Received-SPF: pass (google.com: domain of linux-kernel+bounces-79226-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.80.249 as permitted sender) client-ip=147.75.80.249;
Precedence: bulk
MIME-Version: 1.0
References: <20240221175210.19936-1-khuey@kylehuey.com> <20240221175210.19936-2-khuey@kylehuey.com>
 <CAP-5=fXsv7TJ_SVOZc38fN0gn+7cWBcMWt3FdVLcs5v0_vO=uw@mail.gmail.com>
 <CAP045AoSHWoOP3TN=6Hf2wZj7X9Y41sThBQWCDZ3BEP68qeTBw@mail.gmail.com>
 <ZdemibZepMqWvv6U@x1> <CAP045Aqc6woHGYKJG4d=x4gPd0=PmrLeFv25Rkts1a8xFuocOQ@mail.gmail.com>
 <ZdjdAKIV39A-jpR5@x1>
In-Reply-To: <ZdjdAKIV39A-jpR5@x1>
Reply-To: robert@ocallahan.org
From: "Robert O'Callahan" <robert@ocallahan.org>
Date: Sat, 24 Feb 2024 10:43:38 +1300
Message-ID: <CAOp6jLZKZKyYrJrzjZ90ZdxzzuDV0wp+XS3Rzssd-kvZV45JSQ@mail.gmail.com>
Subject: Re: [PATCH 2/2] perf test: Test FASYNC with watermark wakeups.
To: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Kyle Huey <me@kylehuey.com>, Ian Rogers <irogers@google.com>, Kyle Huey <khuey@kylehuey.com>, 
	Peter Zijlstra <peterz@infradead.org>, Ingo Molnar <mingo@redhat.com>, 
	Namhyung Kim <namhyung@kernel.org>, Mark Rutland <mark.rutland@arm.com>, 
	Alexander Shishkin <alexander.shishkin@linux.intel.com>, Jiri Olsa <jolsa@kernel.org>, 
	Adrian Hunter <adrian.hunter@intel.com>, linux-perf-users@vger.kernel.org, 
	linux-kernel@vger.kernel.org
Content-Type: text/plain; charset="UTF-8"

(I work with Kyle.)

IMHO this is more of a bug fix than a feature. `man perf_event_open`
expects this to work already: "watermark: If set, have an overflow
notification happen when we cross the wakeup_watermark boundary" and
later "Alternatively, the overflow events can be captured via a signal
handler, by enabling I/O signaling".

Bug fixes need regression tests. Such tests should fail on any kernel
where the bug is present. It seems strange to expect each such test to
detect whether the bug "should be fixed" in the kernel it's running on
and skip when that's not the case. I haven't seen any other project
try to do this. Instead (as in kernel selftests) the tests, the code
under test, and any metadata about which tests are expected to pass
are all in the repository together and updated together.

It makes sense that tests for the code in tools/perf should not fail
on older kernels, given that the code in tools/perf is expected to
work on older kernels. But tests for bug fixes in the kernel itself
should be expected to fail on older kernels and therefore should live
somewhere else, IMHO.

Rob
-- 
Su ot deraeppa sah dna Rehtaf eht htiw saw hcihw, efil lanrete eht uoy
ot mialcorp ew dna, ti ot yfitset dna ti nees evah ew; deraeppa efil
eht. Efil fo Drow eht gninrecnoc mialcorp ew siht - dehcuot evah sdnah
ruo dna ta dekool evah ew hcihw, seye ruo htiw nees evah ew hcihw,
draeh evah ew hcihw, gninnigeb eht morf saw hcihw taht.