ports/77488: devel/p5-String-Ediff seg faults

[Anton Berezin]

The following reply was made to PR ports/77488; it has been noted by GNATS.

From: Anton Berezin <tobez at tobez.org>
To: Jonathan Noack <noackjr at alumni.rice.edu>
Cc: FreeBSD-gnats-submit at FreeBSD.org, knu at FreeBSD.org,
	boxzou at yahoo.com
Subject: Re: ports/77488: devel/p5-String-Ediff seg faults
Date: Mon, 14 Feb 2005 11:38:13 +0100

 On Mon, Feb 14, 2005 at 03:54:44AM -0600, Jonathan Noack wrote:
 
 > While tracking down a problem with devel/cvsweb3, I discovered the
 > issue was really that ediff was seg faulting.  The attached test.pl
 > simulates a call to ediff in cvsweb that failed (this is merely one
 > example; I can provide many).  I tested this on two different machines
 > with identical results.  Please let me know if I can provide further
 > info.
 
 > #0  ediff (s1=0x8054ba8 " * $Id$", s2=0x80680a8 " * $Header$") at st.c:747
 > 747       ret[0] = 0;
 > (gdb) l 747
 > 742         ix++;
 > 743         tmp_seg = tmp_seg->m_next;
 > 744       }
 > 745     #define INT_LEN 11
 > 746       ret = (char*)malloc(sizeof(char) * INT_LEN * ix * 8);
 > 747       ret[0] = 0;
 > 748       tmp_seg = equals;
 > 749       while (tmp_seg) {
 > 750         char buff[4 * INT_LEN + 1];
 > 751         if (tmp_seg->m_begin_line_num1 < 0) {
 
 This clearly happens because ix is 0, which happens when equals is NULL.
 There is no protection for this condition in the code.  It is probably
 better to contact the module author, boxzou at yahoo dot com, directly
 (CCed).
 
 >>How-To-Repeat:
 >        Run attached test.pl
 >>Fix:
 >        *shrug*
 >--- test.pl begins here ---
 >#!/usr/bin/perl -wT
 >
 >require String::Ediff;
 >
 >my $left_diff = " * \$Id\$";
 >my $right_diff = " * \$Header\$";
 >my $diff_str = String::Ediff::ediff($left_diff, $right_diff);
 >
 >printf "diff_str = $diff_str";
 >--- test.pl ends here ---
 
 \Anton.
 -- 
 The moronity of the universe is a monotonically increasing function. --
 Jarkko Hietaniemi