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 int running_pending; 47 u64 generation; 48 49 int writeable; 50 int in_fs_metadata; 51 int missing; 52 int can_discard; 53 int is_tgtdev_for_dev_replace; 54 55 spinlock_t io_lock; 56 57 struct block_device *bdev; 58 59 /* the mode sent to blkdev_get */ 60 fmode_t mode; 61 62 struct rcu_string *name; 63 64 /* the internal btrfs device id */ 65 u64 devid; 66 67 /* size of the device */ 68 u64 total_bytes; 69 70 /* size of the disk */ 71 u64 disk_total_bytes; 72 73 /* bytes used */ 74 u64 bytes_used; 75 76 /* optimal io alignment for this device */ 77 u32 io_align; 78 79 /* optimal io width for this device */ 80 u32 io_width; 81 82 /* minimal io size for this device */ 83 u32 sector_size; 84 85 /* type and info about this device */ 86 u64 type; 87 88 /* physical drive uuid (or lvm uuid) */ 89 u8 uuid[BTRFS_UUID_SIZE]; 90 91 /* per-device scrub information */ 92 struct scrub_ctx *scrub_device; 93 94 struct btrfs_work work; 95 struct rcu_head rcu; 96 struct work_struct rcu_work; 97 98 /* readahead state */ 99 spinlock_t reada_lock; 100 atomic_t reada_in_flight; 101 u64 reada_next; 102 struct reada_zone *reada_curr_zone; 103 struct radix_tree_root reada_zones; 104 struct radix_tree_root reada_extents; 105 106 /* for sending down flush barriers */ 107 struct bio *flush_bio; 108 struct completion flush_wait; 109 int nobarriers; 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 */ 137 struct mutex device_list_mutex; 138 struct list_head devices; 139 140 /* devices not currently being allocated */ 141 struct list_head alloc_list; 142 struct list_head list; 143 144 struct btrfs_fs_devices *seed; 145 int seeding; 146 147 int opened; 148 149 /* set when we find or add a device that doesn't have the 150 * nonrot flag set 151 */ 152 int rotating; 153 }; 154 155 /* 156 * we need the mirror number and stripe index to be passed around 157 * the call chain while we are processing end_io (especially errors). 158 * Really, what we need is a btrfs_bio structure that has this info 159 * and is properly sized with its stripe array, but we're not there 160 * quite yet. We have our own btrfs bioset, and all of the bios 161 * we allocate are actually btrfs_io_bios. We'll cram as much of 162 * struct btrfs_bio as we can into this over time. 163 */ 164 struct btrfs_io_bio { 165 unsigned long mirror_num; 166 unsigned long stripe_index; 167 struct bio bio; 168 }; 169 170 static inline struct btrfs_io_bio *btrfs_io_bio(struct bio *bio) 171 { 172 return container_of(bio, struct btrfs_io_bio, bio); 173 } 174 175 struct btrfs_bio_stripe { 176 struct btrfs_device *dev; 177 u64 physical; 178 u64 length; /* only used for discard mappings */ 179 }; 180 181 struct btrfs_bio; 182 typedef void (btrfs_bio_end_io_t) (struct btrfs_bio *bio, int err); 183 184 struct btrfs_bio { 185 atomic_t stripes_pending; 186 bio_end_io_t *end_io; 187 struct bio *orig_bio; 188 void *private; 189 atomic_t error; 190 int max_errors; 191 int num_stripes; 192 int mirror_num; 193 struct btrfs_bio_stripe stripes[]; 194 }; 195 196 struct btrfs_device_info { 197 struct btrfs_device *dev; 198 u64 dev_offset; 199 u64 max_avail; 200 u64 total_avail; 201 }; 202 203 struct btrfs_raid_attr { 204 int sub_stripes; /* sub_stripes info for map */ 205 int dev_stripes; /* stripes per dev */ 206 int devs_max; /* max devs to use */ 207 int devs_min; /* min devs needed */ 208 int devs_increment; /* ndevs has to be a multiple of this */ 209 int ncopies; /* how many copies to data has */ 210 }; 211 212 struct map_lookup { 213 u64 type; 214 int io_align; 215 int io_width; 216 int stripe_len; 217 int sector_size; 218 int num_stripes; 219 int sub_stripes; 220 struct btrfs_bio_stripe stripes[]; 221 }; 222 223 #define map_lookup_size(n) (sizeof(struct map_lookup) + \ 224 (sizeof(struct btrfs_bio_stripe) * (n))) 225 226 /* 227 * Restriper's general type filter 228 */ 229 #define BTRFS_BALANCE_DATA (1ULL << 0) 230 #define BTRFS_BALANCE_SYSTEM (1ULL << 1) 231 #define BTRFS_BALANCE_METADATA (1ULL << 2) 232 233 #define BTRFS_BALANCE_TYPE_MASK (BTRFS_BALANCE_DATA | \ 234 BTRFS_BALANCE_SYSTEM | \ 235 BTRFS_BALANCE_METADATA) 236 237 #define BTRFS_BALANCE_FORCE (1ULL << 3) 238 #define BTRFS_BALANCE_RESUME (1ULL << 4) 239 240 /* 241 * Balance filters 242 */ 243 #define BTRFS_BALANCE_ARGS_PROFILES (1ULL << 0) 244 #define BTRFS_BALANCE_ARGS_USAGE (1ULL << 1) 245 #define BTRFS_BALANCE_ARGS_DEVID (1ULL << 2) 246 #define BTRFS_BALANCE_ARGS_DRANGE (1ULL << 3) 247 #define BTRFS_BALANCE_ARGS_VRANGE (1ULL << 4) 248 249 /* 250 * Profile changing flags. When SOFT is set we won't relocate chunk if 251 * it already has the target profile (even though it may be 252 * half-filled). 253 */ 254 #define BTRFS_BALANCE_ARGS_CONVERT (1ULL << 8) 255 #define BTRFS_BALANCE_ARGS_SOFT (1ULL << 9) 256 257 struct btrfs_balance_args; 258 struct btrfs_balance_progress; 259 struct btrfs_balance_control { 260 struct btrfs_fs_info *fs_info; 261 262 struct btrfs_balance_args data; 263 struct btrfs_balance_args meta; 264 struct btrfs_balance_args sys; 265 266 u64 flags; 267 268 struct btrfs_balance_progress stat; 269 }; 270 271 int btrfs_account_dev_extents_size(struct btrfs_device *device, u64 start, 272 u64 end, u64 *length); 273 274 #define btrfs_bio_size(n) (sizeof(struct btrfs_bio) + \ 275 (sizeof(struct btrfs_bio_stripe) * (n))) 276 277 int btrfs_map_block(struct btrfs_fs_info *fs_info, int rw, 278 u64 logical, u64 *length, 279 struct btrfs_bio **bbio_ret, int mirror_num); 280 int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree, 281 u64 chunk_start, u64 physical, u64 devid, 282 u64 **logical, int *naddrs, int *stripe_len); 283 int btrfs_read_sys_array(struct btrfs_root *root); 284 int btrfs_read_chunk_tree(struct btrfs_root *root); 285 int btrfs_alloc_chunk(struct btrfs_trans_handle *trans, 286 struct btrfs_root *extent_root, u64 type); 287 void btrfs_mapping_init(struct btrfs_mapping_tree *tree); 288 void btrfs_mapping_tree_free(struct btrfs_mapping_tree *tree); 289 int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio, 290 int mirror_num, int async_submit); 291 int btrfs_open_devices(struct btrfs_fs_devices *fs_devices, 292 fmode_t flags, void *holder); 293 int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder, 294 struct btrfs_fs_devices **fs_devices_ret); 295 int btrfs_close_devices(struct btrfs_fs_devices *fs_devices); 296 void btrfs_close_extra_devices(struct btrfs_fs_info *fs_info, 297 struct btrfs_fs_devices *fs_devices, int step); 298 int btrfs_find_device_missing_or_by_path(struct btrfs_root *root, 299 char *device_path, 300 struct btrfs_device **device); 301 int btrfs_rm_device(struct btrfs_root *root, char *device_path); 302 void btrfs_cleanup_fs_uuids(void); 303 int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len); 304 int btrfs_grow_device(struct btrfs_trans_handle *trans, 305 struct btrfs_device *device, u64 new_size); 306 struct btrfs_device *btrfs_find_device(struct btrfs_fs_info *fs_info, u64 devid, 307 u8 *uuid, u8 *fsid); 308 int btrfs_shrink_device(struct btrfs_device *device, u64 new_size); 309 int btrfs_init_new_device(struct btrfs_root *root, char *path); 310 int btrfs_init_dev_replace_tgtdev(struct btrfs_root *root, char *device_path, 311 struct btrfs_device **device_out); 312 int btrfs_balance(struct btrfs_balance_control *bctl, 313 struct btrfs_ioctl_balance_args *bargs); 314 int btrfs_resume_balance_async(struct btrfs_fs_info *fs_info); 315 int btrfs_recover_balance(struct btrfs_fs_info *fs_info); 316 int btrfs_pause_balance(struct btrfs_fs_info *fs_info); 317 int btrfs_cancel_balance(struct btrfs_fs_info *fs_info); 318 int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset); 319 int find_free_dev_extent(struct btrfs_trans_handle *trans, 320 struct btrfs_device *device, u64 num_bytes, 321 u64 *start, u64 *max_avail); 322 void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index); 323 int btrfs_get_dev_stats(struct btrfs_root *root, 324 struct btrfs_ioctl_get_dev_stats *stats); 325 void btrfs_init_devices_late(struct btrfs_fs_info *fs_info); 326 int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info); 327 int btrfs_run_dev_stats(struct btrfs_trans_handle *trans, 328 struct btrfs_fs_info *fs_info); 329 void btrfs_rm_dev_replace_srcdev(struct btrfs_fs_info *fs_info, 330 struct btrfs_device *srcdev); 331 void btrfs_destroy_dev_replace_tgtdev(struct btrfs_fs_info *fs_info, 332 struct btrfs_device *tgtdev); 333 void btrfs_init_dev_replace_tgtdev_for_resume(struct btrfs_fs_info *fs_info, 334 struct btrfs_device *tgtdev); 335 int btrfs_scratch_superblock(struct btrfs_device *device); 336 int btrfs_is_parity_mirror(struct btrfs_mapping_tree *map_tree, 337 u64 logical, u64 len, int mirror_num); 338 unsigned long btrfs_full_stripe_len(struct btrfs_root *root, 339 struct btrfs_mapping_tree *map_tree, 340 u64 logical); 341 int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans, 342 struct btrfs_root *extent_root, 343 u64 chunk_offset, u64 chunk_size); 344 static inline void btrfs_dev_stat_inc(struct btrfs_device *dev, 345 int index) 346 { 347 atomic_inc(dev->dev_stat_values + index); 348 dev->dev_stats_dirty = 1; 349 } 350 351 static inline int btrfs_dev_stat_read(struct btrfs_device *dev, 352 int index) 353 { 354 return atomic_read(dev->dev_stat_values + index); 355 } 356 357 static inline int btrfs_dev_stat_read_and_reset(struct btrfs_device *dev, 358 int index) 359 { 360 int ret; 361 362 ret = atomic_xchg(dev->dev_stat_values + index, 0); 363 dev->dev_stats_dirty = 1; 364 return ret; 365 } 366 367 static inline void btrfs_dev_stat_set(struct btrfs_device *dev, 368 int index, unsigned long val) 369 { 370 atomic_set(dev->dev_stat_values + index, val); 371 dev->dev_stats_dirty = 1; 372 } 373 374 static inline void btrfs_dev_stat_reset(struct btrfs_device *dev, 375 int index) 376 { 377 btrfs_dev_stat_set(dev, index, 0); 378 } 379 #endif 380