NVIDIA drivers

Dwaine Garden DwaineGarden at rogers.com
Tue Sep 28 01:45:52 UTC 2004


Ricardo Veguilla wrote:

> Nave, Ronald wrote:
>
>> As of my latest update to FC3-test2 on 9/27/2004  , The
>> NVIDIA-Linux-1.0-6111 driver still does not load properly on start up.
>>
>>  
>>
> If I'm not mistaken, this problem (no driver autoloading)  is caused 
> by the changed to udev.  I think they are working on this.
>
>> It is necessary to run the installation program each time the machine is
>> booted.
>>  
>>
> You should only need to re-run the nvidia installer after a kernel 
> upgrade.
>
>> Also it is necessary to run   /sbin/modprobe  nvidia   manually to 
>> get the
>> nvidia module to load.
>>
>> Once the installation process is repeated  X windows will respond to
>> startx and the machine
>> runs OK.
>>  
>>
> Workaround #1
>
> modify /etc/rc.d/rc.sysinit where it says
> echo -n $"Initializing hardware... "
>
> ide=""
> scsi=""
> network=""
> audio=""
> other="nvidia" <--- add this
>
> Pros: rhgb will use the nvidia driver (which is kind of useless)
> Cons: upgrading the initscript package may remove the modification.
>
> Workaround #2
>
> now rhgb accepts an alternate configuration file (/etc/rhgb/xorg.conf )
> so now you can use the safe x.org driver "nv" for rhgb
>
> and call "modprobe nvidia" from /etc/rc.d/rc.local
>
> Pros: rhgb will use the x.org driver, upgrading initscript will not 
> change the setting
> Con: a little more work
>
>> It also appears that  up2date and yum are not working properly.
>>
>> Both yum and up2date will startup , but they are so slow that they are
>> nearly useless.
>>
>> For some reason I can use Mozilla and go directly to the yum mirror 
>> server
>> and then manually download the new RPM files using FTP and then install
>> using  rpm -Uvh   *.rpm.
>>
>> Perhaps there are so many people updating that yum and up2date can not
>> handle the load.
>>
> I also experienced some connection problem a few days ago while trying 
> to upgrade (rawhide) using both yum and up2date. I tried a few hours 
> later and it worked fine, so I ssume was a connection problem on my 
> side. But now that you mention it, I did download RPMS from the 
> repository using the browser....weird.
>
> -- 
> Ricardo Veguilla
>
>
You are right.  It's the udev stuff working.   I had to compile the 
binary drivers twice to get xorg to fire up.   It's seems that the 
nvidia0 and nvidiactl get removed from /dev when the computer is 
rebooted.   The only way is to load the driver manually or put 
references in rc.local and create them on every reboot.

Here is my copy rc.local.

#!/bin/sh
#
# This script will be executed *after* all the other init scripts.
# You can put your own initialization stuff in here if you don't
# want to do the full Sys V style init stuff.

touch /var/lock/subsys/local
mknod /dev/nvidia0 c 195 0
mknod /dev/nvidiactl c 195 255
chmod 666 /dev/nvidia0 /dev/nvidiactl




More information about the fedora-test-list mailing list