Re: [syzbot] WARNING: locking bug in hfa384x_usbctlx_completion_task

[Greg KH]

On Mon, Jun 27, 2022 at 10:08:38AM -0700, Davidlohr Bueso wrote:
> It's nice to see this sort of stuff actually tested :)
> 
> On Sun, 26 Jun 2022, syzbot wrote:
> 
> > ------------[ cut here ]------------
> > DEBUG_LOCKS_WARN_ON(1)
> > WARNING: CPU: 1 PID: 21 at kernel/locking/lockdep.c:231 hlock_class kernel/locking/lockdep.c:231 [inline]
> > WARNING: CPU: 1 PID: 21 at kernel/locking/lockdep.c:231 hlock_class kernel/locking/lockdep.c:220 [inline]
> > WARNING: CPU: 1 PID: 21 at kernel/locking/lockdep.c:231 check_wait_context kernel/locking/lockdep.c:4727 [inline]
> > WARNING: CPU: 1 PID: 21 at kernel/locking/lockdep.c:231 __lock_acquire+0x1356/0x5660 kernel/locking/lockdep.c:5003
> > Modules linked in:
> > CPU: 1 PID: 21 Comm: kworker/1:0 Not tainted 5.19.0-rc3-syzkaller-00071-g105f3fd2f789 #0
> > Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
> > Workqueue: events hfa384x_usbctlx_completion_task
> 
> Sorry about that, the below should fix it. Thanks.
> 
> ---8<-------------------------------
> From: Davidlohr Bueso <dave@xxxxxxxxxxxx>
> Subject: [PATCH] staging/wlan-ng: get the correct struct hfa384x in work callback
> 
> hfa384x_usbctlx_completion_task() is bogusly using the reaper BH when
> in fact this is the completion_bh. This was reflected when trying
> to acquire the hw->ctlxq.lock and getting a failed lockdep class
> initialized to it.
> 
> Reported-by: syzbot+ce3408364c4a234dd90c@xxxxxxxxxxxxxxxxxxxxxxxxx
> Fixes: 9442e81d7e7 (staging/wlan-ng, prism2usb: replace completion_bh tasklet with work)
> Signed-off-by: Davidlohr Bueso <dave@xxxxxxxxxxxx>
> ---
>  drivers/staging/wlan-ng/hfa384x_usb.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)

Can you submit this properly so that I can accept it?

thanks,

greg k-h