linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Arnaldo Carvalho de Melo <acme@kernel.org>
To: Andi Kleen <andi@firstfloor.org>
Cc: Jiri Olsa <jolsa@redhat.com>, Namhyung Kim <namhyung@kernel.org>,
	Jiri Olsa <jolsa@kernel.org>,
	linux-kernel@vger.kernel.org, "Liang, Kan" <kan.liang@intel.com>
Subject: Re: Implement lbr-as-callgraph v10
Date: Fri, 21 Nov 2014 17:30:57 -0300	[thread overview]
Message-ID: <20141121203057.GD18625@kernel.org> (raw)
In-Reply-To: <20141120193347.GL3790@kernel.org>

Em Thu, Nov 20, 2014 at 04:33:47PM -0300, Arnaldo Carvalho de Melo escreveu:
> Em Wed, Nov 19, 2014 at 10:48:22PM +0100, Andi Kleen escreveu:
> > > > directory since I'm comparing the output of --stdio, --tui and --gtk,
> > > > since it looks --gtk is wrong, still unsure about what the problem is in
> > > > that case, but stumbled on:

> > > I need to investigate this further, so I created a perf/branch-history
> > > branch that has the patches I need to test more rebased on top of my
> > > perf/core branch I just pushed out to Ingo.

> > I tested --gtk and I don't see any differences to the console mode
> > with --branch-history.  What problem do you see?

> The difference is with --tui, but I haven't checked if this is a problem
> introduced by your patchkit or if this is something that was there
> before it was applied.

So, here it is, --gtk looks like --stdio:

$ perf report --no-children --branch-history --stdio

# Samples: 43  of event 'cycles'
# Event count (approx.): 26843162
#
# Overhead  Source:Line  Symbol                     Shared Object   
# ........  ...........  .........................  ................
#
    68.42%  tcall.c:5    [.] f2                     tcall           
            |          
            |--87.65%-- f2 tcall.c:4
            |          |          
            |          |--67.41%-- f1 tcall.c:10
            |          |          f1 tcall.c:9
            |          |          main tcall.c:17
            |          |          main tcall.c:17
            |          |          main tcall.c:16
            |          |          main tcall.c:16
            |          |          f1 tcall.c:12
            |          |          f1 tcall.c:12
            |          |          f2 tcall.c:6
            |          |          f2 tcall.c:4
            |          |          f1 tcall.c:11
            |          |          f1 tcall.c:11
            |          |          f2 tcall.c:6
            |          |          f2 tcall.c:4
            |          |          f1 tcall.c:10
            |          |          
            |           --32.59%-- f1 tcall.c:11
            |                     f1 tcall.c:11
            |                     f1 tcall.c:11
            |                     f2 tcall.c:6
            |                     f2 tcall.c:4
            |                     f1 tcall.c:10
            |                     f1 tcall.c:9
            |                     main tcall.c:17
            |                     main tcall.c:17
            |                     main tcall.c:16
            |                     main tcall.c:16
            |                     f1 tcall.c:12
            |                     f1 tcall.c:12
            |                     f2 tcall.c:6
            |                     f2 tcall.c:4
            |                     f1 tcall.c:11
            |          
             --12.35%-- f1 tcall.c:9
                       main tcall.c:17
                       main tcall.c:17
                       main tcall.c:16
                       main tcall.c:16
                       f1 tcall.c:12
                       f1 tcall.c:12
                       f2 tcall.c:6
                       f2 tcall.c:4
                       f1 tcall.c:11
                       f1 tcall.c:11
                       f2 tcall.c:6
                       f2 tcall.c:4
                       f1 tcall.c:10
                       f1 tcall.c:9
                       main tcall.c:17
<SNIP>

But:

$ perf report --no-children --branch-history --tui
# Then expand a few callchains and press 'P' to generate a perf.hist.0
# file:
-   68.42%  tcall.c:5    [.] f2                     tcall
   - f2 tcall.c:4
      - 67.41% f1 tcall.c:10
           f1 tcall.c:9
           main tcall.c:17
           main tcall.c:17
           main tcall.c:16
           main tcall.c:16
           f1 tcall.c:12
           f1 tcall.c:12
           f2 tcall.c:6
           f2 tcall.c:4
           f1 tcall.c:11
           f1 tcall.c:11
           f2 tcall.c:6
           f2 tcall.c:4
           f1 tcall.c:10
      - 32.59% f1 tcall.c:11
           f1 tcall.c:11
           f2 tcall.c:6
           f2 tcall.c:4
           f1 tcall.c:10
           f1 tcall.c:9
           main tcall.c:17
           main tcall.c:17
           main tcall.c:16
           main tcall.c:16
           f1 tcall.c:12
           f1 tcall.c:12
           f2 tcall.c:6
           f2 tcall.c:4
           f1 tcall.c:11
     f1 tcall.c:9
     main tcall.c:17
     main tcall.c:17
     main tcall.c:16
     main tcall.c:16
     f1 tcall.c:12
     f1 tcall.c:12
     f2 tcall.c:6
     f2 tcall.c:4
     f1 tcall.c:11
     f1 tcall.c:11
     f2 tcall.c:6
     f2 tcall.c:4
     f1 tcall.c:10
     f1 tcall.c:9
     main tcall.c:17
<SNIP>


Do you see the diff?  The 87.65% and 12.35% doesn't appear on the --tui
output.

But I don't know if this is due to your patchkit, trying to check.

- Arnaldo

  parent reply	other threads:[~2014-11-21 20:31 UTC|newest]

Thread overview: 55+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-11-13  2:05 Implement lbr-as-callgraph v10 Andi Kleen
2014-11-13  2:05 ` [PATCH 01/10] perf, tools: Factor out adding new call chain entries Andi Kleen
2014-11-13 19:14   ` Arnaldo Carvalho de Melo
2014-11-20  7:37   ` [tip:perf/core] perf callchain: " tip-bot for Andi Kleen
2014-11-13  2:05 ` [PATCH 02/10] perf, tools: Support handling complete branch stacks as histograms Andi Kleen
2014-11-13 19:14   ` Arnaldo Carvalho de Melo
2014-11-13 19:52     ` Andi Kleen
2014-11-13 20:08       ` Arnaldo Carvalho de Melo
2014-11-13 20:15         ` Andi Kleen
2014-11-13 20:42           ` Arnaldo Carvalho de Melo
2014-12-08  6:53   ` [tip:perf/core] perf callchain: " tip-bot for Andi Kleen
2014-11-13  2:05 ` [PATCH 03/10] perf, tools: Use al.addr to set up call chain Andi Kleen
2014-11-13 19:16   ` Arnaldo Carvalho de Melo
2014-12-11 21:46     ` Jiri Olsa
2014-12-11 22:27       ` Andi Kleen
2014-11-20  7:38   ` [tip:perf/core] perf callchain: " tip-bot for Andi Kleen
2014-11-13  2:05 ` [PATCH 04/10] perf, tools: Add --branch-history option to report Andi Kleen
2014-12-08  6:53   ` [tip:perf/core] perf report: Add --branch-history option tip-bot for Andi Kleen
2014-11-13  2:05 ` [PATCH 05/10] perf, tools: Use a common function to resolve symbol or name Andi Kleen
2014-11-13 19:17   ` Arnaldo Carvalho de Melo
2014-11-20  7:38   ` [tip:perf/core] perf callchain: " tip-bot for Andi Kleen
2014-11-13  2:05 ` [PATCH 06/10] perf, tools: Enable printing the srcline in the history Andi Kleen
2014-11-13 19:20   ` Arnaldo Carvalho de Melo
2014-12-08  6:48   ` [tip:perf/core] perf callchain: " tip-bot for Andi Kleen
2014-11-13  2:05 ` [PATCH 07/10] perf, tools: Only print base source file for srcline Andi Kleen
2014-11-13 19:22   ` Arnaldo Carvalho de Melo
2014-11-20  7:38   ` [tip:perf/core] perf " tip-bot for Andi Kleen
2014-11-13  2:05 ` [PATCH 08/10] perf, tools: Support source line numbers in annotate Andi Kleen
2014-11-13 20:52   ` Arnaldo Carvalho de Melo
2014-11-20  7:39   ` [tip:perf/core] perf annotate: " tip-bot for Andi Kleen
2014-11-13  2:05 ` [PATCH 09/10] tools, perf: Make get_srcline fall back to sym+offset Andi Kleen
2014-12-08  6:49   ` [tip:perf/core] perf callchain: " tip-bot for Andi Kleen
2014-11-13  2:05 ` [PATCH 10/10] tools, perf: Add asprintf replacement Andi Kleen
2014-11-13 20:53   ` Arnaldo Carvalho de Melo
2014-11-13 21:14     ` Andi Kleen
2014-11-17 21:34 ` Implement lbr-as-callgraph v10 Arnaldo Carvalho de Melo
2014-11-18  1:56   ` Andi Kleen
2014-11-18 10:44   ` Jiri Olsa
2014-11-18 11:00     ` Jiri Olsa
2014-11-18 13:37       ` Arnaldo Carvalho de Melo
2014-11-19 15:31         ` Andi Kleen
2014-11-19  6:21       ` Namhyung Kim
2014-11-19  9:23         ` Jiri Olsa
2014-11-19 10:54           ` Jiri Olsa
2014-11-19 14:10             ` Arnaldo Carvalho de Melo
2014-11-19 16:04               ` Arnaldo Carvalho de Melo
2014-11-19 21:48                 ` Andi Kleen
2014-11-20 19:33                   ` Arnaldo Carvalho de Melo
2014-11-20 20:46                     ` Andi Kleen
2014-11-21 20:30                     ` Arnaldo Carvalho de Melo [this message]
2014-11-22  1:25                       ` Andi Kleen
2014-11-24  7:40                         ` Namhyung Kim
2014-11-19 21:50               ` Andi Kleen
2014-11-20 20:36                 ` Arnaldo Carvalho de Melo
  -- strict thread matches above, loose matches on Subject: below --
2014-09-26 23:37 Andi Kleen

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20141121203057.GD18625@kernel.org \
    --to=acme@kernel.org \
    --cc=andi@firstfloor.org \
    --cc=jolsa@kernel.org \
    --cc=jolsa@redhat.com \
    --cc=kan.liang@intel.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=namhyung@kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).