linux-kernel - Re: [BUG] perf report --pid not reporting correctly

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: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Date:   Fri, 23 Sep 2016 12:31:33 +0900
From:   Namhyung Kim <namhyung@...nel.org>
To:     Jiri Olsa <jolsa@...hat.com>
Cc:     Stephane Eranian <eranian@...gle.com>,
        Arnaldo Carvalho de Melo <acme@...hat.com>,
        LKML <linux-kernel@...r.kernel.org>,
        Peter Zijlstra <peterz@...radead.org>,
        "mingo@...e.hu" <mingo@...e.hu>,
        "ak@...ux.intel.com" <ak@...ux.intel.com>,
        Vince Weaver <vincent.weaver@...ne.edu>,
        Arnaldo Carvalho de Melo <acme@...nel.org>
Subject: Re: [BUG] perf report --pid not reporting correctly

Hi Jiri,

On Thu, Sep 22, 2016 at 10:36 PM, Jiri Olsa <jolsa@...hat.com> wrote:
> On Thu, Sep 22, 2016 at 10:34:57AM +0900, Namhyung Kim wrote:
>
> SNIP
>
>>   $ perf report -s pid,cpu --stdio --hierarchy
>>   #
>>   #    Overhead  Pid:Command / CPU
>>   # ...........  .......................
>>   #
>>       31.21%     0:swapper
>>          12.16%     000
>>           3.09%     001
>>           2.76%     002
>>           2.23%     003
>>           1.65%     007
>>           1.65%     008
>>           1.52%     009
>>           1.51%     006
>>           1.46%     004
>>           1.34%     005
>>           0.94%     010
>>           0.90%     011
>>       19.15%     8618:getmail
>>          ...
>>
>>
>>
>>
>>
>> -----------------------8<-------------------------------
>> diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h
>> index 8d363d5e65a2..42b1bfd29ef8 100644
>> --- a/tools/perf/util/event.h
>> +++ b/tools/perf/util/event.h
>> @@ -262,6 +262,7 @@ enum auxtrace_error_type {
>>   */
>>  struct events_stats {
>>       u64 total_period;
>> +     u64 total_early_filtered_period;
>
> looks good, but why the word early? wouldn't total_filtered_period be just fine?

I worried about it'd be confused to (wrongly) match to
"total_non_filtered_period".  One might think that the sum of filtered
+ non_filtered period should be the total period, but it's not.

I named it 'early' since it's filtered at sample level before creating
hist entry.  Later we can also filter on hist entries
(hists__filter_by_XXX) but it cannot affect the ones filtered earlier.

Thanks,
Namhyung


>
> thanks,
> jirka
>
>>       u64 total_non_filtered_period;
>>       u64 total_lost;
>>       u64 total_lost_samples;
>> diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c
>> index 37a08f20730a..c7045411cce2 100644
>> --- a/tools/perf/util/hist.c
>> +++ b/tools/perf/util/hist.c
>> @@ -1017,12 +1017,21 @@ int hist_entry_iter__add(struct hist_entry_iter *iter, struct addr_location *al,
>>                        int max_stack_depth, void *arg)
>>  {
>>       int err, err2;
>> +     struct hists *hists = evsel__hists(iter->evsel);
>>
>>       err = sample__resolve_callchain(iter->sample, &callchain_cursor, &iter->parent,
>>                                       iter->evsel, al, max_stack_depth);
>>       if (err)
>>               return err;
>>
>> +     if (symbol__parent_filter(iter->parent))
>> +             al->filtered |= symbol__parent_filter(iter->parent);
>> +
>> +     if (al->filtered) {
>> +             hists->stats.total_early_filtered_period += iter->sample->period;
>
> SNIP
>



-- 
Thanks,
Namhyung