Hi Eric, Andy, > On Tue, Sep 11, 2018 at 5:01 PM Andy Lutomirski <luto@xxxxxxxxxxxxxx> wrote: > > I think Ard’s point is valid: in the long run we don’t want two competing software implementations of each primitive. It clearly *should* be possible to make crypto API call into zinc for synchronous software operations, but a demonstration of how this actually works and that there isn’t some change to zinc to make it would well would be in order, I think. > > > > IMO the right approach is do one conversion right away and save the rest for later. It turned out to be fairly straight forward and quick to do. I'm on a transatlantic plane so the connection is super spotty, but you can check out the three commits added to: <https://git.kernel.org/pub/scm/linux/kernel/git/zx2c4/linux.git/log?h=jd/wireguard> -- the two "crypto:" commits and the one "random:" commit. Alternatively, you can look directly at <https://git.kernel.org/pub/scm/linux/kernel/git/zx2c4/linux.git/tree/crypto/poly1305_zinc.c?h=jd/wireguard> and <https://git.kernel.org/pub/scm/linux/kernel/git/zx2c4/linux.git/tree/crypto/chacha20_zinc.c?h=jd/wireguard>. I'll be refining and checking these over the next week or so, but the basics were super easy to do. Jason