[Dshield] The end of buffer overrun vulns?

Ruigrok , Jeroen jeroen_ruigrok at epson-europe.com
Fri Jan 9 14:00:45 GMT 2004


I am actually wondering if it can be that simple to do so.

The NX idea seems to be akin to W^X from the OpenBSD folks.
I wonder if it fixes both the heap and stack exploit vector.
>From what I read it does that for heaps, stacks, and memory
pools.

What concerns me more is whether or not this is a step up
to going to the whole DRM/trusted computing nightmare.

Yes, it might help, but in my opinion this is taking a
hardware shortcut to avoid problems which come from careless
programming.

I mean, most CPUs already have the ability to work in
multiple rings with scaling privileges.  Also most CPUs
nowadays have the ability to mark memory pages with
attributes (read, write, execute).  But they almost never
get used.

It's great in some aspects, but I also have to place my
question marks since it seems the hardware people are
once again saving the proverbial ass of Microsoft (they
have an extremely bad reputation due to ease of which
their OS and executable format can be abused).
They could've started using canaries in their executable
format and compilers a long while ago.

-- 
Jeroen Ruigrok <jeroen_ruigrok at epson-europe.com>




More information about the list mailing list