raid0.c 19.9 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 29
static int raid0_congested(void *data, int bits)
{
30
	struct mddev *mddev = data;
31
	struct r0conf *conf = mddev->private;
32
	struct md_rdev **devlist = conf->devlist;
33
	int raid_disks = conf->strip_zone[0].nb_dev;
34 35
	int i, ret = 0;

36 37 38
	if (mddev_congested(mddev, bits))
		return 1;

39
	for (i = 0; i < raid_disks && !ret ; i++) {
40
		struct request_queue *q = bdev_get_queue(devlist[i]->bdev);
41 42 43 44 45 46

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

47 48 49
/*
 * inform the user of the raid configuration
*/
50
static void dump_zones(struct mddev *mddev)
51
{
52
	int j, k;
53 54 55
	sector_t zone_size = 0;
	sector_t zone_start = 0;
	char b[BDEVNAME_SIZE];
56
	struct r0conf *conf = mddev->private;
57
	int raid_disks = conf->strip_zone[0].nb_dev;
58 59 60
	printk(KERN_INFO "md: RAID0 configuration for %s - %d zone%s\n",
	       mdname(mddev),
	       conf->nr_strip_zones, conf->nr_strip_zones==1?"":"s");
61
	for (j = 0; j < conf->nr_strip_zones; j++) {
62
		printk(KERN_INFO "md: zone%d=[", j);
63
		for (k = 0; k < conf->strip_zone[j].nb_dev; k++)
64
			printk(KERN_CONT "%s%s", k?"/":"",
65
			bdevname(conf->devlist[j*raid_disks
66
						+ k]->bdev, b));
N
NeilBrown 已提交
67
		printk(KERN_CONT "]\n");
68 69

		zone_size  = conf->strip_zone[j].zone_end - zone_start;
70 71
		printk(KERN_INFO "      zone-offset=%10lluKB, "
				"device-offset=%10lluKB, size=%10lluKB\n",
72 73 74 75 76
			(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;
	}
77
	printk(KERN_INFO "\n");
78 79
}

80
static int create_strip_zones(struct mddev *mddev, struct r0conf **private_conf)
L
Linus Torvalds 已提交
81
{
82
	int i, c, err;
83
	sector_t curr_zone_end, sectors;
84
	struct md_rdev *smallest, *rdev1, *rdev2, *rdev, **dev;
L
Linus Torvalds 已提交
85 86 87
	struct strip_zone *zone;
	int cnt;
	char b[BDEVNAME_SIZE];
88
	char b2[BDEVNAME_SIZE];
89
	struct r0conf *conf = kzalloc(sizeof(*conf), GFP_KERNEL);
S
Shaohua Li 已提交
90
	bool discard_supported = false;
91 92 93

	if (!conf)
		return -ENOMEM;
N
NeilBrown 已提交
94
	rdev_for_each(rdev1, mddev) {
95 96 97
		pr_debug("md/raid0:%s: looking at %s\n",
			 mdname(mddev),
			 bdevname(rdev1->bdev, b));
L
Linus Torvalds 已提交
98
		c = 0;
99 100 101 102 103 104

		/* 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
105
		rdev_for_each(rdev2, mddev) {
106 107 108 109 110 111 112
			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
113
			if (rdev2 == rdev1) {
114 115
				pr_debug("md/raid0:%s:   END\n",
					 mdname(mddev));
L
Linus Torvalds 已提交
116 117
				break;
			}
118
			if (rdev2->sectors == rdev1->sectors) {
L
Linus Torvalds 已提交
119 120 121 122
				/*
				 * Not unique, don't count it as a new
				 * group
				 */
123 124
				pr_debug("md/raid0:%s:   EQUAL\n",
					 mdname(mddev));
L
Linus Torvalds 已提交
125 126 127
				c = 1;
				break;
			}
128 129
			pr_debug("md/raid0:%s:   NOT EQUAL\n",
				 mdname(mddev));
L
Linus Torvalds 已提交
130 131
		}
		if (!c) {
132 133
			pr_debug("md/raid0:%s:   ==> UNIQUE\n",
				 mdname(mddev));
L
Linus Torvalds 已提交
134
			conf->nr_strip_zones++;
135 136
			pr_debug("md/raid0:%s: %d zones\n",
				 mdname(mddev), conf->nr_strip_zones);
L
Linus Torvalds 已提交
137 138
		}
	}
139 140
	pr_debug("md/raid0:%s: FINAL %d zones\n",
		 mdname(mddev), conf->nr_strip_zones);
141
	err = -ENOMEM;
142
	conf->strip_zone = kzalloc(sizeof(struct strip_zone)*
L
Linus Torvalds 已提交
143 144
				conf->nr_strip_zones, GFP_KERNEL);
	if (!conf->strip_zone)
145
		goto abort;
146
	conf->devlist = kzalloc(sizeof(struct md_rdev*)*
L
Linus Torvalds 已提交
147 148 149
				conf->nr_strip_zones*mddev->raid_disks,
				GFP_KERNEL);
	if (!conf->devlist)
150
		goto abort;
L
Linus Torvalds 已提交
151 152 153 154 155 156 157

	/* 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;
158
	dev = conf->devlist;
159
	err = -EINVAL;
N
NeilBrown 已提交
160
	rdev_for_each(rdev1, mddev) {
L
Linus Torvalds 已提交
161 162
		int j = rdev1->raid_disk;

163
		if (mddev->level == 10) {
164 165
			/* taking over a raid10-n2 array */
			j /= 2;
166 167
			rdev1->new_raid_disk = j;
		}
168

169 170 171 172 173 174 175 176
		if (mddev->level == 1) {
			/* taiking over a raid1 array-
			 * we have only one active disk
			 */
			j = 0;
			rdev1->new_raid_disk = j;
		}

177 178 179 180 181 182 183
		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
184 185
			printk(KERN_ERR "md/raid0:%s: bad disk number %d - "
			       "aborting!\n", mdname(mddev), j);
L
Linus Torvalds 已提交
186 187
			goto abort;
		}
188
		if (dev[j]) {
N
NeilBrown 已提交
189 190
			printk(KERN_ERR "md/raid0:%s: multiple devices for %d - "
			       "aborting!\n", mdname(mddev), j);
L
Linus Torvalds 已提交
191 192
			goto abort;
		}
193
		dev[j] = rdev1;
L
Linus Torvalds 已提交
194

195 196
		disk_stack_limits(mddev->gendisk, rdev1->bdev,
				  rdev1->data_offset << 9);
L
Linus Torvalds 已提交
197

198 199 200
		if (rdev1->bdev->bd_disk->queue->merge_bvec_fn)
			conf->has_merge_bvec = 1;

201
		if (!smallest || (rdev1->sectors < smallest->sectors))
L
Linus Torvalds 已提交
202 203
			smallest = rdev1;
		cnt++;
S
Shaohua Li 已提交
204 205 206

		if (blk_queue_discard(bdev_get_queue(rdev1->bdev)))
			discard_supported = true;
L
Linus Torvalds 已提交
207 208
	}
	if (cnt != mddev->raid_disks) {
N
NeilBrown 已提交
209 210
		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
211 212 213
		goto abort;
	}
	zone->nb_dev = cnt;
214
	zone->zone_end = smallest->sectors * cnt;
L
Linus Torvalds 已提交
215

216
	curr_zone_end = zone->zone_end;
L
Linus Torvalds 已提交
217 218 219 220

	/* now do the other zones */
	for (i = 1; i < conf->nr_strip_zones; i++)
	{
221 222
		int j;

L
Linus Torvalds 已提交
223
		zone = conf->strip_zone + i;
224
		dev = conf->devlist + i * mddev->raid_disks;
L
Linus Torvalds 已提交
225

226
		pr_debug("md/raid0:%s: zone %d\n", mdname(mddev), i);
227
		zone->dev_start = smallest->sectors;
L
Linus Torvalds 已提交
228 229 230 231
		smallest = NULL;
		c = 0;

		for (j=0; j<cnt; j++) {
232
			rdev = conf->devlist[j];
233
			if (rdev->sectors <= zone->dev_start) {
234 235 236
				pr_debug("md/raid0:%s: checking %s ... nope\n",
					 mdname(mddev),
					 bdevname(rdev->bdev, b));
237 238
				continue;
			}
239 240 241 242
			pr_debug("md/raid0:%s: checking %s ..."
				 " contained as device %d\n",
				 mdname(mddev),
				 bdevname(rdev->bdev, b), c);
243
			dev[c] = rdev;
244 245 246
			c++;
			if (!smallest || rdev->sectors < smallest->sectors) {
				smallest = rdev;
247 248 249
				pr_debug("md/raid0:%s:  (%llu) is smallest!.\n",
					 mdname(mddev),
					 (unsigned long long)rdev->sectors);
250
			}
L
Linus Torvalds 已提交
251 252 253
		}

		zone->nb_dev = c;
254
		sectors = (smallest->sectors - zone->dev_start) * c;
255 256 257
		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
258

259
		curr_zone_end += sectors;
260
		zone->zone_end = curr_zone_end;
L
Linus Torvalds 已提交
261

262 263 264
		pr_debug("md/raid0:%s: current zone start: %llu\n",
			 mdname(mddev),
			 (unsigned long long)smallest->sectors);
L
Linus Torvalds 已提交
265
	}
266 267
	mddev->queue->backing_dev_info.congested_fn = raid0_congested;
	mddev->queue->backing_dev_info.congested_data = mddev;
L
Linus Torvalds 已提交
268

269 270 271 272
	/*
	 * now since we have the hard sector sizes, we can make sure
	 * chunk size is a multiple of that sector size
	 */
273
	if ((mddev->chunk_sectors << 9) % queue_logical_block_size(mddev->queue)) {
N
NeilBrown 已提交
274
		printk(KERN_ERR "md/raid0:%s: chunk_size of %d not valid\n",
275
		       mdname(mddev),
276
		       mddev->chunk_sectors << 9);
277 278
		goto abort;
	}
279 280 281 282 283

	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
284 285 286 287 288
	if (!discard_supported)
		queue_flag_clear_unlocked(QUEUE_FLAG_DISCARD, mddev->queue);
	else
		queue_flag_set_unlocked(QUEUE_FLAG_DISCARD, mddev->queue);

289
	pr_debug("md/raid0:%s: done.\n", mdname(mddev));
290 291
	*private_conf = conf;

L
Linus Torvalds 已提交
292
	return 0;
293
abort:
294 295 296
	kfree(conf->strip_zone);
	kfree(conf->devlist);
	kfree(conf);
297
	*private_conf = ERR_PTR(err);
298
	return err;
L
Linus Torvalds 已提交
299 300
}

301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356
/* 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
357
/**
358
 *	raid0_mergeable_bvec -- tell bio layer if two requests can be merged
L
Linus Torvalds 已提交
359
 *	@q: request queue
360
 *	@bvm: properties of new bio
L
Linus Torvalds 已提交
361 362 363 364
 *	@biovec: the request that could be merged to it.
 *
 *	Return amount of bytes we can accept at this offset
 */
365 366 367
static int raid0_mergeable_bvec(struct request_queue *q,
				struct bvec_merge_data *bvm,
				struct bio_vec *biovec)
L
Linus Torvalds 已提交
368
{
369
	struct mddev *mddev = q->queuedata;
370
	struct r0conf *conf = mddev->private;
371
	sector_t sector = bvm->bi_sector + get_start_sect(bvm->bi_bdev);
372
	sector_t sector_offset = sector;
L
Linus Torvalds 已提交
373
	int max;
374
	unsigned int chunk_sectors = mddev->chunk_sectors;
375
	unsigned int bio_sectors = bvm->bi_size >> 9;
376 377 378
	struct strip_zone *zone;
	struct md_rdev *rdev;
	struct request_queue *subq;
L
Linus Torvalds 已提交
379

N
NeilBrown 已提交
380
	if (is_power_of_2(chunk_sectors))
381 382 383 384 385
		max =  (chunk_sectors - ((sector & (chunk_sectors-1))
						+ bio_sectors)) << 9;
	else
		max =  (chunk_sectors - (sector_div(sector, chunk_sectors)
						+ bio_sectors)) << 9;
386 387
	if (max < 0)
		max = 0; /* bio_add cannot handle a negative return */
L
Linus Torvalds 已提交
388 389
	if (max <= biovec->bv_len && bio_sectors == 0)
		return biovec->bv_len;
390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406
	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
407 408 409
		return max;
}

410
static sector_t raid0_size(struct mddev *mddev, sector_t sectors, int raid_disks)
411 412
{
	sector_t array_sectors = 0;
413
	struct md_rdev *rdev;
414 415 416 417

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

N
NeilBrown 已提交
418
	rdev_for_each(rdev, mddev)
419 420
		array_sectors += (rdev->sectors &
				  ~(sector_t)(mddev->chunk_sectors-1));
421 422 423 424

	return array_sectors;
}

425 426
static int raid0_stop(struct mddev *mddev);

427
static int raid0_run(struct mddev *mddev)
L
Linus Torvalds 已提交
428
{
429
	struct r0conf *conf;
430
	int ret;
L
Linus Torvalds 已提交
431

432
	if (mddev->chunk_sectors == 0) {
N
NeilBrown 已提交
433 434
		printk(KERN_ERR "md/raid0:%s: chunk size must be set.\n",
		       mdname(mddev));
435 436
		return -EINVAL;
	}
437 438
	if (md_check_no_bitmap(mddev))
		return -EINVAL;
439
	blk_queue_max_hw_sectors(mddev->queue, mddev->chunk_sectors);
440
	blk_queue_max_write_same_sectors(mddev->queue, mddev->chunk_sectors);
S
Shaohua Li 已提交
441
	blk_queue_max_discard_sectors(mddev->queue, mddev->chunk_sectors);
L
Linus Torvalds 已提交
442

443 444 445 446 447 448 449 450
	/* 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
451 452

	/* calculate array device size */
453
	md_set_array_sectors(mddev, raid0_size(mddev, 0, 0));
L
Linus Torvalds 已提交
454

N
NeilBrown 已提交
455 456 457
	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
458 459 460 461 462 463 464 465 466 467
	/* 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.
	 */
	{
468 469
		int stripe = mddev->raid_disks *
			(mddev->chunk_sectors << 9) / PAGE_SIZE;
L
Linus Torvalds 已提交
470 471 472 473 474
		if (mddev->queue->backing_dev_info.ra_pages < 2* stripe)
			mddev->queue->backing_dev_info.ra_pages = 2* stripe;
	}

	blk_queue_merge_bvec(mddev->queue, raid0_mergeable_bvec);
475
	dump_zones(mddev);
476 477 478 479 480 481

	ret = md_integrity_register(mddev);
	if (ret)
		raid0_stop(mddev);

	return ret;
L
Linus Torvalds 已提交
482 483
}

484
static int raid0_stop(struct mddev *mddev)
L
Linus Torvalds 已提交
485
{
486
	struct r0conf *conf = mddev->private;
L
Linus Torvalds 已提交
487 488

	blk_sync_queue(mddev->queue); /* the unplug fn references 'conf'*/
489
	kfree(conf->strip_zone);
490
	kfree(conf->devlist);
491
	kfree(conf);
L
Linus Torvalds 已提交
492 493 494 495
	mddev->private = NULL;
	return 0;
}

496 497 498
/*
 * Is io distribute over 1 or more chunks ?
*/
499
static inline int is_io_in_chunk_boundary(struct mddev *mddev,
500 501
			unsigned int chunk_sects, struct bio *bio)
{
N
NeilBrown 已提交
502
	if (likely(is_power_of_2(chunk_sects))) {
503 504
		return chunk_sects >=
			((bio->bi_iter.bi_sector & (chunk_sects-1))
505
					+ bio_sectors(bio));
506
	} else{
507
		sector_t sector = bio->bi_iter.bi_sector;
508
		return chunk_sects >= (sector_div(sector, chunk_sects)
509
						+ bio_sectors(bio));
510 511 512
	}
}

513
static void raid0_make_request(struct mddev *mddev, struct bio *bio)
514
{
L
Linus Torvalds 已提交
515
	struct strip_zone *zone;
516
	struct md_rdev *tmp_dev;
K
Kent Overstreet 已提交
517
	struct bio *split;
L
Linus Torvalds 已提交
518

T
Tejun Heo 已提交
519 520
	if (unlikely(bio->bi_rw & REQ_FLUSH)) {
		md_flush_request(mddev, bio);
521
		return;
522 523
	}

K
Kent Overstreet 已提交
524
	do {
525
		sector_t sector = bio->bi_iter.bi_sector;
K
Kent Overstreet 已提交
526 527 528 529 530 531 532 533 534 535 536 537 538
		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
539

K
Kent Overstreet 已提交
540 541 542 543 544 545 546 547 548 549 550 551 552
		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
553
}
N
NeilBrown 已提交
554

555
static void raid0_status(struct seq_file *seq, struct mddev *mddev)
L
Linus Torvalds 已提交
556
{
557
	seq_printf(seq, " %dk chunks", mddev->chunk_sectors / 2);
L
Linus Torvalds 已提交
558 559 560
	return;
}

561
static void *raid0_takeover_raid45(struct mddev *mddev)
562
{
563
	struct md_rdev *rdev;
564
	struct r0conf *priv_conf;
565 566

	if (mddev->degraded != 1) {
N
NeilBrown 已提交
567 568
		printk(KERN_ERR "md/raid0:%s: raid5 must be degraded! Degraded disks: %d\n",
		       mdname(mddev),
569 570 571 572
		       mddev->degraded);
		return ERR_PTR(-EINVAL);
	}

N
NeilBrown 已提交
573
	rdev_for_each(rdev, mddev) {
574 575
		/* check slot number for a disk */
		if (rdev->raid_disk == mddev->raid_disks-1) {
N
NeilBrown 已提交
576 577
			printk(KERN_ERR "md/raid0:%s: raid5 must have missing parity disk!\n",
			       mdname(mddev));
578 579
			return ERR_PTR(-EINVAL);
		}
580
		rdev->sectors = mddev->dev_sectors;
581 582 583 584
	}

	/* Set new parameters */
	mddev->new_level = 0;
585
	mddev->new_layout = 0;
586 587 588 589 590 591 592 593 594 595
	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;
}

596
static void *raid0_takeover_raid10(struct mddev *mddev)
597
{
598
	struct r0conf *priv_conf;
599 600 601 602 603 604 605 606

	/* 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
607 608
		printk(KERN_ERR "md/raid0:%s:: Raid0 cannot takover layout: 0x%x\n",
		       mdname(mddev),
609 610 611 612
		       mddev->layout);
		return ERR_PTR(-EINVAL);
	}
	if (mddev->raid_disks & 1) {
N
NeilBrown 已提交
613 614
		printk(KERN_ERR "md/raid0:%s: Raid0 cannot takover Raid10 with odd disk number.\n",
		       mdname(mddev));
615 616 617
		return ERR_PTR(-EINVAL);
	}
	if (mddev->degraded != (mddev->raid_disks>>1)) {
N
NeilBrown 已提交
618 619
		printk(KERN_ERR "md/raid0:%s: All mirrors must be already degraded!\n",
		       mdname(mddev));
620 621 622 623 624
		return ERR_PTR(-EINVAL);
	}

	/* Set new parameters */
	mddev->new_level = 0;
625
	mddev->new_layout = 0;
626 627 628 629 630 631 632 633 634 635 636
	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;
}

637
static void *raid0_takeover_raid1(struct mddev *mddev)
638
{
639
	struct r0conf *priv_conf;
640
	int chunksect;
641 642 643 644 645 646 647 648 649 650

	/* 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);
	}

651 652 653 654 655 656 657 658 659 660 661 662 663 664
	/*
	 * 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);

665 666 667
	/* Set new parameters */
	mddev->new_level = 0;
	mddev->new_layout = 0;
668 669
	mddev->new_chunk_sectors = chunksect;
	mddev->chunk_sectors = chunksect;
670
	mddev->delta_disks = 1 - mddev->raid_disks;
K
Krzysztof Wojcik 已提交
671
	mddev->raid_disks = 1;
672 673 674 675 676 677 678
	/* make sure it will be not marked as dirty */
	mddev->recovery_cp = MaxSector;

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

679
static void *raid0_takeover(struct mddev *mddev)
680 681
{
	/* raid0 can take over:
M
Maciej Trela 已提交
682
	 *  raid4 - if all data disks are active.
683 684
	 *  raid5 - providing it is Raid4 layout and one disk is faulty
	 *  raid10 - assuming we have all necessary active disks
685
	 *  raid1 - with (N -1) mirror drives faulty
686
	 */
687 688 689 690 691 692

	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
693 694 695
	if (mddev->level == 4)
		return raid0_takeover_raid45(mddev);

696 697
	if (mddev->level == 5) {
		if (mddev->layout == ALGORITHM_PARITY_N)
M
Maciej Trela 已提交
698
			return raid0_takeover_raid45(mddev);
699

N
NeilBrown 已提交
700 701
		printk(KERN_ERR "md/raid0:%s: Raid can only takeover Raid5 with layout: %d\n",
		       mdname(mddev), ALGORITHM_PARITY_N);
702 703 704 705 706
	}

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

707 708 709 710 711 712
	if (mddev->level == 1)
		return raid0_takeover_raid1(mddev);

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

713 714 715
	return ERR_PTR(-EINVAL);
}

716
static void raid0_quiesce(struct mddev *mddev, int state)
717 718 719
{
}

720
static struct md_personality raid0_personality=
L
Linus Torvalds 已提交
721 722
{
	.name		= "raid0",
723
	.level		= 0,
L
Linus Torvalds 已提交
724 725 726 727 728
	.owner		= THIS_MODULE,
	.make_request	= raid0_make_request,
	.run		= raid0_run,
	.stop		= raid0_stop,
	.status		= raid0_status,
729
	.size		= raid0_size,
730 731
	.takeover	= raid0_takeover,
	.quiesce	= raid0_quiesce,
L
Linus Torvalds 已提交
732 733 734 735
};

static int __init raid0_init (void)
{
736
	return register_md_personality (&raid0_personality);
L
Linus Torvalds 已提交
737 738 739 740
}

static void raid0_exit (void)
{
741
	unregister_md_personality (&raid0_personality);
L
Linus Torvalds 已提交
742 743 744 745 746
}

module_init(raid0_init);
module_exit(raid0_exit);
MODULE_LICENSE("GPL");
747
MODULE_DESCRIPTION("RAID0 (striping) personality for MD");
L
Linus Torvalds 已提交
748
MODULE_ALIAS("md-personality-2"); /* RAID0 */
749
MODULE_ALIAS("md-raid0");
750
MODULE_ALIAS("md-level-0");