full-disclosure - Coming soon: CPU fix for buffer overflows

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <Pine.LNX.4.21.0402232013481.22009-100000@kcisp2>
From: mikehome at kcisp.net (Mike Barushok)
Subject: Coming soon: CPU fix for buffer overflows

On Mon, 23 Feb 2004, hybriz wrote:

> >
> >Subject: RE: [Full-Disclosure] Coming soon: CPU fix for buffer overflows
> >   From: "Richard M. Smith" <rms@...puterbytesman.com>
> >   Date: Mon, 23 Feb 2004 15:39:10 -0500
> >     To: <hybriz@...o-security.com>
> >
> >Let's get to the bottom line.  Would this page execution bit scheme stop
> >stuff like the Blaster worm?
> >
> >Richard 
> 
> IMHO,
> if the page-wise non-exec stack was implemented in win2k>= during the
> blaster period the blaster worm as we know it obviously would not exist.
> thing is, the timeline substitute would use diferent exploiting techniq
> to have the same effect. non-exec stack doesnt stop ALL buffer overflow
> attacks/techniqs, win2k+3 has a stack protection and it has been proven
> to be bypassable. The blaster worm wouldnt exist as we know it, an
> analogous substitute would.
> 
> the execution bit exists on other archs but it doesnt mean that
> exploitation of stack based overflows isnt possible, it's just slightly
> (IMHO) more difficult and there are less possible attack vectors
> (for example, the ret-into-libc techniq will fail if the binary is
> stacticly linked).
> 
> btw, in my country brought better 'security' to overall networks and home
> users since many started using pseudo-well configured by default firewalls
> and in a way that wouldnt happen if the stupid worm didnt have broken
> shellcode and 'non-universal' offsets.
> 
> regards,
> hybriz
> 

If a system built with a particulare new CPU cannot be made to
perfectly emulate any other 'general purpose computer' that is
equivalent to a Turing machine, then the new system (with the
new CPU) is no longer a general purpose computer.

There is also the mathematical theorem by Kurt Godel (anglicized
spelling, apologies in advance) that implies that no amount of
adding axioms to a sufficiently general and consistent system of
Logic makes the unproveablility of all false theorems possible.

So, the easiest way to reconcile the apparent fallacy is to deduce
that adding an 'execute' flag will only make the next generation
of buffer overflows a little more difficult.

(Note to the non-mathematically inclined: Yes, it might seem
difficult to believe, but a 4004 processor combined with
unlimited (countable) storage, can emulate the fastest super
computer, just not in 'real time').