On Wed, May 4, 2022 at 11:47 AM Milan Broz <gmazyland@xxxxxxxxx> wrote: > BTW we use exactly the same code from Mikulas in cryptsetup now (actually the report > was initiated from here :) and I added some tests for this code, > you can probably adapt it (we just use generic wrapper around it): I use something pretty similar in wireguard-tools: https://git.zx2c4.com/wireguard-tools/tree/src/encoding.c#n74 The code is fine. This is looking like a different issue somewhere else in the OpenRISC stack...