Re: Hard disk Encryption

[Alexander Klimov <alserkli () inbox ru>]

On Sun, 15 Apr 2007, Ali, Saqib wrote:
> > The main problem is how you store (or derive) the key. For
> > example, if you use TPM to store the key, it will be easily
> > extracted by an attacker with hardware tools for bus access.
>
> Can you please elaborate on this point? I was under the impression
> that in a properly implemented system, the "TPM wrapped and bound
> encryption keys" have NO need to leave the TPM. So I am not sure how
> tapping in into the bus will reveal the keys.

Since TPM does not have enough processing power to decrypt the
raw data, the key (used to decrypt the data) has to be
transmitted outside of the TPM.

Let us try to make assumptions that have to be true even in "a
properly implemented system":

 * CPU does not have any secrets (at least immediately after
   a reboot)

 * after communication over an attacker-accessible bus with some
   external entities (TPM, hard disk, etc.) CPU computes the key
   needed to decrypt the data

Due to the first assumption, the CPU can be emulated by an
attacker so that external entities cannot distinguish it from
the "normal" CPU; due to the second assumption, after such an
emulation the attacker may have exactly the same secret key.

-- 
Regards,
ASK