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, 2018 by Delphix. All rights reserved.
24 * Copyright (c) 2011, 2019 by Delphix. All rights reserved.
25 * Copyright (c) 2017, Intel Corporation.
26 * Copyright 2019 Joyent, Inc.
27 */
28
29#ifndef _SYS_VDEV_IMPL_H
30#define	_SYS_VDEV_IMPL_H
31
32#include <sys/avl.h>
33#include <sys/bpobj.h>
34#include <sys/dmu.h>
35#include <sys/metaslab.h>
36#include <sys/nvpair.h>
37#include <sys/space_map.h>
38#include <sys/vdev.h>
39#include <sys/dkio.h>
40#include <sys/uberblock_impl.h>
41#include <sys/vdev_indirect_mapping.h>
42#include <sys/vdev_indirect_births.h>
43#include <sys/vdev_removal.h>
44
45#ifdef	__cplusplus
46extern "C" {
47#endif
48
49/*
50 * Virtual device descriptors.
51 *
52 * All storage pool operations go through the virtual device framework,
53 * which provides data replication and I/O scheduling.
54 */
55
56/*
57 * Forward declarations that lots of things need.
58 */
59typedef struct vdev_queue vdev_queue_t;
60typedef struct vdev_cache vdev_cache_t;
61typedef struct vdev_cache_entry vdev_cache_entry_t;
62struct abd;
63
64extern int zfs_vdev_queue_depth_pct;
65extern int zfs_vdev_def_queue_depth;
66extern uint32_t zfs_vdev_async_write_max_active;
67
68/*
69 * Virtual device operations
70 */
71typedef int	vdev_open_func_t(vdev_t *vd, uint64_t *size, uint64_t *max_size,
72    uint64_t *ashift);
73typedef void	vdev_close_func_t(vdev_t *vd);
74typedef uint64_t vdev_asize_func_t(vdev_t *vd, uint64_t psize);
75typedef void	vdev_io_start_func_t(zio_t *zio);
76typedef void	vdev_io_done_func_t(zio_t *zio);
77typedef void	vdev_state_change_func_t(vdev_t *vd, int, int);
78typedef boolean_t vdev_need_resilver_func_t(vdev_t *vd, uint64_t, size_t);
79typedef void	vdev_hold_func_t(vdev_t *vd);
80typedef void	vdev_rele_func_t(vdev_t *vd);
81
82typedef void	vdev_remap_cb_t(uint64_t inner_offset, vdev_t *vd,
83    uint64_t offset, uint64_t size, void *arg);
84typedef void	vdev_remap_func_t(vdev_t *vd, uint64_t offset, uint64_t size,
85    vdev_remap_cb_t callback, void *arg);
86typedef int	vdev_dumpio_func_t(vdev_t *vd, caddr_t data, size_t size,
87    uint64_t offset, uint64_t origoffset, boolean_t doread, boolean_t isdump);
88/*
89 * Given a target vdev, translates the logical range "in" to the physical
90 * range "res"
91 */
92typedef void vdev_xlation_func_t(vdev_t *cvd, const range_seg64_t *in,
93    range_seg64_t *res);
94
95typedef struct vdev_ops {
96	vdev_open_func_t		*vdev_op_open;
97	vdev_close_func_t		*vdev_op_close;
98	vdev_asize_func_t		*vdev_op_asize;
99	vdev_io_start_func_t		*vdev_op_io_start;
100	vdev_io_done_func_t		*vdev_op_io_done;
101	vdev_state_change_func_t	*vdev_op_state_change;
102	vdev_need_resilver_func_t	*vdev_op_need_resilver;
103	vdev_hold_func_t		*vdev_op_hold;
104	vdev_rele_func_t		*vdev_op_rele;
105	vdev_remap_func_t		*vdev_op_remap;
106	/*
107	 * For translating ranges from non-leaf vdevs (e.g. raidz) to leaves.
108	 * Used when initializing vdevs. Isn't used by leaf ops.
109	 */
110	vdev_xlation_func_t		*vdev_op_xlate;
111	vdev_dumpio_func_t		*vdev_op_dumpio;
112	char				vdev_op_type[16];
113	boolean_t			vdev_op_leaf;
114} vdev_ops_t;
115
116/*
117 * Virtual device properties
118 */
119struct vdev_cache_entry {
120	struct abd	*ve_abd;
121	uint64_t	ve_offset;
122	uint64_t	ve_lastused;
123	avl_node_t	ve_offset_node;
124	avl_node_t	ve_lastused_node;
125	uint32_t	ve_hits;
126	uint16_t	ve_missed_update;
127	zio_t		*ve_fill_io;
128};
129
130struct vdev_cache {
131	avl_tree_t	vc_offset_tree;
132	avl_tree_t	vc_lastused_tree;
133	kmutex_t	vc_lock;
134};
135
136typedef struct vdev_queue_class {
137	uint32_t	vqc_active;
138
139	/*
140	 * Sorted by offset or timestamp, depending on if the queue is
141	 * LBA-ordered vs FIFO.
142	 */
143	avl_tree_t	vqc_queued_tree;
144} vdev_queue_class_t;
145
146struct vdev_queue {
147	vdev_t		*vq_vdev;
148	vdev_queue_class_t vq_class[ZIO_PRIORITY_NUM_QUEUEABLE];
149	avl_tree_t	vq_active_tree;
150	avl_tree_t	vq_read_offset_tree;
151	avl_tree_t	vq_write_offset_tree;
152	avl_tree_t	vq_trim_offset_tree;
153	uint64_t	vq_last_offset;
154	hrtime_t	vq_io_complete_ts; /* time last i/o completed */
155	kmutex_t	vq_lock;
156};
157
158typedef enum vdev_alloc_bias {
159	VDEV_BIAS_NONE,
160	VDEV_BIAS_LOG,		/* dedicated to ZIL data (SLOG) */
161	VDEV_BIAS_SPECIAL,	/* dedicated to ddt, metadata, and small blks */
162	VDEV_BIAS_DEDUP		/* dedicated to dedup metadata */
163} vdev_alloc_bias_t;
164
165
166/*
167 * On-disk indirect vdev state.
168 *
169 * An indirect vdev is described exclusively in the MOS config of a pool.
170 * The config for an indirect vdev includes several fields, which are
171 * accessed in memory by a vdev_indirect_config_t.
172 */
173typedef struct vdev_indirect_config {
174	/*
175	 * Object (in MOS) which contains the indirect mapping. This object
176	 * contains an array of vdev_indirect_mapping_entry_phys_t ordered by
177	 * vimep_src. The bonus buffer for this object is a
178	 * vdev_indirect_mapping_phys_t. This object is allocated when a vdev
179	 * removal is initiated.
180	 *
181	 * Note that this object can be empty if none of the data on the vdev
182	 * has been copied yet.
183	 */
184	uint64_t	vic_mapping_object;
185
186	/*
187	 * Object (in MOS) which contains the birth times for the mapping
188	 * entries. This object contains an array of
189	 * vdev_indirect_birth_entry_phys_t sorted by vibe_offset. The bonus
190	 * buffer for this object is a vdev_indirect_birth_phys_t. This object
191	 * is allocated when a vdev removal is initiated.
192	 *
193	 * Note that this object can be empty if none of the vdev has yet been
194	 * copied.
195	 */
196	uint64_t	vic_births_object;
197
198	/*
199	 * This is the vdev ID which was removed previous to this vdev, or
200	 * UINT64_MAX if there are no previously removed vdevs.
201	 */
202	uint64_t	vic_prev_indirect_vdev;
203} vdev_indirect_config_t;
204
205/*
206 * Virtual device descriptor
207 */
208struct vdev {
209	/*
210	 * Common to all vdev types.
211	 */
212	uint64_t	vdev_id;	/* child number in vdev parent	*/
213	uint64_t	vdev_guid;	/* unique ID for this vdev	*/
214	uint64_t	vdev_guid_sum;	/* self guid + all child guids	*/
215	uint64_t	vdev_orig_guid;	/* orig. guid prior to remove	*/
216	uint64_t	vdev_asize;	/* allocatable device capacity	*/
217	uint64_t	vdev_min_asize;	/* min acceptable asize		*/
218	uint64_t	vdev_max_asize;	/* max acceptable asize		*/
219	uint64_t	vdev_ashift;	/* block alignment shift	*/
220	uint64_t	vdev_state;	/* see VDEV_STATE_* #defines	*/
221	uint64_t	vdev_prevstate;	/* used when reopening a vdev	*/
222	vdev_ops_t	*vdev_ops;	/* vdev operations		*/
223	spa_t		*vdev_spa;	/* spa for this vdev		*/
224	void		*vdev_tsd;	/* type-specific data		*/
225	vnode_t		*vdev_name_vp;	/* vnode for pathname		*/
226	vnode_t		*vdev_devid_vp;	/* vnode for devid		*/
227	vdev_t		*vdev_top;	/* top-level vdev		*/
228	vdev_t		*vdev_parent;	/* parent vdev			*/
229	vdev_t		**vdev_child;	/* array of children		*/
230	uint64_t	vdev_children;	/* number of children		*/
231	vdev_stat_t	vdev_stat;	/* virtual device statistics	*/
232	vdev_stat_ex_t	vdev_stat_ex;	/* extended statistics		*/
233	boolean_t	vdev_expanding;	/* expand the vdev?		*/
234	boolean_t	vdev_reopening;	/* reopen in progress?		*/
235	boolean_t	vdev_nonrot;	/* true if solid state		*/
236	int		vdev_open_error; /* error on last open		*/
237	kthread_t	*vdev_open_thread; /* thread opening children	*/
238	uint64_t	vdev_crtxg;	/* txg when top-level was added */
239
240	/*
241	 * Top-level vdev state.
242	 */
243	uint64_t	vdev_ms_array;	/* metaslab array object	*/
244	uint64_t	vdev_ms_shift;	/* metaslab size shift		*/
245	uint64_t	vdev_ms_count;	/* number of metaslabs		*/
246	metaslab_group_t *vdev_mg;	/* metaslab group		*/
247	metaslab_t	**vdev_ms;	/* metaslab array		*/
248	txg_list_t	vdev_ms_list;	/* per-txg dirty metaslab lists	*/
249	txg_list_t	vdev_dtl_list;	/* per-txg dirty DTL lists	*/
250	txg_node_t	vdev_txg_node;	/* per-txg dirty vdev linkage	*/
251	boolean_t	vdev_remove_wanted; /* async remove wanted?	*/
252	boolean_t	vdev_probe_wanted; /* async probe wanted?	*/
253	list_node_t	vdev_config_dirty_node; /* config dirty list	*/
254	list_node_t	vdev_state_dirty_node; /* state dirty list	*/
255	uint64_t	vdev_deflate_ratio; /* deflation ratio (x512)	*/
256	uint64_t	vdev_islog;	/* is an intent log device	*/
257	uint64_t	vdev_removing;	/* device is being removed?	*/
258	boolean_t	vdev_ishole;	/* is a hole in the namespace	*/
259	uint64_t	vdev_top_zap;
260	vdev_alloc_bias_t vdev_alloc_bias; /* metaslab allocation bias	*/
261
262	/* pool checkpoint related */
263	space_map_t	*vdev_checkpoint_sm;	/* contains reserved blocks */
264
265	/* Initialize related */
266	boolean_t	vdev_initialize_exit_wanted;
267	vdev_initializing_state_t	vdev_initialize_state;
268	list_node_t	vdev_initialize_node;
269	kthread_t	*vdev_initialize_thread;
270	/* Protects vdev_initialize_thread and vdev_initialize_state. */
271	kmutex_t	vdev_initialize_lock;
272	kcondvar_t	vdev_initialize_cv;
273	uint64_t	vdev_initialize_offset[TXG_SIZE];
274	uint64_t	vdev_initialize_last_offset;
275	range_tree_t	*vdev_initialize_tree;	/* valid while initializing */
276	uint64_t	vdev_initialize_bytes_est;
277	uint64_t	vdev_initialize_bytes_done;
278	time_t		vdev_initialize_action_time;	/* start and end time */
279
280	/* TRIM related */
281	boolean_t	vdev_trim_exit_wanted;
282	boolean_t	vdev_autotrim_exit_wanted;
283	vdev_trim_state_t	vdev_trim_state;
284	list_node_t	vdev_trim_node;
285	kmutex_t	vdev_autotrim_lock;
286	kcondvar_t	vdev_autotrim_cv;
287	kthread_t	*vdev_autotrim_thread;
288	/* Protects vdev_trim_thread and vdev_trim_state. */
289	kmutex_t	vdev_trim_lock;
290	kcondvar_t	vdev_trim_cv;
291	kthread_t	*vdev_trim_thread;
292	uint64_t	vdev_trim_offset[TXG_SIZE];
293	uint64_t	vdev_trim_last_offset;
294	uint64_t	vdev_trim_bytes_est;
295	uint64_t	vdev_trim_bytes_done;
296	uint64_t	vdev_trim_rate;		/* requested rate (bytes/sec) */
297	uint64_t	vdev_trim_partial;	/* requested partial TRIM */
298	uint64_t	vdev_trim_secure;	/* requested secure TRIM */
299	time_t		vdev_trim_action_time;	/* start and end time */
300
301	/* The following is not in ZoL, but used for auto-trim test progress */
302	uint64_t	vdev_autotrim_bytes_done;
303
304	/* for limiting outstanding I/Os (initialize and TRIM) */
305	kmutex_t	vdev_initialize_io_lock;
306	kcondvar_t	vdev_initialize_io_cv;
307	uint64_t	vdev_initialize_inflight;
308	kmutex_t	vdev_trim_io_lock;
309	kcondvar_t	vdev_trim_io_cv;
310	uint64_t	vdev_trim_inflight[2];
311
312	/*
313	 * Values stored in the config for an indirect or removing vdev.
314	 */
315	vdev_indirect_config_t	vdev_indirect_config;
316
317	/*
318	 * The vdev_indirect_rwlock protects the vdev_indirect_mapping
319	 * pointer from changing on indirect vdevs (when it is condensed).
320	 * Note that removing (not yet indirect) vdevs have different
321	 * access patterns (the mapping is not accessed from open context,
322	 * e.g. from zio_read) and locking strategy (e.g. svr_lock).
323	 */
324	krwlock_t vdev_indirect_rwlock;
325	vdev_indirect_mapping_t *vdev_indirect_mapping;
326	vdev_indirect_births_t *vdev_indirect_births;
327
328	/*
329	 * In memory data structures used to manage the obsolete sm, for
330	 * indirect or removing vdevs.
331	 *
332	 * The vdev_obsolete_segments is the in-core record of the segments
333	 * that are no longer referenced anywhere in the pool (due to
334	 * being freed or remapped and not referenced by any snapshots).
335	 * During a sync, segments are added to vdev_obsolete_segments
336	 * via vdev_indirect_mark_obsolete(); at the end of each sync
337	 * pass, this is appended to vdev_obsolete_sm via
338	 * vdev_indirect_sync_obsolete().  The vdev_obsolete_lock
339	 * protects against concurrent modifications of vdev_obsolete_segments
340	 * from multiple zio threads.
341	 */
342	kmutex_t	vdev_obsolete_lock;
343	range_tree_t	*vdev_obsolete_segments;
344	space_map_t	*vdev_obsolete_sm;
345
346	/*
347	 * Protects the vdev_scan_io_queue field itself as well as the
348	 * structure's contents (when present).
349	 */
350	kmutex_t	vdev_scan_io_queue_lock;
351	struct dsl_scan_io_queue	*vdev_scan_io_queue;
352
353	/*
354	 * Leaf vdev state.
355	 */
356	range_tree_t	*vdev_dtl[DTL_TYPES]; /* dirty time logs	*/
357	space_map_t	*vdev_dtl_sm;	/* dirty time log space map	*/
358	txg_node_t	vdev_dtl_node;	/* per-txg dirty DTL linkage	*/
359	uint64_t	vdev_dtl_object; /* DTL object			*/
360	uint64_t	vdev_psize;	/* physical device capacity	*/
361	uint64_t	vdev_wholedisk;	/* true if this is a whole disk */
362	uint64_t	vdev_offline;	/* persistent offline state	*/
363	uint64_t	vdev_faulted;	/* persistent faulted state	*/
364	uint64_t	vdev_degraded;	/* persistent degraded state	*/
365	uint64_t	vdev_removed;	/* persistent removed state	*/
366	uint64_t	vdev_resilver_txg; /* persistent resilvering state */
367	uint64_t	vdev_nparity;	/* number of parity devices for raidz */
368	char		*vdev_path;	/* vdev path (if any)		*/
369	char		*vdev_devid;	/* vdev devid (if any)		*/
370	char		*vdev_physpath;	/* vdev device path (if any)	*/
371	char		*vdev_fru;	/* physical FRU location	*/
372	uint64_t	vdev_not_present; /* not present during import	*/
373	uint64_t	vdev_unspare;	/* unspare when resilvering done */
374	boolean_t	vdev_nowritecache; /* true if flushwritecache failed */
375	boolean_t	vdev_has_trim;	/* TRIM is supported		*/
376	boolean_t	vdev_has_securetrim; /* secure TRIM is supported */
377	boolean_t	vdev_checkremove; /* temporary online test	*/
378	boolean_t	vdev_forcefault; /* force online fault		*/
379	boolean_t	vdev_splitting;	/* split or repair in progress  */
380	boolean_t	vdev_delayed_close; /* delayed device close?	*/
381	boolean_t	vdev_tmpoffline; /* device taken offline temporarily? */
382	boolean_t	vdev_detached;	/* device detached?		*/
383	boolean_t	vdev_cant_read;	/* vdev is failing all reads	*/
384	boolean_t	vdev_cant_write; /* vdev is failing all writes	*/
385	boolean_t	vdev_isspare;	/* was a hot spare		*/
386	boolean_t	vdev_isl2cache;	/* was a l2cache device		*/
387	boolean_t	vdev_resilver_deferred;  /* resilver deferred */
388	vdev_queue_t	vdev_queue;	/* I/O deadline schedule queue	*/
389	vdev_cache_t	vdev_cache;	/* physical block cache		*/
390	spa_aux_vdev_t	*vdev_aux;	/* for l2cache and spares vdevs	*/
391	zio_t		*vdev_probe_zio; /* root of current probe	*/
392	vdev_aux_t	vdev_label_aux;	/* on-disk aux state		*/
393	uint64_t	vdev_leaf_zap;
394	hrtime_t	vdev_mmp_pending; /* 0 if write finished	*/
395	uint64_t	vdev_mmp_kstat_id;	/* to find kstat entry */
396	list_node_t	vdev_leaf_node;		/* leaf vdev list */
397
398	/*
399	 * For DTrace to work in userland (libzpool) context, these fields must
400	 * remain at the end of the structure.  DTrace will use the kernel's
401	 * CTF definition for 'struct vdev', and since the size of a kmutex_t is
402	 * larger in userland, the offsets for the rest of the fields would be
403	 * incorrect.
404	 */
405	kmutex_t	vdev_dtl_lock;	/* vdev_dtl_{map,resilver}	*/
406	kmutex_t	vdev_stat_lock;	/* vdev_stat			*/
407	kmutex_t	vdev_probe_lock; /* protects vdev_probe_zio	*/
408};
409
410#define	VDEV_RAIDZ_MAXPARITY	3
411
412#define	VDEV_PAD_SIZE		(8 << 10)
413/* 2 padding areas (vl_pad1 and vl_pad2) to skip */
414#define	VDEV_SKIP_SIZE		VDEV_PAD_SIZE * 2
415#define	VDEV_PHYS_SIZE		(112 << 10)
416#define	VDEV_UBERBLOCK_RING	(128 << 10)
417
418/*
419 * MMP blocks occupy the last MMP_BLOCKS_PER_LABEL slots in the uberblock
420 * ring when MMP is enabled.
421 */
422#define	MMP_BLOCKS_PER_LABEL	1
423
424/* The largest uberblock we support is 8k. */
425#define	MAX_UBERBLOCK_SHIFT (13)
426#define	VDEV_UBERBLOCK_SHIFT(vd)	\
427	MIN(MAX((vd)->vdev_top->vdev_ashift, UBERBLOCK_SHIFT), \
428	    MAX_UBERBLOCK_SHIFT)
429#define	VDEV_UBERBLOCK_COUNT(vd)	\
430	(VDEV_UBERBLOCK_RING >> VDEV_UBERBLOCK_SHIFT(vd))
431#define	VDEV_UBERBLOCK_OFFSET(vd, n)	\
432	offsetof(vdev_label_t, vl_uberblock[(n) << VDEV_UBERBLOCK_SHIFT(vd)])
433#define	VDEV_UBERBLOCK_SIZE(vd)		(1ULL << VDEV_UBERBLOCK_SHIFT(vd))
434
435typedef struct vdev_phys {
436	char		vp_nvlist[VDEV_PHYS_SIZE - sizeof (zio_eck_t)];
437	zio_eck_t	vp_zbt;
438} vdev_phys_t;
439
440typedef struct vdev_label {
441	char		vl_pad1[VDEV_PAD_SIZE];			/*  8K */
442	char		vl_pad2[VDEV_PAD_SIZE];			/*  8K */
443	vdev_phys_t	vl_vdev_phys;				/* 112K	*/
444	char		vl_uberblock[VDEV_UBERBLOCK_RING];	/* 128K	*/
445} vdev_label_t;							/* 256K total */
446
447/*
448 * vdev_dirty() flags
449 */
450#define	VDD_METASLAB	0x01
451#define	VDD_DTL		0x02
452
453/* Offset of embedded boot loader region on each label */
454#define	VDEV_BOOT_OFFSET	(2 * sizeof (vdev_label_t))
455/*
456 * Size of embedded boot loader region on each label.
457 * The total size of the first two labels plus the boot area is 4MB.
458 */
459#define	VDEV_BOOT_SIZE		(7ULL << 19)			/* 3.5M */
460
461/*
462 * Size of label regions at the start and end of each leaf device.
463 */
464#define	VDEV_LABEL_START_SIZE	(2 * sizeof (vdev_label_t) + VDEV_BOOT_SIZE)
465#define	VDEV_LABEL_END_SIZE	(2 * sizeof (vdev_label_t))
466#define	VDEV_LABELS		4
467#define	VDEV_BEST_LABEL		VDEV_LABELS
468
469#define	VDEV_ALLOC_LOAD		0
470#define	VDEV_ALLOC_ADD		1
471#define	VDEV_ALLOC_SPARE	2
472#define	VDEV_ALLOC_L2CACHE	3
473#define	VDEV_ALLOC_ROOTPOOL	4
474#define	VDEV_ALLOC_SPLIT	5
475#define	VDEV_ALLOC_ATTACH	6
476
477/*
478 * Allocate or free a vdev
479 */
480extern vdev_t *vdev_alloc_common(spa_t *spa, uint_t id, uint64_t guid,
481    vdev_ops_t *ops);
482extern int vdev_alloc(spa_t *spa, vdev_t **vdp, nvlist_t *config,
483    vdev_t *parent, uint_t id, int alloctype);
484extern void vdev_free(vdev_t *vd);
485
486/*
487 * Add or remove children and parents
488 */
489extern void vdev_add_child(vdev_t *pvd, vdev_t *cvd);
490extern void vdev_remove_child(vdev_t *pvd, vdev_t *cvd);
491extern void vdev_compact_children(vdev_t *pvd);
492extern vdev_t *vdev_add_parent(vdev_t *cvd, vdev_ops_t *ops);
493extern void vdev_remove_parent(vdev_t *cvd);
494
495/*
496 * vdev sync load and sync
497 */
498extern boolean_t vdev_log_state_valid(vdev_t *vd);
499extern int vdev_load(vdev_t *vd);
500extern int vdev_dtl_load(vdev_t *vd);
501extern void vdev_sync(vdev_t *vd, uint64_t txg);
502extern void vdev_sync_done(vdev_t *vd, uint64_t txg);
503extern void vdev_dirty(vdev_t *vd, int flags, void *arg, uint64_t txg);
504extern void vdev_dirty_leaves(vdev_t *vd, int flags, uint64_t txg);
505
506/*
507 * Available vdev types.
508 */
509extern vdev_ops_t vdev_root_ops;
510extern vdev_ops_t vdev_mirror_ops;
511extern vdev_ops_t vdev_replacing_ops;
512extern vdev_ops_t vdev_raidz_ops;
513extern vdev_ops_t vdev_disk_ops;
514extern vdev_ops_t vdev_file_ops;
515extern vdev_ops_t vdev_missing_ops;
516extern vdev_ops_t vdev_hole_ops;
517extern vdev_ops_t vdev_spare_ops;
518extern vdev_ops_t vdev_indirect_ops;
519
520/*
521 * Common size functions
522 */
523extern void vdev_default_xlate(vdev_t *vd, const range_seg64_t *in,
524    range_seg64_t *out);
525extern uint64_t vdev_default_asize(vdev_t *vd, uint64_t psize);
526extern uint64_t vdev_get_min_asize(vdev_t *vd);
527extern void vdev_set_min_asize(vdev_t *vd);
528
529/*
530 * Global variables
531 */
532extern int zfs_vdev_standard_sm_blksz;
533/* zdb uses this tunable, so it must be declared here to make lint happy. */
534extern int zfs_vdev_cache_size;
535
536/*
537 * Functions from vdev_indirect.c
538 */
539extern void vdev_indirect_sync_obsolete(vdev_t *vd, dmu_tx_t *tx);
540extern boolean_t vdev_indirect_should_condense(vdev_t *vd);
541extern void spa_condense_indirect_start_sync(vdev_t *vd, dmu_tx_t *tx);
542extern int vdev_obsolete_sm_object(vdev_t *vd);
543extern boolean_t vdev_obsolete_counts_are_precise(vdev_t *vd);
544
545/*
546 * Other miscellaneous functions
547 */
548int vdev_checkpoint_sm_object(vdev_t *vd);
549
550/*
551 * The vdev_buf_t is used to translate between zio_t and buf_t, and back again.
552 */
553typedef struct vdev_buf {
554	buf_t	vb_buf;		/* buffer that describes the io */
555	zio_t	*vb_io;		/* pointer back to the original zio_t */
556} vdev_buf_t;
557
558#ifdef	__cplusplus
559}
560#endif
561
562#endif	/* _SYS_VDEV_IMPL_H */
563