blob: d59e4676a1b05b37f24898c17d195a48cdfc6f3d [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 -07004typedef struct mirror_info mirror_info_t;
5
6struct mirror_info {
NeilBrown3cb03002011-10-11 16:45:26 +11007 struct md_rdev *rdev;
Linus Torvalds1da177e2005-04-16 15:20:36 -07008 sector_t head_position;
9};
10
11/*
12 * memory pools need a pointer to the mddev, so they can force an unplug
13 * when memory is tight, and a count of the number of drives that the
14 * pool was allocated for, so they know how much to allocate and free.
15 * mddev->raid_disks cannot be used, as it can change while a pool is active
16 * These two datums are stored in a kmalloced struct.
17 */
18
19struct pool_info {
NeilBrownfd01b882011-10-11 16:47:53 +110020 struct mddev *mddev;
Linus Torvalds1da177e2005-04-16 15:20:36 -070021 int raid_disks;
22};
23
Linus Torvalds1da177e2005-04-16 15:20:36 -070024struct r1_private_data_s {
NeilBrownfd01b882011-10-11 16:47:53 +110025 struct mddev *mddev;
Linus Torvalds1da177e2005-04-16 15:20:36 -070026 mirror_info_t *mirrors;
27 int raid_disks;
NeilBrownce550c22011-10-07 14:22:33 +110028
29 /* When choose the best device for a read (read_balance())
30 * we try to keep sequential reads one the same device
31 * using 'last_used' and 'next_seq_sect'
32 */
Linus Torvalds1da177e2005-04-16 15:20:36 -070033 int last_used;
34 sector_t next_seq_sect;
NeilBrownce550c22011-10-07 14:22:33 +110035 /* During resync, read_balancing is only allowed on the part
36 * of the array that has been resynced. 'next_resync' tells us
37 * where that is.
38 */
39 sector_t next_resync;
40
Linus Torvalds1da177e2005-04-16 15:20:36 -070041 spinlock_t device_lock;
42
NeilBrown9f2c9d12011-10-11 16:48:43 +110043 /* list of 'struct r1bio' that need to be processed by raid1d,
44 * whether to retry a read, writeout a resync or recovery
45 * block, or anything else.
NeilBrownce550c22011-10-07 14:22:33 +110046 */
Linus Torvalds1da177e2005-04-16 15:20:36 -070047 struct list_head retry_list;
NeilBrownce550c22011-10-07 14:22:33 +110048
49 /* queue pending writes to be submitted on unplug */
NeilBrown191ea9b2005-06-21 17:17:23 -070050 struct bio_list pending_bio_list;
NeilBrown191ea9b2005-06-21 17:17:23 -070051
NeilBrownce550c22011-10-07 14:22:33 +110052 /* for use when syncing mirrors:
53 * We don't allow both normal IO and resync/recovery IO at
54 * the same time - resync/recovery can only happen when there
55 * is no other IO. So when either is active, the other has to wait.
56 * See more details description in raid1.c near raise_barrier().
57 */
58 wait_queue_head_t wait_barrier;
Linus Torvalds1da177e2005-04-16 15:20:36 -070059 spinlock_t resync_lock;
NeilBrown191ea9b2005-06-21 17:17:23 -070060 int nr_pending;
NeilBrown17999be2006-01-06 00:20:12 -080061 int nr_waiting;
NeilBrownddaf22a2006-01-06 00:20:19 -080062 int nr_queued;
NeilBrown191ea9b2005-06-21 17:17:23 -070063 int barrier;
Linus Torvalds1da177e2005-04-16 15:20:36 -070064
NeilBrownce550c22011-10-07 14:22:33 +110065 /* Set to 1 if a full sync is needed, (fresh device added).
66 * Cleared when a sync completes.
67 */
68 int fullsync;
Linus Torvalds1da177e2005-04-16 15:20:36 -070069
NeilBrownce550c22011-10-07 14:22:33 +110070 /* When the same as mddev->recovery_disabled we don't allow
71 * recovery to be attempted as we expect a read error.
72 */
73 int recovery_disabled;
74
75
76 /* poolinfo contains information about the content of the
77 * mempools - it changes when the array grows or shrinks
78 */
Linus Torvalds1da177e2005-04-16 15:20:36 -070079 struct pool_info *poolinfo;
NeilBrown9f2c9d12011-10-11 16:48:43 +110080 mempool_t *r1bio_pool;
81 mempool_t *r1buf_pool;
NeilBrown709ae482009-12-14 12:49:51 +110082
NeilBrownce550c22011-10-07 14:22:33 +110083 /* temporary buffer to synchronous IO when attempting to repair
84 * a read error.
85 */
86 struct page *tmppage;
87
88
NeilBrown709ae482009-12-14 12:49:51 +110089 /* When taking over an array from a different personality, we store
90 * the new thread here until we fully activate the array.
91 */
NeilBrown2b8bf342011-10-11 16:48:23 +110092 struct md_thread *thread;
Linus Torvalds1da177e2005-04-16 15:20:36 -070093};
94
95typedef struct r1_private_data_s conf_t;
96
97/*
Linus Torvalds1da177e2005-04-16 15:20:36 -070098 * this is our 'private' RAID1 bio.
99 *
100 * it contains information about what kind of IO operations were started
101 * for this RAID1 operation, and about their status:
102 */
103
NeilBrown9f2c9d12011-10-11 16:48:43 +1100104struct r1bio {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700105 atomic_t remaining; /* 'have we finished' count,
106 * used from IRQ handlers
107 */
NeilBrown4b6d2872005-09-09 16:23:47 -0700108 atomic_t behind_remaining; /* number of write-behind ios remaining
109 * in this BehindIO request
110 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700111 sector_t sector;
112 int sectors;
113 unsigned long state;
NeilBrownfd01b882011-10-11 16:47:53 +1100114 struct mddev *mddev;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700115 /*
116 * original bio going to /dev/mdx
117 */
118 struct bio *master_bio;
119 /*
120 * if the IO is in READ direction, then this is where we read
121 */
122 int read_disk;
123
124 struct list_head retry_list;
NeilBrownaf6d7b72011-05-11 14:51:19 +1000125 /* Next two are only valid when R1BIO_BehindIO is set */
NeilBrown2ca68f52011-07-28 11:32:10 +1000126 struct bio_vec *behind_bvecs;
NeilBrownaf6d7b72011-05-11 14:51:19 +1000127 int behind_page_count;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700128 /*
129 * if the IO is in WRITE direction, then multiple bios are used.
130 * We choose the number when they are allocated.
131 */
132 struct bio *bios[0];
NeilBrown191ea9b2005-06-21 17:17:23 -0700133 /* DO NOT PUT ANY NEW FIELDS HERE - bios array is contiguously alloced*/
Linus Torvalds1da177e2005-04-16 15:20:36 -0700134};
135
NeilBrowncf30a472006-01-06 00:20:23 -0800136/* when we get a read error on a read-only array, we redirect to another
137 * device without failing the first device, or trying to over-write to
138 * correct the read error. To keep track of bad blocks on a per-bio
139 * level, we store IO_BLOCKED in the appropriate 'bios' pointer
140 */
NeilBrown4367af52011-07-28 11:31:49 +1000141#define IO_BLOCKED ((struct bio *)1)
142/* When we successfully write to a known bad-block, we need to remove the
143 * bad-block marking which must be done from process context. So we record
144 * the success by setting bios[n] to IO_MADE_GOOD
145 */
146#define IO_MADE_GOOD ((struct bio *)2)
147
148#define BIO_SPECIAL(bio) ((unsigned long)bio <= 2)
NeilBrowncf30a472006-01-06 00:20:23 -0800149
Linus Torvalds1da177e2005-04-16 15:20:36 -0700150/* bits for r1bio.state */
151#define R1BIO_Uptodate 0
152#define R1BIO_IsSync 1
NeilBrown191ea9b2005-06-21 17:17:23 -0700153#define R1BIO_Degraded 2
NeilBrowna9701a32005-11-08 21:39:34 -0800154#define R1BIO_BehindIO 3
NeilBrownd2eb35a2011-07-28 11:31:48 +1000155/* Set ReadError on bios that experience a readerror so that
156 * raid1d knows what to do with them.
157 */
158#define R1BIO_ReadError 4
NeilBrown4b6d2872005-09-09 16:23:47 -0700159/* For write-behind requests, we call bi_end_io when
160 * the last non-write-behind device completes, providing
161 * any write was successful. Otherwise we call when
162 * any write-behind write succeeds, otherwise we call
163 * with failure when last write completes (and all failed).
164 * Record that bi_end_io was called with this flag...
165 */
NeilBrown9e71f9c2006-03-23 02:59:22 -0800166#define R1BIO_Returned 6
NeilBrown4367af52011-07-28 11:31:49 +1000167/* If a write for this request means we can clear some
168 * known-bad-block records, we set this flag
169 */
NeilBrowncd5ff9a12011-07-28 11:32:41 +1000170#define R1BIO_MadeGood 7
171#define R1BIO_WriteError 8
NeilBrown4b6d2872005-09-09 16:23:47 -0700172
NeilBrownfd01b882011-10-11 16:47:53 +1100173extern int md_raid1_congested(struct mddev *mddev, int bits);
Jonathan Brassow1ed72422011-06-07 17:50:35 -0500174
Linus Torvalds1da177e2005-04-16 15:20:36 -0700175#endif