1 /* SPDX-License-Identifier: GPL-2.0 */
3 #ifndef BTRFS_BLOCK_GROUP_H
4 #define BTRFS_BLOCK_GROUP_H
6 #include "free-space-cache.h"
8 enum btrfs_disk_cache_state {
16 * This describes the state of the block_group for async discard. This is due
17 * to the two pass nature of it where extent discarding is prioritized over
18 * bitmap discarding. BTRFS_DISCARD_RESET_CURSOR is set when we are resetting
19 * between lists to prevent contention for discard state variables
20 * (eg. discard_cursor).
22 enum btrfs_discard_state {
23 BTRFS_DISCARD_EXTENTS,
24 BTRFS_DISCARD_BITMAPS,
25 BTRFS_DISCARD_RESET_CURSOR,
29 * Control flags for do_chunk_alloc's force field CHUNK_ALLOC_NO_FORCE means to
30 * only allocate a chunk if we really need one.
32 * CHUNK_ALLOC_LIMITED means to only try and allocate one if we have very few
33 * chunks already allocated. This is used as part of the clustering code to
34 * help make sure we have a good pool of storage to cluster in, without filling
35 * the FS with empty chunks
37 * CHUNK_ALLOC_FORCE means it must try to allocate one
39 * CHUNK_ALLOC_FORCE_FOR_EXTENT like CHUNK_ALLOC_FORCE but called from
40 * find_free_extent() that also activaes the zone
42 enum btrfs_chunk_alloc_enum {
46 CHUNK_ALLOC_FORCE_FOR_EXTENT,
49 /* Block group flags set at runtime */
50 enum btrfs_block_group_flags {
51 BLOCK_GROUP_FLAG_IREF,
52 BLOCK_GROUP_FLAG_REMOVED,
53 BLOCK_GROUP_FLAG_TO_COPY,
54 BLOCK_GROUP_FLAG_RELOCATING_REPAIR,
55 BLOCK_GROUP_FLAG_CHUNK_ITEM_INSERTED,
56 BLOCK_GROUP_FLAG_ZONE_IS_ACTIVE,
57 BLOCK_GROUP_FLAG_ZONED_DATA_RELOC,
60 enum btrfs_caching_type {
67 struct btrfs_caching_control {
68 struct list_head list;
70 wait_queue_head_t wait;
71 struct btrfs_work work;
72 struct btrfs_block_group *block_group;
76 /* Once caching_thread() finds this much free space, it will wake up waiters. */
77 #define CACHING_CTL_WAKE_UP SZ_2M
80 * Tree to record all locked full stripes of a RAID5/6 block group
82 struct btrfs_full_stripe_locks_tree {
87 struct btrfs_block_group {
88 struct btrfs_fs_info *fs_info;
103 * The last committed used bytes of this block group, if the above @used
104 * is still the same as @commit_used, we don't need to update block
105 * group item of this block group.
109 * If the free space extent count exceeds this number, convert the block
112 u32 bitmap_high_thresh;
115 * If the free space extent count drops below this number, convert the
116 * block group back to extents.
118 u32 bitmap_low_thresh;
121 * It is just used for the delayed data space allocation because
122 * only the data space allocation and the relative metadata update
123 * can be done cross the transaction.
125 struct rw_semaphore data_rwsem;
127 /* For raid56, this is a full stripe, without parity */
128 unsigned long full_stripe_len;
129 unsigned long runtime_flags;
133 int disk_cache_state;
135 /* Cache tracking stuff */
137 struct btrfs_caching_control *caching_ctl;
139 struct btrfs_space_info *space_info;
141 /* Free space cache stuff */
142 struct btrfs_free_space_ctl *free_space_ctl;
144 /* Block group cache stuff */
145 struct rb_node cache_node;
147 /* For block groups in the same raid type */
148 struct list_head list;
153 * List of struct btrfs_free_clusters for this block group.
154 * Today it will only have one thing on it, but that may change
156 struct list_head cluster_list;
158 /* For delayed block group creation or deletion of empty block groups */
159 struct list_head bg_list;
161 /* For read-only block groups */
162 struct list_head ro_list;
165 * When non-zero it means the block group's logical address and its
166 * device extents can not be reused for future block group allocations
167 * until the counter goes down to 0. This is to prevent them from being
168 * reused while some task is still using the block group after it was
169 * deleted - we want to make sure they can only be reused for new block
170 * groups after that task is done with the deleted block group.
174 /* For discard operations */
175 struct list_head discard_list;
177 u64 discard_eligible_time;
179 enum btrfs_discard_state discard_state;
181 /* For dirty block groups */
182 struct list_head dirty_list;
183 struct list_head io_list;
185 struct btrfs_io_ctl io_ctl;
188 * Incremented when doing extent allocations and holding a read lock
189 * on the space_info's groups_sem semaphore.
190 * Decremented when an ordered extent that represents an IO against this
191 * block group's range is created (after it's added to its inode's
192 * root's list of ordered extents) or immediately after the allocation
193 * if it's a metadata extent or fallocate extent (for these cases we
194 * don't create ordered extents).
196 atomic_t reservations;
199 * Incremented while holding the spinlock *lock* by a task checking if
200 * it can perform a nocow write (incremented if the value for the *ro*
201 * field is 0). Decremented by such tasks once they create an ordered
202 * extent or before that if some error happens before reaching that step.
203 * This is to prevent races between block group relocation and nocow
204 * writes through direct IO.
206 atomic_t nocow_writers;
208 /* Lock for free space tree operations. */
209 struct mutex free_space_lock;
212 * Does the block group need to be added to the free space tree?
213 * Protected by free_space_lock.
215 int needs_free_space;
217 /* Flag indicating this block group is placed on a sequential zone */
221 * Number of extents in this block group used for swap files.
222 * All accesses protected by the spinlock 'lock'.
226 /* Record locked full stripes for RAID5/6 block group */
227 struct btrfs_full_stripe_locks_tree full_stripe_locks_root;
230 * Allocation offset for the block group to implement sequential
231 * allocation. This is used only on a zoned filesystem.
236 u64 meta_write_pointer;
237 struct map_lookup *physical_map;
238 struct list_head active_bg_list;
239 struct work_struct zone_finish_work;
240 struct extent_buffer *last_eb;
243 static inline u64 btrfs_block_group_end(struct btrfs_block_group *block_group)
245 return (block_group->start + block_group->length);
248 static inline bool btrfs_is_block_group_data_only(
249 struct btrfs_block_group *block_group)
252 * In mixed mode the fragmentation is expected to be high, lowering the
253 * efficiency, so only proper data block groups are considered.
255 return (block_group->flags & BTRFS_BLOCK_GROUP_DATA) &&
256 !(block_group->flags & BTRFS_BLOCK_GROUP_METADATA);
259 #ifdef CONFIG_BTRFS_DEBUG
260 int btrfs_should_fragment_free_space(struct btrfs_block_group *block_group);
263 struct btrfs_block_group *btrfs_lookup_first_block_group(
264 struct btrfs_fs_info *info, u64 bytenr);
265 struct btrfs_block_group *btrfs_lookup_block_group(
266 struct btrfs_fs_info *info, u64 bytenr);
267 struct btrfs_block_group *btrfs_next_block_group(
268 struct btrfs_block_group *cache);
269 void btrfs_get_block_group(struct btrfs_block_group *cache);
270 void btrfs_put_block_group(struct btrfs_block_group *cache);
271 void btrfs_dec_block_group_reservations(struct btrfs_fs_info *fs_info,
273 void btrfs_wait_block_group_reservations(struct btrfs_block_group *bg);
274 struct btrfs_block_group *btrfs_inc_nocow_writers(struct btrfs_fs_info *fs_info,
276 void btrfs_dec_nocow_writers(struct btrfs_block_group *bg);
277 void btrfs_wait_nocow_writers(struct btrfs_block_group *bg);
278 void btrfs_wait_block_group_cache_progress(struct btrfs_block_group *cache,
280 int btrfs_cache_block_group(struct btrfs_block_group *cache, bool wait);
281 void btrfs_put_caching_control(struct btrfs_caching_control *ctl);
282 struct btrfs_caching_control *btrfs_get_caching_control(
283 struct btrfs_block_group *cache);
284 u64 add_new_free_space(struct btrfs_block_group *block_group,
286 struct btrfs_trans_handle *btrfs_start_trans_remove_block_group(
287 struct btrfs_fs_info *fs_info,
288 const u64 chunk_offset);
289 int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
290 u64 group_start, struct extent_map *em);
291 void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info);
292 void btrfs_mark_bg_unused(struct btrfs_block_group *bg);
293 void btrfs_reclaim_bgs_work(struct work_struct *work);
294 void btrfs_reclaim_bgs(struct btrfs_fs_info *fs_info);
295 void btrfs_mark_bg_to_reclaim(struct btrfs_block_group *bg);
296 int btrfs_read_block_groups(struct btrfs_fs_info *info);
297 struct btrfs_block_group *btrfs_make_block_group(struct btrfs_trans_handle *trans,
298 u64 bytes_used, u64 type,
299 u64 chunk_offset, u64 size);
300 void btrfs_create_pending_block_groups(struct btrfs_trans_handle *trans);
301 int btrfs_inc_block_group_ro(struct btrfs_block_group *cache,
302 bool do_chunk_alloc);
303 void btrfs_dec_block_group_ro(struct btrfs_block_group *cache);
304 int btrfs_start_dirty_block_groups(struct btrfs_trans_handle *trans);
305 int btrfs_write_dirty_block_groups(struct btrfs_trans_handle *trans);
306 int btrfs_setup_space_cache(struct btrfs_trans_handle *trans);
307 int btrfs_update_block_group(struct btrfs_trans_handle *trans,
308 u64 bytenr, u64 num_bytes, bool alloc);
309 int btrfs_add_reserved_bytes(struct btrfs_block_group *cache,
310 u64 ram_bytes, u64 num_bytes, int delalloc);
311 void btrfs_free_reserved_bytes(struct btrfs_block_group *cache,
312 u64 num_bytes, int delalloc);
313 int btrfs_chunk_alloc(struct btrfs_trans_handle *trans, u64 flags,
314 enum btrfs_chunk_alloc_enum force);
315 int btrfs_force_chunk_alloc(struct btrfs_trans_handle *trans, u64 type);
316 void check_system_chunk(struct btrfs_trans_handle *trans, const u64 type);
317 void btrfs_reserve_chunk_metadata(struct btrfs_trans_handle *trans,
318 bool is_item_insertion);
319 u64 btrfs_get_alloc_profile(struct btrfs_fs_info *fs_info, u64 orig_flags);
320 void btrfs_put_block_group_cache(struct btrfs_fs_info *info);
321 int btrfs_free_block_groups(struct btrfs_fs_info *info);
322 int btrfs_rmap_block(struct btrfs_fs_info *fs_info, u64 chunk_start,
323 struct block_device *bdev, u64 physical, u64 **logical,
324 int *naddrs, int *stripe_len);
326 static inline u64 btrfs_data_alloc_profile(struct btrfs_fs_info *fs_info)
328 return btrfs_get_alloc_profile(fs_info, BTRFS_BLOCK_GROUP_DATA);
331 static inline u64 btrfs_metadata_alloc_profile(struct btrfs_fs_info *fs_info)
333 return btrfs_get_alloc_profile(fs_info, BTRFS_BLOCK_GROUP_METADATA);
336 static inline u64 btrfs_system_alloc_profile(struct btrfs_fs_info *fs_info)
338 return btrfs_get_alloc_profile(fs_info, BTRFS_BLOCK_GROUP_SYSTEM);
341 static inline int btrfs_block_group_done(struct btrfs_block_group *cache)
344 return cache->cached == BTRFS_CACHE_FINISHED ||
345 cache->cached == BTRFS_CACHE_ERROR;
348 void btrfs_freeze_block_group(struct btrfs_block_group *cache);
349 void btrfs_unfreeze_block_group(struct btrfs_block_group *cache);
351 bool btrfs_inc_block_group_swap_extents(struct btrfs_block_group *bg);
352 void btrfs_dec_block_group_swap_extents(struct btrfs_block_group *bg, int amount);
354 #endif /* BTRFS_BLOCK_GROUP_H */