Re: PCI IRQ routing problem in 2.4.0 (SiS results part 2)

[Linus Torvalds]

On Mon, 29 Jan 2001, Aaron Tiensivu wrote:
> | Which one was it you got a PIRQ conflict for before? as it te device at
> | 00:01.00 with the strange "0x62" entry?
> 
> Yes.

You've got the pirq setup from hell.

Mind doing that "dump_pirq" thing, preferably run on an _unmodified_ 2.4.0
kernel (ie none of my test-hacks) or on a 2.2.x kernel (that doesn't even
_try_ to do any routing at all - or you can get the same effect by just
making "set_sis_pirq()" always just return 0 without doing anything)?

I don't want to get quantum effects of the observer changing what is being
observed..

> | How about you try adding the line
> | pirq = (pirq-1) & 3;
> | at the top of both pirq_sis_get() and pirq_sis_set() (with my "alternate"
> | SiS routines). What happens then?
> 
> Done.

Not any better. The thing just moves around, it seems.

I'll think about this, and try to find the SiS irq routing register spec
somewhere..

		Linus