Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site pur-phy.UUCP
Path: utzoo!linus!security!genrad!grkermit!masscomp!clyde!floyd!harpo!ihnp4!inuxc!pur-ee!CS-Mordred!Pucc-H:Physics:crl
From: Pucc-H:Physics:crl@CS-Mordred.UUCP
Newsgroups: net.bugs.4bsd,net.bugs.2bsd
Subject: bug in sendmail thaw routine on 11's
Message-ID: <1154@pur-phy.UUCP>
Date: Fri, 20-Jan-84 11:07:39 EST
Article-I.D.: pur-phy.1154
Posted: Fri Jan 20 11:07:39 1984
Date-Received: Sat, 21-Jan-84 07:31:09 EST
Organization: Purdue University Physics Dept.
Lines: 14

We recently brought up sendmail (from a 4.2 BSD system) on a pdp-11 running
2.8 BSD.  There is a problem in the thaw() routine.  On Vaxen, the global
variable environ is in initialized data, and in 2.8 BSD, it is in BSS.
Sendmail's frozen configuration file is implemented by writing out from
&edata (end of initialized data) to the end of non-stack space.  Therefore,
environ is overwritten on thawing.  This lead to some users not being able
to send mail because the evironment passed through execv() was garbage
and an "argument list too big" sys error was returned.  I fixed it by 
saving environ at the beginning of thaw() in main.c and restoring it
after the second read().

Charles LaBrec
UUCP:		pur-ee!Physics:crl, purdue!Physics:crl
INTERNET:	crl @ pur-phy.UUCP