2 * Copyright (C) 2011 STRATO. 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 #include <linux/sched.h>
20 #include <linux/pagemap.h>
21 #include <linux/writeback.h>
22 #include <linux/blkdev.h>
23 #include <linux/rbtree.h>
24 #include <linux/slab.h>
25 #include <linux/workqueue.h>
26 #include <linux/btrfs.h>
29 #include "transaction.h"
34 #include "extent_io.h"
37 * - subvol delete -> delete when ref goes to 0? delete limits also?
41 * - copy also limits on subvol creation
43 * - caches fuer ulists
44 * - performance benchmarks
45 * - check all ioctl parameters
49 * one struct for each qgroup, organized in fs_info->qgroup_tree.
57 u64 rfer; /* referenced */
58 u64 rfer_cmpr; /* referenced compressed */
59 u64 excl; /* exclusive */
60 u64 excl_cmpr; /* exclusive compressed */
65 u64 lim_flags; /* which limits are set */
72 * reservation tracking
79 struct list_head groups; /* groups this group is member of */
80 struct list_head members; /* groups that are members of this group */
81 struct list_head dirty; /* dirty groups */
82 struct rb_node node; /* tree of qgroups */
85 * temp variables for accounting operations
92 * glue structure to represent the relations between qgroups.
94 struct btrfs_qgroup_list {
95 struct list_head next_group;
96 struct list_head next_member;
97 struct btrfs_qgroup *group;
98 struct btrfs_qgroup *member;
102 qgroup_rescan_init(struct btrfs_fs_info *fs_info, u64 progress_objectid,
104 static void qgroup_rescan_zero_tracking(struct btrfs_fs_info *fs_info);
106 /* must be called with qgroup_ioctl_lock held */
107 static struct btrfs_qgroup *find_qgroup_rb(struct btrfs_fs_info *fs_info,
110 struct rb_node *n = fs_info->qgroup_tree.rb_node;
111 struct btrfs_qgroup *qgroup;
114 qgroup = rb_entry(n, struct btrfs_qgroup, node);
115 if (qgroup->qgroupid < qgroupid)
117 else if (qgroup->qgroupid > qgroupid)
125 /* must be called with qgroup_lock held */
126 static struct btrfs_qgroup *add_qgroup_rb(struct btrfs_fs_info *fs_info,
129 struct rb_node **p = &fs_info->qgroup_tree.rb_node;
130 struct rb_node *parent = NULL;
131 struct btrfs_qgroup *qgroup;
135 qgroup = rb_entry(parent, struct btrfs_qgroup, node);
137 if (qgroup->qgroupid < qgroupid)
139 else if (qgroup->qgroupid > qgroupid)
145 qgroup = kzalloc(sizeof(*qgroup), GFP_ATOMIC);
147 return ERR_PTR(-ENOMEM);
149 qgroup->qgroupid = qgroupid;
150 INIT_LIST_HEAD(&qgroup->groups);
151 INIT_LIST_HEAD(&qgroup->members);
152 INIT_LIST_HEAD(&qgroup->dirty);
154 rb_link_node(&qgroup->node, parent, p);
155 rb_insert_color(&qgroup->node, &fs_info->qgroup_tree);
160 static void __del_qgroup_rb(struct btrfs_qgroup *qgroup)
162 struct btrfs_qgroup_list *list;
164 list_del(&qgroup->dirty);
165 while (!list_empty(&qgroup->groups)) {
166 list = list_first_entry(&qgroup->groups,
167 struct btrfs_qgroup_list, next_group);
168 list_del(&list->next_group);
169 list_del(&list->next_member);
173 while (!list_empty(&qgroup->members)) {
174 list = list_first_entry(&qgroup->members,
175 struct btrfs_qgroup_list, next_member);
176 list_del(&list->next_group);
177 list_del(&list->next_member);
183 /* must be called with qgroup_lock held */
184 static int del_qgroup_rb(struct btrfs_fs_info *fs_info, u64 qgroupid)
186 struct btrfs_qgroup *qgroup = find_qgroup_rb(fs_info, qgroupid);
191 rb_erase(&qgroup->node, &fs_info->qgroup_tree);
192 __del_qgroup_rb(qgroup);
196 /* must be called with qgroup_lock held */
197 static int add_relation_rb(struct btrfs_fs_info *fs_info,
198 u64 memberid, u64 parentid)
200 struct btrfs_qgroup *member;
201 struct btrfs_qgroup *parent;
202 struct btrfs_qgroup_list *list;
204 member = find_qgroup_rb(fs_info, memberid);
205 parent = find_qgroup_rb(fs_info, parentid);
206 if (!member || !parent)
209 list = kzalloc(sizeof(*list), GFP_ATOMIC);
213 list->group = parent;
214 list->member = member;
215 list_add_tail(&list->next_group, &member->groups);
216 list_add_tail(&list->next_member, &parent->members);
221 /* must be called with qgroup_lock held */
222 static int del_relation_rb(struct btrfs_fs_info *fs_info,
223 u64 memberid, u64 parentid)
225 struct btrfs_qgroup *member;
226 struct btrfs_qgroup *parent;
227 struct btrfs_qgroup_list *list;
229 member = find_qgroup_rb(fs_info, memberid);
230 parent = find_qgroup_rb(fs_info, parentid);
231 if (!member || !parent)
234 list_for_each_entry(list, &member->groups, next_group) {
235 if (list->group == parent) {
236 list_del(&list->next_group);
237 list_del(&list->next_member);
246 * The full config is read in one go, only called from open_ctree()
247 * It doesn't use any locking, as at this point we're still single-threaded
249 int btrfs_read_qgroup_config(struct btrfs_fs_info *fs_info)
251 struct btrfs_key key;
252 struct btrfs_key found_key;
253 struct btrfs_root *quota_root = fs_info->quota_root;
254 struct btrfs_path *path = NULL;
255 struct extent_buffer *l;
259 u64 rescan_progress = 0;
261 if (!fs_info->quota_enabled)
264 fs_info->qgroup_ulist = ulist_alloc(GFP_NOFS);
265 if (!fs_info->qgroup_ulist) {
270 path = btrfs_alloc_path();
276 /* default this to quota off, in case no status key is found */
277 fs_info->qgroup_flags = 0;
280 * pass 1: read status, all qgroup infos and limits
285 ret = btrfs_search_slot_for_read(quota_root, &key, path, 1, 1);
290 struct btrfs_qgroup *qgroup;
292 slot = path->slots[0];
294 btrfs_item_key_to_cpu(l, &found_key, slot);
296 if (found_key.type == BTRFS_QGROUP_STATUS_KEY) {
297 struct btrfs_qgroup_status_item *ptr;
299 ptr = btrfs_item_ptr(l, slot,
300 struct btrfs_qgroup_status_item);
302 if (btrfs_qgroup_status_version(l, ptr) !=
303 BTRFS_QGROUP_STATUS_VERSION) {
305 "btrfs: old qgroup version, quota disabled\n");
308 if (btrfs_qgroup_status_generation(l, ptr) !=
309 fs_info->generation) {
310 flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
312 "btrfs: qgroup generation mismatch, "
313 "marked as inconsistent\n");
315 fs_info->qgroup_flags = btrfs_qgroup_status_flags(l,
317 rescan_progress = btrfs_qgroup_status_rescan(l, ptr);
321 if (found_key.type != BTRFS_QGROUP_INFO_KEY &&
322 found_key.type != BTRFS_QGROUP_LIMIT_KEY)
325 qgroup = find_qgroup_rb(fs_info, found_key.offset);
326 if ((qgroup && found_key.type == BTRFS_QGROUP_INFO_KEY) ||
327 (!qgroup && found_key.type == BTRFS_QGROUP_LIMIT_KEY)) {
328 printk(KERN_ERR "btrfs: inconsitent qgroup config\n");
329 flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
332 qgroup = add_qgroup_rb(fs_info, found_key.offset);
333 if (IS_ERR(qgroup)) {
334 ret = PTR_ERR(qgroup);
338 switch (found_key.type) {
339 case BTRFS_QGROUP_INFO_KEY: {
340 struct btrfs_qgroup_info_item *ptr;
342 ptr = btrfs_item_ptr(l, slot,
343 struct btrfs_qgroup_info_item);
344 qgroup->rfer = btrfs_qgroup_info_rfer(l, ptr);
345 qgroup->rfer_cmpr = btrfs_qgroup_info_rfer_cmpr(l, ptr);
346 qgroup->excl = btrfs_qgroup_info_excl(l, ptr);
347 qgroup->excl_cmpr = btrfs_qgroup_info_excl_cmpr(l, ptr);
348 /* generation currently unused */
351 case BTRFS_QGROUP_LIMIT_KEY: {
352 struct btrfs_qgroup_limit_item *ptr;
354 ptr = btrfs_item_ptr(l, slot,
355 struct btrfs_qgroup_limit_item);
356 qgroup->lim_flags = btrfs_qgroup_limit_flags(l, ptr);
357 qgroup->max_rfer = btrfs_qgroup_limit_max_rfer(l, ptr);
358 qgroup->max_excl = btrfs_qgroup_limit_max_excl(l, ptr);
359 qgroup->rsv_rfer = btrfs_qgroup_limit_rsv_rfer(l, ptr);
360 qgroup->rsv_excl = btrfs_qgroup_limit_rsv_excl(l, ptr);
365 ret = btrfs_next_item(quota_root, path);
371 btrfs_release_path(path);
374 * pass 2: read all qgroup relations
377 key.type = BTRFS_QGROUP_RELATION_KEY;
379 ret = btrfs_search_slot_for_read(quota_root, &key, path, 1, 0);
383 slot = path->slots[0];
385 btrfs_item_key_to_cpu(l, &found_key, slot);
387 if (found_key.type != BTRFS_QGROUP_RELATION_KEY)
390 if (found_key.objectid > found_key.offset) {
391 /* parent <- member, not needed to build config */
392 /* FIXME should we omit the key completely? */
396 ret = add_relation_rb(fs_info, found_key.objectid,
398 if (ret == -ENOENT) {
400 "btrfs: orphan qgroup relation 0x%llx->0x%llx\n",
401 found_key.objectid, found_key.offset);
402 ret = 0; /* ignore the error */
407 ret = btrfs_next_item(quota_root, path);
414 fs_info->qgroup_flags |= flags;
415 if (!(fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_ON)) {
416 fs_info->quota_enabled = 0;
417 fs_info->pending_quota_state = 0;
418 } else if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN &&
420 ret = qgroup_rescan_init(fs_info, rescan_progress, 0);
422 btrfs_free_path(path);
425 ulist_free(fs_info->qgroup_ulist);
426 fs_info->qgroup_ulist = NULL;
427 fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_RESCAN;
430 return ret < 0 ? ret : 0;
434 * This is called from close_ctree() or open_ctree() or btrfs_quota_disable(),
435 * first two are in single-threaded paths.And for the third one, we have set
436 * quota_root to be null with qgroup_lock held before, so it is safe to clean
437 * up the in-memory structures without qgroup_lock held.
439 void btrfs_free_qgroup_config(struct btrfs_fs_info *fs_info)
442 struct btrfs_qgroup *qgroup;
444 while ((n = rb_first(&fs_info->qgroup_tree))) {
445 qgroup = rb_entry(n, struct btrfs_qgroup, node);
446 rb_erase(n, &fs_info->qgroup_tree);
447 __del_qgroup_rb(qgroup);
450 * we call btrfs_free_qgroup_config() when umounting
451 * filesystem and disabling quota, so we set qgroup_ulit
452 * to be null here to avoid double free.
454 ulist_free(fs_info->qgroup_ulist);
455 fs_info->qgroup_ulist = NULL;
458 static int add_qgroup_relation_item(struct btrfs_trans_handle *trans,
459 struct btrfs_root *quota_root,
463 struct btrfs_path *path;
464 struct btrfs_key key;
466 path = btrfs_alloc_path();
471 key.type = BTRFS_QGROUP_RELATION_KEY;
474 ret = btrfs_insert_empty_item(trans, quota_root, path, &key, 0);
476 btrfs_mark_buffer_dirty(path->nodes[0]);
478 btrfs_free_path(path);
482 static int del_qgroup_relation_item(struct btrfs_trans_handle *trans,
483 struct btrfs_root *quota_root,
487 struct btrfs_path *path;
488 struct btrfs_key key;
490 path = btrfs_alloc_path();
495 key.type = BTRFS_QGROUP_RELATION_KEY;
498 ret = btrfs_search_slot(trans, quota_root, &key, path, -1, 1);
507 ret = btrfs_del_item(trans, quota_root, path);
509 btrfs_free_path(path);
513 static int add_qgroup_item(struct btrfs_trans_handle *trans,
514 struct btrfs_root *quota_root, u64 qgroupid)
517 struct btrfs_path *path;
518 struct btrfs_qgroup_info_item *qgroup_info;
519 struct btrfs_qgroup_limit_item *qgroup_limit;
520 struct extent_buffer *leaf;
521 struct btrfs_key key;
523 path = btrfs_alloc_path();
528 key.type = BTRFS_QGROUP_INFO_KEY;
529 key.offset = qgroupid;
531 ret = btrfs_insert_empty_item(trans, quota_root, path, &key,
532 sizeof(*qgroup_info));
536 leaf = path->nodes[0];
537 qgroup_info = btrfs_item_ptr(leaf, path->slots[0],
538 struct btrfs_qgroup_info_item);
539 btrfs_set_qgroup_info_generation(leaf, qgroup_info, trans->transid);
540 btrfs_set_qgroup_info_rfer(leaf, qgroup_info, 0);
541 btrfs_set_qgroup_info_rfer_cmpr(leaf, qgroup_info, 0);
542 btrfs_set_qgroup_info_excl(leaf, qgroup_info, 0);
543 btrfs_set_qgroup_info_excl_cmpr(leaf, qgroup_info, 0);
545 btrfs_mark_buffer_dirty(leaf);
547 btrfs_release_path(path);
549 key.type = BTRFS_QGROUP_LIMIT_KEY;
550 ret = btrfs_insert_empty_item(trans, quota_root, path, &key,
551 sizeof(*qgroup_limit));
555 leaf = path->nodes[0];
556 qgroup_limit = btrfs_item_ptr(leaf, path->slots[0],
557 struct btrfs_qgroup_limit_item);
558 btrfs_set_qgroup_limit_flags(leaf, qgroup_limit, 0);
559 btrfs_set_qgroup_limit_max_rfer(leaf, qgroup_limit, 0);
560 btrfs_set_qgroup_limit_max_excl(leaf, qgroup_limit, 0);
561 btrfs_set_qgroup_limit_rsv_rfer(leaf, qgroup_limit, 0);
562 btrfs_set_qgroup_limit_rsv_excl(leaf, qgroup_limit, 0);
564 btrfs_mark_buffer_dirty(leaf);
568 btrfs_free_path(path);
572 static int del_qgroup_item(struct btrfs_trans_handle *trans,
573 struct btrfs_root *quota_root, u64 qgroupid)
576 struct btrfs_path *path;
577 struct btrfs_key key;
579 path = btrfs_alloc_path();
584 key.type = BTRFS_QGROUP_INFO_KEY;
585 key.offset = qgroupid;
586 ret = btrfs_search_slot(trans, quota_root, &key, path, -1, 1);
595 ret = btrfs_del_item(trans, quota_root, path);
599 btrfs_release_path(path);
601 key.type = BTRFS_QGROUP_LIMIT_KEY;
602 ret = btrfs_search_slot(trans, quota_root, &key, path, -1, 1);
611 ret = btrfs_del_item(trans, quota_root, path);
614 btrfs_free_path(path);
618 static int update_qgroup_limit_item(struct btrfs_trans_handle *trans,
619 struct btrfs_root *root, u64 qgroupid,
620 u64 flags, u64 max_rfer, u64 max_excl,
621 u64 rsv_rfer, u64 rsv_excl)
623 struct btrfs_path *path;
624 struct btrfs_key key;
625 struct extent_buffer *l;
626 struct btrfs_qgroup_limit_item *qgroup_limit;
631 key.type = BTRFS_QGROUP_LIMIT_KEY;
632 key.offset = qgroupid;
634 path = btrfs_alloc_path();
638 ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
646 slot = path->slots[0];
647 qgroup_limit = btrfs_item_ptr(l, path->slots[0],
648 struct btrfs_qgroup_limit_item);
649 btrfs_set_qgroup_limit_flags(l, qgroup_limit, flags);
650 btrfs_set_qgroup_limit_max_rfer(l, qgroup_limit, max_rfer);
651 btrfs_set_qgroup_limit_max_excl(l, qgroup_limit, max_excl);
652 btrfs_set_qgroup_limit_rsv_rfer(l, qgroup_limit, rsv_rfer);
653 btrfs_set_qgroup_limit_rsv_excl(l, qgroup_limit, rsv_excl);
655 btrfs_mark_buffer_dirty(l);
658 btrfs_free_path(path);
662 static int update_qgroup_info_item(struct btrfs_trans_handle *trans,
663 struct btrfs_root *root,
664 struct btrfs_qgroup *qgroup)
666 struct btrfs_path *path;
667 struct btrfs_key key;
668 struct extent_buffer *l;
669 struct btrfs_qgroup_info_item *qgroup_info;
674 key.type = BTRFS_QGROUP_INFO_KEY;
675 key.offset = qgroup->qgroupid;
677 path = btrfs_alloc_path();
681 ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
689 slot = path->slots[0];
690 qgroup_info = btrfs_item_ptr(l, path->slots[0],
691 struct btrfs_qgroup_info_item);
692 btrfs_set_qgroup_info_generation(l, qgroup_info, trans->transid);
693 btrfs_set_qgroup_info_rfer(l, qgroup_info, qgroup->rfer);
694 btrfs_set_qgroup_info_rfer_cmpr(l, qgroup_info, qgroup->rfer_cmpr);
695 btrfs_set_qgroup_info_excl(l, qgroup_info, qgroup->excl);
696 btrfs_set_qgroup_info_excl_cmpr(l, qgroup_info, qgroup->excl_cmpr);
698 btrfs_mark_buffer_dirty(l);
701 btrfs_free_path(path);
705 static int update_qgroup_status_item(struct btrfs_trans_handle *trans,
706 struct btrfs_fs_info *fs_info,
707 struct btrfs_root *root)
709 struct btrfs_path *path;
710 struct btrfs_key key;
711 struct extent_buffer *l;
712 struct btrfs_qgroup_status_item *ptr;
717 key.type = BTRFS_QGROUP_STATUS_KEY;
720 path = btrfs_alloc_path();
724 ret = btrfs_search_slot(trans, root, &key, path, 0, 1);
732 slot = path->slots[0];
733 ptr = btrfs_item_ptr(l, slot, struct btrfs_qgroup_status_item);
734 btrfs_set_qgroup_status_flags(l, ptr, fs_info->qgroup_flags);
735 btrfs_set_qgroup_status_generation(l, ptr, trans->transid);
736 btrfs_set_qgroup_status_rescan(l, ptr,
737 fs_info->qgroup_rescan_progress.objectid);
739 btrfs_mark_buffer_dirty(l);
742 btrfs_free_path(path);
747 * called with qgroup_lock held
749 static int btrfs_clean_quota_tree(struct btrfs_trans_handle *trans,
750 struct btrfs_root *root)
752 struct btrfs_path *path;
753 struct btrfs_key key;
754 struct extent_buffer *leaf = NULL;
758 path = btrfs_alloc_path();
762 path->leave_spinning = 1;
769 ret = btrfs_search_slot(trans, root, &key, path, -1, 1);
772 leaf = path->nodes[0];
773 nr = btrfs_header_nritems(leaf);
777 * delete the leaf one by one
778 * since the whole tree is going
782 ret = btrfs_del_items(trans, root, path, 0, nr);
786 btrfs_release_path(path);
790 root->fs_info->pending_quota_state = 0;
791 btrfs_free_path(path);
795 int btrfs_quota_enable(struct btrfs_trans_handle *trans,
796 struct btrfs_fs_info *fs_info)
798 struct btrfs_root *quota_root;
799 struct btrfs_root *tree_root = fs_info->tree_root;
800 struct btrfs_path *path = NULL;
801 struct btrfs_qgroup_status_item *ptr;
802 struct extent_buffer *leaf;
803 struct btrfs_key key;
804 struct btrfs_key found_key;
805 struct btrfs_qgroup *qgroup = NULL;
809 mutex_lock(&fs_info->qgroup_ioctl_lock);
810 if (fs_info->quota_root) {
811 fs_info->pending_quota_state = 1;
815 fs_info->qgroup_ulist = ulist_alloc(GFP_NOFS);
816 if (!fs_info->qgroup_ulist) {
822 * initially create the quota tree
824 quota_root = btrfs_create_tree(trans, fs_info,
825 BTRFS_QUOTA_TREE_OBJECTID);
826 if (IS_ERR(quota_root)) {
827 ret = PTR_ERR(quota_root);
831 path = btrfs_alloc_path();
838 key.type = BTRFS_QGROUP_STATUS_KEY;
841 ret = btrfs_insert_empty_item(trans, quota_root, path, &key,
846 leaf = path->nodes[0];
847 ptr = btrfs_item_ptr(leaf, path->slots[0],
848 struct btrfs_qgroup_status_item);
849 btrfs_set_qgroup_status_generation(leaf, ptr, trans->transid);
850 btrfs_set_qgroup_status_version(leaf, ptr, BTRFS_QGROUP_STATUS_VERSION);
851 fs_info->qgroup_flags = BTRFS_QGROUP_STATUS_FLAG_ON |
852 BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
853 btrfs_set_qgroup_status_flags(leaf, ptr, fs_info->qgroup_flags);
854 btrfs_set_qgroup_status_rescan(leaf, ptr, 0);
856 btrfs_mark_buffer_dirty(leaf);
859 key.type = BTRFS_ROOT_REF_KEY;
862 btrfs_release_path(path);
863 ret = btrfs_search_slot_for_read(tree_root, &key, path, 1, 0);
871 slot = path->slots[0];
872 leaf = path->nodes[0];
873 btrfs_item_key_to_cpu(leaf, &found_key, slot);
875 if (found_key.type == BTRFS_ROOT_REF_KEY) {
876 ret = add_qgroup_item(trans, quota_root,
881 qgroup = add_qgroup_rb(fs_info, found_key.offset);
882 if (IS_ERR(qgroup)) {
883 ret = PTR_ERR(qgroup);
887 ret = btrfs_next_item(tree_root, path);
895 btrfs_release_path(path);
896 ret = add_qgroup_item(trans, quota_root, BTRFS_FS_TREE_OBJECTID);
900 qgroup = add_qgroup_rb(fs_info, BTRFS_FS_TREE_OBJECTID);
901 if (IS_ERR(qgroup)) {
902 ret = PTR_ERR(qgroup);
905 spin_lock(&fs_info->qgroup_lock);
906 fs_info->quota_root = quota_root;
907 fs_info->pending_quota_state = 1;
908 spin_unlock(&fs_info->qgroup_lock);
910 btrfs_free_path(path);
913 free_extent_buffer(quota_root->node);
914 free_extent_buffer(quota_root->commit_root);
919 ulist_free(fs_info->qgroup_ulist);
920 fs_info->qgroup_ulist = NULL;
922 mutex_unlock(&fs_info->qgroup_ioctl_lock);
926 int btrfs_quota_disable(struct btrfs_trans_handle *trans,
927 struct btrfs_fs_info *fs_info)
929 struct btrfs_root *tree_root = fs_info->tree_root;
930 struct btrfs_root *quota_root;
933 mutex_lock(&fs_info->qgroup_ioctl_lock);
934 if (!fs_info->quota_root)
936 spin_lock(&fs_info->qgroup_lock);
937 fs_info->quota_enabled = 0;
938 fs_info->pending_quota_state = 0;
939 quota_root = fs_info->quota_root;
940 fs_info->quota_root = NULL;
941 spin_unlock(&fs_info->qgroup_lock);
943 btrfs_free_qgroup_config(fs_info);
945 ret = btrfs_clean_quota_tree(trans, quota_root);
949 ret = btrfs_del_root(trans, tree_root, "a_root->root_key);
953 list_del("a_root->dirty_list);
955 btrfs_tree_lock(quota_root->node);
956 clean_tree_block(trans, tree_root, quota_root->node);
957 btrfs_tree_unlock(quota_root->node);
958 btrfs_free_tree_block(trans, quota_root, quota_root->node, 0, 1);
960 free_extent_buffer(quota_root->node);
961 free_extent_buffer(quota_root->commit_root);
964 mutex_unlock(&fs_info->qgroup_ioctl_lock);
968 static void qgroup_dirty(struct btrfs_fs_info *fs_info,
969 struct btrfs_qgroup *qgroup)
971 if (list_empty(&qgroup->dirty))
972 list_add(&qgroup->dirty, &fs_info->dirty_qgroups);
975 int btrfs_add_qgroup_relation(struct btrfs_trans_handle *trans,
976 struct btrfs_fs_info *fs_info, u64 src, u64 dst)
978 struct btrfs_root *quota_root;
979 struct btrfs_qgroup *parent;
980 struct btrfs_qgroup *member;
981 struct btrfs_qgroup_list *list;
984 mutex_lock(&fs_info->qgroup_ioctl_lock);
985 quota_root = fs_info->quota_root;
990 member = find_qgroup_rb(fs_info, src);
991 parent = find_qgroup_rb(fs_info, dst);
992 if (!member || !parent) {
997 /* check if such qgroup relation exist firstly */
998 list_for_each_entry(list, &member->groups, next_group) {
999 if (list->group == parent) {
1005 ret = add_qgroup_relation_item(trans, quota_root, src, dst);
1009 ret = add_qgroup_relation_item(trans, quota_root, dst, src);
1011 del_qgroup_relation_item(trans, quota_root, src, dst);
1015 spin_lock(&fs_info->qgroup_lock);
1016 ret = add_relation_rb(quota_root->fs_info, src, dst);
1017 spin_unlock(&fs_info->qgroup_lock);
1019 mutex_unlock(&fs_info->qgroup_ioctl_lock);
1023 int btrfs_del_qgroup_relation(struct btrfs_trans_handle *trans,
1024 struct btrfs_fs_info *fs_info, u64 src, u64 dst)
1026 struct btrfs_root *quota_root;
1027 struct btrfs_qgroup *parent;
1028 struct btrfs_qgroup *member;
1029 struct btrfs_qgroup_list *list;
1033 mutex_lock(&fs_info->qgroup_ioctl_lock);
1034 quota_root = fs_info->quota_root;
1040 member = find_qgroup_rb(fs_info, src);
1041 parent = find_qgroup_rb(fs_info, dst);
1042 if (!member || !parent) {
1047 /* check if such qgroup relation exist firstly */
1048 list_for_each_entry(list, &member->groups, next_group) {
1049 if (list->group == parent)
1055 ret = del_qgroup_relation_item(trans, quota_root, src, dst);
1056 err = del_qgroup_relation_item(trans, quota_root, dst, src);
1060 spin_lock(&fs_info->qgroup_lock);
1061 del_relation_rb(fs_info, src, dst);
1062 spin_unlock(&fs_info->qgroup_lock);
1064 mutex_unlock(&fs_info->qgroup_ioctl_lock);
1068 int btrfs_create_qgroup(struct btrfs_trans_handle *trans,
1069 struct btrfs_fs_info *fs_info, u64 qgroupid, char *name)
1071 struct btrfs_root *quota_root;
1072 struct btrfs_qgroup *qgroup;
1075 mutex_lock(&fs_info->qgroup_ioctl_lock);
1076 quota_root = fs_info->quota_root;
1081 qgroup = find_qgroup_rb(fs_info, qgroupid);
1087 ret = add_qgroup_item(trans, quota_root, qgroupid);
1091 spin_lock(&fs_info->qgroup_lock);
1092 qgroup = add_qgroup_rb(fs_info, qgroupid);
1093 spin_unlock(&fs_info->qgroup_lock);
1096 ret = PTR_ERR(qgroup);
1098 mutex_unlock(&fs_info->qgroup_ioctl_lock);
1102 int btrfs_remove_qgroup(struct btrfs_trans_handle *trans,
1103 struct btrfs_fs_info *fs_info, u64 qgroupid)
1105 struct btrfs_root *quota_root;
1106 struct btrfs_qgroup *qgroup;
1109 mutex_lock(&fs_info->qgroup_ioctl_lock);
1110 quota_root = fs_info->quota_root;
1116 qgroup = find_qgroup_rb(fs_info, qgroupid);
1121 /* check if there are no relations to this qgroup */
1122 if (!list_empty(&qgroup->groups) ||
1123 !list_empty(&qgroup->members)) {
1128 ret = del_qgroup_item(trans, quota_root, qgroupid);
1130 spin_lock(&fs_info->qgroup_lock);
1131 del_qgroup_rb(quota_root->fs_info, qgroupid);
1132 spin_unlock(&fs_info->qgroup_lock);
1134 mutex_unlock(&fs_info->qgroup_ioctl_lock);
1138 int btrfs_limit_qgroup(struct btrfs_trans_handle *trans,
1139 struct btrfs_fs_info *fs_info, u64 qgroupid,
1140 struct btrfs_qgroup_limit *limit)
1142 struct btrfs_root *quota_root;
1143 struct btrfs_qgroup *qgroup;
1146 mutex_lock(&fs_info->qgroup_ioctl_lock);
1147 quota_root = fs_info->quota_root;
1153 qgroup = find_qgroup_rb(fs_info, qgroupid);
1158 ret = update_qgroup_limit_item(trans, quota_root, qgroupid,
1159 limit->flags, limit->max_rfer,
1160 limit->max_excl, limit->rsv_rfer,
1163 fs_info->qgroup_flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
1164 printk(KERN_INFO "unable to update quota limit for %llu\n",
1168 spin_lock(&fs_info->qgroup_lock);
1169 qgroup->lim_flags = limit->flags;
1170 qgroup->max_rfer = limit->max_rfer;
1171 qgroup->max_excl = limit->max_excl;
1172 qgroup->rsv_rfer = limit->rsv_rfer;
1173 qgroup->rsv_excl = limit->rsv_excl;
1174 spin_unlock(&fs_info->qgroup_lock);
1176 mutex_unlock(&fs_info->qgroup_ioctl_lock);
1181 * btrfs_qgroup_record_ref is called when the ref is added or deleted. it puts
1182 * the modification into a list that's later used by btrfs_end_transaction to
1183 * pass the recorded modifications on to btrfs_qgroup_account_ref.
1185 int btrfs_qgroup_record_ref(struct btrfs_trans_handle *trans,
1186 struct btrfs_delayed_ref_node *node,
1187 struct btrfs_delayed_extent_op *extent_op)
1189 struct qgroup_update *u;
1191 BUG_ON(!trans->delayed_ref_elem.seq);
1192 u = kmalloc(sizeof(*u), GFP_NOFS);
1197 u->extent_op = extent_op;
1198 list_add_tail(&u->list, &trans->qgroup_ref_list);
1203 static int qgroup_account_ref_step1(struct btrfs_fs_info *fs_info,
1204 struct ulist *roots, struct ulist *tmp,
1207 struct ulist_node *unode;
1208 struct ulist_iterator uiter;
1209 struct ulist_node *tmp_unode;
1210 struct ulist_iterator tmp_uiter;
1211 struct btrfs_qgroup *qg;
1214 ULIST_ITER_INIT(&uiter);
1215 while ((unode = ulist_next(roots, &uiter))) {
1216 qg = find_qgroup_rb(fs_info, unode->val);
1221 /* XXX id not needed */
1222 ret = ulist_add(tmp, qg->qgroupid,
1223 (u64)(uintptr_t)qg, GFP_ATOMIC);
1226 ULIST_ITER_INIT(&tmp_uiter);
1227 while ((tmp_unode = ulist_next(tmp, &tmp_uiter))) {
1228 struct btrfs_qgroup_list *glist;
1230 qg = (struct btrfs_qgroup *)(uintptr_t)tmp_unode->aux;
1231 if (qg->refcnt < seq)
1232 qg->refcnt = seq + 1;
1236 list_for_each_entry(glist, &qg->groups, next_group) {
1237 ret = ulist_add(tmp, glist->group->qgroupid,
1238 (u64)(uintptr_t)glist->group,
1249 static int qgroup_account_ref_step2(struct btrfs_fs_info *fs_info,
1250 struct ulist *roots, struct ulist *tmp,
1251 u64 seq, int sgn, u64 num_bytes,
1252 struct btrfs_qgroup *qgroup)
1254 struct ulist_node *unode;
1255 struct ulist_iterator uiter;
1256 struct btrfs_qgroup *qg;
1257 struct btrfs_qgroup_list *glist;
1261 ret = ulist_add(tmp, qgroup->qgroupid, (uintptr_t)qgroup, GFP_ATOMIC);
1265 ULIST_ITER_INIT(&uiter);
1266 while ((unode = ulist_next(tmp, &uiter))) {
1267 qg = (struct btrfs_qgroup *)(uintptr_t)unode->aux;
1268 if (qg->refcnt < seq) {
1269 /* not visited by step 1 */
1270 qg->rfer += sgn * num_bytes;
1271 qg->rfer_cmpr += sgn * num_bytes;
1272 if (roots->nnodes == 0) {
1273 qg->excl += sgn * num_bytes;
1274 qg->excl_cmpr += sgn * num_bytes;
1276 qgroup_dirty(fs_info, qg);
1278 WARN_ON(qg->tag >= seq);
1281 list_for_each_entry(glist, &qg->groups, next_group) {
1282 ret = ulist_add(tmp, glist->group->qgroupid,
1283 (uintptr_t)glist->group, GFP_ATOMIC);
1292 static int qgroup_account_ref_step3(struct btrfs_fs_info *fs_info,
1293 struct ulist *roots, struct ulist *tmp,
1294 u64 seq, int sgn, u64 num_bytes)
1296 struct ulist_node *unode;
1297 struct ulist_iterator uiter;
1298 struct btrfs_qgroup *qg;
1299 struct ulist_node *tmp_unode;
1300 struct ulist_iterator tmp_uiter;
1303 ULIST_ITER_INIT(&uiter);
1304 while ((unode = ulist_next(roots, &uiter))) {
1305 qg = find_qgroup_rb(fs_info, unode->val);
1310 ret = ulist_add(tmp, qg->qgroupid, (uintptr_t)qg, GFP_ATOMIC);
1314 ULIST_ITER_INIT(&tmp_uiter);
1315 while ((tmp_unode = ulist_next(tmp, &tmp_uiter))) {
1316 struct btrfs_qgroup_list *glist;
1318 qg = (struct btrfs_qgroup *)(uintptr_t)tmp_unode->aux;
1322 if (qg->refcnt - seq == roots->nnodes) {
1323 qg->excl -= sgn * num_bytes;
1324 qg->excl_cmpr -= sgn * num_bytes;
1325 qgroup_dirty(fs_info, qg);
1328 list_for_each_entry(glist, &qg->groups, next_group) {
1329 ret = ulist_add(tmp, glist->group->qgroupid,
1330 (uintptr_t)glist->group,
1342 * btrfs_qgroup_account_ref is called for every ref that is added to or deleted
1343 * from the fs. First, all roots referencing the extent are searched, and
1344 * then the space is accounted accordingly to the different roots. The
1345 * accounting algorithm works in 3 steps documented inline.
1347 int btrfs_qgroup_account_ref(struct btrfs_trans_handle *trans,
1348 struct btrfs_fs_info *fs_info,
1349 struct btrfs_delayed_ref_node *node,
1350 struct btrfs_delayed_extent_op *extent_op)
1352 struct btrfs_key ins;
1353 struct btrfs_root *quota_root;
1355 struct btrfs_qgroup *qgroup;
1356 struct ulist *roots = NULL;
1361 if (!fs_info->quota_enabled)
1364 BUG_ON(!fs_info->quota_root);
1366 ins.objectid = node->bytenr;
1367 ins.offset = node->num_bytes;
1368 ins.type = BTRFS_EXTENT_ITEM_KEY;
1370 if (node->type == BTRFS_TREE_BLOCK_REF_KEY ||
1371 node->type == BTRFS_SHARED_BLOCK_REF_KEY) {
1372 struct btrfs_delayed_tree_ref *ref;
1373 ref = btrfs_delayed_node_to_tree_ref(node);
1374 ref_root = ref->root;
1375 } else if (node->type == BTRFS_EXTENT_DATA_REF_KEY ||
1376 node->type == BTRFS_SHARED_DATA_REF_KEY) {
1377 struct btrfs_delayed_data_ref *ref;
1378 ref = btrfs_delayed_node_to_data_ref(node);
1379 ref_root = ref->root;
1384 if (!is_fstree(ref_root)) {
1386 * non-fs-trees are not being accounted
1391 switch (node->action) {
1392 case BTRFS_ADD_DELAYED_REF:
1393 case BTRFS_ADD_DELAYED_EXTENT:
1395 seq = btrfs_tree_mod_seq_prev(node->seq);
1397 case BTRFS_DROP_DELAYED_REF:
1401 case BTRFS_UPDATE_DELAYED_HEAD:
1407 mutex_lock(&fs_info->qgroup_rescan_lock);
1408 if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) {
1409 if (fs_info->qgroup_rescan_progress.objectid <= node->bytenr) {
1410 mutex_unlock(&fs_info->qgroup_rescan_lock);
1414 mutex_unlock(&fs_info->qgroup_rescan_lock);
1417 * the delayed ref sequence number we pass depends on the direction of
1418 * the operation. for add operations, we pass
1419 * tree_mod_log_prev_seq(node->seq) to skip
1420 * the delayed ref's current sequence number, because we need the state
1421 * of the tree before the add operation. for delete operations, we pass
1422 * (node->seq) to include the delayed ref's current sequence number,
1423 * because we need the state of the tree after the delete operation.
1425 ret = btrfs_find_all_roots(trans, fs_info, node->bytenr, seq, &roots);
1429 spin_lock(&fs_info->qgroup_lock);
1431 quota_root = fs_info->quota_root;
1435 qgroup = find_qgroup_rb(fs_info, ref_root);
1440 * step 1: for each old ref, visit all nodes once and inc refcnt
1442 ulist_reinit(fs_info->qgroup_ulist);
1443 seq = fs_info->qgroup_seq;
1444 fs_info->qgroup_seq += roots->nnodes + 1; /* max refcnt */
1446 ret = qgroup_account_ref_step1(fs_info, roots, fs_info->qgroup_ulist,
1452 * step 2: walk from the new root
1454 ret = qgroup_account_ref_step2(fs_info, roots, fs_info->qgroup_ulist,
1455 seq, sgn, node->num_bytes, qgroup);
1460 * step 3: walk again from old refs
1462 ret = qgroup_account_ref_step3(fs_info, roots, fs_info->qgroup_ulist,
1463 seq, sgn, node->num_bytes);
1468 spin_unlock(&fs_info->qgroup_lock);
1475 * called from commit_transaction. Writes all changed qgroups to disk.
1477 int btrfs_run_qgroups(struct btrfs_trans_handle *trans,
1478 struct btrfs_fs_info *fs_info)
1480 struct btrfs_root *quota_root = fs_info->quota_root;
1482 int start_rescan_worker = 0;
1487 if (!fs_info->quota_enabled && fs_info->pending_quota_state)
1488 start_rescan_worker = 1;
1490 fs_info->quota_enabled = fs_info->pending_quota_state;
1492 spin_lock(&fs_info->qgroup_lock);
1493 while (!list_empty(&fs_info->dirty_qgroups)) {
1494 struct btrfs_qgroup *qgroup;
1495 qgroup = list_first_entry(&fs_info->dirty_qgroups,
1496 struct btrfs_qgroup, dirty);
1497 list_del_init(&qgroup->dirty);
1498 spin_unlock(&fs_info->qgroup_lock);
1499 ret = update_qgroup_info_item(trans, quota_root, qgroup);
1501 fs_info->qgroup_flags |=
1502 BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
1503 spin_lock(&fs_info->qgroup_lock);
1505 if (fs_info->quota_enabled)
1506 fs_info->qgroup_flags |= BTRFS_QGROUP_STATUS_FLAG_ON;
1508 fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_ON;
1509 spin_unlock(&fs_info->qgroup_lock);
1511 ret = update_qgroup_status_item(trans, fs_info, quota_root);
1513 fs_info->qgroup_flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
1515 if (!ret && start_rescan_worker) {
1516 ret = qgroup_rescan_init(fs_info, 0, 1);
1518 qgroup_rescan_zero_tracking(fs_info);
1519 btrfs_queue_worker(&fs_info->qgroup_rescan_workers,
1520 &fs_info->qgroup_rescan_work);
1531 * copy the acounting information between qgroups. This is necessary when a
1532 * snapshot or a subvolume is created
1534 int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans,
1535 struct btrfs_fs_info *fs_info, u64 srcid, u64 objectid,
1536 struct btrfs_qgroup_inherit *inherit)
1541 struct btrfs_root *quota_root = fs_info->quota_root;
1542 struct btrfs_qgroup *srcgroup;
1543 struct btrfs_qgroup *dstgroup;
1547 mutex_lock(&fs_info->qgroup_ioctl_lock);
1548 if (!fs_info->quota_enabled)
1557 i_qgroups = (u64 *)(inherit + 1);
1558 nums = inherit->num_qgroups + 2 * inherit->num_ref_copies +
1559 2 * inherit->num_excl_copies;
1560 for (i = 0; i < nums; ++i) {
1561 srcgroup = find_qgroup_rb(fs_info, *i_qgroups);
1571 * create a tracking group for the subvol itself
1573 ret = add_qgroup_item(trans, quota_root, objectid);
1577 if (inherit && inherit->flags & BTRFS_QGROUP_INHERIT_SET_LIMITS) {
1578 ret = update_qgroup_limit_item(trans, quota_root, objectid,
1580 inherit->lim.max_rfer,
1581 inherit->lim.max_excl,
1582 inherit->lim.rsv_rfer,
1583 inherit->lim.rsv_excl);
1589 struct btrfs_root *srcroot;
1590 struct btrfs_key srckey;
1593 srckey.objectid = srcid;
1594 srckey.type = BTRFS_ROOT_ITEM_KEY;
1595 srckey.offset = (u64)-1;
1596 srcroot = btrfs_read_fs_root_no_name(fs_info, &srckey);
1597 if (IS_ERR(srcroot)) {
1598 ret = PTR_ERR(srcroot);
1603 srcroot_level = btrfs_header_level(srcroot->node);
1604 level_size = btrfs_level_size(srcroot, srcroot_level);
1609 * add qgroup to all inherited groups
1612 i_qgroups = (u64 *)(inherit + 1);
1613 for (i = 0; i < inherit->num_qgroups; ++i) {
1614 ret = add_qgroup_relation_item(trans, quota_root,
1615 objectid, *i_qgroups);
1618 ret = add_qgroup_relation_item(trans, quota_root,
1619 *i_qgroups, objectid);
1627 spin_lock(&fs_info->qgroup_lock);
1629 dstgroup = add_qgroup_rb(fs_info, objectid);
1630 if (IS_ERR(dstgroup)) {
1631 ret = PTR_ERR(dstgroup);
1636 srcgroup = find_qgroup_rb(fs_info, srcid);
1639 dstgroup->rfer = srcgroup->rfer - level_size;
1640 dstgroup->rfer_cmpr = srcgroup->rfer_cmpr - level_size;
1641 srcgroup->excl = level_size;
1642 srcgroup->excl_cmpr = level_size;
1643 qgroup_dirty(fs_info, dstgroup);
1644 qgroup_dirty(fs_info, srcgroup);
1650 i_qgroups = (u64 *)(inherit + 1);
1651 for (i = 0; i < inherit->num_qgroups; ++i) {
1652 ret = add_relation_rb(quota_root->fs_info, objectid,
1659 for (i = 0; i < inherit->num_ref_copies; ++i) {
1660 struct btrfs_qgroup *src;
1661 struct btrfs_qgroup *dst;
1663 src = find_qgroup_rb(fs_info, i_qgroups[0]);
1664 dst = find_qgroup_rb(fs_info, i_qgroups[1]);
1671 dst->rfer = src->rfer - level_size;
1672 dst->rfer_cmpr = src->rfer_cmpr - level_size;
1675 for (i = 0; i < inherit->num_excl_copies; ++i) {
1676 struct btrfs_qgroup *src;
1677 struct btrfs_qgroup *dst;
1679 src = find_qgroup_rb(fs_info, i_qgroups[0]);
1680 dst = find_qgroup_rb(fs_info, i_qgroups[1]);
1687 dst->excl = src->excl + level_size;
1688 dst->excl_cmpr = src->excl_cmpr + level_size;
1693 spin_unlock(&fs_info->qgroup_lock);
1695 mutex_unlock(&fs_info->qgroup_ioctl_lock);
1700 * reserve some space for a qgroup and all its parents. The reservation takes
1701 * place with start_transaction or dealloc_reserve, similar to ENOSPC
1702 * accounting. If not enough space is available, EDQUOT is returned.
1703 * We assume that the requested space is new for all qgroups.
1705 int btrfs_qgroup_reserve(struct btrfs_root *root, u64 num_bytes)
1707 struct btrfs_root *quota_root;
1708 struct btrfs_qgroup *qgroup;
1709 struct btrfs_fs_info *fs_info = root->fs_info;
1710 u64 ref_root = root->root_key.objectid;
1712 struct ulist_node *unode;
1713 struct ulist_iterator uiter;
1715 if (!is_fstree(ref_root))
1721 spin_lock(&fs_info->qgroup_lock);
1722 quota_root = fs_info->quota_root;
1726 qgroup = find_qgroup_rb(fs_info, ref_root);
1731 * in a first step, we check all affected qgroups if any limits would
1734 ulist_reinit(fs_info->qgroup_ulist);
1735 ret = ulist_add(fs_info->qgroup_ulist, qgroup->qgroupid,
1736 (uintptr_t)qgroup, GFP_ATOMIC);
1739 ULIST_ITER_INIT(&uiter);
1740 while ((unode = ulist_next(fs_info->qgroup_ulist, &uiter))) {
1741 struct btrfs_qgroup *qg;
1742 struct btrfs_qgroup_list *glist;
1744 qg = (struct btrfs_qgroup *)(uintptr_t)unode->aux;
1746 if ((qg->lim_flags & BTRFS_QGROUP_LIMIT_MAX_RFER) &&
1747 qg->reserved + (s64)qg->rfer + num_bytes >
1753 if ((qg->lim_flags & BTRFS_QGROUP_LIMIT_MAX_EXCL) &&
1754 qg->reserved + (s64)qg->excl + num_bytes >
1760 list_for_each_entry(glist, &qg->groups, next_group) {
1761 ret = ulist_add(fs_info->qgroup_ulist,
1762 glist->group->qgroupid,
1763 (uintptr_t)glist->group, GFP_ATOMIC);
1770 * no limits exceeded, now record the reservation into all qgroups
1772 ULIST_ITER_INIT(&uiter);
1773 while ((unode = ulist_next(fs_info->qgroup_ulist, &uiter))) {
1774 struct btrfs_qgroup *qg;
1776 qg = (struct btrfs_qgroup *)(uintptr_t)unode->aux;
1778 qg->reserved += num_bytes;
1782 spin_unlock(&fs_info->qgroup_lock);
1786 void btrfs_qgroup_free(struct btrfs_root *root, u64 num_bytes)
1788 struct btrfs_root *quota_root;
1789 struct btrfs_qgroup *qgroup;
1790 struct btrfs_fs_info *fs_info = root->fs_info;
1791 struct ulist_node *unode;
1792 struct ulist_iterator uiter;
1793 u64 ref_root = root->root_key.objectid;
1796 if (!is_fstree(ref_root))
1802 spin_lock(&fs_info->qgroup_lock);
1804 quota_root = fs_info->quota_root;
1808 qgroup = find_qgroup_rb(fs_info, ref_root);
1812 ulist_reinit(fs_info->qgroup_ulist);
1813 ret = ulist_add(fs_info->qgroup_ulist, qgroup->qgroupid,
1814 (uintptr_t)qgroup, GFP_ATOMIC);
1817 ULIST_ITER_INIT(&uiter);
1818 while ((unode = ulist_next(fs_info->qgroup_ulist, &uiter))) {
1819 struct btrfs_qgroup *qg;
1820 struct btrfs_qgroup_list *glist;
1822 qg = (struct btrfs_qgroup *)(uintptr_t)unode->aux;
1824 qg->reserved -= num_bytes;
1826 list_for_each_entry(glist, &qg->groups, next_group) {
1827 ret = ulist_add(fs_info->qgroup_ulist,
1828 glist->group->qgroupid,
1829 (uintptr_t)glist->group, GFP_ATOMIC);
1836 spin_unlock(&fs_info->qgroup_lock);
1839 void assert_qgroups_uptodate(struct btrfs_trans_handle *trans)
1841 if (list_empty(&trans->qgroup_ref_list) && !trans->delayed_ref_elem.seq)
1843 pr_err("btrfs: qgroups not uptodate in trans handle %p: list is%s empty, seq is %#x.%x\n",
1844 trans, list_empty(&trans->qgroup_ref_list) ? "" : " not",
1845 (u32)(trans->delayed_ref_elem.seq >> 32),
1846 (u32)trans->delayed_ref_elem.seq);
1851 * returns < 0 on error, 0 when more leafs are to be scanned.
1852 * returns 1 when done, 2 when done and FLAG_INCONSISTENT was cleared.
1855 qgroup_rescan_leaf(struct btrfs_fs_info *fs_info, struct btrfs_path *path,
1856 struct btrfs_trans_handle *trans, struct ulist *tmp,
1857 struct extent_buffer *scratch_leaf)
1859 struct btrfs_key found;
1860 struct ulist *roots = NULL;
1861 struct ulist_node *unode;
1862 struct ulist_iterator uiter;
1863 struct seq_list tree_mod_seq_elem = {};
1868 path->leave_spinning = 1;
1869 mutex_lock(&fs_info->qgroup_rescan_lock);
1870 ret = btrfs_search_slot_for_read(fs_info->extent_root,
1871 &fs_info->qgroup_rescan_progress,
1874 pr_debug("current progress key (%llu %u %llu), search_slot ret %d\n",
1875 fs_info->qgroup_rescan_progress.objectid,
1876 fs_info->qgroup_rescan_progress.type,
1877 fs_info->qgroup_rescan_progress.offset, ret);
1881 * The rescan is about to end, we will not be scanning any
1882 * further blocks. We cannot unset the RESCAN flag here, because
1883 * we want to commit the transaction if everything went well.
1884 * To make the live accounting work in this phase, we set our
1885 * scan progress pointer such that every real extent objectid
1888 fs_info->qgroup_rescan_progress.objectid = (u64)-1;
1889 btrfs_release_path(path);
1890 mutex_unlock(&fs_info->qgroup_rescan_lock);
1894 btrfs_item_key_to_cpu(path->nodes[0], &found,
1895 btrfs_header_nritems(path->nodes[0]) - 1);
1896 fs_info->qgroup_rescan_progress.objectid = found.objectid + 1;
1898 btrfs_get_tree_mod_seq(fs_info, &tree_mod_seq_elem);
1899 memcpy(scratch_leaf, path->nodes[0], sizeof(*scratch_leaf));
1900 slot = path->slots[0];
1901 btrfs_release_path(path);
1902 mutex_unlock(&fs_info->qgroup_rescan_lock);
1904 for (; slot < btrfs_header_nritems(scratch_leaf); ++slot) {
1905 btrfs_item_key_to_cpu(scratch_leaf, &found, slot);
1906 if (found.type != BTRFS_EXTENT_ITEM_KEY)
1908 ret = btrfs_find_all_roots(trans, fs_info, found.objectid,
1909 tree_mod_seq_elem.seq, &roots);
1912 spin_lock(&fs_info->qgroup_lock);
1913 seq = fs_info->qgroup_seq;
1914 fs_info->qgroup_seq += roots->nnodes + 1; /* max refcnt */
1916 ret = qgroup_account_ref_step1(fs_info, roots, tmp, seq);
1918 spin_unlock(&fs_info->qgroup_lock);
1924 * step2 of btrfs_qgroup_account_ref works from a single root,
1925 * we're doing all at once here.
1928 ULIST_ITER_INIT(&uiter);
1929 while ((unode = ulist_next(roots, &uiter))) {
1930 struct btrfs_qgroup *qg;
1932 qg = find_qgroup_rb(fs_info, unode->val);
1936 ret = ulist_add(tmp, qg->qgroupid, (uintptr_t)qg,
1939 spin_unlock(&fs_info->qgroup_lock);
1945 /* this loop is similar to step 2 of btrfs_qgroup_account_ref */
1946 ULIST_ITER_INIT(&uiter);
1947 while ((unode = ulist_next(tmp, &uiter))) {
1948 struct btrfs_qgroup *qg;
1949 struct btrfs_qgroup_list *glist;
1951 qg = (struct btrfs_qgroup *)(uintptr_t) unode->aux;
1952 qg->rfer += found.offset;
1953 qg->rfer_cmpr += found.offset;
1954 WARN_ON(qg->tag >= seq);
1955 if (qg->refcnt - seq == roots->nnodes) {
1956 qg->excl += found.offset;
1957 qg->excl_cmpr += found.offset;
1959 qgroup_dirty(fs_info, qg);
1961 list_for_each_entry(glist, &qg->groups, next_group) {
1962 ret = ulist_add(tmp, glist->group->qgroupid,
1963 (uintptr_t)glist->group,
1966 spin_unlock(&fs_info->qgroup_lock);
1973 spin_unlock(&fs_info->qgroup_lock);
1979 btrfs_put_tree_mod_seq(fs_info, &tree_mod_seq_elem);
1984 static void btrfs_qgroup_rescan_worker(struct btrfs_work *work)
1986 struct btrfs_fs_info *fs_info = container_of(work, struct btrfs_fs_info,
1987 qgroup_rescan_work);
1988 struct btrfs_path *path;
1989 struct btrfs_trans_handle *trans = NULL;
1990 struct ulist *tmp = NULL;
1991 struct extent_buffer *scratch_leaf = NULL;
1994 path = btrfs_alloc_path();
1997 tmp = ulist_alloc(GFP_NOFS);
2000 scratch_leaf = kmalloc(sizeof(*scratch_leaf), GFP_NOFS);
2006 trans = btrfs_start_transaction(fs_info->fs_root, 0);
2007 if (IS_ERR(trans)) {
2008 err = PTR_ERR(trans);
2011 if (!fs_info->quota_enabled) {
2014 err = qgroup_rescan_leaf(fs_info, path, trans,
2018 btrfs_commit_transaction(trans, fs_info->fs_root);
2020 btrfs_end_transaction(trans, fs_info->fs_root);
2024 kfree(scratch_leaf);
2026 btrfs_free_path(path);
2028 mutex_lock(&fs_info->qgroup_rescan_lock);
2029 fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_RESCAN;
2032 fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT) {
2033 fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
2034 } else if (err < 0) {
2035 fs_info->qgroup_flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
2037 mutex_unlock(&fs_info->qgroup_rescan_lock);
2040 pr_info("btrfs: qgroup scan completed%s\n",
2041 err == 2 ? " (inconsistency flag cleared)" : "");
2043 pr_err("btrfs: qgroup scan failed with %d\n", err);
2046 complete_all(&fs_info->qgroup_rescan_completion);
2050 * Checks that (a) no rescan is running and (b) quota is enabled. Allocates all
2051 * memory required for the rescan context.
2054 qgroup_rescan_init(struct btrfs_fs_info *fs_info, u64 progress_objectid,
2060 (!(fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) ||
2061 !(fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_ON))) {
2066 mutex_lock(&fs_info->qgroup_rescan_lock);
2067 spin_lock(&fs_info->qgroup_lock);
2070 if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN)
2072 else if (!(fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_ON))
2076 spin_unlock(&fs_info->qgroup_lock);
2077 mutex_unlock(&fs_info->qgroup_rescan_lock);
2081 fs_info->qgroup_flags |= BTRFS_QGROUP_STATUS_FLAG_RESCAN;
2084 memset(&fs_info->qgroup_rescan_progress, 0,
2085 sizeof(fs_info->qgroup_rescan_progress));
2086 fs_info->qgroup_rescan_progress.objectid = progress_objectid;
2088 spin_unlock(&fs_info->qgroup_lock);
2089 mutex_unlock(&fs_info->qgroup_rescan_lock);
2091 init_completion(&fs_info->qgroup_rescan_completion);
2093 memset(&fs_info->qgroup_rescan_work, 0,
2094 sizeof(fs_info->qgroup_rescan_work));
2095 fs_info->qgroup_rescan_work.func = btrfs_qgroup_rescan_worker;
2099 pr_info("btrfs: qgroup_rescan_init failed with %d\n", ret);
2107 qgroup_rescan_zero_tracking(struct btrfs_fs_info *fs_info)
2110 struct btrfs_qgroup *qgroup;
2112 spin_lock(&fs_info->qgroup_lock);
2113 /* clear all current qgroup tracking information */
2114 for (n = rb_first(&fs_info->qgroup_tree); n; n = rb_next(n)) {
2115 qgroup = rb_entry(n, struct btrfs_qgroup, node);
2117 qgroup->rfer_cmpr = 0;
2119 qgroup->excl_cmpr = 0;
2121 spin_unlock(&fs_info->qgroup_lock);
2125 btrfs_qgroup_rescan(struct btrfs_fs_info *fs_info)
2128 struct btrfs_trans_handle *trans;
2130 ret = qgroup_rescan_init(fs_info, 0, 1);
2135 * We have set the rescan_progress to 0, which means no more
2136 * delayed refs will be accounted by btrfs_qgroup_account_ref.
2137 * However, btrfs_qgroup_account_ref may be right after its call
2138 * to btrfs_find_all_roots, in which case it would still do the
2140 * To solve this, we're committing the transaction, which will
2141 * ensure we run all delayed refs and only after that, we are
2142 * going to clear all tracking information for a clean start.
2145 trans = btrfs_join_transaction(fs_info->fs_root);
2146 if (IS_ERR(trans)) {
2147 fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_RESCAN;
2148 return PTR_ERR(trans);
2150 ret = btrfs_commit_transaction(trans, fs_info->fs_root);
2152 fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_RESCAN;
2156 qgroup_rescan_zero_tracking(fs_info);
2158 btrfs_queue_worker(&fs_info->qgroup_rescan_workers,
2159 &fs_info->qgroup_rescan_work);
2164 int btrfs_qgroup_wait_for_completion(struct btrfs_fs_info *fs_info)
2169 mutex_lock(&fs_info->qgroup_rescan_lock);
2170 spin_lock(&fs_info->qgroup_lock);
2171 running = fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN;
2172 spin_unlock(&fs_info->qgroup_lock);
2173 mutex_unlock(&fs_info->qgroup_rescan_lock);
2176 ret = wait_for_completion_interruptible(
2177 &fs_info->qgroup_rescan_completion);
2183 * this is only called from open_ctree where we're still single threaded, thus
2184 * locking is omitted here.
2187 btrfs_qgroup_rescan_resume(struct btrfs_fs_info *fs_info)
2189 if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN)
2190 btrfs_queue_worker(&fs_info->qgroup_rescan_workers,
2191 &fs_info->qgroup_rescan_work);