Re: [PATCH RFC V9 2/3] perf,tools: per-event callgraph support

[Arnaldo Carvalho de Melo <acme@kernel.org>]

Em Sat, Aug 08, 2015 at 06:45:39PM +0200, Jiri Olsa escreveu:
> On Fri, Aug 07, 2015 at 12:49:38PM -0300, Arnaldo Carvalho de Melo wrote:
> > Em Fri, Aug 07, 2015 at 12:38:43PM -0300, Arnaldo Carvalho de Melo escreveu:
> > > Em Thu, Aug 06, 2015 at 03:44:53PM -0400, kan.liang@intel.com escreveu:
> > > > From: Kan Liang <kan.liang@intel.com>
> > > > Here are some examples and test results.
> > > > 
> > > > 1. Comparing the elapsed time and perf.data size from "kernbench -M -H".
> > > > 
> > > >  The test command for FULL callgraph and time support.
> > > >    "perf record -e
> > > >    '{cpu/cpu-cycles,period=100000/,cpu/instructions,period=20000/p}'
> > > >    --call-graph fp --time"
> > 
> > Some other oddity I noticed while testing:
> > 
> 
> SNIP
> 
> > This is in the 'perf report' TUI, why, for an event with 'callgraph=no', we
> > get callchains? How come?
> > 
> > If I try it with the --stdio, well, its there as well:
> > 
> > # Samples: 2K of event 'cpu/instructions,call-graph=no,time=0,period=20000/p'
> > # Event count (approx.): 46956518
> > #
> > # Children      Self  Command          Shared Object               Symbol                                        
> > # ........  ........  ...............  ..........................  ..............................................
> > #
> >     67.56%     0.00%  qemu-system-x86  [unknown]                   [.] 0xad5e258d4c544155                        
> >             |
> >             ---0xad5e258d4c544155
> > 
> >     67.56%     0.00%  qemu-system-x86  libc-2.20.so                [.] __libc_start_main                         
> >             |
> >             ---__libc_start_main
> >                0xad5e258d4c544155
> > 
> >     67.56%     0.00%  qemu-system-x86  perf                        [.] main                                      
> >             |
> >             ---main
> >                __libc_start_main
> >                0xad5e258d4c544155
> > 
> >     67.56%     0.00%  qemu-system-x86  perf                        [.] run_builtin                               
> >             |
> >             ---run_builtin
> >                main
> >                __libc_start_main
> >                0xad5e258d4c544155
> > 
> > Can you take a look? What I have, i.e. the patches from Jiri and Kan
> > moving stuff out of the way of the python binding and this patch is at
> > the tmp.perf/core branch in my git tree:
> > 
> >   git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git
> 
> hum, tried this branch and got proper behaviour for both stdio and TUI,
> attaching stdio output
> 
> the record command was:
>   $ ./perf record -e 'cpu/cpu-cycles,call-graph=fp,time,period=10000/pp,cpu/instructions,call-graph=no,time=0,period=20000/p' ls
> 
> got the events strings from your report output.. what was different in your record command?

Was in my message, here it is:

# cmdline : /home/acme/bin/perf record -e {cpu/cpu-cycles,call-graph=fp,time,period=10000/pp,cpu/instructions,call-graph=no,time=0,period=20000/p}
-a

I'll try this again later.

- Arnaldo