raid0.c 19.6 KB
Newer Older
L
Linus Torvalds 已提交
1 2
/*
   raid0.c : Multiple Devices driver for Linux
3
	     Copyright (C) 1994-96 Marc ZYNGIER
L
Linus Torvalds 已提交
4 5
	     <zyngier@ufr-info-p7.ibp.fr> or
	     <maz@gloups.fdn.fr>
6
	     Copyright (C) 1999, 2000 Ingo Molnar, Red Hat
L
Linus Torvalds 已提交
7 8 9 10 11 12 13

   RAID-0 management functions.

   This program is free software; you can redistribute it and/or modify
   it under the terms of the GNU General Public License as published by
   the Free Software Foundation; either version 2, or (at your option)
   any later version.
14

L
Linus Torvalds 已提交
15 16
   You should have received a copy of the GNU General Public License
   (for example /usr/src/linux/COPYING); if not, write to the Free
17
   Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
L
Linus Torvalds 已提交
18 19
*/

20 21
#include <linux/blkdev.h>
#include <linux/seq_file.h>
22
#include <linux/module.h>
23
#include <linux/slab.h>
24
#include "md.h"
25
#include "raid0.h"
26
#include "raid5.h"
L
Linus Torvalds 已提交
27

28
static int raid0_congested(struct mddev *mddev, int bits)
29
{
30
	struct r0conf *conf = mddev->private;
31
	struct md_rdev **devlist = conf->devlist;
32
	int raid_disks = conf->strip_zone[0].nb_dev;
33 34
	int i, ret = 0;

35
	for (i = 0; i < raid_disks && !ret ; i++) {
36
		struct request_queue *q = bdev_get_queue(devlist[i]->bdev);
37 38 39 40 41 42

		ret |= bdi_congested(&q->backing_dev_info, bits);
	}
	return ret;
}

43 44 45
/*
 * inform the user of the raid configuration
*/
46
static void dump_zones(struct mddev *mddev)
47
{
48
	int j, k;
49 50 51
	sector_t zone_size = 0;
	sector_t zone_start = 0;
	char b[BDEVNAME_SIZE];
52
	struct r0conf *conf = mddev->private;
53
	int raid_disks = conf->strip_zone[0].nb_dev;
54 55 56
	printk(KERN_INFO "md: RAID0 configuration for %s - %d zone%s\n",
	       mdname(mddev),
	       conf->nr_strip_zones, conf->nr_strip_zones==1?"":"s");
57
	for (j = 0; j < conf->nr_strip_zones; j++) {
58
		printk(KERN_INFO "md: zone%d=[", j);
59
		for (k = 0; k < conf->strip_zone[j].nb_dev; k++)
60
			printk(KERN_CONT "%s%s", k?"/":"",
61
			bdevname(conf->devlist[j*raid_disks
62
						+ k]->bdev, b));
N
NeilBrown 已提交
63
		printk(KERN_CONT "]\n");
64 65

		zone_size  = conf->strip_zone[j].zone_end - zone_start;
66 67
		printk(KERN_INFO "      zone-offset=%10lluKB, "
				"device-offset=%10lluKB, size=%10lluKB\n",
68 69 70 71 72
			(unsigned long long)zone_start>>1,
			(unsigned long long)conf->strip_zone[j].dev_start>>1,
			(unsigned long long)zone_size>>1);
		zone_start = conf->strip_zone[j].zone_end;
	}
73
	printk(KERN_INFO "\n");
74 75
}

76
static int create_strip_zones(struct mddev *mddev, struct r0conf **private_conf)
L
Linus Torvalds 已提交
77
{
78
	int i, c, err;
79
	sector_t curr_zone_end, sectors;
80
	struct md_rdev *smallest, *rdev1, *rdev2, *rdev, **dev;
L
Linus Torvalds 已提交
81 82 83
	struct strip_zone *zone;
	int cnt;
	char b[BDEVNAME_SIZE];
84
	char b2[BDEVNAME_SIZE];
85
	struct r0conf *conf = kzalloc(sizeof(*conf), GFP_KERNEL);
S
Shaohua Li 已提交
86
	bool discard_supported = false;
87 88 89

	if (!conf)
		return -ENOMEM;
N
NeilBrown 已提交
90
	rdev_for_each(rdev1, mddev) {
91 92 93
		pr_debug("md/raid0:%s: looking at %s\n",
			 mdname(mddev),
			 bdevname(rdev1->bdev, b));
L
Linus Torvalds 已提交
94
		c = 0;
95 96 97 98 99 100

		/* round size to chunk_size */
		sectors = rdev1->sectors;
		sector_div(sectors, mddev->chunk_sectors);
		rdev1->sectors = sectors * mddev->chunk_sectors;

N
NeilBrown 已提交
101
		rdev_for_each(rdev2, mddev) {
102 103 104 105 106 107 108
			pr_debug("md/raid0:%s:   comparing %s(%llu)"
				 " with %s(%llu)\n",
				 mdname(mddev),
				 bdevname(rdev1->bdev,b),
				 (unsigned long long)rdev1->sectors,
				 bdevname(rdev2->bdev,b2),
				 (unsigned long long)rdev2->sectors);
L
Linus Torvalds 已提交
109
			if (rdev2 == rdev1) {
110 111
				pr_debug("md/raid0:%s:   END\n",
					 mdname(mddev));
L
Linus Torvalds 已提交
112 113
				break;
			}
114
			if (rdev2->sectors == rdev1->sectors) {
L
Linus Torvalds 已提交
115 116 117 118
				/*
				 * Not unique, don't count it as a new
				 * group
				 */
119 120
				pr_debug("md/raid0:%s:   EQUAL\n",
					 mdname(mddev));
L
Linus Torvalds 已提交
121 122 123
				c = 1;
				break;
			}
124 125
			pr_debug("md/raid0:%s:   NOT EQUAL\n",
				 mdname(mddev));
L
Linus Torvalds 已提交
126 127
		}
		if (!c) {
128 129
			pr_debug("md/raid0:%s:   ==> UNIQUE\n",
				 mdname(mddev));
L
Linus Torvalds 已提交
130
			conf->nr_strip_zones++;
131 132
			pr_debug("md/raid0:%s: %d zones\n",
				 mdname(mddev), conf->nr_strip_zones);
L
Linus Torvalds 已提交
133 134
		}
	}
135 136
	pr_debug("md/raid0:%s: FINAL %d zones\n",
		 mdname(mddev), conf->nr_strip_zones);
137
	err = -ENOMEM;
138
	conf->strip_zone = kzalloc(sizeof(struct strip_zone)*
L
Linus Torvalds 已提交
139 140
				conf->nr_strip_zones, GFP_KERNEL);
	if (!conf->strip_zone)
141
		goto abort;
142
	conf->devlist = kzalloc(sizeof(struct md_rdev*)*
L
Linus Torvalds 已提交
143 144 145
				conf->nr_strip_zones*mddev->raid_disks,
				GFP_KERNEL);
	if (!conf->devlist)
146
		goto abort;
L
Linus Torvalds 已提交
147 148 149 150 151 152 153

	/* The first zone must contain all devices, so here we check that
	 * there is a proper alignment of slots to devices and find them all
	 */
	zone = &conf->strip_zone[0];
	cnt = 0;
	smallest = NULL;
154
	dev = conf->devlist;
155
	err = -EINVAL;
N
NeilBrown 已提交
156
	rdev_for_each(rdev1, mddev) {
L
Linus Torvalds 已提交
157 158
		int j = rdev1->raid_disk;

159
		if (mddev->level == 10) {
160 161
			/* taking over a raid10-n2 array */
			j /= 2;
162 163
			rdev1->new_raid_disk = j;
		}
164

165 166 167 168 169 170 171 172
		if (mddev->level == 1) {
			/* taiking over a raid1 array-
			 * we have only one active disk
			 */
			j = 0;
			rdev1->new_raid_disk = j;
		}

173 174 175 176 177 178 179
		if (j < 0) {
			printk(KERN_ERR
			       "md/raid0:%s: remove inactive devices before converting to RAID0\n",
			       mdname(mddev));
			goto abort;
		}
		if (j >= mddev->raid_disks) {
N
NeilBrown 已提交
180 181
			printk(KERN_ERR "md/raid0:%s: bad disk number %d - "
			       "aborting!\n", mdname(mddev), j);
L
Linus Torvalds 已提交
182 183
			goto abort;
		}
184
		if (dev[j]) {
N
NeilBrown 已提交
185 186
			printk(KERN_ERR "md/raid0:%s: multiple devices for %d - "
			       "aborting!\n", mdname(mddev), j);
L
Linus Torvalds 已提交
187 188
			goto abort;
		}
189
		dev[j] = rdev1;
L
Linus Torvalds 已提交
190

191 192
		disk_stack_limits(mddev->gendisk, rdev1->bdev,
				  rdev1->data_offset << 9);
L
Linus Torvalds 已提交
193

194 195 196
		if (rdev1->bdev->bd_disk->queue->merge_bvec_fn)
			conf->has_merge_bvec = 1;

197
		if (!smallest || (rdev1->sectors < smallest->sectors))
L
Linus Torvalds 已提交
198 199
			smallest = rdev1;
		cnt++;
S
Shaohua Li 已提交
200 201 202

		if (blk_queue_discard(bdev_get_queue(rdev1->bdev)))
			discard_supported = true;
L
Linus Torvalds 已提交
203 204
	}
	if (cnt != mddev->raid_disks) {
N
NeilBrown 已提交
205 206
		printk(KERN_ERR "md/raid0:%s: too few disks (%d of %d) - "
		       "aborting!\n", mdname(mddev), cnt, mddev->raid_disks);
L
Linus Torvalds 已提交
207 208 209
		goto abort;
	}
	zone->nb_dev = cnt;
210
	zone->zone_end = smallest->sectors * cnt;
L
Linus Torvalds 已提交
211

212
	curr_zone_end = zone->zone_end;
L
Linus Torvalds 已提交
213 214 215 216

	/* now do the other zones */
	for (i = 1; i < conf->nr_strip_zones; i++)
	{
217 218
		int j;

L
Linus Torvalds 已提交
219
		zone = conf->strip_zone + i;
220
		dev = conf->devlist + i * mddev->raid_disks;
L
Linus Torvalds 已提交
221

222
		pr_debug("md/raid0:%s: zone %d\n", mdname(mddev), i);
223
		zone->dev_start = smallest->sectors;
L
Linus Torvalds 已提交
224 225 226 227
		smallest = NULL;
		c = 0;

		for (j=0; j<cnt; j++) {
228
			rdev = conf->devlist[j];
229
			if (rdev->sectors <= zone->dev_start) {
230 231 232
				pr_debug("md/raid0:%s: checking %s ... nope\n",
					 mdname(mddev),
					 bdevname(rdev->bdev, b));
233 234
				continue;
			}
235 236 237 238
			pr_debug("md/raid0:%s: checking %s ..."
				 " contained as device %d\n",
				 mdname(mddev),
				 bdevname(rdev->bdev, b), c);
239
			dev[c] = rdev;
240 241 242
			c++;
			if (!smallest || rdev->sectors < smallest->sectors) {
				smallest = rdev;
243 244 245
				pr_debug("md/raid0:%s:  (%llu) is smallest!.\n",
					 mdname(mddev),
					 (unsigned long long)rdev->sectors);
246
			}
L
Linus Torvalds 已提交
247 248 249
		}

		zone->nb_dev = c;
250
		sectors = (smallest->sectors - zone->dev_start) * c;
251 252 253
		pr_debug("md/raid0:%s: zone->nb_dev: %d, sectors: %llu\n",
			 mdname(mddev),
			 zone->nb_dev, (unsigned long long)sectors);
L
Linus Torvalds 已提交
254

255
		curr_zone_end += sectors;
256
		zone->zone_end = curr_zone_end;
L
Linus Torvalds 已提交
257

258 259 260
		pr_debug("md/raid0:%s: current zone start: %llu\n",
			 mdname(mddev),
			 (unsigned long long)smallest->sectors);
L
Linus Torvalds 已提交
261 262
	}

263 264 265 266
	/*
	 * now since we have the hard sector sizes, we can make sure
	 * chunk size is a multiple of that sector size
	 */
267
	if ((mddev->chunk_sectors << 9) % queue_logical_block_size(mddev->queue)) {
N
NeilBrown 已提交
268
		printk(KERN_ERR "md/raid0:%s: chunk_size of %d not valid\n",
269
		       mdname(mddev),
270
		       mddev->chunk_sectors << 9);
271 272
		goto abort;
	}
273 274 275 276 277

	blk_queue_io_min(mddev->queue, mddev->chunk_sectors << 9);
	blk_queue_io_opt(mddev->queue,
			 (mddev->chunk_sectors << 9) * mddev->raid_disks);

S
Shaohua Li 已提交
278 279 280 281 282
	if (!discard_supported)
		queue_flag_clear_unlocked(QUEUE_FLAG_DISCARD, mddev->queue);
	else
		queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, mddev->queue);

283
	pr_debug("md/raid0:%s: done.\n", mdname(mddev));
284 285
	*private_conf = conf;

L
Linus Torvalds 已提交
286
	return 0;
287
abort:
288 289 290
	kfree(conf->strip_zone);
	kfree(conf->devlist);
	kfree(conf);
291
	*private_conf = ERR_PTR(err);
292
	return err;
L
Linus Torvalds 已提交
293 294
}

295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350
/* Find the zone which holds a particular offset
 * Update *sectorp to be an offset in that zone
 */
static struct strip_zone *find_zone(struct r0conf *conf,
				    sector_t *sectorp)
{
	int i;
	struct strip_zone *z = conf->strip_zone;
	sector_t sector = *sectorp;

	for (i = 0; i < conf->nr_strip_zones; i++)
		if (sector < z[i].zone_end) {
			if (i)
				*sectorp = sector - z[i-1].zone_end;
			return z + i;
		}
	BUG();
}

/*
 * remaps the bio to the target device. we separate two flows.
 * power 2 flow and a general flow for the sake of perfromance
*/
static struct md_rdev *map_sector(struct mddev *mddev, struct strip_zone *zone,
				sector_t sector, sector_t *sector_offset)
{
	unsigned int sect_in_chunk;
	sector_t chunk;
	struct r0conf *conf = mddev->private;
	int raid_disks = conf->strip_zone[0].nb_dev;
	unsigned int chunk_sects = mddev->chunk_sectors;

	if (is_power_of_2(chunk_sects)) {
		int chunksect_bits = ffz(~chunk_sects);
		/* find the sector offset inside the chunk */
		sect_in_chunk  = sector & (chunk_sects - 1);
		sector >>= chunksect_bits;
		/* chunk in zone */
		chunk = *sector_offset;
		/* quotient is the chunk in real device*/
		sector_div(chunk, zone->nb_dev << chunksect_bits);
	} else{
		sect_in_chunk = sector_div(sector, chunk_sects);
		chunk = *sector_offset;
		sector_div(chunk, chunk_sects * zone->nb_dev);
	}
	/*
	*  position the bio over the real device
	*  real sector = chunk in device + starting of zone
	*	+ the position in the chunk
	*/
	*sector_offset = (chunk * chunk_sects) + sect_in_chunk;
	return conf->devlist[(zone - conf->strip_zone)*raid_disks
			     + sector_div(sector, zone->nb_dev)];
}

L
Linus Torvalds 已提交
351
/**
352
 *	raid0_mergeable_bvec -- tell bio layer if two requests can be merged
353
 *	@mddev: the md device
354
 *	@bvm: properties of new bio
L
Linus Torvalds 已提交
355 356 357 358
 *	@biovec: the request that could be merged to it.
 *
 *	Return amount of bytes we can accept at this offset
 */
359
static int raid0_mergeable_bvec(struct mddev *mddev,
360 361
				struct bvec_merge_data *bvm,
				struct bio_vec *biovec)
L
Linus Torvalds 已提交
362
{
363
	struct r0conf *conf = mddev->private;
364
	sector_t sector = bvm->bi_sector + get_start_sect(bvm->bi_bdev);
365
	sector_t sector_offset = sector;
L
Linus Torvalds 已提交
366
	int max;
367
	unsigned int chunk_sectors = mddev->chunk_sectors;
368
	unsigned int bio_sectors = bvm->bi_size >> 9;
369 370 371
	struct strip_zone *zone;
	struct md_rdev *rdev;
	struct request_queue *subq;
L
Linus Torvalds 已提交
372

N
NeilBrown 已提交
373
	if (is_power_of_2(chunk_sectors))
374 375 376 377 378
		max =  (chunk_sectors - ((sector & (chunk_sectors-1))
						+ bio_sectors)) << 9;
	else
		max =  (chunk_sectors - (sector_div(sector, chunk_sectors)
						+ bio_sectors)) << 9;
379 380
	if (max < 0)
		max = 0; /* bio_add cannot handle a negative return */
L
Linus Torvalds 已提交
381 382
	if (max <= biovec->bv_len && bio_sectors == 0)
		return biovec->bv_len;
383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399
	if (max < biovec->bv_len)
		/* too small already, no need to check further */
		return max;
	if (!conf->has_merge_bvec)
		return max;

	/* May need to check subordinate device */
	sector = sector_offset;
	zone = find_zone(mddev->private, &sector_offset);
	rdev = map_sector(mddev, zone, sector, &sector_offset);
	subq = bdev_get_queue(rdev->bdev);
	if (subq->merge_bvec_fn) {
		bvm->bi_bdev = rdev->bdev;
		bvm->bi_sector = sector_offset + zone->dev_start +
			rdev->data_offset;
		return min(max, subq->merge_bvec_fn(subq, bvm, biovec));
	} else
L
Linus Torvalds 已提交
400 401 402
		return max;
}

403
static sector_t raid0_size(struct mddev *mddev, sector_t sectors, int raid_disks)
404 405
{
	sector_t array_sectors = 0;
406
	struct md_rdev *rdev;
407 408 409 410

	WARN_ONCE(sectors || raid_disks,
		  "%s does not support generic reshape\n", __func__);

N
NeilBrown 已提交
411
	rdev_for_each(rdev, mddev)
412 413
		array_sectors += (rdev->sectors &
				  ~(sector_t)(mddev->chunk_sectors-1));
414 415 416 417

	return array_sectors;
}

N
NeilBrown 已提交
418
static void raid0_free(struct mddev *mddev, void *priv);
419

420
static int raid0_run(struct mddev *mddev)
L
Linus Torvalds 已提交
421
{
422
	struct r0conf *conf;
423
	int ret;
L
Linus Torvalds 已提交
424

425
	if (mddev->chunk_sectors == 0) {
N
NeilBrown 已提交
426 427
		printk(KERN_ERR "md/raid0:%s: chunk size must be set.\n",
		       mdname(mddev));
428 429
		return -EINVAL;
	}
430 431
	if (md_check_no_bitmap(mddev))
		return -EINVAL;
432
	blk_queue_max_hw_sectors(mddev->queue, mddev->chunk_sectors);
433
	blk_queue_max_write_same_sectors(mddev->queue, mddev->chunk_sectors);
S
Shaohua Li 已提交
434
	blk_queue_max_discard_sectors(mddev->queue, mddev->chunk_sectors);
L
Linus Torvalds 已提交
435

436 437 438 439 440 441 442 443
	/* if private is not null, we are here after takeover */
	if (mddev->private == NULL) {
		ret = create_strip_zones(mddev, &conf);
		if (ret < 0)
			return ret;
		mddev->private = conf;
	}
	conf = mddev->private;
L
Linus Torvalds 已提交
444 445

	/* calculate array device size */
446
	md_set_array_sectors(mddev, raid0_size(mddev, 0, 0));
L
Linus Torvalds 已提交
447

N
NeilBrown 已提交
448 449 450
	printk(KERN_INFO "md/raid0:%s: md_size is %llu sectors.\n",
	       mdname(mddev),
	       (unsigned long long)mddev->array_sectors);
L
Linus Torvalds 已提交
451 452 453 454 455 456 457 458 459 460
	/* calculate the max read-ahead size.
	 * For read-ahead of large files to be effective, we need to
	 * readahead at least twice a whole stripe. i.e. number of devices
	 * multiplied by chunk size times 2.
	 * If an individual device has an ra_pages greater than the
	 * chunk size, then we will not drive that device as hard as it
	 * wants.  We consider this a configuration error: a larger
	 * chunksize should be used in that case.
	 */
	{
461 462
		int stripe = mddev->raid_disks *
			(mddev->chunk_sectors << 9) / PAGE_SIZE;
L
Linus Torvalds 已提交
463 464 465 466
		if (mddev->queue->backing_dev_info.ra_pages < 2* stripe)
			mddev->queue->backing_dev_info.ra_pages = 2* stripe;
	}

467
	dump_zones(mddev);
468 469 470 471

	ret = md_integrity_register(mddev);

	return ret;
L
Linus Torvalds 已提交
472 473
}

N
NeilBrown 已提交
474
static void raid0_free(struct mddev *mddev, void *priv)
L
Linus Torvalds 已提交
475
{
N
NeilBrown 已提交
476
	struct r0conf *conf = priv;
L
Linus Torvalds 已提交
477

478
	kfree(conf->strip_zone);
479
	kfree(conf->devlist);
480
	kfree(conf);
L
Linus Torvalds 已提交
481 482
}

483 484 485
/*
 * Is io distribute over 1 or more chunks ?
*/
486
static inline int is_io_in_chunk_boundary(struct mddev *mddev,
487 488
			unsigned int chunk_sects, struct bio *bio)
{
N
NeilBrown 已提交
489
	if (likely(is_power_of_2(chunk_sects))) {
490 491
		return chunk_sects >=
			((bio->bi_iter.bi_sector & (chunk_sects-1))
492
					+ bio_sectors(bio));
493
	} else{
494
		sector_t sector = bio->bi_iter.bi_sector;
495
		return chunk_sects >= (sector_div(sector, chunk_sects)
496
						+ bio_sectors(bio));
497 498 499
	}
}

500
static void raid0_make_request(struct mddev *mddev, struct bio *bio)
501
{
L
Linus Torvalds 已提交
502
	struct strip_zone *zone;
503
	struct md_rdev *tmp_dev;
K
Kent Overstreet 已提交
504
	struct bio *split;
L
Linus Torvalds 已提交
505

T
Tejun Heo 已提交
506 507
	if (unlikely(bio->bi_rw & REQ_FLUSH)) {
		md_flush_request(mddev, bio);
508
		return;
509 510
	}

K
Kent Overstreet 已提交
511
	do {
512
		sector_t sector = bio->bi_iter.bi_sector;
K
Kent Overstreet 已提交
513 514 515 516 517 518 519 520 521 522 523 524 525
		unsigned chunk_sects = mddev->chunk_sectors;

		unsigned sectors = chunk_sects -
			(likely(is_power_of_2(chunk_sects))
			 ? (sector & (chunk_sects-1))
			 : sector_div(sector, chunk_sects));

		if (sectors < bio_sectors(bio)) {
			split = bio_split(bio, sectors, GFP_NOIO, fs_bio_set);
			bio_chain(split, bio);
		} else {
			split = bio;
		}
L
Linus Torvalds 已提交
526

K
Kent Overstreet 已提交
527 528 529 530 531 532 533 534 535 536 537 538 539
		zone = find_zone(mddev->private, &sector);
		tmp_dev = map_sector(mddev, zone, sector, &sector);
		split->bi_bdev = tmp_dev->bdev;
		split->bi_iter.bi_sector = sector + zone->dev_start +
			tmp_dev->data_offset;

		if (unlikely((split->bi_rw & REQ_DISCARD) &&
			 !blk_queue_discard(bdev_get_queue(split->bi_bdev)))) {
			/* Just ignore it */
			bio_endio(split, 0);
		} else
			generic_make_request(split);
	} while (split != bio);
L
Linus Torvalds 已提交
540
}
N
NeilBrown 已提交
541

542
static void raid0_status(struct seq_file *seq, struct mddev *mddev)
L
Linus Torvalds 已提交
543
{
544
	seq_printf(seq, " %dk chunks", mddev->chunk_sectors / 2);
L
Linus Torvalds 已提交
545 546 547
	return;
}

548
static void *raid0_takeover_raid45(struct mddev *mddev)
549
{
550
	struct md_rdev *rdev;
551
	struct r0conf *priv_conf;
552 553

	if (mddev->degraded != 1) {
N
NeilBrown 已提交
554 555
		printk(KERN_ERR "md/raid0:%s: raid5 must be degraded! Degraded disks: %d\n",
		       mdname(mddev),
556 557 558 559
		       mddev->degraded);
		return ERR_PTR(-EINVAL);
	}

N
NeilBrown 已提交
560
	rdev_for_each(rdev, mddev) {
561 562
		/* check slot number for a disk */
		if (rdev->raid_disk == mddev->raid_disks-1) {
N
NeilBrown 已提交
563 564
			printk(KERN_ERR "md/raid0:%s: raid5 must have missing parity disk!\n",
			       mdname(mddev));
565 566
			return ERR_PTR(-EINVAL);
		}
567
		rdev->sectors = mddev->dev_sectors;
568 569 570 571
	}

	/* Set new parameters */
	mddev->new_level = 0;
572
	mddev->new_layout = 0;
573 574 575 576 577 578 579 580 581 582
	mddev->new_chunk_sectors = mddev->chunk_sectors;
	mddev->raid_disks--;
	mddev->delta_disks = -1;
	/* make sure it will be not marked as dirty */
	mddev->recovery_cp = MaxSector;

	create_strip_zones(mddev, &priv_conf);
	return priv_conf;
}

583
static void *raid0_takeover_raid10(struct mddev *mddev)
584
{
585
	struct r0conf *priv_conf;
586 587 588 589 590 591 592 593

	/* Check layout:
	 *  - far_copies must be 1
	 *  - near_copies must be 2
	 *  - disks number must be even
	 *  - all mirrors must be already degraded
	 */
	if (mddev->layout != ((1 << 8) + 2)) {
N
NeilBrown 已提交
594 595
		printk(KERN_ERR "md/raid0:%s:: Raid0 cannot takover layout: 0x%x\n",
		       mdname(mddev),
596 597 598 599
		       mddev->layout);
		return ERR_PTR(-EINVAL);
	}
	if (mddev->raid_disks & 1) {
N
NeilBrown 已提交
600 601
		printk(KERN_ERR "md/raid0:%s: Raid0 cannot takover Raid10 with odd disk number.\n",
		       mdname(mddev));
602 603 604
		return ERR_PTR(-EINVAL);
	}
	if (mddev->degraded != (mddev->raid_disks>>1)) {
N
NeilBrown 已提交
605 606
		printk(KERN_ERR "md/raid0:%s: All mirrors must be already degraded!\n",
		       mdname(mddev));
607 608 609 610 611
		return ERR_PTR(-EINVAL);
	}

	/* Set new parameters */
	mddev->new_level = 0;
612
	mddev->new_layout = 0;
613 614 615 616 617 618 619 620 621 622 623
	mddev->new_chunk_sectors = mddev->chunk_sectors;
	mddev->delta_disks = - mddev->raid_disks / 2;
	mddev->raid_disks += mddev->delta_disks;
	mddev->degraded = 0;
	/* make sure it will be not marked as dirty */
	mddev->recovery_cp = MaxSector;

	create_strip_zones(mddev, &priv_conf);
	return priv_conf;
}

624
static void *raid0_takeover_raid1(struct mddev *mddev)
625
{
626
	struct r0conf *priv_conf;
627
	int chunksect;
628 629 630 631 632 633 634 635 636 637

	/* Check layout:
	 *  - (N - 1) mirror drives must be already faulty
	 */
	if ((mddev->raid_disks - 1) != mddev->degraded) {
		printk(KERN_ERR "md/raid0:%s: (N - 1) mirrors drives must be already faulty!\n",
		       mdname(mddev));
		return ERR_PTR(-EINVAL);
	}

638 639 640 641 642 643 644 645 646 647 648 649 650 651
	/*
	 * a raid1 doesn't have the notion of chunk size, so
	 * figure out the largest suitable size we can use.
	 */
	chunksect = 64 * 2; /* 64K by default */

	/* The array must be an exact multiple of chunksize */
	while (chunksect && (mddev->array_sectors & (chunksect - 1)))
		chunksect >>= 1;

	if ((chunksect << 9) < PAGE_SIZE)
		/* array size does not allow a suitable chunk size */
		return ERR_PTR(-EINVAL);

652 653 654
	/* Set new parameters */
	mddev->new_level = 0;
	mddev->new_layout = 0;
655 656
	mddev->new_chunk_sectors = chunksect;
	mddev->chunk_sectors = chunksect;
657
	mddev->delta_disks = 1 - mddev->raid_disks;
K
Krzysztof Wojcik 已提交
658
	mddev->raid_disks = 1;
659 660 661 662 663 664 665
	/* make sure it will be not marked as dirty */
	mddev->recovery_cp = MaxSector;

	create_strip_zones(mddev, &priv_conf);
	return priv_conf;
}

666
static void *raid0_takeover(struct mddev *mddev)
667 668
{
	/* raid0 can take over:
M
Maciej Trela 已提交
669
	 *  raid4 - if all data disks are active.
670 671
	 *  raid5 - providing it is Raid4 layout and one disk is faulty
	 *  raid10 - assuming we have all necessary active disks
672
	 *  raid1 - with (N -1) mirror drives faulty
673
	 */
674 675 676 677 678 679

	if (mddev->bitmap) {
		printk(KERN_ERR "md/raid0: %s: cannot takeover array with bitmap\n",
		       mdname(mddev));
		return ERR_PTR(-EBUSY);
	}
M
Maciej Trela 已提交
680 681 682
	if (mddev->level == 4)
		return raid0_takeover_raid45(mddev);

683 684
	if (mddev->level == 5) {
		if (mddev->layout == ALGORITHM_PARITY_N)
M
Maciej Trela 已提交
685
			return raid0_takeover_raid45(mddev);
686

N
NeilBrown 已提交
687 688
		printk(KERN_ERR "md/raid0:%s: Raid can only takeover Raid5 with layout: %d\n",
		       mdname(mddev), ALGORITHM_PARITY_N);
689 690 691 692 693
	}

	if (mddev->level == 10)
		return raid0_takeover_raid10(mddev);

694 695 696 697 698 699
	if (mddev->level == 1)
		return raid0_takeover_raid1(mddev);

	printk(KERN_ERR "Takeover from raid%i to raid0 not supported\n",
		mddev->level);

700 701 702
	return ERR_PTR(-EINVAL);
}

703
static void raid0_quiesce(struct mddev *mddev, int state)
704 705 706
{
}

707
static struct md_personality raid0_personality=
L
Linus Torvalds 已提交
708 709
{
	.name		= "raid0",
710
	.level		= 0,
L
Linus Torvalds 已提交
711 712 713
	.owner		= THIS_MODULE,
	.make_request	= raid0_make_request,
	.run		= raid0_run,
N
NeilBrown 已提交
714
	.free		= raid0_free,
L
Linus Torvalds 已提交
715
	.status		= raid0_status,
716
	.size		= raid0_size,
717 718
	.takeover	= raid0_takeover,
	.quiesce	= raid0_quiesce,
719
	.congested	= raid0_congested,
720
	.mergeable_bvec	= raid0_mergeable_bvec,
L
Linus Torvalds 已提交
721 722 723 724
};

static int __init raid0_init (void)
{
725
	return register_md_personality (&raid0_personality);
L
Linus Torvalds 已提交
726 727 728 729
}

static void raid0_exit (void)
{
730
	unregister_md_personality (&raid0_personality);
L
Linus Torvalds 已提交
731 732 733 734 735
}

module_init(raid0_init);
module_exit(raid0_exit);
MODULE_LICENSE("GPL");
736
MODULE_DESCRIPTION("RAID0 (striping) personality for MD");
L
Linus Torvalds 已提交
737
MODULE_ALIAS("md-personality-2"); /* RAID0 */
738
MODULE_ALIAS("md-raid0");
739
MODULE_ALIAS("md-level-0");