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

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/btrfs-unstable:
  Btrfs: make sure fallocate properly starts a transaction
  Btrfs: make metadata chunks smaller
  Btrfs: Show discard option in /proc/mounts
  Btrfs: deny sys_link across subvolumes.
  Btrfs: fail mount on bad mount options
  Btrfs: don't add extent 0 to the free space cache v2
  Btrfs: Fix per root used space accounting
  Btrfs: Fix btrfs_drop_extent_cache for skip pinned case
  Btrfs: Add delayed iput
  Btrfs: Pass transaction handle to security and ACL initialization functions
  Btrfs: Make truncate(2) more ENOSPC friendly
  Btrfs: Make fallocate(2) more ENOSPC friendly
  Btrfs: Avoid orphan inodes cleanup during committing transaction
  Btrfs: Avoid orphan inodes cleanup while replaying log
  Btrfs: Fix disk_i_size update corner case
  Btrfs: Rewrite btrfs_drop_extents
  Btrfs: Add btrfs_duplicate_item
  Btrfs: Avoid superfluous tree-log writeout

fs/btrfs/acl.c

@@ -94,7 +94,8 @@ static int btrfs_xattr_acl_get(struct dentry *dentry, const char *name,
 /*
  * Needs to be called with fs_mutex held
  */
-static int btrfs_set_acl(struct inode *inode, struct posix_acl *acl, int type)
+static int btrfs_set_acl(struct btrfs_trans_handle *trans,
+			 struct inode *inode, struct posix_acl *acl, int type)
 {
 	int ret, size = 0;
 	const char *name;
@@ -140,8 +141,7 @@ static int btrfs_set_acl(struct inode *inode, struct posix_acl *acl, int type)
 			goto out;
 	}
 
-	ret = __btrfs_setxattr(inode, name, value, size, 0);
-
+	ret = __btrfs_setxattr(trans, inode, name, value, size, 0);
 out:
 	kfree(value);
 
@@ -154,7 +154,7 @@ static int btrfs_set_acl(struct inode *inode, struct posix_acl *acl, int type)
 static int btrfs_xattr_acl_set(struct dentry *dentry, const char *name,
 		const void *value, size_t size, int flags, int type)
 {
-	int ret = 0;
+	int ret;
 	struct posix_acl *acl = NULL;
 
 	if (value) {
@@ -167,7 +167,7 @@ static int btrfs_xattr_acl_set(struct dentry *dentry, const char *name,
 		}
 	}
 
-	ret = btrfs_set_acl(dentry->d_inode, acl, type);
+	ret = btrfs_set_acl(NULL, dentry->d_inode, acl, type);
 
 	posix_acl_release(acl);
 
@@ -196,7 +196,8 @@ int btrfs_check_acl(struct inode *inode, int mask)
  * stuff has been fixed to work with that.  If the locking stuff changes, we
  * need to re-evaluate the acl locking stuff.
  */
-int btrfs_init_acl(struct inode *inode, struct inode *dir)
+int btrfs_init_acl(struct btrfs_trans_handle *trans,
+		   struct inode *inode, struct inode *dir)
 {
 	struct posix_acl *acl = NULL;
 	int ret = 0;
@@ -221,7 +222,8 @@ int btrfs_init_acl(struct inode *inode, struct inode *dir)
 		mode_t mode;
 
 		if (S_ISDIR(inode->i_mode)) {
-			ret = btrfs_set_acl(inode, acl, ACL_TYPE_DEFAULT);
+			ret = btrfs_set_acl(trans, inode, acl,
+					    ACL_TYPE_DEFAULT);
 			if (ret)
 				goto failed;
 		}
@@ -236,7 +238,7 @@ int btrfs_init_acl(struct inode *inode, struct inode *dir)
 			inode->i_mode = mode;
 			if (ret > 0) {
 				/* we need an acl */
-				ret = btrfs_set_acl(inode, clone,
+				ret = btrfs_set_acl(trans, inode, clone,
 						    ACL_TYPE_ACCESS);
 			}
 		}
@@ -269,7 +271,7 @@ int btrfs_acl_chmod(struct inode *inode)
 
 	ret = posix_acl_chmod_masq(clone, inode->i_mode);
 	if (!ret)
-		ret = btrfs_set_acl(inode, clone, ACL_TYPE_ACCESS);
+		ret = btrfs_set_acl(NULL, inode, clone, ACL_TYPE_ACCESS);
 
 	posix_acl_release(clone);
 
@@ -297,7 +299,8 @@ int btrfs_acl_chmod(struct inode *inode)
 	return 0;
 }
 
-int btrfs_init_acl(struct inode *inode, struct inode *dir)
+int btrfs_init_acl(struct btrfs_trans_handle *trans,
+		   struct inode *inode, struct inode *dir)
 {
 	return 0;
 }

fs/btrfs/btrfs_inode.h

@@ -44,9 +44,6 @@ struct btrfs_inode {
 	 */
 	struct extent_io_tree io_failure_tree;
 
-	/* held while inesrting or deleting extents from files */
-	struct mutex extent_mutex;
-
 	/* held while logging the inode in tree-log.c */
 	struct mutex log_mutex;
 
@@ -166,7 +163,7 @@ static inline struct btrfs_inode *BTRFS_I(struct inode *inode)
 
 static inline void btrfs_i_size_write(struct inode *inode, u64 size)
 {
-	inode->i_size = size;
+	i_size_write(inode, size);
 	BTRFS_I(inode)->disk_i_size = size;
 }
 

fs/btrfs/ctree.h

@@ -310,6 +310,9 @@ struct btrfs_header {
 #define BTRFS_MAX_INLINE_DATA_SIZE(r) (BTRFS_LEAF_DATA_SIZE(r) - \
 					sizeof(struct btrfs_item) - \
 					sizeof(struct btrfs_file_extent_item))
+#define BTRFS_MAX_XATTR_SIZE(r)	(BTRFS_LEAF_DATA_SIZE(r) - \
+				 sizeof(struct btrfs_item) -\
+				 sizeof(struct btrfs_dir_item))
 
 
 /*
@@ -859,8 +862,9 @@ struct btrfs_fs_info {
 	struct mutex ordered_operations_mutex;
 	struct rw_semaphore extent_commit_sem;
 
-	struct rw_semaphore subvol_sem;
+	struct rw_semaphore cleanup_work_sem;
 
+	struct rw_semaphore subvol_sem;
 	struct srcu_struct subvol_srcu;
 
 	struct list_head trans_list;
@@ -868,6 +872,9 @@ struct btrfs_fs_info {
 	struct list_head dead_roots;
 	struct list_head caching_block_groups;
 
+	spinlock_t delayed_iput_lock;
+	struct list_head delayed_iputs;
+
 	atomic_t nr_async_submits;
 	atomic_t async_submit_draining;
 	atomic_t nr_async_bios;
@@ -1034,12 +1041,12 @@ struct btrfs_root {
 	int ref_cows;
 	int track_dirty;
 	int in_radix;
+	int clean_orphans;
 
 	u64 defrag_trans_start;
 	struct btrfs_key defrag_progress;
 	struct btrfs_key defrag_max;
 	int defrag_running;
-	int defrag_level;
 	char *name;
 	int in_sysfs;
 
@@ -1975,6 +1982,10 @@ struct extent_buffer *btrfs_alloc_free_block(struct btrfs_trans_handle *trans,
 					u64 parent, u64 root_objectid,
 					struct btrfs_disk_key *key, int level,
 					u64 hint, u64 empty_size);
+int btrfs_free_tree_block(struct btrfs_trans_handle *trans,
+			  struct btrfs_root *root,
+			  u64 bytenr, u32 blocksize,
+			  u64 parent, u64 root_objectid, int level);
 struct extent_buffer *btrfs_init_new_buffer(struct btrfs_trans_handle *trans,
 					    struct btrfs_root *root,
 					    u64 bytenr, u32 blocksize,
@@ -2089,6 +2100,10 @@ int btrfs_split_item(struct btrfs_trans_handle *trans,
 		     struct btrfs_path *path,
 		     struct btrfs_key *new_key,
 		     unsigned long split_offset);
+int btrfs_duplicate_item(struct btrfs_trans_handle *trans,
+			 struct btrfs_root *root,
+			 struct btrfs_path *path,
+			 struct btrfs_key *new_key);
 int btrfs_search_slot(struct btrfs_trans_handle *trans, struct btrfs_root
 		      *root, struct btrfs_key *key, struct btrfs_path *p, int
 		      ins_len, int cow);
@@ -2196,9 +2211,10 @@ int btrfs_delete_one_dir_name(struct btrfs_trans_handle *trans,
 			      struct btrfs_path *path,
 			      struct btrfs_dir_item *di);
 int btrfs_insert_xattr_item(struct btrfs_trans_handle *trans,
-			    struct btrfs_root *root, const char *name,
-			    u16 name_len, const void *data, u16 data_len,
-			    u64 dir);
+			    struct btrfs_root *root,
+			    struct btrfs_path *path, u64 objectid,
+			    const char *name, u16 name_len,
+			    const void *data, u16 data_len);
 struct btrfs_dir_item *btrfs_lookup_xattr(struct btrfs_trans_handle *trans,
 					  struct btrfs_root *root,
 					  struct btrfs_path *path, u64 dir,
@@ -2292,7 +2308,7 @@ int btrfs_truncate_inode_items(struct btrfs_trans_handle *trans,
 			       struct inode *inode, u64 new_size,
 			       u32 min_type);
 
-int btrfs_start_delalloc_inodes(struct btrfs_root *root);
+int btrfs_start_delalloc_inodes(struct btrfs_root *root, int delay_iput);
 int btrfs_set_extent_delalloc(struct inode *inode, u64 start, u64 end);
 int btrfs_writepages(struct address_space *mapping,
 		     struct writeback_control *wbc);
@@ -2332,6 +2348,8 @@ int btrfs_orphan_del(struct btrfs_trans_handle *trans, struct inode *inode);
 void btrfs_orphan_cleanup(struct btrfs_root *root);
 int btrfs_cont_expand(struct inode *inode, loff_t size);
 int btrfs_invalidate_inodes(struct btrfs_root *root);
+void btrfs_add_delayed_iput(struct inode *inode);
+void btrfs_run_delayed_iputs(struct btrfs_root *root);
 extern const struct dentry_operations btrfs_dentry_operations;
 
 /* ioctl.c */
@@ -2345,12 +2363,9 @@ int btrfs_drop_extent_cache(struct inode *inode, u64 start, u64 end,
 			    int skip_pinned);
 int btrfs_check_file(struct btrfs_root *root, struct inode *inode);
 extern const struct file_operations btrfs_file_operations;
-int btrfs_drop_extents(struct btrfs_trans_handle *trans,
-		       struct btrfs_root *root, struct inode *inode,
-		       u64 start, u64 end, u64 locked_end,
-		       u64 inline_limit, u64 *hint_block, int drop_cache);
+int btrfs_drop_extents(struct btrfs_trans_handle *trans, struct inode *inode,
+		       u64 start, u64 end, u64 *hint_byte, int drop_cache);
 int btrfs_mark_extent_written(struct btrfs_trans_handle *trans,
-			      struct btrfs_root *root,
 			      struct inode *inode, u64 start, u64 end);
 int btrfs_release_file(struct inode *inode, struct file *file);
 
@@ -2380,7 +2395,8 @@ int btrfs_check_acl(struct inode *inode, int mask);
 #else
 #define btrfs_check_acl NULL
 #endif
-int btrfs_init_acl(struct inode *inode, struct inode *dir);
+int btrfs_init_acl(struct btrfs_trans_handle *trans,
+		   struct inode *inode, struct inode *dir);
 int btrfs_acl_chmod(struct inode *inode);
 
 /* relocation.c */

fs/btrfs/dir-item.c

@@ -68,12 +68,12 @@ static struct btrfs_dir_item *insert_with_overflow(struct btrfs_trans_handle
  * into the tree
  */
 int btrfs_insert_xattr_item(struct btrfs_trans_handle *trans,
-			    struct btrfs_root *root, const char *name,
-			    u16 name_len, const void *data, u16 data_len,
-			    u64 dir)
+			    struct btrfs_root *root,
+			    struct btrfs_path *path, u64 objectid,
+			    const char *name, u16 name_len,
+			    const void *data, u16 data_len)
 {
 	int ret = 0;
-	struct btrfs_path *path;
 	struct btrfs_dir_item *dir_item;
 	unsigned long name_ptr, data_ptr;
 	struct btrfs_key key, location;
@@ -81,15 +81,11 @@ int btrfs_insert_xattr_item(struct btrfs_trans_handle *trans,
 	struct extent_buffer *leaf;
 	u32 data_size;
 
-	key.objectid = dir;
+	BUG_ON(name_len + data_len > BTRFS_MAX_XATTR_SIZE(root));
+
+	key.objectid = objectid;
 	btrfs_set_key_type(&key, BTRFS_XATTR_ITEM_KEY);
 	key.offset = btrfs_name_hash(name, name_len);
-	path = btrfs_alloc_path();
-	if (!path)
-		return -ENOMEM;
-	if (name_len + data_len + sizeof(struct btrfs_dir_item) >
-	    BTRFS_LEAF_DATA_SIZE(root) - sizeof(struct btrfs_item))
-		return -ENOSPC;
 
 	data_size = sizeof(*dir_item) + name_len + data_len;
 	dir_item = insert_with_overflow(trans, root, path, &key, data_size,
@@ -117,7 +113,6 @@ int btrfs_insert_xattr_item(struct btrfs_trans_handle *trans,
 	write_extent_buffer(leaf, data, data_ptr, data_len);
 	btrfs_mark_buffer_dirty(path->nodes[0]);
 
-	btrfs_free_path(path);
 	return ret;
 }
 

fs/btrfs/disk-io.c

@@ -892,6 +892,8 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
 	root->stripesize = stripesize;
 	root->ref_cows = 0;
 	root->track_dirty = 0;
+	root->in_radix = 0;
+	root->clean_orphans = 0;
 
 	root->fs_info = fs_info;
 	root->objectid = objectid;
@@ -928,7 +930,6 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
 	root->defrag_trans_start = fs_info->generation;
 	init_completion(&root->kobj_unregister);
 	root->defrag_running = 0;
-	root->defrag_level = 0;
 	root->root_key.objectid = objectid;
 	root->anon_super.s_root = NULL;
 	root->anon_super.s_dev = 0;
@@ -980,12 +981,12 @@ int btrfs_free_log_root_tree(struct btrfs_trans_handle *trans,
 
 	while (1) {
 		ret = find_first_extent_bit(&log_root_tree->dirty_log_pages,
-				    0, &start, &end, EXTENT_DIRTY);
+				0, &start, &end, EXTENT_DIRTY | EXTENT_NEW);
 		if (ret)
 			break;
 
-		clear_extent_dirty(&log_root_tree->dirty_log_pages,
-				   start, end, GFP_NOFS);
+		clear_extent_bits(&log_root_tree->dirty_log_pages, start, end,
+				  EXTENT_DIRTY | EXTENT_NEW, GFP_NOFS);
 	}
 	eb = fs_info->log_root_tree->node;
 
@@ -1210,8 +1211,10 @@ struct btrfs_root *btrfs_read_fs_root_no_name(struct btrfs_fs_info *fs_info,
 	ret = radix_tree_insert(&fs_info->fs_roots_radix,
 				(unsigned long)root->root_key.objectid,
 				root);
-	if (ret == 0)
+	if (ret == 0) {
 		root->in_radix = 1;
+		root->clean_orphans = 1;
+	}
 	spin_unlock(&fs_info->fs_roots_radix_lock);
 	radix_tree_preload_end();
 	if (ret) {
@@ -1225,10 +1228,6 @@ struct btrfs_root *btrfs_read_fs_root_no_name(struct btrfs_fs_info *fs_info,
 	ret = btrfs_find_dead_roots(fs_info->tree_root,
 				    root->root_key.objectid);
 	WARN_ON(ret);
-
-	if (!(fs_info->sb->s_flags & MS_RDONLY))
-		btrfs_orphan_cleanup(root);
-
 	return root;
 fail:
 	free_fs_root(root);
@@ -1477,6 +1476,7 @@ static int cleaner_kthread(void *arg)
 
 		if (!(root->fs_info->sb->s_flags & MS_RDONLY) &&
 		    mutex_trylock(&root->fs_info->cleaner_mutex)) {
+			btrfs_run_delayed_iputs(root);
 			btrfs_clean_old_snapshots(root);
 			mutex_unlock(&root->fs_info->cleaner_mutex);
 		}
@@ -1606,6 +1606,7 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	INIT_RADIX_TREE(&fs_info->fs_roots_radix, GFP_ATOMIC);
 	INIT_LIST_HEAD(&fs_info->trans_list);
 	INIT_LIST_HEAD(&fs_info->dead_roots);
+	INIT_LIST_HEAD(&fs_info->delayed_iputs);
 	INIT_LIST_HEAD(&fs_info->hashers);
 	INIT_LIST_HEAD(&fs_info->delalloc_inodes);
 	INIT_LIST_HEAD(&fs_info->ordered_operations);
@@ -1614,6 +1615,7 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	spin_lock_init(&fs_info->new_trans_lock);
 	spin_lock_init(&fs_info->ref_cache_lock);
 	spin_lock_init(&fs_info->fs_roots_radix_lock);
+	spin_lock_init(&fs_info->delayed_iput_lock);
 
 	init_completion(&fs_info->kobj_unregister);
 	fs_info->tree_root = tree_root;
@@ -1689,6 +1691,7 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	mutex_init(&fs_info->cleaner_mutex);
 	mutex_init(&fs_info->volume_mutex);
 	init_rwsem(&fs_info->extent_commit_sem);
+	init_rwsem(&fs_info->cleanup_work_sem);
 	init_rwsem(&fs_info->subvol_sem);
 
 	btrfs_init_free_cluster(&fs_info->meta_alloc_cluster);
@@ -2386,8 +2389,14 @@ int btrfs_commit_super(struct btrfs_root *root)
 	int ret;
 
 	mutex_lock(&root->fs_info->cleaner_mutex);
+	btrfs_run_delayed_iputs(root);
 	btrfs_clean_old_snapshots(root);
 	mutex_unlock(&root->fs_info->cleaner_mutex);
+
+	/* wait until ongoing cleanup work done */
+	down_write(&root->fs_info->cleanup_work_sem);
+	up_write(&root->fs_info->cleanup_work_sem);
+
 	trans = btrfs_start_transaction(root, 1);
 	ret = btrfs_commit_transaction(trans, root);
 	BUG_ON(ret);

fs/btrfs/extent-tree.c

@@ -195,6 +195,14 @@ static int exclude_super_stripes(struct btrfs_root *root,
 	int stripe_len;
 	int i, nr, ret;
 
+	if (cache->key.objectid < BTRFS_SUPER_INFO_OFFSET) {
+		stripe_len = BTRFS_SUPER_INFO_OFFSET - cache->key.objectid;
+		cache->bytes_super += stripe_len;
+		ret = add_excluded_extent(root, cache->key.objectid,
+					  stripe_len);
+		BUG_ON(ret);
+	}
+
 	for (i = 0; i < BTRFS_SUPER_MIRROR_MAX; i++) {
 		bytenr = btrfs_sb_offset(i);
 		ret = btrfs_rmap_block(&root->fs_info->mapping_tree,
@@ -255,7 +263,7 @@ static u64 add_new_free_space(struct btrfs_block_group_cache *block_group,
 		if (ret)
 			break;
 
-		if (extent_start == start) {
+		if (extent_start <= start) {
 			start = extent_end + 1;
 		} else if (extent_start > start && extent_start < end) {
 			size = extent_start - start;
@@ -2880,9 +2888,9 @@ static noinline void flush_delalloc_async(struct btrfs_work *work)
 	root = async->root;
 	info = async->info;
 
-	btrfs_start_delalloc_inodes(root);
+	btrfs_start_delalloc_inodes(root, 0);
 	wake_up(&info->flush_wait);
-	btrfs_wait_ordered_extents(root, 0);
+	btrfs_wait_ordered_extents(root, 0, 0);
 
 	spin_lock(&info->lock);
 	info->flushing = 0;
@@ -2956,8 +2964,8 @@ static void flush_delalloc(struct btrfs_root *root,
 	return;
 
 flush:
-	btrfs_start_delalloc_inodes(root);
-	btrfs_wait_ordered_extents(root, 0);
+	btrfs_start_delalloc_inodes(root, 0);
+	btrfs_wait_ordered_extents(root, 0, 0);
 
 	spin_lock(&info->lock);
 	info->flushing = 0;
@@ -3454,14 +3462,6 @@ static int update_block_group(struct btrfs_trans_handle *trans,
 	else
 		old_val -= num_bytes;
 	btrfs_set_super_bytes_used(&info->super_copy, old_val);
-
-	/* block accounting for root item */
-	old_val = btrfs_root_used(&root->root_item);
-	if (alloc)
-		old_val += num_bytes;
-	else
-		old_val -= num_bytes;
-	btrfs_set_root_used(&root->root_item, old_val);
 	spin_unlock(&info->delalloc_lock);
 
 	while (total) {
@@ -4049,6 +4049,21 @@ int btrfs_free_extent(struct btrfs_trans_handle *trans,
 	return ret;
 }
 
+int btrfs_free_tree_block(struct btrfs_trans_handle *trans,
+			  struct btrfs_root *root,
+			  u64 bytenr, u32 blocksize,
+			  u64 parent, u64 root_objectid, int level)
+{
+	u64 used;
+	spin_lock(&root->node_lock);
+	used = btrfs_root_used(&root->root_item) - blocksize;
+	btrfs_set_root_used(&root->root_item, used);
+	spin_unlock(&root->node_lock);
+
+	return btrfs_free_extent(trans, root, bytenr, blocksize,
+				 parent, root_objectid, level, 0);
+}
+
 static u64 stripe_align(struct btrfs_root *root, u64 val)
 {
 	u64 mask = ((u64)root->stripesize - 1);
@@ -4578,7 +4593,6 @@ int btrfs_reserve_extent(struct btrfs_trans_handle *trans,
 {
 	int ret;
 	u64 search_start = 0;
-	struct btrfs_fs_info *info = root->fs_info;
 
 	data = btrfs_get_alloc_profile(root, data);
 again:
@@ -4586,17 +4600,9 @@ int btrfs_reserve_extent(struct btrfs_trans_handle *trans,
 	 * the only place that sets empty_size is btrfs_realloc_node, which
 	 * is not called recursively on allocations
 	 */
-	if (empty_size || root->ref_cows) {
-		if (!(data & BTRFS_BLOCK_GROUP_METADATA)) {
-			ret = do_chunk_alloc(trans, root->fs_info->extent_root,
-				     2 * 1024 * 1024,
-				     BTRFS_BLOCK_GROUP_METADATA |
-				     (info->metadata_alloc_profile &
-				      info->avail_metadata_alloc_bits), 0);
-		}
+	if (empty_size || root->ref_cows)
 		ret = do_chunk_alloc(trans, root->fs_info->extent_root,
 				     num_bytes + 2 * 1024 * 1024, data, 0);
-	}
 
 	WARN_ON(num_bytes < root->sectorsize);
 	ret = find_free_extent(trans, root, num_bytes, empty_size,
@@ -4897,6 +4903,14 @@ static int alloc_tree_block(struct btrfs_trans_handle *trans,
 					extent_op);
 		BUG_ON(ret);
 	}
+
+	if (root_objectid == root->root_key.objectid) {
+		u64 used;
+		spin_lock(&root->node_lock);
+		used = btrfs_root_used(&root->root_item) + num_bytes;
+		btrfs_set_root_used(&root->root_item, used);
+		spin_unlock(&root->node_lock);
+	}
 	return ret;
 }
 
@@ -4919,8 +4933,16 @@ struct extent_buffer *btrfs_init_new_buffer(struct btrfs_trans_handle *trans,
 	btrfs_set_buffer_uptodate(buf);
 
 	if (root->root_key.objectid == BTRFS_TREE_LOG_OBJECTID) {
+		/*
+		 * we allow two log transactions at a time, use different
+		 * EXENT bit to differentiate dirty pages.
+		 */
+		if (root->log_transid % 2 == 0)
 			set_extent_dirty(&root->dirty_log_pages, buf->start,
 					buf->start + buf->len - 1, GFP_NOFS);
+		else
+			set_extent_new(&root->dirty_log_pages, buf->start,
+					buf->start + buf->len - 1, GFP_NOFS);
 	} else {
 		set_extent_dirty(&trans->transaction->dirty_pages, buf->start,
 			 buf->start + buf->len - 1, GFP_NOFS);

fs/btrfs/ioctl.c

@@ -237,7 +237,6 @@ static noinline int create_subvol(struct btrfs_root *root,
 	u64 objectid;
 	u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID;
 	u64 index = 0;
-	unsigned long nr = 1;
 
 	/*
 	 * 1 - inode item
@@ -290,7 +289,7 @@ static noinline int create_subvol(struct btrfs_root *root,
 	btrfs_set_root_generation(&root_item, trans->transid);
 	btrfs_set_root_level(&root_item, 0);
 	btrfs_set_root_refs(&root_item, 1);
-	btrfs_set_root_used(&root_item, 0);
+	btrfs_set_root_used(&root_item, leaf->len);
 	btrfs_set_root_last_snapshot(&root_item, 0);
 
 	memset(&root_item.drop_progress, 0, sizeof(root_item.drop_progress));
@@ -342,24 +341,21 @@ static noinline int create_subvol(struct btrfs_root *root,
 
 	d_instantiate(dentry, btrfs_lookup_dentry(dir, dentry));
 fail:
-	nr = trans->blocks_used;
 	err = btrfs_commit_transaction(trans, root);
 	if (err && !ret)
 		ret = err;
 
 	btrfs_unreserve_metadata_space(root, 6);
-	btrfs_btree_balance_dirty(root, nr);
 	return ret;
 }
 
 static int create_snapshot(struct btrfs_root *root, struct dentry *dentry,
 			   char *name, int namelen)
 {
+	struct inode *inode;
 	struct btrfs_pending_snapshot *pending_snapshot;
 	struct btrfs_trans_handle *trans;
-	int ret = 0;
-	int err;
-	unsigned long nr = 0;
+	int ret;
 
 	if (!root->ref_cows)
 		return -EINVAL;
@@ -372,20 +368,20 @@ static int create_snapshot(struct btrfs_root *root, struct dentry *dentry,
 	 */
 	ret = btrfs_reserve_metadata_space(root, 6);
 	if (ret)
-		goto fail_unlock;
+		goto fail;
 
 	pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_NOFS);
 	if (!pending_snapshot) {
 		ret = -ENOMEM;
 		btrfs_unreserve_metadata_space(root, 6);
-		goto fail_unlock;
+		goto fail;
 	}
 	pending_snapshot->name = kmalloc(namelen + 1, GFP_NOFS);
 	if (!pending_snapshot->name) {
 		ret = -ENOMEM;
 		kfree(pending_snapshot);
 		btrfs_unreserve_metadata_space(root, 6);
-		goto fail_unlock;
+		goto fail;
 	}
 	memcpy(pending_snapshot->name, name, namelen);
 	pending_snapshot->name[namelen] = '\0';
@@ -395,10 +391,19 @@ static int create_snapshot(struct btrfs_root *root, struct dentry *dentry,
 	pending_snapshot->root = root;
 	list_add(&pending_snapshot->list,
 		 &trans->transaction->pending_snapshots);
-	err = btrfs_commit_transaction(trans, root);
+	ret = btrfs_commit_transaction(trans, root);
+	BUG_ON(ret);
+	btrfs_unreserve_metadata_space(root, 6);
 
-fail_unlock:
-	btrfs_btree_balance_dirty(root, nr);
+	inode = btrfs_lookup_dentry(dentry->d_parent->d_inode, dentry);
+	if (IS_ERR(inode)) {
+		ret = PTR_ERR(inode);
+		goto fail;
+	}
+	BUG_ON(!inode);
+	d_instantiate(dentry, inode);
+	ret = 0;
+fail:
 	return ret;
 }
 
@@ -1027,8 +1032,7 @@ static noinline long btrfs_ioctl_clone(struct file *file, unsigned long srcfd,
 	BUG_ON(!trans);
 
 	/* punch hole in destination first */
-	btrfs_drop_extents(trans, root, inode, off, off + len,
-			   off + len, 0, &hint_byte, 1);
+	btrfs_drop_extents(trans, inode, off, off + len, &hint_byte, 1);
 
 	/* clone data */
 	key.objectid = src->i_ino;

fs/btrfs/ordered-data.c

@@ -291,16 +291,16 @@ int btrfs_put_ordered_extent(struct btrfs_ordered_extent *entry)
 
 /*
  * remove an ordered extent from the tree.  No references are dropped
- * but, anyone waiting on this extent is woken up.
+ * and you must wake_up entry->wait.  You must hold the tree mutex
+ * while you call this function.
  */
-int btrfs_remove_ordered_extent(struct inode *inode,
+static int __btrfs_remove_ordered_extent(struct inode *inode,
 				struct btrfs_ordered_extent *entry)
 {
 	struct btrfs_ordered_inode_tree *tree;
 	struct rb_node *node;
 
 	tree = &BTRFS_I(inode)->ordered_tree;
-	mutex_lock(&tree->mutex);
 	node = &entry->rb_node;
 	rb_erase(node, &tree->tree);
 	tree->last = NULL;
@@ -326,16 +326,34 @@ int btrfs_remove_ordered_extent(struct inode *inode,
 	}
 	spin_unlock(&BTRFS_I(inode)->root->fs_info->ordered_extent_lock);
 
+	return 0;
+}
+
+/*
+ * remove an ordered extent from the tree.  No references are dropped
+ * but any waiters are woken.
+ */
+int btrfs_remove_ordered_extent(struct inode *inode,
+				struct btrfs_ordered_extent *entry)
+{
+	struct btrfs_ordered_inode_tree *tree;
+	int ret;
+
+	tree = &BTRFS_I(inode)->ordered_tree;
+	mutex_lock(&tree->mutex);
+	ret = __btrfs_remove_ordered_extent(inode, entry);
 	mutex_unlock(&tree->mutex);
 	wake_up(&entry->wait);
-	return 0;
+
+	return ret;
 }
 
 /*
  * wait for all the ordered extents in a root.  This is done when balancing
  * space between drives.
  */
-int btrfs_wait_ordered_extents(struct btrfs_root *root, int nocow_only)
+int btrfs_wait_ordered_extents(struct btrfs_root *root,
+			       int nocow_only, int delay_iput)
 {
 	struct list_head splice;
 	struct list_head *cur;
@@ -372,6 +390,9 @@ int btrfs_wait_ordered_extents(struct btrfs_root *root, int nocow_only)
 		if (inode) {
 			btrfs_start_ordered_extent(inode, ordered, 1);
 			btrfs_put_ordered_extent(ordered);
+			if (delay_iput)
+				btrfs_add_delayed_iput(inode);
+			else
 				iput(inode);
 		} else {
 			btrfs_put_ordered_extent(ordered);
@@ -430,7 +451,7 @@ int btrfs_run_ordered_operations(struct btrfs_root *root, int wait)
 				btrfs_wait_ordered_range(inode, 0, (u64)-1);
 			else
 				filemap_flush(inode->i_mapping);
-			iput(inode);
+			btrfs_add_delayed_iput(inode);
 		}
 
 		cond_resched();
@@ -589,7 +610,7 @@ btrfs_lookup_first_ordered_extent(struct inode *inode, u64 file_offset)
  * After an extent is done, call this to conditionally update the on disk
  * i_size.  i_size is updated to cover any fully written part of the file.
  */
-int btrfs_ordered_update_i_size(struct inode *inode,
+int btrfs_ordered_update_i_size(struct inode *inode, u64 offset,
 				struct btrfs_ordered_extent *ordered)
 {
 	struct btrfs_ordered_inode_tree *tree = &BTRFS_I(inode)->ordered_tree;
@@ -597,18 +618,30 @@ int btrfs_ordered_update_i_size(struct inode *inode,
 	u64 disk_i_size;
 	u64 new_i_size;
 	u64 i_size_test;
+	u64 i_size = i_size_read(inode);
 	struct rb_node *node;
+	struct rb_node *prev = NULL;
 	struct btrfs_ordered_extent *test;
+	int ret = 1;
+
+	if (ordered)
+		offset = entry_end(ordered);
 
 	mutex_lock(&tree->mutex);
 	disk_i_size = BTRFS_I(inode)->disk_i_size;
 
+	/* truncate file */
+	if (disk_i_size > i_size) {
+		BTRFS_I(inode)->disk_i_size = i_size;
+		ret = 0;
+		goto out;
+	}
+
 	/*
 	 * if the disk i_size is already at the inode->i_size, or
 	 * this ordered extent is inside the disk i_size, we're done
 	 */
-	if (disk_i_size >= inode->i_size ||
-	    ordered->file_offset + ordered->len <= disk_i_size) {
+	if (disk_i_size == i_size || offset <= disk_i_size) {
 		goto out;
 	}
 
@@ -616,8 +649,7 @@ int btrfs_ordered_update_i_size(struct inode *inode,
 	 * we can't update the disk_isize if there are delalloc bytes
 	 * between disk_i_size and  this ordered extent
 	 */
-	if (test_range_bit(io_tree, disk_i_size,
-			   ordered->file_offset + ordered->len - 1,
+	if (test_range_bit(io_tree, disk_i_size, offset - 1,
 			   EXTENT_DELALLOC, 0, NULL)) {
 		goto out;
 	}
@@ -626,20 +658,32 @@ int btrfs_ordered_update_i_size(struct inode *inode,
 	 * if we find an ordered extent then we can't update disk i_size
 	 * yet
 	 */
-	node = &ordered->rb_node;
-	while (1) {
-		node = rb_prev(node);
-		if (!node)
-			break;
+	if (ordered) {
+		node = rb_prev(&ordered->rb_node);
+	} else {
+		prev = tree_search(tree, offset);
+		/*
+		 * we insert file extents without involving ordered struct,
+		 * so there should be no ordered struct cover this offset
+		 */
+		if (prev) {
+			test = rb_entry(prev, struct btrfs_ordered_extent,
+					rb_node);
+			BUG_ON(offset_in_entry(test, offset));
+		}
+		node = prev;
+	}
+	while (node) {
 		test = rb_entry(node, struct btrfs_ordered_extent, rb_node);
 		if (test->file_offset + test->len <= disk_i_size)
 			break;
-		if (test->file_offset >= inode->i_size)
+		if (test->file_offset >= i_size)
 			break;
 		if (test->file_offset >= disk_i_size)
 			goto out;
+		node = rb_prev(node);
 	}
-	new_i_size = min_t(u64, entry_end(ordered), i_size_read(inode));
+	new_i_size = min_t(u64, offset, i_size);
 
 	/*
 	 * at this point, we know we can safely update i_size to at least
@@ -647,7 +691,14 @@ int btrfs_ordered_update_i_size(struct inode *inode,
 	 * walk forward and see if ios from higher up in the file have
 	 * finished.
 	 */
+	if (ordered) {
 		node = rb_next(&ordered->rb_node);
+	} else {
+		if (prev)
+			node = rb_next(prev);
+		else
+			node = rb_first(&tree->tree);
+	}
 	i_size_test = 0;
 	if (node) {
 		/*
@@ -655,10 +706,10 @@ int btrfs_ordered_update_i_size(struct inode *inode,
 		 * between our ordered extent and the next one.
 		 */
 		test = rb_entry(node, struct btrfs_ordered_extent, rb_node);
-		if (test->file_offset > entry_end(ordered))
+		if (test->file_offset > offset)
 			i_size_test = test->file_offset;
 	} else {
-		i_size_test = i_size_read(inode);
+		i_size_test = i_size;
 	}
 
 	/*
@@ -667,15 +718,25 @@ int btrfs_ordered_update_i_size(struct inode *inode,
 	 * are no delalloc bytes in this area, it is safe to update
 	 * disk_i_size to the end of the region.
 	 */
-	if (i_size_test > entry_end(ordered) &&
-	    !test_range_bit(io_tree, entry_end(ordered), i_size_test - 1,
+	if (i_size_test > offset &&
+	    !test_range_bit(io_tree, offset, i_size_test - 1,
 			    EXTENT_DELALLOC, 0, NULL)) {
-		new_i_size = min_t(u64, i_size_test, i_size_read(inode));
+		new_i_size = min_t(u64, i_size_test, i_size);
 	}
 	BTRFS_I(inode)->disk_i_size = new_i_size;
+	ret = 0;
 out:
+	/*
+	 * we need to remove the ordered extent with the tree lock held
+	 * so that other people calling this function don't find our fully
+	 * processed ordered entry and skip updating the i_size
+	 */
+	if (ordered)
+		__btrfs_remove_ordered_extent(inode, ordered);
 	mutex_unlock(&tree->mutex);
-	return 0;
+	if (ordered)
+		wake_up(&ordered->wait);
+	return ret;
 }
 
 /*

fs/btrfs/ordered-data.h

@@ -150,12 +150,13 @@ void btrfs_start_ordered_extent(struct inode *inode,
 int btrfs_wait_ordered_range(struct inode *inode, u64 start, u64 len);
 struct btrfs_ordered_extent *
 btrfs_lookup_first_ordered_extent(struct inode * inode, u64 file_offset);
-int btrfs_ordered_update_i_size(struct inode *inode,
+int btrfs_ordered_update_i_size(struct inode *inode, u64 offset,
 				struct btrfs_ordered_extent *ordered);
 int btrfs_find_ordered_sum(struct inode *inode, u64 offset, u64 disk_bytenr, u32 *sum);
-int btrfs_wait_ordered_extents(struct btrfs_root *root, int nocow_only);
 int btrfs_run_ordered_operations(struct btrfs_root *root, int wait);
 int btrfs_add_ordered_operation(struct btrfs_trans_handle *trans,
 				struct btrfs_root *root,
 				struct inode *inode);
+int btrfs_wait_ordered_extents(struct btrfs_root *root,
+			       int nocow_only, int delay_iput);
 #endif

fs/btrfs/relocation.c

@@ -1561,6 +1561,20 @@ static int invalidate_extent_cache(struct btrfs_root *root,
 	return 0;
 }
 
+static void put_inodes(struct list_head *list)
+{
+	struct inodevec *ivec;
+	while (!list_empty(list)) {
+		ivec = list_entry(list->next, struct inodevec, list);
+		list_del(&ivec->list);
+		while (ivec->nr > 0) {
+			ivec->nr--;
+			iput(ivec->inode[ivec->nr]);
+		}
+		kfree(ivec);
+	}
+}
+
 static int find_next_key(struct btrfs_path *path, int level,
 			 struct btrfs_key *key)
 
@@ -1723,6 +1737,11 @@ static noinline_for_stack int merge_reloc_root(struct reloc_control *rc,
 
 		btrfs_btree_balance_dirty(root, nr);
 
+		/*
+		 * put inodes outside transaction, otherwise we may deadlock.
+		 */
+		put_inodes(&inode_list);
+
 		if (replaced && rc->stage == UPDATE_DATA_PTRS)
 			invalidate_extent_cache(root, &key, &next_key);
 	}
@@ -1752,19 +1771,7 @@ static noinline_for_stack int merge_reloc_root(struct reloc_control *rc,
 
 	btrfs_btree_balance_dirty(root, nr);
 
-	/*
-	 * put inodes while we aren't holding the tree locks
-	 */
-	while (!list_empty(&inode_list)) {
-		struct inodevec *ivec;
-		ivec = list_entry(inode_list.next, struct inodevec, list);
-		list_del(&ivec->list);
-		while (ivec->nr > 0) {
-			ivec->nr--;
-			iput(ivec->inode[ivec->nr]);
-		}
-		kfree(ivec);
-	}
+	put_inodes(&inode_list);
 
 	if (replaced && rc->stage == UPDATE_DATA_PTRS)
 		invalidate_extent_cache(root, &key, &next_key);
@@ -3534,8 +3541,8 @@ int btrfs_relocate_block_group(struct btrfs_root *extent_root, u64 group_start)
 	       (unsigned long long)rc->block_group->key.objectid,
 	       (unsigned long long)rc->block_group->flags);
 
-	btrfs_start_delalloc_inodes(fs_info->tree_root);
-	btrfs_wait_ordered_extents(fs_info->tree_root, 0);
+	btrfs_start_delalloc_inodes(fs_info->tree_root, 0);
+	btrfs_wait_ordered_extents(fs_info->tree_root, 0, 0);
 
 	while (1) {
 		rc->extents_found = 0;
@@ -3755,6 +3762,7 @@ int btrfs_recover_relocation(struct btrfs_root *root)
 				       BTRFS_DATA_RELOC_TREE_OBJECTID);
 		if (IS_ERR(fs_root))
 			err = PTR_ERR(fs_root);
+		btrfs_orphan_cleanup(fs_root);
 	}
 	return err;
 }

fs/btrfs/super.c

@@ -128,6 +128,7 @@ int btrfs_parse_options(struct btrfs_root *root, char *options)
 	substring_t args[MAX_OPT_ARGS];
 	char *p, *num;
 	int intarg;
+	int ret = 0;
 
 	if (!options)
 		return 0;
@@ -262,12 +263,18 @@ int btrfs_parse_options(struct btrfs_root *root, char *options)
 		case Opt_discard:
 			btrfs_set_opt(info->mount_opt, DISCARD);
 			break;
+		case Opt_err:
+			printk(KERN_INFO "btrfs: unrecognized mount option "
+			       "'%s'\n", p);
+			ret = -EINVAL;
+			goto out;
 		default:
 			break;
 		}
 	}
+out:
 	kfree(options);
-	return 0;
+	return ret;
 }
 
 /*
@@ -405,8 +412,8 @@ int btrfs_sync_fs(struct super_block *sb, int wait)
 		return 0;
 	}
 
-	btrfs_start_delalloc_inodes(root);
-	btrfs_wait_ordered_extents(root, 0);
+	btrfs_start_delalloc_inodes(root, 0);
+	btrfs_wait_ordered_extents(root, 0, 0);
 
 	trans = btrfs_start_transaction(root, 1);
 	ret = btrfs_commit_transaction(trans, root);
@@ -450,6 +457,8 @@ static int btrfs_show_options(struct seq_file *seq, struct vfsmount *vfs)
 		seq_puts(seq, ",notreelog");
 	if (btrfs_test_opt(root, FLUSHONCOMMIT))
 		seq_puts(seq, ",flushoncommit");
+	if (btrfs_test_opt(root, DISCARD))
+		seq_puts(seq, ",discard");
 	if (!(root->fs_info->sb->s_flags & MS_POSIXACL))
 		seq_puts(seq, ",noacl");
 	return 0;

fs/btrfs/transaction.c

@@ -333,6 +333,9 @@ static int __btrfs_end_transaction(struct btrfs_trans_handle *trans,
 	memset(trans, 0, sizeof(*trans));
 	kmem_cache_free(btrfs_trans_handle_cachep, trans);
 
+	if (throttle)
+		btrfs_run_delayed_iputs(root);
+
 	return 0;
 }
 
@@ -354,7 +357,7 @@ int btrfs_end_transaction_throttle(struct btrfs_trans_handle *trans,
  * those extents are sent to disk but does not wait on them
  */
 int btrfs_write_marked_extents(struct btrfs_root *root,
-			       struct extent_io_tree *dirty_pages)
+			       struct extent_io_tree *dirty_pages, int mark)
 {
 	int ret;
 	int err = 0;
@@ -367,7 +370,7 @@ int btrfs_write_marked_extents(struct btrfs_root *root,
 
 	while (1) {
 		ret = find_first_extent_bit(dirty_pages, start, &start, &end,
-					    EXTENT_DIRTY);
+					    mark);
 		if (ret)
 			break;
 		while (start <= end) {
@@ -413,7 +416,7 @@ int btrfs_write_marked_extents(struct btrfs_root *root,
  * on all the pages and clear them from the dirty pages state tree
  */
 int btrfs_wait_marked_extents(struct btrfs_root *root,
-			      struct extent_io_tree *dirty_pages)
+			      struct extent_io_tree *dirty_pages, int mark)
 {
 	int ret;
 	int err = 0;
@@ -425,12 +428,12 @@ int btrfs_wait_marked_extents(struct btrfs_root *root,
 	unsigned long index;
 
 	while (1) {
-		ret = find_first_extent_bit(dirty_pages, 0, &start, &end,
-					    EXTENT_DIRTY);
+		ret = find_first_extent_bit(dirty_pages, start, &start, &end,
+					    mark);
 		if (ret)
 			break;
 
-		clear_extent_dirty(dirty_pages, start, end, GFP_NOFS);
+		clear_extent_bits(dirty_pages, start, end, mark, GFP_NOFS);
 		while (start <= end) {
 			index = start >> PAGE_CACHE_SHIFT;
 			start = (u64)(index + 1) << PAGE_CACHE_SHIFT;
@@ -460,13 +463,13 @@ int btrfs_wait_marked_extents(struct btrfs_root *root,
  * those extents are on disk for transaction or log commit
  */
 int btrfs_write_and_wait_marked_extents(struct btrfs_root *root,
-					struct extent_io_tree *dirty_pages)
+				struct extent_io_tree *dirty_pages, int mark)
 {
 	int ret;
 	int ret2;
 
-	ret = btrfs_write_marked_extents(root, dirty_pages);
-	ret2 = btrfs_wait_marked_extents(root, dirty_pages);
+	ret = btrfs_write_marked_extents(root, dirty_pages, mark);
+	ret2 = btrfs_wait_marked_extents(root, dirty_pages, mark);
 	return ret || ret2;
 }
 
@@ -479,7 +482,8 @@ int btrfs_write_and_wait_transaction(struct btrfs_trans_handle *trans,
 		return filemap_write_and_wait(btree_inode->i_mapping);
 	}
 	return btrfs_write_and_wait_marked_extents(root,
-					   &trans->transaction->dirty_pages);
+					   &trans->transaction->dirty_pages,
+					   EXTENT_DIRTY);
 }
 
 /*
@@ -497,13 +501,16 @@ static int update_cowonly_root(struct btrfs_trans_handle *trans,
 {
 	int ret;
 	u64 old_root_bytenr;
+	u64 old_root_used;
 	struct btrfs_root *tree_root = root->fs_info->tree_root;
 
+	old_root_used = btrfs_root_used(&root->root_item);
 	btrfs_write_dirty_block_groups(trans, root);
 
 	while (1) {
 		old_root_bytenr = btrfs_root_bytenr(&root->root_item);
-		if (old_root_bytenr == root->node->start)
+		if (old_root_bytenr == root->node->start &&
+		    old_root_used == btrfs_root_used(&root->root_item))
 			break;
 
 		btrfs_set_root_node(&root->root_item, root->node);
@@ -512,6 +519,7 @@ static int update_cowonly_root(struct btrfs_trans_handle *trans,
 					&root->root_item);
 		BUG_ON(ret);
 
+		old_root_used = btrfs_root_used(&root->root_item);
 		ret = btrfs_write_dirty_block_groups(trans, root);
 		BUG_ON(ret);
 	}
@@ -795,7 +803,6 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	memcpy(&pending->root_key, &key, sizeof(key));
 fail:
 	kfree(new_root_item);
-	btrfs_unreserve_metadata_space(root, 6);
 	return ret;
 }
 
@@ -807,7 +814,6 @@ static noinline int finish_pending_snapshot(struct btrfs_fs_info *fs_info,
 	u64 index = 0;
 	struct btrfs_trans_handle *trans;
 	struct inode *parent_inode;
-	struct inode *inode;
 	struct btrfs_root *parent_root;
 
 	parent_inode = pending->dentry->d_parent->d_inode;
@@ -839,8 +845,6 @@ static noinline int finish_pending_snapshot(struct btrfs_fs_info *fs_info,
 
 	BUG_ON(ret);
 
-	inode = btrfs_lookup_dentry(parent_inode, pending->dentry);
-	d_instantiate(pending->dentry, inode);
 fail:
 	btrfs_end_transaction(trans, fs_info->fs_root);
 	return ret;
@@ -994,11 +998,11 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 		mutex_unlock(&root->fs_info->trans_mutex);
 
 		if (flush_on_commit) {
-			btrfs_start_delalloc_inodes(root);
-			ret = btrfs_wait_ordered_extents(root, 0);
+			btrfs_start_delalloc_inodes(root, 1);
+			ret = btrfs_wait_ordered_extents(root, 0, 1);
 			BUG_ON(ret);
 		} else if (snap_pending) {
-			ret = btrfs_wait_ordered_extents(root, 1);
+			ret = btrfs_wait_ordered_extents(root, 0, 1);
 			BUG_ON(ret);
 		}
 
@@ -1116,6 +1120,10 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 		current->journal_info = NULL;
 
 	kmem_cache_free(btrfs_trans_handle_cachep, trans);
+
+	if (current != root->fs_info->transaction_kthread)
+		btrfs_run_delayed_iputs(root);
+
 	return ret;
 }
 

fs/btrfs/transaction.h

@@ -107,10 +107,10 @@ void btrfs_throttle(struct btrfs_root *root);
 int btrfs_record_root_in_trans(struct btrfs_trans_handle *trans,
 				struct btrfs_root *root);
 int btrfs_write_and_wait_marked_extents(struct btrfs_root *root,
-					struct extent_io_tree *dirty_pages);
+				struct extent_io_tree *dirty_pages, int mark);
 int btrfs_write_marked_extents(struct btrfs_root *root,
-					struct extent_io_tree *dirty_pages);
+				struct extent_io_tree *dirty_pages, int mark);
 int btrfs_wait_marked_extents(struct btrfs_root *root,
-					struct extent_io_tree *dirty_pages);
+				struct extent_io_tree *dirty_pages, int mark);
 int btrfs_transaction_in_commit(struct btrfs_fs_info *info);
 #endif

fs/btrfs/tree-log.c

@@ -542,8 +542,8 @@ static noinline int replay_one_extent(struct btrfs_trans_handle *trans,
 
 	saved_nbytes = inode_get_bytes(inode);
 	/* drop any overlapping extents */
-	ret = btrfs_drop_extents(trans, root, inode,
-			 start, extent_end, extent_end, start, &alloc_hint, 1);
+	ret = btrfs_drop_extents(trans, inode, start, extent_end,
+				 &alloc_hint, 1);
 	BUG_ON(ret);
 
 	if (found_type == BTRFS_FILE_EXTENT_REG ||
@@ -930,6 +930,17 @@ static noinline int add_inode_ref(struct btrfs_trans_handle *trans,
 	return 0;
 }
 
+static int insert_orphan_item(struct btrfs_trans_handle *trans,
+			      struct btrfs_root *root, u64 offset)
+{
+	int ret;
+	ret = btrfs_find_orphan_item(root, offset);
+	if (ret > 0)
+		ret = btrfs_insert_orphan_item(trans, root, offset);
+	return ret;
+}
+
+
 /*
  * There are a few corners where the link count of the file can't
  * be properly maintained during replay.  So, instead of adding
@@ -997,11 +1008,15 @@ static noinline int fixup_inode_link_count(struct btrfs_trans_handle *trans,
 	}
 	BTRFS_I(inode)->index_cnt = (u64)-1;
 
-	if (inode->i_nlink == 0 && S_ISDIR(inode->i_mode)) {
+	if (inode->i_nlink == 0) {
+		if (S_ISDIR(inode->i_mode)) {
 			ret = replay_dir_deletes(trans, root, NULL, path,
 						 inode->i_ino, 1);
 			BUG_ON(ret);
 		}
+		ret = insert_orphan_item(trans, root, inode->i_ino);
+		BUG_ON(ret);
+	}
 	btrfs_free_path(path);
 
 	return 0;
@@ -1587,7 +1602,6 @@ static int replay_one_buffer(struct btrfs_root *log, struct extent_buffer *eb,
 		/* inode keys are done during the first stage */
 		if (key.type == BTRFS_INODE_ITEM_KEY &&
 		    wc->stage == LOG_WALK_REPLAY_INODES) {
-			struct inode *inode;
 			struct btrfs_inode_item *inode_item;
 			u32 mode;
 
@@ -1603,31 +1617,16 @@ static int replay_one_buffer(struct btrfs_root *log, struct extent_buffer *eb,
 					     eb, i, &key);
 			BUG_ON(ret);
 
-			/* for regular files, truncate away
-			 * extents past the new EOF
+			/* for regular files, make sure corresponding
+			 * orhpan item exist. extents past the new EOF
+			 * will be truncated later by orphan cleanup.
 			 */
 			if (S_ISREG(mode)) {
-				inode = read_one_inode(root,
+				ret = insert_orphan_item(wc->trans, root,
 							 key.objectid);
-				BUG_ON(!inode);
-
-				ret = btrfs_truncate_inode_items(wc->trans,
-					root, inode, inode->i_size,
-					BTRFS_EXTENT_DATA_KEY);
 				BUG_ON(ret);
-
-				/* if the nlink count is zero here, the iput
-				 * will free the inode.  We bump it to make
-				 * sure it doesn't get freed until the link
-				 * count fixup is done
-				 */
-				if (inode->i_nlink == 0) {
-					btrfs_inc_nlink(inode);
-					btrfs_update_inode(wc->trans,
-							   root, inode);
-				}
-				iput(inode);
 			}
+
 			ret = link_to_fixup_dir(wc->trans, root,
 						path, key.objectid);
 			BUG_ON(ret);
@@ -1977,10 +1976,11 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 {
 	int index1;
 	int index2;
+	int mark;
 	int ret;
 	struct btrfs_root *log = root->log_root;
 	struct btrfs_root *log_root_tree = root->fs_info->log_root_tree;
-	u64 log_transid = 0;
+	unsigned long log_transid = 0;
 
 	mutex_lock(&root->log_mutex);
 	index1 = root->log_transid % 2;
@@ -2014,24 +2014,29 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 		goto out;
 	}
 
+	log_transid = root->log_transid;
+	if (log_transid % 2 == 0)
+		mark = EXTENT_DIRTY;
+	else
+		mark = EXTENT_NEW;
+
 	/* we start IO on  all the marked extents here, but we don't actually
 	 * wait for them until later.
 	 */
-	ret = btrfs_write_marked_extents(log, &log->dirty_log_pages);
+	ret = btrfs_write_marked_extents(log, &log->dirty_log_pages, mark);
 	BUG_ON(ret);
 
 	btrfs_set_root_node(&log->root_item, log->node);
 
 	root->log_batch = 0;
-	log_transid = root->log_transid;
 	root->log_transid++;
 	log->log_transid = root->log_transid;
 	root->log_start_pid = 0;
 	smp_mb();
 	/*
-	 * log tree has been flushed to disk, new modifications of
-	 * the log will be written to new positions. so it's safe to
-	 * allow log writers to go in.
+	 * IO has been started, blocks of the log tree have WRITTEN flag set
+	 * in their headers. new modifications of the log will be written to
+	 * new positions. so it's safe to allow log writers to go in.
 	 */
 	mutex_unlock(&root->log_mutex);
 
@@ -2052,7 +2057,7 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 
 	index2 = log_root_tree->log_transid % 2;
 	if (atomic_read(&log_root_tree->log_commit[index2])) {
-		btrfs_wait_marked_extents(log, &log->dirty_log_pages);
+		btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
 		wait_log_commit(trans, log_root_tree,
 				log_root_tree->log_transid);
 		mutex_unlock(&log_root_tree->log_mutex);
@@ -2072,16 +2077,17 @@ int btrfs_sync_log(struct btrfs_trans_handle *trans,
 	 * check the full commit flag again
 	 */
 	if (root->fs_info->last_trans_log_full_commit == trans->transid) {
-		btrfs_wait_marked_extents(log, &log->dirty_log_pages);
+		btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
 		mutex_unlock(&log_root_tree->log_mutex);
 		ret = -EAGAIN;
 		goto out_wake_log_root;
 	}
 
 	ret = btrfs_write_and_wait_marked_extents(log_root_tree,
-				&log_root_tree->dirty_log_pages);
+				&log_root_tree->dirty_log_pages,
+				EXTENT_DIRTY | EXTENT_NEW);
 	BUG_ON(ret);
-	btrfs_wait_marked_extents(log, &log->dirty_log_pages);
+	btrfs_wait_marked_extents(log, &log->dirty_log_pages, mark);
 
 	btrfs_set_super_log_root(&root->fs_info->super_for_commit,
 				log_root_tree->node->start);
@@ -2147,12 +2153,12 @@ int btrfs_free_log(struct btrfs_trans_handle *trans, struct btrfs_root *root)
 
 	while (1) {
 		ret = find_first_extent_bit(&log->dirty_log_pages,
-				    0, &start, &end, EXTENT_DIRTY);
+				0, &start, &end, EXTENT_DIRTY | EXTENT_NEW);
 		if (ret)
 			break;
 
-		clear_extent_dirty(&log->dirty_log_pages,
-				   start, end, GFP_NOFS);
+		clear_extent_bits(&log->dirty_log_pages, start, end,
+				  EXTENT_DIRTY | EXTENT_NEW, GFP_NOFS);
 	}
 
 	if (log->log_transid > 0) {

fs/btrfs/volumes.c

@@ -2209,7 +2209,7 @@ static int __btrfs_alloc_chunk(struct btrfs_trans_handle *trans,
 		max_chunk_size = 10 * calc_size;
 		min_stripe_size = 64 * 1024 * 1024;
 	} else if (type & BTRFS_BLOCK_GROUP_METADATA) {
-		max_chunk_size = 4 * calc_size;
+		max_chunk_size = 256 * 1024 * 1024;
 		min_stripe_size = 32 * 1024 * 1024;
 	} else if (type & BTRFS_BLOCK_GROUP_SYSTEM) {
 		calc_size = 8 * 1024 * 1024;

fs/btrfs/xattr.c

@@ -85,22 +85,23 @@ ssize_t __btrfs_getxattr(struct inode *inode, const char *name,
 	return ret;
 }
 
-int __btrfs_setxattr(struct inode *inode, const char *name,
+static int do_setxattr(struct btrfs_trans_handle *trans,
+		       struct inode *inode, const char *name,
 		       const void *value, size_t size, int flags)
 {
 	struct btrfs_dir_item *di;
 	struct btrfs_root *root = BTRFS_I(inode)->root;
-	struct btrfs_trans_handle *trans;
 	struct btrfs_path *path;
-	int ret = 0, mod = 0;
+	size_t name_len = strlen(name);
+	int ret = 0;
+
+	if (name_len + size > BTRFS_MAX_XATTR_SIZE(root))
+		return -ENOSPC;
 
 	path = btrfs_alloc_path();
 	if (!path)
 		return -ENOMEM;
 
-	trans = btrfs_join_transaction(root, 1);
-	btrfs_set_trans_block_group(trans, inode);
-
 	/* first lets see if we already have this xattr */
 	di = btrfs_lookup_xattr(trans, root, path, inode->i_ino, name,
 				strlen(name), -1);
@@ -118,15 +119,12 @@ int __btrfs_setxattr(struct inode *inode, const char *name,
 		}
 
 		ret = btrfs_delete_one_dir_name(trans, root, path, di);
-		if (ret)
-			goto out;
+		BUG_ON(ret);
 		btrfs_release_path(root, path);
 
 		/* if we don't have a value then we are removing the xattr */
-		if (!value) {
-			mod = 1;
+		if (!value)
 			goto out;
-		}
 	} else {
 		btrfs_release_path(root, path);
 
@@ -138,20 +136,45 @@ int __btrfs_setxattr(struct inode *inode, const char *name,
 	}
 
 	/* ok we have to create a completely new xattr */
-	ret = btrfs_insert_xattr_item(trans, root, name, strlen(name),
-				      value, size, inode->i_ino);
+	ret = btrfs_insert_xattr_item(trans, root, path, inode->i_ino,
+				      name, name_len, value, size);
+	BUG_ON(ret);
+out:
+	btrfs_free_path(path);
+	return ret;
+}
+
+int __btrfs_setxattr(struct btrfs_trans_handle *trans,
+		     struct inode *inode, const char *name,
+		     const void *value, size_t size, int flags)
+{
+	struct btrfs_root *root = BTRFS_I(inode)->root;
+	int ret;
+
+	if (trans)
+		return do_setxattr(trans, inode, name, value, size, flags);
+
+	ret = btrfs_reserve_metadata_space(root, 2);
+	if (ret)
+		return ret;
+
+	trans = btrfs_start_transaction(root, 1);
+	if (!trans) {
+		ret = -ENOMEM;
+		goto out;
+	}
+	btrfs_set_trans_block_group(trans, inode);
+
+	ret = do_setxattr(trans, inode, name, value, size, flags);
 	if (ret)
 		goto out;
-	mod = 1;
 
-out:
-	if (mod) {
 	inode->i_ctime = CURRENT_TIME;
 	ret = btrfs_update_inode(trans, root, inode);
-	}
-
-	btrfs_end_transaction(trans, root);
-	btrfs_free_path(path);
+	BUG_ON(ret);
+out:
+	btrfs_end_transaction_throttle(trans, root);
+	btrfs_unreserve_metadata_space(root, 2);
 	return ret;
 }
 
@@ -314,7 +337,9 @@ int btrfs_setxattr(struct dentry *dentry, const char *name, const void *value,
 
 	if (size == 0)
 		value = "";  /* empty EA, do not remove */
-	return __btrfs_setxattr(dentry->d_inode, name, value, size, flags);
+
+	return __btrfs_setxattr(NULL, dentry->d_inode, name, value, size,
+				flags);
 }
 
 int btrfs_removexattr(struct dentry *dentry, const char *name)
@@ -329,10 +354,13 @@ int btrfs_removexattr(struct dentry *dentry, const char *name)
 
 	if (!btrfs_is_valid_xattr(name))
 		return -EOPNOTSUPP;
-	return __btrfs_setxattr(dentry->d_inode, name, NULL, 0, XATTR_REPLACE);
+
+	return __btrfs_setxattr(NULL, dentry->d_inode, name, NULL, 0,
+				XATTR_REPLACE);
 }
 
-int btrfs_xattr_security_init(struct inode *inode, struct inode *dir)
+int btrfs_xattr_security_init(struct btrfs_trans_handle *trans,
+			      struct inode *inode, struct inode *dir)
 {
 	int err;
 	size_t len;
@@ -354,7 +382,7 @@ int btrfs_xattr_security_init(struct inode *inode, struct inode *dir)
 	} else {
 		strcpy(name, XATTR_SECURITY_PREFIX);
 		strcpy(name + XATTR_SECURITY_PREFIX_LEN, suffix);
-		err = __btrfs_setxattr(inode, name, value, len, 0);
+		err = __btrfs_setxattr(trans, inode, name, value, len, 0);
 		kfree(name);
 	}
 

fs/btrfs/xattr.h

@@ -27,15 +27,16 @@ extern struct xattr_handler *btrfs_xattr_handlers[];
 
 extern ssize_t __btrfs_getxattr(struct inode *inode, const char *name,
 		void *buffer, size_t size);
-extern int __btrfs_setxattr(struct inode *inode, const char *name,
+extern int __btrfs_setxattr(struct btrfs_trans_handle *trans,
+			    struct inode *inode, const char *name,
 			    const void *value, size_t size, int flags);
-
 extern ssize_t btrfs_getxattr(struct dentry *dentry, const char *name,
 		void *buffer, size_t size);
 extern int btrfs_setxattr(struct dentry *dentry, const char *name,
 		const void *value, size_t size, int flags);
 extern int btrfs_removexattr(struct dentry *dentry, const char *name);
 
-extern int btrfs_xattr_security_init(struct inode *inode, struct inode *dir);
+extern int btrfs_xattr_security_init(struct btrfs_trans_handle *trans,
+				     struct inode *inode, struct inode *dir);
 
 #endif /* __XATTR__ */