Re: Future of buffer overflows ?

["Bluefish (P.Magnusson)" <11a () GMX NET>]

>       Non-excutable stacks on x86 have been discussed for years.  They
> may be a hack and they may not be common BUT THEY ARE NOT A NEW CONCEPT!
> Discussions of non-executable stacks predate Linux (I worked on MicroPort
> Unix, SCO Unix, and Xenix).  They are nothing new.  They just don't buy
> you anything.  I use to think they contributed to security but a few
> simple working illustration dispersed that illusion real quick.

I'd say they're an important feature ; a basic principle of hardening a
system is to not allow anything which isn't used. Making constants truely
constant, disabling exec on writable areas etc.

It may not be enough, we will see a raise in BO which doesn't rely on
aribitery code but merely changes of executation order (returns to glibc
exec, or jumping within the program itself). But many vulnerabilities will
become much harder to exploit, and in some cases all the attacker can do
is to crash the applications. It's a good start. Then you add stackgaurd
or propolice.

No action alone solves everything. But when mainstream systems becomes
hardened in that way... when mayor Linux distribution, *BSD & NT are
shipped with more than one layor of hardening against insertion of
aribitery code, the rules of the game has been severly changed.

..:::::::::::::::::::::::::::::::::::::::::::::::::..
     http://www.11a.nu || http://bluefish.11a.nu
    eleventh alliance development & security team

             http://www.eff.org/cafe