Re: [PATCH] selftests: default to host arch for LLVM builds

[Valentin Obst]

On 4/29/24 12:04 AM, John Hubbard wrote:
> ...
>>> And yes, the selftests "normal" (non-cross-compile) build is *broken*
>>> right now, for clang. I didn't realize from the patch title that this is
>>> actually a significant fix. Maybe we should change the subject line
>>> (patch
>>> title) to something like:
>>>
>>>      [PATCH] selftests: fix the clang build: default to host arch for
>>> LLVM builds
>>
>> Yes, I agree that the title should contain the word 'fix' somewhere. For
>> me its okay if maintainers reword the title when applying the patch,
>> alternatively I can send a v2. (Is it still a v2 if I change the
>> title, or
>> rather a new patch?).
>
> It would still be a v2, although the cover letter, or the section after the
> "---", would need to point to v1 so that people could make the connection.
>
>>
>> Any thoughts on whether this also needs a 'Cc stable'? Its not quite
>> clear to me if this fix meets the requirements. As above, no
>> objections if
>> maintainers should decide to add it.
>>
>
> Maybe not, because it doesn't seem urgent. But it's a judgment call.
>
> By the way, I've been chipping away at fixing clang selftest build
> failures and warnings that are only visible after clang is working again
> (due to your fix here), and I'm up to 30+ patches, and probably only a
> few more to go to get all of them.

Thanks! There're really a lot of those.

>
> I'm expecting to post the series soon, hopefully this week. And I'm
> thinking maybe I should carry your patch as the first one in the series,
> in order to ensure it gets picked up. Or, I can just refer to it as a
> prerequisite in the cover letter.

Correct me if I'm wrong, but intuitively 30+ patches that touch selftests
from many different subsystems do not sound like something that is going
to be merged fast. Since I'm also planning to send a separate series that
removes explicit setting of `ARCH` from subsystem Makefiles (if they do it
for the sole purpose of working around this issue) I'd prefer to leave this
patch separate for easier reference and potentially faster merging.

Referring to it should hopefully be sufficient to prevent it from being
forgotten.

	- Best Valentin

>
>
> thanks,