blob: 8e4c75c00d4628f76ca32e0aaab243e3aea70d4c [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001/*
2 md_k.h : kernel internal structure of the Linux MD driver
3 Copyright (C) 1996-98 Ingo Molnar, Gadi Oxman
4
5 This program is free software; you can redistribute it and/or modify
6 it under the terms of the GNU General Public License as published by
7 the Free Software Foundation; either version 2, or (at your option)
8 any later version.
9
10 You should have received a copy of the GNU General Public License
11 (for example /usr/src/linux/COPYING); if not, write to the Free
12 Software Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
13*/
14
Christoph Hellwig63fe0812009-04-14 12:01:53 +100015#ifndef _MD_MD_H
16#define _MD_MD_H
Linus Torvalds1da177e2005-04-16 15:20:36 -070017
Christoph Hellwig63fe0812009-04-14 12:01:53 +100018#include <linux/blkdev.h>
19#include <linux/kobject.h>
20#include <linux/list.h>
21#include <linux/mm.h>
22#include <linux/mutex.h>
23#include <linux/timer.h>
24#include <linux/wait.h>
25#include <linux/workqueue.h>
David Howells93614012006-09-30 20:45:40 +020026
Linus Torvalds1da177e2005-04-16 15:20:36 -070027#define MaxSector (~(sector_t)0)
Linus Torvalds1da177e2005-04-16 15:20:36 -070028
Linus Torvalds1da177e2005-04-16 15:20:36 -070029typedef struct mddev_s mddev_t;
30typedef struct mdk_rdev_s mdk_rdev_t;
31
Linus Torvalds1da177e2005-04-16 15:20:36 -070032/*
Linus Torvalds1da177e2005-04-16 15:20:36 -070033 * MD's 'extended' device
34 */
35struct mdk_rdev_s
36{
37 struct list_head same_set; /* RAID devices within the same set */
38
Andre Nolldd8ac332009-03-31 14:33:13 +110039 sector_t sectors; /* Device size (in 512bytes sectors) */
Linus Torvalds1da177e2005-04-16 15:20:36 -070040 mddev_t *mddev; /* RAID array if running */
NeilBrowneea1bf32009-03-31 14:27:02 +110041 int last_events; /* IO event timestamp */
Linus Torvalds1da177e2005-04-16 15:20:36 -070042
43 struct block_device *bdev; /* block device handle */
44
45 struct page *sb_page;
46 int sb_loaded;
NeilBrown42543762006-06-26 00:27:57 -070047 __u64 sb_events;
Linus Torvalds1da177e2005-04-16 15:20:36 -070048 sector_t data_offset; /* start of data in array */
Andre Noll0f420352008-07-11 22:02:23 +100049 sector_t sb_start; /* offset of the super block (in 512byte sectors) */
NeilBrown0002b272005-09-09 16:23:53 -070050 int sb_size; /* bytes in the superblock */
Linus Torvalds1da177e2005-04-16 15:20:36 -070051 int preferred_minor; /* autorun support */
52
NeilBrown86e6ffd2005-11-08 21:39:24 -080053 struct kobject kobj;
54
Linus Torvalds1da177e2005-04-16 15:20:36 -070055 /* A device can be in one of three states based on two flags:
56 * Not working: faulty==1 in_sync==0
57 * Fully working: faulty==0 in_sync==1
58 * Working, but not
59 * in sync with array
60 * faulty==0 in_sync==0
61 *
62 * It can never have faulty==1, in_sync==1
63 * This reduces the burden of testing multiple flags in many cases
64 */
Linus Torvalds1da177e2005-04-16 15:20:36 -070065
NeilBrownb2d444d2005-11-08 21:39:31 -080066 unsigned long flags;
67#define Faulty 1 /* device is known to have a fault */
68#define In_sync 2 /* device is in_sync with rest of array */
NeilBrown8ddf9ef2005-09-09 16:23:45 -070069#define WriteMostly 4 /* Avoid reading if at all possible */
NeilBrowna9701a32005-11-08 21:39:34 -080070#define BarriersNotsupp 5 /* BIO_RW_BARRIER is not supported */
NeilBrownc5d79ad2008-02-06 01:39:54 -080071#define AllReserved 6 /* If whole device is reserved for
72 * one array */
NeilBrownd0fae182008-03-04 14:29:31 -080073#define AutoDetected 7 /* added by auto-detect */
Dan Williams6bfe0b42008-04-30 00:52:32 -070074#define Blocked 8 /* An error occured on an externally
75 * managed array, don't allow writes
76 * until it is cleared */
Neil Brown52664732008-06-28 08:31:44 +100077#define StateChanged 9 /* Faulty or Blocked has changed during
78 * interrupt, so it needs to be
79 * notified by the thread */
Dan Williams6bfe0b42008-04-30 00:52:32 -070080 wait_queue_head_t blocked_wait;
NeilBrown8ddf9ef2005-09-09 16:23:45 -070081
Linus Torvalds1da177e2005-04-16 15:20:36 -070082 int desc_nr; /* descriptor index in the superblock */
83 int raid_disk; /* role of device in array */
NeilBrown41158c72005-06-21 17:17:25 -070084 int saved_raid_disk; /* role that device used to have in the
85 * array and could again if we did a partial
86 * resync from the bitmap
87 */
NeilBrown5fd6c1d2006-06-26 00:27:40 -070088 sector_t recovery_offset;/* If this device has been partially
89 * recovered, this is where we were
90 * up to.
91 */
Linus Torvalds1da177e2005-04-16 15:20:36 -070092
93 atomic_t nr_pending; /* number of pending requests.
94 * only maintained for arrays that
95 * support hot removal
96 */
NeilBrownba22dcb2005-11-08 21:39:31 -080097 atomic_t read_errors; /* number of consecutive read errors that
98 * we have tried to ignore.
99 */
Robert Becker1e509152009-12-14 12:49:58 +1100100 struct timespec last_read_error; /* monotonic time since our
101 * last read error
102 */
NeilBrown4dbcdc72006-01-06 00:20:52 -0800103 atomic_t corrected_errors; /* number of corrected read errors,
104 * for reporting to userspace and storing
105 * in superblock.
106 */
NeilBrown5792a282007-04-04 19:08:18 -0700107 struct work_struct del_work; /* used for delayed sysfs removal */
NeilBrown3c0ee632008-10-21 13:25:28 +1100108
109 struct sysfs_dirent *sysfs_state; /* handle for 'state'
110 * sysfs entry */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700111};
112
Linus Torvalds1da177e2005-04-16 15:20:36 -0700113struct mddev_s
114{
115 void *private;
NeilBrown2604b702006-01-06 00:20:36 -0800116 struct mdk_personality *pers;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700117 dev_t unit;
118 int md_minor;
119 struct list_head disks;
NeilBrown850b2b42006-10-03 01:15:46 -0700120 unsigned long flags;
121#define MD_CHANGE_DEVS 0 /* Some device status has changed */
122#define MD_CHANGE_CLEAN 1 /* transition to or from 'clean' */
123#define MD_CHANGE_PENDING 2 /* superblock update in progress */
124
NeilBrown409c57f2009-03-31 14:39:39 +1100125 int suspended;
126 atomic_t active_io;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700127 int ro;
128
129 struct gendisk *gendisk;
130
NeilBrowneae17012005-11-08 21:39:23 -0800131 struct kobject kobj;
NeilBrownd3374822009-01-09 08:31:10 +1100132 int hold_active;
133#define UNTIL_IOCTL 1
NeilBrownefeb53c2009-01-09 08:31:10 +1100134#define UNTIL_STOP 2
NeilBrowneae17012005-11-08 21:39:23 -0800135
Linus Torvalds1da177e2005-04-16 15:20:36 -0700136 /* Superblock information */
137 int major_version,
138 minor_version,
139 patch_version;
140 int persistent;
NeilBrowne6910632008-02-06 01:39:51 -0800141 int external; /* metadata is
142 * managed externally */
143 char metadata_type[17]; /* externally set*/
Andre Noll9d8f0362009-06-18 08:45:01 +1000144 int chunk_sectors;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700145 time_t ctime, utime;
146 int level, layout;
NeilBrownd9d166c2006-01-06 00:20:51 -0800147 char clevel[16];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700148 int raid_disks;
149 int max_disks;
Andre Noll58c0fed2009-03-31 14:33:13 +1100150 sector_t dev_sectors; /* used size of
151 * component devices */
Andre Nollf233ea52008-07-21 17:05:22 +1000152 sector_t array_sectors; /* exported array size */
Dan Williamsb522adc2009-03-31 15:00:31 +1100153 int external_size; /* size managed
154 * externally */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700155 __u64 events;
156
157 char uuid[16];
158
NeilBrownf6705572006-03-27 01:18:11 -0800159 /* If the array is being reshaped, we need to record the
160 * new shape and an indication of where we are up to.
161 * This is written to the superblock.
162 * If reshape_position is MaxSector, then no reshape is happening (yet).
163 */
164 sector_t reshape_position;
Andre Noll664e7c42009-06-18 08:45:27 +1000165 int delta_disks, new_level, new_layout;
166 int new_chunk_sectors;
NeilBrownf6705572006-03-27 01:18:11 -0800167
Linus Torvalds1da177e2005-04-16 15:20:36 -0700168 struct mdk_thread_s *thread; /* management thread */
169 struct mdk_thread_s *sync_thread; /* doing resync or reconstruct */
NeilBrownff4e8d92006-07-10 04:44:16 -0700170 sector_t curr_resync; /* last block scheduled */
NeilBrown97e4f422009-03-31 14:33:13 +1100171 /* As resync requests can complete out of order, we cannot easily track
172 * how much resync has been completed. So we occasionally pause until
173 * everything completes, then set curr_resync_completed to curr_resync.
174 * As such it may be well behind the real resync mark, but it is a value
175 * we are certain of.
176 */
177 sector_t curr_resync_completed;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700178 unsigned long resync_mark; /* a recent timestamp */
179 sector_t resync_mark_cnt;/* blocks written at resync_mark */
NeilBrownff4e8d92006-07-10 04:44:16 -0700180 sector_t curr_mark_cnt; /* blocks scheduled now */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700181
182 sector_t resync_max_sectors; /* may be set by personality */
NeilBrown9d888832005-11-08 21:39:26 -0800183
184 sector_t resync_mismatches; /* count of sectors where
185 * parity/replica mismatch found
186 */
NeilBrowne464eaf2006-03-27 01:18:14 -0800187
188 /* allow user-space to request suspension of IO to regions of the array */
189 sector_t suspend_lo;
190 sector_t suspend_hi;
NeilBrown88202a02006-01-06 00:21:36 -0800191 /* if zero, use the system-wide default */
192 int sync_speed_min;
193 int sync_speed_max;
194
Bernd Schubert90b08712008-05-23 13:04:38 -0700195 /* resync even though the same disks are shared among md-devices */
196 int parallel_resync;
197
NeilBrown6ff8d8e2006-01-06 00:20:15 -0800198 int ok_start_degraded;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700199 /* recovery/resync flags
200 * NEEDED: we might need to start a resync/recover
201 * RUNNING: a thread is running, or about to be started
202 * SYNC: actually doing a resync, not a recovery
Neil Brown72a23c22008-06-28 08:31:41 +1000203 * RECOVER: doing recovery, or need to try it.
NeilBrowndfc70642008-05-23 13:04:39 -0700204 * INTR: resync needs to be aborted for some reason
Linus Torvalds1da177e2005-04-16 15:20:36 -0700205 * DONE: thread is done and is waiting to be reaped
NeilBrown24dd4692005-11-08 21:39:26 -0800206 * REQUEST: user-space has requested a sync (used with SYNC)
Anand Gadiyar411c9402009-07-07 15:24:23 +0530207 * CHECK: user-space request for check-only, no repair
NeilBrownccfcc3c2006-03-27 01:18:09 -0800208 * RESHAPE: A reshape is happening
209 *
210 * If neither SYNC or RESHAPE are set, then it is a recovery.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700211 */
212#define MD_RECOVERY_RUNNING 0
213#define MD_RECOVERY_SYNC 1
Neil Brown72a23c22008-06-28 08:31:41 +1000214#define MD_RECOVERY_RECOVER 2
Linus Torvalds1da177e2005-04-16 15:20:36 -0700215#define MD_RECOVERY_INTR 3
216#define MD_RECOVERY_DONE 4
217#define MD_RECOVERY_NEEDED 5
NeilBrown24dd4692005-11-08 21:39:26 -0800218#define MD_RECOVERY_REQUESTED 6
219#define MD_RECOVERY_CHECK 7
NeilBrownccfcc3c2006-03-27 01:18:09 -0800220#define MD_RECOVERY_RESHAPE 8
NeilBrown5fd6c1d2006-06-26 00:27:40 -0700221#define MD_RECOVERY_FROZEN 9
222
Linus Torvalds1da177e2005-04-16 15:20:36 -0700223 unsigned long recovery;
NeilBrown4044ba52009-01-09 08:31:11 +1100224 int recovery_disabled; /* if we detect that recovery
225 * will always fail, set this
226 * so we don't loop trying */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700227
228 int in_sync; /* know to not need resync */
NeilBrownc8c00a62009-08-10 12:50:52 +1000229 /* 'open_mutex' avoids races between 'md_open' and 'do_md_stop', so
230 * that we are never stopping an array while it is open.
231 * 'reconfig_mutex' protects all other reconfiguration.
232 * These locks are separate due to conflicting interactions
233 * with bdev->bd_mutex.
234 * Lock ordering is:
235 * reconfig_mutex -> bd_mutex : e.g. do_md_run -> revalidate_disk
236 * bd_mutex -> open_mutex: e.g. __blkdev_get -> md_open
237 */
238 struct mutex open_mutex;
NeilBrowndf5b89b2006-03-27 01:18:20 -0800239 struct mutex reconfig_mutex;
NeilBrownf2ea68c2008-07-21 17:05:25 +1000240 atomic_t active; /* general refcount */
241 atomic_t openers; /* number of active opens */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700242
Linus Torvalds44ce6292007-05-09 18:51:36 -0700243 int changed; /* true if we might need to reread partition info */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700244 int degraded; /* whether md should consider
245 * adding a spare
246 */
NeilBrowna9701a32005-11-08 21:39:34 -0800247 int barriers_work; /* initialised to true, cleared as soon
248 * as a barrier request to slave
249 * fails. Only supported
250 */
251 struct bio *biolist; /* bios that need to be retried
252 * because BIO_RW_BARRIER is not supported
253 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700254
255 atomic_t recovery_active; /* blocks scheduled, but not written */
256 wait_queue_head_t recovery_wait;
257 sector_t recovery_cp;
Neil Brown5e96ee62008-06-28 08:31:24 +1000258 sector_t resync_min; /* user requested sync
259 * starts here */
NeilBrownc6207272008-02-06 01:39:52 -0800260 sector_t resync_max; /* resync should pause
261 * when it gets here */
NeilBrown06d91a52005-06-21 17:17:12 -0700262
NeilBrownb62b7592008-10-21 13:25:21 +1100263 struct sysfs_dirent *sysfs_state; /* handle for 'array_state'
264 * file in sysfs.
265 */
NeilBrown0c3573f2009-01-09 08:31:05 +1100266 struct sysfs_dirent *sysfs_action; /* handle for 'sync_action' */
NeilBrownb62b7592008-10-21 13:25:21 +1100267
NeilBrownd3374822009-01-09 08:31:10 +1100268 struct work_struct del_work; /* used for delayed sysfs removal */
269
NeilBrown06d91a52005-06-21 17:17:12 -0700270 spinlock_t write_lock;
NeilBrown3d310eb2005-06-21 17:17:26 -0700271 wait_queue_head_t sb_wait; /* for waiting on superblock updates */
NeilBrown7bfa19f2005-06-21 17:17:28 -0700272 atomic_t pending_writes; /* number of active superblock writes */
NeilBrown06d91a52005-06-21 17:17:12 -0700273
Linus Torvalds1da177e2005-04-16 15:20:36 -0700274 unsigned int safemode; /* if set, update "clean" superblock
275 * when no writes pending.
276 */
277 unsigned int safemode_delay;
278 struct timer_list safemode_timer;
279 atomic_t writes_pending;
Jens Axboe165125e2007-07-24 09:28:11 +0200280 struct request_queue *queue; /* for plugging ... */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700281
NeilBrown4b6d2872005-09-09 16:23:47 -0700282 atomic_t write_behind; /* outstanding async IO */
283 unsigned int max_write_behind; /* 0 = sync */
284
NeilBrown32a76272005-06-21 17:17:14 -0700285 struct bitmap *bitmap; /* the bitmap for the device */
NeilBrownc3d97142009-12-14 12:49:52 +1100286 struct {
287 struct file *file; /* the bitmap file */
NeilBrownf6af9492009-12-14 12:49:54 +1100288 loff_t offset; /* offset from superblock of
NeilBrownc3d97142009-12-14 12:49:52 +1100289 * start of bitmap. May be
290 * negative, but not '0'
NeilBrownf6af9492009-12-14 12:49:54 +1100291 * For external metadata, offset
292 * from start of device.
NeilBrownc3d97142009-12-14 12:49:52 +1100293 */
NeilBrownf6af9492009-12-14 12:49:54 +1100294 loff_t default_offset; /* this is the offset to use when
NeilBrownc3d97142009-12-14 12:49:52 +1100295 * hot-adding a bitmap. It should
296 * eventually be settable by sysfs.
297 */
298 struct mutex mutex;
NeilBrown42a04b52009-12-14 12:49:53 +1100299 unsigned long chunksize;
300 unsigned long daemon_sleep; /* how many seconds between updates? */
301 unsigned long max_write_behind; /* write-behind mode */
NeilBrownece5cff2009-12-14 12:49:56 +1100302 int external;
NeilBrownc3d97142009-12-14 12:49:52 +1100303 } bitmap_info;
NeilBrown32a76272005-06-21 17:17:14 -0700304
Robert Becker1e509152009-12-14 12:49:58 +1100305 atomic_t max_corr_read_errors; /* max read retries */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700306 struct list_head all_mddevs;
NeilBrowna2826aa2009-12-14 12:49:49 +1100307
308 /* Generic barrier handling.
309 * If there is a pending barrier request, all other
310 * writes are blocked while the devices are flushed.
311 * The last to finish a flush schedules a worker to
312 * submit the barrier request (without the barrier flag),
313 * then submit more flush requests.
314 */
315 struct bio *barrier;
316 atomic_t flush_pending;
317 struct work_struct barrier_work;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700318};
319
320
321static inline void rdev_dec_pending(mdk_rdev_t *rdev, mddev_t *mddev)
322{
NeilBrownb2d444d2005-11-08 21:39:31 -0800323 int faulty = test_bit(Faulty, &rdev->flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700324 if (atomic_dec_and_test(&rdev->nr_pending) && faulty)
325 set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
326}
327
328static inline void md_sync_acct(struct block_device *bdev, unsigned long nr_sectors)
329{
330 atomic_add(nr_sectors, &bdev->bd_contains->bd_disk->sync_io);
331}
332
NeilBrown2604b702006-01-06 00:20:36 -0800333struct mdk_personality
Linus Torvalds1da177e2005-04-16 15:20:36 -0700334{
335 char *name;
NeilBrown2604b702006-01-06 00:20:36 -0800336 int level;
337 struct list_head list;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700338 struct module *owner;
Jens Axboe165125e2007-07-24 09:28:11 +0200339 int (*make_request)(struct request_queue *q, struct bio *bio);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700340 int (*run)(mddev_t *mddev);
341 int (*stop)(mddev_t *mddev);
342 void (*status)(struct seq_file *seq, mddev_t *mddev);
343 /* error_handler must set ->faulty and clear ->in_sync
344 * if appropriate, and should abort recovery if needed
345 */
346 void (*error_handler)(mddev_t *mddev, mdk_rdev_t *rdev);
347 int (*hot_add_disk) (mddev_t *mddev, mdk_rdev_t *rdev);
348 int (*hot_remove_disk) (mddev_t *mddev, int number);
349 int (*spare_active) (mddev_t *mddev);
NeilBrown57afd892005-06-21 17:17:13 -0700350 sector_t (*sync_request)(mddev_t *mddev, sector_t sector_nr, int *skipped, int go_faster);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700351 int (*resize) (mddev_t *mddev, sector_t sectors);
Dan Williams80c3a6c2009-03-17 18:10:40 -0700352 sector_t (*size) (mddev_t *mddev, sector_t sectors, int raid_disks);
NeilBrown63c70c42006-03-27 01:18:13 -0800353 int (*check_reshape) (mddev_t *mddev);
354 int (*start_reshape) (mddev_t *mddev);
NeilBrowncea9c222009-03-31 15:15:05 +1100355 void (*finish_reshape) (mddev_t *mddev);
NeilBrown36fa3062005-09-09 16:23:45 -0700356 /* quiesce moves between quiescence states
357 * 0 - fully active
358 * 1 - no new requests allowed
359 * others - reserved
360 */
361 void (*quiesce) (mddev_t *mddev, int state);
NeilBrown245f46c2009-03-31 14:39:39 +1100362 /* takeover is used to transition an array from one
363 * personality to another. The new personality must be able
364 * to handle the data in the current layout.
365 * e.g. 2drive raid1 -> 2drive raid5
366 * ndrive raid5 -> degraded n+1drive raid6 with special layout
367 * If the takeover succeeds, a new 'private' structure is returned.
368 * This needs to be installed and then ->run used to activate the
369 * array.
370 */
371 void *(*takeover) (mddev_t *mddev);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700372};
373
374
NeilBrown007583c2005-11-08 21:39:30 -0800375struct md_sysfs_entry {
376 struct attribute attr;
377 ssize_t (*show)(mddev_t *, char *);
378 ssize_t (*store)(mddev_t *, const char *, size_t);
379};
NeilBrown43a70502009-12-14 12:49:55 +1100380extern struct attribute_group md_bitmap_group;
NeilBrown007583c2005-11-08 21:39:30 -0800381
Linus Torvalds1da177e2005-04-16 15:20:36 -0700382static inline char * mdname (mddev_t * mddev)
383{
384 return mddev->gendisk ? mddev->gendisk->disk_name : "mdX";
385}
386
Linus Torvalds1da177e2005-04-16 15:20:36 -0700387/*
388 * iterates through some rdev ringlist. It's safe to remove the
389 * current 'rdev'. Dont touch 'tmp' though.
390 */
Cheng Renquan159ec1f2009-01-09 08:31:08 +1100391#define rdev_for_each_list(rdev, tmp, head) \
392 list_for_each_entry_safe(rdev, tmp, head, same_set)
393
Linus Torvalds1da177e2005-04-16 15:20:36 -0700394/*
395 * iterates through the 'same array disks' ringlist
396 */
NeilBrownd089c6a2008-02-06 01:39:59 -0800397#define rdev_for_each(rdev, tmp, mddev) \
Cheng Renquan159ec1f2009-01-09 08:31:08 +1100398 list_for_each_entry_safe(rdev, tmp, &((mddev)->disks), same_set)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700399
NeilBrown4b809912008-07-21 17:05:25 +1000400#define rdev_for_each_rcu(rdev, mddev) \
401 list_for_each_entry_rcu(rdev, &((mddev)->disks), same_set)
402
Linus Torvalds1da177e2005-04-16 15:20:36 -0700403typedef struct mdk_thread_s {
404 void (*run) (mddev_t *mddev);
405 mddev_t *mddev;
406 wait_queue_head_t wqueue;
407 unsigned long flags;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700408 struct task_struct *tsk;
NeilBrown32a76272005-06-21 17:17:14 -0700409 unsigned long timeout;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700410} mdk_thread_t;
411
412#define THREAD_WAKEUP 0
413
414#define __wait_event_lock_irq(wq, condition, lock, cmd) \
415do { \
416 wait_queue_t __wait; \
417 init_waitqueue_entry(&__wait, current); \
418 \
419 add_wait_queue(&wq, &__wait); \
420 for (;;) { \
421 set_current_state(TASK_UNINTERRUPTIBLE); \
422 if (condition) \
423 break; \
424 spin_unlock_irq(&lock); \
425 cmd; \
426 schedule(); \
427 spin_lock_irq(&lock); \
428 } \
429 current->state = TASK_RUNNING; \
430 remove_wait_queue(&wq, &__wait); \
431} while (0)
432
433#define wait_event_lock_irq(wq, condition, lock, cmd) \
434do { \
435 if (condition) \
436 break; \
437 __wait_event_lock_irq(wq, condition, lock, cmd); \
438} while (0)
439
NeilBrown1345b1d2006-01-06 00:20:40 -0800440static inline void safe_put_page(struct page *p)
441{
442 if (p) put_page(p);
443}
444
NeilBrown92022952009-03-31 14:33:13 +1100445extern int register_md_personality(struct mdk_personality *p);
446extern int unregister_md_personality(struct mdk_personality *p);
447extern mdk_thread_t * md_register_thread(void (*run) (mddev_t *mddev),
448 mddev_t *mddev, const char *name);
449extern void md_unregister_thread(mdk_thread_t *thread);
450extern void md_wakeup_thread(mdk_thread_t *thread);
451extern void md_check_recovery(mddev_t *mddev);
452extern void md_write_start(mddev_t *mddev, struct bio *bi);
453extern void md_write_end(mddev_t *mddev);
454extern void md_done_sync(mddev_t *mddev, int blocks, int ok);
455extern void md_error(mddev_t *mddev, mdk_rdev_t *rdev);
456
NeilBrown3fa841d2009-09-23 18:10:29 +1000457extern int mddev_congested(mddev_t *mddev, int bits);
NeilBrowna2826aa2009-12-14 12:49:49 +1100458extern void md_barrier_request(mddev_t *mddev, struct bio *bio);
NeilBrown92022952009-03-31 14:33:13 +1100459extern void md_super_write(mddev_t *mddev, mdk_rdev_t *rdev,
460 sector_t sector, int size, struct page *page);
461extern void md_super_wait(mddev_t *mddev);
462extern int sync_page_io(struct block_device *bdev, sector_t sector, int size,
463 struct page *page, int rw);
464extern void md_do_sync(mddev_t *mddev);
465extern void md_new_event(mddev_t *mddev);
466extern int md_allow_write(mddev_t *mddev);
467extern void md_wait_for_blocked_rdev(mdk_rdev_t *rdev, mddev_t *mddev);
Dan Williams1f403622009-03-31 14:59:03 +1100468extern void md_set_array_sectors(mddev_t *mddev, sector_t array_sectors);
Andre Noll0894cc32009-06-18 08:49:23 +1000469extern int md_check_no_bitmap(mddev_t *mddev);
Andre Nollac5e7112009-08-03 10:59:47 +1000470extern int md_integrity_register(mddev_t *mddev);
NeilBrown72e02072009-12-14 12:49:55 +1100471extern void md_integrity_add_rdev(mdk_rdev_t *rdev, mddev_t *mddev);
472extern int strict_strtoul_scaled(const char *cp, unsigned long *res, int scale);
NeilBrown43a70502009-12-14 12:49:55 +1100473extern void restore_bitmap_write_access(struct file *file);
Christoph Hellwig63fe0812009-04-14 12:01:53 +1000474
475#endif /* _MD_MD_H */