]>
Commit | Line | Data |
---|---|---|
59ec819a NB |
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 (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 | * Copyright 2013 Nexenta Systems, Inc. All rights reserved. | |
26 | */ | |
27 | ||
28 | #ifndef _SYS_ARC_IMPL_H | |
29 | #define _SYS_ARC_IMPL_H | |
30 | ||
31 | #include <sys/arc.h> | |
32 | ||
33 | #ifdef __cplusplus | |
34 | extern "C" { | |
35 | #endif | |
36 | ||
37 | /* | |
38 | * Note that buffers can be in one of 6 states: | |
39 | * ARC_anon - anonymous (discussed below) | |
40 | * ARC_mru - recently used, currently cached | |
41 | * ARC_mru_ghost - recentely used, no longer in cache | |
42 | * ARC_mfu - frequently used, currently cached | |
43 | * ARC_mfu_ghost - frequently used, no longer in cache | |
44 | * ARC_l2c_only - exists in L2ARC but not other states | |
45 | * When there are no active references to the buffer, they are | |
46 | * are linked onto a list in one of these arc states. These are | |
47 | * the only buffers that can be evicted or deleted. Within each | |
48 | * state there are multiple lists, one for meta-data and one for | |
49 | * non-meta-data. Meta-data (indirect blocks, blocks of dnodes, | |
50 | * etc.) is tracked separately so that it can be managed more | |
51 | * explicitly: favored over data, limited explicitly. | |
52 | * | |
53 | * Anonymous buffers are buffers that are not associated with | |
54 | * a DVA. These are buffers that hold dirty block copies | |
55 | * before they are written to stable storage. By definition, | |
56 | * they are "ref'd" and are considered part of arc_mru | |
4e33ba4c | 57 | * that cannot be freed. Generally, they will acquire a DVA |
59ec819a NB |
58 | * as they are written and migrate onto the arc_mru list. |
59 | * | |
60 | * The ARC_l2c_only state is for buffers that are in the second | |
61 | * level ARC but no longer in any of the ARC_m* lists. The second | |
62 | * level ARC itself may also contain buffers that are in any of | |
63 | * the ARC_m* states - meaning that a buffer can exist in two | |
64 | * places. The reason for the ARC_l2c_only state is to keep the | |
65 | * buffer header in the hash table, so that reads that hit the | |
66 | * second level ARC benefit from these fast lookups. | |
67 | */ | |
68 | ||
69 | typedef struct arc_state { | |
ca0bf58d PS |
70 | /* |
71 | * list of evictable buffers | |
72 | */ | |
73 | multilist_t arcs_list[ARC_BUFC_NUMTYPES]; | |
74 | /* | |
75 | * total amount of evictable data in this state | |
76 | */ | |
d3c2ae1c | 77 | refcount_t arcs_esize[ARC_BUFC_NUMTYPES]; |
ca0bf58d PS |
78 | /* |
79 | * total amount of data in this state; this includes: evictable, | |
80 | * non-evictable, ARC_BUFC_DATA, and ARC_BUFC_METADATA. | |
81 | */ | |
36da08ef | 82 | refcount_t arcs_size; |
ca0bf58d PS |
83 | /* |
84 | * supports the "dbufs" kstat | |
85 | */ | |
59ec819a NB |
86 | arc_state_type_t arcs_state; |
87 | } arc_state_t; | |
88 | ||
59ec819a NB |
89 | typedef struct arc_callback arc_callback_t; |
90 | ||
91 | struct arc_callback { | |
92 | void *acb_private; | |
93 | arc_done_func_t *acb_done; | |
94 | arc_buf_t *acb_buf; | |
2aa34383 | 95 | boolean_t acb_compressed; |
59ec819a NB |
96 | zio_t *acb_zio_dummy; |
97 | arc_callback_t *acb_next; | |
98 | }; | |
99 | ||
100 | typedef struct arc_write_callback arc_write_callback_t; | |
101 | ||
102 | struct arc_write_callback { | |
103 | void *awcb_private; | |
104 | arc_done_func_t *awcb_ready; | |
bc77ba73 | 105 | arc_done_func_t *awcb_children_ready; |
59ec819a NB |
106 | arc_done_func_t *awcb_physdone; |
107 | arc_done_func_t *awcb_done; | |
108 | arc_buf_t *awcb_buf; | |
109 | }; | |
110 | ||
b9541d6b CW |
111 | /* |
112 | * ARC buffers are separated into multiple structs as a memory saving measure: | |
113 | * - Common fields struct, always defined, and embedded within it: | |
114 | * - L2-only fields, always allocated but undefined when not in L2ARC | |
115 | * - L1-only fields, only allocated when in L1ARC | |
116 | * | |
117 | * Buffer in L1 Buffer only in L2 | |
118 | * +------------------------+ +------------------------+ | |
119 | * | arc_buf_hdr_t | | arc_buf_hdr_t | | |
120 | * | | | | | |
121 | * | | | | | |
122 | * | | | | | |
123 | * +------------------------+ +------------------------+ | |
124 | * | l2arc_buf_hdr_t | | l2arc_buf_hdr_t | | |
125 | * | (undefined if L1-only) | | | | |
126 | * +------------------------+ +------------------------+ | |
127 | * | l1arc_buf_hdr_t | | |
128 | * | | | |
129 | * | | | |
130 | * | | | |
131 | * | | | |
132 | * +------------------------+ | |
133 | * | |
134 | * Because it's possible for the L2ARC to become extremely large, we can wind | |
135 | * up eating a lot of memory in L2ARC buffer headers, so the size of a header | |
136 | * is minimized by only allocating the fields necessary for an L1-cached buffer | |
137 | * when a header is actually in the L1 cache. The sub-headers (l1arc_buf_hdr and | |
138 | * l2arc_buf_hdr) are embedded rather than allocated separately to save a couple | |
139 | * words in pointers. arc_hdr_realloc() is used to switch a header between | |
140 | * these two allocation states. | |
141 | */ | |
142 | typedef struct l1arc_buf_hdr { | |
59ec819a | 143 | kmutex_t b_freeze_lock; |
d3c2ae1c | 144 | zio_cksum_t *b_freeze_cksum; |
59ec819a | 145 | |
59ec819a | 146 | arc_buf_t *b_buf; |
d3c2ae1c | 147 | uint32_t b_bufcnt; |
b9541d6b | 148 | /* for waiting on writes to complete */ |
59ec819a | 149 | kcondvar_t b_cv; |
d3c2ae1c | 150 | uint8_t b_byteswap; |
59ec819a | 151 | |
59ec819a NB |
152 | |
153 | /* protected by arc state mutex */ | |
154 | arc_state_t *b_state; | |
ca0bf58d | 155 | multilist_node_t b_arc_node; |
59ec819a NB |
156 | |
157 | /* updated atomically */ | |
158 | clock_t b_arc_access; | |
159 | uint32_t b_mru_hits; | |
160 | uint32_t b_mru_ghost_hits; | |
161 | uint32_t b_mfu_hits; | |
162 | uint32_t b_mfu_ghost_hits; | |
163 | uint32_t b_l2_hits; | |
164 | ||
165 | /* self protecting */ | |
166 | refcount_t b_refcnt; | |
167 | ||
b9541d6b | 168 | arc_callback_t *b_acb; |
a6255b7f | 169 | abd_t *b_pabd; |
b9541d6b | 170 | } l1arc_buf_hdr_t; |
59ec819a NB |
171 | |
172 | typedef struct l2arc_dev { | |
173 | vdev_t *l2ad_vdev; /* vdev */ | |
174 | spa_t *l2ad_spa; /* spa */ | |
175 | uint64_t l2ad_hand; /* next write location */ | |
176 | uint64_t l2ad_start; /* first addr on device */ | |
177 | uint64_t l2ad_end; /* last addr on device */ | |
59ec819a NB |
178 | boolean_t l2ad_first; /* first sweep through */ |
179 | boolean_t l2ad_writing; /* currently writing */ | |
b9541d6b CW |
180 | kmutex_t l2ad_mtx; /* lock for buffer list */ |
181 | list_t l2ad_buflist; /* buffer list */ | |
59ec819a | 182 | list_node_t l2ad_node; /* device list node */ |
d962d5da | 183 | refcount_t l2ad_alloc; /* allocated bytes */ |
59ec819a NB |
184 | } l2arc_dev_t; |
185 | ||
b9541d6b CW |
186 | typedef struct l2arc_buf_hdr { |
187 | /* protected by arc_buf_hdr mutex */ | |
188 | l2arc_dev_t *b_dev; /* L2ARC device */ | |
189 | uint64_t b_daddr; /* disk address, offset byte */ | |
b9541d6b | 190 | uint32_t b_hits; |
b9541d6b CW |
191 | |
192 | list_node_t b_l2node; | |
193 | } l2arc_buf_hdr_t; | |
194 | ||
49ee64e5 NB |
195 | typedef struct l2arc_write_callback { |
196 | l2arc_dev_t *l2wcb_dev; /* device info */ | |
197 | arc_buf_hdr_t *l2wcb_head; /* head of write buflist */ | |
198 | } l2arc_write_callback_t; | |
199 | ||
b9541d6b CW |
200 | struct arc_buf_hdr { |
201 | /* protected by hash lock */ | |
202 | dva_t b_dva; | |
203 | uint64_t b_birth; | |
b9541d6b | 204 | |
d3c2ae1c | 205 | arc_buf_contents_t b_type; |
b9541d6b CW |
206 | arc_buf_hdr_t *b_hash_next; |
207 | arc_flags_t b_flags; | |
208 | ||
d3c2ae1c GW |
209 | /* |
210 | * This field stores the size of the data buffer after | |
211 | * compression, and is set in the arc's zio completion handlers. | |
212 | * It is in units of SPA_MINBLOCKSIZE (e.g. 1 == 512 bytes). | |
213 | * | |
214 | * While the block pointers can store up to 32MB in their psize | |
215 | * field, we can only store up to 32MB minus 512B. This is due | |
216 | * to the bp using a bias of 1, whereas we use a bias of 0 (i.e. | |
217 | * a field of zeros represents 512B in the bp). We can't use a | |
218 | * bias of 1 since we need to reserve a psize of zero, here, to | |
219 | * represent holes and embedded blocks. | |
220 | * | |
221 | * This isn't a problem in practice, since the maximum size of a | |
222 | * buffer is limited to 16MB, so we never need to store 32MB in | |
223 | * this field. Even in the upstream illumos code base, the | |
224 | * maximum size of a buffer is limited to 16MB. | |
225 | */ | |
226 | uint16_t b_psize; | |
227 | ||
228 | /* | |
229 | * This field stores the size of the data buffer before | |
230 | * compression, and cannot change once set. It is in units | |
231 | * of SPA_MINBLOCKSIZE (e.g. 2 == 1024 bytes) | |
232 | */ | |
233 | uint16_t b_lsize; /* immutable */ | |
234 | uint64_t b_spa; /* immutable */ | |
b9541d6b CW |
235 | |
236 | /* L2ARC fields. Undefined when not in L2ARC. */ | |
237 | l2arc_buf_hdr_t b_l2hdr; | |
238 | /* L1ARC fields. Undefined when in l2arc_only state */ | |
239 | l1arc_buf_hdr_t b_l1hdr; | |
240 | }; | |
59ec819a NB |
241 | #ifdef __cplusplus |
242 | } | |
243 | #endif | |
244 | ||
245 | #endif /* _SYS_ARC_IMPL_H */ |