On Thursday 24 July 2003 13:58, Rob North wrote:
> Andrew de Quincey wrote:
> > On Wednesday 23 July 2003 08:53, Grover, Andrew wrote:
> >>Andrew,
> >>
> >>You still hacking on this?
> >>
> >>Just got another bug report (completely different HW) that I think might
> >>be the same problem.
> >
> > Yeah, I've been having a wee break for a few days. back on it now.
> >
> > Interesting fact. All my onboard devices work now, but I didn't have any
> > PCI cards. I've just put a wireless PCI card in my machine, and its
> > having the same IRQ problems as I used to with the onoard ones!
>
> I take it that the IRQ polarity issues are now fixed.
> The remaining issues are to handle the cases where ACPI routing fails,
> and cases where the _SRS method fails?
For me, they seem to be fixed for onboard devices, but not for devices plugged
into PCI slots.
I can detect when _SRS fails easily enough as you can do doubt see. What I
need is a way of reliably dropping back to a different configuration method
if _SRS fails. On my board, when I hack it to pretend theres been a failure,
I can't yet get the drop-backed-to mechanism to reconfigure any IRQs which
were configured with _SRS (even if it fails) properly.
|