x86-64 box

[Lennart Sorensen]

On Wed, Jun 22, 2005 at 11:51:03PM -0400, Franco Saliola wrote:
> Especially since hyperthreading isn't implemented correctly!

I have yet to see anyone say it could have been implemented in a way
that prevented this "problem".  The P4 HT is not the first system to
have this type of problem as far as I can tell from the discusions on it
so far.

> "Hyper-Threading, as currently implemented on Intel Pentium Extreme
> Edition, Pentium 4, Mobile Pentium 4, and Xeon processors, suffers
> from a serious security flaw. This flaw permits local information
> disclosure, including allowing an unprivileged user to steal an RSA
> private key being used on the same machine. Administrators of
> multi-user systems are strongly advised to take action to disable
> Hyper-Threading immediately; single-user systems (i.e., desktop
> computers) are not affected."
>     http://www.daemonology.net/hyperthreading-considered-harmful

Many people have also looked at that and said it's a load of crap since
you require the system to be in such an amzingly known state before you
can learn anything at all, that it is essentially impossible to exploit
it.  And apparently it isn't new, it's a normal issue of being able to
try and predict what some data was by how quickly it was retrievied by
attacking the cache of the cpu.  Supposedly multicore cpus with shared
cache can be affected too, as can some SMP systems, and I think someone
claimed that it might even be possible if you poked at the cache on a
single cpu system you could learn something if you were scheduled
alternately with the program you wanted to get information from.  Seems
awful complicated to me, and like way to much work.  Then again there
are people with too much time on their hands that will try anything to
break into systems.  In this case you have to actually break into the
system first (or be a valid user) before you could even hope to begin to
learn any crypto key bits.

I think this particular issue was way overhyped.

Lennart Sorensen
--
The Toronto Linux Users Group.      Meetings: http://tlug.ss.org
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://tlug.ss.org/subscribe.shtml