[CentOS] Problem with nvidia-drv-x11 when upgrading to CentOS 5.2

Tue Jun 24 19:08:35 UTC 2008
Bernhard Gschaider <bgschaid_lists at ice-sf.at>

Hi!

I know that the nvidia-drv-x11-package is from RPMForge, but I thought
I might mention it here.

I have a machine with a NVidia-card and because of the ease of use I
use the rpmforge-package for that.

After upgrading the problem was that during rebooting as soon as the
machine hit the RHGB (which implicitly uses that driver) the machine
froze. I booted into the old kernel removed rhgb, rebooted. Now the
machine froze when starting X. Same procedure: Old Kernel. Remove
nvidia-drv-x11. Reboot into the new kernel and reinstall
nvidia-drv-x11. After a reboot everything works as it used to.

The cause of the problem could have been (sorry, I didn't do any
thorough forensic) that the nvidia.ko that was used in the new kernel
was a symbolic link from the "weak-updates"-directory into the old
kernel "extra"-directory. Is it possible that this only works inside
of kernels with the same x in CentOS 5.x ?

For the other machines I am upgrading I plan to

1. deinstall nvidia-x11-drv
2. do a regular "yum upgrade"
3. reinstall nvidia-x11-drv after reboot

Is there any better way to do it?

Bernhard

PS: the main purpose of this message is to alert others to this
problem, not to get an answer, although that would be nice