Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!mnetor!seismo!rutgers!clyde!cuae2!ihnp4!uniq!rjnoe
From: rjnoe@uniq.UUCP (Roger J. Noe)
Newsgroups: news.software.b
Subject: Re: 2.11 vnews core dumps when reply/followup tried
Message-ID: <211@uniq.UUCP>
Date: Thu, 4-Dec-86 12:39:18 EST
Article-I.D.: uniq.211
Posted: Thu Dec  4 12:39:18 1986
Date-Received: Fri, 5-Dec-86 06:30:46 EST
Organization: Uniq Digital Technologies, Batavia, IL
Lines: 22
Keywords: 2.11 vnews 64K crunch

> Environment is a PDP-11/44 running UNIX System V Release 2.0.  When trying
> the 'r' or 'f' commands in vnews, I get the error message "Memory fault -
> core dump".

What was happening is that our little 64K of data space was filling up whenever
vnews had to exec any other program and so a segmentation violation would kill
the process and dump core.  This is for the shell escape (!), replies (r, R,
and M-r) and follow-up (f).

By the time you input one of these commands, the stack of functions is like
main() -> readr() -> vcmd() -> docmd().  If it's '!' or 'f', then shcmd()
will be called next, otherwise reply() or direct_reply(), as the case may be.
All of these call prun() which, if it's a non-background job, will first
attempt to call botscreen(), which calls _amove(), which eventually calls
_hmove().  It's at this point that my machine bombs.  I think things would
probably be OK if the botscreen() and signal setting for non-background jobs
were done back in docmd() rather than waiting until you're way down in prun().
(By the way, these functions are in visual.c and virtterm.c.)  For now, I think
I'll just use readnews for replies and followups.  What a disappointment!
	Roger Noe			ihnp4!uniq!rjnoe
	Uniq Digital Technologies	(312) 879-1566
	Batavia, Illinois  60510	41:51:10 N.  88:18:25 W.