aboutsummaryrefslogtreecommitdiff
path: root/block/mirror.c
blob: e0b3f4180f302cfa7edb2d82f8a613d16fe150be (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
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
356
357
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
389
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
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
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
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
/*
 * Image mirroring
 *
 * Copyright Red Hat, Inc. 2012
 *
 * Authors:
 *  Paolo Bonzini  <pbonzini@redhat.com>
 *
 * This work is licensed under the terms of the GNU LGPL, version 2 or later.
 * See the COPYING.LIB file in the top-level directory.
 *
 */

#include "qemu/osdep.h"
#include "trace.h"
#include "block/blockjob.h"
#include "block/block_int.h"
#include "sysemu/block-backend.h"
#include "qapi/error.h"
#include "qapi/qmp/qerror.h"
#include "qemu/ratelimit.h"
#include "qemu/bitmap.h"

#define SLICE_TIME    100000000ULL /* ns */
#define MAX_IN_FLIGHT 16
#define MAX_IO_SECTORS ((1 << 20) >> BDRV_SECTOR_BITS) /* 1 Mb */
#define DEFAULT_MIRROR_BUF_SIZE \
    (MAX_IN_FLIGHT * MAX_IO_SECTORS * BDRV_SECTOR_SIZE)

/* The mirroring buffer is a list of granularity-sized chunks.
 * Free chunks are organized in a list.
 */
typedef struct MirrorBuffer {
    QSIMPLEQ_ENTRY(MirrorBuffer) next;
} MirrorBuffer;

typedef struct MirrorBlockJob {
    BlockJob common;
    RateLimit limit;
    BlockBackend *target;
    BlockDriverState *base;
    /* The name of the graph node to replace */
    char *replaces;
    /* The BDS to replace */
    BlockDriverState *to_replace;
    /* Used to block operations on the drive-mirror-replace target */
    Error *replace_blocker;
    bool is_none_mode;
    BlockMirrorBackingMode backing_mode;
    BlockdevOnError on_source_error, on_target_error;
    bool synced;
    bool should_complete;
    int64_t granularity;
    size_t buf_size;
    int64_t bdev_length;
    unsigned long *cow_bitmap;
    BdrvDirtyBitmap *dirty_bitmap;
    HBitmapIter hbi;
    uint8_t *buf;
    QSIMPLEQ_HEAD(, MirrorBuffer) buf_free;
    int buf_free_count;

    uint64_t last_pause_ns;
    unsigned long *in_flight_bitmap;
    int in_flight;
    int64_t sectors_in_flight;
    int ret;
    bool unmap;
    bool waiting_for_io;
    int target_cluster_sectors;
    int max_iov;
} MirrorBlockJob;

typedef struct MirrorOp {
    MirrorBlockJob *s;
    QEMUIOVector qiov;
    int64_t sector_num;
    int nb_sectors;
} MirrorOp;

static BlockErrorAction mirror_error_action(MirrorBlockJob *s, bool read,
                                            int error)
{
    s->synced = false;
    if (read) {
        return block_job_error_action(&s->common, s->on_source_error,
                                      true, error);
    } else {
        return block_job_error_action(&s->common, s->on_target_error,
                                      false, error);
    }
}

static void mirror_iteration_done(MirrorOp *op, int ret)
{
    MirrorBlockJob *s = op->s;
    struct iovec *iov;
    int64_t chunk_num;
    int i, nb_chunks, sectors_per_chunk;

    trace_mirror_iteration_done(s, op->sector_num, op->nb_sectors, ret);

    s->in_flight--;
    s->sectors_in_flight -= op->nb_sectors;
    iov = op->qiov.iov;
    for (i = 0; i < op->qiov.niov; i++) {
        MirrorBuffer *buf = (MirrorBuffer *) iov[i].iov_base;
        QSIMPLEQ_INSERT_TAIL(&s->buf_free, buf, next);
        s->buf_free_count++;
    }

    sectors_per_chunk = s->granularity >> BDRV_SECTOR_BITS;
    chunk_num = op->sector_num / sectors_per_chunk;
    nb_chunks = DIV_ROUND_UP(op->nb_sectors, sectors_per_chunk);
    bitmap_clear(s->in_flight_bitmap, chunk_num, nb_chunks);
    if (ret >= 0) {
        if (s->cow_bitmap) {
            bitmap_set(s->cow_bitmap, chunk_num, nb_chunks);
        }
        s->common.offset += (uint64_t)op->nb_sectors * BDRV_SECTOR_SIZE;
    }

    qemu_iovec_destroy(&op->qiov);
    g_free(op);

    if (s->waiting_for_io) {
        qemu_coroutine_enter(s->common.co);
    }
}

static void mirror_write_complete(void *opaque, int ret)
{
    MirrorOp *op = opaque;
    MirrorBlockJob *s = op->s;
    if (ret < 0) {
        BlockErrorAction action;

        bdrv_set_dirty_bitmap(s->dirty_bitmap, op->sector_num, op->nb_sectors);
        action = mirror_error_action(s, false, -ret);
        if (action == BLOCK_ERROR_ACTION_REPORT && s->ret >= 0) {
            s->ret = ret;
        }
    }
    mirror_iteration_done(op, ret);
}

static void mirror_read_complete(void *opaque, int ret)
{
    MirrorOp *op = opaque;
    MirrorBlockJob *s = op->s;
    if (ret < 0) {
        BlockErrorAction action;

        bdrv_set_dirty_bitmap(s->dirty_bitmap, op->sector_num, op->nb_sectors);
        action = mirror_error_action(s, true, -ret);
        if (action == BLOCK_ERROR_ACTION_REPORT && s->ret >= 0) {
            s->ret = ret;
        }

        mirror_iteration_done(op, ret);
        return;
    }
    blk_aio_pwritev(s->target, op->sector_num * BDRV_SECTOR_SIZE, &op->qiov,
                    0, mirror_write_complete, op);
}

static inline void mirror_clip_sectors(MirrorBlockJob *s,
                                       int64_t sector_num,
                                       int *nb_sectors)
{
    *nb_sectors = MIN(*nb_sectors,
                      s->bdev_length / BDRV_SECTOR_SIZE - sector_num);
}

/* Round sector_num and/or nb_sectors to target cluster if COW is needed, and
 * return the offset of the adjusted tail sector against original. */
static int mirror_cow_align(MirrorBlockJob *s,
                            int64_t *sector_num,
                            int *nb_sectors)
{
    bool need_cow;
    int ret = 0;
    int chunk_sectors = s->granularity >> BDRV_SECTOR_BITS;
    int64_t align_sector_num = *sector_num;
    int align_nb_sectors = *nb_sectors;
    int max_sectors = chunk_sectors * s->max_iov;

    need_cow = !test_bit(*sector_num / chunk_sectors, s->cow_bitmap);
    need_cow |= !test_bit((*sector_num + *nb_sectors - 1) / chunk_sectors,
                          s->cow_bitmap);
    if (need_cow) {
        bdrv_round_sectors_to_clusters(blk_bs(s->target), *sector_num,
                                       *nb_sectors, &align_sector_num,
                                       &align_nb_sectors);
    }

    if (align_nb_sectors > max_sectors) {
        align_nb_sectors = max_sectors;
        if (need_cow) {
            align_nb_sectors = QEMU_ALIGN_DOWN(align_nb_sectors,
                                               s->target_cluster_sectors);
        }
    }
    /* Clipping may result in align_nb_sectors unaligned to chunk boundary, but
     * that doesn't matter because it's already the end of source image. */
    mirror_clip_sectors(s, align_sector_num, &align_nb_sectors);

    ret = align_sector_num + align_nb_sectors - (*sector_num + *nb_sectors);
    *sector_num = align_sector_num;
    *nb_sectors = align_nb_sectors;
    assert(ret >= 0);
    return ret;
}

static inline void mirror_wait_for_io(MirrorBlockJob *s)
{
    assert(!s->waiting_for_io);
    s->waiting_for_io = true;
    qemu_coroutine_yield();
    s->waiting_for_io = false;
}

/* Submit async read while handling COW.
 * Returns: The number of sectors copied after and including sector_num,
 *          excluding any sectors copied prior to sector_num due to alignment.
 *          This will be nb_sectors if no alignment is necessary, or
 *          (new_end - sector_num) if tail is rounded up or down due to
 *          alignment or buffer limit.
 */
static int mirror_do_read(MirrorBlockJob *s, int64_t sector_num,
                          int nb_sectors)
{
    BlockBackend *source = s->common.blk;
    int sectors_per_chunk, nb_chunks;
    int ret;
    MirrorOp *op;
    int max_sectors;

    sectors_per_chunk = s->granularity >> BDRV_SECTOR_BITS;
    max_sectors = sectors_per_chunk * s->max_iov;

    /* We can only handle as much as buf_size at a time. */
    nb_sectors = MIN(s->buf_size >> BDRV_SECTOR_BITS, nb_sectors);
    nb_sectors = MIN(max_sectors, nb_sectors);
    assert(nb_sectors);
    ret = nb_sectors;

    if (s->cow_bitmap) {
        ret += mirror_cow_align(s, &sector_num, &nb_sectors);
    }
    assert(nb_sectors << BDRV_SECTOR_BITS <= s->buf_size);
    /* The sector range must meet granularity because:
     * 1) Caller passes in aligned values;
     * 2) mirror_cow_align is used only when target cluster is larger. */
    assert(!(sector_num % sectors_per_chunk));
    nb_chunks = DIV_ROUND_UP(nb_sectors, sectors_per_chunk);

    while (s->buf_free_count < nb_chunks) {
        trace_mirror_yield_in_flight(s, sector_num, s->in_flight);
        mirror_wait_for_io(s);
    }

    /* Allocate a MirrorOp that is used as an AIO callback.  */
    op = g_new(MirrorOp, 1);
    op->s = s;
    op->sector_num = sector_num;
    op->nb_sectors = nb_sectors;

    /* Now make a QEMUIOVector taking enough granularity-sized chunks
     * from s->buf_free.
     */
    qemu_iovec_init(&op->qiov, nb_chunks);
    while (nb_chunks-- > 0) {
        MirrorBuffer *buf = QSIMPLEQ_FIRST(&s->buf_free);
        size_t remaining = nb_sectors * BDRV_SECTOR_SIZE - op->qiov.size;

        QSIMPLEQ_REMOVE_HEAD(&s->buf_free, next);
        s->buf_free_count--;
        qemu_iovec_add(&op->qiov, buf, MIN(s->granularity, remaining));
    }

    /* Copy the dirty cluster.  */
    s->in_flight++;
    s->sectors_in_flight += nb_sectors;
    trace_mirror_one_iteration(s, sector_num, nb_sectors);

    blk_aio_preadv(source, sector_num * BDRV_SECTOR_SIZE, &op->qiov, 0,
                   mirror_read_complete, op);
    return ret;
}

static void mirror_do_zero_or_discard(MirrorBlockJob *s,
                                      int64_t sector_num,
                                      int nb_sectors,
                                      bool is_discard)
{
    MirrorOp *op;

    /* Allocate a MirrorOp that is used as an AIO callback. The qiov is zeroed
     * so the freeing in mirror_iteration_done is nop. */
    op = g_new0(MirrorOp, 1);
    op->s = s;
    op->sector_num = sector_num;
    op->nb_sectors = nb_sectors;

    s->in_flight++;
    s->sectors_in_flight += nb_sectors;
    if (is_discard) {
        blk_aio_pdiscard(s->target, sector_num << BDRV_SECTOR_BITS,
                         op->nb_sectors << BDRV_SECTOR_BITS,
                         mirror_write_complete, op);
    } else {
        blk_aio_pwrite_zeroes(s->target, sector_num * BDRV_SECTOR_SIZE,
                              op->nb_sectors * BDRV_SECTOR_SIZE,
                              s->unmap ? BDRV_REQ_MAY_UNMAP : 0,
                              mirror_write_complete, op);
    }
}

static uint64_t coroutine_fn mirror_iteration(MirrorBlockJob *s)
{
    BlockDriverState *source = blk_bs(s->common.blk);
    int64_t sector_num, first_chunk;
    uint64_t delay_ns = 0;
    /* At least the first dirty chunk is mirrored in one iteration. */
    int nb_chunks = 1;
    int64_t end = s->bdev_length / BDRV_SECTOR_SIZE;
    int sectors_per_chunk = s->granularity >> BDRV_SECTOR_BITS;
    bool write_zeroes_ok = bdrv_can_write_zeroes_with_unmap(blk_bs(s->target));
    int max_io_sectors = MAX((s->buf_size >> BDRV_SECTOR_BITS) / MAX_IN_FLIGHT,
                             MAX_IO_SECTORS);

    sector_num = hbitmap_iter_next(&s->hbi);
    if (sector_num < 0) {
        bdrv_dirty_iter_init(s->dirty_bitmap, &s->hbi);
        sector_num = hbitmap_iter_next(&s->hbi);
        trace_mirror_restart_iter(s, bdrv_get_dirty_count(s->dirty_bitmap));
        assert(sector_num >= 0);
    }

    first_chunk = sector_num / sectors_per_chunk;
    while (test_bit(first_chunk, s->in_flight_bitmap)) {
        trace_mirror_yield_in_flight(s, sector_num, s->in_flight);
        mirror_wait_for_io(s);
    }

    block_job_pause_point(&s->common);

    /* Find the number of consective dirty chunks following the first dirty
     * one, and wait for in flight requests in them. */
    while (nb_chunks * sectors_per_chunk < (s->buf_size >> BDRV_SECTOR_BITS)) {
        int64_t hbitmap_next;
        int64_t next_sector = sector_num + nb_chunks * sectors_per_chunk;
        int64_t next_chunk = next_sector / sectors_per_chunk;
        if (next_sector >= end ||
            !bdrv_get_dirty(source, s->dirty_bitmap, next_sector)) {
            break;
        }
        if (test_bit(next_chunk, s->in_flight_bitmap)) {
            break;
        }

        hbitmap_next = hbitmap_iter_next(&s->hbi);
        if (hbitmap_next > next_sector || hbitmap_next < 0) {
            /* The bitmap iterator's cache is stale, refresh it */
            bdrv_set_dirty_iter(&s->hbi, next_sector);
            hbitmap_next = hbitmap_iter_next(&s->hbi);
        }
        assert(hbitmap_next == next_sector);
        nb_chunks++;
    }

    /* Clear dirty bits before querying the block status, because
     * calling bdrv_get_block_status_above could yield - if some blocks are
     * marked dirty in this window, we need to know.
     */
    bdrv_reset_dirty_bitmap(s->dirty_bitmap, sector_num,
                            nb_chunks * sectors_per_chunk);
    bitmap_set(s->in_flight_bitmap, sector_num / sectors_per_chunk, nb_chunks);
    while (nb_chunks > 0 && sector_num < end) {
        int ret;
        int io_sectors, io_sectors_acct;
        BlockDriverState *file;
        enum MirrorMethod {
            MIRROR_METHOD_COPY,
            MIRROR_METHOD_ZERO,
            MIRROR_METHOD_DISCARD
        } mirror_method = MIRROR_METHOD_COPY;

        assert(!(sector_num % sectors_per_chunk));
        ret = bdrv_get_block_status_above(source, NULL, sector_num,
                                          nb_chunks * sectors_per_chunk,
                                          &io_sectors, &file);
        if (ret < 0) {
            io_sectors = MIN(nb_chunks * sectors_per_chunk, max_io_sectors);
        } else if (ret & BDRV_BLOCK_DATA) {
            io_sectors = MIN(io_sectors, max_io_sectors);
        }

        io_sectors -= io_sectors % sectors_per_chunk;
        if (io_sectors < sectors_per_chunk) {
            io_sectors = sectors_per_chunk;
        } else if (ret >= 0 && !(ret & BDRV_BLOCK_DATA)) {
            int64_t target_sector_num;
            int target_nb_sectors;
            bdrv_round_sectors_to_clusters(blk_bs(s->target), sector_num,
                                           io_sectors,  &target_sector_num,
                                           &target_nb_sectors);
            if (target_sector_num == sector_num &&
                target_nb_sectors == io_sectors) {
                mirror_method = ret & BDRV_BLOCK_ZERO ?
                                    MIRROR_METHOD_ZERO :
                                    MIRROR_METHOD_DISCARD;
            }
        }

        while (s->in_flight >= MAX_IN_FLIGHT) {
            trace_mirror_yield_in_flight(s, sector_num, s->in_flight);
            mirror_wait_for_io(s);
        }

        if (s->ret < 0) {
            return 0;
        }

        mirror_clip_sectors(s, sector_num, &io_sectors);
        switch (mirror_method) {
        case MIRROR_METHOD_COPY:
            io_sectors = mirror_do_read(s, sector_num, io_sectors);
            io_sectors_acct = io_sectors;
            break;
        case MIRROR_METHOD_ZERO:
        case MIRROR_METHOD_DISCARD:
            mirror_do_zero_or_discard(s, sector_num, io_sectors,
                                      mirror_method == MIRROR_METHOD_DISCARD);
            if (write_zeroes_ok) {
                io_sectors_acct = 0;
            } else {
                io_sectors_acct = io_sectors;
            }
            break;
        default:
            abort();
        }
        assert(io_sectors);
        sector_num += io_sectors;
        nb_chunks -= DIV_ROUND_UP(io_sectors, sectors_per_chunk);
        if (s->common.speed) {
            delay_ns = ratelimit_calculate_delay(&s->limit, io_sectors_acct);
        }
    }
    return delay_ns;
}

static void mirror_free_init(MirrorBlockJob *s)
{
    int granularity = s->granularity;
    size_t buf_size = s->buf_size;
    uint8_t *buf = s->buf;

    assert(s->buf_free_count == 0);
    QSIMPLEQ_INIT(&s->buf_free);
    while (buf_size != 0) {
        MirrorBuffer *cur = (MirrorBuffer *)buf;
        QSIMPLEQ_INSERT_TAIL(&s->buf_free, cur, next);
        s->buf_free_count++;
        buf_size -= granularity;
        buf += granularity;
    }
}

static void mirror_drain(MirrorBlockJob *s)
{
    while (s->in_flight > 0) {
        mirror_wait_for_io(s);
    }
}

typedef struct {
    int ret;
} MirrorExitData;

static void mirror_exit(BlockJob *job, void *opaque)
{
    MirrorBlockJob *s = container_of(job, MirrorBlockJob, common);
    MirrorExitData *data = opaque;
    AioContext *replace_aio_context = NULL;
    BlockDriverState *src = blk_bs(s->common.blk);
    BlockDriverState *target_bs = blk_bs(s->target);

    /* Make sure that the source BDS doesn't go away before we called
     * block_job_completed(). */
    bdrv_ref(src);

    if (s->to_replace) {
        replace_aio_context = bdrv_get_aio_context(s->to_replace);
        aio_context_acquire(replace_aio_context);
    }

    if (s->should_complete && data->ret == 0) {
        BlockDriverState *to_replace = src;
        if (s->to_replace) {
            to_replace = s->to_replace;
        }

        if (bdrv_get_flags(target_bs) != bdrv_get_flags(to_replace)) {
            bdrv_reopen(target_bs, bdrv_get_flags(to_replace), NULL);
        }

        /* The mirror job has no requests in flight any more, but we need to
         * drain potential other users of the BDS before changing the graph. */
        bdrv_drained_begin(target_bs);
        bdrv_replace_in_backing_chain(to_replace, target_bs);
        bdrv_drained_end(target_bs);

        /* We just changed the BDS the job BB refers to */
        blk_remove_bs(job->blk);
        blk_insert_bs(job->blk, src);
    }
    if (s->to_replace) {
        bdrv_op_unblock_all(s->to_replace, s->replace_blocker);
        error_free(s->replace_blocker);
        bdrv_unref(s->to_replace);
    }
    if (replace_aio_context) {
        aio_context_release(replace_aio_context);
    }
    g_free(s->replaces);
    bdrv_op_unblock_all(target_bs, s->common.blocker);
    blk_unref(s->target);
    block_job_completed(&s->common, data->ret);
    g_free(data);
    bdrv_drained_end(src);
    bdrv_unref(src);
}

static void mirror_throttle(MirrorBlockJob *s)
{
    int64_t now = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);

    if (now - s->last_pause_ns > SLICE_TIME) {
        s->last_pause_ns = now;
        block_job_sleep_ns(&s->common, QEMU_CLOCK_REALTIME, 0);
    } else {
        block_job_pause_point(&s->common);
    }
}

static int coroutine_fn mirror_dirty_init(MirrorBlockJob *s)
{
    int64_t sector_num, end;
    BlockDriverState *base = s->base;
    BlockDriverState *bs = blk_bs(s->common.blk);
    BlockDriverState *target_bs = blk_bs(s->target);
    int ret, n;

    end = s->bdev_length / BDRV_SECTOR_SIZE;

    if (base == NULL && !bdrv_has_zero_init(target_bs)) {
        if (!bdrv_can_write_zeroes_with_unmap(target_bs)) {
            bdrv_set_dirty_bitmap(s->dirty_bitmap, 0, end);
            return 0;
        }

        for (sector_num = 0; sector_num < end; ) {
            int nb_sectors = MIN(end - sector_num,
                QEMU_ALIGN_DOWN(INT_MAX, s->granularity) >> BDRV_SECTOR_BITS);

            mirror_throttle(s);

            if (block_job_is_cancelled(&s->common)) {
                return 0;
            }

            if (s->in_flight >= MAX_IN_FLIGHT) {
                trace_mirror_yield(s, s->in_flight, s->buf_free_count, -1);
                mirror_wait_for_io(s);
                continue;
            }

            mirror_do_zero_or_discard(s, sector_num, nb_sectors, false);
            sector_num += nb_sectors;
        }

        mirror_drain(s);
    }

    /* First part, loop on the sectors and initialize the dirty bitmap.  */
    for (sector_num = 0; sector_num < end; ) {
        /* Just to make sure we are not exceeding int limit. */
        int nb_sectors = MIN(INT_MAX >> BDRV_SECTOR_BITS,
                             end - sector_num);

        mirror_throttle(s);

        if (block_job_is_cancelled(&s->common)) {
            return 0;
        }

        ret = bdrv_is_allocated_above(bs, base, sector_num, nb_sectors, &n);
        if (ret < 0) {
            return ret;
        }

        assert(n > 0);
        if (ret == 1) {
            bdrv_set_dirty_bitmap(s->dirty_bitmap, sector_num, n);
        }
        sector_num += n;
    }
    return 0;
}

static void coroutine_fn mirror_run(void *opaque)
{
    MirrorBlockJob *s = opaque;
    MirrorExitData *data;
    BlockDriverState *bs = blk_bs(s->common.blk);
    BlockDriverState *target_bs = blk_bs(s->target);
    int64_t length;
    BlockDriverInfo bdi;
    char backing_filename[2]; /* we only need 2 characters because we are only
                                 checking for a NULL string */
    int ret = 0;
    int target_cluster_size = BDRV_SECTOR_SIZE;

    if (block_job_is_cancelled(&s->common)) {
        goto immediate_exit;
    }

    s->bdev_length = bdrv_getlength(bs);
    if (s->bdev_length < 0) {
        ret = s->bdev_length;
        goto immediate_exit;
    } else if (s->bdev_length == 0) {
        /* Report BLOCK_JOB_READY and wait for complete. */
        block_job_event_ready(&s->common);
        s->synced = true;
        while (!block_job_is_cancelled(&s->common) && !s->should_complete) {
            block_job_yield(&s->common);
        }
        s->common.cancelled = false;
        goto immediate_exit;
    }

    length = DIV_ROUND_UP(s->bdev_length, s->granularity);
    s->in_flight_bitmap = bitmap_new(length);

    /* If we have no backing file yet in the destination, we cannot let
     * the destination do COW.  Instead, we copy sectors around the
     * dirty data if needed.  We need a bitmap to do that.
     */
    bdrv_get_backing_filename(target_bs, backing_filename,
                              sizeof(backing_filename));
    if (!bdrv_get_info(target_bs, &bdi) && bdi.cluster_size) {
        target_cluster_size = bdi.cluster_size;
    }
    if (backing_filename[0] && !target_bs->backing
        && s->granularity < target_cluster_size) {
        s->buf_size = MAX(s->buf_size, target_cluster_size);
        s->cow_bitmap = bitmap_new(length);
    }
    s->target_cluster_sectors = target_cluster_size >> BDRV_SECTOR_BITS;
    s->max_iov = MIN(bs->bl.max_iov, target_bs->bl.max_iov);

    s->buf = qemu_try_blockalign(bs, s->buf_size);
    if (s->buf == NULL) {
        ret = -ENOMEM;
        goto immediate_exit;
    }

    mirror_free_init(s);

    s->last_pause_ns = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
    if (!s->is_none_mode) {
        ret = mirror_dirty_init(s);
        if (ret < 0 || block_job_is_cancelled(&s->common)) {
            goto immediate_exit;
        }
    }

    bdrv_dirty_iter_init(s->dirty_bitmap, &s->hbi);
    for (;;) {
        uint64_t delay_ns = 0;
        int64_t cnt, delta;
        bool should_complete;

        if (s->ret < 0) {
            ret = s->ret;
            goto immediate_exit;
        }

        block_job_pause_point(&s->common);

        cnt = bdrv_get_dirty_count(s->dirty_bitmap);
        /* s->common.offset contains the number of bytes already processed so
         * far, cnt is the number of dirty sectors remaining and
         * s->sectors_in_flight is the number of sectors currently being
         * processed; together those are the current total operation length */
        s->common.len = s->common.offset +
                        (cnt + s->sectors_in_flight) * BDRV_SECTOR_SIZE;

        /* Note that even when no rate limit is applied we need to yield
         * periodically with no pending I/O so that bdrv_drain_all() returns.
         * We do so every SLICE_TIME nanoseconds, or when there is an error,
         * or when the source is clean, whichever comes first.
         */
        delta = qemu_clock_get_ns(QEMU_CLOCK_REALTIME) - s->last_pause_ns;
        if (delta < SLICE_TIME &&
            s->common.iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
            if (s->in_flight >= MAX_IN_FLIGHT || s->buf_free_count == 0 ||
                (cnt == 0 && s->in_flight > 0)) {
                trace_mirror_yield(s, s->in_flight, s->buf_free_count, cnt);
                mirror_wait_for_io(s);
                continue;
            } else if (cnt != 0) {
                delay_ns = mirror_iteration(s);
            }
        }

        should_complete = false;
        if (s->in_flight == 0 && cnt == 0) {
            trace_mirror_before_flush(s);
            ret = blk_flush(s->target);
            if (ret < 0) {
                if (mirror_error_action(s, false, -ret) ==
                    BLOCK_ERROR_ACTION_REPORT) {
                    goto immediate_exit;
                }
            } else {
                /* We're out of the streaming phase.  From now on, if the job
                 * is cancelled we will actually complete all pending I/O and
                 * report completion.  This way, block-job-cancel will leave
                 * the target in a consistent state.
                 */
                if (!s->synced) {
                    block_job_event_ready(&s->common);
                    s->synced = true;
                }

                should_complete = s->should_complete ||
                    block_job_is_cancelled(&s->common);
                cnt = bdrv_get_dirty_count(s->dirty_bitmap);
            }
        }

        if (cnt == 0 && should_complete) {
            /* The dirty bitmap is not updated while operations are pending.
             * If we're about to exit, wait for pending operations before
             * calling bdrv_get_dirty_count(bs), or we may exit while the
             * source has dirty data to copy!
             *
             * Note that I/O can be submitted by the guest while
             * mirror_populate runs.
             */
            trace_mirror_before_drain(s, cnt);
            bdrv_co_drain(bs);
            cnt = bdrv_get_dirty_count(s->dirty_bitmap);
        }

        ret = 0;
        trace_mirror_before_sleep(s, cnt, s->synced, delay_ns);
        if (!s->synced) {
            block_job_sleep_ns(&s->common, QEMU_CLOCK_REALTIME, delay_ns);
            if (block_job_is_cancelled(&s->common)) {
                break;
            }
        } else if (!should_complete) {
            delay_ns = (s->in_flight == 0 && cnt == 0 ? SLICE_TIME : 0);
            block_job_sleep_ns(&s->common, QEMU_CLOCK_REALTIME, delay_ns);
        } else if (cnt == 0) {
            /* The two disks are in sync.  Exit and report successful
             * completion.
             */
            assert(QLIST_EMPTY(&bs->tracked_requests));
            s->common.cancelled = false;
            break;
        }
        s->last_pause_ns = qemu_clock_get_ns(QEMU_CLOCK_REALTIME);
    }

immediate_exit:
    if (s->in_flight > 0) {
        /* We get here only if something went wrong.  Either the job failed,
         * or it was cancelled prematurely so that we do not guarantee that
         * the target is a copy of the source.
         */
        assert(ret < 0 || (!s->synced && block_job_is_cancelled(&s->common)));
        mirror_drain(s);
    }

    assert(s->in_flight == 0);
    qemu_vfree(s->buf);
    g_free(s->cow_bitmap);
    g_free(s->in_flight_bitmap);
    bdrv_release_dirty_bitmap(bs, s->dirty_bitmap);

    data = g_malloc(sizeof(*data));
    data->ret = ret;
    /* Before we switch to target in mirror_exit, make sure data doesn't
     * change. */
    bdrv_drained_begin(bs);
    block_job_defer_to_main_loop(&s->common, mirror_exit, data);
}

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

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

static void mirror_complete(BlockJob *job, Error **errp)
{
    MirrorBlockJob *s = container_of(job, MirrorBlockJob, common);
    BlockDriverState *src, *target;

    src = blk_bs(job->blk);
    target = blk_bs(s->target);

    if (!s->synced) {
        error_setg(errp, "The active block job '%s' cannot be completed",
                   job->id);
        return;
    }

    if (s->backing_mode == MIRROR_OPEN_BACKING_CHAIN) {
        int ret;

        assert(!target->backing);
        ret = bdrv_open_backing_file(target, NULL, "backing", errp);
        if (ret < 0) {
            return;
        }
    }

    /* block all operations on to_replace bs */
    if (s->replaces) {
        AioContext *replace_aio_context;

        s->to_replace = bdrv_find_node(s->replaces);
        if (!s->to_replace) {
            error_setg(errp, "Node name '%s' not found", s->replaces);
            return;
        }

        replace_aio_context = bdrv_get_aio_context(s->to_replace);
        aio_context_acquire(replace_aio_context);

        error_setg(&s->replace_blocker,
                   "block device is in use by block-job-complete");
        bdrv_op_block_all(s->to_replace, s->replace_blocker);
        bdrv_ref(s->to_replace);

        aio_context_release(replace_aio_context);
    }

    if (s->backing_mode == MIRROR_SOURCE_BACKING_CHAIN) {
        BlockDriverState *backing = s->is_none_mode ? src : s->base;
        if (backing_bs(target) != backing) {
            bdrv_set_backing_hd(target, backing);
        }
    }

    s->should_complete = true;
    block_job_enter(&s->common);
}

/* There is no matching mirror_resume() because mirror_run() will begin
 * iterating again when the job is resumed.
 */
static void coroutine_fn mirror_pause(BlockJob *job)
{
    MirrorBlockJob *s = container_of(job, MirrorBlockJob, common);

    mirror_drain(s);
}

static void mirror_attached_aio_context(BlockJob *job, AioContext *new_context)
{
    MirrorBlockJob *s = container_of(job, MirrorBlockJob, common);

    blk_set_aio_context(s->target, new_context);
}

static const BlockJobDriver mirror_job_driver = {
    .instance_size          = sizeof(MirrorBlockJob),
    .job_type               = BLOCK_JOB_TYPE_MIRROR,
    .set_speed              = mirror_set_speed,
    .complete               = mirror_complete,
    .pause                  = mirror_pause,
    .attached_aio_context   = mirror_attached_aio_context,
};

static const BlockJobDriver commit_active_job_driver = {
    .instance_size          = sizeof(MirrorBlockJob),
    .job_type               = BLOCK_JOB_TYPE_COMMIT,
    .set_speed              = mirror_set_speed,
    .complete               = mirror_complete,
    .pause                  = mirror_pause,
    .attached_aio_context   = mirror_attached_aio_context,
};

static void mirror_start_job(const char *job_id, BlockDriverState *bs,
                             BlockDriverState *target, const char *replaces,
                             int64_t speed, uint32_t granularity,
                             int64_t buf_size,
                             BlockMirrorBackingMode backing_mode,
                             BlockdevOnError on_source_error,
                             BlockdevOnError on_target_error,
                             bool unmap,
                             BlockCompletionFunc *cb,
                             void *opaque, Error **errp,
                             const BlockJobDriver *driver,
                             bool is_none_mode, BlockDriverState *base)
{
    MirrorBlockJob *s;

    if (granularity == 0) {
        granularity = bdrv_get_default_bitmap_granularity(target);
    }

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

    if (buf_size < 0) {
        error_setg(errp, "Invalid parameter 'buf-size'");
        return;
    }

    if (buf_size == 0) {
        buf_size = DEFAULT_MIRROR_BUF_SIZE;
    }

    s = block_job_create(job_id, driver, bs, speed, cb, opaque, errp);
    if (!s) {
        return;
    }

    s->target = blk_new();
    blk_insert_bs(s->target, target);

    s->replaces = g_strdup(replaces);
    s->on_source_error = on_source_error;
    s->on_target_error = on_target_error;
    s->is_none_mode = is_none_mode;
    s->backing_mode = backing_mode;
    s->base = base;
    s->granularity = granularity;
    s->buf_size = ROUND_UP(buf_size, granularity);
    s->unmap = unmap;

    s->dirty_bitmap = bdrv_create_dirty_bitmap(bs, granularity, NULL, errp);
    if (!s->dirty_bitmap) {
        g_free(s->replaces);
        blk_unref(s->target);
        block_job_unref(&s->common);
        return;
    }

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

    s->common.co = qemu_coroutine_create(mirror_run, s);
    trace_mirror_start(bs, s, s->common.co, opaque);
    qemu_coroutine_enter(s->common.co);
}

void mirror_start(const char *job_id, BlockDriverState *bs,
                  BlockDriverState *target, const char *replaces,
                  int64_t speed, uint32_t granularity, int64_t buf_size,
                  MirrorSyncMode mode, BlockMirrorBackingMode backing_mode,
                  BlockdevOnError on_source_error,
                  BlockdevOnError on_target_error,
                  bool unmap,
                  BlockCompletionFunc *cb,
                  void *opaque, Error **errp)
{
    bool is_none_mode;
    BlockDriverState *base;

    if (mode == MIRROR_SYNC_MODE_INCREMENTAL) {
        error_setg(errp, "Sync mode 'incremental' not supported");
        return;
    }
    is_none_mode = mode == MIRROR_SYNC_MODE_NONE;
    base = mode == MIRROR_SYNC_MODE_TOP ? backing_bs(bs) : NULL;
    mirror_start_job(job_id, bs, target, replaces,
                     speed, granularity, buf_size, backing_mode,
                     on_source_error, on_target_error, unmap, cb, opaque, errp,
                     &mirror_job_driver, is_none_mode, base);
}

void commit_active_start(const char *job_id, BlockDriverState *bs,
                         BlockDriverState *base, int64_t speed,
                         BlockdevOnError on_error,
                         BlockCompletionFunc *cb,
                         void *opaque, Error **errp)
{
    int64_t length, base_length;
    int orig_base_flags;
    int ret;
    Error *local_err = NULL;

    orig_base_flags = bdrv_get_flags(base);

    if (bdrv_reopen(base, bs->open_flags, errp)) {
        return;
    }

    length = bdrv_getlength(bs);
    if (length < 0) {
        error_setg_errno(errp, -length,
                         "Unable to determine length of %s", bs->filename);
        goto error_restore_flags;
    }

    base_length = bdrv_getlength(base);
    if (base_length < 0) {
        error_setg_errno(errp, -base_length,
                         "Unable to determine length of %s", base->filename);
        goto error_restore_flags;
    }

    if (length > base_length) {
        ret = bdrv_truncate(base, length);
        if (ret < 0) {
            error_setg_errno(errp, -ret,
                            "Top image %s is larger than base image %s, and "
                             "resize of base image failed",
                             bs->filename, base->filename);
            goto error_restore_flags;
        }
    }

    mirror_start_job(job_id, bs, base, NULL, speed, 0, 0,
                     MIRROR_LEAVE_BACKING_CHAIN,
                     on_error, on_error, false, cb, opaque, &local_err,
                     &commit_active_job_driver, false, base);
    if (local_err) {
        error_propagate(errp, local_err);
        goto error_restore_flags;
    }

    return;

error_restore_flags:
    /* ignore error and errp for bdrv_reopen, because we want to propagate
     * the original error */
    bdrv_reopen(base, orig_base_flags, NULL);
    return;
}