On Tue, Feb 07, 2017 at 04:34:44PM +0100, Maxime Ripard wrote: > On Mon, Feb 06, 2017 at 12:29:31PM +0100, Noralf Trønnes wrote: > > > > Den 06.02.2017 11.39, skrev Maxime Ripard: > > > Hi Noralf, > > > > > > On Fri, Feb 03, 2017 at 07:48:51PM +0100, Noralf Trønnes wrote: > > > > Den 03.02.2017 10.59, skrev Maxime Ripard: > > > > > Hi, > > > > > > > > > > Here is an attempt at supporting the ST7789V LCD controller from Sitronix. > > > > What happens if there's another panel driven by ST7789V that needs > > > > a different controller initialization? > > > You know those panels / controllers much better than I do, but why > > > would that be the case? > > > > > > > Maybe it's better to name it after the panel, not the controller. > > > I guess you could also use that panel directly without the controller? > > > > A controller can drive many different panels that can require different > > initializations. I faced that with staging/fbtft, when I wrote > > controller drivers having initialization code, and then came across > > displays with the same controller but with a different initialization. > > > > Trying to write controller drivers for these controllers is very > > difficult with all the possible permutations. On top of that we have > > those undocumented commands/registers. > > > > Some panels come with embedded controllers, in which case it makes > > sense to write a driver for the panel. > > > > But if the panel and controller are separate, then I don't know. Maybe > > the chance of coming across two uncompatible ST7789V and panel > > combinations in drm/panel is extremly low. > > Hmm, I see. If we ever come across that case, I guess we could just > add new optional properties to override the current sequence. I think if that ever happens it's probably best to split out ST7789V code into a helper library that has functions which take parameters. Then we can simply call those functions with parameters specific to a panel from a panel-specific driver. Trying to fit all that into device tree properties is likely going to end up being really messy. Thierry
Attachment:
signature.asc
Description: PGP signature
