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) 2012 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_io vdev_io_t
;
54 typedef struct vdev_cache vdev_cache_t
;
55 typedef struct vdev_cache_entry vdev_cache_entry_t
;
58 * Virtual device operations
60 typedef int vdev_open_func_t(vdev_t
*vd
, uint64_t *size
, uint64_t *max_size
,
62 typedef void vdev_close_func_t(vdev_t
*vd
);
63 typedef uint64_t vdev_asize_func_t(vdev_t
*vd
, uint64_t psize
);
64 typedef int vdev_io_start_func_t(zio_t
*zio
);
65 typedef void vdev_io_done_func_t(zio_t
*zio
);
66 typedef void vdev_state_change_func_t(vdev_t
*vd
, int, int);
67 typedef void vdev_hold_func_t(vdev_t
*vd
);
68 typedef void vdev_rele_func_t(vdev_t
*vd
);
70 typedef const struct vdev_ops
{
71 vdev_open_func_t
*vdev_op_open
;
72 vdev_close_func_t
*vdev_op_close
;
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 vdev_hold_func_t
*vdev_op_hold
;
78 vdev_rele_func_t
*vdev_op_rele
;
79 char vdev_op_type
[16];
80 boolean_t vdev_op_leaf
;
84 * Virtual device properties
86 struct vdev_cache_entry
{
90 avl_node_t ve_offset_node
;
91 avl_node_t ve_lastused_node
;
93 uint16_t ve_missed_update
;
98 avl_tree_t vc_offset_tree
;
99 avl_tree_t vc_lastused_tree
;
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 hrtime_t vq_io_complete_ts
;
109 hrtime_t vq_io_delta_ts
;
115 char vi_buffer
[SPA_MAXBLOCKSIZE
]; /* Must be first */
120 * Virtual device descriptor
124 * Common to all vdev types.
126 uint64_t vdev_id
; /* child number in vdev parent */
127 uint64_t vdev_guid
; /* unique ID for this vdev */
128 uint64_t vdev_guid_sum
; /* self guid + all child guids */
129 uint64_t vdev_orig_guid
; /* orig. guid prior to remove */
130 uint64_t vdev_asize
; /* allocatable device capacity */
131 uint64_t vdev_min_asize
; /* min acceptable asize */
132 uint64_t vdev_max_asize
; /* max acceptable asize */
133 uint64_t vdev_ashift
; /* block alignment shift */
134 uint64_t vdev_state
; /* see VDEV_STATE_* #defines */
135 uint64_t vdev_prevstate
; /* used when reopening a vdev */
136 vdev_ops_t
*vdev_ops
; /* vdev operations */
137 spa_t
*vdev_spa
; /* spa for this vdev */
138 void *vdev_tsd
; /* type-specific data */
139 vnode_t
*vdev_name_vp
; /* vnode for pathname */
140 vnode_t
*vdev_devid_vp
; /* vnode for devid */
141 vdev_t
*vdev_top
; /* top-level vdev */
142 vdev_t
*vdev_parent
; /* parent vdev */
143 vdev_t
**vdev_child
; /* array of children */
144 uint64_t vdev_children
; /* number of children */
145 space_map_t vdev_dtl
[DTL_TYPES
]; /* in-core dirty time logs */
146 vdev_stat_t vdev_stat
; /* virtual device statistics */
147 boolean_t vdev_expanding
; /* expand the vdev? */
148 boolean_t vdev_reopening
; /* reopen in progress? */
149 int vdev_open_error
; /* error on last open */
150 kthread_t
*vdev_open_thread
; /* thread opening children */
151 uint64_t vdev_crtxg
; /* txg when top-level was added */
154 * Top-level vdev state.
156 uint64_t vdev_ms_array
; /* metaslab array object */
157 uint64_t vdev_ms_shift
; /* metaslab size shift */
158 uint64_t vdev_ms_count
; /* number of metaslabs */
159 metaslab_group_t
*vdev_mg
; /* metaslab group */
160 metaslab_t
**vdev_ms
; /* metaslab array */
161 uint64_t vdev_pending_fastwrite
; /* allocated fastwrites */
162 txg_list_t vdev_ms_list
; /* per-txg dirty metaslab lists */
163 txg_list_t vdev_dtl_list
; /* per-txg dirty DTL lists */
164 txg_node_t vdev_txg_node
; /* per-txg dirty vdev linkage */
165 boolean_t vdev_remove_wanted
; /* async remove wanted? */
166 boolean_t vdev_probe_wanted
; /* async probe wanted? */
167 uint64_t vdev_removing
; /* device is being removed? */
168 list_node_t vdev_config_dirty_node
; /* config dirty list */
169 list_node_t vdev_state_dirty_node
; /* state dirty list */
170 uint64_t vdev_deflate_ratio
; /* deflation ratio (x512) */
171 uint64_t vdev_islog
; /* is an intent log device */
172 uint64_t vdev_ishole
; /* is a hole in the namespace */
177 uint64_t vdev_psize
; /* physical device capacity */
178 space_map_obj_t vdev_dtl_smo
; /* dirty time log space map obj */
179 txg_node_t vdev_dtl_node
; /* per-txg dirty DTL linkage */
180 uint64_t vdev_wholedisk
; /* true if this is a whole disk */
181 uint64_t vdev_offline
; /* persistent offline state */
182 uint64_t vdev_faulted
; /* persistent faulted state */
183 uint64_t vdev_degraded
; /* persistent degraded state */
184 uint64_t vdev_removed
; /* persistent removed state */
185 uint64_t vdev_resilvering
; /* persistent resilvering state */
186 uint64_t vdev_nparity
; /* number of parity devices for raidz */
187 char *vdev_path
; /* vdev path (if any) */
188 char *vdev_devid
; /* vdev devid (if any) */
189 char *vdev_physpath
; /* vdev device path (if any) */
190 char *vdev_fru
; /* physical FRU location */
191 uint64_t vdev_not_present
; /* not present during import */
192 uint64_t vdev_unspare
; /* unspare when resilvering done */
193 hrtime_t vdev_last_try
; /* last reopen time */
194 boolean_t vdev_nowritecache
; /* true if flushwritecache failed */
195 boolean_t vdev_checkremove
; /* temporary online test */
196 boolean_t vdev_forcefault
; /* force online fault */
197 boolean_t vdev_splitting
; /* split or repair in progress */
198 boolean_t vdev_delayed_close
; /* delayed device close? */
199 uint8_t vdev_tmpoffline
; /* device taken offline temporarily? */
200 uint8_t vdev_detached
; /* device detached? */
201 uint8_t vdev_cant_read
; /* vdev is failing all reads */
202 uint8_t vdev_cant_write
; /* vdev is failing all writes */
203 uint64_t vdev_isspare
; /* was a hot spare */
204 uint64_t vdev_isl2cache
; /* was a l2cache device */
205 vdev_queue_t vdev_queue
; /* I/O deadline schedule queue */
206 vdev_cache_t vdev_cache
; /* physical block cache */
207 spa_aux_vdev_t
*vdev_aux
; /* for l2cache vdevs */
208 zio_t
*vdev_probe_zio
; /* root of current probe */
209 vdev_aux_t vdev_label_aux
; /* on-disk aux state */
212 * For DTrace to work in userland (libzpool) context, these fields must
213 * remain at the end of the structure. DTrace will use the kernel's
214 * CTF definition for 'struct vdev', and since the size of a kmutex_t is
215 * larger in userland, the offsets for the rest of the fields would be
218 kmutex_t vdev_dtl_lock
; /* vdev_dtl_{map,resilver} */
219 kmutex_t vdev_stat_lock
; /* vdev_stat */
220 kmutex_t vdev_probe_lock
; /* protects vdev_probe_zio */
223 #define VDEV_RAIDZ_MAXPARITY 3
225 #define VDEV_PAD_SIZE (8 << 10)
226 /* 2 padding areas (vl_pad1 and vl_pad2) to skip */
227 #define VDEV_SKIP_SIZE VDEV_PAD_SIZE * 2
228 #define VDEV_PHYS_SIZE (112 << 10)
229 #define VDEV_UBERBLOCK_RING (128 << 10)
231 #define VDEV_UBERBLOCK_SHIFT(vd) \
232 MAX((vd)->vdev_top->vdev_ashift, UBERBLOCK_SHIFT)
233 #define VDEV_UBERBLOCK_COUNT(vd) \
234 (VDEV_UBERBLOCK_RING >> VDEV_UBERBLOCK_SHIFT(vd))
235 #define VDEV_UBERBLOCK_OFFSET(vd, n) \
236 offsetof(vdev_label_t, vl_uberblock[(n) << VDEV_UBERBLOCK_SHIFT(vd)])
237 #define VDEV_UBERBLOCK_SIZE(vd) (1ULL << VDEV_UBERBLOCK_SHIFT(vd))
239 typedef struct vdev_phys
{
240 char vp_nvlist
[VDEV_PHYS_SIZE
- sizeof (zio_eck_t
)];
244 typedef struct vdev_label
{
245 char vl_pad1
[VDEV_PAD_SIZE
]; /* 8K */
246 char vl_pad2
[VDEV_PAD_SIZE
]; /* 8K */
247 vdev_phys_t vl_vdev_phys
; /* 112K */
248 char vl_uberblock
[VDEV_UBERBLOCK_RING
]; /* 128K */
249 } vdev_label_t
; /* 256K total */
254 #define VDD_METASLAB 0x01
258 * Size and offset of embedded boot loader region on each label.
259 * The total size of the first two labels plus the boot area is 4MB.
261 #define VDEV_BOOT_OFFSET (2 * sizeof (vdev_label_t))
262 #define VDEV_BOOT_SIZE (7ULL << 19) /* 3.5M */
265 * Size of label regions at the start and end of each leaf device.
267 #define VDEV_LABEL_START_SIZE (2 * sizeof (vdev_label_t) + VDEV_BOOT_SIZE)
268 #define VDEV_LABEL_END_SIZE (2 * sizeof (vdev_label_t))
269 #define VDEV_LABELS 4
270 #define VDEV_BEST_LABEL VDEV_LABELS
272 #define VDEV_ALLOC_LOAD 0
273 #define VDEV_ALLOC_ADD 1
274 #define VDEV_ALLOC_SPARE 2
275 #define VDEV_ALLOC_L2CACHE 3
276 #define VDEV_ALLOC_ROOTPOOL 4
277 #define VDEV_ALLOC_SPLIT 5
278 #define VDEV_ALLOC_ATTACH 6
281 * Allocate or free a vdev
283 extern vdev_t
*vdev_alloc_common(spa_t
*spa
, uint_t id
, uint64_t guid
,
285 extern int vdev_alloc(spa_t
*spa
, vdev_t
**vdp
, nvlist_t
*config
,
286 vdev_t
*parent
, uint_t id
, int alloctype
);
287 extern void vdev_free(vdev_t
*vd
);
290 * Add or remove children and parents
292 extern void vdev_add_child(vdev_t
*pvd
, vdev_t
*cvd
);
293 extern void vdev_remove_child(vdev_t
*pvd
, vdev_t
*cvd
);
294 extern void vdev_compact_children(vdev_t
*pvd
);
295 extern vdev_t
*vdev_add_parent(vdev_t
*cvd
, vdev_ops_t
*ops
);
296 extern void vdev_remove_parent(vdev_t
*cvd
);
299 * vdev sync load and sync
301 extern void vdev_load_log_state(vdev_t
*nvd
, vdev_t
*ovd
);
302 extern boolean_t
vdev_log_state_valid(vdev_t
*vd
);
303 extern void vdev_load(vdev_t
*vd
);
304 extern void vdev_sync(vdev_t
*vd
, uint64_t txg
);
305 extern void vdev_sync_done(vdev_t
*vd
, uint64_t txg
);
306 extern void vdev_dirty(vdev_t
*vd
, int flags
, void *arg
, uint64_t txg
);
309 * Available vdev types.
311 extern vdev_ops_t vdev_root_ops
;
312 extern vdev_ops_t vdev_mirror_ops
;
313 extern vdev_ops_t vdev_replacing_ops
;
314 extern vdev_ops_t vdev_raidz_ops
;
315 extern vdev_ops_t vdev_disk_ops
;
316 extern vdev_ops_t vdev_file_ops
;
317 extern vdev_ops_t vdev_missing_ops
;
318 extern vdev_ops_t vdev_hole_ops
;
319 extern vdev_ops_t vdev_spare_ops
;
322 * Common size functions
324 extern uint64_t vdev_default_asize(vdev_t
*vd
, uint64_t psize
);
325 extern uint64_t vdev_get_min_asize(vdev_t
*vd
);
326 extern void vdev_set_min_asize(vdev_t
*vd
);
329 * zdb uses this tunable, so it must be declared here to make lint happy.
331 extern int zfs_vdev_cache_size
;
337 #endif /* _SYS_VDEV_IMPL_H */