[RFC 0/3] um/random: reproducible random

From: Johannes Berg

Date: Thu Nov 19 2026 - 08:24:05 EST


So these patches are *clearly* (except the first one maybe)
not something we actually want to apply upstream, but we've
been carrying them for years out-of-tree (no problem there)
and Lingao just asked because they have the same need when
testing BT (we did it for testing WiFi):
https://lore.kernel.org/linux-um/D2219FFB-73EF-4E27-B84A-5C8E550E3553@xxxxxxxxxx/

The idea is to even get things like key derivation into a
reproducible state for tests, to make it easier to compare
and be sure the runs are the same.

So at least I figured I'd send them, maybe we _could_ consider
doing something like that behind a scary Kconfig option or one
that depends already on time-travel in UML, or ...

But even if this is going to fly like a lead balloon then the
patches are now out there for whoever wants to pick them up
and run with them when they have the same need.

johannes