]> git.proxmox.com Git - mirror_zfs.git/blob - include/sys/vdev_impl.h
Turn on/off enclosure slot fault LED even when disk isn't present
[mirror_zfs.git] / include / sys / vdev_impl.h
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) 2011, 2015 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 #include <sys/zfs_ratelimit.h>
38
39 #ifdef __cplusplus
40 extern "C" {
41 #endif
42
43 /*
44 * Virtual device descriptors.
45 *
46 * All storage pool operations go through the virtual device framework,
47 * which provides data replication and I/O scheduling.
48 */
49
50 /*
51 * Forward declarations that lots of things need.
52 */
53 typedef struct vdev_queue vdev_queue_t;
54 typedef struct vdev_cache vdev_cache_t;
55 typedef struct vdev_cache_entry vdev_cache_entry_t;
56
57 extern int zfs_vdev_queue_depth_pct;
58 extern uint32_t zfs_vdev_async_write_max_active;
59
60 /*
61 * Virtual device operations
62 */
63 typedef int vdev_open_func_t(vdev_t *vd, uint64_t *size, uint64_t *max_size,
64 uint64_t *ashift);
65 typedef void vdev_close_func_t(vdev_t *vd);
66 typedef uint64_t vdev_asize_func_t(vdev_t *vd, uint64_t psize);
67 typedef void vdev_io_start_func_t(zio_t *zio);
68 typedef void vdev_io_done_func_t(zio_t *zio);
69 typedef void vdev_state_change_func_t(vdev_t *vd, int, int);
70 typedef void vdev_hold_func_t(vdev_t *vd);
71 typedef void vdev_rele_func_t(vdev_t *vd);
72
73 typedef const struct vdev_ops {
74 vdev_open_func_t *vdev_op_open;
75 vdev_close_func_t *vdev_op_close;
76 vdev_asize_func_t *vdev_op_asize;
77 vdev_io_start_func_t *vdev_op_io_start;
78 vdev_io_done_func_t *vdev_op_io_done;
79 vdev_state_change_func_t *vdev_op_state_change;
80 vdev_hold_func_t *vdev_op_hold;
81 vdev_rele_func_t *vdev_op_rele;
82 char vdev_op_type[16];
83 boolean_t vdev_op_leaf;
84 } vdev_ops_t;
85
86 /*
87 * Virtual device properties
88 */
89 struct vdev_cache_entry {
90 char *ve_data;
91 uint64_t ve_offset;
92 clock_t ve_lastused;
93 avl_node_t ve_offset_node;
94 avl_node_t ve_lastused_node;
95 uint32_t ve_hits;
96 uint16_t ve_missed_update;
97 zio_t *ve_fill_io;
98 };
99
100 struct vdev_cache {
101 avl_tree_t vc_offset_tree;
102 avl_tree_t vc_lastused_tree;
103 kmutex_t vc_lock;
104 };
105
106 typedef struct vdev_queue_class {
107 uint32_t vqc_active;
108
109 /*
110 * Sorted by offset or timestamp, depending on if the queue is
111 * LBA-ordered vs FIFO.
112 */
113 avl_tree_t vqc_queued_tree;
114 } vdev_queue_class_t;
115
116 struct vdev_queue {
117 vdev_t *vq_vdev;
118 vdev_queue_class_t vq_class[ZIO_PRIORITY_NUM_QUEUEABLE];
119 avl_tree_t vq_active_tree;
120 avl_tree_t vq_read_offset_tree;
121 avl_tree_t vq_write_offset_tree;
122 uint64_t vq_last_offset;
123 hrtime_t vq_io_complete_ts; /* time last i/o completed */
124 hrtime_t vq_io_delta_ts;
125 zio_t vq_io_search; /* used as local for stack reduction */
126 kmutex_t vq_lock;
127 uint64_t vq_lastoffset;
128 };
129
130 /*
131 * Virtual device descriptor
132 */
133 struct vdev {
134 /*
135 * Common to all vdev types.
136 */
137 uint64_t vdev_id; /* child number in vdev parent */
138 uint64_t vdev_guid; /* unique ID for this vdev */
139 uint64_t vdev_guid_sum; /* self guid + all child guids */
140 uint64_t vdev_orig_guid; /* orig. guid prior to remove */
141 uint64_t vdev_asize; /* allocatable device capacity */
142 uint64_t vdev_min_asize; /* min acceptable asize */
143 uint64_t vdev_max_asize; /* max acceptable asize */
144 uint64_t vdev_ashift; /* block alignment shift */
145 uint64_t vdev_state; /* see VDEV_STATE_* #defines */
146 uint64_t vdev_prevstate; /* used when reopening a vdev */
147 vdev_ops_t *vdev_ops; /* vdev operations */
148 spa_t *vdev_spa; /* spa for this vdev */
149 void *vdev_tsd; /* type-specific data */
150 vnode_t *vdev_name_vp; /* vnode for pathname */
151 vnode_t *vdev_devid_vp; /* vnode for devid */
152 vdev_t *vdev_top; /* top-level vdev */
153 vdev_t *vdev_parent; /* parent vdev */
154 vdev_t **vdev_child; /* array of children */
155 uint64_t vdev_children; /* number of children */
156 vdev_stat_t vdev_stat; /* virtual device statistics */
157 vdev_stat_ex_t vdev_stat_ex; /* extended statistics */
158 boolean_t vdev_expanding; /* expand the vdev? */
159 boolean_t vdev_reopening; /* reopen in progress? */
160 boolean_t vdev_nonrot; /* true if solid state */
161 int vdev_open_error; /* error on last open */
162 kthread_t *vdev_open_thread; /* thread opening children */
163 uint64_t vdev_crtxg; /* txg when top-level was added */
164
165 /*
166 * Top-level vdev state.
167 */
168 uint64_t vdev_ms_array; /* metaslab array object */
169 uint64_t vdev_ms_shift; /* metaslab size shift */
170 uint64_t vdev_ms_count; /* number of metaslabs */
171 metaslab_group_t *vdev_mg; /* metaslab group */
172 metaslab_t **vdev_ms; /* metaslab array */
173 uint64_t vdev_pending_fastwrite; /* allocated fastwrites */
174 txg_list_t vdev_ms_list; /* per-txg dirty metaslab lists */
175 txg_list_t vdev_dtl_list; /* per-txg dirty DTL lists */
176 txg_node_t vdev_txg_node; /* per-txg dirty vdev linkage */
177 boolean_t vdev_remove_wanted; /* async remove wanted? */
178 boolean_t vdev_probe_wanted; /* async probe wanted? */
179 list_node_t vdev_config_dirty_node; /* config dirty list */
180 list_node_t vdev_state_dirty_node; /* state dirty list */
181 uint64_t vdev_deflate_ratio; /* deflation ratio (x512) */
182 uint64_t vdev_islog; /* is an intent log device */
183 uint64_t vdev_removing; /* device is being removed? */
184 boolean_t vdev_ishole; /* is a hole in the namespace */
185 kmutex_t vdev_queue_lock; /* protects vdev_queue_depth */
186 uint64_t vdev_top_zap;
187
188 /*
189 * The queue depth parameters determine how many async writes are
190 * still pending (i.e. allocated by net yet issued to disk) per
191 * top-level (vdev_async_write_queue_depth) and the maximum allowed
192 * (vdev_max_async_write_queue_depth). These values only apply to
193 * top-level vdevs.
194 */
195 uint64_t vdev_async_write_queue_depth;
196 uint64_t vdev_max_async_write_queue_depth;
197
198 /*
199 * Leaf vdev state.
200 */
201 range_tree_t *vdev_dtl[DTL_TYPES]; /* dirty time logs */
202 space_map_t *vdev_dtl_sm; /* dirty time log space map */
203 txg_node_t vdev_dtl_node; /* per-txg dirty DTL linkage */
204 uint64_t vdev_dtl_object; /* DTL object */
205 uint64_t vdev_psize; /* physical device capacity */
206 uint64_t vdev_wholedisk; /* true if this is a whole disk */
207 uint64_t vdev_offline; /* persistent offline state */
208 uint64_t vdev_faulted; /* persistent faulted state */
209 uint64_t vdev_degraded; /* persistent degraded state */
210 uint64_t vdev_removed; /* persistent removed state */
211 uint64_t vdev_resilver_txg; /* persistent resilvering state */
212 uint64_t vdev_nparity; /* number of parity devices for raidz */
213 char *vdev_path; /* vdev path (if any) */
214 char *vdev_devid; /* vdev devid (if any) */
215 char *vdev_physpath; /* vdev device path (if any) */
216 char *vdev_enc_sysfs_path; /* enclosure sysfs path */
217 char *vdev_fru; /* physical FRU location */
218 uint64_t vdev_not_present; /* not present during import */
219 uint64_t vdev_unspare; /* unspare when resilvering done */
220 boolean_t vdev_nowritecache; /* true if flushwritecache failed */
221 boolean_t vdev_checkremove; /* temporary online test */
222 boolean_t vdev_forcefault; /* force online fault */
223 boolean_t vdev_splitting; /* split or repair in progress */
224 boolean_t vdev_delayed_close; /* delayed device close? */
225 boolean_t vdev_tmpoffline; /* device taken offline temporarily? */
226 boolean_t vdev_detached; /* device detached? */
227 boolean_t vdev_cant_read; /* vdev is failing all reads */
228 boolean_t vdev_cant_write; /* vdev is failing all writes */
229 boolean_t vdev_isspare; /* was a hot spare */
230 boolean_t vdev_isl2cache; /* was a l2cache device */
231 vdev_queue_t vdev_queue; /* I/O deadline schedule queue */
232 vdev_cache_t vdev_cache; /* physical block cache */
233 spa_aux_vdev_t *vdev_aux; /* for l2cache and spares vdevs */
234 zio_t *vdev_probe_zio; /* root of current probe */
235 vdev_aux_t vdev_label_aux; /* on-disk aux state */
236 uint64_t vdev_leaf_zap;
237
238 /*
239 * For DTrace to work in userland (libzpool) context, these fields must
240 * remain at the end of the structure. DTrace will use the kernel's
241 * CTF definition for 'struct vdev', and since the size of a kmutex_t is
242 * larger in userland, the offsets for the rest of the fields would be
243 * incorrect.
244 */
245 kmutex_t vdev_dtl_lock; /* vdev_dtl_{map,resilver} */
246 kmutex_t vdev_stat_lock; /* vdev_stat */
247 kmutex_t vdev_probe_lock; /* protects vdev_probe_zio */
248
249 /*
250 * We rate limit ZIO delay and ZIO checksum events, since they
251 * can flood ZED with tons of events when a drive is acting up.
252 */
253 #define DELAYS_PER_SECOND 5
254 #define CHECKSUMS_PER_SECOND 5
255 zfs_ratelimit_t vdev_delay_rl;
256 zfs_ratelimit_t vdev_checksum_rl;
257 };
258
259 #define VDEV_RAIDZ_MAXPARITY 3
260
261 #define VDEV_PAD_SIZE (8 << 10)
262 /* 2 padding areas (vl_pad1 and vl_pad2) to skip */
263 #define VDEV_SKIP_SIZE VDEV_PAD_SIZE * 2
264 #define VDEV_PHYS_SIZE (112 << 10)
265 #define VDEV_UBERBLOCK_RING (128 << 10)
266
267 /* The largest uberblock we support is 8k. */
268 #define MAX_UBERBLOCK_SHIFT (13)
269 #define VDEV_UBERBLOCK_SHIFT(vd) \
270 MIN(MAX((vd)->vdev_top->vdev_ashift, UBERBLOCK_SHIFT), \
271 MAX_UBERBLOCK_SHIFT)
272 #define VDEV_UBERBLOCK_COUNT(vd) \
273 (VDEV_UBERBLOCK_RING >> VDEV_UBERBLOCK_SHIFT(vd))
274 #define VDEV_UBERBLOCK_OFFSET(vd, n) \
275 offsetof(vdev_label_t, vl_uberblock[(n) << VDEV_UBERBLOCK_SHIFT(vd)])
276 #define VDEV_UBERBLOCK_SIZE(vd) (1ULL << VDEV_UBERBLOCK_SHIFT(vd))
277
278 typedef struct vdev_phys {
279 char vp_nvlist[VDEV_PHYS_SIZE - sizeof (zio_eck_t)];
280 zio_eck_t vp_zbt;
281 } vdev_phys_t;
282
283 typedef struct vdev_label {
284 char vl_pad1[VDEV_PAD_SIZE]; /* 8K */
285 char vl_pad2[VDEV_PAD_SIZE]; /* 8K */
286 vdev_phys_t vl_vdev_phys; /* 112K */
287 char vl_uberblock[VDEV_UBERBLOCK_RING]; /* 128K */
288 } vdev_label_t; /* 256K total */
289
290 /*
291 * vdev_dirty() flags
292 */
293 #define VDD_METASLAB 0x01
294 #define VDD_DTL 0x02
295
296 /* Offset of embedded boot loader region on each label */
297 #define VDEV_BOOT_OFFSET (2 * sizeof (vdev_label_t))
298 /*
299 * Size of embedded boot loader region on each label.
300 * The total size of the first two labels plus the boot area is 4MB.
301 */
302 #define VDEV_BOOT_SIZE (7ULL << 19) /* 3.5M */
303
304 /*
305 * Size of label regions at the start and end of each leaf device.
306 */
307 #define VDEV_LABEL_START_SIZE (2 * sizeof (vdev_label_t) + VDEV_BOOT_SIZE)
308 #define VDEV_LABEL_END_SIZE (2 * sizeof (vdev_label_t))
309 #define VDEV_LABELS 4
310 #define VDEV_BEST_LABEL VDEV_LABELS
311
312 #define VDEV_ALLOC_LOAD 0
313 #define VDEV_ALLOC_ADD 1
314 #define VDEV_ALLOC_SPARE 2
315 #define VDEV_ALLOC_L2CACHE 3
316 #define VDEV_ALLOC_ROOTPOOL 4
317 #define VDEV_ALLOC_SPLIT 5
318 #define VDEV_ALLOC_ATTACH 6
319
320 /*
321 * Allocate or free a vdev
322 */
323 extern vdev_t *vdev_alloc_common(spa_t *spa, uint_t id, uint64_t guid,
324 vdev_ops_t *ops);
325 extern int vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *config,
326 vdev_t *parent, uint_t id, int alloctype);
327 extern void vdev_free(vdev_t *vd);
328
329 /*
330 * Add or remove children and parents
331 */
332 extern void vdev_add_child(vdev_t *pvd, vdev_t *cvd);
333 extern void vdev_remove_child(vdev_t *pvd, vdev_t *cvd);
334 extern void vdev_compact_children(vdev_t *pvd);
335 extern vdev_t *vdev_add_parent(vdev_t *cvd, vdev_ops_t *ops);
336 extern void vdev_remove_parent(vdev_t *cvd);
337
338 /*
339 * vdev sync load and sync
340 */
341 extern void vdev_load_log_state(vdev_t *nvd, vdev_t *ovd);
342 extern boolean_t vdev_log_state_valid(vdev_t *vd);
343 extern void vdev_load(vdev_t *vd);
344 extern int vdev_dtl_load(vdev_t *vd);
345 extern void vdev_sync(vdev_t *vd, uint64_t txg);
346 extern void vdev_sync_done(vdev_t *vd, uint64_t txg);
347 extern void vdev_dirty(vdev_t *vd, int flags, void *arg, uint64_t txg);
348 extern void vdev_dirty_leaves(vdev_t *vd, int flags, uint64_t txg);
349
350 /*
351 * Available vdev types.
352 */
353 extern vdev_ops_t vdev_root_ops;
354 extern vdev_ops_t vdev_mirror_ops;
355 extern vdev_ops_t vdev_replacing_ops;
356 extern vdev_ops_t vdev_raidz_ops;
357 extern vdev_ops_t vdev_disk_ops;
358 extern vdev_ops_t vdev_file_ops;
359 extern vdev_ops_t vdev_missing_ops;
360 extern vdev_ops_t vdev_hole_ops;
361 extern vdev_ops_t vdev_spare_ops;
362
363 /*
364 * Common size functions
365 */
366 extern uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize);
367 extern uint64_t vdev_get_min_asize(vdev_t *vd);
368 extern void vdev_set_min_asize(vdev_t *vd);
369
370 /*
371 * Global variables
372 */
373 /* zdb uses this tunable, so it must be declared here to make lint happy. */
374 extern int zfs_vdev_cache_size;
375
376 #ifdef __cplusplus
377 }
378 #endif
379
380 #endif /* _SYS_VDEV_IMPL_H */