Merging upstream version 6.7.7.

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

1 files changed, 542 insertions, 251 deletions

diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c
index a006f5160e..bbfa44b89b 100644
--- a/fs/btrfs/qgroup.c
+++ b/fs/btrfs/qgroup.c
@@ -30,6 +30,25 @@
 #include "root-tree.h"
 #include "tree-checker.h"
 
+enum btrfs_qgroup_mode btrfs_qgroup_mode(struct btrfs_fs_info *fs_info)
+{
+	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags))
+		return BTRFS_QGROUP_MODE_DISABLED;
+	if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_SIMPLE_MODE)
+		return BTRFS_QGROUP_MODE_SIMPLE;
+	return BTRFS_QGROUP_MODE_FULL;
+}
+
+bool btrfs_qgroup_enabled(struct btrfs_fs_info *fs_info)
+{
+	return btrfs_qgroup_mode(fs_info) != BTRFS_QGROUP_MODE_DISABLED;
+}
+
+bool btrfs_qgroup_full_accounting(struct btrfs_fs_info *fs_info)
+{
+	return btrfs_qgroup_mode(fs_info) == BTRFS_QGROUP_MODE_FULL;
+}
+
 /*
  * Helpers to access qgroup reservation
  *
@@ -146,16 +165,6 @@ struct btrfs_qgroup_list {
 	struct btrfs_qgroup *member;
 };
 
-static inline u64 qgroup_to_aux(struct btrfs_qgroup *qg)
-{
-	return (u64)(uintptr_t)qg;
-}
-
-static inline struct btrfs_qgroup* unode_aux_to_qgroup(struct ulist_node *n)
-{
-	return (struct btrfs_qgroup *)(uintptr_t)n->aux;
-}
-
 static int
 qgroup_rescan_init(struct btrfs_fs_info *fs_info, u64 progress_objectid,
 		   int init_flags);
@@ -180,35 +189,46 @@ static struct btrfs_qgroup *find_qgroup_rb(struct btrfs_fs_info *fs_info,
 	return NULL;
 }
 
-/* must be called with qgroup_lock held */
+/*
+ * Add qgroup to the filesystem's qgroup tree.
+ *
+ * Must be called with qgroup_lock held and @prealloc preallocated.
+ *
+ * The control on the lifespan of @prealloc would be transfered to this
+ * function, thus caller should no longer touch @prealloc.
+ */
 static struct btrfs_qgroup *add_qgroup_rb(struct btrfs_fs_info *fs_info,
+					  struct btrfs_qgroup *prealloc,
 					  u64 qgroupid)
 {
 	struct rb_node **p = &fs_info->qgroup_tree.rb_node;
 	struct rb_node *parent = NULL;
 	struct btrfs_qgroup *qgroup;
 
+	/* Caller must have pre-allocated @prealloc. */
+	ASSERT(prealloc);
+
 	while (*p) {
 		parent = *p;
 		qgroup = rb_entry(parent, struct btrfs_qgroup, node);
 
-		if (qgroup->qgroupid < qgroupid)
+		if (qgroup->qgroupid < qgroupid) {
 			p = &(*p)->rb_left;
-		else if (qgroup->qgroupid > qgroupid)
+		} else if (qgroup->qgroupid > qgroupid) {
 			p = &(*p)->rb_right;
-		else
+		} else {
+			kfree(prealloc);
 			return qgroup;
+		}
 	}
 
-	qgroup = kzalloc(sizeof(*qgroup), GFP_ATOMIC);
-	if (!qgroup)
-		return ERR_PTR(-ENOMEM);
-
+	qgroup = prealloc;
 	qgroup->qgroupid = qgroupid;
 	INIT_LIST_HEAD(&qgroup->groups);
 	INIT_LIST_HEAD(&qgroup->members);
 	INIT_LIST_HEAD(&qgroup->dirty);
 	INIT_LIST_HEAD(&qgroup->iterator);
+	INIT_LIST_HEAD(&qgroup->nested_iterator);
 
 	rb_link_node(&qgroup->node, parent, p);
 	rb_insert_color(&qgroup->node, &fs_info->qgroup_tree);
@@ -255,27 +275,26 @@ static int del_qgroup_rb(struct btrfs_fs_info *fs_info, u64 qgroupid)
 /*
  * Add relation specified by two qgroups.
  *
- * Must be called with qgroup_lock held.
+ * Must be called with qgroup_lock held, the ownership of @prealloc is
+ * transferred to this function and caller should not touch it anymore.
  *
  * Return: 0        on success
  *         -ENOENT  if one of the qgroups is NULL
  *         <0       other errors
  */
-static int __add_relation_rb(struct btrfs_qgroup *member, struct btrfs_qgroup *parent)
+static int __add_relation_rb(struct btrfs_qgroup_list *prealloc,
+			     struct btrfs_qgroup *member,
+			     struct btrfs_qgroup *parent)
 {
-	struct btrfs_qgroup_list *list;
-
-	if (!member || !parent)
+	if (!member || !parent) {
+		kfree(prealloc);
 		return -ENOENT;
+	}
 
-	list = kzalloc(sizeof(*list), GFP_ATOMIC);
-	if (!list)
-		return -ENOMEM;
-
-	list->group = parent;
-	list->member = member;
-	list_add_tail(&list->next_group, &member->groups);
-	list_add_tail(&list->next_member, &parent->members);
+	prealloc->group = parent;
+	prealloc->member = member;
+	list_add_tail(&prealloc->next_group, &member->groups);
+	list_add_tail(&prealloc->next_member, &parent->members);
 
 	return 0;
 }
@@ -289,7 +308,9 @@ static int __add_relation_rb(struct btrfs_qgroup *member, struct btrfs_qgroup *p
  *         -ENOENT  if one of the ids does not exist
  *         <0       other errors
  */
-static int add_relation_rb(struct btrfs_fs_info *fs_info, u64 memberid, u64 parentid)
+static int add_relation_rb(struct btrfs_fs_info *fs_info,
+			   struct btrfs_qgroup_list *prealloc,
+			   u64 memberid, u64 parentid)
 {
 	struct btrfs_qgroup *member;
 	struct btrfs_qgroup *parent;
@@ -297,7 +318,7 @@ static int add_relation_rb(struct btrfs_fs_info *fs_info, u64 memberid, u64 pare
 	member = find_qgroup_rb(fs_info, memberid);
 	parent = find_qgroup_rb(fs_info, parentid);
 
-	return __add_relation_rb(member, parent);
+	return __add_relation_rb(prealloc, member, parent);
 }
 
 /* Must be called with qgroup_lock held */
@@ -341,11 +362,22 @@ int btrfs_verify_qgroup_counts(struct btrfs_fs_info *fs_info, u64 qgroupid,
 
 static void qgroup_mark_inconsistent(struct btrfs_fs_info *fs_info)
 {
+	if (btrfs_qgroup_mode(fs_info) == BTRFS_QGROUP_MODE_SIMPLE)
+		return;
 	fs_info->qgroup_flags |= (BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT |
 				  BTRFS_QGROUP_RUNTIME_FLAG_CANCEL_RESCAN |
 				  BTRFS_QGROUP_RUNTIME_FLAG_NO_ACCOUNTING);
 }
 
+static void qgroup_read_enable_gen(struct btrfs_fs_info *fs_info,
+				   struct extent_buffer *leaf, int slot,
+				   struct btrfs_qgroup_status_item *ptr)
+{
+	ASSERT(btrfs_fs_incompat(fs_info, SIMPLE_QUOTA));
+	ASSERT(btrfs_item_size(leaf, slot) >= sizeof(*ptr));
+	fs_info->qgroup_enable_gen = btrfs_qgroup_status_enable_gen(leaf, ptr);
+}
+
 /*
  * The full config is read in one go, only called from open_ctree()
  * It doesn't use any locking, as at this point we're still single-threaded
@@ -362,7 +394,7 @@ int btrfs_read_qgroup_config(struct btrfs_fs_info *fs_info)
 	u64 flags = 0;
 	u64 rescan_progress = 0;
 
-	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags))
+	if (!fs_info->quota_root)
 		return 0;
 
 	fs_info->qgroup_ulist = ulist_alloc(GFP_KERNEL);
@@ -412,14 +444,14 @@ int btrfs_read_qgroup_config(struct btrfs_fs_info *fs_info)
 				 "old qgroup version, quota disabled");
 				goto out;
 			}
-			if (btrfs_qgroup_status_generation(l, ptr) !=
-			    fs_info->generation) {
+			fs_info->qgroup_flags = btrfs_qgroup_status_flags(l, ptr);
+			if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_SIMPLE_MODE) {
+				qgroup_read_enable_gen(fs_info, l, slot, ptr);
+			} else if (btrfs_qgroup_status_generation(l, ptr) != fs_info->generation) {
 				qgroup_mark_inconsistent(fs_info);
 				btrfs_err(fs_info,
 					"qgroup generation mismatch, marked as inconsistent");
 			}
-			fs_info->qgroup_flags = btrfs_qgroup_status_flags(l,
-									  ptr);
 			rescan_progress = btrfs_qgroup_status_rescan(l, ptr);
 			goto next1;
 		}
@@ -435,11 +467,14 @@ int btrfs_read_qgroup_config(struct btrfs_fs_info *fs_info)
 			qgroup_mark_inconsistent(fs_info);
 		}
 		if (!qgroup) {
-			qgroup = add_qgroup_rb(fs_info, found_key.offset);
-			if (IS_ERR(qgroup)) {
-				ret = PTR_ERR(qgroup);
+			struct btrfs_qgroup *prealloc;
+
+			prealloc = kzalloc(sizeof(*prealloc), GFP_KERNEL);
+			if (!prealloc) {
+				ret = -ENOMEM;
 				goto out;
 			}
+			qgroup = add_qgroup_rb(fs_info, prealloc, found_key.offset);
 		}
 		ret = btrfs_sysfs_add_one_qgroup(fs_info, qgroup);
 		if (ret < 0)
@@ -490,6 +525,8 @@ next1:
 	if (ret)
 		goto out;
 	while (1) {
+		struct btrfs_qgroup_list *list = NULL;
+
 		slot = path->slots[0];
 		l = path->nodes[0];
 		btrfs_item_key_to_cpu(l, &found_key, slot);
@@ -503,8 +540,14 @@ next1:
 			goto next2;
 		}
 
-		ret = add_relation_rb(fs_info, found_key.objectid,
+		list = kzalloc(sizeof(*list), GFP_KERNEL);
+		if (!list) {
+			ret = -ENOMEM;
+			goto out;
+		}
+		ret = add_relation_rb(fs_info, list, found_key.objectid,
 				      found_key.offset);
+		list = NULL;
 		if (ret == -ENOENT) {
 			btrfs_warn(fs_info,
 				"orphan qgroup relation 0x%llx->0x%llx",
@@ -523,13 +566,12 @@ next2:
 out:
 	btrfs_free_path(path);
 	fs_info->qgroup_flags |= flags;
-	if (!(fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_ON))
-		clear_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags);
-	else if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN &&
-		 ret >= 0)
-		ret = qgroup_rescan_init(fs_info, rescan_progress, 0);
-
-	if (ret < 0) {
+	if (ret >= 0) {
+		if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_ON)
+			set_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags);
+		if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN)
+			ret = qgroup_rescan_init(fs_info, rescan_progress, 0);
+	} else {
 		ulist_free(fs_info->qgroup_ulist);
 		fs_info->qgroup_ulist = NULL;
 		fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_RESCAN;
@@ -551,7 +593,7 @@ bool btrfs_check_quota_leak(struct btrfs_fs_info *fs_info)
 	struct rb_node *node;
 	bool ret = false;
 
-	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags))
+	if (btrfs_qgroup_mode(fs_info) == BTRFS_QGROUP_MODE_DISABLED)
 		return ret;
 	/*
 	 * Since we're unmounting, there is no race and no need to grab qgroup
@@ -950,7 +992,8 @@ out:
 	return ret;
 }
 
-int btrfs_quota_enable(struct btrfs_fs_info *fs_info)
+int btrfs_quota_enable(struct btrfs_fs_info *fs_info,
+		       struct btrfs_ioctl_quota_ctl_args *quota_ctl_args)
 {
 	struct btrfs_root *quota_root;
 	struct btrfs_root *tree_root = fs_info->tree_root;
@@ -960,8 +1003,10 @@ int btrfs_quota_enable(struct btrfs_fs_info *fs_info)
 	struct btrfs_key key;
 	struct btrfs_key found_key;
 	struct btrfs_qgroup *qgroup = NULL;
+	struct btrfs_qgroup *prealloc = NULL;
 	struct btrfs_trans_handle *trans = NULL;
 	struct ulist *ulist = NULL;
+	const bool simple = (quota_ctl_args->cmd == BTRFS_QUOTA_CTL_ENABLE_SIMPLE_QUOTA);
 	int ret = 0;
 	int slot;
 
@@ -1064,8 +1109,13 @@ int btrfs_quota_enable(struct btrfs_fs_info *fs_info)
 				 struct btrfs_qgroup_status_item);
 	btrfs_set_qgroup_status_generation(leaf, ptr, trans->transid);
 	btrfs_set_qgroup_status_version(leaf, ptr, BTRFS_QGROUP_STATUS_VERSION);
-	fs_info->qgroup_flags = BTRFS_QGROUP_STATUS_FLAG_ON |
-				BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
+	fs_info->qgroup_flags = BTRFS_QGROUP_STATUS_FLAG_ON;
+	if (simple) {
+		fs_info->qgroup_flags |= BTRFS_QGROUP_STATUS_FLAG_SIMPLE_MODE;
+		btrfs_set_qgroup_status_enable_gen(leaf, ptr, trans->transid);
+	} else {
+		fs_info->qgroup_flags |= BTRFS_QGROUP_STATUS_FLAG_INCONSISTENT;
+	}
 	btrfs_set_qgroup_status_flags(leaf, ptr, fs_info->qgroup_flags &
 				      BTRFS_QGROUP_STATUS_FLAGS_MASK);
 	btrfs_set_qgroup_status_rescan(leaf, ptr, 0);
@@ -1095,6 +1145,15 @@ int btrfs_quota_enable(struct btrfs_fs_info *fs_info)
 			/* Release locks on tree_root before we access quota_root */
 			btrfs_release_path(path);
 
+			/* We should not have a stray @prealloc pointer. */
+			ASSERT(prealloc == NULL);
+			prealloc = kzalloc(sizeof(*prealloc), GFP_NOFS);
+			if (!prealloc) {
+				ret = -ENOMEM;
+				btrfs_abort_transaction(trans, ret);
+				goto out_free_path;
+			}
+
 			ret = add_qgroup_item(trans, quota_root,
 					      found_key.offset);
 			if (ret) {
@@ -1102,7 +1161,8 @@ int btrfs_quota_enable(struct btrfs_fs_info *fs_info)
 				goto out_free_path;
 			}
 
-			qgroup = add_qgroup_rb(fs_info, found_key.offset);
+			qgroup = add_qgroup_rb(fs_info, prealloc, found_key.offset);
+			prealloc = NULL;
 			if (IS_ERR(qgroup)) {
 				ret = PTR_ERR(qgroup);
 				btrfs_abort_transaction(trans, ret);
@@ -1145,18 +1205,22 @@ out_add_root:
 		goto out_free_path;
 	}
 
-	qgroup = add_qgroup_rb(fs_info, BTRFS_FS_TREE_OBJECTID);
-	if (IS_ERR(qgroup)) {
-		ret = PTR_ERR(qgroup);
-		btrfs_abort_transaction(trans, ret);
+	ASSERT(prealloc == NULL);
+	prealloc = kzalloc(sizeof(*prealloc), GFP_NOFS);
+	if (!prealloc) {
+		ret = -ENOMEM;
 		goto out_free_path;
 	}
+	qgroup = add_qgroup_rb(fs_info, prealloc, BTRFS_FS_TREE_OBJECTID);
+	prealloc = NULL;
 	ret = btrfs_sysfs_add_one_qgroup(fs_info, qgroup);
 	if (ret < 0) {
 		btrfs_abort_transaction(trans, ret);
 		goto out_free_path;
 	}
 
+	fs_info->qgroup_enable_gen = trans->transid;
+
 	mutex_unlock(&fs_info->qgroup_ioctl_lock);
 	/*
 	 * Commit the transaction while not holding qgroup_ioctl_lock, to avoid
@@ -1181,8 +1245,14 @@ out_add_root:
 	spin_lock(&fs_info->qgroup_lock);
 	fs_info->quota_root = quota_root;
 	set_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags);
+	if (simple)
+		btrfs_set_fs_incompat(fs_info, SIMPLE_QUOTA);
 	spin_unlock(&fs_info->qgroup_lock);
 
+	/* Skip rescan for simple qgroups. */
+	if (btrfs_qgroup_mode(fs_info) == BTRFS_QGROUP_MODE_SIMPLE)
+		goto out_free_path;
+
 	ret = qgroup_rescan_init(fs_info, 0, 1);
 	if (!ret) {
 	        qgroup_rescan_zero_tracking(fs_info);
@@ -1223,6 +1293,39 @@ out:
 	else if (trans)
 		ret = btrfs_end_transaction(trans);
 	ulist_free(ulist);
+	kfree(prealloc);
+	return ret;
+}
+
+/*
+ * It is possible to have outstanding ordered extents which reserved bytes
+ * before we disabled. We need to fully flush delalloc, ordered extents, and a
+ * commit to ensure that we don't leak such reservations, only to have them
+ * come back if we re-enable.
+ *
+ * - enable simple quotas
+ * - reserve space
+ * - release it, store rsv_bytes in OE
+ * - disable quotas
+ * - enable simple quotas (qgroup rsv are all 0)
+ * - OE finishes
+ * - run delayed refs
+ * - free rsv_bytes, resulting in miscounting or even underflow
+ */
+static int flush_reservations(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_trans_handle *trans;
+	int ret;
+
+	ret = btrfs_start_delalloc_roots(fs_info, LONG_MAX, false);
+	if (ret)
+		return ret;
+	btrfs_wait_ordered_roots(fs_info, U64_MAX, 0, (u64)-1);
+	trans = btrfs_join_transaction(fs_info->tree_root);
+	if (IS_ERR(trans))
+		return PTR_ERR(trans);
+	btrfs_commit_transaction(trans);
+
 	return ret;
 }
 
@@ -1270,6 +1373,10 @@ int btrfs_quota_disable(struct btrfs_fs_info *fs_info)
 	clear_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags);
 	btrfs_qgroup_wait_for_completion(fs_info, false);
 
+	ret = flush_reservations(fs_info);
+	if (ret)
+		goto out_unlock_cleaner;
+
 	/*
 	 * 1 For the root item
 	 *
@@ -1296,6 +1403,7 @@ int btrfs_quota_disable(struct btrfs_fs_info *fs_info)
 	quota_root = fs_info->quota_root;
 	fs_info->quota_root = NULL;
 	fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_ON;
+	fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_SIMPLE_MODE;
 	fs_info->qgroup_drop_subtree_thres = BTRFS_MAX_LEVEL;
 	spin_unlock(&fs_info->qgroup_lock);
 
@@ -1330,7 +1438,8 @@ out:
 	if (ret && trans)
 		btrfs_end_transaction(trans);
 	else if (trans)
-		ret = btrfs_end_transaction(trans);
+		ret = btrfs_commit_transaction(trans);
+out_unlock_cleaner:
 	mutex_unlock(&fs_info->cleaner_mutex);
 
 	return ret;
@@ -1375,14 +1484,12 @@ static void qgroup_iterator_clean(struct list_head *head)
  *
  * Caller should hold fs_info->qgroup_lock.
  */
-static int __qgroup_excl_accounting(struct btrfs_fs_info *fs_info,
-				    struct ulist *tmp, u64 ref_root,
+static int __qgroup_excl_accounting(struct btrfs_fs_info *fs_info, u64 ref_root,
 				    struct btrfs_qgroup *src, int sign)
 {
 	struct btrfs_qgroup *qgroup;
-	struct btrfs_qgroup_list *glist;
-	struct ulist_node *unode;
-	struct ulist_iterator uiter;
+	struct btrfs_qgroup *cur;
+	LIST_HEAD(qgroup_list);
 	u64 num_bytes = src->excl;
 	int ret = 0;
 
@@ -1390,53 +1497,30 @@ static int __qgroup_excl_accounting(struct btrfs_fs_info *fs_info,
 	if (!qgroup)
 		goto out;
 
-	qgroup->rfer += sign * num_bytes;
-	qgroup->rfer_cmpr += sign * num_bytes;
-
-	WARN_ON(sign < 0 && qgroup->excl < num_bytes);
-	qgroup->excl += sign * num_bytes;
-	qgroup->excl_cmpr += sign * num_bytes;
-
-	if (sign > 0)
-		qgroup_rsv_add_by_qgroup(fs_info, qgroup, src);
-	else
-		qgroup_rsv_release_by_qgroup(fs_info, qgroup, src);
-
-	qgroup_dirty(fs_info, qgroup);
-
-	/* Get all of the parent groups that contain this qgroup */
-	list_for_each_entry(glist, &qgroup->groups, next_group) {
-		ret = ulist_add(tmp, glist->group->qgroupid,
-				qgroup_to_aux(glist->group), GFP_ATOMIC);
-		if (ret < 0)
-			goto out;
-	}
+	qgroup_iterator_add(&qgroup_list, qgroup);
+	list_for_each_entry(cur, &qgroup_list, iterator) {
+		struct btrfs_qgroup_list *glist;
 
-	/* Iterate all of the parents and adjust their reference counts */
-	ULIST_ITER_INIT(&uiter);
-	while ((unode = ulist_next(tmp, &uiter))) {
-		qgroup = unode_aux_to_qgroup(unode);
 		qgroup->rfer += sign * num_bytes;
 		qgroup->rfer_cmpr += sign * num_bytes;
+
 		WARN_ON(sign < 0 && qgroup->excl < num_bytes);
 		qgroup->excl += sign * num_bytes;
+		qgroup->excl_cmpr += sign * num_bytes;
+
 		if (sign > 0)
 			qgroup_rsv_add_by_qgroup(fs_info, qgroup, src);
 		else
 			qgroup_rsv_release_by_qgroup(fs_info, qgroup, src);
-		qgroup->excl_cmpr += sign * num_bytes;
 		qgroup_dirty(fs_info, qgroup);
 
-		/* Add any parents of the parents */
-		list_for_each_entry(glist, &qgroup->groups, next_group) {
-			ret = ulist_add(tmp, glist->group->qgroupid,
-					qgroup_to_aux(glist->group), GFP_ATOMIC);
-			if (ret < 0)
-				goto out;
-		}
+		/* Append parent qgroups to @qgroup_list. */
+		list_for_each_entry(glist, &qgroup->groups, next_group)
+			qgroup_iterator_add(&qgroup_list, glist->group);
 	}
 	ret = 0;
 out:
+	qgroup_iterator_clean(&qgroup_list);
 	return ret;
 }
 
@@ -1453,8 +1537,7 @@ out:
  * Return < 0 for other error.
  */
 static int quick_update_accounting(struct btrfs_fs_info *fs_info,
-				   struct ulist *tmp, u64 src, u64 dst,
-				   int sign)
+				   u64 src, u64 dst, int sign)
 {
 	struct btrfs_qgroup *qgroup;
 	int ret = 1;
@@ -1465,8 +1548,7 @@ static int quick_update_accounting(struct btrfs_fs_info *fs_info,
 		goto out;
 	if (qgroup->excl == qgroup->rfer) {
 		ret = 0;
-		err = __qgroup_excl_accounting(fs_info, tmp, dst,
-					       qgroup, sign);
+		err = __qgroup_excl_accounting(fs_info, dst, qgroup, sign);
 		if (err < 0) {
 			ret = err;
 			goto out;
@@ -1478,28 +1560,19 @@ out:
 	return ret;
 }
 
-int btrfs_add_qgroup_relation(struct btrfs_trans_handle *trans, u64 src,
-			      u64 dst)
+int btrfs_add_qgroup_relation(struct btrfs_trans_handle *trans, u64 src, u64 dst)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
 	struct btrfs_qgroup *parent;
 	struct btrfs_qgroup *member;
 	struct btrfs_qgroup_list *list;
-	struct ulist *tmp;
-	unsigned int nofs_flag;
+	struct btrfs_qgroup_list *prealloc = NULL;
 	int ret = 0;
 
 	/* Check the level of src and dst first */
 	if (btrfs_qgroup_level(src) >= btrfs_qgroup_level(dst))
 		return -EINVAL;
 
-	/* We hold a transaction handle open, must do a NOFS allocation. */
-	nofs_flag = memalloc_nofs_save();
-	tmp = ulist_alloc(GFP_KERNEL);
-	memalloc_nofs_restore(nofs_flag);
-	if (!tmp)
-		return -ENOMEM;
-
 	mutex_lock(&fs_info->qgroup_ioctl_lock);
 	if (!fs_info->quota_root) {
 		ret = -ENOTCONN;
@@ -1520,6 +1593,11 @@ int btrfs_add_qgroup_relation(struct btrfs_trans_handle *trans, u64 src,
 		}
 	}
 
+	prealloc = kzalloc(sizeof(*list), GFP_NOFS);
+	if (!prealloc) {
+		ret = -ENOMEM;
+		goto out;
+	}
 	ret = add_qgroup_relation_item(trans, src, dst);
 	if (ret)
 		goto out;
@@ -1531,16 +1609,17 @@ int btrfs_add_qgroup_relation(struct btrfs_trans_handle *trans, u64 src,
 	}
 
 	spin_lock(&fs_info->qgroup_lock);
-	ret = __add_relation_rb(member, parent);
+	ret = __add_relation_rb(prealloc, member, parent);
+	prealloc = NULL;
 	if (ret < 0) {
 		spin_unlock(&fs_info->qgroup_lock);
 		goto out;
 	}
-	ret = quick_update_accounting(fs_info, tmp, src, dst, 1);
+	ret = quick_update_accounting(fs_info, src, dst, 1);
 	spin_unlock(&fs_info->qgroup_lock);
 out:
+	kfree(prealloc);
 	mutex_unlock(&fs_info->qgroup_ioctl_lock);
-	ulist_free(tmp);
 	return ret;
 }
 
@@ -1551,19 +1630,10 @@ static int __del_qgroup_relation(struct btrfs_trans_handle *trans, u64 src,
 	struct btrfs_qgroup *parent;
 	struct btrfs_qgroup *member;
 	struct btrfs_qgroup_list *list;
-	struct ulist *tmp;
 	bool found = false;
-	unsigned int nofs_flag;
 	int ret = 0;
 	int ret2;
 
-	/* We hold a transaction handle open, must do a NOFS allocation. */
-	nofs_flag = memalloc_nofs_save();
-	tmp = ulist_alloc(GFP_KERNEL);
-	memalloc_nofs_restore(nofs_flag);
-	if (!tmp)
-		return -ENOMEM;
-
 	if (!fs_info->quota_root) {
 		ret = -ENOTCONN;
 		goto out;
@@ -1601,11 +1671,10 @@ delete_item:
 	if (found) {
 		spin_lock(&fs_info->qgroup_lock);
 		del_relation_rb(fs_info, src, dst);
-		ret = quick_update_accounting(fs_info, tmp, src, dst, -1);
+		ret = quick_update_accounting(fs_info, src, dst, -1);
 		spin_unlock(&fs_info->qgroup_lock);
 	}
 out:
-	ulist_free(tmp);
 	return ret;
 }
 
@@ -1627,8 +1696,12 @@ int btrfs_create_qgroup(struct btrfs_trans_handle *trans, u64 qgroupid)
 	struct btrfs_fs_info *fs_info = trans->fs_info;
 	struct btrfs_root *quota_root;
 	struct btrfs_qgroup *qgroup;
+	struct btrfs_qgroup *prealloc = NULL;
 	int ret = 0;
 
+	if (btrfs_qgroup_mode(fs_info) == BTRFS_QGROUP_MODE_DISABLED)
+		return 0;
+
 	mutex_lock(&fs_info->qgroup_ioctl_lock);
 	if (!fs_info->quota_root) {
 		ret = -ENOTCONN;
@@ -1641,24 +1714,37 @@ int btrfs_create_qgroup(struct btrfs_trans_handle *trans, u64 qgroupid)
 		goto out;
 	}
 
+	prealloc = kzalloc(sizeof(*prealloc), GFP_NOFS);
+	if (!prealloc) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
 	ret = add_qgroup_item(trans, quota_root, qgroupid);
 	if (ret)
 		goto out;
 
 	spin_lock(&fs_info->qgroup_lock);
-	qgroup = add_qgroup_rb(fs_info, qgroupid);
+	qgroup = add_qgroup_rb(fs_info, prealloc, qgroupid);
 	spin_unlock(&fs_info->qgroup_lock);
+	prealloc = NULL;
 
-	if (IS_ERR(qgroup)) {
-		ret = PTR_ERR(qgroup);
-		goto out;
-	}
 	ret = btrfs_sysfs_add_one_qgroup(fs_info, qgroup);
 out:
 	mutex_unlock(&fs_info->qgroup_ioctl_lock);
+	kfree(prealloc);
 	return ret;
 }
 
+static bool qgroup_has_usage(struct btrfs_qgroup *qgroup)
+{
+	return (qgroup->rfer > 0 || qgroup->rfer_cmpr > 0 ||
+		qgroup->excl > 0 || qgroup->excl_cmpr > 0 ||
+		qgroup->rsv.values[BTRFS_QGROUP_RSV_DATA] > 0 ||
+		qgroup->rsv.values[BTRFS_QGROUP_RSV_META_PREALLOC] > 0 ||
+		qgroup->rsv.values[BTRFS_QGROUP_RSV_META_PERTRANS] > 0);
+}
+
 int btrfs_remove_qgroup(struct btrfs_trans_handle *trans, u64 qgroupid)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
@@ -1678,6 +1764,11 @@ int btrfs_remove_qgroup(struct btrfs_trans_handle *trans, u64 qgroupid)
 		goto out;
 	}
 
+	if (is_fstree(qgroupid) && qgroup_has_usage(qgroup)) {
+		ret = -EBUSY;
+		goto out;
+	}
+
 	/* Check if there are no children of this qgroup */
 	if (!list_empty(&qgroup->members)) {
 		ret = -EBUSY;
@@ -1790,6 +1881,17 @@ out:
 	return ret;
 }
 
+/*
+ * Inform qgroup to trace one dirty extent, its info is recorded in @record.
+ * So qgroup can account it at transaction committing time.
+ *
+ * No lock version, caller must acquire delayed ref lock and allocated memory,
+ * then call btrfs_qgroup_trace_extent_post() after exiting lock context.
+ *
+ * Return 0 for success insert
+ * Return >0 for existing record, caller can free @record safely.
+ * Error is not possible
+ */
 int btrfs_qgroup_trace_extent_nolock(struct btrfs_fs_info *fs_info,
 				struct btrfs_delayed_ref_root *delayed_refs,
 				struct btrfs_qgroup_extent_record *record)
@@ -1799,6 +1901,9 @@ int btrfs_qgroup_trace_extent_nolock(struct btrfs_fs_info *fs_info,
 	struct btrfs_qgroup_extent_record *entry;
 	u64 bytenr = record->bytenr;
 
+	if (!btrfs_qgroup_full_accounting(fs_info))
+		return 1;
+
 	lockdep_assert_held(&delayed_refs->lock);
 	trace_btrfs_qgroup_trace_extent(fs_info, record);
 
@@ -1825,12 +1930,35 @@ int btrfs_qgroup_trace_extent_nolock(struct btrfs_fs_info *fs_info,
 	return 0;
 }
 
+/*
+ * Post handler after qgroup_trace_extent_nolock().
+ *
+ * NOTE: Current qgroup does the expensive backref walk at transaction
+ * committing time with TRANS_STATE_COMMIT_DOING, this blocks incoming
+ * new transaction.
+ * This is designed to allow btrfs_find_all_roots() to get correct new_roots
+ * result.
+ *
+ * However for old_roots there is no need to do backref walk at that time,
+ * since we search commit roots to walk backref and result will always be
+ * correct.
+ *
+ * Due to the nature of no lock version, we can't do backref there.
+ * So we must call btrfs_qgroup_trace_extent_post() after exiting
+ * spinlock context.
+ *
+ * TODO: If we can fix and prove btrfs_find_all_roots() can get correct result
+ * using current root, then we can move all expensive backref walk out of
+ * transaction committing, but not now as qgroup accounting will be wrong again.
+ */
 int btrfs_qgroup_trace_extent_post(struct btrfs_trans_handle *trans,
 				   struct btrfs_qgroup_extent_record *qrecord)
 {
 	struct btrfs_backref_walk_ctx ctx = { 0 };
 	int ret;
 
+	if (!btrfs_qgroup_full_accounting(trans->fs_info))
+		return 0;
 	/*
 	 * We are always called in a context where we are already holding a
 	 * transaction handle. Often we are called when adding a data delayed
@@ -1878,6 +2006,19 @@ int btrfs_qgroup_trace_extent_post(struct btrfs_trans_handle *trans,
 	return 0;
 }
 
+/*
+ * Inform qgroup to trace one dirty extent, specified by @bytenr and
+ * @num_bytes.
+ * So qgroup can account it at commit trans time.
+ *
+ * Better encapsulated version, with memory allocation and backref walk for
+ * commit roots.
+ * So this can sleep.
+ *
+ * Return 0 if the operation is done.
+ * Return <0 for error, like memory allocation failure or invalid parameter
+ * (NULL trans)
+ */
 int btrfs_qgroup_trace_extent(struct btrfs_trans_handle *trans, u64 bytenr,
 			      u64 num_bytes)
 {
@@ -1886,8 +2027,7 @@ int btrfs_qgroup_trace_extent(struct btrfs_trans_handle *trans, u64 bytenr,
 	struct btrfs_delayed_ref_root *delayed_refs;
 	int ret;
 
-	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags)
-	    || bytenr == 0 || num_bytes == 0)
+	if (!btrfs_qgroup_full_accounting(fs_info) || bytenr == 0 || num_bytes == 0)
 		return 0;
 	record = kzalloc(sizeof(*record), GFP_NOFS);
 	if (!record)
@@ -1908,6 +2048,12 @@ int btrfs_qgroup_trace_extent(struct btrfs_trans_handle *trans, u64 bytenr,
 	return btrfs_qgroup_trace_extent_post(trans, record);
 }
 
+/*
+ * Inform qgroup to trace all leaf items of data
+ *
+ * Return 0 for success
+ * Return <0 for error(ENOMEM)
+ */
 int btrfs_qgroup_trace_leaf_items(struct btrfs_trans_handle *trans,
 				  struct extent_buffer *eb)
 {
@@ -1919,7 +2065,7 @@ int btrfs_qgroup_trace_leaf_items(struct btrfs_trans_handle *trans,
 	u64 bytenr, num_bytes;
 
 	/* We can be called directly from walk_up_proc() */
-	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags))
+	if (!btrfs_qgroup_full_accounting(fs_info))
 		return 0;
 
 	for (i = 0; i < nr; i++) {
@@ -2295,7 +2441,7 @@ static int qgroup_trace_subtree_swap(struct btrfs_trans_handle *trans,
 	int level;
 	int ret;
 
-	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags))
+	if (!btrfs_qgroup_full_accounting(fs_info))
 		return 0;
 
 	/* Wrong parameter order */
@@ -2338,6 +2484,16 @@ out:
 	return ret;
 }
 
+/*
+ * Inform qgroup to trace a whole subtree, including all its child tree
+ * blocks and data.
+ * The root tree block is specified by @root_eb.
+ *
+ * Normally used by relocation(tree block swap) and subvolume deletion.
+ *
+ * Return 0 for success
+ * Return <0 for error(ENOMEM or tree search error)
+ */
 int btrfs_qgroup_trace_subtree(struct btrfs_trans_handle *trans,
 			       struct extent_buffer *root_eb,
 			       u64 root_gen, int root_level)
@@ -2352,7 +2508,7 @@ int btrfs_qgroup_trace_subtree(struct btrfs_trans_handle *trans,
 	BUG_ON(root_level < 0 || root_level >= BTRFS_MAX_LEVEL);
 	BUG_ON(root_eb == NULL);
 
-	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags))
+	if (!btrfs_qgroup_full_accounting(fs_info))
 		return 0;
 
 	spin_lock(&fs_info->qgroup_lock);
@@ -2464,62 +2620,64 @@ out:
 	return ret;
 }
 
+static void qgroup_iterator_nested_add(struct list_head *head, struct btrfs_qgroup *qgroup)
+{
+	if (!list_empty(&qgroup->nested_iterator))
+		return;
+
+	list_add_tail(&qgroup->nested_iterator, head);
+}
+
+static void qgroup_iterator_nested_clean(struct list_head *head)
+{
+	while (!list_empty(head)) {
+		struct btrfs_qgroup *qgroup;
+
+		qgroup = list_first_entry(head, struct btrfs_qgroup, nested_iterator);
+		list_del_init(&qgroup->nested_iterator);
+	}
+}
+
 #define UPDATE_NEW	0
 #define UPDATE_OLD	1
 /*
  * Walk all of the roots that points to the bytenr and adjust their refcnts.
  */
-static int qgroup_update_refcnt(struct btrfs_fs_info *fs_info,
-				struct ulist *roots, struct ulist *tmp,
-				struct ulist *qgroups, u64 seq, int update_old)
+static void qgroup_update_refcnt(struct btrfs_fs_info *fs_info,
+				 struct ulist *roots, struct list_head *qgroups,
+				 u64 seq, int update_old)
 {
 	struct ulist_node *unode;
 	struct ulist_iterator uiter;
-	struct ulist_node *tmp_unode;
-	struct ulist_iterator tmp_uiter;
 	struct btrfs_qgroup *qg;
-	int ret = 0;
 
 	if (!roots)
-		return 0;
+		return;
 	ULIST_ITER_INIT(&uiter);
 	while ((unode = ulist_next(roots, &uiter))) {
+		LIST_HEAD(tmp);
+
 		qg = find_qgroup_rb(fs_info, unode->val);
 		if (!qg)
 			continue;
 
-		ulist_reinit(tmp);
-		ret = ulist_add(qgroups, qg->qgroupid, qgroup_to_aux(qg),
-				GFP_ATOMIC);
-		if (ret < 0)
-			return ret;
-		ret = ulist_add(tmp, qg->qgroupid, qgroup_to_aux(qg), GFP_ATOMIC);
-		if (ret < 0)
-			return ret;
-		ULIST_ITER_INIT(&tmp_uiter);
-		while ((tmp_unode = ulist_next(tmp, &tmp_uiter))) {
+		qgroup_iterator_nested_add(qgroups, qg);
+		qgroup_iterator_add(&tmp, qg);
+		list_for_each_entry(qg, &tmp, iterator) {
 			struct btrfs_qgroup_list *glist;
 
-			qg = unode_aux_to_qgroup(tmp_unode);
 			if (update_old)
 				btrfs_qgroup_update_old_refcnt(qg, seq, 1);
 			else
 				btrfs_qgroup_update_new_refcnt(qg, seq, 1);
+
 			list_for_each_entry(glist, &qg->groups, next_group) {
-				ret = ulist_add(qgroups, glist->group->qgroupid,
-						qgroup_to_aux(glist->group),
-						GFP_ATOMIC);
-				if (ret < 0)
-					return ret;
-				ret = ulist_add(tmp, glist->group->qgroupid,
-						qgroup_to_aux(glist->group),
-						GFP_ATOMIC);
-				if (ret < 0)
-					return ret;
+				qgroup_iterator_nested_add(qgroups, glist->group);
+				qgroup_iterator_add(&tmp, glist->group);
 			}
 		}
+		qgroup_iterator_clean(&tmp);
 	}
-	return 0;
 }
 
 /*
@@ -2558,22 +2716,16 @@ static int qgroup_update_refcnt(struct btrfs_fs_info *fs_info,
  * But this time we don't need to consider other things, the codes and logic
  * is easy to understand now.
  */
-static int qgroup_update_counters(struct btrfs_fs_info *fs_info,
-				  struct ulist *qgroups,
-				  u64 nr_old_roots,
-				  u64 nr_new_roots,
-				  u64 num_bytes, u64 seq)
+static void qgroup_update_counters(struct btrfs_fs_info *fs_info,
+				   struct list_head *qgroups, u64 nr_old_roots,
+				   u64 nr_new_roots, u64 num_bytes, u64 seq)
 {
-	struct ulist_node *unode;
-	struct ulist_iterator uiter;
 	struct btrfs_qgroup *qg;
-	u64 cur_new_count, cur_old_count;
 
-	ULIST_ITER_INIT(&uiter);
-	while ((unode = ulist_next(qgroups, &uiter))) {
+	list_for_each_entry(qg, qgroups, nested_iterator) {
+		u64 cur_new_count, cur_old_count;
 		bool dirty = false;
 
-		qg = unode_aux_to_qgroup(unode);
 		cur_old_count = btrfs_qgroup_get_old_refcnt(qg, seq);
 		cur_new_count = btrfs_qgroup_get_new_refcnt(qg, seq);
 
@@ -2644,7 +2796,6 @@ static int qgroup_update_counters(struct btrfs_fs_info *fs_info,
 		if (dirty)
 			qgroup_dirty(fs_info, qg);
 	}
-	return 0;
 }
 
 /*
@@ -2681,8 +2832,7 @@ int btrfs_qgroup_account_extent(struct btrfs_trans_handle *trans, u64 bytenr,
 				struct ulist *new_roots)
 {
 	struct btrfs_fs_info *fs_info = trans->fs_info;
-	struct ulist *qgroups = NULL;
-	struct ulist *tmp = NULL;
+	LIST_HEAD(qgroups);
 	u64 seq;
 	u64 nr_new_roots = 0;
 	u64 nr_old_roots = 0;
@@ -2692,7 +2842,7 @@ int btrfs_qgroup_account_extent(struct btrfs_trans_handle *trans, u64 bytenr,
 	 * If quotas get disabled meanwhile, the resources need to be freed and
 	 * we can't just exit here.
 	 */
-	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags) ||
+	if (!btrfs_qgroup_full_accounting(fs_info) ||
 	    fs_info->qgroup_flags & BTRFS_QGROUP_RUNTIME_FLAG_NO_ACCOUNTING)
 		goto out_free;
 
@@ -2716,17 +2866,6 @@ int btrfs_qgroup_account_extent(struct btrfs_trans_handle *trans, u64 bytenr,
 	trace_btrfs_qgroup_account_extent(fs_info, trans->transid, bytenr,
 					num_bytes, nr_old_roots, nr_new_roots);
 
-	qgroups = ulist_alloc(GFP_NOFS);
-	if (!qgroups) {
-		ret = -ENOMEM;
-		goto out_free;
-	}
-	tmp = ulist_alloc(GFP_NOFS);
-	if (!tmp) {
-		ret = -ENOMEM;
-		goto out_free;
-	}
-
 	mutex_lock(&fs_info->qgroup_rescan_lock);
 	if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) {
 		if (fs_info->qgroup_rescan_progress.objectid <= bytenr) {
@@ -2741,29 +2880,27 @@ int btrfs_qgroup_account_extent(struct btrfs_trans_handle *trans, u64 bytenr,
 	seq = fs_info->qgroup_seq;
 
 	/* Update old refcnts using old_roots */
-	ret = qgroup_update_refcnt(fs_info, old_roots, tmp, qgroups, seq,
-				   UPDATE_OLD);
-	if (ret < 0)
-		goto out;
+	qgroup_update_refcnt(fs_info, old_roots, &qgroups, seq, UPDATE_OLD);
 
 	/* Update new refcnts using new_roots */
-	ret = qgroup_update_refcnt(fs_info, new_roots, tmp, qgroups, seq,
-				   UPDATE_NEW);
-	if (ret < 0)
-		goto out;
+	qgroup_update_refcnt(fs_info, new_roots, &qgroups, seq, UPDATE_NEW);
 
-	qgroup_update_counters(fs_info, qgroups, nr_old_roots, nr_new_roots,
+	qgroup_update_counters(fs_info, &qgroups, nr_old_roots, nr_new_roots,
 			       num_bytes, seq);
 
 	/*
+	 * We're done using the iterator, release all its qgroups while holding
+	 * fs_info->qgroup_lock so that we don't race with btrfs_remove_qgroup()
+	 * and trigger use-after-free accesses to qgroups.
+	 */
+	qgroup_iterator_nested_clean(&qgroups);
+
+	/*
 	 * Bump qgroup_seq to avoid seq overlap
 	 */
 	fs_info->qgroup_seq += max(nr_old_roots, nr_new_roots) + 1;
-out:
 	spin_unlock(&fs_info->qgroup_lock);
 out_free:
-	ulist_free(tmp);
-	ulist_free(qgroups);
 	ulist_free(old_roots);
 	ulist_free(new_roots);
 	return ret;
@@ -2780,6 +2917,9 @@ int btrfs_qgroup_account_extents(struct btrfs_trans_handle *trans)
 	u64 qgroup_to_skip;
 	int ret = 0;
 
+	if (btrfs_qgroup_mode(fs_info) == BTRFS_QGROUP_MODE_SIMPLE)
+		return 0;
+
 	delayed_refs = &trans->transaction->delayed_refs;
 	qgroup_to_skip = delayed_refs->qgroup_to_skip;
 	while ((node = rb_first(&delayed_refs->dirty_extent_root))) {
@@ -2895,7 +3035,7 @@ int btrfs_run_qgroups(struct btrfs_trans_handle *trans)
 			qgroup_mark_inconsistent(fs_info);
 		spin_lock(&fs_info->qgroup_lock);
 	}
-	if (test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags))
+	if (btrfs_qgroup_enabled(fs_info))
 		fs_info->qgroup_flags |= BTRFS_QGROUP_STATUS_FLAG_ON;
 	else
 		fs_info->qgroup_flags &= ~BTRFS_QGROUP_STATUS_FLAG_ON;
@@ -2908,6 +3048,47 @@ int btrfs_run_qgroups(struct btrfs_trans_handle *trans)
 	return ret;
 }
 
+static int qgroup_auto_inherit(struct btrfs_fs_info *fs_info,
+			       u64 inode_rootid,
+			       struct btrfs_qgroup_inherit **inherit)
+{
+	int i = 0;
+	u64 num_qgroups = 0;
+	struct btrfs_qgroup *inode_qg;
+	struct btrfs_qgroup_list *qg_list;
+	struct btrfs_qgroup_inherit *res;
+	size_t struct_sz;
+	u64 *qgids;
+
+	if (*inherit)
+		return -EEXIST;
+
+	inode_qg = find_qgroup_rb(fs_info, inode_rootid);
+	if (!inode_qg)
+		return -ENOENT;
+
+	num_qgroups = list_count_nodes(&inode_qg->groups);
+
+	if (!num_qgroups)
+		return 0;
+
+	struct_sz = struct_size(res, qgroups, num_qgroups);
+	if (struct_sz == SIZE_MAX)
+		return -ERANGE;
+
+	res = kzalloc(struct_sz, GFP_NOFS);
+	if (!res)
+		return -ENOMEM;
+	res->num_qgroups = num_qgroups;
+	qgids = res->qgroups;
+
+	list_for_each_entry(qg_list, &inode_qg->groups, next_group)
+		qgids[i] = qg_list->group->qgroupid;
+
+	*inherit = res;
+	return 0;
+}
+
 /*
  * Copy the accounting information between qgroups. This is necessary
  * when a snapshot or a subvolume is created. Throwing an error will
@@ -2915,7 +3096,8 @@ int btrfs_run_qgroups(struct btrfs_trans_handle *trans)
  * when a readonly fs is a reasonable outcome.
  */
 int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans, u64 srcid,
-			 u64 objectid, struct btrfs_qgroup_inherit *inherit)
+			 u64 objectid, u64 inode_rootid,
+			 struct btrfs_qgroup_inherit *inherit)
 {
 	int ret = 0;
 	int i;
@@ -2925,10 +3107,17 @@ int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans, u64 srcid,
 	struct btrfs_root *quota_root;
 	struct btrfs_qgroup *srcgroup;
 	struct btrfs_qgroup *dstgroup;
+	struct btrfs_qgroup *prealloc;
+	struct btrfs_qgroup_list **qlist_prealloc = NULL;
+	bool free_inherit = false;
 	bool need_rescan = false;
 	u32 level_size = 0;
 	u64 nums;
 
+	prealloc = kzalloc(sizeof(*prealloc), GFP_NOFS);
+	if (!prealloc)
+		return -ENOMEM;
+
 	/*
 	 * There are only two callers of this function.
 	 *
@@ -2948,7 +3137,7 @@ int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans, u64 srcid,
 
 	if (!committing)
 		mutex_lock(&fs_info->qgroup_ioctl_lock);
-	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags))
+	if (!btrfs_qgroup_enabled(fs_info))
 		goto out;
 
 	quota_root = fs_info->quota_root;
@@ -2957,6 +3146,13 @@ int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans, u64 srcid,
 		goto out;
 	}
 
+	if (btrfs_qgroup_mode(fs_info) == BTRFS_QGROUP_MODE_SIMPLE && !inherit) {
+		ret = qgroup_auto_inherit(fs_info, inode_rootid, &inherit);
+		if (ret)
+			goto out;
+		free_inherit = true;
+	}
+
 	if (inherit) {
 		i_qgroups = (u64 *)(inherit + 1);
 		nums = inherit->num_qgroups + 2 * inherit->num_ref_copies +
@@ -3001,16 +3197,28 @@ int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans, u64 srcid,
 				goto out;
 		}
 		ret = 0;
-	}
 
+		qlist_prealloc = kcalloc(inherit->num_qgroups,
+					 sizeof(struct btrfs_qgroup_list *),
+					 GFP_NOFS);
+		if (!qlist_prealloc) {
+			ret = -ENOMEM;
+			goto out;
+		}
+		for (int i = 0; i < inherit->num_qgroups; i++) {
+			qlist_prealloc[i] = kzalloc(sizeof(struct btrfs_qgroup_list),
+						    GFP_NOFS);
+			if (!qlist_prealloc[i]) {
+				ret = -ENOMEM;
+				goto out;
+			}
+		}
+	}
 
 	spin_lock(&fs_info->qgroup_lock);
 
-	dstgroup = add_qgroup_rb(fs_info, objectid);
-	if (IS_ERR(dstgroup)) {
-		ret = PTR_ERR(dstgroup);
-		goto unlock;
-	}
+	dstgroup = add_qgroup_rb(fs_info, prealloc, objectid);
+	prealloc = NULL;
 
 	if (inherit && inherit->flags & BTRFS_QGROUP_INHERIT_SET_LIMITS) {
 		dstgroup->lim_flags = inherit->lim.flags;
@@ -3022,7 +3230,7 @@ int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans, u64 srcid,
 		qgroup_dirty(fs_info, dstgroup);
 	}
 
-	if (srcid) {
+	if (srcid && btrfs_qgroup_mode(fs_info) == BTRFS_QGROUP_MODE_FULL) {
 		srcgroup = find_qgroup_rb(fs_info, srcid);
 		if (!srcgroup)
 			goto unlock;
@@ -3057,7 +3265,9 @@ int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans, u64 srcid,
 	i_qgroups = (u64 *)(inherit + 1);
 	for (i = 0; i < inherit->num_qgroups; ++i) {
 		if (*i_qgroups) {
-			ret = add_relation_rb(fs_info, objectid, *i_qgroups);
+			ret = add_relation_rb(fs_info, qlist_prealloc[i], objectid,
+					      *i_qgroups);
+			qlist_prealloc[i] = NULL;
 			if (ret)
 				goto unlock;
 		}
@@ -3121,6 +3331,14 @@ out:
 		mutex_unlock(&fs_info->qgroup_ioctl_lock);
 	if (need_rescan)
 		qgroup_mark_inconsistent(fs_info);
+	if (qlist_prealloc) {
+		for (int i = 0; i < inherit->num_qgroups; i++)
+			kfree(qlist_prealloc[i]);
+		kfree(qlist_prealloc);
+	}
+	if (free_inherit)
+		kfree(inherit);
+	kfree(prealloc);
 	return ret;
 }
 
@@ -3204,9 +3422,7 @@ void btrfs_qgroup_free_refroot(struct btrfs_fs_info *fs_info,
 			       enum btrfs_qgroup_rsv_type type)
 {
 	struct btrfs_qgroup *qgroup;
-	struct ulist_node *unode;
-	struct ulist_iterator uiter;
-	int ret = 0;
+	LIST_HEAD(qgroup_list);
 
 	if (!is_fstree(ref_root))
 		return;
@@ -3234,30 +3450,17 @@ void btrfs_qgroup_free_refroot(struct btrfs_fs_info *fs_info,
 		 */
 		num_bytes = qgroup->rsv.values[type];
 
-	ulist_reinit(fs_info->qgroup_ulist);
-	ret = ulist_add(fs_info->qgroup_ulist, qgroup->qgroupid,
-			qgroup_to_aux(qgroup), GFP_ATOMIC);
-	if (ret < 0)
-		goto out;
-	ULIST_ITER_INIT(&uiter);
-	while ((unode = ulist_next(fs_info->qgroup_ulist, &uiter))) {
-		struct btrfs_qgroup *qg;
+	qgroup_iterator_add(&qgroup_list, qgroup);
+	list_for_each_entry(qgroup, &qgroup_list, iterator) {
 		struct btrfs_qgroup_list *glist;
 
-		qg = unode_aux_to_qgroup(unode);
-
-		qgroup_rsv_release(fs_info, qg, num_bytes, type);
-
-		list_for_each_entry(glist, &qg->groups, next_group) {
-			ret = ulist_add(fs_info->qgroup_ulist,
-					glist->group->qgroupid,
-					qgroup_to_aux(glist->group), GFP_ATOMIC);
-			if (ret < 0)
-				goto out;
+		qgroup_rsv_release(fs_info, qgroup, num_bytes, type);
+		list_for_each_entry(glist, &qgroup->groups, next_group) {
+			qgroup_iterator_add(&qgroup_list, glist->group);
 		}
 	}
-
 out:
+	qgroup_iterator_clean(&qgroup_list);
 	spin_unlock(&fs_info->qgroup_lock);
 }
 
@@ -3292,6 +3495,9 @@ static int qgroup_rescan_leaf(struct btrfs_trans_handle *trans,
 	int slot;
 	int ret;
 
+	if (!btrfs_qgroup_full_accounting(fs_info))
+		return 1;
+
 	mutex_lock(&fs_info->qgroup_rescan_lock);
 	extent_root = btrfs_extent_root(fs_info,
 				fs_info->qgroup_rescan_progress.objectid);
@@ -3372,10 +3578,15 @@ out:
 
 static bool rescan_should_stop(struct btrfs_fs_info *fs_info)
 {
-	return btrfs_fs_closing(fs_info) ||
-		test_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state) ||
-		!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags) ||
-			  fs_info->qgroup_flags & BTRFS_QGROUP_RUNTIME_FLAG_CANCEL_RESCAN;
+	if (btrfs_fs_closing(fs_info))
+		return true;
+	if (test_bit(BTRFS_FS_STATE_REMOUNTING, &fs_info->fs_state))
+		return true;
+	if (!btrfs_qgroup_enabled(fs_info))
+		return true;
+	if (fs_info->qgroup_flags & BTRFS_QGROUP_RUNTIME_FLAG_CANCEL_RESCAN)
+		return true;
+	return false;
 }
 
 static void btrfs_qgroup_rescan_worker(struct btrfs_work *work)
@@ -3389,6 +3600,9 @@ static void btrfs_qgroup_rescan_worker(struct btrfs_work *work)
 	bool stopped = false;
 	bool did_leaf_rescans = false;
 
+	if (btrfs_qgroup_mode(fs_info) == BTRFS_QGROUP_MODE_SIMPLE)
+		return;
+
 	path = btrfs_alloc_path();
 	if (!path)
 		goto out;
@@ -3492,6 +3706,11 @@ qgroup_rescan_init(struct btrfs_fs_info *fs_info, u64 progress_objectid,
 {
 	int ret = 0;
 
+	if (btrfs_qgroup_mode(fs_info) == BTRFS_QGROUP_MODE_SIMPLE) {
+		btrfs_warn(fs_info, "qgroup rescan init failed, running in simple mode");
+		return -EINVAL;
+	}
+
 	if (!init_flags) {
 		/* we're resuming qgroup rescan at mount time */
 		if (!(fs_info->qgroup_flags &
@@ -3522,7 +3741,7 @@ qgroup_rescan_init(struct btrfs_fs_info *fs_info, u64 progress_objectid,
 			btrfs_warn(fs_info,
 			"qgroup rescan init failed, qgroup is not enabled");
 			ret = -EINVAL;
-		} else if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags)) {
+		} else if (btrfs_qgroup_mode(fs_info) == BTRFS_QGROUP_MODE_DISABLED) {
 			/* Quota disable is in progress */
 			ret = -EBUSY;
 		}
@@ -3543,7 +3762,7 @@ qgroup_rescan_init(struct btrfs_fs_info *fs_info, u64 progress_objectid,
 	mutex_unlock(&fs_info->qgroup_rescan_lock);
 
 	btrfs_init_work(&fs_info->qgroup_rescan_work,
-			btrfs_qgroup_rescan_worker, NULL, NULL);
+			btrfs_qgroup_rescan_worker, NULL);
 	return 0;
 }
 
@@ -3781,7 +4000,7 @@ static int qgroup_reserve_data(struct btrfs_inode *inode,
 	u64 to_reserve;
 	int ret;
 
-	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &root->fs_info->flags) ||
+	if (btrfs_qgroup_mode(root->fs_info) == BTRFS_QGROUP_MODE_DISABLED ||
 	    !is_fstree(root->root_key.objectid) || len == 0)
 		return 0;
 
@@ -3916,8 +4135,12 @@ static int __btrfs_qgroup_release_data(struct btrfs_inode *inode,
 	int trace_op = QGROUP_RELEASE;
 	int ret;
 
-	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &inode->root->fs_info->flags))
-		return 0;
+	if (btrfs_qgroup_mode(inode->root->fs_info) == BTRFS_QGROUP_MODE_DISABLED) {
+		extent_changeset_init(&changeset);
+		return clear_record_extent_bits(&inode->io_tree, start,
+						start + len - 1,
+						EXTENT_QGROUP_RESERVED, &changeset);
+	}
 
 	/* In release case, we shouldn't have @reserved */
 	WARN_ON(!free && reserved);
@@ -4029,7 +4252,7 @@ int btrfs_qgroup_reserve_meta(struct btrfs_root *root, int num_bytes,
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	int ret;
 
-	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags) ||
+	if (btrfs_qgroup_mode(fs_info) == BTRFS_QGROUP_MODE_DISABLED ||
 	    !is_fstree(root->root_key.objectid) || num_bytes == 0)
 		return 0;
 
@@ -4066,11 +4289,15 @@ int __btrfs_qgroup_reserve_meta(struct btrfs_root *root, int num_bytes,
 	return btrfs_qgroup_reserve_meta(root, num_bytes, type, enforce);
 }
 
+/*
+ * Per-transaction meta reservation should be all freed at transaction commit
+ * time
+ */
 void btrfs_qgroup_free_meta_all_pertrans(struct btrfs_root *root)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 
-	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags) ||
+	if (btrfs_qgroup_mode(fs_info) == BTRFS_QGROUP_MODE_DISABLED ||
 	    !is_fstree(root->root_key.objectid))
 		return;
 
@@ -4086,7 +4313,7 @@ void __btrfs_qgroup_free_meta(struct btrfs_root *root, int num_bytes,
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 
-	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags) ||
+	if (btrfs_qgroup_mode(fs_info) == BTRFS_QGROUP_MODE_DISABLED ||
 	    !is_fstree(root->root_key.objectid))
 		return;
 
@@ -4136,11 +4363,17 @@ out:
 	spin_unlock(&fs_info->qgroup_lock);
 }
 
+/*
+ * Convert @num_bytes of META_PREALLOCATED reservation to META_PERTRANS.
+ *
+ * This is called when preallocated meta reservation needs to be used.
+ * Normally after btrfs_join_transaction() call.
+ */
 void btrfs_qgroup_convert_reserved_meta(struct btrfs_root *root, int num_bytes)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
 
-	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags) ||
+	if (btrfs_qgroup_mode(fs_info) == BTRFS_QGROUP_MODE_DISABLED ||
 	    !is_fstree(root->root_key.objectid))
 		return;
 	/* Same as btrfs_qgroup_free_meta_prealloc() */
@@ -4248,7 +4481,7 @@ int btrfs_qgroup_add_swapped_blocks(struct btrfs_trans_handle *trans,
 	int level = btrfs_header_level(subvol_parent) - 1;
 	int ret = 0;
 
-	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags))
+	if (!btrfs_qgroup_full_accounting(fs_info))
 		return 0;
 
 	if (btrfs_node_ptr_generation(subvol_parent, subvol_slot) >
@@ -4358,7 +4591,7 @@ int btrfs_qgroup_trace_subtree_after_cow(struct btrfs_trans_handle *trans,
 	int ret = 0;
 	int i;
 
-	if (!test_bit(BTRFS_FS_QUOTA_ENABLED, &fs_info->flags))
+	if (!btrfs_qgroup_full_accounting(fs_info))
 		return 0;
 	if (!is_fstree(root->root_key.objectid) || !root->reloc_root)
 		return 0;
@@ -4441,3 +4674,61 @@ void btrfs_qgroup_destroy_extent_records(struct btrfs_transaction *trans)
 	}
 	*root = RB_ROOT;
 }
+
+void btrfs_free_squota_rsv(struct btrfs_fs_info *fs_info, u64 root, u64 rsv_bytes)
+{
+	if (btrfs_qgroup_mode(fs_info) != BTRFS_QGROUP_MODE_SIMPLE)
+		return;
+
+	if (!is_fstree(root))
+		return;
+
+	btrfs_qgroup_free_refroot(fs_info, root, rsv_bytes, BTRFS_QGROUP_RSV_DATA);
+}
+
+int btrfs_record_squota_delta(struct btrfs_fs_info *fs_info,
+			      struct btrfs_squota_delta *delta)
+{
+	int ret;
+	struct btrfs_qgroup *qgroup;
+	struct btrfs_qgroup *qg;
+	LIST_HEAD(qgroup_list);
+	u64 root = delta->root;
+	u64 num_bytes = delta->num_bytes;
+	const int sign = (delta->is_inc ? 1 : -1);
+
+	if (btrfs_qgroup_mode(fs_info) != BTRFS_QGROUP_MODE_SIMPLE)
+		return 0;
+
+	if (!is_fstree(root))
+		return 0;
+
+	/* If the extent predates enabling quotas, don't count it. */
+	if (delta->generation < fs_info->qgroup_enable_gen)
+		return 0;
+
+	spin_lock(&fs_info->qgroup_lock);
+	qgroup = find_qgroup_rb(fs_info, root);
+	if (!qgroup) {
+		ret = -ENOENT;
+		goto out;
+	}
+
+	ret = 0;
+	qgroup_iterator_add(&qgroup_list, qgroup);
+	list_for_each_entry(qg, &qgroup_list, iterator) {
+		struct btrfs_qgroup_list *glist;
+
+		qg->excl += num_bytes * sign;
+		qg->rfer += num_bytes * sign;
+		qgroup_dirty(fs_info, qg);
+
+		list_for_each_entry(glist, &qg->groups, next_group)
+			qgroup_iterator_add(&qgroup_list, glist->group);
+	}
+	qgroup_iterator_clean(&qgroup_list);
+
+out:
+	spin_unlock(&fs_info->qgroup_lock);
+	return ret;
+}