blockjob.c

   1 /*
   2  * QEMU System Emulator block driver
   3  *
   4  * Copyright (c) 2011 IBM Corp.
   5  * Copyright (c) 2012 Red Hat, Inc.
   6  *
   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:
  13  *
  14  * The above copyright notice and this permission notice shall be included in
  15  * all copies or substantial portions of the Software.
  16  *
  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.
  24  */
  25
  26 #include "qemu/osdep.h"
  27 #include "block/block.h"
  28 #include "block/blockjob_int.h"
  29 #include "block/block_int.h"
  30 #include "block/trace.h"
  31 #include "sysemu/block-backend.h"
  32 #include "qapi/error.h"
  33 #include "qapi/qapi-events-block-core.h"
  34 #include "qapi/qmp/qerror.h"
  35 #include "qemu/coroutine.h"
  36 #include "qemu/main-loop.h"
  37 #include "qemu/timer.h"
  38
  39 static bool is_block_job(Job *job)
  40 {
  41     return job_type(job) == JOB_TYPE_BACKUP ||
  42            job_type(job) == JOB_TYPE_COMMIT ||
  43            job_type(job) == JOB_TYPE_MIRROR ||
  44            job_type(job) == JOB_TYPE_STREAM;
  45 }
  46
  47 BlockJob *block_job_next_locked(BlockJob *bjob)
  48 {
  49     Job *job = bjob ? &bjob->job : NULL;
  50     GLOBAL_STATE_CODE();
  51
  52     do {
  53         job = job_next_locked(job);
  54     } while (job && !is_block_job(job));
  55
  56     return job ? container_of(job, BlockJob, job) : NULL;
  57 }
  58
  59 BlockJob *block_job_get_locked(const char *id)
  60 {
  61     Job *job = job_get_locked(id);
  62     GLOBAL_STATE_CODE();
  63
  64     if (job && is_block_job(job)) {
  65         return container_of(job, BlockJob, job);
  66     } else {
  67         return NULL;
  68     }
  69 }
  70
  71 BlockJob *block_job_get(const char *id)
  72 {
  73     JOB_LOCK_GUARD();
  74     return block_job_get_locked(id);
  75 }
  76
  77 void block_job_free(Job *job)
  78 {
  79     BlockJob *bjob = container_of(job, BlockJob, job);
  80     GLOBAL_STATE_CODE();
  81
  82     block_job_remove_all_bdrv(bjob);
  83     ratelimit_destroy(&bjob->limit);
  84     error_free(bjob->blocker);
  85 }
  86
  87 static char *child_job_get_parent_desc(BdrvChild *c)
  88 {
  89     BlockJob *job = c->opaque;
  90     return g_strdup_printf("%s job '%s'", job_type_str(&job->job), job->job.id);
  91 }
  92
  93 static void child_job_drained_begin(BdrvChild *c)
  94 {
  95     BlockJob *job = c->opaque;
  96     job_pause(&job->job);
  97 }
  98
  99 static bool child_job_drained_poll(BdrvChild *c)
 100 {
 101     BlockJob *bjob = c->opaque;
 102     Job *job = &bjob->job;
 103     const BlockJobDriver *drv = block_job_driver(bjob);
 104
 105     /* An inactive or completed job doesn't have any pending requests. Jobs
 106      * with !job->busy are either already paused or have a pause point after
 107      * being reentered, so no job driver code will run before they pause. */
 108     WITH_JOB_LOCK_GUARD() {
 109         if (!job->busy || job_is_completed_locked(job)) {
 110             return false;
 111         }
 112     }
 113
 114     /* Otherwise, assume that it isn't fully stopped yet, but allow the job to
 115      * override this assumption. */
 116     if (drv->drained_poll) {
 117         return drv->drained_poll(bjob);
 118     } else {
 119         return true;
 120     }
 121 }
 122
 123 static void child_job_drained_end(BdrvChild *c, int *drained_end_counter)
 124 {
 125     BlockJob *job = c->opaque;
 126     job_resume(&job->job);
 127 }
 128
 129 typedef struct BdrvStateChildJobContext {
 130     AioContext *new_ctx;
 131     BlockJob *job;
 132 } BdrvStateChildJobContext;
 133
 134 static void child_job_set_aio_ctx_commit(void *opaque)
 135 {
 136     BdrvStateChildJobContext *s = opaque;
 137     BlockJob *job = s->job;
 138
 139     job_set_aio_context(&job->job, s->new_ctx);
 140 }
 141
 142 static TransactionActionDrv change_child_job_context = {
 143     .commit = child_job_set_aio_ctx_commit,
 144     .clean = g_free,
 145 };
 146
 147 static bool child_job_change_aio_ctx(BdrvChild *c, AioContext *ctx,
 148                                      GHashTable *visited, Transaction *tran,
 149                                      Error **errp)
 150 {
 151     BlockJob *job = c->opaque;
 152     BdrvStateChildJobContext *s;
 153     GSList *l;
 154
 155     for (l = job->nodes; l; l = l->next) {
 156         BdrvChild *sibling = l->data;
 157         if (!bdrv_child_change_aio_context(sibling, ctx, visited,
 158                                            tran, errp)) {
 159             return false;
 160         }
 161     }
 162
 163     s = g_new(BdrvStateChildJobContext, 1);
 164     *s = (BdrvStateChildJobContext) {
 165         .new_ctx = ctx,
 166         .job = job,
 167     };
 168
 169     tran_add(tran, &change_child_job_context, s);
 170     return true;
 171 }
 172
 173 static bool child_job_can_set_aio_ctx(BdrvChild *c, AioContext *ctx,
 174                                       GSList **ignore, Error **errp)
 175 {
 176     BlockJob *job = c->opaque;
 177     GSList *l;
 178
 179     for (l = job->nodes; l; l = l->next) {
 180         BdrvChild *sibling = l->data;
 181         if (!bdrv_child_can_set_aio_context(sibling, ctx, ignore, errp)) {
 182             return false;
 183         }
 184     }
 185     return true;
 186 }
 187
 188 static void child_job_set_aio_ctx(BdrvChild *c, AioContext *ctx,
 189                                   GSList **ignore)
 190 {
 191     BlockJob *job = c->opaque;
 192     GSList *l;
 193
 194     for (l = job->nodes; l; l = l->next) {
 195         BdrvChild *sibling = l->data;
 196         if (g_slist_find(*ignore, sibling)) {
 197             continue;
 198         }
 199         *ignore = g_slist_prepend(*ignore, sibling);
 200         bdrv_set_aio_context_ignore(sibling->bs, ctx, ignore);
 201     }
 202
 203     job_set_aio_context(&job->job, ctx);
 204 }
 205
 206 static AioContext *child_job_get_parent_aio_context(BdrvChild *c)
 207 {
 208     BlockJob *job = c->opaque;
 209     GLOBAL_STATE_CODE();
 210
 211     return job->job.aio_context;
 212 }
 213
 214 static const BdrvChildClass child_job = {
 215     .get_parent_desc    = child_job_get_parent_desc,
 216     .drained_begin      = child_job_drained_begin,
 217     .drained_poll       = child_job_drained_poll,
 218     .drained_end        = child_job_drained_end,
 219     .can_set_aio_ctx    = child_job_can_set_aio_ctx,
 220     .set_aio_ctx        = child_job_set_aio_ctx,
 221     .change_aio_ctx     = child_job_change_aio_ctx,
 222     .stay_at_node       = true,
 223     .get_parent_aio_context = child_job_get_parent_aio_context,
 224 };
 225
 226 void block_job_remove_all_bdrv(BlockJob *job)
 227 {
 228     GLOBAL_STATE_CODE();
 229     /*
 230      * bdrv_root_unref_child() may reach child_job_[can_]set_aio_ctx(),
 231      * which will also traverse job->nodes, so consume the list one by
 232      * one to make sure that such a concurrent access does not attempt
 233      * to process an already freed BdrvChild.
 234      */
 235     while (job->nodes) {
 236         GSList *l = job->nodes;
 237         BdrvChild *c = l->data;
 238
 239         job->nodes = l->next;
 240
 241         bdrv_op_unblock_all(c->bs, job->blocker);
 242         bdrv_root_unref_child(c);
 243
 244         g_slist_free_1(l);
 245     }
 246 }
 247
 248 bool block_job_has_bdrv(BlockJob *job, BlockDriverState *bs)
 249 {
 250     GSList *el;
 251     GLOBAL_STATE_CODE();
 252
 253     for (el = job->nodes; el; el = el->next) {
 254         BdrvChild *c = el->data;
 255         if (c->bs == bs) {
 256             return true;
 257         }
 258     }
 259
 260     return false;
 261 }
 262
 263 int block_job_add_bdrv(BlockJob *job, const char *name, BlockDriverState *bs,
 264                        uint64_t perm, uint64_t shared_perm, Error **errp)
 265 {
 266     BdrvChild *c;
 267     bool need_context_ops;
 268     GLOBAL_STATE_CODE();
 269
 270     bdrv_ref(bs);
 271
 272     need_context_ops = bdrv_get_aio_context(bs) != job->job.aio_context;
 273
 274     if (need_context_ops && job->job.aio_context != qemu_get_aio_context()) {
 275         aio_context_release(job->job.aio_context);
 276     }
 277     c = bdrv_root_attach_child(bs, name, &child_job, 0, perm, shared_perm, job,
 278                                errp);
 279     if (need_context_ops && job->job.aio_context != qemu_get_aio_context()) {
 280         aio_context_acquire(job->job.aio_context);
 281     }
 282     if (c == NULL) {
 283         return -EPERM;
 284     }
 285
 286     job->nodes = g_slist_prepend(job->nodes, c);
 287     bdrv_op_block_all(bs, job->blocker);
 288
 289     return 0;
 290 }
 291
 292 /* Called with job_mutex lock held. */
 293 static void block_job_on_idle_locked(Notifier *n, void *opaque)
 294 {
 295     aio_wait_kick();
 296 }
 297
 298 bool block_job_is_internal(BlockJob *job)
 299 {
 300     return (job->job.id == NULL);
 301 }
 302
 303 const BlockJobDriver *block_job_driver(BlockJob *job)
 304 {
 305     return container_of(job->job.driver, BlockJobDriver, job_driver);
 306 }
 307
 308 /* Assumes the job_mutex is held */
 309 static bool job_timer_pending(Job *job)
 310 {
 311     return timer_pending(&job->sleep_timer);
 312 }
 313
 314 bool block_job_set_speed_locked(BlockJob *job, int64_t speed, Error **errp)
 315 {
 316     const BlockJobDriver *drv = block_job_driver(job);
 317     int64_t old_speed = job->speed;
 318
 319     GLOBAL_STATE_CODE();
 320
 321     if (job_apply_verb_locked(&job->job, JOB_VERB_SET_SPEED, errp) < 0) {
 322         return false;
 323     }
 324     if (speed < 0) {
 325         error_setg(errp, QERR_INVALID_PARAMETER_VALUE, "speed",
 326                    "a non-negative value");
 327         return false;
 328     }
 329
 330     ratelimit_set_speed(&job->limit, speed, BLOCK_JOB_SLICE_TIME);
 331
 332     job->speed = speed;
 333
 334     if (drv->set_speed) {
 335         job_unlock();
 336         drv->set_speed(job, speed);
 337         job_lock();
 338     }
 339
 340     if (speed && speed <= old_speed) {
 341         return true;
 342     }
 343
 344     /* kick only if a timer is pending */
 345     job_enter_cond_locked(&job->job, job_timer_pending);
 346
 347     return true;
 348 }
 349
 350 static bool block_job_set_speed(BlockJob *job, int64_t speed, Error **errp)
 351 {
 352     JOB_LOCK_GUARD();
 353     return block_job_set_speed_locked(job, speed, errp);
 354 }
 355
 356 int64_t block_job_ratelimit_get_delay(BlockJob *job, uint64_t n)
 357 {
 358     IO_CODE();
 359     return ratelimit_calculate_delay(&job->limit, n);
 360 }
 361
 362 BlockJobInfo *block_job_query_locked(BlockJob *job, Error **errp)
 363 {
 364     BlockJobInfo *info;
 365     uint64_t progress_current, progress_total;
 366
 367     GLOBAL_STATE_CODE();
 368
 369     if (block_job_is_internal(job)) {
 370         error_setg(errp, "Cannot query QEMU internal jobs");
 371         return NULL;
 372     }
 373
 374     progress_get_snapshot(&job->job.progress, &progress_current,
 375                           &progress_total);
 376
 377     info = g_new0(BlockJobInfo, 1);
 378     info->type      = g_strdup(job_type_str(&job->job));
 379     info->device    = g_strdup(job->job.id);
 380     info->busy      = job->job.busy;
 381     info->paused    = job->job.pause_count > 0;
 382     info->offset    = progress_current;
 383     info->len       = progress_total;
 384     info->speed     = job->speed;
 385     info->io_status = job->iostatus;
 386     info->ready     = job_is_ready_locked(&job->job),
 387     info->status    = job->job.status;
 388     info->auto_finalize = job->job.auto_finalize;
 389     info->auto_dismiss  = job->job.auto_dismiss;
 390     if (job->job.ret) {
 391         info->has_error = true;
 392         info->error = job->job.err ?
 393                         g_strdup(error_get_pretty(job->job.err)) :
 394                         g_strdup(strerror(-job->job.ret));
 395     }
 396     return info;
 397 }
 398
 399 /* Called with job lock held */
 400 static void block_job_iostatus_set_err_locked(BlockJob *job, int error)
 401 {
 402     if (job->iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
 403         job->iostatus = error == ENOSPC ? BLOCK_DEVICE_IO_STATUS_NOSPACE :
 404                                           BLOCK_DEVICE_IO_STATUS_FAILED;
 405     }
 406 }
 407
 408 /* Called with job_mutex lock held. */
 409 static void block_job_event_cancelled_locked(Notifier *n, void *opaque)
 410 {
 411     BlockJob *job = opaque;
 412     uint64_t progress_current, progress_total;
 413
 414     if (block_job_is_internal(job)) {
 415         return;
 416     }
 417
 418     progress_get_snapshot(&job->job.progress, &progress_current,
 419                           &progress_total);
 420
 421     qapi_event_send_block_job_cancelled(job_type(&job->job),
 422                                         job->job.id,
 423                                         progress_total,
 424                                         progress_current,
 425                                         job->speed);
 426 }
 427
 428 /* Called with job_mutex lock held. */
 429 static void block_job_event_completed_locked(Notifier *n, void *opaque)
 430 {
 431     BlockJob *job = opaque;
 432     const char *msg = NULL;
 433     uint64_t progress_current, progress_total;
 434
 435     if (block_job_is_internal(job)) {
 436         return;
 437     }
 438
 439     if (job->job.ret < 0) {
 440         msg = error_get_pretty(job->job.err);
 441     }
 442
 443     progress_get_snapshot(&job->job.progress, &progress_current,
 444                           &progress_total);
 445
 446     qapi_event_send_block_job_completed(job_type(&job->job),
 447                                         job->job.id,
 448                                         progress_total,
 449                                         progress_current,
 450                                         job->speed,
 451                                         !!msg,
 452                                         msg);
 453 }
 454
 455 /* Called with job_mutex lock held. */
 456 static void block_job_event_pending_locked(Notifier *n, void *opaque)
 457 {
 458     BlockJob *job = opaque;
 459
 460     if (block_job_is_internal(job)) {
 461         return;
 462     }
 463
 464     qapi_event_send_block_job_pending(job_type(&job->job),
 465                                       job->job.id);
 466 }
 467
 468 /* Called with job_mutex lock held. */
 469 static void block_job_event_ready_locked(Notifier *n, void *opaque)
 470 {
 471     BlockJob *job = opaque;
 472     uint64_t progress_current, progress_total;
 473
 474     if (block_job_is_internal(job)) {
 475         return;
 476     }
 477
 478     progress_get_snapshot(&job->job.progress, &progress_current,
 479                           &progress_total);
 480
 481     qapi_event_send_block_job_ready(job_type(&job->job),
 482                                     job->job.id,
 483                                     progress_total,
 484                                     progress_current,
 485                                     job->speed);
 486 }
 487
 488
 489 void *block_job_create(const char *job_id, const BlockJobDriver *driver,
 490                        JobTxn *txn, BlockDriverState *bs, uint64_t perm,
 491                        uint64_t shared_perm, int64_t speed, int flags,
 492                        BlockCompletionFunc *cb, void *opaque, Error **errp)
 493 {
 494     BlockJob *job;
 495     int ret;
 496     GLOBAL_STATE_CODE();
 497
 498     if (job_id == NULL && !(flags & JOB_INTERNAL)) {
 499         job_id = bdrv_get_device_name(bs);
 500     }
 501
 502     job = job_create(job_id, &driver->job_driver, txn, bdrv_get_aio_context(bs),
 503                      flags, cb, opaque, errp);
 504     if (job == NULL) {
 505         return NULL;
 506     }
 507
 508     assert(is_block_job(&job->job));
 509     assert(job->job.driver->free == &block_job_free);
 510     assert(job->job.driver->user_resume == &block_job_user_resume);
 511
 512     ratelimit_init(&job->limit);
 513
 514     job->finalize_cancelled_notifier.notify = block_job_event_cancelled_locked;
 515     job->finalize_completed_notifier.notify = block_job_event_completed_locked;
 516     job->pending_notifier.notify = block_job_event_pending_locked;
 517     job->ready_notifier.notify = block_job_event_ready_locked;
 518     job->idle_notifier.notify = block_job_on_idle_locked;
 519
 520     WITH_JOB_LOCK_GUARD() {
 521         notifier_list_add(&job->job.on_finalize_cancelled,
 522                           &job->finalize_cancelled_notifier);
 523         notifier_list_add(&job->job.on_finalize_completed,
 524                           &job->finalize_completed_notifier);
 525         notifier_list_add(&job->job.on_pending, &job->pending_notifier);
 526         notifier_list_add(&job->job.on_ready, &job->ready_notifier);
 527         notifier_list_add(&job->job.on_idle, &job->idle_notifier);
 528     }
 529
 530     error_setg(&job->blocker, "block device is in use by block job: %s",
 531                job_type_str(&job->job));
 532
 533     ret = block_job_add_bdrv(job, "main node", bs, perm, shared_perm, errp);
 534     if (ret < 0) {
 535         goto fail;
 536     }
 537
 538     bdrv_op_unblock(bs, BLOCK_OP_TYPE_DATAPLANE, job->blocker);
 539
 540     if (!block_job_set_speed(job, speed, errp)) {
 541         goto fail;
 542     }
 543
 544     return job;
 545
 546 fail:
 547     job_early_fail(&job->job);
 548     return NULL;
 549 }
 550
 551 void block_job_iostatus_reset_locked(BlockJob *job)
 552 {
 553     GLOBAL_STATE_CODE();
 554     if (job->iostatus == BLOCK_DEVICE_IO_STATUS_OK) {
 555         return;
 556     }
 557     assert(job->job.user_paused && job->job.pause_count > 0);
 558     job->iostatus = BLOCK_DEVICE_IO_STATUS_OK;
 559 }
 560
 561 static void block_job_iostatus_reset(BlockJob *job)
 562 {
 563     JOB_LOCK_GUARD();
 564     block_job_iostatus_reset_locked(job);
 565 }
 566
 567 void block_job_user_resume(Job *job)
 568 {
 569     BlockJob *bjob = container_of(job, BlockJob, job);
 570     GLOBAL_STATE_CODE();
 571     block_job_iostatus_reset(bjob);
 572 }
 573
 574 BlockErrorAction block_job_error_action(BlockJob *job, BlockdevOnError on_err,
 575                                         int is_read, int error)
 576 {
 577     BlockErrorAction action;
 578     IO_CODE();
 579
 580     switch (on_err) {
 581     case BLOCKDEV_ON_ERROR_ENOSPC:
 582     case BLOCKDEV_ON_ERROR_AUTO:
 583         action = (error == ENOSPC) ?
 584                  BLOCK_ERROR_ACTION_STOP : BLOCK_ERROR_ACTION_REPORT;
 585         break;
 586     case BLOCKDEV_ON_ERROR_STOP:
 587         action = BLOCK_ERROR_ACTION_STOP;
 588         break;
 589     case BLOCKDEV_ON_ERROR_REPORT:
 590         action = BLOCK_ERROR_ACTION_REPORT;
 591         break;
 592     case BLOCKDEV_ON_ERROR_IGNORE:
 593         action = BLOCK_ERROR_ACTION_IGNORE;
 594         break;
 595     default:
 596         abort();
 597     }
 598     if (!block_job_is_internal(job)) {
 599         qapi_event_send_block_job_error(job->job.id,
 600                                         is_read ? IO_OPERATION_TYPE_READ :
 601                                         IO_OPERATION_TYPE_WRITE,
 602                                         action);
 603     }
 604     if (action == BLOCK_ERROR_ACTION_STOP) {
 605         WITH_JOB_LOCK_GUARD() {
 606             if (!job->job.user_paused) {
 607                 job_pause_locked(&job->job);
 608                 /*
 609                  * make the pause user visible, which will be
 610                  * resumed from QMP.
 611                  */
 612                 job->job.user_paused = true;
 613             }
 614             block_job_iostatus_set_err_locked(job, error);
 615         }
 616     }
 617     return action;
 618 }
 619
 620 AioContext *block_job_get_aio_context(BlockJob *job)
 621 {
 622     GLOBAL_STATE_CODE();
 623     return job->job.aio_context;
 624 }