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.
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.
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]
22 * Copyright 2009 Sun Microsystems, Inc. All rights reserved.
23 * Use is subject to license terms.
27 * Copyright (c) 2013 by Delphix. All rights reserved.
30 #ifndef _SYS_METASLAB_IMPL_H
31 #define _SYS_METASLAB_IMPL_H
33 #include <sys/metaslab.h>
34 #include <sys/space_map.h>
35 #include <sys/range_tree.h>
44 struct metaslab_class {
46 metaslab_group_t *mc_rotor;
47 metaslab_ops_t *mc_ops;
49 uint64_t mc_alloc_groups; /* # of allocatable groups */
50 uint64_t mc_alloc; /* total allocated space */
51 uint64_t mc_deferred; /* total deferred frees */
52 uint64_t mc_space; /* total space (alloc + free) */
53 uint64_t mc_dspace; /* total deflated space */
54 uint64_t mc_minblocksize;
57 struct metaslab_group {
59 avl_tree_t mg_metaslab_tree;
61 uint64_t mg_alloc_failures;
62 boolean_t mg_allocatable; /* can we allocate? */
63 uint64_t mg_free_capacity; /* percentage free */
65 int64_t mg_activation_count;
66 metaslab_class_t *mg_class;
69 metaslab_group_t *mg_prev;
70 metaslab_group_t *mg_next;
74 * This value defines the number of elements in the ms_lbas array. The value
75 * of 64 was chosen as it covers to cover all power of 2 buckets up to
76 * UINT64_MAX. This is the equivalent of highbit(UINT64_MAX).
81 * Each metaslab maintains a set of in-core trees to track metaslab operations.
82 * The in-core free tree (ms_tree) contains the current list of free segments.
83 * As blocks are allocated, the allocated segment are removed from the ms_tree
84 * and added to a per txg allocation tree (ms_alloctree). As blocks are freed,
85 * they are added to the per txg free tree (ms_freetree). These per txg
86 * trees allow us to process all allocations and frees in syncing context
87 * where it is safe to update the on-disk space maps. One additional in-core
88 * tree is maintained to track deferred frees (ms_defertree). Once a block
89 * is freed it will move from the ms_freetree to the ms_defertree. A deferred
90 * free means that a block has been freed but cannot be used by the pool
91 * until TXG_DEFER_SIZE transactions groups later. For example, a block
92 * that is freed in txg 50 will not be available for reallocation until
93 * txg 52 (50 + TXG_DEFER_SIZE). This provides a safety net for uberblock
94 * rollback. A pool could be safely rolled back TXG_DEFERS_SIZE
95 * transactions groups and ensure that no block has been reallocated.
97 * The simplified transition diagram looks like this:
103 * free segment (ms_tree) --------> ms_alloctree ----> (write to space map)
106 * | ms_freetree <--- FREE
110 * +----------- ms_defertree <-------+---------> (write to space map)
113 * Each metaslab's space is tracked in a single space map in the MOS,
114 * which is only updated in syncing context. Each time we sync a txg,
115 * we append the allocs and frees from that txg to the space map.
116 * The pool space is only updated once all metaslabs have finished syncing.
118 * To load the in-core free tree we read the space map from disk.
119 * This object contains a series of alloc and free records that are
120 * combined to make up the list of all free segments in this metaslab. These
121 * segments are represented in-core by the ms_tree and are stored in an
124 * As the space map grows (as a result of the appends) it will
125 * eventually become space-inefficient. When the metaslab's in-core free tree
126 * is zfs_condense_pct/100 times the size of the minimal on-disk
127 * representation, we rewrite it in its minimized form. If a metaslab
128 * needs to condense then we must set the ms_condensing flag to ensure
129 * that allocations are not performed on the metaslab that is being written.
133 kcondvar_t ms_load_cv;
135 metaslab_ops_t *ms_ops;
140 range_tree_t *ms_alloctree[TXG_SIZE];
141 range_tree_t *ms_freetree[TXG_SIZE];
142 range_tree_t *ms_defertree[TXG_DEFER_SIZE];
143 range_tree_t *ms_tree;
145 boolean_t ms_condensing; /* condensing? */
147 boolean_t ms_loading;
149 int64_t ms_deferspace; /* sum of ms_defermap[] space */
150 uint64_t ms_weight; /* weight vs. others in group */
152 uint64_t ms_access_txg;
155 * The metaslab block allocators can optionally use a size-ordered
156 * range tree and/or an array of LBAs. Not all allocators use
157 * this functionality. The ms_size_tree should always contain the
158 * same number of segments as the ms_tree. The only difference
159 * is that the ms_size_tree is ordered by segment sizes.
161 avl_tree_t ms_size_tree;
162 uint64_t ms_lbas[MAX_LBAS];
164 metaslab_group_t *ms_group; /* metaslab group */
165 avl_node_t ms_group_node; /* node in metaslab group tree */
166 txg_node_t ms_txg_node; /* per-txg dirty metaslab links */
173 #endif /* _SYS_METASLAB_IMPL_H */