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, int);
68
69 typedef struct vdev_ops {
70 vdev_open_func_t *vdev_op_open;
71 vdev_close_func_t *vdev_op_close;
72 vdev_probe_func_t *vdev_op_probe;
73 vdev_asize_func_t *vdev_op_asize;
74 vdev_io_start_func_t *vdev_op_io_start;
75 vdev_io_done_func_t *vdev_op_io_done;
76 vdev_state_change_func_t *vdev_op_state_change;
77 char vdev_op_type[16];
78 boolean_t vdev_op_leaf;
79 } vdev_ops_t;
80
81 /*
82 * Virtual device properties
83 */
84 struct vdev_cache_entry {
85 char *ve_data;
86 uint64_t ve_offset;
87 uint64_t ve_lastused;
88 avl_node_t ve_offset_node;
89 avl_node_t ve_lastused_node;
90 uint32_t ve_hits;
91 uint16_t ve_missed_update;
92 zio_t *ve_fill_io;
93 };
94
95 struct vdev_cache {
96 avl_tree_t vc_offset_tree;
97 avl_tree_t vc_lastused_tree;
98 kmutex_t vc_lock;
99 };
100
101 struct vdev_queue {
102 avl_tree_t vq_deadline_tree;
103 avl_tree_t vq_read_tree;
104 avl_tree_t vq_write_tree;
105 avl_tree_t vq_pending_tree;
106 kmutex_t vq_lock;
107 };
108
109 /*
110 * Virtual device descriptor
111 */
112 struct vdev {
113 /*
114 * Common to all vdev types.
115 */
116 uint64_t vdev_id; /* child number in vdev parent */
117 uint64_t vdev_guid; /* unique ID for this vdev */
118 uint64_t vdev_guid_sum; /* self guid + all child guids */
119 uint64_t vdev_asize; /* allocatable device capacity */
120 uint64_t vdev_ashift; /* block alignment shift */
121 uint64_t vdev_state; /* see VDEV_STATE_* #defines */
122 uint64_t vdev_prevstate; /* used when reopening a vdev */
123 vdev_ops_t *vdev_ops; /* vdev operations */
124 spa_t *vdev_spa; /* spa for this vdev */
125 void *vdev_tsd; /* type-specific data */
126 vdev_t *vdev_top; /* top-level vdev */
127 vdev_t *vdev_parent; /* parent vdev */
128 vdev_t **vdev_child; /* array of children */
129 uint64_t vdev_children; /* number of children */
130 space_map_t vdev_dtl_map; /* dirty time log in-core state */
131 space_map_t vdev_dtl_scrub; /* DTL for scrub repair writes */
132 vdev_stat_t vdev_stat; /* virtual device statistics */
133
134 /*
135 * Top-level vdev state.
136 */
137 uint64_t vdev_ms_array; /* metaslab array object */
138 uint64_t vdev_ms_shift; /* metaslab size shift */
139 uint64_t vdev_ms_count; /* number of metaslabs */
140 metaslab_group_t *vdev_mg; /* metaslab group */
141 metaslab_t **vdev_ms; /* metaslab array */
142 txg_list_t vdev_ms_list; /* per-txg dirty metaslab lists */
143 txg_list_t vdev_dtl_list; /* per-txg dirty DTL lists */
144 txg_node_t vdev_txg_node; /* per-txg dirty vdev linkage */
145 boolean_t vdev_remove_wanted; /* async remove wanted? */
146 list_node_t vdev_dirty_node; /* config dirty list */
147 uint64_t vdev_deflate_ratio; /* deflation ratio (x512) */
148 uint64_t vdev_islog; /* is an intent log device */
149
150 /*
151 * Leaf vdev state.
152 */
153 uint64_t vdev_psize; /* physical device capacity */
154 space_map_obj_t vdev_dtl; /* dirty time log on-disk state */
155 txg_node_t vdev_dtl_node; /* per-txg dirty DTL linkage */
156 uint64_t vdev_wholedisk; /* true if this is a whole disk */
157 uint64_t vdev_offline; /* persistent offline state */
158 uint64_t vdev_faulted; /* persistent faulted state */
159 uint64_t vdev_degraded; /* persistent degraded state */
160 uint64_t vdev_removed; /* persistent removed state */
161 uint64_t vdev_nparity; /* number of parity devices for raidz */
162 char *vdev_path; /* vdev path (if any) */
163 char *vdev_devid; /* vdev devid (if any) */
164 char *vdev_physpath; /* vdev device path (if any) */
165 uint64_t vdev_fault_arg; /* fault injection paramater */
166 int vdev_fault_mask; /* zio types to fault */
167 uint8_t vdev_fault_mode; /* fault injection mode */
168 uint8_t vdev_tmpoffline; /* device taken offline temporarily? */
169 uint8_t vdev_detached; /* device detached? */
170 uint64_t vdev_isspare; /* was a hot spare */
171 uint64_t vdev_isl2cache; /* was a l2cache device */
172 vdev_queue_t vdev_queue; /* I/O deadline schedule queue */
173 vdev_cache_t vdev_cache; /* physical block cache */
174 uint64_t vdev_not_present; /* not present during import */
175 hrtime_t vdev_last_try; /* last reopen time */
176 boolean_t vdev_nowritecache; /* true if flushwritecache failed */
177 uint64_t vdev_unspare; /* unspare when resilvering done */
178 boolean_t vdev_checkremove; /* temporary online test */
179 boolean_t vdev_forcefault; /* force online fault */
180 boolean_t vdev_is_failing; /* device errors seen */
181
182 /*
183 * For DTrace to work in userland (libzpool) context, these fields must
184 * remain at the end of the structure. DTrace will use the kernel's
185 * CTF definition for 'struct vdev', and since the size of a kmutex_t is
186 * larger in userland, the offsets for the rest fields would be
187 * incorrect.
188 */
189 kmutex_t vdev_dtl_lock; /* vdev_dtl_{map,resilver} */
190 kmutex_t vdev_stat_lock; /* vdev_stat */
191 };
192
193 #define VDEV_SKIP_SIZE (8 << 10)
194 #define VDEV_BOOT_HEADER_SIZE (8 << 10)
195 #define VDEV_PHYS_SIZE (112 << 10)
196 #define VDEV_UBERBLOCK_RING (128 << 10)
197
198 #define VDEV_UBERBLOCK_SHIFT(vd) \
199 MAX((vd)->vdev_top->vdev_ashift, UBERBLOCK_SHIFT)
200 #define VDEV_UBERBLOCK_COUNT(vd) \
201 (VDEV_UBERBLOCK_RING >> VDEV_UBERBLOCK_SHIFT(vd))
202 #define VDEV_UBERBLOCK_OFFSET(vd, n) \
203 offsetof(vdev_label_t, vl_uberblock[(n) << VDEV_UBERBLOCK_SHIFT(vd)])
204 #define VDEV_UBERBLOCK_SIZE(vd) (1ULL << VDEV_UBERBLOCK_SHIFT(vd))
205
206 /* ZFS boot block */
207 #define VDEV_BOOT_MAGIC 0x2f5b007b10cULL
208 #define VDEV_BOOT_VERSION 1 /* version number */
209
210 typedef struct vdev_boot_header {
211 uint64_t vb_magic; /* VDEV_BOOT_MAGIC */
212 uint64_t vb_version; /* VDEV_BOOT_VERSION */
213 uint64_t vb_offset; /* start offset (bytes) */
214 uint64_t vb_size; /* size (bytes) */
215 char vb_pad[VDEV_BOOT_HEADER_SIZE - 4 * sizeof (uint64_t)];
216 } vdev_boot_header_t;
217
218 typedef struct vdev_phys {
219 char vp_nvlist[VDEV_PHYS_SIZE - sizeof (zio_block_tail_t)];
220 zio_block_tail_t vp_zbt;
221 } vdev_phys_t;
222
223 typedef struct vdev_label {
224 char vl_pad[VDEV_SKIP_SIZE]; /* 8K */
225 vdev_boot_header_t vl_boot_header; /* 8K */
226 vdev_phys_t vl_vdev_phys; /* 112K */
227 char vl_uberblock[VDEV_UBERBLOCK_RING]; /* 128K */
228 } vdev_label_t; /* 256K total */
229
230 /*
231 * vdev_dirty() flags
232 */
233 #define VDD_METASLAB 0x01
234 #define VDD_DTL 0x02
235
236 /*
237 * Size and offset of embedded boot loader region on each label.
238 * The total size of the first two labels plus the boot area is 4MB.
239 */
240 #define VDEV_BOOT_OFFSET (2 * sizeof (vdev_label_t))
241 #define VDEV_BOOT_SIZE (7ULL << 19) /* 3.5M */
242
243 /*
244 * Size of label regions at the start and end of each leaf device.
245 */
246 #define VDEV_LABEL_START_SIZE (2 * sizeof (vdev_label_t) + VDEV_BOOT_SIZE)
247 #define VDEV_LABEL_END_SIZE (2 * sizeof (vdev_label_t))
248 #define VDEV_LABELS 4
249
250 #define VDEV_ALLOC_LOAD 0
251 #define VDEV_ALLOC_ADD 1
252 #define VDEV_ALLOC_SPARE 2
253 #define VDEV_ALLOC_L2CACHE 3
254
255 /*
256 * Allocate or free a vdev
257 */
258 extern int vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *config,
259 vdev_t *parent, uint_t id, int alloctype);
260 extern void vdev_free(vdev_t *vd);
261
262 /*
263 * Add or remove children and parents
264 */
265 extern void vdev_add_child(vdev_t *pvd, vdev_t *cvd);
266 extern void vdev_remove_child(vdev_t *pvd, vdev_t *cvd);
267 extern void vdev_compact_children(vdev_t *pvd);
268 extern vdev_t *vdev_add_parent(vdev_t *cvd, vdev_ops_t *ops);
269 extern void vdev_remove_parent(vdev_t *cvd);
270
271 /*
272 * vdev sync load and sync
273 */
274 extern void vdev_load(vdev_t *vd);
275 extern void vdev_sync(vdev_t *vd, uint64_t txg);
276 extern void vdev_sync_done(vdev_t *vd, uint64_t txg);
277 extern void vdev_dirty(vdev_t *vd, int flags, void *arg, uint64_t txg);
278
279 /*
280 * Available vdev types.
281 */
282 extern vdev_ops_t vdev_root_ops;
283 extern vdev_ops_t vdev_mirror_ops;
284 extern vdev_ops_t vdev_replacing_ops;
285 extern vdev_ops_t vdev_raidz_ops;
286 extern vdev_ops_t vdev_disk_ops;
287 extern vdev_ops_t vdev_file_ops;
288 extern vdev_ops_t vdev_missing_ops;
289 extern vdev_ops_t vdev_spare_ops;
290
291 /*
292 * Common size functions
293 */
294 extern uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize);
295 extern uint64_t vdev_get_rsize(vdev_t *vd);
296
297 /*
298 * zdb uses this tunable, so it must be declared here to make lint happy.
299 */
300 extern int zfs_vdev_cache_size;
301
302 #ifdef __cplusplus
303 }
304 #endif
305
306 #endif /* _SYS_VDEV_IMPL_H */