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.
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.
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]
22 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
23 * Copyright (c) 2013 by Delphix. All rights reserved.
26 #ifndef _SYS_VDEV_IMPL_H
27 #define _SYS_VDEV_IMPL_H
31 #include <sys/metaslab.h>
32 #include <sys/nvpair.h>
33 #include <sys/space_map.h>
36 #include <sys/uberblock_impl.h>
43 * Virtual device descriptors.
45 * All storage pool operations go through the virtual device framework,
46 * which provides data replication and I/O scheduling.
50 * Forward declarations that lots of things need.
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;
57 * Virtual device operations
59 typedef int vdev_open_func_t(vdev_t *vd, uint64_t *size, uint64_t *max_size,
60 uint64_t *logical_ashift, uint64_t *physical_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);
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;
83 * Virtual device properties
85 struct vdev_cache_entry {
89 avl_node_t ve_offset_node;
90 avl_node_t ve_lastused_node;
92 uint16_t ve_missed_update;
97 avl_tree_t vc_offset_tree;
98 avl_tree_t vc_lastused_tree;
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;
112 * Virtual device descriptor
116 * Common to all vdev types.
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 */
127 * Logical block alignment shift
129 * The smallest sized/aligned I/O supported by the device.
131 uint64_t vdev_logical_ashift;
133 * Physical block alignment shift
135 * The device supports logical I/Os with vdev_logical_ashift
136 * size/alignment, but optimum performance will be achieved by
137 * aligning/sizing requests to vdev_physical_ashift. Smaller
138 * requests may be inflated or incur device level read-modify-write
141 * May be 0 to indicate no preference (i.e. use vdev_logical_ashift).
143 uint64_t vdev_physical_ashift;
144 uint64_t vdev_state; /* see VDEV_STATE_* #defines */
145 uint64_t vdev_prevstate; /* used when reopening a vdev */
146 vdev_ops_t *vdev_ops; /* vdev operations */
147 spa_t *vdev_spa; /* spa for this vdev */
148 void *vdev_tsd; /* type-specific data */
149 vnode_t *vdev_name_vp; /* vnode for pathname */
150 vnode_t *vdev_devid_vp; /* vnode for devid */
151 vdev_t *vdev_top; /* top-level vdev */
152 vdev_t *vdev_parent; /* parent vdev */
153 vdev_t **vdev_child; /* array of children */
154 uint64_t vdev_children; /* number of children */
155 space_map_t vdev_dtl[DTL_TYPES]; /* in-core dirty time logs */
156 vdev_stat_t vdev_stat; /* virtual device statistics */
157 boolean_t vdev_expanding; /* expand the vdev? */
158 boolean_t vdev_reopening; /* reopen in progress? */
159 int vdev_open_error; /* error on last open */
160 kthread_t *vdev_open_thread; /* thread opening children */
161 uint64_t vdev_crtxg; /* txg when top-level was added */
164 * Top-level vdev state.
166 uint64_t vdev_ms_array; /* metaslab array object */
167 uint64_t vdev_ms_shift; /* metaslab size shift */
168 uint64_t vdev_ms_count; /* number of metaslabs */
169 metaslab_group_t *vdev_mg; /* metaslab group */
170 metaslab_t **vdev_ms; /* metaslab array */
171 txg_list_t vdev_ms_list; /* per-txg dirty metaslab lists */
172 txg_list_t vdev_dtl_list; /* per-txg dirty DTL lists */
173 txg_node_t vdev_txg_node; /* per-txg dirty vdev linkage */
174 boolean_t vdev_remove_wanted; /* async remove wanted? */
175 boolean_t vdev_probe_wanted; /* async probe wanted? */
176 uint64_t vdev_removing; /* device is being removed? */
177 list_node_t vdev_config_dirty_node; /* config dirty list */
178 list_node_t vdev_state_dirty_node; /* state dirty list */
179 uint64_t vdev_deflate_ratio; /* deflation ratio (x512) */
180 uint64_t vdev_islog; /* is an intent log device */
181 uint64_t vdev_ishole; /* is a hole in the namespace */
186 uint64_t vdev_psize; /* physical device capacity */
187 space_map_obj_t vdev_dtl_smo; /* dirty time log space map obj */
188 txg_node_t vdev_dtl_node; /* per-txg dirty DTL linkage */
189 uint64_t vdev_wholedisk; /* true if this is a whole disk */
190 uint64_t vdev_offline; /* persistent offline state */
191 uint64_t vdev_faulted; /* persistent faulted state */
192 uint64_t vdev_degraded; /* persistent degraded state */
193 uint64_t vdev_removed; /* persistent removed state */
194 uint64_t vdev_resilver_txg; /* persistent resilvering state */
195 uint64_t vdev_nparity; /* number of parity devices for raidz */
196 char *vdev_path; /* vdev path (if any) */
197 char *vdev_devid; /* vdev devid (if any) */
198 char *vdev_physpath; /* vdev device path (if any) */
199 char *vdev_fru; /* physical FRU location */
200 uint64_t vdev_not_present; /* not present during import */
201 uint64_t vdev_unspare; /* unspare when resilvering done */
202 hrtime_t vdev_last_try; /* last reopen time */
203 boolean_t vdev_nowritecache; /* true if flushwritecache failed */
204 boolean_t vdev_notrim; /* true if trim failed */
205 boolean_t vdev_checkremove; /* temporary online test */
206 boolean_t vdev_forcefault; /* force online fault */
207 boolean_t vdev_splitting; /* split or repair in progress */
208 boolean_t vdev_delayed_close; /* delayed device close? */
209 uint8_t vdev_tmpoffline; /* device taken offline temporarily? */
210 uint8_t vdev_detached; /* device detached? */
211 uint8_t vdev_cant_read; /* vdev is failing all reads */
212 uint8_t vdev_cant_write; /* vdev is failing all writes */
213 uint64_t vdev_isspare; /* was a hot spare */
214 uint64_t vdev_isl2cache; /* was a l2cache device */
215 vdev_queue_t vdev_queue; /* I/O deadline schedule queue */
216 vdev_cache_t vdev_cache; /* physical block cache */
217 spa_aux_vdev_t *vdev_aux; /* for l2cache vdevs */
218 zio_t *vdev_probe_zio; /* root of current probe */
219 vdev_aux_t vdev_label_aux; /* on-disk aux state */
220 struct trim_map *vdev_trimmap;
223 * For DTrace to work in userland (libzpool) context, these fields must
224 * remain at the end of the structure. DTrace will use the kernel's
225 * CTF definition for 'struct vdev', and since the size of a kmutex_t is
226 * larger in userland, the offsets for the rest of the fields would be
229 kmutex_t vdev_dtl_lock; /* vdev_dtl_{map,resilver} */
230 kmutex_t vdev_stat_lock; /* vdev_stat */
231 kmutex_t vdev_probe_lock; /* protects vdev_probe_zio */
234 #define VDEV_RAIDZ_MAXPARITY 3
236 #define VDEV_PAD_SIZE (8 << 10)
237 /* 2 padding areas (vl_pad1 and vl_pad2) to skip */
238 #define VDEV_SKIP_SIZE VDEV_PAD_SIZE * 2
239 #define VDEV_PHYS_SIZE (112 << 10)
240 #define VDEV_UBERBLOCK_RING (128 << 10)
242 #define VDEV_UBERBLOCK_SHIFT(vd) \
243 MAX((vd)->vdev_top->vdev_ashift, UBERBLOCK_SHIFT)
244 #define VDEV_UBERBLOCK_COUNT(vd) \
245 (VDEV_UBERBLOCK_RING >> VDEV_UBERBLOCK_SHIFT(vd))
246 #define VDEV_UBERBLOCK_OFFSET(vd, n) \
247 offsetof(vdev_label_t, vl_uberblock[(n) << VDEV_UBERBLOCK_SHIFT(vd)])
248 #define VDEV_UBERBLOCK_SIZE(vd) (1ULL << VDEV_UBERBLOCK_SHIFT(vd))
250 typedef struct vdev_phys {
251 char vp_nvlist[VDEV_PHYS_SIZE - sizeof (zio_eck_t)];
255 typedef struct vdev_label {
256 char vl_pad1[VDEV_PAD_SIZE]; /* 8K */
257 char vl_pad2[VDEV_PAD_SIZE]; /* 8K */
258 vdev_phys_t vl_vdev_phys; /* 112K */
259 char vl_uberblock[VDEV_UBERBLOCK_RING]; /* 128K */
260 } vdev_label_t; /* 256K total */
265 #define VDD_METASLAB 0x01
268 /* Offset of embedded boot loader region on each label */
269 #define VDEV_BOOT_OFFSET (2 * sizeof (vdev_label_t))
271 * Size of embedded boot loader region on each label.
272 * The total size of the first two labels plus the boot area is 4MB.
274 #define VDEV_BOOT_SIZE (7ULL << 19) /* 3.5M */
277 * Size of label regions at the start and end of each leaf device.
279 #define VDEV_LABEL_START_SIZE (2 * sizeof (vdev_label_t) + VDEV_BOOT_SIZE)
280 #define VDEV_LABEL_END_SIZE (2 * sizeof (vdev_label_t))
281 #define VDEV_LABELS 4
282 #define VDEV_BEST_LABEL VDEV_LABELS
284 #define VDEV_ALLOC_LOAD 0
285 #define VDEV_ALLOC_ADD 1
286 #define VDEV_ALLOC_SPARE 2
287 #define VDEV_ALLOC_L2CACHE 3
288 #define VDEV_ALLOC_ROOTPOOL 4
289 #define VDEV_ALLOC_SPLIT 5
290 #define VDEV_ALLOC_ATTACH 6
293 * Allocate or free a vdev
295 extern vdev_t *vdev_alloc_common(spa_t *spa, uint_t id, uint64_t guid,
297 extern int vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *config,
298 vdev_t *parent, uint_t id, int alloctype);
299 extern void vdev_free(vdev_t *vd);
302 * Add or remove children and parents
304 extern void vdev_add_child(vdev_t *pvd, vdev_t *cvd);
305 extern void vdev_remove_child(vdev_t *pvd, vdev_t *cvd);
306 extern void vdev_compact_children(vdev_t *pvd);
307 extern vdev_t *vdev_add_parent(vdev_t *cvd, vdev_ops_t *ops);
308 extern void vdev_remove_parent(vdev_t *cvd);
311 * vdev sync load and sync
313 extern void vdev_load_log_state(vdev_t *nvd, vdev_t *ovd);
314 extern boolean_t vdev_log_state_valid(vdev_t *vd);
315 extern void vdev_load(vdev_t *vd);
316 extern void vdev_sync(vdev_t *vd, uint64_t txg);
317 extern void vdev_sync_done(vdev_t *vd, uint64_t txg);
318 extern void vdev_dirty(vdev_t *vd, int flags, void *arg, uint64_t txg);
321 * Available vdev types.
323 extern vdev_ops_t vdev_root_ops;
324 extern vdev_ops_t vdev_mirror_ops;
325 extern vdev_ops_t vdev_replacing_ops;
326 extern vdev_ops_t vdev_raidz_ops;
328 extern vdev_ops_t vdev_geom_ops;
330 extern vdev_ops_t vdev_disk_ops;
332 extern vdev_ops_t vdev_file_ops;
333 extern vdev_ops_t vdev_missing_ops;
334 extern vdev_ops_t vdev_hole_ops;
335 extern vdev_ops_t vdev_spare_ops;
338 * Common size functions
340 extern uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize);
341 extern uint64_t vdev_get_min_asize(vdev_t *vd);
342 extern void vdev_set_min_asize(vdev_t *vd);
347 /* zdb uses this tunable, so it must be declared here to make lint happy. */
348 extern int zfs_vdev_cache_size;
354 #endif /* _SYS_VDEV_IMPL_H */