Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!iuvax!sahayman
From: sahayman@iuvax.cs.indiana.edu (Steve Hayman)
Newsgroups: comp.sys.apollo
Subject: Re: vt100 woes (really rlogin problems) (and a kludge fix)
Keywords: vt100, rlogin, broken, problems
Message-ID: <23016@iuvax.cs.indiana.edu>
Date: 7 Jul 89 04:16:54 GMT
References: <11406@siemens.siemens.com>
Reply-To: sahayman@iuvax.cs.indiana.edu (Steve Hayman)
Organization: Computer Science Department, Indiana University
Lines: 40

>running sr10.1, bsd4.3, csh
>can hang my vt100 window every time by rlogging into our vax and running
>emacs. The vt100 emulator hangs on exit of emacs, as does screen.

We noticed this too.  You can get around most of the problems by using
"rlogin -8" instead of "rlogin", but there are still a few other ways
to make rlogin hang.  If you have the rlogin source (you can ftp it
from uunet.uu.net) you can make a small patch to give you a more
reliable version of rlogin.  The fix is to have the variable "eight"
always set, and throw in a "0" into one if-statement in the oob()
routine to keep the TIOCPKT_FLUSHWRITE code from being executed.

Change

	int eight;
	
to 
	int eight = 1;

and find the line that says:

 	if(mark & TIOCPKT_FLUSHWRITE) {

and change it to

 	if ( 0 & mark & TIOCPKT_FLUSHWRITE ) {

to disable that part of the oob() code.  I'm not sure exactly what
the ramifications of disabling this are, but the stock Apollo rlogin
hangs in the 'for' loop just after this line, waiting to read
data that's never going to arrive.  We made the above
fix to our rlogin months ago and it hasn't hung since.

Hope this is of some use to someone.

..Steve

-- 
Steve Hayman    Workstation Manager    Computer Science Department   Indiana U.
sahayman@iuvax.cs.indiana.edu                                   (812) 855-6984