Questions Arise Over NVIDIA's Fence Sync Support

As reported this morning, RandR 1.4 is now ready for X.Org Server 1.10 after this next xorg-server release's merge window was kept open to allow this work to be finished and land in the Git tree. RandR 1.4 brings per-CRTC pixmaps, sprite transforms, and a new RandR request that will hopefully allow NVIDIA's binary driver to finally support RandR 1.2+ features. While the merge window is kept open for a short period of time, NVIDIA has been trying to put in their Fence Sync support for the X Server. While it looks like these patches will still be accepted, some objections and questions have arose over this open-source contribution by NVIDIA.

Red Hat's Owen Taylor started out by asking about a broad overview on NVIDIA's Fence Sync, seeing as he is the maintainer of Mutter, the GNOME 3.0 compositing window manager that uses Clutter. "There's already a lot of magic voodoo dances around both Damage and Texture-From-Pixmap, what extra incantations does this add to the picture?" Owen further noted, "I can understand each individual step of the magic voodoo dance, but when I go away from the individual problems and come back 6 months later, I have to work it all out again. And there's a strong sense that only particular code paths that actually are in use are tested and anything else probably doesn't work, at least on some drivers."

Once NVIDIA's James Jones gave the usual spiel on the X Fence Sync Support, Owen came back to seek other clarifications and say, "It worries me to see a pretty complex, somewhat expensive band-aid going in *without* knowing more about that long term picture." Keith Packard then hopped in to say, "The trouble here is that all of the drivers we can look at don't need this to work efficiently, so there's no way to evaluate the design to see if it makes any sense at all. Requiring changes to all compositing managers just to support one driver seems like a failure in design, and it will be fragile as none of it will matter until the compositing manager is run against that driver."

James Jones ended up responding that the technical issues brought up by Owen and Keith were already evaluated and discussed by those developers active on IRC. "For my part, I regret that this work may be seen by some as an NVIDIA-backed attempt to dump functionality only needed by a closed-source driver on an
under-staffed open-source project. That is not my intention, and while the discussion has centered around the one initial application of the changes that currently only benefits our driver, I do think fence objects will be generally useful on all platforms. Further, I'll be around to support this code, answer questions about it, etc, as long as NVIDIA keeps paying me, and probably even if they don't just because I find it interesting...I disagree that this is a failure of design. I always try to strike a balance between the most efficient solution for the end users and the constraints of development and maintenance. And we're used to being the odd man out. We regularly need to provide patches or at least point out bugs in many projects that only happen on our drivers for whatever reason (We support a different set of GL extensions, we accelerate a different set of X render operations, older apps relying on non-compliant SGI weirdness, newer apps relying on non-compliant/undefined DRI/mesa/whatever behaviour, etc.). It's not in our interest to needlessly burden our users, but different isn't necessarily always bad."

Keith Packard though came back with a new proposal for NVIDIA. After stomaching the criticism and it being raised so late (days before the merge window closed and months after the sync fences first came), James prepared a counter offer.

Under this latest offer, the X Fence Sync support would go into X.Org Server 1.10, but then as part of the X.Org Server 1.11 release there would be a new version of X Damage. If not X Server 1.11, even possibly a point release in the 1.10 series. This update would conditional-ize the language and allow time for testing the new DamageSubtractAndTrigger functionality to see whether it adds any significant latency. If DamageSubtractAndTrigger turns out not to be bad, then it would be advertised that compositing window managers could hook into this support. James also challenges an open-source driver to implement the explicit mode in this new extension to see whether it's faster than the status quo while still refraining from any desktop tearing.

In the end, it looks like things will be sorted out and ready for the X.Org Server 1.10 release.