[PATCH v2 0/4] perf tool: Skip invalid hybrid PMU

[PATCH v2 0/4] perf tool: Skip invalid hybrid PMU

[Jin Yao]

On hybrid platform, such as Alderlake, if atom CPUs are offlined,
the kernel still exports the sysfs path '/sys/devices/cpu_atom/' for
'cpu_atom' pmu but the file '/sys/devices/cpu_atom/cpus' is empty,
which indicates this is an invalid pmu. So we need to check and skip
the invalid hybrid pmu.

Also we need to update some perf test cases for core-only system.

Jin Yao (4):
  perf pmu: Skip invalid hybrid pmu
  perf tests: Fix 'Parse event definition strings' on core-only system
  perf tests: Fix 'Roundtrip evsel->name' on core-only system
  perf tests: Fix 'Convert perf time to TSC' on core-only system

 tools/perf/tests/evsel-roundtrip-name.c |  3 ++-
 tools/perf/tests/parse-events.c         | 16 ++++++++++------
 tools/perf/tests/perf-time-to-tsc.c     |  3 ++-
 tools/perf/util/pmu.c                   |  9 ++++++++-
 4 files changed, 22 insertions(+), 9 deletions(-)

-- 
2.17.1

[PATCH v2 1/4] perf pmu: Skip invalid hybrid pmu

[Jin Yao]

On hybrid platform, such as Alderlake, if atom CPUs are offlined,
the kernel still exports the sysfs path '/sys/devices/cpu_atom/' for
'cpu_atom' pmu but the file '/sys/devices/cpu_atom/cpus' is empty,
which indicates this is an invalid pmu.

Need to check and skip the invalid hybrid pmu.

Before:

  # perf list
  ...
  branch-instructions OR cpu_atom/branch-instructions/ [Kernel PMU event]
  branch-instructions OR cpu_core/branch-instructions/ [Kernel PMU event]
  branch-misses OR cpu_atom/branch-misses/           [Kernel PMU event]
  branch-misses OR cpu_core/branch-misses/           [Kernel PMU event]
  bus-cycles OR cpu_atom/bus-cycles/                 [Kernel PMU event]
  bus-cycles OR cpu_core/bus-cycles/                 [Kernel PMU event]
  ...

The cpu_atom events are still displayed even if atom CPUs are offlined.

After:

  # perf list
  ...
  branch-instructions OR cpu_core/branch-instructions/ [Kernel PMU event]
  branch-misses OR cpu_core/branch-misses/           [Kernel PMU event]
  bus-cycles OR cpu_core/bus-cycles/                 [Kernel PMU event]
  ...

Now only cpu_core events are displayed.

Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
---
 tools/perf/util/pmu.c | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/tools/perf/util/pmu.c b/tools/perf/util/pmu.c
index 88c8ecdc60b0..cb20a9f69a04 100644
--- a/tools/perf/util/pmu.c
+++ b/tools/perf/util/pmu.c
@@ -927,6 +927,13 @@ static struct perf_pmu *pmu_lookup(const char *name)
 	LIST_HEAD(format);
 	LIST_HEAD(aliases);
 	__u32 type;
+	bool is_hybrid = perf_pmu__hybrid_mounted(name);
+
+	/*
+	 * Check pmu name for hybrid and the pmu may be invalid in sysfs
+	 */
+	if (!strncmp(name, "cpu_", 4) && !is_hybrid)
+		return NULL;
 
 	/*
 	 * The pmu data we store & need consists of the pmu
@@ -955,7 +962,7 @@ static struct perf_pmu *pmu_lookup(const char *name)
 	pmu->is_uncore = pmu_is_uncore(name);
 	if (pmu->is_uncore)
 		pmu->id = pmu_id(name);
-	pmu->is_hybrid = perf_pmu__hybrid_mounted(name);
+	pmu->is_hybrid = is_hybrid;
 	pmu->max_precise = pmu_max_precise(name);
 	pmu_add_cpu_aliases(&aliases, pmu);
 	pmu_add_sys_aliases(&aliases, pmu);
-- 
2.17.1

[PATCH v2 2/4] perf tests: Fix 'Parse event definition strings' on core-only system

[Jin Yao]

If the atom CPUs are offlined, the 'cpu_atom' is not valid.
We don't need the test case for 'cpu_atom'.

Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
---
 tools/perf/tests/parse-events.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)

diff --git a/tools/perf/tests/parse-events.c b/tools/perf/tests/parse-events.c
index 0f113b2b36a3..3ca515fb141f 100644
--- a/tools/perf/tests/parse-events.c
+++ b/tools/perf/tests/parse-events.c
@@ -6,6 +6,7 @@
 #include "tests.h"
 #include "debug.h"
 #include "pmu.h"
+#include "pmu-hybrid.h"
 #include <dirent.h>
 #include <errno.h>
 #include <sys/types.h>
@@ -1596,6 +1597,13 @@ static int test__hybrid_raw1(struct evlist *evlist)
 {
 	struct evsel *evsel = evlist__first(evlist);
 
+	if (!perf_pmu__hybrid_mounted("cpu_atom")) {
+		TEST_ASSERT_VAL("wrong number of entries", 1 == evlist->core.nr_entries);
+		TEST_ASSERT_VAL("wrong type", PERF_TYPE_RAW == evsel->core.attr.type);
+		TEST_ASSERT_VAL("wrong config", 0x1a == evsel->core.attr.config);
+		return 0;
+	}
+
 	TEST_ASSERT_VAL("wrong number of entries", 2 == evlist->core.nr_entries);
 	TEST_ASSERT_VAL("wrong type", PERF_TYPE_RAW == evsel->core.attr.type);
 	TEST_ASSERT_VAL("wrong config", 0x1a == evsel->core.attr.config);
@@ -1620,13 +1628,9 @@ static int test__hybrid_cache_event(struct evlist *evlist)
 {
 	struct evsel *evsel = evlist__first(evlist);
 
-	TEST_ASSERT_VAL("wrong number of entries", 2 == evlist->core.nr_entries);
+	TEST_ASSERT_VAL("wrong number of entries", 1 == evlist->core.nr_entries);
 	TEST_ASSERT_VAL("wrong type", PERF_TYPE_HW_CACHE == evsel->core.attr.type);
 	TEST_ASSERT_VAL("wrong config", 0x2 == (evsel->core.attr.config & 0xffffffff));
-
-	evsel = evsel__next(evsel);
-	TEST_ASSERT_VAL("wrong type", PERF_TYPE_HW_CACHE == evsel->core.attr.type);
-	TEST_ASSERT_VAL("wrong config", 0x10002 == (evsel->core.attr.config & 0xffffffff));
 	return 0;
 }
 
@@ -2028,7 +2032,7 @@ static struct evlist_test test__hybrid_events[] = {
 		.id    = 7,
 	},
 	{
-		.name  = "cpu_core/LLC-loads/,cpu_atom/LLC-load-misses/",
+		.name  = "cpu_core/LLC-loads/",
 		.check = test__hybrid_cache_event,
 		.id    = 8,
 	},
-- 
2.17.1

[PATCH v2 3/4] perf tests: Fix 'Roundtrip evsel->name' on core-only system

[Jin Yao]

If the atom CPUs are offlined, the 'cpu_atom' is not valid.
Perf will not create two events for one hw event, so the
evsel->idx doesn't need to be divided by 2 before comparing.

Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
---
 tools/perf/tests/evsel-roundtrip-name.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/tools/perf/tests/evsel-roundtrip-name.c b/tools/perf/tests/evsel-roundtrip-name.c
index b74cf80d1f10..26c49dc3e01e 100644
--- a/tools/perf/tests/evsel-roundtrip-name.c
+++ b/tools/perf/tests/evsel-roundtrip-name.c
@@ -5,6 +5,7 @@
 #include "tests.h"
 #include "debug.h"
 #include "pmu.h"
+#include "pmu-hybrid.h"
 #include <errno.h>
 #include <linux/kernel.h>
 
@@ -102,7 +103,7 @@ int test__perf_evsel__roundtrip_name_test(struct test *test __maybe_unused, int
 {
 	int err = 0, ret = 0;
 
-	if (perf_pmu__has_hybrid())
+	if (perf_pmu__has_hybrid() && perf_pmu__hybrid_mounted("cpu_atom"))
 		return perf_evsel__name_array_test(evsel__hw_names, 2);
 
 	err = perf_evsel__name_array_test(evsel__hw_names, 1);
-- 
2.17.1

[PATCH v2 4/4] perf tests: Fix 'Convert perf time to TSC' on core-only system

[Jin Yao]

If the atom CPUs are offlined, the 'cpu_atom' is not valid.
We don't need the test case for 'cpu_atom'.

Signed-off-by: Jin Yao <yao.jin@linux.intel.com>
---
 tools/perf/tests/perf-time-to-tsc.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/tools/perf/tests/perf-time-to-tsc.c b/tools/perf/tests/perf-time-to-tsc.c
index 85d75b9b25a1..7c56bc1f4cff 100644
--- a/tools/perf/tests/perf-time-to-tsc.c
+++ b/tools/perf/tests/perf-time-to-tsc.c
@@ -21,6 +21,7 @@
 #include "mmap.h"
 #include "tests.h"
 #include "pmu.h"
+#include "pmu-hybrid.h"
 
 #define CHECK__(x) {				\
 	while ((x) < 0) {			\
@@ -93,7 +94,7 @@ int test__perf_time_to_tsc(struct test *test __maybe_unused, int subtest __maybe
 	 * For hybrid "cycles:u", it creates two events.
 	 * Init the second evsel here.
 	 */
-	if (perf_pmu__has_hybrid()) {
+	if (perf_pmu__has_hybrid() && perf_pmu__hybrid_mounted("cpu_atom")) {
 		evsel = evsel__next(evsel);
 		evsel->core.attr.comm = 1;
 		evsel->core.attr.disabled = 1;
-- 
2.17.1

Re: [PATCH v2 0/4] perf tool: Skip invalid hybrid PMU

[Jiri Olsa]

On Thu, Jul 08, 2021 at 09:36:57AM +0800, Jin Yao wrote:
> On hybrid platform, such as Alderlake, if atom CPUs are offlined,
> the kernel still exports the sysfs path '/sys/devices/cpu_atom/' for
> 'cpu_atom' pmu but the file '/sys/devices/cpu_atom/cpus' is empty,
> which indicates this is an invalid pmu. So we need to check and skip
> the invalid hybrid pmu.
> 
> Also we need to update some perf test cases for core-only system.
> 
> Jin Yao (4):
>   perf pmu: Skip invalid hybrid pmu
>   perf tests: Fix 'Parse event definition strings' on core-only system
>   perf tests: Fix 'Roundtrip evsel->name' on core-only system
>   perf tests: Fix 'Convert perf time to TSC' on core-only system

Acked-by: Jiri Olsa <jolsa@redhat.com>

thanks,
jirka

> 
>  tools/perf/tests/evsel-roundtrip-name.c |  3 ++-
>  tools/perf/tests/parse-events.c         | 16 ++++++++++------
>  tools/perf/tests/perf-time-to-tsc.c     |  3 ++-
>  tools/perf/util/pmu.c                   |  9 ++++++++-
>  4 files changed, 22 insertions(+), 9 deletions(-)
> 
> -- 
> 2.17.1
> 

Re: [PATCH v2 0/4] perf tool: Skip invalid hybrid PMU

[Arnaldo Carvalho de Melo]

Em Sun, Jul 11, 2021 at 05:52:48PM +0200, Jiri Olsa escreveu:
> On Thu, Jul 08, 2021 at 09:36:57AM +0800, Jin Yao wrote:
> > On hybrid platform, such as Alderlake, if atom CPUs are offlined,
> > the kernel still exports the sysfs path '/sys/devices/cpu_atom/' for
> > 'cpu_atom' pmu but the file '/sys/devices/cpu_atom/cpus' is empty,
> > which indicates this is an invalid pmu. So we need to check and skip
> > the invalid hybrid pmu.
> > 
> > Also we need to update some perf test cases for core-only system.
> > 
> > Jin Yao (4):
> >   perf pmu: Skip invalid hybrid pmu
> >   perf tests: Fix 'Parse event definition strings' on core-only system
> >   perf tests: Fix 'Roundtrip evsel->name' on core-only system
> >   perf tests: Fix 'Convert perf time to TSC' on core-only system
> 
> Acked-by: Jiri Olsa <jolsa@redhat.com>

Thanks, applied to perf/urgent.

- Arnaldo