Re: [PATCH v5] perf/x86/amd/power: Add AMD accumulated power reporting mechanism

[Huang Rui]

On Tue, Mar 01, 2016 at 02:58:39PM +0100, Borislav Petkov wrote:
> On Mon, Feb 29, 2016 at 12:30:47PM +0100, Thomas Gleixner wrote:
> > On Tue, 1 Mar 2016, Huang Rui wrote:
> > > On Fri, Feb 26, 2016 at 11:18:28AM +0100, Thomas Gleixner wrote:
> > > > > +static int __init amd_power_pmu_init(void)
> > > > > +{
> > > > > +	int i, ret;
> > > > > +	u64 tmp;
> > > > > +
> > > > > +	if (!x86_match_cpu(cpu_match))
> > > > > +		return 0;
> > > > > +
> > > > > +	if (!boot_cpu_has(X86_FEATURE_ACC_POWER))
> > > > > +		return -ENODEV;
> > > > > +
> > > > > +	cores_per_cu = amd_get_cores_per_cu();
> > > > > +	cu_num = boot_cpu_data.x86_max_cores / cores_per_cu;
> > > > 
> > > > Please use the new package management functions which are on the way to tip.
> > > > 
> > > Can you give me some hints?
> > 
> > http://git.kernel.org/tip/1f12e32f4cd5243ae46d8b933181be0d022c6793
> > 
> > topology_max_packages() is what you want.
> 
> Rui, I think you want to use smp_num_siblings here. And we did define
> that amd_get_cores_per_cu() but that's redundant:
> 
>                 /* get compute unit information */
>                 smp_num_siblings = ((ebx >> 8) & 3) + 1;
>                 c->compute_unit_id = ebx & 0xff;
>                 cores_per_cu += ((ebx >> 8) & 3);
> 
> So if I'm not missing anything else, we don't need that cores_per_cu
> variable, nor do we need amd_get_cores_per_cu(). I'll zap this patch
> from my queue. You can use smp_num_siblings in your patch instead which
> is exported to everything.
> 

That makes sense for me. Thanks, Boris, Thomas. We just want to know
cores number per compute unit here.

Thanks,
Rui