🚀 go-pugleaf

In article <0elw=Nm00XQL8ycFls@andrew.cmu.edu> Douglas Allen Luce <dl2p+@andrew.cmu.edu> writes:
>I've got a 4-port Async card that has had 16550's shoved into the 8250
>sockets, and am running modified asy drivers (by Tim Theisen, Peter
>Honeyman mainly) so as to keep my SLIP link running.  I've been getting
>some odd crashes that I can't quite figure out:
>
>Sep 28 08:15:13 telerama vmunix: panic: machine check
>
>I'm a bit baffled by this; from what I can tell, the CPU was pulling in
>stuff from an I/O section while this happened (asy0?), but it crashed. 
>The tech docs say that there's supposed to be a bit set to prevent the
>cpu from dying when an I/O error occurs.
>
>Has anyone seen this before?

Do you have the "spurious machine check" patch to trap.c installed?
I don't have any of this available anymore, but Bill ____ (gad, I forget
his name, but he was one of the AOS RT gurus) who was then of IBM's Palo
Alto division simply put a case statement in the main switch statement of
trap() in trap.c which caused the case of a machine check or machine check
in user mode (when no additional bits were set in the machine check status
register).

My RT used to crash several times a day until this change was made.

-- 
Steve Dyer
dyer@ursa-major.spdcc.com aka {ima,harvard,rayssd,linus,m2c}!spdcc!dyer