Skip to contentSkip to conversation

perf test: Pass the verbose option to shell tests

PATCH In review Ian Rogers 2 participants 8 messages Jun 21, 2021 – Jun 22, 2021

Subsystems (2)
lore
Share & tools
lore

Ian Rogers

author reviewer
#0 lore raw

PATCH v2 1/3 perf test: Pass the verbose option to shell tests

Having a verbose option will allow shell tests to provide extra failure details when the fail or skip.

---
 tools/perf/tests/builtin-test.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
tools/perf/tests/builtin-test.c +5 -2
index cbbfe48ab802..e1ed60567b2f 100644
@@ -577,10 +577,13 @@ struct shell_test {
 static int shell_test__run(struct test *test, int subdir __maybe_unused)
 {
 	int err;
-	char script[PATH_MAX];
+	char script[PATH_MAX + 3];
 	struct shell_test *st = test->priv;
 
-	path__join(script, sizeof(script), st->dir, st->file);
+	path__join(script, sizeof(script) - 3, st->dir, st->file);
+
+	if (verbose)
+		strncat(script, " -v", sizeof(script) - strlen(script) - 1);
 
 	err = system(script);
 	if (!err)
Signature · 1 line
2.32.0.288.g62a8d224e6-goog
author reviewer
#1 lore raw

PATCH v2 2/3 perf test: Add verbose skip output for bpf counters

Provide additional context for when the stat bpf counters test skips.

---
 tools/perf/tests/shell/stat_bpf_counters.sh | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
tools/perf/tests/shell/stat_bpf_counters.sh +7 -1
index 22eb31e48ca7..85eb689fe202 100755
@@ -22,7 +22,13 @@ compare_number()
 }
 
 # skip if --bpf-counters is not supported
-perf stat --bpf-counters true > /dev/null 2>&1 || exit 2
+if ! perf stat --bpf-counters true > /dev/null 2>&1; then
+	if [ "$1" == "-v" ]; then
+		echo "Skipping: --bpf-counters not supported"
+		perf --no-pager stat --bpf-counters true || true
+	fi
+	exit 2
+fi
 
 base_cycles=$(perf stat --no-big-num -e cycles -- perf bench sched messaging -g 1 -l 100 -t 2>&1 | awk '/cycles/ {print $1}')
 bpf_cycles=$(perf stat --no-big-num --bpf-counters -e cycles -- perf bench sched messaging -g 1 -l 100 -t 2>&1 | awk '/cycles/ {print $1}')
Signature · 1 line
2.32.0.288.g62a8d224e6-goog
#2 lore raw

Re: [PATCH v2 2/3] perf test: Add verbose skip output for bpf counters

Em Mon, Jun 21, 2021 at 02:56:47PM -0700, Ian Rogers escreveu:

Quoted 1 line

Provide additional context for when the stat bpf counters test skips.

Ditto

Quoted 27 lines
---
 tools/perf/tests/shell/stat_bpf_counters.sh | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
tools/perf/tests/shell/stat_bpf_counters.sh +7 -1
index 22eb31e48ca7..85eb689fe202 100755
@@ -22,7 +22,13 @@ compare_number()
 }
 
 # skip if --bpf-counters is not supported
-perf stat --bpf-counters true > /dev/null 2>&1 || exit 2
+if ! perf stat --bpf-counters true > /dev/null 2>&1; then
+	if [ "$1" == "-v" ]; then
+		echo "Skipping: --bpf-counters not supported"
+		perf --no-pager stat --bpf-counters true || true
+	fi
+	exit 2
+fi
 
 base_cycles=$(perf stat --no-big-num -e cycles -- perf bench sched messaging -g 1 -l 100 -t 2>&1 | awk '/cycles/ {print $1}')
 bpf_cycles=$(perf stat --no-big-num --bpf-counters -e cycles -- perf bench sched messaging -g 1 -l 100 -t 2>&1 | awk '/cycles/ {print $1}')
Signature · 1 line
2.32.0.288.g62a8d224e6-goog
Signature · 1 line
- Arnaldo
author reviewer
#3 lore raw

PATCH v2 3/3 perf test: Make stat bpf counters test more robust

In reply to Ian Rogers

If the test is run on a hypervisor then the cycles event may not be counted, skip the test in this situation. Fail the test if cycles are not counted in the subsequent bpf counter run.

---
 tools/perf/tests/shell/stat_bpf_counters.sh | 8 ++++++++
 1 file changed, 8 insertions(+)
tools/perf/tests/shell/stat_bpf_counters.sh +8 -0
index 85eb689fe202..6b156dd85469 100755
@@ -31,7 +31,15 @@ if ! perf stat --bpf-counters true > /dev/null 2>&1; then
 fi
 
 base_cycles=$(perf stat --no-big-num -e cycles -- perf bench sched messaging -g 1 -l 100 -t 2>&1 | awk '/cycles/ {print $1}')
+if [ "$base_cycles" == "<not" ]; then
+	echo "Skipping: cycles event not counted"
+	exit 2
+fi
 bpf_cycles=$(perf stat --no-big-num --bpf-counters -e cycles -- perf bench sched messaging -g 1 -l 100 -t 2>&1 | awk '/cycles/ {print $1}')
+if [ "$bpf_cycles" == "<not" ]; then
+	echo "Failed: cycles not counted with --bpf-counters"
+	exit 1
+fi
 
 compare_number $base_cycles $bpf_cycles
 exit 0
Signature · 1 line
2.32.0.288.g62a8d224e6-goog
#4 lore raw

Re: [PATCH v2 3/3] perf test: Make stat bpf counters test more robust

Em Mon, Jun 21, 2021 at 02:56:48PM -0700, Ian Rogers escreveu:

Quoted 3 lines

If the test is run on a hypervisor then the cycles event may not be counted, skip the test in this situation. Fail the test if cycles are not counted in the subsequent bpf counter run.

This one was already in,

- Arnaldo

Quoted 28 lines
---
 tools/perf/tests/shell/stat_bpf_counters.sh | 8 ++++++++
 1 file changed, 8 insertions(+)
tools/perf/tests/shell/stat_bpf_counters.sh +8 -0
index 85eb689fe202..6b156dd85469 100755
@@ -31,7 +31,15 @@ if ! perf stat --bpf-counters true > /dev/null 2>&1; then
 fi
 
 base_cycles=$(perf stat --no-big-num -e cycles -- perf bench sched messaging -g 1 -l 100 -t 2>&1 | awk '/cycles/ {print $1}')
+if [ "$base_cycles" == "<not" ]; then
+	echo "Skipping: cycles event not counted"
+	exit 2
+fi
 bpf_cycles=$(perf stat --no-big-num --bpf-counters -e cycles -- perf bench sched messaging -g 1 -l 100 -t 2>&1 | awk '/cycles/ {print $1}')
+if [ "$bpf_cycles" == "<not" ]; then
+	echo "Failed: cycles not counted with --bpf-counters"
+	exit 1
+fi
 
 compare_number $base_cycles $bpf_cycles
 exit 0
Signature · 1 line
2.32.0.288.g62a8d224e6-goog
Signature · 1 line
- Arnaldo
#5 lore raw
In reply to Ian Rogers

Em Mon, Jun 21, 2021 at 02:56:46PM -0700, Ian Rogers escreveu:

Quoted 13 lines

Having a verbose option will allow shell tests to provide extra failure details when the fail or skip.

---
 tools/perf/tests/builtin-test.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
tools/perf/tests/builtin-test.c +0 -0
index cbbfe48ab802..e1ed60567b2f 100644

tools/perf/tests/builtin-test.c struct shell_test { · line 580

Quoted 5 lines
tools/perf/tests/builtin-test.c +1 -1
@@ -577,10 +577,13 @@ struct shell_test {
 static int shell_test__run(struct test *test, int subdir __maybe_unused)
 {
 	int err;
-	char script[PATH_MAX];
+	char script[PATH_MAX + 3];

This looks strange, i.e. if it is a _path_ _MAX_, why add 3 chars past that max when generating a _path_? I'll drop the above hunk and keep the rest, ok?

- Arnaldo

Quoted 10 lines
tools/perf/tests/builtin-test.c +4 -1
 	struct shell_test *st = test->priv;
 
-	path__join(script, sizeof(script), st->dir, st->file);
+	path__join(script, sizeof(script) - 3, st->dir, st->file);
+
+	if (verbose)
+		strncat(script, " -v", sizeof(script) - strlen(script) - 1);
 
 	err = system(script);
 	if (!err)
#6 lore raw

Em Tue, Jun 22, 2021 at 02:45:51PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 9 lines

Em Mon, Jun 21, 2021 at 02:56:46PM -0700, Ian Rogers escreveu:

Quoted 8 lines

Having a verbose option will allow shell tests to provide extra failure details when the fail or skip.

---
 tools/perf/tests/builtin-test.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
Quoted 14 lines

tools/perf/tests/builtin-test.c struct shell_test { · line 580

Quoted 10 lines
tools/perf/tests/builtin-test.c +1 -1
index cbbfe48ab802..e1ed60567b2f 100644
@@ -577,10 +577,13 @@ struct shell_test {
 static int shell_test__run(struct test *test, int subdir __maybe_unused)
 {
 	int err;
-	char script[PATH_MAX];
+	char script[PATH_MAX + 3];

This looks strange, i.e. if it is a _path_ _MAX_, why add 3 chars past that max when generating a _path_? I'll drop the above hunk and keep the rest, ok?

Oh well, its not a path after all, its something that is passed to system(), the use of PATH_MAX seems arbitrary, so your patch wasn't wrong, but since it is arbitrary, I'll keep it at PATH_MAX and reduce the patch size 8-)

- Arnaldo

Quoted 10 lines
Quoted 10 lines
tools/perf/tests/builtin-test.c +4 -1
 	struct shell_test *st = test->priv;
 
-	path__join(script, sizeof(script), st->dir, st->file);
+	path__join(script, sizeof(script) - 3, st->dir, st->file);
+
+	if (verbose)
+		strncat(script, " -v", sizeof(script) - strlen(script) - 1);
 
 	err = system(script);
 	if (!err)
Signature · 1 line
- Arnaldo
author reviewer
#7 lore raw
Quoted 26 lines · Arnaldo Carvalho de Melo

Em Tue, Jun 22, 2021 at 02:45:51PM -0300, Arnaldo Carvalho de Melo escreveu:

Quoted 9 lines

Em Mon, Jun 21, 2021 at 02:56:46PM -0700, Ian Rogers escreveu:

Quoted 8 lines

Having a verbose option will allow shell tests to provide extra failure details when the fail or skip.

---
 tools/perf/tests/builtin-test.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)
Quoted 14 lines

tools/perf/tests/builtin-test.c struct shell_test { · line 580

Quoted 10 lines
tools/perf/tests/builtin-test.c +1 -1
index cbbfe48ab802..e1ed60567b2f 100644
@@ -577,10 +577,13 @@ struct shell_test {
 static int shell_test__run(struct test *test, int subdir __maybe_unused)
 {
    int err;
-   char script[PATH_MAX];
+   char script[PATH_MAX + 3];

This looks strange, i.e. if it is a _path_ _MAX_, why add 3 chars past that max when generating a _path_? I'll drop the above hunk and keep the rest, ok?

Quoted 6 lines

Oh well, its not a path after all, its something that is passed to system(), the use of PATH_MAX seems arbitrary, so your patch wasn't wrong, but since it is arbitrary, I'll keep it at PATH_MAX and reduce the patch size 8-)

- Arnaldo

Works for me. Thanks,

Ian

Quoted 14 lines
Quoted 10 lines

tools/perf/tests/builtin-test.c struct shell_test { · line 589

Quoted 10 lines
tools/perf/tests/builtin-test.c +4 -1
    struct shell_test *st = test->priv;
 
-   path__join(script, sizeof(script), st->dir, st->file);
+   path__join(script, sizeof(script) - 3, st->dir, st->file);
+
+   if (verbose)
+           strncat(script, " -v", sizeof(script) - strlen(script) - 1);
 
    err = system(script);
    if (!err)

--

- Arnaldo