[LTP] [PATCH] crypto/af_alg0[13]: update tests for additional possible errno case

[David Hildenbrand]

On 05.11.24 11:40, Petr Vorel wrote:
> Hi Cyril, Jan, David,
> 
>> kernel behaviour wrt checking invalid algorithms has changed [1] [2]
>> updating the tests to address the additional errno case.
>> Related discussion on the mailing list [3]
> 
> Looking at 57ab2160c0 ("move_pages04: remove special-casing for kernels < 4.3") [4]
> recently reverting errnos for 4.3 d539a004dd ("move_pages04: fix zero page
> status code for kernels >= 4.3") [5] please double check this already merged
> change. I still believe it's a different case thus merging this is correct.
> Also Eric is suggesting this (I should have added Suggested-by: tag for him).
> 
> Maybe we need some rules to clarify when we are ok with different errno and when not.
> 

Right.

Regarding d539a004dd, we pretty much hid kernel bugs: behaving 
differently than expected+documented.

If the kernel starts reporting a different errno it might be a bug: user 
space might not be prepared to handle that. Or it might be expected, 
because nobody really cares about the exact error code.

So if a test starts failing, it's definitely concerning and needs a 
closer look.

> I also thought there would be some rule "don't hide kernel bugs", but I can't
> find it in the docs.

That rule makes sense to me.

-- 
Cheers,

David / dhildenb