[RESEND PATCH 2/2] modetest: only select plane with matching format

[Emil Velikov]

On 11 June 2015 at 14:43, Tobias Jakobi <tjakobi at math.uni-bielefeld.de> wrote:
> Sorry for the late reply!
>
> On 2015-05-28 14:25, Emil Velikov wrote:
>>>
>>> If you
>>> create your primary plane with a XRGB-type format then this patch
>>> doesn't change anything. If you create it with an ARGB-type format then
>>> I can assume that the user wants an alpha-channel (he explicitly asks
>>> for it).
>>>
>> True. But the test might be assuming that alpha is always 255, thus it
>> expects that the output for XRGB and ARGB type formats is the same.
>
> I still don't understand this part at all. What "test" are we actually
> talking about? Do you mean some test application that somehow internally
> uses output from libdrm's modetest? I have trouble seeing how this should
> work at all? If this patch changes anything then it's visual output on the
> screen (assuming that the DRM driver somehow interprets the content of the
> 'X' channel). So only something that a human inspecting the screen could
> detect.
>
> Or am I missing something here?
>
I'm a master of finding some elaborate cases which few sane person
would ever think of. The case I'm thinking is about a hypothetical
broken test(ing scenario), and as such it's not our issue :-) So don't
bother with it at all.

-Emil