Re: [patch, -rc5-mm1] locking validator: special rule: 8390.c disable_irq()
From: Ingo Molnar
Date: Wed May 31 2006 - 17:26:20 EST
* Arjan van de Ven <arjan@xxxxxxxxxxxxx> wrote:
> On Wed, 2006-05-31 at 22:02 +0200, Ingo Molnar wrote:
> > untested on 8390 hardware, but ought to solve the lockdep false
> > positive.
> > -----------------
> > Subject: locking validator: special rule: 8390.c disable_irq()
> > From: Ingo Molnar <mingo@xxxxxxx>
> > 8390.c knows that ei_local->page_lock can only be used by an irq
> > context that it disabled -
> btw I think this is no longer correct with the irq polling stuff Alan
> added to the kernel recently...
hm, indeed. misrouted_irq() goes through all irq descriptors and ignores
IRQ_DISABLED flag - rendering disable_irq() useless in essence, and
introducing the kind of deadlocks that lockdep warned about.
Andrew, as far as i can see with irqfixup this isnt a lockdep false
positive but a real deadlock scenario - a spurious IRQ might arrive
during vortex_timer() execution and might cause the execution of
misrouted_irq(), which could execute vortex_interrupt() => deadlock.
Alan, is this a necessary property of irqpoll/irqfixup? Shouldnt
irqfixup leave irq descriptors alone that are disabled?
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/