Re: [PATCH v3] perf stat: Show percore counts in per CPU output

[Ravi Bangoria <ravi.bangoria@linux.ibm.com>]

On 2/13/20 8:40 PM, Jin, Yao wrote:
> 
> 
> On 2/13/2020 9:20 PM, Ravi Bangoria wrote:
>> Hi Jin,
>>
>> On 2/13/20 12:45 PM, Jin Yao wrote:
>>> With this patch, for example,
>>>
>>>   # perf stat -e cpu/event=cpu-cycles,percore/ -a -A --percore-show-thread  -- sleep 1
>>>
>>>    Performance counter stats for 'system wide':
>>>
>>>   CPU0               2,453,061      cpu/event=cpu-cycles,percore/
>>>   CPU1               1,823,921      cpu/event=cpu-cycles,percore/
>>>   CPU2               1,383,166      cpu/event=cpu-cycles,percore/
>>>   CPU3               1,102,652      cpu/event=cpu-cycles,percore/
>>>   CPU4               2,453,061      cpu/event=cpu-cycles,percore/
>>>   CPU5               1,823,921      cpu/event=cpu-cycles,percore/
>>>   CPU6               1,383,166      cpu/event=cpu-cycles,percore/
>>>   CPU7               1,102,652      cpu/event=cpu-cycles,percore/
>>>
>>> We can see counts are duplicated in CPU pairs
>>> (CPU0/CPU4, CPU1/CPU5, CPU2/CPU6, CPU3/CPU7).
>>>
>>
>> I was trying this patch and I am getting bit weird results when any cpu
>> is offline. Ex,
>>
>>    $ lscpu | grep list
>>    On-line CPU(s) list:             0-4,6,7
>>    Off-line CPU(s) list:            5
>>
>>    $ sudo ./perf stat -e cpu/event=cpu-cycles,percore/ -a -A --percore-show-thread -vv -- sleep 1
>>      ...
>>    cpu/event=cpu-cycles,percore/: 0: 23746491 1001189836 1001189836
>>    cpu/event=cpu-cycles,percore/: 1: 19802666 1001291299 1001291299
>>    cpu/event=cpu-cycles,percore/: 2: 24211983 1001394318 1001394318
>>    cpu/event=cpu-cycles,percore/: 3: 54051396 1001516816 1001516816
>>    cpu/event=cpu-cycles,percore/: 4: 6378825 1001064048 1001064048
>>    cpu/event=cpu-cycles,percore/: 5: 21299840 1001166297 1001166297
>>    cpu/event=cpu-cycles,percore/: 6: 13075410 1001274535 1001274535
>>     Performance counter stats for 'system wide':
>>    CPU0              30,125,316      cpu/event=cpu-cycles,percore/
>>    CPU1              19,802,666      cpu/event=cpu-cycles,percore/
>>    CPU2              45,511,823      cpu/event=cpu-cycles,percore/
>>    CPU3              67,126,806      cpu/event=cpu-cycles,percore/
>>    CPU4              30,125,316      cpu/event=cpu-cycles,percore/
>>    CPU7              67,126,806      cpu/event=cpu-cycles,percore/
>>    CPU0              30,125,316      cpu/event=cpu-cycles,percore/
>>           1.001918764 seconds time elapsed
>>
>> I see proper result without --percore-show-thread:
>>
>>    $ sudo ./perf stat -e cpu/event=cpu-cycles,percore/ -a -A -vv -- sleep 1
>>      ...
>>    cpu/event=cpu-cycles,percore/: 0: 11676414 1001190709 1001190709
>>    cpu/event=cpu-cycles,percore/: 1: 39119617 1001291459 1001291459
>>    cpu/event=cpu-cycles,percore/: 2: 41821512 1001391158 1001391158
>>    cpu/event=cpu-cycles,percore/: 3: 46853730 1001492799 1001492799
>>    cpu/event=cpu-cycles,percore/: 4: 14448274 1001095948 1001095948
>>    cpu/event=cpu-cycles,percore/: 5: 42238217 1001191187 1001191187
>>    cpu/event=cpu-cycles,percore/: 6: 33129641 1001292072 1001292072
>>     Performance counter stats for 'system wide':
>>    S0-D0-C0             26,124,688      cpu/event=cpu-cycles,percore/
>>    S0-D0-C1             39,119,617      cpu/event=cpu-cycles,percore/
>>    S0-D0-C2             84,059,729      cpu/event=cpu-cycles,percore/
>>    S0-D0-C3             79,983,371      cpu/event=cpu-cycles,percore/
>>           1.001961563 seconds time elapsed
>>
>> [...]
>>
> 
> Thanks so much for reporting this issue!
> 
> It looks I should use the cpu idx in print_percore_thread. I can't use the cpu value. I have a fix:
> 
> diff --git a/tools/perf/util/stat-display.c b/tools/perf/util/stat-display.c
> index 7eb3643a97ae..d89cb0da90f8 100644
> --- a/tools/perf/util/stat-display.c
> +++ b/tools/perf/util/stat-display.c
> @@ -1149,13 +1149,11 @@ static void print_footer(struct perf_stat_config *config)
>   static void print_percore_thread(struct perf_stat_config *config,
>                                   struct evsel *counter, char *prefix)
>   {
> -       int cpu, s, s2, id;
> +       int s, s2, id;
>          bool first = true;
> 
>          for (int i = 0; i < perf_evsel__nr_cpus(counter); i++) {
> -               cpu = perf_cpu_map__cpu(evsel__cpus(counter), i);
> -               s2 = config->aggr_get_id(config, evsel__cpus(counter), cpu);
> -
> +               s2 = config->aggr_get_id(config, evsel__cpus(counter), i);
>                  for (s = 0; s < config->aggr_map->nr; s++) {
>                          id = config->aggr_map->map[s];
>                          if (s2 == id)
> @@ -1164,7 +1162,7 @@ static void print_percore_thread(struct perf_stat_config *config,
> 
>                  print_counter_aggrdata(config, counter, s,
>                                         prefix, false,
> -                                      &first, cpu);
> +                                      &first, i);
>          }
>   }

LGTM.

Tested-by: Ravi Bangoria <ravi.bangoria@linux.ibm.com>

Ravi