1 // SPDX-License-Identifier: GPL-2.0
3 * Copyright (C) 2009 Oracle. All rights reserved.
6 #include <linux/sched.h>
7 #include <linux/slab.h>
8 #include <linux/sort.h>
10 #include "delayed-ref.h"
11 #include "transaction.h"
13 #include "space-info.h"
14 #include "tree-mod-log.h"
16 struct kmem_cache *btrfs_delayed_ref_head_cachep;
17 struct kmem_cache *btrfs_delayed_tree_ref_cachep;
18 struct kmem_cache *btrfs_delayed_data_ref_cachep;
19 struct kmem_cache *btrfs_delayed_extent_op_cachep;
21 * delayed back reference update tracking. For subvolume trees
22 * we queue up extent allocations and backref maintenance for
23 * delayed processing. This avoids deep call chains where we
24 * add extents in the middle of btrfs_search_slot, and it allows
25 * us to buffer up frequently modified backrefs in an rb tree instead
26 * of hammering updates on the extent allocation tree.
29 bool btrfs_check_space_for_delayed_refs(struct btrfs_fs_info *fs_info)
31 struct btrfs_block_rsv *delayed_refs_rsv = &fs_info->delayed_refs_rsv;
32 struct btrfs_block_rsv *global_rsv = &fs_info->global_block_rsv;
36 spin_lock(&global_rsv->lock);
37 reserved = global_rsv->reserved;
38 spin_unlock(&global_rsv->lock);
41 * Since the global reserve is just kind of magic we don't really want
42 * to rely on it to save our bacon, so if our size is more than the
43 * delayed_refs_rsv and the global rsv then it's time to think about
46 spin_lock(&delayed_refs_rsv->lock);
47 reserved += delayed_refs_rsv->reserved;
48 if (delayed_refs_rsv->size >= reserved)
50 spin_unlock(&delayed_refs_rsv->lock);
54 int btrfs_should_throttle_delayed_refs(struct btrfs_trans_handle *trans)
57 atomic_read(&trans->transaction->delayed_refs.num_entries);
62 avg_runtime = trans->fs_info->avg_delayed_ref_runtime;
63 val = num_entries * avg_runtime;
64 if (val >= NSEC_PER_SEC)
66 if (val >= NSEC_PER_SEC / 2)
69 return btrfs_check_space_for_delayed_refs(trans->fs_info);
73 * Release a ref head's reservation
75 * @fs_info: the filesystem
76 * @nr: number of items to drop
78 * This drops the delayed ref head's count from the delayed refs rsv and frees
79 * any excess reservation we had.
81 void btrfs_delayed_refs_rsv_release(struct btrfs_fs_info *fs_info, int nr)
83 struct btrfs_block_rsv *block_rsv = &fs_info->delayed_refs_rsv;
84 u64 num_bytes = btrfs_calc_insert_metadata_size(fs_info, nr);
88 * We have to check the mount option here because we could be enabling
89 * the free space tree for the first time and don't have the compat_ro
92 * We need extra reservations if we have the free space tree because
93 * we'll have to modify that tree as well.
95 if (btrfs_test_opt(fs_info, FREE_SPACE_TREE))
98 released = btrfs_block_rsv_release(fs_info, block_rsv, num_bytes, NULL);
100 trace_btrfs_space_reservation(fs_info, "delayed_refs_rsv",
105 * btrfs_update_delayed_refs_rsv - adjust the size of the delayed refs rsv
106 * @trans - the trans that may have generated delayed refs
108 * This is to be called anytime we may have adjusted trans->delayed_ref_updates,
109 * it'll calculate the additional size and add it to the delayed_refs_rsv.
111 void btrfs_update_delayed_refs_rsv(struct btrfs_trans_handle *trans)
113 struct btrfs_fs_info *fs_info = trans->fs_info;
114 struct btrfs_block_rsv *delayed_rsv = &fs_info->delayed_refs_rsv;
117 if (!trans->delayed_ref_updates)
120 num_bytes = btrfs_calc_insert_metadata_size(fs_info,
121 trans->delayed_ref_updates);
123 * We have to check the mount option here because we could be enabling
124 * the free space tree for the first time and don't have the compat_ro
127 * We need extra reservations if we have the free space tree because
128 * we'll have to modify that tree as well.
130 if (btrfs_test_opt(fs_info, FREE_SPACE_TREE))
133 spin_lock(&delayed_rsv->lock);
134 delayed_rsv->size += num_bytes;
135 delayed_rsv->full = 0;
136 spin_unlock(&delayed_rsv->lock);
137 trans->delayed_ref_updates = 0;
141 * Transfer bytes to our delayed refs rsv
143 * @fs_info: the filesystem
144 * @src: source block rsv to transfer from
145 * @num_bytes: number of bytes to transfer
147 * This transfers up to the num_bytes amount from the src rsv to the
148 * delayed_refs_rsv. Any extra bytes are returned to the space info.
150 void btrfs_migrate_to_delayed_refs_rsv(struct btrfs_fs_info *fs_info,
151 struct btrfs_block_rsv *src,
154 struct btrfs_block_rsv *delayed_refs_rsv = &fs_info->delayed_refs_rsv;
157 spin_lock(&src->lock);
158 src->reserved -= num_bytes;
159 src->size -= num_bytes;
160 spin_unlock(&src->lock);
162 spin_lock(&delayed_refs_rsv->lock);
163 if (delayed_refs_rsv->size > delayed_refs_rsv->reserved) {
164 u64 delta = delayed_refs_rsv->size -
165 delayed_refs_rsv->reserved;
166 if (num_bytes > delta) {
167 to_free = num_bytes - delta;
176 delayed_refs_rsv->reserved += num_bytes;
177 if (delayed_refs_rsv->reserved >= delayed_refs_rsv->size)
178 delayed_refs_rsv->full = 1;
179 spin_unlock(&delayed_refs_rsv->lock);
182 trace_btrfs_space_reservation(fs_info, "delayed_refs_rsv",
185 btrfs_space_info_free_bytes_may_use(fs_info,
186 delayed_refs_rsv->space_info, to_free);
190 * Refill based on our delayed refs usage
192 * @fs_info: the filesystem
193 * @flush: control how we can flush for this reservation.
195 * This will refill the delayed block_rsv up to 1 items size worth of space and
196 * will return -ENOSPC if we can't make the reservation.
198 int btrfs_delayed_refs_rsv_refill(struct btrfs_fs_info *fs_info,
199 enum btrfs_reserve_flush_enum flush)
201 struct btrfs_block_rsv *block_rsv = &fs_info->delayed_refs_rsv;
202 u64 limit = btrfs_calc_insert_metadata_size(fs_info, 1);
206 spin_lock(&block_rsv->lock);
207 if (block_rsv->reserved < block_rsv->size) {
208 num_bytes = block_rsv->size - block_rsv->reserved;
209 num_bytes = min(num_bytes, limit);
211 spin_unlock(&block_rsv->lock);
216 ret = btrfs_reserve_metadata_bytes(fs_info, block_rsv, num_bytes, flush);
219 btrfs_block_rsv_add_bytes(block_rsv, num_bytes, 0);
220 trace_btrfs_space_reservation(fs_info, "delayed_refs_rsv",
226 * compare two delayed tree backrefs with same bytenr and type
228 static int comp_tree_refs(struct btrfs_delayed_tree_ref *ref1,
229 struct btrfs_delayed_tree_ref *ref2)
231 if (ref1->node.type == BTRFS_TREE_BLOCK_REF_KEY) {
232 if (ref1->root < ref2->root)
234 if (ref1->root > ref2->root)
237 if (ref1->parent < ref2->parent)
239 if (ref1->parent > ref2->parent)
246 * compare two delayed data backrefs with same bytenr and type
248 static int comp_data_refs(struct btrfs_delayed_data_ref *ref1,
249 struct btrfs_delayed_data_ref *ref2)
251 if (ref1->node.type == BTRFS_EXTENT_DATA_REF_KEY) {
252 if (ref1->root < ref2->root)
254 if (ref1->root > ref2->root)
256 if (ref1->objectid < ref2->objectid)
258 if (ref1->objectid > ref2->objectid)
260 if (ref1->offset < ref2->offset)
262 if (ref1->offset > ref2->offset)
265 if (ref1->parent < ref2->parent)
267 if (ref1->parent > ref2->parent)
273 static int comp_refs(struct btrfs_delayed_ref_node *ref1,
274 struct btrfs_delayed_ref_node *ref2,
279 if (ref1->type < ref2->type)
281 if (ref1->type > ref2->type)
283 if (ref1->type == BTRFS_TREE_BLOCK_REF_KEY ||
284 ref1->type == BTRFS_SHARED_BLOCK_REF_KEY)
285 ret = comp_tree_refs(btrfs_delayed_node_to_tree_ref(ref1),
286 btrfs_delayed_node_to_tree_ref(ref2));
288 ret = comp_data_refs(btrfs_delayed_node_to_data_ref(ref1),
289 btrfs_delayed_node_to_data_ref(ref2));
293 if (ref1->seq < ref2->seq)
295 if (ref1->seq > ref2->seq)
301 /* insert a new ref to head ref rbtree */
302 static struct btrfs_delayed_ref_head *htree_insert(struct rb_root_cached *root,
303 struct rb_node *node)
305 struct rb_node **p = &root->rb_root.rb_node;
306 struct rb_node *parent_node = NULL;
307 struct btrfs_delayed_ref_head *entry;
308 struct btrfs_delayed_ref_head *ins;
310 bool leftmost = true;
312 ins = rb_entry(node, struct btrfs_delayed_ref_head, href_node);
313 bytenr = ins->bytenr;
316 entry = rb_entry(parent_node, struct btrfs_delayed_ref_head,
319 if (bytenr < entry->bytenr) {
321 } else if (bytenr > entry->bytenr) {
329 rb_link_node(node, parent_node, p);
330 rb_insert_color_cached(node, root, leftmost);
334 static struct btrfs_delayed_ref_node* tree_insert(struct rb_root_cached *root,
335 struct btrfs_delayed_ref_node *ins)
337 struct rb_node **p = &root->rb_root.rb_node;
338 struct rb_node *node = &ins->ref_node;
339 struct rb_node *parent_node = NULL;
340 struct btrfs_delayed_ref_node *entry;
341 bool leftmost = true;
347 entry = rb_entry(parent_node, struct btrfs_delayed_ref_node,
349 comp = comp_refs(ins, entry, true);
352 } else if (comp > 0) {
360 rb_link_node(node, parent_node, p);
361 rb_insert_color_cached(node, root, leftmost);
365 static struct btrfs_delayed_ref_head *find_first_ref_head(
366 struct btrfs_delayed_ref_root *dr)
369 struct btrfs_delayed_ref_head *entry;
371 n = rb_first_cached(&dr->href_root);
375 entry = rb_entry(n, struct btrfs_delayed_ref_head, href_node);
381 * Find a head entry based on bytenr. This returns the delayed ref head if it
382 * was able to find one, or NULL if nothing was in that spot. If return_bigger
383 * is given, the next bigger entry is returned if no exact match is found.
385 static struct btrfs_delayed_ref_head *find_ref_head(
386 struct btrfs_delayed_ref_root *dr, u64 bytenr,
389 struct rb_root *root = &dr->href_root.rb_root;
391 struct btrfs_delayed_ref_head *entry;
396 entry = rb_entry(n, struct btrfs_delayed_ref_head, href_node);
398 if (bytenr < entry->bytenr)
400 else if (bytenr > entry->bytenr)
405 if (entry && return_bigger) {
406 if (bytenr > entry->bytenr) {
407 n = rb_next(&entry->href_node);
410 entry = rb_entry(n, struct btrfs_delayed_ref_head,
418 int btrfs_delayed_ref_lock(struct btrfs_delayed_ref_root *delayed_refs,
419 struct btrfs_delayed_ref_head *head)
421 lockdep_assert_held(&delayed_refs->lock);
422 if (mutex_trylock(&head->mutex))
425 refcount_inc(&head->refs);
426 spin_unlock(&delayed_refs->lock);
428 mutex_lock(&head->mutex);
429 spin_lock(&delayed_refs->lock);
430 if (RB_EMPTY_NODE(&head->href_node)) {
431 mutex_unlock(&head->mutex);
432 btrfs_put_delayed_ref_head(head);
435 btrfs_put_delayed_ref_head(head);
439 static inline void drop_delayed_ref(struct btrfs_trans_handle *trans,
440 struct btrfs_delayed_ref_root *delayed_refs,
441 struct btrfs_delayed_ref_head *head,
442 struct btrfs_delayed_ref_node *ref)
444 lockdep_assert_held(&head->lock);
445 rb_erase_cached(&ref->ref_node, &head->ref_tree);
446 RB_CLEAR_NODE(&ref->ref_node);
447 if (!list_empty(&ref->add_list))
448 list_del(&ref->add_list);
450 btrfs_put_delayed_ref(ref);
451 atomic_dec(&delayed_refs->num_entries);
454 static bool merge_ref(struct btrfs_trans_handle *trans,
455 struct btrfs_delayed_ref_root *delayed_refs,
456 struct btrfs_delayed_ref_head *head,
457 struct btrfs_delayed_ref_node *ref,
460 struct btrfs_delayed_ref_node *next;
461 struct rb_node *node = rb_next(&ref->ref_node);
464 while (!done && node) {
467 next = rb_entry(node, struct btrfs_delayed_ref_node, ref_node);
468 node = rb_next(node);
469 if (seq && next->seq >= seq)
471 if (comp_refs(ref, next, false))
474 if (ref->action == next->action) {
477 if (ref->ref_mod < next->ref_mod) {
481 mod = -next->ref_mod;
484 drop_delayed_ref(trans, delayed_refs, head, next);
486 if (ref->ref_mod == 0) {
487 drop_delayed_ref(trans, delayed_refs, head, ref);
491 * Can't have multiples of the same ref on a tree block.
493 WARN_ON(ref->type == BTRFS_TREE_BLOCK_REF_KEY ||
494 ref->type == BTRFS_SHARED_BLOCK_REF_KEY);
501 void btrfs_merge_delayed_refs(struct btrfs_trans_handle *trans,
502 struct btrfs_delayed_ref_root *delayed_refs,
503 struct btrfs_delayed_ref_head *head)
505 struct btrfs_fs_info *fs_info = trans->fs_info;
506 struct btrfs_delayed_ref_node *ref;
507 struct rb_node *node;
510 lockdep_assert_held(&head->lock);
512 if (RB_EMPTY_ROOT(&head->ref_tree.rb_root))
515 /* We don't have too many refs to merge for data. */
519 seq = btrfs_tree_mod_log_lowest_seq(fs_info);
521 for (node = rb_first_cached(&head->ref_tree); node;
522 node = rb_next(node)) {
523 ref = rb_entry(node, struct btrfs_delayed_ref_node, ref_node);
524 if (seq && ref->seq >= seq)
526 if (merge_ref(trans, delayed_refs, head, ref, seq))
531 int btrfs_check_delayed_seq(struct btrfs_fs_info *fs_info, u64 seq)
534 u64 min_seq = btrfs_tree_mod_log_lowest_seq(fs_info);
536 if (min_seq != 0 && seq >= min_seq) {
538 "holding back delayed_ref %llu, lowest is %llu",
546 struct btrfs_delayed_ref_head *btrfs_select_ref_head(
547 struct btrfs_delayed_ref_root *delayed_refs)
549 struct btrfs_delayed_ref_head *head;
552 head = find_ref_head(delayed_refs, delayed_refs->run_delayed_start,
554 if (!head && delayed_refs->run_delayed_start != 0) {
555 delayed_refs->run_delayed_start = 0;
556 head = find_first_ref_head(delayed_refs);
561 while (head->processing) {
562 struct rb_node *node;
564 node = rb_next(&head->href_node);
566 if (delayed_refs->run_delayed_start == 0)
568 delayed_refs->run_delayed_start = 0;
571 head = rb_entry(node, struct btrfs_delayed_ref_head,
575 head->processing = 1;
576 WARN_ON(delayed_refs->num_heads_ready == 0);
577 delayed_refs->num_heads_ready--;
578 delayed_refs->run_delayed_start = head->bytenr +
583 void btrfs_delete_ref_head(struct btrfs_delayed_ref_root *delayed_refs,
584 struct btrfs_delayed_ref_head *head)
586 lockdep_assert_held(&delayed_refs->lock);
587 lockdep_assert_held(&head->lock);
589 rb_erase_cached(&head->href_node, &delayed_refs->href_root);
590 RB_CLEAR_NODE(&head->href_node);
591 atomic_dec(&delayed_refs->num_entries);
592 delayed_refs->num_heads--;
593 if (head->processing == 0)
594 delayed_refs->num_heads_ready--;
598 * Helper to insert the ref_node to the tail or merge with tail.
600 * Return 0 for insert.
601 * Return >0 for merge.
603 static int insert_delayed_ref(struct btrfs_trans_handle *trans,
604 struct btrfs_delayed_ref_root *root,
605 struct btrfs_delayed_ref_head *href,
606 struct btrfs_delayed_ref_node *ref)
608 struct btrfs_delayed_ref_node *exist;
612 spin_lock(&href->lock);
613 exist = tree_insert(&href->ref_tree, ref);
617 /* Now we are sure we can merge */
619 if (exist->action == ref->action) {
622 /* Need to change action */
623 if (exist->ref_mod < ref->ref_mod) {
624 exist->action = ref->action;
625 mod = -exist->ref_mod;
626 exist->ref_mod = ref->ref_mod;
627 if (ref->action == BTRFS_ADD_DELAYED_REF)
628 list_add_tail(&exist->add_list,
629 &href->ref_add_list);
630 else if (ref->action == BTRFS_DROP_DELAYED_REF) {
631 ASSERT(!list_empty(&exist->add_list));
632 list_del(&exist->add_list);
639 exist->ref_mod += mod;
641 /* remove existing tail if its ref_mod is zero */
642 if (exist->ref_mod == 0)
643 drop_delayed_ref(trans, root, href, exist);
644 spin_unlock(&href->lock);
647 if (ref->action == BTRFS_ADD_DELAYED_REF)
648 list_add_tail(&ref->add_list, &href->ref_add_list);
649 atomic_inc(&root->num_entries);
650 spin_unlock(&href->lock);
655 * helper function to update the accounting in the head ref
656 * existing and update must have the same bytenr
658 static noinline void update_existing_head_ref(struct btrfs_trans_handle *trans,
659 struct btrfs_delayed_ref_head *existing,
660 struct btrfs_delayed_ref_head *update)
662 struct btrfs_delayed_ref_root *delayed_refs =
663 &trans->transaction->delayed_refs;
664 struct btrfs_fs_info *fs_info = trans->fs_info;
667 BUG_ON(existing->is_data != update->is_data);
669 spin_lock(&existing->lock);
670 if (update->must_insert_reserved) {
671 /* if the extent was freed and then
672 * reallocated before the delayed ref
673 * entries were processed, we can end up
674 * with an existing head ref without
675 * the must_insert_reserved flag set.
678 existing->must_insert_reserved = update->must_insert_reserved;
681 * update the num_bytes so we make sure the accounting
684 existing->num_bytes = update->num_bytes;
688 if (update->extent_op) {
689 if (!existing->extent_op) {
690 existing->extent_op = update->extent_op;
692 if (update->extent_op->update_key) {
693 memcpy(&existing->extent_op->key,
694 &update->extent_op->key,
695 sizeof(update->extent_op->key));
696 existing->extent_op->update_key = true;
698 if (update->extent_op->update_flags) {
699 existing->extent_op->flags_to_set |=
700 update->extent_op->flags_to_set;
701 existing->extent_op->update_flags = true;
703 btrfs_free_delayed_extent_op(update->extent_op);
707 * update the reference mod on the head to reflect this new operation,
708 * only need the lock for this case cause we could be processing it
709 * currently, for refs we just added we know we're a-ok.
711 old_ref_mod = existing->total_ref_mod;
712 existing->ref_mod += update->ref_mod;
713 existing->total_ref_mod += update->ref_mod;
716 * If we are going to from a positive ref mod to a negative or vice
717 * versa we need to make sure to adjust pending_csums accordingly.
719 if (existing->is_data) {
721 btrfs_csum_bytes_to_leaves(fs_info,
722 existing->num_bytes);
724 if (existing->total_ref_mod >= 0 && old_ref_mod < 0) {
725 delayed_refs->pending_csums -= existing->num_bytes;
726 btrfs_delayed_refs_rsv_release(fs_info, csum_leaves);
728 if (existing->total_ref_mod < 0 && old_ref_mod >= 0) {
729 delayed_refs->pending_csums += existing->num_bytes;
730 trans->delayed_ref_updates += csum_leaves;
734 spin_unlock(&existing->lock);
737 static void init_delayed_ref_head(struct btrfs_delayed_ref_head *head_ref,
738 struct btrfs_qgroup_extent_record *qrecord,
739 u64 bytenr, u64 num_bytes, u64 ref_root,
740 u64 reserved, int action, bool is_data,
744 int must_insert_reserved = 0;
746 /* If reserved is provided, it must be a data extent. */
747 BUG_ON(!is_data && reserved);
750 * The head node stores the sum of all the mods, so dropping a ref
751 * should drop the sum in the head node by one.
753 if (action == BTRFS_UPDATE_DELAYED_HEAD)
755 else if (action == BTRFS_DROP_DELAYED_REF)
759 * BTRFS_ADD_DELAYED_EXTENT means that we need to update the reserved
760 * accounting when the extent is finally added, or if a later
761 * modification deletes the delayed ref without ever inserting the
762 * extent into the extent allocation tree. ref->must_insert_reserved
763 * is the flag used to record that accounting mods are required.
765 * Once we record must_insert_reserved, switch the action to
766 * BTRFS_ADD_DELAYED_REF because other special casing is not required.
768 if (action == BTRFS_ADD_DELAYED_EXTENT)
769 must_insert_reserved = 1;
771 must_insert_reserved = 0;
773 refcount_set(&head_ref->refs, 1);
774 head_ref->bytenr = bytenr;
775 head_ref->num_bytes = num_bytes;
776 head_ref->ref_mod = count_mod;
777 head_ref->must_insert_reserved = must_insert_reserved;
778 head_ref->is_data = is_data;
779 head_ref->is_system = is_system;
780 head_ref->ref_tree = RB_ROOT_CACHED;
781 INIT_LIST_HEAD(&head_ref->ref_add_list);
782 RB_CLEAR_NODE(&head_ref->href_node);
783 head_ref->processing = 0;
784 head_ref->total_ref_mod = count_mod;
785 spin_lock_init(&head_ref->lock);
786 mutex_init(&head_ref->mutex);
789 if (ref_root && reserved) {
790 qrecord->data_rsv = reserved;
791 qrecord->data_rsv_refroot = ref_root;
793 qrecord->bytenr = bytenr;
794 qrecord->num_bytes = num_bytes;
795 qrecord->old_roots = NULL;
800 * helper function to actually insert a head node into the rbtree.
801 * this does all the dirty work in terms of maintaining the correct
802 * overall modification count.
804 static noinline struct btrfs_delayed_ref_head *
805 add_delayed_ref_head(struct btrfs_trans_handle *trans,
806 struct btrfs_delayed_ref_head *head_ref,
807 struct btrfs_qgroup_extent_record *qrecord,
808 int action, int *qrecord_inserted_ret)
810 struct btrfs_delayed_ref_head *existing;
811 struct btrfs_delayed_ref_root *delayed_refs;
812 int qrecord_inserted = 0;
814 delayed_refs = &trans->transaction->delayed_refs;
816 /* Record qgroup extent info if provided */
818 if (btrfs_qgroup_trace_extent_nolock(trans->fs_info,
819 delayed_refs, qrecord))
822 qrecord_inserted = 1;
825 trace_add_delayed_ref_head(trans->fs_info, head_ref, action);
827 existing = htree_insert(&delayed_refs->href_root,
828 &head_ref->href_node);
830 update_existing_head_ref(trans, existing, head_ref);
832 * we've updated the existing ref, free the newly
835 kmem_cache_free(btrfs_delayed_ref_head_cachep, head_ref);
838 if (head_ref->is_data && head_ref->ref_mod < 0) {
839 delayed_refs->pending_csums += head_ref->num_bytes;
840 trans->delayed_ref_updates +=
841 btrfs_csum_bytes_to_leaves(trans->fs_info,
842 head_ref->num_bytes);
844 delayed_refs->num_heads++;
845 delayed_refs->num_heads_ready++;
846 atomic_inc(&delayed_refs->num_entries);
847 trans->delayed_ref_updates++;
849 if (qrecord_inserted_ret)
850 *qrecord_inserted_ret = qrecord_inserted;
856 * init_delayed_ref_common - Initialize the structure which represents a
857 * modification to a an extent.
859 * @fs_info: Internal to the mounted filesystem mount structure.
861 * @ref: The structure which is going to be initialized.
863 * @bytenr: The logical address of the extent for which a modification is
864 * going to be recorded.
866 * @num_bytes: Size of the extent whose modification is being recorded.
868 * @ref_root: The id of the root where this modification has originated, this
869 * can be either one of the well-known metadata trees or the
870 * subvolume id which references this extent.
872 * @action: Can be one of BTRFS_ADD_DELAYED_REF/BTRFS_DROP_DELAYED_REF or
873 * BTRFS_ADD_DELAYED_EXTENT
875 * @ref_type: Holds the type of the extent which is being recorded, can be
876 * one of BTRFS_SHARED_BLOCK_REF_KEY/BTRFS_TREE_BLOCK_REF_KEY
877 * when recording a metadata extent or BTRFS_SHARED_DATA_REF_KEY/
878 * BTRFS_EXTENT_DATA_REF_KEY when recording data extent
880 static void init_delayed_ref_common(struct btrfs_fs_info *fs_info,
881 struct btrfs_delayed_ref_node *ref,
882 u64 bytenr, u64 num_bytes, u64 ref_root,
883 int action, u8 ref_type)
887 if (action == BTRFS_ADD_DELAYED_EXTENT)
888 action = BTRFS_ADD_DELAYED_REF;
890 if (is_fstree(ref_root))
891 seq = atomic64_read(&fs_info->tree_mod_seq);
893 refcount_set(&ref->refs, 1);
894 ref->bytenr = bytenr;
895 ref->num_bytes = num_bytes;
897 ref->action = action;
901 ref->type = ref_type;
902 RB_CLEAR_NODE(&ref->ref_node);
903 INIT_LIST_HEAD(&ref->add_list);
907 * add a delayed tree ref. This does all of the accounting required
908 * to make sure the delayed ref is eventually processed before this
909 * transaction commits.
911 int btrfs_add_delayed_tree_ref(struct btrfs_trans_handle *trans,
912 struct btrfs_ref *generic_ref,
913 struct btrfs_delayed_extent_op *extent_op)
915 struct btrfs_fs_info *fs_info = trans->fs_info;
916 struct btrfs_delayed_tree_ref *ref;
917 struct btrfs_delayed_ref_head *head_ref;
918 struct btrfs_delayed_ref_root *delayed_refs;
919 struct btrfs_qgroup_extent_record *record = NULL;
920 int qrecord_inserted;
922 int action = generic_ref->action;
923 int level = generic_ref->tree_ref.level;
925 u64 bytenr = generic_ref->bytenr;
926 u64 num_bytes = generic_ref->len;
927 u64 parent = generic_ref->parent;
930 is_system = (generic_ref->tree_ref.owning_root == BTRFS_CHUNK_TREE_OBJECTID);
932 ASSERT(generic_ref->type == BTRFS_REF_METADATA && generic_ref->action);
933 ref = kmem_cache_alloc(btrfs_delayed_tree_ref_cachep, GFP_NOFS);
937 head_ref = kmem_cache_alloc(btrfs_delayed_ref_head_cachep, GFP_NOFS);
939 kmem_cache_free(btrfs_delayed_tree_ref_cachep, ref);
943 if (test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags) &&
944 !generic_ref->skip_qgroup) {
945 record = kzalloc(sizeof(*record), GFP_NOFS);
947 kmem_cache_free(btrfs_delayed_tree_ref_cachep, ref);
948 kmem_cache_free(btrfs_delayed_ref_head_cachep, head_ref);
954 ref_type = BTRFS_SHARED_BLOCK_REF_KEY;
956 ref_type = BTRFS_TREE_BLOCK_REF_KEY;
958 init_delayed_ref_common(fs_info, &ref->node, bytenr, num_bytes,
959 generic_ref->tree_ref.owning_root, action,
961 ref->root = generic_ref->tree_ref.owning_root;
962 ref->parent = parent;
965 init_delayed_ref_head(head_ref, record, bytenr, num_bytes,
966 generic_ref->tree_ref.owning_root, 0, action,
968 head_ref->extent_op = extent_op;
970 delayed_refs = &trans->transaction->delayed_refs;
971 spin_lock(&delayed_refs->lock);
974 * insert both the head node and the new ref without dropping
977 head_ref = add_delayed_ref_head(trans, head_ref, record,
978 action, &qrecord_inserted);
980 ret = insert_delayed_ref(trans, delayed_refs, head_ref, &ref->node);
981 spin_unlock(&delayed_refs->lock);
984 * Need to update the delayed_refs_rsv with any changes we may have
987 btrfs_update_delayed_refs_rsv(trans);
989 trace_add_delayed_tree_ref(fs_info, &ref->node, ref,
990 action == BTRFS_ADD_DELAYED_EXTENT ?
991 BTRFS_ADD_DELAYED_REF : action);
993 kmem_cache_free(btrfs_delayed_tree_ref_cachep, ref);
995 if (qrecord_inserted)
996 btrfs_qgroup_trace_extent_post(trans, record);
1002 * add a delayed data ref. it's similar to btrfs_add_delayed_tree_ref.
1004 int btrfs_add_delayed_data_ref(struct btrfs_trans_handle *trans,
1005 struct btrfs_ref *generic_ref,
1008 struct btrfs_fs_info *fs_info = trans->fs_info;
1009 struct btrfs_delayed_data_ref *ref;
1010 struct btrfs_delayed_ref_head *head_ref;
1011 struct btrfs_delayed_ref_root *delayed_refs;
1012 struct btrfs_qgroup_extent_record *record = NULL;
1013 int qrecord_inserted;
1014 int action = generic_ref->action;
1016 u64 bytenr = generic_ref->bytenr;
1017 u64 num_bytes = generic_ref->len;
1018 u64 parent = generic_ref->parent;
1019 u64 ref_root = generic_ref->data_ref.owning_root;
1020 u64 owner = generic_ref->data_ref.ino;
1021 u64 offset = generic_ref->data_ref.offset;
1024 ASSERT(generic_ref->type == BTRFS_REF_DATA && action);
1025 ref = kmem_cache_alloc(btrfs_delayed_data_ref_cachep, GFP_NOFS);
1030 ref_type = BTRFS_SHARED_DATA_REF_KEY;
1032 ref_type = BTRFS_EXTENT_DATA_REF_KEY;
1033 init_delayed_ref_common(fs_info, &ref->node, bytenr, num_bytes,
1034 ref_root, action, ref_type);
1035 ref->root = ref_root;
1036 ref->parent = parent;
1037 ref->objectid = owner;
1038 ref->offset = offset;
1041 head_ref = kmem_cache_alloc(btrfs_delayed_ref_head_cachep, GFP_NOFS);
1043 kmem_cache_free(btrfs_delayed_data_ref_cachep, ref);
1047 if (test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags) &&
1048 !generic_ref->skip_qgroup) {
1049 record = kzalloc(sizeof(*record), GFP_NOFS);
1051 kmem_cache_free(btrfs_delayed_data_ref_cachep, ref);
1052 kmem_cache_free(btrfs_delayed_ref_head_cachep,
1058 init_delayed_ref_head(head_ref, record, bytenr, num_bytes, ref_root,
1059 reserved, action, true, false);
1060 head_ref->extent_op = NULL;
1062 delayed_refs = &trans->transaction->delayed_refs;
1063 spin_lock(&delayed_refs->lock);
1066 * insert both the head node and the new ref without dropping
1069 head_ref = add_delayed_ref_head(trans, head_ref, record,
1070 action, &qrecord_inserted);
1072 ret = insert_delayed_ref(trans, delayed_refs, head_ref, &ref->node);
1073 spin_unlock(&delayed_refs->lock);
1076 * Need to update the delayed_refs_rsv with any changes we may have
1079 btrfs_update_delayed_refs_rsv(trans);
1081 trace_add_delayed_data_ref(trans->fs_info, &ref->node, ref,
1082 action == BTRFS_ADD_DELAYED_EXTENT ?
1083 BTRFS_ADD_DELAYED_REF : action);
1085 kmem_cache_free(btrfs_delayed_data_ref_cachep, ref);
1088 if (qrecord_inserted)
1089 return btrfs_qgroup_trace_extent_post(trans, record);
1093 int btrfs_add_delayed_extent_op(struct btrfs_trans_handle *trans,
1094 u64 bytenr, u64 num_bytes,
1095 struct btrfs_delayed_extent_op *extent_op)
1097 struct btrfs_delayed_ref_head *head_ref;
1098 struct btrfs_delayed_ref_root *delayed_refs;
1100 head_ref = kmem_cache_alloc(btrfs_delayed_ref_head_cachep, GFP_NOFS);
1104 init_delayed_ref_head(head_ref, NULL, bytenr, num_bytes, 0, 0,
1105 BTRFS_UPDATE_DELAYED_HEAD, false, false);
1106 head_ref->extent_op = extent_op;
1108 delayed_refs = &trans->transaction->delayed_refs;
1109 spin_lock(&delayed_refs->lock);
1111 add_delayed_ref_head(trans, head_ref, NULL, BTRFS_UPDATE_DELAYED_HEAD,
1114 spin_unlock(&delayed_refs->lock);
1117 * Need to update the delayed_refs_rsv with any changes we may have
1120 btrfs_update_delayed_refs_rsv(trans);
1125 * This does a simple search for the head node for a given extent. Returns the
1126 * head node if found, or NULL if not.
1128 struct btrfs_delayed_ref_head *
1129 btrfs_find_delayed_ref_head(struct btrfs_delayed_ref_root *delayed_refs, u64 bytenr)
1131 lockdep_assert_held(&delayed_refs->lock);
1133 return find_ref_head(delayed_refs, bytenr, false);
1136 void __cold btrfs_delayed_ref_exit(void)
1138 kmem_cache_destroy(btrfs_delayed_ref_head_cachep);
1139 kmem_cache_destroy(btrfs_delayed_tree_ref_cachep);
1140 kmem_cache_destroy(btrfs_delayed_data_ref_cachep);
1141 kmem_cache_destroy(btrfs_delayed_extent_op_cachep);
1144 int __init btrfs_delayed_ref_init(void)
1146 btrfs_delayed_ref_head_cachep = kmem_cache_create(
1147 "btrfs_delayed_ref_head",
1148 sizeof(struct btrfs_delayed_ref_head), 0,
1149 SLAB_MEM_SPREAD, NULL);
1150 if (!btrfs_delayed_ref_head_cachep)
1153 btrfs_delayed_tree_ref_cachep = kmem_cache_create(
1154 "btrfs_delayed_tree_ref",
1155 sizeof(struct btrfs_delayed_tree_ref), 0,
1156 SLAB_MEM_SPREAD, NULL);
1157 if (!btrfs_delayed_tree_ref_cachep)
1160 btrfs_delayed_data_ref_cachep = kmem_cache_create(
1161 "btrfs_delayed_data_ref",
1162 sizeof(struct btrfs_delayed_data_ref), 0,
1163 SLAB_MEM_SPREAD, NULL);
1164 if (!btrfs_delayed_data_ref_cachep)
1167 btrfs_delayed_extent_op_cachep = kmem_cache_create(
1168 "btrfs_delayed_extent_op",
1169 sizeof(struct btrfs_delayed_extent_op), 0,
1170 SLAB_MEM_SPREAD, NULL);
1171 if (!btrfs_delayed_extent_op_cachep)
1176 btrfs_delayed_ref_exit();