Re: CVS commit: src

[Taylor R Campbell <campbell+netbsd-source-changes-d%mumble.net@localhost>]

> Date: Thu, 10 Aug 2023 07:42:53 +1000
> from: matthew green <mrg%eterna.com.au@localhost>
> 
> > Log Message:
> > viadrmums(4): build legacy VIA DRM UMS driver module for amd64.
> >
> > This driver is not built-in by default, thus loadable module can help (un)lucky
> 
> if it works, why isn't it in GENERIC as well as a module?

Couple reasons:

1. I never adequately tested it.  I started X a few years ago (and a
   couple drm updates ago) but that's it.  (Also only on 32-bit VIA.)
   Maybe andvar can test it more extensively now, though -- glxgears,
   x11perf, firefox, &c.

   Even better if we can get a sampling of code coverage, e.g. maybe
   with dtrace to count function calls at least, to confirm it's
   actually getting exercised.

2. It's a legacy UMS driver, which is inherently sketchy (display
   configuration requires granting userland direct access to pci and
   device registers) and increasingly unmaintained even upstream.

   So I put in some infrastructure to allow UMS drivers to be loaded
   dynamically, so it wouldn't be necessary to have sketchy business
   like that in the kernel by default.

   I kind of intended to give the same treatment to the other legacy
   UMS drivers like mga (matrox), r128, sis, tdfx, &c., but it's been
   a very low priority so I haven't gotten a round tuit (and I only
   have hardware for mga and that's in a build server I don't like to
   mess with).