[v4] drm: two planes with the same zpos have undefined ordering

[Simon Ser]

On Tuesday, October 8, 2019 6:16 PM, Daniel Vetter <daniel at ffwll.ch> wrote:

> On Tue, Oct 8, 2019 at 5:11 PM Simon Ser contact at emersion.fr wrote:
>
> > On Tuesday, October 8, 2019 6:03 PM, Daniel Vetter daniel at ffwll.ch wrote:
> >
> > > > > > > In any case, this doesn't change the patch itself. Probably something worth
> > > > > > > mentionning in the commit message.
> > > > > >
> > > > > > Yes, recording these use cases would be very nice.
> > > > >
> > > > > There's a lot more hw that does the same tricks (qualcom is one for sure).
> > > > > Imo before we encode this we should make sure that:
> > > > > a) everyone is happy with this new uapi
> > > >
> > > > Sorry, what new UAPI?
> > > > We're just trying to make the documentation match what currently
> > > > happens, right?
> > >
> > > It's so much new uapi that I've sent out a few reverts for 5.4 (it
> > > landed in that merge window) to undo the stuff the arm driver team has
> > > done (it didn't come with userspace, proper discussion on dri-devel,
> > > docs or testcases in igt). I also just spotted that a leftover is that
> > > arm/komeda still computes its own version of normalized_zpos, which
> > > probably should be ditched too (it's not actually different from the
> > > one in helpers without the reverted uapi).
> > > So yeah, separate patch :-)
> >
> > I don't get it. Do you want to split the docs changes for user-space,
> > only keeping the doc changes for drivers in this patch?
> > User-space could already see duplicate zpos because of the non-atomic
> > API. I don't think this introduces a new uAPI.
>
> I'm talking specifically about the "duplicated zpos values indicate
> special cloned planes like in the arm example" clarification. Not
> about splitting the zpos documentation any more, that's indeed just
> documenting existing uapi. But assigning the special meaning to
> duplicated zpos values (instead of just "can happen because non-atomic
> legacy userspace"), that is new uapi. Especially if we allow
> duplicated zpos for immutable properties (afaik that doesn't exist
> yet).

Oh, I see. That makes sense, will send a new version.