dm-raid.c 46.8 KB
Newer Older
NeilBrown's avatar
NeilBrown committed
1
2
/*
 * Copyright (C) 2010-2011 Neil Brown
3
 * Copyright (C) 2010-2014 Red Hat, Inc. All rights reserved.
NeilBrown's avatar
NeilBrown committed
4
5
6
7
8
 *
 * This file is released under the GPL.
 */

#include <linux/slab.h>
9
#include <linux/module.h>
NeilBrown's avatar
NeilBrown committed
10
11

#include "md.h"
12
#include "raid1.h"
NeilBrown's avatar
NeilBrown committed
13
#include "raid5.h"
14
#include "raid10.h"
NeilBrown's avatar
NeilBrown committed
15
16
#include "bitmap.h"

Alasdair G Kergon's avatar
Alasdair G Kergon committed
17
18
#include <linux/device-mapper.h>

NeilBrown's avatar
NeilBrown committed
19
20
#define DM_MSG_PREFIX "raid"

21
22
static bool devices_handle_discard_safely = false;

NeilBrown's avatar
NeilBrown committed
23
/*
24
25
 * The following flags are used by dm-raid.c to set up the array state.
 * They must be cleared before md_run is called.
NeilBrown's avatar
NeilBrown committed
26
 */
27
#define FirstUse 10             /* rdev flag */
NeilBrown's avatar
NeilBrown committed
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43

struct raid_dev {
	/*
	 * Two DM devices, one to hold metadata and one to hold the
	 * actual data/parity.  The reason for this is to not confuse
	 * ti->len and give more flexibility in altering size and
	 * characteristics.
	 *
	 * While it is possible for this device to be associated
	 * with a different physical device than the data_dev, it
	 * is intended for it to be the same.
	 *    |--------- Physical Device ---------|
	 *    |- meta_dev -|------ data_dev ------|
	 */
	struct dm_dev *meta_dev;
	struct dm_dev *data_dev;
44
	struct md_rdev rdev;
NeilBrown's avatar
NeilBrown committed
45
46
47
48
49
};

/*
 * Flags for rs->print_flags field.
 */
50
51
52
53
54
55
56
57
#define DMPF_SYNC              0x1
#define DMPF_NOSYNC            0x2
#define DMPF_REBUILD           0x4
#define DMPF_DAEMON_SLEEP      0x8
#define DMPF_MIN_RECOVERY_RATE 0x10
#define DMPF_MAX_RECOVERY_RATE 0x20
#define DMPF_MAX_WRITE_BEHIND  0x40
#define DMPF_STRIPE_CACHE      0x80
58
59
60
61
#define DMPF_REGION_SIZE       0x100
#define DMPF_RAID10_COPIES     0x200
#define DMPF_RAID10_FORMAT     0x400

NeilBrown's avatar
NeilBrown committed
62
63
64
struct raid_set {
	struct dm_target *ti;

65
66
	uint32_t bitmap_loaded;
	uint32_t print_flags;
NeilBrown's avatar
NeilBrown committed
67

68
	struct mddev md;
NeilBrown's avatar
NeilBrown committed
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
	struct raid_type *raid_type;
	struct dm_target_callbacks callbacks;

	struct raid_dev dev[0];
};

/* Supported raid types and properties. */
static struct raid_type {
	const char *name;		/* RAID algorithm. */
	const char *descr;		/* Descriptor text for logging. */
	const unsigned parity_devs;	/* # of parity devices. */
	const unsigned minimal_devs;	/* minimal # of devices in set. */
	const unsigned level;		/* RAID level. */
	const unsigned algorithm;	/* RAID algorithm. */
} raid_types[] = {
84
	{"raid1",    "RAID1 (mirroring)",               0, 2, 1, 0 /* NONE */},
85
	{"raid10",   "RAID10 (striped mirrors)",        0, 2, 10, UINT_MAX /* Varies */},
NeilBrown's avatar
NeilBrown committed
86
87
88
89
90
91
92
93
94
95
	{"raid4",    "RAID4 (dedicated parity disk)",	1, 2, 5, ALGORITHM_PARITY_0},
	{"raid5_la", "RAID5 (left asymmetric)",		1, 2, 5, ALGORITHM_LEFT_ASYMMETRIC},
	{"raid5_ra", "RAID5 (right asymmetric)",	1, 2, 5, ALGORITHM_RIGHT_ASYMMETRIC},
	{"raid5_ls", "RAID5 (left symmetric)",		1, 2, 5, ALGORITHM_LEFT_SYMMETRIC},
	{"raid5_rs", "RAID5 (right symmetric)",		1, 2, 5, ALGORITHM_RIGHT_SYMMETRIC},
	{"raid6_zr", "RAID6 (zero restart)",		2, 4, 6, ALGORITHM_ROTATING_ZERO_RESTART},
	{"raid6_nr", "RAID6 (N restart)",		2, 4, 6, ALGORITHM_ROTATING_N_RESTART},
	{"raid6_nc", "RAID6 (N continue)",		2, 4, 6, ALGORITHM_ROTATING_N_CONTINUE}
};

96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
static char *raid10_md_layout_to_format(int layout)
{
	/*
	 * Bit 16 and 17 stand for "offset" and "use_far_sets"
	 * Refer to MD's raid10.c for details
	 */
	if ((layout & 0x10000) && (layout & 0x20000))
		return "offset";

	if ((layout & 0xFF) > 1)
		return "near";

	return "far";
}

111
112
static unsigned raid10_md_layout_to_copies(int layout)
{
113
114
115
	if ((layout & 0xFF) > 1)
		return layout & 0xFF;
	return (layout >> 8) & 0xFF;
116
117
118
119
}

static int raid10_format_to_md_layout(char *format, unsigned copies)
{
120
121
122
123
124
125
126
127
128
129
130
131
132
133
	unsigned n = 1, f = 1;

	if (!strcmp("near", format))
		n = copies;
	else
		f = copies;

	if (!strcmp("offset", format))
		return 0x30000 | (f << 8) | n;

	if (!strcmp("far", format))
		return 0x20000 | (f << 8) | n;

	return (f << 8) | n;
134
135
}

NeilBrown's avatar
NeilBrown committed
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
static struct raid_type *get_raid_type(char *name)
{
	int i;

	for (i = 0; i < ARRAY_SIZE(raid_types); i++)
		if (!strcmp(raid_types[i].name, name))
			return &raid_types[i];

	return NULL;
}

static struct raid_set *context_alloc(struct dm_target *ti, struct raid_type *raid_type, unsigned raid_devs)
{
	unsigned i;
	struct raid_set *rs;

	if (raid_devs <= raid_type->parity_devs) {
		ti->error = "Insufficient number of devices";
		return ERR_PTR(-EINVAL);
	}

	rs = kzalloc(sizeof(*rs) + raid_devs * sizeof(rs->dev[0]), GFP_KERNEL);
	if (!rs) {
		ti->error = "Cannot allocate raid context";
		return ERR_PTR(-ENOMEM);
	}

	mddev_init(&rs->md);

	rs->ti = ti;
	rs->raid_type = raid_type;
	rs->md.raid_disks = raid_devs;
	rs->md.level = raid_type->level;
	rs->md.new_level = rs->md.level;
	rs->md.layout = raid_type->algorithm;
	rs->md.new_layout = rs->md.layout;
	rs->md.delta_disks = 0;
	rs->md.recovery_cp = 0;

	for (i = 0; i < raid_devs; i++)
		md_rdev_init(&rs->dev[i].rdev);

	/*
	 * Remaining items to be initialized by further RAID params:
	 *  rs->md.persistent
	 *  rs->md.external
	 *  rs->md.chunk_sectors
	 *  rs->md.new_chunk_sectors
184
	 *  rs->md.dev_sectors
NeilBrown's avatar
NeilBrown committed
185
186
187
188
189
190
191
192
193
	 */

	return rs;
}

static void context_free(struct raid_set *rs)
{
	int i;

194
195
196
	for (i = 0; i < rs->md.raid_disks; i++) {
		if (rs->dev[i].meta_dev)
			dm_put_device(rs->ti, rs->dev[i].meta_dev);
197
		md_rdev_clear(&rs->dev[i].rdev);
NeilBrown's avatar
NeilBrown committed
198
199
		if (rs->dev[i].data_dev)
			dm_put_device(rs->ti, rs->dev[i].data_dev);
200
	}
NeilBrown's avatar
NeilBrown committed
201
202
203
204
205
206
207
208
209

	kfree(rs);
}

/*
 * For every device we have two words
 *  <meta_dev>: meta device name or '-' if missing
 *  <data_dev>: data device name or '-' if missing
 *
210
211
212
213
214
215
216
217
218
219
 * The following are permitted:
 *    - -
 *    - <data_dev>
 *    <meta_dev> <data_dev>
 *
 * The following is not allowed:
 *    <meta_dev> -
 *
 * This code parses those words.  If there is a failure,
 * the caller must use context_free to unwind the operations.
NeilBrown's avatar
NeilBrown committed
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
 */
static int dev_parms(struct raid_set *rs, char **argv)
{
	int i;
	int rebuild = 0;
	int metadata_available = 0;
	int ret = 0;

	for (i = 0; i < rs->md.raid_disks; i++, argv += 2) {
		rs->dev[i].rdev.raid_disk = i;

		rs->dev[i].meta_dev = NULL;
		rs->dev[i].data_dev = NULL;

		/*
		 * There are no offsets, since there is a separate device
		 * for data and metadata.
		 */
		rs->dev[i].rdev.data_offset = 0;
		rs->dev[i].rdev.mddev = &rs->md;

		if (strcmp(argv[0], "-")) {
242
243
244
245
246
247
248
249
250
251
			ret = dm_get_device(rs->ti, argv[0],
					    dm_table_get_mode(rs->ti->table),
					    &rs->dev[i].meta_dev);
			rs->ti->error = "RAID metadata device lookup failure";
			if (ret)
				return ret;

			rs->dev[i].rdev.sb_page = alloc_page(GFP_KERNEL);
			if (!rs->dev[i].rdev.sb_page)
				return -ENOMEM;
NeilBrown's avatar
NeilBrown committed
252
253
254
255
256
257
258
259
260
		}

		if (!strcmp(argv[1], "-")) {
			if (!test_bit(In_sync, &rs->dev[i].rdev.flags) &&
			    (!rs->dev[i].rdev.recovery_offset)) {
				rs->ti->error = "Drive designated for rebuild not specified";
				return -EINVAL;
			}

261
262
263
264
			rs->ti->error = "No data device supplied with metadata device";
			if (rs->dev[i].meta_dev)
				return -EINVAL;

NeilBrown's avatar
NeilBrown committed
265
266
267
268
269
270
271
272
273
274
275
			continue;
		}

		ret = dm_get_device(rs->ti, argv[1],
				    dm_table_get_mode(rs->ti->table),
				    &rs->dev[i].data_dev);
		if (ret) {
			rs->ti->error = "RAID device lookup failure";
			return ret;
		}

276
277
278
279
		if (rs->dev[i].meta_dev) {
			metadata_available = 1;
			rs->dev[i].rdev.meta_bdev = rs->dev[i].meta_dev->bdev;
		}
NeilBrown's avatar
NeilBrown committed
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
		rs->dev[i].rdev.bdev = rs->dev[i].data_dev->bdev;
		list_add(&rs->dev[i].rdev.same_set, &rs->md.disks);
		if (!test_bit(In_sync, &rs->dev[i].rdev.flags))
			rebuild++;
	}

	if (metadata_available) {
		rs->md.external = 0;
		rs->md.persistent = 1;
		rs->md.major_version = 2;
	} else if (rebuild && !rs->md.recovery_cp) {
		/*
		 * Without metadata, we will not be able to tell if the array
		 * is in-sync or not - we must assume it is not.  Therefore,
		 * it is impossible to rebuild a drive.
		 *
		 * Even if there is metadata, the on-disk information may
		 * indicate that the array is not in-sync and it will then
		 * fail at that time.
		 *
		 * User could specify 'nosync' option if desperate.
		 */
		DMERR("Unable to rebuild drive while array is not in-sync");
		rs->ti->error = "RAID device lookup failure";
		return -EINVAL;
	}

	return 0;
}

310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
/*
 * validate_region_size
 * @rs
 * @region_size:  region size in sectors.  If 0, pick a size (4MiB default).
 *
 * Set rs->md.bitmap_info.chunksize (which really refers to 'region size').
 * Ensure that (ti->len/region_size < 2^21) - required by MD bitmap.
 *
 * Returns: 0 on success, -EINVAL on failure.
 */
static int validate_region_size(struct raid_set *rs, unsigned long region_size)
{
	unsigned long min_region_size = rs->ti->len / (1 << 21);

	if (!region_size) {
		/*
		 * Choose a reasonable default.  All figures in sectors.
		 */
		if (min_region_size > (1 << 13)) {
329
330
331
			/* If not a power of 2, make it the next power of 2 */
			if (min_region_size & (min_region_size - 1))
				region_size = 1 << fls(region_size);
332
333
334
335
336
337
338
339
340
341
342
343
344
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
			DMINFO("Choosing default region size of %lu sectors",
			       region_size);
		} else {
			DMINFO("Choosing default region size of 4MiB");
			region_size = 1 << 13; /* sectors */
		}
	} else {
		/*
		 * Validate user-supplied value.
		 */
		if (region_size > rs->ti->len) {
			rs->ti->error = "Supplied region size is too large";
			return -EINVAL;
		}

		if (region_size < min_region_size) {
			DMERR("Supplied region_size (%lu sectors) below minimum (%lu)",
			      region_size, min_region_size);
			rs->ti->error = "Supplied region size is too small";
			return -EINVAL;
		}

		if (!is_power_of_2(region_size)) {
			rs->ti->error = "Region size is not a power of 2";
			return -EINVAL;
		}

		if (region_size < rs->md.chunk_sectors) {
			rs->ti->error = "Region size is smaller than the chunk size";
			return -EINVAL;
		}
	}

	/*
	 * Convert sectors to bytes.
	 */
	rs->md.bitmap_info.chunksize = (region_size << 9);

	return 0;
}

373
/*
374
 * validate_raid_redundancy
375
376
 * @rs
 *
377
378
 * Determine if there are enough devices in the array that haven't
 * failed (or are being rebuilt) to form a usable array.
379
380
381
 *
 * Returns: 0 on success, -EINVAL on failure.
 */
382
static int validate_raid_redundancy(struct raid_set *rs)
383
384
{
	unsigned i, rebuild_cnt = 0;
385
	unsigned rebuilds_per_group = 0, copies, d;
386
	unsigned group_size, last_group_start;
387
388

	for (i = 0; i < rs->md.raid_disks; i++)
389
390
		if (!test_bit(In_sync, &rs->dev[i].rdev.flags) ||
		    !rs->dev[i].rdev.sb_page)
391
392
393
394
395
396
397
398
399
400
401
402
403
404
			rebuild_cnt++;

	switch (rs->raid_type->level) {
	case 1:
		if (rebuild_cnt >= rs->md.raid_disks)
			goto too_many;
		break;
	case 4:
	case 5:
	case 6:
		if (rebuild_cnt > rs->raid_type->parity_devs)
			goto too_many;
		break;
	case 10:
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
		copies = raid10_md_layout_to_copies(rs->md.layout);
		if (rebuild_cnt < copies)
			break;

		/*
		 * It is possible to have a higher rebuild count for RAID10,
		 * as long as the failed devices occur in different mirror
		 * groups (i.e. different stripes).
		 *
		 * When checking "near" format, make sure no adjacent devices
		 * have failed beyond what can be handled.  In addition to the
		 * simple case where the number of devices is a multiple of the
		 * number of copies, we must also handle cases where the number
		 * of devices is not a multiple of the number of copies.
		 * E.g.    dev1 dev2 dev3 dev4 dev5
		 *          A    A    B    B    C
		 *          C    D    D    E    E
		 */
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
		if (!strcmp("near", raid10_md_layout_to_format(rs->md.layout))) {
			for (i = 0; i < rs->md.raid_disks * copies; i++) {
				if (!(i % copies))
					rebuilds_per_group = 0;
				d = i % rs->md.raid_disks;
				if ((!rs->dev[d].rdev.sb_page ||
				     !test_bit(In_sync, &rs->dev[d].rdev.flags)) &&
				    (++rebuilds_per_group >= copies))
					goto too_many;
			}
			break;
		}

		/*
		 * When checking "far" and "offset" formats, we need to ensure
		 * that the device that holds its copy is not also dead or
		 * being rebuilt.  (Note that "far" and "offset" formats only
		 * support two copies right now.  These formats also only ever
		 * use the 'use_far_sets' variant.)
		 *
		 * This check is somewhat complicated by the need to account
		 * for arrays that are not a multiple of (far) copies.  This
		 * results in the need to treat the last (potentially larger)
		 * set differently.
		 */
		group_size = (rs->md.raid_disks / copies);
		last_group_start = (rs->md.raid_disks / group_size) - 1;
		last_group_start *= group_size;
		for (i = 0; i < rs->md.raid_disks; i++) {
			if (!(i % copies) && !(i > last_group_start))
453
				rebuilds_per_group = 0;
454
455
			if ((!rs->dev[i].rdev.sb_page ||
			     !test_bit(In_sync, &rs->dev[i].rdev.flags)) &&
456
			    (++rebuilds_per_group >= copies))
457
					goto too_many;
458
459
		}
		break;
460
	default:
461
462
		if (rebuild_cnt)
			return -EINVAL;
463
464
465
466
467
468
469
470
	}

	return 0;

too_many:
	return -EINVAL;
}

NeilBrown's avatar
NeilBrown committed
471
472
473
474
/*
 * Possible arguments are...
 *	<chunk_size> [optional_args]
 *
475
476
477
478
479
 * Argument definitions
 *    <chunk_size>			The number of sectors per disk that
 *                                      will form the "stripe"
 *    [[no]sync]			Force or prevent recovery of the
 *                                      entire array
480
481
 *    [devices_handle_discard_safely]	Allow discards on RAID4/5/6; useful if RAID
 *					member device(s) properly support TRIM/UNMAP
NeilBrown's avatar
NeilBrown committed
482
 *    [rebuild <idx>]			Rebuild the drive indicated by the index
483
484
 *    [daemon_sleep <ms>]		Time between bitmap daemon work to
 *                                      clear bits
NeilBrown's avatar
NeilBrown committed
485
486
 *    [min_recovery_rate <kB/sec/disk>]	Throttle RAID initialization
 *    [max_recovery_rate <kB/sec/disk>]	Throttle RAID initialization
487
 *    [write_mostly <idx>]		Indicate a write mostly drive via index
NeilBrown's avatar
NeilBrown committed
488
489
 *    [max_write_behind <sectors>]	See '-write-behind=' (man mdadm)
 *    [stripe_cache <sectors>]		Stripe cache size for higher RAIDs
490
 *    [region_size <sectors>]           Defines granularity of bitmap
491
492
493
 *
 * RAID10-only options:
 *    [raid10_copies <# copies>]        Number of copies.  (Default: 2)
494
 *    [raid10_format <near|far|offset>] Layout algorithm.  (Default: near)
NeilBrown's avatar
NeilBrown committed
495
496
497
498
 */
static int parse_raid_params(struct raid_set *rs, char **argv,
			     unsigned num_raid_params)
{
499
500
	char *raid10_format = "near";
	unsigned raid10_copies = 2;
501
	unsigned i;
502
	unsigned long value, region_size = 0;
503
	sector_t sectors_per_dev = rs->ti->len;
504
	sector_t max_io_len;
NeilBrown's avatar
NeilBrown committed
505
506
507
508
	char *key;

	/*
	 * First, parse the in-order required arguments
509
	 * "chunk_size" is the only argument of this type.
NeilBrown's avatar
NeilBrown committed
510
	 */
511
	if ((kstrtoul(argv[0], 10, &value) < 0)) {
NeilBrown's avatar
NeilBrown committed
512
513
		rs->ti->error = "Bad chunk size";
		return -EINVAL;
514
515
516
517
518
519
520
521
522
523
	} else if (rs->raid_type->level == 1) {
		if (value)
			DMERR("Ignoring chunk size parameter for RAID 1");
		value = 0;
	} else if (!is_power_of_2(value)) {
		rs->ti->error = "Chunk size must be a power of 2";
		return -EINVAL;
	} else if (value < 8) {
		rs->ti->error = "Chunk size value is too small";
		return -EINVAL;
NeilBrown's avatar
NeilBrown committed
524
525
526
527
528
529
530
	}

	rs->md.new_chunk_sectors = rs->md.chunk_sectors = value;
	argv++;
	num_raid_params--;

	/*
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
	 * We set each individual device as In_sync with a completed
	 * 'recovery_offset'.  If there has been a device failure or
	 * replacement then one of the following cases applies:
	 *
	 *   1) User specifies 'rebuild'.
	 *      - Device is reset when param is read.
	 *   2) A new device is supplied.
	 *      - No matching superblock found, resets device.
	 *   3) Device failure was transient and returns on reload.
	 *      - Failure noticed, resets device for bitmap replay.
	 *   4) Device hadn't completed recovery after previous failure.
	 *      - Superblock is read and overrides recovery_offset.
	 *
	 * What is found in the superblocks of the devices is always
	 * authoritative, unless 'rebuild' or '[no]sync' was specified.
NeilBrown's avatar
NeilBrown committed
546
	 */
547
	for (i = 0; i < rs->md.raid_disks; i++) {
NeilBrown's avatar
NeilBrown committed
548
		set_bit(In_sync, &rs->dev[i].rdev.flags);
549
550
		rs->dev[i].rdev.recovery_offset = MaxSector;
	}
NeilBrown's avatar
NeilBrown committed
551

552
553
554
	/*
	 * Second, parse the unordered optional arguments
	 */
NeilBrown's avatar
NeilBrown committed
555
	for (i = 0; i < num_raid_params; i++) {
556
		if (!strcasecmp(argv[i], "nosync")) {
NeilBrown's avatar
NeilBrown committed
557
558
559
560
			rs->md.recovery_cp = MaxSector;
			rs->print_flags |= DMPF_NOSYNC;
			continue;
		}
561
		if (!strcasecmp(argv[i], "sync")) {
NeilBrown's avatar
NeilBrown committed
562
563
564
565
566
567
568
569
570
571
572
573
			rs->md.recovery_cp = 0;
			rs->print_flags |= DMPF_SYNC;
			continue;
		}

		/* The rest of the optional arguments come in key/value pairs */
		if ((i + 1) >= num_raid_params) {
			rs->ti->error = "Wrong number of raid parameters given";
			return -EINVAL;
		}

		key = argv[i++];
574
575
576
577
578
579
580

		/* Parameters that take a string value are checked here. */
		if (!strcasecmp(key, "raid10_format")) {
			if (rs->raid_type->level != 10) {
				rs->ti->error = "'raid10_format' is an invalid parameter for this RAID type";
				return -EINVAL;
			}
581
582
583
			if (strcmp("near", argv[i]) &&
			    strcmp("far", argv[i]) &&
			    strcmp("offset", argv[i])) {
584
585
586
587
588
589
590
591
				rs->ti->error = "Invalid 'raid10_format' value given";
				return -EINVAL;
			}
			raid10_format = argv[i];
			rs->print_flags |= DMPF_RAID10_FORMAT;
			continue;
		}

592
		if (kstrtoul(argv[i], 10, &value) < 0) {
NeilBrown's avatar
NeilBrown committed
593
594
595
596
			rs->ti->error = "Bad numerical argument given in raid params";
			return -EINVAL;
		}

597
		/* Parameters that take a numeric value are checked here */
598
		if (!strcasecmp(key, "rebuild")) {
599
			if (value >= rs->md.raid_disks) {
NeilBrown's avatar
NeilBrown committed
600
601
602
603
604
				rs->ti->error = "Invalid rebuild index given";
				return -EINVAL;
			}
			clear_bit(In_sync, &rs->dev[value].rdev.flags);
			rs->dev[value].rdev.recovery_offset = 0;
605
			rs->print_flags |= DMPF_REBUILD;
606
607
608
609
610
		} else if (!strcasecmp(key, "write_mostly")) {
			if (rs->raid_type->level != 1) {
				rs->ti->error = "write_mostly option is only valid for RAID1";
				return -EINVAL;
			}
611
			if (value >= rs->md.raid_disks) {
612
613
614
615
				rs->ti->error = "Invalid write_mostly drive index given";
				return -EINVAL;
			}
			set_bit(WriteMostly, &rs->dev[value].rdev.flags);
616
		} else if (!strcasecmp(key, "max_write_behind")) {
617
618
619
620
			if (rs->raid_type->level != 1) {
				rs->ti->error = "max_write_behind option is only valid for RAID1";
				return -EINVAL;
			}
NeilBrown's avatar
NeilBrown committed
621
622
623
624
625
626
627
628
629
630
631
632
			rs->print_flags |= DMPF_MAX_WRITE_BEHIND;

			/*
			 * In device-mapper, we specify things in sectors, but
			 * MD records this value in kB
			 */
			value /= 2;
			if (value > COUNTER_MAX) {
				rs->ti->error = "Max write-behind limit out of range";
				return -EINVAL;
			}
			rs->md.bitmap_info.max_write_behind = value;
633
		} else if (!strcasecmp(key, "daemon_sleep")) {
NeilBrown's avatar
NeilBrown committed
634
635
636
637
638
639
			rs->print_flags |= DMPF_DAEMON_SLEEP;
			if (!value || (value > MAX_SCHEDULE_TIMEOUT)) {
				rs->ti->error = "daemon sleep period out of range";
				return -EINVAL;
			}
			rs->md.bitmap_info.daemon_sleep = value;
640
		} else if (!strcasecmp(key, "stripe_cache")) {
NeilBrown's avatar
NeilBrown committed
641
642
643
644
645
646
647
648
			rs->print_flags |= DMPF_STRIPE_CACHE;

			/*
			 * In device-mapper, we specify things in sectors, but
			 * MD records this value in kB
			 */
			value /= 2;

649
650
			if ((rs->raid_type->level != 5) &&
			    (rs->raid_type->level != 6)) {
NeilBrown's avatar
NeilBrown committed
651
652
653
654
655
656
657
				rs->ti->error = "Inappropriate argument: stripe_cache";
				return -EINVAL;
			}
			if (raid5_set_cache_size(&rs->md, (int)value)) {
				rs->ti->error = "Bad stripe_cache size";
				return -EINVAL;
			}
658
		} else if (!strcasecmp(key, "min_recovery_rate")) {
NeilBrown's avatar
NeilBrown committed
659
660
661
662
663
664
			rs->print_flags |= DMPF_MIN_RECOVERY_RATE;
			if (value > INT_MAX) {
				rs->ti->error = "min_recovery_rate out of range";
				return -EINVAL;
			}
			rs->md.sync_speed_min = (int)value;
665
		} else if (!strcasecmp(key, "max_recovery_rate")) {
NeilBrown's avatar
NeilBrown committed
666
667
668
669
670
671
			rs->print_flags |= DMPF_MAX_RECOVERY_RATE;
			if (value > INT_MAX) {
				rs->ti->error = "max_recovery_rate out of range";
				return -EINVAL;
			}
			rs->md.sync_speed_max = (int)value;
672
673
674
		} else if (!strcasecmp(key, "region_size")) {
			rs->print_flags |= DMPF_REGION_SIZE;
			region_size = value;
675
676
677
678
679
680
681
682
		} else if (!strcasecmp(key, "raid10_copies") &&
			   (rs->raid_type->level == 10)) {
			if ((value < 2) || (value > 0xFF)) {
				rs->ti->error = "Bad value for 'raid10_copies'";
				return -EINVAL;
			}
			rs->print_flags |= DMPF_RAID10_COPIES;
			raid10_copies = value;
NeilBrown's avatar
NeilBrown committed
683
684
685
686
687
688
689
		} else {
			DMERR("Unable to parse RAID parameter: %s", key);
			rs->ti->error = "Unable to parse RAID parameters";
			return -EINVAL;
		}
	}

690
691
692
693
	if (validate_region_size(rs, region_size))
		return -EINVAL;

	if (rs->md.chunk_sectors)
694
		max_io_len = rs->md.chunk_sectors;
695
	else
696
		max_io_len = region_size;
697

698
699
	if (dm_set_target_max_io_len(rs->ti, max_io_len))
		return -EINVAL;
700

701
702
703
704
705
706
	if (rs->raid_type->level == 10) {
		if (raid10_copies > rs->md.raid_disks) {
			rs->ti->error = "Not enough devices to satisfy specification";
			return -EINVAL;
		}

707
708
709
710
711
712
713
714
715
		/*
		 * If the format is not "near", we only support
		 * two copies at the moment.
		 */
		if (strcmp("near", raid10_format) && (raid10_copies > 2)) {
			rs->ti->error = "Too many copies for given RAID10 format.";
			return -EINVAL;
		}

716
717
718
719
720
721
722
723
724
725
		/* (Len * #mirrors) / #devices */
		sectors_per_dev = rs->ti->len * raid10_copies;
		sector_div(sectors_per_dev, rs->md.raid_disks);

		rs->md.layout = raid10_format_to_md_layout(raid10_format,
							   raid10_copies);
		rs->md.new_layout = rs->md.layout;
	} else if ((rs->raid_type->level > 1) &&
		   sector_div(sectors_per_dev,
			      (rs->md.raid_disks - rs->raid_type->parity_devs))) {
726
727
728
729
730
		rs->ti->error = "Target length not divisible by number of data devices";
		return -EINVAL;
	}
	rs->md.dev_sectors = sectors_per_dev;

NeilBrown's avatar
NeilBrown committed
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
	/* Assume there are no metadata devices until the drives are parsed */
	rs->md.persistent = 0;
	rs->md.external = 1;

	return 0;
}

static void do_table_event(struct work_struct *ws)
{
	struct raid_set *rs = container_of(ws, struct raid_set, md.event_work);

	dm_table_event(rs->ti->table);
}

static int raid_is_congested(struct dm_target_callbacks *cb, int bits)
{
	struct raid_set *rs = container_of(cb, struct raid_set, callbacks);

749
	return mddev_congested(&rs->md, bits);
NeilBrown's avatar
NeilBrown committed
750
751
}

752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
/*
 * This structure is never routinely used by userspace, unlike md superblocks.
 * Devices with this superblock should only ever be accessed via device-mapper.
 */
#define DM_RAID_MAGIC 0x64526D44
struct dm_raid_superblock {
	__le32 magic;		/* "DmRd" */
	__le32 features;	/* Used to indicate possible future changes */

	__le32 num_devices;	/* Number of devices in this array. (Max 64) */
	__le32 array_position;	/* The position of this drive in the array */

	__le64 events;		/* Incremented by md when superblock updated */
	__le64 failed_devices;	/* Bit field of devices to indicate failures */

	/*
	 * This offset tracks the progress of the repair or replacement of
	 * an individual drive.
	 */
	__le64 disk_recovery_offset;

	/*
	 * This offset tracks the progress of the initial array
	 * synchronisation/parity calculation.
	 */
	__le64 array_resync_offset;

	/*
	 * RAID characteristics
	 */
	__le32 level;
	__le32 layout;
	__le32 stripe_sectors;

786
	/* Remainder of a logical block is zero-filled when writing (see super_sync()). */
787
788
} __packed;

789
static int read_disk_sb(struct md_rdev *rdev, int size)
790
791
792
793
794
795
796
{
	BUG_ON(!rdev->sb_page);

	if (rdev->sb_loaded)
		return 0;

	if (!sync_page_io(rdev, 0, size, rdev->sb_page, READ, 1)) {
797
798
		DMERR("Failed to read superblock of device at position %d",
		      rdev->raid_disk);
799
		md_error(rdev->mddev, rdev);
800
801
802
803
804
805
806
807
		return -EINVAL;
	}

	rdev->sb_loaded = 1;

	return 0;
}

808
static void super_sync(struct mddev *mddev, struct md_rdev *rdev)
809
{
810
	int i;
811
812
	uint64_t failed_devices;
	struct dm_raid_superblock *sb;
813
	struct raid_set *rs = container_of(mddev, struct raid_set, md);
814
815
816
817

	sb = page_address(rdev->sb_page);
	failed_devices = le64_to_cpu(sb->failed_devices);

818
819
820
821
	for (i = 0; i < mddev->raid_disks; i++)
		if (!rs->dev[i].data_dev ||
		    test_bit(Faulty, &(rs->dev[i].rdev.flags)))
			failed_devices |= (1ULL << i);
822

823
	memset(sb + 1, 0, rdev->sb_size - sizeof(*sb));
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

	sb->magic = cpu_to_le32(DM_RAID_MAGIC);
	sb->features = cpu_to_le32(0);	/* No features yet */

	sb->num_devices = cpu_to_le32(mddev->raid_disks);
	sb->array_position = cpu_to_le32(rdev->raid_disk);

	sb->events = cpu_to_le64(mddev->events);
	sb->failed_devices = cpu_to_le64(failed_devices);

	sb->disk_recovery_offset = cpu_to_le64(rdev->recovery_offset);
	sb->array_resync_offset = cpu_to_le64(mddev->recovery_cp);

	sb->level = cpu_to_le32(mddev->level);
	sb->layout = cpu_to_le32(mddev->layout);
	sb->stripe_sectors = cpu_to_le32(mddev->chunk_sectors);
}

/*
 * super_load
 *
 * This function creates a superblock if one is not found on the device
 * and will decide which superblock to use if there's a choice.
 *
 * Return: 1 if use rdev, 0 if use refdev, -Exxx otherwise
 */
850
static int super_load(struct md_rdev *rdev, struct md_rdev *refdev)
851
852
853
854
855
856
857
{
	int ret;
	struct dm_raid_superblock *sb;
	struct dm_raid_superblock *refsb;
	uint64_t events_sb, events_refsb;

	rdev->sb_start = 0;
858
859
860
861
862
	rdev->sb_size = bdev_logical_block_size(rdev->meta_bdev);
	if (rdev->sb_size < sizeof(*sb) || rdev->sb_size > PAGE_SIZE) {
		DMERR("superblock size of a logical block is no longer valid");
		return -EINVAL;
	}
863
864
865
866
867
868

	ret = read_disk_sb(rdev, rdev->sb_size);
	if (ret)
		return ret;

	sb = page_address(rdev->sb_page);
869
870
871
872
873
874
875
876

	/*
	 * Two cases that we want to write new superblocks and rebuild:
	 * 1) New device (no matching magic number)
	 * 2) Device specified for rebuild (!In_sync w/ offset == 0)
	 */
	if ((sb->magic != cpu_to_le32(DM_RAID_MAGIC)) ||
	    (!test_bit(In_sync, &rdev->flags) && !rdev->recovery_offset)) {
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
		super_sync(rdev->mddev, rdev);

		set_bit(FirstUse, &rdev->flags);

		/* Force writing of superblocks to disk */
		set_bit(MD_CHANGE_DEVS, &rdev->mddev->flags);

		/* Any superblock is better than none, choose that if given */
		return refdev ? 0 : 1;
	}

	if (!refdev)
		return 1;

	events_sb = le64_to_cpu(sb->events);

	refsb = page_address(refdev->sb_page);
	events_refsb = le64_to_cpu(refsb->events);

	return (events_sb > events_refsb) ? 1 : 0;
}

899
static int super_init_validation(struct mddev *mddev, struct md_rdev *rdev)
900
901
902
903
904
905
906
907
{
	int role;
	struct raid_set *rs = container_of(mddev, struct raid_set, md);
	uint64_t events_sb;
	uint64_t failed_devices;
	struct dm_raid_superblock *sb;
	uint32_t new_devs = 0;
	uint32_t rebuilds = 0;
NeilBrown's avatar
NeilBrown committed
908
	struct md_rdev *r;
909
910
911
912
913
914
915
916
917
918
919
920
921
922
	struct dm_raid_superblock *sb2;

	sb = page_address(rdev->sb_page);
	events_sb = le64_to_cpu(sb->events);
	failed_devices = le64_to_cpu(sb->failed_devices);

	/*
	 * Initialise to 1 if this is a new superblock.
	 */
	mddev->events = events_sb ? : 1;

	/*
	 * Reshaping is not currently allowed
	 */
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
	if (le32_to_cpu(sb->level) != mddev->level) {
		DMERR("Reshaping arrays not yet supported. (RAID level change)");
		return -EINVAL;
	}
	if (le32_to_cpu(sb->layout) != mddev->layout) {
		DMERR("Reshaping arrays not yet supported. (RAID layout change)");
		DMERR("  0x%X vs 0x%X", le32_to_cpu(sb->layout), mddev->layout);
		DMERR("  Old layout: %s w/ %d copies",
		      raid10_md_layout_to_format(le32_to_cpu(sb->layout)),
		      raid10_md_layout_to_copies(le32_to_cpu(sb->layout)));
		DMERR("  New layout: %s w/ %d copies",
		      raid10_md_layout_to_format(mddev->layout),
		      raid10_md_layout_to_copies(mddev->layout));
		return -EINVAL;
	}
	if (le32_to_cpu(sb->stripe_sectors) != mddev->chunk_sectors) {
		DMERR("Reshaping arrays not yet supported. (stripe sectors change)");
940
941
942
943
944
945
		return -EINVAL;
	}

	/* We can only change the number of devices in RAID1 right now */
	if ((rs->raid_type->level != 1) &&
	    (le32_to_cpu(sb->num_devices) != mddev->raid_disks)) {
946
		DMERR("Reshaping arrays not yet supported. (device count change)");
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
		return -EINVAL;
	}

	if (!(rs->print_flags & (DMPF_SYNC | DMPF_NOSYNC)))
		mddev->recovery_cp = le64_to_cpu(sb->array_resync_offset);

	/*
	 * During load, we set FirstUse if a new superblock was written.
	 * There are two reasons we might not have a superblock:
	 * 1) The array is brand new - in which case, all of the
	 *    devices must have their In_sync bit set.  Also,
	 *    recovery_cp must be 0, unless forced.
	 * 2) This is a new device being added to an old array
	 *    and the new device needs to be rebuilt - in which
	 *    case the In_sync bit will /not/ be set and
	 *    recovery_cp must be MaxSector.
	 */
NeilBrown's avatar
NeilBrown committed
964
	rdev_for_each(r, mddev) {
965
		if (!test_bit(In_sync, &r->flags)) {
966
967
			DMINFO("Device %d specified for rebuild: "
			       "Clearing superblock", r->raid_disk);
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
			rebuilds++;
		} else if (test_bit(FirstUse, &r->flags))
			new_devs++;
	}

	if (!rebuilds) {
		if (new_devs == mddev->raid_disks) {
			DMINFO("Superblocks created for new array");
			set_bit(MD_ARRAY_FIRST_USE, &mddev->flags);
		} else if (new_devs) {
			DMERR("New device injected "
			      "into existing array without 'rebuild' "
			      "parameter specified");
			return -EINVAL;
		}
	} else if (new_devs) {
		DMERR("'rebuild' devices cannot be "
		      "injected into an array with other first-time devices");
		return -EINVAL;
	} else if (mddev->recovery_cp != MaxSector) {
		DMERR("'rebuild' specified while array is not in-sync");
		return -EINVAL;
	}

	/*
	 * Now we set the Faulty bit for those devices that are
	 * recorded in the superblock as failed.
	 */
NeilBrown's avatar
NeilBrown committed
996
	rdev_for_each(r, mddev) {
997
998
999
1000
		if (!r->sb_page)
			continue;
		sb2 = page_address(r->sb_page);
		sb2->failed_devices = 0;