Path: utzoo!utgpu!news-server.csri.toronto.edu!clyde.concordia.ca!uunet!zaphod.mps.ohio-state.edu!usc!cs.utexas.edu!rice!rice!sun-spots-request
From: dougc@uswat.uswest.com (Doug Corey)
Newsgroups: comp.sys.sun
Subject: Restoring a hung tty driver?
Keywords: Miscellaneous
Message-ID: <1990Nov4.031228.2681@rice.edu>
Date: 4 Nov 90 03:00:00 GMT
Sender: sun-spots-request@rice.edu
Organization: Sun-Spots
Lines: 17
Approved: Sun-Spots@rice.edu
Originator: spots@titan.rice.edu
X-Sun-Spots-Digest: Volume 9, Issue 358, message 14
X-Original-Date: 22 Oct 90 22:56:44 GMT

We are building an application that uses modems attached to serial ports.
The work is on a Sun 4 running the Sun 4.0.3 OS.  No getty is run on these
ports.  In effect, the application takes the place of getty but handles
outgoing calls as well as incoming calls.  Since the application is in the
development stage, it has bugs, one of which (not sure why yet) causes the
port to "hang".  That is the port (tty driver) is no longer communicative
and doesn't respond to ioctl() or stty.  The only way we've found to clear
the problem is to reboot the machine.  "kill -1 1" doesn't work.  I assume
that's because there is no getty or anything else associated with these
ports (the application has been terminated).  If I do "stty </dev/cuaa"
when cuaa is hung (as it is now), the stty doesn't return.  (Curiously, if
I do a ps, I don't see the stty process, just the subshell that spawned
it!  Killing the subshell kills the stty call so I can get the window from
which I did the stty back and I get an "interrupted system call message".)

Can anyone suggest a way to clear the driver/port and return it to some
stable state that doesn't require rebooting the machine?  Thanks.