Luke Small on Sun, May 15 2022: > The current implementation is nothing more than a naive arc4random() % > upper_bound which trashes initial arc4random() calls it doesn’t like,
Yes, that's exactly what it is. > The whole transformation by modulus of perfectly decent random data > seems so awkward. [...] it seems like an ugly HACK Let me point out that this way of doing it is absolutely standard. It is not awkward or a hack. > to simply meet demand This is probably a good way of describing the design goal for this highly sensitive piece of kernel interna. -p PS: I am not an OpenBSD developer, a tiny contributor at most. I speak as a bystander who happens to have some domain knowledge.
