[32676] in bugtraq
Re: [ANNOUNCE] glibc heap protection patch
daemon@ATHENA.MIT.EDU (Stefan Esser)
Thu Dec 4 12:10:36 2003
Message-ID: <3FCF160D.6000003@nopiracy.de>
Date: Thu, 04 Dec 2003 12:10:05 +0100
From: Stefan Esser <se@nopiracy.de>
MIME-Version: 1.0
To: xenophi1e <oliver.lavery@sympatico.ca>
Cc: bugtraq@securityfocus.com
In-Reply-To: <20031203221916.6056.qmail@sf-www1-symnsj.securityfocus.com>
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
xenophi1e wrote:
> This question seems more complex than 'Feel free to demonstrate me an
> unlink exploit that works while my unlink macro is in place'. But I
> have to admit my own ignorance here, I can't say for certain whether
> an attacker who passes the test in your macro is left in a situation
> where an exploit is possible.
Fact is my macro makes arbitrary pointer overwrites with unlink()
impossible. The magic value approach just makes it harder. You need to
guess a 32bit value. Even if this is totally random it is theoreticly
possible to exploit the unlink() macro in that case. And do not forget
the power of information leak exploits.
Just an example: The gamecube was hacked by an information leak exploit.
A crc feature the Phantasy Star Online game allows to request checksums
of arbitrary memory positions (and sizes).
So it was possible for the smart guy who did it, to create a complete
memory dump from
remote. In that case your magic values are worthless...
Stefan Esser
