block.c 176.3 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/block-backend.h"
32
#include "sysemu/sysemu.h"
33
#include "qemu/notify.h"
34
#include "block/coroutine.h"
35
#include "block/qapi.h"
L
Luiz Capitulino 已提交
36
#include "qmp-commands.h"
37
#include "qemu/timer.h"
38
#include "qapi-event.h"
B
bellard 已提交
39

J
Juan Quintela 已提交
40
#ifdef CONFIG_BSD
B
bellard 已提交
41 42 43
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/ioctl.h>
B
Blue Swirl 已提交
44
#include <sys/queue.h>
45
#ifndef __DragonFly__
B
bellard 已提交
46 47
#include <sys/disk.h>
#endif
48
#endif
B
bellard 已提交
49

50 51 52 53
#ifdef _WIN32
#include <windows.h>
#endif

F
Fam Zheng 已提交
54 55 56 57 58
struct BdrvDirtyBitmap {
    HBitmap *bitmap;
    QLIST_ENTRY(BdrvDirtyBitmap) list;
};

59 60
#define NOT_DONE 0x7fffffff /* used while emulated sync operation in progress */

61
static BlockAIOCB *bdrv_aio_readv_em(BlockDriverState *bs,
62
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
63
        BlockCompletionFunc *cb, void *opaque);
64
static BlockAIOCB *bdrv_aio_writev_em(BlockDriverState *bs,
65
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
66
        BlockCompletionFunc *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 83
static BlockAIOCB *bdrv_co_aio_rw_vector(BlockDriverState *bs,
                                         int64_t sector_num,
                                         QEMUIOVector *qiov,
                                         int nb_sectors,
                                         BdrvRequestFlags flags,
84
                                         BlockCompletionFunc *cb,
85 86
                                         void *opaque,
                                         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

100 101 102 103
static void bdrv_set_dirty(BlockDriverState *bs, int64_t cur_sector,
                           int nr_sectors);
static void bdrv_reset_dirty(BlockDriverState *bs, int64_t cur_sector,
                             int nr_sectors);
104 105 106
/* If non-zero, use only whitelisted block drivers */
static int use_bdrv_whitelist;

107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126
#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

127
/* throttling disk I/O limits */
128 129
void bdrv_set_io_limits(BlockDriverState *bs,
                        ThrottleConfig *cfg)
Z
Zhi Yong Wu 已提交
130
{
131
    int i;
Z
Zhi Yong Wu 已提交
132

133
    throttle_config(&bs->throttle_state, cfg);
Z
Zhi Yong Wu 已提交
134

135 136
    for (i = 0; i < 2; i++) {
        qemu_co_enter_next(&bs->throttled_reqs[i]);
Z
Zhi Yong Wu 已提交
137
    }
138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155
}

/* 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
156

157
    return drained;
Z
Zhi Yong Wu 已提交
158 159
}

160
void bdrv_io_limits_disable(BlockDriverState *bs)
161
{
162
    bs->io_limits_enabled = false;
163

164 165 166
    bdrv_start_throttled_reqs(bs);

    throttle_destroy(&bs->throttle_state);
167 168
}

169
static void bdrv_throttle_read_timer_cb(void *opaque)
170
{
171 172
    BlockDriverState *bs = opaque;
    qemu_co_enter_next(&bs->throttled_reqs[0]);
173 174
}

175
static void bdrv_throttle_write_timer_cb(void *opaque)
176
{
177 178
    BlockDriverState *bs = opaque;
    qemu_co_enter_next(&bs->throttled_reqs[1]);
179 180
}

181 182 183 184 185
/* 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,
186
                  bdrv_get_aio_context(bs),
187 188 189 190 191 192 193 194 195 196 197 198
                  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
199
static void bdrv_io_limits_intercept(BlockDriverState *bs,
200
                                     unsigned int bytes,
201
                                     bool is_write)
Z
Zhi Yong Wu 已提交
202
{
203 204
    /* does this io must wait */
    bool must_wait = throttle_schedule_timer(&bs->throttle_state, is_write);
Z
Zhi Yong Wu 已提交
205

206 207 208 209
    /* 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
210 211
    }

212
    /* the IO will be executed, do the accounting */
213 214
    throttle_account(&bs->throttle_state, is_write, bytes);

Z
Zhi Yong Wu 已提交
215

216 217 218
    /* 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
219 220
    }

221 222
    /* else queue next request for execution */
    qemu_co_queue_next(&bs->throttled_reqs[is_write]);
Z
Zhi Yong Wu 已提交
223 224
}

225 226 227 228 229 230 231 232 233 234
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;
}

235
/* check if the path starts with "<protocol>:" */
236
int path_has_protocol(const char *path)
237
{
238 239
    const char *p;

240 241 242 243 244
#ifdef _WIN32
    if (is_windows_drive(path) ||
        is_windows_drive_prefix(path)) {
        return 0;
    }
245 246 247
    p = path + strcspn(path, ":/\\");
#else
    p = path + strcspn(path, ":/");
248 249
#endif

250
    return *p == ':';
251 252
}

B
bellard 已提交
253
int path_is_absolute(const char *path)
254
{
B
bellard 已提交
255 256
#ifdef _WIN32
    /* specific case for names like: "\\.\d:" */
P
Paolo Bonzini 已提交
257
    if (is_windows_drive(path) || is_windows_drive_prefix(path)) {
B
bellard 已提交
258
        return 1;
P
Paolo Bonzini 已提交
259 260
    }
    return (*path == '/' || *path == '\\');
261
#else
P
Paolo Bonzini 已提交
262
    return (*path == '/');
263
#endif
264 265
}

B
bellard 已提交
266 267 268 269 270 271
/* 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)
272
{
B
bellard 已提交
273 274 275 276 277 278 279 280 281 282 283 284 285
    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;
286 287 288 289 290 291 292 293 294
        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
295 296 297 298 299 300 301 302 303 304 305 306
        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);
307 308 309
    }
}

310 311
void bdrv_get_full_backing_filename_from_filename(const char *backed,
                                                  const char *backing,
312 313
                                                  char *dest, size_t sz,
                                                  Error **errp)
314
{
315 316 317
    if (backing[0] == '\0' || path_has_protocol(backing) ||
        path_is_absolute(backing))
    {
318
        pstrcpy(dest, sz, backing);
319 320 321
    } else if (backed[0] == '\0' || strstart(backed, "json:", NULL)) {
        error_setg(errp, "Cannot use relative backing file names for '%s'",
                   backed);
322
    } else {
323
        path_combine(dest, sz, backed, backing);
324 325 326
    }
}

327 328
void bdrv_get_full_backing_filename(BlockDriverState *bs, char *dest, size_t sz,
                                    Error **errp)
329
{
330 331 332 333
    char *backed = bs->exact_filename[0] ? bs->exact_filename : bs->filename;

    bdrv_get_full_backing_filename_from_filename(backed, bs->backing_file,
                                                 dest, sz, errp);
334 335
}

336
void bdrv_register(BlockDriver *bdrv)
B
bellard 已提交
337
{
338 339
    /* Block drivers without coroutine functions need emulation */
    if (!bdrv->bdrv_co_readv) {
340 341 342
        bdrv->bdrv_co_readv = bdrv_co_readv_em;
        bdrv->bdrv_co_writev = bdrv_co_writev_em;

343 344 345
        /* 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.
         */
346 347 348 349 350
        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
351
    }
352

353
    QLIST_INSERT_HEAD(&bdrv_drivers, bdrv, list);
B
bellard 已提交
354
}
B
bellard 已提交
355

356
BlockDriverState *bdrv_new_root(void)
B
bellard 已提交
357
{
358
    BlockDriverState *bs = bdrv_new();
359 360 361 362 363 364 365 366 367 368

    QTAILQ_INSERT_TAIL(&bdrv_states, bs, device_list);
    return bs;
}

BlockDriverState *bdrv_new(void)
{
    BlockDriverState *bs;
    int i;

369
    bs = g_new0(BlockDriverState, 1);
F
Fam Zheng 已提交
370
    QLIST_INIT(&bs->dirty_bitmaps);
371 372 373
    for (i = 0; i < BLOCK_OP_TYPE_MAX; i++) {
        QLIST_INIT(&bs->op_blockers[i]);
    }
374
    bdrv_iostatus_disable(bs);
P
Paolo Bonzini 已提交
375
    notifier_list_init(&bs->close_notifiers);
376
    notifier_with_return_list_init(&bs->before_write_notifiers);
377 378
    qemu_co_queue_init(&bs->throttled_reqs[0]);
    qemu_co_queue_init(&bs->throttled_reqs[1]);
379
    bs->refcnt = 1;
380
    bs->aio_context = qemu_get_aio_context();
P
Paolo Bonzini 已提交
381

B
bellard 已提交
382 383 384
    return bs;
}

P
Paolo Bonzini 已提交
385 386 387 388 389
void bdrv_add_close_notifier(BlockDriverState *bs, Notifier *notify)
{
    notifier_list_add(&bs->close_notifiers, notify);
}

B
bellard 已提交
390 391 392
BlockDriver *bdrv_find_format(const char *format_name)
{
    BlockDriver *drv1;
393 394
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
        if (!strcmp(drv1->format_name, format_name)) {
B
bellard 已提交
395
            return drv1;
396
        }
B
bellard 已提交
397 398 399 400
    }
    return NULL;
}

401
static int bdrv_is_whitelisted(BlockDriver *drv, bool read_only)
402
{
403 404 405 406 407
    static const char *whitelist_rw[] = {
        CONFIG_BDRV_RW_WHITELIST
    };
    static const char *whitelist_ro[] = {
        CONFIG_BDRV_RO_WHITELIST
408 409 410
    };
    const char **p;

411
    if (!whitelist_rw[0] && !whitelist_ro[0]) {
412
        return 1;               /* no whitelist, anything goes */
413
    }
414

415
    for (p = whitelist_rw; *p; p++) {
416 417 418 419
        if (!strcmp(drv->format_name, *p)) {
            return 1;
        }
    }
420 421 422 423 424 425 426
    if (read_only) {
        for (p = whitelist_ro; *p; p++) {
            if (!strcmp(drv->format_name, *p)) {
                return 1;
            }
        }
    }
427 428 429
    return 0;
}

430 431
BlockDriver *bdrv_find_whitelisted_format(const char *format_name,
                                          bool read_only)
432 433
{
    BlockDriver *drv = bdrv_find_format(format_name);
434
    return drv && bdrv_is_whitelisted(drv, read_only) ? drv : NULL;
435 436
}

437 438 439
typedef struct CreateCo {
    BlockDriver *drv;
    char *filename;
440
    QemuOpts *opts;
441
    int ret;
442
    Error *err;
443 444 445 446
} CreateCo;

static void coroutine_fn bdrv_create_co_entry(void *opaque)
{
447 448 449
    Error *local_err = NULL;
    int ret;

450 451 452
    CreateCo *cco = opaque;
    assert(cco->drv);

C
Chunyan Liu 已提交
453
    ret = cco->drv->bdrv_create(cco->filename, cco->opts, &local_err);
454
    if (local_err) {
455 456 457
        error_propagate(&cco->err, local_err);
    }
    cco->ret = ret;
458 459
}

460
int bdrv_create(BlockDriver *drv, const char* filename,
461
                QemuOpts *opts, Error **errp)
B
bellard 已提交
462
{
463 464 465 466 467 468
    int ret;

    Coroutine *co;
    CreateCo cco = {
        .drv = drv,
        .filename = g_strdup(filename),
469
        .opts = opts,
470
        .ret = NOT_DONE,
471
        .err = NULL,
472 473
    };

C
Chunyan Liu 已提交
474
    if (!drv->bdrv_create) {
475
        error_setg(errp, "Driver '%s' does not support image creation", drv->format_name);
476 477
        ret = -ENOTSUP;
        goto out;
478 479 480 481 482 483 484 485 486
    }

    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) {
487
            aio_poll(qemu_get_aio_context(), true);
488 489 490 491
        }
    }

    ret = cco.ret;
492
    if (ret < 0) {
493
        if (cco.err) {
494 495 496 497 498
            error_propagate(errp, cco.err);
        } else {
            error_setg_errno(errp, -ret, "Could not create image");
        }
    }
499

500 501
out:
    g_free(cco.filename);
502
    return ret;
B
bellard 已提交
503 504
}

C
Chunyan Liu 已提交
505
int bdrv_create_file(const char *filename, QemuOpts *opts, Error **errp)
506 507
{
    BlockDriver *drv;
508 509
    Error *local_err = NULL;
    int ret;
510

511
    drv = bdrv_find_protocol(filename, true, errp);
512
    if (drv == NULL) {
513
        return -ENOENT;
514 515
    }

C
Chunyan Liu 已提交
516
    ret = bdrv_create(drv, filename, opts, &local_err);
517
    if (local_err) {
518 519 520
        error_propagate(errp, local_err);
    }
    return ret;
521 522
}

523
void bdrv_refresh_limits(BlockDriverState *bs, Error **errp)
524 525
{
    BlockDriver *drv = bs->drv;
526
    Error *local_err = NULL;
527 528 529

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

K
Kevin Wolf 已提交
530
    if (!drv) {
531
        return;
K
Kevin Wolf 已提交
532 533 534 535
    }

    /* Take some limits from the children as a default */
    if (bs->file) {
536 537 538 539 540
        bdrv_refresh_limits(bs->file, &local_err);
        if (local_err) {
            error_propagate(errp, local_err);
            return;
        }
K
Kevin Wolf 已提交
541
        bs->bl.opt_transfer_length = bs->file->bl.opt_transfer_length;
542
        bs->bl.max_transfer_length = bs->file->bl.max_transfer_length;
543 544 545
        bs->bl.opt_mem_alignment = bs->file->bl.opt_mem_alignment;
    } else {
        bs->bl.opt_mem_alignment = 512;
K
Kevin Wolf 已提交
546 547 548
    }

    if (bs->backing_hd) {
549 550 551 552 553
        bdrv_refresh_limits(bs->backing_hd, &local_err);
        if (local_err) {
            error_propagate(errp, local_err);
            return;
        }
K
Kevin Wolf 已提交
554 555 556
        bs->bl.opt_transfer_length =
            MAX(bs->bl.opt_transfer_length,
                bs->backing_hd->bl.opt_transfer_length);
557 558 559
        bs->bl.max_transfer_length =
            MIN_NON_ZERO(bs->bl.max_transfer_length,
                         bs->backing_hd->bl.max_transfer_length);
560 561 562
        bs->bl.opt_mem_alignment =
            MAX(bs->bl.opt_mem_alignment,
                bs->backing_hd->bl.opt_mem_alignment);
K
Kevin Wolf 已提交
563 564 565 566
    }

    /* Then let the driver override it */
    if (drv->bdrv_refresh_limits) {
567
        drv->bdrv_refresh_limits(bs, errp);
568 569 570
    }
}

571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604
/**
 * Try to get @bs's logical and physical block size.
 * On success, store them in @bsz struct and return 0.
 * On failure return -errno.
 * @bs must not be empty.
 */
int bdrv_probe_blocksizes(BlockDriverState *bs, BlockSizes *bsz)
{
    BlockDriver *drv = bs->drv;

    if (drv && drv->bdrv_probe_blocksizes) {
        return drv->bdrv_probe_blocksizes(bs, bsz);
    }

    return -ENOTSUP;
}

/**
 * Try to get @bs's geometry (cyls, heads, sectors).
 * On success, store them in @geo struct and return 0.
 * On failure return -errno.
 * @bs must not be empty.
 */
int bdrv_probe_geometry(BlockDriverState *bs, HDGeometry *geo)
{
    BlockDriver *drv = bs->drv;

    if (drv && drv->bdrv_probe_geometry) {
        return drv->bdrv_probe_geometry(bs, geo);
    }

    return -ENOTSUP;
}

605 606 607 608 609
/*
 * 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
610
{
611
#ifdef _WIN32
612
    char temp_dir[MAX_PATH];
613 614 615 616 617 618
    /* 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
619
#else
B
bellard 已提交
620
    int fd;
621
    const char *tmpdir;
A
aurel32 已提交
622
    tmpdir = getenv("TMPDIR");
623 624 625
    if (!tmpdir) {
        tmpdir = "/var/tmp";
    }
626 627 628
    if (snprintf(filename, size, "%s/vl.XXXXXX", tmpdir) >= size) {
        return -EOVERFLOW;
    }
B
bellard 已提交
629
    fd = mkstemp(filename);
630 631 632 633 634
    if (fd < 0) {
        return -errno;
    }
    if (close(fd) != 0) {
        unlink(filename);
635 636 637
        return -errno;
    }
    return 0;
B
bellard 已提交
638
#endif
639
}
B
bellard 已提交
640

641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662
/*
 * 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;
}

663
BlockDriver *bdrv_find_protocol(const char *filename,
664 665
                                bool allow_protocol_prefix,
                                Error **errp)
B
bellard 已提交
666 667 668
{
    BlockDriver *drv1;
    char protocol[128];
669
    int len;
B
bellard 已提交
670
    const char *p;
B
bellard 已提交
671

672 673
    /* TODO Drivers without bdrv_file_open must be specified explicitly */

674 675 676 677 678 679 680 681 682 683 684 685
    /*
     * 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;
    }

686
    if (!path_has_protocol(filename) || !allow_protocol_prefix) {
687
        return &bdrv_file;
688
    }
689

690 691
    p = strchr(filename, ':');
    assert(p != NULL);
692 693 694 695 696
    len = p - filename;
    if (len > sizeof(protocol) - 1)
        len = sizeof(protocol) - 1;
    memcpy(protocol, filename, len);
    protocol[len] = '\0';
697
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
698
        if (drv1->protocol_name &&
699
            !strcmp(drv1->protocol_name, protocol)) {
B
bellard 已提交
700
            return drv1;
701
        }
B
bellard 已提交
702
    }
703 704

    error_setg(errp, "Unknown protocol '%s'", protocol);
B
bellard 已提交
705 706 707
    return NULL;
}

708 709 710 711 712 713
/*
 * Guess image format by probing its contents.
 * This is not a good idea when your image is raw (CVE-2008-2004), but
 * we do it anyway for backward compatibility.
 *
 * @buf         contains the image's first @buf_size bytes.
714 715
 * @buf_size    is the buffer size in bytes (generally BLOCK_PROBE_BUF_SIZE,
 *              but can be smaller if the image file is smaller)
716 717 718 719 720 721
 * @filename    is its filename.
 *
 * For all block drivers, call the bdrv_probe() method to get its
 * probing score.
 * Return the first block driver with the highest probing score.
 */
722 723
BlockDriver *bdrv_probe_all(const uint8_t *buf, int buf_size,
                            const char *filename)
724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740
{
    int score_max = 0, score;
    BlockDriver *drv = NULL, *d;

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

    return drv;
}

741
static int find_image_format(BlockDriverState *bs, const char *filename,
742
                             BlockDriver **pdrv, Error **errp)
743
{
744
    BlockDriver *drv;
745
    uint8_t buf[BLOCK_PROBE_BUF_SIZE];
746
    int ret = 0;
747

748
    /* Return the raw BlockDriver * to scsi-generic devices or empty drives */
749
    if (bs->sg || !bdrv_is_inserted(bs) || bdrv_getlength(bs) == 0) {
750
        *pdrv = &bdrv_raw;
751
        return ret;
752
    }
753

B
bellard 已提交
754 755
    ret = bdrv_pread(bs, 0, buf, sizeof(buf));
    if (ret < 0) {
756 757
        error_setg_errno(errp, -ret, "Could not read image for determining its "
                         "format");
758 759
        *pdrv = NULL;
        return ret;
B
bellard 已提交
760 761
    }

762
    drv = bdrv_probe_all(buf, ret, filename);
763
    if (!drv) {
764 765
        error_setg(errp, "Could not determine image format: No compatible "
                   "driver found");
766 767 768 769
        ret = -ENOENT;
    }
    *pdrv = drv;
    return ret;
B
bellard 已提交
770 771
}

772 773
/**
 * Set the current 'total_sectors' value
M
Markus Armbruster 已提交
774
 * Return 0 on success, -errno on error.
775 776 777 778 779
 */
static int refresh_total_sectors(BlockDriverState *bs, int64_t hint)
{
    BlockDriver *drv = bs->drv;

780 781 782 783
    /* Do not attempt drv->bdrv_getlength() on scsi-generic devices */
    if (bs->sg)
        return 0;

784 785 786 787 788 789
    /* 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
790
        hint = DIV_ROUND_UP(length, BDRV_SECTOR_SIZE);
791 792 793 794 795 796
    }

    bs->total_sectors = hint;
    return 0;
}

P
Paolo Bonzini 已提交
797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816
/**
 * 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;
}

817 818 819 820 821 822 823 824 825 826 827
/**
 * 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;
828 829
    } else if (!strcmp(mode, "directsync")) {
        *flags |= BDRV_O_NOCACHE;
830 831 832 833 834 835 836 837 838 839 840 841 842 843
    } 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;
}

844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859
/**
 * 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--;
}

860 861 862 863 864 865 866 867 868 869
/*
 * 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;
}

870 871 872 873 874 875 876 877 878 879 880 881 882 883
/*
 * 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 */
884
    flags &= ~(BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING | BDRV_O_COPY_ON_READ);
885 886 887 888

    return flags;
}

K
Kevin Wolf 已提交
889 890 891 892 893 894 895 896 897 898
/*
 * 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 */
899
    flags &= ~(BDRV_O_SNAPSHOT | BDRV_O_TEMPORARY);
K
Kevin Wolf 已提交
900 901 902 903

    return flags;
}

K
Kevin Wolf 已提交
904 905 906 907 908 909 910 911
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.
     */
912
    open_flags &= ~(BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING | BDRV_O_PROTOCOL);
K
Kevin Wolf 已提交
913 914 915 916

    /*
     * Snapshots should be writable.
     */
917
    if (flags & BDRV_O_TEMPORARY) {
K
Kevin Wolf 已提交
918 919 920 921 922 923
        open_flags |= BDRV_O_RDWR;
    }

    return open_flags;
}

924 925 926
static void bdrv_assign_node_name(BlockDriverState *bs,
                                  const char *node_name,
                                  Error **errp)
927 928
{
    if (!node_name) {
929
        return;
930 931
    }

K
Kevin Wolf 已提交
932
    /* Check for empty string or invalid characters */
933
    if (!id_wellformed(node_name)) {
K
Kevin Wolf 已提交
934
        error_setg(errp, "Invalid node name");
935
        return;
936 937
    }

938
    /* takes care of avoiding namespaces collisions */
939
    if (blk_by_name(node_name)) {
940 941
        error_setg(errp, "node-name=%s is conflicting with a device id",
                   node_name);
942
        return;
943 944
    }

945 946 947
    /* takes care of avoiding duplicates node names */
    if (bdrv_find_node(node_name)) {
        error_setg(errp, "Duplicate node name");
948
        return;
949 950 951 952 953 954 955
    }

    /* 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);
}

956 957
/*
 * Common part for opening disk images and files
958 959
 *
 * Removes all processed options from *options.
960
 */
961
static int bdrv_open_common(BlockDriverState *bs, BlockDriverState *file,
962
    QDict *options, int flags, BlockDriver *drv, Error **errp)
963 964
{
    int ret, open_flags;
K
Kevin Wolf 已提交
965
    const char *filename;
966
    const char *node_name = NULL;
967
    Error *local_err = NULL;
968 969

    assert(drv != NULL);
970
    assert(bs->file == NULL);
971
    assert(options != NULL && bs->options != options);
972

973 974 975 976 977 978
    if (file != NULL) {
        filename = file->filename;
    } else {
        filename = qdict_get_try_str(options, "filename");
    }

979 980 981 982 983 984
    if (drv->bdrv_needs_filename && !filename) {
        error_setg(errp, "The '%s' block driver requires a file name",
                   drv->format_name);
        return -EINVAL;
    }

985
    trace_bdrv_open_common(bs, filename ?: "", flags, drv->format_name);
S
Stefan Hajnoczi 已提交
986

987
    node_name = qdict_get_try_str(options, "node-name");
988
    bdrv_assign_node_name(bs, node_name, &local_err);
989
    if (local_err) {
990 991
        error_propagate(errp, local_err);
        return -EINVAL;
992 993 994
    }
    qdict_del(options, "node-name");

995 996 997 998 999 1000 1001 1002
    /* 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;
    }

1003
    bs->open_flags = flags;
1004
    bs->guest_block_size = 512;
1005
    bs->request_alignment = 512;
A
Asias He 已提交
1006
    bs->zero_beyond_eof = true;
1007 1008 1009 1010
    open_flags = bdrv_open_flags(bs, flags);
    bs->read_only = !(open_flags & BDRV_O_RDWR);

    if (use_bdrv_whitelist && !bdrv_is_whitelisted(drv, bs->read_only)) {
1011 1012 1013 1014 1015
        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);
1016 1017
        return -ENOTSUP;
    }
1018

1019
    assert(bs->copy_on_read == 0); /* bdrv_new() and bdrv_close() make it so */
1020 1021 1022 1023 1024 1025 1026
    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;
        }
1027 1028
    }

1029 1030 1031 1032 1033
    if (filename != NULL) {
        pstrcpy(bs->filename, sizeof(bs->filename), filename);
    } else {
        bs->filename[0] = '\0';
    }
M
Max Reitz 已提交
1034
    pstrcpy(bs->exact_filename, sizeof(bs->exact_filename), bs->filename);
1035 1036

    bs->drv = drv;
1037
    bs->opaque = g_malloc0(drv->instance_size);
1038

1039
    bs->enable_write_cache = !!(flags & BDRV_O_CACHE_WB);
1040

1041 1042
    /* Open the image, either directly or using a protocol */
    if (drv->bdrv_file_open) {
1043
        assert(file == NULL);
1044
        assert(!drv->bdrv_needs_filename || filename != NULL);
1045
        ret = drv->bdrv_file_open(bs, options, open_flags, &local_err);
1046
    } else {
1047
        if (file == NULL) {
1048 1049
            error_setg(errp, "Can't use '%s' as a block driver for the "
                       "protocol level", drv->format_name);
1050 1051 1052
            ret = -EINVAL;
            goto free_and_fail;
        }
1053
        bs->file = file;
1054
        ret = drv->bdrv_open(bs, options, open_flags, &local_err);
1055 1056
    }

1057
    if (ret < 0) {
1058
        if (local_err) {
1059
            error_propagate(errp, local_err);
1060 1061
        } else if (bs->filename[0]) {
            error_setg_errno(errp, -ret, "Could not open '%s'", bs->filename);
1062 1063 1064
        } else {
            error_setg_errno(errp, -ret, "Could not open image");
        }
1065 1066 1067
        goto free_and_fail;
    }

1068 1069 1070 1071 1072 1073 1074
    if (bs->encrypted) {
        error_report("Encrypted images are deprecated");
        error_printf("Support for them will be removed in a future release.\n"
                     "You can use 'qemu-img convert' to convert your image"
                     " to an unencrypted one.\n");
    }

1075 1076
    ret = refresh_total_sectors(bs, bs->total_sectors);
    if (ret < 0) {
1077
        error_setg_errno(errp, -ret, "Could not refresh total sector count");
1078
        goto free_and_fail;
1079
    }
1080

1081 1082 1083 1084 1085 1086 1087
    bdrv_refresh_limits(bs, &local_err);
    if (local_err) {
        error_propagate(errp, local_err);
        ret = -EINVAL;
        goto free_and_fail;
    }

1088
    assert(bdrv_opt_mem_align(bs) != 0);
1089
    assert((bs->request_alignment != 0) || bs->sg);
1090 1091 1092
    return 0;

free_and_fail:
1093
    bs->file = NULL;
1094
    g_free(bs->opaque);
1095 1096 1097 1098 1099
    bs->opaque = NULL;
    bs->drv = NULL;
    return ret;
}

1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126
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
1127
/*
K
Kevin Wolf 已提交
1128 1129
 * Fills in default options for opening images and converts the legacy
 * filename/flags pair to option QDict entries.
K
Kevin Wolf 已提交
1130
 */
1131
static int bdrv_fill_options(QDict **options, const char **pfilename, int flags,
1132
                             BlockDriver *drv, Error **errp)
B
bellard 已提交
1133
{
1134
    const char *filename = *pfilename;
1135
    const char *drvname;
1136
    bool protocol = flags & BDRV_O_PROTOCOL;
1137
    bool parse_filename = false;
1138
    Error *local_err = NULL;
B
bellard 已提交
1139

1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154
    /* 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
1155
    /* Fetch the file name from the options QDict if necessary */
1156
    if (protocol && filename) {
K
Kevin Wolf 已提交
1157 1158 1159 1160 1161 1162 1163 1164
        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
1165 1166
    }

1167
    /* Find the right block driver */
K
Kevin Wolf 已提交
1168
    filename = qdict_get_try_str(*options, "filename");
1169
    drvname = qdict_get_try_str(*options, "driver");
K
Kevin Wolf 已提交
1170

1171 1172 1173 1174 1175 1176 1177 1178 1179 1180
    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) {
1181
                drv = bdrv_find_protocol(filename, parse_filename, errp);
1182 1183 1184 1185 1186 1187 1188 1189
                if (!drv) {
                    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
1190 1191
                return -EINVAL;
            }
1192 1193 1194 1195 1196 1197
        } else if (drvname) {
            drv = bdrv_find_format(drvname);
            if (!drv) {
                error_setg(errp, "Unknown driver '%s'", drvname);
                return -ENOENT;
            }
1198
        }
1199 1200
    }

1201
    assert(drv || !protocol);
1202

K
Kevin Wolf 已提交
1203
    /* Driver-specific filename parsing */
1204
    if (drv && drv->bdrv_parse_filename && parse_filename) {
1205
        drv->bdrv_parse_filename(filename, *options, &local_err);
1206
        if (local_err) {
1207
            error_propagate(errp, local_err);
K
Kevin Wolf 已提交
1208
            return -EINVAL;
1209
        }
1210 1211 1212 1213

        if (!drv->bdrv_needs_filename) {
            qdict_del(*options, "filename");
        }
1214 1215
    }

K
Kevin Wolf 已提交
1216 1217 1218
    return 0;
}

F
Fam Zheng 已提交
1219 1220 1221
void bdrv_set_backing_hd(BlockDriverState *bs, BlockDriverState *backing_hd)
{

1222 1223 1224 1225 1226 1227
    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'",
1228
                   bdrv_get_device_name(bs));
1229 1230
    }

F
Fam Zheng 已提交
1231 1232
    bs->backing_hd = backing_hd;
    if (!backing_hd) {
1233 1234
        error_free(bs->backing_blocker);
        bs->backing_blocker = NULL;
F
Fam Zheng 已提交
1235 1236 1237 1238 1239 1240
        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 : "");
1241 1242 1243

    bdrv_op_block_all(bs->backing_hd, bs->backing_blocker);
    /* Otherwise we won't be able to commit due to check in bdrv_commit */
1244
    bdrv_op_unblock(bs->backing_hd, BLOCK_OP_TYPE_COMMIT_TARGET,
1245
                    bs->backing_blocker);
F
Fam Zheng 已提交
1246
out:
1247
    bdrv_refresh_limits(bs, NULL);
F
Fam Zheng 已提交
1248 1249
}

1250 1251 1252 1253 1254 1255 1256 1257
/*
 * 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.
 */
1258
int bdrv_open_backing_file(BlockDriverState *bs, QDict *options, Error **errp)
P
Paolo Bonzini 已提交
1259
{
1260
    char *backing_filename = g_malloc0(PATH_MAX);
K
Kevin Wolf 已提交
1261
    int ret = 0;
F
Fam Zheng 已提交
1262
    BlockDriverState *backing_hd;
1263
    Error *local_err = NULL;
P
Paolo Bonzini 已提交
1264 1265

    if (bs->backing_hd != NULL) {
1266
        QDECREF(options);
1267
        goto free_exit;
P
Paolo Bonzini 已提交
1268 1269
    }

1270 1271 1272 1273 1274
    /* NULL means an empty set of options */
    if (options == NULL) {
        options = qdict_new();
    }

P
Paolo Bonzini 已提交
1275
    bs->open_flags &= ~BDRV_O_NO_BACKING;
1276 1277 1278
    if (qdict_haskey(options, "file.filename")) {
        backing_filename[0] = '\0';
    } else if (bs->backing_file[0] == '\0' && qdict_size(options) == 0) {
1279
        QDECREF(options);
1280
        goto free_exit;
F
Fam Zheng 已提交
1281
    } else {
1282 1283 1284 1285 1286 1287 1288 1289
        bdrv_get_full_backing_filename(bs, backing_filename, PATH_MAX,
                                       &local_err);
        if (local_err) {
            ret = -EINVAL;
            error_propagate(errp, local_err);
            QDECREF(options);
            goto free_exit;
        }
P
Paolo Bonzini 已提交
1290 1291
    }

1292 1293 1294 1295 1296 1297 1298
    if (!bs->drv || !bs->drv->supports_backing) {
        ret = -EINVAL;
        error_setg(errp, "Driver doesn't support backing files");
        QDECREF(options);
        goto free_exit;
    }

1299
    backing_hd = bdrv_new();
F
Fam Zheng 已提交
1300

1301 1302
    if (bs->backing_format[0] != '\0' && !qdict_haskey(options, "driver")) {
        qdict_put(options, "driver", qstring_from_str(bs->backing_format));
P
Paolo Bonzini 已提交
1303 1304
    }

1305
    assert(bs->backing_hd == NULL);
F
Fam Zheng 已提交
1306
    ret = bdrv_open(&backing_hd,
1307
                    *backing_filename ? backing_filename : NULL, NULL, options,
1308
                    bdrv_backing_flags(bs->open_flags), NULL, &local_err);
P
Paolo Bonzini 已提交
1309
    if (ret < 0) {
F
Fam Zheng 已提交
1310 1311
        bdrv_unref(backing_hd);
        backing_hd = NULL;
P
Paolo Bonzini 已提交
1312
        bs->open_flags |= BDRV_O_NO_BACKING;
1313 1314 1315
        error_setg(errp, "Could not open backing file: %s",
                   error_get_pretty(local_err));
        error_free(local_err);
1316
        goto free_exit;
P
Paolo Bonzini 已提交
1317
    }
F
Fam Zheng 已提交
1318
    bdrv_set_backing_hd(bs, backing_hd);
P
Peter Feiner 已提交
1319

1320 1321 1322
free_exit:
    g_free(backing_filename);
    return ret;
P
Paolo Bonzini 已提交
1323 1324
}

M
Max Reitz 已提交
1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337
/*
 * 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.
1338 1339
 *
 * To conform with the behavior of bdrv_open(), *pbs has to be NULL.
M
Max Reitz 已提交
1340 1341 1342
 */
int bdrv_open_image(BlockDriverState **pbs, const char *filename,
                    QDict *options, const char *bdref_key, int flags,
1343
                    bool allow_none, Error **errp)
M
Max Reitz 已提交
1344 1345 1346 1347 1348 1349
{
    QDict *image_options;
    int ret;
    char *bdref_key_dot;
    const char *reference;

1350 1351 1352
    assert(pbs);
    assert(*pbs == NULL);

M
Max Reitz 已提交
1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365
    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;
        }
1366
        QDECREF(image_options);
M
Max Reitz 已提交
1367 1368 1369
        goto done;
    }

1370
    ret = bdrv_open(pbs, filename, reference, image_options, flags, NULL, errp);
M
Max Reitz 已提交
1371 1372 1373 1374 1375 1376

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

1377
int bdrv_append_temp_snapshot(BlockDriverState *bs, int flags, Error **errp)
1378 1379
{
    /* TODO: extra byte is a hack to ensure MAX_PATH space on Windows. */
1380
    char *tmp_filename = g_malloc0(PATH_MAX + 1);
1381
    int64_t total_size;
1382
    QemuOpts *opts = NULL;
1383 1384 1385 1386 1387 1388 1389 1390 1391
    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 */
1392 1393
    total_size = bdrv_getlength(bs);
    if (total_size < 0) {
1394
        ret = total_size;
1395
        error_setg_errno(errp, -total_size, "Could not get image size");
1396
        goto out;
1397
    }
1398 1399

    /* Create the temporary image */
1400
    ret = get_tmp_filename(tmp_filename, PATH_MAX + 1);
1401 1402
    if (ret < 0) {
        error_setg_errno(errp, -ret, "Could not get temporary filename");
1403
        goto out;
1404 1405
    }

1406
    opts = qemu_opts_create(bdrv_qcow2.create_opts, NULL, 0,
C
Chunyan Liu 已提交
1407
                            &error_abort);
1408
    qemu_opt_set_number(opts, BLOCK_OPT_SIZE, total_size, &error_abort);
1409
    ret = bdrv_create(&bdrv_qcow2, tmp_filename, opts, &local_err);
1410
    qemu_opts_del(opts);
1411 1412 1413 1414 1415
    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);
1416
        goto out;
1417 1418 1419 1420 1421 1422 1423 1424 1425
    }

    /* 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));

1426
    bs_snapshot = bdrv_new();
1427 1428

    ret = bdrv_open(&bs_snapshot, NULL, NULL, snapshot_options,
1429
                    flags, &bdrv_qcow2, &local_err);
1430 1431
    if (ret < 0) {
        error_propagate(errp, local_err);
1432
        goto out;
1433 1434 1435
    }

    bdrv_append(bs_snapshot, bs);
1436 1437 1438

out:
    g_free(tmp_filename);
1439
    return ret;
1440 1441
}

K
Kevin Wolf 已提交
1442 1443
/*
 * Opens a disk image (raw, qcow2, vmdk, ...)
1444 1445 1446 1447 1448
 *
 * 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.
1449 1450 1451
 *
 * 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.
1452 1453 1454 1455
 *
 * 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
1456
 */
1457 1458 1459
int bdrv_open(BlockDriverState **pbs, const char *filename,
              const char *reference, QDict *options, int flags,
              BlockDriver *drv, Error **errp)
B
bellard 已提交
1460
{
K
Kevin Wolf 已提交
1461
    int ret;
1462
    BlockDriverState *file = NULL, *bs;
1463
    const char *drvname;
1464
    Error *local_err = NULL;
1465
    int snapshot_flags = 0;
B
bellard 已提交
1466

1467 1468
    assert(pbs);

1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493
    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;
    }

1494 1495 1496
    if (*pbs) {
        bs = *pbs;
    } else {
1497
        bs = bdrv_new();
1498 1499
    }

1500 1501 1502 1503 1504
    /* NULL means an empty set of options */
    if (options == NULL) {
        options = qdict_new();
    }

1505
    ret = bdrv_fill_options(&options, &filename, flags, drv, &local_err);
1506 1507 1508 1509
    if (local_err) {
        goto fail;
    }

1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529
    /* 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;
    }

1530
    bs->options = options;
1531
    options = qdict_clone_shallow(options);
1532

1533
    /* Open image file without format layer */
1534 1535 1536 1537 1538 1539 1540 1541
    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);
        }
1542

1543 1544 1545 1546 1547 1548 1549
        assert(file == NULL);
        ret = bdrv_open_image(&file, filename, options, "file",
                              bdrv_inherited_flags(flags),
                              true, &local_err);
        if (ret < 0) {
            goto fail;
        }
1550 1551
    }

1552
    /* Image format probing */
1553
    bs->probed = !drv;
1554
    if (!drv && file) {
1555 1556
        ret = find_image_format(file, filename, &drv, &local_err);
        if (ret < 0) {
1557
            goto fail;
1558
        }
1559
    } else if (!drv) {
1560 1561
        error_setg(errp, "Must specify either driver or file");
        ret = -EINVAL;
1562
        goto fail;
B
bellard 已提交
1563
    }
K
Kevin Wolf 已提交
1564 1565

    /* Open the image */
1566
    ret = bdrv_open_common(bs, file, options, flags, drv, &local_err);
K
Kevin Wolf 已提交
1567
    if (ret < 0) {
1568
        goto fail;
1569 1570
    }

1571
    if (file && (bs->file != file)) {
F
Fam Zheng 已提交
1572
        bdrv_unref(file);
1573 1574 1575
        file = NULL;
    }

K
Kevin Wolf 已提交
1576
    /* If there is a backing file, use it */
P
Paolo Bonzini 已提交
1577
    if ((flags & BDRV_O_NO_BACKING) == 0) {
1578 1579
        QDict *backing_options;

1580
        qdict_extract_subqdict(options, &backing_options, "backing.");
1581
        ret = bdrv_open_backing_file(bs, backing_options, &local_err);
K
Kevin Wolf 已提交
1582
        if (ret < 0) {
1583
            goto close_and_fail;
K
Kevin Wolf 已提交
1584 1585 1586
        }
    }

M
Max Reitz 已提交
1587 1588
    bdrv_refresh_filename(bs);

1589 1590
    /* For snapshot=on, create a temporary qcow2 overlay. bs points to the
     * temporary snapshot afterwards. */
1591
    if (snapshot_flags) {
1592
        ret = bdrv_append_temp_snapshot(bs, snapshot_flags, &local_err);
1593 1594 1595 1596 1597
        if (local_err) {
            goto close_and_fail;
        }
    }

1598
    /* Check if any unknown options were used */
1599
    if (options && (qdict_size(options) != 0)) {
1600
        const QDictEntry *entry = qdict_first(options);
1601 1602 1603 1604 1605 1606
        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,
1607
                       bdrv_get_device_name(bs), entry->key);
1608
        }
1609 1610 1611 1612 1613

        ret = -EINVAL;
        goto close_and_fail;
    }

K
Kevin Wolf 已提交
1614
    if (!bdrv_key_required(bs)) {
1615 1616 1617
        if (bs->blk) {
            blk_dev_change_media_cb(bs->blk, true);
        }
1618 1619 1620 1621 1622 1623 1624
    } 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
1625 1626
    }

1627
    QDECREF(options);
1628
    *pbs = bs;
K
Kevin Wolf 已提交
1629 1630
    return 0;

1631
fail:
1632
    if (file != NULL) {
F
Fam Zheng 已提交
1633
        bdrv_unref(file);
1634
    }
1635
    QDECREF(bs->options);
1636
    QDECREF(options);
1637
    bs->options = NULL;
1638 1639 1640 1641 1642 1643
    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);
    }
1644
    if (local_err) {
1645 1646
        error_propagate(errp, local_err);
    }
1647
    return ret;
1648

1649
close_and_fail:
1650 1651 1652 1653 1654 1655
    /* See fail path, but now the BDS has to be always closed */
    if (*pbs) {
        bdrv_close(bs);
    } else {
        bdrv_unref(bs);
    }
1656
    QDECREF(options);
1657
    if (local_err) {
1658 1659
        error_propagate(errp, local_err);
    }
K
Kevin Wolf 已提交
1660 1661 1662
    return ret;
}

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
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);
    }

1698 1699 1700
    /* bdrv_open() masks this flag out */
    flags &= ~BDRV_O_PROTOCOL;

1701
    if (bs->file) {
1702
        bdrv_reopen_queue(bs_queue, bs->file, bdrv_inherited_flags(flags));
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 1735 1736 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 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815
    }

    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,
1816
                  bdrv_get_device_name(reopen_state->bs));
1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833
        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 {
1834 1835
                error_setg(errp, "failed while preparing to reopen image '%s'",
                           reopen_state->bs->filename);
1836 1837 1838 1839 1840 1841 1842
            }
            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,
1843
                  drv->format_name, bdrv_get_device_name(reopen_state->bs),
1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877
                 "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);
1878

1879
    bdrv_refresh_limits(reopen_state->bs, NULL);
1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899
}

/*
 * 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
1900 1901
void bdrv_close(BlockDriverState *bs)
{
M
Max Reitz 已提交
1902 1903
    BdrvAioNotifier *ban, *ban_next;

1904 1905 1906
    if (bs->job) {
        block_job_cancel_sync(bs->job);
    }
1907 1908 1909
    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
1910
    notifier_list_notify(&bs->close_notifiers, bs);
K
Kevin Wolf 已提交
1911

1912
    if (bs->drv) {
1913
        if (bs->backing_hd) {
1914 1915 1916
            BlockDriverState *backing_hd = bs->backing_hd;
            bdrv_set_backing_hd(bs, NULL);
            bdrv_unref(backing_hd);
1917
        }
B
bellard 已提交
1918
        bs->drv->bdrv_close(bs);
1919
        g_free(bs->opaque);
B
bellard 已提交
1920 1921
        bs->opaque = NULL;
        bs->drv = NULL;
1922
        bs->copy_on_read = 0;
1923 1924
        bs->backing_file[0] = '\0';
        bs->backing_format[0] = '\0';
1925 1926 1927 1928
        bs->total_sectors = 0;
        bs->encrypted = 0;
        bs->valid_key = 0;
        bs->sg = 0;
A
Asias He 已提交
1929
        bs->zero_beyond_eof = false;
1930 1931
        QDECREF(bs->options);
        bs->options = NULL;
M
Max Reitz 已提交
1932 1933
        QDECREF(bs->full_open_options);
        bs->full_open_options = NULL;
B
bellard 已提交
1934

1935
        if (bs->file != NULL) {
F
Fam Zheng 已提交
1936
            bdrv_unref(bs->file);
1937
            bs->file = NULL;
1938
        }
B
bellard 已提交
1939
    }
Z
Zhi Yong Wu 已提交
1940

1941 1942 1943
    if (bs->blk) {
        blk_dev_change_media_cb(bs->blk, false);
    }
P
Pavel Hrdina 已提交
1944

Z
Zhi Yong Wu 已提交
1945 1946 1947 1948
    /*throttling disk I/O limits*/
    if (bs->io_limits_enabled) {
        bdrv_io_limits_disable(bs);
    }
M
Max Reitz 已提交
1949 1950 1951 1952 1953

    QLIST_FOREACH_SAFE(ban, &bs->aio_notifiers, list, ban_next) {
        g_free(ban);
    }
    QLIST_INIT(&bs->aio_notifiers);
B
bellard 已提交
1954 1955
}

1956 1957 1958 1959
void bdrv_close_all(void)
{
    BlockDriverState *bs;

1960
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
1961 1962 1963
        AioContext *aio_context = bdrv_get_aio_context(bs);

        aio_context_acquire(aio_context);
1964
        bdrv_close(bs);
1965
        aio_context_release(aio_context);
1966 1967 1968
    }
}

1969 1970 1971 1972 1973 1974
/* 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;
    }
1975 1976 1977 1978
    if (!qemu_co_queue_empty(&bs->throttled_reqs[0])) {
        return true;
    }
    if (!qemu_co_queue_empty(&bs->throttled_reqs[1])) {
1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989
        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;
}

S
Stefan Hajnoczi 已提交
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017
static bool bdrv_drain_one(BlockDriverState *bs)
{
    bool bs_busy;

    bdrv_flush_io_queue(bs);
    bdrv_start_throttled_reqs(bs);
    bs_busy = bdrv_requests_pending(bs);
    bs_busy |= aio_poll(bdrv_get_aio_context(bs), bs_busy);
    return bs_busy;
}

/*
 * Wait for pending requests to complete on a single BlockDriverState subtree
 *
 * See the warning in bdrv_drain_all().  This function can only be called if
 * you are sure nothing can generate I/O because you have op blockers
 * installed.
 *
 * Note that unlike bdrv_drain_all(), the caller must hold the BlockDriverState
 * AioContext.
 */
void bdrv_drain(BlockDriverState *bs)
{
    while (bdrv_drain_one(bs)) {
        /* Keep iterating */
    }
}

2018 2019 2020 2021 2022
/*
 * 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.
2023 2024 2025 2026 2027 2028
 *
 * 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.
2029 2030 2031
 */
void bdrv_drain_all(void)
{
2032 2033
    /* Always run first iteration so any pending completion BHs run */
    bool busy = true;
2034 2035
    BlockDriverState *bs;

2036
    while (busy) {
2037 2038
        busy = false;

2039
        QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
2040 2041 2042
            AioContext *aio_context = bdrv_get_aio_context(bs);

            aio_context_acquire(aio_context);
S
Stefan Hajnoczi 已提交
2043
            busy |= bdrv_drain_one(bs);
2044 2045
            aio_context_release(aio_context);
        }
2046 2047 2048
    }
}

2049 2050
/* make a BlockDriverState anonymous by removing from bdrv_state and
 * graph_bdrv_state list.
2051 2052 2053
   Also, NULL terminate the device_name to prevent double remove */
void bdrv_make_anon(BlockDriverState *bs)
{
2054 2055 2056 2057 2058 2059 2060 2061
    /*
     * Take care to remove bs from bdrv_states only when it's actually
     * in it.  Note that bs->device_list.tqe_prev is initially null,
     * and gets set to non-null by QTAILQ_INSERT_TAIL().  Establish
     * the useful invariant "bs in bdrv_states iff bs->tqe_prev" by
     * resetting it to null on remove.
     */
    if (bs->device_list.tqe_prev) {
2062
        QTAILQ_REMOVE(&bdrv_states, bs, device_list);
2063
        bs->device_list.tqe_prev = NULL;
2064
    }
2065 2066 2067 2068
    if (bs->node_name[0] != '\0') {
        QTAILQ_REMOVE(&graph_bdrv_states, bs, node_list);
    }
    bs->node_name[0] = '\0';
2069 2070
}

P
Paolo Bonzini 已提交
2071 2072 2073 2074 2075 2076 2077
static void bdrv_rebind(BlockDriverState *bs)
{
    if (bs->drv && bs->drv->bdrv_rebind) {
        bs->drv->bdrv_rebind(bs);
    }
}

2078 2079
static void bdrv_move_feature_fields(BlockDriverState *bs_dest,
                                     BlockDriverState *bs_src)
2080
{
2081
    /* move some fields that need to stay attached to the device */
2082 2083

    /* dev info */
2084
    bs_dest->guest_block_size   = bs_src->guest_block_size;
2085
    bs_dest->copy_on_read       = bs_src->copy_on_read;
2086

2087
    bs_dest->enable_write_cache = bs_src->enable_write_cache;
2088

2089 2090 2091 2092 2093 2094
    /* 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];
2095
    bs_dest->io_limits_enabled  = bs_src->io_limits_enabled;
2096 2097

    /* r/w error */
2098 2099
    bs_dest->on_read_error      = bs_src->on_read_error;
    bs_dest->on_write_error     = bs_src->on_write_error;
2100 2101

    /* i/o status */
2102 2103
    bs_dest->iostatus_enabled   = bs_src->iostatus_enabled;
    bs_dest->iostatus           = bs_src->iostatus;
2104

2105
    /* dirty bitmap */
F
Fam Zheng 已提交
2106
    bs_dest->dirty_bitmaps      = bs_src->dirty_bitmaps;
2107

2108 2109 2110
    /* reference count */
    bs_dest->refcnt             = bs_src->refcnt;

2111
    /* job */
2112
    bs_dest->job                = bs_src->job;
2113

2114
    /* keep the same entry in bdrv_states */
2115
    bs_dest->device_list = bs_src->device_list;
2116 2117
    bs_dest->blk = bs_src->blk;

2118 2119
    memcpy(bs_dest->op_blockers, bs_src->op_blockers,
           sizeof(bs_dest->op_blockers));
2120
}
2121

2122 2123 2124 2125 2126 2127 2128 2129
/*
 * 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.
 *
2130
 * bs_new must not be attached to a BlockBackend.
2131 2132 2133 2134 2135 2136
 *
 * 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
2137

2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148
    /* 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);
    }

2149
    /* bs_new must be unattached and shouldn't have anything fancy enabled */
2150
    assert(!bs_new->blk);
F
Fam Zheng 已提交
2151
    assert(QLIST_EMPTY(&bs_new->dirty_bitmaps));
2152 2153
    assert(bs_new->job == NULL);
    assert(bs_new->io_limits_enabled == false);
2154
    assert(!throttle_have_timer(&bs_new->throttle_state));
2155

2156 2157 2158
    tmp = *bs_new;
    *bs_new = *bs_old;
    *bs_old = tmp;
2159

2160 2161 2162 2163
    /* 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);
2164

2165
    /* bs_new must remain unattached */
2166
    assert(!bs_new->blk);
2167 2168 2169 2170

    /* Check a few fields that should remain attached to the device */
    assert(bs_new->job == NULL);
    assert(bs_new->io_limits_enabled == false);
2171
    assert(!throttle_have_timer(&bs_new->throttle_state));
P
Paolo Bonzini 已提交
2172

2173 2174 2175 2176 2177 2178 2179 2180
    /* 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
2181
    bdrv_rebind(bs_new);
2182 2183 2184 2185 2186 2187 2188 2189 2190 2191
    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.
 *
2192
 * bs_new must not be attached to a BlockBackend.
2193 2194 2195 2196 2197 2198 2199 2200 2201
 *
 * 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
2202
    bdrv_set_backing_hd(bs_top, bs_new);
2203 2204
}

F
Fam Zheng 已提交
2205
static void bdrv_delete(BlockDriverState *bs)
B
bellard 已提交
2206
{
2207
    assert(!bs->job);
2208
    assert(bdrv_op_blocker_is_empty(bs));
F
Fam Zheng 已提交
2209
    assert(!bs->refcnt);
F
Fam Zheng 已提交
2210
    assert(QLIST_EMPTY(&bs->dirty_bitmaps));
2211

2212 2213
    bdrv_close(bs);

2214
    /* remove from list, if necessary */
2215
    bdrv_make_anon(bs);
2216

2217
    g_free(bs);
B
bellard 已提交
2218 2219
}

A
aliguori 已提交
2220 2221 2222
/*
 * Run consistency checks on an image
 *
2223
 * Returns 0 if the check could be completed (it doesn't mean that the image is
2224
 * free of errors) or -errno when an internal error occurred. The results of the
2225
 * check are stored in res.
A
aliguori 已提交
2226
 */
2227
int bdrv_check(BlockDriverState *bs, BdrvCheckResult *res, BdrvCheckMode fix)
A
aliguori 已提交
2228
{
2229 2230 2231
    if (bs->drv == NULL) {
        return -ENOMEDIUM;
    }
A
aliguori 已提交
2232 2233 2234 2235
    if (bs->drv->bdrv_check == NULL) {
        return -ENOTSUP;
    }

2236
    memset(res, 0, sizeof(*res));
2237
    return bs->drv->bdrv_check(bs, res, fix);
A
aliguori 已提交
2238 2239
}

2240 2241
#define COMMIT_BUF_SECTORS 2048

2242 2243 2244
/* commit COW file into the raw image */
int bdrv_commit(BlockDriverState *bs)
{
B
bellard 已提交
2245
    BlockDriver *drv = bs->drv;
2246
    int64_t sector, total_sectors, length, backing_length;
2247
    int n, ro, open_flags;
2248
    int ret = 0;
2249
    uint8_t *buf = NULL;
2250

B
bellard 已提交
2251 2252
    if (!drv)
        return -ENOMEDIUM;
2253

2254 2255
    if (!bs->backing_hd) {
        return -ENOTSUP;
2256 2257
    }

2258 2259
    if (bdrv_op_is_blocked(bs, BLOCK_OP_TYPE_COMMIT_SOURCE, NULL) ||
        bdrv_op_is_blocked(bs->backing_hd, BLOCK_OP_TYPE_COMMIT_TARGET, NULL)) {
2260 2261 2262
        return -EBUSY;
    }

2263 2264 2265 2266
    ro = bs->backing_hd->read_only;
    open_flags =  bs->backing_hd->open_flags;

    if (ro) {
2267 2268
        if (bdrv_reopen(bs->backing_hd, open_flags | BDRV_O_RDWR, NULL)) {
            return -EACCES;
2269
        }
B
bellard 已提交
2270
    }
2271

2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294
    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;
2295 2296 2297 2298 2299 2300 2301 2302

    /* qemu_try_blockalign() for bs will choose an alignment that works for
     * bs->backing_hd as well, so no need to compare the alignment manually. */
    buf = qemu_try_blockalign(bs, COMMIT_BUF_SECTORS * BDRV_SECTOR_SIZE);
    if (buf == NULL) {
        ret = -ENOMEM;
        goto ro_cleanup;
    }
2303 2304

    for (sector = 0; sector < total_sectors; sector += n) {
2305 2306 2307 2308 2309
        ret = bdrv_is_allocated(bs, sector, COMMIT_BUF_SECTORS, &n);
        if (ret < 0) {
            goto ro_cleanup;
        }
        if (ret) {
K
Kevin Wolf 已提交
2310 2311
            ret = bdrv_read(bs, sector, buf, n);
            if (ret < 0) {
2312 2313 2314
                goto ro_cleanup;
            }

K
Kevin Wolf 已提交
2315 2316
            ret = bdrv_write(bs->backing_hd, sector, buf, n);
            if (ret < 0) {
2317 2318
                goto ro_cleanup;
            }
B
bellard 已提交
2319
        }
2320
    }
2321

2322 2323
    if (drv->bdrv_make_empty) {
        ret = drv->bdrv_make_empty(bs);
K
Kevin Wolf 已提交
2324 2325 2326
        if (ret < 0) {
            goto ro_cleanup;
        }
2327 2328
        bdrv_flush(bs);
    }
2329

2330 2331 2332 2333
    /*
     * Make sure all data we wrote to the backing device is actually
     * stable on disk.
     */
K
Kevin Wolf 已提交
2334
    if (bs->backing_hd) {
2335
        bdrv_flush(bs->backing_hd);
K
Kevin Wolf 已提交
2336
    }
2337

K
Kevin Wolf 已提交
2338
    ret = 0;
2339
ro_cleanup:
2340
    qemu_vfree(buf);
2341 2342

    if (ro) {
2343 2344
        /* ignoring error return here */
        bdrv_reopen(bs->backing_hd, open_flags & ~BDRV_O_RDWR, NULL);
2345 2346
    }

2347
    return ret;
2348 2349
}

2350
int bdrv_commit_all(void)
2351 2352 2353
{
    BlockDriverState *bs;

2354
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
2355 2356 2357
        AioContext *aio_context = bdrv_get_aio_context(bs);

        aio_context_acquire(aio_context);
2358 2359 2360
        if (bs->drv && bs->backing_hd) {
            int ret = bdrv_commit(bs);
            if (ret < 0) {
2361
                aio_context_release(aio_context);
2362 2363
                return ret;
            }
2364
        }
2365
        aio_context_release(aio_context);
2366
    }
2367
    return 0;
2368 2369
}

S
Stefan Hajnoczi 已提交
2370 2371 2372 2373 2374 2375 2376
/**
 * 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)
{
2377 2378 2379 2380
    if (req->serialising) {
        req->bs->serialising_in_flight--;
    }

S
Stefan Hajnoczi 已提交
2381
    QLIST_REMOVE(req, list);
2382
    qemu_co_queue_restart_all(&req->wait_queue);
S
Stefan Hajnoczi 已提交
2383 2384 2385 2386 2387 2388 2389
}

/**
 * Add an active request to the tracked requests list
 */
static void tracked_request_begin(BdrvTrackedRequest *req,
                                  BlockDriverState *bs,
2390 2391
                                  int64_t offset,
                                  unsigned int bytes, bool is_write)
S
Stefan Hajnoczi 已提交
2392 2393 2394
{
    *req = (BdrvTrackedRequest){
        .bs = bs,
2395 2396 2397 2398 2399
        .offset         = offset,
        .bytes          = bytes,
        .is_write       = is_write,
        .co             = qemu_coroutine_self(),
        .serialising    = false,
2400 2401
        .overlap_offset = offset,
        .overlap_bytes  = bytes,
S
Stefan Hajnoczi 已提交
2402 2403
    };

2404 2405
    qemu_co_queue_init(&req->wait_queue);

S
Stefan Hajnoczi 已提交
2406 2407 2408
    QLIST_INSERT_HEAD(&bs->tracked_requests, req, list);
}

2409
static void mark_request_serialising(BdrvTrackedRequest *req, uint64_t align)
2410
{
2411
    int64_t overlap_offset = req->offset & ~(align - 1);
2412 2413
    unsigned int overlap_bytes = ROUND_UP(req->offset + req->bytes, align)
                               - overlap_offset;
2414

2415 2416 2417 2418
    if (!req->serialising) {
        req->bs->serialising_in_flight++;
        req->serialising = true;
    }
2419 2420 2421

    req->overlap_offset = MIN(req->overlap_offset, overlap_offset);
    req->overlap_bytes = MAX(req->overlap_bytes, overlap_bytes);
2422 2423
}

S
Stefan Hajnoczi 已提交
2424 2425 2426
/**
 * Round a region to cluster boundaries
 */
2427 2428 2429 2430
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
2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444
{
    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);
    }
}

2445
static int bdrv_get_cluster_size(BlockDriverState *bs)
2446 2447
{
    BlockDriverInfo bdi;
2448
    int ret;
2449

2450 2451 2452
    ret = bdrv_get_info(bs, &bdi);
    if (ret < 0 || bdi.cluster_size == 0) {
        return bs->request_alignment;
2453
    } else {
2454
        return bdi.cluster_size;
2455 2456 2457
    }
}

2458
static bool tracked_request_overlaps(BdrvTrackedRequest *req,
2459 2460
                                     int64_t offset, unsigned int bytes)
{
S
Stefan Hajnoczi 已提交
2461
    /*        aaaa   bbbb */
2462
    if (offset >= req->overlap_offset + req->overlap_bytes) {
S
Stefan Hajnoczi 已提交
2463 2464 2465
        return false;
    }
    /* bbbb   aaaa        */
2466
    if (req->overlap_offset >= offset + bytes) {
S
Stefan Hajnoczi 已提交
2467 2468 2469
        return false;
    }
    return true;
2470 2471
}

2472
static bool coroutine_fn wait_serialising_requests(BdrvTrackedRequest *self)
2473
{
2474
    BlockDriverState *bs = self->bs;
2475 2476
    BdrvTrackedRequest *req;
    bool retry;
2477
    bool waited = false;
2478

2479
    if (!bs->serialising_in_flight) {
2480
        return false;
2481 2482
    }

2483 2484 2485
    do {
        retry = false;
        QLIST_FOREACH(req, &bs->tracked_requests, list) {
2486
            if (req == self || (!req->serialising && !self->serialising)) {
2487 2488
                continue;
            }
2489 2490 2491
            if (tracked_request_overlaps(req, self->overlap_offset,
                                         self->overlap_bytes))
            {
2492 2493 2494 2495 2496 2497
                /* 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);

2498 2499 2500 2501 2502 2503 2504 2505
                /* 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;
2506
                    waited = true;
2507 2508
                    break;
                }
2509 2510 2511
            }
        }
    } while (retry);
2512 2513

    return waited;
2514 2515
}

K
Kevin Wolf 已提交
2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527
/*
 * 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;
2528
    int ret;
K
Kevin Wolf 已提交
2529

2530 2531 2532 2533 2534
    /* Backing file format doesn't make sense without a backing file */
    if (backing_fmt && !backing_file) {
        return -EINVAL;
    }

K
Kevin Wolf 已提交
2535
    if (drv->bdrv_change_backing_file != NULL) {
2536
        ret = drv->bdrv_change_backing_file(bs, backing_file, backing_fmt);
K
Kevin Wolf 已提交
2537
    } else {
2538
        ret = -ENOTSUP;
K
Kevin Wolf 已提交
2539
    }
2540 2541 2542 2543 2544 2545

    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
2546 2547
}

2548 2549 2550 2551 2552 2553 2554
/*
 * 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.
2555 2556
 *
 * Returns the bottommost base image if bs == NULL.
2557 2558 2559 2560
 */
BlockDriverState *bdrv_find_overlay(BlockDriverState *active,
                                    BlockDriverState *bs)
{
2561 2562
    while (active && bs != active->backing_hd) {
        active = active->backing_hd;
2563 2564
    }

2565 2566
    return active;
}
2567

2568 2569 2570 2571
/* Given a BDS, searches for the base layer. */
BlockDriverState *bdrv_find_base(BlockDriverState *bs)
{
    return bdrv_find_overlay(bs, NULL);
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
}

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
 *
2602 2603 2604
 * If backing_file_str is non-NULL, it will be used when modifying top's
 * overlay image metadata.
 *
2605 2606 2607 2608 2609
 * Error conditions:
 *  if active == top, that is considered an error
 *
 */
int bdrv_drop_intermediate(BlockDriverState *active, BlockDriverState *top,
2610
                           BlockDriverState *base, const char *backing_file_str)
2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644
{
    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) {
2645
        intermediate_state = g_new0(BlkIntermediateStates, 1);
2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661
        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 */
2662 2663
    backing_file_str = backing_file_str ? backing_file_str : base_bs->filename;
    ret = bdrv_change_backing_file(new_top_bs, backing_file_str,
2664 2665 2666 2667
                                   base_bs->drv ? base_bs->drv->format_name : "");
    if (ret) {
        goto exit;
    }
2668
    bdrv_set_backing_hd(new_top_bs, base_bs);
2669 2670 2671

    QSIMPLEQ_FOREACH_SAFE(intermediate_state, &states_to_delete, entry, next) {
        /* so that bdrv_close() does not recursively close the chain */
2672
        bdrv_set_backing_hd(intermediate_state->bs, NULL);
F
Fam Zheng 已提交
2673
        bdrv_unref(intermediate_state->bs);
2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684
    }
    ret = 0;

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


2685 2686 2687
static int bdrv_check_byte_request(BlockDriverState *bs, int64_t offset,
                                   size_t size)
{
2688
    if (size > BDRV_REQUEST_MAX_SECTORS << BDRV_SECTOR_BITS) {
2689 2690 2691
        return -EIO;
    }

M
Max Reitz 已提交
2692
    if (!bdrv_is_inserted(bs)) {
2693
        return -ENOMEDIUM;
M
Max Reitz 已提交
2694
    }
2695

M
Max Reitz 已提交
2696
    if (offset < 0) {
2697
        return -EIO;
M
Max Reitz 已提交
2698
    }
2699 2700 2701 2702 2703 2704 2705

    return 0;
}

static int bdrv_check_request(BlockDriverState *bs, int64_t sector_num,
                              int nb_sectors)
{
2706
    if (nb_sectors < 0 || nb_sectors > BDRV_REQUEST_MAX_SECTORS) {
2707 2708 2709
        return -EIO;
    }

2710 2711
    return bdrv_check_byte_request(bs, sector_num * BDRV_SECTOR_SIZE,
                                   nb_sectors * BDRV_SECTOR_SIZE);
2712 2713
}

2714 2715
typedef struct RwCo {
    BlockDriverState *bs;
2716
    int64_t offset;
2717 2718 2719
    QEMUIOVector *qiov;
    bool is_write;
    int ret;
P
Peter Lieven 已提交
2720
    BdrvRequestFlags flags;
2721 2722 2723
} RwCo;

static void coroutine_fn bdrv_rw_co_entry(void *opaque)
B
bellard 已提交
2724
{
2725
    RwCo *rwco = opaque;
B
bellard 已提交
2726

2727
    if (!rwco->is_write) {
2728 2729
        rwco->ret = bdrv_co_do_preadv(rwco->bs, rwco->offset,
                                      rwco->qiov->size, rwco->qiov,
P
Peter Lieven 已提交
2730
                                      rwco->flags);
2731 2732 2733 2734
    } else {
        rwco->ret = bdrv_co_do_pwritev(rwco->bs, rwco->offset,
                                       rwco->qiov->size, rwco->qiov,
                                       rwco->flags);
2735 2736
    }
}
2737

2738
/*
2739
 * Process a vectored synchronous request using coroutines
2740
 */
2741 2742 2743
static int bdrv_prwv_co(BlockDriverState *bs, int64_t offset,
                        QEMUIOVector *qiov, bool is_write,
                        BdrvRequestFlags flags)
2744 2745 2746 2747
{
    Coroutine *co;
    RwCo rwco = {
        .bs = bs,
2748
        .offset = offset,
2749
        .qiov = qiov,
2750 2751
        .is_write = is_write,
        .ret = NOT_DONE,
P
Peter Lieven 已提交
2752
        .flags = flags,
2753
    };
2754

2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765
    /**
     * 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);
    }

2766 2767 2768 2769
    if (qemu_in_coroutine()) {
        /* Fast-path if already in coroutine context */
        bdrv_rw_co_entry(&rwco);
    } else {
2770 2771
        AioContext *aio_context = bdrv_get_aio_context(bs);

2772 2773 2774
        co = qemu_coroutine_create(bdrv_rw_co_entry);
        qemu_coroutine_enter(co, &rwco);
        while (rwco.ret == NOT_DONE) {
2775
            aio_poll(aio_context, true);
2776 2777 2778 2779
        }
    }
    return rwco.ret;
}
B
bellard 已提交
2780

2781 2782 2783 2784
/*
 * 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
2785
                      int nb_sectors, bool is_write, BdrvRequestFlags flags)
2786 2787 2788 2789 2790 2791 2792
{
    QEMUIOVector qiov;
    struct iovec iov = {
        .iov_base = (void *)buf,
        .iov_len = nb_sectors * BDRV_SECTOR_SIZE,
    };

2793
    if (nb_sectors < 0 || nb_sectors > BDRV_REQUEST_MAX_SECTORS) {
2794 2795 2796
        return -EINVAL;
    }

2797
    qemu_iovec_init_external(&qiov, &iov, 1);
2798 2799
    return bdrv_prwv_co(bs, sector_num << BDRV_SECTOR_BITS,
                        &qiov, is_write, flags);
2800 2801
}

2802 2803 2804 2805
/* 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
2806
    return bdrv_rw_co(bs, sector_num, buf, nb_sectors, false, 0);
B
bellard 已提交
2807 2808
}

2809 2810 2811 2812 2813 2814 2815 2816 2817
/* 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;
2818
    ret = bdrv_read(bs, sector_num, buf, nb_sectors);
2819 2820 2821 2822
    bs->io_limits_enabled = enabled;
    return ret;
}

2823
/* Return < 0 if error. Important errors are:
B
bellard 已提交
2824 2825 2826 2827 2828
  -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
*/
2829
int bdrv_write(BlockDriverState *bs, int64_t sector_num,
B
bellard 已提交
2830 2831
               const uint8_t *buf, int nb_sectors)
{
P
Peter Lieven 已提交
2832
    return bdrv_rw_co(bs, sector_num, (uint8_t *)buf, nb_sectors, true, 0);
B
bellard 已提交
2833 2834
}

2835 2836
int bdrv_write_zeroes(BlockDriverState *bs, int64_t sector_num,
                      int nb_sectors, BdrvRequestFlags flags)
P
Peter Lieven 已提交
2837 2838
{
    return bdrv_rw_co(bs, sector_num, NULL, nb_sectors, true,
2839
                      BDRV_REQ_ZERO_WRITE | flags);
2840 2841
}

P
Peter Lieven 已提交
2842 2843 2844 2845 2846 2847 2848 2849 2850 2851
/*
 * 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)
{
2852
    int64_t target_sectors, ret, nb_sectors, sector_num = 0;
P
Peter Lieven 已提交
2853 2854
    int n;

2855 2856 2857
    target_sectors = bdrv_nb_sectors(bs);
    if (target_sectors < 0) {
        return target_sectors;
2858 2859
    }

P
Peter Lieven 已提交
2860
    for (;;) {
2861
        nb_sectors = MIN(target_sectors - sector_num, BDRV_REQUEST_MAX_SECTORS);
P
Peter Lieven 已提交
2862 2863 2864 2865
        if (nb_sectors <= 0) {
            return 0;
        }
        ret = bdrv_get_block_status(bs, sector_num, nb_sectors, &n);
2866 2867 2868 2869 2870
        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
2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884
        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;
    }
}

2885
int bdrv_pread(BlockDriverState *bs, int64_t offset, void *buf, int bytes)
B
bellard 已提交
2886
{
2887 2888 2889 2890 2891
    QEMUIOVector qiov;
    struct iovec iov = {
        .iov_base = (void *)buf,
        .iov_len = bytes,
    };
2892
    int ret;
B
bellard 已提交
2893

2894 2895
    if (bytes < 0) {
        return -EINVAL;
B
bellard 已提交
2896 2897
    }

2898 2899 2900 2901
    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
2902
    }
2903 2904

    return bytes;
B
bellard 已提交
2905 2906
}

2907
int bdrv_pwritev(BlockDriverState *bs, int64_t offset, QEMUIOVector *qiov)
B
bellard 已提交
2908
{
2909
    int ret;
B
bellard 已提交
2910

2911 2912 2913
    ret = bdrv_prwv_co(bs, offset, qiov, true, 0);
    if (ret < 0) {
        return ret;
B
bellard 已提交
2914 2915
    }

2916 2917 2918 2919
    return qiov->size;
}

int bdrv_pwrite(BlockDriverState *bs, int64_t offset,
2920
                const void *buf, int bytes)
2921 2922 2923 2924
{
    QEMUIOVector qiov;
    struct iovec iov = {
        .iov_base   = (void *) buf,
2925
        .iov_len    = bytes,
2926 2927
    };

2928 2929 2930 2931
    if (bytes < 0) {
        return -EINVAL;
    }

2932 2933
    qemu_iovec_init_external(&qiov, &iov, 1);
    return bdrv_pwritev(bs, offset, &qiov);
B
bellard 已提交
2934 2935
}

K
Kevin Wolf 已提交
2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951
/*
 * 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;
    }

2952 2953
    /* No flush needed for cache modes that already do it */
    if (bs->enable_write_cache) {
K
Kevin Wolf 已提交
2954 2955 2956 2957 2958 2959
        bdrv_flush(bs);
    }

    return 0;
}

2960
static int coroutine_fn bdrv_co_do_copy_on_readv(BlockDriverState *bs,
S
Stefan Hajnoczi 已提交
2961 2962 2963 2964 2965 2966 2967 2968 2969
        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;

2970
    BlockDriver *drv = bs->drv;
S
Stefan Hajnoczi 已提交
2971 2972 2973 2974 2975 2976 2977 2978 2979 2980
    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.
     */
2981 2982
    bdrv_round_to_clusters(bs, sector_num, nb_sectors,
                           &cluster_sector_num, &cluster_nb_sectors);
S
Stefan Hajnoczi 已提交
2983

2984 2985
    trace_bdrv_co_do_copy_on_readv(bs, sector_num, nb_sectors,
                                   cluster_sector_num, cluster_nb_sectors);
S
Stefan Hajnoczi 已提交
2986 2987

    iov.iov_len = cluster_nb_sectors * BDRV_SECTOR_SIZE;
2988 2989 2990 2991 2992 2993
    iov.iov_base = bounce_buffer = qemu_try_blockalign(bs, iov.iov_len);
    if (bounce_buffer == NULL) {
        ret = -ENOMEM;
        goto err;
    }

S
Stefan Hajnoczi 已提交
2994 2995
    qemu_iovec_init_external(&bounce_qiov, &iov, 1);

2996 2997
    ret = drv->bdrv_co_readv(bs, cluster_sector_num, cluster_nb_sectors,
                             &bounce_qiov);
S
Stefan Hajnoczi 已提交
2998 2999 3000 3001
    if (ret < 0) {
        goto err;
    }

3002 3003
    if (drv->bdrv_co_write_zeroes &&
        buffer_is_zero(bounce_buffer, iov.iov_len)) {
K
Kevin Wolf 已提交
3004
        ret = bdrv_co_do_write_zeroes(bs, cluster_sector_num,
3005
                                      cluster_nb_sectors, 0);
3006
    } else {
3007 3008 3009
        /* This does not change the data on the disk, it is not necessary
         * to flush even in cache=writethrough mode.
         */
3010
        ret = drv->bdrv_co_writev(bs, cluster_sector_num, cluster_nb_sectors,
S
Stefan Hajnoczi 已提交
3011
                                  &bounce_qiov);
3012 3013
    }

S
Stefan Hajnoczi 已提交
3014 3015 3016 3017 3018 3019 3020 3021 3022
    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;
3023 3024
    qemu_iovec_from_buf(qiov, 0, bounce_buffer + skip_bytes,
                        nb_sectors * BDRV_SECTOR_SIZE);
S
Stefan Hajnoczi 已提交
3025 3026 3027 3028 3029 3030

err:
    qemu_vfree(bounce_buffer);
    return ret;
}

3031
/*
3032 3033 3034
 * 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.
3035
 */
3036
static int coroutine_fn bdrv_aligned_preadv(BlockDriverState *bs,
3037
    BdrvTrackedRequest *req, int64_t offset, unsigned int bytes,
3038
    int64_t align, QEMUIOVector *qiov, int flags)
K
Kevin Wolf 已提交
3039 3040
{
    BlockDriver *drv = bs->drv;
S
Stefan Hajnoczi 已提交
3041
    int ret;
K
Kevin Wolf 已提交
3042

3043 3044
    int64_t sector_num = offset >> BDRV_SECTOR_BITS;
    unsigned int nb_sectors = bytes >> BDRV_SECTOR_BITS;
K
Kevin Wolf 已提交
3045

3046 3047
    assert((offset & (BDRV_SECTOR_SIZE - 1)) == 0);
    assert((bytes & (BDRV_SECTOR_SIZE - 1)) == 0);
3048
    assert(!qiov || bytes == qiov->size);
3049 3050

    /* Handle Copy on Read and associated serialisation */
3051
    if (flags & BDRV_REQ_COPY_ON_READ) {
3052 3053 3054 3055 3056 3057
        /* 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));
3058 3059
    }

3060
    wait_serialising_requests(req);
3061

3062
    if (flags & BDRV_REQ_COPY_ON_READ) {
S
Stefan Hajnoczi 已提交
3063 3064
        int pnum;

3065
        ret = bdrv_is_allocated(bs, sector_num, nb_sectors, &pnum);
S
Stefan Hajnoczi 已提交
3066 3067 3068 3069 3070
        if (ret < 0) {
            goto out;
        }

        if (!ret || pnum != nb_sectors) {
3071
            ret = bdrv_co_do_copy_on_readv(bs, sector_num, nb_sectors, qiov);
S
Stefan Hajnoczi 已提交
3072 3073 3074 3075
            goto out;
        }
    }

3076
    /* Forward the request to the BlockDriver */
M
Max Reitz 已提交
3077
    if (!bs->zero_beyond_eof) {
3078 3079
        ret = drv->bdrv_co_readv(bs, sector_num, nb_sectors, qiov);
    } else {
M
Max Reitz 已提交
3080
        /* Read zeros after EOF */
3081
        int64_t total_sectors, max_nb_sectors;
3082

3083 3084 3085
        total_sectors = bdrv_nb_sectors(bs);
        if (total_sectors < 0) {
            ret = total_sectors;
3086 3087 3088
            goto out;
        }

3089 3090
        max_nb_sectors = ROUND_UP(MAX(0, total_sectors - sector_num),
                                  align >> BDRV_SECTOR_BITS);
3091 3092 3093
        if (nb_sectors < max_nb_sectors) {
            ret = drv->bdrv_co_readv(bs, sector_num, nb_sectors, qiov);
        } else if (max_nb_sectors > 0) {
3094 3095 3096 3097
            QEMUIOVector local_qiov;

            qemu_iovec_init(&local_qiov, qiov->niov);
            qemu_iovec_concat(&local_qiov, qiov, 0,
3098
                              max_nb_sectors * BDRV_SECTOR_SIZE);
3099

3100
            ret = drv->bdrv_co_readv(bs, sector_num, max_nb_sectors,
3101 3102 3103
                                     &local_qiov);

            qemu_iovec_destroy(&local_qiov);
3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115
        } 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
3116 3117

out:
S
Stefan Hajnoczi 已提交
3118
    return ret;
K
Kevin Wolf 已提交
3119 3120
}

F
Fam Zheng 已提交
3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133
static inline uint64_t bdrv_get_align(BlockDriverState *bs)
{
    /* TODO Lift BDRV_SECTOR_SIZE restriction in BlockDriver interface */
    return MAX(BDRV_SECTOR_SIZE, bs->request_alignment);
}

static inline bool bdrv_req_is_aligned(BlockDriverState *bs,
                                       int64_t offset, size_t bytes)
{
    int64_t align = bdrv_get_align(bs);
    return !(offset & (align - 1) || (bytes & (align - 1)));
}

3134 3135 3136
/*
 * Handle a read request in coroutine context
 */
3137 3138
static int coroutine_fn bdrv_co_do_preadv(BlockDriverState *bs,
    int64_t offset, unsigned int bytes, QEMUIOVector *qiov,
3139 3140 3141
    BdrvRequestFlags flags)
{
    BlockDriver *drv = bs->drv;
3142 3143
    BdrvTrackedRequest req;

F
Fam Zheng 已提交
3144
    uint64_t align = bdrv_get_align(bs);
3145 3146 3147 3148
    uint8_t *head_buf = NULL;
    uint8_t *tail_buf = NULL;
    QEMUIOVector local_qiov;
    bool use_local_qiov = false;
3149 3150 3151 3152 3153
    int ret;

    if (!drv) {
        return -ENOMEDIUM;
    }
3154 3155 3156 3157

    ret = bdrv_check_byte_request(bs, offset, bytes);
    if (ret < 0) {
        return ret;
3158 3159 3160 3161 3162 3163 3164 3165
    }

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

    /* throttling disk I/O */
    if (bs->io_limits_enabled) {
3166
        bdrv_io_limits_intercept(bs, bytes, false);
3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193
    }

    /* 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);
    }

3194
    tracked_request_begin(&req, bs, offset, bytes, false);
3195
    ret = bdrv_aligned_preadv(bs, &req, offset, bytes, align,
3196 3197
                              use_local_qiov ? &local_qiov : qiov,
                              flags);
3198
    tracked_request_end(&req);
3199 3200 3201 3202 3203

    if (use_local_qiov) {
        qemu_iovec_destroy(&local_qiov);
        qemu_vfree(head_buf);
        qemu_vfree(tail_buf);
3204 3205 3206 3207 3208
    }

    return ret;
}

3209 3210 3211 3212
static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags)
{
3213
    if (nb_sectors < 0 || nb_sectors > BDRV_REQUEST_MAX_SECTORS) {
3214 3215 3216 3217 3218 3219 3220
        return -EINVAL;
    }

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

3221
int coroutine_fn bdrv_co_readv(BlockDriverState *bs, int64_t sector_num,
K
Kevin Wolf 已提交
3222 3223
    int nb_sectors, QEMUIOVector *qiov)
{
3224
    trace_bdrv_co_readv(bs, sector_num, nb_sectors);
K
Kevin Wolf 已提交
3225

3226 3227 3228 3229 3230 3231 3232 3233 3234 3235
    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);
3236 3237
}

3238
#define MAX_WRITE_ZEROES_BOUNCE_BUFFER 32768
3239

3240
static int coroutine_fn bdrv_co_do_write_zeroes(BlockDriverState *bs,
3241
    int64_t sector_num, int nb_sectors, BdrvRequestFlags flags)
3242 3243 3244
{
    BlockDriver *drv = bs->drv;
    QEMUIOVector qiov;
3245 3246
    struct iovec iov = {0};
    int ret = 0;
3247

3248 3249
    int max_write_zeroes = MIN_NON_ZERO(bs->bl.max_write_zeroes,
                                        BDRV_REQUEST_MAX_SECTORS);
K
Kevin Wolf 已提交
3250

3251 3252 3253
    while (nb_sectors > 0 && !ret) {
        int num = nb_sectors;

3254 3255 3256 3257 3258 3259 3260
        /* 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.  */
3261
                num = bs->bl.write_zeroes_alignment;
3262 3263 3264 3265 3266 3267
                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;
3268
            }
K
Kevin Wolf 已提交
3269
        }
3270

3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283
        /* 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 */
3284
            int max_xfer_len = MIN_NON_ZERO(bs->bl.max_transfer_length,
3285
                                            MAX_WRITE_ZEROES_BOUNCE_BUFFER);
3286
            num = MIN(num, max_xfer_len);
3287 3288
            iov.iov_len = num * BDRV_SECTOR_SIZE;
            if (iov.iov_base == NULL) {
3289 3290 3291 3292 3293
                iov.iov_base = qemu_try_blockalign(bs, num * BDRV_SECTOR_SIZE);
                if (iov.iov_base == NULL) {
                    ret = -ENOMEM;
                    goto fail;
                }
3294
                memset(iov.iov_base, 0, num * BDRV_SECTOR_SIZE);
3295 3296
            }
            qemu_iovec_init_external(&qiov, &iov, 1);
3297

3298
            ret = drv->bdrv_co_writev(bs, sector_num, num, &qiov);
3299 3300 3301 3302

            /* Keep bounce buffer around if it is big enough for all
             * all future requests.
             */
3303
            if (num < max_xfer_len) {
3304 3305 3306
                qemu_vfree(iov.iov_base);
                iov.iov_base = NULL;
            }
3307 3308 3309 3310 3311
        }

        sector_num += num;
        nb_sectors -= num;
    }
3312

3313
fail:
3314 3315 3316 3317
    qemu_vfree(iov.iov_base);
    return ret;
}

3318
/*
3319
 * Forwards an already correctly aligned write request to the BlockDriver.
3320
 */
3321
static int coroutine_fn bdrv_aligned_pwritev(BlockDriverState *bs,
3322 3323
    BdrvTrackedRequest *req, int64_t offset, unsigned int bytes,
    QEMUIOVector *qiov, int flags)
3324 3325
{
    BlockDriver *drv = bs->drv;
3326
    bool waited;
3327
    int ret;
K
Kevin Wolf 已提交
3328

3329 3330
    int64_t sector_num = offset >> BDRV_SECTOR_BITS;
    unsigned int nb_sectors = bytes >> BDRV_SECTOR_BITS;
3331

3332 3333
    assert((offset & (BDRV_SECTOR_SIZE - 1)) == 0);
    assert((bytes & (BDRV_SECTOR_SIZE - 1)) == 0);
3334
    assert(!qiov || bytes == qiov->size);
3335

3336 3337
    waited = wait_serialising_requests(req);
    assert(!waited || !req->serialising);
3338 3339
    assert(req->overlap_offset <= offset);
    assert(offset + bytes <= req->overlap_offset + req->overlap_bytes);
3340

3341
    ret = notifier_with_return_list_notify(&bs->before_write_notifiers, req);
3342

3343 3344 3345 3346 3347 3348 3349 3350 3351
    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;
        }
    }

3352 3353 3354
    if (ret < 0) {
        /* Do nothing, write notifier decided to fail this request */
    } else if (flags & BDRV_REQ_ZERO_WRITE) {
3355
        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_ZERO);
3356
        ret = bdrv_co_do_write_zeroes(bs, sector_num, nb_sectors, flags);
3357
    } else {
3358
        BLKDBG_EVENT(bs, BLKDBG_PWRITEV);
3359 3360
        ret = drv->bdrv_co_writev(bs, sector_num, nb_sectors, qiov);
    }
3361
    BLKDBG_EVENT(bs, BLKDBG_PWRITEV_DONE);
3362

3363 3364 3365 3366
    if (ret == 0 && !bs->enable_write_cache) {
        ret = bdrv_co_flush(bs);
    }

F
Fam Zheng 已提交
3367
    bdrv_set_dirty(bs, sector_num, nb_sectors);
K
Kevin Wolf 已提交
3368

3369
    block_acct_highest_sector(&bs->stats, sector_num, nb_sectors);
3370

M
Max Reitz 已提交
3371
    if (ret >= 0) {
3372 3373
        bs->total_sectors = MAX(bs->total_sectors, sector_num + nb_sectors);
    }
K
Kevin Wolf 已提交
3374

3375
    return ret;
K
Kevin Wolf 已提交
3376 3377
}

3378 3379 3380
/*
 * Handle a write request in coroutine context
 */
3381 3382
static int coroutine_fn bdrv_co_do_pwritev(BlockDriverState *bs,
    int64_t offset, unsigned int bytes, QEMUIOVector *qiov,
3383 3384
    BdrvRequestFlags flags)
{
3385
    BdrvTrackedRequest req;
F
Fam Zheng 已提交
3386
    uint64_t align = bdrv_get_align(bs);
3387 3388 3389 3390
    uint8_t *head_buf = NULL;
    uint8_t *tail_buf = NULL;
    QEMUIOVector local_qiov;
    bool use_local_qiov = false;
3391 3392 3393 3394 3395 3396 3397 3398
    int ret;

    if (!bs->drv) {
        return -ENOMEDIUM;
    }
    if (bs->read_only) {
        return -EACCES;
    }
3399 3400 3401 3402

    ret = bdrv_check_byte_request(bs, offset, bytes);
    if (ret < 0) {
        return ret;
3403 3404 3405 3406
    }

    /* throttling disk I/O */
    if (bs->io_limits_enabled) {
3407
        bdrv_io_limits_intercept(bs, bytes, true);
3408 3409
    }

3410 3411 3412 3413 3414
    /*
     * 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.
     */
3415
    tracked_request_begin(&req, bs, offset, bytes, true);
3416 3417 3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430

    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);

3431
        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_HEAD);
3432 3433 3434 3435 3436
        ret = bdrv_aligned_preadv(bs, &req, offset & ~(align - 1), align,
                                  align, &head_qiov, 0);
        if (ret < 0) {
            goto fail;
        }
3437
        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_AFTER_HEAD);
3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451

        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;
3452
        bool waited;
3453 3454

        mark_request_serialising(&req, align);
3455 3456
        waited = wait_serialising_requests(&req);
        assert(!waited || !use_local_qiov);
3457 3458 3459 3460 3461 3462 3463 3464

        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);

3465
        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_TAIL);
3466 3467 3468 3469 3470
        ret = bdrv_aligned_preadv(bs, &req, (offset + bytes) & ~(align - 1), align,
                                  align, &tail_qiov, 0);
        if (ret < 0) {
            goto fail;
        }
3471
        BLKDBG_EVENT(bs, BLKDBG_PWRITEV_RMW_AFTER_TAIL);
3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483 3484

        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);
    }

F
Fam Zheng 已提交
3485 3486 3487 3488
    if (use_local_qiov) {
        /* Local buffer may have non-zero data. */
        flags &= ~BDRV_REQ_ZERO_WRITE;
    }
3489 3490 3491 3492 3493
    ret = bdrv_aligned_pwritev(bs, &req, offset, bytes,
                               use_local_qiov ? &local_qiov : qiov,
                               flags);

fail:
3494
    tracked_request_end(&req);
3495

3496 3497 3498
    if (use_local_qiov) {
        qemu_iovec_destroy(&local_qiov);
    }
3499 3500
    qemu_vfree(head_buf);
    qemu_vfree(tail_buf);
3501

3502 3503 3504
    return ret;
}

3505 3506 3507 3508
static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs,
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags)
{
3509
    if (nb_sectors < 0 || nb_sectors > BDRV_REQUEST_MAX_SECTORS) {
3510 3511 3512 3513 3514 3515 3516
        return -EINVAL;
    }

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

3517 3518 3519 3520 3521
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);

3522 3523 3524 3525
    return bdrv_co_do_writev(bs, sector_num, nb_sectors, qiov, 0);
}

int coroutine_fn bdrv_co_write_zeroes(BlockDriverState *bs,
3526 3527
                                      int64_t sector_num, int nb_sectors,
                                      BdrvRequestFlags flags)
3528
{
F
Fam Zheng 已提交
3529 3530
    int ret;

3531
    trace_bdrv_co_write_zeroes(bs, sector_num, nb_sectors, flags);
3532

3533 3534 3535
    if (!(bs->open_flags & BDRV_O_UNMAP)) {
        flags &= ~BDRV_REQ_MAY_UNMAP;
    }
F
Fam Zheng 已提交
3536 3537 3538 3539 3540 3541 3542 3543 3544 3545 3546 3547 3548
    if (bdrv_req_is_aligned(bs, sector_num << BDRV_SECTOR_BITS,
                            nb_sectors << BDRV_SECTOR_BITS)) {
        ret = bdrv_co_do_writev(bs, sector_num, nb_sectors, NULL,
                                BDRV_REQ_ZERO_WRITE | flags);
    } else {
        uint8_t *buf;
        QEMUIOVector local_qiov;
        size_t bytes = nb_sectors << BDRV_SECTOR_BITS;

        buf = qemu_memalign(bdrv_opt_mem_align(bs), bytes);
        memset(buf, 0, bytes);
        qemu_iovec_init(&local_qiov, 1);
        qemu_iovec_add(&local_qiov, buf, bytes);
3549

F
Fam Zheng 已提交
3550 3551 3552 3553 3554
        ret = bdrv_co_do_writev(bs, sector_num, nb_sectors, &local_qiov,
                                BDRV_REQ_ZERO_WRITE | flags);
        qemu_vfree(buf);
    }
    return ret;
3555 3556
}

B
bellard 已提交
3557 3558 3559 3560 3561 3562
/**
 * Truncate file to 'offset' bytes (needed only for file protocols)
 */
int bdrv_truncate(BlockDriverState *bs, int64_t offset)
{
    BlockDriver *drv = bs->drv;
3563
    int ret;
B
bellard 已提交
3564
    if (!drv)
B
bellard 已提交
3565
        return -ENOMEDIUM;
B
bellard 已提交
3566 3567
    if (!drv->bdrv_truncate)
        return -ENOTSUP;
3568 3569
    if (bs->read_only)
        return -EACCES;
3570

3571 3572 3573
    ret = drv->bdrv_truncate(bs, offset);
    if (ret == 0) {
        ret = refresh_total_sectors(bs, offset >> BDRV_SECTOR_BITS);
3574 3575 3576
        if (bs->blk) {
            blk_dev_resize_cb(bs->blk);
        }
3577 3578
    }
    return ret;
B
bellard 已提交
3579 3580
}

3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599
/**
 * 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
3600
/**
M
Markus Armbruster 已提交
3601
 * Return number of sectors on success, -errno on error.
B
bellard 已提交
3602
 */
M
Markus Armbruster 已提交
3603
int64_t bdrv_nb_sectors(BlockDriverState *bs)
B
bellard 已提交
3604 3605
{
    BlockDriver *drv = bs->drv;
M
Markus Armbruster 已提交
3606

B
bellard 已提交
3607
    if (!drv)
B
bellard 已提交
3608
        return -ENOMEDIUM;
3609

3610 3611 3612 3613
    if (drv->has_variable_length) {
        int ret = refresh_total_sectors(bs, bs->total_sectors);
        if (ret < 0) {
            return ret;
3614
        }
B
bellard 已提交
3615
    }
M
Markus Armbruster 已提交
3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627
    return bs->total_sectors;
}

/**
 * Return length in bytes on success, -errno on error.
 * The length is always a multiple of BDRV_SECTOR_SIZE.
 */
int64_t bdrv_getlength(BlockDriverState *bs)
{
    int64_t ret = bdrv_nb_sectors(bs);

    return ret < 0 ? ret : ret * BDRV_SECTOR_SIZE;
B
bellard 已提交
3628 3629
}

B
bellard 已提交
3630
/* return 0 as number of sectors if no device present or error */
3631
void bdrv_get_geometry(BlockDriverState *bs, uint64_t *nb_sectors_ptr)
B
bellard 已提交
3632
{
M
Markus Armbruster 已提交
3633 3634 3635
    int64_t nb_sectors = bdrv_nb_sectors(bs);

    *nb_sectors_ptr = nb_sectors < 0 ? 0 : nb_sectors;
B
bellard 已提交
3636
}
B
bellard 已提交
3637

3638 3639
void bdrv_set_on_error(BlockDriverState *bs, BlockdevOnError on_read_error,
                       BlockdevOnError on_write_error)
3640 3641 3642 3643 3644
{
    bs->on_read_error = on_read_error;
    bs->on_write_error = on_write_error;
}

3645
BlockdevOnError bdrv_get_on_error(BlockDriverState *bs, bool is_read)
3646 3647 3648 3649
{
    return is_read ? bs->on_read_error : bs->on_write_error;
}

3650 3651 3652 3653 3654 3655
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
3656 3657
        return (error == ENOSPC) ?
               BLOCK_ERROR_ACTION_STOP : BLOCK_ERROR_ACTION_REPORT;
3658
    case BLOCKDEV_ON_ERROR_STOP:
W
Wenchao Xia 已提交
3659
        return BLOCK_ERROR_ACTION_STOP;
3660
    case BLOCKDEV_ON_ERROR_REPORT:
W
Wenchao Xia 已提交
3661
        return BLOCK_ERROR_ACTION_REPORT;
3662
    case BLOCKDEV_ON_ERROR_IGNORE:
W
Wenchao Xia 已提交
3663
        return BLOCK_ERROR_ACTION_IGNORE;
3664 3665 3666 3667 3668
    default:
        abort();
    }
}

3669 3670 3671 3672
static void send_qmp_error_event(BlockDriverState *bs,
                                 BlockErrorAction action,
                                 bool is_read, int error)
{
3673
    IoOperationType optype;
3674

3675 3676
    optype = is_read ? IO_OPERATION_TYPE_READ : IO_OPERATION_TYPE_WRITE;
    qapi_event_send_block_io_error(bdrv_get_device_name(bs), optype, action,
3677
                                   bdrv_iostatus_is_enabled(bs),
3678 3679
                                   error == ENOSPC, strerror(error),
                                   &error_abort);
3680 3681
}

3682 3683 3684 3685 3686 3687 3688 3689
/* 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);
3690

W
Wenchao Xia 已提交
3691
    if (action == BLOCK_ERROR_ACTION_STOP) {
3692 3693 3694 3695
        /* 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).
         */
3696
        bdrv_iostatus_set_err(bs, error);
3697 3698 3699 3700 3701 3702 3703 3704 3705 3706

        /* 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();
3707
        send_qmp_error_event(bs, action, is_read, error);
3708 3709
        qemu_system_vmstop_request(RUN_STATE_IO_ERROR);
    } else {
3710
        send_qmp_error_event(bs, action, is_read, error);
3711 3712 3713
    }
}

B
bellard 已提交
3714 3715 3716 3717 3718
int bdrv_is_read_only(BlockDriverState *bs)
{
    return bs->read_only;
}

3719 3720 3721 3722 3723
int bdrv_is_sg(BlockDriverState *bs)
{
    return bs->sg;
}

3724 3725 3726 3727 3728
int bdrv_enable_write_cache(BlockDriverState *bs)
{
    return bs->enable_write_cache;
}

3729 3730 3731
void bdrv_set_enable_write_cache(BlockDriverState *bs, bool wce)
{
    bs->enable_write_cache = wce;
3732 3733 3734 3735 3736 3737 3738

    /* so a reopen() will preserve wce */
    if (wce) {
        bs->open_flags |= BDRV_O_CACHE_WB;
    } else {
        bs->open_flags &= ~BDRV_O_CACHE_WB;
    }
3739 3740
}

B
bellard 已提交
3741 3742 3743 3744 3745 3746 3747
int bdrv_is_encrypted(BlockDriverState *bs)
{
    if (bs->backing_hd && bs->backing_hd->encrypted)
        return 1;
    return bs->encrypted;
}

3748 3749 3750 3751 3752 3753 3754 3755 3756
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
3757 3758 3759 3760 3761 3762 3763 3764 3765 3766
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;
    }
3767 3768 3769 3770 3771
    if (!bs->encrypted) {
        return -EINVAL;
    } else if (!bs->drv || !bs->drv->bdrv_set_key) {
        return -ENOMEDIUM;
    }
3772
    ret = bs->drv->bdrv_set_key(bs, key);
3773 3774 3775 3776
    if (ret < 0) {
        bs->valid_key = 0;
    } else if (!bs->valid_key) {
        bs->valid_key = 1;
3777 3778 3779 3780
        if (bs->blk) {
            /* call the change callback now, we skipped it on open */
            blk_dev_change_media_cb(bs->blk, true);
        }
3781
    }
3782
    return ret;
B
bellard 已提交
3783 3784
}

3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799
/*
 * Provide an encryption key for @bs.
 * If @key is non-null:
 *     If @bs is not encrypted, fail.
 *     Else if the key is invalid, fail.
 *     Else set @bs's key to @key, replacing the existing key, if any.
 * If @key is null:
 *     If @bs is encrypted and still lacks a key, fail.
 *     Else do nothing.
 * On failure, store an error object through @errp if non-null.
 */
void bdrv_add_key(BlockDriverState *bs, const char *key, Error **errp)
{
    if (key) {
        if (!bdrv_is_encrypted(bs)) {
3800
            error_setg(errp, "Device '%s' is not encrypted",
3801 3802 3803 3804 3805 3806
                      bdrv_get_device_name(bs));
        } else if (bdrv_set_key(bs, key) < 0) {
            error_set(errp, QERR_INVALID_PASSWORD);
        }
    } else {
        if (bdrv_key_required(bs)) {
3807 3808
            error_set(errp, ERROR_CLASS_DEVICE_ENCRYPTED,
                      "'%s' (%s) is encrypted",
3809 3810 3811 3812 3813 3814
                      bdrv_get_device_name(bs),
                      bdrv_get_encrypted_filename(bs));
        }
    }
}

3815
const char *bdrv_get_format_name(BlockDriverState *bs)
B
bellard 已提交
3816
{
3817
    return bs->drv ? bs->drv->format_name : NULL;
B
bellard 已提交
3818 3819
}

3820 3821 3822 3823 3824
static int qsort_strcmp(const void *a, const void *b)
{
    return strcmp(a, b);
}

3825
void bdrv_iterate_format(void (*it)(void *opaque, const char *name),
B
bellard 已提交
3826 3827 3828
                         void *opaque)
{
    BlockDriver *drv;
3829
    int count = 0;
3830
    int i;
3831
    const char **formats = NULL;
B
bellard 已提交
3832

3833
    QLIST_FOREACH(drv, &bdrv_drivers, list) {
3834 3835 3836 3837 3838 3839 3840 3841
        if (drv->format_name) {
            bool found = false;
            int i = count;
            while (formats && i && !found) {
                found = !strcmp(formats[--i], drv->format_name);
            }

            if (!found) {
3842
                formats = g_renew(const char *, formats, count + 1);
3843 3844 3845
                formats[count++] = drv->format_name;
            }
        }
B
bellard 已提交
3846
    }
3847 3848 3849 3850 3851 3852 3853

    qsort(formats, count, sizeof(formats[0]), qsort_strcmp);

    for (i = 0; i < count; i++) {
        it(opaque, formats[i]);
    }

3854
    g_free(formats);
B
bellard 已提交
3855 3856
}

3857 3858 3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870 3871
/* 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;
}

3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888
/* 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;
}

3889 3890 3891 3892
BlockDriverState *bdrv_lookup_bs(const char *device,
                                 const char *node_name,
                                 Error **errp)
{
3893 3894
    BlockBackend *blk;
    BlockDriverState *bs;
3895 3896

    if (device) {
3897
        blk = blk_by_name(device);
3898

3899 3900
        if (blk) {
            return blk_bs(blk);
3901 3902 3903
        }
    }

3904 3905
    if (node_name) {
        bs = bdrv_find_node(node_name);
3906

3907 3908 3909
        if (bs) {
            return bs;
        }
3910 3911
    }

3912 3913 3914 3915
    error_setg(errp, "Cannot find device=%s nor node_name=%s",
                     device ? device : "",
                     node_name ? node_name : "");
    return NULL;
3916 3917
}

3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928
/* If 'base' is in the same chain as 'top', return true. Otherwise,
 * return false.  If either argument is NULL, return false. */
bool bdrv_chain_contains(BlockDriverState *top, BlockDriverState *base)
{
    while (top && top != base) {
        top = top->backing_hd;
    }

    return top != NULL;
}

F
Fam Zheng 已提交
3929 3930 3931 3932 3933 3934 3935 3936
BlockDriverState *bdrv_next_node(BlockDriverState *bs)
{
    if (!bs) {
        return QTAILQ_FIRST(&graph_bdrv_states);
    }
    return QTAILQ_NEXT(bs, node_list);
}

M
Markus Armbruster 已提交
3937 3938 3939 3940 3941
BlockDriverState *bdrv_next(BlockDriverState *bs)
{
    if (!bs) {
        return QTAILQ_FIRST(&bdrv_states);
    }
3942
    return QTAILQ_NEXT(bs, device_list);
M
Markus Armbruster 已提交
3943 3944
}

F
Fam Zheng 已提交
3945 3946 3947 3948 3949
const char *bdrv_get_node_name(const BlockDriverState *bs)
{
    return bs->node_name;
}

3950
/* TODO check what callers really want: bs->node_name or blk_name() */
3951
const char *bdrv_get_device_name(const BlockDriverState *bs)
B
bellard 已提交
3952
{
3953
    return bs->blk ? blk_name(bs->blk) : "";
B
bellard 已提交
3954 3955
}

M
Markus Armbruster 已提交
3956 3957 3958 3959 3960
int bdrv_get_flags(BlockDriverState *bs)
{
    return bs->open_flags;
}

3961
int bdrv_flush_all(void)
A
aliguori 已提交
3962 3963
{
    BlockDriverState *bs;
3964
    int result = 0;
A
aliguori 已提交
3965

3966
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
3967 3968 3969 3970 3971
        AioContext *aio_context = bdrv_get_aio_context(bs);
        int ret;

        aio_context_acquire(aio_context);
        ret = bdrv_flush(bs);
3972 3973 3974
        if (ret < 0 && !result) {
            result = ret;
        }
3975
        aio_context_release(aio_context);
3976
    }
3977 3978

    return result;
A
aliguori 已提交
3979 3980
}

3981 3982 3983 3984 3985
int bdrv_has_zero_init_1(BlockDriverState *bs)
{
    return 1;
}

K
Kevin Wolf 已提交
3986 3987 3988 3989
int bdrv_has_zero_init(BlockDriverState *bs)
{
    assert(bs->drv);

3990 3991 3992 3993 3994
    /* 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
3995 3996
    if (bs->drv->bdrv_has_zero_init) {
        return bs->drv->bdrv_has_zero_init(bs);
K
Kevin Wolf 已提交
3997 3998
    }

3999 4000
    /* safe default */
    return 0;
K
Kevin Wolf 已提交
4001 4002
}

4003 4004 4005 4006 4007 4008 4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020 4021 4022 4023 4024 4025 4026 4027 4028 4029 4030 4031 4032
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;
}

4033
typedef struct BdrvCoGetBlockStatusData {
4034
    BlockDriverState *bs;
4035
    BlockDriverState *base;
4036 4037 4038
    int64_t sector_num;
    int nb_sectors;
    int *pnum;
4039
    int64_t ret;
4040
    bool done;
4041
} BdrvCoGetBlockStatusData;
4042

4043
/*
4044 4045 4046
 * Returns the allocation status of the specified sectors.
 * Drivers not implementing the functionality are assumed to not support
 * backing files, hence all their sectors are reported as allocated.
4047
 *
4048 4049 4050
 * If 'sector_num' is beyond the end of the disk image the return value is 0
 * and 'pnum' is set to 0.
 *
4051 4052 4053 4054
 * '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.
 *
4055 4056
 * '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.
4057
 */
4058 4059 4060
static int64_t coroutine_fn bdrv_co_get_block_status(BlockDriverState *bs,
                                                     int64_t sector_num,
                                                     int nb_sectors, int *pnum)
4061
{
4062
    int64_t total_sectors;
4063
    int64_t n;
4064
    int64_t ret, ret2;
4065

4066 4067 4068
    total_sectors = bdrv_nb_sectors(bs);
    if (total_sectors < 0) {
        return total_sectors;
4069 4070
    }

4071
    if (sector_num >= total_sectors) {
4072 4073 4074 4075
        *pnum = 0;
        return 0;
    }

4076
    n = total_sectors - sector_num;
4077 4078 4079 4080
    if (n < nb_sectors) {
        nb_sectors = n;
    }

4081
    if (!bs->drv->bdrv_co_get_block_status) {
4082
        *pnum = nb_sectors;
4083
        ret = BDRV_BLOCK_DATA | BDRV_BLOCK_ALLOCATED;
4084 4085 4086 4087
        if (bs->drv->protocol_name) {
            ret |= BDRV_BLOCK_OFFSET_VALID | (sector_num * BDRV_SECTOR_SIZE);
        }
        return ret;
4088
    }
4089

4090 4091
    ret = bs->drv->bdrv_co_get_block_status(bs, sector_num, nb_sectors, pnum);
    if (ret < 0) {
4092
        *pnum = 0;
4093 4094 4095
        return ret;
    }

4096 4097 4098 4099 4100 4101
    if (ret & BDRV_BLOCK_RAW) {
        assert(ret & BDRV_BLOCK_OFFSET_VALID);
        return bdrv_get_block_status(bs->file, ret >> BDRV_SECTOR_BITS,
                                     *pnum, pnum);
    }

4102 4103 4104 4105
    if (ret & (BDRV_BLOCK_DATA | BDRV_BLOCK_ZERO)) {
        ret |= BDRV_BLOCK_ALLOCATED;
    }

4106 4107
    if (!(ret & BDRV_BLOCK_DATA) && !(ret & BDRV_BLOCK_ZERO)) {
        if (bdrv_unallocated_blocks_are_zero(bs)) {
4108
            ret |= BDRV_BLOCK_ZERO;
4109
        } else if (bs->backing_hd) {
4110
            BlockDriverState *bs2 = bs->backing_hd;
4111 4112
            int64_t nb_sectors2 = bdrv_nb_sectors(bs2);
            if (nb_sectors2 >= 0 && sector_num >= nb_sectors2) {
4113 4114 4115
                ret |= BDRV_BLOCK_ZERO;
            }
        }
4116
    }
4117 4118 4119 4120

    if (bs->file &&
        (ret & BDRV_BLOCK_DATA) && !(ret & BDRV_BLOCK_ZERO) &&
        (ret & BDRV_BLOCK_OFFSET_VALID)) {
M
Max Reitz 已提交
4121 4122
        int file_pnum;

4123
        ret2 = bdrv_co_get_block_status(bs->file, ret >> BDRV_SECTOR_BITS,
M
Max Reitz 已提交
4124
                                        *pnum, &file_pnum);
4125 4126 4127 4128
        if (ret2 >= 0) {
            /* Ignore errors.  This is just providing extra information, it
             * is useful but not necessary.
             */
M
Max Reitz 已提交
4129 4130 4131 4132 4133 4134 4135 4136 4137 4138
            if (!file_pnum) {
                /* !file_pnum indicates an offset at or beyond the EOF; it is
                 * perfectly valid for the format block driver to point to such
                 * offsets, so catch it and mark everything as zero */
                ret |= BDRV_BLOCK_ZERO;
            } else {
                /* Limit request to the range reported by the protocol driver */
                *pnum = file_pnum;
                ret |= (ret2 & BDRV_BLOCK_ZERO);
            }
4139 4140 4141
        }
    }

4142
    return ret;
4143 4144
}

4145 4146
/* Coroutine wrapper for bdrv_get_block_status() */
static void coroutine_fn bdrv_get_block_status_co_entry(void *opaque)
4147
{
4148
    BdrvCoGetBlockStatusData *data = opaque;
4149 4150
    BlockDriverState *bs = data->bs;

4151 4152
    data->ret = bdrv_co_get_block_status(bs, data->sector_num, data->nb_sectors,
                                         data->pnum);
4153 4154 4155 4156
    data->done = true;
}

/*
4157
 * Synchronous wrapper around bdrv_co_get_block_status().
4158
 *
4159
 * See bdrv_co_get_block_status() for details.
4160
 */
4161 4162
int64_t bdrv_get_block_status(BlockDriverState *bs, int64_t sector_num,
                              int nb_sectors, int *pnum)
4163
{
4164
    Coroutine *co;
4165
    BdrvCoGetBlockStatusData data = {
4166 4167 4168 4169 4170 4171 4172
        .bs = bs,
        .sector_num = sector_num,
        .nb_sectors = nb_sectors,
        .pnum = pnum,
        .done = false,
    };

4173 4174
    if (qemu_in_coroutine()) {
        /* Fast-path if already in coroutine context */
4175
        bdrv_get_block_status_co_entry(&data);
4176
    } else {
4177 4178
        AioContext *aio_context = bdrv_get_aio_context(bs);

4179
        co = qemu_coroutine_create(bdrv_get_block_status_co_entry);
4180 4181
        qemu_coroutine_enter(co, &data);
        while (!data.done) {
4182
            aio_poll(aio_context, true);
4183
        }
4184 4185
    }
    return data.ret;
4186 4187
}

4188 4189 4190
int coroutine_fn bdrv_is_allocated(BlockDriverState *bs, int64_t sector_num,
                                   int nb_sectors, int *pnum)
{
4191 4192 4193 4194
    int64_t ret = bdrv_get_block_status(bs, sector_num, nb_sectors, pnum);
    if (ret < 0) {
        return ret;
    }
4195
    return !!(ret & BDRV_BLOCK_ALLOCATED);
4196 4197
}

4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209
/*
 * 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.
 *
 */
4210 4211 4212 4213
int bdrv_is_allocated_above(BlockDriverState *top,
                            BlockDriverState *base,
                            int64_t sector_num,
                            int nb_sectors, int *pnum)
4214 4215 4216 4217 4218 4219 4220
{
    BlockDriverState *intermediate;
    int ret, n = nb_sectors;

    intermediate = top;
    while (intermediate && intermediate != base) {
        int pnum_inter;
4221 4222
        ret = bdrv_is_allocated(intermediate, sector_num, nb_sectors,
                                &pnum_inter);
4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233 4234 4235
        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.
         */
4236 4237 4238
        if (n > pnum_inter &&
            (intermediate == top ||
             sector_num + pnum_inter < intermediate->total_sectors)) {
4239 4240 4241 4242 4243 4244 4245 4246 4247 4248
            n = pnum_inter;
        }

        intermediate = intermediate->backing_hd;
    }

    *pnum = n;
    return 0;
}

4249 4250 4251 4252 4253 4254 4255 4256 4257 4258
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;
}

4259
void bdrv_get_backing_filename(BlockDriverState *bs,
B
bellard 已提交
4260 4261
                               char *filename, int filename_size)
{
K
Kevin Wolf 已提交
4262
    pstrcpy(filename, filename_size, bs->backing_file);
B
bellard 已提交
4263 4264
}

4265
int bdrv_write_compressed(BlockDriverState *bs, int64_t sector_num,
B
bellard 已提交
4266 4267 4268
                          const uint8_t *buf, int nb_sectors)
{
    BlockDriver *drv = bs->drv;
4269 4270 4271
    int ret;

    if (!drv) {
B
bellard 已提交
4272
        return -ENOMEDIUM;
4273 4274
    }
    if (!drv->bdrv_write_compressed) {
B
bellard 已提交
4275
        return -ENOTSUP;
4276 4277 4278 4279 4280
    }
    ret = bdrv_check_request(bs, sector_num, nb_sectors);
    if (ret < 0) {
        return ret;
    }
4281

F
Fam Zheng 已提交
4282
    assert(QLIST_EMPTY(&bs->dirty_bitmaps));
4283

B
bellard 已提交
4284 4285
    return drv->bdrv_write_compressed(bs, sector_num, buf, nb_sectors);
}
4286

B
bellard 已提交
4287 4288 4289 4290
int bdrv_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
4291
        return -ENOMEDIUM;
B
bellard 已提交
4292 4293 4294 4295 4296 4297
    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
4298 4299 4300 4301 4302 4303 4304 4305 4306
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;
}

4307 4308
int bdrv_save_vmstate(BlockDriverState *bs, const uint8_t *buf,
                      int64_t pos, int size)
4309 4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320
{
    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)
4321 4322
{
    BlockDriver *drv = bs->drv;
4323 4324

    if (!drv) {
4325
        return -ENOMEDIUM;
4326 4327 4328 4329 4330 4331
    } 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);
    }

4332
    return -ENOTSUP;
4333 4334
}

4335 4336
int bdrv_load_vmstate(BlockDriverState *bs, uint8_t *buf,
                      int64_t pos, int size)
4337 4338 4339 4340
{
    BlockDriver *drv = bs->drv;
    if (!drv)
        return -ENOMEDIUM;
4341 4342 4343 4344 4345
    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;
4346 4347
}

K
Kevin Wolf 已提交
4348 4349
void bdrv_debug_event(BlockDriverState *bs, BlkDebugEvent event)
{
4350
    if (!bs || !bs->drv || !bs->drv->bdrv_debug_event) {
K
Kevin Wolf 已提交
4351 4352 4353
        return;
    }

4354
    bs->drv->bdrv_debug_event(bs, event);
K
Kevin Wolf 已提交
4355 4356 4357 4358 4359 4360 4361 4362 4363 4364 4365 4366 4367 4368 4369 4370
}

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
4371 4372 4373 4374 4375 4376 4377 4378 4379 4380 4381 4382 4383
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
4384 4385
int bdrv_debug_resume(BlockDriverState *bs, const char *tag)
{
4386
    while (bs && (!bs->drv || !bs->drv->bdrv_debug_resume)) {
K
Kevin Wolf 已提交
4387 4388
        bs = bs->file;
    }
K
Kevin Wolf 已提交
4389

K
Kevin Wolf 已提交
4390 4391 4392 4393 4394 4395 4396 4397 4398 4399 4400 4401 4402 4403 4404 4405 4406 4407
    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
4408 4409
}

4410 4411 4412 4413 4414
int bdrv_is_snapshot(BlockDriverState *bs)
{
    return !!(bs->open_flags & BDRV_O_SNAPSHOT);
}

4415 4416 4417 4418
/* 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. */
4419 4420 4421
BlockDriverState *bdrv_find_backing_image(BlockDriverState *bs,
        const char *backing_file)
{
4422 4423 4424 4425 4426 4427 4428 4429
    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) {
4430 4431 4432
        return NULL;
    }

4433 4434 4435 4436 4437 4438 4439 4440 4441 4442 4443 4444 4445 4446 4447
    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;
            }
4448
        } else {
4449 4450 4451 4452 4453 4454 4455 4456 4457 4458 4459 4460 4461 4462 4463 4464 4465 4466 4467 4468 4469 4470 4471
            /* 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;
            }
4472 4473 4474
        }
    }

4475 4476 4477 4478
    g_free(filename_full);
    g_free(backing_file_full);
    g_free(filename_tmp);
    return retval;
4479 4480
}

4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491 4492 4493
int bdrv_get_backing_file_depth(BlockDriverState *bs)
{
    if (!bs->drv) {
        return 0;
    }

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

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

B
bellard 已提交
4494
/**************************************************************/
B
bellard 已提交
4495
/* async I/Os */
B
bellard 已提交
4496

4497 4498
BlockAIOCB *bdrv_aio_readv(BlockDriverState *bs, int64_t sector_num,
                           QEMUIOVector *qiov, int nb_sectors,
4499
                           BlockCompletionFunc *cb, void *opaque)
B
bellard 已提交
4500
{
4501 4502
    trace_bdrv_aio_readv(bs, sector_num, nb_sectors, opaque);

P
Paolo Bonzini 已提交
4503
    return bdrv_co_aio_rw_vector(bs, sector_num, qiov, nb_sectors, 0,
4504
                                 cb, opaque, false);
B
bellard 已提交
4505 4506
}

4507 4508
BlockAIOCB *bdrv_aio_writev(BlockDriverState *bs, int64_t sector_num,
                            QEMUIOVector *qiov, int nb_sectors,
4509
                            BlockCompletionFunc *cb, void *opaque)
B
bellard 已提交
4510
{
4511 4512
    trace_bdrv_aio_writev(bs, sector_num, nb_sectors, opaque);

P
Paolo Bonzini 已提交
4513
    return bdrv_co_aio_rw_vector(bs, sector_num, qiov, nb_sectors, 0,
4514
                                 cb, opaque, true);
B
bellard 已提交
4515 4516
}

4517
BlockAIOCB *bdrv_aio_write_zeroes(BlockDriverState *bs,
P
Paolo Bonzini 已提交
4518
        int64_t sector_num, int nb_sectors, BdrvRequestFlags flags,
4519
        BlockCompletionFunc *cb, void *opaque)
P
Paolo Bonzini 已提交
4520 4521 4522 4523 4524 4525 4526 4527
{
    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
4528 4529 4530 4531 4532 4533

typedef struct MultiwriteCB {
    int error;
    int num_requests;
    int num_callbacks;
    struct {
4534
        BlockCompletionFunc *cb;
K
Kevin Wolf 已提交
4535 4536 4537 4538 4539 4540 4541 4542 4543 4544 4545
        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);
4546 4547 4548
        if (mcb->callbacks[i].free_qiov) {
            qemu_iovec_destroy(mcb->callbacks[i].free_qiov);
        }
4549
        g_free(mcb->callbacks[i].free_qiov);
K
Kevin Wolf 已提交
4550 4551 4552 4553 4554 4555 4556
    }
}

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

4557 4558
    trace_multiwrite_cb(mcb, ret);

4559
    if (ret < 0 && !mcb->error) {
K
Kevin Wolf 已提交
4560 4561 4562 4563 4564
        mcb->error = ret;
    }

    mcb->num_requests--;
    if (mcb->num_requests == 0) {
4565
        multiwrite_user_cb(mcb);
4566
        g_free(mcb);
K
Kevin Wolf 已提交
4567 4568 4569 4570 4571
    }
}

static int multiwrite_req_compare(const void *a, const void *b)
{
4572 4573 4574 4575 4576 4577 4578 4579 4580 4581 4582 4583 4584
    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
4585 4586 4587 4588 4589 4590 4591 4592 4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604 4605
}

/*
 * 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;

4606
        // Handle exactly sequential writes and overlapping writes.
K
Kevin Wolf 已提交
4607 4608 4609 4610
        if (reqs[i].sector <= oldreq_last) {
            merge = 1;
        }

4611 4612 4613 4614
        if (reqs[outidx].qiov->niov + reqs[i].qiov->niov + 1 > IOV_MAX) {
            merge = 0;
        }

4615 4616 4617 4618 4619
        if (bs->bl.max_transfer_length && reqs[outidx].nb_sectors +
            reqs[i].nb_sectors > bs->bl.max_transfer_length) {
            merge = 0;
        }

K
Kevin Wolf 已提交
4620 4621
        if (merge) {
            size_t size;
4622
            QEMUIOVector *qiov = g_malloc0(sizeof(*qiov));
K
Kevin Wolf 已提交
4623 4624 4625 4626 4627 4628
            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;
4629
            qemu_iovec_concat(qiov, reqs[outidx].qiov, 0, size);
K
Kevin Wolf 已提交
4630

4631 4632
            // We should need to add any zeros between the two requests
            assert (reqs[i].sector <= oldreq_last);
K
Kevin Wolf 已提交
4633 4634

            // Add the second request
4635
            qemu_iovec_concat(qiov, reqs[i].qiov, 0, reqs[i].qiov->size);
K
Kevin Wolf 已提交
4636

4637 4638 4639 4640 4641 4642
            // Add tail of first request, if necessary
            if (qiov->size < reqs[outidx].qiov->size) {
                qemu_iovec_concat(qiov, reqs[outidx].qiov, qiov->size,
                                  reqs[outidx].qiov->size - qiov->size);
            }

4643
            reqs[outidx].nb_sectors = qiov->size >> 9;
K
Kevin Wolf 已提交
4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654
            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;
        }
    }

4655 4656
    block_acct_merge_done(&bs->stats, BLOCK_ACCT_WRITE, num_reqs - outidx - 1);

K
Kevin Wolf 已提交
4657 4658 4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678
    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;

4679 4680 4681 4682 4683 4684 4685 4686
    /* 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
4687 4688 4689 4690 4691
    if (num_reqs == 0) {
        return 0;
    }

    // Create MultiwriteCB structure
4692
    mcb = g_malloc0(sizeof(*mcb) + num_reqs * sizeof(*mcb->callbacks));
K
Kevin Wolf 已提交
4693 4694 4695 4696 4697 4698 4699 4700 4701 4702 4703
    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);

4704 4705
    trace_bdrv_aio_multiwrite(mcb, mcb->num_callbacks, num_reqs);

4706 4707
    /* Run the aio requests. */
    mcb->num_requests = num_reqs;
K
Kevin Wolf 已提交
4708
    for (i = 0; i < num_reqs; i++) {
P
Paolo Bonzini 已提交
4709 4710 4711 4712
        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
4713 4714 4715 4716 4717
    }

    return 0;
}

4718
void bdrv_aio_cancel(BlockAIOCB *acb)
B
bellard 已提交
4719
{
F
Fam Zheng 已提交
4720 4721 4722 4723 4724 4725 4726 4727 4728
    qemu_aio_ref(acb);
    bdrv_aio_cancel_async(acb);
    while (acb->refcnt > 1) {
        if (acb->aiocb_info->get_aio_context) {
            aio_poll(acb->aiocb_info->get_aio_context(acb), true);
        } else if (acb->bs) {
            aio_poll(bdrv_get_aio_context(acb->bs), true);
        } else {
            abort();
F
Fam Zheng 已提交
4729 4730
        }
    }
4731
    qemu_aio_unref(acb);
F
Fam Zheng 已提交
4732 4733 4734 4735 4736
}

/* Async version of aio cancel. The caller is not blocked if the acb implements
 * cancel_async, otherwise we do nothing and let the request normally complete.
 * In either case the completion callback must be called. */
4737
void bdrv_aio_cancel_async(BlockAIOCB *acb)
F
Fam Zheng 已提交
4738 4739 4740 4741
{
    if (acb->aiocb_info->cancel_async) {
        acb->aiocb_info->cancel_async(acb);
    }
B
bellard 已提交
4742 4743 4744 4745 4746
}

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

4747 4748
typedef struct BlockAIOCBSync {
    BlockAIOCB common;
4749 4750 4751 4752 4753 4754
    QEMUBH *bh;
    int ret;
    /* vector translation state */
    QEMUIOVector *qiov;
    uint8_t *bounce;
    int is_write;
4755
} BlockAIOCBSync;
4756

S
Stefan Hajnoczi 已提交
4757
static const AIOCBInfo bdrv_em_aiocb_info = {
4758
    .aiocb_size         = sizeof(BlockAIOCBSync),
4759 4760
};

4761
static void bdrv_aio_bh_cb(void *opaque)
B
bellard 已提交
4762
{
4763
    BlockAIOCBSync *acb = opaque;
4764

4765
    if (!acb->is_write && acb->ret >= 0) {
4766
        qemu_iovec_from_buf(acb->qiov, 0, acb->bounce, acb->qiov->size);
4767
    }
4768
    qemu_vfree(acb->bounce);
4769
    acb->common.cb(acb->common.opaque, acb->ret);
4770
    qemu_bh_delete(acb->bh);
A
Avi Kivity 已提交
4771
    acb->bh = NULL;
4772
    qemu_aio_unref(acb);
B
bellard 已提交
4773
}
4774

4775 4776 4777 4778
static BlockAIOCB *bdrv_aio_rw_vector(BlockDriverState *bs,
                                      int64_t sector_num,
                                      QEMUIOVector *qiov,
                                      int nb_sectors,
4779
                                      BlockCompletionFunc *cb,
4780 4781
                                      void *opaque,
                                      int is_write)
4782

B
bellard 已提交
4783
{
4784
    BlockAIOCBSync *acb;
4785

S
Stefan Hajnoczi 已提交
4786
    acb = qemu_aio_get(&bdrv_em_aiocb_info, bs, cb, opaque);
4787 4788
    acb->is_write = is_write;
    acb->qiov = qiov;
4789
    acb->bounce = qemu_try_blockalign(bs, qiov->size);
4790
    acb->bh = aio_bh_new(bdrv_get_aio_context(bs), bdrv_aio_bh_cb, acb);
4791

4792 4793 4794
    if (acb->bounce == NULL) {
        acb->ret = -ENOMEM;
    } else if (is_write) {
4795
        qemu_iovec_to_buf(acb->qiov, 0, acb->bounce, qiov->size);
4796
        acb->ret = bs->drv->bdrv_write(bs, sector_num, acb->bounce, nb_sectors);
4797
    } else {
4798
        acb->ret = bs->drv->bdrv_read(bs, sector_num, acb->bounce, nb_sectors);
4799 4800
    }

4801
    qemu_bh_schedule(acb->bh);
4802

4803
    return &acb->common;
4804 4805
}

4806
static BlockAIOCB *bdrv_aio_readv_em(BlockDriverState *bs,
4807
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
4808
        BlockCompletionFunc *cb, void *opaque)
4809
{
4810 4811
    return bdrv_aio_rw_vector(bs, sector_num, qiov, nb_sectors, cb, opaque, 0);
}
B
bellard 已提交
4812

4813
static BlockAIOCB *bdrv_aio_writev_em(BlockDriverState *bs,
4814
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
4815
        BlockCompletionFunc *cb, void *opaque)
4816 4817
{
    return bdrv_aio_rw_vector(bs, sector_num, qiov, nb_sectors, cb, opaque, 1);
4818 4819
}

4820

4821 4822
typedef struct BlockAIOCBCoroutine {
    BlockAIOCB common;
4823 4824
    BlockRequest req;
    bool is_write;
K
Kevin Wolf 已提交
4825
    bool *done;
4826
    QEMUBH* bh;
4827
} BlockAIOCBCoroutine;
4828

S
Stefan Hajnoczi 已提交
4829
static const AIOCBInfo bdrv_em_co_aiocb_info = {
4830
    .aiocb_size         = sizeof(BlockAIOCBCoroutine),
4831 4832
};

P
Paolo Bonzini 已提交
4833
static void bdrv_co_em_bh(void *opaque)
4834
{
4835
    BlockAIOCBCoroutine *acb = opaque;
4836 4837

    acb->common.cb(acb->common.opaque, acb->req.error);
K
Kevin Wolf 已提交
4838

4839
    qemu_bh_delete(acb->bh);
4840
    qemu_aio_unref(acb);
4841 4842
}

4843 4844 4845
/* Invoke bdrv_co_do_readv/bdrv_co_do_writev */
static void coroutine_fn bdrv_co_do_rw(void *opaque)
{
4846
    BlockAIOCBCoroutine *acb = opaque;
4847 4848 4849 4850
    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
4851
            acb->req.nb_sectors, acb->req.qiov, acb->req.flags);
4852 4853
    } else {
        acb->req.error = bdrv_co_do_writev(bs, acb->req.sector,
P
Paolo Bonzini 已提交
4854
            acb->req.nb_sectors, acb->req.qiov, acb->req.flags);
4855 4856
    }

4857
    acb->bh = aio_bh_new(bdrv_get_aio_context(bs), bdrv_co_em_bh, acb);
4858 4859 4860
    qemu_bh_schedule(acb->bh);
}

4861 4862 4863 4864 4865
static BlockAIOCB *bdrv_co_aio_rw_vector(BlockDriverState *bs,
                                         int64_t sector_num,
                                         QEMUIOVector *qiov,
                                         int nb_sectors,
                                         BdrvRequestFlags flags,
4866
                                         BlockCompletionFunc *cb,
4867 4868
                                         void *opaque,
                                         bool is_write)
4869 4870
{
    Coroutine *co;
4871
    BlockAIOCBCoroutine *acb;
4872

S
Stefan Hajnoczi 已提交
4873
    acb = qemu_aio_get(&bdrv_em_co_aiocb_info, bs, cb, opaque);
4874 4875 4876
    acb->req.sector = sector_num;
    acb->req.nb_sectors = nb_sectors;
    acb->req.qiov = qiov;
P
Paolo Bonzini 已提交
4877
    acb->req.flags = flags;
4878 4879
    acb->is_write = is_write;

4880
    co = qemu_coroutine_create(bdrv_co_do_rw);
4881 4882 4883 4884 4885
    qemu_coroutine_enter(co, acb);

    return &acb->common;
}

P
Paolo Bonzini 已提交
4886
static void coroutine_fn bdrv_aio_flush_co_entry(void *opaque)
4887
{
4888
    BlockAIOCBCoroutine *acb = opaque;
P
Paolo Bonzini 已提交
4889
    BlockDriverState *bs = acb->common.bs;
4890

P
Paolo Bonzini 已提交
4891
    acb->req.error = bdrv_co_flush(bs);
4892
    acb->bh = aio_bh_new(bdrv_get_aio_context(bs), bdrv_co_em_bh, acb);
4893 4894 4895
    qemu_bh_schedule(acb->bh);
}

4896
BlockAIOCB *bdrv_aio_flush(BlockDriverState *bs,
4897
        BlockCompletionFunc *cb, void *opaque)
4898
{
P
Paolo Bonzini 已提交
4899
    trace_bdrv_aio_flush(bs, opaque);
4900

P
Paolo Bonzini 已提交
4901
    Coroutine *co;
4902
    BlockAIOCBCoroutine *acb;
4903

S
Stefan Hajnoczi 已提交
4904
    acb = qemu_aio_get(&bdrv_em_co_aiocb_info, bs, cb, opaque);
K
Kevin Wolf 已提交
4905

P
Paolo Bonzini 已提交
4906 4907
    co = qemu_coroutine_create(bdrv_aio_flush_co_entry);
    qemu_coroutine_enter(co, acb);
4908 4909 4910 4911

    return &acb->common;
}

4912 4913
static void coroutine_fn bdrv_aio_discard_co_entry(void *opaque)
{
4914
    BlockAIOCBCoroutine *acb = opaque;
4915 4916 4917
    BlockDriverState *bs = acb->common.bs;

    acb->req.error = bdrv_co_discard(bs, acb->req.sector, acb->req.nb_sectors);
4918
    acb->bh = aio_bh_new(bdrv_get_aio_context(bs), bdrv_co_em_bh, acb);
4919 4920 4921
    qemu_bh_schedule(acb->bh);
}

4922
BlockAIOCB *bdrv_aio_discard(BlockDriverState *bs,
4923
        int64_t sector_num, int nb_sectors,
4924
        BlockCompletionFunc *cb, void *opaque)
4925 4926
{
    Coroutine *co;
4927
    BlockAIOCBCoroutine *acb;
4928 4929 4930

    trace_bdrv_aio_discard(bs, sector_num, nb_sectors, opaque);

S
Stefan Hajnoczi 已提交
4931
    acb = qemu_aio_get(&bdrv_em_co_aiocb_info, bs, cb, opaque);
4932 4933 4934 4935 4936 4937 4938 4939
    acb->req.sector = sector_num;
    acb->req.nb_sectors = nb_sectors;
    co = qemu_coroutine_create(bdrv_aio_discard_co_entry);
    qemu_coroutine_enter(co, acb);

    return &acb->common;
}

B
bellard 已提交
4940 4941
void bdrv_init(void)
{
4942
    module_call_init(MODULE_INIT_BLOCK);
B
bellard 已提交
4943
}
4944

4945 4946 4947 4948 4949 4950
void bdrv_init_with_whitelist(void)
{
    use_bdrv_whitelist = 1;
    bdrv_init();
}

S
Stefan Hajnoczi 已提交
4951
void *qemu_aio_get(const AIOCBInfo *aiocb_info, BlockDriverState *bs,
4952
                   BlockCompletionFunc *cb, void *opaque)
4953
{
4954
    BlockAIOCB *acb;
4955

S
Stefan Hajnoczi 已提交
4956 4957
    acb = g_slice_alloc(aiocb_info->aiocb_size);
    acb->aiocb_info = aiocb_info;
4958 4959 4960
    acb->bs = bs;
    acb->cb = cb;
    acb->opaque = opaque;
4961
    acb->refcnt = 1;
4962 4963 4964
    return acb;
}

4965 4966
void qemu_aio_ref(void *p)
{
4967
    BlockAIOCB *acb = p;
4968 4969 4970
    acb->refcnt++;
}

4971
void qemu_aio_unref(void *p)
4972
{
4973
    BlockAIOCB *acb = p;
4974 4975 4976 4977
    assert(acb->refcnt > 0);
    if (--acb->refcnt == 0) {
        g_slice_free1(acb->aiocb_info->aiocb_size, acb);
    }
4978
}
B
bellard 已提交
4979

4980 4981 4982 4983 4984 4985 4986 4987 4988 4989 4990 4991 4992 4993 4994 4995 4996 4997 4998 4999 5000 5001 5002
/**************************************************************/
/* 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(),
    };
5003
    BlockAIOCB *acb;
5004 5005

    if (is_write) {
5006 5007
        acb = bs->drv->bdrv_aio_writev(bs, sector_num, iov, nb_sectors,
                                       bdrv_co_io_em_complete, &co);
5008
    } else {
5009 5010
        acb = bs->drv->bdrv_aio_readv(bs, sector_num, iov, nb_sectors,
                                      bdrv_co_io_em_complete, &co);
5011 5012
    }

5013
    trace_bdrv_co_io_em(bs, sector_num, nb_sectors, is_write, acb);
5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035
    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
5036
static void coroutine_fn bdrv_flush_co_entry(void *opaque)
5037
{
P
Paolo Bonzini 已提交
5038 5039 5040 5041 5042 5043 5044
    RwCo *rwco = opaque;

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

int coroutine_fn bdrv_co_flush(BlockDriverState *bs)
{
K
Kevin Wolf 已提交
5045 5046
    int ret;

5047
    if (!bs || !bdrv_is_inserted(bs) || bdrv_is_read_only(bs)) {
P
Paolo Bonzini 已提交
5048
        return 0;
K
Kevin Wolf 已提交
5049 5050
    }

5051
    /* Write back cached data to the OS even with cache=unsafe */
5052
    BLKDBG_EVENT(bs->file, BLKDBG_FLUSH_TO_OS);
K
Kevin Wolf 已提交
5053 5054 5055 5056 5057 5058 5059
    if (bs->drv->bdrv_co_flush_to_os) {
        ret = bs->drv->bdrv_co_flush_to_os(bs);
        if (ret < 0) {
            return ret;
        }
    }

5060 5061
    /* But don't actually force it to the disk with cache=unsafe */
    if (bs->open_flags & BDRV_O_NO_FLUSH) {
5062
        goto flush_parent;
5063 5064
    }

5065
    BLKDBG_EVENT(bs->file, BLKDBG_FLUSH_TO_DISK);
K
Kevin Wolf 已提交
5066
    if (bs->drv->bdrv_co_flush_to_disk) {
5067
        ret = bs->drv->bdrv_co_flush_to_disk(bs);
P
Paolo Bonzini 已提交
5068
    } else if (bs->drv->bdrv_aio_flush) {
5069
        BlockAIOCB *acb;
P
Paolo Bonzini 已提交
5070 5071 5072 5073 5074 5075
        CoroutineIOCompletion co = {
            .coroutine = qemu_coroutine_self(),
        };

        acb = bs->drv->bdrv_aio_flush(bs, bdrv_co_io_em_complete, &co);
        if (acb == NULL) {
5076
            ret = -EIO;
P
Paolo Bonzini 已提交
5077 5078
        } else {
            qemu_coroutine_yield();
5079
            ret = co.ret;
P
Paolo Bonzini 已提交
5080 5081 5082 5083 5084 5085 5086 5087 5088 5089 5090 5091 5092
        }
    } 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.
         */
5093
        ret = 0;
P
Paolo Bonzini 已提交
5094
    }
5095 5096 5097 5098 5099 5100 5101
    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.
     */
5102
flush_parent:
5103
    return bdrv_co_flush(bs->file);
P
Paolo Bonzini 已提交
5104 5105
}

5106
void bdrv_invalidate_cache(BlockDriverState *bs, Error **errp)
5107
{
5108 5109 5110
    Error *local_err = NULL;
    int ret;

5111 5112 5113 5114
    if (!bs->drv)  {
        return;
    }

5115 5116 5117 5118 5119
    if (!(bs->open_flags & BDRV_O_INCOMING)) {
        return;
    }
    bs->open_flags &= ~BDRV_O_INCOMING;

5120
    if (bs->drv->bdrv_invalidate_cache) {
5121
        bs->drv->bdrv_invalidate_cache(bs, &local_err);
5122
    } else if (bs->file) {
5123 5124 5125 5126 5127
        bdrv_invalidate_cache(bs->file, &local_err);
    }
    if (local_err) {
        error_propagate(errp, local_err);
        return;
5128
    }
5129

5130 5131 5132 5133 5134
    ret = refresh_total_sectors(bs, bs->total_sectors);
    if (ret < 0) {
        error_setg_errno(errp, -ret, "Could not refresh total sector count");
        return;
    }
5135 5136
}

5137
void bdrv_invalidate_cache_all(Error **errp)
5138 5139
{
    BlockDriverState *bs;
5140
    Error *local_err = NULL;
5141

5142
    QTAILQ_FOREACH(bs, &bdrv_states, device_list) {
5143 5144 5145
        AioContext *aio_context = bdrv_get_aio_context(bs);

        aio_context_acquire(aio_context);
5146
        bdrv_invalidate_cache(bs, &local_err);
5147
        aio_context_release(aio_context);
5148 5149 5150 5151
        if (local_err) {
            error_propagate(errp, local_err);
            return;
        }
5152 5153 5154
    }
}

P
Paolo Bonzini 已提交
5155 5156 5157 5158 5159 5160
int bdrv_flush(BlockDriverState *bs)
{
    Coroutine *co;
    RwCo rwco = {
        .bs = bs,
        .ret = NOT_DONE,
5161 5162
    };

P
Paolo Bonzini 已提交
5163 5164 5165 5166
    if (qemu_in_coroutine()) {
        /* Fast-path if already in coroutine context */
        bdrv_flush_co_entry(&rwco);
    } else {
5167 5168
        AioContext *aio_context = bdrv_get_aio_context(bs);

P
Paolo Bonzini 已提交
5169 5170 5171
        co = qemu_coroutine_create(bdrv_flush_co_entry);
        qemu_coroutine_enter(co, &rwco);
        while (rwco.ret == NOT_DONE) {
5172
            aio_poll(aio_context, true);
P
Paolo Bonzini 已提交
5173
        }
5174
    }
P
Paolo Bonzini 已提交
5175 5176

    return rwco.ret;
5177 5178
}

5179 5180 5181 5182 5183 5184
typedef struct DiscardCo {
    BlockDriverState *bs;
    int64_t sector_num;
    int nb_sectors;
    int ret;
} DiscardCo;
5185 5186
static void coroutine_fn bdrv_discard_co_entry(void *opaque)
{
5187
    DiscardCo *rwco = opaque;
5188 5189 5190 5191 5192 5193 5194

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

int coroutine_fn bdrv_co_discard(BlockDriverState *bs, int64_t sector_num,
                                 int nb_sectors)
{
5195
    int max_discard, ret;
5196

5197 5198
    if (!bs->drv) {
        return -ENOMEDIUM;
5199 5200 5201 5202 5203
    }

    ret = bdrv_check_request(bs, sector_num, nb_sectors);
    if (ret < 0) {
        return ret;
5204 5205
    } else if (bs->read_only) {
        return -EROFS;
5206 5207
    }

F
Fam Zheng 已提交
5208
    bdrv_reset_dirty(bs, sector_num, nb_sectors);
5209

P
Paolo Bonzini 已提交
5210 5211 5212 5213 5214
    /* Do nothing if disabled.  */
    if (!(bs->open_flags & BDRV_O_UNMAP)) {
        return 0;
    }

5215 5216 5217
    if (!bs->drv->bdrv_co_discard && !bs->drv->bdrv_aio_discard) {
        return 0;
    }
5218

5219
    max_discard = MIN_NON_ZERO(bs->bl.max_discard, BDRV_REQUEST_MAX_SECTORS);
5220 5221 5222
    while (nb_sectors > 0) {
        int ret;
        int num = nb_sectors;
5223

5224 5225 5226 5227 5228 5229
        /* 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;
5230
            }
5231 5232
            num -= sector_num % bs->bl.discard_alignment;
        }
5233

5234 5235 5236 5237
        /* limit request size */
        if (num > max_discard) {
            num = max_discard;
        }
5238

5239
        if (bs->drv->bdrv_co_discard) {
5240
            ret = bs->drv->bdrv_co_discard(bs, sector_num, num);
5241
        } else {
5242
            BlockAIOCB *acb;
5243 5244 5245 5246 5247 5248 5249 5250 5251 5252 5253
            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;
5254 5255
            }
        }
5256
        if (ret && ret != -ENOTSUP) {
5257
            return ret;
5258
        }
5259 5260 5261

        sector_num += num;
        nb_sectors -= num;
5262
    }
5263
    return 0;
5264 5265 5266 5267 5268
}

int bdrv_discard(BlockDriverState *bs, int64_t sector_num, int nb_sectors)
{
    Coroutine *co;
5269
    DiscardCo rwco = {
5270 5271 5272 5273 5274 5275 5276 5277 5278 5279
        .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 {
5280 5281
        AioContext *aio_context = bdrv_get_aio_context(bs);

5282 5283 5284
        co = qemu_coroutine_create(bdrv_discard_co_entry);
        qemu_coroutine_enter(co, &rwco);
        while (rwco.ret == NOT_DONE) {
5285
            aio_poll(aio_context, true);
5286 5287 5288 5289 5290 5291
        }
    }

    return rwco.ret;
}

B
bellard 已提交
5292 5293 5294 5295 5296 5297 5298 5299 5300
/**************************************************************/
/* removable device support */

/**
 * Return TRUE if the media is present
 */
int bdrv_is_inserted(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;
5301

B
bellard 已提交
5302 5303 5304
    if (!drv)
        return 0;
    if (!drv->bdrv_is_inserted)
5305 5306
        return 1;
    return drv->bdrv_is_inserted(bs);
B
bellard 已提交
5307 5308 5309
}

/**
5310 5311
 * 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
5312 5313 5314 5315 5316
 */
int bdrv_media_changed(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;

5317 5318 5319 5320
    if (drv && drv->bdrv_media_changed) {
        return drv->bdrv_media_changed(bs);
    }
    return -ENOTSUP;
B
bellard 已提交
5321 5322 5323 5324 5325
}

/**
 * If eject_flag is TRUE, eject the media. Otherwise, close the tray
 */
5326
void bdrv_eject(BlockDriverState *bs, bool eject_flag)
B
bellard 已提交
5327 5328
{
    BlockDriver *drv = bs->drv;
5329
    const char *device_name;
B
bellard 已提交
5330

5331 5332
    if (drv && drv->bdrv_eject) {
        drv->bdrv_eject(bs, eject_flag);
B
bellard 已提交
5333
    }
5334

5335 5336 5337
    device_name = bdrv_get_device_name(bs);
    if (device_name[0] != '\0') {
        qapi_event_send_device_tray_moved(device_name,
5338
                                          eject_flag, &error_abort);
5339
    }
B
bellard 已提交
5340 5341 5342 5343 5344 5345
}

/**
 * Lock or unlock the media (if it is locked, the user won't be able
 * to eject it manually).
 */
5346
void bdrv_lock_medium(BlockDriverState *bs, bool locked)
B
bellard 已提交
5347 5348 5349
{
    BlockDriver *drv = bs->drv;

5350
    trace_bdrv_lock_medium(bs, locked);
S
Stefan Hajnoczi 已提交
5351

5352 5353
    if (drv && drv->bdrv_lock_medium) {
        drv->bdrv_lock_medium(bs, locked);
B
bellard 已提交
5354 5355
    }
}
5356 5357 5358 5359 5360 5361 5362 5363 5364 5365 5366

/* 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;
}
5367

5368
BlockAIOCB *bdrv_aio_ioctl(BlockDriverState *bs,
5369
        unsigned long int req, void *buf,
5370
        BlockCompletionFunc *cb, void *opaque)
5371
{
5372
    BlockDriver *drv = bs->drv;
5373

5374 5375 5376
    if (drv && drv->bdrv_aio_ioctl)
        return drv->bdrv_aio_ioctl(bs, req, buf, cb, opaque);
    return NULL;
5377
}
5378

5379
void bdrv_set_guest_block_size(BlockDriverState *bs, int align)
5380
{
5381
    bs->guest_block_size = align;
5382
}
5383

5384 5385
void *qemu_blockalign(BlockDriverState *bs, size_t size)
{
5386
    return qemu_memalign(bdrv_opt_mem_align(bs), size);
5387
}
5388

M
Max Reitz 已提交
5389 5390 5391 5392 5393
void *qemu_blockalign0(BlockDriverState *bs, size_t size)
{
    return memset(qemu_blockalign(bs, size), 0, size);
}

5394 5395 5396 5397 5398 5399 5400 5401 5402 5403 5404 5405 5406
void *qemu_try_blockalign(BlockDriverState *bs, size_t size)
{
    size_t align = bdrv_opt_mem_align(bs);

    /* Ensure that NULL is never returned on success */
    assert(align > 0);
    if (size == 0) {
        size = align;
    }

    return qemu_try_memalign(align, size);
}

M
Max Reitz 已提交
5407 5408 5409 5410 5411 5412 5413 5414 5415 5416 5417
void *qemu_try_blockalign0(BlockDriverState *bs, size_t size)
{
    void *mem = qemu_try_blockalign(bs, size);

    if (mem) {
        memset(mem, 0, size);
    }

    return mem;
}

5418 5419 5420 5421 5422 5423
/*
 * Check if all memory in this vector is sector aligned.
 */
bool bdrv_qiov_is_aligned(BlockDriverState *bs, QEMUIOVector *qiov)
{
    int i;
5424
    size_t alignment = bdrv_opt_mem_align(bs);
5425 5426

    for (i = 0; i < qiov->niov; i++) {
5427
        if ((uintptr_t) qiov->iov[i].iov_base % alignment) {
5428
            return false;
5429
        }
5430
        if (qiov->iov[i].iov_len % alignment) {
5431
            return false;
5432 5433 5434 5435 5436 5437
        }
    }

    return true;
}

5438 5439
BdrvDirtyBitmap *bdrv_create_dirty_bitmap(BlockDriverState *bs, int granularity,
                                          Error **errp)
5440 5441
{
    int64_t bitmap_size;
F
Fam Zheng 已提交
5442
    BdrvDirtyBitmap *bitmap;
5443

5444 5445
    assert((granularity & (granularity - 1)) == 0);

F
Fam Zheng 已提交
5446 5447
    granularity >>= BDRV_SECTOR_BITS;
    assert(granularity);
5448
    bitmap_size = bdrv_nb_sectors(bs);
5449 5450 5451 5452 5453
    if (bitmap_size < 0) {
        error_setg_errno(errp, -bitmap_size, "could not get length of device");
        errno = -bitmap_size;
        return NULL;
    }
5454
    bitmap = g_new0(BdrvDirtyBitmap, 1);
F
Fam Zheng 已提交
5455 5456 5457 5458 5459 5460 5461 5462 5463 5464 5465 5466 5467 5468
    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;
5469
        }
5470 5471 5472
    }
}

F
Fam Zheng 已提交
5473 5474 5475 5476 5477 5478 5479
BlockDirtyInfoList *bdrv_query_dirty_bitmaps(BlockDriverState *bs)
{
    BdrvDirtyBitmap *bm;
    BlockDirtyInfoList *list = NULL;
    BlockDirtyInfoList **plist = &list;

    QLIST_FOREACH(bm, &bs->dirty_bitmaps, list) {
5480 5481
        BlockDirtyInfo *info = g_new0(BlockDirtyInfo, 1);
        BlockDirtyInfoList *entry = g_new0(BlockDirtyInfoList, 1);
F
Fam Zheng 已提交
5482 5483 5484 5485 5486 5487 5488 5489 5490 5491 5492
        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
5493
int bdrv_get_dirty(BlockDriverState *bs, BdrvDirtyBitmap *bitmap, int64_t sector)
5494
{
F
Fam Zheng 已提交
5495 5496
    if (bitmap) {
        return hbitmap_get(bitmap->bitmap, sector);
5497 5498 5499 5500 5501
    } else {
        return 0;
    }
}

F
Fam Zheng 已提交
5502 5503
void bdrv_dirty_iter_init(BlockDriverState *bs,
                          BdrvDirtyBitmap *bitmap, HBitmapIter *hbi)
5504
{
F
Fam Zheng 已提交
5505
    hbitmap_iter_init(hbi, bitmap->bitmap, 0);
5506 5507
}

5508 5509 5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521
void bdrv_set_dirty_bitmap(BlockDriverState *bs, BdrvDirtyBitmap *bitmap,
                           int64_t cur_sector, int nr_sectors)
{
    hbitmap_set(bitmap->bitmap, cur_sector, nr_sectors);
}

void bdrv_reset_dirty_bitmap(BlockDriverState *bs, BdrvDirtyBitmap *bitmap,
                             int64_t cur_sector, int nr_sectors)
{
    hbitmap_reset(bitmap->bitmap, cur_sector, nr_sectors);
}

static void bdrv_set_dirty(BlockDriverState *bs, int64_t cur_sector,
                           int nr_sectors)
5522
{
F
Fam Zheng 已提交
5523 5524 5525 5526
    BdrvDirtyBitmap *bitmap;
    QLIST_FOREACH(bitmap, &bs->dirty_bitmaps, list) {
        hbitmap_set(bitmap->bitmap, cur_sector, nr_sectors);
    }
5527 5528
}

5529 5530
static void bdrv_reset_dirty(BlockDriverState *bs, int64_t cur_sector,
                             int nr_sectors)
5531
{
F
Fam Zheng 已提交
5532 5533 5534 5535
    BdrvDirtyBitmap *bitmap;
    QLIST_FOREACH(bitmap, &bs->dirty_bitmaps, list) {
        hbitmap_reset(bitmap->bitmap, cur_sector, nr_sectors);
    }
5536
}
5537

F
Fam Zheng 已提交
5538
int64_t bdrv_get_dirty_count(BlockDriverState *bs, BdrvDirtyBitmap *bitmap)
5539
{
F
Fam Zheng 已提交
5540
    return hbitmap_count(bitmap->bitmap);
5541
}
J
Jes Sorensen 已提交
5542

5543 5544 5545 5546 5547 5548 5549 5550 5551 5552 5553
/* 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)
{
5554 5555 5556
    if (!bs) {
        return;
    }
5557 5558 5559 5560 5561 5562
    assert(bs->refcnt > 0);
    if (--bs->refcnt == 0) {
        bdrv_delete(bs);
    }
}

5563 5564 5565 5566 5567 5568 5569 5570 5571 5572 5573 5574 5575
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",
5576 5577
                       bdrv_get_device_name(bs),
                       error_get_pretty(blocker->reason));
5578 5579 5580 5581 5582 5583 5584 5585 5586 5587 5588
        }
        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);

5589
    blocker = g_new0(BdrvOpBlocker, 1);
5590 5591 5592 5593 5594 5595 5596 5597 5598 5599 5600 5601 5602 5603 5604 5605 5606 5607 5608 5609 5610 5611 5612 5613 5614 5615 5616 5617 5618 5619 5620 5621 5622 5623 5624 5625 5626 5627 5628 5629 5630 5631 5632 5633
    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;
}

5634 5635
void bdrv_iostatus_enable(BlockDriverState *bs)
{
5636
    bs->iostatus_enabled = true;
5637
    bs->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
5638 5639 5640 5641 5642 5643
}

/* 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)
{
5644
    return (bs->iostatus_enabled &&
5645 5646 5647
           (bs->on_write_error == BLOCKDEV_ON_ERROR_ENOSPC ||
            bs->on_write_error == BLOCKDEV_ON_ERROR_STOP   ||
            bs->on_read_error == BLOCKDEV_ON_ERROR_STOP));
5648 5649 5650 5651
}

void bdrv_iostatus_disable(BlockDriverState *bs)
{
5652
    bs->iostatus_enabled = false;
5653 5654 5655 5656 5657
}

void bdrv_iostatus_reset(BlockDriverState *bs)
{
    if (bdrv_iostatus_is_enabled(bs)) {
5658
        bs->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
5659 5660 5661
        if (bs->job) {
            block_job_iostatus_reset(bs->job);
        }
5662 5663 5664 5665 5666
    }
}

void bdrv_iostatus_set_err(BlockDriverState *bs, int error)
{
5667 5668
    assert(bdrv_iostatus_is_enabled(bs));
    if (bs->iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
5669 5670
        bs->iostatus = error == ENOSPC ? BLOCK_DEVICE_IO_STATUS_NOSPACE :
                                         BLOCK_DEVICE_IO_STATUS_FAILED;
5671 5672 5673
    }
}

5674 5675
void bdrv_img_create(const char *filename, const char *fmt,
                     const char *base_filename, const char *base_fmt,
5676 5677
                     char *options, uint64_t img_size, int flags,
                     Error **errp, bool quiet)
J
Jes Sorensen 已提交
5678
{
5679 5680 5681 5682
    QemuOptsList *create_opts = NULL;
    QemuOpts *opts = NULL;
    const char *backing_fmt, *backing_file;
    int64_t size;
J
Jes Sorensen 已提交
5683
    BlockDriver *drv, *proto_drv;
5684
    BlockDriver *backing_drv = NULL;
5685
    Error *local_err = NULL;
J
Jes Sorensen 已提交
5686 5687 5688 5689 5690
    int ret = 0;

    /* Find driver and parse its options */
    drv = bdrv_find_format(fmt);
    if (!drv) {
5691
        error_setg(errp, "Unknown file format '%s'", fmt);
5692
        return;
J
Jes Sorensen 已提交
5693 5694
    }

5695
    proto_drv = bdrv_find_protocol(filename, true, errp);
J
Jes Sorensen 已提交
5696
    if (!proto_drv) {
5697
        return;
J
Jes Sorensen 已提交
5698 5699
    }

5700 5701 5702 5703 5704 5705 5706 5707 5708 5709 5710 5711
    if (!drv->create_opts) {
        error_setg(errp, "Format driver '%s' does not support image creation",
                   drv->format_name);
        return;
    }

    if (!proto_drv->create_opts) {
        error_setg(errp, "Protocol driver '%s' does not support image creation",
                   proto_drv->format_name);
        return;
    }

C
Chunyan Liu 已提交
5712 5713
    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
5714 5715

    /* Create parameter list with default values */
5716
    opts = qemu_opts_create(create_opts, NULL, 0, &error_abort);
5717
    qemu_opt_set_number(opts, BLOCK_OPT_SIZE, img_size, &error_abort);
J
Jes Sorensen 已提交
5718 5719 5720

    /* Parse -o options */
    if (options) {
5721 5722 5723 5724
        qemu_opts_do_parse(opts, options, NULL, &local_err);
        if (local_err) {
            error_report_err(local_err);
            local_err = NULL;
5725
            error_setg(errp, "Invalid options for file format '%s'", fmt);
J
Jes Sorensen 已提交
5726 5727 5728 5729 5730
            goto out;
        }
    }

    if (base_filename) {
5731
        qemu_opt_set(opts, BLOCK_OPT_BACKING_FILE, base_filename, &local_err);
5732
        if (local_err) {
5733 5734
            error_setg(errp, "Backing file not supported for file format '%s'",
                       fmt);
J
Jes Sorensen 已提交
5735 5736 5737 5738 5739
            goto out;
        }
    }

    if (base_fmt) {
5740
        qemu_opt_set(opts, BLOCK_OPT_BACKING_FMT, base_fmt, &local_err);
5741
        if (local_err) {
5742 5743
            error_setg(errp, "Backing file format not supported for file "
                             "format '%s'", fmt);
J
Jes Sorensen 已提交
5744 5745 5746 5747
            goto out;
        }
    }

5748 5749 5750
    backing_file = qemu_opt_get(opts, BLOCK_OPT_BACKING_FILE);
    if (backing_file) {
        if (!strcmp(filename, backing_file)) {
5751 5752
            error_setg(errp, "Error: Trying to create an image with the "
                             "same filename as the backing file");
5753 5754 5755 5756
            goto out;
        }
    }

5757 5758 5759
    backing_fmt = qemu_opt_get(opts, BLOCK_OPT_BACKING_FMT);
    if (backing_fmt) {
        backing_drv = bdrv_find_format(backing_fmt);
5760
        if (!backing_drv) {
5761
            error_setg(errp, "Unknown backing file format '%s'",
5762
                       backing_fmt);
J
Jes Sorensen 已提交
5763 5764 5765 5766 5767 5768
            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
5769 5770 5771
    size = qemu_opt_get_size(opts, BLOCK_OPT_SIZE, 0);
    if (size == -1) {
        if (backing_file) {
5772
            BlockDriverState *bs;
5773
            char *full_backing = g_new0(char, PATH_MAX);
5774
            int64_t size;
5775 5776
            int back_flags;

5777 5778 5779 5780 5781 5782 5783 5784
            bdrv_get_full_backing_filename_from_filename(filename, backing_file,
                                                         full_backing, PATH_MAX,
                                                         &local_err);
            if (local_err) {
                g_free(full_backing);
                goto out;
            }

5785 5786 5787
            /* 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
5788

5789
            bs = NULL;
5790
            ret = bdrv_open(&bs, full_backing, NULL, NULL, back_flags,
5791
                            backing_drv, &local_err);
5792
            g_free(full_backing);
J
Jes Sorensen 已提交
5793 5794 5795
            if (ret < 0) {
                goto out;
            }
5796 5797 5798 5799 5800 5801 5802
            size = bdrv_getlength(bs);
            if (size < 0) {
                error_setg_errno(errp, -size, "Could not get size of '%s'",
                                 backing_file);
                bdrv_unref(bs);
                goto out;
            }
J
Jes Sorensen 已提交
5803

5804
            qemu_opt_set_number(opts, BLOCK_OPT_SIZE, size, &error_abort);
5805 5806

            bdrv_unref(bs);
J
Jes Sorensen 已提交
5807
        } else {
5808
            error_setg(errp, "Image creation needs a size parameter");
J
Jes Sorensen 已提交
5809 5810 5811 5812
            goto out;
        }
    }

5813
    if (!quiet) {
5814 5815
        printf("Formatting '%s', fmt=%s", filename, fmt);
        qemu_opts_print(opts, " ");
5816 5817
        puts("");
    }
5818

C
Chunyan Liu 已提交
5819
    ret = bdrv_create(drv, filename, opts, &local_err);
5820

5821 5822 5823 5824 5825
    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 = "";
5826
        if (qemu_opt_get_size(opts, BLOCK_OPT_CLUSTER_SIZE, 0)) {
5827
            cluster_size_hint = " (try using a larger cluster size)";
J
Jes Sorensen 已提交
5828
        }
5829 5830 5831 5832
        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
5833 5834 5835
    }

out:
5836 5837
    qemu_opts_del(opts);
    qemu_opts_free(create_opts);
5838
    if (local_err) {
5839 5840
        error_propagate(errp, local_err);
    }
J
Jes Sorensen 已提交
5841
}
5842 5843 5844

AioContext *bdrv_get_aio_context(BlockDriverState *bs)
{
5845 5846 5847 5848 5849
    return bs->aio_context;
}

void bdrv_detach_aio_context(BlockDriverState *bs)
{
M
Max Reitz 已提交
5850 5851
    BdrvAioNotifier *baf;

5852 5853 5854 5855
    if (!bs->drv) {
        return;
    }

M
Max Reitz 已提交
5856 5857 5858 5859
    QLIST_FOREACH(baf, &bs->aio_notifiers, list) {
        baf->detach_aio_context(baf->opaque);
    }

5860 5861 5862
    if (bs->io_limits_enabled) {
        throttle_detach_aio_context(&bs->throttle_state);
    }
5863 5864 5865 5866 5867 5868 5869 5870 5871 5872 5873 5874 5875 5876 5877 5878
    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)
{
M
Max Reitz 已提交
5879 5880
    BdrvAioNotifier *ban;

5881 5882 5883 5884 5885 5886 5887 5888 5889 5890 5891 5892 5893 5894 5895
    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);
    }
5896 5897 5898
    if (bs->io_limits_enabled) {
        throttle_attach_aio_context(&bs->throttle_state, new_context);
    }
M
Max Reitz 已提交
5899 5900 5901 5902

    QLIST_FOREACH(ban, &bs->aio_notifiers, list) {
        ban->attached_aio_context(new_context, ban->opaque);
    }
5903 5904 5905 5906 5907 5908 5909 5910 5911 5912 5913 5914 5915 5916
}

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);
5917
}
5918

M
Max Reitz 已提交
5919 5920 5921 5922 5923 5924 5925 5926 5927 5928 5929 5930 5931 5932 5933 5934 5935 5936 5937 5938 5939 5940 5941 5942 5943 5944 5945 5946 5947 5948 5949 5950 5951 5952 5953 5954 5955
void bdrv_add_aio_context_notifier(BlockDriverState *bs,
        void (*attached_aio_context)(AioContext *new_context, void *opaque),
        void (*detach_aio_context)(void *opaque), void *opaque)
{
    BdrvAioNotifier *ban = g_new(BdrvAioNotifier, 1);
    *ban = (BdrvAioNotifier){
        .attached_aio_context = attached_aio_context,
        .detach_aio_context   = detach_aio_context,
        .opaque               = opaque
    };

    QLIST_INSERT_HEAD(&bs->aio_notifiers, ban, list);
}

void bdrv_remove_aio_context_notifier(BlockDriverState *bs,
                                      void (*attached_aio_context)(AioContext *,
                                                                   void *),
                                      void (*detach_aio_context)(void *),
                                      void *opaque)
{
    BdrvAioNotifier *ban, *ban_next;

    QLIST_FOREACH_SAFE(ban, &bs->aio_notifiers, list, ban_next) {
        if (ban->attached_aio_context == attached_aio_context &&
            ban->detach_aio_context   == detach_aio_context   &&
            ban->opaque               == opaque)
        {
            QLIST_REMOVE(ban, list);
            g_free(ban);

            return;
        }
    }

    abort();
}

5956 5957 5958 5959 5960
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
5961

5962 5963
int bdrv_amend_options(BlockDriverState *bs, QemuOpts *opts,
                       BlockDriverAmendStatusCB *status_cb)
M
Max Reitz 已提交
5964
{
C
Chunyan Liu 已提交
5965
    if (!bs->drv->bdrv_amend_options) {
M
Max Reitz 已提交
5966 5967
        return -ENOTSUP;
    }
5968
    return bs->drv->bdrv_amend_options(bs, opts, status_cb);
M
Max Reitz 已提交
5969
}
5970

5971 5972 5973 5974
/* 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.
5975
 */
5976
bool bdrv_recurse_is_first_non_filter(BlockDriverState *bs,
5977
                                      BlockDriverState *candidate)
5978
{
5979 5980
    /* return false if basic checks fails */
    if (!bs || !bs->drv) {
5981
        return false;
5982 5983
    }

5984 5985 5986 5987 5988
    /* 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;
5989
    }
5990
    /* Down this path the driver is a block filter driver */
5991

5992 5993 5994 5995
    /* 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) {
5996
        return bs->drv->bdrv_recurse_is_first_non_filter(bs, candidate);
5997 5998
    }

5999 6000 6001
    /* the driver is a block filter but don't allow to recurse -> return false
     */
    return false;
6002 6003
}

6004 6005 6006 6007 6008
/* 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)
6009
{
6010 6011 6012 6013 6014 6015
    BlockDriverState *bs;

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

6016
        /* try to recurse in this top level bs */
6017
        perm = bdrv_recurse_is_first_non_filter(bs, candidate);
6018 6019 6020 6021 6022 6023 6024 6025

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

    return false;
6026
}
6027 6028 6029 6030

BlockDriverState *check_to_replace_node(const char *node_name, Error **errp)
{
    BlockDriverState *to_replace_bs = bdrv_find_node(node_name);
6031 6032
    AioContext *aio_context;

6033 6034 6035 6036 6037
    if (!to_replace_bs) {
        error_setg(errp, "Node name '%s' not found", node_name);
        return NULL;
    }

6038 6039 6040
    aio_context = bdrv_get_aio_context(to_replace_bs);
    aio_context_acquire(aio_context);

6041
    if (bdrv_op_is_blocked(to_replace_bs, BLOCK_OP_TYPE_REPLACE, errp)) {
6042 6043
        to_replace_bs = NULL;
        goto out;
6044 6045 6046 6047 6048 6049 6050 6051 6052
    }

    /* We don't want arbitrary node of the BDS chain to be replaced only the top
     * most non filter in order to prevent data corruption.
     * Another benefit is that this tests exclude backing files which are
     * blocked by the backing blockers.
     */
    if (!bdrv_is_first_non_filter(to_replace_bs)) {
        error_setg(errp, "Only top most non filter can be replaced");
6053 6054
        to_replace_bs = NULL;
        goto out;
6055 6056
    }

6057 6058
out:
    aio_context_release(aio_context);
6059 6060
    return to_replace_bs;
}
6061 6062 6063 6064 6065 6066 6067 6068 6069 6070 6071 6072 6073 6074 6075 6076 6077 6078 6079 6080 6081 6082 6083 6084 6085 6086 6087 6088 6089 6090

void bdrv_io_plug(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;
    if (drv && drv->bdrv_io_plug) {
        drv->bdrv_io_plug(bs);
    } else if (bs->file) {
        bdrv_io_plug(bs->file);
    }
}

void bdrv_io_unplug(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;
    if (drv && drv->bdrv_io_unplug) {
        drv->bdrv_io_unplug(bs);
    } else if (bs->file) {
        bdrv_io_unplug(bs->file);
    }
}

void bdrv_flush_io_queue(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;
    if (drv && drv->bdrv_flush_io_queue) {
        drv->bdrv_flush_io_queue(bs);
    } else if (bs->file) {
        bdrv_flush_io_queue(bs->file);
    }
}
M
Max Reitz 已提交
6091 6092 6093 6094 6095 6096 6097 6098 6099 6100 6101 6102 6103 6104 6105 6106 6107 6108 6109 6110 6111 6112 6113 6114 6115 6116 6117 6118 6119 6120 6121 6122 6123 6124 6125 6126 6127 6128 6129 6130 6131 6132 6133 6134 6135 6136 6137 6138 6139 6140 6141 6142 6143 6144 6145 6146 6147 6148 6149 6150 6151 6152 6153 6154 6155 6156 6157 6158 6159 6160 6161 6162 6163 6164 6165 6166 6167 6168 6169 6170 6171 6172 6173 6174 6175 6176 6177 6178 6179 6180 6181 6182 6183 6184 6185 6186 6187 6188 6189 6190 6191 6192 6193 6194 6195 6196 6197 6198 6199 6200 6201 6202 6203 6204 6205 6206 6207 6208 6209 6210 6211 6212 6213 6214 6215 6216 6217 6218 6219 6220

static bool append_open_options(QDict *d, BlockDriverState *bs)
{
    const QDictEntry *entry;
    bool found_any = false;

    for (entry = qdict_first(bs->options); entry;
         entry = qdict_next(bs->options, entry))
    {
        /* Only take options for this level and exclude all non-driver-specific
         * options */
        if (!strchr(qdict_entry_key(entry), '.') &&
            strcmp(qdict_entry_key(entry), "node-name"))
        {
            qobject_incref(qdict_entry_value(entry));
            qdict_put_obj(d, qdict_entry_key(entry), qdict_entry_value(entry));
            found_any = true;
        }
    }

    return found_any;
}

/* Updates the following BDS fields:
 *  - exact_filename: A filename which may be used for opening a block device
 *                    which (mostly) equals the given BDS (even without any
 *                    other options; so reading and writing must return the same
 *                    results, but caching etc. may be different)
 *  - full_open_options: Options which, when given when opening a block device
 *                       (without a filename), result in a BDS (mostly)
 *                       equalling the given one
 *  - filename: If exact_filename is set, it is copied here. Otherwise,
 *              full_open_options is converted to a JSON object, prefixed with
 *              "json:" (for use through the JSON pseudo protocol) and put here.
 */
void bdrv_refresh_filename(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;
    QDict *opts;

    if (!drv) {
        return;
    }

    /* This BDS's file name will most probably depend on its file's name, so
     * refresh that first */
    if (bs->file) {
        bdrv_refresh_filename(bs->file);
    }

    if (drv->bdrv_refresh_filename) {
        /* Obsolete information is of no use here, so drop the old file name
         * information before refreshing it */
        bs->exact_filename[0] = '\0';
        if (bs->full_open_options) {
            QDECREF(bs->full_open_options);
            bs->full_open_options = NULL;
        }

        drv->bdrv_refresh_filename(bs);
    } else if (bs->file) {
        /* Try to reconstruct valid information from the underlying file */
        bool has_open_options;

        bs->exact_filename[0] = '\0';
        if (bs->full_open_options) {
            QDECREF(bs->full_open_options);
            bs->full_open_options = NULL;
        }

        opts = qdict_new();
        has_open_options = append_open_options(opts, bs);

        /* If no specific options have been given for this BDS, the filename of
         * the underlying file should suffice for this one as well */
        if (bs->file->exact_filename[0] && !has_open_options) {
            strcpy(bs->exact_filename, bs->file->exact_filename);
        }
        /* Reconstructing the full options QDict is simple for most format block
         * drivers, as long as the full options are known for the underlying
         * file BDS. The full options QDict of that file BDS should somehow
         * contain a representation of the filename, therefore the following
         * suffices without querying the (exact_)filename of this BDS. */
        if (bs->file->full_open_options) {
            qdict_put_obj(opts, "driver",
                          QOBJECT(qstring_from_str(drv->format_name)));
            QINCREF(bs->file->full_open_options);
            qdict_put_obj(opts, "file", QOBJECT(bs->file->full_open_options));

            bs->full_open_options = opts;
        } else {
            QDECREF(opts);
        }
    } else if (!bs->full_open_options && qdict_size(bs->options)) {
        /* There is no underlying file BDS (at least referenced by BDS.file),
         * so the full options QDict should be equal to the options given
         * specifically for this block device when it was opened (plus the
         * driver specification).
         * Because those options don't change, there is no need to update
         * full_open_options when it's already set. */

        opts = qdict_new();
        append_open_options(opts, bs);
        qdict_put_obj(opts, "driver",
                      QOBJECT(qstring_from_str(drv->format_name)));

        if (bs->exact_filename[0]) {
            /* This may not work for all block protocol drivers (some may
             * require this filename to be parsed), but we have to find some
             * default solution here, so just include it. If some block driver
             * does not support pure options without any filename at all or
             * needs some special format of the options QDict, it needs to
             * implement the driver-specific bdrv_refresh_filename() function.
             */
            qdict_put_obj(opts, "filename",
                          QOBJECT(qstring_from_str(bs->exact_filename)));
        }

        bs->full_open_options = opts;
    }

    if (bs->exact_filename[0]) {
        pstrcpy(bs->filename, sizeof(bs->filename), bs->exact_filename);
    } else if (bs->full_open_options) {
        QString *json = qobject_to_json(QOBJECT(bs->full_open_options));
        snprintf(bs->filename, sizeof(bs->filename), "json:%s",
                 qstring_get_str(json));
        QDECREF(json);
    }
}
6221 6222 6223 6224 6225 6226 6227 6228 6229 6230 6231

/* This accessor function purpose is to allow the device models to access the
 * BlockAcctStats structure embedded inside a BlockDriverState without being
 * aware of the BlockDriverState structure layout.
 * It will go away when the BlockAcctStats structure will be moved inside
 * the device models.
 */
BlockAcctStats *bdrv_get_stats(BlockDriverState *bs)
{
    return &bs->stats;
}