dm.c 60.2 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 18 19
#include <linux/moduleparam.h>
#include <linux/blkpg.h>
#include <linux/bio.h>
#include <linux/mempool.h>
#include <linux/slab.h>
#include <linux/idr.h>
D
Darrick J. Wong 已提交
20
#include <linux/hdreg.h>
21
#include <linux/delay.h>
22 23

#include <trace/events/block.h>
L
Linus Torvalds 已提交
24

25 26
#define DM_MSG_PREFIX "core"

N
Namhyung Kim 已提交
27 28 29 30 31 32 33 34 35 36
#ifdef CONFIG_PRINTK
/*
 * ratelimit state to be used in DMXXX_LIMIT().
 */
DEFINE_RATELIMIT_STATE(dm_ratelimit_state,
		       DEFAULT_RATELIMIT_INTERVAL,
		       DEFAULT_RATELIMIT_BURST);
EXPORT_SYMBOL(dm_ratelimit_state);
#endif

M
Milan Broz 已提交
37 38 39 40 41 42 43
/*
 * 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
44 45 46 47 48
static const char *_name = DM_NAME;

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

49 50
static DEFINE_IDR(_minor_idr);

51
static DEFINE_SPINLOCK(_minor_lock);
L
Linus Torvalds 已提交
52
/*
K
Kiyoshi Ueda 已提交
53
 * For bio-based dm.
L
Linus Torvalds 已提交
54 55 56 57 58 59
 * 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
60
	struct bio *bio;
61
	unsigned long start_time;
62
	spinlock_t endio_lock;
L
Linus Torvalds 已提交
63 64 65
};

/*
K
Kiyoshi Ueda 已提交
66
 * For bio-based dm.
L
Linus Torvalds 已提交
67 68 69
 * 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
70
struct dm_target_io {
L
Linus Torvalds 已提交
71 72 73 74 75
	struct dm_io *io;
	struct dm_target *ti;
	union map_info info;
};

K
Kiyoshi Ueda 已提交
76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93
/*
 * 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;
94
	struct dm_rq_target_io *tio;
K
Kiyoshi Ueda 已提交
95 96
};

L
Linus Torvalds 已提交
97 98
union map_info *dm_get_mapinfo(struct bio *bio)
{
A
Alasdair G Kergon 已提交
99
	if (bio && bio->bi_private)
A
Alasdair G Kergon 已提交
100
		return &((struct dm_target_io *)bio->bi_private)->info;
A
Alasdair G Kergon 已提交
101
	return NULL;
L
Linus Torvalds 已提交
102 103
}

104 105 106 107 108 109 110 111
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);

112 113
#define MINOR_ALLOCED ((void *)-1)

L
Linus Torvalds 已提交
114 115 116
/*
 * Bits for the md->flags field.
 */
117
#define DMF_BLOCK_IO_FOR_SUSPEND 0
L
Linus Torvalds 已提交
118
#define DMF_SUSPENDED 1
119
#define DMF_FROZEN 2
J
Jeff Mahoney 已提交
120
#define DMF_FREEING 3
121
#define DMF_DELETING 4
122
#define DMF_NOFLUSH_SUSPENDING 5
123
#define DMF_MERGE_IS_OPTIONAL 6
L
Linus Torvalds 已提交
124

125 126 127
/*
 * Work processed by per-device workqueue.
 */
L
Linus Torvalds 已提交
128
struct mapped_device {
129
	struct rw_semaphore io_lock;
130
	struct mutex suspend_lock;
L
Linus Torvalds 已提交
131 132
	rwlock_t map_lock;
	atomic_t holders;
133
	atomic_t open_count;
L
Linus Torvalds 已提交
134 135 136

	unsigned long flags;

137
	struct request_queue *queue;
138
	unsigned type;
139
	/* Protect queue and type against concurrent access. */
140 141
	struct mutex type_lock;

142 143
	struct target_type *immutable_target_type;

L
Linus Torvalds 已提交
144
	struct gendisk *disk;
M
Mike Anderson 已提交
145
	char name[16];
L
Linus Torvalds 已提交
146 147 148 149 150 151

	void *interface_ptr;

	/*
	 * A list of ios that arrived while we were suspended.
	 */
152
	atomic_t pending[2];
L
Linus Torvalds 已提交
153
	wait_queue_head_t wait;
154
	struct work_struct work;
K
Kiyoshi Ueda 已提交
155
	struct bio_list deferred;
156
	spinlock_t deferred_lock;
L
Linus Torvalds 已提交
157

158
	/*
159
	 * Processing queue (flush)
160 161 162
	 */
	struct workqueue_struct *wq;

L
Linus Torvalds 已提交
163 164 165 166 167 168 169 170 171 172 173
	/*
	 * 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
174 175
	struct bio_set *bs;

L
Linus Torvalds 已提交
176 177 178 179 180
	/*
	 * Event handling.
	 */
	atomic_t event_nr;
	wait_queue_head_t eventq;
M
Mike Anderson 已提交
181 182 183
	atomic_t uevent_seq;
	struct list_head uevent_list;
	spinlock_t uevent_lock; /* Protect access to uevent_list */
L
Linus Torvalds 已提交
184 185 186 187 188

	/*
	 * freeze/thaw support require holding onto a super block
	 */
	struct super_block *frozen_sb;
189
	struct block_device *bdev;
D
Darrick J. Wong 已提交
190 191 192

	/* forced geometry settings */
	struct hd_geometry geometry;
M
Milan Broz 已提交
193 194 195

	/* sysfs handle */
	struct kobject kobj;
196

197 198
	/* zero-length flush that will be cloned and submitted to targets */
	struct bio flush_bio;
L
Linus Torvalds 已提交
199 200
};

K
Kiyoshi Ueda 已提交
201 202 203 204 205 206 207 208 209
/*
 * 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
210
#define MIN_IOS 256
211 212
static struct kmem_cache *_io_cache;
static struct kmem_cache *_tio_cache;
K
Kiyoshi Ueda 已提交
213 214
static struct kmem_cache *_rq_tio_cache;
static struct kmem_cache *_rq_bio_info_cache;
L
Linus Torvalds 已提交
215 216 217

static int __init local_init(void)
{
K
Kiyoshi Ueda 已提交
218
	int r = -ENOMEM;
L
Linus Torvalds 已提交
219 220

	/* allocate a slab for the dm_ios */
A
Alasdair G Kergon 已提交
221
	_io_cache = KMEM_CACHE(dm_io, 0);
L
Linus Torvalds 已提交
222
	if (!_io_cache)
K
Kiyoshi Ueda 已提交
223
		return r;
L
Linus Torvalds 已提交
224 225

	/* allocate a slab for the target ios */
A
Alasdair G Kergon 已提交
226
	_tio_cache = KMEM_CACHE(dm_target_io, 0);
K
Kiyoshi Ueda 已提交
227 228
	if (!_tio_cache)
		goto out_free_io_cache;
L
Linus Torvalds 已提交
229

K
Kiyoshi Ueda 已提交
230 231 232 233 234 235 236 237
	_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
238
	r = dm_uevent_init();
K
Kiyoshi Ueda 已提交
239
	if (r)
K
Kiyoshi Ueda 已提交
240
		goto out_free_rq_bio_info_cache;
M
Mike Anderson 已提交
241

L
Linus Torvalds 已提交
242 243
	_major = major;
	r = register_blkdev(_major, _name);
K
Kiyoshi Ueda 已提交
244 245
	if (r < 0)
		goto out_uevent_exit;
L
Linus Torvalds 已提交
246 247 248 249 250

	if (!_major)
		_major = r;

	return 0;
K
Kiyoshi Ueda 已提交
251 252 253

out_uevent_exit:
	dm_uevent_exit();
K
Kiyoshi Ueda 已提交
254 255 256 257
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
258 259 260 261 262 263
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
264 265 266 267
}

static void local_exit(void)
{
K
Kiyoshi Ueda 已提交
268 269
	kmem_cache_destroy(_rq_bio_info_cache);
	kmem_cache_destroy(_rq_tio_cache);
L
Linus Torvalds 已提交
270 271
	kmem_cache_destroy(_tio_cache);
	kmem_cache_destroy(_io_cache);
272
	unregister_blkdev(_major, _name);
M
Mike Anderson 已提交
273
	dm_uevent_exit();
L
Linus Torvalds 已提交
274 275 276 277 278 279

	_major = 0;

	DMINFO("cleaned up");
}

280
static int (*_inits[])(void) __initdata = {
L
Linus Torvalds 已提交
281 282 283 284
	local_init,
	dm_target_init,
	dm_linear_init,
	dm_stripe_init,
M
Mikulas Patocka 已提交
285
	dm_io_init,
286
	dm_kcopyd_init,
L
Linus Torvalds 已提交
287 288 289
	dm_interface_init,
};

290
static void (*_exits[])(void) = {
L
Linus Torvalds 已提交
291 292 293 294
	local_exit,
	dm_target_exit,
	dm_linear_exit,
	dm_stripe_exit,
M
Mikulas Patocka 已提交
295
	dm_io_exit,
296
	dm_kcopyd_exit,
L
Linus Torvalds 已提交
297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326
	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]();
327 328 329 330 331 332

	/*
	 * Should be empty by this point.
	 */
	idr_remove_all(&_minor_idr);
	idr_destroy(&_minor_idr);
L
Linus Torvalds 已提交
333 334 335 336 337
}

/*
 * Block device functions
 */
M
Mike Anderson 已提交
338 339 340 341 342
int dm_deleting_md(struct mapped_device *md)
{
	return test_bit(DMF_DELETING, &md->flags);
}

A
Al Viro 已提交
343
static int dm_blk_open(struct block_device *bdev, fmode_t mode)
L
Linus Torvalds 已提交
344 345 346
{
	struct mapped_device *md;

J
Jeff Mahoney 已提交
347 348
	spin_lock(&_minor_lock);

A
Al Viro 已提交
349
	md = bdev->bd_disk->private_data;
J
Jeff Mahoney 已提交
350 351 352
	if (!md)
		goto out;

353
	if (test_bit(DMF_FREEING, &md->flags) ||
M
Mike Anderson 已提交
354
	    dm_deleting_md(md)) {
J
Jeff Mahoney 已提交
355 356 357 358
		md = NULL;
		goto out;
	}

L
Linus Torvalds 已提交
359
	dm_get(md);
360
	atomic_inc(&md->open_count);
J
Jeff Mahoney 已提交
361 362 363 364 365

out:
	spin_unlock(&_minor_lock);

	return md ? 0 : -ENXIO;
L
Linus Torvalds 已提交
366 367
}

A
Al Viro 已提交
368
static int dm_blk_close(struct gendisk *disk, fmode_t mode)
L
Linus Torvalds 已提交
369
{
A
Al Viro 已提交
370
	struct mapped_device *md = disk->private_data;
371

372 373
	spin_lock(&_minor_lock);

374
	atomic_dec(&md->open_count);
L
Linus Torvalds 已提交
375
	dm_put(md);
376 377

	spin_unlock(&_minor_lock);
378

L
Linus Torvalds 已提交
379 380 381
	return 0;
}

382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405
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
406 407 408 409 410 411 412
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
413
static int dm_blk_ioctl(struct block_device *bdev, fmode_t mode,
414 415
			unsigned int cmd, unsigned long arg)
{
A
Al Viro 已提交
416
	struct mapped_device *md = bdev->bd_disk->private_data;
417
	struct dm_table *map = dm_get_live_table(md);
418 419 420 421 422 423 424 425 426 427 428 429
	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);

430
	if (dm_suspended_md(md)) {
431 432 433 434 435
		r = -EAGAIN;
		goto out;
	}

	if (tgt->type->ioctl)
436
		r = tgt->type->ioctl(tgt, cmd, arg);
437 438 439 440 441 442 443

out:
	dm_table_put(map);

	return r;
}

A
Alasdair G Kergon 已提交
444
static struct dm_io *alloc_io(struct mapped_device *md)
L
Linus Torvalds 已提交
445 446 447 448
{
	return mempool_alloc(md->io_pool, GFP_NOIO);
}

A
Alasdair G Kergon 已提交
449
static void free_io(struct mapped_device *md, struct dm_io *io)
L
Linus Torvalds 已提交
450 451 452 453
{
	mempool_free(io, md->io_pool);
}

A
Alasdair G Kergon 已提交
454
static void free_tio(struct mapped_device *md, struct dm_target_io *tio)
L
Linus Torvalds 已提交
455 456 457 458
{
	mempool_free(tio, md->tio_pool);
}

K
Kiyoshi Ueda 已提交
459 460
static struct dm_rq_target_io *alloc_rq_tio(struct mapped_device *md,
					    gfp_t gfp_mask)
461
{
K
Kiyoshi Ueda 已提交
462
	return mempool_alloc(md->tio_pool, gfp_mask);
463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479
}

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
480 481 482 483 484 485
static int md_in_flight(struct mapped_device *md)
{
	return atomic_read(&md->pending[READ]) +
	       atomic_read(&md->pending[WRITE]);
}

486 487 488
static void start_io_acct(struct dm_io *io)
{
	struct mapped_device *md = io->md;
T
Tejun Heo 已提交
489
	int cpu;
490
	int rw = bio_data_dir(io->bio);
491 492 493

	io->start_time = jiffies;

T
Tejun Heo 已提交
494 495 496
	cpu = part_stat_lock();
	part_round_stats(cpu, &dm_disk(md)->part0);
	part_stat_unlock();
497 498
	atomic_set(&dm_disk(md)->part0.in_flight[rw],
		atomic_inc_return(&md->pending[rw]));
499 500
}

501
static void end_io_acct(struct dm_io *io)
502 503 504 505
{
	struct mapped_device *md = io->md;
	struct bio *bio = io->bio;
	unsigned long duration = jiffies - io->start_time;
T
Tejun Heo 已提交
506
	int pending, cpu;
507 508
	int rw = bio_data_dir(bio);

T
Tejun Heo 已提交
509 510 511 512
	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();
513

514 515
	/*
	 * After this is decremented the bio must not be touched if it is
516
	 * a flush.
517
	 */
518 519
	pending = atomic_dec_return(&md->pending[rw]);
	atomic_set(&dm_disk(md)->part0.in_flight[rw], pending);
520
	pending += atomic_read(&md->pending[rw^0x1]);
521

522 523 524
	/* nudge anyone waiting on suspend queue */
	if (!pending)
		wake_up(&md->wait);
525 526
}

L
Linus Torvalds 已提交
527 528 529
/*
 * Add the bio to the list of deferred io.
 */
M
Mikulas Patocka 已提交
530
static void queue_io(struct mapped_device *md, struct bio *bio)
L
Linus Torvalds 已提交
531
{
532
	unsigned long flags;
L
Linus Torvalds 已提交
533

534
	spin_lock_irqsave(&md->deferred_lock, flags);
L
Linus Torvalds 已提交
535
	bio_list_add(&md->deferred, bio);
536
	spin_unlock_irqrestore(&md->deferred_lock, flags);
537
	queue_work(md->wq, &md->work);
L
Linus Torvalds 已提交
538 539 540 541 542 543 544
}

/*
 * 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.
 */
545
struct dm_table *dm_get_live_table(struct mapped_device *md)
L
Linus Torvalds 已提交
546 547
{
	struct dm_table *t;
548
	unsigned long flags;
L
Linus Torvalds 已提交
549

550
	read_lock_irqsave(&md->map_lock, flags);
L
Linus Torvalds 已提交
551 552 553
	t = md->map;
	if (t)
		dm_table_get(t);
554
	read_unlock_irqrestore(&md->map_lock, flags);
L
Linus Torvalds 已提交
555 556 557 558

	return t;
}

D
Darrick J. Wong 已提交
559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585
/*
 * 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
586 587 588 589 590 591 592 593 594
/*-----------------------------------------------------------------
 * 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.
 *---------------------------------------------------------------*/

595 596 597 598 599
static int __noflush_suspending(struct mapped_device *md)
{
	return test_bit(DMF_NOFLUSH_SUSPENDING, &md->flags);
}

L
Linus Torvalds 已提交
600 601 602 603
/*
 * Decrements the number of outstanding ios that a bio has been
 * cloned into, completing the original io if necc.
 */
604
static void dec_pending(struct dm_io *io, int error)
L
Linus Torvalds 已提交
605
{
606
	unsigned long flags;
607 608 609
	int io_error;
	struct bio *bio;
	struct mapped_device *md = io->md;
610 611

	/* Push-back supersedes any I/O errors */
612 613 614 615 616 617
	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
618 619

	if (atomic_dec_and_test(&io->io_count)) {
620 621 622 623
		if (io->error == DM_ENDIO_REQUEUE) {
			/*
			 * Target requested pushing back the I/O.
			 */
624
			spin_lock_irqsave(&md->deferred_lock, flags);
625 626 627
			if (__noflush_suspending(md))
				bio_list_add_head(&md->deferred, io->bio);
			else
628 629
				/* noflush suspend was interrupted. */
				io->error = -EIO;
630
			spin_unlock_irqrestore(&md->deferred_lock, flags);
631 632
		}

633 634
		io_error = io->error;
		bio = io->bio;
635 636 637 638 639
		end_io_acct(io);
		free_io(md, io);

		if (io_error == DM_ENDIO_REQUEUE)
			return;
640

641
		if ((bio->bi_rw & REQ_FLUSH) && bio->bi_size) {
642
			/*
643 644
			 * Preflush done for flush with data, reissue
			 * without REQ_FLUSH.
645
			 */
646 647
			bio->bi_rw &= ~REQ_FLUSH;
			queue_io(md, bio);
648
		} else {
649
			/* done with normal IO or empty flush */
650
			trace_block_bio_complete(md->queue, bio, io_error);
651
			bio_endio(bio, io_error);
652
		}
L
Linus Torvalds 已提交
653 654 655
	}
}

656
static void clone_endio(struct bio *bio, int error)
L
Linus Torvalds 已提交
657 658
{
	int r = 0;
A
Alasdair G Kergon 已提交
659
	struct dm_target_io *tio = bio->bi_private;
660
	struct dm_io *io = tio->io;
S
Stefan Bader 已提交
661
	struct mapped_device *md = tio->io->md;
L
Linus Torvalds 已提交
662 663 664 665 666 667 668
	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);
669 670 671 672 673
		if (r < 0 || r == DM_ENDIO_REQUEUE)
			/*
			 * error and requeue request are handled
			 * in dec_pending().
			 */
L
Linus Torvalds 已提交
674
			error = r;
675 676
		else if (r == DM_ENDIO_INCOMPLETE)
			/* The target will handle the io */
677
			return;
678 679 680 681
		else if (r) {
			DMWARN("unimplemented target endio return value: %d", r);
			BUG();
		}
L
Linus Torvalds 已提交
682 683
	}

S
Stefan Bader 已提交
684
	free_tio(md, tio);
685 686
	bio_put(bio);
	dec_pending(io, error);
L
Linus Torvalds 已提交
687 688
}

689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743
/*
 * 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.
 */
744
static void rq_completed(struct mapped_device *md, int rw, int run_queue)
745
{
746
	atomic_dec(&md->pending[rw]);
747 748

	/* nudge anyone waiting on suspend queue */
749
	if (!md_in_flight(md))
750 751 752
		wake_up(&md->wait);

	if (run_queue)
753
		blk_run_queue(md->queue);
754 755 756 757 758 759 760

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

761 762 763 764 765 766 767 768
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
769 770 771 772 773 774 775 776 777 778 779
/*
 * 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;

780
	if (rq->cmd_type == REQ_TYPE_BLOCK_PC) {
K
Kiyoshi Ueda 已提交
781 782 783 784 785 786 787 788 789 790 791 792 793
		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);
794 795
	blk_end_request_all(rq, error);
	rq_completed(md, rw, true);
K
Kiyoshi Ueda 已提交
796 797
}

798 799 800 801 802 803 804
static void dm_unprep_request(struct request *rq)
{
	struct request *clone = rq->special;

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

805
	free_rq_clone(clone);
806 807 808 809 810 811 812
}

/*
 * Requeue the original request of a clone.
 */
void dm_requeue_unmapped_request(struct request *clone)
{
813
	int rw = rq_data_dir(clone);
814 815 816 817 818 819 820 821 822 823 824 825
	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);
	blk_requeue_request(q, rq);
	spin_unlock_irqrestore(q->queue_lock, flags);

826
	rq_completed(md, rw, 0);
827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858
}
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);
}

859
static void dm_done(struct request *clone, int error, bool mapped)
860
{
861
	int r = error;
862 863 864
	struct dm_rq_target_io *tio = clone->end_io_data;
	dm_request_endio_fn rq_end_io = tio->ti->type->rq_end_io;

865 866
	if (mapped && rq_end_io)
		r = rq_end_io(tio->ti, clone, error, &tio->info);
867

868
	if (r <= 0)
869
		/* The target wants to complete the I/O */
870 871
		dm_end_request(clone, r);
	else if (r == DM_ENDIO_INCOMPLETE)
872 873
		/* The target will handle the I/O */
		return;
874
	else if (r == DM_ENDIO_REQUEUE)
875 876 877
		/* The target wants to requeue the I/O */
		dm_requeue_unmapped_request(clone);
	else {
878
		DMWARN("unimplemented target endio return value: %d", r);
879 880 881 882
		BUG();
	}
}

883 884 885 886 887 888 889 890 891 892 893 894 895 896 897
/*
 * 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);
}

898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951
/*
 * 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);
}

952 953 954 955 956 957 958 959 960 961 962 963
/*
 * 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
964
{
965
	sector_t len = max_io_len_target_boundary(sector, ti);
966
	sector_t offset, max_len;
L
Linus Torvalds 已提交
967 968

	/*
969
	 * Does the target need to split even further?
L
Linus Torvalds 已提交
970
	 */
971 972 973 974 975 976 977 978 979 980
	if (ti->max_io_len) {
		offset = dm_target_offset(ti, sector);
		if (unlikely(ti->max_io_len & (ti->max_io_len - 1)))
			max_len = sector_div(offset, ti->max_io_len);
		else
			max_len = offset & (ti->max_io_len - 1);
		max_len = ti->max_io_len - max_len;

		if (len > max_len)
			len = max_len;
L
Linus Torvalds 已提交
981 982 983 984 985
	}

	return len;
}

986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000
int dm_set_target_max_io_len(struct dm_target *ti, sector_t len)
{
	if (len > UINT_MAX) {
		DMERR("Specified maximum size of target IO (%llu) exceeds limit (%u)",
		      (unsigned long long)len, UINT_MAX);
		ti->error = "Maximum size of target IO is too large";
		return -EINVAL;
	}

	ti->max_io_len = (uint32_t) len;

	return 0;
}
EXPORT_SYMBOL_GPL(dm_set_target_max_io_len);

L
Linus Torvalds 已提交
1001
static void __map_bio(struct dm_target *ti, struct bio *clone,
A
Alasdair G Kergon 已提交
1002
		      struct dm_target_io *tio)
L
Linus Torvalds 已提交
1003 1004
{
	int r;
1005
	sector_t sector;
S
Stefan Bader 已提交
1006
	struct mapped_device *md;
L
Linus Torvalds 已提交
1007 1008 1009 1010 1011 1012 1013 1014 1015 1016

	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);
1017
	sector = clone->bi_sector;
L
Linus Torvalds 已提交
1018
	r = ti->type->map(ti, clone, &tio->info);
1019
	if (r == DM_MAPIO_REMAPPED) {
L
Linus Torvalds 已提交
1020
		/* the bio has been remapped so dispatch it */
1021

1022 1023
		trace_block_bio_remap(bdev_get_queue(clone->bi_bdev), clone,
				      tio->io->bio->bi_bdev->bd_dev, sector);
1024

L
Linus Torvalds 已提交
1025
		generic_make_request(clone);
1026 1027
	} 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
1028 1029
		md = tio->io->md;
		dec_pending(tio->io, r);
L
Linus Torvalds 已提交
1030
		bio_put(clone);
S
Stefan Bader 已提交
1031
		free_tio(md, tio);
1032 1033 1034
	} else if (r) {
		DMWARN("unimplemented target map return value: %d", r);
		BUG();
L
Linus Torvalds 已提交
1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048
	}
}

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;
};

/*
1049
 * Creates a little bio that just does part of a bvec.
L
Linus Torvalds 已提交
1050 1051 1052
 */
static struct bio *split_bvec(struct bio *bio, sector_t sector,
			      unsigned short idx, unsigned int offset,
S
Stefan Bader 已提交
1053
			      unsigned int len, struct bio_set *bs)
L
Linus Torvalds 已提交
1054 1055 1056 1057
{
	struct bio *clone;
	struct bio_vec *bv = bio->bi_io_vec + idx;

S
Stefan Bader 已提交
1058
	clone = bio_alloc_bioset(GFP_NOIO, 1, bs);
L
Linus Torvalds 已提交
1059 1060 1061 1062
	*clone->bi_io_vec = *bv;

	clone->bi_sector = sector;
	clone->bi_bdev = bio->bi_bdev;
1063
	clone->bi_rw = bio->bi_rw;
L
Linus Torvalds 已提交
1064 1065 1066 1067
	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
1068
	clone->bi_flags |= 1 << BIO_CLONED;
L
Linus Torvalds 已提交
1069

M
Martin K. Petersen 已提交
1070
	if (bio_integrity(bio)) {
1071
		bio_integrity_clone(clone, bio, GFP_NOIO);
M
Martin K. Petersen 已提交
1072 1073 1074 1075
		bio_integrity_trim(clone,
				   bio_sector_offset(bio, idx, offset), len);
	}

L
Linus Torvalds 已提交
1076 1077 1078 1079 1080 1081 1082 1083
	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
1084
			     unsigned int len, struct bio_set *bs)
L
Linus Torvalds 已提交
1085 1086 1087
{
	struct bio *clone;

S
Stefan Bader 已提交
1088 1089
	clone = bio_alloc_bioset(GFP_NOIO, bio->bi_max_vecs, bs);
	__bio_clone(clone, bio);
L
Linus Torvalds 已提交
1090 1091 1092 1093 1094 1095
	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
1096
	if (bio_integrity(bio)) {
1097
		bio_integrity_clone(clone, bio, GFP_NOIO);
M
Martin K. Petersen 已提交
1098 1099 1100 1101 1102 1103

		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
1104 1105 1106
	return clone;
}

1107 1108
static struct dm_target_io *alloc_tio(struct clone_info *ci,
				      struct dm_target *ti)
1109
{
1110
	struct dm_target_io *tio = mempool_alloc(ci->md->tio_pool, GFP_NOIO);
1111 1112 1113 1114

	tio->io = ci->io;
	tio->ti = ti;
	memset(&tio->info, 0, sizeof(tio->info));
1115 1116 1117 1118

	return tio;
}

1119
static void __issue_target_request(struct clone_info *ci, struct dm_target *ti,
1120
				   unsigned request_nr, sector_t len)
1121 1122 1123 1124
{
	struct dm_target_io *tio = alloc_tio(ci, ti);
	struct bio *clone;

1125
	tio->info.target_request_nr = request_nr;
1126

1127 1128 1129 1130 1131 1132
	/*
	 * 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);
1133
	__bio_clone(clone, ci->bio);
1134 1135 1136 1137
	if (len) {
		clone->bi_sector = ci->sector;
		clone->bi_size = to_bytes(len);
	}
1138 1139 1140 1141

	__map_bio(ti, clone, tio);
}

1142
static void __issue_target_requests(struct clone_info *ci, struct dm_target *ti,
1143
				    unsigned num_requests, sector_t len)
1144 1145 1146 1147
{
	unsigned request_nr;

	for (request_nr = 0; request_nr < num_requests; request_nr++)
1148
		__issue_target_request(ci, ti, request_nr, len);
1149 1150
}

1151
static int __clone_and_map_empty_flush(struct clone_info *ci)
1152
{
1153
	unsigned target_nr = 0;
1154 1155
	struct dm_target *ti;

1156
	BUG_ON(bio_has_data(ci->bio));
1157
	while ((ti = dm_table_get_target(ci->map, target_nr++)))
1158
		__issue_target_requests(ci, ti, ti->num_flush_requests, 0);
1159 1160 1161 1162

	return 0;
}

M
Mike Snitzer 已提交
1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181
/*
 * 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;
1182
	sector_t len;
M
Mike Snitzer 已提交
1183

1184 1185 1186 1187
	do {
		ti = dm_table_find_target(ci->map, ci->sector);
		if (!dm_target_is_valid(ti))
			return -EIO;
M
Mike Snitzer 已提交
1188 1189

		/*
1190
		 * Even though the device advertised discard support,
M
Mike Snitzer 已提交
1191 1192
		 * that does not mean every target supports it, and
		 * reconfiguration might also have changed that since the
1193
		 * check was performed.
M
Mike Snitzer 已提交
1194
		 */
1195 1196
		if (!ti->num_discard_requests)
			return -EOPNOTSUPP;
M
Mike Snitzer 已提交
1197

1198 1199 1200 1201
		if (!ti->split_discard_requests)
			len = min(ci->sector_count, max_io_len_target_boundary(ci->sector, ti));
		else
			len = min(ci->sector_count, max_io_len(ci->sector, ti));
1202

1203 1204 1205 1206
		__issue_target_requests(ci, ti, ti->num_discard_requests, len);

		ci->sector += len;
	} while (ci->sector_count -= len);
M
Mike Snitzer 已提交
1207 1208 1209 1210

	return 0;
}

1211
static int __clone_and_map(struct clone_info *ci)
L
Linus Torvalds 已提交
1212 1213
{
	struct bio *clone, *bio = ci->bio;
1214 1215
	struct dm_target *ti;
	sector_t len = 0, max;
A
Alasdair G Kergon 已提交
1216
	struct dm_target_io *tio;
L
Linus Torvalds 已提交
1217

M
Mike Snitzer 已提交
1218 1219 1220
	if (unlikely(bio->bi_rw & REQ_DISCARD))
		return __clone_and_map_discard(ci);

1221 1222 1223 1224
	ti = dm_table_find_target(ci->map, ci->sector);
	if (!dm_target_is_valid(ti))
		return -EIO;

1225
	max = max_io_len(ci->sector, ti);
1226

L
Linus Torvalds 已提交
1227 1228 1229 1230 1231
	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
1232
		__clone_and_map_simple(ci, ti);
L
Linus Torvalds 已提交
1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252

	} 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
1253
		tio = alloc_tio(ci, ti);
S
Stefan Bader 已提交
1254 1255
		clone = clone_bio(bio, ci->sector, ci->idx, i - ci->idx, len,
				  ci->md->bs);
L
Linus Torvalds 已提交
1256 1257 1258 1259 1260 1261 1262 1263
		__map_bio(ti, clone, tio);

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

	} else {
		/*
1264
		 * Handle a bvec that must be split between two or more targets.
L
Linus Torvalds 已提交
1265 1266
		 */
		struct bio_vec *bv = bio->bi_io_vec + ci->idx;
1267 1268
		sector_t remaining = to_sector(bv->bv_len);
		unsigned int offset = 0;
L
Linus Torvalds 已提交
1269

1270 1271 1272
		do {
			if (offset) {
				ti = dm_table_find_target(ci->map, ci->sector);
1273 1274 1275
				if (!dm_target_is_valid(ti))
					return -EIO;

1276
				max = max_io_len(ci->sector, ti);
1277 1278 1279 1280
			}

			len = min(remaining, max);

M
Mike Snitzer 已提交
1281
			tio = alloc_tio(ci, ti);
1282
			clone = split_bvec(bio, ci->sector, ci->idx,
S
Stefan Bader 已提交
1283 1284
					   bv->bv_offset + offset, len,
					   ci->md->bs);
1285 1286 1287 1288 1289 1290 1291

			__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
1292 1293 1294

		ci->idx++;
	}
1295 1296

	return 0;
L
Linus Torvalds 已提交
1297 1298 1299
}

/*
M
Mikulas Patocka 已提交
1300
 * Split the bio into several clones and submit it to targets.
L
Linus Torvalds 已提交
1301
 */
1302
static void __split_and_process_bio(struct mapped_device *md, struct bio *bio)
L
Linus Torvalds 已提交
1303 1304
{
	struct clone_info ci;
1305
	int error = 0;
L
Linus Torvalds 已提交
1306

1307
	ci.map = dm_get_live_table(md);
1308
	if (unlikely(!ci.map)) {
1309
		bio_io_error(bio);
1310 1311
		return;
	}
1312

L
Linus Torvalds 已提交
1313 1314 1315 1316 1317 1318
	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;
1319
	spin_lock_init(&ci.io->endio_lock);
L
Linus Torvalds 已提交
1320 1321 1322
	ci.sector = bio->bi_sector;
	ci.idx = bio->bi_idx;

1323
	start_io_acct(ci.io);
1324 1325 1326 1327 1328 1329
	if (bio->bi_rw & REQ_FLUSH) {
		ci.bio = &ci.md->flush_bio;
		ci.sector_count = 0;
		error = __clone_and_map_empty_flush(&ci);
		/* dec_pending submits any data associated with flush */
	} else {
1330
		ci.bio = bio;
1331
		ci.sector_count = bio_sectors(bio);
1332
		while (ci.sector_count && !error)
1333 1334
			error = __clone_and_map(&ci);
	}
L
Linus Torvalds 已提交
1335 1336

	/* drop the extra reference count */
1337
	dec_pending(ci.io, error);
L
Linus Torvalds 已提交
1338 1339 1340 1341 1342 1343
	dm_table_put(ci.map);
}
/*-----------------------------------------------------------------
 * CRUD END
 *---------------------------------------------------------------*/

M
Milan Broz 已提交
1344 1345 1346 1347 1348
static int dm_merge_bvec(struct request_queue *q,
			 struct bvec_merge_data *bvm,
			 struct bio_vec *biovec)
{
	struct mapped_device *md = q->queuedata;
1349
	struct dm_table *map = dm_get_live_table(md);
M
Milan Broz 已提交
1350 1351
	struct dm_target *ti;
	sector_t max_sectors;
1352
	int max_size = 0;
M
Milan Broz 已提交
1353 1354

	if (unlikely(!map))
1355
		goto out;
M
Milan Broz 已提交
1356 1357

	ti = dm_table_find_target(map, bvm->bi_sector);
1358 1359
	if (!dm_target_is_valid(ti))
		goto out_table;
M
Milan Broz 已提交
1360 1361 1362 1363

	/*
	 * Find maximum amount of I/O that won't need splitting
	 */
1364
	max_sectors = min(max_io_len(bvm->bi_sector, ti),
M
Milan Broz 已提交
1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376
			  (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);
1377 1378 1379 1380 1381 1382 1383 1384 1385 1386
	/*
	 * 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
1387

1388
out_table:
1389 1390 1391
	dm_table_put(map);

out:
M
Milan Broz 已提交
1392 1393 1394 1395 1396 1397 1398 1399 1400
	/*
	 * 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
1401 1402 1403 1404
/*
 * The request function that just remaps the bio built up by
 * dm_merge_bvec.
 */
1405
static void _dm_request(struct request_queue *q, struct bio *bio)
L
Linus Torvalds 已提交
1406
{
1407
	int rw = bio_data_dir(bio);
L
Linus Torvalds 已提交
1408
	struct mapped_device *md = q->queuedata;
T
Tejun Heo 已提交
1409
	int cpu;
L
Linus Torvalds 已提交
1410

1411
	down_read(&md->io_lock);
L
Linus Torvalds 已提交
1412

T
Tejun Heo 已提交
1413 1414 1415 1416
	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();
1417

1418 1419
	/* if we're suspended, we have to queue this io for later */
	if (unlikely(test_bit(DMF_BLOCK_IO_FOR_SUSPEND, &md->flags))) {
1420
		up_read(&md->io_lock);
L
Linus Torvalds 已提交
1421

1422 1423 1424
		if (bio_rw(bio) != READA)
			queue_io(md, bio);
		else
A
Alasdair G Kergon 已提交
1425
			bio_io_error(bio);
1426
		return;
L
Linus Torvalds 已提交
1427 1428
	}

1429
	__split_and_process_bio(md, bio);
1430
	up_read(&md->io_lock);
1431
	return;
1432 1433 1434 1435 1436 1437 1438
}

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

1439
static void dm_request(struct request_queue *q, struct bio *bio)
1440 1441 1442 1443
{
	struct mapped_device *md = q->queuedata;

	if (dm_request_based(md))
1444 1445 1446
		blk_queue_bio(q, bio);
	else
		_dm_request(q, bio);
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 1486 1487 1488 1489 1490 1491 1492 1493
}

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)
{
1494
	int r;
1495

1496 1497 1498 1499
	r = blk_rq_prep_clone(clone, rq, tio->md->bs, GFP_ATOMIC,
			      dm_rq_bio_constructor, tio);
	if (r)
		return r;
1500

1501 1502 1503 1504
	clone->cmd = rq->cmd;
	clone->cmd_len = rq->cmd_len;
	clone->sense = rq->sense;
	clone->buffer = rq->buffer;
1505 1506 1507 1508 1509 1510
	clone->end_io = end_clone_request;
	clone->end_io_data = tio;

	return 0;
}

K
Kiyoshi Ueda 已提交
1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536
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;
}

1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549
/*
 * 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
1550 1551
	clone = clone_rq(rq, md, GFP_ATOMIC);
	if (!clone)
1552 1553 1554 1555 1556 1557 1558 1559
		return BLKPREP_DEFER;

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

	return BLKPREP_OK;
}

1560 1561 1562 1563 1564 1565 1566
/*
 * 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)
1567
{
1568
	int r, requeued = 0;
1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587
	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 */
1588 1589
		trace_block_rq_remap(clone->q, clone, disk_devt(dm_disk(md)),
				     blk_rq_pos(tio->orig));
1590 1591 1592 1593 1594
		dm_dispatch_request(clone);
		break;
	case DM_MAPIO_REQUEUE:
		/* The target wants to requeue the I/O */
		dm_requeue_unmapped_request(clone);
1595
		requeued = 1;
1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606
		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;
	}
1607 1608

	return requeued;
1609 1610 1611 1612 1613 1614 1615 1616 1617
}

/*
 * 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;
1618
	struct dm_table *map = dm_get_live_table(md);
1619
	struct dm_target *ti;
1620
	struct request *rq, *clone;
1621
	sector_t pos;
1622 1623

	/*
1624 1625 1626 1627
	 * 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().
1628
	 */
J
Jens Axboe 已提交
1629
	while (!blk_queue_stopped(q)) {
1630 1631
		rq = blk_peek_request(q);
		if (!rq)
J
Jens Axboe 已提交
1632
			goto delay_and_out;
1633

1634 1635 1636 1637 1638 1639 1640
		/* 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));
1641

1642
		if (ti->type->busy && ti->type->busy(ti))
J
Jens Axboe 已提交
1643
			goto delay_and_out;
1644 1645

		blk_start_request(rq);
1646 1647 1648
		clone = rq->special;
		atomic_inc(&md->pending[rq_data_dir(clone)]);

1649
		spin_unlock(q->queue_lock);
1650 1651 1652
		if (map_request(ti, clone, md))
			goto requeued;

1653 1654
		BUG_ON(!irqs_disabled());
		spin_lock(q->queue_lock);
1655 1656 1657 1658
	}

	goto out;

1659
requeued:
1660 1661
	BUG_ON(!irqs_disabled());
	spin_lock(q->queue_lock);
1662

J
Jens Axboe 已提交
1663 1664
delay_and_out:
	blk_delay_queue(q, HZ / 10);
1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680
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;
1681
	struct dm_table *map = dm_get_live_table(md);
1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692

	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;
}

L
Linus Torvalds 已提交
1693 1694
static int dm_any_congested(void *congested_data, int bdi_bits)
{
1695 1696 1697
	int r = bdi_bits;
	struct mapped_device *md = congested_data;
	struct dm_table *map;
L
Linus Torvalds 已提交
1698

1699
	if (!test_bit(DMF_BLOCK_IO_FOR_SUSPEND, &md->flags)) {
1700
		map = dm_get_live_table(md);
1701
		if (map) {
1702 1703 1704 1705 1706 1707 1708 1709 1710 1711
			/*
			 * 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);

1712 1713 1714 1715
			dm_table_put(map);
		}
	}

L
Linus Torvalds 已提交
1716 1717 1718 1719 1720 1721
	return r;
}

/*-----------------------------------------------------------------
 * An IDR is used to keep track of allocated minor numbers.
 *---------------------------------------------------------------*/
1722
static void free_minor(int minor)
L
Linus Torvalds 已提交
1723
{
1724
	spin_lock(&_minor_lock);
L
Linus Torvalds 已提交
1725
	idr_remove(&_minor_idr, minor);
1726
	spin_unlock(&_minor_lock);
L
Linus Torvalds 已提交
1727 1728 1729 1730 1731
}

/*
 * See if the device with a specific minor # is free.
 */
1732
static int specific_minor(int minor)
L
Linus Torvalds 已提交
1733 1734 1735 1736 1737 1738
{
	int r, m;

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

J
Jeff Mahoney 已提交
1739 1740 1741 1742
	r = idr_pre_get(&_minor_idr, GFP_KERNEL);
	if (!r)
		return -ENOMEM;

1743
	spin_lock(&_minor_lock);
L
Linus Torvalds 已提交
1744 1745 1746 1747 1748 1749

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

1750
	r = idr_get_new_above(&_minor_idr, MINOR_ALLOCED, minor, &m);
J
Jeff Mahoney 已提交
1751
	if (r)
L
Linus Torvalds 已提交
1752 1753 1754 1755 1756 1757 1758 1759 1760
		goto out;

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

out:
1761
	spin_unlock(&_minor_lock);
L
Linus Torvalds 已提交
1762 1763 1764
	return r;
}

1765
static int next_free_minor(int *minor)
L
Linus Torvalds 已提交
1766
{
1767
	int r, m;
L
Linus Torvalds 已提交
1768 1769

	r = idr_pre_get(&_minor_idr, GFP_KERNEL);
J
Jeff Mahoney 已提交
1770 1771 1772
	if (!r)
		return -ENOMEM;

1773
	spin_lock(&_minor_lock);
L
Linus Torvalds 已提交
1774

1775
	r = idr_get_new(&_minor_idr, MINOR_ALLOCED, &m);
1776
	if (r)
L
Linus Torvalds 已提交
1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787
		goto out;

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

	*minor = m;

out:
1788
	spin_unlock(&_minor_lock);
L
Linus Torvalds 已提交
1789 1790 1791
	return r;
}

1792
static const struct block_device_operations dm_blk_dops;
L
Linus Torvalds 已提交
1793

1794 1795
static void dm_wq_work(struct work_struct *work);

1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816
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);
	blk_queue_merge_bvec(md->queue, dm_merge_bvec);
}

L
Linus Torvalds 已提交
1817 1818 1819
/*
 * Allocate and initialise a blank device with a given minor.
 */
1820
static struct mapped_device *alloc_dev(int minor)
L
Linus Torvalds 已提交
1821 1822
{
	int r;
1823
	struct mapped_device *md = kzalloc(sizeof(*md), GFP_KERNEL);
1824
	void *old_md;
L
Linus Torvalds 已提交
1825 1826 1827 1828 1829 1830

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

1831
	if (!try_module_get(THIS_MODULE))
M
Milan Broz 已提交
1832
		goto bad_module_get;
1833

L
Linus Torvalds 已提交
1834
	/* get a minor number for the dev */
1835
	if (minor == DM_ANY_MINOR)
1836
		r = next_free_minor(&minor);
1837
	else
1838
		r = specific_minor(minor);
L
Linus Torvalds 已提交
1839
	if (r < 0)
M
Milan Broz 已提交
1840
		goto bad_minor;
L
Linus Torvalds 已提交
1841

1842
	md->type = DM_TYPE_NONE;
1843
	init_rwsem(&md->io_lock);
1844
	mutex_init(&md->suspend_lock);
1845
	mutex_init(&md->type_lock);
1846
	spin_lock_init(&md->deferred_lock);
L
Linus Torvalds 已提交
1847 1848
	rwlock_init(&md->map_lock);
	atomic_set(&md->holders, 1);
1849
	atomic_set(&md->open_count, 0);
L
Linus Torvalds 已提交
1850
	atomic_set(&md->event_nr, 0);
M
Mike Anderson 已提交
1851 1852 1853
	atomic_set(&md->uevent_seq, 0);
	INIT_LIST_HEAD(&md->uevent_list);
	spin_lock_init(&md->uevent_lock);
L
Linus Torvalds 已提交
1854

1855
	md->queue = blk_alloc_queue(GFP_KERNEL);
L
Linus Torvalds 已提交
1856
	if (!md->queue)
M
Milan Broz 已提交
1857
		goto bad_queue;
L
Linus Torvalds 已提交
1858

1859
	dm_init_md_queue(md);
S
Stefan Bader 已提交
1860

L
Linus Torvalds 已提交
1861 1862
	md->disk = alloc_disk(1);
	if (!md->disk)
M
Milan Broz 已提交
1863
		goto bad_disk;
L
Linus Torvalds 已提交
1864

1865 1866
	atomic_set(&md->pending[0], 0);
	atomic_set(&md->pending[1], 0);
1867
	init_waitqueue_head(&md->wait);
1868
	INIT_WORK(&md->work, dm_wq_work);
1869 1870
	init_waitqueue_head(&md->eventq);

L
Linus Torvalds 已提交
1871 1872 1873 1874 1875 1876 1877
	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
1878
	format_dev_t(md->name, MKDEV(_major, minor));
L
Linus Torvalds 已提交
1879

1880 1881
	md->wq = alloc_workqueue("kdmflush",
				 WQ_NON_REENTRANT | WQ_MEM_RECLAIM, 0);
1882 1883 1884
	if (!md->wq)
		goto bad_thread;

M
Mikulas Patocka 已提交
1885 1886 1887 1888
	md->bdev = bdget_disk(md->disk, 0);
	if (!md->bdev)
		goto bad_bdev;

1889 1890 1891 1892
	bio_init(&md->flush_bio);
	md->flush_bio.bi_bdev = md->bdev;
	md->flush_bio.bi_rw = WRITE_FLUSH;

1893
	/* Populate the mapping, nobody knows we exist yet */
1894
	spin_lock(&_minor_lock);
1895
	old_md = idr_replace(&_minor_idr, md, minor);
1896
	spin_unlock(&_minor_lock);
1897 1898 1899

	BUG_ON(old_md != MINOR_ALLOCED);

L
Linus Torvalds 已提交
1900 1901
	return md;

M
Mikulas Patocka 已提交
1902 1903
bad_bdev:
	destroy_workqueue(md->wq);
1904
bad_thread:
1905
	del_gendisk(md->disk);
1906
	put_disk(md->disk);
M
Milan Broz 已提交
1907
bad_disk:
1908
	blk_cleanup_queue(md->queue);
M
Milan Broz 已提交
1909
bad_queue:
L
Linus Torvalds 已提交
1910
	free_minor(minor);
M
Milan Broz 已提交
1911
bad_minor:
1912
	module_put(THIS_MODULE);
M
Milan Broz 已提交
1913
bad_module_get:
L
Linus Torvalds 已提交
1914 1915 1916 1917
	kfree(md);
	return NULL;
}

J
Jun'ichi Nomura 已提交
1918 1919
static void unlock_fs(struct mapped_device *md);

L
Linus Torvalds 已提交
1920 1921
static void free_dev(struct mapped_device *md)
{
1922
	int minor = MINOR(disk_devt(md->disk));
1923

M
Mikulas Patocka 已提交
1924 1925
	unlock_fs(md);
	bdput(md->bdev);
1926
	destroy_workqueue(md->wq);
K
Kiyoshi Ueda 已提交
1927 1928 1929 1930 1931 1932
	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
1933
	blk_integrity_unregister(md->disk);
L
Linus Torvalds 已提交
1934
	del_gendisk(md->disk);
1935
	free_minor(minor);
J
Jeff Mahoney 已提交
1936 1937 1938 1939 1940

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

L
Linus Torvalds 已提交
1941
	put_disk(md->disk);
1942
	blk_cleanup_queue(md->queue);
1943
	module_put(THIS_MODULE);
L
Linus Torvalds 已提交
1944 1945 1946
	kfree(md);
}

K
Kiyoshi Ueda 已提交
1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969
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
1970 1971 1972 1973 1974
/*
 * Bind a table to the device.
 */
static void event_callback(void *context)
{
M
Mike Anderson 已提交
1975 1976
	unsigned long flags;
	LIST_HEAD(uevents);
L
Linus Torvalds 已提交
1977 1978
	struct mapped_device *md = (struct mapped_device *) context;

M
Mike Anderson 已提交
1979 1980 1981 1982
	spin_lock_irqsave(&md->uevent_lock, flags);
	list_splice_init(&md->uevent_list, &uevents);
	spin_unlock_irqrestore(&md->uevent_lock, flags);

1983
	dm_send_uevents(&uevents, &disk_to_dev(md->disk)->kobj);
M
Mike Anderson 已提交
1984

L
Linus Torvalds 已提交
1985 1986 1987 1988
	atomic_inc(&md->event_nr);
	wake_up(&md->eventq);
}

1989 1990 1991
/*
 * Protected by md->suspend_lock obtained by dm_swap_table().
 */
1992
static void __set_size(struct mapped_device *md, sector_t size)
L
Linus Torvalds 已提交
1993
{
1994
	set_capacity(md->disk, size);
L
Linus Torvalds 已提交
1995

1996
	i_size_write(md->bdev->bd_inode, (loff_t)size << SECTOR_SHIFT);
L
Linus Torvalds 已提交
1997 1998
}

1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051
/*
 * Return 1 if the queue has a compulsory merge_bvec_fn function.
 *
 * If this function returns 0, then the device is either a non-dm
 * device without a merge_bvec_fn, or it is a dm device that is
 * able to split any bios it receives that are too big.
 */
int dm_queue_merge_is_compulsory(struct request_queue *q)
{
	struct mapped_device *dev_md;

	if (!q->merge_bvec_fn)
		return 0;

	if (q->make_request_fn == dm_request) {
		dev_md = q->queuedata;
		if (test_bit(DMF_MERGE_IS_OPTIONAL, &dev_md->flags))
			return 0;
	}

	return 1;
}

static int dm_device_merge_is_compulsory(struct dm_target *ti,
					 struct dm_dev *dev, sector_t start,
					 sector_t len, void *data)
{
	struct block_device *bdev = dev->bdev;
	struct request_queue *q = bdev_get_queue(bdev);

	return dm_queue_merge_is_compulsory(q);
}

/*
 * Return 1 if it is acceptable to ignore merge_bvec_fn based
 * on the properties of the underlying devices.
 */
static int dm_table_merge_is_optional(struct dm_table *table)
{
	unsigned i = 0;
	struct dm_target *ti;

	while (i < dm_table_get_num_targets(table)) {
		ti = dm_table_get_target(table, i++);

		if (ti->type->iterate_devices &&
		    ti->type->iterate_devices(ti, dm_device_merge_is_compulsory, NULL))
			return 0;
	}

	return 1;
}

2052 2053 2054 2055 2056
/*
 * 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
2057
{
2058
	struct dm_table *old_map;
2059
	struct request_queue *q = md->queue;
L
Linus Torvalds 已提交
2060
	sector_t size;
2061
	unsigned long flags;
2062
	int merge_is_optional;
L
Linus Torvalds 已提交
2063 2064

	size = dm_table_get_size(t);
D
Darrick J. Wong 已提交
2065 2066 2067 2068 2069 2070 2071

	/*
	 * 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
2072
	__set_size(md, size);
2073

2074 2075
	dm_table_event_callback(t, event_callback, md);

K
Kiyoshi Ueda 已提交
2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087
	/*
	 * 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);

2088 2089
	merge_is_optional = dm_table_merge_is_optional(t);

2090
	write_lock_irqsave(&md->map_lock, flags);
2091
	old_map = md->map;
L
Linus Torvalds 已提交
2092
	md->map = t;
2093 2094
	md->immutable_target_type = dm_table_get_immutable_target_type(t);

2095
	dm_table_set_restrictions(t, q, limits);
2096 2097 2098 2099
	if (merge_is_optional)
		set_bit(DMF_MERGE_IS_OPTIONAL, &md->flags);
	else
		clear_bit(DMF_MERGE_IS_OPTIONAL, &md->flags);
2100
	write_unlock_irqrestore(&md->map_lock, flags);
L
Linus Torvalds 已提交
2101

2102
	return old_map;
L
Linus Torvalds 已提交
2103 2104
}

2105 2106 2107 2108
/*
 * Returns unbound table for the caller to free.
 */
static struct dm_table *__unbind(struct mapped_device *md)
L
Linus Torvalds 已提交
2109 2110
{
	struct dm_table *map = md->map;
2111
	unsigned long flags;
L
Linus Torvalds 已提交
2112 2113

	if (!map)
2114
		return NULL;
L
Linus Torvalds 已提交
2115 2116

	dm_table_event_callback(map, NULL, NULL);
2117
	write_lock_irqsave(&md->map_lock, flags);
L
Linus Torvalds 已提交
2118
	md->map = NULL;
2119
	write_unlock_irqrestore(&md->map_lock, flags);
2120 2121

	return map;
L
Linus Torvalds 已提交
2122 2123 2124 2125 2126
}

/*
 * Constructor for a new device.
 */
2127
int dm_create(int minor, struct mapped_device **result)
L
Linus Torvalds 已提交
2128 2129 2130
{
	struct mapped_device *md;

2131
	md = alloc_dev(minor);
L
Linus Torvalds 已提交
2132 2133 2134
	if (!md)
		return -ENXIO;

M
Milan Broz 已提交
2135 2136
	dm_sysfs_init(md);

L
Linus Torvalds 已提交
2137 2138 2139 2140
	*result = md;
	return 0;
}

2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164
/*
 * 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;
}

2165 2166 2167 2168 2169
struct target_type *dm_get_immutable_target_type(struct mapped_device *md)
{
	return md->immutable_target_type;
}

2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209
/*
 * 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;
	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;
}

2210
static struct mapped_device *dm_find_md(dev_t dev)
L
Linus Torvalds 已提交
2211 2212 2213 2214 2215 2216 2217
{
	struct mapped_device *md;
	unsigned minor = MINOR(dev);

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

2218
	spin_lock(&_minor_lock);
L
Linus Torvalds 已提交
2219 2220

	md = idr_find(&_minor_idr, minor);
J
Jeff Mahoney 已提交
2221
	if (md && (md == MINOR_ALLOCED ||
2222
		   (MINOR(disk_devt(dm_disk(md))) != minor) ||
2223
		   dm_deleting_md(md) ||
A
Alasdair G Kergon 已提交
2224
		   test_bit(DMF_FREEING, &md->flags))) {
2225
		md = NULL;
J
Jeff Mahoney 已提交
2226 2227
		goto out;
	}
L
Linus Torvalds 已提交
2228

J
Jeff Mahoney 已提交
2229
out:
2230
	spin_unlock(&_minor_lock);
L
Linus Torvalds 已提交
2231

2232 2233 2234
	return md;
}

2235 2236 2237 2238 2239 2240 2241 2242 2243
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
2244
EXPORT_SYMBOL_GPL(dm_get_md);
2245

A
Alasdair G Kergon 已提交
2246
void *dm_get_mdptr(struct mapped_device *md)
2247
{
A
Alasdair G Kergon 已提交
2248
	return md->interface_ptr;
L
Linus Torvalds 已提交
2249 2250 2251 2252 2253 2254 2255 2256 2257 2258
}

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);
2259
	BUG_ON(test_bit(DMF_FREEING, &md->flags));
L
Linus Torvalds 已提交
2260 2261
}

2262 2263 2264 2265 2266 2267
const char *dm_device_name(struct mapped_device *md)
{
	return md->name;
}
EXPORT_SYMBOL_GPL(dm_device_name);

2268
static void __dm_destroy(struct mapped_device *md, bool wait)
L
Linus Torvalds 已提交
2269
{
M
Mike Anderson 已提交
2270
	struct dm_table *map;
L
Linus Torvalds 已提交
2271

2272
	might_sleep();
J
Jeff Mahoney 已提交
2273

2274 2275 2276 2277 2278 2279 2280 2281 2282
	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
2283
	}
2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316

	/*
	 * 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
2317
}
E
Edward Goggin 已提交
2318
EXPORT_SYMBOL_GPL(dm_put);
L
Linus Torvalds 已提交
2319

2320
static int dm_wait_for_completion(struct mapped_device *md, int interruptible)
2321 2322
{
	int r = 0;
2323 2324 2325
	DECLARE_WAITQUEUE(wait, current);

	add_wait_queue(&md->wait, &wait);
2326 2327

	while (1) {
2328
		set_current_state(interruptible);
2329

2330
		if (!md_in_flight(md))
2331 2332
			break;

2333 2334
		if (interruptible == TASK_INTERRUPTIBLE &&
		    signal_pending(current)) {
2335 2336 2337 2338 2339 2340 2341 2342
			r = -EINTR;
			break;
		}

		io_schedule();
	}
	set_current_state(TASK_RUNNING);

2343 2344
	remove_wait_queue(&md->wait, &wait);

2345 2346 2347
	return r;
}

L
Linus Torvalds 已提交
2348 2349 2350
/*
 * Process the deferred bios
 */
2351
static void dm_wq_work(struct work_struct *work)
L
Linus Torvalds 已提交
2352
{
2353 2354
	struct mapped_device *md = container_of(work, struct mapped_device,
						work);
2355
	struct bio *c;
L
Linus Torvalds 已提交
2356

2357
	down_read(&md->io_lock);
2358

2359
	while (!test_bit(DMF_BLOCK_IO_FOR_SUSPEND, &md->flags)) {
A
Alasdair G Kergon 已提交
2360 2361 2362 2363
		spin_lock_irq(&md->deferred_lock);
		c = bio_list_pop(&md->deferred);
		spin_unlock_irq(&md->deferred_lock);

2364
		if (!c)
A
Alasdair G Kergon 已提交
2365
			break;
2366

2367
		up_read(&md->io_lock);
2368

K
Kiyoshi Ueda 已提交
2369 2370
		if (dm_request_based(md))
			generic_make_request(c);
2371 2372
		else
			__split_and_process_bio(md, c);
2373

2374
		down_read(&md->io_lock);
2375
	}
M
Milan Broz 已提交
2376

2377
	up_read(&md->io_lock);
L
Linus Torvalds 已提交
2378 2379
}

2380
static void dm_queue_flush(struct mapped_device *md)
2381
{
2382 2383
	clear_bit(DMF_BLOCK_IO_FOR_SUSPEND, &md->flags);
	smp_mb__after_clear_bit();
2384
	queue_work(md->wq, &md->work);
2385 2386
}

L
Linus Torvalds 已提交
2387
/*
2388
 * Swap in a new table, returning the old one for the caller to destroy.
L
Linus Torvalds 已提交
2389
 */
2390
struct dm_table *dm_swap_table(struct mapped_device *md, struct dm_table *table)
L
Linus Torvalds 已提交
2391
{
2392
	struct dm_table *map = ERR_PTR(-EINVAL);
2393
	struct queue_limits limits;
2394
	int r;
L
Linus Torvalds 已提交
2395

2396
	mutex_lock(&md->suspend_lock);
L
Linus Torvalds 已提交
2397 2398

	/* device must be suspended */
2399
	if (!dm_suspended_md(md))
2400
		goto out;
L
Linus Torvalds 已提交
2401

2402
	r = dm_calculate_queue_limits(table, &limits);
2403 2404
	if (r) {
		map = ERR_PTR(r);
2405
		goto out;
2406
	}
2407

2408
	map = __bind(md, table, &limits);
L
Linus Torvalds 已提交
2409

2410
out:
2411
	mutex_unlock(&md->suspend_lock);
2412
	return map;
L
Linus Torvalds 已提交
2413 2414 2415 2416 2417 2418
}

/*
 * Functions to lock and unlock any filesystem running on the
 * device.
 */
2419
static int lock_fs(struct mapped_device *md)
L
Linus Torvalds 已提交
2420
{
2421
	int r;
L
Linus Torvalds 已提交
2422 2423

	WARN_ON(md->frozen_sb);
2424

2425
	md->frozen_sb = freeze_bdev(md->bdev);
2426
	if (IS_ERR(md->frozen_sb)) {
2427
		r = PTR_ERR(md->frozen_sb);
2428 2429
		md->frozen_sb = NULL;
		return r;
2430 2431
	}

2432 2433
	set_bit(DMF_FROZEN, &md->flags);

L
Linus Torvalds 已提交
2434 2435 2436
	return 0;
}

2437
static void unlock_fs(struct mapped_device *md)
L
Linus Torvalds 已提交
2438
{
2439 2440 2441
	if (!test_bit(DMF_FROZEN, &md->flags))
		return;

2442
	thaw_bdev(md->bdev, md->frozen_sb);
L
Linus Torvalds 已提交
2443
	md->frozen_sb = NULL;
2444
	clear_bit(DMF_FROZEN, &md->flags);
L
Linus Torvalds 已提交
2445 2446 2447 2448 2449 2450 2451 2452 2453
}

/*
 * 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.
 */
2454 2455 2456
/*
 * Suspend mechanism in request-based dm.
 *
K
Kiyoshi Ueda 已提交
2457 2458 2459
 * 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.
2460
 *
K
Kiyoshi Ueda 已提交
2461
 * To abort suspend, start the request_queue.
2462
 */
2463
int dm_suspend(struct mapped_device *md, unsigned suspend_flags)
L
Linus Torvalds 已提交
2464
{
2465
	struct dm_table *map = NULL;
2466
	int r = 0;
2467
	int do_lockfs = suspend_flags & DM_SUSPEND_LOCKFS_FLAG ? 1 : 0;
2468
	int noflush = suspend_flags & DM_SUSPEND_NOFLUSH_FLAG ? 1 : 0;
L
Linus Torvalds 已提交
2469

2470
	mutex_lock(&md->suspend_lock);
2471

2472
	if (dm_suspended_md(md)) {
M
Milan Broz 已提交
2473
		r = -EINVAL;
2474
		goto out_unlock;
M
Milan Broz 已提交
2475
	}
L
Linus Torvalds 已提交
2476

2477
	map = dm_get_live_table(md);
L
Linus Torvalds 已提交
2478

2479 2480 2481 2482 2483 2484 2485
	/*
	 * 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);

2486 2487 2488
	/* This does not get reverted if there's an error later. */
	dm_table_presuspend_targets(map);

M
Mikulas Patocka 已提交
2489
	/*
K
Kiyoshi Ueda 已提交
2490 2491 2492 2493
	 * 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
2494 2495 2496 2497
	 */
	if (!noflush && do_lockfs) {
		r = lock_fs(md);
		if (r)
K
Kiyoshi Ueda 已提交
2498
			goto out;
2499
	}
L
Linus Torvalds 已提交
2500 2501

	/*
2502 2503 2504 2505 2506 2507 2508
	 * 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
2509 2510 2511
	 * __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
2512
	 */
2513
	down_write(&md->io_lock);
2514
	set_bit(DMF_BLOCK_IO_FOR_SUSPEND, &md->flags);
2515
	up_write(&md->io_lock);
L
Linus Torvalds 已提交
2516

2517
	/*
2518 2519
	 * Stop md->queue before flushing md->wq in case request-based
	 * dm defers requests to md->wq from md->queue.
2520
	 */
2521
	if (dm_request_based(md))
K
Kiyoshi Ueda 已提交
2522
		stop_queue(md->queue);
2523

2524 2525
	flush_workqueue(md->wq);

L
Linus Torvalds 已提交
2526
	/*
2527 2528 2529
	 * 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
2530
	 */
2531
	r = dm_wait_for_completion(md, TASK_INTERRUPTIBLE);
L
Linus Torvalds 已提交
2532

2533
	down_write(&md->io_lock);
2534
	if (noflush)
2535
		clear_bit(DMF_NOFLUSH_SUSPENDING, &md->flags);
2536
	up_write(&md->io_lock);
2537

L
Linus Torvalds 已提交
2538
	/* were we interrupted ? */
2539
	if (r < 0) {
2540
		dm_queue_flush(md);
M
Milan Broz 已提交
2541

2542
		if (dm_request_based(md))
K
Kiyoshi Ueda 已提交
2543
			start_queue(md->queue);
2544

2545
		unlock_fs(md);
2546
		goto out; /* pushback list is already flushed, so skip flush */
2547
	}
L
Linus Torvalds 已提交
2548

2549 2550 2551 2552 2553 2554
	/*
	 * 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.
	 */

2555
	set_bit(DMF_SUSPENDED, &md->flags);
2556

2557 2558
	dm_table_postsuspend_targets(map);

2559 2560
out:
	dm_table_put(map);
2561 2562

out_unlock:
2563
	mutex_unlock(&md->suspend_lock);
2564
	return r;
L
Linus Torvalds 已提交
2565 2566 2567 2568
}

int dm_resume(struct mapped_device *md)
{
2569 2570
	int r = -EINVAL;
	struct dm_table *map = NULL;
L
Linus Torvalds 已提交
2571

2572
	mutex_lock(&md->suspend_lock);
2573
	if (!dm_suspended_md(md))
2574 2575
		goto out;

2576
	map = dm_get_live_table(md);
2577
	if (!map || !dm_table_get_size(map))
2578
		goto out;
L
Linus Torvalds 已提交
2579

2580 2581 2582
	r = dm_table_resume_targets(map);
	if (r)
		goto out;
2583

2584
	dm_queue_flush(md);
2585

2586 2587 2588 2589 2590 2591 2592 2593
	/*
	 * 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);

2594 2595 2596 2597
	unlock_fs(md);

	clear_bit(DMF_SUSPENDED, &md->flags);

2598 2599 2600
	r = 0;
out:
	dm_table_put(map);
2601
	mutex_unlock(&md->suspend_lock);
2602

2603
	return r;
L
Linus Torvalds 已提交
2604 2605 2606 2607 2608
}

/*-----------------------------------------------------------------
 * Event notification.
 *---------------------------------------------------------------*/
2609
int dm_kobject_uevent(struct mapped_device *md, enum kobject_action action,
M
Milan Broz 已提交
2610
		       unsigned cookie)
2611
{
M
Milan Broz 已提交
2612 2613 2614 2615
	char udev_cookie[DM_COOKIE_LENGTH];
	char *envp[] = { udev_cookie, NULL };

	if (!cookie)
2616
		return kobject_uevent(&disk_to_dev(md->disk)->kobj, action);
M
Milan Broz 已提交
2617 2618 2619
	else {
		snprintf(udev_cookie, DM_COOKIE_LENGTH, "%s=%u",
			 DM_COOKIE_ENV_VAR_NAME, cookie);
2620 2621
		return kobject_uevent_env(&disk_to_dev(md->disk)->kobj,
					  action, envp);
M
Milan Broz 已提交
2622
	}
2623 2624
}

M
Mike Anderson 已提交
2625 2626 2627 2628 2629
uint32_t dm_next_uevent_seq(struct mapped_device *md)
{
	return atomic_add_return(1, &md->uevent_seq);
}

L
Linus Torvalds 已提交
2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640
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
2641 2642 2643 2644 2645 2646 2647 2648 2649
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
2650 2651 2652 2653 2654 2655 2656 2657 2658
/*
 * 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
2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675
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;

2676
	if (test_bit(DMF_FREEING, &md->flags) ||
M
Mike Anderson 已提交
2677
	    dm_deleting_md(md))
2678 2679
		return NULL;

M
Milan Broz 已提交
2680 2681 2682 2683
	dm_get(md);
	return md;
}

2684
int dm_suspended_md(struct mapped_device *md)
L
Linus Torvalds 已提交
2685 2686 2687 2688
{
	return test_bit(DMF_SUSPENDED, &md->flags);
}

2689 2690
int dm_suspended(struct dm_target *ti)
{
2691
	return dm_suspended_md(dm_table_get_md(ti->table));
2692 2693 2694
}
EXPORT_SYMBOL_GPL(dm_suspended);

2695 2696
int dm_noflush_suspending(struct dm_target *ti)
{
2697
	return __noflush_suspending(dm_table_get_md(ti->table));
2698 2699 2700
}
EXPORT_SYMBOL_GPL(dm_noflush_suspending);

2701
struct dm_md_mempools *dm_alloc_md_mempools(unsigned type, unsigned integrity)
K
Kiyoshi Ueda 已提交
2702 2703
{
	struct dm_md_mempools *pools = kmalloc(sizeof(*pools), GFP_KERNEL);
2704
	unsigned int pool_size = (type == DM_TYPE_BIO_BASED) ? 16 : MIN_IOS;
K
Kiyoshi Ueda 已提交
2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720

	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;

2721
	pools->bs = bioset_create(pool_size, 0);
K
Kiyoshi Ueda 已提交
2722 2723 2724
	if (!pools->bs)
		goto free_tio_pool_and_out;

2725 2726 2727
	if (integrity && bioset_integrity_create(pools->bs, pool_size))
		goto free_bioset_and_out;

K
Kiyoshi Ueda 已提交
2728 2729
	return pools;

2730 2731 2732
free_bioset_and_out:
	bioset_free(pools->bs);

K
Kiyoshi Ueda 已提交
2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761
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);
}

2762
static const struct block_device_operations dm_blk_dops = {
L
Linus Torvalds 已提交
2763 2764
	.open = dm_blk_open,
	.release = dm_blk_close,
2765
	.ioctl = dm_blk_ioctl,
D
Darrick J. Wong 已提交
2766
	.getgeo = dm_blk_getgeo,
L
Linus Torvalds 已提交
2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782
	.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");