Hallo Mark,
there is one thing, that is some kind of curious:
using 2.4.0-test12pre5 I've many apic errors with CPU1 a n d CPU0:
Dec 7 06:52:04 nmb kernel: APIC error on CPU1: 04(00)
Dec 7 06:52:04 nmb kernel: APIC error on CPU0: 02(00)
Dec 7 06:52:04 nmb kernel: APIC error on CPU0: 04(00)
Dec 7 06:52:07 nmb kernel: APIC error on CPU0: 02(00)
Dec 7 06:52:07 nmb kernel: APIC error on CPU1: 04(00)
running 2.4.0-test11ac4 apic errors only appear with CPU0:
Dec 6 13:41:33 nmb kernel: APIC error on CPU0: 01(00)
Dec 6 13:41:33 nmb kernel: APIC error on CPU0: 02(00)
Dec 6 13:42:07 nmb last message repeated 2 times
Dec 6 13:42:28 nmb last message repeated 10 times
Dec 6 13:42:32 nmb kernel: APIC error on CPU0: 04(00)
Dec 6 13:42:34 nmb kernel: APIC error on CPU0: 02(00)
Dec 6 13:42:39 nmb last message repeated 4 times
Dec 6 13:42:40 nmb kernel: APIC error on CPU0: 08(00)
Dec 6 13:42:49 nmb kernel: APIC error on CPU0: 02(00)
Dec 6 13:42:55 nmb kernel: APIC error on CPU0: 08(00)
Dec 6 13:43:02 nmb kernel: APIC error on CPU0: 02(00)
and n e v e r on CPU1 !
today I will compile 2.4.0-test12pre7 and 'll see....
kind regards
Norbert
On Friday 01 December 2000 08:36, Mark Hahn wrote:
> > > unfortunately, many SMP boards have noisy APIC lines.
> >
> > Hallo Mark,
> >
> > thank you for your immediate reply!
> > Do APIC errors mean => hardware errors? With the actual kernel
>
> they are corrected errors. that is, the apic checksums its messages
> and retries. obviously, if the corruption is bad enough, it'll hang.
>
> > (2.4.0-test12pre3) they apear less often...
>
> there have been tweaks to the kernel code that handles the error
> reports, and there have been changes in how the kernel uses apics.
>
> > Can I do anything to solve this problem besides buying a new board (I do
> > not need a new computer ;o))?
>
> nope. among people who have abit bp6's (very prone to this problem),
> a number of voodoo techniques are used, including >=300W power,
> fan on the BX chipset, bios updates, etc.