[ltp] tpm chip

[Ulrich Kuehn]

On Wed, 2005-03-02 at 20:43 +0100, Michael Gaber wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Ulrich Kuehn wrote:
> | On Wed, 2005-03-02 at 17:46 +0100, Michael Gaber wrote:
> |
> |>as far as i know this chip implements a AES-256 encryption.
> |>it would be nice to use this chip for encrypting the whole hard-drive on
> |>the fly, cause it would really decrease cpu-time needed for a crypto-fs
> |>and also should be faster.
> |>
> |>is there a way to use this chip for that matter?
> |
> |
> | Hi Michael,
> |
> | No, there is no AES in it. Instead, it implements the version 1.1b of
> | the Trusted Computing Group's spec, see www.trustedcomputinggroup.org.
> |
> | Actually, the functions are quite complicated, so I am not going into
> | details here...
> |
> | Ulrich
> 
> sorry, after the reading of the linux-journals page i see my mistake.
> 
> but it should be possible to use it for encrypting a crypto-fs.
> 
The chip to too slow to be helpfull in mass-encryption. But it can
potentially help keeping the key secret.


> actually i'm having some problems inserting the tpm-module i downloaded
> from the ibm-page
> 

Yes, its a change in the return value of pci_register_driver() in the
kernel starting with 2.6.10. In the source

static int __init
init_tpm(void)
{       
----->   if (!pci_register_driver(&tpm_pci_driver)) {
                pci_unregister_driver(&tpm_pci_driver);
                return -ENODEV;
        }
     .....


change the line to 

      if (pci_register_driver(&tpm_pci_driver) < 0) {

At least that allowed me to run the driver again.

You might also try to the tpmdd driver on sourceforge. google should
give you the correct url.

Ulrich