[Intel-gfx] [PATCH v3 6/9] drm/i915/mso: add splitter state readout for platforms that support it

[Ville Syrjälä]

On Tue, Mar 02, 2021 at 12:25:00PM +0200, Jani Nikula wrote:
> On Mon, 22 Feb 2021, Jani Nikula <jani.nikula at intel.com> wrote:
> > On Mon, 22 Feb 2021, "Shankar, Uma" <uma.shankar at intel.com> wrote:
> >>> diff --git a/drivers/gpu/drm/i915/display/intel_display_types.h
> >>> b/drivers/gpu/drm/i915/display/intel_display_types.h
> >>> index 71611b596c88..5564db512d22 100644
> >>> --- a/drivers/gpu/drm/i915/display/intel_display_types.h
> >>> +++ b/drivers/gpu/drm/i915/display/intel_display_types.h
> >>> @@ -1161,6 +1161,13 @@ struct intel_crtc_state {
> >>>  		u8 pipeline_full;
> >>>  		u16 flipline, vmin, vmax;
> >>>  	} vrr;
> >>> +
> >>> +	/* Stream Splitter for eDP MSO */
> >>> +	struct {
> >>> +		bool enable;
> >>> +		u8 link_count;
> >>> +		u8 pixel_overlap;
> >>> +	} splitter;
> >>
> >> For DSI which also has this in common along with MSO, may be we can
> >> take these link_count and pixel_overlap out of splitter which is more
> >> of a MSO feature. Thoughts ?
> >
> > Ville suggested the same I think.
> 
> Coming back to this. DSI does not actually use crtc state for this
> currently. But it does use the display stream splitter. The register is
> different, but the functionality is roughly the same.
> 
> I suggest we keep the "splitter" substruct as above, and convert DSI
> code to use it in follow-up.
> 
> In early versions of the patch the substruct was called "mso"; I think
> "splitter" is better, and captures both MSO and DSI cases.

The "splitter" name is rather icl+ specific. Earlier DSI or LVDS do not
have such a hw block (at least in name).

-- 
Ville Syrjälä
Intel