vpc.c 27.2 KB
Newer Older
1
/*
S
Stefan Weil 已提交
2
 * Block driver for Connectix / Microsoft Virtual PC images
3
 *
4
 * Copyright (c) 2005 Alex Beregszaszi
A
aliguori 已提交
5
 * Copyright (c) 2009 Kevin Wolf <kwolf@suse.de>
6
 *
7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
 * 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
25
#include "qemu-common.h"
26
#include "block/block_int.h"
27
#include "qemu/module.h"
28
#include "migration/migration.h"
29 30 31
#if defined(CONFIG_UUID)
#include <uuid/uuid.h>
#endif
32 33 34 35 36 37 38

/**************************************************************/

#define HEADER_SIZE 512

//#define CACHE

39 40 41 42 43 44
enum vhd_type {
    VHD_FIXED           = 2,
    VHD_DYNAMIC         = 3,
    VHD_DIFFERENCING    = 4,
};

A
aliguori 已提交
45 46 47
// Seconds since Jan 1, 2000 0:00:00 (UTC)
#define VHD_TIMESTAMP_BASE 946684800

48 49
#define VHD_MAX_SECTORS       (65535LL * 255 * 255)

50
// always big-endian
51
typedef struct vhd_footer {
52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83
    char        creator[8]; // "conectix"
    uint32_t    features;
    uint32_t    version;

    // Offset of next header structure, 0xFFFFFFFF if none
    uint64_t    data_offset;

    // Seconds since Jan 1, 2000 0:00:00 (UTC)
    uint32_t    timestamp;

    char        creator_app[4]; // "vpc "
    uint16_t    major;
    uint16_t    minor;
    char        creator_os[4]; // "Wi2k"

    uint64_t    orig_size;
    uint64_t    size;

    uint16_t    cyls;
    uint8_t     heads;
    uint8_t     secs_per_cyl;

    uint32_t    type;

    // Checksum of the Hard Disk Footer ("one's complement of the sum of all
    // the bytes in the footer without the checksum field")
    uint32_t    checksum;

    // UUID used to identify a parent hard disk (backing file)
    uint8_t     uuid[16];

    uint8_t     in_saved_state;
84
} QEMU_PACKED VHDFooter;
85

86
typedef struct vhd_dyndisk_header {
87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115
    char        magic[8]; // "cxsparse"

    // Offset of next header structure, 0xFFFFFFFF if none
    uint64_t    data_offset;

    // Offset of the Block Allocation Table (BAT)
    uint64_t    table_offset;

    uint32_t    version;
    uint32_t    max_table_entries; // 32bit/entry

    // 2 MB by default, must be a power of two
    uint32_t    block_size;

    uint32_t    checksum;
    uint8_t     parent_uuid[16];
    uint32_t    parent_timestamp;
    uint32_t    reserved;

    // Backing file name (in UTF-16)
    uint8_t     parent_name[512];

    struct {
        uint32_t    platform;
        uint32_t    data_space;
        uint32_t    data_length;
        uint32_t    reserved;
        uint64_t    data_offset;
    } parent_locator[8];
116
} QEMU_PACKED VHDDynDiskHeader;
117 118

typedef struct BDRVVPCState {
119
    CoMutex lock;
A
aliguori 已提交
120 121
    uint8_t footer_buf[HEADER_SIZE];
    uint64_t free_data_block_offset;
122
    int max_table_entries;
123
    uint32_t *pagetable;
A
aliguori 已提交
124 125
    uint64_t bat_offset;
    uint64_t last_bitmap_offset;
126

127
    uint32_t block_size;
A
aliguori 已提交
128 129
    uint32_t bitmap_size;

130 131 132 133
#ifdef CACHE
    uint8_t *pageentry_u8;
    uint32_t *pageentry_u32;
    uint16_t *pageentry_u16;
134

135 136
    uint64_t last_bitmap;
#endif
K
Kevin Wolf 已提交
137 138

    Error *migration_blocker;
139 140
} BDRVVPCState;

A
aliguori 已提交
141 142 143 144 145 146 147 148 149 150 151 152
static uint32_t vpc_checksum(uint8_t* buf, size_t size)
{
    uint32_t res = 0;
    int i;

    for (i = 0; i < size; i++)
        res += buf[i];

    return ~res;
}


153 154
static int vpc_probe(const uint8_t *buf, int buf_size, const char *filename)
{
T
ths 已提交
155
    if (buf_size >= 8 && !strncmp((char *)buf, "conectix", 8))
156 157 158 159
	return 100;
    return 0;
}

M
Max Reitz 已提交
160 161
static int vpc_open(BlockDriverState *bs, QDict *options, int flags,
                    Error **errp)
162 163
{
    BDRVVPCState *s = bs->opaque;
164
    int i;
165 166
    VHDFooter *footer;
    VHDDynDiskHeader *dyndisk_header;
167
    uint8_t buf[HEADER_SIZE];
A
aliguori 已提交
168
    uint32_t checksum;
169
    uint64_t computed_size;
170
    int disk_type = VHD_DYNAMIC;
K
Kevin Wolf 已提交
171
    int ret;
172

K
Kevin Wolf 已提交
173 174
    ret = bdrv_pread(bs->file, 0, s->footer_buf, HEADER_SIZE);
    if (ret < 0) {
175
        goto fail;
K
Kevin Wolf 已提交
176
    }
177

178
    footer = (VHDFooter *) s->footer_buf;
179 180
    if (strncmp(footer->creator, "conectix", 8)) {
        int64_t offset = bdrv_getlength(bs->file);
K
Kevin Wolf 已提交
181 182 183 184 185
        if (offset < 0) {
            ret = offset;
            goto fail;
        } else if (offset < HEADER_SIZE) {
            ret = -EINVAL;
186 187
            goto fail;
        }
K
Kevin Wolf 已提交
188

189
        /* If a fixed disk, the footer is found only at the end of the file */
K
Kevin Wolf 已提交
190 191 192
        ret = bdrv_pread(bs->file, offset-HEADER_SIZE, s->footer_buf,
                         HEADER_SIZE);
        if (ret < 0) {
193 194 195
            goto fail;
        }
        if (strncmp(footer->creator, "conectix", 8)) {
P
Paolo Bonzini 已提交
196 197
            error_setg(errp, "invalid VPC image");
            ret = -EINVAL;
198 199 200 201
            goto fail;
        }
        disk_type = VHD_FIXED;
    }
202

A
aliguori 已提交
203 204 205 206
    checksum = be32_to_cpu(footer->checksum);
    footer->checksum = 0;
    if (vpc_checksum(s->footer_buf, HEADER_SIZE) != checksum)
        fprintf(stderr, "block-vpc: The header checksum of '%s' is "
207
            "incorrect.\n", bs->filename);
A
aliguori 已提交
208

209
    /* Write 'checksum' back to footer, or else will leave it with zero. */
210
    footer->checksum = cpu_to_be32(checksum);
211

212 213 214 215 216
    // The visible size of a image in Virtual PC depends on the geometry
    // rather than on the size stored in the footer (the size in the footer
    // is too large usually)
    bs->total_sectors = (int64_t)
        be16_to_cpu(footer->cyls) * footer->heads * footer->secs_per_cyl;
A
aliguori 已提交
217

218 219 220 221
    /* Images that have exactly the maximum geometry are probably bigger and
     * would be truncated if we adhered to the geometry for them. Rely on
     * footer->size for them. */
    if (bs->total_sectors == 65535ULL * 16 * 255) {
222 223 224
        bs->total_sectors = be64_to_cpu(footer->size) / BDRV_SECTOR_SIZE;
    }

225
    /* Allow a maximum disk size of approximately 2 TB */
226
    if (bs->total_sectors >= VHD_MAX_SECTORS) {
K
Kevin Wolf 已提交
227
        ret = -EFBIG;
228 229 230
        goto fail;
    }

231
    if (disk_type == VHD_DYNAMIC) {
K
Kevin Wolf 已提交
232 233 234
        ret = bdrv_pread(bs->file, be64_to_cpu(footer->data_offset), buf,
                         HEADER_SIZE);
        if (ret < 0) {
235 236
            goto fail;
        }
237

238
        dyndisk_header = (VHDDynDiskHeader *) buf;
239

240
        if (strncmp(dyndisk_header->magic, "cxsparse", 8)) {
K
Kevin Wolf 已提交
241
            ret = -EINVAL;
242 243
            goto fail;
        }
244

245
        s->block_size = be32_to_cpu(dyndisk_header->block_size);
246 247 248 249 250
        if (!is_power_of_2(s->block_size) || s->block_size < BDRV_SECTOR_SIZE) {
            error_setg(errp, "Invalid block size %" PRIu32, s->block_size);
            ret = -EINVAL;
            goto fail;
        }
251
        s->bitmap_size = ((s->block_size / (8 * 512)) + 511) & ~511;
A
aliguori 已提交
252

253
        s->max_table_entries = be32_to_cpu(dyndisk_header->max_table_entries);
254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269

        if ((bs->total_sectors * 512) / s->block_size > 0xffffffffU) {
            ret = -EINVAL;
            goto fail;
        }
        if (s->max_table_entries > (VHD_MAX_SECTORS * 512) / s->block_size) {
            ret = -EINVAL;
            goto fail;
        }

        computed_size = (uint64_t) s->max_table_entries * s->block_size;
        if (computed_size < bs->total_sectors * 512) {
            ret = -EINVAL;
            goto fail;
        }

270 271 272 273 274
        s->pagetable = qemu_try_blockalign(bs->file, s->max_table_entries * 4);
        if (s->pagetable == NULL) {
            ret = -ENOMEM;
            goto fail;
        }
275

276
        s->bat_offset = be64_to_cpu(dyndisk_header->table_offset);
K
Kevin Wolf 已提交
277 278 279 280

        ret = bdrv_pread(bs->file, s->bat_offset, s->pagetable,
                         s->max_table_entries * 4);
        if (ret < 0) {
281 282
            goto fail;
        }
283

284 285
        s->free_data_block_offset =
            (s->bat_offset + (s->max_table_entries * 4) + 511) & ~511;
A
aliguori 已提交
286

287 288 289 290 291
        for (i = 0; i < s->max_table_entries; i++) {
            be32_to_cpus(&s->pagetable[i]);
            if (s->pagetable[i] != 0xFFFFFFFF) {
                int64_t next = (512 * (int64_t) s->pagetable[i]) +
                    s->bitmap_size + s->block_size;
A
aliguori 已提交
292

293 294 295 296
                if (next > s->free_data_block_offset) {
                    s->free_data_block_offset = next;
                }
            }
A
aliguori 已提交
297 298
        }

299 300 301 302 303 304 305
        if (s->free_data_block_offset > bdrv_getlength(bs->file)) {
            error_setg(errp, "block-vpc: free_data_block_offset points after "
                             "the end of file. The image has been truncated.");
            ret = -EINVAL;
            goto fail;
        }

306
        s->last_bitmap_offset = (int64_t) -1;
307 308

#ifdef CACHE
309 310 311 312
        s->pageentry_u8 = g_malloc(512);
        s->pageentry_u32 = s->pageentry_u8;
        s->pageentry_u16 = s->pageentry_u8;
        s->last_pagetable = -1;
313
#endif
314
    }
315

316
    qemu_co_mutex_init(&s->lock);
K
Kevin Wolf 已提交
317 318 319 320

    /* Disable migration when VHD images are used */
    error_set(&s->migration_blocker,
              QERR_BLOCK_FORMAT_FEATURE_NOT_SUPPORTED,
321
              "vpc", bdrv_get_device_name(bs), "live migration");
K
Kevin Wolf 已提交
322 323
    migrate_add_blocker(s->migration_blocker);

324
    return 0;
K
Kevin Wolf 已提交
325 326

fail:
327
    qemu_vfree(s->pagetable);
K
Kevin Wolf 已提交
328 329 330 331
#ifdef CACHE
    g_free(s->pageentry_u8);
#endif
    return ret;
332 333
}

J
Jeff Cody 已提交
334 335 336 337 338 339
static int vpc_reopen_prepare(BDRVReopenState *state,
                              BlockReopenQueue *queue, Error **errp)
{
    return 0;
}

340 341 342
/*
 * Returns the absolute byte offset of the given sector in the image file.
 * If the sector is not allocated, -1 is returned instead.
A
aliguori 已提交
343 344 345
 *
 * The parameter write must be 1 if the offset will be used for a write
 * operation (the block bitmaps is updated then), 0 otherwise.
346
 */
A
aliguori 已提交
347 348
static inline int64_t get_sector_offset(BlockDriverState *bs,
    int64_t sector_num, int write)
349 350 351 352 353 354
{
    BDRVVPCState *s = bs->opaque;
    uint64_t offset = sector_num * 512;
    uint64_t bitmap_offset, block_offset;
    uint32_t pagetable_index, pageentry_index;

355 356
    pagetable_index = offset / s->block_size;
    pageentry_index = (offset % s->block_size) / 512;
357

A
aliguori 已提交
358 359
    if (pagetable_index >= s->max_table_entries || s->pagetable[pagetable_index] == 0xffffffff)
        return -1; // not allocated
360

361
    bitmap_offset = 512 * (uint64_t) s->pagetable[pagetable_index];
A
aliguori 已提交
362 363 364 365 366 367 368 369 370 371 372 373
    block_offset = bitmap_offset + s->bitmap_size + (512 * pageentry_index);

    // We must ensure that we don't write to any sectors which are marked as
    // unused in the bitmap. We get away with setting all bits in the block
    // bitmap each time we write to a new block. This might cause Virtual PC to
    // miss sparse read optimization, but it's not a problem in terms of
    // correctness.
    if (write && (s->last_bitmap_offset != bitmap_offset)) {
        uint8_t bitmap[s->bitmap_size];

        s->last_bitmap_offset = bitmap_offset;
        memset(bitmap, 0xff, s->bitmap_size);
374
        bdrv_pwrite_sync(bs->file, bitmap_offset, bitmap, s->bitmap_size);
A
aliguori 已提交
375
    }
376

B
bellard 已提交
377
//    printf("sector: %" PRIx64 ", index: %x, offset: %x, bioff: %" PRIx64 ", bloff: %" PRIx64 "\n",
378 379 380 381 382 383 384 385 386 387 388
//	sector_num, pagetable_index, pageentry_index,
//	bitmap_offset, block_offset);

// disabled by reason
#if 0
#ifdef CACHE
    if (bitmap_offset != s->last_bitmap)
    {
	lseek(s->fd, bitmap_offset, SEEK_SET);

	s->last_bitmap = bitmap_offset;
389

390 391 392 393 394 395 396 397 398 399 400
	// Scary! Bitmap is stored as big endian 32bit entries,
	// while we used to look it up byte by byte
	read(s->fd, s->pageentry_u8, 512);
	for (i = 0; i < 128; i++)
	    be32_to_cpus(&s->pageentry_u32[i]);
    }

    if ((s->pageentry_u8[pageentry_index / 8] >> (pageentry_index % 8)) & 1)
	return -1;
#else
    lseek(s->fd, bitmap_offset + (pageentry_index / 8), SEEK_SET);
401

402 403 404 405 406 407 408
    read(s->fd, &bitmap_entry, 1);

    if ((bitmap_entry >> (pageentry_index % 8)) & 1)
	return -1; // not allocated
#endif
#endif

409
    return block_offset;
410 411
}

A
aliguori 已提交
412 413 414 415 416 417 418 419 420 421 422 423
/*
 * Writes the footer to the end of the image file. This is needed when the
 * file grows as it overwrites the old footer
 *
 * Returns 0 on success and < 0 on error
 */
static int rewrite_footer(BlockDriverState* bs)
{
    int ret;
    BDRVVPCState *s = bs->opaque;
    int64_t offset = s->free_data_block_offset;

424
    ret = bdrv_pwrite_sync(bs->file, offset, s->footer_buf, HEADER_SIZE);
A
aliguori 已提交
425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458
    if (ret < 0)
        return ret;

    return 0;
}

/*
 * Allocates a new block. This involves writing a new footer and updating
 * the Block Allocation Table to use the space at the old end of the image
 * file (overwriting the old footer)
 *
 * Returns the sectors' offset in the image file on success and < 0 on error
 */
static int64_t alloc_block(BlockDriverState* bs, int64_t sector_num)
{
    BDRVVPCState *s = bs->opaque;
    int64_t bat_offset;
    uint32_t index, bat_value;
    int ret;
    uint8_t bitmap[s->bitmap_size];

    // Check if sector_num is valid
    if ((sector_num < 0) || (sector_num > bs->total_sectors))
        return -1;

    // Write entry into in-memory BAT
    index = (sector_num * 512) / s->block_size;
    if (s->pagetable[index] != 0xFFFFFFFF)
        return -1;

    s->pagetable[index] = s->free_data_block_offset / 512;

    // Initialize the block's bitmap
    memset(bitmap, 0xff, s->bitmap_size);
459
    ret = bdrv_pwrite_sync(bs->file, s->free_data_block_offset, bitmap,
460
        s->bitmap_size);
461 462 463
    if (ret < 0) {
        return ret;
    }
A
aliguori 已提交
464 465 466 467 468 469 470 471 472

    // Write new footer (the old one will be overwritten)
    s->free_data_block_offset += s->block_size + s->bitmap_size;
    ret = rewrite_footer(bs);
    if (ret < 0)
        goto fail;

    // Write BAT entry to disk
    bat_offset = s->bat_offset + (4 * index);
473
    bat_value = cpu_to_be32(s->pagetable[index]);
474
    ret = bdrv_pwrite_sync(bs->file, bat_offset, &bat_value, 4);
A
aliguori 已提交
475 476 477 478 479 480 481 482 483 484
    if (ret < 0)
        goto fail;

    return get_sector_offset(bs, sector_num, 0);

fail:
    s->free_data_block_offset -= (s->block_size + s->bitmap_size);
    return -1;
}

P
Paolo Bonzini 已提交
485 486 487 488 489
static int vpc_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
{
    BDRVVPCState *s = (BDRVVPCState *)bs->opaque;
    VHDFooter *footer = (VHDFooter *) s->footer_buf;

490
    if (be32_to_cpu(footer->type) != VHD_FIXED) {
P
Paolo Bonzini 已提交
491 492 493
        bdi->cluster_size = s->block_size;
    }

494
    bdi->unallocated_blocks_are_zero = true;
P
Paolo Bonzini 已提交
495 496 497
    return 0;
}

498
static int vpc_read(BlockDriverState *bs, int64_t sector_num,
499 500
                    uint8_t *buf, int nb_sectors)
{
501
    BDRVVPCState *s = bs->opaque;
502
    int ret;
503
    int64_t offset;
504
    int64_t sectors, sectors_per_block;
505
    VHDFooter *footer = (VHDFooter *) s->footer_buf;
506

507
    if (be32_to_cpu(footer->type) == VHD_FIXED) {
508 509
        return bdrv_read(bs->file, sector_num, buf, nb_sectors);
    }
510
    while (nb_sectors > 0) {
A
aliguori 已提交
511
        offset = get_sector_offset(bs, sector_num, 0);
512

513 514 515 516 517 518
        sectors_per_block = s->block_size >> BDRV_SECTOR_BITS;
        sectors = sectors_per_block - (sector_num % sectors_per_block);
        if (sectors > nb_sectors) {
            sectors = nb_sectors;
        }

519
        if (offset == -1) {
520
            memset(buf, 0, sectors * BDRV_SECTOR_SIZE);
521
        } else {
522 523 524
            ret = bdrv_pread(bs->file, offset, buf,
                sectors * BDRV_SECTOR_SIZE);
            if (ret != sectors * BDRV_SECTOR_SIZE) {
525
                return -1;
526
            }
527 528
        }

529 530 531
        nb_sectors -= sectors;
        sector_num += sectors;
        buf += sectors * BDRV_SECTOR_SIZE;
532 533 534 535
    }
    return 0;
}

536 537 538 539 540 541 542 543 544 545 546
static coroutine_fn int vpc_co_read(BlockDriverState *bs, int64_t sector_num,
                                    uint8_t *buf, int nb_sectors)
{
    int ret;
    BDRVVPCState *s = bs->opaque;
    qemu_co_mutex_lock(&s->lock);
    ret = vpc_read(bs, sector_num, buf, nb_sectors);
    qemu_co_mutex_unlock(&s->lock);
    return ret;
}

A
aliguori 已提交
547 548 549
static int vpc_write(BlockDriverState *bs, int64_t sector_num,
    const uint8_t *buf, int nb_sectors)
{
550
    BDRVVPCState *s = bs->opaque;
A
aliguori 已提交
551
    int64_t offset;
552
    int64_t sectors, sectors_per_block;
A
aliguori 已提交
553
    int ret;
554
    VHDFooter *footer =  (VHDFooter *) s->footer_buf;
A
aliguori 已提交
555

556
    if (be32_to_cpu(footer->type) == VHD_FIXED) {
557 558
        return bdrv_write(bs->file, sector_num, buf, nb_sectors);
    }
A
aliguori 已提交
559 560 561
    while (nb_sectors > 0) {
        offset = get_sector_offset(bs, sector_num, 1);

562 563 564 565 566 567
        sectors_per_block = s->block_size >> BDRV_SECTOR_BITS;
        sectors = sectors_per_block - (sector_num % sectors_per_block);
        if (sectors > nb_sectors) {
            sectors = nb_sectors;
        }

A
aliguori 已提交
568 569 570 571 572 573
        if (offset == -1) {
            offset = alloc_block(bs, sector_num);
            if (offset < 0)
                return -1;
        }

574 575
        ret = bdrv_pwrite(bs->file, offset, buf, sectors * BDRV_SECTOR_SIZE);
        if (ret != sectors * BDRV_SECTOR_SIZE) {
A
aliguori 已提交
576
            return -1;
577
        }
A
aliguori 已提交
578

579 580 581
        nb_sectors -= sectors;
        sector_num += sectors;
        buf += sectors * BDRV_SECTOR_SIZE;
A
aliguori 已提交
582 583 584 585 586
    }

    return 0;
}

587 588 589 590 591 592 593 594 595 596 597
static coroutine_fn int vpc_co_write(BlockDriverState *bs, int64_t sector_num,
                                     const uint8_t *buf, int nb_sectors)
{
    int ret;
    BDRVVPCState *s = bs->opaque;
    qemu_co_mutex_lock(&s->lock);
    ret = vpc_write(bs, sector_num, buf, nb_sectors);
    qemu_co_mutex_unlock(&s->lock);
    return ret;
}

598 599 600 601 602
static int64_t coroutine_fn vpc_co_get_block_status(BlockDriverState *bs,
        int64_t sector_num, int nb_sectors, int *pnum)
{
    BDRVVPCState *s = bs->opaque;
    VHDFooter *footer = (VHDFooter*) s->footer_buf;
603
    int64_t start, offset;
604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626
    bool allocated;
    int n;

    if (be32_to_cpu(footer->type) == VHD_FIXED) {
        *pnum = nb_sectors;
        return BDRV_BLOCK_RAW | BDRV_BLOCK_OFFSET_VALID | BDRV_BLOCK_DATA |
               (sector_num << BDRV_SECTOR_BITS);
    }

    offset = get_sector_offset(bs, sector_num, 0);
    start = offset;
    allocated = (offset != -1);
    *pnum = 0;

    do {
        /* All sectors in a block are contiguous (without using the bitmap) */
        n = ROUND_UP(sector_num + 1, s->block_size / BDRV_SECTOR_SIZE)
          - sector_num;
        n = MIN(n, nb_sectors);

        *pnum += n;
        sector_num += n;
        nb_sectors -= n;
627 628 629 630 631
        /* *pnum can't be greater than one block for allocated
         * sectors since there is always a bitmap in between. */
        if (allocated) {
            return BDRV_BLOCK_DATA | BDRV_BLOCK_OFFSET_VALID | start;
        }
632 633 634 635
        if (nb_sectors == 0) {
            break;
        }
        offset = get_sector_offset(bs, sector_num, 0);
636
    } while (offset == -1);
637

638
    return 0;
639 640
}

A
aliguori 已提交
641 642 643 644 645 646 647
/*
 * Calculates the number of cylinders, heads and sectors per cylinder
 * based on a given number of sectors. This is the algorithm described
 * in the VHD specification.
 *
 * Note that the geometry doesn't always exactly match total_sectors but
 * may round it down.
648
 *
649 650 651
 * Returns 0 on success, -EFBIG if the size is larger than ~2 TB. Override
 * the hardware EIDE and ATA-2 limit of 16 heads (max disk size of 127 GB)
 * and instead allow up to 255 heads.
A
aliguori 已提交
652
 */
653
static int calculate_geometry(int64_t total_sectors, uint16_t* cyls,
A
aliguori 已提交
654 655 656 657
    uint8_t* heads, uint8_t* secs_per_cyl)
{
    uint32_t cyls_times_heads;

658 659
    /* Allow a maximum disk size of approximately 2 TB */
    if (total_sectors > 65535LL * 255 * 255) {
660
        return -EFBIG;
661
    }
A
aliguori 已提交
662 663 664

    if (total_sectors > 65535 * 16 * 63) {
        *secs_per_cyl = 255;
665 666 667 668 669
        if (total_sectors > 65535 * 16 * 255) {
            *heads = 255;
        } else {
            *heads = 16;
        }
A
aliguori 已提交
670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691
        cyls_times_heads = total_sectors / *secs_per_cyl;
    } else {
        *secs_per_cyl = 17;
        cyls_times_heads = total_sectors / *secs_per_cyl;
        *heads = (cyls_times_heads + 1023) / 1024;

        if (*heads < 4)
            *heads = 4;

        if (cyls_times_heads >= (*heads * 1024) || *heads > 16) {
            *secs_per_cyl = 31;
            *heads = 16;
            cyls_times_heads = total_sectors / *secs_per_cyl;
        }

        if (cyls_times_heads >= (*heads * 1024)) {
            *secs_per_cyl = 63;
            *heads = 16;
            cyls_times_heads = total_sectors / *secs_per_cyl;
        }
    }

692
    *cyls = cyls_times_heads / *heads;
693 694

    return 0;
A
aliguori 已提交
695 696
}

697 698
static int create_dynamic_disk(BlockDriverState *bs, uint8_t *buf,
                               int64_t total_sectors)
A
aliguori 已提交
699
{
700 701
    VHDDynDiskHeader *dyndisk_header =
        (VHDDynDiskHeader *) buf;
A
aliguori 已提交
702
    size_t block_size, num_bat_entries;
703
    int i;
704 705
    int ret;
    int64_t offset = 0;
A
aliguori 已提交
706 707 708 709 710

    // Write the footer (twice: at the beginning and at the end)
    block_size = 0x200000;
    num_bat_entries = (total_sectors + block_size / 512) / (block_size / 512);

711 712
    ret = bdrv_pwrite_sync(bs, offset, buf, HEADER_SIZE);
    if (ret) {
B
Blue Swirl 已提交
713 714
        goto fail;
    }
A
aliguori 已提交
715

716 717 718
    offset = 1536 + ((num_bat_entries * 4 + 511) & ~511);
    ret = bdrv_pwrite_sync(bs, offset, buf, HEADER_SIZE);
    if (ret < 0) {
B
Blue Swirl 已提交
719 720
        goto fail;
    }
A
aliguori 已提交
721 722

    // Write the initial BAT
723
    offset = 3 * 512;
A
aliguori 已提交
724 725

    memset(buf, 0xFF, 512);
B
Blue Swirl 已提交
726
    for (i = 0; i < (num_bat_entries * 4 + 511) / 512; i++) {
727 728
        ret = bdrv_pwrite_sync(bs, offset, buf, 512);
        if (ret < 0) {
B
Blue Swirl 已提交
729 730
            goto fail;
        }
731
        offset += 512;
B
Blue Swirl 已提交
732
    }
A
aliguori 已提交
733 734 735 736

    // Prepare the Dynamic Disk Header
    memset(buf, 0, 1024);

737
    memcpy(dyndisk_header->magic, "cxsparse", 8);
A
aliguori 已提交
738

739 740 741 742
    /*
     * Note: The spec is actually wrong here for data_offset, it says
     * 0xFFFFFFFF, but MS tools expect all 64 bits to be set.
     */
743 744 745 746 747
    dyndisk_header->data_offset = cpu_to_be64(0xFFFFFFFFFFFFFFFFULL);
    dyndisk_header->table_offset = cpu_to_be64(3 * 512);
    dyndisk_header->version = cpu_to_be32(0x00010000);
    dyndisk_header->block_size = cpu_to_be32(block_size);
    dyndisk_header->max_table_entries = cpu_to_be32(num_bat_entries);
A
aliguori 已提交
748

749
    dyndisk_header->checksum = cpu_to_be32(vpc_checksum(buf, 1024));
A
aliguori 已提交
750 751

    // Write the header
752
    offset = 512;
A
aliguori 已提交
753

754 755
    ret = bdrv_pwrite_sync(bs, offset, buf, 1024);
    if (ret < 0) {
B
Blue Swirl 已提交
756 757 758
        goto fail;
    }

759 760 761 762
 fail:
    return ret;
}

763 764
static int create_fixed_disk(BlockDriverState *bs, uint8_t *buf,
                             int64_t total_size)
765
{
766
    int ret;
767 768

    /* Add footer to total size */
769 770 771 772 773
    total_size += HEADER_SIZE;

    ret = bdrv_truncate(bs, total_size);
    if (ret < 0) {
        return ret;
774 775
    }

776 777 778 779
    ret = bdrv_pwrite_sync(bs, total_size - HEADER_SIZE, buf, HEADER_SIZE);
    if (ret < 0) {
        return ret;
    }
780 781 782 783

    return ret;
}

784
static int vpc_create(const char *filename, QemuOpts *opts, Error **errp)
785 786
{
    uint8_t buf[1024];
787
    VHDFooter *footer = (VHDFooter *) buf;
788
    char *disk_type_param;
789
    int i;
790 791 792 793 794 795 796
    uint16_t cyls = 0;
    uint8_t heads = 0;
    uint8_t secs_per_cyl = 0;
    int64_t total_sectors;
    int64_t total_size;
    int disk_type;
    int ret = -EIO;
797 798
    Error *local_err = NULL;
    BlockDriverState *bs = NULL;
799 800

    /* Read out options */
801 802
    total_size = ROUND_UP(qemu_opt_get_size_del(opts, BLOCK_OPT_SIZE, 0),
                          BDRV_SECTOR_SIZE);
803 804 805
    disk_type_param = qemu_opt_get_del(opts, BLOCK_OPT_SUBFMT);
    if (disk_type_param) {
        if (!strcmp(disk_type_param, "dynamic")) {
806
            disk_type = VHD_DYNAMIC;
807
        } else if (!strcmp(disk_type_param, "fixed")) {
808 809
            disk_type = VHD_FIXED;
        } else {
810 811
            ret = -EINVAL;
            goto out;
812 813 814 815 816
        }
    } else {
        disk_type = VHD_DYNAMIC;
    }

817 818 819
    ret = bdrv_create_file(filename, opts, &local_err);
    if (ret < 0) {
        error_propagate(errp, local_err);
820
        goto out;
821
    }
822 823 824 825 826
    ret = bdrv_open(&bs, filename, NULL, NULL, BDRV_O_RDWR | BDRV_O_PROTOCOL,
                    NULL, &local_err);
    if (ret < 0) {
        error_propagate(errp, local_err);
        goto out;
827 828
    }

829 830 831 832 833
    /*
     * Calculate matching total_size and geometry. Increase the number of
     * sectors requested until we get enough (or fail). This ensures that
     * qemu-img convert doesn't truncate images, but rather rounds up.
     */
834
    total_sectors = total_size / BDRV_SECTOR_SIZE;
835 836 837 838
    for (i = 0; total_sectors > (int64_t)cyls * heads * secs_per_cyl; i++) {
        if (calculate_geometry(total_sectors + i, &cyls, &heads,
                               &secs_per_cyl))
        {
839
            ret = -EFBIG;
840
            goto out;
841 842
        }
    }
843

844
    total_sectors = (int64_t) cyls * heads * secs_per_cyl;
845
    total_size = total_sectors * BDRV_SECTOR_SIZE;
846 847 848 849 850 851 852 853 854

    /* Prepare the Hard Disk Footer */
    memset(buf, 0, 1024);

    memcpy(footer->creator, "conectix", 8);
    /* TODO Check if "qemu" creator_app is ok for VPC */
    memcpy(footer->creator_app, "qemu", 4);
    memcpy(footer->creator_os, "Wi2k", 4);

855 856
    footer->features = cpu_to_be32(0x02);
    footer->version = cpu_to_be32(0x00010000);
857
    if (disk_type == VHD_DYNAMIC) {
858
        footer->data_offset = cpu_to_be64(HEADER_SIZE);
859
    } else {
860
        footer->data_offset = cpu_to_be64(0xFFFFFFFFFFFFFFFFULL);
861
    }
862
    footer->timestamp = cpu_to_be32(time(NULL) - VHD_TIMESTAMP_BASE);
863 864

    /* Version of Virtual PC 2007 */
865 866
    footer->major = cpu_to_be16(0x0005);
    footer->minor = cpu_to_be16(0x0003);
867 868
    footer->orig_size = cpu_to_be64(total_size);
    footer->size = cpu_to_be64(total_size);
869
    footer->cyls = cpu_to_be16(cyls);
870 871 872
    footer->heads = heads;
    footer->secs_per_cyl = secs_per_cyl;

873
    footer->type = cpu_to_be32(disk_type);
874

875 876 877
#if defined(CONFIG_UUID)
    uuid_generate(footer->uuid);
#endif
878

879
    footer->checksum = cpu_to_be32(vpc_checksum(buf, HEADER_SIZE));
880 881

    if (disk_type == VHD_DYNAMIC) {
882
        ret = create_dynamic_disk(bs, buf, total_sectors);
883
    } else {
884
        ret = create_fixed_disk(bs, buf, total_size);
885 886
    }

887
out:
888
    bdrv_unref(bs);
889
    g_free(disk_type_param);
B
Blue Swirl 已提交
890
    return ret;
A
aliguori 已提交
891 892
}

K
Kevin Wolf 已提交
893 894 895
static int vpc_has_zero_init(BlockDriverState *bs)
{
    BDRVVPCState *s = bs->opaque;
896
    VHDFooter *footer =  (VHDFooter *) s->footer_buf;
K
Kevin Wolf 已提交
897

898
    if (be32_to_cpu(footer->type) == VHD_FIXED) {
K
Kevin Wolf 已提交
899 900 901 902 903 904
        return bdrv_has_zero_init(bs->file);
    } else {
        return 1;
    }
}

905 906 907
static void vpc_close(BlockDriverState *bs)
{
    BDRVVPCState *s = bs->opaque;
908
    qemu_vfree(s->pagetable);
909
#ifdef CACHE
910
    g_free(s->pageentry_u8);
911
#endif
K
Kevin Wolf 已提交
912 913 914

    migrate_del_blocker(s->migration_blocker);
    error_free(s->migration_blocker);
915 916
}

917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934
static QemuOptsList vpc_create_opts = {
    .name = "vpc-create-opts",
    .head = QTAILQ_HEAD_INITIALIZER(vpc_create_opts.head),
    .desc = {
        {
            .name = BLOCK_OPT_SIZE,
            .type = QEMU_OPT_SIZE,
            .help = "Virtual disk size"
        },
        {
            .name = BLOCK_OPT_SUBFMT,
            .type = QEMU_OPT_STRING,
            .help =
                "Type of virtual hard disk format. Supported formats are "
                "{dynamic (default) | fixed} "
        },
        { /* end of list */ }
    }
935 936
};

937
static BlockDriver bdrv_vpc = {
K
Kevin Wolf 已提交
938 939
    .format_name    = "vpc",
    .instance_size  = sizeof(BDRVVPCState),
940

K
Kevin Wolf 已提交
941 942 943 944
    .bdrv_probe             = vpc_probe,
    .bdrv_open              = vpc_open,
    .bdrv_close             = vpc_close,
    .bdrv_reopen_prepare    = vpc_reopen_prepare,
C
Chunyan Liu 已提交
945
    .bdrv_create            = vpc_create,
946

947 948 949
    .bdrv_read                  = vpc_co_read,
    .bdrv_write                 = vpc_co_write,
    .bdrv_co_get_block_status   = vpc_co_get_block_status,
950

P
Paolo Bonzini 已提交
951 952
    .bdrv_get_info          = vpc_get_info,

953
    .create_opts            = &vpc_create_opts,
K
Kevin Wolf 已提交
954
    .bdrv_has_zero_init     = vpc_has_zero_init,
955
};
956 957 958 959 960 961 962

static void bdrv_vpc_init(void)
{
    bdrv_register(&bdrv_vpc);
}

block_init(bdrv_vpc_init);