blk-core.c 46.7 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0
Linus Torvalds's avatar
Linus Torvalds committed
2 3 4 5 6
/*
 * Copyright (C) 1991, 1992 Linus Torvalds
 * Copyright (C) 1994,      Karl Keyte: Added support for disk statistics
 * Elevator latency, (C) 2000  Andrea Arcangeli <andrea@suse.de> SuSE
 * Queue request tables / lock, selectable elevator, Jens Axboe <axboe@suse.de>
7 8
 * kernel-doc documentation started by NeilBrown <neilb@cse.unsw.edu.au>
 *	-  July2000
Linus Torvalds's avatar
Linus Torvalds committed
9 10 11 12 13 14 15 16 17 18 19
 * bio rewrite, highmem i/o, etc, Jens Axboe <axboe@suse.de> - may 2001
 */

/*
 * This handles all read/write requests to block devices
 */
#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/backing-dev.h>
#include <linux/bio.h>
#include <linux/blkdev.h>
20
#include <linux/blk-mq.h>
Linus Torvalds's avatar
Linus Torvalds committed
21 22 23 24 25 26 27 28 29
#include <linux/highmem.h>
#include <linux/mm.h>
#include <linux/kernel_stat.h>
#include <linux/string.h>
#include <linux/init.h>
#include <linux/completion.h>
#include <linux/slab.h>
#include <linux/swap.h>
#include <linux/writeback.h>
30
#include <linux/task_io_accounting_ops.h>
31
#include <linux/fault-inject.h>
32
#include <linux/list_sort.h>
Tejun Heo's avatar
Tejun Heo committed
33
#include <linux/delay.h>
34
#include <linux/ratelimit.h>
Lin Ming's avatar
Lin Ming committed
35
#include <linux/pm_runtime.h>
36
#include <linux/blk-cgroup.h>
37
#include <linux/debugfs.h>
38
#include <linux/bpf.h>
39 40 41

#define CREATE_TRACE_POINTS
#include <trace/events/block.h>
Linus Torvalds's avatar
Linus Torvalds committed
42

43
#include "blk.h"
44
#include "blk-mq.h"
45
#include "blk-mq-sched.h"
46
#include "blk-pm.h"
47
#include "blk-rq-qos.h"
48

49 50 51 52
#ifdef CONFIG_DEBUG_FS
struct dentry *blk_debugfs_root;
#endif

53
EXPORT_TRACEPOINT_SYMBOL_GPL(block_bio_remap);
54
EXPORT_TRACEPOINT_SYMBOL_GPL(block_rq_remap);
55
EXPORT_TRACEPOINT_SYMBOL_GPL(block_bio_complete);
Keith Busch's avatar
Keith Busch committed
56
EXPORT_TRACEPOINT_SYMBOL_GPL(block_split);
57
EXPORT_TRACEPOINT_SYMBOL_GPL(block_unplug);
58

59 60
DEFINE_IDA(blk_queue_ida);

Linus Torvalds's avatar
Linus Torvalds committed
61 62 63
/*
 * For queue allocation
 */
64
struct kmem_cache *blk_requestq_cachep;
Linus Torvalds's avatar
Linus Torvalds committed
65 66 67 68

/*
 * Controlling structure to kblockd
 */
69
static struct workqueue_struct *kblockd_workqueue;
Linus Torvalds's avatar
Linus Torvalds committed
70

71 72 73 74 75 76 77
/**
 * blk_queue_flag_set - atomically set a queue flag
 * @flag: flag to be set
 * @q: request queue
 */
void blk_queue_flag_set(unsigned int flag, struct request_queue *q)
{
78
	set_bit(flag, &q->queue_flags);
79 80 81 82 83 84 85 86 87 88
}
EXPORT_SYMBOL(blk_queue_flag_set);

/**
 * blk_queue_flag_clear - atomically clear a queue flag
 * @flag: flag to be cleared
 * @q: request queue
 */
void blk_queue_flag_clear(unsigned int flag, struct request_queue *q)
{
89
	clear_bit(flag, &q->queue_flags);
90 91 92 93 94 95 96 97 98 99 100 101 102
}
EXPORT_SYMBOL(blk_queue_flag_clear);

/**
 * blk_queue_flag_test_and_set - atomically test and set a queue flag
 * @flag: flag to be set
 * @q: request queue
 *
 * Returns the previous value of @flag - 0 if the flag was not set and 1 if
 * the flag was already set.
 */
bool blk_queue_flag_test_and_set(unsigned int flag, struct request_queue *q)
{
103
	return test_and_set_bit(flag, &q->queue_flags);
104 105 106
}
EXPORT_SYMBOL_GPL(blk_queue_flag_test_and_set);

107
void blk_rq_init(struct request_queue *q, struct request *rq)
Linus Torvalds's avatar
Linus Torvalds committed
108
{
109 110
	memset(rq, 0, sizeof(*rq));

Linus Torvalds's avatar
Linus Torvalds committed
111
	INIT_LIST_HEAD(&rq->queuelist);
Jens Axboe's avatar
Jens Axboe committed
112
	rq->q = q;
113
	rq->__sector = (sector_t) -1;
114 115
	INIT_HLIST_NODE(&rq->hash);
	RB_CLEAR_NODE(&rq->rb_node);
Jens Axboe's avatar
Jens Axboe committed
116
	rq->tag = -1;
117
	rq->internal_tag = -1;
118
	rq->start_time_ns = ktime_get_ns();
119
	rq->part = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
120
}
121
EXPORT_SYMBOL(blk_rq_init);
Linus Torvalds's avatar
Linus Torvalds committed
122

123 124 125 126 127 128 129 130 131 132 133 134 135 136
static const struct {
	int		errno;
	const char	*name;
} blk_errors[] = {
	[BLK_STS_OK]		= { 0,		"" },
	[BLK_STS_NOTSUPP]	= { -EOPNOTSUPP, "operation not supported" },
	[BLK_STS_TIMEOUT]	= { -ETIMEDOUT,	"timeout" },
	[BLK_STS_NOSPC]		= { -ENOSPC,	"critical space allocation" },
	[BLK_STS_TRANSPORT]	= { -ENOLINK,	"recoverable transport" },
	[BLK_STS_TARGET]	= { -EREMOTEIO,	"critical target" },
	[BLK_STS_NEXUS]		= { -EBADE,	"critical nexus" },
	[BLK_STS_MEDIUM]	= { -ENODATA,	"critical medium" },
	[BLK_STS_PROTECTION]	= { -EILSEQ,	"protection" },
	[BLK_STS_RESOURCE]	= { -ENOMEM,	"kernel resource" },
137
	[BLK_STS_DEV_RESOURCE]	= { -EBUSY,	"device resource" },
138
	[BLK_STS_AGAIN]		= { -EAGAIN,	"nonblocking retry" },
139

140 141 142
	/* device mapper special case, should not leak out: */
	[BLK_STS_DM_REQUEUE]	= { -EREMCHG, "dm internal retry" },

143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163
	/* everything else not covered above: */
	[BLK_STS_IOERR]		= { -EIO,	"I/O" },
};

blk_status_t errno_to_blk_status(int errno)
{
	int i;

	for (i = 0; i < ARRAY_SIZE(blk_errors); i++) {
		if (blk_errors[i].errno == errno)
			return (__force blk_status_t)i;
	}

	return BLK_STS_IOERR;
}
EXPORT_SYMBOL_GPL(errno_to_blk_status);

int blk_status_to_errno(blk_status_t status)
{
	int idx = (__force int)status;

164
	if (WARN_ON_ONCE(idx >= ARRAY_SIZE(blk_errors)))
165 166 167 168 169 170 171 172 173
		return -EIO;
	return blk_errors[idx].errno;
}
EXPORT_SYMBOL_GPL(blk_status_to_errno);

static void print_req_error(struct request *req, blk_status_t status)
{
	int idx = (__force int)status;

174
	if (WARN_ON_ONCE(idx >= ARRAY_SIZE(blk_errors)))
175 176
		return;

177 178 179 180 181
	printk_ratelimited(KERN_ERR "%s: %s error, dev %s, sector %llu flags %x\n",
				__func__, blk_errors[idx].name,
				req->rq_disk ?  req->rq_disk->disk_name : "?",
				(unsigned long long)blk_rq_pos(req),
				req->cmd_flags);
182 183
}

184
static void req_bio_endio(struct request *rq, struct bio *bio,
185
			  unsigned int nbytes, blk_status_t error)
Linus Torvalds's avatar
Linus Torvalds committed
186
{
187
	if (error)
188
		bio->bi_status = error;
189

190
	if (unlikely(rq->rq_flags & RQF_QUIET))
191
		bio_set_flag(bio, BIO_QUIET);
192

193
	bio_advance(bio, nbytes);
194

195
	/* don't actually finish bio if it's part of flush sequence */
196
	if (bio->bi_iter.bi_size == 0 && !(rq->rq_flags & RQF_FLUSH_SEQ))
197
		bio_endio(bio);
Linus Torvalds's avatar
Linus Torvalds committed
198 199 200 201
}

void blk_dump_rq_flags(struct request *rq, char *msg)
{
202 203
	printk(KERN_INFO "%s: dev %s: flags=%llx\n", msg,
		rq->rq_disk ? rq->rq_disk->disk_name : "?",
204
		(unsigned long long) rq->cmd_flags);
Linus Torvalds's avatar
Linus Torvalds committed
205

206 207 208
	printk(KERN_INFO "  sector %llu, nr/cnr %u/%u\n",
	       (unsigned long long)blk_rq_pos(rq),
	       blk_rq_sectors(rq), blk_rq_cur_sectors(rq));
209 210
	printk(KERN_INFO "  bio %p, biotail %p, len %u\n",
	       rq->bio, rq->biotail, blk_rq_bytes(rq));
Linus Torvalds's avatar
Linus Torvalds committed
211 212 213 214 215 216 217 218 219 220 221 222
}
EXPORT_SYMBOL(blk_dump_rq_flags);

/**
 * blk_sync_queue - cancel any pending callbacks on a queue
 * @q: the queue
 *
 * Description:
 *     The block layer may perform asynchronous callback activity
 *     on a queue, such as calling the unplug function after a timeout.
 *     A block device may call blk_sync_queue to ensure that any
 *     such activity is cancelled, thus allowing it to release resources
223
 *     that the callbacks might use. The caller must already have made sure
Linus Torvalds's avatar
Linus Torvalds committed
224 225 226
 *     that its ->make_request_fn will not re-add plugging prior to calling
 *     this function.
 *
227
 *     This function does not cancel any asynchronous activity arising
228
 *     out of elevator or throttling code. That would require elevator_exit()
229
 *     and blkcg_exit_queue() to be called with queue lock initialized.
230
 *
Linus Torvalds's avatar
Linus Torvalds committed
231 232 233
 */
void blk_sync_queue(struct request_queue *q)
{
234
	del_timer_sync(&q->timeout);
235
	cancel_work_sync(&q->timeout_work);
Linus Torvalds's avatar
Linus Torvalds committed
236 237 238
}
EXPORT_SYMBOL(blk_sync_queue);

239
/**
240
 * blk_set_pm_only - increment pm_only counter
241 242
 * @q: request queue pointer
 */
243
void blk_set_pm_only(struct request_queue *q)
244
{
245
	atomic_inc(&q->pm_only);
246
}
247
EXPORT_SYMBOL_GPL(blk_set_pm_only);
248

249
void blk_clear_pm_only(struct request_queue *q)
250
{
251 252 253 254 255 256
	int pm_only;

	pm_only = atomic_dec_return(&q->pm_only);
	WARN_ON_ONCE(pm_only < 0);
	if (pm_only == 0)
		wake_up_all(&q->mq_freeze_wq);
257
}
258
EXPORT_SYMBOL_GPL(blk_clear_pm_only);
259

260
void blk_put_queue(struct request_queue *q)
261 262 263
{
	kobject_put(&q->kobj);
}
264
EXPORT_SYMBOL(blk_put_queue);
265

266 267
void blk_set_queue_dying(struct request_queue *q)
{
268
	blk_queue_flag_set(QUEUE_FLAG_DYING, q);
269

270 271 272 273 274 275 276
	/*
	 * When queue DYING flag is set, we need to block new req
	 * entering queue, so we call blk_freeze_queue_start() to
	 * prevent I/O from crossing blk_queue_enter().
	 */
	blk_freeze_queue_start(q);

Jens Axboe's avatar
Jens Axboe committed
277
	if (queue_is_mq(q))
278
		blk_mq_wake_waiters(q);
279 280 281

	/* Make blk_queue_enter() reexamine the DYING flag. */
	wake_up_all(&q->mq_freeze_wq);
282 283 284
}
EXPORT_SYMBOL_GPL(blk_set_queue_dying);

285 286 287 288
/**
 * blk_cleanup_queue - shutdown a request queue
 * @q: request queue to shutdown
 *
289 290
 * Mark @q DYING, drain all pending requests, mark @q DEAD, destroy and
 * put it.  All future requests will be failed immediately with -ENODEV.
291
 */
292
void blk_cleanup_queue(struct request_queue *q)
293
{
294
	/* mark @q DYING, no new request or merges will be allowed afterwards */
295
	mutex_lock(&q->sysfs_lock);
296
	blk_set_queue_dying(q);
297

298 299 300
	blk_queue_flag_set(QUEUE_FLAG_NOMERGES, q);
	blk_queue_flag_set(QUEUE_FLAG_NOXMERGES, q);
	blk_queue_flag_set(QUEUE_FLAG_DYING, q);
301 302
	mutex_unlock(&q->sysfs_lock);

303 304 305 306
	/*
	 * Drain all requests queued before DYING marking. Set DEAD flag to
	 * prevent that q->request_fn() gets invoked after draining finished.
	 */
307
	blk_freeze_queue(q);
308 309 310

	rq_qos_exit(q);

311
	blk_queue_flag_set(QUEUE_FLAG_DEAD, q);
312

313 314 315
	/* for synchronous bio-based driver finish in-flight integrity i/o */
	blk_flush_integrity();

316
	/* @q won't process any more request, flush async actions */
317
	del_timer_sync(&q->backing_dev_info->laptop_mode_wb_timer);
318 319
	blk_sync_queue(q);

Jens Axboe's avatar
Jens Axboe committed
320
	if (queue_is_mq(q))
321
		blk_mq_exit_queue(q);
322

323 324 325 326 327 328 329 330 331 332 333 334 335
	/*
	 * In theory, request pool of sched_tags belongs to request queue.
	 * However, the current implementation requires tag_set for freeing
	 * requests, so free the pool now.
	 *
	 * Queue has become frozen, there can't be any in-queue requests, so
	 * it is safe to free requests now.
	 */
	mutex_lock(&q->sysfs_lock);
	if (q->elevator)
		blk_mq_sched_free_requests(q);
	mutex_unlock(&q->sysfs_lock);

336
	percpu_ref_exit(&q->q_usage_counter);
337

338
	/* @q is and will stay empty, shutdown and put */
339 340
	blk_put_queue(q);
}
Linus Torvalds's avatar
Linus Torvalds committed
341 342
EXPORT_SYMBOL(blk_cleanup_queue);

343
struct request_queue *blk_alloc_queue(gfp_t gfp_mask)
Linus Torvalds's avatar
Linus Torvalds committed
344
{
345
	return blk_alloc_queue_node(gfp_mask, NUMA_NO_NODE);
346 347
}
EXPORT_SYMBOL(blk_alloc_queue);
Linus Torvalds's avatar
Linus Torvalds committed
348

349 350 351 352 353
/**
 * blk_queue_enter() - try to increase q->q_usage_counter
 * @q: request queue pointer
 * @flags: BLK_MQ_REQ_NOWAIT and/or BLK_MQ_REQ_PREEMPT
 */
354
int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags)
355
{
356
	const bool pm = flags & BLK_MQ_REQ_PREEMPT;
357

358
	while (true) {
359
		bool success = false;
360

361
		rcu_read_lock();
362 363
		if (percpu_ref_tryget_live(&q->q_usage_counter)) {
			/*
364 365 366
			 * The code that increments the pm_only counter is
			 * responsible for ensuring that that counter is
			 * globally visible before the queue is unfrozen.
367
			 */
368
			if (pm || !blk_queue_pm_only(q)) {
369 370 371 372 373
				success = true;
			} else {
				percpu_ref_put(&q->q_usage_counter);
			}
		}
374
		rcu_read_unlock();
375 376

		if (success)
377 378
			return 0;

379
		if (flags & BLK_MQ_REQ_NOWAIT)
380 381
			return -EBUSY;

382
		/*
383
		 * read pair of barrier in blk_freeze_queue_start(),
384
		 * we need to order reading __PERCPU_REF_DEAD flag of
385 386 387
		 * .q_usage_counter and reading .mq_freeze_depth or
		 * queue dying flag, otherwise the following wait may
		 * never return if the two reads are reordered.
388 389 390
		 */
		smp_rmb();

391
		wait_event(q->mq_freeze_wq,
392
			   (!q->mq_freeze_depth &&
393 394
			    (pm || (blk_pm_request_resume(q),
				    !blk_queue_pm_only(q)))) ||
395
			   blk_queue_dying(q));
396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413
		if (blk_queue_dying(q))
			return -ENODEV;
	}
}

void blk_queue_exit(struct request_queue *q)
{
	percpu_ref_put(&q->q_usage_counter);
}

static void blk_queue_usage_counter_release(struct percpu_ref *ref)
{
	struct request_queue *q =
		container_of(ref, struct request_queue, q_usage_counter);

	wake_up_all(&q->mq_freeze_wq);
}

414
static void blk_rq_timed_out_timer(struct timer_list *t)
415
{
416
	struct request_queue *q = from_timer(q, t, timeout);
417 418 419 420

	kblockd_schedule_work(&q->timeout_work);
}

421 422 423 424
static void blk_timeout_work(struct work_struct *work)
{
}

425 426 427 428 429
/**
 * blk_alloc_queue_node - allocate a request queue
 * @gfp_mask: memory allocation flags
 * @node_id: NUMA node to allocate memory from
 */
430
struct request_queue *blk_alloc_queue_node(gfp_t gfp_mask, int node_id)
431
{
432
	struct request_queue *q;
433
	int ret;
434

435
	q = kmem_cache_alloc_node(blk_requestq_cachep,
436
				gfp_mask | __GFP_ZERO, node_id);
Linus Torvalds's avatar
Linus Torvalds committed
437 438 439
	if (!q)
		return NULL;

440 441 442
	INIT_LIST_HEAD(&q->queue_head);
	q->last_merge = NULL;

443
	q->id = ida_simple_get(&blk_queue_ida, 0, 0, gfp_mask);
444
	if (q->id < 0)
445
		goto fail_q;
446

447 448
	ret = bioset_init(&q->bio_split, BIO_POOL_SIZE, 0, BIOSET_NEED_BVECS);
	if (ret)
449 450
		goto fail_id;

451 452 453 454
	q->backing_dev_info = bdi_alloc_node(gfp_mask, node_id);
	if (!q->backing_dev_info)
		goto fail_split;

455 456 457 458
	q->stats = blk_alloc_queue_stats();
	if (!q->stats)
		goto fail_stats;

459
	q->backing_dev_info->ra_pages = VM_READAHEAD_PAGES;
460 461
	q->backing_dev_info->capabilities = BDI_CAP_CGROUP_WRITEBACK;
	q->backing_dev_info->name = "block";
462
	q->node = node_id;
463

464 465 466
	timer_setup(&q->backing_dev_info->laptop_mode_wb_timer,
		    laptop_mode_timer_fn, 0);
	timer_setup(&q->timeout, blk_rq_timed_out_timer, 0);
467
	INIT_WORK(&q->timeout_work, blk_timeout_work);
468
	INIT_LIST_HEAD(&q->icq_list);
469
#ifdef CONFIG_BLK_CGROUP
470
	INIT_LIST_HEAD(&q->blkg_list);
471
#endif
472

473
	kobject_init(&q->kobj, &blk_queue_ktype);
Linus Torvalds's avatar
Linus Torvalds committed
474

475 476 477
#ifdef CONFIG_BLK_DEV_IO_TRACE
	mutex_init(&q->blk_trace_mutex);
#endif
478
	mutex_init(&q->sysfs_lock);
479
	spin_lock_init(&q->queue_lock);
480

481
	init_waitqueue_head(&q->mq_freeze_wq);
482
	mutex_init(&q->mq_freeze_lock);
483

484 485 486 487 488 489 490
	/*
	 * Init percpu_ref in atomic mode so that it's faster to shutdown.
	 * See blk_register_queue() for details.
	 */
	if (percpu_ref_init(&q->q_usage_counter,
				blk_queue_usage_counter_release,
				PERCPU_REF_INIT_ATOMIC, GFP_KERNEL))
491
		goto fail_bdi;
492

493 494 495
	if (blkcg_init_queue(q))
		goto fail_ref;

Linus Torvalds's avatar
Linus Torvalds committed
496
	return q;
497

498 499
fail_ref:
	percpu_ref_exit(&q->q_usage_counter);
500
fail_bdi:
501 502
	blk_free_queue_stats(q->stats);
fail_stats:
503
	bdi_put(q->backing_dev_info);
504
fail_split:
505
	bioset_exit(&q->bio_split);
506 507 508 509 510
fail_id:
	ida_simple_remove(&blk_queue_ida, q->id);
fail_q:
	kmem_cache_free(blk_requestq_cachep, q);
	return NULL;
Linus Torvalds's avatar
Linus Torvalds committed
511
}
512
EXPORT_SYMBOL(blk_alloc_queue_node);
Linus Torvalds's avatar
Linus Torvalds committed
513

514
bool blk_get_queue(struct request_queue *q)
Linus Torvalds's avatar
Linus Torvalds committed
515
{
516
	if (likely(!blk_queue_dying(q))) {
517 518
		__blk_get_queue(q);
		return true;
Linus Torvalds's avatar
Linus Torvalds committed
519 520
	}

521
	return false;
Linus Torvalds's avatar
Linus Torvalds committed
522
}
523
EXPORT_SYMBOL(blk_get_queue);
Linus Torvalds's avatar
Linus Torvalds committed
524

525 526 527 528 529
/**
 * blk_get_request - allocate a request
 * @q: request queue to allocate a request for
 * @op: operation (REQ_OP_*) and REQ_* flags, e.g. REQ_SYNC.
 * @flags: BLK_MQ_REQ_* flags, e.g. BLK_MQ_REQ_NOWAIT.
Linus Torvalds's avatar
Linus Torvalds committed
530
 */
531 532
struct request *blk_get_request(struct request_queue *q, unsigned int op,
				blk_mq_req_flags_t flags)
Linus Torvalds's avatar
Linus Torvalds committed
533
{
534
	struct request *req;
Linus Torvalds's avatar
Linus Torvalds committed
535

536 537
	WARN_ON_ONCE(op & REQ_NOWAIT);
	WARN_ON_ONCE(flags & ~(BLK_MQ_REQ_NOWAIT | BLK_MQ_REQ_PREEMPT));
Linus Torvalds's avatar
Linus Torvalds committed
538

539 540 541
	req = blk_mq_alloc_request(q, op, flags);
	if (!IS_ERR(req) && q->mq_ops->initialize_rq_fn)
		q->mq_ops->initialize_rq_fn(req);
Linus Torvalds's avatar
Linus Torvalds committed
542

543
	return req;
Linus Torvalds's avatar
Linus Torvalds committed
544
}
545
EXPORT_SYMBOL(blk_get_request);
Linus Torvalds's avatar
Linus Torvalds committed
546 547 548

void blk_put_request(struct request *req)
{
549
	blk_mq_free_request(req);
Linus Torvalds's avatar
Linus Torvalds committed
550 551 552
}
EXPORT_SYMBOL(blk_put_request);

553 554
bool bio_attempt_back_merge(struct request *req, struct bio *bio,
		unsigned int nr_segs)
555
{
556
	const int ff = bio->bi_opf & REQ_FAILFAST_MASK;
557

558
	if (!ll_back_merge_fn(req, bio, nr_segs))
559 560
		return false;

561
	trace_block_bio_backmerge(req->q, req, bio);
562 563 564 565 566 567

	if ((req->cmd_flags & REQ_FAILFAST_MASK) != ff)
		blk_rq_set_mixed_merge(req);

	req->biotail->bi_next = bio;
	req->biotail = bio;
568
	req->__data_len += bio->bi_iter.bi_size;
569

570
	blk_account_io_start(req, false);
571 572 573
	return true;
}

574 575
bool bio_attempt_front_merge(struct request *req, struct bio *bio,
		unsigned int nr_segs)
576
{
577
	const int ff = bio->bi_opf & REQ_FAILFAST_MASK;
578

579
	if (!ll_front_merge_fn(req, bio, nr_segs))
580 581
		return false;

582
	trace_block_bio_frontmerge(req->q, req, bio);
583 584 585 586 587 588 589

	if ((req->cmd_flags & REQ_FAILFAST_MASK) != ff)
		blk_rq_set_mixed_merge(req);

	bio->bi_next = req->bio;
	req->bio = bio;

590 591
	req->__sector = bio->bi_iter.bi_sector;
	req->__data_len += bio->bi_iter.bi_size;
592

593
	blk_account_io_start(req, false);
594 595 596
	return true;
}

597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619
bool bio_attempt_discard_merge(struct request_queue *q, struct request *req,
		struct bio *bio)
{
	unsigned short segments = blk_rq_nr_discard_segments(req);

	if (segments >= queue_max_discard_segments(q))
		goto no_merge;
	if (blk_rq_sectors(req) + bio_sectors(bio) >
	    blk_rq_get_max_sectors(req, blk_rq_pos(req)))
		goto no_merge;

	req->biotail->bi_next = bio;
	req->biotail = bio;
	req->__data_len += bio->bi_iter.bi_size;
	req->nr_phys_segments = segments + 1;

	blk_account_io_start(req, false);
	return true;
no_merge:
	req_set_nomerge(q, req);
	return false;
}

620
/**
621
 * blk_attempt_plug_merge - try to merge with %current's plugged list
622 623
 * @q: request_queue new bio is being queued at
 * @bio: new bio being queued
624
 * @nr_segs: number of segments in @bio
625 626 627
 * @same_queue_rq: pointer to &struct request that gets filled in when
 * another request associated with @q is found on the plug list
 * (optional, may be %NULL)
628 629 630 631 632
 *
 * Determine whether @bio being queued on @q can be merged with a request
 * on %current's plugged list.  Returns %true if merge was successful,
 * otherwise %false.
 *
633 634 635 636 637 638
 * Plugging coalesces IOs from the same issuer for the same purpose without
 * going through @q->queue_lock.  As such it's more of an issuing mechanism
 * than scheduling, and the request, while may have elvpriv data, is not
 * added on the elevator at this point.  In addition, we don't have
 * reliable access to the elevator outside queue lock.  Only check basic
 * merging parameters without querying the elevator.
639 640
 *
 * Caller must ensure !blk_queue_nomerges(q) beforehand.
641
 */
642
bool blk_attempt_plug_merge(struct request_queue *q, struct bio *bio,
643
		unsigned int nr_segs, struct request **same_queue_rq)
644 645 646
{
	struct blk_plug *plug;
	struct request *rq;
Shaohua Li's avatar
Shaohua Li committed
647
	struct list_head *plug_list;
648

649
	plug = current->plug;
650
	if (!plug)
651
		return false;
652

653
	plug_list = &plug->mq_list;
Shaohua Li's avatar
Shaohua Li committed
654 655

	list_for_each_entry_reverse(rq, plug_list, queuelist) {
656
		bool merged = false;
657

658
		if (rq->q == q && same_queue_rq) {
659 660 661 662 663
			/*
			 * Only blk-mq multiple hardware queues case checks the
			 * rq in the same queue, there should be only one such
			 * rq in a queue
			 **/
664
			*same_queue_rq = rq;
665
		}
666

667
		if (rq->q != q || !blk_rq_merge_ok(rq, bio))
668 669
			continue;

670 671
		switch (blk_try_merge(rq, bio)) {
		case ELEVATOR_BACK_MERGE:
672
			merged = bio_attempt_back_merge(rq, bio, nr_segs);
673 674
			break;
		case ELEVATOR_FRONT_MERGE:
675
			merged = bio_attempt_front_merge(rq, bio, nr_segs);
676
			break;
677 678 679
		case ELEVATOR_DISCARD_MERGE:
			merged = bio_attempt_discard_merge(q, rq, bio);
			break;
680 681
		default:
			break;
682
		}
683 684 685

		if (merged)
			return true;
686
	}
687 688

	return false;
689 690
}

691
static void handle_bad_sector(struct bio *bio, sector_t maxsector)
Linus Torvalds's avatar
Linus Torvalds committed
692 693 694 695
{
	char b[BDEVNAME_SIZE];

	printk(KERN_INFO "attempt to access beyond end of device\n");
696
	printk(KERN_INFO "%s: rw=%d, want=%Lu, limit=%Lu\n",
697
			bio_devname(bio, b), bio->bi_opf,
698
			(unsigned long long)bio_end_sector(bio),
699
			(long long)maxsector);
Linus Torvalds's avatar
Linus Torvalds committed
700 701
}

702 703 704 705 706 707 708 709 710 711
#ifdef CONFIG_FAIL_MAKE_REQUEST

static DECLARE_FAULT_ATTR(fail_make_request);

static int __init setup_fail_make_request(char *str)
{
	return setup_fault_attr(&fail_make_request, str);
}
__setup("fail_make_request=", setup_fail_make_request);

712
static bool should_fail_request(struct hd_struct *part, unsigned int bytes)
713
{
714
	return part->make_it_fail && should_fail(&fail_make_request, bytes);
715 716 717 718
}

static int __init fail_make_request_debugfs(void)
{
719 720 721
	struct dentry *dir = fault_create_debugfs_attr("fail_make_request",
						NULL, &fail_make_request);

722
	return PTR_ERR_OR_ZERO(dir);
723 724 725 726 727 728
}

late_initcall(fail_make_request_debugfs);

#else /* CONFIG_FAIL_MAKE_REQUEST */

729 730
static inline bool should_fail_request(struct hd_struct *part,
					unsigned int bytes)
731
{
732
	return false;
733 734 735 736
}

#endif /* CONFIG_FAIL_MAKE_REQUEST */

737 738
static inline bool bio_check_ro(struct bio *bio, struct hd_struct *part)
{
739 740
	const int op = bio_op(bio);

741
	if (part->policy && op_is_write(op)) {
742 743
		char b[BDEVNAME_SIZE];

744 745 746
		if (op_is_flush(bio->bi_opf) && !bio_sectors(bio))
			return false;

747
		WARN_ONCE(1,
748 749 750
		       "generic_make_request: Trying to write "
			"to read-only block-device %s (partno %d)\n",
			bio_devname(bio, b), part->partno);
751 752
		/* Older lvm-tools actually trigger this */
		return false;
753 754 755 756 757
	}

	return false;
}

758 759 760 761 762 763 764 765
static noinline int should_fail_bio(struct bio *bio)
{
	if (should_fail_request(&bio->bi_disk->part0, bio->bi_iter.bi_size))
		return -EIO;
	return 0;
}
ALLOW_ERROR_INJECTION(should_fail_bio, ERRNO);

766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783
/*
 * Check whether this bio extends beyond the end of the device or partition.
 * This may well happen - the kernel calls bread() without checking the size of
 * the device, e.g., when mounting a file system.
 */
static inline int bio_check_eod(struct bio *bio, sector_t maxsector)
{
	unsigned int nr_sectors = bio_sectors(bio);

	if (nr_sectors && maxsector &&
	    (nr_sectors > maxsector ||
	     bio->bi_iter.bi_sector > maxsector - nr_sectors)) {
		handle_bad_sector(bio, maxsector);
		return -EIO;
	}
	return 0;
}

784 785 786 787 788 789
/*
 * Remap block n of partition p to block n+start(p) of the disk.
 */
static inline int blk_partition_remap(struct bio *bio)
{
	struct hd_struct *p;
790
	int ret = -EIO;
791

792 793
	rcu_read_lock();
	p = __disk_get_part(bio->bi_disk, bio->bi_partno);
794 795 796 797 798
	if (unlikely(!p))
		goto out;
	if (unlikely(should_fail_request(p, bio->bi_iter.bi_size)))
		goto out;
	if (unlikely(bio_check_ro(bio, p)))
799 800
		goto out;

801 802 803 804
	/*
	 * Zone reset does not include bi_size so bio_sectors() is always 0.
	 * Include a test for the reset op code and perform the remap if needed.
	 */
805 806 807 808 809 810 811
	if (bio_sectors(bio) || bio_op(bio) == REQ_OP_ZONE_RESET) {
		if (bio_check_eod(bio, part_nr_sects_read(p)))
			goto out;
		bio->bi_iter.bi_sector += p->start_sect;
		trace_block_bio_remap(bio->bi_disk->queue, bio, part_devt(p),
				      bio->bi_iter.bi_sector - p->start_sect);
	}
812
	bio->bi_partno = 0;
813
	ret = 0;
814 815
out:
	rcu_read_unlock();
816 817 818
	return ret;
}

819 820
static noinline_for_stack bool
generic_make_request_checks(struct bio *bio)
Linus Torvalds's avatar
Linus Torvalds committed
821
{
822
	struct request_queue *q;
823
	int nr_sectors = bio_sectors(bio);
824
	blk_status_t status = BLK_STS_IOERR;
825
	char b[BDEVNAME_SIZE];
Linus Torvalds's avatar
Linus Torvalds committed
826 827 828

	might_sleep();

829
	q = bio->bi_disk->queue;
830 831 832 833
	if (unlikely(!q)) {
		printk(KERN_ERR
		       "generic_make_request: Trying to access "
			"nonexistent block-device %s (%Lu)\n",
834
			bio_devname(bio, b), (long long)bio->bi_iter.bi_sector);
835 836
		goto end_io;
	}
837

838 839 840 841
	/*
	 * For a REQ_NOWAIT based request, return -EOPNOTSUPP
	 * if queue is not a request based queue.
	 */
Jens Axboe's avatar
Jens Axboe committed
842
	if ((bio->bi_opf & REQ_NOWAIT) && !queue_is_mq(q))
843 844
		goto not_supported;

845
	if (should_fail_bio(bio))
846
		goto end_io;
847

848 849
	if (bio->bi_partno) {
		if (unlikely(blk_partition_remap(bio)))
850 851
			goto end_io;
	} else {
852 853 854
		if (unlikely(bio_check_ro(bio, &bio->bi_disk->part0)))
			goto end_io;
		if (unlikely(bio_check_eod(bio, get_capacity(bio->bi_disk))))
855 856
			goto end_io;
	}
857

858 859 860 861 862
	/*
	 * Filter flush bio's early so that make_request based
	 * drivers without flush support don't have to worry
	 * about them.
	 */
863
	if (op_is_flush(bio->bi_opf) &&
Jens Axboe's avatar
Jens Axboe committed
864
	    !test_bit(QUEUE_FLAG_WC, &q->queue_flags)) {
865
		bio->bi_opf &= ~(REQ_PREFLUSH | REQ_FUA);
866
		if (!nr_sectors) {
867
			status = BLK_STS_OK;
868 869
			goto end_io;
		}
870
	}
871

872 873 874
	if (!test_bit(QUEUE_FLAG_POLL, &q->queue_flags))
		bio->bi_opf &= ~REQ_HIPRI;

875 876 877 878 879 880 881 882 883 884
	switch (bio_op(bio)) {
	case REQ_OP_DISCARD:
		if (!blk_queue_discard(q))
			goto not_supported;
		break;
	case REQ_OP_SECURE_ERASE:
		if (!blk_queue_secure_erase(q))
			goto not_supported;
		break;
	case REQ_OP_WRITE_SAME:
885
		if (!q->limits.max_write_same_sectors)
886
			goto not_supported;
887
		break;
888
	case REQ_OP_ZONE_RESET:
889
		if (!blk_queue_is_zoned(q))
890
			goto not_supported;
891
		break;
892
	case REQ_OP_WRITE_ZEROES:
893
		if (!q->limits.max_write_zeroes_sectors)
894 895
			goto not_supported;
		break;
896 897
	default:
		break;
898
	}
899

900 901 902 903 904 905 906 907
	/*
	 * Various block parts want %current->io_context and lazy ioc
	 * allocation ends up trading a lot of pain for a small amount of
	 * memory.  Just allocate it upfront.  This may fail and block
	 * layer knows how to live with it.
	 */
	create_io_context(GFP_ATOMIC, q->node);

908 909
	if (!blkcg_bio_issue_check(q, bio))
		return false;
910

911 912 913 914 915 916 917
	if (!bio_flagged(bio, BIO_TRACE_COMPLETION)) {
		trace_block_bio_queue(q, bio);
		/* Now that enqueuing has been traced, we need to trace
		 * completion as well.
		 */
		bio_set_flag(bio, BIO_TRACE_COMPLETION);
	}
918
	return true;
919

920
not_supported:
921
	status = BLK_STS_NOTSUPP;
922
end_io:
923
	bio->bi_status = status;
924
	bio_endio(bio);
925
	return false;
Linus Torvalds's avatar
Linus Torvalds committed
926 927
}

928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950
/**
 * generic_make_request - hand a buffer to its device driver for I/O
 * @bio:  The bio describing the location in memory and on the device.
 *
 * generic_make_request() is used to make I/O requests of block
 * devices. It is passed a &struct bio, which describes the I/O that needs
 * to be done.
 *
 * generic_make_request() does not return any status.  The
 * success/failure status of the request, along with notification of
 * completion, is delivered asynchronously through the bio->bi_end_io
 * function described (one day) else where.
 *
 * The caller of generic_make_request must make sure that bi_io_vec
 * are set to describe the memory buffer, and that bi_dev and bi_sector are
 * set to describe the device address, and the
 * bi_end_io and optionally bi_private are set to describe how
 * completion notification should be signaled.
 *
 * generic_make_request and the drivers it calls may use bi_next if this
 * bio happens to be merged with someone else, and may resubmit the bio to
 * a lower device by calling into generic_make_request recursively, which
 * means the bio should NOT be touched after the call to ->make_request_fn.
951
 */
952
blk_qc_t generic_make_request(struct bio *bio)
953
{
954 955 956 957 958 959 960 961
	/*
	 * bio_list_on_stack[0] contains bios submitted by the current
	 * make_request_fn.
	 * bio_list_on_stack[1] contains bios that were submitted before
	 * the current make_request_fn, but that haven't been processed
	 * yet.
	 */
	struct bio_list bio_list_on_stack[2];
962
	blk_qc_t ret = BLK_QC_T_NONE;
963

964
	if (!generic_make_request_checks(bio))
965
		goto out;
966 967 968 969 970 971 972 973 974 975 976

	/*
	 * We only want one ->make_request_fn to be active at a time, else
	 * stack usage with stacked devices could be a problem.  So use
	 * current->bio_list to keep a list of requests submited by a
	 * make_request_fn function.  current->bio_list is also used as a
	 * flag to say if generic_make_request is currently active in this
	 * task or not.  If it is NULL, then no make_request is active.  If
	 * it is non-NULL, then a make_request is active, and new requests
	 * should be added at the tail
	 */
977
	if (current->bio_list) {
978
		bio_list_add(&current->bio_list[0], bio);
979
		goto out;
980
	}
981

982 983 984 985 986
	/* following loop may be a bit non-obvious, and so deserves some
	 * explanation.
	 * Before entering the loop, bio->bi_next is NULL (as all callers
	 * ensure that) so we have a list with a single bio.
	 * We pretend that we have just taken it off a longer list, so
987 988
	 * we assign bio_list to a pointer to the bio_list_on_stack,
	 * thus initialising the bio_list of new bios to be
989
	 * added.  ->make_request() may indeed add some more bios
990 991 992
	 * through a recursive call to generic_make_request.  If it
	 * did, we find a non-NULL value in bio_list and re-enter the loop
	 * from the top.  In this case we really did just take the bio
993
	 * of the top of the list (no pretending) and so remove it from
994
	 * bio_list, and call into ->make_request() again.
995 996
	 */
	BUG_ON(bio->bi_next);
997 998
	bio_list_init(&bio_list_on_stack[0]);
	current->bio_list = bio_list_on_stack;
999
	do {
1000 1001 1002
		struct request_queue *q = bio->bi_disk->queue;
		blk_mq_req_flags_t flags = bio->bi_opf & REQ_NOWAIT ?
			BLK_MQ_REQ_NOWAIT : 0;
1003

1004
		if (likely(blk_queue_enter(q, flags) == 0)) {
1005 1006 1007
			struct bio_list lower, same;

			/* Create a fresh bio_list for all subordinate requests */
1008 1009
			bio_list_on_stack[1] = bio_list_on_stack[0];
			bio_list_init(&bio_list_on_stack[0]);
1010
			ret = q->make_request_fn(q, bio);
1011

1012 1013
			blk_queue_exit(q);

1014 1015 1016 1017 1018
			/* sort new bios into those for a lower level
			 * and those for the same level
			 */
			bio_list_init(&lower);
			bio_list_init(&same);
1019
			while ((bio = bio_list_pop(&bio_list_on_stack[0])) != NULL)
1020
				if (q == bio->bi_disk->queue)
1021 1022 1023