blob: 23a642619bedfbbe76137f2d1e8d174782622f4e [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
2 * Copyright (C) 2003 Sistina Software Limited.
3 *
4 * This file is released under the GPL.
5 */
6
7#include "dm.h"
8#include "dm-bio-list.h"
9#include "dm-io.h"
10#include "dm-log.h"
11#include "kcopyd.h"
12
13#include <linux/ctype.h>
14#include <linux/init.h>
15#include <linux/mempool.h>
16#include <linux/module.h>
17#include <linux/pagemap.h>
18#include <linux/slab.h>
19#include <linux/time.h>
20#include <linux/vmalloc.h>
21#include <linux/workqueue.h>
22
Alasdair G Kergon72d94862006-06-26 00:27:35 -070023#define DM_MSG_PREFIX "raid1"
24
Linus Torvalds1da177e2005-04-16 15:20:36 -070025static struct workqueue_struct *_kmirrord_wq;
26static struct work_struct _kmirrord_work;
Jonathan E Brassow33184042006-11-08 17:44:44 -080027static DECLARE_WAIT_QUEUE_HEAD(_kmirrord_recovery_stopped);
Linus Torvalds1da177e2005-04-16 15:20:36 -070028
29static inline void wake(void)
30{
31 queue_work(_kmirrord_wq, &_kmirrord_work);
32}
33
34/*-----------------------------------------------------------------
35 * Region hash
36 *
37 * The mirror splits itself up into discrete regions. Each
38 * region can be in one of three states: clean, dirty,
39 * nosync. There is no need to put clean regions in the hash.
40 *
41 * In addition to being present in the hash table a region _may_
42 * be present on one of three lists.
43 *
44 * clean_regions: Regions on this list have no io pending to
45 * them, they are in sync, we are no longer interested in them,
46 * they are dull. rh_update_states() will remove them from the
47 * hash table.
48 *
49 * quiesced_regions: These regions have been spun down, ready
50 * for recovery. rh_recovery_start() will remove regions from
51 * this list and hand them to kmirrord, which will schedule the
52 * recovery io with kcopyd.
53 *
54 * recovered_regions: Regions that kcopyd has successfully
55 * recovered. rh_update_states() will now schedule any delayed
56 * io, up the recovery_count, and remove the region from the
57 * hash.
58 *
59 * There are 2 locks:
60 * A rw spin lock 'hash_lock' protects just the hash table,
61 * this is never held in write mode from interrupt context,
62 * which I believe means that we only have to disable irqs when
63 * doing a write lock.
64 *
65 * An ordinary spin lock 'region_lock' that protects the three
66 * lists in the region_hash, with the 'state', 'list' and
67 * 'bhs_delayed' fields of the regions. This is used from irq
68 * context, so all other uses will have to suspend local irqs.
69 *---------------------------------------------------------------*/
70struct mirror_set;
71struct region_hash {
72 struct mirror_set *ms;
73 uint32_t region_size;
74 unsigned region_shift;
75
76 /* holds persistent region state */
77 struct dirty_log *log;
78
79 /* hash table */
80 rwlock_t hash_lock;
81 mempool_t *region_pool;
82 unsigned int mask;
83 unsigned int nr_buckets;
84 struct list_head *buckets;
85
86 spinlock_t region_lock;
Jonathan E Brassow33184042006-11-08 17:44:44 -080087 atomic_t recovery_in_flight;
Linus Torvalds1da177e2005-04-16 15:20:36 -070088 struct semaphore recovery_count;
89 struct list_head clean_regions;
90 struct list_head quiesced_regions;
91 struct list_head recovered_regions;
92};
93
94enum {
95 RH_CLEAN,
96 RH_DIRTY,
97 RH_NOSYNC,
98 RH_RECOVERING
99};
100
101struct region {
102 struct region_hash *rh; /* FIXME: can we get rid of this ? */
103 region_t key;
104 int state;
105
106 struct list_head hash_list;
107 struct list_head list;
108
109 atomic_t pending;
110 struct bio_list delayed_bios;
111};
112
Neil Browne4c8b3b2006-06-26 00:27:26 -0700113
114/*-----------------------------------------------------------------
115 * Mirror set structures.
116 *---------------------------------------------------------------*/
117struct mirror {
118 atomic_t error_count;
119 struct dm_dev *dev;
120 sector_t offset;
121};
122
123struct mirror_set {
124 struct dm_target *ti;
125 struct list_head list;
126 struct region_hash rh;
127 struct kcopyd_client *kcopyd_client;
128
129 spinlock_t lock; /* protects the next two lists */
130 struct bio_list reads;
131 struct bio_list writes;
132
133 /* recovery */
134 region_t nr_regions;
135 int in_sync;
136
137 struct mirror *default_mirror; /* Default mirror */
138
139 unsigned int nr_mirrors;
140 struct mirror mirror[0];
141};
142
Linus Torvalds1da177e2005-04-16 15:20:36 -0700143/*
144 * Conversion fns
145 */
146static inline region_t bio_to_region(struct region_hash *rh, struct bio *bio)
147{
Neil Browne4c8b3b2006-06-26 00:27:26 -0700148 return (bio->bi_sector - rh->ms->ti->begin) >> rh->region_shift;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700149}
150
151static inline sector_t region_to_sector(struct region_hash *rh, region_t region)
152{
153 return region << rh->region_shift;
154}
155
156/* FIXME move this */
157static void queue_bio(struct mirror_set *ms, struct bio *bio, int rw);
158
Linus Torvalds1da177e2005-04-16 15:20:36 -0700159#define MIN_REGIONS 64
160#define MAX_RECOVERY 1
161static int rh_init(struct region_hash *rh, struct mirror_set *ms,
162 struct dirty_log *log, uint32_t region_size,
163 region_t nr_regions)
164{
165 unsigned int nr_buckets, max_buckets;
166 size_t i;
167
168 /*
169 * Calculate a suitable number of buckets for our hash
170 * table.
171 */
172 max_buckets = nr_regions >> 6;
173 for (nr_buckets = 128u; nr_buckets < max_buckets; nr_buckets <<= 1)
174 ;
175 nr_buckets >>= 1;
176
177 rh->ms = ms;
178 rh->log = log;
179 rh->region_size = region_size;
180 rh->region_shift = ffs(region_size) - 1;
181 rwlock_init(&rh->hash_lock);
182 rh->mask = nr_buckets - 1;
183 rh->nr_buckets = nr_buckets;
184
185 rh->buckets = vmalloc(nr_buckets * sizeof(*rh->buckets));
186 if (!rh->buckets) {
187 DMERR("unable to allocate region hash memory");
188 return -ENOMEM;
189 }
190
191 for (i = 0; i < nr_buckets; i++)
192 INIT_LIST_HEAD(rh->buckets + i);
193
194 spin_lock_init(&rh->region_lock);
195 sema_init(&rh->recovery_count, 0);
Jonathan E Brassow33184042006-11-08 17:44:44 -0800196 atomic_set(&rh->recovery_in_flight, 0);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700197 INIT_LIST_HEAD(&rh->clean_regions);
198 INIT_LIST_HEAD(&rh->quiesced_regions);
199 INIT_LIST_HEAD(&rh->recovered_regions);
200
Matthew Dobson0eaae62a2006-03-26 01:37:47 -0800201 rh->region_pool = mempool_create_kmalloc_pool(MIN_REGIONS,
202 sizeof(struct region));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700203 if (!rh->region_pool) {
204 vfree(rh->buckets);
205 rh->buckets = NULL;
206 return -ENOMEM;
207 }
208
209 return 0;
210}
211
212static void rh_exit(struct region_hash *rh)
213{
214 unsigned int h;
215 struct region *reg, *nreg;
216
217 BUG_ON(!list_empty(&rh->quiesced_regions));
218 for (h = 0; h < rh->nr_buckets; h++) {
219 list_for_each_entry_safe(reg, nreg, rh->buckets + h, hash_list) {
220 BUG_ON(atomic_read(&reg->pending));
221 mempool_free(reg, rh->region_pool);
222 }
223 }
224
225 if (rh->log)
226 dm_destroy_dirty_log(rh->log);
227 if (rh->region_pool)
228 mempool_destroy(rh->region_pool);
229 vfree(rh->buckets);
230}
231
232#define RH_HASH_MULT 2654435387U
233
234static inline unsigned int rh_hash(struct region_hash *rh, region_t region)
235{
236 return (unsigned int) ((region * RH_HASH_MULT) >> 12) & rh->mask;
237}
238
239static struct region *__rh_lookup(struct region_hash *rh, region_t region)
240{
241 struct region *reg;
242
243 list_for_each_entry (reg, rh->buckets + rh_hash(rh, region), hash_list)
244 if (reg->key == region)
245 return reg;
246
247 return NULL;
248}
249
250static void __rh_insert(struct region_hash *rh, struct region *reg)
251{
252 unsigned int h = rh_hash(rh, reg->key);
253 list_add(&reg->hash_list, rh->buckets + h);
254}
255
256static struct region *__rh_alloc(struct region_hash *rh, region_t region)
257{
258 struct region *reg, *nreg;
259
260 read_unlock(&rh->hash_lock);
Daniel Kobrasc06aad82006-08-27 01:23:24 -0700261 nreg = mempool_alloc(rh->region_pool, GFP_ATOMIC);
262 if (unlikely(!nreg))
263 nreg = kmalloc(sizeof(struct region), GFP_NOIO);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700264 nreg->state = rh->log->type->in_sync(rh->log, region, 1) ?
265 RH_CLEAN : RH_NOSYNC;
266 nreg->rh = rh;
267 nreg->key = region;
268
269 INIT_LIST_HEAD(&nreg->list);
270
271 atomic_set(&nreg->pending, 0);
272 bio_list_init(&nreg->delayed_bios);
273 write_lock_irq(&rh->hash_lock);
274
275 reg = __rh_lookup(rh, region);
276 if (reg)
277 /* we lost the race */
278 mempool_free(nreg, rh->region_pool);
279
280 else {
281 __rh_insert(rh, nreg);
282 if (nreg->state == RH_CLEAN) {
283 spin_lock(&rh->region_lock);
284 list_add(&nreg->list, &rh->clean_regions);
285 spin_unlock(&rh->region_lock);
286 }
287 reg = nreg;
288 }
289 write_unlock_irq(&rh->hash_lock);
290 read_lock(&rh->hash_lock);
291
292 return reg;
293}
294
295static inline struct region *__rh_find(struct region_hash *rh, region_t region)
296{
297 struct region *reg;
298
299 reg = __rh_lookup(rh, region);
300 if (!reg)
301 reg = __rh_alloc(rh, region);
302
303 return reg;
304}
305
306static int rh_state(struct region_hash *rh, region_t region, int may_block)
307{
308 int r;
309 struct region *reg;
310
311 read_lock(&rh->hash_lock);
312 reg = __rh_lookup(rh, region);
313 read_unlock(&rh->hash_lock);
314
315 if (reg)
316 return reg->state;
317
318 /*
319 * The region wasn't in the hash, so we fall back to the
320 * dirty log.
321 */
322 r = rh->log->type->in_sync(rh->log, region, may_block);
323
324 /*
325 * Any error from the dirty log (eg. -EWOULDBLOCK) gets
326 * taken as a RH_NOSYNC
327 */
328 return r == 1 ? RH_CLEAN : RH_NOSYNC;
329}
330
331static inline int rh_in_sync(struct region_hash *rh,
332 region_t region, int may_block)
333{
334 int state = rh_state(rh, region, may_block);
335 return state == RH_CLEAN || state == RH_DIRTY;
336}
337
338static void dispatch_bios(struct mirror_set *ms, struct bio_list *bio_list)
339{
340 struct bio *bio;
341
342 while ((bio = bio_list_pop(bio_list))) {
343 queue_bio(ms, bio, WRITE);
344 }
345}
346
Jonathan E Brassowf3ee6b22006-12-08 02:41:11 -0800347static void complete_resync_work(struct region *reg, int success)
348{
349 struct region_hash *rh = reg->rh;
350
351 rh->log->type->set_region_sync(rh->log, reg->key, success);
352 dispatch_bios(rh->ms, &reg->delayed_bios);
353 if (atomic_dec_and_test(&rh->recovery_in_flight))
354 wake_up_all(&_kmirrord_recovery_stopped);
355 up(&rh->recovery_count);
356}
357
Linus Torvalds1da177e2005-04-16 15:20:36 -0700358static void rh_update_states(struct region_hash *rh)
359{
360 struct region *reg, *next;
361
362 LIST_HEAD(clean);
363 LIST_HEAD(recovered);
364
365 /*
366 * Quickly grab the lists.
367 */
368 write_lock_irq(&rh->hash_lock);
369 spin_lock(&rh->region_lock);
370 if (!list_empty(&rh->clean_regions)) {
371 list_splice(&rh->clean_regions, &clean);
372 INIT_LIST_HEAD(&rh->clean_regions);
373
374 list_for_each_entry (reg, &clean, list) {
375 rh->log->type->clear_region(rh->log, reg->key);
376 list_del(&reg->hash_list);
377 }
378 }
379
380 if (!list_empty(&rh->recovered_regions)) {
381 list_splice(&rh->recovered_regions, &recovered);
382 INIT_LIST_HEAD(&rh->recovered_regions);
383
384 list_for_each_entry (reg, &recovered, list)
385 list_del(&reg->hash_list);
386 }
387 spin_unlock(&rh->region_lock);
388 write_unlock_irq(&rh->hash_lock);
389
390 /*
391 * All the regions on the recovered and clean lists have
392 * now been pulled out of the system, so no need to do
393 * any more locking.
394 */
395 list_for_each_entry_safe (reg, next, &recovered, list) {
396 rh->log->type->clear_region(rh->log, reg->key);
Jonathan E Brassowf3ee6b22006-12-08 02:41:11 -0800397 complete_resync_work(reg, 1);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700398 mempool_free(reg, rh->region_pool);
399 }
400
401 if (!list_empty(&recovered))
402 rh->log->type->flush(rh->log);
403
404 list_for_each_entry_safe (reg, next, &clean, list)
405 mempool_free(reg, rh->region_pool);
406}
407
408static void rh_inc(struct region_hash *rh, region_t region)
409{
410 struct region *reg;
411
412 read_lock(&rh->hash_lock);
413 reg = __rh_find(rh, region);
Jun'ichi Nomura844e8d92005-09-09 16:23:42 -0700414
Jonathan E Brassow7692c5d2005-11-21 21:32:37 -0800415 spin_lock_irq(&rh->region_lock);
Jun'ichi Nomura844e8d92005-09-09 16:23:42 -0700416 atomic_inc(&reg->pending);
417
Linus Torvalds1da177e2005-04-16 15:20:36 -0700418 if (reg->state == RH_CLEAN) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700419 reg->state = RH_DIRTY;
420 list_del_init(&reg->list); /* take off the clean list */
Jonathan E Brassow7692c5d2005-11-21 21:32:37 -0800421 spin_unlock_irq(&rh->region_lock);
422
423 rh->log->type->mark_region(rh->log, reg->key);
424 } else
425 spin_unlock_irq(&rh->region_lock);
426
Linus Torvalds1da177e2005-04-16 15:20:36 -0700427
Linus Torvalds1da177e2005-04-16 15:20:36 -0700428 read_unlock(&rh->hash_lock);
429}
430
431static void rh_inc_pending(struct region_hash *rh, struct bio_list *bios)
432{
433 struct bio *bio;
434
435 for (bio = bios->head; bio; bio = bio->bi_next)
436 rh_inc(rh, bio_to_region(rh, bio));
437}
438
439static void rh_dec(struct region_hash *rh, region_t region)
440{
441 unsigned long flags;
442 struct region *reg;
443 int should_wake = 0;
444
445 read_lock(&rh->hash_lock);
446 reg = __rh_lookup(rh, region);
447 read_unlock(&rh->hash_lock);
448
Jonathan E Brassow7692c5d2005-11-21 21:32:37 -0800449 spin_lock_irqsave(&rh->region_lock, flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700450 if (atomic_dec_and_test(&reg->pending)) {
Jun'ichi Nomura930d3322006-03-27 01:17:47 -0800451 /*
452 * There is no pending I/O for this region.
453 * We can move the region to corresponding list for next action.
454 * At this point, the region is not yet connected to any list.
455 *
456 * If the state is RH_NOSYNC, the region should be kept off
457 * from clean list.
458 * The hash entry for RH_NOSYNC will remain in memory
459 * until the region is recovered or the map is reloaded.
460 */
461
462 /* do nothing for RH_NOSYNC */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700463 if (reg->state == RH_RECOVERING) {
464 list_add_tail(&reg->list, &rh->quiesced_regions);
Jun'ichi Nomura930d3322006-03-27 01:17:47 -0800465 } else if (reg->state == RH_DIRTY) {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700466 reg->state = RH_CLEAN;
467 list_add(&reg->list, &rh->clean_regions);
468 }
Linus Torvalds1da177e2005-04-16 15:20:36 -0700469 should_wake = 1;
470 }
Jonathan E Brassow7692c5d2005-11-21 21:32:37 -0800471 spin_unlock_irqrestore(&rh->region_lock, flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700472
473 if (should_wake)
474 wake();
475}
476
477/*
478 * Starts quiescing a region in preparation for recovery.
479 */
480static int __rh_recovery_prepare(struct region_hash *rh)
481{
482 int r;
483 struct region *reg;
484 region_t region;
485
486 /*
487 * Ask the dirty log what's next.
488 */
489 r = rh->log->type->get_resync_work(rh->log, &region);
490 if (r <= 0)
491 return r;
492
493 /*
494 * Get this region, and start it quiescing by setting the
495 * recovering flag.
496 */
497 read_lock(&rh->hash_lock);
498 reg = __rh_find(rh, region);
499 read_unlock(&rh->hash_lock);
500
501 spin_lock_irq(&rh->region_lock);
502 reg->state = RH_RECOVERING;
503
504 /* Already quiesced ? */
505 if (atomic_read(&reg->pending))
506 list_del_init(&reg->list);
Akinobu Mita179e0912006-06-26 00:24:41 -0700507 else
508 list_move(&reg->list, &rh->quiesced_regions);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700509
Linus Torvalds1da177e2005-04-16 15:20:36 -0700510 spin_unlock_irq(&rh->region_lock);
511
512 return 1;
513}
514
515static void rh_recovery_prepare(struct region_hash *rh)
516{
Jonathan E Brassow33184042006-11-08 17:44:44 -0800517 /* Extra reference to avoid race with rh_stop_recovery */
518 atomic_inc(&rh->recovery_in_flight);
519
520 while (!down_trylock(&rh->recovery_count)) {
521 atomic_inc(&rh->recovery_in_flight);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700522 if (__rh_recovery_prepare(rh) <= 0) {
Jonathan E Brassow33184042006-11-08 17:44:44 -0800523 atomic_dec(&rh->recovery_in_flight);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700524 up(&rh->recovery_count);
525 break;
526 }
Jonathan E Brassow33184042006-11-08 17:44:44 -0800527 }
528
529 /* Drop the extra reference */
530 if (atomic_dec_and_test(&rh->recovery_in_flight))
531 wake_up_all(&_kmirrord_recovery_stopped);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700532}
533
534/*
535 * Returns any quiesced regions.
536 */
537static struct region *rh_recovery_start(struct region_hash *rh)
538{
539 struct region *reg = NULL;
540
541 spin_lock_irq(&rh->region_lock);
542 if (!list_empty(&rh->quiesced_regions)) {
543 reg = list_entry(rh->quiesced_regions.next,
544 struct region, list);
545 list_del_init(&reg->list); /* remove from the quiesced list */
546 }
547 spin_unlock_irq(&rh->region_lock);
548
549 return reg;
550}
551
552/* FIXME: success ignored for now */
553static void rh_recovery_end(struct region *reg, int success)
554{
555 struct region_hash *rh = reg->rh;
556
557 spin_lock_irq(&rh->region_lock);
558 list_add(&reg->list, &reg->rh->recovered_regions);
559 spin_unlock_irq(&rh->region_lock);
560
561 wake();
562}
563
564static void rh_flush(struct region_hash *rh)
565{
566 rh->log->type->flush(rh->log);
567}
568
569static void rh_delay(struct region_hash *rh, struct bio *bio)
570{
571 struct region *reg;
572
573 read_lock(&rh->hash_lock);
574 reg = __rh_find(rh, bio_to_region(rh, bio));
575 bio_list_add(&reg->delayed_bios, bio);
576 read_unlock(&rh->hash_lock);
577}
578
579static void rh_stop_recovery(struct region_hash *rh)
580{
581 int i;
582
583 /* wait for any recovering regions */
584 for (i = 0; i < MAX_RECOVERY; i++)
585 down(&rh->recovery_count);
586}
587
588static void rh_start_recovery(struct region_hash *rh)
589{
590 int i;
591
592 for (i = 0; i < MAX_RECOVERY; i++)
593 up(&rh->recovery_count);
594
595 wake();
596}
597
Linus Torvalds1da177e2005-04-16 15:20:36 -0700598/*
599 * Every mirror should look like this one.
600 */
601#define DEFAULT_MIRROR 0
602
603/*
604 * This is yucky. We squirrel the mirror_set struct away inside
605 * bi_next for write buffers. This is safe since the bh
606 * doesn't get submitted to the lower levels of block layer.
607 */
608static struct mirror_set *bio_get_ms(struct bio *bio)
609{
610 return (struct mirror_set *) bio->bi_next;
611}
612
613static void bio_set_ms(struct bio *bio, struct mirror_set *ms)
614{
615 bio->bi_next = (struct bio *) ms;
616}
617
618/*-----------------------------------------------------------------
619 * Recovery.
620 *
621 * When a mirror is first activated we may find that some regions
622 * are in the no-sync state. We have to recover these by
623 * recopying from the default mirror to all the others.
624 *---------------------------------------------------------------*/
625static void recovery_complete(int read_err, unsigned int write_err,
626 void *context)
627{
628 struct region *reg = (struct region *) context;
629
630 /* FIXME: better error handling */
Jonathan Brassowce503f52006-06-26 00:27:30 -0700631 rh_recovery_end(reg, !(read_err || write_err));
Linus Torvalds1da177e2005-04-16 15:20:36 -0700632}
633
634static int recover(struct mirror_set *ms, struct region *reg)
635{
636 int r;
637 unsigned int i;
638 struct io_region from, to[KCOPYD_MAX_REGIONS], *dest;
639 struct mirror *m;
640 unsigned long flags = 0;
641
642 /* fill in the source */
Jonathan E Brassowa1a19082006-01-06 00:20:05 -0800643 m = ms->default_mirror;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700644 from.bdev = m->dev->bdev;
645 from.sector = m->offset + region_to_sector(reg->rh, reg->key);
646 if (reg->key == (ms->nr_regions - 1)) {
647 /*
648 * The final region may be smaller than
649 * region_size.
650 */
651 from.count = ms->ti->len & (reg->rh->region_size - 1);
652 if (!from.count)
653 from.count = reg->rh->region_size;
654 } else
655 from.count = reg->rh->region_size;
656
657 /* fill in the destinations */
658 for (i = 0, dest = to; i < ms->nr_mirrors; i++) {
Jonathan E Brassowa1a19082006-01-06 00:20:05 -0800659 if (&ms->mirror[i] == ms->default_mirror)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700660 continue;
661
662 m = ms->mirror + i;
663 dest->bdev = m->dev->bdev;
664 dest->sector = m->offset + region_to_sector(reg->rh, reg->key);
665 dest->count = from.count;
666 dest++;
667 }
668
669 /* hand to kcopyd */
670 set_bit(KCOPYD_IGNORE_ERROR, &flags);
671 r = kcopyd_copy(ms->kcopyd_client, &from, ms->nr_mirrors - 1, to, flags,
672 recovery_complete, reg);
673
674 return r;
675}
676
677static void do_recovery(struct mirror_set *ms)
678{
679 int r;
680 struct region *reg;
681 struct dirty_log *log = ms->rh.log;
682
683 /*
684 * Start quiescing some regions.
685 */
686 rh_recovery_prepare(&ms->rh);
687
688 /*
689 * Copy any already quiesced regions.
690 */
691 while ((reg = rh_recovery_start(&ms->rh))) {
692 r = recover(ms, reg);
693 if (r)
694 rh_recovery_end(reg, 0);
695 }
696
697 /*
698 * Update the in sync flag.
699 */
700 if (!ms->in_sync &&
701 (log->type->get_sync_count(log) == ms->nr_regions)) {
702 /* the sync is complete */
703 dm_table_event(ms->ti->table);
704 ms->in_sync = 1;
705 }
706}
707
708/*-----------------------------------------------------------------
709 * Reads
710 *---------------------------------------------------------------*/
711static struct mirror *choose_mirror(struct mirror_set *ms, sector_t sector)
712{
713 /* FIXME: add read balancing */
Jonathan E Brassowa1a19082006-01-06 00:20:05 -0800714 return ms->default_mirror;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700715}
716
717/*
718 * remap a buffer to a particular mirror.
719 */
720static void map_bio(struct mirror_set *ms, struct mirror *m, struct bio *bio)
721{
722 bio->bi_bdev = m->dev->bdev;
723 bio->bi_sector = m->offset + (bio->bi_sector - ms->ti->begin);
724}
725
726static void do_reads(struct mirror_set *ms, struct bio_list *reads)
727{
728 region_t region;
729 struct bio *bio;
730 struct mirror *m;
731
732 while ((bio = bio_list_pop(reads))) {
733 region = bio_to_region(&ms->rh, bio);
734
735 /*
736 * We can only read balance if the region is in sync.
737 */
738 if (rh_in_sync(&ms->rh, region, 0))
739 m = choose_mirror(ms, bio->bi_sector);
740 else
Jonathan E Brassowa1a19082006-01-06 00:20:05 -0800741 m = ms->default_mirror;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700742
743 map_bio(ms, m, bio);
744 generic_make_request(bio);
745 }
746}
747
748/*-----------------------------------------------------------------
749 * Writes.
750 *
751 * We do different things with the write io depending on the
752 * state of the region that it's in:
753 *
754 * SYNC: increment pending, use kcopyd to write to *all* mirrors
755 * RECOVERING: delay the io until recovery completes
756 * NOSYNC: increment pending, just write to the default mirror
757 *---------------------------------------------------------------*/
758static void write_callback(unsigned long error, void *context)
759{
760 unsigned int i;
761 int uptodate = 1;
762 struct bio *bio = (struct bio *) context;
763 struct mirror_set *ms;
764
765 ms = bio_get_ms(bio);
766 bio_set_ms(bio, NULL);
767
768 /*
769 * NOTE: We don't decrement the pending count here,
770 * instead it is done by the targets endio function.
771 * This way we handle both writes to SYNC and NOSYNC
772 * regions with the same code.
773 */
774
775 if (error) {
776 /*
777 * only error the io if all mirrors failed.
778 * FIXME: bogus
779 */
780 uptodate = 0;
781 for (i = 0; i < ms->nr_mirrors; i++)
782 if (!test_bit(i, &error)) {
783 uptodate = 1;
784 break;
785 }
786 }
787 bio_endio(bio, bio->bi_size, 0);
788}
789
790static void do_write(struct mirror_set *ms, struct bio *bio)
791{
792 unsigned int i;
793 struct io_region io[KCOPYD_MAX_REGIONS+1];
794 struct mirror *m;
795
796 for (i = 0; i < ms->nr_mirrors; i++) {
797 m = ms->mirror + i;
798
799 io[i].bdev = m->dev->bdev;
800 io[i].sector = m->offset + (bio->bi_sector - ms->ti->begin);
801 io[i].count = bio->bi_size >> 9;
802 }
803
804 bio_set_ms(bio, ms);
805 dm_io_async_bvec(ms->nr_mirrors, io, WRITE,
806 bio->bi_io_vec + bio->bi_idx,
807 write_callback, bio);
808}
809
810static void do_writes(struct mirror_set *ms, struct bio_list *writes)
811{
812 int state;
813 struct bio *bio;
814 struct bio_list sync, nosync, recover, *this_list = NULL;
815
816 if (!writes->head)
817 return;
818
819 /*
820 * Classify each write.
821 */
822 bio_list_init(&sync);
823 bio_list_init(&nosync);
824 bio_list_init(&recover);
825
826 while ((bio = bio_list_pop(writes))) {
827 state = rh_state(&ms->rh, bio_to_region(&ms->rh, bio), 1);
828 switch (state) {
829 case RH_CLEAN:
830 case RH_DIRTY:
831 this_list = &sync;
832 break;
833
834 case RH_NOSYNC:
835 this_list = &nosync;
836 break;
837
838 case RH_RECOVERING:
839 this_list = &recover;
840 break;
841 }
842
843 bio_list_add(this_list, bio);
844 }
845
846 /*
847 * Increment the pending counts for any regions that will
848 * be written to (writes to recover regions are going to
849 * be delayed).
850 */
851 rh_inc_pending(&ms->rh, &sync);
852 rh_inc_pending(&ms->rh, &nosync);
853 rh_flush(&ms->rh);
854
855 /*
856 * Dispatch io.
857 */
858 while ((bio = bio_list_pop(&sync)))
859 do_write(ms, bio);
860
861 while ((bio = bio_list_pop(&recover)))
862 rh_delay(&ms->rh, bio);
863
864 while ((bio = bio_list_pop(&nosync))) {
Jonathan E Brassowa1a19082006-01-06 00:20:05 -0800865 map_bio(ms, ms->default_mirror, bio);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700866 generic_make_request(bio);
867 }
868}
869
870/*-----------------------------------------------------------------
871 * kmirrord
872 *---------------------------------------------------------------*/
873static LIST_HEAD(_mirror_sets);
874static DECLARE_RWSEM(_mirror_sets_lock);
875
876static void do_mirror(struct mirror_set *ms)
877{
878 struct bio_list reads, writes;
879
880 spin_lock(&ms->lock);
881 reads = ms->reads;
882 writes = ms->writes;
883 bio_list_init(&ms->reads);
884 bio_list_init(&ms->writes);
885 spin_unlock(&ms->lock);
886
887 rh_update_states(&ms->rh);
888 do_recovery(ms);
889 do_reads(ms, &reads);
890 do_writes(ms, &writes);
891}
892
David Howellsc4028952006-11-22 14:57:56 +0000893static void do_work(struct work_struct *ignored)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700894{
895 struct mirror_set *ms;
896
897 down_read(&_mirror_sets_lock);
898 list_for_each_entry (ms, &_mirror_sets, list)
899 do_mirror(ms);
900 up_read(&_mirror_sets_lock);
901}
902
903/*-----------------------------------------------------------------
904 * Target functions
905 *---------------------------------------------------------------*/
906static struct mirror_set *alloc_context(unsigned int nr_mirrors,
907 uint32_t region_size,
908 struct dm_target *ti,
909 struct dirty_log *dl)
910{
911 size_t len;
912 struct mirror_set *ms = NULL;
913
914 if (array_too_big(sizeof(*ms), sizeof(ms->mirror[0]), nr_mirrors))
915 return NULL;
916
917 len = sizeof(*ms) + (sizeof(ms->mirror[0]) * nr_mirrors);
918
919 ms = kmalloc(len, GFP_KERNEL);
920 if (!ms) {
Alasdair G Kergon72d94862006-06-26 00:27:35 -0700921 ti->error = "Cannot allocate mirror context";
Linus Torvalds1da177e2005-04-16 15:20:36 -0700922 return NULL;
923 }
924
925 memset(ms, 0, len);
926 spin_lock_init(&ms->lock);
927
928 ms->ti = ti;
929 ms->nr_mirrors = nr_mirrors;
930 ms->nr_regions = dm_sector_div_up(ti->len, region_size);
931 ms->in_sync = 0;
Jonathan E Brassowa1a19082006-01-06 00:20:05 -0800932 ms->default_mirror = &ms->mirror[DEFAULT_MIRROR];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700933
934 if (rh_init(&ms->rh, ms, dl, region_size, ms->nr_regions)) {
Alasdair G Kergon72d94862006-06-26 00:27:35 -0700935 ti->error = "Error creating dirty region hash";
Linus Torvalds1da177e2005-04-16 15:20:36 -0700936 kfree(ms);
937 return NULL;
938 }
939
940 return ms;
941}
942
943static void free_context(struct mirror_set *ms, struct dm_target *ti,
944 unsigned int m)
945{
946 while (m--)
947 dm_put_device(ti, ms->mirror[m].dev);
948
949 rh_exit(&ms->rh);
950 kfree(ms);
951}
952
953static inline int _check_region_size(struct dm_target *ti, uint32_t size)
954{
955 return !(size % (PAGE_SIZE >> 9) || (size & (size - 1)) ||
956 size > ti->len);
957}
958
959static int get_mirror(struct mirror_set *ms, struct dm_target *ti,
960 unsigned int mirror, char **argv)
961{
Andrew Morton4ee218c2006-03-27 01:17:48 -0800962 unsigned long long offset;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700963
Andrew Morton4ee218c2006-03-27 01:17:48 -0800964 if (sscanf(argv[1], "%llu", &offset) != 1) {
Alasdair G Kergon72d94862006-06-26 00:27:35 -0700965 ti->error = "Invalid offset";
Linus Torvalds1da177e2005-04-16 15:20:36 -0700966 return -EINVAL;
967 }
968
969 if (dm_get_device(ti, argv[0], offset, ti->len,
970 dm_table_get_mode(ti->table),
971 &ms->mirror[mirror].dev)) {
Alasdair G Kergon72d94862006-06-26 00:27:35 -0700972 ti->error = "Device lookup failure";
Linus Torvalds1da177e2005-04-16 15:20:36 -0700973 return -ENXIO;
974 }
975
976 ms->mirror[mirror].offset = offset;
977
978 return 0;
979}
980
981static int add_mirror_set(struct mirror_set *ms)
982{
983 down_write(&_mirror_sets_lock);
984 list_add_tail(&ms->list, &_mirror_sets);
985 up_write(&_mirror_sets_lock);
986 wake();
987
988 return 0;
989}
990
991static void del_mirror_set(struct mirror_set *ms)
992{
993 down_write(&_mirror_sets_lock);
994 list_del(&ms->list);
995 up_write(&_mirror_sets_lock);
996}
997
998/*
999 * Create dirty log: log_type #log_params <log_params>
1000 */
1001static struct dirty_log *create_dirty_log(struct dm_target *ti,
1002 unsigned int argc, char **argv,
1003 unsigned int *args_used)
1004{
1005 unsigned int param_count;
1006 struct dirty_log *dl;
1007
1008 if (argc < 2) {
Alasdair G Kergon72d94862006-06-26 00:27:35 -07001009 ti->error = "Insufficient mirror log arguments";
Linus Torvalds1da177e2005-04-16 15:20:36 -07001010 return NULL;
1011 }
1012
1013 if (sscanf(argv[1], "%u", &param_count) != 1) {
Alasdair G Kergon72d94862006-06-26 00:27:35 -07001014 ti->error = "Invalid mirror log argument count";
Linus Torvalds1da177e2005-04-16 15:20:36 -07001015 return NULL;
1016 }
1017
1018 *args_used = 2 + param_count;
1019
1020 if (argc < *args_used) {
Alasdair G Kergon72d94862006-06-26 00:27:35 -07001021 ti->error = "Insufficient mirror log arguments";
Linus Torvalds1da177e2005-04-16 15:20:36 -07001022 return NULL;
1023 }
1024
1025 dl = dm_create_dirty_log(argv[0], ti, param_count, argv + 2);
1026 if (!dl) {
Alasdair G Kergon72d94862006-06-26 00:27:35 -07001027 ti->error = "Error creating mirror dirty log";
Linus Torvalds1da177e2005-04-16 15:20:36 -07001028 return NULL;
1029 }
1030
1031 if (!_check_region_size(ti, dl->type->get_region_size(dl))) {
Alasdair G Kergon72d94862006-06-26 00:27:35 -07001032 ti->error = "Invalid region size";
Linus Torvalds1da177e2005-04-16 15:20:36 -07001033 dm_destroy_dirty_log(dl);
1034 return NULL;
1035 }
1036
1037 return dl;
1038}
1039
1040/*
1041 * Construct a mirror mapping:
1042 *
1043 * log_type #log_params <log_params>
1044 * #mirrors [mirror_path offset]{2,}
1045 *
1046 * log_type is "core" or "disk"
1047 * #log_params is between 1 and 3
1048 */
1049#define DM_IO_PAGES 64
1050static int mirror_ctr(struct dm_target *ti, unsigned int argc, char **argv)
1051{
1052 int r;
1053 unsigned int nr_mirrors, m, args_used;
1054 struct mirror_set *ms;
1055 struct dirty_log *dl;
1056
1057 dl = create_dirty_log(ti, argc, argv, &args_used);
1058 if (!dl)
1059 return -EINVAL;
1060
1061 argv += args_used;
1062 argc -= args_used;
1063
1064 if (!argc || sscanf(argv[0], "%u", &nr_mirrors) != 1 ||
1065 nr_mirrors < 2 || nr_mirrors > KCOPYD_MAX_REGIONS + 1) {
Alasdair G Kergon72d94862006-06-26 00:27:35 -07001066 ti->error = "Invalid number of mirrors";
Linus Torvalds1da177e2005-04-16 15:20:36 -07001067 dm_destroy_dirty_log(dl);
1068 return -EINVAL;
1069 }
1070
1071 argv++, argc--;
1072
1073 if (argc != nr_mirrors * 2) {
Alasdair G Kergon72d94862006-06-26 00:27:35 -07001074 ti->error = "Wrong number of mirror arguments";
Linus Torvalds1da177e2005-04-16 15:20:36 -07001075 dm_destroy_dirty_log(dl);
1076 return -EINVAL;
1077 }
1078
1079 ms = alloc_context(nr_mirrors, dl->type->get_region_size(dl), ti, dl);
1080 if (!ms) {
1081 dm_destroy_dirty_log(dl);
1082 return -ENOMEM;
1083 }
1084
1085 /* Get the mirror parameter sets */
1086 for (m = 0; m < nr_mirrors; m++) {
1087 r = get_mirror(ms, ti, m, argv);
1088 if (r) {
1089 free_context(ms, ti, m);
1090 return r;
1091 }
1092 argv += 2;
1093 argc -= 2;
1094 }
1095
1096 ti->private = ms;
Alasdair G Kergond88854f2005-07-07 17:59:34 -07001097 ti->split_io = ms->rh.region_size;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001098
1099 r = kcopyd_client_create(DM_IO_PAGES, &ms->kcopyd_client);
1100 if (r) {
1101 free_context(ms, ti, ms->nr_mirrors);
1102 return r;
1103 }
1104
1105 add_mirror_set(ms);
1106 return 0;
1107}
1108
1109static void mirror_dtr(struct dm_target *ti)
1110{
1111 struct mirror_set *ms = (struct mirror_set *) ti->private;
1112
1113 del_mirror_set(ms);
1114 kcopyd_client_destroy(ms->kcopyd_client);
1115 free_context(ms, ti, ms->nr_mirrors);
1116}
1117
1118static void queue_bio(struct mirror_set *ms, struct bio *bio, int rw)
1119{
1120 int should_wake = 0;
1121 struct bio_list *bl;
1122
1123 bl = (rw == WRITE) ? &ms->writes : &ms->reads;
1124 spin_lock(&ms->lock);
1125 should_wake = !(bl->head);
1126 bio_list_add(bl, bio);
1127 spin_unlock(&ms->lock);
1128
1129 if (should_wake)
1130 wake();
1131}
1132
1133/*
1134 * Mirror mapping function
1135 */
1136static int mirror_map(struct dm_target *ti, struct bio *bio,
1137 union map_info *map_context)
1138{
1139 int r, rw = bio_rw(bio);
1140 struct mirror *m;
1141 struct mirror_set *ms = ti->private;
1142
Neil Browne4c8b3b2006-06-26 00:27:26 -07001143 map_context->ll = bio_to_region(&ms->rh, bio);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001144
1145 if (rw == WRITE) {
1146 queue_bio(ms, bio, rw);
Kiyoshi Uedad2a7ad22006-12-08 02:41:06 -08001147 return DM_MAPIO_SUBMITTED;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001148 }
1149
1150 r = ms->rh.log->type->in_sync(ms->rh.log,
1151 bio_to_region(&ms->rh, bio), 0);
1152 if (r < 0 && r != -EWOULDBLOCK)
1153 return r;
1154
1155 if (r == -EWOULDBLOCK) /* FIXME: ugly */
Kiyoshi Uedad2a7ad22006-12-08 02:41:06 -08001156 r = DM_MAPIO_SUBMITTED;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001157
1158 /*
1159 * We don't want to fast track a recovery just for a read
1160 * ahead. So we just let it silently fail.
1161 * FIXME: get rid of this.
1162 */
1163 if (!r && rw == READA)
1164 return -EIO;
1165
1166 if (!r) {
1167 /* Pass this io over to the daemon */
1168 queue_bio(ms, bio, rw);
Kiyoshi Uedad2a7ad22006-12-08 02:41:06 -08001169 return DM_MAPIO_SUBMITTED;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001170 }
1171
1172 m = choose_mirror(ms, bio->bi_sector);
1173 if (!m)
1174 return -EIO;
1175
1176 map_bio(ms, m, bio);
Kiyoshi Uedad2a7ad22006-12-08 02:41:06 -08001177 return DM_MAPIO_REMAPPED;
Linus Torvalds1da177e2005-04-16 15:20:36 -07001178}
1179
1180static int mirror_end_io(struct dm_target *ti, struct bio *bio,
1181 int error, union map_info *map_context)
1182{
1183 int rw = bio_rw(bio);
1184 struct mirror_set *ms = (struct mirror_set *) ti->private;
1185 region_t region = map_context->ll;
1186
1187 /*
1188 * We need to dec pending if this was a write.
1189 */
1190 if (rw == WRITE)
1191 rh_dec(&ms->rh, region);
1192
1193 return 0;
1194}
1195
1196static void mirror_postsuspend(struct dm_target *ti)
1197{
1198 struct mirror_set *ms = (struct mirror_set *) ti->private;
1199 struct dirty_log *log = ms->rh.log;
1200
1201 rh_stop_recovery(&ms->rh);
Jonathan E Brassow33184042006-11-08 17:44:44 -08001202
1203 /* Wait for all I/O we generated to complete */
1204 wait_event(_kmirrord_recovery_stopped,
1205 !atomic_read(&ms->rh.recovery_in_flight));
1206
Linus Torvalds1da177e2005-04-16 15:20:36 -07001207 if (log->type->suspend && log->type->suspend(log))
1208 /* FIXME: need better error handling */
1209 DMWARN("log suspend failed");
1210}
1211
1212static void mirror_resume(struct dm_target *ti)
1213{
1214 struct mirror_set *ms = (struct mirror_set *) ti->private;
1215 struct dirty_log *log = ms->rh.log;
1216 if (log->type->resume && log->type->resume(log))
1217 /* FIXME: need better error handling */
1218 DMWARN("log resume failed");
1219 rh_start_recovery(&ms->rh);
1220}
1221
1222static int mirror_status(struct dm_target *ti, status_type_t type,
1223 char *result, unsigned int maxlen)
1224{
1225 unsigned int m, sz;
1226 struct mirror_set *ms = (struct mirror_set *) ti->private;
1227
1228 sz = ms->rh.log->type->status(ms->rh.log, type, result, maxlen);
1229
1230 switch (type) {
1231 case STATUSTYPE_INFO:
1232 DMEMIT("%d ", ms->nr_mirrors);
1233 for (m = 0; m < ms->nr_mirrors; m++)
1234 DMEMIT("%s ", ms->mirror[m].dev->name);
1235
Andrew Morton4ee218c2006-03-27 01:17:48 -08001236 DMEMIT("%llu/%llu",
1237 (unsigned long long)ms->rh.log->type->
1238 get_sync_count(ms->rh.log),
1239 (unsigned long long)ms->nr_regions);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001240 break;
1241
1242 case STATUSTYPE_TABLE:
Jonathan Brassowe52b8f62006-10-03 01:15:32 -07001243 DMEMIT("%d", ms->nr_mirrors);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001244 for (m = 0; m < ms->nr_mirrors; m++)
Jonathan Brassowe52b8f62006-10-03 01:15:32 -07001245 DMEMIT(" %s %llu", ms->mirror[m].dev->name,
Andrew Morton4ee218c2006-03-27 01:17:48 -08001246 (unsigned long long)ms->mirror[m].offset);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001247 }
1248
1249 return 0;
1250}
1251
1252static struct target_type mirror_target = {
1253 .name = "mirror",
Kevin Corry702ca6f2006-06-26 00:27:28 -07001254 .version = {1, 0, 2},
Linus Torvalds1da177e2005-04-16 15:20:36 -07001255 .module = THIS_MODULE,
1256 .ctr = mirror_ctr,
1257 .dtr = mirror_dtr,
1258 .map = mirror_map,
1259 .end_io = mirror_end_io,
1260 .postsuspend = mirror_postsuspend,
1261 .resume = mirror_resume,
1262 .status = mirror_status,
1263};
1264
1265static int __init dm_mirror_init(void)
1266{
1267 int r;
1268
1269 r = dm_dirty_log_init();
1270 if (r)
1271 return r;
1272
Alasdair G Kergon48f1f532005-08-04 12:53:37 -07001273 _kmirrord_wq = create_singlethread_workqueue("kmirrord");
Linus Torvalds1da177e2005-04-16 15:20:36 -07001274 if (!_kmirrord_wq) {
1275 DMERR("couldn't start kmirrord");
1276 dm_dirty_log_exit();
1277 return r;
1278 }
David Howellsc4028952006-11-22 14:57:56 +00001279 INIT_WORK(&_kmirrord_work, do_work);
Linus Torvalds1da177e2005-04-16 15:20:36 -07001280
1281 r = dm_register_target(&mirror_target);
1282 if (r < 0) {
1283 DMERR("%s: Failed to register mirror target",
1284 mirror_target.name);
1285 dm_dirty_log_exit();
1286 destroy_workqueue(_kmirrord_wq);
1287 }
1288
1289 return r;
1290}
1291
1292static void __exit dm_mirror_exit(void)
1293{
1294 int r;
1295
1296 r = dm_unregister_target(&mirror_target);
1297 if (r < 0)
1298 DMERR("%s: unregister failed %d", mirror_target.name, r);
1299
1300 destroy_workqueue(_kmirrord_wq);
1301 dm_dirty_log_exit();
1302}
1303
1304/* Module hooks */
1305module_init(dm_mirror_init);
1306module_exit(dm_mirror_exit);
1307
1308MODULE_DESCRIPTION(DM_NAME " mirror target");
1309MODULE_AUTHOR("Joe Thornber");
1310MODULE_LICENSE("GPL");