diff options
| author | Paul Buetow <paul@buetow.org> | 2026-05-30 09:55:38 +0300 |
|---|---|---|
| committer | Paul Buetow <paul@buetow.org> | 2026-05-30 09:55:38 +0300 |
| commit | d685ed30e772a3f04858a235d59405d624ac98f4 (patch) | |
| tree | 639140ce9cb50848f9363297d1c2dfb25cb7bfea /integrationtests | |
| parent | ff63b2c2fc99c9f4a2bb99248bd53d09df91712a (diff) | |
test(generate): lock in times(2) syscall classification
Audit of times(2) (clock_t times(struct tms *buf)) confirmed its
classification is correct and consistent with its time/clock siblings:
- sys_enter_times: KindNull (single userspace output struct tms *buf;
no fd or pathname argument).
- sys_exit_times: ret_event UNCLASSIFIED — times() returns a clock_t
tick count (or (clock_t)-1 on error), which is a tick tally, not a
transferred byte count.
- family: FamilyTime, alongside gettimeofday/clock_gettime, and NOT
FamilyProcess where getrusage lives.
No misclassification was found; docs/syscall-tracing-plan.md already
lists times under the Time family and the null kind. Add lock-in tests
so any stray reclassification trips a unit test:
- family_test.go: assert sys_enter/exit_times == FamilyTime.
- retclassify_test.go: assert sys_exit_times stays UNCLASSIFIED.
KindNull is already covered by TestClassifyM7NameOnlyKinds.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Diffstat (limited to 'integrationtests')
0 files changed, 0 insertions, 0 deletions
