]> git.proxmox.com Git - mirror_zfs.git/blame - include/sys/vdev_impl.h
Move metaslab_group_alloc_update() call
[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;
0b75bdb3 89 clock_t ve_lastused;
34dc7c2f
BB
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
e8b96c60
MA
103typedef struct vdev_queue_class {
104 uint32_t vqc_active;
105
106 /*
107 * Sorted by offset or timestamp, depending on if the queue is
108 * LBA-ordered vs FIFO.
109 */
110 avl_tree_t vqc_queued_tree;
111} vdev_queue_class_t;
112
34dc7c2f 113struct vdev_queue {
e8b96c60
MA
114 vdev_t *vq_vdev;
115 vdev_queue_class_t vq_class[ZIO_PRIORITY_NUM_QUEUEABLE];
116 avl_tree_t vq_active_tree;
117 uint64_t vq_last_offset;
118 hrtime_t vq_io_complete_ts; /* time last i/o completed */
cb682a17 119 hrtime_t vq_io_delta_ts;
86dd0fd9 120 list_t vq_io_list;
34dc7c2f
BB
121 kmutex_t vq_lock;
122};
123
86dd0fd9
BB
124struct vdev_io {
125 char vi_buffer[SPA_MAXBLOCKSIZE]; /* Must be first */
126 list_node_t vi_node;
127};
128
34dc7c2f
BB
129/*
130 * Virtual device descriptor
131 */
132struct vdev {
133 /*
134 * Common to all vdev types.
135 */
136 uint64_t vdev_id; /* child number in vdev parent */
137 uint64_t vdev_guid; /* unique ID for this vdev */
138 uint64_t vdev_guid_sum; /* self guid + all child guids */
428870ff 139 uint64_t vdev_orig_guid; /* orig. guid prior to remove */
34dc7c2f 140 uint64_t vdev_asize; /* allocatable device capacity */
9babb374 141 uint64_t vdev_min_asize; /* min acceptable asize */
1bd201e7 142 uint64_t vdev_max_asize; /* max acceptable asize */
34dc7c2f
BB
143 uint64_t vdev_ashift; /* block alignment shift */
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 */
428870ff
BB
149 vnode_t *vdev_name_vp; /* vnode for pathname */
150 vnode_t *vdev_devid_vp; /* vnode for devid */
34dc7c2f
BB
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 */
fb5f0bc8 155 space_map_t vdev_dtl[DTL_TYPES]; /* in-core dirty time logs */
34dc7c2f 156 vdev_stat_t vdev_stat; /* virtual device statistics */
9babb374 157 boolean_t vdev_expanding; /* expand the vdev? */
428870ff 158 boolean_t vdev_reopening; /* reopen in progress? */
45d1cae3
BB
159 int vdev_open_error; /* error on last open */
160 kthread_t *vdev_open_thread; /* thread opening children */
428870ff 161 uint64_t vdev_crtxg; /* txg when top-level was added */
34dc7c2f
BB
162
163 /*
164 * Top-level vdev state.
165 */
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 */
920dd524 171 uint64_t vdev_pending_fastwrite; /* allocated fastwrites */
34dc7c2f
BB
172 txg_list_t vdev_ms_list; /* per-txg dirty metaslab lists */
173 txg_list_t vdev_dtl_list; /* per-txg dirty DTL lists */
174 txg_node_t vdev_txg_node; /* per-txg dirty vdev linkage */
175 boolean_t vdev_remove_wanted; /* async remove wanted? */
b128c09f 176 boolean_t vdev_probe_wanted; /* async probe wanted? */
428870ff 177 uint64_t vdev_removing; /* device is being removed? */
b128c09f
BB
178 list_node_t vdev_config_dirty_node; /* config dirty list */
179 list_node_t vdev_state_dirty_node; /* state dirty list */
34dc7c2f
BB
180 uint64_t vdev_deflate_ratio; /* deflation ratio (x512) */
181 uint64_t vdev_islog; /* is an intent log device */
428870ff 182 uint64_t vdev_ishole; /* is a hole in the namespace */
34dc7c2f
BB
183
184 /*
185 * Leaf vdev state.
186 */
187 uint64_t vdev_psize; /* physical device capacity */
fb5f0bc8 188 space_map_obj_t vdev_dtl_smo; /* dirty time log space map obj */
34dc7c2f
BB
189 txg_node_t vdev_dtl_node; /* per-txg dirty DTL linkage */
190 uint64_t vdev_wholedisk; /* true if this is a whole disk */
191 uint64_t vdev_offline; /* persistent offline state */
192 uint64_t vdev_faulted; /* persistent faulted state */
193 uint64_t vdev_degraded; /* persistent degraded state */
194 uint64_t vdev_removed; /* persistent removed state */
5d1f7fb6 195 uint64_t vdev_resilver_txg; /* persistent resilvering state */
34dc7c2f
BB
196 uint64_t vdev_nparity; /* number of parity devices for raidz */
197 char *vdev_path; /* vdev path (if any) */
198 char *vdev_devid; /* vdev devid (if any) */
199 char *vdev_physpath; /* vdev device path (if any) */
9babb374 200 char *vdev_fru; /* physical FRU location */
b128c09f
BB
201 uint64_t vdev_not_present; /* not present during import */
202 uint64_t vdev_unspare; /* unspare when resilvering done */
203 hrtime_t vdev_last_try; /* last reopen time */
204 boolean_t vdev_nowritecache; /* true if flushwritecache failed */
205 boolean_t vdev_checkremove; /* temporary online test */
206 boolean_t vdev_forcefault; /* force online fault */
428870ff
BB
207 boolean_t vdev_splitting; /* split or repair in progress */
208 boolean_t vdev_delayed_close; /* delayed device close? */
34dc7c2f
BB
209 uint8_t vdev_tmpoffline; /* device taken offline temporarily? */
210 uint8_t vdev_detached; /* device detached? */
b128c09f
BB
211 uint8_t vdev_cant_read; /* vdev is failing all reads */
212 uint8_t vdev_cant_write; /* vdev is failing all writes */
34dc7c2f
BB
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 */
b128c09f
BB
217 spa_aux_vdev_t *vdev_aux; /* for l2cache vdevs */
218 zio_t *vdev_probe_zio; /* root of current probe */
428870ff 219 vdev_aux_t vdev_label_aux; /* on-disk aux state */
34dc7c2f
BB
220
221 /*
222 * For DTrace to work in userland (libzpool) context, these fields must
223 * remain at the end of the structure. DTrace will use the kernel's
224 * CTF definition for 'struct vdev', and since the size of a kmutex_t is
9ae529ec 225 * larger in userland, the offsets for the rest of the fields would be
34dc7c2f
BB
226 * incorrect.
227 */
228 kmutex_t vdev_dtl_lock; /* vdev_dtl_{map,resilver} */
229 kmutex_t vdev_stat_lock; /* vdev_stat */
b128c09f 230 kmutex_t vdev_probe_lock; /* protects vdev_probe_zio */
34dc7c2f
BB
231};
232
428870ff
BB
233#define VDEV_RAIDZ_MAXPARITY 3
234
9babb374
BB
235#define VDEV_PAD_SIZE (8 << 10)
236/* 2 padding areas (vl_pad1 and vl_pad2) to skip */
237#define VDEV_SKIP_SIZE VDEV_PAD_SIZE * 2
34dc7c2f
BB
238#define VDEV_PHYS_SIZE (112 << 10)
239#define VDEV_UBERBLOCK_RING (128 << 10)
240
241#define VDEV_UBERBLOCK_SHIFT(vd) \
242 MAX((vd)->vdev_top->vdev_ashift, UBERBLOCK_SHIFT)
243#define VDEV_UBERBLOCK_COUNT(vd) \
244 (VDEV_UBERBLOCK_RING >> VDEV_UBERBLOCK_SHIFT(vd))
245#define VDEV_UBERBLOCK_OFFSET(vd, n) \
246 offsetof(vdev_label_t, vl_uberblock[(n) << VDEV_UBERBLOCK_SHIFT(vd)])
247#define VDEV_UBERBLOCK_SIZE(vd) (1ULL << VDEV_UBERBLOCK_SHIFT(vd))
248
34dc7c2f 249typedef struct vdev_phys {
428870ff
BB
250 char vp_nvlist[VDEV_PHYS_SIZE - sizeof (zio_eck_t)];
251 zio_eck_t vp_zbt;
34dc7c2f
BB
252} vdev_phys_t;
253
254typedef struct vdev_label {
9babb374
BB
255 char vl_pad1[VDEV_PAD_SIZE]; /* 8K */
256 char vl_pad2[VDEV_PAD_SIZE]; /* 8K */
34dc7c2f
BB
257 vdev_phys_t vl_vdev_phys; /* 112K */
258 char vl_uberblock[VDEV_UBERBLOCK_RING]; /* 128K */
259} vdev_label_t; /* 256K total */
260
261/*
262 * vdev_dirty() flags
263 */
264#define VDD_METASLAB 0x01
265#define VDD_DTL 0x02
266
d3cc8b15
WA
267/* Offset of embedded boot loader region on each label */
268#define VDEV_BOOT_OFFSET (2 * sizeof (vdev_label_t))
34dc7c2f 269/*
d3cc8b15 270 * Size of embedded boot loader region on each label.
34dc7c2f
BB
271 * The total size of the first two labels plus the boot area is 4MB.
272 */
d3cc8b15 273#define VDEV_BOOT_SIZE (7ULL << 19) /* 3.5M */
34dc7c2f
BB
274
275/*
276 * Size of label regions at the start and end of each leaf device.
277 */
278#define VDEV_LABEL_START_SIZE (2 * sizeof (vdev_label_t) + VDEV_BOOT_SIZE)
279#define VDEV_LABEL_END_SIZE (2 * sizeof (vdev_label_t))
280#define VDEV_LABELS 4
9ae529ec 281#define VDEV_BEST_LABEL VDEV_LABELS
34dc7c2f
BB
282
283#define VDEV_ALLOC_LOAD 0
284#define VDEV_ALLOC_ADD 1
285#define VDEV_ALLOC_SPARE 2
286#define VDEV_ALLOC_L2CACHE 3
9babb374 287#define VDEV_ALLOC_ROOTPOOL 4
428870ff 288#define VDEV_ALLOC_SPLIT 5
5ffb9d1d 289#define VDEV_ALLOC_ATTACH 6
34dc7c2f
BB
290
291/*
292 * Allocate or free a vdev
293 */
428870ff
BB
294extern vdev_t *vdev_alloc_common(spa_t *spa, uint_t id, uint64_t guid,
295 vdev_ops_t *ops);
34dc7c2f
BB
296extern int vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *config,
297 vdev_t *parent, uint_t id, int alloctype);
298extern void vdev_free(vdev_t *vd);
299
300/*
301 * Add or remove children and parents
302 */
303extern void vdev_add_child(vdev_t *pvd, vdev_t *cvd);
304extern void vdev_remove_child(vdev_t *pvd, vdev_t *cvd);
305extern void vdev_compact_children(vdev_t *pvd);
306extern vdev_t *vdev_add_parent(vdev_t *cvd, vdev_ops_t *ops);
307extern void vdev_remove_parent(vdev_t *cvd);
308
309/*
310 * vdev sync load and sync
311 */
428870ff 312extern void vdev_load_log_state(vdev_t *nvd, vdev_t *ovd);
572e2857 313extern boolean_t vdev_log_state_valid(vdev_t *vd);
34dc7c2f
BB
314extern void vdev_load(vdev_t *vd);
315extern void vdev_sync(vdev_t *vd, uint64_t txg);
316extern void vdev_sync_done(vdev_t *vd, uint64_t txg);
317extern void vdev_dirty(vdev_t *vd, int flags, void *arg, uint64_t txg);
318
319/*
320 * Available vdev types.
321 */
322extern vdev_ops_t vdev_root_ops;
323extern vdev_ops_t vdev_mirror_ops;
324extern vdev_ops_t vdev_replacing_ops;
325extern vdev_ops_t vdev_raidz_ops;
326extern vdev_ops_t vdev_disk_ops;
327extern vdev_ops_t vdev_file_ops;
328extern vdev_ops_t vdev_missing_ops;
428870ff 329extern vdev_ops_t vdev_hole_ops;
34dc7c2f
BB
330extern vdev_ops_t vdev_spare_ops;
331
332/*
333 * Common size functions
334 */
335extern uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize);
9babb374
BB
336extern uint64_t vdev_get_min_asize(vdev_t *vd);
337extern void vdev_set_min_asize(vdev_t *vd);
34dc7c2f
BB
338
339/*
d3cc8b15 340 * Global variables
34dc7c2f 341 */
d3cc8b15 342/* zdb uses this tunable, so it must be declared here to make lint happy. */
34dc7c2f
BB
343extern int zfs_vdev_cache_size;
344
345#ifdef __cplusplus
346}
347#endif
348
349#endif /* _SYS_VDEV_IMPL_H */