On Wed, 26 Jul 2000, Scott Balmos wrote:
> Hi again fellow cookers,
>
> I am trying to install Helium onto an HP Vectra XU 6/200 series server,
> equipped with dual Pentium Pro-200/256k cache chips. Both of the
> processors are Stepping 9, but the main processor is a model SL22V, and
> the secondary is model SL22T.
>
Was the machine bundled as is? Looks strange.
> When the system boots, it looks like all is going well, until it hits the
> synchronization part for the two procs. Without trying to hand type what's
> on the screen, I can basically say that it looks like the kernel
> calibrates the first processor fine, detects the second processor and
> calibrates it fine, is able to add together the two BogoMIPS (like we
> care), and then craps out on this setting the IO-APIC IRQ line.
>
> I've had a large amount of experience with computers, but not really with
> multiproc ones. What the heck is IO-APIC? I'm thinking something like
> Automatic InterProcessor Communication?
>
Advanced Programmable Interrupt Controller - an evolution of PCs' PICs. Allows
interrupts to be distributed among processors. No APICs, no SMP.
> Anyway, here are the last few lines before the kernel locks:
>
> enabling symmetric IO mode... ...done.
> ENABLING IO-APIC IRQs
> ...changing IO-APIC physical APIC ID to 16
> Kernel anic: could not set ID
> in swapper task - not syncing
>
> It goes to note, obviously, that I can easily boot into the uniproc
> version of the multiproc kernel, which is 2.2.15-mdk4secure, if it
> requires a recompilation.
>
> Please let me know if this is a kernel problem, a BIOS thing maybe, or the
> fact that the two procs have to be both the same stepping and model? Maybe
> I have to flip them to get the "newer" SL22V in the secondary position?
>
I'd say that yes, the two processors have to be the same. I've no idea about
the differences between the two models.
You can try and compile a vanilla SMP kernel (ie, one from ftp.kernel.org) and
see if you can reproduce the problem.
--
Francis Galiegue, [EMAIL PROTECTED]
"Programming is a race between programmers, who try and make more and more
idiot-proof software, and universe, which produces more and more remarkable
idiots. Until now, universe leads the race" -- R. Cook
