lkml.org 
[lkml]   [2015]   [Apr]   [22]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH v3] perf tools: Document --children option in more detail
Hi Arnaldo,

On Tue, Apr 21, 2015 at 01:46:40PM -0300, Arnaldo Carvalho de Melo wrote:
> Em Wed, Apr 22, 2015 at 01:16:29AM +0900, Namhyung Kim escreveu:
> > On Tue, Apr 21, 2015 at 12:41:33PM -0300, Arnaldo Carvalho de Melo wrote:
> > > > +++ b/tools/perf/Documentation/overhead-calculation.txt
> > > I think Ingo suggested that you renamed this file to include the word
> > > "callchain" in it, no? looking at "overhead-calculation" I feel like I
> > > first have to open it to figure out what kind of overhead is this,
> > > perhaps it would be better named:
>
> > > tools/perf/Documentation/callchain-overhead.txt
> > > ?
>
> > Please see my reply to the Ingo's post. I think he agreed on this name.
>
> I still find it confusing for the file name, where there is no context,
> from just the file name when one does a 'ls tools/perf/Documentatoin' to
> figure out about what overhead that is referring to.
>
> So, perhaps a longer name:
>
> tools/perf/Documentation/callchain-overhead-calculation.txt
>
> ?

OK, will change.

>
> Inside perf-{record,top}.txt, yeah, we have context, we know that this
> is about post processing, formatting, etc.

Right.

>
> <SNIP>
>
> > > > +--no-children option on the command line or by adding 'report.children
> > > > += false' or 'top.children = false' in the perf config file.
> > >
> > > One can as well use the OPTION_FOO shortening mechanism and instead use:
> > >
> > > perf report --no-ch
> > >
> > > Which is enough to disambiguate it from "--no-column-widths" and "--no-cpu".
> >
> > Are you saying that you want to add the short form instead of the full
> > --no-chlidren name? I think we need to verbose in the manpage at
> > least and it might not work in the future if some --chxxx option is
> > added.
>
> Perhaps:
>
> "--no-children option on the command line or by adding 'report.children = false'
> or 'top.children = false' in the perf config file.
>
> A shorter form on the command line can be used, for instance '--no-ch'
> is unambiguous at the time of this writing."

I don't think it belongs here. The shorter form is not only for the
--children, so it should be described in different place.


>
> > > > diff --git a/tools/perf/Documentation/perf-report.txt b/tools/perf/Documentation/perf-report.txt
> > > > index 4879cf638824..b7bb298deee3 100644
> > > > --- a/tools/perf/Documentation/perf-report.txt
> > > > +++ b/tools/perf/Documentation/perf-report.txt
> > > > @@ -193,6 +193,7 @@ OPTIONS
> > > > Accumulate callchain of children to parent entry so that then can
> > > > show up in the output. The output will have a new "Children" column
> > > > and will be sorted on the data. It requires callchains are recorded.
> > > > + See the `overhead calculation' section for more details.
> > >
> > > `callchain overhead'
> >
> > Do you prefer this name to 'overhead calculation'? For me, it looks
>
> It is ok with me "overhead calculation", as mentioned previously in this
> message, the context in this perf-report.txt file should make it clear
> that the overhead is about callchains.

OK, I'll leave it as is.

Thanks,
Namhyung


>
> > like saying about how much overhead will be added if we enabled
> > callchains at perf record time or processing them at perf report time.
>
> Ok.
>
> - Arnaldo


\
 
 \ /
  Last update: 2015-04-22 08:41    [W:0.049 / U:4.444 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site