Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site oddjob.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxr!mhuxn!ihnp4!gargoyle!oddjob!matt
From: matt@oddjob.UUCP (Matt Crawford)
Newsgroups: net.bugs.uucp,net.bugs.4bsd
Subject: uucico is core-dumping
Message-ID: <1069@oddjob.UUCP>
Date: Fri, 6-Dec-85 14:41:35 EST
Article-I.D.: oddjob.1069
Posted: Fri Dec  6 14:41:35 1985
Date-Received: Sat, 7-Dec-85 19:49:40 EST
Reply-To: matt@oddjob.UUCP (Matt Crawford)
Followup-To: net.bugs.uucp
Organization: U. Chicago, Astronomy & Astrophysics
Lines: 24
Xref: watmath net.bugs.uucp:627 net.bugs.4bsd:1871

Help!  In the past few weeks our uucico has been dying about
once per day of a SIGSEGV.  It is happening in pk1.c, routine
pkgetpack(), called from pkaccept(), called from pkread().

	static char sccsid[] = "@(#)pk1.c	5.1 (Berkeley) 7/2/83";
	...
	pkgetpack(ipk)
	struct pack *ipk;
	{
	...
			bp = pk->p_ipool;
			pk->p_ipool = (char **) *bp;
	...
	}

The SIGSEGV comes at the last assignment shown, presumably
because bp has the value 0x3d110f00 which, on a vax, does not
point to valid memory.  The memory locations near p_ipool have
not been clobbered.  The pk->p_state is RXMIT, if that helps.
This problem only started recently.  The binary of uucico is not
corrupted and there are no silo overrun messages.  Any ideas?
_____________________________________________________
Matt		University	crawford@anl-mcs.arpa
Crawford	of Chicago	ihnp4!oddjob!matt