[Cryptech Tech] Noise board on Novena

[Fredrik Thulin]

On Tuesday, September 23, 2014 06:04:58 AM Benedikt Stockebrand wrote:
> Hi Joachim and list,
> 
> Joachim Strömbergson <joachim at secworks.se> writes:
> > [Why SPI is a pain]
> > 
> > That (and a few other details) is what i mean with "don't like". ;-)
> > 
> > I'm spending a lot of time in another project with SPI. The positive
> > thing with SPI is that it is commonly supported by many devices.
> 
> that's the point: I just wanted to make sure this didn't look like an
> unjustified personal dislike, but like a justified one:-)

I agree - SPI has it's downsides. I'll be easily swayed in the matter if you 
can argue why some other alternative is better.

As I see it, the alternatives on this levels for the speeds we are talking 
about are SPI, I2C, UART, and bit-banging.

The appealing thing about SPI, I2C and UART in my mind is that they are well 
supported (often in hardware) among MCUs, and other people could take a 
entropy module speaking one of those, connect it to a Raspberry Pi (for 
example) and start interacting with it in about 10 lines of Python code.

This makes it easier for others to test, develop and audit entropy sources for 
Cryptech HSMs. The entropy sources would also be easier to reuse in other 
applications.

Bit-banging, although preferably with less than 32 parallel data lines, would 
probably be easier if we didn't have to care about other people, but I think 
we should care.

Let me hear your arguments for something else than SPI =)

/Fredrik