blk-core.c 48.6 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1 2 3 4 5
/*
 * Copyright (C) 1991, 1992 Linus Torvalds
 * Copyright (C) 1994,      Karl Keyte: Added support for disk statistics
 * Elevator latency, (C) 2000  Andrea Arcangeli <[email protected]> SuSE
 * Queue request tables / lock, selectable elevator, Jens Axboe <[email protected]>
6 7
 * kernel-doc documentation started by NeilBrown <[email protected]>
 *	-  July2000
Linus Torvalds's avatar
Linus Torvalds committed
8 9 10 11 12 13 14 15 16 17 18
 * bio rewrite, highmem i/o, etc, Jens Axboe <[email protected]> - 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>
19
#include <linux/blk-mq.h>
Linus Torvalds's avatar
Linus Torvalds committed
20 21 22 23 24 25 26 27 28
#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>
29
#include <linux/task_io_accounting_ops.h>
30
#include <linux/fault-inject.h>
31
#include <linux/list_sort.h>
Tejun Heo's avatar
Tejun Heo committed
32
#include <linux/delay.h>
33
#include <linux/ratelimit.h>
Lin Ming's avatar
Lin Ming committed
34
#include <linux/pm_runtime.h>
35
#include <linux/blk-cgroup.h>
36
#include <linux/debugfs.h>
37
#include <linux/bpf.h>
38 39 40

#define CREATE_TRACE_POINTS
#include <trace/events/block.h>
Linus Torvalds's avatar
Linus Torvalds committed
41

42
#include "blk.h"
43
#include "blk-mq.h"
44
#include "blk-mq-sched.h"
45
#include "blk-pm.h"
46
#include "blk-rq-qos.h"
47

48 49 50 51
#ifdef CONFIG_DEBUG_FS
struct dentry *blk_debugfs_root;
#endif

52
EXPORT_TRACEPOINT_SYMBOL_GPL(block_bio_remap);
53
EXPORT_TRACEPOINT_SYMBOL_GPL(block_rq_remap);
54
EXPORT_TRACEPOINT_SYMBOL_GPL(block_bio_complete);
Keith Busch's avatar
Keith Busch committed
55
EXPORT_TRACEPOINT_SYMBOL_GPL(block_split);
56
EXPORT_TRACEPOINT_SYMBOL_GPL(block_unplug);
57

58 59
DEFINE_IDA(blk_queue_ida);

Linus Torvalds's avatar
Linus Torvalds committed
60 61 62
/*
 * For queue allocation
 */
63
struct kmem_cache *blk_requestq_cachep;
Linus Torvalds's avatar
Linus Torvalds committed
64 65 66 67

/*
 * Controlling structure to kblockd
 */
68
static struct workqueue_struct *kblockd_workqueue;
Linus Torvalds's avatar
Linus Torvalds committed
69

70 71 72 73 74 75 76
/**
 * 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)
{
77
	set_bit(flag, &q->queue_flags);
78 79 80 81 82 83 84 85 86 87
}
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)
{
88
	clear_bit(flag, &q->queue_flags);
89 90 91 92 93 94 95 96 97 98 99 100 101
}
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)
{
102
	return test_and_set_bit(flag, &q->queue_flags);
103 104 105
}
EXPORT_SYMBOL_GPL(blk_queue_flag_test_and_set);

106
void blk_rq_init(struct request_queue *q, struct request *rq)
Linus Torvalds's avatar
Linus Torvalds committed
107
{
108 109
	memset(rq, 0, sizeof(*rq));

Linus Torvalds's avatar
Linus Torvalds committed
110
	INIT_LIST_HEAD(&rq->queuelist);
Jens Axboe's avatar
Jens Axboe committed
111
	rq->q = q;
112
	rq->__sector = (sector_t) -1;
113 114
	INIT_HLIST_NODE(&rq->hash);
	RB_CLEAR_NODE(&rq->rb_node);
Jens Axboe's avatar
Jens Axboe committed
115
	rq->tag = -1;
116
	rq->internal_tag = -1;
117
	rq->start_time_ns = ktime_get_ns();
118
	rq->part = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
119
}
120
EXPORT_SYMBOL(blk_rq_init);
Linus Torvalds's avatar
Linus Torvalds committed
121

122 123 124 125 126 127 128 129 130 131 132 133 134 135
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" },
136
	[BLK_STS_DEV_RESOURCE]	= { -EBUSY,	"device resource" },
137
	[BLK_STS_AGAIN]		= { -EAGAIN,	"nonblocking retry" },
138

139 140 141
	/* device mapper special case, should not leak out: */
	[BLK_STS_DM_REQUEUE]	= { -EREMCHG, "dm internal retry" },

142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162
	/* 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;

163
	if (WARN_ON_ONCE(idx >= ARRAY_SIZE(blk_errors)))
164 165 166 167 168 169 170 171 172
		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;

173
	if (WARN_ON_ONCE(idx >= ARRAY_SIZE(blk_errors)))
174 175
		return;

176 177 178 179 180
	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);
181 182
}

183
static void req_bio_endio(struct request *rq, struct bio *bio,
184
			  unsigned int nbytes, blk_status_t error)
Linus Torvalds's avatar
Linus Torvalds committed
185
{
186
	if (error)
187
		bio->bi_status = error;
188

189
	if (unlikely(rq->rq_flags & RQF_QUIET))
190
		bio_set_flag(bio, BIO_QUIET);
191

192
	bio_advance(bio, nbytes);
193

194
	/* don't actually finish bio if it's part of flush sequence */
195
	if (bio->bi_iter.bi_size == 0 && !(rq->rq_flags & RQF_FLUSH_SEQ))
196
		bio_endio(bio);
Linus Torvalds's avatar
Linus Torvalds committed
197 198 199 200
}

void blk_dump_rq_flags(struct request *rq, char *msg)
{
201 202
	printk(KERN_INFO "%s: dev %s: flags=%llx\n", msg,
		rq->rq_disk ? rq->rq_disk->disk_name : "?",
203
		(unsigned long long) rq->cmd_flags);
Linus Torvalds's avatar
Linus Torvalds committed
204

205 206 207
	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));
208 209
	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
210 211 212 213 214 215 216 217 218 219 220 221
}
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
222
 *     that the callbacks might use. The caller must already have made sure
Linus Torvalds's avatar
Linus Torvalds committed
223 224 225
 *     that its ->make_request_fn will not re-add plugging prior to calling
 *     this function.
 *
226
 *     This function does not cancel any asynchronous activity arising
227
 *     out of elevator or throttling code. That would require elevator_exit()
228
 *     and blkcg_exit_queue() to be called with queue lock initialized.
229
 *
Linus Torvalds's avatar
Linus Torvalds committed
230 231 232
 */
void blk_sync_queue(struct request_queue *q)
{
233
	del_timer_sync(&q->timeout);
234
	cancel_work_sync(&q->timeout_work);
235

Jens Axboe's avatar
Jens Axboe committed
236
	if (queue_is_mq(q)) {
237 238 239
		struct blk_mq_hw_ctx *hctx;
		int i;

240
		cancel_delayed_work_sync(&q->requeue_work);
241
		queue_for_each_hw_ctx(q, hctx, i)
242
			cancel_delayed_work_sync(&hctx->run_work);
243
	}
Linus Torvalds's avatar
Linus Torvalds committed
244 245 246
}
EXPORT_SYMBOL(blk_sync_queue);

247
/**
248
 * blk_set_pm_only - increment pm_only counter
249 250
 * @q: request queue pointer
 */
251
void blk_set_pm_only(struct request_queue *q)
252
{
253
	atomic_inc(&q->pm_only);
254
}
255
EXPORT_SYMBOL_GPL(blk_set_pm_only);
256

257
void blk_clear_pm_only(struct request_queue *q)
258
{
259 260 261 262 263 264
	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);
265
}
266
EXPORT_SYMBOL_GPL(blk_clear_pm_only);
267

268
void blk_put_queue(struct request_queue *q)
269 270 271
{
	kobject_put(&q->kobj);
}
272
EXPORT_SYMBOL(blk_put_queue);
273

274 275
void blk_set_queue_dying(struct request_queue *q)
{
276
	blk_queue_flag_set(QUEUE_FLAG_DYING, q);
277

278 279 280 281 282 283 284
	/*
	 * 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
285
	if (queue_is_mq(q))
286
		blk_mq_wake_waiters(q);
287 288 289

	/* Make blk_queue_enter() reexamine the DYING flag. */
	wake_up_all(&q->mq_freeze_wq);
290 291 292
}
EXPORT_SYMBOL_GPL(blk_set_queue_dying);

293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321
/* Unconfigure the I/O scheduler and dissociate from the cgroup controller. */
void blk_exit_queue(struct request_queue *q)
{
	/*
	 * Since the I/O scheduler exit code may access cgroup information,
	 * perform I/O scheduler exit before disassociating from the block
	 * cgroup controller.
	 */
	if (q->elevator) {
		ioc_clear_queue(q);
		elevator_exit(q, q->elevator);
		q->elevator = NULL;
	}

	/*
	 * Remove all references to @q from the block cgroup controller before
	 * restoring @q->queue_lock to avoid that restoring this pointer causes
	 * e.g. blkcg_print_blkgs() to crash.
	 */
	blkcg_exit_queue(q);

	/*
	 * Since the cgroup code may dereference the @q->backing_dev_info
	 * pointer, only decrease its reference count after having removed the
	 * association with the block cgroup controller.
	 */
	bdi_put(q->backing_dev_info);
}

322 323 324 325
/**
 * blk_cleanup_queue - shutdown a request queue
 * @q: request queue to shutdown
 *
326 327
 * Mark @q DYING, drain all pending requests, mark @q DEAD, destroy and
 * put it.  All future requests will be failed immediately with -ENODEV.
328
 */
329
void blk_cleanup_queue(struct request_queue *q)
330
{
331
	/* mark @q DYING, no new request or merges will be allowed afterwards */
332
	mutex_lock(&q->sysfs_lock);
333
	blk_set_queue_dying(q);
334

335 336 337
	blk_queue_flag_set(QUEUE_FLAG_NOMERGES, q);
	blk_queue_flag_set(QUEUE_FLAG_NOXMERGES, q);
	blk_queue_flag_set(QUEUE_FLAG_DYING, q);
338 339
	mutex_unlock(&q->sysfs_lock);

340 341 342 343
	/*
	 * Drain all requests queued before DYING marking. Set DEAD flag to
	 * prevent that q->request_fn() gets invoked after draining finished.
	 */
344
	blk_freeze_queue(q);
345 346 347

	rq_qos_exit(q);

348
	blk_queue_flag_set(QUEUE_FLAG_DEAD, q);
349

350 351 352 353
	/*
	 * make sure all in-progress dispatch are completed because
	 * blk_freeze_queue() can only complete all requests, and
	 * dispatch may still be in-progress since we dispatch requests
354 355
	 * from more than one contexts.
	 *
356 357
	 * We rely on driver to deal with the race in case that queue
	 * initialization isn't done.
358
	 */
Jens Axboe's avatar
Jens Axboe committed
359
	if (queue_is_mq(q) && blk_queue_init_done(q))
360 361
		blk_mq_quiesce_queue(q);

362 363 364
	/* for synchronous bio-based driver finish in-flight integrity i/o */
	blk_flush_integrity();

365
	/* @q won't process any more request, flush async actions */
366
	del_timer_sync(&q->backing_dev_info->laptop_mode_wb_timer);
367 368
	blk_sync_queue(q);

369 370 371 372 373 374
	/*
	 * I/O scheduler exit is only safe after the sysfs scheduler attribute
	 * has been removed.
	 */
	WARN_ON_ONCE(q->kobj.state_in_sysfs);

375
	blk_exit_queue(q);
376

Jens Axboe's avatar
Jens Axboe committed
377
	if (queue_is_mq(q))
378
		blk_mq_free_queue(q);
379

380
	percpu_ref_exit(&q->q_usage_counter);
381

382
	/* @q is and will stay empty, shutdown and put */
383 384
	blk_put_queue(q);
}
Linus Torvalds's avatar
Linus Torvalds committed
385 386
EXPORT_SYMBOL(blk_cleanup_queue);

387
struct request_queue *blk_alloc_queue(gfp_t gfp_mask)
Linus Torvalds's avatar
Linus Torvalds committed
388
{
389
	return blk_alloc_queue_node(gfp_mask, NUMA_NO_NODE);
390 391
}
EXPORT_SYMBOL(blk_alloc_queue);
Linus Torvalds's avatar
Linus Torvalds committed
392

393 394 395 396 397
/**
 * 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
 */
398
int blk_queue_enter(struct request_queue *q, blk_mq_req_flags_t flags)
399
{
400
	const bool pm = flags & BLK_MQ_REQ_PREEMPT;
401

402
	while (true) {
403
		bool success = false;
404

405
		rcu_read_lock();
406 407
		if (percpu_ref_tryget_live(&q->q_usage_counter)) {
			/*
408 409 410
			 * The code that increments the pm_only counter is
			 * responsible for ensuring that that counter is
			 * globally visible before the queue is unfrozen.
411
			 */
412
			if (pm || !blk_queue_pm_only(q)) {
413 414 415 416 417
				success = true;
			} else {
				percpu_ref_put(&q->q_usage_counter);
			}
		}
418
		rcu_read_unlock();
419 420

		if (success)
421 422
			return 0;

423
		if (flags & BLK_MQ_REQ_NOWAIT)
424 425
			return -EBUSY;

426
		/*
427
		 * read pair of barrier in blk_freeze_queue_start(),
428
		 * we need to order reading __PERCPU_REF_DEAD flag of
429 430 431
		 * .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.
432 433 434
		 */
		smp_rmb();

435 436
		wait_event(q->mq_freeze_wq,
			   (atomic_read(&q->mq_freeze_depth) == 0 &&
437 438
			    (pm || (blk_pm_request_resume(q),
				    !blk_queue_pm_only(q)))) ||
439
			   blk_queue_dying(q));
440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457
		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);
}

458
static void blk_rq_timed_out_timer(struct timer_list *t)
459
{
460
	struct request_queue *q = from_timer(q, t, timeout);
461 462 463 464

	kblockd_schedule_work(&q->timeout_work);
}

465 466 467 468 469
/**
 * blk_alloc_queue_node - allocate a request queue
 * @gfp_mask: memory allocation flags
 * @node_id: NUMA node to allocate memory from
 */
470
struct request_queue *blk_alloc_queue_node(gfp_t gfp_mask, int node_id)
471
{
472
	struct request_queue *q;
473
	int ret;
474

475
	q = kmem_cache_alloc_node(blk_requestq_cachep,
476
				gfp_mask | __GFP_ZERO, node_id);
Linus Torvalds's avatar
Linus Torvalds committed
477 478 479
	if (!q)
		return NULL;

480 481 482
	INIT_LIST_HEAD(&q->queue_head);
	q->last_merge = NULL;

483
	q->id = ida_simple_get(&blk_queue_ida, 0, 0, gfp_mask);
484
	if (q->id < 0)
485
		goto fail_q;
486

487 488
	ret = bioset_init(&q->bio_split, BIO_POOL_SIZE, 0, BIOSET_NEED_BVECS);
	if (ret)
489 490
		goto fail_id;

491 492 493 494
	q->backing_dev_info = bdi_alloc_node(gfp_mask, node_id);
	if (!q->backing_dev_info)
		goto fail_split;

495 496 497 498
	q->stats = blk_alloc_queue_stats();
	if (!q->stats)
		goto fail_stats;

499
	q->backing_dev_info->ra_pages =
500
			(VM_MAX_READAHEAD * 1024) / PAGE_SIZE;
501 502
	q->backing_dev_info->capabilities = BDI_CAP_CGROUP_WRITEBACK;
	q->backing_dev_info->name = "block";
503
	q->node = node_id;
504

505 506 507
	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);
508
	INIT_WORK(&q->timeout_work, NULL);
509
	INIT_LIST_HEAD(&q->icq_list);
510
#ifdef CONFIG_BLK_CGROUP
511
	INIT_LIST_HEAD(&q->blkg_list);
512
#endif
513

514
	kobject_init(&q->kobj, &blk_queue_ktype);
Linus Torvalds's avatar
Linus Torvalds committed
515

516 517 518
#ifdef CONFIG_BLK_DEV_IO_TRACE
	mutex_init(&q->blk_trace_mutex);
#endif
519
	mutex_init(&q->sysfs_lock);
520
	spin_lock_init(&q->queue_lock);
521

522 523
	init_waitqueue_head(&q->mq_freeze_wq);

524 525 526 527 528 529 530
	/*
	 * 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))
531
		goto fail_bdi;
532

533 534 535
	if (blkcg_init_queue(q))
		goto fail_ref;

Linus Torvalds's avatar
Linus Torvalds committed
536
	return q;
537

538 539
fail_ref:
	percpu_ref_exit(&q->q_usage_counter);
540
fail_bdi:
541 542
	blk_free_queue_stats(q->stats);
fail_stats:
543
	bdi_put(q->backing_dev_info);
544
fail_split:
545
	bioset_exit(&q->bio_split);
546 547 548 549 550
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
551
}
552
EXPORT_SYMBOL(blk_alloc_queue_node);
Linus Torvalds's avatar
Linus Torvalds committed
553

554
bool blk_get_queue(struct request_queue *q)
Linus Torvalds's avatar
Linus Torvalds committed
555
{
556
	if (likely(!blk_queue_dying(q))) {
557 558
		__blk_get_queue(q);
		return true;
Linus Torvalds's avatar
Linus Torvalds committed
559 560
	}

561
	return false;
Linus Torvalds's avatar
Linus Torvalds committed
562
}
563
EXPORT_SYMBOL(blk_get_queue);
Linus Torvalds's avatar
Linus Torvalds committed
564

565 566 567 568 569
/**
 * 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
570
 */
571 572
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
573
{
574
	struct request *req;
Linus Torvalds's avatar
Linus Torvalds committed
575

576 577
	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
578

579 580 581
	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
582

583
	return req;
Linus Torvalds's avatar
Linus Torvalds committed
584
}
585
EXPORT_SYMBOL(blk_get_request);
Linus Torvalds's avatar
Linus Torvalds committed
586 587 588

void blk_put_request(struct request *req)
{
589
	blk_mq_free_request(req);
Linus Torvalds's avatar
Linus Torvalds committed
590 591 592
}
EXPORT_SYMBOL(blk_put_request);

593 594
bool bio_attempt_back_merge(struct request_queue *q, struct request *req,
			    struct bio *bio)
595
{
596
	const int ff = bio->bi_opf & REQ_FAILFAST_MASK;
597 598 599 600

	if (!ll_back_merge_fn(q, req, bio))
		return false;

601
	trace_block_bio_backmerge(q, req, bio);
602 603 604 605 606 607

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

	req->biotail->bi_next = bio;
	req->biotail = bio;
608
	req->__data_len += bio->bi_iter.bi_size;
609

610
	blk_account_io_start(req, false);
611 612 613
	return true;
}

614 615
bool bio_attempt_front_merge(struct request_queue *q, struct request *req,
			     struct bio *bio)
616
{
617
	const int ff = bio->bi_opf & REQ_FAILFAST_MASK;
618 619 620 621

	if (!ll_front_merge_fn(q, req, bio))
		return false;

622
	trace_block_bio_frontmerge(q, req, bio);
623 624 625 626 627 628 629

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

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

630 631
	req->__sector = bio->bi_iter.bi_sector;
	req->__data_len += bio->bi_iter.bi_size;
632

633
	blk_account_io_start(req, false);
634 635 636
	return true;
}

637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659
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;
}

660
/**
661
 * blk_attempt_plug_merge - try to merge with %current's plugged list
662 663 664
 * @q: request_queue new bio is being queued at
 * @bio: new bio being queued
 * @request_count: out parameter for number of traversed plugged requests
665 666 667
 * @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)
668 669 670 671 672
 *
 * 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.
 *
673 674 675 676 677 678
 * 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.
679 680
 *
 * Caller must ensure !blk_queue_nomerges(q) beforehand.
681
 */
682
bool blk_attempt_plug_merge(struct request_queue *q, struct bio *bio,
683
			    struct request **same_queue_rq)
684 685 686
{
	struct blk_plug *plug;
	struct request *rq;
Shaohua Li's avatar
Shaohua Li committed
687
	struct list_head *plug_list;
688

689
	plug = current->plug;
690
	if (!plug)
691
		return false;
692

693
	plug_list = &plug->mq_list;
Shaohua Li's avatar
Shaohua Li committed
694 695

	list_for_each_entry_reverse(rq, plug_list, queuelist) {
696
		bool merged = false;
697

698
		if (rq->q == q && same_queue_rq) {
699 700 701 702 703
			/*
			 * Only blk-mq multiple hardware queues case checks the
			 * rq in the same queue, there should be only one such
			 * rq in a queue
			 **/
704
			*same_queue_rq = rq;
705
		}
706

707
		if (rq->q != q || !blk_rq_merge_ok(rq, bio))
708 709
			continue;

710 711 712 713 714 715 716
		switch (blk_try_merge(rq, bio)) {
		case ELEVATOR_BACK_MERGE:
			merged = bio_attempt_back_merge(q, rq, bio);
			break;
		case ELEVATOR_FRONT_MERGE:
			merged = bio_attempt_front_merge(q, rq, bio);
			break;
717 718 719
		case ELEVATOR_DISCARD_MERGE:
			merged = bio_attempt_discard_merge(q, rq, bio);
			break;
720 721
		default:
			break;
722
		}
723 724 725

		if (merged)
			return true;
726
	}
727 728

	return false;
729 730
}

731
void blk_init_request_from_bio(struct request *req, struct bio *bio)
732
{
733
	if (bio->bi_opf & REQ_RAHEAD)
734
		req->cmd_flags |= REQ_FAILFAST_MASK;
Jens Axboe's avatar
Jens Axboe committed
735

736
	req->__sector = bio->bi_iter.bi_sector;
737
	req->ioprio = bio_prio(bio);
738
	req->write_hint = bio->bi_write_hint;
739
	blk_rq_bio_prep(req->q, req, bio);
740
}
741
EXPORT_SYMBOL_GPL(blk_init_request_from_bio);
742

743
static void handle_bad_sector(struct bio *bio, sector_t maxsector)
Linus Torvalds's avatar
Linus Torvalds committed
744 745 746 747
{
	char b[BDEVNAME_SIZE];

	printk(KERN_INFO "attempt to access beyond end of device\n");
748
	printk(KERN_INFO "%s: rw=%d, want=%Lu, limit=%Lu\n",
749
			bio_devname(bio, b), bio->bi_opf,
750
			(unsigned long long)bio_end_sector(bio),
751
			(long long)maxsector);
Linus Torvalds's avatar
Linus Torvalds committed
752 753
}

754 755 756 757 758 759 760 761 762 763
#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);

764
static bool should_fail_request(struct hd_struct *part, unsigned int bytes)
765
{
766
	return part->make_it_fail && should_fail(&fail_make_request, bytes);
767 768 769 770
}

static int __init fail_make_request_debugfs(void)
{
771 772 773
	struct dentry *dir = fault_create_debugfs_attr("fail_make_request",
						NULL, &fail_make_request);

774
	return PTR_ERR_OR_ZERO(dir);
775 776 777 778 779 780
}

late_initcall(fail_make_request_debugfs);

#else /* CONFIG_FAIL_MAKE_REQUEST */

781 782
static inline bool should_fail_request(struct hd_struct *part,
					unsigned int bytes)
783
{
784
	return false;
785 786 787 788
}

#endif /* CONFIG_FAIL_MAKE_REQUEST */

789 790
static inline bool bio_check_ro(struct bio *bio, struct hd_struct *part)
{
791 792
	const int op = bio_op(bio);

793
	if (part->policy && op_is_write(op)) {
794 795
		char b[BDEVNAME_SIZE];

796 797 798
		if (op_is_flush(bio->bi_opf) && !bio_sectors(bio))
			return false;

799
		WARN_ONCE(1,
800 801 802
		       "generic_make_request: Trying to write "
			"to read-only block-device %s (partno %d)\n",
			bio_devname(bio, b), part->partno);
803 804
		/* Older lvm-tools actually trigger this */
		return false;
805 806 807 808 809
	}

	return false;
}

810 811 812 813 814 815 816 817
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);

818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835
/*
 * 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;
}

836 837 838 839 840 841
/*
 * 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;
842
	int ret = -EIO;
843

844 845
	rcu_read_lock();
	p = __disk_get_part(bio->bi_disk, bio->bi_partno);
846 847 848 849 850
	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)))
851 852
		goto out;

853 854 855 856
	/*
	 * 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.
	 */
857 858 859 860 861 862 863
	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);
	}
864
	bio->bi_partno = 0;
865
	ret = 0;
866 867
out:
	rcu_read_unlock();
868 869 870
	return ret;
}

871 872
static noinline_for_stack bool
generic_make_request_checks(struct bio *bio)
Linus Torvalds's avatar
Linus Torvalds committed
873
{
874
	struct request_queue *q;
875
	int nr_sectors = bio_sectors(bio);
876
	blk_status_t status = BLK_STS_IOERR;
877
	char b[BDEVNAME_SIZE];
Linus Torvalds's avatar
Linus Torvalds committed
878 879 880

	might_sleep();

881
	q = bio->bi_disk->queue;
882 883 884 885
	if (unlikely(!q)) {
		printk(KERN_ERR
		       "generic_make_request: Trying to access "
			"nonexistent block-device %s (%Lu)\n",
886
			bio_devname(bio, b), (long long)bio->bi_iter.bi_sector);
887 888
		goto end_io;
	}
889

890 891 892 893
	/*
	 * For a REQ_NOWAIT based request, return -EOPNOTSUPP
	 * if queue is not a request based queue.
	 */
Jens Axboe's avatar
Jens Axboe committed
894
	if ((bio->bi_opf & REQ_NOWAIT) && !queue_is_mq(q))
895 896
		goto not_supported;

897
	if (should_fail_bio(bio))
898
		goto end_io;
899

900 901
	if (bio->bi_partno) {
		if (unlikely(blk_partition_remap(bio)))
902 903
			goto end_io;
	} else {
904 905 906
		if (unlikely(bio_check_ro(bio, &bio->bi_disk->part0)))
			goto end_io;
		if (unlikely(bio_check_eod(bio, get_capacity(bio->bi_disk))))
907 908
			goto end_io;
	}
909

910 911 912 913 914
	/*
	 * Filter flush bio's early so that make_request based
	 * drivers without flush support don't have to worry
	 * about them.
	 */
915
	if (op_is_flush(bio->bi_opf) &&
Jens Axboe's avatar
Jens Axboe committed
916
	    !test_bit(QUEUE_FLAG_WC, &q->queue_flags)) {
917
		bio->bi_opf &= ~(REQ_PREFLUSH | REQ_FUA);
918
		if (!nr_sectors) {
919
			status = BLK_STS_OK;
920 921
			goto end_io;
		}
922
	}
923

924 925 926
	if (!test_bit(QUEUE_FLAG_POLL, &q->queue_flags))
		bio->bi_opf &= ~REQ_HIPRI;

927 928 929 930 931 932 933 934 935 936
	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:
937
		if (!q->limits.max_write_same_sectors)
938
			goto not_supported;
939
		break;
940
	case REQ_OP_ZONE_RESET:
941
		if (!blk_queue_is_zoned(q))
942
			goto not_supported;
943
		break;
944
	case REQ_OP_WRITE_ZEROES:
945
		if (!q->limits.max_write_zeroes_sectors)
946 947
			goto not_supported;
		break;
948 949
	default:
		break;
950
	}
951

952 953 954 955 956 957 958 959
	/*
	 * 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);

960 961
	if (!blkcg_bio_issue_check(q, bio))
		return false;
962

963 964 965 966 967 968 969
	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);
	}
970
	return true;
971

972
not_supported:
973
	status = BLK_STS_NOTSUPP;
974
end_io:
975
	bio->bi_status = status;
976
	bio_endio(bio);
977
	return false;
Linus Torvalds's avatar
Linus Torvalds committed
978 979
}

980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002
/**
 * 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