Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1 | #ifndef _RAID1_H |
| 2 | #define _RAID1_H |
| 3 | |
Jonathan Brassow | 0eaf822 | 2012-07-31 10:03:52 +1000 | [diff] [blame] | 4 | struct raid1_info { |
NeilBrown | 3cb0300 | 2011-10-11 16:45:26 +1100 | [diff] [blame] | 5 | struct md_rdev *rdev; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 6 | sector_t head_position; |
Shaohua Li | be4d328 | 2012-07-31 10:03:53 +1000 | [diff] [blame] | 7 | |
| 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 Li | 12cee5a | 2012-07-31 10:03:53 +1000 | [diff] [blame] | 12 | sector_t seq_start; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 13 | }; |
| 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. |
NeilBrown | 8f19ccb | 2011-12-23 10:17:56 +1100 | [diff] [blame] | 21 | * 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 Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 24 | */ |
| 25 | |
| 26 | struct pool_info { |
NeilBrown | fd01b88 | 2011-10-11 16:47:53 +1100 | [diff] [blame] | 27 | struct mddev *mddev; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 28 | int raid_disks; |
| 29 | }; |
| 30 | |
NeilBrown | e809636 | 2011-10-11 16:49:05 +1100 | [diff] [blame] | 31 | struct r1conf { |
NeilBrown | fd01b88 | 2011-10-11 16:47:53 +1100 | [diff] [blame] | 32 | struct mddev *mddev; |
Jonathan Brassow | 0eaf822 | 2012-07-31 10:03:52 +1000 | [diff] [blame] | 33 | struct raid1_info *mirrors; /* twice 'raid_disks' to |
NeilBrown | 8f19ccb | 2011-12-23 10:17:56 +1100 | [diff] [blame] | 34 | * allow for replacements. |
| 35 | */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 36 | int raid_disks; |
NeilBrown | ce550c2 | 2011-10-07 14:22:33 +1100 | [diff] [blame] | 37 | |
NeilBrown | ce550c2 | 2011-10-07 14:22:33 +1100 | [diff] [blame] | 38 | /* 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 Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 44 | spinlock_t device_lock; |
| 45 | |
NeilBrown | 9f2c9d1 | 2011-10-11 16:48:43 +1100 | [diff] [blame] | 46 | /* 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. |
NeilBrown | ce550c2 | 2011-10-07 14:22:33 +1100 | [diff] [blame] | 49 | */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 50 | struct list_head retry_list; |
NeilBrown | ce550c2 | 2011-10-07 14:22:33 +1100 | [diff] [blame] | 51 | |
| 52 | /* queue pending writes to be submitted on unplug */ |
NeilBrown | 191ea9b | 2005-06-21 17:17:23 -0700 | [diff] [blame] | 53 | struct bio_list pending_bio_list; |
NeilBrown | 34db0cd | 2011-10-11 16:50:01 +1100 | [diff] [blame] | 54 | int pending_count; |
NeilBrown | 191ea9b | 2005-06-21 17:17:23 -0700 | [diff] [blame] | 55 | |
NeilBrown | ce550c2 | 2011-10-07 14:22:33 +1100 | [diff] [blame] | 56 | /* 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 Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 63 | spinlock_t resync_lock; |
NeilBrown | 191ea9b | 2005-06-21 17:17:23 -0700 | [diff] [blame] | 64 | int nr_pending; |
NeilBrown | 17999be | 2006-01-06 00:20:12 -0800 | [diff] [blame] | 65 | int nr_waiting; |
NeilBrown | ddaf22a | 2006-01-06 00:20:19 -0800 | [diff] [blame] | 66 | int nr_queued; |
NeilBrown | 191ea9b | 2005-06-21 17:17:23 -0700 | [diff] [blame] | 67 | int barrier; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 68 | |
NeilBrown | ce550c2 | 2011-10-07 14:22:33 +1100 | [diff] [blame] | 69 | /* Set to 1 if a full sync is needed, (fresh device added). |
| 70 | * Cleared when a sync completes. |
| 71 | */ |
| 72 | int fullsync; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 73 | |
NeilBrown | ce550c2 | 2011-10-07 14:22:33 +1100 | [diff] [blame] | 74 | /* 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 Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 83 | struct pool_info *poolinfo; |
NeilBrown | 9f2c9d1 | 2011-10-11 16:48:43 +1100 | [diff] [blame] | 84 | mempool_t *r1bio_pool; |
| 85 | mempool_t *r1buf_pool; |
NeilBrown | 709ae48 | 2009-12-14 12:49:51 +1100 | [diff] [blame] | 86 | |
NeilBrown | ce550c2 | 2011-10-07 14:22:33 +1100 | [diff] [blame] | 87 | /* temporary buffer to synchronous IO when attempting to repair |
| 88 | * a read error. |
| 89 | */ |
| 90 | struct page *tmppage; |
| 91 | |
| 92 | |
NeilBrown | 709ae48 | 2009-12-14 12:49:51 +1100 | [diff] [blame] | 93 | /* When taking over an array from a different personality, we store |
| 94 | * the new thread here until we fully activate the array. |
| 95 | */ |
NeilBrown | 2b8bf34 | 2011-10-11 16:48:23 +1100 | [diff] [blame] | 96 | struct md_thread *thread; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 97 | }; |
| 98 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 99 | /* |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 100 | * 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 | |
NeilBrown | 9f2c9d1 | 2011-10-11 16:48:43 +1100 | [diff] [blame] | 106 | struct r1bio { |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 107 | atomic_t remaining; /* 'have we finished' count, |
| 108 | * used from IRQ handlers |
| 109 | */ |
NeilBrown | 4b6d287 | 2005-09-09 16:23:47 -0700 | [diff] [blame] | 110 | atomic_t behind_remaining; /* number of write-behind ios remaining |
| 111 | * in this BehindIO request |
| 112 | */ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 113 | sector_t sector; |
| 114 | int sectors; |
| 115 | unsigned long state; |
NeilBrown | fd01b88 | 2011-10-11 16:47:53 +1100 | [diff] [blame] | 116 | struct mddev *mddev; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 117 | /* |
| 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; |
NeilBrown | af6d7b7 | 2011-05-11 14:51:19 +1000 | [diff] [blame] | 127 | /* Next two are only valid when R1BIO_BehindIO is set */ |
NeilBrown | 2ca68f5 | 2011-07-28 11:32:10 +1000 | [diff] [blame] | 128 | struct bio_vec *behind_bvecs; |
NeilBrown | af6d7b7 | 2011-05-11 14:51:19 +1000 | [diff] [blame] | 129 | int behind_page_count; |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 130 | /* |
| 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]; |
NeilBrown | 191ea9b | 2005-06-21 17:17:23 -0700 | [diff] [blame] | 135 | /* DO NOT PUT ANY NEW FIELDS HERE - bios array is contiguously alloced*/ |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 136 | }; |
| 137 | |
| 138 | /* bits for r1bio.state */ |
| 139 | #define R1BIO_Uptodate 0 |
| 140 | #define R1BIO_IsSync 1 |
NeilBrown | 191ea9b | 2005-06-21 17:17:23 -0700 | [diff] [blame] | 141 | #define R1BIO_Degraded 2 |
NeilBrown | a9701a3 | 2005-11-08 21:39:34 -0800 | [diff] [blame] | 142 | #define R1BIO_BehindIO 3 |
NeilBrown | d2eb35a | 2011-07-28 11:31:48 +1000 | [diff] [blame] | 143 | /* Set ReadError on bios that experience a readerror so that |
| 144 | * raid1d knows what to do with them. |
| 145 | */ |
| 146 | #define R1BIO_ReadError 4 |
NeilBrown | 4b6d287 | 2005-09-09 16:23:47 -0700 | [diff] [blame] | 147 | /* 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 | */ |
NeilBrown | 9e71f9c | 2006-03-23 02:59:22 -0800 | [diff] [blame] | 154 | #define R1BIO_Returned 6 |
NeilBrown | 4367af5 | 2011-07-28 11:31:49 +1000 | [diff] [blame] | 155 | /* If a write for this request means we can clear some |
| 156 | * known-bad-block records, we set this flag |
| 157 | */ |
NeilBrown | cd5ff9a1 | 2011-07-28 11:32:41 +1000 | [diff] [blame] | 158 | #define R1BIO_MadeGood 7 |
| 159 | #define R1BIO_WriteError 8 |
NeilBrown | 4b6d287 | 2005-09-09 16:23:47 -0700 | [diff] [blame] | 160 | |
NeilBrown | fd01b88 | 2011-10-11 16:47:53 +1100 | [diff] [blame] | 161 | extern int md_raid1_congested(struct mddev *mddev, int bits); |
Jonathan Brassow | 1ed7242 | 2011-06-07 17:50:35 -0500 | [diff] [blame] | 162 | |
Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 163 | #endif |