Path: utzoo!attcan!uunet!samsung!zaphod.mps.ohio-state.edu!ub!boulder!hartzell
From: hartzell@boulder.colorado.edu (George Hartzell)
Newsgroups: comp.sys.mips
Subject: Have you had problems with ...
Message-ID: <27420@boulder.Colorado.EDU>
Date: 2 Oct 90 19:43:59 GMT
Sender: news@boulder.Colorado.EDU
Reply-To: hartzell@boulder.colorado.edu (George Hartzell)
Organization: University of Colorado, Boulder
Lines: 67


****************Background

I have been having problems with the Integrated Solutions serial
boards that MIPS uses(used) in the M-2000.  

Under RISC/os 4.0x we saw:
 "cptimer: dropped tx_intr on unit cpx line x"
poping up every once and a while.  We were never able to correlate it
with anything.

When we upgraded to RISC/os 4.5****beta**** we started seeing panics:
   "PANIC: assertion failed: (((int)cs + POFFMASK) + len - 1) < NBPP,
   file cp.c, line 535"
***Since this was beta software we considered this part of the deal.

When we upgraded to RISC/o 4.50FCS the panic message became:
   "PANIC: assertion failed: (((int)cs + POFFMASK) + len - 1) < NBPP,
   file cp.c, line 533"

MIPS looked at it a while, and informed us that it was a problem with
the driver for the serial boards.  At about that time we started
noticing that the crashes correlated with sending long postscript jobs
to our printers (NEC LC-890, Apple IINT).

MIPS gave us a patched driver; now the machine is able to recover from
whatever was panicing it before, but has the unfortunate side effect
of dumping part of the stream going to the printer.  At best this
results in blown print jobs, at worst it hangs the printer and someone
with root needs to use lpc to abort and start it.  We run 2, 3, and 7,
and our cable is too long on one of the printers (but we get errors
from a very close printer too...).  We originally had both printers on
one (of our three) serial boards, the problems have continued when we
moved each printer to its own serial board (and didn't use the first
one in case it had a problem).

On a related note, we have never had very good luck with our dial-in
modems and our inbound connections with an asynch. network (ISN) that
also depends on modem control.  Originally I (and MIPS) blamed the
modem flakyness on our "signalman" hayes clones, and we blamed the ISN
flakiness on the ISN.  I recently bought US Robotics 2400 baud modems,
and cabled them to MIPS' specs and still have them hang every other
day or so.  They will answer the phone, but the M-2000 getty never
seems to notice.  We get similar behavior on the ISN, with more flames
since it is the primary access method for our biochemistry subculture.


************** The question (finally)

I would like to hear about people's experiences (good and bad) with
serial I/O on their MIPS systems (in general and on the M-2000 in
particular).  Has anyone else seen this panic or had problems with
long jobs to postscript printers?  Am I the only person with this
problem (maybe it's me???), or are others affected

Does anyone reliably use US Robotics 2400 baud modems on their MIPS?
How did you do it?  I'd love to hear about your cables, switch
settings, getty setup, etc...

Can anyone confirm/deny that MIPS does/does not run RISC/os on the
internal machine that they use as a modem server?

g.

George Hartzell			                  (303) 492-4535
 MCD Biology, University of Colorado-Boulder, Boulder, CO 80309
hartzell@Boulder.Colorado.EDU           ..!ncar!boulder!hartzell