Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 11 Nov 1998 06:40:01 -0800 (PST)
From:      Peter Wemm <peter@netplex.com.au>
To:        freebsd-bugs@FreeBSD.ORG
Subject:   Re: bin/8646: Implement rlogind -a option 
Message-ID:  <199811111440.GAA20509@freefall.freebsd.org>

next in thread | raw e-mail | index | archive | help
The following reply was made to PR bin/8646; it has been noted by GNATS.

From: Peter Wemm <peter@netplex.com.au>
To: Cy Schubert - ITSD Open Systems Group <cschuber@uumail.gov.bc.ca>
Cc: FreeBSD-gnats-submit@FreeBSD.ORG
Subject: Re: bin/8646: Implement rlogind -a option 
Date: Wed, 11 Nov 1998 22:36:47 +0800

 Cy Schubert - ITSD Open Systems Group wrote:
 > In message <199811110504.NAA08334@spinner.netplex.com.au>, Peter Wemm 
 > writes:
 > > Cy Schubert wrote:
 > > 
 > > > >Synopsis:       Implement rlogind -a option
 > > 
 > > > >Description:
 > > > 
 > > > 	Implement rshd's -a option in rlogind.  Hopefully this will
 > > > 	provide a little better security.
 > > 
 > > I'm not sure that this is the right thing..  What is it to protect? 
 > > Hostname spoofing for .rhosts?  If so, that is already taken care of 
 > > within the ruserok() and iruserok() code in libc which deals with .rhosts.
 > > 
 > > All that I can see that it does is verify the hostname for utmp purposes.. 
 > > What it should do in this case is log the IP address instead of the 
 > > hostname if there is a mismatch, and let ruserok() decide what to do.  
 > > 
 > > There is no need to refuse a connection from an incorrectly configured 
 > > client if that client has it's IP address (not hostname) explicitly listed 
 > > in the .rhosts file.
 > > 
 > > Refusing service solely because of DNS problems is bad.  Refusing to 
 > > *trust* DNS if there is a problem is much better.  The logging should 
 > > switch to IP addresses if there is any doubt about the DNS integrity.
 >
 > I looked through the ruserok() and iruserok() code.  You are indeed 
 > correct.  I stand corrected.  I suppose that the -a option in rshd 
 > should also be deprecated.  Wietse Venema's comments about BSD-style 
 > "paranoid" checks in BUGTRAQ are false.
 
 Wietse's comments could well be correct for older BSD code, I don't have 
 it to check, but I was pretty sure I'd seen it somewhere in there.
 
 The -a checks do have some purpose, but are not implemented the way I'd 
 like.  They are used for logging the hostname in log files and the utmp/
 wtmp.
 
 What I would prefer:
  - the checks are on permanently
  - a mismatch is not a fatal error, but instead of using the hostname, we 
 use the IP address in ascii form in the logs and utmp/wtmp.
 
 ie: if we get a connection from 1.2.3.4 and "1.2.3.4 -> foo.bar.org", but
 "foo.bar.org -> 2.3.4.5", then rather than dropping the connection, we 
 just use "1.2.3.4" in the logs and the utmp files.  And this should be on 
 always. :-)  After all, there is little point in refusing a connection if 
 the user's .rhosts has explicitly got:  "1.2.3.4 otheruser" in it.
 
 Cheers,
 -Peter
 
 

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



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199811111440.GAA20509>