| 1 | #ifndef _RAID1_H |
| 2 | #define _RAID1_H |
| 3 | |
| 4 | /* |
| 5 | * each barrier unit size is 64MB fow now |
| 6 | * note: it must be larger than RESYNC_DEPTH |
| 7 | */ |
| 8 | #define BARRIER_UNIT_SECTOR_BITS 17 |
| 9 | #define BARRIER_UNIT_SECTOR_SIZE (1<<17) |
| 10 | /* |
| 11 | * In struct r1conf, the following members are related to I/O barrier |
| 12 | * buckets, |
| 13 | * atomic_t *nr_pending; |
| 14 | * atomic_t *nr_waiting; |
| 15 | * atomic_t *nr_queued; |
| 16 | * atomic_t *barrier; |
| 17 | * Each of them points to array of atomic_t variables, each array is |
| 18 | * designed to have BARRIER_BUCKETS_NR elements and occupy a single |
| 19 | * memory page. The data width of atomic_t variables is 4 bytes, equal |
| 20 | * to 1<<(ilog2(sizeof(atomic_t))), BARRIER_BUCKETS_NR_BITS is defined |
| 21 | * as (PAGE_SHIFT - ilog2(sizeof(int))) to make sure an array of |
| 22 | * atomic_t variables with BARRIER_BUCKETS_NR elements just exactly |
| 23 | * occupies a single memory page. |
| 24 | */ |
| 25 | #define BARRIER_BUCKETS_NR_BITS (PAGE_SHIFT - ilog2(sizeof(atomic_t))) |
| 26 | #define BARRIER_BUCKETS_NR (1<<BARRIER_BUCKETS_NR_BITS) |
| 27 | |
| 28 | struct raid1_info { |
| 29 | struct md_rdev *rdev; |
| 30 | sector_t head_position; |
| 31 | |
| 32 | /* When choose the best device for a read (read_balance()) |
| 33 | * we try to keep sequential reads one the same device |
| 34 | */ |
| 35 | sector_t next_seq_sect; |
| 36 | sector_t seq_start; |
| 37 | }; |
| 38 | |
| 39 | /* |
| 40 | * memory pools need a pointer to the mddev, so they can force an unplug |
| 41 | * when memory is tight, and a count of the number of drives that the |
| 42 | * pool was allocated for, so they know how much to allocate and free. |
| 43 | * mddev->raid_disks cannot be used, as it can change while a pool is active |
| 44 | * These two datums are stored in a kmalloced struct. |
| 45 | * The 'raid_disks' here is twice the raid_disks in r1conf. |
| 46 | * This allows space for each 'real' device can have a replacement in the |
| 47 | * second half of the array. |
| 48 | */ |
| 49 | |
| 50 | struct pool_info { |
| 51 | struct mddev *mddev; |
| 52 | int raid_disks; |
| 53 | }; |
| 54 | |
| 55 | struct r1conf { |
| 56 | struct mddev *mddev; |
| 57 | struct raid1_info *mirrors; /* twice 'raid_disks' to |
| 58 | * allow for replacements. |
| 59 | */ |
| 60 | int raid_disks; |
| 61 | |
| 62 | spinlock_t device_lock; |
| 63 | |
| 64 | /* list of 'struct r1bio' that need to be processed by raid1d, |
| 65 | * whether to retry a read, writeout a resync or recovery |
| 66 | * block, or anything else. |
| 67 | */ |
| 68 | struct list_head retry_list; |
| 69 | /* A separate list of r1bio which just need raid_end_bio_io called. |
| 70 | * This mustn't happen for writes which had any errors if the superblock |
| 71 | * needs to be written. |
| 72 | */ |
| 73 | struct list_head bio_end_io_list; |
| 74 | |
| 75 | /* queue pending writes to be submitted on unplug */ |
| 76 | struct bio_list pending_bio_list; |
| 77 | int pending_count; |
| 78 | |
| 79 | /* for use when syncing mirrors: |
| 80 | * We don't allow both normal IO and resync/recovery IO at |
| 81 | * the same time - resync/recovery can only happen when there |
| 82 | * is no other IO. So when either is active, the other has to wait. |
| 83 | * See more details description in raid1.c near raise_barrier(). |
| 84 | */ |
| 85 | wait_queue_head_t wait_barrier; |
| 86 | spinlock_t resync_lock; |
| 87 | atomic_t nr_sync_pending; |
| 88 | atomic_t *nr_pending; |
| 89 | atomic_t *nr_waiting; |
| 90 | atomic_t *nr_queued; |
| 91 | atomic_t *barrier; |
| 92 | int array_frozen; |
| 93 | |
| 94 | /* Set to 1 if a full sync is needed, (fresh device added). |
| 95 | * Cleared when a sync completes. |
| 96 | */ |
| 97 | int fullsync; |
| 98 | |
| 99 | /* When the same as mddev->recovery_disabled we don't allow |
| 100 | * recovery to be attempted as we expect a read error. |
| 101 | */ |
| 102 | int recovery_disabled; |
| 103 | |
| 104 | /* poolinfo contains information about the content of the |
| 105 | * mempools - it changes when the array grows or shrinks |
| 106 | */ |
| 107 | struct pool_info *poolinfo; |
| 108 | mempool_t *r1bio_pool; |
| 109 | mempool_t *r1buf_pool; |
| 110 | |
| 111 | struct bio_set *bio_split; |
| 112 | |
| 113 | /* temporary buffer to synchronous IO when attempting to repair |
| 114 | * a read error. |
| 115 | */ |
| 116 | struct page *tmppage; |
| 117 | |
| 118 | /* When taking over an array from a different personality, we store |
| 119 | * the new thread here until we fully activate the array. |
| 120 | */ |
| 121 | struct md_thread *thread; |
| 122 | |
| 123 | /* Keep track of cluster resync window to send to other |
| 124 | * nodes. |
| 125 | */ |
| 126 | sector_t cluster_sync_low; |
| 127 | sector_t cluster_sync_high; |
| 128 | |
| 129 | }; |
| 130 | |
| 131 | /* |
| 132 | * this is our 'private' RAID1 bio. |
| 133 | * |
| 134 | * it contains information about what kind of IO operations were started |
| 135 | * for this RAID1 operation, and about their status: |
| 136 | */ |
| 137 | |
| 138 | struct r1bio { |
| 139 | atomic_t remaining; /* 'have we finished' count, |
| 140 | * used from IRQ handlers |
| 141 | */ |
| 142 | atomic_t behind_remaining; /* number of write-behind ios remaining |
| 143 | * in this BehindIO request |
| 144 | */ |
| 145 | sector_t sector; |
| 146 | int sectors; |
| 147 | unsigned long state; |
| 148 | struct mddev *mddev; |
| 149 | /* |
| 150 | * original bio going to /dev/mdx |
| 151 | */ |
| 152 | struct bio *master_bio; |
| 153 | /* |
| 154 | * if the IO is in READ direction, then this is where we read |
| 155 | */ |
| 156 | int read_disk; |
| 157 | |
| 158 | struct list_head retry_list; |
| 159 | |
| 160 | /* |
| 161 | * When R1BIO_BehindIO is set, we store pages for write behind |
| 162 | * in behind_master_bio. |
| 163 | */ |
| 164 | struct bio *behind_master_bio; |
| 165 | |
| 166 | /* |
| 167 | * if the IO is in WRITE direction, then multiple bios are used. |
| 168 | * We choose the number when they are allocated. |
| 169 | */ |
| 170 | struct bio *bios[0]; |
| 171 | /* DO NOT PUT ANY NEW FIELDS HERE - bios array is contiguously alloced*/ |
| 172 | }; |
| 173 | |
| 174 | /* bits for r1bio.state */ |
| 175 | enum r1bio_state { |
| 176 | R1BIO_Uptodate, |
| 177 | R1BIO_IsSync, |
| 178 | R1BIO_Degraded, |
| 179 | R1BIO_BehindIO, |
| 180 | /* Set ReadError on bios that experience a readerror so that |
| 181 | * raid1d knows what to do with them. |
| 182 | */ |
| 183 | R1BIO_ReadError, |
| 184 | /* For write-behind requests, we call bi_end_io when |
| 185 | * the last non-write-behind device completes, providing |
| 186 | * any write was successful. Otherwise we call when |
| 187 | * any write-behind write succeeds, otherwise we call |
| 188 | * with failure when last write completes (and all failed). |
| 189 | * Record that bi_end_io was called with this flag... |
| 190 | */ |
| 191 | R1BIO_Returned, |
| 192 | /* If a write for this request means we can clear some |
| 193 | * known-bad-block records, we set this flag |
| 194 | */ |
| 195 | R1BIO_MadeGood, |
| 196 | R1BIO_WriteError, |
| 197 | R1BIO_FailFast, |
| 198 | }; |
| 199 | |
| 200 | static inline int sector_to_idx(sector_t sector) |
| 201 | { |
| 202 | return hash_long(sector >> BARRIER_UNIT_SECTOR_BITS, |
| 203 | BARRIER_BUCKETS_NR_BITS); |
| 204 | } |
| 205 | #endif |