Re: Software leaves encryption keys, passwords lying around in memory

[pgut001 () cs auckland ac nz (Peter Gutmann)]

Dan Kaminsky <dan () doxpara com> writes:

> Yes, but here you *hope* the compiler has the same semantics for "volatile"
> that you do.  The "keys to the kingdom"(sufficient context to zap your
> memset) are left in place; you just hope the compiler bothers to ignore it.
> I'd rather *know*, at least at the same level of confidence I have that I
> know anything else about the compiler.

This is what makes it such a tough problem, and why it may need compiler-level
assistance.  While I was looking for the version of gcc which removes the
memset() (it appears to be a 3.x-only thing, but I can't get to the machine
with 3.x on it at the moment) I noticed that every version of gcc I tried
produced different output for the test source code.  You really can't rely on
a kludge which just happens to work for one version of the compiler (and you
have to be careful when reporting a "problem" which only affects one version
of the compiler :-).

Peter.