raid1.c 60.4 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7 8 9 10 11
/*
 * raid1.c : Multiple Devices driver for Linux
 *
 * Copyright (C) 1999, 2000, 2001 Ingo Molnar, Red Hat
 *
 * Copyright (C) 1996, 1997, 1998 Ingo Molnar, Miguel de Icaza, Gadi Oxman
 *
 * RAID-1 management functions.
 *
 * Better read-balancing code written by Mika Kuoppala <miku@iki.fi>, 2000
 *
12
 * Fixes to reconstruction by Jakob Østergaard" <jakob@ostenfeld.dk>
L
Linus Torvalds 已提交
13 14
 * Various fixes by Neil Brown <neilb@cse.unsw.edu.au>
 *
15 16 17 18 19 20 21 22 23
 * Changes by Peter T. Breuer <ptb@it.uc3m.es> 31/1/2003 to support
 * bitmapped intelligence in resync:
 *
 *      - bitmap marked during normal i/o
 *      - bitmap used to skip nondirty blocks during sync
 *
 * Additions to bitmap code, (C) 2003-2004 Paul Clements, SteelEye Technology:
 * - persistent bitmap code
 *
L
Linus Torvalds 已提交
24 25 26 27 28 29 30 31 32 33
 * 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.
 */

34
#include <linux/slab.h>
35
#include <linux/delay.h>
36 37
#include <linux/blkdev.h>
#include <linux/seq_file.h>
38
#include "md.h"
39 40
#include "raid1.h"
#include "bitmap.h"
41 42 43 44 45 46 47

#define DEBUG 0
#if DEBUG
#define PRINTK(x...) printk(x)
#else
#define PRINTK(x...)
#endif
L
Linus Torvalds 已提交
48 49 50 51 52 53 54 55 56

/*
 * Number of guaranteed r1bios in case of extreme VM load:
 */
#define	NR_RAID1_BIOS 256


static void unplug_slaves(mddev_t *mddev);

57 58
static void allow_barrier(conf_t *conf);
static void lower_barrier(conf_t *conf);
L
Linus Torvalds 已提交
59

A
Al Viro 已提交
60
static void * r1bio_pool_alloc(gfp_t gfp_flags, void *data)
L
Linus Torvalds 已提交
61 62 63 64 65 66
{
	struct pool_info *pi = data;
	r1bio_t *r1_bio;
	int size = offsetof(r1bio_t, bios[pi->raid_disks]);

	/* allocate a r1bio with room for raid_disks entries in the bios array */
67
	r1_bio = kzalloc(size, gfp_flags);
68
	if (!r1_bio && pi->mddev)
L
Linus Torvalds 已提交
69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84
		unplug_slaves(pi->mddev);

	return r1_bio;
}

static void r1bio_pool_free(void *r1_bio, void *data)
{
	kfree(r1_bio);
}

#define RESYNC_BLOCK_SIZE (64*1024)
//#define RESYNC_BLOCK_SIZE PAGE_SIZE
#define RESYNC_SECTORS (RESYNC_BLOCK_SIZE >> 9)
#define RESYNC_PAGES ((RESYNC_BLOCK_SIZE + PAGE_SIZE-1) / PAGE_SIZE)
#define RESYNC_WINDOW (2048*1024)

A
Al Viro 已提交
85
static void * r1buf_pool_alloc(gfp_t gfp_flags, void *data)
L
Linus Torvalds 已提交
86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109
{
	struct pool_info *pi = data;
	struct page *page;
	r1bio_t *r1_bio;
	struct bio *bio;
	int i, j;

	r1_bio = r1bio_pool_alloc(gfp_flags, pi);
	if (!r1_bio) {
		unplug_slaves(pi->mddev);
		return NULL;
	}

	/*
	 * Allocate bios : 1 for reading, n-1 for writing
	 */
	for (j = pi->raid_disks ; j-- ; ) {
		bio = bio_alloc(gfp_flags, RESYNC_PAGES);
		if (!bio)
			goto out_free_bio;
		r1_bio->bios[j] = bio;
	}
	/*
	 * Allocate RESYNC_PAGES data pages and attach them to
110 111 112
	 * the first bio.
	 * If this is a user-requested check/repair, allocate
	 * RESYNC_PAGES for each bio.
L
Linus Torvalds 已提交
113
	 */
114 115 116 117 118 119 120 121 122 123 124 125
	if (test_bit(MD_RECOVERY_REQUESTED, &pi->mddev->recovery))
		j = pi->raid_disks;
	else
		j = 1;
	while(j--) {
		bio = r1_bio->bios[j];
		for (i = 0; i < RESYNC_PAGES; i++) {
			page = alloc_page(gfp_flags);
			if (unlikely(!page))
				goto out_free_pages;

			bio->bi_io_vec[i].bv_page = page;
126
			bio->bi_vcnt = i+1;
127 128 129 130 131 132 133 134
		}
	}
	/* If not user-requests, copy the page pointers to all bios */
	if (!test_bit(MD_RECOVERY_REQUESTED, &pi->mddev->recovery)) {
		for (i=0; i<RESYNC_PAGES ; i++)
			for (j=1; j<pi->raid_disks; j++)
				r1_bio->bios[j]->bi_io_vec[i].bv_page =
					r1_bio->bios[0]->bi_io_vec[i].bv_page;
L
Linus Torvalds 已提交
135 136 137 138 139 140 141
	}

	r1_bio->master_bio = NULL;

	return r1_bio;

out_free_pages:
142 143 144
	for (j=0 ; j < pi->raid_disks; j++)
		for (i=0; i < r1_bio->bios[j]->bi_vcnt ; i++)
			put_page(r1_bio->bios[j]->bi_io_vec[i].bv_page);
145
	j = -1;
L
Linus Torvalds 已提交
146 147 148 149 150 151 152 153 154 155
out_free_bio:
	while ( ++j < pi->raid_disks )
		bio_put(r1_bio->bios[j]);
	r1bio_pool_free(r1_bio, data);
	return NULL;
}

static void r1buf_pool_free(void *__r1_bio, void *data)
{
	struct pool_info *pi = data;
156
	int i,j;
L
Linus Torvalds 已提交
157 158
	r1bio_t *r1bio = __r1_bio;

159 160 161 162 163
	for (i = 0; i < RESYNC_PAGES; i++)
		for (j = pi->raid_disks; j-- ;) {
			if (j == 0 ||
			    r1bio->bios[j]->bi_io_vec[i].bv_page !=
			    r1bio->bios[0]->bi_io_vec[i].bv_page)
164
				safe_put_page(r1bio->bios[j]->bi_io_vec[i].bv_page);
165
		}
L
Linus Torvalds 已提交
166 167 168 169 170 171 172 173 174 175 176 177
	for (i=0 ; i < pi->raid_disks; i++)
		bio_put(r1bio->bios[i]);

	r1bio_pool_free(r1bio, data);
}

static void put_all_bios(conf_t *conf, r1bio_t *r1_bio)
{
	int i;

	for (i = 0; i < conf->raid_disks; i++) {
		struct bio **bio = r1_bio->bios + i;
178
		if (*bio && *bio != IO_BLOCKED)
L
Linus Torvalds 已提交
179 180 181 182 183
			bio_put(*bio);
		*bio = NULL;
	}
}

184
static void free_r1bio(r1bio_t *r1_bio)
L
Linus Torvalds 已提交
185
{
186
	conf_t *conf = r1_bio->mddev->private;
L
Linus Torvalds 已提交
187 188 189 190 191

	/*
	 * Wake up any possible resync thread that waits for the device
	 * to go idle.
	 */
192
	allow_barrier(conf);
L
Linus Torvalds 已提交
193 194 195 196 197

	put_all_bios(conf, r1_bio);
	mempool_free(r1_bio, conf->r1bio_pool);
}

198
static void put_buf(r1bio_t *r1_bio)
L
Linus Torvalds 已提交
199
{
200
	conf_t *conf = r1_bio->mddev->private;
201 202 203 204 205 206 207
	int i;

	for (i=0; i<conf->raid_disks; i++) {
		struct bio *bio = r1_bio->bios[i];
		if (bio->bi_end_io)
			rdev_dec_pending(conf->mirrors[i].rdev, r1_bio->mddev);
	}
L
Linus Torvalds 已提交
208 209 210

	mempool_free(r1_bio, conf->r1buf_pool);

211
	lower_barrier(conf);
L
Linus Torvalds 已提交
212 213 214 215 216 217
}

static void reschedule_retry(r1bio_t *r1_bio)
{
	unsigned long flags;
	mddev_t *mddev = r1_bio->mddev;
218
	conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
219 220 221

	spin_lock_irqsave(&conf->device_lock, flags);
	list_add(&r1_bio->retry_list, &conf->retry_list);
222
	conf->nr_queued ++;
L
Linus Torvalds 已提交
223 224
	spin_unlock_irqrestore(&conf->device_lock, flags);

225
	wake_up(&conf->wait_barrier);
L
Linus Torvalds 已提交
226 227 228 229 230 231 232 233 234 235 236 237
	md_wakeup_thread(mddev->thread);
}

/*
 * raid_end_bio_io() is called when we have finished servicing a mirrored
 * operation and are ready to return a success/failure code to the buffer
 * cache layer.
 */
static void raid_end_bio_io(r1bio_t *r1_bio)
{
	struct bio *bio = r1_bio->master_bio;

238 239 240 241 242 243 244 245
	/* if nobody has done the final endio yet, do it now */
	if (!test_and_set_bit(R1BIO_Returned, &r1_bio->state)) {
		PRINTK(KERN_DEBUG "raid1: sync end %s on sectors %llu-%llu\n",
			(bio_data_dir(bio) == WRITE) ? "write" : "read",
			(unsigned long long) bio->bi_sector,
			(unsigned long long) bio->bi_sector +
				(bio->bi_size >> 9) - 1);

246
		bio_endio(bio,
247 248
			test_bit(R1BIO_Uptodate, &r1_bio->state) ? 0 : -EIO);
	}
L
Linus Torvalds 已提交
249 250 251 252 253 254 255 256
	free_r1bio(r1_bio);
}

/*
 * Update disk head position estimator based on IRQ completion info.
 */
static inline void update_head_pos(int disk, r1bio_t *r1_bio)
{
257
	conf_t *conf = r1_bio->mddev->private;
L
Linus Torvalds 已提交
258 259 260 261 262

	conf->mirrors[disk].head_position =
		r1_bio->sector + (r1_bio->sectors);
}

263
static void raid1_end_read_request(struct bio *bio, int error)
L
Linus Torvalds 已提交
264 265
{
	int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
266
	r1bio_t *r1_bio = bio->bi_private;
L
Linus Torvalds 已提交
267
	int mirror;
268
	conf_t *conf = r1_bio->mddev->private;
L
Linus Torvalds 已提交
269 270 271 272 273

	mirror = r1_bio->read_disk;
	/*
	 * this branch is our 'one mirror IO has finished' event handler:
	 */
274 275
	update_head_pos(mirror, r1_bio);

276 277 278 279 280 281
	if (uptodate)
		set_bit(R1BIO_Uptodate, &r1_bio->state);
	else {
		/* If all other devices have failed, we want to return
		 * the error upwards rather than fail the last device.
		 * Here we redefine "uptodate" to mean "Don't want to retry"
L
Linus Torvalds 已提交
282
		 */
283 284 285 286 287 288 289 290
		unsigned long flags;
		spin_lock_irqsave(&conf->device_lock, flags);
		if (r1_bio->mddev->degraded == conf->raid_disks ||
		    (r1_bio->mddev->degraded == conf->raid_disks-1 &&
		     !test_bit(Faulty, &conf->mirrors[mirror].rdev->flags)))
			uptodate = 1;
		spin_unlock_irqrestore(&conf->device_lock, flags);
	}
L
Linus Torvalds 已提交
291

292
	if (uptodate)
L
Linus Torvalds 已提交
293
		raid_end_bio_io(r1_bio);
294
	else {
L
Linus Torvalds 已提交
295 296 297 298 299
		/*
		 * oops, read error:
		 */
		char b[BDEVNAME_SIZE];
		if (printk_ratelimit())
N
NeilBrown 已提交
300 301
			printk(KERN_ERR "md/raid1:%s: %s: rescheduling sector %llu\n",
			       mdname(conf->mddev),
L
Linus Torvalds 已提交
302 303 304 305 306 307 308
			       bdevname(conf->mirrors[mirror].rdev->bdev,b), (unsigned long long)r1_bio->sector);
		reschedule_retry(r1_bio);
	}

	rdev_dec_pending(conf->mirrors[mirror].rdev, conf->mddev);
}

309
static void raid1_end_write_request(struct bio *bio, int error)
L
Linus Torvalds 已提交
310 311
{
	int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
312
	r1bio_t *r1_bio = bio->bi_private;
313
	int mirror, behind = test_bit(R1BIO_BehindIO, &r1_bio->state);
314
	conf_t *conf = r1_bio->mddev->private;
315
	struct bio *to_put = NULL;
L
Linus Torvalds 已提交
316 317 318 319 320 321


	for (mirror = 0; mirror < conf->raid_disks; mirror++)
		if (r1_bio->bios[mirror] == bio)
			break;

T
Tejun Heo 已提交
322 323 324 325 326 327 328 329 330 331
	/*
	 * 'one mirror IO has finished' event handler:
	 */
	r1_bio->bios[mirror] = NULL;
	to_put = bio;
	if (!uptodate) {
		md_error(r1_bio->mddev, conf->mirrors[mirror].rdev);
		/* an I/O failed, we can't clear the bitmap */
		set_bit(R1BIO_Degraded, &r1_bio->state);
	} else
L
Linus Torvalds 已提交
332
		/*
T
Tejun Heo 已提交
333 334 335 336 337 338 339 340
		 * Set R1BIO_Uptodate in our master bio, so that we
		 * will return a good error code for to the higher
		 * levels even if IO on some other mirrored buffer
		 * fails.
		 *
		 * The 'master' represents the composite IO operation
		 * to user-side. So if something waits for IO, then it
		 * will wait for the 'master' bio.
L
Linus Torvalds 已提交
341
		 */
T
Tejun Heo 已提交
342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366
		set_bit(R1BIO_Uptodate, &r1_bio->state);

	update_head_pos(mirror, r1_bio);

	if (behind) {
		if (test_bit(WriteMostly, &conf->mirrors[mirror].rdev->flags))
			atomic_dec(&r1_bio->behind_remaining);

		/*
		 * In behind mode, we ACK the master bio once the I/O
		 * has safely reached all non-writemostly
		 * disks. Setting the Returned bit ensures that this
		 * gets done only once -- we don't ever want to return
		 * -EIO here, instead we'll wait
		 */
		if (atomic_read(&r1_bio->behind_remaining) >= (atomic_read(&r1_bio->remaining)-1) &&
		    test_bit(R1BIO_Uptodate, &r1_bio->state)) {
			/* Maybe we can return now */
			if (!test_and_set_bit(R1BIO_Returned, &r1_bio->state)) {
				struct bio *mbio = r1_bio->master_bio;
				PRINTK(KERN_DEBUG "raid1: behind end write sectors %llu-%llu\n",
				       (unsigned long long) mbio->bi_sector,
				       (unsigned long long) mbio->bi_sector +
				       (mbio->bi_size >> 9) - 1);
				bio_endio(mbio, 0);
367 368 369
			}
		}
	}
T
Tejun Heo 已提交
370 371
	rdev_dec_pending(conf->mirrors[mirror].rdev, conf->mddev);

L
Linus Torvalds 已提交
372 373 374 375 376
	/*
	 * Let's see if all mirrored write operations have finished
	 * already.
	 */
	if (atomic_dec_and_test(&r1_bio->remaining)) {
T
Tejun Heo 已提交
377 378 379 380 381
		if (test_bit(R1BIO_BehindIO, &r1_bio->state)) {
			/* free extra copy of the data pages */
			int i = bio->bi_vcnt;
			while (i--)
				safe_put_page(bio->bi_io_vec[i].bv_page);
382
		}
T
Tejun Heo 已提交
383 384 385 386 387 388 389
		/* clear the bitmap if all writes complete successfully */
		bitmap_endwrite(r1_bio->mddev->bitmap, r1_bio->sector,
				r1_bio->sectors,
				!test_bit(R1BIO_Degraded, &r1_bio->state),
				behind);
		md_write_end(r1_bio->mddev);
		raid_end_bio_io(r1_bio);
L
Linus Torvalds 已提交
390
	}
391

392 393
	if (to_put)
		bio_put(to_put);
L
Linus Torvalds 已提交
394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412
}


/*
 * This routine returns the disk from which the requested read should
 * be done. There is a per-array 'next expected sequential IO' sector
 * number - if this matches on the next IO then we use the last disk.
 * There is also a per-disk 'last know head position' sector that is
 * maintained from IRQ contexts, both the normal and the resync IO
 * completion handlers update this position correctly. If there is no
 * perfect sequential match then we pick the disk whose head is closest.
 *
 * If there are 2 mirrors in the same 2 devices, performance degrades
 * because position is mirror, not device based.
 *
 * The rdev for the device selected will have nr_pending incremented.
 */
static int read_balance(conf_t *conf, r1bio_t *r1_bio)
{
413
	const sector_t this_sector = r1_bio->sector;
L
Linus Torvalds 已提交
414
	int new_disk = conf->last_used, disk = new_disk;
415
	int wonly_disk = -1;
L
Linus Torvalds 已提交
416 417
	const int sectors = r1_bio->sectors;
	sector_t new_distance, current_distance;
418
	mdk_rdev_t *rdev;
L
Linus Torvalds 已提交
419 420 421

	rcu_read_lock();
	/*
422
	 * Check if we can balance. We can balance on the whole
L
Linus Torvalds 已提交
423 424 425 426 427 428
	 * device if no resync is going on, or below the resync window.
	 * We take the first readable disk when above the resync window.
	 */
 retry:
	if (conf->mddev->recovery_cp < MaxSector &&
	    (this_sector + sectors >= conf->next_resync)) {
429
		/* Choose the first operational device, for consistancy */
L
Linus Torvalds 已提交
430 431
		new_disk = 0;

432
		for (rdev = rcu_dereference(conf->mirrors[new_disk].rdev);
433
		     r1_bio->bios[new_disk] == IO_BLOCKED ||
434
		     !rdev || !test_bit(In_sync, &rdev->flags)
435
			     || test_bit(WriteMostly, &rdev->flags);
436
		     rdev = rcu_dereference(conf->mirrors[++new_disk].rdev)) {
437

438 439
			if (rdev && test_bit(In_sync, &rdev->flags) &&
				r1_bio->bios[new_disk] != IO_BLOCKED)
440 441 442 443
				wonly_disk = new_disk;

			if (new_disk == conf->raid_disks - 1) {
				new_disk = wonly_disk;
L
Linus Torvalds 已提交
444 445 446 447 448 449 450 451
				break;
			}
		}
		goto rb_out;
	}


	/* make sure the disk is operational */
452
	for (rdev = rcu_dereference(conf->mirrors[new_disk].rdev);
453
	     r1_bio->bios[new_disk] == IO_BLOCKED ||
454
	     !rdev || !test_bit(In_sync, &rdev->flags) ||
455
		     test_bit(WriteMostly, &rdev->flags);
456
	     rdev = rcu_dereference(conf->mirrors[new_disk].rdev)) {
457

458 459
		if (rdev && test_bit(In_sync, &rdev->flags) &&
		    r1_bio->bios[new_disk] != IO_BLOCKED)
460 461
			wonly_disk = new_disk;

L
Linus Torvalds 已提交
462 463 464 465
		if (new_disk <= 0)
			new_disk = conf->raid_disks;
		new_disk--;
		if (new_disk == disk) {
466 467
			new_disk = wonly_disk;
			break;
L
Linus Torvalds 已提交
468 469
		}
	}
470 471 472 473

	if (new_disk < 0)
		goto rb_out;

L
Linus Torvalds 已提交
474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493
	disk = new_disk;
	/* now disk == new_disk == starting point for search */

	/*
	 * Don't change to another disk for sequential reads:
	 */
	if (conf->next_seq_sect == this_sector)
		goto rb_out;
	if (this_sector == conf->mirrors[new_disk].head_position)
		goto rb_out;

	current_distance = abs(this_sector - conf->mirrors[disk].head_position);

	/* Find the disk whose head is closest */

	do {
		if (disk <= 0)
			disk = conf->raid_disks;
		disk--;

494
		rdev = rcu_dereference(conf->mirrors[disk].rdev);
495

496
		if (!rdev || r1_bio->bios[disk] == IO_BLOCKED ||
497
		    !test_bit(In_sync, &rdev->flags) ||
498
		    test_bit(WriteMostly, &rdev->flags))
L
Linus Torvalds 已提交
499 500 501 502 503 504 505 506 507 508 509 510 511
			continue;

		if (!atomic_read(&rdev->nr_pending)) {
			new_disk = disk;
			break;
		}
		new_distance = abs(this_sector - conf->mirrors[disk].head_position);
		if (new_distance < current_distance) {
			current_distance = new_distance;
			new_disk = disk;
		}
	} while (disk != conf->last_used);

512
 rb_out:
L
Linus Torvalds 已提交
513 514 515


	if (new_disk >= 0) {
516
		rdev = rcu_dereference(conf->mirrors[new_disk].rdev);
517 518 519
		if (!rdev)
			goto retry;
		atomic_inc(&rdev->nr_pending);
520
		if (!test_bit(In_sync, &rdev->flags)) {
L
Linus Torvalds 已提交
521 522 523
			/* cannot risk returning a device that failed
			 * before we inc'ed nr_pending
			 */
524
			rdev_dec_pending(rdev, conf->mddev);
L
Linus Torvalds 已提交
525 526
			goto retry;
		}
527 528
		conf->next_seq_sect = this_sector + sectors;
		conf->last_used = new_disk;
L
Linus Torvalds 已提交
529 530 531 532 533 534 535 536
	}
	rcu_read_unlock();

	return new_disk;
}

static void unplug_slaves(mddev_t *mddev)
{
537
	conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
538 539 540 541
	int i;

	rcu_read_lock();
	for (i=0; i<mddev->raid_disks; i++) {
542
		mdk_rdev_t *rdev = rcu_dereference(conf->mirrors[i].rdev);
543
		if (rdev && !test_bit(Faulty, &rdev->flags) && atomic_read(&rdev->nr_pending)) {
544
			struct request_queue *r_queue = bdev_get_queue(rdev->bdev);
L
Linus Torvalds 已提交
545 546 547 548

			atomic_inc(&rdev->nr_pending);
			rcu_read_unlock();

549
			blk_unplug(r_queue);
L
Linus Torvalds 已提交
550 551 552 553 554 555 556 557

			rdev_dec_pending(rdev, mddev);
			rcu_read_lock();
		}
	}
	rcu_read_unlock();
}

558
static void raid1_unplug(struct request_queue *q)
L
Linus Torvalds 已提交
559
{
560 561 562 563
	mddev_t *mddev = q->queuedata;

	unplug_slaves(mddev);
	md_wakeup_thread(mddev->thread);
L
Linus Torvalds 已提交
564 565
}

566 567 568
static int raid1_congested(void *data, int bits)
{
	mddev_t *mddev = data;
569
	conf_t *conf = mddev->private;
570 571
	int i, ret = 0;

572 573 574
	if (mddev_congested(mddev, bits))
		return 1;

575 576 577 578
	rcu_read_lock();
	for (i = 0; i < mddev->raid_disks; i++) {
		mdk_rdev_t *rdev = rcu_dereference(conf->mirrors[i].rdev);
		if (rdev && !test_bit(Faulty, &rdev->flags)) {
579
			struct request_queue *q = bdev_get_queue(rdev->bdev);
580 581 582 583

			/* Note the '|| 1' - when read_balance prefers
			 * non-congested targets, it can be removed
			 */
584
			if ((bits & (1<<BDI_async_congested)) || 1)
585 586 587 588 589 590 591 592 593 594
				ret |= bdi_congested(&q->backing_dev_info, bits);
			else
				ret &= bdi_congested(&q->backing_dev_info, bits);
		}
	}
	rcu_read_unlock();
	return ret;
}


595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625
static int flush_pending_writes(conf_t *conf)
{
	/* Any writes that have been queued but are awaiting
	 * bitmap updates get flushed here.
	 * We return 1 if any requests were actually submitted.
	 */
	int rv = 0;

	spin_lock_irq(&conf->device_lock);

	if (conf->pending_bio_list.head) {
		struct bio *bio;
		bio = bio_list_get(&conf->pending_bio_list);
		blk_remove_plug(conf->mddev->queue);
		spin_unlock_irq(&conf->device_lock);
		/* flush any pending bitmap writes to
		 * disk before proceeding w/ I/O */
		bitmap_unplug(conf->mddev->bitmap);

		while (bio) { /* submit pending writes */
			struct bio *next = bio->bi_next;
			bio->bi_next = NULL;
			generic_make_request(bio);
			bio = next;
		}
		rv = 1;
	} else
		spin_unlock_irq(&conf->device_lock);
	return rv;
}

626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645
/* Barriers....
 * Sometimes we need to suspend IO while we do something else,
 * either some resync/recovery, or reconfigure the array.
 * To do this we raise a 'barrier'.
 * The 'barrier' is a counter that can be raised multiple times
 * to count how many activities are happening which preclude
 * normal IO.
 * We can only raise the barrier if there is no pending IO.
 * i.e. if nr_pending == 0.
 * We choose only to raise the barrier if no-one is waiting for the
 * barrier to go down.  This means that as soon as an IO request
 * is ready, no other operations which require a barrier will start
 * until the IO request has had a chance.
 *
 * So: regular IO calls 'wait_barrier'.  When that returns there
 *    is no backgroup IO happening,  It must arrange to call
 *    allow_barrier when it has finished its IO.
 * backgroup IO calls must call raise_barrier.  Once that returns
 *    there is no normal IO happeing.  It must arrange to call
 *    lower_barrier when the particular background IO completes.
L
Linus Torvalds 已提交
646 647 648
 */
#define RESYNC_DEPTH 32

649
static void raise_barrier(conf_t *conf)
L
Linus Torvalds 已提交
650 651
{
	spin_lock_irq(&conf->resync_lock);
652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672

	/* Wait until no block IO is waiting */
	wait_event_lock_irq(conf->wait_barrier, !conf->nr_waiting,
			    conf->resync_lock,
			    raid1_unplug(conf->mddev->queue));

	/* block any new IO from starting */
	conf->barrier++;

	/* No wait for all pending IO to complete */
	wait_event_lock_irq(conf->wait_barrier,
			    !conf->nr_pending && conf->barrier < RESYNC_DEPTH,
			    conf->resync_lock,
			    raid1_unplug(conf->mddev->queue));

	spin_unlock_irq(&conf->resync_lock);
}

static void lower_barrier(conf_t *conf)
{
	unsigned long flags;
673
	BUG_ON(conf->barrier <= 0);
674 675 676 677 678 679 680 681 682 683 684 685 686 687 688
	spin_lock_irqsave(&conf->resync_lock, flags);
	conf->barrier--;
	spin_unlock_irqrestore(&conf->resync_lock, flags);
	wake_up(&conf->wait_barrier);
}

static void wait_barrier(conf_t *conf)
{
	spin_lock_irq(&conf->resync_lock);
	if (conf->barrier) {
		conf->nr_waiting++;
		wait_event_lock_irq(conf->wait_barrier, !conf->barrier,
				    conf->resync_lock,
				    raid1_unplug(conf->mddev->queue));
		conf->nr_waiting--;
L
Linus Torvalds 已提交
689
	}
690
	conf->nr_pending++;
L
Linus Torvalds 已提交
691 692 693
	spin_unlock_irq(&conf->resync_lock);
}

694 695 696 697 698 699 700 701 702
static void allow_barrier(conf_t *conf)
{
	unsigned long flags;
	spin_lock_irqsave(&conf->resync_lock, flags);
	conf->nr_pending--;
	spin_unlock_irqrestore(&conf->resync_lock, flags);
	wake_up(&conf->wait_barrier);
}

703 704 705 706 707
static void freeze_array(conf_t *conf)
{
	/* stop syncio and normal IO and wait for everything to
	 * go quite.
	 * We increment barrier and nr_waiting, and then
708 709 710 711 712 713 714 715
	 * wait until nr_pending match nr_queued+1
	 * This is called in the context of one normal IO request
	 * that has failed. Thus any sync request that might be pending
	 * will be blocked by nr_pending, and we need to wait for
	 * pending IO requests to complete or be queued for re-try.
	 * Thus the number queued (nr_queued) plus this request (1)
	 * must match the number of pending IOs (nr_pending) before
	 * we continue.
716 717 718 719 720
	 */
	spin_lock_irq(&conf->resync_lock);
	conf->barrier++;
	conf->nr_waiting++;
	wait_event_lock_irq(conf->wait_barrier,
721
			    conf->nr_pending == conf->nr_queued+1,
722
			    conf->resync_lock,
723 724
			    ({ flush_pending_writes(conf);
			       raid1_unplug(conf->mddev->queue); }));
725 726 727 728 729 730 731 732 733 734 735 736
	spin_unlock_irq(&conf->resync_lock);
}
static void unfreeze_array(conf_t *conf)
{
	/* reverse the effect of the freeze */
	spin_lock_irq(&conf->resync_lock);
	conf->barrier--;
	conf->nr_waiting--;
	wake_up(&conf->wait_barrier);
	spin_unlock_irq(&conf->resync_lock);
}

737

738 739 740 741 742
/* duplicate the data pages for behind I/O */
static struct page **alloc_behind_pages(struct bio *bio)
{
	int i;
	struct bio_vec *bvec;
743
	struct page **pages = kzalloc(bio->bi_vcnt * sizeof(struct page *),
744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762
					GFP_NOIO);
	if (unlikely(!pages))
		goto do_sync_io;

	bio_for_each_segment(bvec, bio, i) {
		pages[i] = alloc_page(GFP_NOIO);
		if (unlikely(!pages[i]))
			goto do_sync_io;
		memcpy(kmap(pages[i]) + bvec->bv_offset,
			kmap(bvec->bv_page) + bvec->bv_offset, bvec->bv_len);
		kunmap(pages[i]);
		kunmap(bvec->bv_page);
	}

	return pages;

do_sync_io:
	if (pages)
		for (i = 0; i < bio->bi_vcnt && pages[i]; i++)
763
			put_page(pages[i]);
764 765 766 767 768
	kfree(pages);
	PRINTK("%dB behind alloc failed, doing sync I/O\n", bio->bi_size);
	return NULL;
}

769
static int make_request(mddev_t *mddev, struct bio * bio)
L
Linus Torvalds 已提交
770
{
771
	conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
772 773 774
	mirror_info_t *mirror;
	r1bio_t *r1_bio;
	struct bio *read_bio;
775
	int i, targets = 0, disks;
776
	struct bitmap *bitmap;
777 778
	unsigned long flags;
	struct bio_list bl;
779
	struct page **behind_pages = NULL;
780
	const int rw = bio_data_dir(bio);
781
	const unsigned long do_sync = (bio->bi_rw & REQ_SYNC);
T
Tejun Heo 已提交
782
	const unsigned long do_flush_fua = (bio->bi_rw & (REQ_FLUSH | REQ_FUA));
783
	mdk_rdev_t *blocked_rdev;
784

L
Linus Torvalds 已提交
785 786 787 788 789
	/*
	 * Register the new request and wait if the reconstruction
	 * thread has put up a bar for new requests.
	 * Continue immediately if no resync is active currently.
	 */
790

791 792
	md_write_start(mddev, bio); /* wait on superblock update early */

793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811
	if (bio_data_dir(bio) == WRITE &&
	    bio->bi_sector + bio->bi_size/512 > mddev->suspend_lo &&
	    bio->bi_sector < mddev->suspend_hi) {
		/* As the suspend_* range is controlled by
		 * userspace, we want an interruptible
		 * wait.
		 */
		DEFINE_WAIT(w);
		for (;;) {
			flush_signals(current);
			prepare_to_wait(&conf->wait_barrier,
					&w, TASK_INTERRUPTIBLE);
			if (bio->bi_sector + bio->bi_size/512 <= mddev->suspend_lo ||
			    bio->bi_sector >= mddev->suspend_hi)
				break;
			schedule();
		}
		finish_wait(&conf->wait_barrier, &w);
	}
812

813
	wait_barrier(conf);
L
Linus Torvalds 已提交
814

815 816
	bitmap = mddev->bitmap;

L
Linus Torvalds 已提交
817 818 819 820 821 822 823 824 825
	/*
	 * make_request() can abort the operation when READA is being
	 * used and no empty request is available.
	 *
	 */
	r1_bio = mempool_alloc(conf->r1bio_pool, GFP_NOIO);

	r1_bio->master_bio = bio;
	r1_bio->sectors = bio->bi_size >> 9;
826
	r1_bio->state = 0;
L
Linus Torvalds 已提交
827 828 829
	r1_bio->mddev = mddev;
	r1_bio->sector = bio->bi_sector;

830
	if (rw == READ) {
L
Linus Torvalds 已提交
831 832 833 834 835 836 837 838 839 840 841 842
		/*
		 * read balancing logic:
		 */
		int rdisk = read_balance(conf, r1_bio);

		if (rdisk < 0) {
			/* couldn't find anywhere to read from */
			raid_end_bio_io(r1_bio);
			return 0;
		}
		mirror = conf->mirrors + rdisk;

843 844 845 846 847 848 849 850 851
		if (test_bit(WriteMostly, &mirror->rdev->flags) &&
		    bitmap) {
			/* Reading from a write-mostly device must
			 * take care not to over-take any writes
			 * that are 'behind'
			 */
			wait_event(bitmap->behind_wait,
				   atomic_read(&bitmap->behind_writes) == 0);
		}
L
Linus Torvalds 已提交
852 853 854 855 856 857 858 859 860
		r1_bio->read_disk = rdisk;

		read_bio = bio_clone(bio, GFP_NOIO);

		r1_bio->bios[rdisk] = read_bio;

		read_bio->bi_sector = r1_bio->sector + mirror->rdev->data_offset;
		read_bio->bi_bdev = mirror->rdev->bdev;
		read_bio->bi_end_io = raid1_end_read_request;
861
		read_bio->bi_rw = READ | do_sync;
L
Linus Torvalds 已提交
862 863 864 865 866 867 868 869 870 871 872 873 874 875
		read_bio->bi_private = r1_bio;

		generic_make_request(read_bio);
		return 0;
	}

	/*
	 * WRITE:
	 */
	/* first select target devices under spinlock and
	 * inc refcount on their rdev.  Record them by setting
	 * bios[x] to bio
	 */
	disks = conf->raid_disks;
876 877 878 879 880 881 882
#if 0
	{ static int first=1;
	if (first) printk("First Write sector %llu disks %d\n",
			  (unsigned long long)r1_bio->sector, disks);
	first = 0;
	}
#endif
883 884
 retry_write:
	blocked_rdev = NULL;
L
Linus Torvalds 已提交
885 886
	rcu_read_lock();
	for (i = 0;  i < disks; i++) {
887 888 889 890 891 892 893
		mdk_rdev_t *rdev = rcu_dereference(conf->mirrors[i].rdev);
		if (rdev && unlikely(test_bit(Blocked, &rdev->flags))) {
			atomic_inc(&rdev->nr_pending);
			blocked_rdev = rdev;
			break;
		}
		if (rdev && !test_bit(Faulty, &rdev->flags)) {
L
Linus Torvalds 已提交
894
			atomic_inc(&rdev->nr_pending);
895
			if (test_bit(Faulty, &rdev->flags)) {
896
				rdev_dec_pending(rdev, mddev);
L
Linus Torvalds 已提交
897
				r1_bio->bios[i] = NULL;
898
			} else {
L
Linus Torvalds 已提交
899
				r1_bio->bios[i] = bio;
900 901
				targets++;
			}
L
Linus Torvalds 已提交
902 903 904 905 906
		} else
			r1_bio->bios[i] = NULL;
	}
	rcu_read_unlock();

907 908 909 910 911 912 913 914 915 916 917 918 919 920
	if (unlikely(blocked_rdev)) {
		/* Wait for this device to become unblocked */
		int j;

		for (j = 0; j < i; j++)
			if (r1_bio->bios[j])
				rdev_dec_pending(conf->mirrors[j].rdev, mddev);

		allow_barrier(conf);
		md_wait_for_blocked_rdev(blocked_rdev, mddev);
		wait_barrier(conf);
		goto retry_write;
	}

921 922
	BUG_ON(targets == 0); /* we never fail the last device */

923 924 925 926 927 928
	if (targets < conf->raid_disks) {
		/* array is degraded, we will not clear the bitmap
		 * on I/O completion (see raid1_end_write_request) */
		set_bit(R1BIO_Degraded, &r1_bio->state);
	}

929 930 931 932
	/* do behind I/O ?
	 * Not if there are too many, or cannot allocate memory,
	 * or a reader on WriteMostly is waiting for behind writes 
	 * to flush */
933
	if (bitmap &&
934 935
	    (atomic_read(&bitmap->behind_writes)
	     < mddev->bitmap_info.max_write_behind) &&
936
	    !waitqueue_active(&bitmap->behind_wait) &&
937 938 939
	    (behind_pages = alloc_behind_pages(bio)) != NULL)
		set_bit(R1BIO_BehindIO, &r1_bio->state);

940
	atomic_set(&r1_bio->remaining, 0);
941
	atomic_set(&r1_bio->behind_remaining, 0);
942

943
	bio_list_init(&bl);
L
Linus Torvalds 已提交
944 945 946 947 948 949 950 951 952 953 954
	for (i = 0; i < disks; i++) {
		struct bio *mbio;
		if (!r1_bio->bios[i])
			continue;

		mbio = bio_clone(bio, GFP_NOIO);
		r1_bio->bios[i] = mbio;

		mbio->bi_sector	= r1_bio->sector + conf->mirrors[i].rdev->data_offset;
		mbio->bi_bdev = conf->mirrors[i].rdev->bdev;
		mbio->bi_end_io	= raid1_end_write_request;
T
Tejun Heo 已提交
955
		mbio->bi_rw = WRITE | do_flush_fua | do_sync;
L
Linus Torvalds 已提交
956 957
		mbio->bi_private = r1_bio;

958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974
		if (behind_pages) {
			struct bio_vec *bvec;
			int j;

			/* Yes, I really want the '__' version so that
			 * we clear any unused pointer in the io_vec, rather
			 * than leave them unchanged.  This is important
			 * because when we come to free the pages, we won't
			 * know the originial bi_idx, so we just free
			 * them all
			 */
			__bio_for_each_segment(bvec, mbio, j, 0)
				bvec->bv_page = behind_pages[j];
			if (test_bit(WriteMostly, &conf->mirrors[i].rdev->flags))
				atomic_inc(&r1_bio->behind_remaining);
		}

L
Linus Torvalds 已提交
975 976
		atomic_inc(&r1_bio->remaining);

977
		bio_list_add(&bl, mbio);
L
Linus Torvalds 已提交
978
	}
979
	kfree(behind_pages); /* the behind pages are attached to the bios now */
L
Linus Torvalds 已提交
980

981 982
	bitmap_startwrite(bitmap, bio->bi_sector, r1_bio->sectors,
				test_bit(R1BIO_BehindIO, &r1_bio->state));
983 984 985 986 987 988 989
	spin_lock_irqsave(&conf->device_lock, flags);
	bio_list_merge(&conf->pending_bio_list, &bl);
	bio_list_init(&bl);

	blk_plug_device(mddev->queue);
	spin_unlock_irqrestore(&conf->device_lock, flags);

990 991 992
	/* In case raid1d snuck into freeze_array */
	wake_up(&conf->wait_barrier);

993 994
	if (do_sync)
		md_wakeup_thread(mddev->thread);
995 996 997 998 999
#if 0
	while ((bio = bio_list_pop(&bl)) != NULL)
		generic_make_request(bio);
#endif

L
Linus Torvalds 已提交
1000 1001 1002 1003 1004
	return 0;
}

static void status(struct seq_file *seq, mddev_t *mddev)
{
1005
	conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
1006 1007 1008
	int i;

	seq_printf(seq, " [%d/%d] [", conf->raid_disks,
1009
		   conf->raid_disks - mddev->degraded);
1010 1011 1012
	rcu_read_lock();
	for (i = 0; i < conf->raid_disks; i++) {
		mdk_rdev_t *rdev = rcu_dereference(conf->mirrors[i].rdev);
L
Linus Torvalds 已提交
1013
		seq_printf(seq, "%s",
1014 1015 1016
			   rdev && test_bit(In_sync, &rdev->flags) ? "U" : "_");
	}
	rcu_read_unlock();
L
Linus Torvalds 已提交
1017 1018 1019 1020 1021 1022 1023
	seq_printf(seq, "]");
}


static void error(mddev_t *mddev, mdk_rdev_t *rdev)
{
	char b[BDEVNAME_SIZE];
1024
	conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
1025 1026 1027 1028 1029 1030 1031

	/*
	 * If it is not operational, then we have already marked it as dead
	 * else if it is the last working disks, ignore the error, let the
	 * next level up know.
	 * else mark the drive as failed
	 */
1032
	if (test_bit(In_sync, &rdev->flags)
1033
	    && (conf->raid_disks - mddev->degraded) == 1) {
L
Linus Torvalds 已提交
1034 1035
		/*
		 * Don't fail the drive, act as though we were just a
1036 1037 1038
		 * normal single drive.
		 * However don't try a recovery from this drive as
		 * it is very likely to fail.
L
Linus Torvalds 已提交
1039
		 */
1040
		mddev->recovery_disabled = 1;
L
Linus Torvalds 已提交
1041
		return;
1042
	}
1043 1044 1045
	if (test_and_clear_bit(In_sync, &rdev->flags)) {
		unsigned long flags;
		spin_lock_irqsave(&conf->device_lock, flags);
L
Linus Torvalds 已提交
1046
		mddev->degraded++;
1047
		set_bit(Faulty, &rdev->flags);
1048
		spin_unlock_irqrestore(&conf->device_lock, flags);
L
Linus Torvalds 已提交
1049 1050 1051
		/*
		 * if recovery is running, make sure it aborts.
		 */
1052
		set_bit(MD_RECOVERY_INTR, &mddev->recovery);
1053 1054
	} else
		set_bit(Faulty, &rdev->flags);
1055
	set_bit(MD_CHANGE_DEVS, &mddev->flags);
N
NeilBrown 已提交
1056 1057 1058 1059
	printk(KERN_ALERT "md/raid1:%s: Disk failure on %s, disabling device.\n"
	       KERN_ALERT "md/raid1:%s: Operation continuing on %d devices.\n",
	       mdname(mddev), bdevname(rdev->bdev, b),
	       mdname(mddev), conf->raid_disks - mddev->degraded);
L
Linus Torvalds 已提交
1060 1061 1062 1063 1064 1065
}

static void print_conf(conf_t *conf)
{
	int i;

N
NeilBrown 已提交
1066
	printk(KERN_DEBUG "RAID1 conf printout:\n");
L
Linus Torvalds 已提交
1067
	if (!conf) {
N
NeilBrown 已提交
1068
		printk(KERN_DEBUG "(!conf)\n");
L
Linus Torvalds 已提交
1069 1070
		return;
	}
N
NeilBrown 已提交
1071
	printk(KERN_DEBUG " --- wd:%d rd:%d\n", conf->raid_disks - conf->mddev->degraded,
L
Linus Torvalds 已提交
1072 1073
		conf->raid_disks);

1074
	rcu_read_lock();
L
Linus Torvalds 已提交
1075 1076
	for (i = 0; i < conf->raid_disks; i++) {
		char b[BDEVNAME_SIZE];
1077 1078
		mdk_rdev_t *rdev = rcu_dereference(conf->mirrors[i].rdev);
		if (rdev)
N
NeilBrown 已提交
1079
			printk(KERN_DEBUG " disk %d, wo:%d, o:%d, dev:%s\n",
1080 1081 1082
			       i, !test_bit(In_sync, &rdev->flags),
			       !test_bit(Faulty, &rdev->flags),
			       bdevname(rdev->bdev,b));
L
Linus Torvalds 已提交
1083
	}
1084
	rcu_read_unlock();
L
Linus Torvalds 已提交
1085 1086 1087 1088
}

static void close_sync(conf_t *conf)
{
1089 1090
	wait_barrier(conf);
	allow_barrier(conf);
L
Linus Torvalds 已提交
1091 1092 1093 1094 1095 1096 1097 1098 1099

	mempool_destroy(conf->r1buf_pool);
	conf->r1buf_pool = NULL;
}

static int raid1_spare_active(mddev_t *mddev)
{
	int i;
	conf_t *conf = mddev->private;
1100 1101
	int count = 0;
	unsigned long flags;
L
Linus Torvalds 已提交
1102 1103 1104

	/*
	 * Find all failed disks within the RAID1 configuration 
1105 1106
	 * and mark them readable.
	 * Called under mddev lock, so rcu protection not needed.
L
Linus Torvalds 已提交
1107 1108
	 */
	for (i = 0; i < conf->raid_disks; i++) {
1109 1110 1111
		mdk_rdev_t *rdev = conf->mirrors[i].rdev;
		if (rdev
		    && !test_bit(Faulty, &rdev->flags)
1112
		    && !test_and_set_bit(In_sync, &rdev->flags)) {
1113
			count++;
1114
			sysfs_notify_dirent(rdev->sysfs_state);
L
Linus Torvalds 已提交
1115 1116
		}
	}
1117 1118 1119
	spin_lock_irqsave(&conf->device_lock, flags);
	mddev->degraded -= count;
	spin_unlock_irqrestore(&conf->device_lock, flags);
L
Linus Torvalds 已提交
1120 1121

	print_conf(conf);
1122
	return count;
L
Linus Torvalds 已提交
1123 1124 1125 1126 1127 1128
}


static int raid1_add_disk(mddev_t *mddev, mdk_rdev_t *rdev)
{
	conf_t *conf = mddev->private;
1129
	int err = -EEXIST;
1130
	int mirror = 0;
L
Linus Torvalds 已提交
1131
	mirror_info_t *p;
1132 1133
	int first = 0;
	int last = mddev->raid_disks - 1;
L
Linus Torvalds 已提交
1134

1135 1136 1137 1138
	if (rdev->raid_disk >= 0)
		first = last = rdev->raid_disk;

	for (mirror = first; mirror <= last; mirror++)
L
Linus Torvalds 已提交
1139 1140
		if ( !(p=conf->mirrors+mirror)->rdev) {

1141 1142
			disk_stack_limits(mddev->gendisk, rdev->bdev,
					  rdev->data_offset << 9);
1143 1144 1145 1146 1147
			/* as we don't honour merge_bvec_fn, we must
			 * never risk violating it, so limit
			 * ->max_segments to one lying with a single
			 * page, as a one page request is never in
			 * violation.
L
Linus Torvalds 已提交
1148
			 */
1149 1150 1151 1152 1153
			if (rdev->bdev->bd_disk->queue->merge_bvec_fn) {
				blk_queue_max_segments(mddev->queue, 1);
				blk_queue_segment_boundary(mddev->queue,
							   PAGE_CACHE_SIZE - 1);
			}
L
Linus Torvalds 已提交
1154 1155 1156

			p->head_position = 0;
			rdev->raid_disk = mirror;
1157
			err = 0;
1158 1159 1160 1161
			/* As all devices are equivalent, we don't need a full recovery
			 * if this was recently any drive of the array
			 */
			if (rdev->saved_raid_disk < 0)
1162
				conf->fullsync = 1;
1163
			rcu_assign_pointer(p->rdev, rdev);
L
Linus Torvalds 已提交
1164 1165
			break;
		}
1166
	md_integrity_add_rdev(rdev, mddev);
L
Linus Torvalds 已提交
1167
	print_conf(conf);
1168
	return err;
L
Linus Torvalds 已提交
1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180
}

static int raid1_remove_disk(mddev_t *mddev, int number)
{
	conf_t *conf = mddev->private;
	int err = 0;
	mdk_rdev_t *rdev;
	mirror_info_t *p = conf->mirrors+ number;

	print_conf(conf);
	rdev = p->rdev;
	if (rdev) {
1181
		if (test_bit(In_sync, &rdev->flags) ||
L
Linus Torvalds 已提交
1182 1183 1184 1185
		    atomic_read(&rdev->nr_pending)) {
			err = -EBUSY;
			goto abort;
		}
1186 1187 1188 1189 1190 1191 1192 1193
		/* Only remove non-faulty devices is recovery
		 * is not possible.
		 */
		if (!test_bit(Faulty, &rdev->flags) &&
		    mddev->degraded < conf->raid_disks) {
			err = -EBUSY;
			goto abort;
		}
L
Linus Torvalds 已提交
1194
		p->rdev = NULL;
1195
		synchronize_rcu();
L
Linus Torvalds 已提交
1196 1197 1198 1199
		if (atomic_read(&rdev->nr_pending)) {
			/* lost the race, try later */
			err = -EBUSY;
			p->rdev = rdev;
1200
			goto abort;
L
Linus Torvalds 已提交
1201
		}
1202
		md_integrity_register(mddev);
L
Linus Torvalds 已提交
1203 1204 1205 1206 1207 1208 1209 1210
	}
abort:

	print_conf(conf);
	return err;
}


1211
static void end_sync_read(struct bio *bio, int error)
L
Linus Torvalds 已提交
1212
{
1213
	r1bio_t *r1_bio = bio->bi_private;
1214
	int i;
L
Linus Torvalds 已提交
1215

1216 1217 1218 1219 1220
	for (i=r1_bio->mddev->raid_disks; i--; )
		if (r1_bio->bios[i] == bio)
			break;
	BUG_ON(i < 0);
	update_head_pos(i, r1_bio);
L
Linus Torvalds 已提交
1221 1222 1223 1224 1225
	/*
	 * we have read a block, now it needs to be re-written,
	 * or re-read if the read failed.
	 * We don't do much here, just schedule handling by raid1d
	 */
1226
	if (test_bit(BIO_UPTODATE, &bio->bi_flags))
L
Linus Torvalds 已提交
1227
		set_bit(R1BIO_Uptodate, &r1_bio->state);
1228 1229 1230

	if (atomic_dec_and_test(&r1_bio->remaining))
		reschedule_retry(r1_bio);
L
Linus Torvalds 已提交
1231 1232
}

1233
static void end_sync_write(struct bio *bio, int error)
L
Linus Torvalds 已提交
1234 1235
{
	int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
1236
	r1bio_t *r1_bio = bio->bi_private;
L
Linus Torvalds 已提交
1237
	mddev_t *mddev = r1_bio->mddev;
1238
	conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
1239 1240 1241 1242 1243 1244 1245 1246
	int i;
	int mirror=0;

	for (i = 0; i < conf->raid_disks; i++)
		if (r1_bio->bios[i] == bio) {
			mirror = i;
			break;
		}
1247
	if (!uptodate) {
N
NeilBrown 已提交
1248
		sector_t sync_blocks = 0;
1249 1250 1251 1252
		sector_t s = r1_bio->sector;
		long sectors_to_go = r1_bio->sectors;
		/* make sure these bits doesn't get cleared. */
		do {
1253
			bitmap_end_sync(mddev->bitmap, s,
1254 1255 1256 1257
					&sync_blocks, 1);
			s += sync_blocks;
			sectors_to_go -= sync_blocks;
		} while (sectors_to_go > 0);
L
Linus Torvalds 已提交
1258
		md_error(mddev, conf->mirrors[mirror].rdev);
1259
	}
1260

L
Linus Torvalds 已提交
1261 1262 1263
	update_head_pos(mirror, r1_bio);

	if (atomic_dec_and_test(&r1_bio->remaining)) {
1264
		sector_t s = r1_bio->sectors;
L
Linus Torvalds 已提交
1265
		put_buf(r1_bio);
1266
		md_done_sync(mddev, s, uptodate);
L
Linus Torvalds 已提交
1267 1268 1269 1270 1271
	}
}

static void sync_request_write(mddev_t *mddev, r1bio_t *r1_bio)
{
1272
	conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
1273 1274 1275 1276 1277 1278
	int i;
	int disks = conf->raid_disks;
	struct bio *bio, *wbio;

	bio = r1_bio->bios[r1_bio->read_disk];

1279

1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301
	if (test_bit(MD_RECOVERY_REQUESTED, &mddev->recovery)) {
		/* We have read all readable devices.  If we haven't
		 * got the block, then there is no hope left.
		 * If we have, then we want to do a comparison
		 * and skip the write if everything is the same.
		 * If any blocks failed to read, then we need to
		 * attempt an over-write
		 */
		int primary;
		if (!test_bit(R1BIO_Uptodate, &r1_bio->state)) {
			for (i=0; i<mddev->raid_disks; i++)
				if (r1_bio->bios[i]->bi_end_io == end_sync_read)
					md_error(mddev, conf->mirrors[i].rdev);

			md_done_sync(mddev, r1_bio->sectors, 1);
			put_buf(r1_bio);
			return;
		}
		for (primary=0; primary<mddev->raid_disks; primary++)
			if (r1_bio->bios[primary]->bi_end_io == end_sync_read &&
			    test_bit(BIO_UPTODATE, &r1_bio->bios[primary]->bi_flags)) {
				r1_bio->bios[primary]->bi_end_io = NULL;
1302
				rdev_dec_pending(conf->mirrors[primary].rdev, mddev);
1303 1304 1305 1306
				break;
			}
		r1_bio->read_disk = primary;
		for (i=0; i<mddev->raid_disks; i++)
1307
			if (r1_bio->bios[i]->bi_end_io == end_sync_read) {
1308 1309 1310 1311
				int j;
				int vcnt = r1_bio->sectors >> (PAGE_SHIFT- 9);
				struct bio *pbio = r1_bio->bios[primary];
				struct bio *sbio = r1_bio->bios[i];
1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324

				if (test_bit(BIO_UPTODATE, &sbio->bi_flags)) {
					for (j = vcnt; j-- ; ) {
						struct page *p, *s;
						p = pbio->bi_io_vec[j].bv_page;
						s = sbio->bi_io_vec[j].bv_page;
						if (memcmp(page_address(p),
							   page_address(s),
							   PAGE_SIZE))
							break;
					}
				} else
					j = 0;
1325 1326
				if (j >= 0)
					mddev->resync_mismatches += r1_bio->sectors;
1327 1328
				if (j < 0 || (test_bit(MD_RECOVERY_CHECK, &mddev->recovery)
					      && test_bit(BIO_UPTODATE, &sbio->bi_flags))) {
1329
					sbio->bi_end_io = NULL;
1330 1331
					rdev_dec_pending(conf->mirrors[i].rdev, mddev);
				} else {
1332
					/* fixup the bio for reuse */
1333
					int size;
1334 1335 1336 1337 1338 1339 1340 1341 1342 1343
					sbio->bi_vcnt = vcnt;
					sbio->bi_size = r1_bio->sectors << 9;
					sbio->bi_idx = 0;
					sbio->bi_phys_segments = 0;
					sbio->bi_flags &= ~(BIO_POOL_MASK - 1);
					sbio->bi_flags |= 1 << BIO_UPTODATE;
					sbio->bi_next = NULL;
					sbio->bi_sector = r1_bio->sector +
						conf->mirrors[i].rdev->data_offset;
					sbio->bi_bdev = conf->mirrors[i].rdev->bdev;
1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354
					size = sbio->bi_size;
					for (j = 0; j < vcnt ; j++) {
						struct bio_vec *bi;
						bi = &sbio->bi_io_vec[j];
						bi->bv_offset = 0;
						if (size > PAGE_SIZE)
							bi->bv_len = PAGE_SIZE;
						else
							bi->bv_len = size;
						size -= PAGE_SIZE;
						memcpy(page_address(bi->bv_page),
1355 1356
						       page_address(pbio->bi_io_vec[j].bv_page),
						       PAGE_SIZE);
1357
					}
1358

1359 1360 1361
				}
			}
	}
L
Linus Torvalds 已提交
1362
	if (!test_bit(R1BIO_Uptodate, &r1_bio->state)) {
1363 1364 1365
		/* ouch - failed to read all of that.
		 * Try some synchronous reads of other devices to get
		 * good data, much like with normal read errors.  Only
1366
		 * read into the pages we already have so we don't
1367 1368 1369 1370
		 * need to re-issue the read request.
		 * We don't need to freeze the array, because being in an
		 * active sync request, there is no normal IO, and
		 * no overlapping syncs.
L
Linus Torvalds 已提交
1371
		 */
1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385
		sector_t sect = r1_bio->sector;
		int sectors = r1_bio->sectors;
		int idx = 0;

		while(sectors) {
			int s = sectors;
			int d = r1_bio->read_disk;
			int success = 0;
			mdk_rdev_t *rdev;

			if (s > (PAGE_SIZE>>9))
				s = PAGE_SIZE >> 9;
			do {
				if (r1_bio->bios[d]->bi_end_io == end_sync_read) {
1386 1387 1388 1389
					/* No rcu protection needed here devices
					 * can only be removed when no resync is
					 * active, and resync is currently active
					 */
1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405
					rdev = conf->mirrors[d].rdev;
					if (sync_page_io(rdev->bdev,
							 sect + rdev->data_offset,
							 s<<9,
							 bio->bi_io_vec[idx].bv_page,
							 READ)) {
						success = 1;
						break;
					}
				}
				d++;
				if (d == conf->raid_disks)
					d = 0;
			} while (!success && d != r1_bio->read_disk);

			if (success) {
1406
				int start = d;
1407 1408 1409 1410 1411 1412 1413 1414 1415
				/* write it back and re-read */
				set_bit(R1BIO_Uptodate, &r1_bio->state);
				while (d != r1_bio->read_disk) {
					if (d == 0)
						d = conf->raid_disks;
					d--;
					if (r1_bio->bios[d]->bi_end_io != end_sync_read)
						continue;
					rdev = conf->mirrors[d].rdev;
1416
					atomic_add(s, &rdev->corrected_errors);
1417 1418 1419 1420
					if (sync_page_io(rdev->bdev,
							 sect + rdev->data_offset,
							 s<<9,
							 bio->bi_io_vec[idx].bv_page,
1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432
							 WRITE) == 0)
						md_error(mddev, rdev);
				}
				d = start;
				while (d != r1_bio->read_disk) {
					if (d == 0)
						d = conf->raid_disks;
					d--;
					if (r1_bio->bios[d]->bi_end_io != end_sync_read)
						continue;
					rdev = conf->mirrors[d].rdev;
					if (sync_page_io(rdev->bdev,
1433 1434 1435
							 sect + rdev->data_offset,
							 s<<9,
							 bio->bi_io_vec[idx].bv_page,
1436
							 READ) == 0)
1437 1438 1439 1440 1441 1442
						md_error(mddev, rdev);
				}
			} else {
				char b[BDEVNAME_SIZE];
				/* Cannot read from anywhere, array is toast */
				md_error(mddev, conf->mirrors[r1_bio->read_disk].rdev);
N
NeilBrown 已提交
1443
				printk(KERN_ALERT "md/raid1:%s: %s: unrecoverable I/O read error"
1444
				       " for block %llu\n",
N
NeilBrown 已提交
1445 1446
				       mdname(mddev),
				       bdevname(bio->bi_bdev, b),
1447 1448 1449 1450 1451 1452 1453 1454 1455
				       (unsigned long long)r1_bio->sector);
				md_done_sync(mddev, r1_bio->sectors, 0);
				put_buf(r1_bio);
				return;
			}
			sectors -= s;
			sect += s;
			idx ++;
		}
L
Linus Torvalds 已提交
1456
	}
1457 1458 1459 1460

	/*
	 * schedule writes
	 */
L
Linus Torvalds 已提交
1461 1462 1463
	atomic_set(&r1_bio->remaining, 1);
	for (i = 0; i < disks ; i++) {
		wbio = r1_bio->bios[i];
1464 1465 1466 1467
		if (wbio->bi_end_io == NULL ||
		    (wbio->bi_end_io == end_sync_read &&
		     (i == r1_bio->read_disk ||
		      !test_bit(MD_RECOVERY_SYNC, &mddev->recovery))))
L
Linus Torvalds 已提交
1468 1469
			continue;

1470 1471
		wbio->bi_rw = WRITE;
		wbio->bi_end_io = end_sync_write;
L
Linus Torvalds 已提交
1472 1473
		atomic_inc(&r1_bio->remaining);
		md_sync_acct(conf->mirrors[i].rdev->bdev, wbio->bi_size >> 9);
1474

L
Linus Torvalds 已提交
1475 1476 1477 1478
		generic_make_request(wbio);
	}

	if (atomic_dec_and_test(&r1_bio->remaining)) {
1479
		/* if we're here, all write(s) have completed, so clean up */
L
Linus Torvalds 已提交
1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492
		md_done_sync(mddev, r1_bio->sectors, 1);
		put_buf(r1_bio);
	}
}

/*
 * This is a kernel thread which:
 *
 *	1.	Retries failed read operations on working mirrors.
 *	2.	Updates the raid superblock when problems encounter.
 *	3.	Performs writes following reads for array syncronising.
 */

1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567
static void fix_read_error(conf_t *conf, int read_disk,
			   sector_t sect, int sectors)
{
	mddev_t *mddev = conf->mddev;
	while(sectors) {
		int s = sectors;
		int d = read_disk;
		int success = 0;
		int start;
		mdk_rdev_t *rdev;

		if (s > (PAGE_SIZE>>9))
			s = PAGE_SIZE >> 9;

		do {
			/* Note: no rcu protection needed here
			 * as this is synchronous in the raid1d thread
			 * which is the thread that might remove
			 * a device.  If raid1d ever becomes multi-threaded....
			 */
			rdev = conf->mirrors[d].rdev;
			if (rdev &&
			    test_bit(In_sync, &rdev->flags) &&
			    sync_page_io(rdev->bdev,
					 sect + rdev->data_offset,
					 s<<9,
					 conf->tmppage, READ))
				success = 1;
			else {
				d++;
				if (d == conf->raid_disks)
					d = 0;
			}
		} while (!success && d != read_disk);

		if (!success) {
			/* Cannot read from anywhere -- bye bye array */
			md_error(mddev, conf->mirrors[read_disk].rdev);
			break;
		}
		/* write it back and re-read */
		start = d;
		while (d != read_disk) {
			if (d==0)
				d = conf->raid_disks;
			d--;
			rdev = conf->mirrors[d].rdev;
			if (rdev &&
			    test_bit(In_sync, &rdev->flags)) {
				if (sync_page_io(rdev->bdev,
						 sect + rdev->data_offset,
						 s<<9, conf->tmppage, WRITE)
				    == 0)
					/* Well, this device is dead */
					md_error(mddev, rdev);
			}
		}
		d = start;
		while (d != read_disk) {
			char b[BDEVNAME_SIZE];
			if (d==0)
				d = conf->raid_disks;
			d--;
			rdev = conf->mirrors[d].rdev;
			if (rdev &&
			    test_bit(In_sync, &rdev->flags)) {
				if (sync_page_io(rdev->bdev,
						 sect + rdev->data_offset,
						 s<<9, conf->tmppage, READ)
				    == 0)
					/* Well, this device is dead */
					md_error(mddev, rdev);
				else {
					atomic_add(s, &rdev->corrected_errors);
					printk(KERN_INFO
N
NeilBrown 已提交
1568
					       "md/raid1:%s: read error corrected "
1569 1570
					       "(%d sectors at %llu on %s)\n",
					       mdname(mddev), s,
1571 1572
					       (unsigned long long)(sect +
					           rdev->data_offset),
1573 1574 1575 1576 1577 1578 1579 1580 1581
					       bdevname(rdev->bdev, b));
				}
			}
		}
		sectors -= s;
		sect += s;
	}
}

L
Linus Torvalds 已提交
1582 1583 1584 1585 1586
static void raid1d(mddev_t *mddev)
{
	r1bio_t *r1_bio;
	struct bio *bio;
	unsigned long flags;
1587
	conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
1588 1589 1590 1591 1592 1593 1594 1595
	struct list_head *head = &conf->retry_list;
	int unplug=0;
	mdk_rdev_t *rdev;

	md_check_recovery(mddev);
	
	for (;;) {
		char b[BDEVNAME_SIZE];
1596

1597
		unplug += flush_pending_writes(conf);
1598

1599 1600 1601
		spin_lock_irqsave(&conf->device_lock, flags);
		if (list_empty(head)) {
			spin_unlock_irqrestore(&conf->device_lock, flags);
L
Linus Torvalds 已提交
1602
			break;
1603
		}
L
Linus Torvalds 已提交
1604 1605
		r1_bio = list_entry(head->prev, r1bio_t, retry_list);
		list_del(head->prev);
1606
		conf->nr_queued--;
L
Linus Torvalds 已提交
1607 1608 1609
		spin_unlock_irqrestore(&conf->device_lock, flags);

		mddev = r1_bio->mddev;
1610
		conf = mddev->private;
L
Linus Torvalds 已提交
1611 1612 1613 1614 1615
		if (test_bit(R1BIO_IsSync, &r1_bio->state)) {
			sync_request_write(mddev, r1_bio);
			unplug = 1;
		} else {
			int disk;
1616 1617 1618 1619 1620 1621 1622 1623 1624

			/* we got a read error. Maybe the drive is bad.  Maybe just
			 * the block and we can fix it.
			 * We freeze all other IO, and try reading the block from
			 * other devices.  When we find one, we re-write
			 * and check it that fixes the read error.
			 * This is all done synchronously while the array is
			 * frozen
			 */
1625 1626 1627 1628 1629 1630
			if (mddev->ro == 0) {
				freeze_array(conf);
				fix_read_error(conf, r1_bio->read_disk,
					       r1_bio->sector,
					       r1_bio->sectors);
				unfreeze_array(conf);
1631 1632 1633
			} else
				md_error(mddev,
					 conf->mirrors[r1_bio->read_disk].rdev);
1634

L
Linus Torvalds 已提交
1635
			bio = r1_bio->bios[r1_bio->read_disk];
1636
			if ((disk=read_balance(conf, r1_bio)) == -1) {
N
NeilBrown 已提交
1637
				printk(KERN_ALERT "md/raid1:%s: %s: unrecoverable I/O"
L
Linus Torvalds 已提交
1638
				       " read error for block %llu\n",
N
NeilBrown 已提交
1639
				       mdname(mddev),
L
Linus Torvalds 已提交
1640 1641 1642 1643
				       bdevname(bio->bi_bdev,b),
				       (unsigned long long)r1_bio->sector);
				raid_end_bio_io(r1_bio);
			} else {
1644
				const unsigned long do_sync = r1_bio->master_bio->bi_rw & REQ_SYNC;
1645 1646
				r1_bio->bios[r1_bio->read_disk] =
					mddev->ro ? IO_BLOCKED : NULL;
L
Linus Torvalds 已提交
1647 1648 1649 1650 1651 1652
				r1_bio->read_disk = disk;
				bio_put(bio);
				bio = bio_clone(r1_bio->master_bio, GFP_NOIO);
				r1_bio->bios[r1_bio->read_disk] = bio;
				rdev = conf->mirrors[disk].rdev;
				if (printk_ratelimit())
N
NeilBrown 已提交
1653
					printk(KERN_ERR "md/raid1:%s: redirecting sector %llu to"
1654
					       " other mirror: %s\n",
N
NeilBrown 已提交
1655
					       mdname(mddev),
1656 1657
					       (unsigned long long)r1_bio->sector,
					       bdevname(rdev->bdev,b));
L
Linus Torvalds 已提交
1658 1659 1660
				bio->bi_sector = r1_bio->sector + rdev->data_offset;
				bio->bi_bdev = rdev->bdev;
				bio->bi_end_io = raid1_end_read_request;
1661
				bio->bi_rw = READ | do_sync;
L
Linus Torvalds 已提交
1662 1663 1664 1665 1666
				bio->bi_private = r1_bio;
				unplug = 1;
				generic_make_request(bio);
			}
		}
N
NeilBrown 已提交
1667
		cond_resched();
L
Linus Torvalds 已提交
1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678
	}
	if (unplug)
		unplug_slaves(mddev);
}


static int init_resync(conf_t *conf)
{
	int buffs;

	buffs = RESYNC_WINDOW / RESYNC_BLOCK_SIZE;
1679
	BUG_ON(conf->r1buf_pool);
L
Linus Torvalds 已提交
1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697
	conf->r1buf_pool = mempool_create(buffs, r1buf_pool_alloc, r1buf_pool_free,
					  conf->poolinfo);
	if (!conf->r1buf_pool)
		return -ENOMEM;
	conf->next_resync = 0;
	return 0;
}

/*
 * perform a "sync" on one "block"
 *
 * We need to make sure that no normal I/O request - particularly write
 * requests - conflict with active sync requests.
 *
 * This is achieved by tracking pending requests and a 'barrier' concept
 * that can be installed to exclude normal IO requests.
 */

1698
static sector_t sync_request(mddev_t *mddev, sector_t sector_nr, int *skipped, int go_faster)
L
Linus Torvalds 已提交
1699
{
1700
	conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
1701 1702 1703
	r1bio_t *r1_bio;
	struct bio *bio;
	sector_t max_sector, nr_sectors;
1704
	int disk = -1;
L
Linus Torvalds 已提交
1705
	int i;
1706 1707
	int wonly = -1;
	int write_targets = 0, read_targets = 0;
N
NeilBrown 已提交
1708
	sector_t sync_blocks;
1709
	int still_degraded = 0;
L
Linus Torvalds 已提交
1710 1711 1712

	if (!conf->r1buf_pool)
		if (init_resync(conf))
1713
			return 0;
L
Linus Torvalds 已提交
1714

A
Andre Noll 已提交
1715
	max_sector = mddev->dev_sectors;
L
Linus Torvalds 已提交
1716
	if (sector_nr >= max_sector) {
1717 1718 1719 1720 1721
		/* If we aborted, we need to abort the
		 * sync on the 'current' bitmap chunk (there will
		 * only be one in raid1 resync.
		 * We can find the current addess in mddev->curr_resync
		 */
1722 1723
		if (mddev->curr_resync < max_sector) /* aborted */
			bitmap_end_sync(mddev->bitmap, mddev->curr_resync,
1724
						&sync_blocks, 1);
1725
		else /* completed sync */
1726
			conf->fullsync = 0;
1727 1728

		bitmap_close_sync(mddev->bitmap);
L
Linus Torvalds 已提交
1729 1730 1731 1732
		close_sync(conf);
		return 0;
	}

1733 1734
	if (mddev->bitmap == NULL &&
	    mddev->recovery_cp == MaxSector &&
1735
	    !test_bit(MD_RECOVERY_REQUESTED, &mddev->recovery) &&
1736 1737 1738 1739
	    conf->fullsync == 0) {
		*skipped = 1;
		return max_sector - sector_nr;
	}
1740 1741 1742
	/* before building a request, check if we can skip these blocks..
	 * This call the bitmap_start_sync doesn't actually record anything
	 */
1743
	if (!bitmap_start_sync(mddev->bitmap, sector_nr, &sync_blocks, 1) &&
1744
	    !conf->fullsync && !test_bit(MD_RECOVERY_REQUESTED, &mddev->recovery)) {
1745 1746 1747 1748
		/* We can skip this block, and probably several more */
		*skipped = 1;
		return sync_blocks;
	}
L
Linus Torvalds 已提交
1749
	/*
1750 1751 1752
	 * If there is non-resync activity waiting for a turn,
	 * and resync is going fast enough,
	 * then let it though before starting on this new sync request.
L
Linus Torvalds 已提交
1753
	 */
1754
	if (!go_faster && conf->nr_waiting)
L
Linus Torvalds 已提交
1755
		msleep_interruptible(1000);
1756

N
NeilBrown 已提交
1757
	bitmap_cond_end_sync(mddev->bitmap, sector_nr);
1758 1759 1760
	raise_barrier(conf);

	conf->next_resync = sector_nr;
L
Linus Torvalds 已提交
1761

1762 1763
	r1_bio = mempool_alloc(conf->r1buf_pool, GFP_NOIO);
	rcu_read_lock();
L
Linus Torvalds 已提交
1764
	/*
1765 1766 1767 1768 1769 1770
	 * If we get a correctably read error during resync or recovery,
	 * we might want to read from a different device.  So we
	 * flag all drives that could conceivably be read from for READ,
	 * and any others (which will be non-In_sync devices) for WRITE.
	 * If a read fails, we try reading from something else for which READ
	 * is OK.
L
Linus Torvalds 已提交
1771 1772 1773 1774
	 */

	r1_bio->mddev = mddev;
	r1_bio->sector = sector_nr;
1775
	r1_bio->state = 0;
L
Linus Torvalds 已提交
1776 1777 1778
	set_bit(R1BIO_IsSync, &r1_bio->state);

	for (i=0; i < conf->raid_disks; i++) {
1779
		mdk_rdev_t *rdev;
L
Linus Torvalds 已提交
1780 1781 1782 1783
		bio = r1_bio->bios[i];

		/* take from bio_init */
		bio->bi_next = NULL;
1784
		bio->bi_flags &= ~(BIO_POOL_MASK-1);
L
Linus Torvalds 已提交
1785
		bio->bi_flags |= 1 << BIO_UPTODATE;
1786
		bio->bi_comp_cpu = -1;
1787
		bio->bi_rw = READ;
L
Linus Torvalds 已提交
1788 1789 1790 1791 1792 1793 1794
		bio->bi_vcnt = 0;
		bio->bi_idx = 0;
		bio->bi_phys_segments = 0;
		bio->bi_size = 0;
		bio->bi_end_io = NULL;
		bio->bi_private = NULL;

1795 1796 1797
		rdev = rcu_dereference(conf->mirrors[i].rdev);
		if (rdev == NULL ||
			   test_bit(Faulty, &rdev->flags)) {
1798 1799
			still_degraded = 1;
			continue;
1800
		} else if (!test_bit(In_sync, &rdev->flags)) {
L
Linus Torvalds 已提交
1801 1802 1803
			bio->bi_rw = WRITE;
			bio->bi_end_io = end_sync_write;
			write_targets ++;
1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819
		} else {
			/* may need to read from here */
			bio->bi_rw = READ;
			bio->bi_end_io = end_sync_read;
			if (test_bit(WriteMostly, &rdev->flags)) {
				if (wonly < 0)
					wonly = i;
			} else {
				if (disk < 0)
					disk = i;
			}
			read_targets++;
		}
		atomic_inc(&rdev->nr_pending);
		bio->bi_sector = sector_nr + rdev->data_offset;
		bio->bi_bdev = rdev->bdev;
L
Linus Torvalds 已提交
1820 1821
		bio->bi_private = r1_bio;
	}
1822 1823 1824 1825
	rcu_read_unlock();
	if (disk < 0)
		disk = wonly;
	r1_bio->read_disk = disk;
1826

1827 1828 1829 1830 1831
	if (test_bit(MD_RECOVERY_SYNC, &mddev->recovery) && read_targets > 0)
		/* extra read targets are also write targets */
		write_targets += read_targets-1;

	if (write_targets == 0 || read_targets == 0) {
L
Linus Torvalds 已提交
1832 1833 1834
		/* There is nowhere to write, so all non-sync
		 * drives must be failed - so we are finished
		 */
1835 1836
		sector_t rv = max_sector - sector_nr;
		*skipped = 1;
L
Linus Torvalds 已提交
1837 1838 1839 1840
		put_buf(r1_bio);
		return rv;
	}

1841 1842
	if (max_sector > mddev->resync_max)
		max_sector = mddev->resync_max; /* Don't do IO beyond here */
L
Linus Torvalds 已提交
1843
	nr_sectors = 0;
1844
	sync_blocks = 0;
L
Linus Torvalds 已提交
1845 1846 1847 1848 1849 1850 1851
	do {
		struct page *page;
		int len = PAGE_SIZE;
		if (sector_nr + (len>>9) > max_sector)
			len = (max_sector - sector_nr) << 9;
		if (len == 0)
			break;
1852 1853
		if (sync_blocks == 0) {
			if (!bitmap_start_sync(mddev->bitmap, sector_nr,
1854 1855 1856
					       &sync_blocks, still_degraded) &&
			    !conf->fullsync &&
			    !test_bit(MD_RECOVERY_REQUESTED, &mddev->recovery))
1857
				break;
1858
			BUG_ON(sync_blocks < (PAGE_SIZE>>9));
1859
			if ((len >> 9) > sync_blocks)
1860
				len = sync_blocks<<9;
1861
		}
1862

L
Linus Torvalds 已提交
1863 1864 1865
		for (i=0 ; i < conf->raid_disks; i++) {
			bio = r1_bio->bios[i];
			if (bio->bi_end_io) {
1866
				page = bio->bi_io_vec[bio->bi_vcnt].bv_page;
L
Linus Torvalds 已提交
1867 1868
				if (bio_add_page(bio, page, len, 0) == 0) {
					/* stop here */
1869
					bio->bi_io_vec[bio->bi_vcnt].bv_page = page;
L
Linus Torvalds 已提交
1870 1871 1872
					while (i > 0) {
						i--;
						bio = r1_bio->bios[i];
1873 1874
						if (bio->bi_end_io==NULL)
							continue;
L
Linus Torvalds 已提交
1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885
						/* remove last page from this bio */
						bio->bi_vcnt--;
						bio->bi_size -= len;
						bio->bi_flags &= ~(1<< BIO_SEG_VALID);
					}
					goto bio_full;
				}
			}
		}
		nr_sectors += len>>9;
		sector_nr += len>>9;
1886
		sync_blocks -= (len>>9);
L
Linus Torvalds 已提交
1887 1888 1889 1890
	} while (r1_bio->bios[disk]->bi_vcnt < RESYNC_PAGES);
 bio_full:
	r1_bio->sectors = nr_sectors;

1891 1892 1893 1894 1895 1896 1897 1898
	/* For a user-requested sync, we read all readable devices and do a
	 * compare
	 */
	if (test_bit(MD_RECOVERY_REQUESTED, &mddev->recovery)) {
		atomic_set(&r1_bio->remaining, read_targets);
		for (i=0; i<conf->raid_disks; i++) {
			bio = r1_bio->bios[i];
			if (bio->bi_end_io == end_sync_read) {
1899
				md_sync_acct(bio->bi_bdev, nr_sectors);
1900 1901 1902 1903 1904 1905
				generic_make_request(bio);
			}
		}
	} else {
		atomic_set(&r1_bio->remaining, 1);
		bio = r1_bio->bios[r1_bio->read_disk];
1906
		md_sync_acct(bio->bi_bdev, nr_sectors);
1907
		generic_make_request(bio);
L
Linus Torvalds 已提交
1908

1909
	}
L
Linus Torvalds 已提交
1910 1911 1912
	return nr_sectors;
}

1913 1914 1915 1916 1917 1918 1919 1920
static sector_t raid1_size(mddev_t *mddev, sector_t sectors, int raid_disks)
{
	if (sectors)
		return sectors;

	return mddev->dev_sectors;
}

1921
static conf_t *setup_conf(mddev_t *mddev)
L
Linus Torvalds 已提交
1922 1923
{
	conf_t *conf;
1924
	int i;
L
Linus Torvalds 已提交
1925 1926
	mirror_info_t *disk;
	mdk_rdev_t *rdev;
1927
	int err = -ENOMEM;
L
Linus Torvalds 已提交
1928

1929
	conf = kzalloc(sizeof(conf_t), GFP_KERNEL);
L
Linus Torvalds 已提交
1930
	if (!conf)
1931
		goto abort;
L
Linus Torvalds 已提交
1932

1933
	conf->mirrors = kzalloc(sizeof(struct mirror_info)*mddev->raid_disks,
L
Linus Torvalds 已提交
1934 1935
				 GFP_KERNEL);
	if (!conf->mirrors)
1936
		goto abort;
L
Linus Torvalds 已提交
1937

1938 1939
	conf->tmppage = alloc_page(GFP_KERNEL);
	if (!conf->tmppage)
1940
		goto abort;
1941

1942
	conf->poolinfo = kzalloc(sizeof(*conf->poolinfo), GFP_KERNEL);
L
Linus Torvalds 已提交
1943
	if (!conf->poolinfo)
1944
		goto abort;
L
Linus Torvalds 已提交
1945 1946 1947 1948 1949
	conf->poolinfo->raid_disks = mddev->raid_disks;
	conf->r1bio_pool = mempool_create(NR_RAID1_BIOS, r1bio_pool_alloc,
					  r1bio_pool_free,
					  conf->poolinfo);
	if (!conf->r1bio_pool)
1950 1951
		goto abort;

1952
	conf->poolinfo->mddev = mddev;
L
Linus Torvalds 已提交
1953

1954
	spin_lock_init(&conf->device_lock);
1955
	list_for_each_entry(rdev, &mddev->disks, same_set) {
1956
		int disk_idx = rdev->raid_disk;
L
Linus Torvalds 已提交
1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970
		if (disk_idx >= mddev->raid_disks
		    || disk_idx < 0)
			continue;
		disk = conf->mirrors + disk_idx;

		disk->rdev = rdev;

		disk->head_position = 0;
	}
	conf->raid_disks = mddev->raid_disks;
	conf->mddev = mddev;
	INIT_LIST_HEAD(&conf->retry_list);

	spin_lock_init(&conf->resync_lock);
1971
	init_waitqueue_head(&conf->wait_barrier);
L
Linus Torvalds 已提交
1972

1973 1974 1975
	bio_list_init(&conf->pending_bio_list);
	bio_list_init(&conf->flushing_bio_list);

1976
	conf->last_used = -1;
L
Linus Torvalds 已提交
1977 1978 1979 1980
	for (i = 0; i < conf->raid_disks; i++) {

		disk = conf->mirrors + i;

1981 1982
		if (!disk->rdev ||
		    !test_bit(In_sync, &disk->rdev->flags)) {
L
Linus Torvalds 已提交
1983
			disk->head_position = 0;
1984 1985
			if (disk->rdev)
				conf->fullsync = 1;
1986 1987 1988 1989 1990 1991
		} else if (conf->last_used < 0)
			/*
			 * The first working device is used as a
			 * starting point to read balancing.
			 */
			conf->last_used = i;
L
Linus Torvalds 已提交
1992
	}
1993 1994 1995

	err = -EIO;
	if (conf->last_used < 0) {
N
NeilBrown 已提交
1996
		printk(KERN_ERR "md/raid1:%s: no operational mirrors\n",
1997 1998 1999 2000 2001 2002 2003
		       mdname(mddev));
		goto abort;
	}
	err = -ENOMEM;
	conf->thread = md_register_thread(raid1d, mddev, NULL);
	if (!conf->thread) {
		printk(KERN_ERR
N
NeilBrown 已提交
2004
		       "md/raid1:%s: couldn't allocate thread\n",
2005 2006
		       mdname(mddev));
		goto abort;
2007
	}
L
Linus Torvalds 已提交
2008

2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029
	return conf;

 abort:
	if (conf) {
		if (conf->r1bio_pool)
			mempool_destroy(conf->r1bio_pool);
		kfree(conf->mirrors);
		safe_put_page(conf->tmppage);
		kfree(conf->poolinfo);
		kfree(conf);
	}
	return ERR_PTR(err);
}

static int run(mddev_t *mddev)
{
	conf_t *conf;
	int i;
	mdk_rdev_t *rdev;

	if (mddev->level != 1) {
N
NeilBrown 已提交
2030
		printk(KERN_ERR "md/raid1:%s: raid level not set to mirroring (%d)\n",
2031 2032 2033 2034
		       mdname(mddev), mddev->level);
		return -EIO;
	}
	if (mddev->reshape_position != MaxSector) {
N
NeilBrown 已提交
2035
		printk(KERN_ERR "md/raid1:%s: reshape_position set but not supported\n",
2036 2037 2038
		       mdname(mddev));
		return -EIO;
	}
L
Linus Torvalds 已提交
2039
	/*
2040 2041 2042
	 * copy the already verified devices into our private RAID1
	 * bookkeeping area. [whatever we allocate in run(),
	 * should be freed in stop()]
L
Linus Torvalds 已提交
2043
	 */
2044 2045 2046 2047
	if (mddev->private == NULL)
		conf = setup_conf(mddev);
	else
		conf = mddev->private;
L
Linus Torvalds 已提交
2048

2049 2050
	if (IS_ERR(conf))
		return PTR_ERR(conf);
L
Linus Torvalds 已提交
2051

2052 2053 2054 2055 2056
	mddev->queue->queue_lock = &conf->device_lock;
	list_for_each_entry(rdev, &mddev->disks, same_set) {
		disk_stack_limits(mddev->gendisk, rdev->bdev,
				  rdev->data_offset << 9);
		/* as we don't honour merge_bvec_fn, we must never risk
2057 2058
		 * violating it, so limit ->max_segments to 1 lying within
		 * a single page, as a one page request is never in violation.
2059
		 */
2060 2061 2062 2063 2064
		if (rdev->bdev->bd_disk->queue->merge_bvec_fn) {
			blk_queue_max_segments(mddev->queue, 1);
			blk_queue_segment_boundary(mddev->queue,
						   PAGE_CACHE_SIZE - 1);
		}
L
Linus Torvalds 已提交
2065
	}
2066

2067 2068 2069 2070 2071 2072 2073 2074 2075 2076
	mddev->degraded = 0;
	for (i=0; i < conf->raid_disks; i++)
		if (conf->mirrors[i].rdev == NULL ||
		    !test_bit(In_sync, &conf->mirrors[i].rdev->flags) ||
		    test_bit(Faulty, &conf->mirrors[i].rdev->flags))
			mddev->degraded++;

	if (conf->raid_disks - mddev->degraded == 1)
		mddev->recovery_cp = MaxSector;

2077
	if (mddev->recovery_cp != MaxSector)
N
NeilBrown 已提交
2078
		printk(KERN_NOTICE "md/raid1:%s: not clean"
2079 2080
		       " -- starting background reconstruction\n",
		       mdname(mddev));
L
Linus Torvalds 已提交
2081
	printk(KERN_INFO 
N
NeilBrown 已提交
2082
		"md/raid1:%s: active with %d out of %d mirrors\n",
L
Linus Torvalds 已提交
2083 2084
		mdname(mddev), mddev->raid_disks - mddev->degraded, 
		mddev->raid_disks);
2085

L
Linus Torvalds 已提交
2086 2087 2088
	/*
	 * Ok, everything is just fine now
	 */
2089 2090 2091 2092
	mddev->thread = conf->thread;
	conf->thread = NULL;
	mddev->private = conf;

2093
	md_set_array_sectors(mddev, raid1_size(mddev, 0, 0));
L
Linus Torvalds 已提交
2094

2095
	mddev->queue->unplug_fn = raid1_unplug;
2096 2097
	mddev->queue->backing_dev_info.congested_fn = raid1_congested;
	mddev->queue->backing_dev_info.congested_data = mddev;
2098
	md_integrity_register(mddev);
L
Linus Torvalds 已提交
2099 2100 2101 2102 2103
	return 0;
}

static int stop(mddev_t *mddev)
{
2104
	conf_t *conf = mddev->private;
2105 2106 2107
	struct bitmap *bitmap = mddev->bitmap;

	/* wait for behind writes to complete */
2108
	if (bitmap && atomic_read(&bitmap->behind_writes) > 0) {
N
NeilBrown 已提交
2109 2110
		printk(KERN_INFO "md/raid1:%s: behind writes in progress - waiting to stop.\n",
		       mdname(mddev));
2111
		/* need to kick something here to make sure I/O goes? */
2112 2113
		wait_event(bitmap->behind_wait,
			   atomic_read(&bitmap->behind_writes) == 0);
2114
	}
L
Linus Torvalds 已提交
2115

2116 2117 2118
	raise_barrier(conf);
	lower_barrier(conf);

L
Linus Torvalds 已提交
2119 2120 2121 2122 2123
	md_unregister_thread(mddev->thread);
	mddev->thread = NULL;
	blk_sync_queue(mddev->queue); /* the unplug fn references 'conf'*/
	if (conf->r1bio_pool)
		mempool_destroy(conf->r1bio_pool);
2124 2125
	kfree(conf->mirrors);
	kfree(conf->poolinfo);
L
Linus Torvalds 已提交
2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139
	kfree(conf);
	mddev->private = NULL;
	return 0;
}

static int raid1_resize(mddev_t *mddev, sector_t sectors)
{
	/* no resync is happening, and there is enough space
	 * on all devices, so we can resize.
	 * We need to make sure resync covers any new space.
	 * If the array is shrinking we should possibly wait until
	 * any io in the removed space completes, but it hardly seems
	 * worth it.
	 */
2140
	md_set_array_sectors(mddev, raid1_size(mddev, sectors, 0));
D
Dan Williams 已提交
2141 2142
	if (mddev->array_sectors > raid1_size(mddev, sectors, 0))
		return -EINVAL;
2143
	set_capacity(mddev->gendisk, mddev->array_sectors);
2144
	revalidate_disk(mddev->gendisk);
D
Dan Williams 已提交
2145
	if (sectors > mddev->dev_sectors &&
2146
	    mddev->recovery_cp == MaxSector) {
A
Andre Noll 已提交
2147
		mddev->recovery_cp = mddev->dev_sectors;
L
Linus Torvalds 已提交
2148 2149
		set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
	}
D
Dan Williams 已提交
2150
	mddev->dev_sectors = sectors;
2151
	mddev->resync_max_sectors = sectors;
L
Linus Torvalds 已提交
2152 2153 2154
	return 0;
}

2155
static int raid1_reshape(mddev_t *mddev)
L
Linus Torvalds 已提交
2156 2157 2158 2159 2160 2161 2162 2163
{
	/* We need to:
	 * 1/ resize the r1bio_pool
	 * 2/ resize conf->mirrors
	 *
	 * We allocate a new r1bio_pool if we can.
	 * Then raise a device barrier and wait until all IO stops.
	 * Then resize conf->mirrors and swap in the new r1bio pool.
2164 2165 2166
	 *
	 * At the same time, we "pack" the devices so that all the missing
	 * devices have the higher raid_disk numbers.
L
Linus Torvalds 已提交
2167 2168 2169 2170
	 */
	mempool_t *newpool, *oldpool;
	struct pool_info *newpoolinfo;
	mirror_info_t *newmirrors;
2171
	conf_t *conf = mddev->private;
2172
	int cnt, raid_disks;
2173
	unsigned long flags;
2174
	int d, d2, err;
L
Linus Torvalds 已提交
2175

2176
	/* Cannot change chunk_size, layout, or level */
2177
	if (mddev->chunk_sectors != mddev->new_chunk_sectors ||
2178 2179
	    mddev->layout != mddev->new_layout ||
	    mddev->level != mddev->new_level) {
2180
		mddev->new_chunk_sectors = mddev->chunk_sectors;
2181 2182 2183 2184 2185
		mddev->new_layout = mddev->layout;
		mddev->new_level = mddev->level;
		return -EINVAL;
	}

2186 2187 2188
	err = md_allow_write(mddev);
	if (err)
		return err;
2189

2190 2191
	raid_disks = mddev->raid_disks + mddev->delta_disks;

2192 2193 2194 2195 2196 2197
	if (raid_disks < conf->raid_disks) {
		cnt=0;
		for (d= 0; d < conf->raid_disks; d++)
			if (conf->mirrors[d].rdev)
				cnt++;
		if (cnt > raid_disks)
L
Linus Torvalds 已提交
2198
			return -EBUSY;
2199
	}
L
Linus Torvalds 已提交
2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212

	newpoolinfo = kmalloc(sizeof(*newpoolinfo), GFP_KERNEL);
	if (!newpoolinfo)
		return -ENOMEM;
	newpoolinfo->mddev = mddev;
	newpoolinfo->raid_disks = raid_disks;

	newpool = mempool_create(NR_RAID1_BIOS, r1bio_pool_alloc,
				 r1bio_pool_free, newpoolinfo);
	if (!newpool) {
		kfree(newpoolinfo);
		return -ENOMEM;
	}
2213
	newmirrors = kzalloc(sizeof(struct mirror_info) * raid_disks, GFP_KERNEL);
L
Linus Torvalds 已提交
2214 2215 2216 2217 2218 2219
	if (!newmirrors) {
		kfree(newpoolinfo);
		mempool_destroy(newpool);
		return -ENOMEM;
	}

2220
	raise_barrier(conf);
L
Linus Torvalds 已提交
2221 2222 2223 2224

	/* ok, everything is stopped */
	oldpool = conf->r1bio_pool;
	conf->r1bio_pool = newpool;
2225

2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237
	for (d = d2 = 0; d < conf->raid_disks; d++) {
		mdk_rdev_t *rdev = conf->mirrors[d].rdev;
		if (rdev && rdev->raid_disk != d2) {
			char nm[20];
			sprintf(nm, "rd%d", rdev->raid_disk);
			sysfs_remove_link(&mddev->kobj, nm);
			rdev->raid_disk = d2;
			sprintf(nm, "rd%d", rdev->raid_disk);
			sysfs_remove_link(&mddev->kobj, nm);
			if (sysfs_create_link(&mddev->kobj,
					      &rdev->kobj, nm))
				printk(KERN_WARNING
N
NeilBrown 已提交
2238 2239 2240
				       "md/raid1:%s: cannot register "
				       "%s\n",
				       mdname(mddev), nm);
2241
		}
2242 2243 2244
		if (rdev)
			newmirrors[d2++].rdev = rdev;
	}
L
Linus Torvalds 已提交
2245 2246 2247 2248 2249
	kfree(conf->mirrors);
	conf->mirrors = newmirrors;
	kfree(conf->poolinfo);
	conf->poolinfo = newpoolinfo;

2250
	spin_lock_irqsave(&conf->device_lock, flags);
L
Linus Torvalds 已提交
2251
	mddev->degraded += (raid_disks - conf->raid_disks);
2252
	spin_unlock_irqrestore(&conf->device_lock, flags);
L
Linus Torvalds 已提交
2253
	conf->raid_disks = mddev->raid_disks = raid_disks;
2254
	mddev->delta_disks = 0;
L
Linus Torvalds 已提交
2255

2256
	conf->last_used = 0; /* just make sure it is in-range */
2257
	lower_barrier(conf);
L
Linus Torvalds 已提交
2258 2259 2260 2261 2262 2263 2264 2265

	set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
	md_wakeup_thread(mddev->thread);

	mempool_destroy(oldpool);
	return 0;
}

2266
static void raid1_quiesce(mddev_t *mddev, int state)
2267
{
2268
	conf_t *conf = mddev->private;
2269 2270

	switch(state) {
2271 2272 2273
	case 2: /* wake for suspend */
		wake_up(&conf->wait_barrier);
		break;
2274
	case 1:
2275
		raise_barrier(conf);
2276
		break;
2277
	case 0:
2278
		lower_barrier(conf);
2279 2280 2281 2282
		break;
	}
}

2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299
static void *raid1_takeover(mddev_t *mddev)
{
	/* raid1 can take over:
	 *  raid5 with 2 devices, any layout or chunk size
	 */
	if (mddev->level == 5 && mddev->raid_disks == 2) {
		conf_t *conf;
		mddev->new_level = 1;
		mddev->new_layout = 0;
		mddev->new_chunk_sectors = 0;
		conf = setup_conf(mddev);
		if (!IS_ERR(conf))
			conf->barrier = 1;
		return conf;
	}
	return ERR_PTR(-EINVAL);
}
L
Linus Torvalds 已提交
2300

2301
static struct mdk_personality raid1_personality =
L
Linus Torvalds 已提交
2302 2303
{
	.name		= "raid1",
2304
	.level		= 1,
L
Linus Torvalds 已提交
2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315
	.owner		= THIS_MODULE,
	.make_request	= make_request,
	.run		= run,
	.stop		= stop,
	.status		= status,
	.error_handler	= error,
	.hot_add_disk	= raid1_add_disk,
	.hot_remove_disk= raid1_remove_disk,
	.spare_active	= raid1_spare_active,
	.sync_request	= sync_request,
	.resize		= raid1_resize,
2316
	.size		= raid1_size,
2317
	.check_reshape	= raid1_reshape,
2318
	.quiesce	= raid1_quiesce,
2319
	.takeover	= raid1_takeover,
L
Linus Torvalds 已提交
2320 2321 2322 2323
};

static int __init raid_init(void)
{
2324
	return register_md_personality(&raid1_personality);
L
Linus Torvalds 已提交
2325 2326 2327 2328
}

static void raid_exit(void)
{
2329
	unregister_md_personality(&raid1_personality);
L
Linus Torvalds 已提交
2330 2331 2332 2333 2334
}

module_init(raid_init);
module_exit(raid_exit);
MODULE_LICENSE("GPL");
2335
MODULE_DESCRIPTION("RAID1 (mirroring) personality for MD");
L
Linus Torvalds 已提交
2336
MODULE_ALIAS("md-personality-3"); /* RAID1 */
2337
MODULE_ALIAS("md-raid1");
2338
MODULE_ALIAS("md-level-1");