blob: c732b6cce935283a6adb826a641bb023accf081e [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001#ifndef _RAID1_H
2#define _RAID1_H
3
Linus Torvalds1da177e2005-04-16 15:20:36 -07004struct mirror_info {
NeilBrown3cb03002011-10-11 16:45:26 +11005 struct md_rdev *rdev;
Linus Torvalds1da177e2005-04-16 15:20:36 -07006 sector_t head_position;
7};
8
9/*
10 * memory pools need a pointer to the mddev, so they can force an unplug
11 * when memory is tight, and a count of the number of drives that the
12 * pool was allocated for, so they know how much to allocate and free.
13 * mddev->raid_disks cannot be used, as it can change while a pool is active
14 * These two datums are stored in a kmalloced struct.
15 */
16
17struct pool_info {
NeilBrownfd01b882011-10-11 16:47:53 +110018 struct mddev *mddev;
Linus Torvalds1da177e2005-04-16 15:20:36 -070019 int raid_disks;
20};
21
NeilBrowne8096362011-10-11 16:49:05 +110022struct r1conf {
NeilBrownfd01b882011-10-11 16:47:53 +110023 struct mddev *mddev;
NeilBrown0f6d02d2011-10-11 16:48:46 +110024 struct mirror_info *mirrors;
Linus Torvalds1da177e2005-04-16 15:20:36 -070025 int raid_disks;
NeilBrownce550c22011-10-07 14:22:33 +110026
27 /* When choose the best device for a read (read_balance())
28 * we try to keep sequential reads one the same device
29 * using 'last_used' and 'next_seq_sect'
30 */
Linus Torvalds1da177e2005-04-16 15:20:36 -070031 int last_used;
32 sector_t next_seq_sect;
NeilBrownce550c22011-10-07 14:22:33 +110033 /* During resync, read_balancing is only allowed on the part
34 * of the array that has been resynced. 'next_resync' tells us
35 * where that is.
36 */
37 sector_t next_resync;
38
Linus Torvalds1da177e2005-04-16 15:20:36 -070039 spinlock_t device_lock;
40
NeilBrown9f2c9d12011-10-11 16:48:43 +110041 /* list of 'struct r1bio' that need to be processed by raid1d,
42 * whether to retry a read, writeout a resync or recovery
43 * block, or anything else.
NeilBrownce550c22011-10-07 14:22:33 +110044 */
Linus Torvalds1da177e2005-04-16 15:20:36 -070045 struct list_head retry_list;
NeilBrownce550c22011-10-07 14:22:33 +110046
47 /* queue pending writes to be submitted on unplug */
NeilBrown191ea9b2005-06-21 17:17:23 -070048 struct bio_list pending_bio_list;
NeilBrown34db0cd2011-10-11 16:50:01 +110049 int pending_count;
NeilBrown191ea9b2005-06-21 17:17:23 -070050
NeilBrownce550c22011-10-07 14:22:33 +110051 /* for use when syncing mirrors:
52 * We don't allow both normal IO and resync/recovery IO at
53 * the same time - resync/recovery can only happen when there
54 * is no other IO. So when either is active, the other has to wait.
55 * See more details description in raid1.c near raise_barrier().
56 */
57 wait_queue_head_t wait_barrier;
Linus Torvalds1da177e2005-04-16 15:20:36 -070058 spinlock_t resync_lock;
NeilBrown191ea9b2005-06-21 17:17:23 -070059 int nr_pending;
NeilBrown17999be2006-01-06 00:20:12 -080060 int nr_waiting;
NeilBrownddaf22a2006-01-06 00:20:19 -080061 int nr_queued;
NeilBrown191ea9b2005-06-21 17:17:23 -070062 int barrier;
Linus Torvalds1da177e2005-04-16 15:20:36 -070063
NeilBrownce550c22011-10-07 14:22:33 +110064 /* Set to 1 if a full sync is needed, (fresh device added).
65 * Cleared when a sync completes.
66 */
67 int fullsync;
Linus Torvalds1da177e2005-04-16 15:20:36 -070068
NeilBrownce550c22011-10-07 14:22:33 +110069 /* When the same as mddev->recovery_disabled we don't allow
70 * recovery to be attempted as we expect a read error.
71 */
72 int recovery_disabled;
73
74
75 /* poolinfo contains information about the content of the
76 * mempools - it changes when the array grows or shrinks
77 */
Linus Torvalds1da177e2005-04-16 15:20:36 -070078 struct pool_info *poolinfo;
NeilBrown9f2c9d12011-10-11 16:48:43 +110079 mempool_t *r1bio_pool;
80 mempool_t *r1buf_pool;
NeilBrown709ae482009-12-14 12:49:51 +110081
NeilBrownce550c22011-10-07 14:22:33 +110082 /* temporary buffer to synchronous IO when attempting to repair
83 * a read error.
84 */
85 struct page *tmppage;
86
87
NeilBrown709ae482009-12-14 12:49:51 +110088 /* When taking over an array from a different personality, we store
89 * the new thread here until we fully activate the array.
90 */
NeilBrown2b8bf342011-10-11 16:48:23 +110091 struct md_thread *thread;
Linus Torvalds1da177e2005-04-16 15:20:36 -070092};
93
Linus Torvalds1da177e2005-04-16 15:20:36 -070094/*
Linus Torvalds1da177e2005-04-16 15:20:36 -070095 * this is our 'private' RAID1 bio.
96 *
97 * it contains information about what kind of IO operations were started
98 * for this RAID1 operation, and about their status:
99 */
100
NeilBrown9f2c9d12011-10-11 16:48:43 +1100101struct r1bio {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700102 atomic_t remaining; /* 'have we finished' count,
103 * used from IRQ handlers
104 */
NeilBrown4b6d2872005-09-09 16:23:47 -0700105 atomic_t behind_remaining; /* number of write-behind ios remaining
106 * in this BehindIO request
107 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700108 sector_t sector;
109 int sectors;
110 unsigned long state;
NeilBrownfd01b882011-10-11 16:47:53 +1100111 struct mddev *mddev;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700112 /*
113 * original bio going to /dev/mdx
114 */
115 struct bio *master_bio;
116 /*
117 * if the IO is in READ direction, then this is where we read
118 */
119 int read_disk;
120
121 struct list_head retry_list;
NeilBrownaf6d7b72011-05-11 14:51:19 +1000122 /* Next two are only valid when R1BIO_BehindIO is set */
NeilBrown2ca68f52011-07-28 11:32:10 +1000123 struct bio_vec *behind_bvecs;
NeilBrownaf6d7b72011-05-11 14:51:19 +1000124 int behind_page_count;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700125 /*
126 * if the IO is in WRITE direction, then multiple bios are used.
127 * We choose the number when they are allocated.
128 */
129 struct bio *bios[0];
NeilBrown191ea9b2005-06-21 17:17:23 -0700130 /* DO NOT PUT ANY NEW FIELDS HERE - bios array is contiguously alloced*/
Linus Torvalds1da177e2005-04-16 15:20:36 -0700131};
132
NeilBrowncf30a472006-01-06 00:20:23 -0800133/* when we get a read error on a read-only array, we redirect to another
134 * device without failing the first device, or trying to over-write to
135 * correct the read error. To keep track of bad blocks on a per-bio
136 * level, we store IO_BLOCKED in the appropriate 'bios' pointer
137 */
NeilBrown4367af52011-07-28 11:31:49 +1000138#define IO_BLOCKED ((struct bio *)1)
139/* When we successfully write to a known bad-block, we need to remove the
140 * bad-block marking which must be done from process context. So we record
141 * the success by setting bios[n] to IO_MADE_GOOD
142 */
143#define IO_MADE_GOOD ((struct bio *)2)
144
145#define BIO_SPECIAL(bio) ((unsigned long)bio <= 2)
NeilBrowncf30a472006-01-06 00:20:23 -0800146
Linus Torvalds1da177e2005-04-16 15:20:36 -0700147/* bits for r1bio.state */
148#define R1BIO_Uptodate 0
149#define R1BIO_IsSync 1
NeilBrown191ea9b2005-06-21 17:17:23 -0700150#define R1BIO_Degraded 2
NeilBrowna9701a32005-11-08 21:39:34 -0800151#define R1BIO_BehindIO 3
NeilBrownd2eb35a2011-07-28 11:31:48 +1000152/* Set ReadError on bios that experience a readerror so that
153 * raid1d knows what to do with them.
154 */
155#define R1BIO_ReadError 4
NeilBrown4b6d2872005-09-09 16:23:47 -0700156/* For write-behind requests, we call bi_end_io when
157 * the last non-write-behind device completes, providing
158 * any write was successful. Otherwise we call when
159 * any write-behind write succeeds, otherwise we call
160 * with failure when last write completes (and all failed).
161 * Record that bi_end_io was called with this flag...
162 */
NeilBrown9e71f9c2006-03-23 02:59:22 -0800163#define R1BIO_Returned 6
NeilBrown4367af52011-07-28 11:31:49 +1000164/* If a write for this request means we can clear some
165 * known-bad-block records, we set this flag
166 */
NeilBrowncd5ff9a2011-07-28 11:32:41 +1000167#define R1BIO_MadeGood 7
168#define R1BIO_WriteError 8
NeilBrown4b6d2872005-09-09 16:23:47 -0700169
NeilBrownfd01b882011-10-11 16:47:53 +1100170extern int md_raid1_congested(struct mddev *mddev, int bits);
Jonathan Brassow1ed72422011-06-07 17:50:35 -0500171
Linus Torvalds1da177e2005-04-16 15:20:36 -0700172#endif