1 // SPDX-License-Identifier: GPL-2.0
2 /*
3  * Copyright (C) 1991, 1992 Linus Torvalds
4  * Copyright (C) 1994,      Karl Keyte: Added support for disk statistics
5  * Elevator latency, (C) 2000  Andrea Arcangeli <andrea@suse.de> SuSE
6  * Queue request tables / lock, selectable elevator, Jens Axboe <axboe@suse.de>
7  * kernel-doc documentation started by NeilBrown <neilb@cse.unsw.edu.au>
8  *	-  July2000
9  * bio rewrite, highmem i/o, etc, Jens Axboe <axboe@suse.de> - may 2001
10  */
11 
12 /*
13  * This handles all read/write requests to block devices
14  */
15 #include <linux/kernel.h>
16 #include <linux/module.h>
17 #include <linux/bio.h>
18 #include <linux/blkdev.h>
19 #include <linux/blk-pm.h>
20 #include <linux/blk-integrity.h>
21 #include <linux/highmem.h>
22 #include <linux/mm.h>
23 #include <linux/pagemap.h>
24 #include <linux/kernel_stat.h>
25 #include <linux/string.h>
26 #include <linux/init.h>
27 #include <linux/completion.h>
28 #include <linux/slab.h>
29 #include <linux/swap.h>
30 #include <linux/writeback.h>
31 #include <linux/task_io_accounting_ops.h>
32 #include <linux/fault-inject.h>
33 #include <linux/list_sort.h>
34 #include <linux/delay.h>
35 #include <linux/ratelimit.h>
36 #include <linux/pm_runtime.h>
37 #include <linux/t10-pi.h>
38 #include <linux/debugfs.h>
39 #include <linux/bpf.h>
40 #include <linux/part_stat.h>
41 #include <linux/sched/sysctl.h>
42 #include <linux/blk-crypto.h>
43 
44 #define CREATE_TRACE_POINTS
45 #include <trace/events/block.h>
46 
47 #include "blk.h"
48 #include "blk-mq-sched.h"
49 #include "blk-pm.h"
50 #include "blk-cgroup.h"
51 #include "blk-throttle.h"
52 
53 struct dentry *blk_debugfs_root;
54 
55 EXPORT_TRACEPOINT_SYMBOL_GPL(block_bio_remap);
56 EXPORT_TRACEPOINT_SYMBOL_GPL(block_rq_remap);
57 EXPORT_TRACEPOINT_SYMBOL_GPL(block_bio_complete);
58 EXPORT_TRACEPOINT_SYMBOL_GPL(block_split);
59 EXPORT_TRACEPOINT_SYMBOL_GPL(block_unplug);
60 EXPORT_TRACEPOINT_SYMBOL_GPL(block_rq_insert);
61 
62 static DEFINE_IDA(blk_queue_ida);
63 
64 /*
65  * For queue allocation
66  */
67 static struct kmem_cache *blk_requestq_cachep;
68 
69 /*
70  * Controlling structure to kblockd
71  */
72 static struct workqueue_struct *kblockd_workqueue;
73 
74 /**
75  * blk_queue_flag_set - atomically set a queue flag
76  * @flag: flag to be set
77  * @q: request queue
78  */
blk_queue_flag_set(unsigned int flag,struct request_queue * q)79 void blk_queue_flag_set(unsigned int flag, struct request_queue *q)
80 {
81 	set_bit(flag, &q->queue_flags);
82 }
83 EXPORT_SYMBOL(blk_queue_flag_set);
84 
85 /**
86  * blk_queue_flag_clear - atomically clear a queue flag
87  * @flag: flag to be cleared
88  * @q: request queue
89  */
blk_queue_flag_clear(unsigned int flag,struct request_queue * q)90 void blk_queue_flag_clear(unsigned int flag, struct request_queue *q)
91 {
92 	clear_bit(flag, &q->queue_flags);
93 }
94 EXPORT_SYMBOL(blk_queue_flag_clear);
95 
96 /**
97  * blk_queue_flag_test_and_set - atomically test and set a queue flag
98  * @flag: flag to be set
99  * @q: request queue
100  *
101  * Returns the previous value of @flag - 0 if the flag was not set and 1 if
102  * the flag was already set.
103  */
blk_queue_flag_test_and_set(unsigned int flag,struct request_queue * q)104 bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q)
105 {
106 	return test_and_set_bit(flag, &q->queue_flags);
107 }
108 EXPORT_SYMBOL_GPL(blk_queue_flag_test_and_set);
109 
110 #define REQ_OP_NAME(name) [REQ_OP_##name] = #name
111 static const char *const blk_op_name[] = {
112 	REQ_OP_NAME(READ),
113 	REQ_OP_NAME(WRITE),
114 	REQ_OP_NAME(FLUSH),
115 	REQ_OP_NAME(DISCARD),
116 	REQ_OP_NAME(SECURE_ERASE),
117 	REQ_OP_NAME(ZONE_RESET),
118 	REQ_OP_NAME(ZONE_RESET_ALL),
119 	REQ_OP_NAME(ZONE_OPEN),
120 	REQ_OP_NAME(ZONE_CLOSE),
121 	REQ_OP_NAME(ZONE_FINISH),
122 	REQ_OP_NAME(ZONE_APPEND),
123 	REQ_OP_NAME(WRITE_ZEROES),
124 	REQ_OP_NAME(DRV_IN),
125 	REQ_OP_NAME(DRV_OUT),
126 };
127 #undef REQ_OP_NAME
128 
129 /**
130  * blk_op_str - Return string XXX in the REQ_OP_XXX.
131  * @op: REQ_OP_XXX.
132  *
133  * Description: Centralize block layer function to convert REQ_OP_XXX into
134  * string format. Useful in the debugging and tracing bio or request. For
135  * invalid REQ_OP_XXX it returns string "UNKNOWN".
136  */
blk_op_str(enum req_op op)137 inline const char *blk_op_str(enum req_op op)
138 {
139 	const char *op_str = "UNKNOWN";
140 
141 	if (op < ARRAY_SIZE(blk_op_name) && blk_op_name[op])
142 		op_str = blk_op_name[op];
143 
144 	return op_str;
145 }
146 EXPORT_SYMBOL_GPL(blk_op_str);
147 
148 static const struct {
149 	int		errno;
150 	const char	*name;
151 } blk_errors[] = {
152 	[BLK_STS_OK]		= { 0,		"" },
153 	[BLK_STS_NOTSUPP]	= { -EOPNOTSUPP, "operation not supported" },
154 	[BLK_STS_TIMEOUT]	= { -ETIMEDOUT,	"timeout" },
155 	[BLK_STS_NOSPC]		= { -ENOSPC,	"critical space allocation" },
156 	[BLK_STS_TRANSPORT]	= { -ENOLINK,	"recoverable transport" },
157 	[BLK_STS_TARGET]	= { -EREMOTEIO,	"critical target" },
158 	[BLK_STS_NEXUS]		= { -EBADE,	"critical nexus" },
159 	[BLK_STS_MEDIUM]	= { -ENODATA,	"critical medium" },
160 	[BLK_STS_PROTECTION]	= { -EILSEQ,	"protection" },
161 	[BLK_STS_RESOURCE]	= { -ENOMEM,	"kernel resource" },
162 	[BLK_STS_DEV_RESOURCE]	= { -EBUSY,	"device resource" },
163 	[BLK_STS_AGAIN]		= { -EAGAIN,	"nonblocking retry" },
164 	[BLK_STS_OFFLINE]	= { -ENODEV,	"device offline" },
165 
166 	/* device mapper special case, should not leak out: */
167 	[BLK_STS_DM_REQUEUE]	= { -EREMCHG, "dm internal retry" },
168 
169 	/* zone device specific errors */
170 	[BLK_STS_ZONE_OPEN_RESOURCE]	= { -ETOOMANYREFS, "open zones exceeded" },
171 	[BLK_STS_ZONE_ACTIVE_RESOURCE]	= { -EOVERFLOW, "active zones exceeded" },
172 
173 	/* everything else not covered above: */
174 	[BLK_STS_IOERR]		= { -EIO,	"I/O" },
175 };
176 
errno_to_blk_status(int errno)177 blk_status_t errno_to_blk_status(int errno)
178 {
179 	int i;
180 
181 	for (i = 0; i < ARRAY_SIZE(blk_errors); i++) {
182 		if (blk_errors[i].errno == errno)
183 			return (__force blk_status_t)i;
184 	}
185 
186 	return BLK_STS_IOERR;
187 }
188 EXPORT_SYMBOL_GPL(errno_to_blk_status);
189 
blk_status_to_errno(blk_status_t status)190 int blk_status_to_errno(blk_status_t status)
191 {
192 	int idx = (__force int)status;
193 
194 	if (WARN_ON_ONCE(idx >= ARRAY_SIZE(blk_errors)))
195 		return -EIO;
196 	return blk_errors[idx].errno;
197 }
198 EXPORT_SYMBOL_GPL(blk_status_to_errno);
199 
blk_status_to_str(blk_status_t status)200 const char *blk_status_to_str(blk_status_t status)
201 {
202 	int idx = (__force int)status;
203 
204 	if (WARN_ON_ONCE(idx >= ARRAY_SIZE(blk_errors)))
205 		return "<null>";
206 	return blk_errors[idx].name;
207 }
208 
209 /**
210  * blk_sync_queue - cancel any pending callbacks on a queue
211  * @q: the queue
212  *
213  * Description:
214  *     The block layer may perform asynchronous callback activity
215  *     on a queue, such as calling the unplug function after a timeout.
216  *     A block device may call blk_sync_queue to ensure that any
217  *     such activity is cancelled, thus allowing it to release resources
218  *     that the callbacks might use. The caller must already have made sure
219  *     that its ->submit_bio will not re-add plugging prior to calling
220  *     this function.
221  *
222  *     This function does not cancel any asynchronous activity arising
223  *     out of elevator or throttling code. That would require elevator_exit()
224  *     and blkcg_exit_queue() to be called with queue lock initialized.
225  *
226  */
blk_sync_queue(struct request_queue * q)227 void blk_sync_queue(struct request_queue *q)
228 {
229 	del_timer_sync(&q->timeout);
230 	cancel_work_sync(&q->timeout_work);
231 }
232 EXPORT_SYMBOL(blk_sync_queue);
233 
234 /**
235  * blk_set_pm_only - increment pm_only counter
236  * @q: request queue pointer
237  */
blk_set_pm_only(struct request_queue * q)238 void blk_set_pm_only(struct request_queue *q)
239 {
240 	atomic_inc(&q->pm_only);
241 }
242 EXPORT_SYMBOL_GPL(blk_set_pm_only);
243 
blk_clear_pm_only(struct request_queue * q)244 void blk_clear_pm_only(struct request_queue *q)
245 {
246 	int pm_only;
247 
248 	pm_only = atomic_dec_return(&q->pm_only);
249 	WARN_ON_ONCE(pm_only < 0);
250 	if (pm_only == 0)
251 		wake_up_all(&q->mq_freeze_wq);
252 }
253 EXPORT_SYMBOL_GPL(blk_clear_pm_only);
254 
blk_free_queue_rcu(struct rcu_head * rcu_head)255 static void blk_free_queue_rcu(struct rcu_head *rcu_head)
256 {
257 	struct request_queue *q = container_of(rcu_head,
258 			struct request_queue, rcu_head);
259 
260 	percpu_ref_exit(&q->q_usage_counter);
261 	kmem_cache_free(blk_requestq_cachep, q);
262 }
263 
blk_free_queue(struct request_queue * q)264 static void blk_free_queue(struct request_queue *q)
265 {
266 	if (q->poll_stat)
267 		blk_stat_remove_callback(q, q->poll_cb);
268 	blk_stat_free_callback(q->poll_cb);
269 
270 	blk_free_queue_stats(q->stats);
271 	kfree(q->poll_stat);
272 
273 	if (queue_is_mq(q))
274 		blk_mq_release(q);
275 
276 	ida_free(&blk_queue_ida, q->id);
277 	call_rcu(&q->rcu_head, blk_free_queue_rcu);
278 }
279 
280 /**
281  * blk_put_queue - decrement the request_queue refcount
282  * @q: the request_queue structure to decrement the refcount for
283  *
284  * Decrements the refcount of the request_queue and free it when the refcount
285  * reaches 0.
286  */
blk_put_queue(struct request_queue * q)287 void blk_put_queue(struct request_queue *q)
288 {
289 	if (refcount_dec_and_test(&q->refs))
290 		blk_free_queue(q);
291 }
292 EXPORT_SYMBOL(blk_put_queue);
293 
blk_queue_start_drain(struct request_queue * q)294 void blk_queue_start_drain(struct request_queue *q)
295 {
296 	/*
297 	 * When queue DYING flag is set, we need to block new req
298 	 * entering queue, so we call blk_freeze_queue_start() to
299 	 * prevent I/O from crossing blk_queue_enter().
300 	 */
301 	blk_freeze_queue_start(q);
302 	if (queue_is_mq(q))
303 		blk_mq_wake_waiters(q);
304 	/* Make blk_queue_enter() reexamine the DYING flag. */
305 	wake_up_all(&q->mq_freeze_wq);
306 }
307 
308 /**
309  * blk_queue_enter() - try to increase q->q_usage_counter
310  * @q: request queue pointer
311  * @flags: BLK_MQ_REQ_NOWAIT and/or BLK_MQ_REQ_PM
312  */
blk_queue_enter(struct request_queue * q,blk_mq_req_flags_t flags)313 int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags)
314 {
315 	const bool pm = flags & BLK_MQ_REQ_PM;
316 
317 	while (!blk_try_enter_queue(q, pm)) {
318 		if (flags & BLK_MQ_REQ_NOWAIT)
319 			return -EAGAIN;
320 
321 		/*
322 		 * read pair of barrier in blk_freeze_queue_start(), we need to
323 		 * order reading __PERCPU_REF_DEAD flag of .q_usage_counter and
324 		 * reading .mq_freeze_depth or queue dying flag, otherwise the
325 		 * following wait may never return if the two reads are
326 		 * reordered.
327 		 */
328 		smp_rmb();
329 		wait_event(q->mq_freeze_wq,
330 			   (!q->mq_freeze_depth &&
331 			    blk_pm_resume_queue(pm, q)) ||
332 			   blk_queue_dying(q));
333 		if (blk_queue_dying(q))
334 			return -ENODEV;
335 	}
336 
337 	return 0;
338 }
339 
__bio_queue_enter(struct request_queue * q,struct bio * bio)340 int __bio_queue_enter(struct request_queue *q, struct bio *bio)
341 {
342 	while (!blk_try_enter_queue(q, false)) {
343 		struct gendisk *disk = bio->bi_bdev->bd_disk;
344 
345 		if (bio->bi_opf & REQ_NOWAIT) {
346 			if (test_bit(GD_DEAD, &disk->state))
347 				goto dead;
348 			bio_wouldblock_error(bio);
349 			return -EAGAIN;
350 		}
351 
352 		/*
353 		 * read pair of barrier in blk_freeze_queue_start(), we need to
354 		 * order reading __PERCPU_REF_DEAD flag of .q_usage_counter and
355 		 * reading .mq_freeze_depth or queue dying flag, otherwise the
356 		 * following wait may never return if the two reads are
357 		 * reordered.
358 		 */
359 		smp_rmb();
360 		wait_event(q->mq_freeze_wq,
361 			   (!q->mq_freeze_depth &&
362 			    blk_pm_resume_queue(false, q)) ||
363 			   test_bit(GD_DEAD, &disk->state));
364 		if (test_bit(GD_DEAD, &disk->state))
365 			goto dead;
366 	}
367 
368 	return 0;
369 dead:
370 	bio_io_error(bio);
371 	return -ENODEV;
372 }
373 
blk_queue_exit(struct request_queue * q)374 void blk_queue_exit(struct request_queue *q)
375 {
376 	percpu_ref_put(&q->q_usage_counter);
377 }
378 
blk_queue_usage_counter_release(struct percpu_ref * ref)379 static void blk_queue_usage_counter_release(struct percpu_ref *ref)
380 {
381 	struct request_queue *q =
382 		container_of(ref, struct request_queue, q_usage_counter);
383 
384 	wake_up_all(&q->mq_freeze_wq);
385 }
386 
blk_rq_timed_out_timer(struct timer_list * t)387 static void blk_rq_timed_out_timer(struct timer_list *t)
388 {
389 	struct request_queue *q = from_timer(q, t, timeout);
390 
391 	kblockd_schedule_work(&q->timeout_work);
392 }
393 
blk_timeout_work(struct work_struct * work)394 static void blk_timeout_work(struct work_struct *work)
395 {
396 }
397 
blk_alloc_queue(int node_id)398 struct request_queue *blk_alloc_queue(int node_id)
399 {
400 	struct request_queue *q;
401 
402 	q = kmem_cache_alloc_node(blk_requestq_cachep, GFP_KERNEL | __GFP_ZERO,
403 				  node_id);
404 	if (!q)
405 		return NULL;
406 
407 	q->last_merge = NULL;
408 
409 	q->id = ida_alloc(&blk_queue_ida, GFP_KERNEL);
410 	if (q->id < 0)
411 		goto fail_q;
412 
413 	q->stats = blk_alloc_queue_stats();
414 	if (!q->stats)
415 		goto fail_id;
416 
417 	q->node = node_id;
418 
419 	atomic_set(&q->nr_active_requests_shared_tags, 0);
420 
421 	timer_setup(&q->timeout, blk_rq_timed_out_timer, 0);
422 	INIT_WORK(&q->timeout_work, blk_timeout_work);
423 	INIT_LIST_HEAD(&q->icq_list);
424 
425 	refcount_set(&q->refs, 1);
426 	mutex_init(&q->debugfs_mutex);
427 	mutex_init(&q->sysfs_lock);
428 	mutex_init(&q->sysfs_dir_lock);
429 	spin_lock_init(&q->queue_lock);
430 
431 	init_waitqueue_head(&q->mq_freeze_wq);
432 	mutex_init(&q->mq_freeze_lock);
433 
434 	/*
435 	 * Init percpu_ref in atomic mode so that it's faster to shutdown.
436 	 * See blk_register_queue() for details.
437 	 */
438 	if (percpu_ref_init(&q->q_usage_counter,
439 				blk_queue_usage_counter_release,
440 				PERCPU_REF_INIT_ATOMIC, GFP_KERNEL))
441 		goto fail_stats;
442 
443 	blk_set_default_limits(&q->limits);
444 	q->nr_requests = BLKDEV_DEFAULT_RQ;
445 
446 	return q;
447 
448 fail_stats:
449 	blk_free_queue_stats(q->stats);
450 fail_id:
451 	ida_free(&blk_queue_ida, q->id);
452 fail_q:
453 	kmem_cache_free(blk_requestq_cachep, q);
454 	return NULL;
455 }
456 
457 /**
458  * blk_get_queue - increment the request_queue refcount
459  * @q: the request_queue structure to increment the refcount for
460  *
461  * Increment the refcount of the request_queue kobject.
462  *
463  * Context: Any context.
464  */
blk_get_queue(struct request_queue * q)465 bool blk_get_queue(struct request_queue *q)
466 {
467 	if (unlikely(blk_queue_dying(q)))
468 		return false;
469 	refcount_inc(&q->refs);
470 	return true;
471 }
472 EXPORT_SYMBOL(blk_get_queue);
473 
474 #ifdef CONFIG_FAIL_MAKE_REQUEST
475 
476 static DECLARE_FAULT_ATTR(fail_make_request);
477 
setup_fail_make_request(char * str)478 static int __init setup_fail_make_request(char *str)
479 {
480 	return setup_fault_attr(&fail_make_request, str);
481 }
482 __setup("fail_make_request=", setup_fail_make_request);
483 
should_fail_request(struct block_device * part,unsigned int bytes)484 bool should_fail_request(struct block_device *part, unsigned int bytes)
485 {
486 	return part->bd_make_it_fail && should_fail(&fail_make_request, bytes);
487 }
488 
fail_make_request_debugfs(void)489 static int __init fail_make_request_debugfs(void)
490 {
491 	struct dentry *dir = fault_create_debugfs_attr("fail_make_request",
492 						NULL, &fail_make_request);
493 
494 	return PTR_ERR_OR_ZERO(dir);
495 }
496 
497 late_initcall(fail_make_request_debugfs);
498 #endif /* CONFIG_FAIL_MAKE_REQUEST */
499 
bio_check_ro(struct bio * bio)500 static inline void bio_check_ro(struct bio *bio)
501 {
502 	if (op_is_write(bio_op(bio)) && bdev_read_only(bio->bi_bdev)) {
503 		if (op_is_flush(bio->bi_opf) && !bio_sectors(bio))
504 			return;
505 		pr_warn("Trying to write to read-only block-device %pg\n",
506 			bio->bi_bdev);
507 		/* Older lvm-tools actually trigger this */
508 	}
509 }
510 
should_fail_bio(struct bio * bio)511 static noinline int should_fail_bio(struct bio *bio)
512 {
513 	if (should_fail_request(bdev_whole(bio->bi_bdev), bio->bi_iter.bi_size))
514 		return -EIO;
515 	return 0;
516 }
517 ALLOW_ERROR_INJECTION(should_fail_bio, ERRNO);
518 
519 /*
520  * Check whether this bio extends beyond the end of the device or partition.
521  * This may well happen - the kernel calls bread() without checking the size of
522  * the device, e.g., when mounting a file system.
523  */
bio_check_eod(struct bio * bio)524 static inline int bio_check_eod(struct bio *bio)
525 {
526 	sector_t maxsector = bdev_nr_sectors(bio->bi_bdev);
527 	unsigned int nr_sectors = bio_sectors(bio);
528 
529 	if (nr_sectors && maxsector &&
530 	    (nr_sectors > maxsector ||
531 	     bio->bi_iter.bi_sector > maxsector - nr_sectors)) {
532 		pr_info_ratelimited("%s: attempt to access beyond end of device\n"
533 				    "%pg: rw=%d, sector=%llu, nr_sectors = %u limit=%llu\n",
534 				    current->comm, bio->bi_bdev, bio->bi_opf,
535 				    bio->bi_iter.bi_sector, nr_sectors, maxsector);
536 		return -EIO;
537 	}
538 	return 0;
539 }
540 
541 /*
542  * Remap block n of partition p to block n+start(p) of the disk.
543  */
blk_partition_remap(struct bio * bio)544 static int blk_partition_remap(struct bio *bio)
545 {
546 	struct block_device *p = bio->bi_bdev;
547 
548 	if (unlikely(should_fail_request(p, bio->bi_iter.bi_size)))
549 		return -EIO;
550 	if (bio_sectors(bio)) {
551 		bio->bi_iter.bi_sector += p->bd_start_sect;
552 		trace_block_bio_remap(bio, p->bd_dev,
553 				      bio->bi_iter.bi_sector -
554 				      p->bd_start_sect);
555 	}
556 	bio_set_flag(bio, BIO_REMAPPED);
557 	return 0;
558 }
559 
560 /*
561  * Check write append to a zoned block device.
562  */
blk_check_zone_append(struct request_queue * q,struct bio * bio)563 static inline blk_status_t blk_check_zone_append(struct request_queue *q,
564 						 struct bio *bio)
565 {
566 	int nr_sectors = bio_sectors(bio);
567 
568 	/* Only applicable to zoned block devices */
569 	if (!bdev_is_zoned(bio->bi_bdev))
570 		return BLK_STS_NOTSUPP;
571 
572 	/* The bio sector must point to the start of a sequential zone */
573 	if (!bdev_is_zone_start(bio->bi_bdev, bio->bi_iter.bi_sector) ||
574 	    !bio_zone_is_seq(bio))
575 		return BLK_STS_IOERR;
576 
577 	/*
578 	 * Not allowed to cross zone boundaries. Otherwise, the BIO will be
579 	 * split and could result in non-contiguous sectors being written in
580 	 * different zones.
581 	 */
582 	if (nr_sectors > q->limits.chunk_sectors)
583 		return BLK_STS_IOERR;
584 
585 	/* Make sure the BIO is small enough and will not get split */
586 	if (nr_sectors > q->limits.max_zone_append_sectors)
587 		return BLK_STS_IOERR;
588 
589 	bio->bi_opf |= REQ_NOMERGE;
590 
591 	return BLK_STS_OK;
592 }
593 
__submit_bio(struct bio * bio)594 static void __submit_bio(struct bio *bio)
595 {
596 	struct gendisk *disk = bio->bi_bdev->bd_disk;
597 
598 	if (unlikely(!blk_crypto_bio_prep(&bio)))
599 		return;
600 
601 	if (!disk->fops->submit_bio) {
602 		blk_mq_submit_bio(bio);
603 	} else if (likely(bio_queue_enter(bio) == 0)) {
604 		disk->fops->submit_bio(bio);
605 		blk_queue_exit(disk->queue);
606 	}
607 }
608 
609 /*
610  * The loop in this function may be a bit non-obvious, and so deserves some
611  * explanation:
612  *
613  *  - Before entering the loop, bio->bi_next is NULL (as all callers ensure
614  *    that), so we have a list with a single bio.
615  *  - We pretend that we have just taken it off a longer list, so we assign
616  *    bio_list to a pointer to the bio_list_on_stack, thus initialising the
617  *    bio_list of new bios to be added.  ->submit_bio() may indeed add some more
618  *    bios through a recursive call to submit_bio_noacct.  If it did, we find a
619  *    non-NULL value in bio_list and re-enter the loop from the top.
620  *  - In this case we really did just take the bio of the top of the list (no
621  *    pretending) and so remove it from bio_list, and call into ->submit_bio()
622  *    again.
623  *
624  * bio_list_on_stack[0] contains bios submitted by the current ->submit_bio.
625  * bio_list_on_stack[1] contains bios that were submitted before the current
626  *	->submit_bio, but that haven't been processed yet.
627  */
__submit_bio_noacct(struct bio * bio)628 static void __submit_bio_noacct(struct bio *bio)
629 {
630 	struct bio_list bio_list_on_stack[2];
631 
632 	BUG_ON(bio->bi_next);
633 
634 	bio_list_init(&bio_list_on_stack[0]);
635 	current->bio_list = bio_list_on_stack;
636 
637 	do {
638 		struct request_queue *q = bdev_get_queue(bio->bi_bdev);
639 		struct bio_list lower, same;
640 
641 		/*
642 		 * Create a fresh bio_list for all subordinate requests.
643 		 */
644 		bio_list_on_stack[1] = bio_list_on_stack[0];
645 		bio_list_init(&bio_list_on_stack[0]);
646 
647 		__submit_bio(bio);
648 
649 		/*
650 		 * Sort new bios into those for a lower level and those for the
651 		 * same level.
652 		 */
653 		bio_list_init(&lower);
654 		bio_list_init(&same);
655 		while ((bio = bio_list_pop(&bio_list_on_stack[0])) != NULL)
656 			if (q == bdev_get_queue(bio->bi_bdev))
657 				bio_list_add(&same, bio);
658 			else
659 				bio_list_add(&lower, bio);
660 
661 		/*
662 		 * Now assemble so we handle the lowest level first.
663 		 */
664 		bio_list_merge(&bio_list_on_stack[0], &lower);
665 		bio_list_merge(&bio_list_on_stack[0], &same);
666 		bio_list_merge(&bio_list_on_stack[0], &bio_list_on_stack[1]);
667 	} while ((bio = bio_list_pop(&bio_list_on_stack[0])));
668 
669 	current->bio_list = NULL;
670 }
671 
__submit_bio_noacct_mq(struct bio * bio)672 static void __submit_bio_noacct_mq(struct bio *bio)
673 {
674 	struct bio_list bio_list[2] = { };
675 
676 	current->bio_list = bio_list;
677 
678 	do {
679 		__submit_bio(bio);
680 	} while ((bio = bio_list_pop(&bio_list[0])));
681 
682 	current->bio_list = NULL;
683 }
684 
submit_bio_noacct_nocheck(struct bio * bio)685 void submit_bio_noacct_nocheck(struct bio *bio)
686 {
687 	blk_cgroup_bio_start(bio);
688 	blkcg_bio_issue_init(bio);
689 
690 	if (!bio_flagged(bio, BIO_TRACE_COMPLETION)) {
691 		trace_block_bio_queue(bio);
692 		/*
693 		 * Now that enqueuing has been traced, we need to trace
694 		 * completion as well.
695 		 */
696 		bio_set_flag(bio, BIO_TRACE_COMPLETION);
697 	}
698 
699 	/*
700 	 * We only want one ->submit_bio to be active at a time, else stack
701 	 * usage with stacked devices could be a problem.  Use current->bio_list
702 	 * to collect a list of requests submited by a ->submit_bio method while
703 	 * it is active, and then process them after it returned.
704 	 */
705 	if (current->bio_list)
706 		bio_list_add(&current->bio_list[0], bio);
707 	else if (!bio->bi_bdev->bd_disk->fops->submit_bio)
708 		__submit_bio_noacct_mq(bio);
709 	else
710 		__submit_bio_noacct(bio);
711 }
712 
713 /**
714  * submit_bio_noacct - re-submit a bio to the block device layer for I/O
715  * @bio:  The bio describing the location in memory and on the device.
716  *
717  * This is a version of submit_bio() that shall only be used for I/O that is
718  * resubmitted to lower level drivers by stacking block drivers.  All file
719  * systems and other upper level users of the block layer should use
720  * submit_bio() instead.
721  */
submit_bio_noacct(struct bio * bio)722 void submit_bio_noacct(struct bio *bio)
723 {
724 	struct block_device *bdev = bio->bi_bdev;
725 	struct request_queue *q = bdev_get_queue(bdev);
726 	blk_status_t status = BLK_STS_IOERR;
727 	struct blk_plug *plug;
728 
729 	might_sleep();
730 
731 	plug = blk_mq_plug(bio);
732 	if (plug && plug->nowait)
733 		bio->bi_opf |= REQ_NOWAIT;
734 
735 	/*
736 	 * For a REQ_NOWAIT based request, return -EOPNOTSUPP
737 	 * if queue does not support NOWAIT.
738 	 */
739 	if ((bio->bi_opf & REQ_NOWAIT) && !bdev_nowait(bdev))
740 		goto not_supported;
741 
742 	if (should_fail_bio(bio))
743 		goto end_io;
744 	bio_check_ro(bio);
745 	if (!bio_flagged(bio, BIO_REMAPPED)) {
746 		if (unlikely(bio_check_eod(bio)))
747 			goto end_io;
748 		if (bdev->bd_partno && unlikely(blk_partition_remap(bio)))
749 			goto end_io;
750 	}
751 
752 	/*
753 	 * Filter flush bio's early so that bio based drivers without flush
754 	 * support don't have to worry about them.
755 	 */
756 	if (op_is_flush(bio->bi_opf)) {
757 		if (WARN_ON_ONCE(bio_op(bio) != REQ_OP_WRITE &&
758 				 bio_op(bio) != REQ_OP_ZONE_APPEND))
759 			goto end_io;
760 		if (!test_bit(QUEUE_FLAG_WC, &q->queue_flags)) {
761 			bio->bi_opf &= ~(REQ_PREFLUSH | REQ_FUA);
762 			if (!bio_sectors(bio)) {
763 				status = BLK_STS_OK;
764 				goto end_io;
765 			}
766 		}
767 	}
768 
769 	if (!test_bit(QUEUE_FLAG_POLL, &q->queue_flags))
770 		bio_clear_polled(bio);
771 
772 	switch (bio_op(bio)) {
773 	case REQ_OP_DISCARD:
774 		if (!bdev_max_discard_sectors(bdev))
775 			goto not_supported;
776 		break;
777 	case REQ_OP_SECURE_ERASE:
778 		if (!bdev_max_secure_erase_sectors(bdev))
779 			goto not_supported;
780 		break;
781 	case REQ_OP_ZONE_APPEND:
782 		status = blk_check_zone_append(q, bio);
783 		if (status != BLK_STS_OK)
784 			goto end_io;
785 		break;
786 	case REQ_OP_ZONE_RESET:
787 	case REQ_OP_ZONE_OPEN:
788 	case REQ_OP_ZONE_CLOSE:
789 	case REQ_OP_ZONE_FINISH:
790 		if (!bdev_is_zoned(bio->bi_bdev))
791 			goto not_supported;
792 		break;
793 	case REQ_OP_ZONE_RESET_ALL:
794 		if (!bdev_is_zoned(bio->bi_bdev) || !blk_queue_zone_resetall(q))
795 			goto not_supported;
796 		break;
797 	case REQ_OP_WRITE_ZEROES:
798 		if (!q->limits.max_write_zeroes_sectors)
799 			goto not_supported;
800 		break;
801 	default:
802 		break;
803 	}
804 
805 	if (blk_throtl_bio(bio))
806 		return;
807 	submit_bio_noacct_nocheck(bio);
808 	return;
809 
810 not_supported:
811 	status = BLK_STS_NOTSUPP;
812 end_io:
813 	bio->bi_status = status;
814 	bio_endio(bio);
815 }
816 EXPORT_SYMBOL(submit_bio_noacct);
817 
818 /**
819  * submit_bio - submit a bio to the block device layer for I/O
820  * @bio: The &struct bio which describes the I/O
821  *
822  * submit_bio() is used to submit I/O requests to block devices.  It is passed a
823  * fully set up &struct bio that describes the I/O that needs to be done.  The
824  * bio will be send to the device described by the bi_bdev field.
825  *
826  * The success/failure status of the request, along with notification of
827  * completion, is delivered asynchronously through the ->bi_end_io() callback
828  * in @bio.  The bio must NOT be touched by the caller until ->bi_end_io() has
829  * been called.
830  */
submit_bio(struct bio * bio)831 void submit_bio(struct bio *bio)
832 {
833 	if (blkcg_punt_bio_submit(bio))
834 		return;
835 
836 	if (bio_op(bio) == REQ_OP_READ) {
837 		task_io_account_read(bio->bi_iter.bi_size);
838 		count_vm_events(PGPGIN, bio_sectors(bio));
839 	} else if (bio_op(bio) == REQ_OP_WRITE) {
840 		count_vm_events(PGPGOUT, bio_sectors(bio));
841 	}
842 
843 	submit_bio_noacct(bio);
844 }
845 EXPORT_SYMBOL(submit_bio);
846 
847 /**
848  * bio_poll - poll for BIO completions
849  * @bio: bio to poll for
850  * @iob: batches of IO
851  * @flags: BLK_POLL_* flags that control the behavior
852  *
853  * Poll for completions on queue associated with the bio. Returns number of
854  * completed entries found.
855  *
856  * Note: the caller must either be the context that submitted @bio, or
857  * be in a RCU critical section to prevent freeing of @bio.
858  */
bio_poll(struct bio * bio,struct io_comp_batch * iob,unsigned int flags)859 int bio_poll(struct bio *bio, struct io_comp_batch *iob, unsigned int flags)
860 {
861 	blk_qc_t cookie = READ_ONCE(bio->bi_cookie);
862 	struct block_device *bdev;
863 	struct request_queue *q;
864 	int ret = 0;
865 
866 	bdev = READ_ONCE(bio->bi_bdev);
867 	if (!bdev)
868 		return 0;
869 
870 	q = bdev_get_queue(bdev);
871 	if (cookie == BLK_QC_T_NONE ||
872 	    !test_bit(QUEUE_FLAG_POLL, &q->queue_flags))
873 		return 0;
874 
875 	/*
876 	 * As the requests that require a zone lock are not plugged in the
877 	 * first place, directly accessing the plug instead of using
878 	 * blk_mq_plug() should not have any consequences during flushing for
879 	 * zoned devices.
880 	 */
881 	blk_flush_plug(current->plug, false);
882 
883 	/*
884 	 * We need to be able to enter a frozen queue, similar to how
885 	 * timeouts also need to do that. If that is blocked, then we can
886 	 * have pending IO when a queue freeze is started, and then the
887 	 * wait for the freeze to finish will wait for polled requests to
888 	 * timeout as the poller is preventer from entering the queue and
889 	 * completing them. As long as we prevent new IO from being queued,
890 	 * that should be all that matters.
891 	 */
892 	if (!percpu_ref_tryget(&q->q_usage_counter))
893 		return 0;
894 	if (queue_is_mq(q)) {
895 		ret = blk_mq_poll(q, cookie, iob, flags);
896 	} else {
897 		struct gendisk *disk = q->disk;
898 
899 		if (disk && disk->fops->poll_bio)
900 			ret = disk->fops->poll_bio(bio, iob, flags);
901 	}
902 	blk_queue_exit(q);
903 	return ret;
904 }
905 EXPORT_SYMBOL_GPL(bio_poll);
906 
907 /*
908  * Helper to implement file_operations.iopoll.  Requires the bio to be stored
909  * in iocb->private, and cleared before freeing the bio.
910  */
iocb_bio_iopoll(struct kiocb * kiocb,struct io_comp_batch * iob,unsigned int flags)911 int iocb_bio_iopoll(struct kiocb *kiocb, struct io_comp_batch *iob,
912 		    unsigned int flags)
913 {
914 	struct bio *bio;
915 	int ret = 0;
916 
917 	/*
918 	 * Note: the bio cache only uses SLAB_TYPESAFE_BY_RCU, so bio can
919 	 * point to a freshly allocated bio at this point.  If that happens
920 	 * we have a few cases to consider:
921 	 *
922 	 *  1) the bio is beeing initialized and bi_bdev is NULL.  We can just
923 	 *     simply nothing in this case
924 	 *  2) the bio points to a not poll enabled device.  bio_poll will catch
925 	 *     this and return 0
926 	 *  3) the bio points to a poll capable device, including but not
927 	 *     limited to the one that the original bio pointed to.  In this
928 	 *     case we will call into the actual poll method and poll for I/O,
929 	 *     even if we don't need to, but it won't cause harm either.
930 	 *
931 	 * For cases 2) and 3) above the RCU grace period ensures that bi_bdev
932 	 * is still allocated. Because partitions hold a reference to the whole
933 	 * device bdev and thus disk, the disk is also still valid.  Grabbing
934 	 * a reference to the queue in bio_poll() ensures the hctxs and requests
935 	 * are still valid as well.
936 	 */
937 	rcu_read_lock();
938 	bio = READ_ONCE(kiocb->private);
939 	if (bio)
940 		ret = bio_poll(bio, iob, flags);
941 	rcu_read_unlock();
942 
943 	return ret;
944 }
945 EXPORT_SYMBOL_GPL(iocb_bio_iopoll);
946 
update_io_ticks(struct block_device * part,unsigned long now,bool end)947 void update_io_ticks(struct block_device *part, unsigned long now, bool end)
948 {
949 	unsigned long stamp;
950 again:
951 	stamp = READ_ONCE(part->bd_stamp);
952 	if (unlikely(time_after(now, stamp))) {
953 		if (likely(try_cmpxchg(&part->bd_stamp, &stamp, now)))
954 			__part_stat_add(part, io_ticks, end ? now - stamp : 1);
955 	}
956 	if (part->bd_partno) {
957 		part = bdev_whole(part);
958 		goto again;
959 	}
960 }
961 
bdev_start_io_acct(struct block_device * bdev,unsigned int sectors,enum req_op op,unsigned long start_time)962 unsigned long bdev_start_io_acct(struct block_device *bdev,
963 				 unsigned int sectors, enum req_op op,
964 				 unsigned long start_time)
965 {
966 	const int sgrp = op_stat_group(op);
967 
968 	part_stat_lock();
969 	update_io_ticks(bdev, start_time, false);
970 	part_stat_inc(bdev, ios[sgrp]);
971 	part_stat_add(bdev, sectors[sgrp], sectors);
972 	part_stat_local_inc(bdev, in_flight[op_is_write(op)]);
973 	part_stat_unlock();
974 
975 	return start_time;
976 }
977 EXPORT_SYMBOL(bdev_start_io_acct);
978 
979 /**
980  * bio_start_io_acct - start I/O accounting for bio based drivers
981  * @bio:	bio to start account for
982  *
983  * Returns the start time that should be passed back to bio_end_io_acct().
984  */
bio_start_io_acct(struct bio * bio)985 unsigned long bio_start_io_acct(struct bio *bio)
986 {
987 	return bdev_start_io_acct(bio->bi_bdev, bio_sectors(bio),
988 				  bio_op(bio), jiffies);
989 }
990 EXPORT_SYMBOL_GPL(bio_start_io_acct);
991 
bdev_end_io_acct(struct block_device * bdev,enum req_op op,unsigned long start_time)992 void bdev_end_io_acct(struct block_device *bdev, enum req_op op,
993 		      unsigned long start_time)
994 {
995 	const int sgrp = op_stat_group(op);
996 	unsigned long now = READ_ONCE(jiffies);
997 	unsigned long duration = now - start_time;
998 
999 	part_stat_lock();
1000 	update_io_ticks(bdev, now, true);
1001 	part_stat_add(bdev, nsecs[sgrp], jiffies_to_nsecs(duration));
1002 	part_stat_local_dec(bdev, in_flight[op_is_write(op)]);
1003 	part_stat_unlock();
1004 }
1005 EXPORT_SYMBOL(bdev_end_io_acct);
1006 
bio_end_io_acct_remapped(struct bio * bio,unsigned long start_time,struct block_device * orig_bdev)1007 void bio_end_io_acct_remapped(struct bio *bio, unsigned long start_time,
1008 			      struct block_device *orig_bdev)
1009 {
1010 	bdev_end_io_acct(orig_bdev, bio_op(bio), start_time);
1011 }
1012 EXPORT_SYMBOL_GPL(bio_end_io_acct_remapped);
1013 
1014 /**
1015  * blk_lld_busy - Check if underlying low-level drivers of a device are busy
1016  * @q : the queue of the device being checked
1017  *
1018  * Description:
1019  *    Check if underlying low-level drivers of a device are busy.
1020  *    If the drivers want to export their busy state, they must set own
1021  *    exporting function using blk_queue_lld_busy() first.
1022  *
1023  *    Basically, this function is used only by request stacking drivers
1024  *    to stop dispatching requests to underlying devices when underlying
1025  *    devices are busy.  This behavior helps more I/O merging on the queue
1026  *    of the request stacking driver and prevents I/O throughput regression
1027  *    on burst I/O load.
1028  *
1029  * Return:
1030  *    0 - Not busy (The request stacking driver should dispatch request)
1031  *    1 - Busy (The request stacking driver should stop dispatching request)
1032  */
blk_lld_busy(struct request_queue * q)1033 int blk_lld_busy(struct request_queue *q)
1034 {
1035 	if (queue_is_mq(q) && q->mq_ops->busy)
1036 		return q->mq_ops->busy(q);
1037 
1038 	return 0;
1039 }
1040 EXPORT_SYMBOL_GPL(blk_lld_busy);
1041 
kblockd_schedule_work(struct work_struct * work)1042 int kblockd_schedule_work(struct work_struct *work)
1043 {
1044 	return queue_work(kblockd_workqueue, work);
1045 }
1046 EXPORT_SYMBOL(kblockd_schedule_work);
1047 
kblockd_mod_delayed_work_on(int cpu,struct delayed_work * dwork,unsigned long delay)1048 int kblockd_mod_delayed_work_on(int cpu, struct delayed_work *dwork,
1049 				unsigned long delay)
1050 {
1051 	return mod_delayed_work_on(cpu, kblockd_workqueue, dwork, delay);
1052 }
1053 EXPORT_SYMBOL(kblockd_mod_delayed_work_on);
1054 
blk_start_plug_nr_ios(struct blk_plug * plug,unsigned short nr_ios)1055 void blk_start_plug_nr_ios(struct blk_plug *plug, unsigned short nr_ios)
1056 {
1057 	struct task_struct *tsk = current;
1058 
1059 	/*
1060 	 * If this is a nested plug, don't actually assign it.
1061 	 */
1062 	if (tsk->plug)
1063 		return;
1064 
1065 	plug->mq_list = NULL;
1066 	plug->cached_rq = NULL;
1067 	plug->nr_ios = min_t(unsigned short, nr_ios, BLK_MAX_REQUEST_COUNT);
1068 	plug->rq_count = 0;
1069 	plug->multiple_queues = false;
1070 	plug->has_elevator = false;
1071 	plug->nowait = false;
1072 	INIT_LIST_HEAD(&plug->cb_list);
1073 
1074 	/*
1075 	 * Store ordering should not be needed here, since a potential
1076 	 * preempt will imply a full memory barrier
1077 	 */
1078 	tsk->plug = plug;
1079 }
1080 
1081 /**
1082  * blk_start_plug - initialize blk_plug and track it inside the task_struct
1083  * @plug:	The &struct blk_plug that needs to be initialized
1084  *
1085  * Description:
1086  *   blk_start_plug() indicates to the block layer an intent by the caller
1087  *   to submit multiple I/O requests in a batch.  The block layer may use
1088  *   this hint to defer submitting I/Os from the caller until blk_finish_plug()
1089  *   is called.  However, the block layer may choose to submit requests
1090  *   before a call to blk_finish_plug() if the number of queued I/Os
1091  *   exceeds %BLK_MAX_REQUEST_COUNT, or if the size of the I/O is larger than
1092  *   %BLK_PLUG_FLUSH_SIZE.  The queued I/Os may also be submitted early if
1093  *   the task schedules (see below).
1094  *
1095  *   Tracking blk_plug inside the task_struct will help with auto-flushing the
1096  *   pending I/O should the task end up blocking between blk_start_plug() and
1097  *   blk_finish_plug(). This is important from a performance perspective, but
1098  *   also ensures that we don't deadlock. For instance, if the task is blocking
1099  *   for a memory allocation, memory reclaim could end up wanting to free a
1100  *   page belonging to that request that is currently residing in our private
1101  *   plug. By flushing the pending I/O when the process goes to sleep, we avoid
1102  *   this kind of deadlock.
1103  */
blk_start_plug(struct blk_plug * plug)1104 void blk_start_plug(struct blk_plug *plug)
1105 {
1106 	blk_start_plug_nr_ios(plug, 1);
1107 }
1108 EXPORT_SYMBOL(blk_start_plug);
1109 
flush_plug_callbacks(struct blk_plug * plug,bool from_schedule)1110 static void flush_plug_callbacks(struct blk_plug *plug, bool from_schedule)
1111 {
1112 	LIST_HEAD(callbacks);
1113 
1114 	while (!list_empty(&plug->cb_list)) {
1115 		list_splice_init(&plug->cb_list, &callbacks);
1116 
1117 		while (!list_empty(&callbacks)) {
1118 			struct blk_plug_cb *cb = list_first_entry(&callbacks,
1119 							  struct blk_plug_cb,
1120 							  list);
1121 			list_del(&cb->list);
1122 			cb->callback(cb, from_schedule);
1123 		}
1124 	}
1125 }
1126 
blk_check_plugged(blk_plug_cb_fn unplug,void * data,int size)1127 struct blk_plug_cb *blk_check_plugged(blk_plug_cb_fn unplug, void *data,
1128 				      int size)
1129 {
1130 	struct blk_plug *plug = current->plug;
1131 	struct blk_plug_cb *cb;
1132 
1133 	if (!plug)
1134 		return NULL;
1135 
1136 	list_for_each_entry(cb, &plug->cb_list, list)
1137 		if (cb->callback == unplug && cb->data == data)
1138 			return cb;
1139 
1140 	/* Not currently on the callback list */
1141 	BUG_ON(size < sizeof(*cb));
1142 	cb = kzalloc(size, GFP_ATOMIC);
1143 	if (cb) {
1144 		cb->data = data;
1145 		cb->callback = unplug;
1146 		list_add(&cb->list, &plug->cb_list);
1147 	}
1148 	return cb;
1149 }
1150 EXPORT_SYMBOL(blk_check_plugged);
1151 
__blk_flush_plug(struct blk_plug * plug,bool from_schedule)1152 void __blk_flush_plug(struct blk_plug *plug, bool from_schedule)
1153 {
1154 	if (!list_empty(&plug->cb_list))
1155 		flush_plug_callbacks(plug, from_schedule);
1156 	if (!rq_list_empty(plug->mq_list))
1157 		blk_mq_flush_plug_list(plug, from_schedule);
1158 	/*
1159 	 * Unconditionally flush out cached requests, even if the unplug
1160 	 * event came from schedule. Since we know hold references to the
1161 	 * queue for cached requests, we don't want a blocked task holding
1162 	 * up a queue freeze/quiesce event.
1163 	 */
1164 	if (unlikely(!rq_list_empty(plug->cached_rq)))
1165 		blk_mq_free_plug_rqs(plug);
1166 }
1167 
1168 /**
1169  * blk_finish_plug - mark the end of a batch of submitted I/O
1170  * @plug:	The &struct blk_plug passed to blk_start_plug()
1171  *
1172  * Description:
1173  * Indicate that a batch of I/O submissions is complete.  This function
1174  * must be paired with an initial call to blk_start_plug().  The intent
1175  * is to allow the block layer to optimize I/O submission.  See the
1176  * documentation for blk_start_plug() for more information.
1177  */
blk_finish_plug(struct blk_plug * plug)1178 void blk_finish_plug(struct blk_plug *plug)
1179 {
1180 	if (plug == current->plug) {
1181 		__blk_flush_plug(plug, false);
1182 		current->plug = NULL;
1183 	}
1184 }
1185 EXPORT_SYMBOL(blk_finish_plug);
1186 
blk_io_schedule(void)1187 void blk_io_schedule(void)
1188 {
1189 	/* Prevent hang_check timer from firing at us during very long I/O */
1190 	unsigned long timeout = sysctl_hung_task_timeout_secs * HZ / 2;
1191 
1192 	if (timeout)
1193 		io_schedule_timeout(timeout);
1194 	else
1195 		io_schedule();
1196 }
1197 EXPORT_SYMBOL_GPL(blk_io_schedule);
1198 
blk_dev_init(void)1199 int __init blk_dev_init(void)
1200 {
1201 	BUILD_BUG_ON((__force u32)REQ_OP_LAST >= (1 << REQ_OP_BITS));
1202 	BUILD_BUG_ON(REQ_OP_BITS + REQ_FLAG_BITS > 8 *
1203 			sizeof_field(struct request, cmd_flags));
1204 	BUILD_BUG_ON(REQ_OP_BITS + REQ_FLAG_BITS > 8 *
1205 			sizeof_field(struct bio, bi_opf));
1206 
1207 	/* used for unplugging and affects IO latency/throughput - HIGHPRI */
1208 	kblockd_workqueue = alloc_workqueue("kblockd",
1209 					    WQ_MEM_RECLAIM | WQ_HIGHPRI, 0);
1210 	if (!kblockd_workqueue)
1211 		panic("Failed to create kblockd\n");
1212 
1213 	blk_requestq_cachep = kmem_cache_create("request_queue",
1214 			sizeof(struct request_queue), 0, SLAB_PANIC, NULL);
1215 
1216 	blk_debugfs_root = debugfs_create_dir("block", NULL);
1217 
1218 	return 0;
1219 }
1220