block.c 144.1 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"
27
#include "monitor/monitor.h"
28 29
#include "block/block_int.h"
#include "block/blockjob.h"
30
#include "qemu/module.h"
31
#include "qapi/qmp/qjson.h"
32
#include "sysemu/sysemu.h"
33
#include "qemu/notify.h"
34
#include "block/coroutine.h"
35
#include "block/qapi.h"
L
Luiz Capitulino 已提交
36
#include "qmp-commands.h"
37
#include "qemu/timer.h"
B
bellard 已提交
38

J
Juan Quintela 已提交
39
#ifdef CONFIG_BSD
B
bellard 已提交
40 41 42
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/ioctl.h>
B
Blue Swirl 已提交
43
#include <sys/queue.h>
44
#ifndef __DragonFly__
B
bellard 已提交
45 46
#include <sys/disk.h>
#endif
47
#endif
B
bellard 已提交
48

49 50 51 52
#ifdef _WIN32
#include <windows.h>
#endif

F
Fam Zheng 已提交
53 54 55 56 57
struct BdrvDirtyBitmap {
    HBitmap *bitmap;
    QLIST_ENTRY(BdrvDirtyBitmap) list;
};

58 59
#define NOT_DONE 0x7fffffff /* used while emulated sync operation in progress */

60
static void bdrv_dev_change_media_cb(BlockDriverState *bs, bool load);
61 62
static BlockDriverAIOCB *bdrv_aio_readv_em(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
63
        BlockDriverCompletionFunc *cb, void *opaque);
64 65
static BlockDriverAIOCB *bdrv_aio_writev_em(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
66
        BlockDriverCompletionFunc *cb, void *opaque);
67 68 69 70 71 72
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);
73
static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,
74 75
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags);
76
static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs,
77 78
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags);
79 80 81 82
static BlockDriverAIOCB *bdrv_co_aio_rw_vector(BlockDriverState *bs,
                                               int64_t sector_num,
                                               QEMUIOVector *qiov,
                                               int nb_sectors,
P
Paolo Bonzini 已提交
83
                                               BdrvRequestFlags flags,
84 85
                                               BlockDriverCompletionFunc *cb,
                                               void *opaque,
86
                                               bool is_write);
87
static void coroutine_fn bdrv_co_do_rw(void *opaque);
K
Kevin Wolf 已提交
88
static int coroutine_fn bdrv_co_do_write_zeroes(BlockDriverState *bs,
89
    int64_t sector_num, int nb_sectors, BdrvRequestFlags flags);
B
bellard 已提交
90

91 92
static QTAILQ_HEAD(, BlockDriverState) bdrv_states =
    QTAILQ_HEAD_INITIALIZER(bdrv_states);
93

94 95 96
static QTAILQ_HEAD(, BlockDriverState) graph_bdrv_states =
    QTAILQ_HEAD_INITIALIZER(graph_bdrv_states);

97 98
static QLIST_HEAD(, BlockDriver) bdrv_drivers =
    QLIST_HEAD_INITIALIZER(bdrv_drivers);
B
bellard 已提交
99

100 101 102
/* If non-zero, use only whitelisted block drivers */
static int use_bdrv_whitelist;

103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122
#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

123
/* throttling disk I/O limits */
124 125
void bdrv_set_io_limits(BlockDriverState *bs,
                        ThrottleConfig *cfg)
Z
Zhi Yong Wu 已提交
126
{
127
    int i;
Z
Zhi Yong Wu 已提交
128

129
    throttle_config(&bs->throttle_state, cfg);
Z
Zhi Yong Wu 已提交
130

131 132
    for (i = 0; i < 2; i++) {
        qemu_co_enter_next(&bs->throttled_reqs[i]);
Z
Zhi Yong Wu 已提交
133
    }
134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151
}

/* this function drain all the throttled IOs */
static bool bdrv_start_throttled_reqs(BlockDriverState *bs)
{
    bool drained = false;
    bool enabled = bs->io_limits_enabled;
    int i;

    bs->io_limits_enabled = false;

    for (i = 0; i < 2; i++) {
        while (qemu_co_enter_next(&bs->throttled_reqs[i])) {
            drained = true;
        }
    }

    bs->io_limits_enabled = enabled;
Z
Zhi Yong Wu 已提交
152

153
    return drained;
Z
Zhi Yong Wu 已提交
154 155
}

156
void bdrv_io_limits_disable(BlockDriverState *bs)
157
{
158
    bs->io_limits_enabled = false;
159

160 161 162
    bdrv_start_throttled_reqs(bs);

    throttle_destroy(&bs->throttle_state);
163 164
}

165
static void bdrv_throttle_read_timer_cb(void *opaque)
166
{
167 168
    BlockDriverState *bs = opaque;
    qemu_co_enter_next(&bs->throttled_reqs[0]);
169 170
}

171
static void bdrv_throttle_write_timer_cb(void *opaque)
172
{
173 174
    BlockDriverState *bs = opaque;
    qemu_co_enter_next(&bs->throttled_reqs[1]);
175 176
}

177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193
/* should be called before bdrv_set_io_limits if a limit is set */
void bdrv_io_limits_enable(BlockDriverState *bs)
{
    assert(!bs->io_limits_enabled);
    throttle_init(&bs->throttle_state,
                  QEMU_CLOCK_VIRTUAL,
                  bdrv_throttle_read_timer_cb,
                  bdrv_throttle_write_timer_cb,
                  bs);
    bs->io_limits_enabled = true;
}

/* This function makes an IO wait if needed
 *
 * @nb_sectors: the number of sectors of the IO
 * @is_write:   is the IO a write
 */
Z
Zhi Yong Wu 已提交
194
static void bdrv_io_limits_intercept(BlockDriverState *bs,
195 196
                                     int nb_sectors,
                                     bool is_write)
Z
Zhi Yong Wu 已提交
197
{
198 199
    /* does this io must wait */
    bool must_wait = throttle_schedule_timer(&bs->throttle_state, is_write);
Z
Zhi Yong Wu 已提交
200

201 202 203 204
    /* if must wait or any request of this type throttled queue the IO */
    if (must_wait ||
        !qemu_co_queue_empty(&bs->throttled_reqs[is_write])) {
        qemu_co_queue_wait(&bs->throttled_reqs[is_write]);
Z
Zhi Yong Wu 已提交
205 206
    }

207 208 209 210
    /* the IO will be executed, do the accounting */
    throttle_account(&bs->throttle_state,
                     is_write,
                     nb_sectors * BDRV_SECTOR_SIZE);
Z
Zhi Yong Wu 已提交
211

212 213 214
    /* if the next request must wait -> do nothing */
    if (throttle_schedule_timer(&bs->throttle_state, is_write)) {
        return;
Z
Zhi Yong Wu 已提交
215 216
    }

217 218
    /* else queue next request for execution */
    qemu_co_queue_next(&bs->throttled_reqs[is_write]);
Z
Zhi Yong Wu 已提交
219 220
}

221 222 223
/* check if the path starts with "<protocol>:" */
static int path_has_protocol(const char *path)
{
224 225
    const char *p;

226 227 228 229 230
#ifdef _WIN32
    if (is_windows_drive(path) ||
        is_windows_drive_prefix(path)) {
        return 0;
    }
231 232 233
    p = path + strcspn(path, ":/\\");
#else
    p = path + strcspn(path, ":/");
234 235
#endif

236
    return *p == ':';
237 238
}

B
bellard 已提交
239
int path_is_absolute(const char *path)
240
{
B
bellard 已提交
241 242
#ifdef _WIN32
    /* specific case for names like: "\\.\d:" */
P
Paolo Bonzini 已提交
243
    if (is_windows_drive(path) || is_windows_drive_prefix(path)) {
B
bellard 已提交
244
        return 1;
P
Paolo Bonzini 已提交
245 246
    }
    return (*path == '/' || *path == '\\');
247
#else
P
Paolo Bonzini 已提交
248
    return (*path == '/');
249
#endif
250 251
}

B
bellard 已提交
252 253 254 255 256 257
/* 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)
258
{
B
bellard 已提交
259 260 261 262 263 264 265 266 267 268 269 270 271
    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;
272 273 274 275 276 277 278 279 280
        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
281 282 283 284 285 286 287 288 289 290 291 292
        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);
293 294 295
    }
}

296 297 298 299 300 301 302 303 304
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);
    }
}

305
void bdrv_register(BlockDriver *bdrv)
B
bellard 已提交
306
{
307 308
    /* Block drivers without coroutine functions need emulation */
    if (!bdrv->bdrv_co_readv) {
309 310 311
        bdrv->bdrv_co_readv = bdrv_co_readv_em;
        bdrv->bdrv_co_writev = bdrv_co_writev_em;

312 313 314
        /* 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.
         */
315 316 317 318 319
        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
320
    }
321

322
    QLIST_INSERT_HEAD(&bdrv_drivers, bdrv, list);
B
bellard 已提交
323
}
B
bellard 已提交
324 325 326 327

/* create a new block device (by default it is empty) */
BlockDriverState *bdrv_new(const char *device_name)
{
328
    BlockDriverState *bs;
B
bellard 已提交
329

330
    bs = g_malloc0(sizeof(BlockDriverState));
F
Fam Zheng 已提交
331
    QLIST_INIT(&bs->dirty_bitmaps);
B
bellard 已提交
332
    pstrcpy(bs->device_name, sizeof(bs->device_name), device_name);
B
bellard 已提交
333
    if (device_name[0] != '\0') {
334
        QTAILQ_INSERT_TAIL(&bdrv_states, bs, device_list);
B
bellard 已提交
335
    }
336
    bdrv_iostatus_disable(bs);
P
Paolo Bonzini 已提交
337
    notifier_list_init(&bs->close_notifiers);
338
    notifier_with_return_list_init(&bs->before_write_notifiers);
339 340
    qemu_co_queue_init(&bs->throttled_reqs[0]);
    qemu_co_queue_init(&bs->throttled_reqs[1]);
341
    bs->refcnt = 1;
P
Paolo Bonzini 已提交
342

B
bellard 已提交
343 344 345
    return bs;
}

P
Paolo Bonzini 已提交
346 347 348 349 350
void bdrv_add_close_notifier(BlockDriverState *bs, Notifier *notify)
{
    notifier_list_add(&bs->close_notifiers, notify);
}

B
bellard 已提交
351 352 353
BlockDriver *bdrv_find_format(const char *format_name)
{
    BlockDriver *drv1;
354 355
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
        if (!strcmp(drv1->format_name, format_name)) {
B
bellard 已提交
356
            return drv1;
357
        }
B
bellard 已提交
358 359 360 361
    }
    return NULL;
}

362
static int bdrv_is_whitelisted(BlockDriver *drv, bool read_only)
363
{
364 365 366 367 368
    static const char *whitelist_rw[] = {
        CONFIG_BDRV_RW_WHITELIST
    };
    static const char *whitelist_ro[] = {
        CONFIG_BDRV_RO_WHITELIST
369 370 371
    };
    const char **p;

372
    if (!whitelist_rw[0] && !whitelist_ro[0]) {
373
        return 1;               /* no whitelist, anything goes */
374
    }
375

376
    for (p = whitelist_rw; *p; p++) {
377 378 379 380
        if (!strcmp(drv->format_name, *p)) {
            return 1;
        }
    }
381 382 383 384 385 386 387
    if (read_only) {
        for (p = whitelist_ro; *p; p++) {
            if (!strcmp(drv->format_name, *p)) {
                return 1;
            }
        }
    }
388 389 390
    return 0;
}

391 392
BlockDriver *bdrv_find_whitelisted_format(const char *format_name,
                                          bool read_only)
393 394
{
    BlockDriver *drv = bdrv_find_format(format_name);
395
    return drv && bdrv_is_whitelisted(drv, read_only) ? drv : NULL;
396 397
}

398 399 400 401 402
typedef struct CreateCo {
    BlockDriver *drv;
    char *filename;
    QEMUOptionParameter *options;
    int ret;
403
    Error *err;
404 405 406 407
} CreateCo;

static void coroutine_fn bdrv_create_co_entry(void *opaque)
{
408 409 410
    Error *local_err = NULL;
    int ret;

411 412 413
    CreateCo *cco = opaque;
    assert(cco->drv);

414 415 416 417 418
    ret = cco->drv->bdrv_create(cco->filename, cco->options, &local_err);
    if (error_is_set(&local_err)) {
        error_propagate(&cco->err, local_err);
    }
    cco->ret = ret;
419 420
}

421
int bdrv_create(BlockDriver *drv, const char* filename,
422
    QEMUOptionParameter *options, Error **errp)
B
bellard 已提交
423
{
424 425 426 427 428 429 430 431
    int ret;

    Coroutine *co;
    CreateCo cco = {
        .drv = drv,
        .filename = g_strdup(filename),
        .options = options,
        .ret = NOT_DONE,
432
        .err = NULL,
433 434 435
    };

    if (!drv->bdrv_create) {
436
        error_setg(errp, "Driver '%s' does not support image creation", drv->format_name);
437 438
        ret = -ENOTSUP;
        goto out;
439 440 441 442 443 444 445 446 447 448 449 450 451 452
    }

    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;
453 454 455 456 457 458 459
    if (ret < 0) {
        if (error_is_set(&cco.err)) {
            error_propagate(errp, cco.err);
        } else {
            error_setg_errno(errp, -ret, "Could not create image");
        }
    }
460

461 462
out:
    g_free(cco.filename);
463
    return ret;
B
bellard 已提交
464 465
}

466 467
int bdrv_create_file(const char* filename, QEMUOptionParameter *options,
                     Error **errp)
468 469
{
    BlockDriver *drv;
470 471
    Error *local_err = NULL;
    int ret;
472

473
    drv = bdrv_find_protocol(filename, true);
474
    if (drv == NULL) {
475
        error_setg(errp, "Could not find protocol for file '%s'", filename);
476
        return -ENOENT;
477 478
    }

479 480 481 482 483
    ret = bdrv_create(drv, filename, options, &local_err);
    if (error_is_set(&local_err)) {
        error_propagate(errp, local_err);
    }
    return ret;
484 485
}

486 487 488 489 490 491 492 493 494 495 496 497 498
static int bdrv_refresh_limits(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;

    memset(&bs->bl, 0, sizeof(bs->bl));

    if (drv && drv->bdrv_refresh_limits) {
        return drv->bdrv_refresh_limits(bs);
    }

    return 0;
}

499 500 501 502 503
/*
 * 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
504
{
505
#ifdef _WIN32
506
    char temp_dir[MAX_PATH];
507 508 509 510 511 512
    /* 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
513
#else
B
bellard 已提交
514
    int fd;
515
    const char *tmpdir;
A
aurel32 已提交
516 517 518
    tmpdir = getenv("TMPDIR");
    if (!tmpdir)
        tmpdir = "/tmp";
519 520 521
    if (snprintf(filename, size, "%s/vl.XXXXXX", tmpdir) >= size) {
        return -EOVERFLOW;
    }
B
bellard 已提交
522
    fd = mkstemp(filename);
523 524 525 526 527
    if (fd < 0) {
        return -errno;
    }
    if (close(fd) != 0) {
        unlink(filename);
528 529 530
        return -errno;
    }
    return 0;
B
bellard 已提交
531
#endif
532
}
B
bellard 已提交
533

534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555
/*
 * 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;
}

556 557
BlockDriver *bdrv_find_protocol(const char *filename,
                                bool allow_protocol_prefix)
B
bellard 已提交
558 559 560
{
    BlockDriver *drv1;
    char protocol[128];
561
    int len;
B
bellard 已提交
562
    const char *p;
B
bellard 已提交
563

564 565
    /* TODO Drivers without bdrv_file_open must be specified explicitly */

566 567 568 569 570 571 572 573 574 575 576 577
    /*
     * 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;
    }

578
    if (!path_has_protocol(filename) || !allow_protocol_prefix) {
579
        return bdrv_find_format("file");
580
    }
581

582 583
    p = strchr(filename, ':');
    assert(p != NULL);
584 585 586 587 588
    len = p - filename;
    if (len > sizeof(protocol) - 1)
        len = sizeof(protocol) - 1;
    memcpy(protocol, filename, len);
    protocol[len] = '\0';
589
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
590
        if (drv1->protocol_name &&
591
            !strcmp(drv1->protocol_name, protocol)) {
B
bellard 已提交
592
            return drv1;
593
        }
B
bellard 已提交
594 595 596 597
    }
    return NULL;
}

598
static int find_image_format(BlockDriverState *bs, const char *filename,
599
                             BlockDriver **pdrv, Error **errp)
600
{
601
    int score, score_max;
602 603
    BlockDriver *drv1, *drv;
    uint8_t buf[2048];
604
    int ret = 0;
605

606
    /* Return the raw BlockDriver * to scsi-generic devices or empty drives */
607
    if (bs->sg || !bdrv_is_inserted(bs) || bdrv_getlength(bs) == 0) {
608 609
        drv = bdrv_find_format("raw");
        if (!drv) {
610
            error_setg(errp, "Could not find raw image format");
611 612 613 614
            ret = -ENOENT;
        }
        *pdrv = drv;
        return ret;
615
    }
616

B
bellard 已提交
617 618
    ret = bdrv_pread(bs, 0, buf, sizeof(buf));
    if (ret < 0) {
619 620
        error_setg_errno(errp, -ret, "Could not read image for determining its "
                         "format");
621 622
        *pdrv = NULL;
        return ret;
B
bellard 已提交
623 624
    }

B
bellard 已提交
625
    score_max = 0;
626
    drv = NULL;
627
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
B
bellard 已提交
628 629 630 631 632 633
        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
634
        }
B
bellard 已提交
635
    }
636
    if (!drv) {
637 638
        error_setg(errp, "Could not determine image format: No compatible "
                   "driver found");
639 640 641 642
        ret = -ENOENT;
    }
    *pdrv = drv;
    return ret;
B
bellard 已提交
643 644
}

645 646 647 648 649 650 651
/**
 * Set the current 'total_sectors' value
 */
static int refresh_total_sectors(BlockDriverState *bs, int64_t hint)
{
    BlockDriver *drv = bs->drv;

652 653 654 655
    /* Do not attempt drv->bdrv_getlength() on scsi-generic devices */
    if (bs->sg)
        return 0;

656 657 658 659 660 661
    /* 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;
        }
F
Fam Zheng 已提交
662
        hint = DIV_ROUND_UP(length, BDRV_SECTOR_SIZE);
663 664 665 666 667 668
    }

    bs->total_sectors = hint;
    return 0;
}

P
Paolo Bonzini 已提交
669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688
/**
 * Set open flags for a given discard mode
 *
 * Return 0 on success, -1 if the discard mode was invalid.
 */
int bdrv_parse_discard_flags(const char *mode, int *flags)
{
    *flags &= ~BDRV_O_UNMAP;

    if (!strcmp(mode, "off") || !strcmp(mode, "ignore")) {
        /* do nothing */
    } else if (!strcmp(mode, "on") || !strcmp(mode, "unmap")) {
        *flags |= BDRV_O_UNMAP;
    } else {
        return -1;
    }

    return 0;
}

689 690 691 692 693 694 695 696 697 698 699
/**
 * 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;
700 701
    } else if (!strcmp(mode, "directsync")) {
        *flags |= BDRV_O_NOCACHE;
702 703 704 705 706 707 708 709 710 711 712 713 714 715
    } 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;
}

716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731
/**
 * 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--;
}

K
Kevin Wolf 已提交
732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751
static int bdrv_open_flags(BlockDriverState *bs, int flags)
{
    int open_flags = flags | BDRV_O_CACHE_WB;

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

    /*
     * Snapshots should be writable.
     */
    if (bs->is_temporary) {
        open_flags |= BDRV_O_RDWR;
    }

    return open_flags;
}

752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778
static int bdrv_assign_node_name(BlockDriverState *bs,
                                 const char *node_name,
                                 Error **errp)
{
    if (!node_name) {
        return 0;
    }

    /* empty string node name is invalid */
    if (node_name[0] == '\0') {
        error_setg(errp, "Empty node name");
        return -EINVAL;
    }

    /* takes care of avoiding duplicates node names */
    if (bdrv_find_node(node_name)) {
        error_setg(errp, "Duplicate node name");
        return -EINVAL;
    }

    /* copy node name into the bs and insert it into the graph list */
    pstrcpy(bs->node_name, sizeof(bs->node_name), node_name);
    QTAILQ_INSERT_TAIL(&graph_bdrv_states, bs, node_list);

    return 0;
}

779 780
/*
 * Common part for opening disk images and files
781 782
 *
 * Removes all processed options from *options.
783
 */
784
static int bdrv_open_common(BlockDriverState *bs, BlockDriverState *file,
785
    QDict *options, int flags, BlockDriver *drv, Error **errp)
786 787
{
    int ret, open_flags;
K
Kevin Wolf 已提交
788
    const char *filename;
789
    const char *node_name = NULL;
790
    Error *local_err = NULL;
791 792

    assert(drv != NULL);
793
    assert(bs->file == NULL);
794
    assert(options != NULL && bs->options != options);
795

796 797 798 799 800 801 802
    if (file != NULL) {
        filename = file->filename;
    } else {
        filename = qdict_get_try_str(options, "filename");
    }

    trace_bdrv_open_common(bs, filename ?: "", flags, drv->format_name);
S
Stefan Hajnoczi 已提交
803

804 805 806 807 808 809 810
    node_name = qdict_get_try_str(options, "node-name");
    ret = bdrv_assign_node_name(bs, node_name, errp);
    if (ret < 0) {
        return ret;
    }
    qdict_del(options, "node-name");

811 812 813 814 815 816 817 818
    /* bdrv_open() with directly using a protocol as drv. This layer is already
     * opened, so assign it to bs (while file becomes a closed BlockDriverState)
     * and return immediately. */
    if (file != NULL && drv->bdrv_file_open) {
        bdrv_swap(file, bs);
        return 0;
    }

819 820
    bs->open_flags = flags;
    bs->buffer_alignment = 512;
A
Asias He 已提交
821
    bs->zero_beyond_eof = true;
822 823 824 825
    open_flags = bdrv_open_flags(bs, flags);
    bs->read_only = !(open_flags & BDRV_O_RDWR);

    if (use_bdrv_whitelist && !bdrv_is_whitelisted(drv, bs->read_only)) {
826 827 828 829 830
        error_setg(errp,
                   !bs->read_only && bdrv_is_whitelisted(drv, true)
                        ? "Driver '%s' can only be used for read-only devices"
                        : "Driver '%s' is not whitelisted",
                   drv->format_name);
831 832
        return -ENOTSUP;
    }
833

834
    assert(bs->copy_on_read == 0); /* bdrv_new() and bdrv_close() make it so */
835 836 837 838 839 840 841
    if (flags & BDRV_O_COPY_ON_READ) {
        if (!bs->read_only) {
            bdrv_enable_copy_on_read(bs);
        } else {
            error_setg(errp, "Can't use copy-on-read on read-only device");
            return -EINVAL;
        }
842 843
    }

844 845 846 847 848
    if (filename != NULL) {
        pstrcpy(bs->filename, sizeof(bs->filename), filename);
    } else {
        bs->filename[0] = '\0';
    }
849 850

    bs->drv = drv;
851
    bs->opaque = g_malloc0(drv->instance_size);
852

853
    bs->enable_write_cache = !!(flags & BDRV_O_CACHE_WB);
854

855 856
    /* Open the image, either directly or using a protocol */
    if (drv->bdrv_file_open) {
857
        assert(file == NULL);
858
        assert(!drv->bdrv_needs_filename || filename != NULL);
859
        ret = drv->bdrv_file_open(bs, options, open_flags, &local_err);
860
    } else {
861
        if (file == NULL) {
862 863
            error_setg(errp, "Can't use '%s' as a block driver for the "
                       "protocol level", drv->format_name);
864 865 866
            ret = -EINVAL;
            goto free_and_fail;
        }
867
        bs->file = file;
868
        ret = drv->bdrv_open(bs, options, open_flags, &local_err);
869 870
    }

871
    if (ret < 0) {
872 873
        if (error_is_set(&local_err)) {
            error_propagate(errp, local_err);
874 875
        } else if (bs->filename[0]) {
            error_setg_errno(errp, -ret, "Could not open '%s'", bs->filename);
876 877 878
        } else {
            error_setg_errno(errp, -ret, "Could not open image");
        }
879 880 881
        goto free_and_fail;
    }

882 883
    ret = refresh_total_sectors(bs, bs->total_sectors);
    if (ret < 0) {
884
        error_setg_errno(errp, -ret, "Could not refresh total sector count");
885
        goto free_and_fail;
886
    }
887

888 889
    bdrv_refresh_limits(bs);

890 891
#ifndef _WIN32
    if (bs->is_temporary) {
D
Dunrong Huang 已提交
892 893
        assert(bs->filename[0] != '\0');
        unlink(bs->filename);
894 895 896 897 898
    }
#endif
    return 0;

free_and_fail:
899
    bs->file = NULL;
900
    g_free(bs->opaque);
901 902 903 904 905
    bs->opaque = NULL;
    bs->drv = NULL;
    return ret;
}

K
Kevin Wolf 已提交
906 907
/*
 * Opens a file using a protocol (file, host_device, nbd, ...)
908 909 910 911 912
 *
 * options is a QDict of options to pass to the block drivers, or NULL for an
 * empty set of options. The reference to the QDict belongs to the block layer
 * after the call (even on failure), so if the caller intends to reuse the
 * dictionary, it needs to use QINCREF() before calling bdrv_file_open.
K
Kevin Wolf 已提交
913
 */
914
int bdrv_file_open(BlockDriverState **pbs, const char *filename,
915 916
                   const char *reference, QDict *options, int flags,
                   Error **errp)
B
bellard 已提交
917
{
918
    BlockDriverState *bs = NULL;
919
    BlockDriver *drv;
920
    const char *drvname;
921
    bool allow_protocol_prefix = false;
922
    Error *local_err = NULL;
B
bellard 已提交
923 924
    int ret;

925 926 927 928 929
    /* NULL means an empty set of options */
    if (options == NULL) {
        options = qdict_new();
    }

930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947
    if (reference) {
        if (filename || qdict_size(options)) {
            error_setg(errp, "Cannot reference an existing block device with "
                       "additional options or a new filename");
            return -EINVAL;
        }
        QDECREF(options);

        bs = bdrv_find(reference);
        if (!bs) {
            error_setg(errp, "Cannot find block device '%s'", reference);
            return -ENODEV;
        }
        bdrv_ref(bs);
        *pbs = bs;
        return 0;
    }

B
bellard 已提交
948
    bs = bdrv_new("");
949 950 951
    bs->options = options;
    options = qdict_clone_shallow(options);

K
Kevin Wolf 已提交
952 953 954 955 956
    /* Fetch the file name from the options QDict if necessary */
    if (!filename) {
        filename = qdict_get_try_str(options, "filename");
    } else if (filename && !qdict_haskey(options, "filename")) {
        qdict_put(options, "filename", qstring_from_str(filename));
957
        allow_protocol_prefix = true;
K
Kevin Wolf 已提交
958
    } else {
959 960
        error_setg(errp, "Can't specify 'file' and 'filename' options at the "
                   "same time");
K
Kevin Wolf 已提交
961 962 963 964
        ret = -EINVAL;
        goto fail;
    }

965 966 967
    /* Find the right block driver */
    drvname = qdict_get_try_str(options, "driver");
    if (drvname) {
968
        drv = bdrv_find_format(drvname);
969 970 971
        if (!drv) {
            error_setg(errp, "Unknown driver '%s'", drvname);
        }
972 973
        qdict_del(options, "driver");
    } else if (filename) {
974 975
        drv = bdrv_find_protocol(filename, allow_protocol_prefix);
        if (!drv) {
976
            error_setg(errp, "Unknown protocol");
977
        }
978
    } else {
979
        error_setg(errp, "Must specify either driver or file");
980 981 982 983
        drv = NULL;
    }

    if (!drv) {
984
        /* errp has been set already */
985 986 987 988 989 990
        ret = -ENOENT;
        goto fail;
    }

    /* Parse the filename and open it */
    if (drv->bdrv_parse_filename && filename) {
991 992
        drv->bdrv_parse_filename(filename, options, &local_err);
        if (error_is_set(&local_err)) {
993
            error_propagate(errp, local_err);
994 995 996
            ret = -EINVAL;
            goto fail;
        }
997
        qdict_del(options, "filename");
998
    } else if (drv->bdrv_needs_filename && !filename) {
999 1000
        error_setg(errp, "The '%s' block driver requires a file name",
                   drv->format_name);
1001 1002
        ret = -EINVAL;
        goto fail;
1003 1004
    }

M
Max Reitz 已提交
1005 1006 1007 1008 1009 1010
    if (!drv->bdrv_file_open) {
        ret = bdrv_open(bs, filename, options, flags, drv, &local_err);
        options = NULL;
    } else {
        ret = bdrv_open_common(bs, NULL, options, flags, drv, &local_err);
    }
B
bellard 已提交
1011
    if (ret < 0) {
1012
        error_propagate(errp, local_err);
1013 1014 1015 1016
        goto fail;
    }

    /* Check if any unknown options were used */
M
Max Reitz 已提交
1017
    if (options && (qdict_size(options) != 0)) {
1018
        const QDictEntry *entry = qdict_first(options);
1019 1020
        error_setg(errp, "Block protocol '%s' doesn't support the option '%s'",
                   drv->format_name, entry->key);
1021 1022
        ret = -EINVAL;
        goto fail;
1023
    }
1024 1025
    QDECREF(options);

1026
    bs->growable = 1;
B
bellard 已提交
1027 1028
    *pbs = bs;
    return 0;
1029 1030 1031 1032 1033 1034

fail:
    QDECREF(options);
    if (!bs->drv) {
        QDECREF(bs->options);
    }
F
Fam Zheng 已提交
1035
    bdrv_unref(bs);
1036
    return ret;
B
bellard 已提交
1037 1038
}

1039 1040 1041 1042 1043 1044 1045 1046
/*
 * Opens the backing file for a BlockDriverState if not yet open
 *
 * options is a QDict of options to pass to the block drivers, or NULL for an
 * empty set of options. The reference to the QDict is transferred to this
 * function (even on failure), so if the caller intends to reuse the dictionary,
 * it needs to use QINCREF() before calling bdrv_file_open.
 */
1047
int bdrv_open_backing_file(BlockDriverState *bs, QDict *options, Error **errp)
P
Paolo Bonzini 已提交
1048 1049 1050 1051
{
    char backing_filename[PATH_MAX];
    int back_flags, ret;
    BlockDriver *back_drv = NULL;
1052
    Error *local_err = NULL;
P
Paolo Bonzini 已提交
1053 1054

    if (bs->backing_hd != NULL) {
1055
        QDECREF(options);
P
Paolo Bonzini 已提交
1056 1057 1058
        return 0;
    }

1059 1060 1061 1062 1063
    /* NULL means an empty set of options */
    if (options == NULL) {
        options = qdict_new();
    }

P
Paolo Bonzini 已提交
1064
    bs->open_flags &= ~BDRV_O_NO_BACKING;
1065 1066 1067
    if (qdict_haskey(options, "file.filename")) {
        backing_filename[0] = '\0';
    } else if (bs->backing_file[0] == '\0' && qdict_size(options) == 0) {
1068
        QDECREF(options);
P
Paolo Bonzini 已提交
1069
        return 0;
F
Fam Zheng 已提交
1070 1071 1072
    } else {
        bdrv_get_full_backing_filename(bs, backing_filename,
                                       sizeof(backing_filename));
P
Paolo Bonzini 已提交
1073 1074 1075 1076 1077 1078 1079 1080 1081
    }

    bs->backing_hd = bdrv_new("");

    if (bs->backing_format[0] != '\0') {
        back_drv = bdrv_find_format(bs->backing_format);
    }

    /* backing files always opened read-only */
1082 1083
    back_flags = bs->open_flags & ~(BDRV_O_RDWR | BDRV_O_SNAPSHOT |
                                    BDRV_O_COPY_ON_READ);
P
Paolo Bonzini 已提交
1084

1085 1086
    ret = bdrv_open(bs->backing_hd,
                    *backing_filename ? backing_filename : NULL, options,
1087
                    back_flags, back_drv, &local_err);
P
Paolo Bonzini 已提交
1088
    if (ret < 0) {
F
Fam Zheng 已提交
1089
        bdrv_unref(bs->backing_hd);
P
Paolo Bonzini 已提交
1090 1091
        bs->backing_hd = NULL;
        bs->open_flags |= BDRV_O_NO_BACKING;
1092 1093 1094
        error_setg(errp, "Could not open backing file: %s",
                   error_get_pretty(local_err));
        error_free(local_err);
P
Paolo Bonzini 已提交
1095 1096
        return ret;
    }
P
Peter Feiner 已提交
1097 1098 1099 1100 1101 1102

    if (bs->backing_hd->file) {
        pstrcpy(bs->backing_file, sizeof(bs->backing_file),
                bs->backing_hd->file->filename);
    }

1103 1104 1105
    /* Recalculate the BlockLimits with the backing file */
    bdrv_refresh_limits(bs);

P
Paolo Bonzini 已提交
1106 1107 1108
    return 0;
}

M
Max Reitz 已提交
1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 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
/*
 * Opens a disk image whose options are given as BlockdevRef in another block
 * device's options.
 *
 * If force_raw is true, bdrv_file_open() will be used, thereby preventing any
 * image format auto-detection. If it is false and a filename is given,
 * bdrv_open() will be used for auto-detection.
 *
 * If allow_none is true, no image will be opened if filename is false and no
 * BlockdevRef is given. *pbs will remain unchanged and 0 will be returned.
 *
 * bdrev_key specifies the key for the image's BlockdevRef in the options QDict.
 * That QDict has to be flattened; therefore, if the BlockdevRef is a QDict
 * itself, all options starting with "${bdref_key}." are considered part of the
 * BlockdevRef.
 *
 * The BlockdevRef will be removed from the options QDict.
 */
int bdrv_open_image(BlockDriverState **pbs, const char *filename,
                    QDict *options, const char *bdref_key, int flags,
                    bool force_raw, bool allow_none, Error **errp)
{
    QDict *image_options;
    int ret;
    char *bdref_key_dot;
    const char *reference;

    bdref_key_dot = g_strdup_printf("%s.", bdref_key);
    qdict_extract_subqdict(options, &image_options, bdref_key_dot);
    g_free(bdref_key_dot);

    reference = qdict_get_try_str(options, bdref_key);
    if (!filename && !reference && !qdict_size(image_options)) {
        if (allow_none) {
            ret = 0;
        } else {
            error_setg(errp, "A block device must be specified for \"%s\"",
                       bdref_key);
            ret = -EINVAL;
        }
        goto done;
    }

    if (filename && !force_raw) {
        /* If a filename is given and the block driver should be detected
           automatically (instead of using none), use bdrv_open() in order to do
           that auto-detection. */
        BlockDriverState *bs;

        if (reference) {
            error_setg(errp, "Cannot reference an existing block device while "
                       "giving a filename");
            ret = -EINVAL;
            goto done;
        }

        bs = bdrv_new("");
        ret = bdrv_open(bs, filename, image_options, flags, NULL, errp);
        if (ret < 0) {
            bdrv_unref(bs);
        } else {
            *pbs = bs;
        }
    } else {
        ret = bdrv_file_open(pbs, filename, reference, image_options, flags,
                             errp);
    }

done:
    qdict_del(options, bdref_key);
    return ret;
}

K
Kevin Wolf 已提交
1182 1183
/*
 * Opens a disk image (raw, qcow2, vmdk, ...)
1184 1185 1186 1187 1188
 *
 * options is a QDict of options to pass to the block drivers, or NULL for an
 * empty set of options. The reference to the QDict belongs to the block layer
 * after the call (even on failure), so if the caller intends to reuse the
 * dictionary, it needs to use QINCREF() before calling bdrv_open.
K
Kevin Wolf 已提交
1189
 */
1190
int bdrv_open(BlockDriverState *bs, const char *filename, QDict *options,
1191
              int flags, BlockDriver *drv, Error **errp)
B
bellard 已提交
1192
{
K
Kevin Wolf 已提交
1193
    int ret;
1194 1195
    /* TODO: extra byte is a hack to ensure MAX_PATH space on Windows. */
    char tmp_filename[PATH_MAX + 1];
1196
    BlockDriverState *file = NULL;
1197
    const char *drvname;
1198
    Error *local_err = NULL;
B
bellard 已提交
1199

1200 1201 1202 1203 1204 1205
    /* NULL means an empty set of options */
    if (options == NULL) {
        options = qdict_new();
    }

    bs->options = options;
1206
    options = qdict_clone_shallow(options);
1207 1208

    /* For snapshot=on, create a temporary qcow2 overlay */
B
bellard 已提交
1209
    if (flags & BDRV_O_SNAPSHOT) {
B
bellard 已提交
1210 1211
        BlockDriverState *bs1;
        int64_t total_size;
K
Kevin Wolf 已提交
1212
        BlockDriver *bdrv_qcow2;
1213
        QEMUOptionParameter *create_options;
1214
        QDict *snapshot_options;
1215

B
bellard 已提交
1216 1217
        /* if snapshot, we create a temporary backing file and open it
           instead of opening 'filename' directly */
1218

1219
        /* Get the required size from the image */
B
bellard 已提交
1220
        bs1 = bdrv_new("");
1221
        QINCREF(options);
1222 1223
        ret = bdrv_open(bs1, filename, options, BDRV_O_NO_BACKING,
                        drv, &local_err);
1224
        if (ret < 0) {
F
Fam Zheng 已提交
1225
            bdrv_unref(bs1);
1226
            goto fail;
B
bellard 已提交
1227
        }
1228
        total_size = bdrv_getlength(bs1) & BDRV_SECTOR_MASK;
A
aliguori 已提交
1229

F
Fam Zheng 已提交
1230
        bdrv_unref(bs1);
1231

1232
        /* Create the temporary image */
1233 1234
        ret = get_tmp_filename(tmp_filename, sizeof(tmp_filename));
        if (ret < 0) {
1235
            error_setg_errno(errp, -ret, "Could not get temporary filename");
1236
            goto fail;
1237
        }
A
aliguori 已提交
1238

K
Kevin Wolf 已提交
1239
        bdrv_qcow2 = bdrv_find_format("qcow2");
1240 1241
        create_options = parse_option_parameters("", bdrv_qcow2->create_options,
                                                 NULL);
K
Kevin Wolf 已提交
1242

1243
        set_option_parameter_int(create_options, BLOCK_OPT_SIZE, total_size);
K
Kevin Wolf 已提交
1244

1245
        ret = bdrv_create(bdrv_qcow2, tmp_filename, create_options, &local_err);
1246
        free_option_parameters(create_options);
1247
        if (ret < 0) {
1248
            error_setg_errno(errp, -ret, "Could not create temporary overlay "
1249 1250 1251 1252
                             "'%s': %s", tmp_filename,
                             error_get_pretty(local_err));
            error_free(local_err);
            local_err = NULL;
1253
            goto fail;
B
bellard 已提交
1254
        }
K
Kevin Wolf 已提交
1255

1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271
        /* Prepare a new options QDict for the temporary file, where user
         * options refer to the backing file */
        if (filename) {
            qdict_put(options, "file.filename", qstring_from_str(filename));
        }
        if (drv) {
            qdict_put(options, "driver", qstring_from_str(drv->format_name));
        }

        snapshot_options = qdict_new();
        qdict_put(snapshot_options, "backing", options);
        qdict_flatten(snapshot_options);

        bs->options = snapshot_options;
        options = qdict_clone_shallow(bs->options);

B
bellard 已提交
1272
        filename = tmp_filename;
K
Kevin Wolf 已提交
1273
        drv = bdrv_qcow2;
B
bellard 已提交
1274 1275
        bs->is_temporary = 1;
    }
B
bellard 已提交
1276

1277 1278 1279 1280 1281
    /* Open image file without format layer */
    if (flags & BDRV_O_RDWR) {
        flags |= BDRV_O_ALLOW_RDWR;
    }

1282 1283 1284 1285 1286
    ret = bdrv_open_image(&file, filename, options, "file",
                          bdrv_open_flags(bs, flags | BDRV_O_UNMAP), true, true,
                          &local_err);
    if (ret < 0) {
        goto fail;
1287 1288
    }

K
Kevin Wolf 已提交
1289
    /* Find the right image format driver */
1290 1291
    drvname = qdict_get_try_str(options, "driver");
    if (drvname) {
1292
        drv = bdrv_find_format(drvname);
1293
        qdict_del(options, "driver");
1294 1295 1296 1297 1298
        if (!drv) {
            error_setg(errp, "Invalid driver: '%s'", drvname);
            ret = -EINVAL;
            goto unlink_and_fail;
        }
1299 1300
    }

1301
    if (!drv) {
1302 1303 1304 1305 1306 1307 1308
        if (file) {
            ret = find_image_format(file, filename, &drv, &local_err);
        } else {
            error_setg(errp, "Must specify either driver or file");
            ret = -EINVAL;
            goto unlink_and_fail;
        }
1309
    }
1310

1311 1312
    if (!drv) {
        goto unlink_and_fail;
B
bellard 已提交
1313
    }
K
Kevin Wolf 已提交
1314 1315

    /* Open the image */
1316
    ret = bdrv_open_common(bs, file, options, flags, drv, &local_err);
K
Kevin Wolf 已提交
1317
    if (ret < 0) {
1318 1319 1320
        goto unlink_and_fail;
    }

1321
    if (file && (bs->file != file)) {
F
Fam Zheng 已提交
1322
        bdrv_unref(file);
1323 1324 1325
        file = NULL;
    }

K
Kevin Wolf 已提交
1326
    /* If there is a backing file, use it */
P
Paolo Bonzini 已提交
1327
    if ((flags & BDRV_O_NO_BACKING) == 0) {
1328 1329
        QDict *backing_options;

1330
        qdict_extract_subqdict(options, &backing_options, "backing.");
1331
        ret = bdrv_open_backing_file(bs, backing_options, &local_err);
K
Kevin Wolf 已提交
1332
        if (ret < 0) {
1333
            goto close_and_fail;
K
Kevin Wolf 已提交
1334 1335 1336
        }
    }

1337 1338 1339
    /* Check if any unknown options were used */
    if (qdict_size(options) != 0) {
        const QDictEntry *entry = qdict_first(options);
1340 1341 1342
        error_setg(errp, "Block format '%s' used by device '%s' doesn't "
                   "support the option '%s'", drv->format_name, bs->device_name,
                   entry->key);
1343 1344 1345 1346 1347 1348

        ret = -EINVAL;
        goto close_and_fail;
    }
    QDECREF(options);

K
Kevin Wolf 已提交
1349
    if (!bdrv_key_required(bs)) {
1350
        bdrv_dev_change_media_cb(bs, true);
K
Kevin Wolf 已提交
1351 1352 1353 1354 1355
    }

    return 0;

unlink_and_fail:
1356
    if (file != NULL) {
F
Fam Zheng 已提交
1357
        bdrv_unref(file);
1358
    }
K
Kevin Wolf 已提交
1359 1360 1361
    if (bs->is_temporary) {
        unlink(filename);
    }
1362 1363
fail:
    QDECREF(bs->options);
1364
    QDECREF(options);
1365
    bs->options = NULL;
1366 1367 1368
    if (error_is_set(&local_err)) {
        error_propagate(errp, local_err);
    }
1369
    return ret;
1370

1371 1372 1373
close_and_fail:
    bdrv_close(bs);
    QDECREF(options);
1374 1375 1376
    if (error_is_set(&local_err)) {
        error_propagate(errp, local_err);
    }
K
Kevin Wolf 已提交
1377 1378 1379
    return ret;
}

1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547
typedef struct BlockReopenQueueEntry {
     bool prepared;
     BDRVReopenState state;
     QSIMPLEQ_ENTRY(BlockReopenQueueEntry) entry;
} BlockReopenQueueEntry;

/*
 * Adds a BlockDriverState to a simple queue for an atomic, transactional
 * reopen of multiple devices.
 *
 * bs_queue can either be an existing BlockReopenQueue that has had QSIMPLE_INIT
 * already performed, or alternatively may be NULL a new BlockReopenQueue will
 * be created and initialized. This newly created BlockReopenQueue should be
 * passed back in for subsequent calls that are intended to be of the same
 * atomic 'set'.
 *
 * bs is the BlockDriverState to add to the reopen queue.
 *
 * flags contains the open flags for the associated bs
 *
 * returns a pointer to bs_queue, which is either the newly allocated
 * bs_queue, or the existing bs_queue being used.
 *
 */
BlockReopenQueue *bdrv_reopen_queue(BlockReopenQueue *bs_queue,
                                    BlockDriverState *bs, int flags)
{
    assert(bs != NULL);

    BlockReopenQueueEntry *bs_entry;
    if (bs_queue == NULL) {
        bs_queue = g_new0(BlockReopenQueue, 1);
        QSIMPLEQ_INIT(bs_queue);
    }

    if (bs->file) {
        bdrv_reopen_queue(bs_queue, bs->file, flags);
    }

    bs_entry = g_new0(BlockReopenQueueEntry, 1);
    QSIMPLEQ_INSERT_TAIL(bs_queue, bs_entry, entry);

    bs_entry->state.bs = bs;
    bs_entry->state.flags = flags;

    return bs_queue;
}

/*
 * Reopen multiple BlockDriverStates atomically & transactionally.
 *
 * The queue passed in (bs_queue) must have been built up previous
 * via bdrv_reopen_queue().
 *
 * Reopens all BDS specified in the queue, with the appropriate
 * flags.  All devices are prepared for reopen, and failure of any
 * device will cause all device changes to be abandonded, and intermediate
 * data cleaned up.
 *
 * If all devices prepare successfully, then the changes are committed
 * to all devices.
 *
 */
int bdrv_reopen_multiple(BlockReopenQueue *bs_queue, Error **errp)
{
    int ret = -1;
    BlockReopenQueueEntry *bs_entry, *next;
    Error *local_err = NULL;

    assert(bs_queue != NULL);

    bdrv_drain_all();

    QSIMPLEQ_FOREACH(bs_entry, bs_queue, entry) {
        if (bdrv_reopen_prepare(&bs_entry->state, bs_queue, &local_err)) {
            error_propagate(errp, local_err);
            goto cleanup;
        }
        bs_entry->prepared = true;
    }

    /* If we reach this point, we have success and just need to apply the
     * changes
     */
    QSIMPLEQ_FOREACH(bs_entry, bs_queue, entry) {
        bdrv_reopen_commit(&bs_entry->state);
    }

    ret = 0;

cleanup:
    QSIMPLEQ_FOREACH_SAFE(bs_entry, bs_queue, entry, next) {
        if (ret && bs_entry->prepared) {
            bdrv_reopen_abort(&bs_entry->state);
        }
        g_free(bs_entry);
    }
    g_free(bs_queue);
    return ret;
}


/* Reopen a single BlockDriverState with the specified flags. */
int bdrv_reopen(BlockDriverState *bs, int bdrv_flags, Error **errp)
{
    int ret = -1;
    Error *local_err = NULL;
    BlockReopenQueue *queue = bdrv_reopen_queue(NULL, bs, bdrv_flags);

    ret = bdrv_reopen_multiple(queue, &local_err);
    if (local_err != NULL) {
        error_propagate(errp, local_err);
    }
    return ret;
}


/*
 * Prepares a BlockDriverState for reopen. All changes are staged in the
 * 'opaque' field of the BDRVReopenState, which is used and allocated by
 * the block driver layer .bdrv_reopen_prepare()
 *
 * bs is the BlockDriverState to reopen
 * flags are the new open flags
 * queue is the reopen queue
 *
 * Returns 0 on success, non-zero on error.  On error errp will be set
 * as well.
 *
 * On failure, bdrv_reopen_abort() will be called to clean up any data.
 * It is the responsibility of the caller to then call the abort() or
 * commit() for any other BDS that have been left in a prepare() state
 *
 */
int bdrv_reopen_prepare(BDRVReopenState *reopen_state, BlockReopenQueue *queue,
                        Error **errp)
{
    int ret = -1;
    Error *local_err = NULL;
    BlockDriver *drv;

    assert(reopen_state != NULL);
    assert(reopen_state->bs->drv != NULL);
    drv = reopen_state->bs->drv;

    /* if we are to stay read-only, do not allow permission change
     * to r/w */
    if (!(reopen_state->bs->open_flags & BDRV_O_ALLOW_RDWR) &&
        reopen_state->flags & BDRV_O_RDWR) {
        error_set(errp, QERR_DEVICE_IS_READ_ONLY,
                  reopen_state->bs->device_name);
        goto error;
    }


    ret = bdrv_flush(reopen_state->bs);
    if (ret) {
        error_set(errp, ERROR_CLASS_GENERIC_ERROR, "Error (%s) flushing drive",
                  strerror(-ret));
        goto error;
    }

    if (drv->bdrv_reopen_prepare) {
        ret = drv->bdrv_reopen_prepare(reopen_state, queue, &local_err);
        if (ret) {
            if (local_err != NULL) {
                error_propagate(errp, local_err);
            } else {
1548 1549
                error_setg(errp, "failed while preparing to reopen image '%s'",
                           reopen_state->bs->filename);
1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611
            }
            goto error;
        }
    } else {
        /* It is currently mandatory to have a bdrv_reopen_prepare()
         * handler for each supported drv. */
        error_set(errp, QERR_BLOCK_FORMAT_FEATURE_NOT_SUPPORTED,
                  drv->format_name, reopen_state->bs->device_name,
                 "reopening of file");
        ret = -1;
        goto error;
    }

    ret = 0;

error:
    return ret;
}

/*
 * Takes the staged changes for the reopen from bdrv_reopen_prepare(), and
 * makes them final by swapping the staging BlockDriverState contents into
 * the active BlockDriverState contents.
 */
void bdrv_reopen_commit(BDRVReopenState *reopen_state)
{
    BlockDriver *drv;

    assert(reopen_state != NULL);
    drv = reopen_state->bs->drv;
    assert(drv != NULL);

    /* If there are any driver level actions to take */
    if (drv->bdrv_reopen_commit) {
        drv->bdrv_reopen_commit(reopen_state);
    }

    /* set BDS specific flags now */
    reopen_state->bs->open_flags         = reopen_state->flags;
    reopen_state->bs->enable_write_cache = !!(reopen_state->flags &
                                              BDRV_O_CACHE_WB);
    reopen_state->bs->read_only = !(reopen_state->flags & BDRV_O_RDWR);
}

/*
 * Abort the reopen, and delete and free the staged changes in
 * reopen_state
 */
void bdrv_reopen_abort(BDRVReopenState *reopen_state)
{
    BlockDriver *drv;

    assert(reopen_state != NULL);
    drv = reopen_state->bs->drv;
    assert(drv != NULL);

    if (drv->bdrv_reopen_abort) {
        drv->bdrv_reopen_abort(reopen_state);
    }
}


B
bellard 已提交
1612 1613
void bdrv_close(BlockDriverState *bs)
{
1614 1615 1616
    if (bs->job) {
        block_job_cancel_sync(bs->job);
    }
1617 1618 1619
    bdrv_drain_all(); /* complete I/O */
    bdrv_flush(bs);
    bdrv_drain_all(); /* in case flush left pending I/O */
P
Paolo Bonzini 已提交
1620
    notifier_list_notify(&bs->close_notifiers, bs);
K
Kevin Wolf 已提交
1621

1622
    if (bs->drv) {
1623
        if (bs->backing_hd) {
F
Fam Zheng 已提交
1624
            bdrv_unref(bs->backing_hd);
1625 1626
            bs->backing_hd = NULL;
        }
B
bellard 已提交
1627
        bs->drv->bdrv_close(bs);
1628
        g_free(bs->opaque);
B
bellard 已提交
1629 1630 1631 1632
#ifdef _WIN32
        if (bs->is_temporary) {
            unlink(bs->filename);
        }
B
bellard 已提交
1633
#endif
B
bellard 已提交
1634 1635
        bs->opaque = NULL;
        bs->drv = NULL;
1636
        bs->copy_on_read = 0;
1637 1638
        bs->backing_file[0] = '\0';
        bs->backing_format[0] = '\0';
1639 1640 1641 1642 1643
        bs->total_sectors = 0;
        bs->encrypted = 0;
        bs->valid_key = 0;
        bs->sg = 0;
        bs->growable = 0;
A
Asias He 已提交
1644
        bs->zero_beyond_eof = false;
1645 1646
        QDECREF(bs->options);
        bs->options = NULL;
B
bellard 已提交
1647

1648
        if (bs->file != NULL) {
F
Fam Zheng 已提交
1649
            bdrv_unref(bs->file);
1650
            bs->file = NULL;
1651
        }
B
bellard 已提交
1652
    }
Z
Zhi Yong Wu 已提交
1653

P
Pavel Hrdina 已提交
1654 1655
    bdrv_dev_change_media_cb(bs, false);

Z
Zhi Yong Wu 已提交
1656 1657 1658 1659
    /*throttling disk I/O limits*/
    if (bs->io_limits_enabled) {
        bdrv_io_limits_disable(bs);
    }
B
bellard 已提交
1660 1661
}

1662 1663 1664 1665
void bdrv_close_all(void)
{
    BlockDriverState *bs;

1666
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
1667 1668 1669 1670
        bdrv_close(bs);
    }
}

1671 1672 1673 1674 1675 1676
/* Check if any requests are in-flight (including throttled requests) */
static bool bdrv_requests_pending(BlockDriverState *bs)
{
    if (!QLIST_EMPTY(&bs->tracked_requests)) {
        return true;
    }
1677 1678 1679 1680
    if (!qemu_co_queue_empty(&bs->throttled_reqs[0])) {
        return true;
    }
    if (!qemu_co_queue_empty(&bs->throttled_reqs[1])) {
1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694
        return true;
    }
    if (bs->file && bdrv_requests_pending(bs->file)) {
        return true;
    }
    if (bs->backing_hd && bdrv_requests_pending(bs->backing_hd)) {
        return true;
    }
    return false;
}

static bool bdrv_requests_pending_all(void)
{
    BlockDriverState *bs;
1695
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
1696 1697 1698 1699 1700 1701 1702
        if (bdrv_requests_pending(bs)) {
            return true;
        }
    }
    return false;
}

1703 1704 1705 1706 1707
/*
 * 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.
1708 1709 1710 1711 1712 1713
 *
 * 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.
1714 1715 1716
 */
void bdrv_drain_all(void)
{
1717 1718
    /* Always run first iteration so any pending completion BHs run */
    bool busy = true;
1719 1720
    BlockDriverState *bs;

1721
    while (busy) {
1722
        QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
1723
            bdrv_start_throttled_reqs(bs);
1724
        }
1725

1726 1727
        busy = bdrv_requests_pending_all();
        busy |= aio_poll(qemu_get_aio_context(), busy);
1728 1729 1730
    }
}

1731 1732
/* make a BlockDriverState anonymous by removing from bdrv_state and
 * graph_bdrv_state list.
1733 1734 1735 1736
   Also, NULL terminate the device_name to prevent double remove */
void bdrv_make_anon(BlockDriverState *bs)
{
    if (bs->device_name[0] != '\0') {
1737
        QTAILQ_REMOVE(&bdrv_states, bs, device_list);
1738 1739
    }
    bs->device_name[0] = '\0';
1740 1741 1742 1743
    if (bs->node_name[0] != '\0') {
        QTAILQ_REMOVE(&graph_bdrv_states, bs, node_list);
    }
    bs->node_name[0] = '\0';
1744 1745
}

P
Paolo Bonzini 已提交
1746 1747 1748 1749 1750 1751 1752
static void bdrv_rebind(BlockDriverState *bs)
{
    if (bs->drv && bs->drv->bdrv_rebind) {
        bs->drv->bdrv_rebind(bs);
    }
}

1753 1754
static void bdrv_move_feature_fields(BlockDriverState *bs_dest,
                                     BlockDriverState *bs_src)
1755
{
1756 1757
    /* move some fields that need to stay attached to the device */
    bs_dest->open_flags         = bs_src->open_flags;
1758 1759

    /* dev info */
1760 1761 1762 1763 1764
    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;
1765

1766
    bs_dest->enable_write_cache = bs_src->enable_write_cache;
1767

1768 1769 1770 1771 1772 1773
    /* i/o throttled req */
    memcpy(&bs_dest->throttle_state,
           &bs_src->throttle_state,
           sizeof(ThrottleState));
    bs_dest->throttled_reqs[0]  = bs_src->throttled_reqs[0];
    bs_dest->throttled_reqs[1]  = bs_src->throttled_reqs[1];
1774
    bs_dest->io_limits_enabled  = bs_src->io_limits_enabled;
1775 1776

    /* r/w error */
1777 1778
    bs_dest->on_read_error      = bs_src->on_read_error;
    bs_dest->on_write_error     = bs_src->on_write_error;
1779 1780

    /* i/o status */
1781 1782
    bs_dest->iostatus_enabled   = bs_src->iostatus_enabled;
    bs_dest->iostatus           = bs_src->iostatus;
1783

1784
    /* dirty bitmap */
F
Fam Zheng 已提交
1785
    bs_dest->dirty_bitmaps      = bs_src->dirty_bitmaps;
1786

1787 1788 1789
    /* reference count */
    bs_dest->refcnt             = bs_src->refcnt;

1790
    /* job */
1791 1792
    bs_dest->in_use             = bs_src->in_use;
    bs_dest->job                = bs_src->job;
1793

1794
    /* keep the same entry in bdrv_states */
1795 1796
    pstrcpy(bs_dest->device_name, sizeof(bs_dest->device_name),
            bs_src->device_name);
1797 1798 1799 1800 1801 1802
    bs_dest->device_list = bs_src->device_list;

    /* keep the same entry in graph_bdrv_states
     * We do want to swap name but don't want to swap linked list entries
     */
    bs_dest->node_list   = bs_src->node_list;
1803
}
1804

1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819
/*
 * 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
1820

1821 1822
    /* bs_new must be anonymous and shouldn't have anything fancy enabled */
    assert(bs_new->device_name[0] == '\0');
F
Fam Zheng 已提交
1823
    assert(QLIST_EMPTY(&bs_new->dirty_bitmaps));
1824 1825 1826 1827
    assert(bs_new->job == NULL);
    assert(bs_new->dev == NULL);
    assert(bs_new->in_use == 0);
    assert(bs_new->io_limits_enabled == false);
1828
    assert(!throttle_have_timer(&bs_new->throttle_state));
1829

1830 1831 1832
    tmp = *bs_new;
    *bs_new = *bs_old;
    *bs_old = tmp;
1833

1834 1835 1836 1837
    /* 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);
1838

1839 1840 1841 1842 1843 1844 1845 1846
    /* 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);
1847
    assert(!throttle_have_timer(&bs_new->throttle_state));
P
Paolo Bonzini 已提交
1848 1849

    bdrv_rebind(bs_new);
1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875
    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 : "");
1876 1877
}

F
Fam Zheng 已提交
1878
static void bdrv_delete(BlockDriverState *bs)
B
bellard 已提交
1879
{
1880
    assert(!bs->dev);
1881 1882
    assert(!bs->job);
    assert(!bs->in_use);
F
Fam Zheng 已提交
1883
    assert(!bs->refcnt);
F
Fam Zheng 已提交
1884
    assert(QLIST_EMPTY(&bs->dirty_bitmaps));
1885

1886 1887
    bdrv_close(bs);

1888
    /* remove from list, if necessary */
1889
    bdrv_make_anon(bs);
1890

1891
    g_free(bs);
B
bellard 已提交
1892 1893
}

1894 1895
int bdrv_attach_dev(BlockDriverState *bs, void *dev)
/* TODO change to DeviceState *dev when all users are qdevified */
1896
{
1897
    if (bs->dev) {
1898 1899
        return -EBUSY;
    }
1900
    bs->dev = dev;
1901
    bdrv_iostatus_reset(bs);
1902 1903 1904
    return 0;
}

1905 1906
/* TODO qdevified devices don't use this, remove when devices are qdevified */
void bdrv_attach_dev_nofail(BlockDriverState *bs, void *dev)
1907
{
1908 1909 1910 1911 1912 1913 1914 1915 1916 1917
    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;
1918 1919
    bs->dev_ops = NULL;
    bs->dev_opaque = NULL;
1920
    bs->buffer_alignment = 512;
1921 1922
}

1923 1924
/* TODO change to return DeviceState * when all users are qdevified */
void *bdrv_get_attached_dev(BlockDriverState *bs)
1925
{
1926
    return bs->dev;
1927 1928
}

1929 1930 1931 1932 1933 1934 1935
void bdrv_set_dev_ops(BlockDriverState *bs, const BlockDevOps *ops,
                      void *opaque)
{
    bs->dev_ops = ops;
    bs->dev_opaque = opaque;
}

P
Paolo Bonzini 已提交
1936 1937 1938
void bdrv_emit_qmp_error_event(const BlockDriverState *bdrv,
                               enum MonitorEvent ev,
                               BlockErrorAction action, bool is_read)
1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960
{
    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");
P
Paolo Bonzini 已提交
1961
    monitor_protocol_event(ev, data);
1962 1963 1964 1965

    qobject_decref(data);
}

1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976
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);
}

1977
static void bdrv_dev_change_media_cb(BlockDriverState *bs, bool load)
1978
{
1979
    if (bs->dev_ops && bs->dev_ops->change_media_cb) {
1980
        bool tray_was_closed = !bdrv_dev_is_tray_open(bs);
1981
        bs->dev_ops->change_media_cb(bs->dev_opaque, load);
1982 1983 1984 1985 1986 1987 1988 1989
        if (tray_was_closed) {
            /* tray open */
            bdrv_emit_qmp_eject_event(bs, true);
        }
        if (load) {
            /* tray close */
            bdrv_emit_qmp_eject_event(bs, false);
        }
1990 1991 1992
    }
}

1993 1994 1995 1996 1997
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
1998 1999 2000 2001 2002 2003 2004
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);
    }
}

2005 2006 2007 2008 2009 2010 2011 2012
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;
}

2013 2014 2015 2016
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);
2017 2018 2019
    }
}

2020 2021 2022 2023 2024 2025 2026 2027
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
2028 2029 2030
/*
 * Run consistency checks on an image
 *
2031
 * Returns 0 if the check could be completed (it doesn't mean that the image is
2032
 * free of errors) or -errno when an internal error occurred. The results of the
2033
 * check are stored in res.
A
aliguori 已提交
2034
 */
2035
int bdrv_check(BlockDriverState *bs, BdrvCheckResult *res, BdrvCheckMode fix)
A
aliguori 已提交
2036 2037 2038 2039 2040
{
    if (bs->drv->bdrv_check == NULL) {
        return -ENOTSUP;
    }

2041
    memset(res, 0, sizeof(*res));
2042
    return bs->drv->bdrv_check(bs, res, fix);
A
aliguori 已提交
2043 2044
}

2045 2046
#define COMMIT_BUF_SECTORS 2048

2047 2048 2049
/* commit COW file into the raw image */
int bdrv_commit(BlockDriverState *bs)
{
B
bellard 已提交
2050
    BlockDriver *drv = bs->drv;
2051
    int64_t sector, total_sectors, length, backing_length;
2052
    int n, ro, open_flags;
2053
    int ret = 0;
2054
    uint8_t *buf = NULL;
2055
    char filename[PATH_MAX];
2056

B
bellard 已提交
2057 2058
    if (!drv)
        return -ENOMEDIUM;
2059 2060 2061
    
    if (!bs->backing_hd) {
        return -ENOTSUP;
2062 2063
    }

2064 2065 2066 2067
    if (bdrv_in_use(bs) || bdrv_in_use(bs->backing_hd)) {
        return -EBUSY;
    }

2068
    ro = bs->backing_hd->read_only;
2069 2070
    /* Use pstrcpy (not strncpy): filename must be NUL-terminated. */
    pstrcpy(filename, sizeof(filename), bs->backing_hd->filename);
2071 2072 2073
    open_flags =  bs->backing_hd->open_flags;

    if (ro) {
2074 2075
        if (bdrv_reopen(bs->backing_hd, open_flags | BDRV_O_RDWR, NULL)) {
            return -EACCES;
2076
        }
B
bellard 已提交
2077
    }
2078

2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101
    length = bdrv_getlength(bs);
    if (length < 0) {
        ret = length;
        goto ro_cleanup;
    }

    backing_length = bdrv_getlength(bs->backing_hd);
    if (backing_length < 0) {
        ret = backing_length;
        goto ro_cleanup;
    }

    /* If our top snapshot is larger than the backing file image,
     * grow the backing file image if possible.  If not possible,
     * we must return an error */
    if (length > backing_length) {
        ret = bdrv_truncate(bs->backing_hd, length);
        if (ret < 0) {
            goto ro_cleanup;
        }
    }

    total_sectors = length >> BDRV_SECTOR_BITS;
2102
    buf = g_malloc(COMMIT_BUF_SECTORS * BDRV_SECTOR_SIZE);
2103 2104

    for (sector = 0; sector < total_sectors; sector += n) {
2105 2106 2107 2108 2109
        ret = bdrv_is_allocated(bs, sector, COMMIT_BUF_SECTORS, &n);
        if (ret < 0) {
            goto ro_cleanup;
        }
        if (ret) {
K
Kevin Wolf 已提交
2110 2111
            ret = bdrv_read(bs, sector, buf, n);
            if (ret < 0) {
2112 2113 2114
                goto ro_cleanup;
            }

K
Kevin Wolf 已提交
2115 2116
            ret = bdrv_write(bs->backing_hd, sector, buf, n);
            if (ret < 0) {
2117 2118
                goto ro_cleanup;
            }
B
bellard 已提交
2119
        }
2120
    }
2121

2122 2123
    if (drv->bdrv_make_empty) {
        ret = drv->bdrv_make_empty(bs);
K
Kevin Wolf 已提交
2124 2125 2126
        if (ret < 0) {
            goto ro_cleanup;
        }
2127 2128
        bdrv_flush(bs);
    }
2129

2130 2131 2132 2133
    /*
     * Make sure all data we wrote to the backing device is actually
     * stable on disk.
     */
K
Kevin Wolf 已提交
2134
    if (bs->backing_hd) {
2135
        bdrv_flush(bs->backing_hd);
K
Kevin Wolf 已提交
2136
    }
2137

K
Kevin Wolf 已提交
2138
    ret = 0;
2139
ro_cleanup:
2140
    g_free(buf);
2141 2142

    if (ro) {
2143 2144
        /* ignoring error return here */
        bdrv_reopen(bs->backing_hd, open_flags & ~BDRV_O_RDWR, NULL);
2145 2146
    }

2147
    return ret;
2148 2149
}

2150
int bdrv_commit_all(void)
2151 2152 2153
{
    BlockDriverState *bs;

2154
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
2155 2156 2157 2158 2159
        if (bs->drv && bs->backing_hd) {
            int ret = bdrv_commit(bs);
            if (ret < 0) {
                return ret;
            }
2160
        }
2161
    }
2162
    return 0;
2163 2164
}

S
Stefan Hajnoczi 已提交
2165 2166 2167 2168 2169 2170 2171 2172
/**
 * 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);
2173
    qemu_co_queue_restart_all(&req->wait_queue);
S
Stefan Hajnoczi 已提交
2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188
}

/**
 * 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,
2189
        .co = qemu_coroutine_self(),
S
Stefan Hajnoczi 已提交
2190 2191
    };

2192 2193
    qemu_co_queue_init(&req->wait_queue);

S
Stefan Hajnoczi 已提交
2194 2195 2196
    QLIST_INSERT_HEAD(&bs->tracked_requests, req, list);
}

S
Stefan Hajnoczi 已提交
2197 2198 2199
/**
 * Round a region to cluster boundaries
 */
2200 2201 2202 2203
void bdrv_round_to_clusters(BlockDriverState *bs,
                            int64_t sector_num, int nb_sectors,
                            int64_t *cluster_sector_num,
                            int *cluster_nb_sectors)
S
Stefan Hajnoczi 已提交
2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217
{
    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);
    }
}

2218 2219
static bool tracked_request_overlaps(BdrvTrackedRequest *req,
                                     int64_t sector_num, int nb_sectors) {
S
Stefan Hajnoczi 已提交
2220 2221 2222 2223 2224 2225 2226 2227 2228
    /*        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;
2229 2230 2231 2232 2233 2234
}

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
2235 2236
    int64_t cluster_sector_num;
    int cluster_nb_sectors;
2237 2238
    bool retry;

S
Stefan Hajnoczi 已提交
2239 2240 2241 2242 2243 2244
    /* 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.
     */
2245 2246
    bdrv_round_to_clusters(bs, sector_num, nb_sectors,
                           &cluster_sector_num, &cluster_nb_sectors);
S
Stefan Hajnoczi 已提交
2247

2248 2249 2250
    do {
        retry = false;
        QLIST_FOREACH(req, &bs->tracked_requests, list) {
S
Stefan Hajnoczi 已提交
2251 2252
            if (tracked_request_overlaps(req, cluster_sector_num,
                                         cluster_nb_sectors)) {
2253 2254 2255 2256 2257 2258
                /* 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);

2259 2260 2261 2262 2263 2264 2265 2266
                qemu_co_queue_wait(&req->wait_queue);
                retry = true;
                break;
            }
        }
    } while (retry);
}

K
Kevin Wolf 已提交
2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278
/*
 * 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;
2279
    int ret;
K
Kevin Wolf 已提交
2280

2281 2282 2283 2284 2285
    /* Backing file format doesn't make sense without a backing file */
    if (backing_fmt && !backing_file) {
        return -EINVAL;
    }

K
Kevin Wolf 已提交
2286
    if (drv->bdrv_change_backing_file != NULL) {
2287
        ret = drv->bdrv_change_backing_file(bs, backing_file, backing_fmt);
K
Kevin Wolf 已提交
2288
    } else {
2289
        ret = -ENOTSUP;
K
Kevin Wolf 已提交
2290
    }
2291 2292 2293 2294 2295 2296

    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
2297 2298
}

2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 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 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429
/*
 * Finds the image layer in the chain that has 'bs' as its backing file.
 *
 * active is the current topmost image.
 *
 * Returns NULL if bs is not found in active's image chain,
 * or if active == bs.
 */
BlockDriverState *bdrv_find_overlay(BlockDriverState *active,
                                    BlockDriverState *bs)
{
    BlockDriverState *overlay = NULL;
    BlockDriverState *intermediate;

    assert(active != NULL);
    assert(bs != NULL);

    /* if bs is the same as active, then by definition it has no overlay
     */
    if (active == bs) {
        return NULL;
    }

    intermediate = active;
    while (intermediate->backing_hd) {
        if (intermediate->backing_hd == bs) {
            overlay = intermediate;
            break;
        }
        intermediate = intermediate->backing_hd;
    }

    return overlay;
}

typedef struct BlkIntermediateStates {
    BlockDriverState *bs;
    QSIMPLEQ_ENTRY(BlkIntermediateStates) entry;
} BlkIntermediateStates;


/*
 * Drops images above 'base' up to and including 'top', and sets the image
 * above 'top' to have base as its backing file.
 *
 * Requires that the overlay to 'top' is opened r/w, so that the backing file
 * information in 'bs' can be properly updated.
 *
 * E.g., this will convert the following chain:
 * bottom <- base <- intermediate <- top <- active
 *
 * to
 *
 * bottom <- base <- active
 *
 * It is allowed for bottom==base, in which case it converts:
 *
 * base <- intermediate <- top <- active
 *
 * to
 *
 * base <- active
 *
 * Error conditions:
 *  if active == top, that is considered an error
 *
 */
int bdrv_drop_intermediate(BlockDriverState *active, BlockDriverState *top,
                           BlockDriverState *base)
{
    BlockDriverState *intermediate;
    BlockDriverState *base_bs = NULL;
    BlockDriverState *new_top_bs = NULL;
    BlkIntermediateStates *intermediate_state, *next;
    int ret = -EIO;

    QSIMPLEQ_HEAD(states_to_delete, BlkIntermediateStates) states_to_delete;
    QSIMPLEQ_INIT(&states_to_delete);

    if (!top->drv || !base->drv) {
        goto exit;
    }

    new_top_bs = bdrv_find_overlay(active, top);

    if (new_top_bs == NULL) {
        /* we could not find the image above 'top', this is an error */
        goto exit;
    }

    /* special case of new_top_bs->backing_hd already pointing to base - nothing
     * to do, no intermediate images */
    if (new_top_bs->backing_hd == base) {
        ret = 0;
        goto exit;
    }

    intermediate = top;

    /* now we will go down through the list, and add each BDS we find
     * into our deletion queue, until we hit the 'base'
     */
    while (intermediate) {
        intermediate_state = g_malloc0(sizeof(BlkIntermediateStates));
        intermediate_state->bs = intermediate;
        QSIMPLEQ_INSERT_TAIL(&states_to_delete, intermediate_state, entry);

        if (intermediate->backing_hd == base) {
            base_bs = intermediate->backing_hd;
            break;
        }
        intermediate = intermediate->backing_hd;
    }
    if (base_bs == NULL) {
        /* something went wrong, we did not end at the base. safely
         * unravel everything, and exit with error */
        goto exit;
    }

    /* success - we can delete the intermediate states, and link top->base */
    ret = bdrv_change_backing_file(new_top_bs, base_bs->filename,
                                   base_bs->drv ? base_bs->drv->format_name : "");
    if (ret) {
        goto exit;
    }
    new_top_bs->backing_hd = base_bs;


    QSIMPLEQ_FOREACH_SAFE(intermediate_state, &states_to_delete, entry, next) {
        /* so that bdrv_close() does not recursively close the chain */
        intermediate_state->bs->backing_hd = NULL;
F
Fam Zheng 已提交
2430
        bdrv_unref(intermediate_state->bs);
2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441
    }
    ret = 0;

exit:
    QSIMPLEQ_FOREACH_SAFE(intermediate_state, &states_to_delete, entry, next) {
        g_free(intermediate_state);
    }
    return ret;
}


2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454
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
2455 2456 2457 2458
    if (offset < 0)
        return -EIO;

    if ((offset > len) || (len - offset < size))
2459 2460 2461 2462 2463 2464 2465 2466
        return -EIO;

    return 0;
}

static int bdrv_check_request(BlockDriverState *bs, int64_t sector_num,
                              int nb_sectors)
{
2467 2468
    return bdrv_check_byte_request(bs, sector_num * BDRV_SECTOR_SIZE,
                                   nb_sectors * BDRV_SECTOR_SIZE);
2469 2470
}

2471 2472 2473 2474 2475 2476 2477
typedef struct RwCo {
    BlockDriverState *bs;
    int64_t sector_num;
    int nb_sectors;
    QEMUIOVector *qiov;
    bool is_write;
    int ret;
P
Peter Lieven 已提交
2478
    BdrvRequestFlags flags;
2479 2480 2481
} RwCo;

static void coroutine_fn bdrv_rw_co_entry(void *opaque)
B
bellard 已提交
2482
{
2483
    RwCo *rwco = opaque;
B
bellard 已提交
2484

2485 2486
    if (!rwco->is_write) {
        rwco->ret = bdrv_co_do_readv(rwco->bs, rwco->sector_num,
P
Peter Lieven 已提交
2487 2488
                                     rwco->nb_sectors, rwco->qiov,
                                     rwco->flags);
2489 2490
    } else {
        rwco->ret = bdrv_co_do_writev(rwco->bs, rwco->sector_num,
P
Peter Lieven 已提交
2491 2492
                                      rwco->nb_sectors, rwco->qiov,
                                      rwco->flags);
2493 2494
    }
}
2495

2496
/*
2497
 * Process a vectored synchronous request using coroutines
2498
 */
2499
static int bdrv_rwv_co(BlockDriverState *bs, int64_t sector_num,
P
Peter Lieven 已提交
2500 2501
                       QEMUIOVector *qiov, bool is_write,
                       BdrvRequestFlags flags)
2502 2503 2504 2505 2506
{
    Coroutine *co;
    RwCo rwco = {
        .bs = bs,
        .sector_num = sector_num,
2507 2508
        .nb_sectors = qiov->size >> BDRV_SECTOR_BITS,
        .qiov = qiov,
2509 2510
        .is_write = is_write,
        .ret = NOT_DONE,
P
Peter Lieven 已提交
2511
        .flags = flags,
2512
    };
2513
    assert((qiov->size & (BDRV_SECTOR_SIZE - 1)) == 0);
2514

2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525
    /**
     * 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);
    }

2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537
    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
2538

2539 2540 2541 2542
/*
 * Process a synchronous request using coroutines
 */
static int bdrv_rw_co(BlockDriverState *bs, int64_t sector_num, uint8_t *buf,
P
Peter Lieven 已提交
2543
                      int nb_sectors, bool is_write, BdrvRequestFlags flags)
2544 2545 2546 2547 2548 2549 2550 2551
{
    QEMUIOVector qiov;
    struct iovec iov = {
        .iov_base = (void *)buf,
        .iov_len = nb_sectors * BDRV_SECTOR_SIZE,
    };

    qemu_iovec_init_external(&qiov, &iov, 1);
P
Peter Lieven 已提交
2552
    return bdrv_rwv_co(bs, sector_num, &qiov, is_write, flags);
2553 2554
}

2555 2556 2557 2558
/* 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)
{
P
Peter Lieven 已提交
2559
    return bdrv_rw_co(bs, sector_num, buf, nb_sectors, false, 0);
B
bellard 已提交
2560 2561
}

2562 2563 2564 2565 2566 2567 2568 2569 2570
/* 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;
2571
    ret = bdrv_read(bs, sector_num, buf, nb_sectors);
2572 2573 2574 2575
    bs->io_limits_enabled = enabled;
    return ret;
}

2576
/* Return < 0 if error. Important errors are:
B
bellard 已提交
2577 2578 2579 2580 2581
  -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
*/
2582
int bdrv_write(BlockDriverState *bs, int64_t sector_num,
B
bellard 已提交
2583 2584
               const uint8_t *buf, int nb_sectors)
{
P
Peter Lieven 已提交
2585
    return bdrv_rw_co(bs, sector_num, (uint8_t *)buf, nb_sectors, true, 0);
B
bellard 已提交
2586 2587
}

2588 2589
int bdrv_writev(BlockDriverState *bs, int64_t sector_num, QEMUIOVector *qiov)
{
P
Peter Lieven 已提交
2590 2591 2592
    return bdrv_rwv_co(bs, sector_num, qiov, true, 0);
}

2593 2594
int bdrv_write_zeroes(BlockDriverState *bs, int64_t sector_num,
                      int nb_sectors, BdrvRequestFlags flags)
P
Peter Lieven 已提交
2595 2596
{
    return bdrv_rw_co(bs, sector_num, NULL, nb_sectors, true,
2597
                      BDRV_REQ_ZERO_WRITE | flags);
2598 2599
}

P
Peter Lieven 已提交
2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622
/*
 * Completely zero out a block device with the help of bdrv_write_zeroes.
 * The operation is sped up by checking the block status and only writing
 * zeroes to the device if they currently do not return zeroes. Optional
 * flags are passed through to bdrv_write_zeroes (e.g. BDRV_REQ_MAY_UNMAP).
 *
 * Returns < 0 on error, 0 on success. For error codes see bdrv_write().
 */
int bdrv_make_zero(BlockDriverState *bs, BdrvRequestFlags flags)
{
    int64_t target_size = bdrv_getlength(bs) / BDRV_SECTOR_SIZE;
    int64_t ret, nb_sectors, sector_num = 0;
    int n;

    for (;;) {
        nb_sectors = target_size - sector_num;
        if (nb_sectors <= 0) {
            return 0;
        }
        if (nb_sectors > INT_MAX) {
            nb_sectors = INT_MAX;
        }
        ret = bdrv_get_block_status(bs, sector_num, nb_sectors, &n);
2623 2624 2625 2626 2627
        if (ret < 0) {
            error_report("error getting block status at sector %" PRId64 ": %s",
                         sector_num, strerror(-ret));
            return ret;
        }
P
Peter Lieven 已提交
2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641
        if (ret & BDRV_BLOCK_ZERO) {
            sector_num += n;
            continue;
        }
        ret = bdrv_write_zeroes(bs, sector_num, n, flags);
        if (ret < 0) {
            error_report("error writing zeroes at sector %" PRId64 ": %s",
                         sector_num, strerror(-ret));
            return ret;
        }
        sector_num += n;
    }
}

2642 2643
int bdrv_pread(BlockDriverState *bs, int64_t offset,
               void *buf, int count1)
B
bellard 已提交
2644
{
J
Jan Kiszka 已提交
2645
    uint8_t tmp_buf[BDRV_SECTOR_SIZE];
B
bellard 已提交
2646 2647
    int len, nb_sectors, count;
    int64_t sector_num;
2648
    int ret;
B
bellard 已提交
2649 2650 2651

    count = count1;
    /* first read to align to sector start */
J
Jan Kiszka 已提交
2652
    len = (BDRV_SECTOR_SIZE - offset) & (BDRV_SECTOR_SIZE - 1);
B
bellard 已提交
2653 2654
    if (len > count)
        len = count;
J
Jan Kiszka 已提交
2655
    sector_num = offset >> BDRV_SECTOR_BITS;
B
bellard 已提交
2656
    if (len > 0) {
2657 2658
        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
J
Jan Kiszka 已提交
2659
        memcpy(buf, tmp_buf + (offset & (BDRV_SECTOR_SIZE - 1)), len);
B
bellard 已提交
2660 2661 2662 2663 2664 2665 2666 2667
        count -= len;
        if (count == 0)
            return count1;
        sector_num++;
        buf += len;
    }

    /* read the sectors "in place" */
J
Jan Kiszka 已提交
2668
    nb_sectors = count >> BDRV_SECTOR_BITS;
B
bellard 已提交
2669
    if (nb_sectors > 0) {
2670 2671
        if ((ret = bdrv_read(bs, sector_num, buf, nb_sectors)) < 0)
            return ret;
B
bellard 已提交
2672
        sector_num += nb_sectors;
J
Jan Kiszka 已提交
2673
        len = nb_sectors << BDRV_SECTOR_BITS;
B
bellard 已提交
2674 2675 2676 2677 2678 2679
        buf += len;
        count -= len;
    }

    /* add data from the last sector */
    if (count > 0) {
2680 2681
        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
B
bellard 已提交
2682 2683 2684 2685 2686
        memcpy(buf, tmp_buf, count);
    }
    return count1;
}

2687
int bdrv_pwritev(BlockDriverState *bs, int64_t offset, QEMUIOVector *qiov)
B
bellard 已提交
2688
{
J
Jan Kiszka 已提交
2689
    uint8_t tmp_buf[BDRV_SECTOR_SIZE];
B
bellard 已提交
2690 2691
    int len, nb_sectors, count;
    int64_t sector_num;
2692
    int ret;
B
bellard 已提交
2693

2694 2695
    count = qiov->size;

B
bellard 已提交
2696
    /* first write to align to sector start */
J
Jan Kiszka 已提交
2697
    len = (BDRV_SECTOR_SIZE - offset) & (BDRV_SECTOR_SIZE - 1);
B
bellard 已提交
2698 2699
    if (len > count)
        len = count;
J
Jan Kiszka 已提交
2700
    sector_num = offset >> BDRV_SECTOR_BITS;
B
bellard 已提交
2701
    if (len > 0) {
2702 2703
        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
2704 2705
        qemu_iovec_to_buf(qiov, 0, tmp_buf + (offset & (BDRV_SECTOR_SIZE - 1)),
                          len);
2706 2707
        if ((ret = bdrv_write(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
B
bellard 已提交
2708 2709
        count -= len;
        if (count == 0)
2710
            return qiov->size;
B
bellard 已提交
2711 2712 2713 2714
        sector_num++;
    }

    /* write the sectors "in place" */
J
Jan Kiszka 已提交
2715
    nb_sectors = count >> BDRV_SECTOR_BITS;
B
bellard 已提交
2716
    if (nb_sectors > 0) {
2717 2718 2719 2720 2721 2722 2723 2724
        QEMUIOVector qiov_inplace;

        qemu_iovec_init(&qiov_inplace, qiov->niov);
        qemu_iovec_concat(&qiov_inplace, qiov, len,
                          nb_sectors << BDRV_SECTOR_BITS);
        ret = bdrv_writev(bs, sector_num, &qiov_inplace);
        qemu_iovec_destroy(&qiov_inplace);
        if (ret < 0) {
2725
            return ret;
2726 2727
        }

B
bellard 已提交
2728
        sector_num += nb_sectors;
J
Jan Kiszka 已提交
2729
        len = nb_sectors << BDRV_SECTOR_BITS;
B
bellard 已提交
2730 2731 2732 2733 2734
        count -= len;
    }

    /* add data from the last sector */
    if (count > 0) {
2735 2736
        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
2737
        qemu_iovec_to_buf(qiov, qiov->size - count, tmp_buf, count);
2738 2739
        if ((ret = bdrv_write(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
B
bellard 已提交
2740
    }
2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754
    return qiov->size;
}

int bdrv_pwrite(BlockDriverState *bs, int64_t offset,
                const void *buf, int count1)
{
    QEMUIOVector qiov;
    struct iovec iov = {
        .iov_base   = (void *) buf,
        .iov_len    = count1,
    };

    qemu_iovec_init_external(&qiov, &iov, 1);
    return bdrv_pwritev(bs, offset, &qiov);
B
bellard 已提交
2755 2756
}

K
Kevin Wolf 已提交
2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772
/*
 * 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;
    }

2773 2774
    /* No flush needed for cache modes that already do it */
    if (bs->enable_write_cache) {
K
Kevin Wolf 已提交
2775 2776 2777 2778 2779 2780
        bdrv_flush(bs);
    }

    return 0;
}

2781
static int coroutine_fn bdrv_co_do_copy_on_readv(BlockDriverState *bs,
S
Stefan Hajnoczi 已提交
2782 2783 2784 2785 2786 2787 2788 2789 2790
        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;

2791
    BlockDriver *drv = bs->drv;
S
Stefan Hajnoczi 已提交
2792 2793 2794 2795 2796 2797 2798 2799 2800 2801
    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.
     */
2802 2803
    bdrv_round_to_clusters(bs, sector_num, nb_sectors,
                           &cluster_sector_num, &cluster_nb_sectors);
S
Stefan Hajnoczi 已提交
2804

2805 2806
    trace_bdrv_co_do_copy_on_readv(bs, sector_num, nb_sectors,
                                   cluster_sector_num, cluster_nb_sectors);
S
Stefan Hajnoczi 已提交
2807 2808 2809 2810 2811

    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);

2812 2813
    ret = drv->bdrv_co_readv(bs, cluster_sector_num, cluster_nb_sectors,
                             &bounce_qiov);
S
Stefan Hajnoczi 已提交
2814 2815 2816 2817
    if (ret < 0) {
        goto err;
    }

2818 2819
    if (drv->bdrv_co_write_zeroes &&
        buffer_is_zero(bounce_buffer, iov.iov_len)) {
K
Kevin Wolf 已提交
2820
        ret = bdrv_co_do_write_zeroes(bs, cluster_sector_num,
2821
                                      cluster_nb_sectors, 0);
2822
    } else {
2823 2824 2825
        /* This does not change the data on the disk, it is not necessary
         * to flush even in cache=writethrough mode.
         */
2826
        ret = drv->bdrv_co_writev(bs, cluster_sector_num, cluster_nb_sectors,
S
Stefan Hajnoczi 已提交
2827
                                  &bounce_qiov);
2828 2829
    }

S
Stefan Hajnoczi 已提交
2830 2831 2832 2833 2834 2835 2836 2837 2838
    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;
2839 2840
    qemu_iovec_from_buf(qiov, 0, bounce_buffer + skip_bytes,
                        nb_sectors * BDRV_SECTOR_SIZE);
S
Stefan Hajnoczi 已提交
2841 2842 2843 2844 2845 2846

err:
    qemu_vfree(bounce_buffer);
    return ret;
}

2847 2848 2849 2850
/*
 * Handle a read request in coroutine context
 */
static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,
2851 2852
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags)
K
Kevin Wolf 已提交
2853 2854
{
    BlockDriver *drv = bs->drv;
S
Stefan Hajnoczi 已提交
2855 2856
    BdrvTrackedRequest req;
    int ret;
K
Kevin Wolf 已提交
2857 2858 2859 2860 2861 2862 2863 2864

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

2865
    if (bs->copy_on_read) {
2866 2867 2868 2869 2870 2871 2872
        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) {
2873 2874 2875
        wait_for_overlapping_requests(bs, sector_num, nb_sectors);
    }

2876 2877 2878 2879 2880
    /* throttling disk I/O */
    if (bs->io_limits_enabled) {
        bdrv_io_limits_intercept(bs, nb_sectors, false);
    }

S
Stefan Hajnoczi 已提交
2881
    tracked_request_begin(&req, bs, sector_num, nb_sectors, false);
S
Stefan Hajnoczi 已提交
2882

2883
    if (flags & BDRV_REQ_COPY_ON_READ) {
S
Stefan Hajnoczi 已提交
2884 2885
        int pnum;

2886
        ret = bdrv_is_allocated(bs, sector_num, nb_sectors, &pnum);
S
Stefan Hajnoczi 已提交
2887 2888 2889 2890 2891
        if (ret < 0) {
            goto out;
        }

        if (!ret || pnum != nb_sectors) {
2892
            ret = bdrv_co_do_copy_on_readv(bs, sector_num, nb_sectors, qiov);
S
Stefan Hajnoczi 已提交
2893 2894 2895 2896
            goto out;
        }
    }

2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908
    if (!(bs->zero_beyond_eof && bs->growable)) {
        ret = drv->bdrv_co_readv(bs, sector_num, nb_sectors, qiov);
    } else {
        /* Read zeros after EOF of growable BDSes */
        int64_t len, total_sectors, max_nb_sectors;

        len = bdrv_getlength(bs);
        if (len < 0) {
            ret = len;
            goto out;
        }

2909
        total_sectors = DIV_ROUND_UP(len, BDRV_SECTOR_SIZE);
2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925
        max_nb_sectors = MAX(0, total_sectors - sector_num);
        if (max_nb_sectors > 0) {
            ret = drv->bdrv_co_readv(bs, sector_num,
                                     MIN(nb_sectors, max_nb_sectors), qiov);
        } else {
            ret = 0;
        }

        /* Reading beyond end of file is supposed to produce zeroes */
        if (ret == 0 && total_sectors < sector_num + nb_sectors) {
            uint64_t offset = MAX(0, total_sectors - sector_num);
            uint64_t bytes = (sector_num + nb_sectors - offset) *
                              BDRV_SECTOR_SIZE;
            qemu_iovec_memset(qiov, offset * BDRV_SECTOR_SIZE, 0, bytes);
        }
    }
S
Stefan Hajnoczi 已提交
2926 2927

out:
S
Stefan Hajnoczi 已提交
2928
    tracked_request_end(&req);
2929 2930 2931 2932 2933

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

S
Stefan Hajnoczi 已提交
2934
    return ret;
K
Kevin Wolf 已提交
2935 2936
}

2937
int coroutine_fn bdrv_co_readv(BlockDriverState *bs, int64_t sector_num,
K
Kevin Wolf 已提交
2938 2939
    int nb_sectors, QEMUIOVector *qiov)
{
2940
    trace_bdrv_co_readv(bs, sector_num, nb_sectors);
K
Kevin Wolf 已提交
2941

2942 2943 2944 2945 2946 2947 2948 2949 2950 2951
    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);
2952 2953
}

2954 2955 2956 2957 2958
/* if no limit is specified in the BlockLimits use a default
 * of 32768 512-byte sectors (16 MiB) per request.
 */
#define MAX_WRITE_ZEROES_DEFAULT 32768

2959
static int coroutine_fn bdrv_co_do_write_zeroes(BlockDriverState *bs,
2960
    int64_t sector_num, int nb_sectors, BdrvRequestFlags flags)
2961 2962 2963
{
    BlockDriver *drv = bs->drv;
    QEMUIOVector qiov;
2964 2965
    struct iovec iov = {0};
    int ret = 0;
2966

2967 2968
    int max_write_zeroes = bs->bl.max_write_zeroes ?
                           bs->bl.max_write_zeroes : MAX_WRITE_ZEROES_DEFAULT;
K
Kevin Wolf 已提交
2969

2970 2971 2972
    while (nb_sectors > 0 && !ret) {
        int num = nb_sectors;

2973 2974 2975 2976 2977 2978 2979
        /* Align request.  Block drivers can expect the "bulk" of the request
         * to be aligned.
         */
        if (bs->bl.write_zeroes_alignment
            && num > bs->bl.write_zeroes_alignment) {
            if (sector_num % bs->bl.write_zeroes_alignment != 0) {
                /* Make a small request up to the first aligned sector.  */
2980
                num = bs->bl.write_zeroes_alignment;
2981 2982 2983 2984 2985 2986
                num -= sector_num % bs->bl.write_zeroes_alignment;
            } else if ((sector_num + num) % bs->bl.write_zeroes_alignment != 0) {
                /* Shorten the request to the last aligned sector.  num cannot
                 * underflow because num > bs->bl.write_zeroes_alignment.
                 */
                num -= (sector_num + num) % bs->bl.write_zeroes_alignment;
2987
            }
K
Kevin Wolf 已提交
2988
        }
2989

2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004
        /* limit request size */
        if (num > max_write_zeroes) {
            num = max_write_zeroes;
        }

        ret = -ENOTSUP;
        /* First try the efficient write zeroes operation */
        if (drv->bdrv_co_write_zeroes) {
            ret = drv->bdrv_co_write_zeroes(bs, sector_num, num, flags);
        }

        if (ret == -ENOTSUP) {
            /* Fall back to bounce buffer if write zeroes is unsupported */
            iov.iov_len = num * BDRV_SECTOR_SIZE;
            if (iov.iov_base == NULL) {
3005 3006
                iov.iov_base = qemu_blockalign(bs, num * BDRV_SECTOR_SIZE);
                memset(iov.iov_base, 0, num * BDRV_SECTOR_SIZE);
3007 3008
            }
            qemu_iovec_init_external(&qiov, &iov, 1);
3009

3010
            ret = drv->bdrv_co_writev(bs, sector_num, num, &qiov);
3011 3012 3013 3014 3015 3016 3017 3018

            /* Keep bounce buffer around if it is big enough for all
             * all future requests.
             */
            if (num < max_write_zeroes) {
                qemu_vfree(iov.iov_base);
                iov.iov_base = NULL;
            }
3019 3020 3021 3022 3023
        }

        sector_num += num;
        nb_sectors -= num;
    }
3024 3025 3026 3027 3028

    qemu_vfree(iov.iov_base);
    return ret;
}

3029 3030 3031 3032
/*
 * Handle a write request in coroutine context
 */
static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs,
3033 3034
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags)
3035 3036
{
    BlockDriver *drv = bs->drv;
S
Stefan Hajnoczi 已提交
3037
    BdrvTrackedRequest req;
3038
    int ret;
K
Kevin Wolf 已提交
3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049

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

3050
    if (bs->copy_on_read_in_flight) {
3051 3052 3053
        wait_for_overlapping_requests(bs, sector_num, nb_sectors);
    }

3054 3055 3056 3057 3058
    /* throttling disk I/O */
    if (bs->io_limits_enabled) {
        bdrv_io_limits_intercept(bs, nb_sectors, true);
    }

S
Stefan Hajnoczi 已提交
3059 3060
    tracked_request_begin(&req, bs, sector_num, nb_sectors, true);

3061 3062 3063 3064 3065
    ret = notifier_with_return_list_notify(&bs->before_write_notifiers, &req);

    if (ret < 0) {
        /* Do nothing, write notifier decided to fail this request */
    } else if (flags & BDRV_REQ_ZERO_WRITE) {
3066
        ret = bdrv_co_do_write_zeroes(bs, sector_num, nb_sectors, flags);
3067 3068 3069
    } else {
        ret = drv->bdrv_co_writev(bs, sector_num, nb_sectors, qiov);
    }
3070

3071 3072 3073 3074
    if (ret == 0 && !bs->enable_write_cache) {
        ret = bdrv_co_flush(bs);
    }

F
Fam Zheng 已提交
3075
    bdrv_set_dirty(bs, sector_num, nb_sectors);
K
Kevin Wolf 已提交
3076 3077 3078 3079

    if (bs->wr_highest_sector < sector_num + nb_sectors - 1) {
        bs->wr_highest_sector = sector_num + nb_sectors - 1;
    }
3080 3081 3082
    if (bs->growable && ret >= 0) {
        bs->total_sectors = MAX(bs->total_sectors, sector_num + nb_sectors);
    }
K
Kevin Wolf 已提交
3083

S
Stefan Hajnoczi 已提交
3084 3085
    tracked_request_end(&req);

3086
    return ret;
K
Kevin Wolf 已提交
3087 3088
}

3089 3090 3091 3092 3093
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);

3094 3095 3096 3097
    return bdrv_co_do_writev(bs, sector_num, nb_sectors, qiov, 0);
}

int coroutine_fn bdrv_co_write_zeroes(BlockDriverState *bs,
3098 3099
                                      int64_t sector_num, int nb_sectors,
                                      BdrvRequestFlags flags)
3100
{
3101
    trace_bdrv_co_write_zeroes(bs, sector_num, nb_sectors, flags);
3102

3103 3104 3105 3106
    if (!(bs->open_flags & BDRV_O_UNMAP)) {
        flags &= ~BDRV_REQ_MAY_UNMAP;
    }

3107
    return bdrv_co_do_writev(bs, sector_num, nb_sectors, NULL,
3108
                             BDRV_REQ_ZERO_WRITE | flags);
3109 3110
}

B
bellard 已提交
3111 3112 3113 3114 3115 3116
/**
 * Truncate file to 'offset' bytes (needed only for file protocols)
 */
int bdrv_truncate(BlockDriverState *bs, int64_t offset)
{
    BlockDriver *drv = bs->drv;
3117
    int ret;
B
bellard 已提交
3118
    if (!drv)
B
bellard 已提交
3119
        return -ENOMEDIUM;
B
bellard 已提交
3120 3121
    if (!drv->bdrv_truncate)
        return -ENOTSUP;
3122 3123
    if (bs->read_only)
        return -EACCES;
M
Marcelo Tosatti 已提交
3124 3125
    if (bdrv_in_use(bs))
        return -EBUSY;
3126 3127 3128
    ret = drv->bdrv_truncate(bs, offset);
    if (ret == 0) {
        ret = refresh_total_sectors(bs, offset >> BDRV_SECTOR_BITS);
3129
        bdrv_dev_resize_cb(bs);
3130 3131
    }
    return ret;
B
bellard 已提交
3132 3133
}

3134 3135 3136 3137 3138 3139 3140 3141 3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152
/**
 * 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
3153 3154 3155 3156 3157 3158 3159
/**
 * 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
3160
        return -ENOMEDIUM;
3161

3162 3163 3164 3165
    if (drv->has_variable_length) {
        int ret = refresh_total_sectors(bs, bs->total_sectors);
        if (ret < 0) {
            return ret;
3166
        }
B
bellard 已提交
3167
    }
3168
    return bs->total_sectors * BDRV_SECTOR_SIZE;
B
bellard 已提交
3169 3170
}

B
bellard 已提交
3171
/* return 0 as number of sectors if no device present or error */
3172
void bdrv_get_geometry(BlockDriverState *bs, uint64_t *nb_sectors_ptr)
B
bellard 已提交
3173
{
B
bellard 已提交
3174 3175 3176 3177 3178
    int64_t length;
    length = bdrv_getlength(bs);
    if (length < 0)
        length = 0;
    else
J
Jan Kiszka 已提交
3179
        length = length >> BDRV_SECTOR_BITS;
B
bellard 已提交
3180
    *nb_sectors_ptr = length;
B
bellard 已提交
3181
}
B
bellard 已提交
3182

3183 3184
void bdrv_set_on_error(BlockDriverState *bs, BlockdevOnError on_read_error,
                       BlockdevOnError on_write_error)
3185 3186 3187 3188 3189
{
    bs->on_read_error = on_read_error;
    bs->on_write_error = on_write_error;
}

3190
BlockdevOnError bdrv_get_on_error(BlockDriverState *bs, bool is_read)
3191 3192 3193 3194
{
    return is_read ? bs->on_read_error : bs->on_write_error;
}

3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220
BlockErrorAction bdrv_get_error_action(BlockDriverState *bs, bool is_read, int error)
{
    BlockdevOnError on_err = is_read ? bs->on_read_error : bs->on_write_error;

    switch (on_err) {
    case BLOCKDEV_ON_ERROR_ENOSPC:
        return (error == ENOSPC) ? BDRV_ACTION_STOP : BDRV_ACTION_REPORT;
    case BLOCKDEV_ON_ERROR_STOP:
        return BDRV_ACTION_STOP;
    case BLOCKDEV_ON_ERROR_REPORT:
        return BDRV_ACTION_REPORT;
    case BLOCKDEV_ON_ERROR_IGNORE:
        return BDRV_ACTION_IGNORE;
    default:
        abort();
    }
}

/* This is done by device models because, while the block layer knows
 * about the error, it does not know whether an operation comes from
 * the device or the block layer (from a job, for example).
 */
void bdrv_error_action(BlockDriverState *bs, BlockErrorAction action,
                       bool is_read, int error)
{
    assert(error >= 0);
P
Paolo Bonzini 已提交
3221
    bdrv_emit_qmp_error_event(bs, QEVENT_BLOCK_IO_ERROR, action, is_read);
3222 3223 3224 3225 3226 3227
    if (action == BDRV_ACTION_STOP) {
        vm_stop(RUN_STATE_IO_ERROR);
        bdrv_iostatus_set_err(bs, error);
    }
}

B
bellard 已提交
3228 3229 3230 3231 3232
int bdrv_is_read_only(BlockDriverState *bs)
{
    return bs->read_only;
}

3233 3234 3235 3236 3237
int bdrv_is_sg(BlockDriverState *bs)
{
    return bs->sg;
}

3238 3239 3240 3241 3242
int bdrv_enable_write_cache(BlockDriverState *bs)
{
    return bs->enable_write_cache;
}

3243 3244 3245
void bdrv_set_enable_write_cache(BlockDriverState *bs, bool wce)
{
    bs->enable_write_cache = wce;
3246 3247 3248 3249 3250 3251 3252

    /* so a reopen() will preserve wce */
    if (wce) {
        bs->open_flags |= BDRV_O_CACHE_WB;
    } else {
        bs->open_flags &= ~BDRV_O_CACHE_WB;
    }
3253 3254
}

B
bellard 已提交
3255 3256 3257 3258 3259 3260 3261
int bdrv_is_encrypted(BlockDriverState *bs)
{
    if (bs->backing_hd && bs->backing_hd->encrypted)
        return 1;
    return bs->encrypted;
}

3262 3263 3264 3265 3266 3267 3268 3269 3270
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
3271 3272 3273 3274 3275 3276 3277 3278 3279 3280
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;
    }
3281 3282 3283 3284 3285
    if (!bs->encrypted) {
        return -EINVAL;
    } else if (!bs->drv || !bs->drv->bdrv_set_key) {
        return -ENOMEDIUM;
    }
3286
    ret = bs->drv->bdrv_set_key(bs, key);
3287 3288 3289 3290 3291
    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 */
3292
        bdrv_dev_change_media_cb(bs, true);
3293
    }
3294
    return ret;
B
bellard 已提交
3295 3296
}

3297
const char *bdrv_get_format_name(BlockDriverState *bs)
B
bellard 已提交
3298
{
3299
    return bs->drv ? bs->drv->format_name : NULL;
B
bellard 已提交
3300 3301
}

3302
void bdrv_iterate_format(void (*it)(void *opaque, const char *name),
B
bellard 已提交
3303 3304 3305 3306
                         void *opaque)
{
    BlockDriver *drv;

3307
    QLIST_FOREACH(drv, &bdrv_drivers, list) {
B
bellard 已提交
3308 3309 3310 3311
        it(opaque, drv->format_name);
    }
}

3312
/* This function is to find block backend bs */
B
bellard 已提交
3313 3314 3315 3316
BlockDriverState *bdrv_find(const char *name)
{
    BlockDriverState *bs;

3317
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
3318
        if (!strcmp(name, bs->device_name)) {
B
bellard 已提交
3319
            return bs;
3320
        }
B
bellard 已提交
3321 3322 3323 3324
    }
    return NULL;
}

3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339
/* This function is to find a node in the bs graph */
BlockDriverState *bdrv_find_node(const char *node_name)
{
    BlockDriverState *bs;

    assert(node_name);

    QTAILQ_FOREACH(bs, &graph_bdrv_states, node_list) {
        if (!strcmp(node_name, bs->node_name)) {
            return bs;
        }
    }
    return NULL;
}

3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356
/* Put this QMP function here so it can access the static graph_bdrv_states. */
BlockDeviceInfoList *bdrv_named_nodes_list(void)
{
    BlockDeviceInfoList *list, *entry;
    BlockDriverState *bs;

    list = NULL;
    QTAILQ_FOREACH(bs, &graph_bdrv_states, node_list) {
        entry = g_malloc0(sizeof(*entry));
        entry->value = bdrv_block_device_info(bs);
        entry->next = list;
        list = entry;
    }

    return list;
}

3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388
BlockDriverState *bdrv_lookup_bs(const char *device,
                                 const char *node_name,
                                 Error **errp)
{
    BlockDriverState *bs = NULL;

    if ((!device && !node_name) || (device && node_name)) {
        error_setg(errp, "Use either device or node-name but not both");
        return NULL;
    }

    if (device) {
        bs = bdrv_find(device);

        if (!bs) {
            error_set(errp, QERR_DEVICE_NOT_FOUND, device);
            return NULL;
        }

        return bs;
    }

    bs = bdrv_find_node(node_name);

    if (!bs) {
        error_set(errp, QERR_DEVICE_NOT_FOUND, node_name);
        return NULL;
    }

    return bs;
}

M
Markus Armbruster 已提交
3389 3390 3391 3392 3393
BlockDriverState *bdrv_next(BlockDriverState *bs)
{
    if (!bs) {
        return QTAILQ_FIRST(&bdrv_states);
    }
3394
    return QTAILQ_NEXT(bs, device_list);
M
Markus Armbruster 已提交
3395 3396
}

3397
void bdrv_iterate(void (*it)(void *opaque, BlockDriverState *bs), void *opaque)
B
bellard 已提交
3398 3399 3400
{
    BlockDriverState *bs;

3401
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
3402
        it(opaque, bs);
B
bellard 已提交
3403 3404 3405
    }
}

B
bellard 已提交
3406 3407 3408 3409 3410
const char *bdrv_get_device_name(BlockDriverState *bs)
{
    return bs->device_name;
}

M
Markus Armbruster 已提交
3411 3412 3413 3414 3415
int bdrv_get_flags(BlockDriverState *bs)
{
    return bs->open_flags;
}

3416
int bdrv_flush_all(void)
A
aliguori 已提交
3417 3418
{
    BlockDriverState *bs;
3419
    int result = 0;
A
aliguori 已提交
3420

3421
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
3422 3423 3424 3425
        int ret = bdrv_flush(bs);
        if (ret < 0 && !result) {
            result = ret;
        }
3426
    }
3427 3428

    return result;
A
aliguori 已提交
3429 3430
}

3431 3432 3433 3434 3435
int bdrv_has_zero_init_1(BlockDriverState *bs)
{
    return 1;
}

K
Kevin Wolf 已提交
3436 3437 3438 3439
int bdrv_has_zero_init(BlockDriverState *bs)
{
    assert(bs->drv);

3440 3441 3442 3443 3444
    /* If BS is a copy on write image, it is initialized to
       the contents of the base image, which may not be zeroes.  */
    if (bs->backing_hd) {
        return 0;
    }
K
Kevin Wolf 已提交
3445 3446
    if (bs->drv->bdrv_has_zero_init) {
        return bs->drv->bdrv_has_zero_init(bs);
K
Kevin Wolf 已提交
3447 3448
    }

3449 3450
    /* safe default */
    return 0;
K
Kevin Wolf 已提交
3451 3452
}

3453 3454 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 3482
bool bdrv_unallocated_blocks_are_zero(BlockDriverState *bs)
{
    BlockDriverInfo bdi;

    if (bs->backing_hd) {
        return false;
    }

    if (bdrv_get_info(bs, &bdi) == 0) {
        return bdi.unallocated_blocks_are_zero;
    }

    return false;
}

bool bdrv_can_write_zeroes_with_unmap(BlockDriverState *bs)
{
    BlockDriverInfo bdi;

    if (bs->backing_hd || !(bs->open_flags & BDRV_O_UNMAP)) {
        return false;
    }

    if (bdrv_get_info(bs, &bdi) == 0) {
        return bdi.can_write_zeroes_with_unmap;
    }

    return false;
}

3483
typedef struct BdrvCoGetBlockStatusData {
3484
    BlockDriverState *bs;
3485
    BlockDriverState *base;
3486 3487 3488
    int64_t sector_num;
    int nb_sectors;
    int *pnum;
3489
    int64_t ret;
3490
    bool done;
3491
} BdrvCoGetBlockStatusData;
3492

3493 3494 3495 3496 3497
/*
 * 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.
 *
3498 3499 3500
 * If 'sector_num' is beyond the end of the disk image the return value is 0
 * and 'pnum' is set to 0.
 *
3501 3502 3503 3504
 * '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.
 *
3505 3506
 * '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.
3507
 */
3508 3509 3510
static int64_t coroutine_fn bdrv_co_get_block_status(BlockDriverState *bs,
                                                     int64_t sector_num,
                                                     int nb_sectors, int *pnum)
3511
{
3512
    int64_t length;
3513
    int64_t n;
3514
    int64_t ret, ret2;
3515

3516 3517 3518 3519 3520 3521
    length = bdrv_getlength(bs);
    if (length < 0) {
        return length;
    }

    if (sector_num >= (length >> BDRV_SECTOR_BITS)) {
3522 3523 3524 3525 3526 3527 3528 3529 3530
        *pnum = 0;
        return 0;
    }

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

3531
    if (!bs->drv->bdrv_co_get_block_status) {
3532
        *pnum = nb_sectors;
3533 3534 3535 3536 3537
        ret = BDRV_BLOCK_DATA;
        if (bs->drv->protocol_name) {
            ret |= BDRV_BLOCK_OFFSET_VALID | (sector_num * BDRV_SECTOR_SIZE);
        }
        return ret;
3538
    }
3539

3540 3541
    ret = bs->drv->bdrv_co_get_block_status(bs, sector_num, nb_sectors, pnum);
    if (ret < 0) {
3542
        *pnum = 0;
3543 3544 3545
        return ret;
    }

3546 3547 3548 3549 3550 3551
    if (ret & BDRV_BLOCK_RAW) {
        assert(ret & BDRV_BLOCK_OFFSET_VALID);
        return bdrv_get_block_status(bs->file, ret >> BDRV_SECTOR_BITS,
                                     *pnum, pnum);
    }

3552 3553
    if (!(ret & BDRV_BLOCK_DATA) && !(ret & BDRV_BLOCK_ZERO)) {
        if (bdrv_unallocated_blocks_are_zero(bs)) {
3554
            ret |= BDRV_BLOCK_ZERO;
3555
        } else if (bs->backing_hd) {
3556 3557 3558 3559 3560 3561
            BlockDriverState *bs2 = bs->backing_hd;
            int64_t length2 = bdrv_getlength(bs2);
            if (length2 >= 0 && sector_num >= (length2 >> BDRV_SECTOR_BITS)) {
                ret |= BDRV_BLOCK_ZERO;
            }
        }
3562
    }
3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576

    if (bs->file &&
        (ret & BDRV_BLOCK_DATA) && !(ret & BDRV_BLOCK_ZERO) &&
        (ret & BDRV_BLOCK_OFFSET_VALID)) {
        ret2 = bdrv_co_get_block_status(bs->file, ret >> BDRV_SECTOR_BITS,
                                        *pnum, pnum);
        if (ret2 >= 0) {
            /* Ignore errors.  This is just providing extra information, it
             * is useful but not necessary.
             */
            ret |= (ret2 & BDRV_BLOCK_ZERO);
        }
    }

3577
    return ret;
3578 3579
}

3580 3581
/* Coroutine wrapper for bdrv_get_block_status() */
static void coroutine_fn bdrv_get_block_status_co_entry(void *opaque)
3582
{
3583
    BdrvCoGetBlockStatusData *data = opaque;
3584 3585
    BlockDriverState *bs = data->bs;

3586 3587
    data->ret = bdrv_co_get_block_status(bs, data->sector_num, data->nb_sectors,
                                         data->pnum);
3588 3589 3590 3591
    data->done = true;
}

/*
3592
 * Synchronous wrapper around bdrv_co_get_block_status().
3593
 *
3594
 * See bdrv_co_get_block_status() for details.
3595
 */
3596 3597
int64_t bdrv_get_block_status(BlockDriverState *bs, int64_t sector_num,
                              int nb_sectors, int *pnum)
3598
{
3599
    Coroutine *co;
3600
    BdrvCoGetBlockStatusData data = {
3601 3602 3603 3604 3605 3606 3607
        .bs = bs,
        .sector_num = sector_num,
        .nb_sectors = nb_sectors,
        .pnum = pnum,
        .done = false,
    };

3608 3609
    if (qemu_in_coroutine()) {
        /* Fast-path if already in coroutine context */
3610
        bdrv_get_block_status_co_entry(&data);
3611
    } else {
3612
        co = qemu_coroutine_create(bdrv_get_block_status_co_entry);
3613 3614 3615 3616
        qemu_coroutine_enter(co, &data);
        while (!data.done) {
            qemu_aio_wait();
        }
3617 3618
    }
    return data.ret;
3619 3620
}

3621 3622 3623
int coroutine_fn bdrv_is_allocated(BlockDriverState *bs, int64_t sector_num,
                                   int nb_sectors, int *pnum)
{
3624 3625 3626 3627 3628 3629 3630
    int64_t ret = bdrv_get_block_status(bs, sector_num, nb_sectors, pnum);
    if (ret < 0) {
        return ret;
    }
    return
        (ret & BDRV_BLOCK_DATA) ||
        ((ret & BDRV_BLOCK_ZERO) && !bdrv_has_zero_init(bs));
3631 3632
}

3633 3634 3635 3636 3637 3638 3639 3640 3641 3642 3643 3644
/*
 * 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.
 *
 */
3645 3646 3647 3648
int bdrv_is_allocated_above(BlockDriverState *top,
                            BlockDriverState *base,
                            int64_t sector_num,
                            int nb_sectors, int *pnum)
3649 3650 3651 3652 3653 3654 3655
{
    BlockDriverState *intermediate;
    int ret, n = nb_sectors;

    intermediate = top;
    while (intermediate && intermediate != base) {
        int pnum_inter;
3656 3657
        ret = bdrv_is_allocated(intermediate, sector_num, nb_sectors,
                                &pnum_inter);
3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670
        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.
         */
3671 3672 3673
        if (n > pnum_inter &&
            (intermediate == top ||
             sector_num + pnum_inter < intermediate->total_sectors)) {
3674 3675 3676 3677 3678 3679 3680 3681 3682 3683
            n = pnum_inter;
        }

        intermediate = intermediate->backing_hd;
    }

    *pnum = n;
    return 0;
}

3684 3685 3686 3687 3688 3689 3690 3691 3692 3693
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;
}

3694
void bdrv_get_backing_filename(BlockDriverState *bs,
B
bellard 已提交
3695 3696
                               char *filename, int filename_size)
{
K
Kevin Wolf 已提交
3697
    pstrcpy(filename, filename_size, bs->backing_file);
B
bellard 已提交
3698 3699
}

3700
int bdrv_write_compressed(BlockDriverState *bs, int64_t sector_num,
B
bellard 已提交
3701 3702 3703 3704
                          const uint8_t *buf, int nb_sectors)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
3705
        return -ENOMEDIUM;
B
bellard 已提交
3706 3707
    if (!drv->bdrv_write_compressed)
        return -ENOTSUP;
K
Kevin Wolf 已提交
3708 3709
    if (bdrv_check_request(bs, sector_num, nb_sectors))
        return -EIO;
3710

F
Fam Zheng 已提交
3711
    assert(QLIST_EMPTY(&bs->dirty_bitmaps));
3712

B
bellard 已提交
3713 3714
    return drv->bdrv_write_compressed(bs, sector_num, buf, nb_sectors);
}
3715

B
bellard 已提交
3716 3717 3718 3719
int bdrv_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
3720
        return -ENOMEDIUM;
B
bellard 已提交
3721 3722 3723 3724 3725 3726
    if (!drv->bdrv_get_info)
        return -ENOTSUP;
    memset(bdi, 0, sizeof(*bdi));
    return drv->bdrv_get_info(bs, bdi);
}

M
Max Reitz 已提交
3727 3728 3729 3730 3731 3732 3733 3734 3735
ImageInfoSpecific *bdrv_get_specific_info(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;
    if (drv && drv->bdrv_get_specific_info) {
        return drv->bdrv_get_specific_info(bs);
    }
    return NULL;
}

3736 3737
int bdrv_save_vmstate(BlockDriverState *bs, const uint8_t *buf,
                      int64_t pos, int size)
3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749
{
    QEMUIOVector qiov;
    struct iovec iov = {
        .iov_base   = (void *) buf,
        .iov_len    = size,
    };

    qemu_iovec_init_external(&qiov, &iov, 1);
    return bdrv_writev_vmstate(bs, &qiov, pos);
}

int bdrv_writev_vmstate(BlockDriverState *bs, QEMUIOVector *qiov, int64_t pos)
3750 3751
{
    BlockDriver *drv = bs->drv;
3752 3753

    if (!drv) {
3754
        return -ENOMEDIUM;
3755 3756 3757 3758 3759 3760
    } else if (drv->bdrv_save_vmstate) {
        return drv->bdrv_save_vmstate(bs, qiov, pos);
    } else if (bs->file) {
        return bdrv_writev_vmstate(bs->file, qiov, pos);
    }

3761
    return -ENOTSUP;
3762 3763
}

3764 3765
int bdrv_load_vmstate(BlockDriverState *bs, uint8_t *buf,
                      int64_t pos, int size)
3766 3767 3768 3769
{
    BlockDriver *drv = bs->drv;
    if (!drv)
        return -ENOMEDIUM;
3770 3771 3772 3773 3774
    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;
3775 3776
}

K
Kevin Wolf 已提交
3777 3778
void bdrv_debug_event(BlockDriverState *bs, BlkDebugEvent event)
{
3779
    if (!bs || !bs->drv || !bs->drv->bdrv_debug_event) {
K
Kevin Wolf 已提交
3780 3781 3782
        return;
    }

3783
    bs->drv->bdrv_debug_event(bs, event);
K
Kevin Wolf 已提交
3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799
}

int bdrv_debug_breakpoint(BlockDriverState *bs, const char *event,
                          const char *tag)
{
    while (bs && bs->drv && !bs->drv->bdrv_debug_breakpoint) {
        bs = bs->file;
    }

    if (bs && bs->drv && bs->drv->bdrv_debug_breakpoint) {
        return bs->drv->bdrv_debug_breakpoint(bs, event, tag);
    }

    return -ENOTSUP;
}

F
Fam Zheng 已提交
3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812
int bdrv_debug_remove_breakpoint(BlockDriverState *bs, const char *tag)
{
    while (bs && bs->drv && !bs->drv->bdrv_debug_remove_breakpoint) {
        bs = bs->file;
    }

    if (bs && bs->drv && bs->drv->bdrv_debug_remove_breakpoint) {
        return bs->drv->bdrv_debug_remove_breakpoint(bs, tag);
    }

    return -ENOTSUP;
}

K
Kevin Wolf 已提交
3813 3814 3815 3816 3817
int bdrv_debug_resume(BlockDriverState *bs, const char *tag)
{
    while (bs && bs->drv && !bs->drv->bdrv_debug_resume) {
        bs = bs->file;
    }
K
Kevin Wolf 已提交
3818

K
Kevin Wolf 已提交
3819 3820 3821 3822 3823 3824 3825 3826 3827 3828 3829 3830 3831 3832 3833 3834 3835 3836
    if (bs && bs->drv && bs->drv->bdrv_debug_resume) {
        return bs->drv->bdrv_debug_resume(bs, tag);
    }

    return -ENOTSUP;
}

bool bdrv_debug_is_suspended(BlockDriverState *bs, const char *tag)
{
    while (bs && bs->drv && !bs->drv->bdrv_debug_is_suspended) {
        bs = bs->file;
    }

    if (bs && bs->drv && bs->drv->bdrv_debug_is_suspended) {
        return bs->drv->bdrv_debug_is_suspended(bs, tag);
    }

    return false;
K
Kevin Wolf 已提交
3837 3838
}

3839 3840 3841 3842 3843
int bdrv_is_snapshot(BlockDriverState *bs)
{
    return !!(bs->open_flags & BDRV_O_SNAPSHOT);
}

3844 3845 3846 3847
/* backing_file can either be relative, or absolute, or a protocol.  If it is
 * relative, it must be relative to the chain.  So, passing in bs->filename
 * from a BDS as backing_file should not be done, as that may be relative to
 * the CWD rather than the chain. */
3848 3849 3850
BlockDriverState *bdrv_find_backing_image(BlockDriverState *bs,
        const char *backing_file)
{
3851 3852 3853 3854 3855 3856 3857 3858
    char *filename_full = NULL;
    char *backing_file_full = NULL;
    char *filename_tmp = NULL;
    int is_protocol = 0;
    BlockDriverState *curr_bs = NULL;
    BlockDriverState *retval = NULL;

    if (!bs || !bs->drv || !backing_file) {
3859 3860 3861
        return NULL;
    }

3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876
    filename_full     = g_malloc(PATH_MAX);
    backing_file_full = g_malloc(PATH_MAX);
    filename_tmp      = g_malloc(PATH_MAX);

    is_protocol = path_has_protocol(backing_file);

    for (curr_bs = bs; curr_bs->backing_hd; curr_bs = curr_bs->backing_hd) {

        /* If either of the filename paths is actually a protocol, then
         * compare unmodified paths; otherwise make paths relative */
        if (is_protocol || path_has_protocol(curr_bs->backing_file)) {
            if (strcmp(backing_file, curr_bs->backing_file) == 0) {
                retval = curr_bs->backing_hd;
                break;
            }
3877
        } else {
3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892 3893 3894 3895 3896 3897 3898 3899 3900
            /* If not an absolute filename path, make it relative to the current
             * image's filename path */
            path_combine(filename_tmp, PATH_MAX, curr_bs->filename,
                         backing_file);

            /* We are going to compare absolute pathnames */
            if (!realpath(filename_tmp, filename_full)) {
                continue;
            }

            /* We need to make sure the backing filename we are comparing against
             * is relative to the current image filename (or absolute) */
            path_combine(filename_tmp, PATH_MAX, curr_bs->filename,
                         curr_bs->backing_file);

            if (!realpath(filename_tmp, backing_file_full)) {
                continue;
            }

            if (strcmp(backing_file_full, filename_full) == 0) {
                retval = curr_bs->backing_hd;
                break;
            }
3901 3902 3903
        }
    }

3904 3905 3906 3907
    g_free(filename_full);
    g_free(backing_file_full);
    g_free(filename_tmp);
    return retval;
3908 3909
}

3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922
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);
}

3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938
BlockDriverState *bdrv_find_base(BlockDriverState *bs)
{
    BlockDriverState *curr_bs = NULL;

    if (!bs) {
        return NULL;
    }

    curr_bs = bs;

    while (curr_bs->backing_hd) {
        curr_bs = curr_bs->backing_hd;
    }
    return curr_bs;
}

B
bellard 已提交
3939
/**************************************************************/
B
bellard 已提交
3940
/* async I/Os */
B
bellard 已提交
3941

3942
BlockDriverAIOCB *bdrv_aio_readv(BlockDriverState *bs, int64_t sector_num,
3943
                                 QEMUIOVector *qiov, int nb_sectors,
3944
                                 BlockDriverCompletionFunc *cb, void *opaque)
B
bellard 已提交
3945
{
3946 3947
    trace_bdrv_aio_readv(bs, sector_num, nb_sectors, opaque);

P
Paolo Bonzini 已提交
3948
    return bdrv_co_aio_rw_vector(bs, sector_num, qiov, nb_sectors, 0,
3949
                                 cb, opaque, false);
B
bellard 已提交
3950 3951
}

3952 3953 3954
BlockDriverAIOCB *bdrv_aio_writev(BlockDriverState *bs, int64_t sector_num,
                                  QEMUIOVector *qiov, int nb_sectors,
                                  BlockDriverCompletionFunc *cb, void *opaque)
B
bellard 已提交
3955
{
3956 3957
    trace_bdrv_aio_writev(bs, sector_num, nb_sectors, opaque);

P
Paolo Bonzini 已提交
3958
    return bdrv_co_aio_rw_vector(bs, sector_num, qiov, nb_sectors, 0,
3959
                                 cb, opaque, true);
B
bellard 已提交
3960 3961
}

P
Paolo Bonzini 已提交
3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972
BlockDriverAIOCB *bdrv_aio_write_zeroes(BlockDriverState *bs,
        int64_t sector_num, int nb_sectors, BdrvRequestFlags flags,
        BlockDriverCompletionFunc *cb, void *opaque)
{
    trace_bdrv_aio_write_zeroes(bs, sector_num, nb_sectors, flags, opaque);

    return bdrv_co_aio_rw_vector(bs, sector_num, NULL, nb_sectors,
                                 BDRV_REQ_ZERO_WRITE | flags,
                                 cb, opaque, true);
}

K
Kevin Wolf 已提交
3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990

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);
3991 3992 3993
        if (mcb->callbacks[i].free_qiov) {
            qemu_iovec_destroy(mcb->callbacks[i].free_qiov);
        }
3994
        g_free(mcb->callbacks[i].free_qiov);
K
Kevin Wolf 已提交
3995 3996 3997 3998 3999 4000 4001
    }
}

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

4002 4003
    trace_multiwrite_cb(mcb, ret);

4004
    if (ret < 0 && !mcb->error) {
K
Kevin Wolf 已提交
4005 4006 4007 4008 4009
        mcb->error = ret;
    }

    mcb->num_requests--;
    if (mcb->num_requests == 0) {
4010
        multiwrite_user_cb(mcb);
4011
        g_free(mcb);
K
Kevin Wolf 已提交
4012 4013 4014 4015 4016
    }
}

static int multiwrite_req_compare(const void *a, const void *b)
{
4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029
    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
4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050
}

/*
 * 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;

4051
        // Handle exactly sequential writes and overlapping writes.
K
Kevin Wolf 已提交
4052 4053 4054 4055
        if (reqs[i].sector <= oldreq_last) {
            merge = 1;
        }

4056 4057 4058 4059
        if (reqs[outidx].qiov->niov + reqs[i].qiov->niov + 1 > IOV_MAX) {
            merge = 0;
        }

K
Kevin Wolf 已提交
4060 4061
        if (merge) {
            size_t size;
4062
            QEMUIOVector *qiov = g_malloc0(sizeof(*qiov));
K
Kevin Wolf 已提交
4063 4064 4065 4066 4067 4068
            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;
4069
            qemu_iovec_concat(qiov, reqs[outidx].qiov, 0, size);
K
Kevin Wolf 已提交
4070

4071 4072
            // We should need to add any zeros between the two requests
            assert (reqs[i].sector <= oldreq_last);
K
Kevin Wolf 已提交
4073 4074

            // Add the second request
4075
            qemu_iovec_concat(qiov, reqs[i].qiov, 0, reqs[i].qiov->size);
K
Kevin Wolf 已提交
4076

4077
            reqs[outidx].nb_sectors = qiov->size >> 9;
K
Kevin Wolf 已提交
4078 4079 4080 4081 4082 4083 4084 4085 4086 4087 4088 4089 4090 4091 4092 4093 4094 4095 4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110
            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;

4111 4112 4113 4114 4115 4116 4117 4118
    /* 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
4119 4120 4121 4122 4123
    if (num_reqs == 0) {
        return 0;
    }

    // Create MultiwriteCB structure
4124
    mcb = g_malloc0(sizeof(*mcb) + num_reqs * sizeof(*mcb->callbacks));
K
Kevin Wolf 已提交
4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135
    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);

4136 4137
    trace_bdrv_aio_multiwrite(mcb, mcb->num_callbacks, num_reqs);

4138 4139
    /* Run the aio requests. */
    mcb->num_requests = num_reqs;
K
Kevin Wolf 已提交
4140
    for (i = 0; i < num_reqs; i++) {
P
Paolo Bonzini 已提交
4141 4142 4143 4144
        bdrv_co_aio_rw_vector(bs, reqs[i].sector, reqs[i].qiov,
                              reqs[i].nb_sectors, reqs[i].flags,
                              multiwrite_cb, mcb,
                              true);
K
Kevin Wolf 已提交
4145 4146 4147 4148 4149
    }

    return 0;
}

B
bellard 已提交
4150 4151
void bdrv_aio_cancel(BlockDriverAIOCB *acb)
{
S
Stefan Hajnoczi 已提交
4152
    acb->aiocb_info->cancel(acb);
B
bellard 已提交
4153 4154 4155 4156 4157
}

/**************************************************************/
/* async block device emulation */

4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169
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)
{
4170 4171
    BlockDriverAIOCBSync *acb =
        container_of(blockacb, BlockDriverAIOCBSync, common);
4172
    qemu_bh_delete(acb->bh);
A
Avi Kivity 已提交
4173
    acb->bh = NULL;
4174 4175 4176
    qemu_aio_release(acb);
}

S
Stefan Hajnoczi 已提交
4177
static const AIOCBInfo bdrv_em_aiocb_info = {
4178 4179 4180 4181
    .aiocb_size         = sizeof(BlockDriverAIOCBSync),
    .cancel             = bdrv_aio_cancel_em,
};

4182
static void bdrv_aio_bh_cb(void *opaque)
B
bellard 已提交
4183
{
4184
    BlockDriverAIOCBSync *acb = opaque;
4185 4186

    if (!acb->is_write)
4187
        qemu_iovec_from_buf(acb->qiov, 0, acb->bounce, acb->qiov->size);
4188
    qemu_vfree(acb->bounce);
4189
    acb->common.cb(acb->common.opaque, acb->ret);
4190
    qemu_bh_delete(acb->bh);
A
Avi Kivity 已提交
4191
    acb->bh = NULL;
4192
    qemu_aio_release(acb);
B
bellard 已提交
4193
}
4194

4195 4196 4197 4198 4199 4200 4201 4202
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
4203
{
4204 4205
    BlockDriverAIOCBSync *acb;

S
Stefan Hajnoczi 已提交
4206
    acb = qemu_aio_get(&bdrv_em_aiocb_info, bs, cb, opaque);
4207 4208
    acb->is_write = is_write;
    acb->qiov = qiov;
4209
    acb->bounce = qemu_blockalign(bs, qiov->size);
4210
    acb->bh = qemu_bh_new(bdrv_aio_bh_cb, acb);
4211 4212

    if (is_write) {
4213
        qemu_iovec_to_buf(acb->qiov, 0, acb->bounce, qiov->size);
4214
        acb->ret = bs->drv->bdrv_write(bs, sector_num, acb->bounce, nb_sectors);
4215
    } else {
4216
        acb->ret = bs->drv->bdrv_read(bs, sector_num, acb->bounce, nb_sectors);
4217 4218
    }

4219
    qemu_bh_schedule(acb->bh);
4220

4221
    return &acb->common;
4222 4223
}

4224 4225
static BlockDriverAIOCB *bdrv_aio_readv_em(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
4226
        BlockDriverCompletionFunc *cb, void *opaque)
4227
{
4228 4229
    return bdrv_aio_rw_vector(bs, sector_num, qiov, nb_sectors, cb, opaque, 0);
}
B
bellard 已提交
4230

4231 4232 4233 4234 4235
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);
4236 4237
}

4238 4239 4240 4241 4242

typedef struct BlockDriverAIOCBCoroutine {
    BlockDriverAIOCB common;
    BlockRequest req;
    bool is_write;
K
Kevin Wolf 已提交
4243
    bool *done;
4244 4245 4246 4247 4248
    QEMUBH* bh;
} BlockDriverAIOCBCoroutine;

static void bdrv_aio_co_cancel_em(BlockDriverAIOCB *blockacb)
{
K
Kevin Wolf 已提交
4249 4250 4251 4252 4253 4254 4255 4256
    BlockDriverAIOCBCoroutine *acb =
        container_of(blockacb, BlockDriverAIOCBCoroutine, common);
    bool done = false;

    acb->done = &done;
    while (!done) {
        qemu_aio_wait();
    }
4257 4258
}

S
Stefan Hajnoczi 已提交
4259
static const AIOCBInfo bdrv_em_co_aiocb_info = {
4260 4261 4262 4263
    .aiocb_size         = sizeof(BlockDriverAIOCBCoroutine),
    .cancel             = bdrv_aio_co_cancel_em,
};

P
Paolo Bonzini 已提交
4264
static void bdrv_co_em_bh(void *opaque)
4265 4266 4267 4268
{
    BlockDriverAIOCBCoroutine *acb = opaque;

    acb->common.cb(acb->common.opaque, acb->req.error);
K
Kevin Wolf 已提交
4269 4270 4271 4272 4273

    if (acb->done) {
        *acb->done = true;
    }

4274 4275 4276 4277
    qemu_bh_delete(acb->bh);
    qemu_aio_release(acb);
}

4278 4279 4280 4281 4282 4283 4284 4285
/* 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,
P
Paolo Bonzini 已提交
4286
            acb->req.nb_sectors, acb->req.qiov, acb->req.flags);
4287 4288
    } else {
        acb->req.error = bdrv_co_do_writev(bs, acb->req.sector,
P
Paolo Bonzini 已提交
4289
            acb->req.nb_sectors, acb->req.qiov, acb->req.flags);
4290 4291
    }

P
Paolo Bonzini 已提交
4292
    acb->bh = qemu_bh_new(bdrv_co_em_bh, acb);
4293 4294 4295
    qemu_bh_schedule(acb->bh);
}

4296 4297 4298 4299
static BlockDriverAIOCB *bdrv_co_aio_rw_vector(BlockDriverState *bs,
                                               int64_t sector_num,
                                               QEMUIOVector *qiov,
                                               int nb_sectors,
P
Paolo Bonzini 已提交
4300
                                               BdrvRequestFlags flags,
4301 4302
                                               BlockDriverCompletionFunc *cb,
                                               void *opaque,
4303
                                               bool is_write)
4304 4305 4306 4307
{
    Coroutine *co;
    BlockDriverAIOCBCoroutine *acb;

S
Stefan Hajnoczi 已提交
4308
    acb = qemu_aio_get(&bdrv_em_co_aiocb_info, bs, cb, opaque);
4309 4310 4311
    acb->req.sector = sector_num;
    acb->req.nb_sectors = nb_sectors;
    acb->req.qiov = qiov;
P
Paolo Bonzini 已提交
4312
    acb->req.flags = flags;
4313
    acb->is_write = is_write;
K
Kevin Wolf 已提交
4314
    acb->done = NULL;
4315

4316
    co = qemu_coroutine_create(bdrv_co_do_rw);
4317 4318 4319 4320 4321
    qemu_coroutine_enter(co, acb);

    return &acb->common;
}

P
Paolo Bonzini 已提交
4322
static void coroutine_fn bdrv_aio_flush_co_entry(void *opaque)
4323
{
P
Paolo Bonzini 已提交
4324 4325
    BlockDriverAIOCBCoroutine *acb = opaque;
    BlockDriverState *bs = acb->common.bs;
4326

P
Paolo Bonzini 已提交
4327 4328
    acb->req.error = bdrv_co_flush(bs);
    acb->bh = qemu_bh_new(bdrv_co_em_bh, acb);
4329 4330 4331
    qemu_bh_schedule(acb->bh);
}

P
Paolo Bonzini 已提交
4332
BlockDriverAIOCB *bdrv_aio_flush(BlockDriverState *bs,
4333 4334
        BlockDriverCompletionFunc *cb, void *opaque)
{
P
Paolo Bonzini 已提交
4335
    trace_bdrv_aio_flush(bs, opaque);
4336

P
Paolo Bonzini 已提交
4337 4338
    Coroutine *co;
    BlockDriverAIOCBCoroutine *acb;
4339

S
Stefan Hajnoczi 已提交
4340
    acb = qemu_aio_get(&bdrv_em_co_aiocb_info, bs, cb, opaque);
K
Kevin Wolf 已提交
4341 4342
    acb->done = NULL;

P
Paolo Bonzini 已提交
4343 4344
    co = qemu_coroutine_create(bdrv_aio_flush_co_entry);
    qemu_coroutine_enter(co, acb);
4345 4346 4347 4348

    return &acb->common;
}

4349 4350 4351 4352 4353 4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367
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);

S
Stefan Hajnoczi 已提交
4368
    acb = qemu_aio_get(&bdrv_em_co_aiocb_info, bs, cb, opaque);
4369 4370
    acb->req.sector = sector_num;
    acb->req.nb_sectors = nb_sectors;
K
Kevin Wolf 已提交
4371
    acb->done = NULL;
4372 4373 4374 4375 4376 4377
    co = qemu_coroutine_create(bdrv_aio_discard_co_entry);
    qemu_coroutine_enter(co, acb);

    return &acb->common;
}

B
bellard 已提交
4378 4379
void bdrv_init(void)
{
4380
    module_call_init(MODULE_INIT_BLOCK);
B
bellard 已提交
4381
}
4382

4383 4384 4385 4386 4387 4388
void bdrv_init_with_whitelist(void)
{
    use_bdrv_whitelist = 1;
    bdrv_init();
}

S
Stefan Hajnoczi 已提交
4389
void *qemu_aio_get(const AIOCBInfo *aiocb_info, BlockDriverState *bs,
4390
                   BlockDriverCompletionFunc *cb, void *opaque)
4391 4392 4393
{
    BlockDriverAIOCB *acb;

S
Stefan Hajnoczi 已提交
4394 4395
    acb = g_slice_alloc(aiocb_info->aiocb_size);
    acb->aiocb_info = aiocb_info;
4396 4397 4398 4399 4400 4401 4402 4403
    acb->bs = bs;
    acb->cb = cb;
    acb->opaque = opaque;
    return acb;
}

void qemu_aio_release(void *p)
{
4404
    BlockDriverAIOCB *acb = p;
S
Stefan Hajnoczi 已提交
4405
    g_slice_free1(acb->aiocb_info->aiocb_size, acb);
4406
}
B
bellard 已提交
4407

4408 4409 4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422 4423 4424 4425 4426 4427 4428 4429 4430 4431 4432 4433
/**************************************************************/
/* 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) {
4434 4435
        acb = bs->drv->bdrv_aio_writev(bs, sector_num, iov, nb_sectors,
                                       bdrv_co_io_em_complete, &co);
4436
    } else {
4437 4438
        acb = bs->drv->bdrv_aio_readv(bs, sector_num, iov, nb_sectors,
                                      bdrv_co_io_em_complete, &co);
4439 4440
    }

4441
    trace_bdrv_co_io_em(bs, sector_num, nb_sectors, is_write, acb);
4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463
    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
4464
static void coroutine_fn bdrv_flush_co_entry(void *opaque)
4465
{
P
Paolo Bonzini 已提交
4466 4467 4468 4469 4470 4471 4472
    RwCo *rwco = opaque;

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

int coroutine_fn bdrv_co_flush(BlockDriverState *bs)
{
K
Kevin Wolf 已提交
4473 4474
    int ret;

4475
    if (!bs || !bdrv_is_inserted(bs) || bdrv_is_read_only(bs)) {
P
Paolo Bonzini 已提交
4476
        return 0;
K
Kevin Wolf 已提交
4477 4478
    }

4479
    /* Write back cached data to the OS even with cache=unsafe */
4480
    BLKDBG_EVENT(bs->file, BLKDBG_FLUSH_TO_OS);
K
Kevin Wolf 已提交
4481 4482 4483 4484 4485 4486 4487
    if (bs->drv->bdrv_co_flush_to_os) {
        ret = bs->drv->bdrv_co_flush_to_os(bs);
        if (ret < 0) {
            return ret;
        }
    }

4488 4489
    /* But don't actually force it to the disk with cache=unsafe */
    if (bs->open_flags & BDRV_O_NO_FLUSH) {
4490
        goto flush_parent;
4491 4492
    }

4493
    BLKDBG_EVENT(bs->file, BLKDBG_FLUSH_TO_DISK);
K
Kevin Wolf 已提交
4494
    if (bs->drv->bdrv_co_flush_to_disk) {
4495
        ret = bs->drv->bdrv_co_flush_to_disk(bs);
P
Paolo Bonzini 已提交
4496 4497 4498 4499 4500 4501 4502 4503
    } 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) {
4504
            ret = -EIO;
P
Paolo Bonzini 已提交
4505 4506
        } else {
            qemu_coroutine_yield();
4507
            ret = co.ret;
P
Paolo Bonzini 已提交
4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520
        }
    } 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.
         */
4521
        ret = 0;
P
Paolo Bonzini 已提交
4522
    }
4523 4524 4525 4526 4527 4528 4529
    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.
     */
4530
flush_parent:
4531
    return bdrv_co_flush(bs->file);
P
Paolo Bonzini 已提交
4532 4533
}

4534 4535 4536 4537 4538 4539 4540 4541 4542 4543 4544
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;

4545
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
4546 4547 4548 4549
        bdrv_invalidate_cache(bs);
    }
}

4550 4551 4552 4553
void bdrv_clear_incoming_migration_all(void)
{
    BlockDriverState *bs;

4554
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
4555 4556 4557 4558
        bs->open_flags = bs->open_flags & ~(BDRV_O_INCOMING);
    }
}

P
Paolo Bonzini 已提交
4559 4560 4561 4562 4563 4564
int bdrv_flush(BlockDriverState *bs)
{
    Coroutine *co;
    RwCo rwco = {
        .bs = bs,
        .ret = NOT_DONE,
4565 4566
    };

P
Paolo Bonzini 已提交
4567 4568 4569 4570 4571 4572 4573 4574 4575
    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();
        }
4576
    }
P
Paolo Bonzini 已提交
4577 4578

    return rwco.ret;
4579 4580
}

4581 4582 4583 4584 4585 4586 4587
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);
}

4588 4589 4590 4591 4592
/* if no limit is specified in the BlockLimits use a default
 * of 32768 512-byte sectors (16 MiB) per request.
 */
#define MAX_DISCARD_DEFAULT 32768

4593 4594 4595
int coroutine_fn bdrv_co_discard(BlockDriverState *bs, int64_t sector_num,
                                 int nb_sectors)
{
4596 4597
    int max_discard;

4598 4599 4600 4601 4602 4603
    if (!bs->drv) {
        return -ENOMEDIUM;
    } else if (bdrv_check_request(bs, sector_num, nb_sectors)) {
        return -EIO;
    } else if (bs->read_only) {
        return -EROFS;
4604 4605
    }

F
Fam Zheng 已提交
4606
    bdrv_reset_dirty(bs, sector_num, nb_sectors);
4607

P
Paolo Bonzini 已提交
4608 4609 4610 4611 4612
    /* Do nothing if disabled.  */
    if (!(bs->open_flags & BDRV_O_UNMAP)) {
        return 0;
    }

4613 4614 4615
    if (!bs->drv->bdrv_co_discard && !bs->drv->bdrv_aio_discard) {
        return 0;
    }
4616

4617 4618 4619 4620
    max_discard = bs->bl.max_discard ?  bs->bl.max_discard : MAX_DISCARD_DEFAULT;
    while (nb_sectors > 0) {
        int ret;
        int num = nb_sectors;
4621

4622 4623 4624 4625 4626 4627
        /* align request */
        if (bs->bl.discard_alignment &&
            num >= bs->bl.discard_alignment &&
            sector_num % bs->bl.discard_alignment) {
            if (num > bs->bl.discard_alignment) {
                num = bs->bl.discard_alignment;
4628
            }
4629 4630
            num -= sector_num % bs->bl.discard_alignment;
        }
4631

4632 4633 4634 4635
        /* limit request size */
        if (num > max_discard) {
            num = max_discard;
        }
4636

4637
        if (bs->drv->bdrv_co_discard) {
4638
            ret = bs->drv->bdrv_co_discard(bs, sector_num, num);
4639 4640 4641 4642 4643 4644 4645 4646 4647 4648 4649 4650 4651
        } else {
            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();
                ret = co.ret;
4652 4653
            }
        }
4654
        if (ret && ret != -ENOTSUP) {
4655
            return ret;
4656
        }
4657 4658 4659

        sector_num += num;
        nb_sectors -= num;
4660
    }
4661
    return 0;
4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687
}

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
4688 4689 4690 4691 4692 4693 4694 4695 4696
/**************************************************************/
/* removable device support */

/**
 * Return TRUE if the media is present
 */
int bdrv_is_inserted(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;
4697

B
bellard 已提交
4698 4699 4700
    if (!drv)
        return 0;
    if (!drv->bdrv_is_inserted)
4701 4702
        return 1;
    return drv->bdrv_is_inserted(bs);
B
bellard 已提交
4703 4704 4705
}

/**
4706 4707
 * 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
4708 4709 4710 4711 4712
 */
int bdrv_media_changed(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;

4713 4714 4715 4716
    if (drv && drv->bdrv_media_changed) {
        return drv->bdrv_media_changed(bs);
    }
    return -ENOTSUP;
B
bellard 已提交
4717 4718 4719 4720 4721
}

/**
 * If eject_flag is TRUE, eject the media. Otherwise, close the tray
 */
4722
void bdrv_eject(BlockDriverState *bs, bool eject_flag)
B
bellard 已提交
4723 4724 4725
{
    BlockDriver *drv = bs->drv;

4726 4727
    if (drv && drv->bdrv_eject) {
        drv->bdrv_eject(bs, eject_flag);
B
bellard 已提交
4728
    }
4729 4730 4731 4732

    if (bs->device_name[0] != '\0') {
        bdrv_emit_qmp_eject_event(bs, eject_flag);
    }
B
bellard 已提交
4733 4734 4735 4736 4737 4738
}

/**
 * Lock or unlock the media (if it is locked, the user won't be able
 * to eject it manually).
 */
4739
void bdrv_lock_medium(BlockDriverState *bs, bool locked)
B
bellard 已提交
4740 4741 4742
{
    BlockDriver *drv = bs->drv;

4743
    trace_bdrv_lock_medium(bs, locked);
S
Stefan Hajnoczi 已提交
4744

4745 4746
    if (drv && drv->bdrv_lock_medium) {
        drv->bdrv_lock_medium(bs, locked);
B
bellard 已提交
4747 4748
    }
}
4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759

/* 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;
}
4760

4761 4762 4763
BlockDriverAIOCB *bdrv_aio_ioctl(BlockDriverState *bs,
        unsigned long int req, void *buf,
        BlockDriverCompletionFunc *cb, void *opaque)
4764
{
4765
    BlockDriver *drv = bs->drv;
4766

4767 4768 4769
    if (drv && drv->bdrv_aio_ioctl)
        return drv->bdrv_aio_ioctl(bs, req, buf, cb, opaque);
    return NULL;
4770
}
4771

4772 4773 4774 4775
void bdrv_set_buffer_alignment(BlockDriverState *bs, int align)
{
    bs->buffer_alignment = align;
}
4776

4777 4778 4779 4780
void *qemu_blockalign(BlockDriverState *bs, size_t size)
{
    return qemu_memalign((bs && bs->buffer_alignment) ? bs->buffer_alignment : 512, size);
}
4781

4782 4783 4784 4785 4786 4787 4788 4789 4790 4791 4792 4793 4794 4795 4796 4797
/*
 * Check if all memory in this vector is sector aligned.
 */
bool bdrv_qiov_is_aligned(BlockDriverState *bs, QEMUIOVector *qiov)
{
    int i;

    for (i = 0; i < qiov->niov; i++) {
        if ((uintptr_t) qiov->iov[i].iov_base % bs->buffer_alignment) {
            return false;
        }
    }

    return true;
}

F
Fam Zheng 已提交
4798
BdrvDirtyBitmap *bdrv_create_dirty_bitmap(BlockDriverState *bs, int granularity)
4799 4800
{
    int64_t bitmap_size;
F
Fam Zheng 已提交
4801
    BdrvDirtyBitmap *bitmap;
4802

4803 4804
    assert((granularity & (granularity - 1)) == 0);

F
Fam Zheng 已提交
4805 4806 4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822
    granularity >>= BDRV_SECTOR_BITS;
    assert(granularity);
    bitmap_size = (bdrv_getlength(bs) >> BDRV_SECTOR_BITS);
    bitmap = g_malloc0(sizeof(BdrvDirtyBitmap));
    bitmap->bitmap = hbitmap_alloc(bitmap_size, ffs(granularity) - 1);
    QLIST_INSERT_HEAD(&bs->dirty_bitmaps, bitmap, list);
    return bitmap;
}

void bdrv_release_dirty_bitmap(BlockDriverState *bs, BdrvDirtyBitmap *bitmap)
{
    BdrvDirtyBitmap *bm, *next;
    QLIST_FOREACH_SAFE(bm, &bs->dirty_bitmaps, list, next) {
        if (bm == bitmap) {
            QLIST_REMOVE(bitmap, list);
            hbitmap_free(bitmap->bitmap);
            g_free(bitmap);
            return;
4823
        }
4824 4825 4826
    }
}

F
Fam Zheng 已提交
4827 4828 4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841 4842 4843 4844 4845 4846
BlockDirtyInfoList *bdrv_query_dirty_bitmaps(BlockDriverState *bs)
{
    BdrvDirtyBitmap *bm;
    BlockDirtyInfoList *list = NULL;
    BlockDirtyInfoList **plist = &list;

    QLIST_FOREACH(bm, &bs->dirty_bitmaps, list) {
        BlockDirtyInfo *info = g_malloc0(sizeof(BlockDirtyInfo));
        BlockDirtyInfoList *entry = g_malloc0(sizeof(BlockDirtyInfoList));
        info->count = bdrv_get_dirty_count(bs, bm);
        info->granularity =
            ((int64_t) BDRV_SECTOR_SIZE << hbitmap_granularity(bm->bitmap));
        entry->value = info;
        *plist = entry;
        plist = &entry->next;
    }

    return list;
}

F
Fam Zheng 已提交
4847
int bdrv_get_dirty(BlockDriverState *bs, BdrvDirtyBitmap *bitmap, int64_t sector)
4848
{
F
Fam Zheng 已提交
4849 4850
    if (bitmap) {
        return hbitmap_get(bitmap->bitmap, sector);
4851 4852 4853 4854 4855
    } else {
        return 0;
    }
}

F
Fam Zheng 已提交
4856 4857
void bdrv_dirty_iter_init(BlockDriverState *bs,
                          BdrvDirtyBitmap *bitmap, HBitmapIter *hbi)
4858
{
F
Fam Zheng 已提交
4859
    hbitmap_iter_init(hbi, bitmap->bitmap, 0);
4860 4861 4862 4863 4864
}

void bdrv_set_dirty(BlockDriverState *bs, int64_t cur_sector,
                    int nr_sectors)
{
F
Fam Zheng 已提交
4865 4866 4867 4868
    BdrvDirtyBitmap *bitmap;
    QLIST_FOREACH(bitmap, &bs->dirty_bitmaps, list) {
        hbitmap_set(bitmap->bitmap, cur_sector, nr_sectors);
    }
4869 4870
}

F
Fam Zheng 已提交
4871
void bdrv_reset_dirty(BlockDriverState *bs, int64_t cur_sector, int nr_sectors)
4872
{
F
Fam Zheng 已提交
4873 4874 4875 4876
    BdrvDirtyBitmap *bitmap;
    QLIST_FOREACH(bitmap, &bs->dirty_bitmaps, list) {
        hbitmap_reset(bitmap->bitmap, cur_sector, nr_sectors);
    }
4877
}
4878

F
Fam Zheng 已提交
4879
int64_t bdrv_get_dirty_count(BlockDriverState *bs, BdrvDirtyBitmap *bitmap)
4880
{
F
Fam Zheng 已提交
4881
    return hbitmap_count(bitmap->bitmap);
4882
}
J
Jes Sorensen 已提交
4883

4884 4885 4886 4887 4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900
/* Get a reference to bs */
void bdrv_ref(BlockDriverState *bs)
{
    bs->refcnt++;
}

/* Release a previously grabbed reference to bs.
 * If after releasing, reference count is zero, the BlockDriverState is
 * deleted. */
void bdrv_unref(BlockDriverState *bs)
{
    assert(bs->refcnt > 0);
    if (--bs->refcnt == 0) {
        bdrv_delete(bs);
    }
}

4901 4902 4903 4904 4905 4906 4907 4908 4909 4910 4911
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;
}

4912 4913
void bdrv_iostatus_enable(BlockDriverState *bs)
{
4914
    bs->iostatus_enabled = true;
4915
    bs->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
4916 4917 4918 4919 4920 4921
}

/* 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)
{
4922
    return (bs->iostatus_enabled &&
4923 4924 4925
           (bs->on_write_error == BLOCKDEV_ON_ERROR_ENOSPC ||
            bs->on_write_error == BLOCKDEV_ON_ERROR_STOP   ||
            bs->on_read_error == BLOCKDEV_ON_ERROR_STOP));
4926 4927 4928 4929
}

void bdrv_iostatus_disable(BlockDriverState *bs)
{
4930
    bs->iostatus_enabled = false;
4931 4932 4933 4934 4935
}

void bdrv_iostatus_reset(BlockDriverState *bs)
{
    if (bdrv_iostatus_is_enabled(bs)) {
4936
        bs->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
4937 4938 4939
        if (bs->job) {
            block_job_iostatus_reset(bs->job);
        }
4940 4941 4942 4943 4944
    }
}

void bdrv_iostatus_set_err(BlockDriverState *bs, int error)
{
4945 4946
    assert(bdrv_iostatus_is_enabled(bs));
    if (bs->iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
4947 4948
        bs->iostatus = error == ENOSPC ? BLOCK_DEVICE_IO_STATUS_NOSPACE :
                                         BLOCK_DEVICE_IO_STATUS_FAILED;
4949 4950 4951
    }
}

4952 4953 4954 4955 4956 4957 4958
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
4959
    cookie->start_time_ns = get_clock();
4960 4961 4962 4963 4964 4965 4966 4967 4968 4969
    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
4970
    bs->total_time_ns[cookie->type] += get_clock() - cookie->start_time_ns;
4971 4972
}

4973 4974
void bdrv_img_create(const char *filename, const char *fmt,
                     const char *base_filename, const char *base_fmt,
4975 4976
                     char *options, uint64_t img_size, int flags,
                     Error **errp, bool quiet)
J
Jes Sorensen 已提交
4977 4978
{
    QEMUOptionParameter *param = NULL, *create_options = NULL;
K
Kevin Wolf 已提交
4979
    QEMUOptionParameter *backing_fmt, *backing_file, *size;
J
Jes Sorensen 已提交
4980
    BlockDriver *drv, *proto_drv;
4981
    BlockDriver *backing_drv = NULL;
4982
    Error *local_err = NULL;
J
Jes Sorensen 已提交
4983 4984 4985 4986 4987
    int ret = 0;

    /* Find driver and parse its options */
    drv = bdrv_find_format(fmt);
    if (!drv) {
4988
        error_setg(errp, "Unknown file format '%s'", fmt);
4989
        return;
J
Jes Sorensen 已提交
4990 4991
    }

4992
    proto_drv = bdrv_find_protocol(filename, true);
J
Jes Sorensen 已提交
4993
    if (!proto_drv) {
4994
        error_setg(errp, "Unknown protocol '%s'", filename);
4995
        return;
J
Jes Sorensen 已提交
4996 4997 4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011
    }

    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) {
5012
            error_setg(errp, "Invalid options for file format '%s'.", fmt);
J
Jes Sorensen 已提交
5013 5014 5015 5016 5017 5018 5019
            goto out;
        }
    }

    if (base_filename) {
        if (set_option_parameter(param, BLOCK_OPT_BACKING_FILE,
                                 base_filename)) {
5020 5021
            error_setg(errp, "Backing file not supported for file format '%s'",
                       fmt);
J
Jes Sorensen 已提交
5022 5023 5024 5025 5026 5027
            goto out;
        }
    }

    if (base_fmt) {
        if (set_option_parameter(param, BLOCK_OPT_BACKING_FMT, base_fmt)) {
5028 5029
            error_setg(errp, "Backing file format not supported for file "
                             "format '%s'", fmt);
J
Jes Sorensen 已提交
5030 5031 5032 5033
            goto out;
        }
    }

5034 5035 5036
    backing_file = get_option_parameter(param, BLOCK_OPT_BACKING_FILE);
    if (backing_file && backing_file->value.s) {
        if (!strcmp(filename, backing_file->value.s)) {
5037 5038
            error_setg(errp, "Error: Trying to create an image with the "
                             "same filename as the backing file");
5039 5040 5041 5042
            goto out;
        }
    }

J
Jes Sorensen 已提交
5043 5044
    backing_fmt = get_option_parameter(param, BLOCK_OPT_BACKING_FMT);
    if (backing_fmt && backing_fmt->value.s) {
5045 5046
        backing_drv = bdrv_find_format(backing_fmt->value.s);
        if (!backing_drv) {
5047 5048
            error_setg(errp, "Unknown backing file format '%s'",
                       backing_fmt->value.s);
J
Jes Sorensen 已提交
5049 5050 5051 5052 5053 5054
            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
5055 5056
    size = get_option_parameter(param, BLOCK_OPT_SIZE);
    if (size && size->value.n == -1) {
J
Jes Sorensen 已提交
5057
        if (backing_file && backing_file->value.s) {
5058
            BlockDriverState *bs;
J
Jes Sorensen 已提交
5059 5060
            uint64_t size;
            char buf[32];
5061 5062 5063 5064 5065
            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
5066 5067 5068

            bs = bdrv_new("");

5069
            ret = bdrv_open(bs, backing_file->value.s, NULL, back_flags,
5070
                            backing_drv, &local_err);
J
Jes Sorensen 已提交
5071
            if (ret < 0) {
5072 5073 5074 5075 5076
                error_setg_errno(errp, -ret, "Could not open '%s': %s",
                                 backing_file->value.s,
                                 error_get_pretty(local_err));
                error_free(local_err);
                local_err = NULL;
5077
                bdrv_unref(bs);
J
Jes Sorensen 已提交
5078 5079 5080 5081 5082 5083 5084
                goto out;
            }
            bdrv_get_geometry(bs, &size);
            size *= 512;

            snprintf(buf, sizeof(buf), "%" PRId64, size);
            set_option_parameter(param, BLOCK_OPT_SIZE, buf);
5085 5086

            bdrv_unref(bs);
J
Jes Sorensen 已提交
5087
        } else {
5088
            error_setg(errp, "Image creation needs a size parameter");
J
Jes Sorensen 已提交
5089 5090 5091 5092
            goto out;
        }
    }

5093 5094 5095 5096 5097
    if (!quiet) {
        printf("Formatting '%s', fmt=%s ", filename, fmt);
        print_option_parameters(param);
        puts("");
    }
5098 5099 5100 5101 5102 5103 5104 5105
    ret = bdrv_create(drv, filename, param, &local_err);
    if (ret == -EFBIG) {
        /* This is generally a better message than whatever the driver would
         * deliver (especially because of the cluster_size_hint), since that
         * is most probably not much different from "image too large". */
        const char *cluster_size_hint = "";
        if (get_option_parameter(create_options, BLOCK_OPT_CLUSTER_SIZE)) {
            cluster_size_hint = " (try using a larger cluster size)";
J
Jes Sorensen 已提交
5106
        }
5107 5108 5109 5110
        error_setg(errp, "The image size is too large for file format '%s'"
                   "%s", fmt, cluster_size_hint);
        error_free(local_err);
        local_err = NULL;
J
Jes Sorensen 已提交
5111 5112 5113 5114 5115 5116
    }

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

5117 5118 5119
    if (error_is_set(&local_err)) {
        error_propagate(errp, local_err);
    }
J
Jes Sorensen 已提交
5120
}
5121 5122 5123 5124 5125 5126

AioContext *bdrv_get_aio_context(BlockDriverState *bs)
{
    /* Currently BlockDriverState always uses the main loop AioContext */
    return qemu_get_aio_context();
}
5127 5128 5129 5130 5131 5132

void bdrv_add_before_write_notifier(BlockDriverState *bs,
                                    NotifierWithReturn *notifier)
{
    notifier_with_return_list_add(&bs->before_write_notifiers, notifier);
}
M
Max Reitz 已提交
5133 5134 5135 5136 5137 5138 5139 5140

int bdrv_amend_options(BlockDriverState *bs, QEMUOptionParameter *options)
{
    if (bs->drv->bdrv_amend_options == NULL) {
        return -ENOTSUP;
    }
    return bs->drv->bdrv_amend_options(bs, options);
}
5141

5142 5143 5144 5145 5146
/* Used to recurse on single child block filters.
 * Single child block filter will store their child in bs->file.
 */
bool bdrv_generic_is_first_non_filter(BlockDriverState *bs,
                                      BlockDriverState *candidate)
5147
{
5148 5149 5150 5151 5152 5153 5154 5155 5156 5157 5158 5159 5160 5161
    if (!bs->drv) {
        return false;
    }

    if (!bs->drv->authorizations[BS_IS_A_FILTER]) {
        if (bs == candidate) {
            return true;
        } else {
            return false;
        }
    }

    if (!bs->drv->authorizations[BS_FILTER_PASS_DOWN]) {
        return false;
5162 5163
    }

5164 5165 5166 5167 5168 5169 5170 5171 5172 5173 5174 5175
    if (!bs->file) {
        return false;
    }

    return bdrv_recurse_is_first_non_filter(bs->file, candidate);
}

bool bdrv_recurse_is_first_non_filter(BlockDriverState *bs,
                                      BlockDriverState *candidate)
{
    if (bs->drv && bs->drv->bdrv_recurse_is_first_non_filter) {
        return bs->drv->bdrv_recurse_is_first_non_filter(bs, candidate);
5176 5177
    }

5178
    return bdrv_generic_is_first_non_filter(bs, candidate);
5179 5180
}

5181 5182 5183 5184 5185
/* This function checks if the candidate is the first non filter bs down it's
 * bs chain. Since we don't have pointers to parents it explore all bs chains
 * from the top. Some filters can choose not to pass down the recursion.
 */
bool bdrv_is_first_non_filter(BlockDriverState *candidate)
5186
{
5187 5188 5189 5190 5191 5192 5193 5194 5195 5196 5197 5198 5199 5200 5201 5202 5203 5204 5205
    BlockDriverState *bs;

    /* walk down the bs forest recursively */
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
        bool perm;

        if (!bs->file) {
            continue;
        }

        perm = bdrv_recurse_is_first_non_filter(bs->file, candidate);

        /* candidate is the first non filter */
        if (perm) {
            return true;
        }
    }

    return false;
5206
}