2007-09-17 13:08:19

by Konstantin Sharlaimov

[permalink] [raw]
Subject: Re: 2.6.22.1: Enabling IO-APIC = APIC error on CPU0: 40(40)

I am experiencing the similar problem on my Acer Aspire 5000 laptop -
once in a while a bunch of "APIC error on CPU0: 40(40)" messages are
showing up in dmesg.

After few experiments and a lot of googling, I identified a cause - it
was my built-in wireless card! Disabling IO-APIC helped, besides a
single "Spurious 8259A interrupt: IRQ7" message while booting up there
was nothing unusual.

A did some further experiments on various hardware and from my
experience I can tell that "APIC error" messages are entirely hardware
related.

Regards,
Konstantin

On Sat, 2007-07-14 at 08:09 -0400, Justin Piszcz wrote:
> I have an older MSI motherboard and when I enable the following option:
>
> >From lshw:
> description: Motherboard
> product: MS-6730
>
> [*] IO-APIC support on uniprocessors
>
> I get the following errors in dmesg:
>
> [ 247.177372] APIC error on CPU0: 00(40)
> [ 247.556741] APIC error on CPU0: 40(40)
> [ 262.010412] APIC error on CPU0: 40(40)
> [ 262.765553] APIC error on CPU0: 40(40)
> [ 263.876481] APIC error on CPU0: 40(40)
> [ 264.191962] APIC error on CPU0: 40(40)
> [ 265.040566] APIC error on CPU0: 40(40)
> [ 267.959689] APIC error on CPU0: 40(40)
> [ 269.446029] APIC error on CPU0: 40(40)
> [ 269.853344] APIC error on CPU0: 40(40)
> [ 271.122825] APIC error on CPU0: 40(40)
> [ 271.287051] APIC error on CPU0: 40(40)
> [ 273.825237] APIC error on CPU0: 40(40)
> [ 274.373737] APIC error on CPU0: 40(40)
> [ 275.355775] APIC error on CPU0: 40(40)
> [ 275.982087] APIC error on CPU0: 40(40)
> [ 276.101181] APIC error on CPU0: 40(40)
> [ 276.173611] APIC error on CPU0: 40(40)
> [ 276.970970] APIC error on CPU0: 40(40)
> [ 277.561575] APIC error on CPU0: 40(40)
> [ 278.694486] APIC error on CPU0: 40(40)
> [ 279.353573] APIC error on CPU0: 40(40)
> [ 281.779398] APIC error on CPU0: 40(40)
> [ 283.625457] APIC error on CPU0: 40(40)
> [ 284.749322] APIC error on CPU0: 40(40)
> [ 284.983013] APIC error on CPU0: 40(40)
> [ 285.126394] APIC error on CPU0: 40(40)
>
> I normally do not use this option (which also gets rid of the errors*1),
> but I am curious, why do these errors occur? I remember them with the early
> 2.6.x series and they are still re-occurring in 2.6.22.1.
>
> The weird part-- is when IO-APIC is disabled, I see this on boot:
>
> IO-APIC DISABLED:
>
> [ 46.902712] XFS mounting filesystem md0
> [ 47.017322] spurious 8259A interrupt: IRQ7.
> [ 47.273009] Ending clean XFS mount for filesystem: md0
>
> IO-APIC ENABLED:
>
> [ 46.851012] XFS mounting filesystem md0
> [ 47.103066] Ending clean XFS mount for filesystem: md0
>
> * 1 Anytime I see a spurious interrupt on (usually older hardware)
> especially during file I/O this usually means corruption (from what I
> have seen) that is why I don't really want to trust anything on this host.
>
> Anyone have any idea what's going on here?
>
> Justin.
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>
>


2007-09-17 13:11:00

by Justin Piszcz

[permalink] [raw]
Subject: Re: 2.6.22.1: Enabling IO-APIC = APIC error on CPU0: 40(40)



On Tue, 18 Sep 2007, Konstantin Sharlaimov wrote:

> I am experiencing the similar problem on my Acer Aspire 5000 laptop -
> once in a while a bunch of "APIC error on CPU0: 40(40)" messages are
> showing up in dmesg.
>
> After few experiments and a lot of googling, I identified a cause - it
> was my built-in wireless card! Disabling IO-APIC helped, besides a
> single "Spurious 8259A interrupt: IRQ7" message while booting up there
> was nothing unusual.
>
> A did some further experiments on various hardware and from my
> experience I can tell that "APIC error" messages are entirely hardware
> related.
>
> Regards,
> Konstantin

Thanks for the follow-up, yeah its probably one of the Promise
controllers or some such that is causing the problem, although I
remmeber I used to get the error with only the motherboard, memory and
video card, something tells me that MSI board with (SiS) chipset is
not quite supported in Linux with APIC enabled.

Justin.


2007-09-17 13:16:49

by Konstantin Sharlaimov

[permalink] [raw]
Subject: Re: 2.6.22.1: Enabling IO-APIC = APIC error on CPU0: 40(40)

I agree, this is probably somehow related with SiS chipset - my laptop
has one of these SiS 760 host controllers inside.

I'll probably do some digging to figure out if this is indeed chipset
related. I wonder if this error occurs under Windows (Win does not
report much, yet it's kernel must be aware of such error).

Konstantin

On Mon, 2007-09-17 at 09:10 -0400, Justin Piszcz wrote:
>
> On Tue, 18 Sep 2007, Konstantin Sharlaimov wrote:
>
> > I am experiencing the similar problem on my Acer Aspire 5000 laptop -
> > once in a while a bunch of "APIC error on CPU0: 40(40)" messages are
> > showing up in dmesg.
> >
> > After few experiments and a lot of googling, I identified a cause - it
> > was my built-in wireless card! Disabling IO-APIC helped, besides a
> > single "Spurious 8259A interrupt: IRQ7" message while booting up there
> > was nothing unusual.
> >
> > A did some further experiments on various hardware and from my
> > experience I can tell that "APIC error" messages are entirely hardware
> > related.
> >
> > Regards,
> > Konstantin
>
> Thanks for the follow-up, yeah its probably one of the Promise
> controllers or some such that is causing the problem, although I
> remmeber I used to get the error with only the motherboard, memory and
> video card, something tells me that MSI board with (SiS) chipset is
> not quite supported in Linux with APIC enabled.
>
> Justin.
>
>

2007-09-18 17:39:19

by Phillip Susi

[permalink] [raw]
Subject: Re: 2.6.22.1: Enabling IO-APIC = APIC error on CPU0: 40(40)

Konstantin Sharlaimov wrote:
> I agree, this is probably somehow related with SiS chipset - my laptop
> has one of these SiS 760 host controllers inside.
>
> I'll probably do some digging to figure out if this is indeed chipset
> related. I wonder if this error occurs under Windows (Win does not
> report much, yet it's kernel must be aware of such error).

Windows does not use IO-APICs AFAIK.