raid1.c 54.4 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1 2 3 4 5 6 7 8 9 10 11 12 13 14
/*
 * 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
 *
 * Fixes to reconstruction by Jakob stergaard" <jakob@ostenfeld.dk>
 * 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
 *
Linus Torvalds's avatar
Linus Torvalds committed
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 "dm-bio-list.h"
Linus Torvalds's avatar
Linus Torvalds committed
35
#include <linux/raid/raid1.h>
36 37 38 39 40 41 42 43
#include <linux/raid/bitmap.h>

#define DEBUG 0
#if DEBUG
#define PRINTK(x...) printk(x)
#else
#define PRINTK(x...)
#endif
Linus Torvalds's avatar
Linus Torvalds committed
44 45 46 47 48 49 50 51 52

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


static void unplug_slaves(mddev_t *mddev);

53 54
static void allow_barrier(conf_t *conf);
static void lower_barrier(conf_t *conf);
Linus Torvalds's avatar
Linus Torvalds committed
55

56
static void * r1bio_pool_alloc(gfp_t gfp_flags, void *data)
Linus Torvalds's avatar
Linus Torvalds committed
57 58 59 60 61 62
{
	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 */
63 64
	r1_bio = kzalloc(size, gfp_flags);
	if (!r1_bio)
Linus Torvalds's avatar
Linus Torvalds committed
65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80
		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)

81
static void * r1buf_pool_alloc(gfp_t gfp_flags, void *data)
Linus Torvalds's avatar
Linus Torvalds committed
82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105
{
	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
106 107 108
	 * the first bio.
	 * If this is a user-requested check/repair, allocate
	 * RESYNC_PAGES for each bio.
Linus Torvalds's avatar
Linus Torvalds committed
109
	 */
110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129
	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;
		}
	}
	/* 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;
Linus Torvalds's avatar
Linus Torvalds committed
130 131 132 133 134 135 136
	}

	r1_bio->master_bio = NULL;

	return r1_bio;

out_free_pages:
137 138
	for (i=0; i < RESYNC_PAGES ; i++)
		for (j=0 ; j < pi->raid_disks; j++)
139
			safe_put_page(r1_bio->bios[j]->bi_io_vec[i].bv_page);
140
	j = -1;
Linus Torvalds's avatar
Linus Torvalds committed
141 142 143 144 145 146 147 148 149 150
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;
151
	int i,j;
Linus Torvalds's avatar
Linus Torvalds committed
152 153
	r1bio_t *r1bio = __r1_bio;

154 155 156 157 158
	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)
159
				safe_put_page(r1bio->bios[j]->bi_io_vec[i].bv_page);
160
		}
Linus Torvalds's avatar
Linus Torvalds committed
161 162 163 164 165 166 167 168 169 170 171 172
	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;
173
		if (*bio && *bio != IO_BLOCKED)
Linus Torvalds's avatar
Linus Torvalds committed
174 175 176 177 178
			bio_put(*bio);
		*bio = NULL;
	}
}

179
static void free_r1bio(r1bio_t *r1_bio)
Linus Torvalds's avatar
Linus Torvalds committed
180 181 182 183 184 185 186
{
	conf_t *conf = mddev_to_conf(r1_bio->mddev);

	/*
	 * Wake up any possible resync thread that waits for the device
	 * to go idle.
	 */
187
	allow_barrier(conf);
Linus Torvalds's avatar
Linus Torvalds committed
188 189 190 191 192

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

193
static void put_buf(r1bio_t *r1_bio)
Linus Torvalds's avatar
Linus Torvalds committed
194 195
{
	conf_t *conf = mddev_to_conf(r1_bio->mddev);
196 197 198 199 200 201 202
	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);
	}
Linus Torvalds's avatar
Linus Torvalds committed
203 204 205

	mempool_free(r1_bio, conf->r1buf_pool);

206
	lower_barrier(conf);
Linus Torvalds's avatar
Linus Torvalds committed
207 208 209 210 211 212 213 214 215 216
}

static void reschedule_retry(r1bio_t *r1_bio)
{
	unsigned long flags;
	mddev_t *mddev = r1_bio->mddev;
	conf_t *conf = mddev_to_conf(mddev);

	spin_lock_irqsave(&conf->device_lock, flags);
	list_add(&r1_bio->retry_list, &conf->retry_list);
217
	conf->nr_queued ++;
Linus Torvalds's avatar
Linus Torvalds committed
218 219
	spin_unlock_irqrestore(&conf->device_lock, flags);

220
	wake_up(&conf->wait_barrier);
Linus Torvalds's avatar
Linus Torvalds committed
221 222 223 224 225 226 227 228 229 230 231 232
	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;

233 234 235 236 237 238 239 240 241 242 243
	/* 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);

		bio_endio(bio, bio->bi_size,
			test_bit(R1BIO_Uptodate, &r1_bio->state) ? 0 : -EIO);
	}
Linus Torvalds's avatar
Linus Torvalds committed
244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271
	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)
{
	conf_t *conf = mddev_to_conf(r1_bio->mddev);

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

static int raid1_end_read_request(struct bio *bio, unsigned int bytes_done, int error)
{
	int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
	r1bio_t * r1_bio = (r1bio_t *)(bio->bi_private);
	int mirror;
	conf_t *conf = mddev_to_conf(r1_bio->mddev);

	if (bio->bi_size)
		return 1;
	
	mirror = r1_bio->read_disk;
	/*
	 * this branch is our 'one mirror IO has finished' event handler:
	 */
272 273 274
	update_head_pos(mirror, r1_bio);

	if (uptodate || conf->working_disks <= 1) {
Linus Torvalds's avatar
Linus Torvalds committed
275 276 277 278 279 280 281 282 283
		/*
		 * 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.
		 */
284 285
		if (uptodate)
			set_bit(R1BIO_Uptodate, &r1_bio->state);
Linus Torvalds's avatar
Linus Torvalds committed
286 287

		raid_end_bio_io(r1_bio);
288
	} else {
Linus Torvalds's avatar
Linus Torvalds committed
289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306
		/*
		 * oops, read error:
		 */
		char b[BDEVNAME_SIZE];
		if (printk_ratelimit())
			printk(KERN_ERR "raid1: %s: rescheduling sector %llu\n",
			       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);
	return 0;
}

static int raid1_end_write_request(struct bio *bio, unsigned int bytes_done, int error)
{
	int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
	r1bio_t * r1_bio = (r1bio_t *)(bio->bi_private);
307
	int mirror, behind = test_bit(R1BIO_BehindIO, &r1_bio->state);
Linus Torvalds's avatar
Linus Torvalds committed
308 309 310 311 312 313 314 315 316
	conf_t *conf = mddev_to_conf(r1_bio->mddev);

	if (bio->bi_size)
		return 1;

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

317 318 319 320 321
	if (error == -ENOTSUPP && test_bit(R1BIO_Barrier, &r1_bio->state)) {
		set_bit(BarriersNotsupp, &conf->mirrors[mirror].rdev->flags);
		set_bit(R1BIO_BarrierRetry, &r1_bio->state);
		r1_bio->mddev->barriers_work = 0;
	} else {
Linus Torvalds's avatar
Linus Torvalds committed
322
		/*
323
		 * this branch is our 'one mirror IO has finished' event handler:
Linus Torvalds's avatar
Linus Torvalds committed
324
		 */
325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363
		r1_bio->bios[mirror] = NULL;
		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);
					bio_endio(mbio, mbio->bi_size, 0);
				}
364 365 366
			}
		}
	}
Linus Torvalds's avatar
Linus Torvalds committed
367 368 369 370 371 372
	/*
	 *
	 * Let's see if all mirrored write operations have finished
	 * already.
	 */
	if (atomic_dec_and_test(&r1_bio->remaining)) {
373 374 375 376 377 378 379
		if (test_bit(R1BIO_BarrierRetry, &r1_bio->state)) {
			reschedule_retry(r1_bio);
			/* Don't dec_pending yet, we want to hold
			 * the reference over the retry
			 */
			return 0;
		}
380 381 382 383
		if (test_bit(R1BIO_BehindIO, &r1_bio->state)) {
			/* free extra copy of the data pages */
			int i = bio->bi_vcnt;
			while (i--)
384
				safe_put_page(bio->bi_io_vec[i].bv_page);
385
		}
386 387 388
		/* clear the bitmap if all writes complete successfully */
		bitmap_endwrite(r1_bio->mddev->bitmap, r1_bio->sector,
				r1_bio->sectors,
389 390
				!test_bit(R1BIO_Degraded, &r1_bio->state),
				behind);
Linus Torvalds's avatar
Linus Torvalds committed
391 392 393 394
		md_write_end(r1_bio->mddev);
		raid_end_bio_io(r1_bio);
	}

395 396 397
	if (r1_bio->bios[mirror]==NULL)
		bio_put(bio);

Linus Torvalds's avatar
Linus Torvalds committed
398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420
	rdev_dec_pending(conf->mirrors[mirror].rdev, conf->mddev);
	return 0;
}


/*
 * 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)
{
	const unsigned long this_sector = r1_bio->sector;
	int new_disk = conf->last_used, disk = new_disk;
421
	int wonly_disk = -1;
Linus Torvalds's avatar
Linus Torvalds committed
422 423
	const int sectors = r1_bio->sectors;
	sector_t new_distance, current_distance;
424
	mdk_rdev_t *rdev;
Linus Torvalds's avatar
Linus Torvalds committed
425 426 427

	rcu_read_lock();
	/*
428
	 * Check if we can balance. We can balance on the whole
Linus Torvalds's avatar
Linus Torvalds committed
429 430 431 432 433 434 435 436 437
	 * 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)) {
		/* Choose the first operation device, for consistancy */
		new_disk = 0;

438
		for (rdev = rcu_dereference(conf->mirrors[new_disk].rdev);
439
		     r1_bio->bios[new_disk] == IO_BLOCKED ||
440
		     !rdev || !test_bit(In_sync, &rdev->flags)
441
			     || test_bit(WriteMostly, &rdev->flags);
442
		     rdev = rcu_dereference(conf->mirrors[++new_disk].rdev)) {
443

444 445
			if (rdev && test_bit(In_sync, &rdev->flags) &&
				r1_bio->bios[new_disk] != IO_BLOCKED)
446 447 448 449
				wonly_disk = new_disk;

			if (new_disk == conf->raid_disks - 1) {
				new_disk = wonly_disk;
Linus Torvalds's avatar
Linus Torvalds committed
450 451 452 453 454 455 456 457
				break;
			}
		}
		goto rb_out;
	}


	/* make sure the disk is operational */
458
	for (rdev = rcu_dereference(conf->mirrors[new_disk].rdev);
459
	     r1_bio->bios[new_disk] == IO_BLOCKED ||
460
	     !rdev || !test_bit(In_sync, &rdev->flags) ||
461
		     test_bit(WriteMostly, &rdev->flags);
462
	     rdev = rcu_dereference(conf->mirrors[new_disk].rdev)) {
463

464 465
		if (rdev && test_bit(In_sync, &rdev->flags) &&
		    r1_bio->bios[new_disk] != IO_BLOCKED)
466 467
			wonly_disk = new_disk;

Linus Torvalds's avatar
Linus Torvalds committed
468 469 470 471
		if (new_disk <= 0)
			new_disk = conf->raid_disks;
		new_disk--;
		if (new_disk == disk) {
472 473
			new_disk = wonly_disk;
			break;
Linus Torvalds's avatar
Linus Torvalds committed
474 475
		}
	}
476 477 478 479

	if (new_disk < 0)
		goto rb_out;

Linus Torvalds's avatar
Linus Torvalds committed
480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499
	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--;

500
		rdev = rcu_dereference(conf->mirrors[disk].rdev);
501

502
		if (!rdev || r1_bio->bios[disk] == IO_BLOCKED ||
503
		    !test_bit(In_sync, &rdev->flags) ||
504
		    test_bit(WriteMostly, &rdev->flags))
Linus Torvalds's avatar
Linus Torvalds committed
505 506 507 508 509 510 511 512 513 514 515 516 517
			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);

518
 rb_out:
Linus Torvalds's avatar
Linus Torvalds committed
519 520 521


	if (new_disk >= 0) {
522
		rdev = rcu_dereference(conf->mirrors[new_disk].rdev);
523 524 525
		if (!rdev)
			goto retry;
		atomic_inc(&rdev->nr_pending);
526
		if (!test_bit(In_sync, &rdev->flags)) {
Linus Torvalds's avatar
Linus Torvalds committed
527 528 529
			/* cannot risk returning a device that failed
			 * before we inc'ed nr_pending
			 */
530
			rdev_dec_pending(rdev, conf->mddev);
Linus Torvalds's avatar
Linus Torvalds committed
531 532
			goto retry;
		}
533 534
		conf->next_seq_sect = this_sector + sectors;
		conf->last_used = new_disk;
Linus Torvalds's avatar
Linus Torvalds committed
535 536 537 538 539 540 541 542 543 544 545 546 547
	}
	rcu_read_unlock();

	return new_disk;
}

static void unplug_slaves(mddev_t *mddev)
{
	conf_t *conf = mddev_to_conf(mddev);
	int i;

	rcu_read_lock();
	for (i=0; i<mddev->raid_disks; i++) {
548
		mdk_rdev_t *rdev = rcu_dereference(conf->mirrors[i].rdev);
549
		if (rdev && !test_bit(Faulty, &rdev->flags) && atomic_read(&rdev->nr_pending)) {
Linus Torvalds's avatar
Linus Torvalds committed
550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566
			request_queue_t *r_queue = bdev_get_queue(rdev->bdev);

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

			if (r_queue->unplug_fn)
				r_queue->unplug_fn(r_queue);

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

static void raid1_unplug(request_queue_t *q)
{
567 568 569 570
	mddev_t *mddev = q->queuedata;

	unplug_slaves(mddev);
	md_wakeup_thread(mddev->thread);
Linus Torvalds's avatar
Linus Torvalds committed
571 572 573 574 575 576 577 578 579 580 581
}

static int raid1_issue_flush(request_queue_t *q, struct gendisk *disk,
			     sector_t *error_sector)
{
	mddev_t *mddev = q->queuedata;
	conf_t *conf = mddev_to_conf(mddev);
	int i, ret = 0;

	rcu_read_lock();
	for (i=0; i<mddev->raid_disks && ret == 0; i++) {
582
		mdk_rdev_t *rdev = rcu_dereference(conf->mirrors[i].rdev);
583
		if (rdev && !test_bit(Faulty, &rdev->flags)) {
Linus Torvalds's avatar
Linus Torvalds committed
584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602
			struct block_device *bdev = rdev->bdev;
			request_queue_t *r_queue = bdev_get_queue(bdev);

			if (!r_queue->issue_flush_fn)
				ret = -EOPNOTSUPP;
			else {
				atomic_inc(&rdev->nr_pending);
				rcu_read_unlock();
				ret = r_queue->issue_flush_fn(r_queue, bdev->bd_disk,
							      error_sector);
				rdev_dec_pending(rdev, mddev);
				rcu_read_lock();
			}
		}
	}
	rcu_read_unlock();
	return ret;
}

603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622
/* 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.
Linus Torvalds's avatar
Linus Torvalds committed
623 624 625
 */
#define RESYNC_DEPTH 32

626
static void raise_barrier(conf_t *conf)
Linus Torvalds's avatar
Linus Torvalds committed
627 628
{
	spin_lock_irq(&conf->resync_lock);
629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664

	/* 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;
	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--;
Linus Torvalds's avatar
Linus Torvalds committed
665
	}
666
	conf->nr_pending++;
Linus Torvalds's avatar
Linus Torvalds committed
667 668 669
	spin_unlock_irq(&conf->resync_lock);
}

670 671 672 673 674 675 676 677 678
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);
}

679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704
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
	 * wait until barrier+nr_pending match nr_queued+2
	 */
	spin_lock_irq(&conf->resync_lock);
	conf->barrier++;
	conf->nr_waiting++;
	wait_event_lock_irq(conf->wait_barrier,
			    conf->barrier+conf->nr_pending == conf->nr_queued+2,
			    conf->resync_lock,
			    raid1_unplug(conf->mddev->queue));
	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);
}

705

706 707 708 709 710
/* duplicate the data pages for behind I/O */
static struct page **alloc_behind_pages(struct bio *bio)
{
	int i;
	struct bio_vec *bvec;
711
	struct page **pages = kzalloc(bio->bi_vcnt * sizeof(struct page *),
712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730
					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++)
731
			put_page(pages[i]);
732 733 734 735 736
	kfree(pages);
	PRINTK("%dB behind alloc failed, doing sync I/O\n", bio->bi_size);
	return NULL;
}

Linus Torvalds's avatar
Linus Torvalds committed
737 738 739 740 741 742 743
static int make_request(request_queue_t *q, struct bio * bio)
{
	mddev_t *mddev = q->queuedata;
	conf_t *conf = mddev_to_conf(mddev);
	mirror_info_t *mirror;
	r1bio_t *r1_bio;
	struct bio *read_bio;
744
	int i, targets = 0, disks;
Linus Torvalds's avatar
Linus Torvalds committed
745
	mdk_rdev_t *rdev;
746 747 748
	struct bitmap *bitmap = mddev->bitmap;
	unsigned long flags;
	struct bio_list bl;
749
	struct page **behind_pages = NULL;
750
	const int rw = bio_data_dir(bio);
751
	int do_barriers;
752

753
	if (unlikely(!mddev->barriers_work && bio_barrier(bio))) {
754 755 756
		bio_endio(bio, bio->bi_size, -EOPNOTSUPP);
		return 0;
	}
Linus Torvalds's avatar
Linus Torvalds committed
757 758 759 760 761 762

	/*
	 * 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.
	 */
763 764
	md_write_start(mddev, bio); /* wait on superblock update early */

765
	wait_barrier(conf);
Linus Torvalds's avatar
Linus Torvalds committed
766

767 768
	disk_stat_inc(mddev->gendisk, ios[rw]);
	disk_stat_add(mddev->gendisk, sectors[rw], bio_sectors(bio));
Linus Torvalds's avatar
Linus Torvalds committed
769 770 771 772 773 774 775 776 777 778

	/*
	 * 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;
779
	r1_bio->state = 0;
Linus Torvalds's avatar
Linus Torvalds committed
780 781 782
	r1_bio->mddev = mddev;
	r1_bio->sector = bio->bi_sector;

783
	if (rw == READ) {
Linus Torvalds's avatar
Linus Torvalds committed
784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819
		/*
		 * 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;

		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;
		read_bio->bi_rw = READ;
		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;
820 821 822 823 824 825 826
#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
Linus Torvalds's avatar
Linus Torvalds committed
827 828
	rcu_read_lock();
	for (i = 0;  i < disks; i++) {
829
		if ((rdev=rcu_dereference(conf->mirrors[i].rdev)) != NULL &&
830
		    !test_bit(Faulty, &rdev->flags)) {
Linus Torvalds's avatar
Linus Torvalds committed
831
			atomic_inc(&rdev->nr_pending);
832
			if (test_bit(Faulty, &rdev->flags)) {
833
				rdev_dec_pending(rdev, mddev);
Linus Torvalds's avatar
Linus Torvalds committed
834 835 836
				r1_bio->bios[i] = NULL;
			} else
				r1_bio->bios[i] = bio;
837
			targets++;
Linus Torvalds's avatar
Linus Torvalds committed
838 839 840 841 842
		} else
			r1_bio->bios[i] = NULL;
	}
	rcu_read_unlock();

843 844
	BUG_ON(targets == 0); /* we never fail the last device */

845 846 847 848 849 850
	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);
	}

851 852 853 854 855 856
	/* do behind I/O ? */
	if (bitmap &&
	    atomic_read(&bitmap->behind_writes) < bitmap->max_write_behind &&
	    (behind_pages = alloc_behind_pages(bio)) != NULL)
		set_bit(R1BIO_BehindIO, &r1_bio->state);

857
	atomic_set(&r1_bio->remaining, 0);
858
	atomic_set(&r1_bio->behind_remaining, 0);
859

860 861 862 863
	do_barriers = bio->bi_rw & BIO_RW_BARRIER;
	if (do_barriers)
		set_bit(R1BIO_Barrier, &r1_bio->state);

864
	bio_list_init(&bl);
Linus Torvalds's avatar
Linus Torvalds committed
865 866 867 868 869 870 871 872 873 874 875
	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;
876
		mbio->bi_rw = WRITE | do_barriers;
Linus Torvalds's avatar
Linus Torvalds committed
877 878
		mbio->bi_private = r1_bio;

879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895
		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);
		}

Linus Torvalds's avatar
Linus Torvalds committed
896 897
		atomic_inc(&r1_bio->remaining);

898
		bio_list_add(&bl, mbio);
Linus Torvalds's avatar
Linus Torvalds committed
899
	}
900
	kfree(behind_pages); /* the behind pages are attached to the bios now */
Linus Torvalds's avatar
Linus Torvalds committed
901

902 903
	bitmap_startwrite(bitmap, bio->bi_sector, r1_bio->sectors,
				test_bit(R1BIO_BehindIO, &r1_bio->state));
904 905 906 907 908 909 910 911 912 913 914 915
	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);

#if 0
	while ((bio = bio_list_pop(&bl)) != NULL)
		generic_make_request(bio);
#endif

Linus Torvalds's avatar
Linus Torvalds committed
916 917 918 919 920 921 922 923 924 925 926 927 928
	return 0;
}

static void status(struct seq_file *seq, mddev_t *mddev)
{
	conf_t *conf = mddev_to_conf(mddev);
	int i;

	seq_printf(seq, " [%d/%d] [", conf->raid_disks,
						conf->working_disks);
	for (i = 0; i < conf->raid_disks; i++)
		seq_printf(seq, "%s",
			      conf->mirrors[i].rdev &&
929
			      test_bit(In_sync, &conf->mirrors[i].rdev->flags) ? "U" : "_");
Linus Torvalds's avatar
Linus Torvalds committed
930 931 932 933 934 935 936 937 938 939 940 941 942 943 944
	seq_printf(seq, "]");
}


static void error(mddev_t *mddev, mdk_rdev_t *rdev)
{
	char b[BDEVNAME_SIZE];
	conf_t *conf = mddev_to_conf(mddev);

	/*
	 * 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
	 */
945
	if (test_bit(In_sync, &rdev->flags)
Linus Torvalds's avatar
Linus Torvalds committed
946 947 948 949 950 951
	    && conf->working_disks == 1)
		/*
		 * Don't fail the drive, act as though we were just a
		 * normal single drive
		 */
		return;
952
	if (test_bit(In_sync, &rdev->flags)) {
Linus Torvalds's avatar
Linus Torvalds committed
953 954 955 956 957 958 959
		mddev->degraded++;
		conf->working_disks--;
		/*
		 * if recovery is running, make sure it aborts.
		 */
		set_bit(MD_RECOVERY_ERR, &mddev->recovery);
	}
960 961
	clear_bit(In_sync, &rdev->flags);
	set_bit(Faulty, &rdev->flags);
Linus Torvalds's avatar
Linus Torvalds committed
962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985
	mddev->sb_dirty = 1;
	printk(KERN_ALERT "raid1: Disk failure on %s, disabling device. \n"
		"	Operation continuing on %d devices\n",
		bdevname(rdev->bdev,b), conf->working_disks);
}

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

	printk("RAID1 conf printout:\n");
	if (!conf) {
		printk("(!conf)\n");
		return;
	}
	printk(" --- wd:%d rd:%d\n", conf->working_disks,
		conf->raid_disks);

	for (i = 0; i < conf->raid_disks; i++) {
		char b[BDEVNAME_SIZE];
		tmp = conf->mirrors + i;
		if (tmp->rdev)
			printk(" disk %d, wo:%d, o:%d, dev:%s\n",
986
				i, !test_bit(In_sync, &tmp->rdev->flags), !test_bit(Faulty, &tmp->rdev->flags),
Linus Torvalds's avatar
Linus Torvalds committed
987 988 989 990 991 992
				bdevname(tmp->rdev->bdev,b));
	}
}

static void close_sync(conf_t *conf)
{
993 994
	wait_barrier(conf);
	allow_barrier(conf);
Linus Torvalds's avatar
Linus Torvalds committed
995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012

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

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

	/*
	 * Find all failed disks within the RAID1 configuration 
	 * and mark them readable
	 */
	for (i = 0; i < conf->raid_disks; i++) {
		tmp = conf->mirrors + i;
		if (tmp->rdev 
1013 1014
		    && !test_bit(Faulty, &tmp->rdev->flags)
		    && !test_bit(In_sync, &tmp->rdev->flags)) {
Linus Torvalds's avatar
Linus Torvalds committed
1015 1016
			conf->working_disks++;
			mddev->degraded--;
1017
			set_bit(In_sync, &tmp->rdev->flags);
Linus Torvalds's avatar
Linus Torvalds committed
1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029
		}
	}

	print_conf(conf);
	return 0;
}


static int raid1_add_disk(mddev_t *mddev, mdk_rdev_t *rdev)
{
	conf_t *conf = mddev->private;
	int found = 0;
1030
	int mirror = 0;
Linus Torvalds's avatar
Linus Torvalds committed
1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048
	mirror_info_t *p;

	for (mirror=0; mirror < mddev->raid_disks; mirror++)
		if ( !(p=conf->mirrors+mirror)->rdev) {

			blk_queue_stack_limits(mddev->queue,
					       rdev->bdev->bd_disk->queue);
			/* as we don't honour merge_bvec_fn, we must never risk
			 * violating it, so limit ->max_sector to one PAGE, as
			 * a one page request is never in violation.
			 */
			if (rdev->bdev->bd_disk->queue->merge_bvec_fn &&
			    mddev->queue->max_sectors > (PAGE_SIZE>>9))
				blk_queue_max_sectors(mddev->queue, PAGE_SIZE>>9);

			p->head_position = 0;
			rdev->raid_disk = mirror;
			found = 1;
1049 1050 1051 1052
			/* 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)
1053
				conf->fullsync = 1;
1054
			rcu_assign_pointer(p->rdev, rdev);
Linus Torvalds's avatar
Linus Torvalds committed
1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071
			break;
		}

	print_conf(conf);
	return found;
}

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) {
1072
		if (test_bit(In_sync, &rdev->flags) ||
Linus Torvalds's avatar
Linus Torvalds committed
1073 1074 1075 1076 1077
		    atomic_read(&rdev->nr_pending)) {
			err = -EBUSY;
			goto abort;
		}
		p->rdev = NULL;
1078
		synchronize_rcu();
Linus Torvalds's avatar
Linus Torvalds committed
1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094
		if (atomic_read(&rdev->nr_pending)) {
			/* lost the race, try later */
			err = -EBUSY;
			p->rdev = rdev;
		}
	}
abort:

	print_conf(conf);
	return err;
}


static int end_sync_read(struct bio *bio, unsigned int bytes_done, int error)
{
	r1bio_t * r1_bio = (r1bio_t *)(bio->bi_private);
1095
	int i;
Linus Torvalds's avatar
Linus Torvalds committed
1096 1097 1098 1099

	if (bio->bi_size)
		return 1;

1100 1101 1102 1103 1104
	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);
Linus Torvalds's avatar
Linus Torvalds committed
1105 1106 1107 1108 1109
	/*
	 * 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
	 */
1110
	if (test_bit(BIO_UPTODATE, &bio->bi_flags))
Linus Torvalds's avatar
Linus Torvalds committed
1111
		set_bit(R1BIO_Uptodate, &r1_bio->state);
1112 1113 1114

	if (atomic_dec_and_test(&r1_bio->remaining))
		reschedule_retry(r1_bio);
Linus Torvalds's avatar
Linus Torvalds committed
1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134
	return 0;
}

static int end_sync_write(struct bio *bio, unsigned int bytes_done, int error)
{
	int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
	r1bio_t * r1_bio = (r1bio_t *)(bio->bi_private);
	mddev_t *mddev = r1_bio->mddev;
	conf_t *conf = mddev_to_conf(mddev);
	int i;
	int mirror=0;

	if (bio->bi_size)
		return 1;

	for (i = 0; i < conf->raid_disks; i++)
		if (r1_bio->bios[i] == bio) {
			mirror = i;
			break;
		}
1135
	if (!uptodate)
Linus Torvalds's avatar
Linus Torvalds committed
1136
		md_error(mddev, conf->mirrors[mirror].rdev);
1137

Linus Torvalds's avatar
Linus Torvalds committed
1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155
	update_head_pos(mirror, r1_bio);

	if (atomic_dec_and_test(&r1_bio->remaining)) {
		md_done_sync(mddev, r1_bio->sectors, uptodate);
		put_buf(r1_bio);
	}
	return 0;
}

static void sync_request_write(mddev_t *mddev, r1bio_t *r1_bio)
{
	conf_t *conf = mddev_to_conf(mddev);
	int i;
	int disks = conf->raid_disks;
	struct bio *bio, *wbio;

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

1156

1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178
	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;
1179
				rdev_dec_pending(conf->mirrors[primary].rdev, mddev);
1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196
				break;
			}
		r1_bio->read_disk = primary;
		for (i=0; i<mddev->raid_disks; i++)
			if (r1_bio->bios[i]->bi_end_io == end_sync_read &&
			    test_bit(BIO_UPTODATE, &r1_bio->bios[i]->bi_flags)) {
				int j;
				int vcnt = r1_bio->sectors >> (PAGE_SHIFT- 9);
				struct bio *pbio = r1_bio->bios[primary];
				struct bio *sbio = r1_bio->bios[i];
				for (j = vcnt; j-- ; )
					if (memcmp(page_address(pbio->bi_io_vec[j].bv_page),
						   page_address(sbio->bi_io_vec[j].bv_page),
						   PAGE_SIZE))
						break;
				if (j >= 0)
					mddev->resync_mismatches += r1_bio->sectors;
1197
				if (j < 0 || test_bit(MD_RECOVERY_CHECK, &mddev->recovery)) {
1198
					sbio->bi_end_io = NULL;
1199 1200
					rdev_dec_pending(conf->mirrors[i].rdev, mddev);
				} else {
1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217
					/* fixup the bio for reuse */
					sbio->bi_vcnt = vcnt;
					sbio->bi_size = r1_bio->sectors << 9;
					sbio->bi_idx = 0;
					sbio->bi_phys_segments = 0;
					sbio->bi_hw_segments = 0;
					sbio->bi_hw_front_size = 0;
					sbio->bi_hw_back_size = 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;
				}
			}
	}
Linus Torvalds's avatar
Linus Torvalds committed
1218
	if (!test_bit(R1BIO_Uptodate, &r1_bio->state)) {
1219 1220 1221 1222 1223 1224 1225 1226
		/* 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
		 * read into the pages we already have so they we don't
		 * 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.
Linus Torvalds's avatar
Linus Torvalds committed
1227
		 */
1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257
		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) {
					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) {
1258
				int start = d;
1259 1260 1261 1262 1263 1264 1265 1266 1267
				/* 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;
1268
					atomic_add(s, &rdev->corrected_errors);
1269 1270 1271 1272
					if (sync_page_io(rdev->bdev,
							 sect + rdev->data_offset,
							 s<<9,
							 bio->bi_io_vec[idx].bv_page,
1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284
							 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,
1285 1286 1287
							 sect + rdev->data_offset,
							 s<<9,
							 bio->bi_io_vec[idx].bv_page,
1288
							 READ) == 0)
1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306
						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);
				printk(KERN_ALERT "raid1: %s: unrecoverable I/O read error"
				       " for block %llu\n",
				       bdevname(bio->bi_bdev,b),
				       (unsigned long long)r1_bio->sector);
				md_done_sync(mddev, r1_bio->sectors, 0);
				put_buf(r1_bio);
				return;
			}
			sectors -= s;
			sect += s;
			idx ++;
		}
Linus Torvalds's avatar
Linus Torvalds committed
1307
	}
1308 1309 1310 1311

	/*
	 * schedule writes
	 */
Linus Torvalds's avatar
Linus Torvalds committed
1312 1313 1314
	atomic_set(&r1_bio->remaining, 1);
	for (i = 0; i < disks ; i++) {
		wbio = r1_bio->bios[i];
1315 1316 1317 1318
		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))))
Linus Torvalds's avatar
Linus Torvalds committed
1319 1320
			continue;

1321 1322
		wbio->bi_rw = WRITE;
		wbio->bi_end_io = end_sync_write;
Linus Torvalds's avatar
Linus Torvalds committed
1323 1324
		atomic_inc(&r1_bio->remaining);
		md_sync_acct(conf->mirrors[i].rdev->bdev, wbio->bi_size >> 9);
1325

Linus Torvalds's avatar
Linus Torvalds committed
1326 1327 1328 1329
		generic_make_request(wbio);
	}

	if (atomic_dec_and_test(&r1_bio->remaining)) {
1330
		/* if we're here, all write(s) have completed, so clean up */
Linus Torvalds's avatar
Linus Torvalds committed
1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358
		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.
 */

static void raid1d(mddev_t *mddev)
{
	r1bio_t *r1_bio;
	struct bio *bio;
	unsigned long flags;
	conf_t *conf = mddev_to_conf(mddev);
	struct list_head *head = &conf->retry_list;
	int unplug=0;
	mdk_rdev_t *rdev;

	md_check_recovery(mddev);
	
	for (;;) {
		char b[BDEVNAME_SIZE];
		spin_lock_irqsave(&conf->device_lock, flags);
1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378

		if (conf->pending_bio_list.head) {
			bio = bio_list_get(&conf->pending_bio_list);
			blk_remove_plug(mddev->queue);
			spin_unlock_irqrestore(&conf->device_lock, flags);
			/* flush any pending bitmap writes to disk before proceeding w/ I/O */
			if (bitmap_unplug(mddev->bitmap) != 0)
				printk("%s: bitmap file write failed!\n", mdname(mddev));

			while (bio) { /* submit pending writes */
				struct bio *next = bio->bi_next;
				bio->bi_next = NULL;
				generic_make_request(bio);
				bio = next;
			}
			unplug = 1;

			continue;
		}

Linus Torvalds's avatar
Linus Torvalds committed
1379 1380 1381 1382
		if (list_empty(head))
			break;
		r1_bio = list_entry(head->prev, r1bio_t, retry_list);
		list_del(head->prev);
1383
		conf->nr_queued--;
Linus Torvalds's avatar
Linus Torvalds committed
1384 1385 1386 1387 1388 1389 1390
		spin_unlock_irqrestore(&conf->device_lock, flags);

		mddev = r1_bio->mddev;
		conf = mddev_to_conf(mddev);
		if (test_bit(R1BIO_IsSync, &r1_bio->state)) {
			sync_request_write(mddev, r1_bio);
			unplug = 1;
1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420
		} else if (test_bit(R1BIO_BarrierRetry, &r1_bio->state)) {
			/* some requests in the r1bio were BIO_RW_BARRIER
			 * requests which failed with -ENOTSUPP.  Hohumm..
			 * Better resubmit without the barrier.
			 * We know which devices to resubmit for, because
			 * all others have had their bios[] entry cleared.
			 */
			int i;
			clear_bit(R1BIO_BarrierRetry, &r1_bio->state);
			clear_bit(R1BIO_Barrier, &r1_bio->state);
			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;
					bio->bi_rw = WRITE;
					bio->bi_private = r1_bio;
					r1_bio->bios[i] = bio;
					generic_make_request(bio);
				}
Linus Torvalds's avatar
Linus Torvalds committed
1421 1422
		} else {
			int disk;
1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434

			/* 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
			 */
			sector_t sect = r1_bio->sector;
			int sectors = r1_bio->sectors;
			freeze_array(conf);
1435
			if (mddev->ro == 0) while(sectors) {
1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460
				int s = sectors;
				int d = r1_bio->read_disk;
				int success = 0;

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

				do {
					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 != r1_bio->read_disk);

				if (success) {
					/* write it back and re-read */
1461
					int start = d;
1462 1463 1464 1465 1466
					while (d != r1_bio->read_disk) {
						if (d==0)
							d = conf->raid_disks;
						d--;
						rdev = conf->mirrors[d].rdev;
1467
						atomic_add(s, &rdev->corrected_errors);
1468 1469 1470 1471
						if (rdev &&
						    test_bit(In_sync, &rdev->flags)) {
							if (sync_page_io(rdev->bdev,
									 sect + rdev->data_offset,
1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485
									 s<<9, conf->tmppage, WRITE) == 0)
								/* Well, this device is dead */
								md_error(mddev, rdev);
						}
					}
					d = start;
					while (d != r1_bio->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,
1486
									 sect + rdev->data_offset,
1487
									 s<<9, conf->tmppage, READ) == 0)
1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502
								/* Well, this device is dead */
								md_error(mddev, rdev);
						}
					}
				} else {
					/* Cannot read from anywhere -- bye bye array */
					md_error(mddev, conf->mirrors[r1_bio->read_disk].rdev);
					break;
				}
				sectors -= s;
				sect += s;
			}

			unfreeze_array(conf);

Linus Torvalds's avatar
Linus Torvalds committed
1503 1504 1505 1506 1507 1508 1509 1510
			bio = r1_bio->bios[r1_bio->read_disk];
			if ((disk=read_balance(conf, r1_bio)) == -1) {
				printk(KERN_ALERT "raid1: %s: unrecoverable I/O"
				       " read error for block %llu\n",
				       bdevname(bio->bi_bdev,b),
				       (unsigned long long)r1_bio->sector);
				raid_end_bio_io(r1_bio);
			} else {
1511 1512
				r1_bio->bios[r1_bio->read_disk] =
					mddev->ro ? IO_BLOCKED : NULL;
Linus Torvalds's avatar
Linus Torvalds committed
1513 1514 1515 1516 1517 1518 1519 1520