[PATCH 2/2] drm/xe: drop display/ subdir from include directories

[Lucas De Marchi]

On Tue, Jan 30, 2024 at 11:19:45AM +0200, Jani Nikula wrote:
>On Tue, 23 Jan 2024, Jani Nikula <jani.nikula at intel.com> wrote:
>> On Tue, 23 Jan 2024, Lucas De Marchi <lucas.demarchi at intel.com> wrote:
>>> On Mon, Jan 22, 2024 at 07:39:56PM +0200, Jani Nikula wrote:
>>>>On Mon, 22 Jan 2024, Lucas De Marchi <lucas.demarchi at intel.com> wrote:
>>>>> the downside of patch 1 is now that core xe code can include any of the
>>>>> display/ headers, but only xe_display.h is acceptable to keep the
>>>>> interface sane.
>>>>
>>>>I'd say xe_display.h remains the only interface towards xe core, I don't
>>>>think this patch changes that, and its location doesn't really give any
>>>>guarantees. It's been a matter of sticking display/ in the include
>>>>anyway, and enforcing that is a matter of maintainer vigilance.
>>>
>>> I just thought the previous split was slightly better: No code in xe
>>> should include display/ and should rather use the xe_display.[hc]
>>> interface.
>>>
>>> Now with #include "display/xe_display.h" spread throughout the code,
>>> this could serve as example for people to start including stuff they
>>> shouldn't.
>>>
>>> I'm not entirely opposed, so if you and others agree, please go ahead.
>>
>> Well, at the moment you can include *anything* from under display/
>> *without* the prefix, because it's all in the include path. (Although
>> that will fail for DRM_XE_DISPLAY=n, but does CI even build that combo
>> regularly?)
>>
>> I'm fine with dropping this too. I just think it makes it easier to
>> check that nothing outside of display/ does any displayish things.
>>
>> Your call.
>
>I've held off on pushing this. What's the verdict?

Let's push this. If it doesn't work out due to people including other
headers, we may think of a way to block that without reverting.

Acked-by: Lucas De Marchi <lucas.demarchi at intel.com>

Lucas De Marchi

>
>BR,
>Jani.
>
>
>-- 
>Jani Nikula, Intel