Re: Wanted: Secure-delete utility for Linux

Jason Burrell (jburrell@crl.com)
Tue, 22 Dec 1998 10:40:40 -0600


--YZ5djTAD1cGYuMQK
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: quoted-printable

On Tue, Dec 22, 1998 at 10:36:19AM +0100, Steffen Grunewald wrote:

> Why urandom ? The safest way to destroy all information on disk should
> be to reverse _all_ magnetic particles (flip and flip back)=20
> |>=20
> |> AFAIK DoD military grade wiping is zero->random->zero->random->zero.
>=20
> A better scheme would be 10101010 - 01010101 - 11111111 - 00000000
>=20
> BUT:
>=20
> if I write a bit pattern to the disk - how is it stored ?
> Remember, CD data are re-encoded to avoid long 1 or 0 patterns.
> I'd guess, with fuzzy bit stuffing nowadays harddisk use similar
> techniques... so perhaps a "all ones / all zeroes" pattern is
> re-encoded to something that is near ...01010101010...
>=20

The kernel, filesystem, and wiper code all have to support this
properly. Forgive my ignorance on the below; I've never really looked
into the lower levels of FS code.

Correct me if I'm wrong, but if I write a file, then shrink it,=20
ext2 doesn't know where the space went, so a secure deletion utility=20
can't overwrite it. If we're talking about files written by text=20
editors, utilities like patch, and other such things, the application=20
doing the shrinking doesn't know to wipe the data before shrinking.=20

Is this a case for the (not working, AFAIK) "s" ext2fs file attribute?
It could probably be modified to overwrite the file with whatever
patterns are deemed best.=20

--YZ5djTAD1cGYuMQK
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: PGPfreeware 5.0i for non-commercial use
MessageID: WUGAkmaQ0MHFfF9od533oyIz3O4kJR1I

iQA/AwUBNn+9dest32QKzeTmEQL2kgCfR+lmEAwoanypcyvm+6oR66caj2UAoOlo
KAaO3JjGnnPRoEzIxL/+6pW8
=CxKk
-----END PGP SIGNATURE-----

--YZ5djTAD1cGYuMQK--

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/