blob: 1c26c7a08ae6c053524cbee6c751c84935c5cfe3 [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
Jonathan Brassowa6ff7e02011-01-14 09:14:34 +110043 /*
44 * If meta_bdev is non-NULL, it means that a separate device is
45 * being used to store the metadata (superblock/bitmap) which
46 * would otherwise be contained on the same device as the data (bdev).
47 */
48 struct block_device *meta_bdev;
Linus Torvalds1da177e2005-04-16 15:20:36 -070049 struct block_device *bdev; /* block device handle */
50
51 struct page *sb_page;
52 int sb_loaded;
NeilBrown42543762006-06-26 00:27:57 -070053 __u64 sb_events;
Linus Torvalds1da177e2005-04-16 15:20:36 -070054 sector_t data_offset; /* start of data in array */
Andre Noll0f420352008-07-11 22:02:23 +100055 sector_t sb_start; /* offset of the super block (in 512byte sectors) */
NeilBrown0002b272005-09-09 16:23:53 -070056 int sb_size; /* bytes in the superblock */
Linus Torvalds1da177e2005-04-16 15:20:36 -070057 int preferred_minor; /* autorun support */
58
NeilBrown86e6ffd2005-11-08 21:39:24 -080059 struct kobject kobj;
60
Linus Torvalds1da177e2005-04-16 15:20:36 -070061 /* A device can be in one of three states based on two flags:
62 * Not working: faulty==1 in_sync==0
63 * Fully working: faulty==0 in_sync==1
64 * Working, but not
65 * in sync with array
66 * faulty==0 in_sync==0
67 *
68 * It can never have faulty==1, in_sync==1
69 * This reduces the burden of testing multiple flags in many cases
70 */
Linus Torvalds1da177e2005-04-16 15:20:36 -070071
NeilBrownb2d444d2005-11-08 21:39:31 -080072 unsigned long flags;
73#define Faulty 1 /* device is known to have a fault */
74#define In_sync 2 /* device is in_sync with rest of array */
NeilBrown8ddf9ef2005-09-09 16:23:45 -070075#define WriteMostly 4 /* Avoid reading if at all possible */
NeilBrownd0fae182008-03-04 14:29:31 -080076#define AutoDetected 7 /* added by auto-detect */
Lucas De Marchi25985ed2011-03-30 22:57:33 -030077#define Blocked 8 /* An error occurred on an externally
Dan Williams6bfe0b42008-04-30 00:52:32 -070078 * managed array, don't allow writes
79 * until it is cleared */
80 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 */
NeilBrowne93f68a2010-06-15 09:36:03 +010084 int new_raid_disk; /* role that the device will have in
85 * the array after a level-change completes.
86 */
NeilBrown41158c72005-06-21 17:17:25 -070087 int saved_raid_disk; /* role that device used to have in the
88 * array and could again if we did a partial
89 * resync from the bitmap
90 */
NeilBrown5fd6c1d2006-06-26 00:27:40 -070091 sector_t recovery_offset;/* If this device has been partially
92 * recovered, this is where we were
93 * up to.
94 */
Linus Torvalds1da177e2005-04-16 15:20:36 -070095
96 atomic_t nr_pending; /* number of pending requests.
97 * only maintained for arrays that
98 * support hot removal
99 */
NeilBrownba22dcb2005-11-08 21:39:31 -0800100 atomic_t read_errors; /* number of consecutive read errors that
101 * we have tried to ignore.
102 */
Robert Becker1e509152009-12-14 12:49:58 +1100103 struct timespec last_read_error; /* monotonic time since our
104 * last read error
105 */
NeilBrown4dbcdc72006-01-06 00:20:52 -0800106 atomic_t corrected_errors; /* number of corrected read errors,
107 * for reporting to userspace and storing
108 * in superblock.
109 */
NeilBrown5792a282007-04-04 19:08:18 -0700110 struct work_struct del_work; /* used for delayed sysfs removal */
NeilBrown3c0ee632008-10-21 13:25:28 +1100111
112 struct sysfs_dirent *sysfs_state; /* handle for 'state'
113 * sysfs entry */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700114};
115
Linus Torvalds1da177e2005-04-16 15:20:36 -0700116struct mddev_s
117{
118 void *private;
NeilBrown2604b702006-01-06 00:20:36 -0800119 struct mdk_personality *pers;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700120 dev_t unit;
121 int md_minor;
122 struct list_head disks;
NeilBrown850b2b42006-10-03 01:15:46 -0700123 unsigned long flags;
124#define MD_CHANGE_DEVS 0 /* Some device status has changed */
125#define MD_CHANGE_CLEAN 1 /* transition to or from 'clean' */
NeilBrown070dc6d2010-08-30 17:33:34 +1000126#define MD_CHANGE_PENDING 2 /* switch from 'clean' to 'active' in progress */
Jonathan Brassow9c810752011-06-08 17:59:30 -0500127#define MD_ARRAY_FIRST_USE 3 /* First use of array, needs initialization */
NeilBrown850b2b42006-10-03 01:15:46 -0700128
NeilBrown409c57f2009-03-31 14:39:39 +1100129 int suspended;
130 atomic_t active_io;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700131 int ro;
NeilBrownbb4f1e92010-08-08 21:18:03 +1000132 int sysfs_active; /* set when sysfs deletes
133 * are happening, so run/
134 * takeover/stop are not safe
135 */
NeilBrown0ca69882011-01-14 09:14:33 +1100136 int ready; /* See when safe to pass
137 * IO requests down */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700138 struct gendisk *gendisk;
139
NeilBrowneae17012005-11-08 21:39:23 -0800140 struct kobject kobj;
NeilBrownd3374822009-01-09 08:31:10 +1100141 int hold_active;
142#define UNTIL_IOCTL 1
NeilBrownefeb53c2009-01-09 08:31:10 +1100143#define UNTIL_STOP 2
NeilBrowneae17012005-11-08 21:39:23 -0800144
Linus Torvalds1da177e2005-04-16 15:20:36 -0700145 /* Superblock information */
146 int major_version,
147 minor_version,
148 patch_version;
149 int persistent;
NeilBrowne6910632008-02-06 01:39:51 -0800150 int external; /* metadata is
151 * managed externally */
152 char metadata_type[17]; /* externally set*/
Andre Noll9d8f0362009-06-18 08:45:01 +1000153 int chunk_sectors;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700154 time_t ctime, utime;
155 int level, layout;
NeilBrownd9d166c2006-01-06 00:20:51 -0800156 char clevel[16];
Linus Torvalds1da177e2005-04-16 15:20:36 -0700157 int raid_disks;
158 int max_disks;
Andre Noll58c0fed2009-03-31 14:33:13 +1100159 sector_t dev_sectors; /* used size of
160 * component devices */
Andre Nollf233ea52008-07-21 17:05:22 +1000161 sector_t array_sectors; /* exported array size */
Dan Williamsb522adc2009-03-31 15:00:31 +1100162 int external_size; /* size managed
163 * externally */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700164 __u64 events;
NeilBrowna8707c02010-05-18 09:28:43 +1000165 /* If the last 'event' was simply a clean->dirty transition, and
166 * we didn't write it to the spares, then it is safe and simple
167 * to just decrement the event count on a dirty->clean transition.
168 * So we record that possibility here.
169 */
170 int can_decrease_events;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700171
172 char uuid[16];
173
NeilBrownf6705572006-03-27 01:18:11 -0800174 /* If the array is being reshaped, we need to record the
175 * new shape and an indication of where we are up to.
176 * This is written to the superblock.
177 * If reshape_position is MaxSector, then no reshape is happening (yet).
178 */
179 sector_t reshape_position;
Andre Noll664e7c42009-06-18 08:45:27 +1000180 int delta_disks, new_level, new_layout;
181 int new_chunk_sectors;
NeilBrownf6705572006-03-27 01:18:11 -0800182
NeilBrown97658cd2011-04-18 18:25:42 +1000183 atomic_t plug_cnt; /* If device is expecting
184 * more bios soon.
185 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700186 struct mdk_thread_s *thread; /* management thread */
187 struct mdk_thread_s *sync_thread; /* doing resync or reconstruct */
NeilBrownff4e8d92006-07-10 04:44:16 -0700188 sector_t curr_resync; /* last block scheduled */
NeilBrown97e4f422009-03-31 14:33:13 +1100189 /* As resync requests can complete out of order, we cannot easily track
190 * how much resync has been completed. So we occasionally pause until
191 * everything completes, then set curr_resync_completed to curr_resync.
192 * As such it may be well behind the real resync mark, but it is a value
193 * we are certain of.
194 */
195 sector_t curr_resync_completed;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700196 unsigned long resync_mark; /* a recent timestamp */
197 sector_t resync_mark_cnt;/* blocks written at resync_mark */
NeilBrownff4e8d92006-07-10 04:44:16 -0700198 sector_t curr_mark_cnt; /* blocks scheduled now */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700199
200 sector_t resync_max_sectors; /* may be set by personality */
NeilBrown9d888832005-11-08 21:39:26 -0800201
202 sector_t resync_mismatches; /* count of sectors where
203 * parity/replica mismatch found
204 */
NeilBrowne464eaf2006-03-27 01:18:14 -0800205
206 /* allow user-space to request suspension of IO to regions of the array */
207 sector_t suspend_lo;
208 sector_t suspend_hi;
NeilBrown88202a02006-01-06 00:21:36 -0800209 /* if zero, use the system-wide default */
210 int sync_speed_min;
211 int sync_speed_max;
212
Bernd Schubert90b08712008-05-23 13:04:38 -0700213 /* resync even though the same disks are shared among md-devices */
214 int parallel_resync;
215
NeilBrown6ff8d8ec2006-01-06 00:20:15 -0800216 int ok_start_degraded;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700217 /* recovery/resync flags
218 * NEEDED: we might need to start a resync/recover
219 * RUNNING: a thread is running, or about to be started
220 * SYNC: actually doing a resync, not a recovery
Neil Brown72a23c22008-06-28 08:31:41 +1000221 * RECOVER: doing recovery, or need to try it.
NeilBrowndfc70642008-05-23 13:04:39 -0700222 * INTR: resync needs to be aborted for some reason
Linus Torvalds1da177e2005-04-16 15:20:36 -0700223 * DONE: thread is done and is waiting to be reaped
NeilBrown24dd4692005-11-08 21:39:26 -0800224 * REQUEST: user-space has requested a sync (used with SYNC)
Anand Gadiyar411c9402009-07-07 15:24:23 +0530225 * CHECK: user-space request for check-only, no repair
NeilBrownccfcc3c2006-03-27 01:18:09 -0800226 * RESHAPE: A reshape is happening
227 *
228 * If neither SYNC or RESHAPE are set, then it is a recovery.
Linus Torvalds1da177e2005-04-16 15:20:36 -0700229 */
230#define MD_RECOVERY_RUNNING 0
231#define MD_RECOVERY_SYNC 1
Neil Brown72a23c22008-06-28 08:31:41 +1000232#define MD_RECOVERY_RECOVER 2
Linus Torvalds1da177e2005-04-16 15:20:36 -0700233#define MD_RECOVERY_INTR 3
234#define MD_RECOVERY_DONE 4
235#define MD_RECOVERY_NEEDED 5
NeilBrown24dd4692005-11-08 21:39:26 -0800236#define MD_RECOVERY_REQUESTED 6
237#define MD_RECOVERY_CHECK 7
NeilBrownccfcc3c2006-03-27 01:18:09 -0800238#define MD_RECOVERY_RESHAPE 8
NeilBrown5fd6c1d2006-06-26 00:27:40 -0700239#define MD_RECOVERY_FROZEN 9
240
Linus Torvalds1da177e2005-04-16 15:20:36 -0700241 unsigned long recovery;
NeilBrown4044ba52009-01-09 08:31:11 +1100242 int recovery_disabled; /* if we detect that recovery
243 * will always fail, set this
244 * so we don't loop trying */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700245
246 int in_sync; /* know to not need resync */
NeilBrownc8c00a62009-08-10 12:50:52 +1000247 /* 'open_mutex' avoids races between 'md_open' and 'do_md_stop', so
248 * that we are never stopping an array while it is open.
249 * 'reconfig_mutex' protects all other reconfiguration.
250 * These locks are separate due to conflicting interactions
251 * with bdev->bd_mutex.
252 * Lock ordering is:
253 * reconfig_mutex -> bd_mutex : e.g. do_md_run -> revalidate_disk
254 * bd_mutex -> open_mutex: e.g. __blkdev_get -> md_open
255 */
256 struct mutex open_mutex;
NeilBrowndf5b89b2006-03-27 01:18:20 -0800257 struct mutex reconfig_mutex;
NeilBrownf2ea68c2008-07-21 17:05:25 +1000258 atomic_t active; /* general refcount */
259 atomic_t openers; /* number of active opens */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700260
NeilBrownf0b4f7e2011-02-24 17:26:41 +1100261 int changed; /* True if we might need to
262 * reread partition info */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700263 int degraded; /* whether md should consider
264 * adding a spare
265 */
266
267 atomic_t recovery_active; /* blocks scheduled, but not written */
268 wait_queue_head_t recovery_wait;
269 sector_t recovery_cp;
Neil Brown5e96ee62008-06-28 08:31:24 +1000270 sector_t resync_min; /* user requested sync
271 * starts here */
NeilBrownc6207272008-02-06 01:39:52 -0800272 sector_t resync_max; /* resync should pause
273 * when it gets here */
NeilBrown06d91a52005-06-21 17:17:12 -0700274
NeilBrownb62b7592008-10-21 13:25:21 +1100275 struct sysfs_dirent *sysfs_state; /* handle for 'array_state'
276 * file in sysfs.
277 */
NeilBrown0c3573f2009-01-09 08:31:05 +1100278 struct sysfs_dirent *sysfs_action; /* handle for 'sync_action' */
NeilBrownb62b7592008-10-21 13:25:21 +1100279
NeilBrownd3374822009-01-09 08:31:10 +1100280 struct work_struct del_work; /* used for delayed sysfs removal */
281
NeilBrown06d91a52005-06-21 17:17:12 -0700282 spinlock_t write_lock;
NeilBrown3d310eb2005-06-21 17:17:26 -0700283 wait_queue_head_t sb_wait; /* for waiting on superblock updates */
NeilBrown7bfa19f2005-06-21 17:17:28 -0700284 atomic_t pending_writes; /* number of active superblock writes */
NeilBrown06d91a52005-06-21 17:17:12 -0700285
Linus Torvalds1da177e2005-04-16 15:20:36 -0700286 unsigned int safemode; /* if set, update "clean" superblock
287 * when no writes pending.
288 */
289 unsigned int safemode_delay;
290 struct timer_list safemode_timer;
291 atomic_t writes_pending;
Jens Axboe165125e2007-07-24 09:28:11 +0200292 struct request_queue *queue; /* for plugging ... */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700293
NeilBrown32a76272005-06-21 17:17:14 -0700294 struct bitmap *bitmap; /* the bitmap for the device */
NeilBrownc3d97142009-12-14 12:49:52 +1100295 struct {
296 struct file *file; /* the bitmap file */
NeilBrownf6af9492009-12-14 12:49:54 +1100297 loff_t offset; /* offset from superblock of
NeilBrownc3d97142009-12-14 12:49:52 +1100298 * start of bitmap. May be
299 * negative, but not '0'
NeilBrownf6af9492009-12-14 12:49:54 +1100300 * For external metadata, offset
301 * from start of device.
NeilBrownc3d97142009-12-14 12:49:52 +1100302 */
NeilBrownf6af9492009-12-14 12:49:54 +1100303 loff_t default_offset; /* this is the offset to use when
NeilBrownc3d97142009-12-14 12:49:52 +1100304 * hot-adding a bitmap. It should
305 * eventually be settable by sysfs.
306 */
NeilBrowne384e582010-06-01 19:37:34 +1000307 /* When md is serving under dm, it might use a
308 * dirty_log to store the bits.
309 */
310 struct dm_dirty_log *log;
311
NeilBrownc3d97142009-12-14 12:49:52 +1100312 struct mutex mutex;
NeilBrown42a04b52009-12-14 12:49:53 +1100313 unsigned long chunksize;
NeilBrownac2f40b2010-06-01 19:37:31 +1000314 unsigned long daemon_sleep; /* how many jiffies between updates? */
NeilBrown42a04b52009-12-14 12:49:53 +1100315 unsigned long max_write_behind; /* write-behind mode */
NeilBrownece5cff2009-12-14 12:49:56 +1100316 int external;
NeilBrownc3d97142009-12-14 12:49:52 +1100317 } bitmap_info;
NeilBrown32a76272005-06-21 17:17:14 -0700318
Robert Becker1e509152009-12-14 12:49:58 +1100319 atomic_t max_corr_read_errors; /* max read retries */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700320 struct list_head all_mddevs;
NeilBrowna2826aa2009-12-14 12:49:49 +1100321
NeilBrowna64c8762010-04-14 17:15:37 +1000322 struct attribute_group *to_remove;
NeilBrown252ac522010-06-01 19:37:29 +1000323
NeilBrowna167f662010-10-26 18:31:13 +1100324 struct bio_set *bio_set;
325
Tejun Heoe9c74692010-09-03 11:56:18 +0200326 /* Generic flush handling.
327 * The last to finish preflush schedules a worker to submit
328 * the rest of the request (without the REQ_FLUSH flag).
NeilBrowna2826aa2009-12-14 12:49:49 +1100329 */
Tejun Heoe9c74692010-09-03 11:56:18 +0200330 struct bio *flush_bio;
NeilBrowna2826aa2009-12-14 12:49:49 +1100331 atomic_t flush_pending;
Tejun Heoe9c74692010-09-03 11:56:18 +0200332 struct work_struct flush_work;
NeilBrown768a4182010-07-26 11:49:55 +1000333 struct work_struct event_work; /* used by dm to report failure event */
Jonathan Brassow076f9682011-06-07 17:51:30 -0500334 void (*sync_super)(mddev_t *mddev, mdk_rdev_t *rdev);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700335};
336
337
338static inline void rdev_dec_pending(mdk_rdev_t *rdev, mddev_t *mddev)
339{
NeilBrownb2d444d2005-11-08 21:39:31 -0800340 int faulty = test_bit(Faulty, &rdev->flags);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700341 if (atomic_dec_and_test(&rdev->nr_pending) && faulty)
342 set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
343}
344
345static inline void md_sync_acct(struct block_device *bdev, unsigned long nr_sectors)
346{
347 atomic_add(nr_sectors, &bdev->bd_contains->bd_disk->sync_io);
348}
349
NeilBrown2604b702006-01-06 00:20:36 -0800350struct mdk_personality
Linus Torvalds1da177e2005-04-16 15:20:36 -0700351{
352 char *name;
NeilBrown2604b702006-01-06 00:20:36 -0800353 int level;
354 struct list_head list;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700355 struct module *owner;
NeilBrown21a52c62010-04-01 15:02:13 +1100356 int (*make_request)(mddev_t *mddev, struct bio *bio);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700357 int (*run)(mddev_t *mddev);
358 int (*stop)(mddev_t *mddev);
359 void (*status)(struct seq_file *seq, mddev_t *mddev);
360 /* error_handler must set ->faulty and clear ->in_sync
361 * if appropriate, and should abort recovery if needed
362 */
363 void (*error_handler)(mddev_t *mddev, mdk_rdev_t *rdev);
364 int (*hot_add_disk) (mddev_t *mddev, mdk_rdev_t *rdev);
365 int (*hot_remove_disk) (mddev_t *mddev, int number);
366 int (*spare_active) (mddev_t *mddev);
NeilBrown57afd892005-06-21 17:17:13 -0700367 sector_t (*sync_request)(mddev_t *mddev, sector_t sector_nr, int *skipped, int go_faster);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700368 int (*resize) (mddev_t *mddev, sector_t sectors);
Dan Williams80c3a6c2009-03-17 18:10:40 -0700369 sector_t (*size) (mddev_t *mddev, sector_t sectors, int raid_disks);
NeilBrown63c70c42006-03-27 01:18:13 -0800370 int (*check_reshape) (mddev_t *mddev);
371 int (*start_reshape) (mddev_t *mddev);
NeilBrowncea9c222009-03-31 15:15:05 +1100372 void (*finish_reshape) (mddev_t *mddev);
NeilBrown36fa3062005-09-09 16:23:45 -0700373 /* quiesce moves between quiescence states
374 * 0 - fully active
375 * 1 - no new requests allowed
376 * others - reserved
377 */
378 void (*quiesce) (mddev_t *mddev, int state);
NeilBrown245f46c2009-03-31 14:39:39 +1100379 /* takeover is used to transition an array from one
380 * personality to another. The new personality must be able
381 * to handle the data in the current layout.
382 * e.g. 2drive raid1 -> 2drive raid5
383 * ndrive raid5 -> degraded n+1drive raid6 with special layout
384 * If the takeover succeeds, a new 'private' structure is returned.
385 * This needs to be installed and then ->run used to activate the
386 * array.
387 */
388 void *(*takeover) (mddev_t *mddev);
Linus Torvalds1da177e2005-04-16 15:20:36 -0700389};
390
391
NeilBrown007583c2005-11-08 21:39:30 -0800392struct md_sysfs_entry {
393 struct attribute attr;
394 ssize_t (*show)(mddev_t *, char *);
395 ssize_t (*store)(mddev_t *, const char *, size_t);
396};
NeilBrown43a70502009-12-14 12:49:55 +1100397extern struct attribute_group md_bitmap_group;
NeilBrown007583c2005-11-08 21:39:30 -0800398
NeilBrown00bcb4a2010-06-01 19:37:23 +1000399static inline struct sysfs_dirent *sysfs_get_dirent_safe(struct sysfs_dirent *sd, char *name)
400{
401 if (sd)
402 return sysfs_get_dirent(sd, NULL, name);
403 return sd;
404}
405static inline void sysfs_notify_dirent_safe(struct sysfs_dirent *sd)
406{
407 if (sd)
408 sysfs_notify_dirent(sd);
409}
410
Linus Torvalds1da177e2005-04-16 15:20:36 -0700411static inline char * mdname (mddev_t * mddev)
412{
413 return mddev->gendisk ? mddev->gendisk->disk_name : "mdX";
414}
415
Linus Torvalds1da177e2005-04-16 15:20:36 -0700416/*
417 * iterates through some rdev ringlist. It's safe to remove the
418 * current 'rdev'. Dont touch 'tmp' though.
419 */
Cheng Renquan159ec1f2009-01-09 08:31:08 +1100420#define rdev_for_each_list(rdev, tmp, head) \
421 list_for_each_entry_safe(rdev, tmp, head, same_set)
422
Linus Torvalds1da177e2005-04-16 15:20:36 -0700423/*
424 * iterates through the 'same array disks' ringlist
425 */
NeilBrownd089c6a2008-02-06 01:39:59 -0800426#define rdev_for_each(rdev, tmp, mddev) \
Cheng Renquan159ec1f2009-01-09 08:31:08 +1100427 list_for_each_entry_safe(rdev, tmp, &((mddev)->disks), same_set)
Linus Torvalds1da177e2005-04-16 15:20:36 -0700428
NeilBrown4b809912008-07-21 17:05:25 +1000429#define rdev_for_each_rcu(rdev, mddev) \
430 list_for_each_entry_rcu(rdev, &((mddev)->disks), same_set)
431
Linus Torvalds1da177e2005-04-16 15:20:36 -0700432typedef struct mdk_thread_s {
433 void (*run) (mddev_t *mddev);
434 mddev_t *mddev;
435 wait_queue_head_t wqueue;
436 unsigned long flags;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700437 struct task_struct *tsk;
NeilBrown32a76272005-06-21 17:17:14 -0700438 unsigned long timeout;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700439} mdk_thread_t;
440
441#define THREAD_WAKEUP 0
442
443#define __wait_event_lock_irq(wq, condition, lock, cmd) \
444do { \
445 wait_queue_t __wait; \
446 init_waitqueue_entry(&__wait, current); \
447 \
448 add_wait_queue(&wq, &__wait); \
449 for (;;) { \
450 set_current_state(TASK_UNINTERRUPTIBLE); \
451 if (condition) \
452 break; \
453 spin_unlock_irq(&lock); \
454 cmd; \
455 schedule(); \
456 spin_lock_irq(&lock); \
457 } \
458 current->state = TASK_RUNNING; \
459 remove_wait_queue(&wq, &__wait); \
460} while (0)
461
462#define wait_event_lock_irq(wq, condition, lock, cmd) \
463do { \
464 if (condition) \
465 break; \
466 __wait_event_lock_irq(wq, condition, lock, cmd); \
467} while (0)
468
NeilBrown1345b1d2006-01-06 00:20:40 -0800469static inline void safe_put_page(struct page *p)
470{
471 if (p) put_page(p);
472}
473
NeilBrown92022952009-03-31 14:33:13 +1100474extern int register_md_personality(struct mdk_personality *p);
475extern int unregister_md_personality(struct mdk_personality *p);
476extern mdk_thread_t * md_register_thread(void (*run) (mddev_t *mddev),
477 mddev_t *mddev, const char *name);
478extern void md_unregister_thread(mdk_thread_t *thread);
479extern void md_wakeup_thread(mdk_thread_t *thread);
480extern void md_check_recovery(mddev_t *mddev);
481extern void md_write_start(mddev_t *mddev, struct bio *bi);
482extern void md_write_end(mddev_t *mddev);
483extern void md_done_sync(mddev_t *mddev, int blocks, int ok);
484extern void md_error(mddev_t *mddev, mdk_rdev_t *rdev);
485
NeilBrown3fa841d2009-09-23 18:10:29 +1000486extern int mddev_congested(mddev_t *mddev, int bits);
Tejun Heoe9c74692010-09-03 11:56:18 +0200487extern void md_flush_request(mddev_t *mddev, struct bio *bio);
NeilBrown92022952009-03-31 14:33:13 +1100488extern void md_super_write(mddev_t *mddev, mdk_rdev_t *rdev,
489 sector_t sector, int size, struct page *page);
490extern void md_super_wait(mddev_t *mddev);
Jonathan Brassowccebd4c2011-01-14 09:14:33 +1100491extern int sync_page_io(mdk_rdev_t *rdev, sector_t sector, int size,
492 struct page *page, int rw, bool metadata_op);
NeilBrown92022952009-03-31 14:33:13 +1100493extern void md_do_sync(mddev_t *mddev);
494extern void md_new_event(mddev_t *mddev);
495extern int md_allow_write(mddev_t *mddev);
496extern void md_wait_for_blocked_rdev(mdk_rdev_t *rdev, mddev_t *mddev);
Dan Williams1f403622009-03-31 14:59:03 +1100497extern void md_set_array_sectors(mddev_t *mddev, sector_t array_sectors);
Andre Noll0894cc32009-06-18 08:49:23 +1000498extern int md_check_no_bitmap(mddev_t *mddev);
Andre Nollac5e7112009-08-03 10:59:47 +1000499extern int md_integrity_register(mddev_t *mddev);
NeilBrown72e02072009-12-14 12:49:55 +1100500extern void md_integrity_add_rdev(mdk_rdev_t *rdev, mddev_t *mddev);
501extern int strict_strtoul_scaled(const char *cp, unsigned long *res, int scale);
NeilBrown43a70502009-12-14 12:49:55 +1100502extern void restore_bitmap_write_access(struct file *file);
Christoph Hellwig63fe0812009-04-14 12:01:53 +1000503
NeilBrown390ee602010-06-01 19:37:27 +1000504extern void mddev_init(mddev_t *mddev);
505extern int md_run(mddev_t *mddev);
506extern void md_stop(mddev_t *mddev);
507extern void md_stop_writes(mddev_t *mddev);
NeilBrowne8bb9a82010-06-01 19:37:26 +1000508extern void md_rdev_init(mdk_rdev_t *rdev);
NeilBrown390ee602010-06-01 19:37:27 +1000509
510extern void mddev_suspend(mddev_t *mddev);
511extern void mddev_resume(mddev_t *mddev);
NeilBrowna167f662010-10-26 18:31:13 +1100512extern struct bio *bio_clone_mddev(struct bio *bio, gfp_t gfp_mask,
513 mddev_t *mddev);
514extern struct bio *bio_alloc_mddev(gfp_t gfp_mask, int nr_iovecs,
515 mddev_t *mddev);
NeilBrown97658cd2011-04-18 18:25:42 +1000516extern int mddev_check_plugged(mddev_t *mddev);
Christoph Hellwig63fe0812009-04-14 12:01:53 +1000517#endif /* _MD_MD_H */