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 2009 Sun Microsystems, Inc. All rights reserved.
23 * Use is subject to license terms.
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 *ashift);
60 typedef void vdev_close_func_t(vdev_t *vd);
61 typedef uint64_t vdev_asize_func_t(vdev_t *vd, uint64_t psize);
62 typedef int vdev_io_start_func_t(zio_t *zio);
63 typedef void vdev_io_done_func_t(zio_t *zio);
64 typedef void vdev_state_change_func_t(vdev_t *vd, int, int);
66 typedef struct vdev_ops {
67 vdev_open_func_t *vdev_op_open;
68 vdev_close_func_t *vdev_op_close;
69 vdev_asize_func_t *vdev_op_asize;
70 vdev_io_start_func_t *vdev_op_io_start;
71 vdev_io_done_func_t *vdev_op_io_done;
72 vdev_state_change_func_t *vdev_op_state_change;
73 char vdev_op_type[16];
74 boolean_t vdev_op_leaf;
78 * Virtual device properties
80 struct vdev_cache_entry {
84 avl_node_t ve_offset_node;
85 avl_node_t ve_lastused_node;
87 uint16_t ve_missed_update;
92 avl_tree_t vc_offset_tree;
93 avl_tree_t vc_lastused_tree;
98 avl_tree_t vq_deadline_tree;
99 avl_tree_t vq_read_tree;
100 avl_tree_t vq_write_tree;
101 avl_tree_t vq_pending_tree;
106 * Virtual device descriptor
110 * Common to all vdev types.
112 uint64_t vdev_id; /* child number in vdev parent */
113 uint64_t vdev_guid; /* unique ID for this vdev */
114 uint64_t vdev_guid_sum; /* self guid + all child guids */
115 uint64_t vdev_asize; /* allocatable device capacity */
116 uint64_t vdev_min_asize; /* min acceptable asize */
117 uint64_t vdev_ashift; /* block alignment shift */
118 uint64_t vdev_state; /* see VDEV_STATE_* #defines */
119 uint64_t vdev_prevstate; /* used when reopening a vdev */
120 vdev_ops_t *vdev_ops; /* vdev operations */
121 spa_t *vdev_spa; /* spa for this vdev */
122 void *vdev_tsd; /* type-specific data */
123 vdev_t *vdev_top; /* top-level vdev */
124 vdev_t *vdev_parent; /* parent vdev */
125 vdev_t **vdev_child; /* array of children */
126 uint64_t vdev_children; /* number of children */
127 space_map_t vdev_dtl[DTL_TYPES]; /* in-core dirty time logs */
128 vdev_stat_t vdev_stat; /* virtual device statistics */
129 boolean_t vdev_expanding; /* expand the vdev? */
132 * Top-level vdev state.
134 uint64_t vdev_ms_array; /* metaslab array object */
135 uint64_t vdev_ms_shift; /* metaslab size shift */
136 uint64_t vdev_ms_count; /* number of metaslabs */
137 metaslab_group_t *vdev_mg; /* metaslab group */
138 metaslab_t **vdev_ms; /* metaslab array */
139 txg_list_t vdev_ms_list; /* per-txg dirty metaslab lists */
140 txg_list_t vdev_dtl_list; /* per-txg dirty DTL lists */
141 txg_node_t vdev_txg_node; /* per-txg dirty vdev linkage */
142 boolean_t vdev_remove_wanted; /* async remove wanted? */
143 boolean_t vdev_probe_wanted; /* async probe wanted? */
144 list_node_t vdev_config_dirty_node; /* config dirty list */
145 list_node_t vdev_state_dirty_node; /* state dirty list */
146 uint64_t vdev_deflate_ratio; /* deflation ratio (x512) */
147 uint64_t vdev_islog; /* is an intent log device */
152 uint64_t vdev_psize; /* physical device capacity */
153 space_map_obj_t vdev_dtl_smo; /* dirty time log space map obj */
154 txg_node_t vdev_dtl_node; /* per-txg dirty DTL linkage */
155 uint64_t vdev_wholedisk; /* true if this is a whole disk */
156 uint64_t vdev_offline; /* persistent offline state */
157 uint64_t vdev_faulted; /* persistent faulted state */
158 uint64_t vdev_degraded; /* persistent degraded state */
159 uint64_t vdev_removed; /* persistent removed state */
160 uint64_t vdev_nparity; /* number of parity devices for raidz */
161 char *vdev_path; /* vdev path (if any) */
162 char *vdev_devid; /* vdev devid (if any) */
163 char *vdev_physpath; /* vdev device path (if any) */
164 char *vdev_fru; /* physical FRU location */
165 uint64_t vdev_not_present; /* not present during import */
166 uint64_t vdev_unspare; /* unspare when resilvering done */
167 hrtime_t vdev_last_try; /* last reopen time */
168 boolean_t vdev_nowritecache; /* true if flushwritecache failed */
169 boolean_t vdev_checkremove; /* temporary online test */
170 boolean_t vdev_forcefault; /* force online fault */
171 uint8_t vdev_tmpoffline; /* device taken offline temporarily? */
172 uint8_t vdev_detached; /* device detached? */
173 uint8_t vdev_cant_read; /* vdev is failing all reads */
174 uint8_t vdev_cant_write; /* vdev is failing all writes */
175 uint64_t vdev_isspare; /* was a hot spare */
176 uint64_t vdev_isl2cache; /* was a l2cache device */
177 vdev_queue_t vdev_queue; /* I/O deadline schedule queue */
178 vdev_cache_t vdev_cache; /* physical block cache */
179 spa_aux_vdev_t *vdev_aux; /* for l2cache vdevs */
180 zio_t *vdev_probe_zio; /* root of current probe */
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
189 kmutex_t vdev_dtl_lock; /* vdev_dtl_{map,resilver} */
190 kmutex_t vdev_stat_lock; /* vdev_stat */
191 kmutex_t vdev_probe_lock; /* protects vdev_probe_zio */
194 #define VDEV_PAD_SIZE (8 << 10)
195 /* 2 padding areas (vl_pad1 and vl_pad2) to skip */
196 #define VDEV_SKIP_SIZE VDEV_PAD_SIZE * 2
197 #define VDEV_PHYS_SIZE (112 << 10)
198 #define VDEV_UBERBLOCK_RING (128 << 10)
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))
208 typedef struct vdev_phys {
209 char vp_nvlist[VDEV_PHYS_SIZE - sizeof (zio_block_tail_t)];
210 zio_block_tail_t vp_zbt;
213 typedef struct vdev_label {
214 char vl_pad1[VDEV_PAD_SIZE]; /* 8K */
215 char vl_pad2[VDEV_PAD_SIZE]; /* 8K */
216 vdev_phys_t vl_vdev_phys; /* 112K */
217 char vl_uberblock[VDEV_UBERBLOCK_RING]; /* 128K */
218 } vdev_label_t; /* 256K total */
223 #define VDD_METASLAB 0x01
227 * Size and offset of embedded boot loader region on each label.
228 * The total size of the first two labels plus the boot area is 4MB.
230 #define VDEV_BOOT_OFFSET (2 * sizeof (vdev_label_t))
231 #define VDEV_BOOT_SIZE (7ULL << 19) /* 3.5M */
234 * Size of label regions at the start and end of each leaf device.
236 #define VDEV_LABEL_START_SIZE (2 * sizeof (vdev_label_t) + VDEV_BOOT_SIZE)
237 #define VDEV_LABEL_END_SIZE (2 * sizeof (vdev_label_t))
238 #define VDEV_LABELS 4
240 #define VDEV_ALLOC_LOAD 0
241 #define VDEV_ALLOC_ADD 1
242 #define VDEV_ALLOC_SPARE 2
243 #define VDEV_ALLOC_L2CACHE 3
244 #define VDEV_ALLOC_ROOTPOOL 4
247 * Allocate or free a vdev
249 extern int vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *config,
250 vdev_t *parent, uint_t id, int alloctype);
251 extern void vdev_free(vdev_t *vd);
254 * Add or remove children and parents
256 extern void vdev_add_child(vdev_t *pvd, vdev_t *cvd);
257 extern void vdev_remove_child(vdev_t *pvd, vdev_t *cvd);
258 extern void vdev_compact_children(vdev_t *pvd);
259 extern vdev_t *vdev_add_parent(vdev_t *cvd, vdev_ops_t *ops);
260 extern void vdev_remove_parent(vdev_t *cvd);
263 * vdev sync load and sync
265 extern void vdev_load_log_state(vdev_t *vd, nvlist_t *nv);
266 extern void vdev_load(vdev_t *vd);
267 extern void vdev_sync(vdev_t *vd, uint64_t txg);
268 extern void vdev_sync_done(vdev_t *vd, uint64_t txg);
269 extern void vdev_dirty(vdev_t *vd, int flags, void *arg, uint64_t txg);
272 * Available vdev types.
274 extern vdev_ops_t vdev_root_ops;
275 extern vdev_ops_t vdev_mirror_ops;
276 extern vdev_ops_t vdev_replacing_ops;
277 extern vdev_ops_t vdev_raidz_ops;
278 extern vdev_ops_t vdev_disk_ops;
279 extern vdev_ops_t vdev_file_ops;
280 extern vdev_ops_t vdev_missing_ops;
281 extern vdev_ops_t vdev_spare_ops;
284 * Common size functions
286 extern uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize);
287 extern uint64_t vdev_get_min_asize(vdev_t *vd);
288 extern void vdev_set_min_asize(vdev_t *vd);
291 * zdb uses this tunable, so it must be declared here to make lint happy.
293 extern int zfs_vdev_cache_size;
299 #endif /* _SYS_VDEV_IMPL_H */