Make sure all changes to md/dev-XX/state are notified
[GitHub/mt8127/android_kernel_alcatel_ttab.git] / include / linux / raid / md_k.h
CommitLineData
1da177e4
LT
1/*
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
15#ifndef _MD_K_H
16#define _MD_K_H
17
06d91a5f
N
18/* and dm-bio-list.h is not under include/linux because.... ??? */
19#include "../../../drivers/md/dm-bio-list.h"
20
9361401e
DH
21#ifdef CONFIG_BLOCK
22
1da177e4
LT
23#define LEVEL_MULTIPATH (-4)
24#define LEVEL_LINEAR (-1)
25#define LEVEL_FAULTY (-5)
26
2604b703
N
27/* we need a value for 'no level specified' and 0
28 * means 'raid0', so we need something else. This is
29 * for internal use only
30 */
31#define LEVEL_NONE (-1000000)
32
1da177e4 33#define MaxSector (~(sector_t)0)
1da177e4 34
1da177e4
LT
35typedef struct mddev_s mddev_t;
36typedef struct mdk_rdev_s mdk_rdev_t;
37
1da177e4
LT
38/*
39 * options passed in raidrun:
40 */
41
e8703fe1 42/* Currently this must fit in an 'int' */
8932c2e0 43#define MAX_CHUNK_SIZE (1<<30)
1da177e4 44
1da177e4
LT
45/*
46 * MD's 'extended' device
47 */
48struct mdk_rdev_s
49{
50 struct list_head same_set; /* RAID devices within the same set */
51
52 sector_t size; /* Device size (in blocks) */
53 mddev_t *mddev; /* RAID array if running */
713f6ab1 54 long last_events; /* IO event timestamp */
1da177e4
LT
55
56 struct block_device *bdev; /* block device handle */
57
58 struct page *sb_page;
59 int sb_loaded;
42543769 60 __u64 sb_events;
1da177e4
LT
61 sector_t data_offset; /* start of data in array */
62 sector_t sb_offset;
0002b271 63 int sb_size; /* bytes in the superblock */
1da177e4
LT
64 int preferred_minor; /* autorun support */
65
86e6ffdd
N
66 struct kobject kobj;
67
1da177e4
LT
68 /* A device can be in one of three states based on two flags:
69 * Not working: faulty==1 in_sync==0
70 * Fully working: faulty==0 in_sync==1
71 * Working, but not
72 * in sync with array
73 * faulty==0 in_sync==0
74 *
75 * It can never have faulty==1, in_sync==1
76 * This reduces the burden of testing multiple flags in many cases
77 */
1da177e4 78
b2d444d7
N
79 unsigned long flags;
80#define Faulty 1 /* device is known to have a fault */
81#define In_sync 2 /* device is in_sync with rest of array */
8ddf9efe 82#define WriteMostly 4 /* Avoid reading if at all possible */
a9701a30 83#define BarriersNotsupp 5 /* BIO_RW_BARRIER is not supported */
c5d79adb
N
84#define AllReserved 6 /* If whole device is reserved for
85 * one array */
d0fae18f 86#define AutoDetected 7 /* added by auto-detect */
6bfe0b49
DW
87#define Blocked 8 /* An error occured on an externally
88 * managed array, don't allow writes
89 * until it is cleared */
52664732
NB
90#define StateChanged 9 /* Faulty or Blocked has changed during
91 * interrupt, so it needs to be
92 * notified by the thread */
6bfe0b49 93 wait_queue_head_t blocked_wait;
8ddf9efe 94
1da177e4
LT
95 int desc_nr; /* descriptor index in the superblock */
96 int raid_disk; /* role of device in array */
41158c7e
N
97 int saved_raid_disk; /* role that device used to have in the
98 * array and could again if we did a partial
99 * resync from the bitmap
100 */
5fd6c1dc
N
101 sector_t recovery_offset;/* If this device has been partially
102 * recovered, this is where we were
103 * up to.
104 */
1da177e4
LT
105
106 atomic_t nr_pending; /* number of pending requests.
107 * only maintained for arrays that
108 * support hot removal
109 */
ba22dcbf
N
110 atomic_t read_errors; /* number of consecutive read errors that
111 * we have tried to ignore.
112 */
4dbcdc75
N
113 atomic_t corrected_errors; /* number of corrected read errors,
114 * for reporting to userspace and storing
115 * in superblock.
116 */
5792a285 117 struct work_struct del_work; /* used for delayed sysfs removal */
1da177e4
LT
118};
119
1da177e4
LT
120struct mddev_s
121{
122 void *private;
2604b703 123 struct mdk_personality *pers;
1da177e4
LT
124 dev_t unit;
125 int md_minor;
126 struct list_head disks;
850b2b42
N
127 unsigned long flags;
128#define MD_CHANGE_DEVS 0 /* Some device status has changed */
129#define MD_CHANGE_CLEAN 1 /* transition to or from 'clean' */
130#define MD_CHANGE_PENDING 2 /* superblock update in progress */
131
1da177e4
LT
132 int ro;
133
134 struct gendisk *gendisk;
135
eae1701f
N
136 struct kobject kobj;
137
1da177e4
LT
138 /* Superblock information */
139 int major_version,
140 minor_version,
141 patch_version;
142 int persistent;
e691063a
N
143 int external; /* metadata is
144 * managed externally */
145 char metadata_type[17]; /* externally set*/
1da177e4
LT
146 int chunk_size;
147 time_t ctime, utime;
148 int level, layout;
d9d166c2 149 char clevel[16];
1da177e4
LT
150 int raid_disks;
151 int max_disks;
152 sector_t size; /* used size of component devices */
153 sector_t array_size; /* exported array size */
154 __u64 events;
155
156 char uuid[16];
157
f6705578
N
158 /* If the array is being reshaped, we need to record the
159 * new shape and an indication of where we are up to.
160 * This is written to the superblock.
161 * If reshape_position is MaxSector, then no reshape is happening (yet).
162 */
163 sector_t reshape_position;
164 int delta_disks, new_level, new_layout, new_chunk;
165
1da177e4
LT
166 struct mdk_thread_s *thread; /* management thread */
167 struct mdk_thread_s *sync_thread; /* doing resync or reconstruct */
ff4e8d9a 168 sector_t curr_resync; /* last block scheduled */
1da177e4
LT
169 unsigned long resync_mark; /* a recent timestamp */
170 sector_t resync_mark_cnt;/* blocks written at resync_mark */
ff4e8d9a 171 sector_t curr_mark_cnt; /* blocks scheduled now */
1da177e4
LT
172
173 sector_t resync_max_sectors; /* may be set by personality */
9d88883e
N
174
175 sector_t resync_mismatches; /* count of sectors where
176 * parity/replica mismatch found
177 */
e464eafd
N
178
179 /* allow user-space to request suspension of IO to regions of the array */
180 sector_t suspend_lo;
181 sector_t suspend_hi;
88202a0c
N
182 /* if zero, use the system-wide default */
183 int sync_speed_min;
184 int sync_speed_max;
185
90b08710
BS
186 /* resync even though the same disks are shared among md-devices */
187 int parallel_resync;
188
6ff8d8ec 189 int ok_start_degraded;
1da177e4
LT
190 /* recovery/resync flags
191 * NEEDED: we might need to start a resync/recover
192 * RUNNING: a thread is running, or about to be started
193 * SYNC: actually doing a resync, not a recovery
72a23c21 194 * RECOVER: doing recovery, or need to try it.
dfc70645 195 * INTR: resync needs to be aborted for some reason
1da177e4 196 * DONE: thread is done and is waiting to be reaped
24dd469d
N
197 * REQUEST: user-space has requested a sync (used with SYNC)
198 * CHECK: user-space request for for check-only, no repair
ccfcc3c1
N
199 * RESHAPE: A reshape is happening
200 *
201 * If neither SYNC or RESHAPE are set, then it is a recovery.
1da177e4
LT
202 */
203#define MD_RECOVERY_RUNNING 0
204#define MD_RECOVERY_SYNC 1
72a23c21 205#define MD_RECOVERY_RECOVER 2
1da177e4
LT
206#define MD_RECOVERY_INTR 3
207#define MD_RECOVERY_DONE 4
208#define MD_RECOVERY_NEEDED 5
24dd469d
N
209#define MD_RECOVERY_REQUESTED 6
210#define MD_RECOVERY_CHECK 7
ccfcc3c1 211#define MD_RECOVERY_RESHAPE 8
5fd6c1dc
N
212#define MD_RECOVERY_FROZEN 9
213
1da177e4
LT
214 unsigned long recovery;
215
216 int in_sync; /* know to not need resync */
df5b89b3 217 struct mutex reconfig_mutex;
1da177e4
LT
218 atomic_t active;
219
44ce6294 220 int changed; /* true if we might need to reread partition info */
1da177e4
LT
221 int degraded; /* whether md should consider
222 * adding a spare
223 */
a9701a30
N
224 int barriers_work; /* initialised to true, cleared as soon
225 * as a barrier request to slave
226 * fails. Only supported
227 */
228 struct bio *biolist; /* bios that need to be retried
229 * because BIO_RW_BARRIER is not supported
230 */
1da177e4
LT
231
232 atomic_t recovery_active; /* blocks scheduled, but not written */
233 wait_queue_head_t recovery_wait;
234 sector_t recovery_cp;
5e96ee65
NB
235 sector_t resync_min; /* user requested sync
236 * starts here */
c6207277
N
237 sector_t resync_max; /* resync should pause
238 * when it gets here */
06d91a5f
N
239
240 spinlock_t write_lock;
3d310eb7 241 wait_queue_head_t sb_wait; /* for waiting on superblock updates */
7bfa19f2 242 atomic_t pending_writes; /* number of active superblock writes */
06d91a5f 243
1da177e4
LT
244 unsigned int safemode; /* if set, update "clean" superblock
245 * when no writes pending.
246 */
247 unsigned int safemode_delay;
248 struct timer_list safemode_timer;
249 atomic_t writes_pending;
165125e1 250 struct request_queue *queue; /* for plugging ... */
1da177e4 251
4b6d287f
N
252 atomic_t write_behind; /* outstanding async IO */
253 unsigned int max_write_behind; /* 0 = sync */
254
32a7627c
N
255 struct bitmap *bitmap; /* the bitmap for the device */
256 struct file *bitmap_file; /* the bitmap file */
a654b9d8
N
257 long bitmap_offset; /* offset from superblock of
258 * start of bitmap. May be
259 * negative, but not '0'
260 */
36fa3063
N
261 long default_bitmap_offset; /* this is the offset to use when
262 * hot-adding a bitmap. It should
263 * eventually be settable by sysfs.
264 */
32a7627c 265
1da177e4
LT
266 struct list_head all_mddevs;
267};
268
269
270static inline void rdev_dec_pending(mdk_rdev_t *rdev, mddev_t *mddev)
271{
b2d444d7 272 int faulty = test_bit(Faulty, &rdev->flags);
1da177e4
LT
273 if (atomic_dec_and_test(&rdev->nr_pending) && faulty)
274 set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
275}
276
277static inline void md_sync_acct(struct block_device *bdev, unsigned long nr_sectors)
278{
279 atomic_add(nr_sectors, &bdev->bd_contains->bd_disk->sync_io);
280}
281
2604b703 282struct mdk_personality
1da177e4
LT
283{
284 char *name;
2604b703
N
285 int level;
286 struct list_head list;
1da177e4 287 struct module *owner;
165125e1 288 int (*make_request)(struct request_queue *q, struct bio *bio);
1da177e4
LT
289 int (*run)(mddev_t *mddev);
290 int (*stop)(mddev_t *mddev);
291 void (*status)(struct seq_file *seq, mddev_t *mddev);
292 /* error_handler must set ->faulty and clear ->in_sync
293 * if appropriate, and should abort recovery if needed
294 */
295 void (*error_handler)(mddev_t *mddev, mdk_rdev_t *rdev);
296 int (*hot_add_disk) (mddev_t *mddev, mdk_rdev_t *rdev);
297 int (*hot_remove_disk) (mddev_t *mddev, int number);
298 int (*spare_active) (mddev_t *mddev);
57afd89f 299 sector_t (*sync_request)(mddev_t *mddev, sector_t sector_nr, int *skipped, int go_faster);
1da177e4 300 int (*resize) (mddev_t *mddev, sector_t sectors);
63c70c4f
N
301 int (*check_reshape) (mddev_t *mddev);
302 int (*start_reshape) (mddev_t *mddev);
1da177e4 303 int (*reconfig) (mddev_t *mddev, int layout, int chunk_size);
36fa3063
N
304 /* quiesce moves between quiescence states
305 * 0 - fully active
306 * 1 - no new requests allowed
307 * others - reserved
308 */
309 void (*quiesce) (mddev_t *mddev, int state);
1da177e4
LT
310};
311
312
007583c9
N
313struct md_sysfs_entry {
314 struct attribute attr;
315 ssize_t (*show)(mddev_t *, char *);
316 ssize_t (*store)(mddev_t *, const char *, size_t);
317};
318
319
1da177e4
LT
320static inline char * mdname (mddev_t * mddev)
321{
322 return mddev->gendisk ? mddev->gendisk->disk_name : "mdX";
323}
324
1da177e4
LT
325/*
326 * iterates through some rdev ringlist. It's safe to remove the
327 * current 'rdev'. Dont touch 'tmp' though.
328 */
73c34431 329#define rdev_for_each_list(rdev, tmp, list) \
1da177e4 330 \
73c34431 331 for ((tmp) = (list).next; \
1da177e4 332 (rdev) = (list_entry((tmp), mdk_rdev_t, same_set)), \
73c34431 333 (tmp) = (tmp)->next, (tmp)->prev != &(list) \
1da177e4
LT
334 ; )
335/*
336 * iterates through the 'same array disks' ringlist
337 */
d089c6af 338#define rdev_for_each(rdev, tmp, mddev) \
73c34431 339 rdev_for_each_list(rdev, tmp, (mddev)->disks)
1da177e4
LT
340
341typedef struct mdk_thread_s {
342 void (*run) (mddev_t *mddev);
343 mddev_t *mddev;
344 wait_queue_head_t wqueue;
345 unsigned long flags;
1da177e4 346 struct task_struct *tsk;
32a7627c 347 unsigned long timeout;
1da177e4
LT
348} mdk_thread_t;
349
350#define THREAD_WAKEUP 0
351
352#define __wait_event_lock_irq(wq, condition, lock, cmd) \
353do { \
354 wait_queue_t __wait; \
355 init_waitqueue_entry(&__wait, current); \
356 \
357 add_wait_queue(&wq, &__wait); \
358 for (;;) { \
359 set_current_state(TASK_UNINTERRUPTIBLE); \
360 if (condition) \
361 break; \
362 spin_unlock_irq(&lock); \
363 cmd; \
364 schedule(); \
365 spin_lock_irq(&lock); \
366 } \
367 current->state = TASK_RUNNING; \
368 remove_wait_queue(&wq, &__wait); \
369} while (0)
370
371#define wait_event_lock_irq(wq, condition, lock, cmd) \
372do { \
373 if (condition) \
374 break; \
375 __wait_event_lock_irq(wq, condition, lock, cmd); \
376} while (0)
377
1345b1d8
N
378static inline void safe_put_page(struct page *p)
379{
380 if (p) put_page(p);
381}
382
9361401e 383#endif /* CONFIG_BLOCK */
1da177e4
LT
384#endif
385