On Mon, 30 Nov 2009, Uwe Kleine-König wrote:
> I think there is
> 
>  3) you can only benefit from decent priority hardware if irqs are
>     processed while irqs are enabled.
> 
> I think 
> 
> 	git grep handle_fasteoi_irq
> 
> gives an overview here: some hits in arch/powerpc, arch/sparc and
> arch/x86/kernel/apic/io_apic.c.  (There is handle_prio_irq in

No. That handler is not an indicator for prio hardware actively used
in the sense of allowing higher prio interrupts to interrupt a current
running lower priority one. It can be used when the irq controller
does not fire the interrupt again before the eoi acknowledge has been
done.

Thanks,

	tglx