[PATCH 0/2] drm: revert some framebuffer API tests

[Guenter Roeck]

On 9/25/24 05:59, Maxime Ripard wrote:
...

>> All I'm saying it generates an extra burden for current real world CI
>> systems that run tests on a massive scale and even have paid
>> maintainers. It's not trivial to sort out expected and unexpected
>> warnings, and keep the filters updated every time the warnings
>> change. It's not without a cost. And the end result might just be that
>> the unit tests won't get run at all. Or warnings just get completely
>> ignored for kunit tests.
> 
> I realise it must take a significant amount of resources, but it's also
> self inflicted. You could also stop looking for warnings when running
> kunit.
> 

FWIW, that doesn't work if kunit tests are built into the kernel and
run when booting the system. It also doesn't work well if kunit tests
trigger a WARN_ONCE() because that causes any real subsequent warning to be
dropped. Also, ignoring warnings while running a kunit test means that _real_
unexpected warnings triggered by those tests will be ignored as well.

Guenter