diff options
Diffstat (limited to 'include/linux/raid')
| -rw-r--r-- | include/linux/raid/bitmap.h | 273 | ||||
| -rw-r--r-- | include/linux/raid/md.h | 17 | ||||
| -rw-r--r-- | include/linux/raid/md_k.h | 22 | ||||
| -rw-r--r-- | include/linux/raid/md_p.h | 9 | ||||
| -rw-r--r-- | include/linux/raid/md_u.h | 7 | ||||
| -rw-r--r-- | include/linux/raid/raid1.h | 16 |
6 files changed, 337 insertions, 7 deletions
diff --git a/include/linux/raid/bitmap.h b/include/linux/raid/bitmap.h new file mode 100644 index 0000000000..e24b74b111 --- /dev/null +++ b/include/linux/raid/bitmap.h | |||
| @@ -0,0 +1,273 @@ | |||
| 1 | /* | ||
| 2 | * bitmap.h: Copyright (C) Peter T. Breuer (ptb@ot.uc3m.es) 2003 | ||
| 3 | * | ||
| 4 | * additions: Copyright (C) 2003-2004, Paul Clements, SteelEye Technology, Inc. | ||
| 5 | */ | ||
| 6 | #ifndef BITMAP_H | ||
| 7 | #define BITMAP_H 1 | ||
| 8 | |||
| 9 | #define BITMAP_MAJOR 3 | ||
| 10 | #define BITMAP_MINOR 38 | ||
| 11 | |||
| 12 | /* | ||
| 13 | * in-memory bitmap: | ||
| 14 | * | ||
| 15 | * Use 16 bit block counters to track pending writes to each "chunk". | ||
| 16 | * The 2 high order bits are special-purpose, the first is a flag indicating | ||
| 17 | * whether a resync is needed. The second is a flag indicating whether a | ||
| 18 | * resync is active. | ||
| 19 | * This means that the counter is actually 14 bits: | ||
| 20 | * | ||
| 21 | * +--------+--------+------------------------------------------------+ | ||
| 22 | * | resync | resync | counter | | ||
| 23 | * | needed | active | | | ||
| 24 | * | (0-1) | (0-1) | (0-16383) | | ||
| 25 | * +--------+--------+------------------------------------------------+ | ||
| 26 | * | ||
| 27 | * The "resync needed" bit is set when: | ||
| 28 | * a '1' bit is read from storage at startup. | ||
| 29 | * a write request fails on some drives | ||
| 30 | * a resync is aborted on a chunk with 'resync active' set | ||
| 31 | * It is cleared (and resync-active set) when a resync starts across all drives | ||
| 32 | * of the chunk. | ||
| 33 | * | ||
| 34 | * | ||
| 35 | * The "resync active" bit is set when: | ||
| 36 | * a resync is started on all drives, and resync_needed is set. | ||
| 37 | * resync_needed will be cleared (as long as resync_active wasn't already set). | ||
| 38 | * It is cleared when a resync completes. | ||
| 39 | * | ||
| 40 | * The counter counts pending write requests, plus the on-disk bit. | ||
| 41 | * When the counter is '1' and the resync bits are clear, the on-disk | ||
| 42 | * bit can be cleared aswell, thus setting the counter to 0. | ||
| 43 | * When we set a bit, or in the counter (to start a write), if the fields is | ||
| 44 | * 0, we first set the disk bit and set the counter to 1. | ||
| 45 | * | ||
| 46 | * If the counter is 0, the on-disk bit is clear and the stipe is clean | ||
| 47 | * Anything that dirties the stipe pushes the counter to 2 (at least) | ||
| 48 | * and sets the on-disk bit (lazily). | ||
| 49 | * If a periodic sweep find the counter at 2, it is decremented to 1. | ||
| 50 | * If the sweep find the counter at 1, the on-disk bit is cleared and the | ||
| 51 | * counter goes to zero. | ||
| 52 | * | ||
| 53 | * Also, we'll hijack the "map" pointer itself and use it as two 16 bit block | ||
| 54 | * counters as a fallback when "page" memory cannot be allocated: | ||
| 55 | * | ||
| 56 | * Normal case (page memory allocated): | ||
| 57 | * | ||
| 58 | * page pointer (32-bit) | ||
| 59 | * | ||
| 60 | * [ ] ------+ | ||
| 61 | * | | ||
| 62 | * +-------> [ ][ ]..[ ] (4096 byte page == 2048 counters) | ||
| 63 | * c1 c2 c2048 | ||
| 64 | * | ||
| 65 | * Hijacked case (page memory allocation failed): | ||
| 66 | * | ||
| 67 | * hijacked page pointer (32-bit) | ||
| 68 | * | ||
| 69 | * [ ][ ] (no page memory allocated) | ||
| 70 | * counter #1 (16-bit) counter #2 (16-bit) | ||
| 71 | * | ||
| 72 | */ | ||
| 73 | |||
| 74 | #ifdef __KERNEL__ | ||
| 75 | |||
| 76 | #define PAGE_BITS (PAGE_SIZE << 3) | ||
| 77 | #define PAGE_BIT_SHIFT (PAGE_SHIFT + 3) | ||
| 78 | |||
| 79 | typedef __u16 bitmap_counter_t; | ||
| 80 | #define COUNTER_BITS 16 | ||
| 81 | #define COUNTER_BIT_SHIFT 4 | ||
| 82 | #define COUNTER_BYTE_RATIO (COUNTER_BITS / 8) | ||
| 83 | #define COUNTER_BYTE_SHIFT (COUNTER_BIT_SHIFT - 3) | ||
| 84 | |||
| 85 | #define NEEDED_MASK ((bitmap_counter_t) (1 << (COUNTER_BITS - 1))) | ||
| 86 | #define RESYNC_MASK ((bitmap_counter_t) (1 << (COUNTER_BITS - 2))) | ||
| 87 | #define COUNTER_MAX ((bitmap_counter_t) RESYNC_MASK - 1) | ||
| 88 | #define NEEDED(x) (((bitmap_counter_t) x) & NEEDED_MASK) | ||
| 89 | #define RESYNC(x) (((bitmap_counter_t) x) & RESYNC_MASK) | ||
| 90 | #define COUNTER(x) (((bitmap_counter_t) x) & COUNTER_MAX) | ||
| 91 | |||
| 92 | /* how many counters per page? */ | ||
| 93 | #define PAGE_COUNTER_RATIO (PAGE_BITS / COUNTER_BITS) | ||
| 94 | /* same, except a shift value for more efficient bitops */ | ||
| 95 | #define PAGE_COUNTER_SHIFT (PAGE_BIT_SHIFT - COUNTER_BIT_SHIFT) | ||
| 96 | /* same, except a mask value for more efficient bitops */ | ||
| 97 | #define PAGE_COUNTER_MASK (PAGE_COUNTER_RATIO - 1) | ||
| 98 | |||
| 99 | #define BITMAP_BLOCK_SIZE 512 | ||
| 100 | #define BITMAP_BLOCK_SHIFT 9 | ||
| 101 | |||
| 102 | /* how many blocks per chunk? (this is variable) */ | ||
| 103 | #define CHUNK_BLOCK_RATIO(bitmap) ((bitmap)->chunksize >> BITMAP_BLOCK_SHIFT) | ||
| 104 | #define CHUNK_BLOCK_SHIFT(bitmap) ((bitmap)->chunkshift - BITMAP_BLOCK_SHIFT) | ||
| 105 | #define CHUNK_BLOCK_MASK(bitmap) (CHUNK_BLOCK_RATIO(bitmap) - 1) | ||
| 106 | |||
| 107 | /* when hijacked, the counters and bits represent even larger "chunks" */ | ||
| 108 | /* there will be 1024 chunks represented by each counter in the page pointers */ | ||
| 109 | #define PAGEPTR_BLOCK_RATIO(bitmap) \ | ||
| 110 | (CHUNK_BLOCK_RATIO(bitmap) << PAGE_COUNTER_SHIFT >> 1) | ||
| 111 | #define PAGEPTR_BLOCK_SHIFT(bitmap) \ | ||
| 112 | (CHUNK_BLOCK_SHIFT(bitmap) + PAGE_COUNTER_SHIFT - 1) | ||
| 113 | #define PAGEPTR_BLOCK_MASK(bitmap) (PAGEPTR_BLOCK_RATIO(bitmap) - 1) | ||
| 114 | |||
| 115 | /* | ||
| 116 | * on-disk bitmap: | ||
| 117 | * | ||
| 118 | * Use one bit per "chunk" (block set). We do the disk I/O on the bitmap | ||
| 119 | * file a page at a time. There's a superblock at the start of the file. | ||
| 120 | */ | ||
| 121 | |||
| 122 | /* map chunks (bits) to file pages - offset by the size of the superblock */ | ||
| 123 | #define CHUNK_BIT_OFFSET(chunk) ((chunk) + (sizeof(bitmap_super_t) << 3)) | ||
| 124 | |||
| 125 | #endif | ||
| 126 | |||
| 127 | /* | ||
| 128 | * bitmap structures: | ||
| 129 | */ | ||
| 130 | |||
| 131 | #define BITMAP_MAGIC 0x6d746962 | ||
| 132 | |||
| 133 | /* use these for bitmap->flags and bitmap->sb->state bit-fields */ | ||
| 134 | enum bitmap_state { | ||
| 135 | BITMAP_ACTIVE = 0x001, /* the bitmap is in use */ | ||
| 136 | BITMAP_STALE = 0x002 /* the bitmap file is out of date or had -EIO */ | ||
| 137 | }; | ||
| 138 | |||
| 139 | /* the superblock at the front of the bitmap file -- little endian */ | ||
| 140 | typedef struct bitmap_super_s { | ||
| 141 | __u32 magic; /* 0 BITMAP_MAGIC */ | ||
| 142 | __u32 version; /* 4 the bitmap major for now, could change... */ | ||
| 143 | __u8 uuid[16]; /* 8 128 bit uuid - must match md device uuid */ | ||
| 144 | __u64 events; /* 24 event counter for the bitmap (1)*/ | ||
| 145 | __u64 events_cleared;/*32 event counter when last bit cleared (2) */ | ||
| 146 | __u64 sync_size; /* 40 the size of the md device's sync range(3) */ | ||
| 147 | __u32 state; /* 48 bitmap state information */ | ||
| 148 | __u32 chunksize; /* 52 the bitmap chunk size in bytes */ | ||
| 149 | __u32 daemon_sleep; /* 56 seconds between disk flushes */ | ||
| 150 | |||
| 151 | __u8 pad[256 - 60]; /* set to zero */ | ||
| 152 | } bitmap_super_t; | ||
| 153 | |||
| 154 | /* notes: | ||
| 155 | * (1) This event counter is updated before the eventcounter in the md superblock | ||
| 156 | * When a bitmap is loaded, it is only accepted if this event counter is equal | ||
| 157 | * to, or one greater than, the event counter in the superblock. | ||
| 158 | * (2) This event counter is updated when the other one is *if*and*only*if* the | ||
| 159 | * array is not degraded. As bits are not cleared when the array is degraded, | ||
| 160 | * this represents the last time that any bits were cleared. | ||
| 161 | * If a device is being added that has an event count with this value or | ||
| 162 | * higher, it is accepted as conforming to the bitmap. | ||
| 163 | * (3)This is the number of sectors represented by the bitmap, and is the range that | ||
| 164 | * resync happens across. For raid1 and raid5/6 it is the size of individual | ||
| 165 | * devices. For raid10 it is the size of the array. | ||
| 166 | */ | ||
| 167 | |||
| 168 | #ifdef __KERNEL__ | ||
| 169 | |||
| 170 | /* the in-memory bitmap is represented by bitmap_pages */ | ||
| 171 | struct bitmap_page { | ||
| 172 | /* | ||
| 173 | * map points to the actual memory page | ||
| 174 | */ | ||
| 175 | char *map; | ||
| 176 | /* | ||
| 177 | * in emergencies (when map cannot be alloced), hijack the map | ||
| 178 | * pointer and use it as two counters itself | ||
| 179 | */ | ||
| 180 | unsigned int hijacked:1; | ||
| 181 | /* | ||
| 182 | * count of dirty bits on the page | ||
| 183 | */ | ||
| 184 | unsigned int count:31; | ||
| 185 | }; | ||
| 186 | |||
| 187 | /* keep track of bitmap file pages that have pending writes on them */ | ||
| 188 | struct page_list { | ||
| 189 | struct list_head list; | ||
| 190 | struct page *page; | ||
| 191 | }; | ||
| 192 | |||
| 193 | /* the main bitmap structure - one per mddev */ | ||
| 194 | struct bitmap { | ||
| 195 | struct bitmap_page *bp; | ||
| 196 | unsigned long pages; /* total number of pages in the bitmap */ | ||
| 197 | unsigned long missing_pages; /* number of pages not yet allocated */ | ||
| 198 | |||
| 199 | mddev_t *mddev; /* the md device that the bitmap is for */ | ||
| 200 | |||
| 201 | int counter_bits; /* how many bits per block counter */ | ||
| 202 | |||
| 203 | /* bitmap chunksize -- how much data does each bit represent? */ | ||
| 204 | unsigned long chunksize; | ||
| 205 | unsigned long chunkshift; /* chunksize = 2^chunkshift (for bitops) */ | ||
| 206 | unsigned long chunks; /* total number of data chunks for the array */ | ||
| 207 | |||
| 208 | /* We hold a count on the chunk currently being synced, and drop | ||
| 209 | * it when the last block is started. If the resync is aborted | ||
| 210 | * midway, we need to be able to drop that count, so we remember | ||
| 211 | * the counted chunk.. | ||
| 212 | */ | ||
| 213 | unsigned long syncchunk; | ||
| 214 | |||
| 215 | __u64 events_cleared; | ||
| 216 | |||
| 217 | /* bitmap spinlock */ | ||
| 218 | spinlock_t lock; | ||
| 219 | |||
| 220 | long offset; /* offset from superblock if file is NULL */ | ||
| 221 | struct file *file; /* backing disk file */ | ||
| 222 | struct page *sb_page; /* cached copy of the bitmap file superblock */ | ||
| 223 | struct page **filemap; /* list of cache pages for the file */ | ||
| 224 | unsigned long *filemap_attr; /* attributes associated w/ filemap pages */ | ||
| 225 | unsigned long file_pages; /* number of pages in the file */ | ||
| 226 | |||
| 227 | unsigned long flags; | ||
| 228 | |||
| 229 | /* | ||
| 230 | * the bitmap daemon - periodically wakes up and sweeps the bitmap | ||
| 231 | * file, cleaning up bits and flushing out pages to disk as necessary | ||
| 232 | */ | ||
| 233 | unsigned long daemon_lastrun; /* jiffies of last run */ | ||
| 234 | unsigned long daemon_sleep; /* how many seconds between updates? */ | ||
| 235 | |||
| 236 | /* | ||
| 237 | * bitmap_writeback_daemon waits for file-pages that have been written, | ||
| 238 | * as there is no way to get a call-back when a page write completes. | ||
| 239 | */ | ||
| 240 | mdk_thread_t *writeback_daemon; | ||
| 241 | spinlock_t write_lock; | ||
| 242 | wait_queue_head_t write_wait; | ||
| 243 | struct list_head complete_pages; | ||
| 244 | mempool_t *write_pool; | ||
| 245 | }; | ||
| 246 | |||
| 247 | /* the bitmap API */ | ||
| 248 | |||
| 249 | /* these are used only by md/bitmap */ | ||
| 250 | int bitmap_create(mddev_t *mddev); | ||
| 251 | void bitmap_destroy(mddev_t *mddev); | ||
| 252 | int bitmap_active(struct bitmap *bitmap); | ||
| 253 | |||
| 254 | char *file_path(struct file *file, char *buf, int count); | ||
| 255 | void bitmap_print_sb(struct bitmap *bitmap); | ||
| 256 | int bitmap_update_sb(struct bitmap *bitmap); | ||
| 257 | |||
| 258 | int bitmap_setallbits(struct bitmap *bitmap); | ||
| 259 | void bitmap_write_all(struct bitmap *bitmap); | ||
| 260 | |||
| 261 | /* these are exported */ | ||
| 262 | int bitmap_startwrite(struct bitmap *bitmap, sector_t offset, unsigned long sectors); | ||
| 263 | void bitmap_endwrite(struct bitmap *bitmap, sector_t offset, unsigned long sectors, | ||
| 264 | int success); | ||
| 265 | int bitmap_start_sync(struct bitmap *bitmap, sector_t offset, int *blocks); | ||
| 266 | void bitmap_end_sync(struct bitmap *bitmap, sector_t offset, int *blocks, int aborted); | ||
| 267 | void bitmap_close_sync(struct bitmap *bitmap); | ||
| 268 | |||
| 269 | int bitmap_unplug(struct bitmap *bitmap); | ||
| 270 | int bitmap_daemon_work(struct bitmap *bitmap); | ||
| 271 | #endif | ||
| 272 | |||
| 273 | #endif | ||
diff --git a/include/linux/raid/md.h b/include/linux/raid/md.h index a6a67d102b..ffa316ce4d 100644 --- a/include/linux/raid/md.h +++ b/include/linux/raid/md.h | |||
| @@ -60,7 +60,14 @@ | |||
| 60 | */ | 60 | */ |
| 61 | #define MD_MAJOR_VERSION 0 | 61 | #define MD_MAJOR_VERSION 0 |
| 62 | #define MD_MINOR_VERSION 90 | 62 | #define MD_MINOR_VERSION 90 |
| 63 | #define MD_PATCHLEVEL_VERSION 1 | 63 | /* |
| 64 | * MD_PATCHLEVEL_VERSION indicates kernel functionality. | ||
| 65 | * >=1 means different superblock formats are selectable using SET_ARRAY_INFO | ||
| 66 | * and major_version/minor_version accordingly | ||
| 67 | * >=2 means that Internal bitmaps are supported by setting MD_SB_BITMAP_PRESENT | ||
| 68 | * in the super status byte | ||
| 69 | */ | ||
| 70 | #define MD_PATCHLEVEL_VERSION 2 | ||
| 64 | 71 | ||
| 65 | extern int register_md_personality (int p_num, mdk_personality_t *p); | 72 | extern int register_md_personality (int p_num, mdk_personality_t *p); |
| 66 | extern int unregister_md_personality (int p_num); | 73 | extern int unregister_md_personality (int p_num); |
| @@ -69,7 +76,7 @@ extern mdk_thread_t * md_register_thread (void (*run) (mddev_t *mddev), | |||
| 69 | extern void md_unregister_thread (mdk_thread_t *thread); | 76 | extern void md_unregister_thread (mdk_thread_t *thread); |
| 70 | extern void md_wakeup_thread(mdk_thread_t *thread); | 77 | extern void md_wakeup_thread(mdk_thread_t *thread); |
| 71 | extern void md_check_recovery(mddev_t *mddev); | 78 | extern void md_check_recovery(mddev_t *mddev); |
| 72 | extern void md_write_start(mddev_t *mddev); | 79 | extern void md_write_start(mddev_t *mddev, struct bio *bi); |
| 73 | extern void md_write_end(mddev_t *mddev); | 80 | extern void md_write_end(mddev_t *mddev); |
| 74 | extern void md_handle_safemode(mddev_t *mddev); | 81 | extern void md_handle_safemode(mddev_t *mddev); |
| 75 | extern void md_done_sync(mddev_t *mddev, int blocks, int ok); | 82 | extern void md_done_sync(mddev_t *mddev, int blocks, int ok); |
| @@ -78,6 +85,12 @@ extern void md_unplug_mddev(mddev_t *mddev); | |||
| 78 | 85 | ||
| 79 | extern void md_print_devices (void); | 86 | extern void md_print_devices (void); |
| 80 | 87 | ||
| 88 | extern void md_super_write(mddev_t *mddev, mdk_rdev_t *rdev, | ||
| 89 | sector_t sector, int size, struct page *page); | ||
| 90 | extern int sync_page_io(struct block_device *bdev, sector_t sector, int size, | ||
| 91 | struct page *page, int rw); | ||
| 92 | |||
| 93 | |||
| 81 | #define MD_BUG(x...) { printk("md: bug in file %s, line %d\n", __FILE__, __LINE__); md_print_devices(); } | 94 | #define MD_BUG(x...) { printk("md: bug in file %s, line %d\n", __FILE__, __LINE__); md_print_devices(); } |
| 82 | 95 | ||
| 83 | #endif | 96 | #endif |
diff --git a/include/linux/raid/md_k.h b/include/linux/raid/md_k.h index c9a0d4013b..8c14ba565a 100644 --- a/include/linux/raid/md_k.h +++ b/include/linux/raid/md_k.h | |||
| @@ -15,6 +15,9 @@ | |||
| 15 | #ifndef _MD_K_H | 15 | #ifndef _MD_K_H |
| 16 | #define _MD_K_H | 16 | #define _MD_K_H |
| 17 | 17 | ||
| 18 | /* and dm-bio-list.h is not under include/linux because.... ??? */ | ||
| 19 | #include "../../../drivers/md/dm-bio-list.h" | ||
| 20 | |||
| 18 | #define MD_RESERVED 0UL | 21 | #define MD_RESERVED 0UL |
| 19 | #define LINEAR 1UL | 22 | #define LINEAR 1UL |
| 20 | #define RAID0 2UL | 23 | #define RAID0 2UL |
| @@ -180,6 +183,10 @@ struct mdk_rdev_s | |||
| 180 | 183 | ||
| 181 | int desc_nr; /* descriptor index in the superblock */ | 184 | int desc_nr; /* descriptor index in the superblock */ |
| 182 | int raid_disk; /* role of device in array */ | 185 | int raid_disk; /* role of device in array */ |
| 186 | int saved_raid_disk; /* role that device used to have in the | ||
| 187 | * array and could again if we did a partial | ||
| 188 | * resync from the bitmap | ||
| 189 | */ | ||
| 183 | 190 | ||
| 184 | atomic_t nr_pending; /* number of pending requests. | 191 | atomic_t nr_pending; /* number of pending requests. |
| 185 | * only maintained for arrays that | 192 | * only maintained for arrays that |
| @@ -252,6 +259,11 @@ struct mddev_s | |||
| 252 | atomic_t recovery_active; /* blocks scheduled, but not written */ | 259 | atomic_t recovery_active; /* blocks scheduled, but not written */ |
| 253 | wait_queue_head_t recovery_wait; | 260 | wait_queue_head_t recovery_wait; |
| 254 | sector_t recovery_cp; | 261 | sector_t recovery_cp; |
| 262 | |||
| 263 | spinlock_t write_lock; | ||
| 264 | wait_queue_head_t sb_wait; /* for waiting on superblock updates */ | ||
| 265 | atomic_t pending_writes; /* number of active superblock writes */ | ||
| 266 | |||
| 255 | unsigned int safemode; /* if set, update "clean" superblock | 267 | unsigned int safemode; /* if set, update "clean" superblock |
| 256 | * when no writes pending. | 268 | * when no writes pending. |
| 257 | */ | 269 | */ |
| @@ -260,6 +272,13 @@ struct mddev_s | |||
| 260 | atomic_t writes_pending; | 272 | atomic_t writes_pending; |
| 261 | request_queue_t *queue; /* for plugging ... */ | 273 | request_queue_t *queue; /* for plugging ... */ |
| 262 | 274 | ||
| 275 | struct bitmap *bitmap; /* the bitmap for the device */ | ||
| 276 | struct file *bitmap_file; /* the bitmap file */ | ||
| 277 | long bitmap_offset; /* offset from superblock of | ||
| 278 | * start of bitmap. May be | ||
| 279 | * negative, but not '0' | ||
| 280 | */ | ||
| 281 | |||
| 263 | struct list_head all_mddevs; | 282 | struct list_head all_mddevs; |
| 264 | }; | 283 | }; |
| 265 | 284 | ||
| @@ -291,7 +310,7 @@ struct mdk_personality_s | |||
| 291 | int (*hot_add_disk) (mddev_t *mddev, mdk_rdev_t *rdev); | 310 | int (*hot_add_disk) (mddev_t *mddev, mdk_rdev_t *rdev); |
| 292 | int (*hot_remove_disk) (mddev_t *mddev, int number); | 311 | int (*hot_remove_disk) (mddev_t *mddev, int number); |
| 293 | int (*spare_active) (mddev_t *mddev); | 312 | int (*spare_active) (mddev_t *mddev); |
| 294 | int (*sync_request)(mddev_t *mddev, sector_t sector_nr, int go_faster); | 313 | sector_t (*sync_request)(mddev_t *mddev, sector_t sector_nr, int *skipped, int go_faster); |
| 295 | int (*resize) (mddev_t *mddev, sector_t sectors); | 314 | int (*resize) (mddev_t *mddev, sector_t sectors); |
| 296 | int (*reshape) (mddev_t *mddev, int raid_disks); | 315 | int (*reshape) (mddev_t *mddev, int raid_disks); |
| 297 | int (*reconfig) (mddev_t *mddev, int layout, int chunk_size); | 316 | int (*reconfig) (mddev_t *mddev, int layout, int chunk_size); |
| @@ -334,6 +353,7 @@ typedef struct mdk_thread_s { | |||
| 334 | unsigned long flags; | 353 | unsigned long flags; |
| 335 | struct completion *event; | 354 | struct completion *event; |
| 336 | struct task_struct *tsk; | 355 | struct task_struct *tsk; |
| 356 | unsigned long timeout; | ||
| 337 | const char *name; | 357 | const char *name; |
| 338 | } mdk_thread_t; | 358 | } mdk_thread_t; |
| 339 | 359 | ||
diff --git a/include/linux/raid/md_p.h b/include/linux/raid/md_p.h index 8ba95d6732..dc65cd4354 100644 --- a/include/linux/raid/md_p.h +++ b/include/linux/raid/md_p.h | |||
| @@ -96,6 +96,7 @@ typedef struct mdp_device_descriptor_s { | |||
| 96 | #define MD_SB_CLEAN 0 | 96 | #define MD_SB_CLEAN 0 |
| 97 | #define MD_SB_ERRORS 1 | 97 | #define MD_SB_ERRORS 1 |
| 98 | 98 | ||
| 99 | #define MD_SB_BITMAP_PRESENT 8 /* bitmap may be present nearby */ | ||
| 99 | typedef struct mdp_superblock_s { | 100 | typedef struct mdp_superblock_s { |
| 100 | /* | 101 | /* |
| 101 | * Constant generic information | 102 | * Constant generic information |
| @@ -184,7 +185,7 @@ struct mdp_superblock_1 { | |||
| 184 | /* constant array information - 128 bytes */ | 185 | /* constant array information - 128 bytes */ |
| 185 | __u32 magic; /* MD_SB_MAGIC: 0xa92b4efc - little endian */ | 186 | __u32 magic; /* MD_SB_MAGIC: 0xa92b4efc - little endian */ |
| 186 | __u32 major_version; /* 1 */ | 187 | __u32 major_version; /* 1 */ |
| 187 | __u32 feature_map; /* 0 for now */ | 188 | __u32 feature_map; /* bit 0 set if 'bitmap_offset' is meaningful */ |
| 188 | __u32 pad0; /* always set to 0 when writing */ | 189 | __u32 pad0; /* always set to 0 when writing */ |
| 189 | 190 | ||
| 190 | __u8 set_uuid[16]; /* user-space generated. */ | 191 | __u8 set_uuid[16]; /* user-space generated. */ |
| @@ -197,7 +198,11 @@ struct mdp_superblock_1 { | |||
| 197 | 198 | ||
| 198 | __u32 chunksize; /* in 512byte sectors */ | 199 | __u32 chunksize; /* in 512byte sectors */ |
| 199 | __u32 raid_disks; | 200 | __u32 raid_disks; |
| 200 | __u8 pad1[128-96]; /* set to 0 when written */ | 201 | __u32 bitmap_offset; /* sectors after start of superblock that bitmap starts |
| 202 | * NOTE: signed, so bitmap can be before superblock | ||
| 203 | * only meaningful of feature_map[0] is set. | ||
| 204 | */ | ||
| 205 | __u8 pad1[128-100]; /* set to 0 when written */ | ||
| 201 | 206 | ||
| 202 | /* constant this-device information - 64 bytes */ | 207 | /* constant this-device information - 64 bytes */ |
| 203 | __u64 data_offset; /* sector start of data, often 0 */ | 208 | __u64 data_offset; /* sector start of data, often 0 */ |
diff --git a/include/linux/raid/md_u.h b/include/linux/raid/md_u.h index a2df5c2a42..81da20ccec 100644 --- a/include/linux/raid/md_u.h +++ b/include/linux/raid/md_u.h | |||
| @@ -23,6 +23,7 @@ | |||
| 23 | #define GET_DISK_INFO _IOR (MD_MAJOR, 0x12, mdu_disk_info_t) | 23 | #define GET_DISK_INFO _IOR (MD_MAJOR, 0x12, mdu_disk_info_t) |
| 24 | #define PRINT_RAID_DEBUG _IO (MD_MAJOR, 0x13) | 24 | #define PRINT_RAID_DEBUG _IO (MD_MAJOR, 0x13) |
| 25 | #define RAID_AUTORUN _IO (MD_MAJOR, 0x14) | 25 | #define RAID_AUTORUN _IO (MD_MAJOR, 0x14) |
| 26 | #define GET_BITMAP_FILE _IOR (MD_MAJOR, 0x15, mdu_bitmap_file_t) | ||
| 26 | 27 | ||
| 27 | /* configuration */ | 28 | /* configuration */ |
| 28 | #define CLEAR_ARRAY _IO (MD_MAJOR, 0x20) | 29 | #define CLEAR_ARRAY _IO (MD_MAJOR, 0x20) |
| @@ -36,6 +37,7 @@ | |||
| 36 | #define HOT_ADD_DISK _IO (MD_MAJOR, 0x28) | 37 | #define HOT_ADD_DISK _IO (MD_MAJOR, 0x28) |
| 37 | #define SET_DISK_FAULTY _IO (MD_MAJOR, 0x29) | 38 | #define SET_DISK_FAULTY _IO (MD_MAJOR, 0x29) |
| 38 | #define HOT_GENERATE_ERROR _IO (MD_MAJOR, 0x2a) | 39 | #define HOT_GENERATE_ERROR _IO (MD_MAJOR, 0x2a) |
| 40 | #define SET_BITMAP_FILE _IOW (MD_MAJOR, 0x2b, int) | ||
| 39 | 41 | ||
| 40 | /* usage */ | 42 | /* usage */ |
| 41 | #define RUN_ARRAY _IOW (MD_MAJOR, 0x30, mdu_param_t) | 43 | #define RUN_ARRAY _IOW (MD_MAJOR, 0x30, mdu_param_t) |
| @@ -106,6 +108,11 @@ typedef struct mdu_start_info_s { | |||
| 106 | 108 | ||
| 107 | } mdu_start_info_t; | 109 | } mdu_start_info_t; |
| 108 | 110 | ||
| 111 | typedef struct mdu_bitmap_file_s | ||
| 112 | { | ||
| 113 | char pathname[4096]; | ||
| 114 | } mdu_bitmap_file_t; | ||
| 115 | |||
| 109 | typedef struct mdu_param_s | 116 | typedef struct mdu_param_s |
| 110 | { | 117 | { |
| 111 | int personality; /* 1,2,3,4 */ | 118 | int personality; /* 1,2,3,4 */ |
diff --git a/include/linux/raid/raid1.h b/include/linux/raid/raid1.h index abbfdd9afe..9d93cf12e8 100644 --- a/include/linux/raid/raid1.h +++ b/include/linux/raid/raid1.h | |||
| @@ -36,12 +36,21 @@ struct r1_private_data_s { | |||
| 36 | spinlock_t device_lock; | 36 | spinlock_t device_lock; |
| 37 | 37 | ||
| 38 | struct list_head retry_list; | 38 | struct list_head retry_list; |
| 39 | /* queue pending writes and submit them on unplug */ | ||
| 40 | struct bio_list pending_bio_list; | ||
| 41 | /* queue of writes that have been unplugged */ | ||
| 42 | struct bio_list flushing_bio_list; | ||
| 43 | |||
| 39 | /* for use when syncing mirrors: */ | 44 | /* for use when syncing mirrors: */ |
| 40 | 45 | ||
| 41 | spinlock_t resync_lock; | 46 | spinlock_t resync_lock; |
| 42 | int nr_pending; | 47 | int nr_pending; |
| 43 | int barrier; | 48 | int barrier; |
| 44 | sector_t next_resync; | 49 | sector_t next_resync; |
| 50 | int fullsync; /* set to 1 if a full sync is needed, | ||
| 51 | * (fresh device added). | ||
| 52 | * Cleared when a sync completes. | ||
| 53 | */ | ||
| 45 | 54 | ||
| 46 | wait_queue_head_t wait_idle; | 55 | wait_queue_head_t wait_idle; |
| 47 | wait_queue_head_t wait_resume; | 56 | wait_queue_head_t wait_resume; |
| @@ -85,14 +94,17 @@ struct r1bio_s { | |||
| 85 | int read_disk; | 94 | int read_disk; |
| 86 | 95 | ||
| 87 | struct list_head retry_list; | 96 | struct list_head retry_list; |
| 97 | struct bitmap_update *bitmap_update; | ||
| 88 | /* | 98 | /* |
| 89 | * if the IO is in WRITE direction, then multiple bios are used. | 99 | * if the IO is in WRITE direction, then multiple bios are used. |
| 90 | * We choose the number when they are allocated. | 100 | * We choose the number when they are allocated. |
| 91 | */ | 101 | */ |
| 92 | struct bio *bios[0]; | 102 | struct bio *bios[0]; |
| 103 | /* DO NOT PUT ANY NEW FIELDS HERE - bios array is contiguously alloced*/ | ||
| 93 | }; | 104 | }; |
| 94 | 105 | ||
| 95 | /* bits for r1bio.state */ | 106 | /* bits for r1bio.state */ |
| 96 | #define R1BIO_Uptodate 0 | 107 | #define R1BIO_Uptodate 0 |
| 97 | #define R1BIO_IsSync 1 | 108 | #define R1BIO_IsSync 1 |
| 109 | #define R1BIO_Degraded 2 | ||
| 98 | #endif | 110 | #endif |
