2004-04-15 23:08:06

by Andi Kleen

[permalink] [raw]
Subject: Re: IO-APIC on nforce2 [PATCH] + [PATCH] for nmi_debug=1 + [PATCH] for idle=C1halt, 2.6.5

Len Brown <[email protected]> writes:

> While I don't want to get into the business of maintaining
> a dmi_scan entry for every system with this issue, I think
> it might be a good idea to add a couple of example entries
> for high volume systems for which there is no BIOS fix available.

Or do a generic fix: check for the PCI-ID of the Nforce2 and when
it is true and the timer is wrong just correct it. That's ugly,
but it's probably the best solution for such a common issue
(and the IO-APIC code is already filled with workarounds anyways)

One problem is that this likely must happen before the PCI quirks
run. In the x86-64 code I have special "early PCI scanning" code
for this that could be copied. I don't have a Nforce2, but when
someone is willing to test I can do a patch for this.

-Andi

P.S.: This problem of reference BIOS bugs getting haunting Linux
even after they are long fixed is unfortunately common :-(


2004-04-21 22:01:00

by Brown, Len

[permalink] [raw]
Subject: Re: IO-APIC on nforce2 [PATCH] + [PATCH] for nmi_debug=1 + [PATCH] for idle=C1halt, 2.6.5

On Thu, 2004-04-15 at 19:07, Andi Kleen wrote:
> Len Brown <[email protected]> writes:
>
> > While I don't want to get into the business of maintaining
> > a dmi_scan entry for every system with this issue, I think
> > it might be a good idea to add a couple of example entries
> > for high volume systems for which there is no BIOS fix available.
>
> Or do a generic fix: check for the PCI-ID of the Nforce2 and when
> it is true and the timer is wrong just correct it. That's ugly,
> but it's probably the best solution for such a common issue
> (and the IO-APIC code is already filled with workarounds anyways)

IMO the fact that the IOAPIC code is full of workarounds is a reason NOT
to add another one.

> One problem is that this likely must happen before the PCI quirks
> run. In the x86-64 code I have special "early PCI scanning" code
> for this that could be copied. I don't have a Nforce2, but when
> someone is willing to test I can do a patch for this.

If this issue had no other fix, I'd agree that the complexity is worth
it. But a BIOS upgrade fixes this -- so I think dmi-scan simplicity
is the way to go.

-Len