[net.unix-wizards] Translation Buffer Parity faults on 11/750

Jskud@sri-unix (09/04/82)

I haven't seen all the related articles.  But I would like to say
that the cp tbug par fault is somewhat of misnomer, since it is, according
to a photocopy of the fiche listings (provided by DEC on request), also
indicative of cache parity errors and (non-existent) memory timeouts.
I have seen software generate this fault (due to bad memory references),
and suggest that you check the machine check summary register (mcsr)
to determine if bit 3 is on.  If so, you've got non-existent memory,
not a parity fault, which could be a hardware fault, rumored to be caused
by a vendor's chips (I don't (yet) know who) failing spec.

/Jskud

lee.usc-cse@Udel-Relay@sri-unix (09/11/82)

Date:         1 Sep 1982 12:04-PDT
Yes, we have had those same problems, and I have also heard that others
have.  However, I haven't learned much about how to avoid them.  We did
find that they diminished in number for some time after we fixed a
memory board that was getting frequent soft errors, but I can't draw a
connection in logic between those two phenomena.
						-- Lee