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;
57 struct btrfs_root *dev_root;
59 struct rcu_string *name;
63 spinlock_t io_lock ____cacheline_aligned;
65 /* When true means this device has pending chunk alloc in
66 * current transaction. Protected by chunk_mutex.
68 bool has_pending_chunks;
70 /* regular prio bios */
71 struct btrfs_pending_bios pending_bios;
73 struct btrfs_pending_bios pending_sync_bios;
75 struct block_device *bdev;
77 /* the mode sent to blkdev_get */
84 int is_tgtdev_for_dev_replace;
86 #ifdef __BTRFS_NEED_DEVICE_DATA_ORDERED
87 seqcount_t data_seqcount;
90 /* the internal btrfs device id */
93 /* size of the device in memory */
96 /* size of the device on disk */
102 /* optimal io alignment for this device */
105 /* optimal io width for this device */
107 /* type and info about this device */
110 /* minimal io size for this device */
113 /* physical drive uuid (or lvm uuid) */
114 u8 uuid[BTRFS_UUID_SIZE];
117 * size of the device on the current transaction
119 * This variant is update when committing the transaction,
120 * and protected by device_list_mutex
122 u64 commit_total_bytes;
124 /* bytes used on the current transaction */
125 u64 commit_bytes_used;
127 * used to manage the device which is resized
129 * It is protected by chunk_lock.
131 struct list_head resized_list;
133 /* for sending down flush barriers */
135 struct bio *flush_bio;
136 struct completion flush_wait;
138 /* per-device scrub information */
139 struct scrub_ctx *scrub_device;
141 struct btrfs_work work;
143 struct work_struct rcu_work;
145 /* readahead state */
146 spinlock_t reada_lock;
147 atomic_t reada_in_flight;
149 struct reada_zone *reada_curr_zone;
150 struct radix_tree_root reada_zones;
151 struct radix_tree_root reada_extents;
153 /* disk I/O failure stats. For detailed description refer to
154 * enum btrfs_dev_stat_values in ioctl.h */
157 /* Counter to record the change of device stats */
158 atomic_t dev_stats_ccnt;
159 atomic_t dev_stat_values[BTRFS_DEV_STAT_VALUES_MAX];
163 * If we read those variants at the context of their own lock, we needn't
164 * use the following helpers, reading them directly is safe.
166 #if BITS_PER_LONG==32 && defined(CONFIG_SMP)
167 #define BTRFS_DEVICE_GETSET_FUNCS(name) \
169 btrfs_device_get_##name(const struct btrfs_device *dev) \
175 seq = read_seqcount_begin(&dev->data_seqcount); \
177 } while (read_seqcount_retry(&dev->data_seqcount, seq)); \
182 btrfs_device_set_##name(struct btrfs_device *dev, u64 size) \
185 write_seqcount_begin(&dev->data_seqcount); \
187 write_seqcount_end(&dev->data_seqcount); \
190 #elif BITS_PER_LONG==32 && defined(CONFIG_PREEMPT)
191 #define BTRFS_DEVICE_GETSET_FUNCS(name) \
193 btrfs_device_get_##name(const struct btrfs_device *dev) \
204 btrfs_device_set_##name(struct btrfs_device *dev, u64 size) \
211 #define BTRFS_DEVICE_GETSET_FUNCS(name) \
213 btrfs_device_get_##name(const struct btrfs_device *dev) \
219 btrfs_device_set_##name(struct btrfs_device *dev, u64 size) \
225 BTRFS_DEVICE_GETSET_FUNCS(total_bytes);
226 BTRFS_DEVICE_GETSET_FUNCS(disk_total_bytes);
227 BTRFS_DEVICE_GETSET_FUNCS(bytes_used);
229 struct btrfs_fs_devices {
230 u8 fsid[BTRFS_FSID_SIZE]; /* FS specific uuid */
238 struct block_device *latest_bdev;
240 /* all of the devices in the FS, protected by a mutex
241 * so we can safely walk it to write out the supers without
242 * worrying about add/remove by the multi-device code.
243 * Scrubbing super can kick off supers writing by holding
246 struct mutex device_list_mutex;
247 struct list_head devices;
249 struct list_head resized_devices;
250 /* devices not currently being allocated */
251 struct list_head alloc_list;
252 struct list_head list;
254 struct btrfs_fs_devices *seed;
259 /* set when we find or add a device that doesn't have the
264 struct btrfs_fs_info *fs_info;
266 struct kobject fsid_kobj;
267 struct kobject *device_dir_kobj;
268 struct completion kobj_unregister;
271 #define BTRFS_BIO_INLINE_CSUM_SIZE 64
274 * we need the mirror number and stripe index to be passed around
275 * the call chain while we are processing end_io (especially errors).
276 * Really, what we need is a btrfs_bio structure that has this info
277 * and is properly sized with its stripe array, but we're not there
278 * quite yet. We have our own btrfs bioset, and all of the bios
279 * we allocate are actually btrfs_io_bios. We'll cram as much of
280 * struct btrfs_bio as we can into this over time.
282 typedef void (btrfs_io_bio_end_io_t) (struct btrfs_io_bio *bio, int err);
283 struct btrfs_io_bio {
284 unsigned int mirror_num;
285 unsigned int stripe_index;
288 u8 csum_inline[BTRFS_BIO_INLINE_CSUM_SIZE];
290 btrfs_io_bio_end_io_t *end_io;
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;
323 * logical block numbers for the start of each stripe
324 * The last one or two are p/q. These are sorted,
325 * so raid_map[0] is the start of our full stripe
328 struct btrfs_bio_stripe stripes[];
331 struct btrfs_device_info {
332 struct btrfs_device *dev;
338 struct btrfs_raid_attr {
339 int sub_stripes; /* sub_stripes info for map */
340 int dev_stripes; /* stripes per dev */
341 int devs_max; /* max devs to use */
342 int devs_min; /* min devs needed */
343 int tolerated_failures; /* max tolerated fail devs */
344 int devs_increment; /* ndevs has to be a multiple of this */
345 int ncopies; /* how many copies to data has */
348 extern const struct btrfs_raid_attr btrfs_raid_array[BTRFS_NR_RAID_TYPES];
349 extern const int btrfs_raid_mindev_error[BTRFS_NR_RAID_TYPES];
350 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;
380 int btrfs_account_dev_extents_size(struct btrfs_device *device, u64 start,
381 u64 end, u64 *length);
382 void btrfs_get_bbio(struct btrfs_bio *bbio);
383 void btrfs_put_bbio(struct btrfs_bio *bbio);
384 int btrfs_map_block(struct btrfs_fs_info *fs_info, int op,
385 u64 logical, u64 *length,
386 struct btrfs_bio **bbio_ret, int mirror_num);
387 int btrfs_map_sblock(struct btrfs_fs_info *fs_info, int op,
388 u64 logical, u64 *length,
389 struct btrfs_bio **bbio_ret, int mirror_num,
391 int btrfs_rmap_block(struct btrfs_fs_info *fs_info,
392 u64 chunk_start, u64 physical, u64 devid,
393 u64 **logical, int *naddrs, int *stripe_len);
394 int btrfs_read_sys_array(struct btrfs_root *root);
395 int btrfs_read_chunk_tree(struct btrfs_root *root);
396 int btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
397 struct btrfs_root *extent_root, u64 type);
398 void btrfs_mapping_init(struct btrfs_mapping_tree *tree);
399 void btrfs_mapping_tree_free(struct btrfs_mapping_tree *tree);
400 int btrfs_map_bio(struct btrfs_root *root, struct bio *bio,
401 int mirror_num, int async_submit);
402 int btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
403 fmode_t flags, void *holder);
404 int btrfs_scan_one_device(const char *path, fmode_t flags, void *holder,
405 struct btrfs_fs_devices **fs_devices_ret);
406 int btrfs_close_devices(struct btrfs_fs_devices *fs_devices);
407 void btrfs_close_extra_devices(struct btrfs_fs_devices *fs_devices, int step);
408 void btrfs_assign_next_active_device(struct btrfs_fs_info *fs_info,
409 struct btrfs_device *device, struct btrfs_device *this_dev);
410 int btrfs_find_device_missing_or_by_path(struct btrfs_root *root,
412 struct btrfs_device **device);
413 int btrfs_find_device_by_devspec(struct btrfs_root *root, u64 devid,
415 struct btrfs_device **device);
416 struct btrfs_device *btrfs_alloc_device(struct btrfs_fs_info *fs_info,
419 int btrfs_rm_device(struct btrfs_root *root, char *device_path, u64 devid);
420 void btrfs_cleanup_fs_uuids(void);
421 int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len);
422 int btrfs_grow_device(struct btrfs_trans_handle *trans,
423 struct btrfs_device *device, u64 new_size);
424 struct btrfs_device *btrfs_find_device(struct btrfs_fs_info *fs_info, u64 devid,
426 int btrfs_shrink_device(struct btrfs_device *device, u64 new_size);
427 int btrfs_init_new_device(struct btrfs_root *root, char *path);
428 int btrfs_init_dev_replace_tgtdev(struct btrfs_root *root, char *device_path,
429 struct btrfs_device *srcdev,
430 struct btrfs_device **device_out);
431 int btrfs_balance(struct btrfs_balance_control *bctl,
432 struct btrfs_ioctl_balance_args *bargs);
433 int btrfs_resume_balance_async(struct btrfs_fs_info *fs_info);
434 int btrfs_recover_balance(struct btrfs_fs_info *fs_info);
435 int btrfs_pause_balance(struct btrfs_fs_info *fs_info);
436 int btrfs_cancel_balance(struct btrfs_fs_info *fs_info);
437 int btrfs_create_uuid_tree(struct btrfs_fs_info *fs_info);
438 int btrfs_check_uuid_tree(struct btrfs_fs_info *fs_info);
439 int btrfs_chunk_readonly(struct btrfs_root *root, u64 chunk_offset);
440 int find_free_dev_extent_start(struct btrfs_transaction *transaction,
441 struct btrfs_device *device, u64 num_bytes,
442 u64 search_start, u64 *start, u64 *max_avail);
443 int find_free_dev_extent(struct btrfs_trans_handle *trans,
444 struct btrfs_device *device, u64 num_bytes,
445 u64 *start, u64 *max_avail);
446 void btrfs_dev_stat_inc_and_print(struct btrfs_device *dev, int index);
447 int btrfs_get_dev_stats(struct btrfs_root *root,
448 struct btrfs_ioctl_get_dev_stats *stats);
449 void btrfs_init_devices_late(struct btrfs_fs_info *fs_info);
450 int btrfs_init_dev_stats(struct btrfs_fs_info *fs_info);
451 int btrfs_run_dev_stats(struct btrfs_trans_handle *trans,
452 struct btrfs_fs_info *fs_info);
453 void btrfs_rm_dev_replace_remove_srcdev(struct btrfs_fs_info *fs_info,
454 struct btrfs_device *srcdev);
455 void btrfs_rm_dev_replace_free_srcdev(struct btrfs_fs_info *fs_info,
456 struct btrfs_device *srcdev);
457 void btrfs_destroy_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
458 struct btrfs_device *tgtdev);
459 void btrfs_init_dev_replace_tgtdev_for_resume(struct btrfs_fs_info *fs_info,
460 struct btrfs_device *tgtdev);
461 void btrfs_scratch_superblocks(struct block_device *bdev, char *device_path);
462 int btrfs_is_parity_mirror(struct btrfs_mapping_tree *map_tree,
463 u64 logical, u64 len, int mirror_num);
464 unsigned long btrfs_full_stripe_len(struct btrfs_root *root,
465 struct btrfs_mapping_tree *map_tree,
467 int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans,
468 struct btrfs_root *extent_root,
469 u64 chunk_offset, u64 chunk_size);
470 int btrfs_remove_chunk(struct btrfs_trans_handle *trans,
471 struct btrfs_root *root, u64 chunk_offset);
473 static inline int btrfs_dev_stats_dirty(struct btrfs_device *dev)
475 return atomic_read(&dev->dev_stats_ccnt);
478 static inline void btrfs_dev_stat_inc(struct btrfs_device *dev,
481 atomic_inc(dev->dev_stat_values + index);
482 smp_mb__before_atomic();
483 atomic_inc(&dev->dev_stats_ccnt);
486 static inline int btrfs_dev_stat_read(struct btrfs_device *dev,
489 return atomic_read(dev->dev_stat_values + index);
492 static inline int btrfs_dev_stat_read_and_reset(struct btrfs_device *dev,
497 ret = atomic_xchg(dev->dev_stat_values + index, 0);
498 smp_mb__before_atomic();
499 atomic_inc(&dev->dev_stats_ccnt);
503 static inline void btrfs_dev_stat_set(struct btrfs_device *dev,
504 int index, unsigned long val)
506 atomic_set(dev->dev_stat_values + index, val);
507 smp_mb__before_atomic();
508 atomic_inc(&dev->dev_stats_ccnt);
511 static inline void btrfs_dev_stat_reset(struct btrfs_device *dev,
514 btrfs_dev_stat_set(dev, index, 0);
517 void btrfs_update_commit_device_size(struct btrfs_fs_info *fs_info);
518 void btrfs_update_commit_device_bytes_used(struct btrfs_root *root,
519 struct btrfs_transaction *transaction);
521 static inline void lock_chunks(struct btrfs_root *root)
523 mutex_lock(&root->fs_info->chunk_mutex);
526 static inline void unlock_chunks(struct btrfs_root *root)
528 mutex_unlock(&root->fs_info->chunk_mutex);
531 struct list_head *btrfs_get_fs_uuids(void);
532 void btrfs_set_fs_info_ptr(struct btrfs_fs_info *fs_info);
533 void btrfs_reset_fs_info_ptr(struct btrfs_fs_info *fs_info);