dm.c 28.5 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1 2
/*
 * Copyright (C) 2001, 2002 Sistina Software (UK) Limited.
3
 * Copyright (C) 2004-2006 Red Hat, Inc. All rights reserved.
Linus Torvalds's avatar
Linus Torvalds committed
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>
13
#include <linux/mutex.h>
Linus Torvalds's avatar
Linus Torvalds committed
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>
Darrick J. Wong's avatar
Darrick J. Wong committed
21
#include <linux/hdreg.h>
22
#include <linux/blktrace_api.h>
23
#include <linux/smp_lock.h>
Linus Torvalds's avatar
Linus Torvalds committed
24

25 26
#define DM_MSG_PREFIX "core"

Linus Torvalds's avatar
Linus Torvalds committed
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);
Linus Torvalds's avatar
Linus Torvalds committed
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;
Linus Torvalds's avatar
Linus Torvalds committed
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)
{
Alasdair G Kergon's avatar
Alasdair G Kergon committed
56 57 58
	if (bio && bio->bi_private)
		return &((struct target_io *)bio->bi_private)->info;
	return NULL;
Linus Torvalds's avatar
Linus Torvalds committed
59 60
}

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

Linus Torvalds's avatar
Linus Torvalds committed
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
Jeff Mahoney's avatar
Jeff Mahoney committed
69
#define DMF_FREEING 3
70
#define DMF_DELETING 4
Linus Torvalds's avatar
Linus Torvalds committed
71 72

struct mapped_device {
73 74
	struct rw_semaphore io_lock;
	struct semaphore suspend_lock;
Linus Torvalds's avatar
Linus Torvalds committed
75 76
	rwlock_t map_lock;
	atomic_t holders;
77
	atomic_t open_count;
Linus Torvalds's avatar
Linus Torvalds committed
78 79 80 81 82

	unsigned long flags;

	request_queue_t *queue;
	struct gendisk *disk;
Mike Anderson's avatar
Mike Anderson committed
83
	char name[16];
Linus Torvalds's avatar
Linus Torvalds committed
84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104

	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;

Stefan Bader's avatar
Stefan Bader committed
105 106
	struct bio_set *bs;

Linus Torvalds's avatar
Linus Torvalds committed
107 108 109 110 111 112 113 114 115 116
	/*
	 * 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;
117
	struct block_device *suspended_bdev;
Darrick J. Wong's avatar
Darrick J. Wong committed
118 119 120

	/* forced geometry settings */
	struct hd_geometry geometry;
Linus Torvalds's avatar
Linus Torvalds committed
121 122 123
};

#define MIN_IOS 256
124 125
static struct kmem_cache *_io_cache;
static struct kmem_cache *_tio_cache;
Linus Torvalds's avatar
Linus Torvalds committed
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

static int __init local_init(void)
{
	int r;

	/* 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);

	if (unregister_blkdev(_major, _name) < 0)
165
		DMERR("unregister_blkdev failed");
Linus Torvalds's avatar
Linus Torvalds committed
166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223

	_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;

Jeff Mahoney's avatar
Jeff Mahoney committed
224 225
	spin_lock(&_minor_lock);

Linus Torvalds's avatar
Linus Torvalds committed
226
	md = inode->i_bdev->bd_disk->private_data;
Jeff Mahoney's avatar
Jeff Mahoney committed
227 228 229
	if (!md)
		goto out;

230 231
	if (test_bit(DMF_FREEING, &md->flags) ||
	    test_bit(DMF_DELETING, &md->flags)) {
Jeff Mahoney's avatar
Jeff Mahoney committed
232 233 234 235
		md = NULL;
		goto out;
	}

Linus Torvalds's avatar
Linus Torvalds committed
236
	dm_get(md);
237
	atomic_inc(&md->open_count);
Jeff Mahoney's avatar
Jeff Mahoney committed
238 239 240 241 242

out:
	spin_unlock(&_minor_lock);

	return md ? 0 : -ENXIO;
Linus Torvalds's avatar
Linus Torvalds committed
243 244 245 246 247 248 249
}

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

	md = inode->i_bdev->bd_disk->private_data;
250
	atomic_dec(&md->open_count);
Linus Torvalds's avatar
Linus Torvalds committed
251 252 253 254
	dm_put(md);
	return 0;
}

255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278
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;
}

Darrick J. Wong's avatar
Darrick J. Wong committed
279 280 281 282 283 284 285
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);
}

286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324
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;
}

Linus Torvalds's avatar
Linus Torvalds committed
325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344
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);
}

345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374
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;
}

Linus Torvalds's avatar
Linus Torvalds committed
375 376 377 378 379
/*
 * Add the bio to the list of deferred io.
 */
static int queue_io(struct mapped_device *md, struct bio *bio)
{
380
	down_write(&md->io_lock);
Linus Torvalds's avatar
Linus Torvalds committed
381 382

	if (!test_bit(DMF_BLOCK_IO, &md->flags)) {
383
		up_write(&md->io_lock);
Linus Torvalds's avatar
Linus Torvalds committed
384 385 386 387 388
		return 1;
	}

	bio_list_add(&md->deferred, bio);

389
	up_write(&md->io_lock);
Linus Torvalds's avatar
Linus Torvalds committed
390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410
	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;
}

Darrick J. Wong's avatar
Darrick J. Wong committed
411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437
/*
 * 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;
}

Linus Torvalds's avatar
Linus Torvalds committed
438 439 440 441 442 443 444 445 446 447 448 449 450
/*-----------------------------------------------------------------
 * 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.
 */
451
static void dec_pending(struct dm_io *io, int error)
Linus Torvalds's avatar
Linus Torvalds committed
452 453 454 455 456
{
	if (error)
		io->error = error;

	if (atomic_dec_and_test(&io->io_count)) {
457
		if (end_io_acct(io))
Linus Torvalds's avatar
Linus Torvalds committed
458 459 460
			/* nudge anyone waiting on suspend queue */
			wake_up(&io->md->wait);

461 462
		blk_add_trace_bio(io->md->queue, io->bio, BLK_TA_COMPLETE);

Linus Torvalds's avatar
Linus Torvalds committed
463 464 465 466 467 468 469 470 471
		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;
Stefan Bader's avatar
Stefan Bader committed
472
	struct mapped_device *md = tio->io->md;
Linus Torvalds's avatar
Linus Torvalds committed
473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490
	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;
	}

Stefan Bader's avatar
Stefan Bader committed
491 492 493 494 495 496 497
	dec_pending(tio->io, error);

	/*
	 * Store md for cleanup instead of tio which is about to get freed.
	 */
	bio->bi_private = md->bs;

Linus Torvalds's avatar
Linus Torvalds committed
498
	bio_put(bio);
Stefan Bader's avatar
Stefan Bader committed
499
	free_tio(md, tio);
Linus Torvalds's avatar
Linus Torvalds committed
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
	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;
Stefan Bader's avatar
Stefan Bader committed
528
	struct mapped_device *md;
Linus Torvalds's avatar
Linus Torvalds committed
529 530 531 532 533 534 535 536 537 538 539 540 541 542 543

	/*
	 * 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);
544
	sector = clone->bi_sector;
Linus Torvalds's avatar
Linus Torvalds committed
545
	r = ti->type->map(ti, clone, &tio->info);
546
	if (r > 0) {
Linus Torvalds's avatar
Linus Torvalds committed
547
		/* the bio has been remapped so dispatch it */
548

Alasdair G Kergon's avatar
Alasdair G Kergon committed
549 550
		blk_add_trace_remap(bdev_get_queue(clone->bi_bdev), clone,
				    tio->io->bio->bi_bdev->bd_dev, sector,
551 552
				    clone->bi_sector);

Linus Torvalds's avatar
Linus Torvalds committed
553
		generic_make_request(clone);
554
	}
Linus Torvalds's avatar
Linus Torvalds committed
555 556 557

	else if (r < 0) {
		/* error the io and bail out */
Stefan Bader's avatar
Stefan Bader committed
558 559 560 561 562 563
		md = tio->io->md;
		dec_pending(tio->io, r);
		/*
		 * Store bio_set for cleanup.
		 */
		clone->bi_private = md->bs;
Linus Torvalds's avatar
Linus Torvalds committed
564
		bio_put(clone);
Stefan Bader's avatar
Stefan Bader committed
565
		free_tio(md, tio);
Linus Torvalds's avatar
Linus Torvalds committed
566 567 568 569 570 571 572 573 574 575 576 577 578
	}
}

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;
};

Peter Osterlund's avatar
Peter Osterlund committed
579 580
static void dm_bio_destructor(struct bio *bio)
{
Stefan Bader's avatar
Stefan Bader committed
581 582 583
	struct bio_set *bs = bio->bi_private;

	bio_free(bio, bs);
Peter Osterlund's avatar
Peter Osterlund committed
584 585
}

Linus Torvalds's avatar
Linus Torvalds committed
586 587 588 589 590
/*
 * 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,
Stefan Bader's avatar
Stefan Bader committed
591
			      unsigned int len, struct bio_set *bs)
Linus Torvalds's avatar
Linus Torvalds committed
592 593 594 595
{
	struct bio *clone;
	struct bio_vec *bv = bio->bi_io_vec + idx;

Stefan Bader's avatar
Stefan Bader committed
596
	clone = bio_alloc_bioset(GFP_NOIO, 1, bs);
Peter Osterlund's avatar
Peter Osterlund committed
597
	clone->bi_destructor = dm_bio_destructor;
Linus Torvalds's avatar
Linus Torvalds committed
598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615
	*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,
Stefan Bader's avatar
Stefan Bader committed
616
			     unsigned int len, struct bio_set *bs)
Linus Torvalds's avatar
Linus Torvalds committed
617 618 619
{
	struct bio *clone;

Stefan Bader's avatar
Stefan Bader committed
620 621 622
	clone = bio_alloc_bioset(GFP_NOIO, bio->bi_max_vecs, bs);
	__bio_clone(clone, bio);
	clone->bi_destructor = dm_bio_destructor;
Linus Torvalds's avatar
Linus Torvalds committed
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
	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,
Stefan Bader's avatar
Stefan Bader committed
653 654
				  bio->bi_vcnt - ci->idx, ci->sector_count,
				  ci->md->bs);
Linus Torvalds's avatar
Linus Torvalds committed
655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676
		__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;
		}

Stefan Bader's avatar
Stefan Bader committed
677 678
		clone = clone_bio(bio, ci->sector, ci->idx, i - ci->idx, len,
				  ci->md->bs);
Linus Torvalds's avatar
Linus Torvalds committed
679 680 681 682 683 684 685 686
		__map_bio(ti, clone, tio);

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

	} else {
		/*
687
		 * Handle a bvec that must be split between two or more targets.
Linus Torvalds's avatar
Linus Torvalds committed
688 689
		 */
		struct bio_vec *bv = bio->bi_io_vec + ci->idx;
690 691
		sector_t remaining = to_sector(bv->bv_len);
		unsigned int offset = 0;
Linus Torvalds's avatar
Linus Torvalds committed
692

693 694 695 696
		do {
			if (offset) {
				ti = dm_table_find_target(ci->map, ci->sector);
				max = max_io_len(ci->md, ci->sector, ti);
Linus Torvalds's avatar
Linus Torvalds committed
697

698 699 700 701 702 703 704 705 706
				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,
Stefan Bader's avatar
Stefan Bader committed
707 708
					   bv->bv_offset + offset, len,
					   ci->md->bs);
709 710 711 712 713 714 715

			__map_bio(ti, clone, tio);

			ci->sector += len;
			ci->sector_count -= len;
			offset += to_bytes(len);
		} while (remaining -= len);
Linus Torvalds's avatar
Linus Torvalds committed
716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744

		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;

745
	start_io_acct(ci.io);
Linus Torvalds's avatar
Linus Torvalds committed
746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763
	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;
764
	int rw = bio_data_dir(bio);
Linus Torvalds's avatar
Linus Torvalds committed
765 766
	struct mapped_device *md = q->queuedata;

767
	down_read(&md->io_lock);
Linus Torvalds's avatar
Linus Torvalds committed
768

769 770 771
	disk_stat_inc(dm_disk(md), ios[rw]);
	disk_stat_add(dm_disk(md), sectors[rw], bio_sectors(bio));

Linus Torvalds's avatar
Linus Torvalds committed
772 773 774 775 776
	/*
	 * If we're suspended we have to queue
	 * this io for later.
	 */
	while (test_bit(DMF_BLOCK_IO, &md->flags)) {
777
		up_read(&md->io_lock);
Linus Torvalds's avatar
Linus Torvalds committed
778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795

		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.
		 */
796
		down_read(&md->io_lock);
Linus Torvalds's avatar
Linus Torvalds committed
797 798 799
	}

	__split_bio(md, bio);
800
	up_read(&md->io_lock);
Linus Torvalds's avatar
Linus Torvalds committed
801 802 803 804 805 806 807 808 809 810 811
	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) {
812
		ret = dm_table_flush_all(map);
Linus Torvalds's avatar
Linus Torvalds committed
813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849
		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);

850
static void free_minor(int minor)
Linus Torvalds's avatar
Linus Torvalds committed
851
{
852
	spin_lock(&_minor_lock);
Linus Torvalds's avatar
Linus Torvalds committed
853
	idr_remove(&_minor_idr, minor);
854
	spin_unlock(&_minor_lock);
Linus Torvalds's avatar
Linus Torvalds committed
855 856 857 858 859
}

/*
 * See if the device with a specific minor # is free.
 */
860
static int specific_minor(struct mapped_device *md, int minor)
Linus Torvalds's avatar
Linus Torvalds committed
861 862 863 864 865 866
{
	int r, m;

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

Jeff Mahoney's avatar
Jeff Mahoney committed
867 868 869 870
	r = idr_pre_get(&_minor_idr, GFP_KERNEL);
	if (!r)
		return -ENOMEM;

871
	spin_lock(&_minor_lock);
Linus Torvalds's avatar
Linus Torvalds committed
872 873 874 875 876 877

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

878
	r = idr_get_new_above(&_minor_idr, MINOR_ALLOCED, minor, &m);
Jeff Mahoney's avatar
Jeff Mahoney committed
879
	if (r)
Linus Torvalds's avatar
Linus Torvalds committed
880 881 882 883 884 885 886 887 888
		goto out;

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

out:
889
	spin_unlock(&_minor_lock);
Linus Torvalds's avatar
Linus Torvalds committed
890 891 892
	return r;
}

893
static int next_free_minor(struct mapped_device *md, int *minor)
Linus Torvalds's avatar
Linus Torvalds committed
894
{
895
	int r, m;
Linus Torvalds's avatar
Linus Torvalds committed
896 897

	r = idr_pre_get(&_minor_idr, GFP_KERNEL);
Jeff Mahoney's avatar
Jeff Mahoney committed
898 899 900
	if (!r)
		return -ENOMEM;

901
	spin_lock(&_minor_lock);
Linus Torvalds's avatar
Linus Torvalds committed
902

903
	r = idr_get_new(&_minor_idr, MINOR_ALLOCED, &m);
Linus Torvalds's avatar
Linus Torvalds committed
904 905 906 907 908 909 910 911 912 913 914 915 916
	if (r) {
		goto out;
	}

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

	*minor = m;

out:
917
	spin_unlock(&_minor_lock);
Linus Torvalds's avatar
Linus Torvalds committed
918 919 920 921 922 923 924 925
	return r;
}

static struct block_device_operations dm_blk_dops;

/*
 * Allocate and initialise a blank device with a given minor.
 */
926
static struct mapped_device *alloc_dev(int minor)
Linus Torvalds's avatar
Linus Torvalds committed
927 928 929
{
	int r;
	struct mapped_device *md = kmalloc(sizeof(*md), GFP_KERNEL);
930
	void *old_md;
Linus Torvalds's avatar
Linus Torvalds committed
931 932 933 934 935 936

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

937 938 939
	if (!try_module_get(THIS_MODULE))
		goto bad0;

Linus Torvalds's avatar
Linus Torvalds committed
940
	/* get a minor number for the dev */
941 942 943 944
	if (minor == DM_ANY_MINOR)
		r = next_free_minor(md, &minor);
	else
		r = specific_minor(md, minor);
Linus Torvalds's avatar
Linus Torvalds committed
945 946 947 948
	if (r < 0)
		goto bad1;

	memset(md, 0, sizeof(*md));
949 950
	init_rwsem(&md->io_lock);
	init_MUTEX(&md->suspend_lock);
Linus Torvalds's avatar
Linus Torvalds committed
951 952
	rwlock_init(&md->map_lock);
	atomic_set(&md->holders, 1);
953
	atomic_set(&md->open_count, 0);
Linus Torvalds's avatar
Linus Torvalds committed
954 955 956 957
	atomic_set(&md->event_nr, 0);

	md->queue = blk_alloc_queue(GFP_KERNEL);
	if (!md->queue)
958
		goto bad1_free_minor;
Linus Torvalds's avatar
Linus Torvalds committed
959 960 961 962 963

	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);
964
	blk_queue_bounce_limit(md->queue, BLK_BOUNCE_ANY);
Linus Torvalds's avatar
Linus Torvalds committed
965 966 967
	md->queue->unplug_fn = dm_unplug_all;
	md->queue->issue_flush_fn = dm_flush_all;

968
	md->io_pool = mempool_create_slab_pool(MIN_IOS, _io_cache);
Linus Torvalds's avatar
Linus Torvalds committed
969 970 971
 	if (!md->io_pool)
 		goto bad2;

972
	md->tio_pool = mempool_create_slab_pool(MIN_IOS, _tio_cache);
Linus Torvalds's avatar
Linus Torvalds committed
973 974 975
	if (!md->tio_pool)
		goto bad3;

Stefan Bader's avatar
Stefan Bader committed
976 977 978 979
	md->bs = bioset_create(16, 16, 4);
	if (!md->bs)
		goto bad_no_bioset;

Linus Torvalds's avatar
Linus Torvalds committed
980 981 982 983
	md->disk = alloc_disk(1);
	if (!md->disk)
		goto bad4;

984 985 986 987
	atomic_set(&md->pending, 0);
	init_waitqueue_head(&md->wait);
	init_waitqueue_head(&md->eventq);

Linus Torvalds's avatar
Linus Torvalds committed
988 989 990 991 992 993 994
	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);
Mike Anderson's avatar
Mike Anderson committed
995
	format_dev_t(md->name, MKDEV(_major, minor));
Linus Torvalds's avatar
Linus Torvalds committed
996

997
	/* Populate the mapping, nobody knows we exist yet */
998
	spin_lock(&_minor_lock);
999
	old_md = idr_replace(&_minor_idr, md, minor);
1000
	spin_unlock(&_minor_lock);
1001 1002 1003

	BUG_ON(old_md != MINOR_ALLOCED);

Linus Torvalds's avatar
Linus Torvalds committed
1004 1005 1006
	return md;

 bad4:
Stefan Bader's avatar
Stefan Bader committed
1007 1008
	bioset_free(md->bs);
 bad_no_bioset:
Linus Torvalds's avatar
Linus Torvalds committed
1009 1010 1011 1012
	mempool_destroy(md->tio_pool);
 bad3:
	mempool_destroy(md->io_pool);
 bad2:
1013
	blk_cleanup_queue(md->queue);
1014
 bad1_free_minor:
Linus Torvalds's avatar
Linus Torvalds committed
1015 1016
	free_minor(minor);
 bad1:
1017 1018
	module_put(THIS_MODULE);
 bad0:
Linus Torvalds's avatar
Linus Torvalds committed
1019 1020 1021 1022 1023 1024
	kfree(md);
	return NULL;
}

static void free_dev(struct mapped_device *md)
{
1025
	int minor = md->disk->first_minor;
1026

1027 1028 1029 1030
	if (md->suspended_bdev) {
		thaw_bdev(md->suspended_bdev, NULL);
		bdput(md->suspended_bdev);
	}
Linus Torvalds's avatar
Linus Torvalds committed
1031 1032
	mempool_destroy(md->tio_pool);
	mempool_destroy(md->io_pool);
Stefan Bader's avatar
Stefan Bader committed
1033
	bioset_free(md->bs);
Linus Torvalds's avatar
Linus Torvalds committed
1034
	del_gendisk(md->disk);
1035
	free_minor(minor);
Jeff Mahoney's avatar
Jeff Mahoney committed
1036 1037 1038 1039 1040

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

Linus Torvalds's avatar
Linus Torvalds committed
1041
	put_disk(md->disk);
1042
	blk_cleanup_queue(md->queue);
1043
	module_put(THIS_MODULE);
Linus Torvalds's avatar
Linus Torvalds committed
1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057
	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);
}

1058
static void __set_size(struct mapped_device *md, sector_t size)
Linus Torvalds's avatar
Linus Torvalds committed
1059
{
1060
	set_capacity(md->disk, size);
Linus Torvalds's avatar
Linus Torvalds committed
1061

1062
	mutex_lock(&md->suspended_bdev->bd_inode->i_mutex);
1063
	i_size_write(md->suspended_bdev->bd_inode, (loff_t)size << SECTOR_SHIFT);
1064
	mutex_unlock(&md->suspended_bdev->bd_inode->i_mutex);
Linus Torvalds's avatar
Linus Torvalds committed
1065 1066 1067 1068 1069 1070 1071 1072
}

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);
Darrick J. Wong's avatar
Darrick J. Wong committed
1073 1074 1075 1076 1077 1078 1079

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

1080
	__set_size(md, size);
Linus Torvalds's avatar
Linus Torvalds committed
1081 1082 1083
	if (size == 0)
		return 0;

1084 1085 1086
	dm_table_get(t);
	dm_table_event_callback(t, event_callback, md);

Linus Torvalds's avatar
Linus Torvalds committed
1087 1088
	write_lock(&md->map_lock);
	md->map = t;
1089
	dm_table_set_restrictions(t, q);
Linus Torvalds's avatar
Linus Torvalds committed
1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111
	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.
 */
1112
int dm_create(int minor, struct mapped_device **result)
Linus Torvalds's avatar
Linus Torvalds committed
1113 1114 1115
{
	struct mapped_device *md;

1116
	md = alloc_dev(minor);
Linus Torvalds's avatar
Linus Torvalds committed
1117 1118 1119 1120 1121 1122 1123
	if (!md)
		return -ENXIO;

	*result = md;
	return 0;
}

1124
static struct mapped_device *dm_find_md(dev_t dev)
Linus Torvalds's avatar
Linus Torvalds committed
1125 1126 1127 1128 1129 1130 1131
{
	struct mapped_device *md;
	unsigned minor = MINOR(dev);

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

1132
	spin_lock(&_minor_lock);
Linus Torvalds's avatar
Linus Torvalds committed
1133 1134

	md = idr_find(&_minor_idr, minor);
Jeff Mahoney's avatar
Jeff Mahoney committed
1135 1136
	if (md && (md == MINOR_ALLOCED ||
		   (dm_disk(md)->first_minor != minor) ||
Alasdair G Kergon's avatar
Alasdair G Kergon committed
1137
		   test_bit(DMF_FREEING, &md->flags))) {
1138
		md = NULL;
Jeff Mahoney's avatar
Jeff Mahoney committed
1139 1140
		goto out;
	}
Linus Torvalds's avatar
Linus Torvalds committed
1141

Jeff Mahoney's avatar
Jeff Mahoney committed
1142
out:
1143
	spin_unlock(&_minor_lock);
Linus Torvalds's avatar
Linus Torvalds committed
1144

1145 1146 1147
	return md;
}

1148 1149 1150 1151 1152 1153 1154 1155 1156 1157
struct mapped_device *dm_get_md(dev_t dev)
{
	struct mapped_device *md = dm_find_md(dev);

	if (md)
		dm_get(md);

	return md;
}

Alasdair G Kergon's avatar
Alasdair G Kergon committed
1158
void *dm_get_mdptr(struct mapped_device *md)
1159
{
Alasdair G Kergon's avatar
Alasdair G Kergon committed
1160
	return md->interface_ptr;
Linus Torvalds's avatar
Linus Torvalds committed
1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172
}

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);
}

1173 1174 1175 1176 1177 1178
const char *dm_device_name(struct mapped_device *md)
{
	return md->name;
}
EXPORT_SYMBOL_GPL(dm_device_name);

Linus Torvalds's avatar
Linus Torvalds committed
1179 1180
void dm_put(struct mapped_device *md)
{
Mike Anderson's avatar
Mike Anderson committed
1181
	struct dm_table *map;
Linus Torvalds's avatar
Linus Torvalds committed
1182

Jeff Mahoney's avatar
Jeff Mahoney committed
1183 1184
	BUG_ON(test_bit(DMF_FREEING, &md->flags));

1185
	if (atomic_dec_and_lock(&md->holders, &_minor_lock)) {
Mike Anderson's avatar
Mike Anderson committed
1186
		map = dm_get_table(md);
1187
		idr_replace(&_minor_idr, MINOR_ALLOCED, dm_disk(md)->first_minor);
Jeff Mahoney's avatar
Jeff Mahoney committed
1188
		set_bit(DMF_FREEING, &md->flags);
1189
		spin_unlock(&_minor_lock);
1190
		if (!dm_suspended(md)) {
Linus Torvalds's avatar
Linus Torvalds committed
1191 1192 1193 1194
			dm_table_presuspend_targets(map);
			dm_table_postsuspend_targets(map);
		}
		__unbind(md);
Mike Anderson's avatar
Mike Anderson committed
1195
		dm_table_put(map);
Linus Torvalds's avatar
Linus Torvalds committed
1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219
		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)
{
1220
	int r = -EINVAL;
Linus Torvalds's avatar
Linus Torvalds committed
1221

1222
	down(&md->suspend_lock);
Linus Torvalds's avatar
Linus Torvalds committed
1223 1224

	/* device must be suspended */
1225
	if (!dm_suspended(md))
1226
		goto out;
Linus Torvalds's avatar
Linus Torvalds committed
1227 1228 1229 1230

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

1231
out:
1232
	up(&md->suspend_lock);
1233
	return r;
Linus Torvalds's avatar
Linus Torvalds committed
1234 1235 1236 1237 1238 1239
}

/*
 * Functions to lock and unlock any filesystem running on the
 * device.
 */
1240
static int lock_fs(struct mapped_device *md)
Linus Torvalds's avatar
Linus Torvalds committed
1241
{
1242
	int r;
Linus Torvalds's avatar
Linus Torvalds committed
1243 1244

	WARN_ON(md->frozen_sb);
1245

1246
	md->frozen_sb = freeze_bdev(md->suspended_bdev);
1247
	if (IS_ERR(md->frozen_sb)) {
1248
		r = PTR_ERR(md->frozen_sb);
1249 1250
		md->frozen_sb = NULL;
		return r;
1251 1252
	}

1253 1254
	set_bit(DMF_FROZEN, &md->flags);

Linus Torvalds's avatar
Linus Torvalds committed
1255
	/* don't bdput right now, we don't want the bdev
1256
	 * to go away while it is locked.
Linus Torvalds's avatar
Linus Torvalds committed
1257 1258 1259 1260
	 */
	return 0;
}

1261
static void unlock_fs(struct mapped_device *md)
Linus Torvalds's avatar
Linus Torvalds committed
1262
{
1263 1264 1265
	if (!test_bit(DMF_FROZEN, &md->flags))
		return;

1266
	thaw_bdev(md->suspended_bdev, md->frozen_sb);
Linus Torvalds's avatar
Linus Torvalds committed
1267
	md->frozen_sb = NULL;
1268
	clear_bit(DMF_FROZEN, &md->flags);
Linus Torvalds's avatar
Linus Torvalds committed
1269 1270 1271 1272 1273 1274 1275 1276 1277
}

/*
 * 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.
 */
1278
int dm_suspend(struct mapped_device *md, int do_lockfs)
Linus Torvalds's avatar
Linus Torvalds committed
1279
{
1280
	struct dm_table *map = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
1281
	DECLARE_WAITQUEUE(wait, current);
Jun'ichi Nomura's avatar
Jun'ichi Nomura committed
1282
	struct bio *def;
1283
	int r = -EINVAL;
Linus Torvalds's avatar
Linus Torvalds committed
1284

1285 1286 1287
	down(&md->suspend_lock);

	if (dm_suspended(md))
1288
		goto out_unlock;
Linus Torvalds's avatar
Linus Torvalds committed
1289 1290 1291

	map = dm_get_table(md);

1292 1293 1294
	/* This does not get reverted if there's an error later. */
	dm_table_presuspend_targets(map);

1295 1296 1297 1298 1299 1300 1301
	md->suspended_bdev = bdget_disk(md->disk, 0);
	if (!md->suspended_bdev) {
		DMWARN("bdget failed in dm_suspend");
		r = -ENOMEM;
		goto out;
	}

1302
	/* Flush I/O to the device. */
1303 1304 1305 1306 1307
	if (do_lockfs) {
		r = lock_fs(md);
		if (r)
			goto out;
	}
Linus Torvalds's avatar
Linus Torvalds committed
1308 1309

	/*