]> git.proxmox.com Git - mirror_zfs.git/blame - include/sys/vdev_impl.h
Illumos #3956, #3957, #3958, #3959, #3960, #3961, #3962
[mirror_zfs.git] / include / sys / vdev_impl.h
CommitLineData
34dc7c2f
BB
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/*
428870ff 22 * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
5d1f7fb6 23 * Copyright (c) 2013 by Delphix. All rights reserved.
34dc7c2f
BB
24 */
25
26#ifndef _SYS_VDEV_IMPL_H
27#define _SYS_VDEV_IMPL_H
28
34dc7c2f
BB
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
39extern "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 */
52typedef struct vdev_queue vdev_queue_t;
86dd0fd9 53typedef struct vdev_io vdev_io_t;
34dc7c2f
BB
54typedef struct vdev_cache vdev_cache_t;
55typedef struct vdev_cache_entry vdev_cache_entry_t;
56
57/*
58 * Virtual device operations
59 */
1bd201e7
CS
60typedef int vdev_open_func_t(vdev_t *vd, uint64_t *size, uint64_t *max_size,
61 uint64_t *ashift);
34dc7c2f 62typedef void vdev_close_func_t(vdev_t *vd);
34dc7c2f
BB
63typedef uint64_t vdev_asize_func_t(vdev_t *vd, uint64_t psize);
64typedef int vdev_io_start_func_t(zio_t *zio);
b128c09f 65typedef void vdev_io_done_func_t(zio_t *zio);
34dc7c2f 66typedef void vdev_state_change_func_t(vdev_t *vd, int, int);
428870ff
BB
67typedef void vdev_hold_func_t(vdev_t *vd);
68typedef void vdev_rele_func_t(vdev_t *vd);
34dc7c2f 69
b01615d5 70typedef const struct vdev_ops {
34dc7c2f
BB
71 vdev_open_func_t *vdev_op_open;
72 vdev_close_func_t *vdev_op_close;
34dc7c2f
BB
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;
428870ff
BB
77 vdev_hold_func_t *vdev_op_hold;
78 vdev_rele_func_t *vdev_op_rele;
34dc7c2f
BB
79 char vdev_op_type[16];
80 boolean_t vdev_op_leaf;
81} vdev_ops_t;
82
83/*
84 * Virtual device properties
85 */
86struct vdev_cache_entry {
87 char *ve_data;
88 uint64_t ve_offset;
89 uint64_t ve_lastused;
90 avl_node_t ve_offset_node;
91 avl_node_t ve_lastused_node;
92 uint32_t ve_hits;
93 uint16_t ve_missed_update;
94 zio_t *ve_fill_io;
95};
96
97struct vdev_cache {
98 avl_tree_t vc_offset_tree;
99 avl_tree_t vc_lastused_tree;
100 kmutex_t vc_lock;
101};
102
103struct vdev_queue {
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;
cb682a17
MA
108 hrtime_t vq_io_complete_ts;
109 hrtime_t vq_io_delta_ts;
86dd0fd9 110 list_t vq_io_list;
34dc7c2f
BB
111 kmutex_t vq_lock;
112};
113
86dd0fd9
BB
114struct vdev_io {
115 char vi_buffer[SPA_MAXBLOCKSIZE]; /* Must be first */
116 list_node_t vi_node;
117};
118
34dc7c2f
BB
119/*
120 * Virtual device descriptor
121 */
122struct vdev {
123 /*
124 * Common to all vdev types.
125 */
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 */
428870ff 129 uint64_t vdev_orig_guid; /* orig. guid prior to remove */
34dc7c2f 130 uint64_t vdev_asize; /* allocatable device capacity */
9babb374 131 uint64_t vdev_min_asize; /* min acceptable asize */
1bd201e7 132 uint64_t vdev_max_asize; /* max acceptable asize */
34dc7c2f
BB
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 */
428870ff
BB
139 vnode_t *vdev_name_vp; /* vnode for pathname */
140 vnode_t *vdev_devid_vp; /* vnode for devid */
34dc7c2f
BB
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 */
fb5f0bc8 145 space_map_t vdev_dtl[DTL_TYPES]; /* in-core dirty time logs */
34dc7c2f 146 vdev_stat_t vdev_stat; /* virtual device statistics */
9babb374 147 boolean_t vdev_expanding; /* expand the vdev? */
428870ff 148 boolean_t vdev_reopening; /* reopen in progress? */
45d1cae3
BB
149 int vdev_open_error; /* error on last open */
150 kthread_t *vdev_open_thread; /* thread opening children */
428870ff 151 uint64_t vdev_crtxg; /* txg when top-level was added */
34dc7c2f
BB
152
153 /*
154 * Top-level vdev state.
155 */
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 */
920dd524 161 uint64_t vdev_pending_fastwrite; /* allocated fastwrites */
34dc7c2f
BB
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? */
b128c09f 166 boolean_t vdev_probe_wanted; /* async probe wanted? */
428870ff 167 uint64_t vdev_removing; /* device is being removed? */
b128c09f
BB
168 list_node_t vdev_config_dirty_node; /* config dirty list */
169 list_node_t vdev_state_dirty_node; /* state dirty list */
34dc7c2f
BB
170 uint64_t vdev_deflate_ratio; /* deflation ratio (x512) */
171 uint64_t vdev_islog; /* is an intent log device */
428870ff 172 uint64_t vdev_ishole; /* is a hole in the namespace */
34dc7c2f
BB
173
174 /*
175 * Leaf vdev state.
176 */
177 uint64_t vdev_psize; /* physical device capacity */
fb5f0bc8 178 space_map_obj_t vdev_dtl_smo; /* dirty time log space map obj */
34dc7c2f
BB
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 */
5d1f7fb6 185 uint64_t vdev_resilver_txg; /* persistent resilvering state */
34dc7c2f
BB
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) */
9babb374 190 char *vdev_fru; /* physical FRU location */
b128c09f
BB
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 */
428870ff
BB
197 boolean_t vdev_splitting; /* split or repair in progress */
198 boolean_t vdev_delayed_close; /* delayed device close? */
34dc7c2f
BB
199 uint8_t vdev_tmpoffline; /* device taken offline temporarily? */
200 uint8_t vdev_detached; /* device detached? */
b128c09f
BB
201 uint8_t vdev_cant_read; /* vdev is failing all reads */
202 uint8_t vdev_cant_write; /* vdev is failing all writes */
34dc7c2f
BB
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 */
b128c09f
BB
207 spa_aux_vdev_t *vdev_aux; /* for l2cache vdevs */
208 zio_t *vdev_probe_zio; /* root of current probe */
428870ff 209 vdev_aux_t vdev_label_aux; /* on-disk aux state */
34dc7c2f
BB
210
211 /*
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
9ae529ec 215 * larger in userland, the offsets for the rest of the fields would be
34dc7c2f
BB
216 * incorrect.
217 */
218 kmutex_t vdev_dtl_lock; /* vdev_dtl_{map,resilver} */
219 kmutex_t vdev_stat_lock; /* vdev_stat */
b128c09f 220 kmutex_t vdev_probe_lock; /* protects vdev_probe_zio */
34dc7c2f
BB
221};
222
428870ff
BB
223#define VDEV_RAIDZ_MAXPARITY 3
224
9babb374
BB
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
34dc7c2f
BB
228#define VDEV_PHYS_SIZE (112 << 10)
229#define VDEV_UBERBLOCK_RING (128 << 10)
230
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))
238
34dc7c2f 239typedef struct vdev_phys {
428870ff
BB
240 char vp_nvlist[VDEV_PHYS_SIZE - sizeof (zio_eck_t)];
241 zio_eck_t vp_zbt;
34dc7c2f
BB
242} vdev_phys_t;
243
244typedef struct vdev_label {
9babb374
BB
245 char vl_pad1[VDEV_PAD_SIZE]; /* 8K */
246 char vl_pad2[VDEV_PAD_SIZE]; /* 8K */
34dc7c2f
BB
247 vdev_phys_t vl_vdev_phys; /* 112K */
248 char vl_uberblock[VDEV_UBERBLOCK_RING]; /* 128K */
249} vdev_label_t; /* 256K total */
250
251/*
252 * vdev_dirty() flags
253 */
254#define VDD_METASLAB 0x01
255#define VDD_DTL 0x02
256
d3cc8b15
WA
257/* Offset of embedded boot loader region on each label */
258#define VDEV_BOOT_OFFSET (2 * sizeof (vdev_label_t))
34dc7c2f 259/*
d3cc8b15 260 * Size of embedded boot loader region on each label.
34dc7c2f
BB
261 * The total size of the first two labels plus the boot area is 4MB.
262 */
d3cc8b15 263#define VDEV_BOOT_SIZE (7ULL << 19) /* 3.5M */
34dc7c2f
BB
264
265/*
266 * Size of label regions at the start and end of each leaf device.
267 */
268#define VDEV_LABEL_START_SIZE (2 * sizeof (vdev_label_t) + VDEV_BOOT_SIZE)
269#define VDEV_LABEL_END_SIZE (2 * sizeof (vdev_label_t))
270#define VDEV_LABELS 4
9ae529ec 271#define VDEV_BEST_LABEL VDEV_LABELS
34dc7c2f
BB
272
273#define VDEV_ALLOC_LOAD 0
274#define VDEV_ALLOC_ADD 1
275#define VDEV_ALLOC_SPARE 2
276#define VDEV_ALLOC_L2CACHE 3
9babb374 277#define VDEV_ALLOC_ROOTPOOL 4
428870ff 278#define VDEV_ALLOC_SPLIT 5
5ffb9d1d 279#define VDEV_ALLOC_ATTACH 6
34dc7c2f
BB
280
281/*
282 * Allocate or free a vdev
283 */
428870ff
BB
284extern vdev_t *vdev_alloc_common(spa_t *spa, uint_t id, uint64_t guid,
285 vdev_ops_t *ops);
34dc7c2f
BB
286extern int vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *config,
287 vdev_t *parent, uint_t id, int alloctype);
288extern void vdev_free(vdev_t *vd);
289
290/*
291 * Add or remove children and parents
292 */
293extern void vdev_add_child(vdev_t *pvd, vdev_t *cvd);
294extern void vdev_remove_child(vdev_t *pvd, vdev_t *cvd);
295extern void vdev_compact_children(vdev_t *pvd);
296extern vdev_t *vdev_add_parent(vdev_t *cvd, vdev_ops_t *ops);
297extern void vdev_remove_parent(vdev_t *cvd);
298
299/*
300 * vdev sync load and sync
301 */
428870ff 302extern void vdev_load_log_state(vdev_t *nvd, vdev_t *ovd);
572e2857 303extern boolean_t vdev_log_state_valid(vdev_t *vd);
34dc7c2f
BB
304extern void vdev_load(vdev_t *vd);
305extern void vdev_sync(vdev_t *vd, uint64_t txg);
306extern void vdev_sync_done(vdev_t *vd, uint64_t txg);
307extern void vdev_dirty(vdev_t *vd, int flags, void *arg, uint64_t txg);
308
309/*
310 * Available vdev types.
311 */
312extern vdev_ops_t vdev_root_ops;
313extern vdev_ops_t vdev_mirror_ops;
314extern vdev_ops_t vdev_replacing_ops;
315extern vdev_ops_t vdev_raidz_ops;
316extern vdev_ops_t vdev_disk_ops;
317extern vdev_ops_t vdev_file_ops;
318extern vdev_ops_t vdev_missing_ops;
428870ff 319extern vdev_ops_t vdev_hole_ops;
34dc7c2f
BB
320extern vdev_ops_t vdev_spare_ops;
321
322/*
323 * Common size functions
324 */
325extern uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize);
9babb374
BB
326extern uint64_t vdev_get_min_asize(vdev_t *vd);
327extern void vdev_set_min_asize(vdev_t *vd);
34dc7c2f
BB
328
329/*
d3cc8b15 330 * Global variables
34dc7c2f 331 */
d3cc8b15 332/* zdb uses this tunable, so it must be declared here to make lint happy. */
34dc7c2f
BB
333extern int zfs_vdev_cache_size;
334
335#ifdef __cplusplus
336}
337#endif
338
339#endif /* _SYS_VDEV_IMPL_H */