raid1.c 62.6 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;

322
	if (error == -EOPNOTSUPP && test_bit(R1BIO_Barrier, &r1_bio->state)) {
323 324 325
		set_bit(BarriersNotsupp, &conf->mirrors[mirror].rdev->flags);
		set_bit(R1BIO_BarrierRetry, &r1_bio->state);
		r1_bio->mddev->barriers_work = 0;
326
		/* Don't rdev_dec_pending in this branch - keep it for the retry */
327
	} else {
L
Linus Torvalds 已提交
328
		/*
329
		 * this branch is our 'one mirror IO has finished' event handler:
L
Linus Torvalds 已提交
330
		 */
331
		r1_bio->bios[mirror] = NULL;
332
		to_put = bio;
333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368
		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
			/*
			 * 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.
			 */
			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);
369
					bio_endio(mbio, 0);
370
				}
371 372
			}
		}
373
		rdev_dec_pending(conf->mirrors[mirror].rdev, conf->mddev);
374
	}
L
Linus Torvalds 已提交
375 376 377 378 379 380
	/*
	 *
	 * Let's see if all mirrored write operations have finished
	 * already.
	 */
	if (atomic_dec_and_test(&r1_bio->remaining)) {
381
		if (test_bit(R1BIO_BarrierRetry, &r1_bio->state))
382
			reschedule_retry(r1_bio);
383 384 385 386 387 388 389 390 391 392 393 394 395 396 397
		else {
			/* it really is the end of this request */
			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);
			}
			/* 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);
398
		}
L
Linus Torvalds 已提交
399
	}
400

401 402
	if (to_put)
		bio_put(to_put);
L
Linus Torvalds 已提交
403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421
}


/*
 * 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)
{
422
	const sector_t this_sector = r1_bio->sector;
L
Linus Torvalds 已提交
423
	int new_disk = conf->last_used, disk = new_disk;
424
	int wonly_disk = -1;
L
Linus Torvalds 已提交
425 426
	const int sectors = r1_bio->sectors;
	sector_t new_distance, current_distance;
427
	mdk_rdev_t *rdev;
L
Linus Torvalds 已提交
428 429 430

	rcu_read_lock();
	/*
431
	 * Check if we can balance. We can balance on the whole
L
Linus Torvalds 已提交
432 433 434 435 436 437
	 * 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)) {
438
		/* Choose the first operational device, for consistancy */
L
Linus Torvalds 已提交
439 440
		new_disk = 0;

441
		for (rdev = rcu_dereference(conf->mirrors[new_disk].rdev);
442
		     r1_bio->bios[new_disk] == IO_BLOCKED ||
443
		     !rdev || !test_bit(In_sync, &rdev->flags)
444
			     || test_bit(WriteMostly, &rdev->flags);
445
		     rdev = rcu_dereference(conf->mirrors[++new_disk].rdev)) {
446

447 448
			if (rdev && test_bit(In_sync, &rdev->flags) &&
				r1_bio->bios[new_disk] != IO_BLOCKED)
449 450 451 452
				wonly_disk = new_disk;

			if (new_disk == conf->raid_disks - 1) {
				new_disk = wonly_disk;
L
Linus Torvalds 已提交
453 454 455 456 457 458 459 460
				break;
			}
		}
		goto rb_out;
	}


	/* make sure the disk is operational */
461
	for (rdev = rcu_dereference(conf->mirrors[new_disk].rdev);
462
	     r1_bio->bios[new_disk] == IO_BLOCKED ||
463
	     !rdev || !test_bit(In_sync, &rdev->flags) ||
464
		     test_bit(WriteMostly, &rdev->flags);
465
	     rdev = rcu_dereference(conf->mirrors[new_disk].rdev)) {
466

467 468
		if (rdev && test_bit(In_sync, &rdev->flags) &&
		    r1_bio->bios[new_disk] != IO_BLOCKED)
469 470
			wonly_disk = new_disk;

L
Linus Torvalds 已提交
471 472 473 474
		if (new_disk <= 0)
			new_disk = conf->raid_disks;
		new_disk--;
		if (new_disk == disk) {
475 476
			new_disk = wonly_disk;
			break;
L
Linus Torvalds 已提交
477 478
		}
	}
479 480 481 482

	if (new_disk < 0)
		goto rb_out;

L
Linus Torvalds 已提交
483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502
	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--;

503
		rdev = rcu_dereference(conf->mirrors[disk].rdev);
504

505
		if (!rdev || r1_bio->bios[disk] == IO_BLOCKED ||
506
		    !test_bit(In_sync, &rdev->flags) ||
507
		    test_bit(WriteMostly, &rdev->flags))
L
Linus Torvalds 已提交
508 509 510 511 512 513 514 515 516 517 518 519 520
			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);

521
 rb_out:
L
Linus Torvalds 已提交
522 523 524


	if (new_disk >= 0) {
525
		rdev = rcu_dereference(conf->mirrors[new_disk].rdev);
526 527 528
		if (!rdev)
			goto retry;
		atomic_inc(&rdev->nr_pending);
529
		if (!test_bit(In_sync, &rdev->flags)) {
L
Linus Torvalds 已提交
530 531 532
			/* cannot risk returning a device that failed
			 * before we inc'ed nr_pending
			 */
533
			rdev_dec_pending(rdev, conf->mddev);
L
Linus Torvalds 已提交
534 535
			goto retry;
		}
536 537
		conf->next_seq_sect = this_sector + sectors;
		conf->last_used = new_disk;
L
Linus Torvalds 已提交
538 539 540 541 542 543 544 545
	}
	rcu_read_unlock();

	return new_disk;
}

static void unplug_slaves(mddev_t *mddev)
{
546
	conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
547 548 549 550
	int i;

	rcu_read_lock();
	for (i=0; i<mddev->raid_disks; i++) {
551
		mdk_rdev_t *rdev = rcu_dereference(conf->mirrors[i].rdev);
552
		if (rdev && !test_bit(Faulty, &rdev->flags) && atomic_read(&rdev->nr_pending)) {
553
			struct request_queue *r_queue = bdev_get_queue(rdev->bdev);
L
Linus Torvalds 已提交
554 555 556 557

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

558
			blk_unplug(r_queue);
L
Linus Torvalds 已提交
559 560 561 562 563 564 565 566

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

567
static void raid1_unplug(struct request_queue *q)
L
Linus Torvalds 已提交
568
{
569 570 571 572
	mddev_t *mddev = q->queuedata;

	unplug_slaves(mddev);
	md_wakeup_thread(mddev->thread);
L
Linus Torvalds 已提交
573 574
}

575 576 577
static int raid1_congested(void *data, int bits)
{
	mddev_t *mddev = data;
578
	conf_t *conf = mddev->private;
579 580
	int i, ret = 0;

581 582 583
	if (mddev_congested(mddev, bits))
		return 1;

584 585 586 587
	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)) {
588
			struct request_queue *q = bdev_get_queue(rdev->bdev);
589 590 591 592

			/* Note the '|| 1' - when read_balance prefers
			 * non-congested targets, it can be removed
			 */
593
			if ((bits & (1<<BDI_async_congested)) || 1)
594 595 596 597 598 599 600 601 602 603
				ret |= bdi_congested(&q->backing_dev_info, bits);
			else
				ret &= bdi_congested(&q->backing_dev_info, bits);
		}
	}
	rcu_read_unlock();
	return ret;
}


604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634
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;
}

635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654
/* 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
655 656 657
 */
#define RESYNC_DEPTH 32

658
static void raise_barrier(conf_t *conf)
L
Linus Torvalds 已提交
659 660
{
	spin_lock_irq(&conf->resync_lock);
661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681

	/* 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;
682
	BUG_ON(conf->barrier <= 0);
683 684 685 686 687 688 689 690 691 692 693 694 695 696 697
	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
698
	}
699
	conf->nr_pending++;
L
Linus Torvalds 已提交
700 701 702
	spin_unlock_irq(&conf->resync_lock);
}

703 704 705 706 707 708 709 710 711
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);
}

712 713 714 715 716
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
717 718 719 720 721 722 723 724
	 * 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.
725 726 727 728 729
	 */
	spin_lock_irq(&conf->resync_lock);
	conf->barrier++;
	conf->nr_waiting++;
	wait_event_lock_irq(conf->wait_barrier,
730
			    conf->nr_pending == conf->nr_queued+1,
731
			    conf->resync_lock,
732 733
			    ({ flush_pending_writes(conf);
			       raid1_unplug(conf->mddev->queue); }));
734 735 736 737 738 739 740 741 742 743 744 745
	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);
}

746

747 748 749 750 751
/* duplicate the data pages for behind I/O */
static struct page **alloc_behind_pages(struct bio *bio)
{
	int i;
	struct bio_vec *bvec;
752
	struct page **pages = kzalloc(bio->bi_vcnt * sizeof(struct page *),
753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771
					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++)
772
			put_page(pages[i]);
773 774 775 776 777
	kfree(pages);
	PRINTK("%dB behind alloc failed, doing sync I/O\n", bio->bi_size);
	return NULL;
}

778
static int make_request(mddev_t *mddev, struct bio * bio)
L
Linus Torvalds 已提交
779
{
780
	conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
781 782 783
	mirror_info_t *mirror;
	r1bio_t *r1_bio;
	struct bio *read_bio;
784
	int i, targets = 0, disks;
785
	struct bitmap *bitmap;
786 787
	unsigned long flags;
	struct bio_list bl;
788
	struct page **behind_pages = NULL;
789
	const int rw = bio_data_dir(bio);
790 791
	const bool do_sync = bio_rw_flagged(bio, BIO_RW_SYNCIO);
	bool do_barriers;
792
	mdk_rdev_t *blocked_rdev;
793

L
Linus Torvalds 已提交
794 795 796 797
	/*
	 * 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.
798 799 800
	 * We test barriers_work *after* md_write_start as md_write_start
	 * may cause the first superblock write, and that will check out
	 * if barriers work.
L
Linus Torvalds 已提交
801
	 */
802

803 804
	md_write_start(mddev, bio); /* wait on superblock update early */

805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823
	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);
	}
824 825
	if (unlikely(!mddev->barriers_work &&
		     bio_rw_flagged(bio, BIO_RW_BARRIER))) {
826 827
		if (rw == WRITE)
			md_write_end(mddev);
828
		bio_endio(bio, -EOPNOTSUPP);
829 830 831
		return 0;
	}

832
	wait_barrier(conf);
L
Linus Torvalds 已提交
833

834 835
	bitmap = mddev->bitmap;

L
Linus Torvalds 已提交
836 837 838 839 840 841 842 843 844
	/*
	 * 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;
845
	r1_bio->state = 0;
L
Linus Torvalds 已提交
846 847 848
	r1_bio->mddev = mddev;
	r1_bio->sector = bio->bi_sector;

849
	if (rw == READ) {
L
Linus Torvalds 已提交
850 851 852 853 854 855 856 857 858 859 860 861
		/*
		 * 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;

862 863 864 865 866 867 868 869 870
		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
871 872 873 874 875 876 877 878 879
		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;
880
		read_bio->bi_rw = READ | (do_sync << BIO_RW_SYNCIO);
L
Linus Torvalds 已提交
881 882 883 884 885 886 887 888 889 890 891 892 893 894
		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;
895 896 897 898 899 900 901
#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
902 903
 retry_write:
	blocked_rdev = NULL;
L
Linus Torvalds 已提交
904 905
	rcu_read_lock();
	for (i = 0;  i < disks; i++) {
906 907 908 909 910 911 912
		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
913
			atomic_inc(&rdev->nr_pending);
914
			if (test_bit(Faulty, &rdev->flags)) {
915
				rdev_dec_pending(rdev, mddev);
L
Linus Torvalds 已提交
916
				r1_bio->bios[i] = NULL;
917
			} else {
L
Linus Torvalds 已提交
918
				r1_bio->bios[i] = bio;
919 920
				targets++;
			}
L
Linus Torvalds 已提交
921 922 923 924 925
		} else
			r1_bio->bios[i] = NULL;
	}
	rcu_read_unlock();

926 927 928 929 930 931 932 933 934 935 936 937 938 939
	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;
	}

940 941
	BUG_ON(targets == 0); /* we never fail the last device */

942 943 944 945 946 947
	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);
	}

948 949 950 951
	/* 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 */
952
	if (bitmap &&
953 954
	    (atomic_read(&bitmap->behind_writes)
	     < mddev->bitmap_info.max_write_behind) &&
955
	    !waitqueue_active(&bitmap->behind_wait) &&
956 957 958
	    (behind_pages = alloc_behind_pages(bio)) != NULL)
		set_bit(R1BIO_BehindIO, &r1_bio->state);

959
	atomic_set(&r1_bio->remaining, 0);
960
	atomic_set(&r1_bio->behind_remaining, 0);
961

962
	do_barriers = bio_rw_flagged(bio, BIO_RW_BARRIER);
963 964 965
	if (do_barriers)
		set_bit(R1BIO_Barrier, &r1_bio->state);

966
	bio_list_init(&bl);
L
Linus Torvalds 已提交
967 968 969 970 971 972 973 974 975 976 977
	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;
978 979
		mbio->bi_rw = WRITE | (do_barriers << BIO_RW_BARRIER) |
			(do_sync << BIO_RW_SYNCIO);
L
Linus Torvalds 已提交
980 981
		mbio->bi_private = r1_bio;

982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998
		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
999 1000
		atomic_inc(&r1_bio->remaining);

1001
		bio_list_add(&bl, mbio);
L
Linus Torvalds 已提交
1002
	}
1003
	kfree(behind_pages); /* the behind pages are attached to the bios now */
L
Linus Torvalds 已提交
1004

1005 1006
	bitmap_startwrite(bitmap, bio->bi_sector, r1_bio->sectors,
				test_bit(R1BIO_BehindIO, &r1_bio->state));
1007 1008 1009 1010 1011 1012 1013
	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);

1014 1015 1016
	/* In case raid1d snuck into freeze_array */
	wake_up(&conf->wait_barrier);

1017 1018
	if (do_sync)
		md_wakeup_thread(mddev->thread);
1019 1020 1021 1022 1023
#if 0
	while ((bio = bio_list_pop(&bl)) != NULL)
		generic_make_request(bio);
#endif

L
Linus Torvalds 已提交
1024 1025 1026 1027 1028
	return 0;
}

static void status(struct seq_file *seq, mddev_t *mddev)
{
1029
	conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
1030 1031 1032
	int i;

	seq_printf(seq, " [%d/%d] [", conf->raid_disks,
1033
		   conf->raid_disks - mddev->degraded);
1034 1035 1036
	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
1037
		seq_printf(seq, "%s",
1038 1039 1040
			   rdev && test_bit(In_sync, &rdev->flags) ? "U" : "_");
	}
	rcu_read_unlock();
L
Linus Torvalds 已提交
1041 1042 1043 1044 1045 1046 1047
	seq_printf(seq, "]");
}


static void error(mddev_t *mddev, mdk_rdev_t *rdev)
{
	char b[BDEVNAME_SIZE];
1048
	conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
1049 1050 1051 1052 1053 1054 1055

	/*
	 * 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
	 */
1056
	if (test_bit(In_sync, &rdev->flags)
1057
	    && (conf->raid_disks - mddev->degraded) == 1) {
L
Linus Torvalds 已提交
1058 1059
		/*
		 * Don't fail the drive, act as though we were just a
1060 1061 1062
		 * 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
1063
		 */
1064
		mddev->recovery_disabled = 1;
L
Linus Torvalds 已提交
1065
		return;
1066
	}
1067 1068 1069
	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
1070
		mddev->degraded++;
1071
		set_bit(Faulty, &rdev->flags);
1072
		spin_unlock_irqrestore(&conf->device_lock, flags);
L
Linus Torvalds 已提交
1073 1074 1075
		/*
		 * if recovery is running, make sure it aborts.
		 */
1076
		set_bit(MD_RECOVERY_INTR, &mddev->recovery);
1077 1078
	} else
		set_bit(Faulty, &rdev->flags);
1079
	set_bit(MD_CHANGE_DEVS, &mddev->flags);
N
NeilBrown 已提交
1080 1081 1082 1083
	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
1084 1085 1086 1087 1088 1089
}

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

N
NeilBrown 已提交
1090
	printk(KERN_DEBUG "RAID1 conf printout:\n");
L
Linus Torvalds 已提交
1091
	if (!conf) {
N
NeilBrown 已提交
1092
		printk(KERN_DEBUG "(!conf)\n");
L
Linus Torvalds 已提交
1093 1094
		return;
	}
N
NeilBrown 已提交
1095
	printk(KERN_DEBUG " --- wd:%d rd:%d\n", conf->raid_disks - conf->mddev->degraded,
L
Linus Torvalds 已提交
1096 1097
		conf->raid_disks);

1098
	rcu_read_lock();
L
Linus Torvalds 已提交
1099 1100
	for (i = 0; i < conf->raid_disks; i++) {
		char b[BDEVNAME_SIZE];
1101 1102
		mdk_rdev_t *rdev = rcu_dereference(conf->mirrors[i].rdev);
		if (rdev)
N
NeilBrown 已提交
1103
			printk(KERN_DEBUG " disk %d, wo:%d, o:%d, dev:%s\n",
1104 1105 1106
			       i, !test_bit(In_sync, &rdev->flags),
			       !test_bit(Faulty, &rdev->flags),
			       bdevname(rdev->bdev,b));
L
Linus Torvalds 已提交
1107
	}
1108
	rcu_read_unlock();
L
Linus Torvalds 已提交
1109 1110 1111 1112
}

static void close_sync(conf_t *conf)
{
1113 1114
	wait_barrier(conf);
	allow_barrier(conf);
L
Linus Torvalds 已提交
1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126

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

static int raid1_spare_active(mddev_t *mddev)
{
	int i;
	conf_t *conf = mddev->private;

	/*
	 * Find all failed disks within the RAID1 configuration 
1127 1128
	 * and mark them readable.
	 * Called under mddev lock, so rcu protection not needed.
L
Linus Torvalds 已提交
1129 1130
	 */
	for (i = 0; i < conf->raid_disks; i++) {
1131 1132 1133
		mdk_rdev_t *rdev = conf->mirrors[i].rdev;
		if (rdev
		    && !test_bit(Faulty, &rdev->flags)
1134 1135 1136
		    && !test_and_set_bit(In_sync, &rdev->flags)) {
			unsigned long flags;
			spin_lock_irqsave(&conf->device_lock, flags);
L
Linus Torvalds 已提交
1137
			mddev->degraded--;
1138
			spin_unlock_irqrestore(&conf->device_lock, flags);
L
Linus Torvalds 已提交
1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149
		}
	}

	print_conf(conf);
	return 0;
}


static int raid1_add_disk(mddev_t *mddev, mdk_rdev_t *rdev)
{
	conf_t *conf = mddev->private;
1150
	int err = -EEXIST;
1151
	int mirror = 0;
L
Linus Torvalds 已提交
1152
	mirror_info_t *p;
1153 1154
	int first = 0;
	int last = mddev->raid_disks - 1;
L
Linus Torvalds 已提交
1155

1156 1157 1158 1159
	if (rdev->raid_disk >= 0)
		first = last = rdev->raid_disk;

	for (mirror = first; mirror <= last; mirror++)
L
Linus Torvalds 已提交
1160 1161
		if ( !(p=conf->mirrors+mirror)->rdev) {

1162 1163
			disk_stack_limits(mddev->gendisk, rdev->bdev,
					  rdev->data_offset << 9);
1164 1165 1166 1167 1168
			/* 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
1169
			 */
1170 1171 1172 1173 1174
			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
1175 1176 1177

			p->head_position = 0;
			rdev->raid_disk = mirror;
1178
			err = 0;
1179 1180 1181 1182
			/* 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)
1183
				conf->fullsync = 1;
1184
			rcu_assign_pointer(p->rdev, rdev);
L
Linus Torvalds 已提交
1185 1186
			break;
		}
1187
	md_integrity_add_rdev(rdev, mddev);
L
Linus Torvalds 已提交
1188
	print_conf(conf);
1189
	return err;
L
Linus Torvalds 已提交
1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201
}

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) {
1202
		if (test_bit(In_sync, &rdev->flags) ||
L
Linus Torvalds 已提交
1203 1204 1205 1206
		    atomic_read(&rdev->nr_pending)) {
			err = -EBUSY;
			goto abort;
		}
1207 1208 1209 1210 1211 1212 1213 1214
		/* 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
1215
		p->rdev = NULL;
1216
		synchronize_rcu();
L
Linus Torvalds 已提交
1217 1218 1219 1220
		if (atomic_read(&rdev->nr_pending)) {
			/* lost the race, try later */
			err = -EBUSY;
			p->rdev = rdev;
1221
			goto abort;
L
Linus Torvalds 已提交
1222
		}
1223
		md_integrity_register(mddev);
L
Linus Torvalds 已提交
1224 1225 1226 1227 1228 1229 1230 1231
	}
abort:

	print_conf(conf);
	return err;
}


1232
static void end_sync_read(struct bio *bio, int error)
L
Linus Torvalds 已提交
1233
{
1234
	r1bio_t *r1_bio = bio->bi_private;
1235
	int i;
L
Linus Torvalds 已提交
1236

1237 1238 1239 1240 1241
	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
1242 1243 1244 1245 1246
	/*
	 * 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
	 */
1247
	if (test_bit(BIO_UPTODATE, &bio->bi_flags))
L
Linus Torvalds 已提交
1248
		set_bit(R1BIO_Uptodate, &r1_bio->state);
1249 1250 1251

	if (atomic_dec_and_test(&r1_bio->remaining))
		reschedule_retry(r1_bio);
L
Linus Torvalds 已提交
1252 1253
}

1254
static void end_sync_write(struct bio *bio, int error)
L
Linus Torvalds 已提交
1255 1256
{
	int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
1257
	r1bio_t *r1_bio = bio->bi_private;
L
Linus Torvalds 已提交
1258
	mddev_t *mddev = r1_bio->mddev;
1259
	conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
1260 1261 1262 1263 1264 1265 1266 1267
	int i;
	int mirror=0;

	for (i = 0; i < conf->raid_disks; i++)
		if (r1_bio->bios[i] == bio) {
			mirror = i;
			break;
		}
1268 1269 1270 1271 1272 1273
	if (!uptodate) {
		int sync_blocks = 0;
		sector_t s = r1_bio->sector;
		long sectors_to_go = r1_bio->sectors;
		/* make sure these bits doesn't get cleared. */
		do {
1274
			bitmap_end_sync(mddev->bitmap, s,
1275 1276 1277 1278
					&sync_blocks, 1);
			s += sync_blocks;
			sectors_to_go -= sync_blocks;
		} while (sectors_to_go > 0);
L
Linus Torvalds 已提交
1279
		md_error(mddev, conf->mirrors[mirror].rdev);
1280
	}
1281

L
Linus Torvalds 已提交
1282 1283 1284
	update_head_pos(mirror, r1_bio);

	if (atomic_dec_and_test(&r1_bio->remaining)) {
1285
		sector_t s = r1_bio->sectors;
L
Linus Torvalds 已提交
1286
		put_buf(r1_bio);
1287
		md_done_sync(mddev, s, uptodate);
L
Linus Torvalds 已提交
1288 1289 1290 1291 1292
	}
}

static void sync_request_write(mddev_t *mddev, r1bio_t *r1_bio)
{
1293
	conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
1294 1295 1296 1297 1298 1299
	int i;
	int disks = conf->raid_disks;
	struct bio *bio, *wbio;

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

1300

1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322
	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;
1323
				rdev_dec_pending(conf->mirrors[primary].rdev, mddev);
1324 1325 1326 1327
				break;
			}
		r1_bio->read_disk = primary;
		for (i=0; i<mddev->raid_disks; i++)
1328
			if (r1_bio->bios[i]->bi_end_io == end_sync_read) {
1329 1330 1331 1332
				int j;
				int vcnt = r1_bio->sectors >> (PAGE_SHIFT- 9);
				struct bio *pbio = r1_bio->bios[primary];
				struct bio *sbio = r1_bio->bios[i];
1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345

				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;
1346 1347
				if (j >= 0)
					mddev->resync_mismatches += r1_bio->sectors;
1348 1349
				if (j < 0 || (test_bit(MD_RECOVERY_CHECK, &mddev->recovery)
					      && test_bit(BIO_UPTODATE, &sbio->bi_flags))) {
1350
					sbio->bi_end_io = NULL;
1351 1352
					rdev_dec_pending(conf->mirrors[i].rdev, mddev);
				} else {
1353
					/* fixup the bio for reuse */
1354
					int size;
1355 1356 1357 1358 1359 1360 1361 1362 1363 1364
					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;
1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375
					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),
1376 1377
						       page_address(pbio->bi_io_vec[j].bv_page),
						       PAGE_SIZE);
1378
					}
1379

1380 1381 1382
				}
			}
	}
L
Linus Torvalds 已提交
1383
	if (!test_bit(R1BIO_Uptodate, &r1_bio->state)) {
1384 1385 1386
		/* 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
1387
		 * read into the pages we already have so we don't
1388 1389 1390 1391
		 * 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
1392
		 */
1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406
		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) {
1407 1408 1409 1410
					/* No rcu protection needed here devices
					 * can only be removed when no resync is
					 * active, and resync is currently active
					 */
1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426
					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) {
1427
				int start = d;
1428 1429 1430 1431 1432 1433 1434 1435 1436
				/* 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;
1437
					atomic_add(s, &rdev->corrected_errors);
1438 1439 1440 1441
					if (sync_page_io(rdev->bdev,
							 sect + rdev->data_offset,
							 s<<9,
							 bio->bi_io_vec[idx].bv_page,
1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453
							 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,
1454 1455 1456
							 sect + rdev->data_offset,
							 s<<9,
							 bio->bi_io_vec[idx].bv_page,
1457
							 READ) == 0)
1458 1459 1460 1461 1462 1463
						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
1464
				printk(KERN_ALERT "md/raid1:%s: %s: unrecoverable I/O read error"
1465
				       " for block %llu\n",
N
NeilBrown 已提交
1466 1467
				       mdname(mddev),
				       bdevname(bio->bi_bdev, b),
1468 1469 1470 1471 1472 1473 1474 1475 1476
				       (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
1477
	}
1478 1479 1480 1481

	/*
	 * schedule writes
	 */
L
Linus Torvalds 已提交
1482 1483 1484
	atomic_set(&r1_bio->remaining, 1);
	for (i = 0; i < disks ; i++) {
		wbio = r1_bio->bios[i];
1485 1486 1487 1488
		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
1489 1490
			continue;

1491 1492
		wbio->bi_rw = WRITE;
		wbio->bi_end_io = end_sync_write;
L
Linus Torvalds 已提交
1493 1494
		atomic_inc(&r1_bio->remaining);
		md_sync_acct(conf->mirrors[i].rdev->bdev, wbio->bi_size >> 9);
1495

L
Linus Torvalds 已提交
1496 1497 1498 1499
		generic_make_request(wbio);
	}

	if (atomic_dec_and_test(&r1_bio->remaining)) {
1500
		/* if we're here, all write(s) have completed, so clean up */
L
Linus Torvalds 已提交
1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513
		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.
 */

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 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588
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
1589
					       "md/raid1:%s: read error corrected "
1590 1591
					       "(%d sectors at %llu on %s)\n",
					       mdname(mddev), s,
1592 1593
					       (unsigned long long)(sect +
					           rdev->data_offset),
1594 1595 1596 1597 1598 1599 1600 1601 1602
					       bdevname(rdev->bdev, b));
				}
			}
		}
		sectors -= s;
		sect += s;
	}
}

L
Linus Torvalds 已提交
1603 1604 1605 1606 1607
static void raid1d(mddev_t *mddev)
{
	r1bio_t *r1_bio;
	struct bio *bio;
	unsigned long flags;
1608
	conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
1609 1610 1611 1612 1613 1614 1615 1616
	struct list_head *head = &conf->retry_list;
	int unplug=0;
	mdk_rdev_t *rdev;

	md_check_recovery(mddev);
	
	for (;;) {
		char b[BDEVNAME_SIZE];
1617

1618
		unplug += flush_pending_writes(conf);
1619

1620 1621 1622
		spin_lock_irqsave(&conf->device_lock, flags);
		if (list_empty(head)) {
			spin_unlock_irqrestore(&conf->device_lock, flags);
L
Linus Torvalds 已提交
1623
			break;
1624
		}
L
Linus Torvalds 已提交
1625 1626
		r1_bio = list_entry(head->prev, r1bio_t, retry_list);
		list_del(head->prev);
1627
		conf->nr_queued--;
L
Linus Torvalds 已提交
1628 1629 1630
		spin_unlock_irqrestore(&conf->device_lock, flags);

		mddev = r1_bio->mddev;
1631
		conf = mddev->private;
L
Linus Torvalds 已提交
1632 1633 1634
		if (test_bit(R1BIO_IsSync, &r1_bio->state)) {
			sync_request_write(mddev, r1_bio);
			unplug = 1;
1635 1636
		} else if (test_bit(R1BIO_BarrierRetry, &r1_bio->state)) {
			/* some requests in the r1bio were BIO_RW_BARRIER
1637
			 * requests which failed with -EOPNOTSUPP.  Hohumm..
1638 1639 1640
			 * Better resubmit without the barrier.
			 * We know which devices to resubmit for, because
			 * all others have had their bios[] entry cleared.
1641
			 * We already have a nr_pending reference on these rdevs.
1642 1643
			 */
			int i;
1644
			const bool do_sync = bio_rw_flagged(r1_bio->master_bio, BIO_RW_SYNCIO);
1645 1646
			clear_bit(R1BIO_BarrierRetry, &r1_bio->state);
			clear_bit(R1BIO_Barrier, &r1_bio->state);
1647 1648 1649
			for (i=0; i < conf->raid_disks; i++)
				if (r1_bio->bios[i])
					atomic_inc(&r1_bio->remaining);
1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664
			for (i=0; i < conf->raid_disks; i++)
				if (r1_bio->bios[i]) {
					struct bio_vec *bvec;
					int j;

					bio = bio_clone(r1_bio->master_bio, GFP_NOIO);
					/* copy pages from the failed bio, as
					 * this might be a write-behind device */
					__bio_for_each_segment(bvec, bio, j, 0)
						bvec->bv_page = bio_iovec_idx(r1_bio->bios[i], j)->bv_page;
					bio_put(r1_bio->bios[i]);
					bio->bi_sector = r1_bio->sector +
						conf->mirrors[i].rdev->data_offset;
					bio->bi_bdev = conf->mirrors[i].rdev->bdev;
					bio->bi_end_io = raid1_end_write_request;
1665 1666
					bio->bi_rw = WRITE |
						(do_sync << BIO_RW_SYNCIO);
1667 1668 1669 1670
					bio->bi_private = r1_bio;
					r1_bio->bios[i] = bio;
					generic_make_request(bio);
				}
L
Linus Torvalds 已提交
1671 1672
		} else {
			int disk;
1673 1674 1675 1676 1677 1678 1679 1680 1681

			/* 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
			 */
1682 1683 1684 1685 1686 1687
			if (mddev->ro == 0) {
				freeze_array(conf);
				fix_read_error(conf, r1_bio->read_disk,
					       r1_bio->sector,
					       r1_bio->sectors);
				unfreeze_array(conf);
1688 1689 1690
			} else
				md_error(mddev,
					 conf->mirrors[r1_bio->read_disk].rdev);
1691

L
Linus Torvalds 已提交
1692
			bio = r1_bio->bios[r1_bio->read_disk];
1693
			if ((disk=read_balance(conf, r1_bio)) == -1) {
N
NeilBrown 已提交
1694
				printk(KERN_ALERT "md/raid1:%s: %s: unrecoverable I/O"
L
Linus Torvalds 已提交
1695
				       " read error for block %llu\n",
N
NeilBrown 已提交
1696
				       mdname(mddev),
L
Linus Torvalds 已提交
1697 1698 1699 1700
				       bdevname(bio->bi_bdev,b),
				       (unsigned long long)r1_bio->sector);
				raid_end_bio_io(r1_bio);
			} else {
1701
				const bool do_sync = bio_rw_flagged(r1_bio->master_bio, BIO_RW_SYNCIO);
1702 1703
				r1_bio->bios[r1_bio->read_disk] =
					mddev->ro ? IO_BLOCKED : NULL;
L
Linus Torvalds 已提交
1704 1705 1706 1707 1708 1709
				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
1710
					printk(KERN_ERR "md/raid1:%s: redirecting sector %llu to"
1711
					       " other mirror: %s\n",
N
NeilBrown 已提交
1712
					       mdname(mddev),
1713 1714
					       (unsigned long long)r1_bio->sector,
					       bdevname(rdev->bdev,b));
L
Linus Torvalds 已提交
1715 1716 1717
				bio->bi_sector = r1_bio->sector + rdev->data_offset;
				bio->bi_bdev = rdev->bdev;
				bio->bi_end_io = raid1_end_read_request;
1718
				bio->bi_rw = READ | (do_sync << BIO_RW_SYNCIO);
L
Linus Torvalds 已提交
1719 1720 1721 1722 1723
				bio->bi_private = r1_bio;
				unplug = 1;
				generic_make_request(bio);
			}
		}
N
NeilBrown 已提交
1724
		cond_resched();
L
Linus Torvalds 已提交
1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735
	}
	if (unplug)
		unplug_slaves(mddev);
}


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

	buffs = RESYNC_WINDOW / RESYNC_BLOCK_SIZE;
1736
	BUG_ON(conf->r1buf_pool);
L
Linus Torvalds 已提交
1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754
	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.
 */

1755
static sector_t sync_request(mddev_t *mddev, sector_t sector_nr, int *skipped, int go_faster)
L
Linus Torvalds 已提交
1756
{
1757
	conf_t *conf = mddev->private;
L
Linus Torvalds 已提交
1758 1759 1760
	r1bio_t *r1_bio;
	struct bio *bio;
	sector_t max_sector, nr_sectors;
1761
	int disk = -1;
L
Linus Torvalds 已提交
1762
	int i;
1763 1764
	int wonly = -1;
	int write_targets = 0, read_targets = 0;
1765
	int sync_blocks;
1766
	int still_degraded = 0;
L
Linus Torvalds 已提交
1767 1768 1769

	if (!conf->r1buf_pool)
		if (init_resync(conf))
1770
			return 0;
L
Linus Torvalds 已提交
1771

A
Andre Noll 已提交
1772
	max_sector = mddev->dev_sectors;
L
Linus Torvalds 已提交
1773
	if (sector_nr >= max_sector) {
1774 1775 1776 1777 1778
		/* 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
		 */
1779 1780
		if (mddev->curr_resync < max_sector) /* aborted */
			bitmap_end_sync(mddev->bitmap, mddev->curr_resync,
1781
						&sync_blocks, 1);
1782
		else /* completed sync */
1783
			conf->fullsync = 0;
1784 1785

		bitmap_close_sync(mddev->bitmap);
L
Linus Torvalds 已提交
1786 1787 1788 1789
		close_sync(conf);
		return 0;
	}

1790 1791
	if (mddev->bitmap == NULL &&
	    mddev->recovery_cp == MaxSector &&
1792
	    !test_bit(MD_RECOVERY_REQUESTED, &mddev->recovery) &&
1793 1794 1795 1796
	    conf->fullsync == 0) {
		*skipped = 1;
		return max_sector - sector_nr;
	}
1797 1798 1799
	/* before building a request, check if we can skip these blocks..
	 * This call the bitmap_start_sync doesn't actually record anything
	 */
1800
	if (!bitmap_start_sync(mddev->bitmap, sector_nr, &sync_blocks, 1) &&
1801
	    !conf->fullsync && !test_bit(MD_RECOVERY_REQUESTED, &mddev->recovery)) {
1802 1803 1804 1805
		/* We can skip this block, and probably several more */
		*skipped = 1;
		return sync_blocks;
	}
L
Linus Torvalds 已提交
1806
	/*
1807 1808 1809
	 * 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
1810
	 */
1811
	if (!go_faster && conf->nr_waiting)
L
Linus Torvalds 已提交
1812
		msleep_interruptible(1000);
1813

N
NeilBrown 已提交
1814
	bitmap_cond_end_sync(mddev->bitmap, sector_nr);
1815 1816 1817
	raise_barrier(conf);

	conf->next_resync = sector_nr;
L
Linus Torvalds 已提交
1818

1819 1820
	r1_bio = mempool_alloc(conf->r1buf_pool, GFP_NOIO);
	rcu_read_lock();
L
Linus Torvalds 已提交
1821
	/*
1822 1823 1824 1825 1826 1827
	 * 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
1828 1829 1830 1831
	 */

	r1_bio->mddev = mddev;
	r1_bio->sector = sector_nr;
1832
	r1_bio->state = 0;
L
Linus Torvalds 已提交
1833 1834 1835
	set_bit(R1BIO_IsSync, &r1_bio->state);

	for (i=0; i < conf->raid_disks; i++) {
1836
		mdk_rdev_t *rdev;
L
Linus Torvalds 已提交
1837 1838 1839 1840 1841
		bio = r1_bio->bios[i];

		/* take from bio_init */
		bio->bi_next = NULL;
		bio->bi_flags |= 1 << BIO_UPTODATE;
1842
		bio->bi_rw = READ;
L
Linus Torvalds 已提交
1843 1844 1845 1846 1847 1848 1849
		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;

1850 1851 1852
		rdev = rcu_dereference(conf->mirrors[i].rdev);
		if (rdev == NULL ||
			   test_bit(Faulty, &rdev->flags)) {
1853 1854
			still_degraded = 1;
			continue;
1855
		} else if (!test_bit(In_sync, &rdev->flags)) {
L
Linus Torvalds 已提交
1856 1857 1858
			bio->bi_rw = WRITE;
			bio->bi_end_io = end_sync_write;
			write_targets ++;
1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874
		} 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
1875 1876
		bio->bi_private = r1_bio;
	}
1877 1878 1879 1880
	rcu_read_unlock();
	if (disk < 0)
		disk = wonly;
	r1_bio->read_disk = disk;
1881

1882 1883 1884 1885 1886
	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
1887 1888 1889
		/* There is nowhere to write, so all non-sync
		 * drives must be failed - so we are finished
		 */
1890 1891
		sector_t rv = max_sector - sector_nr;
		*skipped = 1;
L
Linus Torvalds 已提交
1892 1893 1894 1895
		put_buf(r1_bio);
		return rv;
	}

1896 1897
	if (max_sector > mddev->resync_max)
		max_sector = mddev->resync_max; /* Don't do IO beyond here */
L
Linus Torvalds 已提交
1898
	nr_sectors = 0;
1899
	sync_blocks = 0;
L
Linus Torvalds 已提交
1900 1901 1902 1903 1904 1905 1906
	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;
1907 1908
		if (sync_blocks == 0) {
			if (!bitmap_start_sync(mddev->bitmap, sector_nr,
1909 1910 1911
					       &sync_blocks, still_degraded) &&
			    !conf->fullsync &&
			    !test_bit(MD_RECOVERY_REQUESTED, &mddev->recovery))
1912
				break;
1913
			BUG_ON(sync_blocks < (PAGE_SIZE>>9));
1914 1915
			if (len > (sync_blocks<<9))
				len = sync_blocks<<9;
1916
		}
1917

L
Linus Torvalds 已提交
1918 1919 1920
		for (i=0 ; i < conf->raid_disks; i++) {
			bio = r1_bio->bios[i];
			if (bio->bi_end_io) {
1921
				page = bio->bi_io_vec[bio->bi_vcnt].bv_page;
L
Linus Torvalds 已提交
1922 1923
				if (bio_add_page(bio, page, len, 0) == 0) {
					/* stop here */
1924
					bio->bi_io_vec[bio->bi_vcnt].bv_page = page;
L
Linus Torvalds 已提交
1925 1926 1927
					while (i > 0) {
						i--;
						bio = r1_bio->bios[i];
1928 1929
						if (bio->bi_end_io==NULL)
							continue;
L
Linus Torvalds 已提交
1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940
						/* 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;
1941
		sync_blocks -= (len>>9);
L
Linus Torvalds 已提交
1942 1943 1944 1945
	} while (r1_bio->bios[disk]->bi_vcnt < RESYNC_PAGES);
 bio_full:
	r1_bio->sectors = nr_sectors;

1946 1947 1948 1949 1950 1951 1952 1953
	/* 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) {
1954
				md_sync_acct(bio->bi_bdev, nr_sectors);
1955 1956 1957 1958 1959 1960
				generic_make_request(bio);
			}
		}
	} else {
		atomic_set(&r1_bio->remaining, 1);
		bio = r1_bio->bios[r1_bio->read_disk];
1961
		md_sync_acct(bio->bi_bdev, nr_sectors);
1962
		generic_make_request(bio);
L
Linus Torvalds 已提交
1963

1964
	}
L
Linus Torvalds 已提交
1965 1966 1967
	return nr_sectors;
}

1968 1969 1970 1971 1972 1973 1974 1975
static sector_t raid1_size(mddev_t *mddev, sector_t sectors, int raid_disks)
{
	if (sectors)
		return sectors;

	return mddev->dev_sectors;
}

1976
static conf_t *setup_conf(mddev_t *mddev)
L
Linus Torvalds 已提交
1977 1978
{
	conf_t *conf;
1979
	int i;
L
Linus Torvalds 已提交
1980 1981
	mirror_info_t *disk;
	mdk_rdev_t *rdev;
1982
	int err = -ENOMEM;
L
Linus Torvalds 已提交
1983

1984
	conf = kzalloc(sizeof(conf_t), GFP_KERNEL);
L
Linus Torvalds 已提交
1985
	if (!conf)
1986
		goto abort;
L
Linus Torvalds 已提交
1987

1988
	conf->mirrors = kzalloc(sizeof(struct mirror_info)*mddev->raid_disks,
L
Linus Torvalds 已提交
1989 1990
				 GFP_KERNEL);
	if (!conf->mirrors)
1991
		goto abort;
L
Linus Torvalds 已提交
1992

1993 1994
	conf->tmppage = alloc_page(GFP_KERNEL);
	if (!conf->tmppage)
1995
		goto abort;
1996

1997
	conf->poolinfo = kzalloc(sizeof(*conf->poolinfo), GFP_KERNEL);
L
Linus Torvalds 已提交
1998
	if (!conf->poolinfo)
1999
		goto abort;
L
Linus Torvalds 已提交
2000 2001 2002 2003 2004
	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)
2005 2006
		goto abort;

2007
	conf->poolinfo->mddev = mddev;
L
Linus Torvalds 已提交
2008

2009
	spin_lock_init(&conf->device_lock);
2010
	list_for_each_entry(rdev, &mddev->disks, same_set) {
2011
		int disk_idx = rdev->raid_disk;
L
Linus Torvalds 已提交
2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025
		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);
2026
	init_waitqueue_head(&conf->wait_barrier);
L
Linus Torvalds 已提交
2027

2028 2029 2030
	bio_list_init(&conf->pending_bio_list);
	bio_list_init(&conf->flushing_bio_list);

2031
	conf->last_used = -1;
L
Linus Torvalds 已提交
2032 2033 2034 2035
	for (i = 0; i < conf->raid_disks; i++) {

		disk = conf->mirrors + i;

2036 2037
		if (!disk->rdev ||
		    !test_bit(In_sync, &disk->rdev->flags)) {
L
Linus Torvalds 已提交
2038
			disk->head_position = 0;
2039 2040
			if (disk->rdev)
				conf->fullsync = 1;
2041 2042 2043 2044 2045 2046
		} 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
2047
	}
2048 2049 2050

	err = -EIO;
	if (conf->last_used < 0) {
N
NeilBrown 已提交
2051
		printk(KERN_ERR "md/raid1:%s: no operational mirrors\n",
2052 2053 2054 2055 2056 2057 2058
		       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
2059
		       "md/raid1:%s: couldn't allocate thread\n",
2060 2061
		       mdname(mddev));
		goto abort;
2062
	}
L
Linus Torvalds 已提交
2063

2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084
	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
2085
		printk(KERN_ERR "md/raid1:%s: raid level not set to mirroring (%d)\n",
2086 2087 2088 2089
		       mdname(mddev), mddev->level);
		return -EIO;
	}
	if (mddev->reshape_position != MaxSector) {
N
NeilBrown 已提交
2090
		printk(KERN_ERR "md/raid1:%s: reshape_position set but not supported\n",
2091 2092 2093
		       mdname(mddev));
		return -EIO;
	}
L
Linus Torvalds 已提交
2094
	/*
2095 2096 2097
	 * 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
2098
	 */
2099 2100 2101 2102
	if (mddev->private == NULL)
		conf = setup_conf(mddev);
	else
		conf = mddev->private;
L
Linus Torvalds 已提交
2103

2104 2105
	if (IS_ERR(conf))
		return PTR_ERR(conf);
L
Linus Torvalds 已提交
2106

2107 2108 2109 2110 2111
	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
2112 2113
		 * violating it, so limit ->max_segments to 1 lying within
		 * a single page, as a one page request is never in violation.
2114
		 */
2115 2116 2117 2118 2119
		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
2120
	}
2121

2122 2123 2124 2125 2126 2127 2128 2129 2130 2131
	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;

2132
	if (mddev->recovery_cp != MaxSector)
N
NeilBrown 已提交
2133
		printk(KERN_NOTICE "md/raid1:%s: not clean"
2134 2135
		       " -- starting background reconstruction\n",
		       mdname(mddev));
L
Linus Torvalds 已提交
2136
	printk(KERN_INFO 
N
NeilBrown 已提交
2137
		"md/raid1:%s: active with %d out of %d mirrors\n",
L
Linus Torvalds 已提交
2138 2139
		mdname(mddev), mddev->raid_disks - mddev->degraded, 
		mddev->raid_disks);
2140

L
Linus Torvalds 已提交
2141 2142 2143
	/*
	 * Ok, everything is just fine now
	 */
2144 2145 2146 2147
	mddev->thread = conf->thread;
	conf->thread = NULL;
	mddev->private = conf;

2148
	md_set_array_sectors(mddev, raid1_size(mddev, 0, 0));
L
Linus Torvalds 已提交
2149

2150
	mddev->queue->unplug_fn = raid1_unplug;
2151 2152
	mddev->queue->backing_dev_info.congested_fn = raid1_congested;
	mddev->queue->backing_dev_info.congested_data = mddev;
2153
	md_integrity_register(mddev);
L
Linus Torvalds 已提交
2154 2155 2156 2157 2158
	return 0;
}

static int stop(mddev_t *mddev)
{
2159
	conf_t *conf = mddev->private;
2160 2161 2162
	struct bitmap *bitmap = mddev->bitmap;

	/* wait for behind writes to complete */
2163
	if (bitmap && atomic_read(&bitmap->behind_writes) > 0) {
N
NeilBrown 已提交
2164 2165
		printk(KERN_INFO "md/raid1:%s: behind writes in progress - waiting to stop.\n",
		       mdname(mddev));
2166
		/* need to kick something here to make sure I/O goes? */
2167 2168
		wait_event(bitmap->behind_wait,
			   atomic_read(&bitmap->behind_writes) == 0);
2169
	}
L
Linus Torvalds 已提交
2170

2171 2172 2173
	raise_barrier(conf);
	lower_barrier(conf);

L
Linus Torvalds 已提交
2174 2175 2176 2177 2178
	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);
2179 2180
	kfree(conf->mirrors);
	kfree(conf->poolinfo);
L
Linus Torvalds 已提交
2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194
	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.
	 */
2195
	md_set_array_sectors(mddev, raid1_size(mddev, sectors, 0));
D
Dan Williams 已提交
2196 2197
	if (mddev->array_sectors > raid1_size(mddev, sectors, 0))
		return -EINVAL;
2198
	set_capacity(mddev->gendisk, mddev->array_sectors);
2199
	revalidate_disk(mddev->gendisk);
D
Dan Williams 已提交
2200
	if (sectors > mddev->dev_sectors &&
2201
	    mddev->recovery_cp == MaxSector) {
A
Andre Noll 已提交
2202
		mddev->recovery_cp = mddev->dev_sectors;
L
Linus Torvalds 已提交
2203 2204
		set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
	}
D
Dan Williams 已提交
2205
	mddev->dev_sectors = sectors;
2206
	mddev->resync_max_sectors = sectors;
L
Linus Torvalds 已提交
2207 2208 2209
	return 0;
}

2210
static int raid1_reshape(mddev_t *mddev)
L
Linus Torvalds 已提交
2211 2212 2213 2214 2215 2216 2217 2218
{
	/* 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.
2219 2220 2221
	 *
	 * 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
2222 2223 2224 2225
	 */
	mempool_t *newpool, *oldpool;
	struct pool_info *newpoolinfo;
	mirror_info_t *newmirrors;
2226
	conf_t *conf = mddev->private;
2227
	int cnt, raid_disks;
2228
	unsigned long flags;
2229
	int d, d2, err;
L
Linus Torvalds 已提交
2230

2231
	/* Cannot change chunk_size, layout, or level */
2232
	if (mddev->chunk_sectors != mddev->new_chunk_sectors ||
2233 2234
	    mddev->layout != mddev->new_layout ||
	    mddev->level != mddev->new_level) {
2235
		mddev->new_chunk_sectors = mddev->chunk_sectors;
2236 2237 2238 2239 2240
		mddev->new_layout = mddev->layout;
		mddev->new_level = mddev->level;
		return -EINVAL;
	}

2241 2242 2243
	err = md_allow_write(mddev);
	if (err)
		return err;
2244

2245 2246
	raid_disks = mddev->raid_disks + mddev->delta_disks;

2247 2248 2249 2250 2251 2252
	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
2253
			return -EBUSY;
2254
	}
L
Linus Torvalds 已提交
2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267

	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;
	}
2268
	newmirrors = kzalloc(sizeof(struct mirror_info) * raid_disks, GFP_KERNEL);
L
Linus Torvalds 已提交
2269 2270 2271 2272 2273 2274
	if (!newmirrors) {
		kfree(newpoolinfo);
		mempool_destroy(newpool);
		return -ENOMEM;
	}

2275
	raise_barrier(conf);
L
Linus Torvalds 已提交
2276 2277 2278 2279

	/* ok, everything is stopped */
	oldpool = conf->r1bio_pool;
	conf->r1bio_pool = newpool;
2280

2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292
	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
2293 2294 2295
				       "md/raid1:%s: cannot register "
				       "%s\n",
				       mdname(mddev), nm);
2296
		}
2297 2298 2299
		if (rdev)
			newmirrors[d2++].rdev = rdev;
	}
L
Linus Torvalds 已提交
2300 2301 2302 2303 2304
	kfree(conf->mirrors);
	conf->mirrors = newmirrors;
	kfree(conf->poolinfo);
	conf->poolinfo = newpoolinfo;

2305
	spin_lock_irqsave(&conf->device_lock, flags);
L
Linus Torvalds 已提交
2306
	mddev->degraded += (raid_disks - conf->raid_disks);
2307
	spin_unlock_irqrestore(&conf->device_lock, flags);
L
Linus Torvalds 已提交
2308
	conf->raid_disks = mddev->raid_disks = raid_disks;
2309
	mddev->delta_disks = 0;
L
Linus Torvalds 已提交
2310

2311
	conf->last_used = 0; /* just make sure it is in-range */
2312
	lower_barrier(conf);
L
Linus Torvalds 已提交
2313 2314 2315 2316 2317 2318 2319 2320

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

	mempool_destroy(oldpool);
	return 0;
}

2321
static void raid1_quiesce(mddev_t *mddev, int state)
2322
{
2323
	conf_t *conf = mddev->private;
2324 2325

	switch(state) {
2326 2327 2328
	case 2: /* wake for suspend */
		wake_up(&conf->wait_barrier);
		break;
2329
	case 1:
2330
		raise_barrier(conf);
2331
		break;
2332
	case 0:
2333
		lower_barrier(conf);
2334 2335 2336 2337
		break;
	}
}

2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354
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
2355

2356
static struct mdk_personality raid1_personality =
L
Linus Torvalds 已提交
2357 2358
{
	.name		= "raid1",
2359
	.level		= 1,
L
Linus Torvalds 已提交
2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370
	.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,
2371
	.size		= raid1_size,
2372
	.check_reshape	= raid1_reshape,
2373
	.quiesce	= raid1_quiesce,
2374
	.takeover	= raid1_takeover,
L
Linus Torvalds 已提交
2375 2376 2377 2378
};

static int __init raid_init(void)
{
2379
	return register_md_personality(&raid1_personality);
L
Linus Torvalds 已提交
2380 2381 2382 2383
}

static void raid_exit(void)
{
2384
	unregister_md_personality(&raid1_personality);
L
Linus Torvalds 已提交
2385 2386 2387 2388 2389
}

module_init(raid_init);
module_exit(raid_exit);
MODULE_LICENSE("GPL");
2390
MODULE_DESCRIPTION("RAID1 (mirroring) personality for MD");
L
Linus Torvalds 已提交
2391
MODULE_ALIAS("md-personality-3"); /* RAID1 */
2392
MODULE_ALIAS("md-raid1");
2393
MODULE_ALIAS("md-level-1");