Kernel bump from 4.1.3-rt to 4.1.7-rt.
[kvmfornfv.git] / kernel / drivers / md / persistent-data / dm-space-map-metadata.c
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  */
22 struct 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
31 static void threshold_init(struct threshold *t)
32 {
33         t->threshold_set = false;
34         t->value_set = false;
35 }
36
37 static 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
46 static bool below_threshold(struct threshold *t, dm_block_t value)
47 {
48         return t->threshold_set && value <= t->threshold;
49 }
50
51 static bool threshold_already_triggered(struct threshold *t)
52 {
53         return t->value_set && below_threshold(t, t->current_value);
54 }
55
56 static 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
84 enum block_op_type {
85         BOP_INC,
86         BOP_DEC
87 };
88
89 struct block_op {
90         enum block_op_type type;
91         dm_block_t block;
92 };
93
94 struct bop_ring_buffer {
95         unsigned begin;
96         unsigned end;
97         struct block_op bops[MAX_RECURSIVE_ALLOCATIONS + 1];
98 };
99
100 static void brb_init(struct bop_ring_buffer *brb)
101 {
102         brb->begin = 0;
103         brb->end = 0;
104 }
105
106 static bool brb_empty(struct bop_ring_buffer *brb)
107 {
108         return brb->begin == brb->end;
109 }
110
111 static 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
117 static 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
139 static 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
157 struct 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
172 static 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
184 static 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
202 static void in(struct sm_metadata *smm)
203 {
204         smm->recursion_count++;
205 }
206
207 static int apply_bops(struct sm_metadata *smm)
208 {
209         int r = 0;
210
211         while (!brb_empty(&smm->uncommitted)) {
212                 struct block_op bop;
213
214                 r = brb_pop(&smm->uncommitted, &bop);
215                 if (r) {
216                         DMERR("bug in bop ring buffer");
217                         break;
218                 }
219
220                 r = commit_bop(smm, &bop);
221                 if (r)
222                         break;
223         }
224
225         return r;
226 }
227
228 static int out(struct sm_metadata *smm)
229 {
230         int r = 0;
231
232         /*
233          * If we're not recursing then very bad things are happening.
234          */
235         if (!smm->recursion_count) {
236                 DMERR("lost track of recursion depth");
237                 return -ENOMEM;
238         }
239
240         if (smm->recursion_count == 1)
241                 apply_bops(smm);
242
243         smm->recursion_count--;
244
245         return r;
246 }
247
248 /*
249  * When using the out() function above, we often want to combine an error
250  * code for the operation run in the recursive context with that from
251  * out().
252  */
253 static int combine_errors(int r1, int r2)
254 {
255         return r1 ? r1 : r2;
256 }
257
258 static int recursing(struct sm_metadata *smm)
259 {
260         return smm->recursion_count;
261 }
262
263 static void sm_metadata_destroy(struct dm_space_map *sm)
264 {
265         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
266
267         kfree(smm);
268 }
269
270 static int sm_metadata_get_nr_blocks(struct dm_space_map *sm, dm_block_t *count)
271 {
272         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
273
274         *count = smm->ll.nr_blocks;
275
276         return 0;
277 }
278
279 static int sm_metadata_get_nr_free(struct dm_space_map *sm, dm_block_t *count)
280 {
281         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
282
283         *count = smm->old_ll.nr_blocks - smm->old_ll.nr_allocated -
284                  smm->allocated_this_transaction;
285
286         return 0;
287 }
288
289 static int sm_metadata_get_count(struct dm_space_map *sm, dm_block_t b,
290                                  uint32_t *result)
291 {
292         int r;
293         unsigned i;
294         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
295         unsigned adjustment = 0;
296
297         /*
298          * We may have some uncommitted adjustments to add.  This list
299          * should always be really short.
300          */
301         for (i = smm->uncommitted.begin;
302              i != smm->uncommitted.end;
303              i = brb_next(&smm->uncommitted, i)) {
304                 struct block_op *op = smm->uncommitted.bops + i;
305
306                 if (op->block != b)
307                         continue;
308
309                 switch (op->type) {
310                 case BOP_INC:
311                         adjustment++;
312                         break;
313
314                 case BOP_DEC:
315                         adjustment--;
316                         break;
317                 }
318         }
319
320         r = sm_ll_lookup(&smm->ll, b, result);
321         if (r)
322                 return r;
323
324         *result += adjustment;
325
326         return 0;
327 }
328
329 static int sm_metadata_count_is_more_than_one(struct dm_space_map *sm,
330                                               dm_block_t b, int *result)
331 {
332         int r, adjustment = 0;
333         unsigned i;
334         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
335         uint32_t rc;
336
337         /*
338          * We may have some uncommitted adjustments to add.  This list
339          * should always be really short.
340          */
341         for (i = smm->uncommitted.begin;
342              i != smm->uncommitted.end;
343              i = brb_next(&smm->uncommitted, i)) {
344
345                 struct block_op *op = smm->uncommitted.bops + i;
346
347                 if (op->block != b)
348                         continue;
349
350                 switch (op->type) {
351                 case BOP_INC:
352                         adjustment++;
353                         break;
354
355                 case BOP_DEC:
356                         adjustment--;
357                         break;
358                 }
359         }
360
361         if (adjustment > 1) {
362                 *result = 1;
363                 return 0;
364         }
365
366         r = sm_ll_lookup_bitmap(&smm->ll, b, &rc);
367         if (r)
368                 return r;
369
370         if (rc == 3)
371                 /*
372                  * We err on the side of caution, and always return true.
373                  */
374                 *result = 1;
375         else
376                 *result = rc + adjustment > 1;
377
378         return 0;
379 }
380
381 static int sm_metadata_set_count(struct dm_space_map *sm, dm_block_t b,
382                                  uint32_t count)
383 {
384         int r, r2;
385         enum allocation_event ev;
386         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
387
388         if (smm->recursion_count) {
389                 DMERR("cannot recurse set_count()");
390                 return -EINVAL;
391         }
392
393         in(smm);
394         r = sm_ll_insert(&smm->ll, b, count, &ev);
395         r2 = out(smm);
396
397         return combine_errors(r, r2);
398 }
399
400 static int sm_metadata_inc_block(struct dm_space_map *sm, dm_block_t b)
401 {
402         int r, r2 = 0;
403         enum allocation_event ev;
404         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
405
406         if (recursing(smm))
407                 r = add_bop(smm, BOP_INC, b);
408         else {
409                 in(smm);
410                 r = sm_ll_inc(&smm->ll, b, &ev);
411                 r2 = out(smm);
412         }
413
414         return combine_errors(r, r2);
415 }
416
417 static int sm_metadata_dec_block(struct dm_space_map *sm, dm_block_t b)
418 {
419         int r, r2 = 0;
420         enum allocation_event ev;
421         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
422
423         if (recursing(smm))
424                 r = add_bop(smm, BOP_DEC, b);
425         else {
426                 in(smm);
427                 r = sm_ll_dec(&smm->ll, b, &ev);
428                 r2 = out(smm);
429         }
430
431         return combine_errors(r, r2);
432 }
433
434 static int sm_metadata_new_block_(struct dm_space_map *sm, dm_block_t *b)
435 {
436         int r, r2 = 0;
437         enum allocation_event ev;
438         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
439
440         r = sm_ll_find_free_block(&smm->old_ll, smm->begin, smm->old_ll.nr_blocks, b);
441         if (r)
442                 return r;
443
444         smm->begin = *b + 1;
445
446         if (recursing(smm))
447                 r = add_bop(smm, BOP_INC, *b);
448         else {
449                 in(smm);
450                 r = sm_ll_inc(&smm->ll, *b, &ev);
451                 r2 = out(smm);
452         }
453
454         if (!r)
455                 smm->allocated_this_transaction++;
456
457         return combine_errors(r, r2);
458 }
459
460 static int sm_metadata_new_block(struct dm_space_map *sm, dm_block_t *b)
461 {
462         dm_block_t count;
463         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
464
465         int r = sm_metadata_new_block_(sm, b);
466         if (r) {
467                 DMERR_LIMIT("unable to allocate new metadata block");
468                 return r;
469         }
470
471         r = sm_metadata_get_nr_free(sm, &count);
472         if (r) {
473                 DMERR_LIMIT("couldn't get free block count");
474                 return r;
475         }
476
477         check_threshold(&smm->threshold, count);
478
479         return r;
480 }
481
482 static int sm_metadata_commit(struct dm_space_map *sm)
483 {
484         int r;
485         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
486
487         r = sm_ll_commit(&smm->ll);
488         if (r)
489                 return r;
490
491         memcpy(&smm->old_ll, &smm->ll, sizeof(smm->old_ll));
492         smm->begin = 0;
493         smm->allocated_this_transaction = 0;
494
495         return 0;
496 }
497
498 static int sm_metadata_register_threshold_callback(struct dm_space_map *sm,
499                                                    dm_block_t threshold,
500                                                    dm_sm_threshold_fn fn,
501                                                    void *context)
502 {
503         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
504
505         set_threshold(&smm->threshold, threshold, fn, context);
506
507         return 0;
508 }
509
510 static int sm_metadata_root_size(struct dm_space_map *sm, size_t *result)
511 {
512         *result = sizeof(struct disk_sm_root);
513
514         return 0;
515 }
516
517 static int sm_metadata_copy_root(struct dm_space_map *sm, void *where_le, size_t max)
518 {
519         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
520         struct disk_sm_root root_le;
521
522         root_le.nr_blocks = cpu_to_le64(smm->ll.nr_blocks);
523         root_le.nr_allocated = cpu_to_le64(smm->ll.nr_allocated);
524         root_le.bitmap_root = cpu_to_le64(smm->ll.bitmap_root);
525         root_le.ref_count_root = cpu_to_le64(smm->ll.ref_count_root);
526
527         if (max < sizeof(root_le))
528                 return -ENOSPC;
529
530         memcpy(where_le, &root_le, sizeof(root_le));
531
532         return 0;
533 }
534
535 static int sm_metadata_extend(struct dm_space_map *sm, dm_block_t extra_blocks);
536
537 static struct dm_space_map ops = {
538         .destroy = sm_metadata_destroy,
539         .extend = sm_metadata_extend,
540         .get_nr_blocks = sm_metadata_get_nr_blocks,
541         .get_nr_free = sm_metadata_get_nr_free,
542         .get_count = sm_metadata_get_count,
543         .count_is_more_than_one = sm_metadata_count_is_more_than_one,
544         .set_count = sm_metadata_set_count,
545         .inc_block = sm_metadata_inc_block,
546         .dec_block = sm_metadata_dec_block,
547         .new_block = sm_metadata_new_block,
548         .commit = sm_metadata_commit,
549         .root_size = sm_metadata_root_size,
550         .copy_root = sm_metadata_copy_root,
551         .register_threshold_callback = sm_metadata_register_threshold_callback
552 };
553
554 /*----------------------------------------------------------------*/
555
556 /*
557  * When a new space map is created that manages its own space.  We use
558  * this tiny bootstrap allocator.
559  */
560 static void sm_bootstrap_destroy(struct dm_space_map *sm)
561 {
562 }
563
564 static int sm_bootstrap_extend(struct dm_space_map *sm, dm_block_t extra_blocks)
565 {
566         DMERR("bootstrap doesn't support extend");
567
568         return -EINVAL;
569 }
570
571 static int sm_bootstrap_get_nr_blocks(struct dm_space_map *sm, dm_block_t *count)
572 {
573         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
574
575         *count = smm->ll.nr_blocks;
576
577         return 0;
578 }
579
580 static int sm_bootstrap_get_nr_free(struct dm_space_map *sm, dm_block_t *count)
581 {
582         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
583
584         *count = smm->ll.nr_blocks - smm->begin;
585
586         return 0;
587 }
588
589 static int sm_bootstrap_get_count(struct dm_space_map *sm, dm_block_t b,
590                                   uint32_t *result)
591 {
592         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
593
594         *result = (b < smm->begin) ? 1 : 0;
595
596         return 0;
597 }
598
599 static int sm_bootstrap_count_is_more_than_one(struct dm_space_map *sm,
600                                                dm_block_t b, int *result)
601 {
602         *result = 0;
603
604         return 0;
605 }
606
607 static int sm_bootstrap_set_count(struct dm_space_map *sm, dm_block_t b,
608                                   uint32_t count)
609 {
610         DMERR("bootstrap doesn't support set_count");
611
612         return -EINVAL;
613 }
614
615 static int sm_bootstrap_new_block(struct dm_space_map *sm, dm_block_t *b)
616 {
617         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
618
619         /*
620          * We know the entire device is unused.
621          */
622         if (smm->begin == smm->ll.nr_blocks)
623                 return -ENOSPC;
624
625         *b = smm->begin++;
626
627         return 0;
628 }
629
630 static int sm_bootstrap_inc_block(struct dm_space_map *sm, dm_block_t b)
631 {
632         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
633
634         return add_bop(smm, BOP_INC, b);
635 }
636
637 static int sm_bootstrap_dec_block(struct dm_space_map *sm, dm_block_t b)
638 {
639         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
640
641         return add_bop(smm, BOP_DEC, b);
642 }
643
644 static int sm_bootstrap_commit(struct dm_space_map *sm)
645 {
646         return 0;
647 }
648
649 static int sm_bootstrap_root_size(struct dm_space_map *sm, size_t *result)
650 {
651         DMERR("bootstrap doesn't support root_size");
652
653         return -EINVAL;
654 }
655
656 static int sm_bootstrap_copy_root(struct dm_space_map *sm, void *where,
657                                   size_t max)
658 {
659         DMERR("bootstrap doesn't support copy_root");
660
661         return -EINVAL;
662 }
663
664 static struct dm_space_map bootstrap_ops = {
665         .destroy = sm_bootstrap_destroy,
666         .extend = sm_bootstrap_extend,
667         .get_nr_blocks = sm_bootstrap_get_nr_blocks,
668         .get_nr_free = sm_bootstrap_get_nr_free,
669         .get_count = sm_bootstrap_get_count,
670         .count_is_more_than_one = sm_bootstrap_count_is_more_than_one,
671         .set_count = sm_bootstrap_set_count,
672         .inc_block = sm_bootstrap_inc_block,
673         .dec_block = sm_bootstrap_dec_block,
674         .new_block = sm_bootstrap_new_block,
675         .commit = sm_bootstrap_commit,
676         .root_size = sm_bootstrap_root_size,
677         .copy_root = sm_bootstrap_copy_root,
678         .register_threshold_callback = NULL
679 };
680
681 /*----------------------------------------------------------------*/
682
683 static int sm_metadata_extend(struct dm_space_map *sm, dm_block_t extra_blocks)
684 {
685         int r, i;
686         enum allocation_event ev;
687         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
688         dm_block_t old_len = smm->ll.nr_blocks;
689
690         /*
691          * Flick into a mode where all blocks get allocated in the new area.
692          */
693         smm->begin = old_len;
694         memcpy(sm, &bootstrap_ops, sizeof(*sm));
695
696         /*
697          * Extend.
698          */
699         r = sm_ll_extend(&smm->ll, extra_blocks);
700         if (r)
701                 goto out;
702
703         /*
704          * We repeatedly increment then commit until the commit doesn't
705          * allocate any new blocks.
706          */
707         do {
708                 for (i = old_len; !r && i < smm->begin; i++) {
709                         r = sm_ll_inc(&smm->ll, i, &ev);
710                         if (r)
711                                 goto out;
712                 }
713                 old_len = smm->begin;
714
715                 r = apply_bops(smm);
716                 if (r) {
717                         DMERR("%s: apply_bops failed", __func__);
718                         goto out;
719                 }
720
721                 r = sm_ll_commit(&smm->ll);
722                 if (r)
723                         goto out;
724
725         } while (old_len != smm->begin);
726
727 out:
728         /*
729          * Switch back to normal behaviour.
730          */
731         memcpy(sm, &ops, sizeof(*sm));
732         return r;
733 }
734
735 /*----------------------------------------------------------------*/
736
737 struct dm_space_map *dm_sm_metadata_init(void)
738 {
739         struct sm_metadata *smm;
740
741         smm = kmalloc(sizeof(*smm), GFP_KERNEL);
742         if (!smm)
743                 return ERR_PTR(-ENOMEM);
744
745         memcpy(&smm->sm, &ops, sizeof(smm->sm));
746
747         return &smm->sm;
748 }
749
750 int dm_sm_metadata_create(struct dm_space_map *sm,
751                           struct dm_transaction_manager *tm,
752                           dm_block_t nr_blocks,
753                           dm_block_t superblock)
754 {
755         int r;
756         dm_block_t i;
757         enum allocation_event ev;
758         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
759
760         smm->begin = superblock + 1;
761         smm->recursion_count = 0;
762         smm->allocated_this_transaction = 0;
763         brb_init(&smm->uncommitted);
764         threshold_init(&smm->threshold);
765
766         memcpy(&smm->sm, &bootstrap_ops, sizeof(smm->sm));
767
768         r = sm_ll_new_metadata(&smm->ll, tm);
769         if (r)
770                 return r;
771
772         if (nr_blocks > DM_SM_METADATA_MAX_BLOCKS)
773                 nr_blocks = DM_SM_METADATA_MAX_BLOCKS;
774         r = sm_ll_extend(&smm->ll, nr_blocks);
775         if (r)
776                 return r;
777
778         memcpy(&smm->sm, &ops, sizeof(smm->sm));
779
780         /*
781          * Now we need to update the newly created data structures with the
782          * allocated blocks that they were built from.
783          */
784         for (i = superblock; !r && i < smm->begin; i++)
785                 r = sm_ll_inc(&smm->ll, i, &ev);
786
787         if (r)
788                 return r;
789
790         r = apply_bops(smm);
791         if (r) {
792                 DMERR("%s: apply_bops failed", __func__);
793                 return r;
794         }
795
796         return sm_metadata_commit(sm);
797 }
798
799 int dm_sm_metadata_open(struct dm_space_map *sm,
800                         struct dm_transaction_manager *tm,
801                         void *root_le, size_t len)
802 {
803         int r;
804         struct sm_metadata *smm = container_of(sm, struct sm_metadata, sm);
805
806         r = sm_ll_open_metadata(&smm->ll, tm, root_le, len);
807         if (r)
808                 return r;
809
810         smm->begin = 0;
811         smm->recursion_count = 0;
812         smm->allocated_this_transaction = 0;
813         brb_init(&smm->uncommitted);
814         threshold_init(&smm->threshold);
815
816         memcpy(&smm->old_ll, &smm->ll, sizeof(smm->old_ll));
817         return 0;
818 }