14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 * If applicable, add the following below this CDDL HEADER, with the
16 * fields enclosed by brackets "[]" replaced with your own identifying
17 * information: Portions Copyright [yyyy] [name of copyright owner]
18 *
19 * CDDL HEADER END
20 */
21 /*
22 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
23 * Copyright (c) 2013 by Delphix. All rights reserved.
24 * Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
25 */
26
27 #include <sys/zfs_context.h>
28 #include <sys/dmu.h>
29 #include <sys/dmu_tx.h>
30 #include <sys/space_map.h>
31 #include <sys/metaslab_impl.h>
32 #include <sys/vdev_impl.h>
33 #include <sys/zio.h>
34
35 /*
36 * Allow allocations to switch to gang blocks quickly. We do this to
37 * avoid having to load lots of space_maps in a given txg. There are,
38 * however, some cases where we want to avoid "fast" ganging and instead
39 * we want to do an exhaustive search of all metaslabs on this device.
40 * Currently we don't allow any gang, zil, or dump device related allocations
41 * to "fast" gang.
42 */
43 #define CAN_FASTGANG(flags) \
44 (!((flags) & (METASLAB_GANG_CHILD | METASLAB_GANG_HEADER | \
45 METASLAB_GANG_AVOID)))
46
47 uint64_t metaslab_aliquot = 512ULL << 10;
48 uint64_t metaslab_gang_bang = SPA_MAXBLOCKSIZE + 1; /* force gang blocks */
49
50 /*
51 * The in-core space map representation is more compact than its on-disk form.
52 * The zfs_condense_pct determines how much more compact the in-core
53 * space_map representation must be before we compact it on-disk.
54 * Values should be greater than or equal to 100.
55 */
56 int zfs_condense_pct = 200;
57
58 /*
59 * This value defines the number of allowed allocation failures per vdev.
60 * If a device reaches this threshold in a given txg then we consider skipping
61 * allocations on that device. The value of zfs_mg_alloc_failures is computed
62 * in zio_init() unless it has been overridden in /etc/system.
63 */
64 int zfs_mg_alloc_failures = 0;
65
66 /*
67 * The zfs_mg_noalloc_threshold defines which metaslab groups should
68 * be eligible for allocation. The value is defined as a percentage of
69 * a free space. Metaslab groups that have more free space than
70 * zfs_mg_noalloc_threshold are always eligible for allocations. Once
71 * a metaslab group's free space is less than or equal to the
72 * zfs_mg_noalloc_threshold the allocator will avoid allocating to that
73 * group unless all groups in the pool have reached zfs_mg_noalloc_threshold.
74 * Once all groups in the pool reach zfs_mg_noalloc_threshold then all
75 * groups are allowed to accept allocations. Gang blocks are always
76 * eligible to allocate on any metaslab group. The default value of 0 means
77 * no metaslab group will be excluded based on this criterion.
78 */
79 int zfs_mg_noalloc_threshold = 0;
80
81 /*
82 * Metaslab debugging: when set, keeps all space maps in core to verify frees.
83 */
84 static int metaslab_debug = 0;
85
86 /*
87 * Minimum size which forces the dynamic allocator to change
88 * it's allocation strategy. Once the space map cannot satisfy
89 * an allocation of this size then it switches to using more
90 * aggressive strategy (i.e search by size rather than offset).
91 */
92 uint64_t metaslab_df_alloc_threshold = SPA_MAXBLOCKSIZE;
93
94 /*
95 * The minimum free space, in percent, which must be available
96 * in a space map to continue allocations in a first-fit fashion.
97 * Once the space_map's free space drops below this level we dynamically
98 * switch to using best-fit allocations.
99 */
100 int metaslab_df_free_pct = 4;
101
102 /*
103 * A metaslab is considered "free" if it contains a contiguous
104 * segment which is greater than metaslab_min_alloc_size.
105 */
106 uint64_t metaslab_min_alloc_size = DMU_MAX_ACCESS;
107
108 /*
109 * Max number of space_maps to prefetch.
110 */
111 int metaslab_prefetch_limit = SPA_DVAS_PER_BP;
112
113 /*
114 * Percentage bonus multiplier for metaslabs that are in the bonus area.
115 */
116 int metaslab_smo_bonus_pct = 150;
117
118 /*
119 * Should we be willing to write data to degraded vdevs?
120 */
121 boolean_t zfs_write_to_degraded = B_FALSE;
122
123 /*
124 * ==========================================================================
125 * Metaslab classes
126 * ==========================================================================
127 */
128 metaslab_class_t *
129 metaslab_class_create(spa_t *spa, space_map_ops_t *ops)
130 {
131 metaslab_class_t *mc;
132
133 mc = kmem_zalloc(sizeof (metaslab_class_t), KM_SLEEP);
134
135 mc->mc_spa = spa;
136 mc->mc_rotor = NULL;
137 mc->mc_ops = ops;
138
139 return (mc);
140 }
141
142 void
143 metaslab_class_destroy(metaslab_class_t *mc)
144 {
145 ASSERT(mc->mc_rotor == NULL);
146 ASSERT(mc->mc_alloc == 0);
147 ASSERT(mc->mc_deferred == 0);
148 ASSERT(mc->mc_space == 0);
149 ASSERT(mc->mc_dspace == 0);
213
214 /*
215 * ==========================================================================
216 * Metaslab groups
217 * ==========================================================================
218 */
219 static int
220 metaslab_compare(const void *x1, const void *x2)
221 {
222 const metaslab_t *m1 = x1;
223 const metaslab_t *m2 = x2;
224
225 if (m1->ms_weight < m2->ms_weight)
226 return (1);
227 if (m1->ms_weight > m2->ms_weight)
228 return (-1);
229
230 /*
231 * If the weights are identical, use the offset to force uniqueness.
232 */
233 if (m1->ms_map->sm_start < m2->ms_map->sm_start)
234 return (-1);
235 if (m1->ms_map->sm_start > m2->ms_map->sm_start)
236 return (1);
237
238 ASSERT3P(m1, ==, m2);
239
240 return (0);
241 }
242
243 /*
244 * Update the allocatable flag and the metaslab group's capacity.
245 * The allocatable flag is set to true if the capacity is below
246 * the zfs_mg_noalloc_threshold. If a metaslab group transitions
247 * from allocatable to non-allocatable or vice versa then the metaslab
248 * group's class is updated to reflect the transition.
249 */
250 static void
251 metaslab_group_alloc_update(metaslab_group_t *mg)
252 {
253 vdev_t *vd = mg->mg_vd;
254 metaslab_class_t *mc = mg->mg_class;
255 vdev_stat_t *vs = &vd->vdev_stat;
283 if (was_allocatable && !mg->mg_allocatable)
284 mc->mc_alloc_groups--;
285 else if (!was_allocatable && mg->mg_allocatable)
286 mc->mc_alloc_groups++;
287 mutex_exit(&mg->mg_lock);
288 }
289
290 metaslab_group_t *
291 metaslab_group_create(metaslab_class_t *mc, vdev_t *vd)
292 {
293 metaslab_group_t *mg;
294
295 mg = kmem_zalloc(sizeof (metaslab_group_t), KM_SLEEP);
296 mutex_init(&mg->mg_lock, NULL, MUTEX_DEFAULT, NULL);
297 avl_create(&mg->mg_metaslab_tree, metaslab_compare,
298 sizeof (metaslab_t), offsetof(struct metaslab, ms_group_node));
299 mg->mg_vd = vd;
300 mg->mg_class = mc;
301 mg->mg_activation_count = 0;
302
303 return (mg);
304 }
305
306 void
307 metaslab_group_destroy(metaslab_group_t *mg)
308 {
309 ASSERT(mg->mg_prev == NULL);
310 ASSERT(mg->mg_next == NULL);
311 /*
312 * We may have gone below zero with the activation count
313 * either because we never activated in the first place or
314 * because we're done, and possibly removing the vdev.
315 */
316 ASSERT(mg->mg_activation_count <= 0);
317
318 avl_destroy(&mg->mg_metaslab_tree);
319 mutex_destroy(&mg->mg_lock);
320 kmem_free(mg, sizeof (metaslab_group_t));
321 }
322
351 }
352 mc->mc_rotor = mg;
353 }
354
355 void
356 metaslab_group_passivate(metaslab_group_t *mg)
357 {
358 metaslab_class_t *mc = mg->mg_class;
359 metaslab_group_t *mgprev, *mgnext;
360
361 ASSERT(spa_config_held(mc->mc_spa, SCL_ALLOC, RW_WRITER));
362
363 if (--mg->mg_activation_count != 0) {
364 ASSERT(mc->mc_rotor != mg);
365 ASSERT(mg->mg_prev == NULL);
366 ASSERT(mg->mg_next == NULL);
367 ASSERT(mg->mg_activation_count < 0);
368 return;
369 }
370
371 mgprev = mg->mg_prev;
372 mgnext = mg->mg_next;
373
374 if (mg == mgnext) {
375 mc->mc_rotor = NULL;
376 } else {
377 mc->mc_rotor = mgnext;
378 mgprev->mg_next = mgnext;
379 mgnext->mg_prev = mgprev;
380 }
381
382 mg->mg_prev = NULL;
383 mg->mg_next = NULL;
384 }
385
386 static void
387 metaslab_group_add(metaslab_group_t *mg, metaslab_t *msp)
388 {
389 mutex_enter(&mg->mg_lock);
390 ASSERT(msp->ms_group == NULL);
430 */
431 static boolean_t
432 metaslab_group_allocatable(metaslab_group_t *mg)
433 {
434 vdev_t *vd = mg->mg_vd;
435 spa_t *spa = vd->vdev_spa;
436 metaslab_class_t *mc = mg->mg_class;
437
438 /*
439 * A metaslab group is considered allocatable if its free capacity
440 * is greater than the set value of zfs_mg_noalloc_threshold, it's
441 * associated with a slog, or there are no other metaslab groups
442 * with free capacity greater than zfs_mg_noalloc_threshold.
443 */
444 return (mg->mg_free_capacity > zfs_mg_noalloc_threshold ||
445 mc != spa_normal_class(spa) || mc->mc_alloc_groups == 0);
446 }
447
448 /*
449 * ==========================================================================
450 * Common allocator routines
451 * ==========================================================================
452 */
453 static int
454 metaslab_segsize_compare(const void *x1, const void *x2)
455 {
456 const space_seg_t *s1 = x1;
457 const space_seg_t *s2 = x2;
458 uint64_t ss_size1 = s1->ss_end - s1->ss_start;
459 uint64_t ss_size2 = s2->ss_end - s2->ss_start;
460
461 if (ss_size1 < ss_size2)
462 return (-1);
463 if (ss_size1 > ss_size2)
464 return (1);
465
466 if (s1->ss_start < s2->ss_start)
467 return (-1);
468 if (s1->ss_start > s2->ss_start)
469 return (1);
470
471 return (0);
472 }
473
474 /*
475 * This is a helper function that can be used by the allocator to find
476 * a suitable block to allocate. This will search the specified AVL
477 * tree looking for a block that matches the specified criteria.
478 */
479 static uint64_t
480 metaslab_block_picker(avl_tree_t *t, uint64_t *cursor, uint64_t size,
481 uint64_t align)
482 {
483 space_seg_t *ss, ssearch;
484 avl_index_t where;
485
486 ssearch.ss_start = *cursor;
487 ssearch.ss_end = *cursor + size;
488
489 ss = avl_find(t, &ssearch, &where);
490 if (ss == NULL)
491 ss = avl_nearest(t, where, AVL_AFTER);
492
493 while (ss != NULL) {
494 uint64_t offset = P2ROUNDUP(ss->ss_start, align);
495
496 if (offset + size <= ss->ss_end) {
497 *cursor = offset + size;
498 return (offset);
499 }
500 ss = AVL_NEXT(t, ss);
501 }
502
503 /*
504 * If we know we've searched the whole map (*cursor == 0), give up.
505 * Otherwise, reset the cursor to the beginning and try again.
506 */
507 if (*cursor == 0)
508 return (-1ULL);
509
510 *cursor = 0;
511 return (metaslab_block_picker(t, cursor, size, align));
512 }
513
514 static void
515 metaslab_pp_load(space_map_t *sm)
516 {
517 space_seg_t *ss;
518
519 ASSERT(sm->sm_ppd == NULL);
520 sm->sm_ppd = kmem_zalloc(64 * sizeof (uint64_t), KM_SLEEP);
521
522 sm->sm_pp_root = kmem_alloc(sizeof (avl_tree_t), KM_SLEEP);
523 avl_create(sm->sm_pp_root, metaslab_segsize_compare,
524 sizeof (space_seg_t), offsetof(struct space_seg, ss_pp_node));
525
526 for (ss = avl_first(&sm->sm_root); ss; ss = AVL_NEXT(&sm->sm_root, ss))
527 avl_add(sm->sm_pp_root, ss);
528 }
529
530 static void
531 metaslab_pp_unload(space_map_t *sm)
532 {
533 void *cookie = NULL;
534
535 kmem_free(sm->sm_ppd, 64 * sizeof (uint64_t));
536 sm->sm_ppd = NULL;
537
538 while (avl_destroy_nodes(sm->sm_pp_root, &cookie) != NULL) {
539 /* tear down the tree */
540 }
541
542 avl_destroy(sm->sm_pp_root);
543 kmem_free(sm->sm_pp_root, sizeof (avl_tree_t));
544 sm->sm_pp_root = NULL;
545 }
546
547 /* ARGSUSED */
548 static void
549 metaslab_pp_claim(space_map_t *sm, uint64_t start, uint64_t size)
550 {
551 /* No need to update cursor */
552 }
553
554 /* ARGSUSED */
555 static void
556 metaslab_pp_free(space_map_t *sm, uint64_t start, uint64_t size)
557 {
558 /* No need to update cursor */
559 }
560
561 /*
562 * Return the maximum contiguous segment within the metaslab.
563 */
564 uint64_t
565 metaslab_pp_maxsize(space_map_t *sm)
566 {
567 avl_tree_t *t = sm->sm_pp_root;
568 space_seg_t *ss;
569
570 if (t == NULL || (ss = avl_last(t)) == NULL)
571 return (0ULL);
572
573 return (ss->ss_end - ss->ss_start);
574 }
575
576 /*
577 * ==========================================================================
578 * The first-fit block allocator
579 * ==========================================================================
580 */
581 static uint64_t
582 metaslab_ff_alloc(space_map_t *sm, uint64_t size)
583 {
584 avl_tree_t *t = &sm->sm_root;
585 uint64_t align = size & -size;
586 uint64_t *cursor = (uint64_t *)sm->sm_ppd + highbit(align) - 1;
587
588 return (metaslab_block_picker(t, cursor, size, align));
589 }
590
591 /* ARGSUSED */
592 boolean_t
593 metaslab_ff_fragmented(space_map_t *sm)
594 {
595 return (B_TRUE);
596 }
597
598 static space_map_ops_t metaslab_ff_ops = {
599 metaslab_pp_load,
600 metaslab_pp_unload,
601 metaslab_ff_alloc,
602 metaslab_pp_claim,
603 metaslab_pp_free,
604 metaslab_pp_maxsize,
605 metaslab_ff_fragmented
606 };
607
608 /*
609 * ==========================================================================
610 * Dynamic block allocator -
611 * Uses the first fit allocation scheme until space get low and then
612 * adjusts to a best fit allocation method. Uses metaslab_df_alloc_threshold
613 * and metaslab_df_free_pct to determine when to switch the allocation scheme.
614 * ==========================================================================
615 */
616 static uint64_t
617 metaslab_df_alloc(space_map_t *sm, uint64_t size)
618 {
619 avl_tree_t *t = &sm->sm_root;
620 uint64_t align = size & -size;
621 uint64_t *cursor = (uint64_t *)sm->sm_ppd + highbit(align) - 1;
622 uint64_t max_size = metaslab_pp_maxsize(sm);
623 int free_pct = sm->sm_space * 100 / sm->sm_size;
624
625 ASSERT(MUTEX_HELD(sm->sm_lock));
626 ASSERT3U(avl_numnodes(&sm->sm_root), ==, avl_numnodes(sm->sm_pp_root));
627
628 if (max_size < size)
629 return (-1ULL);
630
631 /*
632 * If we're running low on space switch to using the size
633 * sorted AVL tree (best-fit).
634 */
635 if (max_size < metaslab_df_alloc_threshold ||
636 free_pct < metaslab_df_free_pct) {
637 t = sm->sm_pp_root;
638 *cursor = 0;
639 }
640
641 return (metaslab_block_picker(t, cursor, size, 1ULL));
642 }
643
644 static boolean_t
645 metaslab_df_fragmented(space_map_t *sm)
646 {
647 uint64_t max_size = metaslab_pp_maxsize(sm);
648 int free_pct = sm->sm_space * 100 / sm->sm_size;
649
650 if (max_size >= metaslab_df_alloc_threshold &&
651 free_pct >= metaslab_df_free_pct)
652 return (B_FALSE);
653
654 return (B_TRUE);
655 }
656
657 static space_map_ops_t metaslab_df_ops = {
658 metaslab_pp_load,
659 metaslab_pp_unload,
660 metaslab_df_alloc,
661 metaslab_pp_claim,
662 metaslab_pp_free,
663 metaslab_pp_maxsize,
664 metaslab_df_fragmented
665 };
666
667 /*
668 * ==========================================================================
669 * Other experimental allocators
670 * ==========================================================================
671 */
672 static uint64_t
673 metaslab_cdf_alloc(space_map_t *sm, uint64_t size)
674 {
675 avl_tree_t *t = &sm->sm_root;
676 uint64_t *cursor = (uint64_t *)sm->sm_ppd;
677 uint64_t *extent_end = (uint64_t *)sm->sm_ppd + 1;
678 uint64_t max_size = metaslab_pp_maxsize(sm);
679 uint64_t rsize = size;
680 uint64_t offset = 0;
681
682 ASSERT(MUTEX_HELD(sm->sm_lock));
683 ASSERT3U(avl_numnodes(&sm->sm_root), ==, avl_numnodes(sm->sm_pp_root));
684
685 if (max_size < size)
686 return (-1ULL);
687
688 ASSERT3U(*extent_end, >=, *cursor);
689
690 /*
691 * If we're running low on space switch to using the size
692 * sorted AVL tree (best-fit).
693 */
694 if ((*cursor + size) > *extent_end) {
695
696 t = sm->sm_pp_root;
697 *cursor = *extent_end = 0;
698
699 if (max_size > 2 * SPA_MAXBLOCKSIZE)
700 rsize = MIN(metaslab_min_alloc_size, max_size);
701 offset = metaslab_block_picker(t, extent_end, rsize, 1ULL);
702 if (offset != -1)
703 *cursor = offset + size;
704 } else {
705 offset = metaslab_block_picker(t, cursor, rsize, 1ULL);
706 }
707 ASSERT3U(*cursor, <=, *extent_end);
708 return (offset);
709 }
710
711 static boolean_t
712 metaslab_cdf_fragmented(space_map_t *sm)
713 {
714 uint64_t max_size = metaslab_pp_maxsize(sm);
715
716 if (max_size > (metaslab_min_alloc_size * 10))
717 return (B_FALSE);
718 return (B_TRUE);
719 }
720
721 static space_map_ops_t metaslab_cdf_ops = {
722 metaslab_pp_load,
723 metaslab_pp_unload,
724 metaslab_cdf_alloc,
725 metaslab_pp_claim,
726 metaslab_pp_free,
727 metaslab_pp_maxsize,
728 metaslab_cdf_fragmented
729 };
730
731 uint64_t metaslab_ndf_clump_shift = 4;
732
733 static uint64_t
734 metaslab_ndf_alloc(space_map_t *sm, uint64_t size)
735 {
736 avl_tree_t *t = &sm->sm_root;
737 avl_index_t where;
738 space_seg_t *ss, ssearch;
739 uint64_t hbit = highbit(size);
740 uint64_t *cursor = (uint64_t *)sm->sm_ppd + hbit - 1;
741 uint64_t max_size = metaslab_pp_maxsize(sm);
742
743 ASSERT(MUTEX_HELD(sm->sm_lock));
744 ASSERT3U(avl_numnodes(&sm->sm_root), ==, avl_numnodes(sm->sm_pp_root));
745
746 if (max_size < size)
747 return (-1ULL);
748
749 ssearch.ss_start = *cursor;
750 ssearch.ss_end = *cursor + size;
751
752 ss = avl_find(t, &ssearch, &where);
753 if (ss == NULL || (ss->ss_start + size > ss->ss_end)) {
754 t = sm->sm_pp_root;
755
756 ssearch.ss_start = 0;
757 ssearch.ss_end = MIN(max_size,
758 1ULL << (hbit + metaslab_ndf_clump_shift));
759 ss = avl_find(t, &ssearch, &where);
760 if (ss == NULL)
761 ss = avl_nearest(t, where, AVL_AFTER);
762 ASSERT(ss != NULL);
763 }
764
765 if (ss != NULL) {
766 if (ss->ss_start + size <= ss->ss_end) {
767 *cursor = ss->ss_start + size;
768 return (ss->ss_start);
769 }
770 }
771 return (-1ULL);
772 }
773
774 static boolean_t
775 metaslab_ndf_fragmented(space_map_t *sm)
776 {
777 uint64_t max_size = metaslab_pp_maxsize(sm);
778
779 if (max_size > (metaslab_min_alloc_size << metaslab_ndf_clump_shift))
780 return (B_FALSE);
781 return (B_TRUE);
782 }
783
784
785 static space_map_ops_t metaslab_ndf_ops = {
786 metaslab_pp_load,
787 metaslab_pp_unload,
788 metaslab_ndf_alloc,
789 metaslab_pp_claim,
790 metaslab_pp_free,
791 metaslab_pp_maxsize,
792 metaslab_ndf_fragmented
793 };
794
795 space_map_ops_t *zfs_metaslab_ops = &metaslab_df_ops;
796
797 /*
798 * ==========================================================================
799 * Metaslabs
800 * ==========================================================================
801 */
802 metaslab_t *
803 metaslab_init(metaslab_group_t *mg, space_map_obj_t *smo,
804 uint64_t start, uint64_t size, uint64_t txg)
805 {
806 vdev_t *vd = mg->mg_vd;
807 metaslab_t *msp;
808
809 msp = kmem_zalloc(sizeof (metaslab_t), KM_SLEEP);
810 mutex_init(&msp->ms_lock, NULL, MUTEX_DEFAULT, NULL);
811
812 msp->ms_smo_syncing = *smo;
813
814 /*
815 * We create the main space map here, but we don't create the
816 * allocmaps and freemaps until metaslab_sync_done(). This serves
817 * two purposes: it allows metaslab_sync_done() to detect the
818 * addition of new space; and for debugging, it ensures that we'd
819 * data fault on any attempt to use this metaslab before it's ready.
820 */
821 msp->ms_map = kmem_zalloc(sizeof (space_map_t), KM_SLEEP);
822 space_map_create(msp->ms_map, start, size,
823 vd->vdev_ashift, &msp->ms_lock);
824
825 metaslab_group_add(mg, msp);
826
827 if (metaslab_debug && smo->smo_object != 0) {
828 mutex_enter(&msp->ms_lock);
829 VERIFY(space_map_load(msp->ms_map, mg->mg_class->mc_ops,
830 SM_FREE, smo, spa_meta_objset(vd->vdev_spa)) == 0);
831 mutex_exit(&msp->ms_lock);
832 }
833
834 /*
835 * If we're opening an existing pool (txg == 0) or creating
836 * a new one (txg == TXG_INITIAL), all space is available now.
837 * If we're adding space to an existing pool, the new space
838 * does not become available until after this txg has synced.
839 */
840 if (txg <= TXG_INITIAL)
841 metaslab_sync_done(msp, 0);
842
843 if (txg != 0) {
844 vdev_dirty(vd, 0, NULL, txg);
845 vdev_dirty(vd, VDD_METASLAB, msp, txg);
846 }
847
848 return (msp);
849 }
850
851 void
852 metaslab_fini(metaslab_t *msp)
853 {
854 metaslab_group_t *mg = msp->ms_group;
855
856 vdev_space_update(mg->mg_vd,
857 -msp->ms_smo.smo_alloc, 0, -msp->ms_map->sm_size);
858
859 metaslab_group_remove(mg, msp);
860
861 mutex_enter(&msp->ms_lock);
862
863 space_map_unload(msp->ms_map);
864 space_map_destroy(msp->ms_map);
865 kmem_free(msp->ms_map, sizeof (*msp->ms_map));
866
867 for (int t = 0; t < TXG_SIZE; t++) {
868 space_map_destroy(msp->ms_allocmap[t]);
869 space_map_destroy(msp->ms_freemap[t]);
870 kmem_free(msp->ms_allocmap[t], sizeof (*msp->ms_allocmap[t]));
871 kmem_free(msp->ms_freemap[t], sizeof (*msp->ms_freemap[t]));
872 }
873
874 for (int t = 0; t < TXG_DEFER_SIZE; t++) {
875 space_map_destroy(msp->ms_defermap[t]);
876 kmem_free(msp->ms_defermap[t], sizeof (*msp->ms_defermap[t]));
877 }
878
879 ASSERT0(msp->ms_deferspace);
880
881 mutex_exit(&msp->ms_lock);
882 mutex_destroy(&msp->ms_lock);
883
884 kmem_free(msp, sizeof (metaslab_t));
885 }
886
887 #define METASLAB_WEIGHT_PRIMARY (1ULL << 63)
888 #define METASLAB_WEIGHT_SECONDARY (1ULL << 62)
889 #define METASLAB_ACTIVE_MASK \
890 (METASLAB_WEIGHT_PRIMARY | METASLAB_WEIGHT_SECONDARY)
891
892 static uint64_t
893 metaslab_weight(metaslab_t *msp)
894 {
895 metaslab_group_t *mg = msp->ms_group;
896 space_map_t *sm = msp->ms_map;
897 space_map_obj_t *smo = &msp->ms_smo;
898 vdev_t *vd = mg->mg_vd;
899 uint64_t weight, space;
900
901 ASSERT(MUTEX_HELD(&msp->ms_lock));
902
903 /*
904 * This vdev is in the process of being removed so there is nothing
905 * for us to do here.
906 */
907 if (vd->vdev_removing) {
908 ASSERT0(smo->smo_alloc);
909 ASSERT0(vd->vdev_ms_shift);
910 return (0);
911 }
912
913 /*
914 * The baseline weight is the metaslab's free space.
915 */
916 space = sm->sm_size - smo->smo_alloc;
917 weight = space;
918
919 /*
920 * Modern disks have uniform bit density and constant angular velocity.
921 * Therefore, the outer recording zones are faster (higher bandwidth)
922 * than the inner zones by the ratio of outer to inner track diameter,
923 * which is typically around 2:1. We account for this by assigning
924 * higher weight to lower metaslabs (multiplier ranging from 2x to 1x).
925 * In effect, this means that we'll select the metaslab with the most
926 * free bandwidth rather than simply the one with the most free space.
927 */
928 weight = 2 * weight -
929 ((sm->sm_start >> vd->vdev_ms_shift) * weight) / vd->vdev_ms_count;
930 ASSERT(weight >= space && weight <= 2 * space);
931
932 /*
933 * For locality, assign higher weight to metaslabs which have
934 * a lower offset than what we've already activated.
935 */
936 if (sm->sm_start <= mg->mg_bonus_area)
937 weight *= (metaslab_smo_bonus_pct / 100);
938 ASSERT(weight >= space &&
939 weight <= 2 * (metaslab_smo_bonus_pct / 100) * space);
940
941 if (sm->sm_loaded && !sm->sm_ops->smop_fragmented(sm)) {
942 /*
943 * If this metaslab is one we're actively using, adjust its
944 * weight to make it preferable to any inactive metaslab so
945 * we'll polish it off.
946 */
947 weight |= (msp->ms_weight & METASLAB_ACTIVE_MASK);
948 }
949 return (weight);
950 }
951
952 static void
953 metaslab_prefetch(metaslab_group_t *mg)
954 {
955 spa_t *spa = mg->mg_vd->vdev_spa;
956 metaslab_t *msp;
957 avl_tree_t *t = &mg->mg_metaslab_tree;
958 int m;
959
960 mutex_enter(&mg->mg_lock);
961
962 /*
963 * Prefetch the next potential metaslabs
964 */
965 for (msp = avl_first(t), m = 0; msp; msp = AVL_NEXT(t, msp), m++) {
966 space_map_t *sm = msp->ms_map;
967 space_map_obj_t *smo = &msp->ms_smo;
968
969 /* If we have reached our prefetch limit then we're done */
970 if (m >= metaslab_prefetch_limit)
971 break;
972
973 if (!sm->sm_loaded && smo->smo_object != 0) {
974 mutex_exit(&mg->mg_lock);
975 dmu_prefetch(spa_meta_objset(spa), smo->smo_object,
976 0ULL, smo->smo_objsize);
977 mutex_enter(&mg->mg_lock);
978 }
979 }
980 mutex_exit(&mg->mg_lock);
981 }
982
983 static int
984 metaslab_activate(metaslab_t *msp, uint64_t activation_weight)
985 {
986 metaslab_group_t *mg = msp->ms_group;
987 space_map_t *sm = msp->ms_map;
988 space_map_ops_t *sm_ops = msp->ms_group->mg_class->mc_ops;
989
990 ASSERT(MUTEX_HELD(&msp->ms_lock));
991
992 if ((msp->ms_weight & METASLAB_ACTIVE_MASK) == 0) {
993 space_map_load_wait(sm);
994 if (!sm->sm_loaded) {
995 space_map_obj_t *smo = &msp->ms_smo;
996
997 int error = space_map_load(sm, sm_ops, SM_FREE, smo,
998 spa_meta_objset(msp->ms_group->mg_vd->vdev_spa));
999 if (error) {
1000 metaslab_group_sort(msp->ms_group, msp, 0);
1001 return (error);
1002 }
1003 for (int t = 0; t < TXG_DEFER_SIZE; t++)
1004 space_map_walk(msp->ms_defermap[t],
1005 space_map_claim, sm);
1006
1007 }
1008
1009 /*
1010 * Track the bonus area as we activate new metaslabs.
1011 */
1012 if (sm->sm_start > mg->mg_bonus_area) {
1013 mutex_enter(&mg->mg_lock);
1014 mg->mg_bonus_area = sm->sm_start;
1015 mutex_exit(&mg->mg_lock);
1016 }
1017
1018 metaslab_group_sort(msp->ms_group, msp,
1019 msp->ms_weight | activation_weight);
1020 }
1021 ASSERT(sm->sm_loaded);
1022 ASSERT(msp->ms_weight & METASLAB_ACTIVE_MASK);
1023
1024 return (0);
1025 }
1026
1027 static void
1028 metaslab_passivate(metaslab_t *msp, uint64_t size)
1029 {
1030 /*
1031 * If size < SPA_MINBLOCKSIZE, then we will not allocate from
1032 * this metaslab again. In that case, it had better be empty,
1033 * or we would be leaving space on the table.
1034 */
1035 ASSERT(size >= SPA_MINBLOCKSIZE || msp->ms_map->sm_space == 0);
1036 metaslab_group_sort(msp->ms_group, msp, MIN(msp->ms_weight, size));
1037 ASSERT((msp->ms_weight & METASLAB_ACTIVE_MASK) == 0);
1038 }
1039
1040 /*
1041 * Determine if the in-core space map representation can be condensed on-disk.
1042 * We would like to use the following criteria to make our decision:
1043 *
1044 * 1. The size of the space map object should not dramatically increase as a
1045 * result of writing out our in-core free map.
1046 *
1047 * 2. The minimal on-disk space map representation is zfs_condense_pct/100
1048 * times the size than the in-core representation (i.e. zfs_condense_pct = 110
1049 * and in-core = 1MB, minimal = 1.1.MB).
1050 *
1051 * Checking the first condition is tricky since we don't want to walk
1052 * the entire AVL tree calculating the estimated on-disk size. Instead we
1053 * use the size-ordered AVL tree in the space map and calculate the
1054 * size required for the largest segment in our in-core free map. If the
1055 * size required to represent that segment on disk is larger than the space
1056 * map object then we avoid condensing this map.
1057 *
1058 * To determine the second criterion we use a best-case estimate and assume
1059 * each segment can be represented on-disk as a single 64-bit entry. We refer
1060 * to this best-case estimate as the space map's minimal form.
1061 */
1062 static boolean_t
1063 metaslab_should_condense(metaslab_t *msp)
1064 {
1065 space_map_t *sm = msp->ms_map;
1066 space_map_obj_t *smo = &msp->ms_smo_syncing;
1067 space_seg_t *ss;
1068 uint64_t size, entries, segsz;
1069
1070 ASSERT(MUTEX_HELD(&msp->ms_lock));
1071 ASSERT(sm->sm_loaded);
1072
1073 /*
1074 * Use the sm_pp_root AVL tree, which is ordered by size, to obtain
1075 * the largest segment in the in-core free map. If the tree is
1076 * empty then we should condense the map.
1077 */
1078 ss = avl_last(sm->sm_pp_root);
1079 if (ss == NULL)
1080 return (B_TRUE);
1081
1082 /*
1083 * Calculate the number of 64-bit entries this segment would
1084 * require when written to disk. If this single segment would be
1085 * larger on-disk than the entire current on-disk structure, then
1086 * clearly condensing will increase the on-disk structure size.
1087 */
1088 size = (ss->ss_end - ss->ss_start) >> sm->sm_shift;
1089 entries = size / (MIN(size, SM_RUN_MAX));
1090 segsz = entries * sizeof (uint64_t);
1091
1092 return (segsz <= smo->smo_objsize &&
1093 smo->smo_objsize >= (zfs_condense_pct *
1094 sizeof (uint64_t) * avl_numnodes(&sm->sm_root)) / 100);
1095 }
1096
1097 /*
1098 * Condense the on-disk space map representation to its minimized form.
1099 * The minimized form consists of a small number of allocations followed by
1100 * the in-core free map.
1101 */
1102 static void
1103 metaslab_condense(metaslab_t *msp, uint64_t txg, dmu_tx_t *tx)
1104 {
1105 spa_t *spa = msp->ms_group->mg_vd->vdev_spa;
1106 space_map_t *freemap = msp->ms_freemap[txg & TXG_MASK];
1107 space_map_t condense_map;
1108 space_map_t *sm = msp->ms_map;
1109 objset_t *mos = spa_meta_objset(spa);
1110 space_map_obj_t *smo = &msp->ms_smo_syncing;
1111
1112 ASSERT(MUTEX_HELD(&msp->ms_lock));
1113 ASSERT3U(spa_sync_pass(spa), ==, 1);
1114 ASSERT(sm->sm_loaded);
1115
1116 spa_dbgmsg(spa, "condensing: txg %llu, msp[%llu] %p, "
1117 "smo size %llu, segments %lu", txg,
1118 (msp->ms_map->sm_start / msp->ms_map->sm_size), msp,
1119 smo->smo_objsize, avl_numnodes(&sm->sm_root));
1120
1121 /*
1122 * Create an map that is a 100% allocated map. We remove segments
1123 * that have been freed in this txg, any deferred frees that exist,
1124 * and any allocation in the future. Removing segments should be
1125 * a relatively inexpensive operation since we expect these maps to
1126 * a small number of nodes.
1127 */
1128 space_map_create(&condense_map, sm->sm_start, sm->sm_size,
1129 sm->sm_shift, sm->sm_lock);
1130 space_map_add(&condense_map, condense_map.sm_start,
1131 condense_map.sm_size);
1132
1133 /*
1134 * Remove what's been freed in this txg from the condense_map.
1135 * Since we're in sync_pass 1, we know that all the frees from
1136 * this txg are in the freemap.
1137 */
1138 space_map_walk(freemap, space_map_remove, &condense_map);
1139
1140 for (int t = 0; t < TXG_DEFER_SIZE; t++)
1141 space_map_walk(msp->ms_defermap[t],
1142 space_map_remove, &condense_map);
1143
1144 for (int t = 1; t < TXG_CONCURRENT_STATES; t++)
1145 space_map_walk(msp->ms_allocmap[(txg + t) & TXG_MASK],
1146 space_map_remove, &condense_map);
1147
1148 /*
1149 * We're about to drop the metaslab's lock thus allowing
1150 * other consumers to change it's content. Set the
1151 * space_map's sm_condensing flag to ensure that
1152 * allocations on this metaslab do not occur while we're
1153 * in the middle of committing it to disk. This is only critical
1154 * for the ms_map as all other space_maps use per txg
1155 * views of their content.
1156 */
1157 sm->sm_condensing = B_TRUE;
1158
1159 mutex_exit(&msp->ms_lock);
1160 space_map_truncate(smo, mos, tx);
1161 mutex_enter(&msp->ms_lock);
1162
1163 /*
1164 * While we would ideally like to create a space_map representation
1165 * that consists only of allocation records, doing so can be
1166 * prohibitively expensive because the in-core free map can be
1167 * large, and therefore computationally expensive to subtract
1168 * from the condense_map. Instead we sync out two maps, a cheap
1169 * allocation only map followed by the in-core free map. While not
1170 * optimal, this is typically close to optimal, and much cheaper to
1171 * compute.
1172 */
1173 space_map_sync(&condense_map, SM_ALLOC, smo, mos, tx);
1174 space_map_vacate(&condense_map, NULL, NULL);
1175 space_map_destroy(&condense_map);
1176
1177 space_map_sync(sm, SM_FREE, smo, mos, tx);
1178 sm->sm_condensing = B_FALSE;
1179
1180 spa_dbgmsg(spa, "condensed: txg %llu, msp[%llu] %p, "
1181 "smo size %llu", txg,
1182 (msp->ms_map->sm_start / msp->ms_map->sm_size), msp,
1183 smo->smo_objsize);
1184 }
1185
1186 /*
1187 * Write a metaslab to disk in the context of the specified transaction group.
1188 */
1189 void
1190 metaslab_sync(metaslab_t *msp, uint64_t txg)
1191 {
1192 vdev_t *vd = msp->ms_group->mg_vd;
1193 spa_t *spa = vd->vdev_spa;
1194 objset_t *mos = spa_meta_objset(spa);
1195 space_map_t *allocmap = msp->ms_allocmap[txg & TXG_MASK];
1196 space_map_t **freemap = &msp->ms_freemap[txg & TXG_MASK];
1197 space_map_t **freed_map = &msp->ms_freemap[TXG_CLEAN(txg) & TXG_MASK];
1198 space_map_t *sm = msp->ms_map;
1199 space_map_obj_t *smo = &msp->ms_smo_syncing;
1200 dmu_buf_t *db;
1201 dmu_tx_t *tx;
1202
1203 ASSERT(!vd->vdev_ishole);
1204
1205 /*
1206 * This metaslab has just been added so there's no work to do now.
1207 */
1208 if (*freemap == NULL) {
1209 ASSERT3P(allocmap, ==, NULL);
1210 return;
1211 }
1212
1213 ASSERT3P(allocmap, !=, NULL);
1214 ASSERT3P(*freemap, !=, NULL);
1215 ASSERT3P(*freed_map, !=, NULL);
1216
1217 if (allocmap->sm_space == 0 && (*freemap)->sm_space == 0)
1218 return;
1219
1220 /*
1221 * The only state that can actually be changing concurrently with
1222 * metaslab_sync() is the metaslab's ms_map. No other thread can
1223 * be modifying this txg's allocmap, freemap, freed_map, or smo.
1224 * Therefore, we only hold ms_lock to satify space_map ASSERTs.
1225 * We drop it whenever we call into the DMU, because the DMU
1226 * can call down to us (e.g. via zio_free()) at any time.
1227 */
1228
1229 tx = dmu_tx_create_assigned(spa_get_dsl(spa), txg);
1230
1231 if (smo->smo_object == 0) {
1232 ASSERT(smo->smo_objsize == 0);
1233 ASSERT(smo->smo_alloc == 0);
1234 smo->smo_object = dmu_object_alloc(mos,
1235 DMU_OT_SPACE_MAP, 1 << SPACE_MAP_BLOCKSHIFT,
1236 DMU_OT_SPACE_MAP_HEADER, sizeof (*smo), tx);
1237 ASSERT(smo->smo_object != 0);
1238 dmu_write(mos, vd->vdev_ms_array, sizeof (uint64_t) *
1239 (sm->sm_start >> vd->vdev_ms_shift),
1240 sizeof (uint64_t), &smo->smo_object, tx);
1241 }
1242
1243 mutex_enter(&msp->ms_lock);
1244
1245 if (sm->sm_loaded && spa_sync_pass(spa) == 1 &&
1246 metaslab_should_condense(msp)) {
1247 metaslab_condense(msp, txg, tx);
1248 } else {
1249 space_map_sync(allocmap, SM_ALLOC, smo, mos, tx);
1250 space_map_sync(*freemap, SM_FREE, smo, mos, tx);
1251 }
1252
1253 space_map_vacate(allocmap, NULL, NULL);
1254
1255 /*
1256 * For sync pass 1, we avoid walking the entire space map and
1257 * instead will just swap the pointers for freemap and
1258 * freed_map. We can safely do this since the freed_map is
1259 * guaranteed to be empty on the initial pass.
1260 */
1261 if (spa_sync_pass(spa) == 1) {
1262 ASSERT0((*freed_map)->sm_space);
1263 ASSERT0(avl_numnodes(&(*freed_map)->sm_root));
1264 space_map_swap(freemap, freed_map);
1265 } else {
1266 space_map_vacate(*freemap, space_map_add, *freed_map);
1267 }
1268
1269 ASSERT0(msp->ms_allocmap[txg & TXG_MASK]->sm_space);
1270 ASSERT0(msp->ms_freemap[txg & TXG_MASK]->sm_space);
1271
1272 mutex_exit(&msp->ms_lock);
1273
1274 VERIFY0(dmu_bonus_hold(mos, smo->smo_object, FTAG, &db));
1275 dmu_buf_will_dirty(db, tx);
1276 ASSERT3U(db->db_size, >=, sizeof (*smo));
1277 bcopy(smo, db->db_data, sizeof (*smo));
1278 dmu_buf_rele(db, FTAG);
1279
1280 dmu_tx_commit(tx);
1281 }
1282
1283 /*
1284 * Called after a transaction group has completely synced to mark
1285 * all of the metaslab's free space as usable.
1286 */
1287 void
1288 metaslab_sync_done(metaslab_t *msp, uint64_t txg)
1289 {
1290 space_map_obj_t *smo = &msp->ms_smo;
1291 space_map_obj_t *smosync = &msp->ms_smo_syncing;
1292 space_map_t *sm = msp->ms_map;
1293 space_map_t **freed_map = &msp->ms_freemap[TXG_CLEAN(txg) & TXG_MASK];
1294 space_map_t **defer_map = &msp->ms_defermap[txg % TXG_DEFER_SIZE];
1295 metaslab_group_t *mg = msp->ms_group;
1296 vdev_t *vd = mg->mg_vd;
1297 int64_t alloc_delta, defer_delta;
1298
1299 ASSERT(!vd->vdev_ishole);
1300
1301 mutex_enter(&msp->ms_lock);
1302
1303 /*
1304 * If this metaslab is just becoming available, initialize its
1305 * allocmaps, freemaps, and defermap and add its capacity to the vdev.
1306 */
1307 if (*freed_map == NULL) {
1308 ASSERT(*defer_map == NULL);
1309 for (int t = 0; t < TXG_SIZE; t++) {
1310 msp->ms_allocmap[t] = kmem_zalloc(sizeof (space_map_t),
1311 KM_SLEEP);
1312 space_map_create(msp->ms_allocmap[t], sm->sm_start,
1313 sm->sm_size, sm->sm_shift, sm->sm_lock);
1314 msp->ms_freemap[t] = kmem_zalloc(sizeof (space_map_t),
1315 KM_SLEEP);
1316 space_map_create(msp->ms_freemap[t], sm->sm_start,
1317 sm->sm_size, sm->sm_shift, sm->sm_lock);
1318 }
1319
1320 for (int t = 0; t < TXG_DEFER_SIZE; t++) {
1321 msp->ms_defermap[t] = kmem_zalloc(sizeof (space_map_t),
1322 KM_SLEEP);
1323 space_map_create(msp->ms_defermap[t], sm->sm_start,
1324 sm->sm_size, sm->sm_shift, sm->sm_lock);
1325 }
1326
1327 freed_map = &msp->ms_freemap[TXG_CLEAN(txg) & TXG_MASK];
1328 defer_map = &msp->ms_defermap[txg % TXG_DEFER_SIZE];
1329
1330 vdev_space_update(vd, 0, 0, sm->sm_size);
1331 }
1332
1333 alloc_delta = smosync->smo_alloc - smo->smo_alloc;
1334 defer_delta = (*freed_map)->sm_space - (*defer_map)->sm_space;
1335
1336 vdev_space_update(vd, alloc_delta + defer_delta, defer_delta, 0);
1337
1338 ASSERT(msp->ms_allocmap[txg & TXG_MASK]->sm_space == 0);
1339 ASSERT(msp->ms_freemap[txg & TXG_MASK]->sm_space == 0);
1340
1341 /*
1342 * If there's a space_map_load() in progress, wait for it to complete
1343 * so that we have a consistent view of the in-core space map.
1344 */
1345 space_map_load_wait(sm);
1346
1347 /*
1348 * Move the frees from the defer_map to this map (if it's loaded).
1349 * Swap the freed_map and the defer_map -- this is safe to do
1350 * because we've just emptied out the defer_map.
1351 */
1352 space_map_vacate(*defer_map, sm->sm_loaded ? space_map_free : NULL, sm);
1353 ASSERT0((*defer_map)->sm_space);
1354 ASSERT0(avl_numnodes(&(*defer_map)->sm_root));
1355 space_map_swap(freed_map, defer_map);
1356
1357 *smo = *smosync;
1358
1359 msp->ms_deferspace += defer_delta;
1360 ASSERT3S(msp->ms_deferspace, >=, 0);
1361 ASSERT3S(msp->ms_deferspace, <=, sm->sm_size);
1362 if (msp->ms_deferspace != 0) {
1363 /*
1364 * Keep syncing this metaslab until all deferred frees
1365 * are back in circulation.
1366 */
1367 vdev_dirty(vd, VDD_METASLAB, msp, txg + 1);
1368 }
1369
1370 /*
1371 * If the map is loaded but no longer active, evict it as soon as all
1372 * future allocations have synced. (If we unloaded it now and then
1373 * loaded a moment later, the map wouldn't reflect those allocations.)
1374 */
1375 if (sm->sm_loaded && (msp->ms_weight & METASLAB_ACTIVE_MASK) == 0) {
1376 int evictable = 1;
1377
1378 for (int t = 1; t < TXG_CONCURRENT_STATES; t++)
1379 if (msp->ms_allocmap[(txg + t) & TXG_MASK]->sm_space)
1380 evictable = 0;
1381
1382 if (evictable && !metaslab_debug)
1383 space_map_unload(sm);
1384 }
1385
1386 metaslab_group_sort(mg, msp, metaslab_weight(msp));
1387
1388 mutex_exit(&msp->ms_lock);
1389 }
1390
1391 void
1392 metaslab_sync_reassess(metaslab_group_t *mg)
1393 {
1394 vdev_t *vd = mg->mg_vd;
1395 int64_t failures = mg->mg_alloc_failures;
1396
1397 metaslab_group_alloc_update(mg);
1398
1399 /*
1400 * Re-evaluate all metaslabs which have lower offsets than the
1401 * bonus area.
1402 */
1403 for (int m = 0; m < vd->vdev_ms_count; m++) {
1404 metaslab_t *msp = vd->vdev_ms[m];
1405
1406 if (msp->ms_map->sm_start > mg->mg_bonus_area)
1407 break;
1408
1409 mutex_enter(&msp->ms_lock);
1410 metaslab_group_sort(mg, msp, metaslab_weight(msp));
1411 mutex_exit(&msp->ms_lock);
1412 }
1413
1414 atomic_add_64(&mg->mg_alloc_failures, -failures);
1415
1416 /*
1417 * Prefetch the next potential metaslabs
1418 */
1419 metaslab_prefetch(mg);
1420 }
1421
1422 static uint64_t
1423 metaslab_distance(metaslab_t *msp, dva_t *dva)
1424 {
1425 uint64_t ms_shift = msp->ms_group->mg_vd->vdev_ms_shift;
1426 uint64_t offset = DVA_GET_OFFSET(dva) >> ms_shift;
1427 uint64_t start = msp->ms_map->sm_start >> ms_shift;
1428
1429 if (msp->ms_group->mg_vd->vdev_id != DVA_GET_VDEV(dva))
1430 return (1ULL << 63);
1431
1432 if (offset < start)
1433 return ((start - offset) << ms_shift);
1434 if (offset > start)
1435 return ((offset - start) << ms_shift);
1436 return (0);
1437 }
1438
1439 static uint64_t
1440 metaslab_group_alloc(metaslab_group_t *mg, uint64_t psize, uint64_t asize,
1441 uint64_t txg, uint64_t min_distance, dva_t *dva, int d, int flags)
1442 {
1443 spa_t *spa = mg->mg_vd->vdev_spa;
1444 metaslab_t *msp = NULL;
1445 uint64_t offset = -1ULL;
1446 avl_tree_t *t = &mg->mg_metaslab_tree;
1447 uint64_t activation_weight;
1459 for (;;) {
1460 boolean_t was_active;
1461
1462 mutex_enter(&mg->mg_lock);
1463 for (msp = avl_first(t); msp; msp = AVL_NEXT(t, msp)) {
1464 if (msp->ms_weight < asize) {
1465 spa_dbgmsg(spa, "%s: failed to meet weight "
1466 "requirement: vdev %llu, txg %llu, mg %p, "
1467 "msp %p, psize %llu, asize %llu, "
1468 "failures %llu, weight %llu",
1469 spa_name(spa), mg->mg_vd->vdev_id, txg,
1470 mg, msp, psize, asize,
1471 mg->mg_alloc_failures, msp->ms_weight);
1472 mutex_exit(&mg->mg_lock);
1473 return (-1ULL);
1474 }
1475
1476 /*
1477 * If the selected metaslab is condensing, skip it.
1478 */
1479 if (msp->ms_map->sm_condensing)
1480 continue;
1481
1482 was_active = msp->ms_weight & METASLAB_ACTIVE_MASK;
1483 if (activation_weight == METASLAB_WEIGHT_PRIMARY)
1484 break;
1485
1486 target_distance = min_distance +
1487 (msp->ms_smo.smo_alloc ? 0 : min_distance >> 1);
1488
1489 for (i = 0; i < d; i++)
1490 if (metaslab_distance(msp, &dva[i]) <
1491 target_distance)
1492 break;
1493 if (i == d)
1494 break;
1495 }
1496 mutex_exit(&mg->mg_lock);
1497 if (msp == NULL)
1498 return (-1ULL);
1499
1500 mutex_enter(&msp->ms_lock);
1501
1502 /*
1503 * If we've already reached the allowable number of failed
1504 * allocation attempts on this metaslab group then we
1505 * consider skipping it. We skip it only if we're allowed
1506 * to "fast" gang, the physical size is larger than
1507 * a gang block, and we're attempting to allocate from
1508 * the primary metaslab.
1509 */
1510 if (mg->mg_alloc_failures > zfs_mg_alloc_failures &&
1511 CAN_FASTGANG(flags) && psize > SPA_GANGBLOCKSIZE &&
1512 activation_weight == METASLAB_WEIGHT_PRIMARY) {
1513 spa_dbgmsg(spa, "%s: skipping metaslab group: "
1514 "vdev %llu, txg %llu, mg %p, psize %llu, "
1515 "asize %llu, failures %llu", spa_name(spa),
1516 mg->mg_vd->vdev_id, txg, mg, psize, asize,
1517 mg->mg_alloc_failures);
1518 mutex_exit(&msp->ms_lock);
1519 return (-1ULL);
1520 }
1521
1522 /*
1523 * Ensure that the metaslab we have selected is still
1524 * capable of handling our request. It's possible that
1525 * another thread may have changed the weight while we
1526 * were blocked on the metaslab lock.
1527 */
1528 if (msp->ms_weight < asize || (was_active &&
1529 !(msp->ms_weight & METASLAB_ACTIVE_MASK) &&
1530 activation_weight == METASLAB_WEIGHT_PRIMARY)) {
1531 mutex_exit(&msp->ms_lock);
1532 continue;
1533 }
1534
1535 if ((msp->ms_weight & METASLAB_WEIGHT_SECONDARY) &&
1536 activation_weight == METASLAB_WEIGHT_PRIMARY) {
1537 metaslab_passivate(msp,
1538 msp->ms_weight & ~METASLAB_ACTIVE_MASK);
1539 mutex_exit(&msp->ms_lock);
1540 continue;
1541 }
1542
1543 if (metaslab_activate(msp, activation_weight) != 0) {
1544 mutex_exit(&msp->ms_lock);
1545 continue;
1546 }
1547
1548 /*
1549 * If this metaslab is currently condensing then pick again as
1550 * we can't manipulate this metaslab until it's committed
1551 * to disk.
1552 */
1553 if (msp->ms_map->sm_condensing) {
1554 mutex_exit(&msp->ms_lock);
1555 continue;
1556 }
1557
1558 if ((offset = space_map_alloc(msp->ms_map, asize)) != -1ULL)
1559 break;
1560
1561 atomic_inc_64(&mg->mg_alloc_failures);
1562
1563 metaslab_passivate(msp, space_map_maxsize(msp->ms_map));
1564
1565 mutex_exit(&msp->ms_lock);
1566 }
1567
1568 if (msp->ms_allocmap[txg & TXG_MASK]->sm_space == 0)
1569 vdev_dirty(mg->mg_vd, VDD_METASLAB, msp, txg);
1570
1571 space_map_add(msp->ms_allocmap[txg & TXG_MASK], offset, asize);
1572
1573 mutex_exit(&msp->ms_lock);
1574
1575 return (offset);
1576 }
1577
1578 /*
1579 * Allocate a block for the specified i/o.
1580 */
1581 static int
1582 metaslab_alloc_dva(spa_t *spa, metaslab_class_t *mc, uint64_t psize,
1583 dva_t *dva, int d, dva_t *hintdva, uint64_t txg, int flags)
1584 {
1585 metaslab_group_t *mg, *rotor;
1586 vdev_t *vd;
1587 int dshift = 3;
1588 int all_zero;
1589 int zio_lock = B_FALSE;
1590 boolean_t allocatable;
1591 uint64_t offset = -1ULL;
1798
1799 if (txg > spa_freeze_txg(spa))
1800 return;
1801
1802 if ((vd = vdev_lookup_top(spa, vdev)) == NULL ||
1803 (offset >> vd->vdev_ms_shift) >= vd->vdev_ms_count) {
1804 cmn_err(CE_WARN, "metaslab_free_dva(): bad DVA %llu:%llu",
1805 (u_longlong_t)vdev, (u_longlong_t)offset);
1806 ASSERT(0);
1807 return;
1808 }
1809
1810 msp = vd->vdev_ms[offset >> vd->vdev_ms_shift];
1811
1812 if (DVA_GET_GANG(dva))
1813 size = vdev_psize_to_asize(vd, SPA_GANGBLOCKSIZE);
1814
1815 mutex_enter(&msp->ms_lock);
1816
1817 if (now) {
1818 space_map_remove(msp->ms_allocmap[txg & TXG_MASK],
1819 offset, size);
1820 space_map_free(msp->ms_map, offset, size);
1821 } else {
1822 if (msp->ms_freemap[txg & TXG_MASK]->sm_space == 0)
1823 vdev_dirty(vd, VDD_METASLAB, msp, txg);
1824 space_map_add(msp->ms_freemap[txg & TXG_MASK], offset, size);
1825 }
1826
1827 mutex_exit(&msp->ms_lock);
1828 }
1829
1830 /*
1831 * Intent log support: upon opening the pool after a crash, notify the SPA
1832 * of blocks that the intent log has allocated for immediate write, but
1833 * which are still considered free by the SPA because the last transaction
1834 * group didn't commit yet.
1835 */
1836 static int
1837 metaslab_claim_dva(spa_t *spa, const dva_t *dva, uint64_t txg)
1838 {
1839 uint64_t vdev = DVA_GET_VDEV(dva);
1840 uint64_t offset = DVA_GET_OFFSET(dva);
1841 uint64_t size = DVA_GET_ASIZE(dva);
1842 vdev_t *vd;
1843 metaslab_t *msp;
1844 int error = 0;
1845
1846 ASSERT(DVA_IS_VALID(dva));
1847
1848 if ((vd = vdev_lookup_top(spa, vdev)) == NULL ||
1849 (offset >> vd->vdev_ms_shift) >= vd->vdev_ms_count)
1850 return (SET_ERROR(ENXIO));
1851
1852 msp = vd->vdev_ms[offset >> vd->vdev_ms_shift];
1853
1854 if (DVA_GET_GANG(dva))
1855 size = vdev_psize_to_asize(vd, SPA_GANGBLOCKSIZE);
1856
1857 mutex_enter(&msp->ms_lock);
1858
1859 if ((txg != 0 && spa_writeable(spa)) || !msp->ms_map->sm_loaded)
1860 error = metaslab_activate(msp, METASLAB_WEIGHT_SECONDARY);
1861
1862 if (error == 0 && !space_map_contains(msp->ms_map, offset, size))
1863 error = SET_ERROR(ENOENT);
1864
1865 if (error || txg == 0) { /* txg == 0 indicates dry run */
1866 mutex_exit(&msp->ms_lock);
1867 return (error);
1868 }
1869
1870 space_map_claim(msp->ms_map, offset, size);
1871
1872 if (spa_writeable(spa)) { /* don't dirty if we're zdb(1M) */
1873 if (msp->ms_allocmap[txg & TXG_MASK]->sm_space == 0)
1874 vdev_dirty(vd, VDD_METASLAB, msp, txg);
1875 space_map_add(msp->ms_allocmap[txg & TXG_MASK], offset, size);
1876 }
1877
1878 mutex_exit(&msp->ms_lock);
1879
1880 return (0);
1881 }
1882
1883 int
1884 metaslab_alloc(spa_t *spa, metaslab_class_t *mc, uint64_t psize, blkptr_t *bp,
1885 int ndvas, uint64_t txg, blkptr_t *hintbp, int flags)
1886 {
1887 dva_t *dva = bp->blk_dva;
1888 dva_t *hintdva = hintbp->blk_dva;
1889 int error = 0;
1890
1891 ASSERT(bp->blk_birth == 0);
1892 ASSERT(BP_PHYSICAL_BIRTH(bp) == 0);
1893
1894 spa_config_enter(spa, SCL_ALLOC, FTAG, RW_READER);
1895
1896 if (mc->mc_rotor == NULL) { /* no vdevs in this class */
1897 spa_config_exit(spa, SCL_ALLOC, FTAG);
1898 return (SET_ERROR(ENOSPC));
1899 }
1900
1901 ASSERT(ndvas > 0 && ndvas <= spa_max_replication(spa));
1902 ASSERT(BP_GET_NDVAS(bp) == 0);
1903 ASSERT(hintbp == NULL || ndvas <= BP_GET_NDVAS(hintbp));
1904
1905 for (int d = 0; d < ndvas; d++) {
1906 error = metaslab_alloc_dva(spa, mc, psize, dva, d, hintdva,
1907 txg, flags);
1908 if (error) {
1909 for (d--; d >= 0; d--) {
1910 metaslab_free_dva(spa, &dva[d], txg, B_TRUE);
1911 bzero(&dva[d], sizeof (dva_t));
1912 }
1913 spa_config_exit(spa, SCL_ALLOC, FTAG);
1914 return (error);
1915 }
1916 }
1917 ASSERT(error == 0);
1918 ASSERT(BP_GET_NDVAS(bp) == ndvas);
1919
1920 spa_config_exit(spa, SCL_ALLOC, FTAG);
1921
1922 BP_SET_BIRTH(bp, txg, txg);
1923
1924 return (0);
1925 }
1926
1927 void
1928 metaslab_free(spa_t *spa, const blkptr_t *bp, uint64_t txg, boolean_t now)
1955 * First do a dry run to make sure all DVAs are claimable,
1956 * so we don't have to unwind from partial failures below.
1957 */
1958 if ((error = metaslab_claim(spa, bp, 0)) != 0)
1959 return (error);
1960 }
1961
1962 spa_config_enter(spa, SCL_ALLOC, FTAG, RW_READER);
1963
1964 for (int d = 0; d < ndvas; d++)
1965 if ((error = metaslab_claim_dva(spa, &dva[d], txg)) != 0)
1966 break;
1967
1968 spa_config_exit(spa, SCL_ALLOC, FTAG);
1969
1970 ASSERT(error == 0 || txg == 0);
1971
1972 return (error);
1973 }
1974
1975 static void
1976 checkmap(space_map_t *sm, uint64_t off, uint64_t size)
1977 {
1978 space_seg_t *ss;
1979 avl_index_t where;
1980
1981 mutex_enter(sm->sm_lock);
1982 ss = space_map_find(sm, off, size, &where);
1983 if (ss != NULL)
1984 panic("freeing free block; ss=%p", (void *)ss);
1985 mutex_exit(sm->sm_lock);
1986 }
1987
1988 void
1989 metaslab_check_free(spa_t *spa, const blkptr_t *bp)
1990 {
1991 if ((zfs_flags & ZFS_DEBUG_ZIO_FREE) == 0)
1992 return;
1993
1994 spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER);
1995 for (int i = 0; i < BP_GET_NDVAS(bp); i++) {
1996 uint64_t vdid = DVA_GET_VDEV(&bp->blk_dva[i]);
1997 vdev_t *vd = vdev_lookup_top(spa, vdid);
1998 uint64_t off = DVA_GET_OFFSET(&bp->blk_dva[i]);
1999 uint64_t size = DVA_GET_ASIZE(&bp->blk_dva[i]);
2000 metaslab_t *ms = vd->vdev_ms[off >> vd->vdev_ms_shift];
2001
2002 if (ms->ms_map->sm_loaded)
2003 checkmap(ms->ms_map, off, size);
2004
2005 for (int j = 0; j < TXG_SIZE; j++)
2006 checkmap(ms->ms_freemap[j], off, size);
2007 for (int j = 0; j < TXG_DEFER_SIZE; j++)
2008 checkmap(ms->ms_defermap[j], off, size);
2009 }
2010 spa_config_exit(spa, SCL_VDEV, FTAG);
2011 }
|
14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 * If applicable, add the following below this CDDL HEADER, with the
16 * fields enclosed by brackets "[]" replaced with your own identifying
17 * information: Portions Copyright [yyyy] [name of copyright owner]
18 *
19 * CDDL HEADER END
20 */
21 /*
22 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
23 * Copyright (c) 2013 by Delphix. All rights reserved.
24 * Copyright (c) 2013 by Saso Kiselkov. All rights reserved.
25 */
26
27 #include <sys/zfs_context.h>
28 #include <sys/dmu.h>
29 #include <sys/dmu_tx.h>
30 #include <sys/space_map.h>
31 #include <sys/metaslab_impl.h>
32 #include <sys/vdev_impl.h>
33 #include <sys/zio.h>
34 #include <sys/spa_impl.h>
35
36 /*
37 * Allow allocations to switch to gang blocks quickly. We do this to
38 * avoid having to load lots of space_maps in a given txg. There are,
39 * however, some cases where we want to avoid "fast" ganging and instead
40 * we want to do an exhaustive search of all metaslabs on this device.
41 * Currently we don't allow any gang, zil, or dump device related allocations
42 * to "fast" gang.
43 */
44 #define CAN_FASTGANG(flags) \
45 (!((flags) & (METASLAB_GANG_CHILD | METASLAB_GANG_HEADER | \
46 METASLAB_GANG_AVOID)))
47
48 #define METASLAB_WEIGHT_PRIMARY (1ULL << 63)
49 #define METASLAB_WEIGHT_SECONDARY (1ULL << 62)
50 #define METASLAB_ACTIVE_MASK \
51 (METASLAB_WEIGHT_PRIMARY | METASLAB_WEIGHT_SECONDARY)
52
53 uint64_t metaslab_aliquot = 512ULL << 10;
54 uint64_t metaslab_gang_bang = SPA_MAXBLOCKSIZE + 1; /* force gang blocks */
55
56 /*
57 * The in-core space map representation is more compact than its on-disk form.
58 * The zfs_condense_pct determines how much more compact the in-core
59 * space_map representation must be before we compact it on-disk.
60 * Values should be greater than or equal to 100.
61 */
62 int zfs_condense_pct = 200;
63
64 /*
65 * This value defines the number of allowed allocation failures per vdev.
66 * If a device reaches this threshold in a given txg then we consider skipping
67 * allocations on that device. The value of zfs_mg_alloc_failures is computed
68 * in zio_init() unless it has been overridden in /etc/system.
69 */
70 int zfs_mg_alloc_failures = 0;
71
72 /*
73 * The zfs_mg_noalloc_threshold defines which metaslab groups should
74 * be eligible for allocation. The value is defined as a percentage of
75 * a free space. Metaslab groups that have more free space than
76 * zfs_mg_noalloc_threshold are always eligible for allocations. Once
77 * a metaslab group's free space is less than or equal to the
78 * zfs_mg_noalloc_threshold the allocator will avoid allocating to that
79 * group unless all groups in the pool have reached zfs_mg_noalloc_threshold.
80 * Once all groups in the pool reach zfs_mg_noalloc_threshold then all
81 * groups are allowed to accept allocations. Gang blocks are always
82 * eligible to allocate on any metaslab group. The default value of 0 means
83 * no metaslab group will be excluded based on this criterion.
84 */
85 int zfs_mg_noalloc_threshold = 0;
86
87 /*
88 * When set will load all metaslabs when pool is first opened.
89 */
90 int metaslab_debug_load = 0;
91
92 /*
93 * When set will prevent metaslabs from being unloaded.
94 */
95 int metaslab_debug_unload = 0;
96
97 /*
98 * Minimum size which forces the dynamic allocator to change
99 * it's allocation strategy. Once the space map cannot satisfy
100 * an allocation of this size then it switches to using more
101 * aggressive strategy (i.e search by size rather than offset).
102 */
103 uint64_t metaslab_df_alloc_threshold = SPA_MAXBLOCKSIZE;
104
105 /*
106 * The minimum free space, in percent, which must be available
107 * in a space map to continue allocations in a first-fit fashion.
108 * Once the space_map's free space drops below this level we dynamically
109 * switch to using best-fit allocations.
110 */
111 int metaslab_df_free_pct = 4;
112
113 /*
114 * A metaslab is considered "free" if it contains a contiguous
115 * segment which is greater than metaslab_min_alloc_size.
116 */
117 uint64_t metaslab_min_alloc_size = DMU_MAX_ACCESS;
118
119 /*
120 * Percentage of all cpus that can be used by the metaslab taskq.
121 */
122 int metaslab_load_pct = 50;
123
124 /*
125 * Determines how many txgs a metaslab may remain loaded without having any
126 * allocations from it. As long as a metaslab continues to be used we will
127 * keep it loaded.
128 */
129 int metaslab_unload_delay = TXG_SIZE * 2;
130
131 /*
132 * Should we be willing to write data to degraded vdevs?
133 */
134 boolean_t zfs_write_to_degraded = B_FALSE;
135
136 /*
137 * Max number of metaslabs per group to preload.
138 */
139 int metaslab_preload_limit = SPA_DVAS_PER_BP;
140
141 /*
142 * Enable/disable preloading of metaslab.
143 */
144 boolean_t metaslab_preload_enabled = B_TRUE;
145
146 /*
147 * Enable/disable additional weight factor for each metaslab.
148 */
149 boolean_t metaslab_weight_factor_enable = B_FALSE;
150
151
152 /*
153 * ==========================================================================
154 * Metaslab classes
155 * ==========================================================================
156 */
157 metaslab_class_t *
158 metaslab_class_create(spa_t *spa, metaslab_ops_t *ops)
159 {
160 metaslab_class_t *mc;
161
162 mc = kmem_zalloc(sizeof (metaslab_class_t), KM_SLEEP);
163
164 mc->mc_spa = spa;
165 mc->mc_rotor = NULL;
166 mc->mc_ops = ops;
167
168 return (mc);
169 }
170
171 void
172 metaslab_class_destroy(metaslab_class_t *mc)
173 {
174 ASSERT(mc->mc_rotor == NULL);
175 ASSERT(mc->mc_alloc == 0);
176 ASSERT(mc->mc_deferred == 0);
177 ASSERT(mc->mc_space == 0);
178 ASSERT(mc->mc_dspace == 0);
242
243 /*
244 * ==========================================================================
245 * Metaslab groups
246 * ==========================================================================
247 */
248 static int
249 metaslab_compare(const void *x1, const void *x2)
250 {
251 const metaslab_t *m1 = x1;
252 const metaslab_t *m2 = x2;
253
254 if (m1->ms_weight < m2->ms_weight)
255 return (1);
256 if (m1->ms_weight > m2->ms_weight)
257 return (-1);
258
259 /*
260 * If the weights are identical, use the offset to force uniqueness.
261 */
262 if (m1->ms_start < m2->ms_start)
263 return (-1);
264 if (m1->ms_start > m2->ms_start)
265 return (1);
266
267 ASSERT3P(m1, ==, m2);
268
269 return (0);
270 }
271
272 /*
273 * Update the allocatable flag and the metaslab group's capacity.
274 * The allocatable flag is set to true if the capacity is below
275 * the zfs_mg_noalloc_threshold. If a metaslab group transitions
276 * from allocatable to non-allocatable or vice versa then the metaslab
277 * group's class is updated to reflect the transition.
278 */
279 static void
280 metaslab_group_alloc_update(metaslab_group_t *mg)
281 {
282 vdev_t *vd = mg->mg_vd;
283 metaslab_class_t *mc = mg->mg_class;
284 vdev_stat_t *vs = &vd->vdev_stat;
312 if (was_allocatable && !mg->mg_allocatable)
313 mc->mc_alloc_groups--;
314 else if (!was_allocatable && mg->mg_allocatable)
315 mc->mc_alloc_groups++;
316 mutex_exit(&mg->mg_lock);
317 }
318
319 metaslab_group_t *
320 metaslab_group_create(metaslab_class_t *mc, vdev_t *vd)
321 {
322 metaslab_group_t *mg;
323
324 mg = kmem_zalloc(sizeof (metaslab_group_t), KM_SLEEP);
325 mutex_init(&mg->mg_lock, NULL, MUTEX_DEFAULT, NULL);
326 avl_create(&mg->mg_metaslab_tree, metaslab_compare,
327 sizeof (metaslab_t), offsetof(struct metaslab, ms_group_node));
328 mg->mg_vd = vd;
329 mg->mg_class = mc;
330 mg->mg_activation_count = 0;
331
332 mg->mg_taskq = taskq_create("metaslab_group_tasksq", metaslab_load_pct,
333 minclsyspri, 10, INT_MAX, TASKQ_THREADS_CPU_PCT);
334
335 return (mg);
336 }
337
338 void
339 metaslab_group_destroy(metaslab_group_t *mg)
340 {
341 ASSERT(mg->mg_prev == NULL);
342 ASSERT(mg->mg_next == NULL);
343 /*
344 * We may have gone below zero with the activation count
345 * either because we never activated in the first place or
346 * because we're done, and possibly removing the vdev.
347 */
348 ASSERT(mg->mg_activation_count <= 0);
349
350 avl_destroy(&mg->mg_metaslab_tree);
351 mutex_destroy(&mg->mg_lock);
352 kmem_free(mg, sizeof (metaslab_group_t));
353 }
354
383 }
384 mc->mc_rotor = mg;
385 }
386
387 void
388 metaslab_group_passivate(metaslab_group_t *mg)
389 {
390 metaslab_class_t *mc = mg->mg_class;
391 metaslab_group_t *mgprev, *mgnext;
392
393 ASSERT(spa_config_held(mc->mc_spa, SCL_ALLOC, RW_WRITER));
394
395 if (--mg->mg_activation_count != 0) {
396 ASSERT(mc->mc_rotor != mg);
397 ASSERT(mg->mg_prev == NULL);
398 ASSERT(mg->mg_next == NULL);
399 ASSERT(mg->mg_activation_count < 0);
400 return;
401 }
402
403 taskq_wait(mg->mg_taskq);
404
405 mgprev = mg->mg_prev;
406 mgnext = mg->mg_next;
407
408 if (mg == mgnext) {
409 mc->mc_rotor = NULL;
410 } else {
411 mc->mc_rotor = mgnext;
412 mgprev->mg_next = mgnext;
413 mgnext->mg_prev = mgprev;
414 }
415
416 mg->mg_prev = NULL;
417 mg->mg_next = NULL;
418 }
419
420 static void
421 metaslab_group_add(metaslab_group_t *mg, metaslab_t *msp)
422 {
423 mutex_enter(&mg->mg_lock);
424 ASSERT(msp->ms_group == NULL);
464 */
465 static boolean_t
466 metaslab_group_allocatable(metaslab_group_t *mg)
467 {
468 vdev_t *vd = mg->mg_vd;
469 spa_t *spa = vd->vdev_spa;
470 metaslab_class_t *mc = mg->mg_class;
471
472 /*
473 * A metaslab group is considered allocatable if its free capacity
474 * is greater than the set value of zfs_mg_noalloc_threshold, it's
475 * associated with a slog, or there are no other metaslab groups
476 * with free capacity greater than zfs_mg_noalloc_threshold.
477 */
478 return (mg->mg_free_capacity > zfs_mg_noalloc_threshold ||
479 mc != spa_normal_class(spa) || mc->mc_alloc_groups == 0);
480 }
481
482 /*
483 * ==========================================================================
484 * Range tree callbacks
485 * ==========================================================================
486 */
487
488 /*
489 * Comparison function for the private size-ordered tree. Tree is sorted
490 * by size, larger sizes at the end of the tree.
491 */
492 static int
493 metaslab_rangesize_compare(const void *x1, const void *x2)
494 {
495 const range_seg_t *r1 = x1;
496 const range_seg_t *r2 = x2;
497 uint64_t rs_size1 = r1->rs_end - r1->rs_start;
498 uint64_t rs_size2 = r2->rs_end - r2->rs_start;
499
500 if (rs_size1 < rs_size2)
501 return (-1);
502 if (rs_size1 > rs_size2)
503 return (1);
504
505 if (r1->rs_start < r2->rs_start)
506 return (-1);
507
508 if (r1->rs_start > r2->rs_start)
509 return (1);
510
511 return (0);
512 }
513
514 /*
515 * Create any block allocator specific components. The current allocators
516 * rely on using both a size-ordered range_tree_t and an array of uint64_t's.
517 */
518 static void
519 metaslab_rt_create(range_tree_t *rt, void *arg)
520 {
521 metaslab_t *msp = arg;
522
523 ASSERT3P(rt->rt_arg, ==, msp);
524 ASSERT(msp->ms_tree == NULL);
525
526 avl_create(&msp->ms_size_tree, metaslab_rangesize_compare,
527 sizeof (range_seg_t), offsetof(range_seg_t, rs_pp_node));
528 }
529
530 /*
531 * Destroy the block allocator specific components.
532 */
533 static void
534 metaslab_rt_destroy(range_tree_t *rt, void *arg)
535 {
536 metaslab_t *msp = arg;
537
538 ASSERT3P(rt->rt_arg, ==, msp);
539 ASSERT3P(msp->ms_tree, ==, rt);
540 ASSERT0(avl_numnodes(&msp->ms_size_tree));
541
542 avl_destroy(&msp->ms_size_tree);
543 }
544
545 static void
546 metaslab_rt_add(range_tree_t *rt, range_seg_t *rs, void *arg)
547 {
548 metaslab_t *msp = arg;
549
550 ASSERT3P(rt->rt_arg, ==, msp);
551 ASSERT3P(msp->ms_tree, ==, rt);
552 VERIFY(!msp->ms_condensing);
553 avl_add(&msp->ms_size_tree, rs);
554 }
555
556 static void
557 metaslab_rt_remove(range_tree_t *rt, range_seg_t *rs, void *arg)
558 {
559 metaslab_t *msp = arg;
560
561 ASSERT3P(rt->rt_arg, ==, msp);
562 ASSERT3P(msp->ms_tree, ==, rt);
563 VERIFY(!msp->ms_condensing);
564 avl_remove(&msp->ms_size_tree, rs);
565 }
566
567 static void
568 metaslab_rt_vacate(range_tree_t *rt, void *arg)
569 {
570 metaslab_t *msp = arg;
571
572 ASSERT3P(rt->rt_arg, ==, msp);
573 ASSERT3P(msp->ms_tree, ==, rt);
574
575 /*
576 * Normally one would walk the tree freeing nodes along the way.
577 * Since the nodes are shared with the range trees we can avoid
578 * walking all nodes and just reinitialize the avl tree. The nodes
579 * will be freed by the range tree, so we don't want to free them here.
580 */
581 avl_create(&msp->ms_size_tree, metaslab_rangesize_compare,
582 sizeof (range_seg_t), offsetof(range_seg_t, rs_pp_node));
583 }
584
585 static range_tree_ops_t metaslab_rt_ops = {
586 metaslab_rt_create,
587 metaslab_rt_destroy,
588 metaslab_rt_add,
589 metaslab_rt_remove,
590 metaslab_rt_vacate
591 };
592
593 /*
594 * ==========================================================================
595 * Metaslab block operations
596 * ==========================================================================
597 */
598
599 /*
600 * Return the maximum contiguous segment within the metaslab.
601 */
602 uint64_t
603 metaslab_block_maxsize(metaslab_t *msp)
604 {
605 avl_tree_t *t = &msp->ms_size_tree;
606 range_seg_t *rs;
607
608 if (t == NULL || (rs = avl_last(t)) == NULL)
609 return (0ULL);
610
611 return (rs->rs_end - rs->rs_start);
612 }
613
614 uint64_t
615 metaslab_block_alloc(metaslab_t *msp, uint64_t size)
616 {
617 uint64_t start;
618 range_tree_t *rt = msp->ms_tree;
619
620 VERIFY(!msp->ms_condensing);
621
622 start = msp->ms_ops->msop_alloc(msp, size);
623 if (start != -1ULL) {
624 vdev_t *vd = msp->ms_group->mg_vd;
625
626 VERIFY0(P2PHASE(start, 1ULL << vd->vdev_ashift));
627 VERIFY0(P2PHASE(size, 1ULL << vd->vdev_ashift));
628 VERIFY3U(range_tree_space(rt) - size, <=, msp->ms_size);
629 range_tree_remove(rt, start, size);
630 }
631 return (start);
632 }
633
634 /*
635 * ==========================================================================
636 * Common allocator routines
637 * ==========================================================================
638 */
639
640 /*
641 * This is a helper function that can be used by the allocator to find
642 * a suitable block to allocate. This will search the specified AVL
643 * tree looking for a block that matches the specified criteria.
644 */
645 static uint64_t
646 metaslab_block_picker(avl_tree_t *t, uint64_t *cursor, uint64_t size,
647 uint64_t align)
648 {
649 range_seg_t *rs, rsearch;
650 avl_index_t where;
651
652 rsearch.rs_start = *cursor;
653 rsearch.rs_end = *cursor + size;
654
655 rs = avl_find(t, &rsearch, &where);
656 if (rs == NULL)
657 rs = avl_nearest(t, where, AVL_AFTER);
658
659 while (rs != NULL) {
660 uint64_t offset = P2ROUNDUP(rs->rs_start, align);
661
662 if (offset + size <= rs->rs_end) {
663 *cursor = offset + size;
664 return (offset);
665 }
666 rs = AVL_NEXT(t, rs);
667 }
668
669 /*
670 * If we know we've searched the whole map (*cursor == 0), give up.
671 * Otherwise, reset the cursor to the beginning and try again.
672 */
673 if (*cursor == 0)
674 return (-1ULL);
675
676 *cursor = 0;
677 return (metaslab_block_picker(t, cursor, size, align));
678 }
679
680 /*
681 * ==========================================================================
682 * The first-fit block allocator
683 * ==========================================================================
684 */
685 static uint64_t
686 metaslab_ff_alloc(metaslab_t *msp, uint64_t size)
687 {
688 /*
689 * Find the largest power of 2 block size that evenly divides the
690 * requested size. This is used to try to allocate blocks with similar
691 * alignment from the same area of the metaslab (i.e. same cursor
692 * bucket) but it does not guarantee that other allocations sizes
693 * may exist in the same region.
694 */
695 uint64_t align = size & -size;
696 uint64_t *cursor = &msp->ms_lbas[highbit(align) - 1];
697 avl_tree_t *t = &msp->ms_tree->rt_root;
698
699 return (metaslab_block_picker(t, cursor, size, align));
700 }
701
702 /* ARGSUSED */
703 static boolean_t
704 metaslab_ff_fragmented(metaslab_t *msp)
705 {
706 return (B_TRUE);
707 }
708
709 static metaslab_ops_t metaslab_ff_ops = {
710 metaslab_ff_alloc,
711 metaslab_ff_fragmented
712 };
713
714 /*
715 * ==========================================================================
716 * Dynamic block allocator -
717 * Uses the first fit allocation scheme until space get low and then
718 * adjusts to a best fit allocation method. Uses metaslab_df_alloc_threshold
719 * and metaslab_df_free_pct to determine when to switch the allocation scheme.
720 * ==========================================================================
721 */
722 static uint64_t
723 metaslab_df_alloc(metaslab_t *msp, uint64_t size)
724 {
725 /*
726 * Find the largest power of 2 block size that evenly divides the
727 * requested size. This is used to try to allocate blocks with similar
728 * alignment from the same area of the metaslab (i.e. same cursor
729 * bucket) but it does not guarantee that other allocations sizes
730 * may exist in the same region.
731 */
732 uint64_t align = size & -size;
733 uint64_t *cursor = &msp->ms_lbas[highbit(align) - 1];
734 range_tree_t *rt = msp->ms_tree;
735 avl_tree_t *t = &rt->rt_root;
736 uint64_t max_size = metaslab_block_maxsize(msp);
737 int free_pct = range_tree_space(rt) * 100 / msp->ms_size;
738
739 ASSERT(MUTEX_HELD(&msp->ms_lock));
740 ASSERT3U(avl_numnodes(t), ==, avl_numnodes(&msp->ms_size_tree));
741
742 if (max_size < size)
743 return (-1ULL);
744
745 /*
746 * If we're running low on space switch to using the size
747 * sorted AVL tree (best-fit).
748 */
749 if (max_size < metaslab_df_alloc_threshold ||
750 free_pct < metaslab_df_free_pct) {
751 t = &msp->ms_size_tree;
752 *cursor = 0;
753 }
754
755 return (metaslab_block_picker(t, cursor, size, 1ULL));
756 }
757
758 static boolean_t
759 metaslab_df_fragmented(metaslab_t *msp)
760 {
761 range_tree_t *rt = msp->ms_tree;
762 uint64_t max_size = metaslab_block_maxsize(msp);
763 int free_pct = range_tree_space(rt) * 100 / msp->ms_size;
764
765 if (max_size >= metaslab_df_alloc_threshold &&
766 free_pct >= metaslab_df_free_pct)
767 return (B_FALSE);
768
769 return (B_TRUE);
770 }
771
772 static metaslab_ops_t metaslab_df_ops = {
773 metaslab_df_alloc,
774 metaslab_df_fragmented
775 };
776
777 /*
778 * ==========================================================================
779 * Cursor fit block allocator -
780 * Select the largest region in the metaslab, set the cursor to the beginning
781 * of the range and the cursor_end to the end of the range. As allocations
782 * are made advance the cursor. Continue allocating from the cursor until
783 * the range is exhausted and then find a new range.
784 * ==========================================================================
785 */
786 static uint64_t
787 metaslab_cf_alloc(metaslab_t *msp, uint64_t size)
788 {
789 range_tree_t *rt = msp->ms_tree;
790 avl_tree_t *t = &msp->ms_size_tree;
791 uint64_t *cursor = &msp->ms_lbas[0];
792 uint64_t *cursor_end = &msp->ms_lbas[1];
793 uint64_t offset = 0;
794
795 ASSERT(MUTEX_HELD(&msp->ms_lock));
796 ASSERT3U(avl_numnodes(t), ==, avl_numnodes(&rt->rt_root));
797
798 ASSERT3U(*cursor_end, >=, *cursor);
799
800 if ((*cursor + size) > *cursor_end) {
801 range_seg_t *rs;
802
803 rs = avl_last(&msp->ms_size_tree);
804 if (rs == NULL || (rs->rs_end - rs->rs_start) < size)
805 return (-1ULL);
806
807 *cursor = rs->rs_start;
808 *cursor_end = rs->rs_end;
809 }
810
811 offset = *cursor;
812 *cursor += size;
813
814 return (offset);
815 }
816
817 static boolean_t
818 metaslab_cf_fragmented(metaslab_t *msp)
819 {
820 return (metaslab_block_maxsize(msp) < metaslab_min_alloc_size);
821 }
822
823 static metaslab_ops_t metaslab_cf_ops = {
824 metaslab_cf_alloc,
825 metaslab_cf_fragmented
826 };
827
828 /*
829 * ==========================================================================
830 * New dynamic fit allocator -
831 * Select a region that is large enough to allocate 2^metaslab_ndf_clump_shift
832 * contiguous blocks. If no region is found then just use the largest segment
833 * that remains.
834 * ==========================================================================
835 */
836
837 /*
838 * Determines desired number of contiguous blocks (2^metaslab_ndf_clump_shift)
839 * to request from the allocator.
840 */
841 uint64_t metaslab_ndf_clump_shift = 4;
842
843 static uint64_t
844 metaslab_ndf_alloc(metaslab_t *msp, uint64_t size)
845 {
846 avl_tree_t *t = &msp->ms_tree->rt_root;
847 avl_index_t where;
848 range_seg_t *rs, rsearch;
849 uint64_t hbit = highbit(size);
850 uint64_t *cursor = &msp->ms_lbas[hbit - 1];
851 uint64_t max_size = metaslab_block_maxsize(msp);
852
853 ASSERT(MUTEX_HELD(&msp->ms_lock));
854 ASSERT3U(avl_numnodes(t), ==, avl_numnodes(&msp->ms_size_tree));
855
856 if (max_size < size)
857 return (-1ULL);
858
859 rsearch.rs_start = *cursor;
860 rsearch.rs_end = *cursor + size;
861
862 rs = avl_find(t, &rsearch, &where);
863 if (rs == NULL || (rs->rs_end - rs->rs_start) < size) {
864 t = &msp->ms_size_tree;
865
866 rsearch.rs_start = 0;
867 rsearch.rs_end = MIN(max_size,
868 1ULL << (hbit + metaslab_ndf_clump_shift));
869 rs = avl_find(t, &rsearch, &where);
870 if (rs == NULL)
871 rs = avl_nearest(t, where, AVL_AFTER);
872 ASSERT(rs != NULL);
873 }
874
875 if ((rs->rs_end - rs->rs_start) >= size) {
876 *cursor = rs->rs_start + size;
877 return (rs->rs_start);
878 }
879 return (-1ULL);
880 }
881
882 static boolean_t
883 metaslab_ndf_fragmented(metaslab_t *msp)
884 {
885 return (metaslab_block_maxsize(msp) <=
886 (metaslab_min_alloc_size << metaslab_ndf_clump_shift));
887 }
888
889 static metaslab_ops_t metaslab_ndf_ops = {
890 metaslab_ndf_alloc,
891 metaslab_ndf_fragmented
892 };
893
894 metaslab_ops_t *zfs_metaslab_ops = &metaslab_df_ops;
895
896 /*
897 * ==========================================================================
898 * Metaslabs
899 * ==========================================================================
900 */
901
902 /*
903 * Wait for any in-progress metaslab loads to complete.
904 */
905 void
906 metaslab_load_wait(metaslab_t *msp)
907 {
908 ASSERT(MUTEX_HELD(&msp->ms_lock));
909
910 while (msp->ms_loading) {
911 ASSERT(!msp->ms_loaded);
912 cv_wait(&msp->ms_load_cv, &msp->ms_lock);
913 }
914 }
915
916 int
917 metaslab_load(metaslab_t *msp)
918 {
919 int error = 0;
920
921 ASSERT(MUTEX_HELD(&msp->ms_lock));
922 ASSERT(!msp->ms_loaded);
923 ASSERT(!msp->ms_loading);
924
925 msp->ms_loading = B_TRUE;
926
927 /*
928 * If the space map has not been allocated yet, then treat
929 * all the space in the metaslab as free and add it to the
930 * ms_tree.
931 */
932 if (msp->ms_sm != NULL)
933 error = space_map_load(msp->ms_sm, msp->ms_tree, SM_FREE);
934 else
935 range_tree_add(msp->ms_tree, msp->ms_start, msp->ms_size);
936
937 msp->ms_loaded = (error == 0);
938 msp->ms_loading = B_FALSE;
939
940 if (msp->ms_loaded) {
941 for (int t = 0; t < TXG_DEFER_SIZE; t++) {
942 range_tree_walk(msp->ms_defertree[t],
943 range_tree_remove, msp->ms_tree);
944 }
945 }
946 cv_broadcast(&msp->ms_load_cv);
947 return (error);
948 }
949
950 void
951 metaslab_unload(metaslab_t *msp)
952 {
953 ASSERT(MUTEX_HELD(&msp->ms_lock));
954 range_tree_vacate(msp->ms_tree, NULL, NULL);
955 msp->ms_loaded = B_FALSE;
956 msp->ms_weight &= ~METASLAB_ACTIVE_MASK;
957 }
958
959 metaslab_t *
960 metaslab_init(metaslab_group_t *mg, uint64_t id, uint64_t object, uint64_t txg)
961 {
962 vdev_t *vd = mg->mg_vd;
963 objset_t *mos = vd->vdev_spa->spa_meta_objset;
964 metaslab_t *msp;
965
966 msp = kmem_zalloc(sizeof (metaslab_t), KM_SLEEP);
967 mutex_init(&msp->ms_lock, NULL, MUTEX_DEFAULT, NULL);
968 cv_init(&msp->ms_load_cv, NULL, CV_DEFAULT, NULL);
969 msp->ms_id = id;
970 msp->ms_start = id << vd->vdev_ms_shift;
971 msp->ms_size = 1ULL << vd->vdev_ms_shift;
972
973 /*
974 * We only open space map objects that already exist. All others
975 * will be opened when we finally allocate an object for it.
976 */
977 if (object != 0) {
978 VERIFY0(space_map_open(&msp->ms_sm, mos, object, msp->ms_start,
979 msp->ms_size, vd->vdev_ashift, &msp->ms_lock));
980 ASSERT(msp->ms_sm != NULL);
981 }
982
983 /*
984 * We create the main range tree here, but we don't create the
985 * alloctree and freetree until metaslab_sync_done(). This serves
986 * two purposes: it allows metaslab_sync_done() to detect the
987 * addition of new space; and for debugging, it ensures that we'd
988 * data fault on any attempt to use this metaslab before it's ready.
989 */
990 msp->ms_tree = range_tree_create(&metaslab_rt_ops, msp, &msp->ms_lock);
991 metaslab_group_add(mg, msp);
992
993 msp->ms_ops = mg->mg_class->mc_ops;
994
995 /*
996 * If we're opening an existing pool (txg == 0) or creating
997 * a new one (txg == TXG_INITIAL), all space is available now.
998 * If we're adding space to an existing pool, the new space
999 * does not become available until after this txg has synced.
1000 */
1001 if (txg <= TXG_INITIAL)
1002 metaslab_sync_done(msp, 0);
1003
1004 /*
1005 * If metaslab_debug_load is set and we're initializing a metaslab
1006 * that has an allocated space_map object then load the its space
1007 * map so that can verify frees.
1008 */
1009 if (metaslab_debug_load && msp->ms_sm != NULL) {
1010 mutex_enter(&msp->ms_lock);
1011 VERIFY0(metaslab_load(msp));
1012 mutex_exit(&msp->ms_lock);
1013 }
1014
1015 if (txg != 0) {
1016 vdev_dirty(vd, 0, NULL, txg);
1017 vdev_dirty(vd, VDD_METASLAB, msp, txg);
1018 }
1019
1020 return (msp);
1021 }
1022
1023 void
1024 metaslab_fini(metaslab_t *msp)
1025 {
1026 metaslab_group_t *mg = msp->ms_group;
1027
1028 metaslab_group_remove(mg, msp);
1029
1030 mutex_enter(&msp->ms_lock);
1031
1032 VERIFY(msp->ms_group == NULL);
1033 vdev_space_update(mg->mg_vd, -space_map_allocated(msp->ms_sm),
1034 0, -msp->ms_size);
1035 space_map_close(msp->ms_sm);
1036
1037 metaslab_unload(msp);
1038 range_tree_destroy(msp->ms_tree);
1039
1040 for (int t = 0; t < TXG_SIZE; t++) {
1041 range_tree_destroy(msp->ms_alloctree[t]);
1042 range_tree_destroy(msp->ms_freetree[t]);
1043 }
1044
1045 for (int t = 0; t < TXG_DEFER_SIZE; t++) {
1046 range_tree_destroy(msp->ms_defertree[t]);
1047 }
1048
1049 ASSERT0(msp->ms_deferspace);
1050
1051 mutex_exit(&msp->ms_lock);
1052 cv_destroy(&msp->ms_load_cv);
1053 mutex_destroy(&msp->ms_lock);
1054
1055 kmem_free(msp, sizeof (metaslab_t));
1056 }
1057
1058 /*
1059 * Apply a weighting factor based on the histogram information for this
1060 * metaslab. The current weighting factor is somewhat arbitrary and requires
1061 * additional investigation. The implementation provides a measure of
1062 * "weighted" free space and gives a higher weighting for larger contiguous
1063 * regions. The weighting factor is determined by counting the number of
1064 * sm_shift sectors that exist in each region represented by the histogram.
1065 * That value is then multiplied by the power of 2 exponent and the sm_shift
1066 * value.
1067 *
1068 * For example, assume the 2^21 histogram bucket has 4 2MB regions and the
1069 * metaslab has an sm_shift value of 9 (512B):
1070 *
1071 * 1) calculate the number of sm_shift sectors in the region:
1072 * 2^21 / 2^9 = 2^12 = 4096 * 4 (number of regions) = 16384
1073 * 2) multiply by the power of 2 exponent and the sm_shift value:
1074 * 16384 * 21 * 9 = 3096576
1075 * This value will be added to the weighting of the metaslab.
1076 */
1077 static uint64_t
1078 metaslab_weight_factor(metaslab_t *msp)
1079 {
1080 uint64_t factor = 0;
1081 uint64_t sectors;
1082 int i;
1083
1084 /*
1085 * A null space map means that the entire metaslab is free,
1086 * calculate a weight factor that spans the entire size of the
1087 * metaslab.
1088 */
1089 if (msp->ms_sm == NULL) {
1090 vdev_t *vd = msp->ms_group->mg_vd;
1091
1092 i = highbit(msp->ms_size) - 1;
1093 sectors = msp->ms_size >> vd->vdev_ashift;
1094 return (sectors * i * vd->vdev_ashift);
1095 }
1096
1097 if (msp->ms_sm->sm_dbuf->db_size != sizeof (space_map_phys_t))
1098 return (0);
1099
1100 for (i = 0; i < SPACE_MAP_HISTOGRAM_SIZE(msp->ms_sm); i++) {
1101 if (msp->ms_sm->sm_phys->smp_histogram[i] == 0)
1102 continue;
1103
1104 /*
1105 * Determine the number of sm_shift sectors in the region
1106 * indicated by the histogram. For example, given an
1107 * sm_shift value of 9 (512 bytes) and i = 4 then we know
1108 * that we're looking at an 8K region in the histogram
1109 * (i.e. 9 + 4 = 13, 2^13 = 8192). To figure out the
1110 * number of sm_shift sectors (512 bytes in this example),
1111 * we would take 8192 / 512 = 16. Since the histogram
1112 * is offset by sm_shift we can simply use the value of
1113 * of i to calculate this (i.e. 2^i = 16 where i = 4).
1114 */
1115 sectors = msp->ms_sm->sm_phys->smp_histogram[i] << i;
1116 factor += (i + msp->ms_sm->sm_shift) * sectors;
1117 }
1118 return (factor * msp->ms_sm->sm_shift);
1119 }
1120
1121 static uint64_t
1122 metaslab_weight(metaslab_t *msp)
1123 {
1124 metaslab_group_t *mg = msp->ms_group;
1125 vdev_t *vd = mg->mg_vd;
1126 uint64_t weight, space;
1127
1128 ASSERT(MUTEX_HELD(&msp->ms_lock));
1129
1130 /*
1131 * This vdev is in the process of being removed so there is nothing
1132 * for us to do here.
1133 */
1134 if (vd->vdev_removing) {
1135 ASSERT0(space_map_allocated(msp->ms_sm));
1136 ASSERT0(vd->vdev_ms_shift);
1137 return (0);
1138 }
1139
1140 /*
1141 * The baseline weight is the metaslab's free space.
1142 */
1143 space = msp->ms_size - space_map_allocated(msp->ms_sm);
1144 weight = space;
1145
1146 /*
1147 * Modern disks have uniform bit density and constant angular velocity.
1148 * Therefore, the outer recording zones are faster (higher bandwidth)
1149 * than the inner zones by the ratio of outer to inner track diameter,
1150 * which is typically around 2:1. We account for this by assigning
1151 * higher weight to lower metaslabs (multiplier ranging from 2x to 1x).
1152 * In effect, this means that we'll select the metaslab with the most
1153 * free bandwidth rather than simply the one with the most free space.
1154 */
1155 weight = 2 * weight - (msp->ms_id * weight) / vd->vdev_ms_count;
1156 ASSERT(weight >= space && weight <= 2 * space);
1157
1158 msp->ms_factor = metaslab_weight_factor(msp);
1159 if (metaslab_weight_factor_enable)
1160 weight += msp->ms_factor;
1161
1162 if (msp->ms_loaded && !msp->ms_ops->msop_fragmented(msp)) {
1163 /*
1164 * If this metaslab is one we're actively using, adjust its
1165 * weight to make it preferable to any inactive metaslab so
1166 * we'll polish it off.
1167 */
1168 weight |= (msp->ms_weight & METASLAB_ACTIVE_MASK);
1169 }
1170
1171 return (weight);
1172 }
1173
1174 static int
1175 metaslab_activate(metaslab_t *msp, uint64_t activation_weight)
1176 {
1177 ASSERT(MUTEX_HELD(&msp->ms_lock));
1178
1179 if ((msp->ms_weight & METASLAB_ACTIVE_MASK) == 0) {
1180 metaslab_load_wait(msp);
1181 if (!msp->ms_loaded) {
1182 int error = metaslab_load(msp);
1183 if (error) {
1184 metaslab_group_sort(msp->ms_group, msp, 0);
1185 return (error);
1186 }
1187 }
1188
1189 metaslab_group_sort(msp->ms_group, msp,
1190 msp->ms_weight | activation_weight);
1191 }
1192 ASSERT(msp->ms_loaded);
1193 ASSERT(msp->ms_weight & METASLAB_ACTIVE_MASK);
1194
1195 return (0);
1196 }
1197
1198 static void
1199 metaslab_passivate(metaslab_t *msp, uint64_t size)
1200 {
1201 /*
1202 * If size < SPA_MINBLOCKSIZE, then we will not allocate from
1203 * this metaslab again. In that case, it had better be empty,
1204 * or we would be leaving space on the table.
1205 */
1206 ASSERT(size >= SPA_MINBLOCKSIZE || range_tree_space(msp->ms_tree) == 0);
1207 metaslab_group_sort(msp->ms_group, msp, MIN(msp->ms_weight, size));
1208 ASSERT((msp->ms_weight & METASLAB_ACTIVE_MASK) == 0);
1209 }
1210
1211 static void
1212 metaslab_preload(void *arg)
1213 {
1214 metaslab_t *msp = arg;
1215 spa_t *spa = msp->ms_group->mg_vd->vdev_spa;
1216
1217 mutex_enter(&msp->ms_lock);
1218 metaslab_load_wait(msp);
1219 if (!msp->ms_loaded)
1220 (void) metaslab_load(msp);
1221
1222 /*
1223 * Set the ms_access_txg value so that we don't unload it right away.
1224 */
1225 msp->ms_access_txg = spa_syncing_txg(spa) + metaslab_unload_delay + 1;
1226 mutex_exit(&msp->ms_lock);
1227 }
1228
1229 static void
1230 metaslab_group_preload(metaslab_group_t *mg)
1231 {
1232 spa_t *spa = mg->mg_vd->vdev_spa;
1233 metaslab_t *msp;
1234 avl_tree_t *t = &mg->mg_metaslab_tree;
1235 int m = 0;
1236
1237 if (spa_shutting_down(spa) || !metaslab_preload_enabled) {
1238 taskq_wait(mg->mg_taskq);
1239 return;
1240 }
1241 mutex_enter(&mg->mg_lock);
1242
1243 /*
1244 * Prefetch the next potential metaslabs
1245 */
1246 for (msp = avl_first(t); msp != NULL; msp = AVL_NEXT(t, msp)) {
1247
1248 /* If we have reached our preload limit then we're done */
1249 if (++m > metaslab_preload_limit)
1250 break;
1251
1252 VERIFY(taskq_dispatch(mg->mg_taskq, metaslab_preload,
1253 msp, TQ_SLEEP) != NULL);
1254 }
1255 mutex_exit(&mg->mg_lock);
1256 }
1257
1258 /*
1259 * Determine if the space map's on-disk footprint is past our tolerance
1260 * for inefficiency. We would like to use the following criteria to make
1261 * our decision:
1262 *
1263 * 1. The size of the space map object should not dramatically increase as a
1264 * result of writing out the free space range tree.
1265 *
1266 * 2. The minimal on-disk space map representation is zfs_condense_pct/100
1267 * times the size than the free space range tree representation
1268 * (i.e. zfs_condense_pct = 110 and in-core = 1MB, minimal = 1.1.MB).
1269 *
1270 * Checking the first condition is tricky since we don't want to walk
1271 * the entire AVL tree calculating the estimated on-disk size. Instead we
1272 * use the size-ordered range tree in the metaslab and calculate the
1273 * size required to write out the largest segment in our free tree. If the
1274 * size required to represent that segment on disk is larger than the space
1275 * map object then we avoid condensing this map.
1276 *
1277 * To determine the second criterion we use a best-case estimate and assume
1278 * each segment can be represented on-disk as a single 64-bit entry. We refer
1279 * to this best-case estimate as the space map's minimal form.
1280 */
1281 static boolean_t
1282 metaslab_should_condense(metaslab_t *msp)
1283 {
1284 space_map_t *sm = msp->ms_sm;
1285 range_seg_t *rs;
1286 uint64_t size, entries, segsz;
1287
1288 ASSERT(MUTEX_HELD(&msp->ms_lock));
1289 ASSERT(msp->ms_loaded);
1290
1291 /*
1292 * Use the ms_size_tree range tree, which is ordered by size, to
1293 * obtain the largest segment in the free tree. If the tree is empty
1294 * then we should condense the map.
1295 */
1296 rs = avl_last(&msp->ms_size_tree);
1297 if (rs == NULL)
1298 return (B_TRUE);
1299
1300 /*
1301 * Calculate the number of 64-bit entries this segment would
1302 * require when written to disk. If this single segment would be
1303 * larger on-disk than the entire current on-disk structure, then
1304 * clearly condensing will increase the on-disk structure size.
1305 */
1306 size = (rs->rs_end - rs->rs_start) >> sm->sm_shift;
1307 entries = size / (MIN(size, SM_RUN_MAX));
1308 segsz = entries * sizeof (uint64_t);
1309
1310 return (segsz <= space_map_length(msp->ms_sm) &&
1311 space_map_length(msp->ms_sm) >= (zfs_condense_pct *
1312 sizeof (uint64_t) * avl_numnodes(&msp->ms_tree->rt_root)) / 100);
1313 }
1314
1315 /*
1316 * Condense the on-disk space map representation to its minimized form.
1317 * The minimized form consists of a small number of allocations followed by
1318 * the entries of the free range tree.
1319 */
1320 static void
1321 metaslab_condense(metaslab_t *msp, uint64_t txg, dmu_tx_t *tx)
1322 {
1323 spa_t *spa = msp->ms_group->mg_vd->vdev_spa;
1324 range_tree_t *freetree = msp->ms_freetree[txg & TXG_MASK];
1325 range_tree_t *condense_tree;
1326 space_map_t *sm = msp->ms_sm;
1327
1328 ASSERT(MUTEX_HELD(&msp->ms_lock));
1329 ASSERT3U(spa_sync_pass(spa), ==, 1);
1330 ASSERT(msp->ms_loaded);
1331
1332 spa_dbgmsg(spa, "condensing: txg %llu, msp[%llu] %p, "
1333 "smp size %llu, segments %lu", txg, msp->ms_id, msp,
1334 space_map_length(msp->ms_sm), avl_numnodes(&msp->ms_tree->rt_root));
1335
1336 /*
1337 * Create an range tree that is 100% allocated. We remove segments
1338 * that have been freed in this txg, any deferred frees that exist,
1339 * and any allocation in the future. Removing segments should be
1340 * a relatively inexpensive operation since we expect these trees to
1341 * have a small number of nodes.
1342 */
1343 condense_tree = range_tree_create(NULL, NULL, &msp->ms_lock);
1344 range_tree_add(condense_tree, msp->ms_start, msp->ms_size);
1345
1346 /*
1347 * Remove what's been freed in this txg from the condense_tree.
1348 * Since we're in sync_pass 1, we know that all the frees from
1349 * this txg are in the freetree.
1350 */
1351 range_tree_walk(freetree, range_tree_remove, condense_tree);
1352
1353 for (int t = 0; t < TXG_DEFER_SIZE; t++) {
1354 range_tree_walk(msp->ms_defertree[t],
1355 range_tree_remove, condense_tree);
1356 }
1357
1358 for (int t = 1; t < TXG_CONCURRENT_STATES; t++) {
1359 range_tree_walk(msp->ms_alloctree[(txg + t) & TXG_MASK],
1360 range_tree_remove, condense_tree);
1361 }
1362
1363 /*
1364 * We're about to drop the metaslab's lock thus allowing
1365 * other consumers to change it's content. Set the
1366 * metaslab's ms_condensing flag to ensure that
1367 * allocations on this metaslab do not occur while we're
1368 * in the middle of committing it to disk. This is only critical
1369 * for the ms_tree as all other range trees use per txg
1370 * views of their content.
1371 */
1372 msp->ms_condensing = B_TRUE;
1373
1374 mutex_exit(&msp->ms_lock);
1375 space_map_truncate(sm, tx);
1376 mutex_enter(&msp->ms_lock);
1377
1378 /*
1379 * While we would ideally like to create a space_map representation
1380 * that consists only of allocation records, doing so can be
1381 * prohibitively expensive because the in-core free tree can be
1382 * large, and therefore computationally expensive to subtract
1383 * from the condense_tree. Instead we sync out two trees, a cheap
1384 * allocation only tree followed by the in-core free tree. While not
1385 * optimal, this is typically close to optimal, and much cheaper to
1386 * compute.
1387 */
1388 space_map_write(sm, condense_tree, SM_ALLOC, tx);
1389 range_tree_vacate(condense_tree, NULL, NULL);
1390 range_tree_destroy(condense_tree);
1391
1392 space_map_write(sm, msp->ms_tree, SM_FREE, tx);
1393 msp->ms_condensing = B_FALSE;
1394 }
1395
1396 /*
1397 * Write a metaslab to disk in the context of the specified transaction group.
1398 */
1399 void
1400 metaslab_sync(metaslab_t *msp, uint64_t txg)
1401 {
1402 metaslab_group_t *mg = msp->ms_group;
1403 vdev_t *vd = mg->mg_vd;
1404 spa_t *spa = vd->vdev_spa;
1405 objset_t *mos = spa_meta_objset(spa);
1406 range_tree_t *alloctree = msp->ms_alloctree[txg & TXG_MASK];
1407 range_tree_t **freetree = &msp->ms_freetree[txg & TXG_MASK];
1408 range_tree_t **freed_tree =
1409 &msp->ms_freetree[TXG_CLEAN(txg) & TXG_MASK];
1410 dmu_tx_t *tx;
1411 uint64_t object = space_map_object(msp->ms_sm);
1412
1413 ASSERT(!vd->vdev_ishole);
1414
1415 /*
1416 * This metaslab has just been added so there's no work to do now.
1417 */
1418 if (*freetree == NULL) {
1419 ASSERT3P(alloctree, ==, NULL);
1420 return;
1421 }
1422
1423 ASSERT3P(alloctree, !=, NULL);
1424 ASSERT3P(*freetree, !=, NULL);
1425 ASSERT3P(*freed_tree, !=, NULL);
1426
1427 if (range_tree_space(alloctree) == 0 &&
1428 range_tree_space(*freetree) == 0)
1429 return;
1430
1431 /*
1432 * The only state that can actually be changing concurrently with
1433 * metaslab_sync() is the metaslab's ms_tree. No other thread can
1434 * be modifying this txg's alloctree, freetree, freed_tree, or
1435 * space_map_phys_t. Therefore, we only hold ms_lock to satify
1436 * space_map ASSERTs. We drop it whenever we call into the DMU,
1437 * because the DMU can call down to us (e.g. via zio_free()) at
1438 * any time.
1439 */
1440
1441 tx = dmu_tx_create_assigned(spa_get_dsl(spa), txg);
1442
1443 if (msp->ms_sm == NULL) {
1444 uint64_t new_object;
1445
1446 new_object = space_map_alloc(mos, tx);
1447 VERIFY3U(new_object, !=, 0);
1448
1449 VERIFY0(space_map_open(&msp->ms_sm, mos, new_object,
1450 msp->ms_start, msp->ms_size, vd->vdev_ashift,
1451 &msp->ms_lock));
1452 ASSERT(msp->ms_sm != NULL);
1453 }
1454
1455 mutex_enter(&msp->ms_lock);
1456
1457 if (msp->ms_loaded && spa_sync_pass(spa) == 1 &&
1458 metaslab_should_condense(msp)) {
1459 metaslab_condense(msp, txg, tx);
1460 } else {
1461 space_map_write(msp->ms_sm, alloctree, SM_ALLOC, tx);
1462 space_map_write(msp->ms_sm, *freetree, SM_FREE, tx);
1463 }
1464
1465 range_tree_vacate(alloctree, NULL, NULL);
1466
1467 if (msp->ms_loaded) {
1468 /*
1469 * When the space map is loaded, we have an accruate
1470 * histogram in the range tree. This gives us an opportunity
1471 * to bring the space map's histogram up-to-date so we clear
1472 * it first before updating it.
1473 */
1474 space_map_histogram_clear(msp->ms_sm);
1475 space_map_histogram_add(msp->ms_sm, msp->ms_tree, tx);
1476 } else {
1477 /*
1478 * Since the space map is not loaded we simply update the
1479 * exisiting histogram with what was freed in this txg. This
1480 * means that the on-disk histogram may not have an accurate
1481 * view of the free space but it's close enough to allow
1482 * us to make allocation decisions.
1483 */
1484 space_map_histogram_add(msp->ms_sm, *freetree, tx);
1485 }
1486
1487 /*
1488 * For sync pass 1, we avoid traversing this txg's free range tree
1489 * and instead will just swap the pointers for freetree and
1490 * freed_tree. We can safely do this since the freed_tree is
1491 * guaranteed to be empty on the initial pass.
1492 */
1493 if (spa_sync_pass(spa) == 1) {
1494 range_tree_swap(freetree, freed_tree);
1495 } else {
1496 range_tree_vacate(*freetree, range_tree_add, *freed_tree);
1497 }
1498
1499 ASSERT0(range_tree_space(msp->ms_alloctree[txg & TXG_MASK]));
1500 ASSERT0(range_tree_space(msp->ms_freetree[txg & TXG_MASK]));
1501
1502 mutex_exit(&msp->ms_lock);
1503
1504 if (object != space_map_object(msp->ms_sm)) {
1505 object = space_map_object(msp->ms_sm);
1506 dmu_write(mos, vd->vdev_ms_array, sizeof (uint64_t) *
1507 msp->ms_id, sizeof (uint64_t), &object, tx);
1508 }
1509 dmu_tx_commit(tx);
1510 }
1511
1512 /*
1513 * Called after a transaction group has completely synced to mark
1514 * all of the metaslab's free space as usable.
1515 */
1516 void
1517 metaslab_sync_done(metaslab_t *msp, uint64_t txg)
1518 {
1519 metaslab_group_t *mg = msp->ms_group;
1520 vdev_t *vd = mg->mg_vd;
1521 range_tree_t **freed_tree;
1522 range_tree_t **defer_tree;
1523 int64_t alloc_delta, defer_delta;
1524
1525 ASSERT(!vd->vdev_ishole);
1526
1527 mutex_enter(&msp->ms_lock);
1528
1529 /*
1530 * If this metaslab is just becoming available, initialize its
1531 * alloctrees, freetrees, and defertree and add its capacity to
1532 * the vdev.
1533 */
1534 if (msp->ms_freetree[TXG_CLEAN(txg) & TXG_MASK] == NULL) {
1535 for (int t = 0; t < TXG_SIZE; t++) {
1536 ASSERT(msp->ms_alloctree[t] == NULL);
1537 ASSERT(msp->ms_freetree[t] == NULL);
1538
1539 msp->ms_alloctree[t] = range_tree_create(NULL, msp,
1540 &msp->ms_lock);
1541 msp->ms_freetree[t] = range_tree_create(NULL, msp,
1542 &msp->ms_lock);
1543 }
1544
1545 for (int t = 0; t < TXG_DEFER_SIZE; t++) {
1546 ASSERT(msp->ms_defertree[t] == NULL);
1547
1548 msp->ms_defertree[t] = range_tree_create(NULL, msp,
1549 &msp->ms_lock);
1550 }
1551
1552 vdev_space_update(vd, 0, 0, msp->ms_size);
1553 }
1554
1555 freed_tree = &msp->ms_freetree[TXG_CLEAN(txg) & TXG_MASK];
1556 defer_tree = &msp->ms_defertree[txg % TXG_DEFER_SIZE];
1557
1558 alloc_delta = space_map_alloc_delta(msp->ms_sm);
1559 defer_delta = range_tree_space(*freed_tree) -
1560 range_tree_space(*defer_tree);
1561
1562 vdev_space_update(vd, alloc_delta + defer_delta, defer_delta, 0);
1563
1564 ASSERT0(range_tree_space(msp->ms_alloctree[txg & TXG_MASK]));
1565 ASSERT0(range_tree_space(msp->ms_freetree[txg & TXG_MASK]));
1566
1567 /*
1568 * If there's a metaslab_load() in progress, wait for it to complete
1569 * so that we have a consistent view of the in-core space map.
1570 */
1571 metaslab_load_wait(msp);
1572
1573 /*
1574 * Move the frees from the defer_tree back to the free
1575 * range tree (if it's loaded). Swap the freed_tree and the
1576 * defer_tree -- this is safe to do because we've just emptied out
1577 * the defer_tree.
1578 */
1579 range_tree_vacate(*defer_tree,
1580 msp->ms_loaded ? range_tree_add : NULL, msp->ms_tree);
1581 range_tree_swap(freed_tree, defer_tree);
1582
1583 space_map_update(msp->ms_sm);
1584
1585 msp->ms_deferspace += defer_delta;
1586 ASSERT3S(msp->ms_deferspace, >=, 0);
1587 ASSERT3S(msp->ms_deferspace, <=, msp->ms_size);
1588 if (msp->ms_deferspace != 0) {
1589 /*
1590 * Keep syncing this metaslab until all deferred frees
1591 * are back in circulation.
1592 */
1593 vdev_dirty(vd, VDD_METASLAB, msp, txg + 1);
1594 }
1595
1596 if (msp->ms_loaded && msp->ms_access_txg < txg) {
1597 for (int t = 1; t < TXG_CONCURRENT_STATES; t++) {
1598 VERIFY0(range_tree_space(
1599 msp->ms_alloctree[(txg + t) & TXG_MASK]));
1600 }
1601
1602 if (!metaslab_debug_unload)
1603 metaslab_unload(msp);
1604 }
1605
1606 metaslab_group_sort(mg, msp, metaslab_weight(msp));
1607 mutex_exit(&msp->ms_lock);
1608
1609 }
1610
1611 void
1612 metaslab_sync_reassess(metaslab_group_t *mg)
1613 {
1614 int64_t failures = mg->mg_alloc_failures;
1615
1616 metaslab_group_alloc_update(mg);
1617 atomic_add_64(&mg->mg_alloc_failures, -failures);
1618
1619 /*
1620 * Preload the next potential metaslabs
1621 */
1622 metaslab_group_preload(mg);
1623 }
1624
1625 static uint64_t
1626 metaslab_distance(metaslab_t *msp, dva_t *dva)
1627 {
1628 uint64_t ms_shift = msp->ms_group->mg_vd->vdev_ms_shift;
1629 uint64_t offset = DVA_GET_OFFSET(dva) >> ms_shift;
1630 uint64_t start = msp->ms_id;
1631
1632 if (msp->ms_group->mg_vd->vdev_id != DVA_GET_VDEV(dva))
1633 return (1ULL << 63);
1634
1635 if (offset < start)
1636 return ((start - offset) << ms_shift);
1637 if (offset > start)
1638 return ((offset - start) << ms_shift);
1639 return (0);
1640 }
1641
1642 static uint64_t
1643 metaslab_group_alloc(metaslab_group_t *mg, uint64_t psize, uint64_t asize,
1644 uint64_t txg, uint64_t min_distance, dva_t *dva, int d, int flags)
1645 {
1646 spa_t *spa = mg->mg_vd->vdev_spa;
1647 metaslab_t *msp = NULL;
1648 uint64_t offset = -1ULL;
1649 avl_tree_t *t = &mg->mg_metaslab_tree;
1650 uint64_t activation_weight;
1662 for (;;) {
1663 boolean_t was_active;
1664
1665 mutex_enter(&mg->mg_lock);
1666 for (msp = avl_first(t); msp; msp = AVL_NEXT(t, msp)) {
1667 if (msp->ms_weight < asize) {
1668 spa_dbgmsg(spa, "%s: failed to meet weight "
1669 "requirement: vdev %llu, txg %llu, mg %p, "
1670 "msp %p, psize %llu, asize %llu, "
1671 "failures %llu, weight %llu",
1672 spa_name(spa), mg->mg_vd->vdev_id, txg,
1673 mg, msp, psize, asize,
1674 mg->mg_alloc_failures, msp->ms_weight);
1675 mutex_exit(&mg->mg_lock);
1676 return (-1ULL);
1677 }
1678
1679 /*
1680 * If the selected metaslab is condensing, skip it.
1681 */
1682 if (msp->ms_condensing)
1683 continue;
1684
1685 was_active = msp->ms_weight & METASLAB_ACTIVE_MASK;
1686 if (activation_weight == METASLAB_WEIGHT_PRIMARY)
1687 break;
1688
1689 target_distance = min_distance +
1690 (space_map_allocated(msp->ms_sm) != 0 ? 0 :
1691 min_distance >> 1);
1692
1693 for (i = 0; i < d; i++)
1694 if (metaslab_distance(msp, &dva[i]) <
1695 target_distance)
1696 break;
1697 if (i == d)
1698 break;
1699 }
1700 mutex_exit(&mg->mg_lock);
1701 if (msp == NULL)
1702 return (-1ULL);
1703
1704 mutex_enter(&msp->ms_lock);
1705
1706 /*
1707 * If we've already reached the allowable number of failed
1708 * allocation attempts on this metaslab group then we
1709 * consider skipping it. We skip it only if we're allowed
1710 * to "fast" gang, the physical size is larger than
1711 * a gang block, and we're attempting to allocate from
1712 * the primary metaslab.
1713 */
1714 if (mg->mg_alloc_failures > zfs_mg_alloc_failures &&
1715 CAN_FASTGANG(flags) && psize > SPA_GANGBLOCKSIZE &&
1716 activation_weight == METASLAB_WEIGHT_PRIMARY) {
1717 spa_dbgmsg(spa, "%s: skipping metaslab group: "
1718 "vdev %llu, txg %llu, mg %p, msp[%llu] %p, "
1719 "psize %llu, asize %llu, failures %llu",
1720 spa_name(spa), mg->mg_vd->vdev_id, txg, mg,
1721 msp->ms_id, msp, psize, asize,
1722 mg->mg_alloc_failures);
1723 mutex_exit(&msp->ms_lock);
1724 return (-1ULL);
1725 }
1726
1727 /*
1728 * Ensure that the metaslab we have selected is still
1729 * capable of handling our request. It's possible that
1730 * another thread may have changed the weight while we
1731 * were blocked on the metaslab lock.
1732 */
1733 if (msp->ms_weight < asize || (was_active &&
1734 !(msp->ms_weight & METASLAB_ACTIVE_MASK) &&
1735 activation_weight == METASLAB_WEIGHT_PRIMARY)) {
1736 mutex_exit(&msp->ms_lock);
1737 continue;
1738 }
1739
1740 if ((msp->ms_weight & METASLAB_WEIGHT_SECONDARY) &&
1741 activation_weight == METASLAB_WEIGHT_PRIMARY) {
1742 metaslab_passivate(msp,
1743 msp->ms_weight & ~METASLAB_ACTIVE_MASK);
1744 mutex_exit(&msp->ms_lock);
1745 continue;
1746 }
1747
1748 if (metaslab_activate(msp, activation_weight) != 0) {
1749 mutex_exit(&msp->ms_lock);
1750 continue;
1751 }
1752
1753 /*
1754 * If this metaslab is currently condensing then pick again as
1755 * we can't manipulate this metaslab until it's committed
1756 * to disk.
1757 */
1758 if (msp->ms_condensing) {
1759 mutex_exit(&msp->ms_lock);
1760 continue;
1761 }
1762
1763 if ((offset = metaslab_block_alloc(msp, asize)) != -1ULL)
1764 break;
1765
1766 atomic_inc_64(&mg->mg_alloc_failures);
1767
1768 metaslab_passivate(msp, metaslab_block_maxsize(msp));
1769 mutex_exit(&msp->ms_lock);
1770 }
1771
1772 if (range_tree_space(msp->ms_alloctree[txg & TXG_MASK]) == 0)
1773 vdev_dirty(mg->mg_vd, VDD_METASLAB, msp, txg);
1774
1775 range_tree_add(msp->ms_alloctree[txg & TXG_MASK], offset, asize);
1776 msp->ms_access_txg = txg + metaslab_unload_delay;
1777
1778 mutex_exit(&msp->ms_lock);
1779
1780 return (offset);
1781 }
1782
1783 /*
1784 * Allocate a block for the specified i/o.
1785 */
1786 static int
1787 metaslab_alloc_dva(spa_t *spa, metaslab_class_t *mc, uint64_t psize,
1788 dva_t *dva, int d, dva_t *hintdva, uint64_t txg, int flags)
1789 {
1790 metaslab_group_t *mg, *rotor;
1791 vdev_t *vd;
1792 int dshift = 3;
1793 int all_zero;
1794 int zio_lock = B_FALSE;
1795 boolean_t allocatable;
1796 uint64_t offset = -1ULL;
2003
2004 if (txg > spa_freeze_txg(spa))
2005 return;
2006
2007 if ((vd = vdev_lookup_top(spa, vdev)) == NULL ||
2008 (offset >> vd->vdev_ms_shift) >= vd->vdev_ms_count) {
2009 cmn_err(CE_WARN, "metaslab_free_dva(): bad DVA %llu:%llu",
2010 (u_longlong_t)vdev, (u_longlong_t)offset);
2011 ASSERT(0);
2012 return;
2013 }
2014
2015 msp = vd->vdev_ms[offset >> vd->vdev_ms_shift];
2016
2017 if (DVA_GET_GANG(dva))
2018 size = vdev_psize_to_asize(vd, SPA_GANGBLOCKSIZE);
2019
2020 mutex_enter(&msp->ms_lock);
2021
2022 if (now) {
2023 range_tree_remove(msp->ms_alloctree[txg & TXG_MASK],
2024 offset, size);
2025
2026 VERIFY(!msp->ms_condensing);
2027 VERIFY3U(offset, >=, msp->ms_start);
2028 VERIFY3U(offset + size, <=, msp->ms_start + msp->ms_size);
2029 VERIFY3U(range_tree_space(msp->ms_tree) + size, <=,
2030 msp->ms_size);
2031 VERIFY0(P2PHASE(offset, 1ULL << vd->vdev_ashift));
2032 VERIFY0(P2PHASE(size, 1ULL << vd->vdev_ashift));
2033 range_tree_add(msp->ms_tree, offset, size);
2034 } else {
2035 if (range_tree_space(msp->ms_freetree[txg & TXG_MASK]) == 0)
2036 vdev_dirty(vd, VDD_METASLAB, msp, txg);
2037 range_tree_add(msp->ms_freetree[txg & TXG_MASK],
2038 offset, size);
2039 }
2040
2041 mutex_exit(&msp->ms_lock);
2042 }
2043
2044 /*
2045 * Intent log support: upon opening the pool after a crash, notify the SPA
2046 * of blocks that the intent log has allocated for immediate write, but
2047 * which are still considered free by the SPA because the last transaction
2048 * group didn't commit yet.
2049 */
2050 static int
2051 metaslab_claim_dva(spa_t *spa, const dva_t *dva, uint64_t txg)
2052 {
2053 uint64_t vdev = DVA_GET_VDEV(dva);
2054 uint64_t offset = DVA_GET_OFFSET(dva);
2055 uint64_t size = DVA_GET_ASIZE(dva);
2056 vdev_t *vd;
2057 metaslab_t *msp;
2058 int error = 0;
2059
2060 ASSERT(DVA_IS_VALID(dva));
2061
2062 if ((vd = vdev_lookup_top(spa, vdev)) == NULL ||
2063 (offset >> vd->vdev_ms_shift) >= vd->vdev_ms_count)
2064 return (SET_ERROR(ENXIO));
2065
2066 msp = vd->vdev_ms[offset >> vd->vdev_ms_shift];
2067
2068 if (DVA_GET_GANG(dva))
2069 size = vdev_psize_to_asize(vd, SPA_GANGBLOCKSIZE);
2070
2071 mutex_enter(&msp->ms_lock);
2072
2073 if ((txg != 0 && spa_writeable(spa)) || !msp->ms_loaded)
2074 error = metaslab_activate(msp, METASLAB_WEIGHT_SECONDARY);
2075
2076 if (error == 0 && !range_tree_contains(msp->ms_tree, offset, size))
2077 error = SET_ERROR(ENOENT);
2078
2079 if (error || txg == 0) { /* txg == 0 indicates dry run */
2080 mutex_exit(&msp->ms_lock);
2081 return (error);
2082 }
2083
2084 VERIFY(!msp->ms_condensing);
2085 VERIFY0(P2PHASE(offset, 1ULL << vd->vdev_ashift));
2086 VERIFY0(P2PHASE(size, 1ULL << vd->vdev_ashift));
2087 VERIFY3U(range_tree_space(msp->ms_tree) - size, <=, msp->ms_size);
2088 range_tree_remove(msp->ms_tree, offset, size);
2089
2090 if (spa_writeable(spa)) { /* don't dirty if we're zdb(1M) */
2091 if (range_tree_space(msp->ms_alloctree[txg & TXG_MASK]) == 0)
2092 vdev_dirty(vd, VDD_METASLAB, msp, txg);
2093 range_tree_add(msp->ms_alloctree[txg & TXG_MASK], offset, size);
2094 }
2095
2096 mutex_exit(&msp->ms_lock);
2097
2098 return (0);
2099 }
2100
2101 int
2102 metaslab_alloc(spa_t *spa, metaslab_class_t *mc, uint64_t psize, blkptr_t *bp,
2103 int ndvas, uint64_t txg, blkptr_t *hintbp, int flags)
2104 {
2105 dva_t *dva = bp->blk_dva;
2106 dva_t *hintdva = hintbp->blk_dva;
2107 int error = 0;
2108
2109 ASSERT(bp->blk_birth == 0);
2110 ASSERT(BP_PHYSICAL_BIRTH(bp) == 0);
2111
2112 spa_config_enter(spa, SCL_ALLOC, FTAG, RW_READER);
2113
2114 if (mc->mc_rotor == NULL) { /* no vdevs in this class */
2115 spa_config_exit(spa, SCL_ALLOC, FTAG);
2116 return (SET_ERROR(ENOSPC));
2117 }
2118
2119 ASSERT(ndvas > 0 && ndvas <= spa_max_replication(spa));
2120 ASSERT(BP_GET_NDVAS(bp) == 0);
2121 ASSERT(hintbp == NULL || ndvas <= BP_GET_NDVAS(hintbp));
2122
2123 for (int d = 0; d < ndvas; d++) {
2124 error = metaslab_alloc_dva(spa, mc, psize, dva, d, hintdva,
2125 txg, flags);
2126 if (error != 0) {
2127 for (d--; d >= 0; d--) {
2128 metaslab_free_dva(spa, &dva[d], txg, B_TRUE);
2129 bzero(&dva[d], sizeof (dva_t));
2130 }
2131 spa_config_exit(spa, SCL_ALLOC, FTAG);
2132 return (error);
2133 }
2134 }
2135 ASSERT(error == 0);
2136 ASSERT(BP_GET_NDVAS(bp) == ndvas);
2137
2138 spa_config_exit(spa, SCL_ALLOC, FTAG);
2139
2140 BP_SET_BIRTH(bp, txg, txg);
2141
2142 return (0);
2143 }
2144
2145 void
2146 metaslab_free(spa_t *spa, const blkptr_t *bp, uint64_t txg, boolean_t now)
2173 * First do a dry run to make sure all DVAs are claimable,
2174 * so we don't have to unwind from partial failures below.
2175 */
2176 if ((error = metaslab_claim(spa, bp, 0)) != 0)
2177 return (error);
2178 }
2179
2180 spa_config_enter(spa, SCL_ALLOC, FTAG, RW_READER);
2181
2182 for (int d = 0; d < ndvas; d++)
2183 if ((error = metaslab_claim_dva(spa, &dva[d], txg)) != 0)
2184 break;
2185
2186 spa_config_exit(spa, SCL_ALLOC, FTAG);
2187
2188 ASSERT(error == 0 || txg == 0);
2189
2190 return (error);
2191 }
2192
2193 void
2194 metaslab_check_free(spa_t *spa, const blkptr_t *bp)
2195 {
2196 if ((zfs_flags & ZFS_DEBUG_ZIO_FREE) == 0)
2197 return;
2198
2199 spa_config_enter(spa, SCL_VDEV, FTAG, RW_READER);
2200 for (int i = 0; i < BP_GET_NDVAS(bp); i++) {
2201 uint64_t vdev = DVA_GET_VDEV(&bp->blk_dva[i]);
2202 vdev_t *vd = vdev_lookup_top(spa, vdev);
2203 uint64_t offset = DVA_GET_OFFSET(&bp->blk_dva[i]);
2204 uint64_t size = DVA_GET_ASIZE(&bp->blk_dva[i]);
2205 metaslab_t *msp = vd->vdev_ms[offset >> vd->vdev_ms_shift];
2206
2207 if (msp->ms_loaded)
2208 range_tree_verify(msp->ms_tree, offset, size);
2209
2210 for (int j = 0; j < TXG_SIZE; j++)
2211 range_tree_verify(msp->ms_freetree[j], offset, size);
2212 for (int j = 0; j < TXG_DEFER_SIZE; j++)
2213 range_tree_verify(msp->ms_defertree[j], offset, size);
2214 }
2215 spa_config_exit(spa, SCL_VDEV, FTAG);
2216 }
|