raid0.c 19.1 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
/*
   raid0.c : Multiple Devices driver for Linux
             Copyright (C) 1994-96 Marc ZYNGIER
	     <zyngier@ufr-info-p7.ibp.fr> or
	     <maz@gloups.fdn.fr>
             Copyright (C) 1999, 2000 Ingo Molnar, Red Hat


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

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

27
static void raid0_unplug(struct request_queue *q)
L
Linus Torvalds 已提交
28 29
{
	mddev_t *mddev = q->queuedata;
30
	raid0_conf_t *conf = mddev->private;
31
	mdk_rdev_t **devlist = conf->devlist;
32
	int raid_disks = conf->strip_zone[0].nb_dev;
L
Linus Torvalds 已提交
33 34
	int i;

35
	for (i=0; i < raid_disks; i++) {
36
		struct request_queue *r_queue = bdev_get_queue(devlist[i]->bdev);
L
Linus Torvalds 已提交
37

38
		blk_unplug(r_queue);
L
Linus Torvalds 已提交
39 40 41
	}
}

42 43 44
static int raid0_congested(void *data, int bits)
{
	mddev_t *mddev = data;
45
	raid0_conf_t *conf = mddev->private;
46
	mdk_rdev_t **devlist = conf->devlist;
47
	int raid_disks = conf->strip_zone[0].nb_dev;
48 49
	int i, ret = 0;

50 51 52
	if (mddev_congested(mddev, bits))
		return 1;

53
	for (i = 0; i < raid_disks && !ret ; i++) {
54
		struct request_queue *q = bdev_get_queue(devlist[i]->bdev);
55 56 57 58 59 60

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

61 62 63 64 65 66 67 68 69 70
/*
 * inform the user of the raid configuration
*/
static void dump_zones(mddev_t *mddev)
{
	int j, k, h;
	sector_t zone_size = 0;
	sector_t zone_start = 0;
	char b[BDEVNAME_SIZE];
	raid0_conf_t *conf = mddev->private;
71
	int raid_disks = conf->strip_zone[0].nb_dev;
72 73 74 75 76 77
	printk(KERN_INFO "******* %s configuration *********\n",
		mdname(mddev));
	h = 0;
	for (j = 0; j < conf->nr_strip_zones; j++) {
		printk(KERN_INFO "zone%d=[", j);
		for (k = 0; k < conf->strip_zone[j].nb_dev; k++)
78
			printk(KERN_CONT "%s/",
79
			bdevname(conf->devlist[j*raid_disks
80
						+ k]->bdev, b));
81
		printk(KERN_CONT "]\n");
82 83 84 85 86 87 88 89 90 91 92 93

		zone_size  = conf->strip_zone[j].zone_end - zone_start;
		printk(KERN_INFO "        zone offset=%llukb "
				"device offset=%llukb size=%llukb\n",
			(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;
	}
	printk(KERN_INFO "**********************************\n\n");
}

94
static int create_strip_zones(mddev_t *mddev, raid0_conf_t **private_conf)
L
Linus Torvalds 已提交
95
{
96
	int i, c, err;
97
	sector_t curr_zone_end, sectors;
98
	mdk_rdev_t *smallest, *rdev1, *rdev2, *rdev, **dev;
L
Linus Torvalds 已提交
99 100 101
	struct strip_zone *zone;
	int cnt;
	char b[BDEVNAME_SIZE];
102 103 104 105
	raid0_conf_t *conf = kzalloc(sizeof(*conf), GFP_KERNEL);

	if (!conf)
		return -ENOMEM;
106
	list_for_each_entry(rdev1, &mddev->disks, same_set) {
107 108 109
		printk(KERN_INFO "md/raid0:%s: looking at %s\n",
		       mdname(mddev),
		       bdevname(rdev1->bdev, b));
L
Linus Torvalds 已提交
110
		c = 0;
111 112 113 114 115 116

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

117
		list_for_each_entry(rdev2, &mddev->disks, same_set) {
118 119
			printk(KERN_INFO "md/raid0:%s:   comparing %s(%llu)",
			       mdname(mddev),
L
Linus Torvalds 已提交
120
			       bdevname(rdev1->bdev,b),
121
			       (unsigned long long)rdev1->sectors);
122
			printk(KERN_CONT " with %s(%llu)\n",
L
Linus Torvalds 已提交
123
			       bdevname(rdev2->bdev,b),
124
			       (unsigned long long)rdev2->sectors);
L
Linus Torvalds 已提交
125
			if (rdev2 == rdev1) {
126 127
				printk(KERN_INFO "md/raid0:%s:   END\n",
				       mdname(mddev));
L
Linus Torvalds 已提交
128 129
				break;
			}
130
			if (rdev2->sectors == rdev1->sectors) {
L
Linus Torvalds 已提交
131 132 133 134
				/*
				 * Not unique, don't count it as a new
				 * group
				 */
135 136
				printk(KERN_INFO "md/raid0:%s:   EQUAL\n",
				       mdname(mddev));
L
Linus Torvalds 已提交
137 138 139
				c = 1;
				break;
			}
140 141
			printk(KERN_INFO "md/raid0:%s:   NOT EQUAL\n",
			       mdname(mddev));
L
Linus Torvalds 已提交
142 143
		}
		if (!c) {
144 145
			printk(KERN_INFO "md/raid0:%s:   ==> UNIQUE\n",
			       mdname(mddev));
L
Linus Torvalds 已提交
146
			conf->nr_strip_zones++;
147 148
			printk(KERN_INFO "md/raid0:%s: %d zones\n",
			       mdname(mddev), conf->nr_strip_zones);
L
Linus Torvalds 已提交
149 150
		}
	}
151 152
	printk(KERN_INFO "md/raid0:%s: FINAL %d zones\n",
	       mdname(mddev), conf->nr_strip_zones);
153
	err = -ENOMEM;
154
	conf->strip_zone = kzalloc(sizeof(struct strip_zone)*
L
Linus Torvalds 已提交
155 156
				conf->nr_strip_zones, GFP_KERNEL);
	if (!conf->strip_zone)
157
		goto abort;
158
	conf->devlist = kzalloc(sizeof(mdk_rdev_t*)*
L
Linus Torvalds 已提交
159 160 161
				conf->nr_strip_zones*mddev->raid_disks,
				GFP_KERNEL);
	if (!conf->devlist)
162
		goto abort;
L
Linus Torvalds 已提交
163 164 165 166 167 168 169

	/* 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;
170
	dev = conf->devlist;
171
	err = -EINVAL;
172
	list_for_each_entry(rdev1, &mddev->disks, same_set) {
L
Linus Torvalds 已提交
173 174
		int j = rdev1->raid_disk;

175 176 177 178
		if (mddev->level == 10)
			/* taking over a raid10-n2 array */
			j /= 2;

L
Linus Torvalds 已提交
179
		if (j < 0 || j >= mddev->raid_disks) {
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]) {
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
		/* as we don't honour merge_bvec_fn, we must never risk
194 195
		 * violating it, so limit ->max_segments to 1, lying within
		 * a single page.
L
Linus Torvalds 已提交
196 197
		 */

198 199 200 201 202
		if (rdev1->bdev->bd_disk->queue->merge_bvec_fn) {
			blk_queue_max_segments(mddev->queue, 1);
			blk_queue_segment_boundary(mddev->queue,
						   PAGE_CACHE_SIZE - 1);
		}
203
		if (!smallest || (rdev1->sectors < smallest->sectors))
L
Linus Torvalds 已提交
204 205 206 207
			smallest = rdev1;
		cnt++;
	}
	if (cnt != mddev->raid_disks) {
208 209
		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
210 211 212
		goto abort;
	}
	zone->nb_dev = cnt;
213
	zone->zone_end = smallest->sectors * cnt;
L
Linus Torvalds 已提交
214

215
	curr_zone_end = zone->zone_end;
L
Linus Torvalds 已提交
216 217 218 219

	/* now do the other zones */
	for (i = 1; i < conf->nr_strip_zones; i++)
	{
220 221
		int j;

L
Linus Torvalds 已提交
222
		zone = conf->strip_zone + i;
223
		dev = conf->devlist + i * mddev->raid_disks;
L
Linus Torvalds 已提交
224

225 226
		printk(KERN_INFO "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];
233 234 235
			printk(KERN_INFO "md/raid0:%s: checking %s ...",
			       mdname(mddev),
			       bdevname(rdev->bdev, b));
236
			if (rdev->sectors <= zone->dev_start) {
237
				printk(KERN_CONT " nope.\n");
238 239
				continue;
			}
240
			printk(KERN_CONT " contained as device %d\n", c);
241
			dev[c] = rdev;
242 243 244
			c++;
			if (!smallest || rdev->sectors < smallest->sectors) {
				smallest = rdev;
245 246 247
				printk(KERN_INFO "md/raid0:%s:  (%llu) is smallest!.\n",
				       mdname(mddev),
				       (unsigned long long)rdev->sectors);
248
			}
L
Linus Torvalds 已提交
249 250 251
		}

		zone->nb_dev = c;
252
		sectors = (smallest->sectors - zone->dev_start) * c;
253 254 255
		printk(KERN_INFO "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
256

257
		curr_zone_end += sectors;
258
		zone->zone_end = curr_zone_end;
L
Linus Torvalds 已提交
259

260 261 262
		printk(KERN_INFO "md/raid0:%s: current zone start: %llu\n",
		       mdname(mddev),
		       (unsigned long long)smallest->sectors);
L
Linus Torvalds 已提交
263 264
	}
	mddev->queue->unplug_fn = raid0_unplug;
265 266
	mddev->queue->backing_dev_info.congested_fn = raid0_congested;
	mddev->queue->backing_dev_info.congested_data = mddev;
L
Linus Torvalds 已提交
267

268 269 270 271
	/*
	 * now since we have the hard sector sizes, we can make sure
	 * chunk size is a multiple of that sector size
	 */
272
	if ((mddev->chunk_sectors << 9) % queue_logical_block_size(mddev->queue)) {
273
		printk(KERN_ERR "md/raid0:%s: chunk_size of %d not valid\n",
274
		       mdname(mddev),
275
		       mddev->chunk_sectors << 9);
276 277
		goto abort;
	}
278 279 280 281 282

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

283
	printk(KERN_INFO "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 = NULL;
292
	return err;
L
Linus Torvalds 已提交
293 294 295 296 297
}

/**
 *	raid0_mergeable_bvec -- tell bio layer if a two requests can be merged
 *	@q: request queue
298
 *	@bvm: properties of new bio
L
Linus Torvalds 已提交
299 300 301 302
 *	@biovec: the request that could be merged to it.
 *
 *	Return amount of bytes we can accept at this offset
 */
303 304 305
static int raid0_mergeable_bvec(struct request_queue *q,
				struct bvec_merge_data *bvm,
				struct bio_vec *biovec)
L
Linus Torvalds 已提交
306 307
{
	mddev_t *mddev = q->queuedata;
308
	sector_t sector = bvm->bi_sector + get_start_sect(bvm->bi_bdev);
L
Linus Torvalds 已提交
309
	int max;
310
	unsigned int chunk_sectors = mddev->chunk_sectors;
311
	unsigned int bio_sectors = bvm->bi_size >> 9;
L
Linus Torvalds 已提交
312

313
	if (is_power_of_2(chunk_sectors))
314 315 316 317 318
		max =  (chunk_sectors - ((sector & (chunk_sectors-1))
						+ bio_sectors)) << 9;
	else
		max =  (chunk_sectors - (sector_div(sector, chunk_sectors)
						+ bio_sectors)) << 9;
L
Linus Torvalds 已提交
319 320 321 322 323 324 325
	if (max < 0) max = 0; /* bio_add cannot handle a negative return */
	if (max <= biovec->bv_len && bio_sectors == 0)
		return biovec->bv_len;
	else 
		return max;
}

326 327 328 329 330 331 332 333 334 335 336 337 338 339
static sector_t raid0_size(mddev_t *mddev, sector_t sectors, int raid_disks)
{
	sector_t array_sectors = 0;
	mdk_rdev_t *rdev;

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

	list_for_each_entry(rdev, &mddev->disks, same_set)
		array_sectors += rdev->sectors;

	return array_sectors;
}

340
static int raid0_run(mddev_t *mddev)
L
Linus Torvalds 已提交
341
{
342
	raid0_conf_t *conf;
343
	int ret;
L
Linus Torvalds 已提交
344

345
	if (mddev->chunk_sectors == 0) {
346 347
		printk(KERN_ERR "md/raid0:%s: chunk size must be set.\n",
		       mdname(mddev));
348 349
		return -EINVAL;
	}
350 351
	if (md_check_no_bitmap(mddev))
		return -EINVAL;
352
	blk_queue_max_hw_sectors(mddev->queue, mddev->chunk_sectors);
353
	mddev->queue->queue_lock = &mddev->queue->__queue_lock;
L
Linus Torvalds 已提交
354

355 356 357 358 359 360 361 362 363 364 365 366 367 368
	/* 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;
	if (conf->scale_raid_disks) {
		int i;
		for (i=0; i < conf->strip_zone[0].nb_dev; i++)
			conf->devlist[i]->raid_disk /= conf->scale_raid_disks;
		/* FIXME update sysfs rd links */
	}
L
Linus Torvalds 已提交
369 370

	/* calculate array device size */
371
	md_set_array_sectors(mddev, raid0_size(mddev, 0, 0));
L
Linus Torvalds 已提交
372

373 374 375
	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
376 377 378 379 380 381 382 383 384 385
	/* 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.
	 */
	{
386 387
		int stripe = mddev->raid_disks *
			(mddev->chunk_sectors << 9) / PAGE_SIZE;
L
Linus Torvalds 已提交
388 389 390 391 392
		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);
393
	dump_zones(mddev);
394
	md_integrity_register(mddev);
L
Linus Torvalds 已提交
395 396 397
	return 0;
}

398
static int raid0_stop(mddev_t *mddev)
L
Linus Torvalds 已提交
399
{
400
	raid0_conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
401 402

	blk_sync_queue(mddev->queue); /* the unplug fn references 'conf'*/
403
	kfree(conf->strip_zone);
404
	kfree(conf->devlist);
405
	kfree(conf);
L
Linus Torvalds 已提交
406 407 408 409
	mddev->private = NULL;
	return 0;
}

410 411 412
/* Find the zone which holds a particular offset
 * Update *sectorp to be an offset in that zone
 */
413
static struct strip_zone *find_zone(struct raid0_private_data *conf,
414
				    sector_t *sectorp)
415 416 417
{
	int i;
	struct strip_zone *z = conf->strip_zone;
418
	sector_t sector = *sectorp;
419 420

	for (i = 0; i < conf->nr_strip_zones; i++)
421 422 423
		if (sector < z[i].zone_end) {
			if (i)
				*sectorp = sector - z[i-1].zone_end;
424
			return z + i;
425
		}
426 427 428
	BUG();
}

429 430 431 432 433 434
/*
 * remaps the bio to the target device. we separate two flows.
 * power 2 flow and a general flow for the sake of perfromance
*/
static mdk_rdev_t *map_sector(mddev_t *mddev, struct strip_zone *zone,
				sector_t sector, sector_t *sector_offset)
L
Linus Torvalds 已提交
435
{
436 437
	unsigned int sect_in_chunk;
	sector_t chunk;
438
	raid0_conf_t *conf = mddev->private;
439
	int raid_disks = conf->strip_zone[0].nb_dev;
440
	unsigned int chunk_sects = mddev->chunk_sectors;
441

442
	if (is_power_of_2(chunk_sects)) {
443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461
		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;
462
	return conf->devlist[(zone - conf->strip_zone)*raid_disks
463 464 465 466 467 468 469 470 471
			     + sector_div(sector, zone->nb_dev)];
}

/*
 * Is io distribute over 1 or more chunks ?
*/
static inline int is_io_in_chunk_boundary(mddev_t *mddev,
			unsigned int chunk_sects, struct bio *bio)
{
472
	if (likely(is_power_of_2(chunk_sects))) {
473 474 475 476 477 478 479 480 481
		return chunk_sects >= ((bio->bi_sector & (chunk_sects-1))
					+ (bio->bi_size >> 9));
	} else{
		sector_t sector = bio->bi_sector;
		return chunk_sects >= (sector_div(sector, chunk_sects)
						+ (bio->bi_size >> 9));
	}
}

482
static int raid0_make_request(mddev_t *mddev, struct bio *bio)
483 484 485
{
	unsigned int chunk_sects;
	sector_t sector_offset;
L
Linus Torvalds 已提交
486 487 488
	struct strip_zone *zone;
	mdk_rdev_t *tmp_dev;

489
	if (unlikely(bio_rw_flagged(bio, BIO_RW_BARRIER))) {
490
		md_barrier_request(mddev, bio);
491 492 493
		return 0;
	}

494
	chunk_sects = mddev->chunk_sectors;
495 496
	if (unlikely(!is_io_in_chunk_boundary(mddev, chunk_sects, bio))) {
		sector_t sector = bio->bi_sector;
L
Linus Torvalds 已提交
497 498 499 500 501 502 503 504
		struct bio_pair *bp;
		/* Sanity check -- queue functions should prevent this happening */
		if (bio->bi_vcnt != 1 ||
		    bio->bi_idx != 0)
			goto bad_map;
		/* This is a one page bio that upper layers
		 * refuse to split for us, so we need to split it.
		 */
505
		if (likely(is_power_of_2(chunk_sects)))
506 507 508 509 510
			bp = bio_split(bio, chunk_sects - (sector &
							   (chunk_sects-1)));
		else
			bp = bio_split(bio, chunk_sects -
				       sector_div(sector, chunk_sects));
511
		if (raid0_make_request(mddev, &bp->bio1))
L
Linus Torvalds 已提交
512
			generic_make_request(&bp->bio1);
513
		if (raid0_make_request(mddev, &bp->bio2))
L
Linus Torvalds 已提交
514 515 516 517 518 519
			generic_make_request(&bp->bio2);

		bio_pair_release(bp);
		return 0;
	}

520 521 522 523
	sector_offset = bio->bi_sector;
	zone =  find_zone(mddev->private, &sector_offset);
	tmp_dev = map_sector(mddev, zone, bio->bi_sector,
			     &sector_offset);
L
Linus Torvalds 已提交
524
	bio->bi_bdev = tmp_dev->bdev;
525 526
	bio->bi_sector = sector_offset + zone->dev_start +
		tmp_dev->data_offset;
L
Linus Torvalds 已提交
527 528 529 530 531 532
	/*
	 * Let the main block layer submit the IO and resolve recursion:
	 */
	return 1;

bad_map:
533 534 535 536
	printk("md/raid0:%s: make_request bug: can't convert block across chunks"
	       " or bigger than %dk %llu %d\n",
	       mdname(mddev), chunk_sects / 2,
	       (unsigned long long)bio->bi_sector, bio->bi_size >> 10);
L
Linus Torvalds 已提交
537

538
	bio_io_error(bio);
L
Linus Torvalds 已提交
539 540
	return 0;
}
541

542
static void raid0_status(struct seq_file *seq, mddev_t *mddev)
L
Linus Torvalds 已提交
543 544 545 546 547
{
#undef MD_DEBUG
#ifdef MD_DEBUG
	int j, k, h;
	char b[BDEVNAME_SIZE];
548
	raid0_conf_t *conf = mddev->private;
549
	int raid_disks = conf->strip_zone[0].nb_dev;
550

551 552
	sector_t zone_size;
	sector_t zone_start = 0;
L
Linus Torvalds 已提交
553
	h = 0;
554

L
Linus Torvalds 已提交
555 556 557 558
	for (j = 0; j < conf->nr_strip_zones; j++) {
		seq_printf(seq, "      z%d", j);
		seq_printf(seq, "=[");
		for (k = 0; k < conf->strip_zone[j].nb_dev; k++)
559
			seq_printf(seq, "%s/", bdevname(
560
				conf->devlist[j*raid_disks + k]
561 562 563 564 565 566 567 568
						->bdev, b));

		zone_size  = conf->strip_zone[j].zone_end - zone_start;
		seq_printf(seq, "] ze=%lld ds=%lld s=%lld\n",
			(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;
L
Linus Torvalds 已提交
569 570
	}
#endif
571
	seq_printf(seq, " %dk chunks", mddev->chunk_sectors / 2);
L
Linus Torvalds 已提交
572 573 574
	return;
}

575 576 577 578 579 580
static void *raid0_takeover_raid5(mddev_t *mddev)
{
	mdk_rdev_t *rdev;
	raid0_conf_t *priv_conf;

	if (mddev->degraded != 1) {
581 582
		printk(KERN_ERR "md/raid0:%s: raid5 must be degraded! Degraded disks: %d\n",
		       mdname(mddev),
583 584 585 586 587 588 589
		       mddev->degraded);
		return ERR_PTR(-EINVAL);
	}

	list_for_each_entry(rdev, &mddev->disks, same_set) {
		/* check slot number for a disk */
		if (rdev->raid_disk == mddev->raid_disks-1) {
590 591
			printk(KERN_ERR "md/raid0:%s: raid5 must have missing parity disk!\n",
			       mdname(mddev));
592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618
			return ERR_PTR(-EINVAL);
		}
	}

	/* Set new parameters */
	mddev->new_level = 0;
	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;
}

static void *raid0_takeover_raid10(mddev_t *mddev)
{
	raid0_conf_t *priv_conf;

	/* 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)) {
619 620
		printk(KERN_ERR "md/raid0:%s:: Raid0 cannot takover layout: 0x%x\n",
		       mdname(mddev),
621 622 623 624
		       mddev->layout);
		return ERR_PTR(-EINVAL);
	}
	if (mddev->raid_disks & 1) {
625 626
		printk(KERN_ERR "md/raid0:%s: Raid0 cannot takover Raid10 with odd disk number.\n",
		       mdname(mddev));
627 628 629
		return ERR_PTR(-EINVAL);
	}
	if (mddev->degraded != (mddev->raid_disks>>1)) {
630 631
		printk(KERN_ERR "md/raid0:%s: All mirrors must be already degraded!\n",
		       mdname(mddev));
632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658
		return ERR_PTR(-EINVAL);
	}

	/* Set new parameters */
	mddev->new_level = 0;
	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);
	priv_conf->scale_raid_disks = 2;
	return priv_conf;
}

static void *raid0_takeover(mddev_t *mddev)
{
	/* raid0 can take over:
	 *  raid5 - providing it is Raid4 layout and one disk is faulty
	 *  raid10 - assuming we have all necessary active disks
	 */
	if (mddev->level == 5) {
		if (mddev->layout == ALGORITHM_PARITY_N)
			return raid0_takeover_raid5(mddev);

659 660
		printk(KERN_ERR "md/raid0:%s: Raid can only takeover Raid5 with layout: %d\n",
		       mdname(mddev), ALGORITHM_PARITY_N);
661 662 663 664 665 666 667 668 669 670 671 672
	}

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

	return ERR_PTR(-EINVAL);
}

static void raid0_quiesce(mddev_t *mddev, int state)
{
}

673
static struct mdk_personality raid0_personality=
L
Linus Torvalds 已提交
674 675
{
	.name		= "raid0",
676
	.level		= 0,
L
Linus Torvalds 已提交
677 678 679 680 681
	.owner		= THIS_MODULE,
	.make_request	= raid0_make_request,
	.run		= raid0_run,
	.stop		= raid0_stop,
	.status		= raid0_status,
682
	.size		= raid0_size,
683 684
	.takeover	= raid0_takeover,
	.quiesce	= raid0_quiesce,
L
Linus Torvalds 已提交
685 686 687 688
};

static int __init raid0_init (void)
{
689
	return register_md_personality (&raid0_personality);
L
Linus Torvalds 已提交
690 691 692 693
}

static void raid0_exit (void)
{
694
	unregister_md_personality (&raid0_personality);
L
Linus Torvalds 已提交
695 696 697 698 699
}

module_init(raid0_init);
module_exit(raid0_exit);
MODULE_LICENSE("GPL");
700
MODULE_DESCRIPTION("RAID0 (striping) personality for MD");
L
Linus Torvalds 已提交
701
MODULE_ALIAS("md-personality-2"); /* RAID0 */
702
MODULE_ALIAS("md-raid0");
703
MODULE_ALIAS("md-level-0");
新手
引导
客服 返回
顶部