看板FB_security
标 题Re: md(4) (swap-base) disks not cleaned on creation
发信站NCTU CS FreeBSD Server (Wed Nov 7 23:44:47 2012)
转信站ptt!csnews.cs.nctu!news.cs.nctu!.cs.nctucs.nctu!.org!ownorg!owner-free
--ZjlDoLon7m/fln42
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable
On Wed, Nov 07, 2012 at 02:14:36PM +0100, Paul Schenkeveld wrote:
> On Wed, Nov 07, 2012 at 06:03:46PM +1100, Dewayne Geraghty wrote:
> > An excellent example of where swap shouldn't be used. It isn't the use=
of the swap file that is the issue, it is how the output of
> > using swap is used. PHK was right in his advice to not use swap.
> >=20
> > Good catch, nanobsd.sh should be changed.
>=20
> I tend to disagree. Nanobsd.sh is just an example but there may be more
> uses of swap-based md(4) devices where ultimately swap contents are
> leaked to unprivileged users or processes. Des@ mentioned md(4) devices
> made available to jails where the root inside the jail is definately not
> the same as the root outside the jail.
>=20
> All of us (I hope) have been educated with the wisdom that memory
> returned by malloc() and friends is safe to use which may raise the
> expectation (at least it did to me) that mdconfig'd memory follows the
> same principles of security.
It is reverse, malloc-ed memory is not guaranteed to have any predefined
content. But is content does not cross security boundaries.
--ZjlDoLon7m/fln42
Content-Type: application/pgp-signature
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.19 (FreeBSD)
iEYEARECAAYFAlCaZc8ACgkQC3+MBN1Mb4iuUwCfRMHpeqVcwmRoX3rCGgR0XJHK
MpkAoMd+C6Jd3gIjWxVFMwfu68MoiTPI
=fF/Q
-----END PGP SIGNATURE-----
--ZjlDoLon7m/fln42--