On Thu, 14 Mar 2024 at 00:10, syzbot
<syzbot+e525d9be15a106e48379@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote:
> -> #3 (&ovl_i_mutex_dir_key[depth]){++++}-{3:3}:
> lock_acquire+0x1e3/0x530 kernel/locking/lockdep.c:5754
> down_read+0xb1/0xa40 kernel/locking/rwsem.c:1526
> inode_lock_shared include/linux/fs.h:814 [inline]
> lookup_slow+0x45/0x70 fs/namei.c:1709
> walk_component+0x2e1/0x410 fs/namei.c:2005
> lookup_last fs/namei.c:2462 [inline]
> path_lookupat+0x16f/0x450 fs/namei.c:2486
> filename_lookup+0x255/0x610 fs/namei.c:2515
> kern_path+0x35/0x50 fs/namei.c:2623
This is what ultimately closes the locking loop: doing path lookup
(taking directory i_mutex) with kernfs of->mutex held.
#syz dup possible deadlock in seq_read_iter (3)
Thanks,
Miklos
