Skip to contentSkip to conversation

perf test: Do not compare overheads in the zstd comp test

PATCH Applied Namhyung Kim 3 participants 3 messages Aug 12, 2021 – Aug 13, 2021

Subsystems (2)
lore
Share & tools
lore

Namhyung Kim

author maintainer
#0 lore raw

PATCH 1/1 perf test: Do not compare overheads in the zstd comp test

The overhead can vary on each run so it'd make the test failed sometimes. Also order of hist entry can change.

Use perf report -F option to omit the overhead field and sort the result alphabetically.

---
 tools/perf/tests/shell/record+zstd_comp_decomp.sh | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
tools/perf/tests/shell/record+zstd_comp_decomp.sh +2 -2
index 045723b3d992..8a168cf8bacc 100755
@@ -25,8 +25,8 @@ check_compressed_stats() {
 
 check_compressed_output() {
 	$perf_tool inject -i $trace_file -o $trace_file.decomp &&
-	$perf_tool report -i $trace_file --stdio | head -n -3 > $trace_file.comp.output &&
-	$perf_tool report -i $trace_file.decomp --stdio | head -n -3 > $trace_file.decomp.output &&
+	$perf_tool report -i $trace_file --stdio -F comm,dso,sym | head -n -3 > $trace_file.comp.output &&
+	$perf_tool report -i $trace_file.decomp --stdio -F comm,dso,sym | head -n -3 > $trace_file.decomp.output &&
 	diff $trace_file.comp.output $trace_file.decomp.output
 }
 
Signature · 1 line
2.33.0.rc1.237.g0d66db33f3-goog
reviewer Acked-by on v1 1/1
#1 lore raw
Quoted 8 lines · Namhyung Kim

The overhead can vary on each run so it'd make the test failed sometimes. Also order of hist entry can change.

Use perf report -F option to omit the overhead field and sort the result alphabetically.

Thanks,
Ian

Quoted 22 lines
---
 tools/perf/tests/shell/record+zstd_comp_decomp.sh | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
tools/perf/tests/shell/record+zstd_comp_decomp.sh +2 -3
index 045723b3d992..8a168cf8bacc 100755
@@ -25,8 +25,8 @@ check_compressed_stats() {
 
 check_compressed_output() {
        $perf_tool inject -i $trace_file -o $trace_file.decomp &&
-       $perf_tool report -i $trace_file --stdio | head -n -3 > $trace_file.comp.output &&
-       $perf_tool report -i $trace_file.decomp --stdio | head -n -3 > $trace_file.decomp.output &&
+       $perf_tool report -i $trace_file --stdio -F comm,dso,sym | head -n -3 > $trace_file.comp.output &&
+       $perf_tool report -i $trace_file.decomp --stdio -F comm,dso,sym | head -n -3 > $trace_file.decomp.output &&
        diff $trace_file.comp.output $trace_file.decomp.output
 }
 
--

2.33.0.rc1.237.g0d66db33f3-goog

maintainer Applied
#2 lore raw

Em Thu, Aug 12, 2021 at 05:09:20PM -0700, Ian Rogers escreveu:

Quoted 11 lines · Namhyung Kim
Quoted 8 lines

The overhead can vary on each run so it'd make the test failed sometimes. Also order of hist entry can change.

Use perf report -F option to omit the overhead field and sort the result alphabetically.

Thanks, applied.

- Arnaldo

Quoted 25 lines

Thanks,
Ian

Quoted 22 lines
---
 tools/perf/tests/shell/record+zstd_comp_decomp.sh | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
tools/perf/tests/shell/record+zstd_comp_decomp.sh +2 -3
index 045723b3d992..8a168cf8bacc 100755
@@ -25,8 +25,8 @@ check_compressed_stats() {
 
 check_compressed_output() {
        $perf_tool inject -i $trace_file -o $trace_file.decomp &&
-       $perf_tool report -i $trace_file --stdio | head -n -3 > $trace_file.comp.output &&
-       $perf_tool report -i $trace_file.decomp --stdio | head -n -3 > $trace_file.decomp.output &&
+       $perf_tool report -i $trace_file --stdio -F comm,dso,sym | head -n -3 > $trace_file.comp.output &&
+       $perf_tool report -i $trace_file.decomp --stdio -F comm,dso,sym | head -n -3 > $trace_file.decomp.output &&
        diff $trace_file.comp.output $trace_file.decomp.output
 }
 
--

2.33.0.rc1.237.g0d66db33f3-goog

Signature · 1 line
- Arnaldo