[LTP] [PATCH v2 2/3] Add fanotify_get_supported_init_flags() helper function

[Martin Doucha]

On 25. 10. 22 18:53, Amir Goldstein wrote:
> All right.
> 
> Let's see which flag combinations we have in the relevant tests in fanotify14:
> 
> FAN_REPORT_DFID_FID,
> FAN_REPORT_DFID_NAME,
> FAN_REPORT_DFID_NAME_TARGET,
> 
> That's all.
> 
> Support for FAN_REPORT_FID is a requirement for the entire test.
> 
> FAN_REPORT_TARGET_FID depends on all the rest of the flags
> and support for it is already checked explicitly to skip test cases.
> 
> FAN_REPORT_NAME depends on FAN_REPORT_DIR_FID.
> 
> So effectively fanotify_get_supported_init_flags() only checks
> FAN_REPORT_DIR_FID separately from the combination
> FAN_REPORT_DFID_FID.
> 
> fanotify16, which tests *legal* combinations of these flags
> already checks the separate flag FAN_REPORT_DIR_FID
> so fanotify test cases with FAN_REPORT_DFID_FID would
> fail if a kernel that supports FAN_REPORT_DIR_FID does not
> support the combination FAN_REPORT_DFID_FID.
> 
> Bottom line:
> fanotify_get_supported_init_flags() did not add any test coverage.
> 
> This is why it is a slippery slope to suggest fixes without
> proving that there is a bug.

If this is a slippery slope, then we're already sliding down since 
commit fe31bfca which broke test case #8 because it's painfully hard to 
maintain multiple separate feature checks in setup() as you insist. I 
want to fix this mess with one simple helper function that will be 
reusable in all fanotify tests. The helper function is an up-to-date 
replacement for the old deprecated FAN_ALL_INIT_FLAGS macro. Nothing 
more, nothing less.

-- 
Martin Doucha   mdoucha@suse.cz
QA Engineer for Software Maintenance
SUSE LINUX, s.r.o.
CORSO IIa
Krizikova 148/34
186 00 Prague 8
Czech Republic