Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!usc!elroy.jpl.nasa.gov!turnkey!orchard.la.locus.com!elrond.locus.com!ghb
From: ghb@locus.com (George Bray)
Newsgroups: comp.mail.mush
Subject: 'longjmp botch' in 7.0.4 on 4.3Tahoe
Summary: dies if SIGINTR during message composition
Message-ID: <253@elrond.locus.com>
Date: 8 Mar 90 23:06:13 GMT
Reply-To: ghb@elrond.locus.com (George Bray)
Organization: Locus Computing Corp., Los Angeles
Lines: 23

I am using mush 7.0.4 on a VAX running 4.3Tahoe from Berkeley.  This
machine is *terribly* overloaded, so is a good place to find race
conditions.

If I start to reply to a message, and decide I don't really want to,
and hit ^C, I sometimes get "longjmp botch" and a core dump.  I don't
have a non-stripped binary around, so I don't have any useful info
from the core dump (if the problem seems unreproduceable, let me know
and I'll do some debugging).  The problem mostly occurs during the
*second* ^C (the first one prints "another interrupt to terminate" (or
words to that effect)).

I have also seen a longjmp botch if I ^Z (suspend) mush when it isn't
at a prompt (like while it's updating a folder) and then resume it.

Both of these problems happen quite intermittently, and seem related
to system load.

-- 
George Bray
Locus Computing Corporation, 9800 La Cienega Blvd, Inglewood, CA  90301-4440
213-337-5171
lcc!ghb@seas.ucla.edu		       {uunet,ucla-se,elroy!turnkey}!lcc!ghb