Re: [PATCH 3/9] perf tools: Account entry stats when it's added to the output tree

[Namhyung Kim]

On Tue, 22 Apr 2014 19:10:49 +0200, Jiri Olsa wrote:
> On Tue, Apr 22, 2014 at 05:49:45PM +0900, Namhyung Kim wrote:
>
> SNIP
>
>>  
>> diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c
>> index 1c77714f668d..f955ae5a41c5 100644
>> --- a/tools/perf/util/hist.c
>> +++ b/tools/perf/util/hist.c
>> @@ -344,9 +344,11 @@ void hists__inc_nr_entries(struct hists *hists, struct hist_entry *h)
>>  		hists__calc_col_len(hists, h);
>>  		hists->nr_non_filtered_entries++;
>>  		hists->stats.total_non_filtered_period += h->stat.period;
>> +		hists->stats.nr_non_filtered_samples += h->stat.nr_events;
>>  	}
>>  	hists->nr_entries++;
>>  	hists->stats.total_period += h->stat.period;
>> +	hists__add_nr_events(hists, PERF_RECORD_SAMPLE, h->stat.nr_events);
>
> while at it.. could hists__remove_entry_filter call hists__inc_nr_entries too?

Nope, I (and Arnaldo) want to the default stats invariant - they are not
changed by any filter.  As you can see hists__remove_entry_filter() only
updates non-filtered stats.

>
> not sure about those 2 extra lines in hists__remove_entry_filter:
>
>         if (h->ms.unfolded)
>                 hists->nr_non_filtered_entries += h->nr_rows;
>         h->row_offset = 0;
>
> maybe they should be added into hists__inc_nr_entries?

Hmm.. this seems like a mix of generic code and UI-specific code.  The
->nr_rows, ->row_offset and ->ms.unfolded are only used by TUI browser
code AFAIK and it takes the folding state into account.

I think we should just make it folded to make things simple - and
finally get rid of the UI-specific bits in the generic code.

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