bin/100443 : awk(1) dies with SIGBUS when processing INDEX-6

[Ruslan Ermilov]

The following reply was made to PR bin/100443; it has been noted by GNATS.

From: Ruslan Ermilov <ru at FreeBSD.org>
To: "Devon H. O'Dell" <devon.odell at coyotepoint.com>,
        Dmitry Marakasov <amdmi3 at mail.ru>
Cc: bug-followup at FreeBSD.org
Subject: Re: bin/100443 : awk(1) dies with SIGBUS when processing INDEX-6
Date: Fri, 15 Sep 2006 14:14:17 +0400

 On Thu, Sep 14, 2006 at 07:15:00PM -0400, Devon H. O'Dell wrote:
 > Ruslan Ermilov wrote:
 > >On Thu, Sep 14, 2006 at 09:40:24PM +0000, Devon H. O'Dell wrote:
 > [snip]
 > >> Do we need to send this upstream?
 > >> 
 > >Yes, of course.
 > 
 > I sent an email to BWK a couple hours ago -- hopefully he'll get it soon :)
 > 
 There's another function, refldbld(), that needs a similar fix.
 Actually, your patch makes it so that if "n == fieldssize" then
 malloc() will be redundantly called, when we enter this function
 again and again.  I think the following should fix it better:
 
 %%%
 Index: lib.c
 ===================================================================
 RCS file: /home/ncvs/src/contrib/one-true-awk/lib.c,v
 retrieving revision 1.1.1.4
 diff -u -p -r1.1.1.4 lib.c
 --- lib.c	16 May 2005 19:11:33 -0000	1.1.1.4
 +++ lib.c	15 Sep 2006 10:10:59 -0000
 @@ -58,7 +58,7 @@ static Cell dollar1 = { OCELL, CFLD, NUL
  void recinit(unsigned int n)
  {
  	if ( (record = (char *) malloc(n)) == NULL
 -	  || (fields = (char *) malloc(n)) == NULL
 +	  || (fields = (char *) malloc(n+1)) == NULL
  	  || (fldtab = (Cell **) malloc((nfields+1) * sizeof(Cell *))) == NULL
  	  || (fldtab[0] = (Cell *) malloc(sizeof(Cell))) == NULL )
  		FATAL("out of space for $0 and fields");
 %%%
 
 Could you please send me the code/scripts to reproduce a segfault?
 
 
 Cheers,
 -- 
 Ruslan Ermilov
 ru at FreeBSD.org
 FreeBSD committer