lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite for Android: free password hash cracker in your pocket
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Date:	Thu, 15 Mar 2012 14:46:09 -0700
From:	Arun Sharma <asharma@...com>
To:	Namhyung Kim <namhyung.kim@....com>
CC:	Ingo Molnar <mingo@...e.hu>, Namhyung Kim <namhyung@...il.com>,
	Arnaldo Carvalho de Melo <acme@...hat.com>,
	Peter Zijlstra <peterz@...radead.org>,
	Stephane Eranian <eranian@...gle.com>,
	<linux-kernel@...r.kernel.org>, <dsahern@...il.com>,
	<ravitillo@....gov>, <khandual@...ux.vnet.ibm.com>,
	<robert.richter@....com>, <ming.m.lin@...el.com>,
	<vweaver1@...s.utk.edu>, <andi@...stfloor.org>
Subject: Re: [RFC] perf report: Implement symbol filtering on TUI

On 3/14/12 5:44 PM, Namhyung Kim wrote:

> Thanks for reviewing and sending the patch.
>
> However this was already fixed on my new patch set. Please see below:
>
> https://lkml.org/lkml/2012/3/13/73
>

I still get different behaviors with and without my two line patch.

Try:

perf record -g
perf report c -G -g graph,0.5,caller -s inclusive --stdio

I don't see where hists__filter_by_symbol() gets called when using --stdio.

But let's examine what I think is a more important general issue (not 
specific to your patch):

Here's the behavior my coworkers are requesting. With the test program I 
mailed earlier with this callgraph:

./perf report -G -g graph,0.5,caller -s pid --stdio

   100.00%            foo:30804
             |
             --- __libc_start_main
                |
                 --99.87%-- main
                           |
                           |--50.01%-- a
                           |          |
                           |           --42.71%-- c
                           |                     |
                           |                     |--14.36%-- e
                           |                     |          |
                           |                     |           --7.09%-- f
                           |                     |
                           |                      --14.33%-- d
                           |                                |
                           |                                 --6.81%-- f
                           |
                            --49.86%-- b
                                      |
                                       --42.72%-- c
                                                 |
                                                 |--14.43%-- d
                                                 |          |
                                                 |           --7.30%-- f
                                                 |
                                                  --14.13%-- e
                                                            |
                                                             --7.05%-- f

If the user filters by "d", we should:

a) Construct a callgraph that has only callchains containing "d"
b) Express percentages as a fraction of the unfiltered total_period
c) Allow users to expand callchains and see both the callers and callees 
of d.

Right now, I can do this in two steps:

# Get the callgraph leading to "d"
perf report -S d -G -g graph,0.5,callee

  100.00%      foo  foo
                 |
                 --- __libc_start_main
                     main
                    |
                    |--51.37%-- a
                    |          c
                    |          d
                    |
                     --48.63%-- b
                               c
                               d

# Get the callgraph below "d" (i.e. where d is the caller)

perf report d -G -g graph,0.5,caller -s inclusive

    95.56%  [.] d
             |
             --- d
                |
                 --46.88%-- f

and the percentages are not expressed as a percentages of the total_period.

I'll send out a RFC patch next.

  -Arun
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ