xref: /qemu/block/replication.c (revision 75ac231c)
1 /*
2  * Replication Block filter
3  *
4  * Copyright (c) 2016 HUAWEI TECHNOLOGIES CO., LTD.
5  * Copyright (c) 2016 Intel Corporation
6  * Copyright (c) 2016 FUJITSU LIMITED
7  *
8  * Author:
9  *   Wen Congyang <wency@cn.fujitsu.com>
10  *
11  * This work is licensed under the terms of the GNU GPL, version 2 or later.
12  * See the COPYING file in the top-level directory.
13  */
14 
15 #include "qemu/osdep.h"
16 #include "qemu/module.h"
17 #include "qemu/option.h"
18 #include "block/nbd.h"
19 #include "block/blockjob.h"
20 #include "block/block_int.h"
21 #include "block/block_backup.h"
22 #include "sysemu/block-backend.h"
23 #include "qapi/error.h"
24 #include "qapi/qmp/qdict.h"
25 #include "block/replication.h"
26 
27 typedef enum {
28     BLOCK_REPLICATION_NONE,             /* block replication is not started */
29     BLOCK_REPLICATION_RUNNING,          /* block replication is running */
30     BLOCK_REPLICATION_FAILOVER,         /* failover is running in background */
31     BLOCK_REPLICATION_FAILOVER_FAILED,  /* failover failed */
32     BLOCK_REPLICATION_DONE,             /* block replication is done */
33 } ReplicationStage;
34 
35 typedef struct BDRVReplicationState {
36     ReplicationMode mode;
37     ReplicationStage stage;
38     BlockJob *commit_job;
39     BdrvChild *hidden_disk;
40     BdrvChild *secondary_disk;
41     BlockJob *backup_job;
42     char *top_id;
43     ReplicationState *rs;
44     Error *blocker;
45     bool orig_hidden_read_only;
46     bool orig_secondary_read_only;
47     int error;
48 } BDRVReplicationState;
49 
50 static void replication_start(ReplicationState *rs, ReplicationMode mode,
51                               Error **errp);
52 static void replication_do_checkpoint(ReplicationState *rs, Error **errp);
53 static void replication_get_error(ReplicationState *rs, Error **errp);
54 static void replication_stop(ReplicationState *rs, bool failover,
55                              Error **errp);
56 
57 #define REPLICATION_MODE        "mode"
58 #define REPLICATION_TOP_ID      "top-id"
59 static QemuOptsList replication_runtime_opts = {
60     .name = "replication",
61     .head = QTAILQ_HEAD_INITIALIZER(replication_runtime_opts.head),
62     .desc = {
63         {
64             .name = REPLICATION_MODE,
65             .type = QEMU_OPT_STRING,
66         },
67         {
68             .name = REPLICATION_TOP_ID,
69             .type = QEMU_OPT_STRING,
70         },
71         { /* end of list */ }
72     },
73 };
74 
75 static ReplicationOps replication_ops = {
76     .start = replication_start,
77     .checkpoint = replication_do_checkpoint,
78     .get_error = replication_get_error,
79     .stop = replication_stop,
80 };
81 
82 static int replication_open(BlockDriverState *bs, QDict *options,
83                             int flags, Error **errp)
84 {
85     int ret;
86     BDRVReplicationState *s = bs->opaque;
87     QemuOpts *opts = NULL;
88     const char *mode;
89     const char *top_id;
90 
91     ret = bdrv_open_file_child(NULL, options, "file", bs, errp);
92     if (ret < 0) {
93         return ret;
94     }
95 
96     ret = -EINVAL;
97     opts = qemu_opts_create(&replication_runtime_opts, NULL, 0, &error_abort);
98     if (!qemu_opts_absorb_qdict(opts, options, errp)) {
99         goto fail;
100     }
101 
102     mode = qemu_opt_get(opts, REPLICATION_MODE);
103     if (!mode) {
104         error_setg(errp, "Missing the option mode");
105         goto fail;
106     }
107 
108     if (!strcmp(mode, "primary")) {
109         s->mode = REPLICATION_MODE_PRIMARY;
110         top_id = qemu_opt_get(opts, REPLICATION_TOP_ID);
111         if (top_id) {
112             error_setg(errp,
113                        "The primary side does not support option top-id");
114             goto fail;
115         }
116     } else if (!strcmp(mode, "secondary")) {
117         s->mode = REPLICATION_MODE_SECONDARY;
118         top_id = qemu_opt_get(opts, REPLICATION_TOP_ID);
119         s->top_id = g_strdup(top_id);
120         if (!s->top_id) {
121             error_setg(errp, "Missing the option top-id");
122             goto fail;
123         }
124     } else {
125         error_setg(errp,
126                    "The option mode's value should be primary or secondary");
127         goto fail;
128     }
129 
130     s->rs = replication_new(bs, &replication_ops);
131 
132     ret = 0;
133 
134 fail:
135     qemu_opts_del(opts);
136     return ret;
137 }
138 
139 static void replication_close(BlockDriverState *bs)
140 {
141     BDRVReplicationState *s = bs->opaque;
142     Job *commit_job;
143     GLOBAL_STATE_CODE();
144 
145     if (s->stage == BLOCK_REPLICATION_RUNNING) {
146         replication_stop(s->rs, false, NULL);
147     }
148     if (s->stage == BLOCK_REPLICATION_FAILOVER) {
149         commit_job = &s->commit_job->job;
150         assert(commit_job->aio_context == qemu_get_current_aio_context());
151         job_cancel_sync(commit_job, false);
152     }
153 
154     if (s->mode == REPLICATION_MODE_SECONDARY) {
155         g_free(s->top_id);
156     }
157 
158     replication_remove(s->rs);
159 }
160 
161 static void replication_child_perm(BlockDriverState *bs, BdrvChild *c,
162                                    BdrvChildRole role,
163                                    BlockReopenQueue *reopen_queue,
164                                    uint64_t perm, uint64_t shared,
165                                    uint64_t *nperm, uint64_t *nshared)
166 {
167     if (role & BDRV_CHILD_PRIMARY) {
168         *nperm = BLK_PERM_CONSISTENT_READ;
169     } else {
170         *nperm = 0;
171     }
172 
173     if ((bs->open_flags & (BDRV_O_INACTIVE | BDRV_O_RDWR)) == BDRV_O_RDWR) {
174         *nperm |= BLK_PERM_WRITE;
175     }
176     *nshared = BLK_PERM_CONSISTENT_READ
177                | BLK_PERM_WRITE
178                | BLK_PERM_WRITE_UNCHANGED;
179     return;
180 }
181 
182 static int64_t replication_getlength(BlockDriverState *bs)
183 {
184     return bdrv_getlength(bs->file->bs);
185 }
186 
187 static int replication_get_io_status(BDRVReplicationState *s)
188 {
189     switch (s->stage) {
190     case BLOCK_REPLICATION_NONE:
191         return -EIO;
192     case BLOCK_REPLICATION_RUNNING:
193         return 0;
194     case BLOCK_REPLICATION_FAILOVER:
195         return s->mode == REPLICATION_MODE_PRIMARY ? -EIO : 0;
196     case BLOCK_REPLICATION_FAILOVER_FAILED:
197         return s->mode == REPLICATION_MODE_PRIMARY ? -EIO : 1;
198     case BLOCK_REPLICATION_DONE:
199         /*
200          * active commit job completes, and active disk and secondary_disk
201          * is swapped, so we can operate bs->file directly
202          */
203         return s->mode == REPLICATION_MODE_PRIMARY ? -EIO : 0;
204     default:
205         abort();
206     }
207 }
208 
209 static int replication_return_value(BDRVReplicationState *s, int ret)
210 {
211     if (s->mode == REPLICATION_MODE_SECONDARY) {
212         return ret;
213     }
214 
215     if (ret < 0) {
216         s->error = ret;
217         ret = 0;
218     }
219 
220     return ret;
221 }
222 
223 static coroutine_fn int replication_co_readv(BlockDriverState *bs,
224                                              int64_t sector_num,
225                                              int remaining_sectors,
226                                              QEMUIOVector *qiov)
227 {
228     BDRVReplicationState *s = bs->opaque;
229     int ret;
230 
231     if (s->mode == REPLICATION_MODE_PRIMARY) {
232         /* We only use it to forward primary write requests */
233         return -EIO;
234     }
235 
236     ret = replication_get_io_status(s);
237     if (ret < 0) {
238         return ret;
239     }
240 
241     ret = bdrv_co_preadv(bs->file, sector_num * BDRV_SECTOR_SIZE,
242                          remaining_sectors * BDRV_SECTOR_SIZE, qiov, 0);
243 
244     return replication_return_value(s, ret);
245 }
246 
247 static coroutine_fn int replication_co_writev(BlockDriverState *bs,
248                                               int64_t sector_num,
249                                               int remaining_sectors,
250                                               QEMUIOVector *qiov,
251                                               int flags)
252 {
253     BDRVReplicationState *s = bs->opaque;
254     QEMUIOVector hd_qiov;
255     uint64_t bytes_done = 0;
256     BdrvChild *top = bs->file;
257     BdrvChild *base = s->secondary_disk;
258     BdrvChild *target;
259     int ret;
260     int64_t n;
261 
262     ret = replication_get_io_status(s);
263     if (ret < 0) {
264         goto out;
265     }
266 
267     if (ret == 0) {
268         ret = bdrv_co_pwritev(top, sector_num * BDRV_SECTOR_SIZE,
269                               remaining_sectors * BDRV_SECTOR_SIZE, qiov, 0);
270         return replication_return_value(s, ret);
271     }
272 
273     /*
274      * Failover failed, only write to active disk if the sectors
275      * have already been allocated in active disk/hidden disk.
276      */
277     qemu_iovec_init(&hd_qiov, qiov->niov);
278     while (remaining_sectors > 0) {
279         int64_t count;
280 
281         ret = bdrv_is_allocated_above(top->bs, base->bs, false,
282                                       sector_num * BDRV_SECTOR_SIZE,
283                                       remaining_sectors * BDRV_SECTOR_SIZE,
284                                       &count);
285         if (ret < 0) {
286             goto out1;
287         }
288 
289         assert(QEMU_IS_ALIGNED(count, BDRV_SECTOR_SIZE));
290         n = count >> BDRV_SECTOR_BITS;
291         qemu_iovec_reset(&hd_qiov);
292         qemu_iovec_concat(&hd_qiov, qiov, bytes_done, count);
293 
294         target = ret ? top : base;
295         ret = bdrv_co_pwritev(target, sector_num * BDRV_SECTOR_SIZE,
296                               n * BDRV_SECTOR_SIZE, &hd_qiov, 0);
297         if (ret < 0) {
298             goto out1;
299         }
300 
301         remaining_sectors -= n;
302         sector_num += n;
303         bytes_done += count;
304     }
305 
306 out1:
307     qemu_iovec_destroy(&hd_qiov);
308 out:
309     return ret;
310 }
311 
312 static void secondary_do_checkpoint(BlockDriverState *bs, Error **errp)
313 {
314     BDRVReplicationState *s = bs->opaque;
315     BdrvChild *active_disk = bs->file;
316     Error *local_err = NULL;
317     int ret;
318 
319     if (!s->backup_job) {
320         error_setg(errp, "Backup job was cancelled unexpectedly");
321         return;
322     }
323 
324     backup_do_checkpoint(s->backup_job, &local_err);
325     if (local_err) {
326         error_propagate(errp, local_err);
327         return;
328     }
329 
330     if (!active_disk->bs->drv) {
331         error_setg(errp, "Active disk %s is ejected",
332                    active_disk->bs->node_name);
333         return;
334     }
335 
336     ret = bdrv_make_empty(active_disk, errp);
337     if (ret < 0) {
338         return;
339     }
340 
341     if (!s->hidden_disk->bs->drv) {
342         error_setg(errp, "Hidden disk %s is ejected",
343                    s->hidden_disk->bs->node_name);
344         return;
345     }
346 
347     ret = bdrv_make_empty(s->hidden_disk, errp);
348     if (ret < 0) {
349         return;
350     }
351 }
352 
353 /* This function is supposed to be called twice:
354  * first with writable = true, then with writable = false.
355  * The first call puts s->hidden_disk and s->secondary_disk in
356  * r/w mode, and the second puts them back in their original state.
357  */
358 static void reopen_backing_file(BlockDriverState *bs, bool writable,
359                                 Error **errp)
360 {
361     BDRVReplicationState *s = bs->opaque;
362     BdrvChild *hidden_disk, *secondary_disk;
363     BlockReopenQueue *reopen_queue = NULL;
364 
365     /*
366      * s->hidden_disk and s->secondary_disk may not be set yet, as they will
367      * only be set after the children are writable.
368      */
369     hidden_disk = bs->file->bs->backing;
370     secondary_disk = hidden_disk->bs->backing;
371 
372     if (writable) {
373         s->orig_hidden_read_only = bdrv_is_read_only(hidden_disk->bs);
374         s->orig_secondary_read_only = bdrv_is_read_only(secondary_disk->bs);
375     }
376 
377     bdrv_subtree_drained_begin(hidden_disk->bs);
378     bdrv_subtree_drained_begin(secondary_disk->bs);
379 
380     if (s->orig_hidden_read_only) {
381         QDict *opts = qdict_new();
382         qdict_put_bool(opts, BDRV_OPT_READ_ONLY, !writable);
383         reopen_queue = bdrv_reopen_queue(reopen_queue, hidden_disk->bs,
384                                          opts, true);
385     }
386 
387     if (s->orig_secondary_read_only) {
388         QDict *opts = qdict_new();
389         qdict_put_bool(opts, BDRV_OPT_READ_ONLY, !writable);
390         reopen_queue = bdrv_reopen_queue(reopen_queue, secondary_disk->bs,
391                                          opts, true);
392     }
393 
394     if (reopen_queue) {
395         AioContext *ctx = bdrv_get_aio_context(bs);
396         if (ctx != qemu_get_aio_context()) {
397             aio_context_release(ctx);
398         }
399         bdrv_reopen_multiple(reopen_queue, errp);
400         if (ctx != qemu_get_aio_context()) {
401             aio_context_acquire(ctx);
402         }
403     }
404 
405     bdrv_subtree_drained_end(hidden_disk->bs);
406     bdrv_subtree_drained_end(secondary_disk->bs);
407 }
408 
409 static void backup_job_cleanup(BlockDriverState *bs)
410 {
411     BDRVReplicationState *s = bs->opaque;
412     BlockDriverState *top_bs;
413 
414     s->backup_job = NULL;
415 
416     top_bs = bdrv_lookup_bs(s->top_id, s->top_id, NULL);
417     if (!top_bs) {
418         return;
419     }
420     bdrv_op_unblock_all(top_bs, s->blocker);
421     error_free(s->blocker);
422     reopen_backing_file(bs, false, NULL);
423 }
424 
425 static void backup_job_completed(void *opaque, int ret)
426 {
427     BlockDriverState *bs = opaque;
428     BDRVReplicationState *s = bs->opaque;
429 
430     if (s->stage != BLOCK_REPLICATION_FAILOVER) {
431         /* The backup job is cancelled unexpectedly */
432         s->error = -EIO;
433     }
434 
435     backup_job_cleanup(bs);
436 }
437 
438 static bool check_top_bs(BlockDriverState *top_bs, BlockDriverState *bs)
439 {
440     BdrvChild *child;
441 
442     /* The bs itself is the top_bs */
443     if (top_bs == bs) {
444         return true;
445     }
446 
447     /* Iterate over top_bs's children */
448     QLIST_FOREACH(child, &top_bs->children, next) {
449         if (child->bs == bs || check_top_bs(child->bs, bs)) {
450             return true;
451         }
452     }
453 
454     return false;
455 }
456 
457 static void replication_start(ReplicationState *rs, ReplicationMode mode,
458                               Error **errp)
459 {
460     BlockDriverState *bs = rs->opaque;
461     BDRVReplicationState *s;
462     BlockDriverState *top_bs;
463     BdrvChild *active_disk, *hidden_disk, *secondary_disk;
464     int64_t active_length, hidden_length, disk_length;
465     AioContext *aio_context;
466     Error *local_err = NULL;
467     BackupPerf perf = { .use_copy_range = true, .max_workers = 1 };
468 
469     aio_context = bdrv_get_aio_context(bs);
470     aio_context_acquire(aio_context);
471     s = bs->opaque;
472 
473     if (s->stage == BLOCK_REPLICATION_DONE ||
474         s->stage == BLOCK_REPLICATION_FAILOVER) {
475         /*
476          * This case happens when a secondary is promoted to primary.
477          * Ignore the request because the secondary side of replication
478          * doesn't have to do anything anymore.
479          */
480         aio_context_release(aio_context);
481         return;
482     }
483 
484     if (s->stage != BLOCK_REPLICATION_NONE) {
485         error_setg(errp, "Block replication is running or done");
486         aio_context_release(aio_context);
487         return;
488     }
489 
490     if (s->mode != mode) {
491         error_setg(errp, "The parameter mode's value is invalid, needs %d,"
492                    " but got %d", s->mode, mode);
493         aio_context_release(aio_context);
494         return;
495     }
496 
497     switch (s->mode) {
498     case REPLICATION_MODE_PRIMARY:
499         break;
500     case REPLICATION_MODE_SECONDARY:
501         active_disk = bs->file;
502         if (!active_disk || !active_disk->bs || !active_disk->bs->backing) {
503             error_setg(errp, "Active disk doesn't have backing file");
504             aio_context_release(aio_context);
505             return;
506         }
507 
508         hidden_disk = active_disk->bs->backing;
509         if (!hidden_disk->bs || !hidden_disk->bs->backing) {
510             error_setg(errp, "Hidden disk doesn't have backing file");
511             aio_context_release(aio_context);
512             return;
513         }
514 
515         secondary_disk = hidden_disk->bs->backing;
516         if (!secondary_disk->bs || !bdrv_has_blk(secondary_disk->bs)) {
517             error_setg(errp, "The secondary disk doesn't have block backend");
518             aio_context_release(aio_context);
519             return;
520         }
521 
522         /* verify the length */
523         active_length = bdrv_getlength(active_disk->bs);
524         hidden_length = bdrv_getlength(hidden_disk->bs);
525         disk_length = bdrv_getlength(secondary_disk->bs);
526         if (active_length < 0 || hidden_length < 0 || disk_length < 0 ||
527             active_length != hidden_length || hidden_length != disk_length) {
528             error_setg(errp, "Active disk, hidden disk, secondary disk's length"
529                        " are not the same");
530             aio_context_release(aio_context);
531             return;
532         }
533 
534         /* Must be true, or the bdrv_getlength() calls would have failed */
535         assert(active_disk->bs->drv && hidden_disk->bs->drv);
536 
537         if (!active_disk->bs->drv->bdrv_make_empty ||
538             !hidden_disk->bs->drv->bdrv_make_empty) {
539             error_setg(errp,
540                        "Active disk or hidden disk doesn't support make_empty");
541             aio_context_release(aio_context);
542             return;
543         }
544 
545         /* reopen the backing file in r/w mode */
546         reopen_backing_file(bs, true, &local_err);
547         if (local_err) {
548             error_propagate(errp, local_err);
549             aio_context_release(aio_context);
550             return;
551         }
552 
553         bdrv_ref(hidden_disk->bs);
554         s->hidden_disk = bdrv_attach_child(bs, hidden_disk->bs, "hidden disk",
555                                            &child_of_bds, BDRV_CHILD_DATA,
556                                            &local_err);
557         if (local_err) {
558             error_propagate(errp, local_err);
559             aio_context_release(aio_context);
560             return;
561         }
562 
563         bdrv_ref(secondary_disk->bs);
564         s->secondary_disk = bdrv_attach_child(bs, secondary_disk->bs,
565                                               "secondary disk", &child_of_bds,
566                                               BDRV_CHILD_DATA, &local_err);
567         if (local_err) {
568             error_propagate(errp, local_err);
569             aio_context_release(aio_context);
570             return;
571         }
572 
573         /* start backup job now */
574         error_setg(&s->blocker,
575                    "Block device is in use by internal backup job");
576 
577         top_bs = bdrv_lookup_bs(s->top_id, s->top_id, NULL);
578         if (!top_bs || !bdrv_is_root_node(top_bs) ||
579             !check_top_bs(top_bs, bs)) {
580             error_setg(errp, "No top_bs or it is invalid");
581             reopen_backing_file(bs, false, NULL);
582             aio_context_release(aio_context);
583             return;
584         }
585         bdrv_op_block_all(top_bs, s->blocker);
586         bdrv_op_unblock(top_bs, BLOCK_OP_TYPE_DATAPLANE, s->blocker);
587 
588         s->backup_job = backup_job_create(
589                                 NULL, s->secondary_disk->bs, s->hidden_disk->bs,
590                                 0, MIRROR_SYNC_MODE_NONE, NULL, 0, false, NULL,
591                                 &perf,
592                                 BLOCKDEV_ON_ERROR_REPORT,
593                                 BLOCKDEV_ON_ERROR_REPORT, JOB_INTERNAL,
594                                 backup_job_completed, bs, NULL, &local_err);
595         if (local_err) {
596             error_propagate(errp, local_err);
597             backup_job_cleanup(bs);
598             aio_context_release(aio_context);
599             return;
600         }
601         job_start(&s->backup_job->job);
602         break;
603     default:
604         aio_context_release(aio_context);
605         abort();
606     }
607 
608     s->stage = BLOCK_REPLICATION_RUNNING;
609 
610     if (s->mode == REPLICATION_MODE_SECONDARY) {
611         secondary_do_checkpoint(bs, errp);
612     }
613 
614     s->error = 0;
615     aio_context_release(aio_context);
616 }
617 
618 static void replication_do_checkpoint(ReplicationState *rs, Error **errp)
619 {
620     BlockDriverState *bs = rs->opaque;
621     BDRVReplicationState *s;
622     AioContext *aio_context;
623 
624     aio_context = bdrv_get_aio_context(bs);
625     aio_context_acquire(aio_context);
626     s = bs->opaque;
627 
628     if (s->stage == BLOCK_REPLICATION_DONE ||
629         s->stage == BLOCK_REPLICATION_FAILOVER) {
630         /*
631          * This case happens when a secondary was promoted to primary.
632          * Ignore the request because the secondary side of replication
633          * doesn't have to do anything anymore.
634          */
635         aio_context_release(aio_context);
636         return;
637     }
638 
639     if (s->mode == REPLICATION_MODE_SECONDARY) {
640         secondary_do_checkpoint(bs, errp);
641     }
642     aio_context_release(aio_context);
643 }
644 
645 static void replication_get_error(ReplicationState *rs, Error **errp)
646 {
647     BlockDriverState *bs = rs->opaque;
648     BDRVReplicationState *s;
649     AioContext *aio_context;
650 
651     aio_context = bdrv_get_aio_context(bs);
652     aio_context_acquire(aio_context);
653     s = bs->opaque;
654 
655     if (s->stage == BLOCK_REPLICATION_NONE) {
656         error_setg(errp, "Block replication is not running");
657         aio_context_release(aio_context);
658         return;
659     }
660 
661     if (s->error) {
662         error_setg(errp, "I/O error occurred");
663         aio_context_release(aio_context);
664         return;
665     }
666     aio_context_release(aio_context);
667 }
668 
669 static void replication_done(void *opaque, int ret)
670 {
671     BlockDriverState *bs = opaque;
672     BDRVReplicationState *s = bs->opaque;
673 
674     if (ret == 0) {
675         s->stage = BLOCK_REPLICATION_DONE;
676 
677         bdrv_unref_child(bs, s->secondary_disk);
678         s->secondary_disk = NULL;
679         bdrv_unref_child(bs, s->hidden_disk);
680         s->hidden_disk = NULL;
681         s->error = 0;
682     } else {
683         s->stage = BLOCK_REPLICATION_FAILOVER_FAILED;
684         s->error = -EIO;
685     }
686 }
687 
688 static void replication_stop(ReplicationState *rs, bool failover, Error **errp)
689 {
690     BlockDriverState *bs = rs->opaque;
691     BDRVReplicationState *s;
692     AioContext *aio_context;
693 
694     aio_context = bdrv_get_aio_context(bs);
695     aio_context_acquire(aio_context);
696     s = bs->opaque;
697 
698     if (s->stage == BLOCK_REPLICATION_DONE ||
699         s->stage == BLOCK_REPLICATION_FAILOVER) {
700         /*
701          * This case happens when a secondary was promoted to primary.
702          * Ignore the request because the secondary side of replication
703          * doesn't have to do anything anymore.
704          */
705         aio_context_release(aio_context);
706         return;
707     }
708 
709     if (s->stage != BLOCK_REPLICATION_RUNNING) {
710         error_setg(errp, "Block replication is not running");
711         aio_context_release(aio_context);
712         return;
713     }
714 
715     switch (s->mode) {
716     case REPLICATION_MODE_PRIMARY:
717         s->stage = BLOCK_REPLICATION_DONE;
718         s->error = 0;
719         break;
720     case REPLICATION_MODE_SECONDARY:
721         /*
722          * This BDS will be closed, and the job should be completed
723          * before the BDS is closed, because we will access hidden
724          * disk, secondary disk in backup_job_completed().
725          */
726         if (s->backup_job) {
727             aio_context_release(aio_context);
728             job_cancel_sync(&s->backup_job->job, true);
729             aio_context_acquire(aio_context);
730         }
731 
732         if (!failover) {
733             secondary_do_checkpoint(bs, errp);
734             s->stage = BLOCK_REPLICATION_DONE;
735             aio_context_release(aio_context);
736             return;
737         }
738 
739         s->stage = BLOCK_REPLICATION_FAILOVER;
740         s->commit_job = commit_active_start(
741                             NULL, bs->file->bs, s->secondary_disk->bs,
742                             JOB_INTERNAL, 0, BLOCKDEV_ON_ERROR_REPORT,
743                             NULL, replication_done, bs, true, errp);
744         break;
745     default:
746         aio_context_release(aio_context);
747         abort();
748     }
749     aio_context_release(aio_context);
750 }
751 
752 static const char *const replication_strong_runtime_opts[] = {
753     REPLICATION_MODE,
754     REPLICATION_TOP_ID,
755 
756     NULL
757 };
758 
759 static BlockDriver bdrv_replication = {
760     .format_name                = "replication",
761     .instance_size              = sizeof(BDRVReplicationState),
762 
763     .bdrv_open                  = replication_open,
764     .bdrv_close                 = replication_close,
765     .bdrv_child_perm            = replication_child_perm,
766 
767     .bdrv_getlength             = replication_getlength,
768     .bdrv_co_readv              = replication_co_readv,
769     .bdrv_co_writev             = replication_co_writev,
770 
771     .is_filter                  = true,
772 
773     .has_variable_length        = true,
774     .strong_runtime_opts        = replication_strong_runtime_opts,
775 };
776 
777 static void bdrv_replication_init(void)
778 {
779     bdrv_register(&bdrv_replication);
780 }
781 
782 block_init(bdrv_replication_init);
783