Re: Multiple OOPS with Linux-2.5.40

[Johannes Erdfelt]

On Tue, Oct 01, 2002, Greg KH <[e-mail]> wrote:
> On Tue, Oct 01, 2002 at 08:25:17PM -0500, Krishnakumar B wrote:
> > hcd-pci.c: uhci-hcd @ 00:1f.2, Intel Corp. 82801AA USB
> > hcd-pci.c: irq 19, io base 0000ff80
> > hcd.c: new USB bus registered, assigned bus number 1
> > hub.c: USB hub found at 0
> > hub.c: 2 ports detected
> > Debug: sleeping function called from illegal context at /u/scratch/downloads/kernel/linux-2.5.40/include/asm/semaphore.h:119
> > df675f70 e2886692 e28926c0 00000077 df675fa8 df808760 df664f88 df66e0d0 
> >        df674000 dfb36100 00000286 df674000 df674000 df675fc0 df674000 e28869f5 
> >        dfa1b860 dfb36100 df674000 df81df34 00000000 dfa1b860 c0118260 00000000 
> > Call Trace:
> >  [<e2886692>]usb_hub_events+0x82/0x3b0 [usbcore]
> >  [<e28926c0>].rodata.str1.32+0xd20/0x28e2 [usbcore]
> >  [<e28869f5>]usb_hub_thread+0x35/0x100 [usbcore]
> >  [<c0118260>]default_wake_function+0x0/0x40
> >  [<e28869c0>]usb_hub_thread+0x0/0x100 [usbcore]
> >  [<c0105665>]kernel_thread_helper+0x5/0x10
> 
> I think this is due to the following lines of code in
> drivers/usb/core/hub.c:
> 		spin_lock_irqsave(&hub_event_lock, flags);
> 
> 		....
> 
> 		down(&hub->khubd_sem); /* never blocks, we were on list */
> 		spin_unlock_irqrestore(&hub_event_lock, flags);
> 
> Johannes, any reason for the down() when we have a spinlock held?

It was a patch from Pete to fix a race condition in the hub driver. It
looks to be safe, but obviously should be cleaned up.

I'm working on a reference counting patch to remove this bit.

See the thread on linux-usb-devel that David started about this problem.

JE