[PATCH] drm/fourcc: add LINEAR modifiers with an exact pitch alignment

[Simona Vetter]

On Wed, Dec 18, 2024 at 10:44:17AM +0100, Michel Dänzer wrote:
> On 2024-12-17 12:03, Brian Starkey wrote:
> > On Tue, Dec 17, 2024 at 11:13:05AM +0000, Michel Dänzer wrote:
> >> On 2024-12-17 10:14, Brian Starkey wrote:
> >>
> >>> Modifiers are meant to describe framebuffers, and this pitch alignment
> >>> requirement isn't really a framebuffer property - it's a device
> >>> constraint. It feels out of place to overload modifiers with it.
> 
> FWIW, KMS framebuffers aren't the only use case for sharing buffers
> between devices.
> 
> 
> >>> I'm not saying we don't need a way to describe constraints to
> >>> allocators, but I question if modifiers the right mechanism to
> >>> communicate them?
> >> While I agree with your concern in general, AFAIK there's no other
> >> solution for this even on the horizon, after years of talking about
> >> it. The solution proposed here seems like an acceptable stop gap,
> >> assuming it won't result in a gazillion linear modifiers.
> > 
> > UAPI is baked forever, so it's worth being a little wary IMO.
> > 
> > This sets a precedent for describing constraints via modifiers. The
> > reason no other proposal is on the horizon is because describing the
> > plethora of constraints across devices is a hard problem; and the
> > answer so far has been "userspace needs to know" (à la Android's
> > gralloc).
> > 
> > If we start down the road of describing constraints with modifiers, I
> > fear we'll end up in a mess. The full enumeration of modifiers is
> > already horrendous for parameterized types, please let's not
> > combinatorially multiply those by constraints.
> 
> I agree there's a slippery slope.
> 
> That said, linear buffers are special in that they're the only
> possibility which can work for sharing buffers between devices in many
> cases, in particular when the devices are from different vendors or even
> different generations from the same vendor.
> 
> So as long as device vendors don't get too creative with their linear
> pitch alignment restrictions, it still seems like this might be workable
> stop-gap solution for that specific purpose alone, until a better
> solution for handling constraints arrives.
> 
> 
> > P.S. "is the only modifier that has a chance of not working" is
> > fundamentally false.
> 
> My reading of that part of the comment is that pitch alignment shouldn't
> be an issue with non-linear modifiers, since the constraints for pitch
> should be part of the modifier definition. Maybe that could be clarified
> in the comment.

Yeah with all other modifiers pitch alignment or other alignment/sizing
requirements are generally implied. Mostly by stuff like tile size, but
there's others where the hw requirement flat out is that the buffer must
have a power-of-two stride (and maybe we should document these when they
pop up, but the only one I know of are the legacy i915 modifiers, which
are kinda busted anyway for interop).

For that reason I think linear modifiers with explicit pitch/size
alignment constraints is a sound concept and fits into how modifiers work
overall.
-Sima

> 
> 
> -- 
> Earthling Michel Dänzer       \        GNOME / Xwayland / Mesa developer
> https://redhat.com             \               Libre software enthusiast

-- 
Simona Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch