Re: [PATCH] tools/perf/metricgroup: Fix printing event names of metric group with multiple events incase of overlapping events

[kajoljain]

On 1/8/20 12:55 PM, Jin, Yao wrote:
> On 1/8/2020 2:58 PM, Kajol Jain wrote:
> > Commit f01642e4912b ("perf metricgroup: Support multiple
> > events for metricgroup") introduced support for multiple events
> > in a metric group. But with the current upstream, metric events
> > names are not printed properly incase we try to run multiple
> > metric groups with overlapping event.
> >
> > With current upstream version, incase of overlapping metric events
> > issue is, we always start our comparision logic from start.
> > So, the events which already matched with some metric group also
> > take part in comparision logic. Because of that when we have overlapping
> > events, we end up matching current metric group event with already 
> > matched
> > one.
> >
> > For example, in skylake machine we have metric event CoreIPC and
> > Instructions. Both of them need 'inst_retired.any' event value.
> > As events in Instructions is subset of events in CoreIPC, they
> > endup in pointing to same 'inst_retired.any' value.
> >
> > In skylake platform:
> >
> > command:# ./perf stat -M CoreIPC,Instructions -C 0 sleep 1
> >
> > Â Performance counter stats for 'CPU(s) 0':
> >
> > ÂÂÂÂÂ 1,254,992,790ÂÂÂÂÂ inst_retired.anyÂÂÂÂÂÂÂÂÂ # 1254992790.0
> > ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ Instructions
> > ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ #ÂÂÂÂÂ 1.3 CoreIPC
> > ÂÂÂÂÂÂÂ 977,172,805ÂÂÂÂÂ cycles
> > ÂÂÂÂÂ 1,254,992,756ÂÂÂÂÂ inst_retired.any
> >
> > ÂÂÂÂÂÂÂ 1.000802596 seconds time elapsed
> >
> > command:# sudo ./perf stat -M UPI,IPC sleep 1
> >
> > ÂÂÂ Performance counter stats for 'sleep 1':
> >
> > ÂÂÂÂÂÂÂÂÂÂÂ 948,650ÂÂÂÂÂ uops_retired.retire_slots
> > ÂÂÂÂÂÂÂÂÂÂÂ 866,182ÂÂÂÂÂ inst_retired.anyÂÂÂÂÂÂÂÂÂ #ÂÂÂÂÂ 0.7 IPC
> > ÂÂÂÂÂÂÂÂÂÂÂ 866,182ÂÂÂÂÂ inst_retired.any
> > ÂÂÂÂÂÂÂÂÂ 1,175,671ÂÂÂÂÂ cpu_clk_unhalted.thread
> >
> > Patch fixes the issue by adding a static variable 'iterator_perf_evlist'
> > to keep track of events which already matched with some group. It points
> > to event in perf_evlist from where next match should start. Because we
> > need to make sure, we match correct set of events belongs to
> > corresponding metric group.
> >
> > With this patch:
> > In skylake platform:
> >
> > command:# ./perf stat -M CoreIPC,Instructions -C 0 sleep 1
> >
> > Â Performance counter stats for 'CPU(s) 0':
> >
> > ÂÂÂÂÂÂÂ 149,481,533ÂÂÂÂÂ inst_retired.anyÂÂÂÂÂÂÂÂÂ #ÂÂÂÂÂ 0.8 CoreIPC
> > ÂÂÂÂÂÂÂ 186,244,218ÂÂÂÂÂ cycles
> > ÂÂÂÂÂÂÂ 149,479,362ÂÂÂÂÂ inst_retired.anyÂÂÂÂÂÂÂÂÂ # 149479362.0
> > ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ Instructions
> >
> > ÂÂÂÂÂÂÂ 1.001655885 seconds time elapsed
> >
> > command:# ./perf stat -M UPI,IPC sleep 1
> > Â Performance counter stats for 'CPU(s) 0':
> >
> > ÂÂÂÂÂÂÂÂ 16,858,849ÂÂÂÂÂ uops_retired.retire_slots #ÂÂÂÂÂ 1.3 UPI
> > ÂÂÂÂÂÂÂÂ 12,529,178ÂÂÂÂÂ inst_retired.any
> > ÂÂÂÂÂÂÂÂ 12,529,558ÂÂÂÂÂ inst_retired.anyÂÂÂÂÂÂÂÂÂ #ÂÂÂÂÂ 0.3 IPC
> > ÂÂÂÂÂÂÂÂ 39,936,071ÂÂÂÂÂ cpu_clk_unhalted.thread
> >
> > ÂÂÂÂÂÂÂ 1.001413978 seconds time elapsed
> >
> >
> > Signed-off-by: Kajol Jain <kjain@xxxxxxxxxxxxx>
> > Cc: Alexander Shishkin <alexander.shishkin@xxxxxxxxxxxxxxx>
> > Cc: Andi Kleen <ak@xxxxxxxxxxxxxxx>
> > Cc: Jiri Olsa <jolsa@xxxxxxxxxx>
> > Cc: Kan Liang <kan.liang@xxxxxxxxxxxxxxx>
> > Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> > Cc: Jin Yao <yao.jin@xxxxxxxxxxxxxxx>
> > Cc: Madhavan Srinivasan <maddy@xxxxxxxxxxxxxxxxxx>
> > Cc: Anju T Sudhakar <anju@xxxxxxxxxxxxxxxxxx>
> > Cc: Ravi Bangoria <ravi.bangoria@xxxxxxxxxxxxx>
> > ---
> > Â tools/perf/util/metricgroup.c | 8 +++++++-
> > Â 1 file changed, 7 insertions(+), 1 deletion(-)
> >
> > diff --git a/tools/perf/util/metricgroup.c 
> > b/tools/perf/util/metricgroup.c
> > index 35e151b8359b..58889b0496fb 100644
> > --- a/tools/perf/util/metricgroup.c
> > +++ b/tools/perf/util/metricgroup.c
> > @@ -90,16 +90,21 @@ struct egroup {
> > ÂÂÂÂÂ const char *metric_unit;
> > Â };
> > Â +static int iterator_perf_evlist;
> > +
> > Â static struct evsel *find_evsel_group(struct evlist *perf_evlist,
> > ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ const char **ids,
> > ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ int idnum,
> > ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ struct evsel **metric_events)
> > Â {
> > ÂÂÂÂÂ struct evsel *ev;
> > -ÂÂÂ int i = 0;
> > +ÂÂÂ int i = 0, j = 0;
> > ÂÂÂÂÂ bool leader_found;
> > Â ÂÂÂÂÂ evlist__for_each_entry (perf_evlist, ev) {
> > +ÂÂÂÂÂÂÂ j++;
> > +ÂÂÂÂÂÂÂ if (j <= iterator_perf_evlist)
> > +ÂÂÂÂÂÂÂÂÂÂÂ continue;
> > ÂÂÂÂÂÂÂÂÂ if (!strcmp(ev->name, ids[i])) {
> > ÂÂÂÂÂÂÂÂÂÂÂÂÂ if (!metric_events[i])
> > ÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂÂ metric_events[i] = ev;
> > @@ -146,6 +151,7 @@ static struct evsel *find_evsel_group(struct 
> > evlist *perf_evlist,
> > ÂÂÂÂÂÂÂÂÂÂÂÂÂ }
> > ÂÂÂÂÂÂÂÂÂ }
> > ÂÂÂÂÂ }
> > +ÂÂÂ iterator_perf_evlist = j;
> > Â ÂÂÂÂÂ return metric_events[0];
> > Â }
>
> Thanks for reporting and fixing this issue.
>
> I just have one question, do we really need a *static variable* to 
> track the matched events? Perhaps using an input parameter?

Hi Jin,

The other way I come up with to solve this issue is, making change in 
perf_evlist itself by adding some flag in event name, to keep track of 
matched events.

As if we change event name itself, next time when we compare it won't 
matched. But in that case we need to remove those flag later. Which will 
increase the

complexity. If you have any suggestions, please let me know.

Thanks,

Kajol

> Thanks
> Jin Yao