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