Merge tag 'v6.5/vfs.fixes.2' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs

Pull vfs fixes from Christian Brauner:
 "This contains two minor fixes for Jan's rename locking work:

   - Unlocking the source inode was guarded by a check whether source
     was non-NULL. This doesn't make sense because source must be
     non-NULL and the commit message explains in detail why

   - The lock_two_nondirectories() helper called WARN_ON_ONCE() and
     dereferenced the inodes unconditionally but the underlying
     lock_two_inodes() helper and the kernel documentation for that
     function are clear that it is valid to pass NULL arguments, so a
     non-NULL check is needed. No callers currently pass NULL arguments
     but let's not knowingly leave landmines around"

* tag 'v6.5/vfs.fixes.2' of git://git.kernel.org/pub/scm/linux/kernel/git/vfs/vfs:
  fs: don't assume arguments are non-NULL
  fs: no need to check source

Author: torvalds

fs/inode.c

@@ -1156,8 +1156,10 @@ void lock_two_inodes(struct inode *inode1, struct inode *inode2,
  */
 void lock_two_nondirectories(struct inode *inode1, struct inode *inode2)
 {
-	WARN_ON_ONCE(S_ISDIR(inode1->i_mode));
-	WARN_ON_ONCE(S_ISDIR(inode2->i_mode));
+	if (inode1)
+		WARN_ON_ONCE(S_ISDIR(inode1->i_mode));
+	if (inode2)
+		WARN_ON_ONCE(S_ISDIR(inode2->i_mode));
 	lock_two_inodes(inode1, inode2, I_MUTEX_NORMAL, I_MUTEX_NONDIR2);
 }
 EXPORT_SYMBOL(lock_two_nondirectories);

fs/namei.c

@@ -4874,8 +4874,7 @@ int vfs_rename(struct renamedata *rd)
 			d_exchange(old_dentry, new_dentry);
 	}
 out:
-	if (source)
-		inode_unlock(source);
+	inode_unlock(source);
 	if (target)
 		inode_unlock(target);
 	dput(new_dentry);