Messages in this thread | | | Date | Thu, 22 May 2014 14:13:28 +0200 | From | Borislav Petkov <> | Subject | Re: [PATCH] intel_rapl: Correct hotplug correction |
| |
On Thu, May 22, 2014 at 05:24:33PM +0530, Srivatsa S. Bhat wrote: > That was just me trying to explain the current mess, not justifying > it! :-/
Yes, it is a mess - thanks for explaining it.
> I think Oleg had a proposed patch to use per-cpu rwsem in CPU hotplug to > drastically simplify this whole locking scheme. I think we could look at > that again.
And that is my question: why can't all be made to use a single dumb lock allowing only one task and lock everything hotplug with it?
Maybe it is an oversimplification but why do I care about hotplug operations scaling - they're not on the fastpath anyway.
And yes, we're trying to remove CPU_POST_DEAD - I have one user in MCE which I'm testing a removal patch for - and then we can all use get_/put_online_cpus() like we used to do and be happy.
Having 2 + 1 aliased hotplug sync APIs is beyond insane and is simply not needed IMHO.
-- Regards/Gruss, Boris.
Sent from a fat crate under my desk. Formatting is fine. --
| |