From owner-freebsd-arch  Tue Jan 23  2:18:38 2001
Delivered-To: freebsd-arch@freebsd.org
Received: from ringworld.nanolink.com (ringworld.nanolink.com [195.24.48.189])
	by hub.freebsd.org (Postfix) with SMTP id DE29D37B402
	for <freebsd-arch@FreeBSD.org>; Tue, 23 Jan 2001 02:18:06 -0800 (PST)
Received: (qmail 3427 invoked by uid 1000); 23 Jan 2001 10:16:37 -0000
Date: Tue, 23 Jan 2001 12:16:37 +0200
From: Peter Pentchev <roam@orbitel.bg>
To: Garrett Wollman <wollman@khavrinen.lcs.mit.edu>
Cc: freebsd-bugs@FreeBSD.org, freebsd-arch@FreeBSD.org
Subject: Re: bin/24540: add '-c' flag to src/bin/domainname to clear domainname
Message-ID: <20010123121637.E2376@ringworld.oblivion.bg>
Mail-Followup-To: Garrett Wollman <wollman@khavrinen.lcs.mit.edu>,
	freebsd-bugs@FreeBSD.org, freebsd-arch@FreeBSD.org
References: <200101222010.f0MKA2185498@freefall.freebsd.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
User-Agent: Mutt/1.2.5i
In-Reply-To: <200101222010.f0MKA2185498@freefall.freebsd.org>; from wollman@khavrinen.lcs.mit.edu on Mon, Jan 22, 2001 at 12:10:02PM -0800
Sender: owner-freebsd-arch@FreeBSD.ORG
Precedence: bulk
X-Loop: FreeBSD.ORG

On Mon, Jan 22, 2001 at 12:10:02PM -0800, Garrett Wollman wrote:
> The following reply was made to PR bin/24540; it has been noted by GNATS.
> 
> From: Garrett Wollman <wollman@khavrinen.lcs.mit.edu>
> To: Pete Fritchman <petef@databits.net>
> Cc: freebsd-gnats-submit@FreeBSD.ORG, phk@FreeBSD.ORG
> Subject: Re: bin/24540: add '-c' flag to src/bin/domainname to clear domainname
> Date: Mon, 22 Jan 2001 15:02:57 -0500 (EST)
> 
>  <<On Mon, 22 Jan 2001 11:10:02 -0800 (PST), Pete Fritchman <petef@databits.net> said:
>  
>  >  electron# ./domainname foo.bar
>  >  electron# ./domainname ''
>  >  electron# ./domainname
>  >  foo.bar
>  
>  I think this is a bug in sysctl.  Note the following:
>  
>  root@khavrinen(6)# sysctl -w kern.domainname=''
>  kern.domainname: foo.bar -> foo.bar
>  
>  What is happening is that the sysctl() system call is interpreting the
>  request to set the MIB variable to a zero-length object as if it were
>  an indication that setting is not requested.  Here is an (untested)
>  fix (beware cut&paste has bogotified whitespace):
>  
>  
>  
>  --- kern_sysctl.c       2000/07/28 22:40:04     1.100
>  +++ kern_sysctl.c       2001/01/22 20:01:38
>  @@ -862,7 +862,7 @@
>                  req.oldptr= old;
>          }
>   
>  -       if (newlen) {
>  +       if (new) {
>                  req.newlen = newlen;
>                  req.newptr = new;
>          }
>  @@ -1101,7 +1101,7 @@
>                  req.oldptr= old;
>          }
>   
>  -       if (newlen) {
>  +       if (new) {
>                  if (!useracc(new, req.newlen, VM_PROT_READ))
>                          return (EFAULT);
>                  req.newlen = newlen;
>  
>  -GAWollman

This works for me; still, how about changing the test to

if (newlen || (new != NULL))
	
in both places?  I have a -current system which has been running fine
for three hours now with this fix, and there have been no signs of
breakage (there shouldn't be, unless somewhere someone passes uninit'd
values to new or newval).

-arch CC'd because of the proposed kern_sysctl change :)
 
G'luck,
Peter

-- 
Nostalgia ain't what it used to be.


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-arch" in the body of the message