[Cryptech Tech] USB interface

[Peter Stuge]

Rob Austein wrote:
> 2) I don't yet really understand what the data transfer between the
>    SADM11 and the STM32 would look like with your proposal.
..
>    Does software on the STM32 still just treat it as a fast UART?

The STM32 doesn't have to change and can continue using the UART and
the RPC protocol.

This is the premise Linus and I use in development with Alpha+daugtherboard2.


For a new Cryptech design, the STM32 interface *could* change. I think
I remember that Pavel and I looked up that our Alpha STM32 UART can do
4Mbps. A different electrical interface, say SPI, would approach 10x that,
perhaps making some new symmetric crypto streaming use cases viable.

But changing from UART to anything else on STM32 requires even more software.


However, it would also be possible to wire both UART and e.g. SPI up,
and only ever use UART to begin with. That works with no software change.


>    Need to understand this to make any kind of informed decision about
>    security properties, so please forgive if this is a dumb question
>    or one that you think you already answered.

Thanks for asking. I should have been more explicit.


> 1) As you note, somebody would have to write software.  We're on a
>    very tight budget at the moment, with fairly aggressive deadlines
>    (Joachim can speak to this better than I), so I'm concerned that if
>    we jump off this particular cliff we must be able to follow through
>    to completion and produce a usable board on schedule.

Since the STM32 doesn't need to change, and an initial firmware version
could just pass bytes between USB and UART, at least there is only a
single, single-purpose software to write. (muxd would change too, but
that's a matter of a few hours.)

Ideally I'd like to have such an initial firmware version working already
by the time we must decide if we jump off the cliff. Do we know when that is?


Many thanks

//Peter