In message <26650.1022702207@eterna.com.au>matthew green writes
[...]

>
>
>would someone please explain to me why our _kernel_ random() needs
>to worry about POSIX or being reproducible (in general) ?

Verification that the in-kernel security subsystem or cryptosystem
(or other tool using random(9)) really does what it's supposed to do.
By reverse-engineering the behaviour and comparing to behaviour
on the (known, reproducible)  random(9) stream.

And for debugging, when it doesn't match.