Print this page
expandable RAID-Z
| Split |
Close |
| Expand all |
| Collapse all |
--- old/usr/src/uts/common/fs/zfs/sys/vdev_impl.h
+++ new/usr/src/uts/common/fs/zfs/sys/vdev_impl.h
1 1 /*
2 2 * CDDL HEADER START
3 3 *
4 4 * The contents of this file are subject to the terms of the
5 5 * Common Development and Distribution License (the "License").
6 6 * You may not use this file except in compliance with the License.
7 7 *
8 8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9 9 * or http://www.opensolaris.org/os/licensing.
10 10 * See the License for the specific language governing permissions
11 11 * and limitations under the License.
12 12 *
13 13 * When distributing Covered Code, include this CDDL HEADER in each
14 14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 15 * If applicable, add the following below this CDDL HEADER, with the
16 16 * fields enclosed by brackets "[]" replaced with your own identifying
17 17 * information: Portions Copyright [yyyy] [name of copyright owner]
18 18 *
19 19 * CDDL HEADER END
20 20 */
21 21 /*
22 22 * Copyright 2007 Sun Microsystems, Inc. All rights reserved.
23 23 * Use is subject to license terms.
24 24 */
25 25
26 26 #ifndef _SYS_VDEV_IMPL_H
27 27 #define _SYS_VDEV_IMPL_H
28 28
29 29 #pragma ident "%Z%%M% %I% %E% SMI"
30 30
31 31 #include <sys/avl.h>
32 32 #include <sys/dmu.h>
33 33 #include <sys/metaslab.h>
34 34 #include <sys/nvpair.h>
35 35 #include <sys/space_map.h>
36 36 #include <sys/vdev.h>
37 37 #include <sys/dkio.h>
38 38 #include <sys/uberblock_impl.h>
39 39
40 40 #ifdef __cplusplus
41 41 extern "C" {
42 42 #endif
43 43
44 44 /*
45 45 * Virtual device descriptors.
46 46 *
47 47 * All storage pool operations go through the virtual device framework,
48 48 * which provides data replication and I/O scheduling.
49 49 */
50 50
51 51 /*
52 52 * Forward declarations that lots of things need.
53 53 */
54 54 typedef struct vdev_queue vdev_queue_t;
55 55 typedef struct vdev_cache vdev_cache_t;
56 56 typedef struct vdev_cache_entry vdev_cache_entry_t;
|
↓ open down ↓ |
56 lines elided |
↑ open up ↑ |
57 57
58 58 /*
59 59 * Virtual device operations
60 60 */
61 61 typedef int vdev_open_func_t(vdev_t *vd, uint64_t *size, uint64_t *ashift);
62 62 typedef void vdev_close_func_t(vdev_t *vd);
63 63 typedef int vdev_probe_func_t(vdev_t *vd);
64 64 typedef uint64_t vdev_asize_func_t(vdev_t *vd, uint64_t psize);
65 65 typedef int vdev_io_start_func_t(zio_t *zio);
66 66 typedef int vdev_io_done_func_t(zio_t *zio);
67 -typedef void vdev_state_change_func_t(vdev_t *vd, int, int);
67 +typedef void vdev_state_change_func_t(vdev_t *vd, int fauled, int degraded);
68 +typedef uint8_t vdev_grid_func_t(vdev_t *vd);
68 69
69 70 typedef struct vdev_ops {
70 71 vdev_open_func_t *vdev_op_open;
71 72 vdev_close_func_t *vdev_op_close;
72 73 vdev_probe_func_t *vdev_op_probe;
73 74 vdev_asize_func_t *vdev_op_asize;
74 75 vdev_io_start_func_t *vdev_op_io_start;
75 76 vdev_io_done_func_t *vdev_op_io_done;
76 77 vdev_state_change_func_t *vdev_op_state_change;
78 + vdev_grid_func_t *vdev_op_grid;
77 79 char vdev_op_type[16];
78 80 boolean_t vdev_op_leaf;
79 81 } vdev_ops_t;
80 82
81 83 /*
82 84 * Virtual device properties
83 85 */
84 86 struct vdev_cache_entry {
85 87 char *ve_data;
86 88 uint64_t ve_offset;
87 89 uint64_t ve_lastused;
88 90 avl_node_t ve_offset_node;
89 91 avl_node_t ve_lastused_node;
90 92 uint32_t ve_hits;
91 93 uint16_t ve_missed_update;
92 94 zio_t *ve_fill_io;
93 95 };
94 96
95 97 struct vdev_cache {
96 98 avl_tree_t vc_offset_tree;
97 99 avl_tree_t vc_lastused_tree;
98 100 kmutex_t vc_lock;
99 101 };
100 102
101 103 struct vdev_queue {
102 104 avl_tree_t vq_deadline_tree;
103 105 avl_tree_t vq_read_tree;
104 106 avl_tree_t vq_write_tree;
105 107 avl_tree_t vq_pending_tree;
106 108 kmutex_t vq_lock;
107 109 };
108 110
109 111 /*
110 112 * Virtual device descriptor
111 113 */
112 114 struct vdev {
113 115 /*
114 116 * Common to all vdev types.
115 117 */
116 118 uint64_t vdev_id; /* child number in vdev parent */
117 119 uint64_t vdev_guid; /* unique ID for this vdev */
118 120 uint64_t vdev_guid_sum; /* self guid + all child guids */
119 121 uint64_t vdev_asize; /* allocatable device capacity */
120 122 uint64_t vdev_ashift; /* block alignment shift */
121 123 uint64_t vdev_state; /* see VDEV_STATE_* #defines */
122 124 uint64_t vdev_prevstate; /* used when reopening a vdev */
123 125 vdev_ops_t *vdev_ops; /* vdev operations */
124 126 spa_t *vdev_spa; /* spa for this vdev */
125 127 void *vdev_tsd; /* type-specific data */
126 128 vdev_t *vdev_top; /* top-level vdev */
127 129 vdev_t *vdev_parent; /* parent vdev */
128 130 vdev_t **vdev_child; /* array of children */
129 131 uint64_t vdev_children; /* number of children */
130 132 space_map_t vdev_dtl_map; /* dirty time log in-core state */
131 133 space_map_t vdev_dtl_scrub; /* DTL for scrub repair writes */
132 134 vdev_stat_t vdev_stat; /* virtual device statistics */
133 135
134 136 /*
135 137 * Top-level vdev state.
136 138 */
137 139 uint64_t vdev_ms_array; /* metaslab array object */
138 140 uint64_t vdev_ms_shift; /* metaslab size shift */
139 141 uint64_t vdev_ms_count; /* number of metaslabs */
140 142 metaslab_group_t *vdev_mg; /* metaslab group */
141 143 metaslab_t **vdev_ms; /* metaslab array */
142 144 txg_list_t vdev_ms_list; /* per-txg dirty metaslab lists */
143 145 txg_list_t vdev_dtl_list; /* per-txg dirty DTL lists */
144 146 txg_node_t vdev_txg_node; /* per-txg dirty vdev linkage */
145 147 boolean_t vdev_remove_wanted; /* async remove wanted? */
146 148 list_node_t vdev_dirty_node; /* config dirty list */
147 149 uint64_t vdev_deflate_ratio; /* deflation ratio (x512) */
148 150 uint64_t vdev_islog; /* is an intent log device */
149 151
150 152 /*
151 153 * Leaf vdev state.
152 154 */
153 155 uint64_t vdev_psize; /* physical device capacity */
154 156 space_map_obj_t vdev_dtl; /* dirty time log on-disk state */
155 157 txg_node_t vdev_dtl_node; /* per-txg dirty DTL linkage */
156 158 uint64_t vdev_wholedisk; /* true if this is a whole disk */
157 159 uint64_t vdev_offline; /* persistent offline state */
158 160 uint64_t vdev_faulted; /* persistent faulted state */
159 161 uint64_t vdev_degraded; /* persistent degraded state */
160 162 uint64_t vdev_removed; /* persistent removed state */
161 163 uint64_t vdev_nparity; /* number of parity devices for raidz */
162 164 char *vdev_path; /* vdev path (if any) */
163 165 char *vdev_devid; /* vdev devid (if any) */
164 166 char *vdev_physpath; /* vdev device path (if any) */
165 167 uint64_t vdev_fault_arg; /* fault injection paramater */
166 168 int vdev_fault_mask; /* zio types to fault */
167 169 uint8_t vdev_fault_mode; /* fault injection mode */
168 170 uint8_t vdev_tmpoffline; /* device taken offline temporarily? */
169 171 uint8_t vdev_detached; /* device detached? */
170 172 uint64_t vdev_isspare; /* was a hot spare */
171 173 uint64_t vdev_isl2cache; /* was a l2cache device */
172 174 vdev_queue_t vdev_queue; /* I/O deadline schedule queue */
173 175 vdev_cache_t vdev_cache; /* physical block cache */
174 176 uint64_t vdev_not_present; /* not present during import */
175 177 hrtime_t vdev_last_try; /* last reopen time */
176 178 boolean_t vdev_nowritecache; /* true if flushwritecache failed */
177 179 uint64_t vdev_unspare; /* unspare when resilvering done */
178 180 boolean_t vdev_checkremove; /* temporary online test */
179 181 boolean_t vdev_forcefault; /* force online fault */
180 182 boolean_t vdev_is_failing; /* device errors seen */
181 183
182 184 /*
183 185 * For DTrace to work in userland (libzpool) context, these fields must
184 186 * remain at the end of the structure. DTrace will use the kernel's
185 187 * CTF definition for 'struct vdev', and since the size of a kmutex_t is
186 188 * larger in userland, the offsets for the rest fields would be
187 189 * incorrect.
188 190 */
189 191 kmutex_t vdev_dtl_lock; /* vdev_dtl_{map,resilver} */
190 192 kmutex_t vdev_stat_lock; /* vdev_stat */
191 193 };
192 194
193 195 #define VDEV_SKIP_SIZE (8 << 10)
194 196 #define VDEV_BOOT_HEADER_SIZE (8 << 10)
195 197 #define VDEV_PHYS_SIZE (112 << 10)
196 198 #define VDEV_UBERBLOCK_RING (128 << 10)
197 199
198 200 #define VDEV_UBERBLOCK_SHIFT(vd) \
199 201 MAX((vd)->vdev_top->vdev_ashift, UBERBLOCK_SHIFT)
200 202 #define VDEV_UBERBLOCK_COUNT(vd) \
201 203 (VDEV_UBERBLOCK_RING >> VDEV_UBERBLOCK_SHIFT(vd))
202 204 #define VDEV_UBERBLOCK_OFFSET(vd, n) \
203 205 offsetof(vdev_label_t, vl_uberblock[(n) << VDEV_UBERBLOCK_SHIFT(vd)])
204 206 #define VDEV_UBERBLOCK_SIZE(vd) (1ULL << VDEV_UBERBLOCK_SHIFT(vd))
205 207
206 208 /* ZFS boot block */
207 209 #define VDEV_BOOT_MAGIC 0x2f5b007b10cULL
208 210 #define VDEV_BOOT_VERSION 1 /* version number */
209 211
210 212 typedef struct vdev_boot_header {
211 213 uint64_t vb_magic; /* VDEV_BOOT_MAGIC */
212 214 uint64_t vb_version; /* VDEV_BOOT_VERSION */
213 215 uint64_t vb_offset; /* start offset (bytes) */
214 216 uint64_t vb_size; /* size (bytes) */
215 217 char vb_pad[VDEV_BOOT_HEADER_SIZE - 4 * sizeof (uint64_t)];
216 218 } vdev_boot_header_t;
217 219
218 220 typedef struct vdev_phys {
219 221 char vp_nvlist[VDEV_PHYS_SIZE - sizeof (zio_block_tail_t)];
220 222 zio_block_tail_t vp_zbt;
221 223 } vdev_phys_t;
222 224
223 225 typedef struct vdev_label {
224 226 char vl_pad[VDEV_SKIP_SIZE]; /* 8K */
225 227 vdev_boot_header_t vl_boot_header; /* 8K */
226 228 vdev_phys_t vl_vdev_phys; /* 112K */
227 229 char vl_uberblock[VDEV_UBERBLOCK_RING]; /* 128K */
228 230 } vdev_label_t; /* 256K total */
229 231
230 232 /*
231 233 * vdev_dirty() flags
232 234 */
233 235 #define VDD_METASLAB 0x01
234 236 #define VDD_DTL 0x02
235 237
236 238 /*
237 239 * Size and offset of embedded boot loader region on each label.
238 240 * The total size of the first two labels plus the boot area is 4MB.
239 241 */
240 242 #define VDEV_BOOT_OFFSET (2 * sizeof (vdev_label_t))
241 243 #define VDEV_BOOT_SIZE (7ULL << 19) /* 3.5M */
242 244
243 245 /*
244 246 * Size of label regions at the start and end of each leaf device.
245 247 */
246 248 #define VDEV_LABEL_START_SIZE (2 * sizeof (vdev_label_t) + VDEV_BOOT_SIZE)
247 249 #define VDEV_LABEL_END_SIZE (2 * sizeof (vdev_label_t))
248 250 #define VDEV_LABELS 4
249 251
250 252 #define VDEV_ALLOC_LOAD 0
251 253 #define VDEV_ALLOC_ADD 1
252 254 #define VDEV_ALLOC_SPARE 2
253 255 #define VDEV_ALLOC_L2CACHE 3
254 256
255 257 /*
256 258 * Allocate or free a vdev
257 259 */
258 260 extern int vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *config,
259 261 vdev_t *parent, uint_t id, int alloctype);
260 262 extern void vdev_free(vdev_t *vd);
261 263
262 264 /*
263 265 * Add or remove children and parents
264 266 */
265 267 extern void vdev_add_child(vdev_t *pvd, vdev_t *cvd);
266 268 extern void vdev_remove_child(vdev_t *pvd, vdev_t *cvd);
267 269 extern void vdev_compact_children(vdev_t *pvd);
268 270 extern vdev_t *vdev_add_parent(vdev_t *cvd, vdev_ops_t *ops);
269 271 extern void vdev_remove_parent(vdev_t *cvd);
270 272
271 273 /*
272 274 * vdev sync load and sync
273 275 */
274 276 extern void vdev_load(vdev_t *vd);
275 277 extern void vdev_sync(vdev_t *vd, uint64_t txg);
276 278 extern void vdev_sync_done(vdev_t *vd, uint64_t txg);
277 279 extern void vdev_dirty(vdev_t *vd, int flags, void *arg, uint64_t txg);
278 280
279 281 /*
280 282 * Available vdev types.
281 283 */
282 284 extern vdev_ops_t vdev_root_ops;
283 285 extern vdev_ops_t vdev_mirror_ops;
284 286 extern vdev_ops_t vdev_replacing_ops;
285 287 extern vdev_ops_t vdev_raidz_ops;
286 288 extern vdev_ops_t vdev_disk_ops;
287 289 extern vdev_ops_t vdev_file_ops;
288 290 extern vdev_ops_t vdev_missing_ops;
289 291 extern vdev_ops_t vdev_spare_ops;
290 292
291 293 /*
292 294 * Common size functions
293 295 */
294 296 extern uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize);
295 297 extern uint64_t vdev_get_rsize(vdev_t *vd);
296 298
297 299 /*
298 300 * zdb uses this tunable, so it must be declared here to make lint happy.
299 301 */
300 302 extern int zfs_vdev_cache_size;
301 303
302 304 #ifdef __cplusplus
303 305 }
304 306 #endif
305 307
306 308 #endif /* _SYS_VDEV_IMPL_H */
|
↓ open down ↓ |
220 lines elided |
↑ open up ↑ |
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX