Re: [PATCH] perf record/arm-spe: Override attr->sample_period for non-libpfm4 events

From: Ian Rogers <irogers@google.com>
To: German Gomez <german.gomez@arm.com>
Cc: James Clark <james.clark@arm.com>,
	linux-kernel@vger.kernel.org, linux-perf-users@vger.kernel.org,
	Chase Conklin <chase.conklin@arm.com>,
	Mark Rutland <mark.rutland@arm.com>,
	Alexander Shishkin <alexander.shishkin@linux.intel.com>,
	Jiri Olsa <jolsa@redhat.com>, Namhyung Kim <namhyung@kernel.org>,
	Martin KaFai Lau <kafai@fb.com>, Song Liu <songliubraving@fb.com>,
	Yonghong Song <yhs@fb.com>,
	John Fastabend <john.fastabend@gmail.com>,
	KP Singh <kpsingh@kernel.org>,
	Arnaldo Carvalho de Melo <acme@redhat.com>,
	Stephane Eranian <eranian@google.com>,
	netdev@vger.kernel.org, bpf@vger.kernel.org,
	"acme@kernel.org" <acme@kernel.org>
Subject: Re: [PATCH] perf record/arm-spe: Override attr->sample_period for non-libpfm4 events
Date: Mon, 17 Jan 2022 08:28:07 -0800	[thread overview]
Message-ID: <CAP-5=fUHT29Z8Y5pMdTWK4mLKAXrNTtC5RBpet6UsAy4TLDfDw@mail.gmail.com> (raw)
In-Reply-To: <35a4f70f-d7ef-6e3c-dc79-aa09d87f0271@arm.com>

On Mon, Jan 17, 2022 at 2:27 AM German Gomez <german.gomez@arm.com> wrote:
>
> Hi James,
>
> On 17/01/2022 09:59, James Clark wrote:
> >
> > On 14/01/2022 21:21, German Gomez wrote:
> >> A previous commit preventing attr->sample_period values from being
> >> overridden in pfm events changed a related behaviour in arm_spe.
> >>
> >> Before this patch:
> >> perf record -c 10000 -e arm_spe_0// -- sleep 1
> >>
> >> Would not yield an SPE event with period=10000, because the arm-spe code
> > Just to clarify, this seems like it should say "Would yield", not "Would not yield",
> > as in it was previously working?
>
> "this patch" refers to the patch I'm sending, not the one it's fixing.
> I might have to rewrite this to make it more clear. How about:
>
> ===
> A previous patch preventing "attr->sample_period" values from being
> overridden in pfm events changed a related behaviour in arm-spe.
>
> Before said patch:
> perf record -c 10000 -e arm_spe_0// -- sleep 1
>
> Would yield an SPE event with period=10000. After the patch, the period
> in "-c 10000" was being ignored because the arm-spe code initializes
> sample_period to a non-zero value.
>
> This patch restores the previous behaviour for non-libpfm4 events.
> ===

Thanks for fixing this, I can add an acked-by for the v2 patch. Could
we add a test for this to avoid future regressions? There are similar
tests for frequency like:
https://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git/tree/tools/perf/tests/attr/test-record-freq
based on the attr.py test:
https://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git/tree/tools/perf/tests/attr.py
The test specifies a base type of event attribute and then what is
modified by the test. It takes a little to get your head around but
having a test for this would be a welcome addition.

Thanks!
Ian

> Thanks for the review,
> German