raid0.c 18.9 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
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 <linux/slab.h>
24
#include "md.h"
25
#include "raid0.h"
26
#include "raid5.h"
Linus Torvalds's avatar
Linus Torvalds committed
27

28 29 30
static int raid0_congested(void *data, int bits)
{
	mddev_t *mddev = data;
31
	raid0_conf_t *conf = mddev->private;
32
	mdk_rdev_t **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 50 51 52 53 54 55 56
/*
 * 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;
57
	int raid_disks = conf->strip_zone[0].nb_dev;
58 59 60 61 62 63
	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++)
NeilBrown's avatar
NeilBrown committed
64
			printk(KERN_CONT "%s/",
65
			bdevname(conf->devlist[j*raid_disks
66
						+ k]->bdev, b));
NeilBrown's avatar
NeilBrown committed
67
		printk(KERN_CONT "]\n");
68 69 70 71 72 73 74 75 76 77 78 79

		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");
}

80
static int create_strip_zones(mddev_t *mddev, raid0_conf_t **private_conf)
Linus Torvalds's avatar
Linus Torvalds committed
81
{
82
	int i, c, err;
83
	sector_t curr_zone_end, sectors;
84
	mdk_rdev_t *smallest, *rdev1, *rdev2, *rdev, **dev;
Linus Torvalds's avatar
Linus Torvalds committed
85 86 87
	struct strip_zone *zone;
	int cnt;
	char b[BDEVNAME_SIZE];
88 89 90 91
	raid0_conf_t *conf = kzalloc(sizeof(*conf), GFP_KERNEL);

	if (!conf)
		return -ENOMEM;
92
	list_for_each_entry(rdev1, &mddev->disks, same_set) {
NeilBrown's avatar
NeilBrown committed
93 94 95
		printk(KERN_INFO "md/raid0:%s: looking at %s\n",
		       mdname(mddev),
		       bdevname(rdev1->bdev, b));
Linus Torvalds's avatar
Linus Torvalds committed
96
		c = 0;
97 98 99 100 101 102

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

103
		list_for_each_entry(rdev2, &mddev->disks, same_set) {
NeilBrown's avatar
NeilBrown committed
104 105
			printk(KERN_INFO "md/raid0:%s:   comparing %s(%llu)",
			       mdname(mddev),
Linus Torvalds's avatar
Linus Torvalds committed
106
			       bdevname(rdev1->bdev,b),
107
			       (unsigned long long)rdev1->sectors);
NeilBrown's avatar
NeilBrown committed
108
			printk(KERN_CONT " with %s(%llu)\n",
Linus Torvalds's avatar
Linus Torvalds committed
109
			       bdevname(rdev2->bdev,b),
110
			       (unsigned long long)rdev2->sectors);
Linus Torvalds's avatar
Linus Torvalds committed
111
			if (rdev2 == rdev1) {
NeilBrown's avatar
NeilBrown committed
112 113
				printk(KERN_INFO "md/raid0:%s:   END\n",
				       mdname(mddev));
Linus Torvalds's avatar
Linus Torvalds committed
114 115
				break;
			}
116
			if (rdev2->sectors == rdev1->sectors) {
Linus Torvalds's avatar
Linus Torvalds committed
117 118 119 120
				/*
				 * Not unique, don't count it as a new
				 * group
				 */
NeilBrown's avatar
NeilBrown committed
121 122
				printk(KERN_INFO "md/raid0:%s:   EQUAL\n",
				       mdname(mddev));
Linus Torvalds's avatar
Linus Torvalds committed
123 124 125
				c = 1;
				break;
			}
NeilBrown's avatar
NeilBrown committed
126 127
			printk(KERN_INFO "md/raid0:%s:   NOT EQUAL\n",
			       mdname(mddev));
Linus Torvalds's avatar
Linus Torvalds committed
128 129
		}
		if (!c) {
NeilBrown's avatar
NeilBrown committed
130 131
			printk(KERN_INFO "md/raid0:%s:   ==> UNIQUE\n",
			       mdname(mddev));
Linus Torvalds's avatar
Linus Torvalds committed
132
			conf->nr_strip_zones++;
NeilBrown's avatar
NeilBrown committed
133 134
			printk(KERN_INFO "md/raid0:%s: %d zones\n",
			       mdname(mddev), conf->nr_strip_zones);
Linus Torvalds's avatar
Linus Torvalds committed
135 136
		}
	}
NeilBrown's avatar
NeilBrown committed
137 138
	printk(KERN_INFO "md/raid0:%s: FINAL %d zones\n",
	       mdname(mddev), conf->nr_strip_zones);
139
	err = -ENOMEM;
140
	conf->strip_zone = kzalloc(sizeof(struct strip_zone)*
Linus Torvalds's avatar
Linus Torvalds committed
141 142
				conf->nr_strip_zones, GFP_KERNEL);
	if (!conf->strip_zone)
143
		goto abort;
144
	conf->devlist = kzalloc(sizeof(mdk_rdev_t*)*
Linus Torvalds's avatar
Linus Torvalds committed
145 146 147
				conf->nr_strip_zones*mddev->raid_disks,
				GFP_KERNEL);
	if (!conf->devlist)
148
		goto abort;
Linus Torvalds's avatar
Linus Torvalds committed
149 150 151 152 153 154 155

	/* 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;
156
	dev = conf->devlist;
157
	err = -EINVAL;
158
	list_for_each_entry(rdev1, &mddev->disks, same_set) {
Linus Torvalds's avatar
Linus Torvalds committed
159 160
		int j = rdev1->raid_disk;

161
		if (mddev->level == 10) {
162 163
			/* taking over a raid10-n2 array */
			j /= 2;
164 165
			rdev1->new_raid_disk = j;
		}
166

167 168 169 170 171 172 173 174
		if (mddev->level == 1) {
			/* taiking over a raid1 array-
			 * we have only one active disk
			 */
			j = 0;
			rdev1->new_raid_disk = j;
		}

Linus Torvalds's avatar
Linus Torvalds committed
175
		if (j < 0 || j >= mddev->raid_disks) {
NeilBrown's avatar
NeilBrown committed
176 177
			printk(KERN_ERR "md/raid0:%s: bad disk number %d - "
			       "aborting!\n", mdname(mddev), j);
Linus Torvalds's avatar
Linus Torvalds committed
178 179
			goto abort;
		}
180
		if (dev[j]) {
NeilBrown's avatar
NeilBrown committed
181 182
			printk(KERN_ERR "md/raid0:%s: multiple devices for %d - "
			       "aborting!\n", mdname(mddev), j);
Linus Torvalds's avatar
Linus Torvalds committed
183 184
			goto abort;
		}
185
		dev[j] = rdev1;
Linus Torvalds's avatar
Linus Torvalds committed
186

187 188
		disk_stack_limits(mddev->gendisk, rdev1->bdev,
				  rdev1->data_offset << 9);
Linus Torvalds's avatar
Linus Torvalds committed
189
		/* as we don't honour merge_bvec_fn, we must never risk
190 191
		 * violating it, so limit ->max_segments to 1, lying within
		 * a single page.
Linus Torvalds's avatar
Linus Torvalds committed
192 193
		 */

194 195 196 197 198
		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);
		}
199
		if (!smallest || (rdev1->sectors < smallest->sectors))
Linus Torvalds's avatar
Linus Torvalds committed
200 201 202 203
			smallest = rdev1;
		cnt++;
	}
	if (cnt != mddev->raid_disks) {
NeilBrown's avatar
NeilBrown committed
204 205
		printk(KERN_ERR "md/raid0:%s: too few disks (%d of %d) - "
		       "aborting!\n", mdname(mddev), cnt, mddev->raid_disks);
Linus Torvalds's avatar
Linus Torvalds committed
206 207 208
		goto abort;
	}
	zone->nb_dev = cnt;
209
	zone->zone_end = smallest->sectors * cnt;
Linus Torvalds's avatar
Linus Torvalds committed
210

211
	curr_zone_end = zone->zone_end;
Linus Torvalds's avatar
Linus Torvalds committed
212 213 214 215

	/* now do the other zones */
	for (i = 1; i < conf->nr_strip_zones; i++)
	{
216 217
		int j;

Linus Torvalds's avatar
Linus Torvalds committed
218
		zone = conf->strip_zone + i;
219
		dev = conf->devlist + i * mddev->raid_disks;
Linus Torvalds's avatar
Linus Torvalds committed
220

NeilBrown's avatar
NeilBrown committed
221 222
		printk(KERN_INFO "md/raid0:%s: zone %d\n",
		       mdname(mddev), i);
223
		zone->dev_start = smallest->sectors;
Linus Torvalds's avatar
Linus Torvalds committed
224 225 226 227
		smallest = NULL;
		c = 0;

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

		zone->nb_dev = c;
248
		sectors = (smallest->sectors - zone->dev_start) * c;
NeilBrown's avatar
NeilBrown committed
249 250 251
		printk(KERN_INFO "md/raid0:%s: zone->nb_dev: %d, sectors: %llu\n",
		       mdname(mddev),
		       zone->nb_dev, (unsigned long long)sectors);
Linus Torvalds's avatar
Linus Torvalds committed
252

253
		curr_zone_end += sectors;
254
		zone->zone_end = curr_zone_end;
Linus Torvalds's avatar
Linus Torvalds committed
255

NeilBrown's avatar
NeilBrown committed
256 257 258
		printk(KERN_INFO "md/raid0:%s: current zone start: %llu\n",
		       mdname(mddev),
		       (unsigned long long)smallest->sectors);
Linus Torvalds's avatar
Linus Torvalds committed
259
	}
260 261
	mddev->queue->backing_dev_info.congested_fn = raid0_congested;
	mddev->queue->backing_dev_info.congested_data = mddev;
Linus Torvalds's avatar
Linus Torvalds committed
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)) {
NeilBrown's avatar
NeilBrown committed
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);

NeilBrown's avatar
NeilBrown committed
278
	printk(KERN_INFO "md/raid0:%s: done.\n", mdname(mddev));
279 280
	*private_conf = conf;

Linus Torvalds's avatar
Linus Torvalds committed
281
	return 0;
282
abort:
283 284 285
	kfree(conf->strip_zone);
	kfree(conf->devlist);
	kfree(conf);
286
	*private_conf = NULL;
287
	return err;
Linus Torvalds's avatar
Linus Torvalds committed
288 289 290 291 292
}

/**
 *	raid0_mergeable_bvec -- tell bio layer if a two requests can be merged
 *	@q: request queue
293
 *	@bvm: properties of new bio
Linus Torvalds's avatar
Linus Torvalds committed
294 295 296 297
 *	@biovec: the request that could be merged to it.
 *
 *	Return amount of bytes we can accept at this offset
 */
298 299 300
static int raid0_mergeable_bvec(struct request_queue *q,
				struct bvec_merge_data *bvm,
				struct bio_vec *biovec)
Linus Torvalds's avatar
Linus Torvalds committed
301 302
{
	mddev_t *mddev = q->queuedata;
303
	sector_t sector = bvm->bi_sector + get_start_sect(bvm->bi_bdev);
Linus Torvalds's avatar
Linus Torvalds committed
304
	int max;
305
	unsigned int chunk_sectors = mddev->chunk_sectors;
306
	unsigned int bio_sectors = bvm->bi_size >> 9;
Linus Torvalds's avatar
Linus Torvalds committed
307

NeilBrown's avatar
NeilBrown committed
308
	if (is_power_of_2(chunk_sectors))
309 310 311 312 313
		max =  (chunk_sectors - ((sector & (chunk_sectors-1))
						+ bio_sectors)) << 9;
	else
		max =  (chunk_sectors - (sector_div(sector, chunk_sectors)
						+ bio_sectors)) << 9;
Linus Torvalds's avatar
Linus Torvalds committed
314 315 316 317 318 319 320
	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;
}

321 322 323 324 325 326 327 328 329 330 331 332 333 334
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;
}

335
static int raid0_run(mddev_t *mddev)
Linus Torvalds's avatar
Linus Torvalds committed
336
{
337
	raid0_conf_t *conf;
338
	int ret;
Linus Torvalds's avatar
Linus Torvalds committed
339

340
	if (mddev->chunk_sectors == 0) {
NeilBrown's avatar
NeilBrown committed
341 342
		printk(KERN_ERR "md/raid0:%s: chunk size must be set.\n",
		       mdname(mddev));
343 344
		return -EINVAL;
	}
345 346
	if (md_check_no_bitmap(mddev))
		return -EINVAL;
347
	blk_queue_max_hw_sectors(mddev->queue, mddev->chunk_sectors);
Linus Torvalds's avatar
Linus Torvalds committed
348

349 350 351 352 353 354 355 356
	/* 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;
Linus Torvalds's avatar
Linus Torvalds committed
357 358

	/* calculate array device size */
359
	md_set_array_sectors(mddev, raid0_size(mddev, 0, 0));
Linus Torvalds's avatar
Linus Torvalds committed
360

NeilBrown's avatar
NeilBrown committed
361 362 363
	printk(KERN_INFO "md/raid0:%s: md_size is %llu sectors.\n",
	       mdname(mddev),
	       (unsigned long long)mddev->array_sectors);
Linus Torvalds's avatar
Linus Torvalds committed
364 365 366 367 368 369 370 371 372 373
	/* 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.
	 */
	{
374 375
		int stripe = mddev->raid_disks *
			(mddev->chunk_sectors << 9) / PAGE_SIZE;
Linus Torvalds's avatar
Linus Torvalds committed
376 377 378 379 380
		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);
381
	dump_zones(mddev);
382
	return md_integrity_register(mddev);
Linus Torvalds's avatar
Linus Torvalds committed
383 384
}

385
static int raid0_stop(mddev_t *mddev)
Linus Torvalds's avatar
Linus Torvalds committed
386
{
387
	raid0_conf_t *conf = mddev->private;
Linus Torvalds's avatar
Linus Torvalds committed
388 389

	blk_sync_queue(mddev->queue); /* the unplug fn references 'conf'*/
390
	kfree(conf->strip_zone);
391
	kfree(conf->devlist);
392
	kfree(conf);
Linus Torvalds's avatar
Linus Torvalds committed
393 394 395 396
	mddev->private = NULL;
	return 0;
}

397 398 399
/* Find the zone which holds a particular offset
 * Update *sectorp to be an offset in that zone
 */
400
static struct strip_zone *find_zone(struct raid0_private_data *conf,
401
				    sector_t *sectorp)
402 403 404
{
	int i;
	struct strip_zone *z = conf->strip_zone;
405
	sector_t sector = *sectorp;
406 407

	for (i = 0; i < conf->nr_strip_zones; i++)
408 409 410
		if (sector < z[i].zone_end) {
			if (i)
				*sectorp = sector - z[i-1].zone_end;
411
			return z + i;
412
		}
413 414 415
	BUG();
}

416 417 418 419 420 421
/*
 * 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)
Linus Torvalds's avatar
Linus Torvalds committed
422
{
423 424
	unsigned int sect_in_chunk;
	sector_t chunk;
425
	raid0_conf_t *conf = mddev->private;
426
	int raid_disks = conf->strip_zone[0].nb_dev;
427
	unsigned int chunk_sects = mddev->chunk_sectors;
428

NeilBrown's avatar
NeilBrown committed
429
	if (is_power_of_2(chunk_sects)) {
430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448
		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;
449
	return conf->devlist[(zone - conf->strip_zone)*raid_disks
450 451 452 453 454 455 456 457 458
			     + 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)
{
NeilBrown's avatar
NeilBrown committed
459
	if (likely(is_power_of_2(chunk_sects))) {
460 461 462 463 464 465 466 467 468
		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));
	}
}

469
static int raid0_make_request(mddev_t *mddev, struct bio *bio)
470 471 472
{
	unsigned int chunk_sects;
	sector_t sector_offset;
Linus Torvalds's avatar
Linus Torvalds committed
473 474 475
	struct strip_zone *zone;
	mdk_rdev_t *tmp_dev;

Tejun Heo's avatar
Tejun Heo committed
476 477
	if (unlikely(bio->bi_rw & REQ_FLUSH)) {
		md_flush_request(mddev, bio);
478 479 480
		return 0;
	}

481
	chunk_sects = mddev->chunk_sectors;
482 483
	if (unlikely(!is_io_in_chunk_boundary(mddev, chunk_sects, bio))) {
		sector_t sector = bio->bi_sector;
Linus Torvalds's avatar
Linus Torvalds committed
484 485 486 487 488 489 490 491
		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.
		 */
NeilBrown's avatar
NeilBrown committed
492
		if (likely(is_power_of_2(chunk_sects)))
493 494 495 496 497
			bp = bio_split(bio, chunk_sects - (sector &
							   (chunk_sects-1)));
		else
			bp = bio_split(bio, chunk_sects -
				       sector_div(sector, chunk_sects));
498
		if (raid0_make_request(mddev, &bp->bio1))
Linus Torvalds's avatar
Linus Torvalds committed
499
			generic_make_request(&bp->bio1);
500
		if (raid0_make_request(mddev, &bp->bio2))
Linus Torvalds's avatar
Linus Torvalds committed
501 502 503 504 505 506
			generic_make_request(&bp->bio2);

		bio_pair_release(bp);
		return 0;
	}

507 508 509 510
	sector_offset = bio->bi_sector;
	zone =  find_zone(mddev->private, &sector_offset);
	tmp_dev = map_sector(mddev, zone, bio->bi_sector,
			     &sector_offset);
Linus Torvalds's avatar
Linus Torvalds committed
511
	bio->bi_bdev = tmp_dev->bdev;
512 513
	bio->bi_sector = sector_offset + zone->dev_start +
		tmp_dev->data_offset;
Linus Torvalds's avatar
Linus Torvalds committed
514 515 516 517 518 519
	/*
	 * Let the main block layer submit the IO and resolve recursion:
	 */
	return 1;

bad_map:
NeilBrown's avatar
NeilBrown committed
520 521 522 523
	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);
Linus Torvalds's avatar
Linus Torvalds committed
524

525
	bio_io_error(bio);
Linus Torvalds's avatar
Linus Torvalds committed
526 527
	return 0;
}
528

529
static void raid0_status(struct seq_file *seq, mddev_t *mddev)
Linus Torvalds's avatar
Linus Torvalds committed
530
{
531
	seq_printf(seq, " %dk chunks", mddev->chunk_sectors / 2);
Linus Torvalds's avatar
Linus Torvalds committed
532 533 534
	return;
}

535
static void *raid0_takeover_raid45(mddev_t *mddev)
536 537 538 539 540
{
	mdk_rdev_t *rdev;
	raid0_conf_t *priv_conf;

	if (mddev->degraded != 1) {
NeilBrown's avatar
NeilBrown committed
541 542
		printk(KERN_ERR "md/raid0:%s: raid5 must be degraded! Degraded disks: %d\n",
		       mdname(mddev),
543 544 545 546 547 548 549
		       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) {
NeilBrown's avatar
NeilBrown committed
550 551
			printk(KERN_ERR "md/raid0:%s: raid5 must have missing parity disk!\n",
			       mdname(mddev));
552 553 554 555 556 557
			return ERR_PTR(-EINVAL);
		}
	}

	/* Set new parameters */
	mddev->new_level = 0;
558
	mddev->new_layout = 0;
559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579
	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)) {
NeilBrown's avatar
NeilBrown committed
580 581
		printk(KERN_ERR "md/raid0:%s:: Raid0 cannot takover layout: 0x%x\n",
		       mdname(mddev),
582 583 584 585
		       mddev->layout);
		return ERR_PTR(-EINVAL);
	}
	if (mddev->raid_disks & 1) {
NeilBrown's avatar
NeilBrown committed
586 587
		printk(KERN_ERR "md/raid0:%s: Raid0 cannot takover Raid10 with odd disk number.\n",
		       mdname(mddev));
588 589 590
		return ERR_PTR(-EINVAL);
	}
	if (mddev->degraded != (mddev->raid_disks>>1)) {
NeilBrown's avatar
NeilBrown committed
591 592
		printk(KERN_ERR "md/raid0:%s: All mirrors must be already degraded!\n",
		       mdname(mddev));
593 594 595 596 597
		return ERR_PTR(-EINVAL);
	}

	/* Set new parameters */
	mddev->new_level = 0;
598
	mddev->new_layout = 0;
599 600 601 602 603 604 605 606 607 608 609
	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;
}

610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627
static void *raid0_takeover_raid1(mddev_t *mddev)
{
	raid0_conf_t *priv_conf;

	/* 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);
	}

	/* Set new parameters */
	mddev->new_level = 0;
	mddev->new_layout = 0;
	mddev->new_chunk_sectors = 128; /* by default set chunk size to 64k */
	mddev->delta_disks = 1 - mddev->raid_disks;
628
	mddev->raid_disks = 1;
629 630 631 632 633 634 635
	/* make sure it will be not marked as dirty */
	mddev->recovery_cp = MaxSector;

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

636 637 638
static void *raid0_takeover(mddev_t *mddev)
{
	/* raid0 can take over:
639
	 *  raid4 - if all data disks are active.
640 641
	 *  raid5 - providing it is Raid4 layout and one disk is faulty
	 *  raid10 - assuming we have all necessary active disks
642
	 *  raid1 - with (N -1) mirror drives faulty
643
	 */
644 645 646
	if (mddev->level == 4)
		return raid0_takeover_raid45(mddev);

647 648
	if (mddev->level == 5) {
		if (mddev->layout == ALGORITHM_PARITY_N)
649
			return raid0_takeover_raid45(mddev);
650

NeilBrown's avatar
NeilBrown committed
651 652
		printk(KERN_ERR "md/raid0:%s: Raid can only takeover Raid5 with layout: %d\n",
		       mdname(mddev), ALGORITHM_PARITY_N);
653 654 655 656 657
	}

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

658 659 660 661 662 663
	if (mddev->level == 1)
		return raid0_takeover_raid1(mddev);

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

664 665 666 667 668 669 670
	return ERR_PTR(-EINVAL);
}

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

671
static struct mdk_personality raid0_personality=
Linus Torvalds's avatar
Linus Torvalds committed
672 673
{
	.name		= "raid0",
674
	.level		= 0,
Linus Torvalds's avatar
Linus Torvalds committed
675 676 677 678 679
	.owner		= THIS_MODULE,
	.make_request	= raid0_make_request,
	.run		= raid0_run,
	.stop		= raid0_stop,
	.status		= raid0_status,
680
	.size		= raid0_size,
681 682
	.takeover	= raid0_takeover,
	.quiesce	= raid0_quiesce,
Linus Torvalds's avatar
Linus Torvalds committed
683 684 685 686
};

static int __init raid0_init (void)
{
687
	return register_md_personality (&raid0_personality);
Linus Torvalds's avatar
Linus Torvalds committed
688 689 690 691
}

static void raid0_exit (void)
{
692
	unregister_md_personality (&raid0_personality);
Linus Torvalds's avatar
Linus Torvalds committed
693 694 695 696 697
}

module_init(raid0_init);
module_exit(raid0_exit);
MODULE_LICENSE("GPL");
698
MODULE_DESCRIPTION("RAID0 (striping) personality for MD");
Linus Torvalds's avatar
Linus Torvalds committed
699
MODULE_ALIAS("md-personality-2"); /* RAID0 */
700
MODULE_ALIAS("md-raid0");
701
MODULE_ALIAS("md-level-0");