linux-kernel - Re: [PATCH v2 9/9] perf annotate: Use the sample period when calculating the percentage

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]

Message-ID: <b0fa2a64-367d-b313-fe2c-c112a7b50c6e@gmail.com>
Date:   Thu, 20 Jul 2017 02:34:04 +0900
From:   Taeung Song <treeze.taeung@...il.com>
To:     Namhyung Kim <namhyung@...nel.org>
Cc:     Arnaldo Carvalho de Melo <acme@...nel.org>,
        linux-kernel@...r.kernel.org, Jiri Olsa <jolsa@...hat.com>,
        kernel-team@....com
Subject: Re: [PATCH v2 9/9] perf annotate: Use the sample period when
 calculating the percentage



On 07/19/2017 01:23 AM, Namhyung Kim wrote:
> On Fri, Jul 14, 2017 at 02:46:20AM +0900, Taeung Song wrote:
>> Currently the percentages of perf-annotate are calculated
>> with number of samples, not the sample period.
>> So fix it to correspond with perf-report using the sample period
>> for the calculation.
> 
> Not sure someone still wants the old behavior.  Maybe it'd be better
> to honor --show-nr-samples option.
> 
> 

You mean that if users use --show-nr-samples option,
we can calculate the percent values with number of samples, right ?

>>
>> Cc: Namhyung Kim <namhyung@...nel.org>
>> Cc: Jiri Olsa <jolsa@...hat.com>
>> Signed-off-by: Taeung Song <treeze.taeung@...il.com>
>> ---
>>   tools/perf/util/annotate.c | 11 ++++++-----
>>   1 file changed, 6 insertions(+), 5 deletions(-)
>>
>> diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c
>> index d9bdedf..28a6d11 100644
>> --- a/tools/perf/util/annotate.c
>> +++ b/tools/perf/util/annotate.c
>> @@ -988,7 +988,7 @@ double disasm__calc_percent(struct annotation *notes, int evidx, s64 offset,
>>   		if (h->total_samples) {
> 
> It seems checking total_period here is clearer and safer.  But I guess
> it would have same effect 99.9% but still..
> 
> 

I got it.


>>   			sample->nr_samples = hits;
>>   			sample->period = p;
>> -			percent = 100.0 * hits / h->total_samples;
>> +			percent = 100.0 * p / h->total_period;
>>   		}
>>   	}
>>   
>> @@ -1730,8 +1730,9 @@ static int symbol__get_source_line(struct symbol *sym, struct map *map,
>>   	start = map__rip_2objdump(map, sym->start);
>>   
>>   	for (i = 0; i < len; i++) {
>> -		u64 offset, nr_samples;
>> +		u64 offset;
>>   		double percent_max = 0.0;
>> +		struct sym_hist_entry sample;
>>   
>>   		src_line->nr_pcnt = nr_pcnt;
>>   
>> @@ -1739,15 +1740,15 @@ static int symbol__get_source_line(struct symbol *sym, struct map *map,
>>   			double percent = 0.0;
>>   
>>   			h = annotation__histogram(notes, evidx + k);
>> -			nr_samples = h->addr[i].nr_samples;
>> +			sample = h->addr[i];
>>   
>>   			if (h->total_samples)
> 
> Ditto.
> 
> Thanks,
> Namhyung
> 
> 

ok

Thanks,
Taeung

>> -				percent = 100.0 * nr_samples / h->total_samples;
>> +				percent = 100.0 * sample.period / h->total_period;
>>   
>>   			if (percent > percent_max)
>>   				percent_max = percent;
>>   			src_line->samples[k].percent = percent;
>> -			src_line->samples[k].nr = nr_samples;
>> +			src_line->samples[k].nr = sample.nr_samples;
>>   		}
>>   
>>   		if (percent_max <= 0.5)
>> -- 
>> 2.7.4
>>