ALM-2 hang

[gihill at storm.lakeheadu.ca]

Hello SUN users,

I would like to ask the advice of all experienced SUN users about an major
problem we are having with our new ALM-2 board.  To put it briefly, none
of the serial ports have worked properly since we installed it over a
month ago, and our student users are naturally getting restless.  Customer
Support at SUN Canada know about our problem, but despite a lot of their
time and best efforts, no solution has been found so far.

The details are as follows:

Configuration:

The ALM-2 is installed in our SUN 3/160 running OS v4.0-3.  This was
replacing an ALM-1 which worked fine. In fact the ALM-2 is using the same
backplane slot as the ALM-1.

We have some other SUN machines connected via ethernet.

The ALM-2 is configured as device mcp0 (same as in the generic config
file), with terminal devices /dev/ttyh*.

The ports are connected to a mixture of hard-wired terminals and modem
lines. (the exact same ones that were connected to the ALM-1)

Symptoms:

At seemingly random periods (anywhere from 1 hour to 3 days after boot),
*ALL* the serial ports on the ALM-2 hang simultaneously and become
unusable.  Users who are currently on the system are locked out.  New
users trying to login just get the prompt:

                        'machine' login:

and nothing more. i.e. no echoing, no password prompt.

The system itself is however still usable.  The console is OK, and users
can do remote loggins over the ethernet from other machines.

No error messages appear on the operators console or in the system log
files.

Workaround:     Re-boot the system.

What we have tried:

Replace the ALM-2 with a new one from SUN - same problem

Put the ALM-2 on our other SUN machine, a 3/180 - same problem

Re-seat the board in the backplane, check for damaged pins, etc..

Install the latest kernel patches for the mcp driver for bugs found since
version 4.0-3.   - same problem.

Any clues, hints or whatever would be greatly appreciated.