[Intel-gfx] [PATCH 05/16] drm/i915: s/pipe/transcoder/ when dealing with PIPECONF/TRANSCONF

[Jani Nikula]

On Wed, 15 Sep 2021, Ville Syrjälä <ville.syrjala at linux.intel.com> wrote:
> On Wed, Sep 15, 2021 at 01:16:58PM +0300, Jani Nikula wrote:
>> On Mon, 13 Sep 2021, Ville Syrjala <ville.syrjala at linux.intel.com> wrote:
>> > From: Ville Syrjälä <ville.syrjala at linux.intel.com>
>> >
>> > PIPECONF becamse TRANSCONF when HSW introduced the EDP transcoder.
>> > Bigjoiner is making life even more confusing by introducing
>> > a N:1 relationship between pipes and transcoders. In that case
>> > we only enable/configure the transcoder corresponding to the
>> > master pipe. Let's do some renames to make it clear we're dealing
>> > with the transcoder rather than pipe when it comes to
>> > PIPECONF/TRANSCONF.
>> >
>> > I decided to leave the _cpu_ part out from the function/macro
>> > names since the PCH transcoder related stuff already has a
>> > _pch_ in their name. So shouldn't be possible to confuse them.
>> 
>> Wondering about flipping the names to intel_transcoder_enable and
>> intel_transcoder_disable, with a potential move to a separate file.
>
> If you're thinking about intel_transcoder.c then I doubts
> about it making much sense. Seems a bit too specific,
> and the transcoder vs. pipe split has been a rather fluid
> over the years so not all platforms may even fit into the
> same mold.
>
> I think what we want to do to the modeset code in
> intel_display.c is mainly just split on the high level vs.
> low level boundary somewhere. But splitting the low level
> code further is probably going to require more thought.

Yeah, just throwing ideas around, really.

We could start with a new intel_modeset.[ch] and start moving the high
level modeset functions there? Because moving everything else out from
intel_display.c is going to take forever, so it's perhaps easier to move
the high level stuff instead. *shrug*

BR,
Jani.


-- 
Jani Nikula, Intel Open Source Graphics Center