[LTP] [PATCH v2] userfaultfd: Minor fixes

[Petr Vorel]

> > Yes, IMHO TFAIL.

> I guess we need to agree on the solution here, otherwise Ricard will be
> puzzled :-)

Yeah, I'm sorry for the noise.

> I see TBROK as the right solution, since TFAIL should be used on the subject of
> our tests. Also, if we receive the wrong event there's no purpose to continue
> with the test because that means we have a kernel bug.

OK, tst_brk(TBROK | TERRNO is probably better (and we definitely want TERRNO).

I overlooked Ricardo suggests tst_brk(TFAIL) not tst_res(TFAIL).
The only concern I have is that we skip testing on userfaultfd01.c due tst_brk()
but it's probably quite rare.

Also, this is the problem when test have a lot of duplicity => tests diverge
sometimes in subtle details. But, similarly to fanotify tests, userfaultfd tests
are quite different that it's questionable if factoring out small parts improves
things or complicate.

Kind regards,
Petr