Re: Posssible bug in kernel/irq/handle.c

[Benjamin Herrenschmidt <benh@kernel.crashing.org>]

> Be vewy vewy caweful when changing that code, though. If you end up with a 
> patch, please try to give it some nice stress-testing (both on ppc and 
> x86), and then post it for comments, ok? Maybe the arch mailing list and 
> Ingo (who else has touched that logic?)

Ok, I'll try to avoid touching that code. In a perfect world, we should
have proper handlers for those firmware interrupts anyway, it's just
that the "spec" says we should call the firmware for any interrupt we
don't handle...

I suppose it should be enough for us to test for desc->action before
calling __do_IRQ() and eventually do the firmware trick then, since I
doubt that if it matters at all, it will happen on shared interrupts...

Ben.