blk-core.c 53.5 KB
Newer Older
L
Linus Torvalds 已提交
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 <andrea@suse.de> SuSE
 * Queue request tables / lock, selectable elevator, Jens Axboe <axboe@suse.de>
6 7
 * kernel-doc documentation started by NeilBrown <neilb@cse.unsw.edu.au>
 *	-  July2000
L
Linus Torvalds 已提交
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
 * 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>
#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>
28
#include <linux/task_io_accounting_ops.h>
29 30
#include <linux/interrupt.h>
#include <linux/cpu.h>
31
#include <linux/blktrace_api.h>
32
#include <linux/fault-inject.h>
L
Linus Torvalds 已提交
33

34 35
#include "blk.h"

36
static int __make_request(struct request_queue *q, struct bio *bio);
L
Linus Torvalds 已提交
37 38 39 40

/*
 * For the allocated request tables
 */
41
static struct kmem_cache *request_cachep;
L
Linus Torvalds 已提交
42 43 44 45

/*
 * For queue allocation
 */
46
struct kmem_cache *blk_requestq_cachep;
L
Linus Torvalds 已提交
47 48 49 50

/*
 * Controlling structure to kblockd
 */
51
static struct workqueue_struct *kblockd_workqueue;
L
Linus Torvalds 已提交
52

53 54
static DEFINE_PER_CPU(struct list_head, blk_cpu_done);

55 56
static void drive_stat_acct(struct request *rq, int new_io)
{
57
	struct hd_struct *part;
58 59 60 61 62
	int rw = rq_data_dir(rq);

	if (!blk_fs_request(rq) || !rq->rq_disk)
		return;

63 64 65 66
	part = get_part(rq->rq_disk, rq->sector);
	if (!new_io)
		__all_stat_inc(rq->rq_disk, part, merges[rw], rq->sector);
	else {
67 68
		disk_round_stats(rq->rq_disk);
		rq->rq_disk->in_flight++;
69 70 71 72
		if (part) {
			part_round_stats(part);
			part->in_flight++;
		}
73 74 75
	}
}

76
void blk_queue_congestion_threshold(struct request_queue *q)
L
Linus Torvalds 已提交
77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102
{
	int nr;

	nr = q->nr_requests - (q->nr_requests / 8) + 1;
	if (nr > q->nr_requests)
		nr = q->nr_requests;
	q->nr_congestion_on = nr;

	nr = q->nr_requests - (q->nr_requests / 8) - (q->nr_requests / 16) - 1;
	if (nr < 1)
		nr = 1;
	q->nr_congestion_off = nr;
}

/**
 * blk_get_backing_dev_info - get the address of a queue's backing_dev_info
 * @bdev:	device
 *
 * Locates the passed device's request queue and returns the address of its
 * backing_dev_info
 *
 * Will return NULL if the request queue cannot be located.
 */
struct backing_dev_info *blk_get_backing_dev_info(struct block_device *bdev)
{
	struct backing_dev_info *ret = NULL;
103
	struct request_queue *q = bdev_get_queue(bdev);
L
Linus Torvalds 已提交
104 105 106 107 108 109 110

	if (q)
		ret = &q->backing_dev_info;
	return ret;
}
EXPORT_SYMBOL(blk_get_backing_dev_info);

111
void blk_rq_init(struct request_queue *q, struct request *rq)
L
Linus Torvalds 已提交
112
{
113 114
	memset(rq, 0, sizeof(*rq));

L
Linus Torvalds 已提交
115
	INIT_LIST_HEAD(&rq->queuelist);
116
	INIT_LIST_HEAD(&rq->donelist);
J
Jens Axboe 已提交
117 118
	rq->q = q;
	rq->sector = rq->hard_sector = (sector_t) -1;
119 120
	INIT_HLIST_NODE(&rq->hash);
	RB_CLEAR_NODE(&rq->rb_node);
121
	rq->cmd = rq->__cmd;
J
Jens Axboe 已提交
122
	rq->tag = -1;
L
Linus Torvalds 已提交
123 124
	rq->ref_count = 1;
}
125
EXPORT_SYMBOL(blk_rq_init);
L
Linus Torvalds 已提交
126

N
NeilBrown 已提交
127 128
static void req_bio_endio(struct request *rq, struct bio *bio,
			  unsigned int nbytes, int error)
L
Linus Torvalds 已提交
129
{
130
	struct request_queue *q = rq->q;
131

N
NeilBrown 已提交
132 133 134 135 136
	if (&q->bar_rq != rq) {
		if (error)
			clear_bit(BIO_UPTODATE, &bio->bi_flags);
		else if (!test_bit(BIO_UPTODATE, &bio->bi_flags))
			error = -EIO;
137

N
NeilBrown 已提交
138
		if (unlikely(nbytes > bio->bi_size)) {
139
			printk(KERN_ERR "%s: want %u bytes done, %u left\n",
140
			       __func__, nbytes, bio->bi_size);
N
NeilBrown 已提交
141 142
			nbytes = bio->bi_size;
		}
143

N
NeilBrown 已提交
144 145
		bio->bi_size -= nbytes;
		bio->bi_sector += (nbytes >> 9);
146 147 148 149

		if (bio_integrity(bio))
			bio_integrity_advance(bio, nbytes);

N
NeilBrown 已提交
150
		if (bio->bi_size == 0)
151
			bio_endio(bio, error);
N
NeilBrown 已提交
152 153 154 155 156 157 158 159 160
	} else {

		/*
		 * Okay, this is the barrier request in progress, just
		 * record the error;
		 */
		if (error && !q->orderr)
			q->orderr = error;
	}
L
Linus Torvalds 已提交
161 162 163 164 165 166
}

void blk_dump_rq_flags(struct request *rq, char *msg)
{
	int bit;

167
	printk(KERN_INFO "%s: dev %s: type=%x, flags=%x\n", msg,
168 169
		rq->rq_disk ? rq->rq_disk->disk_name : "?", rq->cmd_type,
		rq->cmd_flags);
L
Linus Torvalds 已提交
170

171 172 173 174 175 176 177 178
	printk(KERN_INFO "  sector %llu, nr/cnr %lu/%u\n",
						(unsigned long long)rq->sector,
						rq->nr_sectors,
						rq->current_nr_sectors);
	printk(KERN_INFO "  bio %p, biotail %p, buffer %p, data %p, len %u\n",
						rq->bio, rq->biotail,
						rq->buffer, rq->data,
						rq->data_len);
L
Linus Torvalds 已提交
179

180
	if (blk_pc_request(rq)) {
181
		printk(KERN_INFO "  cdb: ");
182
		for (bit = 0; bit < BLK_MAX_CDB; bit++)
L
Linus Torvalds 已提交
183 184 185 186 187 188 189 190 191 192 193 194 195 196
			printk("%02x ", rq->cmd[bit]);
		printk("\n");
	}
}
EXPORT_SYMBOL(blk_dump_rq_flags);

/*
 * "plug" the device if there are no outstanding requests: this will
 * force the transfer to start only after we have put all the requests
 * on the list.
 *
 * This is called with interrupts off and no requests on the queue and
 * with the queue lock held.
 */
197
void blk_plug_device(struct request_queue *q)
L
Linus Torvalds 已提交
198 199 200 201 202 203 204
{
	WARN_ON(!irqs_disabled());

	/*
	 * don't plug a stopped queue, it must be paired with blk_start_queue()
	 * which will restart the queueing
	 */
205
	if (blk_queue_stopped(q))
L
Linus Torvalds 已提交
206 207
		return;

J
Jens Axboe 已提交
208
	if (!queue_flag_test_and_set(QUEUE_FLAG_PLUGGED, q)) {
L
Linus Torvalds 已提交
209
		mod_timer(&q->unplug_timer, jiffies + q->unplug_delay);
210 211
		blk_add_trace_generic(q, NULL, 0, BLK_TA_PLUG);
	}
L
Linus Torvalds 已提交
212 213 214 215 216 217 218
}
EXPORT_SYMBOL(blk_plug_device);

/*
 * remove the queue from the plugged list, if present. called with
 * queue lock held and interrupts disabled.
 */
219
int blk_remove_plug(struct request_queue *q)
L
Linus Torvalds 已提交
220 221 222
{
	WARN_ON(!irqs_disabled());

J
Jens Axboe 已提交
223
	if (!queue_flag_test_and_clear(QUEUE_FLAG_PLUGGED, q))
L
Linus Torvalds 已提交
224 225 226 227 228 229 230 231 232 233
		return 0;

	del_timer(&q->unplug_timer);
	return 1;
}
EXPORT_SYMBOL(blk_remove_plug);

/*
 * remove the plug and let it rip..
 */
234
void __generic_unplug_device(struct request_queue *q)
L
Linus Torvalds 已提交
235
{
236
	if (unlikely(blk_queue_stopped(q)))
L
Linus Torvalds 已提交
237 238 239 240 241
		return;

	if (!blk_remove_plug(q))
		return;

242
	q->request_fn(q);
L
Linus Torvalds 已提交
243 244 245 246 247
}
EXPORT_SYMBOL(__generic_unplug_device);

/**
 * generic_unplug_device - fire a request queue
248
 * @q:    The &struct request_queue in question
L
Linus Torvalds 已提交
249 250 251 252 253 254 255 256
 *
 * Description:
 *   Linux uses plugging to build bigger requests queues before letting
 *   the device have at them. If a queue is plugged, the I/O scheduler
 *   is still adding and merging requests on the queue. Once the queue
 *   gets unplugged, the request_fn defined for the queue is invoked and
 *   transfers started.
 **/
257
void generic_unplug_device(struct request_queue *q)
L
Linus Torvalds 已提交
258
{
259 260 261 262 263
	if (blk_queue_plugged(q)) {
		spin_lock_irq(q->queue_lock);
		__generic_unplug_device(q);
		spin_unlock_irq(q->queue_lock);
	}
L
Linus Torvalds 已提交
264 265 266 267 268 269
}
EXPORT_SYMBOL(generic_unplug_device);

static void blk_backing_dev_unplug(struct backing_dev_info *bdi,
				   struct page *page)
{
270
	struct request_queue *q = bdi->unplug_io_data;
L
Linus Torvalds 已提交
271

272
	blk_unplug(q);
L
Linus Torvalds 已提交
273 274
}

J
Jens Axboe 已提交
275
void blk_unplug_work(struct work_struct *work)
L
Linus Torvalds 已提交
276
{
277 278
	struct request_queue *q =
		container_of(work, struct request_queue, unplug_work);
L
Linus Torvalds 已提交
279

280 281 282
	blk_add_trace_pdu_int(q, BLK_TA_UNPLUG_IO, NULL,
				q->rq.count[READ] + q->rq.count[WRITE]);

L
Linus Torvalds 已提交
283 284 285
	q->unplug_fn(q);
}

J
Jens Axboe 已提交
286
void blk_unplug_timeout(unsigned long data)
L
Linus Torvalds 已提交
287
{
288
	struct request_queue *q = (struct request_queue *)data;
L
Linus Torvalds 已提交
289

290 291 292
	blk_add_trace_pdu_int(q, BLK_TA_UNPLUG_TIMER, NULL,
				q->rq.count[READ] + q->rq.count[WRITE]);

L
Linus Torvalds 已提交
293 294 295
	kblockd_schedule_work(&q->unplug_work);
}

296 297 298 299 300 301 302 303 304 305 306 307 308 309
void blk_unplug(struct request_queue *q)
{
	/*
	 * devices don't necessarily have an ->unplug_fn defined
	 */
	if (q->unplug_fn) {
		blk_add_trace_pdu_int(q, BLK_TA_UNPLUG_IO, NULL,
					q->rq.count[READ] + q->rq.count[WRITE]);

		q->unplug_fn(q);
	}
}
EXPORT_SYMBOL(blk_unplug);

L
Linus Torvalds 已提交
310 311
/**
 * blk_start_queue - restart a previously stopped queue
312
 * @q:    The &struct request_queue in question
L
Linus Torvalds 已提交
313 314 315 316 317 318
 *
 * Description:
 *   blk_start_queue() will clear the stop flag on the queue, and call
 *   the request_fn for the queue if it was in a stopped state when
 *   entered. Also see blk_stop_queue(). Queue lock must be held.
 **/
319
void blk_start_queue(struct request_queue *q)
L
Linus Torvalds 已提交
320
{
321 322
	WARN_ON(!irqs_disabled());

N
Nick Piggin 已提交
323
	queue_flag_clear(QUEUE_FLAG_STOPPED, q);
L
Linus Torvalds 已提交
324 325 326 327 328

	/*
	 * one level of recursion is ok and is much faster than kicking
	 * the unplug handling
	 */
J
Jens Axboe 已提交
329
	if (!queue_flag_test_and_set(QUEUE_FLAG_REENTER, q)) {
L
Linus Torvalds 已提交
330
		q->request_fn(q);
N
Nick Piggin 已提交
331
		queue_flag_clear(QUEUE_FLAG_REENTER, q);
L
Linus Torvalds 已提交
332 333 334 335 336 337 338 339 340
	} else {
		blk_plug_device(q);
		kblockd_schedule_work(&q->unplug_work);
	}
}
EXPORT_SYMBOL(blk_start_queue);

/**
 * blk_stop_queue - stop a queue
341
 * @q:    The &struct request_queue in question
L
Linus Torvalds 已提交
342 343 344 345 346 347 348 349 350 351 352
 *
 * Description:
 *   The Linux block layer assumes that a block driver will consume all
 *   entries on the request queue when the request_fn strategy is called.
 *   Often this will not happen, because of hardware limitations (queue
 *   depth settings). If a device driver gets a 'queue full' response,
 *   or if it simply chooses not to queue more I/O at one point, it can
 *   call this function to prevent the request_fn from being called until
 *   the driver has signalled it's ready to go again. This happens by calling
 *   blk_start_queue() to restart queue operations. Queue lock must be held.
 **/
353
void blk_stop_queue(struct request_queue *q)
L
Linus Torvalds 已提交
354 355
{
	blk_remove_plug(q);
N
Nick Piggin 已提交
356
	queue_flag_set(QUEUE_FLAG_STOPPED, q);
L
Linus Torvalds 已提交
357 358 359 360 361 362 363 364 365 366 367 368
}
EXPORT_SYMBOL(blk_stop_queue);

/**
 * 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
369
 *     that the callbacks might use. The caller must already have made sure
L
Linus Torvalds 已提交
370 371 372 373 374 375 376
 *     that its ->make_request_fn will not re-add plugging prior to calling
 *     this function.
 *
 */
void blk_sync_queue(struct request_queue *q)
{
	del_timer_sync(&q->unplug_timer);
377
	kblockd_flush_work(&q->unplug_work);
L
Linus Torvalds 已提交
378 379 380 381 382 383 384
}
EXPORT_SYMBOL(blk_sync_queue);

/**
 * blk_run_queue - run a single device queue
 * @q:	The queue to run
 */
N
Nick Piggin 已提交
385
void __blk_run_queue(struct request_queue *q)
L
Linus Torvalds 已提交
386 387
{
	blk_remove_plug(q);
J
Jens Axboe 已提交
388 389 390 391 392 393

	/*
	 * Only recurse once to avoid overrunning the stack, let the unplug
	 * handling reinvoke the handler shortly if we already got there.
	 */
	if (!elv_queue_empty(q)) {
J
Jens Axboe 已提交
394
		if (!queue_flag_test_and_set(QUEUE_FLAG_REENTER, q)) {
J
Jens Axboe 已提交
395
			q->request_fn(q);
N
Nick Piggin 已提交
396
			queue_flag_clear(QUEUE_FLAG_REENTER, q);
J
Jens Axboe 已提交
397 398 399 400 401
		} else {
			blk_plug_device(q);
			kblockd_schedule_work(&q->unplug_work);
		}
	}
N
Nick Piggin 已提交
402 403
}
EXPORT_SYMBOL(__blk_run_queue);
J
Jens Axboe 已提交
404

N
Nick Piggin 已提交
405 406 407 408 409 410 411 412 413 414
/**
 * blk_run_queue - run a single device queue
 * @q: The queue to run
 */
void blk_run_queue(struct request_queue *q)
{
	unsigned long flags;

	spin_lock_irqsave(q->queue_lock, flags);
	__blk_run_queue(q);
L
Linus Torvalds 已提交
415 416 417 418
	spin_unlock_irqrestore(q->queue_lock, flags);
}
EXPORT_SYMBOL(blk_run_queue);

419
void blk_put_queue(struct request_queue *q)
420 421 422 423
{
	kobject_put(&q->kobj);
}

424
void blk_cleanup_queue(struct request_queue *q)
425 426
{
	mutex_lock(&q->sysfs_lock);
N
Nick Piggin 已提交
427
	queue_flag_set_unlocked(QUEUE_FLAG_DEAD, q);
428 429 430 431 432 433 434
	mutex_unlock(&q->sysfs_lock);

	if (q->elevator)
		elevator_exit(q->elevator);

	blk_put_queue(q);
}
L
Linus Torvalds 已提交
435 436
EXPORT_SYMBOL(blk_cleanup_queue);

437
static int blk_init_free_list(struct request_queue *q)
L
Linus Torvalds 已提交
438 439 440 441 442
{
	struct request_list *rl = &q->rq;

	rl->count[READ] = rl->count[WRITE] = 0;
	rl->starved[READ] = rl->starved[WRITE] = 0;
T
Tejun Heo 已提交
443
	rl->elvpriv = 0;
L
Linus Torvalds 已提交
444 445 446
	init_waitqueue_head(&rl->wait[READ]);
	init_waitqueue_head(&rl->wait[WRITE]);

447 448
	rl->rq_pool = mempool_create_node(BLKDEV_MIN_RQ, mempool_alloc_slab,
				mempool_free_slab, request_cachep, q->node);
L
Linus Torvalds 已提交
449 450 451 452 453 454 455

	if (!rl->rq_pool)
		return -ENOMEM;

	return 0;
}

456
struct request_queue *blk_alloc_queue(gfp_t gfp_mask)
L
Linus Torvalds 已提交
457
{
458 459 460
	return blk_alloc_queue_node(gfp_mask, -1);
}
EXPORT_SYMBOL(blk_alloc_queue);
L
Linus Torvalds 已提交
461

462
struct request_queue *blk_alloc_queue_node(gfp_t gfp_mask, int node_id)
463
{
464
	struct request_queue *q;
P
Peter Zijlstra 已提交
465
	int err;
466

467
	q = kmem_cache_alloc_node(blk_requestq_cachep,
468
				gfp_mask | __GFP_ZERO, node_id);
L
Linus Torvalds 已提交
469 470 471
	if (!q)
		return NULL;

P
Peter Zijlstra 已提交
472 473 474 475
	q->backing_dev_info.unplug_io_fn = blk_backing_dev_unplug;
	q->backing_dev_info.unplug_io_data = q;
	err = bdi_init(&q->backing_dev_info);
	if (err) {
476
		kmem_cache_free(blk_requestq_cachep, q);
P
Peter Zijlstra 已提交
477 478 479
		return NULL;
	}

L
Linus Torvalds 已提交
480
	init_timer(&q->unplug_timer);
481

482
	kobject_init(&q->kobj, &blk_queue_ktype);
L
Linus Torvalds 已提交
483

484
	mutex_init(&q->sysfs_lock);
485
	spin_lock_init(&q->__queue_lock);
486

L
Linus Torvalds 已提交
487 488
	return q;
}
489
EXPORT_SYMBOL(blk_alloc_queue_node);
L
Linus Torvalds 已提交
490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512

/**
 * blk_init_queue  - prepare a request queue for use with a block device
 * @rfn:  The function to be called to process requests that have been
 *        placed on the queue.
 * @lock: Request queue spin lock
 *
 * Description:
 *    If a block device wishes to use the standard request handling procedures,
 *    which sorts requests and coalesces adjacent requests, then it must
 *    call blk_init_queue().  The function @rfn will be called when there
 *    are requests on the queue that need to be processed.  If the device
 *    supports plugging, then @rfn may not be called immediately when requests
 *    are available on the queue, but may be called at some time later instead.
 *    Plugged queues are generally unplugged when a buffer belonging to one
 *    of the requests on the queue is needed, or due to memory pressure.
 *
 *    @rfn is not required, or even expected, to remove all requests off the
 *    queue, but only as many as it can handle at a time.  If it does leave
 *    requests on the queue, it is responsible for arranging that the requests
 *    get dealt with eventually.
 *
 *    The queue spin lock must be held while manipulating the requests on the
513 514
 *    request queue; this lock will be taken also from interrupt context, so irq
 *    disabling is needed for it.
L
Linus Torvalds 已提交
515 516 517 518 519 520 521 522
 *
 *    Function returns a pointer to the initialized request queue, or NULL if
 *    it didn't succeed.
 *
 * Note:
 *    blk_init_queue() must be paired with a blk_cleanup_queue() call
 *    when the block device is deactivated (such as at module unload).
 **/
523

524
struct request_queue *blk_init_queue(request_fn_proc *rfn, spinlock_t *lock)
L
Linus Torvalds 已提交
525
{
526 527 528 529
	return blk_init_queue_node(rfn, lock, -1);
}
EXPORT_SYMBOL(blk_init_queue);

530
struct request_queue *
531 532
blk_init_queue_node(request_fn_proc *rfn, spinlock_t *lock, int node_id)
{
533
	struct request_queue *q = blk_alloc_queue_node(GFP_KERNEL, node_id);
L
Linus Torvalds 已提交
534 535 536 537

	if (!q)
		return NULL;

538
	q->node = node_id;
539
	if (blk_init_free_list(q)) {
540
		kmem_cache_free(blk_requestq_cachep, q);
541 542
		return NULL;
	}
L
Linus Torvalds 已提交
543

已提交
544 545 546 547
	/*
	 * if caller didn't supply a lock, they get per-queue locking with
	 * our embedded lock
	 */
548
	if (!lock)
已提交
549 550
		lock = &q->__queue_lock;

L
Linus Torvalds 已提交
551 552 553 554 555 556 557 558 559 560 561 562 563 564
	q->request_fn		= rfn;
	q->prep_rq_fn		= NULL;
	q->unplug_fn		= generic_unplug_device;
	q->queue_flags		= (1 << QUEUE_FLAG_CLUSTER);
	q->queue_lock		= lock;

	blk_queue_segment_boundary(q, 0xffffffff);

	blk_queue_make_request(q, __make_request);
	blk_queue_max_segment_size(q, MAX_SEGMENT_SIZE);

	blk_queue_max_hw_segments(q, MAX_HW_SEGMENTS);
	blk_queue_max_phys_segments(q, MAX_PHYS_SEGMENTS);

565 566
	q->sg_reserved_size = INT_MAX;

L
Linus Torvalds 已提交
567 568 569 570 571 572 573 574
	/*
	 * all done
	 */
	if (!elevator_init(q, NULL)) {
		blk_queue_congestion_threshold(q);
		return q;
	}

575
	blk_put_queue(q);
L
Linus Torvalds 已提交
576 577
	return NULL;
}
578
EXPORT_SYMBOL(blk_init_queue_node);
L
Linus Torvalds 已提交
579

580
int blk_get_queue(struct request_queue *q)
L
Linus Torvalds 已提交
581
{
N
Nick Piggin 已提交
582
	if (likely(!test_bit(QUEUE_FLAG_DEAD, &q->queue_flags))) {
583
		kobject_get(&q->kobj);
L
Linus Torvalds 已提交
584 585 586 587 588 589
		return 0;
	}

	return 1;
}

590
static inline void blk_free_request(struct request_queue *q, struct request *rq)
L
Linus Torvalds 已提交
591
{
592
	if (rq->cmd_flags & REQ_ELVPRIV)
T
Tejun Heo 已提交
593
		elv_put_request(q, rq);
L
Linus Torvalds 已提交
594 595 596
	mempool_free(rq, q->rq.rq_pool);
}

J
Jens Axboe 已提交
597
static struct request *
598
blk_alloc_request(struct request_queue *q, int rw, int priv, gfp_t gfp_mask)
L
Linus Torvalds 已提交
599 600 601 602 603 604
{
	struct request *rq = mempool_alloc(q->rq.rq_pool, gfp_mask);

	if (!rq)
		return NULL;

605
	blk_rq_init(q, rq);
606

L
Linus Torvalds 已提交
607
	/*
608
	 * first three bits are identical in rq->cmd_flags and bio->bi_rw,
L
Linus Torvalds 已提交
609 610
	 * see bio.h and blkdev.h
	 */
611
	rq->cmd_flags = rw | REQ_ALLOCED;
L
Linus Torvalds 已提交
612

T
Tejun Heo 已提交
613
	if (priv) {
614
		if (unlikely(elv_set_request(q, rq, gfp_mask))) {
T
Tejun Heo 已提交
615 616 617
			mempool_free(rq, q->rq.rq_pool);
			return NULL;
		}
618
		rq->cmd_flags |= REQ_ELVPRIV;
T
Tejun Heo 已提交
619
	}
L
Linus Torvalds 已提交
620

T
Tejun Heo 已提交
621
	return rq;
L
Linus Torvalds 已提交
622 623 624 625 626 627
}

/*
 * ioc_batching returns true if the ioc is a valid batching request and
 * should be given priority access to a request.
 */
628
static inline int ioc_batching(struct request_queue *q, struct io_context *ioc)
L
Linus Torvalds 已提交
629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648
{
	if (!ioc)
		return 0;

	/*
	 * Make sure the process is able to allocate at least 1 request
	 * even if the batch times out, otherwise we could theoretically
	 * lose wakeups.
	 */
	return ioc->nr_batch_requests == q->nr_batching ||
		(ioc->nr_batch_requests > 0
		&& time_before(jiffies, ioc->last_waited + BLK_BATCH_TIME));
}

/*
 * ioc_set_batching sets ioc to be a new "batcher" if it is not one. This
 * will cause the process to be a "batcher" on all queues in the system. This
 * is the behaviour we want though - once it gets a wakeup it should be given
 * a nice run.
 */
649
static void ioc_set_batching(struct request_queue *q, struct io_context *ioc)
L
Linus Torvalds 已提交
650 651 652 653 654 655 656 657
{
	if (!ioc || ioc_batching(q, ioc))
		return;

	ioc->nr_batch_requests = q->nr_batching;
	ioc->last_waited = jiffies;
}

658
static void __freed_request(struct request_queue *q, int rw)
L
Linus Torvalds 已提交
659 660 661 662
{
	struct request_list *rl = &q->rq;

	if (rl->count[rw] < queue_congestion_off_threshold(q))
663
		blk_clear_queue_congested(q, rw);
L
Linus Torvalds 已提交
664 665 666 667 668 669 670 671 672 673 674 675 676

	if (rl->count[rw] + 1 <= q->nr_requests) {
		if (waitqueue_active(&rl->wait[rw]))
			wake_up(&rl->wait[rw]);

		blk_clear_queue_full(q, rw);
	}
}

/*
 * A request has just been released.  Account for it, update the full and
 * congestion status, wake up any waiters.   Called under q->queue_lock.
 */
677
static void freed_request(struct request_queue *q, int rw, int priv)
L
Linus Torvalds 已提交
678 679 680 681
{
	struct request_list *rl = &q->rq;

	rl->count[rw]--;
T
Tejun Heo 已提交
682 683
	if (priv)
		rl->elvpriv--;
L
Linus Torvalds 已提交
684 685 686 687 688 689 690 691 692

	__freed_request(q, rw);

	if (unlikely(rl->starved[rw ^ 1]))
		__freed_request(q, rw ^ 1);
}

#define blkdev_free_rq(list) list_entry((list)->next, struct request, queuelist)
/*
N
Nick Piggin 已提交
693 694 695
 * Get a free request, queue_lock must be held.
 * Returns NULL on failure, with queue_lock held.
 * Returns !NULL on success, with queue_lock *not held*.
L
Linus Torvalds 已提交
696
 */
697
static struct request *get_request(struct request_queue *q, int rw_flags,
698
				   struct bio *bio, gfp_t gfp_mask)
L
Linus Torvalds 已提交
699 700 701
{
	struct request *rq = NULL;
	struct request_list *rl = &q->rq;
702
	struct io_context *ioc = NULL;
703
	const int rw = rw_flags & 0x01;
704 705
	int may_queue, priv;

706
	may_queue = elv_may_queue(q, rw_flags);
707 708 709 710 711
	if (may_queue == ELV_MQUEUE_NO)
		goto rq_starved;

	if (rl->count[rw]+1 >= queue_congestion_on_threshold(q)) {
		if (rl->count[rw]+1 >= q->nr_requests) {
712
			ioc = current_io_context(GFP_ATOMIC, q->node);
713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732
			/*
			 * The queue will fill after this allocation, so set
			 * it as full, and mark this process as "batching".
			 * This process will be allowed to complete a batch of
			 * requests, others will be blocked.
			 */
			if (!blk_queue_full(q, rw)) {
				ioc_set_batching(q, ioc);
				blk_set_queue_full(q, rw);
			} else {
				if (may_queue != ELV_MQUEUE_MUST
						&& !ioc_batching(q, ioc)) {
					/*
					 * The queue is full and the allocating
					 * process is not a "batcher", and not
					 * exempted by the IO scheduler
					 */
					goto out;
				}
			}
L
Linus Torvalds 已提交
733
		}
734
		blk_set_queue_congested(q, rw);
L
Linus Torvalds 已提交
735 736
	}

737 738 739 740 741
	/*
	 * Only allow batching queuers to allocate up to 50% over the defined
	 * limit of requests, otherwise we could have thousands of requests
	 * allocated with any setting of ->nr_requests
	 */
H
Hugh Dickins 已提交
742
	if (rl->count[rw] >= (3 * q->nr_requests / 2))
743
		goto out;
H
Hugh Dickins 已提交
744

L
Linus Torvalds 已提交
745 746
	rl->count[rw]++;
	rl->starved[rw] = 0;
T
Tejun Heo 已提交
747

J
Jens Axboe 已提交
748
	priv = !test_bit(QUEUE_FLAG_ELVSWITCH, &q->queue_flags);
T
Tejun Heo 已提交
749 750 751
	if (priv)
		rl->elvpriv++;

L
Linus Torvalds 已提交
752 753
	spin_unlock_irq(q->queue_lock);

754
	rq = blk_alloc_request(q, rw_flags, priv, gfp_mask);
755
	if (unlikely(!rq)) {
L
Linus Torvalds 已提交
756 757 758 759 760 761 762 763
		/*
		 * Allocation failed presumably due to memory. Undo anything
		 * we might have messed up.
		 *
		 * Allocating task should really be put onto the front of the
		 * wait queue, but this is pretty rare.
		 */
		spin_lock_irq(q->queue_lock);
T
Tejun Heo 已提交
764
		freed_request(q, rw, priv);
L
Linus Torvalds 已提交
765 766 767 768 769 770 771 772 773 774 775 776 777 778 779

		/*
		 * in the very unlikely event that allocation failed and no
		 * requests for this direction was pending, mark us starved
		 * so that freeing of a request in the other direction will
		 * notice us. another possible fix would be to split the
		 * rq mempool into READ and WRITE
		 */
rq_starved:
		if (unlikely(rl->count[rw] == 0))
			rl->starved[rw] = 1;

		goto out;
	}

780 781 782 783 784 785
	/*
	 * ioc may be NULL here, and ioc_batching will be false. That's
	 * OK, if the queue is under the request limit then requests need
	 * not count toward the nr_batch_requests limit. There will always
	 * be some limit enforced by BLK_BATCH_TIME.
	 */
L
Linus Torvalds 已提交
786 787
	if (ioc_batching(q, ioc))
		ioc->nr_batch_requests--;
788

789
	blk_add_trace_generic(q, bio, rw, BLK_TA_GETRQ);
L
Linus Torvalds 已提交
790 791 792 793 794 795 796
out:
	return rq;
}

/*
 * No available requests for this queue, unplug the device and wait for some
 * requests to become available.
N
Nick Piggin 已提交
797 798
 *
 * Called with q->queue_lock held, and returns with it unlocked.
L
Linus Torvalds 已提交
799
 */
800
static struct request *get_request_wait(struct request_queue *q, int rw_flags,
801
					struct bio *bio)
L
Linus Torvalds 已提交
802
{
803
	const int rw = rw_flags & 0x01;
L
Linus Torvalds 已提交
804 805
	struct request *rq;

806
	rq = get_request(q, rw_flags, bio, GFP_NOIO);
807 808
	while (!rq) {
		DEFINE_WAIT(wait);
809
		struct io_context *ioc;
L
Linus Torvalds 已提交
810 811 812 813 814
		struct request_list *rl = &q->rq;

		prepare_to_wait_exclusive(&rl->wait[rw], &wait,
				TASK_UNINTERRUPTIBLE);

815
		blk_add_trace_generic(q, bio, rw, BLK_TA_SLEEPRQ);
L
Linus Torvalds 已提交
816

817 818 819
		__generic_unplug_device(q);
		spin_unlock_irq(q->queue_lock);
		io_schedule();
L
Linus Torvalds 已提交
820

821 822 823 824 825 826 827 828
		/*
		 * After sleeping, we become a "batching" process and
		 * will be able to allocate at least one request, and
		 * up to a big batch of them for a small period time.
		 * See ioc_batching, ioc_set_batching
		 */
		ioc = current_io_context(GFP_NOIO, q->node);
		ioc_set_batching(q, ioc);
N
Nick Piggin 已提交
829

830
		spin_lock_irq(q->queue_lock);
L
Linus Torvalds 已提交
831
		finish_wait(&rl->wait[rw], &wait);
832 833 834

		rq = get_request(q, rw_flags, bio, GFP_NOIO);
	};
L
Linus Torvalds 已提交
835 836 837 838

	return rq;
}

839
struct request *blk_get_request(struct request_queue *q, int rw, gfp_t gfp_mask)
L
Linus Torvalds 已提交
840 841 842 843 844
{
	struct request *rq;

	BUG_ON(rw != READ && rw != WRITE);

N
Nick Piggin 已提交
845 846
	spin_lock_irq(q->queue_lock);
	if (gfp_mask & __GFP_WAIT) {
847
		rq = get_request_wait(q, rw, NULL);
N
Nick Piggin 已提交
848
	} else {
849
		rq = get_request(q, rw, NULL, gfp_mask);
N
Nick Piggin 已提交
850 851 852 853
		if (!rq)
			spin_unlock_irq(q->queue_lock);
	}
	/* q->queue_lock is unlocked at this point */
L
Linus Torvalds 已提交
854 855 856 857 858

	return rq;
}
EXPORT_SYMBOL(blk_get_request);

859 860 861 862 863 864 865 866 867 868
/**
 * blk_start_queueing - initiate dispatch of requests to device
 * @q:		request queue to kick into gear
 *
 * This is basically a helper to remove the need to know whether a queue
 * is plugged or not if someone just wants to initiate dispatch of requests
 * for this queue.
 *
 * The queue lock must be held with interrupts disabled.
 */
869
void blk_start_queueing(struct request_queue *q)
870 871 872 873 874 875 876 877
{
	if (!blk_queue_plugged(q))
		q->request_fn(q);
	else
		__generic_unplug_device(q);
}
EXPORT_SYMBOL(blk_start_queueing);

L
Linus Torvalds 已提交
878 879 880 881 882 883 884 885 886 887
/**
 * blk_requeue_request - put a request back on queue
 * @q:		request queue where request should be inserted
 * @rq:		request to be inserted
 *
 * Description:
 *    Drivers often keep queueing requests until the hardware cannot accept
 *    more, when that condition happens we need to put the request back
 *    on the queue. Must be called with queue lock held.
 */
888
void blk_requeue_request(struct request_queue *q, struct request *rq)
L
Linus Torvalds 已提交
889
{
890 891
	blk_add_trace_rq(q, rq, BLK_TA_REQUEUE);

L
Linus Torvalds 已提交
892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917
	if (blk_rq_tagged(rq))
		blk_queue_end_tag(q, rq);

	elv_requeue_request(q, rq);
}
EXPORT_SYMBOL(blk_requeue_request);

/**
 * blk_insert_request - insert a special request in to a request queue
 * @q:		request queue where request should be inserted
 * @rq:		request to be inserted
 * @at_head:	insert request at head or tail of queue
 * @data:	private data
 *
 * Description:
 *    Many block devices need to execute commands asynchronously, so they don't
 *    block the whole kernel from preemption during request execution.  This is
 *    accomplished normally by inserting aritficial requests tagged as
 *    REQ_SPECIAL in to the corresponding request queue, and letting them be
 *    scheduled for actual execution by the request queue.
 *
 *    We have the option of inserting the head or the tail of the queue.
 *    Typically we use the tail for new ioctls and so forth.  We use the head
 *    of the queue for things like a QUEUE_FULL message from a device, or a
 *    host that is unable to accept a particular command.
 */
918
void blk_insert_request(struct request_queue *q, struct request *rq,
919
			int at_head, void *data)
L
Linus Torvalds 已提交
920
{
921
	int where = at_head ? ELEVATOR_INSERT_FRONT : ELEVATOR_INSERT_BACK;
L
Linus Torvalds 已提交
922 923 924 925 926 927 928
	unsigned long flags;

	/*
	 * tell I/O scheduler that this isn't a regular read/write (ie it
	 * must not attempt merges on this) and that it acts as a soft
	 * barrier
	 */
929 930
	rq->cmd_type = REQ_TYPE_SPECIAL;
	rq->cmd_flags |= REQ_SOFTBARRIER;
L
Linus Torvalds 已提交
931 932 933 934 935 936 937 938

	rq->special = data;

	spin_lock_irqsave(q->queue_lock, flags);

	/*
	 * If command is tagged, release the tag
	 */
939 940
	if (blk_rq_tagged(rq))
		blk_queue_end_tag(q, rq);
L
Linus Torvalds 已提交
941

942
	drive_stat_acct(rq, 1);
943
	__elv_add_request(q, rq, where, 0);
944
	blk_start_queueing(q);
L
Linus Torvalds 已提交
945 946 947 948 949 950 951 952 953
	spin_unlock_irqrestore(q->queue_lock, flags);
}
EXPORT_SYMBOL(blk_insert_request);

/*
 * add-request adds a request to the linked list.
 * queue lock is held and interrupts disabled, as we muck with the
 * request queue list.
 */
954
static inline void add_request(struct request_queue *q, struct request *req)
L
Linus Torvalds 已提交
955
{
956
	drive_stat_acct(req, 1);
L
Linus Torvalds 已提交
957 958 959 960 961 962 963

	/*
	 * elevator indicated where it wants this request to be
	 * inserted at elevator_merge time
	 */
	__elv_add_request(q, req, ELEVATOR_INSERT_SORT, 0);
}
964

L
Linus Torvalds 已提交
965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983
/*
 * disk_round_stats()	- Round off the performance stats on a struct
 * disk_stats.
 *
 * The average IO queue length and utilisation statistics are maintained
 * by observing the current state of the queue length and the amount of
 * time it has been in this state for.
 *
 * Normally, that accounting is done on IO completion, but that can result
 * in more than a second's worth of IO being accounted for within any one
 * second, leading to >100% utilisation.  To deal with that, we call this
 * function to do a round-off before returning the results when reading
 * /proc/diskstats.  This accounts immediately for all queue usage up to
 * the current jiffies and restarts the counters again.
 */
void disk_round_stats(struct gendisk *disk)
{
	unsigned long now = jiffies;

984 985
	if (now == disk->stamp)
		return;
L
Linus Torvalds 已提交
986

987 988 989 990 991
	if (disk->in_flight) {
		__disk_stat_add(disk, time_in_queue,
				disk->in_flight * (now - disk->stamp));
		__disk_stat_add(disk, io_ticks, (now - disk->stamp));
	}
L
Linus Torvalds 已提交
992 993
	disk->stamp = now;
}
994 995
EXPORT_SYMBOL_GPL(disk_round_stats);

996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010
void part_round_stats(struct hd_struct *part)
{
	unsigned long now = jiffies;

	if (now == part->stamp)
		return;

	if (part->in_flight) {
		__part_stat_add(part, time_in_queue,
				part->in_flight * (now - part->stamp));
		__part_stat_add(part, io_ticks, (now - part->stamp));
	}
	part->stamp = now;
}

L
Linus Torvalds 已提交
1011 1012 1013
/*
 * queue lock must be held
 */
1014
void __blk_put_request(struct request_queue *q, struct request *req)
L
Linus Torvalds 已提交
1015 1016 1017 1018 1019 1020
{
	if (unlikely(!q))
		return;
	if (unlikely(--req->ref_count))
		return;

1021 1022
	elv_completed_request(q, req);

L
Linus Torvalds 已提交
1023 1024 1025 1026
	/*
	 * Request may not have originated from ll_rw_blk. if not,
	 * it didn't come out of our reserved rq pools
	 */
1027
	if (req->cmd_flags & REQ_ALLOCED) {
L
Linus Torvalds 已提交
1028
		int rw = rq_data_dir(req);
1029
		int priv = req->cmd_flags & REQ_ELVPRIV;
L
Linus Torvalds 已提交
1030 1031

		BUG_ON(!list_empty(&req->queuelist));
1032
		BUG_ON(!hlist_unhashed(&req->hash));
L
Linus Torvalds 已提交
1033 1034

		blk_free_request(q, req);
T
Tejun Heo 已提交
1035
		freed_request(q, rw, priv);
L
Linus Torvalds 已提交
1036 1037
	}
}
1038 1039
EXPORT_SYMBOL_GPL(__blk_put_request);

L
Linus Torvalds 已提交
1040 1041
void blk_put_request(struct request *req)
{
1042
	unsigned long flags;
1043
	struct request_queue *q = req->q;
1044

1045 1046 1047
	spin_lock_irqsave(q->queue_lock, flags);
	__blk_put_request(q, req);
	spin_unlock_irqrestore(q->queue_lock, flags);
L
Linus Torvalds 已提交
1048 1049 1050
}
EXPORT_SYMBOL(blk_put_request);

J
Jens Axboe 已提交
1051
void init_request_from_bio(struct request *req, struct bio *bio)
1052
{
1053
	req->cmd_type = REQ_TYPE_FS;
1054 1055 1056 1057 1058

	/*
	 * inherit FAILFAST from bio (for read-ahead, and explicit FAILFAST)
	 */
	if (bio_rw_ahead(bio) || bio_failfast(bio))
1059
		req->cmd_flags |= REQ_FAILFAST;
1060 1061 1062 1063 1064

	/*
	 * REQ_BARRIER implies no merging, but lets make it explicit
	 */
	if (unlikely(bio_barrier(bio)))
1065
		req->cmd_flags |= (REQ_HARDBARRIER | REQ_NOMERGE);
1066

J
Jens Axboe 已提交
1067
	if (bio_sync(bio))
1068
		req->cmd_flags |= REQ_RW_SYNC;
1069 1070
	if (bio_rw_meta(bio))
		req->cmd_flags |= REQ_RW_META;
J
Jens Axboe 已提交
1071

1072 1073 1074 1075
	req->errors = 0;
	req->hard_sector = req->sector = bio->bi_sector;
	req->ioprio = bio_prio(bio);
	req->start_time = jiffies;
1076
	blk_rq_bio_prep(req->q, req, bio);
1077 1078
}

1079
static int __make_request(struct request_queue *q, struct bio *bio)
L
Linus Torvalds 已提交
1080
{
1081
	struct request *req;
1082 1083 1084
	int el_ret, nr_sectors, barrier, err;
	const unsigned short prio = bio_prio(bio);
	const int sync = bio_sync(bio);
1085
	int rw_flags;
L
Linus Torvalds 已提交
1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096

	nr_sectors = bio_sectors(bio);

	/*
	 * low level driver can indicate that it wants pages above a
	 * certain limit bounced to low memory (ie for highmem, or even
	 * ISA dma in theory)
	 */
	blk_queue_bounce(q, &bio);

	barrier = bio_barrier(bio);
1097
	if (unlikely(barrier) && (q->next_ordered == QUEUE_ORDERED_NONE)) {
L
Linus Torvalds 已提交
1098 1099 1100 1101 1102 1103
		err = -EOPNOTSUPP;
		goto end_io;
	}

	spin_lock_irq(q->queue_lock);

1104
	if (unlikely(barrier) || elv_queue_empty(q))
L
Linus Torvalds 已提交
1105 1106 1107 1108
		goto get_rq;

	el_ret = elv_merge(q, &req, bio);
	switch (el_ret) {
1109 1110
	case ELEVATOR_BACK_MERGE:
		BUG_ON(!rq_mergeable(req));
L
Linus Torvalds 已提交
1111

1112 1113
		if (!ll_back_merge_fn(q, req, bio))
			break;
L
Linus Torvalds 已提交
1114

1115
		blk_add_trace_bio(q, bio, BLK_TA_BACKMERGE);
1116

1117 1118 1119 1120 1121 1122 1123 1124
		req->biotail->bi_next = bio;
		req->biotail = bio;
		req->nr_sectors = req->hard_nr_sectors += nr_sectors;
		req->ioprio = ioprio_best(req->ioprio, prio);
		drive_stat_acct(req, 0);
		if (!attempt_back_merge(q, req))
			elv_merged_request(q, req, el_ret);
		goto out;
L
Linus Torvalds 已提交
1125

1126 1127
	case ELEVATOR_FRONT_MERGE:
		BUG_ON(!rq_mergeable(req));
L
Linus Torvalds 已提交
1128

1129 1130
		if (!ll_front_merge_fn(q, req, bio))
			break;
L
Linus Torvalds 已提交
1131

1132
		blk_add_trace_bio(q, bio, BLK_TA_FRONTMERGE);
1133

1134 1135
		bio->bi_next = req->bio;
		req->bio = bio;
L
Linus Torvalds 已提交
1136

1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155
		/*
		 * may not be valid. if the low level driver said
		 * it didn't need a bounce buffer then it better
		 * not touch req->buffer either...
		 */
		req->buffer = bio_data(bio);
		req->current_nr_sectors = bio_cur_sectors(bio);
		req->hard_cur_sectors = req->current_nr_sectors;
		req->sector = req->hard_sector = bio->bi_sector;
		req->nr_sectors = req->hard_nr_sectors += nr_sectors;
		req->ioprio = ioprio_best(req->ioprio, prio);
		drive_stat_acct(req, 0);
		if (!attempt_front_merge(q, req))
			elv_merged_request(q, req, el_ret);
		goto out;

	/* ELV_NO_MERGE: elevator says don't/can't merge. */
	default:
		;
L
Linus Torvalds 已提交
1156 1157
	}

1158
get_rq:
1159 1160 1161 1162 1163 1164 1165 1166 1167
	/*
	 * This sync check and mask will be re-done in init_request_from_bio(),
	 * but we need to set it earlier to expose the sync flag to the
	 * rq allocator and io schedulers.
	 */
	rw_flags = bio_data_dir(bio);
	if (sync)
		rw_flags |= REQ_RW_SYNC;

L
Linus Torvalds 已提交
1168
	/*
1169
	 * Grab a free request. This is might sleep but can not fail.
N
Nick Piggin 已提交
1170
	 * Returns with the queue unlocked.
1171
	 */
1172
	req = get_request_wait(q, rw_flags, bio);
N
Nick Piggin 已提交
1173

1174 1175 1176 1177 1178
	/*
	 * After dropping the lock and possibly sleeping here, our request
	 * may now be mergeable after it had proven unmergeable (above).
	 * We don't worry about that case for efficiency. It won't happen
	 * often, and the elevators are able to handle it.
L
Linus Torvalds 已提交
1179
	 */
1180
	init_request_from_bio(req, bio);
L
Linus Torvalds 已提交
1181

1182 1183 1184
	spin_lock_irq(q->queue_lock);
	if (elv_queue_empty(q))
		blk_plug_device(q);
L
Linus Torvalds 已提交
1185 1186
	add_request(q, req);
out:
1187
	if (sync)
L
Linus Torvalds 已提交
1188 1189 1190 1191 1192 1193
		__generic_unplug_device(q);

	spin_unlock_irq(q->queue_lock);
	return 0;

end_io:
1194
	bio_endio(bio, err);
L
Linus Torvalds 已提交
1195 1196 1197 1198 1199 1200 1201 1202 1203 1204
	return 0;
}

/*
 * If bio->bi_dev is a partition, remap the location
 */
static inline void blk_partition_remap(struct bio *bio)
{
	struct block_device *bdev = bio->bi_bdev;

1205
	if (bio_sectors(bio) && bdev != bdev->bd_contains) {
L
Linus Torvalds 已提交
1206 1207 1208 1209
		struct hd_struct *p = bdev->bd_part;

		bio->bi_sector += p->start_sect;
		bio->bi_bdev = bdev->bd_contains;
1210 1211 1212 1213

		blk_add_trace_remap(bdev_get_queue(bio->bi_bdev), bio,
				    bdev->bd_dev, bio->bi_sector,
				    bio->bi_sector - p->start_sect);
L
Linus Torvalds 已提交
1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230
	}
}

static void handle_bad_sector(struct bio *bio)
{
	char b[BDEVNAME_SIZE];

	printk(KERN_INFO "attempt to access beyond end of device\n");
	printk(KERN_INFO "%s: rw=%ld, want=%Lu, limit=%Lu\n",
			bdevname(bio->bi_bdev, b),
			bio->bi_rw,
			(unsigned long long)bio->bi_sector + bio_sectors(bio),
			(long long)(bio->bi_bdev->bd_inode->i_size >> 9));

	set_bit(BIO_EOF, &bio->bi_flags);
}

1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266
#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);

static int should_fail_request(struct bio *bio)
{
	if ((bio->bi_bdev->bd_disk->flags & GENHD_FL_FAIL) ||
	    (bio->bi_bdev->bd_part && bio->bi_bdev->bd_part->make_it_fail))
		return should_fail(&fail_make_request, bio->bi_size);

	return 0;
}

static int __init fail_make_request_debugfs(void)
{
	return init_fault_attr_dentries(&fail_make_request,
					"fail_make_request");
}

late_initcall(fail_make_request_debugfs);

#else /* CONFIG_FAIL_MAKE_REQUEST */

static inline int should_fail_request(struct bio *bio)
{
	return 0;
}

#endif /* CONFIG_FAIL_MAKE_REQUEST */

J
Jens Axboe 已提交
1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295
/*
 * Check whether this bio extends beyond the end of the device.
 */
static inline int bio_check_eod(struct bio *bio, unsigned int nr_sectors)
{
	sector_t maxsector;

	if (!nr_sectors)
		return 0;

	/* Test device or partition size, when known. */
	maxsector = bio->bi_bdev->bd_inode->i_size >> 9;
	if (maxsector) {
		sector_t sector = bio->bi_sector;

		if (maxsector < nr_sectors || maxsector - nr_sectors < sector) {
			/*
			 * This may well happen - the kernel calls bread()
			 * without checking the size of the device, e.g., when
			 * mounting a device.
			 */
			handle_bad_sector(bio);
			return 1;
		}
	}

	return 0;
}

L
Linus Torvalds 已提交
1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319
/**
 * 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 change bi_dev and
 * bi_sector for remaps as it sees fit.  So the values of these fields
 * should NOT be depended on after the call to generic_make_request.
 */
1320
static inline void __generic_make_request(struct bio *bio)
L
Linus Torvalds 已提交
1321
{
1322
	struct request_queue *q;
1323
	sector_t old_sector;
L
Linus Torvalds 已提交
1324
	int ret, nr_sectors = bio_sectors(bio);
1325
	dev_t old_dev;
1326
	int err = -EIO;
L
Linus Torvalds 已提交
1327 1328 1329

	might_sleep();

J
Jens Axboe 已提交
1330 1331
	if (bio_check_eod(bio, nr_sectors))
		goto end_io;
L
Linus Torvalds 已提交
1332 1333 1334 1335 1336 1337 1338 1339 1340

	/*
	 * Resolve the mapping until finished. (drivers are
	 * still free to implement/resolve their own stacking
	 * by explicitly returning 0)
	 *
	 * NOTE: we don't repeat the blk_size check for each new device.
	 * Stacking drivers are expected to know what they are doing.
	 */
1341
	old_sector = -1;
1342
	old_dev = 0;
L
Linus Torvalds 已提交
1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353
	do {
		char b[BDEVNAME_SIZE];

		q = bdev_get_queue(bio->bi_bdev);
		if (!q) {
			printk(KERN_ERR
			       "generic_make_request: Trying to access "
				"nonexistent block-device %s (%Lu)\n",
				bdevname(bio->bi_bdev, b),
				(long long) bio->bi_sector);
end_io:
1354
			bio_endio(bio, err);
L
Linus Torvalds 已提交
1355 1356 1357
			break;
		}

J
Jens Axboe 已提交
1358
		if (unlikely(nr_sectors > q->max_hw_sectors)) {
1359
			printk(KERN_ERR "bio too big device %s (%u > %u)\n",
L
Linus Torvalds 已提交
1360 1361 1362 1363 1364 1365
				bdevname(bio->bi_bdev, b),
				bio_sectors(bio),
				q->max_hw_sectors);
			goto end_io;
		}

N
Nick Piggin 已提交
1366
		if (unlikely(test_bit(QUEUE_FLAG_DEAD, &q->queue_flags)))
L
Linus Torvalds 已提交
1367 1368
			goto end_io;

1369 1370 1371
		if (should_fail_request(bio))
			goto end_io;

L
Linus Torvalds 已提交
1372 1373 1374 1375 1376 1377
		/*
		 * If this device has partitions, remap block n
		 * of partition p to block n+start(p) of the disk.
		 */
		blk_partition_remap(bio);

1378 1379 1380
		if (bio_integrity_enabled(bio) && bio_integrity_prep(bio))
			goto end_io;

1381
		if (old_sector != -1)
J
Jens Axboe 已提交
1382
			blk_add_trace_remap(q, bio, old_dev, bio->bi_sector,
1383
					    old_sector);
1384 1385 1386

		blk_add_trace_bio(q, bio, BLK_TA_QUEUE);

1387
		old_sector = bio->bi_sector;
1388 1389
		old_dev = bio->bi_bdev->bd_dev;

J
Jens Axboe 已提交
1390 1391
		if (bio_check_eod(bio, nr_sectors))
			goto end_io;
1392 1393 1394 1395
		if (bio_empty_barrier(bio) && !q->prepare_flush_fn) {
			err = -EOPNOTSUPP;
			goto end_io;
		}
1396

L
Linus Torvalds 已提交
1397 1398 1399 1400
		ret = q->make_request_fn(q, bio);
	} while (ret);
}

1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450
/*
 * 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,tail} to keep a list of requests
 * submited by a make_request_fn function.
 * current->bio_tail 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
 */
void generic_make_request(struct bio *bio)
{
	if (current->bio_tail) {
		/* make_request is active */
		*(current->bio_tail) = bio;
		bio->bi_next = NULL;
		current->bio_tail = &bio->bi_next;
		return;
	}
	/* 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
	 * we assign bio_list to the next (which is NULL) and bio_tail
	 * to &bio_list, thus initialising the bio_list of new bios to be
	 * added.  __generic_make_request may indeed add some more bios
	 * 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
	 * of the top of the list (no pretending) and so fixup bio_list and
	 * bio_tail or bi_next, and call into __generic_make_request again.
	 *
	 * The loop was structured like this to make only one call to
	 * __generic_make_request (which is important as it is large and
	 * inlined) and to keep the structure simple.
	 */
	BUG_ON(bio->bi_next);
	do {
		current->bio_list = bio->bi_next;
		if (bio->bi_next == NULL)
			current->bio_tail = &current->bio_list;
		else
			bio->bi_next = NULL;
		__generic_make_request(bio);
		bio = current->bio_list;
	} while (bio);
	current->bio_tail = NULL; /* deactivate */
}
L
Linus Torvalds 已提交
1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466
EXPORT_SYMBOL(generic_make_request);

/**
 * submit_bio: submit a bio to the block device layer for I/O
 * @rw: whether to %READ or %WRITE, or maybe to %READA (read ahead)
 * @bio: The &struct bio which describes the I/O
 *
 * submit_bio() is very similar in purpose to generic_make_request(), and
 * uses that function to do most of the work. Both are fairly rough
 * interfaces, @bio must be presetup and ready for I/O.
 *
 */
void submit_bio(int rw, struct bio *bio)
{
	int count = bio_sectors(bio);

1467
	bio->bi_rw |= rw;
L
Linus Torvalds 已提交
1468

1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487
	/*
	 * If it's a regular read/write or a barrier with data attached,
	 * go through the normal accounting stuff before submission.
	 */
	if (!bio_empty_barrier(bio)) {

		BIO_BUG_ON(!bio->bi_size);
		BIO_BUG_ON(!bio->bi_io_vec);

		if (rw & WRITE) {
			count_vm_events(PGPGOUT, count);
		} else {
			task_io_account_read(bio->bi_size);
			count_vm_events(PGPGIN, count);
		}

		if (unlikely(block_dump)) {
			char b[BDEVNAME_SIZE];
			printk(KERN_DEBUG "%s(%d): %s block %Lu on %s\n",
1488
			current->comm, task_pid_nr(current),
1489 1490
				(rw & WRITE) ? "WRITE" : "READ",
				(unsigned long long)bio->bi_sector,
1491
				bdevname(bio->bi_bdev, b));
1492
		}
L
Linus Torvalds 已提交
1493 1494 1495 1496 1497 1498
	}

	generic_make_request(bio);
}
EXPORT_SYMBOL(submit_bio);

1499 1500 1501
/**
 * __end_that_request_first - end I/O on a request
 * @req:      the request being processed
1502
 * @error:    0 for success, < 0 for error
1503 1504 1505 1506 1507 1508 1509 1510 1511 1512
 * @nr_bytes: number of bytes to complete
 *
 * Description:
 *     Ends I/O on a number of bytes attached to @req, and sets it up
 *     for the next range of segments (if any) in the cluster.
 *
 * Return:
 *     0 - we are done with this request, call end_that_request_last()
 *     1 - still buffers pending for this request
 **/
1513
static int __end_that_request_first(struct request *req, int error,
L
Linus Torvalds 已提交
1514 1515
				    int nr_bytes)
{
1516
	int total_bytes, bio_nbytes, next_idx = 0;
L
Linus Torvalds 已提交
1517 1518
	struct bio *bio;

1519 1520
	blk_add_trace_rq(req->q, req, BLK_TA_COMPLETE);

L
Linus Torvalds 已提交
1521 1522 1523 1524 1525 1526 1527
	/*
	 * for a REQ_BLOCK_PC request, we want to carry any eventual
	 * sense key with us all the way through
	 */
	if (!blk_pc_request(req))
		req->errors = 0;

1528 1529
	if (error && (blk_fs_request(req) && !(req->cmd_flags & REQ_QUIET))) {
		printk(KERN_ERR "end_request: I/O error, dev %s, sector %llu\n",
L
Linus Torvalds 已提交
1530 1531 1532 1533
				req->rq_disk ? req->rq_disk->disk_name : "?",
				(unsigned long long)req->sector);
	}

1534
	if (blk_fs_request(req) && req->rq_disk) {
1535
		struct hd_struct *part = get_part(req->rq_disk, req->sector);
1536 1537
		const int rw = rq_data_dir(req);

1538 1539
		all_stat_add(req->rq_disk, part, sectors[rw],
				nr_bytes >> 9, req->sector);
1540 1541
	}

L
Linus Torvalds 已提交
1542 1543 1544 1545
	total_bytes = bio_nbytes = 0;
	while ((bio = req->bio) != NULL) {
		int nbytes;

1546 1547 1548 1549 1550 1551 1552 1553
		/*
		 * For an empty barrier request, the low level driver must
		 * store a potential error location in ->sector. We pass
		 * that back up in ->bi_sector.
		 */
		if (blk_empty_barrier(req))
			bio->bi_sector = req->sector;

L
Linus Torvalds 已提交
1554 1555 1556
		if (nr_bytes >= bio->bi_size) {
			req->bio = bio->bi_next;
			nbytes = bio->bi_size;
N
NeilBrown 已提交
1557
			req_bio_endio(req, bio, nbytes, error);
L
Linus Torvalds 已提交
1558 1559 1560 1561 1562 1563 1564
			next_idx = 0;
			bio_nbytes = 0;
		} else {
			int idx = bio->bi_idx + next_idx;

			if (unlikely(bio->bi_idx >= bio->bi_vcnt)) {
				blk_dump_rq_flags(req, "__end_that");
1565
				printk(KERN_ERR "%s: bio idx %d >= vcnt %d\n",
1566
				       __func__, bio->bi_idx, bio->bi_vcnt);
L
Linus Torvalds 已提交
1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591
				break;
			}

			nbytes = bio_iovec_idx(bio, idx)->bv_len;
			BIO_BUG_ON(nbytes > bio->bi_size);

			/*
			 * not a complete bvec done
			 */
			if (unlikely(nbytes > nr_bytes)) {
				bio_nbytes += nr_bytes;
				total_bytes += nr_bytes;
				break;
			}

			/*
			 * advance to the next vector
			 */
			next_idx++;
			bio_nbytes += nbytes;
		}

		total_bytes += nbytes;
		nr_bytes -= nbytes;

1592 1593
		bio = req->bio;
		if (bio) {
L
Linus Torvalds 已提交
1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611
			/*
			 * end more in this run, or just return 'not-done'
			 */
			if (unlikely(nr_bytes <= 0))
				break;
		}
	}

	/*
	 * completely done
	 */
	if (!req->bio)
		return 0;

	/*
	 * if the request wasn't completed, update state
	 */
	if (bio_nbytes) {
N
NeilBrown 已提交
1612
		req_bio_endio(req, bio, bio_nbytes, error);
L
Linus Torvalds 已提交
1613 1614 1615 1616 1617 1618 1619 1620 1621 1622
		bio->bi_idx += next_idx;
		bio_iovec(bio)->bv_offset += nr_bytes;
		bio_iovec(bio)->bv_len -= nr_bytes;
	}

	blk_recalc_rq_sectors(req, total_bytes >> 9);
	blk_recalc_rq_segments(req);
	return 1;
}

1623 1624 1625 1626 1627 1628
/*
 * splice the completion data to a local structure and hand off to
 * process_completion_queue() to complete the requests
 */
static void blk_done_softirq(struct softirq_action *h)
{
1629
	struct list_head *cpu_list, local_list;
1630 1631 1632

	local_irq_disable();
	cpu_list = &__get_cpu_var(blk_cpu_done);
1633
	list_replace_init(cpu_list, &local_list);
1634 1635 1636
	local_irq_enable();

	while (!list_empty(&local_list)) {
1637
		struct request *rq;
1638

1639
		rq = list_entry(local_list.next, struct request, donelist);
1640 1641 1642 1643 1644
		list_del_init(&rq->donelist);
		rq->q->softirq_done_fn(rq);
	}
}

1645 1646
static int __cpuinit blk_cpu_notify(struct notifier_block *self,
				    unsigned long action, void *hcpu)
1647 1648 1649 1650 1651
{
	/*
	 * If a CPU goes away, splice its entries to the current CPU
	 * and trigger a run of the softirq
	 */
1652
	if (action == CPU_DEAD || action == CPU_DEAD_FROZEN) {
1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665
		int cpu = (unsigned long) hcpu;

		local_irq_disable();
		list_splice_init(&per_cpu(blk_cpu_done, cpu),
				 &__get_cpu_var(blk_cpu_done));
		raise_softirq_irqoff(BLOCK_SOFTIRQ);
		local_irq_enable();
	}

	return NOTIFY_OK;
}


1666
static struct notifier_block blk_cpu_notifier __cpuinitdata = {
1667 1668 1669 1670 1671 1672 1673 1674 1675
	.notifier_call	= blk_cpu_notify,
};

/**
 * blk_complete_request - end I/O on a request
 * @req:      the request being processed
 *
 * Description:
 *     Ends all I/O on a request. It does not handle partial completions,
A
Andreas Mohr 已提交
1676
 *     unless the driver actually implements this in its completion callback
J
Jens Axboe 已提交
1677
 *     through requeueing. The actual completion happens out-of-order,
1678 1679 1680 1681 1682 1683 1684 1685 1686 1687
 *     through a softirq handler. The user must have registered a completion
 *     callback through blk_queue_softirq_done().
 **/

void blk_complete_request(struct request *req)
{
	struct list_head *cpu_list;
	unsigned long flags;

	BUG_ON(!req->q->softirq_done_fn);
1688

1689 1690 1691 1692 1693 1694 1695 1696 1697
	local_irq_save(flags);

	cpu_list = &__get_cpu_var(blk_cpu_done);
	list_add_tail(&req->donelist, cpu_list);
	raise_softirq_irqoff(BLOCK_SOFTIRQ);

	local_irq_restore(flags);
}
EXPORT_SYMBOL(blk_complete_request);
1698

L
Linus Torvalds 已提交
1699 1700 1701
/*
 * queue lock must be held
 */
1702
static void end_that_request_last(struct request *req, int error)
L
Linus Torvalds 已提交
1703 1704
{
	struct gendisk *disk = req->rq_disk;
1705

1706 1707 1708 1709 1710
	if (blk_rq_tagged(req))
		blk_queue_end_tag(req->q, req);

	if (blk_queued_rq(req))
		blkdev_dequeue_request(req);
L
Linus Torvalds 已提交
1711 1712 1713 1714

	if (unlikely(laptop_mode) && blk_fs_request(req))
		laptop_io_completion();

1715 1716 1717 1718 1719 1720
	/*
	 * Account IO completion.  bar_rq isn't accounted as a normal
	 * IO on queueing nor completion.  Accounting the containing
	 * request is enough.
	 */
	if (disk && blk_fs_request(req) && req != &req->q->bar_rq) {
L
Linus Torvalds 已提交
1721
		unsigned long duration = jiffies - req->start_time;
1722
		const int rw = rq_data_dir(req);
1723
		struct hd_struct *part = get_part(disk, req->sector);
1724

1725 1726
		__all_stat_inc(disk, part, ios[rw], req->sector);
		__all_stat_add(disk, part, ticks[rw], duration, req->sector);
L
Linus Torvalds 已提交
1727 1728
		disk_round_stats(disk);
		disk->in_flight--;
1729 1730 1731 1732
		if (part) {
			part_round_stats(part);
			part->in_flight--;
		}
L
Linus Torvalds 已提交
1733
	}
1734

L
Linus Torvalds 已提交
1735
	if (req->end_io)
1736
		req->end_io(req, error);
1737 1738 1739 1740
	else {
		if (blk_bidi_rq(req))
			__blk_put_request(req->next_rq->q, req->next_rq);

L
Linus Torvalds 已提交
1741
		__blk_put_request(req->q, req);
1742
	}
L
Linus Torvalds 已提交
1743 1744
}

1745
static inline void __end_request(struct request *rq, int uptodate,
1746
				 unsigned int nr_bytes)
L
Linus Torvalds 已提交
1747
{
1748 1749 1750 1751 1752 1753
	int error = 0;

	if (uptodate <= 0)
		error = uptodate ? uptodate : -EIO;

	__blk_end_request(rq, error, nr_bytes);
L
Linus Torvalds 已提交
1754 1755
}

1756 1757
/**
 * blk_rq_bytes - Returns bytes left to complete in the entire request
1758
 * @rq: the request being processed
1759 1760
 **/
unsigned int blk_rq_bytes(struct request *rq)
1761 1762 1763 1764 1765 1766
{
	if (blk_fs_request(rq))
		return rq->hard_nr_sectors << 9;

	return rq->data_len;
}
1767 1768 1769 1770
EXPORT_SYMBOL_GPL(blk_rq_bytes);

/**
 * blk_rq_cur_bytes - Returns bytes left to complete in the current segment
1771
 * @rq: the request being processed
1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783
 **/
unsigned int blk_rq_cur_bytes(struct request *rq)
{
	if (blk_fs_request(rq))
		return rq->current_nr_sectors << 9;

	if (rq->bio)
		return rq->bio->bi_size;

	return rq->data_len;
}
EXPORT_SYMBOL_GPL(blk_rq_cur_bytes);
1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797

/**
 * end_queued_request - end all I/O on a queued request
 * @rq:		the request being processed
 * @uptodate:	error value or 0/1 uptodate flag
 *
 * Description:
 *     Ends all I/O on a request, and removes it from the block layer queues.
 *     Not suitable for normal IO completion, unless the driver still has
 *     the request attached to the block layer.
 *
 **/
void end_queued_request(struct request *rq, int uptodate)
{
1798
	__end_request(rq, uptodate, blk_rq_bytes(rq));
1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814
}
EXPORT_SYMBOL(end_queued_request);

/**
 * end_dequeued_request - end all I/O on a dequeued request
 * @rq:		the request being processed
 * @uptodate:	error value or 0/1 uptodate flag
 *
 * Description:
 *     Ends all I/O on a request. The request must already have been
 *     dequeued using blkdev_dequeue_request(), as is normally the case
 *     for most drivers.
 *
 **/
void end_dequeued_request(struct request *rq, int uptodate)
{
1815
	__end_request(rq, uptodate, blk_rq_bytes(rq));
1816 1817 1818 1819 1820 1821
}
EXPORT_SYMBOL(end_dequeued_request);


/**
 * end_request - end I/O on the current segment of the request
1822
 * @req:	the request being processed
1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840
 * @uptodate:	error value or 0/1 uptodate flag
 *
 * Description:
 *     Ends I/O on the current segment of a request. If that is the only
 *     remaining segment, the request is also completed and freed.
 *
 *     This is a remnant of how older block drivers handled IO completions.
 *     Modern drivers typically end IO on the full request in one go, unless
 *     they have a residual value to account for. For that case this function
 *     isn't really useful, unless the residual just happens to be the
 *     full current segment. In other words, don't use this function in new
 *     code. Either use end_request_completely(), or the
 *     end_that_request_chunk() (along with end_that_request_last()) for
 *     partial completions.
 *
 **/
void end_request(struct request *req, int uptodate)
{
1841
	__end_request(req, uptodate, req->hard_cur_sectors << 9);
1842
}
L
Linus Torvalds 已提交
1843 1844
EXPORT_SYMBOL(end_request);

1845
/**
1846 1847 1848
 * blk_end_io - Generic end_io function to complete a request.
 * @rq:           the request being processed
 * @error:        0 for success, < 0 for error
1849 1850
 * @nr_bytes:     number of bytes to complete @rq
 * @bidi_bytes:   number of bytes to complete @rq->next_rq
1851 1852 1853 1854
 * @drv_callback: function called between completion of bios in the request
 *                and completion of the request.
 *                If the callback returns non 0, this helper returns without
 *                completion of the request.
1855 1856
 *
 * Description:
1857
 *     Ends I/O on a number of bytes attached to @rq and @rq->next_rq.
1858 1859 1860 1861
 *     If @rq has leftover, sets it up for the next range of segments.
 *
 * Return:
 *     0 - we are done with this request
1862
 *     1 - this request is not freed yet, it still has pending buffers.
1863
 **/
1864 1865 1866
static int blk_end_io(struct request *rq, int error, unsigned int nr_bytes,
		      unsigned int bidi_bytes,
		      int (drv_callback)(struct request *))
1867 1868 1869 1870 1871
{
	struct request_queue *q = rq->q;
	unsigned long flags = 0UL;

	if (blk_fs_request(rq) || blk_pc_request(rq)) {
1872
		if (__end_that_request_first(rq, error, nr_bytes))
1873
			return 1;
1874 1875 1876

		/* Bidi request must be completed as a whole */
		if (blk_bidi_rq(rq) &&
1877
		    __end_that_request_first(rq->next_rq, error, bidi_bytes))
1878
			return 1;
1879 1880
	}

1881 1882 1883 1884
	/* Special feature for tricky drivers */
	if (drv_callback && drv_callback(rq))
		return 1;

1885 1886 1887
	add_disk_randomness(rq->rq_disk);

	spin_lock_irqsave(q->queue_lock, flags);
1888
	end_that_request_last(rq, error);
1889 1890 1891 1892
	spin_unlock_irqrestore(q->queue_lock, flags);

	return 0;
}
1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907

/**
 * blk_end_request - Helper function for drivers to complete the request.
 * @rq:       the request being processed
 * @error:    0 for success, < 0 for error
 * @nr_bytes: number of bytes to complete
 *
 * Description:
 *     Ends I/O on a number of bytes attached to @rq.
 *     If @rq has leftover, sets it up for the next range of segments.
 *
 * Return:
 *     0 - we are done with this request
 *     1 - still buffers pending for this request
 **/
1908
int blk_end_request(struct request *rq, int error, unsigned int nr_bytes)
1909
{
1910
	return blk_end_io(rq, error, nr_bytes, 0, NULL);
1911
}
1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926
EXPORT_SYMBOL_GPL(blk_end_request);

/**
 * __blk_end_request - Helper function for drivers to complete the request.
 * @rq:       the request being processed
 * @error:    0 for success, < 0 for error
 * @nr_bytes: number of bytes to complete
 *
 * Description:
 *     Must be called with queue lock held unlike blk_end_request().
 *
 * Return:
 *     0 - we are done with this request
 *     1 - still buffers pending for this request
 **/
1927
int __blk_end_request(struct request *rq, int error, unsigned int nr_bytes)
1928 1929
{
	if (blk_fs_request(rq) || blk_pc_request(rq)) {
1930
		if (__end_that_request_first(rq, error, nr_bytes))
1931 1932 1933 1934 1935
			return 1;
	}

	add_disk_randomness(rq->rq_disk);

1936
	end_that_request_last(rq, error);
1937 1938 1939 1940 1941

	return 0;
}
EXPORT_SYMBOL_GPL(__blk_end_request);

1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955
/**
 * blk_end_bidi_request - Helper function for drivers to complete bidi request.
 * @rq:         the bidi request being processed
 * @error:      0 for success, < 0 for error
 * @nr_bytes:   number of bytes to complete @rq
 * @bidi_bytes: number of bytes to complete @rq->next_rq
 *
 * Description:
 *     Ends I/O on a number of bytes attached to @rq and @rq->next_rq.
 *
 * Return:
 *     0 - we are done with this request
 *     1 - still buffers pending for this request
 **/
1956 1957
int blk_end_bidi_request(struct request *rq, int error, unsigned int nr_bytes,
			 unsigned int bidi_bytes)
1958 1959 1960 1961 1962
{
	return blk_end_io(rq, error, nr_bytes, bidi_bytes, NULL);
}
EXPORT_SYMBOL_GPL(blk_end_bidi_request);

1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987
/**
 * blk_end_request_callback - Special helper function for tricky drivers
 * @rq:           the request being processed
 * @error:        0 for success, < 0 for error
 * @nr_bytes:     number of bytes to complete
 * @drv_callback: function called between completion of bios in the request
 *                and completion of the request.
 *                If the callback returns non 0, this helper returns without
 *                completion of the request.
 *
 * Description:
 *     Ends I/O on a number of bytes attached to @rq.
 *     If @rq has leftover, sets it up for the next range of segments.
 *
 *     This special helper function is used only for existing tricky drivers.
 *     (e.g. cdrom_newpc_intr() of ide-cd)
 *     This interface will be removed when such drivers are rewritten.
 *     Don't use this interface in other places anymore.
 *
 * Return:
 *     0 - we are done with this request
 *     1 - this request is not freed yet.
 *         this request still has pending buffers or
 *         the driver doesn't want to finish this request yet.
 **/
1988 1989
int blk_end_request_callback(struct request *rq, int error,
			     unsigned int nr_bytes,
1990 1991
			     int (drv_callback)(struct request *))
{
1992
	return blk_end_io(rq, error, nr_bytes, 0, drv_callback);
1993 1994 1995
}
EXPORT_SYMBOL_GPL(blk_end_request_callback);

J
Jens Axboe 已提交
1996 1997
void blk_rq_bio_prep(struct request_queue *q, struct request *rq,
		     struct bio *bio)
L
Linus Torvalds 已提交
1998
{
1999 2000
	/* first two bits are identical in rq->cmd_flags and bio->bi_rw */
	rq->cmd_flags |= (bio->bi_rw & 3);
L
Linus Torvalds 已提交
2001 2002 2003 2004 2005 2006 2007

	rq->nr_phys_segments = bio_phys_segments(q, bio);
	rq->nr_hw_segments = bio_hw_segments(q, bio);
	rq->current_nr_sectors = bio_cur_sectors(bio);
	rq->hard_cur_sectors = rq->current_nr_sectors;
	rq->hard_nr_sectors = rq->nr_sectors = bio_sectors(bio);
	rq->buffer = bio_data(bio);
2008
	rq->data_len = bio->bi_size;
L
Linus Torvalds 已提交
2009 2010 2011

	rq->bio = rq->biotail = bio;

N
NeilBrown 已提交
2012 2013 2014
	if (bio->bi_bdev)
		rq->rq_disk = bio->bi_bdev->bd_disk;
}
L
Linus Torvalds 已提交
2015 2016 2017 2018 2019 2020 2021

int kblockd_schedule_work(struct work_struct *work)
{
	return queue_work(kblockd_workqueue, work);
}
EXPORT_SYMBOL(kblockd_schedule_work);

A
Andrew Morton 已提交
2022
void kblockd_flush_work(struct work_struct *work)
L
Linus Torvalds 已提交
2023
{
2024
	cancel_work_sync(work);
L
Linus Torvalds 已提交
2025
}
A
Andrew Morton 已提交
2026
EXPORT_SYMBOL(kblockd_flush_work);
L
Linus Torvalds 已提交
2027 2028 2029

int __init blk_dev_init(void)
{
2030 2031
	int i;

L
Linus Torvalds 已提交
2032 2033 2034 2035 2036
	kblockd_workqueue = create_workqueue("kblockd");
	if (!kblockd_workqueue)
		panic("Failed to create kblockd\n");

	request_cachep = kmem_cache_create("blkdev_requests",
2037
			sizeof(struct request), 0, SLAB_PANIC, NULL);
L
Linus Torvalds 已提交
2038

2039
	blk_requestq_cachep = kmem_cache_create("blkdev_queue",
2040
			sizeof(struct request_queue), 0, SLAB_PANIC, NULL);
L
Linus Torvalds 已提交
2041

2042
	for_each_possible_cpu(i)
2043 2044
		INIT_LIST_HEAD(&per_cpu(blk_cpu_done, i));

2045
	open_softirq(BLOCK_SOFTIRQ, blk_done_softirq);
2046
	register_hotcpu_notifier(&blk_cpu_notifier);
2047

2048
	return 0;
L
Linus Torvalds 已提交
2049 2050
}