[PATCH v1 0/4] Test fixes and debug logging

[PATCH v1 0/4] Test fixes and debug logging

[Ian Rogers]

Testing in a hypervisor guest showed some issues that these patches
try to address, or improve the debug situation upon.

Ian Rogers (4):
  perf test: Be tolerant of missing json metric none value
  perf parse-events: Add debug logging to perf_event
  perf test: Don't fail if user rdpmc returns 0 when disabled
  perf stat-shadow: Display metric-only for 0 counters

 tools/perf/tests/mmap-basic.c                       |  2 +-
 tools/perf/tests/shell/lib/perf_json_output_lint.py |  7 +++++--
 tools/perf/util/parse-events.c                      |  2 ++
 tools/perf/util/stat-display.c                      |  3 +++
 tools/perf/util/stat-shadow.c                       | 11 ++++-------
 5 files changed, 15 insertions(+), 10 deletions(-)

-- 
2.51.0.384.g4c02a37b29-goog

[PATCH v1 1/4] perf test: Be tolerant of missing json metric none value

[Ian Rogers]

print_metric_only_json and print_metric_end in stat-display.c may
create a metric value of "none" which fails validation as isfloat. Add
a helper to properly validate metric numeric values.

Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/perf/tests/shell/lib/perf_json_output_lint.py | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/tools/perf/tests/shell/lib/perf_json_output_lint.py b/tools/perf/tests/shell/lib/perf_json_output_lint.py
index c6750ef06c0f..ff90c23d1d62 100644
--- a/tools/perf/tests/shell/lib/perf_json_output_lint.py
+++ b/tools/perf/tests/shell/lib/perf_json_output_lint.py
@@ -43,6 +43,9 @@ def isint(num):
 def is_counter_value(num):
   return isfloat(num) or num == '<not counted>' or num == '<not supported>'
 
+def is_metric_value(num):
+  return isfloat(num) or num == 'none'
+
 def check_json_output(expected_items):
   checks = {
       'counters': lambda x: isfloat(x),
@@ -57,7 +60,7 @@ def check_json_output(expected_items):
       'event-runtime': lambda x: isfloat(x),
       'interval': lambda x: isfloat(x),
       'metric-unit': lambda x: True,
-      'metric-value': lambda x: isfloat(x),
+      'metric-value': lambda x: is_metric_value(x),
       'metric-threshold': lambda x: x in ['unknown', 'good', 'less good', 'nearly bad', 'bad'],
       'metricgroup': lambda x: True,
       'node': lambda x: True,
@@ -65,7 +68,7 @@ def check_json_output(expected_items):
       'socket': lambda x: True,
       'thread': lambda x: True,
       'unit': lambda x: True,
-      'insn per cycle': lambda x: isfloat(x),
+      'insn per cycle': lambda x: is_metric_value(x),
       'GHz': lambda x: True,  # FIXME: it seems unintended for --metric-only
   }
   input = '[\n' + ','.join(Lines) + '\n]'
-- 
2.51.0.384.g4c02a37b29-goog

[PATCH v1 2/4] perf parse-events: Add debug logging to perf_event

[Ian Rogers]

If verbose is enabled and parse_event is called, typically by tests,
log failures.

Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/perf/util/parse-events.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c
index 5dbbede45363..f96f2e750fe7 100644
--- a/tools/perf/util/parse-events.c
+++ b/tools/perf/util/parse-events.c
@@ -2211,6 +2211,8 @@ int parse_event(struct evlist *evlist, const char *str)
 
 	parse_events_error__init(&err);
 	ret = parse_events(evlist, str, &err);
+	if (ret && verbose > 0)
+		parse_events_error__print(&err, str);
 	parse_events_error__exit(&err);
 	return ret;
 }
-- 
2.51.0.384.g4c02a37b29-goog

[PATCH v1 3/4] perf test: Don't fail if user rdpmc returns 0 when disabled

[Ian Rogers]

In certain hypervisor set ups the value 0 may be returned but this is
only erroneous if the user rdpmc isn't disabled.

Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/perf/tests/mmap-basic.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/tools/perf/tests/mmap-basic.c b/tools/perf/tests/mmap-basic.c
index 3c89d3001887..3313c236104e 100644
--- a/tools/perf/tests/mmap-basic.c
+++ b/tools/perf/tests/mmap-basic.c
@@ -322,7 +322,7 @@ static int test_stat_user_read(u64 event, enum user_read_state enabled)
 		}
 
 		perf_evsel__read(evsel, 0, 0, &counts);
-		if (counts.val == 0) {
+		if (rdpmc_supported && counts.val == 0) {
 			pr_err("User space counter reading for PMU %s [Failed read]\n", pmu->name);
 			ret = TEST_FAIL;
 			goto cleanup;
-- 
2.51.0.384.g4c02a37b29-goog

[PATCH v1 4/4] perf stat-shadow: Display metric-only for 0 counters

[Ian Rogers]

0 counters may occur in hypervisor settings but metric-only output is
always expected. This resolves an issue in the "perf stat STD output
linter" test.

Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/perf/util/stat-display.c |  3 +++
 tools/perf/util/stat-shadow.c  | 11 ++++-------
 2 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/tools/perf/util/stat-display.c b/tools/perf/util/stat-display.c
index a67b991f4e81..8a72e019e6c7 100644
--- a/tools/perf/util/stat-display.c
+++ b/tools/perf/util/stat-display.c
@@ -944,6 +944,9 @@ static bool should_skip_zero_counter(struct perf_stat_config *config,
 	if (verbose == 0 && counter->skippable && !counter->supported)
 		return true;
 
+	/* Metric only counts won't be displayed but the metric wants to be computed. */
+	if (config->metric_only)
+		return false;
 	/*
 	 * Skip value 0 when enabling --per-thread globally,
 	 * otherwise it will have too many 0 output.
diff --git a/tools/perf/util/stat-shadow.c b/tools/perf/util/stat-shadow.c
index abaf6b579bfc..dd3a15b17eb9 100644
--- a/tools/perf/util/stat-shadow.c
+++ b/tools/perf/util/stat-shadow.c
@@ -320,13 +320,10 @@ static void print_instructions(struct perf_stat_config *config,
 	double max_stalled = max(find_stat(evsel, aggr_idx, STAT_STALLED_CYCLES_FRONT),
 				find_stat(evsel, aggr_idx, STAT_STALLED_CYCLES_BACK));
 
-	if (cycles) {
-		print_metric(config, ctxp, METRIC_THRESHOLD_UNKNOWN, "%7.2f ",
-			     "insn per cycle", instructions / cycles);
-	} else {
-		print_metric(config, ctxp, METRIC_THRESHOLD_UNKNOWN, /*fmt=*/NULL,
-			     "insn per cycle", 0);
-	}
+        print_metric(config, ctxp, METRIC_THRESHOLD_UNKNOWN, "%7.2f ",
+                     "insn per cycle",
+                     fpclassify(cycles) == FP_ZERO ? 0 : instructions / cycles);
+
 	if (max_stalled && instructions) {
 		if (out->new_line)
 			out->new_line(config, ctxp);
-- 
2.51.0.384.g4c02a37b29-goog

Re: [PATCH v1 0/4] Test fixes and debug logging

[Arnaldo Carvalho de Melo]

On Sun, Sep 14, 2025 at 11:28:56AM -0700, Ian Rogers wrote:
> Testing in a hypervisor guest showed some issues that these patches
> try to address, or improve the debug situation upon.
> 
> Ian Rogers (4):
>   perf test: Be tolerant of missing json metric none value
>   perf parse-events: Add debug logging to perf_event
>   perf test: Don't fail if user rdpmc returns 0 when disabled
>   perf stat-shadow: Display metric-only for 0 counters

Applied, but while checking before/after on a host, I see that these are
quite unstable:

root@x1:~# perf test -r10 linter
 94: perf stat CSV output linter                                     : FAILED!
 94: perf stat CSV output linter                                     : FAILED!
 94: perf stat CSV output linter                                     : FAILED!
 94: perf stat CSV output linter                                     : FAILED!
 94: perf stat CSV output linter                                     : FAILED!
 94: perf stat CSV output linter                                     : FAILED!
 94: perf stat CSV output linter                                     : FAILED!
 94: perf stat CSV output linter                                     : FAILED!
 94: perf stat CSV output linter                                     : FAILED!
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 97: perf stat JSON output linter                                    : Ok
 97: perf stat JSON output linter                                    : Ok
 97: perf stat JSON output linter                                    : Ok
 97: perf stat JSON output linter                                    : Ok
 97: perf stat JSON output linter                                    : Ok
 97: perf stat JSON output linter                                    : Ok
 97: perf stat JSON output linter                                    : Ok
 97: perf stat JSON output linter                                    : Ok
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
 99: perf stat STD output linter                                     : Ok
 99: perf stat STD output linter                                     : Ok
 99: perf stat STD output linter                                     : FAILED!
 99: perf stat STD output linter                                     : FAILED!
 99: perf stat STD output linter                                     : Ok
 99: perf stat STD output linter                                     : FAILED!
 99: perf stat STD output linter                                     : Ok
 99: perf stat STD output linter                                     : Ok
 99: perf stat STD output linter                                     : FAILED!
root@x1:~#

root@x1:~# for a in $(seq 20) ; do perf test linter ; echo "------" ; done
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : FAILED!
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : FAILED!
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
------
 94: perf stat CSV output linter                                     : FAILED!
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
------
root@x1:~#

root@x1:~# uname -a
Linux x1 6.16.5-200.fc42.x86_64 #1 SMP PREEMPT_DYNAMIC Thu Sep  4 16:37:21 UTC 2025 x86_64 GNU/Linux
root@x1:~# grep -m1 "model name" /proc/cpuinfo 
model name	: 13th Gen Intel(R) Core(TM) i7-1365U
root@x1:~#

All is well on AMD land:

root@number:~# perf test linter
 94: perf stat CSV output linter                                     : Ok
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
root@number:~# perf test -r 5 linter
 94: perf stat CSV output linter                                     : Ok
 94: perf stat CSV output linter                                     : Ok
 94: perf stat CSV output linter                                     : Ok
 94: perf stat CSV output linter                                     : Ok
 94: perf stat CSV output linter                                     : Ok
 97: perf stat JSON output linter                                    : Ok
 97: perf stat JSON output linter                                    : Ok
 97: perf stat JSON output linter                                    : Ok
 97: perf stat JSON output linter                                    : Ok
 97: perf stat JSON output linter                                    : Ok
 99: perf stat STD output linter                                     : Ok
 99: perf stat STD output linter                                     : Ok
 99: perf stat STD output linter                                     : Ok
 99: perf stat STD output linter                                     : Ok
 99: perf stat STD output linter                                     : Ok
root@number:~# uname -a
Linux number 6.16.7-200.fc42.x86_64 #1 SMP PREEMPT_DYNAMIC Thu Sep 11 17:46:54 UTC 2025 x86_64 GNU/Linux
root@number:~# grep -m1 "model name" /proc/cpuinfo 
model name	: AMD Ryzen 9 9950X3D 16-Core Processor
root@number:~#

acme@number:~/git/perf-tools-next$ git log --oneline -5
58c26aeac5132152 (HEAD -> perf-tools-next, perf-tools-next/tmp.perf-tools-next) perf stat-shadow: Display metric-only for 0 counters
f46533cc149e39bd perf test: Don't fail if user rdpmc returns 0 when disabled
9bb146fd8a342ae1 perf parse-events: Add debug logging to perf_event
64b3eae288bb4370 perf test: Be tolerant of missing json metric none value
0dc96cae063cbf9e perf build-id: Ensure snprintf string is empty when size is 0
acme@number:~/git/perf-tools-next$ sudo su -
root@number:~# perf -v
perf version 6.17.rc6.g58c26aeac513
root@number:~#

- Arnaldo

Re: [PATCH v1 0/4] Test fixes and debug logging

[Arnaldo Carvalho de Melo]

On Sun, Sep 14, 2025 at 11:28:56AM -0700, Ian Rogers wrote:
> Testing in a hypervisor guest showed some issues that these patches
> try to address, or improve the debug situation upon.
> 
> Ian Rogers (4):
>   perf test: Be tolerant of missing json metric none value
>   perf parse-events: Add debug logging to perf_event
>   perf test: Don't fail if user rdpmc returns 0 when disabled
>   perf stat-shadow: Display metric-only for 0 counters

Thanks, applied to perf-tools-next,

- Arnaldo

Re: [PATCH v1 0/4] Test fixes and debug logging

[Ian Rogers]

On Fri, Sep 19, 2025 at 1:46 PM Arnaldo Carvalho de Melo
<acme@kernel.org> wrote:
>
> On Sun, Sep 14, 2025 at 11:28:56AM -0700, Ian Rogers wrote:
> > Testing in a hypervisor guest showed some issues that these patches
> > try to address, or improve the debug situation upon.
> >
> > Ian Rogers (4):
> >   perf test: Be tolerant of missing json metric none value
> >   perf parse-events: Add debug logging to perf_event
> >   perf test: Don't fail if user rdpmc returns 0 when disabled
> >   perf stat-shadow: Display metric-only for 0 counters
>
> Thanks, applied to perf-tools-next,

Thanks Arnaldo,

there was some follow up in:
https://lore.kernel.org/lkml/20250914182140.1958530-1-irogers@google.com/
I think I've also noticed that logging introducing newlines that
impacts the tests column counts. I'll take a further look as we tried
to harden against that with an output file option.

Thanks,
Ian

Re: [PATCH v1 0/4] Test fixes and debug logging

[Arnaldo Carvalho de Melo]

On Sat, Sep 20, 2025 at 08:21:43AM -0700, Ian Rogers wrote:
> On Fri, Sep 19, 2025 at 1:46 PM Arnaldo Carvalho de Melo <acme@kernel.org> wrote:
> > On Sun, Sep 14, 2025 at 11:28:56AM -0700, Ian Rogers wrote:
> > > Ian Rogers (4):
> > >   perf test: Be tolerant of missing json metric none value
> > >   perf parse-events: Add debug logging to perf_event
> > >   perf test: Don't fail if user rdpmc returns 0 when disabled
> > >   perf stat-shadow: Display metric-only for 0 counters

> > Thanks, applied to perf-tools-next,
 
> there was some follow up in:
> https://lore.kernel.org/lkml/20250914182140.1958530-1-irogers@google.com/
> I think I've also noticed that logging introducing newlines that
> impacts the tests column counts. I'll take a further look as we tried
> to harden against that with an output file option.

I'll make sure I get the most uptodate version, if/when available.

I'm attending Kernel Recipes, so there is an additional delay on
processing patches, but I should be back processing patches RSN.

- Arnaldo