block.c 125.1 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"
L
Luiz Capitulino 已提交
35
#include "qmp-commands.h"
36
#include "qemu/timer.h"
B
bellard 已提交
37

J
Juan Quintela 已提交
38
#ifdef CONFIG_BSD
B
bellard 已提交
39 40 41
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/ioctl.h>
B
Blue Swirl 已提交
42
#include <sys/queue.h>
43
#ifndef __DragonFly__
B
bellard 已提交
44 45
#include <sys/disk.h>
#endif
46
#endif
B
bellard 已提交
47

48 49 50 51
#ifdef _WIN32
#include <windows.h>
#endif

52 53
#define NOT_DONE 0x7fffffff /* used while emulated sync operation in progress */

54 55
typedef enum {
    BDRV_REQ_COPY_ON_READ = 0x1,
56
    BDRV_REQ_ZERO_WRITE   = 0x2,
57 58
} BdrvRequestFlags;

59
static void bdrv_dev_change_media_cb(BlockDriverState *bs, bool load);
60 61
static BlockDriverAIOCB *bdrv_aio_readv_em(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
62
        BlockDriverCompletionFunc *cb, void *opaque);
63 64
static BlockDriverAIOCB *bdrv_aio_writev_em(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
65
        BlockDriverCompletionFunc *cb, void *opaque);
66 67 68 69 70 71
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);
72
static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,
73 74
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags);
75
static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs,
76 77
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags);
78 79 80 81 82 83
static BlockDriverAIOCB *bdrv_co_aio_rw_vector(BlockDriverState *bs,
                                               int64_t sector_num,
                                               QEMUIOVector *qiov,
                                               int nb_sectors,
                                               BlockDriverCompletionFunc *cb,
                                               void *opaque,
84
                                               bool is_write);
85
static void coroutine_fn bdrv_co_do_rw(void *opaque);
K
Kevin Wolf 已提交
86 87
static int coroutine_fn bdrv_co_do_write_zeroes(BlockDriverState *bs,
    int64_t sector_num, int nb_sectors);
B
bellard 已提交
88

89 90
static QTAILQ_HEAD(, BlockDriverState) bdrv_states =
    QTAILQ_HEAD_INITIALIZER(bdrv_states);
91

92 93
static QLIST_HEAD(, BlockDriver) bdrv_drivers =
    QLIST_HEAD_INITIALIZER(bdrv_drivers);
B
bellard 已提交
94

95 96 97
/* If non-zero, use only whitelisted block drivers */
static int use_bdrv_whitelist;

98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117
#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

118
/* throttling disk I/O limits */
119 120
void bdrv_set_io_limits(BlockDriverState *bs,
                        ThrottleConfig *cfg)
Z
Zhi Yong Wu 已提交
121
{
122
    int i;
Z
Zhi Yong Wu 已提交
123

124
    throttle_config(&bs->throttle_state, cfg);
Z
Zhi Yong Wu 已提交
125

126 127
    for (i = 0; i < 2; i++) {
        qemu_co_enter_next(&bs->throttled_reqs[i]);
Z
Zhi Yong Wu 已提交
128
    }
129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146
}

/* 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
147

148
    return drained;
Z
Zhi Yong Wu 已提交
149 150
}

151
void bdrv_io_limits_disable(BlockDriverState *bs)
152
{
153
    bs->io_limits_enabled = false;
154

155 156 157
    bdrv_start_throttled_reqs(bs);

    throttle_destroy(&bs->throttle_state);
158 159
}

160
static void bdrv_throttle_read_timer_cb(void *opaque)
161
{
162 163
    BlockDriverState *bs = opaque;
    qemu_co_enter_next(&bs->throttled_reqs[0]);
164 165
}

166
static void bdrv_throttle_write_timer_cb(void *opaque)
167
{
168 169
    BlockDriverState *bs = opaque;
    qemu_co_enter_next(&bs->throttled_reqs[1]);
170 171
}

172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188
/* 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
189
static void bdrv_io_limits_intercept(BlockDriverState *bs,
190 191
                                     int nb_sectors,
                                     bool is_write)
Z
Zhi Yong Wu 已提交
192
{
193 194
    /* does this io must wait */
    bool must_wait = throttle_schedule_timer(&bs->throttle_state, is_write);
Z
Zhi Yong Wu 已提交
195

196 197 198 199
    /* 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
200 201
    }

202 203 204 205
    /* 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
206

207 208 209
    /* 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
210 211
    }

212 213
    /* else queue next request for execution */
    qemu_co_queue_next(&bs->throttled_reqs[is_write]);
Z
Zhi Yong Wu 已提交
214 215
}

216 217 218
/* check if the path starts with "<protocol>:" */
static int path_has_protocol(const char *path)
{
219 220
    const char *p;

221 222 223 224 225
#ifdef _WIN32
    if (is_windows_drive(path) ||
        is_windows_drive_prefix(path)) {
        return 0;
    }
226 227 228
    p = path + strcspn(path, ":/\\");
#else
    p = path + strcspn(path, ":/");
229 230
#endif

231
    return *p == ':';
232 233
}

B
bellard 已提交
234
int path_is_absolute(const char *path)
235
{
B
bellard 已提交
236 237
#ifdef _WIN32
    /* specific case for names like: "\\.\d:" */
P
Paolo Bonzini 已提交
238
    if (is_windows_drive(path) || is_windows_drive_prefix(path)) {
B
bellard 已提交
239
        return 1;
P
Paolo Bonzini 已提交
240 241
    }
    return (*path == '/' || *path == '\\');
242
#else
P
Paolo Bonzini 已提交
243
    return (*path == '/');
244
#endif
245 246
}

B
bellard 已提交
247 248 249 250 251 252
/* 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)
253
{
B
bellard 已提交
254 255 256 257 258 259 260 261 262 263 264 265 266
    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;
267 268 269 270 271 272 273 274 275
        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
276 277 278 279 280 281 282 283 284 285 286 287
        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);
288 289 290
    }
}

291 292 293 294 295 296 297 298 299
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);
    }
}

300
void bdrv_register(BlockDriver *bdrv)
B
bellard 已提交
301
{
302 303
    /* Block drivers without coroutine functions need emulation */
    if (!bdrv->bdrv_co_readv) {
304 305 306
        bdrv->bdrv_co_readv = bdrv_co_readv_em;
        bdrv->bdrv_co_writev = bdrv_co_writev_em;

307 308 309
        /* 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.
         */
310 311 312 313 314
        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
315
    }
316

317
    QLIST_INSERT_HEAD(&bdrv_drivers, bdrv, list);
B
bellard 已提交
318
}
B
bellard 已提交
319 320 321 322

/* create a new block device (by default it is empty) */
BlockDriverState *bdrv_new(const char *device_name)
{
323
    BlockDriverState *bs;
B
bellard 已提交
324

325
    bs = g_malloc0(sizeof(BlockDriverState));
B
bellard 已提交
326
    pstrcpy(bs->device_name, sizeof(bs->device_name), device_name);
B
bellard 已提交
327
    if (device_name[0] != '\0') {
328
        QTAILQ_INSERT_TAIL(&bdrv_states, bs, list);
B
bellard 已提交
329
    }
330
    bdrv_iostatus_disable(bs);
P
Paolo Bonzini 已提交
331
    notifier_list_init(&bs->close_notifiers);
332
    notifier_with_return_list_init(&bs->before_write_notifiers);
333 334
    qemu_co_queue_init(&bs->throttled_reqs[0]);
    qemu_co_queue_init(&bs->throttled_reqs[1]);
335
    bs->refcnt = 1;
P
Paolo Bonzini 已提交
336

B
bellard 已提交
337 338 339
    return bs;
}

P
Paolo Bonzini 已提交
340 341 342 343 344
void bdrv_add_close_notifier(BlockDriverState *bs, Notifier *notify)
{
    notifier_list_add(&bs->close_notifiers, notify);
}

B
bellard 已提交
345 346 347
BlockDriver *bdrv_find_format(const char *format_name)
{
    BlockDriver *drv1;
348 349
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
        if (!strcmp(drv1->format_name, format_name)) {
B
bellard 已提交
350
            return drv1;
351
        }
B
bellard 已提交
352 353 354 355
    }
    return NULL;
}

356
static int bdrv_is_whitelisted(BlockDriver *drv, bool read_only)
357
{
358 359 360 361 362
    static const char *whitelist_rw[] = {
        CONFIG_BDRV_RW_WHITELIST
    };
    static const char *whitelist_ro[] = {
        CONFIG_BDRV_RO_WHITELIST
363 364 365
    };
    const char **p;

366
    if (!whitelist_rw[0] && !whitelist_ro[0]) {
367
        return 1;               /* no whitelist, anything goes */
368
    }
369

370
    for (p = whitelist_rw; *p; p++) {
371 372 373 374
        if (!strcmp(drv->format_name, *p)) {
            return 1;
        }
    }
375 376 377 378 379 380 381
    if (read_only) {
        for (p = whitelist_ro; *p; p++) {
            if (!strcmp(drv->format_name, *p)) {
                return 1;
            }
        }
    }
382 383 384
    return 0;
}

385 386
BlockDriver *bdrv_find_whitelisted_format(const char *format_name,
                                          bool read_only)
387 388
{
    BlockDriver *drv = bdrv_find_format(format_name);
389
    return drv && bdrv_is_whitelisted(drv, read_only) ? drv : NULL;
390 391
}

392 393 394 395 396 397 398 399 400 401 402 403 404 405 406
typedef struct CreateCo {
    BlockDriver *drv;
    char *filename;
    QEMUOptionParameter *options;
    int ret;
} CreateCo;

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

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

407 408
int bdrv_create(BlockDriver *drv, const char* filename,
    QEMUOptionParameter *options)
B
bellard 已提交
409
{
410 411 412 413 414 415 416 417 418 419 420
    int ret;

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

    if (!drv->bdrv_create) {
421 422
        ret = -ENOTSUP;
        goto out;
423 424 425 426 427 428 429 430 431 432 433 434 435 436
    }

    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;
437

438 439
out:
    g_free(cco.filename);
440
    return ret;
B
bellard 已提交
441 442
}

443 444 445 446
int bdrv_create_file(const char* filename, QEMUOptionParameter *options)
{
    BlockDriver *drv;

447
    drv = bdrv_find_protocol(filename, true);
448
    if (drv == NULL) {
449
        return -ENOENT;
450 451 452 453 454
    }

    return bdrv_create(drv, filename, options);
}

455 456 457 458 459
/*
 * 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
460
{
461
#ifdef _WIN32
462
    char temp_dir[MAX_PATH];
463 464 465 466 467 468
    /* 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
469
#else
B
bellard 已提交
470
    int fd;
471
    const char *tmpdir;
A
aurel32 已提交
472 473 474
    tmpdir = getenv("TMPDIR");
    if (!tmpdir)
        tmpdir = "/tmp";
475 476 477
    if (snprintf(filename, size, "%s/vl.XXXXXX", tmpdir) >= size) {
        return -EOVERFLOW;
    }
B
bellard 已提交
478
    fd = mkstemp(filename);
479 480 481 482 483
    if (fd < 0) {
        return -errno;
    }
    if (close(fd) != 0) {
        unlink(filename);
484 485 486
        return -errno;
    }
    return 0;
B
bellard 已提交
487
#endif
488
}
B
bellard 已提交
489

490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511
/*
 * 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;
}

512 513
BlockDriver *bdrv_find_protocol(const char *filename,
                                bool allow_protocol_prefix)
B
bellard 已提交
514 515 516
{
    BlockDriver *drv1;
    char protocol[128];
517
    int len;
B
bellard 已提交
518
    const char *p;
B
bellard 已提交
519

520 521
    /* TODO Drivers without bdrv_file_open must be specified explicitly */

522 523 524 525 526 527 528 529 530 531 532 533
    /*
     * 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;
    }

534
    if (!path_has_protocol(filename) || !allow_protocol_prefix) {
535
        return bdrv_find_format("file");
536
    }
537

538 539
    p = strchr(filename, ':');
    assert(p != NULL);
540 541 542 543 544
    len = p - filename;
    if (len > sizeof(protocol) - 1)
        len = sizeof(protocol) - 1;
    memcpy(protocol, filename, len);
    protocol[len] = '\0';
545
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
546
        if (drv1->protocol_name &&
547
            !strcmp(drv1->protocol_name, protocol)) {
B
bellard 已提交
548
            return drv1;
549
        }
B
bellard 已提交
550 551 552 553
    }
    return NULL;
}

554 555
static int find_image_format(BlockDriverState *bs, const char *filename,
                             BlockDriver **pdrv)
556
{
557
    int score, score_max;
558 559
    BlockDriver *drv1, *drv;
    uint8_t buf[2048];
560
    int ret = 0;
561

562
    /* Return the raw BlockDriver * to scsi-generic devices or empty drives */
563
    if (bs->sg || !bdrv_is_inserted(bs) || bdrv_getlength(bs) == 0) {
564 565 566 567 568 569
        drv = bdrv_find_format("raw");
        if (!drv) {
            ret = -ENOENT;
        }
        *pdrv = drv;
        return ret;
570
    }
571

B
bellard 已提交
572 573
    ret = bdrv_pread(bs, 0, buf, sizeof(buf));
    if (ret < 0) {
574 575
        *pdrv = NULL;
        return ret;
B
bellard 已提交
576 577
    }

B
bellard 已提交
578
    score_max = 0;
579
    drv = NULL;
580
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
B
bellard 已提交
581 582 583 584 585 586
        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
587
        }
B
bellard 已提交
588
    }
589 590 591 592 593
    if (!drv) {
        ret = -ENOENT;
    }
    *pdrv = drv;
    return ret;
B
bellard 已提交
594 595
}

596 597 598 599 600 601 602
/**
 * Set the current 'total_sectors' value
 */
static int refresh_total_sectors(BlockDriverState *bs, int64_t hint)
{
    BlockDriver *drv = bs->drv;

603 604 605 606
    /* Do not attempt drv->bdrv_getlength() on scsi-generic devices */
    if (bs->sg)
        return 0;

607 608 609 610 611 612 613 614 615 616 617 618 619
    /* query actual device if possible, otherwise just trust the hint */
    if (drv->bdrv_getlength) {
        int64_t length = drv->bdrv_getlength(bs);
        if (length < 0) {
            return length;
        }
        hint = length >> BDRV_SECTOR_BITS;
    }

    bs->total_sectors = hint;
    return 0;
}

P
Paolo Bonzini 已提交
620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639
/**
 * 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;
}

640 641 642 643 644 645 646 647 648 649 650
/**
 * 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;
651 652
    } else if (!strcmp(mode, "directsync")) {
        *flags |= BDRV_O_NOCACHE;
653 654 655 656 657 658 659 660 661 662 663 664 665 666
    } 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;
}

667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682
/**
 * 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
683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702
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;
}

703 704
/*
 * Common part for opening disk images and files
705 706
 *
 * Removes all processed options from *options.
707
 */
708
static int bdrv_open_common(BlockDriverState *bs, BlockDriverState *file,
K
Kevin Wolf 已提交
709
    QDict *options, int flags, BlockDriver *drv)
710 711
{
    int ret, open_flags;
K
Kevin Wolf 已提交
712
    const char *filename;
713 714

    assert(drv != NULL);
715
    assert(bs->file == NULL);
716
    assert(options != NULL && bs->options != options);
717

718 719 720 721 722 723 724
    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
725

726 727 728 729 730 731 732 733
    /* 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;
    }

734 735
    bs->open_flags = flags;
    bs->buffer_alignment = 512;
A
Asias He 已提交
736
    bs->zero_beyond_eof = true;
737 738 739 740 741 742
    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)) {
        return -ENOTSUP;
    }
743

744
    assert(bs->copy_on_read == 0); /* bdrv_new() and bdrv_close() make it so */
745
    if (!bs->read_only && (flags & BDRV_O_COPY_ON_READ)) {
746 747 748
        bdrv_enable_copy_on_read(bs);
    }

749 750 751 752 753
    if (filename != NULL) {
        pstrcpy(bs->filename, sizeof(bs->filename), filename);
    } else {
        bs->filename[0] = '\0';
    }
754 755

    bs->drv = drv;
756
    bs->opaque = g_malloc0(drv->instance_size);
757

758
    bs->enable_write_cache = !!(flags & BDRV_O_CACHE_WB);
759

760 761
    /* Open the image, either directly or using a protocol */
    if (drv->bdrv_file_open) {
762 763
        assert(file == NULL);
        assert(drv->bdrv_parse_filename || filename != NULL);
764
        ret = drv->bdrv_file_open(bs, options, open_flags);
765
    } else {
766 767 768 769 770 771 772
        if (file == NULL) {
            qerror_report(ERROR_CLASS_GENERIC_ERROR, "Can't use '%s' as a "
                          "block driver for the protocol level",
                          drv->format_name);
            ret = -EINVAL;
            goto free_and_fail;
        }
773
        bs->file = file;
774
        ret = drv->bdrv_open(bs, options, open_flags);
775 776
    }

777 778 779 780
    if (ret < 0) {
        goto free_and_fail;
    }

781 782 783
    ret = refresh_total_sectors(bs, bs->total_sectors);
    if (ret < 0) {
        goto free_and_fail;
784
    }
785

786 787
#ifndef _WIN32
    if (bs->is_temporary) {
788
        assert(filename != NULL);
789 790 791 792 793 794
        unlink(filename);
    }
#endif
    return 0;

free_and_fail:
795
    bs->file = NULL;
796
    g_free(bs->opaque);
797 798 799 800 801
    bs->opaque = NULL;
    bs->drv = NULL;
    return ret;
}

K
Kevin Wolf 已提交
802 803
/*
 * Opens a file using a protocol (file, host_device, nbd, ...)
804 805 806 807 808
 *
 * 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
809
 */
810 811
int bdrv_file_open(BlockDriverState **pbs, const char *filename,
                   QDict *options, int flags)
B
bellard 已提交
812
{
B
bellard 已提交
813
    BlockDriverState *bs;
814
    BlockDriver *drv;
815
    const char *drvname;
816
    bool allow_protocol_prefix = false;
B
bellard 已提交
817 818
    int ret;

819 820 821 822 823
    /* NULL means an empty set of options */
    if (options == NULL) {
        options = qdict_new();
    }

B
bellard 已提交
824
    bs = bdrv_new("");
825 826 827
    bs->options = options;
    options = qdict_clone_shallow(options);

K
Kevin Wolf 已提交
828 829 830 831 832
    /* 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));
833
        allow_protocol_prefix = true;
K
Kevin Wolf 已提交
834 835 836 837 838 839 840
    } else {
        qerror_report(ERROR_CLASS_GENERIC_ERROR, "Can't specify 'file' and "
                      "'filename' options at the same time");
        ret = -EINVAL;
        goto fail;
    }

841 842 843
    /* Find the right block driver */
    drvname = qdict_get_try_str(options, "driver");
    if (drvname) {
844
        drv = bdrv_find_whitelisted_format(drvname, !(flags & BDRV_O_RDWR));
845 846
        qdict_del(options, "driver");
    } else if (filename) {
847 848 849 850
        drv = bdrv_find_protocol(filename, allow_protocol_prefix);
        if (!drv) {
            qerror_report(ERROR_CLASS_GENERIC_ERROR, "Unknown protocol");
        }
851 852 853 854 855 856 857 858 859 860 861 862 863
    } else {
        qerror_report(ERROR_CLASS_GENERIC_ERROR,
                      "Must specify either driver or file");
        drv = NULL;
    }

    if (!drv) {
        ret = -ENOENT;
        goto fail;
    }

    /* Parse the filename and open it */
    if (drv->bdrv_parse_filename && filename) {
864 865 866 867 868 869 870 871
        Error *local_err = NULL;
        drv->bdrv_parse_filename(filename, options, &local_err);
        if (error_is_set(&local_err)) {
            qerror_report_err(local_err);
            error_free(local_err);
            ret = -EINVAL;
            goto fail;
        }
872
        qdict_del(options, "filename");
873 874 875 876 877 878
    } else if (!drv->bdrv_parse_filename && !filename) {
        qerror_report(ERROR_CLASS_GENERIC_ERROR,
                      "The '%s' block driver requires a file name",
                      drv->format_name);
        ret = -EINVAL;
        goto fail;
879 880
    }

K
Kevin Wolf 已提交
881
    ret = bdrv_open_common(bs, NULL, options, flags, drv);
B
bellard 已提交
882
    if (ret < 0) {
883 884 885 886 887 888 889 890 891 892 893
        goto fail;
    }

    /* Check if any unknown options were used */
    if (qdict_size(options) != 0) {
        const QDictEntry *entry = qdict_first(options);
        qerror_report(ERROR_CLASS_GENERIC_ERROR, "Block protocol '%s' doesn't "
                      "support the option '%s'",
                      drv->format_name, entry->key);
        ret = -EINVAL;
        goto fail;
894
    }
895 896
    QDECREF(options);

897
    bs->growable = 1;
B
bellard 已提交
898 899
    *pbs = bs;
    return 0;
900 901 902 903 904 905

fail:
    QDECREF(options);
    if (!bs->drv) {
        QDECREF(bs->options);
    }
F
Fam Zheng 已提交
906
    bdrv_unref(bs);
907
    return ret;
B
bellard 已提交
908 909
}

910 911 912 913 914 915 916 917 918
/*
 * 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.
 */
int bdrv_open_backing_file(BlockDriverState *bs, QDict *options)
P
Paolo Bonzini 已提交
919 920 921 922 923 924
{
    char backing_filename[PATH_MAX];
    int back_flags, ret;
    BlockDriver *back_drv = NULL;

    if (bs->backing_hd != NULL) {
925
        QDECREF(options);
P
Paolo Bonzini 已提交
926 927 928
        return 0;
    }

929 930 931 932 933
    /* NULL means an empty set of options */
    if (options == NULL) {
        options = qdict_new();
    }

P
Paolo Bonzini 已提交
934
    bs->open_flags &= ~BDRV_O_NO_BACKING;
935 936 937
    if (qdict_haskey(options, "file.filename")) {
        backing_filename[0] = '\0';
    } else if (bs->backing_file[0] == '\0' && qdict_size(options) == 0) {
938
        QDECREF(options);
P
Paolo Bonzini 已提交
939 940 941 942 943 944 945 946 947 948 949 950 951 952
        return 0;
    }

    bs->backing_hd = bdrv_new("");
    bdrv_get_full_backing_filename(bs, backing_filename,
                                   sizeof(backing_filename));

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

    /* backing files always opened read-only */
    back_flags = bs->open_flags & ~(BDRV_O_RDWR | BDRV_O_SNAPSHOT);

953 954
    ret = bdrv_open(bs->backing_hd,
                    *backing_filename ? backing_filename : NULL, options,
955
                    back_flags, back_drv);
P
Paolo Bonzini 已提交
956
    if (ret < 0) {
F
Fam Zheng 已提交
957
        bdrv_unref(bs->backing_hd);
P
Paolo Bonzini 已提交
958 959 960 961 962 963 964
        bs->backing_hd = NULL;
        bs->open_flags |= BDRV_O_NO_BACKING;
        return ret;
    }
    return 0;
}

965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983
static void extract_subqdict(QDict *src, QDict **dst, const char *start)
{
    const QDictEntry *entry, *next;
    const char *p;

    *dst = qdict_new();
    entry = qdict_first(src);

    while (entry != NULL) {
        next = qdict_next(src, entry);
        if (strstart(entry->key, start, &p)) {
            qobject_incref(entry->value);
            qdict_put_obj(*dst, p, entry->value);
            qdict_del(src, entry->key);
        }
        entry = next;
    }
}

K
Kevin Wolf 已提交
984 985
/*
 * Opens a disk image (raw, qcow2, vmdk, ...)
986 987 988 989 990
 *
 * 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
991
 */
992 993
int bdrv_open(BlockDriverState *bs, const char *filename, QDict *options,
              int flags, BlockDriver *drv)
B
bellard 已提交
994
{
K
Kevin Wolf 已提交
995
    int ret;
996 997
    /* TODO: extra byte is a hack to ensure MAX_PATH space on Windows. */
    char tmp_filename[PATH_MAX + 1];
998
    BlockDriverState *file = NULL;
999
    QDict *file_options = NULL;
1000
    const char *drvname;
B
bellard 已提交
1001

1002 1003 1004 1005 1006 1007
    /* NULL means an empty set of options */
    if (options == NULL) {
        options = qdict_new();
    }

    bs->options = options;
1008
    options = qdict_clone_shallow(options);
1009 1010

    /* For snapshot=on, create a temporary qcow2 overlay */
B
bellard 已提交
1011
    if (flags & BDRV_O_SNAPSHOT) {
B
bellard 已提交
1012 1013
        BlockDriverState *bs1;
        int64_t total_size;
K
Kevin Wolf 已提交
1014
        BlockDriver *bdrv_qcow2;
1015
        QEMUOptionParameter *create_options;
K
Kevin Wolf 已提交
1016
        char backing_filename[PATH_MAX];
1017

1018 1019 1020 1021 1022 1023 1024
        if (qdict_size(options) != 0) {
            error_report("Can't use snapshot=on with driver-specific options");
            ret = -EINVAL;
            goto fail;
        }
        assert(filename != NULL);

B
bellard 已提交
1025 1026
        /* if snapshot, we create a temporary backing file and open it
           instead of opening 'filename' directly */
1027

B
bellard 已提交
1028 1029
        /* if there is a backing file, use it */
        bs1 = bdrv_new("");
1030
        ret = bdrv_open(bs1, filename, NULL, 0, drv);
1031
        if (ret < 0) {
F
Fam Zheng 已提交
1032
            bdrv_unref(bs1);
1033
            goto fail;
B
bellard 已提交
1034
        }
1035
        total_size = bdrv_getlength(bs1) & BDRV_SECTOR_MASK;
A
aliguori 已提交
1036

F
Fam Zheng 已提交
1037
        bdrv_unref(bs1);
1038

1039 1040
        ret = get_tmp_filename(tmp_filename, sizeof(tmp_filename));
        if (ret < 0) {
1041
            goto fail;
1042
        }
A
aliguori 已提交
1043 1044

        /* Real path is meaningless for protocols */
1045
        if (path_has_protocol(filename)) {
A
aliguori 已提交
1046 1047
            snprintf(backing_filename, sizeof(backing_filename),
                     "%s", filename);
1048 1049 1050 1051
        } else if (!realpath(filename, backing_filename)) {
            ret = -errno;
            goto fail;
        }
A
aliguori 已提交
1052

K
Kevin Wolf 已提交
1053
        bdrv_qcow2 = bdrv_find_format("qcow2");
1054 1055
        create_options = parse_option_parameters("", bdrv_qcow2->create_options,
                                                 NULL);
K
Kevin Wolf 已提交
1056

1057 1058 1059
        set_option_parameter_int(create_options, BLOCK_OPT_SIZE, total_size);
        set_option_parameter(create_options, BLOCK_OPT_BACKING_FILE,
                             backing_filename);
K
Kevin Wolf 已提交
1060
        if (drv) {
1061
            set_option_parameter(create_options, BLOCK_OPT_BACKING_FMT,
K
Kevin Wolf 已提交
1062 1063 1064
                drv->format_name);
        }

1065 1066
        ret = bdrv_create(bdrv_qcow2, tmp_filename, create_options);
        free_option_parameters(create_options);
1067
        if (ret < 0) {
1068
            goto fail;
B
bellard 已提交
1069
        }
K
Kevin Wolf 已提交
1070

B
bellard 已提交
1071
        filename = tmp_filename;
K
Kevin Wolf 已提交
1072
        drv = bdrv_qcow2;
B
bellard 已提交
1073 1074
        bs->is_temporary = 1;
    }
B
bellard 已提交
1075

1076 1077 1078 1079 1080
    /* Open image file without format layer */
    if (flags & BDRV_O_RDWR) {
        flags |= BDRV_O_ALLOW_RDWR;
    }

1081 1082 1083
    extract_subqdict(options, &file_options, "file.");

    ret = bdrv_file_open(&file, filename, file_options,
1084
                         bdrv_open_flags(bs, flags | BDRV_O_UNMAP));
1085
    if (ret < 0) {
1086
        goto fail;
1087 1088
    }

K
Kevin Wolf 已提交
1089
    /* Find the right image format driver */
1090 1091 1092 1093 1094 1095
    drvname = qdict_get_try_str(options, "driver");
    if (drvname) {
        drv = bdrv_find_whitelisted_format(drvname, !(flags & BDRV_O_RDWR));
        qdict_del(options, "driver");
    }

1096
    if (!drv) {
1097
        ret = find_image_format(file, filename, &drv);
1098
    }
1099

1100 1101
    if (!drv) {
        goto unlink_and_fail;
B
bellard 已提交
1102
    }
K
Kevin Wolf 已提交
1103 1104

    /* Open the image */
K
Kevin Wolf 已提交
1105
    ret = bdrv_open_common(bs, file, options, flags, drv);
K
Kevin Wolf 已提交
1106
    if (ret < 0) {
1107 1108 1109
        goto unlink_and_fail;
    }

1110
    if (bs->file != file) {
F
Fam Zheng 已提交
1111
        bdrv_unref(file);
1112 1113 1114
        file = NULL;
    }

K
Kevin Wolf 已提交
1115
    /* If there is a backing file, use it */
P
Paolo Bonzini 已提交
1116
    if ((flags & BDRV_O_NO_BACKING) == 0) {
1117 1118 1119 1120
        QDict *backing_options;

        extract_subqdict(options, &backing_options, "backing.");
        ret = bdrv_open_backing_file(bs, backing_options);
K
Kevin Wolf 已提交
1121
        if (ret < 0) {
1122
            goto close_and_fail;
K
Kevin Wolf 已提交
1123 1124 1125
        }
    }

1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137
    /* Check if any unknown options were used */
    if (qdict_size(options) != 0) {
        const QDictEntry *entry = qdict_first(options);
        qerror_report(ERROR_CLASS_GENERIC_ERROR, "Block format '%s' used by "
            "device '%s' doesn't support the option '%s'",
            drv->format_name, bs->device_name, entry->key);

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

K
Kevin Wolf 已提交
1138
    if (!bdrv_key_required(bs)) {
1139
        bdrv_dev_change_media_cb(bs, true);
K
Kevin Wolf 已提交
1140 1141 1142 1143 1144
    }

    return 0;

unlink_and_fail:
1145
    if (file != NULL) {
F
Fam Zheng 已提交
1146
        bdrv_unref(file);
1147
    }
K
Kevin Wolf 已提交
1148 1149 1150
    if (bs->is_temporary) {
        unlink(filename);
    }
1151 1152
fail:
    QDECREF(bs->options);
1153
    QDECREF(options);
1154
    bs->options = NULL;
1155
    return ret;
1156

1157 1158 1159
close_and_fail:
    bdrv_close(bs);
    QDECREF(options);
K
Kevin Wolf 已提交
1160 1161 1162
    return ret;
}

1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330
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 {
1331 1332
                error_setg(errp, "failed while preparing to reopen image '%s'",
                           reopen_state->bs->filename);
1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394
            }
            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
1395 1396
void bdrv_close(BlockDriverState *bs)
{
1397 1398 1399
    if (bs->job) {
        block_job_cancel_sync(bs->job);
    }
1400 1401 1402
    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
1403
    notifier_list_notify(&bs->close_notifiers, bs);
K
Kevin Wolf 已提交
1404

1405
    if (bs->drv) {
1406
        if (bs->backing_hd) {
F
Fam Zheng 已提交
1407
            bdrv_unref(bs->backing_hd);
1408 1409
            bs->backing_hd = NULL;
        }
B
bellard 已提交
1410
        bs->drv->bdrv_close(bs);
1411
        g_free(bs->opaque);
B
bellard 已提交
1412 1413 1414 1415
#ifdef _WIN32
        if (bs->is_temporary) {
            unlink(bs->filename);
        }
B
bellard 已提交
1416
#endif
B
bellard 已提交
1417 1418
        bs->opaque = NULL;
        bs->drv = NULL;
1419
        bs->copy_on_read = 0;
1420 1421
        bs->backing_file[0] = '\0';
        bs->backing_format[0] = '\0';
1422 1423 1424 1425 1426
        bs->total_sectors = 0;
        bs->encrypted = 0;
        bs->valid_key = 0;
        bs->sg = 0;
        bs->growable = 0;
A
Asias He 已提交
1427
        bs->zero_beyond_eof = false;
1428 1429
        QDECREF(bs->options);
        bs->options = NULL;
B
bellard 已提交
1430

1431
        if (bs->file != NULL) {
F
Fam Zheng 已提交
1432
            bdrv_unref(bs->file);
1433
            bs->file = NULL;
1434
        }
B
bellard 已提交
1435
    }
Z
Zhi Yong Wu 已提交
1436

P
Pavel Hrdina 已提交
1437 1438
    bdrv_dev_change_media_cb(bs, false);

Z
Zhi Yong Wu 已提交
1439 1440 1441 1442
    /*throttling disk I/O limits*/
    if (bs->io_limits_enabled) {
        bdrv_io_limits_disable(bs);
    }
B
bellard 已提交
1443 1444
}

1445 1446 1447 1448 1449 1450 1451 1452 1453
void bdrv_close_all(void)
{
    BlockDriverState *bs;

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

1454 1455 1456 1457 1458 1459
/* 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;
    }
1460 1461 1462 1463
    if (!qemu_co_queue_empty(&bs->throttled_reqs[0])) {
        return true;
    }
    if (!qemu_co_queue_empty(&bs->throttled_reqs[1])) {
1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485
        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;
    QTAILQ_FOREACH(bs, &bdrv_states, list) {
        if (bdrv_requests_pending(bs)) {
            return true;
        }
    }
    return false;
}

1486 1487 1488 1489 1490
/*
 * 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.
1491 1492 1493 1494 1495 1496
 *
 * 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.
1497 1498 1499
 */
void bdrv_drain_all(void)
{
1500 1501
    /* Always run first iteration so any pending completion BHs run */
    bool busy = true;
1502 1503
    BlockDriverState *bs;

1504
    while (busy) {
1505 1506 1507 1508
        /* FIXME: We do not have timer support here, so this is effectively
         * a busy wait.
         */
        QTAILQ_FOREACH(bs, &bdrv_states, list) {
1509
            if (bdrv_start_throttled_reqs(bs)) {
1510 1511 1512
                busy = true;
            }
        }
1513

1514 1515
        busy = bdrv_requests_pending_all();
        busy |= aio_poll(qemu_get_aio_context(), busy);
1516 1517 1518
    }
}

1519 1520 1521 1522 1523 1524 1525 1526 1527 1528
/* make a BlockDriverState anonymous by removing from bdrv_state list.
   Also, NULL terminate the device_name to prevent double remove */
void bdrv_make_anon(BlockDriverState *bs)
{
    if (bs->device_name[0] != '\0') {
        QTAILQ_REMOVE(&bdrv_states, bs, list);
    }
    bs->device_name[0] = '\0';
}

P
Paolo Bonzini 已提交
1529 1530 1531 1532 1533 1534 1535
static void bdrv_rebind(BlockDriverState *bs)
{
    if (bs->drv && bs->drv->bdrv_rebind) {
        bs->drv->bdrv_rebind(bs);
    }
}

1536 1537
static void bdrv_move_feature_fields(BlockDriverState *bs_dest,
                                     BlockDriverState *bs_src)
1538
{
1539 1540
    /* move some fields that need to stay attached to the device */
    bs_dest->open_flags         = bs_src->open_flags;
1541 1542

    /* dev info */
1543 1544 1545 1546 1547
    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;
1548

1549
    bs_dest->enable_write_cache = bs_src->enable_write_cache;
1550

1551 1552 1553 1554 1555 1556
    /* 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];
1557
    bs_dest->io_limits_enabled  = bs_src->io_limits_enabled;
1558 1559

    /* r/w error */
1560 1561
    bs_dest->on_read_error      = bs_src->on_read_error;
    bs_dest->on_write_error     = bs_src->on_write_error;
1562 1563

    /* i/o status */
1564 1565
    bs_dest->iostatus_enabled   = bs_src->iostatus_enabled;
    bs_dest->iostatus           = bs_src->iostatus;
1566

1567
    /* dirty bitmap */
1568
    bs_dest->dirty_bitmap       = bs_src->dirty_bitmap;
1569

1570 1571 1572
    /* reference count */
    bs_dest->refcnt             = bs_src->refcnt;

1573
    /* job */
1574 1575
    bs_dest->in_use             = bs_src->in_use;
    bs_dest->job                = bs_src->job;
1576

1577
    /* keep the same entry in bdrv_states */
1578 1579 1580 1581
    pstrcpy(bs_dest->device_name, sizeof(bs_dest->device_name),
            bs_src->device_name);
    bs_dest->list = bs_src->list;
}
1582

1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597
/*
 * 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
1598

1599 1600 1601 1602 1603 1604 1605
    /* bs_new must be anonymous and shouldn't have anything fancy enabled */
    assert(bs_new->device_name[0] == '\0');
    assert(bs_new->dirty_bitmap == NULL);
    assert(bs_new->job == NULL);
    assert(bs_new->dev == NULL);
    assert(bs_new->in_use == 0);
    assert(bs_new->io_limits_enabled == false);
1606
    assert(!throttle_have_timer(&bs_new->throttle_state));
1607

1608 1609 1610
    tmp = *bs_new;
    *bs_new = *bs_old;
    *bs_old = tmp;
1611

1612 1613 1614 1615
    /* 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);
1616

1617 1618 1619 1620 1621 1622 1623 1624
    /* 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);
1625
    assert(!throttle_have_timer(&bs_new->throttle_state));
P
Paolo Bonzini 已提交
1626 1627

    bdrv_rebind(bs_new);
1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653
    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 : "");
1654 1655
}

F
Fam Zheng 已提交
1656
static void bdrv_delete(BlockDriverState *bs)
B
bellard 已提交
1657
{
1658
    assert(!bs->dev);
1659 1660
    assert(!bs->job);
    assert(!bs->in_use);
F
Fam Zheng 已提交
1661
    assert(!bs->refcnt);
1662

1663 1664
    bdrv_close(bs);

1665
    /* remove from list, if necessary */
1666
    bdrv_make_anon(bs);
1667

1668
    g_free(bs);
B
bellard 已提交
1669 1670
}

1671 1672
int bdrv_attach_dev(BlockDriverState *bs, void *dev)
/* TODO change to DeviceState *dev when all users are qdevified */
1673
{
1674
    if (bs->dev) {
1675 1676
        return -EBUSY;
    }
1677
    bs->dev = dev;
1678
    bdrv_iostatus_reset(bs);
1679 1680 1681
    return 0;
}

1682 1683
/* TODO qdevified devices don't use this, remove when devices are qdevified */
void bdrv_attach_dev_nofail(BlockDriverState *bs, void *dev)
1684
{
1685 1686 1687 1688 1689 1690 1691 1692 1693 1694
    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;
1695 1696
    bs->dev_ops = NULL;
    bs->dev_opaque = NULL;
1697
    bs->buffer_alignment = 512;
1698 1699
}

1700 1701
/* TODO change to return DeviceState * when all users are qdevified */
void *bdrv_get_attached_dev(BlockDriverState *bs)
1702
{
1703
    return bs->dev;
1704 1705
}

1706 1707 1708 1709 1710 1711 1712
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
1713 1714 1715
void bdrv_emit_qmp_error_event(const BlockDriverState *bdrv,
                               enum MonitorEvent ev,
                               BlockErrorAction action, bool is_read)
1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737
{
    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
1738
    monitor_protocol_event(ev, data);
1739 1740 1741 1742

    qobject_decref(data);
}

1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753
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);
}

1754
static void bdrv_dev_change_media_cb(BlockDriverState *bs, bool load)
1755
{
1756
    if (bs->dev_ops && bs->dev_ops->change_media_cb) {
1757
        bool tray_was_closed = !bdrv_dev_is_tray_open(bs);
1758
        bs->dev_ops->change_media_cb(bs->dev_opaque, load);
1759 1760 1761 1762 1763 1764 1765 1766
        if (tray_was_closed) {
            /* tray open */
            bdrv_emit_qmp_eject_event(bs, true);
        }
        if (load) {
            /* tray close */
            bdrv_emit_qmp_eject_event(bs, false);
        }
1767 1768 1769
    }
}

1770 1771 1772 1773 1774
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
1775 1776 1777 1778 1779 1780 1781
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);
    }
}

1782 1783 1784 1785 1786 1787 1788 1789
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;
}

1790 1791 1792 1793
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);
1794 1795 1796
    }
}

1797 1798 1799 1800 1801 1802 1803 1804
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
1805 1806 1807
/*
 * Run consistency checks on an image
 *
1808
 * Returns 0 if the check could be completed (it doesn't mean that the image is
1809
 * free of errors) or -errno when an internal error occurred. The results of the
1810
 * check are stored in res.
A
aliguori 已提交
1811
 */
1812
int bdrv_check(BlockDriverState *bs, BdrvCheckResult *res, BdrvCheckMode fix)
A
aliguori 已提交
1813 1814 1815 1816 1817
{
    if (bs->drv->bdrv_check == NULL) {
        return -ENOTSUP;
    }

1818
    memset(res, 0, sizeof(*res));
1819
    return bs->drv->bdrv_check(bs, res, fix);
A
aliguori 已提交
1820 1821
}

1822 1823
#define COMMIT_BUF_SECTORS 2048

1824 1825 1826
/* commit COW file into the raw image */
int bdrv_commit(BlockDriverState *bs)
{
B
bellard 已提交
1827
    BlockDriver *drv = bs->drv;
1828 1829
    int64_t sector, total_sectors;
    int n, ro, open_flags;
1830
    int ret = 0;
1831
    uint8_t *buf;
1832
    char filename[PATH_MAX];
1833

B
bellard 已提交
1834 1835
    if (!drv)
        return -ENOMEDIUM;
1836 1837 1838
    
    if (!bs->backing_hd) {
        return -ENOTSUP;
1839 1840
    }

1841 1842 1843 1844
    if (bdrv_in_use(bs) || bdrv_in_use(bs->backing_hd)) {
        return -EBUSY;
    }

1845
    ro = bs->backing_hd->read_only;
1846 1847
    /* Use pstrcpy (not strncpy): filename must be NUL-terminated. */
    pstrcpy(filename, sizeof(filename), bs->backing_hd->filename);
1848 1849 1850
    open_flags =  bs->backing_hd->open_flags;

    if (ro) {
1851 1852
        if (bdrv_reopen(bs->backing_hd, open_flags | BDRV_O_RDWR, NULL)) {
            return -EACCES;
1853
        }
B
bellard 已提交
1854
    }
1855

J
Jan Kiszka 已提交
1856
    total_sectors = bdrv_getlength(bs) >> BDRV_SECTOR_BITS;
1857
    buf = g_malloc(COMMIT_BUF_SECTORS * BDRV_SECTOR_SIZE);
1858 1859

    for (sector = 0; sector < total_sectors; sector += n) {
1860 1861 1862 1863 1864
        ret = bdrv_is_allocated(bs, sector, COMMIT_BUF_SECTORS, &n);
        if (ret < 0) {
            goto ro_cleanup;
        }
        if (ret) {
1865 1866 1867 1868 1869 1870 1871 1872 1873
            if (bdrv_read(bs, sector, buf, n) != 0) {
                ret = -EIO;
                goto ro_cleanup;
            }

            if (bdrv_write(bs->backing_hd, sector, buf, n) != 0) {
                ret = -EIO;
                goto ro_cleanup;
            }
B
bellard 已提交
1874
        }
1875
    }
1876

1877 1878 1879 1880
    if (drv->bdrv_make_empty) {
        ret = drv->bdrv_make_empty(bs);
        bdrv_flush(bs);
    }
1881

1882 1883 1884 1885 1886 1887
    /*
     * Make sure all data we wrote to the backing device is actually
     * stable on disk.
     */
    if (bs->backing_hd)
        bdrv_flush(bs->backing_hd);
1888 1889

ro_cleanup:
1890
    g_free(buf);
1891 1892

    if (ro) {
1893 1894
        /* ignoring error return here */
        bdrv_reopen(bs->backing_hd, open_flags & ~BDRV_O_RDWR, NULL);
1895 1896
    }

1897
    return ret;
1898 1899
}

1900
int bdrv_commit_all(void)
1901 1902 1903 1904
{
    BlockDriverState *bs;

    QTAILQ_FOREACH(bs, &bdrv_states, list) {
1905 1906 1907 1908 1909
        if (bs->drv && bs->backing_hd) {
            int ret = bdrv_commit(bs);
            if (ret < 0) {
                return ret;
            }
1910
        }
1911
    }
1912
    return 0;
1913 1914
}

S
Stefan Hajnoczi 已提交
1915 1916 1917 1918 1919 1920 1921 1922
/**
 * 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);
1923
    qemu_co_queue_restart_all(&req->wait_queue);
S
Stefan Hajnoczi 已提交
1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938
}

/**
 * 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,
1939
        .co = qemu_coroutine_self(),
S
Stefan Hajnoczi 已提交
1940 1941
    };

1942 1943
    qemu_co_queue_init(&req->wait_queue);

S
Stefan Hajnoczi 已提交
1944 1945 1946
    QLIST_INSERT_HEAD(&bs->tracked_requests, req, list);
}

S
Stefan Hajnoczi 已提交
1947 1948 1949
/**
 * Round a region to cluster boundaries
 */
1950 1951 1952 1953
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
1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967
{
    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);
    }
}

1968 1969
static bool tracked_request_overlaps(BdrvTrackedRequest *req,
                                     int64_t sector_num, int nb_sectors) {
S
Stefan Hajnoczi 已提交
1970 1971 1972 1973 1974 1975 1976 1977 1978
    /*        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;
1979 1980 1981 1982 1983 1984
}

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
1985 1986
    int64_t cluster_sector_num;
    int cluster_nb_sectors;
1987 1988
    bool retry;

S
Stefan Hajnoczi 已提交
1989 1990 1991 1992 1993 1994
    /* 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.
     */
1995 1996
    bdrv_round_to_clusters(bs, sector_num, nb_sectors,
                           &cluster_sector_num, &cluster_nb_sectors);
S
Stefan Hajnoczi 已提交
1997

1998 1999 2000
    do {
        retry = false;
        QLIST_FOREACH(req, &bs->tracked_requests, list) {
S
Stefan Hajnoczi 已提交
2001 2002
            if (tracked_request_overlaps(req, cluster_sector_num,
                                         cluster_nb_sectors)) {
2003 2004 2005 2006 2007 2008
                /* 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);

2009 2010 2011 2012 2013 2014 2015 2016
                qemu_co_queue_wait(&req->wait_queue);
                retry = true;
                break;
            }
        }
    } while (retry);
}

K
Kevin Wolf 已提交
2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028
/*
 * 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;
2029
    int ret;
K
Kevin Wolf 已提交
2030

2031 2032 2033 2034 2035
    /* Backing file format doesn't make sense without a backing file */
    if (backing_fmt && !backing_file) {
        return -EINVAL;
    }

K
Kevin Wolf 已提交
2036
    if (drv->bdrv_change_backing_file != NULL) {
2037
        ret = drv->bdrv_change_backing_file(bs, backing_file, backing_fmt);
K
Kevin Wolf 已提交
2038
    } else {
2039
        ret = -ENOTSUP;
K
Kevin Wolf 已提交
2040
    }
2041 2042 2043 2044 2045 2046

    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
2047 2048
}

2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 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 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179
/*
 * 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
2180
        bdrv_unref(intermediate_state->bs);
2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191
    }
    ret = 0;

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


2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204
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
2205 2206 2207 2208
    if (offset < 0)
        return -EIO;

    if ((offset > len) || (len - offset < size))
2209 2210 2211 2212 2213 2214 2215 2216
        return -EIO;

    return 0;
}

static int bdrv_check_request(BlockDriverState *bs, int64_t sector_num,
                              int nb_sectors)
{
2217 2218
    return bdrv_check_byte_request(bs, sector_num * BDRV_SECTOR_SIZE,
                                   nb_sectors * BDRV_SECTOR_SIZE);
2219 2220
}

2221 2222 2223 2224 2225 2226 2227
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
2228
    BdrvRequestFlags flags;
2229 2230 2231
} RwCo;

static void coroutine_fn bdrv_rw_co_entry(void *opaque)
B
bellard 已提交
2232
{
2233
    RwCo *rwco = opaque;
B
bellard 已提交
2234

2235 2236
    if (!rwco->is_write) {
        rwco->ret = bdrv_co_do_readv(rwco->bs, rwco->sector_num,
P
Peter Lieven 已提交
2237 2238
                                     rwco->nb_sectors, rwco->qiov,
                                     rwco->flags);
2239 2240
    } else {
        rwco->ret = bdrv_co_do_writev(rwco->bs, rwco->sector_num,
P
Peter Lieven 已提交
2241 2242
                                      rwco->nb_sectors, rwco->qiov,
                                      rwco->flags);
2243 2244
    }
}
2245

2246
/*
2247
 * Process a vectored synchronous request using coroutines
2248
 */
2249
static int bdrv_rwv_co(BlockDriverState *bs, int64_t sector_num,
P
Peter Lieven 已提交
2250 2251
                       QEMUIOVector *qiov, bool is_write,
                       BdrvRequestFlags flags)
2252 2253 2254 2255 2256
{
    Coroutine *co;
    RwCo rwco = {
        .bs = bs,
        .sector_num = sector_num,
2257 2258
        .nb_sectors = qiov->size >> BDRV_SECTOR_BITS,
        .qiov = qiov,
2259 2260
        .is_write = is_write,
        .ret = NOT_DONE,
P
Peter Lieven 已提交
2261
        .flags = flags,
2262
    };
2263
    assert((qiov->size & (BDRV_SECTOR_SIZE - 1)) == 0);
2264

2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275
    /**
     * 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);
    }

2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287
    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
2288

2289 2290 2291 2292
/*
 * 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
2293
                      int nb_sectors, bool is_write, BdrvRequestFlags flags)
2294 2295 2296 2297 2298 2299 2300 2301
{
    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
2302
    return bdrv_rwv_co(bs, sector_num, &qiov, is_write, flags);
2303 2304
}

2305 2306 2307 2308
/* 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
2309
    return bdrv_rw_co(bs, sector_num, buf, nb_sectors, false, 0);
B
bellard 已提交
2310 2311
}

2312 2313 2314 2315 2316 2317 2318 2319 2320
/* 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;
2321
    ret = bdrv_read(bs, sector_num, buf, nb_sectors);
2322 2323 2324 2325
    bs->io_limits_enabled = enabled;
    return ret;
}

2326
/* Return < 0 if error. Important errors are:
B
bellard 已提交
2327 2328 2329 2330 2331
  -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
*/
2332
int bdrv_write(BlockDriverState *bs, int64_t sector_num,
B
bellard 已提交
2333 2334
               const uint8_t *buf, int nb_sectors)
{
P
Peter Lieven 已提交
2335
    return bdrv_rw_co(bs, sector_num, (uint8_t *)buf, nb_sectors, true, 0);
B
bellard 已提交
2336 2337
}

2338 2339
int bdrv_writev(BlockDriverState *bs, int64_t sector_num, QEMUIOVector *qiov)
{
P
Peter Lieven 已提交
2340 2341 2342 2343 2344 2345 2346
    return bdrv_rwv_co(bs, sector_num, qiov, true, 0);
}

int bdrv_write_zeroes(BlockDriverState *bs, int64_t sector_num, int nb_sectors)
{
    return bdrv_rw_co(bs, sector_num, NULL, nb_sectors, true,
                      BDRV_REQ_ZERO_WRITE);
2347 2348
}

2349 2350
int bdrv_pread(BlockDriverState *bs, int64_t offset,
               void *buf, int count1)
B
bellard 已提交
2351
{
J
Jan Kiszka 已提交
2352
    uint8_t tmp_buf[BDRV_SECTOR_SIZE];
B
bellard 已提交
2353 2354
    int len, nb_sectors, count;
    int64_t sector_num;
2355
    int ret;
B
bellard 已提交
2356 2357 2358

    count = count1;
    /* first read to align to sector start */
J
Jan Kiszka 已提交
2359
    len = (BDRV_SECTOR_SIZE - offset) & (BDRV_SECTOR_SIZE - 1);
B
bellard 已提交
2360 2361
    if (len > count)
        len = count;
J
Jan Kiszka 已提交
2362
    sector_num = offset >> BDRV_SECTOR_BITS;
B
bellard 已提交
2363
    if (len > 0) {
2364 2365
        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
J
Jan Kiszka 已提交
2366
        memcpy(buf, tmp_buf + (offset & (BDRV_SECTOR_SIZE - 1)), len);
B
bellard 已提交
2367 2368 2369 2370 2371 2372 2373 2374
        count -= len;
        if (count == 0)
            return count1;
        sector_num++;
        buf += len;
    }

    /* read the sectors "in place" */
J
Jan Kiszka 已提交
2375
    nb_sectors = count >> BDRV_SECTOR_BITS;
B
bellard 已提交
2376
    if (nb_sectors > 0) {
2377 2378
        if ((ret = bdrv_read(bs, sector_num, buf, nb_sectors)) < 0)
            return ret;
B
bellard 已提交
2379
        sector_num += nb_sectors;
J
Jan Kiszka 已提交
2380
        len = nb_sectors << BDRV_SECTOR_BITS;
B
bellard 已提交
2381 2382 2383 2384 2385 2386
        buf += len;
        count -= len;
    }

    /* add data from the last sector */
    if (count > 0) {
2387 2388
        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
B
bellard 已提交
2389 2390 2391 2392 2393
        memcpy(buf, tmp_buf, count);
    }
    return count1;
}

2394
int bdrv_pwritev(BlockDriverState *bs, int64_t offset, QEMUIOVector *qiov)
B
bellard 已提交
2395
{
J
Jan Kiszka 已提交
2396
    uint8_t tmp_buf[BDRV_SECTOR_SIZE];
B
bellard 已提交
2397 2398
    int len, nb_sectors, count;
    int64_t sector_num;
2399
    int ret;
B
bellard 已提交
2400

2401 2402
    count = qiov->size;

B
bellard 已提交
2403
    /* first write to align to sector start */
J
Jan Kiszka 已提交
2404
    len = (BDRV_SECTOR_SIZE - offset) & (BDRV_SECTOR_SIZE - 1);
B
bellard 已提交
2405 2406
    if (len > count)
        len = count;
J
Jan Kiszka 已提交
2407
    sector_num = offset >> BDRV_SECTOR_BITS;
B
bellard 已提交
2408
    if (len > 0) {
2409 2410
        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
2411 2412
        qemu_iovec_to_buf(qiov, 0, tmp_buf + (offset & (BDRV_SECTOR_SIZE - 1)),
                          len);
2413 2414
        if ((ret = bdrv_write(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
B
bellard 已提交
2415 2416
        count -= len;
        if (count == 0)
2417
            return qiov->size;
B
bellard 已提交
2418 2419 2420 2421
        sector_num++;
    }

    /* write the sectors "in place" */
J
Jan Kiszka 已提交
2422
    nb_sectors = count >> BDRV_SECTOR_BITS;
B
bellard 已提交
2423
    if (nb_sectors > 0) {
2424 2425 2426 2427 2428 2429 2430 2431
        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) {
2432
            return ret;
2433 2434
        }

B
bellard 已提交
2435
        sector_num += nb_sectors;
J
Jan Kiszka 已提交
2436
        len = nb_sectors << BDRV_SECTOR_BITS;
B
bellard 已提交
2437 2438 2439 2440 2441
        count -= len;
    }

    /* add data from the last sector */
    if (count > 0) {
2442 2443
        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
2444
        qemu_iovec_to_buf(qiov, qiov->size - count, tmp_buf, count);
2445 2446
        if ((ret = bdrv_write(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
B
bellard 已提交
2447
    }
2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461
    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
2462 2463
}

K
Kevin Wolf 已提交
2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479
/*
 * 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;
    }

2480 2481
    /* No flush needed for cache modes that already do it */
    if (bs->enable_write_cache) {
K
Kevin Wolf 已提交
2482 2483 2484 2485 2486 2487
        bdrv_flush(bs);
    }

    return 0;
}

2488
static int coroutine_fn bdrv_co_do_copy_on_readv(BlockDriverState *bs,
S
Stefan Hajnoczi 已提交
2489 2490 2491 2492 2493 2494 2495 2496 2497
        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;

2498
    BlockDriver *drv = bs->drv;
S
Stefan Hajnoczi 已提交
2499 2500 2501 2502 2503 2504 2505 2506 2507 2508
    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.
     */
2509 2510
    bdrv_round_to_clusters(bs, sector_num, nb_sectors,
                           &cluster_sector_num, &cluster_nb_sectors);
S
Stefan Hajnoczi 已提交
2511

2512 2513
    trace_bdrv_co_do_copy_on_readv(bs, sector_num, nb_sectors,
                                   cluster_sector_num, cluster_nb_sectors);
S
Stefan Hajnoczi 已提交
2514 2515 2516 2517 2518

    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);

2519 2520
    ret = drv->bdrv_co_readv(bs, cluster_sector_num, cluster_nb_sectors,
                             &bounce_qiov);
S
Stefan Hajnoczi 已提交
2521 2522 2523 2524
    if (ret < 0) {
        goto err;
    }

2525 2526
    if (drv->bdrv_co_write_zeroes &&
        buffer_is_zero(bounce_buffer, iov.iov_len)) {
K
Kevin Wolf 已提交
2527 2528
        ret = bdrv_co_do_write_zeroes(bs, cluster_sector_num,
                                      cluster_nb_sectors);
2529
    } else {
2530 2531 2532
        /* This does not change the data on the disk, it is not necessary
         * to flush even in cache=writethrough mode.
         */
2533
        ret = drv->bdrv_co_writev(bs, cluster_sector_num, cluster_nb_sectors,
S
Stefan Hajnoczi 已提交
2534
                                  &bounce_qiov);
2535 2536
    }

S
Stefan Hajnoczi 已提交
2537 2538 2539 2540 2541 2542 2543 2544 2545
    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;
2546 2547
    qemu_iovec_from_buf(qiov, 0, bounce_buffer + skip_bytes,
                        nb_sectors * BDRV_SECTOR_SIZE);
S
Stefan Hajnoczi 已提交
2548 2549 2550 2551 2552 2553

err:
    qemu_vfree(bounce_buffer);
    return ret;
}

2554 2555 2556 2557
/*
 * Handle a read request in coroutine context
 */
static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,
2558 2559
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags)
K
Kevin Wolf 已提交
2560 2561
{
    BlockDriver *drv = bs->drv;
S
Stefan Hajnoczi 已提交
2562 2563
    BdrvTrackedRequest req;
    int ret;
K
Kevin Wolf 已提交
2564 2565 2566 2567 2568 2569 2570 2571

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

2572
    if (bs->copy_on_read) {
2573 2574 2575 2576 2577 2578 2579
        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) {
2580 2581 2582
        wait_for_overlapping_requests(bs, sector_num, nb_sectors);
    }

2583 2584 2585 2586 2587
    /* throttling disk I/O */
    if (bs->io_limits_enabled) {
        bdrv_io_limits_intercept(bs, nb_sectors, false);
    }

S
Stefan Hajnoczi 已提交
2588
    tracked_request_begin(&req, bs, sector_num, nb_sectors, false);
S
Stefan Hajnoczi 已提交
2589

2590
    if (flags & BDRV_REQ_COPY_ON_READ) {
S
Stefan Hajnoczi 已提交
2591 2592
        int pnum;

2593
        ret = bdrv_is_allocated(bs, sector_num, nb_sectors, &pnum);
S
Stefan Hajnoczi 已提交
2594 2595 2596 2597 2598
        if (ret < 0) {
            goto out;
        }

        if (!ret || pnum != nb_sectors) {
2599
            ret = bdrv_co_do_copy_on_readv(bs, sector_num, nb_sectors, qiov);
S
Stefan Hajnoczi 已提交
2600 2601 2602 2603
            goto out;
        }
    }

2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632
    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;
        }

        total_sectors = len >> BDRV_SECTOR_BITS;
        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
2633 2634

out:
S
Stefan Hajnoczi 已提交
2635
    tracked_request_end(&req);
2636 2637 2638 2639 2640

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

S
Stefan Hajnoczi 已提交
2641
    return ret;
K
Kevin Wolf 已提交
2642 2643
}

2644
int coroutine_fn bdrv_co_readv(BlockDriverState *bs, int64_t sector_num,
K
Kevin Wolf 已提交
2645 2646
    int nb_sectors, QEMUIOVector *qiov)
{
2647
    trace_bdrv_co_readv(bs, sector_num, nb_sectors);
K
Kevin Wolf 已提交
2648

2649 2650 2651 2652 2653 2654 2655 2656 2657 2658
    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);
2659 2660
}

2661 2662 2663 2664 2665 2666 2667 2668
static int coroutine_fn bdrv_co_do_write_zeroes(BlockDriverState *bs,
    int64_t sector_num, int nb_sectors)
{
    BlockDriver *drv = bs->drv;
    QEMUIOVector qiov;
    struct iovec iov;
    int ret;

K
Kevin Wolf 已提交
2669 2670 2671
    /* TODO Emulate only part of misaligned requests instead of letting block
     * drivers return -ENOTSUP and emulate everything */

2672 2673
    /* First try the efficient write zeroes operation */
    if (drv->bdrv_co_write_zeroes) {
K
Kevin Wolf 已提交
2674 2675 2676 2677
        ret = drv->bdrv_co_write_zeroes(bs, sector_num, nb_sectors);
        if (ret != -ENOTSUP) {
            return ret;
        }
2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691
    }

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

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

    qemu_vfree(iov.iov_base);
    return ret;
}

2692 2693 2694 2695
/*
 * Handle a write request in coroutine context
 */
static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs,
2696 2697
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags)
2698 2699
{
    BlockDriver *drv = bs->drv;
S
Stefan Hajnoczi 已提交
2700
    BdrvTrackedRequest req;
2701
    int ret;
K
Kevin Wolf 已提交
2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712

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

2713
    if (bs->copy_on_read_in_flight) {
2714 2715 2716
        wait_for_overlapping_requests(bs, sector_num, nb_sectors);
    }

2717 2718 2719 2720 2721
    /* throttling disk I/O */
    if (bs->io_limits_enabled) {
        bdrv_io_limits_intercept(bs, nb_sectors, true);
    }

S
Stefan Hajnoczi 已提交
2722 2723
    tracked_request_begin(&req, bs, sector_num, nb_sectors, true);

2724 2725 2726 2727 2728
    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) {
2729 2730 2731 2732
        ret = bdrv_co_do_write_zeroes(bs, sector_num, nb_sectors);
    } else {
        ret = drv->bdrv_co_writev(bs, sector_num, nb_sectors, qiov);
    }
2733

2734 2735 2736 2737
    if (ret == 0 && !bs->enable_write_cache) {
        ret = bdrv_co_flush(bs);
    }

K
Kevin Wolf 已提交
2738
    if (bs->dirty_bitmap) {
2739
        bdrv_set_dirty(bs, sector_num, nb_sectors);
K
Kevin Wolf 已提交
2740 2741 2742 2743 2744
    }

    if (bs->wr_highest_sector < sector_num + nb_sectors - 1) {
        bs->wr_highest_sector = sector_num + nb_sectors - 1;
    }
2745 2746 2747
    if (bs->growable && ret >= 0) {
        bs->total_sectors = MAX(bs->total_sectors, sector_num + nb_sectors);
    }
K
Kevin Wolf 已提交
2748

S
Stefan Hajnoczi 已提交
2749 2750
    tracked_request_end(&req);

2751
    return ret;
K
Kevin Wolf 已提交
2752 2753
}

2754 2755 2756 2757 2758
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);

2759 2760 2761 2762 2763 2764 2765 2766 2767 2768
    return bdrv_co_do_writev(bs, sector_num, nb_sectors, qiov, 0);
}

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

    return bdrv_co_do_writev(bs, sector_num, nb_sectors, NULL,
                             BDRV_REQ_ZERO_WRITE);
2769 2770
}

B
bellard 已提交
2771 2772 2773 2774 2775 2776
/**
 * Truncate file to 'offset' bytes (needed only for file protocols)
 */
int bdrv_truncate(BlockDriverState *bs, int64_t offset)
{
    BlockDriver *drv = bs->drv;
2777
    int ret;
B
bellard 已提交
2778
    if (!drv)
B
bellard 已提交
2779
        return -ENOMEDIUM;
B
bellard 已提交
2780 2781
    if (!drv->bdrv_truncate)
        return -ENOTSUP;
2782 2783
    if (bs->read_only)
        return -EACCES;
M
Marcelo Tosatti 已提交
2784 2785
    if (bdrv_in_use(bs))
        return -EBUSY;
2786 2787 2788
    ret = drv->bdrv_truncate(bs, offset);
    if (ret == 0) {
        ret = refresh_total_sectors(bs, offset >> BDRV_SECTOR_BITS);
2789
        bdrv_dev_resize_cb(bs);
2790 2791
    }
    return ret;
B
bellard 已提交
2792 2793
}

2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812
/**
 * 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
2813 2814 2815 2816 2817 2818 2819
/**
 * 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
2820
        return -ENOMEDIUM;
2821

2822
    if (bdrv_dev_has_removable_media(bs)) {
2823 2824 2825
        if (drv->bdrv_getlength) {
            return drv->bdrv_getlength(bs);
        }
B
bellard 已提交
2826
    }
2827
    return bs->total_sectors * BDRV_SECTOR_SIZE;
B
bellard 已提交
2828 2829
}

B
bellard 已提交
2830
/* return 0 as number of sectors if no device present or error */
2831
void bdrv_get_geometry(BlockDriverState *bs, uint64_t *nb_sectors_ptr)
B
bellard 已提交
2832
{
B
bellard 已提交
2833 2834 2835 2836 2837
    int64_t length;
    length = bdrv_getlength(bs);
    if (length < 0)
        length = 0;
    else
J
Jan Kiszka 已提交
2838
        length = length >> BDRV_SECTOR_BITS;
B
bellard 已提交
2839
    *nb_sectors_ptr = length;
B
bellard 已提交
2840
}
B
bellard 已提交
2841

2842 2843
void bdrv_set_on_error(BlockDriverState *bs, BlockdevOnError on_read_error,
                       BlockdevOnError on_write_error)
2844 2845 2846 2847 2848
{
    bs->on_read_error = on_read_error;
    bs->on_write_error = on_write_error;
}

2849
BlockdevOnError bdrv_get_on_error(BlockDriverState *bs, bool is_read)
2850 2851 2852 2853
{
    return is_read ? bs->on_read_error : bs->on_write_error;
}

2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879
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
2880
    bdrv_emit_qmp_error_event(bs, QEVENT_BLOCK_IO_ERROR, action, is_read);
2881 2882 2883 2884 2885 2886
    if (action == BDRV_ACTION_STOP) {
        vm_stop(RUN_STATE_IO_ERROR);
        bdrv_iostatus_set_err(bs, error);
    }
}

B
bellard 已提交
2887 2888 2889 2890 2891
int bdrv_is_read_only(BlockDriverState *bs)
{
    return bs->read_only;
}

2892 2893 2894 2895 2896
int bdrv_is_sg(BlockDriverState *bs)
{
    return bs->sg;
}

2897 2898 2899 2900 2901
int bdrv_enable_write_cache(BlockDriverState *bs)
{
    return bs->enable_write_cache;
}

2902 2903 2904
void bdrv_set_enable_write_cache(BlockDriverState *bs, bool wce)
{
    bs->enable_write_cache = wce;
2905 2906 2907 2908 2909 2910 2911

    /* so a reopen() will preserve wce */
    if (wce) {
        bs->open_flags |= BDRV_O_CACHE_WB;
    } else {
        bs->open_flags &= ~BDRV_O_CACHE_WB;
    }
2912 2913
}

B
bellard 已提交
2914 2915 2916 2917 2918 2919 2920
int bdrv_is_encrypted(BlockDriverState *bs)
{
    if (bs->backing_hd && bs->backing_hd->encrypted)
        return 1;
    return bs->encrypted;
}

2921 2922 2923 2924 2925 2926 2927 2928 2929
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
2930 2931 2932 2933 2934 2935 2936 2937 2938 2939
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;
    }
2940 2941 2942 2943 2944
    if (!bs->encrypted) {
        return -EINVAL;
    } else if (!bs->drv || !bs->drv->bdrv_set_key) {
        return -ENOMEDIUM;
    }
2945
    ret = bs->drv->bdrv_set_key(bs, key);
2946 2947 2948 2949 2950
    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 */
2951
        bdrv_dev_change_media_cb(bs, true);
2952
    }
2953
    return ret;
B
bellard 已提交
2954 2955
}

2956
const char *bdrv_get_format_name(BlockDriverState *bs)
B
bellard 已提交
2957
{
2958
    return bs->drv ? bs->drv->format_name : NULL;
B
bellard 已提交
2959 2960
}

2961
void bdrv_iterate_format(void (*it)(void *opaque, const char *name),
B
bellard 已提交
2962 2963 2964 2965
                         void *opaque)
{
    BlockDriver *drv;

2966
    QLIST_FOREACH(drv, &bdrv_drivers, list) {
B
bellard 已提交
2967 2968 2969 2970
        it(opaque, drv->format_name);
    }
}

B
bellard 已提交
2971 2972 2973 2974
BlockDriverState *bdrv_find(const char *name)
{
    BlockDriverState *bs;

2975 2976
    QTAILQ_FOREACH(bs, &bdrv_states, list) {
        if (!strcmp(name, bs->device_name)) {
B
bellard 已提交
2977
            return bs;
2978
        }
B
bellard 已提交
2979 2980 2981 2982
    }
    return NULL;
}

M
Markus Armbruster 已提交
2983 2984 2985 2986 2987 2988 2989 2990
BlockDriverState *bdrv_next(BlockDriverState *bs)
{
    if (!bs) {
        return QTAILQ_FIRST(&bdrv_states);
    }
    return QTAILQ_NEXT(bs, list);
}

2991
void bdrv_iterate(void (*it)(void *opaque, BlockDriverState *bs), void *opaque)
B
bellard 已提交
2992 2993 2994
{
    BlockDriverState *bs;

2995
    QTAILQ_FOREACH(bs, &bdrv_states, list) {
2996
        it(opaque, bs);
B
bellard 已提交
2997 2998 2999
    }
}

B
bellard 已提交
3000 3001 3002 3003 3004
const char *bdrv_get_device_name(BlockDriverState *bs)
{
    return bs->device_name;
}

M
Markus Armbruster 已提交
3005 3006 3007 3008 3009
int bdrv_get_flags(BlockDriverState *bs)
{
    return bs->open_flags;
}

3010
int bdrv_flush_all(void)
A
aliguori 已提交
3011 3012
{
    BlockDriverState *bs;
3013
    int result = 0;
A
aliguori 已提交
3014

3015
    QTAILQ_FOREACH(bs, &bdrv_states, list) {
3016 3017 3018 3019
        int ret = bdrv_flush(bs);
        if (ret < 0 && !result) {
            result = ret;
        }
3020
    }
3021 3022

    return result;
A
aliguori 已提交
3023 3024
}

3025 3026 3027 3028 3029
int bdrv_has_zero_init_1(BlockDriverState *bs)
{
    return 1;
}

K
Kevin Wolf 已提交
3030 3031 3032 3033
int bdrv_has_zero_init(BlockDriverState *bs)
{
    assert(bs->drv);

3034 3035 3036 3037 3038
    /* 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
3039 3040
    if (bs->drv->bdrv_has_zero_init) {
        return bs->drv->bdrv_has_zero_init(bs);
K
Kevin Wolf 已提交
3041 3042
    }

3043 3044
    /* safe default */
    return 0;
K
Kevin Wolf 已提交
3045 3046
}

3047
typedef struct BdrvCoGetBlockStatusData {
3048
    BlockDriverState *bs;
3049
    BlockDriverState *base;
3050 3051 3052
    int64_t sector_num;
    int nb_sectors;
    int *pnum;
3053
    int64_t ret;
3054
    bool done;
3055
} BdrvCoGetBlockStatusData;
3056

3057 3058 3059 3060 3061
/*
 * 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.
 *
3062 3063 3064
 * If 'sector_num' is beyond the end of the disk image the return value is 0
 * and 'pnum' is set to 0.
 *
3065 3066 3067 3068
 * '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.
 *
3069 3070
 * '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.
3071
 */
3072 3073 3074
static int64_t coroutine_fn bdrv_co_get_block_status(BlockDriverState *bs,
                                                     int64_t sector_num,
                                                     int nb_sectors, int *pnum)
3075
{
3076
    int64_t length;
3077
    int64_t n;
3078
    int64_t ret, ret2;
3079

3080 3081 3082 3083 3084 3085
    length = bdrv_getlength(bs);
    if (length < 0) {
        return length;
    }

    if (sector_num >= (length >> BDRV_SECTOR_BITS)) {
3086 3087 3088 3089 3090 3091 3092 3093 3094
        *pnum = 0;
        return 0;
    }

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

3095
    if (!bs->drv->bdrv_co_get_block_status) {
3096
        *pnum = nb_sectors;
3097 3098 3099 3100 3101
        ret = BDRV_BLOCK_DATA;
        if (bs->drv->protocol_name) {
            ret |= BDRV_BLOCK_OFFSET_VALID | (sector_num * BDRV_SECTOR_SIZE);
        }
        return ret;
3102
    }
3103

3104 3105 3106 3107 3108
    ret = bs->drv->bdrv_co_get_block_status(bs, sector_num, nb_sectors, pnum);
    if (ret < 0) {
        return ret;
    }

3109 3110 3111 3112 3113 3114 3115 3116 3117 3118
    if (!(ret & BDRV_BLOCK_DATA)) {
        if (bdrv_has_zero_init(bs)) {
            ret |= BDRV_BLOCK_ZERO;
        } else {
            BlockDriverState *bs2 = bs->backing_hd;
            int64_t length2 = bdrv_getlength(bs2);
            if (length2 >= 0 && sector_num >= (length2 >> BDRV_SECTOR_BITS)) {
                ret |= BDRV_BLOCK_ZERO;
            }
        }
3119
    }
3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133

    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);
        }
    }

3134
    return ret;
3135 3136
}

3137 3138
/* Coroutine wrapper for bdrv_get_block_status() */
static void coroutine_fn bdrv_get_block_status_co_entry(void *opaque)
3139
{
3140
    BdrvCoGetBlockStatusData *data = opaque;
3141 3142
    BlockDriverState *bs = data->bs;

3143 3144
    data->ret = bdrv_co_get_block_status(bs, data->sector_num, data->nb_sectors,
                                         data->pnum);
3145 3146 3147 3148
    data->done = true;
}

/*
3149
 * Synchronous wrapper around bdrv_co_get_block_status().
3150
 *
3151
 * See bdrv_co_get_block_status() for details.
3152
 */
3153 3154
int64_t bdrv_get_block_status(BlockDriverState *bs, int64_t sector_num,
                              int nb_sectors, int *pnum)
3155
{
3156
    Coroutine *co;
3157
    BdrvCoGetBlockStatusData data = {
3158 3159 3160 3161 3162 3163 3164
        .bs = bs,
        .sector_num = sector_num,
        .nb_sectors = nb_sectors,
        .pnum = pnum,
        .done = false,
    };

3165 3166
    if (qemu_in_coroutine()) {
        /* Fast-path if already in coroutine context */
3167
        bdrv_get_block_status_co_entry(&data);
3168
    } else {
3169
        co = qemu_coroutine_create(bdrv_get_block_status_co_entry);
3170 3171 3172 3173
        qemu_coroutine_enter(co, &data);
        while (!data.done) {
            qemu_aio_wait();
        }
3174 3175
    }
    return data.ret;
3176 3177
}

3178 3179 3180
int coroutine_fn bdrv_is_allocated(BlockDriverState *bs, int64_t sector_num,
                                   int nb_sectors, int *pnum)
{
3181 3182 3183 3184 3185 3186 3187
    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));
3188 3189
}

3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201
/*
 * 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.
 *
 */
3202 3203 3204 3205
int bdrv_is_allocated_above(BlockDriverState *top,
                            BlockDriverState *base,
                            int64_t sector_num,
                            int nb_sectors, int *pnum)
3206 3207 3208 3209 3210 3211 3212
{
    BlockDriverState *intermediate;
    int ret, n = nb_sectors;

    intermediate = top;
    while (intermediate && intermediate != base) {
        int pnum_inter;
3213 3214
        ret = bdrv_is_allocated(intermediate, sector_num, nb_sectors,
                                &pnum_inter);
3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227
        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.
         */
3228 3229 3230
        if (n > pnum_inter &&
            (intermediate == top ||
             sector_num + pnum_inter < intermediate->total_sectors)) {
3231 3232 3233 3234 3235 3236 3237 3238 3239 3240
            n = pnum_inter;
        }

        intermediate = intermediate->backing_hd;
    }

    *pnum = n;
    return 0;
}

3241 3242 3243 3244 3245 3246 3247 3248 3249 3250
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;
}

3251
void bdrv_get_backing_filename(BlockDriverState *bs,
B
bellard 已提交
3252 3253
                               char *filename, int filename_size)
{
K
Kevin Wolf 已提交
3254
    pstrcpy(filename, filename_size, bs->backing_file);
B
bellard 已提交
3255 3256
}

3257
int bdrv_write_compressed(BlockDriverState *bs, int64_t sector_num,
B
bellard 已提交
3258 3259 3260 3261
                          const uint8_t *buf, int nb_sectors)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
3262
        return -ENOMEDIUM;
B
bellard 已提交
3263 3264
    if (!drv->bdrv_write_compressed)
        return -ENOTSUP;
K
Kevin Wolf 已提交
3265 3266
    if (bdrv_check_request(bs, sector_num, nb_sectors))
        return -EIO;
3267

3268
    assert(!bs->dirty_bitmap);
3269

B
bellard 已提交
3270 3271
    return drv->bdrv_write_compressed(bs, sector_num, buf, nb_sectors);
}
3272

B
bellard 已提交
3273 3274 3275 3276
int bdrv_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
3277
        return -ENOMEDIUM;
B
bellard 已提交
3278 3279 3280 3281 3282 3283
    if (!drv->bdrv_get_info)
        return -ENOTSUP;
    memset(bdi, 0, sizeof(*bdi));
    return drv->bdrv_get_info(bs, bdi);
}

3284 3285
int bdrv_save_vmstate(BlockDriverState *bs, const uint8_t *buf,
                      int64_t pos, int size)
3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297
{
    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)
3298 3299
{
    BlockDriver *drv = bs->drv;
3300 3301

    if (!drv) {
3302
        return -ENOMEDIUM;
3303 3304 3305 3306 3307 3308
    } 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);
    }

3309
    return -ENOTSUP;
3310 3311
}

3312 3313
int bdrv_load_vmstate(BlockDriverState *bs, uint8_t *buf,
                      int64_t pos, int size)
3314 3315 3316 3317
{
    BlockDriver *drv = bs->drv;
    if (!drv)
        return -ENOMEDIUM;
3318 3319 3320 3321 3322
    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;
3323 3324
}

K
Kevin Wolf 已提交
3325 3326
void bdrv_debug_event(BlockDriverState *bs, BlkDebugEvent event)
{
3327
    if (!bs || !bs->drv || !bs->drv->bdrv_debug_event) {
K
Kevin Wolf 已提交
3328 3329 3330
        return;
    }

3331
    bs->drv->bdrv_debug_event(bs, event);
K
Kevin Wolf 已提交
3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352
}

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;
}

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
3353

K
Kevin Wolf 已提交
3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368 3369 3370 3371
    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
3372 3373
}

3374 3375 3376 3377 3378
int bdrv_is_snapshot(BlockDriverState *bs)
{
    return !!(bs->open_flags & BDRV_O_SNAPSHOT);
}

3379 3380 3381 3382
/* 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. */
3383 3384 3385
BlockDriverState *bdrv_find_backing_image(BlockDriverState *bs,
        const char *backing_file)
{
3386 3387 3388 3389 3390 3391 3392 3393
    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) {
3394 3395 3396
        return NULL;
    }

3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411
    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;
            }
3412
        } else {
3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435
            /* 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;
            }
3436 3437 3438
        }
    }

3439 3440 3441 3442
    g_free(filename_full);
    g_free(backing_file_full);
    g_free(filename_tmp);
    return retval;
3443 3444
}

3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457
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);
}

3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473
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
3474
/**************************************************************/
B
bellard 已提交
3475
/* async I/Os */
B
bellard 已提交
3476

3477
BlockDriverAIOCB *bdrv_aio_readv(BlockDriverState *bs, int64_t sector_num,
3478
                                 QEMUIOVector *qiov, int nb_sectors,
3479
                                 BlockDriverCompletionFunc *cb, void *opaque)
B
bellard 已提交
3480
{
3481 3482
    trace_bdrv_aio_readv(bs, sector_num, nb_sectors, opaque);

3483
    return bdrv_co_aio_rw_vector(bs, sector_num, qiov, nb_sectors,
3484
                                 cb, opaque, false);
B
bellard 已提交
3485 3486
}

3487 3488 3489
BlockDriverAIOCB *bdrv_aio_writev(BlockDriverState *bs, int64_t sector_num,
                                  QEMUIOVector *qiov, int nb_sectors,
                                  BlockDriverCompletionFunc *cb, void *opaque)
B
bellard 已提交
3490
{
3491 3492
    trace_bdrv_aio_writev(bs, sector_num, nb_sectors, opaque);

3493
    return bdrv_co_aio_rw_vector(bs, sector_num, qiov, nb_sectors,
3494
                                 cb, opaque, true);
B
bellard 已提交
3495 3496
}

K
Kevin Wolf 已提交
3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514

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);
3515 3516 3517
        if (mcb->callbacks[i].free_qiov) {
            qemu_iovec_destroy(mcb->callbacks[i].free_qiov);
        }
3518
        g_free(mcb->callbacks[i].free_qiov);
K
Kevin Wolf 已提交
3519 3520 3521 3522 3523 3524 3525
    }
}

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

3526 3527
    trace_multiwrite_cb(mcb, ret);

3528
    if (ret < 0 && !mcb->error) {
K
Kevin Wolf 已提交
3529 3530 3531 3532 3533
        mcb->error = ret;
    }

    mcb->num_requests--;
    if (mcb->num_requests == 0) {
3534
        multiwrite_user_cb(mcb);
3535
        g_free(mcb);
K
Kevin Wolf 已提交
3536 3537 3538 3539 3540
    }
}

static int multiwrite_req_compare(const void *a, const void *b)
{
3541 3542 3543 3544 3545 3546 3547 3548 3549 3550 3551 3552 3553
    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
3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574
}

/*
 * 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;

3575
        // Handle exactly sequential writes and overlapping writes.
K
Kevin Wolf 已提交
3576 3577 3578 3579
        if (reqs[i].sector <= oldreq_last) {
            merge = 1;
        }

3580 3581 3582 3583
        if (reqs[outidx].qiov->niov + reqs[i].qiov->niov + 1 > IOV_MAX) {
            merge = 0;
        }

K
Kevin Wolf 已提交
3584 3585
        if (merge) {
            size_t size;
3586
            QEMUIOVector *qiov = g_malloc0(sizeof(*qiov));
K
Kevin Wolf 已提交
3587 3588 3589 3590 3591 3592
            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;
3593
            qemu_iovec_concat(qiov, reqs[outidx].qiov, 0, size);
K
Kevin Wolf 已提交
3594

3595 3596
            // We should need to add any zeros between the two requests
            assert (reqs[i].sector <= oldreq_last);
K
Kevin Wolf 已提交
3597 3598

            // Add the second request
3599
            qemu_iovec_concat(qiov, reqs[i].qiov, 0, reqs[i].qiov->size);
K
Kevin Wolf 已提交
3600

3601
            reqs[outidx].nb_sectors = qiov->size >> 9;
K
Kevin Wolf 已提交
3602 3603 3604 3605 3606 3607 3608 3609 3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631 3632 3633 3634
            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;

3635 3636 3637 3638 3639 3640 3641 3642
    /* 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
3643 3644 3645 3646 3647
    if (num_reqs == 0) {
        return 0;
    }

    // Create MultiwriteCB structure
3648
    mcb = g_malloc0(sizeof(*mcb) + num_reqs * sizeof(*mcb->callbacks));
K
Kevin Wolf 已提交
3649 3650 3651 3652 3653 3654 3655 3656 3657 3658 3659
    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);

3660 3661
    trace_bdrv_aio_multiwrite(mcb, mcb->num_callbacks, num_reqs);

3662 3663
    /* Run the aio requests. */
    mcb->num_requests = num_reqs;
K
Kevin Wolf 已提交
3664
    for (i = 0; i < num_reqs; i++) {
3665
        bdrv_aio_writev(bs, reqs[i].sector, reqs[i].qiov,
K
Kevin Wolf 已提交
3666 3667 3668 3669 3670 3671
            reqs[i].nb_sectors, multiwrite_cb, mcb);
    }

    return 0;
}

B
bellard 已提交
3672 3673
void bdrv_aio_cancel(BlockDriverAIOCB *acb)
{
S
Stefan Hajnoczi 已提交
3674
    acb->aiocb_info->cancel(acb);
B
bellard 已提交
3675 3676 3677 3678 3679
}

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

3680 3681 3682 3683 3684 3685 3686 3687 3688 3689 3690 3691
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)
{
3692 3693
    BlockDriverAIOCBSync *acb =
        container_of(blockacb, BlockDriverAIOCBSync, common);
3694
    qemu_bh_delete(acb->bh);
A
Avi Kivity 已提交
3695
    acb->bh = NULL;
3696 3697 3698
    qemu_aio_release(acb);
}

S
Stefan Hajnoczi 已提交
3699
static const AIOCBInfo bdrv_em_aiocb_info = {
3700 3701 3702 3703
    .aiocb_size         = sizeof(BlockDriverAIOCBSync),
    .cancel             = bdrv_aio_cancel_em,
};

3704
static void bdrv_aio_bh_cb(void *opaque)
B
bellard 已提交
3705
{
3706
    BlockDriverAIOCBSync *acb = opaque;
3707 3708

    if (!acb->is_write)
3709
        qemu_iovec_from_buf(acb->qiov, 0, acb->bounce, acb->qiov->size);
3710
    qemu_vfree(acb->bounce);
3711
    acb->common.cb(acb->common.opaque, acb->ret);
3712
    qemu_bh_delete(acb->bh);
A
Avi Kivity 已提交
3713
    acb->bh = NULL;
3714
    qemu_aio_release(acb);
B
bellard 已提交
3715
}
3716

3717 3718 3719 3720 3721 3722 3723 3724
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
3725
{
3726 3727
    BlockDriverAIOCBSync *acb;

S
Stefan Hajnoczi 已提交
3728
    acb = qemu_aio_get(&bdrv_em_aiocb_info, bs, cb, opaque);
3729 3730
    acb->is_write = is_write;
    acb->qiov = qiov;
3731
    acb->bounce = qemu_blockalign(bs, qiov->size);
3732
    acb->bh = qemu_bh_new(bdrv_aio_bh_cb, acb);
3733 3734

    if (is_write) {
3735
        qemu_iovec_to_buf(acb->qiov, 0, acb->bounce, qiov->size);
3736
        acb->ret = bs->drv->bdrv_write(bs, sector_num, acb->bounce, nb_sectors);
3737
    } else {
3738
        acb->ret = bs->drv->bdrv_read(bs, sector_num, acb->bounce, nb_sectors);
3739 3740
    }

3741
    qemu_bh_schedule(acb->bh);
3742

3743
    return &acb->common;
3744 3745
}

3746 3747
static BlockDriverAIOCB *bdrv_aio_readv_em(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
3748
        BlockDriverCompletionFunc *cb, void *opaque)
3749
{
3750 3751
    return bdrv_aio_rw_vector(bs, sector_num, qiov, nb_sectors, cb, opaque, 0);
}
B
bellard 已提交
3752

3753 3754 3755 3756 3757
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);
3758 3759
}

3760 3761 3762 3763 3764

typedef struct BlockDriverAIOCBCoroutine {
    BlockDriverAIOCB common;
    BlockRequest req;
    bool is_write;
K
Kevin Wolf 已提交
3765
    bool *done;
3766 3767 3768 3769 3770
    QEMUBH* bh;
} BlockDriverAIOCBCoroutine;

static void bdrv_aio_co_cancel_em(BlockDriverAIOCB *blockacb)
{
K
Kevin Wolf 已提交
3771 3772 3773 3774 3775 3776 3777 3778
    BlockDriverAIOCBCoroutine *acb =
        container_of(blockacb, BlockDriverAIOCBCoroutine, common);
    bool done = false;

    acb->done = &done;
    while (!done) {
        qemu_aio_wait();
    }
3779 3780
}

S
Stefan Hajnoczi 已提交
3781
static const AIOCBInfo bdrv_em_co_aiocb_info = {
3782 3783 3784 3785
    .aiocb_size         = sizeof(BlockDriverAIOCBCoroutine),
    .cancel             = bdrv_aio_co_cancel_em,
};

P
Paolo Bonzini 已提交
3786
static void bdrv_co_em_bh(void *opaque)
3787 3788 3789 3790
{
    BlockDriverAIOCBCoroutine *acb = opaque;

    acb->common.cb(acb->common.opaque, acb->req.error);
K
Kevin Wolf 已提交
3791 3792 3793 3794 3795

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

3796 3797 3798 3799
    qemu_bh_delete(acb->bh);
    qemu_aio_release(acb);
}

3800 3801 3802 3803 3804 3805 3806 3807
/* 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,
3808
            acb->req.nb_sectors, acb->req.qiov, 0);
3809 3810
    } else {
        acb->req.error = bdrv_co_do_writev(bs, acb->req.sector,
3811
            acb->req.nb_sectors, acb->req.qiov, 0);
3812 3813
    }

P
Paolo Bonzini 已提交
3814
    acb->bh = qemu_bh_new(bdrv_co_em_bh, acb);
3815 3816 3817
    qemu_bh_schedule(acb->bh);
}

3818 3819 3820 3821 3822 3823
static BlockDriverAIOCB *bdrv_co_aio_rw_vector(BlockDriverState *bs,
                                               int64_t sector_num,
                                               QEMUIOVector *qiov,
                                               int nb_sectors,
                                               BlockDriverCompletionFunc *cb,
                                               void *opaque,
3824
                                               bool is_write)
3825 3826 3827 3828
{
    Coroutine *co;
    BlockDriverAIOCBCoroutine *acb;

S
Stefan Hajnoczi 已提交
3829
    acb = qemu_aio_get(&bdrv_em_co_aiocb_info, bs, cb, opaque);
3830 3831 3832 3833
    acb->req.sector = sector_num;
    acb->req.nb_sectors = nb_sectors;
    acb->req.qiov = qiov;
    acb->is_write = is_write;
K
Kevin Wolf 已提交
3834
    acb->done = NULL;
3835

3836
    co = qemu_coroutine_create(bdrv_co_do_rw);
3837 3838 3839 3840 3841
    qemu_coroutine_enter(co, acb);

    return &acb->common;
}

P
Paolo Bonzini 已提交
3842
static void coroutine_fn bdrv_aio_flush_co_entry(void *opaque)
3843
{
P
Paolo Bonzini 已提交
3844 3845
    BlockDriverAIOCBCoroutine *acb = opaque;
    BlockDriverState *bs = acb->common.bs;
3846

P
Paolo Bonzini 已提交
3847 3848
    acb->req.error = bdrv_co_flush(bs);
    acb->bh = qemu_bh_new(bdrv_co_em_bh, acb);
3849 3850 3851
    qemu_bh_schedule(acb->bh);
}

P
Paolo Bonzini 已提交
3852
BlockDriverAIOCB *bdrv_aio_flush(BlockDriverState *bs,
3853 3854
        BlockDriverCompletionFunc *cb, void *opaque)
{
P
Paolo Bonzini 已提交
3855
    trace_bdrv_aio_flush(bs, opaque);
3856

P
Paolo Bonzini 已提交
3857 3858
    Coroutine *co;
    BlockDriverAIOCBCoroutine *acb;
3859

S
Stefan Hajnoczi 已提交
3860
    acb = qemu_aio_get(&bdrv_em_co_aiocb_info, bs, cb, opaque);
K
Kevin Wolf 已提交
3861 3862
    acb->done = NULL;

P
Paolo Bonzini 已提交
3863 3864
    co = qemu_coroutine_create(bdrv_aio_flush_co_entry);
    qemu_coroutine_enter(co, acb);
3865 3866 3867 3868

    return &acb->common;
}

3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887
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
3888
    acb = qemu_aio_get(&bdrv_em_co_aiocb_info, bs, cb, opaque);
3889 3890
    acb->req.sector = sector_num;
    acb->req.nb_sectors = nb_sectors;
K
Kevin Wolf 已提交
3891
    acb->done = NULL;
3892 3893 3894 3895 3896 3897
    co = qemu_coroutine_create(bdrv_aio_discard_co_entry);
    qemu_coroutine_enter(co, acb);

    return &acb->common;
}

B
bellard 已提交
3898 3899
void bdrv_init(void)
{
3900
    module_call_init(MODULE_INIT_BLOCK);
B
bellard 已提交
3901
}
3902

3903 3904 3905 3906 3907 3908
void bdrv_init_with_whitelist(void)
{
    use_bdrv_whitelist = 1;
    bdrv_init();
}

S
Stefan Hajnoczi 已提交
3909
void *qemu_aio_get(const AIOCBInfo *aiocb_info, BlockDriverState *bs,
3910
                   BlockDriverCompletionFunc *cb, void *opaque)
3911 3912 3913
{
    BlockDriverAIOCB *acb;

S
Stefan Hajnoczi 已提交
3914 3915
    acb = g_slice_alloc(aiocb_info->aiocb_size);
    acb->aiocb_info = aiocb_info;
3916 3917 3918 3919 3920 3921 3922 3923
    acb->bs = bs;
    acb->cb = cb;
    acb->opaque = opaque;
    return acb;
}

void qemu_aio_release(void *p)
{
3924
    BlockDriverAIOCB *acb = p;
S
Stefan Hajnoczi 已提交
3925
    g_slice_free1(acb->aiocb_info->aiocb_size, acb);
3926
}
B
bellard 已提交
3927

3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949 3950 3951 3952 3953
/**************************************************************/
/* 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) {
3954 3955
        acb = bs->drv->bdrv_aio_writev(bs, sector_num, iov, nb_sectors,
                                       bdrv_co_io_em_complete, &co);
3956
    } else {
3957 3958
        acb = bs->drv->bdrv_aio_readv(bs, sector_num, iov, nb_sectors,
                                      bdrv_co_io_em_complete, &co);
3959 3960
    }

3961
    trace_bdrv_co_io_em(bs, sector_num, nb_sectors, is_write, acb);
3962 3963 3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983
    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
3984
static void coroutine_fn bdrv_flush_co_entry(void *opaque)
3985
{
P
Paolo Bonzini 已提交
3986 3987 3988 3989 3990 3991 3992
    RwCo *rwco = opaque;

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

int coroutine_fn bdrv_co_flush(BlockDriverState *bs)
{
K
Kevin Wolf 已提交
3993 3994
    int ret;

3995
    if (!bs || !bdrv_is_inserted(bs) || bdrv_is_read_only(bs)) {
P
Paolo Bonzini 已提交
3996
        return 0;
K
Kevin Wolf 已提交
3997 3998
    }

3999
    /* Write back cached data to the OS even with cache=unsafe */
4000
    BLKDBG_EVENT(bs->file, BLKDBG_FLUSH_TO_OS);
K
Kevin Wolf 已提交
4001 4002 4003 4004 4005 4006 4007
    if (bs->drv->bdrv_co_flush_to_os) {
        ret = bs->drv->bdrv_co_flush_to_os(bs);
        if (ret < 0) {
            return ret;
        }
    }

4008 4009
    /* But don't actually force it to the disk with cache=unsafe */
    if (bs->open_flags & BDRV_O_NO_FLUSH) {
4010
        goto flush_parent;
4011 4012
    }

4013
    BLKDBG_EVENT(bs->file, BLKDBG_FLUSH_TO_DISK);
K
Kevin Wolf 已提交
4014
    if (bs->drv->bdrv_co_flush_to_disk) {
4015
        ret = bs->drv->bdrv_co_flush_to_disk(bs);
P
Paolo Bonzini 已提交
4016 4017 4018 4019 4020 4021 4022 4023
    } 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) {
4024
            ret = -EIO;
P
Paolo Bonzini 已提交
4025 4026
        } else {
            qemu_coroutine_yield();
4027
            ret = co.ret;
P
Paolo Bonzini 已提交
4028 4029 4030 4031 4032 4033 4034 4035 4036 4037 4038 4039 4040
        }
    } 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.
         */
4041
        ret = 0;
P
Paolo Bonzini 已提交
4042
    }
4043 4044 4045 4046 4047 4048 4049
    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.
     */
4050
flush_parent:
4051
    return bdrv_co_flush(bs->file);
P
Paolo Bonzini 已提交
4052 4053
}

4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069
void bdrv_invalidate_cache(BlockDriverState *bs)
{
    if (bs->drv && bs->drv->bdrv_invalidate_cache) {
        bs->drv->bdrv_invalidate_cache(bs);
    }
}

void bdrv_invalidate_cache_all(void)
{
    BlockDriverState *bs;

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

4070 4071 4072 4073 4074 4075 4076 4077 4078
void bdrv_clear_incoming_migration_all(void)
{
    BlockDriverState *bs;

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

P
Paolo Bonzini 已提交
4079 4080 4081 4082 4083 4084
int bdrv_flush(BlockDriverState *bs)
{
    Coroutine *co;
    RwCo rwco = {
        .bs = bs,
        .ret = NOT_DONE,
4085 4086
    };

P
Paolo Bonzini 已提交
4087 4088 4089 4090 4091 4092 4093 4094 4095
    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();
        }
4096
    }
P
Paolo Bonzini 已提交
4097 4098

    return rwco.ret;
4099 4100
}

4101 4102 4103 4104 4105 4106 4107 4108 4109 4110 4111 4112 4113 4114 4115 4116
static void coroutine_fn bdrv_discard_co_entry(void *opaque)
{
    RwCo *rwco = opaque;

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

int coroutine_fn bdrv_co_discard(BlockDriverState *bs, int64_t sector_num,
                                 int nb_sectors)
{
    if (!bs->drv) {
        return -ENOMEDIUM;
    } else if (bdrv_check_request(bs, sector_num, nb_sectors)) {
        return -EIO;
    } else if (bs->read_only) {
        return -EROFS;
4117 4118 4119
    }

    if (bs->dirty_bitmap) {
4120
        bdrv_reset_dirty(bs, sector_num, nb_sectors);
4121 4122
    }

P
Paolo Bonzini 已提交
4123 4124 4125 4126 4127
    /* Do nothing if disabled.  */
    if (!(bs->open_flags & BDRV_O_UNMAP)) {
        return 0;
    }

4128
    if (bs->drv->bdrv_co_discard) {
4129 4130 4131 4132 4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172
        return bs->drv->bdrv_co_discard(bs, sector_num, nb_sectors);
    } else if (bs->drv->bdrv_aio_discard) {
        BlockDriverAIOCB *acb;
        CoroutineIOCompletion co = {
            .coroutine = qemu_coroutine_self(),
        };

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

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

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

    return rwco.ret;
}

B
bellard 已提交
4173 4174 4175 4176 4177 4178 4179 4180 4181
/**************************************************************/
/* removable device support */

/**
 * Return TRUE if the media is present
 */
int bdrv_is_inserted(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;
4182

B
bellard 已提交
4183 4184 4185
    if (!drv)
        return 0;
    if (!drv->bdrv_is_inserted)
4186 4187
        return 1;
    return drv->bdrv_is_inserted(bs);
B
bellard 已提交
4188 4189 4190
}

/**
4191 4192
 * 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
4193 4194 4195 4196 4197
 */
int bdrv_media_changed(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;

4198 4199 4200 4201
    if (drv && drv->bdrv_media_changed) {
        return drv->bdrv_media_changed(bs);
    }
    return -ENOTSUP;
B
bellard 已提交
4202 4203 4204 4205 4206
}

/**
 * If eject_flag is TRUE, eject the media. Otherwise, close the tray
 */
4207
void bdrv_eject(BlockDriverState *bs, bool eject_flag)
B
bellard 已提交
4208 4209 4210
{
    BlockDriver *drv = bs->drv;

4211 4212
    if (drv && drv->bdrv_eject) {
        drv->bdrv_eject(bs, eject_flag);
B
bellard 已提交
4213
    }
4214 4215 4216 4217

    if (bs->device_name[0] != '\0') {
        bdrv_emit_qmp_eject_event(bs, eject_flag);
    }
B
bellard 已提交
4218 4219 4220 4221 4222 4223
}

/**
 * Lock or unlock the media (if it is locked, the user won't be able
 * to eject it manually).
 */
4224
void bdrv_lock_medium(BlockDriverState *bs, bool locked)
B
bellard 已提交
4225 4226 4227
{
    BlockDriver *drv = bs->drv;

4228
    trace_bdrv_lock_medium(bs, locked);
S
Stefan Hajnoczi 已提交
4229

4230 4231
    if (drv && drv->bdrv_lock_medium) {
        drv->bdrv_lock_medium(bs, locked);
B
bellard 已提交
4232 4233
    }
}
4234 4235 4236 4237 4238 4239 4240 4241 4242 4243 4244

/* 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;
}
4245

4246 4247 4248
BlockDriverAIOCB *bdrv_aio_ioctl(BlockDriverState *bs,
        unsigned long int req, void *buf,
        BlockDriverCompletionFunc *cb, void *opaque)
4249
{
4250
    BlockDriver *drv = bs->drv;
4251

4252 4253 4254
    if (drv && drv->bdrv_aio_ioctl)
        return drv->bdrv_aio_ioctl(bs, req, buf, cb, opaque);
    return NULL;
4255
}
4256

4257 4258 4259 4260
void bdrv_set_buffer_alignment(BlockDriverState *bs, int align)
{
    bs->buffer_alignment = align;
}
4261

4262 4263 4264 4265
void *qemu_blockalign(BlockDriverState *bs, size_t size)
{
    return qemu_memalign((bs && bs->buffer_alignment) ? bs->buffer_alignment : 512, size);
}
4266

4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282
/*
 * 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;
}

4283
void bdrv_set_dirty_tracking(BlockDriverState *bs, int granularity)
4284 4285
{
    int64_t bitmap_size;
4286

4287 4288 4289 4290 4291 4292 4293
    assert((granularity & (granularity - 1)) == 0);

    if (granularity) {
        granularity >>= BDRV_SECTOR_BITS;
        assert(!bs->dirty_bitmap);
        bitmap_size = (bdrv_getlength(bs) >> BDRV_SECTOR_BITS);
        bs->dirty_bitmap = hbitmap_alloc(bitmap_size, ffs(granularity) - 1);
4294
    } else {
4295
        if (bs->dirty_bitmap) {
4296
            hbitmap_free(bs->dirty_bitmap);
4297
            bs->dirty_bitmap = NULL;
4298
        }
4299 4300 4301 4302 4303
    }
}

int bdrv_get_dirty(BlockDriverState *bs, int64_t sector)
{
4304 4305
    if (bs->dirty_bitmap) {
        return hbitmap_get(bs->dirty_bitmap, sector);
4306 4307 4308 4309 4310
    } else {
        return 0;
    }
}

4311
void bdrv_dirty_iter_init(BlockDriverState *bs, HBitmapIter *hbi)
4312
{
4313
    hbitmap_iter_init(hbi, bs->dirty_bitmap, 0);
4314 4315 4316 4317 4318
}

void bdrv_set_dirty(BlockDriverState *bs, int64_t cur_sector,
                    int nr_sectors)
{
4319
    hbitmap_set(bs->dirty_bitmap, cur_sector, nr_sectors);
4320 4321
}

4322 4323
void bdrv_reset_dirty(BlockDriverState *bs, int64_t cur_sector,
                      int nr_sectors)
4324
{
4325
    hbitmap_reset(bs->dirty_bitmap, cur_sector, nr_sectors);
4326
}
4327 4328 4329

int64_t bdrv_get_dirty_count(BlockDriverState *bs)
{
4330
    if (bs->dirty_bitmap) {
4331
        return hbitmap_count(bs->dirty_bitmap);
4332 4333 4334
    } else {
        return 0;
    }
4335
}
J
Jes Sorensen 已提交
4336

4337 4338 4339 4340 4341 4342 4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353
/* 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);
    }
}

4354 4355 4356 4357 4358 4359 4360 4361 4362 4363 4364
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;
}

4365 4366
void bdrv_iostatus_enable(BlockDriverState *bs)
{
4367
    bs->iostatus_enabled = true;
4368
    bs->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
4369 4370 4371 4372 4373 4374
}

/* 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)
{
4375
    return (bs->iostatus_enabled &&
4376 4377 4378
           (bs->on_write_error == BLOCKDEV_ON_ERROR_ENOSPC ||
            bs->on_write_error == BLOCKDEV_ON_ERROR_STOP   ||
            bs->on_read_error == BLOCKDEV_ON_ERROR_STOP));
4379 4380 4381 4382
}

void bdrv_iostatus_disable(BlockDriverState *bs)
{
4383
    bs->iostatus_enabled = false;
4384 4385 4386 4387 4388
}

void bdrv_iostatus_reset(BlockDriverState *bs)
{
    if (bdrv_iostatus_is_enabled(bs)) {
4389
        bs->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
4390 4391 4392
        if (bs->job) {
            block_job_iostatus_reset(bs->job);
        }
4393 4394 4395 4396 4397
    }
}

void bdrv_iostatus_set_err(BlockDriverState *bs, int error)
{
4398 4399
    assert(bdrv_iostatus_is_enabled(bs));
    if (bs->iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
4400 4401
        bs->iostatus = error == ENOSPC ? BLOCK_DEVICE_IO_STATUS_NOSPACE :
                                         BLOCK_DEVICE_IO_STATUS_FAILED;
4402 4403 4404
    }
}

4405 4406 4407 4408 4409 4410 4411
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
4412
    cookie->start_time_ns = get_clock();
4413 4414 4415 4416 4417 4418 4419 4420 4421 4422
    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
4423
    bs->total_time_ns[cookie->type] += get_clock() - cookie->start_time_ns;
4424 4425
}

4426 4427
void bdrv_img_create(const char *filename, const char *fmt,
                     const char *base_filename, const char *base_fmt,
4428 4429
                     char *options, uint64_t img_size, int flags,
                     Error **errp, bool quiet)
J
Jes Sorensen 已提交
4430 4431
{
    QEMUOptionParameter *param = NULL, *create_options = NULL;
K
Kevin Wolf 已提交
4432
    QEMUOptionParameter *backing_fmt, *backing_file, *size;
J
Jes Sorensen 已提交
4433 4434
    BlockDriverState *bs = NULL;
    BlockDriver *drv, *proto_drv;
4435
    BlockDriver *backing_drv = NULL;
J
Jes Sorensen 已提交
4436 4437 4438 4439 4440
    int ret = 0;

    /* Find driver and parse its options */
    drv = bdrv_find_format(fmt);
    if (!drv) {
4441
        error_setg(errp, "Unknown file format '%s'", fmt);
4442
        return;
J
Jes Sorensen 已提交
4443 4444
    }

4445
    proto_drv = bdrv_find_protocol(filename, true);
J
Jes Sorensen 已提交
4446
    if (!proto_drv) {
4447
        error_setg(errp, "Unknown protocol '%s'", filename);
4448
        return;
J
Jes Sorensen 已提交
4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463 4464
    }

    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) {
4465
            error_setg(errp, "Invalid options for file format '%s'.", fmt);
J
Jes Sorensen 已提交
4466 4467 4468 4469 4470 4471 4472
            goto out;
        }
    }

    if (base_filename) {
        if (set_option_parameter(param, BLOCK_OPT_BACKING_FILE,
                                 base_filename)) {
4473 4474
            error_setg(errp, "Backing file not supported for file format '%s'",
                       fmt);
J
Jes Sorensen 已提交
4475 4476 4477 4478 4479 4480
            goto out;
        }
    }

    if (base_fmt) {
        if (set_option_parameter(param, BLOCK_OPT_BACKING_FMT, base_fmt)) {
4481 4482
            error_setg(errp, "Backing file format not supported for file "
                             "format '%s'", fmt);
J
Jes Sorensen 已提交
4483 4484 4485 4486
            goto out;
        }
    }

4487 4488 4489
    backing_file = get_option_parameter(param, BLOCK_OPT_BACKING_FILE);
    if (backing_file && backing_file->value.s) {
        if (!strcmp(filename, backing_file->value.s)) {
4490 4491
            error_setg(errp, "Error: Trying to create an image with the "
                             "same filename as the backing file");
4492 4493 4494 4495
            goto out;
        }
    }

J
Jes Sorensen 已提交
4496 4497
    backing_fmt = get_option_parameter(param, BLOCK_OPT_BACKING_FMT);
    if (backing_fmt && backing_fmt->value.s) {
4498 4499
        backing_drv = bdrv_find_format(backing_fmt->value.s);
        if (!backing_drv) {
4500 4501
            error_setg(errp, "Unknown backing file format '%s'",
                       backing_fmt->value.s);
J
Jes Sorensen 已提交
4502 4503 4504 4505 4506 4507
            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
4508 4509
    size = get_option_parameter(param, BLOCK_OPT_SIZE);
    if (size && size->value.n == -1) {
J
Jes Sorensen 已提交
4510 4511 4512
        if (backing_file && backing_file->value.s) {
            uint64_t size;
            char buf[32];
4513 4514 4515 4516 4517
            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
4518 4519 4520

            bs = bdrv_new("");

4521 4522
            ret = bdrv_open(bs, backing_file->value.s, NULL, back_flags,
                            backing_drv);
J
Jes Sorensen 已提交
4523
            if (ret < 0) {
4524 4525
                error_setg_errno(errp, -ret, "Could not open '%s'",
                                 backing_file->value.s);
J
Jes Sorensen 已提交
4526 4527 4528 4529 4530 4531 4532 4533
                goto out;
            }
            bdrv_get_geometry(bs, &size);
            size *= 512;

            snprintf(buf, sizeof(buf), "%" PRId64, size);
            set_option_parameter(param, BLOCK_OPT_SIZE, buf);
        } else {
4534
            error_setg(errp, "Image creation needs a size parameter");
J
Jes Sorensen 已提交
4535 4536 4537 4538
            goto out;
        }
    }

4539 4540 4541 4542 4543
    if (!quiet) {
        printf("Formatting '%s', fmt=%s ", filename, fmt);
        print_option_parameters(param);
        puts("");
    }
J
Jes Sorensen 已提交
4544 4545 4546
    ret = bdrv_create(drv, filename, param);
    if (ret < 0) {
        if (ret == -ENOTSUP) {
4547 4548
            error_setg(errp,"Formatting or formatting option not supported for "
                            "file format '%s'", fmt);
J
Jes Sorensen 已提交
4549
        } else if (ret == -EFBIG) {
4550 4551 4552 4553 4554 4555
            const char *cluster_size_hint = "";
            if (get_option_parameter(create_options, BLOCK_OPT_CLUSTER_SIZE)) {
                cluster_size_hint = " (try using a larger cluster size)";
            }
            error_setg(errp, "The image size is too large for file format '%s'%s",
                       fmt, cluster_size_hint);
J
Jes Sorensen 已提交
4556
        } else {
4557 4558
            error_setg(errp, "%s: error while creating %s: %s", filename, fmt,
                       strerror(-ret));
J
Jes Sorensen 已提交
4559 4560 4561 4562 4563 4564 4565 4566
        }
    }

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

    if (bs) {
F
Fam Zheng 已提交
4567
        bdrv_unref(bs);
J
Jes Sorensen 已提交
4568 4569
    }
}
4570 4571 4572 4573 4574 4575

AioContext *bdrv_get_aio_context(BlockDriverState *bs)
{
    /* Currently BlockDriverState always uses the main loop AioContext */
    return qemu_get_aio_context();
}
4576 4577 4578 4579 4580 4581

void bdrv_add_before_write_notifier(BlockDriverState *bs,
                                    NotifierWithReturn *notifier)
{
    notifier_with_return_list_add(&bs->before_write_notifiers, notifier);
}