raid5.c 160.5 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4
/*
 * raid5.c : Multiple Devices driver for Linux
 *	   Copyright (C) 1996, 1997 Ingo Molnar, Miguel de Icaza, Gadi Oxman
 *	   Copyright (C) 1999, 2000 Ingo Molnar
5
 *	   Copyright (C) 2002, 2003 H. Peter Anvin
L
Linus Torvalds 已提交
6
 *
7 8 9
 * RAID-4/5/6 management functions.
 * Thanks to Penguin Computing for making the RAID-6 development possible
 * by donating a test server!
L
Linus Torvalds 已提交
10 11 12 13 14 15 16 17 18 19 20
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2, or (at your option)
 * any later version.
 *
 * You should have received a copy of the GNU General Public License
 * (for example /usr/src/linux/COPYING); if not, write to the Free
 * Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 */

21 22 23 24 25 26 27 28 29
/*
 * BITMAP UNPLUGGING:
 *
 * The sequencing for updating the bitmap reliably is a little
 * subtle (and I got it wrong the first time) so it deserves some
 * explanation.
 *
 * We group bitmap updates into batches.  Each batch has a number.
 * We may write out several batches at once, but that isn't very important.
30 31
 * conf->seq_write is the number of the last batch successfully written.
 * conf->seq_flush is the number of the last batch that was closed to
32 33 34
 *    new additions.
 * When we discover that we will need to write to any block in a stripe
 * (in add_stripe_bio) we update the in-memory bitmap and record in sh->bm_seq
35
 * the number of the batch it will be in. This is seq_flush+1.
36 37 38 39 40 41 42 43 44
 * When we are ready to do a write, if that batch hasn't been written yet,
 *   we plug the array and queue the stripe for later.
 * When an unplug happens, we increment bm_flush, thus closing the current
 *   batch.
 * When we notice that bm_flush > bm_write, we write out all pending updates
 * to the bitmap, and advance bm_write to where bm_flush was.
 * This may occasionally write a bit out twice, but is sure never to
 * miss any bits.
 */
L
Linus Torvalds 已提交
45

46
#include <linux/blkdev.h>
47
#include <linux/kthread.h>
48
#include <linux/raid/pq.h>
49
#include <linux/async_tx.h>
50
#include <linux/module.h>
51
#include <linux/async.h>
52
#include <linux/seq_file.h>
53
#include <linux/cpu.h>
54
#include <linux/slab.h>
55
#include <linux/ratelimit.h>
56
#include "md.h"
57
#include "raid5.h"
58
#include "raid0.h"
59
#include "bitmap.h"
60

L
Linus Torvalds 已提交
61 62 63 64 65 66 67 68 69
/*
 * Stripe cache
 */

#define NR_STRIPES		256
#define STRIPE_SIZE		PAGE_SIZE
#define STRIPE_SHIFT		(PAGE_SHIFT - 9)
#define STRIPE_SECTORS		(STRIPE_SIZE>>9)
#define	IO_THRESHOLD		1
70
#define BYPASS_THRESHOLD	1
71
#define NR_HASH			(PAGE_SIZE / sizeof(struct hlist_head))
L
Linus Torvalds 已提交
72 73
#define HASH_MASK		(NR_HASH - 1)

74
static inline struct hlist_head *stripe_hash(struct r5conf *conf, sector_t sect)
75 76 77 78
{
	int hash = (sect >> STRIPE_SHIFT) & HASH_MASK;
	return &conf->stripe_hashtbl[hash];
}
L
Linus Torvalds 已提交
79 80 81 82 83 84 85

/* bio's attached to a stripe+device for I/O are linked together in bi_sector
 * order without overlap.  There may be several bio's per stripe+device, and
 * a bio could span several devices.
 * When walking this list for a particular stripe+device, we must never proceed
 * beyond a bio that extends past this device, as the next bio might no longer
 * be valid.
86
 * This function is used to determine the 'next' bio in the list, given the sector
L
Linus Torvalds 已提交
87 88
 * of the current stripe+device
 */
89 90 91 92 93 94 95 96
static inline struct bio *r5_next_bio(struct bio *bio, sector_t sector)
{
	int sectors = bio->bi_size >> 9;
	if (bio->bi_sector + sectors < sector + STRIPE_SECTORS)
		return bio->bi_next;
	else
		return NULL;
}
L
Linus Torvalds 已提交
97

98
/*
99 100
 * We maintain a biased count of active stripes in the bottom 16 bits of
 * bi_phys_segments, and a count of processed stripes in the upper 16 bits
101 102 103
 */
static inline int raid5_bi_phys_segments(struct bio *bio)
{
104
	return bio->bi_phys_segments & 0xffff;
105 106 107 108
}

static inline int raid5_bi_hw_segments(struct bio *bio)
{
109
	return (bio->bi_phys_segments >> 16) & 0xffff;
110 111 112 113 114 115 116 117 118 119 120 121 122
}

static inline int raid5_dec_bi_phys_segments(struct bio *bio)
{
	--bio->bi_phys_segments;
	return raid5_bi_phys_segments(bio);
}

static inline int raid5_dec_bi_hw_segments(struct bio *bio)
{
	unsigned short val = raid5_bi_hw_segments(bio);

	--val;
123
	bio->bi_phys_segments = (val << 16) | raid5_bi_phys_segments(bio);
124 125 126 127 128
	return val;
}

static inline void raid5_set_bi_hw_segments(struct bio *bio, unsigned int cnt)
{
129
	bio->bi_phys_segments = raid5_bi_phys_segments(bio) | (cnt << 16);
130 131
}

132 133 134
/* Find first data disk in a raid6 stripe */
static inline int raid6_d0(struct stripe_head *sh)
{
135 136 137 138
	if (sh->ddf_layout)
		/* ddf always start from first device */
		return 0;
	/* md starts just after Q block */
139 140 141 142 143
	if (sh->qd_idx == sh->disks - 1)
		return 0;
	else
		return sh->qd_idx + 1;
}
144 145 146 147 148
static inline int raid6_next_disk(int disk, int raid_disks)
{
	disk++;
	return (disk < raid_disks) ? disk : 0;
}
149

150 151 152 153 154
/* When walking through the disks in a raid5, starting at raid6_d0,
 * We need to map each disk to a 'slot', where the data disks are slot
 * 0 .. raid_disks-3, the parity disk is raid_disks-2 and the Q disk
 * is raid_disks-1.  This help does that mapping.
 */
155 156
static int raid6_idx_to_slot(int idx, struct stripe_head *sh,
			     int *count, int syndrome_disks)
157
{
158
	int slot = *count;
159

160
	if (sh->ddf_layout)
161
		(*count)++;
162
	if (idx == sh->pd_idx)
163
		return syndrome_disks;
164
	if (idx == sh->qd_idx)
165
		return syndrome_disks + 1;
166
	if (!sh->ddf_layout)
167
		(*count)++;
168 169 170
	return slot;
}

171 172 173 174 175 176 177 178
static void return_io(struct bio *return_bi)
{
	struct bio *bi = return_bi;
	while (bi) {

		return_bi = bi->bi_next;
		bi->bi_next = NULL;
		bi->bi_size = 0;
179
		bio_endio(bi, 0);
180 181 182 183
		bi = return_bi;
	}
}

184
static void print_raid5_conf (struct r5conf *conf);
L
Linus Torvalds 已提交
185

186 187 188 189 190 191 192
static int stripe_operations_active(struct stripe_head *sh)
{
	return sh->check_state || sh->reconstruct_state ||
	       test_bit(STRIPE_BIOFILL_RUN, &sh->state) ||
	       test_bit(STRIPE_COMPUTE_RUN, &sh->state);
}

193
static void __release_stripe(struct r5conf *conf, struct stripe_head *sh)
L
Linus Torvalds 已提交
194 195
{
	if (atomic_dec_and_test(&sh->count)) {
196 197
		BUG_ON(!list_empty(&sh->lru));
		BUG_ON(atomic_read(&conf->active_stripes)==0);
L
Linus Torvalds 已提交
198
		if (test_bit(STRIPE_HANDLE, &sh->state)) {
N
NeilBrown 已提交
199
			if (test_bit(STRIPE_DELAYED, &sh->state))
L
Linus Torvalds 已提交
200
				list_add_tail(&sh->lru, &conf->delayed_list);
N
NeilBrown 已提交
201 202
			else if (test_bit(STRIPE_BIT_DELAY, &sh->state) &&
				   sh->bm_seq - conf->seq_write > 0)
203
				list_add_tail(&sh->lru, &conf->bitmap_list);
N
NeilBrown 已提交
204
			else {
205
				clear_bit(STRIPE_BIT_DELAY, &sh->state);
L
Linus Torvalds 已提交
206
				list_add_tail(&sh->lru, &conf->handle_list);
207
			}
L
Linus Torvalds 已提交
208 209
			md_wakeup_thread(conf->mddev->thread);
		} else {
210
			BUG_ON(stripe_operations_active(sh));
L
Linus Torvalds 已提交
211 212 213 214 215 216
			if (test_and_clear_bit(STRIPE_PREREAD_ACTIVE, &sh->state)) {
				atomic_dec(&conf->preread_active_stripes);
				if (atomic_read(&conf->preread_active_stripes) < IO_THRESHOLD)
					md_wakeup_thread(conf->mddev->thread);
			}
			atomic_dec(&conf->active_stripes);
217 218
			if (!test_bit(STRIPE_EXPANDING, &sh->state)) {
				list_add_tail(&sh->lru, &conf->inactive_list);
L
Linus Torvalds 已提交
219
				wake_up(&conf->wait_for_stripe);
220 221
				if (conf->retry_read_aligned)
					md_wakeup_thread(conf->mddev->thread);
222
			}
L
Linus Torvalds 已提交
223 224 225
		}
	}
}
226

L
Linus Torvalds 已提交
227 228
static void release_stripe(struct stripe_head *sh)
{
229
	struct r5conf *conf = sh->raid_conf;
L
Linus Torvalds 已提交
230
	unsigned long flags;
231

L
Linus Torvalds 已提交
232 233 234 235 236
	spin_lock_irqsave(&conf->device_lock, flags);
	__release_stripe(conf, sh);
	spin_unlock_irqrestore(&conf->device_lock, flags);
}

237
static inline void remove_hash(struct stripe_head *sh)
L
Linus Torvalds 已提交
238
{
239 240
	pr_debug("remove_hash(), stripe %llu\n",
		(unsigned long long)sh->sector);
L
Linus Torvalds 已提交
241

242
	hlist_del_init(&sh->hash);
L
Linus Torvalds 已提交
243 244
}

245
static inline void insert_hash(struct r5conf *conf, struct stripe_head *sh)
L
Linus Torvalds 已提交
246
{
247
	struct hlist_head *hp = stripe_hash(conf, sh->sector);
L
Linus Torvalds 已提交
248

249 250
	pr_debug("insert_hash(), stripe %llu\n",
		(unsigned long long)sh->sector);
L
Linus Torvalds 已提交
251

252
	hlist_add_head(&sh->hash, hp);
L
Linus Torvalds 已提交
253 254 255 256
}


/* find an idle stripe, make sure it is unhashed, and return it. */
257
static struct stripe_head *get_free_stripe(struct r5conf *conf)
L
Linus Torvalds 已提交
258 259 260 261 262 263 264 265 266 267 268 269 270 271 272
{
	struct stripe_head *sh = NULL;
	struct list_head *first;

	if (list_empty(&conf->inactive_list))
		goto out;
	first = conf->inactive_list.next;
	sh = list_entry(first, struct stripe_head, lru);
	list_del_init(first);
	remove_hash(sh);
	atomic_inc(&conf->active_stripes);
out:
	return sh;
}

273
static void shrink_buffers(struct stripe_head *sh)
L
Linus Torvalds 已提交
274 275 276
{
	struct page *p;
	int i;
277
	int num = sh->raid_conf->pool_size;
L
Linus Torvalds 已提交
278

279
	for (i = 0; i < num ; i++) {
L
Linus Torvalds 已提交
280 281 282 283
		p = sh->dev[i].page;
		if (!p)
			continue;
		sh->dev[i].page = NULL;
284
		put_page(p);
L
Linus Torvalds 已提交
285 286 287
	}
}

288
static int grow_buffers(struct stripe_head *sh)
L
Linus Torvalds 已提交
289 290
{
	int i;
291
	int num = sh->raid_conf->pool_size;
L
Linus Torvalds 已提交
292

293
	for (i = 0; i < num; i++) {
L
Linus Torvalds 已提交
294 295 296 297 298 299 300 301 302 303
		struct page *page;

		if (!(page = alloc_page(GFP_KERNEL))) {
			return 1;
		}
		sh->dev[i].page = page;
	}
	return 0;
}

304
static void raid5_build_block(struct stripe_head *sh, int i, int previous);
305
static void stripe_set_idx(sector_t stripe, struct r5conf *conf, int previous,
306
			    struct stripe_head *sh);
L
Linus Torvalds 已提交
307

308
static void init_stripe(struct stripe_head *sh, sector_t sector, int previous)
L
Linus Torvalds 已提交
309
{
310
	struct r5conf *conf = sh->raid_conf;
311
	int i;
L
Linus Torvalds 已提交
312

313 314
	BUG_ON(atomic_read(&sh->count) != 0);
	BUG_ON(test_bit(STRIPE_HANDLE, &sh->state));
315
	BUG_ON(stripe_operations_active(sh));
316

317
	pr_debug("init_stripe called, stripe %llu\n",
L
Linus Torvalds 已提交
318 319 320
		(unsigned long long)sh->sector);

	remove_hash(sh);
321

322
	sh->generation = conf->generation - previous;
323
	sh->disks = previous ? conf->previous_raid_disks : conf->raid_disks;
L
Linus Torvalds 已提交
324
	sh->sector = sector;
325
	stripe_set_idx(sector, conf, previous, sh);
L
Linus Torvalds 已提交
326 327
	sh->state = 0;

328 329

	for (i = sh->disks; i--; ) {
L
Linus Torvalds 已提交
330 331
		struct r5dev *dev = &sh->dev[i];

332
		if (dev->toread || dev->read || dev->towrite || dev->written ||
L
Linus Torvalds 已提交
333
		    test_bit(R5_LOCKED, &dev->flags)) {
334
			printk(KERN_ERR "sector=%llx i=%d %p %p %p %p %d\n",
L
Linus Torvalds 已提交
335
			       (unsigned long long)sh->sector, i, dev->toread,
336
			       dev->read, dev->towrite, dev->written,
L
Linus Torvalds 已提交
337
			       test_bit(R5_LOCKED, &dev->flags));
338
			WARN_ON(1);
L
Linus Torvalds 已提交
339 340
		}
		dev->flags = 0;
341
		raid5_build_block(sh, i, previous);
L
Linus Torvalds 已提交
342 343 344 345
	}
	insert_hash(conf, sh);
}

346
static struct stripe_head *__find_stripe(struct r5conf *conf, sector_t sector,
347
					 short generation)
L
Linus Torvalds 已提交
348 349
{
	struct stripe_head *sh;
350
	struct hlist_node *hn;
L
Linus Torvalds 已提交
351

352
	pr_debug("__find_stripe, sector %llu\n", (unsigned long long)sector);
353
	hlist_for_each_entry(sh, hn, stripe_hash(conf, sector), hash)
354
		if (sh->sector == sector && sh->generation == generation)
L
Linus Torvalds 已提交
355
			return sh;
356
	pr_debug("__stripe %llu not in cache\n", (unsigned long long)sector);
L
Linus Torvalds 已提交
357 358 359
	return NULL;
}

360 361 362 363 364 365 366 367 368 369 370 371 372
/*
 * Need to check if array has failed when deciding whether to:
 *  - start an array
 *  - remove non-faulty devices
 *  - add a spare
 *  - allow a reshape
 * This determination is simple when no reshape is happening.
 * However if there is a reshape, we need to carefully check
 * both the before and after sections.
 * This is because some failed devices may only affect one
 * of the two sections, and some non-in_sync devices may
 * be insync in the section most affected by failed devices.
 */
373
static int calc_degraded(struct r5conf *conf)
374
{
375
	int degraded, degraded2;
376 377 378 379 380
	int i;

	rcu_read_lock();
	degraded = 0;
	for (i = 0; i < conf->previous_raid_disks; i++) {
381
		struct md_rdev *rdev = rcu_dereference(conf->disks[i].rdev);
382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399
		if (!rdev || test_bit(Faulty, &rdev->flags))
			degraded++;
		else if (test_bit(In_sync, &rdev->flags))
			;
		else
			/* not in-sync or faulty.
			 * If the reshape increases the number of devices,
			 * this is being recovered by the reshape, so
			 * this 'previous' section is not in_sync.
			 * If the number of devices is being reduced however,
			 * the device can only be part of the array if
			 * we are reverting a reshape, so this section will
			 * be in-sync.
			 */
			if (conf->raid_disks >= conf->previous_raid_disks)
				degraded++;
	}
	rcu_read_unlock();
400 401
	if (conf->raid_disks == conf->previous_raid_disks)
		return degraded;
402
	rcu_read_lock();
403
	degraded2 = 0;
404
	for (i = 0; i < conf->raid_disks; i++) {
405
		struct md_rdev *rdev = rcu_dereference(conf->disks[i].rdev);
406
		if (!rdev || test_bit(Faulty, &rdev->flags))
407
			degraded2++;
408 409 410 411 412 413 414 415 416
		else if (test_bit(In_sync, &rdev->flags))
			;
		else
			/* not in-sync or faulty.
			 * If reshape increases the number of devices, this
			 * section has already been recovered, else it
			 * almost certainly hasn't.
			 */
			if (conf->raid_disks <= conf->previous_raid_disks)
417
				degraded2++;
418 419
	}
	rcu_read_unlock();
420 421 422 423 424 425 426 427 428 429 430 431 432
	if (degraded2 > degraded)
		return degraded2;
	return degraded;
}

static int has_failed(struct r5conf *conf)
{
	int degraded;

	if (conf->mddev->reshape_position == MaxSector)
		return conf->mddev->degraded > conf->max_degraded;

	degraded = calc_degraded(conf);
433 434 435 436 437
	if (degraded > conf->max_degraded)
		return 1;
	return 0;
}

438
static struct stripe_head *
439
get_active_stripe(struct r5conf *conf, sector_t sector,
440
		  int previous, int noblock, int noquiesce)
L
Linus Torvalds 已提交
441 442 443
{
	struct stripe_head *sh;

444
	pr_debug("get_stripe, sector %llu\n", (unsigned long long)sector);
L
Linus Torvalds 已提交
445 446 447 448

	spin_lock_irq(&conf->device_lock);

	do {
449
		wait_event_lock_irq(conf->wait_for_stripe,
450
				    conf->quiesce == 0 || noquiesce,
451
				    conf->device_lock, /* nothing */);
452
		sh = __find_stripe(conf, sector, conf->generation - previous);
L
Linus Torvalds 已提交
453 454 455 456 457 458 459 460 461
		if (!sh) {
			if (!conf->inactive_blocked)
				sh = get_free_stripe(conf);
			if (noblock && sh == NULL)
				break;
			if (!sh) {
				conf->inactive_blocked = 1;
				wait_event_lock_irq(conf->wait_for_stripe,
						    !list_empty(&conf->inactive_list) &&
462 463
						    (atomic_read(&conf->active_stripes)
						     < (conf->max_nr_stripes *3/4)
L
Linus Torvalds 已提交
464 465
						     || !conf->inactive_blocked),
						    conf->device_lock,
466
						    );
L
Linus Torvalds 已提交
467 468
				conf->inactive_blocked = 0;
			} else
469
				init_stripe(sh, sector, previous);
L
Linus Torvalds 已提交
470 471
		} else {
			if (atomic_read(&sh->count)) {
472 473
				BUG_ON(!list_empty(&sh->lru)
				    && !test_bit(STRIPE_EXPANDING, &sh->state));
L
Linus Torvalds 已提交
474 475 476
			} else {
				if (!test_bit(STRIPE_HANDLE, &sh->state))
					atomic_inc(&conf->active_stripes);
477 478
				if (list_empty(&sh->lru) &&
				    !test_bit(STRIPE_EXPANDING, &sh->state))
479 480
					BUG();
				list_del_init(&sh->lru);
L
Linus Torvalds 已提交
481 482 483 484 485 486 487 488 489 490 491
			}
		}
	} while (sh == NULL);

	if (sh)
		atomic_inc(&sh->count);

	spin_unlock_irq(&conf->device_lock);
	return sh;
}

492 493 494 495
static void
raid5_end_read_request(struct bio *bi, int error);
static void
raid5_end_write_request(struct bio *bi, int error);
496

497
static void ops_run_io(struct stripe_head *sh, struct stripe_head_state *s)
498
{
499
	struct r5conf *conf = sh->raid_conf;
500 501 502 503 504 505 506
	int i, disks = sh->disks;

	might_sleep();

	for (i = disks; i--; ) {
		int rw;
		struct bio *bi;
507
		struct md_rdev *rdev;
T
Tejun Heo 已提交
508 509 510 511 512 513
		if (test_and_clear_bit(R5_Wantwrite, &sh->dev[i].flags)) {
			if (test_and_clear_bit(R5_WantFUA, &sh->dev[i].flags))
				rw = WRITE_FUA;
			else
				rw = WRITE;
		} else if (test_and_clear_bit(R5_Wantread, &sh->dev[i].flags))
514 515 516 517 518 519 520
			rw = READ;
		else
			continue;

		bi = &sh->dev[i].req;

		bi->bi_rw = rw;
521
		if (rw & WRITE)
522 523 524 525 526
			bi->bi_end_io = raid5_end_write_request;
		else
			bi->bi_end_io = raid5_end_read_request;

		rcu_read_lock();
527 528 529 530 531
		if (rw == READ &&
		    test_bit(R5_ReadRepl, &sh->dev[i].flags))
			rdev = rcu_dereference(conf->disks[i].replacement);
		else
			rdev = rcu_dereference(conf->disks[i].rdev);
532 533 534 535 536 537
		if (rdev && test_bit(Faulty, &rdev->flags))
			rdev = NULL;
		if (rdev)
			atomic_inc(&rdev->nr_pending);
		rcu_read_unlock();

538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567
		/* We have already checked bad blocks for reads.  Now
		 * need to check for writes.
		 */
		while ((rw & WRITE) && rdev &&
		       test_bit(WriteErrorSeen, &rdev->flags)) {
			sector_t first_bad;
			int bad_sectors;
			int bad = is_badblock(rdev, sh->sector, STRIPE_SECTORS,
					      &first_bad, &bad_sectors);
			if (!bad)
				break;

			if (bad < 0) {
				set_bit(BlockedBadBlocks, &rdev->flags);
				if (!conf->mddev->external &&
				    conf->mddev->flags) {
					/* It is very unlikely, but we might
					 * still need to write out the
					 * bad block log - better give it
					 * a chance*/
					md_check_recovery(conf->mddev);
				}
				md_wait_for_blocked_rdev(rdev, conf->mddev);
			} else {
				/* Acknowledged bad block - skip the write */
				rdev_dec_pending(rdev, conf->mddev);
				rdev = NULL;
			}
		}

568
		if (rdev) {
569
			if (s->syncing || s->expanding || s->expanded)
570 571
				md_sync_acct(rdev->bdev, STRIPE_SECTORS);

D
Dan Williams 已提交
572 573
			set_bit(STRIPE_IO_STARTED, &sh->state);

574 575
			bi->bi_bdev = rdev->bdev;
			pr_debug("%s: for %llu schedule op %ld on disc %d\n",
576
				__func__, (unsigned long long)sh->sector,
577 578 579 580 581 582 583 584 585 586 587
				bi->bi_rw, i);
			atomic_inc(&sh->count);
			bi->bi_sector = sh->sector + rdev->data_offset;
			bi->bi_flags = 1 << BIO_UPTODATE;
			bi->bi_idx = 0;
			bi->bi_io_vec[0].bv_len = STRIPE_SIZE;
			bi->bi_io_vec[0].bv_offset = 0;
			bi->bi_size = STRIPE_SIZE;
			bi->bi_next = NULL;
			generic_make_request(bi);
		} else {
588
			if (rw & WRITE)
589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605
				set_bit(STRIPE_DEGRADED, &sh->state);
			pr_debug("skip op %ld on disc %d for sector %llu\n",
				bi->bi_rw, i, (unsigned long long)sh->sector);
			clear_bit(R5_LOCKED, &sh->dev[i].flags);
			set_bit(STRIPE_HANDLE, &sh->state);
		}
	}
}

static struct dma_async_tx_descriptor *
async_copy_data(int frombio, struct bio *bio, struct page *page,
	sector_t sector, struct dma_async_tx_descriptor *tx)
{
	struct bio_vec *bvl;
	struct page *bio_page;
	int i;
	int page_offset;
606
	struct async_submit_ctl submit;
D
Dan Williams 已提交
607
	enum async_tx_flags flags = 0;
608 609 610 611 612

	if (bio->bi_sector >= sector)
		page_offset = (signed)(bio->bi_sector - sector) * 512;
	else
		page_offset = (signed)(sector - bio->bi_sector) * -512;
613

D
Dan Williams 已提交
614 615 616 617
	if (frombio)
		flags |= ASYNC_TX_FENCE;
	init_async_submit(&submit, flags, tx, NULL, NULL, NULL);

618
	bio_for_each_segment(bvl, bio, i) {
619
		int len = bvl->bv_len;
620 621 622 623 624 625 626 627 628 629 630 631 632 633 634
		int clen;
		int b_offset = 0;

		if (page_offset < 0) {
			b_offset = -page_offset;
			page_offset += b_offset;
			len -= b_offset;
		}

		if (len > 0 && page_offset + len > STRIPE_SIZE)
			clen = STRIPE_SIZE - page_offset;
		else
			clen = len;

		if (clen > 0) {
635 636
			b_offset += bvl->bv_offset;
			bio_page = bvl->bv_page;
637 638
			if (frombio)
				tx = async_memcpy(page, bio_page, page_offset,
639
						  b_offset, clen, &submit);
640 641
			else
				tx = async_memcpy(bio_page, page, b_offset,
642
						  page_offset, clen, &submit);
643
		}
644 645 646
		/* chain the operations */
		submit.depend_tx = tx;

647 648 649 650 651 652 653 654 655 656 657 658
		if (clen < len) /* hit end of page */
			break;
		page_offset +=  len;
	}

	return tx;
}

static void ops_complete_biofill(void *stripe_head_ref)
{
	struct stripe_head *sh = stripe_head_ref;
	struct bio *return_bi = NULL;
659
	struct r5conf *conf = sh->raid_conf;
660
	int i;
661

662
	pr_debug("%s: stripe %llu\n", __func__,
663 664 665
		(unsigned long long)sh->sector);

	/* clear completed biofills */
666
	spin_lock_irq(&conf->device_lock);
667 668 669 670
	for (i = sh->disks; i--; ) {
		struct r5dev *dev = &sh->dev[i];

		/* acknowledge completion of a biofill operation */
671 672
		/* and check if we need to reply to a read request,
		 * new R5_Wantfill requests are held off until
673
		 * !STRIPE_BIOFILL_RUN
674 675
		 */
		if (test_and_clear_bit(R5_Wantfill, &dev->flags)) {
676 677 678 679 680 681 682 683
			struct bio *rbi, *rbi2;

			BUG_ON(!dev->read);
			rbi = dev->read;
			dev->read = NULL;
			while (rbi && rbi->bi_sector <
				dev->sector + STRIPE_SECTORS) {
				rbi2 = r5_next_bio(rbi, dev->sector);
684
				if (!raid5_dec_bi_phys_segments(rbi)) {
685 686 687 688 689 690 691
					rbi->bi_next = return_bi;
					return_bi = rbi;
				}
				rbi = rbi2;
			}
		}
	}
692 693
	spin_unlock_irq(&conf->device_lock);
	clear_bit(STRIPE_BIOFILL_RUN, &sh->state);
694 695 696

	return_io(return_bi);

697
	set_bit(STRIPE_HANDLE, &sh->state);
698 699 700 701 702 703
	release_stripe(sh);
}

static void ops_run_biofill(struct stripe_head *sh)
{
	struct dma_async_tx_descriptor *tx = NULL;
704
	struct r5conf *conf = sh->raid_conf;
705
	struct async_submit_ctl submit;
706 707
	int i;

708
	pr_debug("%s: stripe %llu\n", __func__,
709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728
		(unsigned long long)sh->sector);

	for (i = sh->disks; i--; ) {
		struct r5dev *dev = &sh->dev[i];
		if (test_bit(R5_Wantfill, &dev->flags)) {
			struct bio *rbi;
			spin_lock_irq(&conf->device_lock);
			dev->read = rbi = dev->toread;
			dev->toread = NULL;
			spin_unlock_irq(&conf->device_lock);
			while (rbi && rbi->bi_sector <
				dev->sector + STRIPE_SECTORS) {
				tx = async_copy_data(0, rbi, dev->page,
					dev->sector, tx);
				rbi = r5_next_bio(rbi, dev->sector);
			}
		}
	}

	atomic_inc(&sh->count);
729 730
	init_async_submit(&submit, ASYNC_TX_ACK, tx, ops_complete_biofill, sh, NULL);
	async_trigger_callback(&submit);
731 732
}

733
static void mark_target_uptodate(struct stripe_head *sh, int target)
734
{
735
	struct r5dev *tgt;
736

737 738
	if (target < 0)
		return;
739

740
	tgt = &sh->dev[target];
741 742 743
	set_bit(R5_UPTODATE, &tgt->flags);
	BUG_ON(!test_bit(R5_Wantcompute, &tgt->flags));
	clear_bit(R5_Wantcompute, &tgt->flags);
744 745
}

746
static void ops_complete_compute(void *stripe_head_ref)
747 748 749
{
	struct stripe_head *sh = stripe_head_ref;

750
	pr_debug("%s: stripe %llu\n", __func__,
751 752
		(unsigned long long)sh->sector);

753
	/* mark the computed target(s) as uptodate */
754
	mark_target_uptodate(sh, sh->ops.target);
755
	mark_target_uptodate(sh, sh->ops.target2);
756

757 758 759
	clear_bit(STRIPE_COMPUTE_RUN, &sh->state);
	if (sh->check_state == check_state_compute_run)
		sh->check_state = check_state_compute_result;
760 761 762 763
	set_bit(STRIPE_HANDLE, &sh->state);
	release_stripe(sh);
}

764 765 766 767 768 769 770 771 772
/* return a pointer to the address conversion region of the scribble buffer */
static addr_conv_t *to_addr_conv(struct stripe_head *sh,
				 struct raid5_percpu *percpu)
{
	return percpu->scribble + sizeof(struct page *) * (sh->disks + 2);
}

static struct dma_async_tx_descriptor *
ops_run_compute5(struct stripe_head *sh, struct raid5_percpu *percpu)
773 774
{
	int disks = sh->disks;
775
	struct page **xor_srcs = percpu->scribble;
776 777 778 779 780
	int target = sh->ops.target;
	struct r5dev *tgt = &sh->dev[target];
	struct page *xor_dest = tgt->page;
	int count = 0;
	struct dma_async_tx_descriptor *tx;
781
	struct async_submit_ctl submit;
782 783 784
	int i;

	pr_debug("%s: stripe %llu block: %d\n",
785
		__func__, (unsigned long long)sh->sector, target);
786 787 788 789 790 791 792 793
	BUG_ON(!test_bit(R5_Wantcompute, &tgt->flags));

	for (i = disks; i--; )
		if (i != target)
			xor_srcs[count++] = sh->dev[i].page;

	atomic_inc(&sh->count);

D
Dan Williams 已提交
794
	init_async_submit(&submit, ASYNC_TX_FENCE|ASYNC_TX_XOR_ZERO_DST, NULL,
795
			  ops_complete_compute, sh, to_addr_conv(sh, percpu));
796
	if (unlikely(count == 1))
797
		tx = async_memcpy(xor_dest, xor_srcs[0], 0, 0, STRIPE_SIZE, &submit);
798
	else
799
		tx = async_xor(xor_dest, xor_srcs, 0, count, STRIPE_SIZE, &submit);
800 801 802 803

	return tx;
}

804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821
/* set_syndrome_sources - populate source buffers for gen_syndrome
 * @srcs - (struct page *) array of size sh->disks
 * @sh - stripe_head to parse
 *
 * Populates srcs in proper layout order for the stripe and returns the
 * 'count' of sources to be used in a call to async_gen_syndrome.  The P
 * destination buffer is recorded in srcs[count] and the Q destination
 * is recorded in srcs[count+1]].
 */
static int set_syndrome_sources(struct page **srcs, struct stripe_head *sh)
{
	int disks = sh->disks;
	int syndrome_disks = sh->ddf_layout ? disks : (disks - 2);
	int d0_idx = raid6_d0(sh);
	int count;
	int i;

	for (i = 0; i < disks; i++)
822
		srcs[i] = NULL;
823 824 825 826 827 828 829 830 831 832

	count = 0;
	i = d0_idx;
	do {
		int slot = raid6_idx_to_slot(i, sh, &count, syndrome_disks);

		srcs[slot] = sh->dev[i].page;
		i = raid6_next_disk(i, disks);
	} while (i != d0_idx);

833
	return syndrome_disks;
834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853
}

static struct dma_async_tx_descriptor *
ops_run_compute6_1(struct stripe_head *sh, struct raid5_percpu *percpu)
{
	int disks = sh->disks;
	struct page **blocks = percpu->scribble;
	int target;
	int qd_idx = sh->qd_idx;
	struct dma_async_tx_descriptor *tx;
	struct async_submit_ctl submit;
	struct r5dev *tgt;
	struct page *dest;
	int i;
	int count;

	if (sh->ops.target < 0)
		target = sh->ops.target2;
	else if (sh->ops.target2 < 0)
		target = sh->ops.target;
854
	else
855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870
		/* we should only have one valid target */
		BUG();
	BUG_ON(target < 0);
	pr_debug("%s: stripe %llu block: %d\n",
		__func__, (unsigned long long)sh->sector, target);

	tgt = &sh->dev[target];
	BUG_ON(!test_bit(R5_Wantcompute, &tgt->flags));
	dest = tgt->page;

	atomic_inc(&sh->count);

	if (target == qd_idx) {
		count = set_syndrome_sources(blocks, sh);
		blocks[count] = NULL; /* regenerating p is not necessary */
		BUG_ON(blocks[count+1] != dest); /* q should already be set */
D
Dan Williams 已提交
871 872
		init_async_submit(&submit, ASYNC_TX_FENCE, NULL,
				  ops_complete_compute, sh,
873 874 875 876 877 878 879 880 881 882 883
				  to_addr_conv(sh, percpu));
		tx = async_gen_syndrome(blocks, 0, count+2, STRIPE_SIZE, &submit);
	} else {
		/* Compute any data- or p-drive using XOR */
		count = 0;
		for (i = disks; i-- ; ) {
			if (i == target || i == qd_idx)
				continue;
			blocks[count++] = sh->dev[i].page;
		}

D
Dan Williams 已提交
884 885
		init_async_submit(&submit, ASYNC_TX_FENCE|ASYNC_TX_XOR_ZERO_DST,
				  NULL, ops_complete_compute, sh,
886 887 888
				  to_addr_conv(sh, percpu));
		tx = async_xor(dest, blocks, 0, count, STRIPE_SIZE, &submit);
	}
889 890 891 892

	return tx;
}

893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913
static struct dma_async_tx_descriptor *
ops_run_compute6_2(struct stripe_head *sh, struct raid5_percpu *percpu)
{
	int i, count, disks = sh->disks;
	int syndrome_disks = sh->ddf_layout ? disks : disks-2;
	int d0_idx = raid6_d0(sh);
	int faila = -1, failb = -1;
	int target = sh->ops.target;
	int target2 = sh->ops.target2;
	struct r5dev *tgt = &sh->dev[target];
	struct r5dev *tgt2 = &sh->dev[target2];
	struct dma_async_tx_descriptor *tx;
	struct page **blocks = percpu->scribble;
	struct async_submit_ctl submit;

	pr_debug("%s: stripe %llu block1: %d block2: %d\n",
		 __func__, (unsigned long long)sh->sector, target, target2);
	BUG_ON(target < 0 || target2 < 0);
	BUG_ON(!test_bit(R5_Wantcompute, &tgt->flags));
	BUG_ON(!test_bit(R5_Wantcompute, &tgt2->flags));

914
	/* we need to open-code set_syndrome_sources to handle the
915 916 917
	 * slot number conversion for 'faila' and 'failb'
	 */
	for (i = 0; i < disks ; i++)
918
		blocks[i] = NULL;
919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944
	count = 0;
	i = d0_idx;
	do {
		int slot = raid6_idx_to_slot(i, sh, &count, syndrome_disks);

		blocks[slot] = sh->dev[i].page;

		if (i == target)
			faila = slot;
		if (i == target2)
			failb = slot;
		i = raid6_next_disk(i, disks);
	} while (i != d0_idx);

	BUG_ON(faila == failb);
	if (failb < faila)
		swap(faila, failb);
	pr_debug("%s: stripe: %llu faila: %d failb: %d\n",
		 __func__, (unsigned long long)sh->sector, faila, failb);

	atomic_inc(&sh->count);

	if (failb == syndrome_disks+1) {
		/* Q disk is one of the missing disks */
		if (faila == syndrome_disks) {
			/* Missing P+Q, just recompute */
D
Dan Williams 已提交
945 946 947
			init_async_submit(&submit, ASYNC_TX_FENCE, NULL,
					  ops_complete_compute, sh,
					  to_addr_conv(sh, percpu));
948
			return async_gen_syndrome(blocks, 0, syndrome_disks+2,
949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967
						  STRIPE_SIZE, &submit);
		} else {
			struct page *dest;
			int data_target;
			int qd_idx = sh->qd_idx;

			/* Missing D+Q: recompute D from P, then recompute Q */
			if (target == qd_idx)
				data_target = target2;
			else
				data_target = target;

			count = 0;
			for (i = disks; i-- ; ) {
				if (i == data_target || i == qd_idx)
					continue;
				blocks[count++] = sh->dev[i].page;
			}
			dest = sh->dev[data_target].page;
D
Dan Williams 已提交
968 969 970 971
			init_async_submit(&submit,
					  ASYNC_TX_FENCE|ASYNC_TX_XOR_ZERO_DST,
					  NULL, NULL, NULL,
					  to_addr_conv(sh, percpu));
972 973 974 975
			tx = async_xor(dest, blocks, 0, count, STRIPE_SIZE,
				       &submit);

			count = set_syndrome_sources(blocks, sh);
D
Dan Williams 已提交
976 977 978
			init_async_submit(&submit, ASYNC_TX_FENCE, tx,
					  ops_complete_compute, sh,
					  to_addr_conv(sh, percpu));
979 980 981 982
			return async_gen_syndrome(blocks, 0, count+2,
						  STRIPE_SIZE, &submit);
		}
	} else {
983 984 985 986 987 988 989 990 991 992 993 994 995 996
		init_async_submit(&submit, ASYNC_TX_FENCE, NULL,
				  ops_complete_compute, sh,
				  to_addr_conv(sh, percpu));
		if (failb == syndrome_disks) {
			/* We're missing D+P. */
			return async_raid6_datap_recov(syndrome_disks+2,
						       STRIPE_SIZE, faila,
						       blocks, &submit);
		} else {
			/* We're missing D+D. */
			return async_raid6_2data_recov(syndrome_disks+2,
						       STRIPE_SIZE, faila, failb,
						       blocks, &submit);
		}
997 998 999 1000
	}
}


1001 1002 1003 1004
static void ops_complete_prexor(void *stripe_head_ref)
{
	struct stripe_head *sh = stripe_head_ref;

1005
	pr_debug("%s: stripe %llu\n", __func__,
1006 1007 1008 1009
		(unsigned long long)sh->sector);
}

static struct dma_async_tx_descriptor *
1010 1011
ops_run_prexor(struct stripe_head *sh, struct raid5_percpu *percpu,
	       struct dma_async_tx_descriptor *tx)
1012 1013
{
	int disks = sh->disks;
1014
	struct page **xor_srcs = percpu->scribble;
1015
	int count = 0, pd_idx = sh->pd_idx, i;
1016
	struct async_submit_ctl submit;
1017 1018 1019 1020

	/* existing parity data subtracted */
	struct page *xor_dest = xor_srcs[count++] = sh->dev[pd_idx].page;

1021
	pr_debug("%s: stripe %llu\n", __func__,
1022 1023 1024 1025 1026
		(unsigned long long)sh->sector);

	for (i = disks; i--; ) {
		struct r5dev *dev = &sh->dev[i];
		/* Only process blocks that are known to be uptodate */
1027
		if (test_bit(R5_Wantdrain, &dev->flags))
1028 1029 1030
			xor_srcs[count++] = dev->page;
	}

D
Dan Williams 已提交
1031
	init_async_submit(&submit, ASYNC_TX_FENCE|ASYNC_TX_XOR_DROP_DST, tx,
1032
			  ops_complete_prexor, sh, to_addr_conv(sh, percpu));
1033
	tx = async_xor(xor_dest, xor_srcs, 0, count, STRIPE_SIZE, &submit);
1034 1035 1036 1037 1038

	return tx;
}

static struct dma_async_tx_descriptor *
1039
ops_run_biodrain(struct stripe_head *sh, struct dma_async_tx_descriptor *tx)
1040 1041
{
	int disks = sh->disks;
1042
	int i;
1043

1044
	pr_debug("%s: stripe %llu\n", __func__,
1045 1046 1047 1048 1049 1050
		(unsigned long long)sh->sector);

	for (i = disks; i--; ) {
		struct r5dev *dev = &sh->dev[i];
		struct bio *chosen;

1051
		if (test_and_clear_bit(R5_Wantdrain, &dev->flags)) {
1052 1053
			struct bio *wbi;

1054
			spin_lock_irq(&sh->raid_conf->device_lock);
1055 1056 1057 1058
			chosen = dev->towrite;
			dev->towrite = NULL;
			BUG_ON(dev->written);
			wbi = dev->written = chosen;
1059
			spin_unlock_irq(&sh->raid_conf->device_lock);
1060 1061 1062

			while (wbi && wbi->bi_sector <
				dev->sector + STRIPE_SECTORS) {
T
Tejun Heo 已提交
1063 1064
				if (wbi->bi_rw & REQ_FUA)
					set_bit(R5_WantFUA, &dev->flags);
1065 1066 1067 1068 1069 1070 1071 1072 1073 1074
				tx = async_copy_data(1, wbi, dev->page,
					dev->sector, tx);
				wbi = r5_next_bio(wbi, dev->sector);
			}
		}
	}

	return tx;
}

1075
static void ops_complete_reconstruct(void *stripe_head_ref)
1076 1077
{
	struct stripe_head *sh = stripe_head_ref;
1078 1079 1080 1081
	int disks = sh->disks;
	int pd_idx = sh->pd_idx;
	int qd_idx = sh->qd_idx;
	int i;
T
Tejun Heo 已提交
1082
	bool fua = false;
1083

1084
	pr_debug("%s: stripe %llu\n", __func__,
1085 1086
		(unsigned long long)sh->sector);

T
Tejun Heo 已提交
1087 1088 1089
	for (i = disks; i--; )
		fua |= test_bit(R5_WantFUA, &sh->dev[i].flags);

1090 1091
	for (i = disks; i--; ) {
		struct r5dev *dev = &sh->dev[i];
1092

T
Tejun Heo 已提交
1093
		if (dev->written || i == pd_idx || i == qd_idx) {
1094
			set_bit(R5_UPTODATE, &dev->flags);
T
Tejun Heo 已提交
1095 1096 1097
			if (fua)
				set_bit(R5_WantFUA, &dev->flags);
		}
1098 1099
	}

1100 1101 1102 1103 1104 1105 1106 1107
	if (sh->reconstruct_state == reconstruct_state_drain_run)
		sh->reconstruct_state = reconstruct_state_drain_result;
	else if (sh->reconstruct_state == reconstruct_state_prexor_drain_run)
		sh->reconstruct_state = reconstruct_state_prexor_drain_result;
	else {
		BUG_ON(sh->reconstruct_state != reconstruct_state_run);
		sh->reconstruct_state = reconstruct_state_result;
	}
1108 1109 1110 1111 1112 1113

	set_bit(STRIPE_HANDLE, &sh->state);
	release_stripe(sh);
}

static void
1114 1115
ops_run_reconstruct5(struct stripe_head *sh, struct raid5_percpu *percpu,
		     struct dma_async_tx_descriptor *tx)
1116 1117
{
	int disks = sh->disks;
1118
	struct page **xor_srcs = percpu->scribble;
1119
	struct async_submit_ctl submit;
1120 1121
	int count = 0, pd_idx = sh->pd_idx, i;
	struct page *xor_dest;
1122
	int prexor = 0;
1123 1124
	unsigned long flags;

1125
	pr_debug("%s: stripe %llu\n", __func__,
1126 1127 1128 1129 1130
		(unsigned long long)sh->sector);

	/* check if prexor is active which means only process blocks
	 * that are part of a read-modify-write (written)
	 */
1131 1132
	if (sh->reconstruct_state == reconstruct_state_prexor_drain_run) {
		prexor = 1;
1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152
		xor_dest = xor_srcs[count++] = sh->dev[pd_idx].page;
		for (i = disks; i--; ) {
			struct r5dev *dev = &sh->dev[i];
			if (dev->written)
				xor_srcs[count++] = dev->page;
		}
	} else {
		xor_dest = sh->dev[pd_idx].page;
		for (i = disks; i--; ) {
			struct r5dev *dev = &sh->dev[i];
			if (i != pd_idx)
				xor_srcs[count++] = dev->page;
		}
	}

	/* 1/ if we prexor'd then the dest is reused as a source
	 * 2/ if we did not prexor then we are redoing the parity
	 * set ASYNC_TX_XOR_DROP_DST and ASYNC_TX_XOR_ZERO_DST
	 * for the synchronous xor case
	 */
1153
	flags = ASYNC_TX_ACK |
1154 1155 1156 1157
		(prexor ? ASYNC_TX_XOR_DROP_DST : ASYNC_TX_XOR_ZERO_DST);

	atomic_inc(&sh->count);

1158
	init_async_submit(&submit, flags, tx, ops_complete_reconstruct, sh,
1159
			  to_addr_conv(sh, percpu));
1160 1161 1162 1163
	if (unlikely(count == 1))
		tx = async_memcpy(xor_dest, xor_srcs[0], 0, 0, STRIPE_SIZE, &submit);
	else
		tx = async_xor(xor_dest, xor_srcs, 0, count, STRIPE_SIZE, &submit);
1164 1165
}

1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182
static void
ops_run_reconstruct6(struct stripe_head *sh, struct raid5_percpu *percpu,
		     struct dma_async_tx_descriptor *tx)
{
	struct async_submit_ctl submit;
	struct page **blocks = percpu->scribble;
	int count;

	pr_debug("%s: stripe %llu\n", __func__, (unsigned long long)sh->sector);

	count = set_syndrome_sources(blocks, sh);

	atomic_inc(&sh->count);

	init_async_submit(&submit, ASYNC_TX_ACK, tx, ops_complete_reconstruct,
			  sh, to_addr_conv(sh, percpu));
	async_gen_syndrome(blocks, 0, count+2, STRIPE_SIZE,  &submit);
1183 1184 1185 1186 1187 1188
}

static void ops_complete_check(void *stripe_head_ref)
{
	struct stripe_head *sh = stripe_head_ref;

1189
	pr_debug("%s: stripe %llu\n", __func__,
1190 1191
		(unsigned long long)sh->sector);

1192
	sh->check_state = check_state_check_result;
1193 1194 1195 1196
	set_bit(STRIPE_HANDLE, &sh->state);
	release_stripe(sh);
}

1197
static void ops_run_check_p(struct stripe_head *sh, struct raid5_percpu *percpu)
1198 1199
{
	int disks = sh->disks;
1200 1201 1202
	int pd_idx = sh->pd_idx;
	int qd_idx = sh->qd_idx;
	struct page *xor_dest;
1203
	struct page **xor_srcs = percpu->scribble;
1204
	struct dma_async_tx_descriptor *tx;
1205
	struct async_submit_ctl submit;
1206 1207
	int count;
	int i;
1208

1209
	pr_debug("%s: stripe %llu\n", __func__,
1210 1211
		(unsigned long long)sh->sector);

1212 1213 1214
	count = 0;
	xor_dest = sh->dev[pd_idx].page;
	xor_srcs[count++] = xor_dest;
1215
	for (i = disks; i--; ) {
1216 1217 1218
		if (i == pd_idx || i == qd_idx)
			continue;
		xor_srcs[count++] = sh->dev[i].page;
1219 1220
	}

1221 1222
	init_async_submit(&submit, 0, NULL, NULL, NULL,
			  to_addr_conv(sh, percpu));
D
Dan Williams 已提交
1223
	tx = async_xor_val(xor_dest, xor_srcs, 0, count, STRIPE_SIZE,
1224
			   &sh->ops.zero_sum_result, &submit);
1225 1226

	atomic_inc(&sh->count);
1227 1228
	init_async_submit(&submit, ASYNC_TX_ACK, tx, ops_complete_check, sh, NULL);
	tx = async_trigger_callback(&submit);
1229 1230
}

1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242
static void ops_run_check_pq(struct stripe_head *sh, struct raid5_percpu *percpu, int checkp)
{
	struct page **srcs = percpu->scribble;
	struct async_submit_ctl submit;
	int count;

	pr_debug("%s: stripe %llu checkp: %d\n", __func__,
		(unsigned long long)sh->sector, checkp);

	count = set_syndrome_sources(srcs, sh);
	if (!checkp)
		srcs[count] = NULL;
1243 1244

	atomic_inc(&sh->count);
1245 1246 1247 1248
	init_async_submit(&submit, ASYNC_TX_ACK, NULL, ops_complete_check,
			  sh, to_addr_conv(sh, percpu));
	async_syndrome_val(srcs, 0, count+2, STRIPE_SIZE,
			   &sh->ops.zero_sum_result, percpu->spare_page, &submit);
1249 1250
}

1251
static void __raid_run_ops(struct stripe_head *sh, unsigned long ops_request)
1252 1253 1254
{
	int overlap_clear = 0, i, disks = sh->disks;
	struct dma_async_tx_descriptor *tx = NULL;
1255
	struct r5conf *conf = sh->raid_conf;
1256
	int level = conf->level;
1257 1258
	struct raid5_percpu *percpu;
	unsigned long cpu;
1259

1260 1261
	cpu = get_cpu();
	percpu = per_cpu_ptr(conf->percpu, cpu);
1262
	if (test_bit(STRIPE_OP_BIOFILL, &ops_request)) {
1263 1264 1265 1266
		ops_run_biofill(sh);
		overlap_clear++;
	}

1267
	if (test_bit(STRIPE_OP_COMPUTE_BLK, &ops_request)) {
1268 1269 1270 1271 1272 1273 1274 1275 1276 1277
		if (level < 6)
			tx = ops_run_compute5(sh, percpu);
		else {
			if (sh->ops.target2 < 0 || sh->ops.target < 0)
				tx = ops_run_compute6_1(sh, percpu);
			else
				tx = ops_run_compute6_2(sh, percpu);
		}
		/* terminate the chain if reconstruct is not set to be run */
		if (tx && !test_bit(STRIPE_OP_RECONSTRUCT, &ops_request))
1278 1279
			async_tx_ack(tx);
	}
1280

1281
	if (test_bit(STRIPE_OP_PREXOR, &ops_request))
1282
		tx = ops_run_prexor(sh, percpu, tx);
1283

1284
	if (test_bit(STRIPE_OP_BIODRAIN, &ops_request)) {
1285
		tx = ops_run_biodrain(sh, tx);
1286 1287 1288
		overlap_clear++;
	}

1289 1290 1291 1292 1293 1294
	if (test_bit(STRIPE_OP_RECONSTRUCT, &ops_request)) {
		if (level < 6)
			ops_run_reconstruct5(sh, percpu, tx);
		else
			ops_run_reconstruct6(sh, percpu, tx);
	}
1295

1296 1297 1298 1299 1300 1301 1302 1303 1304 1305
	if (test_bit(STRIPE_OP_CHECK, &ops_request)) {
		if (sh->check_state == check_state_run)
			ops_run_check_p(sh, percpu);
		else if (sh->check_state == check_state_run_q)
			ops_run_check_pq(sh, percpu, 0);
		else if (sh->check_state == check_state_run_pq)
			ops_run_check_pq(sh, percpu, 1);
		else
			BUG();
	}
1306 1307 1308 1309 1310 1311 1312

	if (overlap_clear)
		for (i = disks; i--; ) {
			struct r5dev *dev = &sh->dev[i];
			if (test_and_clear_bit(R5_Overlap, &dev->flags))
				wake_up(&sh->raid_conf->wait_for_overlap);
		}
1313
	put_cpu();
1314 1315
}

1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345
#ifdef CONFIG_MULTICORE_RAID456
static void async_run_ops(void *param, async_cookie_t cookie)
{
	struct stripe_head *sh = param;
	unsigned long ops_request = sh->ops.request;

	clear_bit_unlock(STRIPE_OPS_REQ_PENDING, &sh->state);
	wake_up(&sh->ops.wait_for_ops);

	__raid_run_ops(sh, ops_request);
	release_stripe(sh);
}

static void raid_run_ops(struct stripe_head *sh, unsigned long ops_request)
{
	/* since handle_stripe can be called outside of raid5d context
	 * we need to ensure sh->ops.request is de-staged before another
	 * request arrives
	 */
	wait_event(sh->ops.wait_for_ops,
		   !test_and_set_bit_lock(STRIPE_OPS_REQ_PENDING, &sh->state));
	sh->ops.request = ops_request;

	atomic_inc(&sh->count);
	async_schedule(async_run_ops, sh);
}
#else
#define raid_run_ops __raid_run_ops
#endif

1346
static int grow_one_stripe(struct r5conf *conf)
L
Linus Torvalds 已提交
1347 1348
{
	struct stripe_head *sh;
N
Namhyung Kim 已提交
1349
	sh = kmem_cache_zalloc(conf->slab_cache, GFP_KERNEL);
1350 1351
	if (!sh)
		return 0;
N
Namhyung Kim 已提交
1352

1353
	sh->raid_conf = conf;
1354 1355 1356
	#ifdef CONFIG_MULTICORE_RAID456
	init_waitqueue_head(&sh->ops.wait_for_ops);
	#endif
1357

1358 1359
	if (grow_buffers(sh)) {
		shrink_buffers(sh);
1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370
		kmem_cache_free(conf->slab_cache, sh);
		return 0;
	}
	/* we just created an active stripe so... */
	atomic_set(&sh->count, 1);
	atomic_inc(&conf->active_stripes);
	INIT_LIST_HEAD(&sh->lru);
	release_stripe(sh);
	return 1;
}

1371
static int grow_stripes(struct r5conf *conf, int num)
1372
{
1373
	struct kmem_cache *sc;
1374
	int devs = max(conf->raid_disks, conf->previous_raid_disks);
L
Linus Torvalds 已提交
1375

1376 1377 1378 1379 1380 1381 1382 1383
	if (conf->mddev->gendisk)
		sprintf(conf->cache_name[0],
			"raid%d-%s", conf->level, mdname(conf->mddev));
	else
		sprintf(conf->cache_name[0],
			"raid%d-%p", conf->level, conf->mddev);
	sprintf(conf->cache_name[1], "%s-alt", conf->cache_name[0]);

1384 1385
	conf->active_name = 0;
	sc = kmem_cache_create(conf->cache_name[conf->active_name],
L
Linus Torvalds 已提交
1386
			       sizeof(struct stripe_head)+(devs-1)*sizeof(struct r5dev),
1387
			       0, 0, NULL);
L
Linus Torvalds 已提交
1388 1389 1390
	if (!sc)
		return 1;
	conf->slab_cache = sc;
1391
	conf->pool_size = devs;
1392
	while (num--)
1393
		if (!grow_one_stripe(conf))
L
Linus Torvalds 已提交
1394 1395 1396
			return 1;
	return 0;
}
1397

1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419
/**
 * scribble_len - return the required size of the scribble region
 * @num - total number of disks in the array
 *
 * The size must be enough to contain:
 * 1/ a struct page pointer for each device in the array +2
 * 2/ room to convert each entry in (1) to its corresponding dma
 *    (dma_map_page()) or page (page_address()) address.
 *
 * Note: the +2 is for the destination buffers of the ddf/raid6 case where we
 * calculate over all devices (not just the data blocks), using zeros in place
 * of the P and Q blocks.
 */
static size_t scribble_len(int num)
{
	size_t len;

	len = sizeof(struct page *) * (num+2) + sizeof(addr_conv_t) * (num+2);

	return len;
}

1420
static int resize_stripes(struct r5conf *conf, int newsize)
1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447
{
	/* Make all the stripes able to hold 'newsize' devices.
	 * New slots in each stripe get 'page' set to a new page.
	 *
	 * This happens in stages:
	 * 1/ create a new kmem_cache and allocate the required number of
	 *    stripe_heads.
	 * 2/ gather all the old stripe_heads and tranfer the pages across
	 *    to the new stripe_heads.  This will have the side effect of
	 *    freezing the array as once all stripe_heads have been collected,
	 *    no IO will be possible.  Old stripe heads are freed once their
	 *    pages have been transferred over, and the old kmem_cache is
	 *    freed when all stripes are done.
	 * 3/ reallocate conf->disks to be suitable bigger.  If this fails,
	 *    we simple return a failre status - no need to clean anything up.
	 * 4/ allocate new pages for the new slots in the new stripe_heads.
	 *    If this fails, we don't bother trying the shrink the
	 *    stripe_heads down again, we just leave them as they are.
	 *    As each stripe_head is processed the new one is released into
	 *    active service.
	 *
	 * Once step2 is started, we cannot afford to wait for a write,
	 * so we use GFP_NOIO allocations.
	 */
	struct stripe_head *osh, *nsh;
	LIST_HEAD(newstripes);
	struct disk_info *ndisks;
1448
	unsigned long cpu;
1449
	int err;
1450
	struct kmem_cache *sc;
1451 1452 1453 1454 1455
	int i;

	if (newsize <= conf->pool_size)
		return 0; /* never bother to shrink */

1456 1457 1458
	err = md_allow_write(conf->mddev);
	if (err)
		return err;
1459

1460 1461 1462
	/* Step 1 */
	sc = kmem_cache_create(conf->cache_name[1-conf->active_name],
			       sizeof(struct stripe_head)+(newsize-1)*sizeof(struct r5dev),
1463
			       0, 0, NULL);
1464 1465 1466 1467
	if (!sc)
		return -ENOMEM;

	for (i = conf->max_nr_stripes; i; i--) {
N
Namhyung Kim 已提交
1468
		nsh = kmem_cache_zalloc(sc, GFP_KERNEL);
1469 1470 1471 1472
		if (!nsh)
			break;

		nsh->raid_conf = conf;
1473 1474 1475
		#ifdef CONFIG_MULTICORE_RAID456
		init_waitqueue_head(&nsh->ops.wait_for_ops);
		#endif
1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497

		list_add(&nsh->lru, &newstripes);
	}
	if (i) {
		/* didn't get enough, give up */
		while (!list_empty(&newstripes)) {
			nsh = list_entry(newstripes.next, struct stripe_head, lru);
			list_del(&nsh->lru);
			kmem_cache_free(sc, nsh);
		}
		kmem_cache_destroy(sc);
		return -ENOMEM;
	}
	/* Step 2 - Must use GFP_NOIO now.
	 * OK, we have enough stripes, start collecting inactive
	 * stripes and copying them over
	 */
	list_for_each_entry(nsh, &newstripes, lru) {
		spin_lock_irq(&conf->device_lock);
		wait_event_lock_irq(conf->wait_for_stripe,
				    !list_empty(&conf->inactive_list),
				    conf->device_lock,
N
NeilBrown 已提交
1498
				    );
1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512
		osh = get_free_stripe(conf);
		spin_unlock_irq(&conf->device_lock);
		atomic_set(&nsh->count, 1);
		for(i=0; i<conf->pool_size; i++)
			nsh->dev[i].page = osh->dev[i].page;
		for( ; i<newsize; i++)
			nsh->dev[i].page = NULL;
		kmem_cache_free(conf->slab_cache, osh);
	}
	kmem_cache_destroy(conf->slab_cache);

	/* Step 3.
	 * At this point, we are holding all the stripes so the array
	 * is completely stalled, so now is a good time to resize
1513
	 * conf->disks and the scribble region
1514 1515 1516 1517 1518 1519 1520 1521 1522 1523
	 */
	ndisks = kzalloc(newsize * sizeof(struct disk_info), GFP_NOIO);
	if (ndisks) {
		for (i=0; i<conf->raid_disks; i++)
			ndisks[i] = conf->disks[i];
		kfree(conf->disks);
		conf->disks = ndisks;
	} else
		err = -ENOMEM;

1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542
	get_online_cpus();
	conf->scribble_len = scribble_len(newsize);
	for_each_present_cpu(cpu) {
		struct raid5_percpu *percpu;
		void *scribble;

		percpu = per_cpu_ptr(conf->percpu, cpu);
		scribble = kmalloc(conf->scribble_len, GFP_NOIO);

		if (scribble) {
			kfree(percpu->scribble);
			percpu->scribble = scribble;
		} else {
			err = -ENOMEM;
			break;
		}
	}
	put_online_cpus();

1543 1544 1545 1546
	/* Step 4, return new stripes to service */
	while(!list_empty(&newstripes)) {
		nsh = list_entry(newstripes.next, struct stripe_head, lru);
		list_del_init(&nsh->lru);
1547

1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563
		for (i=conf->raid_disks; i < newsize; i++)
			if (nsh->dev[i].page == NULL) {
				struct page *p = alloc_page(GFP_NOIO);
				nsh->dev[i].page = p;
				if (!p)
					err = -ENOMEM;
			}
		release_stripe(nsh);
	}
	/* critical section pass, GFP_NOIO no longer needed */

	conf->slab_cache = sc;
	conf->active_name = 1-conf->active_name;
	conf->pool_size = newsize;
	return err;
}
L
Linus Torvalds 已提交
1564

1565
static int drop_one_stripe(struct r5conf *conf)
L
Linus Torvalds 已提交
1566 1567 1568
{
	struct stripe_head *sh;

1569 1570 1571 1572 1573
	spin_lock_irq(&conf->device_lock);
	sh = get_free_stripe(conf);
	spin_unlock_irq(&conf->device_lock);
	if (!sh)
		return 0;
1574
	BUG_ON(atomic_read(&sh->count));
1575
	shrink_buffers(sh);
1576 1577 1578 1579 1580
	kmem_cache_free(conf->slab_cache, sh);
	atomic_dec(&conf->active_stripes);
	return 1;
}

1581
static void shrink_stripes(struct r5conf *conf)
1582 1583 1584 1585
{
	while (drop_one_stripe(conf))
		;

N
NeilBrown 已提交
1586 1587
	if (conf->slab_cache)
		kmem_cache_destroy(conf->slab_cache);
L
Linus Torvalds 已提交
1588 1589 1590
	conf->slab_cache = NULL;
}

1591
static void raid5_end_read_request(struct bio * bi, int error)
L
Linus Torvalds 已提交
1592
{
1593
	struct stripe_head *sh = bi->bi_private;
1594
	struct r5conf *conf = sh->raid_conf;
1595
	int disks = sh->disks, i;
L
Linus Torvalds 已提交
1596
	int uptodate = test_bit(BIO_UPTODATE, &bi->bi_flags);
1597
	char b[BDEVNAME_SIZE];
1598
	struct md_rdev *rdev;
L
Linus Torvalds 已提交
1599 1600 1601 1602 1603 1604


	for (i=0 ; i<disks; i++)
		if (bi == &sh->dev[i].req)
			break;

1605 1606
	pr_debug("end_read_request %llu/%d, count: %d, uptodate %d.\n",
		(unsigned long long)sh->sector, i, atomic_read(&sh->count),
L
Linus Torvalds 已提交
1607 1608 1609
		uptodate);
	if (i == disks) {
		BUG();
1610
		return;
L
Linus Torvalds 已提交
1611
	}
1612 1613 1614 1615
	if (test_bit(R5_ReadRepl, &sh->dev[i].flags))
		rdev = conf->disks[i].replacement;
	else
		rdev = conf->disks[i].rdev;
L
Linus Torvalds 已提交
1616 1617 1618

	if (uptodate) {
		set_bit(R5_UPTODATE, &sh->dev[i].flags);
1619
		if (test_bit(R5_ReadError, &sh->dev[i].flags)) {
1620 1621 1622 1623
			/* Note that this cannot happen on a
			 * replacement device.  We just fail those on
			 * any error
			 */
1624 1625 1626 1627 1628 1629 1630 1631
			printk_ratelimited(
				KERN_INFO
				"md/raid:%s: read error corrected"
				" (%lu sectors at %llu on %s)\n",
				mdname(conf->mddev), STRIPE_SECTORS,
				(unsigned long long)(sh->sector
						     + rdev->data_offset),
				bdevname(rdev->bdev, b));
1632
			atomic_add(STRIPE_SECTORS, &rdev->corrected_errors);
1633 1634 1635
			clear_bit(R5_ReadError, &sh->dev[i].flags);
			clear_bit(R5_ReWrite, &sh->dev[i].flags);
		}
1636 1637
		if (atomic_read(&rdev->read_errors))
			atomic_set(&rdev->read_errors, 0);
L
Linus Torvalds 已提交
1638
	} else {
1639
		const char *bdn = bdevname(rdev->bdev, b);
1640
		int retry = 0;
1641

L
Linus Torvalds 已提交
1642
		clear_bit(R5_UPTODATE, &sh->dev[i].flags);
1643
		atomic_inc(&rdev->read_errors);
1644 1645 1646 1647 1648 1649 1650 1651 1652 1653
		if (test_bit(R5_ReadRepl, &sh->dev[i].flags))
			printk_ratelimited(
				KERN_WARNING
				"md/raid:%s: read error on replacement device "
				"(sector %llu on %s).\n",
				mdname(conf->mddev),
				(unsigned long long)(sh->sector
						     + rdev->data_offset),
				bdn);
		else if (conf->mddev->degraded >= conf->max_degraded)
1654 1655 1656 1657 1658 1659 1660 1661
			printk_ratelimited(
				KERN_WARNING
				"md/raid:%s: read error not correctable "
				"(sector %llu on %s).\n",
				mdname(conf->mddev),
				(unsigned long long)(sh->sector
						     + rdev->data_offset),
				bdn);
1662
		else if (test_bit(R5_ReWrite, &sh->dev[i].flags))
1663
			/* Oh, no!!! */
1664 1665 1666 1667 1668 1669 1670 1671
			printk_ratelimited(
				KERN_WARNING
				"md/raid:%s: read error NOT corrected!! "
				"(sector %llu on %s).\n",
				mdname(conf->mddev),
				(unsigned long long)(sh->sector
						     + rdev->data_offset),
				bdn);
1672
		else if (atomic_read(&rdev->read_errors)
1673
			 > conf->max_nr_stripes)
N
NeilBrown 已提交
1674
			printk(KERN_WARNING
1675
			       "md/raid:%s: Too many read errors, failing device %s.\n",
1676
			       mdname(conf->mddev), bdn);
1677 1678 1679 1680 1681
		else
			retry = 1;
		if (retry)
			set_bit(R5_ReadError, &sh->dev[i].flags);
		else {
1682 1683
			clear_bit(R5_ReadError, &sh->dev[i].flags);
			clear_bit(R5_ReWrite, &sh->dev[i].flags);
1684
			md_error(conf->mddev, rdev);
1685
		}
L
Linus Torvalds 已提交
1686
	}
1687
	rdev_dec_pending(rdev, conf->mddev);
L
Linus Torvalds 已提交
1688 1689 1690 1691 1692
	clear_bit(R5_LOCKED, &sh->dev[i].flags);
	set_bit(STRIPE_HANDLE, &sh->state);
	release_stripe(sh);
}

1693
static void raid5_end_write_request(struct bio *bi, int error)
L
Linus Torvalds 已提交
1694
{
1695
	struct stripe_head *sh = bi->bi_private;
1696
	struct r5conf *conf = sh->raid_conf;
1697
	int disks = sh->disks, i;
L
Linus Torvalds 已提交
1698
	int uptodate = test_bit(BIO_UPTODATE, &bi->bi_flags);
1699 1700
	sector_t first_bad;
	int bad_sectors;
L
Linus Torvalds 已提交
1701 1702 1703 1704 1705

	for (i=0 ; i<disks; i++)
		if (bi == &sh->dev[i].req)
			break;

1706
	pr_debug("end_write_request %llu/%d, count %d, uptodate: %d.\n",
L
Linus Torvalds 已提交
1707 1708 1709 1710
		(unsigned long long)sh->sector, i, atomic_read(&sh->count),
		uptodate);
	if (i == disks) {
		BUG();
1711
		return;
L
Linus Torvalds 已提交
1712 1713
	}

1714 1715 1716
	if (!uptodate) {
		set_bit(WriteErrorSeen, &conf->disks[i].rdev->flags);
		set_bit(R5_WriteError, &sh->dev[i].flags);
1717 1718 1719
	} else if (is_badblock(conf->disks[i].rdev, sh->sector, STRIPE_SECTORS,
			       &first_bad, &bad_sectors))
		set_bit(R5_MadeGood, &sh->dev[i].flags);
L
Linus Torvalds 已提交
1720 1721 1722 1723 1724

	rdev_dec_pending(conf->disks[i].rdev, conf->mddev);
	
	clear_bit(R5_LOCKED, &sh->dev[i].flags);
	set_bit(STRIPE_HANDLE, &sh->state);
1725
	release_stripe(sh);
L
Linus Torvalds 已提交
1726 1727 1728
}


1729
static sector_t compute_blocknr(struct stripe_head *sh, int i, int previous);
L
Linus Torvalds 已提交
1730
	
1731
static void raid5_build_block(struct stripe_head *sh, int i, int previous)
L
Linus Torvalds 已提交
1732 1733 1734 1735 1736 1737 1738 1739
{
	struct r5dev *dev = &sh->dev[i];

	bio_init(&dev->req);
	dev->req.bi_io_vec = &dev->vec;
	dev->req.bi_vcnt++;
	dev->req.bi_max_vecs++;
	dev->req.bi_private = sh;
1740
	dev->vec.bv_page = dev->page;
L
Linus Torvalds 已提交
1741 1742

	dev->flags = 0;
1743
	dev->sector = compute_blocknr(sh, i, previous);
L
Linus Torvalds 已提交
1744 1745
}

1746
static void error(struct mddev *mddev, struct md_rdev *rdev)
L
Linus Torvalds 已提交
1747 1748
{
	char b[BDEVNAME_SIZE];
1749
	struct r5conf *conf = mddev->private;
1750
	unsigned long flags;
1751
	pr_debug("raid456: error called\n");
L
Linus Torvalds 已提交
1752

1753 1754 1755 1756 1757 1758
	spin_lock_irqsave(&conf->device_lock, flags);
	clear_bit(In_sync, &rdev->flags);
	mddev->degraded = calc_degraded(conf);
	spin_unlock_irqrestore(&conf->device_lock, flags);
	set_bit(MD_RECOVERY_INTR, &mddev->recovery);

1759
	set_bit(Blocked, &rdev->flags);
1760 1761 1762 1763 1764 1765 1766 1767 1768
	set_bit(Faulty, &rdev->flags);
	set_bit(MD_CHANGE_DEVS, &mddev->flags);
	printk(KERN_ALERT
	       "md/raid:%s: Disk failure on %s, disabling device.\n"
	       "md/raid:%s: Operation continuing on %d devices.\n",
	       mdname(mddev),
	       bdevname(rdev->bdev, b),
	       mdname(mddev),
	       conf->raid_disks - mddev->degraded);
1769
}
L
Linus Torvalds 已提交
1770 1771 1772 1773 1774

/*
 * Input: a 'big' sector number,
 * Output: index of the data and parity disk, and the sector # in them.
 */
1775
static sector_t raid5_compute_sector(struct r5conf *conf, sector_t r_sector,
1776 1777
				     int previous, int *dd_idx,
				     struct stripe_head *sh)
L
Linus Torvalds 已提交
1778
{
N
NeilBrown 已提交
1779
	sector_t stripe, stripe2;
1780
	sector_t chunk_number;
L
Linus Torvalds 已提交
1781
	unsigned int chunk_offset;
1782
	int pd_idx, qd_idx;
1783
	int ddf_layout = 0;
L
Linus Torvalds 已提交
1784
	sector_t new_sector;
1785 1786
	int algorithm = previous ? conf->prev_algo
				 : conf->algorithm;
1787 1788
	int sectors_per_chunk = previous ? conf->prev_chunk_sectors
					 : conf->chunk_sectors;
1789 1790 1791
	int raid_disks = previous ? conf->previous_raid_disks
				  : conf->raid_disks;
	int data_disks = raid_disks - conf->max_degraded;
L
Linus Torvalds 已提交
1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803

	/* First compute the information on this sector */

	/*
	 * Compute the chunk number and the sector offset inside the chunk
	 */
	chunk_offset = sector_div(r_sector, sectors_per_chunk);
	chunk_number = r_sector;

	/*
	 * Compute the stripe number
	 */
1804 1805
	stripe = chunk_number;
	*dd_idx = sector_div(stripe, data_disks);
N
NeilBrown 已提交
1806
	stripe2 = stripe;
L
Linus Torvalds 已提交
1807 1808 1809
	/*
	 * Select the parity disk based on the user selected algorithm.
	 */
1810
	pd_idx = qd_idx = -1;
1811 1812
	switch(conf->level) {
	case 4:
1813
		pd_idx = data_disks;
1814 1815
		break;
	case 5:
1816
		switch (algorithm) {
L
Linus Torvalds 已提交
1817
		case ALGORITHM_LEFT_ASYMMETRIC:
N
NeilBrown 已提交
1818
			pd_idx = data_disks - sector_div(stripe2, raid_disks);
1819
			if (*dd_idx >= pd_idx)
L
Linus Torvalds 已提交
1820 1821 1822
				(*dd_idx)++;
			break;
		case ALGORITHM_RIGHT_ASYMMETRIC:
N
NeilBrown 已提交
1823
			pd_idx = sector_div(stripe2, raid_disks);
1824
			if (*dd_idx >= pd_idx)
L
Linus Torvalds 已提交
1825 1826 1827
				(*dd_idx)++;
			break;
		case ALGORITHM_LEFT_SYMMETRIC:
N
NeilBrown 已提交
1828
			pd_idx = data_disks - sector_div(stripe2, raid_disks);
1829
			*dd_idx = (pd_idx + 1 + *dd_idx) % raid_disks;
L
Linus Torvalds 已提交
1830 1831
			break;
		case ALGORITHM_RIGHT_SYMMETRIC:
N
NeilBrown 已提交
1832
			pd_idx = sector_div(stripe2, raid_disks);
1833
			*dd_idx = (pd_idx + 1 + *dd_idx) % raid_disks;
L
Linus Torvalds 已提交
1834
			break;
1835 1836 1837 1838 1839 1840 1841
		case ALGORITHM_PARITY_0:
			pd_idx = 0;
			(*dd_idx)++;
			break;
		case ALGORITHM_PARITY_N:
			pd_idx = data_disks;
			break;
L
Linus Torvalds 已提交
1842
		default:
1843
			BUG();
1844 1845 1846 1847
		}
		break;
	case 6:

1848
		switch (algorithm) {
1849
		case ALGORITHM_LEFT_ASYMMETRIC:
N
NeilBrown 已提交
1850
			pd_idx = raid_disks - 1 - sector_div(stripe2, raid_disks);
1851 1852
			qd_idx = pd_idx + 1;
			if (pd_idx == raid_disks-1) {
1853
				(*dd_idx)++;	/* Q D D D P */
1854 1855
				qd_idx = 0;
			} else if (*dd_idx >= pd_idx)
1856 1857 1858
				(*dd_idx) += 2; /* D D P Q D */
			break;
		case ALGORITHM_RIGHT_ASYMMETRIC:
N
NeilBrown 已提交
1859
			pd_idx = sector_div(stripe2, raid_disks);
1860 1861
			qd_idx = pd_idx + 1;
			if (pd_idx == raid_disks-1) {
1862
				(*dd_idx)++;	/* Q D D D P */
1863 1864
				qd_idx = 0;
			} else if (*dd_idx >= pd_idx)
1865 1866 1867
				(*dd_idx) += 2; /* D D P Q D */
			break;
		case ALGORITHM_LEFT_SYMMETRIC:
N
NeilBrown 已提交
1868
			pd_idx = raid_disks - 1 - sector_div(stripe2, raid_disks);
1869 1870
			qd_idx = (pd_idx + 1) % raid_disks;
			*dd_idx = (pd_idx + 2 + *dd_idx) % raid_disks;
1871 1872
			break;
		case ALGORITHM_RIGHT_SYMMETRIC:
N
NeilBrown 已提交
1873
			pd_idx = sector_div(stripe2, raid_disks);
1874 1875
			qd_idx = (pd_idx + 1) % raid_disks;
			*dd_idx = (pd_idx + 2 + *dd_idx) % raid_disks;
1876
			break;
1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891

		case ALGORITHM_PARITY_0:
			pd_idx = 0;
			qd_idx = 1;
			(*dd_idx) += 2;
			break;
		case ALGORITHM_PARITY_N:
			pd_idx = data_disks;
			qd_idx = data_disks + 1;
			break;

		case ALGORITHM_ROTATING_ZERO_RESTART:
			/* Exactly the same as RIGHT_ASYMMETRIC, but or
			 * of blocks for computing Q is different.
			 */
N
NeilBrown 已提交
1892
			pd_idx = sector_div(stripe2, raid_disks);
1893 1894 1895 1896 1897 1898
			qd_idx = pd_idx + 1;
			if (pd_idx == raid_disks-1) {
				(*dd_idx)++;	/* Q D D D P */
				qd_idx = 0;
			} else if (*dd_idx >= pd_idx)
				(*dd_idx) += 2; /* D D P Q D */
1899
			ddf_layout = 1;
1900 1901 1902 1903 1904 1905 1906
			break;

		case ALGORITHM_ROTATING_N_RESTART:
			/* Same a left_asymmetric, by first stripe is
			 * D D D P Q  rather than
			 * Q D D D P
			 */
N
NeilBrown 已提交
1907 1908
			stripe2 += 1;
			pd_idx = raid_disks - 1 - sector_div(stripe2, raid_disks);
1909 1910 1911 1912 1913 1914
			qd_idx = pd_idx + 1;
			if (pd_idx == raid_disks-1) {
				(*dd_idx)++;	/* Q D D D P */
				qd_idx = 0;
			} else if (*dd_idx >= pd_idx)
				(*dd_idx) += 2; /* D D P Q D */
1915
			ddf_layout = 1;
1916 1917 1918 1919
			break;

		case ALGORITHM_ROTATING_N_CONTINUE:
			/* Same as left_symmetric but Q is before P */
N
NeilBrown 已提交
1920
			pd_idx = raid_disks - 1 - sector_div(stripe2, raid_disks);
1921 1922
			qd_idx = (pd_idx + raid_disks - 1) % raid_disks;
			*dd_idx = (pd_idx + 1 + *dd_idx) % raid_disks;
1923
			ddf_layout = 1;
1924 1925 1926 1927
			break;

		case ALGORITHM_LEFT_ASYMMETRIC_6:
			/* RAID5 left_asymmetric, with Q on last device */
N
NeilBrown 已提交
1928
			pd_idx = data_disks - sector_div(stripe2, raid_disks-1);
1929 1930 1931 1932 1933 1934
			if (*dd_idx >= pd_idx)
				(*dd_idx)++;
			qd_idx = raid_disks - 1;
			break;

		case ALGORITHM_RIGHT_ASYMMETRIC_6:
N
NeilBrown 已提交
1935
			pd_idx = sector_div(stripe2, raid_disks-1);
1936 1937 1938 1939 1940 1941
			if (*dd_idx >= pd_idx)
				(*dd_idx)++;
			qd_idx = raid_disks - 1;
			break;

		case ALGORITHM_LEFT_SYMMETRIC_6:
N
NeilBrown 已提交
1942
			pd_idx = data_disks - sector_div(stripe2, raid_disks-1);
1943 1944 1945 1946 1947
			*dd_idx = (pd_idx + 1 + *dd_idx) % (raid_disks-1);
			qd_idx = raid_disks - 1;
			break;

		case ALGORITHM_RIGHT_SYMMETRIC_6:
N
NeilBrown 已提交
1948
			pd_idx = sector_div(stripe2, raid_disks-1);
1949 1950 1951 1952 1953 1954 1955 1956 1957 1958
			*dd_idx = (pd_idx + 1 + *dd_idx) % (raid_disks-1);
			qd_idx = raid_disks - 1;
			break;

		case ALGORITHM_PARITY_0_6:
			pd_idx = 0;
			(*dd_idx)++;
			qd_idx = raid_disks - 1;
			break;

1959
		default:
1960
			BUG();
1961 1962
		}
		break;
L
Linus Torvalds 已提交
1963 1964
	}

1965 1966 1967
	if (sh) {
		sh->pd_idx = pd_idx;
		sh->qd_idx = qd_idx;
1968
		sh->ddf_layout = ddf_layout;
1969
	}
L
Linus Torvalds 已提交
1970 1971 1972 1973 1974 1975 1976 1977
	/*
	 * Finally, compute the new sector number
	 */
	new_sector = (sector_t)stripe * sectors_per_chunk + chunk_offset;
	return new_sector;
}


1978
static sector_t compute_blocknr(struct stripe_head *sh, int i, int previous)
L
Linus Torvalds 已提交
1979
{
1980
	struct r5conf *conf = sh->raid_conf;
1981 1982
	int raid_disks = sh->disks;
	int data_disks = raid_disks - conf->max_degraded;
L
Linus Torvalds 已提交
1983
	sector_t new_sector = sh->sector, check;
1984 1985
	int sectors_per_chunk = previous ? conf->prev_chunk_sectors
					 : conf->chunk_sectors;
1986 1987
	int algorithm = previous ? conf->prev_algo
				 : conf->algorithm;
L
Linus Torvalds 已提交
1988 1989
	sector_t stripe;
	int chunk_offset;
1990 1991
	sector_t chunk_number;
	int dummy1, dd_idx = i;
L
Linus Torvalds 已提交
1992
	sector_t r_sector;
1993
	struct stripe_head sh2;
L
Linus Torvalds 已提交
1994

1995

L
Linus Torvalds 已提交
1996 1997 1998
	chunk_offset = sector_div(new_sector, sectors_per_chunk);
	stripe = new_sector;

1999 2000 2001 2002 2003
	if (i == sh->pd_idx)
		return 0;
	switch(conf->level) {
	case 4: break;
	case 5:
2004
		switch (algorithm) {
L
Linus Torvalds 已提交
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
		case ALGORITHM_LEFT_ASYMMETRIC:
		case ALGORITHM_RIGHT_ASYMMETRIC:
			if (i > sh->pd_idx)
				i--;
			break;
		case ALGORITHM_LEFT_SYMMETRIC:
		case ALGORITHM_RIGHT_SYMMETRIC:
			if (i < sh->pd_idx)
				i += raid_disks;
			i -= (sh->pd_idx + 1);
			break;
2016 2017 2018 2019 2020
		case ALGORITHM_PARITY_0:
			i -= 1;
			break;
		case ALGORITHM_PARITY_N:
			break;
L
Linus Torvalds 已提交
2021
		default:
2022
			BUG();
2023 2024 2025
		}
		break;
	case 6:
2026
		if (i == sh->qd_idx)
2027
			return 0; /* It is the Q disk */
2028
		switch (algorithm) {
2029 2030
		case ALGORITHM_LEFT_ASYMMETRIC:
		case ALGORITHM_RIGHT_ASYMMETRIC:
2031 2032 2033 2034
		case ALGORITHM_ROTATING_ZERO_RESTART:
		case ALGORITHM_ROTATING_N_RESTART:
			if (sh->pd_idx == raid_disks-1)
				i--;	/* Q D D D P */
2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048
			else if (i > sh->pd_idx)
				i -= 2; /* D D P Q D */
			break;
		case ALGORITHM_LEFT_SYMMETRIC:
		case ALGORITHM_RIGHT_SYMMETRIC:
			if (sh->pd_idx == raid_disks-1)
				i--; /* Q D D D P */
			else {
				/* D D P Q D */
				if (i < sh->pd_idx)
					i += raid_disks;
				i -= (sh->pd_idx + 2);
			}
			break;
2049 2050 2051 2052 2053 2054
		case ALGORITHM_PARITY_0:
			i -= 2;
			break;
		case ALGORITHM_PARITY_N:
			break;
		case ALGORITHM_ROTATING_N_CONTINUE:
2055
			/* Like left_symmetric, but P is before Q */
2056 2057
			if (sh->pd_idx == 0)
				i--;	/* P D D D Q */
2058 2059 2060 2061 2062 2063
			else {
				/* D D Q P D */
				if (i < sh->pd_idx)
					i += raid_disks;
				i -= (sh->pd_idx + 1);
			}
2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078
			break;
		case ALGORITHM_LEFT_ASYMMETRIC_6:
		case ALGORITHM_RIGHT_ASYMMETRIC_6:
			if (i > sh->pd_idx)
				i--;
			break;
		case ALGORITHM_LEFT_SYMMETRIC_6:
		case ALGORITHM_RIGHT_SYMMETRIC_6:
			if (i < sh->pd_idx)
				i += data_disks + 1;
			i -= (sh->pd_idx + 1);
			break;
		case ALGORITHM_PARITY_0_6:
			i -= 1;
			break;
2079
		default:
2080
			BUG();
2081 2082
		}
		break;
L
Linus Torvalds 已提交
2083 2084 2085
	}

	chunk_number = stripe * data_disks + i;
2086
	r_sector = chunk_number * sectors_per_chunk + chunk_offset;
L
Linus Torvalds 已提交
2087

2088
	check = raid5_compute_sector(conf, r_sector,
2089
				     previous, &dummy1, &sh2);
2090 2091
	if (check != sh->sector || dummy1 != dd_idx || sh2.pd_idx != sh->pd_idx
		|| sh2.qd_idx != sh->qd_idx) {
2092 2093
		printk(KERN_ERR "md/raid:%s: compute_blocknr: map not correct\n",
		       mdname(conf->mddev));
L
Linus Torvalds 已提交
2094 2095 2096 2097 2098 2099
		return 0;
	}
	return r_sector;
}


2100
static void
2101
schedule_reconstruction(struct stripe_head *sh, struct stripe_head_state *s,
2102
			 int rcw, int expand)
2103 2104
{
	int i, pd_idx = sh->pd_idx, disks = sh->disks;
2105
	struct r5conf *conf = sh->raid_conf;
2106
	int level = conf->level;
2107 2108 2109 2110 2111 2112 2113

	if (rcw) {
		/* if we are not expanding this is a proper write request, and
		 * there will be bios with new data to be drained into the
		 * stripe cache
		 */
		if (!expand) {
2114 2115 2116 2117
			sh->reconstruct_state = reconstruct_state_drain_run;
			set_bit(STRIPE_OP_BIODRAIN, &s->ops_request);
		} else
			sh->reconstruct_state = reconstruct_state_run;
2118

2119
		set_bit(STRIPE_OP_RECONSTRUCT, &s->ops_request);
2120 2121 2122 2123 2124 2125

		for (i = disks; i--; ) {
			struct r5dev *dev = &sh->dev[i];

			if (dev->towrite) {
				set_bit(R5_LOCKED, &dev->flags);
2126
				set_bit(R5_Wantdrain, &dev->flags);
2127 2128
				if (!expand)
					clear_bit(R5_UPTODATE, &dev->flags);
2129
				s->locked++;
2130 2131
			}
		}
2132
		if (s->locked + conf->max_degraded == disks)
2133
			if (!test_and_set_bit(STRIPE_FULL_WRITE, &sh->state))
2134
				atomic_inc(&conf->pending_full_writes);
2135
	} else {
2136
		BUG_ON(level == 6);
2137 2138 2139
		BUG_ON(!(test_bit(R5_UPTODATE, &sh->dev[pd_idx].flags) ||
			test_bit(R5_Wantcompute, &sh->dev[pd_idx].flags)));

2140
		sh->reconstruct_state = reconstruct_state_prexor_drain_run;
2141 2142
		set_bit(STRIPE_OP_PREXOR, &s->ops_request);
		set_bit(STRIPE_OP_BIODRAIN, &s->ops_request);
2143
		set_bit(STRIPE_OP_RECONSTRUCT, &s->ops_request);
2144 2145 2146 2147 2148 2149 2150 2151

		for (i = disks; i--; ) {
			struct r5dev *dev = &sh->dev[i];
			if (i == pd_idx)
				continue;

			if (dev->towrite &&
			    (test_bit(R5_UPTODATE, &dev->flags) ||
2152 2153
			     test_bit(R5_Wantcompute, &dev->flags))) {
				set_bit(R5_Wantdrain, &dev->flags);
2154 2155
				set_bit(R5_LOCKED, &dev->flags);
				clear_bit(R5_UPTODATE, &dev->flags);
2156
				s->locked++;
2157 2158 2159 2160
			}
		}
	}

2161
	/* keep the parity disk(s) locked while asynchronous operations
2162 2163 2164 2165
	 * are in flight
	 */
	set_bit(R5_LOCKED, &sh->dev[pd_idx].flags);
	clear_bit(R5_UPTODATE, &sh->dev[pd_idx].flags);
2166
	s->locked++;
2167

2168 2169 2170 2171 2172 2173 2174 2175 2176
	if (level == 6) {
		int qd_idx = sh->qd_idx;
		struct r5dev *dev = &sh->dev[qd_idx];

		set_bit(R5_LOCKED, &dev->flags);
		clear_bit(R5_UPTODATE, &dev->flags);
		s->locked++;
	}

2177
	pr_debug("%s: stripe %llu locked: %d ops_request: %lx\n",
2178
		__func__, (unsigned long long)sh->sector,
2179
		s->locked, s->ops_request);
2180
}
2181

L
Linus Torvalds 已提交
2182 2183
/*
 * Each stripe/dev can have one or more bion attached.
2184
 * toread/towrite point to the first in a chain.
L
Linus Torvalds 已提交
2185 2186 2187 2188 2189
 * The bi_next chain must be in order.
 */
static int add_stripe_bio(struct stripe_head *sh, struct bio *bi, int dd_idx, int forwrite)
{
	struct bio **bip;
2190
	struct r5conf *conf = sh->raid_conf;
2191
	int firstwrite=0;
L
Linus Torvalds 已提交
2192

2193
	pr_debug("adding bi b#%llu to stripe s#%llu\n",
L
Linus Torvalds 已提交
2194 2195 2196 2197 2198
		(unsigned long long)bi->bi_sector,
		(unsigned long long)sh->sector);


	spin_lock_irq(&conf->device_lock);
2199
	if (forwrite) {
L
Linus Torvalds 已提交
2200
		bip = &sh->dev[dd_idx].towrite;
2201 2202 2203
		if (*bip == NULL && sh->dev[dd_idx].written == NULL)
			firstwrite = 1;
	} else
L
Linus Torvalds 已提交
2204 2205 2206 2207 2208 2209 2210 2211 2212
		bip = &sh->dev[dd_idx].toread;
	while (*bip && (*bip)->bi_sector < bi->bi_sector) {
		if ((*bip)->bi_sector + ((*bip)->bi_size >> 9) > bi->bi_sector)
			goto overlap;
		bip = & (*bip)->bi_next;
	}
	if (*bip && (*bip)->bi_sector < bi->bi_sector + ((bi->bi_size)>>9))
		goto overlap;

2213
	BUG_ON(*bip && bi->bi_next && (*bip) != bi->bi_next);
L
Linus Torvalds 已提交
2214 2215 2216
	if (*bip)
		bi->bi_next = *bip;
	*bip = bi;
2217
	bi->bi_phys_segments++;
2218

L
Linus Torvalds 已提交
2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231
	if (forwrite) {
		/* check if page is covered */
		sector_t sector = sh->dev[dd_idx].sector;
		for (bi=sh->dev[dd_idx].towrite;
		     sector < sh->dev[dd_idx].sector + STRIPE_SECTORS &&
			     bi && bi->bi_sector <= sector;
		     bi = r5_next_bio(bi, sh->dev[dd_idx].sector)) {
			if (bi->bi_sector + (bi->bi_size>>9) >= sector)
				sector = bi->bi_sector + (bi->bi_size>>9);
		}
		if (sector >= sh->dev[dd_idx].sector + STRIPE_SECTORS)
			set_bit(R5_OVERWRITE, &sh->dev[dd_idx].flags);
	}
2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243
	spin_unlock_irq(&conf->device_lock);

	pr_debug("added bi b#%llu to stripe s#%llu, disk %d.\n",
		(unsigned long long)(*bip)->bi_sector,
		(unsigned long long)sh->sector, dd_idx);

	if (conf->mddev->bitmap && firstwrite) {
		bitmap_startwrite(conf->mddev->bitmap, sh->sector,
				  STRIPE_SECTORS, 0);
		sh->bm_seq = conf->seq_flush+1;
		set_bit(STRIPE_BIT_DELAY, &sh->state);
	}
L
Linus Torvalds 已提交
2244 2245 2246 2247 2248 2249 2250 2251
	return 1;

 overlap:
	set_bit(R5_Overlap, &sh->dev[dd_idx].flags);
	spin_unlock_irq(&conf->device_lock);
	return 0;
}

2252
static void end_reshape(struct r5conf *conf);
2253

2254
static void stripe_set_idx(sector_t stripe, struct r5conf *conf, int previous,
2255
			    struct stripe_head *sh)
2256
{
2257
	int sectors_per_chunk =
2258
		previous ? conf->prev_chunk_sectors : conf->chunk_sectors;
2259
	int dd_idx;
2260
	int chunk_offset = sector_div(stripe, sectors_per_chunk);
2261
	int disks = previous ? conf->previous_raid_disks : conf->raid_disks;
2262

2263 2264
	raid5_compute_sector(conf,
			     stripe * (disks - conf->max_degraded)
2265
			     *sectors_per_chunk + chunk_offset,
2266
			     previous,
2267
			     &dd_idx, sh);
2268 2269
}

2270
static void
2271
handle_failed_stripe(struct r5conf *conf, struct stripe_head *sh,
2272 2273 2274 2275 2276 2277 2278 2279 2280
				struct stripe_head_state *s, int disks,
				struct bio **return_bi)
{
	int i;
	for (i = disks; i--; ) {
		struct bio *bi;
		int bitmap_end = 0;

		if (test_bit(R5_ReadError, &sh->dev[i].flags)) {
2281
			struct md_rdev *rdev;
2282 2283 2284
			rcu_read_lock();
			rdev = rcu_dereference(conf->disks[i].rdev);
			if (rdev && test_bit(In_sync, &rdev->flags))
2285 2286 2287
				atomic_inc(&rdev->nr_pending);
			else
				rdev = NULL;
2288
			rcu_read_unlock();
2289 2290 2291 2292 2293 2294 2295 2296
			if (rdev) {
				if (!rdev_set_badblocks(
					    rdev,
					    sh->sector,
					    STRIPE_SECTORS, 0))
					md_error(conf->mddev, rdev);
				rdev_dec_pending(rdev, conf->mddev);
			}
2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313
		}
		spin_lock_irq(&conf->device_lock);
		/* fail all writes first */
		bi = sh->dev[i].towrite;
		sh->dev[i].towrite = NULL;
		if (bi) {
			s->to_write--;
			bitmap_end = 1;
		}

		if (test_and_clear_bit(R5_Overlap, &sh->dev[i].flags))
			wake_up(&conf->wait_for_overlap);

		while (bi && bi->bi_sector <
			sh->dev[i].sector + STRIPE_SECTORS) {
			struct bio *nextbi = r5_next_bio(bi, sh->dev[i].sector);
			clear_bit(BIO_UPTODATE, &bi->bi_flags);
2314
			if (!raid5_dec_bi_phys_segments(bi)) {
2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328
				md_write_end(conf->mddev);
				bi->bi_next = *return_bi;
				*return_bi = bi;
			}
			bi = nextbi;
		}
		/* and fail all 'written' */
		bi = sh->dev[i].written;
		sh->dev[i].written = NULL;
		if (bi) bitmap_end = 1;
		while (bi && bi->bi_sector <
		       sh->dev[i].sector + STRIPE_SECTORS) {
			struct bio *bi2 = r5_next_bio(bi, sh->dev[i].sector);
			clear_bit(BIO_UPTODATE, &bi->bi_flags);
2329
			if (!raid5_dec_bi_phys_segments(bi)) {
2330 2331 2332 2333 2334 2335 2336
				md_write_end(conf->mddev);
				bi->bi_next = *return_bi;
				*return_bi = bi;
			}
			bi = bi2;
		}

2337 2338 2339 2340 2341 2342
		/* fail any reads if this device is non-operational and
		 * the data has not reached the cache yet.
		 */
		if (!test_bit(R5_Wantfill, &sh->dev[i].flags) &&
		    (!test_bit(R5_Insync, &sh->dev[i].flags) ||
		      test_bit(R5_ReadError, &sh->dev[i].flags))) {
2343 2344 2345 2346 2347 2348 2349 2350 2351 2352
			bi = sh->dev[i].toread;
			sh->dev[i].toread = NULL;
			if (test_and_clear_bit(R5_Overlap, &sh->dev[i].flags))
				wake_up(&conf->wait_for_overlap);
			if (bi) s->to_read--;
			while (bi && bi->bi_sector <
			       sh->dev[i].sector + STRIPE_SECTORS) {
				struct bio *nextbi =
					r5_next_bio(bi, sh->dev[i].sector);
				clear_bit(BIO_UPTODATE, &bi->bi_flags);
2353
				if (!raid5_dec_bi_phys_segments(bi)) {
2354 2355 2356 2357 2358 2359 2360 2361 2362 2363
					bi->bi_next = *return_bi;
					*return_bi = bi;
				}
				bi = nextbi;
			}
		}
		spin_unlock_irq(&conf->device_lock);
		if (bitmap_end)
			bitmap_endwrite(conf->mddev->bitmap, sh->sector,
					STRIPE_SECTORS, 0, 0);
2364 2365 2366 2367
		/* If we were in the middle of a write the parity block might
		 * still be locked - so just clear all R5_LOCKED flags
		 */
		clear_bit(R5_LOCKED, &sh->dev[i].flags);
2368 2369
	}

2370 2371 2372
	if (test_and_clear_bit(STRIPE_FULL_WRITE, &sh->state))
		if (atomic_dec_and_test(&conf->pending_full_writes))
			md_wakeup_thread(conf->mddev->thread);
2373 2374
}

2375
static void
2376
handle_failed_sync(struct r5conf *conf, struct stripe_head *sh,
2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394
		   struct stripe_head_state *s)
{
	int abort = 0;
	int i;

	md_done_sync(conf->mddev, STRIPE_SECTORS, 0);
	clear_bit(STRIPE_SYNCING, &sh->state);
	s->syncing = 0;
	/* There is nothing more to do for sync/check/repair.
	 * For recover we need to record a bad block on all
	 * non-sync devices, or abort the recovery
	 */
	if (!test_bit(MD_RECOVERY_RECOVER, &conf->mddev->recovery))
		return;
	/* During recovery devices cannot be removed, so locking and
	 * refcounting of rdevs is not needed
	 */
	for (i = 0; i < conf->raid_disks; i++) {
2395
		struct md_rdev *rdev = conf->disks[i].rdev;
2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409
		if (!rdev
		    || test_bit(Faulty, &rdev->flags)
		    || test_bit(In_sync, &rdev->flags))
			continue;
		if (!rdev_set_badblocks(rdev, sh->sector,
					STRIPE_SECTORS, 0))
			abort = 1;
	}
	if (abort) {
		conf->recovery_disabled = conf->mddev->recovery_disabled;
		set_bit(MD_RECOVERY_INTR, &conf->mddev->recovery);
	}
}

2410
/* fetch_block - checks the given member device to see if its data needs
2411 2412 2413
 * to be read or computed to satisfy a request.
 *
 * Returns 1 when no more member devices need to be checked, otherwise returns
2414
 * 0 to tell the loop in handle_stripe_fill to continue
2415
 */
2416 2417
static int fetch_block(struct stripe_head *sh, struct stripe_head_state *s,
		       int disk_idx, int disks)
2418
{
2419
	struct r5dev *dev = &sh->dev[disk_idx];
2420 2421
	struct r5dev *fdev[2] = { &sh->dev[s->failed_num[0]],
				  &sh->dev[s->failed_num[1]] };
2422

2423
	/* is the data in this block needed, and can we get it? */
2424 2425 2426 2427 2428
	if (!test_bit(R5_LOCKED, &dev->flags) &&
	    !test_bit(R5_UPTODATE, &dev->flags) &&
	    (dev->toread ||
	     (dev->towrite && !test_bit(R5_OVERWRITE, &dev->flags)) ||
	     s->syncing || s->expanding ||
2429 2430
	     (s->failed >= 1 && fdev[0]->toread) ||
	     (s->failed >= 2 && fdev[1]->toread) ||
2431 2432 2433
	     (sh->raid_conf->level <= 5 && s->failed && fdev[0]->towrite &&
	      !test_bit(R5_OVERWRITE, &fdev[0]->flags)) ||
	     (sh->raid_conf->level == 6 && s->failed && s->to_write))) {
2434 2435 2436 2437 2438 2439
		/* we would like to get this block, possibly by computing it,
		 * otherwise read it if the backing disk is insync
		 */
		BUG_ON(test_bit(R5_Wantcompute, &dev->flags));
		BUG_ON(test_bit(R5_Wantread, &dev->flags));
		if ((s->uptodate == disks - 1) &&
2440 2441
		    (s->failed && (disk_idx == s->failed_num[0] ||
				   disk_idx == s->failed_num[1]))) {
2442 2443
			/* have disk failed, and we're requested to fetch it;
			 * do compute it
2444
			 */
2445 2446 2447 2448 2449 2450 2451 2452
			pr_debug("Computing stripe %llu block %d\n",
			       (unsigned long long)sh->sector, disk_idx);
			set_bit(STRIPE_COMPUTE_RUN, &sh->state);
			set_bit(STRIPE_OP_COMPUTE_BLK, &s->ops_request);
			set_bit(R5_Wantcompute, &dev->flags);
			sh->ops.target = disk_idx;
			sh->ops.target2 = -1; /* no 2nd target */
			s->req_compute = 1;
2453 2454 2455 2456 2457 2458
			/* Careful: from this point on 'uptodate' is in the eye
			 * of raid_run_ops which services 'compute' operations
			 * before writes. R5_Wantcompute flags a block that will
			 * be R5_UPTODATE by the time it is needed for a
			 * subsequent operation.
			 */
2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471
			s->uptodate++;
			return 1;
		} else if (s->uptodate == disks-2 && s->failed >= 2) {
			/* Computing 2-failure is *very* expensive; only
			 * do it if failed >= 2
			 */
			int other;
			for (other = disks; other--; ) {
				if (other == disk_idx)
					continue;
				if (!test_bit(R5_UPTODATE,
				      &sh->dev[other].flags))
					break;
2472
			}
2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491
			BUG_ON(other < 0);
			pr_debug("Computing stripe %llu blocks %d,%d\n",
			       (unsigned long long)sh->sector,
			       disk_idx, other);
			set_bit(STRIPE_COMPUTE_RUN, &sh->state);
			set_bit(STRIPE_OP_COMPUTE_BLK, &s->ops_request);
			set_bit(R5_Wantcompute, &sh->dev[disk_idx].flags);
			set_bit(R5_Wantcompute, &sh->dev[other].flags);
			sh->ops.target = disk_idx;
			sh->ops.target2 = other;
			s->uptodate += 2;
			s->req_compute = 1;
			return 1;
		} else if (test_bit(R5_Insync, &dev->flags)) {
			set_bit(R5_LOCKED, &dev->flags);
			set_bit(R5_Wantread, &dev->flags);
			s->locked++;
			pr_debug("Reading block %d (sync=%d)\n",
				disk_idx, s->syncing);
2492 2493
		}
	}
2494 2495 2496 2497 2498

	return 0;
}

/**
2499
 * handle_stripe_fill - read or compute data to satisfy pending requests.
2500
 */
2501 2502 2503
static void handle_stripe_fill(struct stripe_head *sh,
			       struct stripe_head_state *s,
			       int disks)
2504 2505 2506 2507 2508 2509 2510 2511 2512 2513
{
	int i;

	/* look for blocks to read/compute, skip this if a compute
	 * is already in flight, or if the stripe contents are in the
	 * midst of changing due to a write
	 */
	if (!test_bit(STRIPE_COMPUTE_RUN, &sh->state) && !sh->check_state &&
	    !sh->reconstruct_state)
		for (i = disks; i--; )
2514
			if (fetch_block(sh, s, i, disks))
2515
				break;
2516 2517 2518 2519
	set_bit(STRIPE_HANDLE, &sh->state);
}


2520
/* handle_stripe_clean_event
2521 2522 2523 2524
 * any written block on an uptodate or failed drive can be returned.
 * Note that if we 'wrote' to a failed drive, it will be UPTODATE, but
 * never LOCKED, so we don't need to test 'failed' directly.
 */
2525
static void handle_stripe_clean_event(struct r5conf *conf,
2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538
	struct stripe_head *sh, int disks, struct bio **return_bi)
{
	int i;
	struct r5dev *dev;

	for (i = disks; i--; )
		if (sh->dev[i].written) {
			dev = &sh->dev[i];
			if (!test_bit(R5_LOCKED, &dev->flags) &&
				test_bit(R5_UPTODATE, &dev->flags)) {
				/* We can return any write requests */
				struct bio *wbi, *wbi2;
				int bitmap_end = 0;
2539
				pr_debug("Return write for disc %d\n", i);
2540 2541 2542 2543 2544 2545
				spin_lock_irq(&conf->device_lock);
				wbi = dev->written;
				dev->written = NULL;
				while (wbi && wbi->bi_sector <
					dev->sector + STRIPE_SECTORS) {
					wbi2 = r5_next_bio(wbi, dev->sector);
2546
					if (!raid5_dec_bi_phys_segments(wbi)) {
2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563
						md_write_end(conf->mddev);
						wbi->bi_next = *return_bi;
						*return_bi = wbi;
					}
					wbi = wbi2;
				}
				if (dev->towrite == NULL)
					bitmap_end = 1;
				spin_unlock_irq(&conf->device_lock);
				if (bitmap_end)
					bitmap_endwrite(conf->mddev->bitmap,
							sh->sector,
							STRIPE_SECTORS,
					 !test_bit(STRIPE_DEGRADED, &sh->state),
							0);
			}
		}
2564 2565 2566 2567

	if (test_and_clear_bit(STRIPE_FULL_WRITE, &sh->state))
		if (atomic_dec_and_test(&conf->pending_full_writes))
			md_wakeup_thread(conf->mddev->thread);
2568 2569
}

2570
static void handle_stripe_dirtying(struct r5conf *conf,
2571 2572 2573
				   struct stripe_head *sh,
				   struct stripe_head_state *s,
				   int disks)
2574 2575
{
	int rmw = 0, rcw = 0, i;
2576 2577 2578 2579 2580 2581 2582
	if (conf->max_degraded == 2) {
		/* RAID6 requires 'rcw' in current implementation
		 * Calculate the real rcw later - for now fake it
		 * look like rcw is cheaper
		 */
		rcw = 1; rmw = 2;
	} else for (i = disks; i--; ) {
2583 2584 2585 2586
		/* would I have to read this buffer for read_modify_write */
		struct r5dev *dev = &sh->dev[i];
		if ((dev->towrite || i == sh->pd_idx) &&
		    !test_bit(R5_LOCKED, &dev->flags) &&
2587 2588
		    !(test_bit(R5_UPTODATE, &dev->flags) ||
		      test_bit(R5_Wantcompute, &dev->flags))) {
2589 2590 2591 2592 2593 2594 2595 2596
			if (test_bit(R5_Insync, &dev->flags))
				rmw++;
			else
				rmw += 2*disks;  /* cannot read it */
		}
		/* Would I have to read this buffer for reconstruct_write */
		if (!test_bit(R5_OVERWRITE, &dev->flags) && i != sh->pd_idx &&
		    !test_bit(R5_LOCKED, &dev->flags) &&
2597 2598 2599
		    !(test_bit(R5_UPTODATE, &dev->flags) ||
		    test_bit(R5_Wantcompute, &dev->flags))) {
			if (test_bit(R5_Insync, &dev->flags)) rcw++;
2600 2601 2602 2603
			else
				rcw += 2*disks;
		}
	}
2604
	pr_debug("for sector %llu, rmw=%d rcw=%d\n",
2605 2606 2607 2608 2609 2610 2611 2612
		(unsigned long long)sh->sector, rmw, rcw);
	set_bit(STRIPE_HANDLE, &sh->state);
	if (rmw < rcw && rmw > 0)
		/* prefer read-modify-write, but need to get some data */
		for (i = disks; i--; ) {
			struct r5dev *dev = &sh->dev[i];
			if ((dev->towrite || i == sh->pd_idx) &&
			    !test_bit(R5_LOCKED, &dev->flags) &&
2613 2614
			    !(test_bit(R5_UPTODATE, &dev->flags) ||
			    test_bit(R5_Wantcompute, &dev->flags)) &&
2615 2616 2617
			    test_bit(R5_Insync, &dev->flags)) {
				if (
				  test_bit(STRIPE_PREREAD_ACTIVE, &sh->state)) {
2618
					pr_debug("Read_old block "
2619 2620 2621 2622 2623 2624 2625 2626 2627 2628
						"%d for r-m-w\n", i);
					set_bit(R5_LOCKED, &dev->flags);
					set_bit(R5_Wantread, &dev->flags);
					s->locked++;
				} else {
					set_bit(STRIPE_DELAYED, &sh->state);
					set_bit(STRIPE_HANDLE, &sh->state);
				}
			}
		}
2629
	if (rcw <= rmw && rcw > 0) {
2630
		/* want reconstruct write, but need to get some data */
2631
		rcw = 0;
2632 2633 2634
		for (i = disks; i--; ) {
			struct r5dev *dev = &sh->dev[i];
			if (!test_bit(R5_OVERWRITE, &dev->flags) &&
2635
			    i != sh->pd_idx && i != sh->qd_idx &&
2636
			    !test_bit(R5_LOCKED, &dev->flags) &&
2637
			    !(test_bit(R5_UPTODATE, &dev->flags) ||
2638 2639 2640 2641
			      test_bit(R5_Wantcompute, &dev->flags))) {
				rcw++;
				if (!test_bit(R5_Insync, &dev->flags))
					continue; /* it's a failed drive */
2642 2643
				if (
				  test_bit(STRIPE_PREREAD_ACTIVE, &sh->state)) {
2644
					pr_debug("Read_old block "
2645 2646 2647 2648 2649 2650 2651 2652 2653 2654
						"%d for Reconstruct\n", i);
					set_bit(R5_LOCKED, &dev->flags);
					set_bit(R5_Wantread, &dev->flags);
					s->locked++;
				} else {
					set_bit(STRIPE_DELAYED, &sh->state);
					set_bit(STRIPE_HANDLE, &sh->state);
				}
			}
		}
2655
	}
2656 2657 2658
	/* now if nothing is locked, and if we have enough data,
	 * we can start a write request
	 */
2659 2660
	/* since handle_stripe can be called at any time we need to handle the
	 * case where a compute block operation has been submitted and then a
2661 2662
	 * subsequent call wants to start a write request.  raid_run_ops only
	 * handles the case where compute block and reconstruct are requested
2663 2664 2665
	 * simultaneously.  If this is not the case then new writes need to be
	 * held off until the compute completes.
	 */
2666 2667 2668
	if ((s->req_compute || !test_bit(STRIPE_COMPUTE_RUN, &sh->state)) &&
	    (s->locked == 0 && (rcw == 0 || rmw == 0) &&
	    !test_bit(STRIPE_BIT_DELAY, &sh->state)))
2669
		schedule_reconstruction(sh, s, rcw == 0, 0);
2670 2671
}

2672
static void handle_parity_checks5(struct r5conf *conf, struct stripe_head *sh,
2673 2674
				struct stripe_head_state *s, int disks)
{
2675
	struct r5dev *dev = NULL;
2676

2677
	set_bit(STRIPE_HANDLE, &sh->state);
2678

2679 2680 2681
	switch (sh->check_state) {
	case check_state_idle:
		/* start a new check operation if there are no failures */
2682 2683
		if (s->failed == 0) {
			BUG_ON(s->uptodate != disks);
2684 2685
			sh->check_state = check_state_run;
			set_bit(STRIPE_OP_CHECK, &s->ops_request);
2686 2687
			clear_bit(R5_UPTODATE, &sh->dev[sh->pd_idx].flags);
			s->uptodate--;
2688
			break;
2689
		}
2690
		dev = &sh->dev[s->failed_num[0]];
2691 2692 2693 2694 2695 2696 2697 2698 2699
		/* fall through */
	case check_state_compute_result:
		sh->check_state = check_state_idle;
		if (!dev)
			dev = &sh->dev[sh->pd_idx];

		/* check that a write has not made the stripe insync */
		if (test_bit(STRIPE_INSYNC, &sh->state))
			break;
D
Dan Williams 已提交
2700

2701 2702 2703 2704 2705
		/* either failed parity check, or recovery is happening */
		BUG_ON(!test_bit(R5_UPTODATE, &dev->flags));
		BUG_ON(s->uptodate != disks);

		set_bit(R5_LOCKED, &dev->flags);
2706
		s->locked++;
2707
		set_bit(R5_Wantwrite, &dev->flags);
2708

2709 2710
		clear_bit(STRIPE_DEGRADED, &sh->state);
		set_bit(STRIPE_INSYNC, &sh->state);
2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726
		break;
	case check_state_run:
		break; /* we will be called again upon completion */
	case check_state_check_result:
		sh->check_state = check_state_idle;

		/* if a failure occurred during the check operation, leave
		 * STRIPE_INSYNC not set and let the stripe be handled again
		 */
		if (s->failed)
			break;

		/* handle a successful check operation, if parity is correct
		 * we are done.  Otherwise update the mismatch count and repair
		 * parity if !MD_RECOVERY_CHECK
		 */
D
Dan Williams 已提交
2727
		if ((sh->ops.zero_sum_result & SUM_CHECK_P_RESULT) == 0)
2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738
			/* parity is correct (on disc,
			 * not in buffer any more)
			 */
			set_bit(STRIPE_INSYNC, &sh->state);
		else {
			conf->mddev->resync_mismatches += STRIPE_SECTORS;
			if (test_bit(MD_RECOVERY_CHECK, &conf->mddev->recovery))
				/* don't try to repair!! */
				set_bit(STRIPE_INSYNC, &sh->state);
			else {
				sh->check_state = check_state_compute_run;
2739
				set_bit(STRIPE_COMPUTE_RUN, &sh->state);
2740 2741 2742 2743
				set_bit(STRIPE_OP_COMPUTE_BLK, &s->ops_request);
				set_bit(R5_Wantcompute,
					&sh->dev[sh->pd_idx].flags);
				sh->ops.target = sh->pd_idx;
2744
				sh->ops.target2 = -1;
2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755
				s->uptodate++;
			}
		}
		break;
	case check_state_compute_run:
		break;
	default:
		printk(KERN_ERR "%s: unknown check_state: %d sector: %llu\n",
		       __func__, sh->check_state,
		       (unsigned long long) sh->sector);
		BUG();
2756 2757 2758 2759
	}
}


2760
static void handle_parity_checks6(struct r5conf *conf, struct stripe_head *sh,
2761
				  struct stripe_head_state *s,
2762
				  int disks)
2763 2764
{
	int pd_idx = sh->pd_idx;
N
NeilBrown 已提交
2765
	int qd_idx = sh->qd_idx;
2766
	struct r5dev *dev;
2767 2768 2769 2770

	set_bit(STRIPE_HANDLE, &sh->state);

	BUG_ON(s->failed > 2);
2771

2772 2773 2774 2775 2776 2777
	/* Want to check and possibly repair P and Q.
	 * However there could be one 'failed' device, in which
	 * case we can only check one of them, possibly using the
	 * other to generate missing data
	 */

2778 2779 2780
	switch (sh->check_state) {
	case check_state_idle:
		/* start a new check operation if there are < 2 failures */
2781
		if (s->failed == s->q_failed) {
2782
			/* The only possible failed device holds Q, so it
2783 2784 2785
			 * makes sense to check P (If anything else were failed,
			 * we would have used P to recreate it).
			 */
2786
			sh->check_state = check_state_run;
2787
		}
2788
		if (!s->q_failed && s->failed < 2) {
2789
			/* Q is not failed, and we didn't use it to generate
2790 2791
			 * anything, so it makes sense to check it
			 */
2792 2793 2794 2795
			if (sh->check_state == check_state_run)
				sh->check_state = check_state_run_pq;
			else
				sh->check_state = check_state_run_q;
2796 2797
		}

2798 2799
		/* discard potentially stale zero_sum_result */
		sh->ops.zero_sum_result = 0;
2800

2801 2802 2803 2804
		if (sh->check_state == check_state_run) {
			/* async_xor_zero_sum destroys the contents of P */
			clear_bit(R5_UPTODATE, &sh->dev[pd_idx].flags);
			s->uptodate--;
2805
		}
2806 2807 2808 2809 2810 2811 2812
		if (sh->check_state >= check_state_run &&
		    sh->check_state <= check_state_run_pq) {
			/* async_syndrome_zero_sum preserves P and Q, so
			 * no need to mark them !uptodate here
			 */
			set_bit(STRIPE_OP_CHECK, &s->ops_request);
			break;
2813 2814
		}

2815 2816 2817 2818 2819
		/* we have 2-disk failure */
		BUG_ON(s->failed != 2);
		/* fall through */
	case check_state_compute_result:
		sh->check_state = check_state_idle;
2820

2821 2822 2823
		/* check that a write has not made the stripe insync */
		if (test_bit(STRIPE_INSYNC, &sh->state))
			break;
2824 2825

		/* now write out any block on a failed drive,
2826
		 * or P or Q if they were recomputed
2827
		 */
2828
		BUG_ON(s->uptodate < disks - 1); /* We don't need Q to recover */
2829
		if (s->failed == 2) {
2830
			dev = &sh->dev[s->failed_num[1]];
2831 2832 2833 2834 2835
			s->locked++;
			set_bit(R5_LOCKED, &dev->flags);
			set_bit(R5_Wantwrite, &dev->flags);
		}
		if (s->failed >= 1) {
2836
			dev = &sh->dev[s->failed_num[0]];
2837 2838 2839 2840
			s->locked++;
			set_bit(R5_LOCKED, &dev->flags);
			set_bit(R5_Wantwrite, &dev->flags);
		}
2841
		if (sh->ops.zero_sum_result & SUM_CHECK_P_RESULT) {
2842 2843 2844 2845 2846
			dev = &sh->dev[pd_idx];
			s->locked++;
			set_bit(R5_LOCKED, &dev->flags);
			set_bit(R5_Wantwrite, &dev->flags);
		}
2847
		if (sh->ops.zero_sum_result & SUM_CHECK_Q_RESULT) {
2848 2849 2850 2851 2852 2853 2854 2855
			dev = &sh->dev[qd_idx];
			s->locked++;
			set_bit(R5_LOCKED, &dev->flags);
			set_bit(R5_Wantwrite, &dev->flags);
		}
		clear_bit(STRIPE_DEGRADED, &sh->state);

		set_bit(STRIPE_INSYNC, &sh->state);
2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919
		break;
	case check_state_run:
	case check_state_run_q:
	case check_state_run_pq:
		break; /* we will be called again upon completion */
	case check_state_check_result:
		sh->check_state = check_state_idle;

		/* handle a successful check operation, if parity is correct
		 * we are done.  Otherwise update the mismatch count and repair
		 * parity if !MD_RECOVERY_CHECK
		 */
		if (sh->ops.zero_sum_result == 0) {
			/* both parities are correct */
			if (!s->failed)
				set_bit(STRIPE_INSYNC, &sh->state);
			else {
				/* in contrast to the raid5 case we can validate
				 * parity, but still have a failure to write
				 * back
				 */
				sh->check_state = check_state_compute_result;
				/* Returning at this point means that we may go
				 * off and bring p and/or q uptodate again so
				 * we make sure to check zero_sum_result again
				 * to verify if p or q need writeback
				 */
			}
		} else {
			conf->mddev->resync_mismatches += STRIPE_SECTORS;
			if (test_bit(MD_RECOVERY_CHECK, &conf->mddev->recovery))
				/* don't try to repair!! */
				set_bit(STRIPE_INSYNC, &sh->state);
			else {
				int *target = &sh->ops.target;

				sh->ops.target = -1;
				sh->ops.target2 = -1;
				sh->check_state = check_state_compute_run;
				set_bit(STRIPE_COMPUTE_RUN, &sh->state);
				set_bit(STRIPE_OP_COMPUTE_BLK, &s->ops_request);
				if (sh->ops.zero_sum_result & SUM_CHECK_P_RESULT) {
					set_bit(R5_Wantcompute,
						&sh->dev[pd_idx].flags);
					*target = pd_idx;
					target = &sh->ops.target2;
					s->uptodate++;
				}
				if (sh->ops.zero_sum_result & SUM_CHECK_Q_RESULT) {
					set_bit(R5_Wantcompute,
						&sh->dev[qd_idx].flags);
					*target = qd_idx;
					s->uptodate++;
				}
			}
		}
		break;
	case check_state_compute_run:
		break;
	default:
		printk(KERN_ERR "%s: unknown check_state: %d sector: %llu\n",
		       __func__, sh->check_state,
		       (unsigned long long) sh->sector);
		BUG();
2920 2921 2922
	}
}

2923
static void handle_stripe_expansion(struct r5conf *conf, struct stripe_head *sh)
2924 2925 2926 2927 2928 2929
{
	int i;

	/* We have read all the blocks in this stripe and now we need to
	 * copy some of them into a target stripe for expand.
	 */
2930
	struct dma_async_tx_descriptor *tx = NULL;
2931 2932
	clear_bit(STRIPE_EXPAND_SOURCE, &sh->state);
	for (i = 0; i < sh->disks; i++)
N
NeilBrown 已提交
2933
		if (i != sh->pd_idx && i != sh->qd_idx) {
2934
			int dd_idx, j;
2935
			struct stripe_head *sh2;
2936
			struct async_submit_ctl submit;
2937

2938
			sector_t bn = compute_blocknr(sh, i, 1);
2939 2940
			sector_t s = raid5_compute_sector(conf, bn, 0,
							  &dd_idx, NULL);
2941
			sh2 = get_active_stripe(conf, s, 0, 1, 1);
2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953
			if (sh2 == NULL)
				/* so far only the early blocks of this stripe
				 * have been requested.  When later blocks
				 * get requested, we will try again
				 */
				continue;
			if (!test_bit(STRIPE_EXPANDING, &sh2->state) ||
			   test_bit(R5_Expanded, &sh2->dev[dd_idx].flags)) {
				/* must have already done this block */
				release_stripe(sh2);
				continue;
			}
2954 2955

			/* place all the copies on one channel */
2956
			init_async_submit(&submit, 0, tx, NULL, NULL, NULL);
2957
			tx = async_memcpy(sh2->dev[dd_idx].page,
2958
					  sh->dev[i].page, 0, 0, STRIPE_SIZE,
2959
					  &submit);
2960

2961 2962 2963 2964
			set_bit(R5_Expanded, &sh2->dev[dd_idx].flags);
			set_bit(R5_UPTODATE, &sh2->dev[dd_idx].flags);
			for (j = 0; j < conf->raid_disks; j++)
				if (j != sh2->pd_idx &&
2965
				    j != sh2->qd_idx &&
2966 2967 2968 2969 2970 2971 2972
				    !test_bit(R5_Expanded, &sh2->dev[j].flags))
					break;
			if (j == conf->raid_disks) {
				set_bit(STRIPE_EXPAND_READY, &sh2->state);
				set_bit(STRIPE_HANDLE, &sh2->state);
			}
			release_stripe(sh2);
2973

2974
		}
2975 2976 2977 2978 2979
	/* done submitting copies, wait for them to complete */
	if (tx) {
		async_tx_ack(tx);
		dma_wait_for_async_tx(tx);
	}
2980
}
L
Linus Torvalds 已提交
2981

2982

L
Linus Torvalds 已提交
2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998
/*
 * handle_stripe - do things to a stripe.
 *
 * We lock the stripe and then examine the state of various bits
 * to see what needs to be done.
 * Possible results:
 *    return some read request which now have data
 *    return some write requests which are safely on disc
 *    schedule a read on some buffers
 *    schedule a write of some buffers
 *    return confirmation of parity correctness
 *
 * buffers are taken off read_list or write_list, and bh_cache buffers
 * get BH_Lock set before the stripe lock is released.
 *
 */
2999

3000
static void analyse_stripe(struct stripe_head *sh, struct stripe_head_state *s)
L
Linus Torvalds 已提交
3001
{
3002
	struct r5conf *conf = sh->raid_conf;
3003
	int disks = sh->disks;
3004 3005
	struct r5dev *dev;
	int i;
L
Linus Torvalds 已提交
3006

3007 3008 3009 3010 3011 3012 3013
	memset(s, 0, sizeof(*s));

	s->syncing = test_bit(STRIPE_SYNCING, &sh->state);
	s->expanding = test_bit(STRIPE_EXPAND_SOURCE, &sh->state);
	s->expanded = test_bit(STRIPE_EXPAND_READY, &sh->state);
	s->failed_num[0] = -1;
	s->failed_num[1] = -1;
L
Linus Torvalds 已提交
3014

3015
	/* Now to look around and see what can be done */
L
Linus Torvalds 已提交
3016
	rcu_read_lock();
3017
	spin_lock_irq(&conf->device_lock);
3018
	for (i=disks; i--; ) {
3019
		struct md_rdev *rdev;
3020 3021 3022
		sector_t first_bad;
		int bad_sectors;
		int is_bad = 0;
3023

3024
		dev = &sh->dev[i];
L
Linus Torvalds 已提交
3025

3026
		pr_debug("check %d: state 0x%lx read %p write %p written %p\n",
3027
			i, dev->flags, dev->toread, dev->towrite, dev->written);
3028 3029 3030 3031 3032 3033 3034 3035
		/* maybe we can reply to a read
		 *
		 * new wantfill requests are only permitted while
		 * ops_complete_biofill is guaranteed to be inactive
		 */
		if (test_bit(R5_UPTODATE, &dev->flags) && dev->toread &&
		    !test_bit(STRIPE_BIOFILL_RUN, &sh->state))
			set_bit(R5_Wantfill, &dev->flags);
L
Linus Torvalds 已提交
3036

3037
		/* now count some things */
3038 3039 3040 3041
		if (test_bit(R5_LOCKED, &dev->flags))
			s->locked++;
		if (test_bit(R5_UPTODATE, &dev->flags))
			s->uptodate++;
3042
		if (test_bit(R5_Wantcompute, &dev->flags)) {
3043 3044
			s->compute++;
			BUG_ON(s->compute > 2);
3045
		}
L
Linus Torvalds 已提交
3046

3047
		if (test_bit(R5_Wantfill, &dev->flags))
3048
			s->to_fill++;
3049
		else if (dev->toread)
3050
			s->to_read++;
3051
		if (dev->towrite) {
3052
			s->to_write++;
3053
			if (!test_bit(R5_OVERWRITE, &dev->flags))
3054
				s->non_overwrite++;
3055
		}
3056
		if (dev->written)
3057
			s->written++;
3058 3059 3060 3061 3062 3063 3064 3065 3066 3067 3068 3069 3070
		/* Prefer to use the replacement for reads, but only
		 * if it is recovered enough and has no bad blocks.
		 */
		rdev = rcu_dereference(conf->disks[i].replacement);
		if (rdev && !test_bit(Faulty, &rdev->flags) &&
		    rdev->recovery_offset >= sh->sector + STRIPE_SECTORS &&
		    !is_badblock(rdev, sh->sector, STRIPE_SECTORS,
				 &first_bad, &bad_sectors))
			set_bit(R5_ReadRepl, &dev->flags);
		else {
			rdev = rcu_dereference(conf->disks[i].rdev);
			clear_bit(R5_ReadRepl, &dev->flags);
		}
3071 3072
		if (rdev && test_bit(Faulty, &rdev->flags))
			rdev = NULL;
3073 3074 3075 3076 3077 3078 3079 3080 3081 3082 3083 3084
		if (rdev) {
			is_bad = is_badblock(rdev, sh->sector, STRIPE_SECTORS,
					     &first_bad, &bad_sectors);
			if (s->blocked_rdev == NULL
			    && (test_bit(Blocked, &rdev->flags)
				|| is_bad < 0)) {
				if (is_bad < 0)
					set_bit(BlockedBadBlocks,
						&rdev->flags);
				s->blocked_rdev = rdev;
				atomic_inc(&rdev->nr_pending);
			}
3085
		}
3086 3087 3088
		clear_bit(R5_Insync, &dev->flags);
		if (!rdev)
			/* Not in-sync */;
3089 3090 3091 3092 3093 3094 3095 3096 3097 3098
		else if (is_bad) {
			/* also not in-sync */
			if (!test_bit(WriteErrorSeen, &rdev->flags)) {
				/* treat as in-sync, but with a read error
				 * which we can now try to correct
				 */
				set_bit(R5_Insync, &dev->flags);
				set_bit(R5_ReadError, &dev->flags);
			}
		} else if (test_bit(In_sync, &rdev->flags))
3099
			set_bit(R5_Insync, &dev->flags);
3100
		else if (sh->sector + STRIPE_SECTORS <= rdev->recovery_offset)
3101
			/* in sync if before recovery_offset */
3102 3103 3104 3105 3106 3107 3108 3109 3110
			set_bit(R5_Insync, &dev->flags);
		else if (test_bit(R5_UPTODATE, &dev->flags) &&
			 test_bit(R5_Expanded, &dev->flags))
			/* If we've reshaped into here, we assume it is Insync.
			 * We will shortly update recovery_offset to make
			 * it official.
			 */
			set_bit(R5_Insync, &dev->flags);

A
Adam Kwolek 已提交
3111
		if (rdev && test_bit(R5_WriteError, &dev->flags)) {
3112 3113 3114 3115 3116 3117 3118
			/* This flag does not apply to '.replacement'
			 * only to .rdev, so make sure to check that*/
			struct md_rdev *rdev2 = rcu_dereference(
				conf->disks[i].rdev);
			if (rdev2 == rdev)
				clear_bit(R5_Insync, &dev->flags);
			if (rdev2 && !test_bit(Faulty, &rdev2->flags)) {
3119
				s->handle_bad_blocks = 1;
3120
				atomic_inc(&rdev2->nr_pending);
3121 3122 3123
			} else
				clear_bit(R5_WriteError, &dev->flags);
		}
A
Adam Kwolek 已提交
3124
		if (rdev && test_bit(R5_MadeGood, &dev->flags)) {
3125 3126 3127 3128 3129
			/* This flag does not apply to '.replacement'
			 * only to .rdev, so make sure to check that*/
			struct md_rdev *rdev2 = rcu_dereference(
				conf->disks[i].rdev);
			if (rdev2 && !test_bit(Faulty, &rdev2->flags)) {
3130
				s->handle_bad_blocks = 1;
3131
				atomic_inc(&rdev2->nr_pending);
3132 3133 3134
			} else
				clear_bit(R5_MadeGood, &dev->flags);
		}
3135
		if (!test_bit(R5_Insync, &dev->flags)) {
3136 3137 3138
			/* The ReadError flag will just be confusing now */
			clear_bit(R5_ReadError, &dev->flags);
			clear_bit(R5_ReWrite, &dev->flags);
L
Linus Torvalds 已提交
3139
		}
3140 3141 3142
		if (test_bit(R5_ReadError, &dev->flags))
			clear_bit(R5_Insync, &dev->flags);
		if (!test_bit(R5_Insync, &dev->flags)) {
3143 3144 3145
			if (s->failed < 2)
				s->failed_num[s->failed] = i;
			s->failed++;
3146
		}
L
Linus Torvalds 已提交
3147
	}
3148
	spin_unlock_irq(&conf->device_lock);
L
Linus Torvalds 已提交
3149
	rcu_read_unlock();
3150 3151 3152 3153 3154
}

static void handle_stripe(struct stripe_head *sh)
{
	struct stripe_head_state s;
3155
	struct r5conf *conf = sh->raid_conf;
3156
	int i;
3157 3158
	int prexor;
	int disks = sh->disks;
3159
	struct r5dev *pdev, *qdev;
3160 3161

	clear_bit(STRIPE_HANDLE, &sh->state);
3162
	if (test_and_set_bit_lock(STRIPE_ACTIVE, &sh->state)) {
3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179
		/* already being handled, ensure it gets handled
		 * again when current action finishes */
		set_bit(STRIPE_HANDLE, &sh->state);
		return;
	}

	if (test_and_clear_bit(STRIPE_SYNC_REQUESTED, &sh->state)) {
		set_bit(STRIPE_SYNCING, &sh->state);
		clear_bit(STRIPE_INSYNC, &sh->state);
	}
	clear_bit(STRIPE_DELAYED, &sh->state);

	pr_debug("handling stripe %llu, state=%#lx cnt=%d, "
		"pd_idx=%d, qd_idx=%d\n, check:%d, reconstruct:%d\n",
	       (unsigned long long)sh->sector, sh->state,
	       atomic_read(&sh->count), sh->pd_idx, sh->qd_idx,
	       sh->check_state, sh->reconstruct_state);
3180

3181
	analyse_stripe(sh, &s);
3182

3183 3184 3185 3186 3187
	if (s.handle_bad_blocks) {
		set_bit(STRIPE_HANDLE, &sh->state);
		goto finish;
	}

3188 3189 3190 3191 3192 3193 3194 3195 3196 3197 3198 3199 3200 3201 3202 3203 3204 3205 3206 3207 3208 3209 3210
	if (unlikely(s.blocked_rdev)) {
		if (s.syncing || s.expanding || s.expanded ||
		    s.to_write || s.written) {
			set_bit(STRIPE_HANDLE, &sh->state);
			goto finish;
		}
		/* There is nothing for the blocked_rdev to block */
		rdev_dec_pending(s.blocked_rdev, conf->mddev);
		s.blocked_rdev = NULL;
	}

	if (s.to_fill && !test_bit(STRIPE_BIOFILL_RUN, &sh->state)) {
		set_bit(STRIPE_OP_BIOFILL, &s.ops_request);
		set_bit(STRIPE_BIOFILL_RUN, &sh->state);
	}

	pr_debug("locked=%d uptodate=%d to_read=%d"
	       " to_write=%d failed=%d failed_num=%d,%d\n",
	       s.locked, s.uptodate, s.to_read, s.to_write, s.failed,
	       s.failed_num[0], s.failed_num[1]);
	/* check if the array has lost more than max_degraded devices and,
	 * if so, some requests might need to be failed.
	 */
3211 3212 3213 3214 3215 3216 3217 3218
	if (s.failed > conf->max_degraded) {
		sh->check_state = 0;
		sh->reconstruct_state = 0;
		if (s.to_read+s.to_write+s.written)
			handle_failed_stripe(conf, sh, &s, disks, &s.return_bi);
		if (s.syncing)
			handle_failed_sync(conf, sh, &s);
	}
3219 3220 3221 3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232 3233 3234 3235 3236 3237 3238 3239 3240 3241 3242 3243 3244 3245 3246 3247 3248 3249

	/*
	 * might be able to return some write requests if the parity blocks
	 * are safe, or on a failed drive
	 */
	pdev = &sh->dev[sh->pd_idx];
	s.p_failed = (s.failed >= 1 && s.failed_num[0] == sh->pd_idx)
		|| (s.failed >= 2 && s.failed_num[1] == sh->pd_idx);
	qdev = &sh->dev[sh->qd_idx];
	s.q_failed = (s.failed >= 1 && s.failed_num[0] == sh->qd_idx)
		|| (s.failed >= 2 && s.failed_num[1] == sh->qd_idx)
		|| conf->level < 6;

	if (s.written &&
	    (s.p_failed || ((test_bit(R5_Insync, &pdev->flags)
			     && !test_bit(R5_LOCKED, &pdev->flags)
			     && test_bit(R5_UPTODATE, &pdev->flags)))) &&
	    (s.q_failed || ((test_bit(R5_Insync, &qdev->flags)
			     && !test_bit(R5_LOCKED, &qdev->flags)
			     && test_bit(R5_UPTODATE, &qdev->flags)))))
		handle_stripe_clean_event(conf, sh, disks, &s.return_bi);

	/* Now we might consider reading some blocks, either to check/generate
	 * parity, or to satisfy requests
	 * or to load a block that is being partially written.
	 */
	if (s.to_read || s.non_overwrite
	    || (conf->level == 6 && s.to_write && s.failed)
	    || (s.syncing && (s.uptodate + s.compute < disks)) || s.expanding)
		handle_stripe_fill(sh, &s, disks);

3250 3251 3252 3253 3254 3255 3256 3257 3258 3259 3260 3261 3262 3263 3264 3265 3266 3267 3268 3269 3270 3271 3272 3273 3274 3275 3276 3277 3278 3279 3280 3281 3282 3283 3284 3285 3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296 3297 3298 3299 3300 3301 3302 3303 3304 3305 3306 3307
	/* Now we check to see if any write operations have recently
	 * completed
	 */
	prexor = 0;
	if (sh->reconstruct_state == reconstruct_state_prexor_drain_result)
		prexor = 1;
	if (sh->reconstruct_state == reconstruct_state_drain_result ||
	    sh->reconstruct_state == reconstruct_state_prexor_drain_result) {
		sh->reconstruct_state = reconstruct_state_idle;

		/* All the 'written' buffers and the parity block are ready to
		 * be written back to disk
		 */
		BUG_ON(!test_bit(R5_UPTODATE, &sh->dev[sh->pd_idx].flags));
		BUG_ON(sh->qd_idx >= 0 &&
		       !test_bit(R5_UPTODATE, &sh->dev[sh->qd_idx].flags));
		for (i = disks; i--; ) {
			struct r5dev *dev = &sh->dev[i];
			if (test_bit(R5_LOCKED, &dev->flags) &&
				(i == sh->pd_idx || i == sh->qd_idx ||
				 dev->written)) {
				pr_debug("Writing block %d\n", i);
				set_bit(R5_Wantwrite, &dev->flags);
				if (prexor)
					continue;
				if (!test_bit(R5_Insync, &dev->flags) ||
				    ((i == sh->pd_idx || i == sh->qd_idx)  &&
				     s.failed == 0))
					set_bit(STRIPE_INSYNC, &sh->state);
			}
		}
		if (test_and_clear_bit(STRIPE_PREREAD_ACTIVE, &sh->state))
			s.dec_preread_active = 1;
	}

	/* Now to consider new write requests and what else, if anything
	 * should be read.  We do not handle new writes when:
	 * 1/ A 'write' operation (copy+xor) is already in flight.
	 * 2/ A 'check' operation is in flight, as it may clobber the parity
	 *    block.
	 */
	if (s.to_write && !sh->reconstruct_state && !sh->check_state)
		handle_stripe_dirtying(conf, sh, &s, disks);

	/* maybe we need to check and possibly fix the parity for this stripe
	 * Any reads will already have been scheduled, so we just see if enough
	 * data is available.  The parity check is held off while parity
	 * dependent operations are in flight.
	 */
	if (sh->check_state ||
	    (s.syncing && s.locked == 0 &&
	     !test_bit(STRIPE_COMPUTE_RUN, &sh->state) &&
	     !test_bit(STRIPE_INSYNC, &sh->state))) {
		if (conf->level == 6)
			handle_parity_checks6(conf, sh, &s, disks);
		else
			handle_parity_checks5(conf, sh, &s, disks);
	}
3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320 3321 3322 3323 3324 3325 3326 3327 3328 3329 3330 3331 3332 3333 3334 3335 3336 3337 3338

	if (s.syncing && s.locked == 0 && test_bit(STRIPE_INSYNC, &sh->state)) {
		md_done_sync(conf->mddev, STRIPE_SECTORS, 1);
		clear_bit(STRIPE_SYNCING, &sh->state);
	}

	/* If the failed drives are just a ReadError, then we might need
	 * to progress the repair/check process
	 */
	if (s.failed <= conf->max_degraded && !conf->mddev->ro)
		for (i = 0; i < s.failed; i++) {
			struct r5dev *dev = &sh->dev[s.failed_num[i]];
			if (test_bit(R5_ReadError, &dev->flags)
			    && !test_bit(R5_LOCKED, &dev->flags)
			    && test_bit(R5_UPTODATE, &dev->flags)
				) {
				if (!test_bit(R5_ReWrite, &dev->flags)) {
					set_bit(R5_Wantwrite, &dev->flags);
					set_bit(R5_ReWrite, &dev->flags);
					set_bit(R5_LOCKED, &dev->flags);
					s.locked++;
				} else {
					/* let's read it back */
					set_bit(R5_Wantread, &dev->flags);
					set_bit(R5_LOCKED, &dev->flags);
					s.locked++;
				}
			}
		}


3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353 3354 3355 3356 3357 3358 3359 3360 3361 3362 3363 3364 3365
	/* Finish reconstruct operations initiated by the expansion process */
	if (sh->reconstruct_state == reconstruct_state_result) {
		struct stripe_head *sh_src
			= get_active_stripe(conf, sh->sector, 1, 1, 1);
		if (sh_src && test_bit(STRIPE_EXPAND_SOURCE, &sh_src->state)) {
			/* sh cannot be written until sh_src has been read.
			 * so arrange for sh to be delayed a little
			 */
			set_bit(STRIPE_DELAYED, &sh->state);
			set_bit(STRIPE_HANDLE, &sh->state);
			if (!test_and_set_bit(STRIPE_PREREAD_ACTIVE,
					      &sh_src->state))
				atomic_inc(&conf->preread_active_stripes);
			release_stripe(sh_src);
			goto finish;
		}
		if (sh_src)
			release_stripe(sh_src);

		sh->reconstruct_state = reconstruct_state_idle;
		clear_bit(STRIPE_EXPANDING, &sh->state);
		for (i = conf->raid_disks; i--; ) {
			set_bit(R5_Wantwrite, &sh->dev[i].flags);
			set_bit(R5_LOCKED, &sh->dev[i].flags);
			s.locked++;
		}
	}
3366

3367 3368 3369 3370 3371 3372 3373 3374 3375 3376 3377 3378 3379 3380 3381 3382
	if (s.expanded && test_bit(STRIPE_EXPANDING, &sh->state) &&
	    !sh->reconstruct_state) {
		/* Need to write out all blocks after computing parity */
		sh->disks = conf->raid_disks;
		stripe_set_idx(sh->sector, conf, 0, sh);
		schedule_reconstruction(sh, &s, 1, 1);
	} else if (s.expanded && !sh->reconstruct_state && s.locked == 0) {
		clear_bit(STRIPE_EXPAND_READY, &sh->state);
		atomic_dec(&conf->reshape_stripes);
		wake_up(&conf->wait_for_overlap);
		md_done_sync(conf->mddev, STRIPE_SECTORS, 1);
	}

	if (s.expanding && s.locked == 0 &&
	    !test_bit(STRIPE_COMPUTE_RUN, &sh->state))
		handle_stripe_expansion(conf, sh);
3383

3384
finish:
3385
	/* wait for this device to become unblocked */
3386
	if (conf->mddev->external && unlikely(s.blocked_rdev))
3387
		md_wait_for_blocked_rdev(s.blocked_rdev, conf->mddev);
3388

3389 3390
	if (s.handle_bad_blocks)
		for (i = disks; i--; ) {
3391
			struct md_rdev *rdev;
3392 3393 3394 3395 3396 3397 3398 3399 3400
			struct r5dev *dev = &sh->dev[i];
			if (test_and_clear_bit(R5_WriteError, &dev->flags)) {
				/* We own a safe reference to the rdev */
				rdev = conf->disks[i].rdev;
				if (!rdev_set_badblocks(rdev, sh->sector,
							STRIPE_SECTORS, 0))
					md_error(conf->mddev, rdev);
				rdev_dec_pending(rdev, conf->mddev);
			}
3401 3402 3403 3404 3405 3406
			if (test_and_clear_bit(R5_MadeGood, &dev->flags)) {
				rdev = conf->disks[i].rdev;
				rdev_clear_badblocks(rdev, sh->sector,
						     STRIPE_SECTORS);
				rdev_dec_pending(rdev, conf->mddev);
			}
3407 3408
		}

3409 3410 3411
	if (s.ops_request)
		raid_run_ops(sh, s.ops_request);

D
Dan Williams 已提交
3412
	ops_run_io(sh, &s);
3413

3414
	if (s.dec_preread_active) {
3415
		/* We delay this until after ops_run_io so that if make_request
T
Tejun Heo 已提交
3416
		 * is waiting on a flush, it won't continue until the writes
3417 3418 3419 3420 3421 3422 3423 3424
		 * have actually been submitted.
		 */
		atomic_dec(&conf->preread_active_stripes);
		if (atomic_read(&conf->preread_active_stripes) <
		    IO_THRESHOLD)
			md_wakeup_thread(conf->mddev->thread);
	}

3425
	return_io(s.return_bi);
3426

3427
	clear_bit_unlock(STRIPE_ACTIVE, &sh->state);
3428 3429
}

3430
static void raid5_activate_delayed(struct r5conf *conf)
3431 3432 3433 3434 3435 3436 3437 3438 3439 3440
{
	if (atomic_read(&conf->preread_active_stripes) < IO_THRESHOLD) {
		while (!list_empty(&conf->delayed_list)) {
			struct list_head *l = conf->delayed_list.next;
			struct stripe_head *sh;
			sh = list_entry(l, struct stripe_head, lru);
			list_del_init(l);
			clear_bit(STRIPE_DELAYED, &sh->state);
			if (!test_and_set_bit(STRIPE_PREREAD_ACTIVE, &sh->state))
				atomic_inc(&conf->preread_active_stripes);
3441
			list_add_tail(&sh->lru, &conf->hold_list);
3442
		}
N
NeilBrown 已提交
3443
	}
3444 3445
}

3446
static void activate_bit_delay(struct r5conf *conf)
3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459
{
	/* device_lock is held */
	struct list_head head;
	list_add(&head, &conf->bitmap_list);
	list_del_init(&conf->bitmap_list);
	while (!list_empty(&head)) {
		struct stripe_head *sh = list_entry(head.next, struct stripe_head, lru);
		list_del_init(&sh->lru);
		atomic_inc(&sh->count);
		__release_stripe(conf, sh);
	}
}

3460
int md_raid5_congested(struct mddev *mddev, int bits)
3461
{
3462
	struct r5conf *conf = mddev->private;
3463 3464 3465 3466

	/* No difference between reads and writes.  Just check
	 * how busy the stripe_cache is
	 */
3467

3468 3469 3470 3471 3472 3473 3474 3475 3476
	if (conf->inactive_blocked)
		return 1;
	if (conf->quiesce)
		return 1;
	if (list_empty_careful(&conf->inactive_list))
		return 1;

	return 0;
}
N
NeilBrown 已提交
3477 3478 3479 3480
EXPORT_SYMBOL_GPL(md_raid5_congested);

static int raid5_congested(void *data, int bits)
{
3481
	struct mddev *mddev = data;
N
NeilBrown 已提交
3482 3483 3484 3485

	return mddev_congested(mddev, bits) ||
		md_raid5_congested(mddev, bits);
}
3486

3487 3488 3489
/* We want read requests to align with chunks where possible,
 * but write requests don't need to.
 */
3490 3491 3492
static int raid5_mergeable_bvec(struct request_queue *q,
				struct bvec_merge_data *bvm,
				struct bio_vec *biovec)
3493
{
3494
	struct mddev *mddev = q->queuedata;
3495
	sector_t sector = bvm->bi_sector + get_start_sect(bvm->bi_bdev);
3496
	int max;
3497
	unsigned int chunk_sectors = mddev->chunk_sectors;
3498
	unsigned int bio_sectors = bvm->bi_size >> 9;
3499

3500
	if ((bvm->bi_rw & 1) == WRITE)
3501 3502
		return biovec->bv_len; /* always allow writes to be mergeable */

3503 3504
	if (mddev->new_chunk_sectors < mddev->chunk_sectors)
		chunk_sectors = mddev->new_chunk_sectors;
3505 3506 3507 3508 3509 3510 3511 3512
	max =  (chunk_sectors - ((sector & (chunk_sectors - 1)) + bio_sectors)) << 9;
	if (max < 0) max = 0;
	if (max <= biovec->bv_len && bio_sectors == 0)
		return biovec->bv_len;
	else
		return max;
}

3513

3514
static int in_chunk_boundary(struct mddev *mddev, struct bio *bio)
3515 3516
{
	sector_t sector = bio->bi_sector + get_start_sect(bio->bi_bdev);
3517
	unsigned int chunk_sectors = mddev->chunk_sectors;
3518 3519
	unsigned int bio_sectors = bio->bi_size >> 9;

3520 3521
	if (mddev->new_chunk_sectors < mddev->chunk_sectors)
		chunk_sectors = mddev->new_chunk_sectors;
3522 3523 3524 3525
	return  chunk_sectors >=
		((sector & (chunk_sectors - 1)) + bio_sectors);
}

3526 3527 3528 3529
/*
 *  add bio to the retry LIFO  ( in O(1) ... we are in interrupt )
 *  later sampled by raid5d.
 */
3530
static void add_bio_to_retry(struct bio *bi,struct r5conf *conf)
3531 3532 3533 3534 3535 3536 3537 3538 3539 3540 3541 3542 3543
{
	unsigned long flags;

	spin_lock_irqsave(&conf->device_lock, flags);

	bi->bi_next = conf->retry_read_aligned_list;
	conf->retry_read_aligned_list = bi;

	spin_unlock_irqrestore(&conf->device_lock, flags);
	md_wakeup_thread(conf->mddev->thread);
}


3544
static struct bio *remove_bio_from_retry(struct r5conf *conf)
3545 3546 3547 3548 3549 3550 3551 3552 3553 3554
{
	struct bio *bi;

	bi = conf->retry_read_aligned;
	if (bi) {
		conf->retry_read_aligned = NULL;
		return bi;
	}
	bi = conf->retry_read_aligned_list;
	if(bi) {
3555
		conf->retry_read_aligned_list = bi->bi_next;
3556
		bi->bi_next = NULL;
3557 3558 3559 3560
		/*
		 * this sets the active strip count to 1 and the processed
		 * strip count to zero (upper 8 bits)
		 */
3561 3562 3563 3564 3565 3566 3567
		bi->bi_phys_segments = 1; /* biased count of active stripes */
	}

	return bi;
}


3568 3569 3570 3571 3572 3573
/*
 *  The "raid5_align_endio" should check if the read succeeded and if it
 *  did, call bio_endio on the original bio (having bio_put the new bio
 *  first).
 *  If the read failed..
 */
3574
static void raid5_align_endio(struct bio *bi, int error)
3575 3576
{
	struct bio* raid_bi  = bi->bi_private;
3577
	struct mddev *mddev;
3578
	struct r5conf *conf;
3579
	int uptodate = test_bit(BIO_UPTODATE, &bi->bi_flags);
3580
	struct md_rdev *rdev;
3581

3582
	bio_put(bi);
3583 3584 3585

	rdev = (void*)raid_bi->bi_next;
	raid_bi->bi_next = NULL;
3586 3587
	mddev = rdev->mddev;
	conf = mddev->private;
3588 3589 3590 3591

	rdev_dec_pending(rdev, conf->mddev);

	if (!error && uptodate) {
3592
		bio_endio(raid_bi, 0);
3593 3594
		if (atomic_dec_and_test(&conf->active_aligned_reads))
			wake_up(&conf->wait_for_stripe);
3595
		return;
3596 3597 3598
	}


3599
	pr_debug("raid5_align_endio : io error...handing IO for a retry\n");
3600 3601

	add_bio_to_retry(raid_bi, conf);
3602 3603
}

3604 3605
static int bio_fits_rdev(struct bio *bi)
{
3606
	struct request_queue *q = bdev_get_queue(bi->bi_bdev);
3607

3608
	if ((bi->bi_size>>9) > queue_max_sectors(q))
3609 3610
		return 0;
	blk_recount_segments(q, bi);
3611
	if (bi->bi_phys_segments > queue_max_segments(q))
3612 3613 3614 3615 3616 3617 3618 3619 3620 3621 3622 3623
		return 0;

	if (q->merge_bvec_fn)
		/* it's too hard to apply the merge_bvec_fn at this stage,
		 * just just give up
		 */
		return 0;

	return 1;
}


3624
static int chunk_aligned_read(struct mddev *mddev, struct bio * raid_bio)
3625
{
3626
	struct r5conf *conf = mddev->private;
N
NeilBrown 已提交
3627
	int dd_idx;
3628
	struct bio* align_bi;
3629
	struct md_rdev *rdev;
3630
	sector_t end_sector;
3631 3632

	if (!in_chunk_boundary(mddev, raid_bio)) {
3633
		pr_debug("chunk_aligned_read : non aligned\n");
3634 3635 3636
		return 0;
	}
	/*
3637
	 * use bio_clone_mddev to make a copy of the bio
3638
	 */
3639
	align_bi = bio_clone_mddev(raid_bio, GFP_NOIO, mddev);
3640 3641 3642 3643 3644 3645 3646 3647 3648 3649 3650
	if (!align_bi)
		return 0;
	/*
	 *   set bi_end_io to a new function, and set bi_private to the
	 *     original bio.
	 */
	align_bi->bi_end_io  = raid5_align_endio;
	align_bi->bi_private = raid_bio;
	/*
	 *	compute position
	 */
3651 3652
	align_bi->bi_sector =  raid5_compute_sector(conf, raid_bio->bi_sector,
						    0,
3653
						    &dd_idx, NULL);
3654

3655
	end_sector = align_bi->bi_sector + (align_bi->bi_size >> 9);
3656
	rcu_read_lock();
3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667
	rdev = rcu_dereference(conf->disks[dd_idx].replacement);
	if (!rdev || test_bit(Faulty, &rdev->flags) ||
	    rdev->recovery_offset < end_sector) {
		rdev = rcu_dereference(conf->disks[dd_idx].rdev);
		if (rdev &&
		    (test_bit(Faulty, &rdev->flags) ||
		    !(test_bit(In_sync, &rdev->flags) ||
		      rdev->recovery_offset >= end_sector)))
			rdev = NULL;
	}
	if (rdev) {
3668 3669 3670
		sector_t first_bad;
		int bad_sectors;

3671 3672
		atomic_inc(&rdev->nr_pending);
		rcu_read_unlock();
3673 3674 3675 3676 3677
		raid_bio->bi_next = (void*)rdev;
		align_bi->bi_bdev =  rdev->bdev;
		align_bi->bi_flags &= ~(1 << BIO_SEG_VALID);
		align_bi->bi_sector += rdev->data_offset;

3678 3679 3680 3681
		if (!bio_fits_rdev(align_bi) ||
		    is_badblock(rdev, align_bi->bi_sector, align_bi->bi_size>>9,
				&first_bad, &bad_sectors)) {
			/* too big in some way, or has a known bad block */
3682 3683 3684 3685 3686
			bio_put(align_bi);
			rdev_dec_pending(rdev, mddev);
			return 0;
		}

3687 3688 3689 3690 3691 3692 3693
		spin_lock_irq(&conf->device_lock);
		wait_event_lock_irq(conf->wait_for_stripe,
				    conf->quiesce == 0,
				    conf->device_lock, /* nothing */);
		atomic_inc(&conf->active_aligned_reads);
		spin_unlock_irq(&conf->device_lock);

3694 3695 3696 3697
		generic_make_request(align_bi);
		return 1;
	} else {
		rcu_read_unlock();
3698
		bio_put(align_bi);
3699 3700 3701 3702
		return 0;
	}
}

3703 3704 3705 3706 3707 3708 3709 3710 3711 3712
/* __get_priority_stripe - get the next stripe to process
 *
 * Full stripe writes are allowed to pass preread active stripes up until
 * the bypass_threshold is exceeded.  In general the bypass_count
 * increments when the handle_list is handled before the hold_list; however, it
 * will not be incremented when STRIPE_IO_STARTED is sampled set signifying a
 * stripe with in flight i/o.  The bypass_count will be reset when the
 * head of the hold_list has changed, i.e. the head was promoted to the
 * handle_list.
 */
3713
static struct stripe_head *__get_priority_stripe(struct r5conf *conf)
3714 3715 3716 3717 3718 3719 3720 3721 3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733 3734 3735 3736 3737 3738 3739 3740 3741 3742 3743 3744 3745 3746 3747 3748 3749 3750 3751 3752 3753 3754
{
	struct stripe_head *sh;

	pr_debug("%s: handle: %s hold: %s full_writes: %d bypass_count: %d\n",
		  __func__,
		  list_empty(&conf->handle_list) ? "empty" : "busy",
		  list_empty(&conf->hold_list) ? "empty" : "busy",
		  atomic_read(&conf->pending_full_writes), conf->bypass_count);

	if (!list_empty(&conf->handle_list)) {
		sh = list_entry(conf->handle_list.next, typeof(*sh), lru);

		if (list_empty(&conf->hold_list))
			conf->bypass_count = 0;
		else if (!test_bit(STRIPE_IO_STARTED, &sh->state)) {
			if (conf->hold_list.next == conf->last_hold)
				conf->bypass_count++;
			else {
				conf->last_hold = conf->hold_list.next;
				conf->bypass_count -= conf->bypass_threshold;
				if (conf->bypass_count < 0)
					conf->bypass_count = 0;
			}
		}
	} else if (!list_empty(&conf->hold_list) &&
		   ((conf->bypass_threshold &&
		     conf->bypass_count > conf->bypass_threshold) ||
		    atomic_read(&conf->pending_full_writes) == 0)) {
		sh = list_entry(conf->hold_list.next,
				typeof(*sh), lru);
		conf->bypass_count -= conf->bypass_threshold;
		if (conf->bypass_count < 0)
			conf->bypass_count = 0;
	} else
		return NULL;

	list_del_init(&sh->lru);
	atomic_inc(&sh->count);
	BUG_ON(atomic_read(&sh->count) != 1);
	return sh;
}
3755

3756
static void make_request(struct mddev *mddev, struct bio * bi)
L
Linus Torvalds 已提交
3757
{
3758
	struct r5conf *conf = mddev->private;
3759
	int dd_idx;
L
Linus Torvalds 已提交
3760 3761 3762
	sector_t new_sector;
	sector_t logical_sector, last_sector;
	struct stripe_head *sh;
3763
	const int rw = bio_data_dir(bi);
3764
	int remaining;
3765
	int plugged;
L
Linus Torvalds 已提交
3766

T
Tejun Heo 已提交
3767 3768
	if (unlikely(bi->bi_rw & REQ_FLUSH)) {
		md_flush_request(mddev, bi);
3769
		return;
3770 3771
	}

3772
	md_write_start(mddev, bi);
3773

3774
	if (rw == READ &&
3775
	     mddev->reshape_position == MaxSector &&
3776
	     chunk_aligned_read(mddev,bi))
3777
		return;
3778

L
Linus Torvalds 已提交
3779 3780 3781 3782
	logical_sector = bi->bi_sector & ~((sector_t)STRIPE_SECTORS-1);
	last_sector = bi->bi_sector + (bi->bi_size>>9);
	bi->bi_next = NULL;
	bi->bi_phys_segments = 1;	/* over-loaded to count active stripes */
3783

3784
	plugged = mddev_check_plugged(mddev);
L
Linus Torvalds 已提交
3785 3786
	for (;logical_sector < last_sector; logical_sector += STRIPE_SECTORS) {
		DEFINE_WAIT(w);
3787
		int disks, data_disks;
3788
		int previous;
3789

3790
	retry:
3791
		previous = 0;
3792
		disks = conf->raid_disks;
3793
		prepare_to_wait(&conf->wait_for_overlap, &w, TASK_UNINTERRUPTIBLE);
3794
		if (unlikely(conf->reshape_progress != MaxSector)) {
3795
			/* spinlock is needed as reshape_progress may be
3796 3797
			 * 64bit on a 32bit platform, and so it might be
			 * possible to see a half-updated value
3798
			 * Of course reshape_progress could change after
3799 3800 3801 3802
			 * the lock is dropped, so once we get a reference
			 * to the stripe that we think it is, we will have
			 * to check again.
			 */
3803
			spin_lock_irq(&conf->device_lock);
3804 3805 3806
			if (mddev->delta_disks < 0
			    ? logical_sector < conf->reshape_progress
			    : logical_sector >= conf->reshape_progress) {
3807
				disks = conf->previous_raid_disks;
3808 3809
				previous = 1;
			} else {
3810 3811 3812
				if (mddev->delta_disks < 0
				    ? logical_sector < conf->reshape_safe
				    : logical_sector >= conf->reshape_safe) {
3813 3814 3815 3816 3817
					spin_unlock_irq(&conf->device_lock);
					schedule();
					goto retry;
				}
			}
3818 3819
			spin_unlock_irq(&conf->device_lock);
		}
3820 3821
		data_disks = disks - conf->max_degraded;

3822 3823
		new_sector = raid5_compute_sector(conf, logical_sector,
						  previous,
3824
						  &dd_idx, NULL);
3825
		pr_debug("raid456: make_request, sector %llu logical %llu\n",
L
Linus Torvalds 已提交
3826 3827 3828
			(unsigned long long)new_sector, 
			(unsigned long long)logical_sector);

3829
		sh = get_active_stripe(conf, new_sector, previous,
3830
				       (bi->bi_rw&RWA_MASK), 0);
L
Linus Torvalds 已提交
3831
		if (sh) {
3832
			if (unlikely(previous)) {
3833
				/* expansion might have moved on while waiting for a
3834 3835 3836 3837 3838 3839
				 * stripe, so we must do the range check again.
				 * Expansion could still move past after this
				 * test, but as we are holding a reference to
				 * 'sh', we know that if that happens,
				 *  STRIPE_EXPANDING will get set and the expansion
				 * won't proceed until we finish with the stripe.
3840 3841 3842
				 */
				int must_retry = 0;
				spin_lock_irq(&conf->device_lock);
3843 3844 3845
				if (mddev->delta_disks < 0
				    ? logical_sector >= conf->reshape_progress
				    : logical_sector < conf->reshape_progress)
3846 3847 3848 3849 3850
					/* mismatch, need to try again */
					must_retry = 1;
				spin_unlock_irq(&conf->device_lock);
				if (must_retry) {
					release_stripe(sh);
3851
					schedule();
3852 3853 3854
					goto retry;
				}
			}
3855

3856
			if (rw == WRITE &&
3857
			    logical_sector >= mddev->suspend_lo &&
3858 3859
			    logical_sector < mddev->suspend_hi) {
				release_stripe(sh);
3860 3861 3862 3863 3864 3865 3866 3867 3868 3869
				/* As the suspend_* range is controlled by
				 * userspace, we want an interruptible
				 * wait.
				 */
				flush_signals(current);
				prepare_to_wait(&conf->wait_for_overlap,
						&w, TASK_INTERRUPTIBLE);
				if (logical_sector >= mddev->suspend_lo &&
				    logical_sector < mddev->suspend_hi)
					schedule();
3870 3871
				goto retry;
			}
3872 3873

			if (test_bit(STRIPE_EXPANDING, &sh->state) ||
3874
			    !add_stripe_bio(sh, bi, dd_idx, rw)) {
3875 3876
				/* Stripe is busy expanding or
				 * add failed due to overlap.  Flush everything
L
Linus Torvalds 已提交
3877 3878
				 * and wait a while
				 */
N
NeilBrown 已提交
3879
				md_wakeup_thread(mddev->thread);
L
Linus Torvalds 已提交
3880 3881 3882 3883 3884
				release_stripe(sh);
				schedule();
				goto retry;
			}
			finish_wait(&conf->wait_for_overlap, &w);
3885 3886
			set_bit(STRIPE_HANDLE, &sh->state);
			clear_bit(STRIPE_DELAYED, &sh->state);
T
Tejun Heo 已提交
3887
			if ((bi->bi_rw & REQ_SYNC) &&
3888 3889
			    !test_and_set_bit(STRIPE_PREREAD_ACTIVE, &sh->state))
				atomic_inc(&conf->preread_active_stripes);
L
Linus Torvalds 已提交
3890 3891 3892 3893 3894 3895 3896 3897 3898
			release_stripe(sh);
		} else {
			/* cannot get stripe for read-ahead, just give-up */
			clear_bit(BIO_UPTODATE, &bi->bi_flags);
			finish_wait(&conf->wait_for_overlap, &w);
			break;
		}
			
	}
3899 3900 3901
	if (!plugged)
		md_wakeup_thread(mddev->thread);

L
Linus Torvalds 已提交
3902
	spin_lock_irq(&conf->device_lock);
3903
	remaining = raid5_dec_bi_phys_segments(bi);
3904 3905
	spin_unlock_irq(&conf->device_lock);
	if (remaining == 0) {
L
Linus Torvalds 已提交
3906

3907
		if ( rw == WRITE )
L
Linus Torvalds 已提交
3908
			md_write_end(mddev);
3909

3910
		bio_endio(bi, 0);
L
Linus Torvalds 已提交
3911 3912 3913
	}
}

3914
static sector_t raid5_size(struct mddev *mddev, sector_t sectors, int raid_disks);
D
Dan Williams 已提交
3915

3916
static sector_t reshape_request(struct mddev *mddev, sector_t sector_nr, int *skipped)
L
Linus Torvalds 已提交
3917
{
3918 3919 3920 3921 3922 3923 3924 3925 3926
	/* reshaping is quite different to recovery/resync so it is
	 * handled quite separately ... here.
	 *
	 * On each call to sync_request, we gather one chunk worth of
	 * destination stripes and flag them as expanding.
	 * Then we find all the source stripes and request reads.
	 * As the reads complete, handle_stripe will copy the data
	 * into the destination stripe and release that stripe.
	 */
3927
	struct r5conf *conf = mddev->private;
L
Linus Torvalds 已提交
3928
	struct stripe_head *sh;
3929
	sector_t first_sector, last_sector;
3930 3931 3932
	int raid_disks = conf->previous_raid_disks;
	int data_disks = raid_disks - conf->max_degraded;
	int new_data_disks = conf->raid_disks - conf->max_degraded;
3933 3934
	int i;
	int dd_idx;
3935
	sector_t writepos, readpos, safepos;
3936
	sector_t stripe_addr;
3937
	int reshape_sectors;
3938
	struct list_head stripes;
3939

3940 3941 3942 3943 3944 3945
	if (sector_nr == 0) {
		/* If restarting in the middle, skip the initial sectors */
		if (mddev->delta_disks < 0 &&
		    conf->reshape_progress < raid5_size(mddev, 0, 0)) {
			sector_nr = raid5_size(mddev, 0, 0)
				- conf->reshape_progress;
3946
		} else if (mddev->delta_disks >= 0 &&
3947 3948
			   conf->reshape_progress > 0)
			sector_nr = conf->reshape_progress;
3949
		sector_div(sector_nr, new_data_disks);
3950
		if (sector_nr) {
3951 3952
			mddev->curr_resync_completed = sector_nr;
			sysfs_notify(&mddev->kobj, NULL, "sync_completed");
3953 3954 3955
			*skipped = 1;
			return sector_nr;
		}
3956 3957
	}

3958 3959 3960 3961
	/* We need to process a full chunk at a time.
	 * If old and new chunk sizes differ, we need to process the
	 * largest of these
	 */
3962 3963
	if (mddev->new_chunk_sectors > mddev->chunk_sectors)
		reshape_sectors = mddev->new_chunk_sectors;
3964
	else
3965
		reshape_sectors = mddev->chunk_sectors;
3966

3967 3968 3969 3970 3971
	/* we update the metadata when there is more than 3Meg
	 * in the block range (that is rather arbitrary, should
	 * probably be time based) or when the data about to be
	 * copied would over-write the source of the data at
	 * the front of the range.
3972 3973
	 * i.e. one new_stripe along from reshape_progress new_maps
	 * to after where reshape_safe old_maps to
3974
	 */
3975
	writepos = conf->reshape_progress;
3976
	sector_div(writepos, new_data_disks);
3977 3978
	readpos = conf->reshape_progress;
	sector_div(readpos, data_disks);
3979
	safepos = conf->reshape_safe;
3980
	sector_div(safepos, data_disks);
3981
	if (mddev->delta_disks < 0) {
3982
		writepos -= min_t(sector_t, reshape_sectors, writepos);
3983
		readpos += reshape_sectors;
3984
		safepos += reshape_sectors;
3985
	} else {
3986
		writepos += reshape_sectors;
3987 3988
		readpos -= min_t(sector_t, reshape_sectors, readpos);
		safepos -= min_t(sector_t, reshape_sectors, safepos);
3989
	}
3990

3991 3992 3993 3994 3995 3996 3997 3998 3999 4000 4001 4002 4003 4004 4005 4006 4007
	/* 'writepos' is the most advanced device address we might write.
	 * 'readpos' is the least advanced device address we might read.
	 * 'safepos' is the least address recorded in the metadata as having
	 *     been reshaped.
	 * If 'readpos' is behind 'writepos', then there is no way that we can
	 * ensure safety in the face of a crash - that must be done by userspace
	 * making a backup of the data.  So in that case there is no particular
	 * rush to update metadata.
	 * Otherwise if 'safepos' is behind 'writepos', then we really need to
	 * update the metadata to advance 'safepos' to match 'readpos' so that
	 * we can be safe in the event of a crash.
	 * So we insist on updating metadata if safepos is behind writepos and
	 * readpos is beyond writepos.
	 * In any case, update the metadata every 10 seconds.
	 * Maybe that number should be configurable, but I'm not sure it is
	 * worth it.... maybe it could be a multiple of safemode_delay???
	 */
4008
	if ((mddev->delta_disks < 0
4009 4010 4011
	     ? (safepos > writepos && readpos < writepos)
	     : (safepos < writepos && readpos > writepos)) ||
	    time_after(jiffies, conf->reshape_checkpoint + 10*HZ)) {
4012 4013 4014
		/* Cannot proceed until we've updated the superblock... */
		wait_event(conf->wait_for_overlap,
			   atomic_read(&conf->reshape_stripes)==0);
4015
		mddev->reshape_position = conf->reshape_progress;
4016
		mddev->curr_resync_completed = sector_nr;
4017
		conf->reshape_checkpoint = jiffies;
4018
		set_bit(MD_CHANGE_DEVS, &mddev->flags);
4019
		md_wakeup_thread(mddev->thread);
4020
		wait_event(mddev->sb_wait, mddev->flags == 0 ||
4021 4022
			   kthread_should_stop());
		spin_lock_irq(&conf->device_lock);
4023
		conf->reshape_safe = mddev->reshape_position;
4024 4025
		spin_unlock_irq(&conf->device_lock);
		wake_up(&conf->wait_for_overlap);
4026
		sysfs_notify(&mddev->kobj, NULL, "sync_completed");
4027 4028
	}

4029 4030 4031 4032
	if (mddev->delta_disks < 0) {
		BUG_ON(conf->reshape_progress == 0);
		stripe_addr = writepos;
		BUG_ON((mddev->dev_sectors &
4033 4034
			~((sector_t)reshape_sectors - 1))
		       - reshape_sectors - stripe_addr
4035 4036
		       != sector_nr);
	} else {
4037
		BUG_ON(writepos != sector_nr + reshape_sectors);
4038 4039
		stripe_addr = sector_nr;
	}
4040
	INIT_LIST_HEAD(&stripes);
4041
	for (i = 0; i < reshape_sectors; i += STRIPE_SECTORS) {
4042
		int j;
4043
		int skipped_disk = 0;
4044
		sh = get_active_stripe(conf, stripe_addr+i, 0, 0, 1);
4045 4046 4047 4048 4049 4050 4051 4052 4053
		set_bit(STRIPE_EXPANDING, &sh->state);
		atomic_inc(&conf->reshape_stripes);
		/* If any of this stripe is beyond the end of the old
		 * array, then we need to zero those blocks
		 */
		for (j=sh->disks; j--;) {
			sector_t s;
			if (j == sh->pd_idx)
				continue;
4054
			if (conf->level == 6 &&
4055
			    j == sh->qd_idx)
4056
				continue;
4057
			s = compute_blocknr(sh, j, 0);
D
Dan Williams 已提交
4058
			if (s < raid5_size(mddev, 0, 0)) {
4059
				skipped_disk = 1;
4060 4061 4062 4063 4064 4065
				continue;
			}
			memset(page_address(sh->dev[j].page), 0, STRIPE_SIZE);
			set_bit(R5_Expanded, &sh->dev[j].flags);
			set_bit(R5_UPTODATE, &sh->dev[j].flags);
		}
4066
		if (!skipped_disk) {
4067 4068 4069
			set_bit(STRIPE_EXPAND_READY, &sh->state);
			set_bit(STRIPE_HANDLE, &sh->state);
		}
4070
		list_add(&sh->lru, &stripes);
4071 4072
	}
	spin_lock_irq(&conf->device_lock);
4073
	if (mddev->delta_disks < 0)
4074
		conf->reshape_progress -= reshape_sectors * new_data_disks;
4075
	else
4076
		conf->reshape_progress += reshape_sectors * new_data_disks;
4077 4078 4079 4080 4081 4082 4083
	spin_unlock_irq(&conf->device_lock);
	/* Ok, those stripe are ready. We can start scheduling
	 * reads on the source stripes.
	 * The source stripes are determined by mapping the first and last
	 * block on the destination stripes.
	 */
	first_sector =
4084
		raid5_compute_sector(conf, stripe_addr*(new_data_disks),
4085
				     1, &dd_idx, NULL);
4086
	last_sector =
4087
		raid5_compute_sector(conf, ((stripe_addr+reshape_sectors)
4088
					    * new_data_disks - 1),
4089
				     1, &dd_idx, NULL);
A
Andre Noll 已提交
4090 4091
	if (last_sector >= mddev->dev_sectors)
		last_sector = mddev->dev_sectors - 1;
4092
	while (first_sector <= last_sector) {
4093
		sh = get_active_stripe(conf, first_sector, 1, 0, 1);
4094 4095 4096 4097 4098
		set_bit(STRIPE_EXPAND_SOURCE, &sh->state);
		set_bit(STRIPE_HANDLE, &sh->state);
		release_stripe(sh);
		first_sector += STRIPE_SECTORS;
	}
4099 4100 4101 4102 4103 4104 4105 4106
	/* Now that the sources are clearly marked, we can release
	 * the destination stripes
	 */
	while (!list_empty(&stripes)) {
		sh = list_entry(stripes.next, struct stripe_head, lru);
		list_del_init(&sh->lru);
		release_stripe(sh);
	}
4107 4108 4109
	/* If this takes us to the resync_max point where we have to pause,
	 * then we need to write out the superblock.
	 */
4110
	sector_nr += reshape_sectors;
4111 4112
	if ((sector_nr - mddev->curr_resync_completed) * 2
	    >= mddev->resync_max - mddev->curr_resync_completed) {
4113 4114 4115
		/* Cannot proceed until we've updated the superblock... */
		wait_event(conf->wait_for_overlap,
			   atomic_read(&conf->reshape_stripes) == 0);
4116
		mddev->reshape_position = conf->reshape_progress;
4117
		mddev->curr_resync_completed = sector_nr;
4118
		conf->reshape_checkpoint = jiffies;
4119 4120 4121 4122 4123 4124
		set_bit(MD_CHANGE_DEVS, &mddev->flags);
		md_wakeup_thread(mddev->thread);
		wait_event(mddev->sb_wait,
			   !test_bit(MD_CHANGE_DEVS, &mddev->flags)
			   || kthread_should_stop());
		spin_lock_irq(&conf->device_lock);
4125
		conf->reshape_safe = mddev->reshape_position;
4126 4127
		spin_unlock_irq(&conf->device_lock);
		wake_up(&conf->wait_for_overlap);
4128
		sysfs_notify(&mddev->kobj, NULL, "sync_completed");
4129
	}
4130
	return reshape_sectors;
4131 4132 4133
}

/* FIXME go_faster isn't used */
4134
static inline sector_t sync_request(struct mddev *mddev, sector_t sector_nr, int *skipped, int go_faster)
4135
{
4136
	struct r5conf *conf = mddev->private;
4137
	struct stripe_head *sh;
A
Andre Noll 已提交
4138
	sector_t max_sector = mddev->dev_sectors;
N
NeilBrown 已提交
4139
	sector_t sync_blocks;
4140 4141
	int still_degraded = 0;
	int i;
L
Linus Torvalds 已提交
4142

4143
	if (sector_nr >= max_sector) {
L
Linus Torvalds 已提交
4144
		/* just being told to finish up .. nothing much to do */
4145

4146 4147 4148 4149
		if (test_bit(MD_RECOVERY_RESHAPE, &mddev->recovery)) {
			end_reshape(conf);
			return 0;
		}
4150 4151 4152 4153

		if (mddev->curr_resync < max_sector) /* aborted */
			bitmap_end_sync(mddev->bitmap, mddev->curr_resync,
					&sync_blocks, 1);
4154
		else /* completed sync */
4155 4156 4157
			conf->fullsync = 0;
		bitmap_close_sync(mddev->bitmap);

L
Linus Torvalds 已提交
4158 4159
		return 0;
	}
4160

4161 4162 4163
	/* Allow raid5_quiesce to complete */
	wait_event(conf->wait_for_overlap, conf->quiesce != 2);

4164 4165
	if (test_bit(MD_RECOVERY_RESHAPE, &mddev->recovery))
		return reshape_request(mddev, sector_nr, skipped);
4166

4167 4168 4169 4170 4171 4172
	/* No need to check resync_max as we never do more than one
	 * stripe, and as resync_max will always be on a chunk boundary,
	 * if the check in md_do_sync didn't fire, there is no chance
	 * of overstepping resync_max here
	 */

4173
	/* if there is too many failed drives and we are trying
L
Linus Torvalds 已提交
4174 4175 4176
	 * to resync, then assert that we are finished, because there is
	 * nothing we can do.
	 */
4177
	if (mddev->degraded >= conf->max_degraded &&
4178
	    test_bit(MD_RECOVERY_SYNC, &mddev->recovery)) {
A
Andre Noll 已提交
4179
		sector_t rv = mddev->dev_sectors - sector_nr;
4180
		*skipped = 1;
L
Linus Torvalds 已提交
4181 4182
		return rv;
	}
4183
	if (!bitmap_start_sync(mddev->bitmap, sector_nr, &sync_blocks, 1) &&
4184
	    !test_bit(MD_RECOVERY_REQUESTED, &mddev->recovery) &&
4185 4186 4187 4188 4189 4190
	    !conf->fullsync && sync_blocks >= STRIPE_SECTORS) {
		/* we can skip this block, and probably more */
		sync_blocks /= STRIPE_SECTORS;
		*skipped = 1;
		return sync_blocks * STRIPE_SECTORS; /* keep things rounded to whole stripes */
	}
L
Linus Torvalds 已提交
4191

N
NeilBrown 已提交
4192 4193 4194

	bitmap_cond_end_sync(mddev->bitmap, sector_nr);

4195
	sh = get_active_stripe(conf, sector_nr, 0, 1, 0);
L
Linus Torvalds 已提交
4196
	if (sh == NULL) {
4197
		sh = get_active_stripe(conf, sector_nr, 0, 0, 0);
L
Linus Torvalds 已提交
4198
		/* make sure we don't swamp the stripe cache if someone else
4199
		 * is trying to get access
L
Linus Torvalds 已提交
4200
		 */
4201
		schedule_timeout_uninterruptible(1);
L
Linus Torvalds 已提交
4202
	}
4203 4204 4205 4206
	/* Need to check if array will still be degraded after recovery/resync
	 * We don't need to check the 'failed' flag as when that gets set,
	 * recovery aborts.
	 */
4207
	for (i = 0; i < conf->raid_disks; i++)
4208 4209 4210 4211 4212
		if (conf->disks[i].rdev == NULL)
			still_degraded = 1;

	bitmap_start_sync(mddev->bitmap, sector_nr, &sync_blocks, still_degraded);

4213
	set_bit(STRIPE_SYNC_REQUESTED, &sh->state);
L
Linus Torvalds 已提交
4214

4215
	handle_stripe(sh);
L
Linus Torvalds 已提交
4216 4217 4218 4219 4220
	release_stripe(sh);

	return STRIPE_SECTORS;
}

4221
static int  retry_aligned_read(struct r5conf *conf, struct bio *raid_bio)
4222 4223 4224 4225 4226 4227 4228 4229 4230 4231 4232 4233
{
	/* We may not be able to submit a whole bio at once as there
	 * may not be enough stripe_heads available.
	 * We cannot pre-allocate enough stripe_heads as we may need
	 * more than exist in the cache (if we allow ever large chunks).
	 * So we do one stripe head at a time and record in
	 * ->bi_hw_segments how many have been done.
	 *
	 * We *know* that this entire raid_bio is in one chunk, so
	 * it will be only one 'dd_idx' and only need one call to raid5_compute_sector.
	 */
	struct stripe_head *sh;
4234
	int dd_idx;
4235 4236 4237 4238 4239 4240
	sector_t sector, logical_sector, last_sector;
	int scnt = 0;
	int remaining;
	int handled = 0;

	logical_sector = raid_bio->bi_sector & ~((sector_t)STRIPE_SECTORS-1);
4241
	sector = raid5_compute_sector(conf, logical_sector,
4242
				      0, &dd_idx, NULL);
4243 4244 4245
	last_sector = raid_bio->bi_sector + (raid_bio->bi_size>>9);

	for (; logical_sector < last_sector;
4246 4247 4248
	     logical_sector += STRIPE_SECTORS,
		     sector += STRIPE_SECTORS,
		     scnt++) {
4249

4250
		if (scnt < raid5_bi_hw_segments(raid_bio))
4251 4252 4253
			/* already done this stripe */
			continue;

4254
		sh = get_active_stripe(conf, sector, 0, 1, 0);
4255 4256 4257

		if (!sh) {
			/* failed to get a stripe - must wait */
4258
			raid5_set_bi_hw_segments(raid_bio, scnt);
4259 4260 4261 4262
			conf->retry_read_aligned = raid_bio;
			return handled;
		}

4263 4264
		if (!add_stripe_bio(sh, raid_bio, dd_idx, 0)) {
			release_stripe(sh);
4265
			raid5_set_bi_hw_segments(raid_bio, scnt);
4266 4267 4268 4269
			conf->retry_read_aligned = raid_bio;
			return handled;
		}

4270
		handle_stripe(sh);
4271 4272 4273 4274
		release_stripe(sh);
		handled++;
	}
	spin_lock_irq(&conf->device_lock);
4275
	remaining = raid5_dec_bi_phys_segments(raid_bio);
4276
	spin_unlock_irq(&conf->device_lock);
4277 4278
	if (remaining == 0)
		bio_endio(raid_bio, 0);
4279 4280 4281 4282 4283 4284
	if (atomic_dec_and_test(&conf->active_aligned_reads))
		wake_up(&conf->wait_for_stripe);
	return handled;
}


L
Linus Torvalds 已提交
4285 4286 4287 4288 4289 4290 4291
/*
 * This is our raid5 kernel thread.
 *
 * We scan the hash table for stripes which can be handled now.
 * During the scan, completed stripes are saved for us by the interrupt
 * handler, so that they will not have to wait for our next wakeup.
 */
4292
static void raid5d(struct mddev *mddev)
L
Linus Torvalds 已提交
4293 4294
{
	struct stripe_head *sh;
4295
	struct r5conf *conf = mddev->private;
L
Linus Torvalds 已提交
4296
	int handled;
4297
	struct blk_plug plug;
L
Linus Torvalds 已提交
4298

4299
	pr_debug("+++ raid5d active\n");
L
Linus Torvalds 已提交
4300 4301 4302

	md_check_recovery(mddev);

4303
	blk_start_plug(&plug);
L
Linus Torvalds 已提交
4304 4305 4306
	handled = 0;
	spin_lock_irq(&conf->device_lock);
	while (1) {
4307
		struct bio *bio;
L
Linus Torvalds 已提交
4308

4309 4310 4311 4312
		if (atomic_read(&mddev->plug_cnt) == 0 &&
		    !list_empty(&conf->bitmap_list)) {
			/* Now is a good time to flush some bitmap updates */
			conf->seq_flush++;
4313
			spin_unlock_irq(&conf->device_lock);
4314
			bitmap_unplug(mddev->bitmap);
4315
			spin_lock_irq(&conf->device_lock);
4316
			conf->seq_write = conf->seq_flush;
4317 4318
			activate_bit_delay(conf);
		}
4319 4320
		if (atomic_read(&mddev->plug_cnt) == 0)
			raid5_activate_delayed(conf);
4321

4322 4323 4324 4325 4326 4327 4328 4329 4330 4331
		while ((bio = remove_bio_from_retry(conf))) {
			int ok;
			spin_unlock_irq(&conf->device_lock);
			ok = retry_aligned_read(conf, bio);
			spin_lock_irq(&conf->device_lock);
			if (!ok)
				break;
			handled++;
		}

4332 4333
		sh = __get_priority_stripe(conf);

4334
		if (!sh)
L
Linus Torvalds 已提交
4335 4336 4337 4338
			break;
		spin_unlock_irq(&conf->device_lock);
		
		handled++;
4339 4340 4341
		handle_stripe(sh);
		release_stripe(sh);
		cond_resched();
L
Linus Torvalds 已提交
4342

4343 4344 4345
		if (mddev->flags & ~(1<<MD_CHANGE_PENDING))
			md_check_recovery(mddev);

L
Linus Torvalds 已提交
4346 4347
		spin_lock_irq(&conf->device_lock);
	}
4348
	pr_debug("%d stripes handled\n", handled);
L
Linus Torvalds 已提交
4349 4350 4351

	spin_unlock_irq(&conf->device_lock);

4352
	async_tx_issue_pending_all();
4353
	blk_finish_plug(&plug);
L
Linus Torvalds 已提交
4354

4355
	pr_debug("--- raid5d inactive\n");
L
Linus Torvalds 已提交
4356 4357
}

4358
static ssize_t
4359
raid5_show_stripe_cache_size(struct mddev *mddev, char *page)
4360
{
4361
	struct r5conf *conf = mddev->private;
4362 4363 4364 4365
	if (conf)
		return sprintf(page, "%d\n", conf->max_nr_stripes);
	else
		return 0;
4366 4367
}

4368
int
4369
raid5_set_cache_size(struct mddev *mddev, int size)
4370
{
4371
	struct r5conf *conf = mddev->private;
4372 4373
	int err;

4374
	if (size <= 16 || size > 32768)
4375
		return -EINVAL;
4376
	while (size < conf->max_nr_stripes) {
4377 4378 4379 4380 4381
		if (drop_one_stripe(conf))
			conf->max_nr_stripes--;
		else
			break;
	}
4382 4383 4384
	err = md_allow_write(mddev);
	if (err)
		return err;
4385
	while (size > conf->max_nr_stripes) {
4386 4387 4388 4389
		if (grow_one_stripe(conf))
			conf->max_nr_stripes++;
		else break;
	}
4390 4391 4392 4393 4394
	return 0;
}
EXPORT_SYMBOL(raid5_set_cache_size);

static ssize_t
4395
raid5_store_stripe_cache_size(struct mddev *mddev, const char *page, size_t len)
4396
{
4397
	struct r5conf *conf = mddev->private;
4398 4399 4400 4401 4402 4403 4404 4405 4406 4407 4408 4409 4410
	unsigned long new;
	int err;

	if (len >= PAGE_SIZE)
		return -EINVAL;
	if (!conf)
		return -ENODEV;

	if (strict_strtoul(page, 10, &new))
		return -EINVAL;
	err = raid5_set_cache_size(mddev, new);
	if (err)
		return err;
4411 4412
	return len;
}
4413

4414 4415 4416 4417
static struct md_sysfs_entry
raid5_stripecache_size = __ATTR(stripe_cache_size, S_IRUGO | S_IWUSR,
				raid5_show_stripe_cache_size,
				raid5_store_stripe_cache_size);
4418

4419
static ssize_t
4420
raid5_show_preread_threshold(struct mddev *mddev, char *page)
4421
{
4422
	struct r5conf *conf = mddev->private;
4423 4424 4425 4426 4427 4428 4429
	if (conf)
		return sprintf(page, "%d\n", conf->bypass_threshold);
	else
		return 0;
}

static ssize_t
4430
raid5_store_preread_threshold(struct mddev *mddev, const char *page, size_t len)
4431
{
4432
	struct r5conf *conf = mddev->private;
4433
	unsigned long new;
4434 4435 4436 4437 4438
	if (len >= PAGE_SIZE)
		return -EINVAL;
	if (!conf)
		return -ENODEV;

4439
	if (strict_strtoul(page, 10, &new))
4440
		return -EINVAL;
4441
	if (new > conf->max_nr_stripes)
4442 4443 4444 4445 4446 4447 4448 4449 4450 4451 4452
		return -EINVAL;
	conf->bypass_threshold = new;
	return len;
}

static struct md_sysfs_entry
raid5_preread_bypass_threshold = __ATTR(preread_bypass_threshold,
					S_IRUGO | S_IWUSR,
					raid5_show_preread_threshold,
					raid5_store_preread_threshold);

4453
static ssize_t
4454
stripe_cache_active_show(struct mddev *mddev, char *page)
4455
{
4456
	struct r5conf *conf = mddev->private;
4457 4458 4459 4460
	if (conf)
		return sprintf(page, "%d\n", atomic_read(&conf->active_stripes));
	else
		return 0;
4461 4462
}

4463 4464
static struct md_sysfs_entry
raid5_stripecache_active = __ATTR_RO(stripe_cache_active);
4465

4466
static struct attribute *raid5_attrs[] =  {
4467 4468
	&raid5_stripecache_size.attr,
	&raid5_stripecache_active.attr,
4469
	&raid5_preread_bypass_threshold.attr,
4470 4471
	NULL,
};
4472 4473 4474
static struct attribute_group raid5_attrs_group = {
	.name = NULL,
	.attrs = raid5_attrs,
4475 4476
};

4477
static sector_t
4478
raid5_size(struct mddev *mddev, sector_t sectors, int raid_disks)
4479
{
4480
	struct r5conf *conf = mddev->private;
4481 4482 4483

	if (!sectors)
		sectors = mddev->dev_sectors;
4484
	if (!raid_disks)
4485
		/* size is defined by the smallest of previous and new size */
4486
		raid_disks = min(conf->raid_disks, conf->previous_raid_disks);
4487

4488
	sectors &= ~((sector_t)mddev->chunk_sectors - 1);
4489
	sectors &= ~((sector_t)mddev->new_chunk_sectors - 1);
4490 4491 4492
	return sectors * (raid_disks - conf->max_degraded);
}

4493
static void raid5_free_percpu(struct r5conf *conf)
4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504
{
	struct raid5_percpu *percpu;
	unsigned long cpu;

	if (!conf->percpu)
		return;

	get_online_cpus();
	for_each_possible_cpu(cpu) {
		percpu = per_cpu_ptr(conf->percpu, cpu);
		safe_put_page(percpu->spare_page);
4505
		kfree(percpu->scribble);
4506 4507 4508 4509 4510 4511 4512 4513 4514
	}
#ifdef CONFIG_HOTPLUG_CPU
	unregister_cpu_notifier(&conf->cpu_notify);
#endif
	put_online_cpus();

	free_percpu(conf->percpu);
}

4515
static void free_conf(struct r5conf *conf)
4516 4517
{
	shrink_stripes(conf);
4518
	raid5_free_percpu(conf);
4519 4520 4521 4522 4523
	kfree(conf->disks);
	kfree(conf->stripe_hashtbl);
	kfree(conf);
}

4524 4525 4526 4527
#ifdef CONFIG_HOTPLUG_CPU
static int raid456_cpu_notify(struct notifier_block *nfb, unsigned long action,
			      void *hcpu)
{
4528
	struct r5conf *conf = container_of(nfb, struct r5conf, cpu_notify);
4529 4530 4531 4532 4533 4534
	long cpu = (long)hcpu;
	struct raid5_percpu *percpu = per_cpu_ptr(conf->percpu, cpu);

	switch (action) {
	case CPU_UP_PREPARE:
	case CPU_UP_PREPARE_FROZEN:
4535
		if (conf->level == 6 && !percpu->spare_page)
4536
			percpu->spare_page = alloc_page(GFP_KERNEL);
4537 4538 4539 4540 4541 4542 4543
		if (!percpu->scribble)
			percpu->scribble = kmalloc(conf->scribble_len, GFP_KERNEL);

		if (!percpu->scribble ||
		    (conf->level == 6 && !percpu->spare_page)) {
			safe_put_page(percpu->spare_page);
			kfree(percpu->scribble);
4544 4545
			pr_err("%s: failed memory allocation for cpu%ld\n",
			       __func__, cpu);
4546
			return notifier_from_errno(-ENOMEM);
4547 4548 4549 4550 4551
		}
		break;
	case CPU_DEAD:
	case CPU_DEAD_FROZEN:
		safe_put_page(percpu->spare_page);
4552
		kfree(percpu->scribble);
4553
		percpu->spare_page = NULL;
4554
		percpu->scribble = NULL;
4555 4556 4557 4558 4559 4560 4561 4562
		break;
	default:
		break;
	}
	return NOTIFY_OK;
}
#endif

4563
static int raid5_alloc_percpu(struct r5conf *conf)
4564 4565 4566
{
	unsigned long cpu;
	struct page *spare_page;
4567
	struct raid5_percpu __percpu *allcpus;
4568
	void *scribble;
4569 4570 4571 4572 4573 4574 4575 4576 4577 4578
	int err;

	allcpus = alloc_percpu(struct raid5_percpu);
	if (!allcpus)
		return -ENOMEM;
	conf->percpu = allcpus;

	get_online_cpus();
	err = 0;
	for_each_present_cpu(cpu) {
4579 4580 4581 4582 4583 4584 4585 4586
		if (conf->level == 6) {
			spare_page = alloc_page(GFP_KERNEL);
			if (!spare_page) {
				err = -ENOMEM;
				break;
			}
			per_cpu_ptr(conf->percpu, cpu)->spare_page = spare_page;
		}
4587
		scribble = kmalloc(conf->scribble_len, GFP_KERNEL);
4588
		if (!scribble) {
4589 4590 4591
			err = -ENOMEM;
			break;
		}
4592
		per_cpu_ptr(conf->percpu, cpu)->scribble = scribble;
4593 4594 4595 4596 4597 4598 4599 4600 4601 4602 4603 4604
	}
#ifdef CONFIG_HOTPLUG_CPU
	conf->cpu_notify.notifier_call = raid456_cpu_notify;
	conf->cpu_notify.priority = 0;
	if (err == 0)
		err = register_cpu_notifier(&conf->cpu_notify);
#endif
	put_online_cpus();

	return err;
}

4605
static struct r5conf *setup_conf(struct mddev *mddev)
L
Linus Torvalds 已提交
4606
{
4607
	struct r5conf *conf;
4608
	int raid_disk, memory, max_disks;
4609
	struct md_rdev *rdev;
L
Linus Torvalds 已提交
4610 4611
	struct disk_info *disk;

N
NeilBrown 已提交
4612 4613 4614
	if (mddev->new_level != 5
	    && mddev->new_level != 4
	    && mddev->new_level != 6) {
4615
		printk(KERN_ERR "md/raid:%s: raid level not set to 4/5/6 (%d)\n",
N
NeilBrown 已提交
4616 4617
		       mdname(mddev), mddev->new_level);
		return ERR_PTR(-EIO);
L
Linus Torvalds 已提交
4618
	}
N
NeilBrown 已提交
4619 4620 4621 4622
	if ((mddev->new_level == 5
	     && !algorithm_valid_raid5(mddev->new_layout)) ||
	    (mddev->new_level == 6
	     && !algorithm_valid_raid6(mddev->new_layout))) {
4623
		printk(KERN_ERR "md/raid:%s: layout %d not supported\n",
N
NeilBrown 已提交
4624 4625
		       mdname(mddev), mddev->new_layout);
		return ERR_PTR(-EIO);
4626
	}
N
NeilBrown 已提交
4627
	if (mddev->new_level == 6 && mddev->raid_disks < 4) {
4628
		printk(KERN_ERR "md/raid:%s: not enough configured devices (%d, minimum 4)\n",
N
NeilBrown 已提交
4629 4630
		       mdname(mddev), mddev->raid_disks);
		return ERR_PTR(-EINVAL);
4631 4632
	}

4633 4634 4635
	if (!mddev->new_chunk_sectors ||
	    (mddev->new_chunk_sectors << 9) % PAGE_SIZE ||
	    !is_power_of_2(mddev->new_chunk_sectors)) {
4636 4637
		printk(KERN_ERR "md/raid:%s: invalid chunk size %d\n",
		       mdname(mddev), mddev->new_chunk_sectors << 9);
N
NeilBrown 已提交
4638
		return ERR_PTR(-EINVAL);
4639 4640
	}

4641
	conf = kzalloc(sizeof(struct r5conf), GFP_KERNEL);
N
NeilBrown 已提交
4642
	if (conf == NULL)
L
Linus Torvalds 已提交
4643
		goto abort;
4644 4645 4646 4647 4648 4649 4650 4651 4652 4653 4654 4655
	spin_lock_init(&conf->device_lock);
	init_waitqueue_head(&conf->wait_for_stripe);
	init_waitqueue_head(&conf->wait_for_overlap);
	INIT_LIST_HEAD(&conf->handle_list);
	INIT_LIST_HEAD(&conf->hold_list);
	INIT_LIST_HEAD(&conf->delayed_list);
	INIT_LIST_HEAD(&conf->bitmap_list);
	INIT_LIST_HEAD(&conf->inactive_list);
	atomic_set(&conf->active_stripes, 0);
	atomic_set(&conf->preread_active_stripes, 0);
	atomic_set(&conf->active_aligned_reads, 0);
	conf->bypass_threshold = BYPASS_THRESHOLD;
4656
	conf->recovery_disabled = mddev->recovery_disabled - 1;
N
NeilBrown 已提交
4657 4658 4659 4660 4661

	conf->raid_disks = mddev->raid_disks;
	if (mddev->reshape_position == MaxSector)
		conf->previous_raid_disks = mddev->raid_disks;
	else
4662
		conf->previous_raid_disks = mddev->raid_disks - mddev->delta_disks;
4663 4664
	max_disks = max(conf->raid_disks, conf->previous_raid_disks);
	conf->scribble_len = scribble_len(max_disks);
4665

4666
	conf->disks = kzalloc(max_disks * sizeof(struct disk_info),
4667 4668 4669
			      GFP_KERNEL);
	if (!conf->disks)
		goto abort;
4670

L
Linus Torvalds 已提交
4671 4672
	conf->mddev = mddev;

4673
	if ((conf->stripe_hashtbl = kzalloc(PAGE_SIZE, GFP_KERNEL)) == NULL)
L
Linus Torvalds 已提交
4674 4675
		goto abort;

4676 4677 4678 4679
	conf->level = mddev->new_level;
	if (raid5_alloc_percpu(conf) != 0)
		goto abort;

4680
	pr_debug("raid456: run(%s) called.\n", mdname(mddev));
L
Linus Torvalds 已提交
4681

4682
	list_for_each_entry(rdev, &mddev->disks, same_set) {
L
Linus Torvalds 已提交
4683
		raid_disk = rdev->raid_disk;
4684
		if (raid_disk >= max_disks
L
Linus Torvalds 已提交
4685 4686 4687 4688 4689 4690
		    || raid_disk < 0)
			continue;
		disk = conf->disks + raid_disk;

		disk->rdev = rdev;

4691
		if (test_bit(In_sync, &rdev->flags)) {
L
Linus Torvalds 已提交
4692
			char b[BDEVNAME_SIZE];
4693 4694 4695
			printk(KERN_INFO "md/raid:%s: device %s operational as raid"
			       " disk %d\n",
			       mdname(mddev), bdevname(rdev->bdev, b), raid_disk);
J
Jonathan Brassow 已提交
4696
		} else if (rdev->saved_raid_disk != raid_disk)
4697 4698
			/* Cannot rely on bitmap to complete recovery */
			conf->fullsync = 1;
L
Linus Torvalds 已提交
4699 4700
	}

4701
	conf->chunk_sectors = mddev->new_chunk_sectors;
N
NeilBrown 已提交
4702
	conf->level = mddev->new_level;
4703 4704 4705 4706
	if (conf->level == 6)
		conf->max_degraded = 2;
	else
		conf->max_degraded = 1;
N
NeilBrown 已提交
4707
	conf->algorithm = mddev->new_layout;
L
Linus Torvalds 已提交
4708
	conf->max_nr_stripes = NR_STRIPES;
4709
	conf->reshape_progress = mddev->reshape_position;
4710
	if (conf->reshape_progress != MaxSector) {
4711
		conf->prev_chunk_sectors = mddev->chunk_sectors;
4712 4713
		conf->prev_algo = mddev->layout;
	}
L
Linus Torvalds 已提交
4714

N
NeilBrown 已提交
4715
	memory = conf->max_nr_stripes * (sizeof(struct stripe_head) +
4716
		 max_disks * ((sizeof(struct bio) + PAGE_SIZE))) / 1024;
N
NeilBrown 已提交
4717 4718
	if (grow_stripes(conf, conf->max_nr_stripes)) {
		printk(KERN_ERR
4719 4720
		       "md/raid:%s: couldn't allocate %dkB for buffers\n",
		       mdname(mddev), memory);
N
NeilBrown 已提交
4721 4722
		goto abort;
	} else
4723 4724
		printk(KERN_INFO "md/raid:%s: allocated %dkB\n",
		       mdname(mddev), memory);
L
Linus Torvalds 已提交
4725

4726
	conf->thread = md_register_thread(raid5d, mddev, NULL);
N
NeilBrown 已提交
4727 4728
	if (!conf->thread) {
		printk(KERN_ERR
4729
		       "md/raid:%s: couldn't allocate thread.\n",
N
NeilBrown 已提交
4730
		       mdname(mddev));
4731 4732
		goto abort;
	}
N
NeilBrown 已提交
4733 4734 4735 4736 4737

	return conf;

 abort:
	if (conf) {
4738
		free_conf(conf);
N
NeilBrown 已提交
4739 4740 4741 4742 4743
		return ERR_PTR(-EIO);
	} else
		return ERR_PTR(-ENOMEM);
}

4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754 4755 4756 4757 4758 4759 4760 4761 4762 4763 4764 4765 4766 4767 4768 4769 4770

static int only_parity(int raid_disk, int algo, int raid_disks, int max_degraded)
{
	switch (algo) {
	case ALGORITHM_PARITY_0:
		if (raid_disk < max_degraded)
			return 1;
		break;
	case ALGORITHM_PARITY_N:
		if (raid_disk >= raid_disks - max_degraded)
			return 1;
		break;
	case ALGORITHM_PARITY_0_6:
		if (raid_disk == 0 || 
		    raid_disk == raid_disks - 1)
			return 1;
		break;
	case ALGORITHM_LEFT_ASYMMETRIC_6:
	case ALGORITHM_RIGHT_ASYMMETRIC_6:
	case ALGORITHM_LEFT_SYMMETRIC_6:
	case ALGORITHM_RIGHT_SYMMETRIC_6:
		if (raid_disk == raid_disks - 1)
			return 1;
	}
	return 0;
}

4771
static int run(struct mddev *mddev)
N
NeilBrown 已提交
4772
{
4773
	struct r5conf *conf;
4774
	int working_disks = 0;
4775
	int dirty_parity_disks = 0;
4776
	struct md_rdev *rdev;
4777
	sector_t reshape_offset = 0;
N
NeilBrown 已提交
4778

4779
	if (mddev->recovery_cp != MaxSector)
4780
		printk(KERN_NOTICE "md/raid:%s: not clean"
4781 4782
		       " -- starting background reconstruction\n",
		       mdname(mddev));
N
NeilBrown 已提交
4783 4784 4785 4786 4787 4788 4789 4790
	if (mddev->reshape_position != MaxSector) {
		/* Check that we can continue the reshape.
		 * Currently only disks can change, it must
		 * increase, and we must be past the point where
		 * a stripe over-writes itself
		 */
		sector_t here_new, here_old;
		int old_disks;
4791
		int max_degraded = (mddev->level == 6 ? 2 : 1);
N
NeilBrown 已提交
4792

4793
		if (mddev->new_level != mddev->level) {
4794
			printk(KERN_ERR "md/raid:%s: unsupported reshape "
N
NeilBrown 已提交
4795 4796 4797 4798 4799 4800 4801 4802 4803 4804
			       "required - aborting.\n",
			       mdname(mddev));
			return -EINVAL;
		}
		old_disks = mddev->raid_disks - mddev->delta_disks;
		/* reshape_position must be on a new-stripe boundary, and one
		 * further up in new geometry must map after here in old
		 * geometry.
		 */
		here_new = mddev->reshape_position;
4805
		if (sector_div(here_new, mddev->new_chunk_sectors *
N
NeilBrown 已提交
4806
			       (mddev->raid_disks - max_degraded))) {
4807 4808
			printk(KERN_ERR "md/raid:%s: reshape_position not "
			       "on a stripe boundary\n", mdname(mddev));
N
NeilBrown 已提交
4809 4810
			return -EINVAL;
		}
4811
		reshape_offset = here_new * mddev->new_chunk_sectors;
N
NeilBrown 已提交
4812 4813
		/* here_new is the stripe we will write to */
		here_old = mddev->reshape_position;
4814
		sector_div(here_old, mddev->chunk_sectors *
N
NeilBrown 已提交
4815 4816 4817
			   (old_disks-max_degraded));
		/* here_old is the first stripe that we might need to read
		 * from */
4818 4819 4820 4821 4822 4823 4824 4825 4826 4827 4828
		if (mddev->delta_disks == 0) {
			/* We cannot be sure it is safe to start an in-place
			 * reshape.  It is only safe if user-space if monitoring
			 * and taking constant backups.
			 * mdadm always starts a situation like this in
			 * readonly mode so it can take control before
			 * allowing any writes.  So just check for that.
			 */
			if ((here_new * mddev->new_chunk_sectors != 
			     here_old * mddev->chunk_sectors) ||
			    mddev->ro == 0) {
4829 4830 4831
				printk(KERN_ERR "md/raid:%s: in-place reshape must be started"
				       " in read-only mode - aborting\n",
				       mdname(mddev));
4832 4833 4834 4835 4836 4837 4838
				return -EINVAL;
			}
		} else if (mddev->delta_disks < 0
		    ? (here_new * mddev->new_chunk_sectors <=
		       here_old * mddev->chunk_sectors)
		    : (here_new * mddev->new_chunk_sectors >=
		       here_old * mddev->chunk_sectors)) {
N
NeilBrown 已提交
4839
			/* Reading from the same stripe as writing to - bad */
4840 4841 4842
			printk(KERN_ERR "md/raid:%s: reshape_position too early for "
			       "auto-recovery - aborting.\n",
			       mdname(mddev));
N
NeilBrown 已提交
4843 4844
			return -EINVAL;
		}
4845 4846
		printk(KERN_INFO "md/raid:%s: reshape will continue\n",
		       mdname(mddev));
N
NeilBrown 已提交
4847 4848 4849 4850
		/* OK, we should be able to continue; */
	} else {
		BUG_ON(mddev->level != mddev->new_level);
		BUG_ON(mddev->layout != mddev->new_layout);
4851
		BUG_ON(mddev->chunk_sectors != mddev->new_chunk_sectors);
N
NeilBrown 已提交
4852
		BUG_ON(mddev->delta_disks != 0);
L
Linus Torvalds 已提交
4853
	}
N
NeilBrown 已提交
4854

4855 4856 4857 4858 4859
	if (mddev->private == NULL)
		conf = setup_conf(mddev);
	else
		conf = mddev->private;

N
NeilBrown 已提交
4860 4861 4862 4863 4864 4865 4866 4867 4868 4869
	if (IS_ERR(conf))
		return PTR_ERR(conf);

	mddev->thread = conf->thread;
	conf->thread = NULL;
	mddev->private = conf;

	/*
	 * 0 for a fully functional array, 1 or 2 for a degraded array.
	 */
4870 4871 4872
	list_for_each_entry(rdev, &mddev->disks, same_set) {
		if (rdev->raid_disk < 0)
			continue;
4873
		if (test_bit(In_sync, &rdev->flags)) {
N
NeilBrown 已提交
4874
			working_disks++;
4875 4876
			continue;
		}
4877 4878 4879 4880 4881 4882 4883 4884 4885 4886 4887 4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898 4899 4900 4901 4902 4903 4904
		/* This disc is not fully in-sync.  However if it
		 * just stored parity (beyond the recovery_offset),
		 * when we don't need to be concerned about the
		 * array being dirty.
		 * When reshape goes 'backwards', we never have
		 * partially completed devices, so we only need
		 * to worry about reshape going forwards.
		 */
		/* Hack because v0.91 doesn't store recovery_offset properly. */
		if (mddev->major_version == 0 &&
		    mddev->minor_version > 90)
			rdev->recovery_offset = reshape_offset;
			
		if (rdev->recovery_offset < reshape_offset) {
			/* We need to check old and new layout */
			if (!only_parity(rdev->raid_disk,
					 conf->algorithm,
					 conf->raid_disks,
					 conf->max_degraded))
				continue;
		}
		if (!only_parity(rdev->raid_disk,
				 conf->prev_algo,
				 conf->previous_raid_disks,
				 conf->max_degraded))
			continue;
		dirty_parity_disks++;
	}
N
NeilBrown 已提交
4905

4906
	mddev->degraded = calc_degraded(conf);
N
NeilBrown 已提交
4907

4908
	if (has_failed(conf)) {
4909
		printk(KERN_ERR "md/raid:%s: not enough operational devices"
L
Linus Torvalds 已提交
4910
			" (%d/%d failed)\n",
4911
			mdname(mddev), mddev->degraded, conf->raid_disks);
L
Linus Torvalds 已提交
4912 4913 4914
		goto abort;
	}

N
NeilBrown 已提交
4915
	/* device size must be a multiple of chunk size */
4916
	mddev->dev_sectors &= ~(mddev->chunk_sectors - 1);
N
NeilBrown 已提交
4917 4918
	mddev->resync_max_sectors = mddev->dev_sectors;

4919
	if (mddev->degraded > dirty_parity_disks &&
L
Linus Torvalds 已提交
4920
	    mddev->recovery_cp != MaxSector) {
4921 4922
		if (mddev->ok_start_degraded)
			printk(KERN_WARNING
4923 4924
			       "md/raid:%s: starting dirty degraded array"
			       " - data corruption possible.\n",
4925 4926 4927
			       mdname(mddev));
		else {
			printk(KERN_ERR
4928
			       "md/raid:%s: cannot start dirty degraded array.\n",
4929 4930 4931
			       mdname(mddev));
			goto abort;
		}
L
Linus Torvalds 已提交
4932 4933 4934
	}

	if (mddev->degraded == 0)
4935 4936
		printk(KERN_INFO "md/raid:%s: raid level %d active with %d out of %d"
		       " devices, algorithm %d\n", mdname(mddev), conf->level,
4937 4938
		       mddev->raid_disks-mddev->degraded, mddev->raid_disks,
		       mddev->new_layout);
L
Linus Torvalds 已提交
4939
	else
4940 4941 4942 4943 4944
		printk(KERN_ALERT "md/raid:%s: raid level %d active with %d"
		       " out of %d devices, algorithm %d\n",
		       mdname(mddev), conf->level,
		       mddev->raid_disks - mddev->degraded,
		       mddev->raid_disks, mddev->new_layout);
L
Linus Torvalds 已提交
4945 4946 4947

	print_raid5_conf(conf);

4948 4949
	if (conf->reshape_progress != MaxSector) {
		conf->reshape_safe = conf->reshape_progress;
4950 4951 4952 4953 4954 4955
		atomic_set(&conf->reshape_stripes, 0);
		clear_bit(MD_RECOVERY_SYNC, &mddev->recovery);
		clear_bit(MD_RECOVERY_CHECK, &mddev->recovery);
		set_bit(MD_RECOVERY_RESHAPE, &mddev->recovery);
		set_bit(MD_RECOVERY_RUNNING, &mddev->recovery);
		mddev->sync_thread = md_register_thread(md_do_sync, mddev,
4956
							"reshape");
4957 4958
	}

L
Linus Torvalds 已提交
4959 4960

	/* Ok, everything is just fine now */
4961 4962
	if (mddev->to_remove == &raid5_attrs_group)
		mddev->to_remove = NULL;
N
NeilBrown 已提交
4963 4964
	else if (mddev->kobj.sd &&
	    sysfs_create_group(&mddev->kobj, &raid5_attrs_group))
4965
		printk(KERN_WARNING
4966
		       "raid5: failed to create sysfs attributes for %s\n",
4967
		       mdname(mddev));
4968
	md_set_array_sectors(mddev, raid5_size(mddev, 0, 0));
4969

4970
	if (mddev->queue) {
4971
		int chunk_size;
4972 4973 4974 4975 4976 4977 4978 4979 4980
		/* read-ahead size must cover two whole stripes, which
		 * is 2 * (datadisks) * chunksize where 'n' is the
		 * number of raid devices
		 */
		int data_disks = conf->previous_raid_disks - conf->max_degraded;
		int stripe = data_disks *
			((mddev->chunk_sectors << 9) / PAGE_SIZE);
		if (mddev->queue->backing_dev_info.ra_pages < 2 * stripe)
			mddev->queue->backing_dev_info.ra_pages = 2 * stripe;
N
NeilBrown 已提交
4981

4982
		blk_queue_merge_bvec(mddev->queue, raid5_mergeable_bvec);
4983

N
NeilBrown 已提交
4984 4985
		mddev->queue->backing_dev_info.congested_data = mddev;
		mddev->queue->backing_dev_info.congested_fn = raid5_congested;
4986

4987 4988 4989 4990
		chunk_size = mddev->chunk_sectors << 9;
		blk_queue_io_min(mddev->queue, chunk_size);
		blk_queue_io_opt(mddev->queue, chunk_size *
				 (conf->raid_disks - conf->max_degraded));
4991

4992 4993 4994 4995
		list_for_each_entry(rdev, &mddev->disks, same_set)
			disk_stack_limits(mddev->gendisk, rdev->bdev,
					  rdev->data_offset << 9);
	}
4996

L
Linus Torvalds 已提交
4997 4998
	return 0;
abort:
4999
	md_unregister_thread(&mddev->thread);
N
NeilBrown 已提交
5000 5001
	print_raid5_conf(conf);
	free_conf(conf);
L
Linus Torvalds 已提交
5002
	mddev->private = NULL;
5003
	printk(KERN_ALERT "md/raid:%s: failed to run raid set.\n", mdname(mddev));
L
Linus Torvalds 已提交
5004 5005 5006
	return -EIO;
}

5007
static int stop(struct mddev *mddev)
L
Linus Torvalds 已提交
5008
{
5009
	struct r5conf *conf = mddev->private;
L
Linus Torvalds 已提交
5010

5011
	md_unregister_thread(&mddev->thread);
N
NeilBrown 已提交
5012 5013
	if (mddev->queue)
		mddev->queue->backing_dev_info.congested_fn = NULL;
5014
	free_conf(conf);
5015 5016
	mddev->private = NULL;
	mddev->to_remove = &raid5_attrs_group;
L
Linus Torvalds 已提交
5017 5018 5019
	return 0;
}

5020
static void status(struct seq_file *seq, struct mddev *mddev)
L
Linus Torvalds 已提交
5021
{
5022
	struct r5conf *conf = mddev->private;
L
Linus Torvalds 已提交
5023 5024
	int i;

5025 5026
	seq_printf(seq, " level %d, %dk chunk, algorithm %d", mddev->level,
		mddev->chunk_sectors / 2, mddev->layout);
5027
	seq_printf (seq, " [%d/%d] [", conf->raid_disks, conf->raid_disks - mddev->degraded);
L
Linus Torvalds 已提交
5028 5029 5030
	for (i = 0; i < conf->raid_disks; i++)
		seq_printf (seq, "%s",
			       conf->disks[i].rdev &&
5031
			       test_bit(In_sync, &conf->disks[i].rdev->flags) ? "U" : "_");
L
Linus Torvalds 已提交
5032 5033 5034
	seq_printf (seq, "]");
}

5035
static void print_raid5_conf (struct r5conf *conf)
L
Linus Torvalds 已提交
5036 5037 5038 5039
{
	int i;
	struct disk_info *tmp;

5040
	printk(KERN_DEBUG "RAID conf printout:\n");
L
Linus Torvalds 已提交
5041 5042 5043 5044
	if (!conf) {
		printk("(conf==NULL)\n");
		return;
	}
5045 5046 5047
	printk(KERN_DEBUG " --- level:%d rd:%d wd:%d\n", conf->level,
	       conf->raid_disks,
	       conf->raid_disks - conf->mddev->degraded);
L
Linus Torvalds 已提交
5048 5049 5050 5051 5052

	for (i = 0; i < conf->raid_disks; i++) {
		char b[BDEVNAME_SIZE];
		tmp = conf->disks + i;
		if (tmp->rdev)
5053 5054 5055
			printk(KERN_DEBUG " disk %d, o:%d, dev:%s\n",
			       i, !test_bit(Faulty, &tmp->rdev->flags),
			       bdevname(tmp->rdev->bdev, b));
L
Linus Torvalds 已提交
5056 5057 5058
	}
}

5059
static int raid5_spare_active(struct mddev *mddev)
L
Linus Torvalds 已提交
5060 5061
{
	int i;
5062
	struct r5conf *conf = mddev->private;
L
Linus Torvalds 已提交
5063
	struct disk_info *tmp;
5064 5065
	int count = 0;
	unsigned long flags;
L
Linus Torvalds 已提交
5066 5067 5068 5069

	for (i = 0; i < conf->raid_disks; i++) {
		tmp = conf->disks + i;
		if (tmp->rdev
5070
		    && tmp->rdev->recovery_offset == MaxSector
5071
		    && !test_bit(Faulty, &tmp->rdev->flags)
5072
		    && !test_and_set_bit(In_sync, &tmp->rdev->flags)) {
5073
			count++;
5074
			sysfs_notify_dirent_safe(tmp->rdev->sysfs_state);
L
Linus Torvalds 已提交
5075 5076
		}
	}
5077
	spin_lock_irqsave(&conf->device_lock, flags);
5078
	mddev->degraded = calc_degraded(conf);
5079
	spin_unlock_irqrestore(&conf->device_lock, flags);
L
Linus Torvalds 已提交
5080
	print_raid5_conf(conf);
5081
	return count;
L
Linus Torvalds 已提交
5082 5083
}

5084
static int raid5_remove_disk(struct mddev *mddev, struct md_rdev *rdev)
L
Linus Torvalds 已提交
5085
{
5086
	struct r5conf *conf = mddev->private;
L
Linus Torvalds 已提交
5087
	int err = 0;
5088
	int number = rdev->raid_disk;
L
Linus Torvalds 已提交
5089 5090 5091
	struct disk_info *p = conf->disks + number;

	print_raid5_conf(conf);
5092
	if (rdev == p->rdev) {
5093 5094 5095 5096
		if (number >= conf->raid_disks &&
		    conf->reshape_progress == MaxSector)
			clear_bit(In_sync, &rdev->flags);

5097
		if (test_bit(In_sync, &rdev->flags) ||
L
Linus Torvalds 已提交
5098 5099 5100 5101
		    atomic_read(&rdev->nr_pending)) {
			err = -EBUSY;
			goto abort;
		}
5102 5103 5104 5105
		/* Only remove non-faulty devices if recovery
		 * isn't possible.
		 */
		if (!test_bit(Faulty, &rdev->flags) &&
5106
		    mddev->recovery_disabled != conf->recovery_disabled &&
5107
		    !has_failed(conf) &&
5108
		    number < conf->raid_disks) {
5109 5110 5111
			err = -EBUSY;
			goto abort;
		}
L
Linus Torvalds 已提交
5112
		p->rdev = NULL;
5113
		synchronize_rcu();
L
Linus Torvalds 已提交
5114 5115 5116 5117 5118 5119 5120 5121 5122 5123 5124 5125
		if (atomic_read(&rdev->nr_pending)) {
			/* lost the race, try later */
			err = -EBUSY;
			p->rdev = rdev;
		}
	}
abort:

	print_raid5_conf(conf);
	return err;
}

5126
static int raid5_add_disk(struct mddev *mddev, struct md_rdev *rdev)
L
Linus Torvalds 已提交
5127
{
5128
	struct r5conf *conf = mddev->private;
5129
	int err = -EEXIST;
L
Linus Torvalds 已提交
5130 5131
	int disk;
	struct disk_info *p;
5132 5133
	int first = 0;
	int last = conf->raid_disks - 1;
L
Linus Torvalds 已提交
5134

5135 5136 5137
	if (mddev->recovery_disabled == conf->recovery_disabled)
		return -EBUSY;

5138
	if (has_failed(conf))
L
Linus Torvalds 已提交
5139
		/* no point adding a device */
5140
		return -EINVAL;
L
Linus Torvalds 已提交
5141

5142 5143
	if (rdev->raid_disk >= 0)
		first = last = rdev->raid_disk;
L
Linus Torvalds 已提交
5144 5145

	/*
5146 5147
	 * find the disk ... but prefer rdev->saved_raid_disk
	 * if possible.
L
Linus Torvalds 已提交
5148
	 */
5149
	if (rdev->saved_raid_disk >= 0 &&
5150
	    rdev->saved_raid_disk >= first &&
5151 5152 5153
	    conf->disks[rdev->saved_raid_disk].rdev == NULL)
		disk = rdev->saved_raid_disk;
	else
5154 5155
		disk = first;
	for ( ; disk <= last ; disk++)
L
Linus Torvalds 已提交
5156
		if ((p=conf->disks + disk)->rdev == NULL) {
5157
			clear_bit(In_sync, &rdev->flags);
L
Linus Torvalds 已提交
5158
			rdev->raid_disk = disk;
5159
			err = 0;
5160 5161
			if (rdev->saved_raid_disk != disk)
				conf->fullsync = 1;
5162
			rcu_assign_pointer(p->rdev, rdev);
L
Linus Torvalds 已提交
5163 5164 5165
			break;
		}
	print_raid5_conf(conf);
5166
	return err;
L
Linus Torvalds 已提交
5167 5168
}

5169
static int raid5_resize(struct mddev *mddev, sector_t sectors)
L
Linus Torvalds 已提交
5170 5171 5172 5173 5174 5175 5176 5177
{
	/* 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.
	 */
5178
	sectors &= ~((sector_t)mddev->chunk_sectors - 1);
5179 5180
	md_set_array_sectors(mddev, raid5_size(mddev, sectors,
					       mddev->raid_disks));
D
Dan Williams 已提交
5181 5182 5183
	if (mddev->array_sectors >
	    raid5_size(mddev, sectors, mddev->raid_disks))
		return -EINVAL;
5184
	set_capacity(mddev->gendisk, mddev->array_sectors);
5185
	revalidate_disk(mddev->gendisk);
5186 5187
	if (sectors > mddev->dev_sectors &&
	    mddev->recovery_cp > mddev->dev_sectors) {
A
Andre Noll 已提交
5188
		mddev->recovery_cp = mddev->dev_sectors;
L
Linus Torvalds 已提交
5189 5190
		set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
	}
A
Andre Noll 已提交
5191
	mddev->dev_sectors = sectors;
5192
	mddev->resync_max_sectors = sectors;
L
Linus Torvalds 已提交
5193 5194 5195
	return 0;
}

5196
static int check_stripe_cache(struct mddev *mddev)
5197 5198 5199 5200 5201 5202 5203 5204 5205
{
	/* Can only proceed if there are plenty of stripe_heads.
	 * We need a minimum of one full stripe,, and for sensible progress
	 * it is best to have about 4 times that.
	 * If we require 4 times, then the default 256 4K stripe_heads will
	 * allow for chunk sizes up to 256K, which is probably OK.
	 * If the chunk size is greater, user-space should request more
	 * stripe_heads first.
	 */
5206
	struct r5conf *conf = mddev->private;
5207 5208 5209 5210
	if (((mddev->chunk_sectors << 9) / STRIPE_SIZE) * 4
	    > conf->max_nr_stripes ||
	    ((mddev->new_chunk_sectors << 9) / STRIPE_SIZE) * 4
	    > conf->max_nr_stripes) {
5211 5212
		printk(KERN_WARNING "md/raid:%s: reshape: not enough stripes.  Needed %lu\n",
		       mdname(mddev),
5213 5214 5215 5216 5217 5218 5219
		       ((max(mddev->chunk_sectors, mddev->new_chunk_sectors) << 9)
			/ STRIPE_SIZE)*4);
		return 0;
	}
	return 1;
}

5220
static int check_reshape(struct mddev *mddev)
5221
{
5222
	struct r5conf *conf = mddev->private;
5223

5224 5225
	if (mddev->delta_disks == 0 &&
	    mddev->new_layout == mddev->layout &&
5226
	    mddev->new_chunk_sectors == mddev->chunk_sectors)
5227
		return 0; /* nothing to do */
5228 5229 5230
	if (mddev->bitmap)
		/* Cannot grow a bitmap yet */
		return -EBUSY;
5231
	if (has_failed(conf))
5232 5233 5234 5235 5236 5237 5238 5239 5240 5241 5242 5243 5244
		return -EINVAL;
	if (mddev->delta_disks < 0) {
		/* We might be able to shrink, but the devices must
		 * be made bigger first.
		 * For raid6, 4 is the minimum size.
		 * Otherwise 2 is the minimum
		 */
		int min = 2;
		if (mddev->level == 6)
			min = 4;
		if (mddev->raid_disks + mddev->delta_disks < min)
			return -EINVAL;
	}
5245

5246
	if (!check_stripe_cache(mddev))
5247 5248
		return -ENOSPC;

5249
	return resize_stripes(conf, conf->raid_disks + mddev->delta_disks);
5250 5251
}

5252
static int raid5_start_reshape(struct mddev *mddev)
5253
{
5254
	struct r5conf *conf = mddev->private;
5255
	struct md_rdev *rdev;
5256
	int spares = 0;
5257
	unsigned long flags;
5258

5259
	if (test_bit(MD_RECOVERY_RUNNING, &mddev->recovery))
5260 5261
		return -EBUSY;

5262 5263 5264
	if (!check_stripe_cache(mddev))
		return -ENOSPC;

5265
	list_for_each_entry(rdev, &mddev->disks, same_set)
5266 5267
		if (!test_bit(In_sync, &rdev->flags)
		    && !test_bit(Faulty, &rdev->flags))
5268
			spares++;
5269

5270
	if (spares - mddev->degraded < mddev->delta_disks - conf->max_degraded)
5271 5272 5273 5274 5275
		/* Not enough devices even to make a degraded array
		 * of that size
		 */
		return -EINVAL;

5276 5277 5278 5279 5280 5281
	/* Refuse to reduce size of the array.  Any reductions in
	 * array size must be through explicit setting of array_size
	 * attribute.
	 */
	if (raid5_size(mddev, 0, conf->raid_disks + mddev->delta_disks)
	    < mddev->array_sectors) {
5282
		printk(KERN_ERR "md/raid:%s: array size must be reduced "
5283 5284 5285 5286
		       "before number of disks\n", mdname(mddev));
		return -EINVAL;
	}

5287
	atomic_set(&conf->reshape_stripes, 0);
5288 5289
	spin_lock_irq(&conf->device_lock);
	conf->previous_raid_disks = conf->raid_disks;
5290
	conf->raid_disks += mddev->delta_disks;
5291 5292
	conf->prev_chunk_sectors = conf->chunk_sectors;
	conf->chunk_sectors = mddev->new_chunk_sectors;
5293 5294
	conf->prev_algo = conf->algorithm;
	conf->algorithm = mddev->new_layout;
5295 5296 5297 5298 5299
	if (mddev->delta_disks < 0)
		conf->reshape_progress = raid5_size(mddev, 0, 0);
	else
		conf->reshape_progress = 0;
	conf->reshape_safe = conf->reshape_progress;
5300
	conf->generation++;
5301 5302 5303 5304
	spin_unlock_irq(&conf->device_lock);

	/* Add some new drives, as many as will fit.
	 * We know there are enough to make the newly sized array work.
5305 5306 5307 5308
	 * Don't add devices if we are reducing the number of
	 * devices in the array.  This is because it is not possible
	 * to correctly record the "partially reconstructed" state of
	 * such devices during the reshape and confusion could result.
5309
	 */
5310 5311 5312 5313 5314 5315 5316 5317 5318 5319 5320 5321
	if (mddev->delta_disks >= 0) {
		int added_devices = 0;
		list_for_each_entry(rdev, &mddev->disks, same_set)
			if (rdev->raid_disk < 0 &&
			    !test_bit(Faulty, &rdev->flags)) {
				if (raid5_add_disk(mddev, rdev) == 0) {
					if (rdev->raid_disk
					    >= conf->previous_raid_disks) {
						set_bit(In_sync, &rdev->flags);
						added_devices++;
					} else
						rdev->recovery_offset = 0;
5322 5323

					if (sysfs_link_rdev(mddev, rdev))
5324
						/* Failure here is OK */;
5325
				}
5326 5327 5328 5329 5330 5331
			} else if (rdev->raid_disk >= conf->previous_raid_disks
				   && !test_bit(Faulty, &rdev->flags)) {
				/* This is a spare that was manually added */
				set_bit(In_sync, &rdev->flags);
				added_devices++;
			}
5332

5333 5334 5335 5336
		/* When a reshape changes the number of devices,
		 * ->degraded is measured against the larger of the
		 * pre and post number of devices.
		 */
5337
		spin_lock_irqsave(&conf->device_lock, flags);
5338
		mddev->degraded = calc_degraded(conf);
5339 5340
		spin_unlock_irqrestore(&conf->device_lock, flags);
	}
5341
	mddev->raid_disks = conf->raid_disks;
5342
	mddev->reshape_position = conf->reshape_progress;
5343
	set_bit(MD_CHANGE_DEVS, &mddev->flags);
5344

5345 5346 5347 5348 5349
	clear_bit(MD_RECOVERY_SYNC, &mddev->recovery);
	clear_bit(MD_RECOVERY_CHECK, &mddev->recovery);
	set_bit(MD_RECOVERY_RESHAPE, &mddev->recovery);
	set_bit(MD_RECOVERY_RUNNING, &mddev->recovery);
	mddev->sync_thread = md_register_thread(md_do_sync, mddev,
5350
						"reshape");
5351 5352 5353 5354
	if (!mddev->sync_thread) {
		mddev->recovery = 0;
		spin_lock_irq(&conf->device_lock);
		mddev->raid_disks = conf->raid_disks = conf->previous_raid_disks;
5355
		conf->reshape_progress = MaxSector;
5356 5357 5358
		spin_unlock_irq(&conf->device_lock);
		return -EAGAIN;
	}
5359
	conf->reshape_checkpoint = jiffies;
5360 5361 5362 5363 5364
	md_wakeup_thread(mddev->sync_thread);
	md_new_event(mddev);
	return 0;
}

5365 5366 5367
/* This is called from the reshape thread and should make any
 * changes needed in 'conf'
 */
5368
static void end_reshape(struct r5conf *conf)
5369 5370
{

5371 5372 5373
	if (!test_bit(MD_RECOVERY_INTR, &conf->mddev->recovery)) {

		spin_lock_irq(&conf->device_lock);
5374
		conf->previous_raid_disks = conf->raid_disks;
5375
		conf->reshape_progress = MaxSector;
5376
		spin_unlock_irq(&conf->device_lock);
5377
		wake_up(&conf->wait_for_overlap);
5378 5379 5380 5381

		/* read-ahead size must cover two whole stripes, which is
		 * 2 * (datadisks) * chunksize where 'n' is the number of raid devices
		 */
5382
		if (conf->mddev->queue) {
5383
			int data_disks = conf->raid_disks - conf->max_degraded;
5384
			int stripe = data_disks * ((conf->chunk_sectors << 9)
5385
						   / PAGE_SIZE);
5386 5387 5388
			if (conf->mddev->queue->backing_dev_info.ra_pages < 2 * stripe)
				conf->mddev->queue->backing_dev_info.ra_pages = 2 * stripe;
		}
5389 5390 5391
	}
}

5392 5393 5394
/* This is called from the raid5d thread with mddev_lock held.
 * It makes config changes to the device.
 */
5395
static void raid5_finish_reshape(struct mddev *mddev)
5396
{
5397
	struct r5conf *conf = mddev->private;
5398 5399 5400

	if (!test_bit(MD_RECOVERY_INTR, &mddev->recovery)) {

5401 5402 5403
		if (mddev->delta_disks > 0) {
			md_set_array_sectors(mddev, raid5_size(mddev, 0, 0));
			set_capacity(mddev->gendisk, mddev->array_sectors);
5404
			revalidate_disk(mddev->gendisk);
5405 5406
		} else {
			int d;
5407 5408 5409
			spin_lock_irq(&conf->device_lock);
			mddev->degraded = calc_degraded(conf);
			spin_unlock_irq(&conf->device_lock);
5410 5411
			for (d = conf->raid_disks ;
			     d < conf->raid_disks - mddev->delta_disks;
5412
			     d++) {
5413
				struct md_rdev *rdev = conf->disks[d].rdev;
5414 5415
				if (rdev &&
				    raid5_remove_disk(mddev, rdev) == 0) {
5416
					sysfs_unlink_rdev(mddev, rdev);
5417 5418 5419
					rdev->raid_disk = -1;
				}
			}
5420
		}
5421
		mddev->layout = conf->algorithm;
5422
		mddev->chunk_sectors = conf->chunk_sectors;
5423 5424
		mddev->reshape_position = MaxSector;
		mddev->delta_disks = 0;
5425 5426 5427
	}
}

5428
static void raid5_quiesce(struct mddev *mddev, int state)
5429
{
5430
	struct r5conf *conf = mddev->private;
5431 5432

	switch(state) {
5433 5434 5435 5436
	case 2: /* resume for a suspend */
		wake_up(&conf->wait_for_overlap);
		break;

5437 5438
	case 1: /* stop all writes */
		spin_lock_irq(&conf->device_lock);
5439 5440 5441 5442
		/* '2' tells resync/reshape to pause so that all
		 * active stripes can drain
		 */
		conf->quiesce = 2;
5443
		wait_event_lock_irq(conf->wait_for_stripe,
5444 5445
				    atomic_read(&conf->active_stripes) == 0 &&
				    atomic_read(&conf->active_aligned_reads) == 0,
5446
				    conf->device_lock, /* nothing */);
5447
		conf->quiesce = 1;
5448
		spin_unlock_irq(&conf->device_lock);
5449 5450
		/* allow reshape to continue */
		wake_up(&conf->wait_for_overlap);
5451 5452 5453 5454 5455 5456
		break;

	case 0: /* re-enable writes */
		spin_lock_irq(&conf->device_lock);
		conf->quiesce = 0;
		wake_up(&conf->wait_for_stripe);
5457
		wake_up(&conf->wait_for_overlap);
5458 5459 5460 5461
		spin_unlock_irq(&conf->device_lock);
		break;
	}
}
5462

5463

5464
static void *raid45_takeover_raid0(struct mddev *mddev, int level)
5465
{
5466
	struct r0conf *raid0_conf = mddev->private;
5467
	sector_t sectors;
5468

D
Dan Williams 已提交
5469
	/* for raid0 takeover only one zone is supported */
5470
	if (raid0_conf->nr_strip_zones > 1) {
5471 5472
		printk(KERN_ERR "md/raid:%s: cannot takeover raid0 with more than one zone.\n",
		       mdname(mddev));
D
Dan Williams 已提交
5473 5474 5475
		return ERR_PTR(-EINVAL);
	}

5476 5477
	sectors = raid0_conf->strip_zone[0].zone_end;
	sector_div(sectors, raid0_conf->strip_zone[0].nb_dev);
5478
	mddev->dev_sectors = sectors;
D
Dan Williams 已提交
5479
	mddev->new_level = level;
5480 5481 5482 5483 5484 5485 5486 5487 5488 5489 5490
	mddev->new_layout = ALGORITHM_PARITY_N;
	mddev->new_chunk_sectors = mddev->chunk_sectors;
	mddev->raid_disks += 1;
	mddev->delta_disks = 1;
	/* make sure it will be not marked as dirty */
	mddev->recovery_cp = MaxSector;

	return setup_conf(mddev);
}


5491
static void *raid5_takeover_raid1(struct mddev *mddev)
5492 5493 5494 5495 5496 5497 5498 5499 5500 5501 5502 5503 5504 5505 5506 5507 5508 5509 5510 5511 5512
{
	int chunksect;

	if (mddev->raid_disks != 2 ||
	    mddev->degraded > 1)
		return ERR_PTR(-EINVAL);

	/* Should check if there are write-behind devices? */

	chunksect = 64*2; /* 64K by default */

	/* The array must be an exact multiple of chunksize */
	while (chunksect && (mddev->array_sectors & (chunksect-1)))
		chunksect >>= 1;

	if ((chunksect<<9) < STRIPE_SIZE)
		/* array size does not allow a suitable chunk size */
		return ERR_PTR(-EINVAL);

	mddev->new_level = 5;
	mddev->new_layout = ALGORITHM_LEFT_SYMMETRIC;
5513
	mddev->new_chunk_sectors = chunksect;
5514 5515 5516 5517

	return setup_conf(mddev);
}

5518
static void *raid5_takeover_raid6(struct mddev *mddev)
5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541 5542 5543 5544 5545 5546 5547 5548 5549 5550
{
	int new_layout;

	switch (mddev->layout) {
	case ALGORITHM_LEFT_ASYMMETRIC_6:
		new_layout = ALGORITHM_LEFT_ASYMMETRIC;
		break;
	case ALGORITHM_RIGHT_ASYMMETRIC_6:
		new_layout = ALGORITHM_RIGHT_ASYMMETRIC;
		break;
	case ALGORITHM_LEFT_SYMMETRIC_6:
		new_layout = ALGORITHM_LEFT_SYMMETRIC;
		break;
	case ALGORITHM_RIGHT_SYMMETRIC_6:
		new_layout = ALGORITHM_RIGHT_SYMMETRIC;
		break;
	case ALGORITHM_PARITY_0_6:
		new_layout = ALGORITHM_PARITY_0;
		break;
	case ALGORITHM_PARITY_N:
		new_layout = ALGORITHM_PARITY_N;
		break;
	default:
		return ERR_PTR(-EINVAL);
	}
	mddev->new_level = 5;
	mddev->new_layout = new_layout;
	mddev->delta_disks = -1;
	mddev->raid_disks -= 1;
	return setup_conf(mddev);
}

5551

5552
static int raid5_check_reshape(struct mddev *mddev)
5553
{
5554 5555 5556 5557
	/* For a 2-drive array, the layout and chunk size can be changed
	 * immediately as not restriping is needed.
	 * For larger arrays we record the new value - after validation
	 * to be used by a reshape pass.
5558
	 */
5559
	struct r5conf *conf = mddev->private;
5560
	int new_chunk = mddev->new_chunk_sectors;
5561

5562
	if (mddev->new_layout >= 0 && !algorithm_valid_raid5(mddev->new_layout))
5563 5564
		return -EINVAL;
	if (new_chunk > 0) {
5565
		if (!is_power_of_2(new_chunk))
5566
			return -EINVAL;
5567
		if (new_chunk < (PAGE_SIZE>>9))
5568
			return -EINVAL;
5569
		if (mddev->array_sectors & (new_chunk-1))
5570 5571 5572 5573 5574 5575
			/* not factor of array size */
			return -EINVAL;
	}

	/* They look valid */

5576
	if (mddev->raid_disks == 2) {
5577 5578 5579 5580
		/* can make the change immediately */
		if (mddev->new_layout >= 0) {
			conf->algorithm = mddev->new_layout;
			mddev->layout = mddev->new_layout;
5581 5582
		}
		if (new_chunk > 0) {
5583 5584
			conf->chunk_sectors = new_chunk ;
			mddev->chunk_sectors = new_chunk;
5585 5586 5587
		}
		set_bit(MD_CHANGE_DEVS, &mddev->flags);
		md_wakeup_thread(mddev->thread);
5588
	}
5589
	return check_reshape(mddev);
5590 5591
}

5592
static int raid6_check_reshape(struct mddev *mddev)
5593
{
5594
	int new_chunk = mddev->new_chunk_sectors;
5595

5596
	if (mddev->new_layout >= 0 && !algorithm_valid_raid6(mddev->new_layout))
5597
		return -EINVAL;
5598
	if (new_chunk > 0) {
5599
		if (!is_power_of_2(new_chunk))
5600
			return -EINVAL;
5601
		if (new_chunk < (PAGE_SIZE >> 9))
5602
			return -EINVAL;
5603
		if (mddev->array_sectors & (new_chunk-1))
5604 5605
			/* not factor of array size */
			return -EINVAL;
5606
	}
5607 5608

	/* They look valid */
5609
	return check_reshape(mddev);
5610 5611
}

5612
static void *raid5_takeover(struct mddev *mddev)
5613 5614
{
	/* raid5 can take over:
D
Dan Williams 已提交
5615
	 *  raid0 - if there is only one strip zone - make it a raid4 layout
5616 5617 5618 5619
	 *  raid1 - if there are two drives.  We need to know the chunk size
	 *  raid4 - trivial - just use a raid4 layout.
	 *  raid6 - Providing it is a *_6 layout
	 */
D
Dan Williams 已提交
5620 5621
	if (mddev->level == 0)
		return raid45_takeover_raid0(mddev, 5);
5622 5623
	if (mddev->level == 1)
		return raid5_takeover_raid1(mddev);
5624 5625 5626 5627 5628
	if (mddev->level == 4) {
		mddev->new_layout = ALGORITHM_PARITY_N;
		mddev->new_level = 5;
		return setup_conf(mddev);
	}
5629 5630
	if (mddev->level == 6)
		return raid5_takeover_raid6(mddev);
5631 5632 5633 5634

	return ERR_PTR(-EINVAL);
}

5635
static void *raid4_takeover(struct mddev *mddev)
5636
{
D
Dan Williams 已提交
5637 5638 5639
	/* raid4 can take over:
	 *  raid0 - if there is only one strip zone
	 *  raid5 - if layout is right
5640
	 */
D
Dan Williams 已提交
5641 5642
	if (mddev->level == 0)
		return raid45_takeover_raid0(mddev, 4);
5643 5644 5645 5646 5647 5648 5649 5650
	if (mddev->level == 5 &&
	    mddev->layout == ALGORITHM_PARITY_N) {
		mddev->new_layout = 0;
		mddev->new_level = 4;
		return setup_conf(mddev);
	}
	return ERR_PTR(-EINVAL);
}
5651

5652
static struct md_personality raid5_personality;
5653

5654
static void *raid6_takeover(struct mddev *mddev)
5655 5656 5657 5658 5659 5660 5661 5662 5663 5664 5665 5666 5667 5668 5669 5670 5671 5672 5673 5674 5675 5676 5677 5678 5679 5680 5681 5682 5683 5684 5685 5686 5687 5688 5689 5690 5691 5692 5693 5694 5695 5696 5697 5698 5699 5700
{
	/* Currently can only take over a raid5.  We map the
	 * personality to an equivalent raid6 personality
	 * with the Q block at the end.
	 */
	int new_layout;

	if (mddev->pers != &raid5_personality)
		return ERR_PTR(-EINVAL);
	if (mddev->degraded > 1)
		return ERR_PTR(-EINVAL);
	if (mddev->raid_disks > 253)
		return ERR_PTR(-EINVAL);
	if (mddev->raid_disks < 3)
		return ERR_PTR(-EINVAL);

	switch (mddev->layout) {
	case ALGORITHM_LEFT_ASYMMETRIC:
		new_layout = ALGORITHM_LEFT_ASYMMETRIC_6;
		break;
	case ALGORITHM_RIGHT_ASYMMETRIC:
		new_layout = ALGORITHM_RIGHT_ASYMMETRIC_6;
		break;
	case ALGORITHM_LEFT_SYMMETRIC:
		new_layout = ALGORITHM_LEFT_SYMMETRIC_6;
		break;
	case ALGORITHM_RIGHT_SYMMETRIC:
		new_layout = ALGORITHM_RIGHT_SYMMETRIC_6;
		break;
	case ALGORITHM_PARITY_0:
		new_layout = ALGORITHM_PARITY_0_6;
		break;
	case ALGORITHM_PARITY_N:
		new_layout = ALGORITHM_PARITY_N;
		break;
	default:
		return ERR_PTR(-EINVAL);
	}
	mddev->new_level = 6;
	mddev->new_layout = new_layout;
	mddev->delta_disks = 1;
	mddev->raid_disks += 1;
	return setup_conf(mddev);
}


5701
static struct md_personality raid6_personality =
5702 5703 5704 5705 5706 5707 5708 5709 5710 5711 5712 5713 5714 5715
{
	.name		= "raid6",
	.level		= 6,
	.owner		= THIS_MODULE,
	.make_request	= make_request,
	.run		= run,
	.stop		= stop,
	.status		= status,
	.error_handler	= error,
	.hot_add_disk	= raid5_add_disk,
	.hot_remove_disk= raid5_remove_disk,
	.spare_active	= raid5_spare_active,
	.sync_request	= sync_request,
	.resize		= raid5_resize,
5716
	.size		= raid5_size,
5717
	.check_reshape	= raid6_check_reshape,
5718
	.start_reshape  = raid5_start_reshape,
5719
	.finish_reshape = raid5_finish_reshape,
5720
	.quiesce	= raid5_quiesce,
5721
	.takeover	= raid6_takeover,
5722
};
5723
static struct md_personality raid5_personality =
L
Linus Torvalds 已提交
5724 5725
{
	.name		= "raid5",
5726
	.level		= 5,
L
Linus Torvalds 已提交
5727 5728 5729 5730 5731 5732 5733 5734 5735 5736 5737
	.owner		= THIS_MODULE,
	.make_request	= make_request,
	.run		= run,
	.stop		= stop,
	.status		= status,
	.error_handler	= error,
	.hot_add_disk	= raid5_add_disk,
	.hot_remove_disk= raid5_remove_disk,
	.spare_active	= raid5_spare_active,
	.sync_request	= sync_request,
	.resize		= raid5_resize,
5738
	.size		= raid5_size,
5739 5740
	.check_reshape	= raid5_check_reshape,
	.start_reshape  = raid5_start_reshape,
5741
	.finish_reshape = raid5_finish_reshape,
5742
	.quiesce	= raid5_quiesce,
5743
	.takeover	= raid5_takeover,
L
Linus Torvalds 已提交
5744 5745
};

5746
static struct md_personality raid4_personality =
L
Linus Torvalds 已提交
5747
{
5748 5749 5750 5751 5752 5753 5754 5755 5756 5757 5758 5759 5760
	.name		= "raid4",
	.level		= 4,
	.owner		= THIS_MODULE,
	.make_request	= make_request,
	.run		= run,
	.stop		= stop,
	.status		= status,
	.error_handler	= error,
	.hot_add_disk	= raid5_add_disk,
	.hot_remove_disk= raid5_remove_disk,
	.spare_active	= raid5_spare_active,
	.sync_request	= sync_request,
	.resize		= raid5_resize,
5761
	.size		= raid5_size,
5762 5763
	.check_reshape	= raid5_check_reshape,
	.start_reshape  = raid5_start_reshape,
5764
	.finish_reshape = raid5_finish_reshape,
5765
	.quiesce	= raid5_quiesce,
5766
	.takeover	= raid4_takeover,
5767 5768 5769 5770
};

static int __init raid5_init(void)
{
5771
	register_md_personality(&raid6_personality);
5772 5773 5774
	register_md_personality(&raid5_personality);
	register_md_personality(&raid4_personality);
	return 0;
L
Linus Torvalds 已提交
5775 5776
}

5777
static void raid5_exit(void)
L
Linus Torvalds 已提交
5778
{
5779
	unregister_md_personality(&raid6_personality);
5780 5781
	unregister_md_personality(&raid5_personality);
	unregister_md_personality(&raid4_personality);
L
Linus Torvalds 已提交
5782 5783 5784 5785 5786
}

module_init(raid5_init);
module_exit(raid5_exit);
MODULE_LICENSE("GPL");
5787
MODULE_DESCRIPTION("RAID4/5/6 (striping with parity) personality for MD");
L
Linus Torvalds 已提交
5788
MODULE_ALIAS("md-personality-4"); /* RAID5 */
5789 5790
MODULE_ALIAS("md-raid5");
MODULE_ALIAS("md-raid4");
5791 5792
MODULE_ALIAS("md-level-5");
MODULE_ALIAS("md-level-4");
5793 5794 5795 5796 5797 5798 5799
MODULE_ALIAS("md-personality-8"); /* RAID6 */
MODULE_ALIAS("md-raid6");
MODULE_ALIAS("md-level-6");

/* This used to be two separate modules, they were: */
MODULE_ALIAS("raid5");
MODULE_ALIAS("raid6");