[PATCH] md: fix duplicity of levels in md.txt
[linux-block.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
LT
33#define MaxSector (~(sector_t)0)
34#define MD_THREAD_NAME_MAX 14
35
1da177e4
LT
36typedef struct mddev_s mddev_t;
37typedef struct mdk_rdev_s mdk_rdev_t;
38
39#define MAX_MD_DEVS 256 /* Max number of md dev */
40
41/*
42 * options passed in raidrun:
43 */
44
8932c2e0
N
45/* Currently this must fix in an 'int' */
46#define MAX_CHUNK_SIZE (1<<30)
1da177e4 47
1da177e4
LT
48/*
49 * MD's 'extended' device
50 */
51struct mdk_rdev_s
52{
53 struct list_head same_set; /* RAID devices within the same set */
54
55 sector_t size; /* Device size (in blocks) */
56 mddev_t *mddev; /* RAID array if running */
57 unsigned long last_events; /* IO event timestamp */
58
59 struct block_device *bdev; /* block device handle */
60
61 struct page *sb_page;
62 int sb_loaded;
42543769 63 __u64 sb_events;
1da177e4
LT
64 sector_t data_offset; /* start of data in array */
65 sector_t sb_offset;
0002b271 66 int sb_size; /* bytes in the superblock */
1da177e4
LT
67 int preferred_minor; /* autorun support */
68
86e6ffdd
N
69 struct kobject kobj;
70
1da177e4
LT
71 /* A device can be in one of three states based on two flags:
72 * Not working: faulty==1 in_sync==0
73 * Fully working: faulty==0 in_sync==1
74 * Working, but not
75 * in sync with array
76 * faulty==0 in_sync==0
77 *
78 * It can never have faulty==1, in_sync==1
79 * This reduces the burden of testing multiple flags in many cases
80 */
1da177e4 81
b2d444d7
N
82 unsigned long flags;
83#define Faulty 1 /* device is known to have a fault */
84#define In_sync 2 /* device is in_sync with rest of array */
8ddf9efe 85#define WriteMostly 4 /* Avoid reading if at all possible */
a9701a30 86#define BarriersNotsupp 5 /* BIO_RW_BARRIER is not supported */
8ddf9efe 87
1da177e4
LT
88 int desc_nr; /* descriptor index in the superblock */
89 int raid_disk; /* role of device in array */
41158c7e
N
90 int saved_raid_disk; /* role that device used to have in the
91 * array and could again if we did a partial
92 * resync from the bitmap
93 */
5fd6c1dc
N
94 sector_t recovery_offset;/* If this device has been partially
95 * recovered, this is where we were
96 * up to.
97 */
1da177e4
LT
98
99 atomic_t nr_pending; /* number of pending requests.
100 * only maintained for arrays that
101 * support hot removal
102 */
ba22dcbf
N
103 atomic_t read_errors; /* number of consecutive read errors that
104 * we have tried to ignore.
105 */
4dbcdc75
N
106 atomic_t corrected_errors; /* number of corrected read errors,
107 * for reporting to userspace and storing
108 * in superblock.
109 */
1da177e4
LT
110};
111
1da177e4
LT
112struct mddev_s
113{
114 void *private;
2604b703 115 struct mdk_personality *pers;
1da177e4
LT
116 dev_t unit;
117 int md_minor;
118 struct list_head disks;
850b2b42
N
119 unsigned long flags;
120#define MD_CHANGE_DEVS 0 /* Some device status has changed */
121#define MD_CHANGE_CLEAN 1 /* transition to or from 'clean' */
122#define MD_CHANGE_PENDING 2 /* superblock update in progress */
123
1da177e4
LT
124 int ro;
125
126 struct gendisk *gendisk;
127
eae1701f
N
128 struct kobject kobj;
129
1da177e4
LT
130 /* Superblock information */
131 int major_version,
132 minor_version,
133 patch_version;
134 int persistent;
135 int chunk_size;
136 time_t ctime, utime;
137 int level, layout;
d9d166c2 138 char clevel[16];
1da177e4
LT
139 int raid_disks;
140 int max_disks;
141 sector_t size; /* used size of component devices */
142 sector_t array_size; /* exported array size */
143 __u64 events;
144
145 char uuid[16];
146
f6705578
N
147 /* If the array is being reshaped, we need to record the
148 * new shape and an indication of where we are up to.
149 * This is written to the superblock.
150 * If reshape_position is MaxSector, then no reshape is happening (yet).
151 */
152 sector_t reshape_position;
153 int delta_disks, new_level, new_layout, new_chunk;
154
1da177e4
LT
155 struct mdk_thread_s *thread; /* management thread */
156 struct mdk_thread_s *sync_thread; /* doing resync or reconstruct */
ff4e8d9a 157 sector_t curr_resync; /* last block scheduled */
1da177e4
LT
158 unsigned long resync_mark; /* a recent timestamp */
159 sector_t resync_mark_cnt;/* blocks written at resync_mark */
ff4e8d9a 160 sector_t curr_mark_cnt; /* blocks scheduled now */
1da177e4
LT
161
162 sector_t resync_max_sectors; /* may be set by personality */
9d88883e
N
163
164 sector_t resync_mismatches; /* count of sectors where
165 * parity/replica mismatch found
166 */
e464eafd
N
167
168 /* allow user-space to request suspension of IO to regions of the array */
169 sector_t suspend_lo;
170 sector_t suspend_hi;
88202a0c
N
171 /* if zero, use the system-wide default */
172 int sync_speed_min;
173 int sync_speed_max;
174
6ff8d8ec 175 int ok_start_degraded;
1da177e4
LT
176 /* recovery/resync flags
177 * NEEDED: we might need to start a resync/recover
178 * RUNNING: a thread is running, or about to be started
179 * SYNC: actually doing a resync, not a recovery
180 * ERR: and IO error was detected - abort the resync/recovery
181 * INTR: someone requested a (clean) early abort.
182 * DONE: thread is done and is waiting to be reaped
24dd469d
N
183 * REQUEST: user-space has requested a sync (used with SYNC)
184 * CHECK: user-space request for for check-only, no repair
ccfcc3c1
N
185 * RESHAPE: A reshape is happening
186 *
187 * If neither SYNC or RESHAPE are set, then it is a recovery.
1da177e4
LT
188 */
189#define MD_RECOVERY_RUNNING 0
190#define MD_RECOVERY_SYNC 1
191#define MD_RECOVERY_ERR 2
192#define MD_RECOVERY_INTR 3
193#define MD_RECOVERY_DONE 4
194#define MD_RECOVERY_NEEDED 5
24dd469d
N
195#define MD_RECOVERY_REQUESTED 6
196#define MD_RECOVERY_CHECK 7
ccfcc3c1 197#define MD_RECOVERY_RESHAPE 8
5fd6c1dc
N
198#define MD_RECOVERY_FROZEN 9
199
1da177e4
LT
200 unsigned long recovery;
201
202 int in_sync; /* know to not need resync */
df5b89b3 203 struct mutex reconfig_mutex;
1da177e4
LT
204 atomic_t active;
205
206 int changed; /* true if we might need to reread partition info */
207 int degraded; /* whether md should consider
208 * adding a spare
209 */
a9701a30
N
210 int barriers_work; /* initialised to true, cleared as soon
211 * as a barrier request to slave
212 * fails. Only supported
213 */
214 struct bio *biolist; /* bios that need to be retried
215 * because BIO_RW_BARRIER is not supported
216 */
1da177e4
LT
217
218 atomic_t recovery_active; /* blocks scheduled, but not written */
219 wait_queue_head_t recovery_wait;
220 sector_t recovery_cp;
06d91a5f
N
221
222 spinlock_t write_lock;
3d310eb7 223 wait_queue_head_t sb_wait; /* for waiting on superblock updates */
7bfa19f2 224 atomic_t pending_writes; /* number of active superblock writes */
06d91a5f 225
1da177e4
LT
226 unsigned int safemode; /* if set, update "clean" superblock
227 * when no writes pending.
228 */
229 unsigned int safemode_delay;
230 struct timer_list safemode_timer;
231 atomic_t writes_pending;
232 request_queue_t *queue; /* for plugging ... */
233
4b6d287f
N
234 atomic_t write_behind; /* outstanding async IO */
235 unsigned int max_write_behind; /* 0 = sync */
236
32a7627c
N
237 struct bitmap *bitmap; /* the bitmap for the device */
238 struct file *bitmap_file; /* the bitmap file */
a654b9d8
N
239 long bitmap_offset; /* offset from superblock of
240 * start of bitmap. May be
241 * negative, but not '0'
242 */
36fa3063
N
243 long default_bitmap_offset; /* this is the offset to use when
244 * hot-adding a bitmap. It should
245 * eventually be settable by sysfs.
246 */
32a7627c 247
1da177e4
LT
248 struct list_head all_mddevs;
249};
250
251
252static inline void rdev_dec_pending(mdk_rdev_t *rdev, mddev_t *mddev)
253{
b2d444d7 254 int faulty = test_bit(Faulty, &rdev->flags);
1da177e4
LT
255 if (atomic_dec_and_test(&rdev->nr_pending) && faulty)
256 set_bit(MD_RECOVERY_NEEDED, &mddev->recovery);
257}
258
259static inline void md_sync_acct(struct block_device *bdev, unsigned long nr_sectors)
260{
261 atomic_add(nr_sectors, &bdev->bd_contains->bd_disk->sync_io);
262}
263
2604b703 264struct mdk_personality
1da177e4
LT
265{
266 char *name;
2604b703
N
267 int level;
268 struct list_head list;
1da177e4
LT
269 struct module *owner;
270 int (*make_request)(request_queue_t *q, struct bio *bio);
271 int (*run)(mddev_t *mddev);
272 int (*stop)(mddev_t *mddev);
273 void (*status)(struct seq_file *seq, mddev_t *mddev);
274 /* error_handler must set ->faulty and clear ->in_sync
275 * if appropriate, and should abort recovery if needed
276 */
277 void (*error_handler)(mddev_t *mddev, mdk_rdev_t *rdev);
278 int (*hot_add_disk) (mddev_t *mddev, mdk_rdev_t *rdev);
279 int (*hot_remove_disk) (mddev_t *mddev, int number);
280 int (*spare_active) (mddev_t *mddev);
57afd89f 281 sector_t (*sync_request)(mddev_t *mddev, sector_t sector_nr, int *skipped, int go_faster);
1da177e4 282 int (*resize) (mddev_t *mddev, sector_t sectors);
63c70c4f
N
283 int (*check_reshape) (mddev_t *mddev);
284 int (*start_reshape) (mddev_t *mddev);
1da177e4 285 int (*reconfig) (mddev_t *mddev, int layout, int chunk_size);
36fa3063
N
286 /* quiesce moves between quiescence states
287 * 0 - fully active
288 * 1 - no new requests allowed
289 * others - reserved
290 */
291 void (*quiesce) (mddev_t *mddev, int state);
1da177e4
LT
292};
293
294
007583c9
N
295struct md_sysfs_entry {
296 struct attribute attr;
297 ssize_t (*show)(mddev_t *, char *);
298 ssize_t (*store)(mddev_t *, const char *, size_t);
299};
300
301
1da177e4
LT
302static inline char * mdname (mddev_t * mddev)
303{
304 return mddev->gendisk ? mddev->gendisk->disk_name : "mdX";
305}
306
1da177e4
LT
307/*
308 * iterates through some rdev ringlist. It's safe to remove the
309 * current 'rdev'. Dont touch 'tmp' though.
310 */
311#define ITERATE_RDEV_GENERIC(head,rdev,tmp) \
312 \
313 for ((tmp) = (head).next; \
314 (rdev) = (list_entry((tmp), mdk_rdev_t, same_set)), \
315 (tmp) = (tmp)->next, (tmp)->prev != &(head) \
316 ; )
317/*
318 * iterates through the 'same array disks' ringlist
319 */
320#define ITERATE_RDEV(mddev,rdev,tmp) \
321 ITERATE_RDEV_GENERIC((mddev)->disks,rdev,tmp)
322
323/*
324 * Iterates through 'pending RAID disks'
325 */
326#define ITERATE_RDEV_PENDING(rdev,tmp) \
327 ITERATE_RDEV_GENERIC(pending_raid_disks,rdev,tmp)
328
329typedef struct mdk_thread_s {
330 void (*run) (mddev_t *mddev);
331 mddev_t *mddev;
332 wait_queue_head_t wqueue;
333 unsigned long flags;
1da177e4 334 struct task_struct *tsk;
32a7627c 335 unsigned long timeout;
1da177e4
LT
336} mdk_thread_t;
337
338#define THREAD_WAKEUP 0
339
340#define __wait_event_lock_irq(wq, condition, lock, cmd) \
341do { \
342 wait_queue_t __wait; \
343 init_waitqueue_entry(&__wait, current); \
344 \
345 add_wait_queue(&wq, &__wait); \
346 for (;;) { \
347 set_current_state(TASK_UNINTERRUPTIBLE); \
348 if (condition) \
349 break; \
350 spin_unlock_irq(&lock); \
351 cmd; \
352 schedule(); \
353 spin_lock_irq(&lock); \
354 } \
355 current->state = TASK_RUNNING; \
356 remove_wait_queue(&wq, &__wait); \
357} while (0)
358
359#define wait_event_lock_irq(wq, condition, lock, cmd) \
360do { \
361 if (condition) \
362 break; \
363 __wait_event_lock_irq(wq, condition, lock, cmd); \
364} while (0)
365
1345b1d8
N
366static inline void safe_put_page(struct page *p)
367{
368 if (p) put_page(p);
369}
370
9361401e 371#endif /* CONFIG_BLOCK */
1da177e4
LT
372#endif
373