Date:   Fri, 10 Dec 2021 14:55:32 -0300
From:   Arnaldo Carvalho de Melo <arnaldo.melo@gmail.com>
To:     Mathieu Poirier <mathieu.poirier@linaro.org>,
        James Clark <james.clark@arm.com>
CC:     Leo Yan <leo.yan@linaro.org>, coresight@lists.linaro.org,
        suzuki.poulose@arm.com, Mike Leach <mike.leach@linaro.org>,
        John Garry <john.garry@huawei.com>,
        Will Deacon <will@kernel.org>,
        Mark Rutland <mark.rutland@arm.com>,
        Alexander Shishkin <alexander.shishkin@linux.intel.com>,
        Jiri Olsa <jolsa@redhat.com>,
        Namhyung Kim <namhyung@kernel.org>,
        linux-arm-kernel@lists.infradead.org,
        linux-perf-users@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: =?US-ASCII?Q?Re=3A_=5BPATCH=5D_perf_cs-etm=3A_Remove_dup?= =?US-ASCII?Q?licate_and_incorrect_aux_size_checks?=
User-Agent: K-9 Mail for Android
In-Reply-To: <20211210165436.GA1236662@p14s>
References: <20211208115435.610101-1-james.clark@arm.com> <20211208131753.GC273781@leoy-ThinkPad-X240s> <269d2f14-0594-c73e-97b5-82e72f76e826@arm.com> <20211209134413.GA622826@leoy-ThinkPad-X240s> <6a7fd600-91f3-5feb-d21f-ec7cb704f84c@arm.com> <20211210165436.GA1236662@p14s>
Message-ID: <B3677362-D2A0-4341-A6C1-53E50A3B8C49@gmail.com>
MIME-Version: 1.0
Content-Type: text/plain;
 charset=utf-8
Content-Transfer-Encoding: quoted-printable
Precedence: bulk


On December 10, 2021 1:54:36 PM GMT-03:00, Mathieu Poirier <mathieu=2Epoir=
ier@linaro=2Eorg> wrote:
>On Thu, Dec 09, 2021 at 02:16:43PM +0000, James Clark wrote:
>>=20
>>=20
>> On 09/12/2021 13:44, Leo Yan wrote:
>> > On Wed, Dec 08, 2021 at 02:08:04PM +0000, James Clark wrote:
>> >> On 08/12/2021 13:17, Leo Yan wrote:
>> >>> Hi James,
>> >>>
>> >>> On Wed, Dec 08, 2021 at 11:54:35AM +0000, James Clark wrote:
>> >>>> There are two checks, one is for size when running without admin, =
but
>> >>>> this one is covered by the driver and reported on in more detail h=
ere
>> >>>> (builtin-record=2Ec):
>> >>>>
>> >>>>   pr_err("Permission error mapping pages=2E\n"
>> >>>>          "Consider increasing "
>> >>>>          "/proc/sys/kernel/perf_event_mlock_kb,\n"
>> >>>>          "or try again with a smaller value of -m/--mmap_pages=2E\=
n"
>> >>>>          "(current value: %u,%u)\n",
>> >>>
>> >>> I looked into the kernel code and found:
>> >>>
>> >>>   sysctl_perf_event_mlock =3D 512 + (PAGE_SIZE / 1024);  // 512KB +=
 1 page
>> >>>
>> >>> If the system have multiple cores, let's say 8 cores, then kernel e=
ven
>> >>> can relax the limitaion with:
>> >>>
>> >>>   user_lock_limit *=3D num_online_cpus();
>> >>>
>> >>> So means the memory lock limitation is:
>> >>>
>> >>>   (512KB + 1 page) * 8 =3D 4MB + 8 pages=2E
>> >>>
>> >>> Seems to me, it's much relax than the user space's limitaion 128KB=
=2E
>> >>> And let's imagine for Arm server, the permitted buffer size can be =
a
>> >>> huge value (e=2Eg=2E for a system with 128 cores)=2E
>> >>>
>> >>> Could you confirm if this is right?
>> >>
>> >> Yes that seems to be the case=2E And the commit message for that add=
ition
>> >> states the reasoning:
>> >>
>> >>   perf_counter: Increase mmap limit
>> >>  =20
>> >>   In a default 'perf top' run the tool will create a counter for
>> >>   each online CPU=2E With enough CPUs this will eventually exhaust
>> >>   the default limit=2E
>> >>
>> >>   So scale it up with the number of online CPUs=2E
>> >>
>> >> To me that makes sense=2E Normally the memory installed also scales =
with the
>> >> number of cores=2E
>> >>
>> >> Are you saying that we should look into modifying that scaling facto=
r in
>> >> perf_mmap()? Or that we should still add something to userspace for
>> >> coresight to limit user supplied buffer sizes?
>> >=20
>> > I don't think we should modify the scaling factor in perf_mmap(), the
>> > logic is not only used by AUX buffer, it's shared by normal event
>> > ring buffer=2E
>> >=20
>> >> I think it makes sense to allow the user to specify any value that w=
ill work,
>> >> it's up to them=2E
>> >=20
>> > Understand, I verified this patch with below steps:
>> >=20
>> > root@debian:~# echo 0 > /proc/sys/kernel/perf_event_paranoid
>> >=20
>> > leoy@debian:~$ perf record -e cs_etm// -m 4M,8M -o perf_test=2Edata -=
- sleep 1
>> > Permission error mapping pages=2E
>> > Consider increasing /proc/sys/kernel/perf_event_mlock_kb,
>> > or try again with a smaller value of -m/--mmap_pages=2E
>> > (current value: 1024,2048)
>> >=20
>> > leoy@debian:~$ perf record -e cs_etm// -m 4M,4M -o perf_test=2Edata -=
- sleep 1
>> > Couldn't synthesize bpf events=2E
>> > [ perf record: Woken up 1 times to write data ]
>> > [ perf record: Captured and wrote 0=2E607 MB perf_test=2Edata ]
>> >=20
>> > So this patch looks good for me:
>> >=20
>> > Reviewed-by: Leo Yan <leo=2Eyan@linaro=2Eorg>
>> >=20
>> Thanks Leo!
>
>Arnaldo is not on the recipient list and as such he won't see this patch=
=2E=2E=2E
>

I saw it now, can I take this as an acked-by: Matthieu too?

- Arnaldo