 0;
    int flags = 0;
781
    size_t cluster_size = 65536;
782 783 784 785 786 787 788 789 790 791 792

    /* Read out options */
    while (options && options->name) {
        if (!strcmp(options->name, BLOCK_OPT_SIZE)) {
            sectors = options->value.n / 512;
        } else if (!strcmp(options->name, BLOCK_OPT_BACKING_FILE)) {
            backing_file = options->value.s;
        } else if (!strcmp(options->name, BLOCK_OPT_BACKING_FMT)) {
            backing_fmt = options->value.s;
        } else if (!strcmp(options->name, BLOCK_OPT_ENCRYPT)) {
            flags |= options->value.n ? BLOCK_FLAG_ENCRYPT : 0;
793 794 795 796
        } else if (!strcmp(options->name, BLOCK_OPT_CLUSTER_SIZE)) {
            if (options->value.n) {
                cluster_size = options->value.n;
            }
797 798 799 800
        }
        options++;
    }

801 802
    return qcow_create2(filename, sectors, backing_file, backing_fmt, flags,
        cluster_size);
803 804
}

B
bellard 已提交
805 806 807 808 809 810 811 812 813 814
static int qcow_make_empty(BlockDriverState *bs)
{
#if 0
    /* XXX: not correct */
    BDRVQcowState *s = bs->opaque;
    uint32_t l1_length = s->l1_size * sizeof(uint64_t);
    int ret;

    memset(s->l1_table, 0, l1_length);
    if (bdrv_pwrite(s->hd, s->l1_table_offset, s->l1_table, l1_length) < 0)
815
        return -1;
B
bellard 已提交
816 817 818
    ret = bdrv_truncate(s->hd, s->l1_table_offset + l1_length);
    if (ret < 0)
        return ret;
819

B
bellard 已提交
820 821 822 823 824 825 826
    l2_cache_reset(bs);
#endif
    return 0;
}

/* XXX: put compressed sectors first, then all the cluster aligned
   tables to avoid losing bytes in alignment */
827
static int qcow_write_compressed(BlockDriverState *bs, int64_t sector_num,
B
bellard 已提交
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
                                 const uint8_t *buf, int nb_sectors)
{
    BDRVQcowState *s = bs->opaque;
    z_stream strm;
    int ret, out_len;
    uint8_t *out_buf;
    uint64_t cluster_offset;

    if (nb_sectors == 0) {
        /* align end of file to a sector boundary to ease reading with
           sector based I/Os */
        cluster_offset = bdrv_getlength(s->hd);
        cluster_offset = (cluster_offset + 511) & ~511;
        bdrv_truncate(s->hd, cluster_offset);
        return 0;
    }

    if (nb_sectors != s->cluster_sectors)
        return -EINVAL;

    out_buf = qemu_malloc(s->cluster_size + (s->cluster_size / 1000) + 128);

    /* best compression, small window, no zlib header */
    memset(&strm, 0, sizeof(strm));
    ret = deflateInit2(&strm, Z_DEFAULT_COMPRESSION,
853
                       Z_DEFLATED, -12,
B
bellard 已提交
854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876
                       9, Z_DEFAULT_STRATEGY);
    if (ret != 0) {
        qemu_free(out_buf);
        return -1;
    }

    strm.avail_in = s->cluster_size;
    strm.next_in = (uint8_t *)buf;
    strm.avail_out = s->cluster_size;
    strm.next_out = out_buf;

    ret = deflate(&strm, Z_FINISH);
    if (ret != Z_STREAM_END && ret != Z_OK) {
        qemu_free(out_buf);
        deflateEnd(&strm);
        return -1;
    }
    out_len = strm.next_out - out_buf;

    deflateEnd(&strm);

    if (ret != Z_STREAM_END || out_len >= s->cluster_size) {
        /* could not compress: write normal cluster */
877
        bdrv_write(bs, sector_num, buf, s->cluster_sectors);
B
bellard 已提交
878
    } else {
K
Kevin Wolf 已提交
879 880
        cluster_offset = qcow2_alloc_compressed_cluster_offset(bs,
            sector_num << 9, out_len);
881 882
        if (!cluster_offset)
            return -1;
B
bellard 已提交
883 884 885 886 887 888
        cluster_offset &= s->cluster_offset_mask;
        if (bdrv_pwrite(s->hd, cluster_offset, out_buf, out_len) != out_len) {
            qemu_free(out_buf);
            return -1;
        }
    }
889

B
bellard 已提交
890 891 892 893 894 895 896 897 898 899 900 901 902 903
    qemu_free(out_buf);
    return 0;
}

static void qcow_flush(BlockDriverState *bs)
{
    BDRVQcowState *s = bs->opaque;
    bdrv_flush(s->hd);
}

static int qcow_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
{
    BDRVQcowState *s = bs->opaque;
    bdi->cluster_size = s->cluster_size;
904
    bdi->vm_state_offset = (int64_t)s->l1_vm_state_index <<
B
bellard 已提交
905 906 907 908 909
        (s->cluster_bits + s->l2_bits);
    return 0;
}


A
aliguori 已提交
910 911
static int qcow_check(BlockDriverState *bs)
{
K
Kevin Wolf 已提交
912
    return qcow2_check_refcounts(bs);
B
bellard 已提交
913 914 915 916 917 918 919 920 921 922
}

#if 0
static void dump_refcounts(BlockDriverState *bs)
{
    BDRVQcowState *s = bs->opaque;
    int64_t nb_clusters, k, k1, size;
    int refcount;

    size = bdrv_getlength(s->hd);
923
    nb_clusters = size_to_clusters(s, size);
B
bellard 已提交
924 925 926 927 928 929 930 931 932 933 934
    for(k = 0; k < nb_clusters;) {
        k1 = k;
        refcount = get_refcount(bs, k);
        k++;
        while (k < nb_clusters && get_refcount(bs, k) == refcount)
            k++;
        printf("%lld: refcount=%d nb=%lld\n", k, refcount, k - k1);
    }
}
#endif

935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959
static int qcow_put_buffer(BlockDriverState *bs, const uint8_t *buf,
                           int64_t pos, int size)
{
    int growable = bs->growable;

    bs->growable = 1;
    bdrv_pwrite(bs, pos, buf, size);
    bs->growable = growable;

    return size;
}

static int qcow_get_buffer(BlockDriverState *bs, uint8_t *buf,
                           int64_t pos, int size)
{
    int growable = bs->growable;
    int ret;

    bs->growable = 1;
    ret = bdrv_pread(bs, pos, buf, size);
    bs->growable = growable;

    return ret;
}

960
static QEMUOptionParameter qcow_create_options[] = {
961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985
    {
        .name = BLOCK_OPT_SIZE,
        .type = OPT_SIZE,
        .help = "Virtual disk size"
    },
    {
        .name = BLOCK_OPT_BACKING_FILE,
        .type = OPT_STRING,
        .help = "File name of a base image"
    },
    {
        .name = BLOCK_OPT_BACKING_FMT,
        .type = OPT_STRING,
        .help = "Image format of the base image"
    },
    {
        .name = BLOCK_OPT_ENCRYPT,
        .type = OPT_FLAG,
        .help = "Encrypt the image"
    },
    {
        .name = BLOCK_OPT_CLUSTER_SIZE,
        .type = OPT_SIZE,
        .help = "qcow2 cluster size"
    },
986 987 988
    { NULL }
};

989
static BlockDriver bdrv_qcow2 = {
990 991 992 993 994 995 996 997 998 999 1000
    .format_name	= "qcow2",
    .instance_size	= sizeof(BDRVQcowState),
    .bdrv_probe		= qcow_probe,
    .bdrv_open		= qcow_open,
    .bdrv_close		= qcow_close,
    .bdrv_create	= qcow_create,
    .bdrv_flush		= qcow_flush,
    .bdrv_is_allocated	= qcow_is_allocated,
    .bdrv_set_key	= qcow_set_key,
    .bdrv_make_empty	= qcow_make_empty,

1001 1002
    .bdrv_aio_readv	= qcow_aio_readv,
    .bdrv_aio_writev	= qcow_aio_writev,
B
bellard 已提交
1003 1004
    .bdrv_write_compressed = qcow_write_compressed,

K
Kevin Wolf 已提交
1005 1006 1007 1008
    .bdrv_snapshot_create   = qcow2_snapshot_create,
    .bdrv_snapshot_goto     = qcow2_snapshot_goto,
    .bdrv_snapshot_delete   = qcow2_snapshot_delete,
    .bdrv_snapshot_list     = qcow2_snapshot_list,
1009
    .bdrv_get_info	= qcow_get_info,
1010

1011 1012 1013
    .bdrv_put_buffer    = qcow_put_buffer,
    .bdrv_get_buffer    = qcow_get_buffer,

1014
    .create_options = qcow_create_options,
A
aliguori 已提交
1015
    .bdrv_check = qcow_check,
B
bellard 已提交
1016
};
1017 1018 1019 1020 1021 1022 1023

static void bdrv_qcow2_init(void)
{
    bdrv_register(&bdrv_qcow2);
}

block_init(bdrv_qcow2_init);