Ron,
What's the 440 status? Could you bring one up to SC 2001 with a serial
console? (I can use someone's laptop.) These problems look hauntingly
familiar.
It really looks to me like you are misnumbering your IO-APIC and your
Local (on-CPU) APIC for CPU #0.. It should find CPU #1 (APIC-ID=1) as the
boot processor, CPU #0 (APIC-ID=0) as the application processor, and the
IO-APIC (APIC-ID=2). It looks like you have it renumber the App. Proc. to
have ID #2, and hence all sorts of terrible things happen (page faults
aren't handled by the interrupt controller correctly because the CPU
doesn't understand IO-APIC commands).
I had this same problem last Winter -- David H. actually solved it by
asking why my numbering started at #1 and #2 after I told him it should be
#0, #1. The initial numbering is, I believe, motherboard revision
dependent, so it would make sense if your board and Eric's act
differently.
Then again, I could be (and probably am) wrong. I'll see you and talk to
you this week at ALS.
Until then,
James
> Unable to handle kernel paging request at virtual address ffffe0b0
> with one cpu in you get:
...
> Asserting INIT.
> Waiting for send to finish...
> APIC error on CPU0: 00(04)
> APIC error on CPU0: 04(04)
