block.c 161.6 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 28
#include "block/block_int.h"
#include "block/blockjob.h"
29
#include "qemu/module.h"
30
#include "qapi/qmp/qjson.h"
31
#include "sysemu/sysemu.h"
32
#include "qemu/notify.h"
33
#include "block/coroutine.h"
34
#include "block/qapi.h"
L
Luiz Capitulino 已提交
35
#include "qmp-commands.h"
36
#include "qemu/timer.h"
37
#include "qapi-event.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
/* 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,
182
                  bdrv_get_aio_context(bs),
183 184 185 186 187 188 189 190 191 192 193 194
                  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
195
static void bdrv_io_limits_intercept(BlockDriverState *bs,
196
                                     unsigned int bytes,
197
                                     bool is_write)
Z
Zhi Yong Wu 已提交
198
{
199 200
    /* does this io must wait */
    bool must_wait = throttle_schedule_timer(&bs->throttle_state, is_write);
Z
Zhi Yong Wu 已提交
201

202 203 204 205
    /* 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
206 207
    }

208
    /* the IO will be executed, do the accounting */
209 210
    throttle_account(&bs->throttle_state, is_write, bytes);

Z
Zhi Yong Wu 已提交
211

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

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

221 222 223 224 225 226 227 228 229 230
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;
}

231 232 233
/* check if the path starts with "<protocol>:" */
static int path_has_protocol(const char *path)
{
234 235
    const char *p;

236 237 238 239 240
#ifdef _WIN32
    if (is_windows_drive(path) ||
        is_windows_drive_prefix(path)) {
        return 0;
    }
241 242 243
    p = path + strcspn(path, ":/\\");
#else
    p = path + strcspn(path, ":/");
244 245
#endif

246
    return *p == ':';
247 248
}

B
bellard 已提交
249
int path_is_absolute(const char *path)
250
{
B
bellard 已提交
251 252
#ifdef _WIN32
    /* specific case for names like: "\\.\d:" */
P
Paolo Bonzini 已提交
253
    if (is_windows_drive(path) || is_windows_drive_prefix(path)) {
B
bellard 已提交
254
        return 1;
P
Paolo Bonzini 已提交
255 256
    }
    return (*path == '/' || *path == '\\');
257
#else
P
Paolo Bonzini 已提交
258
    return (*path == '/');
259
#endif
260 261
}

B
bellard 已提交
262 263 264 265 266 267
/* 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)
268
{
B
bellard 已提交
269 270 271 272 273 274 275 276 277 278 279 280 281
    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;
282 283 284 285 286 287 288 289 290
        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
291 292 293 294 295 296 297 298 299 300 301 302
        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);
303 304 305
    }
}

306 307 308 309 310 311 312 313 314
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);
    }
}

315
void bdrv_register(BlockDriver *bdrv)
B
bellard 已提交
316
{
317 318
    /* Block drivers without coroutine functions need emulation */
    if (!bdrv->bdrv_co_readv) {
319 320 321
        bdrv->bdrv_co_readv = bdrv_co_readv_em;
        bdrv->bdrv_co_writev = bdrv_co_writev_em;

322 323 324
        /* 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.
         */
325 326 327 328 329
        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
330
    }
331

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

/* create a new block device (by default it is empty) */
K
Kevin Wolf 已提交
336
BlockDriverState *bdrv_new(const char *device_name, Error **errp)
B
bellard 已提交
337
{
338
    BlockDriverState *bs;
339
    int i;
B
bellard 已提交
340

341 342 343 344 345 346 347 348 349 350 351
    if (bdrv_find(device_name)) {
        error_setg(errp, "Device with id '%s' already exists",
                   device_name);
        return NULL;
    }
    if (bdrv_find_node(device_name)) {
        error_setg(errp, "Device with node-name '%s' already exists",
                   device_name);
        return NULL;
    }

352
    bs = g_malloc0(sizeof(BlockDriverState));
F
Fam Zheng 已提交
353
    QLIST_INIT(&bs->dirty_bitmaps);
B
bellard 已提交
354
    pstrcpy(bs->device_name, sizeof(bs->device_name), device_name);
B
bellard 已提交
355
    if (device_name[0] != '\0') {
356
        QTAILQ_INSERT_TAIL(&bdrv_states, bs, device_list);
B
bellard 已提交
357
    }
358 359 360
    for (i = 0; i < BLOCK_OP_TYPE_MAX; i++) {
        QLIST_INIT(&bs->op_blockers[i]);
    }
361
    bdrv_iostatus_disable(bs);
P
Paolo Bonzini 已提交
362
    notifier_list_init(&bs->close_notifiers);
363
    notifier_with_return_list_init(&bs->before_write_notifiers);
364 365
    qemu_co_queue_init(&bs->throttled_reqs[0]);
    qemu_co_queue_init(&bs->throttled_reqs[1]);
366
    bs->refcnt = 1;
367
    bs->aio_context = qemu_get_aio_context();
P
Paolo Bonzini 已提交
368

B
bellard 已提交
369 370 371
    return bs;
}

P
Paolo Bonzini 已提交
372 373 374 375 376
void bdrv_add_close_notifier(BlockDriverState *bs, Notifier *notify)
{
    notifier_list_add(&bs->close_notifiers, notify);
}

B
bellard 已提交
377 378 379
BlockDriver *bdrv_find_format(const char *format_name)
{
    BlockDriver *drv1;
380 381
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
        if (!strcmp(drv1->format_name, format_name)) {
B
bellard 已提交
382
            return drv1;
383
        }
B
bellard 已提交
384 385 386 387
    }
    return NULL;
}

388
static int bdrv_is_whitelisted(BlockDriver *drv, bool read_only)
389
{
390 391 392 393 394
    static const char *whitelist_rw[] = {
        CONFIG_BDRV_RW_WHITELIST
    };
    static const char *whitelist_ro[] = {
        CONFIG_BDRV_RO_WHITELIST
395 396 397
    };
    const char **p;

398
    if (!whitelist_rw[0] && !whitelist_ro[0]) {
399
        return 1;               /* no whitelist, anything goes */
400
    }
401

402
    for (p = whitelist_rw; *p; p++) {
403 404 405 406
        if (!strcmp(drv->format_name, *p)) {
            return 1;
        }
    }
407 408 409 410 411 412 413
    if (read_only) {
        for (p = whitelist_ro; *p; p++) {
            if (!strcmp(drv->format_name, *p)) {
                return 1;
            }
        }
    }
414 415 416
    return 0;
}

417 418
BlockDriver *bdrv_find_whitelisted_format(const char *format_name,
                                          bool read_only)
419 420
{
    BlockDriver *drv = bdrv_find_format(format_name);
421
    return drv && bdrv_is_whitelisted(drv, read_only) ? drv : NULL;
422 423
}

424 425 426
typedef struct CreateCo {
    BlockDriver *drv;
    char *filename;
427
    QemuOpts *opts;
428
    int ret;
429
    Error *err;
430 431 432 433
} CreateCo;

static void coroutine_fn bdrv_create_co_entry(void *opaque)
{
434 435 436
    Error *local_err = NULL;
    int ret;

437 438 439
    CreateCo *cco = opaque;
    assert(cco->drv);

C
Chunyan Liu 已提交
440
    ret = cco->drv->bdrv_create(cco->filename, cco->opts, &local_err);
441
    if (local_err) {
442 443 444
        error_propagate(&cco->err, local_err);
    }
    cco->ret = ret;
445 446
}

447
int bdrv_create(BlockDriver *drv, const char* filename,
448
                QemuOpts *opts, Error **errp)
B
bellard 已提交
449
{
450 451 452 453 454 455
    int ret;

    Coroutine *co;
    CreateCo cco = {
        .drv = drv,
        .filename = g_strdup(filename),
456
        .opts = opts,
457
        .ret = NOT_DONE,
458
        .err = NULL,
459 460
    };

C
Chunyan Liu 已提交
461
    if (!drv->bdrv_create) {
462
        error_setg(errp, "Driver '%s' does not support image creation", drv->format_name);
463 464
        ret = -ENOTSUP;
        goto out;
465 466 467 468 469 470 471 472 473 474 475 476 477 478
    }

    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;
479
    if (ret < 0) {
480
        if (cco.err) {
481 482 483 484 485
            error_propagate(errp, cco.err);
        } else {
            error_setg_errno(errp, -ret, "Could not create image");
        }
    }
486

487 488
out:
    g_free(cco.filename);
489
    return ret;
B
bellard 已提交
490 491
}

C
Chunyan Liu 已提交
492
int bdrv_create_file(const char *filename, QemuOpts *opts, Error **errp)
493 494
{
    BlockDriver *drv;
495 496
    Error *local_err = NULL;
    int ret;
497

498
    drv = bdrv_find_protocol(filename, true);
499
    if (drv == NULL) {
500
        error_setg(errp, "Could not find protocol for file '%s'", filename);
501
        return -ENOENT;
502 503
    }

C
Chunyan Liu 已提交
504
    ret = bdrv_create(drv, filename, opts, &local_err);
505
    if (local_err) {
506 507 508
        error_propagate(errp, local_err);
    }
    return ret;
509 510
}

511
int bdrv_refresh_limits(BlockDriverState *bs)
512 513 514 515 516
{
    BlockDriver *drv = bs->drv;

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

K
Kevin Wolf 已提交
517 518 519 520 521 522 523 524
    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;
525 526 527
        bs->bl.opt_mem_alignment = bs->file->bl.opt_mem_alignment;
    } else {
        bs->bl.opt_mem_alignment = 512;
K
Kevin Wolf 已提交
528 529 530 531 532 533 534
    }

    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);
535 536 537
        bs->bl.opt_mem_alignment =
            MAX(bs->bl.opt_mem_alignment,
                bs->backing_hd->bl.opt_mem_alignment);
K
Kevin Wolf 已提交
538 539 540 541
    }

    /* Then let the driver override it */
    if (drv->bdrv_refresh_limits) {
542 543 544 545 546 547
        return drv->bdrv_refresh_limits(bs);
    }

    return 0;
}

548 549 550 551 552
/*
 * 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
553
{
554
#ifdef _WIN32
555
    char temp_dir[MAX_PATH];
556 557 558 559 560 561
    /* 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
562
#else
B
bellard 已提交
563
    int fd;
564
    const char *tmpdir;
A
aurel32 已提交
565
    tmpdir = getenv("TMPDIR");
566 567 568
    if (!tmpdir) {
        tmpdir = "/var/tmp";
    }
569 570 571
    if (snprintf(filename, size, "%s/vl.XXXXXX", tmpdir) >= size) {
        return -EOVERFLOW;
    }
B
bellard 已提交
572
    fd = mkstemp(filename);
573 574 575 576 577
    if (fd < 0) {
        return -errno;
    }
    if (close(fd) != 0) {
        unlink(filename);
578 579 580
        return -errno;
    }
    return 0;
B
bellard 已提交
581
#endif
582
}
B
bellard 已提交
583

584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605
/*
 * 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;
}

606 607
BlockDriver *bdrv_find_protocol(const char *filename,
                                bool allow_protocol_prefix)
B
bellard 已提交
608 609 610
{
    BlockDriver *drv1;
    char protocol[128];
611
    int len;
B
bellard 已提交
612
    const char *p;
B
bellard 已提交
613

614 615
    /* TODO Drivers without bdrv_file_open must be specified explicitly */

616 617 618 619 620 621 622 623 624 625 626 627
    /*
     * 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;
    }

628
    if (!path_has_protocol(filename) || !allow_protocol_prefix) {
629
        return bdrv_find_format("file");
630
    }
631

632 633
    p = strchr(filename, ':');
    assert(p != NULL);
634 635 636 637 638
    len = p - filename;
    if (len > sizeof(protocol) - 1)
        len = sizeof(protocol) - 1;
    memcpy(protocol, filename, len);
    protocol[len] = '\0';
639
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
640
        if (drv1->protocol_name &&
641
            !strcmp(drv1->protocol_name, protocol)) {
B
bellard 已提交
642
            return drv1;
643
        }
B
bellard 已提交
644 645 646 647
    }
    return NULL;
}

648
static int find_image_format(BlockDriverState *bs, const char *filename,
649
                             BlockDriver **pdrv, Error **errp)
650
{
651
    int score, score_max;
652 653
    BlockDriver *drv1, *drv;
    uint8_t buf[2048];
654
    int ret = 0;
655

656
    /* Return the raw BlockDriver * to scsi-generic devices or empty drives */
657
    if (bs->sg || !bdrv_is_inserted(bs) || bdrv_getlength(bs) == 0) {
658 659
        drv = bdrv_find_format("raw");
        if (!drv) {
660
            error_setg(errp, "Could not find raw image format");
661 662 663 664
            ret = -ENOENT;
        }
        *pdrv = drv;
        return ret;
665
    }
666

B
bellard 已提交
667 668
    ret = bdrv_pread(bs, 0, buf, sizeof(buf));
    if (ret < 0) {
669 670
        error_setg_errno(errp, -ret, "Could not read image for determining its "
                         "format");
671 672
        *pdrv = NULL;
        return ret;
B
bellard 已提交
673 674
    }

B
bellard 已提交
675
    score_max = 0;
676
    drv = NULL;
677
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
B
bellard 已提交
678 679 680 681 682 683
        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
684
        }
B
bellard 已提交
685
    }
686
    if (!drv) {
687 688
        error_setg(errp, "Could not determine image format: No compatible "
                   "driver found");
689 690 691 692
        ret = -ENOENT;
    }
    *pdrv = drv;
    return ret;
B
bellard 已提交
693 694
}

695 696 697 698 699 700 701
/**
 * Set the current 'total_sectors' value
 */
static int refresh_total_sectors(BlockDriverState *bs, int64_t hint)
{
    BlockDriver *drv = bs->drv;

702 703 704 705
    /* Do not attempt drv->bdrv_getlength() on scsi-generic devices */
    if (bs->sg)
        return 0;

706 707 708 709 710 711
    /* 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
712
        hint = DIV_ROUND_UP(length, BDRV_SECTOR_SIZE);
713 714 715 716 717 718
    }

    bs->total_sectors = hint;
    return 0;
}

P
Paolo Bonzini 已提交
719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738
/**
 * 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;
}

739 740 741 742 743 744 745 746 747 748 749
/**
 * 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;
750 751
    } else if (!strcmp(mode, "directsync")) {
        *flags |= BDRV_O_NOCACHE;
752 753 754 755 756 757 758 759 760 761 762 763 764 765
    } 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;
}

766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781
/**
 * 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--;
}

782 783 784 785 786 787 788 789 790 791
/*
 * Returns the flags that a temporary snapshot should get, based on the
 * originally requested flags (the originally requested image will have flags
 * like a backing file)
 */
static int bdrv_temp_snapshot_flags(int flags)
{
    return (flags & ~BDRV_O_SNAPSHOT) | BDRV_O_TEMPORARY;
}

792 793 794 795 796 797 798 799 800 801 802 803 804 805
/*
 * Returns the flags that bs->file should get, based on the given flags for
 * the parent BDS
 */
static int bdrv_inherited_flags(int flags)
{
    /* Enable protocol handling, disable format probing for bs->file */
    flags |= BDRV_O_PROTOCOL;

    /* Our block drivers take care to send flushes and respect unmap policy,
     * so we can enable both unconditionally on lower layers. */
    flags |= BDRV_O_CACHE_WB | BDRV_O_UNMAP;

    /* Clear flags that only apply to the top layer */
806
    flags &= ~(BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING | BDRV_O_COPY_ON_READ);
807 808 809 810

    return flags;
}

K
Kevin Wolf 已提交
811 812 813 814 815 816 817 818 819 820
/*
 * Returns the flags that bs->backing_hd should get, based on the given flags
 * for the parent BDS
 */
static int bdrv_backing_flags(int flags)
{
    /* backing files always opened read-only */
    flags &= ~(BDRV_O_RDWR | BDRV_O_COPY_ON_READ);

    /* snapshot=on is handled on the top layer */
821
    flags &= ~(BDRV_O_SNAPSHOT | BDRV_O_TEMPORARY);
K
Kevin Wolf 已提交
822 823 824 825

    return flags;
}

K
Kevin Wolf 已提交
826 827 828 829 830 831 832 833
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.
     */
834
    open_flags &= ~(BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING | BDRV_O_PROTOCOL);
K
Kevin Wolf 已提交
835 836 837 838

    /*
     * Snapshots should be writable.
     */
839
    if (flags & BDRV_O_TEMPORARY) {
K
Kevin Wolf 已提交
840 841 842 843 844 845
        open_flags |= BDRV_O_RDWR;
    }

    return open_flags;
}

846 847 848
static void bdrv_assign_node_name(BlockDriverState *bs,
                                  const char *node_name,
                                  Error **errp)
849 850
{
    if (!node_name) {
851
        return;
852 853 854 855 856
    }

    /* empty string node name is invalid */
    if (node_name[0] == '\0') {
        error_setg(errp, "Empty node name");
857
        return;
858 859
    }

860 861 862 863
    /* 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);
864
        return;
865 866
    }

867 868 869
    /* takes care of avoiding duplicates node names */
    if (bdrv_find_node(node_name)) {
        error_setg(errp, "Duplicate node name");
870
        return;
871 872 873 874 875 876 877
    }

    /* 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);
}

878 879
/*
 * Common part for opening disk images and files
880 881
 *
 * Removes all processed options from *options.
882
 */
883
static int bdrv_open_common(BlockDriverState *bs, BlockDriverState *file,
884
    QDict *options, int flags, BlockDriver *drv, Error **errp)
885 886
{
    int ret, open_flags;
K
Kevin Wolf 已提交
887
    const char *filename;
888
    const char *node_name = NULL;
889
    Error *local_err = NULL;
890 891

    assert(drv != NULL);
892
    assert(bs->file == NULL);
893
    assert(options != NULL && bs->options != options);
894

895 896 897 898 899 900
    if (file != NULL) {
        filename = file->filename;
    } else {
        filename = qdict_get_try_str(options, "filename");
    }

901 902 903 904 905 906
    if (drv->bdrv_needs_filename && !filename) {
        error_setg(errp, "The '%s' block driver requires a file name",
                   drv->format_name);
        return -EINVAL;
    }

907
    trace_bdrv_open_common(bs, filename ?: "", flags, drv->format_name);
S
Stefan Hajnoczi 已提交
908

909
    node_name = qdict_get_try_str(options, "node-name");
910
    bdrv_assign_node_name(bs, node_name, &local_err);
911
    if (local_err) {
912 913
        error_propagate(errp, local_err);
        return -EINVAL;
914 915 916
    }
    qdict_del(options, "node-name");

917 918 919 920 921 922 923 924
    /* 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;
    }

925
    bs->open_flags = flags;
926
    bs->guest_block_size = 512;
927
    bs->request_alignment = 512;
A
Asias He 已提交
928
    bs->zero_beyond_eof = true;
929 930
    open_flags = bdrv_open_flags(bs, flags);
    bs->read_only = !(open_flags & BDRV_O_RDWR);
931
    bs->growable = !!(flags & BDRV_O_PROTOCOL);
932 933

    if (use_bdrv_whitelist && !bdrv_is_whitelisted(drv, bs->read_only)) {
934 935 936 937 938
        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);
939 940
        return -ENOTSUP;
    }
941

942
    assert(bs->copy_on_read == 0); /* bdrv_new() and bdrv_close() make it so */
943 944 945 946 947 948 949
    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;
        }
950 951
    }

952 953 954 955 956
    if (filename != NULL) {
        pstrcpy(bs->filename, sizeof(bs->filename), filename);
    } else {
        bs->filename[0] = '\0';
    }
957 958

    bs->drv = drv;
959
    bs->opaque = g_malloc0(drv->instance_size);
960

961
    bs->enable_write_cache = !!(flags & BDRV_O_CACHE_WB);
962

963 964
    /* Open the image, either directly or using a protocol */
    if (drv->bdrv_file_open) {
965
        assert(file == NULL);
966
        assert(!drv->bdrv_needs_filename || filename != NULL);
967
        ret = drv->bdrv_file_open(bs, options, open_flags, &local_err);
968
    } else {
969
        if (file == NULL) {
970 971
            error_setg(errp, "Can't use '%s' as a block driver for the "
                       "protocol level", drv->format_name);
972 973 974
            ret = -EINVAL;
            goto free_and_fail;
        }
975
        bs->file = file;
976
        ret = drv->bdrv_open(bs, options, open_flags, &local_err);
977 978
    }

979
    if (ret < 0) {
980
        if (local_err) {
981
            error_propagate(errp, local_err);
982 983
        } else if (bs->filename[0]) {
            error_setg_errno(errp, -ret, "Could not open '%s'", bs->filename);
984 985 986
        } else {
            error_setg_errno(errp, -ret, "Could not open image");
        }
987 988 989
        goto free_and_fail;
    }

990 991
    ret = refresh_total_sectors(bs, bs->total_sectors);
    if (ret < 0) {
992
        error_setg_errno(errp, -ret, "Could not refresh total sector count");
993
        goto free_and_fail;
994
    }
995

996
    bdrv_refresh_limits(bs);
997
    assert(bdrv_opt_mem_align(bs) != 0);
998
    assert((bs->request_alignment != 0) || bs->sg);
999 1000 1001
    return 0;

free_and_fail:
1002
    bs->file = NULL;
1003
    g_free(bs->opaque);
1004 1005 1006 1007 1008
    bs->opaque = NULL;
    bs->drv = NULL;
    return ret;
}

1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035
static QDict *parse_json_filename(const char *filename, Error **errp)
{
    QObject *options_obj;
    QDict *options;
    int ret;

    ret = strstart(filename, "json:", &filename);
    assert(ret);

    options_obj = qobject_from_json(filename);
    if (!options_obj) {
        error_setg(errp, "Could not parse the JSON options");
        return NULL;
    }

    if (qobject_type(options_obj) != QTYPE_QDICT) {
        qobject_decref(options_obj);
        error_setg(errp, "Invalid JSON object given");
        return NULL;
    }

    options = qobject_to_qdict(options_obj);
    qdict_flatten(options);

    return options;
}

K
Kevin Wolf 已提交
1036
/*
K
Kevin Wolf 已提交
1037 1038
 * Fills in default options for opening images and converts the legacy
 * filename/flags pair to option QDict entries.
K
Kevin Wolf 已提交
1039
 */
1040
static int bdrv_fill_options(QDict **options, const char **pfilename, int flags,
1041
                             BlockDriver *drv, Error **errp)
B
bellard 已提交
1042
{
1043
    const char *filename = *pfilename;
1044
    const char *drvname;
1045
    bool protocol = flags & BDRV_O_PROTOCOL;
1046
    bool parse_filename = false;
1047
    Error *local_err = NULL;
B
bellard 已提交
1048

1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063
    /* Parse json: pseudo-protocol */
    if (filename && g_str_has_prefix(filename, "json:")) {
        QDict *json_options = parse_json_filename(filename, &local_err);
        if (local_err) {
            error_propagate(errp, local_err);
            return -EINVAL;
        }

        /* Options given in the filename have lower priority than options
         * specified directly */
        qdict_join(*options, json_options, false);
        QDECREF(json_options);
        *pfilename = filename = NULL;
    }

K
Kevin Wolf 已提交
1064
    /* Fetch the file name from the options QDict if necessary */
1065
    if (protocol && filename) {
K
Kevin Wolf 已提交
1066 1067 1068 1069 1070 1071 1072 1073
        if (!qdict_haskey(*options, "filename")) {
            qdict_put(*options, "filename", qstring_from_str(filename));
            parse_filename = true;
        } else {
            error_setg(errp, "Can't specify 'file' and 'filename' options at "
                             "the same time");
            return -EINVAL;
        }
K
Kevin Wolf 已提交
1074 1075
    }

1076
    /* Find the right block driver */
K
Kevin Wolf 已提交
1077
    filename = qdict_get_try_str(*options, "filename");
1078
    drvname = qdict_get_try_str(*options, "driver");
K
Kevin Wolf 已提交
1079

1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099
    if (drv) {
        if (drvname) {
            error_setg(errp, "Driver specified twice");
            return -EINVAL;
        }
        drvname = drv->format_name;
        qdict_put(*options, "driver", qstring_from_str(drvname));
    } else {
        if (!drvname && protocol) {
            if (filename) {
                drv = bdrv_find_protocol(filename, parse_filename);
                if (!drv) {
                    error_setg(errp, "Unknown protocol");
                    return -EINVAL;
                }

                drvname = drv->format_name;
                qdict_put(*options, "driver", qstring_from_str(drvname));
            } else {
                error_setg(errp, "Must specify either driver or file");
K
Kevin Wolf 已提交
1100 1101
                return -EINVAL;
            }
1102 1103 1104 1105 1106 1107
        } else if (drvname) {
            drv = bdrv_find_format(drvname);
            if (!drv) {
                error_setg(errp, "Unknown driver '%s'", drvname);
                return -ENOENT;
            }
1108
        }
1109 1110
    }

1111
    assert(drv || !protocol);
1112

K
Kevin Wolf 已提交
1113
    /* Driver-specific filename parsing */
1114
    if (drv && drv->bdrv_parse_filename && parse_filename) {
1115
        drv->bdrv_parse_filename(filename, *options, &local_err);
1116
        if (local_err) {
1117
            error_propagate(errp, local_err);
K
Kevin Wolf 已提交
1118
            return -EINVAL;
1119
        }
1120 1121 1122 1123

        if (!drv->bdrv_needs_filename) {
            qdict_del(*options, "filename");
        }
1124 1125
    }

K
Kevin Wolf 已提交
1126 1127 1128
    return 0;
}

F
Fam Zheng 已提交
1129 1130 1131
void bdrv_set_backing_hd(BlockDriverState *bs, BlockDriverState *backing_hd)
{

1132 1133 1134 1135 1136 1137 1138 1139 1140
    if (bs->backing_hd) {
        assert(bs->backing_blocker);
        bdrv_op_unblock_all(bs->backing_hd, bs->backing_blocker);
    } else if (backing_hd) {
        error_setg(&bs->backing_blocker,
                   "device is used as backing hd of '%s'",
                   bs->device_name);
    }

F
Fam Zheng 已提交
1141 1142
    bs->backing_hd = backing_hd;
    if (!backing_hd) {
1143 1144
        error_free(bs->backing_blocker);
        bs->backing_blocker = NULL;
F
Fam Zheng 已提交
1145 1146 1147 1148 1149 1150
        goto out;
    }
    bs->open_flags &= ~BDRV_O_NO_BACKING;
    pstrcpy(bs->backing_file, sizeof(bs->backing_file), backing_hd->filename);
    pstrcpy(bs->backing_format, sizeof(bs->backing_format),
            backing_hd->drv ? backing_hd->drv->format_name : "");
1151 1152 1153 1154 1155

    bdrv_op_block_all(bs->backing_hd, bs->backing_blocker);
    /* Otherwise we won't be able to commit due to check in bdrv_commit */
    bdrv_op_unblock(bs->backing_hd, BLOCK_OP_TYPE_COMMIT,
                    bs->backing_blocker);
F
Fam Zheng 已提交
1156 1157 1158 1159
out:
    bdrv_refresh_limits(bs);
}

1160 1161 1162 1163 1164 1165 1166 1167
/*
 * 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.
 */
1168
int bdrv_open_backing_file(BlockDriverState *bs, QDict *options, Error **errp)
P
Paolo Bonzini 已提交
1169
{
1170
    char *backing_filename = g_malloc0(PATH_MAX);
K
Kevin Wolf 已提交
1171
    int ret = 0;
P
Paolo Bonzini 已提交
1172
    BlockDriver *back_drv = NULL;
F
Fam Zheng 已提交
1173
    BlockDriverState *backing_hd;
1174
    Error *local_err = NULL;
P
Paolo Bonzini 已提交
1175 1176

    if (bs->backing_hd != NULL) {
1177
        QDECREF(options);
1178
        goto free_exit;
P
Paolo Bonzini 已提交
1179 1180
    }

1181 1182 1183 1184 1185
    /* NULL means an empty set of options */
    if (options == NULL) {
        options = qdict_new();
    }

P
Paolo Bonzini 已提交
1186
    bs->open_flags &= ~BDRV_O_NO_BACKING;
1187 1188 1189
    if (qdict_haskey(options, "file.filename")) {
        backing_filename[0] = '\0';
    } else if (bs->backing_file[0] == '\0' && qdict_size(options) == 0) {
1190
        QDECREF(options);
1191
        goto free_exit;
F
Fam Zheng 已提交
1192
    } else {
1193
        bdrv_get_full_backing_filename(bs, backing_filename, PATH_MAX);
P
Paolo Bonzini 已提交
1194 1195
    }

1196 1197 1198 1199 1200 1201 1202
    if (!bs->drv || !bs->drv->supports_backing) {
        ret = -EINVAL;
        error_setg(errp, "Driver doesn't support backing files");
        QDECREF(options);
        goto free_exit;
    }

F
Fam Zheng 已提交
1203 1204
    backing_hd = bdrv_new("", errp);

P
Paolo Bonzini 已提交
1205 1206 1207 1208
    if (bs->backing_format[0] != '\0') {
        back_drv = bdrv_find_format(bs->backing_format);
    }

1209
    assert(bs->backing_hd == NULL);
F
Fam Zheng 已提交
1210
    ret = bdrv_open(&backing_hd,
1211
                    *backing_filename ? backing_filename : NULL, NULL, options,
K
Kevin Wolf 已提交
1212
                    bdrv_backing_flags(bs->open_flags), back_drv, &local_err);
P
Paolo Bonzini 已提交
1213
    if (ret < 0) {
F
Fam Zheng 已提交
1214 1215
        bdrv_unref(backing_hd);
        backing_hd = NULL;
P
Paolo Bonzini 已提交
1216
        bs->open_flags |= BDRV_O_NO_BACKING;
1217 1218 1219
        error_setg(errp, "Could not open backing file: %s",
                   error_get_pretty(local_err));
        error_free(local_err);
1220
        goto free_exit;
P
Paolo Bonzini 已提交
1221
    }
F
Fam Zheng 已提交
1222
    bdrv_set_backing_hd(bs, backing_hd);
P
Peter Feiner 已提交
1223

1224 1225 1226
free_exit:
    g_free(backing_filename);
    return ret;
P
Paolo Bonzini 已提交
1227 1228
}

M
Max Reitz 已提交
1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241
/*
 * Opens a disk image whose options are given as BlockdevRef in another block
 * device's options.
 *
 * 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.
1242 1243
 *
 * To conform with the behavior of bdrv_open(), *pbs has to be NULL.
M
Max Reitz 已提交
1244 1245 1246
 */
int bdrv_open_image(BlockDriverState **pbs, const char *filename,
                    QDict *options, const char *bdref_key, int flags,
1247
                    bool allow_none, Error **errp)
M
Max Reitz 已提交
1248 1249 1250 1251 1252 1253
{
    QDict *image_options;
    int ret;
    char *bdref_key_dot;
    const char *reference;

1254 1255 1256
    assert(pbs);
    assert(*pbs == NULL);

M
Max Reitz 已提交
1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269
    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;
        }
1270
        QDECREF(image_options);
M
Max Reitz 已提交
1271 1272 1273
        goto done;
    }

1274
    ret = bdrv_open(pbs, filename, reference, image_options, flags, NULL, errp);
M
Max Reitz 已提交
1275 1276 1277 1278 1279 1280

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

1281
void bdrv_append_temp_snapshot(BlockDriverState *bs, int flags, Error **errp)
1282 1283
{
    /* TODO: extra byte is a hack to ensure MAX_PATH space on Windows. */
1284
    char *tmp_filename = g_malloc0(PATH_MAX + 1);
1285 1286
    int64_t total_size;
    BlockDriver *bdrv_qcow2;
1287
    QemuOpts *opts = NULL;
1288 1289 1290 1291 1292 1293 1294 1295 1296
    QDict *snapshot_options;
    BlockDriverState *bs_snapshot;
    Error *local_err;
    int ret;

    /* if snapshot, we create a temporary backing file and open it
       instead of opening 'filename' directly */

    /* Get the required size from the image */
1297 1298 1299
    total_size = bdrv_getlength(bs);
    if (total_size < 0) {
        error_setg_errno(errp, -total_size, "Could not get image size");
1300
        goto out;
1301 1302
    }
    total_size &= BDRV_SECTOR_MASK;
1303 1304

    /* Create the temporary image */
1305
    ret = get_tmp_filename(tmp_filename, PATH_MAX + 1);
1306 1307
    if (ret < 0) {
        error_setg_errno(errp, -ret, "Could not get temporary filename");
1308
        goto out;
1309 1310 1311
    }

    bdrv_qcow2 = bdrv_find_format("qcow2");
C
Chunyan Liu 已提交
1312 1313
    opts = qemu_opts_create(bdrv_qcow2->create_opts, NULL, 0,
                            &error_abort);
1314
    qemu_opt_set_number(opts, BLOCK_OPT_SIZE, total_size);
C
Chunyan Liu 已提交
1315
    ret = bdrv_create(bdrv_qcow2, tmp_filename, opts, &local_err);
1316
    qemu_opts_del(opts);
1317 1318 1319 1320 1321
    if (ret < 0) {
        error_setg_errno(errp, -ret, "Could not create temporary overlay "
                         "'%s': %s", tmp_filename,
                         error_get_pretty(local_err));
        error_free(local_err);
1322
        goto out;
1323 1324 1325 1326 1327 1328 1329 1330 1331
    }

    /* Prepare a new options QDict for the temporary file */
    snapshot_options = qdict_new();
    qdict_put(snapshot_options, "file.driver",
              qstring_from_str("file"));
    qdict_put(snapshot_options, "file.filename",
              qstring_from_str(tmp_filename));

K
Kevin Wolf 已提交
1332
    bs_snapshot = bdrv_new("", &error_abort);
1333 1334

    ret = bdrv_open(&bs_snapshot, NULL, NULL, snapshot_options,
1335
                    flags, bdrv_qcow2, &local_err);
1336 1337
    if (ret < 0) {
        error_propagate(errp, local_err);
1338
        goto out;
1339 1340 1341
    }

    bdrv_append(bs_snapshot, bs);
1342 1343 1344

out:
    g_free(tmp_filename);
1345 1346
}

K
Kevin Wolf 已提交
1347 1348
/*
 * Opens a disk image (raw, qcow2, vmdk, ...)
1349 1350 1351 1352 1353
 *
 * 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.
1354 1355 1356
 *
 * 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.
1357 1358 1359 1360
 *
 * The reference parameter may be used to specify an existing block device which
 * should be opened. If specified, neither options nor a filename may be given,
 * nor can an existing BDS be reused (that is, *pbs has to be NULL).
K
Kevin Wolf 已提交
1361
 */
1362 1363 1364
int bdrv_open(BlockDriverState **pbs, const char *filename,
              const char *reference, QDict *options, int flags,
              BlockDriver *drv, Error **errp)
B
bellard 已提交
1365
{
K
Kevin Wolf 已提交
1366
    int ret;
1367
    BlockDriverState *file = NULL, *bs;
1368
    const char *drvname;
1369
    Error *local_err = NULL;
1370
    int snapshot_flags = 0;
B
bellard 已提交
1371

1372 1373
    assert(pbs);

1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398
    if (reference) {
        bool options_non_empty = options ? qdict_size(options) : false;
        QDECREF(options);

        if (*pbs) {
            error_setg(errp, "Cannot reuse an existing BDS when referencing "
                       "another block device");
            return -EINVAL;
        }

        if (filename || options_non_empty) {
            error_setg(errp, "Cannot reference an existing block device with "
                       "additional options or a new filename");
            return -EINVAL;
        }

        bs = bdrv_lookup_bs(reference, reference, errp);
        if (!bs) {
            return -ENODEV;
        }
        bdrv_ref(bs);
        *pbs = bs;
        return 0;
    }

1399 1400 1401
    if (*pbs) {
        bs = *pbs;
    } else {
K
Kevin Wolf 已提交
1402
        bs = bdrv_new("", &error_abort);
1403 1404
    }

1405 1406 1407 1408 1409
    /* NULL means an empty set of options */
    if (options == NULL) {
        options = qdict_new();
    }

1410
    ret = bdrv_fill_options(&options, &filename, flags, drv, &local_err);
1411 1412 1413 1414
    if (local_err) {
        goto fail;
    }

1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434
    /* Find the right image format driver */
    drv = NULL;
    drvname = qdict_get_try_str(options, "driver");
    if (drvname) {
        drv = bdrv_find_format(drvname);
        qdict_del(options, "driver");
        if (!drv) {
            error_setg(errp, "Unknown driver: '%s'", drvname);
            ret = -EINVAL;
            goto fail;
        }
    }

    assert(drvname || !(flags & BDRV_O_PROTOCOL));
    if (drv && !drv->bdrv_file_open) {
        /* If the user explicitly wants a format driver here, we'll need to add
         * another layer for the protocol in bs->file */
        flags &= ~BDRV_O_PROTOCOL;
    }

1435
    bs->options = options;
1436
    options = qdict_clone_shallow(options);
1437

1438
    /* Open image file without format layer */
1439 1440 1441 1442 1443 1444 1445 1446
    if ((flags & BDRV_O_PROTOCOL) == 0) {
        if (flags & BDRV_O_RDWR) {
            flags |= BDRV_O_ALLOW_RDWR;
        }
        if (flags & BDRV_O_SNAPSHOT) {
            snapshot_flags = bdrv_temp_snapshot_flags(flags);
            flags = bdrv_backing_flags(flags);
        }
1447

1448 1449 1450 1451 1452 1453 1454
        assert(file == NULL);
        ret = bdrv_open_image(&file, filename, options, "file",
                              bdrv_inherited_flags(flags),
                              true, &local_err);
        if (ret < 0) {
            goto fail;
        }
1455 1456
    }

1457 1458
    /* Image format probing */
    if (!drv && file) {
1459 1460
        ret = find_image_format(file, filename, &drv, &local_err);
        if (ret < 0) {
1461
            goto fail;
1462
        }
1463
    } else if (!drv) {
1464 1465
        error_setg(errp, "Must specify either driver or file");
        ret = -EINVAL;
1466
        goto fail;
B
bellard 已提交
1467
    }
K
Kevin Wolf 已提交
1468 1469

    /* Open the image */
1470
    ret = bdrv_open_common(bs, file, options, flags, drv, &local_err);
K
Kevin Wolf 已提交
1471
    if (ret < 0) {
1472
        goto fail;
1473 1474
    }

1475
    if (file && (bs->file != file)) {
F
Fam Zheng 已提交
1476
        bdrv_unref(file);
1477 1478 1479
        file = NULL;
    }

K
Kevin Wolf 已提交
1480
    /* If there is a backing file, use it */
P
Paolo Bonzini 已提交
1481
    if ((flags & BDRV_O_NO_BACKING) == 0) {
1482 1483
        QDict *backing_options;

1484
        qdict_extract_subqdict(options, &backing_options, "backing.");
1485
        ret = bdrv_open_backing_file(bs, backing_options, &local_err);
K
Kevin Wolf 已提交
1486
        if (ret < 0) {
1487
            goto close_and_fail;
K
Kevin Wolf 已提交
1488 1489 1490
        }
    }

1491 1492
    /* For snapshot=on, create a temporary qcow2 overlay. bs points to the
     * temporary snapshot afterwards. */
1493 1494
    if (snapshot_flags) {
        bdrv_append_temp_snapshot(bs, snapshot_flags, &local_err);
1495 1496 1497 1498 1499 1500
        if (local_err) {
            error_propagate(errp, local_err);
            goto close_and_fail;
        }
    }

1501
    /* Check if any unknown options were used */
1502
    if (options && (qdict_size(options) != 0)) {
1503
        const QDictEntry *entry = qdict_first(options);
1504 1505 1506 1507 1508 1509 1510 1511
        if (flags & BDRV_O_PROTOCOL) {
            error_setg(errp, "Block protocol '%s' doesn't support the option "
                       "'%s'", drv->format_name, entry->key);
        } else {
            error_setg(errp, "Block format '%s' used by device '%s' doesn't "
                       "support the option '%s'", drv->format_name,
                       bs->device_name, entry->key);
        }
1512 1513 1514 1515 1516

        ret = -EINVAL;
        goto close_and_fail;
    }

K
Kevin Wolf 已提交
1517
    if (!bdrv_key_required(bs)) {
1518
        bdrv_dev_change_media_cb(bs, true);
1519 1520 1521 1522 1523 1524 1525
    } else if (!runstate_check(RUN_STATE_PRELAUNCH)
               && !runstate_check(RUN_STATE_INMIGRATE)
               && !runstate_check(RUN_STATE_PAUSED)) { /* HACK */
        error_setg(errp,
                   "Guest must be stopped for opening of encrypted image");
        ret = -EBUSY;
        goto close_and_fail;
K
Kevin Wolf 已提交
1526 1527
    }

1528
    QDECREF(options);
1529
    *pbs = bs;
K
Kevin Wolf 已提交
1530 1531
    return 0;

1532
fail:
1533
    if (file != NULL) {
F
Fam Zheng 已提交
1534
        bdrv_unref(file);
1535
    }
1536
    QDECREF(bs->options);
1537
    QDECREF(options);
1538
    bs->options = NULL;
1539 1540 1541 1542 1543 1544
    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);
    }
1545
    if (local_err) {
1546 1547
        error_propagate(errp, local_err);
    }
1548
    return ret;
1549

1550
close_and_fail:
1551 1552 1553 1554 1555 1556
    /* See fail path, but now the BDS has to be always closed */
    if (*pbs) {
        bdrv_close(bs);
    } else {
        bdrv_unref(bs);
    }
1557
    QDECREF(options);
1558
    if (local_err) {
1559 1560
        error_propagate(errp, local_err);
    }
K
Kevin Wolf 已提交
1561 1562 1563
    return ret;
}

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
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);
    }

1599 1600 1601
    /* bdrv_open() masks this flag out */
    flags &= ~BDRV_O_PROTOCOL;

1602
    if (bs->file) {
1603
        bdrv_reopen_queue(bs_queue, bs->file, bdrv_inherited_flags(flags));
1604 1605 1606 1607 1608 1609 1610 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 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734
    }

    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 {
1735 1736
                error_setg(errp, "failed while preparing to reopen image '%s'",
                           reopen_state->bs->filename);
1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778
            }
            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);
1779 1780

    bdrv_refresh_limits(reopen_state->bs);
1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800
}

/*
 * 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
1801 1802
void bdrv_close(BlockDriverState *bs)
{
1803 1804 1805
    if (bs->job) {
        block_job_cancel_sync(bs->job);
    }
1806 1807 1808
    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
1809
    notifier_list_notify(&bs->close_notifiers, bs);
K
Kevin Wolf 已提交
1810

1811
    if (bs->drv) {
1812
        if (bs->backing_hd) {
1813 1814 1815
            BlockDriverState *backing_hd = bs->backing_hd;
            bdrv_set_backing_hd(bs, NULL);
            bdrv_unref(backing_hd);
1816
        }
B
bellard 已提交
1817
        bs->drv->bdrv_close(bs);
1818
        g_free(bs->opaque);
B
bellard 已提交
1819 1820
        bs->opaque = NULL;
        bs->drv = NULL;
1821
        bs->copy_on_read = 0;
1822 1823
        bs->backing_file[0] = '\0';
        bs->backing_format[0] = '\0';
1824 1825 1826 1827 1828
        bs->total_sectors = 0;
        bs->encrypted = 0;
        bs->valid_key = 0;
        bs->sg = 0;
        bs->growable = 0;
A
Asias He 已提交
1829
        bs->zero_beyond_eof = false;
1830 1831
        QDECREF(bs->options);
        bs->options = NULL;
B
bellard 已提交
1832

1833
        if (bs->file != NULL) {
F
Fam Zheng 已提交
1834
            bdrv_unref(bs->file);
1835
            bs->file = NULL;
1836
        }
B
bellard 已提交
1837
    }
Z
Zhi Yong Wu 已提交
1838

P
Pavel Hrdina 已提交
1839 1840
    bdrv_dev_change_media_cb(bs, false);

Z
Zhi Yong Wu 已提交
1841 1842 1843 1844
    /*throttling disk I/O limits*/
    if (bs->io_limits_enabled) {
        bdrv_io_limits_disable(bs);
    }
B
bellard 已提交
1845 1846
}

1847 1848 1849 1850
void bdrv_close_all(void)
{
    BlockDriverState *bs;

1851
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
1852 1853 1854
        AioContext *aio_context = bdrv_get_aio_context(bs);

        aio_context_acquire(aio_context);
1855
        bdrv_close(bs);
1856
        aio_context_release(aio_context);
1857 1858 1859
    }
}

1860 1861 1862 1863 1864 1865
/* 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;
    }
1866 1867 1868 1869
    if (!qemu_co_queue_empty(&bs->throttled_reqs[0])) {
        return true;
    }
    if (!qemu_co_queue_empty(&bs->throttled_reqs[1])) {
1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880
        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;
}

1881 1882 1883 1884 1885
/*
 * 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.
1886 1887 1888 1889 1890 1891
 *
 * 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.
1892 1893 1894
 */
void bdrv_drain_all(void)
{
1895 1896
    /* Always run first iteration so any pending completion BHs run */
    bool busy = true;
1897 1898
    BlockDriverState *bs;

1899
    while (busy) {
1900 1901
        busy = false;

1902
        QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
1903 1904 1905 1906
            AioContext *aio_context = bdrv_get_aio_context(bs);
            bool bs_busy;

            aio_context_acquire(aio_context);
1907
            bdrv_start_throttled_reqs(bs);
1908 1909 1910
            bs_busy = bdrv_requests_pending(bs);
            bs_busy |= aio_poll(aio_context, bs_busy);
            aio_context_release(aio_context);
1911

1912 1913
            busy |= bs_busy;
        }
1914 1915 1916
    }
}

1917 1918
/* make a BlockDriverState anonymous by removing from bdrv_state and
 * graph_bdrv_state list.
1919 1920 1921 1922
   Also, NULL terminate the device_name to prevent double remove */
void bdrv_make_anon(BlockDriverState *bs)
{
    if (bs->device_name[0] != '\0') {
1923
        QTAILQ_REMOVE(&bdrv_states, bs, device_list);
1924 1925
    }
    bs->device_name[0] = '\0';
1926 1927 1928 1929
    if (bs->node_name[0] != '\0') {
        QTAILQ_REMOVE(&graph_bdrv_states, bs, node_list);
    }
    bs->node_name[0] = '\0';
1930 1931
}

P
Paolo Bonzini 已提交
1932 1933 1934 1935 1936 1937 1938
static void bdrv_rebind(BlockDriverState *bs)
{
    if (bs->drv && bs->drv->bdrv_rebind) {
        bs->drv->bdrv_rebind(bs);
    }
}

1939 1940
static void bdrv_move_feature_fields(BlockDriverState *bs_dest,
                                     BlockDriverState *bs_src)
1941
{
1942
    /* move some fields that need to stay attached to the device */
1943 1944

    /* dev info */
1945 1946 1947
    bs_dest->dev_ops            = bs_src->dev_ops;
    bs_dest->dev_opaque         = bs_src->dev_opaque;
    bs_dest->dev                = bs_src->dev;
1948
    bs_dest->guest_block_size   = bs_src->guest_block_size;
1949
    bs_dest->copy_on_read       = bs_src->copy_on_read;
1950

1951
    bs_dest->enable_write_cache = bs_src->enable_write_cache;
1952

1953 1954 1955 1956 1957 1958
    /* 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];
1959
    bs_dest->io_limits_enabled  = bs_src->io_limits_enabled;
1960 1961

    /* r/w error */
1962 1963
    bs_dest->on_read_error      = bs_src->on_read_error;
    bs_dest->on_write_error     = bs_src->on_write_error;
1964 1965

    /* i/o status */
1966 1967
    bs_dest->iostatus_enabled   = bs_src->iostatus_enabled;
    bs_dest->iostatus           = bs_src->iostatus;
1968

1969
    /* dirty bitmap */
F
Fam Zheng 已提交
1970
    bs_dest->dirty_bitmaps      = bs_src->dirty_bitmaps;
1971

1972 1973 1974
    /* reference count */
    bs_dest->refcnt             = bs_src->refcnt;

1975
    /* job */
1976
    bs_dest->job                = bs_src->job;
1977

1978
    /* keep the same entry in bdrv_states */
1979 1980
    pstrcpy(bs_dest->device_name, sizeof(bs_dest->device_name),
            bs_src->device_name);
1981
    bs_dest->device_list = bs_src->device_list;
1982 1983
    memcpy(bs_dest->op_blockers, bs_src->op_blockers,
           sizeof(bs_dest->op_blockers));
1984
}
1985

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000
/*
 * 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
2001

2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012
    /* The code needs to swap the node_name but simply swapping node_list won't
     * work so first remove the nodes from the graph list, do the swap then
     * insert them back if needed.
     */
    if (bs_new->node_name[0] != '\0') {
        QTAILQ_REMOVE(&graph_bdrv_states, bs_new, node_list);
    }
    if (bs_old->node_name[0] != '\0') {
        QTAILQ_REMOVE(&graph_bdrv_states, bs_old, node_list);
    }

2013 2014
    /* 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
2015
    assert(QLIST_EMPTY(&bs_new->dirty_bitmaps));
2016 2017 2018
    assert(bs_new->job == NULL);
    assert(bs_new->dev == NULL);
    assert(bs_new->io_limits_enabled == false);
2019
    assert(!throttle_have_timer(&bs_new->throttle_state));
2020

2021 2022 2023
    tmp = *bs_new;
    *bs_new = *bs_old;
    *bs_old = tmp;
2024

2025 2026 2027 2028
    /* 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);
2029

2030 2031 2032 2033 2034 2035 2036
    /* 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->io_limits_enabled == false);
2037
    assert(!throttle_have_timer(&bs_new->throttle_state));
P
Paolo Bonzini 已提交
2038

2039 2040 2041 2042 2043 2044 2045 2046
    /* insert the nodes back into the graph node list if needed */
    if (bs_new->node_name[0] != '\0') {
        QTAILQ_INSERT_TAIL(&graph_bdrv_states, bs_new, node_list);
    }
    if (bs_old->node_name[0] != '\0') {
        QTAILQ_INSERT_TAIL(&graph_bdrv_states, bs_old, node_list);
    }

P
Paolo Bonzini 已提交
2047
    bdrv_rebind(bs_new);
2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067
    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. */
F
Fam Zheng 已提交
2068
    bdrv_set_backing_hd(bs_top, bs_new);
2069 2070
}

F
Fam Zheng 已提交
2071
static void bdrv_delete(BlockDriverState *bs)
B
bellard 已提交
2072
{
2073
    assert(!bs->dev);
2074
    assert(!bs->job);
2075
    assert(bdrv_op_blocker_is_empty(bs));
F
Fam Zheng 已提交
2076
    assert(!bs->refcnt);
F
Fam Zheng 已提交
2077
    assert(QLIST_EMPTY(&bs->dirty_bitmaps));
2078

2079 2080
    bdrv_close(bs);

2081
    /* remove from list, if necessary */
2082
    bdrv_make_anon(bs);
2083

2084
    g_free(bs);
B
bellard 已提交
2085 2086
}

2087 2088
int bdrv_attach_dev(BlockDriverState *bs, void *dev)
/* TODO change to DeviceState *dev when all users are qdevified */
2089
{
2090
    if (bs->dev) {
2091 2092
        return -EBUSY;
    }
2093
    bs->dev = dev;
2094
    bdrv_iostatus_reset(bs);
2095 2096 2097
    return 0;
}

2098 2099
/* TODO qdevified devices don't use this, remove when devices are qdevified */
void bdrv_attach_dev_nofail(BlockDriverState *bs, void *dev)
2100
{
2101 2102 2103 2104 2105 2106 2107 2108 2109 2110
    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;
2111 2112
    bs->dev_ops = NULL;
    bs->dev_opaque = NULL;
2113
    bs->guest_block_size = 512;
2114 2115
}

2116 2117
/* TODO change to return DeviceState * when all users are qdevified */
void *bdrv_get_attached_dev(BlockDriverState *bs)
2118
{
2119
    return bs->dev;
2120 2121
}

2122 2123 2124 2125 2126 2127 2128
void bdrv_set_dev_ops(BlockDriverState *bs, const BlockDevOps *ops,
                      void *opaque)
{
    bs->dev_ops = ops;
    bs->dev_opaque = opaque;
}

2129
static void bdrv_dev_change_media_cb(BlockDriverState *bs, bool load)
2130
{
2131
    if (bs->dev_ops && bs->dev_ops->change_media_cb) {
2132
        bool tray_was_closed = !bdrv_dev_is_tray_open(bs);
2133
        bs->dev_ops->change_media_cb(bs->dev_opaque, load);
2134 2135
        if (tray_was_closed) {
            /* tray open */
2136 2137
            qapi_event_send_device_tray_moved(bdrv_get_device_name(bs),
                                              true, &error_abort);
2138 2139 2140
        }
        if (load) {
            /* tray close */
2141 2142
            qapi_event_send_device_tray_moved(bdrv_get_device_name(bs),
                                              false, &error_abort);
2143
        }
2144 2145 2146
    }
}

2147 2148 2149 2150 2151
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
2152 2153 2154 2155 2156 2157 2158
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);
    }
}

2159 2160 2161 2162 2163 2164 2165 2166
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;
}

2167 2168 2169 2170
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);
2171 2172 2173
    }
}

2174 2175 2176 2177 2178 2179 2180 2181
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
2182 2183 2184
/*
 * Run consistency checks on an image
 *
2185
 * Returns 0 if the check could be completed (it doesn't mean that the image is
2186
 * free of errors) or -errno when an internal error occurred. The results of the
2187
 * check are stored in res.
A
aliguori 已提交
2188
 */
2189
int bdrv_check(BlockDriverState *bs, BdrvCheckResult *res, BdrvCheckMode fix)
A
aliguori 已提交
2190 2191 2192 2193 2194
{
    if (bs->drv->bdrv_check == NULL) {
        return -ENOTSUP;
    }

2195
    memset(res, 0, sizeof(*res));
2196
    return bs->drv->bdrv_check(bs, res, fix);
A
aliguori 已提交
2197 2198
}

2199 2200
#define COMMIT_BUF_SECTORS 2048

2201 2202 2203
/* commit COW file into the raw image */
int bdrv_commit(BlockDriverState *bs)
{
B
bellard 已提交
2204
    BlockDriver *drv = bs->drv;
2205
    int64_t sector, total_sectors, length, backing_length;
2206
    int n, ro, open_flags;
2207
    int ret = 0;
2208
    uint8_t *buf = NULL;
2209
    char filename[PATH_MAX];
2210

B
bellard 已提交
2211 2212
    if (!drv)
        return -ENOMEDIUM;
2213 2214 2215
    
    if (!bs->backing_hd) {
        return -ENOTSUP;
2216 2217
    }

2218 2219
    if (bdrv_op_is_blocked(bs, BLOCK_OP_TYPE_COMMIT, NULL) ||
        bdrv_op_is_blocked(bs->backing_hd, BLOCK_OP_TYPE_COMMIT, NULL)) {
2220 2221 2222
        return -EBUSY;
    }

2223
    ro = bs->backing_hd->read_only;
2224 2225
    /* Use pstrcpy (not strncpy): filename must be NUL-terminated. */
    pstrcpy(filename, sizeof(filename), bs->backing_hd->filename);
2226 2227 2228
    open_flags =  bs->backing_hd->open_flags;

    if (ro) {
2229 2230
        if (bdrv_reopen(bs->backing_hd, open_flags | BDRV_O_RDWR, NULL)) {
            return -EACCES;
2231
        }
B
bellard 已提交
2232
    }
2233

2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256
    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;
2257
    buf = g_malloc(COMMIT_BUF_SECTORS * BDRV_SECTOR_SIZE);
2258 2259

    for (sector = 0; sector < total_sectors; sector += n) {
2260 2261 2262 2263 2264
        ret = bdrv_is_allocated(bs, sector, COMMIT_BUF_SECTORS, &n);
        if (ret < 0) {
            goto ro_cleanup;
        }
        if (ret) {
K
Kevin Wolf 已提交
2265 2266
            ret = bdrv_read(bs, sector, buf, n);
            if (ret < 0) {
2267 2268 2269
                goto ro_cleanup;
            }

K
Kevin Wolf 已提交
2270 2271
            ret = bdrv_write(bs->backing_hd, sector, buf, n);
            if (ret < 0) {
2272 2273
                goto ro_cleanup;
            }
B
bellard 已提交
2274
        }
2275
    }
2276

2277 2278
    if (drv->bdrv_make_empty) {
        ret = drv->bdrv_make_empty(bs);
K
Kevin Wolf 已提交
2279 2280 2281
        if (ret < 0) {
            goto ro_cleanup;
        }
2282 2283
        bdrv_flush(bs);
    }
2284

2285 2286 2287 2288
    /*
     * Make sure all data we wrote to the backing device is actually
     * stable on disk.
     */
K
Kevin Wolf 已提交
2289
    if (bs->backing_hd) {
2290
        bdrv_flush(bs->backing_hd);
K
Kevin Wolf 已提交
2291
    }
2292

K
Kevin Wolf 已提交
2293
    ret = 0;
2294
ro_cleanup:
2295
    g_free(buf);
2296 2297

    if (ro) {
2298 2299
        /* ignoring error return here */
        bdrv_reopen(bs->backing_hd, open_flags & ~BDRV_O_RDWR, NULL);
2300 2301
    }

2302
    return ret;
2303 2304
}

2305
int bdrv_commit_all(void)
2306 2307 2308
{
    BlockDriverState *bs;

2309
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
2310 2311 2312
        AioContext *aio_context = bdrv_get_aio_context(bs);

        aio_context_acquire(aio_context);
2313 2314 2315
        if (bs->drv && bs->backing_hd) {
            int ret = bdrv_commit(bs);
            if (ret < 0) {
2316
                aio_context_release(aio_context);
2317 2318
                return ret;
            }
2319
        }
2320
        aio_context_release(aio_context);
2321
    }
2322
    return 0;
2323 2324
}

S
Stefan Hajnoczi 已提交
2325 2326 2327 2328 2329 2330 2331
/**
 * 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)
{
2332 2333 2334 2335
    if (req->serialising) {
        req->bs->serialising_in_flight--;
    }

S
Stefan Hajnoczi 已提交
2336
    QLIST_REMOVE(req, list);
2337
    qemu_co_queue_restart_all(&req->wait_queue);
S
Stefan Hajnoczi 已提交
2338 2339 2340 2341 2342 2343 2344
}

/**
 * Add an active request to the tracked requests list
 */
static void tracked_request_begin(BdrvTrackedRequest *req,
                                  BlockDriverState *bs,
2345 2346
                                  int64_t offset,
                                  unsigned int bytes, bool is_write)
S
Stefan Hajnoczi 已提交
2347 2348 2349
{
    *req = (BdrvTrackedRequest){
        .bs = bs,
2350 2351 2352 2353 2354
        .offset         = offset,
        .bytes          = bytes,
        .is_write       = is_write,
        .co             = qemu_coroutine_self(),
        .serialising    = false,
2355 2356
        .overlap_offset = offset,
        .overlap_bytes  = bytes,
S
Stefan Hajnoczi 已提交
2357 2358
    };

2359 2360
    qemu_co_queue_init(&req->wait_queue);

S
Stefan Hajnoczi 已提交
2361 2362 2363
    QLIST_INSERT_HEAD(&bs->tracked_requests, req, list);
}

2364
static void mark_request_serialising(BdrvTrackedRequest *req, uint64_t align)
2365
{
2366
    int64_t overlap_offset = req->offset & ~(align - 1);
2367 2368
    unsigned int overlap_bytes = ROUND_UP(req->offset + req->bytes, align)
                               - overlap_offset;
2369

2370 2371 2372 2373
    if (!req->serialising) {
        req->bs->serialising_in_flight++;
        req->serialising = true;
    }
2374 2375 2376

    req->overlap_offset = MIN(req->overlap_offset, overlap_offset);
    req->overlap_bytes = MAX(req->overlap_bytes, overlap_bytes);
2377 2378
}

S
Stefan Hajnoczi 已提交
2379 2380 2381
/**
 * Round a region to cluster boundaries
 */
2382 2383 2384 2385
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
2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399
{
    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);
    }
}

2400
static int bdrv_get_cluster_size(BlockDriverState *bs)
2401 2402
{
    BlockDriverInfo bdi;
2403
    int ret;
2404

2405 2406 2407
    ret = bdrv_get_info(bs, &bdi);
    if (ret < 0 || bdi.cluster_size == 0) {
        return bs->request_alignment;
2408
    } else {
2409
        return bdi.cluster_size;
2410 2411 2412
    }
}

2413
static bool tracked_request_overlaps(BdrvTrackedRequest *req,
2414 2415
                                     int64_t offset, unsigned int bytes)
{
S
Stefan Hajnoczi 已提交
2416
    /*        aaaa   bbbb */
2417
    if (offset >= req->overlap_offset + req->overlap_bytes) {
S
Stefan Hajnoczi 已提交
2418 2419 2420
        return false;
    }
    /* bbbb   aaaa        */
2421
    if (req->overlap_offset >= offset + bytes) {
S
Stefan Hajnoczi 已提交
2422 2423 2424
        return false;
    }
    return true;
2425 2426
}

2427
static bool coroutine_fn wait_serialising_requests(BdrvTrackedRequest *self)
2428
{
2429
    BlockDriverState *bs = self->bs;
2430 2431
    BdrvTrackedRequest *req;
    bool retry;
2432
    bool waited = false;
2433

2434
    if (!bs->serialising_in_flight) {
2435
        return false;
2436 2437
    }

2438 2439 2440
    do {
        retry = false;
        QLIST_FOREACH(req, &bs->tracked_requests, list) {
2441
            if (req == self || (!req->serialising && !self->serialising)) {
2442 2443
                continue;
            }
2444 2445 2446
            if (tracked_request_overlaps(req, self->overlap_offset,
                                         self->overlap_bytes))
            {
2447 2448 2449 2450 2451 2452
                /* 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);

2453 2454 2455 2456 2457 2458 2459 2460
                /* 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;
2461
                    waited = true;
2462 2463
                    break;
                }
2464 2465 2466
            }
        }
    } while (retry);
2467 2468

    return waited;
2469 2470
}

K
Kevin Wolf 已提交
2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482
/*
 * 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;
2483
    int ret;
K
Kevin Wolf 已提交
2484

2485 2486 2487 2488 2489
    /* Backing file format doesn't make sense without a backing file */
    if (backing_fmt && !backing_file) {
        return -EINVAL;
    }

K
Kevin Wolf 已提交
2490
    if (drv->bdrv_change_backing_file != NULL) {
2491
        ret = drv->bdrv_change_backing_file(bs, backing_file, backing_fmt);
K
Kevin Wolf 已提交
2492
    } else {
2493
        ret = -ENOTSUP;
K
Kevin Wolf 已提交
2494
    }
2495 2496 2497 2498 2499 2500

    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
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 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627
/*
 * 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;
    }
2628
    bdrv_set_backing_hd(new_top_bs, base_bs);
2629 2630 2631

    QSIMPLEQ_FOREACH_SAFE(intermediate_state, &states_to_delete, entry, next) {
        /* so that bdrv_close() does not recursively close the chain */
2632
        bdrv_set_backing_hd(intermediate_state->bs, NULL);
F
Fam Zheng 已提交
2633
        bdrv_unref(intermediate_state->bs);
2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644
    }
    ret = 0;

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


2645 2646 2647 2648 2649
static int bdrv_check_byte_request(BlockDriverState *bs, int64_t offset,
                                   size_t size)
{
    int64_t len;

2650 2651 2652 2653
    if (size > INT_MAX) {
        return -EIO;
    }

2654 2655 2656 2657 2658 2659 2660 2661
    if (!bdrv_is_inserted(bs))
        return -ENOMEDIUM;

    if (bs->growable)
        return 0;

    len = bdrv_getlength(bs);

K
Kevin Wolf 已提交
2662 2663 2664 2665
    if (offset < 0)
        return -EIO;

    if ((offset > len) || (len - offset < size))
2666 2667 2668 2669 2670 2671 2672 2673
        return -EIO;

    return 0;
}

static int bdrv_check_request(BlockDriverState *bs, int64_t sector_num,
                              int nb_sectors)
{
2674
    if (nb_sectors < 0 || nb_sectors > INT_MAX / BDRV_SECTOR_SIZE) {
2675 2676 2677
        return -EIO;
    }

2678 2679
    return bdrv_check_byte_request(bs, sector_num * BDRV_SECTOR_SIZE,
                                   nb_sectors * BDRV_SECTOR_SIZE);
2680 2681
}

2682 2683
typedef struct RwCo {
    BlockDriverState *bs;
2684
    int64_t offset;
2685 2686 2687
    QEMUIOVector *qiov;
    bool is_write;
    int ret;
P
Peter Lieven 已提交
2688
    BdrvRequestFlags flags;
2689 2690 2691
} RwCo;

static void coroutine_fn bdrv_rw_co_entry(void *opaque)
B
bellard 已提交
2692
{
2693
    RwCo *rwco = opaque;
B
bellard 已提交
2694

2695
    if (!rwco->is_write) {
2696 2697
        rwco->ret = bdrv_co_do_preadv(rwco->bs, rwco->offset,
                                      rwco->qiov->size, rwco->qiov,
P
Peter Lieven 已提交
2698
                                      rwco->flags);
2699 2700 2701 2702
    } else {
        rwco->ret = bdrv_co_do_pwritev(rwco->bs, rwco->offset,
                                       rwco->qiov->size, rwco->qiov,
                                       rwco->flags);
2703 2704
    }
}
2705

2706
/*
2707
 * Process a vectored synchronous request using coroutines
2708
 */
2709 2710 2711
static int bdrv_prwv_co(BlockDriverState *bs, int64_t offset,
                        QEMUIOVector *qiov, bool is_write,
                        BdrvRequestFlags flags)
2712 2713 2714 2715
{
    Coroutine *co;
    RwCo rwco = {
        .bs = bs,
2716
        .offset = offset,
2717
        .qiov = qiov,
2718 2719
        .is_write = is_write,
        .ret = NOT_DONE,
P
Peter Lieven 已提交
2720
        .flags = flags,
2721
    };
2722

2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733
    /**
     * 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);
    }

2734 2735 2736 2737
    if (qemu_in_coroutine()) {
        /* Fast-path if already in coroutine context */
        bdrv_rw_co_entry(&rwco);
    } else {
2738 2739
        AioContext *aio_context = bdrv_get_aio_context(bs);

2740 2741 2742
        co = qemu_coroutine_create(bdrv_rw_co_entry);
        qemu_coroutine_enter(co, &rwco);
        while (rwco.ret == NOT_DONE) {
2743
            aio_poll(aio_context, true);
2744 2745 2746 2747
        }
    }
    return rwco.ret;
}
B
bellard 已提交
2748

2749 2750 2751 2752
/*
 * 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
2753
                      int nb_sectors, bool is_write, BdrvRequestFlags flags)
2754 2755 2756 2757 2758 2759 2760
{
    QEMUIOVector qiov;
    struct iovec iov = {
        .iov_base = (void *)buf,
        .iov_len = nb_sectors * BDRV_SECTOR_SIZE,
    };

2761 2762 2763 2764
    if (nb_sectors < 0 || nb_sectors > INT_MAX / BDRV_SECTOR_SIZE) {
        return -EINVAL;
    }

2765
    qemu_iovec_init_external(&qiov, &iov, 1);
2766 2767
    return bdrv_prwv_co(bs, sector_num << BDRV_SECTOR_BITS,
                        &qiov, is_write, flags);
2768 2769
}

2770 2771 2772 2773
/* 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
2774
    return bdrv_rw_co(bs, sector_num, buf, nb_sectors, false, 0);
B
bellard 已提交
2775 2776
}

2777 2778 2779 2780 2781 2782 2783 2784 2785
/* 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;
2786
    ret = bdrv_read(bs, sector_num, buf, nb_sectors);
2787 2788 2789 2790
    bs->io_limits_enabled = enabled;
    return ret;
}

2791
/* Return < 0 if error. Important errors are:
B
bellard 已提交
2792 2793 2794 2795 2796
  -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
*/
2797
int bdrv_write(BlockDriverState *bs, int64_t sector_num,
B
bellard 已提交
2798 2799
               const uint8_t *buf, int nb_sectors)
{
P
Peter Lieven 已提交
2800
    return bdrv_rw_co(bs, sector_num, (uint8_t *)buf, nb_sectors, true, 0);
B
bellard 已提交
2801 2802
}

2803 2804
int bdrv_write_zeroes(BlockDriverState *bs, int64_t sector_num,
                      int nb_sectors, BdrvRequestFlags flags)
P
Peter Lieven 已提交
2805 2806
{
    return bdrv_rw_co(bs, sector_num, NULL, nb_sectors, true,
2807
                      BDRV_REQ_ZERO_WRITE | flags);
2808 2809
}

P
Peter Lieven 已提交
2810 2811 2812 2813 2814 2815 2816 2817 2818 2819
/*
 * 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)
{
2820
    int64_t target_size;
P
Peter Lieven 已提交
2821 2822 2823
    int64_t ret, nb_sectors, sector_num = 0;
    int n;

2824 2825 2826 2827 2828 2829
    target_size = bdrv_getlength(bs);
    if (target_size < 0) {
        return target_size;
    }
    target_size /= BDRV_SECTOR_SIZE;

P
Peter Lieven 已提交
2830 2831 2832 2833 2834 2835 2836 2837 2838
    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);
2839 2840 2841 2842 2843
        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
2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857
        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;
    }
}

2858
int bdrv_pread(BlockDriverState *bs, int64_t offset, void *buf, int bytes)
B
bellard 已提交
2859
{
2860 2861 2862 2863 2864
    QEMUIOVector qiov;
    struct iovec iov = {
        .iov_base = (void *)buf,
        .iov_len = bytes,
    };
2865
    int ret;
B
bellard 已提交
2866

2867 2868
    if (bytes < 0) {
        return -EINVAL;
B
bellard 已提交
2869 2870
    }

2871 2872 2873 2874
    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
2875
    }
2876 2877

    return bytes;
B
bellard 已提交
2878 2879
}

2880
int bdrv_pwritev(BlockDriverState *bs, int64_t offset, QEMUIOVector *qiov)
B
bellard 已提交
2881
{
2882
    int ret;
B
bellard 已提交
2883

2884 2885 2886
    ret = bdrv_prwv_co(bs, offset, qiov, true, 0);
    if (ret < 0) {
        return ret;
B
bellard 已提交
2887 2888
    }

2889 2890 2891 2892
    return qiov->size;
}

int bdrv_pwrite(BlockDriverState *bs, int64_t offset,
2893
                const void *buf, int bytes)
2894 2895 2896 2897
{
    QEMUIOVector qiov;
    struct iovec iov = {
        .iov_base   = (void *) buf,
2898
        .iov_len    = bytes,
2899 2900
    };

2901 2902 2903 2904
    if (bytes < 0) {
        return -EINVAL;
    }

2905 2906
    qemu_iovec_init_external(&qiov, &iov, 1);
    return bdrv_pwritev(bs, offset, &qiov);
B
bellard 已提交
2907 2908
}

K
Kevin Wolf 已提交
2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924
/*
 * 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;
    }

2925 2926
    /* No flush needed for cache modes that already do it */
    if (bs->enable_write_cache) {
K
Kevin Wolf 已提交
2927 2928 2929 2930 2931 2932
        bdrv_flush(bs);
    }

    return 0;
}

2933
static int coroutine_fn bdrv_co_do_copy_on_readv(BlockDriverState *bs,
S
Stefan Hajnoczi 已提交
2934 2935 2936 2937 2938 2939 2940 2941 2942
        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;

2943
    BlockDriver *drv = bs->drv;
S
Stefan Hajnoczi 已提交
2944 2945 2946 2947 2948 2949 2950 2951 2952 2953
    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.
     */
2954 2955
    bdrv_round_to_clusters(bs, sector_num, nb_sectors,
                           &cluster_sector_num, &cluster_nb_sectors);
S
Stefan Hajnoczi 已提交
2956

2957 2958
    trace_bdrv_co_do_copy_on_readv(bs, sector_num, nb_sectors,
                                   cluster_sector_num, cluster_nb_sectors);
S
Stefan Hajnoczi 已提交
2959 2960 2961 2962 2963

    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);

2964 2965
    ret = drv->bdrv_co_readv(bs, cluster_sector_num, cluster_nb_sectors,
                             &bounce_qiov);
S
Stefan Hajnoczi 已提交
2966 2967 2968 2969
    if (ret < 0) {
        goto err;
    }

2970 2971
    if (drv->bdrv_co_write_zeroes &&
        buffer_is_zero(bounce_buffer, iov.iov_len)) {
K
Kevin Wolf 已提交
2972
        ret = bdrv_co_do_write_zeroes(bs, cluster_sector_num,
2973
                                      cluster_nb_sectors, 0);
2974
    } else {
2975 2976 2977
        /* This does not change the data on the disk, it is not necessary
         * to flush even in cache=writethrough mode.
         */
2978
        ret = drv->bdrv_co_writev(bs, cluster_sector_num, cluster_nb_sectors,
S
Stefan Hajnoczi 已提交
2979
                                  &bounce_qiov);
2980 2981
    }

S
Stefan Hajnoczi 已提交
2982 2983 2984 2985 2986 2987 2988 2989 2990
    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;
2991 2992
    qemu_iovec_from_buf(qiov, 0, bounce_buffer + skip_bytes,
                        nb_sectors * BDRV_SECTOR_SIZE);
S
Stefan Hajnoczi 已提交
2993 2994 2995 2996 2997 2998

err:
    qemu_vfree(bounce_buffer);
    return ret;
}

2999
/*
3000 3001 3002
 * 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.
3003
 */
3004
static int coroutine_fn bdrv_aligned_preadv(BlockDriverState *bs,
3005
    BdrvTrackedRequest *req, int64_t offset, unsigned int bytes,
3006
    int64_t align, QEMUIOVector *qiov, int flags)
K
Kevin Wolf 已提交
3007 3008
{
    BlockDriver *drv = bs->drv;
S
Stefan Hajnoczi 已提交
3009
    int ret;
K
Kevin Wolf 已提交
3010

3011 3012
    int64_t sector_num = offset >> BDRV_SECTOR_BITS;
    unsigned int nb_sectors = bytes >> BDRV_SECTOR_BITS;
K
Kevin Wolf 已提交
3013

3014 3015 3016 3017
    assert((offset & (BDRV_SECTOR_SIZE - 1)) == 0);
    assert((bytes & (BDRV_SECTOR_SIZE - 1)) == 0);

    /* Handle Copy on Read and associated serialisation */
3018
    if (flags & BDRV_REQ_COPY_ON_READ) {
3019 3020 3021 3022 3023 3024
        /* 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));
3025 3026
    }

3027
    wait_serialising_requests(req);
3028

3029
    if (flags & BDRV_REQ_COPY_ON_READ) {
S
Stefan Hajnoczi 已提交
3030 3031
        int pnum;

3032
        ret = bdrv_is_allocated(bs, sector_num, nb_sectors, &pnum);
S
Stefan Hajnoczi 已提交
3033 3034 3035 3036 3037
        if (ret < 0) {
            goto out;
        }

        if (!ret || pnum != nb_sectors) {
3038
            ret = bdrv_co_do_copy_on_readv(bs, sector_num, nb_sectors, qiov);
S
Stefan Hajnoczi 已提交
3039 3040 3041 3042
            goto out;
        }
    }

3043
    /* Forward the request to the BlockDriver */
3044 3045 3046 3047 3048 3049 3050 3051 3052 3053 3054 3055
    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;
        }

3056
        total_sectors = DIV_ROUND_UP(len, BDRV_SECTOR_SIZE);
3057 3058
        max_nb_sectors = ROUND_UP(MAX(0, total_sectors - sector_num),
                                  align >> BDRV_SECTOR_BITS);
3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070 3071 3072 3073
        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
3074 3075

out:
S
Stefan Hajnoczi 已提交
3076
    return ret;
K
Kevin Wolf 已提交
3077 3078
}

3079 3080 3081
/*
 * Handle a read request in coroutine context
 */
3082 3083
static int coroutine_fn bdrv_co_do_preadv(BlockDriverState *bs,
    int64_t offset, unsigned int bytes, QEMUIOVector *qiov,
3084 3085 3086
    BdrvRequestFlags flags)
{
    BlockDriver *drv = bs->drv;
3087 3088
    BdrvTrackedRequest req;

3089 3090 3091 3092 3093 3094
    /* 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;
3095 3096 3097 3098 3099
    int ret;

    if (!drv) {
        return -ENOMEDIUM;
    }
3100
    if (bdrv_check_byte_request(bs, offset, bytes)) {
3101 3102 3103 3104 3105 3106 3107 3108 3109
        return -EIO;
    }

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

    /* throttling disk I/O */
    if (bs->io_limits_enabled) {
3110
        bdrv_io_limits_intercept(bs, bytes, false);
3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137
    }

    /* 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);
    }

3138
    tracked_request_begin(&req, bs, offset, bytes, false);
3139
    ret = bdrv_aligned_preadv(bs, &req, offset, bytes, align,
3140 3141
                              use_local_qiov ? &local_qiov : qiov,
                              flags);
3142
    tracked_request_end(&req);
3143 3144 3145 3146 3147

    if (use_local_qiov) {
        qemu_iovec_destroy(&local_qiov);
        qemu_vfree(head_buf);
        qemu_vfree(tail_buf);
3148 3149 3150 3151 3152
    }

    return ret;
}

3153 3154 3155 3156 3157 3158 3159 3160 3161 3162 3163 3164
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);
}

3165
int coroutine_fn bdrv_co_readv(BlockDriverState *bs, int64_t sector_num,
K
Kevin Wolf 已提交
3166 3167
    int nb_sectors, QEMUIOVector *qiov)
{
3168
    trace_bdrv_co_readv(bs, sector_num, nb_sectors);
K
Kevin Wolf 已提交
3169

3170 3171 3172 3173 3174 3175 3176 3177 3178 3179
    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);
3180 3181
}

3182 3183 3184 3185 3186
/* 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

3187
static int coroutine_fn bdrv_co_do_write_zeroes(BlockDriverState *bs,
3188
    int64_t sector_num, int nb_sectors, BdrvRequestFlags flags)
3189 3190 3191
{
    BlockDriver *drv = bs->drv;
    QEMUIOVector qiov;
3192 3193
    struct iovec iov = {0};
    int ret = 0;
3194

3195 3196
    int max_write_zeroes = bs->bl.max_write_zeroes ?
                           bs->bl.max_write_zeroes : MAX_WRITE_ZEROES_DEFAULT;
K
Kevin Wolf 已提交
3197

3198 3199 3200
    while (nb_sectors > 0 && !ret) {
        int num = nb_sectors;

3201 3202 3203 3204 3205 3206 3207
        /* 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.  */
3208
                num = bs->bl.write_zeroes_alignment;
3209 3210 3211 3212 3213 3214
                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;
3215
            }
K
Kevin Wolf 已提交
3216
        }
3217

3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232
        /* 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) {
3233 3234
                iov.iov_base = qemu_blockalign(bs, num * BDRV_SECTOR_SIZE);
                memset(iov.iov_base, 0, num * BDRV_SECTOR_SIZE);
3235 3236
            }
            qemu_iovec_init_external(&qiov, &iov, 1);
3237

3238
            ret = drv->bdrv_co_writev(bs, sector_num, num, &qiov);
3239 3240 3241 3242 3243 3244 3245 3246

            /* 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;
            }
3247 3248 3249 3250 3251
        }

        sector_num += num;
        nb_sectors -= num;
    }
3252 3253 3254 3255 3256

    qemu_vfree(iov.iov_base);
    return ret;
}

3257
/*
3258
 * Forwards an already correctly aligned write request to the BlockDriver.
3259
 */
3260
static int coroutine_fn bdrv_aligned_pwritev(BlockDriverState *bs,
3261 3262
    BdrvTrackedRequest *req, int64_t offset, unsigned int bytes,
    QEMUIOVector *qiov, int flags)
3263 3264
{
    BlockDriver *drv = bs->drv;
3265
    bool waited;
3266
    int ret;
K
Kevin Wolf 已提交
3267

3268 3269
    int64_t sector_num = offset >> BDRV_SECTOR_BITS;
    unsigned int nb_sectors = bytes >> BDRV_SECTOR_BITS;
3270

3271 3272
    assert((offset & (BDRV_SECTOR_SIZE - 1)) == 0);
    assert((bytes & (BDRV_SECTOR_SIZE - 1)) == 0);
3273

3274 3275
    waited = wait_serialising_requests(req);
    assert(!waited || !req->serialising);
3276 3277
    assert(req->overlap_offset <= offset);
    assert(offset + bytes <= req->overlap_offset + req->overlap_bytes);
3278

3279
    ret = notifier_with_return_list_notify(&bs->before_write_notifiers, req);
3280

3281 3282 3283 3284 3285 3286 3287 3288 3289
    if (!ret && bs->detect_zeroes != BLOCKDEV_DETECT_ZEROES_OPTIONS_OFF &&
        !(flags & BDRV_REQ_ZERO_WRITE) && drv->bdrv_co_write_zeroes &&
        qemu_iovec_is_zero(qiov)) {
        flags |= BDRV_REQ_ZERO_WRITE;
        if (bs->detect_zeroes == BLOCKDEV_DETECT_ZEROES_OPTIONS_UNMAP) {
            flags |= BDRV_REQ_MAY_UNMAP;
        }
    }

3290 3291 3292
    if (ret < 0) {
        /* Do nothing, write notifier decided to fail this request */
    } else if (flags & BDRV_REQ_ZERO_WRITE) {
3293
        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_ZERO);
3294
        ret = bdrv_co_do_write_zeroes(bs, sector_num, nb_sectors, flags);
3295
    } else {
3296
        BLKDBG_EVENT(bs, BLKDBG_PWRITEV);
3297 3298
        ret = drv->bdrv_co_writev(bs, sector_num, nb_sectors, qiov);
    }
3299
    BLKDBG_EVENT(bs, BLKDBG_PWRITEV_DONE);
3300

3301 3302 3303 3304
    if (ret == 0 && !bs->enable_write_cache) {
        ret = bdrv_co_flush(bs);
    }

F
Fam Zheng 已提交
3305
    bdrv_set_dirty(bs, sector_num, nb_sectors);
K
Kevin Wolf 已提交
3306 3307 3308 3309

    if (bs->wr_highest_sector < sector_num + nb_sectors - 1) {
        bs->wr_highest_sector = sector_num + nb_sectors - 1;
    }
3310 3311 3312
    if (bs->growable && ret >= 0) {
        bs->total_sectors = MAX(bs->total_sectors, sector_num + nb_sectors);
    }
K
Kevin Wolf 已提交
3313

3314
    return ret;
K
Kevin Wolf 已提交
3315 3316
}

3317 3318 3319
/*
 * Handle a write request in coroutine context
 */
3320 3321
static int coroutine_fn bdrv_co_do_pwritev(BlockDriverState *bs,
    int64_t offset, unsigned int bytes, QEMUIOVector *qiov,
3322 3323
    BdrvRequestFlags flags)
{
3324
    BdrvTrackedRequest req;
3325 3326 3327 3328 3329 3330
    /* 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;
3331 3332 3333 3334 3335 3336 3337 3338
    int ret;

    if (!bs->drv) {
        return -ENOMEDIUM;
    }
    if (bs->read_only) {
        return -EACCES;
    }
3339
    if (bdrv_check_byte_request(bs, offset, bytes)) {
3340 3341 3342 3343 3344
        return -EIO;
    }

    /* throttling disk I/O */
    if (bs->io_limits_enabled) {
3345
        bdrv_io_limits_intercept(bs, bytes, true);
3346 3347
    }

3348 3349 3350 3351 3352
    /*
     * 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.
     */
3353
    tracked_request_begin(&req, bs, offset, bytes, true);
3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365 3366 3367 3368

    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);

3369
        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_HEAD);
3370 3371 3372 3373 3374
        ret = bdrv_aligned_preadv(bs, &req, offset & ~(align - 1), align,
                                  align, &head_qiov, 0);
        if (ret < 0) {
            goto fail;
        }
3375
        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_AFTER_HEAD);
3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388 3389

        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;
3390
        bool waited;
3391 3392

        mark_request_serialising(&req, align);
3393 3394
        waited = wait_serialising_requests(&req);
        assert(!waited || !use_local_qiov);
3395 3396 3397 3398 3399 3400 3401 3402

        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);

3403
        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_TAIL);
3404 3405 3406 3407 3408
        ret = bdrv_aligned_preadv(bs, &req, (offset + bytes) & ~(align - 1), align,
                                  align, &tail_qiov, 0);
        if (ret < 0) {
            goto fail;
        }
3409
        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_AFTER_TAIL);
3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427

        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:
3428
    tracked_request_end(&req);
3429

3430 3431 3432
    if (use_local_qiov) {
        qemu_iovec_destroy(&local_qiov);
    }
3433 3434
    qemu_vfree(head_buf);
    qemu_vfree(tail_buf);
3435

3436 3437 3438
    return ret;
}

3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450
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);
}

3451 3452 3453 3454 3455
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);

3456 3457 3458 3459
    return bdrv_co_do_writev(bs, sector_num, nb_sectors, qiov, 0);
}

int coroutine_fn bdrv_co_write_zeroes(BlockDriverState *bs,
3460 3461
                                      int64_t sector_num, int nb_sectors,
                                      BdrvRequestFlags flags)
3462
{
3463
    trace_bdrv_co_write_zeroes(bs, sector_num, nb_sectors, flags);
3464

3465 3466 3467 3468
    if (!(bs->open_flags & BDRV_O_UNMAP)) {
        flags &= ~BDRV_REQ_MAY_UNMAP;
    }

3469
    return bdrv_co_do_writev(bs, sector_num, nb_sectors, NULL,
3470
                             BDRV_REQ_ZERO_WRITE | flags);
3471 3472
}

B
bellard 已提交
3473 3474 3475 3476 3477 3478
/**
 * Truncate file to 'offset' bytes (needed only for file protocols)
 */
int bdrv_truncate(BlockDriverState *bs, int64_t offset)
{
    BlockDriver *drv = bs->drv;
3479
    int ret;
B
bellard 已提交
3480
    if (!drv)
B
bellard 已提交
3481
        return -ENOMEDIUM;
B
bellard 已提交
3482 3483
    if (!drv->bdrv_truncate)
        return -ENOTSUP;
3484 3485
    if (bs->read_only)
        return -EACCES;
3486
    if (bdrv_op_is_blocked(bs, BLOCK_OP_TYPE_RESIZE, NULL)) {
M
Marcelo Tosatti 已提交
3487
        return -EBUSY;
3488
    }
3489 3490 3491
    ret = drv->bdrv_truncate(bs, offset);
    if (ret == 0) {
        ret = refresh_total_sectors(bs, offset >> BDRV_SECTOR_BITS);
3492
        bdrv_dev_resize_cb(bs);
3493 3494
    }
    return ret;
B
bellard 已提交
3495 3496
}

3497 3498 3499 3500 3501 3502 3503 3504 3505 3506 3507 3508 3509 3510 3511 3512 3513 3514 3515
/**
 * 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
3516 3517 3518 3519 3520 3521 3522
/**
 * 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
3523
        return -ENOMEDIUM;
3524

3525 3526 3527 3528
    if (drv->has_variable_length) {
        int ret = refresh_total_sectors(bs, bs->total_sectors);
        if (ret < 0) {
            return ret;
3529
        }
B
bellard 已提交
3530
    }
3531
    return bs->total_sectors * BDRV_SECTOR_SIZE;
B
bellard 已提交
3532 3533
}

B
bellard 已提交
3534
/* return 0 as number of sectors if no device present or error */
3535
void bdrv_get_geometry(BlockDriverState *bs, uint64_t *nb_sectors_ptr)
B
bellard 已提交
3536
{
B
bellard 已提交
3537 3538 3539 3540 3541
    int64_t length;
    length = bdrv_getlength(bs);
    if (length < 0)
        length = 0;
    else
J
Jan Kiszka 已提交
3542
        length = length >> BDRV_SECTOR_BITS;
B
bellard 已提交
3543
    *nb_sectors_ptr = length;
B
bellard 已提交
3544
}
B
bellard 已提交
3545

3546 3547
void bdrv_set_on_error(BlockDriverState *bs, BlockdevOnError on_read_error,
                       BlockdevOnError on_write_error)
3548 3549 3550 3551 3552
{
    bs->on_read_error = on_read_error;
    bs->on_write_error = on_write_error;
}

3553
BlockdevOnError bdrv_get_on_error(BlockDriverState *bs, bool is_read)
3554 3555 3556 3557
{
    return is_read ? bs->on_read_error : bs->on_write_error;
}

3558 3559 3560 3561 3562 3563
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:
W
Wenchao Xia 已提交
3564 3565
        return (error == ENOSPC) ?
               BLOCK_ERROR_ACTION_STOP : BLOCK_ERROR_ACTION_REPORT;
3566
    case BLOCKDEV_ON_ERROR_STOP:
W
Wenchao Xia 已提交
3567
        return BLOCK_ERROR_ACTION_STOP;
3568
    case BLOCKDEV_ON_ERROR_REPORT:
W
Wenchao Xia 已提交
3569
        return BLOCK_ERROR_ACTION_REPORT;
3570
    case BLOCKDEV_ON_ERROR_IGNORE:
W
Wenchao Xia 已提交
3571
        return BLOCK_ERROR_ACTION_IGNORE;
3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584
    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);
3585

W
Wenchao Xia 已提交
3586
    if (action == BLOCK_ERROR_ACTION_STOP) {
3587 3588 3589 3590
        /* First set the iostatus, so that "info block" returns an iostatus
         * that matches the events raised so far (an additional error iostatus
         * is fine, but not a lost one).
         */
3591
        bdrv_iostatus_set_err(bs, error);
3592 3593 3594 3595 3596 3597 3598 3599 3600 3601

        /* Then raise the request to stop the VM and the event.
         * qemu_system_vmstop_request_prepare has two effects.  First,
         * it ensures that the STOP event always comes after the
         * BLOCK_IO_ERROR event.  Second, it ensures that even if management
         * can observe the STOP event and do a "cont" before the STOP
         * event is issued, the VM will not stop.  In this case, vm_start()
         * also ensures that the STOP/RESUME pair of events is emitted.
         */
        qemu_system_vmstop_request_prepare();
3602 3603 3604 3605
        qapi_event_send_block_io_error(bdrv_get_device_name(bs),
                                       is_read ? IO_OPERATION_TYPE_READ :
                                       IO_OPERATION_TYPE_WRITE,
                                       action, &error_abort);
3606 3607
        qemu_system_vmstop_request(RUN_STATE_IO_ERROR);
    } else {
3608 3609 3610 3611
        qapi_event_send_block_io_error(bdrv_get_device_name(bs),
                                       is_read ? IO_OPERATION_TYPE_READ :
                                       IO_OPERATION_TYPE_WRITE,
                                       action, &error_abort);
3612 3613 3614
    }
}

B
bellard 已提交
3615 3616 3617 3618 3619
int bdrv_is_read_only(BlockDriverState *bs)
{
    return bs->read_only;
}

3620 3621 3622 3623 3624
int bdrv_is_sg(BlockDriverState *bs)
{
    return bs->sg;
}

3625 3626 3627 3628 3629
int bdrv_enable_write_cache(BlockDriverState *bs)
{
    return bs->enable_write_cache;
}

3630 3631 3632
void bdrv_set_enable_write_cache(BlockDriverState *bs, bool wce)
{
    bs->enable_write_cache = wce;
3633 3634 3635 3636 3637 3638 3639

    /* so a reopen() will preserve wce */
    if (wce) {
        bs->open_flags |= BDRV_O_CACHE_WB;
    } else {
        bs->open_flags &= ~BDRV_O_CACHE_WB;
    }
3640 3641
}

B
bellard 已提交
3642 3643 3644 3645 3646 3647 3648
int bdrv_is_encrypted(BlockDriverState *bs)
{
    if (bs->backing_hd && bs->backing_hd->encrypted)
        return 1;
    return bs->encrypted;
}

3649 3650 3651 3652 3653 3654 3655 3656 3657
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
3658 3659 3660 3661 3662 3663 3664 3665 3666 3667
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;
    }
3668 3669 3670 3671 3672
    if (!bs->encrypted) {
        return -EINVAL;
    } else if (!bs->drv || !bs->drv->bdrv_set_key) {
        return -ENOMEDIUM;
    }
3673
    ret = bs->drv->bdrv_set_key(bs, key);
3674 3675 3676 3677 3678
    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 */
3679
        bdrv_dev_change_media_cb(bs, true);
3680
    }
3681
    return ret;
B
bellard 已提交
3682 3683
}

3684
const char *bdrv_get_format_name(BlockDriverState *bs)
B
bellard 已提交
3685
{
3686
    return bs->drv ? bs->drv->format_name : NULL;
B
bellard 已提交
3687 3688
}

3689
void bdrv_iterate_format(void (*it)(void *opaque, const char *name),
B
bellard 已提交
3690 3691 3692
                         void *opaque)
{
    BlockDriver *drv;
3693 3694
    int count = 0;
    const char **formats = NULL;
B
bellard 已提交
3695

3696
    QLIST_FOREACH(drv, &bdrv_drivers, list) {
3697 3698 3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709
        if (drv->format_name) {
            bool found = false;
            int i = count;
            while (formats && i && !found) {
                found = !strcmp(formats[--i], drv->format_name);
            }

            if (!found) {
                formats = g_realloc(formats, (count + 1) * sizeof(char *));
                formats[count++] = drv->format_name;
                it(opaque, drv->format_name);
            }
        }
B
bellard 已提交
3710
    }
3711
    g_free(formats);
B
bellard 已提交
3712 3713
}

3714
/* This function is to find block backend bs */
B
bellard 已提交
3715 3716 3717 3718
BlockDriverState *bdrv_find(const char *name)
{
    BlockDriverState *bs;

3719
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
3720
        if (!strcmp(name, bs->device_name)) {
B
bellard 已提交
3721
            return bs;
3722
        }
B
bellard 已提交
3723 3724 3725 3726
    }
    return NULL;
}

3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741
/* 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;
}

3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758
/* 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;
}

3759 3760 3761 3762 3763 3764 3765 3766 3767
BlockDriverState *bdrv_lookup_bs(const char *device,
                                 const char *node_name,
                                 Error **errp)
{
    BlockDriverState *bs = NULL;

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

3768 3769
        if (bs) {
            return bs;
3770 3771 3772
        }
    }

3773 3774
    if (node_name) {
        bs = bdrv_find_node(node_name);
3775

3776 3777 3778
        if (bs) {
            return bs;
        }
3779 3780
    }

3781 3782 3783 3784
    error_setg(errp, "Cannot find device=%s nor node_name=%s",
                     device ? device : "",
                     node_name ? node_name : "");
    return NULL;
3785 3786
}

M
Markus Armbruster 已提交
3787 3788 3789 3790 3791
BlockDriverState *bdrv_next(BlockDriverState *bs)
{
    if (!bs) {
        return QTAILQ_FIRST(&bdrv_states);
    }
3792
    return QTAILQ_NEXT(bs, device_list);
M
Markus Armbruster 已提交
3793 3794
}

3795
void bdrv_iterate(void (*it)(void *opaque, BlockDriverState *bs), void *opaque)
B
bellard 已提交
3796 3797 3798
{
    BlockDriverState *bs;

3799
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
3800
        it(opaque, bs);
B
bellard 已提交
3801 3802 3803
    }
}

B
bellard 已提交
3804 3805 3806 3807 3808
const char *bdrv_get_device_name(BlockDriverState *bs)
{
    return bs->device_name;
}

M
Markus Armbruster 已提交
3809 3810 3811 3812 3813
int bdrv_get_flags(BlockDriverState *bs)
{
    return bs->open_flags;
}

3814
int bdrv_flush_all(void)
A
aliguori 已提交
3815 3816
{
    BlockDriverState *bs;
3817
    int result = 0;
A
aliguori 已提交
3818

3819
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
3820 3821 3822 3823 3824
        AioContext *aio_context = bdrv_get_aio_context(bs);
        int ret;

        aio_context_acquire(aio_context);
        ret = bdrv_flush(bs);
3825 3826 3827
        if (ret < 0 && !result) {
            result = ret;
        }
3828
        aio_context_release(aio_context);
3829
    }
3830 3831

    return result;
A
aliguori 已提交
3832 3833
}

3834 3835 3836 3837 3838
int bdrv_has_zero_init_1(BlockDriverState *bs)
{
    return 1;
}

K
Kevin Wolf 已提交
3839 3840 3841 3842
int bdrv_has_zero_init(BlockDriverState *bs)
{
    assert(bs->drv);

3843 3844 3845 3846 3847
    /* 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
3848 3849
    if (bs->drv->bdrv_has_zero_init) {
        return bs->drv->bdrv_has_zero_init(bs);
K
Kevin Wolf 已提交
3850 3851
    }

3852 3853
    /* safe default */
    return 0;
K
Kevin Wolf 已提交
3854 3855
}

3856 3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885
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;
}

3886
typedef struct BdrvCoGetBlockStatusData {
3887
    BlockDriverState *bs;
3888
    BlockDriverState *base;
3889 3890 3891
    int64_t sector_num;
    int nb_sectors;
    int *pnum;
3892
    int64_t ret;
3893
    bool done;
3894
} BdrvCoGetBlockStatusData;
3895

3896 3897 3898 3899 3900
/*
 * 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.
 *
3901 3902 3903
 * If 'sector_num' is beyond the end of the disk image the return value is 0
 * and 'pnum' is set to 0.
 *
3904 3905 3906 3907
 * '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.
 *
3908 3909
 * '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.
3910
 */
3911 3912 3913
static int64_t coroutine_fn bdrv_co_get_block_status(BlockDriverState *bs,
                                                     int64_t sector_num,
                                                     int nb_sectors, int *pnum)
3914
{
3915
    int64_t length;
3916
    int64_t n;
3917
    int64_t ret, ret2;
3918

3919 3920 3921 3922 3923 3924
    length = bdrv_getlength(bs);
    if (length < 0) {
        return length;
    }

    if (sector_num >= (length >> BDRV_SECTOR_BITS)) {
3925 3926 3927 3928 3929 3930 3931 3932 3933
        *pnum = 0;
        return 0;
    }

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

3934
    if (!bs->drv->bdrv_co_get_block_status) {
3935
        *pnum = nb_sectors;
3936
        ret = BDRV_BLOCK_DATA | BDRV_BLOCK_ALLOCATED;
3937 3938 3939 3940
        if (bs->drv->protocol_name) {
            ret |= BDRV_BLOCK_OFFSET_VALID | (sector_num * BDRV_SECTOR_SIZE);
        }
        return ret;
3941
    }
3942

3943 3944
    ret = bs->drv->bdrv_co_get_block_status(bs, sector_num, nb_sectors, pnum);
    if (ret < 0) {
3945
        *pnum = 0;
3946 3947 3948
        return ret;
    }

3949 3950 3951 3952 3953 3954
    if (ret & BDRV_BLOCK_RAW) {
        assert(ret & BDRV_BLOCK_OFFSET_VALID);
        return bdrv_get_block_status(bs->file, ret >> BDRV_SECTOR_BITS,
                                     *pnum, pnum);
    }

3955 3956 3957 3958
    if (ret & (BDRV_BLOCK_DATA | BDRV_BLOCK_ZERO)) {
        ret |= BDRV_BLOCK_ALLOCATED;
    }

3959 3960
    if (!(ret & BDRV_BLOCK_DATA) && !(ret & BDRV_BLOCK_ZERO)) {
        if (bdrv_unallocated_blocks_are_zero(bs)) {
3961
            ret |= BDRV_BLOCK_ZERO;
3962
        } else if (bs->backing_hd) {
3963 3964 3965 3966 3967 3968
            BlockDriverState *bs2 = bs->backing_hd;
            int64_t length2 = bdrv_getlength(bs2);
            if (length2 >= 0 && sector_num >= (length2 >> BDRV_SECTOR_BITS)) {
                ret |= BDRV_BLOCK_ZERO;
            }
        }
3969
    }
3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983

    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);
        }
    }

3984
    return ret;
3985 3986
}

3987 3988
/* Coroutine wrapper for bdrv_get_block_status() */
static void coroutine_fn bdrv_get_block_status_co_entry(void *opaque)
3989
{
3990
    BdrvCoGetBlockStatusData *data = opaque;
3991 3992
    BlockDriverState *bs = data->bs;

3993 3994
    data->ret = bdrv_co_get_block_status(bs, data->sector_num, data->nb_sectors,
                                         data->pnum);
3995 3996 3997 3998
    data->done = true;
}

/*
3999
 * Synchronous wrapper around bdrv_co_get_block_status().
4000
 *
4001
 * See bdrv_co_get_block_status() for details.
4002
 */
4003 4004
int64_t bdrv_get_block_status(BlockDriverState *bs, int64_t sector_num,
                              int nb_sectors, int *pnum)
4005
{
4006
    Coroutine *co;
4007
    BdrvCoGetBlockStatusData data = {
4008 4009 4010 4011 4012 4013 4014
        .bs = bs,
        .sector_num = sector_num,
        .nb_sectors = nb_sectors,
        .pnum = pnum,
        .done = false,
    };

4015 4016
    if (qemu_in_coroutine()) {
        /* Fast-path if already in coroutine context */
4017
        bdrv_get_block_status_co_entry(&data);
4018
    } else {
4019 4020
        AioContext *aio_context = bdrv_get_aio_context(bs);

4021
        co = qemu_coroutine_create(bdrv_get_block_status_co_entry);
4022 4023
        qemu_coroutine_enter(co, &data);
        while (!data.done) {
4024
            aio_poll(aio_context, true);
4025
        }
4026 4027
    }
    return data.ret;
4028 4029
}

4030 4031 4032
int coroutine_fn bdrv_is_allocated(BlockDriverState *bs, int64_t sector_num,
                                   int nb_sectors, int *pnum)
{
4033 4034 4035 4036
    int64_t ret = bdrv_get_block_status(bs, sector_num, nb_sectors, pnum);
    if (ret < 0) {
        return ret;
    }
4037
    return (ret & BDRV_BLOCK_ALLOCATED);
4038 4039
}

4040 4041 4042 4043 4044 4045 4046 4047 4048 4049 4050 4051
/*
 * 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.
 *
 */
4052 4053 4054 4055
int bdrv_is_allocated_above(BlockDriverState *top,
                            BlockDriverState *base,
                            int64_t sector_num,
                            int nb_sectors, int *pnum)
4056 4057 4058 4059 4060 4061 4062
{
    BlockDriverState *intermediate;
    int ret, n = nb_sectors;

    intermediate = top;
    while (intermediate && intermediate != base) {
        int pnum_inter;
4063 4064
        ret = bdrv_is_allocated(intermediate, sector_num, nb_sectors,
                                &pnum_inter);
4065 4066 4067 4068 4069 4070 4071 4072 4073 4074 4075 4076 4077
        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.
         */
4078 4079 4080
        if (n > pnum_inter &&
            (intermediate == top ||
             sector_num + pnum_inter < intermediate->total_sectors)) {
4081 4082 4083 4084 4085 4086 4087 4088 4089 4090
            n = pnum_inter;
        }

        intermediate = intermediate->backing_hd;
    }

    *pnum = n;
    return 0;
}

4091 4092 4093 4094 4095 4096 4097 4098 4099 4100
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;
}

4101
void bdrv_get_backing_filename(BlockDriverState *bs,
B
bellard 已提交
4102 4103
                               char *filename, int filename_size)
{
K
Kevin Wolf 已提交
4104
    pstrcpy(filename, filename_size, bs->backing_file);
B
bellard 已提交
4105 4106
}

4107
int bdrv_write_compressed(BlockDriverState *bs, int64_t sector_num,
B
bellard 已提交
4108 4109 4110 4111
                          const uint8_t *buf, int nb_sectors)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
4112
        return -ENOMEDIUM;
B
bellard 已提交
4113 4114
    if (!drv->bdrv_write_compressed)
        return -ENOTSUP;
K
Kevin Wolf 已提交
4115 4116
    if (bdrv_check_request(bs, sector_num, nb_sectors))
        return -EIO;
4117

F
Fam Zheng 已提交
4118
    assert(QLIST_EMPTY(&bs->dirty_bitmaps));
4119

B
bellard 已提交
4120 4121
    return drv->bdrv_write_compressed(bs, sector_num, buf, nb_sectors);
}
4122

B
bellard 已提交
4123 4124 4125 4126
int bdrv_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
4127
        return -ENOMEDIUM;
B
bellard 已提交
4128 4129 4130 4131 4132 4133
    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
4134 4135 4136 4137 4138 4139 4140 4141 4142
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;
}

4143 4144
int bdrv_save_vmstate(BlockDriverState *bs, const uint8_t *buf,
                      int64_t pos, int size)
4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156
{
    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)
4157 4158
{
    BlockDriver *drv = bs->drv;
4159 4160

    if (!drv) {
4161
        return -ENOMEDIUM;
4162 4163 4164 4165 4166 4167
    } 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);
    }

4168
    return -ENOTSUP;
4169 4170
}

4171 4172
int bdrv_load_vmstate(BlockDriverState *bs, uint8_t *buf,
                      int64_t pos, int size)
4173 4174 4175 4176
{
    BlockDriver *drv = bs->drv;
    if (!drv)
        return -ENOMEDIUM;
4177 4178 4179 4180 4181
    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;
4182 4183
}

K
Kevin Wolf 已提交
4184 4185
void bdrv_debug_event(BlockDriverState *bs, BlkDebugEvent event)
{
4186
    if (!bs || !bs->drv || !bs->drv->bdrv_debug_event) {
K
Kevin Wolf 已提交
4187 4188 4189
        return;
    }

4190
    bs->drv->bdrv_debug_event(bs, event);
K
Kevin Wolf 已提交
4191 4192 4193 4194 4195 4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206
}

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
4207 4208 4209 4210 4211 4212 4213 4214 4215 4216 4217 4218 4219
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
4220 4221
int bdrv_debug_resume(BlockDriverState *bs, const char *tag)
{
4222
    while (bs && (!bs->drv || !bs->drv->bdrv_debug_resume)) {
K
Kevin Wolf 已提交
4223 4224
        bs = bs->file;
    }
K
Kevin Wolf 已提交
4225

K
Kevin Wolf 已提交
4226 4227 4228 4229 4230 4231 4232 4233 4234 4235 4236 4237 4238 4239 4240 4241 4242 4243
    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
4244 4245
}

4246 4247 4248 4249 4250
int bdrv_is_snapshot(BlockDriverState *bs)
{
    return !!(bs->open_flags & BDRV_O_SNAPSHOT);
}

4251 4252 4253 4254
/* 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. */
4255 4256 4257
BlockDriverState *bdrv_find_backing_image(BlockDriverState *bs,
        const char *backing_file)
{
4258 4259 4260 4261 4262 4263 4264 4265
    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) {
4266 4267 4268
        return NULL;
    }

4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283
    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;
            }
4284
        } else {
4285 4286 4287 4288 4289 4290 4291 4292 4293 4294 4295 4296 4297 4298 4299 4300 4301 4302 4303 4304 4305 4306 4307
            /* 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;
            }
4308 4309 4310
        }
    }

4311 4312 4313 4314
    g_free(filename_full);
    g_free(backing_file_full);
    g_free(filename_tmp);
    return retval;
4315 4316
}

4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329
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);
}

4330 4331 4332 4333 4334 4335 4336 4337 4338 4339 4340 4341 4342 4343 4344 4345
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
4346
/**************************************************************/
B
bellard 已提交
4347
/* async I/Os */
B
bellard 已提交
4348

4349
BlockDriverAIOCB *bdrv_aio_readv(BlockDriverState *bs, int64_t sector_num,
4350
                                 QEMUIOVector *qiov, int nb_sectors,
4351
                                 BlockDriverCompletionFunc *cb, void *opaque)
B
bellard 已提交
4352
{
4353 4354
    trace_bdrv_aio_readv(bs, sector_num, nb_sectors, opaque);

P
Paolo Bonzini 已提交
4355
    return bdrv_co_aio_rw_vector(bs, sector_num, qiov, nb_sectors, 0,
4356
                                 cb, opaque, false);
B
bellard 已提交
4357 4358
}

4359 4360 4361
BlockDriverAIOCB *bdrv_aio_writev(BlockDriverState *bs, int64_t sector_num,
                                  QEMUIOVector *qiov, int nb_sectors,
                                  BlockDriverCompletionFunc *cb, void *opaque)
B
bellard 已提交
4362
{
4363 4364
    trace_bdrv_aio_writev(bs, sector_num, nb_sectors, opaque);

P
Paolo Bonzini 已提交
4365
    return bdrv_co_aio_rw_vector(bs, sector_num, qiov, nb_sectors, 0,
4366
                                 cb, opaque, true);
B
bellard 已提交
4367 4368
}

P
Paolo Bonzini 已提交
4369 4370 4371 4372 4373 4374 4375 4376 4377 4378 4379
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
4380 4381 4382 4383 4384 4385 4386 4387 4388 4389 4390 4391 4392 4393 4394 4395 4396 4397

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);
4398 4399 4400
        if (mcb->callbacks[i].free_qiov) {
            qemu_iovec_destroy(mcb->callbacks[i].free_qiov);
        }
4401
        g_free(mcb->callbacks[i].free_qiov);
K
Kevin Wolf 已提交
4402 4403 4404 4405 4406 4407 4408
    }
}

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

4409 4410
    trace_multiwrite_cb(mcb, ret);

4411
    if (ret < 0 && !mcb->error) {
K
Kevin Wolf 已提交
4412 4413 4414 4415 4416
        mcb->error = ret;
    }

    mcb->num_requests--;
    if (mcb->num_requests == 0) {
4417
        multiwrite_user_cb(mcb);
4418
        g_free(mcb);
K
Kevin Wolf 已提交
4419 4420 4421 4422 4423
    }
}

static int multiwrite_req_compare(const void *a, const void *b)
{
4424 4425 4426 4427 4428 4429 4430 4431 4432 4433 4434 4435 4436
    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
4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452 4453 4454 4455 4456 4457
}

/*
 * 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;

4458
        // Handle exactly sequential writes and overlapping writes.
K
Kevin Wolf 已提交
4459 4460 4461 4462
        if (reqs[i].sector <= oldreq_last) {
            merge = 1;
        }

4463 4464 4465 4466
        if (reqs[outidx].qiov->niov + reqs[i].qiov->niov + 1 > IOV_MAX) {
            merge = 0;
        }

K
Kevin Wolf 已提交
4467 4468
        if (merge) {
            size_t size;
4469
            QEMUIOVector *qiov = g_malloc0(sizeof(*qiov));
K
Kevin Wolf 已提交
4470 4471 4472 4473 4474 4475
            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;
4476
            qemu_iovec_concat(qiov, reqs[outidx].qiov, 0, size);
K
Kevin Wolf 已提交
4477

4478 4479
            // We should need to add any zeros between the two requests
            assert (reqs[i].sector <= oldreq_last);
K
Kevin Wolf 已提交
4480 4481

            // Add the second request
4482
            qemu_iovec_concat(qiov, reqs[i].qiov, 0, reqs[i].qiov->size);
K
Kevin Wolf 已提交
4483

4484
            reqs[outidx].nb_sectors = qiov->size >> 9;
K
Kevin Wolf 已提交
4485 4486 4487 4488 4489 4490 4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513 4514 4515 4516 4517
            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;

4518 4519 4520 4521 4522 4523 4524 4525
    /* 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
4526 4527 4528 4529 4530
    if (num_reqs == 0) {
        return 0;
    }

    // Create MultiwriteCB structure
4531
    mcb = g_malloc0(sizeof(*mcb) + num_reqs * sizeof(*mcb->callbacks));
K
Kevin Wolf 已提交
4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542
    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);

4543 4544
    trace_bdrv_aio_multiwrite(mcb, mcb->num_callbacks, num_reqs);

4545 4546
    /* Run the aio requests. */
    mcb->num_requests = num_reqs;
K
Kevin Wolf 已提交
4547
    for (i = 0; i < num_reqs; i++) {
P
Paolo Bonzini 已提交
4548 4549 4550 4551
        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
4552 4553 4554 4555 4556
    }

    return 0;
}

B
bellard 已提交
4557 4558
void bdrv_aio_cancel(BlockDriverAIOCB *acb)
{
S
Stefan Hajnoczi 已提交
4559
    acb->aiocb_info->cancel(acb);
B
bellard 已提交
4560 4561 4562 4563 4564
}

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

4565 4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576
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)
{
4577 4578
    BlockDriverAIOCBSync *acb =
        container_of(blockacb, BlockDriverAIOCBSync, common);
4579
    qemu_bh_delete(acb->bh);
A
Avi Kivity 已提交
4580
    acb->bh = NULL;
4581 4582 4583
    qemu_aio_release(acb);
}

S
Stefan Hajnoczi 已提交
4584
static const AIOCBInfo bdrv_em_aiocb_info = {
4585 4586 4587 4588
    .aiocb_size         = sizeof(BlockDriverAIOCBSync),
    .cancel             = bdrv_aio_cancel_em,
};

4589
static void bdrv_aio_bh_cb(void *opaque)
B
bellard 已提交
4590
{
4591
    BlockDriverAIOCBSync *acb = opaque;
4592 4593

    if (!acb->is_write)
4594
        qemu_iovec_from_buf(acb->qiov, 0, acb->bounce, acb->qiov->size);
4595
    qemu_vfree(acb->bounce);
4596
    acb->common.cb(acb->common.opaque, acb->ret);
4597
    qemu_bh_delete(acb->bh);
A
Avi Kivity 已提交
4598
    acb->bh = NULL;
4599
    qemu_aio_release(acb);
B
bellard 已提交
4600
}
4601

4602 4603 4604 4605 4606 4607 4608 4609
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
4610
{
4611 4612
    BlockDriverAIOCBSync *acb;

S
Stefan Hajnoczi 已提交
4613
    acb = qemu_aio_get(&bdrv_em_aiocb_info, bs, cb, opaque);
4614 4615
    acb->is_write = is_write;
    acb->qiov = qiov;
4616
    acb->bounce = qemu_blockalign(bs, qiov->size);
4617
    acb->bh = aio_bh_new(bdrv_get_aio_context(bs), bdrv_aio_bh_cb, acb);
4618 4619

    if (is_write) {
4620
        qemu_iovec_to_buf(acb->qiov, 0, acb->bounce, qiov->size);
4621
        acb->ret = bs->drv->bdrv_write(bs, sector_num, acb->bounce, nb_sectors);
4622
    } else {
4623
        acb->ret = bs->drv->bdrv_read(bs, sector_num, acb->bounce, nb_sectors);
4624 4625
    }

4626
    qemu_bh_schedule(acb->bh);
4627

4628
    return &acb->common;
4629 4630
}

4631 4632
static BlockDriverAIOCB *bdrv_aio_readv_em(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
4633
        BlockDriverCompletionFunc *cb, void *opaque)
4634
{
4635 4636
    return bdrv_aio_rw_vector(bs, sector_num, qiov, nb_sectors, cb, opaque, 0);
}
B
bellard 已提交
4637

4638 4639 4640 4641 4642
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);
4643 4644
}

4645 4646 4647 4648 4649

typedef struct BlockDriverAIOCBCoroutine {
    BlockDriverAIOCB common;
    BlockRequest req;
    bool is_write;
K
Kevin Wolf 已提交
4650
    bool *done;
4651 4652 4653 4654 4655
    QEMUBH* bh;
} BlockDriverAIOCBCoroutine;

static void bdrv_aio_co_cancel_em(BlockDriverAIOCB *blockacb)
{
4656
    AioContext *aio_context = bdrv_get_aio_context(blockacb->bs);
K
Kevin Wolf 已提交
4657 4658 4659 4660 4661 4662
    BlockDriverAIOCBCoroutine *acb =
        container_of(blockacb, BlockDriverAIOCBCoroutine, common);
    bool done = false;

    acb->done = &done;
    while (!done) {
4663
        aio_poll(aio_context, true);
K
Kevin Wolf 已提交
4664
    }
4665 4666
}

S
Stefan Hajnoczi 已提交
4667
static const AIOCBInfo bdrv_em_co_aiocb_info = {
4668 4669 4670 4671
    .aiocb_size         = sizeof(BlockDriverAIOCBCoroutine),
    .cancel             = bdrv_aio_co_cancel_em,
};

P
Paolo Bonzini 已提交
4672
static void bdrv_co_em_bh(void *opaque)
4673 4674 4675 4676
{
    BlockDriverAIOCBCoroutine *acb = opaque;

    acb->common.cb(acb->common.opaque, acb->req.error);
K
Kevin Wolf 已提交
4677 4678 4679 4680 4681

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

4682 4683 4684 4685
    qemu_bh_delete(acb->bh);
    qemu_aio_release(acb);
}

4686 4687 4688 4689 4690 4691 4692 4693
/* 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
4694
            acb->req.nb_sectors, acb->req.qiov, acb->req.flags);
4695 4696
    } else {
        acb->req.error = bdrv_co_do_writev(bs, acb->req.sector,
P
Paolo Bonzini 已提交
4697
            acb->req.nb_sectors, acb->req.qiov, acb->req.flags);
4698 4699
    }

4700
    acb->bh = aio_bh_new(bdrv_get_aio_context(bs), bdrv_co_em_bh, acb);
4701 4702 4703
    qemu_bh_schedule(acb->bh);
}

4704 4705 4706 4707
static BlockDriverAIOCB *bdrv_co_aio_rw_vector(BlockDriverState *bs,
                                               int64_t sector_num,
                                               QEMUIOVector *qiov,
                                               int nb_sectors,
P
Paolo Bonzini 已提交
4708
                                               BdrvRequestFlags flags,
4709 4710
                                               BlockDriverCompletionFunc *cb,
                                               void *opaque,
4711
                                               bool is_write)
4712 4713 4714 4715
{
    Coroutine *co;
    BlockDriverAIOCBCoroutine *acb;

S
Stefan Hajnoczi 已提交
4716
    acb = qemu_aio_get(&bdrv_em_co_aiocb_info, bs, cb, opaque);
4717 4718 4719
    acb->req.sector = sector_num;
    acb->req.nb_sectors = nb_sectors;
    acb->req.qiov = qiov;
P
Paolo Bonzini 已提交
4720
    acb->req.flags = flags;
4721
    acb->is_write = is_write;
K
Kevin Wolf 已提交
4722
    acb->done = NULL;
4723

4724
    co = qemu_coroutine_create(bdrv_co_do_rw);
4725 4726 4727 4728 4729
    qemu_coroutine_enter(co, acb);

    return &acb->common;
}

P
Paolo Bonzini 已提交
4730
static void coroutine_fn bdrv_aio_flush_co_entry(void *opaque)
4731
{
P
Paolo Bonzini 已提交
4732 4733
    BlockDriverAIOCBCoroutine *acb = opaque;
    BlockDriverState *bs = acb->common.bs;
4734

P
Paolo Bonzini 已提交
4735
    acb->req.error = bdrv_co_flush(bs);
4736
    acb->bh = aio_bh_new(bdrv_get_aio_context(bs), bdrv_co_em_bh, acb);
4737 4738 4739
    qemu_bh_schedule(acb->bh);
}

P
Paolo Bonzini 已提交
4740
BlockDriverAIOCB *bdrv_aio_flush(BlockDriverState *bs,
4741 4742
        BlockDriverCompletionFunc *cb, void *opaque)
{
P
Paolo Bonzini 已提交
4743
    trace_bdrv_aio_flush(bs, opaque);
4744

P
Paolo Bonzini 已提交
4745 4746
    Coroutine *co;
    BlockDriverAIOCBCoroutine *acb;
4747

S
Stefan Hajnoczi 已提交
4748
    acb = qemu_aio_get(&bdrv_em_co_aiocb_info, bs, cb, opaque);
K
Kevin Wolf 已提交
4749 4750
    acb->done = NULL;

P
Paolo Bonzini 已提交
4751 4752
    co = qemu_coroutine_create(bdrv_aio_flush_co_entry);
    qemu_coroutine_enter(co, acb);
4753 4754 4755 4756

    return &acb->common;
}

4757 4758 4759 4760 4761 4762
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);
4763
    acb->bh = aio_bh_new(bdrv_get_aio_context(bs), bdrv_co_em_bh, acb);
4764 4765 4766 4767 4768 4769 4770 4771 4772 4773 4774 4775
    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
4776
    acb = qemu_aio_get(&bdrv_em_co_aiocb_info, bs, cb, opaque);
4777 4778
    acb->req.sector = sector_num;
    acb->req.nb_sectors = nb_sectors;
K
Kevin Wolf 已提交
4779
    acb->done = NULL;
4780 4781 4782 4783 4784 4785
    co = qemu_coroutine_create(bdrv_aio_discard_co_entry);
    qemu_coroutine_enter(co, acb);

    return &acb->common;
}

B
bellard 已提交
4786 4787
void bdrv_init(void)
{
4788
    module_call_init(MODULE_INIT_BLOCK);
B
bellard 已提交
4789
}
4790

4791 4792 4793 4794 4795 4796
void bdrv_init_with_whitelist(void)
{
    use_bdrv_whitelist = 1;
    bdrv_init();
}

S
Stefan Hajnoczi 已提交
4797
void *qemu_aio_get(const AIOCBInfo *aiocb_info, BlockDriverState *bs,
4798
                   BlockDriverCompletionFunc *cb, void *opaque)
4799 4800 4801
{
    BlockDriverAIOCB *acb;

S
Stefan Hajnoczi 已提交
4802 4803
    acb = g_slice_alloc(aiocb_info->aiocb_size);
    acb->aiocb_info = aiocb_info;
4804 4805 4806 4807 4808 4809 4810 4811
    acb->bs = bs;
    acb->cb = cb;
    acb->opaque = opaque;
    return acb;
}

void qemu_aio_release(void *p)
{
4812
    BlockDriverAIOCB *acb = p;
S
Stefan Hajnoczi 已提交
4813
    g_slice_free1(acb->aiocb_info->aiocb_size, acb);
4814
}
B
bellard 已提交
4815

4816 4817 4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828 4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839 4840 4841
/**************************************************************/
/* 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) {
4842 4843
        acb = bs->drv->bdrv_aio_writev(bs, sector_num, iov, nb_sectors,
                                       bdrv_co_io_em_complete, &co);
4844
    } else {
4845 4846
        acb = bs->drv->bdrv_aio_readv(bs, sector_num, iov, nb_sectors,
                                      bdrv_co_io_em_complete, &co);
4847 4848
    }

4849
    trace_bdrv_co_io_em(bs, sector_num, nb_sectors, is_write, acb);
4850 4851 4852 4853 4854 4855 4856 4857 4858 4859 4860 4861 4862 4863 4864 4865 4866 4867 4868 4869 4870 4871
    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
4872
static void coroutine_fn bdrv_flush_co_entry(void *opaque)
4873
{
P
Paolo Bonzini 已提交
4874 4875 4876 4877 4878 4879 4880
    RwCo *rwco = opaque;

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

int coroutine_fn bdrv_co_flush(BlockDriverState *bs)
{
K
Kevin Wolf 已提交
4881 4882
    int ret;

4883
    if (!bs || !bdrv_is_inserted(bs) || bdrv_is_read_only(bs)) {
P
Paolo Bonzini 已提交
4884
        return 0;
K
Kevin Wolf 已提交
4885 4886
    }

4887
    /* Write back cached data to the OS even with cache=unsafe */
4888
    BLKDBG_EVENT(bs->file, BLKDBG_FLUSH_TO_OS);
K
Kevin Wolf 已提交
4889 4890 4891 4892 4893 4894 4895
    if (bs->drv->bdrv_co_flush_to_os) {
        ret = bs->drv->bdrv_co_flush_to_os(bs);
        if (ret < 0) {
            return ret;
        }
    }

4896 4897
    /* But don't actually force it to the disk with cache=unsafe */
    if (bs->open_flags & BDRV_O_NO_FLUSH) {
4898
        goto flush_parent;
4899 4900
    }

4901
    BLKDBG_EVENT(bs->file, BLKDBG_FLUSH_TO_DISK);
K
Kevin Wolf 已提交
4902
    if (bs->drv->bdrv_co_flush_to_disk) {
4903
        ret = bs->drv->bdrv_co_flush_to_disk(bs);
P
Paolo Bonzini 已提交
4904 4905 4906 4907 4908 4909 4910 4911
    } 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) {
4912
            ret = -EIO;
P
Paolo Bonzini 已提交
4913 4914
        } else {
            qemu_coroutine_yield();
4915
            ret = co.ret;
P
Paolo Bonzini 已提交
4916 4917 4918 4919 4920 4921 4922 4923 4924 4925 4926 4927 4928
        }
    } 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.
         */
4929
        ret = 0;
P
Paolo Bonzini 已提交
4930
    }
4931 4932 4933 4934 4935 4936 4937
    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.
     */
4938
flush_parent:
4939
    return bdrv_co_flush(bs->file);
P
Paolo Bonzini 已提交
4940 4941
}

4942
void bdrv_invalidate_cache(BlockDriverState *bs, Error **errp)
4943
{
4944 4945 4946
    Error *local_err = NULL;
    int ret;

4947 4948 4949 4950 4951
    if (!bs->drv)  {
        return;
    }

    if (bs->drv->bdrv_invalidate_cache) {
4952
        bs->drv->bdrv_invalidate_cache(bs, &local_err);
4953
    } else if (bs->file) {
4954 4955 4956 4957 4958
        bdrv_invalidate_cache(bs->file, &local_err);
    }
    if (local_err) {
        error_propagate(errp, local_err);
        return;
4959
    }
4960

4961 4962 4963 4964 4965
    ret = refresh_total_sectors(bs, bs->total_sectors);
    if (ret < 0) {
        error_setg_errno(errp, -ret, "Could not refresh total sector count");
        return;
    }
4966 4967
}

4968
void bdrv_invalidate_cache_all(Error **errp)
4969 4970
{
    BlockDriverState *bs;
4971
    Error *local_err = NULL;
4972

4973
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
4974 4975 4976
        AioContext *aio_context = bdrv_get_aio_context(bs);

        aio_context_acquire(aio_context);
4977
        bdrv_invalidate_cache(bs, &local_err);
4978
        aio_context_release(aio_context);
4979 4980 4981 4982
        if (local_err) {
            error_propagate(errp, local_err);
            return;
        }
4983 4984 4985
    }
}

4986 4987 4988 4989
void bdrv_clear_incoming_migration_all(void)
{
    BlockDriverState *bs;

4990
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
4991 4992 4993
        AioContext *aio_context = bdrv_get_aio_context(bs);

        aio_context_acquire(aio_context);
4994
        bs->open_flags = bs->open_flags & ~(BDRV_O_INCOMING);
4995
        aio_context_release(aio_context);
4996 4997 4998
    }
}

P
Paolo Bonzini 已提交
4999 5000 5001 5002 5003 5004
int bdrv_flush(BlockDriverState *bs)
{
    Coroutine *co;
    RwCo rwco = {
        .bs = bs,
        .ret = NOT_DONE,
5005 5006
    };

P
Paolo Bonzini 已提交
5007 5008 5009 5010
    if (qemu_in_coroutine()) {
        /* Fast-path if already in coroutine context */
        bdrv_flush_co_entry(&rwco);
    } else {
5011 5012
        AioContext *aio_context = bdrv_get_aio_context(bs);

P
Paolo Bonzini 已提交
5013 5014 5015
        co = qemu_coroutine_create(bdrv_flush_co_entry);
        qemu_coroutine_enter(co, &rwco);
        while (rwco.ret == NOT_DONE) {
5016
            aio_poll(aio_context, true);
P
Paolo Bonzini 已提交
5017
        }
5018
    }
P
Paolo Bonzini 已提交
5019 5020

    return rwco.ret;
5021 5022
}

5023 5024 5025 5026 5027 5028
typedef struct DiscardCo {
    BlockDriverState *bs;
    int64_t sector_num;
    int nb_sectors;
    int ret;
} DiscardCo;
5029 5030
static void coroutine_fn bdrv_discard_co_entry(void *opaque)
{
5031
    DiscardCo *rwco = opaque;
5032 5033 5034 5035

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

5036 5037 5038 5039 5040
/* 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

5041 5042 5043
int coroutine_fn bdrv_co_discard(BlockDriverState *bs, int64_t sector_num,
                                 int nb_sectors)
{
5044 5045
    int max_discard;

5046 5047 5048 5049 5050 5051
    if (!bs->drv) {
        return -ENOMEDIUM;
    } else if (bdrv_check_request(bs, sector_num, nb_sectors)) {
        return -EIO;
    } else if (bs->read_only) {
        return -EROFS;
5052 5053
    }

F
Fam Zheng 已提交
5054
    bdrv_reset_dirty(bs, sector_num, nb_sectors);
5055

P
Paolo Bonzini 已提交
5056 5057 5058 5059 5060
    /* Do nothing if disabled.  */
    if (!(bs->open_flags & BDRV_O_UNMAP)) {
        return 0;
    }

5061 5062 5063
    if (!bs->drv->bdrv_co_discard && !bs->drv->bdrv_aio_discard) {
        return 0;
    }
5064

5065 5066 5067 5068
    max_discard = bs->bl.max_discard ?  bs->bl.max_discard : MAX_DISCARD_DEFAULT;
    while (nb_sectors > 0) {
        int ret;
        int num = nb_sectors;
5069

5070 5071 5072 5073 5074 5075
        /* 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;
5076
            }
5077 5078
            num -= sector_num % bs->bl.discard_alignment;
        }
5079

5080 5081 5082 5083
        /* limit request size */
        if (num > max_discard) {
            num = max_discard;
        }
5084

5085
        if (bs->drv->bdrv_co_discard) {
5086
            ret = bs->drv->bdrv_co_discard(bs, sector_num, num);
5087 5088 5089 5090 5091 5092 5093 5094 5095 5096 5097 5098 5099
        } 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;
5100 5101
            }
        }
5102
        if (ret && ret != -ENOTSUP) {
5103
            return ret;
5104
        }
5105 5106 5107

        sector_num += num;
        nb_sectors -= num;
5108
    }
5109
    return 0;
5110 5111 5112 5113 5114
}

int bdrv_discard(BlockDriverState *bs, int64_t sector_num, int nb_sectors)
{
    Coroutine *co;
5115
    DiscardCo rwco = {
5116 5117 5118 5119 5120 5121 5122 5123 5124 5125
        .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 {
5126 5127
        AioContext *aio_context = bdrv_get_aio_context(bs);

5128 5129 5130
        co = qemu_coroutine_create(bdrv_discard_co_entry);
        qemu_coroutine_enter(co, &rwco);
        while (rwco.ret == NOT_DONE) {
5131
            aio_poll(aio_context, true);
5132 5133 5134 5135 5136 5137
        }
    }

    return rwco.ret;
}

B
bellard 已提交
5138 5139 5140 5141 5142 5143 5144 5145 5146
/**************************************************************/
/* removable device support */

/**
 * Return TRUE if the media is present
 */
int bdrv_is_inserted(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;
5147

B
bellard 已提交
5148 5149 5150
    if (!drv)
        return 0;
    if (!drv->bdrv_is_inserted)
5151 5152
        return 1;
    return drv->bdrv_is_inserted(bs);
B
bellard 已提交
5153 5154 5155
}

/**
5156 5157
 * 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
5158 5159 5160 5161 5162
 */
int bdrv_media_changed(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;

5163 5164 5165 5166
    if (drv && drv->bdrv_media_changed) {
        return drv->bdrv_media_changed(bs);
    }
    return -ENOTSUP;
B
bellard 已提交
5167 5168 5169 5170 5171
}

/**
 * If eject_flag is TRUE, eject the media. Otherwise, close the tray
 */
5172
void bdrv_eject(BlockDriverState *bs, bool eject_flag)
B
bellard 已提交
5173 5174 5175
{
    BlockDriver *drv = bs->drv;

5176 5177
    if (drv && drv->bdrv_eject) {
        drv->bdrv_eject(bs, eject_flag);
B
bellard 已提交
5178
    }
5179 5180

    if (bs->device_name[0] != '\0') {
5181 5182
        qapi_event_send_device_tray_moved(bdrv_get_device_name(bs),
                                          eject_flag, &error_abort);
5183
    }
B
bellard 已提交
5184 5185 5186 5187 5188 5189
}

/**
 * Lock or unlock the media (if it is locked, the user won't be able
 * to eject it manually).
 */
5190
void bdrv_lock_medium(BlockDriverState *bs, bool locked)
B
bellard 已提交
5191 5192 5193
{
    BlockDriver *drv = bs->drv;

5194
    trace_bdrv_lock_medium(bs, locked);
S
Stefan Hajnoczi 已提交
5195

5196 5197
    if (drv && drv->bdrv_lock_medium) {
        drv->bdrv_lock_medium(bs, locked);
B
bellard 已提交
5198 5199
    }
}
5200 5201 5202 5203 5204 5205 5206 5207 5208 5209 5210

/* 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;
}
5211

5212 5213 5214
BlockDriverAIOCB *bdrv_aio_ioctl(BlockDriverState *bs,
        unsigned long int req, void *buf,
        BlockDriverCompletionFunc *cb, void *opaque)
5215
{
5216
    BlockDriver *drv = bs->drv;
5217

5218 5219 5220
    if (drv && drv->bdrv_aio_ioctl)
        return drv->bdrv_aio_ioctl(bs, req, buf, cb, opaque);
    return NULL;
5221
}
5222

5223
void bdrv_set_guest_block_size(BlockDriverState *bs, int align)
5224
{
5225
    bs->guest_block_size = align;
5226
}
5227

5228 5229
void *qemu_blockalign(BlockDriverState *bs, size_t size)
{
5230
    return qemu_memalign(bdrv_opt_mem_align(bs), size);
5231
}
5232

5233 5234 5235 5236 5237 5238
/*
 * Check if all memory in this vector is sector aligned.
 */
bool bdrv_qiov_is_aligned(BlockDriverState *bs, QEMUIOVector *qiov)
{
    int i;
5239
    size_t alignment = bdrv_opt_mem_align(bs);
5240 5241

    for (i = 0; i < qiov->niov; i++) {
5242
        if ((uintptr_t) qiov->iov[i].iov_base % alignment) {
5243
            return false;
5244
        }
5245
        if (qiov->iov[i].iov_len % alignment) {
5246
            return false;
5247 5248 5249 5250 5251 5252
        }
    }

    return true;
}

5253 5254
BdrvDirtyBitmap *bdrv_create_dirty_bitmap(BlockDriverState *bs, int granularity,
                                          Error **errp)
5255 5256
{
    int64_t bitmap_size;
F
Fam Zheng 已提交
5257
    BdrvDirtyBitmap *bitmap;
5258

5259 5260
    assert((granularity & (granularity - 1)) == 0);

F
Fam Zheng 已提交
5261 5262
    granularity >>= BDRV_SECTOR_BITS;
    assert(granularity);
5263 5264 5265 5266 5267 5268 5269
    bitmap_size = bdrv_getlength(bs);
    if (bitmap_size < 0) {
        error_setg_errno(errp, -bitmap_size, "could not get length of device");
        errno = -bitmap_size;
        return NULL;
    }
    bitmap_size >>= BDRV_SECTOR_BITS;
F
Fam Zheng 已提交
5270 5271 5272 5273 5274 5275 5276 5277 5278 5279 5280 5281 5282 5283 5284
    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;
5285
        }
5286 5287 5288
    }
}

F
Fam Zheng 已提交
5289 5290 5291 5292 5293 5294 5295 5296 5297 5298 5299 5300 5301 5302 5303 5304 5305 5306 5307 5308
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
5309
int bdrv_get_dirty(BlockDriverState *bs, BdrvDirtyBitmap *bitmap, int64_t sector)
5310
{
F
Fam Zheng 已提交
5311 5312
    if (bitmap) {
        return hbitmap_get(bitmap->bitmap, sector);
5313 5314 5315 5316 5317
    } else {
        return 0;
    }
}

F
Fam Zheng 已提交
5318 5319
void bdrv_dirty_iter_init(BlockDriverState *bs,
                          BdrvDirtyBitmap *bitmap, HBitmapIter *hbi)
5320
{
F
Fam Zheng 已提交
5321
    hbitmap_iter_init(hbi, bitmap->bitmap, 0);
5322 5323 5324 5325 5326
}

void bdrv_set_dirty(BlockDriverState *bs, int64_t cur_sector,
                    int nr_sectors)
{
F
Fam Zheng 已提交
5327 5328 5329 5330
    BdrvDirtyBitmap *bitmap;
    QLIST_FOREACH(bitmap, &bs->dirty_bitmaps, list) {
        hbitmap_set(bitmap->bitmap, cur_sector, nr_sectors);
    }
5331 5332
}

F
Fam Zheng 已提交
5333
void bdrv_reset_dirty(BlockDriverState *bs, int64_t cur_sector, int nr_sectors)
5334
{
F
Fam Zheng 已提交
5335 5336 5337 5338
    BdrvDirtyBitmap *bitmap;
    QLIST_FOREACH(bitmap, &bs->dirty_bitmaps, list) {
        hbitmap_reset(bitmap->bitmap, cur_sector, nr_sectors);
    }
5339
}
5340

F
Fam Zheng 已提交
5341
int64_t bdrv_get_dirty_count(BlockDriverState *bs, BdrvDirtyBitmap *bitmap)
5342
{
F
Fam Zheng 已提交
5343
    return hbitmap_count(bitmap->bitmap);
5344
}
J
Jes Sorensen 已提交
5345

5346 5347 5348 5349 5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360 5361 5362
/* 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);
    }
}

5363 5364 5365 5366 5367 5368 5369 5370 5371 5372 5373 5374 5375 5376 5377 5378 5379 5380 5381 5382 5383 5384 5385 5386 5387 5388 5389 5390 5391 5392 5393 5394 5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405 5406 5407 5408 5409 5410 5411 5412 5413 5414 5415 5416 5417 5418 5419 5420 5421 5422 5423 5424 5425 5426 5427 5428 5429 5430 5431 5432
struct BdrvOpBlocker {
    Error *reason;
    QLIST_ENTRY(BdrvOpBlocker) list;
};

bool bdrv_op_is_blocked(BlockDriverState *bs, BlockOpType op, Error **errp)
{
    BdrvOpBlocker *blocker;
    assert((int) op >= 0 && op < BLOCK_OP_TYPE_MAX);
    if (!QLIST_EMPTY(&bs->op_blockers[op])) {
        blocker = QLIST_FIRST(&bs->op_blockers[op]);
        if (errp) {
            error_setg(errp, "Device '%s' is busy: %s",
                       bs->device_name, error_get_pretty(blocker->reason));
        }
        return true;
    }
    return false;
}

void bdrv_op_block(BlockDriverState *bs, BlockOpType op, Error *reason)
{
    BdrvOpBlocker *blocker;
    assert((int) op >= 0 && op < BLOCK_OP_TYPE_MAX);

    blocker = g_malloc0(sizeof(BdrvOpBlocker));
    blocker->reason = reason;
    QLIST_INSERT_HEAD(&bs->op_blockers[op], blocker, list);
}

void bdrv_op_unblock(BlockDriverState *bs, BlockOpType op, Error *reason)
{
    BdrvOpBlocker *blocker, *next;
    assert((int) op >= 0 && op < BLOCK_OP_TYPE_MAX);
    QLIST_FOREACH_SAFE(blocker, &bs->op_blockers[op], list, next) {
        if (blocker->reason == reason) {
            QLIST_REMOVE(blocker, list);
            g_free(blocker);
        }
    }
}

void bdrv_op_block_all(BlockDriverState *bs, Error *reason)
{
    int i;
    for (i = 0; i < BLOCK_OP_TYPE_MAX; i++) {
        bdrv_op_block(bs, i, reason);
    }
}

void bdrv_op_unblock_all(BlockDriverState *bs, Error *reason)
{
    int i;
    for (i = 0; i < BLOCK_OP_TYPE_MAX; i++) {
        bdrv_op_unblock(bs, i, reason);
    }
}

bool bdrv_op_blocker_is_empty(BlockDriverState *bs)
{
    int i;

    for (i = 0; i < BLOCK_OP_TYPE_MAX; i++) {
        if (!QLIST_EMPTY(&bs->op_blockers[i])) {
            return false;
        }
    }
    return true;
}

5433 5434
void bdrv_iostatus_enable(BlockDriverState *bs)
{
5435
    bs->iostatus_enabled = true;
5436
    bs->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
5437 5438 5439 5440 5441 5442
}

/* 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)
{
5443
    return (bs->iostatus_enabled &&
5444 5445 5446
           (bs->on_write_error == BLOCKDEV_ON_ERROR_ENOSPC ||
            bs->on_write_error == BLOCKDEV_ON_ERROR_STOP   ||
            bs->on_read_error == BLOCKDEV_ON_ERROR_STOP));
5447 5448 5449 5450
}

void bdrv_iostatus_disable(BlockDriverState *bs)
{
5451
    bs->iostatus_enabled = false;
5452 5453 5454 5455 5456
}

void bdrv_iostatus_reset(BlockDriverState *bs)
{
    if (bdrv_iostatus_is_enabled(bs)) {
5457
        bs->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
5458 5459 5460
        if (bs->job) {
            block_job_iostatus_reset(bs->job);
        }
5461 5462 5463 5464 5465
    }
}

void bdrv_iostatus_set_err(BlockDriverState *bs, int error)
{
5466 5467
    assert(bdrv_iostatus_is_enabled(bs));
    if (bs->iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
5468 5469
        bs->iostatus = error == ENOSPC ? BLOCK_DEVICE_IO_STATUS_NOSPACE :
                                         BLOCK_DEVICE_IO_STATUS_FAILED;
5470 5471 5472
    }
}

5473 5474 5475 5476 5477 5478 5479
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
5480
    cookie->start_time_ns = get_clock();
5481 5482 5483 5484 5485 5486 5487 5488 5489 5490
    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
5491
    bs->total_time_ns[cookie->type] += get_clock() - cookie->start_time_ns;
5492 5493
}

5494 5495
void bdrv_img_create(const char *filename, const char *fmt,
                     const char *base_filename, const char *base_fmt,
5496 5497
                     char *options, uint64_t img_size, int flags,
                     Error **errp, bool quiet)
J
Jes Sorensen 已提交
5498
{
5499 5500 5501 5502
    QemuOptsList *create_opts = NULL;
    QemuOpts *opts = NULL;
    const char *backing_fmt, *backing_file;
    int64_t size;
J
Jes Sorensen 已提交
5503
    BlockDriver *drv, *proto_drv;
5504
    BlockDriver *backing_drv = NULL;
5505
    Error *local_err = NULL;
J
Jes Sorensen 已提交
5506 5507 5508 5509 5510
    int ret = 0;

    /* Find driver and parse its options */
    drv = bdrv_find_format(fmt);
    if (!drv) {
5511
        error_setg(errp, "Unknown file format '%s'", fmt);
5512
        return;
J
Jes Sorensen 已提交
5513 5514
    }

5515
    proto_drv = bdrv_find_protocol(filename, true);
J
Jes Sorensen 已提交
5516
    if (!proto_drv) {
5517
        error_setg(errp, "Unknown protocol '%s'", filename);
5518
        return;
J
Jes Sorensen 已提交
5519 5520
    }

C
Chunyan Liu 已提交
5521 5522
    create_opts = qemu_opts_append(create_opts, drv->create_opts);
    create_opts = qemu_opts_append(create_opts, proto_drv->create_opts);
J
Jes Sorensen 已提交
5523 5524

    /* Create parameter list with default values */
5525 5526
    opts = qemu_opts_create(create_opts, NULL, 0, &error_abort);
    qemu_opt_set_number(opts, BLOCK_OPT_SIZE, img_size);
J
Jes Sorensen 已提交
5527 5528 5529

    /* Parse -o options */
    if (options) {
5530 5531
        if (qemu_opts_do_parse(opts, options, NULL) != 0) {
            error_setg(errp, "Invalid options for file format '%s'", fmt);
J
Jes Sorensen 已提交
5532 5533 5534 5535 5536
            goto out;
        }
    }

    if (base_filename) {
5537
        if (qemu_opt_set(opts, BLOCK_OPT_BACKING_FILE, base_filename)) {
5538 5539
            error_setg(errp, "Backing file not supported for file format '%s'",
                       fmt);
J
Jes Sorensen 已提交
5540 5541 5542 5543 5544
            goto out;
        }
    }

    if (base_fmt) {
5545
        if (qemu_opt_set(opts, BLOCK_OPT_BACKING_FMT, base_fmt)) {
5546 5547
            error_setg(errp, "Backing file format not supported for file "
                             "format '%s'", fmt);
J
Jes Sorensen 已提交
5548 5549 5550 5551
            goto out;
        }
    }

5552 5553 5554
    backing_file = qemu_opt_get(opts, BLOCK_OPT_BACKING_FILE);
    if (backing_file) {
        if (!strcmp(filename, backing_file)) {
5555 5556
            error_setg(errp, "Error: Trying to create an image with the "
                             "same filename as the backing file");
5557 5558 5559 5560
            goto out;
        }
    }

5561 5562 5563
    backing_fmt = qemu_opt_get(opts, BLOCK_OPT_BACKING_FMT);
    if (backing_fmt) {
        backing_drv = bdrv_find_format(backing_fmt);
5564
        if (!backing_drv) {
5565
            error_setg(errp, "Unknown backing file format '%s'",
5566
                       backing_fmt);
J
Jes Sorensen 已提交
5567 5568 5569 5570 5571 5572
            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
5573 5574 5575
    size = qemu_opt_get_size(opts, BLOCK_OPT_SIZE, 0);
    if (size == -1) {
        if (backing_file) {
5576
            BlockDriverState *bs;
J
Jes Sorensen 已提交
5577
            uint64_t size;
5578 5579 5580 5581 5582
            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
5583

5584
            bs = NULL;
5585
            ret = bdrv_open(&bs, backing_file, NULL, NULL, back_flags,
5586
                            backing_drv, &local_err);
J
Jes Sorensen 已提交
5587
            if (ret < 0) {
5588
                error_setg_errno(errp, -ret, "Could not open '%s': %s",
5589
                                 backing_file,
5590 5591 5592
                                 error_get_pretty(local_err));
                error_free(local_err);
                local_err = NULL;
J
Jes Sorensen 已提交
5593 5594 5595 5596 5597
                goto out;
            }
            bdrv_get_geometry(bs, &size);
            size *= 512;

5598
            qemu_opt_set_number(opts, BLOCK_OPT_SIZE, size);
5599 5600

            bdrv_unref(bs);
J
Jes Sorensen 已提交
5601
        } else {
5602
            error_setg(errp, "Image creation needs a size parameter");
J
Jes Sorensen 已提交
5603 5604 5605 5606
            goto out;
        }
    }

5607 5608
    if (!quiet) {
        printf("Formatting '%s', fmt=%s ", filename, fmt);
5609
        qemu_opts_print(opts);
5610 5611
        puts("");
    }
5612

C
Chunyan Liu 已提交
5613
    ret = bdrv_create(drv, filename, opts, &local_err);
5614

5615 5616 5617 5618 5619
    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 = "";
5620
        if (qemu_opt_get_size(opts, BLOCK_OPT_CLUSTER_SIZE, 0)) {
5621
            cluster_size_hint = " (try using a larger cluster size)";
J
Jes Sorensen 已提交
5622
        }
5623 5624 5625 5626
        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
5627 5628 5629
    }

out:
5630 5631
    qemu_opts_del(opts);
    qemu_opts_free(create_opts);
5632
    if (local_err) {
5633 5634
        error_propagate(errp, local_err);
    }
J
Jes Sorensen 已提交
5635
}
5636 5637 5638

AioContext *bdrv_get_aio_context(BlockDriverState *bs)
{
5639 5640 5641 5642 5643 5644 5645 5646 5647
    return bs->aio_context;
}

void bdrv_detach_aio_context(BlockDriverState *bs)
{
    if (!bs->drv) {
        return;
    }

5648 5649 5650
    if (bs->io_limits_enabled) {
        throttle_detach_aio_context(&bs->throttle_state);
    }
5651 5652 5653 5654 5655 5656 5657 5658 5659 5660 5661 5662 5663 5664 5665 5666 5667 5668 5669 5670 5671 5672 5673 5674 5675 5676 5677 5678 5679 5680 5681
    if (bs->drv->bdrv_detach_aio_context) {
        bs->drv->bdrv_detach_aio_context(bs);
    }
    if (bs->file) {
        bdrv_detach_aio_context(bs->file);
    }
    if (bs->backing_hd) {
        bdrv_detach_aio_context(bs->backing_hd);
    }

    bs->aio_context = NULL;
}

void bdrv_attach_aio_context(BlockDriverState *bs,
                             AioContext *new_context)
{
    if (!bs->drv) {
        return;
    }

    bs->aio_context = new_context;

    if (bs->backing_hd) {
        bdrv_attach_aio_context(bs->backing_hd, new_context);
    }
    if (bs->file) {
        bdrv_attach_aio_context(bs->file, new_context);
    }
    if (bs->drv->bdrv_attach_aio_context) {
        bs->drv->bdrv_attach_aio_context(bs, new_context);
    }
5682 5683 5684
    if (bs->io_limits_enabled) {
        throttle_attach_aio_context(&bs->throttle_state, new_context);
    }
5685 5686 5687 5688 5689 5690 5691 5692 5693 5694 5695 5696 5697 5698
}

void bdrv_set_aio_context(BlockDriverState *bs, AioContext *new_context)
{
    bdrv_drain_all(); /* ensure there are no in-flight requests */

    bdrv_detach_aio_context(bs);

    /* This function executes in the old AioContext so acquire the new one in
     * case it runs in a different thread.
     */
    aio_context_acquire(new_context);
    bdrv_attach_aio_context(bs, new_context);
    aio_context_release(new_context);
5699
}
5700 5701 5702 5703 5704 5705

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
5706

C
Chunyan Liu 已提交
5707
int bdrv_amend_options(BlockDriverState *bs, QemuOpts *opts)
M
Max Reitz 已提交
5708
{
C
Chunyan Liu 已提交
5709
    if (!bs->drv->bdrv_amend_options) {
M
Max Reitz 已提交
5710 5711
        return -ENOTSUP;
    }
C
Chunyan Liu 已提交
5712
    return bs->drv->bdrv_amend_options(bs, opts);
M
Max Reitz 已提交
5713
}
5714

5715 5716 5717 5718
/* This function will be called by the bdrv_recurse_is_first_non_filter method
 * of block filter and by bdrv_is_first_non_filter.
 * It is used to test if the given bs is the candidate or recurse more in the
 * node graph.
5719
 */
5720
bool bdrv_recurse_is_first_non_filter(BlockDriverState *bs,
5721
                                      BlockDriverState *candidate)
5722
{
5723 5724
    /* return false if basic checks fails */
    if (!bs || !bs->drv) {
5725
        return false;
5726 5727
    }

5728 5729 5730 5731 5732
    /* the code reached a non block filter driver -> check if the bs is
     * the same as the candidate. It's the recursion termination condition.
     */
    if (!bs->drv->is_filter) {
        return bs == candidate;
5733
    }
5734
    /* Down this path the driver is a block filter driver */
5735

5736 5737 5738 5739
    /* If the block filter recursion method is defined use it to recurse down
     * the node graph.
     */
    if (bs->drv->bdrv_recurse_is_first_non_filter) {
5740
        return bs->drv->bdrv_recurse_is_first_non_filter(bs, candidate);
5741 5742
    }

5743 5744 5745
    /* the driver is a block filter but don't allow to recurse -> return false
     */
    return false;
5746 5747
}

5748 5749 5750 5751 5752
/* 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)
5753
{
5754 5755 5756 5757 5758 5759
    BlockDriverState *bs;

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

5760
        /* try to recurse in this top level bs */
5761
        perm = bdrv_recurse_is_first_non_filter(bs, candidate);
5762 5763 5764 5765 5766 5767 5768 5769

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

    return false;
5770
}