[mirror_qemu.git] / block / backup.c

/*
 * QEMU backup
 *
 * Copyright (C) 2013 Proxmox Server Solutions
 *
 * Authors:
 *  Dietmar Maurer (dietmar@proxmox.com)
 *
 * This work is licensed under the terms of the GNU GPL, version 2 or later.
 * See the COPYING file in the top-level directory.
 *
 */

#include <stdio.h>
#include <errno.h>
#include <unistd.h>

#include "trace.h"
#include "block/block.h"
#include "block/block_int.h"
#include "block/blockjob.h"
#include "qemu/ratelimit.h"

#define BACKUP_CLUSTER_BITS 16
#define BACKUP_CLUSTER_SIZE (1 << BACKUP_CLUSTER_BITS)
#define BACKUP_SECTORS_PER_CLUSTER (BACKUP_CLUSTER_SIZE / BDRV_SECTOR_SIZE)

#define SLICE_TIME 100000000ULL /* ns */

typedef struct CowRequest {
    int64_t start;
    int64_t end;
    QLIST_ENTRY(CowRequest) list;
    CoQueue wait_queue; /* coroutines blocked on this request */
} CowRequest;

typedef struct BackupBlockJob {
    BlockJob common;
    BlockDriverState *target;
    /* bitmap for sync=dirty-bitmap */
    BdrvDirtyBitmap *sync_bitmap;
    MirrorSyncMode sync_mode;
    RateLimit limit;
    BlockdevOnError on_source_error;
    BlockdevOnError on_target_error;
    CoRwlock flush_rwlock;
    uint64_t sectors_read;
    HBitmap *bitmap;
    QLIST_HEAD(, CowRequest) inflight_reqs;
} BackupBlockJob;

/* See if in-flight requests overlap and wait for them to complete */
static void coroutine_fn wait_for_overlapping_requests(BackupBlockJob *job,
                                                       int64_t start,
                                                       int64_t end)
{
    CowRequest *req;
    bool retry;

    do {
        retry = false;
        QLIST_FOREACH(req, &job->inflight_reqs, list) {
            if (end > req->start && start < req->end) {
                qemu_co_queue_wait(&req->wait_queue);
                retry = true;
                break;
            }
        }
    } while (retry);
}

/* Keep track of an in-flight request */
static void cow_request_begin(CowRequest *req, BackupBlockJob *job,
                                     int64_t start, int64_t end)
{
    req->start = start;
    req->end = end;
    qemu_co_queue_init(&req->wait_queue);
    QLIST_INSERT_HEAD(&job->inflight_reqs, req, list);
}

/* Forget about a completed request */
static void cow_request_end(CowRequest *req)
{
    QLIST_REMOVE(req, list);
    qemu_co_queue_restart_all(&req->wait_queue);
}

static int coroutine_fn backup_do_cow(BlockDriverState *bs,
                                      int64_t sector_num, int nb_sectors,
                                      bool *error_is_read)
{
    BackupBlockJob *job = (BackupBlockJob *)bs->job;
    CowRequest cow_request;
    struct iovec iov;
    QEMUIOVector bounce_qiov;
    void *bounce_buffer = NULL;
    int ret = 0;
    int64_t start, end;
    int n;

    qemu_co_rwlock_rdlock(&job->flush_rwlock);

    start = sector_num / BACKUP_SECTORS_PER_CLUSTER;
    end = DIV_ROUND_UP(sector_num + nb_sectors, BACKUP_SECTORS_PER_CLUSTER);

    trace_backup_do_cow_enter(job, start, sector_num, nb_sectors);

    wait_for_overlapping_requests(job, start, end);
    cow_request_begin(&cow_request, job, start, end);

    for (; start < end; start++) {
        if (hbitmap_get(job->bitmap, start)) {
            trace_backup_do_cow_skip(job, start);
            continue; /* already copied */
        }

        trace_backup_do_cow_process(job, start);

        n = MIN(BACKUP_SECTORS_PER_CLUSTER,
                job->common.len / BDRV_SECTOR_SIZE -
                start * BACKUP_SECTORS_PER_CLUSTER);

        if (!bounce_buffer) {
            bounce_buffer = qemu_blockalign(bs, BACKUP_CLUSTER_SIZE);
        }
        iov.iov_base = bounce_buffer;
        iov.iov_len = n * BDRV_SECTOR_SIZE;
        qemu_iovec_init_external(&bounce_qiov, &iov, 1);

        ret = bdrv_co_readv(bs, start * BACKUP_SECTORS_PER_CLUSTER, n,
                            &bounce_qiov);
        if (ret < 0) {
            trace_backup_do_cow_read_fail(job, start, ret);
            if (error_is_read) {
                *error_is_read = true;
            }
            goto out;
        }

        if (buffer_is_zero(iov.iov_base, iov.iov_len)) {
            ret = bdrv_co_write_zeroes(job->target,
                                       start * BACKUP_SECTORS_PER_CLUSTER,
                                       n, BDRV_REQ_MAY_UNMAP);
        } else {
            ret = bdrv_co_writev(job->target,
                                 start * BACKUP_SECTORS_PER_CLUSTER, n,
                                 &bounce_qiov);
        }
        if (ret < 0) {
            trace_backup_do_cow_write_fail(job, start, ret);
            if (error_is_read) {
                *error_is_read = false;
            }
            goto out;
        }

        hbitmap_set(job->bitmap, start, 1);

        /* Publish progress, guest I/O counts as progress too.  Note that the
         * offset field is an opaque progress value, it is not a disk offset.
         */
        job->sectors_read += n;
        job->common.offset += n * BDRV_SECTOR_SIZE;
    }

out:
    if (bounce_buffer) {
        qemu_vfree(bounce_buffer);
    }

    cow_request_end(&cow_request);

    trace_backup_do_cow_return(job, sector_num, nb_sectors, ret);

    qemu_co_rwlock_unlock(&job->flush_rwlock);

    return ret;
}

static int coroutine_fn backup_before_write_notify(
        NotifierWithReturn *notifier,
        void *opaque)
{
    BdrvTrackedRequest *req = opaque;
    int64_t sector_num = req->offset >> BDRV_SECTOR_BITS;
    int nb_sectors = req->bytes >> BDRV_SECTOR_BITS;

    assert((req->offset & (BDRV_SECTOR_SIZE - 1)) == 0);
    assert((req->bytes & (BDRV_SECTOR_SIZE - 1)) == 0);

    return backup_do_cow(req->bs, sector_num, nb_sectors, NULL);
}

static void backup_set_speed(BlockJob *job, int64_t speed, Error **errp)
{
    BackupBlockJob *s = container_of(job, BackupBlockJob, common);

    if (speed < 0) {
        error_set(errp, QERR_INVALID_PARAMETER, "speed");
        return;
    }
    ratelimit_set_speed(&s->limit, speed / BDRV_SECTOR_SIZE, SLICE_TIME);
}

static void backup_iostatus_reset(BlockJob *job)
{
    BackupBlockJob *s = container_of(job, BackupBlockJob, common);

    bdrv_iostatus_reset(s->target);
}

static const BlockJobDriver backup_job_driver = {
    .instance_size  = sizeof(BackupBlockJob),
    .job_type       = BLOCK_JOB_TYPE_BACKUP,
    .set_speed      = backup_set_speed,
    .iostatus_reset = backup_iostatus_reset,
};

static BlockErrorAction backup_error_action(BackupBlockJob *job,
                                            bool read, int error)
{
    if (read) {
        return block_job_error_action(&job->common, job->common.bs,
                                      job->on_source_error, true, error);
    } else {
        return block_job_error_action(&job->common, job->target,
                                      job->on_target_error, false, error);
    }
}

typedef struct {
    int ret;
} BackupCompleteData;

static void backup_complete(BlockJob *job, void *opaque)
{
    BackupBlockJob *s = container_of(job, BackupBlockJob, common);
    BackupCompleteData *data = opaque;

    bdrv_unref(s->target);

    block_job_completed(job, data->ret);
    g_free(data);
}

static bool coroutine_fn yield_and_check(BackupBlockJob *job)
{
    if (block_job_is_cancelled(&job->common)) {
        return true;
    }

    /* we need to yield so that bdrv_drain_all() returns.
     * (without, VM does not reboot)
     */
    if (job->common.speed) {
        uint64_t delay_ns = ratelimit_calculate_delay(&job->limit,
                                                      job->sectors_read);
        job->sectors_read = 0;
        block_job_sleep_ns(&job->common, QEMU_CLOCK_REALTIME, delay_ns);
    } else {
        block_job_sleep_ns(&job->common, QEMU_CLOCK_REALTIME, 0);
    }

    if (block_job_is_cancelled(&job->common)) {
        return true;
    }

    return false;
}

static int coroutine_fn backup_run_incremental(BackupBlockJob *job)
{
    bool error_is_read;
    int ret = 0;
    int clusters_per_iter;
    uint32_t granularity;
    int64_t sector;
    int64_t cluster;
    int64_t end;
    int64_t last_cluster = -1;
    BlockDriverState *bs = job->common.bs;
    HBitmapIter hbi;

    granularity = bdrv_dirty_bitmap_granularity(job->sync_bitmap);
    clusters_per_iter = MAX((granularity / BACKUP_CLUSTER_SIZE), 1);
    bdrv_dirty_iter_init(job->sync_bitmap, &hbi);

    /* Find the next dirty sector(s) */
    while ((sector = hbitmap_iter_next(&hbi)) != -1) {
        cluster = sector / BACKUP_SECTORS_PER_CLUSTER;

        /* Fake progress updates for any clusters we skipped */
        if (cluster != last_cluster + 1) {
            job->common.offset += ((cluster - last_cluster - 1) *
                                   BACKUP_CLUSTER_SIZE);
        }

        for (end = cluster + clusters_per_iter; cluster < end; cluster++) {
            do {
                if (yield_and_check(job)) {
                    return ret;
                }
                ret = backup_do_cow(bs, cluster * BACKUP_SECTORS_PER_CLUSTER,
                                    BACKUP_SECTORS_PER_CLUSTER, &error_is_read);
                if ((ret < 0) &&
                    backup_error_action(job, error_is_read, -ret) ==
                    BLOCK_ERROR_ACTION_REPORT) {
                    return ret;
                }
            } while (ret < 0);
        }

        /* If the bitmap granularity is smaller than the backup granularity,
         * we need to advance the iterator pointer to the next cluster. */
        if (granularity < BACKUP_CLUSTER_SIZE) {
            bdrv_set_dirty_iter(&hbi, cluster * BACKUP_SECTORS_PER_CLUSTER);
        }

        last_cluster = cluster - 1;
    }

    /* Play some final catchup with the progress meter */
    end = DIV_ROUND_UP(job->common.len, BACKUP_CLUSTER_SIZE);
    if (last_cluster + 1 < end) {
        job->common.offset += ((end - last_cluster - 1) * BACKUP_CLUSTER_SIZE);
    }

    return ret;
}

static void coroutine_fn backup_run(void *opaque)
{
    BackupBlockJob *job = opaque;
    BackupCompleteData *data;
    BlockDriverState *bs = job->common.bs;
    BlockDriverState *target = job->target;
    BlockdevOnError on_target_error = job->on_target_error;
    NotifierWithReturn before_write = {
        .notify = backup_before_write_notify,
    };
    int64_t start, end;
    int ret = 0;

    QLIST_INIT(&job->inflight_reqs);
    qemu_co_rwlock_init(&job->flush_rwlock);

    start = 0;
    end = DIV_ROUND_UP(job->common.len, BACKUP_CLUSTER_SIZE);

    job->bitmap = hbitmap_alloc(end, 0);

    bdrv_set_enable_write_cache(target, true);
    bdrv_set_on_error(target, on_target_error, on_target_error);
    bdrv_iostatus_enable(target);

    bdrv_add_before_write_notifier(bs, &before_write);

    if (job->sync_mode == MIRROR_SYNC_MODE_NONE) {
        while (!block_job_is_cancelled(&job->common)) {
            /* Yield until the job is cancelled.  We just let our before_write
             * notify callback service CoW requests. */
            job->common.busy = false;
            qemu_coroutine_yield();
            job->common.busy = true;
        }
    } else if (job->sync_mode == MIRROR_SYNC_MODE_DIRTY_BITMAP) {
        ret = backup_run_incremental(job);
    } else {
        /* Both FULL and TOP SYNC_MODE's require copying.. */
        for (; start < end; start++) {
            bool error_is_read;
            if (yield_and_check(job)) {
                break;
            }

            if (job->sync_mode == MIRROR_SYNC_MODE_TOP) {
                int i, n;
                int alloced = 0;

                /* Check to see if these blocks are already in the
                 * backing file. */

                for (i = 0; i < BACKUP_SECTORS_PER_CLUSTER;) {
                    /* bdrv_is_allocated() only returns true/false based
                     * on the first set of sectors it comes across that
                     * are are all in the same state.
                     * For that reason we must verify each sector in the
                     * backup cluster length.  We end up copying more than
                     * needed but at some point that is always the case. */
                    alloced =
                        bdrv_is_allocated(bs,
                                start * BACKUP_SECTORS_PER_CLUSTER + i,
                                BACKUP_SECTORS_PER_CLUSTER - i, &n);
                    i += n;

                    if (alloced == 1 || n == 0) {
                        break;
                    }
                }

                /* If the above loop never found any sectors that are in
                 * the topmost image, skip this backup. */
                if (alloced == 0) {
                    continue;
                }
            }
            /* FULL sync mode we copy the whole drive. */
            ret = backup_do_cow(bs, start * BACKUP_SECTORS_PER_CLUSTER,
                    BACKUP_SECTORS_PER_CLUSTER, &error_is_read);
            if (ret < 0) {
                /* Depending on error action, fail now or retry cluster */
                BlockErrorAction action =
                    backup_error_action(job, error_is_read, -ret);
                if (action == BLOCK_ERROR_ACTION_REPORT) {
                    break;
                } else {
                    start--;
                    continue;
                }
            }
        }
    }

    notifier_with_return_remove(&before_write);

    /* wait until pending backup_do_cow() calls have completed */
    qemu_co_rwlock_wrlock(&job->flush_rwlock);
    qemu_co_rwlock_unlock(&job->flush_rwlock);

    if (job->sync_bitmap) {
        BdrvDirtyBitmap *bm;
        if (ret < 0) {
            /* Merge the successor back into the parent, delete nothing. */
            bm = bdrv_reclaim_dirty_bitmap(bs, job->sync_bitmap, NULL);
            assert(bm);
        } else {
            /* Everything is fine, delete this bitmap and install the backup. */
            bm = bdrv_dirty_bitmap_abdicate(bs, job->sync_bitmap, NULL);
            assert(bm);
        }
    }
    hbitmap_free(job->bitmap);

    bdrv_iostatus_disable(target);
    bdrv_op_unblock_all(target, job->common.blocker);

    data = g_malloc(sizeof(*data));
    data->ret = ret;
    block_job_defer_to_main_loop(&job->common, backup_complete, data);
}

void backup_start(BlockDriverState *bs, BlockDriverState *target,
                  int64_t speed, MirrorSyncMode sync_mode,
                  BdrvDirtyBitmap *sync_bitmap,
                  BlockdevOnError on_source_error,
                  BlockdevOnError on_target_error,
                  BlockCompletionFunc *cb, void *opaque,
                  Error **errp)
{
    int64_t len;

    assert(bs);
    assert(target);
    assert(cb);

    if (bs == target) {
        error_setg(errp, "Source and target cannot be the same");
        return;
    }

    if ((on_source_error == BLOCKDEV_ON_ERROR_STOP ||
         on_source_error == BLOCKDEV_ON_ERROR_ENOSPC) &&
        !bdrv_iostatus_is_enabled(bs)) {
        error_set(errp, QERR_INVALID_PARAMETER, "on-source-error");
        return;
    }

    if (!bdrv_is_inserted(bs)) {
        error_setg(errp, "Device is not inserted: %s",
                   bdrv_get_device_name(bs));
        return;
    }

    if (!bdrv_is_inserted(target)) {
        error_setg(errp, "Device is not inserted: %s",
                   bdrv_get_device_name(target));
        return;
    }

    if (bdrv_op_is_blocked(bs, BLOCK_OP_TYPE_BACKUP_SOURCE, errp)) {
        return;
    }

    if (bdrv_op_is_blocked(target, BLOCK_OP_TYPE_BACKUP_TARGET, errp)) {
        return;
    }

    if (sync_mode == MIRROR_SYNC_MODE_DIRTY_BITMAP) {
        if (!sync_bitmap) {
            error_setg(errp, "must provide a valid bitmap name for "
                             "\"dirty-bitmap\" sync mode");
            return;
        }

        /* Create a new bitmap, and freeze/disable this one. */
        if (bdrv_dirty_bitmap_create_successor(bs, sync_bitmap, errp) < 0) {
            return;
        }
    } else if (sync_bitmap) {
        error_setg(errp,
                   "a sync_bitmap was provided to backup_run, "
                   "but received an incompatible sync_mode (%s)",
                   MirrorSyncMode_lookup[sync_mode]);
        return;
    }

    len = bdrv_getlength(bs);
    if (len < 0) {
        error_setg_errno(errp, -len, "unable to get length for '%s'",
                         bdrv_get_device_name(bs));
        goto error;
    }

    BackupBlockJob *job = block_job_create(&backup_job_driver, bs, speed,
                                           cb, opaque, errp);
    if (!job) {
        goto error;
    }

    bdrv_op_block_all(target, job->common.blocker);

    job->on_source_error = on_source_error;
    job->on_target_error = on_target_error;
    job->target = target;
    job->sync_mode = sync_mode;
    job->sync_bitmap = sync_mode == MIRROR_SYNC_MODE_DIRTY_BITMAP ?
                       sync_bitmap : NULL;
    job->common.len = len;
    job->common.co = qemu_coroutine_create(backup_run);
    qemu_coroutine_enter(job->common.co, job);
    return;

 error:
    if (sync_bitmap) {
        bdrv_reclaim_dirty_bitmap(bs, sync_bitmap, NULL);
    }
}
Commit	Line	Data
98d2c6f2 DM	1	/*
	2	* QEMU backup
	3	*
	4	* Copyright (C) 2013 Proxmox Server Solutions
	5	*
	6	* Authors:
	7	* Dietmar Maurer (dietmar@proxmox.com)
	8	*
	9	* This work is licensed under the terms of the GNU GPL, version 2 or later.
	10	* See the COPYING file in the top-level directory.
	11	*
	12	*/
	13
	14	#include <stdio.h>
	15	#include <errno.h>
	16	#include <unistd.h>
	17
	18	#include "trace.h"
	19	#include "block/block.h"
	20	#include "block/block_int.h"
	21	#include "block/blockjob.h"
	22	#include "qemu/ratelimit.h"
	23
	24	#define BACKUP_CLUSTER_BITS 16
	25	#define BACKUP_CLUSTER_SIZE (1 << BACKUP_CLUSTER_BITS)
	26	#define BACKUP_SECTORS_PER_CLUSTER (BACKUP_CLUSTER_SIZE / BDRV_SECTOR_SIZE)
	27
	28	#define SLICE_TIME 100000000ULL /* ns */
	29
	30	typedef struct CowRequest {
	31	int64_t start;
	32	int64_t end;
	33	QLIST_ENTRY(CowRequest) list;
	34	CoQueue wait_queue; /* coroutines blocked on this request */
	35	} CowRequest;
	36
	37	typedef struct BackupBlockJob {
	38	BlockJob common;
	39	BlockDriverState *target;
d58d8453 JS	40	/* bitmap for sync=dirty-bitmap */
d58d8453 JS	41	BdrvDirtyBitmap *sync_bitmap;
fc5d3f84	42	MirrorSyncMode sync_mode;
98d2c6f2 DM	43	RateLimit limit;
	44	BlockdevOnError on_source_error;
	45	BlockdevOnError on_target_error;
	46	CoRwlock flush_rwlock;
	47	uint64_t sectors_read;
	48	HBitmap *bitmap;
	49	QLIST_HEAD(, CowRequest) inflight_reqs;
	50	} BackupBlockJob;
	51
	52	/* See if in-flight requests overlap and wait for them to complete */
	53	static void coroutine_fn wait_for_overlapping_requests(BackupBlockJob *job,
	54	int64_t start,
	55	int64_t end)
	56	{
	57	CowRequest *req;
	58	bool retry;
	59
	60	do {
	61	retry = false;
	62	QLIST_FOREACH(req, &job->inflight_reqs, list) {
	63	if (end > req->start && start < req->end) {
	64	qemu_co_queue_wait(&req->wait_queue);
	65	retry = true;
	66	break;
	67	}
	68	}
	69	} while (retry);
	70	}
	71
	72	/* Keep track of an in-flight request */
	73	static void cow_request_begin(CowRequest req, BackupBlockJob job,
	74	int64_t start, int64_t end)
	75	{
	76	req->start = start;
	77	req->end = end;
	78	qemu_co_queue_init(&req->wait_queue);
	79	QLIST_INSERT_HEAD(&job->inflight_reqs, req, list);
	80	}
	81
	82	/* Forget about a completed request */
	83	static void cow_request_end(CowRequest *req)
	84	{
	85	QLIST_REMOVE(req, list);
	86	qemu_co_queue_restart_all(&req->wait_queue);
	87	}
	88
	89	static int coroutine_fn backup_do_cow(BlockDriverState *bs,
	90	int64_t sector_num, int nb_sectors,
	91	bool *error_is_read)
	92	{
	93	BackupBlockJob job = (BackupBlockJob )bs->job;
	94	CowRequest cow_request;
	95	struct iovec iov;
	96	QEMUIOVector bounce_qiov;
	97	void *bounce_buffer = NULL;
	98	int ret = 0;
	99	int64_t start, end;
	100	int n;
	101
	102	qemu_co_rwlock_rdlock(&job->flush_rwlock);
	103
	104	start = sector_num / BACKUP_SECTORS_PER_CLUSTER;
	105	end = DIV_ROUND_UP(sector_num + nb_sectors, BACKUP_SECTORS_PER_CLUSTER);
	106
107	trace_backup_do_cow_enter(job, start, sector_num, nb_sectors);
108
109	wait_for_overlapping_requests(job, start, end);
110	cow_request_begin(&cow_request, job, start, end);
111
112	for (; start < end; start++) {
113	if (hbitmap_get(job->bitmap, start)) {
114	trace_backup_do_cow_skip(job, start);
115	continue; /* already copied */
116	}
117
118	trace_backup_do_cow_process(job, start);
119
120	n = MIN(BACKUP_SECTORS_PER_CLUSTER,
121	job->common.len / BDRV_SECTOR_SIZE -
122	start * BACKUP_SECTORS_PER_CLUSTER);
123
124	if (!bounce_buffer) {
125	bounce_buffer = qemu_blockalign(bs, BACKUP_CLUSTER_SIZE);
126	}
127	iov.iov_base = bounce_buffer;
128	iov.iov_len = n * BDRV_SECTOR_SIZE;
129	qemu_iovec_init_external(&bounce_qiov, &iov, 1);
130
131	ret = bdrv_co_readv(bs, start * BACKUP_SECTORS_PER_CLUSTER, n,
132	&bounce_qiov);
133	if (ret < 0) {
134	trace_backup_do_cow_read_fail(job, start, ret);
135	if (error_is_read) {
136	*error_is_read = true;
137	}
138	goto out;
139	}
140
141	if (buffer_is_zero(iov.iov_base, iov.iov_len)) {
142	ret = bdrv_co_write_zeroes(job->target,
aa7bfbff	143	start * BACKUP_SECTORS_PER_CLUSTER,
d32f35cb	144	n, BDRV_REQ_MAY_UNMAP);
98d2c6f2 DM	145	} else {
	146	ret = bdrv_co_writev(job->target,
	147	start * BACKUP_SECTORS_PER_CLUSTER, n,
	148	&bounce_qiov);
	149	}
	150	if (ret < 0) {
	151	trace_backup_do_cow_write_fail(job, start, ret);
	152	if (error_is_read) {
	153	*error_is_read = false;
	154	}
	155	goto out;
	156	}
	157
	158	hbitmap_set(job->bitmap, start, 1);
	159
	160	/* Publish progress, guest I/O counts as progress too. Note that the
	161	* offset field is an opaque progress value, it is not a disk offset.
	162	*/
	163	job->sectors_read += n;
	164	job->common.offset += n * BDRV_SECTOR_SIZE;
	165	}
	166
	167	out:
	168	if (bounce_buffer) {
	169	qemu_vfree(bounce_buffer);
	170	}
	171
	172	cow_request_end(&cow_request);
	173
	174	trace_backup_do_cow_return(job, sector_num, nb_sectors, ret);
	175
	176	qemu_co_rwlock_unlock(&job->flush_rwlock);
	177
	178	return ret;
	179	}
	180
	181	static int coroutine_fn backup_before_write_notify(
	182	NotifierWithReturn *notifier,
	183	void *opaque)
	184	{
	185	BdrvTrackedRequest *req = opaque;
793ed47a KW	186	int64_t sector_num = req->offset >> BDRV_SECTOR_BITS;
793ed47a KW	187	int nb_sectors = req->bytes >> BDRV_SECTOR_BITS;
98d2c6f2	188
793ed47a KW	189	assert((req->offset & (BDRV_SECTOR_SIZE - 1)) == 0);
	190	assert((req->bytes & (BDRV_SECTOR_SIZE - 1)) == 0);
	191
	192	return backup_do_cow(req->bs, sector_num, nb_sectors, NULL);
98d2c6f2 DM	193	}
	194
	195	static void backup_set_speed(BlockJob job, int64_t speed, Error *errp)
	196	{
	197	BackupBlockJob *s = container_of(job, BackupBlockJob, common);
	198
	199	if (speed < 0) {
	200	error_set(errp, QERR_INVALID_PARAMETER, "speed");
	201	return;
	202	}
	203	ratelimit_set_speed(&s->limit, speed / BDRV_SECTOR_SIZE, SLICE_TIME);
	204	}
	205
	206	static void backup_iostatus_reset(BlockJob *job)
	207	{
	208	BackupBlockJob *s = container_of(job, BackupBlockJob, common);
	209
	210	bdrv_iostatus_reset(s->target);
	211	}
	212
3fc4b10a	213	static const BlockJobDriver backup_job_driver = {
98d2c6f2	214	.instance_size = sizeof(BackupBlockJob),
79e14bf7	215	.job_type = BLOCK_JOB_TYPE_BACKUP,
98d2c6f2 DM	216	.set_speed = backup_set_speed,
	217	.iostatus_reset = backup_iostatus_reset,
	218	};
	219
	220	static BlockErrorAction backup_error_action(BackupBlockJob *job,
	221	bool read, int error)
	222	{
	223	if (read) {
	224	return block_job_error_action(&job->common, job->common.bs,
	225	job->on_source_error, true, error);
	226	} else {
	227	return block_job_error_action(&job->common, job->target,
	228	job->on_target_error, false, error);
	229	}
	230	}
	231
761731b1 SH	232	typedef struct {
	233	int ret;
	234	} BackupCompleteData;
	235
	236	static void backup_complete(BlockJob job, void opaque)
	237	{
	238	BackupBlockJob *s = container_of(job, BackupBlockJob, common);
	239	BackupCompleteData *data = opaque;
	240
	241	bdrv_unref(s->target);
	242
	243	block_job_completed(job, data->ret);
	244	g_free(data);
	245	}
	246
d58d8453 JS	247	static bool coroutine_fn yield_and_check(BackupBlockJob *job)
	248	{
	249	if (block_job_is_cancelled(&job->common)) {
	250	return true;
	251	}
	252
	253	/* we need to yield so that bdrv_drain_all() returns.
	254	* (without, VM does not reboot)
	255	*/
	256	if (job->common.speed) {
	257	uint64_t delay_ns = ratelimit_calculate_delay(&job->limit,
	258	job->sectors_read);
	259	job->sectors_read = 0;
	260	block_job_sleep_ns(&job->common, QEMU_CLOCK_REALTIME, delay_ns);
	261	} else {
	262	block_job_sleep_ns(&job->common, QEMU_CLOCK_REALTIME, 0);
	263	}
	264
	265	if (block_job_is_cancelled(&job->common)) {
	266	return true;
	267	}
	268
	269	return false;
	270	}
	271
	272	static int coroutine_fn backup_run_incremental(BackupBlockJob *job)
	273	{
	274	bool error_is_read;
	275	int ret = 0;
	276	int clusters_per_iter;
	277	uint32_t granularity;
	278	int64_t sector;
	279	int64_t cluster;
	280	int64_t end;
	281	int64_t last_cluster = -1;
	282	BlockDriverState *bs = job->common.bs;
	283	HBitmapIter hbi;
	284
	285	granularity = bdrv_dirty_bitmap_granularity(job->sync_bitmap);
	286	clusters_per_iter = MAX((granularity / BACKUP_CLUSTER_SIZE), 1);
20dca810	287	bdrv_dirty_iter_init(job->sync_bitmap, &hbi);
d58d8453 JS	288
	289	/* Find the next dirty sector(s) */
	290	while ((sector = hbitmap_iter_next(&hbi)) != -1) {
	291	cluster = sector / BACKUP_SECTORS_PER_CLUSTER;
	292
	293	/* Fake progress updates for any clusters we skipped */
	294	if (cluster != last_cluster + 1) {
	295	job->common.offset += ((cluster - last_cluster - 1) *
	296	BACKUP_CLUSTER_SIZE);
	297	}
	298
	299	for (end = cluster + clusters_per_iter; cluster < end; cluster++) {
	300	do {
	301	if (yield_and_check(job)) {
	302	return ret;
	303	}
	304	ret = backup_do_cow(bs, cluster * BACKUP_SECTORS_PER_CLUSTER,
	305	BACKUP_SECTORS_PER_CLUSTER, &error_is_read);
	306	if ((ret < 0) &&
	307	backup_error_action(job, error_is_read, -ret) ==
	308	BLOCK_ERROR_ACTION_REPORT) {
	309	return ret;
	310	}
	311	} while (ret < 0);
	312	}
	313
	314	/* If the bitmap granularity is smaller than the backup granularity,
	315	* we need to advance the iterator pointer to the next cluster. */
	316	if (granularity < BACKUP_CLUSTER_SIZE) {
	317	bdrv_set_dirty_iter(&hbi, cluster * BACKUP_SECTORS_PER_CLUSTER);
	318	}
	319
	320	last_cluster = cluster - 1;
	321	}
	322
	323	/* Play some final catchup with the progress meter */
	324	end = DIV_ROUND_UP(job->common.len, BACKUP_CLUSTER_SIZE);
	325	if (last_cluster + 1 < end) {
	326	job->common.offset += ((end - last_cluster - 1) * BACKUP_CLUSTER_SIZE);
	327	}
	328
	329	return ret;
	330	}
	331
98d2c6f2 DM	332	static void coroutine_fn backup_run(void *opaque)
	333	{
	334	BackupBlockJob *job = opaque;
761731b1	335	BackupCompleteData *data;
98d2c6f2 DM	336	BlockDriverState *bs = job->common.bs;
	337	BlockDriverState *target = job->target;
	338	BlockdevOnError on_target_error = job->on_target_error;
	339	NotifierWithReturn before_write = {
	340	.notify = backup_before_write_notify,
	341	};
	342	int64_t start, end;
	343	int ret = 0;
	344
	345	QLIST_INIT(&job->inflight_reqs);
	346	qemu_co_rwlock_init(&job->flush_rwlock);
	347
	348	start = 0;
d58d8453	349	end = DIV_ROUND_UP(job->common.len, BACKUP_CLUSTER_SIZE);
98d2c6f2 DM	350
	351	job->bitmap = hbitmap_alloc(end, 0);
	352
	353	bdrv_set_enable_write_cache(target, true);
	354	bdrv_set_on_error(target, on_target_error, on_target_error);
	355	bdrv_iostatus_enable(target);
	356
	357	bdrv_add_before_write_notifier(bs, &before_write);
	358
fc5d3f84 IM	359	if (job->sync_mode == MIRROR_SYNC_MODE_NONE) {
	360	while (!block_job_is_cancelled(&job->common)) {
	361	/* Yield until the job is cancelled. We just let our before_write
	362	* notify callback service CoW requests. */
	363	job->common.busy = false;
	364	qemu_coroutine_yield();
	365	job->common.busy = true;
98d2c6f2	366	}
d58d8453 JS	367	} else if (job->sync_mode == MIRROR_SYNC_MODE_DIRTY_BITMAP) {
d58d8453 JS	368	ret = backup_run_incremental(job);
fc5d3f84 IM	369	} else {
	370	/* Both FULL and TOP SYNC_MODE's require copying.. */
	371	for (; start < end; start++) {
	372	bool error_is_read;
d58d8453	373	if (yield_and_check(job)) {
98d2c6f2	374	break;
fc5d3f84 IM	375	}
	376
	377	if (job->sync_mode == MIRROR_SYNC_MODE_TOP) {
	378	int i, n;
	379	int alloced = 0;
	380
	381	/* Check to see if these blocks are already in the
	382	* backing file. */
	383
	384	for (i = 0; i < BACKUP_SECTORS_PER_CLUSTER;) {
bdad13b9	385	/* bdrv_is_allocated() only returns true/false based
4c293dc6	386	* on the first set of sectors it comes across that
fc5d3f84 IM	387	* are are all in the same state.
	388	* For that reason we must verify each sector in the
	389	* backup cluster length. We end up copying more than
	390	* needed but at some point that is always the case. */
	391	alloced =
bdad13b9	392	bdrv_is_allocated(bs,
fc5d3f84 IM	393	start * BACKUP_SECTORS_PER_CLUSTER + i,
	394	BACKUP_SECTORS_PER_CLUSTER - i, &n);
	395	i += n;
	396
d40593dd	397	if (alloced == 1 \|\| n == 0) {
fc5d3f84 IM	398	break;
	399	}
	400	}
	401
	402	/* If the above loop never found any sectors that are in
	403	* the topmost image, skip this backup. */
	404	if (alloced == 0) {
	405	continue;
	406	}
	407	}
	408	/* FULL sync mode we copy the whole drive. */
	409	ret = backup_do_cow(bs, start * BACKUP_SECTORS_PER_CLUSTER,
	410	BACKUP_SECTORS_PER_CLUSTER, &error_is_read);
	411	if (ret < 0) {
	412	/* Depending on error action, fail now or retry cluster */
	413	BlockErrorAction action =
	414	backup_error_action(job, error_is_read, -ret);
a589569f	415	if (action == BLOCK_ERROR_ACTION_REPORT) {
fc5d3f84 IM	416	break;
	417	} else {
	418	start--;
	419	continue;
	420	}
98d2c6f2 DM	421	}
	422	}
	423	}
	424
	425	notifier_with_return_remove(&before_write);
	426
	427	/* wait until pending backup_do_cow() calls have completed */
	428	qemu_co_rwlock_wrlock(&job->flush_rwlock);
	429	qemu_co_rwlock_unlock(&job->flush_rwlock);
	430
d58d8453 JS	431	if (job->sync_bitmap) {
	432	BdrvDirtyBitmap *bm;
	433	if (ret < 0) {
	434	/* Merge the successor back into the parent, delete nothing. */
	435	bm = bdrv_reclaim_dirty_bitmap(bs, job->sync_bitmap, NULL);
	436	assert(bm);
	437	} else {
	438	/* Everything is fine, delete this bitmap and install the backup. */
	439	bm = bdrv_dirty_bitmap_abdicate(bs, job->sync_bitmap, NULL);
	440	assert(bm);
	441	}
	442	}
98d2c6f2 DM	443	hbitmap_free(job->bitmap);
	444
	445	bdrv_iostatus_disable(target);
c29c1dd3	446	bdrv_op_unblock_all(target, job->common.blocker);
98d2c6f2	447
761731b1 SH	448	data = g_malloc(sizeof(*data));
	449	data->ret = ret;
	450	block_job_defer_to_main_loop(&job->common, backup_complete, data);
98d2c6f2 DM	451	}
	452
	453	void backup_start(BlockDriverState bs, BlockDriverState target,
fc5d3f84	454	int64_t speed, MirrorSyncMode sync_mode,
d58d8453	455	BdrvDirtyBitmap *sync_bitmap,
98d2c6f2 DM	456	BlockdevOnError on_source_error,
98d2c6f2 DM	457	BlockdevOnError on_target_error,
097310b5	458	BlockCompletionFunc cb, void opaque,
98d2c6f2 DM	459	Error **errp)
	460	{
	461	int64_t len;
	462
	463	assert(bs);
	464	assert(target);
	465	assert(cb);
	466
c29c1dd3 FZ	467	if (bs == target) {
	468	error_setg(errp, "Source and target cannot be the same");
	469	return;
	470	}
	471
98d2c6f2 DM	472	if ((on_source_error == BLOCKDEV_ON_ERROR_STOP \|\|
	473	on_source_error == BLOCKDEV_ON_ERROR_ENOSPC) &&
	474	!bdrv_iostatus_is_enabled(bs)) {
	475	error_set(errp, QERR_INVALID_PARAMETER, "on-source-error");
	476	return;
	477	}
	478
c29c1dd3 FZ	479	if (!bdrv_is_inserted(bs)) {
	480	error_setg(errp, "Device is not inserted: %s",
	481	bdrv_get_device_name(bs));
	482	return;
	483	}
	484
	485	if (!bdrv_is_inserted(target)) {
	486	error_setg(errp, "Device is not inserted: %s",
	487	bdrv_get_device_name(target));
	488	return;
	489	}
	490
	491	if (bdrv_op_is_blocked(bs, BLOCK_OP_TYPE_BACKUP_SOURCE, errp)) {
	492	return;
	493	}
	494
	495	if (bdrv_op_is_blocked(target, BLOCK_OP_TYPE_BACKUP_TARGET, errp)) {
	496	return;
	497	}
	498
d58d8453 JS	499	if (sync_mode == MIRROR_SYNC_MODE_DIRTY_BITMAP) {
	500	if (!sync_bitmap) {
	501	error_setg(errp, "must provide a valid bitmap name for "
	502	"\"dirty-bitmap\" sync mode");
	503	return;
	504	}
	505
	506	/* Create a new bitmap, and freeze/disable this one. */
	507	if (bdrv_dirty_bitmap_create_successor(bs, sync_bitmap, errp) < 0) {
	508	return;
	509	}
	510	} else if (sync_bitmap) {
	511	error_setg(errp,
	512	"a sync_bitmap was provided to backup_run, "
	513	"but received an incompatible sync_mode (%s)",
	514	MirrorSyncMode_lookup[sync_mode]);
	515	return;
	516	}
	517
98d2c6f2 DM	518	len = bdrv_getlength(bs);
	519	if (len < 0) {
	520	error_setg_errno(errp, -len, "unable to get length for '%s'",
	521	bdrv_get_device_name(bs));
d58d8453	522	goto error;
98d2c6f2 DM	523	}
98d2c6f2 DM	524
3fc4b10a	525	BackupBlockJob *job = block_job_create(&backup_job_driver, bs, speed,
98d2c6f2 DM	526	cb, opaque, errp);
98d2c6f2 DM	527	if (!job) {
d58d8453	528	goto error;
98d2c6f2 DM	529	}
98d2c6f2 DM	530
c29c1dd3 FZ	531	bdrv_op_block_all(target, job->common.blocker);
c29c1dd3 FZ	532
98d2c6f2 DM	533	job->on_source_error = on_source_error;
	534	job->on_target_error = on_target_error;
	535	job->target = target;
fc5d3f84	536	job->sync_mode = sync_mode;
d58d8453 JS	537	job->sync_bitmap = sync_mode == MIRROR_SYNC_MODE_DIRTY_BITMAP ?
d58d8453 JS	538	sync_bitmap : NULL;
98d2c6f2 DM	539	job->common.len = len;
	540	job->common.co = qemu_coroutine_create(backup_run);
	541	qemu_coroutine_enter(job->common.co, job);
d58d8453 JS	542	return;
	543
	544	error:
	545	if (sync_bitmap) {
	546	bdrv_reclaim_dirty_bitmap(bs, sync_bitmap, NULL);
	547	}
98d2c6f2	548	}