dm.c 27.9 KB
Newer Older
L
Linus Torvalds 已提交
1 2
/*
 * Copyright (C) 2001, 2002 Sistina Software (UK) Limited.
3
 * Copyright (C) 2004-2006 Red Hat, Inc. All rights reserved.
L
Linus Torvalds 已提交
4 5 6 7 8 9 10 11 12
 *
 * This file is released under the GPL.
 */

#include "dm.h"
#include "dm-bio-list.h"

#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 20
#include <linux/moduleparam.h>
#include <linux/blkpg.h>
#include <linux/bio.h>
#include <linux/buffer_head.h>
#include <linux/mempool.h>
#include <linux/slab.h>
#include <linux/idr.h>
D
Darrick J. Wong 已提交
21
#include <linux/hdreg.h>
22
#include <linux/blktrace_api.h>
23
#include <linux/smp_lock.h>
L
Linus Torvalds 已提交
24

25 26
#define DM_MSG_PREFIX "core"

L
Linus Torvalds 已提交
27 28 29 30 31
static const char *_name = DM_NAME;

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

32
static DEFINE_SPINLOCK(_minor_lock);
L
Linus Torvalds 已提交
33 34 35 36 37 38 39 40
/*
 * One of these is allocated per bio.
 */
struct dm_io {
	struct mapped_device *md;
	int error;
	struct bio *bio;
	atomic_t io_count;
41
	unsigned long start_time;
L
Linus Torvalds 已提交
42 43 44 45 46 47 48 49 50 51 52 53 54 55
};

/*
 * One of these is allocated per target within a bio.  Hopefully
 * this will be simplified out one day.
 */
struct target_io {
	struct dm_io *io;
	struct dm_target *ti;
	union map_info info;
};

union map_info *dm_get_mapinfo(struct bio *bio)
{
A
Alasdair G Kergon 已提交
56 57 58
	if (bio && bio->bi_private)
		return &((struct target_io *)bio->bi_private)->info;
	return NULL;
L
Linus Torvalds 已提交
59 60
}

61 62
#define MINOR_ALLOCED ((void *)-1)

L
Linus Torvalds 已提交
63 64 65 66 67
/*
 * Bits for the md->flags field.
 */
#define DMF_BLOCK_IO 0
#define DMF_SUSPENDED 1
68
#define DMF_FROZEN 2
J
Jeff Mahoney 已提交
69
#define DMF_FREEING 3
70
#define DMF_DELETING 4
L
Linus Torvalds 已提交
71 72

struct mapped_device {
73 74
	struct rw_semaphore io_lock;
	struct semaphore suspend_lock;
L
Linus Torvalds 已提交
75 76
	rwlock_t map_lock;
	atomic_t holders;
77
	atomic_t open_count;
L
Linus Torvalds 已提交
78 79 80 81 82

	unsigned long flags;

	request_queue_t *queue;
	struct gendisk *disk;
M
Mike Anderson 已提交
83
	char name[16];
L
Linus Torvalds 已提交
84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114

	void *interface_ptr;

	/*
	 * A list of ios that arrived while we were suspended.
	 */
	atomic_t pending;
	wait_queue_head_t wait;
 	struct bio_list deferred;

	/*
	 * The current mapping.
	 */
	struct dm_table *map;

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

	/*
	 * Event handling.
	 */
	atomic_t event_nr;
	wait_queue_head_t eventq;

	/*
	 * freeze/thaw support require holding onto a super block
	 */
	struct super_block *frozen_sb;
115
	struct block_device *suspended_bdev;
D
Darrick J. Wong 已提交
116 117 118

	/* forced geometry settings */
	struct hd_geometry geometry;
L
Linus Torvalds 已提交
119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170
};

#define MIN_IOS 256
static kmem_cache_t *_io_cache;
static kmem_cache_t *_tio_cache;

static struct bio_set *dm_set;

static int __init local_init(void)
{
	int r;

	dm_set = bioset_create(16, 16, 4);
	if (!dm_set)
		return -ENOMEM;

	/* allocate a slab for the dm_ios */
	_io_cache = kmem_cache_create("dm_io",
				      sizeof(struct dm_io), 0, 0, NULL, NULL);
	if (!_io_cache)
		return -ENOMEM;

	/* allocate a slab for the target ios */
	_tio_cache = kmem_cache_create("dm_tio", sizeof(struct target_io),
				       0, 0, NULL, NULL);
	if (!_tio_cache) {
		kmem_cache_destroy(_io_cache);
		return -ENOMEM;
	}

	_major = major;
	r = register_blkdev(_major, _name);
	if (r < 0) {
		kmem_cache_destroy(_tio_cache);
		kmem_cache_destroy(_io_cache);
		return r;
	}

	if (!_major)
		_major = r;

	return 0;
}

static void local_exit(void)
{
	kmem_cache_destroy(_tio_cache);
	kmem_cache_destroy(_io_cache);

	bioset_free(dm_set);

	if (unregister_blkdev(_major, _name) < 0)
171
		DMERR("unregister_blkdev failed");
L
Linus Torvalds 已提交
172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229

	_major = 0;

	DMINFO("cleaned up");
}

int (*_inits[])(void) __initdata = {
	local_init,
	dm_target_init,
	dm_linear_init,
	dm_stripe_init,
	dm_interface_init,
};

void (*_exits[])(void) = {
	local_exit,
	dm_target_exit,
	dm_linear_exit,
	dm_stripe_exit,
	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
 */
static int dm_blk_open(struct inode *inode, struct file *file)
{
	struct mapped_device *md;

J
Jeff Mahoney 已提交
230 231
	spin_lock(&_minor_lock);

L
Linus Torvalds 已提交
232
	md = inode->i_bdev->bd_disk->private_data;
J
Jeff Mahoney 已提交
233 234 235
	if (!md)
		goto out;

236 237
	if (test_bit(DMF_FREEING, &md->flags) ||
	    test_bit(DMF_DELETING, &md->flags)) {
J
Jeff Mahoney 已提交
238 239 240 241
		md = NULL;
		goto out;
	}

L
Linus Torvalds 已提交
242
	dm_get(md);
243
	atomic_inc(&md->open_count);
J
Jeff Mahoney 已提交
244 245 246 247 248

out:
	spin_unlock(&_minor_lock);

	return md ? 0 : -ENXIO;
L
Linus Torvalds 已提交
249 250 251 252 253 254 255
}

static int dm_blk_close(struct inode *inode, struct file *file)
{
	struct mapped_device *md;

	md = inode->i_bdev->bd_disk->private_data;
256
	atomic_dec(&md->open_count);
L
Linus Torvalds 已提交
257 258 259 260
	dm_put(md);
	return 0;
}

261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
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
285 286 287 288 289 290 291
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);
}

292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330
static int dm_blk_ioctl(struct inode *inode, struct file *file,
			unsigned int cmd, unsigned long arg)
{
	struct mapped_device *md;
	struct dm_table *map;
	struct dm_target *tgt;
	int r = -ENOTTY;

	/* We don't really need this lock, but we do need 'inode'. */
	unlock_kernel();

	md = inode->i_bdev->bd_disk->private_data;

	map = dm_get_table(md);

	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);

	if (dm_suspended(md)) {
		r = -EAGAIN;
		goto out;
	}

	if (tgt->type->ioctl)
		r = tgt->type->ioctl(tgt, inode, file, cmd, arg);

out:
	dm_table_put(map);

	lock_kernel();
	return r;
}

L
Linus Torvalds 已提交
331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350
static inline struct dm_io *alloc_io(struct mapped_device *md)
{
	return mempool_alloc(md->io_pool, GFP_NOIO);
}

static inline void free_io(struct mapped_device *md, struct dm_io *io)
{
	mempool_free(io, md->io_pool);
}

static inline struct target_io *alloc_tio(struct mapped_device *md)
{
	return mempool_alloc(md->tio_pool, GFP_NOIO);
}

static inline void free_tio(struct mapped_device *md, struct target_io *tio)
{
	mempool_free(tio, md->tio_pool);
}

351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380
static void start_io_acct(struct dm_io *io)
{
	struct mapped_device *md = io->md;

	io->start_time = jiffies;

	preempt_disable();
	disk_round_stats(dm_disk(md));
	preempt_enable();
	dm_disk(md)->in_flight = atomic_inc_return(&md->pending);
}

static int end_io_acct(struct dm_io *io)
{
	struct mapped_device *md = io->md;
	struct bio *bio = io->bio;
	unsigned long duration = jiffies - io->start_time;
	int pending;
	int rw = bio_data_dir(bio);

	preempt_disable();
	disk_round_stats(dm_disk(md));
	preempt_enable();
	dm_disk(md)->in_flight = pending = atomic_dec_return(&md->pending);

	disk_stat_add(dm_disk(md), ticks[rw], duration);

	return !pending;
}

L
Linus Torvalds 已提交
381 382 383 384 385
/*
 * Add the bio to the list of deferred io.
 */
static int queue_io(struct mapped_device *md, struct bio *bio)
{
386
	down_write(&md->io_lock);
L
Linus Torvalds 已提交
387 388

	if (!test_bit(DMF_BLOCK_IO, &md->flags)) {
389
		up_write(&md->io_lock);
L
Linus Torvalds 已提交
390 391 392 393 394
		return 1;
	}

	bio_list_add(&md->deferred, bio);

395
	up_write(&md->io_lock);
L
Linus Torvalds 已提交
396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416
	return 0;		/* deferred successfully */
}

/*
 * 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.
 */
struct dm_table *dm_get_table(struct mapped_device *md)
{
	struct dm_table *t;

	read_lock(&md->map_lock);
	t = md->map;
	if (t)
		dm_table_get(t);
	read_unlock(&md->map_lock);

	return t;
}

D
Darrick J. Wong 已提交
417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443
/*
 * 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
444 445 446 447 448 449 450 451 452 453 454 455 456
/*-----------------------------------------------------------------
 * 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.
 *---------------------------------------------------------------*/

/*
 * Decrements the number of outstanding ios that a bio has been
 * cloned into, completing the original io if necc.
 */
457
static void dec_pending(struct dm_io *io, int error)
L
Linus Torvalds 已提交
458 459 460 461 462
{
	if (error)
		io->error = error;

	if (atomic_dec_and_test(&io->io_count)) {
463
		if (end_io_acct(io))
L
Linus Torvalds 已提交
464 465 466
			/* nudge anyone waiting on suspend queue */
			wake_up(&io->md->wait);

467 468
		blk_add_trace_bio(io->md->queue, io->bio, BLK_TA_COMPLETE);

L
Linus Torvalds 已提交
469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526
		bio_endio(io->bio, io->bio->bi_size, io->error);
		free_io(io->md, io);
	}
}

static int clone_endio(struct bio *bio, unsigned int done, int error)
{
	int r = 0;
	struct target_io *tio = bio->bi_private;
	struct dm_io *io = tio->io;
	dm_endio_fn endio = tio->ti->type->end_io;

	if (bio->bi_size)
		return 1;

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

	if (endio) {
		r = endio(tio->ti, bio, error, &tio->info);
		if (r < 0)
			error = r;

		else if (r > 0)
			/* the target wants another shot at the io */
			return 1;
	}

	free_tio(io->md, tio);
	dec_pending(io, error);
	bio_put(bio);
	return r;
}

static sector_t max_io_len(struct mapped_device *md,
			   sector_t sector, struct dm_target *ti)
{
	sector_t offset = sector - ti->begin;
	sector_t len = ti->len - offset;

	/*
	 * Does the target need to split even further ?
	 */
	if (ti->split_io) {
		sector_t boundary;
		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,
		      struct target_io *tio)
{
	int r;
527
	sector_t sector;
L
Linus Torvalds 已提交
528 529 530 531 532 533 534 535 536 537 538 539 540 541 542

	/*
	 * Sanity checks.
	 */
	BUG_ON(!clone->bi_size);

	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);
543
	sector = clone->bi_sector;
L
Linus Torvalds 已提交
544
	r = ti->type->map(ti, clone, &tio->info);
545
	if (r > 0) {
L
Linus Torvalds 已提交
546
		/* the bio has been remapped so dispatch it */
547

A
Alasdair G Kergon 已提交
548 549
		blk_add_trace_remap(bdev_get_queue(clone->bi_bdev), clone,
				    tio->io->bio->bi_bdev->bd_dev, sector,
550 551
				    clone->bi_sector);

L
Linus Torvalds 已提交
552
		generic_make_request(clone);
553
	}
L
Linus Torvalds 已提交
554 555 556 557 558

	else if (r < 0) {
		/* error the io and bail out */
		struct dm_io *io = tio->io;
		free_tio(tio->io->md, tio);
559
		dec_pending(io, r);
L
Linus Torvalds 已提交
560 561 562 563 564 565 566 567 568 569 570 571 572 573
		bio_put(clone);
	}
}

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
574 575 576 577 578
static void dm_bio_destructor(struct bio *bio)
{
	bio_free(bio, dm_set);
}

L
Linus Torvalds 已提交
579 580 581 582 583 584 585 586 587 588 589
/*
 * Creates a little bio that is just does part of a bvec.
 */
static struct bio *split_bvec(struct bio *bio, sector_t sector,
			      unsigned short idx, unsigned int offset,
			      unsigned int len)
{
	struct bio *clone;
	struct bio_vec *bv = bio->bi_io_vec + idx;

	clone = bio_alloc_bioset(GFP_NOIO, 1, dm_set);
P
Peter Osterlund 已提交
590
	clone->bi_destructor = dm_bio_destructor;
L
Linus Torvalds 已提交
591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675
	*clone->bi_io_vec = *bv;

	clone->bi_sector = sector;
	clone->bi_bdev = bio->bi_bdev;
	clone->bi_rw = bio->bi_rw;
	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;

	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,
			     unsigned int len)
{
	struct bio *clone;

	clone = bio_clone(bio, GFP_NOIO);
	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);

	return clone;
}

static void __clone_and_map(struct clone_info *ci)
{
	struct bio *clone, *bio = ci->bio;
	struct dm_target *ti = dm_table_find_target(ci->map, ci->sector);
	sector_t len = 0, max = max_io_len(ci->md, ci->sector, ti);
	struct target_io *tio;

	/*
	 * Allocate a target io object.
	 */
	tio = alloc_tio(ci->md);
	tio->io = ci->io;
	tio->ti = ti;
	memset(&tio->info, 0, sizeof(tio->info));

	if (ci->sector_count <= max) {
		/*
		 * Optimise for the simple case where we can do all of
		 * the remaining io with a single clone.
		 */
		clone = clone_bio(bio, ci->sector, ci->idx,
				  bio->bi_vcnt - ci->idx, ci->sector_count);
		__map_bio(ti, clone, tio);
		ci->sector_count = 0;

	} 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;
		}

		clone = clone_bio(bio, ci->sector, ci->idx, i - ci->idx, len);
		__map_bio(ti, clone, tio);

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

	} else {
		/*
676
		 * Handle a bvec that must be split between two or more targets.
L
Linus Torvalds 已提交
677 678
		 */
		struct bio_vec *bv = bio->bi_io_vec + ci->idx;
679 680
		sector_t remaining = to_sector(bv->bv_len);
		unsigned int offset = 0;
L
Linus Torvalds 已提交
681

682 683 684 685
		do {
			if (offset) {
				ti = dm_table_find_target(ci->map, ci->sector);
				max = max_io_len(ci->md, ci->sector, ti);
L
Linus Torvalds 已提交
686

687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703
				tio = alloc_tio(ci->md);
				tio->io = ci->io;
				tio->ti = ti;
				memset(&tio->info, 0, sizeof(tio->info));
			}

			len = min(remaining, max);

			clone = split_bvec(bio, ci->sector, ci->idx,
					   bv->bv_offset + offset, len);

			__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
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

		ci->idx++;
	}
}

/*
 * Split the bio into several clones.
 */
static void __split_bio(struct mapped_device *md, struct bio *bio)
{
	struct clone_info ci;

	ci.map = dm_get_table(md);
	if (!ci.map) {
		bio_io_error(bio, bio->bi_size);
		return;
	}

	ci.md = md;
	ci.bio = bio;
	ci.io = alloc_io(md);
	ci.io->error = 0;
	atomic_set(&ci.io->io_count, 1);
	ci.io->bio = bio;
	ci.io->md = md;
	ci.sector = bio->bi_sector;
	ci.sector_count = bio_sectors(bio);
	ci.idx = bio->bi_idx;

733
	start_io_acct(ci.io);
L
Linus Torvalds 已提交
734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751
	while (ci.sector_count)
		__clone_and_map(&ci);

	/* drop the extra reference count */
	dec_pending(ci.io, 0);
	dm_table_put(ci.map);
}
/*-----------------------------------------------------------------
 * CRUD END
 *---------------------------------------------------------------*/

/*
 * The request function that just remaps the bio built up by
 * dm_merge_bvec.
 */
static int dm_request(request_queue_t *q, struct bio *bio)
{
	int r;
752
	int rw = bio_data_dir(bio);
L
Linus Torvalds 已提交
753 754
	struct mapped_device *md = q->queuedata;

755
	down_read(&md->io_lock);
L
Linus Torvalds 已提交
756

757 758 759
	disk_stat_inc(dm_disk(md), ios[rw]);
	disk_stat_add(dm_disk(md), sectors[rw], bio_sectors(bio));

L
Linus Torvalds 已提交
760 761 762 763 764
	/*
	 * If we're suspended we have to queue
	 * this io for later.
	 */
	while (test_bit(DMF_BLOCK_IO, &md->flags)) {
765
		up_read(&md->io_lock);
L
Linus Torvalds 已提交
766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783

		if (bio_rw(bio) == READA) {
			bio_io_error(bio, bio->bi_size);
			return 0;
		}

		r = queue_io(md, bio);
		if (r < 0) {
			bio_io_error(bio, bio->bi_size);
			return 0;

		} else if (r == 0)
			return 0;	/* deferred successfully */

		/*
		 * We're in a while loop, because someone could suspend
		 * before we get to the following read lock.
		 */
784
		down_read(&md->io_lock);
L
Linus Torvalds 已提交
785 786 787
	}

	__split_bio(md, bio);
788
	up_read(&md->io_lock);
L
Linus Torvalds 已提交
789 790 791 792 793 794 795 796 797 798 799
	return 0;
}

static int dm_flush_all(request_queue_t *q, struct gendisk *disk,
			sector_t *error_sector)
{
	struct mapped_device *md = q->queuedata;
	struct dm_table *map = dm_get_table(md);
	int ret = -ENXIO;

	if (map) {
800
		ret = dm_table_flush_all(map);
L
Linus Torvalds 已提交
801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837
		dm_table_put(map);
	}

	return ret;
}

static void dm_unplug_all(request_queue_t *q)
{
	struct mapped_device *md = q->queuedata;
	struct dm_table *map = dm_get_table(md);

	if (map) {
		dm_table_unplug_all(map);
		dm_table_put(map);
	}
}

static int dm_any_congested(void *congested_data, int bdi_bits)
{
	int r;
	struct mapped_device *md = (struct mapped_device *) congested_data;
	struct dm_table *map = dm_get_table(md);

	if (!map || test_bit(DMF_BLOCK_IO, &md->flags))
		r = bdi_bits;
	else
		r = dm_table_any_congested(map, bdi_bits);

	dm_table_put(map);
	return r;
}

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

838
static void free_minor(int minor)
L
Linus Torvalds 已提交
839
{
840
	spin_lock(&_minor_lock);
L
Linus Torvalds 已提交
841
	idr_remove(&_minor_idr, minor);
842
	spin_unlock(&_minor_lock);
L
Linus Torvalds 已提交
843 844 845 846 847
}

/*
 * See if the device with a specific minor # is free.
 */
848
static int specific_minor(struct mapped_device *md, int minor)
L
Linus Torvalds 已提交
849 850 851 852 853 854
{
	int r, m;

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

J
Jeff Mahoney 已提交
855 856 857 858
	r = idr_pre_get(&_minor_idr, GFP_KERNEL);
	if (!r)
		return -ENOMEM;

859
	spin_lock(&_minor_lock);
L
Linus Torvalds 已提交
860 861 862 863 864 865

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

866
	r = idr_get_new_above(&_minor_idr, MINOR_ALLOCED, minor, &m);
J
Jeff Mahoney 已提交
867
	if (r)
L
Linus Torvalds 已提交
868 869 870 871 872 873 874 875 876
		goto out;

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

out:
877
	spin_unlock(&_minor_lock);
L
Linus Torvalds 已提交
878 879 880
	return r;
}

881
static int next_free_minor(struct mapped_device *md, int *minor)
L
Linus Torvalds 已提交
882
{
883
	int r, m;
L
Linus Torvalds 已提交
884 885

	r = idr_pre_get(&_minor_idr, GFP_KERNEL);
J
Jeff Mahoney 已提交
886 887 888
	if (!r)
		return -ENOMEM;

889
	spin_lock(&_minor_lock);
L
Linus Torvalds 已提交
890

891
	r = idr_get_new(&_minor_idr, MINOR_ALLOCED, &m);
L
Linus Torvalds 已提交
892 893 894 895 896 897 898 899 900 901 902 903 904
	if (r) {
		goto out;
	}

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

	*minor = m;

out:
905
	spin_unlock(&_minor_lock);
L
Linus Torvalds 已提交
906 907 908 909 910 911 912 913
	return r;
}

static struct block_device_operations dm_blk_dops;

/*
 * Allocate and initialise a blank device with a given minor.
 */
914
static struct mapped_device *alloc_dev(int minor)
L
Linus Torvalds 已提交
915 916 917
{
	int r;
	struct mapped_device *md = kmalloc(sizeof(*md), GFP_KERNEL);
918
	void *old_md;
L
Linus Torvalds 已提交
919 920 921 922 923 924

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

925 926 927
	if (!try_module_get(THIS_MODULE))
		goto bad0;

L
Linus Torvalds 已提交
928
	/* get a minor number for the dev */
929 930 931 932
	if (minor == DM_ANY_MINOR)
		r = next_free_minor(md, &minor);
	else
		r = specific_minor(md, minor);
L
Linus Torvalds 已提交
933 934 935 936
	if (r < 0)
		goto bad1;

	memset(md, 0, sizeof(*md));
937 938
	init_rwsem(&md->io_lock);
	init_MUTEX(&md->suspend_lock);
L
Linus Torvalds 已提交
939 940
	rwlock_init(&md->map_lock);
	atomic_set(&md->holders, 1);
941
	atomic_set(&md->open_count, 0);
L
Linus Torvalds 已提交
942 943 944 945 946 947 948 949 950 951
	atomic_set(&md->event_nr, 0);

	md->queue = blk_alloc_queue(GFP_KERNEL);
	if (!md->queue)
		goto bad1;

	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);
952
	blk_queue_bounce_limit(md->queue, BLK_BOUNCE_ANY);
L
Linus Torvalds 已提交
953 954 955
	md->queue->unplug_fn = dm_unplug_all;
	md->queue->issue_flush_fn = dm_flush_all;

956
	md->io_pool = mempool_create_slab_pool(MIN_IOS, _io_cache);
L
Linus Torvalds 已提交
957 958 959
 	if (!md->io_pool)
 		goto bad2;

960
	md->tio_pool = mempool_create_slab_pool(MIN_IOS, _tio_cache);
L
Linus Torvalds 已提交
961 962 963 964 965 966 967
	if (!md->tio_pool)
		goto bad3;

	md->disk = alloc_disk(1);
	if (!md->disk)
		goto bad4;

968 969 970 971
	atomic_set(&md->pending, 0);
	init_waitqueue_head(&md->wait);
	init_waitqueue_head(&md->eventq);

L
Linus Torvalds 已提交
972 973 974 975 976 977 978
	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
979
	format_dev_t(md->name, MKDEV(_major, minor));
L
Linus Torvalds 已提交
980

981
	/* Populate the mapping, nobody knows we exist yet */
982
	spin_lock(&_minor_lock);
983
	old_md = idr_replace(&_minor_idr, md, minor);
984
	spin_unlock(&_minor_lock);
985 986 987

	BUG_ON(old_md != MINOR_ALLOCED);

L
Linus Torvalds 已提交
988 989 990 991 992 993 994
	return md;

 bad4:
	mempool_destroy(md->tio_pool);
 bad3:
	mempool_destroy(md->io_pool);
 bad2:
995
	blk_cleanup_queue(md->queue);
L
Linus Torvalds 已提交
996 997
	free_minor(minor);
 bad1:
998 999
	module_put(THIS_MODULE);
 bad0:
L
Linus Torvalds 已提交
1000 1001 1002 1003 1004 1005
	kfree(md);
	return NULL;
}

static void free_dev(struct mapped_device *md)
{
1006
	int minor = md->disk->first_minor;
1007

1008 1009 1010 1011
	if (md->suspended_bdev) {
		thaw_bdev(md->suspended_bdev, NULL);
		bdput(md->suspended_bdev);
	}
L
Linus Torvalds 已提交
1012 1013 1014
	mempool_destroy(md->tio_pool);
	mempool_destroy(md->io_pool);
	del_gendisk(md->disk);
1015
	free_minor(minor);
J
Jeff Mahoney 已提交
1016 1017 1018 1019 1020

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

L
Linus Torvalds 已提交
1021
	put_disk(md->disk);
1022
	blk_cleanup_queue(md->queue);
1023
	module_put(THIS_MODULE);
L
Linus Torvalds 已提交
1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037
	kfree(md);
}

/*
 * Bind a table to the device.
 */
static void event_callback(void *context)
{
	struct mapped_device *md = (struct mapped_device *) context;

	atomic_inc(&md->event_nr);
	wake_up(&md->eventq);
}

1038
static void __set_size(struct mapped_device *md, sector_t size)
L
Linus Torvalds 已提交
1039
{
1040
	set_capacity(md->disk, size);
L
Linus Torvalds 已提交
1041

1042
	mutex_lock(&md->suspended_bdev->bd_inode->i_mutex);
1043
	i_size_write(md->suspended_bdev->bd_inode, (loff_t)size << SECTOR_SHIFT);
1044
	mutex_unlock(&md->suspended_bdev->bd_inode->i_mutex);
L
Linus Torvalds 已提交
1045 1046 1047 1048 1049 1050 1051 1052
}

static int __bind(struct mapped_device *md, struct dm_table *t)
{
	request_queue_t *q = md->queue;
	sector_t size;

	size = dm_table_get_size(t);
D
Darrick J. Wong 已提交
1053 1054 1055 1056 1057 1058 1059

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

1060
	__set_size(md, size);
L
Linus Torvalds 已提交
1061 1062 1063
	if (size == 0)
		return 0;

1064 1065 1066
	dm_table_get(t);
	dm_table_event_callback(t, event_callback, md);

L
Linus Torvalds 已提交
1067 1068
	write_lock(&md->map_lock);
	md->map = t;
1069
	dm_table_set_restrictions(t, q);
L
Linus Torvalds 已提交
1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091
	write_unlock(&md->map_lock);

	return 0;
}

static void __unbind(struct mapped_device *md)
{
	struct dm_table *map = md->map;

	if (!map)
		return;

	dm_table_event_callback(map, NULL, NULL);
	write_lock(&md->map_lock);
	md->map = NULL;
	write_unlock(&md->map_lock);
	dm_table_put(map);
}

/*
 * Constructor for a new device.
 */
1092
int dm_create(int minor, struct mapped_device **result)
L
Linus Torvalds 已提交
1093 1094 1095
{
	struct mapped_device *md;

1096
	md = alloc_dev(minor);
L
Linus Torvalds 已提交
1097 1098 1099 1100 1101 1102 1103
	if (!md)
		return -ENXIO;

	*result = md;
	return 0;
}

1104
static struct mapped_device *dm_find_md(dev_t dev)
L
Linus Torvalds 已提交
1105 1106 1107 1108 1109 1110 1111
{
	struct mapped_device *md;
	unsigned minor = MINOR(dev);

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

1112
	spin_lock(&_minor_lock);
L
Linus Torvalds 已提交
1113 1114

	md = idr_find(&_minor_idr, minor);
J
Jeff Mahoney 已提交
1115 1116
	if (md && (md == MINOR_ALLOCED ||
		   (dm_disk(md)->first_minor != minor) ||
A
Alasdair G Kergon 已提交
1117
		   test_bit(DMF_FREEING, &md->flags))) {
1118
		md = NULL;
J
Jeff Mahoney 已提交
1119 1120
		goto out;
	}
L
Linus Torvalds 已提交
1121

J
Jeff Mahoney 已提交
1122
out:
1123
	spin_unlock(&_minor_lock);
L
Linus Torvalds 已提交
1124

1125 1126 1127
	return md;
}

1128 1129 1130 1131 1132 1133 1134 1135 1136 1137
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
1138
void *dm_get_mdptr(struct mapped_device *md)
1139
{
A
Alasdair G Kergon 已提交
1140
	return md->interface_ptr;
L
Linus Torvalds 已提交
1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152
}

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);
}

1153 1154 1155 1156 1157 1158
const char *dm_device_name(struct mapped_device *md)
{
	return md->name;
}
EXPORT_SYMBOL_GPL(dm_device_name);

L
Linus Torvalds 已提交
1159 1160
void dm_put(struct mapped_device *md)
{
M
Mike Anderson 已提交
1161
	struct dm_table *map;
L
Linus Torvalds 已提交
1162

J
Jeff Mahoney 已提交
1163 1164
	BUG_ON(test_bit(DMF_FREEING, &md->flags));

1165
	if (atomic_dec_and_lock(&md->holders, &_minor_lock)) {
M
Mike Anderson 已提交
1166
		map = dm_get_table(md);
1167
		idr_replace(&_minor_idr, MINOR_ALLOCED, dm_disk(md)->first_minor);
J
Jeff Mahoney 已提交
1168
		set_bit(DMF_FREEING, &md->flags);
1169
		spin_unlock(&_minor_lock);
1170
		if (!dm_suspended(md)) {
L
Linus Torvalds 已提交
1171 1172 1173 1174
			dm_table_presuspend_targets(map);
			dm_table_postsuspend_targets(map);
		}
		__unbind(md);
M
Mike Anderson 已提交
1175
		dm_table_put(map);
L
Linus Torvalds 已提交
1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199
		free_dev(md);
	}
}

/*
 * Process the deferred bios
 */
static void __flush_deferred_io(struct mapped_device *md, struct bio *c)
{
	struct bio *n;

	while (c) {
		n = c->bi_next;
		c->bi_next = NULL;
		__split_bio(md, c);
		c = n;
	}
}

/*
 * Swap in a new table (destroying old one).
 */
int dm_swap_table(struct mapped_device *md, struct dm_table *table)
{
1200
	int r = -EINVAL;
L
Linus Torvalds 已提交
1201

1202
	down(&md->suspend_lock);
L
Linus Torvalds 已提交
1203 1204

	/* device must be suspended */
1205
	if (!dm_suspended(md))
1206
		goto out;
L
Linus Torvalds 已提交
1207 1208 1209 1210

	__unbind(md);
	r = __bind(md, table);

1211
out:
1212
	up(&md->suspend_lock);
1213
	return r;
L
Linus Torvalds 已提交
1214 1215 1216 1217 1218 1219
}

/*
 * Functions to lock and unlock any filesystem running on the
 * device.
 */
1220
static int lock_fs(struct mapped_device *md)
L
Linus Torvalds 已提交
1221
{
1222
	int r;
L
Linus Torvalds 已提交
1223 1224

	WARN_ON(md->frozen_sb);
1225

1226
	md->frozen_sb = freeze_bdev(md->suspended_bdev);
1227
	if (IS_ERR(md->frozen_sb)) {
1228
		r = PTR_ERR(md->frozen_sb);
1229 1230
		md->frozen_sb = NULL;
		return r;
1231 1232
	}

1233 1234
	set_bit(DMF_FROZEN, &md->flags);

L
Linus Torvalds 已提交
1235
	/* don't bdput right now, we don't want the bdev
1236
	 * to go away while it is locked.
L
Linus Torvalds 已提交
1237 1238 1239 1240
	 */
	return 0;
}

1241
static void unlock_fs(struct mapped_device *md)
L
Linus Torvalds 已提交
1242
{
1243 1244 1245
	if (!test_bit(DMF_FROZEN, &md->flags))
		return;

1246
	thaw_bdev(md->suspended_bdev, md->frozen_sb);
L
Linus Torvalds 已提交
1247
	md->frozen_sb = NULL;
1248
	clear_bit(DMF_FROZEN, &md->flags);
L
Linus Torvalds 已提交
1249 1250 1251 1252 1253 1254 1255 1256 1257
}

/*
 * 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.
 */
1258
int dm_suspend(struct mapped_device *md, int do_lockfs)
L
Linus Torvalds 已提交
1259
{
1260
	struct dm_table *map = NULL;
L
Linus Torvalds 已提交
1261
	DECLARE_WAITQUEUE(wait, current);
J
Jun'ichi Nomura 已提交
1262
	struct bio *def;
1263
	int r = -EINVAL;
L
Linus Torvalds 已提交
1264

1265 1266 1267 1268
	down(&md->suspend_lock);

	if (dm_suspended(md))
		goto out;
L
Linus Torvalds 已提交
1269 1270 1271

	map = dm_get_table(md);

1272 1273 1274
	/* This does not get reverted if there's an error later. */
	dm_table_presuspend_targets(map);

1275 1276 1277 1278 1279 1280 1281
	md->suspended_bdev = bdget_disk(md->disk, 0);
	if (!md->suspended_bdev) {
		DMWARN("bdget failed in dm_suspend");
		r = -ENOMEM;
		goto out;
	}

1282
	/* Flush I/O to the device. */
1283 1284 1285 1286 1287
	if (do_lockfs) {
		r = lock_fs(md);
		if (r)
			goto out;
	}
L
Linus Torvalds 已提交
1288 1289

	/*
1290
	 * First we set the BLOCK_IO flag so no more ios will be mapped.
L
Linus Torvalds 已提交
1291
	 */
1292 1293
	down_write(&md->io_lock);
	set_bit(DMF_BLOCK_IO, &md->flags);
L
Linus Torvalds 已提交
1294 1295

	add_wait_queue(&md->wait, &wait);
1296
	up_write(&md->io_lock);
L
Linus Torvalds 已提交
1297 1298

	/* unplug */
1299
	if (map)
L
Linus Torvalds 已提交
1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315
		dm_table_unplug_all(map);

	/*
	 * Then we wait for the already mapped ios to
	 * complete.
	 */
	while (1) {
		set_current_state(TASK_INTERRUPTIBLE);

		if (!atomic_read(&md->pending) || signal_pending(current))
			break;

		io_schedule();
	}
	set_current_state(TASK_RUNNING);

1316
	down_write(&md->io_lock);
L
Linus Torvalds 已提交
1317 1318 1319
	remove_wait_queue(&md->wait, &wait);

	/* were we interrupted ? */
1320
	r = -EINTR;
1321
	if (atomic_read(&md->pending)) {
J
Jun'ichi Nomura 已提交
1322 1323 1324
		clear_bit(DMF_BLOCK_IO, &md->flags);
		def = bio_list_get(&md->deferred);
		__flush_deferred_io(md, def);
1325 1326 1327 1328 1329
		up_write(&md->io_lock);
		unlock_fs(md);
		goto out;
	}
	up_write(&md->io_lock);
L
Linus Torvalds 已提交
1330

1331
	dm_table_postsuspend_targets(map);
L
Linus Torvalds 已提交
1332

1333
	set_bit(DMF_SUSPENDED, &md->flags);
1334

1335
	r = 0;
1336

1337
out:
1338 1339 1340 1341 1342
	if (r && md->suspended_bdev) {
		bdput(md->suspended_bdev);
		md->suspended_bdev = NULL;
	}

1343 1344
	dm_table_put(map);
	up(&md->suspend_lock);
1345
	return r;
L
Linus Torvalds 已提交
1346 1347 1348 1349
}

int dm_resume(struct mapped_device *md)
{
1350
	int r = -EINVAL;
L
Linus Torvalds 已提交
1351
	struct bio *def;
1352
	struct dm_table *map = NULL;
L
Linus Torvalds 已提交
1353

1354 1355
	down(&md->suspend_lock);
	if (!dm_suspended(md))
1356 1357 1358
		goto out;

	map = dm_get_table(md);
1359
	if (!map || !dm_table_get_size(map))
1360
		goto out;
L
Linus Torvalds 已提交
1361 1362

	dm_table_resume_targets(map);
1363 1364

	down_write(&md->io_lock);
L
Linus Torvalds 已提交
1365 1366 1367 1368
	clear_bit(DMF_BLOCK_IO, &md->flags);

	def = bio_list_get(&md->deferred);
	__flush_deferred_io(md, def);
1369 1370 1371 1372
	up_write(&md->io_lock);

	unlock_fs(md);

1373 1374 1375
	bdput(md->suspended_bdev);
	md->suspended_bdev = NULL;

1376 1377
	clear_bit(DMF_SUSPENDED, &md->flags);

L
Linus Torvalds 已提交
1378 1379
	dm_table_unplug_all(map);

1380
	r = 0;
1381

1382 1383
out:
	dm_table_put(map);
1384 1385
	up(&md->suspend_lock);

1386
	return r;
L
Linus Torvalds 已提交
1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419
}

/*-----------------------------------------------------------------
 * Event notification.
 *---------------------------------------------------------------*/
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)));
}

/*
 * 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;
}

int dm_suspended(struct mapped_device *md)
{
	return test_bit(DMF_SUSPENDED, &md->flags);
}

static struct block_device_operations dm_blk_dops = {
	.open = dm_blk_open,
	.release = dm_blk_close,
1420
	.ioctl = dm_blk_ioctl,
D
Darrick J. Wong 已提交
1421
	.getgeo = dm_blk_getgeo,
L
Linus Torvalds 已提交
1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437
	.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");