[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <200702150649.l1F6nKkP014649@caligula.anu.edu.au>
Date: Thu, 15 Feb 2007 17:49:20 +1100 (Australia/ACT)
From: Darren Reed <avalon@...igula.anu.edu.au>
To: Thierry@...ler.lu
Cc: bugtraq@...urityfocus.com
Subject: Re: Re[2]: Solaris telnet vulnberability - how many on your network?
In some mail from Thierry Zoller, sie said:
>
> CDSC> real back doors are better
> I like that tautologie, "real backdoors", what makes a backdoor more
> real than another one ? Is it the coolness, the stealth ? Or is it
> simply the fact that it gives back door access ?
How about putting a backdoor into your C compiler such that it
generates "special code" when it recognises it is compiling
/bin/login that allows special access?
That doesn't show up in any code audit of /bin/login...
so you think about auditting the code that makes up the compiler..
where does the executable for that come from...
and so on back.
If someone made this kind of programming error 10 or 15 years ago,
nobody would have screamed "backdoor" - and i suspect the only person
(or people) who are doing this now are those with relatively little
experience in computer security.
What makes a "backdoor" a "backdoor" is the intent of the hole.
A backdoor is generally a change made to an existing program and
then all evidence of the change hidden and things made to look like
they are normal when they're not. it's something done after you
have gained access you shouldn't have to keep that access when
someone else thinks they've locked you out.
So, to round this out, any security hole you find in opensolaris
or openbsd or freebsd or netbsd or linux or debian or suse is in
general not a back door.
Darren
Powered by blists - more mailing lists