1 /* 2 * Copyright (C) 2007 Oracle. All rights reserved. 3 * 4 * This program is free software; you can redistribute it and/or 5 * modify it under the terms of the GNU General Public 6 * License v2 as published by the Free Software Foundation. 7 * 8 * This program is distributed in the hope that it will be useful, 9 * but WITHOUT ANY WARRANTY; without even the implied warranty of 10 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU 11 * General Public License for more details. 12 * 13 * You should have received a copy of the GNU General Public 14 * License along with this program; if not, write to the 15 * Free Software Foundation, Inc., 59 Temple Place - Suite 330, 16 * Boston, MA 021110-1307, USA. 17 */ 18 19 #ifndef __BTRFS_VOLUMES_ 20 #define __BTRFS_VOLUMES_ 21 22 #include <linux/bio.h> 23 #include <linux/sort.h> 24 #include <linux/btrfs.h> 25 #include "async-thread.h" 26 27 #define BTRFS_STRIPE_LEN (64 * 1024) 28 29 struct buffer_head; 30 struct btrfs_pending_bios { 31 struct bio *head; 32 struct bio *tail; 33 }; 34 35 struct btrfs_device { 36 struct list_head dev_list; 37 struct list_head dev_alloc_list; 38 struct btrfs_fs_devices *fs_devices; 39 struct btrfs_root *dev_root; 40 41 /* regular prio bios */ 42 struct btrfs_pending_bios pending_bios; 43 /* WRITE_SYNC bios */ 44 struct btrfs_pending_bios pending_sync_bios; 45 46 u64 generation; 47 int running_pending; 48 int writeable; 49 int in_fs_metadata; 50 int missing; 51 int can_discard; 52 int is_tgtdev_for_dev_replace; 53 54 spinlock_t io_lock; 55 /* the mode sent to blkdev_get */ 56 fmode_t mode; 57 58 struct block_device *bdev; 59 60 61 struct rcu_string *name; 62 63 /* the internal btrfs device id */ 64 u64 devid; 65 66 /* size of the device */ 67 u64 total_bytes; 68 69 /* size of the disk */ 70 u64 disk_total_bytes; 71 72 /* bytes used */ 73 u64 bytes_used; 74 75 /* optimal io alignment for this device */ 76 u32 io_align; 77 78 /* optimal io width for this device */ 79 u32 io_width; 80 /* type and info about this device */ 81 u64 type; 82 83 /* minimal io size for this device */ 84 u32 sector_size; 85 86 87 /* physical drive uuid (or lvm uuid) */ 88 u8 uuid[BTRFS_UUID_SIZE]; 89 90 /* for sending down flush barriers */ 91 int nobarriers; 92 struct bio *flush_bio; 93 struct completion flush_wait; 94 95 /* per-device scrub information */ 96 struct scrub_ctx *scrub_device; 97 98 struct btrfs_work work; 99 struct rcu_head rcu; 100 struct work_struct rcu_work; 101 102 /* readahead state */ 103 spinlock_t reada_lock; 104 atomic_t reada_in_flight; 105 u64 reada_next; 106 struct reada_zone *reada_curr_zone; 107 struct radix_tree_root reada_zones; 108 struct radix_tree_root reada_extents; 109 110 111 /* disk I/O failure stats. For detailed description refer to 112 * enum btrfs_dev_stat_values in ioctl.h */ 113 int dev_stats_valid; 114 int dev_stats_dirty; /* counters need to be written to disk */ 115 atomic_t dev_stat_values[BTRFS_DEV_STAT_VALUES_MAX]; 116 }; 117 118 struct btrfs_fs_devices { 119 u8 fsid[BTRFS_FSID_SIZE]; /* FS specific uuid */ 120 121 /* the device with this id has the most recent copy of the super */ 122 u64 latest_devid; 123 u64 latest_trans; 124 u64 num_devices; 125 u64 open_devices; 126 u64 rw_devices; 127 u64 missing_devices; 128 u64 total_rw_bytes; 129 u64 num_can_discard; 130 u64 total_devices; 131 struct block_device *latest_bdev; 132 133 /* all of the devices in the FS, protected by a mutex 134 * so we can safely walk it to write out the supers without 135 * worrying about add/remove by the multi-device code. 136 * Scrubbing super can kick off supers writing by holding 137 * this mutex lock. 138 */ 139 struct mutex device_list_mutex; 140 struct list_head devices; 141 142 /* devices not currently being allocated */ 143 struct list_head alloc_list; 144 struct list_head list; 145 146 struct btrfs_fs_devices *seed; 147 int seeding; 148 149 int opened; 150 151 /* set when we find or add a device that doesn't have the 152 * nonrot flag set 153 */ 154 int rotating; 155 }; 156 157 #define BTRFS_BIO_INLINE_CSUM_SIZE 64 158 159 /* 160 * we need the mirror number and stripe index to be passed around 161 * the call chain while we are processing end_io (especially errors). 162 * Really, what we need is a btrfs_bio structure that has this info 163 * and is properly sized with its stripe array, but we're not there 164 * quite yet. We have our own btrfs bioset, and all of the bios 165 * we allocate are actually btrfs_io_bios. We'll cram as much of 166 * struct btrfs_bio as we can into this over time. 167 */ 168 typedef void (btrfs_io_bio_end_io_t) (struct btrfs_io_bio *bio, int err); 169 struct btrfs_io_bio { 170 unsigned long mirror_num; 171 unsigned long stripe_index; 172 u8 *csum; 173 u8 csum_inline[BTRFS_BIO_INLINE_CSUM_SIZE]; 174 u8 *csum_allocated; 175 btrfs_io_bio_end_io_t *end_io; 176 struct bio bio; 177 }; 178 179 static inline struct btrfs_io_bio *btrfs_io_bio(struct bio *bio) 180 { 181 return container_of(bio, struct btrfs_io_bio, bio); 182 } 183 184 struct btrfs_bio_stripe { 185 struct btrfs_device *dev; 186 u64 physical; 187 u64 length; /* only used for discard mappings */ 188 }; 189 190 struct btrfs_bio; 191 typedef void (btrfs_bio_end_io_t) (struct btrfs_bio *bio, int err); 192 193 struct btrfs_bio { 194 atomic_t stripes_pending; 195 struct btrfs_fs_info *fs_info; 196 bio_end_io_t *end_io; 197 struct bio *orig_bio; 198 void *private; 199 atomic_t error; 200 int max_errors; 201 int num_stripes; 202 int mirror_num; 203 struct btrfs_bio_stripe stripes[]; 204 }; 205 206 struct btrfs_device_info { 207 struct btrfs_device *dev; 208 u64 dev_offset; 209 u64 max_avail; 210 u64 total_avail; 211 }; 212 213 struct btrfs_raid_attr { 214 int sub_stripes; /* sub_stripes info for map */ 215 int dev_stripes; /* stripes per dev */ 216 int devs_max; /* max devs to use */ 217 int devs_min; /* min devs needed */ 218 int devs_increment; /* ndevs has to be a multiple of this */ 219 int ncopies; /* how many copies to data has */ 220 }; 221 222 struct map_lookup { 223 u64 type; 224 int io_align; 225 int io_width; 226 int stripe_len; 227 int sector_size; 228 int num_stripes; 229 int sub_stripes; 230 struct btrfs_bio_stripe stripes[]; 231 }; 232 233 #define map_lookup_size(n) (sizeof(struct map_lookup) + \ 234 (sizeof(struct btrfs_bio_stripe) * (n))) 235 236 /* 237 * Restriper's general type filter 238 */ 239 #define BTRFS_BALANCE_DATA (1ULL << 0) 240 #define BTRFS_BALANCE_SYSTEM (1ULL << 1) 241 #define BTRFS_BALANCE_METADATA (1ULL << 2) 242 243 #define BTRFS_BALANCE_TYPE_MASK (BTRFS_BALANCE_DATA | \ 244 BTRFS_BALANCE_SYSTEM | \ 245 BTRFS_BALANCE_METADATA) 246 247 #define BTRFS_BALANCE_FORCE (1ULL << 3) 248 #define BTRFS_BALANCE_RESUME (1ULL << 4) 249 250 /* 251 * Balance filters 252 */ 253 #define BTRFS_BALANCE_ARGS_PROFILES (1ULL << 0) 254 #define BTRFS_BALANCE_ARGS_USAGE (1ULL << 1) 255 #define BTRFS_BALANCE_ARGS_DEVID (1ULL << 2) 256 #define BTRFS_BALANCE_ARGS_DRANGE (1ULL << 3) 257 #define BTRFS_BALANCE_ARGS_VRANGE (1ULL << 4) 258 #define BTRFS_BALANCE_ARGS_LIMIT (1ULL << 5) 259 260 /* 261 * Profile changing flags. When SOFT is set we won't relocate chunk if 262 * it already has the target profile (even though it may be 263 * half-filled). 264 */ 265 #define BTRFS_BALANCE_ARGS_CONVERT (1ULL << 8) 266 #define BTRFS_BALANCE_ARGS_SOFT (1ULL << 9) 267 268 struct btrfs_balance_args; 269 struct btrfs_balance_progress; 270 struct btrfs_balance_control { 271 struct btrfs_fs_info *fs_info; 272 273 struct btrfs_balance_args data; 274 struct btrfs_balance_args meta; 275 struct btrfs_balance_args sys; 276 277 u64 flags; 278 279 struct btrfs_balance_progress stat; 280 }; 281 282 int btrfs_account_dev_extents_size(struct btrfs_device *device, u64 start, 283 u64 end, u64 *length); 284 285 #define btrfs_bio_size(n) (sizeof(struct btrfs_bio) + \ 286 (sizeof(struct btrfs_bio_stripe) * (n))) 287 288 int btrfs_map_block(struct btrfs_fs_info *fs_info, int rw, 289 u64 logical, u64 *length, 290 struct btrfs_bio **bbio_ret, int mirror_num); 291 int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree, 292 u64 chunk_start, u64 physical, u64 devid, 293 u64 **logical, int *naddrs, int *stripe_len); 294 int btrfs_read_sys_array(struct btrfs_root *root); 295 int btrfs_read_chunk_tree(struct btrfs_root *root); 296 int btrfs_alloc_chunk(struct btrfs_trans_handle *trans, 297 struct btrfs_root *extent_root, u64 type); 298 void btrfs_mapping_init(struct btrfs_mapping_tree *tree); 299 void btrfs_mapping_tree_free(struct btrfs_mapping_tree *tree); 300 int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio, 301 int mirror_num, int async_submit); 302 int btrfs_open_devices(struct btrfs_fs_devices *fs_devices, 303 fmode_t flags, void *holder); 304 int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder, 305 struct btrfs_fs_devices **fs_devices_ret); 306 int btrfs_close_devices(struct btrfs_fs_devices *fs_devices); 307 void btrfs_close_extra_devices(struct btrfs_fs_info *fs_info, 308 struct btrfs_fs_devices *fs_devices, int step); 309 int btrfs_find_device_missing_or_by_path(struct btrfs_root *root, 310 char *device_path, 311 struct btrfs_device **device); 312 struct btrfs_device *btrfs_alloc_device(struct btrfs_fs_info *fs_info, 313 const u64 *devid, 314 const u8 *uuid); 315 int btrfs_rm_device(struct btrfs_root *root, char *device_path); 316 void btrfs_cleanup_fs_uuids(void); 317 int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len); 318 int btrfs_grow_device(struct btrfs_trans_handle *trans, 319 struct btrfs_device *device, u64 new_size); 320 struct btrfs_device *btrfs_find_device(struct btrfs_fs_info *fs_info, u64 devid, 321 u8 *uuid, u8 *fsid); 322 int btrfs_shrink_device(struct btrfs_device *device, u64 new_size); 323 int btrfs_init_new_device(struct btrfs_root *root, char *path); 324 int btrfs_init_dev_replace_tgtdev(struct btrfs_root *root, char *device_path, 325 struct btrfs_device **device_out); 326 int btrfs_balance(struct btrfs_balance_control *bctl, 327 struct btrfs_ioctl_balance_args *bargs); 328 int btrfs_resume_balance_async(struct btrfs_fs_info *fs_info); 329 int btrfs_recover_balance(struct btrfs_fs_info *fs_info); 330 int btrfs_pause_balance(struct btrfs_fs_info *fs_info); 331 int btrfs_cancel_balance(struct btrfs_fs_info *fs_info); 332 int btrfs_create_uuid_tree(struct btrfs_fs_info *fs_info); 333 int btrfs_check_uuid_tree(struct btrfs_fs_info *fs_info); 334 int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset); 335 int find_free_dev_extent(struct btrfs_trans_handle *trans, 336 struct btrfs_device *device, u64 num_bytes, 337 u64 *start, u64 *max_avail); 338 void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index); 339 int btrfs_get_dev_stats(struct btrfs_root *root, 340 struct btrfs_ioctl_get_dev_stats *stats); 341 void btrfs_init_devices_late(struct btrfs_fs_info *fs_info); 342 int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info); 343 int btrfs_run_dev_stats(struct btrfs_trans_handle *trans, 344 struct btrfs_fs_info *fs_info); 345 void btrfs_rm_dev_replace_srcdev(struct btrfs_fs_info *fs_info, 346 struct btrfs_device *srcdev); 347 void btrfs_destroy_dev_replace_tgtdev(struct btrfs_fs_info *fs_info, 348 struct btrfs_device *tgtdev); 349 void btrfs_init_dev_replace_tgtdev_for_resume(struct btrfs_fs_info *fs_info, 350 struct btrfs_device *tgtdev); 351 int btrfs_scratch_superblock(struct btrfs_device *device); 352 int btrfs_is_parity_mirror(struct btrfs_mapping_tree *map_tree, 353 u64 logical, u64 len, int mirror_num); 354 unsigned long btrfs_full_stripe_len(struct btrfs_root *root, 355 struct btrfs_mapping_tree *map_tree, 356 u64 logical); 357 int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans, 358 struct btrfs_root *extent_root, 359 u64 chunk_offset, u64 chunk_size); 360 static inline void btrfs_dev_stat_inc(struct btrfs_device *dev, 361 int index) 362 { 363 atomic_inc(dev->dev_stat_values + index); 364 dev->dev_stats_dirty = 1; 365 } 366 367 static inline int btrfs_dev_stat_read(struct btrfs_device *dev, 368 int index) 369 { 370 return atomic_read(dev->dev_stat_values + index); 371 } 372 373 static inline int btrfs_dev_stat_read_and_reset(struct btrfs_device *dev, 374 int index) 375 { 376 int ret; 377 378 ret = atomic_xchg(dev->dev_stat_values + index, 0); 379 dev->dev_stats_dirty = 1; 380 return ret; 381 } 382 383 static inline void btrfs_dev_stat_set(struct btrfs_device *dev, 384 int index, unsigned long val) 385 { 386 atomic_set(dev->dev_stat_values + index, val); 387 dev->dev_stats_dirty = 1; 388 } 389 390 static inline void btrfs_dev_stat_reset(struct btrfs_device *dev, 391 int index) 392 { 393 btrfs_dev_stat_set(dev, index, 0); 394 } 395 #endif 396