blob: 0ff3715fb7eba5ec4fed61a9922276b07b363aff [file] [log] [blame]
Linus Torvalds1da177e2005-04-16 15:20:36 -07001#ifndef _RAID1_H
2#define _RAID1_H
3
Jonathan Brassow0eaf8222012-07-31 10:03:52 +10004struct raid1_info {
NeilBrown3cb03002011-10-11 16:45:26 +11005 struct md_rdev *rdev;
Linus Torvalds1da177e2005-04-16 15:20:36 -07006 sector_t head_position;
Shaohua Libe4d3282012-07-31 10:03:53 +10007
8 /* When choose the best device for a read (read_balance())
9 * we try to keep sequential reads one the same device
10 */
11 sector_t next_seq_sect;
Shaohua Li12cee5a2012-07-31 10:03:53 +100012 sector_t seq_start;
Linus Torvalds1da177e2005-04-16 15:20:36 -070013};
14
15/*
16 * memory pools need a pointer to the mddev, so they can force an unplug
17 * when memory is tight, and a count of the number of drives that the
18 * pool was allocated for, so they know how much to allocate and free.
19 * mddev->raid_disks cannot be used, as it can change while a pool is active
20 * These two datums are stored in a kmalloced struct.
NeilBrown8f19ccb2011-12-23 10:17:56 +110021 * The 'raid_disks' here is twice the raid_disks in r1conf.
22 * This allows space for each 'real' device can have a replacement in the
23 * second half of the array.
Linus Torvalds1da177e2005-04-16 15:20:36 -070024 */
25
26struct pool_info {
NeilBrownfd01b882011-10-11 16:47:53 +110027 struct mddev *mddev;
Linus Torvalds1da177e2005-04-16 15:20:36 -070028 int raid_disks;
29};
30
NeilBrowne8096362011-10-11 16:49:05 +110031struct r1conf {
NeilBrownfd01b882011-10-11 16:47:53 +110032 struct mddev *mddev;
Jonathan Brassow0eaf8222012-07-31 10:03:52 +100033 struct raid1_info *mirrors; /* twice 'raid_disks' to
NeilBrown8f19ccb2011-12-23 10:17:56 +110034 * allow for replacements.
35 */
Linus Torvalds1da177e2005-04-16 15:20:36 -070036 int raid_disks;
NeilBrownce550c22011-10-07 14:22:33 +110037
NeilBrownce550c22011-10-07 14:22:33 +110038 /* During resync, read_balancing is only allowed on the part
39 * of the array that has been resynced. 'next_resync' tells us
40 * where that is.
41 */
42 sector_t next_resync;
43
Linus Torvalds1da177e2005-04-16 15:20:36 -070044 spinlock_t device_lock;
45
NeilBrown9f2c9d12011-10-11 16:48:43 +110046 /* list of 'struct r1bio' that need to be processed by raid1d,
47 * whether to retry a read, writeout a resync or recovery
48 * block, or anything else.
NeilBrownce550c22011-10-07 14:22:33 +110049 */
Linus Torvalds1da177e2005-04-16 15:20:36 -070050 struct list_head retry_list;
NeilBrownce550c22011-10-07 14:22:33 +110051
52 /* queue pending writes to be submitted on unplug */
NeilBrown191ea9b2005-06-21 17:17:23 -070053 struct bio_list pending_bio_list;
NeilBrown34db0cd2011-10-11 16:50:01 +110054 int pending_count;
NeilBrown191ea9b2005-06-21 17:17:23 -070055
NeilBrownce550c22011-10-07 14:22:33 +110056 /* for use when syncing mirrors:
57 * We don't allow both normal IO and resync/recovery IO at
58 * the same time - resync/recovery can only happen when there
59 * is no other IO. So when either is active, the other has to wait.
60 * See more details description in raid1.c near raise_barrier().
61 */
62 wait_queue_head_t wait_barrier;
Linus Torvalds1da177e2005-04-16 15:20:36 -070063 spinlock_t resync_lock;
NeilBrown191ea9b2005-06-21 17:17:23 -070064 int nr_pending;
NeilBrown17999be2006-01-06 00:20:12 -080065 int nr_waiting;
NeilBrownddaf22a2006-01-06 00:20:19 -080066 int nr_queued;
NeilBrown191ea9b2005-06-21 17:17:23 -070067 int barrier;
Linus Torvalds1da177e2005-04-16 15:20:36 -070068
NeilBrownce550c22011-10-07 14:22:33 +110069 /* Set to 1 if a full sync is needed, (fresh device added).
70 * Cleared when a sync completes.
71 */
72 int fullsync;
Linus Torvalds1da177e2005-04-16 15:20:36 -070073
NeilBrownce550c22011-10-07 14:22:33 +110074 /* When the same as mddev->recovery_disabled we don't allow
75 * recovery to be attempted as we expect a read error.
76 */
77 int recovery_disabled;
78
79
80 /* poolinfo contains information about the content of the
81 * mempools - it changes when the array grows or shrinks
82 */
Linus Torvalds1da177e2005-04-16 15:20:36 -070083 struct pool_info *poolinfo;
NeilBrown9f2c9d12011-10-11 16:48:43 +110084 mempool_t *r1bio_pool;
85 mempool_t *r1buf_pool;
NeilBrown709ae482009-12-14 12:49:51 +110086
NeilBrownce550c22011-10-07 14:22:33 +110087 /* temporary buffer to synchronous IO when attempting to repair
88 * a read error.
89 */
90 struct page *tmppage;
91
92
NeilBrown709ae482009-12-14 12:49:51 +110093 /* When taking over an array from a different personality, we store
94 * the new thread here until we fully activate the array.
95 */
NeilBrown2b8bf342011-10-11 16:48:23 +110096 struct md_thread *thread;
Linus Torvalds1da177e2005-04-16 15:20:36 -070097};
98
Linus Torvalds1da177e2005-04-16 15:20:36 -070099/*
Linus Torvalds1da177e2005-04-16 15:20:36 -0700100 * this is our 'private' RAID1 bio.
101 *
102 * it contains information about what kind of IO operations were started
103 * for this RAID1 operation, and about their status:
104 */
105
NeilBrown9f2c9d12011-10-11 16:48:43 +1100106struct r1bio {
Linus Torvalds1da177e2005-04-16 15:20:36 -0700107 atomic_t remaining; /* 'have we finished' count,
108 * used from IRQ handlers
109 */
NeilBrown4b6d2872005-09-09 16:23:47 -0700110 atomic_t behind_remaining; /* number of write-behind ios remaining
111 * in this BehindIO request
112 */
Linus Torvalds1da177e2005-04-16 15:20:36 -0700113 sector_t sector;
114 int sectors;
115 unsigned long state;
NeilBrownfd01b882011-10-11 16:47:53 +1100116 struct mddev *mddev;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700117 /*
118 * original bio going to /dev/mdx
119 */
120 struct bio *master_bio;
121 /*
122 * if the IO is in READ direction, then this is where we read
123 */
124 int read_disk;
125
126 struct list_head retry_list;
NeilBrownaf6d7b72011-05-11 14:51:19 +1000127 /* Next two are only valid when R1BIO_BehindIO is set */
NeilBrown2ca68f52011-07-28 11:32:10 +1000128 struct bio_vec *behind_bvecs;
NeilBrownaf6d7b72011-05-11 14:51:19 +1000129 int behind_page_count;
Linus Torvalds1da177e2005-04-16 15:20:36 -0700130 /*
131 * if the IO is in WRITE direction, then multiple bios are used.
132 * We choose the number when they are allocated.
133 */
134 struct bio *bios[0];
NeilBrown191ea9b2005-06-21 17:17:23 -0700135 /* DO NOT PUT ANY NEW FIELDS HERE - bios array is contiguously alloced*/
Linus Torvalds1da177e2005-04-16 15:20:36 -0700136};
137
138/* bits for r1bio.state */
139#define R1BIO_Uptodate 0
140#define R1BIO_IsSync 1
NeilBrown191ea9b2005-06-21 17:17:23 -0700141#define R1BIO_Degraded 2
NeilBrowna9701a32005-11-08 21:39:34 -0800142#define R1BIO_BehindIO 3
NeilBrownd2eb35a2011-07-28 11:31:48 +1000143/* Set ReadError on bios that experience a readerror so that
144 * raid1d knows what to do with them.
145 */
146#define R1BIO_ReadError 4
NeilBrown4b6d2872005-09-09 16:23:47 -0700147/* For write-behind requests, we call bi_end_io when
148 * the last non-write-behind device completes, providing
149 * any write was successful. Otherwise we call when
150 * any write-behind write succeeds, otherwise we call
151 * with failure when last write completes (and all failed).
152 * Record that bi_end_io was called with this flag...
153 */
NeilBrown9e71f9c2006-03-23 02:59:22 -0800154#define R1BIO_Returned 6
NeilBrown4367af52011-07-28 11:31:49 +1000155/* If a write for this request means we can clear some
156 * known-bad-block records, we set this flag
157 */
NeilBrowncd5ff9a12011-07-28 11:32:41 +1000158#define R1BIO_MadeGood 7
159#define R1BIO_WriteError 8
NeilBrown4b6d2872005-09-09 16:23:47 -0700160
NeilBrownfd01b882011-10-11 16:47:53 +1100161extern int md_raid1_congested(struct mddev *mddev, int bits);
Jonathan Brassow1ed72422011-06-07 17:50:35 -0500162
Linus Torvalds1da177e2005-04-16 15:20:36 -0700163#endif