I've been running my bp6 stably under 2.4.2 with the NJ bios practically
since the 2.4.2 kernel came out.
Last night I upgraded to the RU bios and ran stably for about an hour before
shutting it down for the night. (I shut it down every night.)
This morning when I booted it up I started getting kernel oops and panics.
I must have rebooted the thing half a dozen times, sometimes getting thru
the boot process and into X, only to die in the wm. Sometimes I didn't
get thru the boot process.
I'd been running dual 366MHz Celerons overclocked to 500MHz. 5.5 mpy, 92MHz,
1:1 AGP.
Cut back to 366MHz, reseated memories and disc cables, etc, etc; nothing
fixed the problem.
As a last gasp measure before ripping the thing apart, I booted 2.4.0. I
haven't had an error since.
It's been running stably now for about half an hour, which is about 100x
longer than it's run under 2.4.2 this morning.
I also notice that the APIC errors, which I started seeing when I installed
2.4.2, are now gone.
I've poked thru the digests for the past couple of months, and see a few
references to APIC errors but none to the RU bios or to problems with 2.4.2
(although I see possibly relevant references to Alan Cox's 2.4.2 patches.)
Is there any evidence that 2.4.2 and APIC interact in some unfortunate way,
especially with the RU bios?
Again, the *only* thing that's changed is that I upgraded the bios.
Suggestions welcomed.
Jim Snyder
[EMAIL PROTECTED]
PS The one other possibility that occurs to me is that I've been running vmware
under 2.4.2 with reserved memory of 128M. This is a 256M system. I don't launch
vmware during boot, but the drivers get installed, so perhaps the drivers soak
up the 128M right away. (I doubt it, but dunno how I'd check.) If that were
the case, then if I've developed memory errors in one bank of RAM (2x128M),
perhaps the reason 2.4.0 is running sans problems is that nothing's touching
the bad memory (because the vmware driver doesn't run under 2.4.0). Doesn't
seem likely, but I thought I'd mention it.
--
=- To unsubscribe, email [EMAIL PROTECTED] with the -=
=- body of "unsubscribe linux-abit". -=
