From owner-freebsd-bugs  Thu Oct  5 18:56:22 1995
Return-Path: owner-bugs
Received: (from root@localhost)
          by freefall.freebsd.org (8.6.12/8.6.6) id SAA03086
          for bugs-outgoing; Thu, 5 Oct 1995 18:56:22 -0700
Received: from dca.net (dca.net [204.183.80.2])
          by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id SAA03081
          for <bugs@freebsd.org>; Thu, 5 Oct 1995 18:56:19 -0700
Received: from dca.net (localhost.dca.net [127.0.0.1]) by dca.net (8.6.12/8.6.12) with SMTP id VAA09296; Thu, 5 Oct 1995 21:56:15 -0400
Date: Thu, 5 Oct 1995 21:56:13 -0400 (EDT)
From: Andrew White <awhite@dca.net>
To: Stefan Esser <se@zpr.uni-koeln.de>
cc: bugs@freebsd.org
Subject: Re: bug with gcc 2.6.2?
In-Reply-To: <199510051000.AA00067@Sysiphos>
Message-ID: <Pine.BSF.3.91.951005215533.9169F-100000@dca.net>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Sender: owner-bugs@freebsd.org
Precedence: bulk

Thanks for your response.  Of course you are right, a null string and a 
null pointer are not the same.

However, it would be nice if the str* functions would not SIGSEV when 
called with a null pointer, but rather failed more nicely.

-aw

On Thu, 5 Oct 1995, Stefan Esser wrote:

> On Oct 5,  2:42, Andrew White wrote:
> } Subject: bug with gcc 2.6.2?
> } FreeBSD bugs list:
> } 
> } The following code generates a segmentation violation using gcc 2.6.2 as 
> } shipped with FreeBSD 2.0-RELEASE.  It shouldn't!  It seems as if any of 
> } the string functions (strcmp, strcpy, etc) bomb when invoked with a null 
> } string.
> 
> This is not a compiler error, but intended behaviour ...
> 
> NULL is not a string, but a pointer to NO string.
> 
> FreeBSD on purpose makes address 0 unavailable, just to
> catch such errors (as just about any other Unix system).
> 
> } I compiled this using gcc 2.3.3 under AIX 3.2.5, and it works as expected 
> } (prints "x is -1" when run).  I am compiling this program (I called it 
> } test.c) with: "gcc -o test test.c"... am I missing something obvious??
> 
> }    s=NULL;
> }  
> }    x=strcmp(s,"test");
> 
> Yes. The test result is undefined, and the system lets
> you know. If you want to use a NULL pointer to imply an
> empty string, then you better code this as:
> 
>      x=strcmp(s ? s : "", "test");
> 
> assuming that the second parameter in fact is not a 
> constant string, but variable and known to not be 
> another NULL pointer ...
> 
> Regards, STefan
> 
> -- 
>  Stefan Esser, Zentrum fuer Paralleles Rechnen		Tel:	+49 221 4706021
>  Universitaet zu Koeln, Weyertal 80, 50931 Koeln	FAX:	+49 221 4705160
>  ==============================================================================
>  http://www.zpr.uni-koeln.de/staff/esser/esser.html	  <se@ZPR.Uni-Koeln.DE>
> 

awhite@dca.net    http://www.dca.net/       (work-related)
Andrew@White.org  http://andrew.white.org/  (personal)
DCANET: The Delaware Common Access Network 
Call (302) 654-1019 or send mail to info@dca.net for more information.