Re: [PATCH 0/4] HiSilicon hip08 uncore PMU events additions

From: John Garry <john.garry@huawei.com>
To: Arnaldo Carvalho de Melo <arnaldo.melo@gmail.com>
Cc: <peterz@infradead.org>, <mingo@redhat.com>,
	<alexander.shishkin@linux.intel.com>, <jolsa@redhat.com>,
	<namhyung@kernel.org>, <linux-kernel@vger.kernel.org>,
	<linuxarm@huawei.com>, <will@kernel.org>, <mark.rutland@arm.com>,
	<zhangshaokun@hisilicon.com>, James Clark <James.Clark@arm.com>
Subject: Re: [PATCH 0/4] HiSilicon hip08 uncore PMU events additions
Date: Fri, 4 Oct 2019 17:30:46 +0100	[thread overview]
Message-ID: <0492ebd9-f867-423d-034c-9fe1c74902e7@huawei.com> (raw)
In-Reply-To: <20191004151853.GC17687@kernel.org>

>
>>>>>> The missing events were originally mentioned in
>>>>>> https://lkml.org/lkml/2019/6/14/645, when upstreaming the JSONs initially.
>
>>>>>> It also includes a fix for a DDRC eventname.
>
>>>>> Could I get these JSON updates picked up please? Maybe they were missed
>>>>> earlier. Let me know if I should re-post.
>
>>>> Looking at them now.
>
>>> It would be really good if somehow we managed to have someone from the
>>> ARM community to check and provide a Reviewed-by for those, i.e. someone
>>> else than the poster to look at it and check that its ok, would that be
>>> possible?
>
>> For this specific case, I'm not sure how much traction or value there would
>> be since we're just adding some missing events for custom IP.
>
> Someone else inside your organization?

For this, sure. Colleague Shaokun (cc'ed) provided me the metadata for 
these JSON additions, so when he returns from national vacation I can 
ask him to provide necessary tags.

  If nobody is available, then so
> be it, let that be clear in the JSON file (see below) and then I
> wouldn't be waiting for acks/reviewed-by messages for such cases.
>
>> But I do agree that more review of JSONs from the community is required - as
>> I brought up here regarding a recent addition:
>> https://lore.kernel.org/lkml/749a0b8e-2bfd-28f6-b34d-dc72ef3d3a74@huawei.com/
>>
>> Can we enforce that at least linux-arm-kernel@lists.infradead.org and/or
>> get_maintainer.pl results is cc'ed on anything ARM specific as a start?
>
> I think this should be the case, would you be willing to add a note to
> that effect at the top of the JSON files?

Adding notes to JSONs would be painful unless the parser is updated to 
to filter them out. And, as I understand, the x86 JSONs are 
autogenerated, so that tooling would need to handle this also.

>
> And an extra note at tools/perf/pmu-events/README telling users to look
> at the json files to figure out what Reviewed-by tags are required for
> something to get merged? One extra Reviewed-by would be ok?Who would be
> the reviewers for each arch? Would that be at the top of the JSON file?

There is no per-arch JSON, and, in addition, I think that would be hard 
to formulate such formal rules.

As an alternative, how about just add a maintainers entry for reviewers 
per arch? As a start, I don't mind being added there for arm64:

--- a/MAINTAINERS
+++ b/MAINTAINERS
@@ -12767,6 +12767,10 @@ F:     arch/*/events/*
  F:     arch/*/events/*/*
  F:     tools/perf/

+PERFORMANCE EVENTS SUBSYSTEM ARM64 PMU EVENTS
+R:     John Garry <john.garry@huawei.com>
+F:     tools/perf/pmu-events/arch/arm64
+

Patches per-arch should have some nod/tag from a member of the 
respective list. Or at very least be cc'ed :)

Thanks,
John

>
> - Arnaldo
>
> .
>


