block.c 113.5 KB
Newer Older
B
bellard 已提交
1 2
/*
 * QEMU System Emulator block driver
3
 *
B
bellard 已提交
4
 * Copyright (c) 2003 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.
 */
24
#include "config-host.h"
P
pbrook 已提交
25
#include "qemu-common.h"
26
#include "trace.h"
A
aliguori 已提交
27
#include "monitor.h"
B
bellard 已提交
28
#include "block_int.h"
29
#include "module.h"
30
#include "qjson.h"
31
#include "qemu-coroutine.h"
L
Luiz Capitulino 已提交
32
#include "qmp-commands.h"
33
#include "qemu-timer.h"
B
bellard 已提交
34

J
Juan Quintela 已提交
35
#ifdef CONFIG_BSD
B
bellard 已提交
36 37 38
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/ioctl.h>
B
Blue Swirl 已提交
39
#include <sys/queue.h>
40
#ifndef __DragonFly__
B
bellard 已提交
41 42
#include <sys/disk.h>
#endif
43
#endif
B
bellard 已提交
44

45 46 47 48
#ifdef _WIN32
#include <windows.h>
#endif

49 50
#define NOT_DONE 0x7fffffff /* used while emulated sync operation in progress */

51 52
typedef enum {
    BDRV_REQ_COPY_ON_READ = 0x1,
53
    BDRV_REQ_ZERO_WRITE   = 0x2,
54 55
} BdrvRequestFlags;

56
static void bdrv_dev_change_media_cb(BlockDriverState *bs, bool load);
57 58
static BlockDriverAIOCB *bdrv_aio_readv_em(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
59
        BlockDriverCompletionFunc *cb, void *opaque);
60 61
static BlockDriverAIOCB *bdrv_aio_writev_em(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
62
        BlockDriverCompletionFunc *cb, void *opaque);
63 64 65 66 67 68
static int coroutine_fn bdrv_co_readv_em(BlockDriverState *bs,
                                         int64_t sector_num, int nb_sectors,
                                         QEMUIOVector *iov);
static int coroutine_fn bdrv_co_writev_em(BlockDriverState *bs,
                                         int64_t sector_num, int nb_sectors,
                                         QEMUIOVector *iov);
69
static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,
70 71
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags);
72
static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs,
73 74
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags);
75 76 77 78 79 80
static BlockDriverAIOCB *bdrv_co_aio_rw_vector(BlockDriverState *bs,
                                               int64_t sector_num,
                                               QEMUIOVector *qiov,
                                               int nb_sectors,
                                               BlockDriverCompletionFunc *cb,
                                               void *opaque,
81
                                               bool is_write);
82
static void coroutine_fn bdrv_co_do_rw(void *opaque);
K
Kevin Wolf 已提交
83 84
static int coroutine_fn bdrv_co_do_write_zeroes(BlockDriverState *bs,
    int64_t sector_num, int nb_sectors);
B
bellard 已提交
85

Z
Zhi Yong Wu 已提交
86 87 88 89 90 91 92
static bool bdrv_exceed_bps_limits(BlockDriverState *bs, int nb_sectors,
        bool is_write, double elapsed_time, uint64_t *wait);
static bool bdrv_exceed_iops_limits(BlockDriverState *bs, bool is_write,
        double elapsed_time, uint64_t *wait);
static bool bdrv_exceed_io_limits(BlockDriverState *bs, int nb_sectors,
        bool is_write, int64_t *wait);

93 94
static QTAILQ_HEAD(, BlockDriverState) bdrv_states =
    QTAILQ_HEAD_INITIALIZER(bdrv_states);
95

96 97
static QLIST_HEAD(, BlockDriver) bdrv_drivers =
    QLIST_HEAD_INITIALIZER(bdrv_drivers);
B
bellard 已提交
98

99 100 101
/* The device to use for VM snapshots */
static BlockDriverState *bs_snapshots;

102 103 104
/* If non-zero, use only whitelisted block drivers */
static int use_bdrv_whitelist;

105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124
#ifdef _WIN32
static int is_windows_drive_prefix(const char *filename)
{
    return (((filename[0] >= 'a' && filename[0] <= 'z') ||
             (filename[0] >= 'A' && filename[0] <= 'Z')) &&
            filename[1] == ':');
}

int is_windows_drive(const char *filename)
{
    if (is_windows_drive_prefix(filename) &&
        filename[2] == '\0')
        return 1;
    if (strstart(filename, "\\\\.\\", NULL) ||
        strstart(filename, "//./", NULL))
        return 1;
    return 0;
}
#endif

125
/* throttling disk I/O limits */
Z
Zhi Yong Wu 已提交
126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143
void bdrv_io_limits_disable(BlockDriverState *bs)
{
    bs->io_limits_enabled = false;

    while (qemu_co_queue_next(&bs->throttled_reqs));

    if (bs->block_timer) {
        qemu_del_timer(bs->block_timer);
        qemu_free_timer(bs->block_timer);
        bs->block_timer = NULL;
    }

    bs->slice_start = 0;
    bs->slice_end   = 0;
    bs->slice_time  = 0;
    memset(&bs->io_base, 0, sizeof(bs->io_base));
}

144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172
static void bdrv_block_timer(void *opaque)
{
    BlockDriverState *bs = opaque;

    qemu_co_queue_next(&bs->throttled_reqs);
}

void bdrv_io_limits_enable(BlockDriverState *bs)
{
    qemu_co_queue_init(&bs->throttled_reqs);
    bs->block_timer = qemu_new_timer_ns(vm_clock, bdrv_block_timer, bs);
    bs->slice_time  = 5 * BLOCK_IO_SLICE_TIME;
    bs->slice_start = qemu_get_clock_ns(vm_clock);
    bs->slice_end   = bs->slice_start + bs->slice_time;
    memset(&bs->io_base, 0, sizeof(bs->io_base));
    bs->io_limits_enabled = true;
}

bool bdrv_io_limits_enabled(BlockDriverState *bs)
{
    BlockIOLimit *io_limits = &bs->io_limits;
    return io_limits->bps[BLOCK_IO_LIMIT_READ]
         || io_limits->bps[BLOCK_IO_LIMIT_WRITE]
         || io_limits->bps[BLOCK_IO_LIMIT_TOTAL]
         || io_limits->iops[BLOCK_IO_LIMIT_READ]
         || io_limits->iops[BLOCK_IO_LIMIT_WRITE]
         || io_limits->iops[BLOCK_IO_LIMIT_TOTAL];
}

Z
Zhi Yong Wu 已提交
173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197
static void bdrv_io_limits_intercept(BlockDriverState *bs,
                                     bool is_write, int nb_sectors)
{
    int64_t wait_time = -1;

    if (!qemu_co_queue_empty(&bs->throttled_reqs)) {
        qemu_co_queue_wait(&bs->throttled_reqs);
    }

    /* In fact, we hope to keep each request's timing, in FIFO mode. The next
     * throttled requests will not be dequeued until the current request is
     * allowed to be serviced. So if the current request still exceeds the
     * limits, it will be inserted to the head. All requests followed it will
     * be still in throttled_reqs queue.
     */

    while (bdrv_exceed_io_limits(bs, nb_sectors, is_write, &wait_time)) {
        qemu_mod_timer(bs->block_timer,
                       wait_time + qemu_get_clock_ns(vm_clock));
        qemu_co_queue_wait_insert_head(&bs->throttled_reqs);
    }

    qemu_co_queue_next(&bs->throttled_reqs);
}

198 199 200
/* check if the path starts with "<protocol>:" */
static int path_has_protocol(const char *path)
{
201 202
    const char *p;

203 204 205 206 207
#ifdef _WIN32
    if (is_windows_drive(path) ||
        is_windows_drive_prefix(path)) {
        return 0;
    }
208 209 210
    p = path + strcspn(path, ":/\\");
#else
    p = path + strcspn(path, ":/");
211 212
#endif

213
    return *p == ':';
214 215
}

B
bellard 已提交
216
int path_is_absolute(const char *path)
217
{
B
bellard 已提交
218 219
#ifdef _WIN32
    /* specific case for names like: "\\.\d:" */
P
Paolo Bonzini 已提交
220
    if (is_windows_drive(path) || is_windows_drive_prefix(path)) {
B
bellard 已提交
221
        return 1;
P
Paolo Bonzini 已提交
222 223
    }
    return (*path == '/' || *path == '\\');
224
#else
P
Paolo Bonzini 已提交
225
    return (*path == '/');
226
#endif
227 228
}

B
bellard 已提交
229 230 231 232 233 234
/* if filename is absolute, just copy it to dest. Otherwise, build a
   path to it by considering it is relative to base_path. URL are
   supported. */
void path_combine(char *dest, int dest_size,
                  const char *base_path,
                  const char *filename)
235
{
B
bellard 已提交
236 237 238 239 240 241 242 243 244 245 246 247 248
    const char *p, *p1;
    int len;

    if (dest_size <= 0)
        return;
    if (path_is_absolute(filename)) {
        pstrcpy(dest, dest_size, filename);
    } else {
        p = strchr(base_path, ':');
        if (p)
            p++;
        else
            p = base_path;
249 250 251 252 253 254 255 256 257
        p1 = strrchr(base_path, '/');
#ifdef _WIN32
        {
            const char *p2;
            p2 = strrchr(base_path, '\\');
            if (!p1 || p2 > p1)
                p1 = p2;
        }
#endif
B
bellard 已提交
258 259 260 261 262 263 264 265 266 267 268 269
        if (p1)
            p1++;
        else
            p1 = base_path;
        if (p1 > p)
            p = p1;
        len = p - base_path;
        if (len > dest_size - 1)
            len = dest_size - 1;
        memcpy(dest, base_path, len);
        dest[len] = '\0';
        pstrcat(dest, dest_size, filename);
270 271 272
    }
}

273 274 275 276 277 278 279 280 281
void bdrv_get_full_backing_filename(BlockDriverState *bs, char *dest, size_t sz)
{
    if (bs->backing_file[0] == '\0' || path_has_protocol(bs->backing_file)) {
        pstrcpy(dest, sz, bs->backing_file);
    } else {
        path_combine(dest, sz, bs->filename, bs->backing_file);
    }
}

282
void bdrv_register(BlockDriver *bdrv)
B
bellard 已提交
283
{
284 285
    /* Block drivers without coroutine functions need emulation */
    if (!bdrv->bdrv_co_readv) {
286 287 288
        bdrv->bdrv_co_readv = bdrv_co_readv_em;
        bdrv->bdrv_co_writev = bdrv_co_writev_em;

289 290 291
        /* bdrv_co_readv_em()/brdv_co_writev_em() work in terms of aio, so if
         * the block driver lacks aio we need to emulate that too.
         */
292 293 294 295 296
        if (!bdrv->bdrv_aio_readv) {
            /* add AIO emulation layer */
            bdrv->bdrv_aio_readv = bdrv_aio_readv_em;
            bdrv->bdrv_aio_writev = bdrv_aio_writev_em;
        }
B
bellard 已提交
297
    }
298

299
    QLIST_INSERT_HEAD(&bdrv_drivers, bdrv, list);
B
bellard 已提交
300
}
B
bellard 已提交
301 302 303 304

/* create a new block device (by default it is empty) */
BlockDriverState *bdrv_new(const char *device_name)
{
305
    BlockDriverState *bs;
B
bellard 已提交
306

307
    bs = g_malloc0(sizeof(BlockDriverState));
B
bellard 已提交
308
    pstrcpy(bs->device_name, sizeof(bs->device_name), device_name);
B
bellard 已提交
309
    if (device_name[0] != '\0') {
310
        QTAILQ_INSERT_TAIL(&bdrv_states, bs, list);
B
bellard 已提交
311
    }
312
    bdrv_iostatus_disable(bs);
B
bellard 已提交
313 314 315
    return bs;
}

B
bellard 已提交
316 317 318
BlockDriver *bdrv_find_format(const char *format_name)
{
    BlockDriver *drv1;
319 320
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
        if (!strcmp(drv1->format_name, format_name)) {
B
bellard 已提交
321
            return drv1;
322
        }
B
bellard 已提交
323 324 325 326
    }
    return NULL;
}

327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350
static int bdrv_is_whitelisted(BlockDriver *drv)
{
    static const char *whitelist[] = {
        CONFIG_BDRV_WHITELIST
    };
    const char **p;

    if (!whitelist[0])
        return 1;               /* no whitelist, anything goes */

    for (p = whitelist; *p; p++) {
        if (!strcmp(drv->format_name, *p)) {
            return 1;
        }
    }
    return 0;
}

BlockDriver *bdrv_find_whitelisted_format(const char *format_name)
{
    BlockDriver *drv = bdrv_find_format(format_name);
    return drv && bdrv_is_whitelisted(drv) ? drv : NULL;
}

351 352 353 354 355 356 357 358 359 360 361 362 363 364 365
typedef struct CreateCo {
    BlockDriver *drv;
    char *filename;
    QEMUOptionParameter *options;
    int ret;
} CreateCo;

static void coroutine_fn bdrv_create_co_entry(void *opaque)
{
    CreateCo *cco = opaque;
    assert(cco->drv);

    cco->ret = cco->drv->bdrv_create(cco->filename, cco->options);
}

366 367
int bdrv_create(BlockDriver *drv, const char* filename,
    QEMUOptionParameter *options)
B
bellard 已提交
368
{
369 370 371 372 373 374 375 376 377 378 379
    int ret;

    Coroutine *co;
    CreateCo cco = {
        .drv = drv,
        .filename = g_strdup(filename),
        .options = options,
        .ret = NOT_DONE,
    };

    if (!drv->bdrv_create) {
B
bellard 已提交
380
        return -ENOTSUP;
381 382 383 384 385 386 387 388 389 390 391 392 393 394 395
    }

    if (qemu_in_coroutine()) {
        /* Fast-path if already in coroutine context */
        bdrv_create_co_entry(&cco);
    } else {
        co = qemu_coroutine_create(bdrv_create_co_entry);
        qemu_coroutine_enter(co, &cco);
        while (cco.ret == NOT_DONE) {
            qemu_aio_wait();
        }
    }

    ret = cco.ret;
    g_free(cco.filename);
396

397
    return ret;
B
bellard 已提交
398 399
}

400 401 402 403
int bdrv_create_file(const char* filename, QEMUOptionParameter *options)
{
    BlockDriver *drv;

404
    drv = bdrv_find_protocol(filename);
405
    if (drv == NULL) {
406
        return -ENOENT;
407 408 409 410 411
    }

    return bdrv_create(drv, filename, options);
}

412 413 414 415 416
/*
 * Create a uniquely-named empty temporary file.
 * Return 0 upon success, otherwise a negative errno value.
 */
int get_tmp_filename(char *filename, int size)
B
bellard 已提交
417
{
418
#ifdef _WIN32
419
    char temp_dir[MAX_PATH];
420 421 422 423 424 425
    /* GetTempFileName requires that its output buffer (4th param)
       have length MAX_PATH or greater.  */
    assert(size >= MAX_PATH);
    return (GetTempPath(MAX_PATH, temp_dir)
            && GetTempFileName(temp_dir, "qem", 0, filename)
            ? 0 : -GetLastError());
B
bellard 已提交
426
#else
B
bellard 已提交
427
    int fd;
428
    const char *tmpdir;
A
aurel32 已提交
429 430 431
    tmpdir = getenv("TMPDIR");
    if (!tmpdir)
        tmpdir = "/tmp";
432 433 434
    if (snprintf(filename, size, "%s/vl.XXXXXX", tmpdir) >= size) {
        return -EOVERFLOW;
    }
B
bellard 已提交
435
    fd = mkstemp(filename);
436 437 438 439
    if (fd < 0 || close(fd)) {
        return -errno;
    }
    return 0;
B
bellard 已提交
440
#endif
441
}
B
bellard 已提交
442

443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464
/*
 * Detect host devices. By convention, /dev/cdrom[N] is always
 * recognized as a host CDROM.
 */
static BlockDriver *find_hdev_driver(const char *filename)
{
    int score_max = 0, score;
    BlockDriver *drv = NULL, *d;

    QLIST_FOREACH(d, &bdrv_drivers, list) {
        if (d->bdrv_probe_device) {
            score = d->bdrv_probe_device(filename);
            if (score > score_max) {
                score_max = score;
                drv = d;
            }
        }
    }

    return drv;
}

465
BlockDriver *bdrv_find_protocol(const char *filename)
B
bellard 已提交
466 467 468
{
    BlockDriver *drv1;
    char protocol[128];
469
    int len;
B
bellard 已提交
470
    const char *p;
B
bellard 已提交
471

472 473
    /* TODO Drivers without bdrv_file_open must be specified explicitly */

474 475 476 477 478 479 480 481 482 483 484 485
    /*
     * XXX(hch): we really should not let host device detection
     * override an explicit protocol specification, but moving this
     * later breaks access to device names with colons in them.
     * Thanks to the brain-dead persistent naming schemes on udev-
     * based Linux systems those actually are quite common.
     */
    drv1 = find_hdev_driver(filename);
    if (drv1) {
        return drv1;
    }

486
    if (!path_has_protocol(filename)) {
487
        return bdrv_find_format("file");
488
    }
489 490
    p = strchr(filename, ':');
    assert(p != NULL);
491 492 493 494 495
    len = p - filename;
    if (len > sizeof(protocol) - 1)
        len = sizeof(protocol) - 1;
    memcpy(protocol, filename, len);
    protocol[len] = '\0';
496
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
497
        if (drv1->protocol_name &&
498
            !strcmp(drv1->protocol_name, protocol)) {
B
bellard 已提交
499
            return drv1;
500
        }
B
bellard 已提交
501 502 503 504
    }
    return NULL;
}

505
static int find_image_format(const char *filename, BlockDriver **pdrv)
506 507 508 509 510 511
{
    int ret, score, score_max;
    BlockDriver *drv1, *drv;
    uint8_t buf[2048];
    BlockDriverState *bs;

512
    ret = bdrv_file_open(&bs, filename, 0);
513 514 515 516
    if (ret < 0) {
        *pdrv = NULL;
        return ret;
    }
517

518 519
    /* Return the raw BlockDriver * to scsi-generic devices or empty drives */
    if (bs->sg || !bdrv_is_inserted(bs)) {
520
        bdrv_delete(bs);
521 522 523 524 525 526
        drv = bdrv_find_format("raw");
        if (!drv) {
            ret = -ENOENT;
        }
        *pdrv = drv;
        return ret;
527
    }
528

B
bellard 已提交
529 530 531
    ret = bdrv_pread(bs, 0, buf, sizeof(buf));
    bdrv_delete(bs);
    if (ret < 0) {
532 533
        *pdrv = NULL;
        return ret;
B
bellard 已提交
534 535
    }

B
bellard 已提交
536
    score_max = 0;
537
    drv = NULL;
538
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
B
bellard 已提交
539 540 541 542 543 544
        if (drv1->bdrv_probe) {
            score = drv1->bdrv_probe(buf, ret, filename);
            if (score > score_max) {
                score_max = score;
                drv = drv1;
            }
B
bellard 已提交
545
        }
B
bellard 已提交
546
    }
547 548 549 550 551
    if (!drv) {
        ret = -ENOENT;
    }
    *pdrv = drv;
    return ret;
B
bellard 已提交
552 553
}

554 555 556 557 558 559 560
/**
 * Set the current 'total_sectors' value
 */
static int refresh_total_sectors(BlockDriverState *bs, int64_t hint)
{
    BlockDriver *drv = bs->drv;

561 562 563 564
    /* Do not attempt drv->bdrv_getlength() on scsi-generic devices */
    if (bs->sg)
        return 0;

565 566 567 568 569 570 571 572 573 574 575 576 577
    /* query actual device if possible, otherwise just trust the hint */
    if (drv->bdrv_getlength) {
        int64_t length = drv->bdrv_getlength(bs);
        if (length < 0) {
            return length;
        }
        hint = length >> BDRV_SECTOR_BITS;
    }

    bs->total_sectors = hint;
    return 0;
}

578 579 580 581 582 583 584 585 586 587 588
/**
 * Set open flags for a given cache mode
 *
 * Return 0 on success, -1 if the cache mode was invalid.
 */
int bdrv_parse_cache_flags(const char *mode, int *flags)
{
    *flags &= ~BDRV_O_CACHE_MASK;

    if (!strcmp(mode, "off") || !strcmp(mode, "none")) {
        *flags |= BDRV_O_NOCACHE | BDRV_O_CACHE_WB;
589 590
    } else if (!strcmp(mode, "directsync")) {
        *flags |= BDRV_O_NOCACHE;
591 592 593 594 595 596 597 598 599 600 601 602 603 604
    } else if (!strcmp(mode, "writeback")) {
        *flags |= BDRV_O_CACHE_WB;
    } else if (!strcmp(mode, "unsafe")) {
        *flags |= BDRV_O_CACHE_WB;
        *flags |= BDRV_O_NO_FLUSH;
    } else if (!strcmp(mode, "writethrough")) {
        /* this is the default */
    } else {
        return -1;
    }

    return 0;
}

605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620
/**
 * The copy-on-read flag is actually a reference count so multiple users may
 * use the feature without worrying about clobbering its previous state.
 * Copy-on-read stays enabled until all users have called to disable it.
 */
void bdrv_enable_copy_on_read(BlockDriverState *bs)
{
    bs->copy_on_read++;
}

void bdrv_disable_copy_on_read(BlockDriverState *bs)
{
    assert(bs->copy_on_read > 0);
    bs->copy_on_read--;
}

621 622 623 624 625 626 627 628 629
/*
 * Common part for opening disk images and files
 */
static int bdrv_open_common(BlockDriverState *bs, const char *filename,
    int flags, BlockDriver *drv)
{
    int ret, open_flags;

    assert(drv != NULL);
630
    assert(bs->file == NULL);
631

S
Stefan Hajnoczi 已提交
632 633
    trace_bdrv_open_common(bs, filename, flags, drv->format_name);

634 635 636
    bs->open_flags = flags;
    bs->buffer_alignment = 512;

637 638 639 640 641
    assert(bs->copy_on_read == 0); /* bdrv_new() and bdrv_close() make it so */
    if ((flags & BDRV_O_RDWR) && (flags & BDRV_O_COPY_ON_READ)) {
        bdrv_enable_copy_on_read(bs);
    }

642 643 644 645 646 647 648
    pstrcpy(bs->filename, sizeof(bs->filename), filename);

    if (use_bdrv_whitelist && !bdrv_is_whitelisted(drv)) {
        return -ENOTSUP;
    }

    bs->drv = drv;
649
    bs->opaque = g_malloc0(drv->instance_size);
650

651
    bs->enable_write_cache = !!(flags & BDRV_O_CACHE_WB);
652
    open_flags = flags | BDRV_O_CACHE_WB;
653 654 655 656 657

    /*
     * Clear flags that are internal to the block layer before opening the
     * image.
     */
658
    open_flags &= ~(BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING);
659 660

    /*
S
Stefan Weil 已提交
661
     * Snapshots should be writable.
662 663 664 665 666
     */
    if (bs->is_temporary) {
        open_flags |= BDRV_O_RDWR;
    }

667 668
    bs->keep_read_only = bs->read_only = !(open_flags & BDRV_O_RDWR);

669 670 671 672 673 674 675 676 677 678
    /* Open the image, either directly or using a protocol */
    if (drv->bdrv_file_open) {
        ret = drv->bdrv_file_open(bs, filename, open_flags);
    } else {
        ret = bdrv_file_open(&bs->file, filename, open_flags);
        if (ret >= 0) {
            ret = drv->bdrv_open(bs, open_flags);
        }
    }

679 680 681 682
    if (ret < 0) {
        goto free_and_fail;
    }

683 684 685
    ret = refresh_total_sectors(bs, bs->total_sectors);
    if (ret < 0) {
        goto free_and_fail;
686
    }
687

688 689 690 691 692 693 694 695
#ifndef _WIN32
    if (bs->is_temporary) {
        unlink(filename);
    }
#endif
    return 0;

free_and_fail:
696 697 698 699
    if (bs->file) {
        bdrv_delete(bs->file);
        bs->file = NULL;
    }
700
    g_free(bs->opaque);
701 702 703 704 705
    bs->opaque = NULL;
    bs->drv = NULL;
    return ret;
}

K
Kevin Wolf 已提交
706 707 708
/*
 * Opens a file using a protocol (file, host_device, nbd, ...)
 */
B
bellard 已提交
709
int bdrv_file_open(BlockDriverState **pbs, const char *filename, int flags)
B
bellard 已提交
710
{
B
bellard 已提交
711
    BlockDriverState *bs;
712
    BlockDriver *drv;
B
bellard 已提交
713 714
    int ret;

715
    drv = bdrv_find_protocol(filename);
716 717 718 719
    if (!drv) {
        return -ENOENT;
    }

B
bellard 已提交
720
    bs = bdrv_new("");
K
Kevin Wolf 已提交
721
    ret = bdrv_open_common(bs, filename, flags, drv);
B
bellard 已提交
722 723 724
    if (ret < 0) {
        bdrv_delete(bs);
        return ret;
725
    }
726
    bs->growable = 1;
B
bellard 已提交
727 728 729 730
    *pbs = bs;
    return 0;
}

K
Kevin Wolf 已提交
731 732 733
/*
 * Opens a disk image (raw, qcow2, vmdk, ...)
 */
K
Kevin Wolf 已提交
734 735
int bdrv_open(BlockDriverState *bs, const char *filename, int flags,
              BlockDriver *drv)
B
bellard 已提交
736
{
K
Kevin Wolf 已提交
737
    int ret;
K
Kevin Wolf 已提交
738
    char tmp_filename[PATH_MAX];
B
bellard 已提交
739

B
bellard 已提交
740
    if (flags & BDRV_O_SNAPSHOT) {
B
bellard 已提交
741 742
        BlockDriverState *bs1;
        int64_t total_size;
A
aliguori 已提交
743
        int is_protocol = 0;
K
Kevin Wolf 已提交
744 745
        BlockDriver *bdrv_qcow2;
        QEMUOptionParameter *options;
K
Kevin Wolf 已提交
746
        char backing_filename[PATH_MAX];
747

B
bellard 已提交
748 749
        /* if snapshot, we create a temporary backing file and open it
           instead of opening 'filename' directly */
750

B
bellard 已提交
751 752
        /* if there is a backing file, use it */
        bs1 = bdrv_new("");
K
Kevin Wolf 已提交
753
        ret = bdrv_open(bs1, filename, 0, drv);
754
        if (ret < 0) {
B
bellard 已提交
755
            bdrv_delete(bs1);
756
            return ret;
B
bellard 已提交
757
        }
758
        total_size = bdrv_getlength(bs1) & BDRV_SECTOR_MASK;
A
aliguori 已提交
759 760 761 762

        if (bs1->drv && bs1->drv->protocol_name)
            is_protocol = 1;

B
bellard 已提交
763
        bdrv_delete(bs1);
764

765 766 767 768
        ret = get_tmp_filename(tmp_filename, sizeof(tmp_filename));
        if (ret < 0) {
            return ret;
        }
A
aliguori 已提交
769 770 771 772 773

        /* Real path is meaningless for protocols */
        if (is_protocol)
            snprintf(backing_filename, sizeof(backing_filename),
                     "%s", filename);
774 775
        else if (!realpath(filename, backing_filename))
            return -errno;
A
aliguori 已提交
776

K
Kevin Wolf 已提交
777 778 779
        bdrv_qcow2 = bdrv_find_format("qcow2");
        options = parse_option_parameters("", bdrv_qcow2->create_options, NULL);

780
        set_option_parameter_int(options, BLOCK_OPT_SIZE, total_size);
K
Kevin Wolf 已提交
781 782 783 784 785 786 787
        set_option_parameter(options, BLOCK_OPT_BACKING_FILE, backing_filename);
        if (drv) {
            set_option_parameter(options, BLOCK_OPT_BACKING_FMT,
                drv->format_name);
        }

        ret = bdrv_create(bdrv_qcow2, tmp_filename, options);
788
        free_option_parameters(options);
789 790
        if (ret < 0) {
            return ret;
B
bellard 已提交
791
        }
K
Kevin Wolf 已提交
792

B
bellard 已提交
793
        filename = tmp_filename;
K
Kevin Wolf 已提交
794
        drv = bdrv_qcow2;
B
bellard 已提交
795 796
        bs->is_temporary = 1;
    }
B
bellard 已提交
797

K
Kevin Wolf 已提交
798
    /* Find the right image format driver */
799
    if (!drv) {
800
        ret = find_image_format(filename, &drv);
801
    }
802

803 804
    if (!drv) {
        goto unlink_and_fail;
B
bellard 已提交
805
    }
K
Kevin Wolf 已提交
806 807 808 809

    /* Open the image */
    ret = bdrv_open_common(bs, filename, flags, drv);
    if (ret < 0) {
810 811 812
        goto unlink_and_fail;
    }

K
Kevin Wolf 已提交
813 814 815 816 817 818 819
    /* If there is a backing file, use it */
    if ((flags & BDRV_O_NO_BACKING) == 0 && bs->backing_file[0] != '\0') {
        char backing_filename[PATH_MAX];
        int back_flags;
        BlockDriver *back_drv = NULL;

        bs->backing_hd = bdrv_new("");
820 821
        bdrv_get_full_backing_filename(bs, backing_filename,
                                       sizeof(backing_filename));
822 823

        if (bs->backing_format[0] != '\0') {
K
Kevin Wolf 已提交
824
            back_drv = bdrv_find_format(bs->backing_format);
825
        }
K
Kevin Wolf 已提交
826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844

        /* backing files always opened read-only */
        back_flags =
            flags & ~(BDRV_O_RDWR | BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING);

        ret = bdrv_open(bs->backing_hd, backing_filename, back_flags, back_drv);
        if (ret < 0) {
            bdrv_close(bs);
            return ret;
        }
        if (bs->is_temporary) {
            bs->backing_hd->keep_read_only = !(flags & BDRV_O_RDWR);
        } else {
            /* base image inherits from "parent" */
            bs->backing_hd->keep_read_only = bs->keep_read_only;
        }
    }

    if (!bdrv_key_required(bs)) {
845
        bdrv_dev_change_media_cb(bs, true);
K
Kevin Wolf 已提交
846 847
    }

Z
Zhi Yong Wu 已提交
848 849 850 851 852
    /* throttling disk I/O limits */
    if (bs->io_limits_enabled) {
        bdrv_io_limits_enable(bs);
    }

K
Kevin Wolf 已提交
853 854 855 856 857 858 859 860 861
    return 0;

unlink_and_fail:
    if (bs->is_temporary) {
        unlink(filename);
    }
    return ret;
}

B
bellard 已提交
862 863
void bdrv_close(BlockDriverState *bs)
{
864
    bdrv_flush(bs);
B
bellard 已提交
865
    if (bs->drv) {
866 867 868
        if (bs->job) {
            block_job_cancel_sync(bs->job);
        }
K
Kevin Wolf 已提交
869 870
        bdrv_drain_all();

871 872 873
        if (bs == bs_snapshots) {
            bs_snapshots = NULL;
        }
874
        if (bs->backing_hd) {
B
bellard 已提交
875
            bdrv_delete(bs->backing_hd);
876 877
            bs->backing_hd = NULL;
        }
B
bellard 已提交
878
        bs->drv->bdrv_close(bs);
879
        g_free(bs->opaque);
B
bellard 已提交
880 881 882 883
#ifdef _WIN32
        if (bs->is_temporary) {
            unlink(bs->filename);
        }
B
bellard 已提交
884
#endif
B
bellard 已提交
885 886
        bs->opaque = NULL;
        bs->drv = NULL;
887
        bs->copy_on_read = 0;
888 889
        bs->backing_file[0] = '\0';
        bs->backing_format[0] = '\0';
890 891 892 893 894
        bs->total_sectors = 0;
        bs->encrypted = 0;
        bs->valid_key = 0;
        bs->sg = 0;
        bs->growable = 0;
B
bellard 已提交
895

896
        if (bs->file != NULL) {
897 898
            bdrv_delete(bs->file);
            bs->file = NULL;
899 900
        }

901
        bdrv_dev_change_media_cb(bs, false);
B
bellard 已提交
902
    }
Z
Zhi Yong Wu 已提交
903 904 905 906 907

    /*throttling disk I/O limits*/
    if (bs->io_limits_enabled) {
        bdrv_io_limits_disable(bs);
    }
B
bellard 已提交
908 909
}

910 911 912 913 914 915 916 917 918
void bdrv_close_all(void)
{
    BlockDriverState *bs;

    QTAILQ_FOREACH(bs, &bdrv_states, list) {
        bdrv_close(bs);
    }
}

919 920 921 922 923
/*
 * Wait for pending requests to complete across all BlockDriverStates
 *
 * This function does not flush data to disk, use bdrv_flush_all() for that
 * after calling this function.
924 925 926 927 928 929
 *
 * Note that completion of an asynchronous I/O operation can trigger any
 * number of other I/O operations on other devices---for example a coroutine
 * can be arbitrarily complex and a constant flow of I/O can come until the
 * coroutine is complete.  Because of this, it is not possible to have a
 * function to drain a single device's I/O queue.
930 931 932 933
 */
void bdrv_drain_all(void)
{
    BlockDriverState *bs;
934 935 936 937
    bool busy;

    do {
        busy = qemu_aio_wait();
938

939 940 941 942 943 944 945 946 947 948
        /* FIXME: We do not have timer support here, so this is effectively
         * a busy wait.
         */
        QTAILQ_FOREACH(bs, &bdrv_states, list) {
            if (!qemu_co_queue_empty(&bs->throttled_reqs)) {
                qemu_co_queue_restart_all(&bs->throttled_reqs);
                busy = true;
            }
        }
    } while (busy);
949 950 951 952 953 954 955 956

    /* If requests are still pending there is a bug somewhere */
    QTAILQ_FOREACH(bs, &bdrv_states, list) {
        assert(QLIST_EMPTY(&bs->tracked_requests));
        assert(qemu_co_queue_empty(&bs->throttled_reqs));
    }
}

957 958 959 960 961 962 963 964 965 966
/* make a BlockDriverState anonymous by removing from bdrv_state list.
   Also, NULL terminate the device_name to prevent double remove */
void bdrv_make_anon(BlockDriverState *bs)
{
    if (bs->device_name[0] != '\0') {
        QTAILQ_REMOVE(&bdrv_states, bs, list);
    }
    bs->device_name[0] = '\0';
}

P
Paolo Bonzini 已提交
967 968 969 970 971 972 973
static void bdrv_rebind(BlockDriverState *bs)
{
    if (bs->drv && bs->drv->bdrv_rebind) {
        bs->drv->bdrv_rebind(bs);
    }
}

974 975
static void bdrv_move_feature_fields(BlockDriverState *bs_dest,
                                     BlockDriverState *bs_src)
976
{
977 978
    /* move some fields that need to stay attached to the device */
    bs_dest->open_flags         = bs_src->open_flags;
979 980

    /* dev info */
981 982 983 984 985
    bs_dest->dev_ops            = bs_src->dev_ops;
    bs_dest->dev_opaque         = bs_src->dev_opaque;
    bs_dest->dev                = bs_src->dev;
    bs_dest->buffer_alignment   = bs_src->buffer_alignment;
    bs_dest->copy_on_read       = bs_src->copy_on_read;
986

987
    bs_dest->enable_write_cache = bs_src->enable_write_cache;
988

989
    /* i/o timing parameters */
990 991 992 993 994 995 996 997
    bs_dest->slice_time         = bs_src->slice_time;
    bs_dest->slice_start        = bs_src->slice_start;
    bs_dest->slice_end          = bs_src->slice_end;
    bs_dest->io_limits          = bs_src->io_limits;
    bs_dest->io_base            = bs_src->io_base;
    bs_dest->throttled_reqs     = bs_src->throttled_reqs;
    bs_dest->block_timer        = bs_src->block_timer;
    bs_dest->io_limits_enabled  = bs_src->io_limits_enabled;
998 999

    /* r/w error */
1000 1001
    bs_dest->on_read_error      = bs_src->on_read_error;
    bs_dest->on_write_error     = bs_src->on_write_error;
1002 1003

    /* i/o status */
1004 1005
    bs_dest->iostatus_enabled   = bs_src->iostatus_enabled;
    bs_dest->iostatus           = bs_src->iostatus;
1006

1007
    /* dirty bitmap */
1008 1009
    bs_dest->dirty_count        = bs_src->dirty_count;
    bs_dest->dirty_bitmap       = bs_src->dirty_bitmap;
1010 1011

    /* job */
1012 1013
    bs_dest->in_use             = bs_src->in_use;
    bs_dest->job                = bs_src->job;
1014

1015
    /* keep the same entry in bdrv_states */
1016 1017 1018 1019
    pstrcpy(bs_dest->device_name, sizeof(bs_dest->device_name),
            bs_src->device_name);
    bs_dest->list = bs_src->list;
}
1020

1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035
/*
 * Swap bs contents for two image chains while they are live,
 * while keeping required fields on the BlockDriverState that is
 * actually attached to a device.
 *
 * This will modify the BlockDriverState fields, and swap contents
 * between bs_new and bs_old. Both bs_new and bs_old are modified.
 *
 * bs_new is required to be anonymous.
 *
 * This function does not create any image files.
 */
void bdrv_swap(BlockDriverState *bs_new, BlockDriverState *bs_old)
{
    BlockDriverState tmp;
J
Jeff Cody 已提交
1036

1037 1038 1039 1040 1041 1042 1043 1044
    /* bs_new must be anonymous and shouldn't have anything fancy enabled */
    assert(bs_new->device_name[0] == '\0');
    assert(bs_new->dirty_bitmap == NULL);
    assert(bs_new->job == NULL);
    assert(bs_new->dev == NULL);
    assert(bs_new->in_use == 0);
    assert(bs_new->io_limits_enabled == false);
    assert(bs_new->block_timer == NULL);
1045

1046 1047 1048
    tmp = *bs_new;
    *bs_new = *bs_old;
    *bs_old = tmp;
1049

1050 1051 1052 1053
    /* there are some fields that should not be swapped, move them back */
    bdrv_move_feature_fields(&tmp, bs_old);
    bdrv_move_feature_fields(bs_old, bs_new);
    bdrv_move_feature_fields(bs_new, &tmp);
1054

1055 1056 1057 1058 1059 1060 1061 1062 1063
    /* bs_new shouldn't be in bdrv_states even after the swap!  */
    assert(bs_new->device_name[0] == '\0');

    /* Check a few fields that should remain attached to the device */
    assert(bs_new->dev == NULL);
    assert(bs_new->job == NULL);
    assert(bs_new->in_use == 0);
    assert(bs_new->io_limits_enabled == false);
    assert(bs_new->block_timer == NULL);
P
Paolo Bonzini 已提交
1064 1065

    bdrv_rebind(bs_new);
1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091
    bdrv_rebind(bs_old);
}

/*
 * Add new bs contents at the top of an image chain while the chain is
 * live, while keeping required fields on the top layer.
 *
 * This will modify the BlockDriverState fields, and swap contents
 * between bs_new and bs_top. Both bs_new and bs_top are modified.
 *
 * bs_new is required to be anonymous.
 *
 * This function does not create any image files.
 */
void bdrv_append(BlockDriverState *bs_new, BlockDriverState *bs_top)
{
    bdrv_swap(bs_new, bs_top);

    /* The contents of 'tmp' will become bs_top, as we are
     * swapping bs_new and bs_top contents. */
    bs_top->backing_hd = bs_new;
    bs_top->open_flags &= ~BDRV_O_NO_BACKING;
    pstrcpy(bs_top->backing_file, sizeof(bs_top->backing_file),
            bs_new->filename);
    pstrcpy(bs_top->backing_format, sizeof(bs_top->backing_format),
            bs_new->drv ? bs_new->drv->format_name : "");
1092 1093
}

B
bellard 已提交
1094 1095
void bdrv_delete(BlockDriverState *bs)
{
1096
    assert(!bs->dev);
1097 1098
    assert(!bs->job);
    assert(!bs->in_use);
1099

1100
    /* remove from list, if necessary */
1101
    bdrv_make_anon(bs);
1102

B
bellard 已提交
1103
    bdrv_close(bs);
1104

1105
    assert(bs != bs_snapshots);
1106
    g_free(bs);
B
bellard 已提交
1107 1108
}

1109 1110
int bdrv_attach_dev(BlockDriverState *bs, void *dev)
/* TODO change to DeviceState *dev when all users are qdevified */
1111
{
1112
    if (bs->dev) {
1113 1114
        return -EBUSY;
    }
1115
    bs->dev = dev;
1116
    bdrv_iostatus_reset(bs);
1117 1118 1119
    return 0;
}

1120 1121
/* TODO qdevified devices don't use this, remove when devices are qdevified */
void bdrv_attach_dev_nofail(BlockDriverState *bs, void *dev)
1122
{
1123 1124 1125 1126 1127 1128 1129 1130 1131 1132
    if (bdrv_attach_dev(bs, dev) < 0) {
        abort();
    }
}

void bdrv_detach_dev(BlockDriverState *bs, void *dev)
/* TODO change to DeviceState *dev when all users are qdevified */
{
    assert(bs->dev == dev);
    bs->dev = NULL;
1133 1134
    bs->dev_ops = NULL;
    bs->dev_opaque = NULL;
1135
    bs->buffer_alignment = 512;
1136 1137
}

1138 1139
/* TODO change to return DeviceState * when all users are qdevified */
void *bdrv_get_attached_dev(BlockDriverState *bs)
1140
{
1141
    return bs->dev;
1142 1143
}

1144 1145 1146 1147 1148
void bdrv_set_dev_ops(BlockDriverState *bs, const BlockDevOps *ops,
                      void *opaque)
{
    bs->dev_ops = ops;
    bs->dev_opaque = opaque;
1149 1150 1151
    if (bdrv_dev_has_removable_media(bs) && bs == bs_snapshots) {
        bs_snapshots = NULL;
    }
1152 1153
}

1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182
void bdrv_emit_qmp_error_event(const BlockDriverState *bdrv,
                               BlockQMPEventAction action, int is_read)
{
    QObject *data;
    const char *action_str;

    switch (action) {
    case BDRV_ACTION_REPORT:
        action_str = "report";
        break;
    case BDRV_ACTION_IGNORE:
        action_str = "ignore";
        break;
    case BDRV_ACTION_STOP:
        action_str = "stop";
        break;
    default:
        abort();
    }

    data = qobject_from_jsonf("{ 'device': %s, 'action': %s, 'operation': %s }",
                              bdrv->device_name,
                              action_str,
                              is_read ? "read" : "write");
    monitor_protocol_event(QEVENT_BLOCK_IO_ERROR, data);

    qobject_decref(data);
}

1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193
static void bdrv_emit_qmp_eject_event(BlockDriverState *bs, bool ejected)
{
    QObject *data;

    data = qobject_from_jsonf("{ 'device': %s, 'tray-open': %i }",
                              bdrv_get_device_name(bs), ejected);
    monitor_protocol_event(QEVENT_DEVICE_TRAY_MOVED, data);

    qobject_decref(data);
}

1194
static void bdrv_dev_change_media_cb(BlockDriverState *bs, bool load)
1195
{
1196
    if (bs->dev_ops && bs->dev_ops->change_media_cb) {
1197
        bool tray_was_closed = !bdrv_dev_is_tray_open(bs);
1198
        bs->dev_ops->change_media_cb(bs->dev_opaque, load);
1199 1200 1201 1202 1203 1204 1205 1206
        if (tray_was_closed) {
            /* tray open */
            bdrv_emit_qmp_eject_event(bs, true);
        }
        if (load) {
            /* tray close */
            bdrv_emit_qmp_eject_event(bs, false);
        }
1207 1208 1209
    }
}

1210 1211 1212 1213 1214
bool bdrv_dev_has_removable_media(BlockDriverState *bs)
{
    return !bs->dev || (bs->dev_ops && bs->dev_ops->change_media_cb);
}

P
Paolo Bonzini 已提交
1215 1216 1217 1218 1219 1220 1221
void bdrv_dev_eject_request(BlockDriverState *bs, bool force)
{
    if (bs->dev_ops && bs->dev_ops->eject_request_cb) {
        bs->dev_ops->eject_request_cb(bs->dev_opaque, force);
    }
}

1222 1223 1224 1225 1226 1227 1228 1229
bool bdrv_dev_is_tray_open(BlockDriverState *bs)
{
    if (bs->dev_ops && bs->dev_ops->is_tray_open) {
        return bs->dev_ops->is_tray_open(bs->dev_opaque);
    }
    return false;
}

1230 1231 1232 1233
static void bdrv_dev_resize_cb(BlockDriverState *bs)
{
    if (bs->dev_ops && bs->dev_ops->resize_cb) {
        bs->dev_ops->resize_cb(bs->dev_opaque);
1234 1235 1236
    }
}

1237 1238 1239 1240 1241 1242 1243 1244
bool bdrv_dev_is_medium_locked(BlockDriverState *bs)
{
    if (bs->dev_ops && bs->dev_ops->is_medium_locked) {
        return bs->dev_ops->is_medium_locked(bs->dev_opaque);
    }
    return false;
}

A
aliguori 已提交
1245 1246 1247
/*
 * Run consistency checks on an image
 *
1248
 * Returns 0 if the check could be completed (it doesn't mean that the image is
1249
 * free of errors) or -errno when an internal error occurred. The results of the
1250
 * check are stored in res.
A
aliguori 已提交
1251
 */
1252
int bdrv_check(BlockDriverState *bs, BdrvCheckResult *res, BdrvCheckMode fix)
A
aliguori 已提交
1253 1254 1255 1256 1257
{
    if (bs->drv->bdrv_check == NULL) {
        return -ENOTSUP;
    }

1258
    memset(res, 0, sizeof(*res));
1259
    return bs->drv->bdrv_check(bs, res, fix);
A
aliguori 已提交
1260 1261
}

1262 1263
#define COMMIT_BUF_SECTORS 2048

1264 1265 1266
/* commit COW file into the raw image */
int bdrv_commit(BlockDriverState *bs)
{
B
bellard 已提交
1267
    BlockDriver *drv = bs->drv;
1268
    BlockDriver *backing_drv;
1269 1270
    int64_t sector, total_sectors;
    int n, ro, open_flags;
1271
    int ret = 0, rw_ret = 0;
1272
    uint8_t *buf;
1273 1274
    char filename[1024];
    BlockDriverState *bs_rw, *bs_ro;
1275

B
bellard 已提交
1276 1277
    if (!drv)
        return -ENOMEDIUM;
1278 1279 1280
    
    if (!bs->backing_hd) {
        return -ENOTSUP;
1281 1282
    }

1283 1284 1285
    if (bs->backing_hd->keep_read_only) {
        return -EACCES;
    }
1286

1287 1288 1289 1290
    if (bdrv_in_use(bs) || bdrv_in_use(bs->backing_hd)) {
        return -EBUSY;
    }

1291
    backing_drv = bs->backing_hd->drv;
1292 1293 1294 1295 1296 1297 1298 1299 1300
    ro = bs->backing_hd->read_only;
    strncpy(filename, bs->backing_hd->filename, sizeof(filename));
    open_flags =  bs->backing_hd->open_flags;

    if (ro) {
        /* re-open as RW */
        bdrv_delete(bs->backing_hd);
        bs->backing_hd = NULL;
        bs_rw = bdrv_new("");
1301 1302
        rw_ret = bdrv_open(bs_rw, filename, open_flags | BDRV_O_RDWR,
            backing_drv);
1303 1304 1305 1306
        if (rw_ret < 0) {
            bdrv_delete(bs_rw);
            /* try to re-open read-only */
            bs_ro = bdrv_new("");
1307 1308
            ret = bdrv_open(bs_ro, filename, open_flags & ~BDRV_O_RDWR,
                backing_drv);
1309 1310 1311 1312 1313 1314 1315 1316 1317 1318
            if (ret < 0) {
                bdrv_delete(bs_ro);
                /* drive not functional anymore */
                bs->drv = NULL;
                return ret;
            }
            bs->backing_hd = bs_ro;
            return rw_ret;
        }
        bs->backing_hd = bs_rw;
B
bellard 已提交
1319
    }
1320

J
Jan Kiszka 已提交
1321
    total_sectors = bdrv_getlength(bs) >> BDRV_SECTOR_BITS;
1322
    buf = g_malloc(COMMIT_BUF_SECTORS * BDRV_SECTOR_SIZE);
1323 1324

    for (sector = 0; sector < total_sectors; sector += n) {
1325
        if (bdrv_is_allocated(bs, sector, COMMIT_BUF_SECTORS, &n)) {
1326 1327 1328 1329 1330 1331 1332 1333 1334 1335

            if (bdrv_read(bs, sector, buf, n) != 0) {
                ret = -EIO;
                goto ro_cleanup;
            }

            if (bdrv_write(bs->backing_hd, sector, buf, n) != 0) {
                ret = -EIO;
                goto ro_cleanup;
            }
B
bellard 已提交
1336
        }
1337
    }
1338

1339 1340 1341 1342
    if (drv->bdrv_make_empty) {
        ret = drv->bdrv_make_empty(bs);
        bdrv_flush(bs);
    }
1343

1344 1345 1346 1347 1348 1349
    /*
     * Make sure all data we wrote to the backing device is actually
     * stable on disk.
     */
    if (bs->backing_hd)
        bdrv_flush(bs->backing_hd);
1350 1351

ro_cleanup:
1352
    g_free(buf);
1353 1354 1355 1356 1357 1358

    if (ro) {
        /* re-open as RO */
        bdrv_delete(bs->backing_hd);
        bs->backing_hd = NULL;
        bs_ro = bdrv_new("");
1359 1360
        ret = bdrv_open(bs_ro, filename, open_flags & ~BDRV_O_RDWR,
            backing_drv);
1361 1362 1363 1364 1365 1366 1367 1368 1369 1370
        if (ret < 0) {
            bdrv_delete(bs_ro);
            /* drive not functional anymore */
            bs->drv = NULL;
            return ret;
        }
        bs->backing_hd = bs_ro;
        bs->backing_hd->keep_read_only = 0;
    }

1371
    return ret;
1372 1373
}

1374
int bdrv_commit_all(void)
1375 1376 1377 1378
{
    BlockDriverState *bs;

    QTAILQ_FOREACH(bs, &bdrv_states, list) {
1379 1380 1381 1382
        int ret = bdrv_commit(bs);
        if (ret < 0) {
            return ret;
        }
1383
    }
1384
    return 0;
1385 1386
}

S
Stefan Hajnoczi 已提交
1387 1388 1389 1390 1391 1392
struct BdrvTrackedRequest {
    BlockDriverState *bs;
    int64_t sector_num;
    int nb_sectors;
    bool is_write;
    QLIST_ENTRY(BdrvTrackedRequest) list;
1393
    Coroutine *co; /* owner, used for deadlock detection */
1394
    CoQueue wait_queue; /* coroutines blocked on this request */
S
Stefan Hajnoczi 已提交
1395 1396 1397 1398 1399 1400 1401 1402 1403 1404
};

/**
 * Remove an active request from the tracked requests list
 *
 * This function should be called when a tracked request is completing.
 */
static void tracked_request_end(BdrvTrackedRequest *req)
{
    QLIST_REMOVE(req, list);
1405
    qemu_co_queue_restart_all(&req->wait_queue);
S
Stefan Hajnoczi 已提交
1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420
}

/**
 * Add an active request to the tracked requests list
 */
static void tracked_request_begin(BdrvTrackedRequest *req,
                                  BlockDriverState *bs,
                                  int64_t sector_num,
                                  int nb_sectors, bool is_write)
{
    *req = (BdrvTrackedRequest){
        .bs = bs,
        .sector_num = sector_num,
        .nb_sectors = nb_sectors,
        .is_write = is_write,
1421
        .co = qemu_coroutine_self(),
S
Stefan Hajnoczi 已提交
1422 1423
    };

1424 1425
    qemu_co_queue_init(&req->wait_queue);

S
Stefan Hajnoczi 已提交
1426 1427 1428
    QLIST_INSERT_HEAD(&bs->tracked_requests, req, list);
}

S
Stefan Hajnoczi 已提交
1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449
/**
 * Round a region to cluster boundaries
 */
static void round_to_clusters(BlockDriverState *bs,
                              int64_t sector_num, int nb_sectors,
                              int64_t *cluster_sector_num,
                              int *cluster_nb_sectors)
{
    BlockDriverInfo bdi;

    if (bdrv_get_info(bs, &bdi) < 0 || bdi.cluster_size == 0) {
        *cluster_sector_num = sector_num;
        *cluster_nb_sectors = nb_sectors;
    } else {
        int64_t c = bdi.cluster_size / BDRV_SECTOR_SIZE;
        *cluster_sector_num = QEMU_ALIGN_DOWN(sector_num, c);
        *cluster_nb_sectors = QEMU_ALIGN_UP(sector_num - *cluster_sector_num +
                                            nb_sectors, c);
    }
}

1450 1451
static bool tracked_request_overlaps(BdrvTrackedRequest *req,
                                     int64_t sector_num, int nb_sectors) {
S
Stefan Hajnoczi 已提交
1452 1453 1454 1455 1456 1457 1458 1459 1460
    /*        aaaa   bbbb */
    if (sector_num >= req->sector_num + req->nb_sectors) {
        return false;
    }
    /* bbbb   aaaa        */
    if (req->sector_num >= sector_num + nb_sectors) {
        return false;
    }
    return true;
1461 1462 1463 1464 1465 1466
}

static void coroutine_fn wait_for_overlapping_requests(BlockDriverState *bs,
        int64_t sector_num, int nb_sectors)
{
    BdrvTrackedRequest *req;
S
Stefan Hajnoczi 已提交
1467 1468
    int64_t cluster_sector_num;
    int cluster_nb_sectors;
1469 1470
    bool retry;

S
Stefan Hajnoczi 已提交
1471 1472 1473 1474 1475 1476 1477 1478 1479
    /* If we touch the same cluster it counts as an overlap.  This guarantees
     * that allocating writes will be serialized and not race with each other
     * for the same cluster.  For example, in copy-on-read it ensures that the
     * CoR read and write operations are atomic and guest writes cannot
     * interleave between them.
     */
    round_to_clusters(bs, sector_num, nb_sectors,
                      &cluster_sector_num, &cluster_nb_sectors);

1480 1481 1482
    do {
        retry = false;
        QLIST_FOREACH(req, &bs->tracked_requests, list) {
S
Stefan Hajnoczi 已提交
1483 1484
            if (tracked_request_overlaps(req, cluster_sector_num,
                                         cluster_nb_sectors)) {
1485 1486 1487 1488 1489 1490
                /* Hitting this means there was a reentrant request, for
                 * example, a block driver issuing nested requests.  This must
                 * never happen since it means deadlock.
                 */
                assert(qemu_coroutine_self() != req->co);

1491 1492 1493 1494 1495 1496 1497 1498
                qemu_co_queue_wait(&req->wait_queue);
                retry = true;
                break;
            }
        }
    } while (retry);
}

K
Kevin Wolf 已提交
1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510
/*
 * Return values:
 * 0        - success
 * -EINVAL  - backing format specified, but no file
 * -ENOSPC  - can't update the backing file because no space is left in the
 *            image file header
 * -ENOTSUP - format driver doesn't support changing the backing file
 */
int bdrv_change_backing_file(BlockDriverState *bs,
    const char *backing_file, const char *backing_fmt)
{
    BlockDriver *drv = bs->drv;
1511
    int ret;
K
Kevin Wolf 已提交
1512

1513 1514 1515 1516 1517
    /* Backing file format doesn't make sense without a backing file */
    if (backing_fmt && !backing_file) {
        return -EINVAL;
    }

K
Kevin Wolf 已提交
1518
    if (drv->bdrv_change_backing_file != NULL) {
1519
        ret = drv->bdrv_change_backing_file(bs, backing_file, backing_fmt);
K
Kevin Wolf 已提交
1520
    } else {
1521
        ret = -ENOTSUP;
K
Kevin Wolf 已提交
1522
    }
1523 1524 1525 1526 1527 1528

    if (ret == 0) {
        pstrcpy(bs->backing_file, sizeof(bs->backing_file), backing_file ?: "");
        pstrcpy(bs->backing_format, sizeof(bs->backing_format), backing_fmt ?: "");
    }
    return ret;
K
Kevin Wolf 已提交
1529 1530
}

1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543
static int bdrv_check_byte_request(BlockDriverState *bs, int64_t offset,
                                   size_t size)
{
    int64_t len;

    if (!bdrv_is_inserted(bs))
        return -ENOMEDIUM;

    if (bs->growable)
        return 0;

    len = bdrv_getlength(bs);

K
Kevin Wolf 已提交
1544 1545 1546 1547
    if (offset < 0)
        return -EIO;

    if ((offset > len) || (len - offset < size))
1548 1549 1550 1551 1552 1553 1554 1555
        return -EIO;

    return 0;
}

static int bdrv_check_request(BlockDriverState *bs, int64_t sector_num,
                              int nb_sectors)
{
1556 1557
    return bdrv_check_byte_request(bs, sector_num * BDRV_SECTOR_SIZE,
                                   nb_sectors * BDRV_SECTOR_SIZE);
1558 1559
}

1560 1561 1562 1563 1564 1565 1566 1567 1568 1569
typedef struct RwCo {
    BlockDriverState *bs;
    int64_t sector_num;
    int nb_sectors;
    QEMUIOVector *qiov;
    bool is_write;
    int ret;
} RwCo;

static void coroutine_fn bdrv_rw_co_entry(void *opaque)
B
bellard 已提交
1570
{
1571
    RwCo *rwco = opaque;
B
bellard 已提交
1572

1573 1574
    if (!rwco->is_write) {
        rwco->ret = bdrv_co_do_readv(rwco->bs, rwco->sector_num,
1575
                                     rwco->nb_sectors, rwco->qiov, 0);
1576 1577
    } else {
        rwco->ret = bdrv_co_do_writev(rwco->bs, rwco->sector_num,
1578
                                      rwco->nb_sectors, rwco->qiov, 0);
1579 1580
    }
}
1581

1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601
/*
 * Process a synchronous request using coroutines
 */
static int bdrv_rw_co(BlockDriverState *bs, int64_t sector_num, uint8_t *buf,
                      int nb_sectors, bool is_write)
{
    QEMUIOVector qiov;
    struct iovec iov = {
        .iov_base = (void *)buf,
        .iov_len = nb_sectors * BDRV_SECTOR_SIZE,
    };
    Coroutine *co;
    RwCo rwco = {
        .bs = bs,
        .sector_num = sector_num,
        .nb_sectors = nb_sectors,
        .qiov = &qiov,
        .is_write = is_write,
        .ret = NOT_DONE,
    };
1602

1603
    qemu_iovec_init_external(&qiov, &iov, 1);
1604

1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615
    /**
     * In sync call context, when the vcpu is blocked, this throttling timer
     * will not fire; so the I/O throttling function has to be disabled here
     * if it has been enabled.
     */
    if (bs->io_limits_enabled) {
        fprintf(stderr, "Disabling I/O throttling on '%s' due "
                        "to synchronous I/O.\n", bdrv_get_device_name(bs));
        bdrv_io_limits_disable(bs);
    }

1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627
    if (qemu_in_coroutine()) {
        /* Fast-path if already in coroutine context */
        bdrv_rw_co_entry(&rwco);
    } else {
        co = qemu_coroutine_create(bdrv_rw_co_entry);
        qemu_coroutine_enter(co, &rwco);
        while (rwco.ret == NOT_DONE) {
            qemu_aio_wait();
        }
    }
    return rwco.ret;
}
B
bellard 已提交
1628

1629 1630 1631 1632 1633
/* return < 0 if error. See bdrv_write() for the return codes */
int bdrv_read(BlockDriverState *bs, int64_t sector_num,
              uint8_t *buf, int nb_sectors)
{
    return bdrv_rw_co(bs, sector_num, buf, nb_sectors, false);
B
bellard 已提交
1634 1635
}

1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649
/* Just like bdrv_read(), but with I/O throttling temporarily disabled */
int bdrv_read_unthrottled(BlockDriverState *bs, int64_t sector_num,
                          uint8_t *buf, int nb_sectors)
{
    bool enabled;
    int ret;

    enabled = bs->io_limits_enabled;
    bs->io_limits_enabled = false;
    ret = bdrv_read(bs, 0, buf, 1);
    bs->io_limits_enabled = enabled;
    return ret;
}

1650 1651
#define BITS_PER_LONG  (sizeof(unsigned long) * 8)

1652
static void set_dirty_bitmap(BlockDriverState *bs, int64_t sector_num,
1653
                             int nb_sectors, int dirty)
1654 1655
{
    int64_t start, end;
1656
    unsigned long val, idx, bit;
1657

J
Jan Kiszka 已提交
1658
    start = sector_num / BDRV_SECTORS_PER_DIRTY_CHUNK;
1659
    end = (sector_num + nb_sectors - 1) / BDRV_SECTORS_PER_DIRTY_CHUNK;
1660 1661

    for (; start <= end; start++) {
1662 1663
        idx = start / BITS_PER_LONG;
        bit = start % BITS_PER_LONG;
1664 1665
        val = bs->dirty_bitmap[idx];
        if (dirty) {
1666
            if (!(val & (1UL << bit))) {
1667
                bs->dirty_count++;
1668
                val |= 1UL << bit;
1669
            }
1670
        } else {
1671
            if (val & (1UL << bit)) {
1672
                bs->dirty_count--;
1673
                val &= ~(1UL << bit);
1674
            }
1675 1676
        }
        bs->dirty_bitmap[idx] = val;
1677 1678 1679
    }
}

1680
/* Return < 0 if error. Important errors are:
B
bellard 已提交
1681 1682 1683 1684 1685
  -EIO         generic I/O error (may happen for all errors)
  -ENOMEDIUM   No media inserted.
  -EINVAL      Invalid sector number or nb_sectors
  -EACCES      Trying to write a read-only device
*/
1686
int bdrv_write(BlockDriverState *bs, int64_t sector_num,
B
bellard 已提交
1687 1688
               const uint8_t *buf, int nb_sectors)
{
1689
    return bdrv_rw_co(bs, sector_num, (uint8_t *)buf, nb_sectors, true);
B
bellard 已提交
1690 1691
}

1692 1693
int bdrv_pread(BlockDriverState *bs, int64_t offset,
               void *buf, int count1)
B
bellard 已提交
1694
{
J
Jan Kiszka 已提交
1695
    uint8_t tmp_buf[BDRV_SECTOR_SIZE];
B
bellard 已提交
1696 1697
    int len, nb_sectors, count;
    int64_t sector_num;
1698
    int ret;
B
bellard 已提交
1699 1700 1701

    count = count1;
    /* first read to align to sector start */
J
Jan Kiszka 已提交
1702
    len = (BDRV_SECTOR_SIZE - offset) & (BDRV_SECTOR_SIZE - 1);
B
bellard 已提交
1703 1704
    if (len > count)
        len = count;
J
Jan Kiszka 已提交
1705
    sector_num = offset >> BDRV_SECTOR_BITS;
B
bellard 已提交
1706
    if (len > 0) {
1707 1708
        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
J
Jan Kiszka 已提交
1709
        memcpy(buf, tmp_buf + (offset & (BDRV_SECTOR_SIZE - 1)), len);
B
bellard 已提交
1710 1711 1712 1713 1714 1715 1716 1717
        count -= len;
        if (count == 0)
            return count1;
        sector_num++;
        buf += len;
    }

    /* read the sectors "in place" */
J
Jan Kiszka 已提交
1718
    nb_sectors = count >> BDRV_SECTOR_BITS;
B
bellard 已提交
1719
    if (nb_sectors > 0) {
1720 1721
        if ((ret = bdrv_read(bs, sector_num, buf, nb_sectors)) < 0)
            return ret;
B
bellard 已提交
1722
        sector_num += nb_sectors;
J
Jan Kiszka 已提交
1723
        len = nb_sectors << BDRV_SECTOR_BITS;
B
bellard 已提交
1724 1725 1726 1727 1728 1729
        buf += len;
        count -= len;
    }

    /* add data from the last sector */
    if (count > 0) {
1730 1731
        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
B
bellard 已提交
1732 1733 1734 1735 1736
        memcpy(buf, tmp_buf, count);
    }
    return count1;
}

1737 1738
int bdrv_pwrite(BlockDriverState *bs, int64_t offset,
                const void *buf, int count1)
B
bellard 已提交
1739
{
J
Jan Kiszka 已提交
1740
    uint8_t tmp_buf[BDRV_SECTOR_SIZE];
B
bellard 已提交
1741 1742
    int len, nb_sectors, count;
    int64_t sector_num;
1743
    int ret;
B
bellard 已提交
1744 1745 1746

    count = count1;
    /* first write to align to sector start */
J
Jan Kiszka 已提交
1747
    len = (BDRV_SECTOR_SIZE - offset) & (BDRV_SECTOR_SIZE - 1);
B
bellard 已提交
1748 1749
    if (len > count)
        len = count;
J
Jan Kiszka 已提交
1750
    sector_num = offset >> BDRV_SECTOR_BITS;
B
bellard 已提交
1751
    if (len > 0) {
1752 1753
        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
J
Jan Kiszka 已提交
1754
        memcpy(tmp_buf + (offset & (BDRV_SECTOR_SIZE - 1)), buf, len);
1755 1756
        if ((ret = bdrv_write(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
B
bellard 已提交
1757 1758 1759 1760 1761 1762 1763 1764
        count -= len;
        if (count == 0)
            return count1;
        sector_num++;
        buf += len;
    }

    /* write the sectors "in place" */
J
Jan Kiszka 已提交
1765
    nb_sectors = count >> BDRV_SECTOR_BITS;
B
bellard 已提交
1766
    if (nb_sectors > 0) {
1767 1768
        if ((ret = bdrv_write(bs, sector_num, buf, nb_sectors)) < 0)
            return ret;
B
bellard 已提交
1769
        sector_num += nb_sectors;
J
Jan Kiszka 已提交
1770
        len = nb_sectors << BDRV_SECTOR_BITS;
B
bellard 已提交
1771 1772 1773 1774 1775 1776
        buf += len;
        count -= len;
    }

    /* add data from the last sector */
    if (count > 0) {
1777 1778
        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
B
bellard 已提交
1779
        memcpy(tmp_buf, buf, count);
1780 1781
        if ((ret = bdrv_write(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
B
bellard 已提交
1782 1783 1784 1785
    }
    return count1;
}

K
Kevin Wolf 已提交
1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801
/*
 * Writes to the file and ensures that no writes are reordered across this
 * request (acts as a barrier)
 *
 * Returns 0 on success, -errno in error cases.
 */
int bdrv_pwrite_sync(BlockDriverState *bs, int64_t offset,
    const void *buf, int count)
{
    int ret;

    ret = bdrv_pwrite(bs, offset, buf, count);
    if (ret < 0) {
        return ret;
    }

1802 1803
    /* No flush needed for cache modes that already do it */
    if (bs->enable_write_cache) {
K
Kevin Wolf 已提交
1804 1805 1806 1807 1808 1809
        bdrv_flush(bs);
    }

    return 0;
}

1810
static int coroutine_fn bdrv_co_do_copy_on_readv(BlockDriverState *bs,
S
Stefan Hajnoczi 已提交
1811 1812 1813 1814 1815 1816 1817 1818 1819
        int64_t sector_num, int nb_sectors, QEMUIOVector *qiov)
{
    /* Perform I/O through a temporary buffer so that users who scribble over
     * their read buffer while the operation is in progress do not end up
     * modifying the image file.  This is critical for zero-copy guest I/O
     * where anything might happen inside guest memory.
     */
    void *bounce_buffer;

1820
    BlockDriver *drv = bs->drv;
S
Stefan Hajnoczi 已提交
1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833
    struct iovec iov;
    QEMUIOVector bounce_qiov;
    int64_t cluster_sector_num;
    int cluster_nb_sectors;
    size_t skip_bytes;
    int ret;

    /* Cover entire cluster so no additional backing file I/O is required when
     * allocating cluster in the image file.
     */
    round_to_clusters(bs, sector_num, nb_sectors,
                      &cluster_sector_num, &cluster_nb_sectors);

1834 1835
    trace_bdrv_co_do_copy_on_readv(bs, sector_num, nb_sectors,
                                   cluster_sector_num, cluster_nb_sectors);
S
Stefan Hajnoczi 已提交
1836 1837 1838 1839 1840

    iov.iov_len = cluster_nb_sectors * BDRV_SECTOR_SIZE;
    iov.iov_base = bounce_buffer = qemu_blockalign(bs, iov.iov_len);
    qemu_iovec_init_external(&bounce_qiov, &iov, 1);

1841 1842
    ret = drv->bdrv_co_readv(bs, cluster_sector_num, cluster_nb_sectors,
                             &bounce_qiov);
S
Stefan Hajnoczi 已提交
1843 1844 1845 1846
    if (ret < 0) {
        goto err;
    }

1847 1848
    if (drv->bdrv_co_write_zeroes &&
        buffer_is_zero(bounce_buffer, iov.iov_len)) {
K
Kevin Wolf 已提交
1849 1850
        ret = bdrv_co_do_write_zeroes(bs, cluster_sector_num,
                                      cluster_nb_sectors);
1851
    } else {
1852 1853 1854
        /* This does not change the data on the disk, it is not necessary
         * to flush even in cache=writethrough mode.
         */
1855
        ret = drv->bdrv_co_writev(bs, cluster_sector_num, cluster_nb_sectors,
S
Stefan Hajnoczi 已提交
1856
                                  &bounce_qiov);
1857 1858
    }

S
Stefan Hajnoczi 已提交
1859 1860 1861 1862 1863 1864 1865 1866 1867
    if (ret < 0) {
        /* It might be okay to ignore write errors for guest requests.  If this
         * is a deliberate copy-on-read then we don't want to ignore the error.
         * Simply report it in all cases.
         */
        goto err;
    }

    skip_bytes = (sector_num - cluster_sector_num) * BDRV_SECTOR_SIZE;
1868 1869
    qemu_iovec_from_buf(qiov, 0, bounce_buffer + skip_bytes,
                        nb_sectors * BDRV_SECTOR_SIZE);
S
Stefan Hajnoczi 已提交
1870 1871 1872 1873 1874 1875

err:
    qemu_vfree(bounce_buffer);
    return ret;
}

1876 1877 1878 1879
/*
 * Handle a read request in coroutine context
 */
static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,
1880 1881
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags)
K
Kevin Wolf 已提交
1882 1883
{
    BlockDriver *drv = bs->drv;
S
Stefan Hajnoczi 已提交
1884 1885
    BdrvTrackedRequest req;
    int ret;
K
Kevin Wolf 已提交
1886 1887 1888 1889 1890 1891 1892 1893

    if (!drv) {
        return -ENOMEDIUM;
    }
    if (bdrv_check_request(bs, sector_num, nb_sectors)) {
        return -EIO;
    }

Z
Zhi Yong Wu 已提交
1894 1895 1896 1897 1898
    /* throttling disk read I/O */
    if (bs->io_limits_enabled) {
        bdrv_io_limits_intercept(bs, false, nb_sectors);
    }

1899
    if (bs->copy_on_read) {
1900 1901 1902 1903 1904 1905 1906
        flags |= BDRV_REQ_COPY_ON_READ;
    }
    if (flags & BDRV_REQ_COPY_ON_READ) {
        bs->copy_on_read_in_flight++;
    }

    if (bs->copy_on_read_in_flight) {
1907 1908 1909
        wait_for_overlapping_requests(bs, sector_num, nb_sectors);
    }

S
Stefan Hajnoczi 已提交
1910
    tracked_request_begin(&req, bs, sector_num, nb_sectors, false);
S
Stefan Hajnoczi 已提交
1911

1912
    if (flags & BDRV_REQ_COPY_ON_READ) {
S
Stefan Hajnoczi 已提交
1913 1914 1915 1916 1917 1918 1919 1920
        int pnum;

        ret = bdrv_co_is_allocated(bs, sector_num, nb_sectors, &pnum);
        if (ret < 0) {
            goto out;
        }

        if (!ret || pnum != nb_sectors) {
1921
            ret = bdrv_co_do_copy_on_readv(bs, sector_num, nb_sectors, qiov);
S
Stefan Hajnoczi 已提交
1922 1923 1924 1925
            goto out;
        }
    }

S
Stefan Hajnoczi 已提交
1926
    ret = drv->bdrv_co_readv(bs, sector_num, nb_sectors, qiov);
S
Stefan Hajnoczi 已提交
1927 1928

out:
S
Stefan Hajnoczi 已提交
1929
    tracked_request_end(&req);
1930 1931 1932 1933 1934

    if (flags & BDRV_REQ_COPY_ON_READ) {
        bs->copy_on_read_in_flight--;
    }

S
Stefan Hajnoczi 已提交
1935
    return ret;
K
Kevin Wolf 已提交
1936 1937
}

1938
int coroutine_fn bdrv_co_readv(BlockDriverState *bs, int64_t sector_num,
K
Kevin Wolf 已提交
1939 1940
    int nb_sectors, QEMUIOVector *qiov)
{
1941
    trace_bdrv_co_readv(bs, sector_num, nb_sectors);
K
Kevin Wolf 已提交
1942

1943 1944 1945 1946 1947 1948 1949 1950 1951 1952
    return bdrv_co_do_readv(bs, sector_num, nb_sectors, qiov, 0);
}

int coroutine_fn bdrv_co_copy_on_readv(BlockDriverState *bs,
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov)
{
    trace_bdrv_co_copy_on_readv(bs, sector_num, nb_sectors);

    return bdrv_co_do_readv(bs, sector_num, nb_sectors, qiov,
                            BDRV_REQ_COPY_ON_READ);
1953 1954
}

1955 1956 1957 1958 1959 1960 1961 1962
static int coroutine_fn bdrv_co_do_write_zeroes(BlockDriverState *bs,
    int64_t sector_num, int nb_sectors)
{
    BlockDriver *drv = bs->drv;
    QEMUIOVector qiov;
    struct iovec iov;
    int ret;

K
Kevin Wolf 已提交
1963 1964 1965
    /* TODO Emulate only part of misaligned requests instead of letting block
     * drivers return -ENOTSUP and emulate everything */

1966 1967
    /* First try the efficient write zeroes operation */
    if (drv->bdrv_co_write_zeroes) {
K
Kevin Wolf 已提交
1968 1969 1970 1971
        ret = drv->bdrv_co_write_zeroes(bs, sector_num, nb_sectors);
        if (ret != -ENOTSUP) {
            return ret;
        }
1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985
    }

    /* Fall back to bounce buffer if write zeroes is unsupported */
    iov.iov_len  = nb_sectors * BDRV_SECTOR_SIZE;
    iov.iov_base = qemu_blockalign(bs, iov.iov_len);
    memset(iov.iov_base, 0, iov.iov_len);
    qemu_iovec_init_external(&qiov, &iov, 1);

    ret = drv->bdrv_co_writev(bs, sector_num, nb_sectors, &qiov);

    qemu_vfree(iov.iov_base);
    return ret;
}

1986 1987 1988 1989
/*
 * Handle a write request in coroutine context
 */
static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs,
1990 1991
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags)
1992 1993
{
    BlockDriver *drv = bs->drv;
S
Stefan Hajnoczi 已提交
1994
    BdrvTrackedRequest req;
1995
    int ret;
K
Kevin Wolf 已提交
1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006

    if (!bs->drv) {
        return -ENOMEDIUM;
    }
    if (bs->read_only) {
        return -EACCES;
    }
    if (bdrv_check_request(bs, sector_num, nb_sectors)) {
        return -EIO;
    }

Z
Zhi Yong Wu 已提交
2007 2008 2009 2010 2011
    /* throttling disk write I/O */
    if (bs->io_limits_enabled) {
        bdrv_io_limits_intercept(bs, true, nb_sectors);
    }

2012
    if (bs->copy_on_read_in_flight) {
2013 2014 2015
        wait_for_overlapping_requests(bs, sector_num, nb_sectors);
    }

S
Stefan Hajnoczi 已提交
2016 2017
    tracked_request_begin(&req, bs, sector_num, nb_sectors, true);

2018 2019 2020 2021 2022
    if (flags & BDRV_REQ_ZERO_WRITE) {
        ret = bdrv_co_do_write_zeroes(bs, sector_num, nb_sectors);
    } else {
        ret = drv->bdrv_co_writev(bs, sector_num, nb_sectors, qiov);
    }
2023

2024 2025 2026 2027
    if (ret == 0 && !bs->enable_write_cache) {
        ret = bdrv_co_flush(bs);
    }

K
Kevin Wolf 已提交
2028 2029 2030 2031 2032 2033 2034 2035
    if (bs->dirty_bitmap) {
        set_dirty_bitmap(bs, sector_num, nb_sectors, 1);
    }

    if (bs->wr_highest_sector < sector_num + nb_sectors - 1) {
        bs->wr_highest_sector = sector_num + nb_sectors - 1;
    }

S
Stefan Hajnoczi 已提交
2036 2037
    tracked_request_end(&req);

2038
    return ret;
K
Kevin Wolf 已提交
2039 2040
}

2041 2042 2043 2044 2045
int coroutine_fn bdrv_co_writev(BlockDriverState *bs, int64_t sector_num,
    int nb_sectors, QEMUIOVector *qiov)
{
    trace_bdrv_co_writev(bs, sector_num, nb_sectors);

2046 2047 2048 2049 2050 2051 2052 2053 2054 2055
    return bdrv_co_do_writev(bs, sector_num, nb_sectors, qiov, 0);
}

int coroutine_fn bdrv_co_write_zeroes(BlockDriverState *bs,
                                      int64_t sector_num, int nb_sectors)
{
    trace_bdrv_co_write_zeroes(bs, sector_num, nb_sectors);

    return bdrv_co_do_writev(bs, sector_num, nb_sectors, NULL,
                             BDRV_REQ_ZERO_WRITE);
2056 2057
}

B
bellard 已提交
2058 2059 2060 2061 2062 2063
/**
 * Truncate file to 'offset' bytes (needed only for file protocols)
 */
int bdrv_truncate(BlockDriverState *bs, int64_t offset)
{
    BlockDriver *drv = bs->drv;
2064
    int ret;
B
bellard 已提交
2065
    if (!drv)
B
bellard 已提交
2066
        return -ENOMEDIUM;
B
bellard 已提交
2067 2068
    if (!drv->bdrv_truncate)
        return -ENOTSUP;
2069 2070
    if (bs->read_only)
        return -EACCES;
M
Marcelo Tosatti 已提交
2071 2072
    if (bdrv_in_use(bs))
        return -EBUSY;
2073 2074 2075
    ret = drv->bdrv_truncate(bs, offset);
    if (ret == 0) {
        ret = refresh_total_sectors(bs, offset >> BDRV_SECTOR_BITS);
2076
        bdrv_dev_resize_cb(bs);
2077 2078
    }
    return ret;
B
bellard 已提交
2079 2080
}

2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099
/**
 * Length of a allocated file in bytes. Sparse files are counted by actual
 * allocated space. Return < 0 if error or unknown.
 */
int64_t bdrv_get_allocated_file_size(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;
    if (!drv) {
        return -ENOMEDIUM;
    }
    if (drv->bdrv_get_allocated_file_size) {
        return drv->bdrv_get_allocated_file_size(bs);
    }
    if (bs->file) {
        return bdrv_get_allocated_file_size(bs->file);
    }
    return -ENOTSUP;
}

B
bellard 已提交
2100 2101 2102 2103 2104 2105 2106
/**
 * Length of a file in bytes. Return < 0 if error or unknown.
 */
int64_t bdrv_getlength(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
2107
        return -ENOMEDIUM;
2108

2109
    if (bs->growable || bdrv_dev_has_removable_media(bs)) {
2110 2111 2112
        if (drv->bdrv_getlength) {
            return drv->bdrv_getlength(bs);
        }
B
bellard 已提交
2113
    }
2114
    return bs->total_sectors * BDRV_SECTOR_SIZE;
B
bellard 已提交
2115 2116
}

B
bellard 已提交
2117
/* return 0 as number of sectors if no device present or error */
2118
void bdrv_get_geometry(BlockDriverState *bs, uint64_t *nb_sectors_ptr)
B
bellard 已提交
2119
{
B
bellard 已提交
2120 2121 2122 2123 2124
    int64_t length;
    length = bdrv_getlength(bs);
    if (length < 0)
        length = 0;
    else
J
Jan Kiszka 已提交
2125
        length = length >> BDRV_SECTOR_BITS;
B
bellard 已提交
2126
    *nb_sectors_ptr = length;
B
bellard 已提交
2127
}
B
bellard 已提交
2128

2129 2130 2131 2132 2133 2134 2135 2136
/* throttling disk io limits */
void bdrv_set_io_limits(BlockDriverState *bs,
                        BlockIOLimit *io_limits)
{
    bs->io_limits = *io_limits;
    bs->io_limits_enabled = bdrv_io_limits_enabled(bs);
}

2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148
void bdrv_set_on_error(BlockDriverState *bs, BlockErrorAction on_read_error,
                       BlockErrorAction on_write_error)
{
    bs->on_read_error = on_read_error;
    bs->on_write_error = on_write_error;
}

BlockErrorAction bdrv_get_on_error(BlockDriverState *bs, int is_read)
{
    return is_read ? bs->on_read_error : bs->on_write_error;
}

B
bellard 已提交
2149 2150 2151 2152 2153
int bdrv_is_read_only(BlockDriverState *bs)
{
    return bs->read_only;
}

2154 2155 2156 2157 2158
int bdrv_is_sg(BlockDriverState *bs)
{
    return bs->sg;
}

2159 2160 2161 2162 2163
int bdrv_enable_write_cache(BlockDriverState *bs)
{
    return bs->enable_write_cache;
}

2164 2165 2166 2167 2168
void bdrv_set_enable_write_cache(BlockDriverState *bs, bool wce)
{
    bs->enable_write_cache = wce;
}

B
bellard 已提交
2169 2170 2171 2172 2173 2174 2175
int bdrv_is_encrypted(BlockDriverState *bs)
{
    if (bs->backing_hd && bs->backing_hd->encrypted)
        return 1;
    return bs->encrypted;
}

2176 2177 2178 2179 2180 2181 2182 2183 2184
int bdrv_key_required(BlockDriverState *bs)
{
    BlockDriverState *backing_hd = bs->backing_hd;

    if (backing_hd && backing_hd->encrypted && !backing_hd->valid_key)
        return 1;
    return (bs->encrypted && !bs->valid_key);
}

B
bellard 已提交
2185 2186 2187 2188 2189 2190 2191 2192 2193 2194
int bdrv_set_key(BlockDriverState *bs, const char *key)
{
    int ret;
    if (bs->backing_hd && bs->backing_hd->encrypted) {
        ret = bdrv_set_key(bs->backing_hd, key);
        if (ret < 0)
            return ret;
        if (!bs->encrypted)
            return 0;
    }
2195 2196 2197 2198 2199
    if (!bs->encrypted) {
        return -EINVAL;
    } else if (!bs->drv || !bs->drv->bdrv_set_key) {
        return -ENOMEDIUM;
    }
2200
    ret = bs->drv->bdrv_set_key(bs, key);
2201 2202 2203 2204 2205
    if (ret < 0) {
        bs->valid_key = 0;
    } else if (!bs->valid_key) {
        bs->valid_key = 1;
        /* call the change callback now, we skipped it on open */
2206
        bdrv_dev_change_media_cb(bs, true);
2207
    }
2208
    return ret;
B
bellard 已提交
2209 2210
}

2211
const char *bdrv_get_format_name(BlockDriverState *bs)
B
bellard 已提交
2212
{
2213
    return bs->drv ? bs->drv->format_name : NULL;
B
bellard 已提交
2214 2215
}

2216
void bdrv_iterate_format(void (*it)(void *opaque, const char *name),
B
bellard 已提交
2217 2218 2219 2220
                         void *opaque)
{
    BlockDriver *drv;

2221
    QLIST_FOREACH(drv, &bdrv_drivers, list) {
B
bellard 已提交
2222 2223 2224 2225
        it(opaque, drv->format_name);
    }
}

B
bellard 已提交
2226 2227 2228 2229
BlockDriverState *bdrv_find(const char *name)
{
    BlockDriverState *bs;

2230 2231
    QTAILQ_FOREACH(bs, &bdrv_states, list) {
        if (!strcmp(name, bs->device_name)) {
B
bellard 已提交
2232
            return bs;
2233
        }
B
bellard 已提交
2234 2235 2236 2237
    }
    return NULL;
}

M
Markus Armbruster 已提交
2238 2239 2240 2241 2242 2243 2244 2245
BlockDriverState *bdrv_next(BlockDriverState *bs)
{
    if (!bs) {
        return QTAILQ_FIRST(&bdrv_states);
    }
    return QTAILQ_NEXT(bs, list);
}

2246
void bdrv_iterate(void (*it)(void *opaque, BlockDriverState *bs), void *opaque)
B
bellard 已提交
2247 2248 2249
{
    BlockDriverState *bs;

2250
    QTAILQ_FOREACH(bs, &bdrv_states, list) {
2251
        it(opaque, bs);
B
bellard 已提交
2252 2253 2254
    }
}

B
bellard 已提交
2255 2256 2257 2258 2259
const char *bdrv_get_device_name(BlockDriverState *bs)
{
    return bs->device_name;
}

M
Markus Armbruster 已提交
2260 2261 2262 2263 2264
int bdrv_get_flags(BlockDriverState *bs)
{
    return bs->open_flags;
}

A
aliguori 已提交
2265 2266 2267 2268
void bdrv_flush_all(void)
{
    BlockDriverState *bs;

2269
    QTAILQ_FOREACH(bs, &bdrv_states, list) {
2270
        bdrv_flush(bs);
2271
    }
A
aliguori 已提交
2272 2273
}

K
Kevin Wolf 已提交
2274 2275 2276 2277
int bdrv_has_zero_init(BlockDriverState *bs)
{
    assert(bs->drv);

K
Kevin Wolf 已提交
2278 2279
    if (bs->drv->bdrv_has_zero_init) {
        return bs->drv->bdrv_has_zero_init(bs);
K
Kevin Wolf 已提交
2280 2281 2282 2283 2284
    }

    return 1;
}

2285 2286 2287 2288 2289 2290 2291 2292 2293
typedef struct BdrvCoIsAllocatedData {
    BlockDriverState *bs;
    int64_t sector_num;
    int nb_sectors;
    int *pnum;
    int ret;
    bool done;
} BdrvCoIsAllocatedData;

2294 2295 2296 2297 2298
/*
 * Returns true iff the specified sector is present in the disk image. Drivers
 * not implementing the functionality are assumed to not support backing files,
 * hence all their sectors are reported as allocated.
 *
2299 2300 2301
 * If 'sector_num' is beyond the end of the disk image the return value is 0
 * and 'pnum' is set to 0.
 *
2302 2303 2304 2305
 * 'pnum' is set to the number of sectors (including and immediately following
 * the specified sector) that are known to be in the same
 * allocated/unallocated state.
 *
2306 2307
 * 'nb_sectors' is the max value 'pnum' should be set to.  If nb_sectors goes
 * beyond the end of the disk image it will be clamped.
2308
 */
2309 2310
int coroutine_fn bdrv_co_is_allocated(BlockDriverState *bs, int64_t sector_num,
                                      int nb_sectors, int *pnum)
2311
{
2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323
    int64_t n;

    if (sector_num >= bs->total_sectors) {
        *pnum = 0;
        return 0;
    }

    n = bs->total_sectors - sector_num;
    if (n < nb_sectors) {
        nb_sectors = n;
    }

2324
    if (!bs->drv->bdrv_co_is_allocated) {
2325
        *pnum = nb_sectors;
2326 2327
        return 1;
    }
2328

2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350
    return bs->drv->bdrv_co_is_allocated(bs, sector_num, nb_sectors, pnum);
}

/* Coroutine wrapper for bdrv_is_allocated() */
static void coroutine_fn bdrv_is_allocated_co_entry(void *opaque)
{
    BdrvCoIsAllocatedData *data = opaque;
    BlockDriverState *bs = data->bs;

    data->ret = bdrv_co_is_allocated(bs, data->sector_num, data->nb_sectors,
                                     data->pnum);
    data->done = true;
}

/*
 * Synchronous wrapper around bdrv_co_is_allocated().
 *
 * See bdrv_co_is_allocated() for details.
 */
int bdrv_is_allocated(BlockDriverState *bs, int64_t sector_num, int nb_sectors,
                      int *pnum)
{
2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365
    Coroutine *co;
    BdrvCoIsAllocatedData data = {
        .bs = bs,
        .sector_num = sector_num,
        .nb_sectors = nb_sectors,
        .pnum = pnum,
        .done = false,
    };

    co = qemu_coroutine_create(bdrv_is_allocated_co_entry);
    qemu_coroutine_enter(co, &data);
    while (!data.done) {
        qemu_aio_wait();
    }
    return data.ret;
2366 2367
}

2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416
/*
 * Given an image chain: ... -> [BASE] -> [INTER1] -> [INTER2] -> [TOP]
 *
 * Return true if the given sector is allocated in any image between
 * BASE and TOP (inclusive).  BASE can be NULL to check if the given
 * sector is allocated in any image of the chain.  Return false otherwise.
 *
 * 'pnum' is set to the number of sectors (including and immediately following
 *  the specified sector) that are known to be in the same
 *  allocated/unallocated state.
 *
 */
int coroutine_fn bdrv_co_is_allocated_above(BlockDriverState *top,
                                            BlockDriverState *base,
                                            int64_t sector_num,
                                            int nb_sectors, int *pnum)
{
    BlockDriverState *intermediate;
    int ret, n = nb_sectors;

    intermediate = top;
    while (intermediate && intermediate != base) {
        int pnum_inter;
        ret = bdrv_co_is_allocated(intermediate, sector_num, nb_sectors,
                                   &pnum_inter);
        if (ret < 0) {
            return ret;
        } else if (ret) {
            *pnum = pnum_inter;
            return 1;
        }

        /*
         * [sector_num, nb_sectors] is unallocated on top but intermediate
         * might have
         *
         * [sector_num+x, nr_sectors] allocated.
         */
        if (n > pnum_inter) {
            n = pnum_inter;
        }

        intermediate = intermediate->backing_hd;
    }

    *pnum = n;
    return 0;
}

L
Luiz Capitulino 已提交
2417
BlockInfoList *qmp_query_block(Error **errp)
B
bellard 已提交
2418
{
L
Luiz Capitulino 已提交
2419
    BlockInfoList *head = NULL, *cur_item = NULL;
B
bellard 已提交
2420 2421
    BlockDriverState *bs;

2422
    QTAILQ_FOREACH(bs, &bdrv_states, list) {
L
Luiz Capitulino 已提交
2423
        BlockInfoList *info = g_malloc0(sizeof(*info));
2424

L
Luiz Capitulino 已提交
2425 2426 2427 2428 2429
        info->value = g_malloc0(sizeof(*info->value));
        info->value->device = g_strdup(bs->device_name);
        info->value->type = g_strdup("unknown");
        info->value->locked = bdrv_dev_is_medium_locked(bs);
        info->value->removable = bdrv_dev_has_removable_media(bs);
2430

2431
        if (bdrv_dev_has_removable_media(bs)) {
L
Luiz Capitulino 已提交
2432 2433
            info->value->has_tray_open = true;
            info->value->tray_open = bdrv_dev_is_tray_open(bs);
2434
        }
2435 2436

        if (bdrv_iostatus_is_enabled(bs)) {
L
Luiz Capitulino 已提交
2437 2438
            info->value->has_io_status = true;
            info->value->io_status = bs->iostatus;
2439 2440
        }

B
bellard 已提交
2441
        if (bs->drv) {
L
Luiz Capitulino 已提交
2442 2443 2444 2445 2446 2447
            info->value->has_inserted = true;
            info->value->inserted = g_malloc0(sizeof(*info->value->inserted));
            info->value->inserted->file = g_strdup(bs->filename);
            info->value->inserted->ro = bs->read_only;
            info->value->inserted->drv = g_strdup(bs->drv->format_name);
            info->value->inserted->encrypted = bs->encrypted;
2448
            info->value->inserted->encryption_key_missing = bdrv_key_required(bs);
L
Luiz Capitulino 已提交
2449 2450 2451
            if (bs->backing_file[0]) {
                info->value->inserted->has_backing_file = true;
                info->value->inserted->backing_file = g_strdup(bs->backing_file);
A
aliguori 已提交
2452
            }
Z
Zhi Yong Wu 已提交
2453

2454 2455 2456
            info->value->inserted->backing_file_depth =
                bdrv_get_backing_file_depth(bs);

Z
Zhi Yong Wu 已提交
2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470
            if (bs->io_limits_enabled) {
                info->value->inserted->bps =
                               bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL];
                info->value->inserted->bps_rd =
                               bs->io_limits.bps[BLOCK_IO_LIMIT_READ];
                info->value->inserted->bps_wr =
                               bs->io_limits.bps[BLOCK_IO_LIMIT_WRITE];
                info->value->inserted->iops =
                               bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL];
                info->value->inserted->iops_rd =
                               bs->io_limits.iops[BLOCK_IO_LIMIT_READ];
                info->value->inserted->iops_wr =
                               bs->io_limits.iops[BLOCK_IO_LIMIT_WRITE];
            }
L
Luiz Capitulino 已提交
2471
        }
2472

L
Luiz Capitulino 已提交
2473 2474 2475 2476 2477 2478
        /* XXX: waiting for the qapi to support GSList */
        if (!cur_item) {
            head = cur_item = info;
        } else {
            cur_item->next = info;
            cur_item = info;
B
bellard 已提交
2479 2480
        }
    }
2481

L
Luiz Capitulino 已提交
2482
    return head;
B
bellard 已提交
2483
}
2484

L
Luiz Capitulino 已提交
2485 2486 2487 2488 2489 2490 2491 2492 2493 2494
/* Consider exposing this as a full fledged QMP command */
static BlockStats *qmp_query_blockstat(const BlockDriverState *bs, Error **errp)
{
    BlockStats *s;

    s = g_malloc0(sizeof(*s));

    if (bs->device_name[0]) {
        s->has_device = true;
        s->device = g_strdup(bs->device_name);
2495 2496
    }

L
Luiz Capitulino 已提交
2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507
    s->stats = g_malloc0(sizeof(*s->stats));
    s->stats->rd_bytes = bs->nr_bytes[BDRV_ACCT_READ];
    s->stats->wr_bytes = bs->nr_bytes[BDRV_ACCT_WRITE];
    s->stats->rd_operations = bs->nr_ops[BDRV_ACCT_READ];
    s->stats->wr_operations = bs->nr_ops[BDRV_ACCT_WRITE];
    s->stats->wr_highest_offset = bs->wr_highest_sector * BDRV_SECTOR_SIZE;
    s->stats->flush_operations = bs->nr_ops[BDRV_ACCT_FLUSH];
    s->stats->wr_total_time_ns = bs->total_time_ns[BDRV_ACCT_WRITE];
    s->stats->rd_total_time_ns = bs->total_time_ns[BDRV_ACCT_READ];
    s->stats->flush_total_time_ns = bs->total_time_ns[BDRV_ACCT_FLUSH];

2508
    if (bs->file) {
L
Luiz Capitulino 已提交
2509 2510
        s->has_parent = true;
        s->parent = qmp_query_blockstat(bs->file, NULL);
2511 2512
    }

L
Luiz Capitulino 已提交
2513
    return s;
2514 2515
}

L
Luiz Capitulino 已提交
2516
BlockStatsList *qmp_query_blockstats(Error **errp)
2517
{
L
Luiz Capitulino 已提交
2518
    BlockStatsList *head = NULL, *cur_item = NULL;
2519 2520
    BlockDriverState *bs;

2521
    QTAILQ_FOREACH(bs, &bdrv_states, list) {
L
Luiz Capitulino 已提交
2522 2523 2524 2525 2526 2527 2528 2529 2530 2531
        BlockStatsList *info = g_malloc0(sizeof(*info));
        info->value = qmp_query_blockstat(bs, NULL);

        /* XXX: waiting for the qapi to support GSList */
        if (!cur_item) {
            head = cur_item = info;
        } else {
            cur_item->next = info;
            cur_item = info;
        }
2532
    }
2533

L
Luiz Capitulino 已提交
2534
    return head;
2535
}
B
bellard 已提交
2536

2537 2538 2539 2540 2541 2542 2543 2544 2545 2546
const char *bdrv_get_encrypted_filename(BlockDriverState *bs)
{
    if (bs->backing_hd && bs->backing_hd->encrypted)
        return bs->backing_file;
    else if (bs->encrypted)
        return bs->filename;
    else
        return NULL;
}

2547
void bdrv_get_backing_filename(BlockDriverState *bs,
B
bellard 已提交
2548 2549
                               char *filename, int filename_size)
{
K
Kevin Wolf 已提交
2550
    pstrcpy(filename, filename_size, bs->backing_file);
B
bellard 已提交
2551 2552
}

2553
int bdrv_write_compressed(BlockDriverState *bs, int64_t sector_num,
B
bellard 已提交
2554 2555 2556 2557
                          const uint8_t *buf, int nb_sectors)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
2558
        return -ENOMEDIUM;
B
bellard 已提交
2559 2560
    if (!drv->bdrv_write_compressed)
        return -ENOTSUP;
K
Kevin Wolf 已提交
2561 2562
    if (bdrv_check_request(bs, sector_num, nb_sectors))
        return -EIO;
2563

2564
    if (bs->dirty_bitmap) {
2565 2566
        set_dirty_bitmap(bs, sector_num, nb_sectors, 1);
    }
2567

B
bellard 已提交
2568 2569
    return drv->bdrv_write_compressed(bs, sector_num, buf, nb_sectors);
}
2570

B
bellard 已提交
2571 2572 2573 2574
int bdrv_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
2575
        return -ENOMEDIUM;
B
bellard 已提交
2576 2577 2578 2579 2580 2581
    if (!drv->bdrv_get_info)
        return -ENOTSUP;
    memset(bdi, 0, sizeof(*bdi));
    return drv->bdrv_get_info(bs, bdi);
}

2582 2583
int bdrv_save_vmstate(BlockDriverState *bs, const uint8_t *buf,
                      int64_t pos, int size)
2584 2585 2586 2587
{
    BlockDriver *drv = bs->drv;
    if (!drv)
        return -ENOMEDIUM;
2588 2589 2590 2591 2592
    if (drv->bdrv_save_vmstate)
        return drv->bdrv_save_vmstate(bs, buf, pos, size);
    if (bs->file)
        return bdrv_save_vmstate(bs->file, buf, pos, size);
    return -ENOTSUP;
2593 2594
}

2595 2596
int bdrv_load_vmstate(BlockDriverState *bs, uint8_t *buf,
                      int64_t pos, int size)
2597 2598 2599 2600
{
    BlockDriver *drv = bs->drv;
    if (!drv)
        return -ENOMEDIUM;
2601 2602 2603 2604 2605
    if (drv->bdrv_load_vmstate)
        return drv->bdrv_load_vmstate(bs, buf, pos, size);
    if (bs->file)
        return bdrv_load_vmstate(bs->file, buf, pos, size);
    return -ENOTSUP;
2606 2607
}

K
Kevin Wolf 已提交
2608 2609 2610 2611 2612 2613 2614 2615
void bdrv_debug_event(BlockDriverState *bs, BlkDebugEvent event)
{
    BlockDriver *drv = bs->drv;

    if (!drv || !drv->bdrv_debug_event) {
        return;
    }

B
Blue Swirl 已提交
2616
    drv->bdrv_debug_event(bs, event);
K
Kevin Wolf 已提交
2617 2618 2619

}

B
bellard 已提交
2620 2621 2622
/**************************************************************/
/* handling of snapshots */

2623 2624 2625
int bdrv_can_snapshot(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;
2626
    if (!drv || !bdrv_is_inserted(bs) || bdrv_is_read_only(bs)) {
2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639
        return 0;
    }

    if (!drv->bdrv_snapshot_create) {
        if (bs->file != NULL) {
            return bdrv_can_snapshot(bs->file);
        }
        return 0;
    }

    return 1;
}

2640 2641 2642 2643 2644
int bdrv_is_snapshot(BlockDriverState *bs)
{
    return !!(bs->open_flags & BDRV_O_SNAPSHOT);
}

2645 2646 2647 2648
BlockDriverState *bdrv_snapshots(void)
{
    BlockDriverState *bs;

2649
    if (bs_snapshots) {
2650
        return bs_snapshots;
2651
    }
2652 2653 2654 2655

    bs = NULL;
    while ((bs = bdrv_next(bs))) {
        if (bdrv_can_snapshot(bs)) {
2656 2657
            bs_snapshots = bs;
            return bs;
2658 2659 2660 2661 2662
        }
    }
    return NULL;
}

2663
int bdrv_snapshot_create(BlockDriverState *bs,
B
bellard 已提交
2664 2665 2666 2667
                         QEMUSnapshotInfo *sn_info)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
2668
        return -ENOMEDIUM;
2669 2670 2671 2672 2673
    if (drv->bdrv_snapshot_create)
        return drv->bdrv_snapshot_create(bs, sn_info);
    if (bs->file)
        return bdrv_snapshot_create(bs->file, sn_info);
    return -ENOTSUP;
B
bellard 已提交
2674 2675
}

2676
int bdrv_snapshot_goto(BlockDriverState *bs,
B
bellard 已提交
2677 2678 2679
                       const char *snapshot_id)
{
    BlockDriver *drv = bs->drv;
2680 2681
    int ret, open_ret;

B
bellard 已提交
2682
    if (!drv)
B
bellard 已提交
2683
        return -ENOMEDIUM;
2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699
    if (drv->bdrv_snapshot_goto)
        return drv->bdrv_snapshot_goto(bs, snapshot_id);

    if (bs->file) {
        drv->bdrv_close(bs);
        ret = bdrv_snapshot_goto(bs->file, snapshot_id);
        open_ret = drv->bdrv_open(bs, bs->open_flags);
        if (open_ret < 0) {
            bdrv_delete(bs->file);
            bs->drv = NULL;
            return open_ret;
        }
        return ret;
    }

    return -ENOTSUP;
B
bellard 已提交
2700 2701 2702 2703 2704 2705
}

int bdrv_snapshot_delete(BlockDriverState *bs, const char *snapshot_id)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
2706
        return -ENOMEDIUM;
2707 2708 2709 2710 2711
    if (drv->bdrv_snapshot_delete)
        return drv->bdrv_snapshot_delete(bs, snapshot_id);
    if (bs->file)
        return bdrv_snapshot_delete(bs->file, snapshot_id);
    return -ENOTSUP;
B
bellard 已提交
2712 2713
}

2714
int bdrv_snapshot_list(BlockDriverState *bs,
B
bellard 已提交
2715 2716 2717 2718
                       QEMUSnapshotInfo **psn_info)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
2719
        return -ENOMEDIUM;
2720 2721 2722 2723 2724
    if (drv->bdrv_snapshot_list)
        return drv->bdrv_snapshot_list(bs, psn_info);
    if (bs->file)
        return bdrv_snapshot_list(bs->file, psn_info);
    return -ENOTSUP;
B
bellard 已提交
2725 2726
}

E
edison 已提交
2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742
int bdrv_snapshot_load_tmp(BlockDriverState *bs,
        const char *snapshot_name)
{
    BlockDriver *drv = bs->drv;
    if (!drv) {
        return -ENOMEDIUM;
    }
    if (!bs->read_only) {
        return -EINVAL;
    }
    if (drv->bdrv_snapshot_load_tmp) {
        return drv->bdrv_snapshot_load_tmp(bs, snapshot_name);
    }
    return -ENOTSUP;
}

2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760
BlockDriverState *bdrv_find_backing_image(BlockDriverState *bs,
        const char *backing_file)
{
    if (!bs->drv) {
        return NULL;
    }

    if (bs->backing_hd) {
        if (strcmp(bs->backing_file, backing_file) == 0) {
            return bs->backing_hd;
        } else {
            return bdrv_find_backing_image(bs->backing_hd, backing_file);
        }
    }

    return NULL;
}

2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773
int bdrv_get_backing_file_depth(BlockDriverState *bs)
{
    if (!bs->drv) {
        return 0;
    }

    if (!bs->backing_hd) {
        return 0;
    }

    return 1 + bdrv_get_backing_file_depth(bs->backing_hd);
}

B
bellard 已提交
2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787
#define NB_SUFFIXES 4

char *get_human_readable_size(char *buf, int buf_size, int64_t size)
{
    static const char suffixes[NB_SUFFIXES] = "KMGT";
    int64_t base;
    int i;

    if (size <= 999) {
        snprintf(buf, buf_size, "%" PRId64, size);
    } else {
        base = 1024;
        for(i = 0; i < NB_SUFFIXES; i++) {
            if (size < (10 * base)) {
2788
                snprintf(buf, buf_size, "%0.1f%c",
B
bellard 已提交
2789 2790 2791 2792
                         (double)size / base,
                         suffixes[i]);
                break;
            } else if (size < (1000 * base) || i == (NB_SUFFIXES - 1)) {
2793
                snprintf(buf, buf_size, "%" PRId64 "%c",
B
bellard 已提交
2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806
                         ((size + (base >> 1)) / base),
                         suffixes[i]);
                break;
            }
            base = base * 1024;
        }
    }
    return buf;
}

char *bdrv_snapshot_dump(char *buf, int buf_size, QEMUSnapshotInfo *sn)
{
    char buf1[128], date_buf[128], clock_buf[128];
2807 2808 2809
#ifdef _WIN32
    struct tm *ptm;
#else
B
bellard 已提交
2810
    struct tm tm;
2811
#endif
B
bellard 已提交
2812 2813 2814 2815
    time_t ti;
    int64_t secs;

    if (!sn) {
2816 2817
        snprintf(buf, buf_size,
                 "%-10s%-20s%7s%20s%15s",
B
bellard 已提交
2818 2819 2820
                 "ID", "TAG", "VM SIZE", "DATE", "VM CLOCK");
    } else {
        ti = sn->date_sec;
2821 2822 2823 2824 2825
#ifdef _WIN32
        ptm = localtime(&ti);
        strftime(date_buf, sizeof(date_buf),
                 "%Y-%m-%d %H:%M:%S", ptm);
#else
B
bellard 已提交
2826 2827 2828
        localtime_r(&ti, &tm);
        strftime(date_buf, sizeof(date_buf),
                 "%Y-%m-%d %H:%M:%S", &tm);
2829
#endif
B
bellard 已提交
2830 2831 2832 2833 2834
        secs = sn->vm_clock_nsec / 1000000000;
        snprintf(clock_buf, sizeof(clock_buf),
                 "%02d:%02d:%02d.%03d",
                 (int)(secs / 3600),
                 (int)((secs / 60) % 60),
2835
                 (int)(secs % 60),
B
bellard 已提交
2836 2837
                 (int)((sn->vm_clock_nsec / 1000000) % 1000));
        snprintf(buf, buf_size,
2838
                 "%-10s%-20s%7s%20s%15s",
B
bellard 已提交
2839 2840 2841 2842 2843 2844 2845 2846
                 sn->id_str, sn->name,
                 get_human_readable_size(buf1, sizeof(buf1), sn->vm_state_size),
                 date_buf,
                 clock_buf);
    }
    return buf;
}

B
bellard 已提交
2847
/**************************************************************/
B
bellard 已提交
2848
/* async I/Os */
B
bellard 已提交
2849

2850
BlockDriverAIOCB *bdrv_aio_readv(BlockDriverState *bs, int64_t sector_num,
2851
                                 QEMUIOVector *qiov, int nb_sectors,
2852
                                 BlockDriverCompletionFunc *cb, void *opaque)
B
bellard 已提交
2853
{
2854 2855
    trace_bdrv_aio_readv(bs, sector_num, nb_sectors, opaque);

2856
    return bdrv_co_aio_rw_vector(bs, sector_num, qiov, nb_sectors,
2857
                                 cb, opaque, false);
B
bellard 已提交
2858 2859
}

2860 2861 2862
BlockDriverAIOCB *bdrv_aio_writev(BlockDriverState *bs, int64_t sector_num,
                                  QEMUIOVector *qiov, int nb_sectors,
                                  BlockDriverCompletionFunc *cb, void *opaque)
B
bellard 已提交
2863
{
2864 2865
    trace_bdrv_aio_writev(bs, sector_num, nb_sectors, opaque);

2866
    return bdrv_co_aio_rw_vector(bs, sector_num, qiov, nb_sectors,
2867
                                 cb, opaque, true);
B
bellard 已提交
2868 2869
}

K
Kevin Wolf 已提交
2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887

typedef struct MultiwriteCB {
    int error;
    int num_requests;
    int num_callbacks;
    struct {
        BlockDriverCompletionFunc *cb;
        void *opaque;
        QEMUIOVector *free_qiov;
    } callbacks[];
} MultiwriteCB;

static void multiwrite_user_cb(MultiwriteCB *mcb)
{
    int i;

    for (i = 0; i < mcb->num_callbacks; i++) {
        mcb->callbacks[i].cb(mcb->callbacks[i].opaque, mcb->error);
2888 2889 2890
        if (mcb->callbacks[i].free_qiov) {
            qemu_iovec_destroy(mcb->callbacks[i].free_qiov);
        }
2891
        g_free(mcb->callbacks[i].free_qiov);
K
Kevin Wolf 已提交
2892 2893 2894 2895 2896 2897 2898
    }
}

static void multiwrite_cb(void *opaque, int ret)
{
    MultiwriteCB *mcb = opaque;

2899 2900
    trace_multiwrite_cb(mcb, ret);

2901
    if (ret < 0 && !mcb->error) {
K
Kevin Wolf 已提交
2902 2903 2904 2905 2906
        mcb->error = ret;
    }

    mcb->num_requests--;
    if (mcb->num_requests == 0) {
2907
        multiwrite_user_cb(mcb);
2908
        g_free(mcb);
K
Kevin Wolf 已提交
2909 2910 2911 2912 2913
    }
}

static int multiwrite_req_compare(const void *a, const void *b)
{
2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926
    const BlockRequest *req1 = a, *req2 = b;

    /*
     * Note that we can't simply subtract req2->sector from req1->sector
     * here as that could overflow the return value.
     */
    if (req1->sector > req2->sector) {
        return 1;
    } else if (req1->sector < req2->sector) {
        return -1;
    } else {
        return 0;
    }
K
Kevin Wolf 已提交
2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947
}

/*
 * Takes a bunch of requests and tries to merge them. Returns the number of
 * requests that remain after merging.
 */
static int multiwrite_merge(BlockDriverState *bs, BlockRequest *reqs,
    int num_reqs, MultiwriteCB *mcb)
{
    int i, outidx;

    // Sort requests by start sector
    qsort(reqs, num_reqs, sizeof(*reqs), &multiwrite_req_compare);

    // Check if adjacent requests touch the same clusters. If so, combine them,
    // filling up gaps with zero sectors.
    outidx = 0;
    for (i = 1; i < num_reqs; i++) {
        int merge = 0;
        int64_t oldreq_last = reqs[outidx].sector + reqs[outidx].nb_sectors;

2948
        // Handle exactly sequential writes and overlapping writes.
K
Kevin Wolf 已提交
2949 2950 2951 2952
        if (reqs[i].sector <= oldreq_last) {
            merge = 1;
        }

2953 2954 2955 2956
        if (reqs[outidx].qiov->niov + reqs[i].qiov->niov + 1 > IOV_MAX) {
            merge = 0;
        }

K
Kevin Wolf 已提交
2957 2958
        if (merge) {
            size_t size;
2959
            QEMUIOVector *qiov = g_malloc0(sizeof(*qiov));
K
Kevin Wolf 已提交
2960 2961 2962 2963 2964 2965
            qemu_iovec_init(qiov,
                reqs[outidx].qiov->niov + reqs[i].qiov->niov + 1);

            // Add the first request to the merged one. If the requests are
            // overlapping, drop the last sectors of the first request.
            size = (reqs[i].sector - reqs[outidx].sector) << 9;
2966
            qemu_iovec_concat(qiov, reqs[outidx].qiov, 0, size);
K
Kevin Wolf 已提交
2967

2968 2969
            // We should need to add any zeros between the two requests
            assert (reqs[i].sector <= oldreq_last);
K
Kevin Wolf 已提交
2970 2971

            // Add the second request
2972
            qemu_iovec_concat(qiov, reqs[i].qiov, 0, reqs[i].qiov->size);
K
Kevin Wolf 已提交
2973

2974
            reqs[outidx].nb_sectors = qiov->size >> 9;
K
Kevin Wolf 已提交
2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007
            reqs[outidx].qiov = qiov;

            mcb->callbacks[i].free_qiov = reqs[outidx].qiov;
        } else {
            outidx++;
            reqs[outidx].sector     = reqs[i].sector;
            reqs[outidx].nb_sectors = reqs[i].nb_sectors;
            reqs[outidx].qiov       = reqs[i].qiov;
        }
    }

    return outidx + 1;
}

/*
 * Submit multiple AIO write requests at once.
 *
 * On success, the function returns 0 and all requests in the reqs array have
 * been submitted. In error case this function returns -1, and any of the
 * requests may or may not be submitted yet. In particular, this means that the
 * callback will be called for some of the requests, for others it won't. The
 * caller must check the error field of the BlockRequest to wait for the right
 * callbacks (if error != 0, no callback will be called).
 *
 * The implementation may modify the contents of the reqs array, e.g. to merge
 * requests. However, the fields opaque and error are left unmodified as they
 * are used to signal failure for a single request to the caller.
 */
int bdrv_aio_multiwrite(BlockDriverState *bs, BlockRequest *reqs, int num_reqs)
{
    MultiwriteCB *mcb;
    int i;

3008 3009 3010 3011 3012 3013 3014 3015
    /* don't submit writes if we don't have a medium */
    if (bs->drv == NULL) {
        for (i = 0; i < num_reqs; i++) {
            reqs[i].error = -ENOMEDIUM;
        }
        return -1;
    }

K
Kevin Wolf 已提交
3016 3017 3018 3019 3020
    if (num_reqs == 0) {
        return 0;
    }

    // Create MultiwriteCB structure
3021
    mcb = g_malloc0(sizeof(*mcb) + num_reqs * sizeof(*mcb->callbacks));
K
Kevin Wolf 已提交
3022 3023 3024 3025 3026 3027 3028 3029 3030 3031 3032
    mcb->num_requests = 0;
    mcb->num_callbacks = num_reqs;

    for (i = 0; i < num_reqs; i++) {
        mcb->callbacks[i].cb = reqs[i].cb;
        mcb->callbacks[i].opaque = reqs[i].opaque;
    }

    // Check for mergable requests
    num_reqs = multiwrite_merge(bs, reqs, num_reqs, mcb);

3033 3034
    trace_bdrv_aio_multiwrite(mcb, mcb->num_callbacks, num_reqs);

3035 3036
    /* Run the aio requests. */
    mcb->num_requests = num_reqs;
K
Kevin Wolf 已提交
3037
    for (i = 0; i < num_reqs; i++) {
3038
        bdrv_aio_writev(bs, reqs[i].sector, reqs[i].qiov,
K
Kevin Wolf 已提交
3039 3040 3041 3042 3043 3044
            reqs[i].nb_sectors, multiwrite_cb, mcb);
    }

    return 0;
}

B
bellard 已提交
3045 3046
void bdrv_aio_cancel(BlockDriverAIOCB *acb)
{
3047
    acb->pool->cancel(acb);
B
bellard 已提交
3048 3049
}

Z
Zhi Yong Wu 已提交
3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084 3085 3086 3087 3088 3089 3090 3091 3092 3093 3094 3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152 3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213
/* block I/O throttling */
static bool bdrv_exceed_bps_limits(BlockDriverState *bs, int nb_sectors,
                 bool is_write, double elapsed_time, uint64_t *wait)
{
    uint64_t bps_limit = 0;
    double   bytes_limit, bytes_base, bytes_res;
    double   slice_time, wait_time;

    if (bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL]) {
        bps_limit = bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL];
    } else if (bs->io_limits.bps[is_write]) {
        bps_limit = bs->io_limits.bps[is_write];
    } else {
        if (wait) {
            *wait = 0;
        }

        return false;
    }

    slice_time = bs->slice_end - bs->slice_start;
    slice_time /= (NANOSECONDS_PER_SECOND);
    bytes_limit = bps_limit * slice_time;
    bytes_base  = bs->nr_bytes[is_write] - bs->io_base.bytes[is_write];
    if (bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL]) {
        bytes_base += bs->nr_bytes[!is_write] - bs->io_base.bytes[!is_write];
    }

    /* bytes_base: the bytes of data which have been read/written; and
     *             it is obtained from the history statistic info.
     * bytes_res: the remaining bytes of data which need to be read/written.
     * (bytes_base + bytes_res) / bps_limit: used to calcuate
     *             the total time for completing reading/writting all data.
     */
    bytes_res   = (unsigned) nb_sectors * BDRV_SECTOR_SIZE;

    if (bytes_base + bytes_res <= bytes_limit) {
        if (wait) {
            *wait = 0;
        }

        return false;
    }

    /* Calc approx time to dispatch */
    wait_time = (bytes_base + bytes_res) / bps_limit - elapsed_time;

    /* When the I/O rate at runtime exceeds the limits,
     * bs->slice_end need to be extended in order that the current statistic
     * info can be kept until the timer fire, so it is increased and tuned
     * based on the result of experiment.
     */
    bs->slice_time = wait_time * BLOCK_IO_SLICE_TIME * 10;
    bs->slice_end += bs->slice_time - 3 * BLOCK_IO_SLICE_TIME;
    if (wait) {
        *wait = wait_time * BLOCK_IO_SLICE_TIME * 10;
    }

    return true;
}

static bool bdrv_exceed_iops_limits(BlockDriverState *bs, bool is_write,
                             double elapsed_time, uint64_t *wait)
{
    uint64_t iops_limit = 0;
    double   ios_limit, ios_base;
    double   slice_time, wait_time;

    if (bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL]) {
        iops_limit = bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL];
    } else if (bs->io_limits.iops[is_write]) {
        iops_limit = bs->io_limits.iops[is_write];
    } else {
        if (wait) {
            *wait = 0;
        }

        return false;
    }

    slice_time = bs->slice_end - bs->slice_start;
    slice_time /= (NANOSECONDS_PER_SECOND);
    ios_limit  = iops_limit * slice_time;
    ios_base   = bs->nr_ops[is_write] - bs->io_base.ios[is_write];
    if (bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL]) {
        ios_base += bs->nr_ops[!is_write] - bs->io_base.ios[!is_write];
    }

    if (ios_base + 1 <= ios_limit) {
        if (wait) {
            *wait = 0;
        }

        return false;
    }

    /* Calc approx time to dispatch */
    wait_time = (ios_base + 1) / iops_limit;
    if (wait_time > elapsed_time) {
        wait_time = wait_time - elapsed_time;
    } else {
        wait_time = 0;
    }

    bs->slice_time = wait_time * BLOCK_IO_SLICE_TIME * 10;
    bs->slice_end += bs->slice_time - 3 * BLOCK_IO_SLICE_TIME;
    if (wait) {
        *wait = wait_time * BLOCK_IO_SLICE_TIME * 10;
    }

    return true;
}

static bool bdrv_exceed_io_limits(BlockDriverState *bs, int nb_sectors,
                           bool is_write, int64_t *wait)
{
    int64_t  now, max_wait;
    uint64_t bps_wait = 0, iops_wait = 0;
    double   elapsed_time;
    int      bps_ret, iops_ret;

    now = qemu_get_clock_ns(vm_clock);
    if ((bs->slice_start < now)
        && (bs->slice_end > now)) {
        bs->slice_end = now + bs->slice_time;
    } else {
        bs->slice_time  =  5 * BLOCK_IO_SLICE_TIME;
        bs->slice_start = now;
        bs->slice_end   = now + bs->slice_time;

        bs->io_base.bytes[is_write]  = bs->nr_bytes[is_write];
        bs->io_base.bytes[!is_write] = bs->nr_bytes[!is_write];

        bs->io_base.ios[is_write]    = bs->nr_ops[is_write];
        bs->io_base.ios[!is_write]   = bs->nr_ops[!is_write];
    }

    elapsed_time  = now - bs->slice_start;
    elapsed_time  /= (NANOSECONDS_PER_SECOND);

    bps_ret  = bdrv_exceed_bps_limits(bs, nb_sectors,
                                      is_write, elapsed_time, &bps_wait);
    iops_ret = bdrv_exceed_iops_limits(bs, is_write,
                                      elapsed_time, &iops_wait);
    if (bps_ret || iops_ret) {
        max_wait = bps_wait > iops_wait ? bps_wait : iops_wait;
        if (wait) {
            *wait = max_wait;
        }

        now = qemu_get_clock_ns(vm_clock);
        if (bs->slice_end < now + max_wait) {
            bs->slice_end = now + max_wait;
        }

        return true;
    }

    if (wait) {
        *wait = 0;
    }

    return false;
}
3214

B
bellard 已提交
3215 3216 3217
/**************************************************************/
/* async block device emulation */

3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229
typedef struct BlockDriverAIOCBSync {
    BlockDriverAIOCB common;
    QEMUBH *bh;
    int ret;
    /* vector translation state */
    QEMUIOVector *qiov;
    uint8_t *bounce;
    int is_write;
} BlockDriverAIOCBSync;

static void bdrv_aio_cancel_em(BlockDriverAIOCB *blockacb)
{
3230 3231
    BlockDriverAIOCBSync *acb =
        container_of(blockacb, BlockDriverAIOCBSync, common);
3232
    qemu_bh_delete(acb->bh);
A
Avi Kivity 已提交
3233
    acb->bh = NULL;
3234 3235 3236 3237 3238 3239 3240 3241
    qemu_aio_release(acb);
}

static AIOPool bdrv_em_aio_pool = {
    .aiocb_size         = sizeof(BlockDriverAIOCBSync),
    .cancel             = bdrv_aio_cancel_em,
};

3242
static void bdrv_aio_bh_cb(void *opaque)
B
bellard 已提交
3243
{
3244
    BlockDriverAIOCBSync *acb = opaque;
3245 3246

    if (!acb->is_write)
3247
        qemu_iovec_from_buf(acb->qiov, 0, acb->bounce, acb->qiov->size);
3248
    qemu_vfree(acb->bounce);
3249
    acb->common.cb(acb->common.opaque, acb->ret);
3250
    qemu_bh_delete(acb->bh);
A
Avi Kivity 已提交
3251
    acb->bh = NULL;
3252
    qemu_aio_release(acb);
B
bellard 已提交
3253
}
3254

3255 3256 3257 3258 3259 3260 3261 3262
static BlockDriverAIOCB *bdrv_aio_rw_vector(BlockDriverState *bs,
                                            int64_t sector_num,
                                            QEMUIOVector *qiov,
                                            int nb_sectors,
                                            BlockDriverCompletionFunc *cb,
                                            void *opaque,
                                            int is_write)

B
bellard 已提交
3263
{
3264 3265
    BlockDriverAIOCBSync *acb;

3266
    acb = qemu_aio_get(&bdrv_em_aio_pool, bs, cb, opaque);
3267 3268
    acb->is_write = is_write;
    acb->qiov = qiov;
3269
    acb->bounce = qemu_blockalign(bs, qiov->size);
3270
    acb->bh = qemu_bh_new(bdrv_aio_bh_cb, acb);
3271 3272

    if (is_write) {
3273
        qemu_iovec_to_buf(acb->qiov, 0, acb->bounce, qiov->size);
3274
        acb->ret = bs->drv->bdrv_write(bs, sector_num, acb->bounce, nb_sectors);
3275
    } else {
3276
        acb->ret = bs->drv->bdrv_read(bs, sector_num, acb->bounce, nb_sectors);
3277 3278
    }

3279
    qemu_bh_schedule(acb->bh);
3280

3281
    return &acb->common;
3282 3283
}

3284 3285
static BlockDriverAIOCB *bdrv_aio_readv_em(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
3286
        BlockDriverCompletionFunc *cb, void *opaque)
3287
{
3288 3289
    return bdrv_aio_rw_vector(bs, sector_num, qiov, nb_sectors, cb, opaque, 0);
}
B
bellard 已提交
3290

3291 3292 3293 3294 3295
static BlockDriverAIOCB *bdrv_aio_writev_em(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
        BlockDriverCompletionFunc *cb, void *opaque)
{
    return bdrv_aio_rw_vector(bs, sector_num, qiov, nb_sectors, cb, opaque, 1);
3296 3297
}

3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315

typedef struct BlockDriverAIOCBCoroutine {
    BlockDriverAIOCB common;
    BlockRequest req;
    bool is_write;
    QEMUBH* bh;
} BlockDriverAIOCBCoroutine;

static void bdrv_aio_co_cancel_em(BlockDriverAIOCB *blockacb)
{
    qemu_aio_flush();
}

static AIOPool bdrv_em_co_aio_pool = {
    .aiocb_size         = sizeof(BlockDriverAIOCBCoroutine),
    .cancel             = bdrv_aio_co_cancel_em,
};

P
Paolo Bonzini 已提交
3316
static void bdrv_co_em_bh(void *opaque)
3317 3318 3319 3320 3321 3322 3323 3324
{
    BlockDriverAIOCBCoroutine *acb = opaque;

    acb->common.cb(acb->common.opaque, acb->req.error);
    qemu_bh_delete(acb->bh);
    qemu_aio_release(acb);
}

3325 3326 3327 3328 3329 3330 3331 3332
/* Invoke bdrv_co_do_readv/bdrv_co_do_writev */
static void coroutine_fn bdrv_co_do_rw(void *opaque)
{
    BlockDriverAIOCBCoroutine *acb = opaque;
    BlockDriverState *bs = acb->common.bs;

    if (!acb->is_write) {
        acb->req.error = bdrv_co_do_readv(bs, acb->req.sector,
3333
            acb->req.nb_sectors, acb->req.qiov, 0);
3334 3335
    } else {
        acb->req.error = bdrv_co_do_writev(bs, acb->req.sector,
3336
            acb->req.nb_sectors, acb->req.qiov, 0);
3337 3338
    }

P
Paolo Bonzini 已提交
3339
    acb->bh = qemu_bh_new(bdrv_co_em_bh, acb);
3340 3341 3342
    qemu_bh_schedule(acb->bh);
}

3343 3344 3345 3346 3347 3348
static BlockDriverAIOCB *bdrv_co_aio_rw_vector(BlockDriverState *bs,
                                               int64_t sector_num,
                                               QEMUIOVector *qiov,
                                               int nb_sectors,
                                               BlockDriverCompletionFunc *cb,
                                               void *opaque,
3349
                                               bool is_write)
3350 3351 3352 3353 3354 3355 3356 3357 3358 3359
{
    Coroutine *co;
    BlockDriverAIOCBCoroutine *acb;

    acb = qemu_aio_get(&bdrv_em_co_aio_pool, bs, cb, opaque);
    acb->req.sector = sector_num;
    acb->req.nb_sectors = nb_sectors;
    acb->req.qiov = qiov;
    acb->is_write = is_write;

3360
    co = qemu_coroutine_create(bdrv_co_do_rw);
3361 3362 3363 3364 3365
    qemu_coroutine_enter(co, acb);

    return &acb->common;
}

P
Paolo Bonzini 已提交
3366
static void coroutine_fn bdrv_aio_flush_co_entry(void *opaque)
3367
{
P
Paolo Bonzini 已提交
3368 3369
    BlockDriverAIOCBCoroutine *acb = opaque;
    BlockDriverState *bs = acb->common.bs;
3370

P
Paolo Bonzini 已提交
3371 3372
    acb->req.error = bdrv_co_flush(bs);
    acb->bh = qemu_bh_new(bdrv_co_em_bh, acb);
3373 3374 3375
    qemu_bh_schedule(acb->bh);
}

P
Paolo Bonzini 已提交
3376
BlockDriverAIOCB *bdrv_aio_flush(BlockDriverState *bs,
3377 3378
        BlockDriverCompletionFunc *cb, void *opaque)
{
P
Paolo Bonzini 已提交
3379
    trace_bdrv_aio_flush(bs, opaque);
3380

P
Paolo Bonzini 已提交
3381 3382
    Coroutine *co;
    BlockDriverAIOCBCoroutine *acb;
3383

P
Paolo Bonzini 已提交
3384 3385 3386
    acb = qemu_aio_get(&bdrv_em_co_aio_pool, bs, cb, opaque);
    co = qemu_coroutine_create(bdrv_aio_flush_co_entry);
    qemu_coroutine_enter(co, acb);
3387 3388 3389 3390

    return &acb->common;
}

3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418
static void coroutine_fn bdrv_aio_discard_co_entry(void *opaque)
{
    BlockDriverAIOCBCoroutine *acb = opaque;
    BlockDriverState *bs = acb->common.bs;

    acb->req.error = bdrv_co_discard(bs, acb->req.sector, acb->req.nb_sectors);
    acb->bh = qemu_bh_new(bdrv_co_em_bh, acb);
    qemu_bh_schedule(acb->bh);
}

BlockDriverAIOCB *bdrv_aio_discard(BlockDriverState *bs,
        int64_t sector_num, int nb_sectors,
        BlockDriverCompletionFunc *cb, void *opaque)
{
    Coroutine *co;
    BlockDriverAIOCBCoroutine *acb;

    trace_bdrv_aio_discard(bs, sector_num, nb_sectors, opaque);

    acb = qemu_aio_get(&bdrv_em_co_aio_pool, bs, cb, opaque);
    acb->req.sector = sector_num;
    acb->req.nb_sectors = nb_sectors;
    co = qemu_coroutine_create(bdrv_aio_discard_co_entry);
    qemu_coroutine_enter(co, acb);

    return &acb->common;
}

B
bellard 已提交
3419 3420
void bdrv_init(void)
{
3421
    module_call_init(MODULE_INIT_BLOCK);
B
bellard 已提交
3422
}
3423

3424 3425 3426 3427 3428 3429
void bdrv_init_with_whitelist(void)
{
    use_bdrv_whitelist = 1;
    bdrv_init();
}

3430 3431
void *qemu_aio_get(AIOPool *pool, BlockDriverState *bs,
                   BlockDriverCompletionFunc *cb, void *opaque)
3432 3433 3434
{
    BlockDriverAIOCB *acb;

3435 3436 3437
    if (pool->free_aiocb) {
        acb = pool->free_aiocb;
        pool->free_aiocb = acb->next;
3438
    } else {
3439
        acb = g_malloc0(pool->aiocb_size);
3440
        acb->pool = pool;
3441 3442 3443 3444 3445 3446 3447 3448 3449
    }
    acb->bs = bs;
    acb->cb = cb;
    acb->opaque = opaque;
    return acb;
}

void qemu_aio_release(void *p)
{
3450 3451 3452 3453
    BlockDriverAIOCB *acb = (BlockDriverAIOCB *)p;
    AIOPool *pool = acb->pool;
    acb->next = pool->free_aiocb;
    pool->free_aiocb = acb;
3454
}
B
bellard 已提交
3455

3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481
/**************************************************************/
/* Coroutine block device emulation */

typedef struct CoroutineIOCompletion {
    Coroutine *coroutine;
    int ret;
} CoroutineIOCompletion;

static void bdrv_co_io_em_complete(void *opaque, int ret)
{
    CoroutineIOCompletion *co = opaque;

    co->ret = ret;
    qemu_coroutine_enter(co->coroutine, NULL);
}

static int coroutine_fn bdrv_co_io_em(BlockDriverState *bs, int64_t sector_num,
                                      int nb_sectors, QEMUIOVector *iov,
                                      bool is_write)
{
    CoroutineIOCompletion co = {
        .coroutine = qemu_coroutine_self(),
    };
    BlockDriverAIOCB *acb;

    if (is_write) {
3482 3483
        acb = bs->drv->bdrv_aio_writev(bs, sector_num, iov, nb_sectors,
                                       bdrv_co_io_em_complete, &co);
3484
    } else {
3485 3486
        acb = bs->drv->bdrv_aio_readv(bs, sector_num, iov, nb_sectors,
                                      bdrv_co_io_em_complete, &co);
3487 3488
    }

3489
    trace_bdrv_co_io_em(bs, sector_num, nb_sectors, is_write, acb);
3490 3491 3492 3493 3494 3495 3496 3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511
    if (!acb) {
        return -EIO;
    }
    qemu_coroutine_yield();

    return co.ret;
}

static int coroutine_fn bdrv_co_readv_em(BlockDriverState *bs,
                                         int64_t sector_num, int nb_sectors,
                                         QEMUIOVector *iov)
{
    return bdrv_co_io_em(bs, sector_num, nb_sectors, iov, false);
}

static int coroutine_fn bdrv_co_writev_em(BlockDriverState *bs,
                                         int64_t sector_num, int nb_sectors,
                                         QEMUIOVector *iov)
{
    return bdrv_co_io_em(bs, sector_num, nb_sectors, iov, true);
}

P
Paolo Bonzini 已提交
3512
static void coroutine_fn bdrv_flush_co_entry(void *opaque)
3513
{
P
Paolo Bonzini 已提交
3514 3515 3516 3517 3518 3519 3520
    RwCo *rwco = opaque;

    rwco->ret = bdrv_co_flush(rwco->bs);
}

int coroutine_fn bdrv_co_flush(BlockDriverState *bs)
{
K
Kevin Wolf 已提交
3521 3522
    int ret;

3523
    if (!bs || !bdrv_is_inserted(bs) || bdrv_is_read_only(bs)) {
P
Paolo Bonzini 已提交
3524
        return 0;
K
Kevin Wolf 已提交
3525 3526
    }

3527
    /* Write back cached data to the OS even with cache=unsafe */
K
Kevin Wolf 已提交
3528 3529 3530 3531 3532 3533 3534
    if (bs->drv->bdrv_co_flush_to_os) {
        ret = bs->drv->bdrv_co_flush_to_os(bs);
        if (ret < 0) {
            return ret;
        }
    }

3535 3536 3537 3538 3539
    /* But don't actually force it to the disk with cache=unsafe */
    if (bs->open_flags & BDRV_O_NO_FLUSH) {
        return 0;
    }

K
Kevin Wolf 已提交
3540
    if (bs->drv->bdrv_co_flush_to_disk) {
3541
        ret = bs->drv->bdrv_co_flush_to_disk(bs);
P
Paolo Bonzini 已提交
3542 3543 3544 3545 3546 3547 3548 3549
    } else if (bs->drv->bdrv_aio_flush) {
        BlockDriverAIOCB *acb;
        CoroutineIOCompletion co = {
            .coroutine = qemu_coroutine_self(),
        };

        acb = bs->drv->bdrv_aio_flush(bs, bdrv_co_io_em_complete, &co);
        if (acb == NULL) {
3550
            ret = -EIO;
P
Paolo Bonzini 已提交
3551 3552
        } else {
            qemu_coroutine_yield();
3553
            ret = co.ret;
P
Paolo Bonzini 已提交
3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566
        }
    } else {
        /*
         * Some block drivers always operate in either writethrough or unsafe
         * mode and don't support bdrv_flush therefore. Usually qemu doesn't
         * know how the server works (because the behaviour is hardcoded or
         * depends on server-side configuration), so we can't ensure that
         * everything is safe on disk. Returning an error doesn't work because
         * that would break guests even if the server operates in writethrough
         * mode.
         *
         * Let's hope the user knows what he's doing.
         */
3567
        ret = 0;
P
Paolo Bonzini 已提交
3568
    }
3569 3570 3571 3572 3573 3574 3575 3576
    if (ret < 0) {
        return ret;
    }

    /* Now flush the underlying protocol.  It will also have BDRV_O_NO_FLUSH
     * in the case of cache=unsafe, so there are no useless flushes.
     */
    return bdrv_co_flush(bs->file);
P
Paolo Bonzini 已提交
3577 3578
}

3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594
void bdrv_invalidate_cache(BlockDriverState *bs)
{
    if (bs->drv && bs->drv->bdrv_invalidate_cache) {
        bs->drv->bdrv_invalidate_cache(bs);
    }
}

void bdrv_invalidate_cache_all(void)
{
    BlockDriverState *bs;

    QTAILQ_FOREACH(bs, &bdrv_states, list) {
        bdrv_invalidate_cache(bs);
    }
}

3595 3596 3597 3598 3599 3600 3601 3602 3603
void bdrv_clear_incoming_migration_all(void)
{
    BlockDriverState *bs;

    QTAILQ_FOREACH(bs, &bdrv_states, list) {
        bs->open_flags = bs->open_flags & ~(BDRV_O_INCOMING);
    }
}

P
Paolo Bonzini 已提交
3604 3605 3606 3607 3608 3609
int bdrv_flush(BlockDriverState *bs)
{
    Coroutine *co;
    RwCo rwco = {
        .bs = bs,
        .ret = NOT_DONE,
3610 3611
    };

P
Paolo Bonzini 已提交
3612 3613 3614 3615 3616 3617 3618 3619 3620
    if (qemu_in_coroutine()) {
        /* Fast-path if already in coroutine context */
        bdrv_flush_co_entry(&rwco);
    } else {
        co = qemu_coroutine_create(bdrv_flush_co_entry);
        qemu_coroutine_enter(co, &rwco);
        while (rwco.ret == NOT_DONE) {
            qemu_aio_wait();
        }
3621
    }
P
Paolo Bonzini 已提交
3622 3623

    return rwco.ret;
3624 3625
}

3626 3627 3628 3629 3630 3631 3632 3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672 3673 3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686
static void coroutine_fn bdrv_discard_co_entry(void *opaque)
{
    RwCo *rwco = opaque;

    rwco->ret = bdrv_co_discard(rwco->bs, rwco->sector_num, rwco->nb_sectors);
}

int coroutine_fn bdrv_co_discard(BlockDriverState *bs, int64_t sector_num,
                                 int nb_sectors)
{
    if (!bs->drv) {
        return -ENOMEDIUM;
    } else if (bdrv_check_request(bs, sector_num, nb_sectors)) {
        return -EIO;
    } else if (bs->read_only) {
        return -EROFS;
    } else if (bs->drv->bdrv_co_discard) {
        return bs->drv->bdrv_co_discard(bs, sector_num, nb_sectors);
    } else if (bs->drv->bdrv_aio_discard) {
        BlockDriverAIOCB *acb;
        CoroutineIOCompletion co = {
            .coroutine = qemu_coroutine_self(),
        };

        acb = bs->drv->bdrv_aio_discard(bs, sector_num, nb_sectors,
                                        bdrv_co_io_em_complete, &co);
        if (acb == NULL) {
            return -EIO;
        } else {
            qemu_coroutine_yield();
            return co.ret;
        }
    } else {
        return 0;
    }
}

int bdrv_discard(BlockDriverState *bs, int64_t sector_num, int nb_sectors)
{
    Coroutine *co;
    RwCo rwco = {
        .bs = bs,
        .sector_num = sector_num,
        .nb_sectors = nb_sectors,
        .ret = NOT_DONE,
    };

    if (qemu_in_coroutine()) {
        /* Fast-path if already in coroutine context */
        bdrv_discard_co_entry(&rwco);
    } else {
        co = qemu_coroutine_create(bdrv_discard_co_entry);
        qemu_coroutine_enter(co, &rwco);
        while (rwco.ret == NOT_DONE) {
            qemu_aio_wait();
        }
    }

    return rwco.ret;
}

B
bellard 已提交
3687 3688 3689 3690 3691 3692 3693 3694 3695
/**************************************************************/
/* removable device support */

/**
 * Return TRUE if the media is present
 */
int bdrv_is_inserted(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;
3696

B
bellard 已提交
3697 3698 3699
    if (!drv)
        return 0;
    if (!drv->bdrv_is_inserted)
3700 3701
        return 1;
    return drv->bdrv_is_inserted(bs);
B
bellard 已提交
3702 3703 3704
}

/**
3705 3706
 * Return whether the media changed since the last call to this
 * function, or -ENOTSUP if we don't know.  Most drivers don't know.
B
bellard 已提交
3707 3708 3709 3710 3711
 */
int bdrv_media_changed(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;

3712 3713 3714 3715
    if (drv && drv->bdrv_media_changed) {
        return drv->bdrv_media_changed(bs);
    }
    return -ENOTSUP;
B
bellard 已提交
3716 3717 3718 3719 3720
}

/**
 * If eject_flag is TRUE, eject the media. Otherwise, close the tray
 */
3721
void bdrv_eject(BlockDriverState *bs, bool eject_flag)
B
bellard 已提交
3722 3723 3724
{
    BlockDriver *drv = bs->drv;

3725 3726
    if (drv && drv->bdrv_eject) {
        drv->bdrv_eject(bs, eject_flag);
B
bellard 已提交
3727
    }
3728 3729 3730 3731

    if (bs->device_name[0] != '\0') {
        bdrv_emit_qmp_eject_event(bs, eject_flag);
    }
B
bellard 已提交
3732 3733 3734 3735 3736 3737
}

/**
 * Lock or unlock the media (if it is locked, the user won't be able
 * to eject it manually).
 */
3738
void bdrv_lock_medium(BlockDriverState *bs, bool locked)
B
bellard 已提交
3739 3740 3741
{
    BlockDriver *drv = bs->drv;

3742
    trace_bdrv_lock_medium(bs, locked);
S
Stefan Hajnoczi 已提交
3743

3744 3745
    if (drv && drv->bdrv_lock_medium) {
        drv->bdrv_lock_medium(bs, locked);
B
bellard 已提交
3746 3747
    }
}
3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758

/* needed for generic scsi interface */

int bdrv_ioctl(BlockDriverState *bs, unsigned long int req, void *buf)
{
    BlockDriver *drv = bs->drv;

    if (drv && drv->bdrv_ioctl)
        return drv->bdrv_ioctl(bs, req, buf);
    return -ENOTSUP;
}
3759

3760 3761 3762
BlockDriverAIOCB *bdrv_aio_ioctl(BlockDriverState *bs,
        unsigned long int req, void *buf,
        BlockDriverCompletionFunc *cb, void *opaque)
3763
{
3764
    BlockDriver *drv = bs->drv;
3765

3766 3767 3768
    if (drv && drv->bdrv_aio_ioctl)
        return drv->bdrv_aio_ioctl(bs, req, buf, cb, opaque);
    return NULL;
3769
}
3770

3771 3772 3773 3774
void bdrv_set_buffer_alignment(BlockDriverState *bs, int align)
{
    bs->buffer_alignment = align;
}
3775

3776 3777 3778 3779
void *qemu_blockalign(BlockDriverState *bs, size_t size)
{
    return qemu_memalign((bs && bs->buffer_alignment) ? bs->buffer_alignment : 512, size);
}
3780 3781 3782 3783

void bdrv_set_dirty_tracking(BlockDriverState *bs, int enable)
{
    int64_t bitmap_size;
3784

3785
    bs->dirty_count = 0;
3786
    if (enable) {
3787 3788
        if (!bs->dirty_bitmap) {
            bitmap_size = (bdrv_getlength(bs) >> BDRV_SECTOR_BITS) +
3789 3790
                    BDRV_SECTORS_PER_DIRTY_CHUNK * BITS_PER_LONG - 1;
            bitmap_size /= BDRV_SECTORS_PER_DIRTY_CHUNK * BITS_PER_LONG;
3791

3792
            bs->dirty_bitmap = g_new0(unsigned long, bitmap_size);
3793
        }
3794
    } else {
3795
        if (bs->dirty_bitmap) {
3796
            g_free(bs->dirty_bitmap);
3797
            bs->dirty_bitmap = NULL;
3798
        }
3799 3800 3801 3802 3803
    }
}

int bdrv_get_dirty(BlockDriverState *bs, int64_t sector)
{
J
Jan Kiszka 已提交
3804
    int64_t chunk = sector / (int64_t)BDRV_SECTORS_PER_DIRTY_CHUNK;
3805

3806 3807
    if (bs->dirty_bitmap &&
        (sector << BDRV_SECTOR_BITS) < bdrv_getlength(bs)) {
3808 3809
        return !!(bs->dirty_bitmap[chunk / (sizeof(unsigned long) * 8)] &
            (1UL << (chunk % (sizeof(unsigned long) * 8))));
3810 3811 3812 3813 3814
    } else {
        return 0;
    }
}

3815 3816
void bdrv_reset_dirty(BlockDriverState *bs, int64_t cur_sector,
                      int nr_sectors)
3817 3818 3819
{
    set_dirty_bitmap(bs, cur_sector, nr_sectors, 0);
}
3820 3821 3822 3823 3824

int64_t bdrv_get_dirty_count(BlockDriverState *bs)
{
    return bs->dirty_count;
}
J
Jes Sorensen 已提交
3825

3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836
void bdrv_set_in_use(BlockDriverState *bs, int in_use)
{
    assert(bs->in_use != in_use);
    bs->in_use = in_use;
}

int bdrv_in_use(BlockDriverState *bs)
{
    return bs->in_use;
}

3837 3838
void bdrv_iostatus_enable(BlockDriverState *bs)
{
3839
    bs->iostatus_enabled = true;
3840
    bs->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
3841 3842 3843 3844 3845 3846
}

/* The I/O status is only enabled if the drive explicitly
 * enables it _and_ the VM is configured to stop on errors */
bool bdrv_iostatus_is_enabled(const BlockDriverState *bs)
{
3847
    return (bs->iostatus_enabled &&
3848 3849 3850 3851 3852 3853 3854
           (bs->on_write_error == BLOCK_ERR_STOP_ENOSPC ||
            bs->on_write_error == BLOCK_ERR_STOP_ANY    ||
            bs->on_read_error == BLOCK_ERR_STOP_ANY));
}

void bdrv_iostatus_disable(BlockDriverState *bs)
{
3855
    bs->iostatus_enabled = false;
3856 3857 3858 3859 3860
}

void bdrv_iostatus_reset(BlockDriverState *bs)
{
    if (bdrv_iostatus_is_enabled(bs)) {
3861
        bs->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
3862 3863 3864 3865 3866 3867 3868 3869
    }
}

/* XXX: Today this is set by device models because it makes the implementation
   quite simple. However, the block layer knows about the error, so it's
   possible to implement this without device models being involved */
void bdrv_iostatus_set_err(BlockDriverState *bs, int error)
{
3870 3871
    if (bdrv_iostatus_is_enabled(bs) &&
        bs->iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
3872
        assert(error >= 0);
3873 3874
        bs->iostatus = error == ENOSPC ? BLOCK_DEVICE_IO_STATUS_NOSPACE :
                                         BLOCK_DEVICE_IO_STATUS_FAILED;
3875 3876 3877
    }
}

3878 3879 3880 3881 3882 3883 3884
void
bdrv_acct_start(BlockDriverState *bs, BlockAcctCookie *cookie, int64_t bytes,
        enum BlockAcctType type)
{
    assert(type < BDRV_MAX_IOTYPE);

    cookie->bytes = bytes;
C
Christoph Hellwig 已提交
3885
    cookie->start_time_ns = get_clock();
3886 3887 3888 3889 3890 3891 3892 3893 3894 3895
    cookie->type = type;
}

void
bdrv_acct_done(BlockDriverState *bs, BlockAcctCookie *cookie)
{
    assert(cookie->type < BDRV_MAX_IOTYPE);

    bs->nr_bytes[cookie->type] += cookie->bytes;
    bs->nr_ops[cookie->type]++;
C
Christoph Hellwig 已提交
3896
    bs->total_time_ns[cookie->type] += get_clock() - cookie->start_time_ns;
3897 3898
}

J
Jes Sorensen 已提交
3899 3900 3901 3902 3903
int bdrv_img_create(const char *filename, const char *fmt,
                    const char *base_filename, const char *base_fmt,
                    char *options, uint64_t img_size, int flags)
{
    QEMUOptionParameter *param = NULL, *create_options = NULL;
K
Kevin Wolf 已提交
3904
    QEMUOptionParameter *backing_fmt, *backing_file, *size;
J
Jes Sorensen 已提交
3905 3906
    BlockDriverState *bs = NULL;
    BlockDriver *drv, *proto_drv;
3907
    BlockDriver *backing_drv = NULL;
J
Jes Sorensen 已提交
3908 3909 3910 3911 3912 3913
    int ret = 0;

    /* Find driver and parse its options */
    drv = bdrv_find_format(fmt);
    if (!drv) {
        error_report("Unknown file format '%s'", fmt);
3914
        ret = -EINVAL;
J
Jes Sorensen 已提交
3915 3916 3917 3918 3919 3920
        goto out;
    }

    proto_drv = bdrv_find_protocol(filename);
    if (!proto_drv) {
        error_report("Unknown protocol '%s'", filename);
3921
        ret = -EINVAL;
J
Jes Sorensen 已提交
3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939
        goto out;
    }

    create_options = append_option_parameters(create_options,
                                              drv->create_options);
    create_options = append_option_parameters(create_options,
                                              proto_drv->create_options);

    /* Create parameter list with default values */
    param = parse_option_parameters("", create_options, param);

    set_option_parameter_int(param, BLOCK_OPT_SIZE, img_size);

    /* Parse -o options */
    if (options) {
        param = parse_option_parameters(options, create_options, param);
        if (param == NULL) {
            error_report("Invalid options for file format '%s'.", fmt);
3940
            ret = -EINVAL;
J
Jes Sorensen 已提交
3941 3942 3943 3944 3945 3946 3947 3948 3949
            goto out;
        }
    }

    if (base_filename) {
        if (set_option_parameter(param, BLOCK_OPT_BACKING_FILE,
                                 base_filename)) {
            error_report("Backing file not supported for file format '%s'",
                         fmt);
3950
            ret = -EINVAL;
J
Jes Sorensen 已提交
3951 3952 3953 3954 3955 3956 3957 3958
            goto out;
        }
    }

    if (base_fmt) {
        if (set_option_parameter(param, BLOCK_OPT_BACKING_FMT, base_fmt)) {
            error_report("Backing file format not supported for file "
                         "format '%s'", fmt);
3959
            ret = -EINVAL;
J
Jes Sorensen 已提交
3960 3961 3962 3963
            goto out;
        }
    }

3964 3965 3966 3967 3968
    backing_file = get_option_parameter(param, BLOCK_OPT_BACKING_FILE);
    if (backing_file && backing_file->value.s) {
        if (!strcmp(filename, backing_file->value.s)) {
            error_report("Error: Trying to create an image with the "
                         "same filename as the backing file");
3969
            ret = -EINVAL;
3970 3971 3972 3973
            goto out;
        }
    }

J
Jes Sorensen 已提交
3974 3975
    backing_fmt = get_option_parameter(param, BLOCK_OPT_BACKING_FMT);
    if (backing_fmt && backing_fmt->value.s) {
3976 3977
        backing_drv = bdrv_find_format(backing_fmt->value.s);
        if (!backing_drv) {
J
Jes Sorensen 已提交
3978 3979
            error_report("Unknown backing file format '%s'",
                         backing_fmt->value.s);
3980
            ret = -EINVAL;
J
Jes Sorensen 已提交
3981 3982 3983 3984 3985 3986
            goto out;
        }
    }

    // The size for the image must always be specified, with one exception:
    // If we are using a backing file, we can obtain the size from there
K
Kevin Wolf 已提交
3987 3988
    size = get_option_parameter(param, BLOCK_OPT_SIZE);
    if (size && size->value.n == -1) {
J
Jes Sorensen 已提交
3989 3990 3991
        if (backing_file && backing_file->value.s) {
            uint64_t size;
            char buf[32];
3992 3993 3994 3995 3996
            int back_flags;

            /* backing files always opened read-only */
            back_flags =
                flags & ~(BDRV_O_RDWR | BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING);
J
Jes Sorensen 已提交
3997 3998 3999

            bs = bdrv_new("");

4000
            ret = bdrv_open(bs, backing_file->value.s, back_flags, backing_drv);
J
Jes Sorensen 已提交
4001
            if (ret < 0) {
4002
                error_report("Could not open '%s'", backing_file->value.s);
J
Jes Sorensen 已提交
4003 4004 4005 4006 4007 4008 4009 4010 4011
                goto out;
            }
            bdrv_get_geometry(bs, &size);
            size *= 512;

            snprintf(buf, sizeof(buf), "%" PRId64, size);
            set_option_parameter(param, BLOCK_OPT_SIZE, buf);
        } else {
            error_report("Image creation needs a size parameter");
4012
            ret = -EINVAL;
J
Jes Sorensen 已提交
4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042
            goto out;
        }
    }

    printf("Formatting '%s', fmt=%s ", filename, fmt);
    print_option_parameters(param);
    puts("");

    ret = bdrv_create(drv, filename, param);

    if (ret < 0) {
        if (ret == -ENOTSUP) {
            error_report("Formatting or formatting option not supported for "
                         "file format '%s'", fmt);
        } else if (ret == -EFBIG) {
            error_report("The image size is too large for file format '%s'",
                         fmt);
        } else {
            error_report("%s: error while creating %s: %s", filename, fmt,
                         strerror(-ret));
        }
    }

out:
    free_option_parameters(create_options);
    free_option_parameters(param);

    if (bs) {
        bdrv_delete(bs);
    }
4043 4044

    return ret;
J
Jes Sorensen 已提交
4045
}
4046 4047

void *block_job_create(const BlockJobType *job_type, BlockDriverState *bs,
4048 4049
                       int64_t speed, BlockDriverCompletionFunc *cb,
                       void *opaque, Error **errp)
4050 4051 4052 4053
{
    BlockJob *job;

    if (bs->job || bdrv_in_use(bs)) {
4054
        error_set(errp, QERR_DEVICE_IN_USE, bdrv_get_device_name(bs));
4055 4056 4057 4058 4059 4060 4061 4062 4063
        return NULL;
    }
    bdrv_set_in_use(bs, 1);

    job = g_malloc0(job_type->instance_size);
    job->job_type      = job_type;
    job->bs            = bs;
    job->cb            = cb;
    job->opaque        = opaque;
P
Paolo Bonzini 已提交
4064
    job->busy          = true;
4065
    bs->job = job;
4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077 4078 4079

    /* Only set speed when necessary to avoid NotSupported error */
    if (speed != 0) {
        Error *local_err = NULL;

        block_job_set_speed(job, speed, &local_err);
        if (error_is_set(&local_err)) {
            bs->job = NULL;
            g_free(job);
            bdrv_set_in_use(bs, 0);
            error_propagate(errp, local_err);
            return NULL;
        }
    }
4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093
    return job;
}

void block_job_complete(BlockJob *job, int ret)
{
    BlockDriverState *bs = job->bs;

    assert(bs->job == job);
    job->cb(job->opaque, ret);
    bs->job = NULL;
    g_free(job);
    bdrv_set_in_use(bs, 0);
}

4094
void block_job_set_speed(BlockJob *job, int64_t speed, Error **errp)
4095
{
4096
    Error *local_err = NULL;
4097

4098
    if (!job->job_type->set_speed) {
4099 4100
        error_set(errp, QERR_NOT_SUPPORTED);
        return;
4101
    }
4102
    job->job_type->set_speed(job, speed, &local_err);
4103 4104 4105
    if (error_is_set(&local_err)) {
        error_propagate(errp, local_err);
        return;
4106
    }
4107

4108
    job->speed = speed;
4109 4110 4111 4112 4113
}

void block_job_cancel(BlockJob *job)
{
    job->cancelled = true;
4114 4115 4116
    if (job->co && !job->busy) {
        qemu_coroutine_enter(job->co, NULL);
    }
4117 4118 4119 4120 4121 4122
}

bool block_job_is_cancelled(BlockJob *job)
{
    return job->cancelled;
}
4123

4124 4125 4126 4127 4128 4129 4130 4131 4132
struct BlockCancelData {
    BlockJob *job;
    BlockDriverCompletionFunc *cb;
    void *opaque;
    bool cancelled;
    int ret;
};

static void block_job_cancel_cb(void *opaque, int ret)
4133
{
4134 4135 4136 4137 4138 4139 4140 4141 4142 4143
    struct BlockCancelData *data = opaque;

    data->cancelled = block_job_is_cancelled(data->job);
    data->ret = ret;
    data->cb(data->opaque, ret);
}

int block_job_cancel_sync(BlockJob *job)
{
    struct BlockCancelData data;
4144 4145 4146
    BlockDriverState *bs = job->bs;

    assert(bs->job == job);
4147 4148 4149 4150 4151 4152 4153 4154 4155 4156

    /* Set up our own callback to store the result and chain to
     * the original callback.
     */
    data.job = job;
    data.cb = job->cb;
    data.opaque = job->opaque;
    data.ret = -EINPROGRESS;
    job->cb = block_job_cancel_cb;
    job->opaque = &data;
4157
    block_job_cancel(job);
4158
    while (data.ret == -EINPROGRESS) {
4159 4160
        qemu_aio_wait();
    }
4161
    return (data.cancelled && data.ret == 0) ? -ECANCELED : data.ret;
4162
}
P
Paolo Bonzini 已提交
4163 4164 4165 4166 4167 4168 4169 4170 4171 4172

void block_job_sleep_ns(BlockJob *job, QEMUClock *clock, int64_t ns)
{
    /* Check cancellation *before* setting busy = false, too!  */
    if (!block_job_is_cancelled(job)) {
        job->busy = false;
        co_sleep_ns(clock, ns);
        job->busy = true;
    }
}