I recall John Carmack contributing to some of these drivers, especially the ATI Rage driver. I think it's the driver being removed.<p>Here's an example of his forum posts from 2000 related to working on some of these drivers:
<a href="https://web.archive.org/web/20000816145733/http://lists.sourceforge.net/pipermail/utah-glx-dev/2000-January/author.html" rel="nofollow">https://web.archive.org/web/20000816145733/http://lists.sour...</a><p>Poke around here for more:<p><a href="https://web.archive.org/web/20000815084204/http://lists.sourceforge.net/pipermail/utah-glx-dev/" rel="nofollow">https://web.archive.org/web/20000815084204/http://lists.sour...</a><p><a href="https://web.archive.org/web/20001019083125/http://utah-glx.sourceforge.net/" rel="nofollow">https://web.archive.org/web/20001019083125/http://utah-glx.s...</a>
This is one issue monolithic kernels have unfortunately. Even if you moved the drivers out of tree, you have the issue of the constantly changing kernel API.
I think Rage 128 was released 25 years ago, so seems pretty reasonable to remove them, no?<p>If anyone still wants to maintain it there shouldn't be any hindrance to keep it as an external patchset somewhere..
The code will always exist in the history, should anyone get a bug up their ass and want to port one of these drivers to a modern kernel (and implicitly become the maintainer).
Would it even be possible to get the userspace drivers for these cards working with a kernel this new? Wouldn't you need to compile a really old version of mesa? In that case I'd imagine you'd need all your libraries and programs to be from 20 years ago as well.
GPU support doesn’t bother me as much as CPU support. For GPUs, if it isn’t modern enough it’s effectively useless for most people beyond VGA/SVGA output. Modern user space requires modern GPUs.