[Intel-gfx] [v8, 2/4] drm/panel: set display info in panel attach
Sean Paul
sean at poorly.run
Tue Oct 8 15:02:57 UTC 2019
/snip
> > > > > diff --git a/include/drm/drm_panel.h b/include/drm/drm_panel.h
> > > > > index d16158deacdc..f3587a54b8ac 100644
> > > > > --- a/include/drm/drm_panel.h
> > > > > +++ b/include/drm/drm_panel.h
> > > > > @@ -141,6 +141,56 @@ struct drm_panel {
> > > > > */
> > > > > const struct drm_panel_funcs *funcs;
> > > > >
> > > >
> > > > All these new added members seems dupliated with drm_display_info,
> > > > So I think, can we add a new drm_plane_funcs func:
> > > >
> > > > int (*set_display_info)(struct drm_panel *panel,
> > > > struct drm_display_info *info);
> > >
> > > I don't disagree personally, since I originally wrote it this way, but
> > > 2 maintainers, Daniel Vetter and Thierry Reding, requested that it be
> > > changed. Unless that decision is reversed, I won't be changing this.
> > >
> >
> > Reading back the feedback on v1, I don't think anyone said they were against
> > storing a drm_display_info struct in drm_panel (no one really weighed in on
> > it one way or another). IMO duplicating a bunch of fields feels pretty icky.
>
> Thanks for the review. Should we duplicate the entire struct, or just
> create a sub-struct, say, physical_properties that will be part of
> drm_display_info and drm_panel?
That's a good idea, but I think you can use the entire struct. Everything has
reasonable default values and I think it might be difficult to draw a line in
the sand as to which fields will or won't be useful for panels going forward.
Best for drivers to just treat the info in drm_display_info as best effort and
have good defaults IMO.
Sean
>
> >
> > I think most fields in drm_display_info have pretty reasonable defaults, so I'd
> > recommend just storing a copy in drm_panel. As Thierry suggests, we can
> > populate the dt parts in probe (orientation) and then copy over all or a subset
> > of the struct to connector on attach.
> >
> > Sean
> >
> > > >
> > > > Then in drm_panel_attach(), via this interface the specific panel
> > > > driver can directly set connector->display_info. like
> > > >
> > > > ...
> > > > if (panel->funcs && panel->funcs->set_display_info)
> > > > panel->funcs->unprepare(panel, connector->display_info);
> > > > ...
> > > >
> > > > Thanks
> > > > James
> > > >
> > > > > + /**
> > > > > + * @width_mm:
> > > > > + *
> > > > > + * Physical width in mm.
> > > > > + */
> > > > > + unsigned int width_mm;
> > > > > +
> > > > > + /**
> > > > > + * @height_mm:
> > > > > + *
> > > > > + * Physical height in mm.
> > > > > + */
> > > > > + unsigned int height_mm;
> > > > > +
> > > > > + /**
> > > > > + * @bpc:
> > > > > + *
> > > > > + * Maximum bits per color channel. Used by HDMI and DP outputs.
> > > > > + */
> > > > > + unsigned int bpc;
> > > > > +
> > > > > + /**
> > > > > + * @orientation
> > > > > + *
> > > > > + * Installation orientation of the panel with respect to the chassis.
> > > > > + */
> > > > > + int orientation;
> > > > > +
> > > > > + /**
> > > > > + * @bus_formats
> > > > > + *
> > > > > + * Pixel data format on the wire.
> > > > > + */
> > > > > + const u32 *bus_formats;
> > > > > +
> > > > > + /**
> > > > > + * @num_bus_formats:
> > > > > + *
> > > > > + * Number of elements pointed to by @bus_formats
> > > > > + */
> > > > > + unsigned int num_bus_formats;
> > > > > +
> > > > > + /**
> > > > > + * @bus_flags:
> > > > > + *
> > > > > + * Additional information (like pixel signal polarity) for the pixel
> > > > > + * data on the bus.
> > > > > + */
> > > > > + u32 bus_flags;
> > > > > +
> > > > > /**
> > > > > * @list:
> > > > > *
> > >
> > > Thanks for the review
> >
> > --
> > Sean Paul, Software Engineer, Google / Chromium OS
--
Sean Paul, Software Engineer, Google / Chromium OS
More information about the Intel-gfx
mailing list