[SGVLUG] Xen and nVidia...

Michael Proctor-Smith mproctor13 at gmail.com
Wed Jun 6 10:02:14 PDT 2007


> The problem isn't with the modules, but rather that I'll need an entirely
> different xorg.conf file [or whatever it's called nowadays] that specifies
> whether to use the nvidia module or the more-or-less standard module; as X
> starts, it loads the module(s) it finds
>
> I'm thinking scpm will do this, but it's been a while since I've dealt with
> it (and then, only for reconfiguring my network based on the selected
> configuration)

First are you sure that nvidia does not work with the first
semi-native linux domain. I think it just does not support
para-virtualized domains. Where you can use VMGL
http://www.cs.toronto.edu/~andreslc/xen-gl/

Well I would try patching the binary driver as it is probably just an
issue with the wrapper code and not an actual change to the binary
blob. See if it is stable if it is perfect. If not then you could run
at a differant run level when running under xen vs. native linux.
Under red hat there is a line in /etc/inittab
# Run xdm in runlevel 5
x:5:respawn:/etc/X11/prefdm -nodaemon

You could have this make a differant call, you can have differant
configs/drivers in the same xorg.conf which you can call by name. Also
you can probably use the open nv driver and get more 2d acceleration.

> -----Original Message-----
> From: sgvlug-bounces at sgvlug.net [mailto:sgvlug-bounces at sgvlug.net] On Behalf
> Of John Jefferson Lowry IV
> Sent: Tuesday, June 05, 2007 2:47 PM
> To: SGVLUG Discussion List.
> Subject: Re: [SGVLUG] Xen and nVidia...
>
> You can pass modules information via your boot loader. Grub will at least, I
> am not that learned on Lilo.
>
>
> On 6/5/07, Emerson, Tom (*IC) < Tom.Emerson at wbconsultant.com> wrote:
> > ...doesn't work (per nVidia)
> >
> > My new laptop has an nVidia card, so naturally I want to use the
> > vendor-supplied driver, but it seems it won't work with a "xen" enabled
> > kernel.  The (open)Suse wiki, however, has an article on how to patch
> > the binary driver so that it will work, but that seems a bit dicey to
> > me...
> >
> > So I'm looking for an alternative, and what comes to mind is to somehow
> > accomplish the following:
> >
> >   1) when booting with the "default" kernel, us the accellerated nvidia
> > driver
> >   2) when booting with the "xen" kernel, revert to the standard VESA/fb
> > driver
> >
> > Any thoughts on how I would go about this?
> >
> >
>
>
>
> --
> John Lowry
>
>


More information about the SGVLUG mailing list