Re: [Rats] draft-richardson-rats-usecases-00 comments

[Laurence Lundblade <[e-mail]>]

> On Jun 20, 2019, at 5:52 AM, Shawn Willden <[e-mail]> wrote:
> 
> Thank you. It was a really interesting read.
> What I learnt is that Android mostly attests to the location of a device
> signing key.
> 
> This is correct.  Originally, keystore attestation was intended *only* to prove that a key (a) lived in a thought-to-be-secure environment and (b) that the key had a certain set of properties.  The scope has expanded a bit to also attest to some bits of information about the device state, at least as it was at boot time.

A few other things that seem useful and may be implemented in some cases:
That the key use is gated by user verification like a fingerprint match. This is immensely valuable IMO.
That there are timer throttles that govern the rate of key use.
Knowing which application (e.g. Android app) requested the attestation.
Allowing the application to include custom data in the attestation (it must be marked as as less secure than other data in the attestation). 
Inventory of some or or all of the SW on the device.

A lot of these are useful as risk signals or cross-checks for payments and such.

LL