On Thu, Jun 16, 2016 at 11:12:18PM +0200, Borislav Petkov wrote:
> On Thu, Jun 16, 2016 at 05:00:04PM -0400, Vince Weaver wrote:
> >
> > According to the BKDG the AMD Family16h Model30h "Jaguar Mullins"
> > also supports the accumulated power interface. I've tested on
> > hardware I have and with this patch I indeed get power readings using
> > perf.
> >
> > Signed-off-by: Vince Weaver <vincent.wea...@maine.edu>
> >
> > diff --git a/arch/x86/events/amd/power.c b/arch/x86/events/amd/power.c
> > index 55a3529..778b77d 100644
> > --- a/arch/x86/events/amd/power.c
> > +++ b/arch/x86/events/amd/power.c
> > @@ -292,6 +292,7 @@ static struct notifier_block power_cpu_notifier_nb = {
> >
> > static const struct x86_cpu_id cpu_match[] = {
> > { .vendor = X86_VENDOR_AMD, .family = 0x15 },
> > + { .vendor = X86_VENDOR_AMD, .family = 0x16 },
> > {},
>
> Actually, I think we remove that table completely and rely solely on the
> CPUID bit:
>
> if (!boot_cpu_has(X86_FEATURE_ACC_POWER))
> return -ENODEV;
>
> Rui?
>
Agree with you. If the some chips are not stable, we can add a check
to ignore them with family and model id.
Thanks,
Rui
