Return-Path: <linux-kernel-owner@vger.kernel.org>
Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand
        id S1751520AbdF1Jxd (ORCPT <rfc822;w@1wt.eu>);
        Wed, 28 Jun 2017 05:53:33 -0400
Received: from mail.kdab.com ([176.9.126.58]:44828 "EHLO mail.kdab.com"
        rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
        id S1751498AbdF1Jx0 (ORCPT <rfc822;linux-kernel@vger.kernel.org>);
        Wed, 28 Jun 2017 05:53:26 -0400
From: Milian Wolff <milian.wolff@kdab.com>
To: Taeung Song <treeze.taeung@gmail.com>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>,
        linux-kernel@vger.kernel.org, Adrian Hunter <adrian.hunter@intel.com>,
        Andi Kleen <ak@linux.intel.com>, David Ahern <dsahern@gmail.com>,
        Jin Yao <yao.jin@linux.intel.com>, Jiri Olsa <jolsa@redhat.com>,
        Kim Phillips <kim.phillips@arm.com>,
        Masami Hiramatsu <mhiramat@kernel.org>,
        Namhyung Kim <namhyung@kernel.org>, Wang Nan <wangnan0@huawei.com>
Subject: Re: [PATCH/RFC 0/4] perf annotate: Add --source-only option and the new source code TUI view
Date: Wed, 28 Jun 2017 11:53:22 +0200
Message-ID: <1571575.U9nG2MaxHg@milian-kdab2>
Organization: KDAB
In-Reply-To: <1498619888-1305-1-git-send-email-treeze.taeung@gmail.com>
References: <1498619888-1305-1-git-send-email-treeze.taeung@gmail.com>
MIME-Version: 1.0
Content-Type: multipart/signed; boundary="nextPart10793721.1pC3Hq5r17"; micalg="sha256"; protocol="application/pkcs7-signature"
Sender: linux-kernel-owner@vger.kernel.org
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org
Content-Length: 12255
Lines: 247

--nextPart10793721.1pC3Hq5r17
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain; charset="UTF-8"

On Wednesday, June 28, 2017 5:18:08 AM CEST Taeung Song wrote:
> Hi,
>=20
> The --source-only option and new source code TUI view can show
> the result of performance analysis based on full source code per
> symbol(function). (Namhyung Kim told me this idea and it was also request=
ed
> by others some time ago..)
>=20
> If someone wants to see the cause, he/she will need to dig into the asm.
> But before that, looking at the source level can give a hint or clue
> for the problem.
>=20
> For example, if target symbol is 'hex2u64' of util/util.c,
> the output is like below.
>=20
>     $ perf annotate --source-only --stdio -s hex2u64
>  Percent |      Source code of util.c for cycles:ppp (42 samples)
> -----------------------------------------------------------------
>     0.00 : 354   * While we find nice hex chars, build a long_val.
>     0.00 : 355   * Return number of chars processed.
>     0.00 : 356   */
>     0.00 : 357  int hex2u64(const char *ptr, u64 *long_val)
>     2.38 : 358  {
>     2.38 : 359          const char *p =3D ptr;
>     0.00 : 360          *long_val =3D 0;
>     0.00 : 361
>    30.95 : 362          while (*p) {
>    23.81 : 363                  const int hex_val =3D hex(*p);
>     0.00 : 364
>    14.29 : 365                  if (hex_val < 0)
>     0.00 : 366                          break;
>     0.00 : 367
>    26.19 : 368                  *long_val =3D (*long_val << 4) | hex_val;
>     0.00 : 369                  p++;
>     0.00 : 370          }
>     0.00 : 371
>     0.00 : 372          return p - ptr;
>     0.00 : 373  }
>=20
> And I added many perf developers into Cc: because I want to listen to your
> opinions about this new feature, if you don't mind.
>=20
> If you give some feedback, I'd appreciate it! :)

Thanks Taeung,

I requested this feature some time ago and it's really cool to see someone=
=20
step up and implement it - much appreciated!

I just tested it out on my pet-example that leverages C++ instead of C:

~~~~~
#include <complex>
#include <cmath>
#include <random>
#include <iostream>

using namespace std;

int main()
{
    uniform_real_distribution<double> uniform(-1E5, 1E5);
    default_random_engine engine;
    double s =3D 0;
    for (int i =3D 0; i < 10000000; ++i) {
        s +=3D norm(complex<double>(uniform(engine), uniform(engine)));
    }
    cout << s << '\n';
    return 0;
}
~~~~~

Compile it with:

g++ -O2 -g -std=3Dc++11 test.cpp -o test

Then record it with perf:

perf record --call-graph dwarf ./test

Then analyse it with `perf report`. You'll see one entry for main with=20
something like:

+  100.00%    39.69%  cpp-inlining  cpp-inlining      [.] main

Select it and annotate it, then switch to your new source-only view:

main  test.cpp
       =E2=94=82  30                                                       =
                                                                           =
                                                                 =20
       =E2=94=82  31    using namespace std;                               =
                                                                           =
                                                                 =20
       =E2=94=82  32                                                       =
                                                                           =
                                                                 =20
       =E2=94=82  33    int main()                                         =
                                                                           =
                                                                 =20
       =E2=94=82+ 34    {                                                  =
                                                                           =
                                                                 =20
       =E2=94=82  35        uniform_real_distribution<double> uniform(-1E5,=
 1E5);                                                                     =
                                                                 =20
       =E2=94=82  36        default_random_engine engine;                  =
                                                                           =
                                                                 =20
       =E2=94=82+ 37        double s =3D 0;                                =
                                                                           =
                                                                   =20
       =E2=94=82+ 38        for (int i =3D 0; i < 10000000; ++i) {         =
                                                                           =
                                                                   =20
  4.88 =E2=94=82+ 39            s +=3D norm(complex<double>(uniform(engine)=
,=20
uniform(engine)));                                                         =
                                                                 =20
       =E2=94=82  40        }                                              =
                                                                           =
                                                                 =20
       =E2=94=82  41        cout << s << '\n';                             =
                                                                           =
                                                                 =20
       =E2=94=82  42        return 0;                                      =
                                                                           =
                                                                 =20
       =E2=94=82+ 43    }=20

Note: the line numbers are off b/c my file contains a file-header on-top.=20
Ignore that.

Note2: There is no column header shown, so it's unclear what the first colu=
mn=20
represents.

Note 3: report showed 39.69% self cost in main, 100.00% inclusive. annotate=
=20
shows 4.88... What is that?

What this shows, is that it's extremely important to visualize inclusive co=
st=20
_and_ self cost in this view. Additionally, we need to account for inlining=
=2E=20
Right now, we only see the self cost that is directly within main, I suspec=
t.=20
=46or C++ this is usually very misleading, and basically makes the annotate=
 view=20
completely useless for application-level profiling. If a second column woul=
d=20
be added with the inclusive cost with the ability to drill down, then I cou=
ld=20
easily see myself using this view.

I would appreciate if you could take this into account.

Thanks a lot


=2D-=20
Milian Wolff | milian.wolff@kdab.com | Senior Software Engineer
KDAB (Deutschland) GmbH&Co KG, a KDAB Group company
Tel: +49-30-521325470
KDAB - The Qt Experts
--nextPart10793721.1pC3Hq5r17
Content-Type: application/pkcs7-signature; name="smime.p7s"
Content-Disposition: attachment; filename="smime.p7s"
Content-Transfer-Encoding: base64
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--nextPart10793721.1pC3Hq5r17--