Re: INFO: rcu detected stall in dummy_timer (3)

[Alan Stern]

On Tue, Jul 14, 2020 at 09:27:18AM -0700, syzbot wrote:
> Hello,
> 
> syzbot found the following crash on:
> 
> HEAD commit:    25051b55 udc: lpc32xx: make symbol 'lpc32xx_usbddata' static
> git tree:       https://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb.git usb-testing
> console output: https://syzkaller.appspot.com/x/log.txt?x=12e0ba00900000
> kernel config:  https://syzkaller.appspot.com/x/.config?x=999be4eb2478ffa5
> dashboard link: https://syzkaller.appspot.com/bug?extid=4d3749e9612c2cfab956
> compiler:       gcc (GCC) 10.1.0-syz 20200507
> syz repro:      https://syzkaller.appspot.com/x/repro.syz?x=127354e7100000
> C reproducer:   https://syzkaller.appspot.com/x/repro.c?x=10ba8500900000
> 
> IMPORTANT: if you fix the bug, please add the following tag to the commit:
> Reported-by: syzbot+4d3749e9612c2cfab956@xxxxxxxxxxxxxxxxxxxxxxxxx
> 
> mceusb 5-1:0.0: Error: urb status = -71
> mceusb 3-1:0.0: Error: urb status = -71
> mceusb 4-1:0.0: Error: urb status = -71
> mceusb 6-1:0.0: Error: urb status = -71
> mceusb 5-1:0.0: Error: urb status = -71
> rcu: INFO: rcu_sched self-detected stall on CPU

This looks like an error in mceusb_dev_recv() in 
drivers/media/rc/mceusb.c.  In the case of a -EPROTO error (indeed, any 
error code it doesn't recognize), the routine immediately resubmits the 
URB.  Instead it should do the same thing as in the other error cases: 
return without resubmitting.

(Incidentally, the calls to usb_unlink_urb() in that routine are useless 
and look very strange.  The URB is already unlinked, since 
mceusb_dev_recv() is the completion handler.)

Alan Stern