Re: [RESEND PATCH] perf/x86/intel: Fix unchecked MSR access error for Alder Lake N

["Liang, Kan" <kan.liang@linux.intel.com>]

On 2022-08-22 9:55 a.m., Peter Zijlstra wrote:
> On Mon, Aug 22, 2022 at 09:24:57AM -0400, Liang, Kan wrote:
>> I think the only issue should be the PMU name. The non-hybrid PMU name
>> is "cpu". The hybrid PMU name is "cpu_$coretype". If we move the
>> ALDERLAKE_N to the non-hybrid PMU, the PMU name will be changed from
>> "cpu_atom" to "cpu". It will be different from the rest of
>> {ALDER,RAPTOP}LAKE.
>>
>> Also, I think we have to update the perf tool for the events because of
>> the PMU name change.
>>
>> But I guess it should be OK, since the ALDERLAKE_N was just added and we
>> know its an Atom-only system.
> 
> cpu/caps/pmu_name should be 'Gracemont', which is exactly like all the
> other !hybrid setups. Surely perf-tools already knows about this
> pattern.
> 
> IOW, if you need to change perf-tools for this, someone did something
> wrong somewhere.

The event list for ADL and RPL is different from the non-hybrid
platforms. We combine the events from big core and small core into a
single file and use the PMU name to distinguish from them. The PMU name
is either cpu_core or cpu_atom.

If we change the ADL-N to non-hybrid, the simplest way is to create a
dedicate gracemont event list. Or we have to specially handle the ADL-N
in the parsing codes. We have to update the tool for either way.

> 
> (also, I just noticed, 'Tremont' is the *only* PMU that has a
> capitalized name, perhaps we don't want Gracemont to follow but instead
> fix tremont if that is still possible)

I don't think the tool rely on the name under cpu/caps/pmu_name.
The event list rely on the CPU model number.
It should be OK to fix the tremont name.


Thanks,
Kan