this patch adds via padlock rng support to pc and pc64. *Warning* I've got no clue about cryptography. I ported https://www.fourmilab.ch/random/ to plan9 here: https://github.com/armeye/ent_random_sequence_tester and wrote a test here: https://github.com/armeye/9front_xstore. the results look right to me...
Tested on my VIA C7 Thinclient, looks good ¯\_(ツ)_/¯

I'm looking for comments about the patch:

* this adds a new line to archctl to show which hwrng is in use, if any
* the via.c is for via c7 specific code, I've got a patch for the cputemp msr in my repository.
  I would like to share the file between architectures but don't know how.
* The xstorebuf function is stupid, but it works. Can this be done better? 
* the xstore instruction is very cpu heavy with the maximum whiteout. This patch has the best value for randomness but
  it eats cpu cycles like it's nothing. I thing we can go down to even the lowest quality because the random data
  gets encrypted anyway, right?

Greetings,
Arne