Linux-libre 4.2-gnu

36 files changed, 2334 insertions, 1783 deletions

diff --git a/fs/btrfs/async-thread.c b/fs/btrfs/async-thread.c
index df9932b00..1ce06c849 100644
--- a/fs/btrfs/async-thread.c
+++ b/fs/btrfs/async-thread.c
@@ -85,6 +85,7 @@ BTRFS_WORK_HELPER(extent_refs_helper);
 BTRFS_WORK_HELPER(scrub_helper);
 BTRFS_WORK_HELPER(scrubwrc_helper);
 BTRFS_WORK_HELPER(scrubnc_helper);
+BTRFS_WORK_HELPER(scrubparity_helper);
 
 static struct __btrfs_workqueue *
 __btrfs_alloc_workqueue(const char *name, unsigned int flags, int max_active,
diff --git a/fs/btrfs/async-thread.h b/fs/btrfs/async-thread.h
index ec2ee477f..b0b093b6a 100644
--- a/fs/btrfs/async-thread.h
+++ b/fs/btrfs/async-thread.h
@@ -64,6 +64,8 @@ BTRFS_WORK_HELPER_PROTO(extent_refs_helper);
 BTRFS_WORK_HELPER_PROTO(scrub_helper);
 BTRFS_WORK_HELPER_PROTO(scrubwrc_helper);
 BTRFS_WORK_HELPER_PROTO(scrubnc_helper);
+BTRFS_WORK_HELPER_PROTO(scrubparity_helper);
+
 
 struct btrfs_workqueue *btrfs_alloc_workqueue(const char *name,
 					      unsigned int flags,
diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c
index 614aaa196..802fabb30 100644
--- a/fs/btrfs/backref.c
+++ b/fs/btrfs/backref.c
@@ -250,8 +250,12 @@ static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
 	 * the first item to check. But sometimes, we may enter it with
 	 * slot==nritems. In that case, go to the next leaf before we continue.
 	 */
-	if (path->slots[0] >= btrfs_header_nritems(path->nodes[0]))
-		ret = btrfs_next_old_leaf(root, path, time_seq);
+	if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
+		if (time_seq == (u64)-1)
+			ret = btrfs_next_leaf(root, path);
+		else
+			ret = btrfs_next_old_leaf(root, path, time_seq);
+	}
 
 	while (!ret && count < total_refs) {
 		eb = path->nodes[0];
@@ -291,7 +295,10 @@ static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
 			eie = NULL;
 		}
 next:
-		ret = btrfs_next_old_item(root, path, time_seq);
+		if (time_seq == (u64)-1)
+			ret = btrfs_next_item(root, path);
+		else
+			ret = btrfs_next_old_item(root, path, time_seq);
 	}
 
 	if (ret > 0)
@@ -334,6 +341,8 @@ static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info,
 
 	if (path->search_commit_root)
 		root_level = btrfs_header_level(root->commit_root);
+	else if (time_seq == (u64)-1)
+		root_level = btrfs_header_level(root->node);
 	else
 		root_level = btrfs_old_root_level(root, time_seq);
 
@@ -343,7 +352,12 @@ static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info,
 	}
 
 	path->lowest_level = level;
-	ret = btrfs_search_old_slot(root, &ref->key_for_search, path, time_seq);
+	if (time_seq == (u64)-1)
+		ret = btrfs_search_slot(NULL, root, &ref->key_for_search, path,
+					0, 0);
+	else
+		ret = btrfs_search_old_slot(root, &ref->key_for_search, path,
+					    time_seq);
 
 	/* root node has been locked, we can release @subvol_srcu safely here */
 	srcu_read_unlock(&fs_info->subvol_srcu, index);
@@ -491,7 +505,9 @@ static int __add_missing_keys(struct btrfs_fs_info *fs_info,
 		BUG_ON(!ref->wanted_disk_byte);
 		eb = read_tree_block(fs_info->tree_root, ref->wanted_disk_byte,
 				     0);
-		if (!eb || !extent_buffer_uptodate(eb)) {
+		if (IS_ERR(eb)) {
+			return PTR_ERR(eb);
+		} else if (!extent_buffer_uptodate(eb)) {
 			free_extent_buffer(eb);
 			return -EIO;
 		}
@@ -507,7 +523,7 @@ static int __add_missing_keys(struct btrfs_fs_info *fs_info,
 }
 
 /*
- * merge two lists of backrefs and adjust counts accordingly
+ * merge backrefs and adjust counts accordingly
  *
  * mode = 1: merge identical keys, if key is set
  *    FIXME: if we add more keys in __add_prelim_ref, we can merge more here.
@@ -535,9 +551,9 @@ static void __merge_refs(struct list_head *head, int mode)
 
 			ref2 = list_entry(pos2, struct __prelim_ref, list);
 
+			if (!ref_for_same_block(ref1, ref2))
+				continue;
 			if (mode == 1) {
-				if (!ref_for_same_block(ref1, ref2))
-					continue;
 				if (!ref1->parent && ref2->parent) {
 					xchg = ref1;
 					ref1 = ref2;
@@ -572,8 +588,8 @@ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
 			      struct list_head *prefs, u64 *total_refs,
 			      u64 inum)
 {
+	struct btrfs_delayed_ref_node *node;
 	struct btrfs_delayed_extent_op *extent_op = head->extent_op;
-	struct rb_node *n = &head->node.rb_node;
 	struct btrfs_key key;
 	struct btrfs_key op_key = {0};
 	int sgn;
@@ -583,12 +599,7 @@ static int __add_delayed_refs(struct btrfs_delayed_ref_head *head, u64 seq,
 		btrfs_disk_key_to_cpu(&op_key, &extent_op->key);
 
 	spin_lock(&head->lock);
-	n = rb_first(&head->ref_root);
-	while (n) {
-		struct btrfs_delayed_ref_node *node;
-		node = rb_entry(n, struct btrfs_delayed_ref_node,
-				rb_node);
-		n = rb_next(n);
+	list_for_each_entry(node, &head->ref_list, list) {
 		if (node->seq > seq)
 			continue;
 
@@ -882,6 +893,11 @@ static int __add_keyed_refs(struct btrfs_fs_info *fs_info,
  *
  * NOTE: This can return values > 0
  *
+ * If time_seq is set to (u64)-1, it will not search delayed_refs, and behave
+ * much like trans == NULL case, the difference only lies in it will not
+ * commit root.
+ * The special case is for qgroup to search roots in commit_transaction().
+ *
  * FIXME some caching might speed things up
  */
 static int find_parent_nodes(struct btrfs_trans_handle *trans,
@@ -920,6 +936,9 @@ static int find_parent_nodes(struct btrfs_trans_handle *trans,
 		path->skip_locking = 1;
 	}
 
+	if (time_seq == (u64)-1)
+		path->skip_locking = 1;
+
 	/*
 	 * grab both a lock on the path and a lock on the delayed ref head.
 	 * We need both to get a consistent picture of how the refs look
@@ -934,9 +953,10 @@ again:
 	BUG_ON(ret == 0);
 
 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
-	if (trans && likely(trans->type != __TRANS_DUMMY)) {
+	if (trans && likely(trans->type != __TRANS_DUMMY) &&
+	    time_seq != (u64)-1) {
 #else
-	if (trans) {
+	if (trans && time_seq != (u64)-1) {
 #endif
 		/*
 		 * look if there are updates for this ref queued and lock the
@@ -1034,7 +1054,10 @@ again:
 
 				eb = read_tree_block(fs_info->extent_root,
 							   ref->parent, 0);
-				if (!eb || !extent_buffer_uptodate(eb)) {
+				if (IS_ERR(eb)) {
+					ret = PTR_ERR(eb);
+					goto out;
+				} else if (!extent_buffer_uptodate(eb)) {
 					free_extent_buffer(eb);
 					ret = -EIO;
 					goto out;
diff --git a/fs/btrfs/btrfs_inode.h b/fs/btrfs/btrfs_inode.h
index 0ef5cc13f..81220b220 100644
--- a/fs/btrfs/btrfs_inode.h
+++ b/fs/btrfs/btrfs_inode.h
@@ -44,6 +44,8 @@
 #define BTRFS_INODE_IN_DELALLOC_LIST		9
 #define BTRFS_INODE_READDIO_NEED_LOCK		10
 #define BTRFS_INODE_HAS_PROPS		        11
+/* DIO is ready to submit */
+#define BTRFS_INODE_DIO_READY		        12
 /*
  * The following 3 bits are meant only for the btree inode.
  * When any of them is set, it means an error happened while writing an
diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c
index 0f11ebc92..54114b488 100644
--- a/fs/btrfs/ctree.c
+++ b/fs/btrfs/ctree.c
@@ -1439,8 +1439,9 @@ get_old_root(struct btrfs_root *root, u64 time_seq)
 		btrfs_tree_read_unlock(eb_root);
 		free_extent_buffer(eb_root);
 		old = read_tree_block(root, logical, 0);
-		if (WARN_ON(!old || !extent_buffer_uptodate(old))) {
-			free_extent_buffer(old);
+		if (WARN_ON(IS_ERR(old) || !extent_buffer_uptodate(old))) {
+			if (!IS_ERR(old))
+				free_extent_buffer(old);
 			btrfs_warn(root->fs_info,
 				"failed to read tree block %llu from get_old_root", logical);
 		} else {
@@ -1685,7 +1686,9 @@ int btrfs_realloc_node(struct btrfs_trans_handle *trans,
 		if (!cur || !uptodate) {
 			if (!cur) {
 				cur = read_tree_block(root, blocknr, gen);
-				if (!cur || !extent_buffer_uptodate(cur)) {
+				if (IS_ERR(cur)) {
+					return PTR_ERR(cur);
+				} else if (!extent_buffer_uptodate(cur)) {
 					free_extent_buffer(cur);
 					return -EIO;
 				}
@@ -1864,8 +1867,9 @@ static noinline struct extent_buffer *read_node_slot(struct btrfs_root *root,
 
 	eb = read_tree_block(root, btrfs_node_blockptr(parent, slot),
 			     btrfs_node_ptr_generation(parent, slot));
-	if (eb && !extent_buffer_uptodate(eb)) {
-		free_extent_buffer(eb);
+	if (IS_ERR(eb) || !extent_buffer_uptodate(eb)) {
+		if (!IS_ERR(eb))
+			free_extent_buffer(eb);
 		eb = NULL;
 	}
 
@@ -2494,7 +2498,7 @@ read_block_for_search(struct btrfs_trans_handle *trans,
 
 	ret = -EAGAIN;
 	tmp = read_tree_block(root, blocknr, 0);
-	if (tmp) {
+	if (!IS_ERR(tmp)) {
 		/*
 		 * If the read above didn't mark this buffer up to date,
 		 * it will never end up being up to date.  Set ret to EIO now
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 6f364e1d8..aac314e14 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -174,7 +174,7 @@ struct btrfs_ordered_sum;
 /* csum types */
 #define BTRFS_CSUM_TYPE_CRC32	0
 
-static int btrfs_csum_sizes[] = { 4, 0 };
+static int btrfs_csum_sizes[] = { 4 };
 
 /* four bytes for CRC32 */
 #define BTRFS_EMPTY_DIR_SIZE 0
@@ -1619,10 +1619,7 @@ struct btrfs_fs_info {
 	struct task_struct *cleaner_kthread;
 	int thread_pool_size;
 
-	struct kobject super_kobj;
 	struct kobject *space_info_kobj;
-	struct kobject *device_dir_kobj;
-	struct completion kobj_unregister;
 	int do_barriers;
 	int closing;
 	int log_root_recovering;
@@ -1698,6 +1695,7 @@ struct btrfs_fs_info {
 	struct btrfs_workqueue *scrub_workers;
 	struct btrfs_workqueue *scrub_wr_completion_workers;
 	struct btrfs_workqueue *scrub_nocow_workers;
+	struct btrfs_workqueue *scrub_parity_workers;
 
 #ifdef CONFIG_BTRFS_FS_CHECK_INTEGRITY
 	u32 check_integrity_print_mask;
@@ -1735,7 +1733,7 @@ struct btrfs_fs_info {
 	/* list of dirty qgroups to be written at next commit */
 	struct list_head dirty_qgroups;
 
-	/* used by btrfs_qgroup_record_ref for an efficient tree traversal */
+	/* used by qgroup for an efficient tree traversal */
 	u64 qgroup_seq;
 
 	/* qgroup rescan items */
@@ -1780,6 +1778,7 @@ struct btrfs_fs_info {
 	spinlock_t unused_bgs_lock;
 	struct list_head unused_bgs;
 	struct mutex unused_bg_unpin_mutex;
+	struct mutex delete_unused_bgs_mutex;
 
 	/* For btrfs to record security options */
 	struct security_mnt_opts security_opts;
@@ -3458,6 +3457,7 @@ int btrfs_check_data_free_space(struct inode *inode, u64 bytes, u64 write_bytes)
 void btrfs_free_reserved_data_space(struct inode *inode, u64 bytes);
 void btrfs_trans_release_metadata(struct btrfs_trans_handle *trans,
 				struct btrfs_root *root);
+void btrfs_trans_release_chunk_metadata(struct btrfs_trans_handle *trans);
 int btrfs_orphan_reserve_metadata(struct btrfs_trans_handle *trans,
 				  struct inode *inode);
 void btrfs_orphan_release_metadata(struct inode *inode);
@@ -3515,6 +3515,9 @@ int btrfs_delayed_refs_qgroup_accounting(struct btrfs_trans_handle *trans,
 int __get_raid_index(u64 flags);
 int btrfs_start_write_no_snapshoting(struct btrfs_root *root);
 void btrfs_end_write_no_snapshoting(struct btrfs_root *root);
+void check_system_chunk(struct btrfs_trans_handle *trans,
+			struct btrfs_root *root,
+			const u64 type);
 /* ctree.c */
 int btrfs_bin_search(struct extent_buffer *eb, struct btrfs_key *key,
 		     int level, int *slot);
@@ -4050,6 +4053,7 @@ void btrfs_printk(const struct btrfs_fs_info *fs_info, const char *fmt, ...)
 
 #ifdef CONFIG_BTRFS_ASSERT
 
+__cold
 static inline void assfail(char *expr, char *file, int line)
 {
 	pr_err("BTRFS: assertion failed: %s, file: %s, line: %d",
@@ -4065,10 +4069,12 @@ static inline void assfail(char *expr, char *file, int line)
 
 #define btrfs_assert()
 __printf(5, 6)
+__cold
 void __btrfs_std_error(struct btrfs_fs_info *fs_info, const char *function,
 		     unsigned int line, int errno, const char *fmt, ...);
 
 
+__cold
 void __btrfs_abort_transaction(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *root, const char *function,
 			       unsigned int line, int errno);
@@ -4111,11 +4117,17 @@ static inline int __btrfs_fs_incompat(struct btrfs_fs_info *fs_info, u64 flag)
  * Call btrfs_abort_transaction as early as possible when an error condition is
  * detected, that way the exact line number is reported.
  */
-
 #define btrfs_abort_transaction(trans, root, errno)		\
 do {								\
-	__btrfs_abort_transaction(trans, root, __func__,	\
-				  __LINE__, errno);		\
+	/* Report first abort since mount */			\
+	if (!test_and_set_bit(BTRFS_FS_STATE_TRANS_ABORTED,	\
+			&((root)->fs_info->fs_state))) {	\
+		WARN(1, KERN_DEBUG				\
+		"BTRFS: Transaction aborted (error %d)\n",	\
+		(errno));					\
+	}							\
+	__btrfs_abort_transaction((trans), (root), __func__,	\
+				  __LINE__, (errno));		\
 } while (0)
 
 #define btrfs_std_error(fs_info, errno)				\
@@ -4132,6 +4144,7 @@ do {								\
 } while (0)
 
 __printf(5, 6)
+__cold
 void __btrfs_panic(struct btrfs_fs_info *fs_info, const char *function,
 		   unsigned int line, int errno, const char *fmt, ...);
 
diff --git a/fs/btrfs/delayed-ref.c b/fs/btrfs/delayed-ref.c
index 8f8ed7d20..ac3e81da6 100644
--- a/fs/btrfs/delayed-ref.c
+++ b/fs/btrfs/delayed-ref.c
@@ -22,6 +22,7 @@
 #include "ctree.h"
 #include "delayed-ref.h"
 #include "transaction.h"
+#include "qgroup.h"
 
 struct kmem_cache *btrfs_delayed_ref_head_cachep;
 struct kmem_cache *btrfs_delayed_tree_ref_cachep;
@@ -84,87 +85,6 @@ static int comp_data_refs(struct btrfs_delayed_data_ref *ref2,
 	return 0;
 }
 
-/*
- * entries in the rb tree are ordered by the byte number of the extent,
- * type of the delayed backrefs and content of delayed backrefs.
- */
-static int comp_entry(struct btrfs_delayed_ref_node *ref2,
-		      struct btrfs_delayed_ref_node *ref1,
-		      bool compare_seq)
-{
-	if (ref1->bytenr < ref2->bytenr)
-		return -1;
-	if (ref1->bytenr > ref2->bytenr)
-		return 1;
-	if (ref1->is_head && ref2->is_head)
-		return 0;
-	if (ref2->is_head)
-		return -1;
-	if (ref1->is_head)
-		return 1;
-	if (ref1->type < ref2->type)
-		return -1;
-	if (ref1->type > ref2->type)
-		return 1;
-	if (ref1->no_quota > ref2->no_quota)
-		return 1;
-	if (ref1->no_quota < ref2->no_quota)
-		return -1;
-	/* merging of sequenced refs is not allowed */
-	if (compare_seq) {
-		if (ref1->seq < ref2->seq)
-			return -1;
-		if (ref1->seq > ref2->seq)
-			return 1;
-	}
-	if (ref1->type == BTRFS_TREE_BLOCK_REF_KEY ||
-	    ref1->type == BTRFS_SHARED_BLOCK_REF_KEY) {
-		return comp_tree_refs(btrfs_delayed_node_to_tree_ref(ref2),
-				      btrfs_delayed_node_to_tree_ref(ref1),
-				      ref1->type);
-	} else if (ref1->type == BTRFS_EXTENT_DATA_REF_KEY ||
-		   ref1->type == BTRFS_SHARED_DATA_REF_KEY) {
-		return comp_data_refs(btrfs_delayed_node_to_data_ref(ref2),
-				      btrfs_delayed_node_to_data_ref(ref1));
-	}
-	BUG();
-	return 0;
-}
-
-/*
- * insert a new ref into the rbtree.  This returns any existing refs
- * for the same (bytenr,parent) tuple, or NULL if the new node was properly
- * inserted.
- */
-static struct btrfs_delayed_ref_node *tree_insert(struct rb_root *root,
-						  struct rb_node *node)
-{
-	struct rb_node **p = &root->rb_node;
-	struct rb_node *parent_node = NULL;
-	struct btrfs_delayed_ref_node *entry;
-	struct btrfs_delayed_ref_node *ins;
-	int cmp;
-
-	ins = rb_entry(node, struct btrfs_delayed_ref_node, rb_node);
-	while (*p) {
-		parent_node = *p;
-		entry = rb_entry(parent_node, struct btrfs_delayed_ref_node,
-				 rb_node);
-
-		cmp = comp_entry(entry, ins, 1);
-		if (cmp < 0)
-			p = &(*p)->rb_left;
-		else if (cmp > 0)
-			p = &(*p)->rb_right;
-		else
-			return entry;
-	}
-
-	rb_link_node(node, parent_node, p);
-	rb_insert_color(node, root);
-	return NULL;
-}
-
 /* insert a new ref to head ref rbtree */
 static struct btrfs_delayed_ref_head *htree_insert(struct rb_root *root,
 						   struct rb_node *node)
@@ -268,7 +188,7 @@ static inline void drop_delayed_ref(struct btrfs_trans_handle *trans,
 		rb_erase(&head->href_node, &delayed_refs->href_root);
 	} else {
 		assert_spin_locked(&head->lock);
-		rb_erase(&ref->rb_node, &head->ref_root);
+		list_del(&ref->list);
 	}
 	ref->in_tree = 0;
 	btrfs_put_delayed_ref(ref);
@@ -277,99 +197,6 @@ static inline void drop_delayed_ref(struct btrfs_trans_handle *trans,
 		trans->delayed_ref_updates--;
 }
 
-static int merge_ref(struct btrfs_trans_handle *trans,
-		     struct btrfs_delayed_ref_root *delayed_refs,
-		     struct btrfs_delayed_ref_head *head,
-		     struct btrfs_delayed_ref_node *ref, u64 seq)
-{
-	struct rb_node *node;
-	int mod = 0;
-	int done = 0;
-
-	node = rb_next(&ref->rb_node);
-	while (!done && node) {
-		struct btrfs_delayed_ref_node *next;
-
-		next = rb_entry(node, struct btrfs_delayed_ref_node, rb_node);
-		node = rb_next(node);
-		if (seq && next->seq >= seq)
-			break;
-		if (comp_entry(ref, next, 0))
-			continue;
-
-		if (ref->action == next->action) {
-			mod = next->ref_mod;
-		} else {
-			if (ref->ref_mod < next->ref_mod) {
-				struct btrfs_delayed_ref_node *tmp;
-
-				tmp = ref;
-				ref = next;
-				next = tmp;
-				done = 1;
-			}
-			mod = -next->ref_mod;
-		}
-
-		drop_delayed_ref(trans, delayed_refs, head, next);
-		ref->ref_mod += mod;
-		if (ref->ref_mod == 0) {
-			drop_delayed_ref(trans, delayed_refs, head, ref);
-			done = 1;
-		} else {
-			/*
-			 * You can't have multiples of the same ref on a tree
-			 * block.
-			 */
-			WARN_ON(ref->type == BTRFS_TREE_BLOCK_REF_KEY ||
-				ref->type == BTRFS_SHARED_BLOCK_REF_KEY);
-		}
-	}
-	return done;
-}
-
-void btrfs_merge_delayed_refs(struct btrfs_trans_handle *trans,
-			      struct btrfs_fs_info *fs_info,
-			      struct btrfs_delayed_ref_root *delayed_refs,
-			      struct btrfs_delayed_ref_head *head)
-{
-	struct rb_node *node;
-	u64 seq = 0;
-
-	assert_spin_locked(&head->lock);
-	/*
-	 * We don't have too much refs to merge in the case of delayed data
-	 * refs.
-	 */
-	if (head->is_data)
-		return;
-
-	spin_lock(&fs_info->tree_mod_seq_lock);
-	if (!list_empty(&fs_info->tree_mod_seq_list)) {
-		struct seq_list *elem;
-
-		elem = list_first_entry(&fs_info->tree_mod_seq_list,
-					struct seq_list, list);
-		seq = elem->seq;
-	}
-	spin_unlock(&fs_info->tree_mod_seq_lock);
-
-	node = rb_first(&head->ref_root);
-	while (node) {
-		struct btrfs_delayed_ref_node *ref;
-
-		ref = rb_entry(node, struct btrfs_delayed_ref_node,
-			       rb_node);
-		/* We can't merge refs that are outside of our seq count */
-		if (seq && ref->seq >= seq)
-			break;
-		if (merge_ref(trans, delayed_refs, head, ref, seq))
-			node = rb_first(&head->ref_root);
-		else
-			node = rb_next(&ref->rb_node);
-	}
-}
-
 int btrfs_check_delayed_seq(struct btrfs_fs_info *fs_info,
 			    struct btrfs_delayed_ref_root *delayed_refs,
 			    u64 seq)
@@ -443,45 +270,71 @@ again:
 }
 
 /*
- * helper function to update an extent delayed ref in the
- * rbtree.  existing and update must both have the same
- * bytenr and parent
+ * Helper to insert the ref_node to the tail or merge with tail.
  *
- * This may free existing if the update cancels out whatever
- * operation it was doing.
+ * Return 0 for insert.
+ * Return >0 for merge.
  */
-static noinline void
-update_existing_ref(struct btrfs_trans_handle *trans,
-		    struct btrfs_delayed_ref_root *delayed_refs,
-		    struct btrfs_delayed_ref_head *head,
-		    struct btrfs_delayed_ref_node *existing,
-		    struct btrfs_delayed_ref_node *update)
+static int
+add_delayed_ref_tail_merge(struct btrfs_trans_handle *trans,
+			   struct btrfs_delayed_ref_root *root,
+			   struct btrfs_delayed_ref_head *href,
+			   struct btrfs_delayed_ref_node *ref)
 {
-	if (update->action != existing->action) {
-		/*
-		 * this is effectively undoing either an add or a
-		 * drop.  We decrement the ref_mod, and if it goes
-		 * down to zero we just delete the entry without
-		 * every changing the extent allocation tree.
-		 */
-		existing->ref_mod--;
-		if (existing->ref_mod == 0)
-			drop_delayed_ref(trans, delayed_refs, head, existing);
-		else
-			WARN_ON(existing->type == BTRFS_TREE_BLOCK_REF_KEY ||
-				existing->type == BTRFS_SHARED_BLOCK_REF_KEY);
+	struct btrfs_delayed_ref_node *exist;
+	int mod;
+	int ret = 0;
+
+	spin_lock(&href->lock);
+	/* Check whether we can merge the tail node with ref */
+	if (list_empty(&href->ref_list))
+		goto add_tail;
+	exist = list_entry(href->ref_list.prev, struct btrfs_delayed_ref_node,
+			   list);
+	/* No need to compare bytenr nor is_head */
+	if (exist->type != ref->type || exist->no_quota != ref->no_quota ||
+	    exist->seq != ref->seq)
+		goto add_tail;
+
+	if ((exist->type == BTRFS_TREE_BLOCK_REF_KEY ||
+	     exist->type == BTRFS_SHARED_BLOCK_REF_KEY) &&
+	    comp_tree_refs(btrfs_delayed_node_to_tree_ref(exist),
+			   btrfs_delayed_node_to_tree_ref(ref),
+			   ref->type))
+		goto add_tail;
+	if ((exist->type == BTRFS_EXTENT_DATA_REF_KEY ||
+	     exist->type == BTRFS_SHARED_DATA_REF_KEY) &&
+	    comp_data_refs(btrfs_delayed_node_to_data_ref(exist),
+			   btrfs_delayed_node_to_data_ref(ref)))
+		goto add_tail;
+
+	/* Now we are sure we can merge */
+	ret = 1;
+	if (exist->action == ref->action) {
+		mod = ref->ref_mod;
 	} else {
-		WARN_ON(existing->type == BTRFS_TREE_BLOCK_REF_KEY ||
-			existing->type == BTRFS_SHARED_BLOCK_REF_KEY);
-		/*
-		 * the action on the existing ref matches
-		 * the action on the ref we're trying to add.
-		 * Bump the ref_mod by one so the backref that
-		 * is eventually added/removed has the correct
-		 * reference count
-		 */
-		existing->ref_mod += update->ref_mod;
+		/* Need to change action */
+		if (exist->ref_mod < ref->ref_mod) {
+			exist->action = ref->action;
+			mod = -exist->ref_mod;
+			exist->ref_mod = ref->ref_mod;
+		} else
+			mod = -ref->ref_mod;
 	}
+	exist->ref_mod += mod;
+
+	/* remove existing tail if its ref_mod is zero */
+	if (exist->ref_mod == 0)
+		drop_delayed_ref(trans, root, href, exist);
+	spin_unlock(&href->lock);
+	return ret;
+
+add_tail:
+	list_add_tail(&ref->list, &href->ref_list);
+	atomic_inc(&root->num_entries);
+	trans->delayed_ref_updates++;
+	spin_unlock(&href->lock);
+	return ret;
 }
 
 /*
@@ -568,12 +421,14 @@ update_existing_head_ref(struct btrfs_delayed_ref_root *delayed_refs,
 static noinline struct btrfs_delayed_ref_head *
 add_delayed_ref_head(struct btrfs_fs_info *fs_info,
 		     struct btrfs_trans_handle *trans,
-		     struct btrfs_delayed_ref_node *ref, u64 bytenr,
-		     u64 num_bytes, int action, int is_data)
+		     struct btrfs_delayed_ref_node *ref,
+		     struct btrfs_qgroup_extent_record *qrecord,
+		     u64 bytenr, u64 num_bytes, int action, int is_data)
 {
 	struct btrfs_delayed_ref_head *existing;
 	struct btrfs_delayed_ref_head *head_ref = NULL;
 	struct btrfs_delayed_ref_root *delayed_refs;
+	struct btrfs_qgroup_extent_record *qexisting;
 	int count_mod = 1;
 	int must_insert_reserved = 0;
 
@@ -618,10 +473,22 @@ add_delayed_ref_head(struct btrfs_fs_info *fs_info,
 	head_ref = btrfs_delayed_node_to_head(ref);
 	head_ref->must_insert_reserved = must_insert_reserved;
 	head_ref->is_data = is_data;
-	head_ref->ref_root = RB_ROOT;
+	INIT_LIST_HEAD(&head_ref->ref_list);
 	head_ref->processing = 0;
 	head_ref->total_ref_mod = count_mod;
 
+	/* Record qgroup extent info if provided */
+	if (qrecord) {
+		qrecord->bytenr = bytenr;
+		qrecord->num_bytes = num_bytes;
+		qrecord->old_roots = NULL;
+
+		qexisting = btrfs_qgroup_insert_dirty_extent(delayed_refs,
+							     qrecord);
+		if (qexisting)
+			kfree(qrecord);
+	}
+
 	spin_lock_init(&head_ref->lock);
 	mutex_init(&head_ref->mutex);
 
@@ -659,10 +526,10 @@ add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 		     u64 num_bytes, u64 parent, u64 ref_root, int level,
 		     int action, int no_quota)
 {
-	struct btrfs_delayed_ref_node *existing;
 	struct btrfs_delayed_tree_ref *full_ref;
 	struct btrfs_delayed_ref_root *delayed_refs;
 	u64 seq = 0;
+	int ret;
 
 	if (action == BTRFS_ADD_DELAYED_EXTENT)
 		action = BTRFS_ADD_DELAYED_REF;
@@ -693,21 +560,14 @@ add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 
 	trace_add_delayed_tree_ref(ref, full_ref, action);
 
-	spin_lock(&head_ref->lock);
-	existing = tree_insert(&head_ref->ref_root, &ref->rb_node);
-	if (existing) {
-		update_existing_ref(trans, delayed_refs, head_ref, existing,
-				    ref);
-		/*
-		 * we've updated the existing ref, free the newly
-		 * allocated ref
-		 */
+	ret = add_delayed_ref_tail_merge(trans, delayed_refs, head_ref, ref);
+
+	/*
+	 * XXX: memory should be freed at the same level allocated.
+	 * But bad practice is anywhere... Follow it now. Need cleanup.
+	 */
+	if (ret > 0)
 		kmem_cache_free(btrfs_delayed_tree_ref_cachep, full_ref);
-	} else {
-		atomic_inc(&delayed_refs->num_entries);
-		trans->delayed_ref_updates++;
-	}
-	spin_unlock(&head_ref->lock);
 }
 
 /*
@@ -721,10 +581,10 @@ add_delayed_data_ref(struct btrfs_fs_info *fs_info,
 		     u64 num_bytes, u64 parent, u64 ref_root, u64 owner,
 		     u64 offset, int action, int no_quota)
 {
-	struct btrfs_delayed_ref_node *existing;
 	struct btrfs_delayed_data_ref *full_ref;
 	struct btrfs_delayed_ref_root *delayed_refs;
 	u64 seq = 0;
+	int ret;
 
 	if (action == BTRFS_ADD_DELAYED_EXTENT)
 		action = BTRFS_ADD_DELAYED_REF;
@@ -758,21 +618,10 @@ add_delayed_data_ref(struct btrfs_fs_info *fs_info,
 
 	trace_add_delayed_data_ref(ref, full_ref, action);
 
-	spin_lock(&head_ref->lock);
-	existing = tree_insert(&head_ref->ref_root, &ref->rb_node);
-	if (existing) {
-		update_existing_ref(trans, delayed_refs, head_ref, existing,
-				    ref);
-		/*
-		 * we've updated the existing ref, free the newly
-		 * allocated ref
-		 */
+	ret = add_delayed_ref_tail_merge(trans, delayed_refs, head_ref, ref);
+
+	if (ret > 0)
 		kmem_cache_free(btrfs_delayed_data_ref_cachep, full_ref);
-	} else {
-		atomic_inc(&delayed_refs->num_entries);
-		trans->delayed_ref_updates++;
-	}
-	spin_unlock(&head_ref->lock);
 }
 
 /*
@@ -790,6 +639,7 @@ int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 	struct btrfs_delayed_tree_ref *ref;
 	struct btrfs_delayed_ref_head *head_ref;
 	struct btrfs_delayed_ref_root *delayed_refs;
+	struct btrfs_qgroup_extent_record *record = NULL;
 
 	if (!is_fstree(ref_root) || !fs_info->quota_enabled)
 		no_quota = 0;
@@ -800,9 +650,13 @@ int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 		return -ENOMEM;
 
 	head_ref = kmem_cache_alloc(btrfs_delayed_ref_head_cachep, GFP_NOFS);
-	if (!head_ref) {
-		kmem_cache_free(btrfs_delayed_tree_ref_cachep, ref);
-		return -ENOMEM;
+	if (!head_ref)
+		goto free_ref;
+
+	if (fs_info->quota_enabled && is_fstree(ref_root)) {
+		record = kmalloc(sizeof(*record), GFP_NOFS);
+		if (!record)
+			goto free_head_ref;
 	}
 
 	head_ref->extent_op = extent_op;
@@ -814,7 +668,7 @@ int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 	 * insert both the head node and the new ref without dropping
 	 * the spin lock
 	 */
-	head_ref = add_delayed_ref_head(fs_info, trans, &head_ref->node,
+	head_ref = add_delayed_ref_head(fs_info, trans, &head_ref->node, record,
 					bytenr, num_bytes, action, 0);
 
 	add_delayed_tree_ref(fs_info, trans, head_ref, &ref->node, bytenr,
@@ -823,6 +677,13 @@ int btrfs_add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 	spin_unlock(&delayed_refs->lock);
 
 	return 0;
+
+free_head_ref:
+	kmem_cache_free(btrfs_delayed_ref_head_cachep, head_ref);
+free_ref:
+	kmem_cache_free(btrfs_delayed_tree_ref_cachep, ref);
+
+	return -ENOMEM;
 }
 
 /*
@@ -839,6 +700,7 @@ int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info,
 	struct btrfs_delayed_data_ref *ref;
 	struct btrfs_delayed_ref_head *head_ref;
 	struct btrfs_delayed_ref_root *delayed_refs;
+	struct btrfs_qgroup_extent_record *record = NULL;
 
 	if (!is_fstree(ref_root) || !fs_info->quota_enabled)
 		no_quota = 0;
@@ -854,6 +716,16 @@ int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info,
 		return -ENOMEM;
 	}
 
+	if (fs_info->quota_enabled && is_fstree(ref_root)) {
+		record = kmalloc(sizeof(*record), GFP_NOFS);
+		if (!record) {
+			kmem_cache_free(btrfs_delayed_data_ref_cachep, ref);
+			kmem_cache_free(btrfs_delayed_ref_head_cachep,
+					head_ref);
+			return -ENOMEM;
+		}
+	}
+
 	head_ref->extent_op = extent_op;
 
 	delayed_refs = &trans->transaction->delayed_refs;
@@ -863,7 +735,7 @@ int btrfs_add_delayed_data_ref(struct btrfs_fs_info *fs_info,
 	 * insert both the head node and the new ref without dropping
 	 * the spin lock
 	 */
-	head_ref = add_delayed_ref_head(fs_info, trans, &head_ref->node,
+	head_ref = add_delayed_ref_head(fs_info, trans, &head_ref->node, record,
 					bytenr, num_bytes, action, 1);
 
 	add_delayed_data_ref(fs_info, trans, head_ref, &ref->node, bytenr,
@@ -891,9 +763,9 @@ int btrfs_add_delayed_extent_op(struct btrfs_fs_info *fs_info,
 	delayed_refs = &trans->transaction->delayed_refs;
 	spin_lock(&delayed_refs->lock);
 
-	add_delayed_ref_head(fs_info, trans, &head_ref->node, bytenr,
-				   num_bytes, BTRFS_UPDATE_DELAYED_HEAD,
-				   extent_op->is_data);
+	add_delayed_ref_head(fs_info, trans, &head_ref->node, NULL, bytenr,
+			     num_bytes, BTRFS_UPDATE_DELAYED_HEAD,
+			     extent_op->is_data);
 
 	spin_unlock(&delayed_refs->lock);
 	return 0;
diff --git a/fs/btrfs/delayed-ref.h b/fs/btrfs/delayed-ref.h
index 5eb089239..13fb5e609 100644
--- a/fs/btrfs/delayed-ref.h
+++ b/fs/btrfs/delayed-ref.h
@@ -24,9 +24,25 @@
 #define BTRFS_ADD_DELAYED_EXTENT 3 /* record a full extent allocation */
 #define BTRFS_UPDATE_DELAYED_HEAD 4 /* not changing ref count on head ref */
 
+/*
+ * XXX: Qu: I really hate the design that ref_head and tree/data ref shares the
+ * same ref_node structure.
+ * Ref_head is in a higher logic level than tree/data ref, and duplicated
+ * bytenr/num_bytes in ref_node is really a waste or memory, they should be
+ * referred from ref_head.
+ * This gets more disgusting after we use list to store tree/data ref in
+ * ref_head. Must clean this mess up later.
+ */
 struct btrfs_delayed_ref_node {
+	/*
+	 * ref_head use rb tree, stored in ref_root->href.
+	 * indexed by bytenr
+	 */
 	struct rb_node rb_node;
 
+	/*data/tree ref use list, stored in ref_head->ref_list. */
+	struct list_head list;
+
 	/* the starting bytenr of the extent */
 	u64 bytenr;
 
@@ -83,7 +99,7 @@ struct btrfs_delayed_ref_head {
 	struct mutex mutex;
 
 	spinlock_t lock;
-	struct rb_root ref_root;
+	struct list_head ref_list;
 
 	struct rb_node href_node;
 
@@ -132,6 +148,9 @@ struct btrfs_delayed_ref_root {
 	/* head ref rbtree */
 	struct rb_root href_root;
 
+	/* dirty extent records */
+	struct rb_root dirty_extent_root;
+
 	/* this spin lock protects the rbtree and the entries inside */
 	spinlock_t lock;
 
@@ -156,6 +175,14 @@ struct btrfs_delayed_ref_root {
 	int flushing;
 
 	u64 run_delayed_start;
+
+	/*
+	 * To make qgroup to skip given root.
+	 * This is for snapshot, as btrfs_qgroup_inherit() will manully
+	 * modify counters for snapshot and its source, so we should skip
+	 * the snapshot in new_root/old_roots or it will get calculated twice
+	 */
+	u64 qgroup_to_skip;
 };
 
 extern struct kmem_cache *btrfs_delayed_ref_head_cachep;
diff --git a/fs/btrfs/dev-replace.c b/fs/btrfs/dev-replace.c
index 0573848c7..564a7de17 100644
--- a/fs/btrfs/dev-replace.c
+++ b/fs/btrfs/dev-replace.c
@@ -376,6 +376,10 @@ int btrfs_dev_replace_start(struct btrfs_root *root,
 	WARN_ON(!tgt_device);
 	dev_replace->tgtdev = tgt_device;
 
+	ret = btrfs_kobj_add_device(tgt_device->fs_devices, tgt_device);
+	if (ret)
+		btrfs_err(root->fs_info, "kobj add dev failed %d\n", ret);
+
 	printk_in_rcu(KERN_INFO
 		      "BTRFS: dev_replace from %s (devid %llu) to %s started\n",
 		      src_device->missing ? "<missing disk>" :
@@ -583,8 +587,7 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 	mutex_unlock(&uuid_mutex);
 
 	/* replace the sysfs entry */
-	btrfs_kobj_rm_device(fs_info, src_device);
-	btrfs_kobj_add_device(fs_info, tgt_device);
+	btrfs_kobj_rm_device(fs_info->fs_devices, src_device);
 	btrfs_rm_dev_replace_free_srcdev(fs_info, src_device);
 
 	/* write back the superblocks */
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 2ef9a4b72..f556c3732 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -1149,12 +1149,12 @@ struct extent_buffer *read_tree_block(struct btrfs_root *root, u64 bytenr,
 
 	buf = btrfs_find_create_tree_block(root, bytenr);
 	if (!buf)
-		return NULL;
+		return ERR_PTR(-ENOMEM);
 
 	ret = btree_read_extent_buffer_pages(root, buf, 0, parent_transid);
 	if (ret) {
 		free_extent_buffer(buf);
-		return NULL;
+		return ERR_PTR(ret);
 	}
 	return buf;
 
@@ -1509,20 +1509,19 @@ static struct btrfs_root *btrfs_read_tree_root(struct btrfs_root *tree_root,
 	generation = btrfs_root_generation(&root->root_item);
 	root->node = read_tree_block(root, btrfs_root_bytenr(&root->root_item),
 				     generation);
-	if (!root->node) {
-		ret = -ENOMEM;
+	if (IS_ERR(root->node)) {
+		ret = PTR_ERR(root->node);
 		goto find_fail;
 	} else if (!btrfs_buffer_uptodate(root->node, generation, 0)) {
 		ret = -EIO;
-		goto read_fail;
+		free_extent_buffer(root->node);
+		goto find_fail;
 	}
 	root->commit_root = btrfs_root_node(root);
 out:
 	btrfs_free_path(path);
 	return root;
 
-read_fail:
-	free_extent_buffer(root->node);
 find_fail:
 	kfree(root);
 alloc_fail:
@@ -1745,13 +1744,14 @@ static void end_workqueue_fn(struct btrfs_work *work)
 	bio->bi_private = end_io_wq->private;
 	bio->bi_end_io = end_io_wq->end_io;
 	kmem_cache_free(btrfs_end_io_wq_cache, end_io_wq);
-	bio_endio_nodec(bio, error);
+	bio_endio(bio, error);
 }
 
 static int cleaner_kthread(void *arg)
 {
 	struct btrfs_root *root = arg;
 	int again;
+	struct btrfs_trans_handle *trans;
 
 	do {
 		again = 0;
@@ -1773,7 +1773,6 @@ static int cleaner_kthread(void *arg)
 		}
 
 		btrfs_run_delayed_iputs(root);
-		btrfs_delete_unused_bgs(root->fs_info);
 		again = btrfs_clean_one_deleted_snapshot(root);
 		mutex_unlock(&root->fs_info->cleaner_mutex);
 
@@ -1782,6 +1781,16 @@ static int cleaner_kthread(void *arg)
 		 * needn't do anything special here.
 		 */
 		btrfs_run_defrag_inodes(root->fs_info);
+
+		/*
+		 * Acquires fs_info->delete_unused_bgs_mutex to avoid racing
+		 * with relocation (btrfs_relocate_chunk) and relocation
+		 * acquires fs_info->cleaner_mutex (btrfs_relocate_block_group)
+		 * after acquiring fs_info->delete_unused_bgs_mutex. So we
+		 * can't hold, nor need to, fs_info->cleaner_mutex when deleting
+		 * unused block groups.
+		 */
+		btrfs_delete_unused_bgs(root->fs_info);
 sleep:
 		if (!try_to_freeze() && !again) {
 			set_current_state(TASK_INTERRUPTIBLE);
@@ -1790,6 +1799,34 @@ sleep:
 			__set_current_state(TASK_RUNNING);
 		}
 	} while (!kthread_should_stop());
+
+	/*
+	 * Transaction kthread is stopped before us and wakes us up.
+	 * However we might have started a new transaction and COWed some
+	 * tree blocks when deleting unused block groups for example. So
+	 * make sure we commit the transaction we started to have a clean
+	 * shutdown when evicting the btree inode - if it has dirty pages
+	 * when we do the final iput() on it, eviction will trigger a
+	 * writeback for it which will fail with null pointer dereferences
+	 * since work queues and other resources were already released and
+	 * destroyed by the time the iput/eviction/writeback is made.
+	 */
+	trans = btrfs_attach_transaction(root);
+	if (IS_ERR(trans)) {
+		if (PTR_ERR(trans) != -ENOENT)
+			btrfs_err(root->fs_info,
+				  "cleaner transaction attach returned %ld",
+				  PTR_ERR(trans));
+	} else {
+		int ret;
+
+		ret = btrfs_commit_transaction(trans, root);
+		if (ret)
+			btrfs_err(root->fs_info,
+				  "cleaner open transaction commit returned %d",
+				  ret);
+	}
+
 	return 0;
 }
 
@@ -2320,8 +2357,12 @@ static int btrfs_replay_log(struct btrfs_fs_info *fs_info,
 
 	log_tree_root->node = read_tree_block(tree_root, bytenr,
 			fs_info->generation + 1);
-	if (!log_tree_root->node ||
-	    !extent_buffer_uptodate(log_tree_root->node)) {
+	if (IS_ERR(log_tree_root->node)) {
+		printk(KERN_ERR "BTRFS: failed to read log tree\n");
+		ret = PTR_ERR(log_tree_root->node);
+		kfree(log_tree_root);
+		return ret;
+	} else if (!extent_buffer_uptodate(log_tree_root->node)) {
 		printk(KERN_ERR "BTRFS: failed to read log tree\n");
 		free_extent_buffer(log_tree_root->node);
 		kfree(log_tree_root);
@@ -2489,12 +2530,12 @@ int open_ctree(struct super_block *sb,
 	spin_lock_init(&fs_info->unused_bgs_lock);
 	rwlock_init(&fs_info->tree_mod_log_lock);
 	mutex_init(&fs_info->unused_bg_unpin_mutex);
+	mutex_init(&fs_info->delete_unused_bgs_mutex);
 	mutex_init(&fs_info->reloc_mutex);
 	mutex_init(&fs_info->delalloc_root_mutex);
 	seqlock_init(&fs_info->profiles_lock);
 	init_rwsem(&fs_info->delayed_iput_sem);
 
-	init_completion(&fs_info->kobj_unregister);
 	INIT_LIST_HEAD(&fs_info->dirty_cowonly_roots);
 	INIT_LIST_HEAD(&fs_info->space_info);
 	INIT_LIST_HEAD(&fs_info->tree_mod_seq_list);
@@ -2797,10 +2838,11 @@ int open_ctree(struct super_block *sb,
 	chunk_root->node = read_tree_block(chunk_root,
 					   btrfs_super_chunk_root(disk_super),
 					   generation);
-	if (!chunk_root->node ||
-	    !test_bit(EXTENT_BUFFER_UPTODATE, &chunk_root->node->bflags)) {
+	if (IS_ERR(chunk_root->node) ||
+	    !extent_buffer_uptodate(chunk_root->node)) {
 		printk(KERN_ERR "BTRFS: failed to read chunk root on %s\n",
 		       sb->s_id);
+		chunk_root->node = NULL;
 		goto fail_tree_roots;
 	}
 	btrfs_set_root_node(&chunk_root->root_item, chunk_root->node);
@@ -2834,11 +2876,11 @@ retry_root_backup:
 	tree_root->node = read_tree_block(tree_root,
 					  btrfs_super_root(disk_super),
 					  generation);
-	if (!tree_root->node ||
-	    !test_bit(EXTENT_BUFFER_UPTODATE, &tree_root->node->bflags)) {
+	if (IS_ERR(tree_root->node) ||
+	    !extent_buffer_uptodate(tree_root->node)) {
 		printk(KERN_WARNING "BTRFS: failed to read tree root on %s\n",
 		       sb->s_id);
-
+		tree_root->node = NULL;
 		goto recovery_tree_root;
 	}
 
@@ -2874,10 +2916,22 @@ retry_root_backup:
 
 	btrfs_close_extra_devices(fs_devices, 1);
 
+	ret = btrfs_sysfs_add_fsid(fs_devices, NULL);
+	if (ret) {
+		pr_err("BTRFS: failed to init sysfs fsid interface: %d\n", ret);
+		goto fail_block_groups;
+	}
+
+	ret = btrfs_sysfs_add_device(fs_devices);
+	if (ret) {
+		pr_err("BTRFS: failed to init sysfs device interface: %d\n", ret);
+		goto fail_fsdev_sysfs;
+	}
+
 	ret = btrfs_sysfs_add_one(fs_info);
 	if (ret) {
 		pr_err("BTRFS: failed to init sysfs interface: %d\n", ret);
-		goto fail_block_groups;
+		goto fail_fsdev_sysfs;
 	}
 
 	ret = btrfs_init_space_info(fs_info);
@@ -3055,6 +3109,9 @@ fail_cleaner:
 fail_sysfs:
 	btrfs_sysfs_remove_one(fs_info);
 
+fail_fsdev_sysfs:
+	btrfs_sysfs_remove_fsid(fs_info->fs_devices);
+
 fail_block_groups:
 	btrfs_put_block_group_cache(fs_info);
 	btrfs_free_block_groups(fs_info);
@@ -3269,11 +3326,8 @@ static int write_dev_supers(struct btrfs_device *device,
  */
 static void btrfs_end_empty_barrier(struct bio *bio, int err)
 {
-	if (err) {
-		if (err == -EOPNOTSUPP)
-			set_bit(BIO_EOPNOTSUPP, &bio->bi_flags);
+	if (err)
 		clear_bit(BIO_UPTODATE, &bio->bi_flags);
-	}
 	if (bio->bi_private)
 		complete(bio->bi_private);
 	bio_put(bio);
@@ -3301,11 +3355,7 @@ static int write_dev_flush(struct btrfs_device *device, int wait)
 
 		wait_for_completion(&device->flush_wait);
 
-		if (bio_flagged(bio, BIO_EOPNOTSUPP)) {
-			printk_in_rcu("BTRFS: disabling barriers on dev %s\n",
-				      rcu_str_deref(device->name));
-			device->nobarriers = 1;
-		} else if (!bio_flagged(bio, BIO_UPTODATE)) {
+		if (!bio_flagged(bio, BIO_UPTODATE)) {
 			ret = -EIO;
 			btrfs_dev_stat_inc_and_print(device,
 				BTRFS_DEV_STAT_FLUSH_ERRS);
@@ -3732,6 +3782,7 @@ void close_ctree(struct btrfs_root *root)
 	}
 
 	btrfs_sysfs_remove_one(fs_info);
+	btrfs_sysfs_remove_fsid(fs_info->fs_devices);
 
 	btrfs_free_fs_roots(fs_info);
 
@@ -4060,6 +4111,7 @@ static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
 
 	while ((node = rb_first(&delayed_refs->href_root)) != NULL) {
 		struct btrfs_delayed_ref_head *head;
+		struct btrfs_delayed_ref_node *tmp;
 		bool pin_bytes = false;
 
 		head = rb_entry(node, struct btrfs_delayed_ref_head,
@@ -4075,11 +4127,10 @@ static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
 			continue;
 		}
 		spin_lock(&head->lock);
-		while ((node = rb_first(&head->ref_root)) != NULL) {
-			ref = rb_entry(node, struct btrfs_delayed_ref_node,
-				       rb_node);
+		list_for_each_entry_safe_reverse(ref, tmp, &head->ref_list,
+						 list) {
 			ref->in_tree = 0;
-			rb_erase(&ref->rb_node, &head->ref_root);
+			list_del(&ref->list);
 			atomic_dec(&delayed_refs->num_entries);
 			btrfs_put_delayed_ref(ref);
 		}
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 0ec3acd14..07204bf60 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -79,11 +79,10 @@ static int update_block_group(struct btrfs_trans_handle *trans,
 			      u64 num_bytes, int alloc);
 static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 				struct btrfs_root *root,
-				u64 bytenr, u64 num_bytes, u64 parent,
+				struct btrfs_delayed_ref_node *node, u64 parent,
 				u64 root_objectid, u64 owner_objectid,
 				u64 owner_offset, int refs_to_drop,
-				struct btrfs_delayed_extent_op *extra_op,
-				int no_quota);
+				struct btrfs_delayed_extent_op *extra_op);
 static void __run_delayed_extent_op(struct btrfs_delayed_extent_op *extent_op,
 				    struct extent_buffer *leaf,
 				    struct btrfs_extent_item *ei);
@@ -1967,10 +1966,9 @@ int btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 
 static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 				  struct btrfs_root *root,
-				  u64 bytenr, u64 num_bytes,
+				  struct btrfs_delayed_ref_node *node,
 				  u64 parent, u64 root_objectid,
 				  u64 owner, u64 offset, int refs_to_add,
-				  int no_quota,
 				  struct btrfs_delayed_extent_op *extent_op)
 {
 	struct btrfs_fs_info *fs_info = root->fs_info;
@@ -1978,9 +1976,11 @@ static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 	struct extent_buffer *leaf;
 	struct btrfs_extent_item *item;
 	struct btrfs_key key;
+	u64 bytenr = node->bytenr;
+	u64 num_bytes = node->num_bytes;
 	u64 refs;
 	int ret;
-	enum btrfs_qgroup_operation_type type = BTRFS_QGROUP_OPER_ADD_EXCL;
+	int no_quota = node->no_quota;
 
 	path = btrfs_alloc_path();
 	if (!path)
@@ -1996,26 +1996,8 @@ static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 					   bytenr, num_bytes, parent,
 					   root_objectid, owner, offset,
 					   refs_to_add, extent_op);
-	if ((ret < 0 && ret != -EAGAIN) || (!ret && no_quota))
+	if ((ret < 0 && ret != -EAGAIN) || !ret)
 		goto out;
-	/*
-	 * Ok we were able to insert an inline extent and it appears to be a new
-	 * reference, deal with the qgroup accounting.
-	 */
-	if (!ret && !no_quota) {
-		ASSERT(root->fs_info->quota_enabled);
-		leaf = path->nodes[0];
-		btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
-		item = btrfs_item_ptr(leaf, path->slots[0],
-				      struct btrfs_extent_item);
-		if (btrfs_extent_refs(leaf, item) > (u64)refs_to_add)
-			type = BTRFS_QGROUP_OPER_ADD_SHARED;
-		btrfs_release_path(path);
-
-		ret = btrfs_qgroup_record_ref(trans, fs_info, root_objectid,
-					      bytenr, num_bytes, type, 0);
-		goto out;
-	}
 
 	/*
 	 * Ok we had -EAGAIN which means we didn't have space to insert and
@@ -2026,8 +2008,6 @@ static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 	btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
 	item = btrfs_item_ptr(leaf, path->slots[0], struct btrfs_extent_item);
 	refs = btrfs_extent_refs(leaf, item);
-	if (refs)
-		type = BTRFS_QGROUP_OPER_ADD_SHARED;
 	btrfs_set_extent_refs(leaf, item, refs + refs_to_add);
 	if (extent_op)
 		__run_delayed_extent_op(extent_op, leaf, item);
@@ -2035,13 +2015,6 @@ static int __btrfs_inc_extent_ref(struct btrfs_trans_handle *trans,
 	btrfs_mark_buffer_dirty(leaf);
 	btrfs_release_path(path);
 
-	if (!no_quota) {
-		ret = btrfs_qgroup_record_ref(trans, fs_info, root_objectid,
-					      bytenr, num_bytes, type, 0);
-		if (ret)
-			goto out;
-	}
-
 	path->reada = 1;
 	path->leave_spinning = 1;
 	/* now insert the actual backref */
@@ -2087,17 +2060,15 @@ static int run_delayed_data_ref(struct btrfs_trans_handle *trans,
 						 ref->objectid, ref->offset,
 						 &ins, node->ref_mod);
 	} else if (node->action == BTRFS_ADD_DELAYED_REF) {
-		ret = __btrfs_inc_extent_ref(trans, root, node->bytenr,
-					     node->num_bytes, parent,
+		ret = __btrfs_inc_extent_ref(trans, root, node, parent,
 					     ref_root, ref->objectid,
 					     ref->offset, node->ref_mod,
-					     node->no_quota, extent_op);
+					     extent_op);
 	} else if (node->action == BTRFS_DROP_DELAYED_REF) {
-		ret = __btrfs_free_extent(trans, root, node->bytenr,
-					  node->num_bytes, parent,
+		ret = __btrfs_free_extent(trans, root, node, parent,
 					  ref_root, ref->objectid,
 					  ref->offset, node->ref_mod,
-					  extent_op, node->no_quota);
+					  extent_op);
 	} else {
 		BUG();
 	}
@@ -2255,15 +2226,14 @@ static int run_delayed_tree_ref(struct btrfs_trans_handle *trans,
 						ref->level, &ins,
 						node->no_quota);
 	} else if (node->action == BTRFS_ADD_DELAYED_REF) {
-		ret = __btrfs_inc_extent_ref(trans, root, node->bytenr,
-					     node->num_bytes, parent, ref_root,
-					     ref->level, 0, 1, node->no_quota,
+		ret = __btrfs_inc_extent_ref(trans, root, node,
+					     parent, ref_root,
+					     ref->level, 0, 1,
 					     extent_op);
 	} else if (node->action == BTRFS_DROP_DELAYED_REF) {
-		ret = __btrfs_free_extent(trans, root, node->bytenr,
-					  node->num_bytes, parent, ref_root,
-					  ref->level, 0, 1, extent_op,
-					  node->no_quota);
+		ret = __btrfs_free_extent(trans, root, node,
+					  parent, ref_root,
+					  ref->level, 0, 1, extent_op);
 	} else {
 		BUG();
 	}
@@ -2323,28 +2293,27 @@ static int run_one_delayed_ref(struct btrfs_trans_handle *trans,
 	return ret;
 }
 
-static noinline struct btrfs_delayed_ref_node *
+static inline struct btrfs_delayed_ref_node *
 select_delayed_ref(struct btrfs_delayed_ref_head *head)
 {
-	struct rb_node *node;
-	struct btrfs_delayed_ref_node *ref, *last = NULL;;
+	struct btrfs_delayed_ref_node *ref;
+
+	if (list_empty(&head->ref_list))
+		return NULL;
 
 	/*
-	 * select delayed ref of type BTRFS_ADD_DELAYED_REF first.
-	 * this prevents ref count from going down to zero when
-	 * there still are pending delayed ref.
+	 * Select a delayed ref of type BTRFS_ADD_DELAYED_REF first.
+	 * This is to prevent a ref count from going down to zero, which deletes
+	 * the extent item from the extent tree, when there still are references
+	 * to add, which would fail because they would not find the extent item.
 	 */
-	node = rb_first(&head->ref_root);
-	while (node) {
-		ref = rb_entry(node, struct btrfs_delayed_ref_node,
-				rb_node);
+	list_for_each_entry(ref, &head->ref_list, list) {
 		if (ref->action == BTRFS_ADD_DELAYED_REF)
 			return ref;
-		else if (last == NULL)
-			last = ref;
-		node = rb_next(node);
 	}
-	return last;
+
+	return list_entry(head->ref_list.next, struct btrfs_delayed_ref_node,
+			  list);
 }
 
 /*
@@ -2396,16 +2365,7 @@ static noinline int __btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
 			}
 		}
 
-		/*
-		 * We need to try and merge add/drops of the same ref since we
-		 * can run into issues with relocate dropping the implicit ref
-		 * and then it being added back again before the drop can
-		 * finish.  If we merged anything we need to re-loop so we can
-		 * get a good ref.
-		 */
 		spin_lock(&locked_ref->lock);
-		btrfs_merge_delayed_refs(trans, fs_info, delayed_refs,
-					 locked_ref);
 
 		/*
 		 * locked_ref is the head node, so we have to go one
@@ -2482,7 +2442,7 @@ static noinline int __btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
 			spin_unlock(&locked_ref->lock);
 			spin_lock(&delayed_refs->lock);
 			spin_lock(&locked_ref->lock);
-			if (rb_first(&locked_ref->ref_root) ||
+			if (!list_empty(&locked_ref->ref_list) ||
 			    locked_ref->extent_op) {
 				spin_unlock(&locked_ref->lock);
 				spin_unlock(&delayed_refs->lock);
@@ -2496,7 +2456,7 @@ static noinline int __btrfs_run_delayed_refs(struct btrfs_trans_handle *trans,
 		} else {
 			actual_count++;
 			ref->in_tree = 0;
-			rb_erase(&ref->rb_node, &locked_ref->ref_root);
+			list_del(&ref->list);
 		}
 		atomic_dec(&delayed_refs->num_entries);
 
@@ -2864,9 +2824,6 @@ again:
 		goto again;
 	}
 out:
-	ret = btrfs_delayed_qgroup_accounting(trans, root->fs_info);
-	if (ret)
-		return ret;
 	assert_qgroups_uptodate(trans);
 	return 0;
 }
@@ -2905,7 +2862,6 @@ static noinline int check_delayed_ref(struct btrfs_trans_handle *trans,
 	struct btrfs_delayed_ref_node *ref;
 	struct btrfs_delayed_data_ref *data_ref;
 	struct btrfs_delayed_ref_root *delayed_refs;
-	struct rb_node *node;
 	int ret = 0;
 
 	delayed_refs = &trans->transaction->delayed_refs;
@@ -2934,11 +2890,7 @@ static noinline int check_delayed_ref(struct btrfs_trans_handle *trans,
 	spin_unlock(&delayed_refs->lock);
 
 	spin_lock(&head->lock);
-	node = rb_first(&head->ref_root);
-	while (node) {
-		ref = rb_entry(node, struct btrfs_delayed_ref_node, rb_node);
-		node = rb_next(node);
-
+	list_for_each_entry(ref, &head->ref_list, list) {
 		/* If it's a shared ref we know a cross reference exists */
 		if (ref->type != BTRFS_EXTENT_DATA_REF_KEY) {
 			ret = 1;
@@ -3693,7 +3645,8 @@ static int update_space_info(struct btrfs_fs_info *info, u64 flags,
 		found->disk_total += total_bytes * factor;
 		found->bytes_used += bytes_used;
 		found->disk_used += bytes_used * factor;
-		found->full = 0;
+		if (total_bytes > 0)
+			found->full = 0;
 		spin_unlock(&found->lock);
 		*space_info = found;
 		return 0;
@@ -3721,7 +3674,10 @@ static int update_space_info(struct btrfs_fs_info *info, u64 flags,
 	found->bytes_reserved = 0;
 	found->bytes_readonly = 0;
 	found->bytes_may_use = 0;
-	found->full = 0;
+	if (total_bytes > 0)
+		found->full = 0;
+	else
+		found->full = 1;
 	found->force_alloc = CHUNK_ALLOC_NO_FORCE;
 	found->chunk_alloc = 0;
 	found->flush = 0;
@@ -3975,6 +3931,9 @@ commit_trans:
 		    !atomic_read(&root->fs_info->open_ioctl_trans)) {
 			need_commit--;
 
+			if (need_commit > 0)
+				btrfs_wait_ordered_roots(fs_info, -1);
+
 			trans = btrfs_join_transaction(root);
 			if (IS_ERR(trans))
 				return PTR_ERR(trans);
@@ -4088,7 +4047,7 @@ static int should_alloc_chunk(struct btrfs_root *root,
 	return 1;
 }
 
-static u64 get_system_chunk_thresh(struct btrfs_root *root, u64 type)
+static u64 get_profile_num_devs(struct btrfs_root *root, u64 type)
 {
 	u64 num_dev;
 
@@ -4102,24 +4061,43 @@ static u64 get_system_chunk_thresh(struct btrfs_root *root, u64 type)
 	else
 		num_dev = 1;	/* DUP or single */
 
-	/* metadata for updaing devices and chunk tree */
-	return btrfs_calc_trans_metadata_size(root, num_dev + 1);
+	return num_dev;
 }
 
-static void check_system_chunk(struct btrfs_trans_handle *trans,
-			       struct btrfs_root *root, u64 type)
+/*
+ * If @is_allocation is true, reserve space in the system space info necessary
+ * for allocating a chunk, otherwise if it's false, reserve space necessary for
+ * removing a chunk.
+ */
+void check_system_chunk(struct btrfs_trans_handle *trans,
+			struct btrfs_root *root,
+			u64 type)
 {
 	struct btrfs_space_info *info;
 	u64 left;
 	u64 thresh;
+	int ret = 0;
+	u64 num_devs;
+
+	/*
+	 * Needed because we can end up allocating a system chunk and for an
+	 * atomic and race free space reservation in the chunk block reserve.
+	 */
+	ASSERT(mutex_is_locked(&root->fs_info->chunk_mutex));
 
 	info = __find_space_info(root->fs_info, BTRFS_BLOCK_GROUP_SYSTEM);
 	spin_lock(&info->lock);
 	left = info->total_bytes - info->bytes_used - info->bytes_pinned -
-		info->bytes_reserved - info->bytes_readonly;
+		info->bytes_reserved - info->bytes_readonly -
+		info->bytes_may_use;
 	spin_unlock(&info->lock);
 
-	thresh = get_system_chunk_thresh(root, type);
+	num_devs = get_profile_num_devs(root, type);
+
+	/* num_devs device items to update and 1 chunk item to add or remove */
+	thresh = btrfs_calc_trunc_metadata_size(root, num_devs) +
+		btrfs_calc_trans_metadata_size(root, 1);
+
 	if (left < thresh && btrfs_test_opt(root, ENOSPC_DEBUG)) {
 		btrfs_info(root->fs_info, "left=%llu, need=%llu, flags=%llu",
 			left, thresh, type);
@@ -4130,7 +4108,21 @@ static void check_system_chunk(struct btrfs_trans_handle *trans,
 		u64 flags;
 
 		flags = btrfs_get_alloc_profile(root->fs_info->chunk_root, 0);
-		btrfs_alloc_chunk(trans, root, flags);
+		/*
+		 * Ignore failure to create system chunk. We might end up not
+		 * needing it, as we might not need to COW all nodes/leafs from
+		 * the paths we visit in the chunk tree (they were already COWed
+		 * or created in the current transaction for example).
+		 */
+		ret = btrfs_alloc_chunk(trans, root, flags);
+	}
+
+	if (!ret) {
+		ret = btrfs_block_rsv_add(root->fs_info->chunk_root,
+					  &root->fs_info->chunk_block_rsv,
+					  thresh, BTRFS_RESERVE_NO_FLUSH);
+		if (!ret)
+			trans->chunk_bytes_reserved += thresh;
 	}
 }
 
@@ -4235,6 +4227,24 @@ out:
 	space_info->chunk_alloc = 0;
 	spin_unlock(&space_info->lock);
 	mutex_unlock(&fs_info->chunk_mutex);
+	/*
+	 * When we allocate a new chunk we reserve space in the chunk block
+	 * reserve to make sure we can COW nodes/leafs in the chunk tree or
+	 * add new nodes/leafs to it if we end up needing to do it when
+	 * inserting the chunk item and updating device items as part of the
+	 * second phase of chunk allocation, performed by
+	 * btrfs_finish_chunk_alloc(). So make sure we don't accumulate a
+	 * large number of new block groups to create in our transaction
+	 * handle's new_bgs list to avoid exhausting the chunk block reserve
+	 * in extreme cases - like having a single transaction create many new
+	 * block groups when starting to write out the free space caches of all
+	 * the block groups that were made dirty during the lifetime of the
+	 * transaction.
+	 */
+	if (trans->chunk_bytes_reserved >= (2 * 1024 * 1024ull)) {
+		btrfs_create_pending_block_groups(trans, trans->root);
+		btrfs_trans_release_chunk_metadata(trans);
+	}
 	return ret;
 }
 
@@ -5188,6 +5198,24 @@ void btrfs_trans_release_metadata(struct btrfs_trans_handle *trans,
 	trans->bytes_reserved = 0;
 }
 
+/*
+ * To be called after all the new block groups attached to the transaction
+ * handle have been created (btrfs_create_pending_block_groups()).
+ */
+void btrfs_trans_release_chunk_metadata(struct btrfs_trans_handle *trans)
+{
+	struct btrfs_fs_info *fs_info = trans->root->fs_info;
+
+	if (!trans->chunk_bytes_reserved)
+		return;
+
+	WARN_ON_ONCE(!list_empty(&trans->new_bgs));
+
+	block_rsv_release_bytes(fs_info, &fs_info->chunk_block_rsv, NULL,
+				trans->chunk_bytes_reserved);
+	trans->chunk_bytes_reserved = 0;
+}
+
 /* Can only return 0 or -ENOSPC */
 int btrfs_orphan_reserve_metadata(struct btrfs_trans_handle *trans,
 				  struct inode *inode)
@@ -6092,11 +6120,10 @@ static void add_pinned_bytes(struct btrfs_fs_info *fs_info, u64 num_bytes,
 
 static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 				struct btrfs_root *root,
-				u64 bytenr, u64 num_bytes, u64 parent,
+				struct btrfs_delayed_ref_node *node, u64 parent,
 				u64 root_objectid, u64 owner_objectid,
 				u64 owner_offset, int refs_to_drop,
-				struct btrfs_delayed_extent_op *extent_op,
-				int no_quota)
+				struct btrfs_delayed_extent_op *extent_op)
 {
 	struct btrfs_key key;
 	struct btrfs_path *path;
@@ -6110,10 +6137,12 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 	int extent_slot = 0;
 	int found_extent = 0;
 	int num_to_del = 1;
+	int no_quota = node->no_quota;
 	u32 item_size;
 	u64 refs;
+	u64 bytenr = node->bytenr;
+	u64 num_bytes = node->num_bytes;
 	int last_ref = 0;
-	enum btrfs_qgroup_operation_type type = BTRFS_QGROUP_OPER_SUB_EXCL;
 	bool skinny_metadata = btrfs_fs_incompat(root->fs_info,
 						 SKINNY_METADATA);
 
@@ -6294,7 +6323,6 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 	refs -= refs_to_drop;
 
 	if (refs > 0) {
-		type = BTRFS_QGROUP_OPER_SUB_SHARED;
 		if (extent_op)
 			__run_delayed_extent_op(extent_op, leaf, ei);
 		/*
@@ -6356,18 +6384,6 @@ static int __btrfs_free_extent(struct btrfs_trans_handle *trans,
 	}
 	btrfs_release_path(path);
 
-	/* Deal with the quota accounting */
-	if (!ret && last_ref && !no_quota) {
-		int mod_seq = 0;
-
-		if (owner_objectid >= BTRFS_FIRST_FREE_OBJECTID &&
-		    type == BTRFS_QGROUP_OPER_SUB_SHARED)
-			mod_seq = 1;
-
-		ret = btrfs_qgroup_record_ref(trans, info, root_objectid,
-					      bytenr, num_bytes, type,
-					      mod_seq);
-	}
 out:
 	btrfs_free_path(path);
 	return ret;
@@ -6393,7 +6409,7 @@ static noinline int check_ref_cleanup(struct btrfs_trans_handle *trans,
 		goto out_delayed_unlock;
 
 	spin_lock(&head->lock);
-	if (rb_first(&head->ref_root))
+	if (!list_empty(&head->ref_list))
 		goto out;
 
 	if (head->extent_op) {
@@ -7303,13 +7319,6 @@ static int alloc_reserved_file_extent(struct btrfs_trans_handle *trans,
 	btrfs_mark_buffer_dirty(path->nodes[0]);
 	btrfs_free_path(path);
 
-	/* Always set parent to 0 here since its exclusive anyway. */
-	ret = btrfs_qgroup_record_ref(trans, fs_info, root_objectid,
-				      ins->objectid, ins->offset,
-				      BTRFS_QGROUP_OPER_ADD_EXCL, 0);
-	if (ret)
-		return ret;
-
 	ret = update_block_group(trans, root, ins->objectid, ins->offset, 1);
 	if (ret) { /* -ENOENT, logic error */
 		btrfs_err(fs_info, "update block group failed for %llu %llu",
@@ -7391,14 +7400,6 @@ static int alloc_reserved_tree_block(struct btrfs_trans_handle *trans,
 	btrfs_mark_buffer_dirty(leaf);
 	btrfs_free_path(path);
 
-	if (!no_quota) {
-		ret = btrfs_qgroup_record_ref(trans, fs_info, root_objectid,
-					      ins->objectid, num_bytes,
-					      BTRFS_QGROUP_OPER_ADD_EXCL, 0);
-		if (ret)
-			return ret;
-	}
-
 	ret = update_block_group(trans, root, ins->objectid, root->nodesize,
 				 1);
 	if (ret) { /* -ENOENT, logic error */
@@ -7755,12 +7756,18 @@ reada:
 	wc->reada_slot = slot;
 }
 
+/*
+ * TODO: Modify related function to add related node/leaf to dirty_extent_root,
+ * for later qgroup accounting.
+ *
+ * Current, this function does nothing.
+ */
 static int account_leaf_items(struct btrfs_trans_handle *trans,
 			      struct btrfs_root *root,
 			      struct extent_buffer *eb)
 {
 	int nr = btrfs_header_nritems(eb);
-	int i, extent_type, ret;
+	int i, extent_type;
 	struct btrfs_key key;
 	struct btrfs_file_extent_item *fi;
 	u64 bytenr, num_bytes;
@@ -7783,13 +7790,6 @@ static int account_leaf_items(struct btrfs_trans_handle *trans,
 			continue;
 
 		num_bytes = btrfs_file_extent_disk_num_bytes(eb, fi);
-
-		ret = btrfs_qgroup_record_ref(trans, root->fs_info,
-					      root->objectid,
-					      bytenr, num_bytes,
-					      BTRFS_QGROUP_OPER_SUB_SUBTREE, 0);
-		if (ret)
-			return ret;
 	}
 	return 0;
 }
@@ -7858,6 +7858,8 @@ static int adjust_slots_upwards(struct btrfs_root *root,
 
 /*
  * root_eb is the subtree root and is locked before this function is called.
+ * TODO: Modify this function to mark all (including complete shared node)
+ * to dirty_extent_root to allow it get accounted in qgroup.
  */
 static int account_shared_subtree(struct btrfs_trans_handle *trans,
 				  struct btrfs_root *root,
@@ -7920,7 +7922,11 @@ walk_down:
 			child_gen = btrfs_node_ptr_generation(eb, parent_slot);
 
 			eb = read_tree_block(root, child_bytenr, child_gen);
-			if (!eb || !extent_buffer_uptodate(eb)) {
+			if (IS_ERR(eb)) {
+				ret = PTR_ERR(eb);
+				goto out;
+			} else if (!extent_buffer_uptodate(eb)) {
+				free_extent_buffer(eb);
 				ret = -EIO;
 				goto out;
 			}
@@ -7931,16 +7937,6 @@ walk_down:
 			btrfs_tree_read_lock(eb);
 			btrfs_set_lock_blocking_rw(eb, BTRFS_READ_LOCK);
 			path->locks[level] = BTRFS_READ_LOCK_BLOCKING;
-
-			ret = btrfs_qgroup_record_ref(trans, root->fs_info,
-						root->objectid,
-						child_bytenr,
-						root->nodesize,
-						BTRFS_QGROUP_OPER_SUB_SUBTREE,
-						0);
-			if (ret)
-				goto out;
-
 		}
 
 		if (level == 0) {
@@ -8151,7 +8147,9 @@ static noinline int do_walk_down(struct btrfs_trans_handle *trans,
 		if (reada && level == 1)
 			reada_walk_down(trans, root, wc, path);
 		next = read_tree_block(root, bytenr, generation);
-		if (!next || !extent_buffer_uptodate(next)) {
+		if (IS_ERR(next)) {
+			return PTR_ERR(next);
+		} else if (!extent_buffer_uptodate(next)) {
 			free_extent_buffer(next);
 			return -EIO;
 		}
@@ -8533,24 +8531,6 @@ int btrfs_drop_snapshot(struct btrfs_root *root,
 				goto out_end_trans;
 			}
 
-			/*
-			 * Qgroup update accounting is run from
-			 * delayed ref handling. This usually works
-			 * out because delayed refs are normally the
-			 * only way qgroup updates are added. However,
-			 * we may have added updates during our tree
-			 * walk so run qgroups here to make sure we
-			 * don't lose any updates.
-			 */
-			ret = btrfs_delayed_qgroup_accounting(trans,
-							      root->fs_info);
-			if (ret)
-				printk_ratelimited(KERN_ERR "BTRFS: Failure %d "
-						   "running qgroup updates "
-						   "during snapshot delete. "
-						   "Quota is out of sync, "
-						   "rescan required.\n", ret);
-
 			btrfs_end_transaction_throttle(trans, tree_root);
 			if (!for_reloc && btrfs_need_cleaner_sleep(root)) {
 				pr_debug("BTRFS: drop snapshot early exit\n");
@@ -8604,14 +8584,6 @@ int btrfs_drop_snapshot(struct btrfs_root *root,
 	}
 	root_dropped = true;
 out_end_trans:
-	ret = btrfs_delayed_qgroup_accounting(trans, tree_root->fs_info);
-	if (ret)
-		printk_ratelimited(KERN_ERR "BTRFS: Failure %d "
-				   "running qgroup updates "
-				   "during snapshot delete. "
-				   "Quota is out of sync, "
-				   "rescan required.\n", ret);
-
 	btrfs_end_transaction_throttle(trans, tree_root);
 out_free:
 	kfree(wc);
@@ -9562,6 +9534,19 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans,
 
 	free_excluded_extents(root, cache);
 
+	/*
+	 * Call to ensure the corresponding space_info object is created and
+	 * assigned to our block group, but don't update its counters just yet.
+	 * We want our bg to be added to the rbtree with its ->space_info set.
+	 */
+	ret = update_space_info(root->fs_info, cache->flags, 0, 0,
+				&cache->space_info);
+	if (ret) {
+		btrfs_remove_free_space_cache(cache);
+		btrfs_put_block_group(cache);
+		return ret;
+	}
+
 	ret = btrfs_add_block_group_cache(root->fs_info, cache);
 	if (ret) {
 		btrfs_remove_free_space_cache(cache);
@@ -9569,6 +9554,10 @@ int btrfs_make_block_group(struct btrfs_trans_handle *trans,
 		return ret;
 	}
 
+	/*
+	 * Now that our block group has its ->space_info set and is inserted in
+	 * the rbtree, update the space info's counters.
+	 */
 	ret = update_space_info(root->fs_info, cache->flags, size, bytes_used,
 				&cache->space_info);
 	if (ret) {
@@ -9931,6 +9920,8 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 		}
 		spin_unlock(&fs_info->unused_bgs_lock);
 
+		mutex_lock(&root->fs_info->delete_unused_bgs_mutex);
+
 		/* Don't want to race with allocators so take the groups_sem */
 		down_write(&space_info->groups_sem);
 		spin_lock(&block_group->lock);
@@ -10025,6 +10016,7 @@ void btrfs_delete_unused_bgs(struct btrfs_fs_info *fs_info)
 end_trans:
 		btrfs_end_transaction(trans, root);
 next:
+		mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 		btrfs_put_block_group(block_group);
 		spin_lock(&fs_info->unused_bgs_lock);
 	}
diff --git a/fs/btrfs/extent-tree.h b/fs/btrfs/extent-tree.h
new file mode 100644
index 000000000..e69de29bb
--- /dev/null
+++ b/fs/btrfs/extent-tree.h
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index c32d226bf..02d05817c 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -1277,7 +1277,12 @@ int set_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
 int clear_extent_bits(struct extent_io_tree *tree, u64 start, u64 end,
 		      unsigned bits, gfp_t mask)
 {
-	return clear_extent_bit(tree, start, end, bits, 0, 0, NULL, mask);
+	int wake = 0;
+
+	if (bits & EXTENT_LOCKED)
+		wake = 1;
+
+	return clear_extent_bit(tree, start, end, bits, wake, 0, NULL, mask);
 }
 
 int set_extent_delalloc(struct extent_io_tree *tree, u64 start, u64 end,
@@ -2767,8 +2772,6 @@ static int __must_check submit_one_bio(int rw, struct bio *bio,
 	else
 		btrfsic_submit_bio(rw, bio);
 
-	if (bio_flagged(bio, BIO_EOPNOTSUPP))
-		ret = -EOPNOTSUPP;
 	bio_put(bio);
 	return ret;
 }
@@ -4492,6 +4495,8 @@ int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo,
 		}
 		if (test_bit(EXTENT_FLAG_COMPRESSED, &em->flags))
 			flags |= FIEMAP_EXTENT_ENCODED;
+		if (test_bit(EXTENT_FLAG_PREALLOC, &em->flags))
+			flags |= FIEMAP_EXTENT_UNWRITTEN;
 
 		free_extent_map(em);
 		em = NULL;
diff --git a/fs/btrfs/file.c b/fs/btrfs/file.c
index b072e1747..b823fac91 100644
--- a/fs/btrfs/file.c
+++ b/fs/btrfs/file.c
@@ -1748,7 +1748,7 @@ static ssize_t btrfs_file_write_iter(struct kiocb *iocb,
 	}
 
 	current->backing_dev_info = inode_to_bdi(inode);
-	err = file_remove_suid(file);
+	err = file_remove_privs(file);
 	if (err) {
 		mutex_unlock(&inode->i_mutex);
 		goto out;
@@ -1868,6 +1868,7 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 	struct btrfs_log_ctx ctx;
 	int ret = 0;
 	bool full_sync = 0;
+	const u64 len = end - start + 1;
 
 	trace_btrfs_sync_file(file, datasync);
 
@@ -1896,7 +1897,7 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 		 * all extents are persisted and the respective file extent
 		 * items are in the fs/subvol btree.
 		 */
-		ret = btrfs_wait_ordered_range(inode, start, end - start + 1);
+		ret = btrfs_wait_ordered_range(inode, start, len);
 	} else {
 		/*
 		 * Start any new ordered operations before starting to log the
@@ -1968,8 +1969,10 @@ int btrfs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
 	 */
 	smp_mb();
 	if (btrfs_inode_in_log(inode, root->fs_info->generation) ||
-	    (full_sync && BTRFS_I(inode)->last_trans <=
-	     root->fs_info->last_trans_committed)) {
+	    (BTRFS_I(inode)->last_trans <=
+	     root->fs_info->last_trans_committed &&
+	     (full_sync ||
+	      !btrfs_have_ordered_extents_in_range(inode, start, len)))) {
 		/*
 		 * We'v had everything committed since the last time we were
 		 * modified so clear this flag in case it was set for whatever
diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c
index 9dbe5b548..fb5a6b1c6 100644
--- a/fs/btrfs/free-space-cache.c
+++ b/fs/btrfs/free-space-cache.c
@@ -231,6 +231,7 @@ int btrfs_truncate_free_space_cache(struct btrfs_root *root,
 {
 	int ret = 0;
 	struct btrfs_path *path = btrfs_alloc_path();
+	bool locked = false;
 
 	if (!path) {
 		ret = -ENOMEM;
@@ -238,6 +239,7 @@ int btrfs_truncate_free_space_cache(struct btrfs_root *root,
 	}
 
 	if (block_group) {
+		locked = true;
 		mutex_lock(&trans->transaction->cache_write_mutex);
 		if (!list_empty(&block_group->io_list)) {
 			list_del_init(&block_group->io_list);
@@ -269,18 +271,14 @@ int btrfs_truncate_free_space_cache(struct btrfs_root *root,
 	 */
 	ret = btrfs_truncate_inode_items(trans, root, inode,
 					 0, BTRFS_EXTENT_DATA_KEY);
-	if (ret) {
-		mutex_unlock(&trans->transaction->cache_write_mutex);
-		btrfs_abort_transaction(trans, root, ret);
-		return ret;
-	}
+	if (ret)
+		goto fail;
 
 	ret = btrfs_update_inode(trans, root, inode);
 
-	if (block_group)
-		mutex_unlock(&trans->transaction->cache_write_mutex);
-
 fail:
+	if (locked)
+		mutex_unlock(&trans->transaction->cache_write_mutex);
 	if (ret)
 		btrfs_abort_transaction(trans, root, ret);
 
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 8bb013672..e33dff356 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -4209,7 +4209,7 @@ int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
 	u64 extent_num_bytes = 0;
 	u64 extent_offset = 0;
 	u64 item_end = 0;
-	u64 last_size = (u64)-1;
+	u64 last_size = new_size;
 	u32 found_type = (u8)-1;
 	int found_extent;
 	int del_item;
@@ -4493,8 +4493,7 @@ out:
 			btrfs_abort_transaction(trans, root, ret);
 	}
 error:
-	if (last_size != (u64)-1 &&
-	    root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID)
+	if (root->root_key.objectid != BTRFS_TREE_LOG_OBJECTID)
 		btrfs_ordered_update_i_size(inode, last_size, NULL);
 
 	btrfs_free_path(path);
@@ -4986,24 +4985,41 @@ static void evict_inode_truncate_pages(struct inode *inode)
 	}
 	write_unlock(&map_tree->lock);
 
+	/*
+	 * Keep looping until we have no more ranges in the io tree.
+	 * We can have ongoing bios started by readpages (called from readahead)
+	 * that have their endio callback (extent_io.c:end_bio_extent_readpage)
+	 * still in progress (unlocked the pages in the bio but did not yet
+	 * unlocked the ranges in the io tree). Therefore this means some
+	 * ranges can still be locked and eviction started because before
+	 * submitting those bios, which are executed by a separate task (work
+	 * queue kthread), inode references (inode->i_count) were not taken
+	 * (which would be dropped in the end io callback of each bio).
+	 * Therefore here we effectively end up waiting for those bios and
+	 * anyone else holding locked ranges without having bumped the inode's
+	 * reference count - if we don't do it, when they access the inode's
+	 * io_tree to unlock a range it may be too late, leading to an
+	 * use-after-free issue.
+	 */
 	spin_lock(&io_tree->lock);
 	while (!RB_EMPTY_ROOT(&io_tree->state)) {
 		struct extent_state *state;
 		struct extent_state *cached_state = NULL;
+		u64 start;
+		u64 end;
 
 		node = rb_first(&io_tree->state);
 		state = rb_entry(node, struct extent_state, rb_node);
-		atomic_inc(&state->refs);
+		start = state->start;
+		end = state->end;
 		spin_unlock(&io_tree->lock);
 
-		lock_extent_bits(io_tree, state->start, state->end,
-				 0, &cached_state);
-		clear_extent_bit(io_tree, state->start, state->end,
+		lock_extent_bits(io_tree, start, end, 0, &cached_state);
+		clear_extent_bit(io_tree, start, end,
 				 EXTENT_LOCKED | EXTENT_DIRTY |
 				 EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING |
 				 EXTENT_DEFRAG, 1, 1,
 				 &cached_state, GFP_NOFS);
-		free_extent_state(state);
 
 		cond_resched();
 		spin_lock(&io_tree->lock);
@@ -7530,6 +7546,7 @@ unlock:
 
 		current->journal_info = outstanding_extents;
 		btrfs_free_reserved_data_space(inode, len);
+		set_bit(BTRFS_INODE_DIO_READY, &BTRFS_I(inode)->runtime_flags);
 	}
 
 	/*
@@ -7855,8 +7872,6 @@ static void btrfs_endio_direct_write(struct bio *bio, int err)
 	struct bio *dio_bio;
 	int ret;
 
-	if (err)
-		goto out_done;
 again:
 	ret = btrfs_dec_test_first_ordered_pending(inode, &ordered,
 						   &ordered_offset,
@@ -7879,7 +7894,6 @@ out_test:
 		ordered = NULL;
 		goto again;
 	}
-out_done:
 	dio_bio = dip->dio_bio;
 
 	kfree(dip);
@@ -8147,9 +8161,8 @@ out_err:
 static void btrfs_submit_direct(int rw, struct bio *dio_bio,
 				struct inode *inode, loff_t file_offset)
 {
-	struct btrfs_root *root = BTRFS_I(inode)->root;
-	struct btrfs_dio_private *dip;
-	struct bio *io_bio;
+	struct btrfs_dio_private *dip = NULL;
+	struct bio *io_bio = NULL;
 	struct btrfs_io_bio *btrfs_bio;
 	int skip_sum;
 	int write = rw & REQ_WRITE;
@@ -8166,7 +8179,7 @@ static void btrfs_submit_direct(int rw, struct bio *dio_bio,
 	dip = kzalloc(sizeof(*dip), GFP_NOFS);
 	if (!dip) {
 		ret = -ENOMEM;
-		goto free_io_bio;
+		goto free_ordered;
 	}
 
 	dip->private = dio_bio->bi_private;
@@ -8194,25 +8207,55 @@ static void btrfs_submit_direct(int rw, struct bio *dio_bio,
 
 	if (btrfs_bio->end_io)
 		btrfs_bio->end_io(btrfs_bio, ret);
-free_io_bio:
-	bio_put(io_bio);
 
 free_ordered:
 	/*
-	 * If this is a write, we need to clean up the reserved space and kill
-	 * the ordered extent.
+	 * If we arrived here it means either we failed to submit the dip
+	 * or we either failed to clone the dio_bio or failed to allocate the
+	 * dip. If we cloned the dio_bio and allocated the dip, we can just
+	 * call bio_endio against our io_bio so that we get proper resource
+	 * cleanup if we fail to submit the dip, otherwise, we must do the
+	 * same as btrfs_endio_direct_[write|read] because we can't call these
+	 * callbacks - they require an allocated dip and a clone of dio_bio.
 	 */
-	if (write) {
-		struct btrfs_ordered_extent *ordered;
-		ordered = btrfs_lookup_ordered_extent(inode, file_offset);
-		if (!test_bit(BTRFS_ORDERED_PREALLOC, &ordered->flags) &&
-		    !test_bit(BTRFS_ORDERED_NOCOW, &ordered->flags))
-			btrfs_free_reserved_extent(root, ordered->start,
-						   ordered->disk_len, 1);
-		btrfs_put_ordered_extent(ordered);
-		btrfs_put_ordered_extent(ordered);
+	if (io_bio && dip) {
+		bio_endio(io_bio, ret);
+		/*
+		 * The end io callbacks free our dip, do the final put on io_bio
+		 * and all the cleanup and final put for dio_bio (through
+		 * dio_end_io()).
+		 */
+		dip = NULL;
+		io_bio = NULL;
+	} else {
+		if (write) {
+			struct btrfs_ordered_extent *ordered;
+
+			ordered = btrfs_lookup_ordered_extent(inode,
+							      file_offset);
+			set_bit(BTRFS_ORDERED_IOERR, &ordered->flags);
+			/*
+			 * Decrements our ref on the ordered extent and removes
+			 * the ordered extent from the inode's ordered tree,
+			 * doing all the proper resource cleanup such as for the
+			 * reserved space and waking up any waiters for this
+			 * ordered extent (through btrfs_remove_ordered_extent).
+			 */
+			btrfs_finish_ordered_io(ordered);
+		} else {
+			unlock_extent(&BTRFS_I(inode)->io_tree, file_offset,
+			      file_offset + dio_bio->bi_iter.bi_size - 1);
+		}
+		clear_bit(BIO_UPTODATE, &dio_bio->bi_flags);
+		/*
+		 * Releases and cleans up our dio_bio, no need to bio_put()
+		 * nor bio_endio()/bio_io_error() against dio_bio.
+		 */
+		dio_end_io(dio_bio, ret);
 	}
-	bio_endio(dio_bio, ret);
+	if (io_bio)
+		bio_put(io_bio);
+	kfree(dip);
 }
 
 static ssize_t check_direct_IO(struct btrfs_root *root, struct kiocb *iocb,
@@ -8314,9 +8357,18 @@ static ssize_t btrfs_direct_IO(struct kiocb *iocb, struct iov_iter *iter,
 				   btrfs_submit_direct, flags);
 	if (iov_iter_rw(iter) == WRITE) {
 		current->journal_info = NULL;
-		if (ret < 0 && ret != -EIOCBQUEUED)
-			btrfs_delalloc_release_space(inode, count);
-		else if (ret >= 0 && (size_t)ret < count)
+		if (ret < 0 && ret != -EIOCBQUEUED) {
+			/*
+			 * If the error comes from submitting stage,
+			 * btrfs_get_blocsk_direct() has free'd data space,
+			 * and metadata space will be handled by
+			 * finish_ordered_fn, don't do that again to make
+			 * sure bytes_may_use is correct.
+			 */
+			if (!test_and_clear_bit(BTRFS_INODE_DIO_READY,
+				     &BTRFS_I(inode)->runtime_flags))
+				btrfs_delalloc_release_space(inode, count);
+		} else if (ret >= 0 && (size_t)ret < count)
 			btrfs_delalloc_release_space(inode,
 						     count - (size_t)ret);
 	}
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index 37d456a9a..0770c9158 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -87,7 +87,8 @@ struct btrfs_ioctl_received_subvol_args_32 {
 
 
 static int btrfs_clone(struct inode *src, struct inode *inode,
-		       u64 off, u64 olen, u64 olen_aligned, u64 destoff);
+		       u64 off, u64 olen, u64 olen_aligned, u64 destoff,
+		       int no_time_update);
 
 /* Mask out flags that are inappropriate for the given type of inode. */
 static inline __u32 btrfs_mask_flags(umode_t mode, __u32 flags)
@@ -553,8 +554,8 @@ static noinline int create_subvol(struct inode *dir,
 	key.offset = (u64)-1;
 	new_root = btrfs_read_fs_root_no_name(root->fs_info, &key);
 	if (IS_ERR(new_root)) {
-		btrfs_abort_transaction(trans, root, PTR_ERR(new_root));
 		ret = PTR_ERR(new_root);
+		btrfs_abort_transaction(trans, root, ret);
 		goto fail;
 	}
 
@@ -1318,7 +1319,7 @@ int btrfs_defrag_file(struct inode *inode, struct file *file,
 		i = range->start >> PAGE_CACHE_SHIFT;
 	}
 	if (!max_to_defrag)
-		max_to_defrag = last_index + 1;
+		max_to_defrag = last_index - i + 1;
 
 	/*
 	 * make writeback starts from i, so the defrag range can be
@@ -1368,7 +1369,7 @@ int btrfs_defrag_file(struct inode *inode, struct file *file,
 			ra_index = max(i, ra_index);
 			btrfs_force_ra(inode->i_mapping, ra, file, ra_index,
 				       cluster);
-			ra_index += max_cluster;
+			ra_index += cluster;
 		}
 
 		mutex_lock(&inode->i_mutex);
@@ -2271,10 +2272,7 @@ static noinline int btrfs_ioctl_ino_lookup(struct file *file,
 {
 	 struct btrfs_ioctl_ino_lookup_args *args;
 	 struct inode *inode;
-	 int ret;
-
-	if (!capable(CAP_SYS_ADMIN))
-		return -EPERM;
+	int ret = 0;
 
 	args = memdup_user(argp, sizeof(*args));
 	if (IS_ERR(args))
@@ -2282,13 +2280,28 @@ static noinline int btrfs_ioctl_ino_lookup(struct file *file,
 
 	inode = file_inode(file);
 
+	/*
+	 * Unprivileged query to obtain the containing subvolume root id. The
+	 * path is reset so it's consistent with btrfs_search_path_in_tree.
+	 */
 	if (args->treeid == 0)
 		args->treeid = BTRFS_I(inode)->root->root_key.objectid;
 
+	if (args->objectid == BTRFS_FIRST_FREE_OBJECTID) {
+		args->name[0] = 0;
+		goto out;
+	}
+
+	if (!capable(CAP_SYS_ADMIN)) {
+		ret = -EPERM;
+		goto out;
+	}
+
 	ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info,
 					args->treeid, args->objectid,
 					args->name);
 
+out:
 	if (ret == 0 && copy_to_user(argp, args, sizeof(*args)))
 		ret = -EFAULT;
 
@@ -2753,14 +2766,11 @@ out:
 	return ret;
 }
 
-static struct page *extent_same_get_page(struct inode *inode, u64 off)
+static struct page *extent_same_get_page(struct inode *inode, pgoff_t index)
 {
 	struct page *page;
-	pgoff_t index;
 	struct extent_io_tree *tree = &BTRFS_I(inode)->io_tree;
 
-	index = off >> PAGE_CACHE_SHIFT;
-
 	page = grab_cache_page(inode->i_mapping, index);
 	if (!page)
 		return NULL;
@@ -2781,6 +2791,20 @@ static struct page *extent_same_get_page(struct inode *inode, u64 off)
 	return page;
 }
 
+static int gather_extent_pages(struct inode *inode, struct page **pages,
+			       int num_pages, u64 off)
+{
+	int i;
+	pgoff_t index = off >> PAGE_CACHE_SHIFT;
+
+	for (i = 0; i < num_pages; i++) {
+		pages[i] = extent_same_get_page(inode, index + i);
+		if (!pages[i])
+			return -ENOMEM;
+	}
+	return 0;
+}
+
 static inline void lock_extent_range(struct inode *inode, u64 off, u64 len)
 {
 	/* do any pending delalloc/csum calc on src, one way or
@@ -2806,52 +2830,120 @@ static inline void lock_extent_range(struct inode *inode, u64 off, u64 len)
 	}
 }
 
-static void btrfs_double_unlock(struct inode *inode1, u64 loff1,
-				struct inode *inode2, u64 loff2, u64 len)
+static void btrfs_double_inode_unlock(struct inode *inode1, struct inode *inode2)
 {
-	unlock_extent(&BTRFS_I(inode1)->io_tree, loff1, loff1 + len - 1);
-	unlock_extent(&BTRFS_I(inode2)->io_tree, loff2, loff2 + len - 1);
-
 	mutex_unlock(&inode1->i_mutex);
 	mutex_unlock(&inode2->i_mutex);
 }
 
-static void btrfs_double_lock(struct inode *inode1, u64 loff1,
-			      struct inode *inode2, u64 loff2, u64 len)
+static void btrfs_double_inode_lock(struct inode *inode1, struct inode *inode2)
+{
+	if (inode1 < inode2)
+		swap(inode1, inode2);
+
+	mutex_lock_nested(&inode1->i_mutex, I_MUTEX_PARENT);
+	if (inode1 != inode2)
+		mutex_lock_nested(&inode2->i_mutex, I_MUTEX_CHILD);
+}
+
+static void btrfs_double_extent_unlock(struct inode *inode1, u64 loff1,
+				      struct inode *inode2, u64 loff2, u64 len)
+{
+	unlock_extent(&BTRFS_I(inode1)->io_tree, loff1, loff1 + len - 1);
+	unlock_extent(&BTRFS_I(inode2)->io_tree, loff2, loff2 + len - 1);
+}
+
+static void btrfs_double_extent_lock(struct inode *inode1, u64 loff1,
+				     struct inode *inode2, u64 loff2, u64 len)
 {
 	if (inode1 < inode2) {
 		swap(inode1, inode2);
 		swap(loff1, loff2);
 	}
-
-	mutex_lock_nested(&inode1->i_mutex, I_MUTEX_PARENT);
 	lock_extent_range(inode1, loff1, len);
-	if (inode1 != inode2) {
-		mutex_lock_nested(&inode2->i_mutex, I_MUTEX_CHILD);
+	if (inode1 != inode2)
 		lock_extent_range(inode2, loff2, len);
+}
+
+struct cmp_pages {
+	int		num_pages;
+	struct page	**src_pages;
+	struct page	**dst_pages;
+};
+
+static void btrfs_cmp_data_free(struct cmp_pages *cmp)
+{
+	int i;
+	struct page *pg;
+
+	for (i = 0; i < cmp->num_pages; i++) {
+		pg = cmp->src_pages[i];
+		if (pg)
+			page_cache_release(pg);
+		pg = cmp->dst_pages[i];
+		if (pg)
+			page_cache_release(pg);
+	}
+	kfree(cmp->src_pages);
+	kfree(cmp->dst_pages);
+}
+
+static int btrfs_cmp_data_prepare(struct inode *src, u64 loff,
+				  struct inode *dst, u64 dst_loff,
+				  u64 len, struct cmp_pages *cmp)
+{
+	int ret;
+	int num_pages = PAGE_CACHE_ALIGN(len) >> PAGE_CACHE_SHIFT;
+	struct page **src_pgarr, **dst_pgarr;
+
+	/*
+	 * We must gather up all the pages before we initiate our
+	 * extent locking. We use an array for the page pointers. Size
+	 * of the array is bounded by len, which is in turn bounded by
+	 * BTRFS_MAX_DEDUPE_LEN.
+	 */
+	src_pgarr = kzalloc(num_pages * sizeof(struct page *), GFP_NOFS);
+	dst_pgarr = kzalloc(num_pages * sizeof(struct page *), GFP_NOFS);
+	if (!src_pgarr || !dst_pgarr) {
+		kfree(src_pgarr);
+		kfree(dst_pgarr);
+		return -ENOMEM;
 	}
+	cmp->num_pages = num_pages;
+	cmp->src_pages = src_pgarr;
+	cmp->dst_pages = dst_pgarr;
+
+	ret = gather_extent_pages(src, cmp->src_pages, cmp->num_pages, loff);
+	if (ret)
+		goto out;
+
+	ret = gather_extent_pages(dst, cmp->dst_pages, cmp->num_pages, dst_loff);
+
+out:
+	if (ret)
+		btrfs_cmp_data_free(cmp);
+	return 0;
 }
 
 static int btrfs_cmp_data(struct inode *src, u64 loff, struct inode *dst,
-			  u64 dst_loff, u64 len)
+			  u64 dst_loff, u64 len, struct cmp_pages *cmp)
 {
 	int ret = 0;
+	int i;
 	struct page *src_page, *dst_page;
 	unsigned int cmp_len = PAGE_CACHE_SIZE;
 	void *addr, *dst_addr;
 
+	i = 0;
 	while (len) {
 		if (len < PAGE_CACHE_SIZE)
 			cmp_len = len;
 
-		src_page = extent_same_get_page(src, loff);
-		if (!src_page)
-			return -EINVAL;
-		dst_page = extent_same_get_page(dst, dst_loff);
-		if (!dst_page) {
-			page_cache_release(src_page);
-			return -EINVAL;
-		}
+		BUG_ON(i >= cmp->num_pages);
+
+		src_page = cmp->src_pages[i];
+		dst_page = cmp->dst_pages[i];
+
 		addr = kmap_atomic(src_page);
 		dst_addr = kmap_atomic(dst_page);
 
@@ -2863,26 +2955,30 @@ static int btrfs_cmp_data(struct inode *src, u64 loff, struct inode *dst,
 
 		kunmap_atomic(addr);
 		kunmap_atomic(dst_addr);
-		page_cache_release(src_page);
-		page_cache_release(dst_page);
 
 		if (ret)
 			break;
 
-		loff += cmp_len;
-		dst_loff += cmp_len;
 		len -= cmp_len;
+		i++;
 	}
 
 	return ret;
 }
 
-static int extent_same_check_offsets(struct inode *inode, u64 off, u64 len)
+static int extent_same_check_offsets(struct inode *inode, u64 off, u64 *plen,
+				     u64 olen)
 {
+	u64 len = *plen;
 	u64 bs = BTRFS_I(inode)->root->fs_info->sb->s_blocksize;
 
-	if (off + len > inode->i_size || off + len < off)
+	if (off + olen > inode->i_size || off + olen < off)
 		return -EINVAL;
+
+	/* if we extend to eof, continue to block boundary */
+	if (off + len == inode->i_size)
+		*plen = len = ALIGN(inode->i_size, bs) - off;
+
 	/* Check that we are block aligned - btrfs_clone() requires this */
 	if (!IS_ALIGNED(off, bs) || !IS_ALIGNED(off + len, bs))
 		return -EINVAL;
@@ -2890,31 +2986,67 @@ static int extent_same_check_offsets(struct inode *inode, u64 off, u64 len)
 	return 0;
 }
 
-static int btrfs_extent_same(struct inode *src, u64 loff, u64 len,
+static int btrfs_extent_same(struct inode *src, u64 loff, u64 olen,
 			     struct inode *dst, u64 dst_loff)
 {
 	int ret;
+	u64 len = olen;
+	struct cmp_pages cmp;
+	int same_inode = 0;
+	u64 same_lock_start = 0;
+	u64 same_lock_len = 0;
 
-	/*
-	 * btrfs_clone() can't handle extents in the same file
-	 * yet. Once that works, we can drop this check and replace it
-	 * with a check for the same inode, but overlapping extents.
-	 */
 	if (src == dst)
-		return -EINVAL;
+		same_inode = 1;
 
 	if (len == 0)
 		return 0;
 
-	btrfs_double_lock(src, loff, dst, dst_loff, len);
+	if (same_inode) {
+		mutex_lock(&src->i_mutex);
 
-	ret = extent_same_check_offsets(src, loff, len);
-	if (ret)
-		goto out_unlock;
+		ret = extent_same_check_offsets(src, loff, &len, olen);
+		if (ret)
+			goto out_unlock;
 
-	ret = extent_same_check_offsets(dst, dst_loff, len);
-	if (ret)
-		goto out_unlock;
+		/*
+		 * Single inode case wants the same checks, except we
+		 * don't want our length pushed out past i_size as
+		 * comparing that data range makes no sense.
+		 *
+		 * extent_same_check_offsets() will do this for an
+		 * unaligned length at i_size, so catch it here and
+		 * reject the request.
+		 *
+		 * This effectively means we require aligned extents
+		 * for the single-inode case, whereas the other cases
+		 * allow an unaligned length so long as it ends at
+		 * i_size.
+		 */
+		if (len != olen) {
+			ret = -EINVAL;
+			goto out_unlock;
+		}
+
+		/* Check for overlapping ranges */
+		if (dst_loff + len > loff && dst_loff < loff + len) {
+			ret = -EINVAL;
+			goto out_unlock;
+		}
+
+		same_lock_start = min_t(u64, loff, dst_loff);
+		same_lock_len = max_t(u64, loff, dst_loff) + len - same_lock_start;
+	} else {
+		btrfs_double_inode_lock(src, dst);
+
+		ret = extent_same_check_offsets(src, loff, &len, olen);
+		if (ret)
+			goto out_unlock;
+
+		ret = extent_same_check_offsets(dst, dst_loff, &len, olen);
+		if (ret)
+			goto out_unlock;
+	}
 
 	/* don't make the dst file partly checksummed */
 	if ((BTRFS_I(src)->flags & BTRFS_INODE_NODATASUM) !=
@@ -2923,12 +3055,32 @@ static int btrfs_extent_same(struct inode *src, u64 loff, u64 len,
 		goto out_unlock;
 	}
 
-	ret = btrfs_cmp_data(src, loff, dst, dst_loff, len);
+	ret = btrfs_cmp_data_prepare(src, loff, dst, dst_loff, olen, &cmp);
+	if (ret)
+		goto out_unlock;
+
+	if (same_inode)
+		lock_extent_range(src, same_lock_start, same_lock_len);
+	else
+		btrfs_double_extent_lock(src, loff, dst, dst_loff, len);
+
+	/* pass original length for comparison so we stay within i_size */
+	ret = btrfs_cmp_data(src, loff, dst, dst_loff, olen, &cmp);
 	if (ret == 0)
-		ret = btrfs_clone(src, dst, loff, len, len, dst_loff);
+		ret = btrfs_clone(src, dst, loff, olen, len, dst_loff, 1);
 
+	if (same_inode)
+		unlock_extent(&BTRFS_I(src)->io_tree, same_lock_start,
+			      same_lock_start + same_lock_len - 1);
+	else
+		btrfs_double_extent_unlock(src, loff, dst, dst_loff, len);
+
+	btrfs_cmp_data_free(&cmp);
 out_unlock:
-	btrfs_double_unlock(src, loff, dst, dst_loff, len);
+	if (same_inode)
+		mutex_unlock(&src->i_mutex);
+	else
+		btrfs_double_inode_unlock(src, dst);
 
 	return ret;
 }
@@ -3082,13 +3234,15 @@ static int clone_finish_inode_update(struct btrfs_trans_handle *trans,
 				     struct inode *inode,
 				     u64 endoff,
 				     const u64 destoff,
-				     const u64 olen)
+				     const u64 olen,
+				     int no_time_update)
 {
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	int ret;
 
 	inode_inc_iversion(inode);
-	inode->i_mtime = inode->i_ctime = CURRENT_TIME;
+	if (!no_time_update)
+		inode->i_mtime = inode->i_ctime = CURRENT_TIME;
 	/*
 	 * We round up to the block size at eof when determining which
 	 * extents to clone above, but shouldn't round up the file size.
@@ -3173,13 +3327,13 @@ static void clone_update_extent_map(struct inode *inode,
  * @inode: Inode to clone to
  * @off: Offset within source to start clone from
  * @olen: Original length, passed by user, of range to clone
- * @olen_aligned: Block-aligned value of olen, extent_same uses
- *               identical values here
+ * @olen_aligned: Block-aligned value of olen
  * @destoff: Offset within @inode to start clone
+ * @no_time_update: Whether to update mtime/ctime on the target inode
  */
 static int btrfs_clone(struct inode *src, struct inode *inode,
 		       const u64 off, const u64 olen, const u64 olen_aligned,
-		       const u64 destoff)
+		       const u64 destoff, int no_time_update)
 {
 	struct btrfs_root *root = BTRFS_I(inode)->root;
 	struct btrfs_path *path = NULL;
@@ -3517,7 +3671,8 @@ process_slot:
 					      root->sectorsize);
 			ret = clone_finish_inode_update(trans, inode,
 							last_dest_end,
-							destoff, olen);
+							destoff, olen,
+							no_time_update);
 			if (ret)
 				goto out;
 			if (new_key.offset + datal >= destoff + len)
@@ -3555,7 +3710,7 @@ process_slot:
 		clone_update_extent_map(inode, trans, NULL, last_dest_end,
 					destoff + len - last_dest_end);
 		ret = clone_finish_inode_update(trans, inode, destoff + len,
-						destoff, olen);
+						destoff, olen, no_time_update);
 	}
 
 out:
@@ -3692,7 +3847,7 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 		lock_extent_range(inode, destoff, len);
 	}
 
-	ret = btrfs_clone(src, inode, off, olen, len, destoff);
+	ret = btrfs_clone(src, inode, off, olen, len, destoff, 0);
 
 	if (same_inode) {
 		u64 lock_start = min_t(u64, off, destoff);
diff --git a/fs/btrfs/ordered-data.c b/fs/btrfs/ordered-data.c
index 760c4a5e0..52170cf17 100644
--- a/fs/btrfs/ordered-data.c
+++ b/fs/btrfs/ordered-data.c
@@ -198,9 +198,6 @@ static int __btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
 	entry->file_offset = file_offset;
 	entry->start = start;
 	entry->len = len;
-	if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM) &&
-	    !(type == BTRFS_ORDERED_NOCOW))
-		entry->csum_bytes_left = disk_len;
 	entry->disk_len = disk_len;
 	entry->bytes_left = len;
 	entry->inode = igrab(inode);
@@ -286,10 +283,6 @@ void btrfs_add_ordered_sum(struct inode *inode,
 	tree = &BTRFS_I(inode)->ordered_tree;
 	spin_lock_irq(&tree->lock);
 	list_add_tail(&sum->list, &entry->list);
-	WARN_ON(entry->csum_bytes_left < sum->len);
-	entry->csum_bytes_left -= sum->len;
-	if (entry->csum_bytes_left == 0)
-		wake_up(&entry->wait);
 	spin_unlock_irq(&tree->lock);
 }
 
@@ -509,7 +502,21 @@ void btrfs_wait_logged_extents(struct btrfs_trans_handle *trans,
 		wait_event(ordered->wait, test_bit(BTRFS_ORDERED_IO_DONE,
 						   &ordered->flags));
 
-		list_add_tail(&ordered->trans_list, &trans->ordered);
+		/*
+		 * If our ordered extent completed it means it updated the
+		 * fs/subvol and csum trees already, so no need to make the
+		 * current transaction's commit wait for it, as we end up
+		 * holding memory unnecessarily and delaying the inode's iput
+		 * until the transaction commit (we schedule an iput for the
+		 * inode when the ordered extent's refcount drops to 0), which
+		 * prevents it from being evictable until the transaction
+		 * commits.
+		 */
+		if (test_bit(BTRFS_ORDERED_COMPLETE, &ordered->flags))
+			btrfs_put_ordered_extent(ordered);
+		else
+			list_add_tail(&ordered->trans_list, &trans->ordered);
+
 		spin_lock_irq(&log->log_extents_lock[index]);
 	}
 	spin_unlock_irq(&log->log_extents_lock[index]);
@@ -545,6 +552,10 @@ void btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry)
 	trace_btrfs_ordered_extent_put(entry->inode, entry);
 
 	if (atomic_dec_and_test(&entry->refs)) {
+		ASSERT(list_empty(&entry->log_list));
+		ASSERT(list_empty(&entry->trans_list));
+		ASSERT(list_empty(&entry->root_extent_list));
+		ASSERT(RB_EMPTY_NODE(&entry->rb_node));
 		if (entry->inode)
 			btrfs_add_delayed_iput(entry->inode);
 		while (!list_empty(&entry->list)) {
@@ -572,6 +583,7 @@ void btrfs_remove_ordered_extent(struct inode *inode,
 	spin_lock_irq(&tree->lock);
 	node = &entry->rb_node;
 	rb_erase(node, &tree->tree);
+	RB_CLEAR_NODE(node);
 	if (tree->last == node)
 		tree->last = NULL;
 	set_bit(BTRFS_ORDERED_COMPLETE, &entry->flags);
@@ -844,6 +856,20 @@ out:
 	return entry;
 }
 
+bool btrfs_have_ordered_extents_in_range(struct inode *inode,
+					 u64 file_offset,
+					 u64 len)
+{
+	struct btrfs_ordered_extent *oe;
+
+	oe = btrfs_lookup_ordered_range(inode, file_offset, len);
+	if (oe) {
+		btrfs_put_ordered_extent(oe);
+		return true;
+	}
+	return false;
+}
+
 /*
  * lookup and return any extent before 'file_offset'.  NULL is returned
  * if none is found
diff --git a/fs/btrfs/ordered-data.h b/fs/btrfs/ordered-data.h
index e96cd4ccd..7176cc0fe 100644
--- a/fs/btrfs/ordered-data.h
+++ b/fs/btrfs/ordered-data.h
@@ -89,9 +89,6 @@ struct btrfs_ordered_extent {
 	/* number of bytes that still need writing */
 	u64 bytes_left;
 
-	/* number of bytes that still need csumming */
-	u64 csum_bytes_left;
-
 	/*
 	 * the end of the ordered extent which is behind it but
 	 * didn't update disk_i_size. Please see the comment of
@@ -191,6 +188,9 @@ btrfs_lookup_first_ordered_extent(struct inode * inode, u64 file_offset);
 struct btrfs_ordered_extent *btrfs_lookup_ordered_range(struct inode *inode,
 							u64 file_offset,
 							u64 len);
+bool btrfs_have_ordered_extents_in_range(struct inode *inode,
+					 u64 file_offset,
+					 u64 len);
 int btrfs_ordered_update_i_size(struct inode *inode, u64 offset,
 				struct btrfs_ordered_extent *ordered);
 int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr,
diff --git a/fs/btrfs/qgroup.c b/fs/btrfs/qgroup.c
index 3d6546581..8a8202956 100644
--- a/fs/btrfs/qgroup.c
+++ b/fs/btrfs/qgroup.c
@@ -34,6 +34,7 @@
 #include "extent_io.h"
 #include "qgroup.h"
 
+
 /* TODO XXX FIXME
  *  - subvol delete -> delete when ref goes to 0? delete limits also?
  *  - reorganize keys
@@ -84,11 +85,42 @@ struct btrfs_qgroup {
 
 	/*
 	 * temp variables for accounting operations
+	 * Refer to qgroup_shared_accouting() for details.
 	 */
 	u64 old_refcnt;
 	u64 new_refcnt;
 };
 
+static void btrfs_qgroup_update_old_refcnt(struct btrfs_qgroup *qg, u64 seq,
+					   int mod)
+{
+	if (qg->old_refcnt < seq)
+		qg->old_refcnt = seq;
+	qg->old_refcnt += mod;
+}
+
+static void btrfs_qgroup_update_new_refcnt(struct btrfs_qgroup *qg, u64 seq,
+					   int mod)
+{
+	if (qg->new_refcnt < seq)
+		qg->new_refcnt = seq;
+	qg->new_refcnt += mod;
+}
+
+static inline u64 btrfs_qgroup_get_old_refcnt(struct btrfs_qgroup *qg, u64 seq)
+{
+	if (qg->old_refcnt < seq)
+		return 0;
+	return qg->old_refcnt - seq;
+}
+
+static inline u64 btrfs_qgroup_get_new_refcnt(struct btrfs_qgroup *qg, u64 seq)
+{
+	if (qg->new_refcnt < seq)
+		return 0;
+	return qg->new_refcnt - seq;
+}
+
 /*
  * glue structure to represent the relations between qgroups.
  */
@@ -1115,14 +1147,14 @@ int btrfs_add_qgroup_relation(struct btrfs_trans_handle *trans,
 	struct ulist *tmp;
 	int ret = 0;
 
-	tmp = ulist_alloc(GFP_NOFS);
-	if (!tmp)
-		return -ENOMEM;
-
 	/* Check the level of src and dst first */
 	if (btrfs_qgroup_level(src) >= btrfs_qgroup_level(dst))
 		return -EINVAL;
 
+	tmp = ulist_alloc(GFP_NOFS);
+	if (!tmp)
+		return -ENOMEM;
+
 	mutex_lock(&fs_info->qgroup_ioctl_lock);
 	quota_root = fs_info->quota_root;
 	if (!quota_root) {
@@ -1317,6 +1349,11 @@ int btrfs_limit_qgroup(struct btrfs_trans_handle *trans,
 	struct btrfs_root *quota_root;
 	struct btrfs_qgroup *qgroup;
 	int ret = 0;
+	/* Sometimes we would want to clear the limit on this qgroup.
+	 * To meet this requirement, we treat the -1 as a special value
+	 * which tell kernel to clear the limit on this qgroup.
+	 */
+	const u64 CLEAR_VALUE = -1;
 
 	mutex_lock(&fs_info->qgroup_ioctl_lock);
 	quota_root = fs_info->quota_root;
@@ -1332,14 +1369,42 @@ int btrfs_limit_qgroup(struct btrfs_trans_handle *trans,
 	}
 
 	spin_lock(&fs_info->qgroup_lock);
-	if (limit->flags & BTRFS_QGROUP_LIMIT_MAX_RFER)
-		qgroup->max_rfer = limit->max_rfer;
-	if (limit->flags & BTRFS_QGROUP_LIMIT_MAX_EXCL)
-		qgroup->max_excl = limit->max_excl;
-	if (limit->flags & BTRFS_QGROUP_LIMIT_RSV_RFER)
-		qgroup->rsv_rfer = limit->rsv_rfer;
-	if (limit->flags & BTRFS_QGROUP_LIMIT_RSV_EXCL)
-		qgroup->rsv_excl = limit->rsv_excl;
+	if (limit->flags & BTRFS_QGROUP_LIMIT_MAX_RFER) {
+		if (limit->max_rfer == CLEAR_VALUE) {
+			qgroup->lim_flags &= ~BTRFS_QGROUP_LIMIT_MAX_RFER;
+			limit->flags &= ~BTRFS_QGROUP_LIMIT_MAX_RFER;
+			qgroup->max_rfer = 0;
+		} else {
+			qgroup->max_rfer = limit->max_rfer;
+		}
+	}
+	if (limit->flags & BTRFS_QGROUP_LIMIT_MAX_EXCL) {
+		if (limit->max_excl == CLEAR_VALUE) {
+			qgroup->lim_flags &= ~BTRFS_QGROUP_LIMIT_MAX_EXCL;
+			limit->flags &= ~BTRFS_QGROUP_LIMIT_MAX_EXCL;
+			qgroup->max_excl = 0;
+		} else {
+			qgroup->max_excl = limit->max_excl;
+		}
+	}
+	if (limit->flags & BTRFS_QGROUP_LIMIT_RSV_RFER) {
+		if (limit->rsv_rfer == CLEAR_VALUE) {
+			qgroup->lim_flags &= ~BTRFS_QGROUP_LIMIT_RSV_RFER;
+			limit->flags &= ~BTRFS_QGROUP_LIMIT_RSV_RFER;
+			qgroup->rsv_rfer = 0;
+		} else {
+			qgroup->rsv_rfer = limit->rsv_rfer;
+		}
+	}
+	if (limit->flags & BTRFS_QGROUP_LIMIT_RSV_EXCL) {
+		if (limit->rsv_excl == CLEAR_VALUE) {
+			qgroup->lim_flags &= ~BTRFS_QGROUP_LIMIT_RSV_EXCL;
+			limit->flags &= ~BTRFS_QGROUP_LIMIT_RSV_EXCL;
+			qgroup->rsv_excl = 0;
+		} else {
+			qgroup->rsv_excl = limit->rsv_excl;
+		}
+	}
 	qgroup->lim_flags |= limit->flags;
 
 	spin_unlock(&fs_info->qgroup_lock);
@@ -1356,239 +1421,86 @@ out:
 	return ret;
 }
 
-static int comp_oper_exist(struct btrfs_qgroup_operation *oper1,
-			   struct btrfs_qgroup_operation *oper2)
+int btrfs_qgroup_prepare_account_extents(struct btrfs_trans_handle *trans,
+					 struct btrfs_fs_info *fs_info)
 {
-	/*
-	 * Ignore seq and type here, we're looking for any operation
-	 * at all related to this extent on that root.
-	 */
-	if (oper1->bytenr < oper2->bytenr)
-		return -1;
-	if (oper1->bytenr > oper2->bytenr)
-		return 1;
-	if (oper1->ref_root < oper2->ref_root)
-		return -1;
-	if (oper1->ref_root > oper2->ref_root)
-		return 1;
-	return 0;
-}
+	struct btrfs_qgroup_extent_record *record;
+	struct btrfs_delayed_ref_root *delayed_refs;
+	struct rb_node *node;
+	u64 qgroup_to_skip;
+	int ret = 0;
 
-static int qgroup_oper_exists(struct btrfs_fs_info *fs_info,
-			      struct btrfs_qgroup_operation *oper)
-{
-	struct rb_node *n;
-	struct btrfs_qgroup_operation *cur;
-	int cmp;
+	delayed_refs = &trans->transaction->delayed_refs;
+	qgroup_to_skip = delayed_refs->qgroup_to_skip;
 
-	spin_lock(&fs_info->qgroup_op_lock);
-	n = fs_info->qgroup_op_tree.rb_node;
-	while (n) {
-		cur = rb_entry(n, struct btrfs_qgroup_operation, n);
-		cmp = comp_oper_exist(cur, oper);
-		if (cmp < 0) {
-			n = n->rb_right;
-		} else if (cmp) {
-			n = n->rb_left;
-		} else {
-			spin_unlock(&fs_info->qgroup_op_lock);
-			return -EEXIST;
-		}
+	/*
+	 * No need to do lock, since this function will only be called in
+	 * btrfs_commmit_transaction().
+	 */
+	node = rb_first(&delayed_refs->dirty_extent_root);
+	while (node) {
+		record = rb_entry(node, struct btrfs_qgroup_extent_record,
+				  node);
+		ret = btrfs_find_all_roots(NULL, fs_info, record->bytenr, 0,
+					   &record->old_roots);
+		if (ret < 0)
+			break;
+		if (qgroup_to_skip)
+			ulist_del(record->old_roots, qgroup_to_skip, 0);
+		node = rb_next(node);
 	}
-	spin_unlock(&fs_info->qgroup_op_lock);
-	return 0;
-}
-
-static int comp_oper(struct btrfs_qgroup_operation *oper1,
-		     struct btrfs_qgroup_operation *oper2)
-{
-	if (oper1->bytenr < oper2->bytenr)
-		return -1;
-	if (oper1->bytenr > oper2->bytenr)
-		return 1;
-	if (oper1->ref_root < oper2->ref_root)
-		return -1;
-	if (oper1->ref_root > oper2->ref_root)
-		return 1;
-	if (oper1->seq < oper2->seq)
-		return -1;
-	if (oper1->seq > oper2->seq)
-		return 1;
-	if (oper1->type < oper2->type)
-		return -1;
-	if (oper1->type > oper2->type)
-		return 1;
-	return 0;
+	return ret;
 }
 
-static int insert_qgroup_oper(struct btrfs_fs_info *fs_info,
-			      struct btrfs_qgroup_operation *oper)
+struct btrfs_qgroup_extent_record
+*btrfs_qgroup_insert_dirty_extent(struct btrfs_delayed_ref_root *delayed_refs,
+				  struct btrfs_qgroup_extent_record *record)
 {
-	struct rb_node **p;
-	struct rb_node *parent = NULL;
-	struct btrfs_qgroup_operation *cur;
-	int cmp;
+	struct rb_node **p = &delayed_refs->dirty_extent_root.rb_node;
+	struct rb_node *parent_node = NULL;
+	struct btrfs_qgroup_extent_record *entry;
+	u64 bytenr = record->bytenr;
 
-	spin_lock(&fs_info->qgroup_op_lock);
-	p = &fs_info->qgroup_op_tree.rb_node;
 	while (*p) {
-		parent = *p;
-		cur = rb_entry(parent, struct btrfs_qgroup_operation, n);
-		cmp = comp_oper(cur, oper);
-		if (cmp < 0) {
-			p = &(*p)->rb_right;
-		} else if (cmp) {
+		parent_node = *p;
+		entry = rb_entry(parent_node, struct btrfs_qgroup_extent_record,
+				 node);
+		if (bytenr < entry->bytenr)
 			p = &(*p)->rb_left;
-		} else {
-			spin_unlock(&fs_info->qgroup_op_lock);
-			return -EEXIST;
-		}
-	}
-	rb_link_node(&oper->n, parent, p);
-	rb_insert_color(&oper->n, &fs_info->qgroup_op_tree);
-	spin_unlock(&fs_info->qgroup_op_lock);
-	return 0;
-}
-
-/*
- * Record a quota operation for processing later on.
- * @trans: the transaction we are adding the delayed op to.
- * @fs_info: the fs_info for this fs.
- * @ref_root: the root of the reference we are acting on,
- * @bytenr: the bytenr we are acting on.
- * @num_bytes: the number of bytes in the reference.
- * @type: the type of operation this is.
- * @mod_seq: do we need to get a sequence number for looking up roots.
- *
- * We just add it to our trans qgroup_ref_list and carry on and process these
- * operations in order at some later point.  If the reference root isn't a fs
- * root then we don't bother with doing anything.
- *
- * MUST BE HOLDING THE REF LOCK.
- */
-int btrfs_qgroup_record_ref(struct btrfs_trans_handle *trans,
-			    struct btrfs_fs_info *fs_info, u64 ref_root,
-			    u64 bytenr, u64 num_bytes,
-			    enum btrfs_qgroup_operation_type type, int mod_seq)
-{
-	struct btrfs_qgroup_operation *oper;
-	int ret;
-
-	if (!is_fstree(ref_root) || !fs_info->quota_enabled)
-		return 0;
-
-	oper = kmalloc(sizeof(*oper), GFP_NOFS);
-	if (!oper)
-		return -ENOMEM;
-
-	oper->ref_root = ref_root;
-	oper->bytenr = bytenr;
-	oper->num_bytes = num_bytes;
-	oper->type = type;
-	oper->seq = atomic_inc_return(&fs_info->qgroup_op_seq);
-	INIT_LIST_HEAD(&oper->elem.list);
-	oper->elem.seq = 0;
-
-	trace_btrfs_qgroup_record_ref(oper);
-
-	if (type == BTRFS_QGROUP_OPER_SUB_SUBTREE) {
-		/*
-		 * If any operation for this bytenr/ref_root combo
-		 * exists, then we know it's not exclusively owned and
-		 * shouldn't be queued up.
-		 *
-		 * This also catches the case where we have a cloned
-		 * extent that gets queued up multiple times during
-		 * drop snapshot.
-		 */
-		if (qgroup_oper_exists(fs_info, oper)) {
-			kfree(oper);
-			return 0;
-		}
-	}
-
-	ret = insert_qgroup_oper(fs_info, oper);
-	if (ret) {
-		/* Shouldn't happen so have an assert for developers */
-		ASSERT(0);
-		kfree(oper);
-		return ret;
+		else if (bytenr > entry->bytenr)
+			p = &(*p)->rb_right;
+		else
+			return entry;
 	}
-	list_add_tail(&oper->list, &trans->qgroup_ref_list);
-
-	if (mod_seq)
-		btrfs_get_tree_mod_seq(fs_info, &oper->elem);
-
-	return 0;
-}
-
-static int qgroup_excl_accounting(struct btrfs_fs_info *fs_info,
-				  struct btrfs_qgroup_operation *oper)
-{
-	struct ulist *tmp;
-	int sign = 0;
-	int ret = 0;
 
-	tmp = ulist_alloc(GFP_NOFS);
-	if (!tmp)
-		return -ENOMEM;
-
-	spin_lock(&fs_info->qgroup_lock);
-	if (!fs_info->quota_root)
-		goto out;
-
-	switch (oper->type) {
-	case BTRFS_QGROUP_OPER_ADD_EXCL:
-		sign = 1;
-		break;
-	case BTRFS_QGROUP_OPER_SUB_EXCL:
-		sign = -1;
-		break;
-	default:
-		ASSERT(0);
-	}
-	ret = __qgroup_excl_accounting(fs_info, tmp, oper->ref_root,
-				       oper->num_bytes, sign);
-out:
-	spin_unlock(&fs_info->qgroup_lock);
-	ulist_free(tmp);
-	return ret;
+	rb_link_node(&record->node, parent_node, p);
+	rb_insert_color(&record->node, &delayed_refs->dirty_extent_root);
+	return NULL;
 }
 
+#define UPDATE_NEW	0
+#define UPDATE_OLD	1
 /*
- * Walk all of the roots that pointed to our bytenr and adjust their refcnts as
- * properly.
+ * Walk all of the roots that points to the bytenr and adjust their refcnts.
  */
-static int qgroup_calc_old_refcnt(struct btrfs_fs_info *fs_info,
-				  u64 root_to_skip, struct ulist *tmp,
-				  struct ulist *roots, struct ulist *qgroups,
-				  u64 seq, int *old_roots, int rescan)
+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)
 {
 	struct ulist_node *unode;
 	struct ulist_iterator uiter;
 	struct ulist_node *tmp_unode;
 	struct ulist_iterator tmp_uiter;
 	struct btrfs_qgroup *qg;
-	int ret;
+	int ret = 0;
 
+	if (!roots)
+		return 0;
 	ULIST_ITER_INIT(&uiter);
 	while ((unode = ulist_next(roots, &uiter))) {
-		/* We don't count our current root here */
-		if (unode->val == root_to_skip)
-			continue;
 		qg = find_qgroup_rb(fs_info, unode->val);
 		if (!qg)
 			continue;
-		/*
-		 * We could have a pending removal of this same ref so we may
-		 * not have actually found our ref root when doing
-		 * btrfs_find_all_roots, so we need to keep track of how many
-		 * old roots we find in case we removed ours and added a
-		 * different one at the same time.  I don't think this could
-		 * happen in practice but that sort of thinking leads to pain
-		 * and suffering and to the dark side.
-		 */
-		(*old_roots)++;
 
 		ulist_reinit(tmp);
 		ret = ulist_add(qgroups, qg->qgroupid, ptr_to_u64(qg),
@@ -1603,29 +1515,10 @@ static int qgroup_calc_old_refcnt(struct btrfs_fs_info *fs_info,
 			struct btrfs_qgroup_list *glist;
 
 			qg = u64_to_ptr(tmp_unode->aux);
-			/*
-			 * We use this sequence number to keep from having to
-			 * run the whole list and 0 out the refcnt every time.
-			 * We basically use sequnce as the known 0 count and
-			 * then add 1 everytime we see a qgroup.  This is how we
-			 * get how many of the roots actually point up to the
-			 * upper level qgroups in order to determine exclusive
-			 * counts.
-			 *
-			 * For rescan we want to set old_refcnt to seq so our
-			 * exclusive calculations end up correct.
-			 */
-			if (rescan)
-				qg->old_refcnt = seq;
-			else if (qg->old_refcnt < seq)
-				qg->old_refcnt = seq + 1;
+			if (update_old)
+				btrfs_qgroup_update_old_refcnt(qg, seq, 1);
 			else
-				qg->old_refcnt++;
-
-			if (qg->new_refcnt < seq)
-				qg->new_refcnt = seq + 1;
-			else
-				qg->new_refcnt++;
+				btrfs_qgroup_update_new_refcnt(qg, seq, 1);
 			list_for_each_entry(glist, &qg->groups, next_group) {
 				ret = ulist_add(qgroups, glist->group->qgroupid,
 						ptr_to_u64(glist->group),
@@ -1644,161 +1537,46 @@ static int qgroup_calc_old_refcnt(struct btrfs_fs_info *fs_info,
 }
 
 /*
- * We need to walk forward in our operation tree and account for any roots that
- * were deleted after we made this operation.
- */
-static int qgroup_account_deleted_refs(struct btrfs_fs_info *fs_info,
-				       struct btrfs_qgroup_operation *oper,
-				       struct ulist *tmp,
-				       struct ulist *qgroups, u64 seq,
-				       int *old_roots)
-{
-	struct ulist_node *unode;
-	struct ulist_iterator uiter;
-	struct btrfs_qgroup *qg;
-	struct btrfs_qgroup_operation *tmp_oper;
-	struct rb_node *n;
-	int ret;
-
-	ulist_reinit(tmp);
-
-	/*
-	 * We only walk forward in the tree since we're only interested in
-	 * removals that happened _after_  our operation.
-	 */
-	spin_lock(&fs_info->qgroup_op_lock);
-	n = rb_next(&oper->n);
-	spin_unlock(&fs_info->qgroup_op_lock);
-	if (!n)
-		return 0;
-	tmp_oper = rb_entry(n, struct btrfs_qgroup_operation, n);
-	while (tmp_oper->bytenr == oper->bytenr) {
-		/*
-		 * If it's not a removal we don't care, additions work out
-		 * properly with our refcnt tracking.
-		 */
-		if (tmp_oper->type != BTRFS_QGROUP_OPER_SUB_SHARED &&
-		    tmp_oper->type != BTRFS_QGROUP_OPER_SUB_EXCL)
-			goto next;
-		qg = find_qgroup_rb(fs_info, tmp_oper->ref_root);
-		if (!qg)
-			goto next;
-		ret = ulist_add(qgroups, qg->qgroupid, ptr_to_u64(qg),
-				GFP_ATOMIC);
-		if (ret) {
-			if (ret < 0)
-				return ret;
-			/*
-			 * We only want to increase old_roots if this qgroup is
-			 * not already in the list of qgroups.  If it is already
-			 * there then that means it must have been re-added or
-			 * the delete will be discarded because we had an
-			 * existing ref that we haven't looked up yet.  In this
-			 * case we don't want to increase old_roots.  So if ret
-			 * == 1 then we know that this is the first time we've
-			 * seen this qgroup and we can bump the old_roots.
-			 */
-			(*old_roots)++;
-			ret = ulist_add(tmp, qg->qgroupid, ptr_to_u64(qg),
-					GFP_ATOMIC);
-			if (ret < 0)
-				return ret;
-		}
-next:
-		spin_lock(&fs_info->qgroup_op_lock);
-		n = rb_next(&tmp_oper->n);
-		spin_unlock(&fs_info->qgroup_op_lock);
-		if (!n)
-			break;
-		tmp_oper = rb_entry(n, struct btrfs_qgroup_operation, n);
-	}
-
-	/* Ok now process the qgroups we found */
-	ULIST_ITER_INIT(&uiter);
-	while ((unode = ulist_next(tmp, &uiter))) {
-		struct btrfs_qgroup_list *glist;
-
-		qg = u64_to_ptr(unode->aux);
-		if (qg->old_refcnt < seq)
-			qg->old_refcnt = seq + 1;
-		else
-			qg->old_refcnt++;
-		if (qg->new_refcnt < seq)
-			qg->new_refcnt = seq + 1;
-		else
-			qg->new_refcnt++;
-		list_for_each_entry(glist, &qg->groups, next_group) {
-			ret = ulist_add(qgroups, glist->group->qgroupid,
-					ptr_to_u64(glist->group), GFP_ATOMIC);
-			if (ret < 0)
-				return ret;
-			ret = ulist_add(tmp, glist->group->qgroupid,
-					ptr_to_u64(glist->group), GFP_ATOMIC);
-			if (ret < 0)
-				return ret;
-		}
-	}
-	return 0;
-}
-
-/* Add refcnt for the newly added reference. */
-static int qgroup_calc_new_refcnt(struct btrfs_fs_info *fs_info,
-				  struct btrfs_qgroup_operation *oper,
-				  struct btrfs_qgroup *qgroup,
-				  struct ulist *tmp, struct ulist *qgroups,
-				  u64 seq)
-{
-	struct ulist_node *unode;
-	struct ulist_iterator uiter;
-	struct btrfs_qgroup *qg;
-	int ret;
-
-	ulist_reinit(tmp);
-	ret = ulist_add(qgroups, qgroup->qgroupid, ptr_to_u64(qgroup),
-			GFP_ATOMIC);
-	if (ret < 0)
-		return ret;
-	ret = ulist_add(tmp, qgroup->qgroupid, ptr_to_u64(qgroup),
-			GFP_ATOMIC);
-	if (ret < 0)
-		return ret;
-	ULIST_ITER_INIT(&uiter);
-	while ((unode = ulist_next(tmp, &uiter))) {
-		struct btrfs_qgroup_list *glist;
-
-		qg = u64_to_ptr(unode->aux);
-		if (oper->type == BTRFS_QGROUP_OPER_ADD_SHARED) {
-			if (qg->new_refcnt < seq)
-				qg->new_refcnt = seq + 1;
-			else
-				qg->new_refcnt++;
-		} else {
-			if (qg->old_refcnt < seq)
-				qg->old_refcnt = seq + 1;
-			else
-				qg->old_refcnt++;
-		}
-		list_for_each_entry(glist, &qg->groups, next_group) {
-			ret = ulist_add(tmp, glist->group->qgroupid,
-					ptr_to_u64(glist->group), GFP_ATOMIC);
-			if (ret < 0)
-				return ret;
-			ret = ulist_add(qgroups, glist->group->qgroupid,
-					ptr_to_u64(glist->group), GFP_ATOMIC);
-			if (ret < 0)
-				return ret;
-		}
-	}
-	return 0;
-}
-
-/*
- * This adjusts the counters for all referenced qgroups if need be.
+ * Update qgroup rfer/excl counters.
+ * Rfer update is easy, codes can explain themselves.
+ *
+ * Excl update is tricky, the update is split into 2 part.
+ * Part 1: Possible exclusive <-> sharing detect:
+ *	|	A	|	!A	|
+ *  -------------------------------------
+ *  B	|	*	|	-	|
+ *  -------------------------------------
+ *  !B	|	+	|	**	|
+ *  -------------------------------------
+ *
+ * Conditions:
+ * A:	cur_old_roots < nr_old_roots	(not exclusive before)
+ * !A:	cur_old_roots == nr_old_roots	(possible exclusive before)
+ * B:	cur_new_roots < nr_new_roots	(not exclusive now)
+ * !B:	cur_new_roots == nr_new_roots	(possible exclsuive now)
+ *
+ * Results:
+ * +: Possible sharing -> exclusive	-: Possible exclusive -> sharing
+ * *: Definitely not changed.		**: Possible unchanged.
+ *
+ * For !A and !B condition, the exception is cur_old/new_roots == 0 case.
+ *
+ * To make the logic clear, we first use condition A and B to split
+ * combination into 4 results.
+ *
+ * Then, for result "+" and "-", check old/new_roots == 0 case, as in them
+ * only on variant maybe 0.
+ *
+ * Lastly, check result **, since there are 2 variants maybe 0, split them
+ * again(2x2).
+ * But this time we don't need to consider other things, the codes and logic
+ * is easy to understand now.
  */
-static int qgroup_adjust_counters(struct btrfs_fs_info *fs_info,
-				  u64 root_to_skip, u64 num_bytes,
-				  struct ulist *qgroups, u64 seq,
-				  int old_roots, int new_roots, int rescan)
+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)
 {
 	struct ulist_node *unode;
 	struct ulist_iterator uiter;
@@ -1810,423 +1588,196 @@ static int qgroup_adjust_counters(struct btrfs_fs_info *fs_info,
 		bool dirty = false;
 
 		qg = u64_to_ptr(unode->aux);
-		/*
-		 * Wasn't referenced before but is now, add to the reference
-		 * counters.
-		 */
-		if (qg->old_refcnt <= seq && qg->new_refcnt > seq) {
+		cur_old_count = btrfs_qgroup_get_old_refcnt(qg, seq);
+		cur_new_count = btrfs_qgroup_get_new_refcnt(qg, seq);
+
+		/* Rfer update part */
+		if (cur_old_count == 0 && cur_new_count > 0) {
 			qg->rfer += num_bytes;
 			qg->rfer_cmpr += num_bytes;
 			dirty = true;
 		}
-
-		/*
-		 * Was referenced before but isn't now, subtract from the
-		 * reference counters.
-		 */
-		if (qg->old_refcnt > seq && qg->new_refcnt <= seq) {
+		if (cur_old_count > 0 && cur_new_count == 0) {
 			qg->rfer -= num_bytes;
 			qg->rfer_cmpr -= num_bytes;
 			dirty = true;
 		}
 
-		if (qg->old_refcnt < seq)
-			cur_old_count = 0;
-		else
-			cur_old_count = qg->old_refcnt - seq;
-		if (qg->new_refcnt < seq)
-			cur_new_count = 0;
-		else
-			cur_new_count = qg->new_refcnt - seq;
+		/* Excl update part */
+		/* Exclusive/none -> shared case */
+		if (cur_old_count == nr_old_roots &&
+		    cur_new_count < nr_new_roots) {
+			/* Exclusive -> shared */
+			if (cur_old_count != 0) {
+				qg->excl -= num_bytes;
+				qg->excl_cmpr -= num_bytes;
+				dirty = true;
+			}
+		}
 
-		/*
-		 * If our refcount was the same as the roots previously but our
-		 * new count isn't the same as the number of roots now then we
-		 * went from having a exclusive reference on this range to not.
-		 */
-		if (old_roots && cur_old_count == old_roots &&
-		    (cur_new_count != new_roots || new_roots == 0)) {
-			WARN_ON(cur_new_count != new_roots && new_roots == 0);
-			qg->excl -= num_bytes;
-			qg->excl_cmpr -= num_bytes;
-			dirty = true;
+		/* Shared -> exclusive/none case */
+		if (cur_old_count < nr_old_roots &&
+		    cur_new_count == nr_new_roots) {
+			/* Shared->exclusive */
+			if (cur_new_count != 0) {
+				qg->excl += num_bytes;
+				qg->excl_cmpr += num_bytes;
+				dirty = true;
+			}
 		}
 
-		/*
-		 * If we didn't reference all the roots before but now we do we
-		 * have an exclusive reference to this range.
-		 */
-		if ((!old_roots || (old_roots && cur_old_count != old_roots))
-		    && cur_new_count == new_roots) {
-			qg->excl += num_bytes;
-			qg->excl_cmpr += num_bytes;
-			dirty = true;
+		/* Exclusive/none -> exclusive/none case */
+		if (cur_old_count == nr_old_roots &&
+		    cur_new_count == nr_new_roots) {
+			if (cur_old_count == 0) {
+				/* None -> exclusive/none */
+
+				if (cur_new_count != 0) {
+					/* None -> exclusive */
+					qg->excl += num_bytes;
+					qg->excl_cmpr += num_bytes;
+					dirty = true;
+				}
+				/* None -> none, nothing changed */
+			} else {
+				/* Exclusive -> exclusive/none */
+
+				if (cur_new_count == 0) {
+					/* Exclusive -> none */
+					qg->excl -= num_bytes;
+					qg->excl_cmpr -= num_bytes;
+					dirty = true;
+				}
+				/* Exclusive -> exclusive, nothing changed */
+			}
 		}
 
+		/* For exclusive extent, free its reserved bytes too */
+		if (nr_old_roots == 0 && nr_new_roots == 1 &&
+		    cur_new_count == nr_new_roots)
+			qg->reserved -= num_bytes;
 		if (dirty)
 			qgroup_dirty(fs_info, qg);
 	}
 	return 0;
 }
 
-/*
- * If we removed a data extent and there were other references for that bytenr
- * then we need to lookup all referenced roots to make sure we still don't
- * reference this bytenr.  If we do then we can just discard this operation.
- */
-static int check_existing_refs(struct btrfs_trans_handle *trans,
-			       struct btrfs_fs_info *fs_info,
-			       struct btrfs_qgroup_operation *oper)
-{
-	struct ulist *roots = NULL;
-	struct ulist_node *unode;
-	struct ulist_iterator uiter;
-	int ret = 0;
-
-	ret = btrfs_find_all_roots(trans, fs_info, oper->bytenr,
-				   oper->elem.seq, &roots);
-	if (ret < 0)
-		return ret;
-	ret = 0;
-
-	ULIST_ITER_INIT(&uiter);
-	while ((unode = ulist_next(roots, &uiter))) {
-		if (unode->val == oper->ref_root) {
-			ret = 1;
-			break;
-		}
-	}
-	ulist_free(roots);
-	btrfs_put_tree_mod_seq(fs_info, &oper->elem);
-
-	return ret;
-}
-
-/*
- * If we share a reference across multiple roots then we may need to adjust
- * various qgroups referenced and exclusive counters.  The basic premise is this
- *
- * 1) We have seq to represent a 0 count.  Instead of looping through all of the
- * qgroups and resetting their refcount to 0 we just constantly bump this
- * sequence number to act as the base reference count.  This means that if
- * anybody is equal to or below this sequence they were never referenced.  We
- * jack this sequence up by the number of roots we found each time in order to
- * make sure we don't have any overlap.
- *
- * 2) We first search all the roots that reference the area _except_ the root
- * we're acting on currently.  This makes up the old_refcnt of all the qgroups
- * before.
- *
- * 3) We walk all of the qgroups referenced by the root we are currently acting
- * on, and will either adjust old_refcnt in the case of a removal or the
- * new_refcnt in the case of an addition.
- *
- * 4) Finally we walk all the qgroups that are referenced by this range
- * including the root we are acting on currently.  We will adjust the counters
- * based on the number of roots we had and will have after this operation.
- *
- * Take this example as an illustration
- *
- *			[qgroup 1/0]
- *		     /         |          \
- *		[qg 0/0]   [qg 0/1]	[qg 0/2]
- *		   \          |            /
- *		  [	   extent	    ]
- *
- * Say we are adding a reference that is covered by qg 0/0.  The first step
- * would give a refcnt of 1 to qg 0/1 and 0/2 and a refcnt of 2 to qg 1/0 with
- * old_roots being 2.  Because it is adding new_roots will be 1.  We then go
- * through qg 0/0 which will get the new_refcnt set to 1 and add 1 to qg 1/0's
- * new_refcnt, bringing it to 3.  We then walk through all of the qgroups, we
- * notice that the old refcnt for qg 0/0 < the new refcnt, so we added a
- * reference and thus must add the size to the referenced bytes.  Everything
- * else is the same so nothing else changes.
- */
-static int qgroup_shared_accounting(struct btrfs_trans_handle *trans,
-				    struct btrfs_fs_info *fs_info,
-				    struct btrfs_qgroup_operation *oper)
+int
+btrfs_qgroup_account_extent(struct btrfs_trans_handle *trans,
+			    struct btrfs_fs_info *fs_info,
+			    u64 bytenr, u64 num_bytes,
+			    struct ulist *old_roots, struct ulist *new_roots)
 {
-	struct ulist *roots = NULL;
-	struct ulist *qgroups, *tmp;
-	struct btrfs_qgroup *qgroup;
-	struct seq_list elem = SEQ_LIST_INIT(elem);
+	struct ulist *qgroups = NULL;
+	struct ulist *tmp = NULL;
 	u64 seq;
-	int old_roots = 0;
-	int new_roots = 0;
+	u64 nr_new_roots = 0;
+	u64 nr_old_roots = 0;
 	int ret = 0;
 
-	if (oper->elem.seq) {
-		ret = check_existing_refs(trans, fs_info, oper);
-		if (ret < 0)
-			return ret;
-		if (ret)
-			return 0;
-	}
+	if (new_roots)
+		nr_new_roots = new_roots->nnodes;
+	if (old_roots)
+		nr_old_roots = old_roots->nnodes;
 
-	qgroups = ulist_alloc(GFP_NOFS);
-	if (!qgroups)
-		return -ENOMEM;
+	if (!fs_info->quota_enabled)
+		goto out_free;
+	BUG_ON(!fs_info->quota_root);
 
+	qgroups = ulist_alloc(GFP_NOFS);
+	if (!qgroups) {
+		ret = -ENOMEM;
+		goto out_free;
+	}
 	tmp = ulist_alloc(GFP_NOFS);
 	if (!tmp) {
-		ulist_free(qgroups);
-		return -ENOMEM;
+		ret = -ENOMEM;
+		goto out_free;
 	}
 
-	btrfs_get_tree_mod_seq(fs_info, &elem);
-	ret = btrfs_find_all_roots(trans, fs_info, oper->bytenr, elem.seq,
-				   &roots);
-	btrfs_put_tree_mod_seq(fs_info, &elem);
-	if (ret < 0) {
-		ulist_free(qgroups);
-		ulist_free(tmp);
-		return ret;
+	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) {
+			mutex_unlock(&fs_info->qgroup_rescan_lock);
+			ret = 0;
+			goto out_free;
+		}
 	}
+	mutex_unlock(&fs_info->qgroup_rescan_lock);
+
 	spin_lock(&fs_info->qgroup_lock);
-	qgroup = find_qgroup_rb(fs_info, oper->ref_root);
-	if (!qgroup)
-		goto out;
 	seq = fs_info->qgroup_seq;
 
-	/*
-	 * So roots is the list of all the roots currently pointing at the
-	 * bytenr, including the ref we are adding if we are adding, or not if
-	 * we are removing a ref.  So we pass in the ref_root to skip that root
-	 * in our calculations.  We set old_refnct and new_refcnt cause who the
-	 * hell knows what everything looked like before, and it doesn't matter
-	 * except...
-	 */
-	ret = qgroup_calc_old_refcnt(fs_info, oper->ref_root, tmp, roots, qgroups,
-				     seq, &old_roots, 0);
-	if (ret < 0)
-		goto out;
-
-	/*
-	 * Now adjust the refcounts of the qgroups that care about this
-	 * reference, either the old_count in the case of removal or new_count
-	 * in the case of an addition.
-	 */
-	ret = qgroup_calc_new_refcnt(fs_info, oper, qgroup, tmp, qgroups,
-				     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;
 
-	/*
-	 * ...in the case of removals.  If we had a removal before we got around
-	 * to processing this operation then we need to find that guy and count
-	 * his references as if they really existed so we don't end up screwing
-	 * up the exclusive counts.  Then whenever we go to process the delete
-	 * everything will be grand and we can account for whatever exclusive
-	 * changes need to be made there.  We also have to pass in old_roots so
-	 * we have an accurate count of the roots as it pertains to this
-	 * operations view of the world.
-	 */
-	ret = qgroup_account_deleted_refs(fs_info, oper, tmp, qgroups, seq,
-					  &old_roots);
+	/* Update new refcnts using new_roots */
+	ret = qgroup_update_refcnt(fs_info, new_roots, tmp, qgroups, seq,
+				   UPDATE_NEW);
 	if (ret < 0)
 		goto out;
 
-	/*
-	 * We are adding our root, need to adjust up the number of roots,
-	 * otherwise old_roots is the number of roots we want.
-	 */
-	if (oper->type == BTRFS_QGROUP_OPER_ADD_SHARED) {
-		new_roots = old_roots + 1;
-	} else {
-		new_roots = old_roots;
-		old_roots++;
-	}
-	fs_info->qgroup_seq += old_roots + 1;
-
+	qgroup_update_counters(fs_info, qgroups, nr_old_roots, nr_new_roots,
+			       num_bytes, seq);
 
 	/*
-	 * And now the magic happens, bless Arne for having a pretty elegant
-	 * solution for this.
+	 * Bump qgroup_seq to avoid seq overlap
 	 */
-	qgroup_adjust_counters(fs_info, oper->ref_root, oper->num_bytes,
-			       qgroups, seq, old_roots, new_roots, 0);
+	fs_info->qgroup_seq += max(nr_old_roots, nr_new_roots) + 1;
 out:
 	spin_unlock(&fs_info->qgroup_lock);
-	ulist_free(qgroups);
-	ulist_free(roots);
+out_free:
 	ulist_free(tmp);
+	ulist_free(qgroups);
+	ulist_free(old_roots);
+	ulist_free(new_roots);
 	return ret;
 }
 
-/*
- * Process a reference to a shared subtree. This type of operation is
- * queued during snapshot removal when we encounter extents which are
- * shared between more than one root.
- */
-static int qgroup_subtree_accounting(struct btrfs_trans_handle *trans,
-				     struct btrfs_fs_info *fs_info,
-				     struct btrfs_qgroup_operation *oper)
-{
-	struct ulist *roots = NULL;
-	struct ulist_node *unode;
-	struct ulist_iterator uiter;
-	struct btrfs_qgroup_list *glist;
-	struct ulist *parents;
-	int ret = 0;
-	int err;
-	struct btrfs_qgroup *qg;
-	u64 root_obj = 0;
-	struct seq_list elem = SEQ_LIST_INIT(elem);
-
-	parents = ulist_alloc(GFP_NOFS);
-	if (!parents)
-		return -ENOMEM;
-
-	btrfs_get_tree_mod_seq(fs_info, &elem);
-	ret = btrfs_find_all_roots(trans, fs_info, oper->bytenr,
-				   elem.seq, &roots);
-	btrfs_put_tree_mod_seq(fs_info, &elem);
-	if (ret < 0)
-		goto out;
-
-	if (roots->nnodes != 1)
-		goto out;
-
-	ULIST_ITER_INIT(&uiter);
-	unode = ulist_next(roots, &uiter); /* Only want 1 so no need to loop */
-	/*
-	 * If we find our ref root then that means all refs
-	 * this extent has to the root have not yet been
-	 * deleted. In that case, we do nothing and let the
-	 * last ref for this bytenr drive our update.
-	 *
-	 * This can happen for example if an extent is
-	 * referenced multiple times in a snapshot (clone,
-	 * etc). If we are in the middle of snapshot removal,
-	 * queued updates for such an extent will find the
-	 * root if we have not yet finished removing the
-	 * snapshot.
-	 */
-	if (unode->val == oper->ref_root)
-		goto out;
-
-	root_obj = unode->val;
-	BUG_ON(!root_obj);
-
-	spin_lock(&fs_info->qgroup_lock);
-	qg = find_qgroup_rb(fs_info, root_obj);
-	if (!qg)
-		goto out_unlock;
-
-	qg->excl += oper->num_bytes;
-	qg->excl_cmpr += oper->num_bytes;
-	qgroup_dirty(fs_info, qg);
-
-	/*
-	 * Adjust counts for parent groups. First we find all
-	 * parents, then in the 2nd loop we do the adjustment
-	 * while adding parents of the parents to our ulist.
-	 */
-	list_for_each_entry(glist, &qg->groups, next_group) {
-		err = ulist_add(parents, glist->group->qgroupid,
-				ptr_to_u64(glist->group), GFP_ATOMIC);
-		if (err < 0) {
-			ret = err;
-			goto out_unlock;
-		}
-	}
-
-	ULIST_ITER_INIT(&uiter);
-	while ((unode = ulist_next(parents, &uiter))) {
-		qg = u64_to_ptr(unode->aux);
-		qg->excl += oper->num_bytes;
-		qg->excl_cmpr += oper->num_bytes;
-		qgroup_dirty(fs_info, qg);
-
-		/* Add any parents of the parents */
-		list_for_each_entry(glist, &qg->groups, next_group) {
-			err = ulist_add(parents, glist->group->qgroupid,
-					ptr_to_u64(glist->group), GFP_ATOMIC);
-			if (err < 0) {
-				ret = err;
-				goto out_unlock;
-			}
-		}
-	}
-
-out_unlock:
-	spin_unlock(&fs_info->qgroup_lock);
-
-out:
-	ulist_free(roots);
-	ulist_free(parents);
-	return ret;
-}
-
-/*
- * btrfs_qgroup_account_ref is called for every ref that is added to or deleted
- * from the fs. First, all roots referencing the extent are searched, and
- * then the space is accounted accordingly to the different roots. The
- * accounting algorithm works in 3 steps documented inline.
- */
-static int btrfs_qgroup_account(struct btrfs_trans_handle *trans,
-				struct btrfs_fs_info *fs_info,
-				struct btrfs_qgroup_operation *oper)
+int btrfs_qgroup_account_extents(struct btrfs_trans_handle *trans,
+				 struct btrfs_fs_info *fs_info)
 {
+	struct btrfs_qgroup_extent_record *record;
+	struct btrfs_delayed_ref_root *delayed_refs;
+	struct ulist *new_roots = NULL;
+	struct rb_node *node;
+	u64 qgroup_to_skip;
 	int ret = 0;
 
-	if (!fs_info->quota_enabled)
-		return 0;
-
-	BUG_ON(!fs_info->quota_root);
+	delayed_refs = &trans->transaction->delayed_refs;
+	qgroup_to_skip = delayed_refs->qgroup_to_skip;
+	while ((node = rb_first(&delayed_refs->dirty_extent_root))) {
+		record = rb_entry(node, struct btrfs_qgroup_extent_record,
+				  node);
 
-	mutex_lock(&fs_info->qgroup_rescan_lock);
-	if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) {
-		if (fs_info->qgroup_rescan_progress.objectid <= oper->bytenr) {
-			mutex_unlock(&fs_info->qgroup_rescan_lock);
-			return 0;
+		if (!ret) {
+			/*
+			 * Use (u64)-1 as time_seq to do special search, which
+			 * doesn't lock tree or delayed_refs and search current
+			 * root. It's safe inside commit_transaction().
+			 */
+			ret = btrfs_find_all_roots(trans, fs_info,
+					record->bytenr, (u64)-1, &new_roots);
+			if (ret < 0)
+				goto cleanup;
+			if (qgroup_to_skip)
+				ulist_del(new_roots, qgroup_to_skip, 0);
+			ret = btrfs_qgroup_account_extent(trans, fs_info,
+					record->bytenr, record->num_bytes,
+					record->old_roots, new_roots);
+			record->old_roots = NULL;
+			new_roots = NULL;
 		}
-	}
-	mutex_unlock(&fs_info->qgroup_rescan_lock);
-
-	ASSERT(is_fstree(oper->ref_root));
-
-	trace_btrfs_qgroup_account(oper);
-
-	switch (oper->type) {
-	case BTRFS_QGROUP_OPER_ADD_EXCL:
-	case BTRFS_QGROUP_OPER_SUB_EXCL:
-		ret = qgroup_excl_accounting(fs_info, oper);
-		break;
-	case BTRFS_QGROUP_OPER_ADD_SHARED:
-	case BTRFS_QGROUP_OPER_SUB_SHARED:
-		ret = qgroup_shared_accounting(trans, fs_info, oper);
-		break;
-	case BTRFS_QGROUP_OPER_SUB_SUBTREE:
-		ret = qgroup_subtree_accounting(trans, fs_info, oper);
-		break;
-	default:
-		ASSERT(0);
-	}
-	return ret;
-}
-
-/*
- * Needs to be called everytime we run delayed refs, even if there is an error
- * in order to cleanup outstanding operations.
- */
-int btrfs_delayed_qgroup_accounting(struct btrfs_trans_handle *trans,
-				    struct btrfs_fs_info *fs_info)
-{
-	struct btrfs_qgroup_operation *oper;
-	int ret = 0;
+cleanup:
+		ulist_free(record->old_roots);
+		ulist_free(new_roots);
+		new_roots = NULL;
+		rb_erase(node, &delayed_refs->dirty_extent_root);
+		kfree(record);
 
-	while (!list_empty(&trans->qgroup_ref_list)) {
-		oper = list_first_entry(&trans->qgroup_ref_list,
-					struct btrfs_qgroup_operation, list);
-		list_del_init(&oper->list);
-		if (!ret || !trans->aborted)
-			ret = btrfs_qgroup_account(trans, fs_info, oper);
-		spin_lock(&fs_info->qgroup_op_lock);
-		rb_erase(&oper->n, &fs_info->qgroup_op_tree);
-		spin_unlock(&fs_info->qgroup_op_lock);
-		btrfs_put_tree_mod_seq(fs_info, &oper->elem);
-		kfree(oper);
 	}
 	return ret;
 }
@@ -2637,15 +2188,13 @@ void assert_qgroups_uptodate(struct btrfs_trans_handle *trans)
  */
 static int
 qgroup_rescan_leaf(struct btrfs_fs_info *fs_info, struct btrfs_path *path,
-		   struct btrfs_trans_handle *trans, struct ulist *qgroups,
-		   struct ulist *tmp, struct extent_buffer *scratch_leaf)
+		   struct btrfs_trans_handle *trans,
+		   struct extent_buffer *scratch_leaf)
 {
 	struct btrfs_key found;
 	struct ulist *roots = NULL;
 	struct seq_list tree_mod_seq_elem = SEQ_LIST_INIT(tree_mod_seq_elem);
 	u64 num_bytes;
-	u64 seq;
-	int new_roots;
 	int slot;
 	int ret;
 
@@ -2695,33 +2244,15 @@ qgroup_rescan_leaf(struct btrfs_fs_info *fs_info, struct btrfs_path *path,
 		else
 			num_bytes = found.offset;
 
-		ulist_reinit(qgroups);
 		ret = btrfs_find_all_roots(NULL, fs_info, found.objectid, 0,
 					   &roots);
 		if (ret < 0)
 			goto out;
-		spin_lock(&fs_info->qgroup_lock);
-		seq = fs_info->qgroup_seq;
-		fs_info->qgroup_seq += roots->nnodes + 1; /* max refcnt */
-
-		new_roots = 0;
-		ret = qgroup_calc_old_refcnt(fs_info, 0, tmp, roots, qgroups,
-					     seq, &new_roots, 1);
-		if (ret < 0) {
-			spin_unlock(&fs_info->qgroup_lock);
-			ulist_free(roots);
-			goto out;
-		}
-
-		ret = qgroup_adjust_counters(fs_info, 0, num_bytes, qgroups,
-					     seq, 0, new_roots, 1);
-		if (ret < 0) {
-			spin_unlock(&fs_info->qgroup_lock);
-			ulist_free(roots);
+		/* For rescan, just pass old_roots as NULL */
+		ret = btrfs_qgroup_account_extent(trans, fs_info,
+				found.objectid, num_bytes, NULL, roots);
+		if (ret < 0)
 			goto out;
-		}
-		spin_unlock(&fs_info->qgroup_lock);
-		ulist_free(roots);
 	}
 out:
 	btrfs_put_tree_mod_seq(fs_info, &tree_mod_seq_elem);
@@ -2735,7 +2266,6 @@ static void btrfs_qgroup_rescan_worker(struct btrfs_work *work)
 						     qgroup_rescan_work);
 	struct btrfs_path *path;
 	struct btrfs_trans_handle *trans = NULL;
-	struct ulist *tmp = NULL, *qgroups = NULL;
 	struct extent_buffer *scratch_leaf = NULL;
 	int err = -ENOMEM;
 	int ret = 0;
@@ -2743,12 +2273,6 @@ static void btrfs_qgroup_rescan_worker(struct btrfs_work *work)
 	path = btrfs_alloc_path();
 	if (!path)
 		goto out;
-	qgroups = ulist_alloc(GFP_NOFS);
-	if (!qgroups)
-		goto out;
-	tmp = ulist_alloc(GFP_NOFS);
-	if (!tmp)
-		goto out;
 	scratch_leaf = kmalloc(sizeof(*scratch_leaf), GFP_NOFS);
 	if (!scratch_leaf)
 		goto out;
@@ -2764,7 +2288,7 @@ static void btrfs_qgroup_rescan_worker(struct btrfs_work *work)
 			err = -EINTR;
 		} else {
 			err = qgroup_rescan_leaf(fs_info, path, trans,
-						 qgroups, tmp, scratch_leaf);
+						 scratch_leaf);
 		}
 		if (err > 0)
 			btrfs_commit_transaction(trans, fs_info->fs_root);
@@ -2774,8 +2298,6 @@ static void btrfs_qgroup_rescan_worker(struct btrfs_work *work)
 
 out:
 	kfree(scratch_leaf);
-	ulist_free(qgroups);
-	ulist_free(tmp);
 	btrfs_free_path(path);
 
 	mutex_lock(&fs_info->qgroup_rescan_lock);
diff --git a/fs/btrfs/qgroup.h b/fs/btrfs/qgroup.h
index c5242aa9a..6387dcfa3 100644
--- a/fs/btrfs/qgroup.h
+++ b/fs/btrfs/qgroup.h
@@ -19,43 +19,18 @@
 #ifndef __BTRFS_QGROUP__
 #define __BTRFS_QGROUP__
 
+#include "ulist.h"
+#include "delayed-ref.h"
+
 /*
- * A description of the operations, all of these operations only happen when we
- * are adding the 1st reference for that subvolume in the case of adding space
- * or on the last reference delete in the case of subtraction.  The only
- * exception is the last one, which is added for confusion.
- *
- * BTRFS_QGROUP_OPER_ADD_EXCL: adding bytes where this subvolume is the only
- * one pointing at the bytes we are adding.  This is called on the first
- * allocation.
- *
- * BTRFS_QGROUP_OPER_ADD_SHARED: adding bytes where this bytenr is going to be
- * shared between subvols.  This is called on the creation of a ref that already
- * has refs from a different subvolume, so basically reflink.
- *
- * BTRFS_QGROUP_OPER_SUB_EXCL: removing bytes where this subvolume is the only
- * one referencing the range.
- *
- * BTRFS_QGROUP_OPER_SUB_SHARED: removing bytes where this subvolume shares with
- * refs with other subvolumes.
+ * Record a dirty extent, and info qgroup to update quota on it
+ * TODO: Use kmem cache to alloc it.
  */
-enum btrfs_qgroup_operation_type {
-	BTRFS_QGROUP_OPER_ADD_EXCL,
-	BTRFS_QGROUP_OPER_ADD_SHARED,
-	BTRFS_QGROUP_OPER_SUB_EXCL,
-	BTRFS_QGROUP_OPER_SUB_SHARED,
-	BTRFS_QGROUP_OPER_SUB_SUBTREE,
-};
-
-struct btrfs_qgroup_operation {
-	u64 ref_root;
+struct btrfs_qgroup_extent_record {
+	struct rb_node node;
 	u64 bytenr;
 	u64 num_bytes;
-	u64 seq;
-	enum btrfs_qgroup_operation_type type;
-	struct seq_list elem;
-	struct rb_node n;
-	struct list_head list;
+	struct ulist *old_roots;
 };
 
 int btrfs_quota_enable(struct btrfs_trans_handle *trans,
@@ -79,16 +54,18 @@ int btrfs_limit_qgroup(struct btrfs_trans_handle *trans,
 int btrfs_read_qgroup_config(struct btrfs_fs_info *fs_info);
 void btrfs_free_qgroup_config(struct btrfs_fs_info *fs_info);
 struct btrfs_delayed_extent_op;
-int btrfs_qgroup_record_ref(struct btrfs_trans_handle *trans,
-			    struct btrfs_fs_info *fs_info, u64 ref_root,
+int btrfs_qgroup_prepare_account_extents(struct btrfs_trans_handle *trans,
+					 struct btrfs_fs_info *fs_info);
+struct btrfs_qgroup_extent_record
+*btrfs_qgroup_insert_dirty_extent(struct btrfs_delayed_ref_root *delayed_refs,
+				  struct btrfs_qgroup_extent_record *record);
+int
+btrfs_qgroup_account_extent(struct btrfs_trans_handle *trans,
+			    struct btrfs_fs_info *fs_info,
 			    u64 bytenr, u64 num_bytes,
-			    enum btrfs_qgroup_operation_type type,
-			    int mod_seq);
-int btrfs_delayed_qgroup_accounting(struct btrfs_trans_handle *trans,
-				    struct btrfs_fs_info *fs_info);
-void btrfs_remove_qgroup_operation(struct btrfs_trans_handle *trans,
-				   struct btrfs_fs_info *fs_info,
-				   struct btrfs_qgroup_operation *oper);
+			    struct ulist *old_roots, struct ulist *new_roots);
+int btrfs_qgroup_account_extents(struct btrfs_trans_handle *trans,
+				 struct btrfs_fs_info *fs_info);
 int btrfs_run_qgroups(struct btrfs_trans_handle *trans,
 		      struct btrfs_fs_info *fs_info);
 int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans,
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index 74b24b01d..88cbb5995 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -1847,8 +1847,10 @@ again:
 			}
 
 			eb = read_tree_block(dest, old_bytenr, old_ptr_gen);
-			if (!eb || !extent_buffer_uptodate(eb)) {
-				ret = (!eb) ? -ENOMEM : -EIO;
+			if (IS_ERR(eb)) {
+				ret = PTR_ERR(eb);
+			} else if (!extent_buffer_uptodate(eb)) {
+				ret = -EIO;
 				free_extent_buffer(eb);
 				break;
 			}
@@ -2002,7 +2004,9 @@ int walk_down_reloc_tree(struct btrfs_root *root, struct btrfs_path *path,
 
 		bytenr = btrfs_node_blockptr(eb, path->slots[i]);
 		eb = read_tree_block(root, bytenr, ptr_gen);
-		if (!eb || !extent_buffer_uptodate(eb)) {
+		if (IS_ERR(eb)) {
+			return PTR_ERR(eb);
+		} else if (!extent_buffer_uptodate(eb)) {
 			free_extent_buffer(eb);
 			return -EIO;
 		}
@@ -2710,7 +2714,10 @@ static int do_relocation(struct btrfs_trans_handle *trans,
 		blocksize = root->nodesize;
 		generation = btrfs_node_ptr_generation(upper->eb, slot);
 		eb = read_tree_block(root, bytenr, generation);
-		if (!eb || !extent_buffer_uptodate(eb)) {
+		if (IS_ERR(eb)) {
+			err = PTR_ERR(eb);
+			goto next;
+		} else if (!extent_buffer_uptodate(eb)) {
 			free_extent_buffer(eb);
 			err = -EIO;
 			goto next;
@@ -2873,7 +2880,9 @@ static int get_tree_block_key(struct reloc_control *rc,
 	BUG_ON(block->key_ready);
 	eb = read_tree_block(rc->extent_root, block->bytenr,
 			     block->key.offset);
-	if (!eb || !extent_buffer_uptodate(eb)) {
+	if (IS_ERR(eb)) {
+		return PTR_ERR(eb);
+	} else if (!extent_buffer_uptodate(eb)) {
 		free_extent_buffer(eb);
 		return -EIO;
 	}
@@ -4040,7 +4049,7 @@ restart:
 	if (trans && progress && err == -ENOSPC) {
 		ret = btrfs_force_chunk_alloc(trans, rc->extent_root,
 					      rc->block_group->flags);
-		if (ret == 0) {
+		if (ret == 1) {
 			err = 0;
 			progress = 0;
 			goto restart;
diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c
index ab5811545..94db0fa52 100644
--- a/fs/btrfs/scrub.c
+++ b/fs/btrfs/scrub.c
@@ -2662,18 +2662,30 @@ static void scrub_free_parity(struct scrub_parity *sparity)
 	kfree(sparity);
 }
 
+static void scrub_parity_bio_endio_worker(struct btrfs_work *work)
+{
+	struct scrub_parity *sparity = container_of(work, struct scrub_parity,
+						    work);
+	struct scrub_ctx *sctx = sparity->sctx;
+
+	scrub_free_parity(sparity);
+	scrub_pending_bio_dec(sctx);
+}
+
 static void scrub_parity_bio_endio(struct bio *bio, int error)
 {
 	struct scrub_parity *sparity = (struct scrub_parity *)bio->bi_private;
-	struct scrub_ctx *sctx = sparity->sctx;
 
 	if (error)
 		bitmap_or(sparity->ebitmap, sparity->ebitmap, sparity->dbitmap,
 			  sparity->nsectors);
 
-	scrub_free_parity(sparity);
-	scrub_pending_bio_dec(sctx);
 	bio_put(bio);
+
+	btrfs_init_work(&sparity->work, btrfs_scrubparity_helper,
+			scrub_parity_bio_endio_worker, NULL, NULL);
+	btrfs_queue_work(sparity->sctx->dev_root->fs_info->scrub_parity_workers,
+			 &sparity->work);
 }
 
 static void scrub_parity_check_and_repair(struct scrub_parity *sparity)
@@ -3559,7 +3571,6 @@ static noinline_for_stack int scrub_supers(struct scrub_ctx *sctx,
 static noinline_for_stack int scrub_workers_get(struct btrfs_fs_info *fs_info,
 						int is_dev_replace)
 {
-	int ret = 0;
 	unsigned int flags = WQ_FREEZABLE | WQ_UNBOUND;
 	int max_active = fs_info->thread_pool_size;
 
@@ -3572,27 +3583,36 @@ static noinline_for_stack int scrub_workers_get(struct btrfs_fs_info *fs_info,
 			fs_info->scrub_workers =
 				btrfs_alloc_workqueue("btrfs-scrub", flags,
 						      max_active, 4);
-		if (!fs_info->scrub_workers) {
-			ret = -ENOMEM;
-			goto out;
-		}
+		if (!fs_info->scrub_workers)
+			goto fail_scrub_workers;
+
 		fs_info->scrub_wr_completion_workers =
 			btrfs_alloc_workqueue("btrfs-scrubwrc", flags,
 					      max_active, 2);
-		if (!fs_info->scrub_wr_completion_workers) {
-			ret = -ENOMEM;
-			goto out;
-		}
+		if (!fs_info->scrub_wr_completion_workers)
+			goto fail_scrub_wr_completion_workers;
+
 		fs_info->scrub_nocow_workers =
 			btrfs_alloc_workqueue("btrfs-scrubnc", flags, 1, 0);
-		if (!fs_info->scrub_nocow_workers) {
-			ret = -ENOMEM;
-			goto out;
-		}
+		if (!fs_info->scrub_nocow_workers)
+			goto fail_scrub_nocow_workers;
+		fs_info->scrub_parity_workers =
+			btrfs_alloc_workqueue("btrfs-scrubparity", flags,
+					      max_active, 2);
+		if (!fs_info->scrub_parity_workers)
+			goto fail_scrub_parity_workers;
 	}
 	++fs_info->scrub_workers_refcnt;
-out:
-	return ret;
+	return 0;
+
+fail_scrub_parity_workers:
+	btrfs_destroy_workqueue(fs_info->scrub_nocow_workers);
+fail_scrub_nocow_workers:
+	btrfs_destroy_workqueue(fs_info->scrub_wr_completion_workers);
+fail_scrub_wr_completion_workers:
+	btrfs_destroy_workqueue(fs_info->scrub_workers);
+fail_scrub_workers:
+	return -ENOMEM;
 }
 
 static noinline_for_stack void scrub_workers_put(struct btrfs_fs_info *fs_info)
@@ -3601,6 +3621,7 @@ static noinline_for_stack void scrub_workers_put(struct btrfs_fs_info *fs_info)
 		btrfs_destroy_workqueue(fs_info->scrub_workers);
 		btrfs_destroy_workqueue(fs_info->scrub_wr_completion_workers);
 		btrfs_destroy_workqueue(fs_info->scrub_nocow_workers);
+		btrfs_destroy_workqueue(fs_info->scrub_parity_workers);
 	}
 	WARN_ON(fs_info->scrub_workers_refcnt < 0);
 }
diff --git a/fs/btrfs/send.c b/fs/btrfs/send.c
index 5cf7838fb..aa72bfd28 100644
--- a/fs/btrfs/send.c
+++ b/fs/btrfs/send.c
@@ -243,6 +243,7 @@ struct waiting_dir_move {
 	 * after this directory is moved, we can try to rmdir the ino rmdir_ino.
 	 */
 	u64 rmdir_ino;
+	bool orphanized;
 };
 
 struct orphan_dir_info {
@@ -1916,8 +1917,13 @@ static int did_overwrite_ref(struct send_ctx *sctx,
 		goto out;
 	}
 
-	/* we know that it is or will be overwritten. check this now */
-	if (ow_inode < sctx->send_progress)
+	/*
+	 * We know that it is or will be overwritten. Check this now.
+	 * The current inode being processed might have been the one that caused
+	 * inode 'ino' to be orphanized, therefore ow_inode can actually be the
+	 * same as sctx->send_progress.
+	 */
+	if (ow_inode <= sctx->send_progress)
 		ret = 1;
 	else
 		ret = 0;
@@ -2239,6 +2245,8 @@ static int get_cur_path(struct send_ctx *sctx, u64 ino, u64 gen,
 	fs_path_reset(dest);
 
 	while (!stop && ino != BTRFS_FIRST_FREE_OBJECTID) {
+		struct waiting_dir_move *wdm;
+
 		fs_path_reset(name);
 
 		if (is_waiting_for_rm(sctx, ino)) {
@@ -2249,7 +2257,11 @@ static int get_cur_path(struct send_ctx *sctx, u64 ino, u64 gen,
 			break;
 		}
 
-		if (is_waiting_for_move(sctx, ino)) {
+		wdm = get_waiting_dir_move(sctx, ino);
+		if (wdm && wdm->orphanized) {
+			ret = gen_unique_name(sctx, ino, gen, name);
+			stop = 1;
+		} else if (wdm) {
 			ret = get_first_ref(sctx->parent_root, ino,
 					    &parent_inode, &parent_gen, name);
 		} else {
@@ -2344,8 +2356,12 @@ static int send_subvol_begin(struct send_ctx *sctx)
 	TLV_PUT_U64(sctx, BTRFS_SEND_A_CTRANSID,
 		    le64_to_cpu(sctx->send_root->root_item.ctransid));
 	if (parent_root) {
-		TLV_PUT_UUID(sctx, BTRFS_SEND_A_CLONE_UUID,
-				sctx->parent_root->root_item.uuid);
+		if (!btrfs_is_empty_uuid(parent_root->root_item.received_uuid))
+			TLV_PUT_UUID(sctx, BTRFS_SEND_A_CLONE_UUID,
+				     parent_root->root_item.received_uuid);
+		else
+			TLV_PUT_UUID(sctx, BTRFS_SEND_A_CLONE_UUID,
+				     parent_root->root_item.uuid);
 		TLV_PUT_U64(sctx, BTRFS_SEND_A_CLONE_CTRANSID,
 			    le64_to_cpu(sctx->parent_root->root_item.ctransid));
 	}
@@ -2939,7 +2955,7 @@ static int is_waiting_for_move(struct send_ctx *sctx, u64 ino)
 	return entry != NULL;
 }
 
-static int add_waiting_dir_move(struct send_ctx *sctx, u64 ino)
+static int add_waiting_dir_move(struct send_ctx *sctx, u64 ino, bool orphanized)
 {
 	struct rb_node **p = &sctx->waiting_dir_moves.rb_node;
 	struct rb_node *parent = NULL;
@@ -2950,6 +2966,7 @@ static int add_waiting_dir_move(struct send_ctx *sctx, u64 ino)
 		return -ENOMEM;
 	dm->ino = ino;
 	dm->rmdir_ino = 0;
+	dm->orphanized = orphanized;
 
 	while (*p) {
 		parent = *p;
@@ -3046,7 +3063,7 @@ static int add_pending_dir_move(struct send_ctx *sctx,
 			goto out;
 	}
 
-	ret = add_waiting_dir_move(sctx, pm->ino);
+	ret = add_waiting_dir_move(sctx, pm->ino, is_orphan);
 	if (ret)
 		goto out;
 
@@ -3369,8 +3386,40 @@ out:
 	return ret;
 }
 
+/*
+ * Check if ino ino1 is an ancestor of inode ino2 in the given root.
+ * Return 1 if true, 0 if false and < 0 on error.
+ */
+static int is_ancestor(struct btrfs_root *root,
+		       const u64 ino1,
+		       const u64 ino1_gen,
+		       const u64 ino2,
+		       struct fs_path *fs_path)
+{
+	u64 ino = ino2;
+
+	while (ino > BTRFS_FIRST_FREE_OBJECTID) {
+		int ret;
+		u64 parent;
+		u64 parent_gen;
+
+		fs_path_reset(fs_path);
+		ret = get_first_ref(root, ino, &parent, &parent_gen, fs_path);
+		if (ret < 0) {
+			if (ret == -ENOENT && ino == ino2)
+				ret = 0;
+			return ret;
+		}
+		if (parent == ino1)
+			return parent_gen == ino1_gen ? 1 : 0;
+		ino = parent;
+	}
+	return 0;
+}
+
 static int wait_for_parent_move(struct send_ctx *sctx,
-				struct recorded_ref *parent_ref)
+				struct recorded_ref *parent_ref,
+				const bool is_orphan)
 {
 	int ret = 0;
 	u64 ino = parent_ref->dir;
@@ -3390,11 +3439,24 @@ static int wait_for_parent_move(struct send_ctx *sctx,
 	 * Our current directory inode may not yet be renamed/moved because some
 	 * ancestor (immediate or not) has to be renamed/moved first. So find if
 	 * such ancestor exists and make sure our own rename/move happens after
-	 * that ancestor is processed.
+	 * that ancestor is processed to avoid path build infinite loops (done
+	 * at get_cur_path()).
 	 */
 	while (ino > BTRFS_FIRST_FREE_OBJECTID) {
 		if (is_waiting_for_move(sctx, ino)) {
-			ret = 1;
+			/*
+			 * If the current inode is an ancestor of ino in the
+			 * parent root, we need to delay the rename of the
+			 * current inode, otherwise don't delayed the rename
+			 * because we can end up with a circular dependency
+			 * of renames, resulting in some directories never
+			 * getting the respective rename operations issued in
+			 * the send stream or getting into infinite path build
+			 * loops.
+			 */
+			ret = is_ancestor(sctx->parent_root,
+					  sctx->cur_ino, sctx->cur_inode_gen,
+					  ino, path_before);
 			break;
 		}
 
@@ -3436,7 +3498,7 @@ out:
 					   ino,
 					   &sctx->new_refs,
 					   &sctx->deleted_refs,
-					   false);
+					   is_orphan);
 		if (!ret)
 			ret = 1;
 	}
@@ -3605,6 +3667,17 @@ verbose_printk("btrfs: process_recorded_refs %llu\n", sctx->cur_ino);
 			}
 		}
 
+		if (S_ISDIR(sctx->cur_inode_mode) && sctx->parent_root &&
+		    can_rename) {
+			ret = wait_for_parent_move(sctx, cur, is_orphan);
+			if (ret < 0)
+				goto out;
+			if (ret == 1) {
+				can_rename = false;
+				*pending_move = 1;
+			}
+		}
+
 		/*
 		 * link/move the ref to the new place. If we have an orphan
 		 * inode, move it and update valid_path. If not, link or move
@@ -3625,18 +3698,11 @@ verbose_printk("btrfs: process_recorded_refs %llu\n", sctx->cur_ino);
 				 * dirs, we always have one new and one deleted
 				 * ref. The deleted ref is ignored later.
 				 */
-				ret = wait_for_parent_move(sctx, cur);
-				if (ret < 0)
-					goto out;
-				if (ret) {
-					*pending_move = 1;
-				} else {
-					ret = send_rename(sctx, valid_path,
-							  cur->full_path);
-					if (!ret)
-						ret = fs_path_copy(valid_path,
-							       cur->full_path);
-				}
+				ret = send_rename(sctx, valid_path,
+						  cur->full_path);
+				if (!ret)
+					ret = fs_path_copy(valid_path,
+							   cur->full_path);
 				if (ret < 0)
 					goto out;
 			} else {
@@ -4524,8 +4590,21 @@ verbose_printk("btrfs: send_clone offset=%llu, len=%d, clone_root=%llu, "
 	if (ret < 0)
 		goto out;
 
-	TLV_PUT_UUID(sctx, BTRFS_SEND_A_CLONE_UUID,
-			clone_root->root->root_item.uuid);
+	/*
+	 * If the parent we're using has a received_uuid set then use that as
+	 * our clone source as that is what we will look for when doing a
+	 * receive.
+	 *
+	 * This covers the case that we create a snapshot off of a received
+	 * subvolume and then use that as the parent and try to receive on a
+	 * different host.
+	 */
+	if (!btrfs_is_empty_uuid(clone_root->root->root_item.received_uuid))
+		TLV_PUT_UUID(sctx, BTRFS_SEND_A_CLONE_UUID,
+			     clone_root->root->root_item.received_uuid);
+	else
+		TLV_PUT_UUID(sctx, BTRFS_SEND_A_CLONE_UUID,
+			     clone_root->root->root_item.uuid);
 	TLV_PUT_U64(sctx, BTRFS_SEND_A_CLONE_CTRANSID,
 		    le64_to_cpu(clone_root->root->root_item.ctransid));
 	TLV_PUT_PATH(sctx, BTRFS_SEND_A_CLONE_PATH, p);
diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
index 9e66f5e72..cd7ef34d2 100644
--- a/fs/btrfs/super.c
+++ b/fs/btrfs/super.c
@@ -135,6 +135,7 @@ static void btrfs_handle_error(struct btrfs_fs_info *fs_info)
  * __btrfs_std_error decodes expected errors from the caller and
  * invokes the approciate error response.
  */
+__cold
 void __btrfs_std_error(struct btrfs_fs_info *fs_info, const char *function,
 		       unsigned int line, int errno, const char *fmt, ...)
 {
@@ -247,18 +248,11 @@ void __btrfs_std_error(struct btrfs_fs_info *fs_info, const char *function,
  * We'll complete the cleanup in btrfs_end_transaction and
  * btrfs_commit_transaction.
  */
+__cold
 void __btrfs_abort_transaction(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *root, const char *function,
 			       unsigned int line, int errno)
 {
-	/*
-	 * Report first abort since mount
-	 */
-	if (!test_and_set_bit(BTRFS_FS_STATE_TRANS_ABORTED,
-				&root->fs_info->fs_state)) {
-		WARN(1, KERN_DEBUG "BTRFS: Transaction aborted (error %d)\n",
-				errno);
-	}
 	trans->aborted = errno;
 	/* Nothing used. The other threads that have joined this
 	 * transaction may be able to continue. */
@@ -281,6 +275,7 @@ void __btrfs_abort_transaction(struct btrfs_trans_handle *trans,
  * __btrfs_panic decodes unexpected, fatal errors from the caller,
  * issues an alert, and either panics or BUGs, depending on mount options.
  */
+__cold
 void __btrfs_panic(struct btrfs_fs_info *fs_info, const char *function,
 		   unsigned int line, int errno, const char *fmt, ...)
 {
@@ -841,33 +836,153 @@ out:
 	return error;
 }
 
-static struct dentry *get_default_root(struct super_block *sb,
-				       u64 subvol_objectid)
+static char *get_subvol_name_from_objectid(struct btrfs_fs_info *fs_info,
+					   u64 subvol_objectid)
 {
-	struct btrfs_fs_info *fs_info = btrfs_sb(sb);
 	struct btrfs_root *root = fs_info->tree_root;
-	struct btrfs_root *new_root;
-	struct btrfs_dir_item *di;
-	struct btrfs_path *path;
-	struct btrfs_key location;
-	struct inode *inode;
-	u64 dir_id;
-	int new = 0;
+	struct btrfs_root *fs_root;
+	struct btrfs_root_ref *root_ref;
+	struct btrfs_inode_ref *inode_ref;
+	struct btrfs_key key;
+	struct btrfs_path *path = NULL;
+	char *name = NULL, *ptr;
+	u64 dirid;
+	int len;
+	int ret;
+
+	path = btrfs_alloc_path();
+	if (!path) {
+		ret = -ENOMEM;
+		goto err;
+	}
+	path->leave_spinning = 1;
+
+	name = kmalloc(PATH_MAX, GFP_NOFS);
+	if (!name) {
+		ret = -ENOMEM;
+		goto err;
+	}
+	ptr = name + PATH_MAX - 1;
+	ptr[0] = '\0';
 
 	/*
-	 * We have a specific subvol we want to mount, just setup location and
-	 * go look up the root.
+	 * Walk up the subvolume trees in the tree of tree roots by root
+	 * backrefs until we hit the top-level subvolume.
 	 */
-	if (subvol_objectid) {
-		location.objectid = subvol_objectid;
-		location.type = BTRFS_ROOT_ITEM_KEY;
-		location.offset = (u64)-1;
-		goto find_root;
+	while (subvol_objectid != BTRFS_FS_TREE_OBJECTID) {
+		key.objectid = subvol_objectid;
+		key.type = BTRFS_ROOT_BACKREF_KEY;
+		key.offset = (u64)-1;
+
+		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+		if (ret < 0) {
+			goto err;
+		} else if (ret > 0) {
+			ret = btrfs_previous_item(root, path, subvol_objectid,
+						  BTRFS_ROOT_BACKREF_KEY);
+			if (ret < 0) {
+				goto err;
+			} else if (ret > 0) {
+				ret = -ENOENT;
+				goto err;
+			}
+		}
+
+		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
+		subvol_objectid = key.offset;
+
+		root_ref = btrfs_item_ptr(path->nodes[0], path->slots[0],
+					  struct btrfs_root_ref);
+		len = btrfs_root_ref_name_len(path->nodes[0], root_ref);
+		ptr -= len + 1;
+		if (ptr < name) {
+			ret = -ENAMETOOLONG;
+			goto err;
+		}
+		read_extent_buffer(path->nodes[0], ptr + 1,
+				   (unsigned long)(root_ref + 1), len);
+		ptr[0] = '/';
+		dirid = btrfs_root_ref_dirid(path->nodes[0], root_ref);
+		btrfs_release_path(path);
+
+		key.objectid = subvol_objectid;
+		key.type = BTRFS_ROOT_ITEM_KEY;
+		key.offset = (u64)-1;
+		fs_root = btrfs_read_fs_root_no_name(fs_info, &key);
+		if (IS_ERR(fs_root)) {
+			ret = PTR_ERR(fs_root);
+			goto err;
+		}
+
+		/*
+		 * Walk up the filesystem tree by inode refs until we hit the
+		 * root directory.
+		 */
+		while (dirid != BTRFS_FIRST_FREE_OBJECTID) {
+			key.objectid = dirid;
+			key.type = BTRFS_INODE_REF_KEY;
+			key.offset = (u64)-1;
+
+			ret = btrfs_search_slot(NULL, fs_root, &key, path, 0, 0);
+			if (ret < 0) {
+				goto err;
+			} else if (ret > 0) {
+				ret = btrfs_previous_item(fs_root, path, dirid,
+							  BTRFS_INODE_REF_KEY);
+				if (ret < 0) {
+					goto err;
+				} else if (ret > 0) {
+					ret = -ENOENT;
+					goto err;
+				}
+			}
+
+			btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
+			dirid = key.offset;
+
+			inode_ref = btrfs_item_ptr(path->nodes[0],
+						   path->slots[0],
+						   struct btrfs_inode_ref);
+			len = btrfs_inode_ref_name_len(path->nodes[0],
+						       inode_ref);
+			ptr -= len + 1;
+			if (ptr < name) {
+				ret = -ENAMETOOLONG;
+				goto err;
+			}
+			read_extent_buffer(path->nodes[0], ptr + 1,
+					   (unsigned long)(inode_ref + 1), len);
+			ptr[0] = '/';
+			btrfs_release_path(path);
+		}
 	}
 
+	btrfs_free_path(path);
+	if (ptr == name + PATH_MAX - 1) {
+		name[0] = '/';
+		name[1] = '\0';
+	} else {
+		memmove(name, ptr, name + PATH_MAX - ptr);
+	}
+	return name;
+
+err:
+	btrfs_free_path(path);
+	kfree(name);
+	return ERR_PTR(ret);
+}
+
+static int get_default_subvol_objectid(struct btrfs_fs_info *fs_info, u64 *objectid)
+{
+	struct btrfs_root *root = fs_info->tree_root;
+	struct btrfs_dir_item *di;
+	struct btrfs_path *path;
+	struct btrfs_key location;
+	u64 dir_id;
+
 	path = btrfs_alloc_path();
 	if (!path)
-		return ERR_PTR(-ENOMEM);
+		return -ENOMEM;
 	path->leave_spinning = 1;
 
 	/*
@@ -879,58 +994,23 @@ static struct dentry *get_default_root(struct super_block *sb,
 	di = btrfs_lookup_dir_item(NULL, root, path, dir_id, "default", 7, 0);
 	if (IS_ERR(di)) {
 		btrfs_free_path(path);
-		return ERR_CAST(di);
+		return PTR_ERR(di);
 	}
 	if (!di) {
 		/*
 		 * Ok the default dir item isn't there.  This is weird since
 		 * it's always been there, but don't freak out, just try and
-		 * mount to root most subvolume.
+		 * mount the top-level subvolume.
 		 */
 		btrfs_free_path(path);
-		dir_id = BTRFS_FIRST_FREE_OBJECTID;
-		new_root = fs_info->fs_root;
-		goto setup_root;
+		*objectid = BTRFS_FS_TREE_OBJECTID;
+		return 0;
 	}
 
 	btrfs_dir_item_key_to_cpu(path->nodes[0], di, &location);
 	btrfs_free_path(path);
-
-find_root:
-	new_root = btrfs_read_fs_root_no_name(fs_info, &location);
-	if (IS_ERR(new_root))
-		return ERR_CAST(new_root);
-
-	if (!(sb->s_flags & MS_RDONLY)) {
-		int ret;
-		down_read(&fs_info->cleanup_work_sem);
-		ret = btrfs_orphan_cleanup(new_root);
-		up_read(&fs_info->cleanup_work_sem);
-		if (ret)
-			return ERR_PTR(ret);
-	}
-
-	dir_id = btrfs_root_dirid(&new_root->root_item);
-setup_root:
-	location.objectid = dir_id;
-	location.type = BTRFS_INODE_ITEM_KEY;
-	location.offset = 0;
-
-	inode = btrfs_iget(sb, &location, new_root, &new);
-	if (IS_ERR(inode))
-		return ERR_CAST(inode);
-
-	/*
-	 * If we're just mounting the root most subvol put the inode and return
-	 * a reference to the dentry.  We will have already gotten a reference
-	 * to the inode in btrfs_fill_super so we're good to go.
-	 */
-	if (!new && d_inode(sb->s_root) == inode) {
-		iput(inode);
-		return dget(sb->s_root);
-	}
-
-	return d_obtain_root(inode);
+	*objectid = location.objectid;
+	return 0;
 }
 
 static int btrfs_fill_super(struct super_block *sb,
@@ -1108,6 +1188,10 @@ static int btrfs_show_options(struct seq_file *seq, struct dentry *dentry)
 		seq_puts(seq, ",fatal_errors=panic");
 	if (info->commit_interval != BTRFS_DEFAULT_COMMIT_INTERVAL)
 		seq_printf(seq, ",commit=%d", info->commit_interval);
+	seq_printf(seq, ",subvolid=%llu",
+		  BTRFS_I(d_inode(dentry))->root->root_key.objectid);
+	seq_puts(seq, ",subvol=");
+	seq_dentry(seq, dentry, " \t\n\\");
 	return 0;
 }
 
@@ -1138,107 +1222,139 @@ static inline int is_subvolume_inode(struct inode *inode)
 }
 
 /*
- * This will strip out the subvol=%s argument for an argument string and add
- * subvolid=0 to make sure we get the actual tree root for path walking to the
- * subvol we want.
+ * This will add subvolid=0 to the argument string while removing any subvol=
+ * and subvolid= arguments to make sure we get the top-level root for path
+ * walking to the subvol we want.
  */
 static char *setup_root_args(char *args)
 {
-	unsigned len = strlen(args) + 2 + 1;
-	char *src, *dst, *buf;
+	char *buf, *dst, *sep;
 
-	/*
-	 * We need the same args as before, but with this substitution:
-	 * s!subvol=[^,]+!subvolid=0!
-	 *
-	 * Since the replacement string is up to 2 bytes longer than the
-	 * original, allocate strlen(args) + 2 + 1 bytes.
-	 */
+	if (!args)
+		return kstrdup("subvolid=0", GFP_NOFS);
 
-	src = strstr(args, "subvol=");
-	/* This shouldn't happen, but just in case.. */
-	if (!src)
-		return NULL;
-
-	buf = dst = kmalloc(len, GFP_NOFS);
+	/* The worst case is that we add ",subvolid=0" to the end. */
+	buf = dst = kmalloc(strlen(args) + strlen(",subvolid=0") + 1, GFP_NOFS);
 	if (!buf)
 		return NULL;
 
-	/*
-	 * If the subvol= arg is not at the start of the string,
-	 * copy whatever precedes it into buf.
-	 */
-	if (src != args) {
-		*src++ = '\0';
-		strcpy(buf, args);
-		dst += strlen(args);
+	while (1) {
+		sep = strchrnul(args, ',');
+		if (!strstarts(args, "subvol=") &&
+		    !strstarts(args, "subvolid=")) {
+			memcpy(dst, args, sep - args);
+			dst += sep - args;
+			*dst++ = ',';
+		}
+		if (*sep)
+			args = sep + 1;
+		else
+			break;
 	}
-
 	strcpy(dst, "subvolid=0");
-	dst += strlen("subvolid=0");
-
-	/*
-	 * If there is a "," after the original subvol=... string,
-	 * copy that suffix into our buffer.  Otherwise, we're done.
-	 */
-	src = strchr(src, ',');
-	if (src)
-		strcpy(dst, src);
 
 	return buf;
 }
 
-static struct dentry *mount_subvol(const char *subvol_name, int flags,
-				   const char *device_name, char *data)
+static struct dentry *mount_subvol(const char *subvol_name, u64 subvol_objectid,
+				   int flags, const char *device_name,
+				   char *data)
 {
 	struct dentry *root;
-	struct vfsmount *mnt;
+	struct vfsmount *mnt = NULL;
 	char *newargs;
+	int ret;
 
 	newargs = setup_root_args(data);
-	if (!newargs)
-		return ERR_PTR(-ENOMEM);
-	mnt = vfs_kern_mount(&btrfs_fs_type, flags, device_name,
-			     newargs);
+	if (!newargs) {
+		root = ERR_PTR(-ENOMEM);
+		goto out;
+	}
 
-	if (PTR_RET(mnt) == -EBUSY) {
+	mnt = vfs_kern_mount(&btrfs_fs_type, flags, device_name, newargs);
+	if (PTR_ERR_OR_ZERO(mnt) == -EBUSY) {
 		if (flags & MS_RDONLY) {
-			mnt = vfs_kern_mount(&btrfs_fs_type, flags & ~MS_RDONLY, device_name,
-					     newargs);
+			mnt = vfs_kern_mount(&btrfs_fs_type, flags & ~MS_RDONLY,
+					     device_name, newargs);
 		} else {
-			int r;
-			mnt = vfs_kern_mount(&btrfs_fs_type, flags | MS_RDONLY, device_name,
-					     newargs);
+			mnt = vfs_kern_mount(&btrfs_fs_type, flags | MS_RDONLY,
+					     device_name, newargs);
 			if (IS_ERR(mnt)) {
-				kfree(newargs);
-				return ERR_CAST(mnt);
+				root = ERR_CAST(mnt);
+				mnt = NULL;
+				goto out;
 			}
 
-			r = btrfs_remount(mnt->mnt_sb, &flags, NULL);
-			if (r < 0) {
-				/* FIXME: release vfsmount mnt ??*/
-				kfree(newargs);
-				return ERR_PTR(r);
+			down_write(&mnt->mnt_sb->s_umount);
+			ret = btrfs_remount(mnt->mnt_sb, &flags, NULL);
+			up_write(&mnt->mnt_sb->s_umount);
+			if (ret < 0) {
+				root = ERR_PTR(ret);
+				goto out;
 			}
 		}
 	}
+	if (IS_ERR(mnt)) {
+		root = ERR_CAST(mnt);
+		mnt = NULL;
+		goto out;
+	}
 
-	kfree(newargs);
+	if (!subvol_name) {
+		if (!subvol_objectid) {
+			ret = get_default_subvol_objectid(btrfs_sb(mnt->mnt_sb),
+							  &subvol_objectid);
+			if (ret) {
+				root = ERR_PTR(ret);
+				goto out;
+			}
+		}
+		subvol_name = get_subvol_name_from_objectid(btrfs_sb(mnt->mnt_sb),
+							    subvol_objectid);
+		if (IS_ERR(subvol_name)) {
+			root = ERR_CAST(subvol_name);
+			subvol_name = NULL;
+			goto out;
+		}
 
-	if (IS_ERR(mnt))
-		return ERR_CAST(mnt);
+	}
 
 	root = mount_subtree(mnt, subvol_name);
+	/* mount_subtree() drops our reference on the vfsmount. */
+	mnt = NULL;
 
-	if (!IS_ERR(root) && !is_subvolume_inode(d_inode(root))) {
+	if (!IS_ERR(root)) {
 		struct super_block *s = root->d_sb;
-		dput(root);
-		root = ERR_PTR(-EINVAL);
-		deactivate_locked_super(s);
-		printk(KERN_ERR "BTRFS: '%s' is not a valid subvolume\n",
-				subvol_name);
+		struct inode *root_inode = d_inode(root);
+		u64 root_objectid = BTRFS_I(root_inode)->root->root_key.objectid;
+
+		ret = 0;
+		if (!is_subvolume_inode(root_inode)) {
+			pr_err("BTRFS: '%s' is not a valid subvolume\n",
+			       subvol_name);
+			ret = -EINVAL;
+		}
+		if (subvol_objectid && root_objectid != subvol_objectid) {
+			/*
+			 * This will also catch a race condition where a
+			 * subvolume which was passed by ID is renamed and
+			 * another subvolume is renamed over the old location.
+			 */
+			pr_err("BTRFS: subvol '%s' does not match subvolid %llu\n",
+			       subvol_name, subvol_objectid);
+			ret = -EINVAL;
+		}
+		if (ret) {
+			dput(root);
+			root = ERR_PTR(ret);
+			deactivate_locked_super(s);
+		}
 	}
 
+out:
+	mntput(mnt);
+	kfree(newargs);
+	kfree(subvol_name);
 	return root;
 }
 
@@ -1303,7 +1419,6 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
 {
 	struct block_device *bdev = NULL;
 	struct super_block *s;
-	struct dentry *root;
 	struct btrfs_fs_devices *fs_devices = NULL;
 	struct btrfs_fs_info *fs_info = NULL;
 	struct security_mnt_opts new_sec_opts;
@@ -1323,10 +1438,10 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
 		return ERR_PTR(error);
 	}
 
-	if (subvol_name) {
-		root = mount_subvol(subvol_name, flags, device_name, data);
-		kfree(subvol_name);
-		return root;
+	if (subvol_name || subvol_objectid != BTRFS_FS_TREE_OBJECTID) {
+		/* mount_subvol() will free subvol_name. */
+		return mount_subvol(subvol_name, subvol_objectid, flags,
+				    device_name, data);
 	}
 
 	security_init_mnt_opts(&new_sec_opts);
@@ -1392,23 +1507,19 @@ static struct dentry *btrfs_mount(struct file_system_type *fs_type, int flags,
 		error = btrfs_fill_super(s, fs_devices, data,
 					 flags & MS_SILENT ? 1 : 0);
 	}
-
-	root = !error ? get_default_root(s, subvol_objectid) : ERR_PTR(error);
-	if (IS_ERR(root)) {
+	if (error) {
 		deactivate_locked_super(s);
-		error = PTR_ERR(root);
 		goto error_sec_opts;
 	}
 
 	fs_info = btrfs_sb(s);
 	error = setup_security_options(fs_info, s, &new_sec_opts);
 	if (error) {
-		dput(root);
 		deactivate_locked_super(s);
 		goto error_sec_opts;
 	}
 
-	return root;
+	return dget(s->s_root);
 
 error_close_devices:
 	btrfs_close_devices(fs_devices);
diff --git a/fs/btrfs/sysfs.c b/fs/btrfs/sysfs.c
index e8a4c86d2..603b0cc2b 100644
--- a/fs/btrfs/sysfs.c
+++ b/fs/btrfs/sysfs.c
@@ -33,6 +33,7 @@
 #include "volumes.h"
 
 static inline struct btrfs_fs_info *to_fs_info(struct kobject *kobj);
+static inline struct btrfs_fs_devices *to_fs_devs(struct kobject *kobj);
 
 static u64 get_features(struct btrfs_fs_info *fs_info,
 			enum btrfs_feature_set set)
@@ -428,7 +429,7 @@ static ssize_t btrfs_clone_alignment_show(struct kobject *kobj,
 
 BTRFS_ATTR(clone_alignment, btrfs_clone_alignment_show);
 
-static struct attribute *btrfs_attrs[] = {
+static const struct attribute *btrfs_attrs[] = {
 	BTRFS_ATTR_PTR(label),
 	BTRFS_ATTR_PTR(nodesize),
 	BTRFS_ATTR_PTR(sectorsize),
@@ -438,21 +439,29 @@ static struct attribute *btrfs_attrs[] = {
 
 static void btrfs_release_super_kobj(struct kobject *kobj)
 {
-	struct btrfs_fs_info *fs_info = to_fs_info(kobj);
-	complete(&fs_info->kobj_unregister);
+	struct btrfs_fs_devices *fs_devs = to_fs_devs(kobj);
+
+	memset(&fs_devs->super_kobj, 0, sizeof(struct kobject));
+	complete(&fs_devs->kobj_unregister);
 }
 
 static struct kobj_type btrfs_ktype = {
 	.sysfs_ops	= &kobj_sysfs_ops,
 	.release	= btrfs_release_super_kobj,
-	.default_attrs	= btrfs_attrs,
 };
 
+static inline struct btrfs_fs_devices *to_fs_devs(struct kobject *kobj)
+{
+	if (kobj->ktype != &btrfs_ktype)
+		return NULL;
+	return container_of(kobj, struct btrfs_fs_devices, super_kobj);
+}
+
 static inline struct btrfs_fs_info *to_fs_info(struct kobject *kobj)
 {
 	if (kobj->ktype != &btrfs_ktype)
 		return NULL;
-	return container_of(kobj, struct btrfs_fs_info, super_kobj);
+	return to_fs_devs(kobj)->fs_info;
 }
 
 #define NUM_FEATURE_BITS 64
@@ -493,12 +502,12 @@ static int addrm_unknown_feature_attrs(struct btrfs_fs_info *fs_info, bool add)
 			attrs[0] = &fa->kobj_attr.attr;
 			if (add) {
 				int ret;
-				ret = sysfs_merge_group(&fs_info->super_kobj,
+				ret = sysfs_merge_group(&fs_info->fs_devices->super_kobj,
 							&agroup);
 				if (ret)
 					return ret;
 			} else
-				sysfs_unmerge_group(&fs_info->super_kobj,
+				sysfs_unmerge_group(&fs_info->fs_devices->super_kobj,
 						    &agroup);
 		}
 
@@ -506,25 +515,49 @@ static int addrm_unknown_feature_attrs(struct btrfs_fs_info *fs_info, bool add)
 	return 0;
 }
 
-static void __btrfs_sysfs_remove_one(struct btrfs_fs_info *fs_info)
+static void __btrfs_sysfs_remove_fsid(struct btrfs_fs_devices *fs_devs)
+{
+	if (fs_devs->device_dir_kobj) {
+		kobject_del(fs_devs->device_dir_kobj);
+		kobject_put(fs_devs->device_dir_kobj);
+		fs_devs->device_dir_kobj = NULL;
+	}
+
+	if (fs_devs->super_kobj.state_initialized) {
+		kobject_del(&fs_devs->super_kobj);
+		kobject_put(&fs_devs->super_kobj);
+		wait_for_completion(&fs_devs->kobj_unregister);
+	}
+}
+
+/* when fs_devs is NULL it will remove all fsid kobject */
+void btrfs_sysfs_remove_fsid(struct btrfs_fs_devices *fs_devs)
 {
-	kobject_del(&fs_info->super_kobj);
-	kobject_put(&fs_info->super_kobj);
-	wait_for_completion(&fs_info->kobj_unregister);
+	struct list_head *fs_uuids = btrfs_get_fs_uuids();
+
+	if (fs_devs) {
+		__btrfs_sysfs_remove_fsid(fs_devs);
+		return;
+	}
+
+	list_for_each_entry(fs_devs, fs_uuids, list) {
+		__btrfs_sysfs_remove_fsid(fs_devs);
+	}
 }
 
 void btrfs_sysfs_remove_one(struct btrfs_fs_info *fs_info)
 {
+	btrfs_reset_fs_info_ptr(fs_info);
+
 	if (fs_info->space_info_kobj) {
 		sysfs_remove_files(fs_info->space_info_kobj, allocation_attrs);
 		kobject_del(fs_info->space_info_kobj);
 		kobject_put(fs_info->space_info_kobj);
 	}
-	kobject_del(fs_info->device_dir_kobj);
-	kobject_put(fs_info->device_dir_kobj);
 	addrm_unknown_feature_attrs(fs_info, false);
-	sysfs_remove_group(&fs_info->super_kobj, &btrfs_feature_attr_group);
-	__btrfs_sysfs_remove_one(fs_info);
+	sysfs_remove_group(&fs_info->fs_devices->super_kobj, &btrfs_feature_attr_group);
+	sysfs_remove_files(&fs_info->fs_devices->super_kobj, btrfs_attrs);
+	btrfs_kobj_rm_device(fs_info->fs_devices, NULL);
 }
 
 const char * const btrfs_feature_set_names[3] = {
@@ -602,40 +635,60 @@ static void init_feature_attrs(void)
 	}
 }
 
-int btrfs_kobj_rm_device(struct btrfs_fs_info *fs_info,
+/* when one_device is NULL, it removes all device links */
+
+int btrfs_kobj_rm_device(struct btrfs_fs_devices *fs_devices,
 		struct btrfs_device *one_device)
 {
 	struct hd_struct *disk;
 	struct kobject *disk_kobj;
 
-	if (!fs_info->device_dir_kobj)
+	if (!fs_devices->device_dir_kobj)
 		return -EINVAL;
 
 	if (one_device && one_device->bdev) {
 		disk = one_device->bdev->bd_part;
 		disk_kobj = &part_to_dev(disk)->kobj;
 
-		sysfs_remove_link(fs_info->device_dir_kobj,
+		sysfs_remove_link(fs_devices->device_dir_kobj,
+						disk_kobj->name);
+	}
+
+	if (one_device)
+		return 0;
+
+	list_for_each_entry(one_device,
+			&fs_devices->devices, dev_list) {
+		if (!one_device->bdev)
+			continue;
+		disk = one_device->bdev->bd_part;
+		disk_kobj = &part_to_dev(disk)->kobj;
+
+		sysfs_remove_link(fs_devices->device_dir_kobj,
 						disk_kobj->name);
 	}
 
 	return 0;
 }
 
-int btrfs_kobj_add_device(struct btrfs_fs_info *fs_info,
-		struct btrfs_device *one_device)
+int btrfs_sysfs_add_device(struct btrfs_fs_devices *fs_devs)
 {
-	int error = 0;
-	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
-	struct btrfs_device *dev;
-
-	if (!fs_info->device_dir_kobj)
-		fs_info->device_dir_kobj = kobject_create_and_add("devices",
-						&fs_info->super_kobj);
+	if (!fs_devs->device_dir_kobj)
+		fs_devs->device_dir_kobj = kobject_create_and_add("devices",
+						&fs_devs->super_kobj);
 
-	if (!fs_info->device_dir_kobj)
+	if (!fs_devs->device_dir_kobj)
 		return -ENOMEM;
 
+	return 0;
+}
+
+int btrfs_kobj_add_device(struct btrfs_fs_devices *fs_devices,
+				struct btrfs_device *one_device)
+{
+	int error = 0;
+	struct btrfs_device *dev;
+
 	list_for_each_entry(dev, &fs_devices->devices, dev_list) {
 		struct hd_struct *disk;
 		struct kobject *disk_kobj;
@@ -649,7 +702,7 @@ int btrfs_kobj_add_device(struct btrfs_fs_info *fs_info,
 		disk = dev->bdev->bd_part;
 		disk_kobj = &part_to_dev(disk)->kobj;
 
-		error = sysfs_create_link(fs_info->device_dir_kobj,
+		error = sysfs_create_link(fs_devices->device_dir_kobj,
 					  disk_kobj, disk_kobj->name);
 		if (error)
 			break;
@@ -667,34 +720,51 @@ static struct dentry *btrfs_debugfs_root_dentry;
 /* Debugging tunables and exported data */
 u64 btrfs_debugfs_test;
 
+/*
+ * Can be called by the device discovery thread.
+ * And parent can be specified for seed device
+ */
+int btrfs_sysfs_add_fsid(struct btrfs_fs_devices *fs_devs,
+				struct kobject *parent)
+{
+	int error;
+
+	init_completion(&fs_devs->kobj_unregister);
+	fs_devs->super_kobj.kset = btrfs_kset;
+	error = kobject_init_and_add(&fs_devs->super_kobj,
+				&btrfs_ktype, parent, "%pU", fs_devs->fsid);
+	return error;
+}
+
 int btrfs_sysfs_add_one(struct btrfs_fs_info *fs_info)
 {
 	int error;
+	struct btrfs_fs_devices *fs_devs = fs_info->fs_devices;
+	struct kobject *super_kobj = &fs_devs->super_kobj;
+
+	btrfs_set_fs_info_ptr(fs_info);
 
-	init_completion(&fs_info->kobj_unregister);
-	fs_info->super_kobj.kset = btrfs_kset;
-	error = kobject_init_and_add(&fs_info->super_kobj, &btrfs_ktype, NULL,
-				     "%pU", fs_info->fsid);
+	error = btrfs_kobj_add_device(fs_devs, NULL);
 	if (error)
 		return error;
 
-	error = sysfs_create_group(&fs_info->super_kobj,
-				   &btrfs_feature_attr_group);
+	error = sysfs_create_files(super_kobj, btrfs_attrs);
 	if (error) {
-		__btrfs_sysfs_remove_one(fs_info);
+		btrfs_kobj_rm_device(fs_devs, NULL);
 		return error;
 	}
 
-	error = addrm_unknown_feature_attrs(fs_info, true);
+	error = sysfs_create_group(super_kobj,
+				   &btrfs_feature_attr_group);
 	if (error)
 		goto failure;
 
-	error = btrfs_kobj_add_device(fs_info, NULL);
+	error = addrm_unknown_feature_attrs(fs_info, true);
 	if (error)
 		goto failure;
 
 	fs_info->space_info_kobj = kobject_create_and_add("allocation",
-						  &fs_info->super_kobj);
+						  super_kobj);
 	if (!fs_info->space_info_kobj) {
 		error = -ENOMEM;
 		goto failure;
diff --git a/fs/btrfs/sysfs.h b/fs/btrfs/sysfs.h
index 3a4bbed72..6392527bc 100644
--- a/fs/btrfs/sysfs.h
+++ b/fs/btrfs/sysfs.h
@@ -82,8 +82,12 @@ char *btrfs_printable_features(enum btrfs_feature_set set, u64 flags);
 extern const char * const btrfs_feature_set_names[3];
 extern struct kobj_type space_info_ktype;
 extern struct kobj_type btrfs_raid_ktype;
-int btrfs_kobj_add_device(struct btrfs_fs_info *fs_info,
+int btrfs_kobj_add_device(struct btrfs_fs_devices *fs_devices,
 		struct btrfs_device *one_device);
-int btrfs_kobj_rm_device(struct btrfs_fs_info *fs_info,
+int btrfs_kobj_rm_device(struct btrfs_fs_devices *fs_devices,
                 struct btrfs_device *one_device);
+int btrfs_sysfs_add_fsid(struct btrfs_fs_devices *fs_devs,
+				struct kobject *parent);
+int btrfs_sysfs_add_device(struct btrfs_fs_devices *fs_devs);
+void btrfs_sysfs_remove_fsid(struct btrfs_fs_devices *fs_devs);
 #endif /* _BTRFS_SYSFS_H_ */
diff --git a/fs/btrfs/tests/qgroup-tests.c b/fs/btrfs/tests/qgroup-tests.c
index c32a7ba76..846d277b1 100644
--- a/fs/btrfs/tests/qgroup-tests.c
+++ b/fs/btrfs/tests/qgroup-tests.c
@@ -21,6 +21,7 @@
 #include "../transaction.h"
 #include "../disk-io.h"
 #include "../qgroup.h"
+#include "../backref.h"
 
 static void init_dummy_trans(struct btrfs_trans_handle *trans)
 {
@@ -227,6 +228,8 @@ static int test_no_shared_qgroup(struct btrfs_root *root)
 {
 	struct btrfs_trans_handle trans;
 	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct ulist *old_roots = NULL;
+	struct ulist *new_roots = NULL;
 	int ret;
 
 	init_dummy_trans(&trans);
@@ -238,10 +241,15 @@ static int test_no_shared_qgroup(struct btrfs_root *root)
 		return ret;
 	}
 
-	ret = btrfs_qgroup_record_ref(&trans, fs_info, 5, 4096, 4096,
-				      BTRFS_QGROUP_OPER_ADD_EXCL, 0);
+	/*
+	 * Since the test trans doesn't havee the complicated delayed refs,
+	 * we can only call btrfs_qgroup_account_extent() directly to test
+	 * quota.
+	 */
+	ret = btrfs_find_all_roots(&trans, fs_info, 4096, 0, &old_roots);
 	if (ret) {
-		test_msg("Couldn't add space to a qgroup %d\n", ret);
+		ulist_free(old_roots);
+		test_msg("Couldn't find old roots: %d\n", ret);
 		return ret;
 	}
 
@@ -249,9 +257,18 @@ static int test_no_shared_qgroup(struct btrfs_root *root)
 	if (ret)
 		return ret;
 
-	ret = btrfs_delayed_qgroup_accounting(&trans, fs_info);
+	ret = btrfs_find_all_roots(&trans, fs_info, 4096, 0, &new_roots);
+	if (ret) {
+		ulist_free(old_roots);
+		ulist_free(new_roots);
+		test_msg("Couldn't find old roots: %d\n", ret);
+		return ret;
+	}
+
+	ret = btrfs_qgroup_account_extent(&trans, fs_info, 4096, 4096,
+					  old_roots, new_roots);
 	if (ret) {
-		test_msg("Delayed qgroup accounting failed %d\n", ret);
+		test_msg("Couldn't account space for a qgroup %d\n", ret);
 		return ret;
 	}
 
@@ -259,21 +276,32 @@ static int test_no_shared_qgroup(struct btrfs_root *root)
 		test_msg("Qgroup counts didn't match expected values\n");
 		return -EINVAL;
 	}
+	old_roots = NULL;
+	new_roots = NULL;
+
+	ret = btrfs_find_all_roots(&trans, fs_info, 4096, 0, &old_roots);
+	if (ret) {
+		ulist_free(old_roots);
+		test_msg("Couldn't find old roots: %d\n", ret);
+		return ret;
+	}
 
 	ret = remove_extent_item(root, 4096, 4096);
 	if (ret)
 		return -EINVAL;
 
-	ret = btrfs_qgroup_record_ref(&trans, fs_info, 5, 4096, 4096,
-				      BTRFS_QGROUP_OPER_SUB_EXCL, 0);
+	ret = btrfs_find_all_roots(&trans, fs_info, 4096, 0, &new_roots);
 	if (ret) {
-		test_msg("Couldn't remove space from the qgroup %d\n", ret);
-		return -EINVAL;
+		ulist_free(old_roots);
+		ulist_free(new_roots);
+		test_msg("Couldn't find old roots: %d\n", ret);
+		return ret;
 	}
 
-	ret = btrfs_delayed_qgroup_accounting(&trans, fs_info);
+	ret = btrfs_qgroup_account_extent(&trans, fs_info, 4096, 4096,
+					  old_roots, new_roots);
 	if (ret) {
-		test_msg("Qgroup accounting failed %d\n", ret);
+		test_msg("Couldn't account space for a qgroup %d\n", ret);
 		return -EINVAL;
 	}
 
@@ -294,6 +322,8 @@ static int test_multiple_refs(struct btrfs_root *root)
 {
 	struct btrfs_trans_handle trans;
 	struct btrfs_fs_info *fs_info = root->fs_info;
+	struct ulist *old_roots = NULL;
+	struct ulist *new_roots = NULL;
 	int ret;
 
 	init_dummy_trans(&trans);
@@ -307,20 +337,29 @@ static int test_multiple_refs(struct btrfs_root *root)
 		return ret;
 	}
 
+	ret = btrfs_find_all_roots(&trans, fs_info, 4096, 0, &old_roots);
+	if (ret) {
+		ulist_free(old_roots);
+		test_msg("Couldn't find old roots: %d\n", ret);
+		return ret;
+	}
+
 	ret = insert_normal_tree_ref(root, 4096, 4096, 0, 5);
 	if (ret)
 		return ret;
 
-	ret = btrfs_qgroup_record_ref(&trans, fs_info, 5, 4096, 4096,
-				      BTRFS_QGROUP_OPER_ADD_EXCL, 0);
+	ret = btrfs_find_all_roots(&trans, fs_info, 4096, 0, &new_roots);
 	if (ret) {
-		test_msg("Couldn't add space to a qgroup %d\n", ret);
+		ulist_free(old_roots);
+		ulist_free(new_roots);
+		test_msg("Couldn't find old roots: %d\n", ret);
 		return ret;
 	}
 
-	ret = btrfs_delayed_qgroup_accounting(&trans, fs_info);
+	ret = btrfs_qgroup_account_extent(&trans, fs_info, 4096, 4096,
+					  old_roots, new_roots);
 	if (ret) {
-		test_msg("Delayed qgroup accounting failed %d\n", ret);
+		test_msg("Couldn't account space for a qgroup %d\n", ret);
 		return ret;
 	}
 
@@ -329,20 +368,29 @@ static int test_multiple_refs(struct btrfs_root *root)
 		return -EINVAL;
 	}
 
+	ret = btrfs_find_all_roots(&trans, fs_info, 4096, 0, &old_roots);
+	if (ret) {
+		ulist_free(old_roots);
+		test_msg("Couldn't find old roots: %d\n", ret);
+		return ret;
+	}
+
 	ret = add_tree_ref(root, 4096, 4096, 0, 256);
 	if (ret)
 		return ret;
 
-	ret = btrfs_qgroup_record_ref(&trans, fs_info, 256, 4096, 4096,
-				      BTRFS_QGROUP_OPER_ADD_SHARED, 0);
+	ret = btrfs_find_all_roots(&trans, fs_info, 4096, 0, &new_roots);
 	if (ret) {
-		test_msg("Qgroup record ref failed %d\n", ret);
+		ulist_free(old_roots);
+		ulist_free(new_roots);
+		test_msg("Couldn't find old roots: %d\n", ret);
 		return ret;
 	}
 
-	ret = btrfs_delayed_qgroup_accounting(&trans, fs_info);
+	ret = btrfs_qgroup_account_extent(&trans, fs_info, 4096, 4096,
+					  old_roots, new_roots);
 	if (ret) {
-		test_msg("Qgroup accounting failed %d\n", ret);
+		test_msg("Couldn't account space for a qgroup %d\n", ret);
 		return ret;
 	}
 
@@ -356,20 +404,29 @@ static int test_multiple_refs(struct btrfs_root *root)
 		return -EINVAL;
 	}
 
+	ret = btrfs_find_all_roots(&trans, fs_info, 4096, 0, &old_roots);
+	if (ret) {
+		ulist_free(old_roots);
+		test_msg("Couldn't find old roots: %d\n", ret);
+		return ret;
+	}
+
 	ret = remove_extent_ref(root, 4096, 4096, 0, 256);
 	if (ret)
 		return ret;
 
-	ret = btrfs_qgroup_record_ref(&trans, fs_info, 256, 4096, 4096,
-				      BTRFS_QGROUP_OPER_SUB_SHARED, 0);
+	ret = btrfs_find_all_roots(&trans, fs_info, 4096, 0, &new_roots);
 	if (ret) {
-		test_msg("Qgroup record ref failed %d\n", ret);
+		ulist_free(old_roots);
+		ulist_free(new_roots);
+		test_msg("Couldn't find old roots: %d\n", ret);
 		return ret;
 	}
 
-	ret = btrfs_delayed_qgroup_accounting(&trans, fs_info);
+	ret = btrfs_qgroup_account_extent(&trans, fs_info, 4096, 4096,
+					  old_roots, new_roots);
 	if (ret) {
-		test_msg("Qgroup accounting failed %d\n", ret);
+		test_msg("Couldn't account space for a qgroup %d\n", ret);
 		return ret;
 	}
 
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 94e909c5a..f5021fcb1 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -225,12 +225,14 @@ loop:
 	cur_trans->dirty_bg_run = 0;
 
 	cur_trans->delayed_refs.href_root = RB_ROOT;
+	cur_trans->delayed_refs.dirty_extent_root = RB_ROOT;
 	atomic_set(&cur_trans->delayed_refs.num_entries, 0);
 	cur_trans->delayed_refs.num_heads_ready = 0;
 	cur_trans->delayed_refs.pending_csums = 0;
 	cur_trans->delayed_refs.num_heads = 0;
 	cur_trans->delayed_refs.flushing = 0;
 	cur_trans->delayed_refs.run_delayed_start = 0;
+	cur_trans->delayed_refs.qgroup_to_skip = 0;
 
 	/*
 	 * although the tree mod log is per file system and not per transaction,
@@ -509,6 +511,7 @@ again:
 	h->transaction = cur_trans;
 	h->blocks_used = 0;
 	h->bytes_reserved = 0;
+	h->chunk_bytes_reserved = 0;
 	h->root = root;
 	h->delayed_ref_updates = 0;
 	h->use_count = 1;
@@ -792,6 +795,8 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
 	if (!list_empty(&trans->new_bgs))
 		btrfs_create_pending_block_groups(trans, root);
 
+	btrfs_trans_release_chunk_metadata(trans);
+
 	if (lock && !atomic_read(&root->fs_info->open_ioctl_trans) &&
 	    should_end_transaction(trans, root) &&
 	    ACCESS_ONCE(cur_trans->state) == TRANS_STATE_RUNNING) {
@@ -1290,6 +1295,12 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	if (pending->error)
 		goto no_free_objectid;
 
+	/*
+	 * Make qgroup to skip current new snapshot's qgroupid, as it is
+	 * accounted by later btrfs_qgroup_inherit().
+	 */
+	btrfs_set_skip_qgroup(trans, objectid);
+
 	btrfs_reloc_pre_snapshot(trans, pending, &to_reserve);
 
 	if (to_reserve > 0) {
@@ -1298,7 +1309,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 						     to_reserve,
 						     BTRFS_RESERVE_NO_FLUSH);
 		if (pending->error)
-			goto no_free_objectid;
+			goto clear_skip_qgroup;
 	}
 
 	key.objectid = objectid;
@@ -1396,25 +1407,6 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 		btrfs_abort_transaction(trans, root, ret);
 		goto fail;
 	}
-
-	/*
-	 * We need to flush delayed refs in order to make sure all of our quota
-	 * operations have been done before we call btrfs_qgroup_inherit.
-	 */
-	ret = btrfs_run_delayed_refs(trans, root, (unsigned long)-1);
-	if (ret) {
-		btrfs_abort_transaction(trans, root, ret);
-		goto fail;
-	}
-
-	ret = btrfs_qgroup_inherit(trans, fs_info,
-				   root->root_key.objectid,
-				   objectid, pending->inherit);
-	if (ret) {
-		btrfs_abort_transaction(trans, root, ret);
-		goto fail;
-	}
-
 	/* see comments in should_cow_block() */
 	set_bit(BTRFS_ROOT_FORCE_COW, &root->state);
 	smp_wmb();
@@ -1497,11 +1489,37 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 			goto fail;
 		}
 	}
+
+	ret = btrfs_run_delayed_refs(trans, root, (unsigned long)-1);
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto fail;
+	}
+
+	/*
+	 * account qgroup counters before qgroup_inherit()
+	 */
+	ret = btrfs_qgroup_prepare_account_extents(trans, fs_info);
+	if (ret)
+		goto fail;
+	ret = btrfs_qgroup_account_extents(trans, fs_info);
+	if (ret)
+		goto fail;
+	ret = btrfs_qgroup_inherit(trans, fs_info,
+				   root->root_key.objectid,
+				   objectid, pending->inherit);
+	if (ret) {
+		btrfs_abort_transaction(trans, root, ret);
+		goto fail;
+	}
+
 fail:
 	pending->error = ret;
 dir_item_existed:
 	trans->block_rsv = rsv;
 	trans->bytes_reserved = 0;
+clear_skip_qgroup:
+	btrfs_clear_skip_qgroup(trans);
 no_free_objectid:
 	kfree(new_root_item);
 root_item_alloc_fail:
@@ -1963,6 +1981,13 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 		goto scrub_continue;
 	}
 
+	/* Reocrd old roots for later qgroup accounting */
+	ret = btrfs_qgroup_prepare_account_extents(trans, root->fs_info);
+	if (ret) {
+		mutex_unlock(&root->fs_info->reloc_mutex);
+		goto scrub_continue;
+	}
+
 	/*
 	 * make sure none of the code above managed to slip in a
 	 * delayed item
@@ -2004,6 +2029,17 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	 */
 	btrfs_free_log_root_tree(trans, root->fs_info);
 
+	/*
+	 * Since fs roots are all committed, we can get a quite accurate
+	 * new_roots. So let's do quota accounting.
+	 */
+	ret = btrfs_qgroup_account_extents(trans, root->fs_info);
+	if (ret < 0) {
+		mutex_unlock(&root->fs_info->tree_log_mutex);
+		mutex_unlock(&root->fs_info->reloc_mutex);
+		goto scrub_continue;
+	}
+
 	ret = commit_cowonly_roots(trans, root);
 	if (ret) {
 		mutex_unlock(&root->fs_info->tree_log_mutex);
@@ -2054,6 +2090,8 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	clear_bit(BTRFS_INODE_BTREE_LOG1_ERR, &btree_ino->runtime_flags);
 	clear_bit(BTRFS_INODE_BTREE_LOG2_ERR, &btree_ino->runtime_flags);
 
+	btrfs_trans_release_chunk_metadata(trans);
+
 	spin_lock(&root->fs_info->trans_lock);
 	cur_trans->state = TRANS_STATE_UNBLOCKED;
 	root->fs_info->running_transaction = NULL;
@@ -2114,7 +2152,8 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 
 	kmem_cache_free(btrfs_trans_handle_cachep, trans);
 
-	if (current != root->fs_info->transaction_kthread)
+	if (current != root->fs_info->transaction_kthread &&
+	    current != root->fs_info->cleaner_kthread)
 		btrfs_run_delayed_iputs(root);
 
 	return ret;
@@ -2123,6 +2162,7 @@ scrub_continue:
 	btrfs_scrub_continue(root);
 cleanup_transaction:
 	btrfs_trans_release_metadata(trans, root);
+	btrfs_trans_release_chunk_metadata(trans);
 	trans->block_rsv = NULL;
 	if (trans->qgroup_reserved) {
 		btrfs_qgroup_free(root, trans->qgroup_reserved);
diff --git a/fs/btrfs/transaction.h b/fs/btrfs/transaction.h
index 0b2475559..eb09c2067 100644
--- a/fs/btrfs/transaction.h
+++ b/fs/btrfs/transaction.h
@@ -102,6 +102,7 @@ struct btrfs_transaction {
 struct btrfs_trans_handle {
 	u64 transid;
 	u64 bytes_reserved;
+	u64 chunk_bytes_reserved;
 	u64 qgroup_reserved;
 	unsigned long use_count;
 	unsigned long blocks_reserved;
@@ -153,6 +154,29 @@ static inline void btrfs_set_inode_last_trans(struct btrfs_trans_handle *trans,
 	spin_unlock(&BTRFS_I(inode)->lock);
 }
 
+/*
+ * Make qgroup codes to skip given qgroupid, means the old/new_roots for
+ * qgroup won't contain the qgroupid in it.
+ */
+static inline void btrfs_set_skip_qgroup(struct btrfs_trans_handle *trans,
+					 u64 qgroupid)
+{
+	struct btrfs_delayed_ref_root *delayed_refs;
+
+	delayed_refs = &trans->transaction->delayed_refs;
+	WARN_ON(delayed_refs->qgroup_to_skip);
+	delayed_refs->qgroup_to_skip = qgroupid;
+}
+
+static inline void btrfs_clear_skip_qgroup(struct btrfs_trans_handle *trans)
+{
+	struct btrfs_delayed_ref_root *delayed_refs;
+
+	delayed_refs = &trans->transaction->delayed_refs;
+	WARN_ON(!delayed_refs->qgroup_to_skip);
+	delayed_refs->qgroup_to_skip = 0;
+}
+
 int btrfs_end_transaction(struct btrfs_trans_handle *trans,
 			  struct btrfs_root *root);
 struct btrfs_trans_handle *btrfs_start_transaction(struct btrfs_root *root,
diff --git a/fs/btrfs/tree-defrag.c b/fs/btrfs/tree-defrag.c
index a63719cc9..a4b9c8b2d 100644
--- a/fs/btrfs/tree-defrag.c
+++ b/fs/btrfs/tree-defrag.c
@@ -52,9 +52,6 @@ int btrfs_defrag_leaves(struct btrfs_trans_handle *trans,
 	if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state))
 		goto out;
 
-	if (btrfs_test_opt(root, SSD))
-		goto out;
-
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c
index 4920fceff..9c45431e6 100644
--- a/fs/btrfs/tree-log.c
+++ b/fs/btrfs/tree-log.c
@@ -3881,12 +3881,6 @@ static int wait_ordered_extents(struct btrfs_trans_handle *trans,
 				     &ordered->flags))
 			continue;
 
-		if (ordered->csum_bytes_left) {
-			btrfs_start_ordered_extent(inode, ordered, 0);
-			wait_event(ordered->wait,
-				   ordered->csum_bytes_left == 0);
-		}
-
 		list_for_each_entry(sum, &ordered->list, list) {
 			ret = btrfs_csum_file_blocks(trans, log, sum);
 			if (ret)
@@ -4123,6 +4117,187 @@ static int logged_inode_size(struct btrfs_root *log, struct inode *inode,
 	return 0;
 }
 
+/*
+ * At the moment we always log all xattrs. This is to figure out at log replay
+ * time which xattrs must have their deletion replayed. If a xattr is missing
+ * in the log tree and exists in the fs/subvol tree, we delete it. This is
+ * because if a xattr is deleted, the inode is fsynced and a power failure
+ * happens, causing the log to be replayed the next time the fs is mounted,
+ * we want the xattr to not exist anymore (same behaviour as other filesystems
+ * with a journal, ext3/4, xfs, f2fs, etc).
+ */
+static int btrfs_log_all_xattrs(struct btrfs_trans_handle *trans,
+				struct btrfs_root *root,
+				struct inode *inode,
+				struct btrfs_path *path,
+				struct btrfs_path *dst_path)
+{
+	int ret;
+	struct btrfs_key key;
+	const u64 ino = btrfs_ino(inode);
+	int ins_nr = 0;
+	int start_slot = 0;
+
+	key.objectid = ino;
+	key.type = BTRFS_XATTR_ITEM_KEY;
+	key.offset = 0;
+
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	if (ret < 0)
+		return ret;
+
+	while (true) {
+		int slot = path->slots[0];
+		struct extent_buffer *leaf = path->nodes[0];
+		int nritems = btrfs_header_nritems(leaf);
+
+		if (slot >= nritems) {
+			if (ins_nr > 0) {
+				u64 last_extent = 0;
+
+				ret = copy_items(trans, inode, dst_path, path,
+						 &last_extent, start_slot,
+						 ins_nr, 1, 0);
+				/* can't be 1, extent items aren't processed */
+				ASSERT(ret <= 0);
+				if (ret < 0)
+					return ret;
+				ins_nr = 0;
+			}
+			ret = btrfs_next_leaf(root, path);
+			if (ret < 0)
+				return ret;
+			else if (ret > 0)
+				break;
+			continue;
+		}
+
+		btrfs_item_key_to_cpu(leaf, &key, slot);
+		if (key.objectid != ino || key.type != BTRFS_XATTR_ITEM_KEY)
+			break;
+
+		if (ins_nr == 0)
+			start_slot = slot;
+		ins_nr++;
+		path->slots[0]++;
+		cond_resched();
+	}
+	if (ins_nr > 0) {
+		u64 last_extent = 0;
+
+		ret = copy_items(trans, inode, dst_path, path,
+				 &last_extent, start_slot,
+				 ins_nr, 1, 0);
+		/* can't be 1, extent items aren't processed */
+		ASSERT(ret <= 0);
+		if (ret < 0)
+			return ret;
+	}
+
+	return 0;
+}
+
+/*
+ * If the no holes feature is enabled we need to make sure any hole between the
+ * last extent and the i_size of our inode is explicitly marked in the log. This
+ * is to make sure that doing something like:
+ *
+ *      1) create file with 128Kb of data
+ *      2) truncate file to 64Kb
+ *      3) truncate file to 256Kb
+ *      4) fsync file
+ *      5) <crash/power failure>
+ *      6) mount fs and trigger log replay
+ *
+ * Will give us a file with a size of 256Kb, the first 64Kb of data match what
+ * the file had in its first 64Kb of data at step 1 and the last 192Kb of the
+ * file correspond to a hole. The presence of explicit holes in a log tree is
+ * what guarantees that log replay will remove/adjust file extent items in the
+ * fs/subvol tree.
+ *
+ * Here we do not need to care about holes between extents, that is already done
+ * by copy_items(). We also only need to do this in the full sync path, where we
+ * lookup for extents from the fs/subvol tree only. In the fast path case, we
+ * lookup the list of modified extent maps and if any represents a hole, we
+ * insert a corresponding extent representing a hole in the log tree.
+ */
+static int btrfs_log_trailing_hole(struct btrfs_trans_handle *trans,
+				   struct btrfs_root *root,
+				   struct inode *inode,
+				   struct btrfs_path *path)
+{
+	int ret;
+	struct btrfs_key key;
+	u64 hole_start;
+	u64 hole_size;
+	struct extent_buffer *leaf;
+	struct btrfs_root *log = root->log_root;
+	const u64 ino = btrfs_ino(inode);
+	const u64 i_size = i_size_read(inode);
+
+	if (!btrfs_fs_incompat(root->fs_info, NO_HOLES))
+		return 0;
+
+	key.objectid = ino;
+	key.type = BTRFS_EXTENT_DATA_KEY;
+	key.offset = (u64)-1;
+
+	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
+	ASSERT(ret != 0);
+	if (ret < 0)
+		return ret;
+
+	ASSERT(path->slots[0] > 0);
+	path->slots[0]--;
+	leaf = path->nodes[0];
+	btrfs_item_key_to_cpu(leaf, &key, path->slots[0]);
+
+	if (key.objectid != ino || key.type != BTRFS_EXTENT_DATA_KEY) {
+		/* inode does not have any extents */
+		hole_start = 0;
+		hole_size = i_size;
+	} else {
+		struct btrfs_file_extent_item *extent;
+		u64 len;
+
+		/*
+		 * If there's an extent beyond i_size, an explicit hole was
+		 * already inserted by copy_items().
+		 */
+		if (key.offset >= i_size)
+			return 0;
+
+		extent = btrfs_item_ptr(leaf, path->slots[0],
+					struct btrfs_file_extent_item);
+
+		if (btrfs_file_extent_type(leaf, extent) ==
+		    BTRFS_FILE_EXTENT_INLINE) {
+			len = btrfs_file_extent_inline_len(leaf,
+							   path->slots[0],
+							   extent);
+			ASSERT(len == i_size);
+			return 0;
+		}
+
+		len = btrfs_file_extent_num_bytes(leaf, extent);
+		/* Last extent goes beyond i_size, no need to log a hole. */
+		if (key.offset + len > i_size)
+			return 0;
+		hole_start = key.offset + len;
+		hole_size = i_size - hole_start;
+	}
+	btrfs_release_path(path);
+
+	/* Last extent ends at i_size. */
+	if (hole_size == 0)
+		return 0;
+
+	hole_size = ALIGN(hole_size, root->sectorsize);
+	ret = btrfs_insert_file_extent(trans, log, ino, hole_start, 0, 0,
+				       hole_size, 0, hole_size, 0, 0, 0);
+	return ret;
+}
+
 /* log a single inode in the tree log.
  * At least one parent directory for this inode must exist in the tree
  * or be logged already.
@@ -4295,6 +4470,25 @@ again:
 		if (min_key.type == BTRFS_INODE_ITEM_KEY)
 			need_log_inode_item = false;
 
+		/* Skip xattrs, we log them later with btrfs_log_all_xattrs() */
+		if (min_key.type == BTRFS_XATTR_ITEM_KEY) {
+			if (ins_nr == 0)
+				goto next_slot;
+			ret = copy_items(trans, inode, dst_path, path,
+					 &last_extent, ins_start_slot,
+					 ins_nr, inode_only, logged_isize);
+			if (ret < 0) {
+				err = ret;
+				goto out_unlock;
+			}
+			ins_nr = 0;
+			if (ret) {
+				btrfs_release_path(path);
+				continue;
+			}
+			goto next_slot;
+		}
+
 		src = path->nodes[0];
 		if (ins_nr && ins_start_slot + ins_nr == path->slots[0]) {
 			ins_nr++;
@@ -4362,6 +4556,18 @@ next_slot:
 		ins_nr = 0;
 	}
 
+	btrfs_release_path(path);
+	btrfs_release_path(dst_path);
+	err = btrfs_log_all_xattrs(trans, root, inode, path, dst_path);
+	if (err)
+		goto out_unlock;
+	if (max_key.type >= BTRFS_EXTENT_DATA_KEY && !fast_search) {
+		btrfs_release_path(path);
+		btrfs_release_path(dst_path);
+		err = btrfs_log_trailing_hole(trans, root, inode, path);
+		if (err)
+			goto out_unlock;
+	}
 log_extents:
 	btrfs_release_path(path);
 	btrfs_release_path(dst_path);
diff --git a/fs/btrfs/ulist.c b/fs/btrfs/ulist.c
index 840a38b27..91feb2bde 100644
--- a/fs/btrfs/ulist.c
+++ b/fs/btrfs/ulist.c
@@ -132,6 +132,15 @@ static struct ulist_node *ulist_rbtree_search(struct ulist *ulist, u64 val)
 	return NULL;
 }
 
+static void ulist_rbtree_erase(struct ulist *ulist, struct ulist_node *node)
+{
+	rb_erase(&node->rb_node, &ulist->root);
+	list_del(&node->list);
+	kfree(node);
+	BUG_ON(ulist->nnodes == 0);
+	ulist->nnodes--;
+}
+
 static int ulist_rbtree_insert(struct ulist *ulist, struct ulist_node *ins)
 {
 	struct rb_node **p = &ulist->root.rb_node;
@@ -197,9 +206,6 @@ int ulist_add_merge(struct ulist *ulist, u64 val, u64 aux,
 
 	node->val = val;
 	node->aux = aux;
-#ifdef CONFIG_BTRFS_DEBUG
-	node->seqnum = ulist->nnodes;
-#endif
 
 	ret = ulist_rbtree_insert(ulist, node);
 	ASSERT(!ret);
@@ -209,6 +215,33 @@ int ulist_add_merge(struct ulist *ulist, u64 val, u64 aux,
 	return 1;
 }
 
+/*
+ * ulist_del - delete one node from ulist
+ * @ulist:	ulist to remove node from
+ * @val:	value to delete
+ * @aux:	aux to delete
+ *
+ * The deletion will only be done when *BOTH* val and aux matches.
+ * Return 0 for successful delete.
+ * Return > 0 for not found.
+ */
+int ulist_del(struct ulist *ulist, u64 val, u64 aux)
+{
+	struct ulist_node *node;
+
+	node = ulist_rbtree_search(ulist, val);
+	/* Not found */
+	if (!node)
+		return 1;
+
+	if (node->aux != aux)
+		return 1;
+
+	/* Found and delete */
+	ulist_rbtree_erase(ulist, node);
+	return 0;
+}
+
 /**
  * ulist_next - iterate ulist
  * @ulist:	ulist to iterate
@@ -237,15 +270,7 @@ struct ulist_node *ulist_next(struct ulist *ulist, struct ulist_iterator *uiter)
 		uiter->cur_list = uiter->cur_list->next;
 	} else {
 		uiter->cur_list = ulist->nodes.next;
-#ifdef CONFIG_BTRFS_DEBUG
-		uiter->i = 0;
-#endif
 	}
 	node = list_entry(uiter->cur_list, struct ulist_node, list);
-#ifdef CONFIG_BTRFS_DEBUG
-	ASSERT(node->seqnum == uiter->i);
-	ASSERT(uiter->i >= 0 && uiter->i < ulist->nnodes);
-	uiter->i++;
-#endif
 	return node;
 }
diff --git a/fs/btrfs/ulist.h b/fs/btrfs/ulist.h
index 4c29db604..a01a2c458 100644
--- a/fs/btrfs/ulist.h
+++ b/fs/btrfs/ulist.h
@@ -57,6 +57,7 @@ void ulist_free(struct ulist *ulist);
 int ulist_add(struct ulist *ulist, u64 val, u64 aux, gfp_t gfp_mask);
 int ulist_add_merge(struct ulist *ulist, u64 val, u64 aux,
 		    u64 *old_aux, gfp_t gfp_mask);
+int ulist_del(struct ulist *ulist, u64 val, u64 aux);
 
 /* just like ulist_add_merge() but take a pointer for the aux data */
 static inline int ulist_add_merge_ptr(struct ulist *ulist, u64 val, void *aux,
diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c
index 174f5e1e0..fbe7c1045 100644
--- a/fs/btrfs/volumes.c
+++ b/fs/btrfs/volumes.c
@@ -52,6 +52,10 @@ static void btrfs_dev_stat_print_on_load(struct btrfs_device *device);
 
 DEFINE_MUTEX(uuid_mutex);
 static LIST_HEAD(fs_uuids);
+struct list_head *btrfs_get_fs_uuids(void)
+{
+	return &fs_uuids;
+}
 
 static struct btrfs_fs_devices *__alloc_fs_devices(void)
 {
@@ -345,7 +349,7 @@ loop_lock:
 		    waitqueue_active(&fs_info->async_submit_wait))
 			wake_up(&fs_info->async_submit_wait);
 
-		BUG_ON(atomic_read(&cur->bi_cnt) == 0);
+		BUG_ON(atomic_read(&cur->__bi_cnt) == 0);
 
 		/*
 		 * if we're doing the sync list, record that our
@@ -441,6 +445,61 @@ static void pending_bios_fn(struct btrfs_work *work)
 	run_scheduled_bios(device);
 }
 
+
+void btrfs_free_stale_device(struct btrfs_device *cur_dev)
+{
+	struct btrfs_fs_devices *fs_devs;
+	struct btrfs_device *dev;
+
+	if (!cur_dev->name)
+		return;
+
+	list_for_each_entry(fs_devs, &fs_uuids, list) {
+		int del = 1;
+
+		if (fs_devs->opened)
+			continue;
+		if (fs_devs->seeding)
+			continue;
+
+		list_for_each_entry(dev, &fs_devs->devices, dev_list) {
+
+			if (dev == cur_dev)
+				continue;
+			if (!dev->name)
+				continue;
+
+			/*
+			 * Todo: This won't be enough. What if the same device
+			 * comes back (with new uuid and) with its mapper path?
+			 * But for now, this does help as mostly an admin will
+			 * either use mapper or non mapper path throughout.
+			 */
+			rcu_read_lock();
+			del = strcmp(rcu_str_deref(dev->name),
+						rcu_str_deref(cur_dev->name));
+			rcu_read_unlock();
+			if (!del)
+				break;
+		}
+
+		if (!del) {
+			/* delete the stale device */
+			if (fs_devs->num_devices == 1) {
+				btrfs_sysfs_remove_fsid(fs_devs);
+				list_del(&fs_devs->list);
+				free_fs_devices(fs_devs);
+			} else {
+				fs_devs->num_devices--;
+				list_del(&dev->dev_list);
+				rcu_string_free(dev->name);
+				kfree(dev);
+			}
+			break;
+		}
+	}
+}
+
 /*
  * Add new device to list of registered devices
  *
@@ -556,6 +615,12 @@ static noinline int device_list_add(const char *path,
 	if (!fs_devices->opened)
 		device->generation = found_transid;
 
+	/*
+	 * if there is new btrfs on an already registered device,
+	 * then remove the stale device entry.
+	 */
+	btrfs_free_stale_device(device);
+
 	*fs_devices_ret = fs_devices;
 
 	return ret;
@@ -693,13 +758,13 @@ static void free_device(struct rcu_head *head)
 
 static int __btrfs_close_devices(struct btrfs_fs_devices *fs_devices)
 {
-	struct btrfs_device *device;
+	struct btrfs_device *device, *tmp;
 
 	if (--fs_devices->opened > 0)
 		return 0;
 
 	mutex_lock(&fs_devices->device_list_mutex);
-	list_for_each_entry(device, &fs_devices->devices, dev_list) {
+	list_for_each_entry_safe(device, tmp, &fs_devices->devices, dev_list) {
 		struct btrfs_device *new_device;
 		struct rcu_string *name;
 
@@ -1067,15 +1132,31 @@ again:
 
 		map = (struct map_lookup *)em->bdev;
 		for (i = 0; i < map->num_stripes; i++) {
+			u64 end;
+
 			if (map->stripes[i].dev != device)
 				continue;
 			if (map->stripes[i].physical >= physical_start + len ||
 			    map->stripes[i].physical + em->orig_block_len <=
 			    physical_start)
 				continue;
-			*start = map->stripes[i].physical +
-				em->orig_block_len;
-			ret = 1;
+			/*
+			 * Make sure that while processing the pinned list we do
+			 * not override our *start with a lower value, because
+			 * we can have pinned chunks that fall within this
+			 * device hole and that have lower physical addresses
+			 * than the pending chunks we processed before. If we
+			 * do not take this special care we can end up getting
+			 * 2 pending chunks that start at the same physical
+			 * device offsets because the end offset of a pinned
+			 * chunk can be equal to the start offset of some
+			 * pending chunk.
+			 */
+			end = map->stripes[i].physical + em->orig_block_len;
+			if (end > *start) {
+				*start = end;
+				ret = 1;
+			}
 		}
 	}
 	if (search_list == &trans->transaction->pending_chunks) {
@@ -1706,7 +1787,7 @@ int btrfs_rm_device(struct btrfs_root *root, char *device_path)
 	if (device->bdev) {
 		device->fs_devices->open_devices--;
 		/* remove sysfs entry */
-		btrfs_kobj_rm_device(root->fs_info, device);
+		btrfs_kobj_rm_device(root->fs_info->fs_devices, device);
 	}
 
 	call_rcu(&device->rcu, free_device);
@@ -1875,6 +1956,9 @@ void btrfs_destroy_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
 	mutex_lock(&uuid_mutex);
 	WARN_ON(!tgtdev);
 	mutex_lock(&fs_info->fs_devices->device_list_mutex);
+
+	btrfs_kobj_rm_device(fs_info->fs_devices, tgtdev);
+
 	if (tgtdev->bdev) {
 		btrfs_scratch_superblock(tgtdev);
 		fs_info->fs_devices->open_devices--;
@@ -2211,7 +2295,7 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
 				    tmp + 1);
 
 	/* add sysfs device entry */
-	btrfs_kobj_add_device(root->fs_info, device);
+	btrfs_kobj_add_device(root->fs_info->fs_devices, device);
 
 	/*
 	 * we've got more storage, clear any full flags on the space
@@ -2252,8 +2336,9 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
 		 */
 		snprintf(fsid_buf, BTRFS_UUID_UNPARSED_SIZE, "%pU",
 						root->fs_info->fsid);
-		if (kobject_rename(&root->fs_info->super_kobj, fsid_buf))
-			goto error_trans;
+		if (kobject_rename(&root->fs_info->fs_devices->super_kobj,
+								fsid_buf))
+			pr_warn("BTRFS: sysfs: failed to create fsid for sprout\n");
 	}
 
 	root->fs_info->num_tolerated_disk_barrier_failures =
@@ -2289,7 +2374,7 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path)
 error_trans:
 	btrfs_end_transaction(trans, root);
 	rcu_string_free(device->name);
-	btrfs_kobj_rm_device(root->fs_info, device);
+	btrfs_kobj_rm_device(root->fs_info->fs_devices, device);
 	kfree(device);
 error:
 	blkdev_put(bdev, FMODE_EXCL);
@@ -2609,6 +2694,9 @@ int btrfs_remove_chunk(struct btrfs_trans_handle *trans,
 		return -EINVAL;
 	}
 	map = (struct map_lookup *)em->bdev;
+	lock_chunks(root->fs_info->chunk_root);
+	check_system_chunk(trans, extent_root, map->type);
+	unlock_chunks(root->fs_info->chunk_root);
 
 	for (i = 0; i < map->num_stripes; i++) {
 		struct btrfs_device *device = map->stripes[i].dev;
@@ -2678,6 +2766,20 @@ static int btrfs_relocate_chunk(struct btrfs_root *root,
 	root = root->fs_info->chunk_root;
 	extent_root = root->fs_info->extent_root;
 
+	/*
+	 * Prevent races with automatic removal of unused block groups.
+	 * After we relocate and before we remove the chunk with offset
+	 * chunk_offset, automatic removal of the block group can kick in,
+	 * resulting in a failure when calling btrfs_remove_chunk() below.
+	 *
+	 * Make sure to acquire this mutex before doing a tree search (dev
+	 * or chunk trees) to find chunks. Otherwise the cleaner kthread might
+	 * call btrfs_remove_chunk() (through btrfs_delete_unused_bgs()) after
+	 * we release the path used to search the chunk/dev tree and before
+	 * the current task acquires this mutex and calls us.
+	 */
+	ASSERT(mutex_is_locked(&root->fs_info->delete_unused_bgs_mutex));
+
 	ret = btrfs_can_relocate(extent_root, chunk_offset);
 	if (ret)
 		return -ENOSPC;
@@ -2726,13 +2828,18 @@ again:
 	key.type = BTRFS_CHUNK_ITEM_KEY;
 
 	while (1) {
+		mutex_lock(&root->fs_info->delete_unused_bgs_mutex);
 		ret = btrfs_search_slot(NULL, chunk_root, &key, path, 0, 0);
-		if (ret < 0)
+		if (ret < 0) {
+			mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 			goto error;
+		}
 		BUG_ON(ret == 0); /* Corruption */
 
 		ret = btrfs_previous_item(chunk_root, path, key.objectid,
 					  key.type);
+		if (ret)
+			mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 		if (ret < 0)
 			goto error;
 		if (ret > 0)
@@ -2755,6 +2862,7 @@ again:
 			else
 				BUG_ON(ret);
 		}
+		mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 
 		if (found_key.offset == 0)
 			break;
@@ -3211,9 +3319,12 @@ again:
 			goto error;
 		}
 
+		mutex_lock(&fs_info->delete_unused_bgs_mutex);
 		ret = btrfs_search_slot(NULL, chunk_root, &key, path, 0, 0);
-		if (ret < 0)
+		if (ret < 0) {
+			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 			goto error;
+		}
 
 		/*
 		 * this shouldn't happen, it means the last relocate
@@ -3225,6 +3336,7 @@ again:
 		ret = btrfs_previous_item(chunk_root, path, 0,
 					  BTRFS_CHUNK_ITEM_KEY);
 		if (ret) {
+			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 			ret = 0;
 			break;
 		}
@@ -3233,8 +3345,10 @@ again:
 		slot = path->slots[0];
 		btrfs_item_key_to_cpu(leaf, &found_key, slot);
 
-		if (found_key.objectid != key.objectid)
+		if (found_key.objectid != key.objectid) {
+			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 			break;
+		}
 
 		chunk = btrfs_item_ptr(leaf, slot, struct btrfs_chunk);
 
@@ -3247,10 +3361,13 @@ again:
 		ret = should_balance_chunk(chunk_root, leaf, chunk,
 					   found_key.offset);
 		btrfs_release_path(path);
-		if (!ret)
+		if (!ret) {
+			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 			goto loop;
+		}
 
 		if (counting) {
+			mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 			spin_lock(&fs_info->balance_lock);
 			bctl->stat.expected++;
 			spin_unlock(&fs_info->balance_lock);
@@ -3260,6 +3377,7 @@ again:
 		ret = btrfs_relocate_chunk(chunk_root,
 					   found_key.objectid,
 					   found_key.offset);
+		mutex_unlock(&fs_info->delete_unused_bgs_mutex);
 		if (ret && ret != -ENOSPC)
 			goto error;
 		if (ret == -ENOSPC) {
@@ -3908,9 +4026,9 @@ int btrfs_create_uuid_tree(struct btrfs_fs_info *fs_info)
 	uuid_root = btrfs_create_tree(trans, fs_info,
 				      BTRFS_UUID_TREE_OBJECTID);
 	if (IS_ERR(uuid_root)) {
-		btrfs_abort_transaction(trans, tree_root,
-					PTR_ERR(uuid_root));
-		return PTR_ERR(uuid_root);
+		ret = PTR_ERR(uuid_root);
+		btrfs_abort_transaction(trans, tree_root, ret);
+		return ret;
 	}
 
 	fs_info->uuid_root = uuid_root;
@@ -3965,6 +4083,7 @@ int btrfs_shrink_device(struct btrfs_device *device, u64 new_size)
 	int slot;
 	int failed = 0;
 	bool retried = false;
+	bool checked_pending_chunks = false;
 	struct extent_buffer *l;
 	struct btrfs_key key;
 	struct btrfs_super_block *super_copy = root->fs_info->super_copy;
@@ -3998,11 +4117,16 @@ again:
 	key.type = BTRFS_DEV_EXTENT_KEY;
 
 	do {
+		mutex_lock(&root->fs_info->delete_unused_bgs_mutex);
 		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
-		if (ret < 0)
+		if (ret < 0) {
+			mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 			goto done;
+		}
 
 		ret = btrfs_previous_item(root, path, 0, key.type);
+		if (ret)
+			mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 		if (ret < 0)
 			goto done;
 		if (ret) {
@@ -4016,6 +4140,7 @@ again:
 		btrfs_item_key_to_cpu(l, &key, path->slots[0]);
 
 		if (key.objectid != device->devid) {
+			mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 			btrfs_release_path(path);
 			break;
 		}
@@ -4024,6 +4149,7 @@ again:
 		length = btrfs_dev_extent_length(l, dev_extent);
 
 		if (key.offset + length <= new_size) {
+			mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 			btrfs_release_path(path);
 			break;
 		}
@@ -4033,6 +4159,7 @@ again:
 		btrfs_release_path(path);
 
 		ret = btrfs_relocate_chunk(root, chunk_objectid, chunk_offset);
+		mutex_unlock(&root->fs_info->delete_unused_bgs_mutex);
 		if (ret && ret != -ENOSPC)
 			goto done;
 		if (ret == -ENOSPC)
@@ -4045,15 +4172,6 @@ again:
 		goto again;
 	} else if (failed && retried) {
 		ret = -ENOSPC;
-		lock_chunks(root);
-
-		btrfs_device_set_total_bytes(device, old_size);
-		if (device->writeable)
-			device->fs_devices->total_rw_bytes += diff;
-		spin_lock(&root->fs_info->free_chunk_lock);
-		root->fs_info->free_chunk_space += diff;
-		spin_unlock(&root->fs_info->free_chunk_lock);
-		unlock_chunks(root);
 		goto done;
 	}
 
@@ -4065,6 +4183,35 @@ again:
 	}
 
 	lock_chunks(root);
+
+	/*
+	 * We checked in the above loop all device extents that were already in
+	 * the device tree. However before we have updated the device's
+	 * total_bytes to the new size, we might have had chunk allocations that
+	 * have not complete yet (new block groups attached to transaction
+	 * handles), and therefore their device extents were not yet in the
+	 * device tree and we missed them in the loop above. So if we have any
+	 * pending chunk using a device extent that overlaps the device range
+	 * that we can not use anymore, commit the current transaction and
+	 * repeat the search on the device tree - this way we guarantee we will
+	 * not have chunks using device extents that end beyond 'new_size'.
+	 */
+	if (!checked_pending_chunks) {
+		u64 start = new_size;
+		u64 len = old_size - new_size;
+
+		if (contains_pending_extent(trans, device, &start, len)) {
+			unlock_chunks(root);
+			checked_pending_chunks = true;
+			failed = 0;
+			retried = false;
+			ret = btrfs_commit_transaction(trans, root);
+			if (ret)
+				goto done;
+			goto again;
+		}
+	}
+
 	btrfs_device_set_disk_total_bytes(device, new_size);
 	if (list_empty(&device->resized_list))
 		list_add_tail(&device->resized_list,
@@ -4079,6 +4226,16 @@ again:
 	btrfs_end_transaction(trans, root);
 done:
 	btrfs_free_path(path);
+	if (ret) {
+		lock_chunks(root);
+		btrfs_device_set_total_bytes(device, old_size);
+		if (device->writeable)
+			device->fs_devices->total_rw_bytes += diff;
+		spin_lock(&root->fs_info->free_chunk_lock);
+		root->fs_info->free_chunk_space += diff;
+		spin_unlock(&root->fs_info->free_chunk_lock);
+		unlock_chunks(root);
+	}
 	return ret;
 }
 
@@ -5586,17 +5743,16 @@ int btrfs_rmap_block(struct btrfs_mapping_tree *map_tree,
 
 static inline void btrfs_end_bbio(struct btrfs_bio *bbio, struct bio *bio, int err)
 {
-	if (likely(bbio->flags & BTRFS_BIO_ORIG_BIO_SUBMITTED))
-		bio_endio_nodec(bio, err);
-	else
-		bio_endio(bio, err);
+	bio->bi_private = bbio->private;
+	bio->bi_end_io = bbio->end_io;
+	bio_endio(bio, err);
+
 	btrfs_put_bbio(bbio);
 }
 
 static void btrfs_end_bio(struct bio *bio, int err)
 {
 	struct btrfs_bio *bbio = bio->bi_private;
-	struct btrfs_device *dev = bbio->stripes[0].dev;
 	int is_orig_bio = 0;
 
 	if (err) {
@@ -5604,6 +5760,7 @@ static void btrfs_end_bio(struct bio *bio, int err)
 		if (err == -EIO || err == -EREMOTEIO) {
 			unsigned int stripe_index =
 				btrfs_io_bio(bio)->stripe_index;
+			struct btrfs_device *dev;
 
 			BUG_ON(stripe_index >= bbio->num_stripes);
 			dev = bbio->stripes[stripe_index].dev;
@@ -5633,8 +5790,6 @@ static void btrfs_end_bio(struct bio *bio, int err)
 			bio = bbio->orig_bio;
 		}
 
-		bio->bi_private = bbio->private;
-		bio->bi_end_io = bbio->end_io;
 		btrfs_io_bio(bio)->mirror_num = bbio->mirror_num;
 		/* only send an error to the higher layers if it is
 		 * beyond the tolerance of the btrfs bio
@@ -5816,8 +5971,6 @@ static void bbio_error(struct btrfs_bio *bbio, struct bio *bio, u64 logical)
 		/* Shoud be the original bio. */
 		WARN_ON(bio != bbio->orig_bio);
 
-		bio->bi_private = bbio->private;
-		bio->bi_end_io = bbio->end_io;
 		btrfs_io_bio(bio)->mirror_num = bbio->mirror_num;
 		bio->bi_iter.bi_sector = logical >> 9;
 
@@ -5898,10 +6051,8 @@ int btrfs_map_bio(struct btrfs_root *root, int rw, struct bio *bio,
 		if (dev_nr < total_devs - 1) {
 			bio = btrfs_bio_clone(first_bio, GFP_NOFS);
 			BUG_ON(!bio); /* -ENOMEM */
-		} else {
+		} else
 			bio = first_bio;
-			bbio->flags |= BTRFS_BIO_ORIG_BIO_SUBMITTED;
-		}
 
 		submit_stripe_bio(root, bbio, bio,
 				  bbio->stripes[dev_nr].physical, dev_nr, rw,
@@ -6078,6 +6229,8 @@ static int read_one_chunk(struct btrfs_root *root, struct btrfs_key *key,
 				free_extent_map(em);
 				return -EIO;
 			}
+			btrfs_warn(root->fs_info, "devid %llu uuid %pU is missing",
+						devid, uuid);
 		}
 		map->stripes[i].dev->in_fs_metadata = 1;
 	}
@@ -6197,10 +6350,11 @@ static int read_one_dev(struct btrfs_root *root,
 		if (!btrfs_test_opt(root, DEGRADED))
 			return -EIO;
 
-		btrfs_warn(root->fs_info, "devid %llu missing", devid);
 		device = add_missing_dev(root, fs_devices, devid, dev_uuid);
 		if (!device)
 			return -ENOMEM;
+		btrfs_warn(root->fs_info, "devid %llu uuid %pU missing",
+				devid, dev_uuid);
 	} else {
 		if (!device->bdev && !btrfs_test_opt(root, DEGRADED))
 			return -EIO;
@@ -6728,3 +6882,21 @@ void btrfs_update_commit_device_bytes_used(struct btrfs_root *root,
 	}
 	unlock_chunks(root);
 }
+
+void btrfs_set_fs_info_ptr(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
+	while (fs_devices) {
+		fs_devices->fs_info = fs_info;
+		fs_devices = fs_devices->seed;
+	}
+}
+
+void btrfs_reset_fs_info_ptr(struct btrfs_fs_info *fs_info)
+{
+	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
+	while (fs_devices) {
+		fs_devices->fs_info = NULL;
+		fs_devices = fs_devices->seed;
+	}
+}
diff --git a/fs/btrfs/volumes.h b/fs/btrfs/volumes.h
index ebc31331a..95842a909 100644
--- a/fs/btrfs/volumes.h
+++ b/fs/btrfs/volumes.h
@@ -253,6 +253,12 @@ struct btrfs_fs_devices {
 	 * nonrot flag set
 	 */
 	int rotating;
+
+	struct btrfs_fs_info *fs_info;
+	/* sysfs kobjects */
+	struct kobject super_kobj;
+	struct kobject *device_dir_kobj;
+	struct completion kobj_unregister;
 };
 
 #define BTRFS_BIO_INLINE_CSUM_SIZE	64
@@ -292,8 +298,6 @@ struct btrfs_bio_stripe {
 struct btrfs_bio;
 typedef void (btrfs_bio_end_io_t) (struct btrfs_bio *bio, int err);
 
-#define BTRFS_BIO_ORIG_BIO_SUBMITTED	(1 << 0)
-
 struct btrfs_bio {
 	atomic_t refs;
 	atomic_t stripes_pending;
@@ -537,5 +541,8 @@ static inline void unlock_chunks(struct btrfs_root *root)
 	mutex_unlock(&root->fs_info->chunk_mutex);
 }
 
+struct list_head *btrfs_get_fs_uuids(void);
+void btrfs_set_fs_info_ptr(struct btrfs_fs_info *fs_info);
+void btrfs_reset_fs_info_ptr(struct btrfs_fs_info *fs_info);
 
 #endif