blob: dd5e66a4163f93080edd4c28610fdfddf8d69472 [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
Linus Torvalds1da177e2005-04-16 15:20:36 -070022struct r1_private_data_s {
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;
NeilBrown191ea9b2005-06-21 17:17:23 -070049
NeilBrownce550c22011-10-07 14:22:33 +110050 /* for use when syncing mirrors:
51 * We don't allow both normal IO and resync/recovery IO at
52 * the same time - resync/recovery can only happen when there
53 * is no other IO. So when either is active, the other has to wait.
54 * See more details description in raid1.c near raise_barrier().
55 */
56 wait_queue_head_t wait_barrier;
Linus Torvalds1da177e2005-04-16 15:20:36 -070057 spinlock_t resync_lock;
NeilBrown191ea9b2005-06-21 17:17:23 -070058 int nr_pending;
NeilBrown17999be2006-01-06 00:20:12 -080059 int nr_waiting;
NeilBrownddaf22a2006-01-06 00:20:19 -080060 int nr_queued;
NeilBrown191ea9b2005-06-21 17:17:23 -070061 int barrier;
Linus Torvalds1da177e2005-04-16 15:20:36 -070062
NeilBrownce550c22011-10-07 14:22:33 +110063 /* Set to 1 if a full sync is needed, (fresh device added).
64 * Cleared when a sync completes.
65 */
66 int fullsync;
Linus Torvalds1da177e2005-04-16 15:20:36 -070067
NeilBrownce550c22011-10-07 14:22:33 +110068 /* When the same as mddev->recovery_disabled we don't allow
69 * recovery to be attempted as we expect a read error.
70 */
71 int recovery_disabled;
72
73
74 /* poolinfo contains information about the content of the
75 * mempools - it changes when the array grows or shrinks
76 */
Linus Torvalds1da177e2005-04-16 15:20:36 -070077 struct pool_info *poolinfo;
NeilBrown9f2c9d12011-10-11 16:48:43 +110078 mempool_t *r1bio_pool;
79 mempool_t *r1buf_pool;
NeilBrown709ae482009-12-14 12:49:51 +110080
NeilBrownce550c22011-10-07 14:22:33 +110081 /* temporary buffer to synchronous IO when attempting to repair
82 * a read error.
83 */
84 struct page *tmppage;
85
86
NeilBrown709ae482009-12-14 12:49:51 +110087 /* When taking over an array from a different personality, we store
88 * the new thread here until we fully activate the array.
89 */
NeilBrown2b8bf342011-10-11 16:48:23 +110090 struct md_thread *thread;
Linus Torvalds1da177e2005-04-16 15:20:36 -070091};
92
93typedef struct r1_private_data_s conf_t;
94
95/*
Linus Torvalds1da177e2005-04-16 15:20:36 -070096 * this is our 'private' RAID1 bio.
97 *
98 * it contains information about what kind of IO operations were started
99 * for this RAID1 operation, and about their status:
100 */
101
NeilBrown9f2c9d12011-10-11 16:48:43 +1100102struct r1bio {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700103 atomic_t remaining; /* 'have we finished' count,
104 * used from IRQ handlers
105 */
NeilBrown4b6d2872005-09-09 16:23:47 -0700106 atomic_t behind_remaining; /* number of write-behind ios remaining
107 * in this BehindIO request
108 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700109 sector_t sector;
110 int sectors;
111 unsigned long state;
NeilBrownfd01b882011-10-11 16:47:53 +1100112 struct mddev *mddev;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700113 /*
114 * original bio going to /dev/mdx
115 */
116 struct bio *master_bio;
117 /*
118 * if the IO is in READ direction, then this is where we read
119 */
120 int read_disk;
121
122 struct list_head retry_list;
NeilBrownaf6d7b72011-05-11 14:51:19 +1000123 /* Next two are only valid when R1BIO_BehindIO is set */
NeilBrown2ca68f52011-07-28 11:32:10 +1000124 struct bio_vec *behind_bvecs;
NeilBrownaf6d7b72011-05-11 14:51:19 +1000125 int behind_page_count;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700126 /*
127 * if the IO is in WRITE direction, then multiple bios are used.
128 * We choose the number when they are allocated.
129 */
130 struct bio *bios[0];
NeilBrown191ea9b2005-06-21 17:17:23 -0700131 /* DO NOT PUT ANY NEW FIELDS HERE - bios array is contiguously alloced*/
Linus Torvalds1da177e2005-04-16 15:20:36 -0700132};
133
NeilBrowncf30a472006-01-06 00:20:23 -0800134/* when we get a read error on a read-only array, we redirect to another
135 * device without failing the first device, or trying to over-write to
136 * correct the read error. To keep track of bad blocks on a per-bio
137 * level, we store IO_BLOCKED in the appropriate 'bios' pointer
138 */
NeilBrown4367af52011-07-28 11:31:49 +1000139#define IO_BLOCKED ((struct bio *)1)
140/* When we successfully write to a known bad-block, we need to remove the
141 * bad-block marking which must be done from process context. So we record
142 * the success by setting bios[n] to IO_MADE_GOOD
143 */
144#define IO_MADE_GOOD ((struct bio *)2)
145
146#define BIO_SPECIAL(bio) ((unsigned long)bio <= 2)
NeilBrowncf30a472006-01-06 00:20:23 -0800147
Linus Torvalds1da177e2005-04-16 15:20:36 -0700148/* bits for r1bio.state */
149#define R1BIO_Uptodate 0
150#define R1BIO_IsSync 1
NeilBrown191ea9b2005-06-21 17:17:23 -0700151#define R1BIO_Degraded 2
NeilBrowna9701a32005-11-08 21:39:34 -0800152#define R1BIO_BehindIO 3
NeilBrownd2eb35a2011-07-28 11:31:48 +1000153/* Set ReadError on bios that experience a readerror so that
154 * raid1d knows what to do with them.
155 */
156#define R1BIO_ReadError 4
NeilBrown4b6d2872005-09-09 16:23:47 -0700157/* For write-behind requests, we call bi_end_io when
158 * the last non-write-behind device completes, providing
159 * any write was successful. Otherwise we call when
160 * any write-behind write succeeds, otherwise we call
161 * with failure when last write completes (and all failed).
162 * Record that bi_end_io was called with this flag...
163 */
NeilBrown9e71f9c2006-03-23 02:59:22 -0800164#define R1BIO_Returned 6
NeilBrown4367af52011-07-28 11:31:49 +1000165/* If a write for this request means we can clear some
166 * known-bad-block records, we set this flag
167 */
NeilBrowncd5ff9a12011-07-28 11:32:41 +1000168#define R1BIO_MadeGood 7
169#define R1BIO_WriteError 8
NeilBrown4b6d2872005-09-09 16:23:47 -0700170
NeilBrownfd01b882011-10-11 16:47:53 +1100171extern int md_raid1_congested(struct mddev *mddev, int bits);
Jonathan Brassow1ed72422011-06-07 17:50:35 -0500172
Linus Torvalds1da177e2005-04-16 15:20:36 -0700173#endif