Re: [Anima] IDevID and TPM devices and draft-richardson-anima-masa-considerations-02

[Michael Richardson <[e-mail]>]

Laurence Lundblade <[e-mail]> wrote:
    > I assume you are referring to slide titled "ECDSA key setup based on
    > 256-bit secret seed”, the first of the three.

Yes, I want to discuss that, as well as the other two slides.

    > I think it is possible to have this partly done by the chip vendor and
    > partly by the board vendor, but that’s not what I’d envision. It seems
    > more complex and a lot less secure because of the transfer of the
    > millions of secret keys from the chip vendor to the board vendor. I
    > think there are business reasons too. I don’t know that chip vendors
    > would share the private keys like this.

Then, what's the point of your slide?
Are CHIP vendors going to run a database of IDevID certificates for each of
their OEMs?  How will they know what serial numbers and what extensions to
put into the certificates?

    > In such a document as this, I’d just describe it as if the board vendor
    > did it all, or the chip vendor did it all to be simpler.

I don't really want to simplify the process, I want to point to real
instances of the process being used by real people.

    > Note that some chips have extra fuses that the board vendor can use to
    > put keys into. Or, they might just put the keys in a TEE app. That will
    > be secure enough for many use cases. The TEE app has to use the key
    > anyway to signing tokens or such.

That still seems to skip a major step: how does the OEM get the TEE app
loaded?   If this can only be done by $1B/year OEMs, then we are doomed.

    > With ECDSA you don’t use a PRNG. The secret seed is just the point on
    > the curve that is the private key. There is no computation.

That's not the case.
The point on the curve is a secret, and it needs to be a random point.
To say that there is no computation is inaccurate:  true, with RSA you have to
test if the key is prime, but once you have it, you have it.
Making it (an RSA private key) deterministically found from a secret seed
takes a bit of care to get all the tests in the right order, etc. but it can
be done.

    > I doubt any of this works for a TPM and you probably don’t even want to
    > to it because a TPM already has an attestation key. That’s part of what
    > you pay so much for.

I thought the TPM had an attestation key, but none of them that I have
purchased seem to.   The OEM has to configure/generate them.  It is this
process that I care about, because the IDevID also needs to be built at that
point.

--
]               Never tell me the odds!                 | ipv6 mesh networks [
]   Michael Richardson, Sandelman Software Works        |    IoT architect   [
]     [e-mail]  http://www.sandelman.ca/        |   ruby on rails    [


--
Michael Richardson <[e-mail]>, Sandelman Software Works
 -= IPv6 IoT consulting =-