Multiple events may have a metric_leader to aggregate into. This happens
for uncore events where, for example, uncore_imc is expanded into
uncore_imc_0, uncore_imc_1, etc. Such events all have the same metric_id
and should aggregate into the first event. The change introducing
metric_ids had a bug where the metric_id was compared to itself,
creating an always true condition. Correct this by comparing the
event in the metric_evlist and the metric_leader.
Fixes: ec5c5b3d2c21 ("perf metric: Encode and use metric-id as qualifier")
Signed-off-by: Ian Rogers <[email protected]>
---
tools/perf/util/metricgroup.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/tools/perf/util/metricgroup.c b/tools/perf/util/metricgroup.c
index 51c99cb08abf..1f1b78b0388f 100644
--- a/tools/perf/util/metricgroup.c
+++ b/tools/perf/util/metricgroup.c
@@ -314,7 +314,7 @@ static int setup_metric_events(struct hashmap *ids,
*/
metric_id = evsel__metric_id(ev);
evlist__for_each_entry_continue(metric_evlist, ev) {
- if (!strcmp(evsel__metric_id(metric_events[i]), metric_id))
+ if (!strcmp(evsel__metric_id(ev), metric_id))
ev->metric_leader = metric_events[i];
}
}
--
2.34.1.703.g22d0c6ccf7-goog
Em Fri, Jan 14, 2022 at 10:28:52PM -0800, Ian Rogers escreveu:
> Multiple events may have a metric_leader to aggregate into. This happens
> for uncore events where, for example, uncore_imc is expanded into
> uncore_imc_0, uncore_imc_1, etc. Such events all have the same metric_id
> and should aggregate into the first event. The change introducing
> metric_ids had a bug where the metric_id was compared to itself,
> creating an always true condition. Correct this by comparing the
> event in the metric_evlist and the metric_leader.
Thanks, applied.
- Arnaldo
> Fixes: ec5c5b3d2c21 ("perf metric: Encode and use metric-id as qualifier")
> Signed-off-by: Ian Rogers <[email protected]>
> ---
> tools/perf/util/metricgroup.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/tools/perf/util/metricgroup.c b/tools/perf/util/metricgroup.c
> index 51c99cb08abf..1f1b78b0388f 100644
> --- a/tools/perf/util/metricgroup.c
> +++ b/tools/perf/util/metricgroup.c
> @@ -314,7 +314,7 @@ static int setup_metric_events(struct hashmap *ids,
> */
> metric_id = evsel__metric_id(ev);
> evlist__for_each_entry_continue(metric_evlist, ev) {
> - if (!strcmp(evsel__metric_id(metric_events[i]), metric_id))
> + if (!strcmp(evsel__metric_id(ev), metric_id))
> ev->metric_leader = metric_events[i];
> }
> }
> --
> 2.34.1.703.g22d0c6ccf7-goog
--
- Arnaldo
On 15/01/2022 06:28, Ian Rogers wrote:
> Multiple events may have a metric_leader to aggregate into. This happens
> for uncore events where, for example, uncore_imc is expanded into
> uncore_imc_0, uncore_imc_1, etc. Such events all have the same metric_id
> and should aggregate into the first event. The change introducing
> metric_ids had a bug where the metric_id was compared to itself,
> creating an always true condition. Correct this by comparing the
> event in the metric_evlist and the metric_leader.
>
> Fixes: ec5c5b3d2c21 ("perf metric: Encode and use metric-id as qualifier")
> Signed-off-by: Ian Rogers<[email protected]>
> ---
Looks ok and I quickly tested it:
Reviewed-by: John Garry <[email protected]>