perf tests: Add auto counter reload (ACR) sampling test
PATCH v4 1/1 perf tests: Add auto counter reload (ACR) sampling test
Add auto counter reload sampling test to verify that the intended event records can be captured and the self-reloaded events won't generate any records.
---
V4: Remove model-specific ACR sampling test in ratio_to_prev test to avoid false positive. Some hybrid platforms only support ACR on E-cores, e.g., PTL. If the test process is scheduled on P-core, it leads to false positive since P-core doesn't support ACR. test_acr_sampling() is already fair enough to validate ACR sampling functionalities.
History:
V3: https://lore.kernel.org/all/[email protected]/
V2: https://lore.kernel.org/all/[email protected]/
v1: https://lore.kernel.org/all/[email protected]/
tools/perf/tests/shell/record.sh | 43 ++++++++++++++++++++++++++++++++ 1 file changed, 43 insertions(+)
tools/perf/tests/shell/record.sh +43 -0
@@ -402,6 +402,48 @@ test_callgraph() { echo "Callgraph test [Success]" } +test_acr_sampling() { + events="{instructions/period=40000,acr_mask=0x2/u,cycles/period=20000,acr_mask=0x3/u}" + pebs_events="{instructions/period=40000,acr_mask=0x2/pu,cycles/period=20000,acr_mask=0x3/u}" + echo "Auto counter reload (ACR) sampling test" + if ! perf record -o "${perfdata}" -e "${events}" ${testprog} 2> /dev/null + then + echo "Auto counter reload sampling [Skipped not supported]" + return + fi + if ! perf script -i "${perfdata}" -F event | grep -q "instructions" + then + echo "Auto counter reload sampling [Failed missing instructions event]" + err=1 + return + fi + if perf script -i "${perfdata}" -F event | grep -q "cycles" + then + echo "Auto counter reload sampling [Failed cycles event shouldn't be sampled]" + err=1 + return + fi + if ! perf record -o "${perfdata}" -e "${pebs_events}" ${testprog} 2> /dev/null + then + echo "Auto counter reload PEBS sampling [Skipped not supported]" + echo "Auto counter reload sampling [Success]" + return + fi + if ! perf script -i "${perfdata}" -F event | grep -q "instructions" + then + echo "Auto counter reload PEBS sampling [Failed missing instructions event]" + err=1 + return + fi + if perf script -i "${perfdata}" -F event | grep -q "cycles" + then + echo "Auto counter reload PEBS sampling [Failed cycles event shouldn't be sampled]" + err=1 + return + fi + echo "Auto counter reload sampling [Success]" +} + test_ratio_to_prev() { echo "ratio-to-prev test" if ! perf record -o /dev/null -e "{instructions, cycles/period=100000,ratio-to-prev=0.5/}" \
@@ -457,6 +499,7 @@ test_leader_sampling test_topdown_leader_sampling test_precise_max test_callgraph +test_acr_sampling test_ratio_to_prev # restore the default value
base-commit: 40db90ac9f66c8246c1746c56d397283d161655c
Signature · 1 line
2.34.1
[email protected]
AI reviewThank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [Medium] Test `test_acr_sampling` is flaky because it structurally relies on the `thloop` workload…
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] Test test_acr_sampling is flaky because it structurally relies on the thloop workload sustaining an IPC > 2.0. --
commit 8d4447d6783576282b44d2b951b516b17d5c46b1
Author: Dapeng Mi <[email protected]>
perf tests: Add auto counter reload (ACR) sampling test
This commit adds a new sampling test to verify that the intended event records can be captured and that self-reloaded events do not generate any records.
Quoted 4 lines
tools/perf/tests/shell/record.sh +0
-0
[ ... ]
tools/perf/tests/shell/record.sh test_callgraph() { · line 407
Quoted 7 lines
tools/perf/tests/shell/record.sh +3
-0@@ -402,6 +402,48 @@ test_callgraph() { echo "Callgraph test [Success]" } +test_acr_sampling() { + events="{instructions/period=40000,acr_mask=0x2/u,cycles/period=20000,acr_mask=0x3/u}" + pebs_events="{instructions/period=40000,acr_mask=0x2/pu,cycles/period=20000,acr_mask=0x3/u}"
[Severity: Medium]
Does this test configuration require the workload to sustain an IPC greater than 2.0 to pass?
With acr_mask=0x3 on the cycles event, if cycles reaches 20000 before instructions reaches 40000, both counters will automatically reload and no samples will be generated.
tools/perf/tests/shell/record.sh test_callgraph() { · line 414
Quoted 7 lines
tools/perf/tests/shell/record.sh +7
-0+ echo "Auto counter reload (ACR) sampling test" + if ! perf record -o "${perfdata}" -e "${events}" ${testprog} 2> /dev/null + then + echo "Auto counter reload sampling [Skipped not supported]" + return + fi + if ! perf script -i "${perfdata}" -F event | grep -q "instructions"
If the actual IPC drops to 2.0 or lower (which might happen due to SMT contention, execution on E-cores, or cache effects), the cycles event will repeatedly overflow first. This means zero samples will be recorded, and the grep check here will fail.
Should the event periods be adjusted to tolerate lower IPC values?
Quoted 5 lines
tools/perf/tests/shell/record.sh +5
-0+ then + echo "Auto counter reload sampling [Failed missing instructions event]" + err=1 + return + fi
[ ... ]
Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
Quoted 5 lines · Dapeng Mi
Add auto counter reload sampling test to verify that the intended event records can be captured and the self-reloaded events won't generate any records.
Applied to perf-tools-next, thanks!
Best regards,
Namhyung