Merge commit 'v3.0-rc5' into perf/core

Merge reason: Pick up the latest fixes.

Signed-off-by: Ingo Molnar <mingo@elte.hu>

12 files changed, 274 insertions, 221 deletions

diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 378b5b4443f3..f30ac05dbda7 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -19,7 +19,6 @@
 #ifndef __BTRFS_CTREE__
 #define __BTRFS_CTREE__
 
-#include <linux/version.h>
 #include <linux/mm.h>
 #include <linux/highmem.h>
 #include <linux/fs.h>
@@ -967,6 +966,12 @@ struct btrfs_fs_info {
 	struct srcu_struct subvol_srcu;
 
 	spinlock_t trans_lock;
+	/*
+	 * the reloc mutex goes with the trans lock, it is taken
+	 * during commit to protect us from the relocation code
+	 */
+	struct mutex reloc_mutex;
+
 	struct list_head trans_list;
 	struct list_head hashers;
 	struct list_head dead_roots;
@@ -1172,6 +1177,14 @@ struct btrfs_root {
 	u32 type;
 
 	u64 highest_objectid;
+
+	/* btrfs_record_root_in_trans is a multi-step process,
+	 * and it can race with the balancing code.   But the
+	 * race is very small, and only the first time the root
+	 * is added to each transaction.  So in_trans_setup
+	 * is used to tell us when more checks are required
+	 */
+	unsigned long in_trans_setup;
 	int ref_cows;
 	int track_dirty;
 	int in_radix;
@@ -1181,7 +1194,6 @@ struct btrfs_root {
 	struct btrfs_key defrag_max;
 	int defrag_running;
 	char *name;
-	int in_sysfs;
 
 	/* the dirty list is only used by non-reference counted roots */
 	struct list_head dirty_list;
diff --git a/fs/btrfs/delayed-inode.c b/fs/btrfs/delayed-inode.c
index 6462c29d2d37..98c68e658a9b 100644
--- a/fs/btrfs/delayed-inode.c
+++ b/fs/btrfs/delayed-inode.c
@@ -82,19 +82,16 @@ static inline struct btrfs_delayed_root *btrfs_get_delayed_root(
 	return root->fs_info->delayed_root;
 }
 
-static struct btrfs_delayed_node *btrfs_get_or_create_delayed_node(
-							struct inode *inode)
+static struct btrfs_delayed_node *btrfs_get_delayed_node(struct inode *inode)
 {
-	struct btrfs_delayed_node *node;
 	struct btrfs_inode *btrfs_inode = BTRFS_I(inode);
 	struct btrfs_root *root = btrfs_inode->root;
 	u64 ino = btrfs_ino(inode);
-	int ret;
+	struct btrfs_delayed_node *node;
 
-again:
 	node = ACCESS_ONCE(btrfs_inode->delayed_node);
 	if (node) {
-		atomic_inc(&node->refs);	/* can be accessed */
+		atomic_inc(&node->refs);
 		return node;
 	}
 
@@ -102,8 +99,10 @@ again:
 	node = radix_tree_lookup(&root->delayed_nodes_tree, ino);
 	if (node) {
 		if (btrfs_inode->delayed_node) {
+			atomic_inc(&node->refs);	/* can be accessed */
+			BUG_ON(btrfs_inode->delayed_node != node);
 			spin_unlock(&root->inode_lock);
-			goto again;
+			return node;
 		}
 		btrfs_inode->delayed_node = node;
 		atomic_inc(&node->refs);	/* can be accessed */
@@ -113,6 +112,23 @@ again:
 	}
 	spin_unlock(&root->inode_lock);
 
+	return NULL;
+}
+
+static struct btrfs_delayed_node *btrfs_get_or_create_delayed_node(
+							struct inode *inode)
+{
+	struct btrfs_delayed_node *node;
+	struct btrfs_inode *btrfs_inode = BTRFS_I(inode);
+	struct btrfs_root *root = btrfs_inode->root;
+	u64 ino = btrfs_ino(inode);
+	int ret;
+
+again:
+	node = btrfs_get_delayed_node(inode);
+	if (node)
+		return node;
+
 	node = kmem_cache_alloc(delayed_node_cache, GFP_NOFS);
 	if (!node)
 		return ERR_PTR(-ENOMEM);
@@ -297,7 +313,6 @@ struct btrfs_delayed_item *btrfs_alloc_delayed_item(u32 data_len)
 		item->data_len = data_len;
 		item->ins_or_del = 0;
 		item->bytes_reserved = 0;
-		item->block_rsv = NULL;
 		item->delayed_node = NULL;
 		atomic_set(&item->refs, 1);
 	}
@@ -549,19 +564,6 @@ struct btrfs_delayed_item *__btrfs_next_delayed_item(
 	return next;
 }
 
-static inline struct btrfs_delayed_node *btrfs_get_delayed_node(
-							struct inode *inode)
-{
-	struct btrfs_inode *btrfs_inode = BTRFS_I(inode);
-	struct btrfs_delayed_node *delayed_node;
-
-	delayed_node = btrfs_inode->delayed_node;
-	if (delayed_node)
-		atomic_inc(&delayed_node->refs);
-
-	return delayed_node;
-}
-
 static inline struct btrfs_root *btrfs_get_fs_root(struct btrfs_root *root,
 						   u64 root_id)
 {
@@ -593,10 +595,8 @@ static int btrfs_delayed_item_reserve_metadata(struct btrfs_trans_handle *trans,
 
 	num_bytes = btrfs_calc_trans_metadata_size(root, 1);
 	ret = btrfs_block_rsv_migrate(src_rsv, dst_rsv, num_bytes);
-	if (!ret) {
+	if (!ret)
 		item->bytes_reserved = num_bytes;
-		item->block_rsv = dst_rsv;
-	}
 
 	return ret;
 }
@@ -604,10 +604,13 @@ static int btrfs_delayed_item_reserve_metadata(struct btrfs_trans_handle *trans,
 static void btrfs_delayed_item_release_metadata(struct btrfs_root *root,
 						struct btrfs_delayed_item *item)
 {
+	struct btrfs_block_rsv *rsv;
+
 	if (!item->bytes_reserved)
 		return;
 
-	btrfs_block_rsv_release(root, item->block_rsv,
+	rsv = &root->fs_info->global_block_rsv;
+	btrfs_block_rsv_release(root, rsv,
 				item->bytes_reserved);
 }
 
@@ -1014,6 +1017,7 @@ int btrfs_run_delayed_items(struct btrfs_trans_handle *trans,
 	struct btrfs_delayed_root *delayed_root;
 	struct btrfs_delayed_node *curr_node, *prev_node;
 	struct btrfs_path *path;
+	struct btrfs_block_rsv *block_rsv;
 	int ret = 0;
 
 	path = btrfs_alloc_path();
@@ -1021,6 +1025,9 @@ int btrfs_run_delayed_items(struct btrfs_trans_handle *trans,
 		return -ENOMEM;
 	path->leave_spinning = 1;
 
+	block_rsv = trans->block_rsv;
+	trans->block_rsv = &root->fs_info->global_block_rsv;
+
 	delayed_root = btrfs_get_delayed_root(root);
 
 	curr_node = btrfs_first_delayed_node(delayed_root);
@@ -1045,6 +1052,7 @@ int btrfs_run_delayed_items(struct btrfs_trans_handle *trans,
 	}
 
 	btrfs_free_path(path);
+	trans->block_rsv = block_rsv;
 	return ret;
 }
 
@@ -1052,6 +1060,7 @@ static int __btrfs_commit_inode_delayed_items(struct btrfs_trans_handle *trans,
 					      struct btrfs_delayed_node *node)
 {
 	struct btrfs_path *path;
+	struct btrfs_block_rsv *block_rsv;
 	int ret;
 
 	path = btrfs_alloc_path();
@@ -1059,6 +1068,9 @@ static int __btrfs_commit_inode_delayed_items(struct btrfs_trans_handle *trans,
 		return -ENOMEM;
 	path->leave_spinning = 1;
 
+	block_rsv = trans->block_rsv;
+	trans->block_rsv = &node->root->fs_info->global_block_rsv;
+
 	ret = btrfs_insert_delayed_items(trans, path, node->root, node);
 	if (!ret)
 		ret = btrfs_delete_delayed_items(trans, path, node->root, node);
@@ -1066,6 +1078,7 @@ static int __btrfs_commit_inode_delayed_items(struct btrfs_trans_handle *trans,
 		ret = btrfs_update_delayed_inode(trans, node->root, path, node);
 	btrfs_free_path(path);
 
+	trans->block_rsv = block_rsv;
 	return ret;
 }
 
@@ -1116,6 +1129,7 @@ static void btrfs_async_run_delayed_node_done(struct btrfs_work *work)
 	struct btrfs_path *path;
 	struct btrfs_delayed_node *delayed_node = NULL;
 	struct btrfs_root *root;
+	struct btrfs_block_rsv *block_rsv;
 	unsigned long nr = 0;
 	int need_requeue = 0;
 	int ret;
@@ -1134,6 +1148,9 @@ static void btrfs_async_run_delayed_node_done(struct btrfs_work *work)
 	if (IS_ERR(trans))
 		goto free_path;
 
+	block_rsv = trans->block_rsv;
+	trans->block_rsv = &root->fs_info->global_block_rsv;
+
 	ret = btrfs_insert_delayed_items(trans, path, root, delayed_node);
 	if (!ret)
 		ret = btrfs_delete_delayed_items(trans, path, root,
@@ -1176,6 +1193,7 @@ static void btrfs_async_run_delayed_node_done(struct btrfs_work *work)
 
 	nr = trans->blocks_used;
 
+	trans->block_rsv = block_rsv;
 	btrfs_end_transaction_dmeta(trans, root);
 	__btrfs_btree_balance_dirty(root, nr);
 free_path:
@@ -1222,6 +1240,13 @@ again:
 	return 0;
 }
 
+void btrfs_assert_delayed_root_empty(struct btrfs_root *root)
+{
+	struct btrfs_delayed_root *delayed_root;
+	delayed_root = btrfs_get_delayed_root(root);
+	WARN_ON(btrfs_first_delayed_node(delayed_root));
+}
+
 void btrfs_balance_delayed_items(struct btrfs_root *root)
 {
 	struct btrfs_delayed_root *delayed_root;
@@ -1382,8 +1407,7 @@ end:
 
 int btrfs_inode_delayed_dir_index_count(struct inode *inode)
 {
-	struct btrfs_delayed_node *delayed_node = BTRFS_I(inode)->delayed_node;
-	int ret = 0;
+	struct btrfs_delayed_node *delayed_node = btrfs_get_delayed_node(inode);
 
 	if (!delayed_node)
 		return -ENOENT;
@@ -1393,11 +1417,14 @@ int btrfs_inode_delayed_dir_index_count(struct inode *inode)
 	 * a new directory index is added into the delayed node and index_cnt
 	 * is updated now. So we needn't lock the delayed node.
 	 */
-	if (!delayed_node->index_cnt)
+	if (!delayed_node->index_cnt) {
+		btrfs_release_delayed_node(delayed_node);
 		return -EINVAL;
+	}
 
 	BTRFS_I(inode)->index_cnt = delayed_node->index_cnt;
-	return ret;
+	btrfs_release_delayed_node(delayed_node);
+	return 0;
 }
 
 void btrfs_get_delayed_items(struct inode *inode, struct list_head *ins_list,
@@ -1591,6 +1618,57 @@ static void fill_stack_inode_item(struct btrfs_trans_handle *trans,
 				      inode->i_ctime.tv_nsec);
 }
 
+int btrfs_fill_inode(struct inode *inode, u32 *rdev)
+{
+	struct btrfs_delayed_node *delayed_node;
+	struct btrfs_inode_item *inode_item;
+	struct btrfs_timespec *tspec;
+
+	delayed_node = btrfs_get_delayed_node(inode);
+	if (!delayed_node)
+		return -ENOENT;
+
+	mutex_lock(&delayed_node->mutex);
+	if (!delayed_node->inode_dirty) {
+		mutex_unlock(&delayed_node->mutex);
+		btrfs_release_delayed_node(delayed_node);
+		return -ENOENT;
+	}
+
+	inode_item = &delayed_node->inode_item;
+
+	inode->i_uid = btrfs_stack_inode_uid(inode_item);
+	inode->i_gid = btrfs_stack_inode_gid(inode_item);
+	btrfs_i_size_write(inode, btrfs_stack_inode_size(inode_item));
+	inode->i_mode = btrfs_stack_inode_mode(inode_item);
+	inode->i_nlink = btrfs_stack_inode_nlink(inode_item);
+	inode_set_bytes(inode, btrfs_stack_inode_nbytes(inode_item));
+	BTRFS_I(inode)->generation = btrfs_stack_inode_generation(inode_item);
+	BTRFS_I(inode)->sequence = btrfs_stack_inode_sequence(inode_item);
+	inode->i_rdev = 0;
+	*rdev = btrfs_stack_inode_rdev(inode_item);
+	BTRFS_I(inode)->flags = btrfs_stack_inode_flags(inode_item);
+
+	tspec = btrfs_inode_atime(inode_item);
+	inode->i_atime.tv_sec = btrfs_stack_timespec_sec(tspec);
+	inode->i_atime.tv_nsec = btrfs_stack_timespec_nsec(tspec);
+
+	tspec = btrfs_inode_mtime(inode_item);
+	inode->i_mtime.tv_sec = btrfs_stack_timespec_sec(tspec);
+	inode->i_mtime.tv_nsec = btrfs_stack_timespec_nsec(tspec);
+
+	tspec = btrfs_inode_ctime(inode_item);
+	inode->i_ctime.tv_sec = btrfs_stack_timespec_sec(tspec);
+	inode->i_ctime.tv_nsec = btrfs_stack_timespec_nsec(tspec);
+
+	inode->i_generation = BTRFS_I(inode)->generation;
+	BTRFS_I(inode)->index_cnt = (u64)-1;
+
+	mutex_unlock(&delayed_node->mutex);
+	btrfs_release_delayed_node(delayed_node);
+	return 0;
+}
+
 int btrfs_delayed_update_inode(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *root, struct inode *inode)
 {
diff --git a/fs/btrfs/delayed-inode.h b/fs/btrfs/delayed-inode.h
index eb7d240aa648..8d27af4bd8b9 100644
--- a/fs/btrfs/delayed-inode.h
+++ b/fs/btrfs/delayed-inode.h
@@ -75,7 +75,6 @@ struct btrfs_delayed_item {
 	struct list_head tree_list;	/* used for batch insert/delete items */
 	struct list_head readdir_list;	/* used for readdir items */
 	u64 bytes_reserved;
-	struct btrfs_block_rsv *block_rsv;
 	struct btrfs_delayed_node *delayed_node;
 	atomic_t refs;
 	int ins_or_del;
@@ -120,6 +119,7 @@ void btrfs_kill_delayed_inode_items(struct inode *inode);
 
 int btrfs_delayed_update_inode(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *root, struct inode *inode);
+int btrfs_fill_inode(struct inode *inode, u32 *rdev);
 
 /* Used for drop dead root */
 void btrfs_kill_all_delayed_nodes(struct btrfs_root *root);
@@ -138,4 +138,8 @@ int btrfs_readdir_delayed_dir_index(struct file *filp, void *dirent,
 /* for init */
 int __init btrfs_delayed_inode_init(void);
 void btrfs_delayed_inode_exit(void);
+
+/* for debugging */
+void btrfs_assert_delayed_root_empty(struct btrfs_root *root);
+
 #endif
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 9f68c6898653..1ac8db5dc0a3 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -1044,7 +1044,6 @@ static int __setup_root(u32 nodesize, u32 leafsize, u32 sectorsize,
 	root->last_trans = 0;
 	root->highest_objectid = 0;
 	root->name = NULL;
-	root->in_sysfs = 0;
 	root->inode_tree = RB_ROOT;
 	INIT_RADIX_TREE(&root->delayed_nodes_tree, GFP_ATOMIC);
 	root->block_rsv = NULL;
@@ -1300,19 +1299,21 @@ again:
 		return root;
 
 	root->free_ino_ctl = kzalloc(sizeof(*root->free_ino_ctl), GFP_NOFS);
-	if (!root->free_ino_ctl)
-		goto fail;
 	root->free_ino_pinned = kzalloc(sizeof(*root->free_ino_pinned),
 					GFP_NOFS);
-	if (!root->free_ino_pinned)
+	if (!root->free_ino_pinned || !root->free_ino_ctl) {
+		ret = -ENOMEM;
 		goto fail;
+	}
 
 	btrfs_init_free_ino_ctl(root);
 	mutex_init(&root->fs_commit_mutex);
 	spin_lock_init(&root->cache_lock);
 	init_waitqueue_head(&root->cache_wait);
 
-	set_anon_super(&root->anon_super, NULL);
+	ret = set_anon_super(&root->anon_super, NULL);
+	if (ret)
+		goto fail;
 
 	if (btrfs_root_refs(&root->root_item) == 0) {
 		ret = -ENOENT;
@@ -1618,6 +1619,7 @@ struct btrfs_root *open_ctree(struct super_block *sb,
 	spin_lock_init(&fs_info->fs_roots_radix_lock);
 	spin_lock_init(&fs_info->delayed_iput_lock);
 	spin_lock_init(&fs_info->defrag_inodes_lock);
+	mutex_init(&fs_info->reloc_mutex);
 
 	init_completion(&fs_info->kobj_unregister);
 	fs_info->tree_root = tree_root;
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index b42efc2ded51..71cd456fdb60 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -3314,10 +3314,6 @@ static int shrink_delalloc(struct btrfs_trans_handle *trans,
 	if (reserved == 0)
 		return 0;
 
-	/* nothing to shrink - nothing to reclaim */
-	if (root->fs_info->delalloc_bytes == 0)
-		return 0;
-
 	max_reclaim = min(reserved, to_reclaim);
 
 	while (loops < 1024) {
@@ -4846,7 +4842,7 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans,
 				     u64 num_bytes, u64 empty_size,
 				     u64 search_start, u64 search_end,
 				     u64 hint_byte, struct btrfs_key *ins,
-				     int data)
+				     u64 data)
 {
 	int ret = 0;
 	struct btrfs_root *root = orig_root->fs_info->extent_root;
@@ -4873,7 +4869,7 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans,
 
 	space_info = __find_space_info(root->fs_info, data);
 	if (!space_info) {
-		printk(KERN_ERR "No space info for %d\n", data);
+		printk(KERN_ERR "No space info for %llu\n", data);
 		return -ENOSPC;
 	}
 
diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c
index 9f985a429877..bf0d61567f3d 100644
--- a/fs/btrfs/free-space-cache.c
+++ b/fs/btrfs/free-space-cache.c
@@ -1893,9 +1893,12 @@ void __btrfs_remove_free_space_cache_locked(struct btrfs_free_space_ctl *ctl)
 
 	while ((node = rb_last(&ctl->free_space_offset)) != NULL) {
 		info = rb_entry(node, struct btrfs_free_space, offset_index);
-		unlink_free_space(ctl, info);
-		kfree(info->bitmap);
-		kmem_cache_free(btrfs_free_space_cachep, info);
+		if (!info->bitmap) {
+			unlink_free_space(ctl, info);
+			kmem_cache_free(btrfs_free_space_cachep, info);
+		} else {
+			free_bitmap(ctl, info);
+		}
 		if (need_resched()) {
 			spin_unlock(&ctl->tree_lock);
 			cond_resched();
diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index 751ddf8fc58a..d340f63d8f07 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -2509,6 +2509,11 @@ static void btrfs_read_locked_inode(struct inode *inode)
 	int maybe_acls;
 	u32 rdev;
 	int ret;
+	bool filled = false;
+
+	ret = btrfs_fill_inode(inode, &rdev);
+	if (!ret)
+		filled = true;
 
 	path = btrfs_alloc_path();
 	BUG_ON(!path);
@@ -2520,6 +2525,10 @@ static void btrfs_read_locked_inode(struct inode *inode)
 		goto make_bad;
 
 	leaf = path->nodes[0];
+
+	if (filled)
+		goto cache_acl;
+
 	inode_item = btrfs_item_ptr(leaf, path->slots[0],
 				    struct btrfs_inode_item);
 	if (!leaf->map_token)
@@ -2556,7 +2565,7 @@ static void btrfs_read_locked_inode(struct inode *inode)
 
 	BTRFS_I(inode)->index_cnt = (u64)-1;
 	BTRFS_I(inode)->flags = btrfs_inode_flags(leaf, inode_item);
-
+cache_acl:
 	/*
 	 * try to precache a NULL acl entry for files that don't have
 	 * any xattrs or acls
@@ -2572,7 +2581,6 @@ static void btrfs_read_locked_inode(struct inode *inode)
 	}
 
 	btrfs_free_path(path);
-	inode_item = NULL;
 
 	switch (inode->i_mode & S_IFMT) {
 	case S_IFREG:
@@ -3076,6 +3084,7 @@ int btrfs_unlink_subvol(struct btrfs_trans_handle *trans,
 	ret = btrfs_update_inode(trans, root, dir);
 	BUG_ON(ret);
 
+	btrfs_free_path(path);
 	return 0;
 }
 
@@ -4519,6 +4528,7 @@ static struct inode *btrfs_new_inode(struct btrfs_trans_handle *trans,
 	inode_tree_add(inode);
 
 	trace_btrfs_inode_new(inode);
+	btrfs_set_inode_last_trans(trans, inode);
 
 	return inode;
 fail:
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index b793d112d1f6..a3c4751e07db 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -482,8 +482,10 @@ static int create_snapshot(struct btrfs_root *root, struct dentry *dentry,
 	ret = btrfs_snap_reserve_metadata(trans, pending_snapshot);
 	BUG_ON(ret);
 
+	spin_lock(&root->fs_info->trans_lock);
 	list_add(&pending_snapshot->list,
 		 &trans->transaction->pending_snapshots);
+	spin_unlock(&root->fs_info->trans_lock);
 	if (async_transid) {
 		*async_transid = trans->transid;
 		ret = btrfs_commit_transaction_async(trans,
diff --git a/fs/btrfs/relocation.c b/fs/btrfs/relocation.c
index b1ef27cc673b..5e0a3dc79a45 100644
--- a/fs/btrfs/relocation.c
+++ b/fs/btrfs/relocation.c
@@ -1368,7 +1368,7 @@ int btrfs_update_reloc_root(struct btrfs_trans_handle *trans,
 	int ret;
 
 	if (!root->reloc_root)
-		return 0;
+		goto out;
 
 	reloc_root = root->reloc_root;
 	root_item = &reloc_root->root_item;
@@ -1390,6 +1390,8 @@ int btrfs_update_reloc_root(struct btrfs_trans_handle *trans,
 	ret = btrfs_update_root(trans, root->fs_info->tree_root,
 				&reloc_root->root_key, root_item);
 	BUG_ON(ret);
+
+out:
 	return 0;
 }
 
@@ -2142,10 +2144,11 @@ int prepare_to_merge(struct reloc_control *rc, int err)
 	u64 num_bytes = 0;
 	int ret;
 
-	spin_lock(&root->fs_info->trans_lock);
+	mutex_lock(&root->fs_info->reloc_mutex);
 	rc->merging_rsv_size += root->nodesize * (BTRFS_MAX_LEVEL - 1) * 2;
 	rc->merging_rsv_size += rc->nodes_relocated * 2;
-	spin_unlock(&root->fs_info->trans_lock);
+	mutex_unlock(&root->fs_info->reloc_mutex);
+
 again:
 	if (!err) {
 		num_bytes = rc->merging_rsv_size;
@@ -2214,9 +2217,16 @@ int merge_reloc_roots(struct reloc_control *rc)
 	int ret;
 again:
 	root = rc->extent_root;
-	spin_lock(&root->fs_info->trans_lock);
+
+	/*
+	 * this serializes us with btrfs_record_root_in_transaction,
+	 * we have to make sure nobody is in the middle of
+	 * adding their roots to the list while we are
+	 * doing this splice
+	 */
+	mutex_lock(&root->fs_info->reloc_mutex);
 	list_splice_init(&rc->reloc_roots, &reloc_roots);
-	spin_unlock(&root->fs_info->trans_lock);
+	mutex_unlock(&root->fs_info->reloc_mutex);
 
 	while (!list_empty(&reloc_roots)) {
 		found = 1;
@@ -3590,17 +3600,19 @@ next:
 static void set_reloc_control(struct reloc_control *rc)
 {
 	struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
-	spin_lock(&fs_info->trans_lock);
+
+	mutex_lock(&fs_info->reloc_mutex);
 	fs_info->reloc_ctl = rc;
-	spin_unlock(&fs_info->trans_lock);
+	mutex_unlock(&fs_info->reloc_mutex);
 }
 
 static void unset_reloc_control(struct reloc_control *rc)
 {
 	struct btrfs_fs_info *fs_info = rc->extent_root->fs_info;
-	spin_lock(&fs_info->trans_lock);
+
+	mutex_lock(&fs_info->reloc_mutex);
 	fs_info->reloc_ctl = NULL;
-	spin_unlock(&fs_info->trans_lock);
+	mutex_unlock(&fs_info->reloc_mutex);
 }
 
 static int check_extent_flags(u64 flags)
diff --git a/fs/btrfs/sysfs.c b/fs/btrfs/sysfs.c
index c3c223ae6691..daac9ae6d731 100644
--- a/fs/btrfs/sysfs.c
+++ b/fs/btrfs/sysfs.c
@@ -28,152 +28,6 @@
 #include "disk-io.h"
 #include "transaction.h"
 
-static ssize_t root_blocks_used_show(struct btrfs_root *root, char *buf)
-{
-	return snprintf(buf, PAGE_SIZE, "%llu\n",
-		(unsigned long long)btrfs_root_used(&root->root_item));
-}
-
-static ssize_t root_block_limit_show(struct btrfs_root *root, char *buf)
-{
-	return snprintf(buf, PAGE_SIZE, "%llu\n",
-		(unsigned long long)btrfs_root_limit(&root->root_item));
-}
-
-static ssize_t super_blocks_used_show(struct btrfs_fs_info *fs, char *buf)
-{
-
-	return snprintf(buf, PAGE_SIZE, "%llu\n",
-		(unsigned long long)btrfs_super_bytes_used(&fs->super_copy));
-}
-
-static ssize_t super_total_blocks_show(struct btrfs_fs_info *fs, char *buf)
-{
-	return snprintf(buf, PAGE_SIZE, "%llu\n",
-		(unsigned long long)btrfs_super_total_bytes(&fs->super_copy));
-}
-
-static ssize_t super_blocksize_show(struct btrfs_fs_info *fs, char *buf)
-{
-	return snprintf(buf, PAGE_SIZE, "%llu\n",
-		(unsigned long long)btrfs_super_sectorsize(&fs->super_copy));
-}
-
-/* this is for root attrs (subvols/snapshots) */
-struct btrfs_root_attr {
-	struct attribute attr;
-	ssize_t (*show)(struct btrfs_root *, char *);
-	ssize_t (*store)(struct btrfs_root *, const char *, size_t);
-};
-
-#define ROOT_ATTR(name, mode, show, store) \
-static struct btrfs_root_attr btrfs_root_attr_##name = __ATTR(name, mode, \
-							      show, store)
-
-ROOT_ATTR(blocks_used,	0444,	root_blocks_used_show,	NULL);
-ROOT_ATTR(block_limit,	0644,	root_block_limit_show,	NULL);
-
-static struct attribute *btrfs_root_attrs[] = {
-	&btrfs_root_attr_blocks_used.attr,
-	&btrfs_root_attr_block_limit.attr,
-	NULL,
-};
-
-/* this is for super attrs (actual full fs) */
-struct btrfs_super_attr {
-	struct attribute attr;
-	ssize_t (*show)(struct btrfs_fs_info *, char *);
-	ssize_t (*store)(struct btrfs_fs_info *, const char *, size_t);
-};
-
-#define SUPER_ATTR(name, mode, show, store) \
-static struct btrfs_super_attr btrfs_super_attr_##name = __ATTR(name, mode, \
-								show, store)
-
-SUPER_ATTR(blocks_used,		0444,	super_blocks_used_show,		NULL);
-SUPER_ATTR(total_blocks,	0444,	super_total_blocks_show,	NULL);
-SUPER_ATTR(blocksize,		0444,	super_blocksize_show,		NULL);
-
-static struct attribute *btrfs_super_attrs[] = {
-	&btrfs_super_attr_blocks_used.attr,
-	&btrfs_super_attr_total_blocks.attr,
-	&btrfs_super_attr_blocksize.attr,
-	NULL,
-};
-
-static ssize_t btrfs_super_attr_show(struct kobject *kobj,
-				    struct attribute *attr, char *buf)
-{
-	struct btrfs_fs_info *fs = container_of(kobj, struct btrfs_fs_info,
-						super_kobj);
-	struct btrfs_super_attr *a = container_of(attr,
-						  struct btrfs_super_attr,
-						  attr);
-
-	return a->show ? a->show(fs, buf) : 0;
-}
-
-static ssize_t btrfs_super_attr_store(struct kobject *kobj,
-				     struct attribute *attr,
-				     const char *buf, size_t len)
-{
-	struct btrfs_fs_info *fs = container_of(kobj, struct btrfs_fs_info,
-						super_kobj);
-	struct btrfs_super_attr *a = container_of(attr,
-						  struct btrfs_super_attr,
-						  attr);
-
-	return a->store ? a->store(fs, buf, len) : 0;
-}
-
-static ssize_t btrfs_root_attr_show(struct kobject *kobj,
-				    struct attribute *attr, char *buf)
-{
-	struct btrfs_root *root = container_of(kobj, struct btrfs_root,
-						root_kobj);
-	struct btrfs_root_attr *a = container_of(attr,
-						 struct btrfs_root_attr,
-						 attr);
-
-	return a->show ? a->show(root, buf) : 0;
-}
-
-static ssize_t btrfs_root_attr_store(struct kobject *kobj,
-				     struct attribute *attr,
-				     const char *buf, size_t len)
-{
-	struct btrfs_root *root = container_of(kobj, struct btrfs_root,
-						root_kobj);
-	struct btrfs_root_attr *a = container_of(attr,
-						 struct btrfs_root_attr,
-						 attr);
-	return a->store ? a->store(root, buf, len) : 0;
-}
-
-static void btrfs_super_release(struct kobject *kobj)
-{
-	struct btrfs_fs_info *fs = container_of(kobj, struct btrfs_fs_info,
-						super_kobj);
-	complete(&fs->kobj_unregister);
-}
-
-static void btrfs_root_release(struct kobject *kobj)
-{
-	struct btrfs_root *root = container_of(kobj, struct btrfs_root,
-						root_kobj);
-	complete(&root->kobj_unregister);
-}
-
-static const struct sysfs_ops btrfs_super_attr_ops = {
-	.show	= btrfs_super_attr_show,
-	.store	= btrfs_super_attr_store,
-};
-
-static const struct sysfs_ops btrfs_root_attr_ops = {
-	.show	= btrfs_root_attr_show,
-	.store	= btrfs_root_attr_store,
-};
-
 /* /sys/fs/btrfs/ entry */
 static struct kset *btrfs_kset;
 
diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c
index 2b3590b9fe98..51dcec86757f 100644
--- a/fs/btrfs/transaction.c
+++ b/fs/btrfs/transaction.c
@@ -126,28 +126,85 @@ static noinline int join_transaction(struct btrfs_root *root, int nofail)
  * to make sure the old root from before we joined the transaction is deleted
  * when the transaction commits
  */
-int btrfs_record_root_in_trans(struct btrfs_trans_handle *trans,
+static int record_root_in_trans(struct btrfs_trans_handle *trans,
 			       struct btrfs_root *root)
 {
 	if (root->ref_cows && root->last_trans < trans->transid) {
 		WARN_ON(root == root->fs_info->extent_root);
 		WARN_ON(root->commit_root != root->node);
 
+		/*
+		 * see below for in_trans_setup usage rules
+		 * we have the reloc mutex held now, so there
+		 * is only one writer in this function
+		 */
+		root->in_trans_setup = 1;
+
+		/* make sure readers find in_trans_setup before
+		 * they find our root->last_trans update
+		 */
+		smp_wmb();
+
 		spin_lock(&root->fs_info->fs_roots_radix_lock);
 		if (root->last_trans == trans->transid) {
 			spin_unlock(&root->fs_info->fs_roots_radix_lock);
 			return 0;
 		}
-		root->last_trans = trans->transid;
 		radix_tree_tag_set(&root->fs_info->fs_roots_radix,
 			   (unsigned long)root->root_key.objectid,
 			   BTRFS_ROOT_TRANS_TAG);
 		spin_unlock(&root->fs_info->fs_roots_radix_lock);
+		root->last_trans = trans->transid;
+
+		/* this is pretty tricky.  We don't want to
+		 * take the relocation lock in btrfs_record_root_in_trans
+		 * unless we're really doing the first setup for this root in
+		 * this transaction.
+		 *
+		 * Normally we'd use root->last_trans as a flag to decide
+		 * if we want to take the expensive mutex.
+		 *
+		 * But, we have to set root->last_trans before we
+		 * init the relocation root, otherwise, we trip over warnings
+		 * in ctree.c.  The solution used here is to flag ourselves
+		 * with root->in_trans_setup.  When this is 1, we're still
+		 * fixing up the reloc trees and everyone must wait.
+		 *
+		 * When this is zero, they can trust root->last_trans and fly
+		 * through btrfs_record_root_in_trans without having to take the
+		 * lock.  smp_wmb() makes sure that all the writes above are
+		 * done before we pop in the zero below
+		 */
 		btrfs_init_reloc_root(trans, root);
+		smp_wmb();
+		root->in_trans_setup = 0;
 	}
 	return 0;
 }
 
+
+int btrfs_record_root_in_trans(struct btrfs_trans_handle *trans,
+			       struct btrfs_root *root)
+{
+	if (!root->ref_cows)
+		return 0;
+
+	/*
+	 * see record_root_in_trans for comments about in_trans_setup usage
+	 * and barriers
+	 */
+	smp_rmb();
+	if (root->last_trans == trans->transid &&
+	    !root->in_trans_setup)
+		return 0;
+
+	mutex_lock(&root->fs_info->reloc_mutex);
+	record_root_in_trans(trans, root);
+	mutex_unlock(&root->fs_info->reloc_mutex);
+
+	return 0;
+}
+
 /* wait for commit against the current transaction to become unblocked
  * when this is done, it is safe to start a new transaction, but the current
  * transaction might not be fully on disk.
@@ -882,7 +939,7 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	parent = dget_parent(dentry);
 	parent_inode = parent->d_inode;
 	parent_root = BTRFS_I(parent_inode)->root;
-	btrfs_record_root_in_trans(trans, parent_root);
+	record_root_in_trans(trans, parent_root);
 
 	/*
 	 * insert the directory item
@@ -900,7 +957,16 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans,
 	ret = btrfs_update_inode(trans, parent_root, parent_inode);
 	BUG_ON(ret);
 
-	btrfs_record_root_in_trans(trans, root);
+	/*
+	 * pull in the delayed directory update
+	 * and the delayed inode item
+	 * otherwise we corrupt the FS during
+	 * snapshot
+	 */
+	ret = btrfs_run_delayed_items(trans, root);
+	BUG_ON(ret);
+
+	record_root_in_trans(trans, root);
 	btrfs_set_root_last_snapshot(&root->root_item, trans->transid);
 	memcpy(new_root_item, &root->root_item, sizeof(*new_root_item));
 	btrfs_check_and_init_root_item(new_root_item);
@@ -961,14 +1027,6 @@ static noinline int create_pending_snapshots(struct btrfs_trans_handle *trans,
 	int ret;
 
 	list_for_each_entry(pending, head, list) {
-		/*
-		 * We must deal with the delayed items before creating
-		 * snapshots, or we will create a snapthot with inconsistent
-		 * information.
-		*/
-		ret = btrfs_run_delayed_items(trans, fs_info->fs_root);
-		BUG_ON(ret);
-
 		ret = create_pending_snapshot(trans, fs_info, pending);
 		BUG_ON(ret);
 	}
@@ -1241,21 +1299,42 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 			schedule_timeout(1);
 
 		finish_wait(&cur_trans->writer_wait, &wait);
-		spin_lock(&root->fs_info->trans_lock);
-		root->fs_info->trans_no_join = 1;
-		spin_unlock(&root->fs_info->trans_lock);
 	} while (atomic_read(&cur_trans->num_writers) > 1 ||
 		 (should_grow && cur_trans->num_joined != joined));
 
-	ret = create_pending_snapshots(trans, root->fs_info);
-	BUG_ON(ret);
+	/*
+	 * Ok now we need to make sure to block out any other joins while we
+	 * commit the transaction.  We could have started a join before setting
+	 * no_join so make sure to wait for num_writers to == 1 again.
+	 */
+	spin_lock(&root->fs_info->trans_lock);
+	root->fs_info->trans_no_join = 1;
+	spin_unlock(&root->fs_info->trans_lock);
+	wait_event(cur_trans->writer_wait,
+		   atomic_read(&cur_trans->num_writers) == 1);
+
+	/*
+	 * the reloc mutex makes sure that we stop
+	 * the balancing code from coming in and moving
+	 * extents around in the middle of the commit
+	 */
+	mutex_lock(&root->fs_info->reloc_mutex);
 
 	ret = btrfs_run_delayed_items(trans, root);
 	BUG_ON(ret);
 
+	ret = create_pending_snapshots(trans, root->fs_info);
+	BUG_ON(ret);
+
 	ret = btrfs_run_delayed_refs(trans, root, (unsigned long)-1);
 	BUG_ON(ret);
 
+	/*
+	 * make sure none of the code above managed to slip in a
+	 * delayed item
+	 */
+	btrfs_assert_delayed_root_empty(root);
+
 	WARN_ON(cur_trans != trans->transaction);
 
 	btrfs_scrub_pause(root);
@@ -1312,6 +1391,7 @@ int btrfs_commit_transaction(struct btrfs_trans_handle *trans,
 	root->fs_info->running_transaction = NULL;
 	root->fs_info->trans_no_join = 0;
 	spin_unlock(&root->fs_info->trans_lock);
+	mutex_unlock(&root->fs_info->reloc_mutex);
 
 	wake_up(&root->fs_info->transaction_wait);
 
diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c
index 592396c6dc47..4ce8a9f41d1e 100644
--- a/fs/btrfs/tree-log.c
+++ b/fs/btrfs/tree-log.c
@@ -3177,7 +3177,7 @@ again:
 		tmp_key.offset = (u64)-1;
 
 		wc.replay_dest = btrfs_read_fs_root_no_name(fs_info, &tmp_key);
-		BUG_ON(!wc.replay_dest);
+		BUG_ON(IS_ERR_OR_NULL(wc.replay_dest));
 
 		wc.replay_dest->log_root = log;
 		btrfs_record_root_in_trans(trans, wc.replay_dest);