xref: /linux/drivers/md/dm-kcopyd.c (revision c79c3c34f75d72a066e292b10aa50fc758c97c89)
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-core.h"
30 
31 #define SPLIT_COUNT	8
32 #define MIN_JOBS	8
33 
34 #define DEFAULT_SUB_JOB_SIZE_KB 512
35 #define MAX_SUB_JOB_SIZE_KB     1024
36 
37 static unsigned kcopyd_subjob_size_kb = DEFAULT_SUB_JOB_SIZE_KB;
38 
39 module_param(kcopyd_subjob_size_kb, uint, S_IRUGO | S_IWUSR);
40 MODULE_PARM_DESC(kcopyd_subjob_size_kb, "Sub-job size for dm-kcopyd clients");
41 
42 static unsigned dm_get_kcopyd_subjob_size(void)
43 {
44 	unsigned sub_job_size_kb;
45 
46 	sub_job_size_kb = __dm_get_module_param(&kcopyd_subjob_size_kb,
47 						DEFAULT_SUB_JOB_SIZE_KB,
48 						MAX_SUB_JOB_SIZE_KB);
49 
50 	return sub_job_size_kb << 1;
51 }
52 
53 /*-----------------------------------------------------------------
54  * Each kcopyd client has its own little pool of preallocated
55  * pages for kcopyd io.
56  *---------------------------------------------------------------*/
57 struct dm_kcopyd_client {
58 	struct page_list *pages;
59 	unsigned nr_reserved_pages;
60 	unsigned nr_free_pages;
61 	unsigned sub_job_size;
62 
63 	struct dm_io_client *io_client;
64 
65 	wait_queue_head_t destroyq;
66 
67 	mempool_t job_pool;
68 
69 	struct workqueue_struct *kcopyd_wq;
70 	struct work_struct kcopyd_work;
71 
72 	struct dm_kcopyd_throttle *throttle;
73 
74 	atomic_t nr_jobs;
75 
76 /*
77  * We maintain four lists of jobs:
78  *
79  * i)   jobs waiting for pages
80  * ii)  jobs that have pages, and are waiting for the io to be issued.
81  * iii) jobs that don't need to do any IO and just run a callback
82  * iv) jobs that have completed.
83  *
84  * All four of these are protected by job_lock.
85  */
86 	spinlock_t job_lock;
87 	struct list_head callback_jobs;
88 	struct list_head complete_jobs;
89 	struct list_head io_jobs;
90 	struct list_head pages_jobs;
91 };
92 
93 static struct page_list zero_page_list;
94 
95 static DEFINE_SPINLOCK(throttle_spinlock);
96 
97 /*
98  * IO/IDLE accounting slowly decays after (1 << ACCOUNT_INTERVAL_SHIFT) period.
99  * When total_period >= (1 << ACCOUNT_INTERVAL_SHIFT) the counters are divided
100  * by 2.
101  */
102 #define ACCOUNT_INTERVAL_SHIFT		SHIFT_HZ
103 
104 /*
105  * Sleep this number of milliseconds.
106  *
107  * The value was decided experimentally.
108  * Smaller values seem to cause an increased copy rate above the limit.
109  * The reason for this is unknown but possibly due to jiffies rounding errors
110  * or read/write cache inside the disk.
111  */
112 #define SLEEP_MSEC			100
113 
114 /*
115  * Maximum number of sleep events. There is a theoretical livelock if more
116  * kcopyd clients do work simultaneously which this limit avoids.
117  */
118 #define MAX_SLEEPS			10
119 
120 static void io_job_start(struct dm_kcopyd_throttle *t)
121 {
122 	unsigned throttle, now, difference;
123 	int slept = 0, skew;
124 
125 	if (unlikely(!t))
126 		return;
127 
128 try_again:
129 	spin_lock_irq(&throttle_spinlock);
130 
131 	throttle = READ_ONCE(t->throttle);
132 
133 	if (likely(throttle >= 100))
134 		goto skip_limit;
135 
136 	now = jiffies;
137 	difference = now - t->last_jiffies;
138 	t->last_jiffies = now;
139 	if (t->num_io_jobs)
140 		t->io_period += difference;
141 	t->total_period += difference;
142 
143 	/*
144 	 * Maintain sane values if we got a temporary overflow.
145 	 */
146 	if (unlikely(t->io_period > t->total_period))
147 		t->io_period = t->total_period;
148 
149 	if (unlikely(t->total_period >= (1 << ACCOUNT_INTERVAL_SHIFT))) {
150 		int shift = fls(t->total_period >> ACCOUNT_INTERVAL_SHIFT);
151 		t->total_period >>= shift;
152 		t->io_period >>= shift;
153 	}
154 
155 	skew = t->io_period - throttle * t->total_period / 100;
156 
157 	if (unlikely(skew > 0) && slept < MAX_SLEEPS) {
158 		slept++;
159 		spin_unlock_irq(&throttle_spinlock);
160 		msleep(SLEEP_MSEC);
161 		goto try_again;
162 	}
163 
164 skip_limit:
165 	t->num_io_jobs++;
166 
167 	spin_unlock_irq(&throttle_spinlock);
168 }
169 
170 static void io_job_finish(struct dm_kcopyd_throttle *t)
171 {
172 	unsigned long flags;
173 
174 	if (unlikely(!t))
175 		return;
176 
177 	spin_lock_irqsave(&throttle_spinlock, flags);
178 
179 	t->num_io_jobs--;
180 
181 	if (likely(READ_ONCE(t->throttle) >= 100))
182 		goto skip_limit;
183 
184 	if (!t->num_io_jobs) {
185 		unsigned now, difference;
186 
187 		now = jiffies;
188 		difference = now - t->last_jiffies;
189 		t->last_jiffies = now;
190 
191 		t->io_period += difference;
192 		t->total_period += difference;
193 
194 		/*
195 		 * Maintain sane values if we got a temporary overflow.
196 		 */
197 		if (unlikely(t->io_period > t->total_period))
198 			t->io_period = t->total_period;
199 	}
200 
201 skip_limit:
202 	spin_unlock_irqrestore(&throttle_spinlock, flags);
203 }
204 
205 
206 static void wake(struct dm_kcopyd_client *kc)
207 {
208 	queue_work(kc->kcopyd_wq, &kc->kcopyd_work);
209 }
210 
211 /*
212  * Obtain one page for the use of kcopyd.
213  */
214 static struct page_list *alloc_pl(gfp_t gfp)
215 {
216 	struct page_list *pl;
217 
218 	pl = kmalloc(sizeof(*pl), gfp);
219 	if (!pl)
220 		return NULL;
221 
222 	pl->page = alloc_page(gfp);
223 	if (!pl->page) {
224 		kfree(pl);
225 		return NULL;
226 	}
227 
228 	return pl;
229 }
230 
231 static void free_pl(struct page_list *pl)
232 {
233 	__free_page(pl->page);
234 	kfree(pl);
235 }
236 
237 /*
238  * Add the provided pages to a client's free page list, releasing
239  * back to the system any beyond the reserved_pages limit.
240  */
241 static void kcopyd_put_pages(struct dm_kcopyd_client *kc, struct page_list *pl)
242 {
243 	struct page_list *next;
244 
245 	do {
246 		next = pl->next;
247 
248 		if (kc->nr_free_pages >= kc->nr_reserved_pages)
249 			free_pl(pl);
250 		else {
251 			pl->next = kc->pages;
252 			kc->pages = pl;
253 			kc->nr_free_pages++;
254 		}
255 
256 		pl = next;
257 	} while (pl);
258 }
259 
260 static int kcopyd_get_pages(struct dm_kcopyd_client *kc,
261 			    unsigned int nr, struct page_list **pages)
262 {
263 	struct page_list *pl;
264 
265 	*pages = NULL;
266 
267 	do {
268 		pl = alloc_pl(__GFP_NOWARN | __GFP_NORETRY | __GFP_KSWAPD_RECLAIM);
269 		if (unlikely(!pl)) {
270 			/* Use reserved pages */
271 			pl = kc->pages;
272 			if (unlikely(!pl))
273 				goto out_of_memory;
274 			kc->pages = pl->next;
275 			kc->nr_free_pages--;
276 		}
277 		pl->next = *pages;
278 		*pages = pl;
279 	} while (--nr);
280 
281 	return 0;
282 
283 out_of_memory:
284 	if (*pages)
285 		kcopyd_put_pages(kc, *pages);
286 	return -ENOMEM;
287 }
288 
289 /*
290  * These three functions resize the page pool.
291  */
292 static void drop_pages(struct page_list *pl)
293 {
294 	struct page_list *next;
295 
296 	while (pl) {
297 		next = pl->next;
298 		free_pl(pl);
299 		pl = next;
300 	}
301 }
302 
303 /*
304  * Allocate and reserve nr_pages for the use of a specific client.
305  */
306 static int client_reserve_pages(struct dm_kcopyd_client *kc, unsigned nr_pages)
307 {
308 	unsigned i;
309 	struct page_list *pl = NULL, *next;
310 
311 	for (i = 0; i < nr_pages; i++) {
312 		next = alloc_pl(GFP_KERNEL);
313 		if (!next) {
314 			if (pl)
315 				drop_pages(pl);
316 			return -ENOMEM;
317 		}
318 		next->next = pl;
319 		pl = next;
320 	}
321 
322 	kc->nr_reserved_pages += nr_pages;
323 	kcopyd_put_pages(kc, pl);
324 
325 	return 0;
326 }
327 
328 static void client_free_pages(struct dm_kcopyd_client *kc)
329 {
330 	BUG_ON(kc->nr_free_pages != kc->nr_reserved_pages);
331 	drop_pages(kc->pages);
332 	kc->pages = NULL;
333 	kc->nr_free_pages = kc->nr_reserved_pages = 0;
334 }
335 
336 /*-----------------------------------------------------------------
337  * kcopyd_jobs need to be allocated by the *clients* of kcopyd,
338  * for this reason we use a mempool to prevent the client from
339  * ever having to do io (which could cause a deadlock).
340  *---------------------------------------------------------------*/
341 struct kcopyd_job {
342 	struct dm_kcopyd_client *kc;
343 	struct list_head list;
344 	unsigned long flags;
345 
346 	/*
347 	 * Error state of the job.
348 	 */
349 	int read_err;
350 	unsigned long write_err;
351 
352 	/*
353 	 * Either READ or WRITE
354 	 */
355 	int rw;
356 	struct dm_io_region source;
357 
358 	/*
359 	 * The destinations for the transfer.
360 	 */
361 	unsigned int num_dests;
362 	struct dm_io_region dests[DM_KCOPYD_MAX_REGIONS];
363 
364 	struct page_list *pages;
365 
366 	/*
367 	 * Set this to ensure you are notified when the job has
368 	 * completed.  'context' is for callback to use.
369 	 */
370 	dm_kcopyd_notify_fn fn;
371 	void *context;
372 
373 	/*
374 	 * These fields are only used if the job has been split
375 	 * into more manageable parts.
376 	 */
377 	struct mutex lock;
378 	atomic_t sub_jobs;
379 	sector_t progress;
380 	sector_t write_offset;
381 
382 	struct kcopyd_job *master_job;
383 };
384 
385 static struct kmem_cache *_job_cache;
386 
387 int __init dm_kcopyd_init(void)
388 {
389 	_job_cache = kmem_cache_create("kcopyd_job",
390 				sizeof(struct kcopyd_job) * (SPLIT_COUNT + 1),
391 				__alignof__(struct kcopyd_job), 0, NULL);
392 	if (!_job_cache)
393 		return -ENOMEM;
394 
395 	zero_page_list.next = &zero_page_list;
396 	zero_page_list.page = ZERO_PAGE(0);
397 
398 	return 0;
399 }
400 
401 void dm_kcopyd_exit(void)
402 {
403 	kmem_cache_destroy(_job_cache);
404 	_job_cache = NULL;
405 }
406 
407 /*
408  * Functions to push and pop a job onto the head of a given job
409  * list.
410  */
411 static struct kcopyd_job *pop_io_job(struct list_head *jobs,
412 				     struct dm_kcopyd_client *kc)
413 {
414 	struct kcopyd_job *job;
415 
416 	/*
417 	 * For I/O jobs, pop any read, any write without sequential write
418 	 * constraint and sequential writes that are at the right position.
419 	 */
420 	list_for_each_entry(job, jobs, list) {
421 		if (job->rw == READ || !test_bit(DM_KCOPYD_WRITE_SEQ, &job->flags)) {
422 			list_del(&job->list);
423 			return job;
424 		}
425 
426 		if (job->write_offset == job->master_job->write_offset) {
427 			job->master_job->write_offset += job->source.count;
428 			list_del(&job->list);
429 			return job;
430 		}
431 	}
432 
433 	return NULL;
434 }
435 
436 static struct kcopyd_job *pop(struct list_head *jobs,
437 			      struct dm_kcopyd_client *kc)
438 {
439 	struct kcopyd_job *job = NULL;
440 	unsigned long flags;
441 
442 	spin_lock_irqsave(&kc->job_lock, flags);
443 
444 	if (!list_empty(jobs)) {
445 		if (jobs == &kc->io_jobs)
446 			job = pop_io_job(jobs, kc);
447 		else {
448 			job = list_entry(jobs->next, struct kcopyd_job, list);
449 			list_del(&job->list);
450 		}
451 	}
452 	spin_unlock_irqrestore(&kc->job_lock, flags);
453 
454 	return job;
455 }
456 
457 static void push(struct list_head *jobs, struct kcopyd_job *job)
458 {
459 	unsigned long flags;
460 	struct dm_kcopyd_client *kc = job->kc;
461 
462 	spin_lock_irqsave(&kc->job_lock, flags);
463 	list_add_tail(&job->list, jobs);
464 	spin_unlock_irqrestore(&kc->job_lock, flags);
465 }
466 
467 
468 static void push_head(struct list_head *jobs, struct kcopyd_job *job)
469 {
470 	unsigned long flags;
471 	struct dm_kcopyd_client *kc = job->kc;
472 
473 	spin_lock_irqsave(&kc->job_lock, flags);
474 	list_add(&job->list, jobs);
475 	spin_unlock_irqrestore(&kc->job_lock, flags);
476 }
477 
478 /*
479  * These three functions process 1 item from the corresponding
480  * job list.
481  *
482  * They return:
483  * < 0: error
484  *   0: success
485  * > 0: can't process yet.
486  */
487 static int run_complete_job(struct kcopyd_job *job)
488 {
489 	void *context = job->context;
490 	int read_err = job->read_err;
491 	unsigned long write_err = job->write_err;
492 	dm_kcopyd_notify_fn fn = job->fn;
493 	struct dm_kcopyd_client *kc = job->kc;
494 
495 	if (job->pages && job->pages != &zero_page_list)
496 		kcopyd_put_pages(kc, job->pages);
497 	/*
498 	 * If this is the master job, the sub jobs have already
499 	 * completed so we can free everything.
500 	 */
501 	if (job->master_job == job) {
502 		mutex_destroy(&job->lock);
503 		mempool_free(job, &kc->job_pool);
504 	}
505 	fn(read_err, write_err, context);
506 
507 	if (atomic_dec_and_test(&kc->nr_jobs))
508 		wake_up(&kc->destroyq);
509 
510 	cond_resched();
511 
512 	return 0;
513 }
514 
515 static void complete_io(unsigned long error, void *context)
516 {
517 	struct kcopyd_job *job = (struct kcopyd_job *) context;
518 	struct dm_kcopyd_client *kc = job->kc;
519 
520 	io_job_finish(kc->throttle);
521 
522 	if (error) {
523 		if (op_is_write(job->rw))
524 			job->write_err |= error;
525 		else
526 			job->read_err = 1;
527 
528 		if (!test_bit(DM_KCOPYD_IGNORE_ERROR, &job->flags)) {
529 			push(&kc->complete_jobs, job);
530 			wake(kc);
531 			return;
532 		}
533 	}
534 
535 	if (op_is_write(job->rw))
536 		push(&kc->complete_jobs, job);
537 
538 	else {
539 		job->rw = WRITE;
540 		push(&kc->io_jobs, job);
541 	}
542 
543 	wake(kc);
544 }
545 
546 /*
547  * Request io on as many buffer heads as we can currently get for
548  * a particular job.
549  */
550 static int run_io_job(struct kcopyd_job *job)
551 {
552 	int r;
553 	struct dm_io_request io_req = {
554 		.bi_op = job->rw,
555 		.bi_op_flags = 0,
556 		.mem.type = DM_IO_PAGE_LIST,
557 		.mem.ptr.pl = job->pages,
558 		.mem.offset = 0,
559 		.notify.fn = complete_io,
560 		.notify.context = job,
561 		.client = job->kc->io_client,
562 	};
563 
564 	/*
565 	 * If we need to write sequentially and some reads or writes failed,
566 	 * no point in continuing.
567 	 */
568 	if (test_bit(DM_KCOPYD_WRITE_SEQ, &job->flags) &&
569 	    job->master_job->write_err) {
570 		job->write_err = job->master_job->write_err;
571 		return -EIO;
572 	}
573 
574 	io_job_start(job->kc->throttle);
575 
576 	if (job->rw == READ)
577 		r = dm_io(&io_req, 1, &job->source, NULL);
578 	else
579 		r = dm_io(&io_req, job->num_dests, job->dests, NULL);
580 
581 	return r;
582 }
583 
584 static int run_pages_job(struct kcopyd_job *job)
585 {
586 	int r;
587 	unsigned nr_pages = dm_div_up(job->dests[0].count, PAGE_SIZE >> 9);
588 
589 	r = kcopyd_get_pages(job->kc, nr_pages, &job->pages);
590 	if (!r) {
591 		/* this job is ready for io */
592 		push(&job->kc->io_jobs, job);
593 		return 0;
594 	}
595 
596 	if (r == -ENOMEM)
597 		/* can't complete now */
598 		return 1;
599 
600 	return r;
601 }
602 
603 /*
604  * Run through a list for as long as possible.  Returns the count
605  * of successful jobs.
606  */
607 static int process_jobs(struct list_head *jobs, struct dm_kcopyd_client *kc,
608 			int (*fn) (struct kcopyd_job *))
609 {
610 	struct kcopyd_job *job;
611 	int r, count = 0;
612 
613 	while ((job = pop(jobs, kc))) {
614 
615 		r = fn(job);
616 
617 		if (r < 0) {
618 			/* error this rogue job */
619 			if (op_is_write(job->rw))
620 				job->write_err = (unsigned long) -1L;
621 			else
622 				job->read_err = 1;
623 			push(&kc->complete_jobs, job);
624 			wake(kc);
625 			break;
626 		}
627 
628 		if (r > 0) {
629 			/*
630 			 * We couldn't service this job ATM, so
631 			 * push this job back onto the list.
632 			 */
633 			push_head(jobs, job);
634 			break;
635 		}
636 
637 		count++;
638 	}
639 
640 	return count;
641 }
642 
643 /*
644  * kcopyd does this every time it's woken up.
645  */
646 static void do_work(struct work_struct *work)
647 {
648 	struct dm_kcopyd_client *kc = container_of(work,
649 					struct dm_kcopyd_client, kcopyd_work);
650 	struct blk_plug plug;
651 	unsigned long flags;
652 
653 	/*
654 	 * The order that these are called is *very* important.
655 	 * complete jobs can free some pages for pages jobs.
656 	 * Pages jobs when successful will jump onto the io jobs
657 	 * list.  io jobs call wake when they complete and it all
658 	 * starts again.
659 	 */
660 	spin_lock_irqsave(&kc->job_lock, flags);
661 	list_splice_tail_init(&kc->callback_jobs, &kc->complete_jobs);
662 	spin_unlock_irqrestore(&kc->job_lock, flags);
663 
664 	blk_start_plug(&plug);
665 	process_jobs(&kc->complete_jobs, kc, run_complete_job);
666 	process_jobs(&kc->pages_jobs, kc, run_pages_job);
667 	process_jobs(&kc->io_jobs, kc, run_io_job);
668 	blk_finish_plug(&plug);
669 }
670 
671 /*
672  * If we are copying a small region we just dispatch a single job
673  * to do the copy, otherwise the io has to be split up into many
674  * jobs.
675  */
676 static void dispatch_job(struct kcopyd_job *job)
677 {
678 	struct dm_kcopyd_client *kc = job->kc;
679 	atomic_inc(&kc->nr_jobs);
680 	if (unlikely(!job->source.count))
681 		push(&kc->callback_jobs, job);
682 	else if (job->pages == &zero_page_list)
683 		push(&kc->io_jobs, job);
684 	else
685 		push(&kc->pages_jobs, job);
686 	wake(kc);
687 }
688 
689 static void segment_complete(int read_err, unsigned long write_err,
690 			     void *context)
691 {
692 	/* FIXME: tidy this function */
693 	sector_t progress = 0;
694 	sector_t count = 0;
695 	struct kcopyd_job *sub_job = (struct kcopyd_job *) context;
696 	struct kcopyd_job *job = sub_job->master_job;
697 	struct dm_kcopyd_client *kc = job->kc;
698 
699 	mutex_lock(&job->lock);
700 
701 	/* update the error */
702 	if (read_err)
703 		job->read_err = 1;
704 
705 	if (write_err)
706 		job->write_err |= write_err;
707 
708 	/*
709 	 * Only dispatch more work if there hasn't been an error.
710 	 */
711 	if ((!job->read_err && !job->write_err) ||
712 	    test_bit(DM_KCOPYD_IGNORE_ERROR, &job->flags)) {
713 		/* get the next chunk of work */
714 		progress = job->progress;
715 		count = job->source.count - progress;
716 		if (count) {
717 			if (count > kc->sub_job_size)
718 				count = kc->sub_job_size;
719 
720 			job->progress += count;
721 		}
722 	}
723 	mutex_unlock(&job->lock);
724 
725 	if (count) {
726 		int i;
727 
728 		*sub_job = *job;
729 		sub_job->write_offset = progress;
730 		sub_job->source.sector += progress;
731 		sub_job->source.count = count;
732 
733 		for (i = 0; i < job->num_dests; i++) {
734 			sub_job->dests[i].sector += progress;
735 			sub_job->dests[i].count = count;
736 		}
737 
738 		sub_job->fn = segment_complete;
739 		sub_job->context = sub_job;
740 		dispatch_job(sub_job);
741 
742 	} else if (atomic_dec_and_test(&job->sub_jobs)) {
743 
744 		/*
745 		 * Queue the completion callback to the kcopyd thread.
746 		 *
747 		 * Some callers assume that all the completions are called
748 		 * from a single thread and don't race with each other.
749 		 *
750 		 * We must not call the callback directly here because this
751 		 * code may not be executing in the thread.
752 		 */
753 		push(&kc->complete_jobs, job);
754 		wake(kc);
755 	}
756 }
757 
758 /*
759  * Create some sub jobs to share the work between them.
760  */
761 static void split_job(struct kcopyd_job *master_job)
762 {
763 	int i;
764 
765 	atomic_inc(&master_job->kc->nr_jobs);
766 
767 	atomic_set(&master_job->sub_jobs, SPLIT_COUNT);
768 	for (i = 0; i < SPLIT_COUNT; i++) {
769 		master_job[i + 1].master_job = master_job;
770 		segment_complete(0, 0u, &master_job[i + 1]);
771 	}
772 }
773 
774 void dm_kcopyd_copy(struct dm_kcopyd_client *kc, struct dm_io_region *from,
775 		    unsigned int num_dests, struct dm_io_region *dests,
776 		    unsigned int flags, dm_kcopyd_notify_fn fn, void *context)
777 {
778 	struct kcopyd_job *job;
779 	int i;
780 
781 	/*
782 	 * Allocate an array of jobs consisting of one master job
783 	 * followed by SPLIT_COUNT sub jobs.
784 	 */
785 	job = mempool_alloc(&kc->job_pool, GFP_NOIO);
786 	mutex_init(&job->lock);
787 
788 	/*
789 	 * set up for the read.
790 	 */
791 	job->kc = kc;
792 	job->flags = flags;
793 	job->read_err = 0;
794 	job->write_err = 0;
795 
796 	job->num_dests = num_dests;
797 	memcpy(&job->dests, dests, sizeof(*dests) * num_dests);
798 
799 	/*
800 	 * If one of the destination is a host-managed zoned block device,
801 	 * we need to write sequentially. If one of the destination is a
802 	 * host-aware device, then leave it to the caller to choose what to do.
803 	 */
804 	if (!test_bit(DM_KCOPYD_WRITE_SEQ, &job->flags)) {
805 		for (i = 0; i < job->num_dests; i++) {
806 			if (bdev_zoned_model(dests[i].bdev) == BLK_ZONED_HM) {
807 				set_bit(DM_KCOPYD_WRITE_SEQ, &job->flags);
808 				break;
809 			}
810 		}
811 	}
812 
813 	/*
814 	 * If we need to write sequentially, errors cannot be ignored.
815 	 */
816 	if (test_bit(DM_KCOPYD_WRITE_SEQ, &job->flags) &&
817 	    test_bit(DM_KCOPYD_IGNORE_ERROR, &job->flags))
818 		clear_bit(DM_KCOPYD_IGNORE_ERROR, &job->flags);
819 
820 	if (from) {
821 		job->source = *from;
822 		job->pages = NULL;
823 		job->rw = READ;
824 	} else {
825 		memset(&job->source, 0, sizeof job->source);
826 		job->source.count = job->dests[0].count;
827 		job->pages = &zero_page_list;
828 
829 		/*
830 		 * Use WRITE ZEROES to optimize zeroing if all dests support it.
831 		 */
832 		job->rw = REQ_OP_WRITE_ZEROES;
833 		for (i = 0; i < job->num_dests; i++)
834 			if (!bdev_write_zeroes_sectors(job->dests[i].bdev)) {
835 				job->rw = WRITE;
836 				break;
837 			}
838 	}
839 
840 	job->fn = fn;
841 	job->context = context;
842 	job->master_job = job;
843 	job->write_offset = 0;
844 
845 	if (job->source.count <= kc->sub_job_size)
846 		dispatch_job(job);
847 	else {
848 		job->progress = 0;
849 		split_job(job);
850 	}
851 }
852 EXPORT_SYMBOL(dm_kcopyd_copy);
853 
854 void dm_kcopyd_zero(struct dm_kcopyd_client *kc,
855 		    unsigned num_dests, struct dm_io_region *dests,
856 		    unsigned flags, dm_kcopyd_notify_fn fn, void *context)
857 {
858 	dm_kcopyd_copy(kc, NULL, num_dests, dests, flags, fn, context);
859 }
860 EXPORT_SYMBOL(dm_kcopyd_zero);
861 
862 void *dm_kcopyd_prepare_callback(struct dm_kcopyd_client *kc,
863 				 dm_kcopyd_notify_fn fn, void *context)
864 {
865 	struct kcopyd_job *job;
866 
867 	job = mempool_alloc(&kc->job_pool, GFP_NOIO);
868 
869 	memset(job, 0, sizeof(struct kcopyd_job));
870 	job->kc = kc;
871 	job->fn = fn;
872 	job->context = context;
873 	job->master_job = job;
874 
875 	atomic_inc(&kc->nr_jobs);
876 
877 	return job;
878 }
879 EXPORT_SYMBOL(dm_kcopyd_prepare_callback);
880 
881 void dm_kcopyd_do_callback(void *j, int read_err, unsigned long write_err)
882 {
883 	struct kcopyd_job *job = j;
884 	struct dm_kcopyd_client *kc = job->kc;
885 
886 	job->read_err = read_err;
887 	job->write_err = write_err;
888 
889 	push(&kc->callback_jobs, job);
890 	wake(kc);
891 }
892 EXPORT_SYMBOL(dm_kcopyd_do_callback);
893 
894 /*
895  * Cancels a kcopyd job, eg. someone might be deactivating a
896  * mirror.
897  */
898 #if 0
899 int kcopyd_cancel(struct kcopyd_job *job, int block)
900 {
901 	/* FIXME: finish */
902 	return -1;
903 }
904 #endif  /*  0  */
905 
906 /*-----------------------------------------------------------------
907  * Client setup
908  *---------------------------------------------------------------*/
909 struct dm_kcopyd_client *dm_kcopyd_client_create(struct dm_kcopyd_throttle *throttle)
910 {
911 	int r;
912 	unsigned reserve_pages;
913 	struct dm_kcopyd_client *kc;
914 
915 	kc = kzalloc(sizeof(*kc), GFP_KERNEL);
916 	if (!kc)
917 		return ERR_PTR(-ENOMEM);
918 
919 	spin_lock_init(&kc->job_lock);
920 	INIT_LIST_HEAD(&kc->callback_jobs);
921 	INIT_LIST_HEAD(&kc->complete_jobs);
922 	INIT_LIST_HEAD(&kc->io_jobs);
923 	INIT_LIST_HEAD(&kc->pages_jobs);
924 	kc->throttle = throttle;
925 
926 	r = mempool_init_slab_pool(&kc->job_pool, MIN_JOBS, _job_cache);
927 	if (r)
928 		goto bad_slab;
929 
930 	INIT_WORK(&kc->kcopyd_work, do_work);
931 	kc->kcopyd_wq = alloc_workqueue("kcopyd", WQ_MEM_RECLAIM, 0);
932 	if (!kc->kcopyd_wq) {
933 		r = -ENOMEM;
934 		goto bad_workqueue;
935 	}
936 
937 	kc->sub_job_size = dm_get_kcopyd_subjob_size();
938 	reserve_pages = DIV_ROUND_UP(kc->sub_job_size << SECTOR_SHIFT, PAGE_SIZE);
939 
940 	kc->pages = NULL;
941 	kc->nr_reserved_pages = kc->nr_free_pages = 0;
942 	r = client_reserve_pages(kc, reserve_pages);
943 	if (r)
944 		goto bad_client_pages;
945 
946 	kc->io_client = dm_io_client_create();
947 	if (IS_ERR(kc->io_client)) {
948 		r = PTR_ERR(kc->io_client);
949 		goto bad_io_client;
950 	}
951 
952 	init_waitqueue_head(&kc->destroyq);
953 	atomic_set(&kc->nr_jobs, 0);
954 
955 	return kc;
956 
957 bad_io_client:
958 	client_free_pages(kc);
959 bad_client_pages:
960 	destroy_workqueue(kc->kcopyd_wq);
961 bad_workqueue:
962 	mempool_exit(&kc->job_pool);
963 bad_slab:
964 	kfree(kc);
965 
966 	return ERR_PTR(r);
967 }
968 EXPORT_SYMBOL(dm_kcopyd_client_create);
969 
970 void dm_kcopyd_client_destroy(struct dm_kcopyd_client *kc)
971 {
972 	/* Wait for completion of all jobs submitted by this client. */
973 	wait_event(kc->destroyq, !atomic_read(&kc->nr_jobs));
974 
975 	BUG_ON(!list_empty(&kc->callback_jobs));
976 	BUG_ON(!list_empty(&kc->complete_jobs));
977 	BUG_ON(!list_empty(&kc->io_jobs));
978 	BUG_ON(!list_empty(&kc->pages_jobs));
979 	destroy_workqueue(kc->kcopyd_wq);
980 	dm_io_client_destroy(kc->io_client);
981 	client_free_pages(kc);
982 	mempool_exit(&kc->job_pool);
983 	kfree(kc);
984 }
985 EXPORT_SYMBOL(dm_kcopyd_client_destroy);
986