On Fri, Dec 13, 2019 at 01:17:59PM +0200, Tomi Valkeinen wrote: > Hi Daniel, > > On 13/12/2019 12:30, Daniel Vetter wrote: > > > > +DRM DRIVERS FOR TI KEYSTONE > > > +M: Jyri Sarha <jsarha@xxxxxx> > > > +M: Tomi Valkeinen <tomi.valkeinen@xxxxxx> > > > +L: dri-devel@xxxxxxxxxxxxxxxxxxxxx > > > +S: Maintained > > > +F: drivers/gpu/drm/tidss/ > > > +F: Documentation/devicetree/bindings/display/ti/ti,k2g-dss.yaml > > > +F: Documentation/devicetree/bindings/display/ti/ti,am65x-dss.yaml > > > +F: Documentation/devicetree/bindings/display/ti/ti,j721e-dss.yaml > > > +T: git git://anongit.freedesktop.org/drm/drm-misc > > > > Is the plan to also move other TI drivers over (like tilcdc) or just an > > experiment to see what happens? Asking since if eventually omapdrm moves > > that might be a bit much (or at least needs a discussion first). > > Hmm, yes, I think we should have a plan for these. > > tilcdc: small, old driver, and I don't see much changes for it. The HW is > very different from the ones supported by omapdrm and tidss (the two of > which have many commonalities). I think drm-misc is fine for tilcdc. > > omapdrm: big changes going on for now, but after the rewrite to get rid of > omapdrm specific drivers is done, I expect the patch count to drop, as the > HW is "legacy". > > tidss: the "new" driver, which should get most attention in the future (after omapdrm rewrite). > > All in all, sometimes there have been very few patches for many months, and > then sometimes there's a big series. Yeah occasionally big series is totally fine for -misc. We're pulling in pretty big amounts of refactoring all the time, so really doesn't make a difference. > I haven't seen a need to have a maintained branch for omapdrm, as multiple > people working on conflicting items has been very rare (there aren't that > many people working on omapdrm). So I've picked patches to my private > branch, which I have rebased as needed. And then I've either pushed via > drm-misc if there's just a few patches, or sent a pull request if there's a > lot. > > Is such a mixed model ok? > > I'm not sure how much is too much for drm-misc, but probably omapdrm and > tidss combined (if pushing everything always via drm-misc) is a bit too > much. So perhaps a maintained TI tree would be an option too, and pushing > everything for omapdrm and tidss via that tree. tbh I dunno either when too much is too much for -misc. I think tilcdc and omapdrm should both easily fit (maybe after the big rewrite). If the shiny new one here becomes too big we can always reconsider. I think the really big drivers like amdgpu or intel don't belong in -misc, that one is clear. In between I guess we'll see. > I'm fine with all options, so I think we can go with whatever is most > acceptable from DRM maintainer point of view. Whatever floats the boat, with a preference for not having tiny trees (simply because those pull request tend to get lost in the noise). -- Daniel Vetter Software Engineer, Intel Corporation http://blog.ffwll.ch
