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 #define BTRFS_MAX_DATA_CHUNK_SIZE (10ULL * SZ_1G)
29 extern struct mutex uuid_mutex;
31 #define BTRFS_STRIPE_LEN SZ_64K
34 struct btrfs_pending_bios {
40 * Use sequence counter to get consistent device stat data on
43 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
44 #include <linux/seqlock.h>
45 #define __BTRFS_NEED_DEVICE_DATA_ORDERED
46 #define btrfs_device_data_ordered_init(device) \
47 seqcount_init(&device->data_seqcount)
49 #define btrfs_device_data_ordered_init(device) do { } while (0)
53 struct list_head dev_list;
54 struct list_head dev_alloc_list;
55 struct btrfs_fs_devices *fs_devices;
56 struct btrfs_fs_info *fs_info;
58 struct rcu_string *name;
62 spinlock_t io_lock ____cacheline_aligned;
64 /* regular prio bios */
65 struct btrfs_pending_bios pending_bios;
67 struct btrfs_pending_bios pending_sync_bios;
69 struct block_device *bdev;
71 /* the mode sent to blkdev_get */
78 int is_tgtdev_for_dev_replace;
79 blk_status_t last_flush_error;
82 #ifdef __BTRFS_NEED_DEVICE_DATA_ORDERED
83 seqcount_t data_seqcount;
86 /* the internal btrfs device id */
89 /* size of the device in memory */
92 /* size of the device on disk */
98 /* optimal io alignment for this device */
101 /* optimal io width for this device */
103 /* type and info about this device */
106 /* minimal io size for this device */
109 /* physical drive uuid (or lvm uuid) */
110 u8 uuid[BTRFS_UUID_SIZE];
113 * size of the device on the current transaction
115 * This variant is update when committing the transaction,
116 * and protected by device_list_mutex
118 u64 commit_total_bytes;
120 /* bytes used on the current transaction */
121 u64 commit_bytes_used;
123 * used to manage the device which is resized
125 * It is protected by chunk_lock.
127 struct list_head resized_list;
129 /* for sending down flush barriers */
130 struct bio *flush_bio;
131 struct completion flush_wait;
133 /* per-device scrub information */
134 struct scrub_ctx *scrub_device;
136 struct btrfs_work work;
138 struct work_struct rcu_work;
140 /* readahead state */
141 spinlock_t reada_lock;
142 atomic_t reada_in_flight;
144 struct reada_zone *reada_curr_zone;
145 struct radix_tree_root reada_zones;
146 struct radix_tree_root reada_extents;
148 /* disk I/O failure stats. For detailed description refer to
149 * enum btrfs_dev_stat_values in ioctl.h */
152 /* Counter to record the change of device stats */
153 atomic_t dev_stats_ccnt;
154 atomic_t dev_stat_values[BTRFS_DEV_STAT_VALUES_MAX];
158 * If we read those variants at the context of their own lock, we needn't
159 * use the following helpers, reading them directly is safe.
161 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
162 #define BTRFS_DEVICE_GETSET_FUNCS(name) \
164 btrfs_device_get_##name(const struct btrfs_device *dev) \
170 seq = read_seqcount_begin(&dev->data_seqcount); \
172 } while (read_seqcount_retry(&dev->data_seqcount, seq)); \
177 btrfs_device_set_##name(struct btrfs_device *dev, u64 size) \
180 write_seqcount_begin(&dev->data_seqcount); \
182 write_seqcount_end(&dev->data_seqcount); \
185 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
186 #define BTRFS_DEVICE_GETSET_FUNCS(name) \
188 btrfs_device_get_##name(const struct btrfs_device *dev) \
199 btrfs_device_set_##name(struct btrfs_device *dev, u64 size) \
206 #define BTRFS_DEVICE_GETSET_FUNCS(name) \
208 btrfs_device_get_##name(const struct btrfs_device *dev) \
214 btrfs_device_set_##name(struct btrfs_device *dev, u64 size) \
220 BTRFS_DEVICE_GETSET_FUNCS(total_bytes);
221 BTRFS_DEVICE_GETSET_FUNCS(disk_total_bytes);
222 BTRFS_DEVICE_GETSET_FUNCS(bytes_used);
224 struct btrfs_fs_devices {
225 u8 fsid[BTRFS_FSID_SIZE]; /* FS specific uuid */
233 struct block_device *latest_bdev;
235 /* all of the devices in the FS, protected by a mutex
236 * so we can safely walk it to write out the supers without
237 * worrying about add/remove by the multi-device code.
238 * Scrubbing super can kick off supers writing by holding
241 struct mutex device_list_mutex;
242 struct list_head devices;
244 struct list_head resized_devices;
245 /* devices not currently being allocated */
246 struct list_head alloc_list;
247 struct list_head list;
249 struct btrfs_fs_devices *seed;
254 /* set when we find or add a device that doesn't have the
259 struct btrfs_fs_info *fs_info;
261 struct kobject fsid_kobj;
262 struct kobject *device_dir_kobj;
263 struct completion kobj_unregister;
266 #define BTRFS_BIO_INLINE_CSUM_SIZE 64
269 * we need the mirror number and stripe index to be passed around
270 * the call chain while we are processing end_io (especially errors).
271 * Really, what we need is a btrfs_bio structure that has this info
272 * and is properly sized with its stripe array, but we're not there
273 * quite yet. We have our own btrfs bioset, and all of the bios
274 * we allocate are actually btrfs_io_bios. We'll cram as much of
275 * struct btrfs_bio as we can into this over time.
277 typedef void (btrfs_io_bio_end_io_t) (struct btrfs_io_bio *bio, int err);
278 struct btrfs_io_bio {
279 unsigned int mirror_num;
280 unsigned int stripe_index;
283 u8 csum_inline[BTRFS_BIO_INLINE_CSUM_SIZE];
285 btrfs_io_bio_end_io_t *end_io;
286 struct bvec_iter iter;
288 * This member must come last, bio_alloc_bioset will allocate enough
289 * bytes for entire btrfs_io_bio but relies on bio being last.
294 static inline struct btrfs_io_bio *btrfs_io_bio(struct bio *bio)
296 return container_of(bio, struct btrfs_io_bio, bio);
299 struct btrfs_bio_stripe {
300 struct btrfs_device *dev;
302 u64 length; /* only used for discard mappings */
306 typedef void (btrfs_bio_end_io_t) (struct btrfs_bio *bio, int err);
310 atomic_t stripes_pending;
311 struct btrfs_fs_info *fs_info;
312 u64 map_type; /* get from map_lookup->type */
313 bio_end_io_t *end_io;
314 struct bio *orig_bio;
324 * logical block numbers for the start of each stripe
325 * The last one or two are p/q. These are sorted,
326 * so raid_map[0] is the start of our full stripe
329 struct btrfs_bio_stripe stripes[];
332 struct btrfs_device_info {
333 struct btrfs_device *dev;
339 struct btrfs_raid_attr {
340 int sub_stripes; /* sub_stripes info for map */
341 int dev_stripes; /* stripes per dev */
342 int devs_max; /* max devs to use */
343 int devs_min; /* min devs needed */
344 int tolerated_failures; /* max tolerated fail devs */
345 int devs_increment; /* ndevs has to be a multiple of this */
346 int ncopies; /* how many copies to data has */
349 extern const struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES];
350 extern const int btrfs_raid_mindev_error[BTRFS_NR_RAID_TYPES];
351 extern const u64 btrfs_raid_group[BTRFS_NR_RAID_TYPES];
360 struct btrfs_bio_stripe stripes[];
363 #define map_lookup_size(n) (sizeof(struct map_lookup) + \
364 (sizeof(struct btrfs_bio_stripe) * (n)))
366 struct btrfs_balance_args;
367 struct btrfs_balance_progress;
368 struct btrfs_balance_control {
369 struct btrfs_fs_info *fs_info;
371 struct btrfs_balance_args data;
372 struct btrfs_balance_args meta;
373 struct btrfs_balance_args sys;
377 struct btrfs_balance_progress stat;
384 BTRFS_MAP_GET_READ_MIRRORS,
387 static inline enum btrfs_map_op btrfs_op(struct bio *bio)
389 switch (bio_op(bio)) {
391 return BTRFS_MAP_DISCARD;
393 return BTRFS_MAP_WRITE;
397 return BTRFS_MAP_READ;
401 int btrfs_account_dev_extents_size(struct btrfs_device *device, u64 start,
402 u64 end, u64 *length);
403 void btrfs_get_bbio(struct btrfs_bio *bbio);
404 void btrfs_put_bbio(struct btrfs_bio *bbio);
405 int btrfs_map_block(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
406 u64 logical, u64 *length,
407 struct btrfs_bio **bbio_ret, int mirror_num);
408 int btrfs_map_sblock(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
409 u64 logical, u64 *length,
410 struct btrfs_bio **bbio_ret);
411 int btrfs_rmap_block(struct btrfs_fs_info *fs_info,
412 u64 chunk_start, u64 physical, u64 devid,
413 u64 **logical, int *naddrs, int *stripe_len);
414 int btrfs_read_sys_array(struct btrfs_fs_info *fs_info);
415 int btrfs_read_chunk_tree(struct btrfs_fs_info *fs_info);
416 int btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
417 struct btrfs_fs_info *fs_info, u64 type);
418 void btrfs_mapping_init(struct btrfs_mapping_tree *tree);
419 void btrfs_mapping_tree_free(struct btrfs_mapping_tree *tree);
420 blk_status_t btrfs_map_bio(struct btrfs_fs_info *fs_info, struct bio *bio,
421 int mirror_num, int async_submit);
422 int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
423 fmode_t flags, void *holder);
424 int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
425 struct btrfs_fs_devices **fs_devices_ret);
426 int btrfs_close_devices(struct btrfs_fs_devices *fs_devices);
427 void btrfs_close_extra_devices(struct btrfs_fs_devices *fs_devices, int step);
428 void btrfs_assign_next_active_device(struct btrfs_fs_info *fs_info,
429 struct btrfs_device *device, struct btrfs_device *this_dev);
430 int btrfs_find_device_missing_or_by_path(struct btrfs_fs_info *fs_info,
431 const char *device_path,
432 struct btrfs_device **device);
433 int btrfs_find_device_by_devspec(struct btrfs_fs_info *fs_info, u64 devid,
435 struct btrfs_device **device);
436 struct btrfs_device *btrfs_alloc_device(struct btrfs_fs_info *fs_info,
439 int btrfs_rm_device(struct btrfs_fs_info *fs_info,
440 const char *device_path, u64 devid);
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_fs_info *fs_info, const char *path);
449 int btrfs_init_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
450 const char *device_path,
451 struct btrfs_device *srcdev,
452 struct btrfs_device **device_out);
453 int btrfs_balance(struct btrfs_balance_control *bctl,
454 struct btrfs_ioctl_balance_args *bargs);
455 int btrfs_resume_balance_async(struct btrfs_fs_info *fs_info);
456 int btrfs_recover_balance(struct btrfs_fs_info *fs_info);
457 int btrfs_pause_balance(struct btrfs_fs_info *fs_info);
458 int btrfs_cancel_balance(struct btrfs_fs_info *fs_info);
459 int btrfs_create_uuid_tree(struct btrfs_fs_info *fs_info);
460 int btrfs_check_uuid_tree(struct btrfs_fs_info *fs_info);
461 int btrfs_chunk_readonly(struct btrfs_fs_info *fs_info, u64 chunk_offset);
462 int find_free_dev_extent_start(struct btrfs_transaction *transaction,
463 struct btrfs_device *device, u64 num_bytes,
464 u64 search_start, u64 *start, u64 *max_avail);
465 int find_free_dev_extent(struct btrfs_trans_handle *trans,
466 struct btrfs_device *device, u64 num_bytes,
467 u64 *start, u64 *max_avail);
468 void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index);
469 int btrfs_get_dev_stats(struct btrfs_fs_info *fs_info,
470 struct btrfs_ioctl_get_dev_stats *stats);
471 void btrfs_init_devices_late(struct btrfs_fs_info *fs_info);
472 int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info);
473 int btrfs_run_dev_stats(struct btrfs_trans_handle *trans,
474 struct btrfs_fs_info *fs_info);
475 void btrfs_rm_dev_replace_remove_srcdev(struct btrfs_fs_info *fs_info,
476 struct btrfs_device *srcdev);
477 void btrfs_rm_dev_replace_free_srcdev(struct btrfs_fs_info *fs_info,
478 struct btrfs_device *srcdev);
479 void btrfs_destroy_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
480 struct btrfs_device *tgtdev);
481 void btrfs_init_dev_replace_tgtdev_for_resume(struct btrfs_fs_info *fs_info,
482 struct btrfs_device *tgtdev);
483 void btrfs_scratch_superblocks(struct block_device *bdev, const char *device_path);
484 int btrfs_is_parity_mirror(struct btrfs_fs_info *fs_info,
485 u64 logical, u64 len);
486 unsigned long btrfs_full_stripe_len(struct btrfs_fs_info *fs_info,
488 int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans,
489 struct btrfs_fs_info *fs_info,
490 u64 chunk_offset, u64 chunk_size);
491 int btrfs_remove_chunk(struct btrfs_trans_handle *trans,
492 struct btrfs_fs_info *fs_info, 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);
504 * This memory barrier orders stores updating statistics before stores
505 * updating dev_stats_ccnt.
507 * It pairs with smp_rmb() in btrfs_run_dev_stats().
509 smp_mb__before_atomic();
510 atomic_inc(&dev->dev_stats_ccnt);
513 static inline int btrfs_dev_stat_read(struct btrfs_device *dev,
516 return atomic_read(dev->dev_stat_values + index);
519 static inline int btrfs_dev_stat_read_and_reset(struct btrfs_device *dev,
524 ret = atomic_xchg(dev->dev_stat_values + index, 0);
525 smp_mb__before_atomic();
526 atomic_inc(&dev->dev_stats_ccnt);
530 static inline void btrfs_dev_stat_set(struct btrfs_device *dev,
531 int index, unsigned long val)
533 atomic_set(dev->dev_stat_values + index, val);
535 * This memory barrier orders stores updating statistics before stores
536 * updating dev_stats_ccnt.
538 * It pairs with smp_rmb() in btrfs_run_dev_stats().
540 smp_mb__before_atomic();
541 atomic_inc(&dev->dev_stats_ccnt);
544 static inline void btrfs_dev_stat_reset(struct btrfs_device *dev,
547 btrfs_dev_stat_set(dev, index, 0);
550 void btrfs_update_commit_device_size(struct btrfs_fs_info *fs_info);
551 void btrfs_update_commit_device_bytes_used(struct btrfs_fs_info *fs_info,
552 struct btrfs_transaction *transaction);
554 struct list_head *btrfs_get_fs_uuids(void);
555 void btrfs_set_fs_info_ptr(struct btrfs_fs_info *fs_info);
556 void btrfs_reset_fs_info_ptr(struct btrfs_fs_info *fs_info);
558 bool btrfs_check_rw_degradable(struct btrfs_fs_info *fs_info);
559 void btrfs_report_missing_device(struct btrfs_fs_info *fs_info, u64 devid,