spa_impl.h revision 6f7938128a2c5e23f4b970ea101137eadd1470a1
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 2011 Nexenta Systems, Inc.  All rights reserved.
25 * Copyright (c) 2014 Spectra Logic Corporation, All rights reserved.
26 * Copyright 2013 Saso Kiselkov. All rights reserved.
27 * Copyright (c) 2017 Datto Inc.
28 */
29
30#ifndef _SYS_SPA_IMPL_H
31#define	_SYS_SPA_IMPL_H
32
33#include <sys/spa.h>
34#include <sys/vdev.h>
35#include <sys/vdev_removal.h>
36#include <sys/metaslab.h>
37#include <sys/dmu.h>
38#include <sys/dsl_pool.h>
39#include <sys/uberblock_impl.h>
40#include <sys/zfs_context.h>
41#include <sys/avl.h>
42#include <sys/refcount.h>
43#include <sys/bplist.h>
44#include <sys/bpobj.h>
45#include <sys/zfeature.h>
46#include <zfeature_common.h>
47
48#ifdef	__cplusplus
49extern "C" {
50#endif
51
52typedef struct spa_error_entry {
53	zbookmark_phys_t	se_bookmark;
54	char			*se_name;
55	avl_node_t		se_avl;
56} spa_error_entry_t;
57
58typedef struct spa_history_phys {
59	uint64_t sh_pool_create_len;	/* ending offset of zpool create */
60	uint64_t sh_phys_max_off;	/* physical EOF */
61	uint64_t sh_bof;		/* logical BOF */
62	uint64_t sh_eof;		/* logical EOF */
63	uint64_t sh_records_lost;	/* num of records overwritten */
64} spa_history_phys_t;
65
66/*
67 * All members must be uint64_t, for byteswap purposes.
68 */
69typedef struct spa_removing_phys {
70	uint64_t sr_state; /* dsl_scan_state_t */
71
72	/*
73	 * The vdev ID that we most recently attempted to remove,
74	 * or -1 if no removal has been attempted.
75	 */
76	uint64_t sr_removing_vdev;
77
78	/*
79	 * The vdev ID that we most recently successfully removed,
80	 * or -1 if no devices have been removed.
81	 */
82	uint64_t sr_prev_indirect_vdev;
83
84	uint64_t sr_start_time;
85	uint64_t sr_end_time;
86
87	/*
88	 * Note that we can not use the space map's or indirect mapping's
89	 * accounting as a substitute for these values, because we need to
90	 * count frees of not-yet-copied data as though it did the copy.
91	 * Otherwise, we could get into a situation where copied > to_copy,
92	 * or we complete before copied == to_copy.
93	 */
94	uint64_t sr_to_copy; /* bytes that need to be copied */
95	uint64_t sr_copied; /* bytes that have been copied or freed */
96} spa_removing_phys_t;
97
98/*
99 * This struct is stored as an entry in the DMU_POOL_DIRECTORY_OBJECT
100 * (with key DMU_POOL_CONDENSING_INDIRECT).  It is present if a condense
101 * of an indirect vdev's mapping object is in progress.
102 */
103typedef struct spa_condensing_indirect_phys {
104	/*
105	 * The vdev ID of the indirect vdev whose indirect mapping is
106	 * being condensed.
107	 */
108	uint64_t	scip_vdev;
109
110	/*
111	 * The vdev's old obsolete spacemap.  This spacemap's contents are
112	 * being integrated into the new mapping.
113	 */
114	uint64_t	scip_prev_obsolete_sm_object;
115
116	/*
117	 * The new mapping object that is being created.
118	 */
119	uint64_t	scip_next_mapping_object;
120} spa_condensing_indirect_phys_t;
121
122struct spa_aux_vdev {
123	uint64_t	sav_object;		/* MOS object for device list */
124	nvlist_t	*sav_config;		/* cached device config */
125	vdev_t		**sav_vdevs;		/* devices */
126	int		sav_count;		/* number devices */
127	boolean_t	sav_sync;		/* sync the device list */
128	nvlist_t	**sav_pending;		/* pending device additions */
129	uint_t		sav_npending;		/* # pending devices */
130};
131
132typedef struct spa_config_lock {
133	kmutex_t	scl_lock;
134	kthread_t	*scl_writer;
135	int		scl_write_wanted;
136	kcondvar_t	scl_cv;
137	refcount_t	scl_count;
138} spa_config_lock_t;
139
140typedef struct spa_config_dirent {
141	list_node_t	scd_link;
142	char		*scd_path;
143} spa_config_dirent_t;
144
145typedef enum zio_taskq_type {
146	ZIO_TASKQ_ISSUE = 0,
147	ZIO_TASKQ_ISSUE_HIGH,
148	ZIO_TASKQ_INTERRUPT,
149	ZIO_TASKQ_INTERRUPT_HIGH,
150	ZIO_TASKQ_TYPES
151} zio_taskq_type_t;
152
153/*
154 * State machine for the zpool-poolname process.  The states transitions
155 * are done as follows:
156 *
157 *	From		   To			Routine
158 *	PROC_NONE	-> PROC_CREATED		spa_activate()
159 *	PROC_CREATED	-> PROC_ACTIVE		spa_thread()
160 *	PROC_ACTIVE	-> PROC_DEACTIVATE	spa_deactivate()
161 *	PROC_DEACTIVATE	-> PROC_GONE		spa_thread()
162 *	PROC_GONE	-> PROC_NONE		spa_deactivate()
163 */
164typedef enum spa_proc_state {
165	SPA_PROC_NONE,		/* spa_proc = &p0, no process created */
166	SPA_PROC_CREATED,	/* spa_activate() has proc, is waiting */
167	SPA_PROC_ACTIVE,	/* taskqs created, spa_proc set */
168	SPA_PROC_DEACTIVATE,	/* spa_deactivate() requests process exit */
169	SPA_PROC_GONE		/* spa_thread() is exiting, spa_proc = &p0 */
170} spa_proc_state_t;
171
172typedef struct spa_taskqs {
173	uint_t stqs_count;
174	taskq_t **stqs_taskq;
175} spa_taskqs_t;
176
177typedef enum spa_all_vdev_zap_action {
178	AVZ_ACTION_NONE = 0,
179	AVZ_ACTION_DESTROY,	/* Destroy all per-vdev ZAPs and the AVZ. */
180	AVZ_ACTION_REBUILD,	/* Populate the new AVZ, see spa_avz_rebuild */
181	AVZ_ACTION_INITIALIZE
182} spa_avz_action_t;
183
184typedef enum spa_config_source {
185	SPA_CONFIG_SRC_NONE = 0,
186	SPA_CONFIG_SRC_SCAN,		/* scan of path (default: /dev/dsk) */
187	SPA_CONFIG_SRC_CACHEFILE,	/* any cachefile */
188	SPA_CONFIG_SRC_TRYIMPORT,	/* returned from call to tryimport */
189	SPA_CONFIG_SRC_SPLIT,		/* new pool in a pool split */
190	SPA_CONFIG_SRC_MOS		/* MOS, but not always from right txg */
191} spa_config_source_t;
192
193struct spa {
194	/*
195	 * Fields protected by spa_namespace_lock.
196	 */
197	char		spa_name[ZFS_MAX_DATASET_NAME_LEN];	/* pool name */
198	char		*spa_comment;		/* comment */
199	avl_node_t	spa_avl;		/* node in spa_namespace_avl */
200	nvlist_t	*spa_config;		/* last synced config */
201	nvlist_t	*spa_config_syncing;	/* currently syncing config */
202	nvlist_t	*spa_config_splitting;	/* config for splitting */
203	nvlist_t	*spa_load_info;		/* info and errors from load */
204	uint64_t	spa_config_txg;		/* txg of last config change */
205	int		spa_sync_pass;		/* iterate-to-convergence */
206	pool_state_t	spa_state;		/* pool state */
207	int		spa_inject_ref;		/* injection references */
208	uint8_t		spa_sync_on;		/* sync threads are running */
209	spa_load_state_t spa_load_state;	/* current load operation */
210	boolean_t	spa_indirect_vdevs_loaded; /* mappings loaded? */
211	boolean_t	spa_trust_config;	/* do we trust vdev tree? */
212	spa_config_source_t spa_config_source;	/* where config comes from? */
213	uint64_t	spa_import_flags;	/* import specific flags */
214	spa_taskqs_t	spa_zio_taskq[ZIO_TYPES][ZIO_TASKQ_TYPES];
215	dsl_pool_t	*spa_dsl_pool;
216	boolean_t	spa_is_initializing;	/* true while opening pool */
217	metaslab_class_t *spa_normal_class;	/* normal data class */
218	metaslab_class_t *spa_log_class;	/* intent log data class */
219	uint64_t	spa_first_txg;		/* first txg after spa_open() */
220	uint64_t	spa_final_txg;		/* txg of export/destroy */
221	uint64_t	spa_freeze_txg;		/* freeze pool at this txg */
222	uint64_t	spa_load_max_txg;	/* best initial ub_txg */
223	uint64_t	spa_claim_max_txg;	/* highest claimed birth txg */
224	timespec_t	spa_loaded_ts;		/* 1st successful open time */
225	objset_t	*spa_meta_objset;	/* copy of dp->dp_meta_objset */
226	kmutex_t	spa_evicting_os_lock;	/* Evicting objset list lock */
227	list_t		spa_evicting_os_list;	/* Objsets being evicted. */
228	kcondvar_t	spa_evicting_os_cv;	/* Objset Eviction Completion */
229	txg_list_t	spa_vdev_txg_list;	/* per-txg dirty vdev list */
230	vdev_t		*spa_root_vdev;		/* top-level vdev container */
231	int		spa_min_ashift;		/* of vdevs in normal class */
232	int		spa_max_ashift;		/* of vdevs in normal class */
233	uint64_t	spa_config_guid;	/* config pool guid */
234	uint64_t	spa_load_guid;		/* spa_load initialized guid */
235	uint64_t	spa_last_synced_guid;	/* last synced guid */
236	list_t		spa_config_dirty_list;	/* vdevs with dirty config */
237	list_t		spa_state_dirty_list;	/* vdevs with dirty state */
238	kmutex_t	spa_alloc_lock;
239	avl_tree_t	spa_alloc_tree;
240	spa_aux_vdev_t	spa_spares;		/* hot spares */
241	spa_aux_vdev_t	spa_l2cache;		/* L2ARC cache devices */
242	nvlist_t	*spa_label_features;	/* Features for reading MOS */
243	uint64_t	spa_config_object;	/* MOS object for pool config */
244	uint64_t	spa_config_generation;	/* config generation number */
245	uint64_t	spa_syncing_txg;	/* txg currently syncing */
246	bpobj_t		spa_deferred_bpobj;	/* deferred-free bplist */
247	bplist_t	spa_free_bplist[TXG_SIZE]; /* bplist of stuff to free */
248	zio_cksum_salt_t spa_cksum_salt;	/* secret salt for cksum */
249	/* checksum context templates */
250	kmutex_t	spa_cksum_tmpls_lock;
251	void		*spa_cksum_tmpls[ZIO_CHECKSUM_FUNCTIONS];
252	uberblock_t	spa_ubsync;		/* last synced uberblock */
253	uberblock_t	spa_uberblock;		/* current uberblock */
254	boolean_t	spa_extreme_rewind;	/* rewind past deferred frees */
255	uint64_t	spa_last_io;		/* lbolt of last non-scan I/O */
256	kmutex_t	spa_scrub_lock;		/* resilver/scrub lock */
257	uint64_t	spa_scrub_inflight;	/* in-flight scrub I/Os */
258	kcondvar_t	spa_scrub_io_cv;	/* scrub I/O completion */
259	uint8_t		spa_scrub_active;	/* active or suspended? */
260	uint8_t		spa_scrub_type;		/* type of scrub we're doing */
261	uint8_t		spa_scrub_finished;	/* indicator to rotate logs */
262	uint8_t		spa_scrub_started;	/* started since last boot */
263	uint8_t		spa_scrub_reopen;	/* scrub doing vdev_reopen */
264	uint64_t	spa_scan_pass_start;	/* start time per pass/reboot */
265	uint64_t	spa_scan_pass_scrub_pause; /* scrub pause time */
266	uint64_t	spa_scan_pass_scrub_spent_paused; /* total paused */
267	uint64_t	spa_scan_pass_exam;	/* examined bytes per pass */
268	kmutex_t	spa_async_lock;		/* protect async state */
269	kthread_t	*spa_async_thread;	/* thread doing async task */
270	int		spa_async_suspended;	/* async tasks suspended */
271	kcondvar_t	spa_async_cv;		/* wait for thread_exit() */
272	uint16_t	spa_async_tasks;	/* async task mask */
273	uint64_t	spa_missing_tvds;	/* unopenable tvds on load */
274	uint64_t	spa_missing_tvds_allowed; /* allow loading spa? */
275
276	spa_removing_phys_t spa_removing_phys;
277	spa_vdev_removal_t *spa_vdev_removal;
278
279	spa_condensing_indirect_phys_t	spa_condensing_indirect_phys;
280	spa_condensing_indirect_t	*spa_condensing_indirect;
281	kthread_t	*spa_condense_thread;	/* thread doing condense. */
282
283	char		*spa_root;		/* alternate root directory */
284	uint64_t	spa_ena;		/* spa-wide ereport ENA */
285	int		spa_last_open_failed;	/* error if last open failed */
286	uint64_t	spa_last_ubsync_txg;	/* "best" uberblock txg */
287	uint64_t	spa_last_ubsync_txg_ts;	/* timestamp from that ub */
288	uint64_t	spa_load_txg;		/* ub txg that loaded */
289	uint64_t	spa_load_txg_ts;	/* timestamp from that ub */
290	uint64_t	spa_load_meta_errors;	/* verify metadata err count */
291	uint64_t	spa_load_data_errors;	/* verify data err count */
292	uint64_t	spa_verify_min_txg;	/* start txg of verify scrub */
293	kmutex_t	spa_errlog_lock;	/* error log lock */
294	uint64_t	spa_errlog_last;	/* last error log object */
295	uint64_t	spa_errlog_scrub;	/* scrub error log object */
296	kmutex_t	spa_errlist_lock;	/* error list/ereport lock */
297	avl_tree_t	spa_errlist_last;	/* last error list */
298	avl_tree_t	spa_errlist_scrub;	/* scrub error list */
299	uint64_t	spa_deflate;		/* should we deflate? */
300	uint64_t	spa_history;		/* history object */
301	kmutex_t	spa_history_lock;	/* history lock */
302	vdev_t		*spa_pending_vdev;	/* pending vdev additions */
303	kmutex_t	spa_props_lock;		/* property lock */
304	uint64_t	spa_pool_props_object;	/* object for properties */
305	uint64_t	spa_bootfs;		/* default boot filesystem */
306	uint64_t	spa_failmode;		/* failure mode for the pool */
307	uint64_t	spa_delegation;		/* delegation on/off */
308	list_t		spa_config_list;	/* previous cache file(s) */
309	/* per-CPU array of root of async I/O: */
310	zio_t		**spa_async_zio_root;
311	zio_t		*spa_suspend_zio_root;	/* root of all suspended I/O */
312	zio_t		*spa_txg_zio[TXG_SIZE]; /* spa_sync() waits for this */
313	kmutex_t	spa_suspend_lock;	/* protects suspend_zio_root */
314	kcondvar_t	spa_suspend_cv;		/* notification of resume */
315	uint8_t		spa_suspended;		/* pool is suspended */
316	uint8_t		spa_claiming;		/* pool is doing zil_claim() */
317	boolean_t	spa_debug;		/* debug enabled? */
318	boolean_t	spa_is_root;		/* pool is root */
319	int		spa_minref;		/* num refs when first opened */
320	int		spa_mode;		/* FREAD | FWRITE */
321	spa_log_state_t spa_log_state;		/* log state */
322	uint64_t	spa_autoexpand;		/* lun expansion on/off */
323	uint64_t	spa_bootsize;		/* efi system partition size */
324	ddt_t		*spa_ddt[ZIO_CHECKSUM_FUNCTIONS]; /* in-core DDTs */
325	uint64_t	spa_ddt_stat_object;	/* DDT statistics */
326	uint64_t	spa_dedup_ditto;	/* dedup ditto threshold */
327	uint64_t	spa_dedup_checksum;	/* default dedup checksum */
328	uint64_t	spa_dspace;		/* dspace in normal class */
329	kmutex_t	spa_vdev_top_lock;	/* dueling offline/remove */
330	kmutex_t	spa_proc_lock;		/* protects spa_proc* */
331	kcondvar_t	spa_proc_cv;		/* spa_proc_state transitions */
332	spa_proc_state_t spa_proc_state;	/* see definition */
333	struct proc	*spa_proc;		/* "zpool-poolname" process */
334	uint64_t	spa_did;		/* if procp != p0, did of t1 */
335	boolean_t	spa_autoreplace;	/* autoreplace set in open */
336	int		spa_vdev_locks;		/* locks grabbed */
337	uint64_t	spa_creation_version;	/* version at pool creation */
338	uint64_t	spa_prev_software_version; /* See ub_software_version */
339	uint64_t	spa_feat_for_write_obj;	/* required to write to pool */
340	uint64_t	spa_feat_for_read_obj;	/* required to read from pool */
341	uint64_t	spa_feat_desc_obj;	/* Feature descriptions */
342	uint64_t	spa_feat_enabled_txg_obj; /* Feature enabled txg */
343	/* cache feature refcounts */
344	uint64_t	spa_feat_refcount_cache[SPA_FEATURES];
345	cyclic_id_t	spa_deadman_cycid;	/* cyclic id */
346	uint64_t	spa_deadman_calls;	/* number of deadman calls */
347	hrtime_t	spa_sync_starttime;	/* starting time fo spa_sync */
348	uint64_t	spa_deadman_synctime;	/* deadman expiration timer */
349	uint64_t	spa_all_vdev_zaps;	/* ZAP of per-vd ZAP obj #s */
350	spa_avz_action_t	spa_avz_action;	/* destroy/rebuild AVZ? */
351
352	/*
353	 * spa_iokstat_lock protects spa_iokstat and
354	 * spa_queue_stats[].
355	 */
356	kmutex_t	spa_iokstat_lock;
357	struct kstat	*spa_iokstat;		/* kstat of io to this pool */
358	struct {
359		int spa_active;
360		int spa_queued;
361	} spa_queue_stats[ZIO_PRIORITY_NUM_QUEUEABLE];
362
363	hrtime_t	spa_ccw_fail_time;	/* Conf cache write fail time */
364
365	/*
366	 * spa_refcount & spa_config_lock must be the last elements
367	 * because refcount_t changes size based on compilation options.
368	 * In order for the MDB module to function correctly, the other
369	 * fields must remain in the same location.
370	 */
371	spa_config_lock_t spa_config_lock[SCL_LOCKS]; /* config changes */
372	refcount_t	spa_refcount;		/* number of opens */
373};
374
375extern const char *spa_config_path;
376
377extern void spa_taskq_dispatch_ent(spa_t *spa, zio_type_t t, zio_taskq_type_t q,
378    task_func_t *func, void *arg, uint_t flags, taskq_ent_t *ent);
379extern void spa_load_spares(spa_t *spa);
380extern void spa_load_l2cache(spa_t *spa);
381
382#ifdef	__cplusplus
383}
384#endif
385
386#endif	/* _SYS_SPA_IMPL_H */
387