Sparcstation panic while reading /dev/ttyb

[Robert Carey]

We have a Sun which crashes regularly while reading a 9600 baud feed
connected to /dev/ttyb. The machine is a Sparcstation 1 with 32
megabytes of memory.   At first I thought that the problem might be
specifically related to the program which reads and parses the feed.
Since then I have found that it can be crashed just as reliably by
running "/bin/cat /dev/ttyb" and waiting for about 2 minutes.
(We have also run "cat /dev/ttyb > /dev/null" with the same effect,
if that makes any difference.)

The only unusual thing about the system is that the console has been
replaced with a Wyse terminal (not my idea) in order to make it fit
better on a shelf of consoles.

Before it crashes I typically see one or more of the following messages on
the console:

     zs1: silo overflow
     xmballoc: out of mblks
     BAD TRAP
     panic: Data fault

Other times when I cat /dev/ttyb the system simply hangs.

Has anyone seen anything like this?  Any idea what is happening?

                                Robert Carey
                                Susquehanna Investment Group
                                carey at eniac.seas.upenn.edu
                                carey at deadlock.uucp