io: add ability to associate an error with a task
[qemu/kevin.git] / blockjob.c
blob513620c199fde95f7448d01e3e3c33fb3724ba21
1 /*
2 * QEMU System Emulator block driver
4 * Copyright (c) 2011 IBM Corp.
5 * Copyright (c) 2012 Red Hat, Inc.
7 * Permission is hereby granted, free of charge, to any person obtaining a copy
8 * of this software and associated documentation files (the "Software"), to deal
9 * in the Software without restriction, including without limitation the rights
10 * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
11 * copies of the Software, and to permit persons to whom the Software is
12 * furnished to do so, subject to the following conditions:
14 * The above copyright notice and this permission notice shall be included in
15 * all copies or substantial portions of the Software.
17 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
18 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
19 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
20 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
21 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
22 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
23 * THE SOFTWARE.
26 #include "qemu/osdep.h"
27 #include "qemu-common.h"
28 #include "trace.h"
29 #include "block/block.h"
30 #include "block/blockjob_int.h"
31 #include "block/block_int.h"
32 #include "sysemu/block-backend.h"
33 #include "qapi/qmp/qerror.h"
34 #include "qapi/qmp/qjson.h"
35 #include "qemu/coroutine.h"
36 #include "qemu/id.h"
37 #include "qmp-commands.h"
38 #include "qemu/timer.h"
39 #include "qapi-event.h"
41 static void block_job_event_cancelled(BlockJob *job);
42 static void block_job_event_completed(BlockJob *job, const char *msg);
44 /* Transactional group of block jobs */
45 struct BlockJobTxn {
47 /* Is this txn being cancelled? */
48 bool aborting;
50 /* List of jobs */
51 QLIST_HEAD(, BlockJob) jobs;
53 /* Reference count */
54 int refcnt;
57 static QLIST_HEAD(, BlockJob) block_jobs = QLIST_HEAD_INITIALIZER(block_jobs);
59 BlockJob *block_job_next(BlockJob *job)
61 if (!job) {
62 return QLIST_FIRST(&block_jobs);
64 return QLIST_NEXT(job, job_list);
67 BlockJob *block_job_get(const char *id)
69 BlockJob *job;
71 QLIST_FOREACH(job, &block_jobs, job_list) {
72 if (job->id && !strcmp(id, job->id)) {
73 return job;
77 return NULL;
80 static void block_job_attached_aio_context(AioContext *new_context,
81 void *opaque)
83 BlockJob *job = opaque;
85 if (job->driver->attached_aio_context) {
86 job->driver->attached_aio_context(job, new_context);
89 block_job_resume(job);
92 static void block_job_drain(BlockJob *job)
94 /* If job is !job->busy this kicks it into the next pause point. */
95 block_job_enter(job);
97 blk_drain(job->blk);
98 if (job->driver->drain) {
99 job->driver->drain(job);
103 static void block_job_detach_aio_context(void *opaque)
105 BlockJob *job = opaque;
107 /* In case the job terminates during aio_poll()... */
108 block_job_ref(job);
110 block_job_pause(job);
112 while (!job->paused && !job->completed) {
113 block_job_drain(job);
116 block_job_unref(job);
119 void block_job_add_bdrv(BlockJob *job, BlockDriverState *bs)
121 job->nodes = g_slist_prepend(job->nodes, bs);
122 bdrv_ref(bs);
123 bdrv_op_block_all(bs, job->blocker);
126 void *block_job_create(const char *job_id, const BlockJobDriver *driver,
127 BlockDriverState *bs, int64_t speed, int flags,
128 BlockCompletionFunc *cb, void *opaque, Error **errp)
130 BlockBackend *blk;
131 BlockJob *job;
133 if (bs->job) {
134 error_setg(errp, QERR_DEVICE_IN_USE, bdrv_get_device_name(bs));
135 return NULL;
138 if (job_id == NULL && !(flags & BLOCK_JOB_INTERNAL)) {
139 job_id = bdrv_get_device_name(bs);
140 if (!*job_id) {
141 error_setg(errp, "An explicit job ID is required for this node");
142 return NULL;
146 if (job_id) {
147 if (flags & BLOCK_JOB_INTERNAL) {
148 error_setg(errp, "Cannot specify job ID for internal block job");
149 return NULL;
152 if (!id_wellformed(job_id)) {
153 error_setg(errp, "Invalid job ID '%s'", job_id);
154 return NULL;
157 if (block_job_get(job_id)) {
158 error_setg(errp, "Job ID '%s' already in use", job_id);
159 return NULL;
163 blk = blk_new();
164 blk_insert_bs(blk, bs);
166 job = g_malloc0(driver->instance_size);
167 error_setg(&job->blocker, "block device is in use by block job: %s",
168 BlockJobType_lookup[driver->job_type]);
169 block_job_add_bdrv(job, bs);
170 bdrv_op_unblock(bs, BLOCK_OP_TYPE_DATAPLANE, job->blocker);
172 job->driver = driver;
173 job->id = g_strdup(job_id);
174 job->blk = blk;
175 job->cb = cb;
176 job->opaque = opaque;
177 job->busy = false;
178 job->paused = true;
179 job->pause_count = 1;
180 job->refcnt = 1;
181 bs->job = job;
183 QLIST_INSERT_HEAD(&block_jobs, job, job_list);
185 blk_add_aio_context_notifier(blk, block_job_attached_aio_context,
186 block_job_detach_aio_context, job);
188 /* Only set speed when necessary to avoid NotSupported error */
189 if (speed != 0) {
190 Error *local_err = NULL;
192 block_job_set_speed(job, speed, &local_err);
193 if (local_err) {
194 block_job_unref(job);
195 error_propagate(errp, local_err);
196 return NULL;
199 return job;
202 bool block_job_is_internal(BlockJob *job)
204 return (job->id == NULL);
207 static bool block_job_started(BlockJob *job)
209 return job->co;
212 void block_job_start(BlockJob *job)
214 assert(job && !block_job_started(job) && job->paused &&
215 !job->busy && job->driver->start);
216 job->co = qemu_coroutine_create(job->driver->start, job);
217 if (--job->pause_count == 0) {
218 job->paused = false;
219 job->busy = true;
220 qemu_coroutine_enter(job->co);
224 void block_job_ref(BlockJob *job)
226 ++job->refcnt;
229 void block_job_unref(BlockJob *job)
231 if (--job->refcnt == 0) {
232 GSList *l;
233 BlockDriverState *bs = blk_bs(job->blk);
234 bs->job = NULL;
235 for (l = job->nodes; l; l = l->next) {
236 bs = l->data;
237 bdrv_op_unblock_all(bs, job->blocker);
238 bdrv_unref(bs);
240 g_slist_free(job->nodes);
241 blk_remove_aio_context_notifier(job->blk,
242 block_job_attached_aio_context,
243 block_job_detach_aio_context, job);
244 blk_unref(job->blk);
245 error_free(job->blocker);
246 g_free(job->id);
247 QLIST_REMOVE(job, job_list);
248 g_free(job);
252 static void block_job_completed_single(BlockJob *job)
254 if (!job->ret) {
255 if (job->driver->commit) {
256 job->driver->commit(job);
258 } else {
259 if (job->driver->abort) {
260 job->driver->abort(job);
263 if (job->driver->clean) {
264 job->driver->clean(job);
267 if (job->cb) {
268 job->cb(job->opaque, job->ret);
271 /* Emit events only if we actually started */
272 if (block_job_started(job)) {
273 if (block_job_is_cancelled(job)) {
274 block_job_event_cancelled(job);
275 } else {
276 const char *msg = NULL;
277 if (job->ret < 0) {
278 msg = strerror(-job->ret);
280 block_job_event_completed(job, msg);
284 if (job->txn) {
285 QLIST_REMOVE(job, txn_list);
286 block_job_txn_unref(job->txn);
288 block_job_unref(job);
291 static void block_job_completed_txn_abort(BlockJob *job)
293 AioContext *ctx;
294 BlockJobTxn *txn = job->txn;
295 BlockJob *other_job, *next;
297 if (txn->aborting) {
299 * We are cancelled by another job, which will handle everything.
301 return;
303 txn->aborting = true;
304 /* We are the first failed job. Cancel other jobs. */
305 QLIST_FOREACH(other_job, &txn->jobs, txn_list) {
306 ctx = blk_get_aio_context(other_job->blk);
307 aio_context_acquire(ctx);
309 QLIST_FOREACH(other_job, &txn->jobs, txn_list) {
310 if (other_job == job || other_job->completed) {
311 /* Other jobs are "effectively" cancelled by us, set the status for
312 * them; this job, however, may or may not be cancelled, depending
313 * on the caller, so leave it. */
314 if (other_job != job) {
315 other_job->cancelled = true;
317 continue;
319 block_job_cancel_sync(other_job);
320 assert(other_job->completed);
322 QLIST_FOREACH_SAFE(other_job, &txn->jobs, txn_list, next) {
323 ctx = blk_get_aio_context(other_job->blk);
324 block_job_completed_single(other_job);
325 aio_context_release(ctx);
329 static void block_job_completed_txn_success(BlockJob *job)
331 AioContext *ctx;
332 BlockJobTxn *txn = job->txn;
333 BlockJob *other_job, *next;
335 * Successful completion, see if there are other running jobs in this
336 * txn.
338 QLIST_FOREACH(other_job, &txn->jobs, txn_list) {
339 if (!other_job->completed) {
340 return;
343 /* We are the last completed job, commit the transaction. */
344 QLIST_FOREACH_SAFE(other_job, &txn->jobs, txn_list, next) {
345 ctx = blk_get_aio_context(other_job->blk);
346 aio_context_acquire(ctx);
347 assert(other_job->ret == 0);
348 block_job_completed_single(other_job);
349 aio_context_release(ctx);
353 void block_job_completed(BlockJob *job, int ret)
355 assert(blk_bs(job->blk)->job == job);
356 assert(!job->completed);
357 job->completed = true;
358 job->ret = ret;
359 if (!job->txn) {
360 block_job_completed_single(job);
361 } else if (ret < 0 || block_job_is_cancelled(job)) {
362 block_job_completed_txn_abort(job);
363 } else {
364 block_job_completed_txn_success(job);
368 void block_job_set_speed(BlockJob *job, int64_t speed, Error **errp)
370 Error *local_err = NULL;
372 if (!job->driver->set_speed) {
373 error_setg(errp, QERR_UNSUPPORTED);
374 return;
376 job->driver->set_speed(job, speed, &local_err);
377 if (local_err) {
378 error_propagate(errp, local_err);
379 return;
382 job->speed = speed;
385 void block_job_complete(BlockJob *job, Error **errp)
387 /* Should not be reachable via external interface for internal jobs */
388 assert(job->id);
389 if (job->pause_count || job->cancelled ||
390 !block_job_started(job) || !job->driver->complete) {
391 error_setg(errp, "The active block job '%s' cannot be completed",
392 job->id);
393 return;
396 job->driver->complete(job, errp);
399 void block_job_pause(BlockJob *job)
401 job->pause_count++;
404 void block_job_user_pause(BlockJob *job)
406 job->user_paused = true;
407 block_job_pause(job);
410 static bool block_job_should_pause(BlockJob *job)
412 return job->pause_count > 0;
415 bool block_job_user_paused(BlockJob *job)
417 return job ? job->user_paused : 0;
420 void coroutine_fn block_job_pause_point(BlockJob *job)
422 assert(job && block_job_started(job));
424 if (!block_job_should_pause(job)) {
425 return;
427 if (block_job_is_cancelled(job)) {
428 return;
431 if (job->driver->pause) {
432 job->driver->pause(job);
435 if (block_job_should_pause(job) && !block_job_is_cancelled(job)) {
436 job->paused = true;
437 job->busy = false;
438 qemu_coroutine_yield(); /* wait for block_job_resume() */
439 job->busy = true;
440 job->paused = false;
443 if (job->driver->resume) {
444 job->driver->resume(job);
448 void block_job_resume(BlockJob *job)
450 assert(job->pause_count > 0);
451 job->pause_count--;
452 if (job->pause_count) {
453 return;
455 block_job_enter(job);
458 void block_job_user_resume(BlockJob *job)
460 if (job && job->user_paused && job->pause_count > 0) {
461 job->user_paused = false;
462 block_job_resume(job);
466 void block_job_enter(BlockJob *job)
468 if (job->co && !job->busy) {
469 qemu_coroutine_enter(job->co);
473 void block_job_cancel(BlockJob *job)
475 if (block_job_started(job)) {
476 job->cancelled = true;
477 block_job_iostatus_reset(job);
478 block_job_enter(job);
479 } else {
480 block_job_completed(job, -ECANCELED);
484 bool block_job_is_cancelled(BlockJob *job)
486 return job->cancelled;
489 void block_job_iostatus_reset(BlockJob *job)
491 job->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
492 if (job->driver->iostatus_reset) {
493 job->driver->iostatus_reset(job);
497 static int block_job_finish_sync(BlockJob *job,
498 void (*finish)(BlockJob *, Error **errp),
499 Error **errp)
501 Error *local_err = NULL;
502 int ret;
504 assert(blk_bs(job->blk)->job == job);
506 block_job_ref(job);
508 finish(job, &local_err);
509 if (local_err) {
510 error_propagate(errp, local_err);
511 block_job_unref(job);
512 return -EBUSY;
514 /* block_job_drain calls block_job_enter, and it should be enough to
515 * induce progress until the job completes or moves to the main thread.
517 while (!job->deferred_to_main_loop && !job->completed) {
518 block_job_drain(job);
520 while (!job->completed) {
521 aio_poll(qemu_get_aio_context(), true);
523 ret = (job->cancelled && job->ret == 0) ? -ECANCELED : job->ret;
524 block_job_unref(job);
525 return ret;
528 /* A wrapper around block_job_cancel() taking an Error ** parameter so it may be
529 * used with block_job_finish_sync() without the need for (rather nasty)
530 * function pointer casts there. */
531 static void block_job_cancel_err(BlockJob *job, Error **errp)
533 block_job_cancel(job);
536 int block_job_cancel_sync(BlockJob *job)
538 return block_job_finish_sync(job, &block_job_cancel_err, NULL);
541 void block_job_cancel_sync_all(void)
543 BlockJob *job;
544 AioContext *aio_context;
546 while ((job = QLIST_FIRST(&block_jobs))) {
547 aio_context = blk_get_aio_context(job->blk);
548 aio_context_acquire(aio_context);
549 block_job_cancel_sync(job);
550 aio_context_release(aio_context);
554 int block_job_complete_sync(BlockJob *job, Error **errp)
556 return block_job_finish_sync(job, &block_job_complete, errp);
559 void block_job_sleep_ns(BlockJob *job, QEMUClockType type, int64_t ns)
561 assert(job->busy);
563 /* Check cancellation *before* setting busy = false, too! */
564 if (block_job_is_cancelled(job)) {
565 return;
568 job->busy = false;
569 if (!block_job_should_pause(job)) {
570 co_aio_sleep_ns(blk_get_aio_context(job->blk), type, ns);
572 job->busy = true;
574 block_job_pause_point(job);
577 void block_job_yield(BlockJob *job)
579 assert(job->busy);
581 /* Check cancellation *before* setting busy = false, too! */
582 if (block_job_is_cancelled(job)) {
583 return;
586 job->busy = false;
587 if (!block_job_should_pause(job)) {
588 qemu_coroutine_yield();
590 job->busy = true;
592 block_job_pause_point(job);
595 BlockJobInfo *block_job_query(BlockJob *job, Error **errp)
597 BlockJobInfo *info;
599 if (block_job_is_internal(job)) {
600 error_setg(errp, "Cannot query QEMU internal jobs");
601 return NULL;
603 info = g_new0(BlockJobInfo, 1);
604 info->type = g_strdup(BlockJobType_lookup[job->driver->job_type]);
605 info->device = g_strdup(job->id);
606 info->len = job->len;
607 info->busy = job->busy;
608 info->paused = job->pause_count > 0;
609 info->offset = job->offset;
610 info->speed = job->speed;
611 info->io_status = job->iostatus;
612 info->ready = job->ready;
613 return info;
616 static void block_job_iostatus_set_err(BlockJob *job, int error)
618 if (job->iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
619 job->iostatus = error == ENOSPC ? BLOCK_DEVICE_IO_STATUS_NOSPACE :
620 BLOCK_DEVICE_IO_STATUS_FAILED;
624 static void block_job_event_cancelled(BlockJob *job)
626 if (block_job_is_internal(job)) {
627 return;
630 qapi_event_send_block_job_cancelled(job->driver->job_type,
631 job->id,
632 job->len,
633 job->offset,
634 job->speed,
635 &error_abort);
638 static void block_job_event_completed(BlockJob *job, const char *msg)
640 if (block_job_is_internal(job)) {
641 return;
644 qapi_event_send_block_job_completed(job->driver->job_type,
645 job->id,
646 job->len,
647 job->offset,
648 job->speed,
649 !!msg,
650 msg,
651 &error_abort);
654 void block_job_event_ready(BlockJob *job)
656 job->ready = true;
658 if (block_job_is_internal(job)) {
659 return;
662 qapi_event_send_block_job_ready(job->driver->job_type,
663 job->id,
664 job->len,
665 job->offset,
666 job->speed, &error_abort);
669 BlockErrorAction block_job_error_action(BlockJob *job, BlockdevOnError on_err,
670 int is_read, int error)
672 BlockErrorAction action;
674 switch (on_err) {
675 case BLOCKDEV_ON_ERROR_ENOSPC:
676 case BLOCKDEV_ON_ERROR_AUTO:
677 action = (error == ENOSPC) ?
678 BLOCK_ERROR_ACTION_STOP : BLOCK_ERROR_ACTION_REPORT;
679 break;
680 case BLOCKDEV_ON_ERROR_STOP:
681 action = BLOCK_ERROR_ACTION_STOP;
682 break;
683 case BLOCKDEV_ON_ERROR_REPORT:
684 action = BLOCK_ERROR_ACTION_REPORT;
685 break;
686 case BLOCKDEV_ON_ERROR_IGNORE:
687 action = BLOCK_ERROR_ACTION_IGNORE;
688 break;
689 default:
690 abort();
692 if (!block_job_is_internal(job)) {
693 qapi_event_send_block_job_error(job->id,
694 is_read ? IO_OPERATION_TYPE_READ :
695 IO_OPERATION_TYPE_WRITE,
696 action, &error_abort);
698 if (action == BLOCK_ERROR_ACTION_STOP) {
699 /* make the pause user visible, which will be resumed from QMP. */
700 block_job_user_pause(job);
701 block_job_iostatus_set_err(job, error);
703 return action;
706 typedef struct {
707 BlockJob *job;
708 AioContext *aio_context;
709 BlockJobDeferToMainLoopFn *fn;
710 void *opaque;
711 } BlockJobDeferToMainLoopData;
713 static void block_job_defer_to_main_loop_bh(void *opaque)
715 BlockJobDeferToMainLoopData *data = opaque;
716 AioContext *aio_context;
718 /* Prevent race with block_job_defer_to_main_loop() */
719 aio_context_acquire(data->aio_context);
721 /* Fetch BDS AioContext again, in case it has changed */
722 aio_context = blk_get_aio_context(data->job->blk);
723 aio_context_acquire(aio_context);
725 data->job->deferred_to_main_loop = false;
726 data->fn(data->job, data->opaque);
728 aio_context_release(aio_context);
730 aio_context_release(data->aio_context);
732 g_free(data);
735 void block_job_defer_to_main_loop(BlockJob *job,
736 BlockJobDeferToMainLoopFn *fn,
737 void *opaque)
739 BlockJobDeferToMainLoopData *data = g_malloc(sizeof(*data));
740 data->job = job;
741 data->aio_context = blk_get_aio_context(job->blk);
742 data->fn = fn;
743 data->opaque = opaque;
744 job->deferred_to_main_loop = true;
746 aio_bh_schedule_oneshot(qemu_get_aio_context(),
747 block_job_defer_to_main_loop_bh, data);
750 BlockJobTxn *block_job_txn_new(void)
752 BlockJobTxn *txn = g_new0(BlockJobTxn, 1);
753 QLIST_INIT(&txn->jobs);
754 txn->refcnt = 1;
755 return txn;
758 static void block_job_txn_ref(BlockJobTxn *txn)
760 txn->refcnt++;
763 void block_job_txn_unref(BlockJobTxn *txn)
765 if (txn && --txn->refcnt == 0) {
766 g_free(txn);
770 void block_job_txn_add_job(BlockJobTxn *txn, BlockJob *job)
772 if (!txn) {
773 return;
776 assert(!job->txn);
777 job->txn = txn;
779 QLIST_INSERT_HEAD(&txn->jobs, job, txn_list);
780 block_job_txn_ref(txn);