blockdev.c 118.1 KB
Newer Older
1 2 3 4 5 6 7
/*
 * QEMU host block devices
 *
 * Copyright (c) 2003-2008 Fabrice Bellard
 *
 * This work is licensed under the terms of the GNU GPL, version 2 or
 * later.  See the COPYING file in the top-level directory.
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
 *
 * This file incorporates work covered by the following copyright and
 * permission notice:
 *
 * Copyright (c) 2003-2008 Fabrice Bellard
 *
 * 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.
31 32
 */

P
Peter Maydell 已提交
33
#include "qemu/osdep.h"
M
Markus Armbruster 已提交
34
#include "sysemu/block-backend.h"
35
#include "sysemu/blockdev.h"
P
Paolo Bonzini 已提交
36
#include "hw/block/block.h"
37
#include "block/blockjob.h"
38
#include "block/throttle-groups.h"
39
#include "monitor/monitor.h"
40
#include "qemu/error-report.h"
41 42
#include "qemu/option.h"
#include "qemu/config-file.h"
43
#include "qapi/qmp/types.h"
K
Kevin Wolf 已提交
44
#include "qapi-visit.h"
45
#include "qapi/qmp/qerror.h"
46
#include "qapi/qobject-output-visitor.h"
47
#include "qapi/util.h"
48
#include "sysemu/sysemu.h"
49
#include "block/block_int.h"
L
Luiz Capitulino 已提交
50
#include "qmp-commands.h"
51
#include "block/trace.h"
52
#include "sysemu/arch_init.h"
53 54
#include "qemu/cutils.h"
#include "qemu/help_option.h"
55
#include "qemu/throttle-options.h"
56

57 58 59
static QTAILQ_HEAD(, BlockDriverState) monitor_bdrv_states =
    QTAILQ_HEAD_INITIALIZER(monitor_bdrv_states);

60 61
static int do_open_tray(const char *blk_name, const char *qdev_id,
                        bool force, Error **errp);
62

63 64 65 66 67 68 69 70 71 72 73 74
static const char *const if_name[IF_COUNT] = {
    [IF_NONE] = "none",
    [IF_IDE] = "ide",
    [IF_SCSI] = "scsi",
    [IF_FLOPPY] = "floppy",
    [IF_PFLASH] = "pflash",
    [IF_MTD] = "mtd",
    [IF_SD] = "sd",
    [IF_VIRTIO] = "virtio",
    [IF_XEN] = "xen",
};

75
static int if_max_devs[IF_COUNT] = {
76 77 78 79
    /*
     * Do not change these numbers!  They govern how drive option
     * index maps to unit and bus.  That mapping is ABI.
     *
W
Wei Jiangang 已提交
80
     * All controllers used to implement if=T drives need to support
81 82 83 84 85 86 87 88 89 90 91
     * if_max_devs[T] units, for any T with if_max_devs[T] != 0.
     * Otherwise, some index values map to "impossible" bus, unit
     * values.
     *
     * For instance, if you change [IF_SCSI] to 255, -drive
     * if=scsi,index=12 no longer means bus=1,unit=5, but
     * bus=0,unit=12.  With an lsi53c895a controller (7 units max),
     * the drive can't be set up.  Regression.
     */
    [IF_IDE] = 2,
    [IF_SCSI] = 7,
92 93
};

94 95 96 97 98 99
/**
 * Boards may call this to offer board-by-board overrides
 * of the default, global values.
 */
void override_max_devs(BlockInterfaceType type, int max_devs)
{
100
    BlockBackend *blk;
101 102 103 104 105 106
    DriveInfo *dinfo;

    if (max_devs <= 0) {
        return;
    }

107 108
    for (blk = blk_next(NULL); blk; blk = blk_next(blk)) {
        dinfo = blk_legacy_dinfo(blk);
109 110 111 112 113 114 115 116 117 118 119
        if (dinfo->type == type) {
            fprintf(stderr, "Cannot override units-per-bus property of"
                    " the %s interface, because a drive of that type has"
                    " already been added.\n", if_name[type]);
            g_assert_not_reached();
        }
    }

    if_max_devs[type] = max_devs;
}

120 121 122 123 124 125 126
/*
 * We automatically delete the drive when a device using it gets
 * unplugged.  Questionable feature, but we can't just drop it.
 * Device models call blockdev_mark_auto_del() to schedule the
 * automatic deletion, and generic qdev code calls blockdev_auto_del()
 * when deletion is actually safe.
 */
127
void blockdev_mark_auto_del(BlockBackend *blk)
128
{
129
    DriveInfo *dinfo = blk_legacy_dinfo(blk);
130
    BlockDriverState *bs = blk_bs(blk);
131
    AioContext *aio_context;
132

133
    if (!dinfo) {
134 135 136
        return;
    }

M
Max Reitz 已提交
137 138 139
    if (bs) {
        aio_context = bdrv_get_aio_context(bs);
        aio_context_acquire(aio_context);
140

M
Max Reitz 已提交
141 142 143
        if (bs->job) {
            block_job_cancel(bs->job);
        }
144

M
Max Reitz 已提交
145 146
        aio_context_release(aio_context);
    }
147

148
    dinfo->auto_del = 1;
149 150
}

151
void blockdev_auto_del(BlockBackend *blk)
152
{
153
    DriveInfo *dinfo = blk_legacy_dinfo(blk);
154

155
    if (dinfo && dinfo->auto_del) {
156
        monitor_remove_blk(blk);
157
        blk_unref(blk);
158 159 160
    }
}

161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177
/**
 * Returns the current mapping of how many units per bus
 * a particular interface can support.
 *
 *  A positive integer indicates n units per bus.
 *  0 implies the mapping has not been established.
 * -1 indicates an invalid BlockInterfaceType was given.
 */
int drive_get_max_devs(BlockInterfaceType type)
{
    if (type >= IF_IDE && type < IF_COUNT) {
        return if_max_devs[type];
    }

    return -1;
}

178 179 180 181 182 183 184 185 186 187 188 189
static int drive_index_to_bus_id(BlockInterfaceType type, int index)
{
    int max_devs = if_max_devs[type];
    return max_devs ? index / max_devs : 0;
}

static int drive_index_to_unit_id(BlockInterfaceType type, int index)
{
    int max_devs = if_max_devs[type];
    return max_devs ? index % max_devs : index;
}

190 191
QemuOpts *drive_def(const char *optstr)
{
192
    return qemu_opts_parse_noisily(qemu_find_opts("drive"), optstr, false);
193 194 195
}

QemuOpts *drive_add(BlockInterfaceType type, int index, const char *file,
196
                    const char *optstr)
197 198 199
{
    QemuOpts *opts;

200
    opts = drive_def(optstr);
201 202 203
    if (!opts) {
        return NULL;
    }
204
    if (type != IF_DEFAULT) {
205
        qemu_opt_set(opts, "if", if_name[type], &error_abort);
206 207
    }
    if (index >= 0) {
208
        qemu_opt_set_number(opts, "index", index, &error_abort);
209
    }
210
    if (file)
211
        qemu_opt_set(opts, "file", file, &error_abort);
212 213 214 215 216
    return opts;
}

DriveInfo *drive_get(BlockInterfaceType type, int bus, int unit)
{
217
    BlockBackend *blk;
218 219
    DriveInfo *dinfo;

220 221 222 223
    for (blk = blk_next(NULL); blk; blk = blk_next(blk)) {
        dinfo = blk_legacy_dinfo(blk);
        if (dinfo && dinfo->type == type
            && dinfo->bus == bus && dinfo->unit == unit) {
224
            return dinfo;
225
        }
226 227 228 229 230
    }

    return NULL;
}

231
void drive_check_orphaned(void)
J
John Snow 已提交
232
{
233
    BlockBackend *blk;
J
John Snow 已提交
234
    DriveInfo *dinfo;
235
    Location loc;
236
    bool orphans = false;
J
John Snow 已提交
237

238 239
    for (blk = blk_next(NULL); blk; blk = blk_next(blk)) {
        dinfo = blk_legacy_dinfo(blk);
240
        if (!blk_get_attached_dev(blk) && !dinfo->is_default &&
J
John Snow 已提交
241
            dinfo->type != IF_NONE) {
242 243
            loc_push_none(&loc);
            qemu_opts_loc_restore(dinfo->opts);
244
            error_report("machine type does not support"
245 246 247
                         " if=%s,bus=%d,unit=%d",
                         if_name[dinfo->type], dinfo->bus, dinfo->unit);
            loc_pop(&loc);
248
            orphans = true;
J
John Snow 已提交
249 250 251
        }
    }

252 253 254
    if (orphans) {
        exit(1);
    }
J
John Snow 已提交
255 256
}

257 258 259 260 261 262 263
DriveInfo *drive_get_by_index(BlockInterfaceType type, int index)
{
    return drive_get(type,
                     drive_index_to_bus_id(type, index),
                     drive_index_to_unit_id(type, index));
}

264 265 266
int drive_get_max_bus(BlockInterfaceType type)
{
    int max_bus;
267
    BlockBackend *blk;
268 269 270
    DriveInfo *dinfo;

    max_bus = -1;
271 272 273
    for (blk = blk_next(NULL); blk; blk = blk_next(blk)) {
        dinfo = blk_legacy_dinfo(blk);
        if (dinfo && dinfo->type == type && dinfo->bus > max_bus) {
274
            max_bus = dinfo->bus;
275
        }
276 277 278 279
    }
    return max_bus;
}

280 281 282 283 284 285 286 287 288 289
/* Get a block device.  This should only be used for single-drive devices
   (e.g. SD/Floppy/MTD).  Multi-disk devices (scsi/ide) should use the
   appropriate bus.  */
DriveInfo *drive_get_next(BlockInterfaceType type)
{
    static int next_block_unit[IF_COUNT];

    return drive_get(type, 0, next_block_unit[type]++);
}

290 291
static void bdrv_format_print(void *opaque, const char *name)
{
292
    error_printf(" %s", name);
293 294
}

295 296
typedef struct {
    QEMUBH *bh;
F
Fam Zheng 已提交
297 298
    BlockDriverState *bs;
} BDRVPutRefBH;
299

300
static int parse_block_error_action(const char *buf, bool is_read, Error **errp)
301 302
{
    if (!strcmp(buf, "ignore")) {
303
        return BLOCKDEV_ON_ERROR_IGNORE;
304
    } else if (!is_read && !strcmp(buf, "enospc")) {
305
        return BLOCKDEV_ON_ERROR_ENOSPC;
306
    } else if (!strcmp(buf, "stop")) {
307
        return BLOCKDEV_ON_ERROR_STOP;
308
    } else if (!strcmp(buf, "report")) {
309
        return BLOCKDEV_ON_ERROR_REPORT;
310
    } else {
311 312
        error_setg(errp, "'%s' invalid %s error action",
                   buf, is_read ? "read" : "write");
313 314 315 316
        return -1;
    }
}

317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355
static bool parse_stats_intervals(BlockAcctStats *stats, QList *intervals,
                                  Error **errp)
{
    const QListEntry *entry;
    for (entry = qlist_first(intervals); entry; entry = qlist_next(entry)) {
        switch (qobject_type(entry->value)) {

        case QTYPE_QSTRING: {
            unsigned long long length;
            const char *str = qstring_get_str(qobject_to_qstring(entry->value));
            if (parse_uint_full(str, &length, 10) == 0 &&
                length > 0 && length <= UINT_MAX) {
                block_acct_add_interval(stats, (unsigned) length);
            } else {
                error_setg(errp, "Invalid interval length: %s", str);
                return false;
            }
            break;
        }

        case QTYPE_QINT: {
            int64_t length = qint_get_int(qobject_to_qint(entry->value));
            if (length > 0 && length <= UINT_MAX) {
                block_acct_add_interval(stats, (unsigned) length);
            } else {
                error_setg(errp, "Invalid interval length: %" PRId64, length);
                return false;
            }
            break;
        }

        default:
            error_setg(errp, "The specification of stats-intervals is invalid");
            return false;
        }
    }
    return true;
}

356 357
typedef enum { MEDIA_DISK, MEDIA_CDROM } DriveMediaType;

358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388
/* All parameters but @opts are optional and may be set to NULL. */
static void extract_common_blockdev_options(QemuOpts *opts, int *bdrv_flags,
    const char **throttling_group, ThrottleConfig *throttle_cfg,
    BlockdevDetectZeroesOptions *detect_zeroes, Error **errp)
{
    Error *local_error = NULL;
    const char *aio;

    if (bdrv_flags) {
        if (qemu_opt_get_bool(opts, "copy-on-read", false)) {
            *bdrv_flags |= BDRV_O_COPY_ON_READ;
        }

        if ((aio = qemu_opt_get(opts, "aio")) != NULL) {
            if (!strcmp(aio, "native")) {
                *bdrv_flags |= BDRV_O_NATIVE_AIO;
            } else if (!strcmp(aio, "threads")) {
                /* this is the default */
            } else {
               error_setg(errp, "invalid aio option");
               return;
            }
        }
    }

    /* disk I/O throttling */
    if (throttling_group) {
        *throttling_group = qemu_opt_get(opts, "throttling.group");
    }

    if (throttle_cfg) {
389
        throttle_config_init(throttle_cfg);
390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415
        throttle_cfg->buckets[THROTTLE_BPS_TOTAL].avg =
            qemu_opt_get_number(opts, "throttling.bps-total", 0);
        throttle_cfg->buckets[THROTTLE_BPS_READ].avg  =
            qemu_opt_get_number(opts, "throttling.bps-read", 0);
        throttle_cfg->buckets[THROTTLE_BPS_WRITE].avg =
            qemu_opt_get_number(opts, "throttling.bps-write", 0);
        throttle_cfg->buckets[THROTTLE_OPS_TOTAL].avg =
            qemu_opt_get_number(opts, "throttling.iops-total", 0);
        throttle_cfg->buckets[THROTTLE_OPS_READ].avg =
            qemu_opt_get_number(opts, "throttling.iops-read", 0);
        throttle_cfg->buckets[THROTTLE_OPS_WRITE].avg =
            qemu_opt_get_number(opts, "throttling.iops-write", 0);

        throttle_cfg->buckets[THROTTLE_BPS_TOTAL].max =
            qemu_opt_get_number(opts, "throttling.bps-total-max", 0);
        throttle_cfg->buckets[THROTTLE_BPS_READ].max  =
            qemu_opt_get_number(opts, "throttling.bps-read-max", 0);
        throttle_cfg->buckets[THROTTLE_BPS_WRITE].max =
            qemu_opt_get_number(opts, "throttling.bps-write-max", 0);
        throttle_cfg->buckets[THROTTLE_OPS_TOTAL].max =
            qemu_opt_get_number(opts, "throttling.iops-total-max", 0);
        throttle_cfg->buckets[THROTTLE_OPS_READ].max =
            qemu_opt_get_number(opts, "throttling.iops-read-max", 0);
        throttle_cfg->buckets[THROTTLE_OPS_WRITE].max =
            qemu_opt_get_number(opts, "throttling.iops-write-max", 0);

416 417 418 419 420 421 422 423 424 425 426 427 428
        throttle_cfg->buckets[THROTTLE_BPS_TOTAL].burst_length =
            qemu_opt_get_number(opts, "throttling.bps-total-max-length", 1);
        throttle_cfg->buckets[THROTTLE_BPS_READ].burst_length  =
            qemu_opt_get_number(opts, "throttling.bps-read-max-length", 1);
        throttle_cfg->buckets[THROTTLE_BPS_WRITE].burst_length =
            qemu_opt_get_number(opts, "throttling.bps-write-max-length", 1);
        throttle_cfg->buckets[THROTTLE_OPS_TOTAL].burst_length =
            qemu_opt_get_number(opts, "throttling.iops-total-max-length", 1);
        throttle_cfg->buckets[THROTTLE_OPS_READ].burst_length =
            qemu_opt_get_number(opts, "throttling.iops-read-max-length", 1);
        throttle_cfg->buckets[THROTTLE_OPS_WRITE].burst_length =
            qemu_opt_get_number(opts, "throttling.iops-write-max-length", 1);

429 430 431
        throttle_cfg->op_size =
            qemu_opt_get_number(opts, "throttling.iops-size", 0);

432
        if (!throttle_is_valid(throttle_cfg, errp)) {
433 434 435 436 437 438 439 440
            return;
        }
    }

    if (detect_zeroes) {
        *detect_zeroes =
            qapi_enum_parse(BlockdevDetectZeroesOptions_lookup,
                            qemu_opt_get(opts, "detect-zeroes"),
441
                            BLOCKDEV_DETECT_ZEROES_OPTIONS__MAX,
442 443 444 445 446 447 448 449 450
                            BLOCKDEV_DETECT_ZEROES_OPTIONS_OFF,
                            &local_error);
        if (local_error) {
            error_propagate(errp, local_error);
            return;
        }
    }
}

451
/* Takes the ownership of bs_opts */
452 453
static BlockBackend *blockdev_init(const char *file, QDict *bs_opts,
                                   Error **errp)
454 455 456 457
{
    const char *buf;
    int bdrv_flags = 0;
    int on_read_error, on_write_error;
458
    bool account_invalid, account_failed;
459
    bool writethrough, read_only;
M
Markus Armbruster 已提交
460
    BlockBackend *blk;
461
    BlockDriverState *bs;
462
    ThrottleConfig cfg;
463
    int snapshot = 0;
464
    Error *error = NULL;
465
    QemuOpts *opts;
466 467
    QDict *interval_dict = NULL;
    QList *interval_list = NULL;
468
    const char *id;
469 470 471
    BlockdevDetectZeroesOptions detect_zeroes =
        BLOCKDEV_DETECT_ZEROES_OPTIONS_OFF;
    const char *throttling_group = NULL;
472

473 474 475
    /* Check common options by copying from bs_opts to opts, all other options
     * stay in bs_opts for processing by bdrv_open(). */
    id = qdict_get_try_str(bs_opts, "id");
476
    opts = qemu_opts_create(&qemu_common_drive_opts, id, 1, &error);
477
    if (error) {
478
        error_propagate(errp, error);
479
        goto err_no_opts;
480 481 482
    }

    qemu_opts_absorb_qdict(opts, bs_opts, &error);
483
    if (error) {
484
        error_propagate(errp, error);
485
        goto early_err;
486 487 488 489 490 491
    }

    if (id) {
        qdict_del(bs_opts, "id");
    }

492 493 494
    /* extract parameters */
    snapshot = qemu_opt_get_bool(opts, "snapshot", 0);

495 496 497
    account_invalid = qemu_opt_get_bool(opts, "stats-account-invalid", true);
    account_failed = qemu_opt_get_bool(opts, "stats-account-failed", true);

498 499
    writethrough = !qemu_opt_get_bool(opts, BDRV_OPT_CACHE_WB, true);

500 501
    id = qemu_opts_id(opts);

502 503 504 505 506 507 508 509
    qdict_extract_subqdict(bs_opts, &interval_dict, "stats-intervals.");
    qdict_array_split(interval_dict, &interval_list);

    if (qdict_size(interval_dict) != 0) {
        error_setg(errp, "Invalid option stats-intervals.%s",
                   qdict_first(interval_dict)->key);
        goto early_err;
    }
510

511 512 513 514 515
    extract_common_blockdev_options(opts, &bdrv_flags, &throttling_group, &cfg,
                                    &detect_zeroes, &error);
    if (error) {
        error_propagate(errp, error);
        goto early_err;
516 517 518
    }

    if ((buf = qemu_opt_get(opts, "format")) != NULL) {
519 520 521 522
        if (is_help_option(buf)) {
            error_printf("Supported formats:");
            bdrv_iterate_format(bdrv_format_print, NULL);
            error_printf("\n");
523
            goto early_err;
524
        }
525

526 527
        if (qdict_haskey(bs_opts, "driver")) {
            error_setg(errp, "Cannot specify both 'driver' and 'format'");
528
            goto early_err;
529
        }
530
        qdict_put(bs_opts, "driver", qstring_from_str(buf));
531 532
    }

533
    on_write_error = BLOCKDEV_ON_ERROR_ENOSPC;
534
    if ((buf = qemu_opt_get(opts, "werror")) != NULL) {
535
        on_write_error = parse_block_error_action(buf, 0, &error);
536
        if (error) {
537
            error_propagate(errp, error);
538
            goto early_err;
539 540 541
        }
    }

542
    on_read_error = BLOCKDEV_ON_ERROR_REPORT;
543
    if ((buf = qemu_opt_get(opts, "rerror")) != NULL) {
544
        on_read_error = parse_block_error_action(buf, 1, &error);
545
        if (error) {
546
            error_propagate(errp, error);
547
            goto early_err;
548 549 550
        }
    }

551
    if (snapshot) {
552
        bdrv_flags |= BDRV_O_SNAPSHOT;
553 554
    }

555 556
    read_only = qemu_opt_get_bool(opts, BDRV_OPT_READ_ONLY, false);

557
    /* init */
558
    if ((!file || !*file) && !qdict_size(bs_opts)) {
559 560
        BlockBackendRootState *blk_rs;

K
Kevin Wolf 已提交
561
        blk = blk_new(0, BLK_PERM_ALL);
562 563
        blk_rs = blk_get_root_state(blk);
        blk_rs->open_flags    = bdrv_flags;
564
        blk_rs->read_only     = read_only;
565 566
        blk_rs->detect_zeroes = detect_zeroes;

567 568 569
        QDECREF(bs_opts);
    } else {
        if (file && !*file) {
570 571
            file = NULL;
        }
572

573 574 575 576 577
        /* bdrv_open() defaults to the values in bdrv_flags (for compatibility
         * with other callers) rather than what we want as the real defaults.
         * Apply the defaults here instead. */
        qdict_set_default_str(bs_opts, BDRV_OPT_CACHE_DIRECT, "off");
        qdict_set_default_str(bs_opts, BDRV_OPT_CACHE_NO_FLUSH, "off");
578 579
        qdict_set_default_str(bs_opts, BDRV_OPT_READ_ONLY,
                              read_only ? "on" : "off");
580
        assert((bdrv_flags & BDRV_O_CACHE_MASK) == 0);
581

582 583 584 585
        if (runstate_check(RUN_STATE_INMIGRATE)) {
            bdrv_flags |= BDRV_O_INACTIVE;
        }

586
        blk = blk_new_open(file, NULL, bs_opts, bdrv_flags, errp);
587 588 589 590
        if (!blk) {
            goto err_no_bs_opts;
        }
        bs = blk_bs(blk);
591

592
        bs->detect_zeroes = detect_zeroes;
593

594 595
        if (bdrv_key_required(bs)) {
            autostart = 0;
596
        }
597 598

        block_acct_init(blk_get_stats(blk), account_invalid, account_failed);
599

600 601 602 603
        if (!parse_stats_intervals(blk_get_stats(blk), interval_list, errp)) {
            blk_unref(blk);
            blk = NULL;
            goto err_no_bs_opts;
604
        }
605 606
    }

607 608 609
    /* disk I/O throttling */
    if (throttle_enabled(&cfg)) {
        if (!throttling_group) {
610
            throttling_group = id;
611 612 613 614 615
        }
        blk_io_limits_enable(blk, throttling_group);
        blk_set_io_limits(blk, &cfg);
    }

616
    blk_set_enable_write_cache(blk, !writethrough);
617
    blk_set_on_error(blk, on_read_error, on_write_error);
618

619
    if (!monitor_add_blk(blk, id, errp)) {
620 621 622 623 624
        blk_unref(blk);
        blk = NULL;
        goto err_no_bs_opts;
    }

625
err_no_bs_opts:
626
    qemu_opts_del(opts);
627 628
    QDECREF(interval_dict);
    QDECREF(interval_list);
629
    return blk;
630

631 632
early_err:
    qemu_opts_del(opts);
633 634
    QDECREF(interval_dict);
    QDECREF(interval_list);
635 636
err_no_opts:
    QDECREF(bs_opts);
637
    return NULL;
638 639
}

640 641 642 643 644
/* Takes the ownership of bs_opts */
static BlockDriverState *bds_tree_init(QDict *bs_opts, Error **errp)
{
    int bdrv_flags = 0;

645 646 647 648 649
    /* bdrv_open() defaults to the values in bdrv_flags (for compatibility
     * with other callers) rather than what we want as the real defaults.
     * Apply the defaults here instead. */
    qdict_set_default_str(bs_opts, BDRV_OPT_CACHE_DIRECT, "off");
    qdict_set_default_str(bs_opts, BDRV_OPT_CACHE_NO_FLUSH, "off");
650
    qdict_set_default_str(bs_opts, BDRV_OPT_READ_ONLY, "off");
651

652 653 654 655
    if (runstate_check(RUN_STATE_INMIGRATE)) {
        bdrv_flags |= BDRV_O_INACTIVE;
    }

K
Kevin Wolf 已提交
656
    return bdrv_open(NULL, NULL, bs_opts, bdrv_flags, errp);
657 658
}

659 660 661 662 663 664 665 666 667 668 669 670 671
void blockdev_close_all_bdrv_states(void)
{
    BlockDriverState *bs, *next_bs;

    QTAILQ_FOREACH_SAFE(bs, &monitor_bdrv_states, monitor_list, next_bs) {
        AioContext *ctx = bdrv_get_aio_context(bs);

        aio_context_acquire(ctx);
        bdrv_unref(bs);
        aio_context_release(ctx);
    }
}

M
Max Reitz 已提交
672 673 674 675 676 677 678
/* Iterates over the list of monitor-owned BlockDriverStates */
BlockDriverState *bdrv_next_monitor_owned(BlockDriverState *bs)
{
    return bs ? QTAILQ_NEXT(bs, monitor_list)
              : QTAILQ_FIRST(&monitor_bdrv_states);
}

679 680
static void qemu_opt_rename(QemuOpts *opts, const char *from, const char *to,
                            Error **errp)
681 682 683 684 685
{
    const char *value;

    value = qemu_opt_get(opts, from);
    if (value) {
686 687 688 689 690
        if (qemu_opt_find(opts, to)) {
            error_setg(errp, "'%s' and its alias '%s' can't be used at the "
                       "same time", to, from);
            return;
        }
691 692 693 694
    }

    /* rename all items in opts */
    while ((value = qemu_opt_get(opts, from))) {
695
        qemu_opt_set(opts, to, value, &error_abort);
696 697 698 699
        qemu_opt_unset(opts, from);
    }
}

700 701 702 703 704
QemuOptsList qemu_legacy_drive_opts = {
    .name = "drive",
    .head = QTAILQ_HEAD_INITIALIZER(qemu_legacy_drive_opts.head),
    .desc = {
        {
705 706 707 708 709 710 711 712 713 714 715 716
            .name = "bus",
            .type = QEMU_OPT_NUMBER,
            .help = "bus number",
        },{
            .name = "unit",
            .type = QEMU_OPT_NUMBER,
            .help = "unit number (i.e. lun for scsi)",
        },{
            .name = "index",
            .type = QEMU_OPT_NUMBER,
            .help = "index number",
        },{
717 718 719
            .name = "media",
            .type = QEMU_OPT_STRING,
            .help = "media type (disk, cdrom)",
720 721 722 723
        },{
            .name = "if",
            .type = QEMU_OPT_STRING,
            .help = "interface (ide, scsi, sd, mtd, floppy, pflash, virtio)",
724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739
        },{
            .name = "cyls",
            .type = QEMU_OPT_NUMBER,
            .help = "number of cylinders (ide disk geometry)",
        },{
            .name = "heads",
            .type = QEMU_OPT_NUMBER,
            .help = "number of heads (ide disk geometry)",
        },{
            .name = "secs",
            .type = QEMU_OPT_NUMBER,
            .help = "number of sectors (ide disk geometry)",
        },{
            .name = "trans",
            .type = QEMU_OPT_STRING,
            .help = "chs translation (auto, lba, none)",
740 741 742 743
        },{
            .name = "boot",
            .type = QEMU_OPT_BOOL,
            .help = "(deprecated, ignored)",
744 745 746 747
        },{
            .name = "addr",
            .type = QEMU_OPT_STRING,
            .help = "pci address (virtio only)",
748 749 750 751
        },{
            .name = "serial",
            .type = QEMU_OPT_STRING,
            .help = "disk serial number",
M
Max Reitz 已提交
752 753 754 755
        },{
            .name = "file",
            .type = QEMU_OPT_STRING,
            .help = "file name",
756
        },
757 758 759

        /* Options that are passed on, but have special semantics with -drive */
        {
760
            .name = BDRV_OPT_READ_ONLY,
761 762
            .type = QEMU_OPT_BOOL,
            .help = "open drive file as read-only",
763 764 765 766 767 768 769 770
        },{
            .name = "rerror",
            .type = QEMU_OPT_STRING,
            .help = "read error action",
        },{
            .name = "werror",
            .type = QEMU_OPT_STRING,
            .help = "write error action",
771 772 773 774 775 776
        },{
            .name = "copy-on-read",
            .type = QEMU_OPT_BOOL,
            .help = "copy read data from backing file into image file",
        },

777 778 779 780
        { /* end of list */ }
    },
};

781
DriveInfo *drive_new(QemuOpts *all_opts, BlockInterfaceType block_default_type)
782
{
K
Kevin Wolf 已提交
783
    const char *value;
784
    BlockBackend *blk;
785
    DriveInfo *dinfo = NULL;
786
    QDict *bs_opts;
787 788
    QemuOpts *legacy_opts;
    DriveMediaType media = MEDIA_DISK;
789
    BlockInterfaceType type;
790
    int cyls, heads, secs, translation;
791
    int max_devs, bus_id, unit_id, index;
792
    const char *devaddr;
793
    const char *werror, *rerror;
F
Fam Zheng 已提交
794 795
    bool read_only = false;
    bool copy_on_read;
796
    const char *serial;
M
Max Reitz 已提交
797
    const char *filename;
798
    Error *local_err = NULL;
799
    int i;
K
Kevin Wolf 已提交
800

801
    /* Change legacy command line options into QMP ones */
802 803 804 805 806 807 808
    static const struct {
        const char *from;
        const char *to;
    } opt_renames[] = {
        { "iops",           "throttling.iops-total" },
        { "iops_rd",        "throttling.iops-read" },
        { "iops_wr",        "throttling.iops-write" },
809

810 811 812
        { "bps",            "throttling.bps-total" },
        { "bps_rd",         "throttling.bps-read" },
        { "bps_wr",         "throttling.bps-write" },
813

814 815 816
        { "iops_max",       "throttling.iops-total-max" },
        { "iops_rd_max",    "throttling.iops-read-max" },
        { "iops_wr_max",    "throttling.iops-write-max" },
817

818 819 820
        { "bps_max",        "throttling.bps-total-max" },
        { "bps_rd_max",     "throttling.bps-read-max" },
        { "bps_wr_max",     "throttling.bps-write-max" },
821

822
        { "iops_size",      "throttling.iops-size" },
823

824 825
        { "group",          "throttling.group" },

826
        { "readonly",       BDRV_OPT_READ_ONLY },
827 828 829
    };

    for (i = 0; i < ARRAY_SIZE(opt_renames); i++) {
830 831 832
        qemu_opt_rename(all_opts, opt_renames[i].from, opt_renames[i].to,
                        &local_err);
        if (local_err) {
833
            error_report_err(local_err);
834 835
            return NULL;
        }
836
    }
837

K
Kevin Wolf 已提交
838 839 840
    value = qemu_opt_get(all_opts, "cache");
    if (value) {
        int flags = 0;
841
        bool writethrough;
K
Kevin Wolf 已提交
842

843
        if (bdrv_parse_cache_mode(value, &flags, &writethrough) != 0) {
K
Kevin Wolf 已提交
844 845 846 847 848
            error_report("invalid cache option");
            return NULL;
        }

        /* Specific options take precedence */
849 850
        if (!qemu_opt_get(all_opts, BDRV_OPT_CACHE_WB)) {
            qemu_opt_set_bool(all_opts, BDRV_OPT_CACHE_WB,
851
                              !writethrough, &error_abort);
K
Kevin Wolf 已提交
852
        }
853 854
        if (!qemu_opt_get(all_opts, BDRV_OPT_CACHE_DIRECT)) {
            qemu_opt_set_bool(all_opts, BDRV_OPT_CACHE_DIRECT,
855
                              !!(flags & BDRV_O_NOCACHE), &error_abort);
K
Kevin Wolf 已提交
856
        }
857 858
        if (!qemu_opt_get(all_opts, BDRV_OPT_CACHE_NO_FLUSH)) {
            qemu_opt_set_bool(all_opts, BDRV_OPT_CACHE_NO_FLUSH,
859
                              !!(flags & BDRV_O_NO_FLUSH), &error_abort);
K
Kevin Wolf 已提交
860 861 862 863
        }
        qemu_opt_unset(all_opts, "cache");
    }

864 865 866 867
    /* Get a QDict for processing the options */
    bs_opts = qdict_new();
    qemu_opts_to_qdict(all_opts, bs_opts);

868 869
    legacy_opts = qemu_opts_create(&qemu_legacy_drive_opts, NULL, 0,
                                   &error_abort);
870
    qemu_opts_absorb_qdict(legacy_opts, bs_opts, &local_err);
871
    if (local_err) {
872
        error_report_err(local_err);
873 874 875
        goto fail;
    }

876 877 878 879 880 881 882
    /* Deprecated option boot=[on|off] */
    if (qemu_opt_get(legacy_opts, "boot") != NULL) {
        fprintf(stderr, "qemu-kvm: boot=on|off is deprecated and will be "
                "ignored. Future versions will reject this parameter. Please "
                "update your scripts.\n");
    }

883 884 885 886 887 888 889
    /* Media type */
    value = qemu_opt_get(legacy_opts, "media");
    if (value) {
        if (!strcmp(value, "disk")) {
            media = MEDIA_DISK;
        } else if (!strcmp(value, "cdrom")) {
            media = MEDIA_CDROM;
F
Fam Zheng 已提交
890
            read_only = true;
891 892 893 894 895 896
        } else {
            error_report("'%s' invalid media", value);
            goto fail;
        }
    }

897
    /* copy-on-read is disabled with a warning for read-only devices */
898
    read_only |= qemu_opt_get_bool(legacy_opts, BDRV_OPT_READ_ONLY, false);
899 900 901 902 903 904 905
    copy_on_read = qemu_opt_get_bool(legacy_opts, "copy-on-read", false);

    if (read_only && copy_on_read) {
        error_report("warning: disabling copy-on-read on read-only drive");
        copy_on_read = false;
    }

906
    qdict_put(bs_opts, BDRV_OPT_READ_ONLY,
907 908 909 910
              qstring_from_str(read_only ? "on" : "off"));
    qdict_put(bs_opts, "copy-on-read",
              qstring_from_str(copy_on_read ? "on" :"off"));

911 912 913 914 915 916 917 918 919 920 921 922 923 924 925
    /* Controller type */
    value = qemu_opt_get(legacy_opts, "if");
    if (value) {
        for (type = 0;
             type < IF_COUNT && strcmp(value, if_name[type]);
             type++) {
        }
        if (type == IF_COUNT) {
            error_report("unsupported bus type '%s'", value);
            goto fail;
        }
    } else {
        type = block_default_type;
    }

926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957
    /* Geometry */
    cyls  = qemu_opt_get_number(legacy_opts, "cyls", 0);
    heads = qemu_opt_get_number(legacy_opts, "heads", 0);
    secs  = qemu_opt_get_number(legacy_opts, "secs", 0);

    if (cyls || heads || secs) {
        if (cyls < 1) {
            error_report("invalid physical cyls number");
            goto fail;
        }
        if (heads < 1) {
            error_report("invalid physical heads number");
            goto fail;
        }
        if (secs < 1) {
            error_report("invalid physical secs number");
            goto fail;
        }
    }

    translation = BIOS_ATA_TRANSLATION_AUTO;
    value = qemu_opt_get(legacy_opts, "trans");
    if (value != NULL) {
        if (!cyls) {
            error_report("'%s' trans must be used with cyls, heads and secs",
                         value);
            goto fail;
        }
        if (!strcmp(value, "none")) {
            translation = BIOS_ATA_TRANSLATION_NONE;
        } else if (!strcmp(value, "lba")) {
            translation = BIOS_ATA_TRANSLATION_LBA;
958 959 960 961
        } else if (!strcmp(value, "large")) {
            translation = BIOS_ATA_TRANSLATION_LARGE;
        } else if (!strcmp(value, "rechs")) {
            translation = BIOS_ATA_TRANSLATION_RECHS;
962 963 964 965 966 967 968 969 970 971 972 973 974 975 976
        } else if (!strcmp(value, "auto")) {
            translation = BIOS_ATA_TRANSLATION_AUTO;
        } else {
            error_report("'%s' invalid translation type", value);
            goto fail;
        }
    }

    if (media == MEDIA_CDROM) {
        if (cyls || secs || heads) {
            error_report("CHS can't be set with media=cdrom");
            goto fail;
        }
    }

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
    /* Device address specified by bus/unit or index.
     * If none was specified, try to find the first free one. */
    bus_id  = qemu_opt_get_number(legacy_opts, "bus", 0);
    unit_id = qemu_opt_get_number(legacy_opts, "unit", -1);
    index   = qemu_opt_get_number(legacy_opts, "index", -1);

    max_devs = if_max_devs[type];

    if (index != -1) {
        if (bus_id != 0 || unit_id != -1) {
            error_report("index cannot be used with bus and unit");
            goto fail;
        }
        bus_id = drive_index_to_bus_id(type, index);
        unit_id = drive_index_to_unit_id(type, index);
    }

    if (unit_id == -1) {
       unit_id = 0;
       while (drive_get(type, bus_id, unit_id) != NULL) {
           unit_id++;
           if (max_devs && unit_id >= max_devs) {
               unit_id -= max_devs;
               bus_id++;
           }
       }
    }

    if (max_devs && unit_id >= max_devs) {
        error_report("unit %d too big (max is %d)", unit_id, max_devs - 1);
        goto fail;
    }

    if (drive_get(type, bus_id, unit_id) != NULL) {
        error_report("drive with bus=%d, unit=%d (index=%d) exists",
                     bus_id, unit_id, index);
        goto fail;
    }

1016 1017 1018
    /* Serial number */
    serial = qemu_opt_get(legacy_opts, "serial");

1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036
    /* no id supplied -> create one */
    if (qemu_opts_id(all_opts) == NULL) {
        char *new_id;
        const char *mediastr = "";
        if (type == IF_IDE || type == IF_SCSI) {
            mediastr = (media == MEDIA_CDROM) ? "-cd" : "-hd";
        }
        if (max_devs) {
            new_id = g_strdup_printf("%s%i%s%i", if_name[type], bus_id,
                                     mediastr, unit_id);
        } else {
            new_id = g_strdup_printf("%s%s%i", if_name[type],
                                     mediastr, unit_id);
        }
        qdict_put(bs_opts, "id", qstring_from_str(new_id));
        g_free(new_id);
    }

1037 1038 1039 1040 1041 1042 1043 1044 1045
    /* Add virtio block device */
    devaddr = qemu_opt_get(legacy_opts, "addr");
    if (devaddr && type != IF_VIRTIO) {
        error_report("addr is not supported by this bus type");
        goto fail;
    }

    if (type == IF_VIRTIO) {
        QemuOpts *devopts;
1046 1047
        devopts = qemu_opts_create(qemu_find_opts("device"), NULL, 0,
                                   &error_abort);
1048
        if (arch_type == QEMU_ARCH_S390X) {
1049
            qemu_opt_set(devopts, "driver", "virtio-blk-ccw", &error_abort);
1050
        } else {
1051
            qemu_opt_set(devopts, "driver", "virtio-blk-pci", &error_abort);
1052
        }
1053 1054
        qemu_opt_set(devopts, "drive", qdict_get_str(bs_opts, "id"),
                     &error_abort);
1055
        if (devaddr) {
1056
            qemu_opt_set(devopts, "addr", devaddr, &error_abort);
1057 1058 1059
        }
    }

M
Max Reitz 已提交
1060 1061
    filename = qemu_opt_get(legacy_opts, "file");

1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082
    /* Check werror/rerror compatibility with if=... */
    werror = qemu_opt_get(legacy_opts, "werror");
    if (werror != NULL) {
        if (type != IF_IDE && type != IF_SCSI && type != IF_VIRTIO &&
            type != IF_NONE) {
            error_report("werror is not supported by this bus type");
            goto fail;
        }
        qdict_put(bs_opts, "werror", qstring_from_str(werror));
    }

    rerror = qemu_opt_get(legacy_opts, "rerror");
    if (rerror != NULL) {
        if (type != IF_IDE && type != IF_VIRTIO && type != IF_SCSI &&
            type != IF_NONE) {
            error_report("rerror is not supported by this bus type");
            goto fail;
        }
        qdict_put(bs_opts, "rerror", qstring_from_str(rerror));
    }

1083
    /* Actual block device init: Functionality shared with blockdev-add */
1084
    blk = blockdev_init(filename, bs_opts, &local_err);
1085
    bs_opts = NULL;
1086
    if (!blk) {
1087
        if (local_err) {
1088
            error_report_err(local_err);
1089
        }
1090
        goto fail;
1091
    } else {
1092
        assert(!local_err);
1093 1094
    }

1095 1096
    /* Create legacy DriveInfo */
    dinfo = g_malloc0(sizeof(*dinfo));
1097
    dinfo->opts = all_opts;
1098

1099 1100 1101 1102 1103
    dinfo->cyls = cyls;
    dinfo->heads = heads;
    dinfo->secs = secs;
    dinfo->trans = translation;

1104
    dinfo->type = type;
1105 1106
    dinfo->bus = bus_id;
    dinfo->unit = unit_id;
1107
    dinfo->devaddr = devaddr;
1108 1109
    dinfo->serial = g_strdup(serial);

1110 1111
    blk_set_legacy_dinfo(blk, dinfo);

1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122
    switch(type) {
    case IF_IDE:
    case IF_SCSI:
    case IF_XEN:
    case IF_NONE:
        dinfo->media_cd = media == MEDIA_CDROM;
        break;
    default:
        break;
    }

1123
fail:
1124
    qemu_opts_del(legacy_opts);
1125
    QDECREF(bs_opts);
1126
    return dinfo;
1127 1128
}

1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150
static BlockDriverState *qmp_get_root_bs(const char *name, Error **errp)
{
    BlockDriverState *bs;

    bs = bdrv_lookup_bs(name, name, errp);
    if (bs == NULL) {
        return NULL;
    }

    if (!bdrv_is_root_node(bs)) {
        error_setg(errp, "Need a root block node");
        return NULL;
    }

    if (!bdrv_is_inserted(bs)) {
        error_setg(errp, "Device has no medium");
        return NULL;
    }

    return bs;
}

1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173
static BlockBackend *qmp_get_blk(const char *blk_name, const char *qdev_id,
                                 Error **errp)
{
    BlockBackend *blk;

    if (!blk_name == !qdev_id) {
        error_setg(errp, "Need exactly one of 'device' and 'id'");
        return NULL;
    }

    if (qdev_id) {
        blk = blk_by_qdev_id(qdev_id, errp);
    } else {
        blk = blk_by_name(blk_name);
        if (blk == NULL) {
            error_set(errp, ERROR_CLASS_DEVICE_NOT_FOUND,
                      "Device '%s' not found", blk_name);
        }
    }

    return blk;
}

1174
void hmp_commit(Monitor *mon, const QDict *qdict)
1175 1176
{
    const char *device = qdict_get_str(qdict, "device");
1177
    BlockBackend *blk;
1178
    int ret;
1179

1180
    if (!strcmp(device, "all")) {
1181
        ret = blk_commit_all();
1182
    } else {
1183 1184 1185
        BlockDriverState *bs;
        AioContext *aio_context;

1186 1187
        blk = blk_by_name(device);
        if (!blk) {
1188
            monitor_printf(mon, "Device '%s' not found\n", device);
1189
            return;
1190
        }
M
Max Reitz 已提交
1191 1192 1193 1194
        if (!blk_is_available(blk)) {
            monitor_printf(mon, "Device '%s' has no medium\n", device);
            return;
        }
1195 1196 1197 1198 1199 1200 1201 1202

        bs = blk_bs(blk);
        aio_context = bdrv_get_aio_context(bs);
        aio_context_acquire(aio_context);

        ret = bdrv_commit(bs);

        aio_context_release(aio_context);
1203 1204 1205 1206
    }
    if (ret < 0) {
        monitor_printf(mon, "'commit' error for '%s': %s\n", device,
                       strerror(-ret));
1207 1208 1209
    }
}

1210
static void blockdev_do_action(TransactionAction *action, Error **errp)
1211
{
1212
    TransactionActionList list;
1213

1214
    list.value = action;
1215
    list.next = NULL;
J
John Snow 已提交
1216
    qmp_transaction(&list, false, NULL, errp);
1217 1218
}

1219 1220 1221 1222 1223
void qmp_blockdev_snapshot_sync(bool has_device, const char *device,
                                bool has_node_name, const char *node_name,
                                const char *snapshot_file,
                                bool has_snapshot_node_name,
                                const char *snapshot_node_name,
1224
                                bool has_format, const char *format,
1225
                                bool has_mode, NewImageMode mode, Error **errp)
1226
{
1227
    BlockdevSnapshotSync snapshot = {
1228
        .has_device = has_device,
1229
        .device = (char *) device,
1230 1231
        .has_node_name = has_node_name,
        .node_name = (char *) node_name,
1232
        .snapshot_file = (char *) snapshot_file,
1233 1234
        .has_snapshot_node_name = has_snapshot_node_name,
        .snapshot_node_name = (char *) snapshot_node_name,
1235 1236 1237 1238 1239
        .has_format = has_format,
        .format = (char *) format,
        .has_mode = has_mode,
        .mode = mode,
    };
1240 1241
    TransactionAction action = {
        .type = TRANSACTION_ACTION_KIND_BLOCKDEV_SNAPSHOT_SYNC,
1242
        .u.blockdev_snapshot_sync.data = &snapshot,
1243 1244
    };
    blockdev_do_action(&action, errp);
1245 1246
}

1247 1248 1249 1250 1251 1252 1253
void qmp_blockdev_snapshot(const char *node, const char *overlay,
                           Error **errp)
{
    BlockdevSnapshot snapshot_data = {
        .node = (char *) node,
        .overlay = (char *) overlay
    };
1254 1255
    TransactionAction action = {
        .type = TRANSACTION_ACTION_KIND_BLOCKDEV_SNAPSHOT,
1256
        .u.blockdev_snapshot.data = &snapshot_data,
1257 1258
    };
    blockdev_do_action(&action, errp);
1259 1260
}

1261 1262 1263 1264 1265 1266 1267 1268
void qmp_blockdev_snapshot_internal_sync(const char *device,
                                         const char *name,
                                         Error **errp)
{
    BlockdevSnapshotInternal snapshot = {
        .device = (char *) device,
        .name = (char *) name
    };
1269 1270
    TransactionAction action = {
        .type = TRANSACTION_ACTION_KIND_BLOCKDEV_SNAPSHOT_INTERNAL_SYNC,
1271
        .u.blockdev_snapshot_internal_sync.data = &snapshot,
1272 1273
    };
    blockdev_do_action(&action, errp);
1274 1275
}

1276 1277 1278 1279 1280 1281 1282
SnapshotInfo *qmp_blockdev_snapshot_delete_internal_sync(const char *device,
                                                         bool has_id,
                                                         const char *id,
                                                         bool has_name,
                                                         const char *name,
                                                         Error **errp)
{
1283
    BlockDriverState *bs;
1284
    AioContext *aio_context;
1285 1286 1287 1288 1289
    QEMUSnapshotInfo sn;
    Error *local_err = NULL;
    SnapshotInfo *info = NULL;
    int ret;

1290 1291
    bs = qmp_get_root_bs(device, errp);
    if (!bs) {
1292 1293
        return NULL;
    }
1294
    aio_context = bdrv_get_aio_context(bs);
M
Max Reitz 已提交
1295
    aio_context_acquire(aio_context);
1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306

    if (!has_id) {
        id = NULL;
    }

    if (!has_name) {
        name = NULL;
    }

    if (!id && !name) {
        error_setg(errp, "Name or id must be provided");
M
Max Reitz 已提交
1307
        goto out_aio_context;
1308 1309
    }

1310 1311 1312 1313
    if (bdrv_op_is_blocked(bs, BLOCK_OP_TYPE_INTERNAL_SNAPSHOT_DELETE, errp)) {
        goto out_aio_context;
    }

1314
    ret = bdrv_snapshot_find_by_id_and_name(bs, id, name, &sn, &local_err);
1315
    if (local_err) {
1316
        error_propagate(errp, local_err);
1317
        goto out_aio_context;
1318 1319 1320 1321 1322 1323
    }
    if (!ret) {
        error_setg(errp,
                   "Snapshot with id '%s' and name '%s' does not exist on "
                   "device '%s'",
                   STR_OR_NULL(id), STR_OR_NULL(name), device);
1324
        goto out_aio_context;
1325 1326 1327
    }

    bdrv_snapshot_delete(bs, id, name, &local_err);
1328
    if (local_err) {
1329
        error_propagate(errp, local_err);
1330
        goto out_aio_context;
1331 1332
    }

1333 1334
    aio_context_release(aio_context);

1335
    info = g_new0(SnapshotInfo, 1);
1336 1337 1338 1339 1340 1341 1342 1343 1344
    info->id = g_strdup(sn.id_str);
    info->name = g_strdup(sn.name);
    info->date_nsec = sn.date_nsec;
    info->date_sec = sn.date_sec;
    info->vm_state_size = sn.vm_state_size;
    info->vm_clock_nsec = sn.vm_clock_nsec % 1000000000;
    info->vm_clock_sec = sn.vm_clock_nsec / 1000000000;

    return info;
1345 1346 1347 1348

out_aio_context:
    aio_context_release(aio_context);
    return NULL;
1349
}
1350

1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412
/**
 * block_dirty_bitmap_lookup:
 * Return a dirty bitmap (if present), after validating
 * the node reference and bitmap names.
 *
 * @node: The name of the BDS node to search for bitmaps
 * @name: The name of the bitmap to search for
 * @pbs: Output pointer for BDS lookup, if desired. Can be NULL.
 * @paio: Output pointer for aio_context acquisition, if desired. Can be NULL.
 * @errp: Output pointer for error information. Can be NULL.
 *
 * @return: A bitmap object on success, or NULL on failure.
 */
static BdrvDirtyBitmap *block_dirty_bitmap_lookup(const char *node,
                                                  const char *name,
                                                  BlockDriverState **pbs,
                                                  AioContext **paio,
                                                  Error **errp)
{
    BlockDriverState *bs;
    BdrvDirtyBitmap *bitmap;
    AioContext *aio_context;

    if (!node) {
        error_setg(errp, "Node cannot be NULL");
        return NULL;
    }
    if (!name) {
        error_setg(errp, "Bitmap name cannot be NULL");
        return NULL;
    }
    bs = bdrv_lookup_bs(node, node, NULL);
    if (!bs) {
        error_setg(errp, "Node '%s' not found", node);
        return NULL;
    }

    aio_context = bdrv_get_aio_context(bs);
    aio_context_acquire(aio_context);

    bitmap = bdrv_find_dirty_bitmap(bs, name);
    if (!bitmap) {
        error_setg(errp, "Dirty bitmap '%s' not found", name);
        goto fail;
    }

    if (pbs) {
        *pbs = bs;
    }
    if (paio) {
        *paio = aio_context;
    } else {
        aio_context_release(aio_context);
    }

    return bitmap;

 fail:
    aio_context_release(aio_context);
    return NULL;
}

1413
/* New and old BlockDriverState structs for atomic group operations */
1414

1415
typedef struct BlkActionState BlkActionState;
1416

1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431
/**
 * BlkActionOps:
 * Table of operations that define an Action.
 *
 * @instance_size: Size of state struct, in bytes.
 * @prepare: Prepare the work, must NOT be NULL.
 * @commit: Commit the changes, can be NULL.
 * @abort: Abort the changes on fail, can be NULL.
 * @clean: Clean up resources after all transaction actions have called
 *         commit() or abort(). Can be NULL.
 *
 * Only prepare() may fail. In a single transaction, only one of commit() or
 * abort() will be called. clean() will always be called if it is present.
 */
typedef struct BlkActionOps {
1432
    size_t instance_size;
1433 1434 1435 1436 1437
    void (*prepare)(BlkActionState *common, Error **errp);
    void (*commit)(BlkActionState *common);
    void (*abort)(BlkActionState *common);
    void (*clean)(BlkActionState *common);
} BlkActionOps;
1438

1439 1440 1441 1442 1443 1444
/**
 * BlkActionState:
 * Describes one Action's state within a Transaction.
 *
 * @action: QAPI-defined enum identifying which Action to perform.
 * @ops: Table of ActionOps this Action can perform.
J
John Snow 已提交
1445
 * @block_job_txn: Transaction which this action belongs to.
1446 1447 1448 1449 1450
 * @entry: List membership for all Actions in this Transaction.
 *
 * This structure must be arranged as first member in a subclassed type,
 * assuming that the compiler will also arrange it to the same offsets as the
 * base class.
1451
 */
1452
struct BlkActionState {
1453
    TransactionAction *action;
1454
    const BlkActionOps *ops;
J
John Snow 已提交
1455 1456
    BlockJobTxn *block_job_txn;
    TransactionProperties *txn_props;
1457
    QSIMPLEQ_ENTRY(BlkActionState) entry;
1458 1459
};

1460 1461
/* internal snapshot private data */
typedef struct InternalSnapshotState {
1462
    BlkActionState common;
1463
    BlockDriverState *bs;
1464
    AioContext *aio_context;
1465
    QEMUSnapshotInfo sn;
1466
    bool created;
1467 1468
} InternalSnapshotState;

J
John Snow 已提交
1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482

static int action_check_completion_mode(BlkActionState *s, Error **errp)
{
    if (s->txn_props->completion_mode != ACTION_COMPLETION_MODE_INDIVIDUAL) {
        error_setg(errp,
                   "Action '%s' does not support Transaction property "
                   "completion-mode = %s",
                   TransactionActionKind_lookup[s->action->type],
                   ActionCompletionMode_lookup[s->txn_props->completion_mode]);
        return -1;
    }
    return 0;
}

1483
static void internal_snapshot_prepare(BlkActionState *common,
1484 1485
                                      Error **errp)
{
1486
    Error *local_err = NULL;
1487 1488 1489 1490 1491 1492 1493 1494 1495 1496
    const char *device;
    const char *name;
    BlockDriverState *bs;
    QEMUSnapshotInfo old_sn, *sn;
    bool ret;
    qemu_timeval tv;
    BlockdevSnapshotInternal *internal;
    InternalSnapshotState *state;
    int ret1;

1497
    g_assert(common->action->type ==
1498
             TRANSACTION_ACTION_KIND_BLOCKDEV_SNAPSHOT_INTERNAL_SYNC);
1499
    internal = common->action->u.blockdev_snapshot_internal_sync.data;
1500 1501 1502 1503 1504 1505 1506
    state = DO_UPCAST(InternalSnapshotState, common, common);

    /* 1. parse input */
    device = internal->device;
    name = internal->name;

    /* 2. check for validation */
J
John Snow 已提交
1507 1508 1509 1510
    if (action_check_completion_mode(common, errp) < 0) {
        return;
    }

1511 1512
    bs = qmp_get_root_bs(device, errp);
    if (!bs) {
1513 1514 1515
        return;
    }

1516
    /* AioContext is released in .clean() */
1517
    state->aio_context = bdrv_get_aio_context(bs);
1518 1519
    aio_context_acquire(state->aio_context);

1520 1521 1522
    state->bs = bs;
    bdrv_drained_begin(bs);

1523 1524 1525 1526
    if (bdrv_op_is_blocked(bs, BLOCK_OP_TYPE_INTERNAL_SNAPSHOT, errp)) {
        return;
    }

1527
    if (bdrv_is_read_only(bs)) {
1528
        error_setg(errp, "Device '%s' is read only", device);
1529 1530 1531 1532
        return;
    }

    if (!bdrv_can_snapshot(bs)) {
1533 1534 1535
        error_setg(errp, "Block format '%s' used by device '%s' "
                   "does not support internal snapshots",
                   bs->drv->format_name, device);
1536 1537 1538 1539 1540 1541 1542 1543 1544
        return;
    }

    if (!strlen(name)) {
        error_setg(errp, "Name is empty");
        return;
    }

    /* check whether a snapshot with name exist */
1545 1546 1547 1548
    ret = bdrv_snapshot_find_by_id_and_name(bs, NULL, name, &old_sn,
                                            &local_err);
    if (local_err) {
        error_propagate(errp, local_err);
1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573
        return;
    } else if (ret) {
        error_setg(errp,
                   "Snapshot with name '%s' already exists on device '%s'",
                   name, device);
        return;
    }

    /* 3. take the snapshot */
    sn = &state->sn;
    pstrcpy(sn->name, sizeof(sn->name), name);
    qemu_gettimeofday(&tv);
    sn->date_sec = tv.tv_sec;
    sn->date_nsec = tv.tv_usec * 1000;
    sn->vm_clock_nsec = qemu_clock_get_ns(QEMU_CLOCK_VIRTUAL);

    ret1 = bdrv_snapshot_create(bs, sn);
    if (ret1 < 0) {
        error_setg_errno(errp, -ret1,
                         "Failed to create snapshot '%s' on device '%s'",
                         name, device);
        return;
    }

    /* 4. succeed, mark a snapshot is created */
1574
    state->created = true;
1575 1576
}

1577
static void internal_snapshot_abort(BlkActionState *common)
1578 1579 1580 1581 1582 1583 1584
{
    InternalSnapshotState *state =
                             DO_UPCAST(InternalSnapshotState, common, common);
    BlockDriverState *bs = state->bs;
    QEMUSnapshotInfo *sn = &state->sn;
    Error *local_error = NULL;

1585
    if (!state->created) {
1586 1587 1588 1589
        return;
    }

    if (bdrv_snapshot_delete(bs, sn->id_str, sn->name, &local_error) < 0) {
1590 1591 1592 1593 1594
        error_reportf_err(local_error,
                          "Failed to delete snapshot with id '%s' and "
                          "name '%s' on device '%s' in abort: ",
                          sn->id_str, sn->name,
                          bdrv_get_device_name(bs));
1595 1596 1597
    }
}

1598
static void internal_snapshot_clean(BlkActionState *common)
1599 1600 1601 1602 1603
{
    InternalSnapshotState *state = DO_UPCAST(InternalSnapshotState,
                                             common, common);

    if (state->aio_context) {
1604 1605 1606
        if (state->bs) {
            bdrv_drained_end(state->bs);
        }
1607 1608 1609 1610
        aio_context_release(state->aio_context);
    }
}

1611
/* external snapshot private data */
1612
typedef struct ExternalSnapshotState {
1613
    BlkActionState common;
1614 1615
    BlockDriverState *old_bs;
    BlockDriverState *new_bs;
1616
    AioContext *aio_context;
1617
} ExternalSnapshotState;
1618

1619
static void external_snapshot_prepare(BlkActionState *common,
1620 1621
                                      Error **errp)
{
M
Max Reitz 已提交
1622
    int flags = 0;
1623
    QDict *options = NULL;
1624
    Error *local_err = NULL;
1625
    /* Device and node name of the image to generate the snapshot from */
1626
    const char *device;
1627
    const char *node_name;
1628 1629 1630
    /* Reference to the new image (for 'blockdev-snapshot') */
    const char *snapshot_ref;
    /* File name of the new image (for 'blockdev-snapshot-sync') */
1631
    const char *new_image_file;
1632 1633
    ExternalSnapshotState *state =
                             DO_UPCAST(ExternalSnapshotState, common, common);
1634
    TransactionAction *action = common->action;
1635

1636 1637 1638 1639 1640
    /* 'blockdev-snapshot' and 'blockdev-snapshot-sync' have similar
     * purpose but a different set of parameters */
    switch (action->type) {
    case TRANSACTION_ACTION_KIND_BLOCKDEV_SNAPSHOT:
        {
1641
            BlockdevSnapshot *s = action->u.blockdev_snapshot.data;
1642 1643 1644 1645 1646 1647 1648 1649
            device = s->node;
            node_name = s->node;
            new_image_file = NULL;
            snapshot_ref = s->overlay;
        }
        break;
    case TRANSACTION_ACTION_KIND_BLOCKDEV_SNAPSHOT_SYNC:
        {
1650
            BlockdevSnapshotSync *s = action->u.blockdev_snapshot_sync.data;
1651 1652 1653 1654 1655 1656 1657 1658
            device = s->has_device ? s->device : NULL;
            node_name = s->has_node_name ? s->node_name : NULL;
            new_image_file = s->snapshot_file;
            snapshot_ref = NULL;
        }
        break;
    default:
        g_assert_not_reached();
1659 1660 1661
    }

    /* start processing */
J
John Snow 已提交
1662 1663 1664 1665
    if (action_check_completion_mode(common, errp) < 0) {
        return;
    }

1666 1667
    state->old_bs = bdrv_lookup_bs(device, node_name, errp);
    if (!state->old_bs) {
1668 1669 1670
        return;
    }

1671 1672 1673
    /* Acquire AioContext now so any threads operating on old_bs stop */
    state->aio_context = bdrv_get_aio_context(state->old_bs);
    aio_context_acquire(state->aio_context);
1674
    bdrv_drained_begin(state->old_bs);
1675

1676
    if (!bdrv_is_inserted(state->old_bs)) {
1677
        error_setg(errp, QERR_DEVICE_HAS_NO_MEDIUM, device);
1678 1679 1680
        return;
    }

1681 1682
    if (bdrv_op_is_blocked(state->old_bs,
                           BLOCK_OP_TYPE_EXTERNAL_SNAPSHOT, errp)) {
1683 1684 1685
        return;
    }

1686 1687
    if (!bdrv_is_read_only(state->old_bs)) {
        if (bdrv_flush(state->old_bs)) {
1688
            error_setg(errp, QERR_IO_ERROR);
1689 1690 1691 1692
            return;
        }
    }

1693
    if (!bdrv_is_first_non_filter(state->old_bs)) {
1694
        error_setg(errp, QERR_FEATURE_DISABLED, "snapshot");
1695 1696 1697
        return;
    }

1698
    if (action->type == TRANSACTION_ACTION_KIND_BLOCKDEV_SNAPSHOT_SYNC) {
1699
        BlockdevSnapshotSync *s = action->u.blockdev_snapshot_sync.data;
1700 1701 1702 1703
        const char *format = s->has_format ? s->format : "qcow2";
        enum NewImageMode mode;
        const char *snapshot_node_name =
            s->has_snapshot_node_name ? s->snapshot_node_name : NULL;
1704

1705 1706
        if (node_name && !snapshot_node_name) {
            error_setg(errp, "New snapshot node name missing");
1707 1708 1709
            return;
        }

1710 1711 1712 1713 1714 1715 1716
        if (snapshot_node_name &&
            bdrv_lookup_bs(snapshot_node_name, snapshot_node_name, NULL)) {
            error_setg(errp, "New snapshot node name already in use");
            return;
        }

        flags = state->old_bs->open_flags;
1717
        flags &= ~(BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING | BDRV_O_COPY_ON_READ);
1718 1719 1720 1721

        /* create new image w/backing file */
        mode = s->has_mode ? s->mode : NEW_IMAGE_MODE_ABSOLUTE_PATHS;
        if (mode != NEW_IMAGE_MODE_EXISTING) {
1722 1723 1724 1725 1726
            int64_t size = bdrv_getlength(state->old_bs);
            if (size < 0) {
                error_setg_errno(errp, -size, "bdrv_getlength failed");
                return;
            }
1727 1728 1729
            bdrv_img_create(new_image_file, format,
                            state->old_bs->filename,
                            state->old_bs->drv->format_name,
1730
                            NULL, size, flags, &local_err, false);
1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744
            if (local_err) {
                error_propagate(errp, local_err);
                return;
            }
        }

        options = qdict_new();
        if (s->has_snapshot_node_name) {
            qdict_put(options, "node-name",
                      qstring_from_str(snapshot_node_name));
        }
        qdict_put(options, "driver", qstring_from_str(format));

        flags |= BDRV_O_NO_BACKING;
1745 1746
    }

M
Max Reitz 已提交
1747 1748
    state->new_bs = bdrv_open(new_image_file, snapshot_ref, options, flags,
                              errp);
1749
    /* We will manually add the backing_hd field to the bs later */
M
Max Reitz 已提交
1750
    if (!state->new_bs) {
1751 1752 1753
        return;
    }

K
Kevin Wolf 已提交
1754
    if (bdrv_has_blk(state->new_bs)) {
1755
        error_setg(errp, "The snapshot is already in use");
1756 1757 1758 1759 1760 1761 1762 1763 1764 1765
        return;
    }

    if (bdrv_op_is_blocked(state->new_bs, BLOCK_OP_TYPE_EXTERNAL_SNAPSHOT,
                           errp)) {
        return;
    }

    if (state->new_bs->backing != NULL) {
        error_setg(errp, "The snapshot already has a backing image");
1766 1767 1768 1769 1770
        return;
    }

    if (!state->new_bs->drv->supports_backing) {
        error_setg(errp, "The snapshot does not support backing images");
1771 1772 1773
    }
}

1774
static void external_snapshot_commit(BlkActionState *common)
1775
{
1776 1777
    ExternalSnapshotState *state =
                             DO_UPCAST(ExternalSnapshotState, common, common);
1778

1779 1780
    bdrv_set_aio_context(state->new_bs, state->aio_context);

1781 1782
    /* This removes our old bs and adds the new bs */
    bdrv_append(state->new_bs, state->old_bs);
1783 1784 1785
    /* We don't need (or want) to use the transactional
     * bdrv_reopen_multiple() across all the entries at once, because we
     * don't want to abort all of them if one of them fails the reopen */
1786 1787 1788 1789
    if (!state->old_bs->copy_on_read) {
        bdrv_reopen(state->old_bs, state->old_bs->open_flags & ~BDRV_O_RDWR,
                    NULL);
    }
1790 1791
}

1792
static void external_snapshot_abort(BlkActionState *common)
1793
{
1794 1795 1796
    ExternalSnapshotState *state =
                             DO_UPCAST(ExternalSnapshotState, common, common);
    if (state->new_bs) {
F
Fam Zheng 已提交
1797
        bdrv_unref(state->new_bs);
1798
    }
1799 1800
}

1801
static void external_snapshot_clean(BlkActionState *common)
1802 1803 1804
{
    ExternalSnapshotState *state =
                             DO_UPCAST(ExternalSnapshotState, common, common);
1805
    if (state->aio_context) {
1806
        bdrv_drained_end(state->old_bs);
1807 1808
        aio_context_release(state->aio_context);
    }
1809 1810
}

1811
typedef struct DriveBackupState {
1812
    BlkActionState common;
1813
    BlockDriverState *bs;
1814
    AioContext *aio_context;
1815 1816 1817
    BlockJob *job;
} DriveBackupState;

1818
static BlockJob *do_drive_backup(DriveBackup *backup, BlockJobTxn *txn,
P
Pavel Butsykin 已提交
1819
                            Error **errp);
1820

1821
static void drive_backup_prepare(BlkActionState *common, Error **errp)
1822 1823
{
    DriveBackupState *state = DO_UPCAST(DriveBackupState, common, common);
1824
    BlockDriverState *bs;
1825 1826 1827
    DriveBackup *backup;
    Error *local_err = NULL;

1828
    assert(common->action->type == TRANSACTION_ACTION_KIND_DRIVE_BACKUP);
1829
    backup = common->action->u.drive_backup.data;
1830

1831 1832
    bs = qmp_get_root_bs(backup->device, errp);
    if (!bs) {
1833 1834 1835
        return;
    }

1836
    /* AioContext is released in .clean() */
1837
    state->aio_context = bdrv_get_aio_context(bs);
1838
    aio_context_acquire(state->aio_context);
1839 1840
    bdrv_drained_begin(bs);
    state->bs = bs;
1841

1842
    state->job = do_drive_backup(backup, common->block_job_txn, &local_err);
1843
    if (local_err) {
1844 1845 1846
        error_propagate(errp, local_err);
        return;
    }
1847
}
1848

1849 1850 1851 1852 1853
static void drive_backup_commit(BlkActionState *common)
{
    DriveBackupState *state = DO_UPCAST(DriveBackupState, common, common);
    assert(state->job);
    block_job_start(state->job);
1854 1855
}

1856
static void drive_backup_abort(BlkActionState *common)
1857 1858 1859
{
    DriveBackupState *state = DO_UPCAST(DriveBackupState, common, common);

1860 1861
    if (state->job) {
        block_job_cancel_sync(state->job);
1862 1863 1864
    }
}

1865
static void drive_backup_clean(BlkActionState *common)
1866 1867 1868 1869
{
    DriveBackupState *state = DO_UPCAST(DriveBackupState, common, common);

    if (state->aio_context) {
1870
        bdrv_drained_end(state->bs);
1871 1872 1873 1874
        aio_context_release(state->aio_context);
    }
}

1875
typedef struct BlockdevBackupState {
1876
    BlkActionState common;
1877 1878 1879 1880 1881
    BlockDriverState *bs;
    BlockJob *job;
    AioContext *aio_context;
} BlockdevBackupState;

1882 1883
static BlockJob *do_blockdev_backup(BlockdevBackup *backup, BlockJobTxn *txn,
                                    Error **errp);
1884

1885
static void blockdev_backup_prepare(BlkActionState *common, Error **errp)
1886 1887 1888
{
    BlockdevBackupState *state = DO_UPCAST(BlockdevBackupState, common, common);
    BlockdevBackup *backup;
1889
    BlockDriverState *bs, *target;
1890 1891
    Error *local_err = NULL;

1892
    assert(common->action->type == TRANSACTION_ACTION_KIND_BLOCKDEV_BACKUP);
1893
    backup = common->action->u.blockdev_backup.data;
1894

1895 1896
    bs = qmp_get_root_bs(backup->device, errp);
    if (!bs) {
1897 1898 1899
        return;
    }

1900
    target = bdrv_lookup_bs(backup->target, backup->target, errp);
M
Max Reitz 已提交
1901
    if (!target) {
1902 1903 1904 1905
        return;
    }

    /* AioContext is released in .clean() */
1906
    state->aio_context = bdrv_get_aio_context(bs);
1907
    if (state->aio_context != bdrv_get_aio_context(target)) {
1908 1909 1910 1911 1912
        state->aio_context = NULL;
        error_setg(errp, "Backup between two IO threads is not implemented");
        return;
    }
    aio_context_acquire(state->aio_context);
1913
    state->bs = bs;
1914
    bdrv_drained_begin(state->bs);
1915

1916
    state->job = do_blockdev_backup(backup, common->block_job_txn, &local_err);
1917 1918 1919 1920
    if (local_err) {
        error_propagate(errp, local_err);
        return;
    }
1921
}
1922

1923 1924 1925 1926 1927
static void blockdev_backup_commit(BlkActionState *common)
{
    BlockdevBackupState *state = DO_UPCAST(BlockdevBackupState, common, common);
    assert(state->job);
    block_job_start(state->job);
1928 1929
}

1930
static void blockdev_backup_abort(BlkActionState *common)
1931 1932 1933
{
    BlockdevBackupState *state = DO_UPCAST(BlockdevBackupState, common, common);

1934 1935
    if (state->job) {
        block_job_cancel_sync(state->job);
1936 1937 1938
    }
}

1939
static void blockdev_backup_clean(BlkActionState *common)
1940 1941 1942 1943
{
    BlockdevBackupState *state = DO_UPCAST(BlockdevBackupState, common, common);

    if (state->aio_context) {
1944
        bdrv_drained_end(state->bs);
1945 1946 1947 1948
        aio_context_release(state->aio_context);
    }
}

F
Fam Zheng 已提交
1949
typedef struct BlockDirtyBitmapState {
1950
    BlkActionState common;
F
Fam Zheng 已提交
1951 1952 1953 1954 1955 1956 1957
    BdrvDirtyBitmap *bitmap;
    BlockDriverState *bs;
    AioContext *aio_context;
    HBitmap *backup;
    bool prepared;
} BlockDirtyBitmapState;

1958
static void block_dirty_bitmap_add_prepare(BlkActionState *common,
F
Fam Zheng 已提交
1959 1960 1961 1962 1963 1964 1965
                                           Error **errp)
{
    Error *local_err = NULL;
    BlockDirtyBitmapAdd *action;
    BlockDirtyBitmapState *state = DO_UPCAST(BlockDirtyBitmapState,
                                             common, common);

J
John Snow 已提交
1966 1967 1968 1969
    if (action_check_completion_mode(common, errp) < 0) {
        return;
    }

1970
    action = common->action->u.block_dirty_bitmap_add.data;
F
Fam Zheng 已提交
1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982
    /* AIO context taken and released within qmp_block_dirty_bitmap_add */
    qmp_block_dirty_bitmap_add(action->node, action->name,
                               action->has_granularity, action->granularity,
                               &local_err);

    if (!local_err) {
        state->prepared = true;
    } else {
        error_propagate(errp, local_err);
    }
}

1983
static void block_dirty_bitmap_add_abort(BlkActionState *common)
F
Fam Zheng 已提交
1984 1985 1986 1987 1988
{
    BlockDirtyBitmapAdd *action;
    BlockDirtyBitmapState *state = DO_UPCAST(BlockDirtyBitmapState,
                                             common, common);

1989
    action = common->action->u.block_dirty_bitmap_add.data;
F
Fam Zheng 已提交
1990 1991 1992 1993 1994 1995 1996 1997
    /* Should not be able to fail: IF the bitmap was added via .prepare(),
     * then the node reference and bitmap name must have been valid.
     */
    if (state->prepared) {
        qmp_block_dirty_bitmap_remove(action->node, action->name, &error_abort);
    }
}

1998
static void block_dirty_bitmap_clear_prepare(BlkActionState *common,
F
Fam Zheng 已提交
1999 2000 2001 2002 2003 2004
                                             Error **errp)
{
    BlockDirtyBitmapState *state = DO_UPCAST(BlockDirtyBitmapState,
                                             common, common);
    BlockDirtyBitmap *action;

J
John Snow 已提交
2005 2006 2007 2008
    if (action_check_completion_mode(common, errp) < 0) {
        return;
    }

2009
    action = common->action->u.block_dirty_bitmap_clear.data;
F
Fam Zheng 已提交
2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030
    state->bitmap = block_dirty_bitmap_lookup(action->node,
                                              action->name,
                                              &state->bs,
                                              &state->aio_context,
                                              errp);
    if (!state->bitmap) {
        return;
    }

    if (bdrv_dirty_bitmap_frozen(state->bitmap)) {
        error_setg(errp, "Cannot modify a frozen bitmap");
        return;
    } else if (!bdrv_dirty_bitmap_enabled(state->bitmap)) {
        error_setg(errp, "Cannot clear a disabled bitmap");
        return;
    }

    bdrv_clear_dirty_bitmap(state->bitmap, &state->backup);
    /* AioContext is released in .clean() */
}

2031
static void block_dirty_bitmap_clear_abort(BlkActionState *common)
F
Fam Zheng 已提交
2032 2033 2034 2035 2036 2037 2038
{
    BlockDirtyBitmapState *state = DO_UPCAST(BlockDirtyBitmapState,
                                             common, common);

    bdrv_undo_clear_dirty_bitmap(state->bitmap, state->backup);
}

2039
static void block_dirty_bitmap_clear_commit(BlkActionState *common)
F
Fam Zheng 已提交
2040 2041 2042 2043 2044 2045 2046
{
    BlockDirtyBitmapState *state = DO_UPCAST(BlockDirtyBitmapState,
                                             common, common);

    hbitmap_free(state->backup);
}

2047
static void block_dirty_bitmap_clear_clean(BlkActionState *common)
F
Fam Zheng 已提交
2048 2049 2050 2051 2052 2053 2054 2055 2056
{
    BlockDirtyBitmapState *state = DO_UPCAST(BlockDirtyBitmapState,
                                             common, common);

    if (state->aio_context) {
        aio_context_release(state->aio_context);
    }
}

2057
static void abort_prepare(BlkActionState *common, Error **errp)
2058 2059 2060 2061
{
    error_setg(errp, "Transaction aborted using Abort action");
}

2062
static void abort_commit(BlkActionState *common)
2063
{
2064
    g_assert_not_reached(); /* this action never succeeds */
2065 2066
}

2067
static const BlkActionOps actions[] = {
2068 2069 2070 2071 2072
    [TRANSACTION_ACTION_KIND_BLOCKDEV_SNAPSHOT] = {
        .instance_size = sizeof(ExternalSnapshotState),
        .prepare  = external_snapshot_prepare,
        .commit   = external_snapshot_commit,
        .abort = external_snapshot_abort,
2073
        .clean = external_snapshot_clean,
2074
    },
2075
    [TRANSACTION_ACTION_KIND_BLOCKDEV_SNAPSHOT_SYNC] = {
2076
        .instance_size = sizeof(ExternalSnapshotState),
2077 2078 2079
        .prepare  = external_snapshot_prepare,
        .commit   = external_snapshot_commit,
        .abort = external_snapshot_abort,
2080
        .clean = external_snapshot_clean,
2081
    },
2082 2083 2084
    [TRANSACTION_ACTION_KIND_DRIVE_BACKUP] = {
        .instance_size = sizeof(DriveBackupState),
        .prepare = drive_backup_prepare,
2085
        .commit = drive_backup_commit,
2086
        .abort = drive_backup_abort,
2087
        .clean = drive_backup_clean,
2088
    },
2089 2090 2091
    [TRANSACTION_ACTION_KIND_BLOCKDEV_BACKUP] = {
        .instance_size = sizeof(BlockdevBackupState),
        .prepare = blockdev_backup_prepare,
2092
        .commit = blockdev_backup_commit,
2093 2094 2095
        .abort = blockdev_backup_abort,
        .clean = blockdev_backup_clean,
    },
2096
    [TRANSACTION_ACTION_KIND_ABORT] = {
2097
        .instance_size = sizeof(BlkActionState),
2098 2099 2100
        .prepare = abort_prepare,
        .commit = abort_commit,
    },
2101 2102 2103 2104
    [TRANSACTION_ACTION_KIND_BLOCKDEV_SNAPSHOT_INTERNAL_SYNC] = {
        .instance_size = sizeof(InternalSnapshotState),
        .prepare  = internal_snapshot_prepare,
        .abort = internal_snapshot_abort,
2105
        .clean = internal_snapshot_clean,
2106
    },
F
Fam Zheng 已提交
2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118
    [TRANSACTION_ACTION_KIND_BLOCK_DIRTY_BITMAP_ADD] = {
        .instance_size = sizeof(BlockDirtyBitmapState),
        .prepare = block_dirty_bitmap_add_prepare,
        .abort = block_dirty_bitmap_add_abort,
    },
    [TRANSACTION_ACTION_KIND_BLOCK_DIRTY_BITMAP_CLEAR] = {
        .instance_size = sizeof(BlockDirtyBitmapState),
        .prepare = block_dirty_bitmap_clear_prepare,
        .commit = block_dirty_bitmap_clear_commit,
        .abort = block_dirty_bitmap_clear_abort,
        .clean = block_dirty_bitmap_clear_clean,
    }
2119 2120
};

J
John Snow 已提交
2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139
/**
 * Allocate a TransactionProperties structure if necessary, and fill
 * that structure with desired defaults if they are unset.
 */
static TransactionProperties *get_transaction_properties(
    TransactionProperties *props)
{
    if (!props) {
        props = g_new0(TransactionProperties, 1);
    }

    if (!props->has_completion_mode) {
        props->has_completion_mode = true;
        props->completion_mode = ACTION_COMPLETION_MODE_INDIVIDUAL;
    }

    return props;
}

2140
/*
2141 2142
 * 'Atomic' group operations.  The operations are performed as a set, and if
 * any fail then we roll back all operations in the group.
2143
 */
J
John Snow 已提交
2144 2145 2146 2147
void qmp_transaction(TransactionActionList *dev_list,
                     bool has_props,
                     struct TransactionProperties *props,
                     Error **errp)
2148
{
2149
    TransactionActionList *dev_entry = dev_list;
J
John Snow 已提交
2150
    BlockJobTxn *block_job_txn = NULL;
2151
    BlkActionState *state, *next;
2152
    Error *local_err = NULL;
2153

2154
    QSIMPLEQ_HEAD(snap_bdrv_states, BlkActionState) snap_bdrv_states;
2155 2156
    QSIMPLEQ_INIT(&snap_bdrv_states);

J
John Snow 已提交
2157 2158 2159 2160 2161 2162 2163 2164
    /* Does this transaction get canceled as a group on failure?
     * If not, we don't really need to make a BlockJobTxn.
     */
    props = get_transaction_properties(props);
    if (props->completion_mode != ACTION_COMPLETION_MODE_INDIVIDUAL) {
        block_job_txn = block_job_txn_new();
    }

2165
    /* drain all i/o before any operations */
2166 2167
    bdrv_drain_all();

2168
    /* We don't do anything in this loop that commits us to the operations */
2169
    while (NULL != dev_entry) {
2170
        TransactionAction *dev_info = NULL;
2171
        const BlkActionOps *ops;
2172

2173 2174 2175
        dev_info = dev_entry->value;
        dev_entry = dev_entry->next;

2176
        assert(dev_info->type < ARRAY_SIZE(actions));
2177

2178
        ops = &actions[dev_info->type];
2179 2180
        assert(ops->instance_size > 0);

2181 2182 2183
        state = g_malloc0(ops->instance_size);
        state->ops = ops;
        state->action = dev_info;
J
John Snow 已提交
2184 2185
        state->block_job_txn = block_job_txn;
        state->txn_props = props;
2186
        QSIMPLEQ_INSERT_TAIL(&snap_bdrv_states, state, entry);
2187

2188
        state->ops->prepare(state, &local_err);
2189
        if (local_err) {
2190 2191
            error_propagate(errp, local_err);
            goto delete_and_fail;
2192 2193 2194
        }
    }

2195
    QSIMPLEQ_FOREACH(state, &snap_bdrv_states, entry) {
2196 2197 2198
        if (state->ops->commit) {
            state->ops->commit(state);
        }
2199 2200 2201 2202 2203 2204
    }

    /* success */
    goto exit;

delete_and_fail:
2205
    /* failure, and it is all-or-none; roll back all operations */
2206 2207 2208
    QSIMPLEQ_FOREACH(state, &snap_bdrv_states, entry) {
        if (state->ops->abort) {
            state->ops->abort(state);
2209
        }
2210 2211
    }
exit:
2212 2213 2214
    QSIMPLEQ_FOREACH_SAFE(state, &snap_bdrv_states, entry, next) {
        if (state->ops->clean) {
            state->ops->clean(state);
2215
        }
2216
        g_free(state);
2217
    }
J
John Snow 已提交
2218 2219 2220 2221
    if (!has_props) {
        qapi_free_TransactionProperties(props);
    }
    block_job_txn_unref(block_job_txn);
2222 2223
}

2224 2225 2226
void qmp_eject(bool has_device, const char *device,
               bool has_id, const char *id,
               bool has_force, bool force, Error **errp)
2227
{
2228
    Error *local_err = NULL;
2229 2230 2231 2232 2233
    int rc;

    if (!has_force) {
        force = false;
    }
2234

2235 2236 2237
    rc = do_open_tray(has_device ? device : NULL,
                      has_id ? id : NULL,
                      force, &local_err);
2238
    if (rc && rc != -ENOSYS) {
2239
        error_propagate(errp, local_err);
L
Luiz Capitulino 已提交
2240
        return;
2241
    }
2242
    error_free(local_err);
2243

2244
    qmp_x_blockdev_remove_medium(has_device, device, has_id, id, errp);
2245 2246
}

2247 2248 2249
void qmp_block_passwd(bool has_device, const char *device,
                      bool has_node_name, const char *node_name,
                      const char *password, Error **errp)
2250
{
2251
    Error *local_err = NULL;
2252
    BlockDriverState *bs;
2253
    AioContext *aio_context;
2254

2255 2256 2257
    bs = bdrv_lookup_bs(has_device ? device : NULL,
                        has_node_name ? node_name : NULL,
                        &local_err);
2258
    if (local_err) {
2259
        error_propagate(errp, local_err);
L
Luiz Capitulino 已提交
2260
        return;
2261 2262
    }

2263 2264 2265
    aio_context = bdrv_get_aio_context(bs);
    aio_context_acquire(aio_context);

2266
    bdrv_add_key(bs, password, errp);
2267 2268

    aio_context_release(aio_context);
2269 2270
}

2271 2272 2273 2274 2275 2276 2277 2278 2279 2280
/*
 * Attempt to open the tray of @device.
 * If @force, ignore its tray lock.
 * Else, if the tray is locked, don't open it, but ask the guest to open it.
 * On error, store an error through @errp and return -errno.
 * If @device does not exist, return -ENODEV.
 * If it has no removable media, return -ENOTSUP.
 * If it has no tray, return -ENOSYS.
 * If the guest was asked to open the tray, return -EINPROGRESS.
 * Else, return 0.
2281
 */
2282 2283
static int do_open_tray(const char *blk_name, const char *qdev_id,
                        bool force, Error **errp)
M
Max Reitz 已提交
2284 2285
{
    BlockBackend *blk;
2286
    const char *device = qdev_id ?: blk_name;
M
Max Reitz 已提交
2287 2288
    bool locked;

2289
    blk = qmp_get_blk(blk_name, qdev_id, errp);
M
Max Reitz 已提交
2290
    if (!blk) {
2291
        return -ENODEV;
M
Max Reitz 已提交
2292 2293 2294 2295
    }

    if (!blk_dev_has_removable_media(blk)) {
        error_setg(errp, "Device '%s' is not removable", device);
2296
        return -ENOTSUP;
M
Max Reitz 已提交
2297 2298
    }

2299
    if (!blk_dev_has_tray(blk)) {
2300 2301
        error_setg(errp, "Device '%s' does not have a tray", device);
        return -ENOSYS;
2302 2303
    }

M
Max Reitz 已提交
2304
    if (blk_dev_is_tray_open(blk)) {
2305
        return 0;
M
Max Reitz 已提交
2306 2307 2308 2309 2310 2311 2312 2313 2314 2315
    }

    locked = blk_dev_is_medium_locked(blk);
    if (locked) {
        blk_dev_eject_request(blk, force);
    }

    if (!locked || force) {
        blk_dev_change_media_cb(blk, false);
    }
2316 2317

    if (locked && !force) {
2318 2319 2320
        error_setg(errp, "Device '%s' is locked and force was not specified, "
                   "wait for tray to open and try again", device);
        return -EINPROGRESS;
2321 2322 2323 2324 2325
    }

    return 0;
}

2326 2327 2328
void qmp_blockdev_open_tray(bool has_device, const char *device,
                            bool has_id, const char *id,
                            bool has_force, bool force,
2329 2330
                            Error **errp)
{
2331 2332 2333
    Error *local_err = NULL;
    int rc;

2334 2335 2336
    if (!has_force) {
        force = false;
    }
2337 2338 2339
    rc = do_open_tray(has_device ? device : NULL,
                      has_id ? id : NULL,
                      force, &local_err);
2340 2341 2342 2343 2344
    if (rc && rc != -ENOSYS && rc != -EINPROGRESS) {
        error_propagate(errp, local_err);
        return;
    }
    error_free(local_err);
M
Max Reitz 已提交
2345 2346
}

2347 2348 2349
void qmp_blockdev_close_tray(bool has_device, const char *device,
                             bool has_id, const char *id,
                             Error **errp)
M
Max Reitz 已提交
2350 2351 2352
{
    BlockBackend *blk;

2353 2354 2355 2356
    device = has_device ? device : NULL;
    id = has_id ? id : NULL;

    blk = qmp_get_blk(device, id, errp);
M
Max Reitz 已提交
2357 2358 2359 2360 2361
    if (!blk) {
        return;
    }

    if (!blk_dev_has_removable_media(blk)) {
2362
        error_setg(errp, "Device '%s' is not removable", device ?: id);
M
Max Reitz 已提交
2363 2364 2365
        return;
    }

2366 2367 2368 2369 2370
    if (!blk_dev_has_tray(blk)) {
        /* Ignore this command on tray-less devices */
        return;
    }

M
Max Reitz 已提交
2371 2372 2373 2374 2375 2376 2377
    if (!blk_dev_is_tray_open(blk)) {
        return;
    }

    blk_dev_change_media_cb(blk, true);
}

2378 2379
void qmp_x_blockdev_remove_medium(bool has_device, const char *device,
                                  bool has_id, const char *id, Error **errp)
M
Max Reitz 已提交
2380 2381 2382 2383
{
    BlockBackend *blk;
    BlockDriverState *bs;
    AioContext *aio_context;
2384
    bool has_attached_device;
M
Max Reitz 已提交
2385

2386 2387 2388 2389
    device = has_device ? device : NULL;
    id = has_id ? id : NULL;

    blk = qmp_get_blk(device, id, errp);
M
Max Reitz 已提交
2390 2391 2392 2393 2394
    if (!blk) {
        return;
    }

    /* For BBs without a device, we can exchange the BDS tree at will */
2395
    has_attached_device = blk_get_attached_dev(blk);
M
Max Reitz 已提交
2396

2397 2398
    if (has_attached_device && !blk_dev_has_removable_media(blk)) {
        error_setg(errp, "Device '%s' is not removable", device ?: id);
M
Max Reitz 已提交
2399 2400 2401
        return;
    }

2402 2403 2404 2405
    if (has_attached_device && blk_dev_has_tray(blk) &&
        !blk_dev_is_tray_open(blk))
    {
        error_setg(errp, "Tray of device '%s' is not open", device ?: id);
M
Max Reitz 已提交
2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422
        return;
    }

    bs = blk_bs(blk);
    if (!bs) {
        return;
    }

    aio_context = bdrv_get_aio_context(bs);
    aio_context_acquire(aio_context);

    if (bdrv_op_is_blocked(bs, BLOCK_OP_TYPE_EJECT, errp)) {
        goto out;
    }

    blk_remove_bs(blk);

2423 2424 2425 2426 2427 2428 2429 2430
    if (!blk_dev_has_tray(blk)) {
        /* For tray-less devices, blockdev-open-tray is a no-op (or may not be
         * called at all); therefore, the medium needs to be ejected here.
         * Do it after blk_remove_bs() so blk_is_inserted(blk) returns the @load
         * value passed here (i.e. false). */
        blk_dev_change_media_cb(blk, false);
    }

M
Max Reitz 已提交
2431 2432 2433 2434
out:
    aio_context_release(aio_context);
}

2435
static void qmp_blockdev_insert_anon_medium(BlockBackend *blk,
M
Max Reitz 已提交
2436 2437 2438 2439 2440 2441 2442 2443
                                            BlockDriverState *bs, Error **errp)
{
    bool has_device;

    /* For BBs without a device, we can exchange the BDS tree at will */
    has_device = blk_get_attached_dev(blk);

    if (has_device && !blk_dev_has_removable_media(blk)) {
2444
        error_setg(errp, "Device is not removable");
M
Max Reitz 已提交
2445 2446 2447
        return;
    }

2448
    if (has_device && blk_dev_has_tray(blk) && !blk_dev_is_tray_open(blk)) {
2449
        error_setg(errp, "Tray of the device is not open");
M
Max Reitz 已提交
2450 2451 2452 2453
        return;
    }

    if (blk_bs(blk)) {
2454
        error_setg(errp, "There already is a medium in the device");
M
Max Reitz 已提交
2455 2456 2457 2458 2459
        return;
    }

    blk_insert_bs(blk, bs);

2460 2461 2462 2463 2464 2465 2466 2467
    if (!blk_dev_has_tray(blk)) {
        /* For tray-less devices, blockdev-close-tray is a no-op (or may not be
         * called at all); therefore, the medium needs to be pushed into the
         * slot here.
         * Do it after blk_insert_bs() so blk_is_inserted(blk) returns the @load
         * value passed here (i.e. true). */
        blk_dev_change_media_cb(blk, true);
    }
M
Max Reitz 已提交
2468 2469
}

2470 2471 2472
void qmp_x_blockdev_insert_medium(bool has_device, const char *device,
                                  bool has_id, const char *id,
                                  const char *node_name, Error **errp)
M
Max Reitz 已提交
2473
{
2474
    BlockBackend *blk;
M
Max Reitz 已提交
2475 2476
    BlockDriverState *bs;

2477 2478 2479 2480 2481 2482 2483
    blk = qmp_get_blk(has_device ? device : NULL,
                      has_id ? id : NULL,
                      errp);
    if (!blk) {
        return;
    }

M
Max Reitz 已提交
2484 2485 2486 2487 2488 2489
    bs = bdrv_find_node(node_name);
    if (!bs) {
        error_setg(errp, "Node '%s' not found", node_name);
        return;
    }

K
Kevin Wolf 已提交
2490
    if (bdrv_has_blk(bs)) {
2491
        error_setg(errp, "Node '%s' is already in use", node_name);
M
Max Reitz 已提交
2492 2493 2494
        return;
    }

2495
    qmp_blockdev_insert_anon_medium(blk, bs, errp);
M
Max Reitz 已提交
2496 2497
}

2498 2499 2500
void qmp_blockdev_change_medium(bool has_device, const char *device,
                                bool has_id, const char *id,
                                const char *filename,
M
Max Reitz 已提交
2501
                                bool has_format, const char *format,
2502 2503
                                bool has_read_only,
                                BlockdevChangeReadOnlyMode read_only,
M
Max Reitz 已提交
2504
                                Error **errp)
2505 2506 2507
{
    BlockBackend *blk;
    BlockDriverState *medium_bs = NULL;
M
Max Reitz 已提交
2508
    int bdrv_flags;
2509
    bool detect_zeroes;
2510
    int rc;
2511 2512 2513
    QDict *options = NULL;
    Error *err = NULL;

2514 2515 2516
    blk = qmp_get_blk(has_device ? device : NULL,
                      has_id ? id : NULL,
                      errp);
2517 2518 2519 2520 2521 2522 2523 2524 2525
    if (!blk) {
        goto fail;
    }

    if (blk_bs(blk)) {
        blk_update_root_state(blk);
    }

    bdrv_flags = blk_get_open_flags_from_root_state(blk);
2526 2527
    bdrv_flags &= ~(BDRV_O_TEMPORARY | BDRV_O_SNAPSHOT | BDRV_O_NO_BACKING |
        BDRV_O_PROTOCOL);
2528

2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548
    if (!has_read_only) {
        read_only = BLOCKDEV_CHANGE_READ_ONLY_MODE_RETAIN;
    }

    switch (read_only) {
    case BLOCKDEV_CHANGE_READ_ONLY_MODE_RETAIN:
        break;

    case BLOCKDEV_CHANGE_READ_ONLY_MODE_READ_ONLY:
        bdrv_flags &= ~BDRV_O_RDWR;
        break;

    case BLOCKDEV_CHANGE_READ_ONLY_MODE_READ_WRITE:
        bdrv_flags |= BDRV_O_RDWR;
        break;

    default:
        abort();
    }

2549 2550 2551 2552 2553
    options = qdict_new();
    detect_zeroes = blk_get_detect_zeroes_from_root_state(blk);
    qdict_put(options, "detect-zeroes",
              qstring_from_str(detect_zeroes ? "on" : "off"));

M
Max Reitz 已提交
2554
    if (has_format) {
2555 2556 2557
        qdict_put(options, "driver", qstring_from_str(format));
    }

M
Max Reitz 已提交
2558 2559
    medium_bs = bdrv_open(filename, NULL, options, bdrv_flags, errp);
    if (!medium_bs) {
2560 2561 2562 2563 2564 2565 2566 2567 2568
        goto fail;
    }

    bdrv_add_key(medium_bs, NULL, &err);
    if (err) {
        error_propagate(errp, err);
        goto fail;
    }

2569 2570 2571
    rc = do_open_tray(has_device ? device : NULL,
                      has_id ? id : NULL,
                      false, &err);
2572
    if (rc && rc != -ENOSYS) {
2573 2574 2575
        error_propagate(errp, err);
        goto fail;
    }
2576 2577
    error_free(err);
    err = NULL;
2578

2579
    qmp_x_blockdev_remove_medium(has_device, device, has_id, id, &err);
2580 2581 2582 2583 2584
    if (err) {
        error_propagate(errp, err);
        goto fail;
    }

2585
    qmp_blockdev_insert_anon_medium(blk, medium_bs, &err);
2586 2587 2588 2589 2590
    if (err) {
        error_propagate(errp, err);
        goto fail;
    }

2591
    qmp_blockdev_close_tray(has_device, device, has_id, id, errp);
2592 2593 2594 2595 2596 2597 2598 2599

fail:
    /* If the medium has been inserted, the device has its own reference, so
     * ours must be relinquished; and if it has not been inserted successfully,
     * the reference must be relinquished anyway */
    bdrv_unref(medium_bs);
}

Z
Zhi Yong Wu 已提交
2600
/* throttling disk I/O limits */
2601
void qmp_block_set_io_throttle(BlockIOThrottle *arg, Error **errp)
Z
Zhi Yong Wu 已提交
2602
{
2603
    ThrottleConfig cfg;
Z
Zhi Yong Wu 已提交
2604
    BlockDriverState *bs;
2605
    BlockBackend *blk;
2606
    AioContext *aio_context;
Z
Zhi Yong Wu 已提交
2607

2608 2609 2610
    blk = qmp_get_blk(arg->has_device ? arg->device : NULL,
                      arg->has_id ? arg->id : NULL,
                      errp);
2611
    if (!blk) {
2612
        return;
Z
Zhi Yong Wu 已提交
2613
    }
M
Max Reitz 已提交
2614 2615 2616 2617

    aio_context = blk_get_aio_context(blk);
    aio_context_acquire(aio_context);

2618
    bs = blk_bs(blk);
M
Max Reitz 已提交
2619
    if (!bs) {
2620
        error_setg(errp, "Device has no medium");
M
Max Reitz 已提交
2621 2622
        goto out;
    }
Z
Zhi Yong Wu 已提交
2623

2624
    throttle_config_init(&cfg);
2625 2626 2627
    cfg.buckets[THROTTLE_BPS_TOTAL].avg = arg->bps;
    cfg.buckets[THROTTLE_BPS_READ].avg  = arg->bps_rd;
    cfg.buckets[THROTTLE_BPS_WRITE].avg = arg->bps_wr;
2628

2629 2630 2631
    cfg.buckets[THROTTLE_OPS_TOTAL].avg = arg->iops;
    cfg.buckets[THROTTLE_OPS_READ].avg  = arg->iops_rd;
    cfg.buckets[THROTTLE_OPS_WRITE].avg = arg->iops_wr;
2632

2633 2634
    if (arg->has_bps_max) {
        cfg.buckets[THROTTLE_BPS_TOTAL].max = arg->bps_max;
2635
    }
2636 2637
    if (arg->has_bps_rd_max) {
        cfg.buckets[THROTTLE_BPS_READ].max = arg->bps_rd_max;
2638
    }
2639 2640
    if (arg->has_bps_wr_max) {
        cfg.buckets[THROTTLE_BPS_WRITE].max = arg->bps_wr_max;
2641
    }
2642 2643
    if (arg->has_iops_max) {
        cfg.buckets[THROTTLE_OPS_TOTAL].max = arg->iops_max;
2644
    }
2645 2646
    if (arg->has_iops_rd_max) {
        cfg.buckets[THROTTLE_OPS_READ].max = arg->iops_rd_max;
2647
    }
2648 2649
    if (arg->has_iops_wr_max) {
        cfg.buckets[THROTTLE_OPS_WRITE].max = arg->iops_wr_max;
2650
    }
Z
Zhi Yong Wu 已提交
2651

2652 2653
    if (arg->has_bps_max_length) {
        cfg.buckets[THROTTLE_BPS_TOTAL].burst_length = arg->bps_max_length;
2654
    }
2655 2656
    if (arg->has_bps_rd_max_length) {
        cfg.buckets[THROTTLE_BPS_READ].burst_length = arg->bps_rd_max_length;
2657
    }
2658 2659
    if (arg->has_bps_wr_max_length) {
        cfg.buckets[THROTTLE_BPS_WRITE].burst_length = arg->bps_wr_max_length;
2660
    }
2661 2662
    if (arg->has_iops_max_length) {
        cfg.buckets[THROTTLE_OPS_TOTAL].burst_length = arg->iops_max_length;
2663
    }
2664 2665
    if (arg->has_iops_rd_max_length) {
        cfg.buckets[THROTTLE_OPS_READ].burst_length = arg->iops_rd_max_length;
2666
    }
2667 2668
    if (arg->has_iops_wr_max_length) {
        cfg.buckets[THROTTLE_OPS_WRITE].burst_length = arg->iops_wr_max_length;
2669 2670
    }

2671 2672
    if (arg->has_iops_size) {
        cfg.op_size = arg->iops_size;
2673
    }
2674

2675
    if (!throttle_is_valid(&cfg, errp)) {
M
Max Reitz 已提交
2676
        goto out;
Z
Zhi Yong Wu 已提交
2677 2678
    }

2679 2680 2681
    if (throttle_enabled(&cfg)) {
        /* Enable I/O limits if they're not enabled yet, otherwise
         * just update the throttling group. */
2682
        if (!blk_get_public(blk)->throttle_state) {
2683
            blk_io_limits_enable(blk,
2684 2685 2686
                                 arg->has_group ? arg->group :
                                 arg->has_device ? arg->device :
                                 arg->id);
2687 2688
        } else if (arg->has_group) {
            blk_io_limits_update_group(blk, arg->group);
2689 2690
        }
        /* Set the new throttling configuration */
2691 2692
        blk_set_io_limits(blk, &cfg);
    } else if (blk_get_public(blk)->throttle_state) {
2693
        /* If all throttling settings are set to 0, disable I/O limits */
2694
        blk_io_limits_disable(blk);
Z
Zhi Yong Wu 已提交
2695
    }
2696

M
Max Reitz 已提交
2697
out:
2698
    aio_context_release(aio_context);
Z
Zhi Yong Wu 已提交
2699 2700
}

2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749
void qmp_block_dirty_bitmap_add(const char *node, const char *name,
                                bool has_granularity, uint32_t granularity,
                                Error **errp)
{
    AioContext *aio_context;
    BlockDriverState *bs;

    if (!name || name[0] == '\0') {
        error_setg(errp, "Bitmap name cannot be empty");
        return;
    }

    bs = bdrv_lookup_bs(node, node, errp);
    if (!bs) {
        return;
    }

    aio_context = bdrv_get_aio_context(bs);
    aio_context_acquire(aio_context);

    if (has_granularity) {
        if (granularity < 512 || !is_power_of_2(granularity)) {
            error_setg(errp, "Granularity must be power of 2 "
                             "and at least 512");
            goto out;
        }
    } else {
        /* Default to cluster size, if available: */
        granularity = bdrv_get_default_bitmap_granularity(bs);
    }

    bdrv_create_dirty_bitmap(bs, granularity, name, errp);

 out:
    aio_context_release(aio_context);
}

void qmp_block_dirty_bitmap_remove(const char *node, const char *name,
                                   Error **errp)
{
    AioContext *aio_context;
    BlockDriverState *bs;
    BdrvDirtyBitmap *bitmap;

    bitmap = block_dirty_bitmap_lookup(node, name, &bs, &aio_context, errp);
    if (!bitmap || !bs) {
        return;
    }

J
John Snow 已提交
2750 2751 2752 2753 2754 2755
    if (bdrv_dirty_bitmap_frozen(bitmap)) {
        error_setg(errp,
                   "Bitmap '%s' is currently frozen and cannot be removed",
                   name);
        goto out;
    }
2756
    bdrv_dirty_bitmap_make_anon(bitmap);
2757 2758
    bdrv_release_dirty_bitmap(bs, bitmap);

J
John Snow 已提交
2759
 out:
2760 2761 2762
    aio_context_release(aio_context);
}

J
John Snow 已提交
2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790
/**
 * Completely clear a bitmap, for the purposes of synchronizing a bitmap
 * immediately after a full backup operation.
 */
void qmp_block_dirty_bitmap_clear(const char *node, const char *name,
                                  Error **errp)
{
    AioContext *aio_context;
    BdrvDirtyBitmap *bitmap;
    BlockDriverState *bs;

    bitmap = block_dirty_bitmap_lookup(node, name, &bs, &aio_context, errp);
    if (!bitmap || !bs) {
        return;
    }

    if (bdrv_dirty_bitmap_frozen(bitmap)) {
        error_setg(errp,
                   "Bitmap '%s' is currently frozen and cannot be modified",
                   name);
        goto out;
    } else if (!bdrv_dirty_bitmap_enabled(bitmap)) {
        error_setg(errp,
                   "Bitmap '%s' is currently disabled and cannot be cleared",
                   name);
        goto out;
    }

F
Fam Zheng 已提交
2791
    bdrv_clear_dirty_bitmap(bitmap, NULL);
J
John Snow 已提交
2792 2793 2794 2795 2796

 out:
    aio_context_release(aio_context);
}

2797
void hmp_drive_del(Monitor *mon, const QDict *qdict)
2798 2799
{
    const char *id = qdict_get_str(qdict, "id");
2800
    BlockBackend *blk;
2801
    BlockDriverState *bs;
2802
    AioContext *aio_context;
2803
    Error *local_err = NULL;
2804

2805 2806
    bs = bdrv_find_node(id);
    if (bs) {
2807
        qmp_x_blockdev_del(id, &local_err);
2808 2809 2810 2811 2812 2813
        if (local_err) {
            error_report_err(local_err);
        }
        return;
    }

2814 2815
    blk = blk_by_name(id);
    if (!blk) {
2816
        error_report("Device '%s' not found", id);
2817
        return;
2818
    }
2819

2820
    if (!blk_legacy_dinfo(blk)) {
2821 2822
        error_report("Deleting device added with blockdev-add"
                     " is not supported");
2823
        return;
2824 2825
    }

M
Max Reitz 已提交
2826
    aio_context = blk_get_aio_context(blk);
2827 2828
    aio_context_acquire(aio_context);

M
Max Reitz 已提交
2829 2830 2831 2832 2833 2834 2835
    bs = blk_bs(blk);
    if (bs) {
        if (bdrv_op_is_blocked(bs, BLOCK_OP_TYPE_DRIVE_DEL, &local_err)) {
            error_report_err(local_err);
            aio_context_release(aio_context);
            return;
        }
2836

2837
        blk_remove_bs(blk);
M
Max Reitz 已提交
2838
    }
2839

2840
    /* Make the BlockBackend and the attached BlockDriverState anonymous */
2841 2842
    monitor_remove_blk(blk);

2843 2844
    /* If this BlockBackend has a device attached to it, its refcount will be
     * decremented when the device is removed; otherwise we have to do so here.
2845
     */
2846
    if (blk_get_attached_dev(blk)) {
2847
        /* Further I/O must not pause the guest */
2848 2849
        blk_set_on_error(blk, BLOCKDEV_ON_ERROR_REPORT,
                         BLOCKDEV_ON_ERROR_REPORT);
2850
    } else {
2851
        blk_unref(blk);
2852 2853
    }

2854
    aio_context_release(aio_context);
2855
}
2856

2857 2858 2859
void qmp_block_resize(bool has_device, const char *device,
                      bool has_node_name, const char *node_name,
                      int64_t size, Error **errp)
2860
{
2861
    Error *local_err = NULL;
2862
    BlockBackend *blk = NULL;
2863
    BlockDriverState *bs;
2864
    AioContext *aio_context;
2865
    int ret;
2866

2867 2868 2869
    bs = bdrv_lookup_bs(has_device ? device : NULL,
                        has_node_name ? node_name : NULL,
                        &local_err);
2870
    if (local_err) {
2871 2872 2873 2874
        error_propagate(errp, local_err);
        return;
    }

2875 2876 2877
    aio_context = bdrv_get_aio_context(bs);
    aio_context_acquire(aio_context);

2878
    if (!bdrv_is_first_non_filter(bs)) {
2879
        error_setg(errp, QERR_FEATURE_DISABLED, "resize");
2880
        goto out;
2881 2882 2883
    }

    if (size < 0) {
2884
        error_setg(errp, QERR_INVALID_PARAMETER_VALUE, "size", "a >0 size");
2885
        goto out;
2886 2887
    }

2888
    if (bdrv_op_is_blocked(bs, BLOCK_OP_TYPE_RESIZE, NULL)) {
2889
        error_setg(errp, QERR_DEVICE_IN_USE, device);
2890
        goto out;
2891 2892
    }

K
Kevin Wolf 已提交
2893
    blk = blk_new(BLK_PERM_RESIZE, BLK_PERM_ALL);
2894 2895
    blk_insert_bs(blk, bs);

2896 2897 2898
    /* complete all in-flight operations before resizing the device */
    bdrv_drain_all();

2899
    ret = blk_truncate(blk, size);
2900
    switch (ret) {
2901 2902 2903
    case 0:
        break;
    case -ENOMEDIUM:
2904
        error_setg(errp, QERR_DEVICE_HAS_NO_MEDIUM, device);
2905 2906
        break;
    case -ENOTSUP:
2907
        error_setg(errp, QERR_UNSUPPORTED);
2908 2909
        break;
    case -EACCES:
2910
        error_setg(errp, "Device '%s' is read only", device);
2911 2912
        break;
    case -EBUSY:
2913
        error_setg(errp, QERR_DEVICE_IN_USE, device);
2914 2915
        break;
    default:
2916
        error_setg_errno(errp, -ret, "Could not resize");
2917
        break;
2918
    }
2919 2920

out:
2921
    blk_unref(blk);
2922
    aio_context_release(aio_context);
2923
}
S
Stefan Hajnoczi 已提交
2924

2925
void qmp_block_stream(bool has_job_id, const char *job_id, const char *device,
2926
                      bool has_base, const char *base,
2927
                      bool has_base_node, const char *base_node,
2928 2929
                      bool has_backing_file, const char *backing_file,
                      bool has_speed, int64_t speed,
P
Paolo Bonzini 已提交
2930 2931
                      bool has_on_error, BlockdevOnError on_error,
                      Error **errp)
S
Stefan Hajnoczi 已提交
2932
{
2933
    BlockDriverState *bs, *iter;
2934
    BlockDriverState *base_bs = NULL;
2935
    AioContext *aio_context;
2936
    Error *local_err = NULL;
2937
    const char *base_name = NULL;
S
Stefan Hajnoczi 已提交
2938

P
Paolo Bonzini 已提交
2939 2940 2941 2942
    if (!has_on_error) {
        on_error = BLOCKDEV_ON_ERROR_REPORT;
    }

2943
    bs = bdrv_lookup_bs(device, device, errp);
2944
    if (!bs) {
S
Stefan Hajnoczi 已提交
2945 2946 2947
        return;
    }

2948
    aio_context = bdrv_get_aio_context(bs);
2949 2950
    aio_context_acquire(aio_context);

2951 2952 2953 2954 2955 2956
    if (has_base && has_base_node) {
        error_setg(errp, "'base' and 'base-node' cannot be specified "
                   "at the same time");
        goto out;
    }

2957
    if (has_base) {
2958 2959
        base_bs = bdrv_find_backing_image(bs, base);
        if (base_bs == NULL) {
2960
            error_setg(errp, QERR_BASE_NOT_FOUND, base);
2961
            goto out;
2962
        }
2963
        assert(bdrv_get_aio_context(base_bs) == aio_context);
2964
        base_name = base;
S
Stefan Hajnoczi 已提交
2965 2966
    }

2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980
    if (has_base_node) {
        base_bs = bdrv_lookup_bs(NULL, base_node, errp);
        if (!base_bs) {
            goto out;
        }
        if (bs == base_bs || !bdrv_chain_contains(bs, base_bs)) {
            error_setg(errp, "Node '%s' is not a backing image of '%s'",
                       base_node, device);
            goto out;
        }
        assert(bdrv_get_aio_context(base_bs) == aio_context);
        base_name = base_bs->filename;
    }

2981 2982 2983 2984 2985 2986 2987
    /* Check for op blockers in the whole chain between bs and base */
    for (iter = bs; iter && iter != base_bs; iter = backing_bs(iter)) {
        if (bdrv_op_is_blocked(iter, BLOCK_OP_TYPE_STREAM, errp)) {
            goto out;
        }
    }

2988 2989 2990 2991 2992
    /* if we are streaming the entire chain, the result will have no backing
     * file, and specifying one is therefore an error */
    if (base_bs == NULL && has_backing_file) {
        error_setg(errp, "backing file specified, but streaming the "
                         "entire chain");
2993
        goto out;
2994 2995 2996 2997 2998
    }

    /* backing_file string overrides base bs filename */
    base_name = has_backing_file ? backing_file : base_name;

2999
    stream_start(has_job_id ? job_id : NULL, bs, base_bs, base_name,
3000
                 has_speed ? speed : 0, on_error, &local_err);
3001
    if (local_err) {
3002
        error_propagate(errp, local_err);
3003
        goto out;
S
Stefan Hajnoczi 已提交
3004 3005 3006
    }

    trace_qmp_block_stream(bs, bs->job);
3007 3008 3009

out:
    aio_context_release(aio_context);
S
Stefan Hajnoczi 已提交
3010
}
3011

3012
void qmp_block_commit(bool has_job_id, const char *job_id, const char *device,
3013 3014
                      bool has_base, const char *base,
                      bool has_top, const char *top,
3015
                      bool has_backing_file, const char *backing_file,
3016 3017 3018 3019
                      bool has_speed, int64_t speed,
                      Error **errp)
{
    BlockDriverState *bs;
3020
    BlockDriverState *iter;
3021
    BlockDriverState *base_bs, *top_bs;
3022
    AioContext *aio_context;
3023 3024 3025 3026
    Error *local_err = NULL;
    /* This will be part of the QMP command, if/when the
     * BlockdevOnError change for blkmirror makes it in
     */
3027
    BlockdevOnError on_error = BLOCKDEV_ON_ERROR_REPORT;
3028

3029 3030 3031 3032
    if (!has_speed) {
        speed = 0;
    }

3033 3034 3035 3036 3037
    /* Important Note:
     *  libvirt relies on the DeviceNotFound error class in order to probe for
     *  live commit feature versions; for this to work, we must make sure to
     *  perform the device lookup before any generic errors that may occur in a
     *  scenario in which all optional arguments are omitted. */
3038 3039 3040 3041 3042 3043 3044 3045 3046 3047
    bs = qmp_get_root_bs(device, &local_err);
    if (!bs) {
        bs = bdrv_lookup_bs(device, device, NULL);
        if (!bs) {
            error_free(local_err);
            error_set(errp, ERROR_CLASS_DEVICE_NOT_FOUND,
                      "Device '%s' not found", device);
        } else {
            error_propagate(errp, local_err);
        }
3048
        return;
3049 3050
    }

3051
    aio_context = bdrv_get_aio_context(bs);
3052 3053
    aio_context_acquire(aio_context);

3054
    if (bdrv_op_is_blocked(bs, BLOCK_OP_TYPE_COMMIT_SOURCE, errp)) {
3055
        goto out;
3056 3057 3058 3059 3060
    }

    /* default top_bs is the active layer */
    top_bs = bs;

3061
    if (has_top && top) {
3062 3063 3064 3065 3066 3067 3068
        if (strcmp(bs->filename, top) != 0) {
            top_bs = bdrv_find_backing_image(bs, top);
        }
    }

    if (top_bs == NULL) {
        error_setg(errp, "Top image file %s not found", top ? top : "NULL");
3069
        goto out;
3070 3071
    }

3072 3073
    assert(bdrv_get_aio_context(top_bs) == aio_context);

3074 3075 3076 3077 3078 3079 3080
    if (has_base && base) {
        base_bs = bdrv_find_backing_image(top_bs, base);
    } else {
        base_bs = bdrv_find_base(top_bs);
    }

    if (base_bs == NULL) {
3081
        error_setg(errp, QERR_BASE_NOT_FOUND, base ? base : "NULL");
3082
        goto out;
3083 3084
    }

3085 3086
    assert(bdrv_get_aio_context(base_bs) == aio_context);

3087 3088 3089 3090
    for (iter = top_bs; iter != backing_bs(base_bs); iter = backing_bs(iter)) {
        if (bdrv_op_is_blocked(iter, BLOCK_OP_TYPE_COMMIT_TARGET, errp)) {
            goto out;
        }
3091 3092
    }

3093 3094 3095
    /* Do not allow attempts to commit an image into itself */
    if (top_bs == base_bs) {
        error_setg(errp, "cannot commit an image into itself");
3096
        goto out;
3097 3098
    }

F
Fam Zheng 已提交
3099
    if (top_bs == bs) {
3100 3101 3102
        if (has_backing_file) {
            error_setg(errp, "'backing-file' specified,"
                             " but 'top' is the active layer");
3103
            goto out;
3104
        }
3105
        commit_active_start(has_job_id ? job_id : NULL, bs, base_bs,
3106 3107
                            BLOCK_JOB_DEFAULT, speed, on_error, NULL, NULL,
                            &local_err, false);
F
Fam Zheng 已提交
3108
    } else {
3109 3110 3111 3112
        BlockDriverState *overlay_bs = bdrv_find_overlay(bs, top_bs);
        if (bdrv_op_is_blocked(overlay_bs, BLOCK_OP_TYPE_COMMIT_TARGET, errp)) {
            goto out;
        }
3113
        commit_start(has_job_id ? job_id : NULL, bs, base_bs, top_bs, speed,
3114 3115
                     on_error, has_backing_file ? backing_file : NULL,
                     &local_err);
F
Fam Zheng 已提交
3116
    }
3117 3118
    if (local_err != NULL) {
        error_propagate(errp, local_err);
3119
        goto out;
3120
    }
3121 3122 3123

out:
    aio_context_release(aio_context);
3124 3125
}

3126 3127
static BlockJob *do_drive_backup(DriveBackup *backup, BlockJobTxn *txn,
                                 Error **errp)
3128 3129 3130
{
    BlockDriverState *bs;
    BlockDriverState *target_bs;
3131
    BlockDriverState *source = NULL;
3132
    BlockJob *job = NULL;
3133
    BdrvDirtyBitmap *bmap = NULL;
3134
    AioContext *aio_context;
3135
    QDict *options = NULL;
3136 3137 3138 3139
    Error *local_err = NULL;
    int flags;
    int64_t size;

P
Pavel Butsykin 已提交
3140 3141
    if (!backup->has_speed) {
        backup->speed = 0;
3142
    }
P
Pavel Butsykin 已提交
3143 3144
    if (!backup->has_on_source_error) {
        backup->on_source_error = BLOCKDEV_ON_ERROR_REPORT;
3145
    }
P
Pavel Butsykin 已提交
3146 3147
    if (!backup->has_on_target_error) {
        backup->on_target_error = BLOCKDEV_ON_ERROR_REPORT;
3148
    }
P
Pavel Butsykin 已提交
3149 3150 3151 3152 3153
    if (!backup->has_mode) {
        backup->mode = NEW_IMAGE_MODE_ABSOLUTE_PATHS;
    }
    if (!backup->has_job_id) {
        backup->job_id = NULL;
3154
    }
3155 3156 3157
    if (!backup->has_compress) {
        backup->compress = false;
    }
3158

P
Pavel Butsykin 已提交
3159
    bs = qmp_get_root_bs(backup->device, errp);
3160
    if (!bs) {
3161
        return NULL;
3162 3163
    }

3164
    aio_context = bdrv_get_aio_context(bs);
3165 3166
    aio_context_acquire(aio_context);

P
Pavel Butsykin 已提交
3167 3168 3169
    if (!backup->has_format) {
        backup->format = backup->mode == NEW_IMAGE_MODE_EXISTING ?
                         NULL : (char*) bs->drv->format_name;
3170 3171
    }

F
Fam Zheng 已提交
3172
    /* Early check to avoid creating target */
3173
    if (bdrv_op_is_blocked(bs, BLOCK_OP_TYPE_BACKUP_SOURCE, errp)) {
3174
        goto out;
3175 3176
    }

K
Kevin Wolf 已提交
3177
    flags = bs->open_flags | BDRV_O_RDWR;
3178

3179 3180
    /* See if we have a backing HD we can use to create our new image
     * on top of. */
P
Pavel Butsykin 已提交
3181
    if (backup->sync == MIRROR_SYNC_MODE_TOP) {
3182
        source = backing_bs(bs);
3183
        if (!source) {
P
Pavel Butsykin 已提交
3184
            backup->sync = MIRROR_SYNC_MODE_FULL;
3185 3186
        }
    }
P
Pavel Butsykin 已提交
3187
    if (backup->sync == MIRROR_SYNC_MODE_NONE) {
3188 3189 3190
        source = bs;
    }

3191 3192 3193
    size = bdrv_getlength(bs);
    if (size < 0) {
        error_setg_errno(errp, -size, "bdrv_getlength failed");
3194
        goto out;
3195 3196
    }

P
Pavel Butsykin 已提交
3197 3198
    if (backup->mode != NEW_IMAGE_MODE_EXISTING) {
        assert(backup->format);
3199
        if (source) {
P
Pavel Butsykin 已提交
3200
            bdrv_img_create(backup->target, backup->format, source->filename,
3201 3202 3203
                            source->drv->format_name, NULL,
                            size, flags, &local_err, false);
        } else {
P
Pavel Butsykin 已提交
3204
            bdrv_img_create(backup->target, backup->format, NULL, NULL, NULL,
3205 3206
                            size, flags, &local_err, false);
        }
3207 3208
    }

3209
    if (local_err) {
3210
        error_propagate(errp, local_err);
3211
        goto out;
3212 3213
    }

P
Pavel Butsykin 已提交
3214
    if (backup->format) {
3215
        options = qdict_new();
P
Pavel Butsykin 已提交
3216
        qdict_put(options, "driver", qstring_from_str(backup->format));
3217 3218
    }

P
Pavel Butsykin 已提交
3219
    target_bs = bdrv_open(backup->target, NULL, options, flags, errp);
M
Max Reitz 已提交
3220
    if (!target_bs) {
3221
        goto out;
3222 3223
    }

3224 3225
    bdrv_set_aio_context(target_bs, aio_context);

P
Pavel Butsykin 已提交
3226 3227
    if (backup->has_bitmap) {
        bmap = bdrv_find_dirty_bitmap(bs, backup->bitmap);
3228
        if (!bmap) {
P
Pavel Butsykin 已提交
3229
            error_setg(errp, "Bitmap '%s' could not be found", backup->bitmap);
3230
            bdrv_unref(target_bs);
3231 3232 3233 3234
            goto out;
        }
    }

3235 3236 3237 3238
    job = backup_job_create(backup->job_id, bs, target_bs, backup->speed,
                            backup->sync, bmap, backup->compress,
                            backup->on_source_error, backup->on_target_error,
                            BLOCK_JOB_DEFAULT, NULL, NULL, txn, &local_err);
K
Kevin Wolf 已提交
3239
    bdrv_unref(target_bs);
3240 3241
    if (local_err != NULL) {
        error_propagate(errp, local_err);
3242
        goto out;
3243
    }
3244 3245 3246

out:
    aio_context_release(aio_context);
3247
    return job;
3248 3249
}

P
Pavel Butsykin 已提交
3250
void qmp_drive_backup(DriveBackup *arg, Error **errp)
3251
{
3252 3253 3254 3255 3256 3257

    BlockJob *job;
    job = do_drive_backup(arg, NULL, errp);
    if (job) {
        block_job_start(job);
    }
3258 3259
}

3260 3261
BlockDeviceInfoList *qmp_query_named_block_nodes(Error **errp)
{
3262
    return bdrv_named_nodes_list(errp);
3263 3264
}

3265 3266
BlockJob *do_blockdev_backup(BlockdevBackup *backup, BlockJobTxn *txn,
                             Error **errp)
F
Fam Zheng 已提交
3267 3268 3269 3270 3271
{
    BlockDriverState *bs;
    BlockDriverState *target_bs;
    Error *local_err = NULL;
    AioContext *aio_context;
3272
    BlockJob *job = NULL;
F
Fam Zheng 已提交
3273

P
Pavel Butsykin 已提交
3274 3275
    if (!backup->has_speed) {
        backup->speed = 0;
F
Fam Zheng 已提交
3276
    }
P
Pavel Butsykin 已提交
3277 3278 3279 3280 3281
    if (!backup->has_on_source_error) {
        backup->on_source_error = BLOCKDEV_ON_ERROR_REPORT;
    }
    if (!backup->has_on_target_error) {
        backup->on_target_error = BLOCKDEV_ON_ERROR_REPORT;
F
Fam Zheng 已提交
3282
    }
P
Pavel Butsykin 已提交
3283 3284
    if (!backup->has_job_id) {
        backup->job_id = NULL;
F
Fam Zheng 已提交
3285
    }
3286 3287 3288
    if (!backup->has_compress) {
        backup->compress = false;
    }
F
Fam Zheng 已提交
3289

P
Pavel Butsykin 已提交
3290
    bs = qmp_get_root_bs(backup->device, errp);
3291
    if (!bs) {
3292
        return NULL;
F
Fam Zheng 已提交
3293 3294
    }

3295
    aio_context = bdrv_get_aio_context(bs);
F
Fam Zheng 已提交
3296 3297
    aio_context_acquire(aio_context);

P
Pavel Butsykin 已提交
3298
    target_bs = bdrv_lookup_bs(backup->target, backup->target, errp);
3299
    if (!target_bs) {
M
Max Reitz 已提交
3300 3301
        goto out;
    }
F
Fam Zheng 已提交
3302

3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313
    if (bdrv_get_aio_context(target_bs) != aio_context) {
        if (!bdrv_has_blk(target_bs)) {
            /* The target BDS is not attached, we can safely move it to another
             * AioContext. */
            bdrv_set_aio_context(target_bs, aio_context);
        } else {
            error_setg(errp, "Target is attached to a different thread from "
                             "source.");
            goto out;
        }
    }
3314 3315 3316 3317
    job = backup_job_create(backup->job_id, bs, target_bs, backup->speed,
                            backup->sync, NULL, backup->compress,
                            backup->on_source_error, backup->on_target_error,
                            BLOCK_JOB_DEFAULT, NULL, NULL, txn, &local_err);
F
Fam Zheng 已提交
3318 3319 3320 3321 3322
    if (local_err != NULL) {
        error_propagate(errp, local_err);
    }
out:
    aio_context_release(aio_context);
3323
    return job;
F
Fam Zheng 已提交
3324 3325
}

P
Pavel Butsykin 已提交
3326 3327
void qmp_blockdev_backup(BlockdevBackup *arg, Error **errp)
{
3328 3329 3330 3331 3332
    BlockJob *job;
    job = do_blockdev_backup(arg, NULL, errp);
    if (job) {
        block_job_start(job);
    }
3333 3334
}

3335 3336 3337
/* Parameter check and block job starting for drive mirroring.
 * Caller should hold @device and @target's aio context (must be the same).
 **/
3338
static void blockdev_mirror_common(const char *job_id, BlockDriverState *bs,
3339 3340 3341
                                   BlockDriverState *target,
                                   bool has_replaces, const char *replaces,
                                   enum MirrorSyncMode sync,
M
Max Reitz 已提交
3342
                                   BlockMirrorBackingMode backing_mode,
3343 3344 3345 3346 3347 3348 3349 3350 3351
                                   bool has_speed, int64_t speed,
                                   bool has_granularity, uint32_t granularity,
                                   bool has_buf_size, int64_t buf_size,
                                   bool has_on_source_error,
                                   BlockdevOnError on_source_error,
                                   bool has_on_target_error,
                                   BlockdevOnError on_target_error,
                                   bool has_unmap, bool unmap,
                                   Error **errp)
P
Paolo Bonzini 已提交
3352 3353 3354 3355 3356
{

    if (!has_speed) {
        speed = 0;
    }
3357 3358 3359 3360 3361 3362
    if (!has_on_source_error) {
        on_source_error = BLOCKDEV_ON_ERROR_REPORT;
    }
    if (!has_on_target_error) {
        on_target_error = BLOCKDEV_ON_ERROR_REPORT;
    }
3363 3364 3365
    if (!has_granularity) {
        granularity = 0;
    }
3366
    if (!has_buf_size) {
W
Wen Congyang 已提交
3367
        buf_size = 0;
3368
    }
3369 3370 3371
    if (!has_unmap) {
        unmap = true;
    }
3372

3373
    if (granularity != 0 && (granularity < 512 || granularity > 1048576 * 64)) {
3374 3375
        error_setg(errp, QERR_INVALID_PARAMETER_VALUE, "granularity",
                   "a value in range [512B, 64MB]");
3376 3377 3378
        return;
    }
    if (granularity & (granularity - 1)) {
3379 3380
        error_setg(errp, QERR_INVALID_PARAMETER_VALUE, "granularity",
                   "power of 2");
3381 3382
        return;
    }
P
Paolo Bonzini 已提交
3383

3384 3385 3386
    if (bdrv_op_is_blocked(bs, BLOCK_OP_TYPE_MIRROR_SOURCE, errp)) {
        return;
    }
F
Fam Zheng 已提交
3387 3388 3389
    if (bdrv_op_is_blocked(target, BLOCK_OP_TYPE_MIRROR_TARGET, errp)) {
        return;
    }
3390 3391 3392 3393 3394 3395 3396 3397

    if (!bs->backing && sync == MIRROR_SYNC_MODE_TOP) {
        sync = MIRROR_SYNC_MODE_FULL;
    }

    /* pass the node name to replace to mirror start since it's loose coupling
     * and will allow to check whether the node still exist at mirror completion
     */
3398
    mirror_start(job_id, bs, target,
3399
                 has_replaces ? replaces : NULL,
M
Max Reitz 已提交
3400
                 speed, granularity, buf_size, sync, backing_mode,
3401
                 on_source_error, on_target_error, unmap, errp);
3402 3403
}

E
Eric Blake 已提交
3404
void qmp_drive_mirror(DriveMirror *arg, Error **errp)
3405 3406 3407 3408
{
    BlockDriverState *bs;
    BlockDriverState *source, *target_bs;
    AioContext *aio_context;
M
Max Reitz 已提交
3409
    BlockMirrorBackingMode backing_mode;
3410 3411 3412 3413
    Error *local_err = NULL;
    QDict *options = NULL;
    int flags;
    int64_t size;
E
Eric Blake 已提交
3414
    const char *format = arg->format;
3415

3416 3417
    bs = qmp_get_root_bs(arg->device, errp);
    if (!bs) {
P
Paolo Bonzini 已提交
3418 3419 3420
        return;
    }

3421
    aio_context = bdrv_get_aio_context(bs);
3422 3423
    aio_context_acquire(aio_context);

E
Eric Blake 已提交
3424 3425
    if (!arg->has_mode) {
        arg->mode = NEW_IMAGE_MODE_ABSOLUTE_PATHS;
3426
    }
P
Paolo Bonzini 已提交
3427

E
Eric Blake 已提交
3428 3429 3430
    if (!arg->has_format) {
        format = (arg->mode == NEW_IMAGE_MODE_EXISTING
                  ? NULL : bs->drv->format_name);
P
Paolo Bonzini 已提交
3431 3432
    }

K
Kevin Wolf 已提交
3433
    flags = bs->open_flags | BDRV_O_RDWR;
3434
    source = backing_bs(bs);
E
Eric Blake 已提交
3435 3436
    if (!source && arg->sync == MIRROR_SYNC_MODE_TOP) {
        arg->sync = MIRROR_SYNC_MODE_FULL;
P
Paolo Bonzini 已提交
3437
    }
E
Eric Blake 已提交
3438
    if (arg->sync == MIRROR_SYNC_MODE_NONE) {
3439 3440
        source = bs;
    }
P
Paolo Bonzini 已提交
3441

3442 3443 3444
    size = bdrv_getlength(bs);
    if (size < 0) {
        error_setg_errno(errp, -size, "bdrv_getlength failed");
3445
        goto out;
3446 3447
    }

E
Eric Blake 已提交
3448
    if (arg->has_replaces) {
3449
        BlockDriverState *to_replace_bs;
3450 3451
        AioContext *replace_aio_context;
        int64_t replace_size;
3452

E
Eric Blake 已提交
3453
        if (!arg->has_node_name) {
3454 3455
            error_setg(errp, "a node-name must be provided when replacing a"
                             " named node of the graph");
3456
            goto out;
3457 3458
        }

E
Eric Blake 已提交
3459
        to_replace_bs = check_to_replace_node(bs, arg->replaces, &local_err);
3460 3461 3462

        if (!to_replace_bs) {
            error_propagate(errp, local_err);
3463
            goto out;
3464 3465
        }

3466 3467 3468 3469 3470 3471
        replace_aio_context = bdrv_get_aio_context(to_replace_bs);
        aio_context_acquire(replace_aio_context);
        replace_size = bdrv_getlength(to_replace_bs);
        aio_context_release(replace_aio_context);

        if (size != replace_size) {
3472 3473
            error_setg(errp, "cannot replace image with a mirror image of "
                             "different size");
3474
            goto out;
3475 3476 3477
        }
    }

E
Eric Blake 已提交
3478
    if (arg->mode == NEW_IMAGE_MODE_ABSOLUTE_PATHS) {
M
Max Reitz 已提交
3479 3480 3481 3482 3483
        backing_mode = MIRROR_SOURCE_BACKING_CHAIN;
    } else {
        backing_mode = MIRROR_OPEN_BACKING_CHAIN;
    }

E
Eric Blake 已提交
3484 3485
    if ((arg->sync == MIRROR_SYNC_MODE_FULL || !source)
        && arg->mode != NEW_IMAGE_MODE_EXISTING)
3486
    {
P
Paolo Bonzini 已提交
3487
        /* create new image w/o backing file */
3488
        assert(format);
E
Eric Blake 已提交
3489
        bdrv_img_create(arg->target, format,
3490
                        NULL, NULL, NULL, size, flags, &local_err, false);
P
Paolo Bonzini 已提交
3491
    } else {
E
Eric Blake 已提交
3492
        switch (arg->mode) {
P
Paolo Bonzini 已提交
3493 3494 3495 3496
        case NEW_IMAGE_MODE_EXISTING:
            break;
        case NEW_IMAGE_MODE_ABSOLUTE_PATHS:
            /* create new image with backing file */
E
Eric Blake 已提交
3497
            bdrv_img_create(arg->target, format,
3498 3499
                            source->filename,
                            source->drv->format_name,
3500
                            NULL, size, flags, &local_err, false);
P
Paolo Bonzini 已提交
3501 3502 3503 3504 3505 3506
            break;
        default:
            abort();
        }
    }

3507
    if (local_err) {
3508
        error_propagate(errp, local_err);
3509
        goto out;
P
Paolo Bonzini 已提交
3510 3511
    }

3512
    options = qdict_new();
E
Eric Blake 已提交
3513 3514
    if (arg->has_node_name) {
        qdict_put(options, "node-name", qstring_from_str(arg->node_name));
3515
    }
3516 3517 3518
    if (format) {
        qdict_put(options, "driver", qstring_from_str(format));
    }
3519

3520 3521 3522
    /* Mirroring takes care of copy-on-write using the source's backing
     * file.
     */
E
Eric Blake 已提交
3523 3524
    target_bs = bdrv_open(arg->target, NULL, options,
                          flags | BDRV_O_NO_BACKING, errp);
M
Max Reitz 已提交
3525
    if (!target_bs) {
3526
        goto out;
P
Paolo Bonzini 已提交
3527 3528
    }

3529 3530
    bdrv_set_aio_context(target_bs, aio_context);

E
Eric Blake 已提交
3531 3532 3533 3534 3535 3536 3537 3538
    blockdev_mirror_common(arg->has_job_id ? arg->job_id : NULL, bs, target_bs,
                           arg->has_replaces, arg->replaces, arg->sync,
                           backing_mode, arg->has_speed, arg->speed,
                           arg->has_granularity, arg->granularity,
                           arg->has_buf_size, arg->buf_size,
                           arg->has_on_source_error, arg->on_source_error,
                           arg->has_on_target_error, arg->on_target_error,
                           arg->has_unmap, arg->unmap,
3539
                           &local_err);
K
Kevin Wolf 已提交
3540
    bdrv_unref(target_bs);
3541
    error_propagate(errp, local_err);
3542 3543
out:
    aio_context_release(aio_context);
P
Paolo Bonzini 已提交
3544 3545
}

3546 3547
void qmp_blockdev_mirror(bool has_job_id, const char *job_id,
                         const char *device, const char *target,
F
Fam Zheng 已提交
3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561
                         bool has_replaces, const char *replaces,
                         MirrorSyncMode sync,
                         bool has_speed, int64_t speed,
                         bool has_granularity, uint32_t granularity,
                         bool has_buf_size, int64_t buf_size,
                         bool has_on_source_error,
                         BlockdevOnError on_source_error,
                         bool has_on_target_error,
                         BlockdevOnError on_target_error,
                         Error **errp)
{
    BlockDriverState *bs;
    BlockDriverState *target_bs;
    AioContext *aio_context;
M
Max Reitz 已提交
3562
    BlockMirrorBackingMode backing_mode = MIRROR_LEAVE_BACKING_CHAIN;
F
Fam Zheng 已提交
3563 3564
    Error *local_err = NULL;

3565
    bs = qmp_get_root_bs(device, errp);
F
Fam Zheng 已提交
3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579
    if (!bs) {
        return;
    }

    target_bs = bdrv_lookup_bs(target, target, errp);
    if (!target_bs) {
        return;
    }

    aio_context = bdrv_get_aio_context(bs);
    aio_context_acquire(aio_context);

    bdrv_set_aio_context(target_bs, aio_context);

3580
    blockdev_mirror_common(has_job_id ? job_id : NULL, bs, target_bs,
M
Max Reitz 已提交
3581
                           has_replaces, replaces, sync, backing_mode,
F
Fam Zheng 已提交
3582 3583 3584 3585 3586 3587 3588
                           has_speed, speed,
                           has_granularity, granularity,
                           has_buf_size, buf_size,
                           has_on_source_error, on_source_error,
                           has_on_target_error, on_target_error,
                           true, true,
                           &local_err);
3589
    error_propagate(errp, local_err);
F
Fam Zheng 已提交
3590 3591 3592 3593

    aio_context_release(aio_context);
}

3594 3595
/* Get a block job using its ID and acquire its AioContext */
static BlockJob *find_block_job(const char *id, AioContext **aio_context,
3596
                                Error **errp)
3597
{
3598
    BlockJob *job;
3599

3600
    assert(id != NULL);
M
Max Reitz 已提交
3601

3602
    *aio_context = NULL;
3603

3604
    job = block_job_get(id);
M
Max Reitz 已提交
3605

3606 3607 3608 3609
    if (!job) {
        error_set(errp, ERROR_CLASS_DEVICE_NOT_ACTIVE,
                  "Block job '%s' not found", id);
        return NULL;
3610
    }
3611

3612 3613
    *aio_context = blk_get_aio_context(job->blk);
    aio_context_acquire(*aio_context);
3614

3615
    return job;
3616 3617
}

3618
void qmp_block_job_set_speed(const char *device, int64_t speed, Error **errp)
3619
{
3620
    AioContext *aio_context;
3621
    BlockJob *job = find_block_job(device, &aio_context, errp);
3622 3623 3624 3625 3626

    if (!job) {
        return;
    }

3627
    block_job_set_speed(job, speed, errp);
3628
    aio_context_release(aio_context);
3629
}
3630

3631 3632
void qmp_block_job_cancel(const char *device,
                          bool has_force, bool force, Error **errp)
3633
{
3634
    AioContext *aio_context;
3635
    BlockJob *job = find_block_job(device, &aio_context, errp);
3636

3637 3638 3639
    if (!job) {
        return;
    }
3640 3641 3642 3643 3644

    if (!has_force) {
        force = false;
    }

3645
    if (block_job_user_paused(job) && !force) {
3646 3647
        error_setg(errp, "The block job for device '%s' is currently paused",
                   device);
3648
        goto out;
3649
    }
3650 3651 3652

    trace_qmp_block_job_cancel(job);
    block_job_cancel(job);
3653 3654
out:
    aio_context_release(aio_context);
3655
}
S
Stefan Hajnoczi 已提交
3656

3657 3658
void qmp_block_job_pause(const char *device, Error **errp)
{
3659
    AioContext *aio_context;
3660
    BlockJob *job = find_block_job(device, &aio_context, errp);
3661

3662
    if (!job || block_job_user_paused(job)) {
3663 3664 3665 3666
        return;
    }

    trace_qmp_block_job_pause(job);
3667
    block_job_user_pause(job);
3668
    aio_context_release(aio_context);
3669 3670 3671 3672
}

void qmp_block_job_resume(const char *device, Error **errp)
{
3673
    AioContext *aio_context;
3674
    BlockJob *job = find_block_job(device, &aio_context, errp);
3675

3676
    if (!job || !block_job_user_paused(job)) {
3677 3678 3679 3680
        return;
    }

    trace_qmp_block_job_resume(job);
3681
    block_job_iostatus_reset(job);
3682
    block_job_user_resume(job);
3683
    aio_context_release(aio_context);
3684 3685
}

P
Paolo Bonzini 已提交
3686 3687
void qmp_block_job_complete(const char *device, Error **errp)
{
3688
    AioContext *aio_context;
3689
    BlockJob *job = find_block_job(device, &aio_context, errp);
P
Paolo Bonzini 已提交
3690 3691 3692 3693 3694 3695 3696

    if (!job) {
        return;
    }

    trace_qmp_block_job_complete(job);
    block_job_complete(job, errp);
3697
    aio_context_release(aio_context);
P
Paolo Bonzini 已提交
3698 3699
}

3700 3701 3702 3703 3704 3705
void qmp_change_backing_file(const char *device,
                             const char *image_node_name,
                             const char *backing_file,
                             Error **errp)
{
    BlockDriverState *bs = NULL;
3706
    AioContext *aio_context;
3707 3708 3709 3710 3711 3712
    BlockDriverState *image_bs = NULL;
    Error *local_err = NULL;
    bool ro;
    int open_flags;
    int ret;

3713 3714
    bs = qmp_get_root_bs(device, errp);
    if (!bs) {
3715 3716 3717
        return;
    }

3718
    aio_context = bdrv_get_aio_context(bs);
3719 3720
    aio_context_acquire(aio_context);

3721 3722 3723
    image_bs = bdrv_lookup_bs(NULL, image_node_name, &local_err);
    if (local_err) {
        error_propagate(errp, local_err);
3724
        goto out;
3725 3726 3727 3728
    }

    if (!image_bs) {
        error_setg(errp, "image file not found");
3729
        goto out;
3730 3731 3732 3733 3734
    }

    if (bdrv_find_base(image_bs) == image_bs) {
        error_setg(errp, "not allowing backing file change on an image "
                         "without a backing file");
3735
        goto out;
3736 3737 3738 3739 3740
    }

    /* even though we are not necessarily operating on bs, we need it to
     * determine if block ops are currently prohibited on the chain */
    if (bdrv_op_is_blocked(bs, BLOCK_OP_TYPE_CHANGE, errp)) {
3741
        goto out;
3742 3743 3744 3745 3746 3747
    }

    /* final sanity check */
    if (!bdrv_chain_contains(bs, image_bs)) {
        error_setg(errp, "'%s' and image file are not in the same chain",
                   device);
3748
        goto out;
3749 3750 3751 3752 3753 3754 3755 3756 3757 3758
    }

    /* if not r/w, reopen to make r/w */
    open_flags = image_bs->open_flags;
    ro = bdrv_is_read_only(image_bs);

    if (ro) {
        bdrv_reopen(image_bs, open_flags | BDRV_O_RDWR, &local_err);
        if (local_err) {
            error_propagate(errp, local_err);
3759
            goto out;
3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774
        }
    }

    ret = bdrv_change_backing_file(image_bs, backing_file,
                               image_bs->drv ? image_bs->drv->format_name : "");

    if (ret < 0) {
        error_setg_errno(errp, -ret, "Could not change backing file to '%s'",
                         backing_file);
        /* don't exit here, so we can try to restore open flags if
         * appropriate */
    }

    if (ro) {
        bdrv_reopen(image_bs, open_flags, &local_err);
3775
        error_propagate(errp, local_err);
3776
    }
3777 3778 3779

out:
    aio_context_release(aio_context);
3780 3781
}

3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795
void hmp_drive_add_node(Monitor *mon, const char *optstr)
{
    QemuOpts *opts;
    QDict *qdict;
    Error *local_err = NULL;

    opts = qemu_opts_parse_noisily(&qemu_drive_opts, optstr, false);
    if (!opts) {
        return;
    }

    qdict = qemu_opts_to_qdict(opts, NULL);

    if (!qdict_get_try_str(qdict, "node-name")) {
3796
        QDECREF(qdict);
3797 3798 3799 3800 3801 3802 3803 3804 3805 3806 3807 3808 3809 3810 3811 3812
        error_report("'node-name' needs to be specified");
        goto out;
    }

    BlockDriverState *bs = bds_tree_init(qdict, &local_err);
    if (!bs) {
        error_report_err(local_err);
        goto out;
    }

    QTAILQ_INSERT_TAIL(&monitor_bdrv_states, bs, monitor_list);

out:
    qemu_opts_del(opts);
}

K
Kevin Wolf 已提交
3813 3814
void qmp_blockdev_add(BlockdevOptions *options, Error **errp)
{
3815
    BlockDriverState *bs;
K
Kevin Wolf 已提交
3816
    QObject *obj;
3817
    Visitor *v = qobject_output_visitor_new(&obj);
K
Kevin Wolf 已提交
3818 3819 3820
    QDict *qdict;
    Error *local_err = NULL;

3821
    visit_type_BlockdevOptions(v, NULL, &options, &local_err);
3822
    if (local_err) {
K
Kevin Wolf 已提交
3823 3824 3825 3826
        error_propagate(errp, local_err);
        goto fail;
    }

3827
    visit_complete(v, &obj);
K
Kevin Wolf 已提交
3828 3829 3830 3831
    qdict = qobject_to_qdict(obj);

    qdict_flatten(qdict);

3832 3833 3834 3835
    if (!qdict_get_try_str(qdict, "node-name")) {
        error_setg(errp, "'node-name' must be specified for the root node");
        goto fail;
    }
3836

3837 3838 3839
    bs = bds_tree_init(qdict, errp);
    if (!bs) {
        goto fail;
K
Kevin Wolf 已提交
3840 3841
    }

3842 3843
    QTAILQ_INSERT_TAIL(&monitor_bdrv_states, bs, monitor_list);

3844
    if (bs && bdrv_key_required(bs)) {
3845 3846
        QTAILQ_REMOVE(&monitor_bdrv_states, bs, monitor_list);
        bdrv_unref(bs);
3847 3848 3849 3850
        error_setg(errp, "blockdev-add doesn't support encrypted devices");
        goto fail;
    }

K
Kevin Wolf 已提交
3851
fail:
3852
    visit_free(v);
K
Kevin Wolf 已提交
3853 3854
}

3855
void qmp_x_blockdev_del(const char *node_name, Error **errp)
3856 3857 3858 3859
{
    AioContext *aio_context;
    BlockDriverState *bs;

3860 3861 3862
    bs = bdrv_find_node(node_name);
    if (!bs) {
        error_setg(errp, "Cannot find node %s", node_name);
3863 3864
        return;
    }
3865 3866 3867
    if (bdrv_has_blk(bs)) {
        error_setg(errp, "Node %s is in use", node_name);
        return;
3868
    }
3869
    aio_context = bdrv_get_aio_context(bs);
3870 3871
    aio_context_acquire(aio_context);

3872 3873 3874
    if (bdrv_op_is_blocked(bs, BLOCK_OP_TYPE_DRIVE_DEL, errp)) {
        goto out;
    }
3875

3876 3877 3878 3879
    if (!bs->monitor_list.tqe_prev) {
        error_setg(errp, "Node %s is not owned by the monitor",
                   bs->node_name);
        goto out;
3880 3881
    }

3882 3883 3884 3885
    if (bs->refcnt > 1) {
        error_setg(errp, "Block device %s is in use",
                   bdrv_get_device_or_node_name(bs));
        goto out;
3886 3887
    }

3888 3889 3890
    QTAILQ_REMOVE(&monitor_bdrv_states, bs, monitor_list);
    bdrv_unref(bs);

3891 3892 3893 3894
out:
    aio_context_release(aio_context);
}

3895 3896 3897 3898 3899 3900 3901 3902 3903 3904 3905 3906 3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917 3918 3919 3920 3921 3922 3923 3924 3925 3926 3927 3928 3929 3930 3931 3932 3933 3934 3935 3936 3937 3938 3939 3940 3941 3942 3943 3944 3945 3946 3947 3948 3949
static BdrvChild *bdrv_find_child(BlockDriverState *parent_bs,
                                  const char *child_name)
{
    BdrvChild *child;

    QLIST_FOREACH(child, &parent_bs->children, next) {
        if (strcmp(child->name, child_name) == 0) {
            return child;
        }
    }

    return NULL;
}

void qmp_x_blockdev_change(const char *parent, bool has_child,
                           const char *child, bool has_node,
                           const char *node, Error **errp)
{
    BlockDriverState *parent_bs, *new_bs = NULL;
    BdrvChild *p_child;

    parent_bs = bdrv_lookup_bs(parent, parent, errp);
    if (!parent_bs) {
        return;
    }

    if (has_child == has_node) {
        if (has_child) {
            error_setg(errp, "The parameters child and node are in conflict");
        } else {
            error_setg(errp, "Either child or node must be specified");
        }
        return;
    }

    if (has_child) {
        p_child = bdrv_find_child(parent_bs, child);
        if (!p_child) {
            error_setg(errp, "Node '%s' does not have child '%s'",
                       parent, child);
            return;
        }
        bdrv_del_child(parent_bs, p_child, errp);
    }

    if (has_node) {
        new_bs = bdrv_find_node(node);
        if (!new_bs) {
            error_setg(errp, "Node '%s' not found", node);
            return;
        }
        bdrv_add_child(parent_bs, new_bs, errp);
    }
}

3950
BlockJobInfoList *qmp_query_block_jobs(Error **errp)
S
Stefan Hajnoczi 已提交
3951
{
3952
    BlockJobInfoList *head = NULL, **p_next = &head;
3953
    BlockJob *job;
S
Stefan Hajnoczi 已提交
3954

3955
    for (job = block_job_next(NULL); job; job = block_job_next(job)) {
3956 3957
        BlockJobInfoList *elem;
        AioContext *aio_context;
3958

3959 3960 3961 3962 3963
        if (block_job_is_internal(job)) {
            continue;
        }
        elem = g_new0(BlockJobInfoList, 1);
        aio_context = blk_get_aio_context(job->blk);
3964
        aio_context_acquire(aio_context);
3965
        elem->value = block_job_query(job, errp);
3966
        aio_context_release(aio_context);
3967 3968 3969 3970 3971
        if (!elem->value) {
            g_free(elem);
            qapi_free_BlockJobInfoList(head);
            return NULL;
        }
3972 3973
        *p_next = elem;
        p_next = &elem->next;
S
Stefan Hajnoczi 已提交
3974 3975
    }

3976
    return head;
S
Stefan Hajnoczi 已提交
3977
}
3978

3979
QemuOptsList qemu_common_drive_opts = {
3980
    .name = "drive",
3981
    .head = QTAILQ_HEAD_INITIALIZER(qemu_common_drive_opts.head),
3982 3983 3984 3985 3986 3987 3988 3989 3990
    .desc = {
        {
            .name = "snapshot",
            .type = QEMU_OPT_BOOL,
            .help = "enable/disable snapshot mode",
        },{
            .name = "aio",
            .type = QEMU_OPT_STRING,
            .help = "host AIO implementation (threads, native)",
3991 3992 3993 3994
        },{
            .name = BDRV_OPT_CACHE_WB,
            .type = QEMU_OPT_BOOL,
            .help = "Enable writeback mode",
3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007
        },{
            .name = "format",
            .type = QEMU_OPT_STRING,
            .help = "disk format (raw, qcow2, ...)",
        },{
            .name = "rerror",
            .type = QEMU_OPT_STRING,
            .help = "read error action",
        },{
            .name = "werror",
            .type = QEMU_OPT_STRING,
            .help = "write error action",
        },{
4008
            .name = BDRV_OPT_READ_ONLY,
4009 4010
            .type = QEMU_OPT_BOOL,
            .help = "open drive file as read-only",
4011 4012 4013 4014 4015
        },

        THROTTLE_OPTS,

        {
4016 4017 4018
            .name = "throttling.group",
            .type = QEMU_OPT_STRING,
            .help = "name of the block throttling group",
4019 4020 4021 4022
        },{
            .name = "copy-on-read",
            .type = QEMU_OPT_BOOL,
            .help = "copy read data from backing file into image file",
4023 4024 4025 4026
        },{
            .name = "detect-zeroes",
            .type = QEMU_OPT_STRING,
            .help = "try to optimize zero writes (off, on, unmap)",
4027 4028 4029 4030 4031 4032 4033 4034 4035 4036
        },{
            .name = "stats-account-invalid",
            .type = QEMU_OPT_BOOL,
            .help = "whether to account for invalid I/O operations "
                    "in the statistics",
        },{
            .name = "stats-account-failed",
            .type = QEMU_OPT_BOOL,
            .help = "whether to account for failed I/O operations "
                    "in the statistics",
4037 4038 4039 4040
        },
        { /* end of list */ }
    },
};
4041 4042 4043 4044 4045

QemuOptsList qemu_drive_opts = {
    .name = "drive",
    .head = QTAILQ_HEAD_INITIALIZER(qemu_drive_opts.head),
    .desc = {
4046 4047 4048 4049
        /*
         * no elements => accept any params
         * validation will happen later
         */
4050 4051 4052
        { /* end of list */ }
    },
};