Comment 488 for bug 1690085

The "freezing while idle" problem does appear to be (at least) worked around by the "Typical Current Idle" option. If that counts as a fix, then the bug is fixed -- or rather, was never a Kernel Bug.

The "Random Soft Lockup" -- the nominal subject of the bug -- on the other hand... who can tell ?

It would be churlish not to be happy that my machine is now reliable and I am a believer in credit where credit is due.

So, as you say, all credit goes to AMD for their total silence up to and after the release of the work around (or fix) -- except for the advice that some (older) power supplies might not work. I don't really need to know why or how the new option does the trick, but I remain curious whether it has a material effect on power consumption.

But I also give credit to this bug's assignee for their total silence -- in particular:

  * it's not clear whether the "freezing while idle" and "Random Soft Lockup"
    problems are separate, related or the same.

  * hence, the CONFIG_RCU_NOCB_CPU and rcu_nocbs=0-15 voodoo may or may
    not still be recommended ?

    For all I know, with the "Typical Current Idle" option, I may be better
    off returning to the default configuration ?

  * and there is talk of kernel changes which reduce power consumption
    (significantly ?) -- which may or may not be related to these issues.

Sadly, the absence of solid information allows half-truths and ill-informed speculation to take its place and to live on in the undamped echo chamber of the interweb :-( [I acknowledge my small part in that noise :-(]