[Cryptech Core] dnssec signer

[Rob Austein]

At Wed, 15 Jul 2015 23:58:39 +0200, Peter Stuge wrote:
> 
> Rob Austein wrote:
> > > Do I get it right, that for SHA-256 internal state is defined by eight 
> > > 32-bit words (h0..h7)?
> > 
> > Er, I'm guessing you mean the H registers (H0..H7); "h" is part of a
> > different series of registers (a..h).
> > 
> > I *think* you're correct that the state we care about is the H
> > registers.
> 
> The internal state in the SHA-256 implementation by Tom St Denis is 8
> 32-bit state words, two 32-bit length words (total bits and intermediate
> bytes) and 0..63 bytes of intermediate state (input bytes) where the
> latter are compressed away into the former 8 state words each time a
> 64th input byte becomes available.

Our hash cores only process complete blocks, so the intermediate state
and intermediate bytes probably go away.  Similarly, total bits is
something we track in software because that's where we construct the
final padding.

So this is consistent with Pavel's theory that the core's state
between blocks is just the H registers.