Skip to contentSkip to conversation

perf test: Skip the extra metric line for stalled_cycles_per_instruction

PATCH New TAKANO Mitsuhiro 2 participants 2 messages Oct 22 perf events

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 perf test: Skip the extra metric line for stalled_cycles_per_instruction

The STD output linter maps one metric to each default event and stops at the first line that shows a different one. stalled_cycles_per_instruction is computed as max(stalled-cycles-frontend, stalled-cycles-backend) / instructions, so its event set differs from the one behind backend_cycles_idle and perf stat prints it on a line of its own: a second stalled-cycles-backend line carrying that metric. The linter rejects it with "wrong event metric. expected backend_cycles_idle in ...".

Metrics printed on their own line are already skipped (skip_metric=("tma_" "TopdownL1")); add this one to that list.

Verified on an Ampere Altra (arm64, Neoverse-N1): perf test "perf stat STD output linter" fails before the change and passes after it. Only the linter is changed; the perf stat output is untouched.

Fixes c7adeb0974f1 ("perf jevents: Add set of common metrics based on default ones")
Assisted-by LLM
---
 tools/perf/tests/shell/stat+std_output.sh | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
tools/perf/tests/shell/stat+std_output.sh +1 -1
index 9c4b92ecf448..d9610f2c92a6 100755
@@ -13,7 +13,7 @@ stat_output=$(mktemp /tmp/__perf_test.stat_output.std.XXXXX)
 
 event_name=(cpu-clock task-clock context-switches cpu-migrations page-faults stalled-cycles-frontend stalled-cycles-backend cycles instructions branches branch-misses)
 event_metric=("CPUs_utilized" "CPUs_utilized" "cs/sec" "migrations/sec" "faults/sec" "frontend_cycles_idle" "backend_cycles_idle" "GHz" "insn_per_cycle" "/sec" "branch_miss_rate")
-skip_metric=("tma_" "TopdownL1")
+skip_metric=("tma_" "TopdownL1" "stalled_cycles_per_instruction")
 
 cleanup() {
   rm -f "${stat_output}"

base-commit: f47d223627f80cb7a1eb14c585fe6cb35de5e984

Signature · 1 line
2.53.0
#1 lore raw
Sashiko has reviewed this patch and found no issues.
Quoted 12 lines

The STD output linter maps one metric to each default event and stops at the first line that shows a different one. stalled_cycles_per_instruction is computed as max(stalled-cycles-frontend, stalled-cycles-backend) / instructions, so its event set differs from the one behind backend_cycles_idle and perf stat prints it on a line of its own: a second stalled-cycles-backend line carrying that metric. The linter rejects it with "wrong event metric. expected backend_cycles_idle in ...".

Metrics printed on their own line are already skipped (skip_metric=("tma_" "TopdownL1")); add this one to that list.

Quoted 7 lines

Verified on an Ampere Altra (arm64, Neoverse-N1): perf test "perf stat STD output linter" fails before the change and passes after it. Only the linter is changed; the perf stat output is untouched.

Fixes c7adeb0974f1 ("perf jevents: Add set of common metrics based on default ones")
Assisted-by LLM

Sashiko has reviewed this patch and found no issues.

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1