Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
From:   Song Liu <songliubraving@fb.com>
To:     Jiri Olsa <jolsa@redhat.com>
CC:     linux-kernel <linux-kernel@vger.kernel.org>,
        Kernel Team <Kernel-team@fb.com>,
        "acme@kernel.org" <acme@kernel.org>,
        "acme@redhat.com" <acme@redhat.com>,
        "namhyung@kernel.org" <namhyung@kernel.org>,
        "jolsa@kernel.org" <jolsa@kernel.org>
Subject: Re: [PATCH] perf-stat: introduce bperf, share hardware PMCs with BPF
Thread-Topic: [PATCH] perf-stat: introduce bperf, share hardware PMCs with BPF
Thread-Index: AQHXFuPifzAWbJ/c0EGfxtA8c/wTpKqAREAAgAA7WYCAAAbkgA==
Date:   Fri, 12 Mar 2021 16:09:53 +0000
Message-ID: <B7934C3F-4414-45AA-BE39-8FE3C64B7E7D@fb.com>
References: <20210312020257.197137-1-songliubraving@fb.com>
 <YEtawFnompBDKpK0@krava> <45BCAC33-1626-42D1-A170-92DC8D7BAAF8@fb.com>
In-Reply-To: <45BCAC33-1626-42D1-A170-92DC8D7BAAF8@fb.com>
Accept-Language: en-US
Content-Language: en-US
Content-Type: text/plain; charset="us-ascii"
Content-ID: <BD7F889D183F3B4D998B019FF6AE32F7@namprd15.prod.outlook.com>
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0
X-MS-Exchange-CrossTenant-Network-Message-Id: 2d36b3ad-38b9-4908-d173-08d8e57145d4
X-MS-Exchange-CrossTenant-originalarrivaltime: 12 Mar 2021 16:09:53.7046
 (UTC)
X-MS-Exchange-CrossTenant-fromentityheader: Hosted
X-MS-Exchange-CrossTenant-id: 8ae927fe-1255-47a7-a2af-5f3a069daaa2
X-MS-Exchange-CrossTenant-mailboxtype: HOSTED
X-MS-Exchange-CrossTenant-userprincipalname: QHMh7Ix5s9h0RlJ3KFkvLmBMkuhLsH7lNerVCtU6kcNLL2y3zTxDxiU1kF5KeovqkI3jiaHna4NH9L1hZJPxsQ==
X-MS-Exchange-Transport-CrossTenantHeadersStamped: BYAPR15MB4246
X-OriginatorOrg: fb.com
X-Proofpoint-Virus-Version: vendor=fsecure engine=2.50.10434:6.0.369,18.0.761
 definitions=2021-03-12_06:2021-03-12,2021-03-12 signatures=0
X-Proofpoint-Spam-Details: rule=fb_default_notspam policy=fb_default score=0 mlxscore=0 malwarescore=0
 mlxlogscore=999 spamscore=0 suspectscore=0 impostorscore=0 bulkscore=0
 lowpriorityscore=0 phishscore=0 clxscore=1015 adultscore=0
 priorityscore=1501 classifier=spam adjust=0 reason=mlx scancount=1
 engine=8.12.0-2009150000 definitions=main-2103120116
X-FB-Internal: deliver
Precedence: bulk
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org


> On Mar 12, 2021, at 7:45 AM, Song Liu <songliubraving@fb.com> wrote:
>=20
>=20
>=20
>> On Mar 12, 2021, at 4:12 AM, Jiri Olsa <jolsa@redhat.com> wrote:
>>=20
>> On Thu, Mar 11, 2021 at 06:02:57PM -0800, Song Liu wrote:
>>> perf uses performance monitoring counters (PMCs) to monitor system
>>> performance. The PMCs are limited hardware resources. For example,
>>> Intel CPUs have 3x fixed PMCs and 4x programmable PMCs per cpu.
>>>=20
>>> Modern data center systems use these PMCs in many different ways:
>>> system level monitoring, (maybe nested) container level monitoring, per
>>> process monitoring, profiling (in sample mode), etc. In some cases,
>>> there are more active perf_events than available hardware PMCs. To allo=
w
>>> all perf_events to have a chance to run, it is necessary to do expensiv=
e
>>> time multiplexing of events.
>>>=20
>>> On the other hand, many monitoring tools count the common metrics (cycl=
es,
>>> instructions). It is a waste to have multiple tools create multiple
>>> perf_events of "cycles" and occupy multiple PMCs.
>>>=20
>>> bperf tries to reduce such wastes by allowing multiple perf_events of
>>> "cycles" or "instructions" (at different scopes) to share PMUs. Instead
>>> of having each perf-stat session to read its own perf_events, bperf use=
s
>>> BPF programs to read the perf_events and aggregate readings to BPF maps=
.
>>> Then, the perf-stat session(s) reads the values from these BPF maps.
>>>=20
>>> Please refer to the comment before the definition of bperf_ops for the
>>> description of bperf architecture.
>>>=20
>>> bperf is off by default. To enable it, pass --use-bpf option to perf-st=
at.
>>> bperf uses a BPF hashmap to share information about BPF programs and ma=
ps
>>> used by bperf. This map is pinned to bpffs. The default address is
>>> /sys/fs/bpf/bperf_attr_map. The user could change the address with opti=
on
>>> --attr-map.
>>=20
>> nice, I recall the presentation about that and was wondering
>> when this will come up ;-)
>=20
> The progress is slower than I expected. But I finished some dependencies =
of=20
> this in the last year:=20
>=20
>  1. BPF_PROG_TEST_RUN for raw_tp event;
>  2. perf-stat -b, which introduced skeleton and bpf_counter;
>  3. BPF task local storage, I didn't use it in this version, but it could=
,
>     help optimize bperf in the future.=20
>=20
>>=20
>>>=20
>>> ---
>>> Known limitations:
>>> 1. Do not support per cgroup events;
>>> 2. Do not support monitoring of BPF program (perf-stat -b);
>>> 3. Do not support event groups.
>>>=20
>>> The following commands have been tested:
>>>=20
>>>  perf stat --use-bpf -e cycles -a
>>>  perf stat --use-bpf -e cycles -C 1,3,4
>>>  perf stat --use-bpf -e cycles -p 123
>>>  perf stat --use-bpf -e cycles -t 100,101
>>=20
>> I assume the output is same as standard perf?

Btw, please give it a try. :)=20

It worked pretty well in my tests. If it doesn't work for some combination=
=20
of options, please let me know.=20

Thanks,
Song

>=20
> Yes, the output is identical to that without --use-bpf option.=20