Re: [PATCH 1/1] drm/format-helper: Add KUnit tests for drm_fb_xrgb8888_to_rgb332()

[Daniel Latypov]

On Mon, Jun 6, 2022 at 6:57 AM Javier Martinez Canillas
<javierm@xxxxxxxxxx> wrote:
>

<snip>

> >>>>  $ ./tools/testing/kunit/kunit.py run --kunitconfig=drivers/gpu/drm \
> >>>>          --kconfig_add CONFIG_VIRTIO_UML=y \
> >>>>          --kconfig_add CONFIG_UML_PCI_OVER_VIRTIO=y
> >>>
> >>> It's not clear to me why you would need VIRTIO here? The Kunit config
> >>> file should be enough to run the tests properly
> >>>
> >>
> >> It's needed or otherwise KUnit will complain with:
> >>
> >> ./tools/testing/kunit/kunit.py run --kunitconfig=drivers/gpu/drm/.kunitconfig
> >> [15:47:31] Configuring KUnit Kernel ...
> >> Regenerating .config ...
> >> Populating config with:
> >> $ make ARCH=um O=.kunit olddefconfig
> >> ERROR:root:Not all Kconfig options selected in kunitconfig were in the generated .config.
> >> This is probably due to unsatisfied dependencies.
> >> Missing: CONFIG_DRM=y, CONFIG_DRM_KUNIT_TEST=y
> >> Note: many Kconfig options aren't available on UML. You can try running on a different architecture with something like "--arch=x86_64".
> >>
> >> The following works correctly but it won't use User Mode Linux:
> >>
> >> ./tools/testing/kunit/kunit.py run --kunitconfig=drivers/gpu/drm/.kunitconfig --arch=x86_64
> >
> > But then, can't we add them to .kunitconfig?
> >
>
> That's what I asked in the previous RFC too. Daniel mentioned that it shouldn't
> go there because is platform specific (AFAIU, one might want to test it on x86,

Slight correction, it was David who explicitly suggested it shouldn't
go in there.
https://lists.freedesktop.org/archives/dri-devel/2022-June/357611.html

> aarch64, etc) but then I asked why we couldn't have a arch/um/.kunitconfig.
>
> The answer was that's not that simple and some agreement on how to do it is needed:

I'm a bit more in favor of having UML-specific options in the drm
.kunitconfig file, but I agree it's a bit unclear.
If people want to easily run with --arch=x86_64 or others, then
they're indeed a liability.

Another option is to perhaps explicitly name the .kunitconfig file
something like drm/uml.kunitconfig, which doesn't solve the problem
but makes it less of a footgun.

Stepping back, I feel like perhaps a cleaner answer lies in adding a
new Kconfig option that selects CONFIG_UML_PCI_OVER_VIRTIO under UML
and just CONFIG_PCI otherwise.
But that's a bigger discussion still.

Daniel