dm.c 58.6 KB
Newer Older
L
Linus Torvalds 已提交
1 2
/*
 * Copyright (C) 2001, 2002 Sistina Software (UK) Limited.
M
Milan Broz 已提交
3
 * Copyright (C) 2004-2008 Red Hat, Inc. All rights reserved.
L
Linus Torvalds 已提交
4 5 6 7 8
 *
 * This file is released under the GPL.
 */

#include "dm.h"
M
Mike Anderson 已提交
9
#include "dm-uevent.h"
L
Linus Torvalds 已提交
10 11 12

#include <linux/init.h>
#include <linux/module.h>
A
Arjan van de Ven 已提交
13
#include <linux/mutex.h>
L
Linus Torvalds 已提交
14 15 16 17
#include <linux/moduleparam.h>
#include <linux/blkpg.h>
#include <linux/bio.h>
#include <linux/buffer_head.h>
18
#include <linux/smp_lock.h>
L
Linus Torvalds 已提交
19 20 21
#include <linux/mempool.h>
#include <linux/slab.h>
#include <linux/idr.h>
D
Darrick J. Wong 已提交
22
#include <linux/hdreg.h>
23
#include <linux/delay.h>
24 25

#include <trace/events/block.h>
L
Linus Torvalds 已提交
26

27 28
#define DM_MSG_PREFIX "core"

M
Milan Broz 已提交
29 30 31 32 33 34 35
/*
 * Cookies are numeric values sent with CHANGE and REMOVE
 * uevents while resuming, removing or renaming the device.
 */
#define DM_COOKIE_ENV_VAR_NAME "DM_COOKIE"
#define DM_COOKIE_LENGTH 24

L
Linus Torvalds 已提交
36 37 38 39 40
static const char *_name = DM_NAME;

static unsigned int major = 0;
static unsigned int _major = 0;

41
static DEFINE_SPINLOCK(_minor_lock);
L
Linus Torvalds 已提交
42
/*
K
Kiyoshi Ueda 已提交
43
 * For bio-based dm.
L
Linus Torvalds 已提交
44 45 46 47 48 49
 * One of these is allocated per bio.
 */
struct dm_io {
	struct mapped_device *md;
	int error;
	atomic_t io_count;
R
Richard Kennedy 已提交
50
	struct bio *bio;
51
	unsigned long start_time;
52
	spinlock_t endio_lock;
L
Linus Torvalds 已提交
53 54 55
};

/*
K
Kiyoshi Ueda 已提交
56
 * For bio-based dm.
L
Linus Torvalds 已提交
57 58 59
 * One of these is allocated per target within a bio.  Hopefully
 * this will be simplified out one day.
 */
A
Alasdair G Kergon 已提交
60
struct dm_target_io {
L
Linus Torvalds 已提交
61 62 63 64 65
	struct dm_io *io;
	struct dm_target *ti;
	union map_info info;
};

K
Kiyoshi Ueda 已提交
66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83
/*
 * For request-based dm.
 * One of these is allocated per request.
 */
struct dm_rq_target_io {
	struct mapped_device *md;
	struct dm_target *ti;
	struct request *orig, clone;
	int error;
	union map_info info;
};

/*
 * For request-based dm.
 * One of these is allocated per bio.
 */
struct dm_rq_clone_bio_info {
	struct bio *orig;
84
	struct dm_rq_target_io *tio;
K
Kiyoshi Ueda 已提交
85 86
};

L
Linus Torvalds 已提交
87 88
union map_info *dm_get_mapinfo(struct bio *bio)
{
A
Alasdair G Kergon 已提交
89
	if (bio && bio->bi_private)
A
Alasdair G Kergon 已提交
90
		return &((struct dm_target_io *)bio->bi_private)->info;
A
Alasdair G Kergon 已提交
91
	return NULL;
L
Linus Torvalds 已提交
92 93
}

94 95 96 97 98 99 100 101
union map_info *dm_get_rq_mapinfo(struct request *rq)
{
	if (rq && rq->end_io_data)
		return &((struct dm_rq_target_io *)rq->end_io_data)->info;
	return NULL;
}
EXPORT_SYMBOL_GPL(dm_get_rq_mapinfo);

102 103
#define MINOR_ALLOCED ((void *)-1)

L
Linus Torvalds 已提交
104 105 106
/*
 * Bits for the md->flags field.
 */
107
#define DMF_BLOCK_IO_FOR_SUSPEND 0
L
Linus Torvalds 已提交
108
#define DMF_SUSPENDED 1
109
#define DMF_FROZEN 2
J
Jeff Mahoney 已提交
110
#define DMF_FREEING 3
111
#define DMF_DELETING 4
112
#define DMF_NOFLUSH_SUSPENDING 5
L
Linus Torvalds 已提交
113

114 115 116
/*
 * Work processed by per-device workqueue.
 */
L
Linus Torvalds 已提交
117
struct mapped_device {
118
	struct rw_semaphore io_lock;
119
	struct mutex suspend_lock;
L
Linus Torvalds 已提交
120 121
	rwlock_t map_lock;
	atomic_t holders;
122
	atomic_t open_count;
L
Linus Torvalds 已提交
123 124 125

	unsigned long flags;

126
	struct request_queue *queue;
127
	unsigned type;
128
	/* Protect queue and type against concurrent access. */
129 130
	struct mutex type_lock;

L
Linus Torvalds 已提交
131
	struct gendisk *disk;
M
Mike Anderson 已提交
132
	char name[16];
L
Linus Torvalds 已提交
133 134 135 136 137 138

	void *interface_ptr;

	/*
	 * A list of ios that arrived while we were suspended.
	 */
139
	atomic_t pending[2];
L
Linus Torvalds 已提交
140
	wait_queue_head_t wait;
141
	struct work_struct work;
K
Kiyoshi Ueda 已提交
142
	struct bio_list deferred;
143
	spinlock_t deferred_lock;
L
Linus Torvalds 已提交
144

145
	/*
146
	 * Processing queue (flush)
147 148 149
	 */
	struct workqueue_struct *wq;

L
Linus Torvalds 已提交
150 151 152 153 154 155 156 157 158 159 160
	/*
	 * The current mapping.
	 */
	struct dm_table *map;

	/*
	 * io objects are allocated from here.
	 */
	mempool_t *io_pool;
	mempool_t *tio_pool;

S
Stefan Bader 已提交
161 162
	struct bio_set *bs;

L
Linus Torvalds 已提交
163 164 165 166 167
	/*
	 * Event handling.
	 */
	atomic_t event_nr;
	wait_queue_head_t eventq;
M
Mike Anderson 已提交
168 169 170
	atomic_t uevent_seq;
	struct list_head uevent_list;
	spinlock_t uevent_lock; /* Protect access to uevent_list */
L
Linus Torvalds 已提交
171 172 173 174 175

	/*
	 * freeze/thaw support require holding onto a super block
	 */
	struct super_block *frozen_sb;
176
	struct block_device *bdev;
D
Darrick J. Wong 已提交
177 178 179

	/* forced geometry settings */
	struct hd_geometry geometry;
M
Milan Broz 已提交
180

181 182 183
	/* For saving the address of __make_request for request based dm */
	make_request_fn *saved_make_request_fn;

M
Milan Broz 已提交
184 185
	/* sysfs handle */
	struct kobject kobj;
186

187 188
	/* zero-length flush that will be cloned and submitted to targets */
	struct bio flush_bio;
L
Linus Torvalds 已提交
189 190
};

K
Kiyoshi Ueda 已提交
191 192 193 194 195 196 197 198 199
/*
 * For mempools pre-allocation at the table loading time.
 */
struct dm_md_mempools {
	mempool_t *io_pool;
	mempool_t *tio_pool;
	struct bio_set *bs;
};

L
Linus Torvalds 已提交
200
#define MIN_IOS 256
201 202
static struct kmem_cache *_io_cache;
static struct kmem_cache *_tio_cache;
K
Kiyoshi Ueda 已提交
203 204
static struct kmem_cache *_rq_tio_cache;
static struct kmem_cache *_rq_bio_info_cache;
L
Linus Torvalds 已提交
205 206 207

static int __init local_init(void)
{
K
Kiyoshi Ueda 已提交
208
	int r = -ENOMEM;
L
Linus Torvalds 已提交
209 210

	/* allocate a slab for the dm_ios */
A
Alasdair G Kergon 已提交
211
	_io_cache = KMEM_CACHE(dm_io, 0);
L
Linus Torvalds 已提交
212
	if (!_io_cache)
K
Kiyoshi Ueda 已提交
213
		return r;
L
Linus Torvalds 已提交
214 215

	/* allocate a slab for the target ios */
A
Alasdair G Kergon 已提交
216
	_tio_cache = KMEM_CACHE(dm_target_io, 0);
K
Kiyoshi Ueda 已提交
217 218
	if (!_tio_cache)
		goto out_free_io_cache;
L
Linus Torvalds 已提交
219

K
Kiyoshi Ueda 已提交
220 221 222 223 224 225 226 227
	_rq_tio_cache = KMEM_CACHE(dm_rq_target_io, 0);
	if (!_rq_tio_cache)
		goto out_free_tio_cache;

	_rq_bio_info_cache = KMEM_CACHE(dm_rq_clone_bio_info, 0);
	if (!_rq_bio_info_cache)
		goto out_free_rq_tio_cache;

M
Mike Anderson 已提交
228
	r = dm_uevent_init();
K
Kiyoshi Ueda 已提交
229
	if (r)
K
Kiyoshi Ueda 已提交
230
		goto out_free_rq_bio_info_cache;
M
Mike Anderson 已提交
231

L
Linus Torvalds 已提交
232 233
	_major = major;
	r = register_blkdev(_major, _name);
K
Kiyoshi Ueda 已提交
234 235
	if (r < 0)
		goto out_uevent_exit;
L
Linus Torvalds 已提交
236 237 238 239 240

	if (!_major)
		_major = r;

	return 0;
K
Kiyoshi Ueda 已提交
241 242 243

out_uevent_exit:
	dm_uevent_exit();
K
Kiyoshi Ueda 已提交
244 245 246 247
out_free_rq_bio_info_cache:
	kmem_cache_destroy(_rq_bio_info_cache);
out_free_rq_tio_cache:
	kmem_cache_destroy(_rq_tio_cache);
K
Kiyoshi Ueda 已提交
248 249 250 251 252 253
out_free_tio_cache:
	kmem_cache_destroy(_tio_cache);
out_free_io_cache:
	kmem_cache_destroy(_io_cache);

	return r;
L
Linus Torvalds 已提交
254 255 256 257
}

static void local_exit(void)
{
K
Kiyoshi Ueda 已提交
258 259
	kmem_cache_destroy(_rq_bio_info_cache);
	kmem_cache_destroy(_rq_tio_cache);
L
Linus Torvalds 已提交
260 261
	kmem_cache_destroy(_tio_cache);
	kmem_cache_destroy(_io_cache);
262
	unregister_blkdev(_major, _name);
M
Mike Anderson 已提交
263
	dm_uevent_exit();
L
Linus Torvalds 已提交
264 265 266 267 268 269

	_major = 0;

	DMINFO("cleaned up");
}

270
static int (*_inits[])(void) __initdata = {
L
Linus Torvalds 已提交
271 272 273 274
	local_init,
	dm_target_init,
	dm_linear_init,
	dm_stripe_init,
M
Mikulas Patocka 已提交
275
	dm_io_init,
276
	dm_kcopyd_init,
L
Linus Torvalds 已提交
277 278 279
	dm_interface_init,
};

280
static void (*_exits[])(void) = {
L
Linus Torvalds 已提交
281 282 283 284
	local_exit,
	dm_target_exit,
	dm_linear_exit,
	dm_stripe_exit,
M
Mikulas Patocka 已提交
285
	dm_io_exit,
286
	dm_kcopyd_exit,
L
Linus Torvalds 已提交
287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321
	dm_interface_exit,
};

static int __init dm_init(void)
{
	const int count = ARRAY_SIZE(_inits);

	int r, i;

	for (i = 0; i < count; i++) {
		r = _inits[i]();
		if (r)
			goto bad;
	}

	return 0;

      bad:
	while (i--)
		_exits[i]();

	return r;
}

static void __exit dm_exit(void)
{
	int i = ARRAY_SIZE(_exits);

	while (i--)
		_exits[i]();
}

/*
 * Block device functions
 */
M
Mike Anderson 已提交
322 323 324 325 326
int dm_deleting_md(struct mapped_device *md)
{
	return test_bit(DMF_DELETING, &md->flags);
}

A
Al Viro 已提交
327
static int dm_blk_open(struct block_device *bdev, fmode_t mode)
L
Linus Torvalds 已提交
328 329 330
{
	struct mapped_device *md;

331
	lock_kernel();
J
Jeff Mahoney 已提交
332 333
	spin_lock(&_minor_lock);

A
Al Viro 已提交
334
	md = bdev->bd_disk->private_data;
J
Jeff Mahoney 已提交
335 336 337
	if (!md)
		goto out;

338
	if (test_bit(DMF_FREEING, &md->flags) ||
M
Mike Anderson 已提交
339
	    dm_deleting_md(md)) {
J
Jeff Mahoney 已提交
340 341 342 343
		md = NULL;
		goto out;
	}

L
Linus Torvalds 已提交
344
	dm_get(md);
345
	atomic_inc(&md->open_count);
J
Jeff Mahoney 已提交
346 347 348

out:
	spin_unlock(&_minor_lock);
349
	unlock_kernel();
J
Jeff Mahoney 已提交
350 351

	return md ? 0 : -ENXIO;
L
Linus Torvalds 已提交
352 353
}

A
Al Viro 已提交
354
static int dm_blk_close(struct gendisk *disk, fmode_t mode)
L
Linus Torvalds 已提交
355
{
A
Al Viro 已提交
356
	struct mapped_device *md = disk->private_data;
357 358

	lock_kernel();
359
	atomic_dec(&md->open_count);
L
Linus Torvalds 已提交
360
	dm_put(md);
361 362
	unlock_kernel();

L
Linus Torvalds 已提交
363 364 365
	return 0;
}

366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389
int dm_open_count(struct mapped_device *md)
{
	return atomic_read(&md->open_count);
}

/*
 * Guarantees nothing is using the device before it's deleted.
 */
int dm_lock_for_deletion(struct mapped_device *md)
{
	int r = 0;

	spin_lock(&_minor_lock);

	if (dm_open_count(md))
		r = -EBUSY;
	else
		set_bit(DMF_DELETING, &md->flags);

	spin_unlock(&_minor_lock);

	return r;
}

D
Darrick J. Wong 已提交
390 391 392 393 394 395 396
static int dm_blk_getgeo(struct block_device *bdev, struct hd_geometry *geo)
{
	struct mapped_device *md = bdev->bd_disk->private_data;

	return dm_get_geometry(md, geo);
}

A
Al Viro 已提交
397
static int dm_blk_ioctl(struct block_device *bdev, fmode_t mode,
398 399
			unsigned int cmd, unsigned long arg)
{
A
Al Viro 已提交
400
	struct mapped_device *md = bdev->bd_disk->private_data;
401
	struct dm_table *map = dm_get_live_table(md);
402 403 404 405 406 407 408 409 410 411 412 413
	struct dm_target *tgt;
	int r = -ENOTTY;

	if (!map || !dm_table_get_size(map))
		goto out;

	/* We only support devices that have a single target */
	if (dm_table_get_num_targets(map) != 1)
		goto out;

	tgt = dm_table_get_target(map, 0);

414
	if (dm_suspended_md(md)) {
415 416 417 418 419
		r = -EAGAIN;
		goto out;
	}

	if (tgt->type->ioctl)
420
		r = tgt->type->ioctl(tgt, cmd, arg);
421 422 423 424 425 426 427

out:
	dm_table_put(map);

	return r;
}

A
Alasdair G Kergon 已提交
428
static struct dm_io *alloc_io(struct mapped_device *md)
L
Linus Torvalds 已提交
429 430 431 432
{
	return mempool_alloc(md->io_pool, GFP_NOIO);
}

A
Alasdair G Kergon 已提交
433
static void free_io(struct mapped_device *md, struct dm_io *io)
L
Linus Torvalds 已提交
434 435 436 437
{
	mempool_free(io, md->io_pool);
}

A
Alasdair G Kergon 已提交
438
static void free_tio(struct mapped_device *md, struct dm_target_io *tio)
L
Linus Torvalds 已提交
439 440 441 442
{
	mempool_free(tio, md->tio_pool);
}

K
Kiyoshi Ueda 已提交
443 444
static struct dm_rq_target_io *alloc_rq_tio(struct mapped_device *md,
					    gfp_t gfp_mask)
445
{
K
Kiyoshi Ueda 已提交
446
	return mempool_alloc(md->tio_pool, gfp_mask);
447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463
}

static void free_rq_tio(struct dm_rq_target_io *tio)
{
	mempool_free(tio, tio->md->tio_pool);
}

static struct dm_rq_clone_bio_info *alloc_bio_info(struct mapped_device *md)
{
	return mempool_alloc(md->io_pool, GFP_ATOMIC);
}

static void free_bio_info(struct dm_rq_clone_bio_info *info)
{
	mempool_free(info, info->tio->md->io_pool);
}

K
Kiyoshi Ueda 已提交
464 465 466 467 468 469
static int md_in_flight(struct mapped_device *md)
{
	return atomic_read(&md->pending[READ]) +
	       atomic_read(&md->pending[WRITE]);
}

470 471 472
static void start_io_acct(struct dm_io *io)
{
	struct mapped_device *md = io->md;
T
Tejun Heo 已提交
473
	int cpu;
474
	int rw = bio_data_dir(io->bio);
475 476 477

	io->start_time = jiffies;

T
Tejun Heo 已提交
478 479 480
	cpu = part_stat_lock();
	part_round_stats(cpu, &dm_disk(md)->part0);
	part_stat_unlock();
481
	dm_disk(md)->part0.in_flight[rw] = atomic_inc_return(&md->pending[rw]);
482 483
}

484
static void end_io_acct(struct dm_io *io)
485 486 487 488
{
	struct mapped_device *md = io->md;
	struct bio *bio = io->bio;
	unsigned long duration = jiffies - io->start_time;
T
Tejun Heo 已提交
489
	int pending, cpu;
490 491
	int rw = bio_data_dir(bio);

T
Tejun Heo 已提交
492 493 494 495
	cpu = part_stat_lock();
	part_round_stats(cpu, &dm_disk(md)->part0);
	part_stat_add(cpu, &dm_disk(md)->part0, ticks[rw], duration);
	part_stat_unlock();
496

497 498
	/*
	 * After this is decremented the bio must not be touched if it is
499
	 * a flush.
500
	 */
501 502 503
	dm_disk(md)->part0.in_flight[rw] = pending =
		atomic_dec_return(&md->pending[rw]);
	pending += atomic_read(&md->pending[rw^0x1]);
504

505 506 507
	/* nudge anyone waiting on suspend queue */
	if (!pending)
		wake_up(&md->wait);
508 509
}

L
Linus Torvalds 已提交
510 511 512
/*
 * Add the bio to the list of deferred io.
 */
M
Mikulas Patocka 已提交
513
static void queue_io(struct mapped_device *md, struct bio *bio)
L
Linus Torvalds 已提交
514
{
515 516 517
	unsigned long flags;

	spin_lock_irqsave(&md->deferred_lock, flags);
L
Linus Torvalds 已提交
518
	bio_list_add(&md->deferred, bio);
519
	spin_unlock_irqrestore(&md->deferred_lock, flags);
520
	queue_work(md->wq, &md->work);
L
Linus Torvalds 已提交
521 522 523 524 525 526 527
}

/*
 * Everyone (including functions in this file), should use this
 * function to access the md->map field, and make sure they call
 * dm_table_put() when finished.
 */
528
struct dm_table *dm_get_live_table(struct mapped_device *md)
L
Linus Torvalds 已提交
529 530
{
	struct dm_table *t;
531
	unsigned long flags;
L
Linus Torvalds 已提交
532

533
	read_lock_irqsave(&md->map_lock, flags);
L
Linus Torvalds 已提交
534 535 536
	t = md->map;
	if (t)
		dm_table_get(t);
537
	read_unlock_irqrestore(&md->map_lock, flags);
L
Linus Torvalds 已提交
538 539 540 541

	return t;
}

D
Darrick J. Wong 已提交
542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568
/*
 * Get the geometry associated with a dm device
 */
int dm_get_geometry(struct mapped_device *md, struct hd_geometry *geo)
{
	*geo = md->geometry;

	return 0;
}

/*
 * Set the geometry of a device.
 */
int dm_set_geometry(struct mapped_device *md, struct hd_geometry *geo)
{
	sector_t sz = (sector_t)geo->cylinders * geo->heads * geo->sectors;

	if (geo->start > sz) {
		DMWARN("Start sector is beyond the geometry limits.");
		return -EINVAL;
	}

	md->geometry = *geo;

	return 0;
}

L
Linus Torvalds 已提交
569 570 571 572 573 574 575 576 577
/*-----------------------------------------------------------------
 * CRUD START:
 *   A more elegant soln is in the works that uses the queue
 *   merge fn, unfortunately there are a couple of changes to
 *   the block layer that I want to make for this.  So in the
 *   interests of getting something for people to use I give
 *   you this clearly demarcated crap.
 *---------------------------------------------------------------*/

578 579 580 581 582
static int __noflush_suspending(struct mapped_device *md)
{
	return test_bit(DMF_NOFLUSH_SUSPENDING, &md->flags);
}

L
Linus Torvalds 已提交
583 584 585 586
/*
 * Decrements the number of outstanding ios that a bio has been
 * cloned into, completing the original io if necc.
 */
587
static void dec_pending(struct dm_io *io, int error)
L
Linus Torvalds 已提交
588
{
589
	unsigned long flags;
590 591 592
	int io_error;
	struct bio *bio;
	struct mapped_device *md = io->md;
593 594

	/* Push-back supersedes any I/O errors */
595 596 597 598 599 600
	if (unlikely(error)) {
		spin_lock_irqsave(&io->endio_lock, flags);
		if (!(io->error > 0 && __noflush_suspending(md)))
			io->error = error;
		spin_unlock_irqrestore(&io->endio_lock, flags);
	}
L
Linus Torvalds 已提交
601 602

	if (atomic_dec_and_test(&io->io_count)) {
603 604 605 606
		if (io->error == DM_ENDIO_REQUEUE) {
			/*
			 * Target requested pushing back the I/O.
			 */
607
			spin_lock_irqsave(&md->deferred_lock, flags);
608 609 610
			if (__noflush_suspending(md))
				bio_list_add_head(&md->deferred, io->bio);
			else
611 612
				/* noflush suspend was interrupted. */
				io->error = -EIO;
613
			spin_unlock_irqrestore(&md->deferred_lock, flags);
614 615
		}

616 617
		io_error = io->error;
		bio = io->bio;
618 619 620 621 622
		end_io_acct(io);
		free_io(md, io);

		if (io_error == DM_ENDIO_REQUEUE)
			return;
623

624 625 626 627
		if (!(bio->bi_rw & REQ_FLUSH) || !bio->bi_size) {
			trace_block_bio_complete(md->queue, bio);
			bio_endio(bio, io_error);
		} else {
628
			/*
629 630
			 * Preflush done for flush with data, reissue
			 * without REQ_FLUSH.
631
			 */
632 633
			bio->bi_rw &= ~REQ_FLUSH;
			queue_io(md, bio);
634
		}
L
Linus Torvalds 已提交
635 636 637
	}
}

638
static void clone_endio(struct bio *bio, int error)
L
Linus Torvalds 已提交
639 640
{
	int r = 0;
A
Alasdair G Kergon 已提交
641
	struct dm_target_io *tio = bio->bi_private;
642
	struct dm_io *io = tio->io;
S
Stefan Bader 已提交
643
	struct mapped_device *md = tio->io->md;
L
Linus Torvalds 已提交
644 645 646 647 648 649 650
	dm_endio_fn endio = tio->ti->type->end_io;

	if (!bio_flagged(bio, BIO_UPTODATE) && !error)
		error = -EIO;

	if (endio) {
		r = endio(tio->ti, bio, error, &tio->info);
651 652 653 654 655
		if (r < 0 || r == DM_ENDIO_REQUEUE)
			/*
			 * error and requeue request are handled
			 * in dec_pending().
			 */
L
Linus Torvalds 已提交
656
			error = r;
657 658
		else if (r == DM_ENDIO_INCOMPLETE)
			/* The target will handle the io */
659
			return;
660 661 662 663
		else if (r) {
			DMWARN("unimplemented target endio return value: %d", r);
			BUG();
		}
L
Linus Torvalds 已提交
664 665
	}

S
Stefan Bader 已提交
666 667 668 669 670 671
	/*
	 * Store md for cleanup instead of tio which is about to get freed.
	 */
	bio->bi_private = md->bs;

	free_tio(md, tio);
672 673
	bio_put(bio);
	dec_pending(io, error);
L
Linus Torvalds 已提交
674 675
}

676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730
/*
 * Partial completion handling for request-based dm
 */
static void end_clone_bio(struct bio *clone, int error)
{
	struct dm_rq_clone_bio_info *info = clone->bi_private;
	struct dm_rq_target_io *tio = info->tio;
	struct bio *bio = info->orig;
	unsigned int nr_bytes = info->orig->bi_size;

	bio_put(clone);

	if (tio->error)
		/*
		 * An error has already been detected on the request.
		 * Once error occurred, just let clone->end_io() handle
		 * the remainder.
		 */
		return;
	else if (error) {
		/*
		 * Don't notice the error to the upper layer yet.
		 * The error handling decision is made by the target driver,
		 * when the request is completed.
		 */
		tio->error = error;
		return;
	}

	/*
	 * I/O for the bio successfully completed.
	 * Notice the data completion to the upper layer.
	 */

	/*
	 * bios are processed from the head of the list.
	 * So the completing bio should always be rq->bio.
	 * If it's not, something wrong is happening.
	 */
	if (tio->orig->bio != bio)
		DMERR("bio completion is going in the middle of the request");

	/*
	 * Update the original request.
	 * Do not use blk_end_request() here, because it may complete
	 * the original request before the clone, and break the ordering.
	 */
	blk_update_request(tio->orig, 0, nr_bytes);
}

/*
 * Don't touch any member of the md after calling this function because
 * the md may be freed in dm_put() at the end of this function.
 * Or do dm_get() before calling this function and dm_put() later.
 */
731
static void rq_completed(struct mapped_device *md, int rw, int run_queue)
732
{
733
	atomic_dec(&md->pending[rw]);
734 735

	/* nudge anyone waiting on suspend queue */
736
	if (!md_in_flight(md))
737 738 739
		wake_up(&md->wait);

	if (run_queue)
740
		blk_run_queue(md->queue);
741 742 743 744 745 746 747

	/*
	 * dm_put() must be at the end of this function. See the comment above
	 */
	dm_put(md);
}

748 749 750 751 752 753 754 755
static void free_rq_clone(struct request *clone)
{
	struct dm_rq_target_io *tio = clone->end_io_data;

	blk_rq_unprep_clone(clone);
	free_rq_tio(tio);
}

K
Kiyoshi Ueda 已提交
756 757 758 759 760 761 762 763 764 765 766
/*
 * Complete the clone and the original request.
 * Must be called without queue lock.
 */
static void dm_end_request(struct request *clone, int error)
{
	int rw = rq_data_dir(clone);
	struct dm_rq_target_io *tio = clone->end_io_data;
	struct mapped_device *md = tio->md;
	struct request *rq = tio->orig;

767
	if (rq->cmd_type == REQ_TYPE_BLOCK_PC) {
K
Kiyoshi Ueda 已提交
768 769 770 771 772 773 774 775 776 777 778 779 780
		rq->errors = clone->errors;
		rq->resid_len = clone->resid_len;

		if (rq->sense)
			/*
			 * We are using the sense buffer of the original
			 * request.
			 * So setting the length of the sense data is enough.
			 */
			rq->sense_len = clone->sense_len;
	}

	free_rq_clone(clone);
781 782
	blk_end_request_all(rq, error);
	rq_completed(md, rw, true);
K
Kiyoshi Ueda 已提交
783 784
}

785 786 787 788 789 790 791
static void dm_unprep_request(struct request *rq)
{
	struct request *clone = rq->special;

	rq->special = NULL;
	rq->cmd_flags &= ~REQ_DONTPREP;

792
	free_rq_clone(clone);
793 794 795 796 797 798 799
}

/*
 * Requeue the original request of a clone.
 */
void dm_requeue_unmapped_request(struct request *clone)
{
800
	int rw = rq_data_dir(clone);
801 802 803 804 805 806 807 808 809 810 811 812 813 814
	struct dm_rq_target_io *tio = clone->end_io_data;
	struct mapped_device *md = tio->md;
	struct request *rq = tio->orig;
	struct request_queue *q = rq->q;
	unsigned long flags;

	dm_unprep_request(rq);

	spin_lock_irqsave(q->queue_lock, flags);
	if (elv_queue_empty(q))
		blk_plug_device(q);
	blk_requeue_request(q, rq);
	spin_unlock_irqrestore(q->queue_lock, flags);

815
	rq_completed(md, rw, 0);
816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847
}
EXPORT_SYMBOL_GPL(dm_requeue_unmapped_request);

static void __stop_queue(struct request_queue *q)
{
	blk_stop_queue(q);
}

static void stop_queue(struct request_queue *q)
{
	unsigned long flags;

	spin_lock_irqsave(q->queue_lock, flags);
	__stop_queue(q);
	spin_unlock_irqrestore(q->queue_lock, flags);
}

static void __start_queue(struct request_queue *q)
{
	if (blk_queue_stopped(q))
		blk_start_queue(q);
}

static void start_queue(struct request_queue *q)
{
	unsigned long flags;

	spin_lock_irqsave(q->queue_lock, flags);
	__start_queue(q);
	spin_unlock_irqrestore(q->queue_lock, flags);
}

848
static void dm_done(struct request *clone, int error, bool mapped)
849
{
850
	int r = error;
851 852 853
	struct dm_rq_target_io *tio = clone->end_io_data;
	dm_request_endio_fn rq_end_io = tio->ti->type->rq_end_io;

854 855
	if (mapped && rq_end_io)
		r = rq_end_io(tio->ti, clone, error, &tio->info);
856

857
	if (r <= 0)
858
		/* The target wants to complete the I/O */
859 860
		dm_end_request(clone, r);
	else if (r == DM_ENDIO_INCOMPLETE)
861 862
		/* The target will handle the I/O */
		return;
863
	else if (r == DM_ENDIO_REQUEUE)
864 865 866
		/* The target wants to requeue the I/O */
		dm_requeue_unmapped_request(clone);
	else {
867
		DMWARN("unimplemented target endio return value: %d", r);
868 869 870 871
		BUG();
	}
}

872 873 874 875 876 877 878 879 880 881 882 883 884 885 886
/*
 * Request completion handler for request-based dm
 */
static void dm_softirq_done(struct request *rq)
{
	bool mapped = true;
	struct request *clone = rq->completion_data;
	struct dm_rq_target_io *tio = clone->end_io_data;

	if (rq->cmd_flags & REQ_FAILED)
		mapped = false;

	dm_done(clone, tio->error, mapped);
}

887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940
/*
 * Complete the clone and the original request with the error status
 * through softirq context.
 */
static void dm_complete_request(struct request *clone, int error)
{
	struct dm_rq_target_io *tio = clone->end_io_data;
	struct request *rq = tio->orig;

	tio->error = error;
	rq->completion_data = clone;
	blk_complete_request(rq);
}

/*
 * Complete the not-mapped clone and the original request with the error status
 * through softirq context.
 * Target's rq_end_io() function isn't called.
 * This may be used when the target's map_rq() function fails.
 */
void dm_kill_unmapped_request(struct request *clone, int error)
{
	struct dm_rq_target_io *tio = clone->end_io_data;
	struct request *rq = tio->orig;

	rq->cmd_flags |= REQ_FAILED;
	dm_complete_request(clone, error);
}
EXPORT_SYMBOL_GPL(dm_kill_unmapped_request);

/*
 * Called with the queue lock held
 */
static void end_clone_request(struct request *clone, int error)
{
	/*
	 * For just cleaning up the information of the queue in which
	 * the clone was dispatched.
	 * The clone is *NOT* freed actually here because it is alloced from
	 * dm own mempool and REQ_ALLOCED isn't set in clone->cmd_flags.
	 */
	__blk_put_request(clone->q, clone);

	/*
	 * Actual request completion is done in a softirq context which doesn't
	 * hold the queue lock.  Otherwise, deadlock could occur because:
	 *     - another request may be submitted by the upper level driver
	 *       of the stacking during the completion
	 *     - the submission which requires queue lock may be done
	 *       against this queue
	 */
	dm_complete_request(clone, error);
}

941 942 943 944 945 946 947 948 949 950 951 952
/*
 * Return maximum size of I/O possible at the supplied sector up to the current
 * target boundary.
 */
static sector_t max_io_len_target_boundary(sector_t sector, struct dm_target *ti)
{
	sector_t target_offset = dm_target_offset(ti, sector);

	return ti->len - target_offset;
}

static sector_t max_io_len(sector_t sector, struct dm_target *ti)
L
Linus Torvalds 已提交
953
{
954
	sector_t len = max_io_len_target_boundary(sector, ti);
L
Linus Torvalds 已提交
955 956 957 958 959 960

	/*
	 * Does the target need to split even further ?
	 */
	if (ti->split_io) {
		sector_t boundary;
961
		sector_t offset = dm_target_offset(ti, sector);
L
Linus Torvalds 已提交
962 963 964 965 966 967 968 969 970 971
		boundary = ((offset + ti->split_io) & ~(ti->split_io - 1))
			   - offset;
		if (len > boundary)
			len = boundary;
	}

	return len;
}

static void __map_bio(struct dm_target *ti, struct bio *clone,
A
Alasdair G Kergon 已提交
972
		      struct dm_target_io *tio)
L
Linus Torvalds 已提交
973 974
{
	int r;
975
	sector_t sector;
S
Stefan Bader 已提交
976
	struct mapped_device *md;
L
Linus Torvalds 已提交
977 978 979 980 981 982 983 984 985 986

	clone->bi_end_io = clone_endio;
	clone->bi_private = tio;

	/*
	 * Map the clone.  If r == 0 we don't need to do
	 * anything, the target has assumed ownership of
	 * this io.
	 */
	atomic_inc(&tio->io->io_count);
987
	sector = clone->bi_sector;
L
Linus Torvalds 已提交
988
	r = ti->type->map(ti, clone, &tio->info);
989
	if (r == DM_MAPIO_REMAPPED) {
L
Linus Torvalds 已提交
990
		/* the bio has been remapped so dispatch it */
991

992
		trace_block_remap(bdev_get_queue(clone->bi_bdev), clone,
993
				    tio->io->bio->bi_bdev->bd_dev, sector);
994

L
Linus Torvalds 已提交
995
		generic_make_request(clone);
996 997
	} else if (r < 0 || r == DM_MAPIO_REQUEUE) {
		/* error the io and bail out, or requeue it if needed */
S
Stefan Bader 已提交
998 999 1000 1001 1002 1003
		md = tio->io->md;
		dec_pending(tio->io, r);
		/*
		 * Store bio_set for cleanup.
		 */
		clone->bi_private = md->bs;
L
Linus Torvalds 已提交
1004
		bio_put(clone);
S
Stefan Bader 已提交
1005
		free_tio(md, tio);
1006 1007 1008
	} else if (r) {
		DMWARN("unimplemented target map return value: %d", r);
		BUG();
L
Linus Torvalds 已提交
1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021
	}
}

struct clone_info {
	struct mapped_device *md;
	struct dm_table *map;
	struct bio *bio;
	struct dm_io *io;
	sector_t sector;
	sector_t sector_count;
	unsigned short idx;
};

P
Peter Osterlund 已提交
1022 1023
static void dm_bio_destructor(struct bio *bio)
{
S
Stefan Bader 已提交
1024 1025 1026
	struct bio_set *bs = bio->bi_private;

	bio_free(bio, bs);
P
Peter Osterlund 已提交
1027 1028
}

L
Linus Torvalds 已提交
1029
/*
1030
 * Creates a little bio that just does part of a bvec.
L
Linus Torvalds 已提交
1031 1032 1033
 */
static struct bio *split_bvec(struct bio *bio, sector_t sector,
			      unsigned short idx, unsigned int offset,
S
Stefan Bader 已提交
1034
			      unsigned int len, struct bio_set *bs)
L
Linus Torvalds 已提交
1035 1036 1037 1038
{
	struct bio *clone;
	struct bio_vec *bv = bio->bi_io_vec + idx;

S
Stefan Bader 已提交
1039
	clone = bio_alloc_bioset(GFP_NOIO, 1, bs);
P
Peter Osterlund 已提交
1040
	clone->bi_destructor = dm_bio_destructor;
L
Linus Torvalds 已提交
1041 1042 1043 1044
	*clone->bi_io_vec = *bv;

	clone->bi_sector = sector;
	clone->bi_bdev = bio->bi_bdev;
1045
	clone->bi_rw = bio->bi_rw;
L
Linus Torvalds 已提交
1046 1047 1048 1049
	clone->bi_vcnt = 1;
	clone->bi_size = to_bytes(len);
	clone->bi_io_vec->bv_offset = offset;
	clone->bi_io_vec->bv_len = clone->bi_size;
M
Martin K. Petersen 已提交
1050
	clone->bi_flags |= 1 << BIO_CLONED;
L
Linus Torvalds 已提交
1051

M
Martin K. Petersen 已提交
1052
	if (bio_integrity(bio)) {
1053
		bio_integrity_clone(clone, bio, GFP_NOIO, bs);
M
Martin K. Petersen 已提交
1054 1055 1056 1057
		bio_integrity_trim(clone,
				   bio_sector_offset(bio, idx, offset), len);
	}

L
Linus Torvalds 已提交
1058 1059 1060 1061 1062 1063 1064 1065
	return clone;
}

/*
 * Creates a bio that consists of range of complete bvecs.
 */
static struct bio *clone_bio(struct bio *bio, sector_t sector,
			     unsigned short idx, unsigned short bv_count,
S
Stefan Bader 已提交
1066
			     unsigned int len, struct bio_set *bs)
L
Linus Torvalds 已提交
1067 1068 1069
{
	struct bio *clone;

S
Stefan Bader 已提交
1070 1071 1072
	clone = bio_alloc_bioset(GFP_NOIO, bio->bi_max_vecs, bs);
	__bio_clone(clone, bio);
	clone->bi_destructor = dm_bio_destructor;
L
Linus Torvalds 已提交
1073 1074 1075 1076 1077 1078
	clone->bi_sector = sector;
	clone->bi_idx = idx;
	clone->bi_vcnt = idx + bv_count;
	clone->bi_size = to_bytes(len);
	clone->bi_flags &= ~(1 << BIO_SEG_VALID);

M
Martin K. Petersen 已提交
1079
	if (bio_integrity(bio)) {
1080
		bio_integrity_clone(clone, bio, GFP_NOIO, bs);
M
Martin K. Petersen 已提交
1081 1082 1083 1084 1085 1086

		if (idx != bio->bi_idx || clone->bi_size < bio->bi_size)
			bio_integrity_trim(clone,
					   bio_sector_offset(bio, idx, 0), len);
	}

L
Linus Torvalds 已提交
1087 1088 1089
	return clone;
}

1090 1091
static struct dm_target_io *alloc_tio(struct clone_info *ci,
				      struct dm_target *ti)
1092
{
1093
	struct dm_target_io *tio = mempool_alloc(ci->md->tio_pool, GFP_NOIO);
1094 1095 1096 1097

	tio->io = ci->io;
	tio->ti = ti;
	memset(&tio->info, 0, sizeof(tio->info));
1098 1099 1100 1101

	return tio;
}

1102
static void __issue_target_request(struct clone_info *ci, struct dm_target *ti,
1103
				   unsigned request_nr, sector_t len)
1104 1105 1106 1107
{
	struct dm_target_io *tio = alloc_tio(ci, ti);
	struct bio *clone;

1108
	tio->info.target_request_nr = request_nr;
1109

1110 1111 1112 1113 1114 1115
	/*
	 * Discard requests require the bio's inline iovecs be initialized.
	 * ci->bio->bi_max_vecs is BIO_INLINE_VECS anyway, for both flush
	 * and discard, so no need for concern about wasted bvec allocations.
	 */
	clone = bio_alloc_bioset(GFP_NOIO, ci->bio->bi_max_vecs, ci->md->bs);
1116 1117
	__bio_clone(clone, ci->bio);
	clone->bi_destructor = dm_bio_destructor;
1118 1119 1120 1121
	if (len) {
		clone->bi_sector = ci->sector;
		clone->bi_size = to_bytes(len);
	}
1122 1123 1124 1125

	__map_bio(ti, clone, tio);
}

1126
static void __issue_target_requests(struct clone_info *ci, struct dm_target *ti,
1127
				    unsigned num_requests, sector_t len)
1128 1129 1130 1131
{
	unsigned request_nr;

	for (request_nr = 0; request_nr < num_requests; request_nr++)
1132
		__issue_target_request(ci, ti, request_nr, len);
1133 1134
}

1135
static int __clone_and_map_flush(struct clone_info *ci)
1136
{
1137
	unsigned target_nr = 0;
1138 1139 1140
	struct dm_target *ti;

	while ((ti = dm_table_get_target(ci->map, target_nr++)))
1141
		__issue_target_requests(ci, ti, ti->num_flush_requests, 0);
1142 1143 1144 1145 1146 1147

	ci->sector_count = 0;

	return 0;
}

M
Mike Snitzer 已提交
1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166
/*
 * Perform all io with a single clone.
 */
static void __clone_and_map_simple(struct clone_info *ci, struct dm_target *ti)
{
	struct bio *clone, *bio = ci->bio;
	struct dm_target_io *tio;

	tio = alloc_tio(ci, ti);
	clone = clone_bio(bio, ci->sector, ci->idx,
			  bio->bi_vcnt - ci->idx, ci->sector_count,
			  ci->md->bs);
	__map_bio(ti, clone, tio);
	ci->sector_count = 0;
}

static int __clone_and_map_discard(struct clone_info *ci)
{
	struct dm_target *ti;
1167
	sector_t len;
M
Mike Snitzer 已提交
1168

1169 1170 1171 1172
	do {
		ti = dm_table_find_target(ci->map, ci->sector);
		if (!dm_target_is_valid(ti))
			return -EIO;
M
Mike Snitzer 已提交
1173 1174

		/*
1175 1176 1177
		 * Even though the device advertised discard support,
		 * reconfiguration might have changed that since the
		 * check was performed.
M
Mike Snitzer 已提交
1178
		 */
1179 1180
		if (!ti->num_discard_requests)
			return -EOPNOTSUPP;
M
Mike Snitzer 已提交
1181

1182
		len = min(ci->sector_count, max_io_len_target_boundary(ci->sector, ti));
1183

1184 1185 1186 1187
		__issue_target_requests(ci, ti, ti->num_discard_requests, len);

		ci->sector += len;
	} while (ci->sector_count -= len);
M
Mike Snitzer 已提交
1188 1189 1190 1191

	return 0;
}

1192
static int __clone_and_map(struct clone_info *ci)
L
Linus Torvalds 已提交
1193 1194
{
	struct bio *clone, *bio = ci->bio;
1195 1196
	struct dm_target *ti;
	sector_t len = 0, max;
A
Alasdair G Kergon 已提交
1197
	struct dm_target_io *tio;
L
Linus Torvalds 已提交
1198

M
Mike Snitzer 已提交
1199 1200 1201
	if (unlikely(bio->bi_rw & REQ_DISCARD))
		return __clone_and_map_discard(ci);

1202 1203 1204 1205
	ti = dm_table_find_target(ci->map, ci->sector);
	if (!dm_target_is_valid(ti))
		return -EIO;

1206
	max = max_io_len(ci->sector, ti);
1207

L
Linus Torvalds 已提交
1208 1209 1210 1211 1212
	if (ci->sector_count <= max) {
		/*
		 * Optimise for the simple case where we can do all of
		 * the remaining io with a single clone.
		 */
M
Mike Snitzer 已提交
1213
		__clone_and_map_simple(ci, ti);
L
Linus Torvalds 已提交
1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233

	} else if (to_sector(bio->bi_io_vec[ci->idx].bv_len) <= max) {
		/*
		 * There are some bvecs that don't span targets.
		 * Do as many of these as possible.
		 */
		int i;
		sector_t remaining = max;
		sector_t bv_len;

		for (i = ci->idx; remaining && (i < bio->bi_vcnt); i++) {
			bv_len = to_sector(bio->bi_io_vec[i].bv_len);

			if (bv_len > remaining)
				break;

			remaining -= bv_len;
			len += bv_len;
		}

M
Mike Snitzer 已提交
1234
		tio = alloc_tio(ci, ti);
S
Stefan Bader 已提交
1235 1236
		clone = clone_bio(bio, ci->sector, ci->idx, i - ci->idx, len,
				  ci->md->bs);
L
Linus Torvalds 已提交
1237 1238 1239 1240 1241 1242 1243 1244
		__map_bio(ti, clone, tio);

		ci->sector += len;
		ci->sector_count -= len;
		ci->idx = i;

	} else {
		/*
1245
		 * Handle a bvec that must be split between two or more targets.
L
Linus Torvalds 已提交
1246 1247
		 */
		struct bio_vec *bv = bio->bi_io_vec + ci->idx;
1248 1249
		sector_t remaining = to_sector(bv->bv_len);
		unsigned int offset = 0;
L
Linus Torvalds 已提交
1250

1251 1252 1253
		do {
			if (offset) {
				ti = dm_table_find_target(ci->map, ci->sector);
1254 1255 1256
				if (!dm_target_is_valid(ti))
					return -EIO;

1257
				max = max_io_len(ci->sector, ti);
1258 1259 1260 1261
			}

			len = min(remaining, max);

M
Mike Snitzer 已提交
1262
			tio = alloc_tio(ci, ti);
1263
			clone = split_bvec(bio, ci->sector, ci->idx,
S
Stefan Bader 已提交
1264 1265
					   bv->bv_offset + offset, len,
					   ci->md->bs);
1266 1267 1268 1269 1270 1271 1272

			__map_bio(ti, clone, tio);

			ci->sector += len;
			ci->sector_count -= len;
			offset += to_bytes(len);
		} while (remaining -= len);
L
Linus Torvalds 已提交
1273 1274 1275

		ci->idx++;
	}
1276 1277

	return 0;
L
Linus Torvalds 已提交
1278 1279 1280
}

/*
M
Mikulas Patocka 已提交
1281
 * Split the bio into several clones and submit it to targets.
L
Linus Torvalds 已提交
1282
 */
1283
static void __split_and_process_bio(struct mapped_device *md, struct bio *bio)
L
Linus Torvalds 已提交
1284
{
1285
	bool is_flush = bio->bi_rw & REQ_FLUSH;
L
Linus Torvalds 已提交
1286
	struct clone_info ci;
1287
	int error = 0;
L
Linus Torvalds 已提交
1288

1289
	ci.map = dm_get_live_table(md);
1290
	if (unlikely(!ci.map)) {
1291
		bio_io_error(bio);
1292 1293
		return;
	}
1294

L
Linus Torvalds 已提交
1295 1296 1297 1298 1299 1300
	ci.md = md;
	ci.io = alloc_io(md);
	ci.io->error = 0;
	atomic_set(&ci.io->io_count, 1);
	ci.io->bio = bio;
	ci.io->md = md;
1301
	spin_lock_init(&ci.io->endio_lock);
L
Linus Torvalds 已提交
1302
	ci.sector = bio->bi_sector;
1303 1304 1305 1306
	ci.idx = bio->bi_idx;

	if (!is_flush) {
		ci.bio = bio;
1307
		ci.sector_count = bio_sectors(bio);
1308 1309
	} else {
		ci.bio = &ci.md->flush_bio;
1310
		ci.sector_count = 1;
1311
	}
L
Linus Torvalds 已提交
1312

1313
	start_io_acct(ci.io);
1314
	while (ci.sector_count && !error) {
1315
		if (!is_flush)
1316 1317 1318 1319
			error = __clone_and_map(&ci);
		else
			error = __clone_and_map_flush(&ci);
	}
L
Linus Torvalds 已提交
1320 1321

	/* drop the extra reference count */
1322
	dec_pending(ci.io, error);
L
Linus Torvalds 已提交
1323 1324 1325 1326 1327 1328
	dm_table_put(ci.map);
}
/*-----------------------------------------------------------------
 * CRUD END
 *---------------------------------------------------------------*/

M
Milan Broz 已提交
1329 1330 1331 1332 1333
static int dm_merge_bvec(struct request_queue *q,
			 struct bvec_merge_data *bvm,
			 struct bio_vec *biovec)
{
	struct mapped_device *md = q->queuedata;
1334
	struct dm_table *map = dm_get_live_table(md);
M
Milan Broz 已提交
1335 1336
	struct dm_target *ti;
	sector_t max_sectors;
1337
	int max_size = 0;
M
Milan Broz 已提交
1338 1339

	if (unlikely(!map))
1340
		goto out;
M
Milan Broz 已提交
1341 1342

	ti = dm_table_find_target(map, bvm->bi_sector);
1343 1344
	if (!dm_target_is_valid(ti))
		goto out_table;
M
Milan Broz 已提交
1345 1346 1347 1348

	/*
	 * Find maximum amount of I/O that won't need splitting
	 */
1349
	max_sectors = min(max_io_len(bvm->bi_sector, ti),
M
Milan Broz 已提交
1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361
			  (sector_t) BIO_MAX_SECTORS);
	max_size = (max_sectors << SECTOR_SHIFT) - bvm->bi_size;
	if (max_size < 0)
		max_size = 0;

	/*
	 * merge_bvec_fn() returns number of bytes
	 * it can accept at this offset
	 * max is precomputed maximal io size
	 */
	if (max_size && ti->type->merge)
		max_size = ti->type->merge(ti, bvm, biovec, max_size);
1362 1363 1364 1365 1366 1367 1368 1369 1370 1371
	/*
	 * If the target doesn't support merge method and some of the devices
	 * provided their merge_bvec method (we know this by looking at
	 * queue_max_hw_sectors), then we can't allow bios with multiple vector
	 * entries.  So always set max_size to 0, and the code below allows
	 * just one page.
	 */
	else if (queue_max_hw_sectors(q) <= PAGE_SIZE >> 9)

		max_size = 0;
M
Milan Broz 已提交
1372

1373
out_table:
1374 1375 1376
	dm_table_put(map);

out:
M
Milan Broz 已提交
1377 1378 1379 1380 1381 1382 1383 1384 1385
	/*
	 * Always allow an entire first page
	 */
	if (max_size <= biovec->bv_len && !(bvm->bi_size >> SECTOR_SHIFT))
		max_size = biovec->bv_len;

	return max_size;
}

L
Linus Torvalds 已提交
1386 1387 1388 1389
/*
 * The request function that just remaps the bio built up by
 * dm_merge_bvec.
 */
1390
static int _dm_request(struct request_queue *q, struct bio *bio)
L
Linus Torvalds 已提交
1391
{
1392
	int rw = bio_data_dir(bio);
L
Linus Torvalds 已提交
1393
	struct mapped_device *md = q->queuedata;
T
Tejun Heo 已提交
1394
	int cpu;
L
Linus Torvalds 已提交
1395

1396
	down_read(&md->io_lock);
L
Linus Torvalds 已提交
1397

T
Tejun Heo 已提交
1398 1399 1400 1401
	cpu = part_stat_lock();
	part_stat_inc(cpu, &dm_disk(md)->part0, ios[rw]);
	part_stat_add(cpu, &dm_disk(md)->part0, sectors[rw], bio_sectors(bio));
	part_stat_unlock();
1402

1403 1404
	/* if we're suspended, we have to queue this io for later */
	if (unlikely(test_bit(DMF_BLOCK_IO_FOR_SUSPEND, &md->flags))) {
1405
		up_read(&md->io_lock);
L
Linus Torvalds 已提交
1406

1407 1408 1409
		if (bio_rw(bio) != READA)
			queue_io(md, bio);
		else
A
Alasdair G Kergon 已提交
1410
			bio_io_error(bio);
M
Mikulas Patocka 已提交
1411
		return 0;
L
Linus Torvalds 已提交
1412 1413
	}

1414
	__split_and_process_bio(md, bio);
1415
	up_read(&md->io_lock);
1416
	return 0;
L
Linus Torvalds 已提交
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 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485
static int dm_make_request(struct request_queue *q, struct bio *bio)
{
	struct mapped_device *md = q->queuedata;

	return md->saved_make_request_fn(q, bio); /* call __make_request() */
}

static int dm_request_based(struct mapped_device *md)
{
	return blk_queue_stackable(md->queue);
}

static int dm_request(struct request_queue *q, struct bio *bio)
{
	struct mapped_device *md = q->queuedata;

	if (dm_request_based(md))
		return dm_make_request(q, bio);

	return _dm_request(q, bio);
}

void dm_dispatch_request(struct request *rq)
{
	int r;

	if (blk_queue_io_stat(rq->q))
		rq->cmd_flags |= REQ_IO_STAT;

	rq->start_time = jiffies;
	r = blk_insert_cloned_request(rq->q, rq);
	if (r)
		dm_complete_request(rq, r);
}
EXPORT_SYMBOL_GPL(dm_dispatch_request);

static void dm_rq_bio_destructor(struct bio *bio)
{
	struct dm_rq_clone_bio_info *info = bio->bi_private;
	struct mapped_device *md = info->tio->md;

	free_bio_info(info);
	bio_free(bio, md->bs);
}

static int dm_rq_bio_constructor(struct bio *bio, struct bio *bio_orig,
				 void *data)
{
	struct dm_rq_target_io *tio = data;
	struct mapped_device *md = tio->md;
	struct dm_rq_clone_bio_info *info = alloc_bio_info(md);

	if (!info)
		return -ENOMEM;

	info->orig = bio_orig;
	info->tio = tio;
	bio->bi_end_io = end_clone_bio;
	bio->bi_private = info;
	bio->bi_destructor = dm_rq_bio_destructor;

	return 0;
}

static int setup_clone(struct request *clone, struct request *rq,
		       struct dm_rq_target_io *tio)
{
1486
	int r;
1487

1488 1489 1490 1491
	r = blk_rq_prep_clone(clone, rq, tio->md->bs, GFP_ATOMIC,
			      dm_rq_bio_constructor, tio);
	if (r)
		return r;
1492

1493 1494 1495 1496
	clone->cmd = rq->cmd;
	clone->cmd_len = rq->cmd_len;
	clone->sense = rq->sense;
	clone->buffer = rq->buffer;
1497 1498 1499 1500 1501 1502
	clone->end_io = end_clone_request;
	clone->end_io_data = tio;

	return 0;
}

K
Kiyoshi Ueda 已提交
1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528
static struct request *clone_rq(struct request *rq, struct mapped_device *md,
				gfp_t gfp_mask)
{
	struct request *clone;
	struct dm_rq_target_io *tio;

	tio = alloc_rq_tio(md, gfp_mask);
	if (!tio)
		return NULL;

	tio->md = md;
	tio->ti = NULL;
	tio->orig = rq;
	tio->error = 0;
	memset(&tio->info, 0, sizeof(tio->info));

	clone = &tio->clone;
	if (setup_clone(clone, rq, tio)) {
		/* -ENOMEM */
		free_rq_tio(tio);
		return NULL;
	}

	return clone;
}

1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541
/*
 * Called with the queue lock held.
 */
static int dm_prep_fn(struct request_queue *q, struct request *rq)
{
	struct mapped_device *md = q->queuedata;
	struct request *clone;

	if (unlikely(rq->special)) {
		DMWARN("Already has something in rq->special.");
		return BLKPREP_KILL;
	}

K
Kiyoshi Ueda 已提交
1542 1543
	clone = clone_rq(rq, md, GFP_ATOMIC);
	if (!clone)
1544 1545 1546 1547 1548 1549 1550 1551
		return BLKPREP_DEFER;

	rq->special = clone;
	rq->cmd_flags |= REQ_DONTPREP;

	return BLKPREP_OK;
}

1552 1553 1554 1555 1556 1557 1558
/*
 * Returns:
 * 0  : the request has been processed (not requeued)
 * !0 : the request has been requeued
 */
static int map_request(struct dm_target *ti, struct request *clone,
		       struct mapped_device *md)
1559
{
1560
	int r, requeued = 0;
1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579
	struct dm_rq_target_io *tio = clone->end_io_data;

	/*
	 * Hold the md reference here for the in-flight I/O.
	 * We can't rely on the reference count by device opener,
	 * because the device may be closed during the request completion
	 * when all bios are completed.
	 * See the comment in rq_completed() too.
	 */
	dm_get(md);

	tio->ti = ti;
	r = ti->type->map_rq(ti, clone, &tio->info);
	switch (r) {
	case DM_MAPIO_SUBMITTED:
		/* The target has taken the I/O to submit by itself later */
		break;
	case DM_MAPIO_REMAPPED:
		/* The target has remapped the I/O so dispatch it */
1580 1581
		trace_block_rq_remap(clone->q, clone, disk_devt(dm_disk(md)),
				     blk_rq_pos(tio->orig));
1582 1583 1584 1585 1586
		dm_dispatch_request(clone);
		break;
	case DM_MAPIO_REQUEUE:
		/* The target wants to requeue the I/O */
		dm_requeue_unmapped_request(clone);
1587
		requeued = 1;
1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598
		break;
	default:
		if (r > 0) {
			DMWARN("unimplemented target map return value: %d", r);
			BUG();
		}

		/* The target wants to complete the I/O */
		dm_kill_unmapped_request(clone, r);
		break;
	}
1599 1600

	return requeued;
1601 1602 1603 1604 1605 1606 1607 1608 1609
}

/*
 * q->request_fn for request-based dm.
 * Called with the queue lock held.
 */
static void dm_request_fn(struct request_queue *q)
{
	struct mapped_device *md = q->queuedata;
1610
	struct dm_table *map = dm_get_live_table(md);
1611
	struct dm_target *ti;
1612
	struct request *rq, *clone;
1613
	sector_t pos;
1614 1615

	/*
1616 1617 1618 1619
	 * For suspend, check blk_queue_stopped() and increment
	 * ->pending within a single queue_lock not to increment the
	 * number of in-flight I/Os after the queue is stopped in
	 * dm_suspend().
1620 1621 1622 1623 1624 1625
	 */
	while (!blk_queue_plugged(q) && !blk_queue_stopped(q)) {
		rq = blk_peek_request(q);
		if (!rq)
			goto plug_and_out;

1626 1627 1628 1629 1630 1631 1632
		/* always use block 0 to find the target for flushes for now */
		pos = 0;
		if (!(rq->cmd_flags & REQ_FLUSH))
			pos = blk_rq_pos(rq);

		ti = dm_table_find_target(map, pos);
		BUG_ON(!dm_target_is_valid(ti));
1633

1634 1635 1636 1637
		if (ti->type->busy && ti->type->busy(ti))
			goto plug_and_out;

		blk_start_request(rq);
1638 1639 1640
		clone = rq->special;
		atomic_inc(&md->pending[rq_data_dir(clone)]);

1641
		spin_unlock(q->queue_lock);
1642 1643 1644
		if (map_request(ti, clone, md))
			goto requeued;

1645 1646 1647 1648 1649
		spin_lock_irq(q->queue_lock);
	}

	goto out;

1650 1651 1652
requeued:
	spin_lock_irq(q->queue_lock);

1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673
plug_and_out:
	if (!elv_queue_empty(q))
		/* Some requests still remain, retry later */
		blk_plug_device(q);

out:
	dm_table_put(map);

	return;
}

int dm_underlying_device_busy(struct request_queue *q)
{
	return blk_lld_busy(q);
}
EXPORT_SYMBOL_GPL(dm_underlying_device_busy);

static int dm_lld_busy(struct request_queue *q)
{
	int r;
	struct mapped_device *md = q->queuedata;
1674
	struct dm_table *map = dm_get_live_table(md);
1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685

	if (!map || test_bit(DMF_BLOCK_IO_FOR_SUSPEND, &md->flags))
		r = 1;
	else
		r = dm_table_any_busy_target(map);

	dm_table_put(map);

	return r;
}

1686
static void dm_unplug_all(struct request_queue *q)
L
Linus Torvalds 已提交
1687 1688
{
	struct mapped_device *md = q->queuedata;
1689
	struct dm_table *map = dm_get_live_table(md);
L
Linus Torvalds 已提交
1690 1691

	if (map) {
1692 1693 1694
		if (dm_request_based(md))
			generic_unplug_device(q);

L
Linus Torvalds 已提交
1695 1696 1697 1698 1699 1700 1701
		dm_table_unplug_all(map);
		dm_table_put(map);
	}
}

static int dm_any_congested(void *congested_data, int bdi_bits)
{
1702 1703 1704
	int r = bdi_bits;
	struct mapped_device *md = congested_data;
	struct dm_table *map;
L
Linus Torvalds 已提交
1705

1706
	if (!test_bit(DMF_BLOCK_IO_FOR_SUSPEND, &md->flags)) {
1707
		map = dm_get_live_table(md);
1708
		if (map) {
1709 1710 1711 1712 1713 1714 1715 1716 1717 1718
			/*
			 * Request-based dm cares about only own queue for
			 * the query about congestion status of request_queue
			 */
			if (dm_request_based(md))
				r = md->queue->backing_dev_info.state &
				    bdi_bits;
			else
				r = dm_table_any_congested(map, bdi_bits);

1719 1720 1721 1722
			dm_table_put(map);
		}
	}

L
Linus Torvalds 已提交
1723 1724 1725 1726 1727 1728 1729 1730
	return r;
}

/*-----------------------------------------------------------------
 * An IDR is used to keep track of allocated minor numbers.
 *---------------------------------------------------------------*/
static DEFINE_IDR(_minor_idr);

1731
static void free_minor(int minor)
L
Linus Torvalds 已提交
1732
{
1733
	spin_lock(&_minor_lock);
L
Linus Torvalds 已提交
1734
	idr_remove(&_minor_idr, minor);
1735
	spin_unlock(&_minor_lock);
L
Linus Torvalds 已提交
1736 1737 1738 1739 1740
}

/*
 * See if the device with a specific minor # is free.
 */
1741
static int specific_minor(int minor)
L
Linus Torvalds 已提交
1742 1743 1744 1745 1746 1747
{
	int r, m;

	if (minor >= (1 << MINORBITS))
		return -EINVAL;

J
Jeff Mahoney 已提交
1748 1749 1750 1751
	r = idr_pre_get(&_minor_idr, GFP_KERNEL);
	if (!r)
		return -ENOMEM;

1752
	spin_lock(&_minor_lock);
L
Linus Torvalds 已提交
1753 1754 1755 1756 1757 1758

	if (idr_find(&_minor_idr, minor)) {
		r = -EBUSY;
		goto out;
	}

1759
	r = idr_get_new_above(&_minor_idr, MINOR_ALLOCED, minor, &m);
J
Jeff Mahoney 已提交
1760
	if (r)
L
Linus Torvalds 已提交
1761 1762 1763 1764 1765 1766 1767 1768 1769
		goto out;

	if (m != minor) {
		idr_remove(&_minor_idr, m);
		r = -EBUSY;
		goto out;
	}

out:
1770
	spin_unlock(&_minor_lock);
L
Linus Torvalds 已提交
1771 1772 1773
	return r;
}

1774
static int next_free_minor(int *minor)
L
Linus Torvalds 已提交
1775
{
1776
	int r, m;
L
Linus Torvalds 已提交
1777 1778

	r = idr_pre_get(&_minor_idr, GFP_KERNEL);
J
Jeff Mahoney 已提交
1779 1780 1781
	if (!r)
		return -ENOMEM;

1782
	spin_lock(&_minor_lock);
L
Linus Torvalds 已提交
1783

1784
	r = idr_get_new(&_minor_idr, MINOR_ALLOCED, &m);
1785
	if (r)
L
Linus Torvalds 已提交
1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796
		goto out;

	if (m >= (1 << MINORBITS)) {
		idr_remove(&_minor_idr, m);
		r = -ENOSPC;
		goto out;
	}

	*minor = m;

out:
1797
	spin_unlock(&_minor_lock);
L
Linus Torvalds 已提交
1798 1799 1800
	return r;
}

1801
static const struct block_device_operations dm_blk_dops;
L
Linus Torvalds 已提交
1802

1803 1804
static void dm_wq_work(struct work_struct *work);

1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824
static void dm_init_md_queue(struct mapped_device *md)
{
	/*
	 * Request-based dm devices cannot be stacked on top of bio-based dm
	 * devices.  The type of this dm device has not been decided yet.
	 * The type is decided at the first table loading time.
	 * To prevent problematic device stacking, clear the queue flag
	 * for request stacking support until then.
	 *
	 * This queue is new, so no concurrency on the queue_flags.
	 */
	queue_flag_clear_unlocked(QUEUE_FLAG_STACKABLE, md->queue);

	md->queue->queuedata = md;
	md->queue->backing_dev_info.congested_fn = dm_any_congested;
	md->queue->backing_dev_info.congested_data = md;
	blk_queue_make_request(md->queue, dm_request);
	blk_queue_bounce_limit(md->queue, BLK_BOUNCE_ANY);
	md->queue->unplug_fn = dm_unplug_all;
	blk_queue_merge_bvec(md->queue, dm_merge_bvec);
1825
	blk_queue_flush(md->queue, REQ_FLUSH | REQ_FUA);
1826 1827
}

L
Linus Torvalds 已提交
1828 1829 1830
/*
 * Allocate and initialise a blank device with a given minor.
 */
1831
static struct mapped_device *alloc_dev(int minor)
L
Linus Torvalds 已提交
1832 1833
{
	int r;
1834
	struct mapped_device *md = kzalloc(sizeof(*md), GFP_KERNEL);
1835
	void *old_md;
L
Linus Torvalds 已提交
1836 1837 1838 1839 1840 1841

	if (!md) {
		DMWARN("unable to allocate device, out of memory.");
		return NULL;
	}

1842
	if (!try_module_get(THIS_MODULE))
M
Milan Broz 已提交
1843
		goto bad_module_get;
1844

L
Linus Torvalds 已提交
1845
	/* get a minor number for the dev */
1846
	if (minor == DM_ANY_MINOR)
1847
		r = next_free_minor(&minor);
1848
	else
1849
		r = specific_minor(minor);
L
Linus Torvalds 已提交
1850
	if (r < 0)
M
Milan Broz 已提交
1851
		goto bad_minor;
L
Linus Torvalds 已提交
1852

1853
	md->type = DM_TYPE_NONE;
1854
	init_rwsem(&md->io_lock);
1855
	mutex_init(&md->suspend_lock);
1856
	mutex_init(&md->type_lock);
1857
	spin_lock_init(&md->deferred_lock);
L
Linus Torvalds 已提交
1858 1859
	rwlock_init(&md->map_lock);
	atomic_set(&md->holders, 1);
1860
	atomic_set(&md->open_count, 0);
L
Linus Torvalds 已提交
1861
	atomic_set(&md->event_nr, 0);
M
Mike Anderson 已提交
1862 1863 1864
	atomic_set(&md->uevent_seq, 0);
	INIT_LIST_HEAD(&md->uevent_list);
	spin_lock_init(&md->uevent_lock);
L
Linus Torvalds 已提交
1865

1866
	md->queue = blk_alloc_queue(GFP_KERNEL);
L
Linus Torvalds 已提交
1867
	if (!md->queue)
M
Milan Broz 已提交
1868
		goto bad_queue;
L
Linus Torvalds 已提交
1869

1870
	dm_init_md_queue(md);
S
Stefan Bader 已提交
1871

L
Linus Torvalds 已提交
1872 1873
	md->disk = alloc_disk(1);
	if (!md->disk)
M
Milan Broz 已提交
1874
		goto bad_disk;
L
Linus Torvalds 已提交
1875

1876 1877
	atomic_set(&md->pending[0], 0);
	atomic_set(&md->pending[1], 0);
1878
	init_waitqueue_head(&md->wait);
1879
	INIT_WORK(&md->work, dm_wq_work);
1880 1881
	init_waitqueue_head(&md->eventq);

L
Linus Torvalds 已提交
1882 1883 1884 1885 1886 1887 1888
	md->disk->major = _major;
	md->disk->first_minor = minor;
	md->disk->fops = &dm_blk_dops;
	md->disk->queue = md->queue;
	md->disk->private_data = md;
	sprintf(md->disk->disk_name, "dm-%d", minor);
	add_disk(md->disk);
M
Mike Anderson 已提交
1889
	format_dev_t(md->name, MKDEV(_major, minor));
L
Linus Torvalds 已提交
1890

1891 1892 1893 1894
	md->wq = create_singlethread_workqueue("kdmflush");
	if (!md->wq)
		goto bad_thread;

M
Mikulas Patocka 已提交
1895 1896 1897 1898
	md->bdev = bdget_disk(md->disk, 0);
	if (!md->bdev)
		goto bad_bdev;

1899 1900 1901 1902
	bio_init(&md->flush_bio);
	md->flush_bio.bi_bdev = md->bdev;
	md->flush_bio.bi_rw = WRITE_FLUSH;

1903
	/* Populate the mapping, nobody knows we exist yet */
1904
	spin_lock(&_minor_lock);
1905
	old_md = idr_replace(&_minor_idr, md, minor);
1906
	spin_unlock(&_minor_lock);
1907 1908 1909

	BUG_ON(old_md != MINOR_ALLOCED);

L
Linus Torvalds 已提交
1910 1911
	return md;

M
Mikulas Patocka 已提交
1912 1913
bad_bdev:
	destroy_workqueue(md->wq);
1914
bad_thread:
1915
	del_gendisk(md->disk);
1916
	put_disk(md->disk);
M
Milan Broz 已提交
1917
bad_disk:
1918
	blk_cleanup_queue(md->queue);
M
Milan Broz 已提交
1919
bad_queue:
L
Linus Torvalds 已提交
1920
	free_minor(minor);
M
Milan Broz 已提交
1921
bad_minor:
1922
	module_put(THIS_MODULE);
M
Milan Broz 已提交
1923
bad_module_get:
L
Linus Torvalds 已提交
1924 1925 1926 1927
	kfree(md);
	return NULL;
}

J
Jun'ichi Nomura 已提交
1928 1929
static void unlock_fs(struct mapped_device *md);

L
Linus Torvalds 已提交
1930 1931
static void free_dev(struct mapped_device *md)
{
1932
	int minor = MINOR(disk_devt(md->disk));
1933

M
Mikulas Patocka 已提交
1934 1935
	unlock_fs(md);
	bdput(md->bdev);
1936
	destroy_workqueue(md->wq);
K
Kiyoshi Ueda 已提交
1937 1938 1939 1940 1941 1942
	if (md->tio_pool)
		mempool_destroy(md->tio_pool);
	if (md->io_pool)
		mempool_destroy(md->io_pool);
	if (md->bs)
		bioset_free(md->bs);
M
Martin K. Petersen 已提交
1943
	blk_integrity_unregister(md->disk);
L
Linus Torvalds 已提交
1944
	del_gendisk(md->disk);
1945
	free_minor(minor);
J
Jeff Mahoney 已提交
1946 1947 1948 1949 1950

	spin_lock(&_minor_lock);
	md->disk->private_data = NULL;
	spin_unlock(&_minor_lock);

L
Linus Torvalds 已提交
1951
	put_disk(md->disk);
1952
	blk_cleanup_queue(md->queue);
1953
	module_put(THIS_MODULE);
L
Linus Torvalds 已提交
1954 1955 1956
	kfree(md);
}

K
Kiyoshi Ueda 已提交
1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979
static void __bind_mempools(struct mapped_device *md, struct dm_table *t)
{
	struct dm_md_mempools *p;

	if (md->io_pool && md->tio_pool && md->bs)
		/* the md already has necessary mempools */
		goto out;

	p = dm_table_get_md_mempools(t);
	BUG_ON(!p || md->io_pool || md->tio_pool || md->bs);

	md->io_pool = p->io_pool;
	p->io_pool = NULL;
	md->tio_pool = p->tio_pool;
	p->tio_pool = NULL;
	md->bs = p->bs;
	p->bs = NULL;

out:
	/* mempool bind completed, now no need any mempools in the table */
	dm_table_free_md_mempools(t);
}

L
Linus Torvalds 已提交
1980 1981 1982 1983 1984
/*
 * Bind a table to the device.
 */
static void event_callback(void *context)
{
M
Mike Anderson 已提交
1985 1986
	unsigned long flags;
	LIST_HEAD(uevents);
L
Linus Torvalds 已提交
1987 1988
	struct mapped_device *md = (struct mapped_device *) context;

M
Mike Anderson 已提交
1989 1990 1991 1992
	spin_lock_irqsave(&md->uevent_lock, flags);
	list_splice_init(&md->uevent_list, &uevents);
	spin_unlock_irqrestore(&md->uevent_lock, flags);

1993
	dm_send_uevents(&uevents, &disk_to_dev(md->disk)->kobj);
M
Mike Anderson 已提交
1994

L
Linus Torvalds 已提交
1995 1996 1997 1998
	atomic_inc(&md->event_nr);
	wake_up(&md->eventq);
}

1999
static void __set_size(struct mapped_device *md, sector_t size)
L
Linus Torvalds 已提交
2000
{
2001
	set_capacity(md->disk, size);
L
Linus Torvalds 已提交
2002

2003 2004 2005
	mutex_lock(&md->bdev->bd_inode->i_mutex);
	i_size_write(md->bdev->bd_inode, (loff_t)size << SECTOR_SHIFT);
	mutex_unlock(&md->bdev->bd_inode->i_mutex);
L
Linus Torvalds 已提交
2006 2007
}

2008 2009 2010 2011 2012
/*
 * Returns old map, which caller must destroy.
 */
static struct dm_table *__bind(struct mapped_device *md, struct dm_table *t,
			       struct queue_limits *limits)
L
Linus Torvalds 已提交
2013
{
2014
	struct dm_table *old_map;
2015
	struct request_queue *q = md->queue;
L
Linus Torvalds 已提交
2016
	sector_t size;
2017
	unsigned long flags;
L
Linus Torvalds 已提交
2018 2019

	size = dm_table_get_size(t);
D
Darrick J. Wong 已提交
2020 2021 2022 2023 2024 2025 2026

	/*
	 * Wipe any geometry if the size of the table changed.
	 */
	if (size != get_capacity(md->disk))
		memset(&md->geometry, 0, sizeof(md->geometry));

M
Mikulas Patocka 已提交
2027
	__set_size(md, size);
2028

2029 2030
	dm_table_event_callback(t, event_callback, md);

K
Kiyoshi Ueda 已提交
2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042
	/*
	 * The queue hasn't been stopped yet, if the old table type wasn't
	 * for request-based during suspension.  So stop it to prevent
	 * I/O mapping before resume.
	 * This must be done before setting the queue restrictions,
	 * because request-based dm may be run just after the setting.
	 */
	if (dm_table_request_based(t) && !blk_queue_stopped(q))
		stop_queue(q);

	__bind_mempools(md, t);

2043
	write_lock_irqsave(&md->map_lock, flags);
2044
	old_map = md->map;
L
Linus Torvalds 已提交
2045
	md->map = t;
2046
	dm_table_set_restrictions(t, q, limits);
2047
	write_unlock_irqrestore(&md->map_lock, flags);
L
Linus Torvalds 已提交
2048

2049
	return old_map;
L
Linus Torvalds 已提交
2050 2051
}

2052 2053 2054 2055
/*
 * Returns unbound table for the caller to free.
 */
static struct dm_table *__unbind(struct mapped_device *md)
L
Linus Torvalds 已提交
2056 2057
{
	struct dm_table *map = md->map;
2058
	unsigned long flags;
L
Linus Torvalds 已提交
2059 2060

	if (!map)
2061
		return NULL;
L
Linus Torvalds 已提交
2062 2063

	dm_table_event_callback(map, NULL, NULL);
2064
	write_lock_irqsave(&md->map_lock, flags);
L
Linus Torvalds 已提交
2065
	md->map = NULL;
2066
	write_unlock_irqrestore(&md->map_lock, flags);
2067 2068

	return map;
L
Linus Torvalds 已提交
2069 2070 2071 2072 2073
}

/*
 * Constructor for a new device.
 */
2074
int dm_create(int minor, struct mapped_device **result)
L
Linus Torvalds 已提交
2075 2076 2077
{
	struct mapped_device *md;

2078
	md = alloc_dev(minor);
L
Linus Torvalds 已提交
2079 2080 2081
	if (!md)
		return -ENXIO;

M
Milan Broz 已提交
2082 2083
	dm_sysfs_init(md);

L
Linus Torvalds 已提交
2084 2085 2086 2087
	*result = md;
	return 0;
}

2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111
/*
 * Functions to manage md->type.
 * All are required to hold md->type_lock.
 */
void dm_lock_md_type(struct mapped_device *md)
{
	mutex_lock(&md->type_lock);
}

void dm_unlock_md_type(struct mapped_device *md)
{
	mutex_unlock(&md->type_lock);
}

void dm_set_md_type(struct mapped_device *md, unsigned type)
{
	md->type = type;
}

unsigned dm_get_md_type(struct mapped_device *md)
{
	return md->type;
}

2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152
/*
 * Fully initialize a request-based queue (->elevator, ->request_fn, etc).
 */
static int dm_init_request_based_queue(struct mapped_device *md)
{
	struct request_queue *q = NULL;

	if (md->queue->elevator)
		return 1;

	/* Fully initialize the queue */
	q = blk_init_allocated_queue(md->queue, dm_request_fn, NULL);
	if (!q)
		return 0;

	md->queue = q;
	md->saved_make_request_fn = md->queue->make_request_fn;
	dm_init_md_queue(md);
	blk_queue_softirq_done(md->queue, dm_softirq_done);
	blk_queue_prep_rq(md->queue, dm_prep_fn);
	blk_queue_lld_busy(md->queue, dm_lld_busy);

	elv_register_queue(md->queue);

	return 1;
}

/*
 * Setup the DM device's queue based on md's type
 */
int dm_setup_md_queue(struct mapped_device *md)
{
	if ((dm_get_md_type(md) == DM_TYPE_REQUEST_BASED) &&
	    !dm_init_request_based_queue(md)) {
		DMWARN("Cannot initialize queue for request-based mapped device");
		return -EINVAL;
	}

	return 0;
}

2153
static struct mapped_device *dm_find_md(dev_t dev)
L
Linus Torvalds 已提交
2154 2155 2156 2157 2158 2159 2160
{
	struct mapped_device *md;
	unsigned minor = MINOR(dev);

	if (MAJOR(dev) != _major || minor >= (1 << MINORBITS))
		return NULL;

2161
	spin_lock(&_minor_lock);
L
Linus Torvalds 已提交
2162 2163

	md = idr_find(&_minor_idr, minor);
J
Jeff Mahoney 已提交
2164
	if (md && (md == MINOR_ALLOCED ||
2165
		   (MINOR(disk_devt(dm_disk(md))) != minor) ||
2166
		   dm_deleting_md(md) ||
A
Alasdair G Kergon 已提交
2167
		   test_bit(DMF_FREEING, &md->flags))) {
2168
		md = NULL;
J
Jeff Mahoney 已提交
2169 2170
		goto out;
	}
L
Linus Torvalds 已提交
2171

J
Jeff Mahoney 已提交
2172
out:
2173
	spin_unlock(&_minor_lock);
L
Linus Torvalds 已提交
2174

2175 2176 2177
	return md;
}

2178 2179 2180 2181 2182 2183 2184 2185 2186 2187
struct mapped_device *dm_get_md(dev_t dev)
{
	struct mapped_device *md = dm_find_md(dev);

	if (md)
		dm_get(md);

	return md;
}

A
Alasdair G Kergon 已提交
2188
void *dm_get_mdptr(struct mapped_device *md)
2189
{
A
Alasdair G Kergon 已提交
2190
	return md->interface_ptr;
L
Linus Torvalds 已提交
2191 2192 2193 2194 2195 2196 2197 2198 2199 2200
}

void dm_set_mdptr(struct mapped_device *md, void *ptr)
{
	md->interface_ptr = ptr;
}

void dm_get(struct mapped_device *md)
{
	atomic_inc(&md->holders);
2201
	BUG_ON(test_bit(DMF_FREEING, &md->flags));
L
Linus Torvalds 已提交
2202 2203
}

2204 2205 2206 2207 2208 2209
const char *dm_device_name(struct mapped_device *md)
{
	return md->name;
}
EXPORT_SYMBOL_GPL(dm_device_name);

2210
static void __dm_destroy(struct mapped_device *md, bool wait)
L
Linus Torvalds 已提交
2211
{
M
Mike Anderson 已提交
2212
	struct dm_table *map;
L
Linus Torvalds 已提交
2213

2214
	might_sleep();
J
Jeff Mahoney 已提交
2215

2216 2217 2218 2219 2220 2221 2222 2223 2224
	spin_lock(&_minor_lock);
	map = dm_get_live_table(md);
	idr_replace(&_minor_idr, MINOR_ALLOCED, MINOR(disk_devt(dm_disk(md))));
	set_bit(DMF_FREEING, &md->flags);
	spin_unlock(&_minor_lock);

	if (!dm_suspended_md(md)) {
		dm_table_presuspend_targets(map);
		dm_table_postsuspend_targets(map);
L
Linus Torvalds 已提交
2225
	}
2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258

	/*
	 * Rare, but there may be I/O requests still going to complete,
	 * for example.  Wait for all references to disappear.
	 * No one should increment the reference count of the mapped_device,
	 * after the mapped_device state becomes DMF_FREEING.
	 */
	if (wait)
		while (atomic_read(&md->holders))
			msleep(1);
	else if (atomic_read(&md->holders))
		DMWARN("%s: Forcibly removing mapped_device still in use! (%d users)",
		       dm_device_name(md), atomic_read(&md->holders));

	dm_sysfs_exit(md);
	dm_table_put(map);
	dm_table_destroy(__unbind(md));
	free_dev(md);
}

void dm_destroy(struct mapped_device *md)
{
	__dm_destroy(md, true);
}

void dm_destroy_immediate(struct mapped_device *md)
{
	__dm_destroy(md, false);
}

void dm_put(struct mapped_device *md)
{
	atomic_dec(&md->holders);
L
Linus Torvalds 已提交
2259
}
E
Edward Goggin 已提交
2260
EXPORT_SYMBOL_GPL(dm_put);
L
Linus Torvalds 已提交
2261

2262
static int dm_wait_for_completion(struct mapped_device *md, int interruptible)
2263 2264
{
	int r = 0;
2265 2266 2267 2268 2269
	DECLARE_WAITQUEUE(wait, current);

	dm_unplug_all(md->queue);

	add_wait_queue(&md->wait, &wait);
2270 2271

	while (1) {
2272
		set_current_state(interruptible);
2273 2274

		smp_mb();
2275
		if (!md_in_flight(md))
2276 2277
			break;

2278 2279
		if (interruptible == TASK_INTERRUPTIBLE &&
		    signal_pending(current)) {
2280 2281 2282 2283 2284 2285 2286 2287
			r = -EINTR;
			break;
		}

		io_schedule();
	}
	set_current_state(TASK_RUNNING);

2288 2289
	remove_wait_queue(&md->wait, &wait);

2290 2291 2292
	return r;
}

L
Linus Torvalds 已提交
2293 2294 2295
/*
 * Process the deferred bios
 */
2296
static void dm_wq_work(struct work_struct *work)
L
Linus Torvalds 已提交
2297
{
2298 2299
	struct mapped_device *md = container_of(work, struct mapped_device,
						work);
2300
	struct bio *c;
L
Linus Torvalds 已提交
2301

2302
	down_read(&md->io_lock);
2303

2304
	while (!test_bit(DMF_BLOCK_IO_FOR_SUSPEND, &md->flags)) {
A
Alasdair G Kergon 已提交
2305 2306 2307 2308
		spin_lock_irq(&md->deferred_lock);
		c = bio_list_pop(&md->deferred);
		spin_unlock_irq(&md->deferred_lock);

2309
		if (!c)
A
Alasdair G Kergon 已提交
2310
			break;
2311

2312
		up_read(&md->io_lock);
2313

K
Kiyoshi Ueda 已提交
2314 2315
		if (dm_request_based(md))
			generic_make_request(c);
2316 2317
		else
			__split_and_process_bio(md, c);
2318

2319
		down_read(&md->io_lock);
2320
	}
M
Milan Broz 已提交
2321

2322
	up_read(&md->io_lock);
L
Linus Torvalds 已提交
2323 2324
}

2325
static void dm_queue_flush(struct mapped_device *md)
2326
{
2327 2328
	clear_bit(DMF_BLOCK_IO_FOR_SUSPEND, &md->flags);
	smp_mb__after_clear_bit();
2329
	queue_work(md->wq, &md->work);
2330 2331
}

L
Linus Torvalds 已提交
2332
/*
2333
 * Swap in a new table, returning the old one for the caller to destroy.
L
Linus Torvalds 已提交
2334
 */
2335
struct dm_table *dm_swap_table(struct mapped_device *md, struct dm_table *table)
L
Linus Torvalds 已提交
2336
{
2337
	struct dm_table *map = ERR_PTR(-EINVAL);
2338
	struct queue_limits limits;
2339
	int r;
L
Linus Torvalds 已提交
2340

2341
	mutex_lock(&md->suspend_lock);
L
Linus Torvalds 已提交
2342 2343

	/* device must be suspended */
2344
	if (!dm_suspended_md(md))
2345
		goto out;
L
Linus Torvalds 已提交
2346

2347
	r = dm_calculate_queue_limits(table, &limits);
2348 2349
	if (r) {
		map = ERR_PTR(r);
2350
		goto out;
2351
	}
2352

2353
	map = __bind(md, table, &limits);
L
Linus Torvalds 已提交
2354

2355
out:
2356
	mutex_unlock(&md->suspend_lock);
2357
	return map;
L
Linus Torvalds 已提交
2358 2359 2360 2361 2362 2363
}

/*
 * Functions to lock and unlock any filesystem running on the
 * device.
 */
2364
static int lock_fs(struct mapped_device *md)
L
Linus Torvalds 已提交
2365
{
2366
	int r;
L
Linus Torvalds 已提交
2367 2368

	WARN_ON(md->frozen_sb);
2369

2370
	md->frozen_sb = freeze_bdev(md->bdev);
2371
	if (IS_ERR(md->frozen_sb)) {
2372
		r = PTR_ERR(md->frozen_sb);
2373 2374
		md->frozen_sb = NULL;
		return r;
2375 2376
	}

2377 2378
	set_bit(DMF_FROZEN, &md->flags);

L
Linus Torvalds 已提交
2379 2380 2381
	return 0;
}

2382
static void unlock_fs(struct mapped_device *md)
L
Linus Torvalds 已提交
2383
{
2384 2385 2386
	if (!test_bit(DMF_FROZEN, &md->flags))
		return;

2387
	thaw_bdev(md->bdev, md->frozen_sb);
L
Linus Torvalds 已提交
2388
	md->frozen_sb = NULL;
2389
	clear_bit(DMF_FROZEN, &md->flags);
L
Linus Torvalds 已提交
2390 2391 2392 2393 2394 2395 2396 2397 2398
}

/*
 * We need to be able to change a mapping table under a mounted
 * filesystem.  For example we might want to move some data in
 * the background.  Before the table can be swapped with
 * dm_bind_table, dm_suspend must be called to flush any in
 * flight bios and ensure that any further io gets deferred.
 */
2399 2400 2401
/*
 * Suspend mechanism in request-based dm.
 *
K
Kiyoshi Ueda 已提交
2402 2403 2404
 * 1. Flush all I/Os by lock_fs() if needed.
 * 2. Stop dispatching any I/O by stopping the request_queue.
 * 3. Wait for all in-flight I/Os to be completed or requeued.
2405
 *
K
Kiyoshi Ueda 已提交
2406
 * To abort suspend, start the request_queue.
2407
 */
2408
int dm_suspend(struct mapped_device *md, unsigned suspend_flags)
L
Linus Torvalds 已提交
2409
{
2410
	struct dm_table *map = NULL;
2411
	int r = 0;
2412
	int do_lockfs = suspend_flags & DM_SUSPEND_LOCKFS_FLAG ? 1 : 0;
2413
	int noflush = suspend_flags & DM_SUSPEND_NOFLUSH_FLAG ? 1 : 0;
L
Linus Torvalds 已提交
2414

2415
	mutex_lock(&md->suspend_lock);
2416

2417
	if (dm_suspended_md(md)) {
M
Milan Broz 已提交
2418
		r = -EINVAL;
2419
		goto out_unlock;
M
Milan Broz 已提交
2420
	}
L
Linus Torvalds 已提交
2421

2422
	map = dm_get_live_table(md);
L
Linus Torvalds 已提交
2423

2424 2425 2426 2427 2428 2429 2430
	/*
	 * DMF_NOFLUSH_SUSPENDING must be set before presuspend.
	 * This flag is cleared before dm_suspend returns.
	 */
	if (noflush)
		set_bit(DMF_NOFLUSH_SUSPENDING, &md->flags);

2431 2432 2433
	/* This does not get reverted if there's an error later. */
	dm_table_presuspend_targets(map);

M
Mikulas Patocka 已提交
2434
	/*
K
Kiyoshi Ueda 已提交
2435 2436 2437 2438
	 * Flush I/O to the device.
	 * Any I/O submitted after lock_fs() may not be flushed.
	 * noflush takes precedence over do_lockfs.
	 * (lock_fs() flushes I/Os and waits for them to complete.)
M
Mikulas Patocka 已提交
2439 2440 2441 2442
	 */
	if (!noflush && do_lockfs) {
		r = lock_fs(md);
		if (r)
K
Kiyoshi Ueda 已提交
2443
			goto out;
2444
	}
L
Linus Torvalds 已提交
2445 2446

	/*
2447 2448 2449 2450 2451 2452 2453
	 * Here we must make sure that no processes are submitting requests
	 * to target drivers i.e. no one may be executing
	 * __split_and_process_bio. This is called from dm_request and
	 * dm_wq_work.
	 *
	 * To get all processes out of __split_and_process_bio in dm_request,
	 * we take the write lock. To prevent any process from reentering
2454 2455 2456
	 * __split_and_process_bio from dm_request and quiesce the thread
	 * (dm_wq_work), we set BMF_BLOCK_IO_FOR_SUSPEND and call
	 * flush_workqueue(md->wq).
L
Linus Torvalds 已提交
2457
	 */
2458
	down_write(&md->io_lock);
2459
	set_bit(DMF_BLOCK_IO_FOR_SUSPEND, &md->flags);
2460
	up_write(&md->io_lock);
L
Linus Torvalds 已提交
2461

2462
	/*
2463 2464
	 * Stop md->queue before flushing md->wq in case request-based
	 * dm defers requests to md->wq from md->queue.
2465
	 */
2466
	if (dm_request_based(md))
K
Kiyoshi Ueda 已提交
2467
		stop_queue(md->queue);
2468

2469 2470
	flush_workqueue(md->wq);

L
Linus Torvalds 已提交
2471
	/*
2472 2473 2474
	 * At this point no more requests are entering target request routines.
	 * We call dm_wait_for_completion to wait for all existing requests
	 * to finish.
L
Linus Torvalds 已提交
2475
	 */
2476
	r = dm_wait_for_completion(md, TASK_INTERRUPTIBLE);
L
Linus Torvalds 已提交
2477

2478
	down_write(&md->io_lock);
2479
	if (noflush)
2480
		clear_bit(DMF_NOFLUSH_SUSPENDING, &md->flags);
2481
	up_write(&md->io_lock);
2482

L
Linus Torvalds 已提交
2483
	/* were we interrupted ? */
2484
	if (r < 0) {
2485
		dm_queue_flush(md);
M
Milan Broz 已提交
2486

2487
		if (dm_request_based(md))
K
Kiyoshi Ueda 已提交
2488
			start_queue(md->queue);
2489

2490
		unlock_fs(md);
2491
		goto out; /* pushback list is already flushed, so skip flush */
2492
	}
L
Linus Torvalds 已提交
2493

2494 2495 2496 2497 2498 2499
	/*
	 * If dm_wait_for_completion returned 0, the device is completely
	 * quiescent now. There is no request-processing activity. All new
	 * requests are being added to md->deferred list.
	 */

2500
	set_bit(DMF_SUSPENDED, &md->flags);
2501

2502 2503
	dm_table_postsuspend_targets(map);

2504 2505
out:
	dm_table_put(map);
2506 2507

out_unlock:
2508
	mutex_unlock(&md->suspend_lock);
2509
	return r;
L
Linus Torvalds 已提交
2510 2511 2512 2513
}

int dm_resume(struct mapped_device *md)
{
2514 2515
	int r = -EINVAL;
	struct dm_table *map = NULL;
L
Linus Torvalds 已提交
2516

2517
	mutex_lock(&md->suspend_lock);
2518
	if (!dm_suspended_md(md))
2519 2520
		goto out;

2521
	map = dm_get_live_table(md);
2522
	if (!map || !dm_table_get_size(map))
2523
		goto out;
L
Linus Torvalds 已提交
2524

2525 2526 2527
	r = dm_table_resume_targets(map);
	if (r)
		goto out;
2528

2529
	dm_queue_flush(md);
2530

2531 2532 2533 2534 2535 2536 2537 2538
	/*
	 * Flushing deferred I/Os must be done after targets are resumed
	 * so that mapping of targets can work correctly.
	 * Request-based dm is queueing the deferred I/Os in its request_queue.
	 */
	if (dm_request_based(md))
		start_queue(md->queue);

2539 2540 2541 2542
	unlock_fs(md);

	clear_bit(DMF_SUSPENDED, &md->flags);

L
Linus Torvalds 已提交
2543
	dm_table_unplug_all(map);
2544 2545 2546
	r = 0;
out:
	dm_table_put(map);
2547
	mutex_unlock(&md->suspend_lock);
2548

2549
	return r;
L
Linus Torvalds 已提交
2550 2551 2552 2553 2554
}

/*-----------------------------------------------------------------
 * Event notification.
 *---------------------------------------------------------------*/
2555
int dm_kobject_uevent(struct mapped_device *md, enum kobject_action action,
M
Milan Broz 已提交
2556
		       unsigned cookie)
2557
{
M
Milan Broz 已提交
2558 2559 2560 2561
	char udev_cookie[DM_COOKIE_LENGTH];
	char *envp[] = { udev_cookie, NULL };

	if (!cookie)
2562
		return kobject_uevent(&disk_to_dev(md->disk)->kobj, action);
M
Milan Broz 已提交
2563 2564 2565
	else {
		snprintf(udev_cookie, DM_COOKIE_LENGTH, "%s=%u",
			 DM_COOKIE_ENV_VAR_NAME, cookie);
2566 2567
		return kobject_uevent_env(&disk_to_dev(md->disk)->kobj,
					  action, envp);
M
Milan Broz 已提交
2568
	}
2569 2570
}

M
Mike Anderson 已提交
2571 2572 2573 2574 2575
uint32_t dm_next_uevent_seq(struct mapped_device *md)
{
	return atomic_add_return(1, &md->uevent_seq);
}

L
Linus Torvalds 已提交
2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586
uint32_t dm_get_event_nr(struct mapped_device *md)
{
	return atomic_read(&md->event_nr);
}

int dm_wait_event(struct mapped_device *md, int event_nr)
{
	return wait_event_interruptible(md->eventq,
			(event_nr != atomic_read(&md->event_nr)));
}

M
Mike Anderson 已提交
2587 2588 2589 2590 2591 2592 2593 2594 2595
void dm_uevent_add(struct mapped_device *md, struct list_head *elist)
{
	unsigned long flags;

	spin_lock_irqsave(&md->uevent_lock, flags);
	list_add(elist, &md->uevent_list);
	spin_unlock_irqrestore(&md->uevent_lock, flags);
}

L
Linus Torvalds 已提交
2596 2597 2598 2599 2600 2601 2602 2603 2604
/*
 * The gendisk is only valid as long as you have a reference
 * count on 'md'.
 */
struct gendisk *dm_disk(struct mapped_device *md)
{
	return md->disk;
}

M
Milan Broz 已提交
2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621
struct kobject *dm_kobject(struct mapped_device *md)
{
	return &md->kobj;
}

/*
 * struct mapped_device should not be exported outside of dm.c
 * so use this check to verify that kobj is part of md structure
 */
struct mapped_device *dm_get_from_kobject(struct kobject *kobj)
{
	struct mapped_device *md;

	md = container_of(kobj, struct mapped_device, kobj);
	if (&md->kobj != kobj)
		return NULL;

2622
	if (test_bit(DMF_FREEING, &md->flags) ||
M
Mike Anderson 已提交
2623
	    dm_deleting_md(md))
2624 2625
		return NULL;

M
Milan Broz 已提交
2626 2627 2628 2629
	dm_get(md);
	return md;
}

2630
int dm_suspended_md(struct mapped_device *md)
L
Linus Torvalds 已提交
2631 2632 2633 2634
{
	return test_bit(DMF_SUSPENDED, &md->flags);
}

2635 2636
int dm_suspended(struct dm_target *ti)
{
2637
	return dm_suspended_md(dm_table_get_md(ti->table));
2638 2639 2640
}
EXPORT_SYMBOL_GPL(dm_suspended);

2641 2642
int dm_noflush_suspending(struct dm_target *ti)
{
2643
	return __noflush_suspending(dm_table_get_md(ti->table));
2644 2645 2646
}
EXPORT_SYMBOL_GPL(dm_noflush_suspending);

K
Kiyoshi Ueda 已提交
2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701
struct dm_md_mempools *dm_alloc_md_mempools(unsigned type)
{
	struct dm_md_mempools *pools = kmalloc(sizeof(*pools), GFP_KERNEL);

	if (!pools)
		return NULL;

	pools->io_pool = (type == DM_TYPE_BIO_BASED) ?
			 mempool_create_slab_pool(MIN_IOS, _io_cache) :
			 mempool_create_slab_pool(MIN_IOS, _rq_bio_info_cache);
	if (!pools->io_pool)
		goto free_pools_and_out;

	pools->tio_pool = (type == DM_TYPE_BIO_BASED) ?
			  mempool_create_slab_pool(MIN_IOS, _tio_cache) :
			  mempool_create_slab_pool(MIN_IOS, _rq_tio_cache);
	if (!pools->tio_pool)
		goto free_io_pool_and_out;

	pools->bs = (type == DM_TYPE_BIO_BASED) ?
		    bioset_create(16, 0) : bioset_create(MIN_IOS, 0);
	if (!pools->bs)
		goto free_tio_pool_and_out;

	return pools;

free_tio_pool_and_out:
	mempool_destroy(pools->tio_pool);

free_io_pool_and_out:
	mempool_destroy(pools->io_pool);

free_pools_and_out:
	kfree(pools);

	return NULL;
}

void dm_free_md_mempools(struct dm_md_mempools *pools)
{
	if (!pools)
		return;

	if (pools->io_pool)
		mempool_destroy(pools->io_pool);

	if (pools->tio_pool)
		mempool_destroy(pools->tio_pool);

	if (pools->bs)
		bioset_free(pools->bs);

	kfree(pools);
}

2702
static const struct block_device_operations dm_blk_dops = {
L
Linus Torvalds 已提交
2703 2704
	.open = dm_blk_open,
	.release = dm_blk_close,
2705
	.ioctl = dm_blk_ioctl,
D
Darrick J. Wong 已提交
2706
	.getgeo = dm_blk_getgeo,
L
Linus Torvalds 已提交
2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722
	.owner = THIS_MODULE
};

EXPORT_SYMBOL(dm_get_mapinfo);

/*
 * module hooks
 */
module_init(dm_init);
module_exit(dm_exit);

module_param(major, uint, 0);
MODULE_PARM_DESC(major, "The major number of the device mapper");
MODULE_DESCRIPTION(DM_NAME " driver");
MODULE_AUTHOR("Joe Thornber <dm-devel@redhat.com>");
MODULE_LICENSE("GPL");