Hi Sam,
On Thu, Aug 08, 2019 at 06:52:53PM +0200, Sam Ravnborg wrote:
Hi Laurent.
As I said in another mail, you have managed to keep me busy...
I took a look at this - it seems simple:
- Update drm_panel.get_modes() to take drm_connector as argument, and fix all callers. All callers already have connector available.
- Drop drm_panel_attach(), drm_panel_detach() and update all callers. In reality just drop all code around attach(), detach(). drm_panel_attach(), drm_panel_detach() will be noops when the connector stored in drm_panel is no longer used.
The semantic difference is that we supply the connector when we call drm_panel_get_modes() and not at panel creation time with an drm_panel_attach().
So it should be doable without any migration from one world to the other.
If someone can say "yes it should be that simple", then I will give it a spin.
Looking forward to that :-)
Almost there.... I have all the preparation patches on dri-devel, with positive feedback on most.
And locally I have updated all get_modes() to take drm_connector as argument.
A few drivers access drm_panel->connector, still need to look into this.
And then for drm_panel_attach(), drm_panel_detach() - so far they are kept but changed to take a drm_device*.
Just sharing this so you do not jump at it and duplicate the work. It will take a little time before I can invest time in this again. Will post patches when something is ready for review.
Thanks for the update. Take your time, this isn't blocking me.