diff options
| author | Paul Buetow <paul@buetow.org> | 2026-05-29 17:53:10 +0300 |
|---|---|---|
| committer | Paul Buetow <paul@buetow.org> | 2026-05-29 17:53:10 +0300 |
| commit | b86b817594ac8a4dc7fe2b80649df2c7a62f1b59 (patch) | |
| tree | bfe09e28771102384cb3f04cae68f7bc907d8eab /docs | |
| parent | 2d87d048eb0c629990445eac81229424fe50c215 (diff) | |
test(fallocate): lock in FS family, KindFd enter, UNCLASSIFIED ret
Audit of fallocate(2) found the tracing correct and consistent with its
fd-based siblings, so add lock-in tests rather than fixing anything:
- fallocate(int fd, int mode, off_t offset, off_t len) returns int 0/-1
(a status code, NOT a transferred byte count). Its exit must stay a
plain ret_event with ret_type UNCLASSIFIED so it is never mistaken for
a READ/WRITE/TRANSFER byte count.
- The enter tracepoint carries a leading fd field (args[0]); only fd is
captured into a fd_event (KindFd), matching fadvise64/ftruncate/
sync_file_range which likewise drop their trailing offset/len/advice
args.
- fallocate belongs to FamilyFS alongside fadvise64/ftruncate/
sync_file_range.
TestClassifyFallocateEnterFd and TestClassifyExitFallocateUnclassifiedRet
assert the per-syscall behavior; TestClassifySyscallFamily now also
covers fallocate/fadvise64/ftruncate so a stray reclassification trips a
test. No classification logic or generated artifacts changed.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Diffstat (limited to 'docs')
0 files changed, 0 insertions, 0 deletions
