Re: Prefer nvidia over nv when available
On Wed, Oct 03, 2007 at 11:08:29AM +0200, Free Ekanayaka wrote:
> Hi David,
>
> |--==> David Nusinow writes:
>
> DN> On Mon, Oct 01, 2007 at 11:30:26PM +0200, Free Ekanayaka wrote:
> >>Hi all,
> >>
> >>is there a way to instruct the script(s) that generate the xorg.conf
> >>file at installation time to prefer nvidia over nv when the first is
> >>available? (and possibly to mark the proper packages for installation
> >>when not already installed).
>
> DN> Not currently but it is definitely something I'm planning to do. Since I'm
> DN> going to scrap that section of the debconfage soon, adding the feature
> DN> there is pretty useless. The solution we currently have is a temporary one
> DN> due to changes upstream, but I'll be implementing things upstream over the
> DN> next few weeks, and I'll follow it up with this sort of feature so that
> DN> nvidia can override nv by default when they're both installed (if the
> DN> nvidia developers decide that's what they want anyway).
>
> Thanks. Just a few questions, if you already know the answer. Beside
> the regular nvidia-glx package and the relevant kernel module package
> would this mechanism also support the various other packages of the
> nvidia driver currently available in sid (e.g. legacy-71xx and
> legacy-96xx)?
>
> These different nvidia packages conflict with each other, so it might
> happen that the correct nvidia package is not installed by the time
> the xorg packages run their debconf script. How would this situation
> be handled?
The idea for the future is that all the drivers will export a symbol table
of the PCI ID's they support. The server will scan them for drivers
supporting the required PCI ID. My plan is to have them export an
additional symbol table consisting of driver names that they override. So
nvidia would export an override table with "nv" in it.
The notable thing about this is that it exists entirely outside the
packaging system, so it's entirely up to upstream. In the case of the
various nvidia drivers, if the legacy drivers are updated to include these
symbols then they'll be automatically loaded if they're available. If not,
then the server will load something else (nv if available, or an
arch-specific fallback like vesa if not).
So, if Nvidia decides to update their drivers, they'll get the benefits of
this. The PCI ID symbol table thing already exists upstream in the xserver
master branch, but the code to auto-load the driver based on it doesn't
exist yet, but that's my goal for next week. In the worst-case scenario,
Nvidia won't add the necessary symbols and users will just have to edit
their xorg.confs the same way they did in the past.
- David Nusinow
Reply to: