block.c 151.1 KB
Newer Older
B
bellard 已提交
1 2
/*
 * QEMU System Emulator block driver
3
 *
B
bellard 已提交
4
 * Copyright (c) 2003 Fabrice Bellard
5
 *
B
bellard 已提交
6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
 * Permission is hereby granted, free of charge, to any person obtaining a copy
 * of this software and associated documentation files (the "Software"), to deal
 * in the Software without restriction, including without limitation the rights
 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
 * copies of the Software, and to permit persons to whom the Software is
 * furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
 * THE SOFTWARE.
 */
24
#include "config-host.h"
P
pbrook 已提交
25
#include "qemu-common.h"
26
#include "trace.h"
27
#include "monitor/monitor.h"
28 29
#include "block/block_int.h"
#include "block/blockjob.h"
30
#include "qemu/module.h"
31
#include "qapi/qmp/qjson.h"
32
#include "sysemu/sysemu.h"
33
#include "qemu/notify.h"
34
#include "block/coroutine.h"
35
#include "block/qapi.h"
L
Luiz Capitulino 已提交
36
#include "qmp-commands.h"
37
#include "qemu/timer.h"
B
bellard 已提交
38

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

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

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

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

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

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

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

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

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

103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122
#ifdef _WIN32
static int is_windows_drive_prefix(const char *filename)
{
    return (((filename[0] >= 'a' && filename[0] <= 'z') ||
             (filename[0] >= 'A' && filename[0] <= 'Z')) &&
            filename[1] == ':');
}

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

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

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

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

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

    bs->io_limits_enabled = false;

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

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

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

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

160 161 162
    bdrv_start_throttled_reqs(bs);

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

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

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

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

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

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

207
    /* the IO will be executed, do the accounting */
208 209
    throttle_account(&bs->throttle_state, is_write, bytes);

Z
Zhi Yong Wu 已提交
210

211 212 213
    /* 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
214 215
    }

216 217
    /* else queue next request for execution */
    qemu_co_queue_next(&bs->throttled_reqs[is_write]);
Z
Zhi Yong Wu 已提交
218 219
}

220 221 222 223 224 225 226 227 228 229
size_t bdrv_opt_mem_align(BlockDriverState *bs)
{
    if (!bs || !bs->drv) {
        /* 4k should be on the safe side */
        return 4096;
    }

    return bs->bl.opt_mem_alignment;
}

230 231 232
/* check if the path starts with "<protocol>:" */
static int path_has_protocol(const char *path)
{
233 234
    const char *p;

235 236 237 238 239
#ifdef _WIN32
    if (is_windows_drive(path) ||
        is_windows_drive_prefix(path)) {
        return 0;
    }
240 241 242
    p = path + strcspn(path, ":/\\");
#else
    p = path + strcspn(path, ":/");
243 244
#endif

245
    return *p == ':';
246 247
}

B
bellard 已提交
248
int path_is_absolute(const char *path)
249
{
B
bellard 已提交
250 251
#ifdef _WIN32
    /* specific case for names like: "\\.\d:" */
P
Paolo Bonzini 已提交
252
    if (is_windows_drive(path) || is_windows_drive_prefix(path)) {
B
bellard 已提交
253
        return 1;
P
Paolo Bonzini 已提交
254 255
    }
    return (*path == '/' || *path == '\\');
256
#else
P
Paolo Bonzini 已提交
257
    return (*path == '/');
258
#endif
259 260
}

B
bellard 已提交
261 262 263 264 265 266
/* 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)
267
{
B
bellard 已提交
268 269 270 271 272 273 274 275 276 277 278 279 280
    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;
281 282 283 284 285 286 287 288 289
        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
290 291 292 293 294 295 296 297 298 299 300 301
        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);
302 303 304
    }
}

305 306 307 308 309 310 311 312 313
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);
    }
}

314
void bdrv_register(BlockDriver *bdrv)
B
bellard 已提交
315
{
316 317
    /* Block drivers without coroutine functions need emulation */
    if (!bdrv->bdrv_co_readv) {
318 319 320
        bdrv->bdrv_co_readv = bdrv_co_readv_em;
        bdrv->bdrv_co_writev = bdrv_co_writev_em;

321 322 323
        /* 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.
         */
324 325 326 327 328
        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
329
    }
330

331
    QLIST_INSERT_HEAD(&bdrv_drivers, bdrv, list);
B
bellard 已提交
332
}
B
bellard 已提交
333 334 335 336

/* create a new block device (by default it is empty) */
BlockDriverState *bdrv_new(const char *device_name)
{
337
    BlockDriverState *bs;
B
bellard 已提交
338

339
    bs = g_malloc0(sizeof(BlockDriverState));
F
Fam Zheng 已提交
340
    QLIST_INIT(&bs->dirty_bitmaps);
B
bellard 已提交
341
    pstrcpy(bs->device_name, sizeof(bs->device_name), device_name);
B
bellard 已提交
342
    if (device_name[0] != '\0') {
343
        QTAILQ_INSERT_TAIL(&bdrv_states, bs, device_list);
B
bellard 已提交
344
    }
345
    bdrv_iostatus_disable(bs);
P
Paolo Bonzini 已提交
346
    notifier_list_init(&bs->close_notifiers);
347
    notifier_with_return_list_init(&bs->before_write_notifiers);
348 349
    qemu_co_queue_init(&bs->throttled_reqs[0]);
    qemu_co_queue_init(&bs->throttled_reqs[1]);
350
    bs->refcnt = 1;
P
Paolo Bonzini 已提交
351

B
bellard 已提交
352 353 354
    return bs;
}

P
Paolo Bonzini 已提交
355 356 357 358 359
void bdrv_add_close_notifier(BlockDriverState *bs, Notifier *notify)
{
    notifier_list_add(&bs->close_notifiers, notify);
}

B
bellard 已提交
360 361 362
BlockDriver *bdrv_find_format(const char *format_name)
{
    BlockDriver *drv1;
363 364
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
        if (!strcmp(drv1->format_name, format_name)) {
B
bellard 已提交
365
            return drv1;
366
        }
B
bellard 已提交
367 368 369 370
    }
    return NULL;
}

371
static int bdrv_is_whitelisted(BlockDriver *drv, bool read_only)
372
{
373 374 375 376 377
    static const char *whitelist_rw[] = {
        CONFIG_BDRV_RW_WHITELIST
    };
    static const char *whitelist_ro[] = {
        CONFIG_BDRV_RO_WHITELIST
378 379 380
    };
    const char **p;

381
    if (!whitelist_rw[0] && !whitelist_ro[0]) {
382
        return 1;               /* no whitelist, anything goes */
383
    }
384

385
    for (p = whitelist_rw; *p; p++) {
386 387 388 389
        if (!strcmp(drv->format_name, *p)) {
            return 1;
        }
    }
390 391 392 393 394 395 396
    if (read_only) {
        for (p = whitelist_ro; *p; p++) {
            if (!strcmp(drv->format_name, *p)) {
                return 1;
            }
        }
    }
397 398 399
    return 0;
}

400 401
BlockDriver *bdrv_find_whitelisted_format(const char *format_name,
                                          bool read_only)
402 403
{
    BlockDriver *drv = bdrv_find_format(format_name);
404
    return drv && bdrv_is_whitelisted(drv, read_only) ? drv : NULL;
405 406
}

407 408 409 410 411
typedef struct CreateCo {
    BlockDriver *drv;
    char *filename;
    QEMUOptionParameter *options;
    int ret;
412
    Error *err;
413 414 415 416
} CreateCo;

static void coroutine_fn bdrv_create_co_entry(void *opaque)
{
417 418 419
    Error *local_err = NULL;
    int ret;

420 421 422
    CreateCo *cco = opaque;
    assert(cco->drv);

423
    ret = cco->drv->bdrv_create(cco->filename, cco->options, &local_err);
424
    if (local_err) {
425 426 427
        error_propagate(&cco->err, local_err);
    }
    cco->ret = ret;
428 429
}

430
int bdrv_create(BlockDriver *drv, const char* filename,
431
    QEMUOptionParameter *options, Error **errp)
B
bellard 已提交
432
{
433 434 435 436 437 438 439 440
    int ret;

    Coroutine *co;
    CreateCo cco = {
        .drv = drv,
        .filename = g_strdup(filename),
        .options = options,
        .ret = NOT_DONE,
441
        .err = NULL,
442 443 444
    };

    if (!drv->bdrv_create) {
445
        error_setg(errp, "Driver '%s' does not support image creation", drv->format_name);
446 447
        ret = -ENOTSUP;
        goto out;
448 449 450 451 452 453 454 455 456 457 458 459 460 461
    }

    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;
462
    if (ret < 0) {
463
        if (cco.err) {
464 465 466 467 468
            error_propagate(errp, cco.err);
        } else {
            error_setg_errno(errp, -ret, "Could not create image");
        }
    }
469

470 471
out:
    g_free(cco.filename);
472
    return ret;
B
bellard 已提交
473 474
}

475 476
int bdrv_create_file(const char* filename, QEMUOptionParameter *options,
                     Error **errp)
477 478
{
    BlockDriver *drv;
479 480
    Error *local_err = NULL;
    int ret;
481

482
    drv = bdrv_find_protocol(filename, true);
483
    if (drv == NULL) {
484
        error_setg(errp, "Could not find protocol for file '%s'", filename);
485
        return -ENOENT;
486 487
    }

488
    ret = bdrv_create(drv, filename, options, &local_err);
489
    if (local_err) {
490 491 492
        error_propagate(errp, local_err);
    }
    return ret;
493 494
}

495
int bdrv_refresh_limits(BlockDriverState *bs)
496 497 498 499 500
{
    BlockDriver *drv = bs->drv;

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

K
Kevin Wolf 已提交
501 502 503 504 505 506 507 508
    if (!drv) {
        return 0;
    }

    /* Take some limits from the children as a default */
    if (bs->file) {
        bdrv_refresh_limits(bs->file);
        bs->bl.opt_transfer_length = bs->file->bl.opt_transfer_length;
509 510 511
        bs->bl.opt_mem_alignment = bs->file->bl.opt_mem_alignment;
    } else {
        bs->bl.opt_mem_alignment = 512;
K
Kevin Wolf 已提交
512 513 514 515 516 517 518
    }

    if (bs->backing_hd) {
        bdrv_refresh_limits(bs->backing_hd);
        bs->bl.opt_transfer_length =
            MAX(bs->bl.opt_transfer_length,
                bs->backing_hd->bl.opt_transfer_length);
519 520 521
        bs->bl.opt_mem_alignment =
            MAX(bs->bl.opt_mem_alignment,
                bs->backing_hd->bl.opt_mem_alignment);
K
Kevin Wolf 已提交
522 523 524 525
    }

    /* Then let the driver override it */
    if (drv->bdrv_refresh_limits) {
526 527 528 529 530 531
        return drv->bdrv_refresh_limits(bs);
    }

    return 0;
}

532 533 534 535 536
/*
 * 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
537
{
538
#ifdef _WIN32
539
    char temp_dir[MAX_PATH];
540 541 542 543 544 545
    /* 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
546
#else
B
bellard 已提交
547
    int fd;
548
    const char *tmpdir;
A
aurel32 已提交
549 550 551
    tmpdir = getenv("TMPDIR");
    if (!tmpdir)
        tmpdir = "/tmp";
552 553 554
    if (snprintf(filename, size, "%s/vl.XXXXXX", tmpdir) >= size) {
        return -EOVERFLOW;
    }
B
bellard 已提交
555
    fd = mkstemp(filename);
556 557 558 559 560
    if (fd < 0) {
        return -errno;
    }
    if (close(fd) != 0) {
        unlink(filename);
561 562 563
        return -errno;
    }
    return 0;
B
bellard 已提交
564
#endif
565
}
B
bellard 已提交
566

567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588
/*
 * 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;
}

589 590
BlockDriver *bdrv_find_protocol(const char *filename,
                                bool allow_protocol_prefix)
B
bellard 已提交
591 592 593
{
    BlockDriver *drv1;
    char protocol[128];
594
    int len;
B
bellard 已提交
595
    const char *p;
B
bellard 已提交
596

597 598
    /* TODO Drivers without bdrv_file_open must be specified explicitly */

599 600 601 602 603 604 605 606 607 608 609 610
    /*
     * 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;
    }

611
    if (!path_has_protocol(filename) || !allow_protocol_prefix) {
612
        return bdrv_find_format("file");
613
    }
614

615 616
    p = strchr(filename, ':');
    assert(p != NULL);
617 618 619 620 621
    len = p - filename;
    if (len > sizeof(protocol) - 1)
        len = sizeof(protocol) - 1;
    memcpy(protocol, filename, len);
    protocol[len] = '\0';
622
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
623
        if (drv1->protocol_name &&
624
            !strcmp(drv1->protocol_name, protocol)) {
B
bellard 已提交
625
            return drv1;
626
        }
B
bellard 已提交
627 628 629 630
    }
    return NULL;
}

631
static int find_image_format(BlockDriverState *bs, const char *filename,
632
                             BlockDriver **pdrv, Error **errp)
633
{
634
    int score, score_max;
635 636
    BlockDriver *drv1, *drv;
    uint8_t buf[2048];
637
    int ret = 0;
638

639
    /* Return the raw BlockDriver * to scsi-generic devices or empty drives */
640
    if (bs->sg || !bdrv_is_inserted(bs) || bdrv_getlength(bs) == 0) {
641 642
        drv = bdrv_find_format("raw");
        if (!drv) {
643
            error_setg(errp, "Could not find raw image format");
644 645 646 647
            ret = -ENOENT;
        }
        *pdrv = drv;
        return ret;
648
    }
649

B
bellard 已提交
650 651
    ret = bdrv_pread(bs, 0, buf, sizeof(buf));
    if (ret < 0) {
652 653
        error_setg_errno(errp, -ret, "Could not read image for determining its "
                         "format");
654 655
        *pdrv = NULL;
        return ret;
B
bellard 已提交
656 657
    }

B
bellard 已提交
658
    score_max = 0;
659
    drv = NULL;
660
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
B
bellard 已提交
661 662 663 664 665 666
        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
667
        }
B
bellard 已提交
668
    }
669
    if (!drv) {
670 671
        error_setg(errp, "Could not determine image format: No compatible "
                   "driver found");
672 673 674 675
        ret = -ENOENT;
    }
    *pdrv = drv;
    return ret;
B
bellard 已提交
676 677
}

678 679 680 681 682 683 684
/**
 * Set the current 'total_sectors' value
 */
static int refresh_total_sectors(BlockDriverState *bs, int64_t hint)
{
    BlockDriver *drv = bs->drv;

685 686 687 688
    /* Do not attempt drv->bdrv_getlength() on scsi-generic devices */
    if (bs->sg)
        return 0;

689 690 691 692 693 694
    /* query actual device if possible, otherwise just trust the hint */
    if (drv->bdrv_getlength) {
        int64_t length = drv->bdrv_getlength(bs);
        if (length < 0) {
            return length;
        }
F
Fam Zheng 已提交
695
        hint = DIV_ROUND_UP(length, BDRV_SECTOR_SIZE);
696 697 698 699 700 701
    }

    bs->total_sectors = hint;
    return 0;
}

P
Paolo Bonzini 已提交
702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721
/**
 * 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;
}

722 723 724 725 726 727 728 729 730 731 732
/**
 * 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;
733 734
    } else if (!strcmp(mode, "directsync")) {
        *flags |= BDRV_O_NOCACHE;
735 736 737 738 739 740 741 742 743 744 745 746 747 748
    } 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;
}

749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764
/**
 * 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
765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784
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;
}

785 786 787 788 789 790 791 792 793 794 795 796 797 798
static int bdrv_assign_node_name(BlockDriverState *bs,
                                 const char *node_name,
                                 Error **errp)
{
    if (!node_name) {
        return 0;
    }

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

799 800 801 802 803 804 805
    /* takes care of avoiding namespaces collisions */
    if (bdrv_find(node_name)) {
        error_setg(errp, "node-name=%s is conflicting with a device id",
                   node_name);
        return -EINVAL;
    }

806 807 808 809 810 811 812 813 814 815 816 817 818
    /* takes care of avoiding duplicates node names */
    if (bdrv_find_node(node_name)) {
        error_setg(errp, "Duplicate node name");
        return -EINVAL;
    }

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

    return 0;
}

819 820
/*
 * Common part for opening disk images and files
821 822
 *
 * Removes all processed options from *options.
823
 */
824
static int bdrv_open_common(BlockDriverState *bs, BlockDriverState *file,
825
    QDict *options, int flags, BlockDriver *drv, Error **errp)
826 827
{
    int ret, open_flags;
K
Kevin Wolf 已提交
828
    const char *filename;
829
    const char *node_name = NULL;
830
    Error *local_err = NULL;
831 832

    assert(drv != NULL);
833
    assert(bs->file == NULL);
834
    assert(options != NULL && bs->options != options);
835

836 837 838 839 840 841
    if (file != NULL) {
        filename = file->filename;
    } else {
        filename = qdict_get_try_str(options, "filename");
    }

842 843 844 845 846 847
    if (drv->bdrv_needs_filename && !filename) {
        error_setg(errp, "The '%s' block driver requires a file name",
                   drv->format_name);
        return -EINVAL;
    }

848
    trace_bdrv_open_common(bs, filename ?: "", flags, drv->format_name);
S
Stefan Hajnoczi 已提交
849

850 851 852 853 854 855 856
    node_name = qdict_get_try_str(options, "node-name");
    ret = bdrv_assign_node_name(bs, node_name, errp);
    if (ret < 0) {
        return ret;
    }
    qdict_del(options, "node-name");

857 858 859 860 861 862 863 864
    /* 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;
    }

865
    bs->open_flags = flags;
866
    bs->guest_block_size = 512;
867
    bs->request_alignment = 512;
A
Asias He 已提交
868
    bs->zero_beyond_eof = true;
869 870 871 872
    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)) {
873 874 875 876 877
        error_setg(errp,
                   !bs->read_only && bdrv_is_whitelisted(drv, true)
                        ? "Driver '%s' can only be used for read-only devices"
                        : "Driver '%s' is not whitelisted",
                   drv->format_name);
878 879
        return -ENOTSUP;
    }
880

881
    assert(bs->copy_on_read == 0); /* bdrv_new() and bdrv_close() make it so */
882 883 884 885 886 887 888
    if (flags & BDRV_O_COPY_ON_READ) {
        if (!bs->read_only) {
            bdrv_enable_copy_on_read(bs);
        } else {
            error_setg(errp, "Can't use copy-on-read on read-only device");
            return -EINVAL;
        }
889 890
    }

891 892 893 894 895
    if (filename != NULL) {
        pstrcpy(bs->filename, sizeof(bs->filename), filename);
    } else {
        bs->filename[0] = '\0';
    }
896 897

    bs->drv = drv;
898
    bs->opaque = g_malloc0(drv->instance_size);
899

900
    bs->enable_write_cache = !!(flags & BDRV_O_CACHE_WB);
901

902 903
    /* Open the image, either directly or using a protocol */
    if (drv->bdrv_file_open) {
904
        assert(file == NULL);
905
        assert(!drv->bdrv_needs_filename || filename != NULL);
906
        ret = drv->bdrv_file_open(bs, options, open_flags, &local_err);
907
    } else {
908
        if (file == NULL) {
909 910
            error_setg(errp, "Can't use '%s' as a block driver for the "
                       "protocol level", drv->format_name);
911 912 913
            ret = -EINVAL;
            goto free_and_fail;
        }
914
        bs->file = file;
915
        ret = drv->bdrv_open(bs, options, open_flags, &local_err);
916 917
    }

918
    if (ret < 0) {
919
        if (local_err) {
920
            error_propagate(errp, local_err);
921 922
        } else if (bs->filename[0]) {
            error_setg_errno(errp, -ret, "Could not open '%s'", bs->filename);
923 924 925
        } else {
            error_setg_errno(errp, -ret, "Could not open image");
        }
926 927 928
        goto free_and_fail;
    }

929 930
    ret = refresh_total_sectors(bs, bs->total_sectors);
    if (ret < 0) {
931
        error_setg_errno(errp, -ret, "Could not refresh total sector count");
932
        goto free_and_fail;
933
    }
934

935
    bdrv_refresh_limits(bs);
936 937
    assert(bdrv_opt_mem_align(bs) != 0);
    assert(bs->request_alignment != 0);
938

939 940
#ifndef _WIN32
    if (bs->is_temporary) {
D
Dunrong Huang 已提交
941 942
        assert(bs->filename[0] != '\0');
        unlink(bs->filename);
943 944 945 946 947
    }
#endif
    return 0;

free_and_fail:
948
    bs->file = NULL;
949
    g_free(bs->opaque);
950 951 952 953 954
    bs->opaque = NULL;
    bs->drv = NULL;
    return ret;
}

K
Kevin Wolf 已提交
955 956
/*
 * Opens a file using a protocol (file, host_device, nbd, ...)
957 958 959 960 961
 *
 * 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
962
 */
963
int bdrv_file_open(BlockDriverState **pbs, const char *filename,
964 965
                   const char *reference, QDict *options, int flags,
                   Error **errp)
B
bellard 已提交
966
{
967
    BlockDriverState *bs = NULL;
968
    BlockDriver *drv;
969
    const char *drvname;
970
    bool allow_protocol_prefix = false;
971
    Error *local_err = NULL;
B
bellard 已提交
972 973
    int ret;

974 975 976 977 978
    /* NULL means an empty set of options */
    if (options == NULL) {
        options = qdict_new();
    }

979 980 981 982 983 984 985 986
    if (reference) {
        if (filename || qdict_size(options)) {
            error_setg(errp, "Cannot reference an existing block device with "
                       "additional options or a new filename");
            return -EINVAL;
        }
        QDECREF(options);

987
        bs = bdrv_lookup_bs(reference, reference, errp);
988 989 990 991 992 993 994 995
        if (!bs) {
            return -ENODEV;
        }
        bdrv_ref(bs);
        *pbs = bs;
        return 0;
    }

B
bellard 已提交
996
    bs = bdrv_new("");
997 998 999
    bs->options = options;
    options = qdict_clone_shallow(options);

K
Kevin Wolf 已提交
1000 1001 1002 1003 1004
    /* 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));
1005
        allow_protocol_prefix = true;
K
Kevin Wolf 已提交
1006
    } else {
1007 1008
        error_setg(errp, "Can't specify 'file' and 'filename' options at the "
                   "same time");
K
Kevin Wolf 已提交
1009 1010 1011 1012
        ret = -EINVAL;
        goto fail;
    }

1013 1014 1015
    /* Find the right block driver */
    drvname = qdict_get_try_str(options, "driver");
    if (drvname) {
1016
        drv = bdrv_find_format(drvname);
1017 1018 1019
        if (!drv) {
            error_setg(errp, "Unknown driver '%s'", drvname);
        }
1020 1021
        qdict_del(options, "driver");
    } else if (filename) {
1022 1023
        drv = bdrv_find_protocol(filename, allow_protocol_prefix);
        if (!drv) {
1024
            error_setg(errp, "Unknown protocol");
1025
        }
1026
    } else {
1027
        error_setg(errp, "Must specify either driver or file");
1028 1029 1030 1031
        drv = NULL;
    }

    if (!drv) {
1032
        /* errp has been set already */
1033 1034 1035 1036 1037 1038
        ret = -ENOENT;
        goto fail;
    }

    /* Parse the filename and open it */
    if (drv->bdrv_parse_filename && filename) {
1039
        drv->bdrv_parse_filename(filename, options, &local_err);
1040
        if (local_err) {
1041
            error_propagate(errp, local_err);
1042 1043 1044
            ret = -EINVAL;
            goto fail;
        }
1045
        qdict_del(options, "filename");
1046 1047
    }

M
Max Reitz 已提交
1048
    if (!drv->bdrv_file_open) {
1049
        ret = bdrv_open(&bs, filename, options, flags, drv, &local_err);
M
Max Reitz 已提交
1050 1051 1052 1053
        options = NULL;
    } else {
        ret = bdrv_open_common(bs, NULL, options, flags, drv, &local_err);
    }
B
bellard 已提交
1054
    if (ret < 0) {
1055
        error_propagate(errp, local_err);
1056 1057 1058 1059
        goto fail;
    }

    /* Check if any unknown options were used */
M
Max Reitz 已提交
1060
    if (options && (qdict_size(options) != 0)) {
1061
        const QDictEntry *entry = qdict_first(options);
1062 1063
        error_setg(errp, "Block protocol '%s' doesn't support the option '%s'",
                   drv->format_name, entry->key);
1064 1065
        ret = -EINVAL;
        goto fail;
1066
    }
1067 1068
    QDECREF(options);

1069
    bs->growable = 1;
B
bellard 已提交
1070 1071
    *pbs = bs;
    return 0;
1072 1073 1074 1075 1076 1077

fail:
    QDECREF(options);
    if (!bs->drv) {
        QDECREF(bs->options);
    }
F
Fam Zheng 已提交
1078
    bdrv_unref(bs);
1079
    return ret;
B
bellard 已提交
1080 1081
}

1082 1083 1084 1085 1086 1087 1088 1089
/*
 * 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.
 */
1090
int bdrv_open_backing_file(BlockDriverState *bs, QDict *options, Error **errp)
P
Paolo Bonzini 已提交
1091 1092 1093 1094
{
    char backing_filename[PATH_MAX];
    int back_flags, ret;
    BlockDriver *back_drv = NULL;
1095
    Error *local_err = NULL;
P
Paolo Bonzini 已提交
1096 1097

    if (bs->backing_hd != NULL) {
1098
        QDECREF(options);
P
Paolo Bonzini 已提交
1099 1100 1101
        return 0;
    }

1102 1103 1104 1105 1106
    /* NULL means an empty set of options */
    if (options == NULL) {
        options = qdict_new();
    }

P
Paolo Bonzini 已提交
1107
    bs->open_flags &= ~BDRV_O_NO_BACKING;
1108 1109 1110
    if (qdict_haskey(options, "file.filename")) {
        backing_filename[0] = '\0';
    } else if (bs->backing_file[0] == '\0' && qdict_size(options) == 0) {
1111
        QDECREF(options);
P
Paolo Bonzini 已提交
1112
        return 0;
F
Fam Zheng 已提交
1113 1114 1115
    } else {
        bdrv_get_full_backing_filename(bs, backing_filename,
                                       sizeof(backing_filename));
P
Paolo Bonzini 已提交
1116 1117 1118 1119 1120 1121 1122
    }

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

    /* backing files always opened read-only */
1123 1124
    back_flags = bs->open_flags & ~(BDRV_O_RDWR | BDRV_O_SNAPSHOT |
                                    BDRV_O_COPY_ON_READ);
P
Paolo Bonzini 已提交
1125

1126 1127
    assert(bs->backing_hd == NULL);
    ret = bdrv_open(&bs->backing_hd,
1128
                    *backing_filename ? backing_filename : NULL, options,
1129
                    back_flags, back_drv, &local_err);
P
Paolo Bonzini 已提交
1130 1131 1132
    if (ret < 0) {
        bs->backing_hd = NULL;
        bs->open_flags |= BDRV_O_NO_BACKING;
1133 1134 1135
        error_setg(errp, "Could not open backing file: %s",
                   error_get_pretty(local_err));
        error_free(local_err);
P
Paolo Bonzini 已提交
1136 1137
        return ret;
    }
P
Peter Feiner 已提交
1138 1139 1140 1141 1142 1143

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

1144 1145 1146
    /* Recalculate the BlockLimits with the backing file */
    bdrv_refresh_limits(bs);

P
Paolo Bonzini 已提交
1147 1148 1149
    return 0;
}

M
Max Reitz 已提交
1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166
/*
 * Opens a disk image whose options are given as BlockdevRef in another block
 * device's options.
 *
 * If force_raw is true, bdrv_file_open() will be used, thereby preventing any
 * image format auto-detection. If it is false and a filename is given,
 * bdrv_open() will be used for auto-detection.
 *
 * If allow_none is true, no image will be opened if filename is false and no
 * BlockdevRef is given. *pbs will remain unchanged and 0 will be returned.
 *
 * bdrev_key specifies the key for the image's BlockdevRef in the options QDict.
 * That QDict has to be flattened; therefore, if the BlockdevRef is a QDict
 * itself, all options starting with "${bdref_key}." are considered part of the
 * BlockdevRef.
 *
 * The BlockdevRef will be removed from the options QDict.
1167 1168
 *
 * To conform with the behavior of bdrv_open(), *pbs has to be NULL.
M
Max Reitz 已提交
1169 1170 1171 1172 1173 1174 1175 1176 1177 1178
 */
int bdrv_open_image(BlockDriverState **pbs, const char *filename,
                    QDict *options, const char *bdref_key, int flags,
                    bool force_raw, bool allow_none, Error **errp)
{
    QDict *image_options;
    int ret;
    char *bdref_key_dot;
    const char *reference;

1179 1180 1181
    assert(pbs);
    assert(*pbs == NULL);

M
Max Reitz 已提交
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
    bdref_key_dot = g_strdup_printf("%s.", bdref_key);
    qdict_extract_subqdict(options, &image_options, bdref_key_dot);
    g_free(bdref_key_dot);

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

    if (filename && !force_raw) {
        /* If a filename is given and the block driver should be detected
           automatically (instead of using none), use bdrv_open() in order to do
           that auto-detection. */
        if (reference) {
            error_setg(errp, "Cannot reference an existing block device while "
                       "giving a filename");
            ret = -EINVAL;
            goto done;
        }

1209
        ret = bdrv_open(pbs, filename, image_options, flags, NULL, errp);
M
Max Reitz 已提交
1210 1211 1212 1213 1214 1215 1216 1217 1218 1219
    } else {
        ret = bdrv_file_open(pbs, filename, reference, image_options, flags,
                             errp);
    }

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

K
Kevin Wolf 已提交
1220 1221
/*
 * Opens a disk image (raw, qcow2, vmdk, ...)
1222 1223 1224 1225 1226
 *
 * 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.
1227 1228 1229
 *
 * If *pbs is NULL, a new BDS will be created with a pointer to it stored there.
 * If it is not NULL, the referenced BDS will be reused.
K
Kevin Wolf 已提交
1230
 */
1231
int bdrv_open(BlockDriverState **pbs, const char *filename, QDict *options,
1232
              int flags, BlockDriver *drv, Error **errp)
B
bellard 已提交
1233
{
K
Kevin Wolf 已提交
1234
    int ret;
1235 1236
    /* TODO: extra byte is a hack to ensure MAX_PATH space on Windows. */
    char tmp_filename[PATH_MAX + 1];
1237
    BlockDriverState *file = NULL, *bs;
1238
    const char *drvname;
1239
    Error *local_err = NULL;
B
bellard 已提交
1240

1241 1242 1243 1244 1245 1246 1247 1248
    assert(pbs);

    if (*pbs) {
        bs = *pbs;
    } else {
        bs = bdrv_new("");
    }

1249 1250 1251 1252 1253 1254
    /* NULL means an empty set of options */
    if (options == NULL) {
        options = qdict_new();
    }

    bs->options = options;
1255
    options = qdict_clone_shallow(options);
1256 1257

    /* For snapshot=on, create a temporary qcow2 overlay */
B
bellard 已提交
1258
    if (flags & BDRV_O_SNAPSHOT) {
B
bellard 已提交
1259 1260
        BlockDriverState *bs1;
        int64_t total_size;
K
Kevin Wolf 已提交
1261
        BlockDriver *bdrv_qcow2;
1262
        QEMUOptionParameter *create_options;
1263
        QDict *snapshot_options;
1264

B
bellard 已提交
1265 1266
        /* if snapshot, we create a temporary backing file and open it
           instead of opening 'filename' directly */
1267

1268 1269
        /* Get the required size from the image */
        QINCREF(options);
1270 1271
        bs1 = NULL;
        ret = bdrv_open(&bs1, filename, options, BDRV_O_NO_BACKING,
1272
                        drv, &local_err);
1273
        if (ret < 0) {
1274
            goto fail;
B
bellard 已提交
1275
        }
1276
        total_size = bdrv_getlength(bs1) & BDRV_SECTOR_MASK;
A
aliguori 已提交
1277

F
Fam Zheng 已提交
1278
        bdrv_unref(bs1);
1279

1280
        /* Create the temporary image */
1281 1282
        ret = get_tmp_filename(tmp_filename, sizeof(tmp_filename));
        if (ret < 0) {
1283
            error_setg_errno(errp, -ret, "Could not get temporary filename");
1284
            goto fail;
1285
        }
A
aliguori 已提交
1286

K
Kevin Wolf 已提交
1287
        bdrv_qcow2 = bdrv_find_format("qcow2");
1288 1289
        create_options = parse_option_parameters("", bdrv_qcow2->create_options,
                                                 NULL);
K
Kevin Wolf 已提交
1290

1291
        set_option_parameter_int(create_options, BLOCK_OPT_SIZE, total_size);
K
Kevin Wolf 已提交
1292

1293
        ret = bdrv_create(bdrv_qcow2, tmp_filename, create_options, &local_err);
1294
        free_option_parameters(create_options);
1295
        if (ret < 0) {
1296
            error_setg_errno(errp, -ret, "Could not create temporary overlay "
1297 1298 1299 1300
                             "'%s': %s", tmp_filename,
                             error_get_pretty(local_err));
            error_free(local_err);
            local_err = NULL;
1301
            goto fail;
B
bellard 已提交
1302
        }
K
Kevin Wolf 已提交
1303

1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319
        /* Prepare a new options QDict for the temporary file, where user
         * options refer to the backing file */
        if (filename) {
            qdict_put(options, "file.filename", qstring_from_str(filename));
        }
        if (drv) {
            qdict_put(options, "driver", qstring_from_str(drv->format_name));
        }

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

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

B
bellard 已提交
1320
        filename = tmp_filename;
K
Kevin Wolf 已提交
1321
        drv = bdrv_qcow2;
B
bellard 已提交
1322 1323
        bs->is_temporary = 1;
    }
B
bellard 已提交
1324

1325 1326 1327 1328 1329
    /* Open image file without format layer */
    if (flags & BDRV_O_RDWR) {
        flags |= BDRV_O_ALLOW_RDWR;
    }

1330
    assert(file == NULL);
1331 1332 1333 1334 1335
    ret = bdrv_open_image(&file, filename, options, "file",
                          bdrv_open_flags(bs, flags | BDRV_O_UNMAP), true, true,
                          &local_err);
    if (ret < 0) {
        goto fail;
1336 1337
    }

K
Kevin Wolf 已提交
1338
    /* Find the right image format driver */
1339 1340
    drvname = qdict_get_try_str(options, "driver");
    if (drvname) {
1341
        drv = bdrv_find_format(drvname);
1342
        qdict_del(options, "driver");
1343 1344 1345 1346 1347
        if (!drv) {
            error_setg(errp, "Invalid driver: '%s'", drvname);
            ret = -EINVAL;
            goto unlink_and_fail;
        }
1348 1349
    }

1350
    if (!drv) {
1351 1352 1353 1354 1355 1356 1357
        if (file) {
            ret = find_image_format(file, filename, &drv, &local_err);
        } else {
            error_setg(errp, "Must specify either driver or file");
            ret = -EINVAL;
            goto unlink_and_fail;
        }
1358
    }
1359

1360 1361
    if (!drv) {
        goto unlink_and_fail;
B
bellard 已提交
1362
    }
K
Kevin Wolf 已提交
1363 1364

    /* Open the image */
1365
    ret = bdrv_open_common(bs, file, options, flags, drv, &local_err);
K
Kevin Wolf 已提交
1366
    if (ret < 0) {
1367 1368 1369
        goto unlink_and_fail;
    }

1370
    if (file && (bs->file != file)) {
F
Fam Zheng 已提交
1371
        bdrv_unref(file);
1372 1373 1374
        file = NULL;
    }

K
Kevin Wolf 已提交
1375
    /* If there is a backing file, use it */
P
Paolo Bonzini 已提交
1376
    if ((flags & BDRV_O_NO_BACKING) == 0) {
1377 1378
        QDict *backing_options;

1379
        qdict_extract_subqdict(options, &backing_options, "backing.");
1380
        ret = bdrv_open_backing_file(bs, backing_options, &local_err);
K
Kevin Wolf 已提交
1381
        if (ret < 0) {
1382
            goto close_and_fail;
K
Kevin Wolf 已提交
1383 1384 1385
        }
    }

1386 1387 1388
    /* Check if any unknown options were used */
    if (qdict_size(options) != 0) {
        const QDictEntry *entry = qdict_first(options);
1389 1390 1391
        error_setg(errp, "Block format '%s' used by device '%s' doesn't "
                   "support the option '%s'", drv->format_name, bs->device_name,
                   entry->key);
1392 1393 1394 1395 1396 1397

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

K
Kevin Wolf 已提交
1398
    if (!bdrv_key_required(bs)) {
1399
        bdrv_dev_change_media_cb(bs, true);
K
Kevin Wolf 已提交
1400 1401
    }

1402
    *pbs = bs;
K
Kevin Wolf 已提交
1403 1404 1405
    return 0;

unlink_and_fail:
1406
    if (file != NULL) {
F
Fam Zheng 已提交
1407
        bdrv_unref(file);
1408
    }
K
Kevin Wolf 已提交
1409 1410 1411
    if (bs->is_temporary) {
        unlink(filename);
    }
1412 1413
fail:
    QDECREF(bs->options);
1414
    QDECREF(options);
1415
    bs->options = NULL;
1416 1417 1418 1419 1420 1421
    if (!*pbs) {
        /* If *pbs is NULL, a new BDS has been created in this function and
           needs to be freed now. Otherwise, it does not need to be closed,
           since it has not really been opened yet. */
        bdrv_unref(bs);
    }
1422
    if (local_err) {
1423 1424
        error_propagate(errp, local_err);
    }
1425
    return ret;
1426

1427
close_and_fail:
1428 1429 1430 1431 1432 1433
    /* See fail path, but now the BDS has to be always closed */
    if (*pbs) {
        bdrv_close(bs);
    } else {
        bdrv_unref(bs);
    }
1434
    QDECREF(options);
1435
    if (local_err) {
1436 1437
        error_propagate(errp, local_err);
    }
K
Kevin Wolf 已提交
1438 1439 1440
    return ret;
}

1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608
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 {
1609 1610
                error_setg(errp, "failed while preparing to reopen image '%s'",
                           reopen_state->bs->filename);
1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 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
            }
            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);
1653 1654

    bdrv_refresh_limits(reopen_state->bs);
1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674
}

/*
 * 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
1675 1676
void bdrv_close(BlockDriverState *bs)
{
1677 1678 1679
    if (bs->job) {
        block_job_cancel_sync(bs->job);
    }
1680 1681 1682
    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
1683
    notifier_list_notify(&bs->close_notifiers, bs);
K
Kevin Wolf 已提交
1684

1685
    if (bs->drv) {
1686
        if (bs->backing_hd) {
F
Fam Zheng 已提交
1687
            bdrv_unref(bs->backing_hd);
1688 1689
            bs->backing_hd = NULL;
        }
B
bellard 已提交
1690
        bs->drv->bdrv_close(bs);
1691
        g_free(bs->opaque);
B
bellard 已提交
1692 1693 1694 1695
#ifdef _WIN32
        if (bs->is_temporary) {
            unlink(bs->filename);
        }
B
bellard 已提交
1696
#endif
B
bellard 已提交
1697 1698
        bs->opaque = NULL;
        bs->drv = NULL;
1699
        bs->copy_on_read = 0;
1700 1701
        bs->backing_file[0] = '\0';
        bs->backing_format[0] = '\0';
1702 1703 1704 1705 1706
        bs->total_sectors = 0;
        bs->encrypted = 0;
        bs->valid_key = 0;
        bs->sg = 0;
        bs->growable = 0;
A
Asias He 已提交
1707
        bs->zero_beyond_eof = false;
1708 1709
        QDECREF(bs->options);
        bs->options = NULL;
B
bellard 已提交
1710

1711
        if (bs->file != NULL) {
F
Fam Zheng 已提交
1712
            bdrv_unref(bs->file);
1713
            bs->file = NULL;
1714
        }
B
bellard 已提交
1715
    }
Z
Zhi Yong Wu 已提交
1716

P
Pavel Hrdina 已提交
1717 1718
    bdrv_dev_change_media_cb(bs, false);

Z
Zhi Yong Wu 已提交
1719 1720 1721 1722
    /*throttling disk I/O limits*/
    if (bs->io_limits_enabled) {
        bdrv_io_limits_disable(bs);
    }
B
bellard 已提交
1723 1724
}

1725 1726 1727 1728
void bdrv_close_all(void)
{
    BlockDriverState *bs;

1729
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
1730 1731 1732 1733
        bdrv_close(bs);
    }
}

1734 1735 1736 1737 1738 1739
/* 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;
    }
1740 1741 1742 1743
    if (!qemu_co_queue_empty(&bs->throttled_reqs[0])) {
        return true;
    }
    if (!qemu_co_queue_empty(&bs->throttled_reqs[1])) {
1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757
        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;
1758
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
1759 1760 1761 1762 1763 1764 1765
        if (bdrv_requests_pending(bs)) {
            return true;
        }
    }
    return false;
}

1766 1767 1768 1769 1770
/*
 * 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.
1771 1772 1773 1774 1775 1776
 *
 * 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.
1777 1778 1779
 */
void bdrv_drain_all(void)
{
1780 1781
    /* Always run first iteration so any pending completion BHs run */
    bool busy = true;
1782 1783
    BlockDriverState *bs;

1784
    while (busy) {
1785
        QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
1786
            bdrv_start_throttled_reqs(bs);
1787
        }
1788

1789 1790
        busy = bdrv_requests_pending_all();
        busy |= aio_poll(qemu_get_aio_context(), busy);
1791 1792 1793
    }
}

1794 1795
/* make a BlockDriverState anonymous by removing from bdrv_state and
 * graph_bdrv_state list.
1796 1797 1798 1799
   Also, NULL terminate the device_name to prevent double remove */
void bdrv_make_anon(BlockDriverState *bs)
{
    if (bs->device_name[0] != '\0') {
1800
        QTAILQ_REMOVE(&bdrv_states, bs, device_list);
1801 1802
    }
    bs->device_name[0] = '\0';
1803 1804 1805 1806
    if (bs->node_name[0] != '\0') {
        QTAILQ_REMOVE(&graph_bdrv_states, bs, node_list);
    }
    bs->node_name[0] = '\0';
1807 1808
}

P
Paolo Bonzini 已提交
1809 1810 1811 1812 1813 1814 1815
static void bdrv_rebind(BlockDriverState *bs)
{
    if (bs->drv && bs->drv->bdrv_rebind) {
        bs->drv->bdrv_rebind(bs);
    }
}

1816 1817
static void bdrv_move_feature_fields(BlockDriverState *bs_dest,
                                     BlockDriverState *bs_src)
1818
{
1819 1820
    /* move some fields that need to stay attached to the device */
    bs_dest->open_flags         = bs_src->open_flags;
1821 1822

    /* dev info */
1823 1824 1825
    bs_dest->dev_ops            = bs_src->dev_ops;
    bs_dest->dev_opaque         = bs_src->dev_opaque;
    bs_dest->dev                = bs_src->dev;
1826
    bs_dest->guest_block_size   = bs_src->guest_block_size;
1827
    bs_dest->copy_on_read       = bs_src->copy_on_read;
1828

1829
    bs_dest->enable_write_cache = bs_src->enable_write_cache;
1830

1831 1832 1833 1834 1835 1836
    /* 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];
1837
    bs_dest->io_limits_enabled  = bs_src->io_limits_enabled;
1838 1839

    /* r/w error */
1840 1841
    bs_dest->on_read_error      = bs_src->on_read_error;
    bs_dest->on_write_error     = bs_src->on_write_error;
1842 1843

    /* i/o status */
1844 1845
    bs_dest->iostatus_enabled   = bs_src->iostatus_enabled;
    bs_dest->iostatus           = bs_src->iostatus;
1846

1847
    /* dirty bitmap */
F
Fam Zheng 已提交
1848
    bs_dest->dirty_bitmaps      = bs_src->dirty_bitmaps;
1849

1850 1851 1852
    /* reference count */
    bs_dest->refcnt             = bs_src->refcnt;

1853
    /* job */
1854 1855
    bs_dest->in_use             = bs_src->in_use;
    bs_dest->job                = bs_src->job;
1856

1857
    /* keep the same entry in bdrv_states */
1858 1859
    pstrcpy(bs_dest->device_name, sizeof(bs_dest->device_name),
            bs_src->device_name);
1860 1861 1862 1863 1864 1865
    bs_dest->device_list = bs_src->device_list;

    /* keep the same entry in graph_bdrv_states
     * We do want to swap name but don't want to swap linked list entries
     */
    bs_dest->node_list   = bs_src->node_list;
1866
}
1867

1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882
/*
 * 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
1883

1884 1885
    /* bs_new must be anonymous and shouldn't have anything fancy enabled */
    assert(bs_new->device_name[0] == '\0');
F
Fam Zheng 已提交
1886
    assert(QLIST_EMPTY(&bs_new->dirty_bitmaps));
1887 1888 1889 1890
    assert(bs_new->job == NULL);
    assert(bs_new->dev == NULL);
    assert(bs_new->in_use == 0);
    assert(bs_new->io_limits_enabled == false);
1891
    assert(!throttle_have_timer(&bs_new->throttle_state));
1892

1893 1894 1895
    tmp = *bs_new;
    *bs_new = *bs_old;
    *bs_old = tmp;
1896

1897 1898 1899 1900
    /* 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);
1901

1902 1903 1904 1905 1906 1907 1908 1909
    /* 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);
1910
    assert(!throttle_have_timer(&bs_new->throttle_state));
P
Paolo Bonzini 已提交
1911 1912

    bdrv_rebind(bs_new);
1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938
    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 : "");
1939 1940
}

F
Fam Zheng 已提交
1941
static void bdrv_delete(BlockDriverState *bs)
B
bellard 已提交
1942
{
1943
    assert(!bs->dev);
1944 1945
    assert(!bs->job);
    assert(!bs->in_use);
F
Fam Zheng 已提交
1946
    assert(!bs->refcnt);
F
Fam Zheng 已提交
1947
    assert(QLIST_EMPTY(&bs->dirty_bitmaps));
1948

1949 1950
    bdrv_close(bs);

1951
    /* remove from list, if necessary */
1952
    bdrv_make_anon(bs);
1953

1954
    g_free(bs);
B
bellard 已提交
1955 1956
}

1957 1958
int bdrv_attach_dev(BlockDriverState *bs, void *dev)
/* TODO change to DeviceState *dev when all users are qdevified */
1959
{
1960
    if (bs->dev) {
1961 1962
        return -EBUSY;
    }
1963
    bs->dev = dev;
1964
    bdrv_iostatus_reset(bs);
1965 1966 1967
    return 0;
}

1968 1969
/* TODO qdevified devices don't use this, remove when devices are qdevified */
void bdrv_attach_dev_nofail(BlockDriverState *bs, void *dev)
1970
{
1971 1972 1973 1974 1975 1976 1977 1978 1979 1980
    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;
1981 1982
    bs->dev_ops = NULL;
    bs->dev_opaque = NULL;
1983
    bs->guest_block_size = 512;
1984 1985
}

1986 1987
/* TODO change to return DeviceState * when all users are qdevified */
void *bdrv_get_attached_dev(BlockDriverState *bs)
1988
{
1989
    return bs->dev;
1990 1991
}

1992 1993 1994 1995 1996 1997 1998
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
1999 2000 2001
void bdrv_emit_qmp_error_event(const BlockDriverState *bdrv,
                               enum MonitorEvent ev,
                               BlockErrorAction action, bool is_read)
2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023
{
    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
2024
    monitor_protocol_event(ev, data);
2025 2026 2027 2028

    qobject_decref(data);
}

2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039
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);
}

2040
static void bdrv_dev_change_media_cb(BlockDriverState *bs, bool load)
2041
{
2042
    if (bs->dev_ops && bs->dev_ops->change_media_cb) {
2043
        bool tray_was_closed = !bdrv_dev_is_tray_open(bs);
2044
        bs->dev_ops->change_media_cb(bs->dev_opaque, load);
2045 2046 2047 2048 2049 2050 2051 2052
        if (tray_was_closed) {
            /* tray open */
            bdrv_emit_qmp_eject_event(bs, true);
        }
        if (load) {
            /* tray close */
            bdrv_emit_qmp_eject_event(bs, false);
        }
2053 2054 2055
    }
}

2056 2057 2058 2059 2060
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
2061 2062 2063 2064 2065 2066 2067
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);
    }
}

2068 2069 2070 2071 2072 2073 2074 2075
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;
}

2076 2077 2078 2079
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);
2080 2081 2082
    }
}

2083 2084 2085 2086 2087 2088 2089 2090
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
2091 2092 2093
/*
 * Run consistency checks on an image
 *
2094
 * Returns 0 if the check could be completed (it doesn't mean that the image is
2095
 * free of errors) or -errno when an internal error occurred. The results of the
2096
 * check are stored in res.
A
aliguori 已提交
2097
 */
2098
int bdrv_check(BlockDriverState *bs, BdrvCheckResult *res, BdrvCheckMode fix)
A
aliguori 已提交
2099 2100 2101 2102 2103
{
    if (bs->drv->bdrv_check == NULL) {
        return -ENOTSUP;
    }

2104
    memset(res, 0, sizeof(*res));
2105
    return bs->drv->bdrv_check(bs, res, fix);
A
aliguori 已提交
2106 2107
}

2108 2109
#define COMMIT_BUF_SECTORS 2048

2110 2111 2112
/* commit COW file into the raw image */
int bdrv_commit(BlockDriverState *bs)
{
B
bellard 已提交
2113
    BlockDriver *drv = bs->drv;
2114
    int64_t sector, total_sectors, length, backing_length;
2115
    int n, ro, open_flags;
2116
    int ret = 0;
2117
    uint8_t *buf = NULL;
2118
    char filename[PATH_MAX];
2119

B
bellard 已提交
2120 2121
    if (!drv)
        return -ENOMEDIUM;
2122 2123 2124
    
    if (!bs->backing_hd) {
        return -ENOTSUP;
2125 2126
    }

2127 2128 2129 2130
    if (bdrv_in_use(bs) || bdrv_in_use(bs->backing_hd)) {
        return -EBUSY;
    }

2131
    ro = bs->backing_hd->read_only;
2132 2133
    /* Use pstrcpy (not strncpy): filename must be NUL-terminated. */
    pstrcpy(filename, sizeof(filename), bs->backing_hd->filename);
2134 2135 2136
    open_flags =  bs->backing_hd->open_flags;

    if (ro) {
2137 2138
        if (bdrv_reopen(bs->backing_hd, open_flags | BDRV_O_RDWR, NULL)) {
            return -EACCES;
2139
        }
B
bellard 已提交
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
    length = bdrv_getlength(bs);
    if (length < 0) {
        ret = length;
        goto ro_cleanup;
    }

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

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

    total_sectors = length >> BDRV_SECTOR_BITS;
2165
    buf = g_malloc(COMMIT_BUF_SECTORS * BDRV_SECTOR_SIZE);
2166 2167

    for (sector = 0; sector < total_sectors; sector += n) {
2168 2169 2170 2171 2172
        ret = bdrv_is_allocated(bs, sector, COMMIT_BUF_SECTORS, &n);
        if (ret < 0) {
            goto ro_cleanup;
        }
        if (ret) {
K
Kevin Wolf 已提交
2173 2174
            ret = bdrv_read(bs, sector, buf, n);
            if (ret < 0) {
2175 2176 2177
                goto ro_cleanup;
            }

K
Kevin Wolf 已提交
2178 2179
            ret = bdrv_write(bs->backing_hd, sector, buf, n);
            if (ret < 0) {
2180 2181
                goto ro_cleanup;
            }
B
bellard 已提交
2182
        }
2183
    }
2184

2185 2186
    if (drv->bdrv_make_empty) {
        ret = drv->bdrv_make_empty(bs);
K
Kevin Wolf 已提交
2187 2188 2189
        if (ret < 0) {
            goto ro_cleanup;
        }
2190 2191
        bdrv_flush(bs);
    }
2192

2193 2194 2195 2196
    /*
     * Make sure all data we wrote to the backing device is actually
     * stable on disk.
     */
K
Kevin Wolf 已提交
2197
    if (bs->backing_hd) {
2198
        bdrv_flush(bs->backing_hd);
K
Kevin Wolf 已提交
2199
    }
2200

K
Kevin Wolf 已提交
2201
    ret = 0;
2202
ro_cleanup:
2203
    g_free(buf);
2204 2205

    if (ro) {
2206 2207
        /* ignoring error return here */
        bdrv_reopen(bs->backing_hd, open_flags & ~BDRV_O_RDWR, NULL);
2208 2209
    }

2210
    return ret;
2211 2212
}

2213
int bdrv_commit_all(void)
2214 2215 2216
{
    BlockDriverState *bs;

2217
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
2218 2219 2220 2221 2222
        if (bs->drv && bs->backing_hd) {
            int ret = bdrv_commit(bs);
            if (ret < 0) {
                return ret;
            }
2223
        }
2224
    }
2225
    return 0;
2226 2227
}

S
Stefan Hajnoczi 已提交
2228 2229 2230 2231 2232 2233 2234
/**
 * 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)
{
2235 2236 2237 2238
    if (req->serialising) {
        req->bs->serialising_in_flight--;
    }

S
Stefan Hajnoczi 已提交
2239
    QLIST_REMOVE(req, list);
2240
    qemu_co_queue_restart_all(&req->wait_queue);
S
Stefan Hajnoczi 已提交
2241 2242 2243 2244 2245 2246 2247
}

/**
 * Add an active request to the tracked requests list
 */
static void tracked_request_begin(BdrvTrackedRequest *req,
                                  BlockDriverState *bs,
2248 2249
                                  int64_t offset,
                                  unsigned int bytes, bool is_write)
S
Stefan Hajnoczi 已提交
2250 2251 2252
{
    *req = (BdrvTrackedRequest){
        .bs = bs,
2253 2254 2255 2256 2257
        .offset         = offset,
        .bytes          = bytes,
        .is_write       = is_write,
        .co             = qemu_coroutine_self(),
        .serialising    = false,
2258 2259
        .overlap_offset = offset,
        .overlap_bytes  = bytes,
S
Stefan Hajnoczi 已提交
2260 2261
    };

2262 2263
    qemu_co_queue_init(&req->wait_queue);

S
Stefan Hajnoczi 已提交
2264 2265 2266
    QLIST_INSERT_HEAD(&bs->tracked_requests, req, list);
}

2267
static void mark_request_serialising(BdrvTrackedRequest *req, uint64_t align)
2268
{
2269
    int64_t overlap_offset = req->offset & ~(align - 1);
2270 2271
    unsigned int overlap_bytes = ROUND_UP(req->offset + req->bytes, align)
                               - overlap_offset;
2272

2273 2274 2275 2276
    if (!req->serialising) {
        req->bs->serialising_in_flight++;
        req->serialising = true;
    }
2277 2278 2279

    req->overlap_offset = MIN(req->overlap_offset, overlap_offset);
    req->overlap_bytes = MAX(req->overlap_bytes, overlap_bytes);
2280 2281
}

S
Stefan Hajnoczi 已提交
2282 2283 2284
/**
 * Round a region to cluster boundaries
 */
2285 2286 2287 2288
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
2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302
{
    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);
    }
}

2303
static int bdrv_get_cluster_size(BlockDriverState *bs)
2304 2305
{
    BlockDriverInfo bdi;
2306
    int ret;
2307

2308 2309 2310
    ret = bdrv_get_info(bs, &bdi);
    if (ret < 0 || bdi.cluster_size == 0) {
        return bs->request_alignment;
2311
    } else {
2312
        return bdi.cluster_size;
2313 2314 2315
    }
}

2316
static bool tracked_request_overlaps(BdrvTrackedRequest *req,
2317 2318
                                     int64_t offset, unsigned int bytes)
{
S
Stefan Hajnoczi 已提交
2319
    /*        aaaa   bbbb */
2320
    if (offset >= req->overlap_offset + req->overlap_bytes) {
S
Stefan Hajnoczi 已提交
2321 2322 2323
        return false;
    }
    /* bbbb   aaaa        */
2324
    if (req->overlap_offset >= offset + bytes) {
S
Stefan Hajnoczi 已提交
2325 2326 2327
        return false;
    }
    return true;
2328 2329
}

2330
static bool coroutine_fn wait_serialising_requests(BdrvTrackedRequest *self)
2331
{
2332
    BlockDriverState *bs = self->bs;
2333 2334
    BdrvTrackedRequest *req;
    bool retry;
2335
    bool waited = false;
2336

2337
    if (!bs->serialising_in_flight) {
2338
        return false;
2339 2340
    }

2341 2342 2343
    do {
        retry = false;
        QLIST_FOREACH(req, &bs->tracked_requests, list) {
2344
            if (req == self || (!req->serialising && !self->serialising)) {
2345 2346
                continue;
            }
2347 2348 2349
            if (tracked_request_overlaps(req, self->overlap_offset,
                                         self->overlap_bytes))
            {
2350 2351 2352 2353 2354 2355
                /* 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);

2356 2357 2358 2359 2360 2361 2362 2363
                /* If the request is already (indirectly) waiting for us, or
                 * will wait for us as soon as it wakes up, then just go on
                 * (instead of producing a deadlock in the former case). */
                if (!req->waiting_for) {
                    self->waiting_for = req;
                    qemu_co_queue_wait(&req->wait_queue);
                    self->waiting_for = NULL;
                    retry = true;
2364
                    waited = true;
2365 2366
                    break;
                }
2367 2368 2369
            }
        }
    } while (retry);
2370 2371

    return waited;
2372 2373
}

K
Kevin Wolf 已提交
2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385
/*
 * 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;
2386
    int ret;
K
Kevin Wolf 已提交
2387

2388 2389 2390 2391 2392
    /* Backing file format doesn't make sense without a backing file */
    if (backing_fmt && !backing_file) {
        return -EINVAL;
    }

K
Kevin Wolf 已提交
2393
    if (drv->bdrv_change_backing_file != NULL) {
2394
        ret = drv->bdrv_change_backing_file(bs, backing_file, backing_fmt);
K
Kevin Wolf 已提交
2395
    } else {
2396
        ret = -ENOTSUP;
K
Kevin Wolf 已提交
2397
    }
2398 2399 2400 2401 2402 2403

    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
2404 2405
}

2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532
/*
 * 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;

2533
    bdrv_refresh_limits(new_top_bs);
2534 2535 2536 2537

    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
2538
        bdrv_unref(intermediate_state->bs);
2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549
    }
    ret = 0;

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


2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562
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
2563 2564 2565 2566
    if (offset < 0)
        return -EIO;

    if ((offset > len) || (len - offset < size))
2567 2568 2569 2570 2571 2572 2573 2574
        return -EIO;

    return 0;
}

static int bdrv_check_request(BlockDriverState *bs, int64_t sector_num,
                              int nb_sectors)
{
2575 2576
    return bdrv_check_byte_request(bs, sector_num * BDRV_SECTOR_SIZE,
                                   nb_sectors * BDRV_SECTOR_SIZE);
2577 2578
}

2579 2580
typedef struct RwCo {
    BlockDriverState *bs;
2581
    int64_t offset;
2582 2583 2584
    QEMUIOVector *qiov;
    bool is_write;
    int ret;
P
Peter Lieven 已提交
2585
    BdrvRequestFlags flags;
2586 2587 2588
} RwCo;

static void coroutine_fn bdrv_rw_co_entry(void *opaque)
B
bellard 已提交
2589
{
2590
    RwCo *rwco = opaque;
B
bellard 已提交
2591

2592
    if (!rwco->is_write) {
2593 2594
        rwco->ret = bdrv_co_do_preadv(rwco->bs, rwco->offset,
                                      rwco->qiov->size, rwco->qiov,
P
Peter Lieven 已提交
2595
                                      rwco->flags);
2596 2597 2598 2599
    } else {
        rwco->ret = bdrv_co_do_pwritev(rwco->bs, rwco->offset,
                                       rwco->qiov->size, rwco->qiov,
                                       rwco->flags);
2600 2601
    }
}
2602

2603
/*
2604
 * Process a vectored synchronous request using coroutines
2605
 */
2606 2607 2608
static int bdrv_prwv_co(BlockDriverState *bs, int64_t offset,
                        QEMUIOVector *qiov, bool is_write,
                        BdrvRequestFlags flags)
2609 2610 2611 2612
{
    Coroutine *co;
    RwCo rwco = {
        .bs = bs,
2613
        .offset = offset,
2614
        .qiov = qiov,
2615 2616
        .is_write = is_write,
        .ret = NOT_DONE,
P
Peter Lieven 已提交
2617
        .flags = flags,
2618
    };
2619

2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630
    /**
     * 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);
    }

2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642
    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
2643

2644 2645 2646 2647
/*
 * 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
2648
                      int nb_sectors, bool is_write, BdrvRequestFlags flags)
2649 2650 2651 2652 2653 2654 2655 2656
{
    QEMUIOVector qiov;
    struct iovec iov = {
        .iov_base = (void *)buf,
        .iov_len = nb_sectors * BDRV_SECTOR_SIZE,
    };

    qemu_iovec_init_external(&qiov, &iov, 1);
2657 2658
    return bdrv_prwv_co(bs, sector_num << BDRV_SECTOR_BITS,
                        &qiov, is_write, flags);
2659 2660
}

2661 2662 2663 2664
/* 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
2665
    return bdrv_rw_co(bs, sector_num, buf, nb_sectors, false, 0);
B
bellard 已提交
2666 2667
}

2668 2669 2670 2671 2672 2673 2674 2675 2676
/* 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;
2677
    ret = bdrv_read(bs, sector_num, buf, nb_sectors);
2678 2679 2680 2681
    bs->io_limits_enabled = enabled;
    return ret;
}

2682
/* Return < 0 if error. Important errors are:
B
bellard 已提交
2683 2684 2685 2686 2687
  -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
*/
2688
int bdrv_write(BlockDriverState *bs, int64_t sector_num,
B
bellard 已提交
2689 2690
               const uint8_t *buf, int nb_sectors)
{
P
Peter Lieven 已提交
2691
    return bdrv_rw_co(bs, sector_num, (uint8_t *)buf, nb_sectors, true, 0);
B
bellard 已提交
2692 2693
}

2694 2695
int bdrv_write_zeroes(BlockDriverState *bs, int64_t sector_num,
                      int nb_sectors, BdrvRequestFlags flags)
P
Peter Lieven 已提交
2696 2697
{
    return bdrv_rw_co(bs, sector_num, NULL, nb_sectors, true,
2698
                      BDRV_REQ_ZERO_WRITE | flags);
2699 2700
}

P
Peter Lieven 已提交
2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723
/*
 * Completely zero out a block device with the help of bdrv_write_zeroes.
 * The operation is sped up by checking the block status and only writing
 * zeroes to the device if they currently do not return zeroes. Optional
 * flags are passed through to bdrv_write_zeroes (e.g. BDRV_REQ_MAY_UNMAP).
 *
 * Returns < 0 on error, 0 on success. For error codes see bdrv_write().
 */
int bdrv_make_zero(BlockDriverState *bs, BdrvRequestFlags flags)
{
    int64_t target_size = bdrv_getlength(bs) / BDRV_SECTOR_SIZE;
    int64_t ret, nb_sectors, sector_num = 0;
    int n;

    for (;;) {
        nb_sectors = target_size - sector_num;
        if (nb_sectors <= 0) {
            return 0;
        }
        if (nb_sectors > INT_MAX) {
            nb_sectors = INT_MAX;
        }
        ret = bdrv_get_block_status(bs, sector_num, nb_sectors, &n);
2724 2725 2726 2727 2728
        if (ret < 0) {
            error_report("error getting block status at sector %" PRId64 ": %s",
                         sector_num, strerror(-ret));
            return ret;
        }
P
Peter Lieven 已提交
2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742
        if (ret & BDRV_BLOCK_ZERO) {
            sector_num += n;
            continue;
        }
        ret = bdrv_write_zeroes(bs, sector_num, n, flags);
        if (ret < 0) {
            error_report("error writing zeroes at sector %" PRId64 ": %s",
                         sector_num, strerror(-ret));
            return ret;
        }
        sector_num += n;
    }
}

2743
int bdrv_pread(BlockDriverState *bs, int64_t offset, void *buf, int bytes)
B
bellard 已提交
2744
{
2745 2746 2747 2748 2749
    QEMUIOVector qiov;
    struct iovec iov = {
        .iov_base = (void *)buf,
        .iov_len = bytes,
    };
2750
    int ret;
B
bellard 已提交
2751

2752 2753
    if (bytes < 0) {
        return -EINVAL;
B
bellard 已提交
2754 2755
    }

2756 2757 2758 2759
    qemu_iovec_init_external(&qiov, &iov, 1);
    ret = bdrv_prwv_co(bs, offset, &qiov, false, 0);
    if (ret < 0) {
        return ret;
B
bellard 已提交
2760
    }
2761 2762

    return bytes;
B
bellard 已提交
2763 2764
}

2765
int bdrv_pwritev(BlockDriverState *bs, int64_t offset, QEMUIOVector *qiov)
B
bellard 已提交
2766
{
2767
    int ret;
B
bellard 已提交
2768

2769 2770 2771
    ret = bdrv_prwv_co(bs, offset, qiov, true, 0);
    if (ret < 0) {
        return ret;
B
bellard 已提交
2772 2773
    }

2774 2775 2776 2777
    return qiov->size;
}

int bdrv_pwrite(BlockDriverState *bs, int64_t offset,
2778
                const void *buf, int bytes)
2779 2780 2781 2782
{
    QEMUIOVector qiov;
    struct iovec iov = {
        .iov_base   = (void *) buf,
2783
        .iov_len    = bytes,
2784 2785
    };

2786 2787 2788 2789
    if (bytes < 0) {
        return -EINVAL;
    }

2790 2791
    qemu_iovec_init_external(&qiov, &iov, 1);
    return bdrv_pwritev(bs, offset, &qiov);
B
bellard 已提交
2792 2793
}

K
Kevin Wolf 已提交
2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809
/*
 * 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;
    }

2810 2811
    /* No flush needed for cache modes that already do it */
    if (bs->enable_write_cache) {
K
Kevin Wolf 已提交
2812 2813 2814 2815 2816 2817
        bdrv_flush(bs);
    }

    return 0;
}

2818
static int coroutine_fn bdrv_co_do_copy_on_readv(BlockDriverState *bs,
S
Stefan Hajnoczi 已提交
2819 2820 2821 2822 2823 2824 2825 2826 2827
        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;

2828
    BlockDriver *drv = bs->drv;
S
Stefan Hajnoczi 已提交
2829 2830 2831 2832 2833 2834 2835 2836 2837 2838
    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.
     */
2839 2840
    bdrv_round_to_clusters(bs, sector_num, nb_sectors,
                           &cluster_sector_num, &cluster_nb_sectors);
S
Stefan Hajnoczi 已提交
2841

2842 2843
    trace_bdrv_co_do_copy_on_readv(bs, sector_num, nb_sectors,
                                   cluster_sector_num, cluster_nb_sectors);
S
Stefan Hajnoczi 已提交
2844 2845 2846 2847 2848

    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);

2849 2850
    ret = drv->bdrv_co_readv(bs, cluster_sector_num, cluster_nb_sectors,
                             &bounce_qiov);
S
Stefan Hajnoczi 已提交
2851 2852 2853 2854
    if (ret < 0) {
        goto err;
    }

2855 2856
    if (drv->bdrv_co_write_zeroes &&
        buffer_is_zero(bounce_buffer, iov.iov_len)) {
K
Kevin Wolf 已提交
2857
        ret = bdrv_co_do_write_zeroes(bs, cluster_sector_num,
2858
                                      cluster_nb_sectors, 0);
2859
    } else {
2860 2861 2862
        /* This does not change the data on the disk, it is not necessary
         * to flush even in cache=writethrough mode.
         */
2863
        ret = drv->bdrv_co_writev(bs, cluster_sector_num, cluster_nb_sectors,
S
Stefan Hajnoczi 已提交
2864
                                  &bounce_qiov);
2865 2866
    }

S
Stefan Hajnoczi 已提交
2867 2868 2869 2870 2871 2872 2873 2874 2875
    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;
2876 2877
    qemu_iovec_from_buf(qiov, 0, bounce_buffer + skip_bytes,
                        nb_sectors * BDRV_SECTOR_SIZE);
S
Stefan Hajnoczi 已提交
2878 2879 2880 2881 2882 2883

err:
    qemu_vfree(bounce_buffer);
    return ret;
}

2884
/*
2885 2886 2887
 * Forwards an already correctly aligned request to the BlockDriver. This
 * handles copy on read and zeroing after EOF; any other features must be
 * implemented by the caller.
2888
 */
2889
static int coroutine_fn bdrv_aligned_preadv(BlockDriverState *bs,
2890
    BdrvTrackedRequest *req, int64_t offset, unsigned int bytes,
2891
    int64_t align, QEMUIOVector *qiov, int flags)
K
Kevin Wolf 已提交
2892 2893
{
    BlockDriver *drv = bs->drv;
S
Stefan Hajnoczi 已提交
2894
    int ret;
K
Kevin Wolf 已提交
2895

2896 2897
    int64_t sector_num = offset >> BDRV_SECTOR_BITS;
    unsigned int nb_sectors = bytes >> BDRV_SECTOR_BITS;
K
Kevin Wolf 已提交
2898

2899 2900 2901 2902
    assert((offset & (BDRV_SECTOR_SIZE - 1)) == 0);
    assert((bytes & (BDRV_SECTOR_SIZE - 1)) == 0);

    /* Handle Copy on Read and associated serialisation */
2903
    if (flags & BDRV_REQ_COPY_ON_READ) {
2904 2905 2906 2907 2908 2909
        /* 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. */
        mark_request_serialising(req, bdrv_get_cluster_size(bs));
2910 2911
    }

2912
    wait_serialising_requests(req);
2913

2914
    if (flags & BDRV_REQ_COPY_ON_READ) {
S
Stefan Hajnoczi 已提交
2915 2916
        int pnum;

2917
        ret = bdrv_is_allocated(bs, sector_num, nb_sectors, &pnum);
S
Stefan Hajnoczi 已提交
2918 2919 2920 2921 2922
        if (ret < 0) {
            goto out;
        }

        if (!ret || pnum != nb_sectors) {
2923
            ret = bdrv_co_do_copy_on_readv(bs, sector_num, nb_sectors, qiov);
S
Stefan Hajnoczi 已提交
2924 2925 2926 2927
            goto out;
        }
    }

2928
    /* Forward the request to the BlockDriver */
2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940
    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;
        }

2941
        total_sectors = DIV_ROUND_UP(len, BDRV_SECTOR_SIZE);
2942 2943
        max_nb_sectors = ROUND_UP(MAX(0, total_sectors - sector_num),
                                  align >> BDRV_SECTOR_BITS);
2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958
        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
2959 2960

out:
S
Stefan Hajnoczi 已提交
2961
    return ret;
K
Kevin Wolf 已提交
2962 2963
}

2964 2965 2966
/*
 * Handle a read request in coroutine context
 */
2967 2968
static int coroutine_fn bdrv_co_do_preadv(BlockDriverState *bs,
    int64_t offset, unsigned int bytes, QEMUIOVector *qiov,
2969 2970 2971
    BdrvRequestFlags flags)
{
    BlockDriver *drv = bs->drv;
2972 2973
    BdrvTrackedRequest req;

2974 2975 2976 2977 2978 2979
    /* TODO Lift BDRV_SECTOR_SIZE restriction in BlockDriver interface */
    uint64_t align = MAX(BDRV_SECTOR_SIZE, bs->request_alignment);
    uint8_t *head_buf = NULL;
    uint8_t *tail_buf = NULL;
    QEMUIOVector local_qiov;
    bool use_local_qiov = false;
2980 2981 2982 2983 2984
    int ret;

    if (!drv) {
        return -ENOMEDIUM;
    }
2985
    if (bdrv_check_byte_request(bs, offset, bytes)) {
2986 2987 2988 2989 2990 2991 2992 2993 2994
        return -EIO;
    }

    if (bs->copy_on_read) {
        flags |= BDRV_REQ_COPY_ON_READ;
    }

    /* throttling disk I/O */
    if (bs->io_limits_enabled) {
2995
        bdrv_io_limits_intercept(bs, bytes, false);
2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014 3015 3016 3017 3018 3019 3020 3021 3022
    }

    /* Align read if necessary by padding qiov */
    if (offset & (align - 1)) {
        head_buf = qemu_blockalign(bs, align);
        qemu_iovec_init(&local_qiov, qiov->niov + 2);
        qemu_iovec_add(&local_qiov, head_buf, offset & (align - 1));
        qemu_iovec_concat(&local_qiov, qiov, 0, qiov->size);
        use_local_qiov = true;

        bytes += offset & (align - 1);
        offset = offset & ~(align - 1);
    }

    if ((offset + bytes) & (align - 1)) {
        if (!use_local_qiov) {
            qemu_iovec_init(&local_qiov, qiov->niov + 1);
            qemu_iovec_concat(&local_qiov, qiov, 0, qiov->size);
            use_local_qiov = true;
        }
        tail_buf = qemu_blockalign(bs, align);
        qemu_iovec_add(&local_qiov, tail_buf,
                       align - ((offset + bytes) & (align - 1)));

        bytes = ROUND_UP(bytes, align);
    }

3023
    tracked_request_begin(&req, bs, offset, bytes, false);
3024
    ret = bdrv_aligned_preadv(bs, &req, offset, bytes, align,
3025 3026
                              use_local_qiov ? &local_qiov : qiov,
                              flags);
3027
    tracked_request_end(&req);
3028 3029 3030 3031 3032

    if (use_local_qiov) {
        qemu_iovec_destroy(&local_qiov);
        qemu_vfree(head_buf);
        qemu_vfree(tail_buf);
3033 3034 3035 3036 3037
    }

    return ret;
}

3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049
static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags)
{
    if (nb_sectors < 0 || nb_sectors > (UINT_MAX >> BDRV_SECTOR_BITS)) {
        return -EINVAL;
    }

    return bdrv_co_do_preadv(bs, sector_num << BDRV_SECTOR_BITS,
                             nb_sectors << BDRV_SECTOR_BITS, qiov, flags);
}

3050
int coroutine_fn bdrv_co_readv(BlockDriverState *bs, int64_t sector_num,
K
Kevin Wolf 已提交
3051 3052
    int nb_sectors, QEMUIOVector *qiov)
{
3053
    trace_bdrv_co_readv(bs, sector_num, nb_sectors);
K
Kevin Wolf 已提交
3054

3055 3056 3057 3058 3059 3060 3061 3062 3063 3064
    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);
3065 3066
}

3067 3068 3069 3070 3071
/* if no limit is specified in the BlockLimits use a default
 * of 32768 512-byte sectors (16 MiB) per request.
 */
#define MAX_WRITE_ZEROES_DEFAULT 32768

3072
static int coroutine_fn bdrv_co_do_write_zeroes(BlockDriverState *bs,
3073
    int64_t sector_num, int nb_sectors, BdrvRequestFlags flags)
3074 3075 3076
{
    BlockDriver *drv = bs->drv;
    QEMUIOVector qiov;
3077 3078
    struct iovec iov = {0};
    int ret = 0;
3079

3080 3081
    int max_write_zeroes = bs->bl.max_write_zeroes ?
                           bs->bl.max_write_zeroes : MAX_WRITE_ZEROES_DEFAULT;
K
Kevin Wolf 已提交
3082

3083 3084 3085
    while (nb_sectors > 0 && !ret) {
        int num = nb_sectors;

3086 3087 3088 3089 3090 3091 3092
        /* Align request.  Block drivers can expect the "bulk" of the request
         * to be aligned.
         */
        if (bs->bl.write_zeroes_alignment
            && num > bs->bl.write_zeroes_alignment) {
            if (sector_num % bs->bl.write_zeroes_alignment != 0) {
                /* Make a small request up to the first aligned sector.  */
3093
                num = bs->bl.write_zeroes_alignment;
3094 3095 3096 3097 3098 3099
                num -= sector_num % bs->bl.write_zeroes_alignment;
            } else if ((sector_num + num) % bs->bl.write_zeroes_alignment != 0) {
                /* Shorten the request to the last aligned sector.  num cannot
                 * underflow because num > bs->bl.write_zeroes_alignment.
                 */
                num -= (sector_num + num) % bs->bl.write_zeroes_alignment;
3100
            }
K
Kevin Wolf 已提交
3101
        }
3102

3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117
        /* limit request size */
        if (num > max_write_zeroes) {
            num = max_write_zeroes;
        }

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

        if (ret == -ENOTSUP) {
            /* Fall back to bounce buffer if write zeroes is unsupported */
            iov.iov_len = num * BDRV_SECTOR_SIZE;
            if (iov.iov_base == NULL) {
3118 3119
                iov.iov_base = qemu_blockalign(bs, num * BDRV_SECTOR_SIZE);
                memset(iov.iov_base, 0, num * BDRV_SECTOR_SIZE);
3120 3121
            }
            qemu_iovec_init_external(&qiov, &iov, 1);
3122

3123
            ret = drv->bdrv_co_writev(bs, sector_num, num, &qiov);
3124 3125 3126 3127 3128 3129 3130 3131

            /* Keep bounce buffer around if it is big enough for all
             * all future requests.
             */
            if (num < max_write_zeroes) {
                qemu_vfree(iov.iov_base);
                iov.iov_base = NULL;
            }
3132 3133 3134 3135 3136
        }

        sector_num += num;
        nb_sectors -= num;
    }
3137 3138 3139 3140 3141

    qemu_vfree(iov.iov_base);
    return ret;
}

3142
/*
3143
 * Forwards an already correctly aligned write request to the BlockDriver.
3144
 */
3145
static int coroutine_fn bdrv_aligned_pwritev(BlockDriverState *bs,
3146 3147
    BdrvTrackedRequest *req, int64_t offset, unsigned int bytes,
    QEMUIOVector *qiov, int flags)
3148 3149
{
    BlockDriver *drv = bs->drv;
3150
    bool waited;
3151
    int ret;
K
Kevin Wolf 已提交
3152

3153 3154
    int64_t sector_num = offset >> BDRV_SECTOR_BITS;
    unsigned int nb_sectors = bytes >> BDRV_SECTOR_BITS;
3155

3156 3157
    assert((offset & (BDRV_SECTOR_SIZE - 1)) == 0);
    assert((bytes & (BDRV_SECTOR_SIZE - 1)) == 0);
3158

3159 3160
    waited = wait_serialising_requests(req);
    assert(!waited || !req->serialising);
3161 3162
    assert(req->overlap_offset <= offset);
    assert(offset + bytes <= req->overlap_offset + req->overlap_bytes);
3163

3164
    ret = notifier_with_return_list_notify(&bs->before_write_notifiers, req);
3165 3166 3167 3168

    if (ret < 0) {
        /* Do nothing, write notifier decided to fail this request */
    } else if (flags & BDRV_REQ_ZERO_WRITE) {
3169
        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_ZERO);
3170
        ret = bdrv_co_do_write_zeroes(bs, sector_num, nb_sectors, flags);
3171
    } else {
3172
        BLKDBG_EVENT(bs, BLKDBG_PWRITEV);
3173 3174
        ret = drv->bdrv_co_writev(bs, sector_num, nb_sectors, qiov);
    }
3175
    BLKDBG_EVENT(bs, BLKDBG_PWRITEV_DONE);
3176

3177 3178 3179 3180
    if (ret == 0 && !bs->enable_write_cache) {
        ret = bdrv_co_flush(bs);
    }

F
Fam Zheng 已提交
3181
    bdrv_set_dirty(bs, sector_num, nb_sectors);
K
Kevin Wolf 已提交
3182 3183 3184 3185

    if (bs->wr_highest_sector < sector_num + nb_sectors - 1) {
        bs->wr_highest_sector = sector_num + nb_sectors - 1;
    }
3186 3187 3188
    if (bs->growable && ret >= 0) {
        bs->total_sectors = MAX(bs->total_sectors, sector_num + nb_sectors);
    }
K
Kevin Wolf 已提交
3189

3190
    return ret;
K
Kevin Wolf 已提交
3191 3192
}

3193 3194 3195
/*
 * Handle a write request in coroutine context
 */
3196 3197
static int coroutine_fn bdrv_co_do_pwritev(BlockDriverState *bs,
    int64_t offset, unsigned int bytes, QEMUIOVector *qiov,
3198 3199
    BdrvRequestFlags flags)
{
3200
    BdrvTrackedRequest req;
3201 3202 3203 3204 3205 3206
    /* TODO Lift BDRV_SECTOR_SIZE restriction in BlockDriver interface */
    uint64_t align = MAX(BDRV_SECTOR_SIZE, bs->request_alignment);
    uint8_t *head_buf = NULL;
    uint8_t *tail_buf = NULL;
    QEMUIOVector local_qiov;
    bool use_local_qiov = false;
3207 3208 3209 3210 3211 3212 3213 3214
    int ret;

    if (!bs->drv) {
        return -ENOMEDIUM;
    }
    if (bs->read_only) {
        return -EACCES;
    }
3215
    if (bdrv_check_byte_request(bs, offset, bytes)) {
3216 3217 3218 3219 3220
        return -EIO;
    }

    /* throttling disk I/O */
    if (bs->io_limits_enabled) {
3221
        bdrv_io_limits_intercept(bs, bytes, true);
3222 3223
    }

3224 3225 3226 3227 3228
    /*
     * Align write if necessary by performing a read-modify-write cycle.
     * Pad qiov with the read parts and be sure to have a tracked request not
     * only for bdrv_aligned_pwritev, but also for the reads of the RMW cycle.
     */
3229
    tracked_request_begin(&req, bs, offset, bytes, true);
3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244

    if (offset & (align - 1)) {
        QEMUIOVector head_qiov;
        struct iovec head_iov;

        mark_request_serialising(&req, align);
        wait_serialising_requests(&req);

        head_buf = qemu_blockalign(bs, align);
        head_iov = (struct iovec) {
            .iov_base   = head_buf,
            .iov_len    = align,
        };
        qemu_iovec_init_external(&head_qiov, &head_iov, 1);

3245
        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_HEAD);
3246 3247 3248 3249 3250
        ret = bdrv_aligned_preadv(bs, &req, offset & ~(align - 1), align,
                                  align, &head_qiov, 0);
        if (ret < 0) {
            goto fail;
        }
3251
        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_AFTER_HEAD);
3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265

        qemu_iovec_init(&local_qiov, qiov->niov + 2);
        qemu_iovec_add(&local_qiov, head_buf, offset & (align - 1));
        qemu_iovec_concat(&local_qiov, qiov, 0, qiov->size);
        use_local_qiov = true;

        bytes += offset & (align - 1);
        offset = offset & ~(align - 1);
    }

    if ((offset + bytes) & (align - 1)) {
        QEMUIOVector tail_qiov;
        struct iovec tail_iov;
        size_t tail_bytes;
3266
        bool waited;
3267 3268

        mark_request_serialising(&req, align);
3269 3270
        waited = wait_serialising_requests(&req);
        assert(!waited || !use_local_qiov);
3271 3272 3273 3274 3275 3276 3277 3278

        tail_buf = qemu_blockalign(bs, align);
        tail_iov = (struct iovec) {
            .iov_base   = tail_buf,
            .iov_len    = align,
        };
        qemu_iovec_init_external(&tail_qiov, &tail_iov, 1);

3279
        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_TAIL);
3280 3281 3282 3283 3284
        ret = bdrv_aligned_preadv(bs, &req, (offset + bytes) & ~(align - 1), align,
                                  align, &tail_qiov, 0);
        if (ret < 0) {
            goto fail;
        }
3285
        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_AFTER_TAIL);
3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303

        if (!use_local_qiov) {
            qemu_iovec_init(&local_qiov, qiov->niov + 1);
            qemu_iovec_concat(&local_qiov, qiov, 0, qiov->size);
            use_local_qiov = true;
        }

        tail_bytes = (offset + bytes) & (align - 1);
        qemu_iovec_add(&local_qiov, tail_buf + tail_bytes, align - tail_bytes);

        bytes = ROUND_UP(bytes, align);
    }

    ret = bdrv_aligned_pwritev(bs, &req, offset, bytes,
                               use_local_qiov ? &local_qiov : qiov,
                               flags);

fail:
3304
    tracked_request_end(&req);
3305

3306 3307 3308
    if (use_local_qiov) {
        qemu_iovec_destroy(&local_qiov);
    }
3309 3310
    qemu_vfree(head_buf);
    qemu_vfree(tail_buf);
3311

3312 3313 3314
    return ret;
}

3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326
static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs,
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags)
{
    if (nb_sectors < 0 || nb_sectors > (INT_MAX >> BDRV_SECTOR_BITS)) {
        return -EINVAL;
    }

    return bdrv_co_do_pwritev(bs, sector_num << BDRV_SECTOR_BITS,
                              nb_sectors << BDRV_SECTOR_BITS, qiov, flags);
}

3327 3328 3329 3330 3331
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);

3332 3333 3334 3335
    return bdrv_co_do_writev(bs, sector_num, nb_sectors, qiov, 0);
}

int coroutine_fn bdrv_co_write_zeroes(BlockDriverState *bs,
3336 3337
                                      int64_t sector_num, int nb_sectors,
                                      BdrvRequestFlags flags)
3338
{
3339
    trace_bdrv_co_write_zeroes(bs, sector_num, nb_sectors, flags);
3340

3341 3342 3343 3344
    if (!(bs->open_flags & BDRV_O_UNMAP)) {
        flags &= ~BDRV_REQ_MAY_UNMAP;
    }

3345
    return bdrv_co_do_writev(bs, sector_num, nb_sectors, NULL,
3346
                             BDRV_REQ_ZERO_WRITE | flags);
3347 3348
}

B
bellard 已提交
3349 3350 3351 3352 3353 3354
/**
 * Truncate file to 'offset' bytes (needed only for file protocols)
 */
int bdrv_truncate(BlockDriverState *bs, int64_t offset)
{
    BlockDriver *drv = bs->drv;
3355
    int ret;
B
bellard 已提交
3356
    if (!drv)
B
bellard 已提交
3357
        return -ENOMEDIUM;
B
bellard 已提交
3358 3359
    if (!drv->bdrv_truncate)
        return -ENOTSUP;
3360 3361
    if (bs->read_only)
        return -EACCES;
M
Marcelo Tosatti 已提交
3362 3363
    if (bdrv_in_use(bs))
        return -EBUSY;
3364 3365 3366
    ret = drv->bdrv_truncate(bs, offset);
    if (ret == 0) {
        ret = refresh_total_sectors(bs, offset >> BDRV_SECTOR_BITS);
3367
        bdrv_dev_resize_cb(bs);
3368 3369
    }
    return ret;
B
bellard 已提交
3370 3371
}

3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389 3390
/**
 * 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
3391 3392 3393 3394 3395 3396 3397
/**
 * 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
3398
        return -ENOMEDIUM;
3399

3400 3401 3402 3403
    if (drv->has_variable_length) {
        int ret = refresh_total_sectors(bs, bs->total_sectors);
        if (ret < 0) {
            return ret;
3404
        }
B
bellard 已提交
3405
    }
3406
    return bs->total_sectors * BDRV_SECTOR_SIZE;
B
bellard 已提交
3407 3408
}

B
bellard 已提交
3409
/* return 0 as number of sectors if no device present or error */
3410
void bdrv_get_geometry(BlockDriverState *bs, uint64_t *nb_sectors_ptr)
B
bellard 已提交
3411
{
B
bellard 已提交
3412 3413 3414 3415 3416
    int64_t length;
    length = bdrv_getlength(bs);
    if (length < 0)
        length = 0;
    else
J
Jan Kiszka 已提交
3417
        length = length >> BDRV_SECTOR_BITS;
B
bellard 已提交
3418
    *nb_sectors_ptr = length;
B
bellard 已提交
3419
}
B
bellard 已提交
3420

3421 3422
void bdrv_set_on_error(BlockDriverState *bs, BlockdevOnError on_read_error,
                       BlockdevOnError on_write_error)
3423 3424 3425 3426 3427
{
    bs->on_read_error = on_read_error;
    bs->on_write_error = on_write_error;
}

3428
BlockdevOnError bdrv_get_on_error(BlockDriverState *bs, bool is_read)
3429 3430 3431 3432
{
    return is_read ? bs->on_read_error : bs->on_write_error;
}

3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458
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
3459
    bdrv_emit_qmp_error_event(bs, QEVENT_BLOCK_IO_ERROR, action, is_read);
3460 3461 3462 3463 3464 3465
    if (action == BDRV_ACTION_STOP) {
        vm_stop(RUN_STATE_IO_ERROR);
        bdrv_iostatus_set_err(bs, error);
    }
}

B
bellard 已提交
3466 3467 3468 3469 3470
int bdrv_is_read_only(BlockDriverState *bs)
{
    return bs->read_only;
}

3471 3472 3473 3474 3475
int bdrv_is_sg(BlockDriverState *bs)
{
    return bs->sg;
}

3476 3477 3478 3479 3480
int bdrv_enable_write_cache(BlockDriverState *bs)
{
    return bs->enable_write_cache;
}

3481 3482 3483
void bdrv_set_enable_write_cache(BlockDriverState *bs, bool wce)
{
    bs->enable_write_cache = wce;
3484 3485 3486 3487 3488 3489 3490

    /* so a reopen() will preserve wce */
    if (wce) {
        bs->open_flags |= BDRV_O_CACHE_WB;
    } else {
        bs->open_flags &= ~BDRV_O_CACHE_WB;
    }
3491 3492
}

B
bellard 已提交
3493 3494 3495 3496 3497 3498 3499
int bdrv_is_encrypted(BlockDriverState *bs)
{
    if (bs->backing_hd && bs->backing_hd->encrypted)
        return 1;
    return bs->encrypted;
}

3500 3501 3502 3503 3504 3505 3506 3507 3508
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
3509 3510 3511 3512 3513 3514 3515 3516 3517 3518
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;
    }
3519 3520 3521 3522 3523
    if (!bs->encrypted) {
        return -EINVAL;
    } else if (!bs->drv || !bs->drv->bdrv_set_key) {
        return -ENOMEDIUM;
    }
3524
    ret = bs->drv->bdrv_set_key(bs, key);
3525 3526 3527 3528 3529
    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 */
3530
        bdrv_dev_change_media_cb(bs, true);
3531
    }
3532
    return ret;
B
bellard 已提交
3533 3534
}

3535
const char *bdrv_get_format_name(BlockDriverState *bs)
B
bellard 已提交
3536
{
3537
    return bs->drv ? bs->drv->format_name : NULL;
B
bellard 已提交
3538 3539
}

3540
void bdrv_iterate_format(void (*it)(void *opaque, const char *name),
B
bellard 已提交
3541 3542 3543 3544
                         void *opaque)
{
    BlockDriver *drv;

3545
    QLIST_FOREACH(drv, &bdrv_drivers, list) {
B
bellard 已提交
3546 3547 3548 3549
        it(opaque, drv->format_name);
    }
}

3550
/* This function is to find block backend bs */
B
bellard 已提交
3551 3552 3553 3554
BlockDriverState *bdrv_find(const char *name)
{
    BlockDriverState *bs;

3555
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
3556
        if (!strcmp(name, bs->device_name)) {
B
bellard 已提交
3557
            return bs;
3558
        }
B
bellard 已提交
3559 3560 3561 3562
    }
    return NULL;
}

3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577
/* This function is to find a node in the bs graph */
BlockDriverState *bdrv_find_node(const char *node_name)
{
    BlockDriverState *bs;

    assert(node_name);

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

3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594
/* Put this QMP function here so it can access the static graph_bdrv_states. */
BlockDeviceInfoList *bdrv_named_nodes_list(void)
{
    BlockDeviceInfoList *list, *entry;
    BlockDriverState *bs;

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

    return list;
}

3595 3596 3597 3598 3599 3600 3601 3602 3603
BlockDriverState *bdrv_lookup_bs(const char *device,
                                 const char *node_name,
                                 Error **errp)
{
    BlockDriverState *bs = NULL;

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

3604 3605
        if (bs) {
            return bs;
3606 3607 3608
        }
    }

3609 3610
    if (node_name) {
        bs = bdrv_find_node(node_name);
3611

3612 3613 3614
        if (bs) {
            return bs;
        }
3615 3616
    }

3617 3618 3619 3620
    error_setg(errp, "Cannot find device=%s nor node_name=%s",
                     device ? device : "",
                     node_name ? node_name : "");
    return NULL;
3621 3622
}

M
Markus Armbruster 已提交
3623 3624 3625 3626 3627
BlockDriverState *bdrv_next(BlockDriverState *bs)
{
    if (!bs) {
        return QTAILQ_FIRST(&bdrv_states);
    }
3628
    return QTAILQ_NEXT(bs, device_list);
M
Markus Armbruster 已提交
3629 3630
}

3631
void bdrv_iterate(void (*it)(void *opaque, BlockDriverState *bs), void *opaque)
B
bellard 已提交
3632 3633 3634
{
    BlockDriverState *bs;

3635
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
3636
        it(opaque, bs);
B
bellard 已提交
3637 3638 3639
    }
}

B
bellard 已提交
3640 3641 3642 3643 3644
const char *bdrv_get_device_name(BlockDriverState *bs)
{
    return bs->device_name;
}

M
Markus Armbruster 已提交
3645 3646 3647 3648 3649
int bdrv_get_flags(BlockDriverState *bs)
{
    return bs->open_flags;
}

3650
int bdrv_flush_all(void)
A
aliguori 已提交
3651 3652
{
    BlockDriverState *bs;
3653
    int result = 0;
A
aliguori 已提交
3654

3655
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
3656 3657 3658 3659
        int ret = bdrv_flush(bs);
        if (ret < 0 && !result) {
            result = ret;
        }
3660
    }
3661 3662

    return result;
A
aliguori 已提交
3663 3664
}

3665 3666 3667 3668 3669
int bdrv_has_zero_init_1(BlockDriverState *bs)
{
    return 1;
}

K
Kevin Wolf 已提交
3670 3671 3672 3673
int bdrv_has_zero_init(BlockDriverState *bs)
{
    assert(bs->drv);

3674 3675 3676 3677 3678
    /* 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
3679 3680
    if (bs->drv->bdrv_has_zero_init) {
        return bs->drv->bdrv_has_zero_init(bs);
K
Kevin Wolf 已提交
3681 3682
    }

3683 3684
    /* safe default */
    return 0;
K
Kevin Wolf 已提交
3685 3686
}

3687 3688 3689 3690 3691 3692 3693 3694 3695 3696 3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714 3715 3716
bool bdrv_unallocated_blocks_are_zero(BlockDriverState *bs)
{
    BlockDriverInfo bdi;

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

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

    return false;
}

bool bdrv_can_write_zeroes_with_unmap(BlockDriverState *bs)
{
    BlockDriverInfo bdi;

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

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

    return false;
}

3717
typedef struct BdrvCoGetBlockStatusData {
3718
    BlockDriverState *bs;
3719
    BlockDriverState *base;
3720 3721 3722
    int64_t sector_num;
    int nb_sectors;
    int *pnum;
3723
    int64_t ret;
3724
    bool done;
3725
} BdrvCoGetBlockStatusData;
3726

3727 3728 3729 3730 3731
/*
 * 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.
 *
3732 3733 3734
 * If 'sector_num' is beyond the end of the disk image the return value is 0
 * and 'pnum' is set to 0.
 *
3735 3736 3737 3738
 * '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.
 *
3739 3740
 * '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.
3741
 */
3742 3743 3744
static int64_t coroutine_fn bdrv_co_get_block_status(BlockDriverState *bs,
                                                     int64_t sector_num,
                                                     int nb_sectors, int *pnum)
3745
{
3746
    int64_t length;
3747
    int64_t n;
3748
    int64_t ret, ret2;
3749

3750 3751 3752 3753 3754 3755
    length = bdrv_getlength(bs);
    if (length < 0) {
        return length;
    }

    if (sector_num >= (length >> BDRV_SECTOR_BITS)) {
3756 3757 3758 3759 3760 3761 3762 3763 3764
        *pnum = 0;
        return 0;
    }

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

3765
    if (!bs->drv->bdrv_co_get_block_status) {
3766
        *pnum = nb_sectors;
3767 3768 3769 3770 3771
        ret = BDRV_BLOCK_DATA;
        if (bs->drv->protocol_name) {
            ret |= BDRV_BLOCK_OFFSET_VALID | (sector_num * BDRV_SECTOR_SIZE);
        }
        return ret;
3772
    }
3773

3774 3775
    ret = bs->drv->bdrv_co_get_block_status(bs, sector_num, nb_sectors, pnum);
    if (ret < 0) {
3776
        *pnum = 0;
3777 3778 3779
        return ret;
    }

3780 3781 3782 3783 3784 3785
    if (ret & BDRV_BLOCK_RAW) {
        assert(ret & BDRV_BLOCK_OFFSET_VALID);
        return bdrv_get_block_status(bs->file, ret >> BDRV_SECTOR_BITS,
                                     *pnum, pnum);
    }

3786 3787
    if (!(ret & BDRV_BLOCK_DATA) && !(ret & BDRV_BLOCK_ZERO)) {
        if (bdrv_unallocated_blocks_are_zero(bs)) {
3788
            ret |= BDRV_BLOCK_ZERO;
3789
        } else if (bs->backing_hd) {
3790 3791 3792 3793 3794 3795
            BlockDriverState *bs2 = bs->backing_hd;
            int64_t length2 = bdrv_getlength(bs2);
            if (length2 >= 0 && sector_num >= (length2 >> BDRV_SECTOR_BITS)) {
                ret |= BDRV_BLOCK_ZERO;
            }
        }
3796
    }
3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810

    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);
        }
    }

3811
    return ret;
3812 3813
}

3814 3815
/* Coroutine wrapper for bdrv_get_block_status() */
static void coroutine_fn bdrv_get_block_status_co_entry(void *opaque)
3816
{
3817
    BdrvCoGetBlockStatusData *data = opaque;
3818 3819
    BlockDriverState *bs = data->bs;

3820 3821
    data->ret = bdrv_co_get_block_status(bs, data->sector_num, data->nb_sectors,
                                         data->pnum);
3822 3823 3824 3825
    data->done = true;
}

/*
3826
 * Synchronous wrapper around bdrv_co_get_block_status().
3827
 *
3828
 * See bdrv_co_get_block_status() for details.
3829
 */
3830 3831
int64_t bdrv_get_block_status(BlockDriverState *bs, int64_t sector_num,
                              int nb_sectors, int *pnum)
3832
{
3833
    Coroutine *co;
3834
    BdrvCoGetBlockStatusData data = {
3835 3836 3837 3838 3839 3840 3841
        .bs = bs,
        .sector_num = sector_num,
        .nb_sectors = nb_sectors,
        .pnum = pnum,
        .done = false,
    };

3842 3843
    if (qemu_in_coroutine()) {
        /* Fast-path if already in coroutine context */
3844
        bdrv_get_block_status_co_entry(&data);
3845
    } else {
3846
        co = qemu_coroutine_create(bdrv_get_block_status_co_entry);
3847 3848 3849 3850
        qemu_coroutine_enter(co, &data);
        while (!data.done) {
            qemu_aio_wait();
        }
3851 3852
    }
    return data.ret;
3853 3854
}

3855 3856 3857
int coroutine_fn bdrv_is_allocated(BlockDriverState *bs, int64_t sector_num,
                                   int nb_sectors, int *pnum)
{
3858 3859 3860 3861 3862 3863 3864
    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));
3865 3866
}

3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878
/*
 * 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.
 *
 */
3879 3880 3881 3882
int bdrv_is_allocated_above(BlockDriverState *top,
                            BlockDriverState *base,
                            int64_t sector_num,
                            int nb_sectors, int *pnum)
3883 3884 3885 3886 3887 3888 3889
{
    BlockDriverState *intermediate;
    int ret, n = nb_sectors;

    intermediate = top;
    while (intermediate && intermediate != base) {
        int pnum_inter;
3890 3891
        ret = bdrv_is_allocated(intermediate, sector_num, nb_sectors,
                                &pnum_inter);
3892 3893 3894 3895 3896 3897 3898 3899 3900 3901 3902 3903 3904
        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.
         */
3905 3906 3907
        if (n > pnum_inter &&
            (intermediate == top ||
             sector_num + pnum_inter < intermediate->total_sectors)) {
3908 3909 3910 3911 3912 3913 3914 3915 3916 3917
            n = pnum_inter;
        }

        intermediate = intermediate->backing_hd;
    }

    *pnum = n;
    return 0;
}

3918 3919 3920 3921 3922 3923 3924 3925 3926 3927
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;
}

3928
void bdrv_get_backing_filename(BlockDriverState *bs,
B
bellard 已提交
3929 3930
                               char *filename, int filename_size)
{
K
Kevin Wolf 已提交
3931
    pstrcpy(filename, filename_size, bs->backing_file);
B
bellard 已提交
3932 3933
}

3934
int bdrv_write_compressed(BlockDriverState *bs, int64_t sector_num,
B
bellard 已提交
3935 3936 3937 3938
                          const uint8_t *buf, int nb_sectors)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
3939
        return -ENOMEDIUM;
B
bellard 已提交
3940 3941
    if (!drv->bdrv_write_compressed)
        return -ENOTSUP;
K
Kevin Wolf 已提交
3942 3943
    if (bdrv_check_request(bs, sector_num, nb_sectors))
        return -EIO;
3944

F
Fam Zheng 已提交
3945
    assert(QLIST_EMPTY(&bs->dirty_bitmaps));
3946

B
bellard 已提交
3947 3948
    return drv->bdrv_write_compressed(bs, sector_num, buf, nb_sectors);
}
3949

B
bellard 已提交
3950 3951 3952 3953
int bdrv_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
3954
        return -ENOMEDIUM;
B
bellard 已提交
3955 3956 3957 3958 3959 3960
    if (!drv->bdrv_get_info)
        return -ENOTSUP;
    memset(bdi, 0, sizeof(*bdi));
    return drv->bdrv_get_info(bs, bdi);
}

M
Max Reitz 已提交
3961 3962 3963 3964 3965 3966 3967 3968 3969
ImageInfoSpecific *bdrv_get_specific_info(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;
    if (drv && drv->bdrv_get_specific_info) {
        return drv->bdrv_get_specific_info(bs);
    }
    return NULL;
}

3970 3971
int bdrv_save_vmstate(BlockDriverState *bs, const uint8_t *buf,
                      int64_t pos, int size)
3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983
{
    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)
3984 3985
{
    BlockDriver *drv = bs->drv;
3986 3987

    if (!drv) {
3988
        return -ENOMEDIUM;
3989 3990 3991 3992 3993 3994
    } 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);
    }

3995
    return -ENOTSUP;
3996 3997
}

3998 3999
int bdrv_load_vmstate(BlockDriverState *bs, uint8_t *buf,
                      int64_t pos, int size)
4000 4001 4002 4003
{
    BlockDriver *drv = bs->drv;
    if (!drv)
        return -ENOMEDIUM;
4004 4005 4006 4007 4008
    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;
4009 4010
}

K
Kevin Wolf 已提交
4011 4012
void bdrv_debug_event(BlockDriverState *bs, BlkDebugEvent event)
{
4013
    if (!bs || !bs->drv || !bs->drv->bdrv_debug_event) {
K
Kevin Wolf 已提交
4014 4015 4016
        return;
    }

4017
    bs->drv->bdrv_debug_event(bs, event);
K
Kevin Wolf 已提交
4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032 4033
}

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

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

    return -ENOTSUP;
}

F
Fam Zheng 已提交
4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044 4045 4046
int bdrv_debug_remove_breakpoint(BlockDriverState *bs, const char *tag)
{
    while (bs && bs->drv && !bs->drv->bdrv_debug_remove_breakpoint) {
        bs = bs->file;
    }

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

    return -ENOTSUP;
}

K
Kevin Wolf 已提交
4047 4048 4049 4050 4051
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
4052

K
Kevin Wolf 已提交
4053 4054 4055 4056 4057 4058 4059 4060 4061 4062 4063 4064 4065 4066 4067 4068 4069 4070
    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
4071 4072
}

4073 4074 4075 4076 4077
int bdrv_is_snapshot(BlockDriverState *bs)
{
    return !!(bs->open_flags & BDRV_O_SNAPSHOT);
}

4078 4079 4080 4081
/* 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. */
4082 4083 4084
BlockDriverState *bdrv_find_backing_image(BlockDriverState *bs,
        const char *backing_file)
{
4085 4086 4087 4088 4089 4090 4091 4092
    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) {
4093 4094 4095
        return NULL;
    }

4096 4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110
    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;
            }
4111
        } else {
4112 4113 4114 4115 4116 4117 4118 4119 4120 4121 4122 4123 4124 4125 4126 4127 4128 4129 4130 4131 4132 4133 4134
            /* 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;
            }
4135 4136 4137
        }
    }

4138 4139 4140 4141
    g_free(filename_full);
    g_free(backing_file_full);
    g_free(filename_tmp);
    return retval;
4142 4143
}

4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156
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);
}

4157 4158 4159 4160 4161 4162 4163 4164 4165 4166 4167 4168 4169 4170 4171 4172
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
4173
/**************************************************************/
B
bellard 已提交
4174
/* async I/Os */
B
bellard 已提交
4175

4176
BlockDriverAIOCB *bdrv_aio_readv(BlockDriverState *bs, int64_t sector_num,
4177
                                 QEMUIOVector *qiov, int nb_sectors,
4178
                                 BlockDriverCompletionFunc *cb, void *opaque)
B
bellard 已提交
4179
{
4180 4181
    trace_bdrv_aio_readv(bs, sector_num, nb_sectors, opaque);

P
Paolo Bonzini 已提交
4182
    return bdrv_co_aio_rw_vector(bs, sector_num, qiov, nb_sectors, 0,
4183
                                 cb, opaque, false);
B
bellard 已提交
4184 4185
}

4186 4187 4188
BlockDriverAIOCB *bdrv_aio_writev(BlockDriverState *bs, int64_t sector_num,
                                  QEMUIOVector *qiov, int nb_sectors,
                                  BlockDriverCompletionFunc *cb, void *opaque)
B
bellard 已提交
4189
{
4190 4191
    trace_bdrv_aio_writev(bs, sector_num, nb_sectors, opaque);

P
Paolo Bonzini 已提交
4192
    return bdrv_co_aio_rw_vector(bs, sector_num, qiov, nb_sectors, 0,
4193
                                 cb, opaque, true);
B
bellard 已提交
4194 4195
}

P
Paolo Bonzini 已提交
4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206
BlockDriverAIOCB *bdrv_aio_write_zeroes(BlockDriverState *bs,
        int64_t sector_num, int nb_sectors, BdrvRequestFlags flags,
        BlockDriverCompletionFunc *cb, void *opaque)
{
    trace_bdrv_aio_write_zeroes(bs, sector_num, nb_sectors, flags, opaque);

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

K
Kevin Wolf 已提交
4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219 4220 4221 4222 4223 4224

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);
4225 4226 4227
        if (mcb->callbacks[i].free_qiov) {
            qemu_iovec_destroy(mcb->callbacks[i].free_qiov);
        }
4228
        g_free(mcb->callbacks[i].free_qiov);
K
Kevin Wolf 已提交
4229 4230 4231 4232 4233 4234 4235
    }
}

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

4236 4237
    trace_multiwrite_cb(mcb, ret);

4238
    if (ret < 0 && !mcb->error) {
K
Kevin Wolf 已提交
4239 4240 4241 4242 4243
        mcb->error = ret;
    }

    mcb->num_requests--;
    if (mcb->num_requests == 0) {
4244
        multiwrite_user_cb(mcb);
4245
        g_free(mcb);
K
Kevin Wolf 已提交
4246 4247 4248 4249 4250
    }
}

static int multiwrite_req_compare(const void *a, const void *b)
{
4251 4252 4253 4254 4255 4256 4257 4258 4259 4260 4261 4262 4263
    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
4264 4265 4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284
}

/*
 * 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;

4285
        // Handle exactly sequential writes and overlapping writes.
K
Kevin Wolf 已提交
4286 4287 4288 4289
        if (reqs[i].sector <= oldreq_last) {
            merge = 1;
        }

4290 4291 4292 4293
        if (reqs[outidx].qiov->niov + reqs[i].qiov->niov + 1 > IOV_MAX) {
            merge = 0;
        }

K
Kevin Wolf 已提交
4294 4295
        if (merge) {
            size_t size;
4296
            QEMUIOVector *qiov = g_malloc0(sizeof(*qiov));
K
Kevin Wolf 已提交
4297 4298 4299 4300 4301 4302
            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;
4303
            qemu_iovec_concat(qiov, reqs[outidx].qiov, 0, size);
K
Kevin Wolf 已提交
4304

4305 4306
            // We should need to add any zeros between the two requests
            assert (reqs[i].sector <= oldreq_last);
K
Kevin Wolf 已提交
4307 4308

            // Add the second request
4309
            qemu_iovec_concat(qiov, reqs[i].qiov, 0, reqs[i].qiov->size);
K
Kevin Wolf 已提交
4310

4311
            reqs[outidx].nb_sectors = qiov->size >> 9;
K
Kevin Wolf 已提交
4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344
            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;

4345 4346 4347 4348 4349 4350 4351 4352
    /* 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
4353 4354 4355 4356 4357
    if (num_reqs == 0) {
        return 0;
    }

    // Create MultiwriteCB structure
4358
    mcb = g_malloc0(sizeof(*mcb) + num_reqs * sizeof(*mcb->callbacks));
K
Kevin Wolf 已提交
4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369
    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);

4370 4371
    trace_bdrv_aio_multiwrite(mcb, mcb->num_callbacks, num_reqs);

4372 4373
    /* Run the aio requests. */
    mcb->num_requests = num_reqs;
K
Kevin Wolf 已提交
4374
    for (i = 0; i < num_reqs; i++) {
P
Paolo Bonzini 已提交
4375 4376 4377 4378
        bdrv_co_aio_rw_vector(bs, reqs[i].sector, reqs[i].qiov,
                              reqs[i].nb_sectors, reqs[i].flags,
                              multiwrite_cb, mcb,
                              true);
K
Kevin Wolf 已提交
4379 4380 4381 4382 4383
    }

    return 0;
}

B
bellard 已提交
4384 4385
void bdrv_aio_cancel(BlockDriverAIOCB *acb)
{
S
Stefan Hajnoczi 已提交
4386
    acb->aiocb_info->cancel(acb);
B
bellard 已提交
4387 4388 4389 4390 4391
}

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

4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403
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)
{
4404 4405
    BlockDriverAIOCBSync *acb =
        container_of(blockacb, BlockDriverAIOCBSync, common);
4406
    qemu_bh_delete(acb->bh);
A
Avi Kivity 已提交
4407
    acb->bh = NULL;
4408 4409 4410
    qemu_aio_release(acb);
}

S
Stefan Hajnoczi 已提交
4411
static const AIOCBInfo bdrv_em_aiocb_info = {
4412 4413 4414 4415
    .aiocb_size         = sizeof(BlockDriverAIOCBSync),
    .cancel             = bdrv_aio_cancel_em,
};

4416
static void bdrv_aio_bh_cb(void *opaque)
B
bellard 已提交
4417
{
4418
    BlockDriverAIOCBSync *acb = opaque;
4419 4420

    if (!acb->is_write)
4421
        qemu_iovec_from_buf(acb->qiov, 0, acb->bounce, acb->qiov->size);
4422
    qemu_vfree(acb->bounce);
4423
    acb->common.cb(acb->common.opaque, acb->ret);
4424
    qemu_bh_delete(acb->bh);
A
Avi Kivity 已提交
4425
    acb->bh = NULL;
4426
    qemu_aio_release(acb);
B
bellard 已提交
4427
}
4428

4429 4430 4431 4432 4433 4434 4435 4436
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
4437
{
4438 4439
    BlockDriverAIOCBSync *acb;

S
Stefan Hajnoczi 已提交
4440
    acb = qemu_aio_get(&bdrv_em_aiocb_info, bs, cb, opaque);
4441 4442
    acb->is_write = is_write;
    acb->qiov = qiov;
4443
    acb->bounce = qemu_blockalign(bs, qiov->size);
4444
    acb->bh = qemu_bh_new(bdrv_aio_bh_cb, acb);
4445 4446

    if (is_write) {
4447
        qemu_iovec_to_buf(acb->qiov, 0, acb->bounce, qiov->size);
4448
        acb->ret = bs->drv->bdrv_write(bs, sector_num, acb->bounce, nb_sectors);
4449
    } else {
4450
        acb->ret = bs->drv->bdrv_read(bs, sector_num, acb->bounce, nb_sectors);
4451 4452
    }

4453
    qemu_bh_schedule(acb->bh);
4454

4455
    return &acb->common;
4456 4457
}

4458 4459
static BlockDriverAIOCB *bdrv_aio_readv_em(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
4460
        BlockDriverCompletionFunc *cb, void *opaque)
4461
{
4462 4463
    return bdrv_aio_rw_vector(bs, sector_num, qiov, nb_sectors, cb, opaque, 0);
}
B
bellard 已提交
4464

4465 4466 4467 4468 4469
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);
4470 4471
}

4472 4473 4474 4475 4476

typedef struct BlockDriverAIOCBCoroutine {
    BlockDriverAIOCB common;
    BlockRequest req;
    bool is_write;
K
Kevin Wolf 已提交
4477
    bool *done;
4478 4479 4480 4481 4482
    QEMUBH* bh;
} BlockDriverAIOCBCoroutine;

static void bdrv_aio_co_cancel_em(BlockDriverAIOCB *blockacb)
{
K
Kevin Wolf 已提交
4483 4484 4485 4486 4487 4488 4489 4490
    BlockDriverAIOCBCoroutine *acb =
        container_of(blockacb, BlockDriverAIOCBCoroutine, common);
    bool done = false;

    acb->done = &done;
    while (!done) {
        qemu_aio_wait();
    }
4491 4492
}

S
Stefan Hajnoczi 已提交
4493
static const AIOCBInfo bdrv_em_co_aiocb_info = {
4494 4495 4496 4497
    .aiocb_size         = sizeof(BlockDriverAIOCBCoroutine),
    .cancel             = bdrv_aio_co_cancel_em,
};

P
Paolo Bonzini 已提交
4498
static void bdrv_co_em_bh(void *opaque)
4499 4500 4501 4502
{
    BlockDriverAIOCBCoroutine *acb = opaque;

    acb->common.cb(acb->common.opaque, acb->req.error);
K
Kevin Wolf 已提交
4503 4504 4505 4506 4507

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

4508 4509 4510 4511
    qemu_bh_delete(acb->bh);
    qemu_aio_release(acb);
}

4512 4513 4514 4515 4516 4517 4518 4519
/* Invoke bdrv_co_do_readv/bdrv_co_do_writev */
static void coroutine_fn bdrv_co_do_rw(void *opaque)
{
    BlockDriverAIOCBCoroutine *acb = opaque;
    BlockDriverState *bs = acb->common.bs;

    if (!acb->is_write) {
        acb->req.error = bdrv_co_do_readv(bs, acb->req.sector,
P
Paolo Bonzini 已提交
4520
            acb->req.nb_sectors, acb->req.qiov, acb->req.flags);
4521 4522
    } else {
        acb->req.error = bdrv_co_do_writev(bs, acb->req.sector,
P
Paolo Bonzini 已提交
4523
            acb->req.nb_sectors, acb->req.qiov, acb->req.flags);
4524 4525
    }

P
Paolo Bonzini 已提交
4526
    acb->bh = qemu_bh_new(bdrv_co_em_bh, acb);
4527 4528 4529
    qemu_bh_schedule(acb->bh);
}

4530 4531 4532 4533
static BlockDriverAIOCB *bdrv_co_aio_rw_vector(BlockDriverState *bs,
                                               int64_t sector_num,
                                               QEMUIOVector *qiov,
                                               int nb_sectors,
P
Paolo Bonzini 已提交
4534
                                               BdrvRequestFlags flags,
4535 4536
                                               BlockDriverCompletionFunc *cb,
                                               void *opaque,
4537
                                               bool is_write)
4538 4539 4540 4541
{
    Coroutine *co;
    BlockDriverAIOCBCoroutine *acb;

S
Stefan Hajnoczi 已提交
4542
    acb = qemu_aio_get(&bdrv_em_co_aiocb_info, bs, cb, opaque);
4543 4544 4545
    acb->req.sector = sector_num;
    acb->req.nb_sectors = nb_sectors;
    acb->req.qiov = qiov;
P
Paolo Bonzini 已提交
4546
    acb->req.flags = flags;
4547
    acb->is_write = is_write;
K
Kevin Wolf 已提交
4548
    acb->done = NULL;
4549

4550
    co = qemu_coroutine_create(bdrv_co_do_rw);
4551 4552 4553 4554 4555
    qemu_coroutine_enter(co, acb);

    return &acb->common;
}

P
Paolo Bonzini 已提交
4556
static void coroutine_fn bdrv_aio_flush_co_entry(void *opaque)
4557
{
P
Paolo Bonzini 已提交
4558 4559
    BlockDriverAIOCBCoroutine *acb = opaque;
    BlockDriverState *bs = acb->common.bs;
4560

P
Paolo Bonzini 已提交
4561 4562
    acb->req.error = bdrv_co_flush(bs);
    acb->bh = qemu_bh_new(bdrv_co_em_bh, acb);
4563 4564 4565
    qemu_bh_schedule(acb->bh);
}

P
Paolo Bonzini 已提交
4566
BlockDriverAIOCB *bdrv_aio_flush(BlockDriverState *bs,
4567 4568
        BlockDriverCompletionFunc *cb, void *opaque)
{
P
Paolo Bonzini 已提交
4569
    trace_bdrv_aio_flush(bs, opaque);
4570

P
Paolo Bonzini 已提交
4571 4572
    Coroutine *co;
    BlockDriverAIOCBCoroutine *acb;
4573

S
Stefan Hajnoczi 已提交
4574
    acb = qemu_aio_get(&bdrv_em_co_aiocb_info, bs, cb, opaque);
K
Kevin Wolf 已提交
4575 4576
    acb->done = NULL;

P
Paolo Bonzini 已提交
4577 4578
    co = qemu_coroutine_create(bdrv_aio_flush_co_entry);
    qemu_coroutine_enter(co, acb);
4579 4580 4581 4582

    return &acb->common;
}

4583 4584 4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601
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
4602
    acb = qemu_aio_get(&bdrv_em_co_aiocb_info, bs, cb, opaque);
4603 4604
    acb->req.sector = sector_num;
    acb->req.nb_sectors = nb_sectors;
K
Kevin Wolf 已提交
4605
    acb->done = NULL;
4606 4607 4608 4609 4610 4611
    co = qemu_coroutine_create(bdrv_aio_discard_co_entry);
    qemu_coroutine_enter(co, acb);

    return &acb->common;
}

B
bellard 已提交
4612 4613
void bdrv_init(void)
{
4614
    module_call_init(MODULE_INIT_BLOCK);
B
bellard 已提交
4615
}
4616

4617 4618 4619 4620 4621 4622
void bdrv_init_with_whitelist(void)
{
    use_bdrv_whitelist = 1;
    bdrv_init();
}

S
Stefan Hajnoczi 已提交
4623
void *qemu_aio_get(const AIOCBInfo *aiocb_info, BlockDriverState *bs,
4624
                   BlockDriverCompletionFunc *cb, void *opaque)
4625 4626 4627
{
    BlockDriverAIOCB *acb;

S
Stefan Hajnoczi 已提交
4628 4629
    acb = g_slice_alloc(aiocb_info->aiocb_size);
    acb->aiocb_info = aiocb_info;
4630 4631 4632 4633 4634 4635 4636 4637
    acb->bs = bs;
    acb->cb = cb;
    acb->opaque = opaque;
    return acb;
}

void qemu_aio_release(void *p)
{
4638
    BlockDriverAIOCB *acb = p;
S
Stefan Hajnoczi 已提交
4639
    g_slice_free1(acb->aiocb_info->aiocb_size, acb);
4640
}
B
bellard 已提交
4641

4642 4643 4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655 4656 4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667
/**************************************************************/
/* 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) {
4668 4669
        acb = bs->drv->bdrv_aio_writev(bs, sector_num, iov, nb_sectors,
                                       bdrv_co_io_em_complete, &co);
4670
    } else {
4671 4672
        acb = bs->drv->bdrv_aio_readv(bs, sector_num, iov, nb_sectors,
                                      bdrv_co_io_em_complete, &co);
4673 4674
    }

4675
    trace_bdrv_co_io_em(bs, sector_num, nb_sectors, is_write, acb);
4676 4677 4678 4679 4680 4681 4682 4683 4684 4685 4686 4687 4688 4689 4690 4691 4692 4693 4694 4695 4696 4697
    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
4698
static void coroutine_fn bdrv_flush_co_entry(void *opaque)
4699
{
P
Paolo Bonzini 已提交
4700 4701 4702 4703 4704 4705 4706
    RwCo *rwco = opaque;

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

int coroutine_fn bdrv_co_flush(BlockDriverState *bs)
{
K
Kevin Wolf 已提交
4707 4708
    int ret;

4709
    if (!bs || !bdrv_is_inserted(bs) || bdrv_is_read_only(bs)) {
P
Paolo Bonzini 已提交
4710
        return 0;
K
Kevin Wolf 已提交
4711 4712
    }

4713
    /* Write back cached data to the OS even with cache=unsafe */
4714
    BLKDBG_EVENT(bs->file, BLKDBG_FLUSH_TO_OS);
K
Kevin Wolf 已提交
4715 4716 4717 4718 4719 4720 4721
    if (bs->drv->bdrv_co_flush_to_os) {
        ret = bs->drv->bdrv_co_flush_to_os(bs);
        if (ret < 0) {
            return ret;
        }
    }

4722 4723
    /* But don't actually force it to the disk with cache=unsafe */
    if (bs->open_flags & BDRV_O_NO_FLUSH) {
4724
        goto flush_parent;
4725 4726
    }

4727
    BLKDBG_EVENT(bs->file, BLKDBG_FLUSH_TO_DISK);
K
Kevin Wolf 已提交
4728
    if (bs->drv->bdrv_co_flush_to_disk) {
4729
        ret = bs->drv->bdrv_co_flush_to_disk(bs);
P
Paolo Bonzini 已提交
4730 4731 4732 4733 4734 4735 4736 4737
    } 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) {
4738
            ret = -EIO;
P
Paolo Bonzini 已提交
4739 4740
        } else {
            qemu_coroutine_yield();
4741
            ret = co.ret;
P
Paolo Bonzini 已提交
4742 4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754
        }
    } 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.
         */
4755
        ret = 0;
P
Paolo Bonzini 已提交
4756
    }
4757 4758 4759 4760 4761 4762 4763
    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.
     */
4764
flush_parent:
4765
    return bdrv_co_flush(bs->file);
P
Paolo Bonzini 已提交
4766 4767
}

4768 4769 4770 4771 4772 4773 4774 4775 4776 4777 4778
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;

4779
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
4780 4781 4782 4783
        bdrv_invalidate_cache(bs);
    }
}

4784 4785 4786 4787
void bdrv_clear_incoming_migration_all(void)
{
    BlockDriverState *bs;

4788
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
4789 4790 4791 4792
        bs->open_flags = bs->open_flags & ~(BDRV_O_INCOMING);
    }
}

P
Paolo Bonzini 已提交
4793 4794 4795 4796 4797 4798
int bdrv_flush(BlockDriverState *bs)
{
    Coroutine *co;
    RwCo rwco = {
        .bs = bs,
        .ret = NOT_DONE,
4799 4800
    };

P
Paolo Bonzini 已提交
4801 4802 4803 4804 4805 4806 4807 4808 4809
    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();
        }
4810
    }
P
Paolo Bonzini 已提交
4811 4812

    return rwco.ret;
4813 4814
}

4815 4816 4817 4818 4819 4820
typedef struct DiscardCo {
    BlockDriverState *bs;
    int64_t sector_num;
    int nb_sectors;
    int ret;
} DiscardCo;
4821 4822
static void coroutine_fn bdrv_discard_co_entry(void *opaque)
{
4823
    DiscardCo *rwco = opaque;
4824 4825 4826 4827

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

4828 4829 4830 4831 4832
/* if no limit is specified in the BlockLimits use a default
 * of 32768 512-byte sectors (16 MiB) per request.
 */
#define MAX_DISCARD_DEFAULT 32768

4833 4834 4835
int coroutine_fn bdrv_co_discard(BlockDriverState *bs, int64_t sector_num,
                                 int nb_sectors)
{
4836 4837
    int max_discard;

4838 4839 4840 4841 4842 4843
    if (!bs->drv) {
        return -ENOMEDIUM;
    } else if (bdrv_check_request(bs, sector_num, nb_sectors)) {
        return -EIO;
    } else if (bs->read_only) {
        return -EROFS;
4844 4845
    }

F
Fam Zheng 已提交
4846
    bdrv_reset_dirty(bs, sector_num, nb_sectors);
4847

P
Paolo Bonzini 已提交
4848 4849 4850 4851 4852
    /* Do nothing if disabled.  */
    if (!(bs->open_flags & BDRV_O_UNMAP)) {
        return 0;
    }

4853 4854 4855
    if (!bs->drv->bdrv_co_discard && !bs->drv->bdrv_aio_discard) {
        return 0;
    }
4856

4857 4858 4859 4860
    max_discard = bs->bl.max_discard ?  bs->bl.max_discard : MAX_DISCARD_DEFAULT;
    while (nb_sectors > 0) {
        int ret;
        int num = nb_sectors;
4861

4862 4863 4864 4865 4866 4867
        /* align request */
        if (bs->bl.discard_alignment &&
            num >= bs->bl.discard_alignment &&
            sector_num % bs->bl.discard_alignment) {
            if (num > bs->bl.discard_alignment) {
                num = bs->bl.discard_alignment;
4868
            }
4869 4870
            num -= sector_num % bs->bl.discard_alignment;
        }
4871

4872 4873 4874 4875
        /* limit request size */
        if (num > max_discard) {
            num = max_discard;
        }
4876

4877
        if (bs->drv->bdrv_co_discard) {
4878
            ret = bs->drv->bdrv_co_discard(bs, sector_num, num);
4879 4880 4881 4882 4883 4884 4885 4886 4887 4888 4889 4890 4891
        } else {
            BlockDriverAIOCB *acb;
            CoroutineIOCompletion co = {
                .coroutine = qemu_coroutine_self(),
            };

            acb = bs->drv->bdrv_aio_discard(bs, sector_num, nb_sectors,
                                            bdrv_co_io_em_complete, &co);
            if (acb == NULL) {
                return -EIO;
            } else {
                qemu_coroutine_yield();
                ret = co.ret;
4892 4893
            }
        }
4894
        if (ret && ret != -ENOTSUP) {
4895
            return ret;
4896
        }
4897 4898 4899

        sector_num += num;
        nb_sectors -= num;
4900
    }
4901
    return 0;
4902 4903 4904 4905 4906
}

int bdrv_discard(BlockDriverState *bs, int64_t sector_num, int nb_sectors)
{
    Coroutine *co;
4907
    DiscardCo rwco = {
4908 4909 4910 4911 4912 4913 4914 4915 4916 4917 4918 4919 4920 4921 4922 4923 4924 4925 4926 4927
        .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
4928 4929 4930 4931 4932 4933 4934 4935 4936
/**************************************************************/
/* removable device support */

/**
 * Return TRUE if the media is present
 */
int bdrv_is_inserted(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;
4937

B
bellard 已提交
4938 4939 4940
    if (!drv)
        return 0;
    if (!drv->bdrv_is_inserted)
4941 4942
        return 1;
    return drv->bdrv_is_inserted(bs);
B
bellard 已提交
4943 4944 4945
}

/**
4946 4947
 * 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
4948 4949 4950 4951 4952
 */
int bdrv_media_changed(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;

4953 4954 4955 4956
    if (drv && drv->bdrv_media_changed) {
        return drv->bdrv_media_changed(bs);
    }
    return -ENOTSUP;
B
bellard 已提交
4957 4958 4959 4960 4961
}

/**
 * If eject_flag is TRUE, eject the media. Otherwise, close the tray
 */
4962
void bdrv_eject(BlockDriverState *bs, bool eject_flag)
B
bellard 已提交
4963 4964 4965
{
    BlockDriver *drv = bs->drv;

4966 4967
    if (drv && drv->bdrv_eject) {
        drv->bdrv_eject(bs, eject_flag);
B
bellard 已提交
4968
    }
4969 4970 4971 4972

    if (bs->device_name[0] != '\0') {
        bdrv_emit_qmp_eject_event(bs, eject_flag);
    }
B
bellard 已提交
4973 4974 4975 4976 4977 4978
}

/**
 * Lock or unlock the media (if it is locked, the user won't be able
 * to eject it manually).
 */
4979
void bdrv_lock_medium(BlockDriverState *bs, bool locked)
B
bellard 已提交
4980 4981 4982
{
    BlockDriver *drv = bs->drv;

4983
    trace_bdrv_lock_medium(bs, locked);
S
Stefan Hajnoczi 已提交
4984

4985 4986
    if (drv && drv->bdrv_lock_medium) {
        drv->bdrv_lock_medium(bs, locked);
B
bellard 已提交
4987 4988
    }
}
4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999

/* 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;
}
5000

5001 5002 5003
BlockDriverAIOCB *bdrv_aio_ioctl(BlockDriverState *bs,
        unsigned long int req, void *buf,
        BlockDriverCompletionFunc *cb, void *opaque)
5004
{
5005
    BlockDriver *drv = bs->drv;
5006

5007 5008 5009
    if (drv && drv->bdrv_aio_ioctl)
        return drv->bdrv_aio_ioctl(bs, req, buf, cb, opaque);
    return NULL;
5010
}
5011

5012
void bdrv_set_guest_block_size(BlockDriverState *bs, int align)
5013
{
5014
    bs->guest_block_size = align;
5015
}
5016

5017 5018
void *qemu_blockalign(BlockDriverState *bs, size_t size)
{
5019
    return qemu_memalign(bdrv_opt_mem_align(bs), size);
5020
}
5021

5022 5023 5024 5025 5026 5027
/*
 * Check if all memory in this vector is sector aligned.
 */
bool bdrv_qiov_is_aligned(BlockDriverState *bs, QEMUIOVector *qiov)
{
    int i;
5028
    size_t alignment = bdrv_opt_mem_align(bs);
5029 5030

    for (i = 0; i < qiov->niov; i++) {
5031
        if ((uintptr_t) qiov->iov[i].iov_base % alignment) {
5032
            return false;
5033
        }
5034
        if (qiov->iov[i].iov_len % alignment) {
5035
            return false;
5036 5037 5038 5039 5040 5041
        }
    }

    return true;
}

F
Fam Zheng 已提交
5042
BdrvDirtyBitmap *bdrv_create_dirty_bitmap(BlockDriverState *bs, int granularity)
5043 5044
{
    int64_t bitmap_size;
F
Fam Zheng 已提交
5045
    BdrvDirtyBitmap *bitmap;
5046

5047 5048
    assert((granularity & (granularity - 1)) == 0);

F
Fam Zheng 已提交
5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059 5060 5061 5062 5063 5064 5065 5066
    granularity >>= BDRV_SECTOR_BITS;
    assert(granularity);
    bitmap_size = (bdrv_getlength(bs) >> BDRV_SECTOR_BITS);
    bitmap = g_malloc0(sizeof(BdrvDirtyBitmap));
    bitmap->bitmap = hbitmap_alloc(bitmap_size, ffs(granularity) - 1);
    QLIST_INSERT_HEAD(&bs->dirty_bitmaps, bitmap, list);
    return bitmap;
}

void bdrv_release_dirty_bitmap(BlockDriverState *bs, BdrvDirtyBitmap *bitmap)
{
    BdrvDirtyBitmap *bm, *next;
    QLIST_FOREACH_SAFE(bm, &bs->dirty_bitmaps, list, next) {
        if (bm == bitmap) {
            QLIST_REMOVE(bitmap, list);
            hbitmap_free(bitmap->bitmap);
            g_free(bitmap);
            return;
5067
        }
5068 5069 5070
    }
}

F
Fam Zheng 已提交
5071 5072 5073 5074 5075 5076 5077 5078 5079 5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090
BlockDirtyInfoList *bdrv_query_dirty_bitmaps(BlockDriverState *bs)
{
    BdrvDirtyBitmap *bm;
    BlockDirtyInfoList *list = NULL;
    BlockDirtyInfoList **plist = &list;

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

    return list;
}

F
Fam Zheng 已提交
5091
int bdrv_get_dirty(BlockDriverState *bs, BdrvDirtyBitmap *bitmap, int64_t sector)
5092
{
F
Fam Zheng 已提交
5093 5094
    if (bitmap) {
        return hbitmap_get(bitmap->bitmap, sector);
5095 5096 5097 5098 5099
    } else {
        return 0;
    }
}

F
Fam Zheng 已提交
5100 5101
void bdrv_dirty_iter_init(BlockDriverState *bs,
                          BdrvDirtyBitmap *bitmap, HBitmapIter *hbi)
5102
{
F
Fam Zheng 已提交
5103
    hbitmap_iter_init(hbi, bitmap->bitmap, 0);
5104 5105 5106 5107 5108
}

void bdrv_set_dirty(BlockDriverState *bs, int64_t cur_sector,
                    int nr_sectors)
{
F
Fam Zheng 已提交
5109 5110 5111 5112
    BdrvDirtyBitmap *bitmap;
    QLIST_FOREACH(bitmap, &bs->dirty_bitmaps, list) {
        hbitmap_set(bitmap->bitmap, cur_sector, nr_sectors);
    }
5113 5114
}

F
Fam Zheng 已提交
5115
void bdrv_reset_dirty(BlockDriverState *bs, int64_t cur_sector, int nr_sectors)
5116
{
F
Fam Zheng 已提交
5117 5118 5119 5120
    BdrvDirtyBitmap *bitmap;
    QLIST_FOREACH(bitmap, &bs->dirty_bitmaps, list) {
        hbitmap_reset(bitmap->bitmap, cur_sector, nr_sectors);
    }
5121
}
5122

F
Fam Zheng 已提交
5123
int64_t bdrv_get_dirty_count(BlockDriverState *bs, BdrvDirtyBitmap *bitmap)
5124
{
F
Fam Zheng 已提交
5125
    return hbitmap_count(bitmap->bitmap);
5126
}
J
Jes Sorensen 已提交
5127

5128 5129 5130 5131 5132 5133 5134 5135 5136 5137 5138 5139 5140 5141 5142 5143 5144
/* 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);
    }
}

5145 5146 5147 5148 5149 5150 5151 5152 5153 5154 5155
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;
}

5156 5157
void bdrv_iostatus_enable(BlockDriverState *bs)
{
5158
    bs->iostatus_enabled = true;
5159
    bs->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
5160 5161 5162 5163 5164 5165
}

/* 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)
{
5166
    return (bs->iostatus_enabled &&
5167 5168 5169
           (bs->on_write_error == BLOCKDEV_ON_ERROR_ENOSPC ||
            bs->on_write_error == BLOCKDEV_ON_ERROR_STOP   ||
            bs->on_read_error == BLOCKDEV_ON_ERROR_STOP));
5170 5171 5172 5173
}

void bdrv_iostatus_disable(BlockDriverState *bs)
{
5174
    bs->iostatus_enabled = false;
5175 5176 5177 5178 5179
}

void bdrv_iostatus_reset(BlockDriverState *bs)
{
    if (bdrv_iostatus_is_enabled(bs)) {
5180
        bs->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
5181 5182 5183
        if (bs->job) {
            block_job_iostatus_reset(bs->job);
        }
5184 5185 5186 5187 5188
    }
}

void bdrv_iostatus_set_err(BlockDriverState *bs, int error)
{
5189 5190
    assert(bdrv_iostatus_is_enabled(bs));
    if (bs->iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
5191 5192
        bs->iostatus = error == ENOSPC ? BLOCK_DEVICE_IO_STATUS_NOSPACE :
                                         BLOCK_DEVICE_IO_STATUS_FAILED;
5193 5194 5195
    }
}

5196 5197 5198 5199 5200 5201 5202
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
5203
    cookie->start_time_ns = get_clock();
5204 5205 5206 5207 5208 5209 5210 5211 5212 5213
    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
5214
    bs->total_time_ns[cookie->type] += get_clock() - cookie->start_time_ns;
5215 5216
}

5217 5218
void bdrv_img_create(const char *filename, const char *fmt,
                     const char *base_filename, const char *base_fmt,
5219 5220
                     char *options, uint64_t img_size, int flags,
                     Error **errp, bool quiet)
J
Jes Sorensen 已提交
5221 5222
{
    QEMUOptionParameter *param = NULL, *create_options = NULL;
K
Kevin Wolf 已提交
5223
    QEMUOptionParameter *backing_fmt, *backing_file, *size;
J
Jes Sorensen 已提交
5224
    BlockDriver *drv, *proto_drv;
5225
    BlockDriver *backing_drv = NULL;
5226
    Error *local_err = NULL;
J
Jes Sorensen 已提交
5227 5228 5229 5230 5231
    int ret = 0;

    /* Find driver and parse its options */
    drv = bdrv_find_format(fmt);
    if (!drv) {
5232
        error_setg(errp, "Unknown file format '%s'", fmt);
5233
        return;
J
Jes Sorensen 已提交
5234 5235
    }

5236
    proto_drv = bdrv_find_protocol(filename, true);
J
Jes Sorensen 已提交
5237
    if (!proto_drv) {
5238
        error_setg(errp, "Unknown protocol '%s'", filename);
5239
        return;
J
Jes Sorensen 已提交
5240 5241 5242 5243 5244 5245 5246 5247 5248 5249 5250 5251 5252 5253 5254 5255
    }

    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) {
5256
            error_setg(errp, "Invalid options for file format '%s'.", fmt);
J
Jes Sorensen 已提交
5257 5258 5259 5260 5261 5262 5263
            goto out;
        }
    }

    if (base_filename) {
        if (set_option_parameter(param, BLOCK_OPT_BACKING_FILE,
                                 base_filename)) {
5264 5265
            error_setg(errp, "Backing file not supported for file format '%s'",
                       fmt);
J
Jes Sorensen 已提交
5266 5267 5268 5269 5270 5271
            goto out;
        }
    }

    if (base_fmt) {
        if (set_option_parameter(param, BLOCK_OPT_BACKING_FMT, base_fmt)) {
5272 5273
            error_setg(errp, "Backing file format not supported for file "
                             "format '%s'", fmt);
J
Jes Sorensen 已提交
5274 5275 5276 5277
            goto out;
        }
    }

5278 5279 5280
    backing_file = get_option_parameter(param, BLOCK_OPT_BACKING_FILE);
    if (backing_file && backing_file->value.s) {
        if (!strcmp(filename, backing_file->value.s)) {
5281 5282
            error_setg(errp, "Error: Trying to create an image with the "
                             "same filename as the backing file");
5283 5284 5285 5286
            goto out;
        }
    }

J
Jes Sorensen 已提交
5287 5288
    backing_fmt = get_option_parameter(param, BLOCK_OPT_BACKING_FMT);
    if (backing_fmt && backing_fmt->value.s) {
5289 5290
        backing_drv = bdrv_find_format(backing_fmt->value.s);
        if (!backing_drv) {
5291 5292
            error_setg(errp, "Unknown backing file format '%s'",
                       backing_fmt->value.s);
J
Jes Sorensen 已提交
5293 5294 5295 5296 5297 5298
            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
5299 5300
    size = get_option_parameter(param, BLOCK_OPT_SIZE);
    if (size && size->value.n == -1) {
J
Jes Sorensen 已提交
5301
        if (backing_file && backing_file->value.s) {
5302
            BlockDriverState *bs;
J
Jes Sorensen 已提交
5303 5304
            uint64_t size;
            char buf[32];
5305 5306 5307 5308 5309
            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
5310

5311 5312
            bs = NULL;
            ret = bdrv_open(&bs, backing_file->value.s, NULL, back_flags,
5313
                            backing_drv, &local_err);
J
Jes Sorensen 已提交
5314
            if (ret < 0) {
5315 5316 5317 5318 5319
                error_setg_errno(errp, -ret, "Could not open '%s': %s",
                                 backing_file->value.s,
                                 error_get_pretty(local_err));
                error_free(local_err);
                local_err = NULL;
J
Jes Sorensen 已提交
5320 5321 5322 5323 5324 5325 5326
                goto out;
            }
            bdrv_get_geometry(bs, &size);
            size *= 512;

            snprintf(buf, sizeof(buf), "%" PRId64, size);
            set_option_parameter(param, BLOCK_OPT_SIZE, buf);
5327 5328

            bdrv_unref(bs);
J
Jes Sorensen 已提交
5329
        } else {
5330
            error_setg(errp, "Image creation needs a size parameter");
J
Jes Sorensen 已提交
5331 5332 5333 5334
            goto out;
        }
    }

5335 5336 5337 5338 5339
    if (!quiet) {
        printf("Formatting '%s', fmt=%s ", filename, fmt);
        print_option_parameters(param);
        puts("");
    }
5340 5341 5342 5343 5344 5345 5346 5347
    ret = bdrv_create(drv, filename, param, &local_err);
    if (ret == -EFBIG) {
        /* This is generally a better message than whatever the driver would
         * deliver (especially because of the cluster_size_hint), since that
         * is most probably not much different from "image too large". */
        const char *cluster_size_hint = "";
        if (get_option_parameter(create_options, BLOCK_OPT_CLUSTER_SIZE)) {
            cluster_size_hint = " (try using a larger cluster size)";
J
Jes Sorensen 已提交
5348
        }
5349 5350 5351 5352
        error_setg(errp, "The image size is too large for file format '%s'"
                   "%s", fmt, cluster_size_hint);
        error_free(local_err);
        local_err = NULL;
J
Jes Sorensen 已提交
5353 5354 5355 5356 5357 5358
    }

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

5359
    if (local_err) {
5360 5361
        error_propagate(errp, local_err);
    }
J
Jes Sorensen 已提交
5362
}
5363 5364 5365 5366 5367 5368

AioContext *bdrv_get_aio_context(BlockDriverState *bs)
{
    /* Currently BlockDriverState always uses the main loop AioContext */
    return qemu_get_aio_context();
}
5369 5370 5371 5372 5373 5374

void bdrv_add_before_write_notifier(BlockDriverState *bs,
                                    NotifierWithReturn *notifier)
{
    notifier_with_return_list_add(&bs->before_write_notifiers, notifier);
}
M
Max Reitz 已提交
5375 5376 5377 5378 5379 5380 5381 5382

int bdrv_amend_options(BlockDriverState *bs, QEMUOptionParameter *options)
{
    if (bs->drv->bdrv_amend_options == NULL) {
        return -ENOTSUP;
    }
    return bs->drv->bdrv_amend_options(bs, options);
}
5383

5384 5385 5386 5387 5388
/* Used to recurse on single child block filters.
 * Single child block filter will store their child in bs->file.
 */
bool bdrv_generic_is_first_non_filter(BlockDriverState *bs,
                                      BlockDriverState *candidate)
5389
{
5390 5391 5392 5393 5394 5395 5396 5397 5398 5399 5400 5401 5402 5403
    if (!bs->drv) {
        return false;
    }

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

    if (!bs->drv->authorizations[BS_FILTER_PASS_DOWN]) {
        return false;
5404 5405
    }

5406 5407 5408 5409 5410 5411 5412 5413 5414 5415 5416 5417
    if (!bs->file) {
        return false;
    }

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

bool bdrv_recurse_is_first_non_filter(BlockDriverState *bs,
                                      BlockDriverState *candidate)
{
    if (bs->drv && bs->drv->bdrv_recurse_is_first_non_filter) {
        return bs->drv->bdrv_recurse_is_first_non_filter(bs, candidate);
5418 5419
    }

5420
    return bdrv_generic_is_first_non_filter(bs, candidate);
5421 5422
}

5423 5424 5425 5426 5427
/* This function checks if the candidate is the first non filter bs down it's
 * bs chain. Since we don't have pointers to parents it explore all bs chains
 * from the top. Some filters can choose not to pass down the recursion.
 */
bool bdrv_is_first_non_filter(BlockDriverState *candidate)
5428
{
5429 5430 5431 5432 5433 5434
    BlockDriverState *bs;

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

5435
        perm = bdrv_recurse_is_first_non_filter(bs, candidate);
5436 5437 5438 5439 5440 5441 5442 5443

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

    return false;
5444
}