Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | #ifndef _RAID1_H |
2 | #define _RAID1_H | |
3 | ||
fd76863e | 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, | |
824e47da | 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. | |
fd76863e | 24 | */ |
824e47da | 25 | #define BARRIER_BUCKETS_NR_BITS (PAGE_SHIFT - ilog2(sizeof(atomic_t))) |
fd76863e | 26 | #define BARRIER_BUCKETS_NR (1<<BARRIER_BUCKETS_NR_BITS) |
27 | ||
0eaf822c | 28 | struct raid1_info { |
3cb03002 | 29 | struct md_rdev *rdev; |
1da177e4 | 30 | sector_t head_position; |
be4d3280 SL |
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; | |
12cee5a8 | 36 | sector_t seq_start; |
1da177e4 LT |
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. | |
8f19ccb2 N |
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. | |
1da177e4 LT |
48 | */ |
49 | ||
50 | struct pool_info { | |
fd01b88c | 51 | struct mddev *mddev; |
1da177e4 LT |
52 | int raid_disks; |
53 | }; | |
54 | ||
e8096360 | 55 | struct r1conf { |
fd01b88c | 56 | struct mddev *mddev; |
0eaf822c | 57 | struct raid1_info *mirrors; /* twice 'raid_disks' to |
8f19ccb2 N |
58 | * allow for replacements. |
59 | */ | |
1da177e4 | 60 | int raid_disks; |
ce550c20 | 61 | |
1da177e4 LT |
62 | spinlock_t device_lock; |
63 | ||
9f2c9d12 N |
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. | |
ce550c20 | 67 | */ |
1da177e4 | 68 | struct list_head retry_list; |
55ce74d4 N |
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; | |
191ea9b2 | 74 | |
ce550c20 N |
75 | /* queue pending writes to be submitted on unplug */ |
76 | struct bio_list pending_bio_list; | |
34db0cd6 | 77 | int pending_count; |
1da177e4 | 78 | |
ce550c20 N |
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; | |
1da177e4 | 86 | spinlock_t resync_lock; |
824e47da | 87 | atomic_t *nr_pending; |
88 | atomic_t *nr_waiting; | |
89 | atomic_t *nr_queued; | |
90 | atomic_t *barrier; | |
b364e3d0 | 91 | int array_frozen; |
1da177e4 | 92 | |
ce550c20 N |
93 | /* Set to 1 if a full sync is needed, (fresh device added). |
94 | * Cleared when a sync completes. | |
95 | */ | |
96 | int fullsync; | |
1da177e4 | 97 | |
ce550c20 N |
98 | /* When the same as mddev->recovery_disabled we don't allow |
99 | * recovery to be attempted as we expect a read error. | |
100 | */ | |
101 | int recovery_disabled; | |
1da177e4 | 102 | |
ce550c20 N |
103 | /* poolinfo contains information about the content of the |
104 | * mempools - it changes when the array grows or shrinks | |
105 | */ | |
106 | struct pool_info *poolinfo; | |
9f2c9d12 N |
107 | mempool_t *r1bio_pool; |
108 | mempool_t *r1buf_pool; | |
709ae487 | 109 | |
ce550c20 N |
110 | /* temporary buffer to synchronous IO when attempting to repair |
111 | * a read error. | |
112 | */ | |
113 | struct page *tmppage; | |
114 | ||
709ae487 N |
115 | /* When taking over an array from a different personality, we store |
116 | * the new thread here until we fully activate the array. | |
117 | */ | |
2b8bf345 | 118 | struct md_thread *thread; |
c40f341f GR |
119 | |
120 | /* Keep track of cluster resync window to send to other | |
121 | * nodes. | |
122 | */ | |
123 | sector_t cluster_sync_low; | |
124 | sector_t cluster_sync_high; | |
125 | ||
1da177e4 LT |
126 | }; |
127 | ||
1da177e4 LT |
128 | /* |
129 | * this is our 'private' RAID1 bio. | |
130 | * | |
131 | * it contains information about what kind of IO operations were started | |
132 | * for this RAID1 operation, and about their status: | |
133 | */ | |
134 | ||
9f2c9d12 | 135 | struct r1bio { |
1da177e4 LT |
136 | atomic_t remaining; /* 'have we finished' count, |
137 | * used from IRQ handlers | |
138 | */ | |
4b6d287f N |
139 | atomic_t behind_remaining; /* number of write-behind ios remaining |
140 | * in this BehindIO request | |
141 | */ | |
1da177e4 LT |
142 | sector_t sector; |
143 | int sectors; | |
144 | unsigned long state; | |
fd01b88c | 145 | struct mddev *mddev; |
1da177e4 LT |
146 | /* |
147 | * original bio going to /dev/mdx | |
148 | */ | |
149 | struct bio *master_bio; | |
150 | /* | |
151 | * if the IO is in READ direction, then this is where we read | |
152 | */ | |
153 | int read_disk; | |
154 | ||
155 | struct list_head retry_list; | |
841c1316 ML |
156 | |
157 | /* | |
158 | * When R1BIO_BehindIO is set, we store pages for write behind | |
159 | * in behind_master_bio. | |
160 | */ | |
161 | struct bio *behind_master_bio; | |
162 | ||
1da177e4 LT |
163 | /* |
164 | * if the IO is in WRITE direction, then multiple bios are used. | |
165 | * We choose the number when they are allocated. | |
166 | */ | |
167 | struct bio *bios[0]; | |
191ea9b2 | 168 | /* DO NOT PUT ANY NEW FIELDS HERE - bios array is contiguously alloced*/ |
1da177e4 LT |
169 | }; |
170 | ||
171 | /* bits for r1bio.state */ | |
be306c29 N |
172 | enum r1bio_state { |
173 | R1BIO_Uptodate, | |
174 | R1BIO_IsSync, | |
175 | R1BIO_Degraded, | |
176 | R1BIO_BehindIO, | |
d2eb35ac N |
177 | /* Set ReadError on bios that experience a readerror so that |
178 | * raid1d knows what to do with them. | |
179 | */ | |
be306c29 | 180 | R1BIO_ReadError, |
4b6d287f N |
181 | /* For write-behind requests, we call bi_end_io when |
182 | * the last non-write-behind device completes, providing | |
183 | * any write was successful. Otherwise we call when | |
184 | * any write-behind write succeeds, otherwise we call | |
185 | * with failure when last write completes (and all failed). | |
186 | * Record that bi_end_io was called with this flag... | |
187 | */ | |
be306c29 | 188 | R1BIO_Returned, |
4367af55 N |
189 | /* If a write for this request means we can clear some |
190 | * known-bad-block records, we set this flag | |
191 | */ | |
be306c29 N |
192 | R1BIO_MadeGood, |
193 | R1BIO_WriteError, | |
2e52d449 | 194 | R1BIO_FailFast, |
be306c29 | 195 | }; |
fd76863e | 196 | |
197 | static inline int sector_to_idx(sector_t sector) | |
198 | { | |
199 | return hash_long(sector >> BARRIER_UNIT_SECTOR_BITS, | |
200 | BARRIER_BUCKETS_NR_BITS); | |
201 | } | |
1da177e4 | 202 | #endif |