2 * Copyright (C) 2007 Oracle. All rights reserved.
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.
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.
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.
19 #ifndef __BTRFS_VOLUMES_
20 #define __BTRFS_VOLUMES_
22 #include <linux/bio.h>
23 #include <linux/sort.h>
24 #include <linux/btrfs.h>
25 #include "async-thread.h"
27 extern struct mutex uuid_mutex;
29 #define BTRFS_STRIPE_LEN (64 * 1024)
32 struct btrfs_pending_bios {
38 * Use sequence counter to get consistent device stat data on
41 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
42 #include <linux/seqlock.h>
43 #define __BTRFS_NEED_DEVICE_DATA_ORDERED
44 #define btrfs_device_data_ordered_init(device) \
45 seqcount_init(&device->data_seqcount)
47 #define btrfs_device_data_ordered_init(device) do { } while (0)
51 struct list_head dev_list;
52 struct list_head dev_alloc_list;
53 struct btrfs_fs_devices *fs_devices;
55 struct btrfs_root *dev_root;
57 struct rcu_string *name;
61 spinlock_t io_lock ____cacheline_aligned;
63 /* regular prio bios */
64 struct btrfs_pending_bios pending_bios;
66 struct btrfs_pending_bios pending_sync_bios;
68 struct block_device *bdev;
70 /* the mode sent to blkdev_get */
77 int is_tgtdev_for_dev_replace;
79 #ifdef __BTRFS_NEED_DEVICE_DATA_ORDERED
80 seqcount_t data_seqcount;
83 /* the internal btrfs device id */
86 /* size of the device in memory */
89 /* size of the device on disk */
95 /* optimal io alignment for this device */
98 /* optimal io width for this device */
100 /* type and info about this device */
103 /* minimal io size for this device */
106 /* physical drive uuid (or lvm uuid) */
107 u8 uuid[BTRFS_UUID_SIZE];
110 * size of the device on the current transaction
112 * This variant is update when committing the transaction,
113 * and protected by device_list_mutex
115 u64 commit_total_bytes;
117 /* bytes used on the current transaction */
118 u64 commit_bytes_used;
120 * used to manage the device which is resized
122 * It is protected by chunk_lock.
124 struct list_head resized_list;
126 /* for sending down flush barriers */
128 struct bio *flush_bio;
129 struct completion flush_wait;
131 /* per-device scrub information */
132 struct scrub_ctx *scrub_device;
134 struct btrfs_work work;
136 struct work_struct rcu_work;
138 /* readahead state */
139 spinlock_t reada_lock;
140 atomic_t reada_in_flight;
142 struct reada_zone *reada_curr_zone;
143 struct radix_tree_root reada_zones;
144 struct radix_tree_root reada_extents;
146 /* disk I/O failure stats. For detailed description refer to
147 * enum btrfs_dev_stat_values in ioctl.h */
150 /* Counter to record the change of device stats */
151 atomic_t dev_stats_ccnt;
152 atomic_t dev_stat_values[BTRFS_DEV_STAT_VALUES_MAX];
156 * If we read those variants at the context of their own lock, we needn't
157 * use the following helpers, reading them directly is safe.
159 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
160 #define BTRFS_DEVICE_GETSET_FUNCS(name) \
162 btrfs_device_get_##name(const struct btrfs_device *dev) \
168 seq = read_seqcount_begin(&dev->data_seqcount); \
170 } while (read_seqcount_retry(&dev->data_seqcount, seq)); \
175 btrfs_device_set_##name(struct btrfs_device *dev, u64 size) \
178 write_seqcount_begin(&dev->data_seqcount); \
180 write_seqcount_end(&dev->data_seqcount); \
183 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
184 #define BTRFS_DEVICE_GETSET_FUNCS(name) \
186 btrfs_device_get_##name(const struct btrfs_device *dev) \
197 btrfs_device_set_##name(struct btrfs_device *dev, u64 size) \
204 #define BTRFS_DEVICE_GETSET_FUNCS(name) \
206 btrfs_device_get_##name(const struct btrfs_device *dev) \
212 btrfs_device_set_##name(struct btrfs_device *dev, u64 size) \
218 BTRFS_DEVICE_GETSET_FUNCS(total_bytes);
219 BTRFS_DEVICE_GETSET_FUNCS(disk_total_bytes);
220 BTRFS_DEVICE_GETSET_FUNCS(bytes_used);
222 struct btrfs_fs_devices {
223 u8 fsid[BTRFS_FSID_SIZE]; /* FS specific uuid */
231 struct block_device *latest_bdev;
233 /* all of the devices in the FS, protected by a mutex
234 * so we can safely walk it to write out the supers without
235 * worrying about add/remove by the multi-device code.
236 * Scrubbing super can kick off supers writing by holding
239 struct mutex device_list_mutex;
240 struct list_head devices;
242 struct list_head resized_devices;
243 /* devices not currently being allocated */
244 struct list_head alloc_list;
245 struct list_head list;
247 struct btrfs_fs_devices *seed;
252 /* set when we find or add a device that doesn't have the
257 struct btrfs_fs_info *fs_info;
259 struct kobject fsid_kobj;
260 struct kobject *device_dir_kobj;
261 struct completion kobj_unregister;
264 #define BTRFS_BIO_INLINE_CSUM_SIZE 64
267 * we need the mirror number and stripe index to be passed around
268 * the call chain while we are processing end_io (especially errors).
269 * Really, what we need is a btrfs_bio structure that has this info
270 * and is properly sized with its stripe array, but we're not there
271 * quite yet. We have our own btrfs bioset, and all of the bios
272 * we allocate are actually btrfs_io_bios. We'll cram as much of
273 * struct btrfs_bio as we can into this over time.
275 typedef void (btrfs_io_bio_end_io_t) (struct btrfs_io_bio *bio, int err);
276 struct btrfs_io_bio {
277 unsigned int mirror_num;
278 unsigned int stripe_index;
281 u8 csum_inline[BTRFS_BIO_INLINE_CSUM_SIZE];
283 btrfs_io_bio_end_io_t *end_io;
287 static inline struct btrfs_io_bio *btrfs_io_bio(struct bio *bio)
289 return container_of(bio, struct btrfs_io_bio, bio);
292 struct btrfs_bio_stripe {
293 struct btrfs_device *dev;
295 u64 length; /* only used for discard mappings */
299 typedef void (btrfs_bio_end_io_t) (struct btrfs_bio *bio, int err);
303 atomic_t stripes_pending;
304 struct btrfs_fs_info *fs_info;
305 u64 map_type; /* get from map_lookup->type */
306 bio_end_io_t *end_io;
307 struct bio *orig_bio;
317 * logical block numbers for the start of each stripe
318 * The last one or two are p/q. These are sorted,
319 * so raid_map[0] is the start of our full stripe
322 struct btrfs_bio_stripe stripes[];
325 struct btrfs_device_info {
326 struct btrfs_device *dev;
332 struct btrfs_raid_attr {
333 int sub_stripes; /* sub_stripes info for map */
334 int dev_stripes; /* stripes per dev */
335 int devs_max; /* max devs to use */
336 int devs_min; /* min devs needed */
337 int tolerated_failures; /* max tolerated fail devs */
338 int devs_increment; /* ndevs has to be a multiple of this */
339 int ncopies; /* how many copies to data has */
342 extern const struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES];
344 extern const u64 btrfs_raid_group[BTRFS_NR_RAID_TYPES];
354 struct btrfs_bio_stripe stripes[];
357 #define map_lookup_size(n) (sizeof(struct map_lookup) + \
358 (sizeof(struct btrfs_bio_stripe) * (n)))
361 * Restriper's general type filter
363 #define BTRFS_BALANCE_DATA (1ULL << 0)
364 #define BTRFS_BALANCE_SYSTEM (1ULL << 1)
365 #define BTRFS_BALANCE_METADATA (1ULL << 2)
367 #define BTRFS_BALANCE_TYPE_MASK (BTRFS_BALANCE_DATA | \
368 BTRFS_BALANCE_SYSTEM | \
369 BTRFS_BALANCE_METADATA)
371 #define BTRFS_BALANCE_FORCE (1ULL << 3)
372 #define BTRFS_BALANCE_RESUME (1ULL << 4)
377 #define BTRFS_BALANCE_ARGS_PROFILES (1ULL << 0)
378 #define BTRFS_BALANCE_ARGS_USAGE (1ULL << 1)
379 #define BTRFS_BALANCE_ARGS_DEVID (1ULL << 2)
380 #define BTRFS_BALANCE_ARGS_DRANGE (1ULL << 3)
381 #define BTRFS_BALANCE_ARGS_VRANGE (1ULL << 4)
382 #define BTRFS_BALANCE_ARGS_LIMIT (1ULL << 5)
385 * Profile changing flags. When SOFT is set we won't relocate chunk if
386 * it already has the target profile (even though it may be
389 #define BTRFS_BALANCE_ARGS_CONVERT (1ULL << 8)
390 #define BTRFS_BALANCE_ARGS_SOFT (1ULL << 9)
392 struct btrfs_balance_args;
393 struct btrfs_balance_progress;
394 struct btrfs_balance_control {
395 struct btrfs_fs_info *fs_info;
397 struct btrfs_balance_args data;
398 struct btrfs_balance_args meta;
399 struct btrfs_balance_args sys;
403 struct btrfs_balance_progress stat;
406 int btrfs_account_dev_extents_size(struct btrfs_device *device, u64 start,
407 u64 end, u64 *length);
408 void btrfs_get_bbio(struct btrfs_bio *bbio);
409 void btrfs_put_bbio(struct btrfs_bio *bbio);
410 int btrfs_map_block(struct btrfs_fs_info *fs_info, int rw,
411 u64 logical, u64 *length,
412 struct btrfs_bio **bbio_ret, int mirror_num);
413 int btrfs_map_sblock(struct btrfs_fs_info *fs_info, int rw,
414 u64 logical, u64 *length,
415 struct btrfs_bio **bbio_ret, int mirror_num,
417 int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree,
418 u64 chunk_start, u64 physical, u64 devid,
419 u64 **logical, int *naddrs, int *stripe_len);
420 int btrfs_read_sys_array(struct btrfs_root *root);
421 int btrfs_read_chunk_tree(struct btrfs_root *root);
422 int btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
423 struct btrfs_root *extent_root, u64 type);
424 void btrfs_mapping_init(struct btrfs_mapping_tree *tree);
425 void btrfs_mapping_tree_free(struct btrfs_mapping_tree *tree);
426 int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
427 int mirror_num, int async_submit);
428 int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
429 fmode_t flags, void *holder);
430 int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
431 struct btrfs_fs_devices **fs_devices_ret);
432 int btrfs_close_devices(struct btrfs_fs_devices *fs_devices);
433 void btrfs_close_extra_devices(struct btrfs_fs_devices *fs_devices, int step);
434 int btrfs_find_device_missing_or_by_path(struct btrfs_root *root,
436 struct btrfs_device **device);
437 struct btrfs_device *btrfs_alloc_device(struct btrfs_fs_info *fs_info,
440 int btrfs_rm_device(struct btrfs_root *root, char *device_path);
441 void btrfs_cleanup_fs_uuids(void);
442 int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len);
443 int btrfs_grow_device(struct btrfs_trans_handle *trans,
444 struct btrfs_device *device, u64 new_size);
445 struct btrfs_device *btrfs_find_device(struct btrfs_fs_info *fs_info, u64 devid,
447 int btrfs_shrink_device(struct btrfs_device *device, u64 new_size);
448 int btrfs_init_new_device(struct btrfs_root *root, char *path);
449 int btrfs_init_dev_replace_tgtdev(struct btrfs_root *root, char *device_path,
450 struct btrfs_device *srcdev,
451 struct btrfs_device **device_out);
452 int btrfs_balance(struct btrfs_balance_control *bctl,
453 struct btrfs_ioctl_balance_args *bargs);
454 int btrfs_resume_balance_async(struct btrfs_fs_info *fs_info);
455 int btrfs_recover_balance(struct btrfs_fs_info *fs_info);
456 int btrfs_pause_balance(struct btrfs_fs_info *fs_info);
457 int btrfs_cancel_balance(struct btrfs_fs_info *fs_info);
458 int btrfs_create_uuid_tree(struct btrfs_fs_info *fs_info);
459 int btrfs_check_uuid_tree(struct btrfs_fs_info *fs_info);
460 int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset);
461 int find_free_dev_extent_start(struct btrfs_transaction *transaction,
462 struct btrfs_device *device, u64 num_bytes,
463 u64 search_start, u64 *start, u64 *max_avail);
464 int find_free_dev_extent(struct btrfs_trans_handle *trans,
465 struct btrfs_device *device, u64 num_bytes,
466 u64 *start, u64 *max_avail);
467 void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index);
468 int btrfs_get_dev_stats(struct btrfs_root *root,
469 struct btrfs_ioctl_get_dev_stats *stats);
470 void btrfs_init_devices_late(struct btrfs_fs_info *fs_info);
471 int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info);
472 int btrfs_run_dev_stats(struct btrfs_trans_handle *trans,
473 struct btrfs_fs_info *fs_info);
474 void btrfs_rm_dev_replace_remove_srcdev(struct btrfs_fs_info *fs_info,
475 struct btrfs_device *srcdev);
476 void btrfs_rm_dev_replace_free_srcdev(struct btrfs_fs_info *fs_info,
477 struct btrfs_device *srcdev);
478 void btrfs_destroy_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
479 struct btrfs_device *tgtdev);
480 void btrfs_init_dev_replace_tgtdev_for_resume(struct btrfs_fs_info *fs_info,
481 struct btrfs_device *tgtdev);
482 void btrfs_scratch_superblocks(struct block_device *bdev, char *device_path);
483 int btrfs_is_parity_mirror(struct btrfs_mapping_tree *map_tree,
484 u64 logical, u64 len, int mirror_num);
485 unsigned long btrfs_full_stripe_len(struct btrfs_root *root,
486 struct btrfs_mapping_tree *map_tree,
488 int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans,
489 struct btrfs_root *extent_root,
490 u64 chunk_offset, u64 chunk_size);
491 int btrfs_remove_chunk(struct btrfs_trans_handle *trans,
492 struct btrfs_root *root, u64 chunk_offset);
494 static inline int btrfs_dev_stats_dirty(struct btrfs_device *dev)
496 return atomic_read(&dev->dev_stats_ccnt);
499 static inline void btrfs_dev_stat_inc(struct btrfs_device *dev,
502 atomic_inc(dev->dev_stat_values + index);
503 smp_mb__before_atomic();
504 atomic_inc(&dev->dev_stats_ccnt);
507 static inline int btrfs_dev_stat_read(struct btrfs_device *dev,
510 return atomic_read(dev->dev_stat_values + index);
513 static inline int btrfs_dev_stat_read_and_reset(struct btrfs_device *dev,
518 ret = atomic_xchg(dev->dev_stat_values + index, 0);
519 smp_mb__before_atomic();
520 atomic_inc(&dev->dev_stats_ccnt);
524 static inline void btrfs_dev_stat_set(struct btrfs_device *dev,
525 int index, unsigned long val)
527 atomic_set(dev->dev_stat_values + index, val);
528 smp_mb__before_atomic();
529 atomic_inc(&dev->dev_stats_ccnt);
532 static inline void btrfs_dev_stat_reset(struct btrfs_device *dev,
535 btrfs_dev_stat_set(dev, index, 0);
538 void btrfs_update_commit_device_size(struct btrfs_fs_info *fs_info);
539 void btrfs_update_commit_device_bytes_used(struct btrfs_root *root,
540 struct btrfs_transaction *transaction);
542 static inline void lock_chunks(struct btrfs_root *root)
544 mutex_lock(&root->fs_info->chunk_mutex);
547 static inline void unlock_chunks(struct btrfs_root *root)
549 mutex_unlock(&root->fs_info->chunk_mutex);
552 struct list_head *btrfs_get_fs_uuids(void);
553 void btrfs_set_fs_info_ptr(struct btrfs_fs_info *fs_info);
554 void btrfs_reset_fs_info_ptr(struct btrfs_fs_info *fs_info);
555 void btrfs_close_one_device(struct btrfs_device *device);