[ltp] using fingerprint reader for encryption or ssh login?

[Jan Kundrát]

This is an OpenPGP/MIME signed message (RFC 2440 and 3156)
--------------enig1011D9C002C4520A760CE064
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

Jiang Qian wrote:
> Thanks a lot JKT for your detailed explanation. I do have a trusted chi=
p=20
> on my thinkpad T43. I'll trust them alright.

An important point to raise here is that if there's any flaw in the TPM
chip, you can't defend against it. If any of the companies producing it
(from chip designers to silicon manufacturers) added a backdoor, you're
screwed.

> How to do hardware crypto using tpm under linux?

No idea except [1] and links from that page. I don't use it myself
(yet?), sorry.

> And what is the advantage of a TPM chip over the following method I'm=20
> considering: I have a usb key with encrypted key files. Only when I nee=
d=20
> to decrypt a file, I plug it in.

Let's assume the TPM chip is really secure and well-designed. (I'm not
sure that you can verify that yourself. If you can, I guess it'd be far
more difficult than verifying a corresponding SW-based solution.) Now
the difference is that the encryption key never leaves your TPM chip.
The TPM generated it and (again, supposing it isn't malfunctioning) it
will never leak it outside of its secure domain.

Now if you want to "use the key", it's probably becasue you want to sign
some data or decrypt them. The application you're using now asks the TPM
like this: "Hello TPM, I've heard you have a key 0x1234, right?" - "Yes,
I do" - "Great, please, use it to sing the following data:
0x33663366..." - "Here you are: 0x99887766...". See, the secret key
never leaves the TPM. When signing data, your "signature key" is the
secret stuff you want to protect.

OTOH if you wanted to use the TPM for stuff like HW encryption, it'd
work like this: "Hi TPM, please decrypt following data that I just read
from the disk: 0x112233..." - "Sure honey, here it is: 0x998877...". (I
guess there'll be some performance issues as well, so it might be worth
to use TPM just for decrypting a real key that you'll use for doing all
crypto stuff with host CPU. This method will protect you only from
someone getting a raw dump of your hard drive, not against someone who
can break your OS.

The important point here is to decide *what* you want to protect -- the
data on the disk, or the key itself? Surely the key is unknown to the OS
so no attacker can steal it, but the OS still sees the real plaintext of
data. There's *nothing* you can do to prevent this.

> I guess in this case a hacker can=20
> syphon off the decrypted key from the memory or just replace the gpg=20
> binary in my system. How is TPM different? Why can't he replace the=20
> binary of whatever program I use to decrypt what's on the TPM chip?

If the question is "can the attacker retrieve the key", then "no, he
can't (provided the TPM chip is safe)". OTOH if you wanted to ask "can
he see my data", then the answer is "yes, when he breaks your OS, he
can", and "no if he can read all your hard drives and USB sticks, but
can't break the OS".

[1] http://thinkwiki.org/wiki/Tpm

If anything in this mail is rubbish, sorry for that. I have no extra
cryptography knowledge.

Cheers,
-jkt

--=20
cd /local/pub && more beer > /dev/mouth


--------------enig1011D9C002C4520A760CE064
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: OpenPGP digital signature
Content-Disposition: attachment; filename="signature.asc"

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFHQ3IZamXfqERyJRcRAlZcAKCH3tLnadhWRvbTZhmcn04vTGsUawCgkatB
GWE78yEbP4O2CDtG77a2y/k=
=7M0M
-----END PGP SIGNATURE-----

--------------enig1011D9C002C4520A760CE064--