>
> > > We may be able to get round this by detecting the error when we try to init
> > > the second CPU. There are frightening warning messages about using APIC error
> > > returns though. Alan - could you comment on this, please?
> >
> > That is the right answer. The Spec doesnt actually say you cant report CPU's
> > you dont have - it sort of implies it and hopes vendors arent acutely
> > lazy..
>
> Well, tell me if I'm wrong, but it sounds like we've decided that in our cases
> (Tyan motherboards being suspect) the kernel thinks there are two CPUs when
> there aren't. Is our error message reporting a failure to start the second
> CPU, or are we seeing only symptoms of not being able to start the other CPU?
> Does anyone have any evidence that a Tyan motherboard with both CPU slots
> populated *does* correctly populate this data structure, or does the kernel
> just end up making a correct assumption that there are two?
Last time i tried SMP on a tyan tomcat III that only had one cpu
it reported one cpu in bois msg (with the ram amount) the kernel reported
two cpu's and there apic versions and when it tried to boot the 2nd cpu it
said error no 2nd cpu and went on to boot and work fine reporting one cpu.
Is the report at the very start of the kernel msg's about having 2
cpu's and their apic versions gathered from the bois or what? If so them i
guess both my M/B dont report things to the kernel correctly. However both
work fine with 1 and 2 cpu's.
Cheers Adam
>
> Mostly rhetorical questions, mind you. I'll be turning on SMP_DEBUG tonight
> in hopes of getting more detailed information...
>
> Rob
>
> --
> Rob Fugina, robf@fergcons.com
> Unix Systems Administrator/Consultant
>