space_reftree.c revision 339158 
1/*
2 * CDDL HEADER START
3 *
4 * The contents of this file are subject to the terms of the
5 * Common Development and Distribution License (the "License").
6 * You may not use this file except in compliance with the License.
7 *
8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9 * or http://www.opensolaris.org/os/licensing.
10 * See the License for the specific language governing permissions
11 * and limitations under the License.
12 *
13 * When distributing Covered Code, include this CDDL HEADER in each
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 2009 Sun Microsystems, Inc.  All rights reserved.
23 * Use is subject to license terms.
24 */
25/*
26 * Copyright (c) 2013, 2015 by Delphix. All rights reserved.
27 */
28
29#include <sys/zfs_context.h>
30#include <sys/range_tree.h>
31#include <sys/space_reftree.h>
32
33/*
34 * Space reference trees.
35 *
36 * A range tree is a collection of integers.  Every integer is either
37 * in the tree, or it's not.  A space reference tree generalizes
38 * the idea: it allows its members to have arbitrary reference counts,
39 * as opposed to the implicit reference count of 0 or 1 in a range tree.
40 * This representation comes in handy when computing the union or
41 * intersection of multiple space maps.  For example, the union of
42 * N range trees is the subset of the reference tree with refcnt >= 1.
43 * The intersection of N range trees is the subset with refcnt >= N.
44 *
45 * [It's very much like a Fourier transform.  Unions and intersections
46 * are hard to perform in the 'range tree domain', so we convert the trees
47 * into the 'reference count domain', where it's trivial, then invert.]
48 *
49 * vdev_dtl_reassess() uses computations of this form to determine
50 * DTL_MISSING and DTL_OUTAGE for interior vdevs -- e.g. a RAID-Z vdev
51 * has an outage wherever refcnt >= vdev_nparity + 1, and a mirror vdev
52 * has an outage wherever refcnt >= vdev_children.
53 */
54static int
55space_reftree_compare(const void *x1, const void *x2)
56{
57	const space_ref_t *sr1 = (const space_ref_t *)x1;
58	const space_ref_t *sr2 = (const space_ref_t *)x2;
59
60	int cmp = AVL_CMP(sr1->sr_offset, sr2->sr_offset);
61	if (likely(cmp))
62		return (cmp);
63
64	return (AVL_PCMP(sr1, sr2));
65}
66
67void
68space_reftree_create(avl_tree_t *t)
69{
70	avl_create(t, space_reftree_compare,
71	    sizeof (space_ref_t), offsetof(space_ref_t, sr_node));
72}
73
74void
75space_reftree_destroy(avl_tree_t *t)
76{
77	space_ref_t *sr;
78	void *cookie = NULL;
79
80	while ((sr = avl_destroy_nodes(t, &cookie)) != NULL)
81		kmem_free(sr, sizeof (*sr));
82
83	avl_destroy(t);
84}
85
86static void
87space_reftree_add_node(avl_tree_t *t, uint64_t offset, int64_t refcnt)
88{
89	space_ref_t *sr;
90
91	sr = kmem_alloc(sizeof (*sr), KM_SLEEP);
92	sr->sr_offset = offset;
93	sr->sr_refcnt = refcnt;
94
95	avl_add(t, sr);
96}
97
98void
99space_reftree_add_seg(avl_tree_t *t, uint64_t start, uint64_t end,
100    int64_t refcnt)
101{
102	space_reftree_add_node(t, start, refcnt);
103	space_reftree_add_node(t, end, -refcnt);
104}
105
106/*
107 * Convert (or add) a range tree into a reference tree.
108 */
109void
110space_reftree_add_map(avl_tree_t *t, range_tree_t *rt, int64_t refcnt)
111{
112	range_seg_t *rs;
113
114	for (rs = avl_first(&rt->rt_root); rs; rs = AVL_NEXT(&rt->rt_root, rs))
115		space_reftree_add_seg(t, rs->rs_start, rs->rs_end, refcnt);
116}
117
118/*
119 * Convert a reference tree into a range tree.  The range tree will contain
120 * all members of the reference tree for which refcnt >= minref.
121 */
122void
123space_reftree_generate_map(avl_tree_t *t, range_tree_t *rt, int64_t minref)
124{
125	uint64_t start = -1ULL;
126	int64_t refcnt = 0;
127	space_ref_t *sr;
128
129	range_tree_vacate(rt, NULL, NULL);
130
131	for (sr = avl_first(t); sr != NULL; sr = AVL_NEXT(t, sr)) {
132		refcnt += sr->sr_refcnt;
133		if (refcnt >= minref) {
134			if (start == -1ULL) {
135				start = sr->sr_offset;
136			}
137		} else {
138			if (start != -1ULL) {
139				uint64_t end = sr->sr_offset;
140				ASSERT(start <= end);
141				if (end > start)
142					range_tree_add(rt, start, end - start);
143				start = -1ULL;
144			}
145		}
146	}
147	ASSERT(refcnt == 0);
148	ASSERT(start == -1ULL);
149}
150