gfs2: don't return ENODATA in __gfs2_xattr_set unless replacing

The function __gfs2_xattr_set() will return -ENODATA when called to
remove a xattr that does not exist. The result is that setfacl will
show an exit status of 1 when called to set only a file's mode bits
(on a file with no ACLs), despite succeeding. A "No data available"
error will be printed as well.

To fix this return 0 instead, except when the XATTR_REPLACE flag is
set, in which case -ENODATA is appropriate. This is consistent with
how most other xattr setting functions work, in other filesystems.

Signed-off-by: Ernesto A. Fernández <ernesto.mnd.fernandez@gmail.com>
Signed-off-by: Bob Peterson <rpeterso@redhat.com>
diff --git a/fs/gfs2/xattr.c b/fs/gfs2/xattr.c
index cf694de..ea09e41 100644
--- a/fs/gfs2/xattr.c
+++ b/fs/gfs2/xattr.c
@@ -1210,8 +1210,12 @@
 	if (namel > GFS2_EA_MAX_NAME_LEN)
 		return -ERANGE;
 
-	if (value == NULL)
-		return gfs2_xattr_remove(ip, type, name);
+	if (value == NULL) {
+		error = gfs2_xattr_remove(ip, type, name);
+		if (error == -ENODATA && !(flags & XATTR_REPLACE))
+			error = 0;
+		return error;
+	}
 
 	if (ea_check_size(sdp, namel, size))
 		return -ERANGE;