instability of timekeeping

[Andriy Gapon]

On 30/10/2015 15:22, Adrian Chadd wrote:
> Hm!
> 
> are you able to come up with a sane way to check whether it does C1E
> the silly way (eg that MSR on that particular platform) so we can
> print/disable it?
> 
> That way noone else has this same problem.

Well, even the silly way can work correctly.  I believe that this particular
BIOS had a bug, but there are many vendors and versions out there.
It might still make sense to print a warning.
It should be as easy as checking bit AMDK8_SMIONCMPHALT in MSR_AMDK8_IPM, see
sys/x86/x86/cpu_machdep.c.

-- 
Andriy Gapon