block.c 118.3 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"
A
aliguori 已提交
27
#include "monitor.h"
B
bellard 已提交
28
#include "block_int.h"
29
#include "module.h"
30
#include "qjson.h"
31
#include "qemu-coroutine.h"
L
Luiz Capitulino 已提交
32
#include "qmp-commands.h"
33
#include "qemu-timer.h"
B
bellard 已提交
34

J
Juan Quintela 已提交
35
#ifdef CONFIG_BSD
B
bellard 已提交
36 37 38
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/ioctl.h>
B
Blue Swirl 已提交
39
#include <sys/queue.h>
40
#ifndef __DragonFly__
B
bellard 已提交
41 42
#include <sys/disk.h>
#endif
43
#endif
B
bellard 已提交
44

45 46 47 48
#ifdef _WIN32
#include <windows.h>
#endif

49 50
#define NOT_DONE 0x7fffffff /* used while emulated sync operation in progress */

51 52
typedef enum {
    BDRV_REQ_COPY_ON_READ = 0x1,
53
    BDRV_REQ_ZERO_WRITE   = 0x2,
54 55
} BdrvRequestFlags;

56
static void bdrv_dev_change_media_cb(BlockDriverState *bs, bool load);
57 58
static BlockDriverAIOCB *bdrv_aio_readv_em(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
59
        BlockDriverCompletionFunc *cb, void *opaque);
60 61
static BlockDriverAIOCB *bdrv_aio_writev_em(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
62
        BlockDriverCompletionFunc *cb, void *opaque);
63 64 65 66 67 68
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);
69
static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,
70 71
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags);
72
static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs,
73 74
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags);
75 76 77 78 79 80
static BlockDriverAIOCB *bdrv_co_aio_rw_vector(BlockDriverState *bs,
                                               int64_t sector_num,
                                               QEMUIOVector *qiov,
                                               int nb_sectors,
                                               BlockDriverCompletionFunc *cb,
                                               void *opaque,
81
                                               bool is_write);
82
static void coroutine_fn bdrv_co_do_rw(void *opaque);
K
Kevin Wolf 已提交
83 84
static int coroutine_fn bdrv_co_do_write_zeroes(BlockDriverState *bs,
    int64_t sector_num, int nb_sectors);
B
bellard 已提交
85

Z
Zhi Yong Wu 已提交
86 87 88 89 90 91 92
static bool bdrv_exceed_bps_limits(BlockDriverState *bs, int nb_sectors,
        bool is_write, double elapsed_time, uint64_t *wait);
static bool bdrv_exceed_iops_limits(BlockDriverState *bs, bool is_write,
        double elapsed_time, uint64_t *wait);
static bool bdrv_exceed_io_limits(BlockDriverState *bs, int nb_sectors,
        bool is_write, int64_t *wait);

93 94
static QTAILQ_HEAD(, BlockDriverState) bdrv_states =
    QTAILQ_HEAD_INITIALIZER(bdrv_states);
95

96 97
static QLIST_HEAD(, BlockDriver) bdrv_drivers =
    QLIST_HEAD_INITIALIZER(bdrv_drivers);
B
bellard 已提交
98

99 100 101
/* The device to use for VM snapshots */
static BlockDriverState *bs_snapshots;

102 103 104
/* If non-zero, use only whitelisted block drivers */
static int use_bdrv_whitelist;

105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124
#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

125
/* throttling disk I/O limits */
Z
Zhi Yong Wu 已提交
126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143
void bdrv_io_limits_disable(BlockDriverState *bs)
{
    bs->io_limits_enabled = false;

    while (qemu_co_queue_next(&bs->throttled_reqs));

    if (bs->block_timer) {
        qemu_del_timer(bs->block_timer);
        qemu_free_timer(bs->block_timer);
        bs->block_timer = NULL;
    }

    bs->slice_start = 0;
    bs->slice_end   = 0;
    bs->slice_time  = 0;
    memset(&bs->io_base, 0, sizeof(bs->io_base));
}

144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172
static void bdrv_block_timer(void *opaque)
{
    BlockDriverState *bs = opaque;

    qemu_co_queue_next(&bs->throttled_reqs);
}

void bdrv_io_limits_enable(BlockDriverState *bs)
{
    qemu_co_queue_init(&bs->throttled_reqs);
    bs->block_timer = qemu_new_timer_ns(vm_clock, bdrv_block_timer, bs);
    bs->slice_time  = 5 * BLOCK_IO_SLICE_TIME;
    bs->slice_start = qemu_get_clock_ns(vm_clock);
    bs->slice_end   = bs->slice_start + bs->slice_time;
    memset(&bs->io_base, 0, sizeof(bs->io_base));
    bs->io_limits_enabled = true;
}

bool bdrv_io_limits_enabled(BlockDriverState *bs)
{
    BlockIOLimit *io_limits = &bs->io_limits;
    return io_limits->bps[BLOCK_IO_LIMIT_READ]
         || io_limits->bps[BLOCK_IO_LIMIT_WRITE]
         || io_limits->bps[BLOCK_IO_LIMIT_TOTAL]
         || io_limits->iops[BLOCK_IO_LIMIT_READ]
         || io_limits->iops[BLOCK_IO_LIMIT_WRITE]
         || io_limits->iops[BLOCK_IO_LIMIT_TOTAL];
}

Z
Zhi Yong Wu 已提交
173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197
static void bdrv_io_limits_intercept(BlockDriverState *bs,
                                     bool is_write, int nb_sectors)
{
    int64_t wait_time = -1;

    if (!qemu_co_queue_empty(&bs->throttled_reqs)) {
        qemu_co_queue_wait(&bs->throttled_reqs);
    }

    /* In fact, we hope to keep each request's timing, in FIFO mode. The next
     * throttled requests will not be dequeued until the current request is
     * allowed to be serviced. So if the current request still exceeds the
     * limits, it will be inserted to the head. All requests followed it will
     * be still in throttled_reqs queue.
     */

    while (bdrv_exceed_io_limits(bs, nb_sectors, is_write, &wait_time)) {
        qemu_mod_timer(bs->block_timer,
                       wait_time + qemu_get_clock_ns(vm_clock));
        qemu_co_queue_wait_insert_head(&bs->throttled_reqs);
    }

    qemu_co_queue_next(&bs->throttled_reqs);
}

198 199 200 201 202 203 204 205 206 207 208 209 210
/* check if the path starts with "<protocol>:" */
static int path_has_protocol(const char *path)
{
#ifdef _WIN32
    if (is_windows_drive(path) ||
        is_windows_drive_prefix(path)) {
        return 0;
    }
#endif

    return strchr(path, ':') != NULL;
}

B
bellard 已提交
211
int path_is_absolute(const char *path)
212
{
B
bellard 已提交
213 214
#ifdef _WIN32
    /* specific case for names like: "\\.\d:" */
P
Paolo Bonzini 已提交
215
    if (is_windows_drive(path) || is_windows_drive_prefix(path)) {
B
bellard 已提交
216
        return 1;
P
Paolo Bonzini 已提交
217 218
    }
    return (*path == '/' || *path == '\\');
219
#else
P
Paolo Bonzini 已提交
220
    return (*path == '/');
221
#endif
222 223
}

B
bellard 已提交
224 225 226 227 228 229
/* 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)
230
{
B
bellard 已提交
231 232 233 234 235 236 237 238 239 240 241 242 243
    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;
244 245 246 247 248 249 250 251 252
        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
253 254 255 256 257 258 259 260 261 262 263 264
        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);
265 266 267
    }
}

268
void bdrv_register(BlockDriver *bdrv)
B
bellard 已提交
269
{
270 271
    /* Block drivers without coroutine functions need emulation */
    if (!bdrv->bdrv_co_readv) {
272 273 274
        bdrv->bdrv_co_readv = bdrv_co_readv_em;
        bdrv->bdrv_co_writev = bdrv_co_writev_em;

275 276 277
        /* 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.
         */
278 279 280 281 282
        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
283
    }
284

285
    QLIST_INSERT_HEAD(&bdrv_drivers, bdrv, list);
B
bellard 已提交
286
}
B
bellard 已提交
287 288 289 290

/* create a new block device (by default it is empty) */
BlockDriverState *bdrv_new(const char *device_name)
{
291
    BlockDriverState *bs;
B
bellard 已提交
292

293
    bs = g_malloc0(sizeof(BlockDriverState));
B
bellard 已提交
294
    pstrcpy(bs->device_name, sizeof(bs->device_name), device_name);
B
bellard 已提交
295
    if (device_name[0] != '\0') {
296
        QTAILQ_INSERT_TAIL(&bdrv_states, bs, list);
B
bellard 已提交
297
    }
298
    bdrv_iostatus_disable(bs);
B
bellard 已提交
299 300 301
    return bs;
}

B
bellard 已提交
302 303 304
BlockDriver *bdrv_find_format(const char *format_name)
{
    BlockDriver *drv1;
305 306
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
        if (!strcmp(drv1->format_name, format_name)) {
B
bellard 已提交
307
            return drv1;
308
        }
B
bellard 已提交
309 310 311 312
    }
    return NULL;
}

313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336
static int bdrv_is_whitelisted(BlockDriver *drv)
{
    static const char *whitelist[] = {
        CONFIG_BDRV_WHITELIST
    };
    const char **p;

    if (!whitelist[0])
        return 1;               /* no whitelist, anything goes */

    for (p = whitelist; *p; p++) {
        if (!strcmp(drv->format_name, *p)) {
            return 1;
        }
    }
    return 0;
}

BlockDriver *bdrv_find_whitelisted_format(const char *format_name)
{
    BlockDriver *drv = bdrv_find_format(format_name);
    return drv && bdrv_is_whitelisted(drv) ? drv : NULL;
}

337 338 339 340 341 342 343 344 345 346 347 348 349 350 351
typedef struct CreateCo {
    BlockDriver *drv;
    char *filename;
    QEMUOptionParameter *options;
    int ret;
} CreateCo;

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

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

352 353
int bdrv_create(BlockDriver *drv, const char* filename,
    QEMUOptionParameter *options)
B
bellard 已提交
354
{
355 356 357 358 359 360 361 362 363 364 365
    int ret;

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

    if (!drv->bdrv_create) {
B
bellard 已提交
366
        return -ENOTSUP;
367 368 369 370 371 372 373 374 375 376 377 378 379 380 381
    }

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

    ret = cco.ret;
    g_free(cco.filename);
382

383
    return ret;
B
bellard 已提交
384 385
}

386 387 388 389
int bdrv_create_file(const char* filename, QEMUOptionParameter *options)
{
    BlockDriver *drv;

390
    drv = bdrv_find_protocol(filename);
391
    if (drv == NULL) {
392
        return -ENOENT;
393 394 395 396 397
    }

    return bdrv_create(drv, filename, options);
}

B
bellard 已提交
398
#ifdef _WIN32
399
void get_tmp_filename(char *filename, int size)
B
bellard 已提交
400
{
401
    char temp_dir[MAX_PATH];
402

403 404
    GetTempPath(MAX_PATH, temp_dir);
    GetTempFileName(temp_dir, "qem", 0, filename);
B
bellard 已提交
405 406
}
#else
407
void get_tmp_filename(char *filename, int size)
B
bellard 已提交
408
{
B
bellard 已提交
409
    int fd;
410
    const char *tmpdir;
B
bellard 已提交
411
    /* XXX: race condition possible */
A
aurel32 已提交
412 413 414 415
    tmpdir = getenv("TMPDIR");
    if (!tmpdir)
        tmpdir = "/tmp";
    snprintf(filename, size, "%s/vl.XXXXXX", tmpdir);
B
bellard 已提交
416 417 418
    fd = mkstemp(filename);
    close(fd);
}
B
bellard 已提交
419
#endif
B
bellard 已提交
420

421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442
/*
 * 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;
}

443
BlockDriver *bdrv_find_protocol(const char *filename)
B
bellard 已提交
444 445 446
{
    BlockDriver *drv1;
    char protocol[128];
447
    int len;
B
bellard 已提交
448
    const char *p;
B
bellard 已提交
449

450 451
    /* TODO Drivers without bdrv_file_open must be specified explicitly */

452 453 454 455 456 457 458 459 460 461 462 463
    /*
     * 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;
    }

464
    if (!path_has_protocol(filename)) {
465
        return bdrv_find_format("file");
466
    }
467 468
    p = strchr(filename, ':');
    assert(p != NULL);
469 470 471 472 473
    len = p - filename;
    if (len > sizeof(protocol) - 1)
        len = sizeof(protocol) - 1;
    memcpy(protocol, filename, len);
    protocol[len] = '\0';
474
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
475
        if (drv1->protocol_name &&
476
            !strcmp(drv1->protocol_name, protocol)) {
B
bellard 已提交
477
            return drv1;
478
        }
B
bellard 已提交
479 480 481 482
    }
    return NULL;
}

483
static int find_image_format(const char *filename, BlockDriver **pdrv)
484 485 486 487 488 489
{
    int ret, score, score_max;
    BlockDriver *drv1, *drv;
    uint8_t buf[2048];
    BlockDriverState *bs;

490
    ret = bdrv_file_open(&bs, filename, 0);
491 492 493 494
    if (ret < 0) {
        *pdrv = NULL;
        return ret;
    }
495

496 497
    /* Return the raw BlockDriver * to scsi-generic devices or empty drives */
    if (bs->sg || !bdrv_is_inserted(bs)) {
498
        bdrv_delete(bs);
499 500 501 502 503 504
        drv = bdrv_find_format("raw");
        if (!drv) {
            ret = -ENOENT;
        }
        *pdrv = drv;
        return ret;
505
    }
506

B
bellard 已提交
507 508 509
    ret = bdrv_pread(bs, 0, buf, sizeof(buf));
    bdrv_delete(bs);
    if (ret < 0) {
510 511
        *pdrv = NULL;
        return ret;
B
bellard 已提交
512 513
    }

B
bellard 已提交
514
    score_max = 0;
515
    drv = NULL;
516
    QLIST_FOREACH(drv1, &bdrv_drivers, list) {
B
bellard 已提交
517 518 519 520 521 522
        if (drv1->bdrv_probe) {
            score = drv1->bdrv_probe(buf, ret, filename);
            if (score > score_max) {
                score_max = score;
                drv = drv1;
            }
B
bellard 已提交
523
        }
B
bellard 已提交
524
    }
525 526 527 528 529
    if (!drv) {
        ret = -ENOENT;
    }
    *pdrv = drv;
    return ret;
B
bellard 已提交
530 531
}

532 533 534 535 536 537 538
/**
 * Set the current 'total_sectors' value
 */
static int refresh_total_sectors(BlockDriverState *bs, int64_t hint)
{
    BlockDriver *drv = bs->drv;

539 540 541 542
    /* Do not attempt drv->bdrv_getlength() on scsi-generic devices */
    if (bs->sg)
        return 0;

543 544 545 546 547 548 549 550 551 552 553 554 555
    /* query actual device if possible, otherwise just trust the hint */
    if (drv->bdrv_getlength) {
        int64_t length = drv->bdrv_getlength(bs);
        if (length < 0) {
            return length;
        }
        hint = length >> BDRV_SECTOR_BITS;
    }

    bs->total_sectors = hint;
    return 0;
}

556 557 558 559 560 561 562 563 564 565 566
/**
 * 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;
567 568
    } else if (!strcmp(mode, "directsync")) {
        *flags |= BDRV_O_NOCACHE;
569 570 571 572 573 574 575 576 577 578 579 580 581 582
    } 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;
}

583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598
/**
 * 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--;
}

599 600 601 602 603 604 605 606 607 608
/*
 * Common part for opening disk images and files
 */
static int bdrv_open_common(BlockDriverState *bs, const char *filename,
    int flags, BlockDriver *drv)
{
    int ret, open_flags;

    assert(drv != NULL);

S
Stefan Hajnoczi 已提交
609 610
    trace_bdrv_open_common(bs, filename, flags, drv->format_name);

611
    bs->file = NULL;
612
    bs->total_sectors = 0;
613 614
    bs->encrypted = 0;
    bs->valid_key = 0;
615
    bs->sg = 0;
616
    bs->open_flags = flags;
617
    bs->growable = 0;
618 619
    bs->buffer_alignment = 512;

620 621 622 623 624
    assert(bs->copy_on_read == 0); /* bdrv_new() and bdrv_close() make it so */
    if ((flags & BDRV_O_RDWR) && (flags & BDRV_O_COPY_ON_READ)) {
        bdrv_enable_copy_on_read(bs);
    }

625
    pstrcpy(bs->filename, sizeof(bs->filename), filename);
626
    bs->backing_file[0] = '\0';
627 628 629 630 631 632

    if (use_bdrv_whitelist && !bdrv_is_whitelisted(drv)) {
        return -ENOTSUP;
    }

    bs->drv = drv;
633
    bs->opaque = g_malloc0(drv->instance_size);
634

635
    bs->enable_write_cache = !!(flags & BDRV_O_CACHE_WB);
636 637 638 639 640 641 642 643

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

    /*
S
Stefan Weil 已提交
644
     * Snapshots should be writable.
645 646 647 648 649
     */
    if (bs->is_temporary) {
        open_flags |= BDRV_O_RDWR;
    }

650 651
    bs->keep_read_only = bs->read_only = !(open_flags & BDRV_O_RDWR);

652 653 654 655 656 657 658 659 660 661
    /* Open the image, either directly or using a protocol */
    if (drv->bdrv_file_open) {
        ret = drv->bdrv_file_open(bs, filename, open_flags);
    } else {
        ret = bdrv_file_open(&bs->file, filename, open_flags);
        if (ret >= 0) {
            ret = drv->bdrv_open(bs, open_flags);
        }
    }

662 663 664 665
    if (ret < 0) {
        goto free_and_fail;
    }

666 667 668
    ret = refresh_total_sectors(bs, bs->total_sectors);
    if (ret < 0) {
        goto free_and_fail;
669
    }
670

671 672 673 674 675 676 677 678
#ifndef _WIN32
    if (bs->is_temporary) {
        unlink(filename);
    }
#endif
    return 0;

free_and_fail:
679 680 681 682
    if (bs->file) {
        bdrv_delete(bs->file);
        bs->file = NULL;
    }
683
    g_free(bs->opaque);
684 685 686 687 688
    bs->opaque = NULL;
    bs->drv = NULL;
    return ret;
}

K
Kevin Wolf 已提交
689 690 691
/*
 * Opens a file using a protocol (file, host_device, nbd, ...)
 */
B
bellard 已提交
692
int bdrv_file_open(BlockDriverState **pbs, const char *filename, int flags)
B
bellard 已提交
693
{
B
bellard 已提交
694
    BlockDriverState *bs;
695
    BlockDriver *drv;
B
bellard 已提交
696 697
    int ret;

698
    drv = bdrv_find_protocol(filename);
699 700 701 702
    if (!drv) {
        return -ENOENT;
    }

B
bellard 已提交
703
    bs = bdrv_new("");
K
Kevin Wolf 已提交
704
    ret = bdrv_open_common(bs, filename, flags, drv);
B
bellard 已提交
705 706 707
    if (ret < 0) {
        bdrv_delete(bs);
        return ret;
708
    }
709
    bs->growable = 1;
B
bellard 已提交
710 711 712 713
    *pbs = bs;
    return 0;
}

K
Kevin Wolf 已提交
714 715 716
/*
 * Opens a disk image (raw, qcow2, vmdk, ...)
 */
K
Kevin Wolf 已提交
717 718
int bdrv_open(BlockDriverState *bs, const char *filename, int flags,
              BlockDriver *drv)
B
bellard 已提交
719
{
K
Kevin Wolf 已提交
720
    int ret;
K
Kevin Wolf 已提交
721
    char tmp_filename[PATH_MAX];
B
bellard 已提交
722

B
bellard 已提交
723
    if (flags & BDRV_O_SNAPSHOT) {
B
bellard 已提交
724 725
        BlockDriverState *bs1;
        int64_t total_size;
A
aliguori 已提交
726
        int is_protocol = 0;
K
Kevin Wolf 已提交
727 728
        BlockDriver *bdrv_qcow2;
        QEMUOptionParameter *options;
K
Kevin Wolf 已提交
729
        char backing_filename[PATH_MAX];
730

B
bellard 已提交
731 732
        /* if snapshot, we create a temporary backing file and open it
           instead of opening 'filename' directly */
733

B
bellard 已提交
734 735
        /* if there is a backing file, use it */
        bs1 = bdrv_new("");
K
Kevin Wolf 已提交
736
        ret = bdrv_open(bs1, filename, 0, drv);
737
        if (ret < 0) {
B
bellard 已提交
738
            bdrv_delete(bs1);
739
            return ret;
B
bellard 已提交
740
        }
741
        total_size = bdrv_getlength(bs1) & BDRV_SECTOR_MASK;
A
aliguori 已提交
742 743 744 745

        if (bs1->drv && bs1->drv->protocol_name)
            is_protocol = 1;

B
bellard 已提交
746
        bdrv_delete(bs1);
747

B
bellard 已提交
748
        get_tmp_filename(tmp_filename, sizeof(tmp_filename));
A
aliguori 已提交
749 750 751 752 753

        /* Real path is meaningless for protocols */
        if (is_protocol)
            snprintf(backing_filename, sizeof(backing_filename),
                     "%s", filename);
754 755
        else if (!realpath(filename, backing_filename))
            return -errno;
A
aliguori 已提交
756

K
Kevin Wolf 已提交
757 758 759
        bdrv_qcow2 = bdrv_find_format("qcow2");
        options = parse_option_parameters("", bdrv_qcow2->create_options, NULL);

760
        set_option_parameter_int(options, BLOCK_OPT_SIZE, total_size);
K
Kevin Wolf 已提交
761 762 763 764 765 766 767
        set_option_parameter(options, BLOCK_OPT_BACKING_FILE, backing_filename);
        if (drv) {
            set_option_parameter(options, BLOCK_OPT_BACKING_FMT,
                drv->format_name);
        }

        ret = bdrv_create(bdrv_qcow2, tmp_filename, options);
768
        free_option_parameters(options);
769 770
        if (ret < 0) {
            return ret;
B
bellard 已提交
771
        }
K
Kevin Wolf 已提交
772

B
bellard 已提交
773
        filename = tmp_filename;
K
Kevin Wolf 已提交
774
        drv = bdrv_qcow2;
B
bellard 已提交
775 776
        bs->is_temporary = 1;
    }
B
bellard 已提交
777

K
Kevin Wolf 已提交
778
    /* Find the right image format driver */
779
    if (!drv) {
780
        ret = find_image_format(filename, &drv);
781
    }
782

783 784
    if (!drv) {
        goto unlink_and_fail;
B
bellard 已提交
785
    }
K
Kevin Wolf 已提交
786 787 788 789

    /* Open the image */
    ret = bdrv_open_common(bs, filename, flags, drv);
    if (ret < 0) {
790 791 792
        goto unlink_and_fail;
    }

K
Kevin Wolf 已提交
793 794 795 796 797 798 799
    /* If there is a backing file, use it */
    if ((flags & BDRV_O_NO_BACKING) == 0 && bs->backing_file[0] != '\0') {
        char backing_filename[PATH_MAX];
        int back_flags;
        BlockDriver *back_drv = NULL;

        bs->backing_hd = bdrv_new("");
800 801 802 803 804 805 806 807 808 809

        if (path_has_protocol(bs->backing_file)) {
            pstrcpy(backing_filename, sizeof(backing_filename),
                    bs->backing_file);
        } else {
            path_combine(backing_filename, sizeof(backing_filename),
                         filename, bs->backing_file);
        }

        if (bs->backing_format[0] != '\0') {
K
Kevin Wolf 已提交
810
            back_drv = bdrv_find_format(bs->backing_format);
811
        }
K
Kevin Wolf 已提交
812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830

        /* backing files always opened read-only */
        back_flags =
            flags & ~(BDRV_O_RDWR | BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING);

        ret = bdrv_open(bs->backing_hd, backing_filename, back_flags, back_drv);
        if (ret < 0) {
            bdrv_close(bs);
            return ret;
        }
        if (bs->is_temporary) {
            bs->backing_hd->keep_read_only = !(flags & BDRV_O_RDWR);
        } else {
            /* base image inherits from "parent" */
            bs->backing_hd->keep_read_only = bs->keep_read_only;
        }
    }

    if (!bdrv_key_required(bs)) {
831
        bdrv_dev_change_media_cb(bs, true);
K
Kevin Wolf 已提交
832 833
    }

Z
Zhi Yong Wu 已提交
834 835 836 837 838
    /* throttling disk I/O limits */
    if (bs->io_limits_enabled) {
        bdrv_io_limits_enable(bs);
    }

K
Kevin Wolf 已提交
839 840 841 842 843 844 845 846 847
    return 0;

unlink_and_fail:
    if (bs->is_temporary) {
        unlink(filename);
    }
    return ret;
}

B
bellard 已提交
848 849
void bdrv_close(BlockDriverState *bs)
{
850
    bdrv_flush(bs);
B
bellard 已提交
851
    if (bs->drv) {
852 853 854
        if (bs->job) {
            block_job_cancel_sync(bs->job);
        }
K
Kevin Wolf 已提交
855 856
        bdrv_drain_all();

857 858 859
        if (bs == bs_snapshots) {
            bs_snapshots = NULL;
        }
860
        if (bs->backing_hd) {
B
bellard 已提交
861
            bdrv_delete(bs->backing_hd);
862 863
            bs->backing_hd = NULL;
        }
B
bellard 已提交
864
        bs->drv->bdrv_close(bs);
865
        g_free(bs->opaque);
B
bellard 已提交
866 867 868 869
#ifdef _WIN32
        if (bs->is_temporary) {
            unlink(bs->filename);
        }
B
bellard 已提交
870
#endif
B
bellard 已提交
871 872
        bs->opaque = NULL;
        bs->drv = NULL;
873
        bs->copy_on_read = 0;
874 875
        bs->backing_file[0] = '\0';
        bs->backing_format[0] = '\0';
B
bellard 已提交
876

877
        if (bs->file != NULL) {
878 879
            bdrv_delete(bs->file);
            bs->file = NULL;
880 881
        }

882
        bdrv_dev_change_media_cb(bs, false);
B
bellard 已提交
883
    }
Z
Zhi Yong Wu 已提交
884 885 886 887 888

    /*throttling disk I/O limits*/
    if (bs->io_limits_enabled) {
        bdrv_io_limits_disable(bs);
    }
B
bellard 已提交
889 890
}

891 892 893 894 895 896 897 898 899
void bdrv_close_all(void)
{
    BlockDriverState *bs;

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

900 901 902 903 904
/*
 * 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.
905 906 907 908 909 910
 *
 * 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.
911 912 913 914
 */
void bdrv_drain_all(void)
{
    BlockDriverState *bs;
915 916 917 918
    bool busy;

    do {
        busy = qemu_aio_wait();
919

920 921 922 923 924 925 926 927 928 929
        /* FIXME: We do not have timer support here, so this is effectively
         * a busy wait.
         */
        QTAILQ_FOREACH(bs, &bdrv_states, list) {
            if (!qemu_co_queue_empty(&bs->throttled_reqs)) {
                qemu_co_queue_restart_all(&bs->throttled_reqs);
                busy = true;
            }
        }
    } while (busy);
930 931 932 933 934 935 936 937

    /* If requests are still pending there is a bug somewhere */
    QTAILQ_FOREACH(bs, &bdrv_states, list) {
        assert(QLIST_EMPTY(&bs->tracked_requests));
        assert(qemu_co_queue_empty(&bs->throttled_reqs));
    }
}

938 939 940 941 942 943 944 945 946 947
/* make a BlockDriverState anonymous by removing from bdrv_state list.
   Also, NULL terminate the device_name to prevent double remove */
void bdrv_make_anon(BlockDriverState *bs)
{
    if (bs->device_name[0] != '\0') {
        QTAILQ_REMOVE(&bdrv_states, bs, list);
    }
    bs->device_name[0] = '\0';
}

P
Paolo Bonzini 已提交
948 949 950 951 952 953 954
static void bdrv_rebind(BlockDriverState *bs)
{
    if (bs->drv && bs->drv->bdrv_rebind) {
        bs->drv->bdrv_rebind(bs);
    }
}

955 956 957 958 959 960 961
/*
 * 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.
 *
J
Jeff Cody 已提交
962 963
 * bs_new is required to be anonymous.
 *
964 965 966 967 968 969
 * This function does not create any image files.
 */
void bdrv_append(BlockDriverState *bs_new, BlockDriverState *bs_top)
{
    BlockDriverState tmp;

J
Jeff Cody 已提交
970 971
    /* bs_new must be anonymous */
    assert(bs_new->device_name[0] == '\0');
972 973 974 975

    tmp = *bs_new;

    /* there are some fields that need to stay on the top layer: */
P
Paolo Bonzini 已提交
976
    tmp.open_flags        = bs_top->open_flags;
977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016

    /* dev info */
    tmp.dev_ops           = bs_top->dev_ops;
    tmp.dev_opaque        = bs_top->dev_opaque;
    tmp.dev               = bs_top->dev;
    tmp.buffer_alignment  = bs_top->buffer_alignment;
    tmp.copy_on_read      = bs_top->copy_on_read;

    /* i/o timing parameters */
    tmp.slice_time        = bs_top->slice_time;
    tmp.slice_start       = bs_top->slice_start;
    tmp.slice_end         = bs_top->slice_end;
    tmp.io_limits         = bs_top->io_limits;
    tmp.io_base           = bs_top->io_base;
    tmp.throttled_reqs    = bs_top->throttled_reqs;
    tmp.block_timer       = bs_top->block_timer;
    tmp.io_limits_enabled = bs_top->io_limits_enabled;

    /* geometry */
    tmp.cyls              = bs_top->cyls;
    tmp.heads             = bs_top->heads;
    tmp.secs              = bs_top->secs;
    tmp.translation       = bs_top->translation;

    /* r/w error */
    tmp.on_read_error     = bs_top->on_read_error;
    tmp.on_write_error    = bs_top->on_write_error;

    /* i/o status */
    tmp.iostatus_enabled  = bs_top->iostatus_enabled;
    tmp.iostatus          = bs_top->iostatus;

    /* keep the same entry in bdrv_states */
    pstrcpy(tmp.device_name, sizeof(tmp.device_name), bs_top->device_name);
    tmp.list = bs_top->list;

    /* The contents of 'tmp' will become bs_top, as we are
     * swapping bs_new and bs_top contents. */
    tmp.backing_hd = bs_new;
    pstrcpy(tmp.backing_file, sizeof(tmp.backing_file), bs_top->filename);
J
Jeff Cody 已提交
1017
    bdrv_get_format(bs_top, tmp.backing_format, sizeof(tmp.backing_format));
1018 1019 1020 1021 1022

    /* swap contents of the fixed new bs and the current top */
    *bs_new = *bs_top;
    *bs_top = tmp;

J
Jeff Cody 已提交
1023 1024 1025 1026 1027 1028
    /* device_name[] was carried over from the old bs_top.  bs_new
     * shouldn't be in bdrv_states, so we need to make device_name[]
     * reflect the anonymity of bs_new
     */
    bs_new->device_name[0] = '\0';

1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043
    /* clear the copied fields in the new backing file */
    bdrv_detach_dev(bs_new, bs_new->dev);

    qemu_co_queue_init(&bs_new->throttled_reqs);
    memset(&bs_new->io_base,   0, sizeof(bs_new->io_base));
    memset(&bs_new->io_limits, 0, sizeof(bs_new->io_limits));
    bdrv_iostatus_disable(bs_new);

    /* we don't use bdrv_io_limits_disable() for this, because we don't want
     * to affect or delete the block_timer, as it has been moved to bs_top */
    bs_new->io_limits_enabled = false;
    bs_new->block_timer       = NULL;
    bs_new->slice_time        = 0;
    bs_new->slice_start       = 0;
    bs_new->slice_end         = 0;
P
Paolo Bonzini 已提交
1044 1045 1046

    bdrv_rebind(bs_new);
    bdrv_rebind(bs_top);
1047 1048
}

B
bellard 已提交
1049 1050
void bdrv_delete(BlockDriverState *bs)
{
1051
    assert(!bs->dev);
1052 1053
    assert(!bs->job);
    assert(!bs->in_use);
1054

1055
    /* remove from list, if necessary */
1056
    bdrv_make_anon(bs);
1057

B
bellard 已提交
1058
    bdrv_close(bs);
1059

1060
    assert(bs != bs_snapshots);
1061
    g_free(bs);
B
bellard 已提交
1062 1063
}

1064 1065
int bdrv_attach_dev(BlockDriverState *bs, void *dev)
/* TODO change to DeviceState *dev when all users are qdevified */
1066
{
1067
    if (bs->dev) {
1068 1069
        return -EBUSY;
    }
1070
    bs->dev = dev;
1071
    bdrv_iostatus_reset(bs);
1072 1073 1074
    return 0;
}

1075 1076
/* TODO qdevified devices don't use this, remove when devices are qdevified */
void bdrv_attach_dev_nofail(BlockDriverState *bs, void *dev)
1077
{
1078 1079 1080 1081 1082 1083 1084 1085 1086 1087
    if (bdrv_attach_dev(bs, dev) < 0) {
        abort();
    }
}

void bdrv_detach_dev(BlockDriverState *bs, void *dev)
/* TODO change to DeviceState *dev when all users are qdevified */
{
    assert(bs->dev == dev);
    bs->dev = NULL;
1088 1089
    bs->dev_ops = NULL;
    bs->dev_opaque = NULL;
1090
    bs->buffer_alignment = 512;
1091 1092
}

1093 1094
/* TODO change to return DeviceState * when all users are qdevified */
void *bdrv_get_attached_dev(BlockDriverState *bs)
1095
{
1096
    return bs->dev;
1097 1098
}

1099 1100 1101 1102 1103
void bdrv_set_dev_ops(BlockDriverState *bs, const BlockDevOps *ops,
                      void *opaque)
{
    bs->dev_ops = ops;
    bs->dev_opaque = opaque;
1104 1105 1106
    if (bdrv_dev_has_removable_media(bs) && bs == bs_snapshots) {
        bs_snapshots = NULL;
    }
1107 1108
}

1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137
void bdrv_emit_qmp_error_event(const BlockDriverState *bdrv,
                               BlockQMPEventAction action, int is_read)
{
    QObject *data;
    const char *action_str;

    switch (action) {
    case BDRV_ACTION_REPORT:
        action_str = "report";
        break;
    case BDRV_ACTION_IGNORE:
        action_str = "ignore";
        break;
    case BDRV_ACTION_STOP:
        action_str = "stop";
        break;
    default:
        abort();
    }

    data = qobject_from_jsonf("{ 'device': %s, 'action': %s, 'operation': %s }",
                              bdrv->device_name,
                              action_str,
                              is_read ? "read" : "write");
    monitor_protocol_event(QEVENT_BLOCK_IO_ERROR, data);

    qobject_decref(data);
}

1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148
static void bdrv_emit_qmp_eject_event(BlockDriverState *bs, bool ejected)
{
    QObject *data;

    data = qobject_from_jsonf("{ 'device': %s, 'tray-open': %i }",
                              bdrv_get_device_name(bs), ejected);
    monitor_protocol_event(QEVENT_DEVICE_TRAY_MOVED, data);

    qobject_decref(data);
}

1149
static void bdrv_dev_change_media_cb(BlockDriverState *bs, bool load)
1150
{
1151
    if (bs->dev_ops && bs->dev_ops->change_media_cb) {
1152
        bool tray_was_closed = !bdrv_dev_is_tray_open(bs);
1153
        bs->dev_ops->change_media_cb(bs->dev_opaque, load);
1154 1155 1156 1157 1158 1159 1160 1161
        if (tray_was_closed) {
            /* tray open */
            bdrv_emit_qmp_eject_event(bs, true);
        }
        if (load) {
            /* tray close */
            bdrv_emit_qmp_eject_event(bs, false);
        }
1162 1163 1164
    }
}

1165 1166 1167 1168 1169
bool bdrv_dev_has_removable_media(BlockDriverState *bs)
{
    return !bs->dev || (bs->dev_ops && bs->dev_ops->change_media_cb);
}

P
Paolo Bonzini 已提交
1170 1171 1172 1173 1174 1175 1176
void bdrv_dev_eject_request(BlockDriverState *bs, bool force)
{
    if (bs->dev_ops && bs->dev_ops->eject_request_cb) {
        bs->dev_ops->eject_request_cb(bs->dev_opaque, force);
    }
}

1177 1178 1179 1180 1181 1182 1183 1184
bool bdrv_dev_is_tray_open(BlockDriverState *bs)
{
    if (bs->dev_ops && bs->dev_ops->is_tray_open) {
        return bs->dev_ops->is_tray_open(bs->dev_opaque);
    }
    return false;
}

1185 1186 1187 1188
static void bdrv_dev_resize_cb(BlockDriverState *bs)
{
    if (bs->dev_ops && bs->dev_ops->resize_cb) {
        bs->dev_ops->resize_cb(bs->dev_opaque);
1189 1190 1191
    }
}

1192 1193 1194 1195 1196 1197 1198 1199
bool bdrv_dev_is_medium_locked(BlockDriverState *bs)
{
    if (bs->dev_ops && bs->dev_ops->is_medium_locked) {
        return bs->dev_ops->is_medium_locked(bs->dev_opaque);
    }
    return false;
}

A
aliguori 已提交
1200 1201 1202
/*
 * Run consistency checks on an image
 *
1203
 * Returns 0 if the check could be completed (it doesn't mean that the image is
1204
 * free of errors) or -errno when an internal error occurred. The results of the
1205
 * check are stored in res.
A
aliguori 已提交
1206
 */
1207
int bdrv_check(BlockDriverState *bs, BdrvCheckResult *res)
A
aliguori 已提交
1208 1209 1210 1211 1212
{
    if (bs->drv->bdrv_check == NULL) {
        return -ENOTSUP;
    }

1213
    memset(res, 0, sizeof(*res));
1214
    return bs->drv->bdrv_check(bs, res);
A
aliguori 已提交
1215 1216
}

1217 1218
#define COMMIT_BUF_SECTORS 2048

1219 1220 1221
/* commit COW file into the raw image */
int bdrv_commit(BlockDriverState *bs)
{
B
bellard 已提交
1222
    BlockDriver *drv = bs->drv;
1223
    BlockDriver *backing_drv;
1224 1225
    int64_t sector, total_sectors;
    int n, ro, open_flags;
1226
    int ret = 0, rw_ret = 0;
1227
    uint8_t *buf;
1228 1229
    char filename[1024];
    BlockDriverState *bs_rw, *bs_ro;
1230

B
bellard 已提交
1231 1232
    if (!drv)
        return -ENOMEDIUM;
1233 1234 1235
    
    if (!bs->backing_hd) {
        return -ENOTSUP;
1236 1237
    }

1238 1239 1240
    if (bs->backing_hd->keep_read_only) {
        return -EACCES;
    }
1241

1242 1243 1244 1245
    if (bdrv_in_use(bs) || bdrv_in_use(bs->backing_hd)) {
        return -EBUSY;
    }

1246
    backing_drv = bs->backing_hd->drv;
1247 1248 1249 1250 1251 1252 1253 1254 1255
    ro = bs->backing_hd->read_only;
    strncpy(filename, bs->backing_hd->filename, sizeof(filename));
    open_flags =  bs->backing_hd->open_flags;

    if (ro) {
        /* re-open as RW */
        bdrv_delete(bs->backing_hd);
        bs->backing_hd = NULL;
        bs_rw = bdrv_new("");
1256 1257
        rw_ret = bdrv_open(bs_rw, filename, open_flags | BDRV_O_RDWR,
            backing_drv);
1258 1259 1260 1261
        if (rw_ret < 0) {
            bdrv_delete(bs_rw);
            /* try to re-open read-only */
            bs_ro = bdrv_new("");
1262 1263
            ret = bdrv_open(bs_ro, filename, open_flags & ~BDRV_O_RDWR,
                backing_drv);
1264 1265 1266 1267 1268 1269 1270 1271 1272 1273
            if (ret < 0) {
                bdrv_delete(bs_ro);
                /* drive not functional anymore */
                bs->drv = NULL;
                return ret;
            }
            bs->backing_hd = bs_ro;
            return rw_ret;
        }
        bs->backing_hd = bs_rw;
B
bellard 已提交
1274
    }
1275

J
Jan Kiszka 已提交
1276
    total_sectors = bdrv_getlength(bs) >> BDRV_SECTOR_BITS;
1277
    buf = g_malloc(COMMIT_BUF_SECTORS * BDRV_SECTOR_SIZE);
1278 1279

    for (sector = 0; sector < total_sectors; sector += n) {
1280
        if (bdrv_is_allocated(bs, sector, COMMIT_BUF_SECTORS, &n)) {
1281 1282 1283 1284 1285 1286 1287 1288 1289 1290

            if (bdrv_read(bs, sector, buf, n) != 0) {
                ret = -EIO;
                goto ro_cleanup;
            }

            if (bdrv_write(bs->backing_hd, sector, buf, n) != 0) {
                ret = -EIO;
                goto ro_cleanup;
            }
B
bellard 已提交
1291
        }
1292
    }
1293

1294 1295 1296 1297
    if (drv->bdrv_make_empty) {
        ret = drv->bdrv_make_empty(bs);
        bdrv_flush(bs);
    }
1298

1299 1300 1301 1302 1303 1304
    /*
     * Make sure all data we wrote to the backing device is actually
     * stable on disk.
     */
    if (bs->backing_hd)
        bdrv_flush(bs->backing_hd);
1305 1306

ro_cleanup:
1307
    g_free(buf);
1308 1309 1310 1311 1312 1313

    if (ro) {
        /* re-open as RO */
        bdrv_delete(bs->backing_hd);
        bs->backing_hd = NULL;
        bs_ro = bdrv_new("");
1314 1315
        ret = bdrv_open(bs_ro, filename, open_flags & ~BDRV_O_RDWR,
            backing_drv);
1316 1317 1318 1319 1320 1321 1322 1323 1324 1325
        if (ret < 0) {
            bdrv_delete(bs_ro);
            /* drive not functional anymore */
            bs->drv = NULL;
            return ret;
        }
        bs->backing_hd = bs_ro;
        bs->backing_hd->keep_read_only = 0;
    }

1326
    return ret;
1327 1328
}

1329
int bdrv_commit_all(void)
1330 1331 1332 1333
{
    BlockDriverState *bs;

    QTAILQ_FOREACH(bs, &bdrv_states, list) {
1334 1335 1336 1337
        int ret = bdrv_commit(bs);
        if (ret < 0) {
            return ret;
        }
1338
    }
1339
    return 0;
1340 1341
}

S
Stefan Hajnoczi 已提交
1342 1343 1344 1345 1346 1347
struct BdrvTrackedRequest {
    BlockDriverState *bs;
    int64_t sector_num;
    int nb_sectors;
    bool is_write;
    QLIST_ENTRY(BdrvTrackedRequest) list;
1348
    Coroutine *co; /* owner, used for deadlock detection */
1349
    CoQueue wait_queue; /* coroutines blocked on this request */
S
Stefan Hajnoczi 已提交
1350 1351 1352 1353 1354 1355 1356 1357 1358 1359
};

/**
 * Remove an active request from the tracked requests list
 *
 * This function should be called when a tracked request is completing.
 */
static void tracked_request_end(BdrvTrackedRequest *req)
{
    QLIST_REMOVE(req, list);
1360
    qemu_co_queue_restart_all(&req->wait_queue);
S
Stefan Hajnoczi 已提交
1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375
}

/**
 * Add an active request to the tracked requests list
 */
static void tracked_request_begin(BdrvTrackedRequest *req,
                                  BlockDriverState *bs,
                                  int64_t sector_num,
                                  int nb_sectors, bool is_write)
{
    *req = (BdrvTrackedRequest){
        .bs = bs,
        .sector_num = sector_num,
        .nb_sectors = nb_sectors,
        .is_write = is_write,
1376
        .co = qemu_coroutine_self(),
S
Stefan Hajnoczi 已提交
1377 1378
    };

1379 1380
    qemu_co_queue_init(&req->wait_queue);

S
Stefan Hajnoczi 已提交
1381 1382 1383
    QLIST_INSERT_HEAD(&bs->tracked_requests, req, list);
}

S
Stefan Hajnoczi 已提交
1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404
/**
 * Round a region to cluster boundaries
 */
static void round_to_clusters(BlockDriverState *bs,
                              int64_t sector_num, int nb_sectors,
                              int64_t *cluster_sector_num,
                              int *cluster_nb_sectors)
{
    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);
    }
}

1405 1406
static bool tracked_request_overlaps(BdrvTrackedRequest *req,
                                     int64_t sector_num, int nb_sectors) {
S
Stefan Hajnoczi 已提交
1407 1408 1409 1410 1411 1412 1413 1414 1415
    /*        aaaa   bbbb */
    if (sector_num >= req->sector_num + req->nb_sectors) {
        return false;
    }
    /* bbbb   aaaa        */
    if (req->sector_num >= sector_num + nb_sectors) {
        return false;
    }
    return true;
1416 1417 1418 1419 1420 1421
}

static void coroutine_fn wait_for_overlapping_requests(BlockDriverState *bs,
        int64_t sector_num, int nb_sectors)
{
    BdrvTrackedRequest *req;
S
Stefan Hajnoczi 已提交
1422 1423
    int64_t cluster_sector_num;
    int cluster_nb_sectors;
1424 1425
    bool retry;

S
Stefan Hajnoczi 已提交
1426 1427 1428 1429 1430 1431 1432 1433 1434
    /* 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.
     */
    round_to_clusters(bs, sector_num, nb_sectors,
                      &cluster_sector_num, &cluster_nb_sectors);

1435 1436 1437
    do {
        retry = false;
        QLIST_FOREACH(req, &bs->tracked_requests, list) {
S
Stefan Hajnoczi 已提交
1438 1439
            if (tracked_request_overlaps(req, cluster_sector_num,
                                         cluster_nb_sectors)) {
1440 1441 1442 1443 1444 1445
                /* 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);

1446 1447 1448 1449 1450 1451 1452 1453
                qemu_co_queue_wait(&req->wait_queue);
                retry = true;
                break;
            }
        }
    } while (retry);
}

K
Kevin Wolf 已提交
1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465
/*
 * 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;
1466
    int ret;
K
Kevin Wolf 已提交
1467

1468 1469 1470 1471 1472
    /* Backing file format doesn't make sense without a backing file */
    if (backing_fmt && !backing_file) {
        return -EINVAL;
    }

K
Kevin Wolf 已提交
1473
    if (drv->bdrv_change_backing_file != NULL) {
1474
        ret = drv->bdrv_change_backing_file(bs, backing_file, backing_fmt);
K
Kevin Wolf 已提交
1475
    } else {
1476
        ret = -ENOTSUP;
K
Kevin Wolf 已提交
1477
    }
1478 1479 1480 1481 1482 1483

    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
1484 1485
}

1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498
static int bdrv_check_byte_request(BlockDriverState *bs, int64_t offset,
                                   size_t size)
{
    int64_t len;

    if (!bdrv_is_inserted(bs))
        return -ENOMEDIUM;

    if (bs->growable)
        return 0;

    len = bdrv_getlength(bs);

K
Kevin Wolf 已提交
1499 1500 1501 1502
    if (offset < 0)
        return -EIO;

    if ((offset > len) || (len - offset < size))
1503 1504 1505 1506 1507 1508 1509 1510
        return -EIO;

    return 0;
}

static int bdrv_check_request(BlockDriverState *bs, int64_t sector_num,
                              int nb_sectors)
{
1511 1512
    return bdrv_check_byte_request(bs, sector_num * BDRV_SECTOR_SIZE,
                                   nb_sectors * BDRV_SECTOR_SIZE);
1513 1514
}

1515 1516 1517 1518 1519 1520 1521 1522 1523 1524
typedef struct RwCo {
    BlockDriverState *bs;
    int64_t sector_num;
    int nb_sectors;
    QEMUIOVector *qiov;
    bool is_write;
    int ret;
} RwCo;

static void coroutine_fn bdrv_rw_co_entry(void *opaque)
B
bellard 已提交
1525
{
1526
    RwCo *rwco = opaque;
B
bellard 已提交
1527

1528 1529
    if (!rwco->is_write) {
        rwco->ret = bdrv_co_do_readv(rwco->bs, rwco->sector_num,
1530
                                     rwco->nb_sectors, rwco->qiov, 0);
1531 1532
    } else {
        rwco->ret = bdrv_co_do_writev(rwco->bs, rwco->sector_num,
1533
                                      rwco->nb_sectors, rwco->qiov, 0);
1534 1535
    }
}
1536

1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556
/*
 * Process a synchronous request using coroutines
 */
static int bdrv_rw_co(BlockDriverState *bs, int64_t sector_num, uint8_t *buf,
                      int nb_sectors, bool is_write)
{
    QEMUIOVector qiov;
    struct iovec iov = {
        .iov_base = (void *)buf,
        .iov_len = nb_sectors * BDRV_SECTOR_SIZE,
    };
    Coroutine *co;
    RwCo rwco = {
        .bs = bs,
        .sector_num = sector_num,
        .nb_sectors = nb_sectors,
        .qiov = &qiov,
        .is_write = is_write,
        .ret = NOT_DONE,
    };
1557

1558
    qemu_iovec_init_external(&qiov, &iov, 1);
1559

1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570
    /**
     * 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);
    }

1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582
    if (qemu_in_coroutine()) {
        /* Fast-path if already in coroutine context */
        bdrv_rw_co_entry(&rwco);
    } else {
        co = qemu_coroutine_create(bdrv_rw_co_entry);
        qemu_coroutine_enter(co, &rwco);
        while (rwco.ret == NOT_DONE) {
            qemu_aio_wait();
        }
    }
    return rwco.ret;
}
B
bellard 已提交
1583

1584 1585 1586 1587 1588
/* 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)
{
    return bdrv_rw_co(bs, sector_num, buf, nb_sectors, false);
B
bellard 已提交
1589 1590
}

1591 1592
#define BITS_PER_LONG  (sizeof(unsigned long) * 8)

1593
static void set_dirty_bitmap(BlockDriverState *bs, int64_t sector_num,
1594
                             int nb_sectors, int dirty)
1595 1596
{
    int64_t start, end;
1597
    unsigned long val, idx, bit;
1598

J
Jan Kiszka 已提交
1599
    start = sector_num / BDRV_SECTORS_PER_DIRTY_CHUNK;
1600
    end = (sector_num + nb_sectors - 1) / BDRV_SECTORS_PER_DIRTY_CHUNK;
1601 1602

    for (; start <= end; start++) {
1603 1604
        idx = start / BITS_PER_LONG;
        bit = start % BITS_PER_LONG;
1605 1606
        val = bs->dirty_bitmap[idx];
        if (dirty) {
1607
            if (!(val & (1UL << bit))) {
1608
                bs->dirty_count++;
1609
                val |= 1UL << bit;
1610
            }
1611
        } else {
1612
            if (val & (1UL << bit)) {
1613
                bs->dirty_count--;
1614
                val &= ~(1UL << bit);
1615
            }
1616 1617
        }
        bs->dirty_bitmap[idx] = val;
1618 1619 1620
    }
}

1621
/* Return < 0 if error. Important errors are:
B
bellard 已提交
1622 1623 1624 1625 1626
  -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
*/
1627
int bdrv_write(BlockDriverState *bs, int64_t sector_num,
B
bellard 已提交
1628 1629
               const uint8_t *buf, int nb_sectors)
{
1630
    return bdrv_rw_co(bs, sector_num, (uint8_t *)buf, nb_sectors, true);
B
bellard 已提交
1631 1632
}

1633 1634
int bdrv_pread(BlockDriverState *bs, int64_t offset,
               void *buf, int count1)
B
bellard 已提交
1635
{
J
Jan Kiszka 已提交
1636
    uint8_t tmp_buf[BDRV_SECTOR_SIZE];
B
bellard 已提交
1637 1638
    int len, nb_sectors, count;
    int64_t sector_num;
1639
    int ret;
B
bellard 已提交
1640 1641 1642

    count = count1;
    /* first read to align to sector start */
J
Jan Kiszka 已提交
1643
    len = (BDRV_SECTOR_SIZE - offset) & (BDRV_SECTOR_SIZE - 1);
B
bellard 已提交
1644 1645
    if (len > count)
        len = count;
J
Jan Kiszka 已提交
1646
    sector_num = offset >> BDRV_SECTOR_BITS;
B
bellard 已提交
1647
    if (len > 0) {
1648 1649
        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
J
Jan Kiszka 已提交
1650
        memcpy(buf, tmp_buf + (offset & (BDRV_SECTOR_SIZE - 1)), len);
B
bellard 已提交
1651 1652 1653 1654 1655 1656 1657 1658
        count -= len;
        if (count == 0)
            return count1;
        sector_num++;
        buf += len;
    }

    /* read the sectors "in place" */
J
Jan Kiszka 已提交
1659
    nb_sectors = count >> BDRV_SECTOR_BITS;
B
bellard 已提交
1660
    if (nb_sectors > 0) {
1661 1662
        if ((ret = bdrv_read(bs, sector_num, buf, nb_sectors)) < 0)
            return ret;
B
bellard 已提交
1663
        sector_num += nb_sectors;
J
Jan Kiszka 已提交
1664
        len = nb_sectors << BDRV_SECTOR_BITS;
B
bellard 已提交
1665 1666 1667 1668 1669 1670
        buf += len;
        count -= len;
    }

    /* add data from the last sector */
    if (count > 0) {
1671 1672
        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
B
bellard 已提交
1673 1674 1675 1676 1677
        memcpy(buf, tmp_buf, count);
    }
    return count1;
}

1678 1679
int bdrv_pwrite(BlockDriverState *bs, int64_t offset,
                const void *buf, int count1)
B
bellard 已提交
1680
{
J
Jan Kiszka 已提交
1681
    uint8_t tmp_buf[BDRV_SECTOR_SIZE];
B
bellard 已提交
1682 1683
    int len, nb_sectors, count;
    int64_t sector_num;
1684
    int ret;
B
bellard 已提交
1685 1686 1687

    count = count1;
    /* first write to align to sector start */
J
Jan Kiszka 已提交
1688
    len = (BDRV_SECTOR_SIZE - offset) & (BDRV_SECTOR_SIZE - 1);
B
bellard 已提交
1689 1690
    if (len > count)
        len = count;
J
Jan Kiszka 已提交
1691
    sector_num = offset >> BDRV_SECTOR_BITS;
B
bellard 已提交
1692
    if (len > 0) {
1693 1694
        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
J
Jan Kiszka 已提交
1695
        memcpy(tmp_buf + (offset & (BDRV_SECTOR_SIZE - 1)), buf, len);
1696 1697
        if ((ret = bdrv_write(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
B
bellard 已提交
1698 1699 1700 1701 1702 1703 1704 1705
        count -= len;
        if (count == 0)
            return count1;
        sector_num++;
        buf += len;
    }

    /* write the sectors "in place" */
J
Jan Kiszka 已提交
1706
    nb_sectors = count >> BDRV_SECTOR_BITS;
B
bellard 已提交
1707
    if (nb_sectors > 0) {
1708 1709
        if ((ret = bdrv_write(bs, sector_num, buf, nb_sectors)) < 0)
            return ret;
B
bellard 已提交
1710
        sector_num += nb_sectors;
J
Jan Kiszka 已提交
1711
        len = nb_sectors << BDRV_SECTOR_BITS;
B
bellard 已提交
1712 1713 1714 1715 1716 1717
        buf += len;
        count -= len;
    }

    /* add data from the last sector */
    if (count > 0) {
1718 1719
        if ((ret = bdrv_read(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
B
bellard 已提交
1720
        memcpy(tmp_buf, buf, count);
1721 1722
        if ((ret = bdrv_write(bs, sector_num, tmp_buf, 1)) < 0)
            return ret;
B
bellard 已提交
1723 1724 1725 1726
    }
    return count1;
}

K
Kevin Wolf 已提交
1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742
/*
 * 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;
    }

1743 1744
    /* No flush needed for cache modes that use O_DSYNC */
    if ((bs->open_flags & BDRV_O_CACHE_WB) != 0) {
K
Kevin Wolf 已提交
1745 1746 1747 1748 1749 1750
        bdrv_flush(bs);
    }

    return 0;
}

1751
static int coroutine_fn bdrv_co_do_copy_on_readv(BlockDriverState *bs,
S
Stefan Hajnoczi 已提交
1752 1753 1754 1755 1756 1757 1758 1759 1760
        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;

1761
    BlockDriver *drv = bs->drv;
S
Stefan Hajnoczi 已提交
1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774
    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.
     */
    round_to_clusters(bs, sector_num, nb_sectors,
                      &cluster_sector_num, &cluster_nb_sectors);

1775 1776
    trace_bdrv_co_do_copy_on_readv(bs, sector_num, nb_sectors,
                                   cluster_sector_num, cluster_nb_sectors);
S
Stefan Hajnoczi 已提交
1777 1778 1779 1780 1781

    iov.iov_len = cluster_nb_sectors * BDRV_SECTOR_SIZE;
    iov.iov_base = bounce_buffer = qemu_blockalign(bs, iov.iov_len);
    qemu_iovec_init_external(&bounce_qiov, &iov, 1);

1782 1783
    ret = drv->bdrv_co_readv(bs, cluster_sector_num, cluster_nb_sectors,
                             &bounce_qiov);
S
Stefan Hajnoczi 已提交
1784 1785 1786 1787
    if (ret < 0) {
        goto err;
    }

1788 1789
    if (drv->bdrv_co_write_zeroes &&
        buffer_is_zero(bounce_buffer, iov.iov_len)) {
K
Kevin Wolf 已提交
1790 1791
        ret = bdrv_co_do_write_zeroes(bs, cluster_sector_num,
                                      cluster_nb_sectors);
1792 1793
    } else {
        ret = drv->bdrv_co_writev(bs, cluster_sector_num, cluster_nb_sectors,
S
Stefan Hajnoczi 已提交
1794
                                  &bounce_qiov);
1795 1796
    }

S
Stefan Hajnoczi 已提交
1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813
    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;
    qemu_iovec_from_buffer(qiov, bounce_buffer + skip_bytes,
                           nb_sectors * BDRV_SECTOR_SIZE);

err:
    qemu_vfree(bounce_buffer);
    return ret;
}

1814 1815 1816 1817
/*
 * Handle a read request in coroutine context
 */
static int coroutine_fn bdrv_co_do_readv(BlockDriverState *bs,
1818 1819
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags)
K
Kevin Wolf 已提交
1820 1821
{
    BlockDriver *drv = bs->drv;
S
Stefan Hajnoczi 已提交
1822 1823
    BdrvTrackedRequest req;
    int ret;
K
Kevin Wolf 已提交
1824 1825 1826 1827 1828 1829 1830 1831

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

Z
Zhi Yong Wu 已提交
1832 1833 1834 1835 1836
    /* throttling disk read I/O */
    if (bs->io_limits_enabled) {
        bdrv_io_limits_intercept(bs, false, nb_sectors);
    }

1837
    if (bs->copy_on_read) {
1838 1839 1840 1841 1842 1843 1844
        flags |= BDRV_REQ_COPY_ON_READ;
    }
    if (flags & BDRV_REQ_COPY_ON_READ) {
        bs->copy_on_read_in_flight++;
    }

    if (bs->copy_on_read_in_flight) {
1845 1846 1847
        wait_for_overlapping_requests(bs, sector_num, nb_sectors);
    }

S
Stefan Hajnoczi 已提交
1848
    tracked_request_begin(&req, bs, sector_num, nb_sectors, false);
S
Stefan Hajnoczi 已提交
1849

1850
    if (flags & BDRV_REQ_COPY_ON_READ) {
S
Stefan Hajnoczi 已提交
1851 1852 1853 1854 1855 1856 1857 1858
        int pnum;

        ret = bdrv_co_is_allocated(bs, sector_num, nb_sectors, &pnum);
        if (ret < 0) {
            goto out;
        }

        if (!ret || pnum != nb_sectors) {
1859
            ret = bdrv_co_do_copy_on_readv(bs, sector_num, nb_sectors, qiov);
S
Stefan Hajnoczi 已提交
1860 1861 1862 1863
            goto out;
        }
    }

S
Stefan Hajnoczi 已提交
1864
    ret = drv->bdrv_co_readv(bs, sector_num, nb_sectors, qiov);
S
Stefan Hajnoczi 已提交
1865 1866

out:
S
Stefan Hajnoczi 已提交
1867
    tracked_request_end(&req);
1868 1869 1870 1871 1872

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

S
Stefan Hajnoczi 已提交
1873
    return ret;
K
Kevin Wolf 已提交
1874 1875
}

1876
int coroutine_fn bdrv_co_readv(BlockDriverState *bs, int64_t sector_num,
K
Kevin Wolf 已提交
1877 1878
    int nb_sectors, QEMUIOVector *qiov)
{
1879
    trace_bdrv_co_readv(bs, sector_num, nb_sectors);
K
Kevin Wolf 已提交
1880

1881 1882 1883 1884 1885 1886 1887 1888 1889 1890
    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);
1891 1892
}

1893 1894 1895 1896 1897 1898 1899 1900
static int coroutine_fn bdrv_co_do_write_zeroes(BlockDriverState *bs,
    int64_t sector_num, int nb_sectors)
{
    BlockDriver *drv = bs->drv;
    QEMUIOVector qiov;
    struct iovec iov;
    int ret;

K
Kevin Wolf 已提交
1901 1902 1903
    /* TODO Emulate only part of misaligned requests instead of letting block
     * drivers return -ENOTSUP and emulate everything */

1904 1905
    /* First try the efficient write zeroes operation */
    if (drv->bdrv_co_write_zeroes) {
K
Kevin Wolf 已提交
1906 1907 1908 1909
        ret = drv->bdrv_co_write_zeroes(bs, sector_num, nb_sectors);
        if (ret != -ENOTSUP) {
            return ret;
        }
1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923
    }

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

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

    qemu_vfree(iov.iov_base);
    return ret;
}

1924 1925 1926 1927
/*
 * Handle a write request in coroutine context
 */
static int coroutine_fn bdrv_co_do_writev(BlockDriverState *bs,
1928 1929
    int64_t sector_num, int nb_sectors, QEMUIOVector *qiov,
    BdrvRequestFlags flags)
1930 1931
{
    BlockDriver *drv = bs->drv;
S
Stefan Hajnoczi 已提交
1932
    BdrvTrackedRequest req;
1933
    int ret;
K
Kevin Wolf 已提交
1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944

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

Z
Zhi Yong Wu 已提交
1945 1946 1947 1948 1949
    /* throttling disk write I/O */
    if (bs->io_limits_enabled) {
        bdrv_io_limits_intercept(bs, true, nb_sectors);
    }

1950
    if (bs->copy_on_read_in_flight) {
1951 1952 1953
        wait_for_overlapping_requests(bs, sector_num, nb_sectors);
    }

S
Stefan Hajnoczi 已提交
1954 1955
    tracked_request_begin(&req, bs, sector_num, nb_sectors, true);

1956 1957 1958 1959 1960
    if (flags & BDRV_REQ_ZERO_WRITE) {
        ret = bdrv_co_do_write_zeroes(bs, sector_num, nb_sectors);
    } else {
        ret = drv->bdrv_co_writev(bs, sector_num, nb_sectors, qiov);
    }
1961

K
Kevin Wolf 已提交
1962 1963 1964 1965 1966 1967 1968 1969
    if (bs->dirty_bitmap) {
        set_dirty_bitmap(bs, sector_num, nb_sectors, 1);
    }

    if (bs->wr_highest_sector < sector_num + nb_sectors - 1) {
        bs->wr_highest_sector = sector_num + nb_sectors - 1;
    }

S
Stefan Hajnoczi 已提交
1970 1971
    tracked_request_end(&req);

1972
    return ret;
K
Kevin Wolf 已提交
1973 1974
}

1975 1976 1977 1978 1979
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);

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989
    return bdrv_co_do_writev(bs, sector_num, nb_sectors, qiov, 0);
}

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

    return bdrv_co_do_writev(bs, sector_num, nb_sectors, NULL,
                             BDRV_REQ_ZERO_WRITE);
1990 1991
}

B
bellard 已提交
1992 1993 1994 1995 1996 1997
/**
 * Truncate file to 'offset' bytes (needed only for file protocols)
 */
int bdrv_truncate(BlockDriverState *bs, int64_t offset)
{
    BlockDriver *drv = bs->drv;
1998
    int ret;
B
bellard 已提交
1999
    if (!drv)
B
bellard 已提交
2000
        return -ENOMEDIUM;
B
bellard 已提交
2001 2002
    if (!drv->bdrv_truncate)
        return -ENOTSUP;
2003 2004
    if (bs->read_only)
        return -EACCES;
M
Marcelo Tosatti 已提交
2005 2006
    if (bdrv_in_use(bs))
        return -EBUSY;
2007 2008 2009
    ret = drv->bdrv_truncate(bs, offset);
    if (ret == 0) {
        ret = refresh_total_sectors(bs, offset >> BDRV_SECTOR_BITS);
2010
        bdrv_dev_resize_cb(bs);
2011 2012
    }
    return ret;
B
bellard 已提交
2013 2014
}

2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033
/**
 * 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
2034 2035 2036 2037 2038 2039 2040
/**
 * Length of a file in bytes. Return < 0 if error or unknown.
 */
int64_t bdrv_getlength(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
2041
        return -ENOMEDIUM;
2042

2043
    if (bs->growable || bdrv_dev_has_removable_media(bs)) {
2044 2045 2046
        if (drv->bdrv_getlength) {
            return drv->bdrv_getlength(bs);
        }
B
bellard 已提交
2047
    }
2048
    return bs->total_sectors * BDRV_SECTOR_SIZE;
B
bellard 已提交
2049 2050
}

B
bellard 已提交
2051
/* return 0 as number of sectors if no device present or error */
2052
void bdrv_get_geometry(BlockDriverState *bs, uint64_t *nb_sectors_ptr)
B
bellard 已提交
2053
{
B
bellard 已提交
2054 2055 2056 2057 2058
    int64_t length;
    length = bdrv_getlength(bs);
    if (length < 0)
        length = 0;
    else
J
Jan Kiszka 已提交
2059
        length = length >> BDRV_SECTOR_BITS;
B
bellard 已提交
2060
    *nb_sectors_ptr = length;
B
bellard 已提交
2061
}
B
bellard 已提交
2062

2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073
struct partition {
        uint8_t boot_ind;           /* 0x80 - active */
        uint8_t head;               /* starting head */
        uint8_t sector;             /* starting sector */
        uint8_t cyl;                /* starting cylinder */
        uint8_t sys_ind;            /* What partition type */
        uint8_t end_head;           /* end head */
        uint8_t end_sector;         /* end sector */
        uint8_t end_cyl;            /* end cylinder */
        uint32_t start_sect;        /* starting sector counting from 0 */
        uint32_t nr_sects;          /* nr of sectors in partition */
2074
} QEMU_PACKED;
2075 2076 2077 2078 2079

/* try to guess the disk logical geometry from the MSDOS partition table. Return 0 if OK, -1 if could not guess */
static int guess_disk_lchs(BlockDriverState *bs,
                           int *pcylinders, int *pheads, int *psectors)
{
2080
    uint8_t buf[BDRV_SECTOR_SIZE];
2081 2082 2083
    int ret, i, heads, sectors, cylinders;
    struct partition *p;
    uint32_t nr_sects;
B
blueswir1 已提交
2084
    uint64_t nb_sectors;
2085
    bool enabled;
2086 2087 2088

    bdrv_get_geometry(bs, &nb_sectors);

2089 2090 2091 2092 2093 2094 2095
    /**
     * The function will be invoked during startup not only in sync I/O mode,
     * but also in async I/O mode. So the I/O throttling function has to
     * be disabled temporarily here, not permanently.
     */
    enabled = bs->io_limits_enabled;
    bs->io_limits_enabled = false;
2096
    ret = bdrv_read(bs, 0, buf, 1);
2097
    bs->io_limits_enabled = enabled;
2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132
    if (ret < 0)
        return -1;
    /* test msdos magic */
    if (buf[510] != 0x55 || buf[511] != 0xaa)
        return -1;
    for(i = 0; i < 4; i++) {
        p = ((struct partition *)(buf + 0x1be)) + i;
        nr_sects = le32_to_cpu(p->nr_sects);
        if (nr_sects && p->end_head) {
            /* We make the assumption that the partition terminates on
               a cylinder boundary */
            heads = p->end_head + 1;
            sectors = p->end_sector & 63;
            if (sectors == 0)
                continue;
            cylinders = nb_sectors / (heads * sectors);
            if (cylinders < 1 || cylinders > 16383)
                continue;
            *pheads = heads;
            *psectors = sectors;
            *pcylinders = cylinders;
#if 0
            printf("guessed geometry: LCHS=%d %d %d\n",
                   cylinders, heads, sectors);
#endif
            return 0;
        }
    }
    return -1;
}

void bdrv_guess_geometry(BlockDriverState *bs, int *pcyls, int *pheads, int *psecs)
{
    int translation, lba_detected = 0;
    int cylinders, heads, secs;
B
blueswir1 已提交
2133
    uint64_t nb_sectors;
2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187

    /* if a geometry hint is available, use it */
    bdrv_get_geometry(bs, &nb_sectors);
    bdrv_get_geometry_hint(bs, &cylinders, &heads, &secs);
    translation = bdrv_get_translation_hint(bs);
    if (cylinders != 0) {
        *pcyls = cylinders;
        *pheads = heads;
        *psecs = secs;
    } else {
        if (guess_disk_lchs(bs, &cylinders, &heads, &secs) == 0) {
            if (heads > 16) {
                /* if heads > 16, it means that a BIOS LBA
                   translation was active, so the default
                   hardware geometry is OK */
                lba_detected = 1;
                goto default_geometry;
            } else {
                *pcyls = cylinders;
                *pheads = heads;
                *psecs = secs;
                /* disable any translation to be in sync with
                   the logical geometry */
                if (translation == BIOS_ATA_TRANSLATION_AUTO) {
                    bdrv_set_translation_hint(bs,
                                              BIOS_ATA_TRANSLATION_NONE);
                }
            }
        } else {
        default_geometry:
            /* if no geometry, use a standard physical disk geometry */
            cylinders = nb_sectors / (16 * 63);

            if (cylinders > 16383)
                cylinders = 16383;
            else if (cylinders < 2)
                cylinders = 2;
            *pcyls = cylinders;
            *pheads = 16;
            *psecs = 63;
            if ((lba_detected == 1) && (translation == BIOS_ATA_TRANSLATION_AUTO)) {
                if ((*pcyls * *pheads) <= 131072) {
                    bdrv_set_translation_hint(bs,
                                              BIOS_ATA_TRANSLATION_LARGE);
                } else {
                    bdrv_set_translation_hint(bs,
                                              BIOS_ATA_TRANSLATION_LBA);
                }
            }
        }
        bdrv_set_geometry_hint(bs, *pcyls, *pheads, *psecs);
    }
}

2188
void bdrv_set_geometry_hint(BlockDriverState *bs,
B
bellard 已提交
2189 2190 2191 2192 2193 2194 2195
                            int cyls, int heads, int secs)
{
    bs->cyls = cyls;
    bs->heads = heads;
    bs->secs = secs;
}

B
bellard 已提交
2196 2197 2198 2199 2200
void bdrv_set_translation_hint(BlockDriverState *bs, int translation)
{
    bs->translation = translation;
}

2201
void bdrv_get_geometry_hint(BlockDriverState *bs,
B
bellard 已提交
2202 2203 2204 2205 2206 2207 2208
                            int *pcyls, int *pheads, int *psecs)
{
    *pcyls = bs->cyls;
    *pheads = bs->heads;
    *psecs = bs->secs;
}

2209 2210 2211 2212 2213 2214 2215 2216
/* throttling disk io limits */
void bdrv_set_io_limits(BlockDriverState *bs,
                        BlockIOLimit *io_limits)
{
    bs->io_limits = *io_limits;
    bs->io_limits_enabled = bdrv_io_limits_enabled(bs);
}

2217 2218 2219 2220 2221 2222
/* Recognize floppy formats */
typedef struct FDFormat {
    FDriveType drive;
    uint8_t last_sect;
    uint8_t max_track;
    uint8_t max_head;
2223
    FDriveRate rate;
2224 2225 2226 2227 2228
} FDFormat;

static const FDFormat fd_formats[] = {
    /* First entry is default format */
    /* 1.44 MB 3"1/2 floppy disks */
2229 2230 2231 2232 2233 2234 2235 2236
    { FDRIVE_DRV_144, 18, 80, 1, FDRIVE_RATE_500K, },
    { FDRIVE_DRV_144, 20, 80, 1, FDRIVE_RATE_500K, },
    { FDRIVE_DRV_144, 21, 80, 1, FDRIVE_RATE_500K, },
    { FDRIVE_DRV_144, 21, 82, 1, FDRIVE_RATE_500K, },
    { FDRIVE_DRV_144, 21, 83, 1, FDRIVE_RATE_500K, },
    { FDRIVE_DRV_144, 22, 80, 1, FDRIVE_RATE_500K, },
    { FDRIVE_DRV_144, 23, 80, 1, FDRIVE_RATE_500K, },
    { FDRIVE_DRV_144, 24, 80, 1, FDRIVE_RATE_500K, },
2237
    /* 2.88 MB 3"1/2 floppy disks */
2238 2239 2240 2241 2242
    { FDRIVE_DRV_288, 36, 80, 1, FDRIVE_RATE_1M, },
    { FDRIVE_DRV_288, 39, 80, 1, FDRIVE_RATE_1M, },
    { FDRIVE_DRV_288, 40, 80, 1, FDRIVE_RATE_1M, },
    { FDRIVE_DRV_288, 44, 80, 1, FDRIVE_RATE_1M, },
    { FDRIVE_DRV_288, 48, 80, 1, FDRIVE_RATE_1M, },
2243
    /* 720 kB 3"1/2 floppy disks */
2244 2245 2246 2247 2248 2249
    { FDRIVE_DRV_144,  9, 80, 1, FDRIVE_RATE_250K, },
    { FDRIVE_DRV_144, 10, 80, 1, FDRIVE_RATE_250K, },
    { FDRIVE_DRV_144, 10, 82, 1, FDRIVE_RATE_250K, },
    { FDRIVE_DRV_144, 10, 83, 1, FDRIVE_RATE_250K, },
    { FDRIVE_DRV_144, 13, 80, 1, FDRIVE_RATE_250K, },
    { FDRIVE_DRV_144, 14, 80, 1, FDRIVE_RATE_250K, },
2250
    /* 1.2 MB 5"1/4 floppy disks */
2251 2252 2253 2254 2255
    { FDRIVE_DRV_120, 15, 80, 1, FDRIVE_RATE_500K, },
    { FDRIVE_DRV_120, 18, 80, 1, FDRIVE_RATE_500K, },
    { FDRIVE_DRV_120, 18, 82, 1, FDRIVE_RATE_500K, },
    { FDRIVE_DRV_120, 18, 83, 1, FDRIVE_RATE_500K, },
    { FDRIVE_DRV_120, 20, 80, 1, FDRIVE_RATE_500K, },
2256
    /* 720 kB 5"1/4 floppy disks */
2257 2258
    { FDRIVE_DRV_120,  9, 80, 1, FDRIVE_RATE_250K, },
    { FDRIVE_DRV_120, 11, 80, 1, FDRIVE_RATE_250K, },
2259
    /* 360 kB 5"1/4 floppy disks */
2260 2261 2262 2263
    { FDRIVE_DRV_120,  9, 40, 1, FDRIVE_RATE_300K, },
    { FDRIVE_DRV_120,  9, 40, 0, FDRIVE_RATE_300K, },
    { FDRIVE_DRV_120, 10, 41, 1, FDRIVE_RATE_300K, },
    { FDRIVE_DRV_120, 10, 42, 1, FDRIVE_RATE_300K, },
2264
    /* 320 kB 5"1/4 floppy disks */
2265 2266
    { FDRIVE_DRV_120,  8, 40, 1, FDRIVE_RATE_250K, },
    { FDRIVE_DRV_120,  8, 40, 0, FDRIVE_RATE_250K, },
2267
    /* 360 kB must match 5"1/4 better than 3"1/2... */
2268
    { FDRIVE_DRV_144,  9, 80, 0, FDRIVE_RATE_250K, },
2269
    /* end */
2270
    { FDRIVE_DRV_NONE, -1, -1, 0, 0, },
2271 2272 2273 2274
};

void bdrv_get_floppy_geometry_hint(BlockDriverState *bs, int *nb_heads,
                                   int *max_track, int *last_sect,
2275 2276
                                   FDriveType drive_in, FDriveType *drive,
                                   FDriveRate *rate)
2277 2278 2279 2280 2281 2282 2283 2284
{
    const FDFormat *parse;
    uint64_t nb_sectors, size;
    int i, first_match, match;

    bdrv_get_geometry_hint(bs, nb_heads, max_track, last_sect);
    if (*nb_heads != 0 && *max_track != 0 && *last_sect != 0) {
        /* User defined disk */
2285
        *rate = FDRIVE_RATE_500K;
2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319
    } else {
        bdrv_get_geometry(bs, &nb_sectors);
        match = -1;
        first_match = -1;
        for (i = 0; ; i++) {
            parse = &fd_formats[i];
            if (parse->drive == FDRIVE_DRV_NONE) {
                break;
            }
            if (drive_in == parse->drive ||
                drive_in == FDRIVE_DRV_NONE) {
                size = (parse->max_head + 1) * parse->max_track *
                    parse->last_sect;
                if (nb_sectors == size) {
                    match = i;
                    break;
                }
                if (first_match == -1) {
                    first_match = i;
                }
            }
        }
        if (match == -1) {
            if (first_match == -1) {
                match = 1;
            } else {
                match = first_match;
            }
            parse = &fd_formats[match];
        }
        *nb_heads = parse->max_head + 1;
        *max_track = parse->max_track;
        *last_sect = parse->last_sect;
        *drive = parse->drive;
2320
        *rate = parse->rate;
2321 2322 2323
    }
}

B
bellard 已提交
2324 2325 2326 2327 2328
int bdrv_get_translation_hint(BlockDriverState *bs)
{
    return bs->translation;
}

2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340
void bdrv_set_on_error(BlockDriverState *bs, BlockErrorAction on_read_error,
                       BlockErrorAction on_write_error)
{
    bs->on_read_error = on_read_error;
    bs->on_write_error = on_write_error;
}

BlockErrorAction bdrv_get_on_error(BlockDriverState *bs, int is_read)
{
    return is_read ? bs->on_read_error : bs->on_write_error;
}

B
bellard 已提交
2341 2342 2343 2344 2345
int bdrv_is_read_only(BlockDriverState *bs)
{
    return bs->read_only;
}

2346 2347 2348 2349 2350
int bdrv_is_sg(BlockDriverState *bs)
{
    return bs->sg;
}

2351 2352 2353 2354 2355
int bdrv_enable_write_cache(BlockDriverState *bs)
{
    return bs->enable_write_cache;
}

B
bellard 已提交
2356 2357 2358 2359 2360 2361 2362
int bdrv_is_encrypted(BlockDriverState *bs)
{
    if (bs->backing_hd && bs->backing_hd->encrypted)
        return 1;
    return bs->encrypted;
}

2363 2364 2365 2366 2367 2368 2369 2370 2371
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
2372 2373 2374 2375 2376 2377 2378 2379 2380 2381
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;
    }
2382 2383 2384 2385 2386
    if (!bs->encrypted) {
        return -EINVAL;
    } else if (!bs->drv || !bs->drv->bdrv_set_key) {
        return -ENOMEDIUM;
    }
2387
    ret = bs->drv->bdrv_set_key(bs, key);
2388 2389 2390 2391 2392
    if (ret < 0) {
        bs->valid_key = 0;
    } else if (!bs->valid_key) {
        bs->valid_key = 1;
        /* call the change callback now, we skipped it on open */
2393
        bdrv_dev_change_media_cb(bs, true);
2394
    }
2395
    return ret;
B
bellard 已提交
2396 2397 2398 2399
}

void bdrv_get_format(BlockDriverState *bs, char *buf, int buf_size)
{
B
bellard 已提交
2400
    if (!bs->drv) {
B
bellard 已提交
2401 2402 2403 2404 2405 2406
        buf[0] = '\0';
    } else {
        pstrcpy(buf, buf_size, bs->drv->format_name);
    }
}

2407
void bdrv_iterate_format(void (*it)(void *opaque, const char *name),
B
bellard 已提交
2408 2409 2410 2411
                         void *opaque)
{
    BlockDriver *drv;

2412
    QLIST_FOREACH(drv, &bdrv_drivers, list) {
B
bellard 已提交
2413 2414 2415 2416
        it(opaque, drv->format_name);
    }
}

B
bellard 已提交
2417 2418 2419 2420
BlockDriverState *bdrv_find(const char *name)
{
    BlockDriverState *bs;

2421 2422
    QTAILQ_FOREACH(bs, &bdrv_states, list) {
        if (!strcmp(name, bs->device_name)) {
B
bellard 已提交
2423
            return bs;
2424
        }
B
bellard 已提交
2425 2426 2427 2428
    }
    return NULL;
}

M
Markus Armbruster 已提交
2429 2430 2431 2432 2433 2434 2435 2436
BlockDriverState *bdrv_next(BlockDriverState *bs)
{
    if (!bs) {
        return QTAILQ_FIRST(&bdrv_states);
    }
    return QTAILQ_NEXT(bs, list);
}

2437
void bdrv_iterate(void (*it)(void *opaque, BlockDriverState *bs), void *opaque)
B
bellard 已提交
2438 2439 2440
{
    BlockDriverState *bs;

2441
    QTAILQ_FOREACH(bs, &bdrv_states, list) {
2442
        it(opaque, bs);
B
bellard 已提交
2443 2444 2445
    }
}

B
bellard 已提交
2446 2447 2448 2449 2450
const char *bdrv_get_device_name(BlockDriverState *bs)
{
    return bs->device_name;
}

A
aliguori 已提交
2451 2452 2453 2454
void bdrv_flush_all(void)
{
    BlockDriverState *bs;

2455
    QTAILQ_FOREACH(bs, &bdrv_states, list) {
2456
        bdrv_flush(bs);
2457
    }
A
aliguori 已提交
2458 2459
}

K
Kevin Wolf 已提交
2460 2461 2462 2463
int bdrv_has_zero_init(BlockDriverState *bs)
{
    assert(bs->drv);

K
Kevin Wolf 已提交
2464 2465
    if (bs->drv->bdrv_has_zero_init) {
        return bs->drv->bdrv_has_zero_init(bs);
K
Kevin Wolf 已提交
2466 2467 2468 2469 2470
    }

    return 1;
}

2471 2472 2473 2474 2475 2476 2477 2478 2479
typedef struct BdrvCoIsAllocatedData {
    BlockDriverState *bs;
    int64_t sector_num;
    int nb_sectors;
    int *pnum;
    int ret;
    bool done;
} BdrvCoIsAllocatedData;

2480 2481 2482 2483 2484
/*
 * Returns true iff the specified sector is present in the disk image. Drivers
 * not implementing the functionality are assumed to not support backing files,
 * hence all their sectors are reported as allocated.
 *
2485 2486 2487
 * If 'sector_num' is beyond the end of the disk image the return value is 0
 * and 'pnum' is set to 0.
 *
2488 2489 2490 2491
 * '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.
 *
2492 2493
 * '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.
2494
 */
2495 2496
int coroutine_fn bdrv_co_is_allocated(BlockDriverState *bs, int64_t sector_num,
                                      int nb_sectors, int *pnum)
2497
{
2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509
    int64_t n;

    if (sector_num >= bs->total_sectors) {
        *pnum = 0;
        return 0;
    }

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

2510
    if (!bs->drv->bdrv_co_is_allocated) {
2511
        *pnum = nb_sectors;
2512 2513
        return 1;
    }
2514

2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536
    return bs->drv->bdrv_co_is_allocated(bs, sector_num, nb_sectors, pnum);
}

/* Coroutine wrapper for bdrv_is_allocated() */
static void coroutine_fn bdrv_is_allocated_co_entry(void *opaque)
{
    BdrvCoIsAllocatedData *data = opaque;
    BlockDriverState *bs = data->bs;

    data->ret = bdrv_co_is_allocated(bs, data->sector_num, data->nb_sectors,
                                     data->pnum);
    data->done = true;
}

/*
 * Synchronous wrapper around bdrv_co_is_allocated().
 *
 * See bdrv_co_is_allocated() for details.
 */
int bdrv_is_allocated(BlockDriverState *bs, int64_t sector_num, int nb_sectors,
                      int *pnum)
{
2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551
    Coroutine *co;
    BdrvCoIsAllocatedData data = {
        .bs = bs,
        .sector_num = sector_num,
        .nb_sectors = nb_sectors,
        .pnum = pnum,
        .done = false,
    };

    co = qemu_coroutine_create(bdrv_is_allocated_co_entry);
    qemu_coroutine_enter(co, &data);
    while (!data.done) {
        qemu_aio_wait();
    }
    return data.ret;
2552 2553
}

L
Luiz Capitulino 已提交
2554
BlockInfoList *qmp_query_block(Error **errp)
B
bellard 已提交
2555
{
L
Luiz Capitulino 已提交
2556
    BlockInfoList *head = NULL, *cur_item = NULL;
B
bellard 已提交
2557 2558
    BlockDriverState *bs;

2559
    QTAILQ_FOREACH(bs, &bdrv_states, list) {
L
Luiz Capitulino 已提交
2560
        BlockInfoList *info = g_malloc0(sizeof(*info));
2561

L
Luiz Capitulino 已提交
2562 2563 2564 2565 2566
        info->value = g_malloc0(sizeof(*info->value));
        info->value->device = g_strdup(bs->device_name);
        info->value->type = g_strdup("unknown");
        info->value->locked = bdrv_dev_is_medium_locked(bs);
        info->value->removable = bdrv_dev_has_removable_media(bs);
2567

2568
        if (bdrv_dev_has_removable_media(bs)) {
L
Luiz Capitulino 已提交
2569 2570
            info->value->has_tray_open = true;
            info->value->tray_open = bdrv_dev_is_tray_open(bs);
2571
        }
2572 2573

        if (bdrv_iostatus_is_enabled(bs)) {
L
Luiz Capitulino 已提交
2574 2575
            info->value->has_io_status = true;
            info->value->io_status = bs->iostatus;
2576 2577
        }

B
bellard 已提交
2578
        if (bs->drv) {
L
Luiz Capitulino 已提交
2579 2580 2581 2582 2583 2584 2585 2586 2587
            info->value->has_inserted = true;
            info->value->inserted = g_malloc0(sizeof(*info->value->inserted));
            info->value->inserted->file = g_strdup(bs->filename);
            info->value->inserted->ro = bs->read_only;
            info->value->inserted->drv = g_strdup(bs->drv->format_name);
            info->value->inserted->encrypted = bs->encrypted;
            if (bs->backing_file[0]) {
                info->value->inserted->has_backing_file = true;
                info->value->inserted->backing_file = g_strdup(bs->backing_file);
A
aliguori 已提交
2588
            }
Z
Zhi Yong Wu 已提交
2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603

            if (bs->io_limits_enabled) {
                info->value->inserted->bps =
                               bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL];
                info->value->inserted->bps_rd =
                               bs->io_limits.bps[BLOCK_IO_LIMIT_READ];
                info->value->inserted->bps_wr =
                               bs->io_limits.bps[BLOCK_IO_LIMIT_WRITE];
                info->value->inserted->iops =
                               bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL];
                info->value->inserted->iops_rd =
                               bs->io_limits.iops[BLOCK_IO_LIMIT_READ];
                info->value->inserted->iops_wr =
                               bs->io_limits.iops[BLOCK_IO_LIMIT_WRITE];
            }
L
Luiz Capitulino 已提交
2604
        }
2605

L
Luiz Capitulino 已提交
2606 2607 2608 2609 2610 2611
        /* XXX: waiting for the qapi to support GSList */
        if (!cur_item) {
            head = cur_item = info;
        } else {
            cur_item->next = info;
            cur_item = info;
B
bellard 已提交
2612 2613
        }
    }
2614

L
Luiz Capitulino 已提交
2615
    return head;
B
bellard 已提交
2616
}
2617

L
Luiz Capitulino 已提交
2618 2619 2620 2621 2622 2623 2624 2625 2626 2627
/* Consider exposing this as a full fledged QMP command */
static BlockStats *qmp_query_blockstat(const BlockDriverState *bs, Error **errp)
{
    BlockStats *s;

    s = g_malloc0(sizeof(*s));

    if (bs->device_name[0]) {
        s->has_device = true;
        s->device = g_strdup(bs->device_name);
2628 2629
    }

L
Luiz Capitulino 已提交
2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640
    s->stats = g_malloc0(sizeof(*s->stats));
    s->stats->rd_bytes = bs->nr_bytes[BDRV_ACCT_READ];
    s->stats->wr_bytes = bs->nr_bytes[BDRV_ACCT_WRITE];
    s->stats->rd_operations = bs->nr_ops[BDRV_ACCT_READ];
    s->stats->wr_operations = bs->nr_ops[BDRV_ACCT_WRITE];
    s->stats->wr_highest_offset = bs->wr_highest_sector * BDRV_SECTOR_SIZE;
    s->stats->flush_operations = bs->nr_ops[BDRV_ACCT_FLUSH];
    s->stats->wr_total_time_ns = bs->total_time_ns[BDRV_ACCT_WRITE];
    s->stats->rd_total_time_ns = bs->total_time_ns[BDRV_ACCT_READ];
    s->stats->flush_total_time_ns = bs->total_time_ns[BDRV_ACCT_FLUSH];

2641
    if (bs->file) {
L
Luiz Capitulino 已提交
2642 2643
        s->has_parent = true;
        s->parent = qmp_query_blockstat(bs->file, NULL);
2644 2645
    }

L
Luiz Capitulino 已提交
2646
    return s;
2647 2648
}

L
Luiz Capitulino 已提交
2649
BlockStatsList *qmp_query_blockstats(Error **errp)
2650
{
L
Luiz Capitulino 已提交
2651
    BlockStatsList *head = NULL, *cur_item = NULL;
2652 2653
    BlockDriverState *bs;

2654
    QTAILQ_FOREACH(bs, &bdrv_states, list) {
L
Luiz Capitulino 已提交
2655 2656 2657 2658 2659 2660 2661 2662 2663 2664
        BlockStatsList *info = g_malloc0(sizeof(*info));
        info->value = qmp_query_blockstat(bs, NULL);

        /* XXX: waiting for the qapi to support GSList */
        if (!cur_item) {
            head = cur_item = info;
        } else {
            cur_item->next = info;
            cur_item = info;
        }
2665
    }
2666

L
Luiz Capitulino 已提交
2667
    return head;
2668
}
B
bellard 已提交
2669

2670 2671 2672 2673 2674 2675 2676 2677 2678 2679
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;
}

2680
void bdrv_get_backing_filename(BlockDriverState *bs,
B
bellard 已提交
2681 2682
                               char *filename, int filename_size)
{
K
Kevin Wolf 已提交
2683
    pstrcpy(filename, filename_size, bs->backing_file);
B
bellard 已提交
2684 2685
}

2686
int bdrv_write_compressed(BlockDriverState *bs, int64_t sector_num,
B
bellard 已提交
2687 2688 2689 2690
                          const uint8_t *buf, int nb_sectors)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
2691
        return -ENOMEDIUM;
B
bellard 已提交
2692 2693
    if (!drv->bdrv_write_compressed)
        return -ENOTSUP;
K
Kevin Wolf 已提交
2694 2695
    if (bdrv_check_request(bs, sector_num, nb_sectors))
        return -EIO;
2696

2697
    if (bs->dirty_bitmap) {
2698 2699
        set_dirty_bitmap(bs, sector_num, nb_sectors, 1);
    }
2700

B
bellard 已提交
2701 2702
    return drv->bdrv_write_compressed(bs, sector_num, buf, nb_sectors);
}
2703

B
bellard 已提交
2704 2705 2706 2707
int bdrv_get_info(BlockDriverState *bs, BlockDriverInfo *bdi)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
2708
        return -ENOMEDIUM;
B
bellard 已提交
2709 2710 2711 2712 2713 2714
    if (!drv->bdrv_get_info)
        return -ENOTSUP;
    memset(bdi, 0, sizeof(*bdi));
    return drv->bdrv_get_info(bs, bdi);
}

2715 2716
int bdrv_save_vmstate(BlockDriverState *bs, const uint8_t *buf,
                      int64_t pos, int size)
2717 2718 2719 2720
{
    BlockDriver *drv = bs->drv;
    if (!drv)
        return -ENOMEDIUM;
2721 2722 2723 2724 2725
    if (drv->bdrv_save_vmstate)
        return drv->bdrv_save_vmstate(bs, buf, pos, size);
    if (bs->file)
        return bdrv_save_vmstate(bs->file, buf, pos, size);
    return -ENOTSUP;
2726 2727
}

2728 2729
int bdrv_load_vmstate(BlockDriverState *bs, uint8_t *buf,
                      int64_t pos, int size)
2730 2731 2732 2733
{
    BlockDriver *drv = bs->drv;
    if (!drv)
        return -ENOMEDIUM;
2734 2735 2736 2737 2738
    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;
2739 2740
}

K
Kevin Wolf 已提交
2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752
void bdrv_debug_event(BlockDriverState *bs, BlkDebugEvent event)
{
    BlockDriver *drv = bs->drv;

    if (!drv || !drv->bdrv_debug_event) {
        return;
    }

    return drv->bdrv_debug_event(bs, event);

}

B
bellard 已提交
2753 2754 2755
/**************************************************************/
/* handling of snapshots */

2756 2757 2758
int bdrv_can_snapshot(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;
2759
    if (!drv || !bdrv_is_inserted(bs) || bdrv_is_read_only(bs)) {
2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772
        return 0;
    }

    if (!drv->bdrv_snapshot_create) {
        if (bs->file != NULL) {
            return bdrv_can_snapshot(bs->file);
        }
        return 0;
    }

    return 1;
}

2773 2774 2775 2776 2777
int bdrv_is_snapshot(BlockDriverState *bs)
{
    return !!(bs->open_flags & BDRV_O_SNAPSHOT);
}

2778 2779 2780 2781
BlockDriverState *bdrv_snapshots(void)
{
    BlockDriverState *bs;

2782
    if (bs_snapshots) {
2783
        return bs_snapshots;
2784
    }
2785 2786 2787 2788

    bs = NULL;
    while ((bs = bdrv_next(bs))) {
        if (bdrv_can_snapshot(bs)) {
2789 2790
            bs_snapshots = bs;
            return bs;
2791 2792 2793 2794 2795
        }
    }
    return NULL;
}

2796
int bdrv_snapshot_create(BlockDriverState *bs,
B
bellard 已提交
2797 2798 2799 2800
                         QEMUSnapshotInfo *sn_info)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
2801
        return -ENOMEDIUM;
2802 2803 2804 2805 2806
    if (drv->bdrv_snapshot_create)
        return drv->bdrv_snapshot_create(bs, sn_info);
    if (bs->file)
        return bdrv_snapshot_create(bs->file, sn_info);
    return -ENOTSUP;
B
bellard 已提交
2807 2808
}

2809
int bdrv_snapshot_goto(BlockDriverState *bs,
B
bellard 已提交
2810 2811 2812
                       const char *snapshot_id)
{
    BlockDriver *drv = bs->drv;
2813 2814
    int ret, open_ret;

B
bellard 已提交
2815
    if (!drv)
B
bellard 已提交
2816
        return -ENOMEDIUM;
2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832
    if (drv->bdrv_snapshot_goto)
        return drv->bdrv_snapshot_goto(bs, snapshot_id);

    if (bs->file) {
        drv->bdrv_close(bs);
        ret = bdrv_snapshot_goto(bs->file, snapshot_id);
        open_ret = drv->bdrv_open(bs, bs->open_flags);
        if (open_ret < 0) {
            bdrv_delete(bs->file);
            bs->drv = NULL;
            return open_ret;
        }
        return ret;
    }

    return -ENOTSUP;
B
bellard 已提交
2833 2834 2835 2836 2837 2838
}

int bdrv_snapshot_delete(BlockDriverState *bs, const char *snapshot_id)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
2839
        return -ENOMEDIUM;
2840 2841 2842 2843 2844
    if (drv->bdrv_snapshot_delete)
        return drv->bdrv_snapshot_delete(bs, snapshot_id);
    if (bs->file)
        return bdrv_snapshot_delete(bs->file, snapshot_id);
    return -ENOTSUP;
B
bellard 已提交
2845 2846
}

2847
int bdrv_snapshot_list(BlockDriverState *bs,
B
bellard 已提交
2848 2849 2850 2851
                       QEMUSnapshotInfo **psn_info)
{
    BlockDriver *drv = bs->drv;
    if (!drv)
B
bellard 已提交
2852
        return -ENOMEDIUM;
2853 2854 2855 2856 2857
    if (drv->bdrv_snapshot_list)
        return drv->bdrv_snapshot_list(bs, psn_info);
    if (bs->file)
        return bdrv_snapshot_list(bs->file, psn_info);
    return -ENOTSUP;
B
bellard 已提交
2858 2859
}

E
edison 已提交
2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875
int bdrv_snapshot_load_tmp(BlockDriverState *bs,
        const char *snapshot_name)
{
    BlockDriver *drv = bs->drv;
    if (!drv) {
        return -ENOMEDIUM;
    }
    if (!bs->read_only) {
        return -EINVAL;
    }
    if (drv->bdrv_snapshot_load_tmp) {
        return drv->bdrv_snapshot_load_tmp(bs, snapshot_name);
    }
    return -ENOTSUP;
}

2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893
BlockDriverState *bdrv_find_backing_image(BlockDriverState *bs,
        const char *backing_file)
{
    if (!bs->drv) {
        return NULL;
    }

    if (bs->backing_hd) {
        if (strcmp(bs->backing_file, backing_file) == 0) {
            return bs->backing_hd;
        } else {
            return bdrv_find_backing_image(bs->backing_hd, backing_file);
        }
    }

    return NULL;
}

B
bellard 已提交
2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907
#define NB_SUFFIXES 4

char *get_human_readable_size(char *buf, int buf_size, int64_t size)
{
    static const char suffixes[NB_SUFFIXES] = "KMGT";
    int64_t base;
    int i;

    if (size <= 999) {
        snprintf(buf, buf_size, "%" PRId64, size);
    } else {
        base = 1024;
        for(i = 0; i < NB_SUFFIXES; i++) {
            if (size < (10 * base)) {
2908
                snprintf(buf, buf_size, "%0.1f%c",
B
bellard 已提交
2909 2910 2911 2912
                         (double)size / base,
                         suffixes[i]);
                break;
            } else if (size < (1000 * base) || i == (NB_SUFFIXES - 1)) {
2913
                snprintf(buf, buf_size, "%" PRId64 "%c",
B
bellard 已提交
2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926
                         ((size + (base >> 1)) / base),
                         suffixes[i]);
                break;
            }
            base = base * 1024;
        }
    }
    return buf;
}

char *bdrv_snapshot_dump(char *buf, int buf_size, QEMUSnapshotInfo *sn)
{
    char buf1[128], date_buf[128], clock_buf[128];
2927 2928 2929
#ifdef _WIN32
    struct tm *ptm;
#else
B
bellard 已提交
2930
    struct tm tm;
2931
#endif
B
bellard 已提交
2932 2933 2934 2935
    time_t ti;
    int64_t secs;

    if (!sn) {
2936 2937
        snprintf(buf, buf_size,
                 "%-10s%-20s%7s%20s%15s",
B
bellard 已提交
2938 2939 2940
                 "ID", "TAG", "VM SIZE", "DATE", "VM CLOCK");
    } else {
        ti = sn->date_sec;
2941 2942 2943 2944 2945
#ifdef _WIN32
        ptm = localtime(&ti);
        strftime(date_buf, sizeof(date_buf),
                 "%Y-%m-%d %H:%M:%S", ptm);
#else
B
bellard 已提交
2946 2947 2948
        localtime_r(&ti, &tm);
        strftime(date_buf, sizeof(date_buf),
                 "%Y-%m-%d %H:%M:%S", &tm);
2949
#endif
B
bellard 已提交
2950 2951 2952 2953 2954
        secs = sn->vm_clock_nsec / 1000000000;
        snprintf(clock_buf, sizeof(clock_buf),
                 "%02d:%02d:%02d.%03d",
                 (int)(secs / 3600),
                 (int)((secs / 60) % 60),
2955
                 (int)(secs % 60),
B
bellard 已提交
2956 2957
                 (int)((sn->vm_clock_nsec / 1000000) % 1000));
        snprintf(buf, buf_size,
2958
                 "%-10s%-20s%7s%20s%15s",
B
bellard 已提交
2959 2960 2961 2962 2963 2964 2965 2966
                 sn->id_str, sn->name,
                 get_human_readable_size(buf1, sizeof(buf1), sn->vm_state_size),
                 date_buf,
                 clock_buf);
    }
    return buf;
}

B
bellard 已提交
2967
/**************************************************************/
B
bellard 已提交
2968
/* async I/Os */
B
bellard 已提交
2969

2970
BlockDriverAIOCB *bdrv_aio_readv(BlockDriverState *bs, int64_t sector_num,
2971
                                 QEMUIOVector *qiov, int nb_sectors,
2972
                                 BlockDriverCompletionFunc *cb, void *opaque)
B
bellard 已提交
2973
{
2974 2975
    trace_bdrv_aio_readv(bs, sector_num, nb_sectors, opaque);

2976
    return bdrv_co_aio_rw_vector(bs, sector_num, qiov, nb_sectors,
2977
                                 cb, opaque, false);
B
bellard 已提交
2978 2979
}

2980 2981 2982
BlockDriverAIOCB *bdrv_aio_writev(BlockDriverState *bs, int64_t sector_num,
                                  QEMUIOVector *qiov, int nb_sectors,
                                  BlockDriverCompletionFunc *cb, void *opaque)
B
bellard 已提交
2983
{
2984 2985
    trace_bdrv_aio_writev(bs, sector_num, nb_sectors, opaque);

2986
    return bdrv_co_aio_rw_vector(bs, sector_num, qiov, nb_sectors,
2987
                                 cb, opaque, true);
B
bellard 已提交
2988 2989
}

K
Kevin Wolf 已提交
2990 2991 2992 2993 2994 2995 2996 2997 2998 2999 3000 3001 3002 3003 3004 3005 3006 3007

typedef struct MultiwriteCB {
    int error;
    int num_requests;
    int num_callbacks;
    struct {
        BlockDriverCompletionFunc *cb;
        void *opaque;
        QEMUIOVector *free_qiov;
    } callbacks[];
} MultiwriteCB;

static void multiwrite_user_cb(MultiwriteCB *mcb)
{
    int i;

    for (i = 0; i < mcb->num_callbacks; i++) {
        mcb->callbacks[i].cb(mcb->callbacks[i].opaque, mcb->error);
3008 3009 3010
        if (mcb->callbacks[i].free_qiov) {
            qemu_iovec_destroy(mcb->callbacks[i].free_qiov);
        }
3011
        g_free(mcb->callbacks[i].free_qiov);
K
Kevin Wolf 已提交
3012 3013 3014 3015 3016 3017 3018
    }
}

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

3019 3020
    trace_multiwrite_cb(mcb, ret);

3021
    if (ret < 0 && !mcb->error) {
K
Kevin Wolf 已提交
3022 3023 3024 3025 3026
        mcb->error = ret;
    }

    mcb->num_requests--;
    if (mcb->num_requests == 0) {
3027
        multiwrite_user_cb(mcb);
3028
        g_free(mcb);
K
Kevin Wolf 已提交
3029 3030 3031 3032 3033
    }
}

static int multiwrite_req_compare(const void *a, const void *b)
{
3034 3035 3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046
    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
3047 3048 3049 3050 3051 3052 3053 3054 3055 3056 3057 3058 3059 3060 3061 3062 3063 3064 3065 3066 3067
}

/*
 * 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;

3068
        // Handle exactly sequential writes and overlapping writes.
K
Kevin Wolf 已提交
3069 3070 3071 3072
        if (reqs[i].sector <= oldreq_last) {
            merge = 1;
        }

3073 3074 3075 3076
        if (reqs[outidx].qiov->niov + reqs[i].qiov->niov + 1 > IOV_MAX) {
            merge = 0;
        }

K
Kevin Wolf 已提交
3077 3078
        if (merge) {
            size_t size;
3079
            QEMUIOVector *qiov = g_malloc0(sizeof(*qiov));
K
Kevin Wolf 已提交
3080 3081 3082 3083 3084 3085 3086 3087
            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;
            qemu_iovec_concat(qiov, reqs[outidx].qiov, size);

3088 3089
            // We should need to add any zeros between the two requests
            assert (reqs[i].sector <= oldreq_last);
K
Kevin Wolf 已提交
3090 3091 3092 3093

            // Add the second request
            qemu_iovec_concat(qiov, reqs[i].qiov, reqs[i].qiov->size);

3094
            reqs[outidx].nb_sectors = qiov->size >> 9;
K
Kevin Wolf 已提交
3095 3096 3097 3098 3099 3100 3101 3102 3103 3104 3105 3106 3107 3108 3109 3110 3111 3112 3113 3114 3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127
            reqs[outidx].qiov = qiov;

            mcb->callbacks[i].free_qiov = reqs[outidx].qiov;
        } else {
            outidx++;
            reqs[outidx].sector     = reqs[i].sector;
            reqs[outidx].nb_sectors = reqs[i].nb_sectors;
            reqs[outidx].qiov       = reqs[i].qiov;
        }
    }

    return outidx + 1;
}

/*
 * Submit multiple AIO write requests at once.
 *
 * On success, the function returns 0 and all requests in the reqs array have
 * been submitted. In error case this function returns -1, and any of the
 * requests may or may not be submitted yet. In particular, this means that the
 * callback will be called for some of the requests, for others it won't. The
 * caller must check the error field of the BlockRequest to wait for the right
 * callbacks (if error != 0, no callback will be called).
 *
 * The implementation may modify the contents of the reqs array, e.g. to merge
 * requests. However, the fields opaque and error are left unmodified as they
 * are used to signal failure for a single request to the caller.
 */
int bdrv_aio_multiwrite(BlockDriverState *bs, BlockRequest *reqs, int num_reqs)
{
    MultiwriteCB *mcb;
    int i;

3128 3129 3130 3131 3132 3133 3134 3135
    /* 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
3136 3137 3138 3139 3140
    if (num_reqs == 0) {
        return 0;
    }

    // Create MultiwriteCB structure
3141
    mcb = g_malloc0(sizeof(*mcb) + num_reqs * sizeof(*mcb->callbacks));
K
Kevin Wolf 已提交
3142 3143 3144 3145 3146 3147 3148 3149 3150 3151 3152
    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);

3153 3154
    trace_bdrv_aio_multiwrite(mcb, mcb->num_callbacks, num_reqs);

3155 3156
    /* Run the aio requests. */
    mcb->num_requests = num_reqs;
K
Kevin Wolf 已提交
3157
    for (i = 0; i < num_reqs; i++) {
3158
        bdrv_aio_writev(bs, reqs[i].sector, reqs[i].qiov,
K
Kevin Wolf 已提交
3159 3160 3161 3162 3163 3164
            reqs[i].nb_sectors, multiwrite_cb, mcb);
    }

    return 0;
}

B
bellard 已提交
3165 3166
void bdrv_aio_cancel(BlockDriverAIOCB *acb)
{
3167
    acb->pool->cancel(acb);
B
bellard 已提交
3168 3169
}

Z
Zhi Yong Wu 已提交
3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210 3211 3212 3213 3214 3215 3216 3217 3218 3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249 3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333
/* block I/O throttling */
static bool bdrv_exceed_bps_limits(BlockDriverState *bs, int nb_sectors,
                 bool is_write, double elapsed_time, uint64_t *wait)
{
    uint64_t bps_limit = 0;
    double   bytes_limit, bytes_base, bytes_res;
    double   slice_time, wait_time;

    if (bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL]) {
        bps_limit = bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL];
    } else if (bs->io_limits.bps[is_write]) {
        bps_limit = bs->io_limits.bps[is_write];
    } else {
        if (wait) {
            *wait = 0;
        }

        return false;
    }

    slice_time = bs->slice_end - bs->slice_start;
    slice_time /= (NANOSECONDS_PER_SECOND);
    bytes_limit = bps_limit * slice_time;
    bytes_base  = bs->nr_bytes[is_write] - bs->io_base.bytes[is_write];
    if (bs->io_limits.bps[BLOCK_IO_LIMIT_TOTAL]) {
        bytes_base += bs->nr_bytes[!is_write] - bs->io_base.bytes[!is_write];
    }

    /* bytes_base: the bytes of data which have been read/written; and
     *             it is obtained from the history statistic info.
     * bytes_res: the remaining bytes of data which need to be read/written.
     * (bytes_base + bytes_res) / bps_limit: used to calcuate
     *             the total time for completing reading/writting all data.
     */
    bytes_res   = (unsigned) nb_sectors * BDRV_SECTOR_SIZE;

    if (bytes_base + bytes_res <= bytes_limit) {
        if (wait) {
            *wait = 0;
        }

        return false;
    }

    /* Calc approx time to dispatch */
    wait_time = (bytes_base + bytes_res) / bps_limit - elapsed_time;

    /* When the I/O rate at runtime exceeds the limits,
     * bs->slice_end need to be extended in order that the current statistic
     * info can be kept until the timer fire, so it is increased and tuned
     * based on the result of experiment.
     */
    bs->slice_time = wait_time * BLOCK_IO_SLICE_TIME * 10;
    bs->slice_end += bs->slice_time - 3 * BLOCK_IO_SLICE_TIME;
    if (wait) {
        *wait = wait_time * BLOCK_IO_SLICE_TIME * 10;
    }

    return true;
}

static bool bdrv_exceed_iops_limits(BlockDriverState *bs, bool is_write,
                             double elapsed_time, uint64_t *wait)
{
    uint64_t iops_limit = 0;
    double   ios_limit, ios_base;
    double   slice_time, wait_time;

    if (bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL]) {
        iops_limit = bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL];
    } else if (bs->io_limits.iops[is_write]) {
        iops_limit = bs->io_limits.iops[is_write];
    } else {
        if (wait) {
            *wait = 0;
        }

        return false;
    }

    slice_time = bs->slice_end - bs->slice_start;
    slice_time /= (NANOSECONDS_PER_SECOND);
    ios_limit  = iops_limit * slice_time;
    ios_base   = bs->nr_ops[is_write] - bs->io_base.ios[is_write];
    if (bs->io_limits.iops[BLOCK_IO_LIMIT_TOTAL]) {
        ios_base += bs->nr_ops[!is_write] - bs->io_base.ios[!is_write];
    }

    if (ios_base + 1 <= ios_limit) {
        if (wait) {
            *wait = 0;
        }

        return false;
    }

    /* Calc approx time to dispatch */
    wait_time = (ios_base + 1) / iops_limit;
    if (wait_time > elapsed_time) {
        wait_time = wait_time - elapsed_time;
    } else {
        wait_time = 0;
    }

    bs->slice_time = wait_time * BLOCK_IO_SLICE_TIME * 10;
    bs->slice_end += bs->slice_time - 3 * BLOCK_IO_SLICE_TIME;
    if (wait) {
        *wait = wait_time * BLOCK_IO_SLICE_TIME * 10;
    }

    return true;
}

static bool bdrv_exceed_io_limits(BlockDriverState *bs, int nb_sectors,
                           bool is_write, int64_t *wait)
{
    int64_t  now, max_wait;
    uint64_t bps_wait = 0, iops_wait = 0;
    double   elapsed_time;
    int      bps_ret, iops_ret;

    now = qemu_get_clock_ns(vm_clock);
    if ((bs->slice_start < now)
        && (bs->slice_end > now)) {
        bs->slice_end = now + bs->slice_time;
    } else {
        bs->slice_time  =  5 * BLOCK_IO_SLICE_TIME;
        bs->slice_start = now;
        bs->slice_end   = now + bs->slice_time;

        bs->io_base.bytes[is_write]  = bs->nr_bytes[is_write];
        bs->io_base.bytes[!is_write] = bs->nr_bytes[!is_write];

        bs->io_base.ios[is_write]    = bs->nr_ops[is_write];
        bs->io_base.ios[!is_write]   = bs->nr_ops[!is_write];
    }

    elapsed_time  = now - bs->slice_start;
    elapsed_time  /= (NANOSECONDS_PER_SECOND);

    bps_ret  = bdrv_exceed_bps_limits(bs, nb_sectors,
                                      is_write, elapsed_time, &bps_wait);
    iops_ret = bdrv_exceed_iops_limits(bs, is_write,
                                      elapsed_time, &iops_wait);
    if (bps_ret || iops_ret) {
        max_wait = bps_wait > iops_wait ? bps_wait : iops_wait;
        if (wait) {
            *wait = max_wait;
        }

        now = qemu_get_clock_ns(vm_clock);
        if (bs->slice_end < now + max_wait) {
            bs->slice_end = now + max_wait;
        }

        return true;
    }

    if (wait) {
        *wait = 0;
    }

    return false;
}
3334

B
bellard 已提交
3335 3336 3337
/**************************************************************/
/* async block device emulation */

3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349
typedef struct BlockDriverAIOCBSync {
    BlockDriverAIOCB common;
    QEMUBH *bh;
    int ret;
    /* vector translation state */
    QEMUIOVector *qiov;
    uint8_t *bounce;
    int is_write;
} BlockDriverAIOCBSync;

static void bdrv_aio_cancel_em(BlockDriverAIOCB *blockacb)
{
3350 3351
    BlockDriverAIOCBSync *acb =
        container_of(blockacb, BlockDriverAIOCBSync, common);
3352
    qemu_bh_delete(acb->bh);
A
Avi Kivity 已提交
3353
    acb->bh = NULL;
3354 3355 3356 3357 3358 3359 3360 3361
    qemu_aio_release(acb);
}

static AIOPool bdrv_em_aio_pool = {
    .aiocb_size         = sizeof(BlockDriverAIOCBSync),
    .cancel             = bdrv_aio_cancel_em,
};

3362
static void bdrv_aio_bh_cb(void *opaque)
B
bellard 已提交
3363
{
3364
    BlockDriverAIOCBSync *acb = opaque;
3365 3366 3367

    if (!acb->is_write)
        qemu_iovec_from_buffer(acb->qiov, acb->bounce, acb->qiov->size);
3368
    qemu_vfree(acb->bounce);
3369
    acb->common.cb(acb->common.opaque, acb->ret);
3370
    qemu_bh_delete(acb->bh);
A
Avi Kivity 已提交
3371
    acb->bh = NULL;
3372
    qemu_aio_release(acb);
B
bellard 已提交
3373
}
3374

3375 3376 3377 3378 3379 3380 3381 3382
static BlockDriverAIOCB *bdrv_aio_rw_vector(BlockDriverState *bs,
                                            int64_t sector_num,
                                            QEMUIOVector *qiov,
                                            int nb_sectors,
                                            BlockDriverCompletionFunc *cb,
                                            void *opaque,
                                            int is_write)

B
bellard 已提交
3383
{
3384 3385
    BlockDriverAIOCBSync *acb;

3386
    acb = qemu_aio_get(&bdrv_em_aio_pool, bs, cb, opaque);
3387 3388
    acb->is_write = is_write;
    acb->qiov = qiov;
3389
    acb->bounce = qemu_blockalign(bs, qiov->size);
3390
    acb->bh = qemu_bh_new(bdrv_aio_bh_cb, acb);
3391 3392 3393

    if (is_write) {
        qemu_iovec_to_buffer(acb->qiov, acb->bounce);
3394
        acb->ret = bs->drv->bdrv_write(bs, sector_num, acb->bounce, nb_sectors);
3395
    } else {
3396
        acb->ret = bs->drv->bdrv_read(bs, sector_num, acb->bounce, nb_sectors);
3397 3398
    }

3399
    qemu_bh_schedule(acb->bh);
3400

3401
    return &acb->common;
3402 3403
}

3404 3405
static BlockDriverAIOCB *bdrv_aio_readv_em(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
3406
        BlockDriverCompletionFunc *cb, void *opaque)
3407
{
3408 3409
    return bdrv_aio_rw_vector(bs, sector_num, qiov, nb_sectors, cb, opaque, 0);
}
B
bellard 已提交
3410

3411 3412 3413 3414 3415
static BlockDriverAIOCB *bdrv_aio_writev_em(BlockDriverState *bs,
        int64_t sector_num, QEMUIOVector *qiov, int nb_sectors,
        BlockDriverCompletionFunc *cb, void *opaque)
{
    return bdrv_aio_rw_vector(bs, sector_num, qiov, nb_sectors, cb, opaque, 1);
3416 3417
}

3418 3419 3420 3421 3422 3423 3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435

typedef struct BlockDriverAIOCBCoroutine {
    BlockDriverAIOCB common;
    BlockRequest req;
    bool is_write;
    QEMUBH* bh;
} BlockDriverAIOCBCoroutine;

static void bdrv_aio_co_cancel_em(BlockDriverAIOCB *blockacb)
{
    qemu_aio_flush();
}

static AIOPool bdrv_em_co_aio_pool = {
    .aiocb_size         = sizeof(BlockDriverAIOCBCoroutine),
    .cancel             = bdrv_aio_co_cancel_em,
};

P
Paolo Bonzini 已提交
3436
static void bdrv_co_em_bh(void *opaque)
3437 3438 3439 3440 3441 3442 3443 3444
{
    BlockDriverAIOCBCoroutine *acb = opaque;

    acb->common.cb(acb->common.opaque, acb->req.error);
    qemu_bh_delete(acb->bh);
    qemu_aio_release(acb);
}

3445 3446 3447 3448 3449 3450 3451 3452
/* Invoke bdrv_co_do_readv/bdrv_co_do_writev */
static void coroutine_fn bdrv_co_do_rw(void *opaque)
{
    BlockDriverAIOCBCoroutine *acb = opaque;
    BlockDriverState *bs = acb->common.bs;

    if (!acb->is_write) {
        acb->req.error = bdrv_co_do_readv(bs, acb->req.sector,
3453
            acb->req.nb_sectors, acb->req.qiov, 0);
3454 3455
    } else {
        acb->req.error = bdrv_co_do_writev(bs, acb->req.sector,
3456
            acb->req.nb_sectors, acb->req.qiov, 0);
3457 3458
    }

P
Paolo Bonzini 已提交
3459
    acb->bh = qemu_bh_new(bdrv_co_em_bh, acb);
3460 3461 3462
    qemu_bh_schedule(acb->bh);
}

3463 3464 3465 3466 3467 3468
static BlockDriverAIOCB *bdrv_co_aio_rw_vector(BlockDriverState *bs,
                                               int64_t sector_num,
                                               QEMUIOVector *qiov,
                                               int nb_sectors,
                                               BlockDriverCompletionFunc *cb,
                                               void *opaque,
3469
                                               bool is_write)
3470 3471 3472 3473 3474 3475 3476 3477 3478 3479
{
    Coroutine *co;
    BlockDriverAIOCBCoroutine *acb;

    acb = qemu_aio_get(&bdrv_em_co_aio_pool, bs, cb, opaque);
    acb->req.sector = sector_num;
    acb->req.nb_sectors = nb_sectors;
    acb->req.qiov = qiov;
    acb->is_write = is_write;

3480
    co = qemu_coroutine_create(bdrv_co_do_rw);
3481 3482 3483 3484 3485
    qemu_coroutine_enter(co, acb);

    return &acb->common;
}

P
Paolo Bonzini 已提交
3486
static void coroutine_fn bdrv_aio_flush_co_entry(void *opaque)
3487
{
P
Paolo Bonzini 已提交
3488 3489
    BlockDriverAIOCBCoroutine *acb = opaque;
    BlockDriverState *bs = acb->common.bs;
3490

P
Paolo Bonzini 已提交
3491 3492
    acb->req.error = bdrv_co_flush(bs);
    acb->bh = qemu_bh_new(bdrv_co_em_bh, acb);
3493 3494 3495
    qemu_bh_schedule(acb->bh);
}

P
Paolo Bonzini 已提交
3496
BlockDriverAIOCB *bdrv_aio_flush(BlockDriverState *bs,
3497 3498
        BlockDriverCompletionFunc *cb, void *opaque)
{
P
Paolo Bonzini 已提交
3499
    trace_bdrv_aio_flush(bs, opaque);
3500

P
Paolo Bonzini 已提交
3501 3502
    Coroutine *co;
    BlockDriverAIOCBCoroutine *acb;
3503

P
Paolo Bonzini 已提交
3504 3505 3506
    acb = qemu_aio_get(&bdrv_em_co_aio_pool, bs, cb, opaque);
    co = qemu_coroutine_create(bdrv_aio_flush_co_entry);
    qemu_coroutine_enter(co, acb);
3507 3508 3509 3510

    return &acb->common;
}

3511 3512 3513 3514 3515 3516 3517 3518 3519 3520 3521 3522 3523 3524 3525 3526 3527 3528 3529 3530 3531 3532 3533 3534 3535 3536 3537 3538
static void coroutine_fn bdrv_aio_discard_co_entry(void *opaque)
{
    BlockDriverAIOCBCoroutine *acb = opaque;
    BlockDriverState *bs = acb->common.bs;

    acb->req.error = bdrv_co_discard(bs, acb->req.sector, acb->req.nb_sectors);
    acb->bh = qemu_bh_new(bdrv_co_em_bh, acb);
    qemu_bh_schedule(acb->bh);
}

BlockDriverAIOCB *bdrv_aio_discard(BlockDriverState *bs,
        int64_t sector_num, int nb_sectors,
        BlockDriverCompletionFunc *cb, void *opaque)
{
    Coroutine *co;
    BlockDriverAIOCBCoroutine *acb;

    trace_bdrv_aio_discard(bs, sector_num, nb_sectors, opaque);

    acb = qemu_aio_get(&bdrv_em_co_aio_pool, bs, cb, opaque);
    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
3539 3540
void bdrv_init(void)
{
3541
    module_call_init(MODULE_INIT_BLOCK);
B
bellard 已提交
3542
}
3543

3544 3545 3546 3547 3548 3549
void bdrv_init_with_whitelist(void)
{
    use_bdrv_whitelist = 1;
    bdrv_init();
}

3550 3551
void *qemu_aio_get(AIOPool *pool, BlockDriverState *bs,
                   BlockDriverCompletionFunc *cb, void *opaque)
3552 3553 3554
{
    BlockDriverAIOCB *acb;

3555 3556 3557
    if (pool->free_aiocb) {
        acb = pool->free_aiocb;
        pool->free_aiocb = acb->next;
3558
    } else {
3559
        acb = g_malloc0(pool->aiocb_size);
3560
        acb->pool = pool;
3561 3562 3563 3564 3565 3566 3567 3568 3569
    }
    acb->bs = bs;
    acb->cb = cb;
    acb->opaque = opaque;
    return acb;
}

void qemu_aio_release(void *p)
{
3570 3571 3572 3573
    BlockDriverAIOCB *acb = (BlockDriverAIOCB *)p;
    AIOPool *pool = acb->pool;
    acb->next = pool->free_aiocb;
    pool->free_aiocb = acb;
3574
}
B
bellard 已提交
3575

3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587 3588 3589 3590 3591 3592 3593 3594 3595 3596 3597 3598 3599 3600 3601
/**************************************************************/
/* Coroutine block device emulation */

typedef struct CoroutineIOCompletion {
    Coroutine *coroutine;
    int ret;
} CoroutineIOCompletion;

static void bdrv_co_io_em_complete(void *opaque, int ret)
{
    CoroutineIOCompletion *co = opaque;

    co->ret = ret;
    qemu_coroutine_enter(co->coroutine, NULL);
}

static int coroutine_fn bdrv_co_io_em(BlockDriverState *bs, int64_t sector_num,
                                      int nb_sectors, QEMUIOVector *iov,
                                      bool is_write)
{
    CoroutineIOCompletion co = {
        .coroutine = qemu_coroutine_self(),
    };
    BlockDriverAIOCB *acb;

    if (is_write) {
3602 3603
        acb = bs->drv->bdrv_aio_writev(bs, sector_num, iov, nb_sectors,
                                       bdrv_co_io_em_complete, &co);
3604
    } else {
3605 3606
        acb = bs->drv->bdrv_aio_readv(bs, sector_num, iov, nb_sectors,
                                      bdrv_co_io_em_complete, &co);
3607 3608
    }

3609
    trace_bdrv_co_io_em(bs, sector_num, nb_sectors, is_write, acb);
3610 3611 3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627 3628 3629 3630 3631
    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
3632
static void coroutine_fn bdrv_flush_co_entry(void *opaque)
3633
{
P
Paolo Bonzini 已提交
3634 3635 3636 3637 3638 3639 3640
    RwCo *rwco = opaque;

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

int coroutine_fn bdrv_co_flush(BlockDriverState *bs)
{
K
Kevin Wolf 已提交
3641 3642
    int ret;

3643
    if (!bs || !bdrv_is_inserted(bs) || bdrv_is_read_only(bs)) {
P
Paolo Bonzini 已提交
3644
        return 0;
K
Kevin Wolf 已提交
3645 3646
    }

3647
    /* Write back cached data to the OS even with cache=unsafe */
K
Kevin Wolf 已提交
3648 3649 3650 3651 3652 3653 3654
    if (bs->drv->bdrv_co_flush_to_os) {
        ret = bs->drv->bdrv_co_flush_to_os(bs);
        if (ret < 0) {
            return ret;
        }
    }

3655 3656 3657 3658 3659
    /* But don't actually force it to the disk with cache=unsafe */
    if (bs->open_flags & BDRV_O_NO_FLUSH) {
        return 0;
    }

K
Kevin Wolf 已提交
3660
    if (bs->drv->bdrv_co_flush_to_disk) {
3661
        ret = bs->drv->bdrv_co_flush_to_disk(bs);
P
Paolo Bonzini 已提交
3662 3663 3664 3665 3666 3667 3668 3669
    } else if (bs->drv->bdrv_aio_flush) {
        BlockDriverAIOCB *acb;
        CoroutineIOCompletion co = {
            .coroutine = qemu_coroutine_self(),
        };

        acb = bs->drv->bdrv_aio_flush(bs, bdrv_co_io_em_complete, &co);
        if (acb == NULL) {
3670
            ret = -EIO;
P
Paolo Bonzini 已提交
3671 3672
        } else {
            qemu_coroutine_yield();
3673
            ret = co.ret;
P
Paolo Bonzini 已提交
3674 3675 3676 3677 3678 3679 3680 3681 3682 3683 3684 3685 3686
        }
    } 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.
         */
3687
        ret = 0;
P
Paolo Bonzini 已提交
3688
    }
3689 3690 3691 3692 3693 3694 3695 3696
    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.
     */
    return bdrv_co_flush(bs->file);
P
Paolo Bonzini 已提交
3697 3698
}

3699 3700 3701 3702 3703 3704 3705 3706 3707 3708 3709 3710 3711 3712 3713 3714
void bdrv_invalidate_cache(BlockDriverState *bs)
{
    if (bs->drv && bs->drv->bdrv_invalidate_cache) {
        bs->drv->bdrv_invalidate_cache(bs);
    }
}

void bdrv_invalidate_cache_all(void)
{
    BlockDriverState *bs;

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

3715 3716 3717 3718 3719 3720 3721 3722 3723
void bdrv_clear_incoming_migration_all(void)
{
    BlockDriverState *bs;

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

P
Paolo Bonzini 已提交
3724 3725 3726 3727 3728 3729
int bdrv_flush(BlockDriverState *bs)
{
    Coroutine *co;
    RwCo rwco = {
        .bs = bs,
        .ret = NOT_DONE,
3730 3731
    };

P
Paolo Bonzini 已提交
3732 3733 3734 3735 3736 3737 3738 3739 3740
    if (qemu_in_coroutine()) {
        /* Fast-path if already in coroutine context */
        bdrv_flush_co_entry(&rwco);
    } else {
        co = qemu_coroutine_create(bdrv_flush_co_entry);
        qemu_coroutine_enter(co, &rwco);
        while (rwco.ret == NOT_DONE) {
            qemu_aio_wait();
        }
3741
    }
P
Paolo Bonzini 已提交
3742 3743

    return rwco.ret;
3744 3745
}

3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799 3800 3801 3802 3803 3804 3805 3806
static void coroutine_fn bdrv_discard_co_entry(void *opaque)
{
    RwCo *rwco = opaque;

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

int coroutine_fn bdrv_co_discard(BlockDriverState *bs, int64_t sector_num,
                                 int nb_sectors)
{
    if (!bs->drv) {
        return -ENOMEDIUM;
    } else if (bdrv_check_request(bs, sector_num, nb_sectors)) {
        return -EIO;
    } else if (bs->read_only) {
        return -EROFS;
    } else if (bs->drv->bdrv_co_discard) {
        return bs->drv->bdrv_co_discard(bs, sector_num, nb_sectors);
    } else if (bs->drv->bdrv_aio_discard) {
        BlockDriverAIOCB *acb;
        CoroutineIOCompletion co = {
            .coroutine = qemu_coroutine_self(),
        };

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

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

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

    return rwco.ret;
}

B
bellard 已提交
3807 3808 3809 3810 3811 3812 3813 3814 3815
/**************************************************************/
/* removable device support */

/**
 * Return TRUE if the media is present
 */
int bdrv_is_inserted(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;
3816

B
bellard 已提交
3817 3818 3819
    if (!drv)
        return 0;
    if (!drv->bdrv_is_inserted)
3820 3821
        return 1;
    return drv->bdrv_is_inserted(bs);
B
bellard 已提交
3822 3823 3824
}

/**
3825 3826
 * 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
3827 3828 3829 3830 3831
 */
int bdrv_media_changed(BlockDriverState *bs)
{
    BlockDriver *drv = bs->drv;

3832 3833 3834 3835
    if (drv && drv->bdrv_media_changed) {
        return drv->bdrv_media_changed(bs);
    }
    return -ENOTSUP;
B
bellard 已提交
3836 3837 3838 3839 3840
}

/**
 * If eject_flag is TRUE, eject the media. Otherwise, close the tray
 */
3841
void bdrv_eject(BlockDriverState *bs, bool eject_flag)
B
bellard 已提交
3842 3843 3844
{
    BlockDriver *drv = bs->drv;

3845 3846
    if (drv && drv->bdrv_eject) {
        drv->bdrv_eject(bs, eject_flag);
B
bellard 已提交
3847
    }
3848 3849 3850 3851

    if (bs->device_name[0] != '\0') {
        bdrv_emit_qmp_eject_event(bs, eject_flag);
    }
B
bellard 已提交
3852 3853 3854 3855 3856 3857
}

/**
 * Lock or unlock the media (if it is locked, the user won't be able
 * to eject it manually).
 */
3858
void bdrv_lock_medium(BlockDriverState *bs, bool locked)
B
bellard 已提交
3859 3860 3861
{
    BlockDriver *drv = bs->drv;

3862
    trace_bdrv_lock_medium(bs, locked);
S
Stefan Hajnoczi 已提交
3863

3864 3865
    if (drv && drv->bdrv_lock_medium) {
        drv->bdrv_lock_medium(bs, locked);
B
bellard 已提交
3866 3867
    }
}
3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878

/* 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;
}
3879

3880 3881 3882
BlockDriverAIOCB *bdrv_aio_ioctl(BlockDriverState *bs,
        unsigned long int req, void *buf,
        BlockDriverCompletionFunc *cb, void *opaque)
3883
{
3884
    BlockDriver *drv = bs->drv;
3885

3886 3887 3888
    if (drv && drv->bdrv_aio_ioctl)
        return drv->bdrv_aio_ioctl(bs, req, buf, cb, opaque);
    return NULL;
3889
}
3890

3891 3892 3893 3894
void bdrv_set_buffer_alignment(BlockDriverState *bs, int align)
{
    bs->buffer_alignment = align;
}
3895

3896 3897 3898 3899
void *qemu_blockalign(BlockDriverState *bs, size_t size)
{
    return qemu_memalign((bs && bs->buffer_alignment) ? bs->buffer_alignment : 512, size);
}
3900 3901 3902 3903

void bdrv_set_dirty_tracking(BlockDriverState *bs, int enable)
{
    int64_t bitmap_size;
3904

3905
    bs->dirty_count = 0;
3906
    if (enable) {
3907 3908
        if (!bs->dirty_bitmap) {
            bitmap_size = (bdrv_getlength(bs) >> BDRV_SECTOR_BITS) +
3909 3910
                    BDRV_SECTORS_PER_DIRTY_CHUNK * BITS_PER_LONG - 1;
            bitmap_size /= BDRV_SECTORS_PER_DIRTY_CHUNK * BITS_PER_LONG;
3911

3912
            bs->dirty_bitmap = g_new0(unsigned long, bitmap_size);
3913
        }
3914
    } else {
3915
        if (bs->dirty_bitmap) {
3916
            g_free(bs->dirty_bitmap);
3917
            bs->dirty_bitmap = NULL;
3918
        }
3919 3920 3921 3922 3923
    }
}

int bdrv_get_dirty(BlockDriverState *bs, int64_t sector)
{
J
Jan Kiszka 已提交
3924
    int64_t chunk = sector / (int64_t)BDRV_SECTORS_PER_DIRTY_CHUNK;
3925

3926 3927
    if (bs->dirty_bitmap &&
        (sector << BDRV_SECTOR_BITS) < bdrv_getlength(bs)) {
3928 3929
        return !!(bs->dirty_bitmap[chunk / (sizeof(unsigned long) * 8)] &
            (1UL << (chunk % (sizeof(unsigned long) * 8))));
3930 3931 3932 3933 3934
    } else {
        return 0;
    }
}

3935 3936
void bdrv_reset_dirty(BlockDriverState *bs, int64_t cur_sector,
                      int nr_sectors)
3937 3938 3939
{
    set_dirty_bitmap(bs, cur_sector, nr_sectors, 0);
}
3940 3941 3942 3943 3944

int64_t bdrv_get_dirty_count(BlockDriverState *bs)
{
    return bs->dirty_count;
}
J
Jes Sorensen 已提交
3945

3946 3947 3948 3949 3950 3951 3952 3953 3954 3955 3956
void bdrv_set_in_use(BlockDriverState *bs, int in_use)
{
    assert(bs->in_use != in_use);
    bs->in_use = in_use;
}

int bdrv_in_use(BlockDriverState *bs)
{
    return bs->in_use;
}

3957 3958
void bdrv_iostatus_enable(BlockDriverState *bs)
{
3959
    bs->iostatus_enabled = true;
3960
    bs->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
3961 3962 3963 3964 3965 3966
}

/* 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)
{
3967
    return (bs->iostatus_enabled &&
3968 3969 3970 3971 3972 3973 3974
           (bs->on_write_error == BLOCK_ERR_STOP_ENOSPC ||
            bs->on_write_error == BLOCK_ERR_STOP_ANY    ||
            bs->on_read_error == BLOCK_ERR_STOP_ANY));
}

void bdrv_iostatus_disable(BlockDriverState *bs)
{
3975
    bs->iostatus_enabled = false;
3976 3977 3978 3979 3980
}

void bdrv_iostatus_reset(BlockDriverState *bs)
{
    if (bdrv_iostatus_is_enabled(bs)) {
3981
        bs->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
3982 3983 3984 3985 3986 3987 3988 3989
    }
}

/* XXX: Today this is set by device models because it makes the implementation
   quite simple. However, the block layer knows about the error, so it's
   possible to implement this without device models being involved */
void bdrv_iostatus_set_err(BlockDriverState *bs, int error)
{
3990 3991
    if (bdrv_iostatus_is_enabled(bs) &&
        bs->iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
3992
        assert(error >= 0);
3993 3994
        bs->iostatus = error == ENOSPC ? BLOCK_DEVICE_IO_STATUS_NOSPACE :
                                         BLOCK_DEVICE_IO_STATUS_FAILED;
3995 3996 3997
    }
}

3998 3999 4000 4001 4002 4003 4004
void
bdrv_acct_start(BlockDriverState *bs, BlockAcctCookie *cookie, int64_t bytes,
        enum BlockAcctType type)
{
    assert(type < BDRV_MAX_IOTYPE);

    cookie->bytes = bytes;
C
Christoph Hellwig 已提交
4005
    cookie->start_time_ns = get_clock();
4006 4007 4008 4009 4010 4011 4012 4013 4014 4015
    cookie->type = type;
}

void
bdrv_acct_done(BlockDriverState *bs, BlockAcctCookie *cookie)
{
    assert(cookie->type < BDRV_MAX_IOTYPE);

    bs->nr_bytes[cookie->type] += cookie->bytes;
    bs->nr_ops[cookie->type]++;
C
Christoph Hellwig 已提交
4016
    bs->total_time_ns[cookie->type] += get_clock() - cookie->start_time_ns;
4017 4018
}

J
Jes Sorensen 已提交
4019 4020 4021 4022 4023
int bdrv_img_create(const char *filename, const char *fmt,
                    const char *base_filename, const char *base_fmt,
                    char *options, uint64_t img_size, int flags)
{
    QEMUOptionParameter *param = NULL, *create_options = NULL;
K
Kevin Wolf 已提交
4024
    QEMUOptionParameter *backing_fmt, *backing_file, *size;
J
Jes Sorensen 已提交
4025 4026
    BlockDriverState *bs = NULL;
    BlockDriver *drv, *proto_drv;
4027
    BlockDriver *backing_drv = NULL;
J
Jes Sorensen 已提交
4028 4029 4030 4031 4032 4033
    int ret = 0;

    /* Find driver and parse its options */
    drv = bdrv_find_format(fmt);
    if (!drv) {
        error_report("Unknown file format '%s'", fmt);
4034
        ret = -EINVAL;
J
Jes Sorensen 已提交
4035 4036 4037 4038 4039 4040
        goto out;
    }

    proto_drv = bdrv_find_protocol(filename);
    if (!proto_drv) {
        error_report("Unknown protocol '%s'", filename);
4041
        ret = -EINVAL;
J
Jes Sorensen 已提交
4042 4043 4044 4045 4046 4047 4048 4049 4050 4051 4052 4053 4054 4055 4056 4057 4058 4059
        goto out;
    }

    create_options = append_option_parameters(create_options,
                                              drv->create_options);
    create_options = append_option_parameters(create_options,
                                              proto_drv->create_options);

    /* Create parameter list with default values */
    param = parse_option_parameters("", create_options, param);

    set_option_parameter_int(param, BLOCK_OPT_SIZE, img_size);

    /* Parse -o options */
    if (options) {
        param = parse_option_parameters(options, create_options, param);
        if (param == NULL) {
            error_report("Invalid options for file format '%s'.", fmt);
4060
            ret = -EINVAL;
J
Jes Sorensen 已提交
4061 4062 4063 4064 4065 4066 4067 4068 4069
            goto out;
        }
    }

    if (base_filename) {
        if (set_option_parameter(param, BLOCK_OPT_BACKING_FILE,
                                 base_filename)) {
            error_report("Backing file not supported for file format '%s'",
                         fmt);
4070
            ret = -EINVAL;
J
Jes Sorensen 已提交
4071 4072 4073 4074 4075 4076 4077 4078
            goto out;
        }
    }

    if (base_fmt) {
        if (set_option_parameter(param, BLOCK_OPT_BACKING_FMT, base_fmt)) {
            error_report("Backing file format not supported for file "
                         "format '%s'", fmt);
4079
            ret = -EINVAL;
J
Jes Sorensen 已提交
4080 4081 4082 4083
            goto out;
        }
    }

4084 4085 4086 4087 4088
    backing_file = get_option_parameter(param, BLOCK_OPT_BACKING_FILE);
    if (backing_file && backing_file->value.s) {
        if (!strcmp(filename, backing_file->value.s)) {
            error_report("Error: Trying to create an image with the "
                         "same filename as the backing file");
4089
            ret = -EINVAL;
4090 4091 4092 4093
            goto out;
        }
    }

J
Jes Sorensen 已提交
4094 4095
    backing_fmt = get_option_parameter(param, BLOCK_OPT_BACKING_FMT);
    if (backing_fmt && backing_fmt->value.s) {
4096 4097
        backing_drv = bdrv_find_format(backing_fmt->value.s);
        if (!backing_drv) {
J
Jes Sorensen 已提交
4098 4099
            error_report("Unknown backing file format '%s'",
                         backing_fmt->value.s);
4100
            ret = -EINVAL;
J
Jes Sorensen 已提交
4101 4102 4103 4104 4105 4106
            goto out;
        }
    }

    // The size for the image must always be specified, with one exception:
    // If we are using a backing file, we can obtain the size from there
K
Kevin Wolf 已提交
4107 4108
    size = get_option_parameter(param, BLOCK_OPT_SIZE);
    if (size && size->value.n == -1) {
J
Jes Sorensen 已提交
4109 4110 4111
        if (backing_file && backing_file->value.s) {
            uint64_t size;
            char buf[32];
4112 4113 4114 4115 4116
            int back_flags;

            /* backing files always opened read-only */
            back_flags =
                flags & ~(BDRV_O_RDWR | BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING);
J
Jes Sorensen 已提交
4117 4118 4119

            bs = bdrv_new("");

4120
            ret = bdrv_open(bs, backing_file->value.s, back_flags, backing_drv);
J
Jes Sorensen 已提交
4121
            if (ret < 0) {
4122
                error_report("Could not open '%s'", backing_file->value.s);
J
Jes Sorensen 已提交
4123 4124 4125 4126 4127 4128 4129 4130 4131
                goto out;
            }
            bdrv_get_geometry(bs, &size);
            size *= 512;

            snprintf(buf, sizeof(buf), "%" PRId64, size);
            set_option_parameter(param, BLOCK_OPT_SIZE, buf);
        } else {
            error_report("Image creation needs a size parameter");
4132
            ret = -EINVAL;
J
Jes Sorensen 已提交
4133 4134 4135 4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162
            goto out;
        }
    }

    printf("Formatting '%s', fmt=%s ", filename, fmt);
    print_option_parameters(param);
    puts("");

    ret = bdrv_create(drv, filename, param);

    if (ret < 0) {
        if (ret == -ENOTSUP) {
            error_report("Formatting or formatting option not supported for "
                         "file format '%s'", fmt);
        } else if (ret == -EFBIG) {
            error_report("The image size is too large for file format '%s'",
                         fmt);
        } else {
            error_report("%s: error while creating %s: %s", filename, fmt,
                         strerror(-ret));
        }
    }

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

    if (bs) {
        bdrv_delete(bs);
    }
4163 4164

    return ret;
J
Jes Sorensen 已提交
4165
}
4166 4167

void *block_job_create(const BlockJobType *job_type, BlockDriverState *bs,
4168 4169
                       int64_t speed, BlockDriverCompletionFunc *cb,
                       void *opaque, Error **errp)
4170 4171 4172 4173
{
    BlockJob *job;

    if (bs->job || bdrv_in_use(bs)) {
4174
        error_set(errp, QERR_DEVICE_IN_USE, bdrv_get_device_name(bs));
4175 4176 4177 4178 4179 4180 4181 4182 4183
        return NULL;
    }
    bdrv_set_in_use(bs, 1);

    job = g_malloc0(job_type->instance_size);
    job->job_type      = job_type;
    job->bs            = bs;
    job->cb            = cb;
    job->opaque        = opaque;
P
Paolo Bonzini 已提交
4184
    job->busy          = true;
4185
    bs->job = job;
4186 4187 4188 4189 4190 4191 4192 4193 4194 4195 4196 4197 4198 4199

    /* Only set speed when necessary to avoid NotSupported error */
    if (speed != 0) {
        Error *local_err = NULL;

        block_job_set_speed(job, speed, &local_err);
        if (error_is_set(&local_err)) {
            bs->job = NULL;
            g_free(job);
            bdrv_set_in_use(bs, 0);
            error_propagate(errp, local_err);
            return NULL;
        }
    }
4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210 4211 4212 4213
    return job;
}

void block_job_complete(BlockJob *job, int ret)
{
    BlockDriverState *bs = job->bs;

    assert(bs->job == job);
    job->cb(job->opaque, ret);
    bs->job = NULL;
    g_free(job);
    bdrv_set_in_use(bs, 0);
}

4214
void block_job_set_speed(BlockJob *job, int64_t speed, Error **errp)
4215
{
4216
    Error *local_err = NULL;
4217

4218
    if (!job->job_type->set_speed) {
4219 4220
        error_set(errp, QERR_NOT_SUPPORTED);
        return;
4221
    }
4222
    job->job_type->set_speed(job, speed, &local_err);
4223 4224 4225
    if (error_is_set(&local_err)) {
        error_propagate(errp, local_err);
        return;
4226
    }
4227

4228
    job->speed = speed;
4229 4230 4231 4232 4233
}

void block_job_cancel(BlockJob *job)
{
    job->cancelled = true;
4234 4235 4236
    if (job->co && !job->busy) {
        qemu_coroutine_enter(job->co, NULL);
    }
4237 4238 4239 4240 4241 4242
}

bool block_job_is_cancelled(BlockJob *job)
{
    return job->cancelled;
}
4243

4244 4245 4246 4247 4248 4249 4250 4251 4252
struct BlockCancelData {
    BlockJob *job;
    BlockDriverCompletionFunc *cb;
    void *opaque;
    bool cancelled;
    int ret;
};

static void block_job_cancel_cb(void *opaque, int ret)
4253
{
4254 4255 4256 4257 4258 4259 4260 4261 4262 4263
    struct BlockCancelData *data = opaque;

    data->cancelled = block_job_is_cancelled(data->job);
    data->ret = ret;
    data->cb(data->opaque, ret);
}

int block_job_cancel_sync(BlockJob *job)
{
    struct BlockCancelData data;
4264 4265 4266
    BlockDriverState *bs = job->bs;

    assert(bs->job == job);
4267 4268 4269 4270 4271 4272 4273 4274 4275 4276

    /* Set up our own callback to store the result and chain to
     * the original callback.
     */
    data.job = job;
    data.cb = job->cb;
    data.opaque = job->opaque;
    data.ret = -EINPROGRESS;
    job->cb = block_job_cancel_cb;
    job->opaque = &data;
4277
    block_job_cancel(job);
4278
    while (data.ret == -EINPROGRESS) {
4279 4280
        qemu_aio_wait();
    }
4281
    return (data.cancelled && data.ret == 0) ? -ECANCELED : data.ret;
4282
}
P
Paolo Bonzini 已提交
4283 4284 4285 4286 4287 4288 4289 4290 4291 4292

void block_job_sleep_ns(BlockJob *job, QEMUClock *clock, int64_t ns)
{
    /* Check cancellation *before* setting busy = false, too!  */
    if (!block_job_is_cancelled(job)) {
        job->busy = false;
        co_sleep_ns(clock, ns);
        job->busy = true;
    }
}