jwz@teak.berkeley.edu (Jamie Zawinski) (03/10/90)
My machine has developed a nasty new affliction. Every now and then, a process called "Nubus Receiver #xF0" wakes up and thrashing away without- interrupts. The mouse tracks, but nothing else is alive. The "network" status light (at least I think that's what it is, the one that's about two inches to the right of the run light) flickers like mad. After a while, the lights go away, but the machine is still dead (except for the mouse). When I warm boot, the only change is that the mouse stops tracking, and about a minute later, it crashes (flipped video). Warm booting after that crashes again immediately. The shutdown reasons are either "Sequence break not allowed" or "A data mismatch has been located in the LEFT VIRTUAL MEMORY MAP RAM M-1: Expected Data M-2: Received Data M-T: Virtual Address UPC-1: PC of the original call to ILLOP" This has happened three times today, and once last week. It might be related to heavy consing, but I'm not certain. I'm running Release 6, Explorer II ucode 429. Extended self-tests all pass. Any ideas? Is some piece of hardware on nubus freaking out? Or do I have a virus? :-)
pf@tan.csc.ti.com (Paul Fuqua) (03/11/90)
In article <114093@ti-csl.csc.ti.com> you write:
My machine has developed a nasty new affliction. Every now and then, a
process called "Nubus Receiver #xF0" wakes up and thrashing away without-
interrupts. The mouse tracks, but nothing else is alive. The "network"
status light (at least I think that's what it is, the one that's about two
inches to the right of the run light) flickers like mad.
That's the paging light. The dashed one on the left is the gc light, the one
under the whostate is the run light, and the one on the right is the paging
light.
The "Nubus Receiver #xF0" process is the one handling the Ethernet board in
slot 0. When I've seen similar trouble, there's some broken machine
elsewhere on the local net that is generating excessive network traffic.
(However, I don't try to warm-boot, so I haven't seen any crashing problems.)
That can suck up 50% to 75% of an Explorer 2, so it probably drowns an
Explorer 1 even if it's not without-interrupts.
I have some crude (very crude) Explorer network monitoring code if you're
interested. Alternatively, call your network management people while the
problem is happening and let them use whatever they have.
Another possibility is a hardware problem in the Ethernet board. I don't
suspect software because I'm running pretty much the same system, but I've
been wrong before.
--
Paul Fuqua pf@csc.ti.com
{smu,texsun,cs.utexas.edu,rice}!ti-csl!pf
Texas Instruments Computer Science Center
PO Box 655474 MS 238, Dallas, Texas 75265