1/*
2 * Copyright (C) 2011 Red Hat, Inc.
3 *
4 * This file is released under the GPL.
5 */
6
7#include "dm-space-map.h"
8#include "dm-space-map-common.h"
9#include "dm-space-map-metadata.h"
10
11#include <linux/list.h>
12#include <linux/slab.h>
13#include <linux/device-mapper.h>
14
15#define DM_MSG_PREFIX "space map metadata"
16
17/*----------------------------------------------------------------*/
18
19/*
20 * An edge triggered threshold.
21 */
22struct threshold {
23	bool threshold_set;
24	bool value_set;
25	dm_block_t threshold;
26	dm_block_t current_value;
27	dm_sm_threshold_fn fn;
28	void *context;
29};
30
31static void threshold_init(struct threshold *t)
32{
33	t->threshold_set = false;
34	t->value_set = false;
35}
36
37static void set_threshold(struct threshold *t, dm_block_t value,
38			  dm_sm_threshold_fn fn, void *context)
39{
40	t->threshold_set = true;
41	t->threshold = value;
42	t->fn = fn;
43	t->context = context;
44}
45
46static bool below_threshold(struct threshold *t, dm_block_t value)
47{
48	return t->threshold_set && value <= t->threshold;
49}
50
51static bool threshold_already_triggered(struct threshold *t)
52{
53	return t->value_set && below_threshold(t, t->current_value);
54}
55
56static void check_threshold(struct threshold *t, dm_block_t value)
57{
58	if (below_threshold(t, value) &&
59	    !threshold_already_triggered(t))
60		t->fn(t->context);
61
62	t->value_set = true;
63	t->current_value = value;
64}
65
66/*----------------------------------------------------------------*/
67
68/*
69 * Space map interface.
70 *
71 * The low level disk format is written using the standard btree and
72 * transaction manager.  This means that performing disk operations may
73 * cause us to recurse into the space map in order to allocate new blocks.
74 * For this reason we have a pool of pre-allocated blocks large enough to
75 * service any metadata_ll_disk operation.
76 */
77
78/*
79 * FIXME: we should calculate this based on the size of the device.
80 * Only the metadata space map needs this functionality.
81 */
82#define MAX_RECURSIVE_ALLOCATIONS 1024
83
84enum block_op_type {
85	BOP_INC,
86	BOP_DEC
87};
88
89struct block_op {
90	enum block_op_type type;
91	dm_block_t block;
92};
93
94struct bop_ring_buffer {
95	unsigned begin;
96	unsigned end;
97	struct block_op bops[MAX_RECURSIVE_ALLOCATIONS + 1];
98};
99
100static void brb_init(struct bop_ring_buffer *brb)
101{
102	brb->begin = 0;
103	brb->end = 0;
104}
105
106static bool brb_empty(struct bop_ring_buffer *brb)
107{
108	return brb->begin == brb->end;
109}
110
111static unsigned brb_next(struct bop_ring_buffer *brb, unsigned old)
112{
113	unsigned r = old + 1;
114	return (r >= (sizeof(brb->bops) / sizeof(*brb->bops))) ? 0 : r;
115}
116
117static int brb_push(struct bop_ring_buffer *brb,
118		    enum block_op_type type, dm_block_t b)
119{
120	struct block_op *bop;
121	unsigned next = brb_next(brb, brb->end);
122
123	/*
124	 * We don't allow the last bop to be filled, this way we can
125	 * differentiate between full and empty.
126	 */
127	if (next == brb->begin)
128		return -ENOMEM;
129
130	bop = brb->bops + brb->end;
131	bop->type = type;
132	bop->block = b;
133
134	brb->end = next;
135
136	return 0;
137}
138
139static int brb_pop(struct bop_ring_buffer *brb, struct block_op *result)
140{
141	struct block_op *bop;
142
143	if (brb_empty(brb))
144		return -ENODATA;
145
146	bop = brb->bops + brb->begin;
147	result->type = bop->type;
148	result->block = bop->block;
149
150	brb->begin = brb_next(brb, brb->begin);
151
152	return 0;
153}
154
155/*----------------------------------------------------------------*/
156
157struct sm_metadata {
158	struct dm_space_map sm;
159
160	struct ll_disk ll;
161	struct ll_disk old_ll;
162
163	dm_block_t begin;
164
165	unsigned recursion_count;
166	unsigned allocated_this_transaction;
167	struct bop_ring_buffer uncommitted;
168
169	struct threshold threshold;
170};
171
172static int add_bop(struct sm_metadata *smm, enum block_op_type type, dm_block_t b)
173{
174	int r = brb_push(&smm->uncommitted, type, b);
175
176	if (r) {
177		DMERR("too many recursive allocations");
178		return -ENOMEM;
179	}
180
181	return 0;
182}
183
184static int commit_bop(struct sm_metadata *smm, struct block_op *op)
185{
186	int r = 0;
187	enum allocation_event ev;
188
189	switch (op->type) {
190	case BOP_INC:
191		r = sm_ll_inc(&smm->ll, op->block, &ev);
192		break;
193
194	case BOP_DEC:
195		r = sm_ll_dec(&smm->ll, op->block, &ev);
196		break;
197	}
198
199	return r;
200}
201
202static void in(struct sm_metadata *smm)
203{
204	smm->recursion_count++;
205}
206
207static int out(struct sm_metadata *smm)
208{
209	int r = 0;
210
211	/*
212	 * If we're not recursing then very bad things are happening.
213	 */
214	if (!smm->recursion_count) {
215		DMERR("lost track of recursion depth");
216		return -ENOMEM;
217	}
218
219	if (smm->recursion_count == 1) {
220		while (!brb_empty(&smm->uncommitted)) {
221			struct block_op bop;
222
223			r = brb_pop(&smm->uncommitted, &bop);
224			if (r) {
225				DMERR("bug in bop ring buffer");
226				break;
227			}
228
229			r = commit_bop(smm, &bop);
230			if (r)
231				break;
232		}
233	}
234
235	smm->recursion_count--;
236
237	return r;
238}
239
240/*
241 * When using the out() function above, we often want to combine an error
242 * code for the operation run in the recursive context with that from
243 * out().
244 */
245static int combine_errors(int r1, int r2)
246{
247	return r1 ? r1 : r2;
248}
249
250static int recursing(struct sm_metadata *smm)
251{
252	return smm->recursion_count;
253}
254
255static void sm_metadata_destroy(struct dm_space_map *sm)
256{
257	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
258
259	kfree(smm);
260}
261
262static int sm_metadata_get_nr_blocks(struct dm_space_map *sm, dm_block_t *count)
263{
264	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
265
266	*count = smm->ll.nr_blocks;
267
268	return 0;
269}
270
271static int sm_metadata_get_nr_free(struct dm_space_map *sm, dm_block_t *count)
272{
273	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
274
275	*count = smm->old_ll.nr_blocks - smm->old_ll.nr_allocated -
276		 smm->allocated_this_transaction;
277
278	return 0;
279}
280
281static int sm_metadata_get_count(struct dm_space_map *sm, dm_block_t b,
282				 uint32_t *result)
283{
284	int r;
285	unsigned i;
286	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
287	unsigned adjustment = 0;
288
289	/*
290	 * We may have some uncommitted adjustments to add.  This list
291	 * should always be really short.
292	 */
293	for (i = smm->uncommitted.begin;
294	     i != smm->uncommitted.end;
295	     i = brb_next(&smm->uncommitted, i)) {
296		struct block_op *op = smm->uncommitted.bops + i;
297
298		if (op->block != b)
299			continue;
300
301		switch (op->type) {
302		case BOP_INC:
303			adjustment++;
304			break;
305
306		case BOP_DEC:
307			adjustment--;
308			break;
309		}
310	}
311
312	r = sm_ll_lookup(&smm->ll, b, result);
313	if (r)
314		return r;
315
316	*result += adjustment;
317
318	return 0;
319}
320
321static int sm_metadata_count_is_more_than_one(struct dm_space_map *sm,
322					      dm_block_t b, int *result)
323{
324	int r, adjustment = 0;
325	unsigned i;
326	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
327	uint32_t rc;
328
329	/*
330	 * We may have some uncommitted adjustments to add.  This list
331	 * should always be really short.
332	 */
333	for (i = smm->uncommitted.begin;
334	     i != smm->uncommitted.end;
335	     i = brb_next(&smm->uncommitted, i)) {
336
337		struct block_op *op = smm->uncommitted.bops + i;
338
339		if (op->block != b)
340			continue;
341
342		switch (op->type) {
343		case BOP_INC:
344			adjustment++;
345			break;
346
347		case BOP_DEC:
348			adjustment--;
349			break;
350		}
351	}
352
353	if (adjustment > 1) {
354		*result = 1;
355		return 0;
356	}
357
358	r = sm_ll_lookup_bitmap(&smm->ll, b, &rc);
359	if (r)
360		return r;
361
362	if (rc == 3)
363		/*
364		 * We err on the side of caution, and always return true.
365		 */
366		*result = 1;
367	else
368		*result = rc + adjustment > 1;
369
370	return 0;
371}
372
373static int sm_metadata_set_count(struct dm_space_map *sm, dm_block_t b,
374				 uint32_t count)
375{
376	int r, r2;
377	enum allocation_event ev;
378	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
379
380	if (smm->recursion_count) {
381		DMERR("cannot recurse set_count()");
382		return -EINVAL;
383	}
384
385	in(smm);
386	r = sm_ll_insert(&smm->ll, b, count, &ev);
387	r2 = out(smm);
388
389	return combine_errors(r, r2);
390}
391
392static int sm_metadata_inc_block(struct dm_space_map *sm, dm_block_t b)
393{
394	int r, r2 = 0;
395	enum allocation_event ev;
396	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
397
398	if (recursing(smm))
399		r = add_bop(smm, BOP_INC, b);
400	else {
401		in(smm);
402		r = sm_ll_inc(&smm->ll, b, &ev);
403		r2 = out(smm);
404	}
405
406	return combine_errors(r, r2);
407}
408
409static int sm_metadata_dec_block(struct dm_space_map *sm, dm_block_t b)
410{
411	int r, r2 = 0;
412	enum allocation_event ev;
413	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
414
415	if (recursing(smm))
416		r = add_bop(smm, BOP_DEC, b);
417	else {
418		in(smm);
419		r = sm_ll_dec(&smm->ll, b, &ev);
420		r2 = out(smm);
421	}
422
423	return combine_errors(r, r2);
424}
425
426static int sm_metadata_new_block_(struct dm_space_map *sm, dm_block_t *b)
427{
428	int r, r2 = 0;
429	enum allocation_event ev;
430	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
431
432	r = sm_ll_find_free_block(&smm->old_ll, smm->begin, smm->old_ll.nr_blocks, b);
433	if (r)
434		return r;
435
436	smm->begin = *b + 1;
437
438	if (recursing(smm))
439		r = add_bop(smm, BOP_INC, *b);
440	else {
441		in(smm);
442		r = sm_ll_inc(&smm->ll, *b, &ev);
443		r2 = out(smm);
444	}
445
446	if (!r)
447		smm->allocated_this_transaction++;
448
449	return combine_errors(r, r2);
450}
451
452static int sm_metadata_new_block(struct dm_space_map *sm, dm_block_t *b)
453{
454	dm_block_t count;
455	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
456
457	int r = sm_metadata_new_block_(sm, b);
458	if (r) {
459		DMERR_LIMIT("unable to allocate new metadata block");
460		return r;
461	}
462
463	r = sm_metadata_get_nr_free(sm, &count);
464	if (r) {
465		DMERR_LIMIT("couldn't get free block count");
466		return r;
467	}
468
469	check_threshold(&smm->threshold, count);
470
471	return r;
472}
473
474static int sm_metadata_commit(struct dm_space_map *sm)
475{
476	int r;
477	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
478
479	r = sm_ll_commit(&smm->ll);
480	if (r)
481		return r;
482
483	memcpy(&smm->old_ll, &smm->ll, sizeof(smm->old_ll));
484	smm->begin = 0;
485	smm->allocated_this_transaction = 0;
486
487	return 0;
488}
489
490static int sm_metadata_register_threshold_callback(struct dm_space_map *sm,
491						   dm_block_t threshold,
492						   dm_sm_threshold_fn fn,
493						   void *context)
494{
495	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
496
497	set_threshold(&smm->threshold, threshold, fn, context);
498
499	return 0;
500}
501
502static int sm_metadata_root_size(struct dm_space_map *sm, size_t *result)
503{
504	*result = sizeof(struct disk_sm_root);
505
506	return 0;
507}
508
509static int sm_metadata_copy_root(struct dm_space_map *sm, void *where_le, size_t max)
510{
511	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
512	struct disk_sm_root root_le;
513
514	root_le.nr_blocks = cpu_to_le64(smm->ll.nr_blocks);
515	root_le.nr_allocated = cpu_to_le64(smm->ll.nr_allocated);
516	root_le.bitmap_root = cpu_to_le64(smm->ll.bitmap_root);
517	root_le.ref_count_root = cpu_to_le64(smm->ll.ref_count_root);
518
519	if (max < sizeof(root_le))
520		return -ENOSPC;
521
522	memcpy(where_le, &root_le, sizeof(root_le));
523
524	return 0;
525}
526
527static int sm_metadata_extend(struct dm_space_map *sm, dm_block_t extra_blocks);
528
529static struct dm_space_map ops = {
530	.destroy = sm_metadata_destroy,
531	.extend = sm_metadata_extend,
532	.get_nr_blocks = sm_metadata_get_nr_blocks,
533	.get_nr_free = sm_metadata_get_nr_free,
534	.get_count = sm_metadata_get_count,
535	.count_is_more_than_one = sm_metadata_count_is_more_than_one,
536	.set_count = sm_metadata_set_count,
537	.inc_block = sm_metadata_inc_block,
538	.dec_block = sm_metadata_dec_block,
539	.new_block = sm_metadata_new_block,
540	.commit = sm_metadata_commit,
541	.root_size = sm_metadata_root_size,
542	.copy_root = sm_metadata_copy_root,
543	.register_threshold_callback = sm_metadata_register_threshold_callback
544};
545
546/*----------------------------------------------------------------*/
547
548/*
549 * When a new space map is created that manages its own space.  We use
550 * this tiny bootstrap allocator.
551 */
552static void sm_bootstrap_destroy(struct dm_space_map *sm)
553{
554}
555
556static int sm_bootstrap_extend(struct dm_space_map *sm, dm_block_t extra_blocks)
557{
558	DMERR("bootstrap doesn't support extend");
559
560	return -EINVAL;
561}
562
563static int sm_bootstrap_get_nr_blocks(struct dm_space_map *sm, dm_block_t *count)
564{
565	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
566
567	return smm->ll.nr_blocks;
568}
569
570static int sm_bootstrap_get_nr_free(struct dm_space_map *sm, dm_block_t *count)
571{
572	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
573
574	*count = smm->ll.nr_blocks - smm->begin;
575
576	return 0;
577}
578
579static int sm_bootstrap_get_count(struct dm_space_map *sm, dm_block_t b,
580				  uint32_t *result)
581{
582	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
583
584	return b < smm->begin ? 1 : 0;
585}
586
587static int sm_bootstrap_count_is_more_than_one(struct dm_space_map *sm,
588					       dm_block_t b, int *result)
589{
590	*result = 0;
591
592	return 0;
593}
594
595static int sm_bootstrap_set_count(struct dm_space_map *sm, dm_block_t b,
596				  uint32_t count)
597{
598	DMERR("bootstrap doesn't support set_count");
599
600	return -EINVAL;
601}
602
603static int sm_bootstrap_new_block(struct dm_space_map *sm, dm_block_t *b)
604{
605	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
606
607	/*
608	 * We know the entire device is unused.
609	 */
610	if (smm->begin == smm->ll.nr_blocks)
611		return -ENOSPC;
612
613	*b = smm->begin++;
614
615	return 0;
616}
617
618static int sm_bootstrap_inc_block(struct dm_space_map *sm, dm_block_t b)
619{
620	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
621
622	return add_bop(smm, BOP_INC, b);
623}
624
625static int sm_bootstrap_dec_block(struct dm_space_map *sm, dm_block_t b)
626{
627	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
628
629	return add_bop(smm, BOP_DEC, b);
630}
631
632static int sm_bootstrap_commit(struct dm_space_map *sm)
633{
634	return 0;
635}
636
637static int sm_bootstrap_root_size(struct dm_space_map *sm, size_t *result)
638{
639	DMERR("bootstrap doesn't support root_size");
640
641	return -EINVAL;
642}
643
644static int sm_bootstrap_copy_root(struct dm_space_map *sm, void *where,
645				  size_t max)
646{
647	DMERR("bootstrap doesn't support copy_root");
648
649	return -EINVAL;
650}
651
652static struct dm_space_map bootstrap_ops = {
653	.destroy = sm_bootstrap_destroy,
654	.extend = sm_bootstrap_extend,
655	.get_nr_blocks = sm_bootstrap_get_nr_blocks,
656	.get_nr_free = sm_bootstrap_get_nr_free,
657	.get_count = sm_bootstrap_get_count,
658	.count_is_more_than_one = sm_bootstrap_count_is_more_than_one,
659	.set_count = sm_bootstrap_set_count,
660	.inc_block = sm_bootstrap_inc_block,
661	.dec_block = sm_bootstrap_dec_block,
662	.new_block = sm_bootstrap_new_block,
663	.commit = sm_bootstrap_commit,
664	.root_size = sm_bootstrap_root_size,
665	.copy_root = sm_bootstrap_copy_root,
666	.register_threshold_callback = NULL
667};
668
669/*----------------------------------------------------------------*/
670
671static int sm_metadata_extend(struct dm_space_map *sm, dm_block_t extra_blocks)
672{
673	int r, i;
674	enum allocation_event ev;
675	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
676	dm_block_t old_len = smm->ll.nr_blocks;
677
678	/*
679	 * Flick into a mode where all blocks get allocated in the new area.
680	 */
681	smm->begin = old_len;
682	memcpy(sm, &bootstrap_ops, sizeof(*sm));
683
684	/*
685	 * Extend.
686	 */
687	r = sm_ll_extend(&smm->ll, extra_blocks);
688	if (r)
689		goto out;
690
691	/*
692	 * We repeatedly increment then commit until the commit doesn't
693	 * allocate any new blocks.
694	 */
695	do {
696		for (i = old_len; !r && i < smm->begin; i++) {
697			r = sm_ll_inc(&smm->ll, i, &ev);
698			if (r)
699				goto out;
700		}
701		old_len = smm->begin;
702
703		r = sm_ll_commit(&smm->ll);
704		if (r)
705			goto out;
706
707	} while (old_len != smm->begin);
708
709out:
710	/*
711	 * Switch back to normal behaviour.
712	 */
713	memcpy(sm, &ops, sizeof(*sm));
714	return r;
715}
716
717/*----------------------------------------------------------------*/
718
719struct dm_space_map *dm_sm_metadata_init(void)
720{
721	struct sm_metadata *smm;
722
723	smm = kmalloc(sizeof(*smm), GFP_KERNEL);
724	if (!smm)
725		return ERR_PTR(-ENOMEM);
726
727	memcpy(&smm->sm, &ops, sizeof(smm->sm));
728
729	return &smm->sm;
730}
731
732int dm_sm_metadata_create(struct dm_space_map *sm,
733			  struct dm_transaction_manager *tm,
734			  dm_block_t nr_blocks,
735			  dm_block_t superblock)
736{
737	int r;
738	dm_block_t i;
739	enum allocation_event ev;
740	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
741
742	smm->begin = superblock + 1;
743	smm->recursion_count = 0;
744	smm->allocated_this_transaction = 0;
745	brb_init(&smm->uncommitted);
746	threshold_init(&smm->threshold);
747
748	memcpy(&smm->sm, &bootstrap_ops, sizeof(smm->sm));
749
750	r = sm_ll_new_metadata(&smm->ll, tm);
751	if (r)
752		return r;
753
754	if (nr_blocks > DM_SM_METADATA_MAX_BLOCKS)
755		nr_blocks = DM_SM_METADATA_MAX_BLOCKS;
756	r = sm_ll_extend(&smm->ll, nr_blocks);
757	if (r)
758		return r;
759
760	memcpy(&smm->sm, &ops, sizeof(smm->sm));
761
762	/*
763	 * Now we need to update the newly created data structures with the
764	 * allocated blocks that they were built from.
765	 */
766	for (i = superblock; !r && i < smm->begin; i++)
767		r = sm_ll_inc(&smm->ll, i, &ev);
768
769	if (r)
770		return r;
771
772	return sm_metadata_commit(sm);
773}
774
775int dm_sm_metadata_open(struct dm_space_map *sm,
776			struct dm_transaction_manager *tm,
777			void *root_le, size_t len)
778{
779	int r;
780	struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
781
782	r = sm_ll_open_metadata(&smm->ll, tm, root_le, len);
783	if (r)
784		return r;
785
786	smm->begin = 0;
787	smm->recursion_count = 0;
788	smm->allocated_this_transaction = 0;
789	brb_init(&smm->uncommitted);
790	threshold_init(&smm->threshold);
791
792	memcpy(&smm->old_ll, &smm->ll, sizeof(smm->old_ll));
793	return 0;
794}
795