Ok so...

There is some sort of bug with my nVidia 7800GT that causes the graphic installer to load up artifacted, frozen, and jumbled. So instead, I did a text-based install. After rebooting, I went straight to a command line, and I found out that I had to modify /etc/inittab so that the default runlevel is 5. I did that, rebooted, and the Fedora load screen came up artifacted and frozen as with Anaconda.

I then started all over, but this time, after changing the default runlevel, I edited /etc/X11/xorg.conf and replaced the video device driver field from nv to vesa. I rebooted and could achieve the log in screen, but I had expected the initial setup program in which I create an account to run, and it didn't. I assume this is because I had already booted once to change xorg.conf and inittab. I can only log in as root, and none of the options that the first time boot program configures were set. Does anyone know how to manually start this utility?

In the mean time, I am trying to install nVidia's drivers so that I don't waste my 7800GT with vesa drivers. I downloaded the latest linux drivers from the nVidia website, copied them to my HD, and switched back to runlevel 3 because the driver installation requires X to be disabled. I entered sh <whateverdrivernamewas>.rpm, but I recieved an error saying that the installer needed to compile an interface with the kernel of some sort. I then downloaded the correct kernel source RPM from the Fedora FTP and rebooted. I entered rpm -Uvh <kernelname>.rpm, and consequently recieved an error that looked something like this:

Failed dependencies: /usr/sbin/hardlink is needed by kernel-devel-blahblahblah.rpm

Note: I also recieved a warning concerning a signature of some sort, but I have since learned that the --nosignature switch will override this.

I am at my nerves' end right now. Am I using the wrong switches when attempting to install the kernel sources? I hope someone knows what is going on here and can provide me with a solution. My linux knowledge is limited... any help would be greatly appreciated. Next up: the nightmare of getting my Linksys WMP54GX wireless PCI adapter to function on FC5. Sigh.

Thanks in Advance,
-JS

To rerun the Firstboot program, as root edit '/etc/sysconfig/firstboot', and change:

RUN_FIRSTBOOT=NO

to

RUN_FIRSTBOOT=YES

First boot will run on your next boot.

The package you need is NOT the kernel source, it's the kernel devel package:

yum -y install kernel-devel

Be aware that some folks are having problems with the current (8xxx series) nvidia driver. Check the nvidia Linux forum for the lastest issues and workarounds.

Thanks.

Also, I had devel and source mixed up... but I did install devel after I found the hardlink rpm.

New problem though:

I'll try to post the exact error, but the nvidia script claimed it could not successfully build nvidia.ko for a number of possible reasons. I'm going to boot back up into linux and write down that message... can't find my USB pendrive, d'oh.

Ok resolved that error message with someone else. He says I need to find a precompiled interface for my kernel, which is kernel-smp-2.6.15-1.2054_FC5.i686

I have no idea where to look on nvidia's ftp site for this (ftp://download.nvidia.com/). By the way, the reason I don't auto fetch the interface is because I have not yet managed to set up wireless. (see: http://www.linuxquestions.org/questi...d.php?t=462262)

Anyways, anyone know where I can find that interface?

The file you posted is not an "interface", it is an older kernel for machines with multiple processors (the "smp" part).

I think he is getting the error that says that the kernel was compiled with a different version of gcc then then one he compiled nvidia.ko with. The only fixes I know of would be:

1.) to use the packages from the livna repo.
2.) to compile a custom kernel and then compile the nvidia driver.

No fix is necessary. The FC5 kernel is compiled with the distributed gcc, which is what he should be using to install the nvidia driver. If he is getting a mismatch, then the system is in a weird, partially updated state, or using files from conflicting repositories. In either case, the system is in an unstable state.

I've only seen this when folks perform a base install (from CD/DVD), then add and enable a bunch of repositories, then perform an update.

Argh.

I followed the directions here:

http://www.mjmwired.net/resources/mjm-fedora-fc5.html

note: the non-yum directions

I ran rpm -ivh kmod-nvidia.rpm \ xorgdriver.rpm (not the actual RPM names) but recieved an error something like:

Failed Dependencies:
selinux-policy conflicts with xorgdriver

SELinux is completely disabled though. Any way around this? Thanks.

Having SELinux is sort of the point of using Fedora. If you are going to disable the security software, you might as well use Ubuntu (which is very similar to FC5, but does use SELinux).

You can't easily remove SELinux, so that third party nvidia package has not been properly created for FC5 installation.

My recommendation:

1- Clean install
2- yum -y update
3- Download and run the nvidia installer from the nvidia website

Stay away from third party repositories until you understand the operation of the system a little better.

Mace, my issue is that I have yet to configure my wireless adapter, and therefore cannot use any network services. I am relying on booting up windows, burning resources to a CDRW, and booting back into Fedora.

I only disabled SELinux after first recieving that dependency error. I have now reenabled it, and used these commands from stanton-finley.net's FC5 installation notes:

Even after using those commands, which I have seen numerous times after googling the dependency error, I still recieve said error.

If you believe the error is SELinux related, you can:

1. setenforce 0
2. run the nvidia installation
3. setenforce 1

The setenforce command temporarily tells SELinux not to enforce violations (it still logs them). Disabling SELinux can cause other problems - files are no longer tagged properly, so when SELinux is re-enabled, access violations can occur. Using setenforce does not cause that problem.

Wait... you haven't applied maintenance before trying to install the nvidia driver? That won't work - the distributed FC5 was broken for loadable modules.

I suggest you download and burn the FC5 re-spin, which has maintenance applied.

What a pain in the arse. Try installing the livna drivers using:

rpm -Uvh --nodeps --force rpmname.rpm

That may work.

That did it. Now on to the next project... http://www.linuxquestions.org/questi...d.php?t=462262