ACPI 100002 IRQ 9 problem.

ACPI 100002 IRQ 9 problem.

[Joshua Schmidlkofer]

Hey all,

   First is there a different list for ACPI questions?

   For the sake of disclosure, the dump I am reporting is 2.5.73, plus
the Davide Libenzis' SiS-96x patch.  I also am currently using the
nvidia driver, I was able to reproduce this in vanilla 2.5.72 (no
nvidia), and I will try tomorrow with a vanilla setup of 2.5.73-bk3.
[unless bk4 is available].

   I have a Soyo P4S-645D, with the SiS 645 chipset.  I have had some
problems w/ the IRQ routing, but 2.5.7[123] have sorted it out (mostly) 
I am having problems ACPI, it is better if I say 'pci=noacpi', but what
happens is when the ACPI interrupt count hit 100002, then I get the
following message on all consoles:

menion kernel: Disabling IRQ #9


Then, I have the following as part of dmesg:

Call Trace: [<c010cad4>]  [<c010cbad>]  [<c010ce46>]  [<c010880e>] 
[<c010880e>]  [<c010b320>]  [<c010880e>]  [<c010880e>]  [<c0108832>] 
[<c010889a>]  [<c0105000>]  [<c041c6bd>]  [<c041c41e>]
[<c0247446>]
Warning (Oops_read): Code line not seen, dumping what data is available


Trace; c010cad4 <__report_bad_irq+2a/7a>
Trace; c010cbad <note_interrupt+6f/a0>
Trace; c010ce46 <do_IRQ+130/140>
Trace; c010880e <default_idle+0/28>
Trace; c010880e <default_idle+0/28>
Trace; c010b320 <common_interrupt+18/20>
Trace; c010880e <default_idle+0/28>
Trace; c010880e <default_idle+0/28>
Trace; c0108832 <default_idle+24/28>
Trace; c010889a <cpu_idle+2e/38>
Trace; c0105000 <_stext+0/0>
Trace; c041c6bd <start_kernel+175/19c>
Trace; c041c41e <unknown_bootoption+0/fc>
Trace; c0247446 <acpi_irq+0/16>

If I say 'pci=noacpi' it take about 20 minutes. The ACPI interrupt count
begins at a reasonable low number and climbs over the course of 20
minutes.

If i don't add the pci line, then I start out w/ ACPI interrupt count at
100000 interrupt count, and about 5 - 10 minutes later, I get 2 ACPI
interrupts and then the message follows.

SiS 645 chipset
P4 2.4 gig
PC2700 ram
3 HD's, 2 CDRW's.
NVidia Geforce4, w/ and w/o binary driver.[4363 + www.minion.de patches]

RedHat 8.0, gcc 3.2, 2.5.72 & 2.5.73.  

2.5.73 only verified w/ NVIDIA driver, not w/o.  

RE: ACPI 100002 IRQ 9 problem.

[Grover, Andrew]

> From: Joshua Schmidlkofer [mailto:menion@asylumwear.com] 
>    First is there a different list for ACPI questions?

>From MAINTAINERS:

ACPI
P:	Andy Grover
M:	andrew.grover@intel.com
L:	acpi-devel@lists.sourceforge.net
W:	http://sf.net/projects/acpi/
S:	Maintained

>    For the sake of disclosure, the dump I am reporting is 2.5.73, plus
> the Davide Libenzis' SiS-96x patch.  I also am currently using the
> nvidia driver, I was able to reproduce this in vanilla 2.5.72 (no
> nvidia), and I will try tomorrow with a vanilla setup of 2.5.73-bk3.
> [unless bk4 is available].
> 
>    I have a Soyo P4S-645D, with the SiS 645 chipset.  I have had some
> problems w/ the IRQ routing, but 2.5.7[123] have sorted it 
> out (mostly) 
> I am having problems ACPI, it is better if I say 
> 'pci=noacpi', but what
> happens is when the ACPI interrupt count hit 100002, then I get the
> following message on all consoles:
> 
> menion kernel: Disabling IRQ #9
> 
> Then, I have the following as part of dmesg:
> 
> Call Trace: [<c010cad4>]  [<c010cbad>]  [<c010ce46>]  [<c010880e>] 
> [<c010880e>]  [<c010b320>]  [<c010880e>]  [<c010880e>]  [<c0108832>] 
> [<c010889a>]  [<c0105000>]  [<c041c6bd>]  [<c041c41e>]
> [<c0247446>]
> Warning (Oops_read): Code line not seen, dumping what data is 
> available

There is a known, as-yet-unfixed problem, but the usual symptom is you
hit 100000 interrupts and then it gets nicely disabled - I'm not sure
why your system oopses.

Regards -- Andy