Re: [PATCH v3 0/3] perf report/annotate: Support average IPC and IPC coverage for function

[Jin, Yao]

On 11/29/2018 6:13 PM, Jiri Olsa wrote:
> On Thu, Nov 29, 2018 at 02:24:27PM +0800, Jin, Yao wrote:
> > On 11/28/2018 6:18 PM, Jiri Olsa wrote:
> > > On Wed, Nov 28, 2018 at 11:17:57AM +0100, Jiri Olsa wrote:
> > > > On Wed, Nov 28, 2018 at 11:14:55PM +0800, Jin Yao wrote:
> > > > > Add supporting of displaying the average IPC and IPC coverage
> > > > > percentage per function.
> > > > >
> > > > > For example,
> > > > >
> > > > > $ perf record -b ...
> > > > > $ perf report -s symbol or
> > > > >     perf report -s symbol --stdio
> > > > >
> > > > > Overhead  Symbol                           IPC   [IPC Coverage]
> > > > >     39.60%  [.] __random                     2.30  [ 54.8%]
> > > > >     18.02%  [.] main                         0.43  [ 54.3%]
> > > > >     14.21%  [.] compute_flag                 2.29  [100.0%]
> > > > >     14.16%  [.] rand                         0.36  [100.0%]
> > > > >      7.06%  [.] __random_r                   2.57  [ 70.5%]
> > > > >      6.85%  [.] rand@plt                     0.00  [  0.0%]
> > > > >     ...
> > > > >
> > > > > $ perf annotate --stdio2
> > > > >
> > > > > Percent  IPC Cycle (Average IPC: 2.30, IPC Coverage: 54.8%)
> > > > >
> > > > >                           Disassembly of section .text:
> > > > >
> > > > >                           000000000003aac0 <random@@GLIBC_2.2.5>:
> > > > >     8.32  3.28              sub    $0x18,%rsp
> > > > >           3.28              mov    $0x1,%esi
> > > > >           3.28              xor    %eax,%eax
> > > > >           3.28              cmpl   $0x0,argp_program_version_hook@@GLIBC_2.2.5+0x1e0
> > > > >    11.57  3.28     1      â je     20
> > > > >                             lock   cmpxchg %esi,__abort_msg@@GLIBC_PRIVATE+0x8a0
> > > > >                           â jne    29
> > > > >                           â jmp    43
> > > > >    11.57  1.10        20:   cmpxchg %esi,__abort_msg@@GLIBC_PRIVATE+0x8a0
> > > > >    ...
> > > > >
> > > > > v3:
> > > > > ---
> > > > >       Remove the sortkey "ipc" from command-line. The columns "IPC"
> > > > >       and "[IPC Coverage]" are automatically enabled when "symbol"
> > > > >       is specified.
> > > > >
> > > > >       Patch "perf report: Display average IPC and IPC coverage per symbol"
> > > > >       is impacted.
> > > > >
> > > > > v2:
> > > > > ---
> > > > >     1. Merge in Jiri's patch to support stdio mode
> > > > >
> > > > >     2. Add a new patch "perf annotate: Create a annotate2 flag
> > > > >        in struct symbol" which records if the symbol has been
> > > > >        annotated yet.
> > > > >
> > > > >     3. Minor update such as adding { } for multiline code in 'if'
> > > > >        condition.
> > > > >
> > > > > Jin Yao (3):
> > > > >     perf annotate: Compute average IPC and IPC coverage per symbol
> > > > >     perf annotate: Create a annotate2 flag in struct symbol
> > > > >     perf report: Display average IPC and IPC coverage per symbol
> > > >
> > > > hi,
> > > > I took he liberty and moved the annotation retrieval into
> > > > resort phase under progress bar scope. It's currently on top
> > > > of my perf/fixes branch, could you please check it?
> > > >
> > > >     git://git.kernel.org/pub/scm/linux/kernel/git/jolsa/perf.git
> > >
> > > commits:
> > > 7f3ffdb9783f perf tools: Move symbol annotation to resort
> > > e87f7d3c4f10 perf tools: Add perf_evsel__output_resort_cb function
> > > 40012b422108 perf tools: Add argument to hists__resort_cb_t callback
> > >
> > > jirka
> >
> > Hi Jiri,
> >
> > Thanks for your patches. I have tested with your repo. Now I can see 2
> > progress bars. One is displayed at the events processing phase, the other is
> > displayed at resorting phase.
> >
> > I have only one concern that is, in my test, much of time is consumed by the
> > event processing phase, for example, 90% of time. Only 10% of time is
> > consumed at resorting phase.
> >
> > So do we really need the second progress bar?
>
> well I did not add it, it's been always there, it just must
> have been real quick for you so u did not notice I guess

Yes, I think so. :)

> it's strange, because for me the resorting takes much longer
> even for small data.. let's have your patchset applied and
> have this discussion when I send out the patches

That's fine! I will post v5 soon.

Thanks
Jin Yao

> thanks,
> jirka