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) 2012 by Delphix. All rights reserved.
24 */
25
26 #ifndef _SYS_VDEV_IMPL_H
27 #define _SYS_VDEV_IMPL_H
28
29 #include <sys/avl.h>
30 #include <sys/dmu.h>
31 #include <sys/metaslab.h>
32 #include <sys/nvpair.h>
33 #include <sys/space_map.h>
34 #include <sys/vdev.h>
35 #include <sys/dkio.h>
36 #include <sys/uberblock_impl.h>
37
38 #ifdef __cplusplus
39 extern "C" {
40 #endif
41
42 /*
43 * Virtual device descriptors.
44 *
45 * All storage pool operations go through the virtual device framework,
46 * which provides data replication and I/O scheduling.
47 */
48
49 /*
50 * Forward declarations that lots of things need.
51 */
52 typedef struct vdev_queue vdev_queue_t;
53 typedef struct vdev_cache vdev_cache_t;
54 typedef struct vdev_cache_entry vdev_cache_entry_t;
55
56 /*
57 * Virtual device operations
58 */
59 typedef int vdev_open_func_t(vdev_t *vd, uint64_t *size, uint64_t *max_size,
60 uint64_t *ashift);
61 typedef void vdev_close_func_t(vdev_t *vd);
62 typedef uint64_t vdev_asize_func_t(vdev_t *vd, uint64_t psize);
63 typedef int vdev_io_start_func_t(zio_t *zio);
64 typedef void vdev_io_done_func_t(zio_t *zio);
65 typedef void vdev_state_change_func_t(vdev_t *vd, int, int);
66 typedef void vdev_hold_func_t(vdev_t *vd);
67 typedef void vdev_rele_func_t(vdev_t *vd);
68
69 typedef struct vdev_ops {
70 vdev_open_func_t *vdev_op_open;
71 vdev_close_func_t *vdev_op_close;
72 vdev_asize_func_t *vdev_op_asize;
73 vdev_io_start_func_t *vdev_op_io_start;
74 vdev_io_done_func_t *vdev_op_io_done;
75 vdev_state_change_func_t *vdev_op_state_change;
76 vdev_hold_func_t *vdev_op_hold;
77 vdev_rele_func_t *vdev_op_rele;
78 char vdev_op_type[16];
79 boolean_t vdev_op_leaf;
80 } vdev_ops_t;
81
82 /*
83 * Virtual device properties
84 */
85 struct vdev_cache_entry {
86 char *ve_data;
87 uint64_t ve_offset;
88 uint64_t ve_lastused;
89 avl_node_t ve_offset_node;
90 avl_node_t ve_lastused_node;
91 uint32_t ve_hits;
92 uint16_t ve_missed_update;
93 zio_t *ve_fill_io;
94 };
95
96 struct vdev_cache {
97 avl_tree_t vc_offset_tree;
98 avl_tree_t vc_lastused_tree;
99 kmutex_t vc_lock;
100 };
101
102 struct vdev_queue {
103 avl_tree_t vq_deadline_tree;
104 avl_tree_t vq_read_tree;
105 avl_tree_t vq_write_tree;
106 avl_tree_t vq_pending_tree;
107 hrtime_t vq_io_complete_ts;
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_orig_guid; /* orig. guid prior to remove */
122 uint64_t vdev_asize; /* allocatable device capacity */
123 uint64_t vdev_min_asize; /* min acceptable asize */
124 uint64_t vdev_max_asize; /* max acceptable asize */
125 uint64_t vdev_ashift; /* block alignment shift */
126 uint64_t vdev_state; /* see VDEV_STATE_* #defines */
127 uint64_t vdev_prevstate; /* used when reopening a vdev */
128 vdev_ops_t *vdev_ops; /* vdev operations */
129 spa_t *vdev_spa; /* spa for this vdev */
130 void *vdev_tsd; /* type-specific data */
131 vnode_t *vdev_name_vp; /* vnode for pathname */
132 vnode_t *vdev_devid_vp; /* vnode for devid */
133 vdev_t *vdev_top; /* top-level vdev */
134 vdev_t *vdev_parent; /* parent vdev */
135 vdev_t **vdev_child; /* array of children */
136 uint64_t vdev_children; /* number of children */
137 space_map_t vdev_dtl[DTL_TYPES]; /* in-core dirty time logs */
138 vdev_stat_t vdev_stat; /* virtual device statistics */
139 boolean_t vdev_expanding; /* expand the vdev? */
140 boolean_t vdev_reopening; /* reopen in progress? */
141 int vdev_open_error; /* error on last open */
142 kthread_t *vdev_open_thread; /* thread opening children */
143 uint64_t vdev_crtxg; /* txg when top-level was added */
144
145 /*
146 * Top-level vdev state.
147 */
148 uint64_t vdev_ms_array; /* metaslab array object */
149 uint64_t vdev_ms_shift; /* metaslab size shift */
150 uint64_t vdev_ms_count; /* number of metaslabs */
151 metaslab_group_t *vdev_mg; /* metaslab group */
152 metaslab_t **vdev_ms; /* metaslab array */
153 txg_list_t vdev_ms_list; /* per-txg dirty metaslab lists */
154 txg_list_t vdev_dtl_list; /* per-txg dirty DTL lists */
155 txg_node_t vdev_txg_node; /* per-txg dirty vdev linkage */
156 boolean_t vdev_remove_wanted; /* async remove wanted? */
157 boolean_t vdev_probe_wanted; /* async probe wanted? */
158 uint64_t vdev_removing; /* device is being removed? */
159 list_node_t vdev_config_dirty_node; /* config dirty list */
160 list_node_t vdev_state_dirty_node; /* state dirty list */
161 uint64_t vdev_deflate_ratio; /* deflation ratio (x512) */
162 uint64_t vdev_islog; /* is an intent log device */
163 uint64_t vdev_ishole; /* is a hole in the namespace */
164
165 /*
166 * Leaf vdev state.
167 */
168 uint64_t vdev_psize; /* physical device capacity */
169 space_map_obj_t vdev_dtl_smo; /* dirty time log space map obj */
170 txg_node_t vdev_dtl_node; /* per-txg dirty DTL linkage */
171 uint64_t vdev_wholedisk; /* true if this is a whole disk */
172 uint64_t vdev_offline; /* persistent offline state */
173 uint64_t vdev_faulted; /* persistent faulted state */
174 uint64_t vdev_degraded; /* persistent degraded state */
175 uint64_t vdev_removed; /* persistent removed state */
176 uint64_t vdev_resilvering; /* persistent resilvering state */
177 uint64_t vdev_nparity; /* number of parity devices for raidz */
178 char *vdev_path; /* vdev path (if any) */
179 char *vdev_devid; /* vdev devid (if any) */
180 char *vdev_physpath; /* vdev device path (if any) */
181 char *vdev_fru; /* physical FRU location */
182 uint64_t vdev_not_present; /* not present during import */
183 uint64_t vdev_unspare; /* unspare when resilvering done */
184 hrtime_t vdev_last_try; /* last reopen time */
185 boolean_t vdev_nowritecache; /* true if flushwritecache failed */
186 boolean_t vdev_checkremove; /* temporary online test */
187 boolean_t vdev_forcefault; /* force online fault */
188 boolean_t vdev_splitting; /* split or repair in progress */
189 boolean_t vdev_delayed_close; /* delayed device close? */
190 uint8_t vdev_tmpoffline; /* device taken offline temporarily? */
191 uint8_t vdev_detached; /* device detached? */
192 uint8_t vdev_cant_read; /* vdev is failing all reads */
193 uint8_t vdev_cant_write; /* vdev is failing all writes */
194 uint64_t vdev_isspare; /* was a hot spare */
195 uint64_t vdev_isl2cache; /* was a l2cache device */
196 vdev_queue_t vdev_queue; /* I/O deadline schedule queue */
197 vdev_cache_t vdev_cache; /* physical block cache */
198 spa_aux_vdev_t *vdev_aux; /* for l2cache vdevs */
199 zio_t *vdev_probe_zio; /* root of current probe */
200 vdev_aux_t vdev_label_aux; /* on-disk aux state */
201
202 /*
203 * For DTrace to work in userland (libzpool) context, these fields must
204 * remain at the end of the structure. DTrace will use the kernel's
205 * CTF definition for 'struct vdev', and since the size of a kmutex_t is
206 * larger in userland, the offsets for the rest of the fields would be
207 * incorrect.
208 */
209 kmutex_t vdev_dtl_lock; /* vdev_dtl_{map,resilver} */
210 kmutex_t vdev_stat_lock; /* vdev_stat */
211 kmutex_t vdev_probe_lock; /* protects vdev_probe_zio */
212 };
213
214 #define VDEV_RAIDZ_MAXPARITY 3
215
216 #define VDEV_PAD_SIZE (8 << 10)
217 /* 2 padding areas (vl_pad1 and vl_pad2) to skip */
218 #define VDEV_SKIP_SIZE VDEV_PAD_SIZE * 2
219 #define VDEV_PHYS_SIZE (112 << 10)
220 #define VDEV_UBERBLOCK_RING (128 << 10)
221
222 #define VDEV_UBERBLOCK_SHIFT(vd) \
223 MAX((vd)->vdev_top->vdev_ashift, UBERBLOCK_SHIFT)
224 #define VDEV_UBERBLOCK_COUNT(vd) \
225 (VDEV_UBERBLOCK_RING >> VDEV_UBERBLOCK_SHIFT(vd))
226 #define VDEV_UBERBLOCK_OFFSET(vd, n) \
227 offsetof(vdev_label_t, vl_uberblock[(n) << VDEV_UBERBLOCK_SHIFT(vd)])
228 #define VDEV_UBERBLOCK_SIZE(vd) (1ULL << VDEV_UBERBLOCK_SHIFT(vd))
229
230 typedef struct vdev_phys {
231 char vp_nvlist[VDEV_PHYS_SIZE - sizeof (zio_eck_t)];
232 zio_eck_t vp_zbt;
233 } vdev_phys_t;
234
235 typedef struct vdev_label {
236 char vl_pad1[VDEV_PAD_SIZE]; /* 8K */
237 char vl_pad2[VDEV_PAD_SIZE]; /* 8K */
238 vdev_phys_t vl_vdev_phys; /* 112K */
239 char vl_uberblock[VDEV_UBERBLOCK_RING]; /* 128K */
240 } vdev_label_t; /* 256K total */
241
242 /*
243 * vdev_dirty() flags
244 */
245 #define VDD_METASLAB 0x01
246 #define VDD_DTL 0x02
247
248 /*
249 * Size and offset of embedded boot loader region on each label.
250 * The total size of the first two labels plus the boot area is 4MB.
251 */
252 #define VDEV_BOOT_OFFSET (2 * sizeof (vdev_label_t))
253 #define VDEV_BOOT_SIZE (7ULL << 19) /* 3.5M */
254
255 /*
256 * Size of label regions at the start and end of each leaf device.
257 */
258 #define VDEV_LABEL_START_SIZE (2 * sizeof (vdev_label_t) + VDEV_BOOT_SIZE)
259 #define VDEV_LABEL_END_SIZE (2 * sizeof (vdev_label_t))
260 #define VDEV_LABELS 4
261 #define VDEV_BEST_LABEL VDEV_LABELS
262
263 #define VDEV_ALLOC_LOAD 0
264 #define VDEV_ALLOC_ADD 1
265 #define VDEV_ALLOC_SPARE 2
266 #define VDEV_ALLOC_L2CACHE 3
267 #define VDEV_ALLOC_ROOTPOOL 4
268 #define VDEV_ALLOC_SPLIT 5
269 #define VDEV_ALLOC_ATTACH 6
270
271 /*
272 * Allocate or free a vdev
273 */
274 extern vdev_t *vdev_alloc_common(spa_t *spa, uint_t id, uint64_t guid,
275 vdev_ops_t *ops);
276 extern int vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *config,
277 vdev_t *parent, uint_t id, int alloctype);
278 extern void vdev_free(vdev_t *vd);
279
280 /*
281 * Add or remove children and parents
282 */
283 extern void vdev_add_child(vdev_t *pvd, vdev_t *cvd);
284 extern void vdev_remove_child(vdev_t *pvd, vdev_t *cvd);
285 extern void vdev_compact_children(vdev_t *pvd);
286 extern vdev_t *vdev_add_parent(vdev_t *cvd, vdev_ops_t *ops);
287 extern void vdev_remove_parent(vdev_t *cvd);
288
289 /*
290 * vdev sync load and sync
291 */
292 extern void vdev_load_log_state(vdev_t *nvd, vdev_t *ovd);
293 extern boolean_t vdev_log_state_valid(vdev_t *vd);
294 extern void vdev_load(vdev_t *vd);
295 extern void vdev_sync(vdev_t *vd, uint64_t txg);
296 extern void vdev_sync_done(vdev_t *vd, uint64_t txg);
297 extern void vdev_dirty(vdev_t *vd, int flags, void *arg, uint64_t txg);
298
299 /*
300 * Available vdev types.
301 */
302 extern vdev_ops_t vdev_root_ops;
303 extern vdev_ops_t vdev_mirror_ops;
304 extern vdev_ops_t vdev_replacing_ops;
305 extern vdev_ops_t vdev_raidz_ops;
306 extern vdev_ops_t vdev_disk_ops;
307 extern vdev_ops_t vdev_file_ops;
308 extern vdev_ops_t vdev_missing_ops;
309 extern vdev_ops_t vdev_hole_ops;
310 extern vdev_ops_t vdev_spare_ops;
311
312 /*
313 * Common size functions
314 */
315 extern uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize);
316 extern uint64_t vdev_get_min_asize(vdev_t *vd);
317 extern void vdev_set_min_asize(vdev_t *vd);
318
319 /*
320 * zdb uses this tunable, so it must be declared here to make lint happy.
321 */
322 extern int zfs_vdev_cache_size;
323
324 /*
325 * The vdev_buf_t is used to translate between zio_t and buf_t, and back again.
326 */
327 typedef struct vdev_buf {
328 buf_t vb_buf; /* buffer that describes the io */
329 zio_t *vb_io; /* pointer back to the original zio_t */
330 } vdev_buf_t;
331
332 #ifdef __cplusplus
333 }
334 #endif
335
336 #endif /* _SYS_VDEV_IMPL_H */