On Tue, 2005-11-08 at 15:04 +0200, Gilboa Davara wrote: > Please do the following: > A. Download the drivers. > http://download.nvidia.com/XFree86/Linux-x86/1.0-7676/NVIDIA-Linux- > x86-1.0-7676-pkg1.run > B. init 3. > C. Execute the installation script. > D. Edit the /etc/X11/xorg.conf > Under section Device, Replace 'Driver "nv"' with 'Driver "nvidia"' > E. init 5. > > If this doesn't work, please post your: > A. /etc/X11/xorg.conf > B. /var/log/Xorg.0.log > C. dmesg. > > Cheers, > Gilboa > ...<sigh>... While this does work on one level, it's definitely NOT the best way if you're using Fedora. The chances of screwing up other video related things in your system are just about 100% even if this install process works. Fedora NVidia drivers work *much* better when you use the RPM files. The issue about upgrades is that, while Livna does provide the latest NVidia driver versions, they don't always keep up with the latest kernel versions, so you often have to rebuild the source RPM to get a kernel module RPM that works. But this is still much better than using the manual install from the script that NVidia provides because (among other things): * There's better fallback to a driver that actually works if a kernel doesn't have the appropriate module installed or available. * Automatic handling of the issues if you use a graphical boot. * Plays much nicer with Mesa and OpenGL stuff. NVidia's script breaks this. Ever wonder why some OpenGL things don't work correctly after using NVidia's install script? It's because they step all over existing Mesa stuff without regard for the integrity of the RPM database or the files that other things depend on. Here's how I keep NVidia drivers up to date from Livna and how you can too: 1) Install Fedora RPM Dev Tools (yum -y install fedora-rpmdevtools) 2) Download the latest source RPM from Livna. At present that is at: http://livna.cat.pdx.edu/fedora/4/i386/SRPMS.lvn/nvidia-glx-1.0.7676-0.lvn.2.4.src.rpm 3) While running the new kernel (assuming the source RPM you downloaded is in the current directory and you have current Pentium processor), run the command: rpmbuild --rebuild --target i686 nvidia-glx-1.0.7676-0.lvn.2.4.src.rpm If your processor isn't i686 class, use the appropriate flag for your system. As I have it listed, this command will rebuild new custom RPMs for your currently running kernel and deposit all of them in the directory /usr/src/redhat/RPMS/i686. You can them move these new RPMs to a directory of your choice for safe keeping. The files you will have will include something like the following list: * kernel-module-nvidia-2.6.13-1.1532_FC4-1.0.7676-0.lvn.2.4.i686.rpm * nvidia-glx-1.0.7676-0.lvn.2.4.i686.rpm * nvidia-glx-debuginfo-1.0.7676-0.lvn.2.4.i686.rpm * nvidia-glx-devel-1.0.7676-0.lvn.2.4.i686.rpm 3) You will only need to install the first two RPMs from the above list to get things working. Once you have installed these, you can either restart X or reboot and enjoy the new drivers. When you upgrade the kernel, follow the same rebuild process. You will then only need to install the updated kernel module RPM. Everything else will already be ready to go. I actually taught my mother to do this process in just a few minutes. She keeps her Livna drivers and kernel upgraded with no muss or fuss. She only calls me now to tell me she did it or if she has trouble because of things like the Livna main server going down. ...of course, if you have installed drivers from the NVidia site using the shell script, all bets are off. You will need to uninstall all of that first, and deal with the Mesa implications along the way, before trying this route. Good luck! Chris -- ====================== "Only two things are infinite, the universe and human stupidity, and I'm not sure about the former." -- Albert Einstein
