Re: [PATCH v2 4/4] drm/format-helper: Add KUnit tests for drm_fb_xrgb8888_to_rgb565()

[José Expósito]

On Wed, Aug 10, 2022 at 09:41:18AM -0700, Daniel Latypov wrote:
> On Sun, Jul 17, 2022 at 10:01 AM José Expósito
> <jose.exposito89@xxxxxxxxx> wrote:
> >
> > José Expósito <jose.exposito89@xxxxxxxxx> wrote:
> > > I already fixed the warning and added the reviewed by tags, however, I
> > > noticed that rebasing the series on the latest drm-misc-next show this
> > > error:
> > > [...]
> >
> > Sorry for the extra email. I forgot to mention that the error is only
> > present in UML. Running in other architectures works as expected.
> > Tested on x86_64 and PowerPC.
> 
> Can you take a look at the raw output?
> 
> It would be .kunit/test.log (or replace .kunit with w/e --build_dir
> you're using).
> You could also run the test with --raw_output to have kunit.py print
> that out instead.
> We might want to compare the output on UML vs x86 and see what looks suspicious.
> 
> These errors
>   [ERROR] Test: xrgb8888_to_rgb332_test: missing subtest result line!
> means that kunit.py can't parse the KTAP output.
> 
> It could be that the output is mangled for some reason.
> I recall running into a UML-specific issue with output mangling on an
> old kernel fork. I doubt it's related to this one, but it shows that
> it's possible there could be something going on.
> 
> -Daniel

Hi!

Sorry for not clarifying the error in this thread.
I fixed it in v3 (already merged).

The issue was in my implementation. I was overwriting a few bytes of
memory due to an out of bounds bug. Thanks to the crash I mentioned,
I detected it before the code got merged.

Thanks a lot for the pointers Daniel, the next time I'll check
.kunit/test.log for usefull information.

Jose