The nv driver and GeForce 7100 (Was: F10 on Homebuilt Box - Absolutely Brutal...!!!!)

Adam Jackson ajax at redhat.com
Wed Nov 19 16:18:06 UTC 2008


On Tue, 2008-11-18 at 16:10 -0500, Adam Jackson wrote:
> On Tue, 2008-11-18 at 10:43 -0700, Christopher A. Williams wrote:
> 
> > > (II) Primary Device is: PCI 00 at 00:10:0
> > > (WW) NV: Ignoring unsupported device 0x10de07e1 (GeForce 7100/nForce
> > > 630i) at 00 at 00:10:0
> > > (WW) Falling back to old probe method for fbdev
> > Unsupported??? (Big, Texas "Hunnnh...?")
> > 
> > After looking at this log in detail (and googling to see others with the
> > same card plus F9/10 and other distros having the identical problem), I
> > guess the real question to ask is why the nv driver seems to support
> > every nVidia card out there _except_ the GeForce 7100 series.
> > 
> > Oversight? If intentional, why, and what can be done to include it?
> 
> Not an oversight, utterly intentional.  You can thank nvidia for this
> one.  Apparently it's slightly weirder than your normal geforce 7, and
> they haven't written the code to support it in the nv driver yet (nor
> told us what needs doing).
> 
> We _ought_ to fall straight back to vesa if we can.  I've stared at this
> code a bit and not figured out why we're not falling back properly, but
> I guess I'll stare more.
> 
> If you're feeling adventurous you could hack out the bit of the nv
> driver that rejects that chip and then bang on things until it works,
> but it's not the sort of thing one can be hand-held through, you really
> will have to poke blindly until it works...

Alright, I've got a (bad) fix for this.  Not the final fix I want to
use, but I'd at least appreciate some testing from people with this kind
of nv driver failure.  It's actually an X server fix, and it can be
found here:

http://koji.fedoraproject.org/scratch/ajax/task_940170/

Run X, either with no config file or with nv selected as the driver in
the config, and then verify in the X log that we fell back to vesa.  I
don't have one of the affected nv chips so I had to synthesize the
failure by hacking the nv driver, but it should work.  Probably.

- ajax
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: This is a digitally signed message part
Url : http://lists.fedoraproject.org/pipermail/test/attachments/20081119/ae5808a4/attachment.bin 


More information about the test mailing list