Yes, I should have done it. Although next patch reconstructs do_rename() which makes this ugly judgement disappered. I will fix it along with other suggestions from other patches in next iteration. Thanks.diff --git a/fs/ubifs/dir.c b/fs/ubifs/dir.c
index 2735ad1affed..6503e6857f6e 100644
--- a/fs/ubifs/dir.c
+++ b/fs/ubifs/dir.c
@@ -1334,6 +1334,8 @@ static int do_rename(struct inode *old_dir, struct dentry *old_dentry,
err = do_tmpfile(old_dir, old_dentry, S_IFCHR | WHITEOUT_MODE, &whiteout);
if (err) {
+ if (whiteout)
+ iput(whiteout);
Should this rather be done in do_tmpfile() directly?
Sascha