From: Chad Perrin <perrin@...>
Date: 2010-02-20T11:16:23+09:00
Subject: Re: using secure memory

--LZvS9be/3tNcYl/X
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Sat, Feb 20, 2010 at 09:45:47AM +0900, Ryan Davis wrote:
>=20
> AFAIK, secured virtual memory is a system-wide needs-a-reboot type of
> thing. So other than good doco, I think the answer defaults to "no". It
> is the route I'd go. The software will be as clean as it can be,
> easy(ier) to security audit, etc. All with the caveat, that you need to
> run it in a secured environment. Given the original requirements, that
> just seems common sense/expected to begin with.

I know that one does not need to reboot a computer to have a guarantee
that data in memory will not be swapped out.  For an example, check the
way GnuPG works on BSD Unix and Linux systems if the SUID bit is set,
where it will start with administrative privileges to secure memory
against swapping to disk then drop administrative privileges to run as
the current user.  That way, passwords, private keys, and cleartext data
that is being encrypted or decrypted will not be swapped to disk where it
could be harvested later by a malicious security cracker more easily than
it could from RAM.

Of course, GnuPG is not written in Ruby, which is why I'm not looking at
GnuPG source code for hints on how to do something similar in Ruby.

--=20
Chad Perrin [ original content licensed OWL: http://owl.apotheon.org ]

--LZvS9be/3tNcYl/X
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.14 (FreeBSD)

iEYEARECAAYFAkt/Q+EACgkQ9mn/Pj01uKXcIACeNgP1uknlafXHH8bOzmyHfTRX
axcAnAzgLznLtbF+RZQ6N4QX8TDNUebF
=JwLO
-----END PGP SIGNATURE-----

--LZvS9be/3tNcYl/X--