Re: 50 Watt idle power regression bisected to Linux-3.10

[Mike Galbraith <bitbucket@online.de>]

On Sat, 2013-12-07 at 11:45 -0500, Len Brown wrote: 
> >> It fixes that, except for my Q6600 box.  Too bad mwait_idle() went away,
> >> beloved old box doesn't play hints game, so it continues to flog itself.
> 
> 
> Thanks for pointing this out, Mike!
> 
> A Q6600 is a Kentsfield.  I dug one of those up.
> Indeed, the only idle capabilities it has are  HALT
> and old style MWAIT, and the latter is much more effective.
> running 3.8 it idles at 75 watts.
> running 3.8 with idle=nomwait it idles at 100 watts,
> which is what it will do with 3.9 and later due to the patch below.
> 
> commit 69fb3676df3329a7142803bb3502fa59dc0db2e3
> Author: Len Brown <len.brown@intel.com>
> Date:   Sun Feb 10 01:38:39 2013 -0500
> 
>     x86 idle: remove mwait_idle() and "idle=mwait" cmdline param
> 
> Kentsfield proves that patch was based on a fault assumption.
> Sweet box in its day, ECC memory and everything -- probably still
> a fair number of them running...
> 
> Plus, I've found another machine that depends on having an idle=mwait
> idle loop (A Sony Vaio BIOS SMM code apparently assumes we use it in
> https://bugzilla.kernel.org/show_bug.cgi?id=60770)
> 
> So it looks like I need to (also) restore the simple idle=mwait idle loop
> to make some machines happy.

Cool, box will definitely be happier.

-Mike