Quite a few UP x86-64 laptops print APIC error 40's repeatedly
when they run an SMP kernel (And Fedora doesn't ship a UP x86-64 kernel
any more). We can suppress this as there's not really anything we
can do about them.
Signed-off-by: Dave Jones <[email protected]>
--- linux-2.6.14/arch/x86_64/kernel/apic.c~ 2005-12-07 15:17:33.000000000 -0500
+++ linux-2.6.14/arch/x86_64/kernel/apic.c 2005-12-07 15:18:16.000000000 -0500
@@ -1032,7 +1032,8 @@ asmlinkage void smp_error_interrupt(void
6: Received illegal vector
7: Illegal register address
*/
- printk (KERN_DEBUG "APIC error on CPU%d: %02x(%02x)\n",
+ if (num_online_cpus() > 1)
+ printk (KERN_DEBUG "APIC error on CPU%d: %02x(%02x)\n",
smp_processor_id(), v , v1);
irq_exit();
}
On Friday 24 February 2006 02:42, Dave Jones wrote:
> Quite a few UP x86-64 laptops print APIC error 40's repeatedly
> when they run an SMP kernel (And Fedora doesn't ship a UP x86-64 kernel
> any more). We can suppress this as there's not really anything we
> can do about them.
No we need to fix the APIC errors, not hide them.
-Andi
>
> Signed-off-by: Dave Jones <[email protected]>
>
> --- linux-2.6.14/arch/x86_64/kernel/apic.c~ 2005-12-07 15:17:33.000000000 -0500
> +++ linux-2.6.14/arch/x86_64/kernel/apic.c 2005-12-07 15:18:16.000000000 -0500
> @@ -1032,7 +1032,8 @@ asmlinkage void smp_error_interrupt(void
> 6: Received illegal vector
> 7: Illegal register address
> */
> - printk (KERN_DEBUG "APIC error on CPU%d: %02x(%02x)\n",
> + if (num_online_cpus() > 1)
> + printk (KERN_DEBUG "APIC error on CPU%d: %02x(%02x)\n",
> smp_processor_id(), v , v1);
> irq_exit();
> }
>
On Fri, Feb 24, 2006 at 02:45:29AM +0100, Andi Kleen wrote:
> On Friday 24 February 2006 02:42, Dave Jones wrote:
> > Quite a few UP x86-64 laptops print APIC error 40's repeatedly
> > when they run an SMP kernel (And Fedora doesn't ship a UP x86-64 kernel
> > any more). We can suppress this as there's not really anything we
> > can do about them.
>
> No we need to fix the APIC errors, not hide them.
What do you need to fix them ? I've got one laptop here that
is affected, and there's a few other examples with dmesg's
in Red Hat bugzilla that I can trawl.
Dave
On Friday 24 February 2006 02:53, Dave Jones wrote:
> On Fri, Feb 24, 2006 at 02:45:29AM +0100, Andi Kleen wrote:
> > On Friday 24 February 2006 02:42, Dave Jones wrote:
> > > Quite a few UP x86-64 laptops print APIC error 40's repeatedly
> > > when they run an SMP kernel (And Fedora doesn't ship a UP x86-64 kernel
> > > any more). We can suppress this as there's not really anything we
> > > can do about them.
> >
> > No we need to fix the APIC errors, not hide them.
>
> What do you need to fix them ? I've got one laptop here that
> is affected, and there's a few other examples with dmesg's
> in Red Hat bugzilla that I can trawl.
Some pattern analysis would be useful. All the same chipset, revision?
Best you collect boot logs.
-Andi
On Fri, Feb 24, 2006 at 03:18:11AM +0100, Andi Kleen wrote:
> On Friday 24 February 2006 02:53, Dave Jones wrote:
> > On Fri, Feb 24, 2006 at 02:45:29AM +0100, Andi Kleen wrote:
> > > On Friday 24 February 2006 02:42, Dave Jones wrote:
> > > > Quite a few UP x86-64 laptops print APIC error 40's repeatedly
> > > > when they run an SMP kernel (And Fedora doesn't ship a UP x86-64 kernel
> > > > any more). We can suppress this as there's not really anything we
> > > > can do about them.
> > >
> > > No we need to fix the APIC errors, not hide them.
> >
> > What do you need to fix them ? I've got one laptop here that
> > is affected, and there's a few other examples with dmesg's
> > in Red Hat bugzilla that I can trawl.
>
> Some pattern analysis would be useful. All the same chipset, revision?
>From first impression, it seems they're all (including mine) HP laptops
with ATI chipsets.
A quick google seems to confirm this.
http://www.google.com/search?&q=HP+%22apic+error%22
I wonder if this is related at all to the 'time goes double speed'
bug that some folks see (incidentally, I don't on mine).
> Best you collect boot logs.
I'll try to gather some more data.
Dave
On Friday 24 February 2006 03:27, Dave Jones wrote:
> From first impression, it seems they're all (including mine) HP laptops
> with ATI chipsets.
Ah that chipset from hell.
> I wonder if this is related at all to the 'time goes double speed'
> bug that some folks see (incidentally, I don't on mine).
>
Very likely is.
-Andi