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

Pull btrfs data corruption fix from Chris Mason:
 "I'm testing a pull with more fixes, but wanted to get this one out so
  Greg can pick it up.

  The corruption isn't easy to hit, you have to do a readonly snapshot
  and have orphans in the snapshot.  But my review and testing missed
  the bug.  Filipe has added a better xfstest to cover it"

* 'for-linus-update' of git://git.kernel.org/pub/scm/linux/kernel/git/mason/linux-btrfs:
  Revert "Btrfs: race free update of commit root for ro snapshots"

2 files changed, 33 insertions, 36 deletions

diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c
index fc9c0439caa3..d23362f4464e 100644
--- a/fs/btrfs/inode.c
+++ b/fs/btrfs/inode.c
@@ -5261,42 +5261,6 @@ struct inode *btrfs_lookup_dentry(struct inode *dir, struct dentry *dentry)
 			iput(inode);
 			inode = ERR_PTR(ret);
 		}
-		/*
-		 * If orphan cleanup did remove any orphans, it means the tree
-		 * was modified and therefore the commit root is not the same as
-		 * the current root anymore. This is a problem, because send
-		 * uses the commit root and therefore can see inode items that
-		 * don't exist in the current root anymore, and for example make
-		 * calls to btrfs_iget, which will do tree lookups based on the
-		 * current root and not on the commit root. Those lookups will
-		 * fail, returning a -ESTALE error, and making send fail with
-		 * that error. So make sure a send does not see any orphans we
-		 * have just removed, and that it will see the same inodes
-		 * regardless of whether a transaction commit happened before
-		 * it started (meaning that the commit root will be the same as
-		 * the current root) or not.
-		 */
-		if (sub_root->node != sub_root->commit_root) {
-			u64 sub_flags = btrfs_root_flags(&sub_root->root_item);
-
-			if (sub_flags & BTRFS_ROOT_SUBVOL_RDONLY) {
-				struct extent_buffer *eb;
-
-				/*
-				 * Assert we can't have races between dentry
-				 * lookup called through the snapshot creation
-				 * ioctl and the VFS.
-				 */
-				ASSERT(mutex_is_locked(&dir->i_mutex));
-
-				down_write(&root->fs_info->commit_root_sem);
-				eb = sub_root->commit_root;
-				sub_root->commit_root =
-					btrfs_root_node(sub_root);
-				up_write(&root->fs_info->commit_root_sem);
-				free_extent_buffer(eb);
-			}
-		}
 	}
 
 	return inode;
diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c
index 0fe1aa047f15..8d2b76e29d3b 100644
--- a/fs/btrfs/ioctl.c
+++ b/fs/btrfs/ioctl.c
@@ -713,6 +713,39 @@ static int create_snapshot(struct btrfs_root *root, struct inode *dir,
 	if (ret)
 		goto fail;
 
+	ret = btrfs_orphan_cleanup(pending_snapshot->snap);
+	if (ret)
+		goto fail;
+
+	/*
+	 * If orphan cleanup did remove any orphans, it means the tree was
+	 * modified and therefore the commit root is not the same as the
+	 * current root anymore. This is a problem, because send uses the
+	 * commit root and therefore can see inode items that don't exist
+	 * in the current root anymore, and for example make calls to
+	 * btrfs_iget, which will do tree lookups based on the current root
+	 * and not on the commit root. Those lookups will fail, returning a
+	 * -ESTALE error, and making send fail with that error. So make sure
+	 * a send does not see any orphans we have just removed, and that it
+	 * will see the same inodes regardless of whether a transaction
+	 * commit happened before it started (meaning that the commit root
+	 * will be the same as the current root) or not.
+	 */
+	if (readonly && pending_snapshot->snap->node !=
+	    pending_snapshot->snap->commit_root) {
+		trans = btrfs_join_transaction(pending_snapshot->snap);
+		if (IS_ERR(trans) && PTR_ERR(trans) != -ENOENT) {
+			ret = PTR_ERR(trans);
+			goto fail;
+		}
+		if (!IS_ERR(trans)) {
+			ret = btrfs_commit_transaction(trans,
+						       pending_snapshot->snap);
+			if (ret)
+				goto fail;
+		}
+	}
+
 	inode = btrfs_lookup_dentry(dentry->d_parent->d_inode, dentry);
 	if (IS_ERR(inode)) {
 		ret = PTR_ERR(inode);