Merge branch 'for-linus-4.12' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs

Pull btrfs updates from Chris Mason:
 "This has fixes and cleanups Dave Sterba collected for the merge
  window.

  The biggest functional fixes are between btrfs raid5/6 and scrub, and
  raid5/6 and device replacement. Some of our pending qgroup fixes are
  included as well while I bash on the rest in testing.

  We also have the usual set of cleanups, including one that makes
  __btrfs_map_block() much more maintainable, and conversions from
  atomic_t to refcount_t"

* 'for-linus-4.12' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs: (71 commits)
  btrfs: fix the gfp_mask for the reada_zones radix tree
  Btrfs: fix reported number of inode blocks
  Btrfs: send, fix file hole not being preserved due to inline extent
  Btrfs: fix extent map leak during fallocate error path
  Btrfs: fix incorrect space accounting after failure to insert inline extent
  Btrfs: fix invalid attempt to free reserved space on failure to cow range
  btrfs: Handle delalloc error correctly to avoid ordered extent hang
  btrfs: Fix metadata underflow caused by btrfs_reloc_clone_csum error
  btrfs: check if the device is flush capable
  btrfs: delete unused member nobarriers
  btrfs: scrub: Fix RAID56 recovery race condition
  btrfs: scrub: Introduce full stripe lock for RAID56
  btrfs: Use ktime_get_real_ts for root ctime
  Btrfs: handle only applicable errors returned by btrfs_get_extent
  btrfs: qgroup: Fix qgroup corruption caused by inode_cache mount option
  btrfs: use q which is already obtained from bdev_get_queue
  Btrfs: switch to div64_u64 if with a u64 divisor
  Btrfs: update scrub_parity to use u64 stripe_len
  Btrfs: enable repair during read for raid56 profile
  btrfs: use clear_page where appropriate
  ...

MAINTAINERS

@@ -2926,6 +2926,8 @@ T:	git git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs.git
 S:	Maintained
 F:	Documentation/filesystems/btrfs.txt
 F:	fs/btrfs/
+F:	include/linux/btrfs*
+F:	include/uapi/linux/btrfs*
 
 BTTV VIDEO4LINUX DRIVER
 M:	Mauro Carvalho Chehab <mchehab@s-opensource.com>

fs/btrfs/backref.c

@@ -26,6 +26,11 @@
 #include "delayed-ref.h"
 #include "locking.h"
 
+enum merge_mode {
+	MERGE_IDENTICAL_KEYS = 1,
+	MERGE_IDENTICAL_PARENTS,
+};
+
 /* Just an arbitrary number so we can be sure this happened */
 #define BACKREF_FOUND_SHARED 6
 
@@ -533,7 +538,7 @@ static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
 	 * slot==nritems. In that case, go to the next leaf before we continue.
 	 */
 	if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
-		if (time_seq == (u64)-1)
+		if (time_seq == SEQ_LAST)
 			ret = btrfs_next_leaf(root, path);
 		else
 			ret = btrfs_next_old_leaf(root, path, time_seq);
@@ -577,7 +582,7 @@ static int add_all_parents(struct btrfs_root *root, struct btrfs_path *path,
 			eie = NULL;
 		}
 next:
-		if (time_seq == (u64)-1)
+		if (time_seq == SEQ_LAST)
 			ret = btrfs_next_item(root, path);
 		else
 			ret = btrfs_next_old_item(root, path, time_seq);
@@ -629,7 +634,7 @@ 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)
+	else if (time_seq == SEQ_LAST)
 		root_level = btrfs_header_level(root->node);
 	else
 		root_level = btrfs_old_root_level(root, time_seq);
@@ -640,7 +645,7 @@ static int __resolve_indirect_ref(struct btrfs_fs_info *fs_info,
 	}
 
 	path->lowest_level = level;
-	if (time_seq == (u64)-1)
+	if (time_seq == SEQ_LAST)
 		ret = btrfs_search_slot(NULL, root, &ref->key_for_search, path,
 					0, 0);
 	else
@@ -809,14 +814,12 @@ static int __add_missing_keys(struct btrfs_fs_info *fs_info,
 /*
  * 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.
- *           additionally, we could even add a key range for the blocks we
- *           looked into to merge even more (-> replace unresolved refs by those
- *           having a parent).
- * mode = 2: merge identical parents
+ *    FIXME: For MERGE_IDENTICAL_KEYS, if we add more keys in __add_prelim_ref
+ *           then we can merge more here. Additionally, we could even add a key
+ *           range for the blocks we looked into to merge even more (-> replace
+ *           unresolved refs by those having a parent).
  */
-static void __merge_refs(struct list_head *head, int mode)
+static void __merge_refs(struct list_head *head, enum merge_mode mode)
 {
 	struct __prelim_ref *pos1;
 
@@ -829,7 +832,7 @@ static void __merge_refs(struct list_head *head, int mode)
 
 			if (!ref_for_same_block(ref1, ref2))
 				continue;
-			if (mode == 1) {
+			if (mode == MERGE_IDENTICAL_KEYS) {
 				if (!ref1->parent && ref2->parent)
 					swap(ref1, ref2);
 			} else {
@@ -1196,7 +1199,7 @@ 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
+ * If time_seq is set to SEQ_LAST, 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().
@@ -1243,7 +1246,7 @@ static int find_parent_nodes(struct btrfs_trans_handle *trans,
 		path->skip_locking = 1;
 	}
 
-	if (time_seq == (u64)-1)
+	if (time_seq == SEQ_LAST)
 		path->skip_locking = 1;
 
 	/*
@@ -1273,9 +1276,9 @@ again:
 
 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
 	if (trans && likely(trans->type != __TRANS_DUMMY) &&
-	    time_seq != (u64)-1) {
+	    time_seq != SEQ_LAST) {
 #else
-	if (trans && time_seq != (u64)-1) {
+	if (trans && time_seq != SEQ_LAST) {
 #endif
 		/*
 		 * look if there are updates for this ref queued and lock the
@@ -1286,7 +1289,7 @@ again:
 		head = btrfs_find_delayed_ref_head(delayed_refs, bytenr);
 		if (head) {
 			if (!mutex_trylock(&head->mutex)) {
-				atomic_inc(&head->node.refs);
+				refcount_inc(&head->node.refs);
 				spin_unlock(&delayed_refs->lock);
 
 				btrfs_release_path(path);
@@ -1374,7 +1377,7 @@ again:
 	if (ret)
 		goto out;
 
-	__merge_refs(&prefs, 1);
+	__merge_refs(&prefs, MERGE_IDENTICAL_KEYS);
 
 	ret = __resolve_indirect_refs(fs_info, path, time_seq, &prefs,
 				      extent_item_pos, total_refs,
@@ -1382,7 +1385,7 @@ again:
 	if (ret)
 		goto out;
 
-	__merge_refs(&prefs, 2);
+	__merge_refs(&prefs, MERGE_IDENTICAL_PARENTS);
 
 	while (!list_empty(&prefs)) {
 		ref = list_first_entry(&prefs, struct __prelim_ref, list);

fs/btrfs/btrfs_inode.h

@@ -124,6 +124,13 @@ struct btrfs_inode {
 	 */
 	u64 delalloc_bytes;
 
+	/*
+	 * Total number of bytes pending delalloc that fall within a file
+	 * range that is either a hole or beyond EOF (and no prealloc extent
+	 * exists in the range). This is always <= delalloc_bytes.
+	 */
+	u64 new_delalloc_bytes;
+
 	/*
 	 * total number of bytes pending defrag, used by stat to check whether
 	 * it needs COW.

fs/btrfs/compression.c

@@ -44,7 +44,7 @@
 
 struct compressed_bio {
 	/* number of bios pending for this compressed extent */
-	atomic_t pending_bios;
+	refcount_t pending_bios;
 
 	/* the pages with the compressed data on them */
 	struct page **compressed_pages;
@@ -161,7 +161,7 @@ static void end_compressed_bio_read(struct bio *bio)
 	/* if there are more bios still pending for this compressed
 	 * extent, just exit
 	 */
-	if (!atomic_dec_and_test(&cb->pending_bios))
+	if (!refcount_dec_and_test(&cb->pending_bios))
 		goto out;
 
 	inode = cb->inode;
@@ -274,7 +274,7 @@ static void end_compressed_bio_write(struct bio *bio)
 	/* if there are more bios still pending for this compressed
 	 * extent, just exit
 	 */
-	if (!atomic_dec_and_test(&cb->pending_bios))
+	if (!refcount_dec_and_test(&cb->pending_bios))
 		goto out;
 
 	/* ok, we're the last bio for this extent, step one is to
@@ -342,7 +342,7 @@ int btrfs_submit_compressed_write(struct inode *inode, u64 start,
 	cb = kmalloc(compressed_bio_size(fs_info, compressed_len), GFP_NOFS);
 	if (!cb)
 		return -ENOMEM;
-	atomic_set(&cb->pending_bios, 0);
+	refcount_set(&cb->pending_bios, 0);
 	cb->errors = 0;
 	cb->inode = inode;
 	cb->start = start;
@@ -363,7 +363,7 @@ int btrfs_submit_compressed_write(struct inode *inode, u64 start,
 	bio_set_op_attrs(bio, REQ_OP_WRITE, 0);
 	bio->bi_private = cb;
 	bio->bi_end_io = end_compressed_bio_write;
-	atomic_inc(&cb->pending_bios);
+	refcount_set(&cb->pending_bios, 1);
 
 	/* create and submit bios for the compressed pages */
 	bytes_left = compressed_len;
@@ -388,7 +388,7 @@ int btrfs_submit_compressed_write(struct inode *inode, u64 start,
 			 * we inc the count.  Otherwise, the cb might get
 			 * freed before we're done setting it up
 			 */
-			atomic_inc(&cb->pending_bios);
+			refcount_inc(&cb->pending_bios);
 			ret = btrfs_bio_wq_end_io(fs_info, bio,
 						  BTRFS_WQ_ENDIO_DATA);
 			BUG_ON(ret); /* -ENOMEM */
@@ -607,7 +607,7 @@ int btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
 	if (!cb)
 		goto out;
 
-	atomic_set(&cb->pending_bios, 0);
+	refcount_set(&cb->pending_bios, 0);
 	cb->errors = 0;
 	cb->inode = inode;
 	cb->mirror_num = mirror_num;
@@ -656,7 +656,7 @@ int btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
 	bio_set_op_attrs (comp_bio, REQ_OP_READ, 0);
 	comp_bio->bi_private = cb;
 	comp_bio->bi_end_io = end_compressed_bio_read;
-	atomic_inc(&cb->pending_bios);
+	refcount_set(&cb->pending_bios, 1);
 
 	for (pg_index = 0; pg_index < nr_pages; pg_index++) {
 		page = cb->compressed_pages[pg_index];
@@ -685,7 +685,7 @@ int btrfs_submit_compressed_read(struct inode *inode, struct bio *bio,
 			 * we inc the count.  Otherwise, the cb might get
 			 * freed before we're done setting it up
 			 */
-			atomic_inc(&cb->pending_bios);
+			refcount_inc(&cb->pending_bios);
 
 			if (!(BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM)) {
 				ret = btrfs_lookup_bio_sums(inode, comp_bio,

fs/btrfs/ctree.c

@@ -567,7 +567,7 @@ tree_mod_log_insert_key(struct btrfs_fs_info *fs_info,
 static noinline int
 tree_mod_log_insert_move(struct btrfs_fs_info *fs_info,
 			 struct extent_buffer *eb, int dst_slot, int src_slot,
-			 int nr_items, gfp_t flags)
+			 int nr_items)
 {
 	struct tree_mod_elem *tm = NULL;
 	struct tree_mod_elem **tm_list = NULL;
@@ -578,11 +578,11 @@ tree_mod_log_insert_move(struct btrfs_fs_info *fs_info,
 	if (!tree_mod_need_log(fs_info, eb))
 		return 0;
 
-	tm_list = kcalloc(nr_items, sizeof(struct tree_mod_elem *), flags);
+	tm_list = kcalloc(nr_items, sizeof(struct tree_mod_elem *), GFP_NOFS);
 	if (!tm_list)
 		return -ENOMEM;
 
-	tm = kzalloc(sizeof(*tm), flags);
+	tm = kzalloc(sizeof(*tm), GFP_NOFS);
 	if (!tm) {
 		ret = -ENOMEM;
 		goto free_tms;
@@ -596,7 +596,7 @@ tree_mod_log_insert_move(struct btrfs_fs_info *fs_info,
 
 	for (i = 0; i + dst_slot < src_slot && i < nr_items; i++) {
 		tm_list[i] = alloc_tree_mod_elem(eb, i + dst_slot,
-		    MOD_LOG_KEY_REMOVE_WHILE_MOVING, flags);
+		    MOD_LOG_KEY_REMOVE_WHILE_MOVING, GFP_NOFS);
 		if (!tm_list[i]) {
 			ret = -ENOMEM;
 			goto free_tms;
@@ -663,7 +663,7 @@ __tree_mod_log_free_eb(struct btrfs_fs_info *fs_info,
 static noinline int
 tree_mod_log_insert_root(struct btrfs_fs_info *fs_info,
 			 struct extent_buffer *old_root,
-			 struct extent_buffer *new_root, gfp_t flags,
+			 struct extent_buffer *new_root,
 			 int log_removal)
 {
 	struct tree_mod_elem *tm = NULL;
@@ -678,14 +678,14 @@ tree_mod_log_insert_root(struct btrfs_fs_info *fs_info,
 	if (log_removal && btrfs_header_level(old_root) > 0) {
 		nritems = btrfs_header_nritems(old_root);
 		tm_list = kcalloc(nritems, sizeof(struct tree_mod_elem *),
-				  flags);
+				  GFP_NOFS);
 		if (!tm_list) {
 			ret = -ENOMEM;
 			goto free_tms;
 		}
 		for (i = 0; i < nritems; i++) {
 			tm_list[i] = alloc_tree_mod_elem(old_root, i,
-			    MOD_LOG_KEY_REMOVE_WHILE_FREEING, flags);
+			    MOD_LOG_KEY_REMOVE_WHILE_FREEING, GFP_NOFS);
 			if (!tm_list[i]) {
 				ret = -ENOMEM;
 				goto free_tms;
@@ -693,7 +693,7 @@ tree_mod_log_insert_root(struct btrfs_fs_info *fs_info,
 		}
 	}
 
-	tm = kzalloc(sizeof(*tm), flags);
+	tm = kzalloc(sizeof(*tm), GFP_NOFS);
 	if (!tm) {
 		ret = -ENOMEM;
 		goto free_tms;
@@ -873,7 +873,7 @@ tree_mod_log_eb_move(struct btrfs_fs_info *fs_info, struct extent_buffer *dst,
 {
 	int ret;
 	ret = tree_mod_log_insert_move(fs_info, dst, dst_offset, src_offset,
-				       nr_items, GFP_NOFS);
+				       nr_items);
 	BUG_ON(ret < 0);
 }
 
@@ -943,7 +943,7 @@ tree_mod_log_set_root_pointer(struct btrfs_root *root,
 {
 	int ret;
 	ret = tree_mod_log_insert_root(root->fs_info, root->node,
-				       new_root_node, GFP_NOFS, log_removal);
+				       new_root_node, log_removal);
 	BUG_ON(ret < 0);
 }
 

fs/btrfs/ctree.h

@@ -39,6 +39,7 @@
 #include <linux/security.h>
 #include <linux/sizes.h>
 #include <linux/dynamic_debug.h>
+#include <linux/refcount.h>
 #include "extent_io.h"
 #include "extent_map.h"
 #include "async-thread.h"
@@ -518,7 +519,7 @@ struct btrfs_caching_control {
 	struct btrfs_work work;
 	struct btrfs_block_group_cache *block_group;
 	u64 progress;
-	atomic_t count;
+	refcount_t count;
 };
 
 /* Once caching_thread() finds this much free space, it will wake up waiters. */
@@ -538,6 +539,14 @@ struct btrfs_io_ctl {
 	unsigned check_crcs:1;
 };
 
+/*
+ * Tree to record all locked full stripes of a RAID5/6 block group
+ */
+struct btrfs_full_stripe_locks_tree {
+	struct rb_root root;
+	struct mutex lock;
+};
+
 struct btrfs_block_group_cache {
 	struct btrfs_key key;
 	struct btrfs_block_group_item item;
@@ -648,6 +657,9 @@ struct btrfs_block_group_cache {
 	 * Protected by free_space_lock.
 	 */
 	int needs_free_space;
+
+	/* Record locked full stripes for RAID5/6 block group */
+	struct btrfs_full_stripe_locks_tree full_stripe_locks_root;
 };
 
 /* delayed seq elem */
@@ -658,6 +670,8 @@ struct seq_list {
 
 #define SEQ_LIST_INIT(name)	{ .list = LIST_HEAD_INIT((name).list), .seq = 0 }
 
+#define SEQ_LAST	((u64)-1)
+
 enum btrfs_orphan_cleanup_state {
 	ORPHAN_CLEANUP_STARTED	= 1,
 	ORPHAN_CLEANUP_DONE	= 2,
@@ -702,6 +716,11 @@ struct btrfs_delayed_root;
 #define BTRFS_FS_BTREE_ERR			11
 #define BTRFS_FS_LOG1_ERR			12
 #define BTRFS_FS_LOG2_ERR			13
+/*
+ * Indicate that a whole-filesystem exclusive operation is running
+ * (device replace, resize, device add/delete, balance)
+ */
+#define BTRFS_FS_EXCL_OP			14
 
 struct btrfs_fs_info {
 	u8 fsid[BTRFS_FSID_SIZE];
@@ -1066,8 +1085,6 @@ struct btrfs_fs_info {
 	/* device replace state */
 	struct btrfs_dev_replace dev_replace;
 
-	atomic_t mutually_exclusive_operation_running;
-
 	struct percpu_counter bio_counter;
 	wait_queue_head_t replace_wait;
 
@@ -1220,7 +1237,7 @@ struct btrfs_root {
 	dev_t anon_dev;
 
 	spinlock_t root_item_lock;
-	atomic_t refs;
+	refcount_t refs;
 
 	struct mutex delalloc_mutex;
 	spinlock_t delalloc_lock;
@@ -3646,6 +3663,12 @@ int btrfs_scrub_cancel_dev(struct btrfs_fs_info *info,
 			   struct btrfs_device *dev);
 int btrfs_scrub_progress(struct btrfs_fs_info *fs_info, u64 devid,
 			 struct btrfs_scrub_progress *progress);
+static inline void btrfs_init_full_stripe_locks_tree(
+			struct btrfs_full_stripe_locks_tree *locks_root)
+{
+	locks_root->root = RB_ROOT;
+	mutex_init(&locks_root->lock);
+}
 
 /* dev-replace.c */
 void btrfs_bio_counter_inc_blocked(struct btrfs_fs_info *fs_info);
@@ -3670,8 +3693,7 @@ struct reada_control *btrfs_reada_add(struct btrfs_root *root,
 			      struct btrfs_key *start, struct btrfs_key *end);
 int btrfs_reada_wait(void *handle);
 void btrfs_reada_detach(void *handle);
-int btree_readahead_hook(struct btrfs_fs_info *fs_info,
-			 struct extent_buffer *eb, int err);
+int btree_readahead_hook(struct extent_buffer *eb, int err);
 
 static inline int is_fstree(u64 rootid)
 {

fs/btrfs/delayed-inode.c

@@ -52,7 +52,7 @@ static inline void btrfs_init_delayed_node(
 {
 	delayed_node->root = root;
 	delayed_node->inode_id = inode_id;
-	atomic_set(&delayed_node->refs, 0);
+	refcount_set(&delayed_node->refs, 0);
 	delayed_node->ins_root = RB_ROOT;
 	delayed_node->del_root = RB_ROOT;
 	mutex_init(&delayed_node->mutex);
@@ -81,7 +81,7 @@ static struct btrfs_delayed_node *btrfs_get_delayed_node(
 
 	node = READ_ONCE(btrfs_inode->delayed_node);
 	if (node) {
-		atomic_inc(&node->refs);
+		refcount_inc(&node->refs);
 		return node;
 	}
 
@@ -89,14 +89,14 @@ static struct btrfs_delayed_node *btrfs_get_delayed_node(
 	node = radix_tree_lookup(&root->delayed_nodes_tree, ino);
 	if (node) {
 		if (btrfs_inode->delayed_node) {
-			atomic_inc(&node->refs);	/* can be accessed */
+			refcount_inc(&node->refs);	/* can be accessed */
 			BUG_ON(btrfs_inode->delayed_node != node);
 			spin_unlock(&root->inode_lock);
 			return node;
 		}
 		btrfs_inode->delayed_node = node;
 		/* can be accessed and cached in the inode */
-		atomic_add(2, &node->refs);
+		refcount_add(2, &node->refs);
 		spin_unlock(&root->inode_lock);
 		return node;
 	}
@@ -125,7 +125,7 @@ again:
 	btrfs_init_delayed_node(node, root, ino);
 
 	/* cached in the btrfs inode and can be accessed */
-	atomic_add(2, &node->refs);
+	refcount_set(&node->refs, 2);
 
 	ret = radix_tree_preload(GFP_NOFS);
 	if (ret) {
@@ -166,7 +166,7 @@ static void btrfs_queue_delayed_node(struct btrfs_delayed_root *root,
 	} else {
 		list_add_tail(&node->n_list, &root->node_list);
 		list_add_tail(&node->p_list, &root->prepare_list);
-		atomic_inc(&node->refs);	/* inserted into list */
+		refcount_inc(&node->refs);	/* inserted into list */
 		root->nodes++;
 		set_bit(BTRFS_DELAYED_NODE_IN_LIST, &node->flags);
 	}
@@ -180,7 +180,7 @@ static void btrfs_dequeue_delayed_node(struct btrfs_delayed_root *root,
 	spin_lock(&root->lock);
 	if (test_bit(BTRFS_DELAYED_NODE_IN_LIST, &node->flags)) {
 		root->nodes--;
-		atomic_dec(&node->refs);	/* not in the list */
+		refcount_dec(&node->refs);	/* not in the list */
 		list_del_init(&node->n_list);
 		if (!list_empty(&node->p_list))
 			list_del_init(&node->p_list);
@@ -201,7 +201,7 @@ static struct btrfs_delayed_node *btrfs_first_delayed_node(
 
 	p = delayed_root->node_list.next;
 	node = list_entry(p, struct btrfs_delayed_node, n_list);
-	atomic_inc(&node->refs);
+	refcount_inc(&node->refs);
 out:
 	spin_unlock(&delayed_root->lock);
 
@@ -228,7 +228,7 @@ static struct btrfs_delayed_node *btrfs_next_delayed_node(
 		p = node->n_list.next;
 
 	next = list_entry(p, struct btrfs_delayed_node, n_list);
-	atomic_inc(&next->refs);
+	refcount_inc(&next->refs);
 out:
 	spin_unlock(&delayed_root->lock);
 
@@ -253,11 +253,11 @@ static void __btrfs_release_delayed_node(
 		btrfs_dequeue_delayed_node(delayed_root, delayed_node);
 	mutex_unlock(&delayed_node->mutex);
 
-	if (atomic_dec_and_test(&delayed_node->refs)) {
+	if (refcount_dec_and_test(&delayed_node->refs)) {
 		bool free = false;
 		struct btrfs_root *root = delayed_node->root;
 		spin_lock(&root->inode_lock);
-		if (atomic_read(&delayed_node->refs) == 0) {
+		if (refcount_read(&delayed_node->refs) == 0) {
 			radix_tree_delete(&root->delayed_nodes_tree,
 					  delayed_node->inode_id);
 			free = true;
@@ -286,7 +286,7 @@ static struct btrfs_delayed_node *btrfs_first_prepared_delayed_node(
 	p = delayed_root->prepare_list.next;
 	list_del_init(p);
 	node = list_entry(p, struct btrfs_delayed_node, p_list);
-	atomic_inc(&node->refs);
+	refcount_inc(&node->refs);
 out:
 	spin_unlock(&delayed_root->lock);
 
@@ -308,7 +308,7 @@ static struct btrfs_delayed_item *btrfs_alloc_delayed_item(u32 data_len)
 		item->ins_or_del = 0;
 		item->bytes_reserved = 0;
 		item->delayed_node = NULL;
-		atomic_set(&item->refs, 1);
+		refcount_set(&item->refs, 1);
 	}
 	return item;
 }
@@ -483,7 +483,7 @@ static void btrfs_release_delayed_item(struct btrfs_delayed_item *item)
 {
 	if (item) {
 		__btrfs_remove_delayed_item(item);
-		if (atomic_dec_and_test(&item->refs))
+		if (refcount_dec_and_test(&item->refs))
 			kfree(item);
 	}
 }
@@ -1600,14 +1600,14 @@ bool btrfs_readdir_get_delayed_items(struct inode *inode,
 	mutex_lock(&delayed_node->mutex);
 	item = __btrfs_first_delayed_insertion_item(delayed_node);
 	while (item) {
-		atomic_inc(&item->refs);
+		refcount_inc(&item->refs);
 		list_add_tail(&item->readdir_list, ins_list);
 		item = __btrfs_next_delayed_item(item);
 	}
 
 	item = __btrfs_first_delayed_deletion_item(delayed_node);
 	while (item) {
-		atomic_inc(&item->refs);
+		refcount_inc(&item->refs);
 		list_add_tail(&item->readdir_list, del_list);
 		item = __btrfs_next_delayed_item(item);
 	}
@@ -1621,7 +1621,7 @@ bool btrfs_readdir_get_delayed_items(struct inode *inode,
 	 * insert/delete delayed items in this period. So we also needn't
 	 * requeue or dequeue this delayed node.
 	 */
-	atomic_dec(&delayed_node->refs);
+	refcount_dec(&delayed_node->refs);
 
 	return true;
 }
@@ -1634,13 +1634,13 @@ void btrfs_readdir_put_delayed_items(struct inode *inode,
 
 	list_for_each_entry_safe(curr, next, ins_list, readdir_list) {
 		list_del(&curr->readdir_list);
-		if (atomic_dec_and_test(&curr->refs))
+		if (refcount_dec_and_test(&curr->refs))
 			kfree(curr);
 	}
 
 	list_for_each_entry_safe(curr, next, del_list, readdir_list) {
 		list_del(&curr->readdir_list);
-		if (atomic_dec_and_test(&curr->refs))
+		if (refcount_dec_and_test(&curr->refs))
 			kfree(curr);
 	}
 
@@ -1667,7 +1667,7 @@ int btrfs_should_delete_dir_index(struct list_head *del_list,
 		list_del(&curr->readdir_list);
 		ret = (curr->key.offset == index);
 
-		if (atomic_dec_and_test(&curr->refs))
+		if (refcount_dec_and_test(&curr->refs))
 			kfree(curr);
 
 		if (ret)
@@ -1705,7 +1705,7 @@ int btrfs_readdir_delayed_dir_index(struct dir_context *ctx,
 		list_del(&curr->readdir_list);
 
 		if (curr->key.offset < ctx->pos) {
-			if (atomic_dec_and_test(&curr->refs))
+			if (refcount_dec_and_test(&curr->refs))
 				kfree(curr);
 			continue;
 		}
@@ -1722,7 +1722,7 @@ int btrfs_readdir_delayed_dir_index(struct dir_context *ctx,
 		over = !dir_emit(ctx, name, name_len,
 			       location.objectid, d_type);
 
-		if (atomic_dec_and_test(&curr->refs))
+		if (refcount_dec_and_test(&curr->refs))
 			kfree(curr);
 
 		if (over)
@@ -1963,7 +1963,7 @@ void btrfs_kill_all_delayed_nodes(struct btrfs_root *root)
 		inode_id = delayed_nodes[n - 1]->inode_id + 1;
 
 		for (i = 0; i < n; i++)
-			atomic_inc(&delayed_nodes[i]->refs);
+			refcount_inc(&delayed_nodes[i]->refs);
 		spin_unlock(&root->inode_lock);
 
 		for (i = 0; i < n; i++) {

fs/btrfs/delayed-inode.h

@@ -26,7 +26,7 @@
 #include <linux/list.h>
 #include <linux/wait.h>
 #include <linux/atomic.h>
-
+#include <linux/refcount.h>
 #include "ctree.h"
 
 /* types of the delayed item */
@@ -67,7 +67,7 @@ struct btrfs_delayed_node {
 	struct rb_root del_root;
 	struct mutex mutex;
 	struct btrfs_inode_item inode_item;
-	atomic_t refs;
+	refcount_t refs;
 	u64 index_cnt;
 	unsigned long flags;
 	int count;
@@ -80,7 +80,7 @@ struct btrfs_delayed_item {
 	struct list_head readdir_list;	/* used for readdir items */
 	u64 bytes_reserved;
 	struct btrfs_delayed_node *delayed_node;
-	atomic_t refs;
+	refcount_t refs;
 	int ins_or_del;
 	u32 data_len;
 	char data[0];

fs/btrfs/delayed-ref.c

@@ -164,7 +164,7 @@ int btrfs_delayed_ref_lock(struct btrfs_trans_handle *trans,
 	if (mutex_trylock(&head->mutex))
 		return 0;
 
-	atomic_inc(&head->node.refs);
+	refcount_inc(&head->node.refs);
 	spin_unlock(&delayed_refs->lock);
 
 	mutex_lock(&head->mutex);
@@ -590,7 +590,7 @@ add_delayed_ref_head(struct btrfs_fs_info *fs_info,
 	delayed_refs = &trans->transaction->delayed_refs;
 
 	/* first set the basic ref node struct up */
-	atomic_set(&ref->refs, 1);
+	refcount_set(&ref->refs, 1);
 	ref->bytenr = bytenr;
 	ref->num_bytes = num_bytes;
 	ref->ref_mod = count_mod;
@@ -682,7 +682,7 @@ add_delayed_tree_ref(struct btrfs_fs_info *fs_info,
 	delayed_refs = &trans->transaction->delayed_refs;
 
 	/* first set the basic ref node struct up */
-	atomic_set(&ref->refs, 1);
+	refcount_set(&ref->refs, 1);
 	ref->bytenr = bytenr;
 	ref->num_bytes = num_bytes;
 	ref->ref_mod = 1;
@@ -739,7 +739,7 @@ add_delayed_data_ref(struct btrfs_fs_info *fs_info,
 		seq = atomic64_read(&fs_info->tree_mod_seq);
 
 	/* first set the basic ref node struct up */
-	atomic_set(&ref->refs, 1);
+	refcount_set(&ref->refs, 1);
 	ref->bytenr = bytenr;
 	ref->num_bytes = num_bytes;
 	ref->ref_mod = 1;

fs/btrfs/delayed-ref.h

@@ -18,6 +18,8 @@
 #ifndef __DELAYED_REF__
 #define __DELAYED_REF__
 
+#include <linux/refcount.h>
+
 /* these are the possible values of struct btrfs_delayed_ref_node->action */
 #define BTRFS_ADD_DELAYED_REF    1 /* add one backref to the tree */
 #define BTRFS_DROP_DELAYED_REF   2 /* delete one backref from the tree */
@@ -53,7 +55,7 @@ struct btrfs_delayed_ref_node {
 	u64 seq;
 
 	/* ref count on this data structure */
-	atomic_t refs;
+	refcount_t refs;
 
 	/*
 	 * how many refs is this entry adding or deleting.  For
@@ -220,8 +222,8 @@ btrfs_free_delayed_extent_op(struct btrfs_delayed_extent_op *op)
 
 static inline void btrfs_put_delayed_ref(struct btrfs_delayed_ref_node *ref)
 {
-	WARN_ON(atomic_read(&ref->refs) == 0);
-	if (atomic_dec_and_test(&ref->refs)) {
+	WARN_ON(refcount_read(&ref->refs) == 0);
+	if (refcount_dec_and_test(&ref->refs)) {
 		WARN_ON(ref->in_tree);
 		switch (ref->type) {
 		case BTRFS_TREE_BLOCK_REF_KEY:

fs/btrfs/dev-replace.c

@@ -546,8 +546,10 @@ static int btrfs_dev_replace_finishing(struct btrfs_fs_info *fs_info,
 		mutex_unlock(&fs_info->chunk_mutex);
 		mutex_unlock(&fs_info->fs_devices->device_list_mutex);
 		mutex_unlock(&uuid_mutex);
+		btrfs_rm_dev_replace_blocked(fs_info);
 		if (tgt_device)
 			btrfs_destroy_dev_replace_tgtdev(fs_info, tgt_device);
+		btrfs_rm_dev_replace_unblocked(fs_info);
 		mutex_unlock(&dev_replace->lock_finishing_cancel_unmount);
 
 		return scrub_ret;
@@ -665,7 +667,7 @@ void btrfs_dev_replace_status(struct btrfs_fs_info *fs_info,
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_STARTED:
 	case BTRFS_IOCTL_DEV_REPLACE_STATE_SUSPENDED:
 		srcdev = dev_replace->srcdev;
-		args->status.progress_1000 = div_u64(dev_replace->cursor_left,
+		args->status.progress_1000 = div64_u64(dev_replace->cursor_left,
 			div_u64(btrfs_device_get_total_bytes(srcdev), 1000));
 		break;
 	}
@@ -784,8 +786,7 @@ int btrfs_resume_dev_replace_async(struct btrfs_fs_info *fs_info)
 	}
 	btrfs_dev_replace_unlock(dev_replace, 1);
 
-	WARN_ON(atomic_xchg(
-		&fs_info->mutually_exclusive_operation_running, 1));
+	WARN_ON(test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags));
 	task = kthread_run(btrfs_dev_replace_kthread, fs_info, "btrfs-devrepl");
 	return PTR_ERR_OR_ZERO(task);
 }
@@ -814,7 +815,7 @@ static int btrfs_dev_replace_kthread(void *data)
 			(unsigned int)progress);
 	}
 	btrfs_dev_replace_continue_on_mount(fs_info);
-	atomic_set(&fs_info->mutually_exclusive_operation_running, 0);
+	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
 
 	return 0;
 }

fs/btrfs/disk-io.c

@@ -762,7 +762,7 @@ static int btree_readpage_end_io_hook(struct btrfs_io_bio *io_bio,
 err:
 	if (reads_done &&
 	    test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
-		btree_readahead_hook(fs_info, eb, ret);
+		btree_readahead_hook(eb, ret);
 
 	if (ret) {
 		/*
@@ -787,7 +787,7 @@ static int btree_io_failed_hook(struct page *page, int failed_mirror)
 	eb->read_mirror = failed_mirror;
 	atomic_dec(&eb->io_pages);
 	if (test_and_clear_bit(EXTENT_BUFFER_READAHEAD, &eb->bflags))
-		btree_readahead_hook(eb->fs_info, eb, -EIO);
+		btree_readahead_hook(eb, -EIO);
 	return -EIO;	/* we fixed nothing */
 }
 
@@ -1340,7 +1340,7 @@ static void __setup_root(struct btrfs_root *root, struct btrfs_fs_info *fs_info,
 	atomic_set(&root->log_writers, 0);
 	atomic_set(&root->log_batch, 0);
 	atomic_set(&root->orphan_inodes, 0);
-	atomic_set(&root->refs, 1);
+	refcount_set(&root->refs, 1);
 	atomic_set(&root->will_be_snapshoted, 0);
 	atomic64_set(&root->qgroup_meta_rsv, 0);
 	root->log_transid = 0;
@@ -3497,10 +3497,11 @@ static void btrfs_end_empty_barrier(struct bio *bio)
  */
 static int write_dev_flush(struct btrfs_device *device, int wait)
 {
+	struct request_queue *q = bdev_get_queue(device->bdev);
 	struct bio *bio;
 	int ret = 0;
 
-	if (device->nobarriers)
+	if (!test_bit(QUEUE_FLAG_WC, &q->queue_flags))
 		return 0;
 
 	if (wait) {
@@ -4321,7 +4322,7 @@ static int btrfs_destroy_delayed_refs(struct btrfs_transaction *trans,
 		head = rb_entry(node, struct btrfs_delayed_ref_head,
 				href_node);
 		if (!mutex_trylock(&head->mutex)) {
-			atomic_inc(&head->node.refs);
+			refcount_inc(&head->node.refs);
 			spin_unlock(&delayed_refs->lock);
 
 			mutex_lock(&head->mutex);
@@ -4593,7 +4594,7 @@ static int btrfs_cleanup_transaction(struct btrfs_fs_info *fs_info)
 		t = list_first_entry(&fs_info->trans_list,
 				     struct btrfs_transaction, list);
 		if (t->state >= TRANS_STATE_COMMIT_START) {
-			atomic_inc(&t->use_count);
+			refcount_inc(&t->use_count);
 			spin_unlock(&fs_info->trans_lock);
 			btrfs_wait_for_commit(fs_info, t->transid);
 			btrfs_put_transaction(t);

fs/btrfs/disk-io.h

@@ -101,14 +101,14 @@ struct btrfs_root *btrfs_alloc_dummy_root(struct btrfs_fs_info *fs_info);
  */
 static inline struct btrfs_root *btrfs_grab_fs_root(struct btrfs_root *root)
 {
-	if (atomic_inc_not_zero(&root->refs))
+	if (refcount_inc_not_zero(&root->refs))
 		return root;
 	return NULL;
 }
 
 static inline void btrfs_put_fs_root(struct btrfs_root *root)
 {
-	if (atomic_dec_and_test(&root->refs))
+	if (refcount_dec_and_test(&root->refs))
 		kfree(root);
 }
 

fs/btrfs/extent-tree.c

@@ -131,6 +131,16 @@ void btrfs_put_block_group(struct btrfs_block_group_cache *cache)
 	if (atomic_dec_and_test(&cache->count)) {
 		WARN_ON(cache->pinned > 0);
 		WARN_ON(cache->reserved > 0);
+
+		/*
+		 * If not empty, someone is still holding mutex of
+		 * full_stripe_lock, which can only be released by caller.
+		 * And it will definitely cause use-after-free when caller
+		 * tries to release full stripe lock.
+		 *
+		 * No better way to resolve, but only to warn.
+		 */
+		WARN_ON(!RB_EMPTY_ROOT(&cache->full_stripe_locks_root.root));
 		kfree(cache->free_space_ctl);
 		kfree(cache);
 	}
@@ -316,14 +326,14 @@ get_caching_control(struct btrfs_block_group_cache *cache)
 	}
 
 	ctl = cache->caching_ctl;
-	atomic_inc(&ctl->count);
+	refcount_inc(&ctl->count);
 	spin_unlock(&cache->lock);
 	return ctl;
 }
 
 static void put_caching_control(struct btrfs_caching_control *ctl)
 {
-	if (atomic_dec_and_test(&ctl->count))
+	if (refcount_dec_and_test(&ctl->count))
 		kfree(ctl);
 }
 
@@ -599,7 +609,7 @@ static int cache_block_group(struct btrfs_block_group_cache *cache,
 	init_waitqueue_head(&caching_ctl->wait);
 	caching_ctl->block_group = cache;
 	caching_ctl->progress = cache->key.objectid;
-	atomic_set(&caching_ctl->count, 1);
+	refcount_set(&caching_ctl->count, 1);
 	btrfs_init_work(&caching_ctl->work, btrfs_cache_helper,
 			caching_thread, NULL, NULL);
 
@@ -620,7 +630,7 @@ static int cache_block_group(struct btrfs_block_group_cache *cache,
 		struct btrfs_caching_control *ctl;
 
 		ctl = cache->caching_ctl;
-		atomic_inc(&ctl->count);
+		refcount_inc(&ctl->count);
 		prepare_to_wait(&ctl->wait, &wait, TASK_UNINTERRUPTIBLE);
 		spin_unlock(&cache->lock);
 
@@ -707,7 +717,7 @@ static int cache_block_group(struct btrfs_block_group_cache *cache,
 	}
 
 	down_write(&fs_info->commit_root_sem);
-	atomic_inc(&caching_ctl->count);
+	refcount_inc(&caching_ctl->count);
 	list_add_tail(&caching_ctl->list, &fs_info->caching_block_groups);
 	up_write(&fs_info->commit_root_sem);
 
@@ -892,7 +902,7 @@ search_again:
 	head = btrfs_find_delayed_ref_head(delayed_refs, bytenr);
 	if (head) {
 		if (!mutex_trylock(&head->mutex)) {
-			atomic_inc(&head->node.refs);
+			refcount_inc(&head->node.refs);
 			spin_unlock(&delayed_refs->lock);
 
 			btrfs_release_path(path);
@@ -2980,7 +2990,7 @@ again:
 				struct btrfs_delayed_ref_node *ref;
 
 				ref = &head->node;
-				atomic_inc(&ref->refs);
+				refcount_inc(&ref->refs);
 
 				spin_unlock(&delayed_refs->lock);
 				/*
@@ -3003,7 +3013,6 @@ again:
 		goto again;
 	}
 out:
-	assert_qgroups_uptodate(trans);
 	trans->can_flush_pending_bgs = can_flush_pending_bgs;
 	return 0;
 }
@@ -3057,7 +3066,7 @@ static noinline int check_delayed_ref(struct btrfs_root *root,
 	}
 
 	if (!mutex_trylock(&head->mutex)) {
-		atomic_inc(&head->node.refs);
+		refcount_inc(&head->node.refs);
 		spin_unlock(&delayed_refs->lock);
 
 		btrfs_release_path(path);
@@ -3443,7 +3452,8 @@ again:
 		/*
 		 * don't bother trying to write stuff out _if_
 		 * a) we're not cached,
-		 * b) we're with nospace_cache mount option.
+		 * b) we're with nospace_cache mount option,
+		 * c) we're with v2 space_cache (FREE_SPACE_TREE).
 		 */
 		dcs = BTRFS_DC_WRITTEN;
 		spin_unlock(&block_group->lock);
@@ -9917,6 +9927,7 @@ btrfs_create_block_group_cache(struct btrfs_fs_info *fs_info,
 	btrfs_init_free_space_ctl(cache);
 	atomic_set(&cache->trimming, 0);
 	mutex_init(&cache->free_space_lock);
+	btrfs_init_full_stripe_locks_tree(&cache->full_stripe_locks_root);
 
 	return cache;
 }
@@ -10416,7 +10427,7 @@ int btrfs_remove_block_group(struct btrfs_trans_handle *trans,
 				    &fs_info->caching_block_groups, list)
 				if (ctl->block_group == block_group) {
 					caching_ctl = ctl;
-					atomic_inc(&caching_ctl->count);
+					refcount_inc(&caching_ctl->count);
 					break;
 				}
 		}
@@ -10850,7 +10861,7 @@ static int btrfs_trim_free_extents(struct btrfs_device *device,
 		spin_lock(&fs_info->trans_lock);
 		trans = fs_info->running_transaction;
 		if (trans)
-			atomic_inc(&trans->use_count);
+			refcount_inc(&trans->use_count);
 		spin_unlock(&fs_info->trans_lock);
 
 		ret = find_free_dev_extent_start(trans, device, minlen, start,

fs/btrfs/extent_io.c

@@ -68,7 +68,7 @@ void btrfs_leak_debug_check(void)
 		pr_err("BTRFS: state leak: start %llu end %llu state %u in tree %d refs %d\n",
 		       state->start, state->end, state->state,
 		       extent_state_in_tree(state),
-		       atomic_read(&state->refs));
+		       refcount_read(&state->refs));
 		list_del(&state->leak_list);
 		kmem_cache_free(extent_state_cache, state);
 	}
@@ -238,7 +238,7 @@ static struct extent_state *alloc_extent_state(gfp_t mask)
 	state->failrec = NULL;
 	RB_CLEAR_NODE(&state->rb_node);
 	btrfs_leak_debug_add(&state->leak_list, &states);
-	atomic_set(&state->refs, 1);
+	refcount_set(&state->refs, 1);
 	init_waitqueue_head(&state->wq);
 	trace_alloc_extent_state(state, mask, _RET_IP_);
 	return state;
@@ -248,7 +248,7 @@ void free_extent_state(struct extent_state *state)
 {
 	if (!state)
 		return;
-	if (atomic_dec_and_test(&state->refs)) {
+	if (refcount_dec_and_test(&state->refs)) {
 		WARN_ON(extent_state_in_tree(state));
 		btrfs_leak_debug_del(&state->leak_list);
 		trace_free_extent_state(state, _RET_IP_);
@@ -641,7 +641,7 @@ again:
 		if (cached && extent_state_in_tree(cached) &&
 		    cached->start <= start && cached->end > start) {
 			if (clear)
-				atomic_dec(&cached->refs);
+				refcount_dec(&cached->refs);
 			state = cached;
 			goto hit_next;
 		}
@@ -793,7 +793,7 @@ process_node:
 
 		if (state->state & bits) {
 			start = state->start;
-			atomic_inc(&state->refs);
+			refcount_inc(&state->refs);
 			wait_on_state(tree, state);
 			free_extent_state(state);
 			goto again;
@@ -834,7 +834,7 @@ static void cache_state_if_flags(struct extent_state *state,
 	if (cached_ptr && !(*cached_ptr)) {
 		if (!flags || (state->state & flags)) {
 			*cached_ptr = state;
-			atomic_inc(&state->refs);
+			refcount_inc(&state->refs);
 		}
 	}
 }
@@ -1538,7 +1538,7 @@ static noinline u64 find_delalloc_range(struct extent_io_tree *tree,
 		if (!found) {
 			*start = state->start;
 			*cached_state = state;
-			atomic_inc(&state->refs);
+			refcount_inc(&state->refs);
 		}
 		found++;
 		*end = state->end;
@@ -2004,16 +2004,11 @@ int repair_io_failure(struct btrfs_inode *inode, u64 start, u64 length,
 	u64 map_length = 0;
 	u64 sector;
 	struct btrfs_bio *bbio = NULL;
-	struct btrfs_mapping_tree *map_tree = &fs_info->mapping_tree;
 	int ret;
 
 	ASSERT(!(fs_info->sb->s_flags & MS_RDONLY));
 	BUG_ON(!mirror_num);
 
-	/* we can't repair anything in raid56 yet */
-	if (btrfs_is_parity_mirror(map_tree, logical, length, mirror_num))
-		return 0;
-
 	bio = btrfs_io_bio_alloc(GFP_NOFS, 1);
 	if (!bio)
 		return -EIO;
@@ -2026,17 +2021,35 @@ int repair_io_failure(struct btrfs_inode *inode, u64 start, u64 length,
 	 * read repair operation.
 	 */
 	btrfs_bio_counter_inc_blocked(fs_info);
-	ret = btrfs_map_block(fs_info, BTRFS_MAP_WRITE, logical,
-			      &map_length, &bbio, mirror_num);
-	if (ret) {
-		btrfs_bio_counter_dec(fs_info);
-		bio_put(bio);
-		return -EIO;
+	if (btrfs_is_parity_mirror(fs_info, logical, length, mirror_num)) {
+		/*
+		 * Note that we don't use BTRFS_MAP_WRITE because it's supposed
+		 * to update all raid stripes, but here we just want to correct
+		 * bad stripe, thus BTRFS_MAP_READ is abused to only get the bad
+		 * stripe's dev and sector.
+		 */
+		ret = btrfs_map_block(fs_info, BTRFS_MAP_READ, logical,
+				      &map_length, &bbio, 0);
+		if (ret) {
+			btrfs_bio_counter_dec(fs_info);
+			bio_put(bio);
+			return -EIO;
+		}
+		ASSERT(bbio->mirror_num == 1);
+	} else {
+		ret = btrfs_map_block(fs_info, BTRFS_MAP_WRITE, logical,
+				      &map_length, &bbio, mirror_num);
+		if (ret) {
+			btrfs_bio_counter_dec(fs_info);
+			bio_put(bio);
+			return -EIO;
+		}
+		BUG_ON(mirror_num != bbio->mirror_num);
 	}
-	BUG_ON(mirror_num != bbio->mirror_num);
-	sector = bbio->stripes[mirror_num-1].physical >> 9;
+
+	sector = bbio->stripes[bbio->mirror_num - 1].physical >> 9;
 	bio->bi_iter.bi_sector = sector;
-	dev = bbio->stripes[mirror_num-1].dev;
+	dev = bbio->stripes[bbio->mirror_num - 1].dev;
 	btrfs_put_bbio(bbio);
 	if (!dev || !dev->bdev || !dev->writeable) {
 		btrfs_bio_counter_dec(fs_info);
@@ -2859,7 +2872,7 @@ __get_extent_map(struct inode *inode, struct page *page, size_t pg_offset,
 		em = *em_cached;
 		if (extent_map_in_tree(em) && start >= em->start &&
 		    start < extent_map_end(em)) {
-			atomic_inc(&em->refs);
+			refcount_inc(&em->refs);
 			return em;
 		}
 
@@ -2870,7 +2883,7 @@ __get_extent_map(struct inode *inode, struct page *page, size_t pg_offset,
 	em = get_extent(BTRFS_I(inode), page, pg_offset, start, len, 0);
 	if (em_cached && !IS_ERR_OR_NULL(em)) {
 		BUG_ON(*em_cached);
-		atomic_inc(&em->refs);
+		refcount_inc(&em->refs);
 		*em_cached = em;
 	}
 	return em;

fs/btrfs/extent_io.h

@@ -2,6 +2,7 @@
 #define __EXTENTIO__
 
 #include <linux/rbtree.h>
+#include <linux/refcount.h>
 #include "ulist.h"
 
 /* bits for the extent state */
@@ -14,14 +15,17 @@
 #define EXTENT_DEFRAG		(1U << 6)
 #define EXTENT_BOUNDARY		(1U << 9)
 #define EXTENT_NODATASUM	(1U << 10)
-#define EXTENT_DO_ACCOUNTING	(1U << 11)
+#define EXTENT_CLEAR_META_RESV	(1U << 11)
 #define EXTENT_FIRST_DELALLOC	(1U << 12)
 #define EXTENT_NEED_WAIT	(1U << 13)
 #define EXTENT_DAMAGED		(1U << 14)
 #define EXTENT_NORESERVE	(1U << 15)
 #define EXTENT_QGROUP_RESERVED	(1U << 16)
 #define EXTENT_CLEAR_DATA_RESV	(1U << 17)
+#define EXTENT_DELALLOC_NEW	(1U << 18)
 #define EXTENT_IOBITS		(EXTENT_LOCKED | EXTENT_WRITEBACK)
+#define EXTENT_DO_ACCOUNTING    (EXTENT_CLEAR_META_RESV | \
+				 EXTENT_CLEAR_DATA_RESV)
 #define EXTENT_CTLBITS		(EXTENT_DO_ACCOUNTING | EXTENT_FIRST_DELALLOC)
 
 /*
@@ -143,7 +147,7 @@ struct extent_state {
 
 	/* ADD NEW ELEMENTS AFTER THIS */
 	wait_queue_head_t wq;
-	atomic_t refs;
+	refcount_t refs;
 	unsigned state;
 
 	struct io_failure_record *failrec;

fs/btrfs/extent_map.c

@@ -55,7 +55,7 @@ struct extent_map *alloc_extent_map(void)
 	em->flags = 0;
 	em->compress_type = BTRFS_COMPRESS_NONE;
 	em->generation = 0;
-	atomic_set(&em->refs, 1);
+	refcount_set(&em->refs, 1);
 	INIT_LIST_HEAD(&em->list);
 	return em;
 }
@@ -71,8 +71,8 @@ void free_extent_map(struct extent_map *em)
 {
 	if (!em)
 		return;
-	WARN_ON(atomic_read(&em->refs) == 0);
-	if (atomic_dec_and_test(&em->refs)) {
+	WARN_ON(refcount_read(&em->refs) == 0);
+	if (refcount_dec_and_test(&em->refs)) {
 		WARN_ON(extent_map_in_tree(em));
 		WARN_ON(!list_empty(&em->list));
 		if (test_bit(EXTENT_FLAG_FS_MAPPING, &em->flags))
@@ -322,7 +322,7 @@ static inline void setup_extent_mapping(struct extent_map_tree *tree,
 					struct extent_map *em,
 					int modified)
 {
-	atomic_inc(&em->refs);
+	refcount_inc(&em->refs);
 	em->mod_start = em->start;
 	em->mod_len = em->len;
 
@@ -381,7 +381,7 @@ __lookup_extent_mapping(struct extent_map_tree *tree,
 	if (strict && !(end > em->start && start < extent_map_end(em)))
 		return NULL;
 
-	atomic_inc(&em->refs);
+	refcount_inc(&em->refs);
 	return em;
 }
 

fs/btrfs/extent_map.h

@@ -2,6 +2,7 @@
 #define __EXTENTMAP__
 
 #include <linux/rbtree.h>
+#include <linux/refcount.h>
 
 #define EXTENT_MAP_LAST_BYTE ((u64)-4)
 #define EXTENT_MAP_HOLE ((u64)-3)
@@ -41,7 +42,7 @@ struct extent_map {
 		 */
 		struct map_lookup *map_lookup;
 	};
-	atomic_t refs;
+	refcount_t refs;
 	unsigned int compress_type;
 	struct list_head list;
 };

fs/btrfs/file.c

@@ -1404,6 +1404,47 @@ fail:
 
 }
 
+static int btrfs_find_new_delalloc_bytes(struct btrfs_inode *inode,
+					 const u64 start,
+					 const u64 len,
+					 struct extent_state **cached_state)
+{
+	u64 search_start = start;
+	const u64 end = start + len - 1;
+
+	while (search_start < end) {
+		const u64 search_len = end - search_start + 1;
+		struct extent_map *em;
+		u64 em_len;
+		int ret = 0;
+
+		em = btrfs_get_extent(inode, NULL, 0, search_start,
+				      search_len, 0);
+		if (IS_ERR(em))
+			return PTR_ERR(em);
+
+		if (em->block_start != EXTENT_MAP_HOLE)
+			goto next;
+
+		em_len = em->len;
+		if (em->start < search_start)
+			em_len -= search_start - em->start;
+		if (em_len > search_len)
+			em_len = search_len;
+
+		ret = set_extent_bit(&inode->io_tree, search_start,
+				     search_start + em_len - 1,
+				     EXTENT_DELALLOC_NEW,
+				     NULL, cached_state, GFP_NOFS);
+next:
+		search_start = extent_map_end(em);
+		free_extent_map(em);
+		if (ret)
+			return ret;
+	}
+	return 0;
+}
+
 /*
  * This function locks the extent and properly waits for data=ordered extents
  * to finish before allowing the pages to be modified if need.
@@ -1432,8 +1473,11 @@ lock_and_cleanup_extent_if_need(struct btrfs_inode *inode, struct page **pages,
 		+ round_up(pos + write_bytes - start_pos,
 			   fs_info->sectorsize) - 1;
 
-	if (start_pos < inode->vfs_inode.i_size) {
+	if (start_pos < inode->vfs_inode.i_size ||
+	    (inode->flags & BTRFS_INODE_PREALLOC)) {
 		struct btrfs_ordered_extent *ordered;
+		unsigned int clear_bits;
+
 		lock_extent_bits(&inode->io_tree, start_pos, last_pos,
 				cached_state);
 		ordered = btrfs_lookup_ordered_range(inode, start_pos,
@@ -1454,11 +1498,19 @@ lock_and_cleanup_extent_if_need(struct btrfs_inode *inode, struct page **pages,
 		}
 		if (ordered)
 			btrfs_put_ordered_extent(ordered);
-
+		ret = btrfs_find_new_delalloc_bytes(inode, start_pos,
+						    last_pos - start_pos + 1,
+						    cached_state);
+		clear_bits = EXTENT_DIRTY | EXTENT_DELALLOC |
+			EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG;
+		if (ret)
+			clear_bits |= EXTENT_DELALLOC_NEW | EXTENT_LOCKED;
 		clear_extent_bit(&inode->io_tree, start_pos,
-				  last_pos, EXTENT_DIRTY | EXTENT_DELALLOC |
-				  EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG,
-				  0, 0, cached_state, GFP_NOFS);
+				 last_pos, clear_bits,
+				 (clear_bits & EXTENT_LOCKED) ? 1 : 0,
+				 0, cached_state, GFP_NOFS);
+		if (ret)
+			return ret;
 		*lockstart = start_pos;
 		*lockend = last_pos;
 		ret = 1;
@@ -2342,13 +2394,8 @@ static int find_first_non_hole(struct inode *inode, u64 *start, u64 *len)
 	int ret = 0;
 
 	em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, *start, *len, 0);
-	if (IS_ERR_OR_NULL(em)) {
-		if (!em)
-			ret = -ENOMEM;
-		else
-			ret = PTR_ERR(em);
-		return ret;
-	}
+	if (IS_ERR(em))
+		return PTR_ERR(em);
 
 	/* Hole or vacuum extent(only exists in no-hole mode) */
 	if (em->block_start == EXTENT_MAP_HOLE) {
@@ -2835,11 +2882,8 @@ static long btrfs_fallocate(struct file *file, int mode,
 	while (1) {
 		em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, cur_offset,
 				      alloc_end - cur_offset, 0);
-		if (IS_ERR_OR_NULL(em)) {
-			if (!em)
-				ret = -ENOMEM;
-			else
-				ret = PTR_ERR(em);
+		if (IS_ERR(em)) {
+			ret = PTR_ERR(em);
 			break;
 		}
 		last_byte = min(extent_map_end(em), alloc_end);
@@ -2856,8 +2900,10 @@ static long btrfs_fallocate(struct file *file, int mode,
 			}
 			ret = btrfs_qgroup_reserve_data(inode, cur_offset,
 					last_byte - cur_offset);
-			if (ret < 0)
+			if (ret < 0) {
+				free_extent_map(em);
 				break;
+			}
 		} else {
 			/*
 			 * Do not need to reserve unwritten extent for this

fs/btrfs/free-space-cache.c

@@ -355,7 +355,7 @@ static void io_ctl_map_page(struct btrfs_io_ctl *io_ctl, int clear)
 	io_ctl->orig = io_ctl->cur;
 	io_ctl->size = PAGE_SIZE;
 	if (clear)
-		memset(io_ctl->cur, 0, PAGE_SIZE);
+		clear_page(io_ctl->cur);
 }
 
 static void io_ctl_drop_pages(struct btrfs_io_ctl *io_ctl)

fs/btrfs/inode.c

@@ -115,6 +115,31 @@ static struct extent_map *create_io_em(struct inode *inode, u64 start, u64 len,
 				       u64 ram_bytes, int compress_type,
 				       int type);
 
+static void __endio_write_update_ordered(struct inode *inode,
+					 const u64 offset, const u64 bytes,
+					 const bool uptodate);
+
+/*
+ * Cleanup all submitted ordered extents in specified range to handle errors
+ * from the fill_dellaloc() callback.
+ *
+ * NOTE: caller must ensure that when an error happens, it can not call
+ * extent_clear_unlock_delalloc() to clear both the bits EXTENT_DO_ACCOUNTING
+ * and EXTENT_DELALLOC simultaneously, because that causes the reserved metadata
+ * to be released, which we want to happen only when finishing the ordered
+ * extent (btrfs_finish_ordered_io()). Also note that the caller of the
+ * fill_delalloc() callback already does proper cleanup for the first page of
+ * the range, that is, it invokes the callback writepage_end_io_hook() for the
+ * range of the first page.
+ */
+static inline void btrfs_cleanup_ordered_extents(struct inode *inode,
+						 const u64 offset,
+						 const u64 bytes)
+{
+	return __endio_write_update_ordered(inode, offset + PAGE_SIZE,
+					    bytes - PAGE_SIZE, false);
+}
+
 static int btrfs_dirty_inode(struct inode *inode);
 
 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
@@ -547,7 +572,7 @@ cont:
 		}
 		if (ret <= 0) {
 			unsigned long clear_flags = EXTENT_DELALLOC |
-				EXTENT_DEFRAG;
+				EXTENT_DELALLOC_NEW | EXTENT_DEFRAG;
 			unsigned long page_error_op;
 
 			clear_flags |= (ret < 0) ? EXTENT_DO_ACCOUNTING : 0;
@@ -565,8 +590,10 @@ cont:
 						     PAGE_SET_WRITEBACK |
 						     page_error_op |
 						     PAGE_END_WRITEBACK);
-			btrfs_free_reserved_data_space_noquota(inode, start,
-						end - start + 1);
+			if (ret == 0)
+				btrfs_free_reserved_data_space_noquota(inode,
+							       start,
+							       end - start + 1);
 			goto free_pages_out;
 		}
 	}
@@ -852,6 +879,7 @@ out_free:
 				     async_extent->start +
 				     async_extent->ram_size - 1,
 				     NULL, EXTENT_LOCKED | EXTENT_DELALLOC |
+				     EXTENT_DELALLOC_NEW |
 				     EXTENT_DEFRAG | EXTENT_DO_ACCOUNTING,
 				     PAGE_UNLOCK | PAGE_CLEAR_DIRTY |
 				     PAGE_SET_WRITEBACK | PAGE_END_WRITEBACK |
@@ -918,10 +946,13 @@ static noinline int cow_file_range(struct inode *inode,
 	u64 num_bytes;
 	unsigned long ram_size;
 	u64 disk_num_bytes;
-	u64 cur_alloc_size;
+	u64 cur_alloc_size = 0;
 	u64 blocksize = fs_info->sectorsize;
 	struct btrfs_key ins;
 	struct extent_map *em;
+	unsigned clear_bits;
+	unsigned long page_ops;
+	bool extent_reserved = false;
 	int ret = 0;
 
 	if (btrfs_is_free_space_inode(BTRFS_I(inode))) {
@@ -944,6 +975,7 @@ static noinline int cow_file_range(struct inode *inode,
 			extent_clear_unlock_delalloc(inode, start, end,
 				     delalloc_end, NULL,
 				     EXTENT_LOCKED | EXTENT_DELALLOC |
+				     EXTENT_DELALLOC_NEW |
 				     EXTENT_DEFRAG, PAGE_UNLOCK |
 				     PAGE_CLEAR_DIRTY | PAGE_SET_WRITEBACK |
 				     PAGE_END_WRITEBACK);
@@ -966,14 +998,14 @@ static noinline int cow_file_range(struct inode *inode,
 			start + num_bytes - 1, 0);
 
 	while (disk_num_bytes > 0) {
-		unsigned long op;
-
 		cur_alloc_size = disk_num_bytes;
 		ret = btrfs_reserve_extent(root, cur_alloc_size, cur_alloc_size,
 					   fs_info->sectorsize, 0, alloc_hint,
 					   &ins, 1, 1);
 		if (ret < 0)
 			goto out_unlock;
+		cur_alloc_size = ins.offset;
+		extent_reserved = true;
 
 		ram_size = ins.offset;
 		em = create_io_em(inode, start, ins.offset, /* len */
@@ -988,7 +1020,6 @@ static noinline int cow_file_range(struct inode *inode,
 			goto out_reserve;
 		free_extent_map(em);
 
-		cur_alloc_size = ins.offset;
 		ret = btrfs_add_ordered_extent(inode, start, ins.objectid,
 					       ram_size, cur_alloc_size, 0);
 		if (ret)
@@ -998,15 +1029,24 @@ static noinline int cow_file_range(struct inode *inode,
 		    BTRFS_DATA_RELOC_TREE_OBJECTID) {
 			ret = btrfs_reloc_clone_csums(inode, start,
 						      cur_alloc_size);
+			/*
+			 * Only drop cache here, and process as normal.
+			 *
+			 * We must not allow extent_clear_unlock_delalloc()
+			 * at out_unlock label to free meta of this ordered
+			 * extent, as its meta should be freed by
+			 * btrfs_finish_ordered_io().
+			 *
+			 * So we must continue until @start is increased to
+			 * skip current ordered extent.
+			 */
 			if (ret)
-				goto out_drop_extent_cache;
+				btrfs_drop_extent_cache(BTRFS_I(inode), start,
+						start + ram_size - 1, 0);
 		}
 
 		btrfs_dec_block_group_reservations(fs_info, ins.objectid);
 
-		if (disk_num_bytes < cur_alloc_size)
-			break;
-
 		/* we're not doing compressed IO, don't unlock the first
 		 * page (which the caller expects to stay locked), don't
 		 * clear any dirty bits and don't set any writeback bits
@@ -1014,18 +1054,30 @@ static noinline int cow_file_range(struct inode *inode,
 		 * Do set the Private2 bit so we know this page was properly
 		 * setup for writepage
 		 */
-		op = unlock ? PAGE_UNLOCK : 0;
-		op |= PAGE_SET_PRIVATE2;
+		page_ops = unlock ? PAGE_UNLOCK : 0;
+		page_ops |= PAGE_SET_PRIVATE2;
 
 		extent_clear_unlock_delalloc(inode, start,
 					     start + ram_size - 1,
 					     delalloc_end, locked_page,
 					     EXTENT_LOCKED | EXTENT_DELALLOC,
-					     op);
-		disk_num_bytes -= cur_alloc_size;
+					     page_ops);
+		if (disk_num_bytes < cur_alloc_size)
+			disk_num_bytes = 0;
+		else
+			disk_num_bytes -= cur_alloc_size;
 		num_bytes -= cur_alloc_size;
 		alloc_hint = ins.objectid + ins.offset;
 		start += cur_alloc_size;
+		extent_reserved = false;
+
+		/*
+		 * btrfs_reloc_clone_csums() error, since start is increased
+		 * extent_clear_unlock_delalloc() at out_unlock label won't
+		 * free metadata of current ordered extent, we're OK to exit.
+		 */
+		if (ret)
+			goto out_unlock;
 	}
 out:
 	return ret;
@@ -1036,12 +1088,35 @@ out_reserve:
 	btrfs_dec_block_group_reservations(fs_info, ins.objectid);
 	btrfs_free_reserved_extent(fs_info, ins.objectid, ins.offset, 1);
 out_unlock:
+	clear_bits = EXTENT_LOCKED | EXTENT_DELALLOC | EXTENT_DELALLOC_NEW |
+		EXTENT_DEFRAG | EXTENT_CLEAR_META_RESV;
+	page_ops = PAGE_UNLOCK | PAGE_CLEAR_DIRTY | PAGE_SET_WRITEBACK |
+		PAGE_END_WRITEBACK;
+	/*
+	 * If we reserved an extent for our delalloc range (or a subrange) and
+	 * failed to create the respective ordered extent, then it means that
+	 * when we reserved the extent we decremented the extent's size from
+	 * the data space_info's bytes_may_use counter and incremented the
+	 * space_info's bytes_reserved counter by the same amount. We must make
+	 * sure extent_clear_unlock_delalloc() does not try to decrement again
+	 * the data space_info's bytes_may_use counter, therefore we do not pass
+	 * it the flag EXTENT_CLEAR_DATA_RESV.
+	 */
+	if (extent_reserved) {
+		extent_clear_unlock_delalloc(inode, start,
+					     start + cur_alloc_size,
+					     start + cur_alloc_size,
+					     locked_page,
+					     clear_bits,
+					     page_ops);
+		start += cur_alloc_size;
+		if (start >= end)
+			goto out;
+	}
 	extent_clear_unlock_delalloc(inode, start, end, delalloc_end,
 				     locked_page,
-				     EXTENT_LOCKED | EXTENT_DO_ACCOUNTING |
-				     EXTENT_DELALLOC | EXTENT_DEFRAG,
-				     PAGE_UNLOCK | PAGE_CLEAR_DIRTY |
-				     PAGE_SET_WRITEBACK | PAGE_END_WRITEBACK);
+				     clear_bits | EXTENT_CLEAR_DATA_RESV,
+				     page_ops);
 	goto out;
 }
 
@@ -1414,15 +1489,14 @@ out_check:
 		BUG_ON(ret); /* -ENOMEM */
 
 		if (root->root_key.objectid ==
-		    BTRFS_DATA_RELOC_TREE_OBJECTID) {
+		    BTRFS_DATA_RELOC_TREE_OBJECTID)
+			/*
+			 * Error handled later, as we must prevent
+			 * extent_clear_unlock_delalloc() in error handler
+			 * from freeing metadata of created ordered extent.
+			 */
 			ret = btrfs_reloc_clone_csums(inode, cur_offset,
 						      num_bytes);
-			if (ret) {
-				if (!nolock && nocow)
-					btrfs_end_write_no_snapshoting(root);
-				goto error;
-			}
-		}
 
 		extent_clear_unlock_delalloc(inode, cur_offset,
 					     cur_offset + num_bytes - 1, end,
@@ -1434,6 +1508,14 @@ out_check:
 		if (!nolock && nocow)
 			btrfs_end_write_no_snapshoting(root);
 		cur_offset = extent_end;
+
+		/*
+		 * btrfs_reloc_clone_csums() error, now we're OK to call error
+		 * handler, as metadata for created ordered extent will only
+		 * be freed by btrfs_finish_ordered_io().
+		 */
+		if (ret)
+			goto error;
 		if (cur_offset > end)
 			break;
 	}
@@ -1509,6 +1591,8 @@ static int run_delalloc_range(struct inode *inode, struct page *locked_page,
 		ret = cow_file_range_async(inode, locked_page, start, end,
 					   page_started, nr_written);
 	}
+	if (ret)
+		btrfs_cleanup_ordered_extents(inode, start, end - start + 1);
 	return ret;
 }
 
@@ -1693,6 +1777,14 @@ static void btrfs_set_bit_hook(struct inode *inode,
 			btrfs_add_delalloc_inodes(root, inode);
 		spin_unlock(&BTRFS_I(inode)->lock);
 	}
+
+	if (!(state->state & EXTENT_DELALLOC_NEW) &&
+	    (*bits & EXTENT_DELALLOC_NEW)) {
+		spin_lock(&BTRFS_I(inode)->lock);
+		BTRFS_I(inode)->new_delalloc_bytes += state->end + 1 -
+			state->start;
+		spin_unlock(&BTRFS_I(inode)->lock);
+	}
 }
 
 /*
@@ -1722,7 +1814,7 @@ static void btrfs_clear_bit_hook(struct btrfs_inode *inode,
 
 		if (*bits & EXTENT_FIRST_DELALLOC) {
 			*bits &= ~EXTENT_FIRST_DELALLOC;
-		} else if (!(*bits & EXTENT_DO_ACCOUNTING)) {
+		} else if (!(*bits & EXTENT_CLEAR_META_RESV)) {
 			spin_lock(&inode->lock);
 			inode->outstanding_extents -= num_extents;
 			spin_unlock(&inode->lock);
@@ -1733,7 +1825,7 @@ static void btrfs_clear_bit_hook(struct btrfs_inode *inode,
 		 * don't need to call dellalloc_release_metadata if there is an
 		 * error.
 		 */
-		if (*bits & EXTENT_DO_ACCOUNTING &&
+		if (*bits & EXTENT_CLEAR_META_RESV &&
 		    root != fs_info->tree_root)
 			btrfs_delalloc_release_metadata(inode, len);
 
@@ -1741,10 +1833,9 @@ static void btrfs_clear_bit_hook(struct btrfs_inode *inode,
 		if (btrfs_is_testing(fs_info))
 			return;
 
-		if (root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID
-		    && do_list && !(state->state & EXTENT_NORESERVE)
-		    && (*bits & (EXTENT_DO_ACCOUNTING |
-		    EXTENT_CLEAR_DATA_RESV)))
+		if (root->root_key.objectid != BTRFS_DATA_RELOC_TREE_OBJECTID &&
+		    do_list && !(state->state & EXTENT_NORESERVE) &&
+		    (*bits & EXTENT_CLEAR_DATA_RESV))
 			btrfs_free_reserved_data_space_noquota(
 					&inode->vfs_inode,
 					state->start, len);
@@ -1759,6 +1850,14 @@ static void btrfs_clear_bit_hook(struct btrfs_inode *inode,
 			btrfs_del_delalloc_inode(root, inode);
 		spin_unlock(&inode->lock);
 	}
+
+	if ((state->state & EXTENT_DELALLOC_NEW) &&
+	    (*bits & EXTENT_DELALLOC_NEW)) {
+		spin_lock(&inode->lock);
+		ASSERT(inode->new_delalloc_bytes >= len);
+		inode->new_delalloc_bytes -= len;
+		spin_unlock(&inode->lock);
+	}
 }
 
 /*
@@ -2791,6 +2890,13 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 	u64 logical_len = ordered_extent->len;
 	bool nolock;
 	bool truncated = false;
+	bool range_locked = false;
+	bool clear_new_delalloc_bytes = false;
+
+	if (!test_bit(BTRFS_ORDERED_NOCOW, &ordered_extent->flags) &&
+	    !test_bit(BTRFS_ORDERED_PREALLOC, &ordered_extent->flags) &&
+	    !test_bit(BTRFS_ORDERED_DIRECT, &ordered_extent->flags))
+		clear_new_delalloc_bytes = true;
 
 	nolock = btrfs_is_free_space_inode(BTRFS_I(inode));
 
@@ -2839,6 +2945,7 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 		goto out;
 	}
 
+	range_locked = true;
 	lock_extent_bits(io_tree, ordered_extent->file_offset,
 			 ordered_extent->file_offset + ordered_extent->len - 1,
 			 &cached_state);
@@ -2864,7 +2971,7 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 	if (IS_ERR(trans)) {
 		ret = PTR_ERR(trans);
 		trans = NULL;
-		goto out_unlock;
+		goto out;
 	}
 
 	trans->block_rsv = &fs_info->delalloc_block_rsv;
@@ -2896,7 +3003,7 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 			   trans->transid);
 	if (ret < 0) {
 		btrfs_abort_transaction(trans, ret);
-		goto out_unlock;
+		goto out;
 	}
 
 	add_pending_csums(trans, inode, &ordered_extent->list);
@@ -2905,14 +3012,26 @@ static int btrfs_finish_ordered_io(struct btrfs_ordered_extent *ordered_extent)
 	ret = btrfs_update_inode_fallback(trans, root, inode);
 	if (ret) { /* -ENOMEM or corruption */
 		btrfs_abort_transaction(trans, ret);
-		goto out_unlock;
+		goto out;
 	}
 	ret = 0;
-out_unlock:
-	unlock_extent_cached(io_tree, ordered_extent->file_offset,
-			     ordered_extent->file_offset +
-			     ordered_extent->len - 1, &cached_state, GFP_NOFS);
 out:
+	if (range_locked || clear_new_delalloc_bytes) {
+		unsigned int clear_bits = 0;
+
+		if (range_locked)
+			clear_bits |= EXTENT_LOCKED;
+		if (clear_new_delalloc_bytes)
+			clear_bits |= EXTENT_DELALLOC_NEW;
+		clear_extent_bit(&BTRFS_I(inode)->io_tree,
+				 ordered_extent->file_offset,
+				 ordered_extent->file_offset +
+				 ordered_extent->len - 1,
+				 clear_bits,
+				 (clear_bits & EXTENT_LOCKED) ? 1 : 0,
+				 0, &cached_state, GFP_NOFS);
+	}
+
 	if (root != fs_info->tree_root)
 		btrfs_delalloc_release_metadata(BTRFS_I(inode),
 				ordered_extent->len);
@@ -4401,9 +4520,17 @@ search_again:
 			if (extent_type != BTRFS_FILE_EXTENT_INLINE) {
 				item_end +=
 				    btrfs_file_extent_num_bytes(leaf, fi);
+
+				trace_btrfs_truncate_show_fi_regular(
+					BTRFS_I(inode), leaf, fi,
+					found_key.offset);
 			} else if (extent_type == BTRFS_FILE_EXTENT_INLINE) {
 				item_end += btrfs_file_extent_inline_len(leaf,
 							 path->slots[0], fi);
+
+				trace_btrfs_truncate_show_fi_inline(
+					BTRFS_I(inode), leaf, fi, path->slots[0],
+					found_key.offset);
 			}
 			item_end--;
 		}
@@ -4603,13 +4730,6 @@ error:
 
 	btrfs_free_path(path);
 
-	if (err == 0) {
-		/* only inline file may have last_size != new_size */
-		if (new_size >= fs_info->sectorsize ||
-		    new_size > fs_info->max_inline)
-			ASSERT(last_size == new_size);
-	}
-
 	if (be_nice && bytes_deleted > SZ_32M) {
 		unsigned long updates = trans->delayed_ref_updates;
 		if (updates) {
@@ -6735,7 +6855,6 @@ static noinline int uncompress_inline(struct btrfs_path *path,
  *
  * This also copies inline extents directly into the page.
  */
-
 struct extent_map *btrfs_get_extent(struct btrfs_inode *inode,
 		struct page *page,
 	    size_t pg_offset, u64 start, u64 len,
@@ -6835,11 +6954,18 @@ again:
 	    found_type == BTRFS_FILE_EXTENT_PREALLOC) {
 		extent_end = extent_start +
 		       btrfs_file_extent_num_bytes(leaf, item);
+
+		trace_btrfs_get_extent_show_fi_regular(inode, leaf, item,
+						       extent_start);
 	} else if (found_type == BTRFS_FILE_EXTENT_INLINE) {
 		size_t size;
 		size = btrfs_file_extent_inline_len(leaf, path->slots[0], item);
 		extent_end = ALIGN(extent_start + size,
 				   fs_info->sectorsize);
+
+		trace_btrfs_get_extent_show_fi_inline(inode, leaf, item,
+						      path->slots[0],
+						      extent_start);
 	}
 next:
 	if (start >= extent_end) {
@@ -7037,19 +7163,17 @@ struct extent_map *btrfs_get_extent_fiemap(struct btrfs_inode *inode,
 	em = btrfs_get_extent(inode, page, pg_offset, start, len, create);
 	if (IS_ERR(em))
 		return em;
-	if (em) {
-		/*
-		 * if our em maps to
-		 * -  a hole or
-		 * -  a pre-alloc extent,
-		 * there might actually be delalloc bytes behind it.
-		 */
-		if (em->block_start != EXTENT_MAP_HOLE &&
-		    !test_bit(EXTENT_FLAG_PREALLOC, &em->flags))
-			return em;
-		else
-			hole_em = em;
-	}
+	/*
+	 * If our em maps to:
+	 * - a hole or
+	 * - a pre-alloc extent,
+	 * there might actually be delalloc bytes behind it.
+	 */
+	if (em->block_start != EXTENT_MAP_HOLE &&
+	    !test_bit(EXTENT_FLAG_PREALLOC, &em->flags))
+		return em;
+	else
+		hole_em = em;
 
 	/* check to see if we've wrapped (len == -1 or similar) */
 	end = start + len;
@@ -8127,17 +8251,26 @@ static void btrfs_endio_direct_read(struct bio *bio)
 	bio_put(bio);
 }
 
-static void btrfs_endio_direct_write_update_ordered(struct inode *inode,
-						    const u64 offset,
-						    const u64 bytes,
-						    const int uptodate)
+static void __endio_write_update_ordered(struct inode *inode,
+					 const u64 offset, const u64 bytes,
+					 const bool uptodate)
 {
 	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
 	struct btrfs_ordered_extent *ordered = NULL;
+	struct btrfs_workqueue *wq;
+	btrfs_work_func_t func;
 	u64 ordered_offset = offset;
 	u64 ordered_bytes = bytes;
 	int ret;
 
+	if (btrfs_is_free_space_inode(BTRFS_I(inode))) {
+		wq = fs_info->endio_freespace_worker;
+		func = btrfs_freespace_write_helper;
+	} else {
+		wq = fs_info->endio_write_workers;
+		func = btrfs_endio_write_helper;
+	}
+
 again:
 	ret = btrfs_dec_test_first_ordered_pending(inode, &ordered,
 						   &ordered_offset,
@@ -8146,9 +8279,8 @@ again:
 	if (!ret)
 		goto out_test;
 
-	btrfs_init_work(&ordered->work, btrfs_endio_write_helper,
-			finish_ordered_fn, NULL, NULL);
-	btrfs_queue_work(fs_info->endio_write_workers, &ordered->work);
+	btrfs_init_work(&ordered->work, func, finish_ordered_fn, NULL, NULL);
+	btrfs_queue_work(wq, &ordered->work);
 out_test:
 	/*
 	 * our bio might span multiple ordered extents.  If we haven't
@@ -8166,10 +8298,8 @@ static void btrfs_endio_direct_write(struct bio *bio)
 	struct btrfs_dio_private *dip = bio->bi_private;
 	struct bio *dio_bio = dip->dio_bio;
 
-	btrfs_endio_direct_write_update_ordered(dip->inode,
-						dip->logical_offset,
-						dip->bytes,
-						!bio->bi_error);
+	__endio_write_update_ordered(dip->inode, dip->logical_offset,
+				     dip->bytes, !bio->bi_error);
 
 	kfree(dip);
 
@@ -8530,10 +8660,10 @@ free_ordered:
 		io_bio = NULL;
 	} else {
 		if (write)
-			btrfs_endio_direct_write_update_ordered(inode,
+			__endio_write_update_ordered(inode,
 						file_offset,
 						dio_bio->bi_iter.bi_size,
-						0);
+						false);
 		else
 			unlock_extent(&BTRFS_I(inode)->io_tree, file_offset,
 			      file_offset + dio_bio->bi_iter.bi_size - 1);
@@ -8668,11 +8798,11 @@ static ssize_t btrfs_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
 			 */
 			if (dio_data.unsubmitted_oe_range_start <
 			    dio_data.unsubmitted_oe_range_end)
-				btrfs_endio_direct_write_update_ordered(inode,
+				__endio_write_update_ordered(inode,
 					dio_data.unsubmitted_oe_range_start,
 					dio_data.unsubmitted_oe_range_end -
 					dio_data.unsubmitted_oe_range_start,
-					0);
+					false);
 		} else if (ret >= 0 && (size_t)ret < count)
 			btrfs_delalloc_release_space(inode, offset,
 						     count - (size_t)ret);
@@ -8819,6 +8949,7 @@ again:
 		if (!inode_evicting)
 			clear_extent_bit(tree, start, end,
 					 EXTENT_DIRTY | EXTENT_DELALLOC |
+					 EXTENT_DELALLOC_NEW |
 					 EXTENT_LOCKED | EXTENT_DO_ACCOUNTING |
 					 EXTENT_DEFRAG, 1, 0, &cached_state,
 					 GFP_NOFS);
@@ -8876,8 +9007,8 @@ again:
 	if (!inode_evicting) {
 		clear_extent_bit(tree, page_start, page_end,
 				 EXTENT_LOCKED | EXTENT_DIRTY |
-				 EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING |
-				 EXTENT_DEFRAG, 1, 1,
+				 EXTENT_DELALLOC | EXTENT_DELALLOC_NEW |
+				 EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 1, 1,
 				 &cached_state, GFP_NOFS);
 
 		__btrfs_releasepage(page, GFP_NOFS);
@@ -9248,6 +9379,7 @@ struct inode *btrfs_alloc_inode(struct super_block *sb)
 	ei->last_sub_trans = 0;
 	ei->logged_trans = 0;
 	ei->delalloc_bytes = 0;
+	ei->new_delalloc_bytes = 0;
 	ei->defrag_bytes = 0;
 	ei->disk_i_size = 0;
 	ei->flags = 0;
@@ -9313,6 +9445,7 @@ void btrfs_destroy_inode(struct inode *inode)
 	WARN_ON(BTRFS_I(inode)->outstanding_extents);
 	WARN_ON(BTRFS_I(inode)->reserved_extents);
 	WARN_ON(BTRFS_I(inode)->delalloc_bytes);
+	WARN_ON(BTRFS_I(inode)->new_delalloc_bytes);
 	WARN_ON(BTRFS_I(inode)->csum_bytes);
 	WARN_ON(BTRFS_I(inode)->defrag_bytes);
 
@@ -9436,7 +9569,7 @@ static int btrfs_getattr(const struct path *path, struct kstat *stat,
 	stat->dev = BTRFS_I(inode)->root->anon_dev;
 
 	spin_lock(&BTRFS_I(inode)->lock);
-	delalloc_bytes = BTRFS_I(inode)->delalloc_bytes;
+	delalloc_bytes = BTRFS_I(inode)->new_delalloc_bytes;
 	spin_unlock(&BTRFS_I(inode)->lock);
 	stat->blocks = (ALIGN(inode_get_bytes(inode), blocksize) +
 			ALIGN(delalloc_bytes, blocksize)) >> 9;

fs/btrfs/ioctl.c

@@ -1504,7 +1504,7 @@ static noinline int btrfs_ioctl_resize(struct file *file,
 	if (ret)
 		return ret;
 
-	if (atomic_xchg(&fs_info->mutually_exclusive_operation_running, 1)) {
+	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
 		mnt_drop_write_file(file);
 		return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
 	}
@@ -1619,7 +1619,7 @@ out_free:
 	kfree(vol_args);
 out:
 	mutex_unlock(&fs_info->volume_mutex);
-	atomic_set(&fs_info->mutually_exclusive_operation_running, 0);
+	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
 	mnt_drop_write_file(file);
 	return ret;
 }
@@ -2661,7 +2661,7 @@ static long btrfs_ioctl_add_dev(struct btrfs_fs_info *fs_info, void __user *arg)
 	if (!capable(CAP_SYS_ADMIN))
 		return -EPERM;
 
-	if (atomic_xchg(&fs_info->mutually_exclusive_operation_running, 1))
+	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags))
 		return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
 
 	mutex_lock(&fs_info->volume_mutex);
@@ -2680,7 +2680,7 @@ static long btrfs_ioctl_add_dev(struct btrfs_fs_info *fs_info, void __user *arg)
 	kfree(vol_args);
 out:
 	mutex_unlock(&fs_info->volume_mutex);
-	atomic_set(&fs_info->mutually_exclusive_operation_running, 0);
+	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
 	return ret;
 }
 
@@ -2708,7 +2708,7 @@ static long btrfs_ioctl_rm_dev_v2(struct file *file, void __user *arg)
 	if (vol_args->flags & ~BTRFS_VOL_ARG_V2_FLAGS_SUPPORTED)
 		return -EOPNOTSUPP;
 
-	if (atomic_xchg(&fs_info->mutually_exclusive_operation_running, 1)) {
+	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
 		ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
 		goto out;
 	}
@@ -2721,7 +2721,7 @@ static long btrfs_ioctl_rm_dev_v2(struct file *file, void __user *arg)
 		ret = btrfs_rm_device(fs_info, vol_args->name, 0);
 	}
 	mutex_unlock(&fs_info->volume_mutex);
-	atomic_set(&fs_info->mutually_exclusive_operation_running, 0);
+	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
 
 	if (!ret) {
 		if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID)
@@ -2752,7 +2752,7 @@ static long btrfs_ioctl_rm_dev(struct file *file, void __user *arg)
 	if (ret)
 		return ret;
 
-	if (atomic_xchg(&fs_info->mutually_exclusive_operation_running, 1)) {
+	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
 		ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
 		goto out_drop_write;
 	}
@@ -2772,7 +2772,7 @@ static long btrfs_ioctl_rm_dev(struct file *file, void __user *arg)
 		btrfs_info(fs_info, "disk deleted %s", vol_args->name);
 	kfree(vol_args);
 out:
-	atomic_set(&fs_info->mutually_exclusive_operation_running, 0);
+	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
 out_drop_write:
 	mnt_drop_write_file(file);
 
@@ -4439,13 +4439,11 @@ static long btrfs_ioctl_dev_replace(struct btrfs_fs_info *fs_info,
 			ret = -EROFS;
 			goto out;
 		}
-		if (atomic_xchg(
-			&fs_info->mutually_exclusive_operation_running, 1)) {
+		if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
 			ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
 		} else {
 			ret = btrfs_dev_replace_by_ioctl(fs_info, p);
-			atomic_set(
-			 &fs_info->mutually_exclusive_operation_running, 0);
+			clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
 		}
 		break;
 	case BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS:
@@ -4640,7 +4638,7 @@ static long btrfs_ioctl_balance(struct file *file, void __user *arg)
 		return ret;
 
 again:
-	if (!atomic_xchg(&fs_info->mutually_exclusive_operation_running, 1)) {
+	if (!test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
 		mutex_lock(&fs_info->volume_mutex);
 		mutex_lock(&fs_info->balance_mutex);
 		need_unlock = true;
@@ -4686,7 +4684,7 @@ again:
 	}
 
 locked:
-	BUG_ON(!atomic_read(&fs_info->mutually_exclusive_operation_running));
+	BUG_ON(!test_bit(BTRFS_FS_EXCL_OP, &fs_info->flags));
 
 	if (arg) {
 		bargs = memdup_user(arg, sizeof(*bargs));
@@ -4742,11 +4740,10 @@ locked:
 
 do_balance:
 	/*
-	 * Ownership of bctl and mutually_exclusive_operation_running
+	 * Ownership of bctl and filesystem flag BTRFS_FS_EXCL_OP
 	 * goes to to btrfs_balance.  bctl is freed in __cancel_balance,
 	 * or, if restriper was paused all the way until unmount, in
-	 * free_fs_info.  mutually_exclusive_operation_running is
-	 * cleared in __cancel_balance.
+	 * free_fs_info.  The flag is cleared in __cancel_balance.
 	 */
 	need_unlock = false;
 
@@ -4766,7 +4763,7 @@ out_unlock:
 	mutex_unlock(&fs_info->balance_mutex);
 	mutex_unlock(&fs_info->volume_mutex);
 	if (need_unlock)
-		atomic_set(&fs_info->mutually_exclusive_operation_running, 0);
+		clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
 out:
 	mnt_drop_write_file(file);
 	return ret;

fs/btrfs/ordered-data.c

@@ -212,7 +212,7 @@ static int __btrfs_add_ordered_extent(struct inode *inode, u64 file_offset,
 		set_bit(BTRFS_ORDERED_DIRECT, &entry->flags);
 
 	/* one ref for the tree */
-	atomic_set(&entry->refs, 1);
+	refcount_set(&entry->refs, 1);
 	init_waitqueue_head(&entry->wait);
 	INIT_LIST_HEAD(&entry->list);
 	INIT_LIST_HEAD(&entry->root_extent_list);
@@ -358,7 +358,7 @@ int btrfs_dec_test_first_ordered_pending(struct inode *inode,
 out:
 	if (!ret && cached && entry) {
 		*cached = entry;
-		atomic_inc(&entry->refs);
+		refcount_inc(&entry->refs);
 	}
 	spin_unlock_irqrestore(&tree->lock, flags);
 	return ret == 0;
@@ -425,7 +425,7 @@ have_entry:
 out:
 	if (!ret && cached && entry) {
 		*cached = entry;
-		atomic_inc(&entry->refs);
+		refcount_inc(&entry->refs);
 	}
 	spin_unlock_irqrestore(&tree->lock, flags);
 	return ret == 0;
@@ -456,7 +456,7 @@ void btrfs_get_logged_extents(struct btrfs_inode *inode,
 		if (test_and_set_bit(BTRFS_ORDERED_LOGGED, &ordered->flags))
 			continue;
 		list_add(&ordered->log_list, logged_list);
-		atomic_inc(&ordered->refs);
+		refcount_inc(&ordered->refs);
 	}
 	spin_unlock_irq(&tree->lock);
 }
@@ -565,7 +565,7 @@ 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)) {
+	if (refcount_dec_and_test(&entry->refs)) {
 		ASSERT(list_empty(&entry->log_list));
 		ASSERT(list_empty(&entry->trans_list));
 		ASSERT(list_empty(&entry->root_extent_list));
@@ -623,7 +623,7 @@ void btrfs_remove_ordered_extent(struct inode *inode,
 		spin_lock(&fs_info->trans_lock);
 		trans = fs_info->running_transaction;
 		if (trans)
-			atomic_inc(&trans->use_count);
+			refcount_inc(&trans->use_count);
 		spin_unlock(&fs_info->trans_lock);
 
 		ASSERT(trans);
@@ -690,7 +690,7 @@ int btrfs_wait_ordered_extents(struct btrfs_root *root, int nr,
 
 		list_move_tail(&ordered->root_extent_list,
 			       &root->ordered_extents);
-		atomic_inc(&ordered->refs);
+		refcount_inc(&ordered->refs);
 		spin_unlock(&root->ordered_extent_lock);
 
 		btrfs_init_work(&ordered->flush_work,
@@ -870,7 +870,7 @@ struct btrfs_ordered_extent *btrfs_lookup_ordered_extent(struct inode *inode,
 	if (!offset_in_entry(entry, file_offset))
 		entry = NULL;
 	if (entry)
-		atomic_inc(&entry->refs);
+		refcount_inc(&entry->refs);
 out:
 	spin_unlock_irq(&tree->lock);
 	return entry;
@@ -911,7 +911,7 @@ struct btrfs_ordered_extent *btrfs_lookup_ordered_range(
 	}
 out:
 	if (entry)
-		atomic_inc(&entry->refs);
+		refcount_inc(&entry->refs);
 	spin_unlock_irq(&tree->lock);
 	return entry;
 }
@@ -948,7 +948,7 @@ btrfs_lookup_first_ordered_extent(struct inode *inode, u64 file_offset)
 		goto out;
 
 	entry = rb_entry(node, struct btrfs_ordered_extent, rb_node);
-	atomic_inc(&entry->refs);
+	refcount_inc(&entry->refs);
 out:
 	spin_unlock_irq(&tree->lock);
 	return entry;

fs/btrfs/ordered-data.h

@@ -113,7 +113,7 @@ struct btrfs_ordered_extent {
 	int compress_type;
 
 	/* reference count */
-	atomic_t refs;
+	refcount_t refs;
 
 	/* the inode we belong to */
 	struct inode *inode;

fs/btrfs/qgroup.c

@@ -47,50 +47,6 @@
  *  - check all ioctl parameters
  */
 
-/*
- * one struct for each qgroup, organized in fs_info->qgroup_tree.
- */
-struct btrfs_qgroup {
-	u64 qgroupid;
-
-	/*
-	 * state
-	 */
-	u64 rfer;	/* referenced */
-	u64 rfer_cmpr;	/* referenced compressed */
-	u64 excl;	/* exclusive */
-	u64 excl_cmpr;	/* exclusive compressed */
-
-	/*
-	 * limits
-	 */
-	u64 lim_flags;	/* which limits are set */
-	u64 max_rfer;
-	u64 max_excl;
-	u64 rsv_rfer;
-	u64 rsv_excl;
-
-	/*
-	 * reservation tracking
-	 */
-	u64 reserved;
-
-	/*
-	 * lists
-	 */
-	struct list_head groups;  /* groups this group is member of */
-	struct list_head members; /* groups that are members of this group */
-	struct list_head dirty;   /* dirty groups */
-	struct rb_node node;	  /* tree of qgroups */
-
-	/*
-	 * temp variables for accounting operations
-	 * Refer to qgroup_shared_accounting() for details.
-	 */
-	u64 old_refcnt;
-	u64 new_refcnt;
-};
-
 static void btrfs_qgroup_update_old_refcnt(struct btrfs_qgroup *qg, u64 seq,
 					   int mod)
 {
@@ -1078,6 +1034,7 @@ static int __qgroup_excl_accounting(struct btrfs_fs_info *fs_info,
 	qgroup->excl += sign * num_bytes;
 	qgroup->excl_cmpr += sign * num_bytes;
 	if (sign > 0) {
+		trace_qgroup_update_reserve(fs_info, qgroup, -(s64)num_bytes);
 		if (qgroup->reserved < num_bytes)
 			report_reserved_underflow(fs_info, qgroup, num_bytes);
 		else
@@ -1103,6 +1060,8 @@ static int __qgroup_excl_accounting(struct btrfs_fs_info *fs_info,
 		WARN_ON(sign < 0 && qgroup->excl < num_bytes);
 		qgroup->excl += sign * num_bytes;
 		if (sign > 0) {
+			trace_qgroup_update_reserve(fs_info, qgroup,
+						    -(s64)num_bytes);
 			if (qgroup->reserved < num_bytes)
 				report_reserved_underflow(fs_info, qgroup,
 							  num_bytes);
@@ -2058,12 +2017,12 @@ int btrfs_qgroup_account_extents(struct btrfs_trans_handle *trans,
 
 		if (!ret) {
 			/*
-			 * Use (u64)-1 as time_seq to do special search, which
+			 * Use SEQ_LAST 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);
+					record->bytenr, SEQ_LAST, &new_roots);
 			if (ret < 0)
 				goto cleanup;
 			if (qgroup_to_skip)
@@ -2370,6 +2329,7 @@ static int qgroup_reserve(struct btrfs_root *root, u64 num_bytes, bool enforce)
 	struct btrfs_fs_info *fs_info = root->fs_info;
 	u64 ref_root = root->root_key.objectid;
 	int ret = 0;
+	int retried = 0;
 	struct ulist_node *unode;
 	struct ulist_iterator uiter;
 
@@ -2378,7 +2338,7 @@ static int qgroup_reserve(struct btrfs_root *root, u64 num_bytes, bool enforce)
 
 	if (num_bytes == 0)
 		return 0;
-
+retry:
 	spin_lock(&fs_info->qgroup_lock);
 	quota_root = fs_info->quota_root;
 	if (!quota_root)
@@ -2405,6 +2365,27 @@ static int qgroup_reserve(struct btrfs_root *root, u64 num_bytes, bool enforce)
 		qg = unode_aux_to_qgroup(unode);
 
 		if (enforce && !qgroup_check_limits(qg, num_bytes)) {
+			/*
+			 * Commit the tree and retry, since we may have
+			 * deletions which would free up space.
+			 */
+			if (!retried && qg->reserved > 0) {
+				struct btrfs_trans_handle *trans;
+
+				spin_unlock(&fs_info->qgroup_lock);
+				ret = btrfs_start_delalloc_inodes(root, 0);
+				if (ret)
+					return ret;
+				btrfs_wait_ordered_extents(root, -1, 0, (u64)-1);
+				trans = btrfs_join_transaction(root);
+				if (IS_ERR(trans))
+					return PTR_ERR(trans);
+				ret = btrfs_commit_transaction(trans);
+				if (ret)
+					return ret;
+				retried++;
+				goto retry;
+			}
 			ret = -EDQUOT;
 			goto out;
 		}
@@ -2427,6 +2408,7 @@ static int qgroup_reserve(struct btrfs_root *root, u64 num_bytes, bool enforce)
 
 		qg = unode_aux_to_qgroup(unode);
 
+		trace_qgroup_update_reserve(fs_info, qg, num_bytes);
 		qg->reserved += num_bytes;
 	}
 
@@ -2472,6 +2454,7 @@ void btrfs_qgroup_free_refroot(struct btrfs_fs_info *fs_info,
 
 		qg = unode_aux_to_qgroup(unode);
 
+		trace_qgroup_update_reserve(fs_info, qg, -(s64)num_bytes);
 		if (qg->reserved < num_bytes)
 			report_reserved_underflow(fs_info, qg, num_bytes);
 		else
@@ -2490,18 +2473,6 @@ out:
 	spin_unlock(&fs_info->qgroup_lock);
 }
 
-void assert_qgroups_uptodate(struct btrfs_trans_handle *trans)
-{
-	if (list_empty(&trans->qgroup_ref_list) && !trans->delayed_ref_elem.seq)
-		return;
-	btrfs_err(trans->fs_info,
-		"qgroups not uptodate in trans handle %p:  list is%s empty, seq is %#x.%x",
-		trans, list_empty(&trans->qgroup_ref_list) ? "" : " not",
-		(u32)(trans->delayed_ref_elem.seq >> 32),
-		(u32)trans->delayed_ref_elem.seq);
-	BUG();
-}
-
 /*
  * returns < 0 on error, 0 when more leafs are to be scanned.
  * returns 1 when done.
@@ -2889,14 +2860,14 @@ static int __btrfs_qgroup_release_data(struct inode *inode, u64 start, u64 len,
 	if (ret < 0)
 		goto out;
 
-	if (free) {
-		btrfs_qgroup_free_refroot(BTRFS_I(inode)->root->fs_info,
-				BTRFS_I(inode)->root->objectid,
-				changeset.bytes_changed);
+	if (free)
 		trace_op = QGROUP_FREE;
-	}
 	trace_btrfs_qgroup_release_data(inode, start, len,
 					changeset.bytes_changed, trace_op);
+	if (free)
+		btrfs_qgroup_free_refroot(BTRFS_I(inode)->root->fs_info,
+				BTRFS_I(inode)->root->objectid,
+				changeset.bytes_changed);
 out:
 	ulist_release(&changeset.range_changed);
 	return ret;
@@ -2948,6 +2919,7 @@ int btrfs_qgroup_reserve_meta(struct btrfs_root *root, int num_bytes,
 		return 0;
 
 	BUG_ON(num_bytes != round_down(num_bytes, fs_info->nodesize));
+	trace_qgroup_meta_reserve(root, (s64)num_bytes);
 	ret = qgroup_reserve(root, num_bytes, enforce);
 	if (ret < 0)
 		return ret;
@@ -2967,6 +2939,7 @@ void btrfs_qgroup_free_meta_all(struct btrfs_root *root)
 	reserved = atomic64_xchg(&root->qgroup_meta_rsv, 0);
 	if (reserved == 0)
 		return;
+	trace_qgroup_meta_reserve(root, -(s64)reserved);
 	btrfs_qgroup_free_refroot(fs_info, root->objectid, reserved);
 }
 
@@ -2981,6 +2954,7 @@ void btrfs_qgroup_free_meta(struct btrfs_root *root, int num_bytes)
 	BUG_ON(num_bytes != round_down(num_bytes, fs_info->nodesize));
 	WARN_ON(atomic64_read(&root->qgroup_meta_rsv) < num_bytes);
 	atomic64_sub(num_bytes, &root->qgroup_meta_rsv);
+	trace_qgroup_meta_reserve(root, -(s64)num_bytes);
 	btrfs_qgroup_free_refroot(fs_info, root->objectid, num_bytes);
 }
 

fs/btrfs/qgroup.h

@@ -61,6 +61,50 @@ struct btrfs_qgroup_extent_record {
 	struct ulist *old_roots;
 };
 
+/*
+ * one struct for each qgroup, organized in fs_info->qgroup_tree.
+ */
+struct btrfs_qgroup {
+	u64 qgroupid;
+
+	/*
+	 * state
+	 */
+	u64 rfer;	/* referenced */
+	u64 rfer_cmpr;	/* referenced compressed */
+	u64 excl;	/* exclusive */
+	u64 excl_cmpr;	/* exclusive compressed */
+
+	/*
+	 * limits
+	 */
+	u64 lim_flags;	/* which limits are set */
+	u64 max_rfer;
+	u64 max_excl;
+	u64 rsv_rfer;
+	u64 rsv_excl;
+
+	/*
+	 * reservation tracking
+	 */
+	u64 reserved;
+
+	/*
+	 * lists
+	 */
+	struct list_head groups;  /* groups this group is member of */
+	struct list_head members; /* groups that are members of this group */
+	struct list_head dirty;   /* dirty groups */
+	struct rb_node node;	  /* tree of qgroups */
+
+	/*
+	 * temp variables for accounting operations
+	 * Refer to qgroup_shared_accounting() for details.
+	 */
+	u64 old_refcnt;
+	u64 new_refcnt;
+};
+
 /*
  * For qgroup event trace points only
  */
@@ -186,17 +230,12 @@ int btrfs_qgroup_inherit(struct btrfs_trans_handle *trans,
 			 struct btrfs_qgroup_inherit *inherit);
 void btrfs_qgroup_free_refroot(struct btrfs_fs_info *fs_info,
 			       u64 ref_root, u64 num_bytes);
-/*
- * TODO: Add proper trace point for it, as btrfs_qgroup_free() is
- * called by everywhere, can't provide good trace for delayed ref case.
- */
 static inline void btrfs_qgroup_free_delayed_ref(struct btrfs_fs_info *fs_info,
 						 u64 ref_root, u64 num_bytes)
 {
-	btrfs_qgroup_free_refroot(fs_info, ref_root, num_bytes);
 	trace_btrfs_qgroup_free_delayed_ref(fs_info, ref_root, num_bytes);
+	btrfs_qgroup_free_refroot(fs_info, ref_root, num_bytes);
 }
-void assert_qgroups_uptodate(struct btrfs_trans_handle *trans);
 
 #ifdef CONFIG_BTRFS_FS_RUN_SANITY_TESTS
 int btrfs_verify_qgroup_counts(struct btrfs_fs_info *fs_info, u64 qgroupid,

fs/btrfs/raid56.c

@@ -149,7 +149,7 @@ struct btrfs_raid_bio {
 
 	int generic_bio_cnt;
 
-	atomic_t refs;
+	refcount_t refs;
 
 	atomic_t stripes_pending;
 
@@ -389,7 +389,7 @@ static void __remove_rbio_from_cache(struct btrfs_raid_bio *rbio)
 		if (bio_list_empty(&rbio->bio_list)) {
 			if (!list_empty(&rbio->hash_list)) {
 				list_del_init(&rbio->hash_list);
-				atomic_dec(&rbio->refs);
+				refcount_dec(&rbio->refs);
 				BUG_ON(!list_empty(&rbio->plug_list));
 			}
 		}
@@ -480,7 +480,7 @@ static void cache_rbio(struct btrfs_raid_bio *rbio)
 
 	/* bump our ref if we were not in the list before */
 	if (!test_and_set_bit(RBIO_CACHE_BIT, &rbio->flags))
-		atomic_inc(&rbio->refs);
+		refcount_inc(&rbio->refs);
 
 	if (!list_empty(&rbio->stripe_cache)){
 		list_move(&rbio->stripe_cache, &table->stripe_cache);
@@ -689,7 +689,7 @@ static noinline int lock_stripe_add(struct btrfs_raid_bio *rbio)
 			    test_bit(RBIO_CACHE_BIT, &cur->flags) &&
 			    !test_bit(RBIO_RMW_LOCKED_BIT, &cur->flags)) {
 				list_del_init(&cur->hash_list);
-				atomic_dec(&cur->refs);
+				refcount_dec(&cur->refs);
 
 				steal_rbio(cur, rbio);
 				cache_drop = cur;
@@ -738,7 +738,7 @@ static noinline int lock_stripe_add(struct btrfs_raid_bio *rbio)
 		}
 	}
 lockit:
-	atomic_inc(&rbio->refs);
+	refcount_inc(&rbio->refs);
 	list_add(&rbio->hash_list, &h->hash_list);
 out:
 	spin_unlock_irqrestore(&h->lock, flags);
@@ -784,7 +784,7 @@ static noinline void unlock_stripe(struct btrfs_raid_bio *rbio)
 		}
 
 		list_del_init(&rbio->hash_list);
-		atomic_dec(&rbio->refs);
+		refcount_dec(&rbio->refs);
 
 		/*
 		 * we use the plug list to hold all the rbios
@@ -801,7 +801,7 @@ static noinline void unlock_stripe(struct btrfs_raid_bio *rbio)
 			list_del_init(&rbio->plug_list);
 
 			list_add(&next->hash_list, &h->hash_list);
-			atomic_inc(&next->refs);
+			refcount_inc(&next->refs);
 			spin_unlock(&rbio->bio_list_lock);
 			spin_unlock_irqrestore(&h->lock, flags);
 
@@ -843,8 +843,7 @@ static void __free_raid_bio(struct btrfs_raid_bio *rbio)
 {
 	int i;
 
-	WARN_ON(atomic_read(&rbio->refs) < 0);
-	if (!atomic_dec_and_test(&rbio->refs))
+	if (!refcount_dec_and_test(&rbio->refs))
 		return;
 
 	WARN_ON(!list_empty(&rbio->stripe_cache));
@@ -997,7 +996,7 @@ static struct btrfs_raid_bio *alloc_rbio(struct btrfs_fs_info *fs_info,
 	rbio->stripe_npages = stripe_npages;
 	rbio->faila = -1;
 	rbio->failb = -1;
-	atomic_set(&rbio->refs, 1);
+	refcount_set(&rbio->refs, 1);
 	atomic_set(&rbio->error, 0);
 	atomic_set(&rbio->stripes_pending, 0);
 
@@ -2118,6 +2117,11 @@ int raid56_parity_recover(struct btrfs_fs_info *fs_info, struct bio *bio,
 	struct btrfs_raid_bio *rbio;
 	int ret;
 
+	if (generic_io) {
+		ASSERT(bbio->mirror_num == mirror_num);
+		btrfs_io_bio(bio)->mirror_num = mirror_num;
+	}
+
 	rbio = alloc_rbio(fs_info, bbio, stripe_len);
 	if (IS_ERR(rbio)) {
 		if (generic_io)
@@ -2194,6 +2198,8 @@ static void read_rebuild_work(struct btrfs_work *work)
 /*
  * The following code is used to scrub/replace the parity stripe
  *
+ * Caller must have already increased bio_counter for getting @bbio.
+ *
  * Note: We need make sure all the pages that add into the scrub/replace
  * raid bio are correct and not be changed during the scrub/replace. That
  * is those pages just hold metadata or file data with checksum.
@@ -2231,6 +2237,12 @@ raid56_parity_alloc_scrub_rbio(struct btrfs_fs_info *fs_info, struct bio *bio,
 	ASSERT(rbio->stripe_npages == stripe_nsectors);
 	bitmap_copy(rbio->dbitmap, dbitmap, stripe_nsectors);
 
+	/*
+	 * We have already increased bio_counter when getting bbio, record it
+	 * so we can free it at rbio_orig_end_io().
+	 */
+	rbio->generic_bio_cnt = 1;
+
 	return rbio;
 }
 
@@ -2673,6 +2685,12 @@ raid56_alloc_missing_rbio(struct btrfs_fs_info *fs_info, struct bio *bio,
 		return NULL;
 	}
 
+	/*
+	 * When we get bbio, we have already increased bio_counter, record it
+	 * so we can free it at rbio_orig_end_io()
+	 */
+	rbio->generic_bio_cnt = 1;
+
 	return rbio;
 }
 

fs/btrfs/reada.c

@@ -209,9 +209,9 @@ cleanup:
 	return;
 }
 
-int btree_readahead_hook(struct btrfs_fs_info *fs_info,
-			 struct extent_buffer *eb, int err)
+int btree_readahead_hook(struct extent_buffer *eb, int err)
 {
+	struct btrfs_fs_info *fs_info = eb->fs_info;
 	int ret = 0;
 	struct reada_extent *re;
 
@@ -235,10 +235,10 @@ start_machine:
 	return ret;
 }
 
-static struct reada_zone *reada_find_zone(struct btrfs_fs_info *fs_info,
-					  struct btrfs_device *dev, u64 logical,
+static struct reada_zone *reada_find_zone(struct btrfs_device *dev, u64 logical,
 					  struct btrfs_bio *bbio)
 {
+	struct btrfs_fs_info *fs_info = dev->fs_info;
 	int ret;
 	struct reada_zone *zone;
 	struct btrfs_block_group_cache *cache = NULL;
@@ -270,6 +270,12 @@ static struct reada_zone *reada_find_zone(struct btrfs_fs_info *fs_info,
 	if (!zone)
 		return NULL;
 
+	ret = radix_tree_preload(GFP_KERNEL);
+	if (ret) {
+		kfree(zone);
+		return NULL;
+	}
+
 	zone->start = start;
 	zone->end = end;
 	INIT_LIST_HEAD(&zone->list);
@@ -299,6 +305,7 @@ static struct reada_zone *reada_find_zone(struct btrfs_fs_info *fs_info,
 			zone = NULL;
 	}
 	spin_unlock(&fs_info->reada_lock);
+	radix_tree_preload_end();
 
 	return zone;
 }
@@ -313,7 +320,6 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 	struct btrfs_bio *bbio = NULL;
 	struct btrfs_device *dev;
 	struct btrfs_device *prev_dev;
-	u32 blocksize;
 	u64 length;
 	int real_stripes;
 	int nzones = 0;
@@ -334,7 +340,6 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 	if (!re)
 		return NULL;
 
-	blocksize = fs_info->nodesize;
 	re->logical = logical;
 	re->top = *top;
 	INIT_LIST_HEAD(&re->extctl);
@@ -344,10 +349,10 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 	/*
 	 * map block
 	 */
-	length = blocksize;
+	length = fs_info->nodesize;
 	ret = btrfs_map_block(fs_info, BTRFS_MAP_GET_READ_MIRRORS, logical,
 			&length, &bbio, 0);
-	if (ret || !bbio || length < blocksize)
+	if (ret || !bbio || length < fs_info->nodesize)
 		goto error;
 
 	if (bbio->num_stripes > BTRFS_MAX_MIRRORS) {
@@ -367,7 +372,7 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 		 if (!dev->bdev)
 			continue;
 
-		zone = reada_find_zone(fs_info, dev, logical, bbio);
+		zone = reada_find_zone(dev, logical, bbio);
 		if (!zone)
 			continue;
 
@@ -386,6 +391,10 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 		goto error;
 	}
 
+	ret = radix_tree_preload(GFP_KERNEL);
+	if (ret)
+		goto error;
+
 	/* insert extent in reada_tree + all per-device trees, all or nothing */
 	btrfs_dev_replace_lock(&fs_info->dev_replace, 0);
 	spin_lock(&fs_info->reada_lock);
@@ -395,13 +404,16 @@ static struct reada_extent *reada_find_extent(struct btrfs_fs_info *fs_info,
 		re_exist->refcnt++;
 		spin_unlock(&fs_info->reada_lock);
 		btrfs_dev_replace_unlock(&fs_info->dev_replace, 0);
+		radix_tree_preload_end();
 		goto error;
 	}
 	if (ret) {
 		spin_unlock(&fs_info->reada_lock);
 		btrfs_dev_replace_unlock(&fs_info->dev_replace, 0);
+		radix_tree_preload_end();
 		goto error;
 	}
+	radix_tree_preload_end();
 	prev_dev = NULL;
 	dev_replace_is_ongoing = btrfs_dev_replace_is_ongoing(
 			&fs_info->dev_replace);
@@ -639,9 +651,9 @@ static int reada_pick_zone(struct btrfs_device *dev)
 	return 1;
 }
 
-static int reada_start_machine_dev(struct btrfs_fs_info *fs_info,
-				   struct btrfs_device *dev)
+static int reada_start_machine_dev(struct btrfs_device *dev)
 {
+	struct btrfs_fs_info *fs_info = dev->fs_info;
 	struct reada_extent *re = NULL;
 	int mirror_num = 0;
 	struct extent_buffer *eb = NULL;
@@ -754,8 +766,7 @@ static void __reada_start_machine(struct btrfs_fs_info *fs_info)
 		list_for_each_entry(device, &fs_devices->devices, dev_list) {
 			if (atomic_read(&device->reada_in_flight) <
 			    MAX_IN_FLIGHT)
-				enqueued += reada_start_machine_dev(fs_info,
-								    device);
+				enqueued += reada_start_machine_dev(device);
 		}
 		mutex_unlock(&fs_devices->device_list_mutex);
 		total += enqueued;

fs/btrfs/root-tree.c

@@ -501,8 +501,9 @@ void btrfs_update_root_times(struct btrfs_trans_handle *trans,
 			     struct btrfs_root *root)
 {
 	struct btrfs_root_item *item = &root->root_item;
-	struct timespec ct = current_fs_time(root->fs_info->sb);
+	struct timespec ct;
 
+	ktime_get_real_ts(&ct);
 	spin_lock(&root->root_item_lock);
 	btrfs_set_root_ctransid(item, trans->transid);
 	btrfs_set_stack_timespec_sec(&item->ctime, ct.tv_sec);

fs/btrfs/scrub.c

@@ -64,7 +64,7 @@ struct scrub_ctx;
 #define SCRUB_MAX_PAGES_PER_BLOCK	16	/* 64k per node/leaf/sector */
 
 struct scrub_recover {
-	atomic_t		refs;
+	refcount_t		refs;
 	struct btrfs_bio	*bbio;
 	u64			map_length;
 };
@@ -112,7 +112,7 @@ struct scrub_block {
 	struct scrub_page	*pagev[SCRUB_MAX_PAGES_PER_BLOCK];
 	int			page_count;
 	atomic_t		outstanding_pages;
-	atomic_t		refs; /* free mem on transition to zero */
+	refcount_t		refs; /* free mem on transition to zero */
 	struct scrub_ctx	*sctx;
 	struct scrub_parity	*sparity;
 	struct {
@@ -140,9 +140,9 @@ struct scrub_parity {
 
 	int			nsectors;
 
-	int			stripe_len;
+	u64			stripe_len;
 
-	atomic_t		refs;
+	refcount_t		refs;
 
 	struct list_head	spages;
 
@@ -202,7 +202,7 @@ struct scrub_ctx {
 	 * doesn't free the scrub context before or while the workers are
 	 * doing the wakeup() call.
 	 */
-	atomic_t                refs;
+	refcount_t              refs;
 };
 
 struct scrub_fixup_nodatasum {
@@ -240,6 +240,13 @@ struct scrub_warning {
 	struct btrfs_device	*dev;
 };
 
+struct full_stripe_lock {
+	struct rb_node node;
+	u64 logical;
+	u64 refs;
+	struct mutex mutex;
+};
+
 static void scrub_pending_bio_inc(struct scrub_ctx *sctx);
 static void scrub_pending_bio_dec(struct scrub_ctx *sctx);
 static void scrub_pending_trans_workers_inc(struct scrub_ctx *sctx);
@@ -305,7 +312,7 @@ static void scrub_put_ctx(struct scrub_ctx *sctx);
 
 static void scrub_pending_bio_inc(struct scrub_ctx *sctx)
 {
-	atomic_inc(&sctx->refs);
+	refcount_inc(&sctx->refs);
 	atomic_inc(&sctx->bios_in_flight);
 }
 
@@ -348,6 +355,222 @@ static void scrub_blocked_if_needed(struct btrfs_fs_info *fs_info)
 	scrub_pause_off(fs_info);
 }
 
+/*
+ * Insert new full stripe lock into full stripe locks tree
+ *
+ * Return pointer to existing or newly inserted full_stripe_lock structure if
+ * everything works well.
+ * Return ERR_PTR(-ENOMEM) if we failed to allocate memory
+ *
+ * NOTE: caller must hold full_stripe_locks_root->lock before calling this
+ * function
+ */
+static struct full_stripe_lock *insert_full_stripe_lock(
+		struct btrfs_full_stripe_locks_tree *locks_root,
+		u64 fstripe_logical)
+{
+	struct rb_node **p;
+	struct rb_node *parent = NULL;
+	struct full_stripe_lock *entry;
+	struct full_stripe_lock *ret;
+
+	WARN_ON(!mutex_is_locked(&locks_root->lock));
+
+	p = &locks_root->root.rb_node;
+	while (*p) {
+		parent = *p;
+		entry = rb_entry(parent, struct full_stripe_lock, node);
+		if (fstripe_logical < entry->logical) {
+			p = &(*p)->rb_left;
+		} else if (fstripe_logical > entry->logical) {
+			p = &(*p)->rb_right;
+		} else {
+			entry->refs++;
+			return entry;
+		}
+	}
+
+	/* Insert new lock */
+	ret = kmalloc(sizeof(*ret), GFP_KERNEL);
+	if (!ret)
+		return ERR_PTR(-ENOMEM);
+	ret->logical = fstripe_logical;
+	ret->refs = 1;
+	mutex_init(&ret->mutex);
+
+	rb_link_node(&ret->node, parent, p);
+	rb_insert_color(&ret->node, &locks_root->root);
+	return ret;
+}
+
+/*
+ * Search for a full stripe lock of a block group
+ *
+ * Return pointer to existing full stripe lock if found
+ * Return NULL if not found
+ */
+static struct full_stripe_lock *search_full_stripe_lock(
+		struct btrfs_full_stripe_locks_tree *locks_root,
+		u64 fstripe_logical)
+{
+	struct rb_node *node;
+	struct full_stripe_lock *entry;
+
+	WARN_ON(!mutex_is_locked(&locks_root->lock));
+
+	node = locks_root->root.rb_node;
+	while (node) {
+		entry = rb_entry(node, struct full_stripe_lock, node);
+		if (fstripe_logical < entry->logical)
+			node = node->rb_left;
+		else if (fstripe_logical > entry->logical)
+			node = node->rb_right;
+		else
+			return entry;
+	}
+	return NULL;
+}
+
+/*
+ * Helper to get full stripe logical from a normal bytenr.
+ *
+ * Caller must ensure @cache is a RAID56 block group.
+ */
+static u64 get_full_stripe_logical(struct btrfs_block_group_cache *cache,
+				   u64 bytenr)
+{
+	u64 ret;
+
+	/*
+	 * Due to chunk item size limit, full stripe length should not be
+	 * larger than U32_MAX. Just a sanity check here.
+	 */
+	WARN_ON_ONCE(cache->full_stripe_len >= U32_MAX);
+
+	/*
+	 * round_down() can only handle power of 2, while RAID56 full
+	 * stripe length can be 64KiB * n, so we need to manually round down.
+	 */
+	ret = div64_u64(bytenr - cache->key.objectid, cache->full_stripe_len) *
+		cache->full_stripe_len + cache->key.objectid;
+	return ret;
+}
+
+/*
+ * Lock a full stripe to avoid concurrency of recovery and read
+ *
+ * It's only used for profiles with parities (RAID5/6), for other profiles it
+ * does nothing.
+ *
+ * Return 0 if we locked full stripe covering @bytenr, with a mutex held.
+ * So caller must call unlock_full_stripe() at the same context.
+ *
+ * Return <0 if encounters error.
+ */
+static int lock_full_stripe(struct btrfs_fs_info *fs_info, u64 bytenr,
+			    bool *locked_ret)
+{
+	struct btrfs_block_group_cache *bg_cache;
+	struct btrfs_full_stripe_locks_tree *locks_root;
+	struct full_stripe_lock *existing;
+	u64 fstripe_start;
+	int ret = 0;
+
+	*locked_ret = false;
+	bg_cache = btrfs_lookup_block_group(fs_info, bytenr);
+	if (!bg_cache) {
+		ASSERT(0);
+		return -ENOENT;
+	}
+
+	/* Profiles not based on parity don't need full stripe lock */
+	if (!(bg_cache->flags & BTRFS_BLOCK_GROUP_RAID56_MASK))
+		goto out;
+	locks_root = &bg_cache->full_stripe_locks_root;
+
+	fstripe_start = get_full_stripe_logical(bg_cache, bytenr);
+
+	/* Now insert the full stripe lock */
+	mutex_lock(&locks_root->lock);
+	existing = insert_full_stripe_lock(locks_root, fstripe_start);
+	mutex_unlock(&locks_root->lock);
+	if (IS_ERR(existing)) {
+		ret = PTR_ERR(existing);
+		goto out;
+	}
+	mutex_lock(&existing->mutex);
+	*locked_ret = true;
+out:
+	btrfs_put_block_group(bg_cache);
+	return ret;
+}
+
+/*
+ * Unlock a full stripe.
+ *
+ * NOTE: Caller must ensure it's the same context calling corresponding
+ * lock_full_stripe().
+ *
+ * Return 0 if we unlock full stripe without problem.
+ * Return <0 for error
+ */
+static int unlock_full_stripe(struct btrfs_fs_info *fs_info, u64 bytenr,
+			      bool locked)
+{
+	struct btrfs_block_group_cache *bg_cache;
+	struct btrfs_full_stripe_locks_tree *locks_root;
+	struct full_stripe_lock *fstripe_lock;
+	u64 fstripe_start;
+	bool freeit = false;
+	int ret = 0;
+
+	/* If we didn't acquire full stripe lock, no need to continue */
+	if (!locked)
+		return 0;
+
+	bg_cache = btrfs_lookup_block_group(fs_info, bytenr);
+	if (!bg_cache) {
+		ASSERT(0);
+		return -ENOENT;
+	}
+	if (!(bg_cache->flags & BTRFS_BLOCK_GROUP_RAID56_MASK))
+		goto out;
+
+	locks_root = &bg_cache->full_stripe_locks_root;
+	fstripe_start = get_full_stripe_logical(bg_cache, bytenr);
+
+	mutex_lock(&locks_root->lock);
+	fstripe_lock = search_full_stripe_lock(locks_root, fstripe_start);
+	/* Unpaired unlock_full_stripe() detected */
+	if (!fstripe_lock) {
+		WARN_ON(1);
+		ret = -ENOENT;
+		mutex_unlock(&locks_root->lock);
+		goto out;
+	}
+
+	if (fstripe_lock->refs == 0) {
+		WARN_ON(1);
+		btrfs_warn(fs_info, "full stripe lock at %llu refcount underflow",
+			fstripe_lock->logical);
+	} else {
+		fstripe_lock->refs--;
+	}
+
+	if (fstripe_lock->refs == 0) {
+		rb_erase(&fstripe_lock->node, &locks_root->root);
+		freeit = true;
+	}
+	mutex_unlock(&locks_root->lock);
+
+	mutex_unlock(&fstripe_lock->mutex);
+	if (freeit)
+		kfree(fstripe_lock);
+out:
+	btrfs_put_block_group(bg_cache);
+	return ret;
+}
+
 /*
  * used for workers that require transaction commits (i.e., for the
  * NOCOW case)
@@ -356,7 +579,7 @@ static void scrub_pending_trans_workers_inc(struct scrub_ctx *sctx)
 {
 	struct btrfs_fs_info *fs_info = sctx->fs_info;
 
-	atomic_inc(&sctx->refs);
+	refcount_inc(&sctx->refs);
 	/*
 	 * increment scrubs_running to prevent cancel requests from
 	 * completing as long as a worker is running. we must also
@@ -447,7 +670,7 @@ static noinline_for_stack void scrub_free_ctx(struct scrub_ctx *sctx)
 
 static void scrub_put_ctx(struct scrub_ctx *sctx)
 {
-	if (atomic_dec_and_test(&sctx->refs))
+	if (refcount_dec_and_test(&sctx->refs))
 		scrub_free_ctx(sctx);
 }
 
@@ -462,7 +685,7 @@ struct scrub_ctx *scrub_setup_ctx(struct btrfs_device *dev, int is_dev_replace)
 	sctx = kzalloc(sizeof(*sctx), GFP_KERNEL);
 	if (!sctx)
 		goto nomem;
-	atomic_set(&sctx->refs, 1);
+	refcount_set(&sctx->refs, 1);
 	sctx->is_dev_replace = is_dev_replace;
 	sctx->pages_per_rd_bio = SCRUB_PAGES_PER_RD_BIO;
 	sctx->curr = -1;
@@ -857,12 +1080,14 @@ out:
 
 static inline void scrub_get_recover(struct scrub_recover *recover)
 {
-	atomic_inc(&recover->refs);
+	refcount_inc(&recover->refs);
 }
 
-static inline void scrub_put_recover(struct scrub_recover *recover)
+static inline void scrub_put_recover(struct btrfs_fs_info *fs_info,
+				     struct scrub_recover *recover)
 {
-	if (atomic_dec_and_test(&recover->refs)) {
+	if (refcount_dec_and_test(&recover->refs)) {
+		btrfs_bio_counter_dec(fs_info);
 		btrfs_put_bbio(recover->bbio);
 		kfree(recover);
 	}
@@ -892,6 +1117,7 @@ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check)
 	int mirror_index;
 	int page_num;
 	int success;
+	bool full_stripe_locked;
 	static DEFINE_RATELIMIT_STATE(_rs, DEFAULT_RATELIMIT_INTERVAL,
 				      DEFAULT_RATELIMIT_BURST);
 
@@ -917,6 +1143,24 @@ static int scrub_handle_errored_block(struct scrub_block *sblock_to_check)
 	have_csum = sblock_to_check->pagev[0]->have_csum;
 	dev = sblock_to_check->pagev[0]->dev;
 
+	/*
+	 * For RAID5/6, race can happen for a different device scrub thread.
+	 * For data corruption, Parity and Data threads will both try
+	 * to recovery the data.
+	 * Race can lead to doubly added csum error, or even unrecoverable
+	 * error.
+	 */
+	ret = lock_full_stripe(fs_info, logical, &full_stripe_locked);
+	if (ret < 0) {
+		spin_lock(&sctx->stat_lock);
+		if (ret == -ENOMEM)
+			sctx->stat.malloc_errors++;
+		sctx->stat.read_errors++;
+		sctx->stat.uncorrectable_errors++;
+		spin_unlock(&sctx->stat_lock);
+		return ret;
+	}
+
 	if (sctx->is_dev_replace && !is_metadata && !have_csum) {
 		sblocks_for_recheck = NULL;
 		goto nodatasum_case;
@@ -1241,7 +1485,7 @@ out:
 				sblock->pagev[page_index]->sblock = NULL;
 				recover = sblock->pagev[page_index]->recover;
 				if (recover) {
-					scrub_put_recover(recover);
+					scrub_put_recover(fs_info, recover);
 					sblock->pagev[page_index]->recover =
 									NULL;
 				}
@@ -1251,6 +1495,9 @@ out:
 		kfree(sblocks_for_recheck);
 	}
 
+	ret = unlock_full_stripe(fs_info, logical, full_stripe_locked);
+	if (ret < 0)
+		return ret;
 	return 0;
 }
 
@@ -1330,20 +1577,23 @@ static int scrub_setup_recheck_block(struct scrub_block *original_sblock,
 		 * with a length of PAGE_SIZE, each returned stripe
 		 * represents one mirror
 		 */
+		btrfs_bio_counter_inc_blocked(fs_info);
 		ret = btrfs_map_sblock(fs_info, BTRFS_MAP_GET_READ_MIRRORS,
-				logical, &mapped_length, &bbio, 0, 1);
+				logical, &mapped_length, &bbio);
 		if (ret || !bbio || mapped_length < sublen) {
 			btrfs_put_bbio(bbio);
+			btrfs_bio_counter_dec(fs_info);
 			return -EIO;
 		}
 
 		recover = kzalloc(sizeof(struct scrub_recover), GFP_NOFS);
 		if (!recover) {
 			btrfs_put_bbio(bbio);
+			btrfs_bio_counter_dec(fs_info);
 			return -ENOMEM;
 		}
 
-		atomic_set(&recover->refs, 1);
+		refcount_set(&recover->refs, 1);
 		recover->bbio = bbio;
 		recover->map_length = mapped_length;
 
@@ -1365,7 +1615,7 @@ leave_nomem:
 				spin_lock(&sctx->stat_lock);
 				sctx->stat.malloc_errors++;
 				spin_unlock(&sctx->stat_lock);
-				scrub_put_recover(recover);
+				scrub_put_recover(fs_info, recover);
 				return -ENOMEM;
 			}
 			scrub_page_get(page);
@@ -1407,7 +1657,7 @@ leave_nomem:
 			scrub_get_recover(recover);
 			page->recover = recover;
 		}
-		scrub_put_recover(recover);
+		scrub_put_recover(fs_info, recover);
 		length -= sublen;
 		logical += sublen;
 		page_index++;
@@ -1497,14 +1747,18 @@ static void scrub_recheck_block(struct btrfs_fs_info *fs_info,
 
 		bio_add_page(bio, page->page, PAGE_SIZE, 0);
 		if (!retry_failed_mirror && scrub_is_page_on_raid56(page)) {
-			if (scrub_submit_raid56_bio_wait(fs_info, bio, page))
+			if (scrub_submit_raid56_bio_wait(fs_info, bio, page)) {
+				page->io_error = 1;
 				sblock->no_io_error_seen = 0;
+			}
 		} else {
 			bio->bi_iter.bi_sector = page->physical >> 9;
 			bio_set_op_attrs(bio, REQ_OP_READ, 0);
 
-			if (btrfsic_submit_bio_wait(bio))
+			if (btrfsic_submit_bio_wait(bio)) {
+				page->io_error = 1;
 				sblock->no_io_error_seen = 0;
+			}
 		}
 
 		bio_put(bio);
@@ -1634,7 +1888,7 @@ static int scrub_write_page_to_dev_replace(struct scrub_block *sblock,
 	if (spage->io_error) {
 		void *mapped_buffer = kmap_atomic(spage->page);
 
-		memset(mapped_buffer, 0, PAGE_SIZE);
+		clear_page(mapped_buffer);
 		flush_dcache_page(spage->page);
 		kunmap_atomic(mapped_buffer);
 	}
@@ -1998,12 +2252,12 @@ static int scrub_checksum_super(struct scrub_block *sblock)
 
 static void scrub_block_get(struct scrub_block *sblock)
 {
-	atomic_inc(&sblock->refs);
+	refcount_inc(&sblock->refs);
 }
 
 static void scrub_block_put(struct scrub_block *sblock)
 {
-	if (atomic_dec_and_test(&sblock->refs)) {
+	if (refcount_dec_and_test(&sblock->refs)) {
 		int i;
 
 		if (sblock->sparity)
@@ -2187,8 +2441,9 @@ static void scrub_missing_raid56_pages(struct scrub_block *sblock)
 	int ret;
 	int i;
 
+	btrfs_bio_counter_inc_blocked(fs_info);
 	ret = btrfs_map_sblock(fs_info, BTRFS_MAP_GET_READ_MIRRORS, logical,
-			&length, &bbio, 0, 1);
+			&length, &bbio);
 	if (ret || !bbio || !bbio->raid_map)
 		goto bbio_out;
 
@@ -2231,6 +2486,7 @@ static void scrub_missing_raid56_pages(struct scrub_block *sblock)
 rbio_out:
 	bio_put(bio);
 bbio_out:
+	btrfs_bio_counter_dec(fs_info);
 	btrfs_put_bbio(bbio);
 	spin_lock(&sctx->stat_lock);
 	sctx->stat.malloc_errors++;
@@ -2255,7 +2511,7 @@ static int scrub_pages(struct scrub_ctx *sctx, u64 logical, u64 len,
 
 	/* one ref inside this function, plus one for each page added to
 	 * a bio later on */
-	atomic_set(&sblock->refs, 1);
+	refcount_set(&sblock->refs, 1);
 	sblock->sctx = sctx;
 	sblock->no_io_error_seen = 1;
 
@@ -2385,7 +2641,7 @@ static inline void __scrub_mark_bitmap(struct scrub_parity *sparity,
 				       unsigned long *bitmap,
 				       u64 start, u64 len)
 {
-	u32 offset;
+	u64 offset;
 	int nsectors;
 	int sectorsize = sparity->sctx->fs_info->sectorsize;
 
@@ -2395,8 +2651,8 @@ static inline void __scrub_mark_bitmap(struct scrub_parity *sparity,
 	}
 
 	start -= sparity->logic_start;
-	start = div_u64_rem(start, sparity->stripe_len, &offset);
-	offset /= sectorsize;
+	start = div64_u64_rem(start, sparity->stripe_len, &offset);
+	offset = div_u64(offset, sectorsize);
 	nsectors = (int)len / sectorsize;
 
 	if (offset + nsectors <= sparity->nsectors) {
@@ -2555,7 +2811,7 @@ static int scrub_pages_for_parity(struct scrub_parity *sparity,
 
 	/* one ref inside this function, plus one for each page added to
 	 * a bio later on */
-	atomic_set(&sblock->refs, 1);
+	refcount_set(&sblock->refs, 1);
 	sblock->sctx = sctx;
 	sblock->no_io_error_seen = 1;
 	sblock->sparity = sparity;
@@ -2694,7 +2950,7 @@ static int get_raid56_logic_offset(u64 physical, int num,
 	for (i = 0; i < nr_data_stripes(map); i++) {
 		*offset = last_offset + i * map->stripe_len;
 
-		stripe_nr = div_u64(*offset, map->stripe_len);
+		stripe_nr = div64_u64(*offset, map->stripe_len);
 		stripe_nr = div_u64(stripe_nr, nr_data_stripes(map));
 
 		/* Work out the disk rotation on this stripe-set */
@@ -2765,7 +3021,6 @@ static void scrub_parity_check_and_repair(struct scrub_parity *sparity)
 	struct btrfs_fs_info *fs_info = sctx->fs_info;
 	struct bio *bio;
 	struct btrfs_raid_bio *rbio;
-	struct scrub_page *spage;
 	struct btrfs_bio *bbio = NULL;
 	u64 length;
 	int ret;
@@ -2775,8 +3030,10 @@ static void scrub_parity_check_and_repair(struct scrub_parity *sparity)
 		goto out;
 
 	length = sparity->logic_end - sparity->logic_start;
+
+	btrfs_bio_counter_inc_blocked(fs_info);
 	ret = btrfs_map_sblock(fs_info, BTRFS_MAP_WRITE, sparity->logic_start,
-			       &length, &bbio, 0, 1);
+			       &length, &bbio);
 	if (ret || !bbio || !bbio->raid_map)
 		goto bbio_out;
 
@@ -2795,9 +3052,6 @@ static void scrub_parity_check_and_repair(struct scrub_parity *sparity)
 	if (!rbio)
 		goto rbio_out;
 
-	list_for_each_entry(spage, &sparity->spages, list)
-		raid56_add_scrub_pages(rbio, spage->page, spage->logical);
-
 	scrub_pending_bio_inc(sctx);
 	raid56_parity_submit_scrub_rbio(rbio);
 	return;
@@ -2805,6 +3059,7 @@ static void scrub_parity_check_and_repair(struct scrub_parity *sparity)
 rbio_out:
 	bio_put(bio);
 bbio_out:
+	btrfs_bio_counter_dec(fs_info);
 	btrfs_put_bbio(bbio);
 	bitmap_or(sparity->ebitmap, sparity->ebitmap, sparity->dbitmap,
 		  sparity->nsectors);
@@ -2822,12 +3077,12 @@ static inline int scrub_calc_parity_bitmap_len(int nsectors)
 
 static void scrub_parity_get(struct scrub_parity *sparity)
 {
-	atomic_inc(&sparity->refs);
+	refcount_inc(&sparity->refs);
 }
 
 static void scrub_parity_put(struct scrub_parity *sparity)
 {
-	if (!atomic_dec_and_test(&sparity->refs))
+	if (!refcount_dec_and_test(&sparity->refs))
 		return;
 
 	scrub_parity_check_and_repair(sparity);
@@ -2879,7 +3134,7 @@ static noinline_for_stack int scrub_raid56_parity(struct scrub_ctx *sctx,
 	sparity->scrub_dev = sdev;
 	sparity->logic_start = logic_start;
 	sparity->logic_end = logic_end;
-	atomic_set(&sparity->refs, 1);
+	refcount_set(&sparity->refs, 1);
 	INIT_LIST_HEAD(&sparity->spages);
 	sparity->dbitmap = sparity->bitmap;
 	sparity->ebitmap = (void *)sparity->bitmap + bitmap_len;
@@ -3098,7 +3353,7 @@ static noinline_for_stack int scrub_stripe(struct scrub_ctx *sctx,
 
 	physical = map->stripes[num].physical;
 	offset = 0;
-	nstripes = div_u64(length, map->stripe_len);
+	nstripes = div64_u64(length, map->stripe_len);
 	if (map->type & BTRFS_BLOCK_GROUP_RAID0) {
 		offset = map->stripe_len * num;
 		increment = map->stripe_len * map->num_stripes;

fs/btrfs/send.c

@@ -5184,13 +5184,19 @@ static int is_extent_unchanged(struct send_ctx *sctx,
 	while (key.offset < ekey->offset + left_len) {
 		ei = btrfs_item_ptr(eb, slot, struct btrfs_file_extent_item);
 		right_type = btrfs_file_extent_type(eb, ei);
-		if (right_type != BTRFS_FILE_EXTENT_REG) {
+		if (right_type != BTRFS_FILE_EXTENT_REG &&
+		    right_type != BTRFS_FILE_EXTENT_INLINE) {
 			ret = 0;
 			goto out;
 		}
 
 		right_disknr = btrfs_file_extent_disk_bytenr(eb, ei);
-		right_len = btrfs_file_extent_num_bytes(eb, ei);
+		if (right_type == BTRFS_FILE_EXTENT_INLINE) {
+			right_len = btrfs_file_extent_inline_len(eb, slot, ei);
+			right_len = PAGE_ALIGN(right_len);
+		} else {
+			right_len = btrfs_file_extent_num_bytes(eb, ei);
+		}
 		right_offset = btrfs_file_extent_offset(eb, ei);
 		right_gen = btrfs_file_extent_generation(eb, ei);
 
@@ -5204,6 +5210,19 @@ static int is_extent_unchanged(struct send_ctx *sctx,
 			goto out;
 		}
 
+		/*
+		 * We just wanted to see if when we have an inline extent, what
+		 * follows it is a regular extent (wanted to check the above
+		 * condition for inline extents too). This should normally not
+		 * happen but it's possible for example when we have an inline
+		 * compressed extent representing data with a size matching
+		 * the page size (currently the same as sector size).
+		 */
+		if (right_type == BTRFS_FILE_EXTENT_INLINE) {
+			ret = 0;
+			goto out;
+		}
+
 		left_offset_fixed = left_offset;
 		if (key.offset < ekey->offset) {
 			/* Fix the right offset for 2a and 7. */

fs/btrfs/super.c

@@ -1795,8 +1795,7 @@ static int btrfs_remount(struct super_block *sb, int *flags, char *data)
 		}
 
 		if (fs_info->fs_devices->missing_devices >
-		     fs_info->num_tolerated_disk_barrier_failures &&
-		    !(*flags & MS_RDONLY)) {
+		     fs_info->num_tolerated_disk_barrier_failures) {
 			btrfs_warn(fs_info,
 				"too many missing devices, writeable remount is not allowed");
 			ret = -EACCES;

fs/btrfs/tests/btrfs-tests.c

@@ -237,7 +237,6 @@ void btrfs_init_dummy_trans(struct btrfs_trans_handle *trans)
 {
 	memset(trans, 0, sizeof(*trans));
 	trans->transid = 1;
-	INIT_LIST_HEAD(&trans->qgroup_ref_list);
 	trans->type = __TRANS_DUMMY;
 }
 

fs/btrfs/transaction.c

@@ -60,8 +60,8 @@ static const unsigned int btrfs_blocked_trans_types[TRANS_STATE_MAX] = {
 
 void btrfs_put_transaction(struct btrfs_transaction *transaction)
 {
-	WARN_ON(atomic_read(&transaction->use_count) == 0);
-	if (atomic_dec_and_test(&transaction->use_count)) {
+	WARN_ON(refcount_read(&transaction->use_count) == 0);
+	if (refcount_dec_and_test(&transaction->use_count)) {
 		BUG_ON(!list_empty(&transaction->list));
 		WARN_ON(!RB_EMPTY_ROOT(&transaction->delayed_refs.href_root));
 		if (transaction->delayed_refs.pending_csums)
@@ -207,7 +207,7 @@ loop:
 			spin_unlock(&fs_info->trans_lock);
 			return -EBUSY;
 		}
-		atomic_inc(&cur_trans->use_count);
+		refcount_inc(&cur_trans->use_count);
 		atomic_inc(&cur_trans->num_writers);
 		extwriter_counter_inc(cur_trans, type);
 		spin_unlock(&fs_info->trans_lock);
@@ -257,7 +257,7 @@ loop:
 	 * One for this trans handle, one so it will live on until we
 	 * commit the transaction.
 	 */
-	atomic_set(&cur_trans->use_count, 2);
+	refcount_set(&cur_trans->use_count, 2);
 	atomic_set(&cur_trans->pending_ordered, 0);
 	cur_trans->flags = 0;
 	cur_trans->start_time = get_seconds();
@@ -432,7 +432,7 @@ static void wait_current_trans(struct btrfs_fs_info *fs_info)
 	spin_lock(&fs_info->trans_lock);
 	cur_trans = fs_info->running_transaction;
 	if (cur_trans && is_transaction_blocked(cur_trans)) {
-		atomic_inc(&cur_trans->use_count);
+		refcount_inc(&cur_trans->use_count);
 		spin_unlock(&fs_info->trans_lock);
 
 		wait_event(fs_info->transaction_wait,
@@ -572,7 +572,6 @@ again:
 
 	h->type = type;
 	h->can_flush_pending_bgs = true;
-	INIT_LIST_HEAD(&h->qgroup_ref_list);
 	INIT_LIST_HEAD(&h->new_bgs);
 
 	smp_mb();
@@ -744,7 +743,7 @@ int btrfs_wait_for_commit(struct btrfs_fs_info *fs_info, u64 transid)
 		list_for_each_entry(t, &fs_info->trans_list, list) {
 			if (t->transid == transid) {
 				cur_trans = t;
-				atomic_inc(&cur_trans->use_count);
+				refcount_inc(&cur_trans->use_count);
 				ret = 0;
 				break;
 			}
@@ -773,7 +772,7 @@ int btrfs_wait_for_commit(struct btrfs_fs_info *fs_info, u64 transid)
 				if (t->state == TRANS_STATE_COMPLETED)
 					break;
 				cur_trans = t;
-				atomic_inc(&cur_trans->use_count);
+				refcount_inc(&cur_trans->use_count);
 				break;
 			}
 		}
@@ -917,7 +916,6 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
 		wake_up_process(info->transaction_kthread);
 		err = -EIO;
 	}
-	assert_qgroups_uptodate(trans);
 
 	kmem_cache_free(btrfs_trans_handle_cachep, trans);
 	if (must_run_delayed_refs) {
@@ -1839,7 +1837,7 @@ int btrfs_commit_transaction_async(struct btrfs_trans_handle *trans,
 
 	/* take transaction reference */
 	cur_trans = trans->transaction;
-	atomic_inc(&cur_trans->use_count);
+	refcount_inc(&cur_trans->use_count);
 
 	btrfs_end_transaction(trans);
 
@@ -2015,7 +2013,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	spin_lock(&fs_info->trans_lock);
 	if (cur_trans->state >= TRANS_STATE_COMMIT_START) {
 		spin_unlock(&fs_info->trans_lock);
-		atomic_inc(&cur_trans->use_count);
+		refcount_inc(&cur_trans->use_count);
 		ret = btrfs_end_transaction(trans);
 
 		wait_for_commit(cur_trans);
@@ -2035,7 +2033,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 		prev_trans = list_entry(cur_trans->list.prev,
 					struct btrfs_transaction, list);
 		if (prev_trans->state != TRANS_STATE_COMPLETED) {
-			atomic_inc(&prev_trans->use_count);
+			refcount_inc(&prev_trans->use_count);
 			spin_unlock(&fs_info->trans_lock);
 
 			wait_for_commit(prev_trans);
@@ -2130,13 +2128,6 @@ 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, fs_info);
-	if (ret) {
-		mutex_unlock(&fs_info->reloc_mutex);
-		goto scrub_continue;
-	}
-
 	/*
 	 * make sure none of the code above managed to slip in a
 	 * delayed item
@@ -2178,6 +2169,24 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 	 */
 	btrfs_free_log_root_tree(trans, fs_info);
 
+	/*
+	 * commit_fs_roots() can call btrfs_save_ino_cache(), which generates
+	 * new delayed refs. Must handle them or qgroup can be wrong.
+	 */
+	ret = btrfs_run_delayed_refs(trans, fs_info, (unsigned long)-1);
+	if (ret) {
+		mutex_unlock(&fs_info->tree_log_mutex);
+		mutex_unlock(&fs_info->reloc_mutex);
+		goto scrub_continue;
+	}
+
+	ret = btrfs_qgroup_prepare_account_extents(trans, fs_info);
+	if (ret) {
+		mutex_unlock(&fs_info->tree_log_mutex);
+		mutex_unlock(&fs_info->reloc_mutex);
+		goto scrub_continue;
+	}
+
 	/*
 	 * Since fs roots are all committed, we can get a quite accurate
 	 * new_roots. So let's do quota accounting.
@@ -2223,7 +2232,6 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans)
 
 	switch_commit_roots(cur_trans, fs_info);
 
-	assert_qgroups_uptodate(trans);
 	ASSERT(list_empty(&cur_trans->dirty_bgs));
 	ASSERT(list_empty(&cur_trans->io_bgs));
 	update_super_roots(fs_info);

fs/btrfs/transaction.h

@@ -18,6 +18,8 @@
 
 #ifndef __BTRFS_TRANSACTION__
 #define __BTRFS_TRANSACTION__
+
+#include <linux/refcount.h>
 #include "btrfs_inode.h"
 #include "delayed-ref.h"
 #include "ctree.h"
@@ -49,7 +51,7 @@ struct btrfs_transaction {
 	 * transaction can end
 	 */
 	atomic_t num_writers;
-	atomic_t use_count;
+	refcount_t use_count;
 	atomic_t pending_ordered;
 
 	unsigned long flags;
@@ -125,8 +127,6 @@ struct btrfs_trans_handle {
 	unsigned int type;
 	struct btrfs_root *root;
 	struct btrfs_fs_info *fs_info;
-	struct seq_list delayed_ref_elem;
-	struct list_head qgroup_ref_list;
 	struct list_head new_bgs;
 };
 

fs/btrfs/tree-log.c

@@ -4196,7 +4196,7 @@ static int btrfs_log_changed_extents(struct btrfs_trans_handle *trans,
 		if (em->generation <= test_gen)
 			continue;
 		/* Need a ref to keep it from getting evicted from cache */
-		atomic_inc(&em->refs);
+		refcount_inc(&em->refs);
 		set_bit(EXTENT_FLAG_LOGGING, &em->flags);
 		list_add_tail(&em->list, &extents);
 		num++;

fs/btrfs/volumes.c

@@ -139,6 +139,11 @@ static int btrfs_relocate_sys_chunks(struct btrfs_fs_info *fs_info);
 static void __btrfs_reset_dev_stats(struct btrfs_device *dev);
 static void btrfs_dev_stat_print_on_error(struct btrfs_device *dev);
 static void btrfs_dev_stat_print_on_load(struct btrfs_device *device);
+static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
+			     enum btrfs_map_op op,
+			     u64 logical, u64 *length,
+			     struct btrfs_bio **bbio_ret,
+			     int mirror_num, int need_raid_map);
 
 DEFINE_MUTEX(uuid_mutex);
 static LIST_HEAD(fs_uuids);
@@ -1008,14 +1013,13 @@ static int __btrfs_open_devices(struct btrfs_fs_devices *fs_devices,
 		q = bdev_get_queue(bdev);
 		if (blk_queue_discard(q))
 			device->can_discard = 1;
+		if (!blk_queue_nonrot(q))
+			fs_devices->rotating = 1;
 
 		device->bdev = bdev;
 		device->in_fs_metadata = 0;
 		device->mode = flags;
 
-		if (!blk_queue_nonrot(bdev_get_queue(bdev)))
-			fs_devices->rotating = 1;
-
 		fs_devices->open_devices++;
 		if (device->writeable &&
 		    device->devid != BTRFS_DEV_REPLACE_DEVID) {
@@ -2417,7 +2421,7 @@ int btrfs_init_new_device(struct btrfs_fs_info *fs_info, const char *device_path
 	fs_info->free_chunk_space += device->total_bytes;
 	spin_unlock(&fs_info->free_chunk_lock);
 
-	if (!blk_queue_nonrot(bdev_get_queue(bdev)))
+	if (!blk_queue_nonrot(q))
 		fs_info->fs_devices->rotating = 1;
 
 	tmp = btrfs_super_total_bytes(fs_info->super_copy);
@@ -2795,10 +2799,38 @@ static int btrfs_del_sys_chunk(struct btrfs_fs_info *fs_info,
 	return ret;
 }
 
+static struct extent_map *get_chunk_map(struct btrfs_fs_info *fs_info,
+					u64 logical, u64 length)
+{
+	struct extent_map_tree *em_tree;
+	struct extent_map *em;
+
+	em_tree = &fs_info->mapping_tree.map_tree;
+	read_lock(&em_tree->lock);
+	em = lookup_extent_mapping(em_tree, logical, length);
+	read_unlock(&em_tree->lock);
+
+	if (!em) {
+		btrfs_crit(fs_info, "unable to find logical %llu length %llu",
+			   logical, length);
+		return ERR_PTR(-EINVAL);
+	}
+
+	if (em->start > logical || em->start + em->len < logical) {
+		btrfs_crit(fs_info,
+			   "found a bad mapping, wanted %llu-%llu, found %llu-%llu",
+			   logical, length, em->start, em->start + em->len);
+		free_extent_map(em);
+		return ERR_PTR(-EINVAL);
+	}
+
+	/* callers are responsible for dropping em's ref. */
+	return em;
+}
+
 int btrfs_remove_chunk(struct btrfs_trans_handle *trans,
 		       struct btrfs_fs_info *fs_info, u64 chunk_offset)
 {
-	struct extent_map_tree *em_tree;
 	struct extent_map *em;
 	struct map_lookup *map;
 	u64 dev_extent_len = 0;
@@ -2806,23 +2838,15 @@ int btrfs_remove_chunk(struct btrfs_trans_handle *trans,
 	int i, ret = 0;
 	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
 
-	em_tree = &fs_info->mapping_tree.map_tree;
-
-	read_lock(&em_tree->lock);
-	em = lookup_extent_mapping(em_tree, chunk_offset, 1);
-	read_unlock(&em_tree->lock);
-
-	if (!em || em->start > chunk_offset ||
-	    em->start + em->len < chunk_offset) {
+	em = get_chunk_map(fs_info, chunk_offset, 1);
+	if (IS_ERR(em)) {
 		/*
 		 * This is a logic error, but we don't want to just rely on the
 		 * user having built with ASSERT enabled, so if ASSERT doesn't
 		 * do anything we still error out.
 		 */
 		ASSERT(0);
-		if (em)
-			free_extent_map(em);
-		return -EINVAL;
+		return PTR_ERR(em);
 	}
 	map = em->map_lookup;
 	mutex_lock(&fs_info->chunk_mutex);
@@ -3736,7 +3760,7 @@ static void __cancel_balance(struct btrfs_fs_info *fs_info)
 	if (ret)
 		btrfs_handle_fs_error(fs_info, ret, NULL);
 
-	atomic_set(&fs_info->mutually_exclusive_operation_running, 0);
+	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
 }
 
 /* Non-zero return value signifies invalidity */
@@ -3755,6 +3779,7 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 		  struct btrfs_ioctl_balance_args *bargs)
 {
 	struct btrfs_fs_info *fs_info = bctl->fs_info;
+	u64 meta_target, data_target;
 	u64 allowed;
 	int mixed = 0;
 	int ret;
@@ -3851,11 +3876,16 @@ int btrfs_balance(struct btrfs_balance_control *bctl,
 		}
 	} while (read_seqretry(&fs_info->profiles_lock, seq));
 
-	if (btrfs_get_num_tolerated_disk_barrier_failures(bctl->meta.target) <
-		btrfs_get_num_tolerated_disk_barrier_failures(bctl->data.target)) {
+	/* if we're not converting, the target field is uninitialized */
+	meta_target = (bctl->meta.flags & BTRFS_BALANCE_ARGS_CONVERT) ?
+		bctl->meta.target : fs_info->avail_metadata_alloc_bits;
+	data_target = (bctl->data.flags & BTRFS_BALANCE_ARGS_CONVERT) ?
+		bctl->data.target : fs_info->avail_data_alloc_bits;
+	if (btrfs_get_num_tolerated_disk_barrier_failures(meta_target) <
+		btrfs_get_num_tolerated_disk_barrier_failures(data_target)) {
 		btrfs_warn(fs_info,
 			   "metadata profile 0x%llx has lower redundancy than data profile 0x%llx",
-			   bctl->meta.target, bctl->data.target);
+			   meta_target, data_target);
 	}
 
 	if (bctl->sys.flags & BTRFS_BALANCE_ARGS_CONVERT) {
@@ -3910,7 +3940,7 @@ out:
 		__cancel_balance(fs_info);
 	else {
 		kfree(bctl);
-		atomic_set(&fs_info->mutually_exclusive_operation_running, 0);
+		clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
 	}
 	return ret;
 }
@@ -4000,7 +4030,7 @@ int btrfs_recover_balance(struct btrfs_fs_info *fs_info)
 	btrfs_balance_sys(leaf, item, &disk_bargs);
 	btrfs_disk_balance_args_to_cpu(&bctl->sys, &disk_bargs);
 
-	WARN_ON(atomic_xchg(&fs_info->mutually_exclusive_operation_running, 1));
+	WARN_ON(test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags));
 
 	mutex_lock(&fs_info->volume_mutex);
 	mutex_lock(&fs_info->balance_mutex);
@@ -4785,7 +4815,7 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 	stripe_size = div_u64(stripe_size, dev_stripes);
 
 	/* align to BTRFS_STRIPE_LEN */
-	stripe_size = div_u64(stripe_size, raid_stripe_len);
+	stripe_size = div64_u64(stripe_size, raid_stripe_len);
 	stripe_size *= raid_stripe_len;
 
 	map = kmalloc(map_lookup_size(num_stripes), GFP_NOFS);
@@ -4833,7 +4863,7 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 	ret = add_extent_mapping(em_tree, em, 0);
 	if (!ret) {
 		list_add_tail(&em->list, &trans->transaction->pending_chunks);
-		atomic_inc(&em->refs);
+		refcount_inc(&em->refs);
 	}
 	write_unlock(&em_tree->lock);
 	if (ret) {
@@ -4888,7 +4918,6 @@ int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans,
 	struct btrfs_device *device;
 	struct btrfs_chunk *chunk;
 	struct btrfs_stripe *stripe;
-	struct extent_map_tree *em_tree;
 	struct extent_map *em;
 	struct map_lookup *map;
 	size_t item_size;
@@ -4897,24 +4926,9 @@ int btrfs_finish_chunk_alloc(struct btrfs_trans_handle *trans,
 	int i = 0;
 	int ret = 0;
 
-	em_tree = &fs_info->mapping_tree.map_tree;
-	read_lock(&em_tree->lock);
-	em = lookup_extent_mapping(em_tree, chunk_offset, chunk_size);
-	read_unlock(&em_tree->lock);
-
-	if (!em) {
-		btrfs_crit(fs_info, "unable to find logical %Lu len %Lu",
-			   chunk_offset, chunk_size);
-		return -EINVAL;
-	}
-
-	if (em->start != chunk_offset || em->len != chunk_size) {
-		btrfs_crit(fs_info,
-			   "found a bad mapping, wanted %Lu-%Lu, found %Lu-%Lu",
-			    chunk_offset, chunk_size, em->start, em->len);
-		free_extent_map(em);
-		return -EINVAL;
-	}
+	em = get_chunk_map(fs_info, chunk_offset, chunk_size);
+	if (IS_ERR(em))
+		return PTR_ERR(em);
 
 	map = em->map_lookup;
 	item_size = btrfs_chunk_item_size(map->num_stripes);
@@ -5055,15 +5069,12 @@ int btrfs_chunk_readonly(struct btrfs_fs_info *fs_info, u64 chunk_offset)
 {
 	struct extent_map *em;
 	struct map_lookup *map;
-	struct btrfs_mapping_tree *map_tree = &fs_info->mapping_tree;
 	int readonly = 0;
 	int miss_ndevs = 0;
 	int i;
 
-	read_lock(&map_tree->map_tree.lock);
-	em = lookup_extent_mapping(&map_tree->map_tree, chunk_offset, 1);
-	read_unlock(&map_tree->map_tree.lock);
-	if (!em)
+	em = get_chunk_map(fs_info, chunk_offset, 1);
+	if (IS_ERR(em))
 		return 1;
 
 	map = em->map_lookup;
@@ -5117,34 +5128,19 @@ void btrfs_mapping_tree_free(struct btrfs_mapping_tree *tree)
 
 int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len)
 {
-	struct btrfs_mapping_tree *map_tree = &fs_info->mapping_tree;
 	struct extent_map *em;
 	struct map_lookup *map;
-	struct extent_map_tree *em_tree = &map_tree->map_tree;
 	int ret;
 
-	read_lock(&em_tree->lock);
-	em = lookup_extent_mapping(em_tree, logical, len);
-	read_unlock(&em_tree->lock);
-
-	/*
-	 * We could return errors for these cases, but that could get ugly and
-	 * we'd probably do the same thing which is just not do anything else
-	 * and exit, so return 1 so the callers don't try to use other copies.
-	 */
-	if (!em) {
-		btrfs_crit(fs_info, "No mapping for %Lu-%Lu", logical,
-			    logical+len);
-		return 1;
-	}
-
-	if (em->start > logical || em->start + em->len < logical) {
-		btrfs_crit(fs_info, "Invalid mapping for %Lu-%Lu, got %Lu-%Lu",
-			   logical, logical+len, em->start,
-			   em->start + em->len);
-		free_extent_map(em);
+	em = get_chunk_map(fs_info, logical, len);
+	if (IS_ERR(em))
+		/*
+		 * We could return errors for these cases, but that could get
+		 * ugly and we'd probably do the same thing which is just not do
+		 * anything else and exit, so return 1 so the callers don't try
+		 * to use other copies.
+		 */
 		return 1;
-	}
 
 	map = em->map_lookup;
 	if (map->type & (BTRFS_BLOCK_GROUP_DUP | BTRFS_BLOCK_GROUP_RAID1))
@@ -5160,7 +5156,8 @@ int btrfs_num_copies(struct btrfs_fs_info *fs_info, u64 logical, u64 len)
 	free_extent_map(em);
 
 	btrfs_dev_replace_lock(&fs_info->dev_replace, 0);
-	if (btrfs_dev_replace_is_ongoing(&fs_info->dev_replace))
+	if (btrfs_dev_replace_is_ongoing(&fs_info->dev_replace) &&
+	    fs_info->dev_replace.tgtdev)
 		ret++;
 	btrfs_dev_replace_unlock(&fs_info->dev_replace, 0);
 
@@ -5173,15 +5170,11 @@ unsigned long btrfs_full_stripe_len(struct btrfs_fs_info *fs_info,
 {
 	struct extent_map *em;
 	struct map_lookup *map;
-	struct extent_map_tree *em_tree = &map_tree->map_tree;
 	unsigned long len = fs_info->sectorsize;
 
-	read_lock(&em_tree->lock);
-	em = lookup_extent_mapping(em_tree, logical, len);
-	read_unlock(&em_tree->lock);
-	BUG_ON(!em);
+	em = get_chunk_map(fs_info, logical, len);
+	WARN_ON(IS_ERR(em));
 
-	BUG_ON(em->start > logical || em->start + em->len < logical);
 	map = em->map_lookup;
 	if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK)
 		len = map->stripe_len * nr_data_stripes(map);
@@ -5189,20 +5182,16 @@ unsigned long btrfs_full_stripe_len(struct btrfs_fs_info *fs_info,
 	return len;
 }
 
-int btrfs_is_parity_mirror(struct btrfs_mapping_tree *map_tree,
+int btrfs_is_parity_mirror(struct btrfs_fs_info *fs_info,
 			   u64 logical, u64 len, int mirror_num)
 {
 	struct extent_map *em;
 	struct map_lookup *map;
-	struct extent_map_tree *em_tree = &map_tree->map_tree;
 	int ret = 0;
 
-	read_lock(&em_tree->lock);
-	em = lookup_extent_mapping(em_tree, logical, len);
-	read_unlock(&em_tree->lock);
-	BUG_ON(!em);
+	em = get_chunk_map(fs_info, logical, len);
+	WARN_ON(IS_ERR(em));
 
-	BUG_ON(em->start > logical || em->start + em->len < logical);
 	map = em->map_lookup;
 	if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK)
 		ret = 1;
@@ -5295,25 +5284,353 @@ static struct btrfs_bio *alloc_btrfs_bio(int total_stripes, int real_stripes)
 		GFP_NOFS|__GFP_NOFAIL);
 
 	atomic_set(&bbio->error, 0);
-	atomic_set(&bbio->refs, 1);
+	refcount_set(&bbio->refs, 1);
 
 	return bbio;
 }
 
 void btrfs_get_bbio(struct btrfs_bio *bbio)
 {
-	WARN_ON(!atomic_read(&bbio->refs));
-	atomic_inc(&bbio->refs);
+	WARN_ON(!refcount_read(&bbio->refs));
+	refcount_inc(&bbio->refs);
 }
 
 void btrfs_put_bbio(struct btrfs_bio *bbio)
 {
 	if (!bbio)
 		return;
-	if (atomic_dec_and_test(&bbio->refs))
+	if (refcount_dec_and_test(&bbio->refs))
 		kfree(bbio);
 }
 
+/* can REQ_OP_DISCARD be sent with other REQ like REQ_OP_WRITE? */
+/*
+ * Please note that, discard won't be sent to target device of device
+ * replace.
+ */
+static int __btrfs_map_block_for_discard(struct btrfs_fs_info *fs_info,
+					 u64 logical, u64 length,
+					 struct btrfs_bio **bbio_ret)
+{
+	struct extent_map *em;
+	struct map_lookup *map;
+	struct btrfs_bio *bbio;
+	u64 offset;
+	u64 stripe_nr;
+	u64 stripe_nr_end;
+	u64 stripe_end_offset;
+	u64 stripe_cnt;
+	u64 stripe_len;
+	u64 stripe_offset;
+	u64 num_stripes;
+	u32 stripe_index;
+	u32 factor = 0;
+	u32 sub_stripes = 0;
+	u64 stripes_per_dev = 0;
+	u32 remaining_stripes = 0;
+	u32 last_stripe = 0;
+	int ret = 0;
+	int i;
+
+	/* discard always return a bbio */
+	ASSERT(bbio_ret);
+
+	em = get_chunk_map(fs_info, logical, length);
+	if (IS_ERR(em))
+		return PTR_ERR(em);
+
+	map = em->map_lookup;
+	/* we don't discard raid56 yet */
+	if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
+		ret = -EOPNOTSUPP;
+		goto out;
+	}
+
+	offset = logical - em->start;
+	length = min_t(u64, em->len - offset, length);
+
+	stripe_len = map->stripe_len;
+	/*
+	 * stripe_nr counts the total number of stripes we have to stride
+	 * to get to this block
+	 */
+	stripe_nr = div64_u64(offset, stripe_len);
+
+	/* stripe_offset is the offset of this block in its stripe */
+	stripe_offset = offset - stripe_nr * stripe_len;
+
+	stripe_nr_end = round_up(offset + length, map->stripe_len);
+	stripe_nr_end = div64_u64(stripe_nr_end, map->stripe_len);
+	stripe_cnt = stripe_nr_end - stripe_nr;
+	stripe_end_offset = stripe_nr_end * map->stripe_len -
+			    (offset + length);
+	/*
+	 * after this, stripe_nr is the number of stripes on this
+	 * device we have to walk to find the data, and stripe_index is
+	 * the number of our device in the stripe array
+	 */
+	num_stripes = 1;
+	stripe_index = 0;
+	if (map->type & (BTRFS_BLOCK_GROUP_RAID0 |
+			 BTRFS_BLOCK_GROUP_RAID10)) {
+		if (map->type & BTRFS_BLOCK_GROUP_RAID0)
+			sub_stripes = 1;
+		else
+			sub_stripes = map->sub_stripes;
+
+		factor = map->num_stripes / sub_stripes;
+		num_stripes = min_t(u64, map->num_stripes,
+				    sub_stripes * stripe_cnt);
+		stripe_nr = div_u64_rem(stripe_nr, factor, &stripe_index);
+		stripe_index *= sub_stripes;
+		stripes_per_dev = div_u64_rem(stripe_cnt, factor,
+					      &remaining_stripes);
+		div_u64_rem(stripe_nr_end - 1, factor, &last_stripe);
+		last_stripe *= sub_stripes;
+	} else if (map->type & (BTRFS_BLOCK_GROUP_RAID1 |
+				BTRFS_BLOCK_GROUP_DUP)) {
+		num_stripes = map->num_stripes;
+	} else {
+		stripe_nr = div_u64_rem(stripe_nr, map->num_stripes,
+					&stripe_index);
+	}
+
+	bbio = alloc_btrfs_bio(num_stripes, 0);
+	if (!bbio) {
+		ret = -ENOMEM;
+		goto out;
+	}
+
+	for (i = 0; i < num_stripes; i++) {
+		bbio->stripes[i].physical =
+			map->stripes[stripe_index].physical +
+			stripe_offset + stripe_nr * map->stripe_len;
+		bbio->stripes[i].dev = map->stripes[stripe_index].dev;
+
+		if (map->type & (BTRFS_BLOCK_GROUP_RAID0 |
+				 BTRFS_BLOCK_GROUP_RAID10)) {
+			bbio->stripes[i].length = stripes_per_dev *
+				map->stripe_len;
+
+			if (i / sub_stripes < remaining_stripes)
+				bbio->stripes[i].length +=
+					map->stripe_len;
+
+			/*
+			 * Special for the first stripe and
+			 * the last stripe:
+			 *
+			 * |-------|...|-------|
+			 *     |----------|
+			 *    off     end_off
+			 */
+			if (i < sub_stripes)
+				bbio->stripes[i].length -=
+					stripe_offset;
+
+			if (stripe_index >= last_stripe &&
+			    stripe_index <= (last_stripe +
+					     sub_stripes - 1))
+				bbio->stripes[i].length -=
+					stripe_end_offset;
+
+			if (i == sub_stripes - 1)
+				stripe_offset = 0;
+		} else {
+			bbio->stripes[i].length = length;
+		}
+
+		stripe_index++;
+		if (stripe_index == map->num_stripes) {
+			stripe_index = 0;
+			stripe_nr++;
+		}
+	}
+
+	*bbio_ret = bbio;
+	bbio->map_type = map->type;
+	bbio->num_stripes = num_stripes;
+out:
+	free_extent_map(em);
+	return ret;
+}
+
+/*
+ * In dev-replace case, for repair case (that's the only case where the mirror
+ * is selected explicitly when calling btrfs_map_block), blocks left of the
+ * left cursor can also be read from the target drive.
+ *
+ * For REQ_GET_READ_MIRRORS, the target drive is added as the last one to the
+ * array of stripes.
+ * For READ, it also needs to be supported using the same mirror number.
+ *
+ * If the requested block is not left of the left cursor, EIO is returned. This
+ * can happen because btrfs_num_copies() returns one more in the dev-replace
+ * case.
+ */
+static int get_extra_mirror_from_replace(struct btrfs_fs_info *fs_info,
+					 u64 logical, u64 length,
+					 u64 srcdev_devid, int *mirror_num,
+					 u64 *physical)
+{
+	struct btrfs_bio *bbio = NULL;
+	int num_stripes;
+	int index_srcdev = 0;
+	int found = 0;
+	u64 physical_of_found = 0;
+	int i;
+	int ret = 0;
+
+	ret = __btrfs_map_block(fs_info, BTRFS_MAP_GET_READ_MIRRORS,
+				logical, &length, &bbio, 0, 0);
+	if (ret) {
+		ASSERT(bbio == NULL);
+		return ret;
+	}
+
+	num_stripes = bbio->num_stripes;
+	if (*mirror_num > num_stripes) {
+		/*
+		 * BTRFS_MAP_GET_READ_MIRRORS does not contain this mirror,
+		 * that means that the requested area is not left of the left
+		 * cursor
+		 */
+		btrfs_put_bbio(bbio);
+		return -EIO;
+	}
+
+	/*
+	 * process the rest of the function using the mirror_num of the source
+	 * drive. Therefore look it up first.  At the end, patch the device
+	 * pointer to the one of the target drive.
+	 */
+	for (i = 0; i < num_stripes; i++) {
+		if (bbio->stripes[i].dev->devid != srcdev_devid)
+			continue;
+
+		/*
+		 * In case of DUP, in order to keep it simple, only add the
+		 * mirror with the lowest physical address
+		 */
+		if (found &&
+		    physical_of_found <= bbio->stripes[i].physical)
+			continue;
+
+		index_srcdev = i;
+		found = 1;
+		physical_of_found = bbio->stripes[i].physical;
+	}
+
+	btrfs_put_bbio(bbio);
+
+	ASSERT(found);
+	if (!found)
+		return -EIO;
+
+	*mirror_num = index_srcdev + 1;
+	*physical = physical_of_found;
+	return ret;
+}
+
+static void handle_ops_on_dev_replace(enum btrfs_map_op op,
+				      struct btrfs_bio **bbio_ret,
+				      struct btrfs_dev_replace *dev_replace,
+				      int *num_stripes_ret, int *max_errors_ret)
+{
+	struct btrfs_bio *bbio = *bbio_ret;
+	u64 srcdev_devid = dev_replace->srcdev->devid;
+	int tgtdev_indexes = 0;
+	int num_stripes = *num_stripes_ret;
+	int max_errors = *max_errors_ret;
+	int i;
+
+	if (op == BTRFS_MAP_WRITE) {
+		int index_where_to_add;
+
+		/*
+		 * duplicate the write operations while the dev replace
+		 * procedure is running. Since the copying of the old disk to
+		 * the new disk takes place at run time while the filesystem is
+		 * mounted writable, the regular write operations to the old
+		 * disk have to be duplicated to go to the new disk as well.
+		 *
+		 * Note that device->missing is handled by the caller, and that
+		 * the write to the old disk is already set up in the stripes
+		 * array.
+		 */
+		index_where_to_add = num_stripes;
+		for (i = 0; i < num_stripes; i++) {
+			if (bbio->stripes[i].dev->devid == srcdev_devid) {
+				/* write to new disk, too */
+				struct btrfs_bio_stripe *new =
+					bbio->stripes + index_where_to_add;
+				struct btrfs_bio_stripe *old =
+					bbio->stripes + i;
+
+				new->physical = old->physical;
+				new->length = old->length;
+				new->dev = dev_replace->tgtdev;
+				bbio->tgtdev_map[i] = index_where_to_add;
+				index_where_to_add++;
+				max_errors++;
+				tgtdev_indexes++;
+			}
+		}
+		num_stripes = index_where_to_add;
+	} else if (op == BTRFS_MAP_GET_READ_MIRRORS) {
+		int index_srcdev = 0;
+		int found = 0;
+		u64 physical_of_found = 0;
+
+		/*
+		 * During the dev-replace procedure, the target drive can also
+		 * be used to read data in case it is needed to repair a corrupt
+		 * block elsewhere. This is possible if the requested area is
+		 * left of the left cursor. In this area, the target drive is a
+		 * full copy of the source drive.
+		 */
+		for (i = 0; i < num_stripes; i++) {
+			if (bbio->stripes[i].dev->devid == srcdev_devid) {
+				/*
+				 * In case of DUP, in order to keep it simple,
+				 * only add the mirror with the lowest physical
+				 * address
+				 */
+				if (found &&
+				    physical_of_found <=
+				     bbio->stripes[i].physical)
+					continue;
+				index_srcdev = i;
+				found = 1;
+				physical_of_found = bbio->stripes[i].physical;
+			}
+		}
+		if (found) {
+			struct btrfs_bio_stripe *tgtdev_stripe =
+				bbio->stripes + num_stripes;
+
+			tgtdev_stripe->physical = physical_of_found;
+			tgtdev_stripe->length =
+				bbio->stripes[index_srcdev].length;
+			tgtdev_stripe->dev = dev_replace->tgtdev;
+			bbio->tgtdev_map[index_srcdev] = num_stripes;
+
+			tgtdev_indexes++;
+			num_stripes++;
+		}
+	}
+
+	*num_stripes_ret = num_stripes;
+	*max_errors_ret = max_errors;
+	bbio->num_tgtdevs = tgtdev_indexes;
+	*bbio_ret = bbio;
+}
+
+static bool need_full_stripe(enum btrfs_map_op op)
+{
+	return (op == BTRFS_MAP_WRITE || op == BTRFS_MAP_GET_READ_MIRRORS);
+}
+
 static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 			     enum btrfs_map_op op,
 			     u64 logical, u64 *length,
@@ -5322,14 +5639,9 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 {
 	struct extent_map *em;
 	struct map_lookup *map;
-	struct btrfs_mapping_tree *map_tree = &fs_info->mapping_tree;
-	struct extent_map_tree *em_tree = &map_tree->map_tree;
 	u64 offset;
 	u64 stripe_offset;
-	u64 stripe_end_offset;
 	u64 stripe_nr;
-	u64 stripe_nr_orig;
-	u64 stripe_nr_end;
 	u64 stripe_len;
 	u32 stripe_index;
 	int i;
@@ -5345,23 +5657,13 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 	u64 physical_to_patch_in_first_stripe = 0;
 	u64 raid56_full_stripe_start = (u64)-1;
 
-	read_lock(&em_tree->lock);
-	em = lookup_extent_mapping(em_tree, logical, *length);
-	read_unlock(&em_tree->lock);
-
-	if (!em) {
-		btrfs_crit(fs_info, "unable to find logical %llu len %llu",
-			logical, *length);
-		return -EINVAL;
-	}
+	if (op == BTRFS_MAP_DISCARD)
+		return __btrfs_map_block_for_discard(fs_info, logical,
+						     *length, bbio_ret);
 
-	if (em->start > logical || em->start + em->len < logical) {
-		btrfs_crit(fs_info,
-			   "found a bad mapping, wanted %Lu, found %Lu-%Lu",
-			   logical, em->start, em->start + em->len);
-		free_extent_map(em);
-		return -EINVAL;
-	}
+	em = get_chunk_map(fs_info, logical, *length);
+	if (IS_ERR(em))
+		return PTR_ERR(em);
 
 	map = em->map_lookup;
 	offset = logical - em->start;
@@ -5400,14 +5702,7 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 		raid56_full_stripe_start *= full_stripe_len;
 	}
 
-	if (op == BTRFS_MAP_DISCARD) {
-		/* we don't discard raid56 yet */
-		if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK) {
-			ret = -EOPNOTSUPP;
-			goto out;
-		}
-		*length = min_t(u64, em->len - offset, *length);
-	} else if (map->type & BTRFS_BLOCK_GROUP_PROFILE_MASK) {
+	if (map->type & BTRFS_BLOCK_GROUP_PROFILE_MASK) {
 		u64 max_len;
 		/* For writes to RAID[56], allow a full stripeset across all disks.
 		   For other RAID types and for RAID[56] reads, just allow a single
@@ -5438,105 +5733,28 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 		btrfs_dev_replace_set_lock_blocking(dev_replace);
 
 	if (dev_replace_is_ongoing && mirror_num == map->num_stripes + 1 &&
-	    op != BTRFS_MAP_WRITE && op != BTRFS_MAP_DISCARD &&
-	    op != BTRFS_MAP_GET_READ_MIRRORS && dev_replace->tgtdev != NULL) {
-		/*
-		 * in dev-replace case, for repair case (that's the only
-		 * case where the mirror is selected explicitly when
-		 * calling btrfs_map_block), blocks left of the left cursor
-		 * can also be read from the target drive.
-		 * For REQ_GET_READ_MIRRORS, the target drive is added as
-		 * the last one to the array of stripes. For READ, it also
-		 * needs to be supported using the same mirror number.
-		 * If the requested block is not left of the left cursor,
-		 * EIO is returned. This can happen because btrfs_num_copies()
-		 * returns one more in the dev-replace case.
-		 */
-		u64 tmp_length = *length;
-		struct btrfs_bio *tmp_bbio = NULL;
-		int tmp_num_stripes;
-		u64 srcdev_devid = dev_replace->srcdev->devid;
-		int index_srcdev = 0;
-		int found = 0;
-		u64 physical_of_found = 0;
-
-		ret = __btrfs_map_block(fs_info, BTRFS_MAP_GET_READ_MIRRORS,
-			     logical, &tmp_length, &tmp_bbio, 0, 0);
-		if (ret) {
-			WARN_ON(tmp_bbio != NULL);
-			goto out;
-		}
-
-		tmp_num_stripes = tmp_bbio->num_stripes;
-		if (mirror_num > tmp_num_stripes) {
-			/*
-			 * BTRFS_MAP_GET_READ_MIRRORS does not contain this
-			 * mirror, that means that the requested area
-			 * is not left of the left cursor
-			 */
-			ret = -EIO;
-			btrfs_put_bbio(tmp_bbio);
-			goto out;
-		}
-
-		/*
-		 * process the rest of the function using the mirror_num
-		 * of the source drive. Therefore look it up first.
-		 * At the end, patch the device pointer to the one of the
-		 * target drive.
-		 */
-		for (i = 0; i < tmp_num_stripes; i++) {
-			if (tmp_bbio->stripes[i].dev->devid != srcdev_devid)
-				continue;
-
-			/*
-			 * In case of DUP, in order to keep it simple, only add
-			 * the mirror with the lowest physical address
-			 */
-			if (found &&
-			    physical_of_found <= tmp_bbio->stripes[i].physical)
-				continue;
-
-			index_srcdev = i;
-			found = 1;
-			physical_of_found = tmp_bbio->stripes[i].physical;
-		}
-
-		btrfs_put_bbio(tmp_bbio);
-
-		if (!found) {
-			WARN_ON(1);
-			ret = -EIO;
+	    !need_full_stripe(op) && dev_replace->tgtdev != NULL) {
+		ret = get_extra_mirror_from_replace(fs_info, logical, *length,
+						    dev_replace->srcdev->devid,
+						    &mirror_num,
+					    &physical_to_patch_in_first_stripe);
+		if (ret)
 			goto out;
-		}
-
-		mirror_num = index_srcdev + 1;
-		patch_the_first_stripe_for_dev_replace = 1;
-		physical_to_patch_in_first_stripe = physical_of_found;
+		else
+			patch_the_first_stripe_for_dev_replace = 1;
 	} else if (mirror_num > map->num_stripes) {
 		mirror_num = 0;
 	}
 
 	num_stripes = 1;
 	stripe_index = 0;
-	stripe_nr_orig = stripe_nr;
-	stripe_nr_end = ALIGN(offset + *length, map->stripe_len);
-	stripe_nr_end = div_u64(stripe_nr_end, map->stripe_len);
-	stripe_end_offset = stripe_nr_end * map->stripe_len -
-			    (offset + *length);
-
 	if (map->type & BTRFS_BLOCK_GROUP_RAID0) {
-		if (op == BTRFS_MAP_DISCARD)
-			num_stripes = min_t(u64, map->num_stripes,
-					    stripe_nr_end - stripe_nr_orig);
 		stripe_nr = div_u64_rem(stripe_nr, map->num_stripes,
 				&stripe_index);
-		if (op != BTRFS_MAP_WRITE && op != BTRFS_MAP_DISCARD &&
-		    op != BTRFS_MAP_GET_READ_MIRRORS)
+		if (op != BTRFS_MAP_WRITE && op != BTRFS_MAP_GET_READ_MIRRORS)
 			mirror_num = 1;
 	} else if (map->type & BTRFS_BLOCK_GROUP_RAID1) {
-		if (op == BTRFS_MAP_WRITE || op == BTRFS_MAP_DISCARD ||
-		    op == BTRFS_MAP_GET_READ_MIRRORS)
+		if (op == BTRFS_MAP_WRITE || op == BTRFS_MAP_GET_READ_MIRRORS)
 			num_stripes = map->num_stripes;
 		else if (mirror_num)
 			stripe_index = mirror_num - 1;
@@ -5549,8 +5767,7 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 		}
 
 	} else if (map->type & BTRFS_BLOCK_GROUP_DUP) {
-		if (op == BTRFS_MAP_WRITE || op == BTRFS_MAP_DISCARD ||
-		    op == BTRFS_MAP_GET_READ_MIRRORS) {
+		if (op == BTRFS_MAP_WRITE || op == BTRFS_MAP_GET_READ_MIRRORS) {
 			num_stripes = map->num_stripes;
 		} else if (mirror_num) {
 			stripe_index = mirror_num - 1;
@@ -5566,10 +5783,6 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 
 		if (op == BTRFS_MAP_WRITE || op == BTRFS_MAP_GET_READ_MIRRORS)
 			num_stripes = map->sub_stripes;
-		else if (op == BTRFS_MAP_DISCARD)
-			num_stripes = min_t(u64, map->sub_stripes *
-					    (stripe_nr_end - stripe_nr_orig),
-					    map->num_stripes);
 		else if (mirror_num)
 			stripe_index += mirror_num - 1;
 		else {
@@ -5587,7 +5800,7 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 		    (op == BTRFS_MAP_WRITE || op == BTRFS_MAP_GET_READ_MIRRORS ||
 		     mirror_num > 1)) {
 			/* push stripe_nr back to the start of the full stripe */
-			stripe_nr = div_u64(raid56_full_stripe_start,
+			stripe_nr = div64_u64(raid56_full_stripe_start,
 					stripe_len * nr_data_stripes(map));
 
 			/* RAID[56] write or recovery. Return all stripes */
@@ -5612,8 +5825,9 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 			/* We distribute the parity blocks across stripes */
 			div_u64_rem(stripe_nr + stripe_index, map->num_stripes,
 					&stripe_index);
-			if ((op != BTRFS_MAP_WRITE && op != BTRFS_MAP_DISCARD &&
-			    op != BTRFS_MAP_GET_READ_MIRRORS) && mirror_num <= 1)
+			if ((op != BTRFS_MAP_WRITE &&
+			     op != BTRFS_MAP_GET_READ_MIRRORS) &&
+			    mirror_num <= 1)
 				mirror_num = 1;
 		}
 	} else {
@@ -5635,8 +5849,8 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 	}
 
 	num_alloc_stripes = num_stripes;
-	if (dev_replace_is_ongoing) {
-		if (op == BTRFS_MAP_WRITE || op == BTRFS_MAP_DISCARD)
+	if (dev_replace_is_ongoing && dev_replace->tgtdev != NULL) {
+		if (op == BTRFS_MAP_WRITE)
 			num_alloc_stripes <<= 1;
 		if (op == BTRFS_MAP_GET_READ_MIRRORS)
 			num_alloc_stripes++;
@@ -5648,14 +5862,12 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 		ret = -ENOMEM;
 		goto out;
 	}
-	if (dev_replace_is_ongoing)
+	if (dev_replace_is_ongoing && dev_replace->tgtdev != NULL)
 		bbio->tgtdev_map = (int *)(bbio->stripes + num_alloc_stripes);
 
 	/* build raid_map */
-	if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK &&
-	    need_raid_map &&
-	    ((op == BTRFS_MAP_WRITE || op == BTRFS_MAP_GET_READ_MIRRORS) ||
-	    mirror_num > 1)) {
+	if (map->type & BTRFS_BLOCK_GROUP_RAID56_MASK && need_raid_map &&
+	    (need_full_stripe(op) || mirror_num > 1)) {
 		u64 tmp;
 		unsigned rot;
 
@@ -5679,173 +5891,27 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 				RAID6_Q_STRIPE;
 	}
 
-	if (op == BTRFS_MAP_DISCARD) {
-		u32 factor = 0;
-		u32 sub_stripes = 0;
-		u64 stripes_per_dev = 0;
-		u32 remaining_stripes = 0;
-		u32 last_stripe = 0;
 
-		if (map->type &
-		    (BTRFS_BLOCK_GROUP_RAID0 | BTRFS_BLOCK_GROUP_RAID10)) {
-			if (map->type & BTRFS_BLOCK_GROUP_RAID0)
-				sub_stripes = 1;
-			else
-				sub_stripes = map->sub_stripes;
-
-			factor = map->num_stripes / sub_stripes;
-			stripes_per_dev = div_u64_rem(stripe_nr_end -
-						      stripe_nr_orig,
-						      factor,
-						      &remaining_stripes);
-			div_u64_rem(stripe_nr_end - 1, factor, &last_stripe);
-			last_stripe *= sub_stripes;
-		}
-
-		for (i = 0; i < num_stripes; i++) {
-			bbio->stripes[i].physical =
-				map->stripes[stripe_index].physical +
-				stripe_offset + stripe_nr * map->stripe_len;
-			bbio->stripes[i].dev = map->stripes[stripe_index].dev;
-
-			if (map->type & (BTRFS_BLOCK_GROUP_RAID0 |
-					 BTRFS_BLOCK_GROUP_RAID10)) {
-				bbio->stripes[i].length = stripes_per_dev *
-							  map->stripe_len;
-
-				if (i / sub_stripes < remaining_stripes)
-					bbio->stripes[i].length +=
-						map->stripe_len;
-
-				/*
-				 * Special for the first stripe and
-				 * the last stripe:
-				 *
-				 * |-------|...|-------|
-				 *     |----------|
-				 *    off     end_off
-				 */
-				if (i < sub_stripes)
-					bbio->stripes[i].length -=
-						stripe_offset;
-
-				if (stripe_index >= last_stripe &&
-				    stripe_index <= (last_stripe +
-						     sub_stripes - 1))
-					bbio->stripes[i].length -=
-						stripe_end_offset;
-
-				if (i == sub_stripes - 1)
-					stripe_offset = 0;
-			} else
-				bbio->stripes[i].length = *length;
-
-			stripe_index++;
-			if (stripe_index == map->num_stripes) {
-				/* This could only happen for RAID0/10 */
-				stripe_index = 0;
-				stripe_nr++;
-			}
-		}
-	} else {
-		for (i = 0; i < num_stripes; i++) {
-			bbio->stripes[i].physical =
-				map->stripes[stripe_index].physical +
-				stripe_offset +
-				stripe_nr * map->stripe_len;
-			bbio->stripes[i].dev =
-				map->stripes[stripe_index].dev;
-			stripe_index++;
-		}
+	for (i = 0; i < num_stripes; i++) {
+		bbio->stripes[i].physical =
+			map->stripes[stripe_index].physical +
+			stripe_offset +
+			stripe_nr * map->stripe_len;
+		bbio->stripes[i].dev =
+			map->stripes[stripe_index].dev;
+		stripe_index++;
 	}
 
-	if (op == BTRFS_MAP_WRITE || op == BTRFS_MAP_GET_READ_MIRRORS)
+	if (need_full_stripe(op))
 		max_errors = btrfs_chunk_max_errors(map);
 
 	if (bbio->raid_map)
 		sort_parity_stripes(bbio, num_stripes);
 
-	tgtdev_indexes = 0;
-	if (dev_replace_is_ongoing &&
-	   (op == BTRFS_MAP_WRITE || op == BTRFS_MAP_DISCARD) &&
-	    dev_replace->tgtdev != NULL) {
-		int index_where_to_add;
-		u64 srcdev_devid = dev_replace->srcdev->devid;
-
-		/*
-		 * duplicate the write operations while the dev replace
-		 * procedure is running. Since the copying of the old disk
-		 * to the new disk takes place at run time while the
-		 * filesystem is mounted writable, the regular write
-		 * operations to the old disk have to be duplicated to go
-		 * to the new disk as well.
-		 * Note that device->missing is handled by the caller, and
-		 * that the write to the old disk is already set up in the
-		 * stripes array.
-		 */
-		index_where_to_add = num_stripes;
-		for (i = 0; i < num_stripes; i++) {
-			if (bbio->stripes[i].dev->devid == srcdev_devid) {
-				/* write to new disk, too */
-				struct btrfs_bio_stripe *new =
-					bbio->stripes + index_where_to_add;
-				struct btrfs_bio_stripe *old =
-					bbio->stripes + i;
-
-				new->physical = old->physical;
-				new->length = old->length;
-				new->dev = dev_replace->tgtdev;
-				bbio->tgtdev_map[i] = index_where_to_add;
-				index_where_to_add++;
-				max_errors++;
-				tgtdev_indexes++;
-			}
-		}
-		num_stripes = index_where_to_add;
-	} else if (dev_replace_is_ongoing &&
-		   op == BTRFS_MAP_GET_READ_MIRRORS &&
-		   dev_replace->tgtdev != NULL) {
-		u64 srcdev_devid = dev_replace->srcdev->devid;
-		int index_srcdev = 0;
-		int found = 0;
-		u64 physical_of_found = 0;
-
-		/*
-		 * During the dev-replace procedure, the target drive can
-		 * also be used to read data in case it is needed to repair
-		 * a corrupt block elsewhere. This is possible if the
-		 * requested area is left of the left cursor. In this area,
-		 * the target drive is a full copy of the source drive.
-		 */
-		for (i = 0; i < num_stripes; i++) {
-			if (bbio->stripes[i].dev->devid == srcdev_devid) {
-				/*
-				 * In case of DUP, in order to keep it
-				 * simple, only add the mirror with the
-				 * lowest physical address
-				 */
-				if (found &&
-				    physical_of_found <=
-				     bbio->stripes[i].physical)
-					continue;
-				index_srcdev = i;
-				found = 1;
-				physical_of_found = bbio->stripes[i].physical;
-			}
-		}
-		if (found) {
-			struct btrfs_bio_stripe *tgtdev_stripe =
-				bbio->stripes + num_stripes;
-
-			tgtdev_stripe->physical = physical_of_found;
-			tgtdev_stripe->length =
-				bbio->stripes[index_srcdev].length;
-			tgtdev_stripe->dev = dev_replace->tgtdev;
-			bbio->tgtdev_map[index_srcdev] = num_stripes;
-
-			tgtdev_indexes++;
-			num_stripes++;
-		}
+	if (dev_replace_is_ongoing && dev_replace->tgtdev != NULL &&
+	    need_full_stripe(op)) {
+		handle_ops_on_dev_replace(op, &bbio, dev_replace, &num_stripes,
+					  &max_errors);
 	}
 
 	*bbio_ret = bbio;
@@ -5853,7 +5919,6 @@ static int __btrfs_map_block(struct btrfs_fs_info *fs_info,
 	bbio->num_stripes = num_stripes;
 	bbio->max_errors = max_errors;
 	bbio->mirror_num = mirror_num;
-	bbio->num_tgtdevs = tgtdev_indexes;
 
 	/*
 	 * this is the case that REQ_READ && dev_replace_is_ongoing &&
@@ -5886,19 +5951,15 @@ int btrfs_map_block(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
 /* For Scrub/replace */
 int btrfs_map_sblock(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
 		     u64 logical, u64 *length,
-		     struct btrfs_bio **bbio_ret, int mirror_num,
-		     int need_raid_map)
+		     struct btrfs_bio **bbio_ret)
 {
-	return __btrfs_map_block(fs_info, op, logical, length, bbio_ret,
-				 mirror_num, need_raid_map);
+	return __btrfs_map_block(fs_info, op, logical, length, bbio_ret, 0, 1);
 }
 
 int btrfs_rmap_block(struct btrfs_fs_info *fs_info,
 		     u64 chunk_start, u64 physical, u64 devid,
 		     u64 **logical, int *naddrs, int *stripe_len)
 {
-	struct btrfs_mapping_tree *map_tree = &fs_info->mapping_tree;
-	struct extent_map_tree *em_tree = &map_tree->map_tree;
 	struct extent_map *em;
 	struct map_lookup *map;
 	u64 *buf;
@@ -5908,24 +5969,11 @@ int btrfs_rmap_block(struct btrfs_fs_info *fs_info,
 	u64 rmap_len;
 	int i, j, nr = 0;
 
-	read_lock(&em_tree->lock);
-	em = lookup_extent_mapping(em_tree, chunk_start, 1);
-	read_unlock(&em_tree->lock);
-
-	if (!em) {
-		btrfs_err(fs_info, "couldn't find em for chunk %Lu",
-			chunk_start);
+	em = get_chunk_map(fs_info, chunk_start, 1);
+	if (IS_ERR(em))
 		return -EIO;
-	}
 
-	if (em->start != chunk_start) {
-		btrfs_err(fs_info, "bad chunk start, em=%Lu, wanted=%Lu",
-		       em->start, chunk_start);
-		free_extent_map(em);
-		return -EIO;
-	}
 	map = em->map_lookup;
-
 	length = em->len;
 	rmap_len = map->stripe_len;
 
@@ -5949,7 +5997,7 @@ int btrfs_rmap_block(struct btrfs_fs_info *fs_info,
 			continue;
 
 		stripe_nr = physical - map->stripes[i].physical;
-		stripe_nr = div_u64(stripe_nr, map->stripe_len);
+		stripe_nr = div64_u64(stripe_nr, map->stripe_len);
 
 		if (map->type & BTRFS_BLOCK_GROUP_RAID10) {
 			stripe_nr = stripe_nr * map->num_stripes + i;

fs/btrfs/volumes.h

@@ -123,7 +123,6 @@ struct btrfs_device {
 	struct list_head resized_list;
 
 	/* for sending down flush barriers */
-	int nobarriers;
 	struct bio *flush_bio;
 	struct completion flush_wait;
 
@@ -298,7 +297,7 @@ struct btrfs_bio;
 typedef void (btrfs_bio_end_io_t) (struct btrfs_bio *bio, int err);
 
 struct btrfs_bio {
-	atomic_t refs;
+	refcount_t refs;
 	atomic_t stripes_pending;
 	struct btrfs_fs_info *fs_info;
 	u64 map_type; /* get from map_lookup->type */
@@ -400,8 +399,7 @@ int btrfs_map_block(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
 		    struct btrfs_bio **bbio_ret, int mirror_num);
 int btrfs_map_sblock(struct btrfs_fs_info *fs_info, enum btrfs_map_op op,
 		     u64 logical, u64 *length,
-		     struct btrfs_bio **bbio_ret, int mirror_num,
-		     int need_raid_map);
+		     struct btrfs_bio **bbio_ret);
 int btrfs_rmap_block(struct btrfs_fs_info *fs_info,
 		     u64 chunk_start, u64 physical, u64 devid,
 		     u64 **logical, int *naddrs, int *stripe_len);
@@ -475,7 +473,7 @@ void btrfs_destroy_dev_replace_tgtdev(struct btrfs_fs_info *fs_info,
 void btrfs_init_dev_replace_tgtdev_for_resume(struct btrfs_fs_info *fs_info,
 					      struct btrfs_device *tgtdev);
 void btrfs_scratch_superblocks(struct block_device *bdev, const char *device_path);
-int btrfs_is_parity_mirror(struct btrfs_mapping_tree *map_tree,
+int btrfs_is_parity_mirror(struct btrfs_fs_info *fs_info,
 			   u64 logical, u64 len, int mirror_num);
 unsigned long btrfs_full_stripe_len(struct btrfs_fs_info *fs_info,
 				    struct btrfs_mapping_tree *map_tree,

include/trace/events/btrfs.h

@@ -12,6 +12,7 @@ struct btrfs_root;
 struct btrfs_fs_info;
 struct btrfs_inode;
 struct extent_map;
+struct btrfs_file_extent_item;
 struct btrfs_ordered_extent;
 struct btrfs_delayed_ref_node;
 struct btrfs_delayed_tree_ref;
@@ -24,6 +25,7 @@ struct extent_buffer;
 struct btrfs_work;
 struct __btrfs_workqueue;
 struct btrfs_qgroup_extent_record;
+struct btrfs_qgroup;
 
 #define show_ref_type(type)						\
 	__print_symbolic(type,						\
@@ -54,6 +56,12 @@ struct btrfs_qgroup_extent_record;
 	      (obj >= BTRFS_ROOT_TREE_OBJECTID &&			\
 	       obj <= BTRFS_QUOTA_TREE_OBJECTID)) ? __show_root_type(obj) : "-"
 
+#define show_fi_type(type)						\
+	__print_symbolic(type,						\
+		 { BTRFS_FILE_EXTENT_INLINE,	"INLINE" },		\
+		 { BTRFS_FILE_EXTENT_REG,	"REG"	 },		\
+		 { BTRFS_FILE_EXTENT_PREALLOC,	"PREALLOC"})
+
 #define BTRFS_GROUP_FLAGS	\
 	{ BTRFS_BLOCK_GROUP_DATA,	"DATA"},	\
 	{ BTRFS_BLOCK_GROUP_SYSTEM,	"SYSTEM"},	\
@@ -213,7 +221,7 @@ TRACE_EVENT_CONDITION(btrfs_get_extent,
 		__entry->block_start	= map->block_start;
 		__entry->block_len	= map->block_len;
 		__entry->flags		= map->flags;
-		__entry->refs		= atomic_read(&map->refs);
+		__entry->refs		= refcount_read(&map->refs);
 		__entry->compress_type	= map->compress_type;
 	),
 
@@ -232,6 +240,138 @@ TRACE_EVENT_CONDITION(btrfs_get_extent,
 		  __entry->refs, __entry->compress_type)
 );
 
+/* file extent item */
+DECLARE_EVENT_CLASS(btrfs__file_extent_item_regular,
+
+	TP_PROTO(struct btrfs_inode *bi, struct extent_buffer *l,
+		 struct btrfs_file_extent_item *fi, u64 start),
+
+	TP_ARGS(bi, l, fi, start),
+
+	TP_STRUCT__entry_btrfs(
+		__field(	u64,	root_obj	)
+		__field(	u64,	ino		)
+		__field(	loff_t,	isize		)
+		__field(	u64,	disk_isize	)
+		__field(	u64,	num_bytes	)
+		__field(	u64,	ram_bytes	)
+		__field(	u64,	disk_bytenr	)
+		__field(	u64,	disk_num_bytes	)
+		__field(	u64,	extent_offset	)
+		__field(	u8,	extent_type	)
+		__field(	u8,	compression	)
+		__field(	u64,	extent_start	)
+		__field(	u64,	extent_end	)
+	),
+
+	TP_fast_assign_btrfs(bi->root->fs_info,
+		__entry->root_obj	= bi->root->objectid;
+		__entry->ino		= btrfs_ino(bi);
+		__entry->isize		= bi->vfs_inode.i_size;
+		__entry->disk_isize	= bi->disk_i_size;
+		__entry->num_bytes	= btrfs_file_extent_num_bytes(l, fi);
+		__entry->ram_bytes	= btrfs_file_extent_ram_bytes(l, fi);
+		__entry->disk_bytenr	= btrfs_file_extent_disk_bytenr(l, fi);
+		__entry->disk_num_bytes	= btrfs_file_extent_disk_num_bytes(l, fi);
+		__entry->extent_offset	= btrfs_file_extent_offset(l, fi);
+		__entry->extent_type	= btrfs_file_extent_type(l, fi);
+		__entry->compression	= btrfs_file_extent_compression(l, fi);
+		__entry->extent_start	= start;
+		__entry->extent_end	= (start + __entry->num_bytes);
+	),
+
+	TP_printk_btrfs(
+		"root=%llu(%s) inode=%llu size=%llu disk_isize=%llu "
+		"file extent range=[%llu %llu] "
+		"(num_bytes=%llu ram_bytes=%llu disk_bytenr=%llu "
+		"disk_num_bytes=%llu extent_offset=%llu type=%s "
+		"compression=%u",
+		show_root_type(__entry->root_obj), __entry->ino,
+		__entry->isize,
+		__entry->disk_isize, __entry->extent_start,
+		__entry->extent_end, __entry->num_bytes, __entry->ram_bytes,
+		__entry->disk_bytenr, __entry->disk_num_bytes,
+		__entry->extent_offset, show_fi_type(__entry->extent_type),
+		__entry->compression)
+);
+
+DECLARE_EVENT_CLASS(
+	btrfs__file_extent_item_inline,
+
+	TP_PROTO(struct btrfs_inode *bi, struct extent_buffer *l,
+		 struct btrfs_file_extent_item *fi, int slot, u64 start),
+
+	TP_ARGS(bi, l, fi, slot,  start),
+
+	TP_STRUCT__entry_btrfs(
+		__field(	u64,	root_obj	)
+		__field(	u64,	ino		)
+		__field(	loff_t,	isize		)
+		__field(	u64,	disk_isize	)
+		__field(	u8,	extent_type	)
+		__field(	u8,	compression	)
+		__field(	u64,	extent_start	)
+		__field(	u64,	extent_end	)
+	),
+
+	TP_fast_assign_btrfs(
+		bi->root->fs_info,
+		__entry->root_obj	= bi->root->objectid;
+		__entry->ino		= btrfs_ino(bi);
+		__entry->isize		= bi->vfs_inode.i_size;
+		__entry->disk_isize	= bi->disk_i_size;
+		__entry->extent_type	= btrfs_file_extent_type(l, fi);
+		__entry->compression	= btrfs_file_extent_compression(l, fi);
+		__entry->extent_start	= start;
+		__entry->extent_end	= (start + btrfs_file_extent_inline_len(l, slot, fi));
+	),
+
+	TP_printk_btrfs(
+		"root=%llu(%s) inode=%llu size=%llu disk_isize=%llu "
+		"file extent range=[%llu %llu] "
+		"extent_type=%s compression=%u",
+		show_root_type(__entry->root_obj), __entry->ino, __entry->isize,
+		__entry->disk_isize, __entry->extent_start,
+		__entry->extent_end, show_fi_type(__entry->extent_type),
+		__entry->compression)
+);
+
+DEFINE_EVENT(
+	btrfs__file_extent_item_regular, btrfs_get_extent_show_fi_regular,
+
+	TP_PROTO(struct btrfs_inode *bi, struct extent_buffer *l,
+		 struct btrfs_file_extent_item *fi, u64 start),
+
+	TP_ARGS(bi, l, fi, start)
+);
+
+DEFINE_EVENT(
+	btrfs__file_extent_item_regular, btrfs_truncate_show_fi_regular,
+
+	TP_PROTO(struct btrfs_inode *bi, struct extent_buffer *l,
+		 struct btrfs_file_extent_item *fi, u64 start),
+
+	TP_ARGS(bi, l, fi, start)
+);
+
+DEFINE_EVENT(
+	btrfs__file_extent_item_inline, btrfs_get_extent_show_fi_inline,
+
+	TP_PROTO(struct btrfs_inode *bi, struct extent_buffer *l,
+		 struct btrfs_file_extent_item *fi, int slot, u64 start),
+
+	TP_ARGS(bi, l, fi, slot, start)
+);
+
+DEFINE_EVENT(
+	btrfs__file_extent_item_inline, btrfs_truncate_show_fi_inline,
+
+	TP_PROTO(struct btrfs_inode *bi, struct extent_buffer *l,
+		 struct btrfs_file_extent_item *fi, int slot, u64 start),
+
+	TP_ARGS(bi, l, fi, slot, start)
+);
+
 #define show_ordered_flags(flags)					   \
 	__print_flags(flags, "|",					   \
 		{ (1 << BTRFS_ORDERED_IO_DONE), 	"IO_DONE" 	}, \
@@ -275,7 +415,7 @@ DECLARE_EVENT_CLASS(btrfs__ordered_extent,
 		__entry->bytes_left	= ordered->bytes_left;
 		__entry->flags		= ordered->flags;
 		__entry->compress_type	= ordered->compress_type;
-		__entry->refs		= atomic_read(&ordered->refs);
+		__entry->refs		= refcount_read(&ordered->refs);
 		__entry->root_objectid	=
 				BTRFS_I(inode)->root->root_key.objectid;
 		__entry->truncated_len	= ordered->truncated_len;
@@ -1475,6 +1615,49 @@ TRACE_EVENT(qgroup_update_counters,
 		  __entry->cur_new_count)
 );
 
+TRACE_EVENT(qgroup_update_reserve,
+
+	TP_PROTO(struct btrfs_fs_info *fs_info, struct btrfs_qgroup *qgroup,
+		 s64 diff),
+
+	TP_ARGS(fs_info, qgroup, diff),
+
+	TP_STRUCT__entry_btrfs(
+		__field(	u64,	qgid			)
+		__field(	u64,	cur_reserved		)
+		__field(	s64,	diff			)
+	),
+
+	TP_fast_assign_btrfs(fs_info,
+		__entry->qgid		= qgroup->qgroupid;
+		__entry->cur_reserved	= qgroup->reserved;
+		__entry->diff		= diff;
+	),
+
+	TP_printk_btrfs("qgid=%llu cur_reserved=%llu diff=%lld",
+		__entry->qgid, __entry->cur_reserved, __entry->diff)
+);
+
+TRACE_EVENT(qgroup_meta_reserve,
+
+	TP_PROTO(struct btrfs_root *root, s64 diff),
+
+	TP_ARGS(root, diff),
+
+	TP_STRUCT__entry_btrfs(
+		__field(	u64,	refroot			)
+		__field(	s64,	diff			)
+	),
+
+	TP_fast_assign_btrfs(root->fs_info,
+		__entry->refroot	= root->objectid;
+		__entry->diff		= diff;
+	),
+
+	TP_printk_btrfs("refroot=%llu(%s) diff=%lld",
+		show_root_type(__entry->refroot), __entry->diff)
+);
+
 #endif /* _TRACE_BTRFS_H */
 
 /* This part must be outside protection */

include/uapi/linux/btrfs.h

@@ -291,10 +291,10 @@ struct btrfs_ioctl_feature_flags {
 struct btrfs_balance_args {
 	__u64 profiles;
 	union {
-		__le64 usage;
+		__u64 usage;
 		struct {
-			__le32 usage_min;
-			__le32 usage_max;
+			__u32 usage_min;
+			__u32 usage_max;
 		};
 	};
 	__u64 devid;
@@ -324,8 +324,8 @@ struct btrfs_balance_args {
 	 * Process chunks that cross stripes_min..stripes_max devices,
 	 * BTRFS_BALANCE_ARGS_STRIPES_RANGE
 	 */
-	__le32 stripes_min;
-	__le32 stripes_max;
+	__u32 stripes_min;
+	__u32 stripes_max;
 
 	__u64 unused[6];
 } __attribute__ ((__packed__));