Twinview problem using fedora 8

[mooninite]

Originally Posted by mrstoney

I've attached the nvidia-bug-report created after I attempted to boot with the the most basic xorg.conf set to "nvidia."

It looks like your install is pretty messed up. It's still linking to your 100.14.19 binaries when the RPM is the 100.14.23 binaries. So that's one problem.

The next problem is that it looks like you don't even have the nvidia kernel module built. Did you reboot before running this bug report? Did you install DKMS?

[mooninite]

Originally Posted by mrstoney

I also tried using the sh script to install again (still would not install w/o calling out the kernel I'm running). Attached are the resulting installer log and bug report.

This still shows that the script didn't build the correct kernel module.

nvidia: version magic '2.6.17-prep SMP mod_unload PENTIUM4 REGPARM gcc-4.1' should be '2.6.23.1-42.fc8 SMP mod_unload 686 4KSTACKS '

[kwizart]

Originally Posted by mooninite

I mentioned freshrpms over livna because DKMS is a superior method of maintaining binary drivers. DKMS does it all automatically for you. You never have to wait for livna to mosey around and build a new package every kernel update and if you install kernels from elsewhere (say kernel.org) you can use the nvidia binary driver just the same, where as you would be stuck without it on the livna package.

Well that's low level troll...

Dkms/kmod are differents scheme indeed. they have weakness and strong...

kmod do not requires gcc compiler to be present... not even kernel-devel.. That makes it really usefull for "standard" use with official Fedora kernels. Pre-build package are targetted for pre-build distribution (Fedora is versus Gentoo or ArchLinux). Packages are usually tested by the maintainer (actually they are automatically built nowadays even for rawhide kernels).
The livna team (i'm part of), usually set the useful general options so you just need to reboot on the newer kernel without tweaking anything...

dkms can rebuild on-fly kernel module but requires gcc compiler and it can fails... (not such often with nVidia's). In the past day, that was usefull for special kernels such as rawhiders was using..

For my personnal use, kmod is better because i can rebuild kernel (rpm packaged of course since i've learnt how to do this - easier than building without rpm), and then rebuild the appropriate kmod:
This is property documented here:
http://rpmfusion.org/Packaging/KernelModules/Kmods1
http://rpmfusion.org/Packaging/KernelModules/Kmods2
Now i can even redistribute the kmod i've made for my special kernel from a yum reposository if I want...

So in my view dkms is a superior method to solve nothing.
But the main problem here, remains to have a kernel module integrated within the kernel...

[mrstoney]

Originally Posted by mooninite

It looks like your install is pretty messed up. It's still linking to your 100.14.19 binaries when the RPM is the 100.14.23 binaries. So that's one problem.


The next problem is that it looks like you don't even have the nvidia kernel module built. Did you reboot before running this bug report? Did you install DKMS?

Yes, I did reboot and yes, I did install DKMS (it actually will not install w/o doing so).

I'm at a loss at how to proceed.