[PATCHv3 00/41] OMAPDSS: DT support v3

Tony Lindgren tony at atomide.com
Tue Mar 11 12:28:16 EDT 2014


* Tomi Valkeinen <tomi.valkeinen at ti.com> [140311 03:19]:
> On 10/03/14 17:41, Tony Lindgren wrote:
> 
> >>> How about do a pull request for just the .dts changes against current
> >>> omap-for-v3.15/dt branch ASAP for me so I can pull it in? That is assuming
> >>> that just the .dts changes alone won't break anything.
> >>
> >> Unfortunately they do break. At least pinmuxing is moved from the global
> >> definitions to be handled by the respective display drivers, and there
> >> are some regulator_name hacks that the DT patches remove.
> > 
> > OK. Will that cause regressions for omap3 as that's still also booting
> > in legacy mode?
> 
> No, I don't think so. The problems revolve around the pdata-quirks, with
> current DSS support when booting with DT. It's rather difficult to split
> the series so that it could be merged freely in multiple parts.
> 
> If I split the series into three parts: 1) .dts changes 2) main DSS DT
> changes 3) removal of pdata-quirks etc, I run into problems. Both 1) and
> 2) work fine individually, but when both are merged, there are two
> competing systems, the proper DT stuff and the pdata-quirks stuff. And I
> haven't found out a simple way to manage that, as the whole display
> support is split into multiple independent devices.
> 
> One option would be to combine 1) and 3), so that when they are merged,
> there would be proper DT data, and the pdata-quirks would be removed so
> that it wouldn't be messing everything up. But that would, of course,
> mean that after merging 1+3, the display wouldn't work on those boards
> that rely on pdata-quirks, until 2) is merged.

OK, best to keep the series together then.
 
> >> I think those changes could be removed from my patches, and then added
> >> back later when everything else has been merged.
> > 
> > Or you could have a second branch that also merges in omap-for-v3.15/dt
> > branch that you can send later towards the merge window after the arm-soc
> > changes have been merged. If you want to do that, then feel free to add
> > my ack also for the .dts changes:
> > 
> > Acked-by: Tony Lindgren <tony at atomide.com>
> > 
> > If however those changes get postponed to v3.16, it's best that you'll
> > redo the branch as I'm sure we'll have other merge issues for v3.16.
> 
> Ok. At the moment I feel that the easiest option would be to keep the
> DSS DT series as it is, but merge omap-for-v3.15/dt to it and solve the
> conflicts. I'd keep that branch separate from the fbdev changes, so that
> I could send the DSS DT pull request later, when arm-soc has been pulled.

OK makes sense to me.
 
> >> The bigger issue is that suddenly there's lots of discussion about the
> >> display DT bindings. If those are not resolved very soon, I guess I have
> >> no choice but to again skip the merge window for the DSS DT changes.
> > 
> > OK, some of these more bindings can take easily six months to reach
> > some kind of resolution. You may be able to use TI specific unstable
> > bindings meanwhile if that makese sense.
> 
> Yep. I don't know... If the whole port/endpoint system that I currently
> use is changed totally, it might be painful to support both the TI
> specific bindings with the old format and the newer format.

OK that's up you guys in the display land, I have not followed the
latest bindings discussion.

Regards,

Tony



More information about the linux-arm-kernel mailing list