[TrouSerS-users] Question on TPM capability

[Arshad Noor]

Hi,

I had a question on whether a specific capability exists on the TPM -
1.2 or 2.0; hopefully, its an easy answer.

Assuming a TPM has been initialized with the following key-hierarchy:

+-------+
| SRK |
+-------+
|
|
+-------------+
| Storage Key |
+-------------+
|
|
+-------------------------------+
| Some symmetric key (AES/TDES) |
+-------------------------------+
|
+---------------+
| |
+-------------------------+ +-------------------------+
| Some encrypted secret-1 | | Some encrypted secret-2 |
+-------------------------+ +-------------------------+


Is there a mechanism to send secrets (1 and 2) into the TPM (with the
encrypted symmetric key) so the encrypted-key and the secrets are
decrypted inside AND the two plaintext secrets compared inside the
TPM with just a boolean result coming out: True for a match, False
otherwise.

Thanks in advance.

Arshad Noor
StrongAuth, Inc.

[Ken Goldman]

Post by Arshad Noor
I had a question on whether a specific capability exists on the TPM -
1.2 or 2.0; hopefully, its an easy answer.

For TPM 1.2, the answer is easy. There is no exposed AES or TDES, or
any symmetric algorithm.

For TPM 2.0, I don't see any direct compare. However, there are some
smart people on the TPM work group. If you're a member, join us and ask.

~~

For your specific case, what does "some encrypted secret" mean? Is it a
loadable object wrapped with "some symmetric key"? Are -1 and -2 in the
clear inside the TPM? Are they encrypted identically, same key and IV?
The question is not clear enough.

My thought is that, e.g., if -1 and -2 are really identical decrypted
secrets inside the TPM, and they can be loaded as HMAC or symmetric
keys, then you can indirectly compare them. HMAC or encrypt some data.
If the results are equal, the keys are equal.