On Wed, 5 Aug 2020 09:39:17 -0700 Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx> wrote: > On Wed, Aug 5, 2020 at 8:44 AM Marc Plumb <lkml.mplumb@xxxxxxxxx> wrote: > > > > I thought net_rand_state was assumed to be insecure and that anyone > > could determine the internal state. Isn't this Working as Designed? > > I was working as designed - because it wasn't really designed to be > "real crypto" - but sadly it's also the only thing that is fast enough > for a lot of networking. > > So it may be _designed_ to be "not real crypto" and to have a > discoverable internal state. But once again, reality interferes, and > it turns out that people really want something very very fast that is > also not deterministic enough to be discoverable at least remotely. If you turn on the wayback machine, the original net_random came out of having ok values for network simulation with netem. In that use case, the point was to just have something with good distribution and reasonably long period. Then the idea of TCP ports have to be random came along and that was both a weak security feature and benchmark sensitive so the net_random got drafted into more areas.
