xref: /linux/drivers/md/dm-kcopyd.c (revision 8c749ce93ee69e789e46b3be98de9e0cbfcf8ed8)
1 /*
2  * Copyright (C) 2002 Sistina Software (UK) Limited.
3  * Copyright (C) 2006 Red Hat GmbH
4  *
5  * This file is released under the GPL.
6  *
7  * Kcopyd provides a simple interface for copying an area of one
8  * block-device to one or more other block-devices, with an asynchronous
9  * completion notification.
10  */
11 
12 #include <linux/types.h>
13 #include <linux/atomic.h>
14 #include <linux/blkdev.h>
15 #include <linux/fs.h>
16 #include <linux/init.h>
17 #include <linux/list.h>
18 #include <linux/mempool.h>
19 #include <linux/module.h>
20 #include <linux/pagemap.h>
21 #include <linux/slab.h>
22 #include <linux/vmalloc.h>
23 #include <linux/workqueue.h>
24 #include <linux/mutex.h>
25 #include <linux/delay.h>
26 #include <linux/device-mapper.h>
27 #include <linux/dm-kcopyd.h>
28 
29 #include "dm.h"
30 
31 #define SUB_JOB_SIZE	128
32 #define SPLIT_COUNT	8
33 #define MIN_JOBS	8
34 #define RESERVE_PAGES	(DIV_ROUND_UP(SUB_JOB_SIZE << SECTOR_SHIFT, PAGE_SIZE))
35 
36 /*-----------------------------------------------------------------
37  * Each kcopyd client has its own little pool of preallocated
38  * pages for kcopyd io.
39  *---------------------------------------------------------------*/
40 struct dm_kcopyd_client {
41 	struct page_list *pages;
42 	unsigned nr_reserved_pages;
43 	unsigned nr_free_pages;
44 
45 	struct dm_io_client *io_client;
46 
47 	wait_queue_head_t destroyq;
48 	atomic_t nr_jobs;
49 
50 	mempool_t *job_pool;
51 
52 	struct workqueue_struct *kcopyd_wq;
53 	struct work_struct kcopyd_work;
54 
55 	struct dm_kcopyd_throttle *throttle;
56 
57 /*
58  * We maintain three lists of jobs:
59  *
60  * i)   jobs waiting for pages
61  * ii)  jobs that have pages, and are waiting for the io to be issued.
62  * iii) jobs that have completed.
63  *
64  * All three of these are protected by job_lock.
65  */
66 	spinlock_t job_lock;
67 	struct list_head complete_jobs;
68 	struct list_head io_jobs;
69 	struct list_head pages_jobs;
70 };
71 
72 static struct page_list zero_page_list;
73 
74 static DEFINE_SPINLOCK(throttle_spinlock);
75 
76 /*
77  * IO/IDLE accounting slowly decays after (1 << ACCOUNT_INTERVAL_SHIFT) period.
78  * When total_period >= (1 << ACCOUNT_INTERVAL_SHIFT) the counters are divided
79  * by 2.
80  */
81 #define ACCOUNT_INTERVAL_SHIFT		SHIFT_HZ
82 
83 /*
84  * Sleep this number of milliseconds.
85  *
86  * The value was decided experimentally.
87  * Smaller values seem to cause an increased copy rate above the limit.
88  * The reason for this is unknown but possibly due to jiffies rounding errors
89  * or read/write cache inside the disk.
90  */
91 #define SLEEP_MSEC			100
92 
93 /*
94  * Maximum number of sleep events. There is a theoretical livelock if more
95  * kcopyd clients do work simultaneously which this limit avoids.
96  */
97 #define MAX_SLEEPS			10
98 
99 static void io_job_start(struct dm_kcopyd_throttle *t)
100 {
101 	unsigned throttle, now, difference;
102 	int slept = 0, skew;
103 
104 	if (unlikely(!t))
105 		return;
106 
107 try_again:
108 	spin_lock_irq(&throttle_spinlock);
109 
110 	throttle = ACCESS_ONCE(t->throttle);
111 
112 	if (likely(throttle >= 100))
113 		goto skip_limit;
114 
115 	now = jiffies;
116 	difference = now - t->last_jiffies;
117 	t->last_jiffies = now;
118 	if (t->num_io_jobs)
119 		t->io_period += difference;
120 	t->total_period += difference;
121 
122 	/*
123 	 * Maintain sane values if we got a temporary overflow.
124 	 */
125 	if (unlikely(t->io_period > t->total_period))
126 		t->io_period = t->total_period;
127 
128 	if (unlikely(t->total_period >= (1 << ACCOUNT_INTERVAL_SHIFT))) {
129 		int shift = fls(t->total_period >> ACCOUNT_INTERVAL_SHIFT);
130 		t->total_period >>= shift;
131 		t->io_period >>= shift;
132 	}
133 
134 	skew = t->io_period - throttle * t->total_period / 100;
135 
136 	if (unlikely(skew > 0) && slept < MAX_SLEEPS) {
137 		slept++;
138 		spin_unlock_irq(&throttle_spinlock);
139 		msleep(SLEEP_MSEC);
140 		goto try_again;
141 	}
142 
143 skip_limit:
144 	t->num_io_jobs++;
145 
146 	spin_unlock_irq(&throttle_spinlock);
147 }
148 
149 static void io_job_finish(struct dm_kcopyd_throttle *t)
150 {
151 	unsigned long flags;
152 
153 	if (unlikely(!t))
154 		return;
155 
156 	spin_lock_irqsave(&throttle_spinlock, flags);
157 
158 	t->num_io_jobs--;
159 
160 	if (likely(ACCESS_ONCE(t->throttle) >= 100))
161 		goto skip_limit;
162 
163 	if (!t->num_io_jobs) {
164 		unsigned now, difference;
165 
166 		now = jiffies;
167 		difference = now - t->last_jiffies;
168 		t->last_jiffies = now;
169 
170 		t->io_period += difference;
171 		t->total_period += difference;
172 
173 		/*
174 		 * Maintain sane values if we got a temporary overflow.
175 		 */
176 		if (unlikely(t->io_period > t->total_period))
177 			t->io_period = t->total_period;
178 	}
179 
180 skip_limit:
181 	spin_unlock_irqrestore(&throttle_spinlock, flags);
182 }
183 
184 
185 static void wake(struct dm_kcopyd_client *kc)
186 {
187 	queue_work(kc->kcopyd_wq, &kc->kcopyd_work);
188 }
189 
190 /*
191  * Obtain one page for the use of kcopyd.
192  */
193 static struct page_list *alloc_pl(gfp_t gfp)
194 {
195 	struct page_list *pl;
196 
197 	pl = kmalloc(sizeof(*pl), gfp);
198 	if (!pl)
199 		return NULL;
200 
201 	pl->page = alloc_page(gfp);
202 	if (!pl->page) {
203 		kfree(pl);
204 		return NULL;
205 	}
206 
207 	return pl;
208 }
209 
210 static void free_pl(struct page_list *pl)
211 {
212 	__free_page(pl->page);
213 	kfree(pl);
214 }
215 
216 /*
217  * Add the provided pages to a client's free page list, releasing
218  * back to the system any beyond the reserved_pages limit.
219  */
220 static void kcopyd_put_pages(struct dm_kcopyd_client *kc, struct page_list *pl)
221 {
222 	struct page_list *next;
223 
224 	do {
225 		next = pl->next;
226 
227 		if (kc->nr_free_pages >= kc->nr_reserved_pages)
228 			free_pl(pl);
229 		else {
230 			pl->next = kc->pages;
231 			kc->pages = pl;
232 			kc->nr_free_pages++;
233 		}
234 
235 		pl = next;
236 	} while (pl);
237 }
238 
239 static int kcopyd_get_pages(struct dm_kcopyd_client *kc,
240 			    unsigned int nr, struct page_list **pages)
241 {
242 	struct page_list *pl;
243 
244 	*pages = NULL;
245 
246 	do {
247 		pl = alloc_pl(__GFP_NOWARN | __GFP_NORETRY | __GFP_KSWAPD_RECLAIM);
248 		if (unlikely(!pl)) {
249 			/* Use reserved pages */
250 			pl = kc->pages;
251 			if (unlikely(!pl))
252 				goto out_of_memory;
253 			kc->pages = pl->next;
254 			kc->nr_free_pages--;
255 		}
256 		pl->next = *pages;
257 		*pages = pl;
258 	} while (--nr);
259 
260 	return 0;
261 
262 out_of_memory:
263 	if (*pages)
264 		kcopyd_put_pages(kc, *pages);
265 	return -ENOMEM;
266 }
267 
268 /*
269  * These three functions resize the page pool.
270  */
271 static void drop_pages(struct page_list *pl)
272 {
273 	struct page_list *next;
274 
275 	while (pl) {
276 		next = pl->next;
277 		free_pl(pl);
278 		pl = next;
279 	}
280 }
281 
282 /*
283  * Allocate and reserve nr_pages for the use of a specific client.
284  */
285 static int client_reserve_pages(struct dm_kcopyd_client *kc, unsigned nr_pages)
286 {
287 	unsigned i;
288 	struct page_list *pl = NULL, *next;
289 
290 	for (i = 0; i < nr_pages; i++) {
291 		next = alloc_pl(GFP_KERNEL);
292 		if (!next) {
293 			if (pl)
294 				drop_pages(pl);
295 			return -ENOMEM;
296 		}
297 		next->next = pl;
298 		pl = next;
299 	}
300 
301 	kc->nr_reserved_pages += nr_pages;
302 	kcopyd_put_pages(kc, pl);
303 
304 	return 0;
305 }
306 
307 static void client_free_pages(struct dm_kcopyd_client *kc)
308 {
309 	BUG_ON(kc->nr_free_pages != kc->nr_reserved_pages);
310 	drop_pages(kc->pages);
311 	kc->pages = NULL;
312 	kc->nr_free_pages = kc->nr_reserved_pages = 0;
313 }
314 
315 /*-----------------------------------------------------------------
316  * kcopyd_jobs need to be allocated by the *clients* of kcopyd,
317  * for this reason we use a mempool to prevent the client from
318  * ever having to do io (which could cause a deadlock).
319  *---------------------------------------------------------------*/
320 struct kcopyd_job {
321 	struct dm_kcopyd_client *kc;
322 	struct list_head list;
323 	unsigned long flags;
324 
325 	/*
326 	 * Error state of the job.
327 	 */
328 	int read_err;
329 	unsigned long write_err;
330 
331 	/*
332 	 * Either READ or WRITE
333 	 */
334 	int rw;
335 	struct dm_io_region source;
336 
337 	/*
338 	 * The destinations for the transfer.
339 	 */
340 	unsigned int num_dests;
341 	struct dm_io_region dests[DM_KCOPYD_MAX_REGIONS];
342 
343 	struct page_list *pages;
344 
345 	/*
346 	 * Set this to ensure you are notified when the job has
347 	 * completed.  'context' is for callback to use.
348 	 */
349 	dm_kcopyd_notify_fn fn;
350 	void *context;
351 
352 	/*
353 	 * These fields are only used if the job has been split
354 	 * into more manageable parts.
355 	 */
356 	struct mutex lock;
357 	atomic_t sub_jobs;
358 	sector_t progress;
359 
360 	struct kcopyd_job *master_job;
361 };
362 
363 static struct kmem_cache *_job_cache;
364 
365 int __init dm_kcopyd_init(void)
366 {
367 	_job_cache = kmem_cache_create("kcopyd_job",
368 				sizeof(struct kcopyd_job) * (SPLIT_COUNT + 1),
369 				__alignof__(struct kcopyd_job), 0, NULL);
370 	if (!_job_cache)
371 		return -ENOMEM;
372 
373 	zero_page_list.next = &zero_page_list;
374 	zero_page_list.page = ZERO_PAGE(0);
375 
376 	return 0;
377 }
378 
379 void dm_kcopyd_exit(void)
380 {
381 	kmem_cache_destroy(_job_cache);
382 	_job_cache = NULL;
383 }
384 
385 /*
386  * Functions to push and pop a job onto the head of a given job
387  * list.
388  */
389 static struct kcopyd_job *pop(struct list_head *jobs,
390 			      struct dm_kcopyd_client *kc)
391 {
392 	struct kcopyd_job *job = NULL;
393 	unsigned long flags;
394 
395 	spin_lock_irqsave(&kc->job_lock, flags);
396 
397 	if (!list_empty(jobs)) {
398 		job = list_entry(jobs->next, struct kcopyd_job, list);
399 		list_del(&job->list);
400 	}
401 	spin_unlock_irqrestore(&kc->job_lock, flags);
402 
403 	return job;
404 }
405 
406 static void push(struct list_head *jobs, struct kcopyd_job *job)
407 {
408 	unsigned long flags;
409 	struct dm_kcopyd_client *kc = job->kc;
410 
411 	spin_lock_irqsave(&kc->job_lock, flags);
412 	list_add_tail(&job->list, jobs);
413 	spin_unlock_irqrestore(&kc->job_lock, flags);
414 }
415 
416 
417 static void push_head(struct list_head *jobs, struct kcopyd_job *job)
418 {
419 	unsigned long flags;
420 	struct dm_kcopyd_client *kc = job->kc;
421 
422 	spin_lock_irqsave(&kc->job_lock, flags);
423 	list_add(&job->list, jobs);
424 	spin_unlock_irqrestore(&kc->job_lock, flags);
425 }
426 
427 /*
428  * These three functions process 1 item from the corresponding
429  * job list.
430  *
431  * They return:
432  * < 0: error
433  *   0: success
434  * > 0: can't process yet.
435  */
436 static int run_complete_job(struct kcopyd_job *job)
437 {
438 	void *context = job->context;
439 	int read_err = job->read_err;
440 	unsigned long write_err = job->write_err;
441 	dm_kcopyd_notify_fn fn = job->fn;
442 	struct dm_kcopyd_client *kc = job->kc;
443 
444 	if (job->pages && job->pages != &zero_page_list)
445 		kcopyd_put_pages(kc, job->pages);
446 	/*
447 	 * If this is the master job, the sub jobs have already
448 	 * completed so we can free everything.
449 	 */
450 	if (job->master_job == job)
451 		mempool_free(job, kc->job_pool);
452 	fn(read_err, write_err, context);
453 
454 	if (atomic_dec_and_test(&kc->nr_jobs))
455 		wake_up(&kc->destroyq);
456 
457 	return 0;
458 }
459 
460 static void complete_io(unsigned long error, void *context)
461 {
462 	struct kcopyd_job *job = (struct kcopyd_job *) context;
463 	struct dm_kcopyd_client *kc = job->kc;
464 
465 	io_job_finish(kc->throttle);
466 
467 	if (error) {
468 		if (job->rw & WRITE)
469 			job->write_err |= error;
470 		else
471 			job->read_err = 1;
472 
473 		if (!test_bit(DM_KCOPYD_IGNORE_ERROR, &job->flags)) {
474 			push(&kc->complete_jobs, job);
475 			wake(kc);
476 			return;
477 		}
478 	}
479 
480 	if (job->rw & WRITE)
481 		push(&kc->complete_jobs, job);
482 
483 	else {
484 		job->rw = WRITE;
485 		push(&kc->io_jobs, job);
486 	}
487 
488 	wake(kc);
489 }
490 
491 /*
492  * Request io on as many buffer heads as we can currently get for
493  * a particular job.
494  */
495 static int run_io_job(struct kcopyd_job *job)
496 {
497 	int r;
498 	struct dm_io_request io_req = {
499 		.bi_rw = job->rw,
500 		.mem.type = DM_IO_PAGE_LIST,
501 		.mem.ptr.pl = job->pages,
502 		.mem.offset = 0,
503 		.notify.fn = complete_io,
504 		.notify.context = job,
505 		.client = job->kc->io_client,
506 	};
507 
508 	io_job_start(job->kc->throttle);
509 
510 	if (job->rw == READ)
511 		r = dm_io(&io_req, 1, &job->source, NULL);
512 	else
513 		r = dm_io(&io_req, job->num_dests, job->dests, NULL);
514 
515 	return r;
516 }
517 
518 static int run_pages_job(struct kcopyd_job *job)
519 {
520 	int r;
521 	unsigned nr_pages = dm_div_up(job->dests[0].count, PAGE_SIZE >> 9);
522 
523 	r = kcopyd_get_pages(job->kc, nr_pages, &job->pages);
524 	if (!r) {
525 		/* this job is ready for io */
526 		push(&job->kc->io_jobs, job);
527 		return 0;
528 	}
529 
530 	if (r == -ENOMEM)
531 		/* can't complete now */
532 		return 1;
533 
534 	return r;
535 }
536 
537 /*
538  * Run through a list for as long as possible.  Returns the count
539  * of successful jobs.
540  */
541 static int process_jobs(struct list_head *jobs, struct dm_kcopyd_client *kc,
542 			int (*fn) (struct kcopyd_job *))
543 {
544 	struct kcopyd_job *job;
545 	int r, count = 0;
546 
547 	while ((job = pop(jobs, kc))) {
548 
549 		r = fn(job);
550 
551 		if (r < 0) {
552 			/* error this rogue job */
553 			if (job->rw & WRITE)
554 				job->write_err = (unsigned long) -1L;
555 			else
556 				job->read_err = 1;
557 			push(&kc->complete_jobs, job);
558 			break;
559 		}
560 
561 		if (r > 0) {
562 			/*
563 			 * We couldn't service this job ATM, so
564 			 * push this job back onto the list.
565 			 */
566 			push_head(jobs, job);
567 			break;
568 		}
569 
570 		count++;
571 	}
572 
573 	return count;
574 }
575 
576 /*
577  * kcopyd does this every time it's woken up.
578  */
579 static void do_work(struct work_struct *work)
580 {
581 	struct dm_kcopyd_client *kc = container_of(work,
582 					struct dm_kcopyd_client, kcopyd_work);
583 	struct blk_plug plug;
584 
585 	/*
586 	 * The order that these are called is *very* important.
587 	 * complete jobs can free some pages for pages jobs.
588 	 * Pages jobs when successful will jump onto the io jobs
589 	 * list.  io jobs call wake when they complete and it all
590 	 * starts again.
591 	 */
592 	blk_start_plug(&plug);
593 	process_jobs(&kc->complete_jobs, kc, run_complete_job);
594 	process_jobs(&kc->pages_jobs, kc, run_pages_job);
595 	process_jobs(&kc->io_jobs, kc, run_io_job);
596 	blk_finish_plug(&plug);
597 }
598 
599 /*
600  * If we are copying a small region we just dispatch a single job
601  * to do the copy, otherwise the io has to be split up into many
602  * jobs.
603  */
604 static void dispatch_job(struct kcopyd_job *job)
605 {
606 	struct dm_kcopyd_client *kc = job->kc;
607 	atomic_inc(&kc->nr_jobs);
608 	if (unlikely(!job->source.count))
609 		push(&kc->complete_jobs, job);
610 	else if (job->pages == &zero_page_list)
611 		push(&kc->io_jobs, job);
612 	else
613 		push(&kc->pages_jobs, job);
614 	wake(kc);
615 }
616 
617 static void segment_complete(int read_err, unsigned long write_err,
618 			     void *context)
619 {
620 	/* FIXME: tidy this function */
621 	sector_t progress = 0;
622 	sector_t count = 0;
623 	struct kcopyd_job *sub_job = (struct kcopyd_job *) context;
624 	struct kcopyd_job *job = sub_job->master_job;
625 	struct dm_kcopyd_client *kc = job->kc;
626 
627 	mutex_lock(&job->lock);
628 
629 	/* update the error */
630 	if (read_err)
631 		job->read_err = 1;
632 
633 	if (write_err)
634 		job->write_err |= write_err;
635 
636 	/*
637 	 * Only dispatch more work if there hasn't been an error.
638 	 */
639 	if ((!job->read_err && !job->write_err) ||
640 	    test_bit(DM_KCOPYD_IGNORE_ERROR, &job->flags)) {
641 		/* get the next chunk of work */
642 		progress = job->progress;
643 		count = job->source.count - progress;
644 		if (count) {
645 			if (count > SUB_JOB_SIZE)
646 				count = SUB_JOB_SIZE;
647 
648 			job->progress += count;
649 		}
650 	}
651 	mutex_unlock(&job->lock);
652 
653 	if (count) {
654 		int i;
655 
656 		*sub_job = *job;
657 		sub_job->source.sector += progress;
658 		sub_job->source.count = count;
659 
660 		for (i = 0; i < job->num_dests; i++) {
661 			sub_job->dests[i].sector += progress;
662 			sub_job->dests[i].count = count;
663 		}
664 
665 		sub_job->fn = segment_complete;
666 		sub_job->context = sub_job;
667 		dispatch_job(sub_job);
668 
669 	} else if (atomic_dec_and_test(&job->sub_jobs)) {
670 
671 		/*
672 		 * Queue the completion callback to the kcopyd thread.
673 		 *
674 		 * Some callers assume that all the completions are called
675 		 * from a single thread and don't race with each other.
676 		 *
677 		 * We must not call the callback directly here because this
678 		 * code may not be executing in the thread.
679 		 */
680 		push(&kc->complete_jobs, job);
681 		wake(kc);
682 	}
683 }
684 
685 /*
686  * Create some sub jobs to share the work between them.
687  */
688 static void split_job(struct kcopyd_job *master_job)
689 {
690 	int i;
691 
692 	atomic_inc(&master_job->kc->nr_jobs);
693 
694 	atomic_set(&master_job->sub_jobs, SPLIT_COUNT);
695 	for (i = 0; i < SPLIT_COUNT; i++) {
696 		master_job[i + 1].master_job = master_job;
697 		segment_complete(0, 0u, &master_job[i + 1]);
698 	}
699 }
700 
701 int dm_kcopyd_copy(struct dm_kcopyd_client *kc, struct dm_io_region *from,
702 		   unsigned int num_dests, struct dm_io_region *dests,
703 		   unsigned int flags, dm_kcopyd_notify_fn fn, void *context)
704 {
705 	struct kcopyd_job *job;
706 	int i;
707 
708 	/*
709 	 * Allocate an array of jobs consisting of one master job
710 	 * followed by SPLIT_COUNT sub jobs.
711 	 */
712 	job = mempool_alloc(kc->job_pool, GFP_NOIO);
713 
714 	/*
715 	 * set up for the read.
716 	 */
717 	job->kc = kc;
718 	job->flags = flags;
719 	job->read_err = 0;
720 	job->write_err = 0;
721 
722 	job->num_dests = num_dests;
723 	memcpy(&job->dests, dests, sizeof(*dests) * num_dests);
724 
725 	if (from) {
726 		job->source = *from;
727 		job->pages = NULL;
728 		job->rw = READ;
729 	} else {
730 		memset(&job->source, 0, sizeof job->source);
731 		job->source.count = job->dests[0].count;
732 		job->pages = &zero_page_list;
733 
734 		/*
735 		 * Use WRITE SAME to optimize zeroing if all dests support it.
736 		 */
737 		job->rw = WRITE | REQ_WRITE_SAME;
738 		for (i = 0; i < job->num_dests; i++)
739 			if (!bdev_write_same(job->dests[i].bdev)) {
740 				job->rw = WRITE;
741 				break;
742 			}
743 	}
744 
745 	job->fn = fn;
746 	job->context = context;
747 	job->master_job = job;
748 
749 	if (job->source.count <= SUB_JOB_SIZE)
750 		dispatch_job(job);
751 	else {
752 		mutex_init(&job->lock);
753 		job->progress = 0;
754 		split_job(job);
755 	}
756 
757 	return 0;
758 }
759 EXPORT_SYMBOL(dm_kcopyd_copy);
760 
761 int dm_kcopyd_zero(struct dm_kcopyd_client *kc,
762 		   unsigned num_dests, struct dm_io_region *dests,
763 		   unsigned flags, dm_kcopyd_notify_fn fn, void *context)
764 {
765 	return dm_kcopyd_copy(kc, NULL, num_dests, dests, flags, fn, context);
766 }
767 EXPORT_SYMBOL(dm_kcopyd_zero);
768 
769 void *dm_kcopyd_prepare_callback(struct dm_kcopyd_client *kc,
770 				 dm_kcopyd_notify_fn fn, void *context)
771 {
772 	struct kcopyd_job *job;
773 
774 	job = mempool_alloc(kc->job_pool, GFP_NOIO);
775 
776 	memset(job, 0, sizeof(struct kcopyd_job));
777 	job->kc = kc;
778 	job->fn = fn;
779 	job->context = context;
780 	job->master_job = job;
781 
782 	atomic_inc(&kc->nr_jobs);
783 
784 	return job;
785 }
786 EXPORT_SYMBOL(dm_kcopyd_prepare_callback);
787 
788 void dm_kcopyd_do_callback(void *j, int read_err, unsigned long write_err)
789 {
790 	struct kcopyd_job *job = j;
791 	struct dm_kcopyd_client *kc = job->kc;
792 
793 	job->read_err = read_err;
794 	job->write_err = write_err;
795 
796 	push(&kc->complete_jobs, job);
797 	wake(kc);
798 }
799 EXPORT_SYMBOL(dm_kcopyd_do_callback);
800 
801 /*
802  * Cancels a kcopyd job, eg. someone might be deactivating a
803  * mirror.
804  */
805 #if 0
806 int kcopyd_cancel(struct kcopyd_job *job, int block)
807 {
808 	/* FIXME: finish */
809 	return -1;
810 }
811 #endif  /*  0  */
812 
813 /*-----------------------------------------------------------------
814  * Client setup
815  *---------------------------------------------------------------*/
816 struct dm_kcopyd_client *dm_kcopyd_client_create(struct dm_kcopyd_throttle *throttle)
817 {
818 	int r = -ENOMEM;
819 	struct dm_kcopyd_client *kc;
820 
821 	kc = kmalloc(sizeof(*kc), GFP_KERNEL);
822 	if (!kc)
823 		return ERR_PTR(-ENOMEM);
824 
825 	spin_lock_init(&kc->job_lock);
826 	INIT_LIST_HEAD(&kc->complete_jobs);
827 	INIT_LIST_HEAD(&kc->io_jobs);
828 	INIT_LIST_HEAD(&kc->pages_jobs);
829 	kc->throttle = throttle;
830 
831 	kc->job_pool = mempool_create_slab_pool(MIN_JOBS, _job_cache);
832 	if (!kc->job_pool)
833 		goto bad_slab;
834 
835 	INIT_WORK(&kc->kcopyd_work, do_work);
836 	kc->kcopyd_wq = alloc_workqueue("kcopyd", WQ_MEM_RECLAIM, 0);
837 	if (!kc->kcopyd_wq)
838 		goto bad_workqueue;
839 
840 	kc->pages = NULL;
841 	kc->nr_reserved_pages = kc->nr_free_pages = 0;
842 	r = client_reserve_pages(kc, RESERVE_PAGES);
843 	if (r)
844 		goto bad_client_pages;
845 
846 	kc->io_client = dm_io_client_create();
847 	if (IS_ERR(kc->io_client)) {
848 		r = PTR_ERR(kc->io_client);
849 		goto bad_io_client;
850 	}
851 
852 	init_waitqueue_head(&kc->destroyq);
853 	atomic_set(&kc->nr_jobs, 0);
854 
855 	return kc;
856 
857 bad_io_client:
858 	client_free_pages(kc);
859 bad_client_pages:
860 	destroy_workqueue(kc->kcopyd_wq);
861 bad_workqueue:
862 	mempool_destroy(kc->job_pool);
863 bad_slab:
864 	kfree(kc);
865 
866 	return ERR_PTR(r);
867 }
868 EXPORT_SYMBOL(dm_kcopyd_client_create);
869 
870 void dm_kcopyd_client_destroy(struct dm_kcopyd_client *kc)
871 {
872 	/* Wait for completion of all jobs submitted by this client. */
873 	wait_event(kc->destroyq, !atomic_read(&kc->nr_jobs));
874 
875 	BUG_ON(!list_empty(&kc->complete_jobs));
876 	BUG_ON(!list_empty(&kc->io_jobs));
877 	BUG_ON(!list_empty(&kc->pages_jobs));
878 	destroy_workqueue(kc->kcopyd_wq);
879 	dm_io_client_destroy(kc->io_client);
880 	client_free_pages(kc);
881 	mempool_destroy(kc->job_pool);
882 	kfree(kc);
883 }
884 EXPORT_SYMBOL(dm_kcopyd_client_destroy);
885