Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: $Revision: 1.6.2.16 $; site inmet.UUCP
Path: utzoo!linus!philabs!cmcl2!seismo!harvard!think!inmet!keenan
From: keenan@inmet.UUCP
Newsgroups: net.sources.bugs
Subject: Re: Orphaned Response
Message-ID: <25600003@inmet.UUCP>
Date: Tue, 28-May-85 18:19:00 EDT
Article-I.D.: inmet.25600003
Posted: Tue May 28 18:19:00 1985
Date-Received: Mon, 3-Jun-85 00:41:58 EDT
References: <1545@reed.UUCP>
Lines: 38
Nf-ID: #R:reed:-154500:inmet:25600003:177600:1226
Nf-From: inmet!keenan    May 28 18:19:00 1985



I ran into the same problems you did (long name and core dump with
patch on BSD2.9).  Its scarry having a core dump in a 1700 line program
you know nothing about. With lots of print statements, I tracked it
down. Seems that the 2.9 compiler has problems with longs being
autodecremented within an array reference. The decrement takes place as
if it was an integer.  Putting the decrement on the next line seems to
fix the problem.  Here is a context diff (that you could apply if patch
worked!) for the core dump problem.  The problem that I found was in
the routine another_hunk. Let me know of others.

*** oldpatch.c	Sun May 26 18:14:49 1985
--- patch.c	Sun May 26 21:03:19 1985
***************
*** 1342,1348
      int context = 0;
  
      while (p_end >= 0) {
! 	free(p_line[p_end--]);
      }
      assert(p_end == -1);
  

--- 1342,1349 -----
      int context = 0;
  
      while (p_end >= 0) {
! 	free(p_line[p_end]);		/* Changed from postdecrement */
! 	p_end--;			/* by Keenan Ross for BSD2.9  */
      }
      assert(p_end == -1);
  

--keenan ross		UUCP:     {ihnp4,harpo}!inmet!keenan
 Intermetrics, Inc.	INTERNET: ima!inmet!keenan@CCA-UNIX.ARPA
 733 Concord Ave.
 Cambridge, MA  02138	PHONE:    (617) 661-1840