Re: [PATCH] perf jevents: make events order more deterministic

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] The `__lt__` method violates strict weak ordering by using flawed boolean logic, breaking Python's sorting guarantees.
--

commit 1d3198ec01401252a35bb75abf7f3dc67eead5ab
Author: Nazar Kazakov <nazar.kazakov@codethink.co.uk>

perf jevents: make events order more deterministic

This commit attempts to fix a non-deterministic sorting order for events
that share the same name by including the description field in the
comparison.

> diff --git a/tools/perf/pmu-events/metric.py b/tools/perf/pmu-events/metric.py
> index a91ccb5977f08..b3ab78a0ecb22 100644
> --- a/tools/perf/pmu-events/metric.py
> +++ b/tools/perf/pmu-events/metric.py
> @@ -623,7 +623,7 @@ class Metric:
>  
>    def __lt__(self, other):
>      """Sort order."""
> -    return self.name < other.name
> +    return self.name < other.name or self.description < other.description

[Severity: Medium]
Does this logic break Python's strict weak ordering requirement for sorting?

If we have two metrics A and B where A.name > B.name, but
A.description < B.description, then A < B will evaluate to True because the
second condition is met. Conversely, B < A will also evaluate to True because
B.name < A.name meets the first condition.

Since A < B and B < A can both evaluate to True, this breaks the asymmetry
requirement for sorting and can lead to inconsistent event ordering or Python
exceptions during the perf build.

Could we compare tuples instead to ensure consistent ordering, such as:
return (self.name, self.description) < (other.name, other.description)

>  
>    def AddToMetricGroup(self, group):
>      """Callback used when being added to a MetricGroup."""

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260706175624.692736-1-nazar.kazakov@codethink.co.uk?part=1