Skip to contentSkip to conversation

perf parse-events: Do not break up AUX event group

PATCH Applied Adrian Hunter 3 participants 7 messages May 8, 2023 – May 10, 2023 perf events

lore
Share & tools
lore

Adrian Hunter

author reviewer
#0 lore raw

v2 cover perf parse-events: Do not break up AUX event group

Hi

Here is a fix needed in v6.4, but dependent on Ian's patch "perf evsel: Modify group pmu name for software events", refer:

https://lore.kernel.org/lkml/[email protected]/

There is also a test amendment to test the AUX event group fix for Intel PT.

Changes in V2:

Add an Intel PT test amendment for the AUX event group fix Change AUX event group fix as suggested by Ian

Adrian Hunter (2):
perf test: test_intel_pt.sh: Test sample mode with named PMU perf parse-events: Do not break up AUX event group

 tools/perf/tests/shell/test_intel_pt.sh | 7 +++++++
 tools/perf/util/evsel.c                 | 9 +++------
 2 files changed, 10 insertions(+), 6 deletions(-)

Regards
Adrian

author reviewer
#1 lore raw

PATCH v2 1/2 perf test: test_intel_pt.sh: Test sample mode with event with PMU name

br_misp_retired.all_branches is supported on processors that support Intel PT, so use it to test sample mode with an event that has been given a PMU name.

Please note, the test fails prior to the fix "perf parse-events: Do not break up AUX event group".

---
 tools/perf/tests/shell/test_intel_pt.sh | 7 +++++++
 1 file changed, 7 insertions(+)
tools/perf/tests/shell/test_intel_pt.sh +7 -0
index 4ddb17cb83c5..3a8b9bffa022 100755
@@ -506,6 +506,13 @@ test_sample()
 		echo "perf record failed with --aux-sample"
 		return 1
 	fi
+	# Check with event with PMU name
+	if perf_record_no_decode -o "${perfdatafile}" -e br_misp_retired.all_branches:u uname ; then
+		if ! perf_record_no_decode -o "${perfdatafile}" -e '{intel_pt//,br_misp_retired.all_branches/aux-sample-size=8192/}:u' uname ; then
+			echo "perf record failed with --aux-sample-size"
+			return 1
+		fi
+	fi
 	echo OK
 	return 0
 }
Signature · 1 line
2.34.1
author reviewer
#2 lore raw

PATCH v2 2/2 perf parse-events: Do not break up AUX event group

In reply to Adrian Hunter

Do not assume which events may have a PMU name, allowing the logic to keep an AUX event group together.

Example:

Before:

    $ perf record --no-bpf-event -c 10 -e '{intel_pt//,tlb_flush.stlb_any/aux-sample-size=8192/pp}:u' -- sleep 0.1
    WARNING: events were regrouped to match PMUs
    Cannot add AUX area sampling to a group leader

After:

    $ perf record --no-bpf-event -c 10 -e '{intel_pt//,tlb_flush.stlb_any/aux-sample-size=8192/pp}:u' -- sleep 0.1
    [ perf record: Woken up 1 times to write data ]
    [ perf record: Captured and wrote 0.078 MB perf.data ]
    $ perf script -F-dso,+addr | grep -C5 tlb_flush.stlb_any | head -11
    sleep 20444 [003]  7939.510243:  1  branches:uH:  7f5350cc82a2 dl_main+0x9a2 => 7f5350cb38f0 _dl_add_to_namespace_list+0x0
    sleep 20444 [003]  7939.510243:  1  branches:uH:  7f5350cb3908 _dl_add_to_namespace_list+0x18 => 7f5350cbb080 rtld_mutex_dummy+0x0
    sleep 20444 [003]  7939.510243:  1  branches:uH:  7f5350cc8350 dl_main+0xa50 => 0 [unknown]
    sleep 20444 [003]  7939.510244:  1  branches:uH:  7f5350cc83ca dl_main+0xaca => 7f5350caeb60 _dl_process_pt_gnu_property+0x0
    sleep 20444 [003]  7939.510245:  1  branches:uH:  7f5350caeb60 _dl_process_pt_gnu_property+0x0 => 0 [unknown]
    sleep 20444  7939.510245:       10 tlb_flush.stlb_any/aux-sample-size=8192/pp: 0 7f5350caeb60 _dl_process_pt_gnu_property+0x0
    sleep 20444 [003]  7939.510254:  1  branches:uH:  7f5350cc87fe dl_main+0xefe => 7f5350ccd240 strcmp+0x0
    sleep 20444 [003]  7939.510254:  1  branches:uH:  7f5350cc8862 dl_main+0xf62 => 0 [unknown]
    sleep 20444 [003]  7939.510255:  1  branches:uH:  7f5350cc9cdc dl_main+0x23dc => 0 [unknown]
    sleep 20444 [003]  7939.510257:  1  branches:uH:  7f5350cc89f6 dl_main+0x10f6 => 7f5350cb9530 _dl_setup_hash+0x0
    sleep 20444 [003]  7939.510257:  1  branches:uH:  7f5350cc8a2d dl_main+0x112d => 7f5350cb3990 _dl_new_object+0x0
Fixes 347c2f0a0988 ("perf parse-events: Sort and group parsed events")
---
 tools/perf/util/evsel.c | 9 +++------
 1 file changed, 3 insertions(+), 6 deletions(-)
tools/perf/util/evsel.c +3 -6
index e18752fd9245..9fdcde5471df 100644
@@ -828,11 +828,9 @@ bool evsel__name_is(struct evsel *evsel, const char *name)
 
 const char *evsel__group_pmu_name(const struct evsel *evsel)
 {
-	struct evsel *leader, *pos;
+	struct evsel *leader = evsel__leader(evsel);
+	struct evsel *pos;
 
-	/* If the pmu_name is set use it. pmu_name isn't set for CPU and software events. */
-	if (evsel->pmu_name)
-		return evsel->pmu_name;
 	/*
 	 * Software events may be in a group with other uncore PMU events. Use
 	 * the pmu_name of the first non-software event to avoid breaking the
@@ -841,7 +839,6 @@ const char *evsel__group_pmu_name(const struct evsel *evsel)
 	 * Aux event leaders, like intel_pt, expect a group with events from
 	 * other PMUs, so substitute the AUX event's PMU in this case.
 	 */
-	leader  = evsel__leader(evsel);
 	if (evsel->core.attr.type == PERF_TYPE_SOFTWARE || evsel__is_aux_event(leader)) {
 		/* Starting with the leader, find the first event with a named PMU. */
 		for_each_group_evsel(pos, leader) {
@@ -850,7 +847,7 @@ const char *evsel__group_pmu_name(const struct evsel *evsel)
 		}
 	}
 
-	return "cpu";
+	return evsel->pmu_name ?: "cpu";
 }
 
 const char *evsel__metric_id(const struct evsel *evsel)
Signature · 1 line
2.34.1
reviewer Tested-by on v2 2/2
#3 lore raw
Quoted 30 lines · Adrian Hunter

Do not assume which events may have a PMU name, allowing the logic to keep an AUX event group together.

Example:

Before:

    $ perf record --no-bpf-event -c 10 -e '{intel_pt//,tlb_flush.stlb_any/aux-sample-size=8192/pp}:u' -- sleep 0.1
    WARNING: events were regrouped to match PMUs
    Cannot add AUX area sampling to a group leader

After:

    $ perf record --no-bpf-event -c 10 -e '{intel_pt//,tlb_flush.stlb_any/aux-sample-size=8192/pp}:u' -- sleep 0.1
    [ perf record: Woken up 1 times to write data ]
    [ perf record: Captured and wrote 0.078 MB perf.data ]
    $ perf script -F-dso,+addr | grep -C5 tlb_flush.stlb_any | head -11
    sleep 20444 [003]  7939.510243:  1  branches:uH:  7f5350cc82a2 dl_main+0x9a2 => 7f5350cb38f0 _dl_add_to_namespace_list+0x0
    sleep 20444 [003]  7939.510243:  1  branches:uH:  7f5350cb3908 _dl_add_to_namespace_list+0x18 => 7f5350cbb080 rtld_mutex_dummy+0x0
    sleep 20444 [003]  7939.510243:  1  branches:uH:  7f5350cc8350 dl_main+0xa50 => 0 [unknown]
    sleep 20444 [003]  7939.510244:  1  branches:uH:  7f5350cc83ca dl_main+0xaca => 7f5350caeb60 _dl_process_pt_gnu_property+0x0
    sleep 20444 [003]  7939.510245:  1  branches:uH:  7f5350caeb60 _dl_process_pt_gnu_property+0x0 => 0 [unknown]
    sleep 20444  7939.510245:       10 tlb_flush.stlb_any/aux-sample-size=8192/pp: 0 7f5350caeb60 _dl_process_pt_gnu_property+0x0
    sleep 20444 [003]  7939.510254:  1  branches:uH:  7f5350cc87fe dl_main+0xefe => 7f5350ccd240 strcmp+0x0
    sleep 20444 [003]  7939.510254:  1  branches:uH:  7f5350cc8862 dl_main+0xf62 => 0 [unknown]
    sleep 20444 [003]  7939.510255:  1  branches:uH:  7f5350cc9cdc dl_main+0x23dc => 0 [unknown]
    sleep 20444 [003]  7939.510257:  1  branches:uH:  7f5350cc89f6 dl_main+0x10f6 => 7f5350cb9530 _dl_setup_hash+0x0
    sleep 20444 [003]  7939.510257:  1  branches:uH:  7f5350cc8a2d dl_main+0x112d => 7f5350cb3990 _dl_new_object+0x0
Quoted 3 lines
Fixes 347c2f0a0988 ("perf parse-events: Sort and group parsed events")

Thanks!
Ian

Quoted 42 lines
---
 tools/perf/util/evsel.c | 9 +++------
 1 file changed, 3 insertions(+), 6 deletions(-)
tools/perf/util/evsel.c +3 -7
index e18752fd9245..9fdcde5471df 100644
@@ -828,11 +828,9 @@ bool evsel__name_is(struct evsel *evsel, const char *name)
 
 const char *evsel__group_pmu_name(const struct evsel *evsel)
 {
-       struct evsel *leader, *pos;
+       struct evsel *leader = evsel__leader(evsel);
+       struct evsel *pos;
 
-       /* If the pmu_name is set use it. pmu_name isn't set for CPU and software events. */
-       if (evsel->pmu_name)
-               return evsel->pmu_name;
        /*
         * Software events may be in a group with other uncore PMU events. Use
         * the pmu_name of the first non-software event to avoid breaking the
@@ -841,7 +839,6 @@ const char *evsel__group_pmu_name(const struct evsel *evsel)
         * Aux event leaders, like intel_pt, expect a group with events from
         * other PMUs, so substitute the AUX event's PMU in this case.
         */
-       leader  = evsel__leader(evsel);
        if (evsel->core.attr.type == PERF_TYPE_SOFTWARE || evsel__is_aux_event(leader)) {
                /* Starting with the leader, find the first event with a named PMU. */
                for_each_group_evsel(pos, leader) {
@@ -850,7 +847,7 @@ const char *evsel__group_pmu_name(const struct evsel *evsel)
                }
        }
 
-       return "cpu";
+       return evsel->pmu_name ?: "cpu";
 }
 
 const char *evsel__metric_id(const struct evsel *evsel)
--

2.34.1

#4 lore raw
In reply to Adrian Hunter

Em Mon, May 08, 2023 at 12:39:50PM +0300, Adrian Hunter escreveu:

Quoted 9 lines

Hi

Here is a fix needed in v6.4, but dependent on Ian's patch "perf evsel: Modify group pmu name for software events", refer:

https://lore.kernel.org/lkml/[email protected]/

There is also a test amendment to test the AUX event group fix for Intel PT.

Ok, so I think I can add your Tested-by for that "perf evsel: Modify group pmu name for software events" patch, ok?

- Arnaldo

Quoted 14 lines

Changes in V2:

Add an Intel PT test amendment for the AUX event group fix Change AUX event group fix as suggested by Ian

Adrian Hunter (2):
perf test: test_intel_pt.sh: Test sample mode with named PMU perf parse-events: Do not break up AUX event group

 tools/perf/tests/shell/test_intel_pt.sh | 7 +++++++
 tools/perf/util/evsel.c                 | 9 +++------
 2 files changed, 10 insertions(+), 6 deletions(-)
author reviewer
#5 lore raw
Quoted 6 lines · Arnaldo Carvalho de Melo

Em Mon, May 08, 2023 at 12:39:50PM +0300, Adrian Hunter escreveu:

Quoted 5 lines

Hi

Here is a fix needed in v6.4, but dependent on Ian's patch "perf evsel: Modify group pmu name for software events", refer:

Quoted 7 lines
Quoted 4 lines

https://lore.kernel.org/lkml/[email protected]/

There is also a test amendment to test the AUX event group fix for Intel PT.

Ok, so I think I can add your Tested-by for that "perf evsel: Modify group pmu name for software events" patch, ok?

Yes, thank you!

Quoted 17 lines

- Arnaldo

Quoted 14 lines

Changes in V2:

Add an Intel PT test amendment for the AUX event group fix Change AUX event group fix as suggested by Ian

Adrian Hunter (2):
perf test: test_intel_pt.sh: Test sample mode with named PMU perf parse-events: Do not break up AUX event group

 tools/perf/tests/shell/test_intel_pt.sh | 7 +++++++
 tools/perf/util/evsel.c                 | 9 +++------
 2 files changed, 10 insertions(+), 6 deletions(-)
maintainer Applied
#6 lore raw
In reply to Adrian Hunter

Em Mon, May 08, 2023 at 12:39:50PM +0300, Adrian Hunter escreveu:

Quoted 4 lines

Hi

Here is a fix needed in v6.4, but dependent on Ian's patch "perf evsel: Modify group pmu name for software events", refer:

Thanks, applied to perf-core, for v6.4.

- Arnaldo

Quoted 23 lines

https://lore.kernel.org/lkml/[email protected]/

There is also a test amendment to test the AUX event group fix for Intel PT.

Changes in V2:

Add an Intel PT test amendment for the AUX event group fix Change AUX event group fix as suggested by Ian

Adrian Hunter (2):
perf test: test_intel_pt.sh: Test sample mode with named PMU perf parse-events: Do not break up AUX event group

 tools/perf/tests/shell/test_intel_pt.sh | 7 +++++++
 tools/perf/util/evsel.c                 | 9 +++------
 2 files changed, 10 insertions(+), 6 deletions(-)

Regards
Adrian

Signature · 1 line
- Arnaldo