Re: [PATCH v1 03/22] perf metrics: Don't scale counts going into metrics

From: Xing Zhengjun
Date: Thu Sep 29 2022 - 04:48:47 EST




On 9/28/2022 3:21 PM, Ian Rogers wrote:
Counts are scaled prior to going into saved_value, reverse the scaling
so that metrics don't double scale values.

Signed-off-by: Ian Rogers <irogers@xxxxxxxxxx>
---
tools/perf/util/stat-shadow.c | 9 +++++++--
1 file changed, 7 insertions(+), 2 deletions(-)

diff --git a/tools/perf/util/stat-shadow.c b/tools/perf/util/stat-shadow.c
index 9e1eddeff21b..b5cedd37588f 100644
--- a/tools/perf/util/stat-shadow.c
+++ b/tools/perf/util/stat-shadow.c
@@ -865,11 +865,16 @@ static int prepare_metric(struct evsel **metric_events,
if (!v)
break;
stats = &v->stats;
- scale = 1.0;
+ /*
+ * If an event was scaled during stat gathering, reverse
+ * the scale before computing the metric.
+ */
+ scale = 1.0 / metric_events[i]->scale;
+
This look likes not work for kernel side events like /sys/devices/uncore_imc_*/events/cas_count_read(write).

source_count = evsel__source_count(metric_events[i]);
if (v->metric_other)
- metric_total = v->metric_total;
+ metric_total = v->metric_total * scale;
}
n = strdup(evsel__metric_id(metric_events[i]));
if (!n)

--
Zhengjun Xing