Re: Races in net_rx_action vs netpoll?

[Olaf Kirch <okir@lst.de>]

On Wednesday 11 July 2007 07:44, David Miller wrote:
> > +#ifdef CONFIG_NETPOLL
> > +	/* Prevent race with netpoll - yes, this is a kludge.
> > +	 * But at least it doesn't penalize the non-netpoll
> > +	 * code path. */
> > +	if (test_bit(__LINK_STATE_POLL_LIST_FROZEN, &dev->state))
> > +		return;
> > +#endif
> > +
> >  	local_irq_save(flags);
> >  	BUG_ON(!test_bit(__LINK_STATE_RX_SCHED, &dev->state));
> >  	list_del(&dev->poll_list);
> 
> That new bit can be set in interrupt context can't it?

It's set and cleared in poll_napi only, and as far as I can tell 
poll_napi will only ever be called from via softirq, but never
from an interrupt handler directly.

I also don't think the test_bit() needs to lock out interrupts.
The only reason we do it for the RX_SCHED bit is that the RX_SCHED
bit and the poll_list change must happen atomically wrt interrupts
from the NIC, right?

Olaf
-- 
Olaf Kirch  |  --- o --- Nous sommes du soleil we love when we play
okir@lst.de |    / | \   sol.dhoop.naytheet.ah kin.ir.samse.qurax