dm-table.c 41.9 KB
Newer Older
L
Linus Torvalds 已提交
1 2
/*
 * Copyright (C) 2001 Sistina Software (UK) Limited.
3
 * Copyright (C) 2004-2008 Red Hat, Inc. All rights reserved.
L
Linus Torvalds 已提交
4 5 6 7
 *
 * This file is released under the GPL.
 */

8
#include "dm-core.h"
L
Linus Torvalds 已提交
9 10 11 12 13 14

#include <linux/module.h>
#include <linux/vmalloc.h>
#include <linux/blkdev.h>
#include <linux/namei.h>
#include <linux/ctype.h>
15
#include <linux/string.h>
L
Linus Torvalds 已提交
16 17
#include <linux/slab.h>
#include <linux/interrupt.h>
A
Arjan van de Ven 已提交
18
#include <linux/mutex.h>
19
#include <linux/delay.h>
A
Arun Sharma 已提交
20
#include <linux/atomic.h>
21
#include <linux/blk-mq.h>
22
#include <linux/mount.h>
L
Linus Torvalds 已提交
23

24 25
#define DM_MSG_PREFIX "table"

L
Linus Torvalds 已提交
26 27 28 29 30 31
#define MAX_DEPTH 16
#define NODE_SIZE L1_CACHE_BYTES
#define KEYS_PER_NODE (NODE_SIZE / sizeof(sector_t))
#define CHILDREN_PER_NODE (KEYS_PER_NODE + 1)

struct dm_table {
M
Mike Anderson 已提交
32
	struct mapped_device *md;
33
	enum dm_queue_mode type;
L
Linus Torvalds 已提交
34 35 36 37 38 39 40 41 42 43 44

	/* btree table */
	unsigned int depth;
	unsigned int counts[MAX_DEPTH];	/* in nodes */
	sector_t *index[MAX_DEPTH];

	unsigned int num_targets;
	unsigned int num_allocated;
	sector_t *highs;
	struct dm_target *targets;

45
	struct target_type *immutable_target_type;
46 47 48 49

	bool integrity_supported:1;
	bool singleton:1;
	bool all_blk_mq:1;
50
	unsigned integrity_added:1;
M
Mike Snitzer 已提交
51

L
Linus Torvalds 已提交
52 53 54 55 56
	/*
	 * Indicates the rw permissions for the new logical
	 * device.  This should be a combination of FMODE_READ
	 * and FMODE_WRITE.
	 */
57
	fmode_t mode;
L
Linus Torvalds 已提交
58 59 60 61 62 63 64

	/* a list of devices used by this table */
	struct list_head devices;

	/* events get handed up using this callback */
	void (*event_fn)(void *);
	void *event_context;
K
Kiyoshi Ueda 已提交
65 66

	struct dm_md_mempools *mempools;
67 68

	struct list_head target_callbacks;
L
Linus Torvalds 已提交
69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148
};

/*
 * Similar to ceiling(log_size(n))
 */
static unsigned int int_log(unsigned int n, unsigned int base)
{
	int result = 0;

	while (n > 1) {
		n = dm_div_up(n, base);
		result++;
	}

	return result;
}

/*
 * Calculate the index of the child node of the n'th node k'th key.
 */
static inline unsigned int get_child(unsigned int n, unsigned int k)
{
	return (n * CHILDREN_PER_NODE) + k;
}

/*
 * Return the n'th node of level l from table t.
 */
static inline sector_t *get_node(struct dm_table *t,
				 unsigned int l, unsigned int n)
{
	return t->index[l] + (n * KEYS_PER_NODE);
}

/*
 * Return the highest key that you could lookup from the n'th
 * node on level l of the btree.
 */
static sector_t high(struct dm_table *t, unsigned int l, unsigned int n)
{
	for (; l < t->depth - 1; l++)
		n = get_child(n, CHILDREN_PER_NODE - 1);

	if (n >= t->counts[l])
		return (sector_t) - 1;

	return get_node(t, l, n)[KEYS_PER_NODE - 1];
}

/*
 * Fills in a level of the btree based on the highs of the level
 * below it.
 */
static int setup_btree_index(unsigned int l, struct dm_table *t)
{
	unsigned int n, k;
	sector_t *node;

	for (n = 0U; n < t->counts[l]; n++) {
		node = get_node(t, l, n);

		for (k = 0U; k < KEYS_PER_NODE; k++)
			node[k] = high(t, l + 1, get_child(n, k));
	}

	return 0;
}

void *dm_vcalloc(unsigned long nmemb, unsigned long elem_size)
{
	unsigned long size;
	void *addr;

	/*
	 * Check that we're not going to overflow.
	 */
	if (nmemb > (ULONG_MAX / elem_size))
		return NULL;

	size = nmemb * elem_size;
J
Joe Perches 已提交
149
	addr = vzalloc(size);
L
Linus Torvalds 已提交
150 151 152

	return addr;
}
153
EXPORT_SYMBOL(dm_vcalloc);
L
Linus Torvalds 已提交
154 155 156 157 158 159 160 161 162 163 164 165

/*
 * highs, and targets are managed as dynamic arrays during a
 * table load.
 */
static int alloc_targets(struct dm_table *t, unsigned int num)
{
	sector_t *n_highs;
	struct dm_target *n_targets;

	/*
	 * Allocate both the target array and offset array at once.
166 167
	 * Append an empty entry to catch sectors beyond the end of
	 * the device.
L
Linus Torvalds 已提交
168
	 */
169
	n_highs = (sector_t *) dm_vcalloc(num + 1, sizeof(struct dm_target) +
L
Linus Torvalds 已提交
170 171 172 173 174 175
					  sizeof(sector_t));
	if (!n_highs)
		return -ENOMEM;

	n_targets = (struct dm_target *) (n_highs + num);

176
	memset(n_highs, -1, sizeof(*n_highs) * num);
L
Linus Torvalds 已提交
177 178 179 180 181 182 183 184 185
	vfree(t->highs);

	t->num_allocated = num;
	t->highs = n_highs;
	t->targets = n_targets;

	return 0;
}

186
int dm_table_create(struct dm_table **result, fmode_t mode,
M
Mike Anderson 已提交
187
		    unsigned num_targets, struct mapped_device *md)
L
Linus Torvalds 已提交
188
{
D
Dmitry Monakhov 已提交
189
	struct dm_table *t = kzalloc(sizeof(*t), GFP_KERNEL);
L
Linus Torvalds 已提交
190 191 192 193 194

	if (!t)
		return -ENOMEM;

	INIT_LIST_HEAD(&t->devices);
195
	INIT_LIST_HEAD(&t->target_callbacks);
L
Linus Torvalds 已提交
196 197 198 199 200 201

	if (!num_targets)
		num_targets = KEYS_PER_NODE;

	num_targets = dm_round_up(num_targets, KEYS_PER_NODE);

202 203 204 205 206
	if (!num_targets) {
		kfree(t);
		return -ENOMEM;
	}

L
Linus Torvalds 已提交
207 208 209 210 211
	if (alloc_targets(t, num_targets)) {
		kfree(t);
		return -ENOMEM;
	}

212
	t->type = DM_TYPE_NONE;
L
Linus Torvalds 已提交
213
	t->mode = mode;
M
Mike Anderson 已提交
214
	t->md = md;
L
Linus Torvalds 已提交
215 216 217 218
	*result = t;
	return 0;
}

219
static void free_devices(struct list_head *devices, struct mapped_device *md)
L
Linus Torvalds 已提交
220 221 222
{
	struct list_head *tmp, *next;

P
Paul Jimenez 已提交
223
	list_for_each_safe(tmp, next, devices) {
M
Mikulas Patocka 已提交
224 225
		struct dm_dev_internal *dd =
		    list_entry(tmp, struct dm_dev_internal, list);
226 227 228
		DMWARN("%s: dm_table_destroy: dm_put_device call missing for %s",
		       dm_device_name(md), dd->dm_dev->name);
		dm_put_table_device(md, dd->dm_dev);
L
Linus Torvalds 已提交
229 230 231 232
		kfree(dd);
	}
}

233
void dm_table_destroy(struct dm_table *t)
L
Linus Torvalds 已提交
234 235 236
{
	unsigned int i;

237 238 239
	if (!t)
		return;

240
	/* free the indexes */
L
Linus Torvalds 已提交
241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256
	if (t->depth >= 2)
		vfree(t->index[t->depth - 2]);

	/* free the targets */
	for (i = 0; i < t->num_targets; i++) {
		struct dm_target *tgt = t->targets + i;

		if (tgt->type->dtr)
			tgt->type->dtr(tgt);

		dm_put_target_type(tgt->type);
	}

	vfree(t->highs);

	/* free the device list */
257
	free_devices(&t->devices, t->md);
L
Linus Torvalds 已提交
258

K
Kiyoshi Ueda 已提交
259 260
	dm_free_md_mempools(t->mempools);

L
Linus Torvalds 已提交
261 262 263 264 265 266
	kfree(t);
}

/*
 * See if we've already got a device in the list.
 */
M
Mikulas Patocka 已提交
267
static struct dm_dev_internal *find_device(struct list_head *l, dev_t dev)
L
Linus Torvalds 已提交
268
{
M
Mikulas Patocka 已提交
269
	struct dm_dev_internal *dd;
L
Linus Torvalds 已提交
270 271

	list_for_each_entry (dd, l, list)
272
		if (dd->dm_dev->bdev->bd_dev == dev)
L
Linus Torvalds 已提交
273 274 275 276 277 278
			return dd;

	return NULL;
}

/*
279
 * If possible, this checks an area of a destination device is invalid.
L
Linus Torvalds 已提交
280
 */
281 282
static int device_area_is_invalid(struct dm_target *ti, struct dm_dev *dev,
				  sector_t start, sector_t len, void *data)
L
Linus Torvalds 已提交
283
{
284
	struct request_queue *q;
285 286 287 288
	struct queue_limits *limits = data;
	struct block_device *bdev = dev->bdev;
	sector_t dev_size =
		i_size_read(bdev->bd_inode) >> SECTOR_SHIFT;
289
	unsigned short logical_block_size_sectors =
290
		limits->logical_block_size >> SECTOR_SHIFT;
291
	char b[BDEVNAME_SIZE];
M
Mike Anderson 已提交
292

293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308
	/*
	 * Some devices exist without request functions,
	 * such as loop devices not yet bound to backing files.
	 * Forbid the use of such devices.
	 */
	q = bdev_get_queue(bdev);
	if (!q || !q->make_request_fn) {
		DMWARN("%s: %s is not yet initialised: "
		       "start=%llu, len=%llu, dev_size=%llu",
		       dm_device_name(ti->table->md), bdevname(bdev, b),
		       (unsigned long long)start,
		       (unsigned long long)len,
		       (unsigned long long)dev_size);
		return 1;
	}

M
Mike Anderson 已提交
309
	if (!dev_size)
310
		return 0;
M
Mike Anderson 已提交
311

312
	if ((start >= dev_size) || (start + len > dev_size)) {
313 314 315 316 317 318
		DMWARN("%s: %s too small for target: "
		       "start=%llu, len=%llu, dev_size=%llu",
		       dm_device_name(ti->table->md), bdevname(bdev, b),
		       (unsigned long long)start,
		       (unsigned long long)len,
		       (unsigned long long)dev_size);
319
		return 1;
320 321 322
	}

	if (logical_block_size_sectors <= 1)
323
		return 0;
324 325 326

	if (start & (logical_block_size_sectors - 1)) {
		DMWARN("%s: start=%llu not aligned to h/w "
327
		       "logical block size %u of %s",
328 329
		       dm_device_name(ti->table->md),
		       (unsigned long long)start,
330
		       limits->logical_block_size, bdevname(bdev, b));
331
		return 1;
332 333
	}

334
	if (len & (logical_block_size_sectors - 1)) {
335
		DMWARN("%s: len=%llu not aligned to h/w "
336
		       "logical block size %u of %s",
337
		       dm_device_name(ti->table->md),
338
		       (unsigned long long)len,
339
		       limits->logical_block_size, bdevname(bdev, b));
340
		return 1;
341 342
	}

343
	return 0;
L
Linus Torvalds 已提交
344 345 346
}

/*
347
 * This upgrades the mode on an already open dm_dev, being
L
Linus Torvalds 已提交
348
 * careful to leave things as they were if we fail to reopen the
349 350
 * device and not to touch the existing bdev field in case
 * it is accessed concurrently inside dm_table_any_congested().
L
Linus Torvalds 已提交
351
 */
352
static int upgrade_mode(struct dm_dev_internal *dd, fmode_t new_mode,
M
Mikulas Patocka 已提交
353
			struct mapped_device *md)
L
Linus Torvalds 已提交
354 355
{
	int r;
356
	struct dm_dev *old_dev, *new_dev;
L
Linus Torvalds 已提交
357

358
	old_dev = dd->dm_dev;
359

360 361
	r = dm_get_table_device(md, dd->dm_dev->bdev->bd_dev,
				dd->dm_dev->mode | new_mode, &new_dev);
362 363
	if (r)
		return r;
L
Linus Torvalds 已提交
364

365 366
	dd->dm_dev = new_dev;
	dm_put_table_device(md, old_dev);
L
Linus Torvalds 已提交
367

368
	return 0;
L
Linus Torvalds 已提交
369 370
}

371 372 373 374 375
/*
 * Convert the path to a device
 */
dev_t dm_get_dev_t(const char *path)
{
376
	dev_t dev;
377 378 379 380 381 382 383 384 385 386 387 388 389 390
	struct block_device *bdev;

	bdev = lookup_bdev(path);
	if (IS_ERR(bdev))
		dev = name_to_dev_t(path);
	else {
		dev = bdev->bd_dev;
		bdput(bdev);
	}

	return dev;
}
EXPORT_SYMBOL_GPL(dm_get_dev_t);

L
Linus Torvalds 已提交
391 392 393 394
/*
 * Add a device to the list, or just increment the usage count if
 * it's already present.
 */
395 396
int dm_get_device(struct dm_target *ti, const char *path, fmode_t mode,
		  struct dm_dev **result)
L
Linus Torvalds 已提交
397 398
{
	int r;
399
	dev_t dev;
M
Mikulas Patocka 已提交
400
	struct dm_dev_internal *dd;
401
	struct dm_table *t = ti->table;
L
Linus Torvalds 已提交
402

403
	BUG_ON(!t);
L
Linus Torvalds 已提交
404

405 406 407
	dev = dm_get_dev_t(path);
	if (!dev)
		return -ENODEV;
L
Linus Torvalds 已提交
408 409 410 411 412 413 414

	dd = find_device(&t->devices, dev);
	if (!dd) {
		dd = kmalloc(sizeof(*dd), GFP_KERNEL);
		if (!dd)
			return -ENOMEM;

415
		if ((r = dm_get_table_device(t->md, dev, mode, &dd->dm_dev))) {
L
Linus Torvalds 已提交
416 417 418 419 420 421 422
			kfree(dd);
			return r;
		}

		atomic_set(&dd->count, 0);
		list_add(&dd->list, &t->devices);

423
	} else if (dd->dm_dev->mode != (mode | dd->dm_dev->mode)) {
424
		r = upgrade_mode(dd, mode, t->md);
L
Linus Torvalds 已提交
425 426 427 428 429
		if (r)
			return r;
	}
	atomic_inc(&dd->count);

430
	*result = dd->dm_dev;
L
Linus Torvalds 已提交
431 432
	return 0;
}
433
EXPORT_SYMBOL(dm_get_device);
L
Linus Torvalds 已提交
434

435 436
static int dm_set_device_limits(struct dm_target *ti, struct dm_dev *dev,
				sector_t start, sector_t len, void *data)
L
Linus Torvalds 已提交
437
{
438 439
	struct queue_limits *limits = data;
	struct block_device *bdev = dev->bdev;
440
	struct request_queue *q = bdev_get_queue(bdev);
A
Alasdair G Kergon 已提交
441 442 443 444 445
	char b[BDEVNAME_SIZE];

	if (unlikely(!q)) {
		DMWARN("%s: Cannot set limits for nonexistent device %s",
		       dm_device_name(ti->table->md), bdevname(bdev, b));
446
		return 0;
A
Alasdair G Kergon 已提交
447
	}
448

449 450
	if (bdev_stack_limits(limits, bdev, start) < 0)
		DMWARN("%s: adding target device %s caused an alignment inconsistency: "
451 452 453 454 455 456
		       "physical_block_size=%u, logical_block_size=%u, "
		       "alignment_offset=%u, start=%llu",
		       dm_device_name(ti->table->md), bdevname(bdev, b),
		       q->limits.physical_block_size,
		       q->limits.logical_block_size,
		       q->limits.alignment_offset,
457
		       (unsigned long long) start << SECTOR_SHIFT);
458

459
	return 0;
460
}
461

L
Linus Torvalds 已提交
462
/*
463
 * Decrement a device's use count and remove it if necessary.
L
Linus Torvalds 已提交
464
 */
M
Mikulas Patocka 已提交
465
void dm_put_device(struct dm_target *ti, struct dm_dev *d)
L
Linus Torvalds 已提交
466
{
467 468 469
	int found = 0;
	struct list_head *devices = &ti->table->devices;
	struct dm_dev_internal *dd;
M
Mikulas Patocka 已提交
470

471 472 473 474 475 476 477 478 479 480 481
	list_for_each_entry(dd, devices, list) {
		if (dd->dm_dev == d) {
			found = 1;
			break;
		}
	}
	if (!found) {
		DMWARN("%s: device %s not in table devices list",
		       dm_device_name(ti->table->md), d->name);
		return;
	}
L
Linus Torvalds 已提交
482
	if (atomic_dec_and_test(&dd->count)) {
483
		dm_put_table_device(ti->table->md, d);
L
Linus Torvalds 已提交
484 485 486 487
		list_del(&dd->list);
		kfree(dd);
	}
}
488
EXPORT_SYMBOL(dm_put_device);
L
Linus Torvalds 已提交
489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505

/*
 * Checks to see if the target joins onto the end of the table.
 */
static int adjoin(struct dm_table *table, struct dm_target *ti)
{
	struct dm_target *prev;

	if (!table->num_targets)
		return !ti->begin;

	prev = &table->targets[table->num_targets - 1];
	return (ti->begin == (prev->begin + prev->len));
}

/*
 * Used to dynamically allocate the arg array.
506 507 508 509 510 511 512
 *
 * We do first allocation with GFP_NOIO because dm-mpath and dm-thin must
 * process messages even if some device is suspended. These messages have a
 * small fixed number of arguments.
 *
 * On the other hand, dm-switch needs to process bulk data using messages and
 * excessive use of GFP_NOIO could cause trouble.
L
Linus Torvalds 已提交
513 514 515 516 517
 */
static char **realloc_argv(unsigned *array_size, char **old_argv)
{
	char **argv;
	unsigned new_size;
518
	gfp_t gfp;
L
Linus Torvalds 已提交
519

520 521 522 523 524 525 526 527
	if (*array_size) {
		new_size = *array_size * 2;
		gfp = GFP_KERNEL;
	} else {
		new_size = 8;
		gfp = GFP_NOIO;
	}
	argv = kmalloc(new_size * sizeof(*argv), gfp);
L
Linus Torvalds 已提交
528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545
	if (argv) {
		memcpy(argv, old_argv, *array_size * sizeof(*argv));
		*array_size = new_size;
	}

	kfree(old_argv);
	return argv;
}

/*
 * Destructively splits up the argument list to pass to ctr.
 */
int dm_split_args(int *argc, char ***argvp, char *input)
{
	char *start, *end = input, *out, **argv = NULL;
	unsigned array_size = 0;

	*argc = 0;
546 547 548 549 550 551

	if (!input) {
		*argvp = NULL;
		return 0;
	}

L
Linus Torvalds 已提交
552 553 554 555 556 557
	argv = realloc_argv(&array_size, argv);
	if (!argv)
		return -ENOMEM;

	while (1) {
		/* Skip whitespace */
558
		start = skip_spaces(end);
L
Linus Torvalds 已提交
559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599

		if (!*start)
			break;	/* success, we hit the end */

		/* 'out' is used to remove any back-quotes */
		end = out = start;
		while (*end) {
			/* Everything apart from '\0' can be quoted */
			if (*end == '\\' && *(end + 1)) {
				*out++ = *(end + 1);
				end += 2;
				continue;
			}

			if (isspace(*end))
				break;	/* end of token */

			*out++ = *end++;
		}

		/* have we already filled the array ? */
		if ((*argc + 1) > array_size) {
			argv = realloc_argv(&array_size, argv);
			if (!argv)
				return -ENOMEM;
		}

		/* we know this is whitespace */
		if (*end)
			end++;

		/* terminate the string and put it in the array */
		*out = '\0';
		argv[*argc] = start;
		(*argc)++;
	}

	*argvp = argv;
	return 0;
}

600 601 602 603 604 605 606
/*
 * Impose necessary and sufficient conditions on a devices's table such
 * that any incoming bio which respects its logical_block_size can be
 * processed successfully.  If it falls across the boundary between
 * two or more targets, the size of each piece it gets split into must
 * be compatible with the logical_block_size of the target processing it.
 */
607 608
static int validate_hardware_logical_block_alignment(struct dm_table *table,
						 struct queue_limits *limits)
609 610 611 612 613 614
{
	/*
	 * This function uses arithmetic modulo the logical_block_size
	 * (in units of 512-byte sectors).
	 */
	unsigned short device_logical_block_size_sects =
615
		limits->logical_block_size >> SECTOR_SHIFT;
616 617 618 619 620 621 622 623 624 625 626 627 628

	/*
	 * Offset of the start of the next table entry, mod logical_block_size.
	 */
	unsigned short next_target_start = 0;

	/*
	 * Given an aligned bio that extends beyond the end of a
	 * target, how many sectors must the next target handle?
	 */
	unsigned short remaining = 0;

	struct dm_target *uninitialized_var(ti);
629
	struct queue_limits ti_limits;
630
	unsigned i;
631 632 633 634

	/*
	 * Check each entry in the table in turn.
	 */
635 636
	for (i = 0; i < dm_table_get_num_targets(table); i++) {
		ti = dm_table_get_target(table, i);
637

638
		blk_set_stacking_limits(&ti_limits);
639 640 641 642 643 644

		/* combine all target devices' limits */
		if (ti->type->iterate_devices)
			ti->type->iterate_devices(ti, dm_set_device_limits,
						  &ti_limits);

645 646 647 648 649
		/*
		 * If the remaining sectors fall entirely within this
		 * table entry are they compatible with its logical_block_size?
		 */
		if (remaining < ti->len &&
650
		    remaining & ((ti_limits.logical_block_size >>
651 652 653 654 655 656 657 658 659 660 661 662
				  SECTOR_SHIFT) - 1))
			break;	/* Error */

		next_target_start =
		    (unsigned short) ((next_target_start + ti->len) &
				      (device_logical_block_size_sects - 1));
		remaining = next_target_start ?
		    device_logical_block_size_sects - next_target_start : 0;
	}

	if (remaining) {
		DMWARN("%s: table line %u (start sect %llu len %llu) "
663
		       "not aligned to h/w logical block size %u",
664 665 666
		       dm_device_name(table->md), i,
		       (unsigned long long) ti->begin,
		       (unsigned long long) ti->len,
667
		       limits->logical_block_size);
668 669 670 671 672 673
		return -EINVAL;
	}

	return 0;
}

L
Linus Torvalds 已提交
674 675 676 677 678 679 680
int dm_table_add_target(struct dm_table *t, const char *type,
			sector_t start, sector_t len, char *params)
{
	int r = -EINVAL, argc;
	char **argv;
	struct dm_target *tgt;

681 682 683 684 685 686
	if (t->singleton) {
		DMERR("%s: target type %s must appear alone in table",
		      dm_device_name(t->md), t->targets->type->name);
		return -EINVAL;
	}

687
	BUG_ON(t->num_targets >= t->num_allocated);
L
Linus Torvalds 已提交
688 689 690 691 692

	tgt = t->targets + t->num_targets;
	memset(tgt, 0, sizeof(*tgt));

	if (!len) {
693
		DMERR("%s: zero-length target", dm_device_name(t->md));
L
Linus Torvalds 已提交
694 695 696 697 698
		return -EINVAL;
	}

	tgt->type = dm_get_target_type(type);
	if (!tgt->type) {
699
		DMERR("%s: %s: unknown target type", dm_device_name(t->md), type);
L
Linus Torvalds 已提交
700 701 702
		return -EINVAL;
	}

703 704
	if (dm_target_needs_singleton(tgt->type)) {
		if (t->num_targets) {
705 706
			tgt->error = "singleton target type must appear alone in table";
			goto bad;
707
		}
708
		t->singleton = true;
709 710
	}

711
	if (dm_target_always_writeable(tgt->type) && !(t->mode & FMODE_WRITE)) {
712 713
		tgt->error = "target type may not be included in a read-only table";
		goto bad;
714 715
	}

716 717
	if (t->immutable_target_type) {
		if (t->immutable_target_type != tgt->type) {
718 719
			tgt->error = "immutable target type cannot be mixed with other target types";
			goto bad;
720 721 722
		}
	} else if (dm_target_is_immutable(tgt->type)) {
		if (t->num_targets) {
723 724
			tgt->error = "immutable target type cannot be mixed with other target types";
			goto bad;
725 726 727 728
		}
		t->immutable_target_type = tgt->type;
	}

729 730 731
	if (dm_target_has_integrity(tgt->type))
		t->integrity_added = 1;

L
Linus Torvalds 已提交
732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757
	tgt->table = t;
	tgt->begin = start;
	tgt->len = len;
	tgt->error = "Unknown error";

	/*
	 * Does this target adjoin the previous one ?
	 */
	if (!adjoin(t, tgt)) {
		tgt->error = "Gap in table";
		goto bad;
	}

	r = dm_split_args(&argc, &argv, params);
	if (r) {
		tgt->error = "couldn't split parameters (insufficient memory)";
		goto bad;
	}

	r = tgt->type->ctr(tgt, argc, argv);
	kfree(argv);
	if (r)
		goto bad;

	t->highs[t->num_targets++] = tgt->begin + tgt->len - 1;

758 759
	if (!tgt->num_discard_bios && tgt->discards_supported)
		DMWARN("%s: %s: ignoring discards_supported because num_discard_bios is zero.",
M
Mike Snitzer 已提交
760
		       dm_device_name(t->md), type);
M
Mike Snitzer 已提交
761

L
Linus Torvalds 已提交
762 763 764
	return 0;

 bad:
765
	DMERR("%s: %s: %s", dm_device_name(t->md), type, tgt->error);
L
Linus Torvalds 已提交
766 767 768 769
	dm_put_target_type(tgt->type);
	return r;
}

770 771 772 773 774 775 776
/*
 * Target argument parsing helpers.
 */
static int validate_next_arg(struct dm_arg *arg, struct dm_arg_set *arg_set,
			     unsigned *value, char **error, unsigned grouped)
{
	const char *arg_str = dm_shift_arg(arg_set);
777
	char dummy;
778 779

	if (!arg_str ||
780
	    (sscanf(arg_str, "%u%c", value, &dummy) != 1) ||
781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827
	    (*value < arg->min) ||
	    (*value > arg->max) ||
	    (grouped && arg_set->argc < *value)) {
		*error = arg->error;
		return -EINVAL;
	}

	return 0;
}

int dm_read_arg(struct dm_arg *arg, struct dm_arg_set *arg_set,
		unsigned *value, char **error)
{
	return validate_next_arg(arg, arg_set, value, error, 0);
}
EXPORT_SYMBOL(dm_read_arg);

int dm_read_arg_group(struct dm_arg *arg, struct dm_arg_set *arg_set,
		      unsigned *value, char **error)
{
	return validate_next_arg(arg, arg_set, value, error, 1);
}
EXPORT_SYMBOL(dm_read_arg_group);

const char *dm_shift_arg(struct dm_arg_set *as)
{
	char *r;

	if (as->argc) {
		as->argc--;
		r = *as->argv;
		as->argv++;
		return r;
	}

	return NULL;
}
EXPORT_SYMBOL(dm_shift_arg);

void dm_consume_args(struct dm_arg_set *as, unsigned num_args)
{
	BUG_ON(as->argc < num_args);
	as->argc -= num_args;
	as->argv += num_args;
}
EXPORT_SYMBOL(dm_consume_args);

828
static bool __table_type_bio_based(enum dm_queue_mode table_type)
829 830 831 832 833
{
	return (table_type == DM_TYPE_BIO_BASED ||
		table_type == DM_TYPE_DAX_BIO_BASED);
}

834
static bool __table_type_request_based(enum dm_queue_mode table_type)
835 836 837 838 839
{
	return (table_type == DM_TYPE_REQUEST_BASED ||
		table_type == DM_TYPE_MQ_REQUEST_BASED);
}

840
void dm_table_set_type(struct dm_table *t, enum dm_queue_mode type)
841 842 843 844 845
{
	t->type = type;
}
EXPORT_SYMBOL_GPL(dm_table_set_type);

846 847 848 849 850 851 852 853 854 855 856
static int device_supports_dax(struct dm_target *ti, struct dm_dev *dev,
			       sector_t start, sector_t len, void *data)
{
	struct request_queue *q = bdev_get_queue(dev->bdev);

	return q && blk_queue_dax(q);
}

static bool dm_table_supports_dax(struct dm_table *t)
{
	struct dm_target *ti;
857
	unsigned i;
858 859

	/* Ensure that all targets support DAX. */
860 861
	for (i = 0; i < dm_table_get_num_targets(t); i++) {
		ti = dm_table_get_target(t, i);
862 863 864 865 866 867 868 869 870 871 872 873

		if (!ti->type->direct_access)
			return false;

		if (!ti->type->iterate_devices ||
		    !ti->type->iterate_devices(ti, device_supports_dax, NULL))
			return false;
	}

	return true;
}

874
static int dm_table_determine_type(struct dm_table *t)
K
Kiyoshi Ueda 已提交
875 876
{
	unsigned i;
877
	unsigned bio_based = 0, request_based = 0, hybrid = 0;
878
	unsigned sq_count = 0, mq_count = 0;
K
Kiyoshi Ueda 已提交
879 880
	struct dm_target *tgt;
	struct dm_dev_internal *dd;
881
	struct list_head *devices = dm_table_get_devices(t);
882
	enum dm_queue_mode live_md_type = dm_get_md_type(t->md);
K
Kiyoshi Ueda 已提交
883

884 885 886 887
	if (t->type != DM_TYPE_NONE) {
		/* target already set the table's type */
		if (t->type == DM_TYPE_BIO_BASED)
			return 0;
888
		BUG_ON(t->type == DM_TYPE_DAX_BIO_BASED);
889 890 891
		goto verify_rq_based;
	}

K
Kiyoshi Ueda 已提交
892 893
	for (i = 0; i < t->num_targets; i++) {
		tgt = t->targets + i;
894 895 896
		if (dm_target_hybrid(tgt))
			hybrid = 1;
		else if (dm_target_request_based(tgt))
K
Kiyoshi Ueda 已提交
897 898 899 900 901 902 903 904 905 906 907
			request_based = 1;
		else
			bio_based = 1;

		if (bio_based && request_based) {
			DMWARN("Inconsistent table: different target types"
			       " can't be mixed up");
			return -EINVAL;
		}
	}

908 909 910 911 912 913
	if (hybrid && !bio_based && !request_based) {
		/*
		 * The targets can work either way.
		 * Determine the type from the live device.
		 * Default to bio-based if device is new.
		 */
914
		if (__table_type_request_based(live_md_type))
915 916 917 918 919
			request_based = 1;
		else
			bio_based = 1;
	}

K
Kiyoshi Ueda 已提交
920 921 922
	if (bio_based) {
		/* We must use this table as bio-based */
		t->type = DM_TYPE_BIO_BASED;
M
Mike Snitzer 已提交
923 924
		if (dm_table_supports_dax(t) ||
		    (list_empty(devices) && live_md_type == DM_TYPE_DAX_BIO_BASED))
925
			t->type = DM_TYPE_DAX_BIO_BASED;
K
Kiyoshi Ueda 已提交
926 927 928 929 930
		return 0;
	}

	BUG_ON(!request_based); /* No targets in this table */

931 932 933 934 935 936 937
	/*
	 * The only way to establish DM_TYPE_MQ_REQUEST_BASED is by
	 * having a compatible target use dm_table_set_type.
	 */
	t->type = DM_TYPE_REQUEST_BASED;

verify_rq_based:
938 939 940 941 942 943 944 945 946 947 948
	/*
	 * Request-based dm supports only tables that have a single target now.
	 * To support multiple targets, request splitting support is needed,
	 * and that needs lots of changes in the block-layer.
	 * (e.g. request completion process for partial completion.)
	 */
	if (t->num_targets > 1) {
		DMWARN("Request-based dm doesn't support multiple targets yet");
		return -EINVAL;
	}

949 950 951 952 953 954 955 956 957 958 959 960 961
	if (list_empty(devices)) {
		int srcu_idx;
		struct dm_table *live_table = dm_get_live_table(t->md, &srcu_idx);

		/* inherit live table's type and all_blk_mq */
		if (live_table) {
			t->type = live_table->type;
			t->all_blk_mq = live_table->all_blk_mq;
		}
		dm_put_live_table(t->md, srcu_idx);
		return 0;
	}

K
Kiyoshi Ueda 已提交
962 963
	/* Non-request-stackable devices can't be used for request-based dm */
	list_for_each_entry(dd, devices, list) {
964 965 966 967 968
		struct request_queue *q = bdev_get_queue(dd->dm_dev->bdev);

		if (!blk_queue_stackable(q)) {
			DMERR("table load rejected: including"
			      " non-request-stackable devices");
K
Kiyoshi Ueda 已提交
969 970
			return -EINVAL;
		}
971 972

		if (q->mq_ops)
973 974 975
			mq_count++;
		else
			sq_count++;
976
	}
977 978 979
	if (sq_count && mq_count) {
		DMERR("table load rejected: not all devices are blk-mq request-stackable");
		return -EINVAL;
980
	}
981
	t->all_blk_mq = mq_count > 0;
K
Kiyoshi Ueda 已提交
982

983 984 985 986 987
	if (t->type == DM_TYPE_MQ_REQUEST_BASED && !t->all_blk_mq) {
		DMERR("table load rejected: all devices are not blk-mq request-stackable");
		return -EINVAL;
	}

K
Kiyoshi Ueda 已提交
988 989 990
	return 0;
}

991
enum dm_queue_mode dm_table_get_type(struct dm_table *t)
K
Kiyoshi Ueda 已提交
992 993 994 995
{
	return t->type;
}

996 997 998 999 1000
struct target_type *dm_table_get_immutable_target_type(struct dm_table *t)
{
	return t->immutable_target_type;
}

M
Mike Snitzer 已提交
1001 1002 1003 1004 1005 1006 1007 1008 1009 1010
struct dm_target *dm_table_get_immutable_target(struct dm_table *t)
{
	/* Immutable target is implicitly a singleton */
	if (t->num_targets > 1 ||
	    !dm_target_is_immutable(t->targets[0].type))
		return NULL;

	return t->targets;
}

1011 1012
struct dm_target *dm_table_get_wildcard_target(struct dm_table *t)
{
1013 1014
	struct dm_target *ti;
	unsigned i;
1015

1016 1017
	for (i = 0; i < dm_table_get_num_targets(t); i++) {
		ti = dm_table_get_target(t, i);
1018 1019 1020 1021 1022 1023 1024
		if (dm_target_is_wildcard(ti->type))
			return ti;
	}

	return NULL;
}

1025 1026 1027 1028 1029
bool dm_table_bio_based(struct dm_table *t)
{
	return __table_type_bio_based(dm_table_get_type(t));
}

K
Kiyoshi Ueda 已提交
1030 1031
bool dm_table_request_based(struct dm_table *t)
{
1032
	return __table_type_request_based(dm_table_get_type(t));
1033 1034
}

1035
bool dm_table_all_blk_mq_devices(struct dm_table *t)
1036
{
1037
	return t->all_blk_mq;
K
Kiyoshi Ueda 已提交
1038 1039
}

1040
static int dm_table_alloc_md_mempools(struct dm_table *t, struct mapped_device *md)
K
Kiyoshi Ueda 已提交
1041
{
1042
	enum dm_queue_mode type = dm_table_get_type(t);
1043
	unsigned per_io_data_size = 0;
1044
	struct dm_target *tgt;
M
Mikulas Patocka 已提交
1045
	unsigned i;
K
Kiyoshi Ueda 已提交
1046

1047
	if (unlikely(type == DM_TYPE_NONE)) {
K
Kiyoshi Ueda 已提交
1048 1049 1050 1051
		DMWARN("no table type is set, can't allocate mempools");
		return -EINVAL;
	}

1052
	if (__table_type_bio_based(type))
1053 1054
		for (i = 0; i < t->num_targets; i++) {
			tgt = t->targets + i;
1055
			per_io_data_size = max(per_io_data_size, tgt->per_io_data_size);
1056 1057
		}

1058
	t->mempools = dm_alloc_md_mempools(md, type, t->integrity_supported, per_io_data_size);
1059 1060
	if (!t->mempools)
		return -ENOMEM;
K
Kiyoshi Ueda 已提交
1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075

	return 0;
}

void dm_table_free_md_mempools(struct dm_table *t)
{
	dm_free_md_mempools(t->mempools);
	t->mempools = NULL;
}

struct dm_md_mempools *dm_table_get_md_mempools(struct dm_table *t)
{
	return t->mempools;
}

L
Linus Torvalds 已提交
1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092
static int setup_indexes(struct dm_table *t)
{
	int i;
	unsigned int total = 0;
	sector_t *indexes;

	/* allocate the space for *all* the indexes */
	for (i = t->depth - 2; i >= 0; i--) {
		t->counts[i] = dm_div_up(t->counts[i + 1], CHILDREN_PER_NODE);
		total += t->counts[i];
	}

	indexes = (sector_t *) dm_vcalloc(total, (unsigned long) NODE_SIZE);
	if (!indexes)
		return -ENOMEM;

	/* set up internal nodes, bottom-up */
J
Jun'ichi Nomura 已提交
1093
	for (i = t->depth - 2; i >= 0; i--) {
L
Linus Torvalds 已提交
1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104
		t->index[i] = indexes;
		indexes += (KEYS_PER_NODE * t->counts[i]);
		setup_btree_index(i, t);
	}

	return 0;
}

/*
 * Builds the btree to index the map.
 */
1105
static int dm_table_build_index(struct dm_table *t)
L
Linus Torvalds 已提交
1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123
{
	int r = 0;
	unsigned int leaf_nodes;

	/* how many indexes will the btree have ? */
	leaf_nodes = dm_div_up(t->num_targets, KEYS_PER_NODE);
	t->depth = 1 + int_log(leaf_nodes, CHILDREN_PER_NODE);

	/* leaf layer has already been set up */
	t->counts[t->depth - 1] = leaf_nodes;
	t->index[t->depth - 1] = t->highs;

	if (t->depth >= 2)
		r = setup_indexes(t);

	return r;
}

1124 1125 1126 1127 1128
static bool integrity_profile_exists(struct gendisk *disk)
{
	return !!blk_get_integrity(disk);
}

1129 1130 1131 1132
/*
 * Get a disk whose integrity profile reflects the table's profile.
 * Returns NULL if integrity support was inconsistent or unavailable.
 */
1133
static struct gendisk * dm_table_get_integrity_disk(struct dm_table *t)
1134 1135 1136 1137
{
	struct list_head *devices = dm_table_get_devices(t);
	struct dm_dev_internal *dd = NULL;
	struct gendisk *prev_disk = NULL, *template_disk = NULL;
1138 1139 1140 1141 1142 1143 1144
	unsigned i;

	for (i = 0; i < dm_table_get_num_targets(t); i++) {
		struct dm_target *ti = dm_table_get_target(t, i);
		if (!dm_target_passes_integrity(ti->type))
			goto no_integrity;
	}
1145 1146

	list_for_each_entry(dd, devices, list) {
1147
		template_disk = dd->dm_dev->bdev->bd_disk;
1148
		if (!integrity_profile_exists(template_disk))
1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166
			goto no_integrity;
		else if (prev_disk &&
			 blk_integrity_compare(prev_disk, template_disk) < 0)
			goto no_integrity;
		prev_disk = template_disk;
	}

	return template_disk;

no_integrity:
	if (prev_disk)
		DMWARN("%s: integrity not set: %s and %s profile mismatch",
		       dm_device_name(t->md),
		       prev_disk->disk_name,
		       template_disk->disk_name);
	return NULL;
}

1167
/*
1168 1169 1170
 * Register the mapped device for blk_integrity support if the
 * underlying devices have an integrity profile.  But all devices may
 * not have matching profiles (checking all devices isn't reliable
1171
 * during table load because this table may use other DM device(s) which
1172 1173 1174 1175
 * must be resumed before they will have an initialized integity
 * profile).  Consequently, stacked DM devices force a 2 stage integrity
 * profile validation: First pass during table load, final pass during
 * resume.
1176
 */
1177
static int dm_table_register_integrity(struct dm_table *t)
1178
{
1179
	struct mapped_device *md = t->md;
1180
	struct gendisk *template_disk = NULL;
1181

1182 1183 1184 1185
	/* If target handles integrity itself do not register it here. */
	if (t->integrity_added)
		return 0;

1186
	template_disk = dm_table_get_integrity_disk(t);
1187 1188
	if (!template_disk)
		return 0;
1189

1190
	if (!integrity_profile_exists(dm_disk(md))) {
1191
		t->integrity_supported = true;
1192 1193 1194 1195 1196 1197 1198
		/*
		 * Register integrity profile during table load; we can do
		 * this because the final profile must match during resume.
		 */
		blk_integrity_register(dm_disk(md),
				       blk_get_integrity(template_disk));
		return 0;
1199 1200 1201
	}

	/*
1202
	 * If DM device already has an initialized integrity
1203 1204
	 * profile the new profile should not conflict.
	 */
1205
	if (blk_integrity_compare(dm_disk(md), template_disk) < 0) {
1206 1207 1208 1209 1210 1211 1212
		DMWARN("%s: conflict with existing integrity profile: "
		       "%s profile mismatch",
		       dm_device_name(t->md),
		       template_disk->disk_name);
		return 1;
	}

1213
	/* Preserve existing integrity profile */
1214
	t->integrity_supported = true;
1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225
	return 0;
}

/*
 * Prepares the table for use by building the indices,
 * setting the type, and allocating mempools.
 */
int dm_table_complete(struct dm_table *t)
{
	int r;

1226
	r = dm_table_determine_type(t);
1227
	if (r) {
1228
		DMERR("unable to determine table type");
1229 1230 1231 1232 1233 1234 1235 1236 1237
		return r;
	}

	r = dm_table_build_index(t);
	if (r) {
		DMERR("unable to build btrees");
		return r;
	}

1238
	r = dm_table_register_integrity(t);
1239 1240 1241 1242 1243
	if (r) {
		DMERR("could not register integrity profile.");
		return r;
	}

1244
	r = dm_table_alloc_md_mempools(t, t->md);
1245 1246 1247 1248 1249 1250
	if (r)
		DMERR("unable to allocate mempools");

	return r;
}

A
Arjan van de Ven 已提交
1251
static DEFINE_MUTEX(_event_lock);
L
Linus Torvalds 已提交
1252 1253 1254
void dm_table_event_callback(struct dm_table *t,
			     void (*fn)(void *), void *context)
{
A
Arjan van de Ven 已提交
1255
	mutex_lock(&_event_lock);
L
Linus Torvalds 已提交
1256 1257
	t->event_fn = fn;
	t->event_context = context;
A
Arjan van de Ven 已提交
1258
	mutex_unlock(&_event_lock);
L
Linus Torvalds 已提交
1259 1260 1261 1262 1263 1264 1265 1266 1267 1268
}

void dm_table_event(struct dm_table *t)
{
	/*
	 * You can no longer call dm_table_event() from interrupt
	 * context, use a bottom half instead.
	 */
	BUG_ON(in_interrupt());

A
Arjan van de Ven 已提交
1269
	mutex_lock(&_event_lock);
L
Linus Torvalds 已提交
1270 1271
	if (t->event_fn)
		t->event_fn(t->event_context);
A
Arjan van de Ven 已提交
1272
	mutex_unlock(&_event_lock);
L
Linus Torvalds 已提交
1273
}
1274
EXPORT_SYMBOL(dm_table_event);
L
Linus Torvalds 已提交
1275 1276 1277 1278 1279

sector_t dm_table_get_size(struct dm_table *t)
{
	return t->num_targets ? (t->highs[t->num_targets - 1] + 1) : 0;
}
1280
EXPORT_SYMBOL(dm_table_get_size);
L
Linus Torvalds 已提交
1281 1282 1283

struct dm_target *dm_table_get_target(struct dm_table *t, unsigned int index)
{
1284
	if (index >= t->num_targets)
L
Linus Torvalds 已提交
1285 1286 1287 1288 1289 1290 1291
		return NULL;

	return t->targets + index;
}

/*
 * Search the btree for the correct target.
1292 1293 1294
 *
 * Caller should check returned pointer with dm_target_is_valid()
 * to trap I/O beyond end of device.
L
Linus Torvalds 已提交
1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312
 */
struct dm_target *dm_table_find_target(struct dm_table *t, sector_t sector)
{
	unsigned int l, n = 0, k = 0;
	sector_t *node;

	for (l = 0; l < t->depth; l++) {
		n = get_child(n, k);
		node = get_node(t, l, n);

		for (k = 0; k < KEYS_PER_NODE; k++)
			if (node[k] >= sector)
				break;
	}

	return &t->targets[(KEYS_PER_NODE * n) + k];
}

1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330
static int count_device(struct dm_target *ti, struct dm_dev *dev,
			sector_t start, sector_t len, void *data)
{
	unsigned *num_devices = data;

	(*num_devices)++;

	return 0;
}

/*
 * Check whether a table has no data devices attached using each
 * target's iterate_devices method.
 * Returns false if the result is unknown because a target doesn't
 * support iterate_devices.
 */
bool dm_table_has_no_data_devices(struct dm_table *table)
{
1331 1332
	struct dm_target *ti;
	unsigned i, num_devices;
1333

1334 1335
	for (i = 0; i < dm_table_get_num_targets(table); i++) {
		ti = dm_table_get_target(table, i);
1336 1337 1338 1339

		if (!ti->type->iterate_devices)
			return false;

1340
		num_devices = 0;
1341 1342 1343 1344 1345 1346 1347 1348
		ti->type->iterate_devices(ti, count_device, &num_devices);
		if (num_devices)
			return false;
	}

	return true;
}

1349 1350 1351 1352 1353 1354
/*
 * Establish the new table's queue_limits and validate them.
 */
int dm_calculate_queue_limits(struct dm_table *table,
			      struct queue_limits *limits)
{
1355
	struct dm_target *ti;
1356
	struct queue_limits ti_limits;
1357
	unsigned i;
1358

1359
	blk_set_stacking_limits(limits);
1360

1361
	for (i = 0; i < dm_table_get_num_targets(table); i++) {
1362
		blk_set_stacking_limits(&ti_limits);
1363

1364
		ti = dm_table_get_target(table, i);
1365 1366 1367 1368 1369 1370 1371 1372 1373 1374

		if (!ti->type->iterate_devices)
			goto combine_limits;

		/*
		 * Combine queue limits of all the devices this target uses.
		 */
		ti->type->iterate_devices(ti, dm_set_device_limits,
					  &ti_limits);

1375 1376 1377 1378
		/* Set I/O hints portion of queue limits */
		if (ti->type->io_hints)
			ti->type->io_hints(ti, &ti_limits);

1379 1380 1381 1382
		/*
		 * Check each device area is consistent with the target's
		 * overall queue limits.
		 */
1383 1384
		if (ti->type->iterate_devices(ti, device_area_is_invalid,
					      &ti_limits))
1385 1386 1387 1388 1389 1390 1391 1392
			return -EINVAL;

combine_limits:
		/*
		 * Merge this target's queue limits into the overall limits
		 * for the table.
		 */
		if (blk_stack_limits(limits, &ti_limits, 0) < 0)
1393
			DMWARN("%s: adding target device "
1394
			       "(start sect %llu len %llu) "
1395
			       "caused an alignment inconsistency",
1396 1397 1398 1399 1400 1401 1402 1403
			       dm_device_name(table->md),
			       (unsigned long long) ti->begin,
			       (unsigned long long) ti->len);
	}

	return validate_hardware_logical_block_alignment(table, limits);
}

M
Martin K. Petersen 已提交
1404
/*
1405 1406 1407
 * Verify that all devices have an integrity profile that matches the
 * DM device's registered integrity profile.  If the profiles don't
 * match then unregister the DM device's integrity profile.
M
Martin K. Petersen 已提交
1408
 */
1409
static void dm_table_verify_integrity(struct dm_table *t)
M
Martin K. Petersen 已提交
1410
{
1411
	struct gendisk *template_disk = NULL;
M
Martin K. Petersen 已提交
1412

1413 1414 1415
	if (t->integrity_added)
		return;

1416 1417 1418 1419 1420 1421 1422 1423 1424 1425
	if (t->integrity_supported) {
		/*
		 * Verify that the original integrity profile
		 * matches all the devices in this table.
		 */
		template_disk = dm_table_get_integrity_disk(t);
		if (template_disk &&
		    blk_integrity_compare(dm_disk(t->md), template_disk) >= 0)
			return;
	}
M
Martin K. Petersen 已提交
1426

1427
	if (integrity_profile_exists(dm_disk(t->md))) {
1428 1429
		DMWARN("%s: unable to establish an integrity profile",
		       dm_device_name(t->md));
1430 1431
		blk_integrity_unregister(dm_disk(t->md));
	}
M
Martin K. Petersen 已提交
1432 1433
}

1434 1435 1436
static int device_flush_capable(struct dm_target *ti, struct dm_dev *dev,
				sector_t start, sector_t len, void *data)
{
J
Jens Axboe 已提交
1437
	unsigned long flush = (unsigned long) data;
1438 1439
	struct request_queue *q = bdev_get_queue(dev->bdev);

J
Jens Axboe 已提交
1440
	return q && (q->queue_flags & flush);
1441 1442
}

J
Jens Axboe 已提交
1443
static bool dm_table_supports_flush(struct dm_table *t, unsigned long flush)
1444 1445
{
	struct dm_target *ti;
1446
	unsigned i;
1447 1448 1449 1450 1451 1452 1453

	/*
	 * Require at least one underlying device to support flushes.
	 * t->devices includes internal dm devices such as mirror logs
	 * so we need to use iterate_devices here, which targets
	 * supporting flushes must provide.
	 */
1454 1455
	for (i = 0; i < dm_table_get_num_targets(t); i++) {
		ti = dm_table_get_target(t, i);
1456

1457
		if (!ti->num_flush_bios)
1458 1459
			continue;

1460
		if (ti->flush_supported)
1461
			return true;
1462

1463
		if (ti->type->iterate_devices &&
J
Jens Axboe 已提交
1464
		    ti->type->iterate_devices(ti, device_flush_capable, (void *) flush))
1465
			return true;
1466 1467
	}

1468
	return false;
1469 1470
}

1471 1472 1473 1474 1475 1476 1477 1478
static int device_is_nonrot(struct dm_target *ti, struct dm_dev *dev,
			    sector_t start, sector_t len, void *data)
{
	struct request_queue *q = bdev_get_queue(dev->bdev);

	return q && blk_queue_nonrot(q);
}

1479 1480 1481 1482 1483 1484 1485 1486
static int device_is_not_random(struct dm_target *ti, struct dm_dev *dev,
			     sector_t start, sector_t len, void *data)
{
	struct request_queue *q = bdev_get_queue(dev->bdev);

	return q && !blk_queue_add_random(q);
}

1487 1488 1489 1490 1491 1492 1493 1494
static int queue_supports_sg_merge(struct dm_target *ti, struct dm_dev *dev,
				   sector_t start, sector_t len, void *data)
{
	struct request_queue *q = bdev_get_queue(dev->bdev);

	return q && !test_bit(QUEUE_FLAG_NO_SG_MERGE, &q->queue_flags);
}

1495 1496
static bool dm_table_all_devices_attribute(struct dm_table *t,
					   iterate_devices_callout_fn func)
1497 1498
{
	struct dm_target *ti;
1499
	unsigned i;
1500

1501 1502
	for (i = 0; i < dm_table_get_num_targets(t); i++) {
		ti = dm_table_get_target(t, i);
1503 1504

		if (!ti->type->iterate_devices ||
1505
		    !ti->type->iterate_devices(ti, func, NULL))
1506
			return false;
1507 1508
	}

1509
	return true;
1510 1511
}

M
Mike Snitzer 已提交
1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522
static int device_not_write_same_capable(struct dm_target *ti, struct dm_dev *dev,
					 sector_t start, sector_t len, void *data)
{
	struct request_queue *q = bdev_get_queue(dev->bdev);

	return q && !q->limits.max_write_same_sectors;
}

static bool dm_table_supports_write_same(struct dm_table *t)
{
	struct dm_target *ti;
1523
	unsigned i;
M
Mike Snitzer 已提交
1524

1525 1526
	for (i = 0; i < dm_table_get_num_targets(t); i++) {
		ti = dm_table_get_target(t, i);
M
Mike Snitzer 已提交
1527

1528
		if (!ti->num_write_same_bios)
M
Mike Snitzer 已提交
1529 1530 1531
			return false;

		if (!ti->type->iterate_devices ||
M
Mike Snitzer 已提交
1532
		    ti->type->iterate_devices(ti, device_not_write_same_capable, NULL))
M
Mike Snitzer 已提交
1533 1534 1535 1536 1537 1538
			return false;
	}

	return true;
}

1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566
static int device_not_write_zeroes_capable(struct dm_target *ti, struct dm_dev *dev,
					   sector_t start, sector_t len, void *data)
{
	struct request_queue *q = bdev_get_queue(dev->bdev);

	return q && !q->limits.max_write_zeroes_sectors;
}

static bool dm_table_supports_write_zeroes(struct dm_table *t)
{
	struct dm_target *ti;
	unsigned i = 0;

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

		if (!ti->num_write_zeroes_bios)
			return false;

		if (!ti->type->iterate_devices ||
		    ti->type->iterate_devices(ti, device_not_write_zeroes_capable, NULL))
			return false;
	}

	return true;
}


1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577
static int device_discard_capable(struct dm_target *ti, struct dm_dev *dev,
				  sector_t start, sector_t len, void *data)
{
	struct request_queue *q = bdev_get_queue(dev->bdev);

	return q && blk_queue_discard(q);
}

static bool dm_table_supports_discards(struct dm_table *t)
{
	struct dm_target *ti;
1578
	unsigned i;
1579 1580 1581 1582 1583 1584 1585 1586

	/*
	 * Unless any target used by the table set discards_supported,
	 * require at least one underlying device to support discards.
	 * t->devices includes internal dm devices such as mirror logs
	 * so we need to use iterate_devices here, which targets
	 * supporting discard selectively must provide.
	 */
1587 1588
	for (i = 0; i < dm_table_get_num_targets(t); i++) {
		ti = dm_table_get_target(t, i);
1589 1590 1591 1592 1593

		if (!ti->num_discard_bios)
			continue;

		if (ti->discards_supported)
1594
			return true;
1595 1596 1597

		if (ti->type->iterate_devices &&
		    ti->type->iterate_devices(ti, device_discard_capable, NULL))
1598
			return true;
1599 1600
	}

1601
	return false;
1602 1603
}

1604 1605
void dm_table_set_restrictions(struct dm_table *t, struct request_queue *q,
			       struct queue_limits *limits)
L
Linus Torvalds 已提交
1606
{
1607
	bool wc = false, fua = false;
1608

L
Linus Torvalds 已提交
1609
	/*
1610
	 * Copy table's limits to the DM device's request_queue
L
Linus Torvalds 已提交
1611
	 */
1612
	q->limits = *limits;
1613

M
Mike Snitzer 已提交
1614 1615 1616 1617 1618
	if (!dm_table_supports_discards(t))
		queue_flag_clear_unlocked(QUEUE_FLAG_DISCARD, q);
	else
		queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, q);

J
Jens Axboe 已提交
1619
	if (dm_table_supports_flush(t, (1UL << QUEUE_FLAG_WC))) {
1620
		wc = true;
J
Jens Axboe 已提交
1621
		if (dm_table_supports_flush(t, (1UL << QUEUE_FLAG_FUA)))
1622
			fua = true;
1623
	}
1624
	blk_queue_write_cache(q, wc, fua);
1625

1626 1627
	/* Ensure that all underlying devices are non-rotational. */
	if (dm_table_all_devices_attribute(t, device_is_nonrot))
1628 1629 1630 1631
		queue_flag_set_unlocked(QUEUE_FLAG_NONROT, q);
	else
		queue_flag_clear_unlocked(QUEUE_FLAG_NONROT, q);

M
Mike Snitzer 已提交
1632 1633
	if (!dm_table_supports_write_same(t))
		q->limits.max_write_same_sectors = 0;
1634 1635
	if (!dm_table_supports_write_zeroes(t))
		q->limits.max_write_zeroes_sectors = 0;
M
Mike Snitzer 已提交
1636

1637 1638 1639 1640 1641
	if (dm_table_all_devices_attribute(t, queue_supports_sg_merge))
		queue_flag_clear_unlocked(QUEUE_FLAG_NO_SG_MERGE, q);
	else
		queue_flag_set_unlocked(QUEUE_FLAG_NO_SG_MERGE, q);

1642
	dm_table_verify_integrity(t);
K
Kiyoshi Ueda 已提交
1643

1644 1645 1646 1647 1648 1649 1650 1651 1652
	/*
	 * Determine whether or not this queue's I/O timings contribute
	 * to the entropy pool, Only request-based targets use this.
	 * Clear QUEUE_FLAG_ADD_RANDOM if any underlying device does not
	 * have it set.
	 */
	if (blk_queue_add_random(q) && dm_table_all_devices_attribute(t, device_is_not_random))
		queue_flag_clear_unlocked(QUEUE_FLAG_ADD_RANDOM, q);

K
Kiyoshi Ueda 已提交
1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664
	/*
	 * QUEUE_FLAG_STACKABLE must be set after all queue settings are
	 * visible to other CPUs because, once the flag is set, incoming bios
	 * are processed by request-based dm, which refers to the queue
	 * settings.
	 * Until the flag set, bios are passed to bio-based dm and queued to
	 * md->deferred where queue settings are not needed yet.
	 * Those bios are passed to request-based dm at the resume time.
	 */
	smp_mb();
	if (dm_table_request_based(t))
		queue_flag_set_unlocked(QUEUE_FLAG_STACKABLE, q);
L
Linus Torvalds 已提交
1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676
}

unsigned int dm_table_get_num_targets(struct dm_table *t)
{
	return t->num_targets;
}

struct list_head *dm_table_get_devices(struct dm_table *t)
{
	return &t->devices;
}

1677
fmode_t dm_table_get_mode(struct dm_table *t)
L
Linus Torvalds 已提交
1678 1679 1680
{
	return t->mode;
}
1681
EXPORT_SYMBOL(dm_table_get_mode);
L
Linus Torvalds 已提交
1682

1683 1684 1685 1686 1687 1688 1689
enum suspend_mode {
	PRESUSPEND,
	PRESUSPEND_UNDO,
	POSTSUSPEND,
};

static void suspend_targets(struct dm_table *t, enum suspend_mode mode)
L
Linus Torvalds 已提交
1690 1691 1692 1693
{
	int i = t->num_targets;
	struct dm_target *ti = t->targets;

1694 1695
	lockdep_assert_held(&t->md->suspend_lock);

L
Linus Torvalds 已提交
1696
	while (i--) {
1697 1698 1699 1700 1701 1702 1703 1704 1705 1706
		switch (mode) {
		case PRESUSPEND:
			if (ti->type->presuspend)
				ti->type->presuspend(ti);
			break;
		case PRESUSPEND_UNDO:
			if (ti->type->presuspend_undo)
				ti->type->presuspend_undo(ti);
			break;
		case POSTSUSPEND:
L
Linus Torvalds 已提交
1707 1708
			if (ti->type->postsuspend)
				ti->type->postsuspend(ti);
1709 1710
			break;
		}
L
Linus Torvalds 已提交
1711 1712 1713 1714 1715 1716
		ti++;
	}
}

void dm_table_presuspend_targets(struct dm_table *t)
{
1717 1718 1719
	if (!t)
		return;

1720 1721 1722 1723 1724 1725 1726 1727 1728
	suspend_targets(t, PRESUSPEND);
}

void dm_table_presuspend_undo_targets(struct dm_table *t)
{
	if (!t)
		return;

	suspend_targets(t, PRESUSPEND_UNDO);
L
Linus Torvalds 已提交
1729 1730 1731 1732
}

void dm_table_postsuspend_targets(struct dm_table *t)
{
1733 1734 1735
	if (!t)
		return;

1736
	suspend_targets(t, POSTSUSPEND);
L
Linus Torvalds 已提交
1737 1738
}

1739
int dm_table_resume_targets(struct dm_table *t)
L
Linus Torvalds 已提交
1740
{
1741 1742
	int i, r = 0;

1743 1744
	lockdep_assert_held(&t->md->suspend_lock);

1745 1746 1747 1748 1749 1750 1751
	for (i = 0; i < t->num_targets; i++) {
		struct dm_target *ti = t->targets + i;

		if (!ti->type->preresume)
			continue;

		r = ti->type->preresume(ti);
1752 1753 1754
		if (r) {
			DMERR("%s: %s: preresume failed, error = %d",
			      dm_device_name(t->md), ti->type->name, r);
1755
			return r;
1756
		}
1757
	}
L
Linus Torvalds 已提交
1758 1759 1760 1761 1762 1763 1764

	for (i = 0; i < t->num_targets; i++) {
		struct dm_target *ti = t->targets + i;

		if (ti->type->resume)
			ti->type->resume(ti);
	}
1765 1766

	return 0;
L
Linus Torvalds 已提交
1767 1768
}

1769 1770 1771 1772 1773 1774
void dm_table_add_target_callbacks(struct dm_table *t, struct dm_target_callbacks *cb)
{
	list_add(&cb->list, &t->target_callbacks);
}
EXPORT_SYMBOL_GPL(dm_table_add_target_callbacks);

L
Linus Torvalds 已提交
1775 1776
int dm_table_any_congested(struct dm_table *t, int bdi_bits)
{
M
Mikulas Patocka 已提交
1777
	struct dm_dev_internal *dd;
P
Paul Jimenez 已提交
1778
	struct list_head *devices = dm_table_get_devices(t);
1779
	struct dm_target_callbacks *cb;
L
Linus Torvalds 已提交
1780 1781
	int r = 0;

P
Paul Jimenez 已提交
1782
	list_for_each_entry(dd, devices, list) {
1783
		struct request_queue *q = bdev_get_queue(dd->dm_dev->bdev);
A
Alasdair G Kergon 已提交
1784 1785 1786
		char b[BDEVNAME_SIZE];

		if (likely(q))
1787
			r |= bdi_congested(q->backing_dev_info, bdi_bits);
A
Alasdair G Kergon 已提交
1788 1789 1790
		else
			DMWARN_LIMIT("%s: any_congested: nonexistent device %s",
				     dm_device_name(t->md),
1791
				     bdevname(dd->dm_dev->bdev, b));
L
Linus Torvalds 已提交
1792 1793
	}

1794 1795 1796 1797
	list_for_each_entry(cb, &t->target_callbacks, list)
		if (cb->congested_fn)
			r |= cb->congested_fn(cb, bdi_bits);

L
Linus Torvalds 已提交
1798 1799 1800
	return r;
}

M
Mike Anderson 已提交
1801 1802 1803 1804
struct mapped_device *dm_table_get_md(struct dm_table *t)
{
	return t->md;
}
1805
EXPORT_SYMBOL(dm_table_get_md);
M
Mike Anderson 已提交
1806

1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818
void dm_table_run_md_queue_async(struct dm_table *t)
{
	struct mapped_device *md;
	struct request_queue *queue;
	unsigned long flags;

	if (!dm_table_request_based(t))
		return;

	md = dm_table_get_md(t);
	queue = dm_get_md_queue(md);
	if (queue) {
1819 1820 1821 1822 1823 1824 1825
		if (queue->mq_ops)
			blk_mq_run_hw_queues(queue, true);
		else {
			spin_lock_irqsave(queue->queue_lock, flags);
			blk_run_queue_async(queue);
			spin_unlock_irqrestore(queue->queue_lock, flags);
		}
1826 1827 1828 1829
	}
}
EXPORT_SYMBOL(dm_table_run_md_queue_async);