[PATCH] perf: qcom_l2: fix column exclusion check

[Neil Leeder]

The check for column exclusion did not verify that the event being
checked was an L2 event, and not a software event.
Software events should not be checked for column exclusion.
This resulted in a group with both software and L2 events sometimes
incorrectly rejecting the L2 event for column exclusion and
not counting it.

Add a check for PMU type before applying column exclusion logic.

Signed-off-by: Neil Leeder <***@codeaurora.org>
---
drivers/perf/qcom_l2_pmu.c | 2 ++
1 file changed, 2 insertions(+)

diff --git a/drivers/perf/qcom_l2_pmu.c b/drivers/perf/qcom_l2_pmu.c
index c259848..b242cce 100644
--- a/drivers/perf/qcom_l2_pmu.c
+++ b/drivers/perf/qcom_l2_pmu.c
@@ -546,6 +546,7 @@ static int l2_cache_event_init(struct perf_event *event)
}

if ((event != event->group_leader) &&
+ !is_software_event(event->group_leader) &&
(L2_EVT_GROUP(event->group_leader->attr.config) ==
L2_EVT_GROUP(event->attr.config))) {
dev_dbg_ratelimited(&l2cache_pmu->pdev->dev,
@@ -558,6 +559,7 @@ static int l2_cache_event_init(struct perf_event *event)
list_for_each_entry(sibling, &event->group_leader->sibling_list,
group_entry) {
if ((sibling != event) &&
+ !is_software_event(sibling) &&
(L2_EVT_GROUP(sibling->attr.config) ==
L2_EVT_GROUP(event->attr.config))) {
dev_dbg_ratelimited(&l2cache_pmu->pdev->dev,

[Mark Rutland]

Post by Neil Leeder
The check for column exclusion did not verify that the event being
checked was an L2 event, and not a software event.
Software events should not be checked for column exclusion.
This resulted in a group with both software and L2 events sometimes
incorrectly rejecting the L2 event for column exclusion and
not counting it.
Add a check for PMU type before applying column exclusion logic.

This looks correct, so:

Acked-by: Mark Rutland <***@arm.com>

Should this have:

Fixes: 21bdbb7102edeaeb ("perf: add qcom l2 cache perf events driver")

... ?

Post by Neil Leeder
---
drivers/perf/qcom_l2_pmu.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/perf/qcom_l2_pmu.c b/drivers/perf/qcom_l2_pmu.c
index c259848..b242cce 100644
--- a/drivers/perf/qcom_l2_pmu.c
+++ b/drivers/perf/qcom_l2_pmu.c
@@ -546,6 +546,7 @@ static int l2_cache_event_init(struct perf_event *event)
}
if ((event != event->group_leader) &&
+ !is_software_event(event->group_leader) &&
(L2_EVT_GROUP(event->group_leader->attr.config) ==
L2_EVT_GROUP(event->attr.config))) {
dev_dbg_ratelimited(&l2cache_pmu->pdev->dev,
@@ -558,6 +559,7 @@ static int l2_cache_event_init(struct perf_event *event)
list_for_each_entry(sibling, &event->group_leader->sibling_list,
group_entry) {
if ((sibling != event) &&
+ !is_software_event(sibling) &&
(L2_EVT_GROUP(sibling->attr.config) ==
L2_EVT_GROUP(event->attr.config))) {
dev_dbg_ratelimited(&l2cache_pmu->pdev->dev,

It's unfortunate that we duplicate the checks for the leader and
siblings, but that's not a new problem, and we can fix that in a
follow-up patch.

Thanks,
Mark.

[Leeder, Neil]

Post by Mark Rutland

Post by Neil Leeder
The check for column exclusion did not verify that the event being
checked was an L2 event, and not a software event.
Software events should not be checked for column exclusion.
This resulted in a group with both software and L2 events sometimes
incorrectly rejecting the L2 event for column exclusion and
not counting it.
Add a check for PMU type before applying column exclusion logic.

Fixes: 21bdbb7102edeaeb ("perf: add qcom l2 cache perf events driver")
... ?

Thanks. I'll re-post with the Fixes tag.

Neil