Initial XWayland Support Looks To Land In X.Org Server 1.16

It looks like with the X.Org Server 1.16 release this summer there might finally be XWayland support for providing legacy X11 support within Wayland environments.

Kristian Høgsberg sent the XWayland series out on Tuesday to the Xorg-devel mailing list. These patches lay the prep work for supporting XWayland with the necessary alterations to the server code and then land the XWayland DDX support.

Originally XWayland served as an X.Org module by which modified DDX hardware drivers could be loaded on the system so they could offer their 2D support. However, given the advancements of GLAMOR, that is being used instead so we can have one unified XWayland DDX without the need for having patched drivers for hardware support and should work on just about any platform that has OpenGL support. GLAMOR tends to still be slower than the hand-written 2D paths in the xf86-video-ati and xf86-video-intel DDX, but there's still a lot of optimizations and code rewrites taking place of the code now that it's moved from being a standalone library to living within the X.Org Server.

The XWayland DDX is now a much cleaner approach, only a few thousand lines of code, and much more self-contained. It will be interesting to see if Canonical now tries to go for a GLAMOR-based approach within their XMir design, especially given their lack of interest by upstream DDX drivers to accept Mir support patches.

While this patch series for review has the XWayland DDX, this series doesn't include the code for GLAMOR integration and is just software-only for the time being. Direct Rendering support is also missing. Kristian will be sending that out as a separate patch series, which he hopes will still be in time for the 1.16 release. The current patches for review and comment are via the X.Org development list.

X.Org Server 1.16 should be released in July and besides hopefully having this XWayland code will also be less compiler warnings, and all the GLAMOR goodness.

In my performance testing of XWayland just a few months ago to see the performance overhead of running 2D apps or even legacy 3D games with OpenGL, the performance impact is small with the newest code and much better than the rough support when XWayland was initially unveiled.