raid1.c 60.8 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
1
2
3
4
5
6
7
8
9
10
11
/*
 * raid1.c : Multiple Devices driver for Linux
 *
 * Copyright (C) 1999, 2000, 2001 Ingo Molnar, Red Hat
 *
 * Copyright (C) 1996, 1997, 1998 Ingo Molnar, Miguel de Icaza, Gadi Oxman
 *
 * RAID-1 management functions.
 *
 * Better read-balancing code written by Mika Kuoppala <miku@iki.fi>, 2000
 *
12
 * Fixes to reconstruction by Jakob Østergaard" <jakob@ostenfeld.dk>
Linus Torvalds's avatar
Linus Torvalds committed
13
14
 * Various fixes by Neil Brown <neilb@cse.unsw.edu.au>
 *
15
16
17
18
19
20
21
22
23
 * Changes by Peter T. Breuer <ptb@it.uc3m.es> 31/1/2003 to support
 * bitmapped intelligence in resync:
 *
 *      - bitmap marked during normal i/o
 *      - bitmap used to skip nondirty blocks during sync
 *
 * Additions to bitmap code, (C) 2003-2004 Paul Clements, SteelEye Technology:
 * - persistent bitmap code
 *
Linus Torvalds's avatar
Linus Torvalds committed
24
25
26
27
28
29
30
31
32
33
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2, or (at your option)
 * any later version.
 *
 * You should have received a copy of the GNU General Public License
 * (for example /usr/src/linux/COPYING); if not, write to the Free
 * Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 */

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

#define DEBUG 0
#if DEBUG
#define PRINTK(x...) printk(x)
#else
#define PRINTK(x...)
#endif
Linus Torvalds's avatar
Linus Torvalds committed
47
48
49
50
51
52
53
54
55

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


static void unplug_slaves(mddev_t *mddev);

56
57
static void allow_barrier(conf_t *conf);
static void lower_barrier(conf_t *conf);
Linus Torvalds's avatar
Linus Torvalds committed
58

59
static void * r1bio_pool_alloc(gfp_t gfp_flags, void *data)
Linus Torvalds's avatar
Linus Torvalds committed
60
61
62
63
64
65
{
	struct pool_info *pi = data;
	r1bio_t *r1_bio;
	int size = offsetof(r1bio_t, bios[pi->raid_disks]);

	/* allocate a r1bio with room for raid_disks entries in the bios array */
66
	r1_bio = kzalloc(size, gfp_flags);
67
	if (!r1_bio && pi->mddev)
Linus Torvalds's avatar
Linus Torvalds committed
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
		unplug_slaves(pi->mddev);

	return r1_bio;
}

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

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

84
static void * r1buf_pool_alloc(gfp_t gfp_flags, void *data)
Linus Torvalds's avatar
Linus Torvalds committed
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
{
	struct pool_info *pi = data;
	struct page *page;
	r1bio_t *r1_bio;
	struct bio *bio;
	int i, j;

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

	/*
	 * Allocate bios : 1 for reading, n-1 for writing
	 */
	for (j = pi->raid_disks ; j-- ; ) {
		bio = bio_alloc(gfp_flags, RESYNC_PAGES);
		if (!bio)
			goto out_free_bio;
		r1_bio->bios[j] = bio;
	}
	/*
	 * Allocate RESYNC_PAGES data pages and attach them to
109
110
111
	 * the first bio.
	 * If this is a user-requested check/repair, allocate
	 * RESYNC_PAGES for each bio.
Linus Torvalds's avatar
Linus Torvalds committed
112
	 */
113
114
115
116
117
118
119
120
121
122
123
124
	if (test_bit(MD_RECOVERY_REQUESTED, &pi->mddev->recovery))
		j = pi->raid_disks;
	else
		j = 1;
	while(j--) {
		bio = r1_bio->bios[j];
		for (i = 0; i < RESYNC_PAGES; i++) {
			page = alloc_page(gfp_flags);
			if (unlikely(!page))
				goto out_free_pages;

			bio->bi_io_vec[i].bv_page = page;
125
			bio->bi_vcnt = i+1;
126
127
128
129
130
131
132
133
		}
	}
	/* If not user-requests, copy the page pointers to all bios */
	if (!test_bit(MD_RECOVERY_REQUESTED, &pi->mddev->recovery)) {
		for (i=0; i<RESYNC_PAGES ; i++)
			for (j=1; j<pi->raid_disks; j++)
				r1_bio->bios[j]->bi_io_vec[i].bv_page =
					r1_bio->bios[0]->bi_io_vec[i].bv_page;
Linus Torvalds's avatar
Linus Torvalds committed
134
135
136
137
138
139
140
	}

	r1_bio->master_bio = NULL;

	return r1_bio;

out_free_pages:
141
142
143
	for (j=0 ; j < pi->raid_disks; j++)
		for (i=0; i < r1_bio->bios[j]->bi_vcnt ; i++)
			put_page(r1_bio->bios[j]->bi_io_vec[i].bv_page);
144
	j = -1;
Linus Torvalds's avatar
Linus Torvalds committed
145
146
147
148
149
150
151
152
153
154
out_free_bio:
	while ( ++j < pi->raid_disks )
		bio_put(r1_bio->bios[j]);
	r1bio_pool_free(r1_bio, data);
	return NULL;
}

static void r1buf_pool_free(void *__r1_bio, void *data)
{
	struct pool_info *pi = data;
155
	int i,j;
Linus Torvalds's avatar
Linus Torvalds committed
156
157
	r1bio_t *r1bio = __r1_bio;

158
159
160
161
162
	for (i = 0; i < RESYNC_PAGES; i++)
		for (j = pi->raid_disks; j-- ;) {
			if (j == 0 ||
			    r1bio->bios[j]->bi_io_vec[i].bv_page !=
			    r1bio->bios[0]->bi_io_vec[i].bv_page)
163
				safe_put_page(r1bio->bios[j]->bi_io_vec[i].bv_page);
164
		}
Linus Torvalds's avatar
Linus Torvalds committed
165
166
167
168
169
170
171
172
173
174
175
176
	for (i=0 ; i < pi->raid_disks; i++)
		bio_put(r1bio->bios[i]);

	r1bio_pool_free(r1bio, data);
}

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

	for (i = 0; i < conf->raid_disks; i++) {
		struct bio **bio = r1_bio->bios + i;
177
		if (*bio && *bio != IO_BLOCKED)
Linus Torvalds's avatar
Linus Torvalds committed
178
179
180
181
182
			bio_put(*bio);
		*bio = NULL;
	}
}

183
static void free_r1bio(r1bio_t *r1_bio)
Linus Torvalds's avatar
Linus Torvalds committed
184
{
185
	conf_t *conf = r1_bio->mddev->private;
Linus Torvalds's avatar
Linus Torvalds committed
186
187
188
189
190

	/*
	 * Wake up any possible resync thread that waits for the device
	 * to go idle.
	 */
191
	allow_barrier(conf);
Linus Torvalds's avatar
Linus Torvalds committed
192
193
194
195
196

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

197
static void put_buf(r1bio_t *r1_bio)
Linus Torvalds's avatar
Linus Torvalds committed
198
{
199
	conf_t *conf = r1_bio->mddev->private;
200
201
202
203
204
205
206
	int i;

	for (i=0; i<conf->raid_disks; i++) {
		struct bio *bio = r1_bio->bios[i];
		if (bio->bi_end_io)
			rdev_dec_pending(conf->mirrors[i].rdev, r1_bio->mddev);
	}
Linus Torvalds's avatar
Linus Torvalds committed
207
208
209

	mempool_free(r1_bio, conf->r1buf_pool);

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

static void reschedule_retry(r1bio_t *r1_bio)
{
	unsigned long flags;
	mddev_t *mddev = r1_bio->mddev;
217
	conf_t *conf = mddev->private;
Linus Torvalds's avatar
Linus Torvalds committed
218
219
220

	spin_lock_irqsave(&conf->device_lock, flags);
	list_add(&r1_bio->retry_list, &conf->retry_list);
221
	conf->nr_queued ++;
Linus Torvalds's avatar
Linus Torvalds committed
222
223
	spin_unlock_irqrestore(&conf->device_lock, flags);

224
	wake_up(&conf->wait_barrier);
Linus Torvalds's avatar
Linus Torvalds committed
225
226
227
228
229
230
231
232
233
234
235
236
	md_wakeup_thread(mddev->thread);
}

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

237
238
239
240
241
242
243
244
	/* if nobody has done the final endio yet, do it now */
	if (!test_and_set_bit(R1BIO_Returned, &r1_bio->state)) {
		PRINTK(KERN_DEBUG "raid1: sync end %s on sectors %llu-%llu\n",
			(bio_data_dir(bio) == WRITE) ? "write" : "read",
			(unsigned long long) bio->bi_sector,
			(unsigned long long) bio->bi_sector +
				(bio->bi_size >> 9) - 1);

245
		bio_endio(bio,
246
247
			test_bit(R1BIO_Uptodate, &r1_bio->state) ? 0 : -EIO);
	}
Linus Torvalds's avatar
Linus Torvalds committed
248
249
250
251
252
253
254
255
	free_r1bio(r1_bio);
}

/*
 * Update disk head position estimator based on IRQ completion info.
 */
static inline void update_head_pos(int disk, r1bio_t *r1_bio)
{
256
	conf_t *conf = r1_bio->mddev->private;
Linus Torvalds's avatar
Linus Torvalds committed
257
258
259
260
261

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

262
static void raid1_end_read_request(struct bio *bio, int error)
Linus Torvalds's avatar
Linus Torvalds committed
263
264
265
266
{
	int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
	r1bio_t * r1_bio = (r1bio_t *)(bio->bi_private);
	int mirror;
267
	conf_t *conf = r1_bio->mddev->private;
Linus Torvalds's avatar
Linus Torvalds committed
268
269
270
271
272

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

275
276
277
278
279
280
	if (uptodate)
		set_bit(R1BIO_Uptodate, &r1_bio->state);
	else {
		/* If all other devices have failed, we want to return
		 * the error upwards rather than fail the last device.
		 * Here we redefine "uptodate" to mean "Don't want to retry"
Linus Torvalds's avatar
Linus Torvalds committed
281
		 */
282
283
284
285
286
287
288
289
		unsigned long flags;
		spin_lock_irqsave(&conf->device_lock, flags);
		if (r1_bio->mddev->degraded == conf->raid_disks ||
		    (r1_bio->mddev->degraded == conf->raid_disks-1 &&
		     !test_bit(Faulty, &conf->mirrors[mirror].rdev->flags)))
			uptodate = 1;
		spin_unlock_irqrestore(&conf->device_lock, flags);
	}
Linus Torvalds's avatar
Linus Torvalds committed
290

291
	if (uptodate)
Linus Torvalds's avatar
Linus Torvalds committed
292
		raid_end_bio_io(r1_bio);
293
	else {
Linus Torvalds's avatar
Linus Torvalds committed
294
295
296
297
298
299
300
301
302
303
304
305
306
		/*
		 * oops, read error:
		 */
		char b[BDEVNAME_SIZE];
		if (printk_ratelimit())
			printk(KERN_ERR "raid1: %s: rescheduling sector %llu\n",
			       bdevname(conf->mirrors[mirror].rdev->bdev,b), (unsigned long long)r1_bio->sector);
		reschedule_retry(r1_bio);
	}

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

307
static void raid1_end_write_request(struct bio *bio, int error)
Linus Torvalds's avatar
Linus Torvalds committed
308
309
310
{
	int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
	r1bio_t * r1_bio = (r1bio_t *)(bio->bi_private);
311
	int mirror, behind = test_bit(R1BIO_BehindIO, &r1_bio->state);
312
	conf_t *conf = r1_bio->mddev->private;
313
	struct bio *to_put = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
314
315
316
317
318
319


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

320
	if (error == -EOPNOTSUPP && test_bit(R1BIO_Barrier, &r1_bio->state)) {
321
322
323
		set_bit(BarriersNotsupp, &conf->mirrors[mirror].rdev->flags);
		set_bit(R1BIO_BarrierRetry, &r1_bio->state);
		r1_bio->mddev->barriers_work = 0;
324
		/* Don't rdev_dec_pending in this branch - keep it for the retry */
325
	} else {
Linus Torvalds's avatar
Linus Torvalds committed
326
		/*
327
		 * this branch is our 'one mirror IO has finished' event handler:
Linus Torvalds's avatar
Linus Torvalds committed
328
		 */
329
		r1_bio->bios[mirror] = NULL;
330
		to_put = bio;
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
		if (!uptodate) {
			md_error(r1_bio->mddev, conf->mirrors[mirror].rdev);
			/* an I/O failed, we can't clear the bitmap */
			set_bit(R1BIO_Degraded, &r1_bio->state);
		} else
			/*
			 * Set R1BIO_Uptodate in our master bio, so that
			 * we will return a good error code for to the higher
			 * levels even if IO on some other mirrored buffer fails.
			 *
			 * The 'master' represents the composite IO operation to
			 * user-side. So if something waits for IO, then it will
			 * wait for the 'master' bio.
			 */
			set_bit(R1BIO_Uptodate, &r1_bio->state);

		update_head_pos(mirror, r1_bio);

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

			/* In behind mode, we ACK the master bio once the I/O has safely
			 * reached all non-writemostly disks. Setting the Returned bit
			 * ensures that this gets done only once -- we don't ever want to
			 * return -EIO here, instead we'll wait */

			if (atomic_read(&r1_bio->behind_remaining) >= (atomic_read(&r1_bio->remaining)-1) &&
			    test_bit(R1BIO_Uptodate, &r1_bio->state)) {
				/* Maybe we can return now */
				if (!test_and_set_bit(R1BIO_Returned, &r1_bio->state)) {
					struct bio *mbio = r1_bio->master_bio;
					PRINTK(KERN_DEBUG "raid1: behind end write sectors %llu-%llu\n",
					       (unsigned long long) mbio->bi_sector,
					       (unsigned long long) mbio->bi_sector +
					       (mbio->bi_size >> 9) - 1);
367
					bio_endio(mbio, 0);
368
				}
369
370
			}
		}
371
		rdev_dec_pending(conf->mirrors[mirror].rdev, conf->mddev);
372
	}
Linus Torvalds's avatar
Linus Torvalds committed
373
374
375
376
377
378
	/*
	 *
	 * Let's see if all mirrored write operations have finished
	 * already.
	 */
	if (atomic_dec_and_test(&r1_bio->remaining)) {
379
		if (test_bit(R1BIO_BarrierRetry, &r1_bio->state))
380
			reschedule_retry(r1_bio);
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
		else {
			/* it really is the end of this request */
			if (test_bit(R1BIO_BehindIO, &r1_bio->state)) {
				/* free extra copy of the data pages */
				int i = bio->bi_vcnt;
				while (i--)
					safe_put_page(bio->bi_io_vec[i].bv_page);
			}
			/* clear the bitmap if all writes complete successfully */
			bitmap_endwrite(r1_bio->mddev->bitmap, r1_bio->sector,
					r1_bio->sectors,
					!test_bit(R1BIO_Degraded, &r1_bio->state),
					behind);
			md_write_end(r1_bio->mddev);
			raid_end_bio_io(r1_bio);
396
		}
Linus Torvalds's avatar
Linus Torvalds committed
397
	}
398

399
400
	if (to_put)
		bio_put(to_put);
Linus Torvalds's avatar
Linus Torvalds committed
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
}


/*
 * This routine returns the disk from which the requested read should
 * be done. There is a per-array 'next expected sequential IO' sector
 * number - if this matches on the next IO then we use the last disk.
 * There is also a per-disk 'last know head position' sector that is
 * maintained from IRQ contexts, both the normal and the resync IO
 * completion handlers update this position correctly. If there is no
 * perfect sequential match then we pick the disk whose head is closest.
 *
 * If there are 2 mirrors in the same 2 devices, performance degrades
 * because position is mirror, not device based.
 *
 * The rdev for the device selected will have nr_pending incremented.
 */
static int read_balance(conf_t *conf, r1bio_t *r1_bio)
{
	const unsigned long this_sector = r1_bio->sector;
	int new_disk = conf->last_used, disk = new_disk;
422
	int wonly_disk = -1;
Linus Torvalds's avatar
Linus Torvalds committed
423
424
	const int sectors = r1_bio->sectors;
	sector_t new_distance, current_distance;
425
	mdk_rdev_t *rdev;
Linus Torvalds's avatar
Linus Torvalds committed
426
427
428

	rcu_read_lock();
	/*
429
	 * Check if we can balance. We can balance on the whole
Linus Torvalds's avatar
Linus Torvalds committed
430
431
432
433
434
435
436
437
438
	 * device if no resync is going on, or below the resync window.
	 * We take the first readable disk when above the resync window.
	 */
 retry:
	if (conf->mddev->recovery_cp < MaxSector &&
	    (this_sector + sectors >= conf->next_resync)) {
		/* Choose the first operation device, for consistancy */
		new_disk = 0;

439
		for (rdev = rcu_dereference(conf->mirrors[new_disk].rdev);
440
		     r1_bio->bios[new_disk] == IO_BLOCKED ||
441
		     !rdev || !test_bit(In_sync, &rdev->flags)
442
			     || test_bit(WriteMostly, &rdev->flags);
443
		     rdev = rcu_dereference(conf->mirrors[++new_disk].rdev)) {
444

445
446
			if (rdev && test_bit(In_sync, &rdev->flags) &&
				r1_bio->bios[new_disk] != IO_BLOCKED)
447
448
449
450
				wonly_disk = new_disk;

			if (new_disk == conf->raid_disks - 1) {
				new_disk = wonly_disk;
Linus Torvalds's avatar
Linus Torvalds committed
451
452
453
454
455
456
457
458
				break;
			}
		}
		goto rb_out;
	}


	/* make sure the disk is operational */
459
	for (rdev = rcu_dereference(conf->mirrors[new_disk].rdev);
460
	     r1_bio->bios[new_disk] == IO_BLOCKED ||
461
	     !rdev || !test_bit(In_sync, &rdev->flags) ||
462
		     test_bit(WriteMostly, &rdev->flags);
463
	     rdev = rcu_dereference(conf->mirrors[new_disk].rdev)) {
464

465
466
		if (rdev && test_bit(In_sync, &rdev->flags) &&
		    r1_bio->bios[new_disk] != IO_BLOCKED)
467
468
			wonly_disk = new_disk;

Linus Torvalds's avatar
Linus Torvalds committed
469
470
471
472
		if (new_disk <= 0)
			new_disk = conf->raid_disks;
		new_disk--;
		if (new_disk == disk) {
473
474
			new_disk = wonly_disk;
			break;
Linus Torvalds's avatar
Linus Torvalds committed
475
476
		}
	}
477
478
479
480

	if (new_disk < 0)
		goto rb_out;

Linus Torvalds's avatar
Linus Torvalds committed
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
	disk = new_disk;
	/* now disk == new_disk == starting point for search */

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

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

	/* Find the disk whose head is closest */

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

501
		rdev = rcu_dereference(conf->mirrors[disk].rdev);
502

503
		if (!rdev || r1_bio->bios[disk] == IO_BLOCKED ||
504
		    !test_bit(In_sync, &rdev->flags) ||
505
		    test_bit(WriteMostly, &rdev->flags))
Linus Torvalds's avatar
Linus Torvalds committed
506
507
508
509
510
511
512
513
514
515
516
517
518
			continue;

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

519
 rb_out:
Linus Torvalds's avatar
Linus Torvalds committed
520
521
522


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

	return new_disk;
}

static void unplug_slaves(mddev_t *mddev)
{
544
	conf_t *conf = mddev->private;
Linus Torvalds's avatar
Linus Torvalds committed
545
546
547
548
	int i;

	rcu_read_lock();
	for (i=0; i<mddev->raid_disks; i++) {
549
		mdk_rdev_t *rdev = rcu_dereference(conf->mirrors[i].rdev);
550
		if (rdev && !test_bit(Faulty, &rdev->flags) && atomic_read(&rdev->nr_pending)) {
551
			struct request_queue *r_queue = bdev_get_queue(rdev->bdev);
Linus Torvalds's avatar
Linus Torvalds committed
552
553
554
555

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

556
			blk_unplug(r_queue);
Linus Torvalds's avatar
Linus Torvalds committed
557
558
559
560
561
562
563
564

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

565
static void raid1_unplug(struct request_queue *q)
Linus Torvalds's avatar
Linus Torvalds committed
566
{
567
568
569
570
	mddev_t *mddev = q->queuedata;

	unplug_slaves(mddev);
	md_wakeup_thread(mddev->thread);
Linus Torvalds's avatar
Linus Torvalds committed
571
572
}

573
574
575
static int raid1_congested(void *data, int bits)
{
	mddev_t *mddev = data;
576
	conf_t *conf = mddev->private;
577
578
	int i, ret = 0;

579
580
581
	if (mddev_congested(mddev, bits))
		return 1;

582
583
584
585
	rcu_read_lock();
	for (i = 0; i < mddev->raid_disks; i++) {
		mdk_rdev_t *rdev = rcu_dereference(conf->mirrors[i].rdev);
		if (rdev && !test_bit(Faulty, &rdev->flags)) {
586
			struct request_queue *q = bdev_get_queue(rdev->bdev);
587
588
589
590

			/* Note the '|| 1' - when read_balance prefers
			 * non-congested targets, it can be removed
			 */
591
			if ((bits & (1<<BDI_async_congested)) || 1)
592
593
594
595
596
597
598
599
600
601
				ret |= bdi_congested(&q->backing_dev_info, bits);
			else
				ret &= bdi_congested(&q->backing_dev_info, bits);
		}
	}
	rcu_read_unlock();
	return ret;
}


602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
static int flush_pending_writes(conf_t *conf)
{
	/* Any writes that have been queued but are awaiting
	 * bitmap updates get flushed here.
	 * We return 1 if any requests were actually submitted.
	 */
	int rv = 0;

	spin_lock_irq(&conf->device_lock);

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

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

633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
/* Barriers....
 * Sometimes we need to suspend IO while we do something else,
 * either some resync/recovery, or reconfigure the array.
 * To do this we raise a 'barrier'.
 * The 'barrier' is a counter that can be raised multiple times
 * to count how many activities are happening which preclude
 * normal IO.
 * We can only raise the barrier if there is no pending IO.
 * i.e. if nr_pending == 0.
 * We choose only to raise the barrier if no-one is waiting for the
 * barrier to go down.  This means that as soon as an IO request
 * is ready, no other operations which require a barrier will start
 * until the IO request has had a chance.
 *
 * So: regular IO calls 'wait_barrier'.  When that returns there
 *    is no backgroup IO happening,  It must arrange to call
 *    allow_barrier when it has finished its IO.
 * backgroup IO calls must call raise_barrier.  Once that returns
 *    there is no normal IO happeing.  It must arrange to call
 *    lower_barrier when the particular background IO completes.
Linus Torvalds's avatar
Linus Torvalds committed
653
654
655
 */
#define RESYNC_DEPTH 32

656
static void raise_barrier(conf_t *conf)
Linus Torvalds's avatar
Linus Torvalds committed
657
658
{
	spin_lock_irq(&conf->resync_lock);
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694

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

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

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

	spin_unlock_irq(&conf->resync_lock);
}

static void lower_barrier(conf_t *conf)
{
	unsigned long flags;
	spin_lock_irqsave(&conf->resync_lock, flags);
	conf->barrier--;
	spin_unlock_irqrestore(&conf->resync_lock, flags);
	wake_up(&conf->wait_barrier);
}

static void wait_barrier(conf_t *conf)
{
	spin_lock_irq(&conf->resync_lock);
	if (conf->barrier) {
		conf->nr_waiting++;
		wait_event_lock_irq(conf->wait_barrier, !conf->barrier,
				    conf->resync_lock,
				    raid1_unplug(conf->mddev->queue));
		conf->nr_waiting--;
Linus Torvalds's avatar
Linus Torvalds committed
695
	}
696
	conf->nr_pending++;
Linus Torvalds's avatar
Linus Torvalds committed
697
698
699
	spin_unlock_irq(&conf->resync_lock);
}

700
701
702
703
704
705
706
707
708
static void allow_barrier(conf_t *conf)
{
	unsigned long flags;
	spin_lock_irqsave(&conf->resync_lock, flags);
	conf->nr_pending--;
	spin_unlock_irqrestore(&conf->resync_lock, flags);
	wake_up(&conf->wait_barrier);
}

709
710
711
712
713
static void freeze_array(conf_t *conf)
{
	/* stop syncio and normal IO and wait for everything to
	 * go quite.
	 * We increment barrier and nr_waiting, and then
714
715
716
717
718
719
720
721
	 * wait until nr_pending match nr_queued+1
	 * This is called in the context of one normal IO request
	 * that has failed. Thus any sync request that might be pending
	 * will be blocked by nr_pending, and we need to wait for
	 * pending IO requests to complete or be queued for re-try.
	 * Thus the number queued (nr_queued) plus this request (1)
	 * must match the number of pending IOs (nr_pending) before
	 * we continue.
722
723
724
725
726
	 */
	spin_lock_irq(&conf->resync_lock);
	conf->barrier++;
	conf->nr_waiting++;
	wait_event_lock_irq(conf->wait_barrier,
727
			    conf->nr_pending == conf->nr_queued+1,
728
			    conf->resync_lock,
729
730
			    ({ flush_pending_writes(conf);
			       raid1_unplug(conf->mddev->queue); }));
731
732
733
734
735
736
737
738
739
740
741
742
	spin_unlock_irq(&conf->resync_lock);
}
static void unfreeze_array(conf_t *conf)
{
	/* reverse the effect of the freeze */
	spin_lock_irq(&conf->resync_lock);
	conf->barrier--;
	conf->nr_waiting--;
	wake_up(&conf->wait_barrier);
	spin_unlock_irq(&conf->resync_lock);
}

743

744
745
746
747
748
/* duplicate the data pages for behind I/O */
static struct page **alloc_behind_pages(struct bio *bio)
{
	int i;
	struct bio_vec *bvec;
749
	struct page **pages = kzalloc(bio->bi_vcnt * sizeof(struct page *),
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
					GFP_NOIO);
	if (unlikely(!pages))
		goto do_sync_io;

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

	return pages;

do_sync_io:
	if (pages)
		for (i = 0; i < bio->bi_vcnt && pages[i]; i++)
769
			put_page(pages[i]);
770
771
772
773
774
	kfree(pages);
	PRINTK("%dB behind alloc failed, doing sync I/O\n", bio->bi_size);
	return NULL;
}

775
static int make_request(struct request_queue *q, struct bio * bio)
Linus Torvalds's avatar
Linus Torvalds committed
776
777
{
	mddev_t *mddev = q->queuedata;
778
	conf_t *conf = mddev->private;
Linus Torvalds's avatar
Linus Torvalds committed
779
780
781
	mirror_info_t *mirror;
	r1bio_t *r1_bio;
	struct bio *read_bio;
782
	int i, targets = 0, disks;
783
	struct bitmap *bitmap;
784
785
	unsigned long flags;
	struct bio_list bl;
786
	struct page **behind_pages = NULL;
787
	const int rw = bio_data_dir(bio);
788
789
790
	const bool do_sync = bio_rw_flagged(bio, BIO_RW_SYNCIO);
	int cpu;
	bool do_barriers;
791
	mdk_rdev_t *blocked_rdev;
792

Linus Torvalds's avatar
Linus Torvalds committed
793
794
795
796
	/*
	 * Register the new request and wait if the reconstruction
	 * thread has put up a bar for new requests.
	 * Continue immediately if no resync is active currently.
797
798
799
	 * We test barriers_work *after* md_write_start as md_write_start
	 * may cause the first superblock write, and that will check out
	 * if barriers work.
Linus Torvalds's avatar
Linus Torvalds committed
800
	 */
801

802
803
	md_write_start(mddev, bio); /* wait on superblock update early */

804
805
	if (unlikely(!mddev->barriers_work &&
		     bio_rw_flagged(bio, BIO_RW_BARRIER))) {
806
807
		if (rw == WRITE)
			md_write_end(mddev);
808
		bio_endio(bio, -EOPNOTSUPP);
809
810
811
		return 0;
	}

812
	wait_barrier(conf);
Linus Torvalds's avatar
Linus Torvalds committed
813

814
815
	bitmap = mddev->bitmap;

816
817
818
819
820
	cpu = part_stat_lock();
	part_stat_inc(cpu, &mddev->gendisk->part0, ios[rw]);
	part_stat_add(cpu, &mddev->gendisk->part0, sectors[rw],
		      bio_sectors(bio));
	part_stat_unlock();
Linus Torvalds's avatar
Linus Torvalds committed
821
822
823
824
825
826
827
828
829
830

	/*
	 * make_request() can abort the operation when READA is being
	 * used and no empty request is available.
	 *
	 */
	r1_bio = mempool_alloc(conf->r1bio_pool, GFP_NOIO);

	r1_bio->master_bio = bio;
	r1_bio->sectors = bio->bi_size >> 9;
831
	r1_bio->state = 0;
Linus Torvalds's avatar
Linus Torvalds committed
832
833
834
	r1_bio->mddev = mddev;
	r1_bio->sector = bio->bi_sector;

835
	if (rw == READ) {
Linus Torvalds's avatar
Linus Torvalds committed
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
		/*
		 * read balancing logic:
		 */
		int rdisk = read_balance(conf, r1_bio);

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

		r1_bio->read_disk = rdisk;

		read_bio = bio_clone(bio, GFP_NOIO);

		r1_bio->bios[rdisk] = read_bio;

		read_bio->bi_sector = r1_bio->sector + mirror->rdev->data_offset;
		read_bio->bi_bdev = mirror->rdev->bdev;
		read_bio->bi_end_io = raid1_end_read_request;
857
		read_bio->bi_rw = READ | (do_sync << BIO_RW_SYNCIO);
Linus Torvalds's avatar
Linus Torvalds committed
858
859
860
861
862
863
864
865
866
867
868
869
870
871
		read_bio->bi_private = r1_bio;

		generic_make_request(read_bio);
		return 0;
	}

	/*
	 * WRITE:
	 */
	/* first select target devices under spinlock and
	 * inc refcount on their rdev.  Record them by setting
	 * bios[x] to bio
	 */
	disks = conf->raid_disks;
872
873
874
875
876
877
878
#if 0
	{ static int first=1;
	if (first) printk("First Write sector %llu disks %d\n",
			  (unsigned long long)r1_bio->sector, disks);
	first = 0;
	}
#endif
879
880
 retry_write:
	blocked_rdev = NULL;
Linus Torvalds's avatar
Linus Torvalds committed
881
882
	rcu_read_lock();
	for (i = 0;  i < disks; i++) {
883
884
885
886
887
888
889
		mdk_rdev_t *rdev = rcu_dereference(conf->mirrors[i].rdev);
		if (rdev && unlikely(test_bit(Blocked, &rdev->flags))) {
			atomic_inc(&rdev->nr_pending);
			blocked_rdev = rdev;
			break;
		}
		if (rdev && !test_bit(Faulty, &rdev->flags)) {
Linus Torvalds's avatar
Linus Torvalds committed
890
			atomic_inc(&rdev->nr_pending);
891
			if (test_bit(Faulty, &rdev->flags)) {
892
				rdev_dec_pending(rdev, mddev);
Linus Torvalds's avatar
Linus Torvalds committed
893
894
895
				r1_bio->bios[i] = NULL;
			} else
				r1_bio->bios[i] = bio;
896
			targets++;
Linus Torvalds's avatar
Linus Torvalds committed
897
898
899
900
901
		} else
			r1_bio->bios[i] = NULL;
	}
	rcu_read_unlock();

902
903
904
905
906
907
908
909
910
911
912
913
914
915
	if (unlikely(blocked_rdev)) {
		/* Wait for this device to become unblocked */
		int j;

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

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

916
917
	BUG_ON(targets == 0); /* we never fail the last device */

918
919
920
921
922
923
	if (targets < conf->raid_disks) {
		/* array is degraded, we will not clear the bitmap
		 * on I/O completion (see raid1_end_write_request) */
		set_bit(R1BIO_Degraded, &r1_bio->state);
	}

924
925
926
927
928
929
	/* do behind I/O ? */
	if (bitmap &&
	    atomic_read(&bitmap->behind_writes) < bitmap->max_write_behind &&
	    (behind_pages = alloc_behind_pages(bio)) != NULL)
		set_bit(R1BIO_BehindIO, &r1_bio->state);

930
	atomic_set(&r1_bio->remaining, 0);
931
	atomic_set(&r1_bio->behind_remaining, 0);
932

933
	do_barriers = bio_rw_flagged(bio, BIO_RW_BARRIER);
934
935
936
	if (do_barriers)
		set_bit(R1BIO_Barrier, &r1_bio->state);

937
	bio_list_init(&bl);
Linus Torvalds's avatar
Linus Torvalds committed
938
939
940
941
942
943
944
945
946
947
948
	for (i = 0; i < disks; i++) {
		struct bio *mbio;
		if (!r1_bio->bios[i])
			continue;

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

		mbio->bi_sector	= r1_bio->sector + conf->mirrors[i].rdev->data_offset;
		mbio->bi_bdev = conf->mirrors[i].rdev->bdev;
		mbio->bi_end_io	= raid1_end_write_request;
949
950
		mbio->bi_rw = WRITE | (do_barriers << BIO_RW_BARRIER) |
			(do_sync << BIO_RW_SYNCIO);
Linus Torvalds's avatar
Linus Torvalds committed
951
952
		mbio->bi_private = r1_bio;

953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
		if (behind_pages) {
			struct bio_vec *bvec;
			int j;

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

Linus Torvalds's avatar
Linus Torvalds committed
970
971
		atomic_inc(&r1_bio->remaining);

972
		bio_list_add(&bl, mbio);
Linus Torvalds's avatar
Linus Torvalds committed
973
	}
974
	kfree(behind_pages); /* the behind pages are attached to the bios now */
Linus Torvalds's avatar
Linus Torvalds committed
975

976
977
	bitmap_startwrite(bitmap, bio->bi_sector, r1_bio->sectors,
				test_bit(R1BIO_BehindIO, &r1_bio->state));
978
979
980
981
982
983
984
	spin_lock_irqsave(&conf->device_lock, flags);
	bio_list_merge(&conf->pending_bio_list, &bl);
	bio_list_init(&bl);

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

985
986
987
	/* In case raid1d snuck into freeze_array */
	wake_up(&conf->wait_barrier);

988
989
	if (do_sync)
		md_wakeup_thread(mddev->thread);
990
991
992
993
994
#if 0
	while ((bio = bio_list_pop(&bl)) != NULL)
		generic_make_request(bio);
#endif

Linus Torvalds's avatar
Linus Torvalds committed
995
996
997
998
999
	return 0;
}

static void status(struct seq_file *seq, mddev_t *mddev)
{
1000
	conf_t *conf = mddev->private;
Linus Torvalds's avatar
Linus Torvalds committed
1001
1002
1003
	int i;

	seq_printf(seq, " [%d/%d] [", conf->raid_disks,
1004
		   conf->raid_disks - mddev->degraded);
1005
1006
1007
	rcu_read_lock();
	for (i = 0; i < conf->raid_disks; i++) {
		mdk_rdev_t *rdev = rcu_dereference(conf->mirrors[i].rdev);
Linus Torvalds's avatar
Linus Torvalds committed
1008
		seq_printf(seq, "%s",
1009
1010
1011
			   rdev && test_bit(In_sync, &rdev->flags) ? "U" : "_");
	}
	rcu_read_unlock();
Linus Torvalds's avatar
Linus Torvalds committed
1012
1013
1014
1015
1016
1017
1018
	seq_printf(seq, "]");
}


static void error(mddev_t *mddev, mdk_rdev_t *rdev)
{
	char b[BDEVNAME_SIZE];
1019
	conf_t *conf = mddev->private;
Linus Torvalds's avatar
Linus Torvalds committed
1020
1021
1022
1023
1024
1025
1026

	/*
	 * If it is not operational, then we have already marked it as dead
	 * else if it is the last working disks, ignore the error, let the
	 * next level up know.
	 * else mark the drive as failed
	 */
1027
	if (test_bit(In_sync, &rdev->flags)
1028
	    && (conf->raid_disks - mddev->degraded) == 1) {
Linus Torvalds's avatar
Linus Torvalds committed
1029
1030
		/*
		 * Don't fail the drive, act as though we were just a
1031
1032
1033
		 * normal single drive.
		 * However don't try a recovery from this drive as
		 * it is very likely to fail.
Linus Torvalds's avatar
Linus Torvalds committed
1034
		 */
1035
		mddev->recovery_disabled = 1;
Linus Torvalds's avatar
Linus Torvalds committed
1036
		return;
1037
	}
1038
1039
1040
	if (test_and_clear_bit(In_sync, &rdev->flags)) {
		unsigned long flags;
		spin_lock_irqsave(&conf->device_lock, flags);
Linus Torvalds's avatar
Linus Torvalds committed
1041
		mddev->degraded++;
1042
		set_bit(Faulty, &rdev->flags);
1043
		spin_unlock_irqrestore(&conf->device_lock, flags);
Linus Torvalds's avatar
Linus Torvalds committed
1044
1045
1046
		/*
		 * if recovery is running, make sure it aborts.
		 */
1047
		set_bit(MD_RECOVERY_INTR, &mddev->recovery);
1048
1049
	} else
		set_bit(Faulty, &rdev->flags);
1050
	set_bit(MD_CHANGE_DEVS, &mddev->flags);
1051
1052
	printk(KERN_ALERT "raid1: Disk failure on %s, disabling device.\n"
		"raid1: Operation continuing on %d devices.\n",
1053
		bdevname(rdev->bdev,b), conf->raid_disks - mddev->degraded);
Linus Torvalds's avatar
Linus Torvalds committed
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
}

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

	printk("RAID1 conf printout:\n");
	if (!conf) {
		printk("(!conf)\n");
		return;
	}
1065
	printk(" --- wd:%d rd:%d\n", conf->raid_disks - conf->mddev->degraded,
Linus Torvalds's avatar
Linus Torvalds committed
1066
1067
		conf->raid_disks);

1068
	rcu_read_lock();
Linus Torvalds's avatar
Linus Torvalds committed
1069
1070
	for (i = 0; i < conf->raid_disks; i++) {
		char b[BDEVNAME_SIZE];
1071
1072
		mdk_rdev_t *rdev = rcu_dereference(conf->mirrors[i].rdev);
		if (rdev)
Linus Torvalds's avatar
Linus Torvalds committed
1073
			printk(" disk %d, wo:%d, o:%d, dev:%s\n",
1074
1075
1076
			       i, !test_bit(In_sync, &rdev->flags),
			       !test_bit(Faulty, &rdev->flags),
			       bdevname(rdev->bdev,b));
Linus Torvalds's avatar
Linus Torvalds committed
1077
	}
1078
	rcu_read_unlock();
Linus Torvalds's avatar
Linus Torvalds committed
1079
1080
1081
1082
}

static void close_sync(conf_t *conf)
{
1083
1084
	wait_barrier(conf);
	allow_barrier(conf);
Linus Torvalds's avatar
Linus Torvalds committed
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096

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

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

	/*
	 * Find all failed disks within the RAID1 configuration 
1097
1098
	 * and mark them readable.
	 * Called under mddev lock, so rcu protection not needed.
Linus Torvalds's avatar
Linus Torvalds committed
1099
1100
	 */
	for (i = 0; i < conf->raid_disks; i++) {
1101
1102
1103
		mdk_rdev_t *rdev = conf->mirrors[i].rdev;
		if (rdev
		    && !test_bit(Faulty, &rdev->flags)
1104
1105
1106
		    && !test_and_set_bit(In_sync, &rdev->flags)) {
			unsigned long flags;
			spin_lock_irqsave(&conf->device_lock, flags);
Linus Torvalds's avatar
Linus Torvalds committed
1107
			mddev->degraded--;
1108
			spin_unlock_irqrestore(&conf->device_lock, flags);
Linus Torvalds's avatar
Linus Torvalds committed
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
		}
	}

	print_conf(conf);
	return 0;
}


static int raid1_add_disk(mddev_t *mddev, mdk_rdev_t *rdev)
{
	conf_t *conf = mddev->private;
1120
	int err = -EEXIST;
1121
	int mirror = 0;
Linus Torvalds's avatar
Linus Torvalds committed
1122
	mirror_info_t *p;
1123
1124
	int first = 0;
	int last = mddev->raid_disks - 1;
Linus Torvalds's avatar
Linus Torvalds committed
1125

1126
1127
1128
1129
	if (rdev->raid_disk >= 0)
		first = last = rdev->raid_disk;

	for (mirror = first; mirror <= last; mirror++)
Linus Torvalds's avatar
Linus Torvalds committed
1130
1131
		if ( !(p=conf->mirrors+mirror)->rdev) {

1132
1133
			disk_stack_limits(mddev->gendisk, rdev->bdev,
					  rdev->data_offset << 9);
Linus Torvalds's avatar
Linus Torvalds committed
1134
1135
1136
1137
1138
			/* as we don't honour merge_bvec_fn, we must never risk
			 * violating it, so limit ->max_sector to one PAGE, as
			 * a one page request is never in violation.
			 */
			if (rdev->bdev->bd_disk->queue->merge_bvec_fn &&
1139
			    queue_max_sectors(mddev->queue) > (PAGE_SIZE>>9))
Linus Torvalds's avatar
Linus Torvalds committed
1140
1141
1142
1143
				blk_queue_max_sectors(mddev->queue, PAGE_SIZE>>9);

			p->head_position = 0;
			rdev->raid_disk = mirror;
1144
			err = 0;
1145
1146
1147
1148
			/* As all devices are equivalent, we don't need a full recovery
			 * if this was recently any drive of the array
			 */
			if (rdev->saved_raid_disk < 0)
1149
				conf->fullsync = 1;
1150
			rcu_assign_pointer(p->rdev, rdev);
Linus Torvalds's avatar
Linus Torvalds committed
1151
1152
			break;
		}
1153
	md_integrity_add_rdev(rdev, mddev);
Linus Torvalds's avatar
Linus Torvalds committed
1154
	print_conf(conf);
1155
	return err;
Linus Torvalds's avatar
Linus Torvalds committed
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
}

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

	print_conf(conf);
	rdev = p->rdev;
	if (rdev) {
1168
		if (test_bit(In_sync, &rdev->flags) ||
Linus Torvalds's avatar
Linus Torvalds committed
1169
1170
1171
1172
		    atomic_read(&rdev->nr_pending)) {
			err = -EBUSY;
			goto abort;
		}
1173
1174
1175
1176
1177
1178
1179
1180
		/* Only remove non-faulty devices is recovery
		 * is not possible.
		 */
		if (!test_bit(Faulty, &rdev->flags) &&
		    mddev->degraded < conf->raid_disks) {
			err = -EBUSY;
			goto abort;
		}
Linus Torvalds's avatar
Linus Torvalds committed
1181
		p->rdev = NULL;
1182
		synchronize_rcu();
Linus Torvalds's avatar
Linus Torvalds committed
1183
1184
1185
1186
		if (atomic_read(&rdev->nr_pending)) {
			/* lost the race, try later */
			err = -EBUSY;
			p->rdev = rdev;
1187
			goto abort;
Linus Torvalds's avatar
Linus Torvalds committed
1188
		}
1189
		md_integrity_register(mddev);
Linus Torvalds's avatar
Linus Torvalds committed
1190
1191
1192
1193
1194
1195
1196
1197
	}
abort:

	print_conf(conf);
	return err;
}


1198
static void end_sync_read(struct bio *bio, int error)
Linus Torvalds's avatar
Linus Torvalds committed
1199
1200
{
	r1bio_t * r1_bio = (r1bio_t *)(bio->bi_private);
1201
	int i;
Linus Torvalds's avatar
Linus Torvalds committed
1202

1203
1204
1205
1206
1207
	for (i=r1_bio->mddev->raid_disks; i--; )
		if (r1_bio->bios[i] == bio)
			break;
	BUG_ON(i < 0);
	update_head_pos(i, r1_bio);
Linus Torvalds's avatar
Linus Torvalds committed
1208
1209
1210
1211
1212
	/*
	 * we have read a block, now it needs to be re-written,
	 * or re-read if the read failed.
	 * We don't do much here, just schedule handling by raid1d
	 */
1213
	if (test_bit(BIO_UPTODATE, &bio->bi_flags))
Linus Torvalds's avatar
Linus Torvalds committed
1214
		set_bit(R1BIO_Uptodate, &r1_bio->state);
1215
1216
1217

	if (atomic_dec_and_test(&r1_bio->remaining))
		reschedule_retry(r1_bio);
Linus Torvalds's avatar
Linus Torvalds committed
1218
1219
}

1220
static void end_sync_write(struct bio *bio, int error)
Linus Torvalds's avatar
Linus Torvalds committed
1221
1222
1223
1224
{
	int uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags);
	r1bio_t * r1_bio = (r1bio_t *)(bio->bi_private);
	mddev_t *mddev = r1_bio->mddev;
1225
	conf_t *conf = mddev->private;
Linus Torvalds's avatar
Linus Torvalds committed
1226
1227
1228
1229
1230
1231
1232
1233
	int i;
	int mirror=0;

	for (i = 0; i < conf->raid_disks; i++)
		if (r1_bio->bios[i] == bio) {
			mirror = i;
			break;
		}
1234
1235
1236
1237
1238
1239
	if (!uptodate) {
		int sync_blocks = 0;
		sector_t s = r1_bio->sector;
		long sectors_to_go = r1_bio->sectors;
		/* make sure these bits doesn't get cleared. */
		do {
1240
			bitmap_end_sync(mddev->bitmap, s,
1241
1242
1243
1244
					&sync_blocks, 1);
			s += sync_blocks;
			sectors_to_go -= sync_blocks;
		} while (sectors_to_go > 0);
Linus Torvalds's avatar
Linus Torvalds committed
1245
		md_error(mddev, conf->mirrors[mirror].rdev);
1246
	}
1247

Linus Torvalds's avatar
Linus Torvalds committed
1248
1249
1250
	update_head_pos(mirror, r1_bio);

	if (atomic_dec_and_test(&r1_bio->remaining)) {
1251
		sector_t s = r1_bio->sectors;
Linus Torvalds's avatar
Linus Torvalds committed
1252
		put_buf(r1_bio);
1253
		md_done_sync(mddev, s, uptodate);
Linus Torvalds's avatar
Linus Torvalds committed
1254
1255
1256
1257
1258
	}
}

static void sync_request_write(mddev_t *mddev, r1bio_t *r1_bio)
{
1259
	conf_t *conf = mddev->private;
Linus Torvalds's avatar
Linus Torvalds committed
1260
1261
1262
1263
1264
1265
	int i;
	int disks = conf->raid_disks;
	struct bio *bio, *wbio;

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

1266

1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
	if (test_bit(MD_RECOVERY_REQUESTED, &mddev->recovery)) {
		/* We have read all readable devices.  If we haven't
		 * got the block, then there is no hope left.
		 * If we have, then we want to do a comparison
		 * and skip the write if everything is the same.
		 * If any blocks failed to read, then we need to
		 * attempt an over-write
		 */
		int primary;
		if (!test_bit(R1BIO_Uptodate, &r1_bio->state)) {
			for (i=0; i<mddev->raid_disks; i++)
				if (r1_bio->bios[i]->bi_end_io == end_sync_read)
					md_error(mddev, conf->mirrors[i].rdev);

			md_done_sync(mddev, r1_bio->sectors, 1);
			put_buf(r1_bio);
			return;
		}
		for (primary=0; primary<mddev->raid_disks; primary++)
			if (r1_bio->bios[primary]->bi_end_io == end_sync_read &&
			    test_bit(BIO_UPTODATE, &r1_bio->bios[primary]->bi_flags)) {
				r1_bio->bios[primary]->bi_end_io = NULL;
1289
				rdev_dec_pending(conf->mirrors[primary].rdev, mddev);
1290
1291
1292
1293
				break;
			}
		r1_bio->read_disk = primary;
		for (i=0; i<mddev->raid_disks; i++)
1294
			if (r1_bio->bios[i]->bi_end_io == end_sync_read) {
1295
1296
1297
1298
				int j;
				int vcnt = r1_bio->sectors >> (PAGE_SHIFT- 9);
				struct bio *pbio = r1_bio->bios[primary];
				struct bio *sbio = r1_bio->bios[i];
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311

				if (test_bit(BIO_UPTODATE, &sbio->bi_flags)) {
					for (j = vcnt; j-- ; ) {
						struct page *p, *s;
						p = pbio->bi_io_vec[j].bv_page;
						s = sbio->bi_io_vec[j].bv_page;
						if (memcmp(page_address(p),
							   page_address(s),
							   PAGE_SIZE))
							break;
					}
				} else
					j = 0;
1312
1313
				if (j >= 0)
					mddev->resync_mismatches += r1_bio->sectors;
1314
1315
				if (j < 0 || (test_bit(MD_RECOVERY_CHECK, &mddev->recovery)
					      && test_bit(BIO_UPTODATE, &sbio->bi_flags))) {
1316
					sbio->bi_end_io = NULL;
1317
1318
					rdev_dec_pending(conf->mirrors[i].rdev, mddev);
				} else {
1319
					/* fixup the bio for reuse */
1320
					int size;
1321
1322
1323
1324
1325
1326
1327
1328
1329
1330
					sbio->bi_vcnt = vcnt;
					sbio->bi_size = r1_bio->sectors << 9;
					sbio->bi_idx = 0;
					sbio->bi_phys_segments = 0;
					sbio->bi_flags &= ~(BIO_POOL_MASK - 1);
					sbio->bi_flags |= 1 << BIO_UPTODATE;
					sbio->bi_next = NULL;
					sbio->bi_sector = r1_bio->sector +
						conf->mirrors[i].rdev->data_offset;
					sbio->bi_bdev = conf->mirrors[i].rdev->bdev;
1331
1332
1333
1334
1335
1336
1337
1338
1339
1340
1341
					size = sbio->bi_size;
					for (j = 0; j < vcnt ; j++) {
						struct bio_vec *bi;
						bi = &sbio->bi_io_vec[j];
						bi->bv_offset = 0;
						if (size > PAGE_SIZE)
							bi->bv_len = PAGE_SIZE;
						else
							bi->bv_len = size;
						size -= PAGE_SIZE;
						memcpy(page_address(bi->bv_page),
1342
1343
						       page_address(pbio->bi_io_vec[j].bv_page),
						       PAGE_SIZE);
1344
					}
1345

1346
1347
1348
				}
			}
	}
Linus Torvalds's avatar
Linus Torvalds committed
1349
	if (!test_bit(R1BIO_Uptodate, &r1_bio->state)) {
1350
1351
1352
		/* ouch - failed to read all of that.
		 * Try some synchronous reads of other devices to get
		 * good data, much like with normal read errors.  Only
1353
		 * read into the pages we already have so we don't
1354
1355
1356
1357
		 * need to re-issue the read request.
		 * We don't need to freeze the array, because being in an
		 * active sync request, there is no normal IO, and
		 * no overlapping syncs.
Linus Torvalds's avatar
Linus Torvalds committed
1358
		 */
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368