Re: [syzbot] [kernfs?] possible deadlock in kernfs_remove

[Hillf Danton <hdanton@sina.com>]

> Date: Mon, 04 Aug 2025 11:08:26 -0700	[thread overview]
> Hello,
> 
> syzbot found the following issue on:
> 
> HEAD commit:    759dfc7d04ba netlink: avoid infinite retry looping in netl..
> git tree:       net
> console+strace: https://syzkaller.appspot.com/x/log.txt?x=11332f82580000
> kernel config:  https://syzkaller.appspot.com/x/.config?x=ac0888b9ad46cd69
> dashboard link: https://syzkaller.appspot.com/bug?extid=2d7d0fbb5fb979113ff3
> syz repro:      https://syzkaller.appspot.com/x/repro.syz?x=1136d9bc580000
> C reproducer:   https://syzkaller.appspot.com/x/repro.c?x=1536d9bc580000

#syz test upstream master

--- x/fs/kernfs/mount.c
+++ y/fs/kernfs/mount.c
@@ -303,9 +303,7 @@ static int kernfs_fill_super(struct supe
 	sb->s_shrink->seeks = 0;
 
 	/* get root inode, initialize and unlock it */
-	down_read(&kf_root->kernfs_rwsem);
 	inode = kernfs_get_inode(sb, info->root->kn);
-	up_read(&kf_root->kernfs_rwsem);
 	if (!inode) {
 		pr_debug("kernfs: could not get root inode\n");
 		return -ENOMEM;
--