Re: [PATCH v3 2/3] tests: add some tests validating skipped functions due to uncertain arg location

["Alexis Lothoré" <alexis.lothore@bootlin.com>]

On Wed Jul 9, 2025 at 6:21 PM CEST, Alan Maguire wrote:
> On 07/07/2025 20:36, Ihor Solodrai wrote:
>> On 7/7/25 7:14 AM, Alexis Lothoré wrote:
>>> On Mon Jul 7, 2025 at 4:02 PM CEST, Alexis Lothoré (eBPF Foundation)
>>> wrote:

[...]

>> I think a proper fix for this is differentiating two variants of
>> LSK__STOP_LOADING: stop because of an error, and stop because there is
>> nothing else to do. That would require a bit of refactoring.
>> 
>> Alan, Arnaldo, what do you think?
>>
>
> Would it suffice to treat LSK__STOP_LOADING as an error in the BTF
> encoding case, and not otherwise? That's a bit of hack; ideally I
> suppose we'd introduce LSK__ABORT (like DWARF_CB_ABORT) and use it for
> all the failure modes, reserving LSK__STOP_LOADING for cases where we
> are done processing rather than we met an error.

Ihor, Alan, is anyone one of you planning to work on it ? If not, do you
want me take a look and implement one of the solution suggested above ? I
guess it's best to aim for Alan's second suggestion first (introducing a
new LSK enum to represent a failure), otherwise the simpler solution
distinguishing reasons for LSK__STOP_LOADING.

Alexis

-- 
Alexis Lothoré, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com