nVidia driver Fedora Core 4 test 2

[dual_boot]

I am having a hell of a time getting the nvidia driver (NVIDIA-Linux-x86-1.0-7174-pkg1.run) to work properly.

I am running FC4 test2 (that is the only version that will work with my SCSI adaptec adapter).

I have a Geforce 6800 GT PCI-X

The driver installs fine (have tried every documented process on the web) but when I go to start the X server, it flashes a few times and then nothing. I hit enter and view the log file and no errors. There are a few warnings about some resolutions, but no (EE).

I am a linux newbie and maybe I am missing something. Or maybe FC4 is too new. :confused:

I REALLY need a fix for this soon!

Thanks for the help/suggestions!!!!

[RIgnazio]

Interesting, I had a SATA hookup on a brand new motherboard and Fedora Core 3 found it no problem.

I also had an NVidia video card, and got it working without a single problem.

One problem may be the kernel included with Fedora Core 4 Test 2, and it may be too new. On FC3 I ran 2.6.10-1.770. In that case, it would have been:

yum install kernel-module-nvidia

or if it's picky and wants the whole thing:

yum install kernel-module-nvidia-2.6.10-1.770_FC3

I am not sure how to help you otherwise, sorry.

_____
Don't mess with success, stay stable!
Fedora Core 3

[dual_boot]

when I install the nVidia driver, it has to update the kernal...(and does successfully) is this the same thing?

when I run yum install kernel-module-nvidia I get an error that says I have no GPG public keys installed and that I need to download them.

[dual_boot]

hang on... i ran rpm --import /usr/share/doc/fedora-release-3*/RPM-GPG-KEY* (worked)

and then did a yum and got "cannot find valid baseurl for repo: development"

am I on the right track... or any track :)

[BrEakbEaTs]

Im running on the new drivers from Nvidia, and FC4 T2.

Install GCC and Dep's.
Install Kernel-devel
Turning off SELinux. (It best to do this anyway, as the SELinux Policy is still in it infancy on FC4, and will cause issue with most software.)
Then compiled the Nvidia drivers
Then modified the Xorg.conf file to use you new module.

Hope this helps..

[RUSTY]

try checkin http://www.nvnews.net/vbulletin/forumdisplay.php?s=&forumid=14 for more solutions :)

[/\/\r.Abdul]

Im running on the new drivers from Nvidia, and FC4 T2.

Install GCC and Dep's.
Install Kernel-devel
Turning off SELinux. (It best to do this anyway, as the SELinux Policy is still in it infancy on FC4, and will cause issue with most software.)
Then compiled the Nvidia drivers
Then modified the Xorg.conf file to use you new module.

Hope this helps..
hi im a linux noob i am having the saqme problem as thread start and when i did the yum thing first suggested i got this at the end: "Added 900 new packages, deleted 0 old in 3.61 seconds
Parsing package install arguments
No Match for argument: kernel-module-nvidia-2.6.10-1.770_FC3
Nothing to do"

can u plz explain ur post to me in noob talk?

EDIT: ok i typed in yum install GCC and its doing that now downloading all the stuff....do i do the same with "dep" and then "Kernel-devel" and how do i turn off selinux?

[BrEakbEaTs]

I honestly dont know were people got the pack "kernel-module-nvidia-blah blah what ever". I doesnt excist!
(Not that I know of in the normal set of yum repos, I could be wrong though, but doubt it)

Installing the nVidia Driver Howto (For FC4 Beta - By Breakbeats)
This is how I got my system to work. Note: All the follow is executed as Root.
Run "yum install gcc" it will install the GCC program that is needed to compile the driver. The dependencies are also needed, so when it prompts you with the list of packages to install, enter "Y" and then hit enter.

Now to compile the drivers you need kernel-devel because your compiling a Module for you video card. A module is like a little "loadable" part of the kernel. Do this by running "yum install kernel-devel". Hit Y again and enter.

Now lets turn off SELinux. Its a Beta, so we are so worried about security so much, as it should only be a test system. You could turn it on after this procedure, but I've the current policy even effected things like mount the swap file on boot.
I think its best turned off for now until the full version is out.
You can go into the "Security Level" GUI under X11 and click on the SELinux tab, then untick "Enabled".
OR
From the command line, edit the file /etc/selinux/config using a editor of choice, and change the entry "SELINUX=Enable" to "SELINUX=Disable", and then reboot.
This will stop the Security Policy effecting the module from compile correctly. This is only the case on FC4 Beta, because the policy that is active is not yet refind.

Lets compile the driver. Its best to compile the driver in RunLevel 2 or 3. I alway reboot here, when the Grub loader comes up, hit enter, then A, this will show you the command that will load the kernel. You should be like in a editing mode (Note: this only modifies the current load, its not a permanent change!! Once you reboot, it will boot as it did before). Hit the space bar, and enter "single" to the end. This will load your system into single user mode, or RunLevel 1. Not you will not be required to authenticate on this level, and Bash prompt will just come up when it boots.
Once your at this prompt, type "init 2" to now load you into the RunLevel 2. Now you will be prompted to login.
Login as Root and go to the directory where you have your driver downloaded. Now run "sh NVIDIA-Linux-x86-1.0-7174-pkg1.run" (This is for the latest driver at the time of this post, it may very on the driver version, as the name of the file will change). Now Keep hitting enter on the prompts, it should start running through some compiling. This will make the nvidia.o which is your new module. Linux comes with one called nv.o, and this is like the generic one.

We now want to modify the xorg.conf file to load our new module.
Edit the file /etc/X11/xorg.conf with you editor of choice, and locate the following area.

Section "Device"
Identifier "Videocard0"
Driver "nv"
VendorName "Videocard vendor"
BoardName "NVIDIA GeForce FX (generic)"
EndSection

Change this to something like this:

Section "Device"
Identifier "Videocard0"
Driver "nvidia"
VendorName "Videocard vendor"
BoardName "NVIDIA GeForce FX (generic)"
EndSection

You can see that we have change the "Driver" from nv to nvidia. Now your X11 should try and load the new module.


OpenGL Driver

Remove the follow entries from the top of the xorg.conf file if they are present.

Load "dri"
Load "GLcore"

And make sure that there is an entry

Load "glx"


Reboot or Restart you X11 session.

Good luck Compiling! :)

[BrEakbEaTs]

Man that was a big post.. Didnt realise!! hehe :)

[Thomas80]

Hello,


I followed BrEakbEaTs Howto for installing the Nvidia drivers.

But it doesn't work for me.
I got the nVIDIA GeForce Go 6600.

The installation of the driver seems to run fine. Anyway; there are no errors.

But if I modprobe nvidia, I get the following message :

FATAL: Module nvidia_1_0_7174 not found.


What do I do wrong ?

[SlowJet]

SELinux is now neck and neck with Sendmail as the pasty of puddin' heads. ;)

SJ

[BrEakbEaTs]

Hello,


I followed BrEakbEaTs Howto for installing the Nvidia drivers.

But it doesn't work for me.
I got the nVIDIA GeForce Go 6600.

The installation of the driver seems to run fine. Anyway; there are no errors.

But if I modprobe nvidia, I get the following message :

FATAL: Module nvidia_1_0_7174 not found.


What do I do wrong ?

Sounds like something is looking for the wrong file name.
I think this could be a problem in the xorg file, or your modprobe.conf file.

/etc/X11/xorg.conf, make sure you have set "Driver" to "nvidia"

and

/etc/modprobe.conf should have an entry "alias char-major-195* nvidia".
If the entry is "alias char-major-195* nvidia-1_0-7174", drop the -1_0-7174.
If there is a entry "alias nvidia nvidia-1_0-7174", remove it.

NOTE: Please back up all config files before modifying them!
Check those files and get back to me.

[BrEakbEaTs]

Also guys! Dont know if you have seen this but I will post in anyway. If you want some speed boosts from you GPU this is a little trick that helpes.

Open the /etc/modprobe.conf file, under the line "alias char-major-195*" add this line.

"options nvidia NVreg_EnableAGPSBA=1 NVreg_EnableAGPFW=1 NVreg_ReqAGPRate=8"

This turns on SideBand Addressing, FastWrites, and AGPRate x8.

[Finalzone]

There is a easy way to install Nvidia driver by using src.rpm from Livna

1. using yum, install fedora-rpmdevtools
2. use the command fedora-buildrpmtree to install a clone of /usr/src/redhat. The advantage for doing this is you can build package from your /home/user instead of using root
3. download Nvidia source rpm for i386 (http://rpm.livna.org/fedora/3/i386/SRPMS.stable/nvidia-glx-1.0.7174-0.lvn.1.3.src.rpm) or Nvidia source rpm for x86-64 (http://rpm.livna.org/fedora/3/x86_64/SRPMS.stable/nvidia-glx-1.0.7167-0.lvn.3.3.src.rpm) depending of your architecture. Move that package inside $HOME/rpmbuild/SRPMS/
4. Rebuild the package by doing "rpmbuild --rebuild $HOME/rpmbuild/SRPMS/sourcepackage"
5. Finally use yum localinstall $HOME/rpmbuild/RPMS/basearch/rebuiltpackage. You will need to install both kernel-module-nvidia and nvidia-glx at the same time. If yum does not install packages due to gpgcheck, disable it on /etc/yum.conf
6. Reboot your X.

Using that method will save you more tweakling as that package will automatically replace "nv" by "nvidia" inside /etc/X11/xorg.conf

[BrEakbEaTs]

These are FC3 packages (ie URL = http://rpm.livna.org/fedora/3/i386/SRPMS.stable) . And the SELinux may still pose a problem here.
Have you successfully got these RPMs to work on FC4 Test ?

[Finalzone]

They are actually src.rpm packages meant to be rebuilt on any i386 architecture.
So yes, I succesfully install these rebuilt packages and I have not encountered a problem with SELinux enabled.

[Thomas80]

Sounds like something is looking for the wrong file name.
I think this could be a problem in the xorg file, or your modprobe.conf file.

/etc/X11/xorg.conf, make sure you have set "Driver" to "nvidia"

and

/etc/modprobe.conf should have an entry "alias char-major-195* nvidia".
If the entry is "alias char-major-195* nvidia-1_0-7174", drop the -1_0-7174.
If there is a entry "alias nvidia nvidia-1_0-7174", remove it.

NOTE: Please back up all config files before modifying them!
Check those files and get back to me.

Thanks !!!
Nvidia driver works nicely !

Now I can choose the transparency feature in KDE without any speed loss :cool:

[BrEakbEaTs]

Glade I could help..

[/\/\r.Abdul]

thanks BrEakbEaTs after reading your post and taking random stabs with the "yum install" command i got all stuff i needed(the post b4 the really long one) then loaded up in telinit 2(and the CD command worked this time WOOT) i installed it all....except the network driverw oudlnt work but the standered are fine and work perfectly so meh.

thanks again

EDIT: and good to see Aussies in this didnt expect to see any myself(im from canberra)

[Nano-Tech]

There is a easy way to install Nvidia driver by using src.rpm from Livna

1. using yum, install fedora-rpmdevtools
2. use the command fedora-buildtree to install a clone of /usr/src/redhat. The advantage for doing this is you can build package from your /home/user instead of using root
3. download Nvidia source rpm for i386 (http://rpm.livna.org/fedora/3/i386/SRPMS.stable/nvidia-glx-1.0.7174-0.lvn.1.3.src.rpm) or Nvidia source rpm for x86-64 (http://rpm.livna.org/fedora/3/x86_64/SRPMS.stable/nvidia-glx-1.0.7167-0.lvn.3.3.src.rpm) depending of your architecture. Move that package inside $HOME/rpmbuild/SRPMS/
4. Rebuild the package by doing "rpmbuild --rebuild $HOME/rpmbuild/SRPMS/sourcepackage"
5. Finally use yum localinstall $HOME/rpmbuild/RPMS/basearch/rebuiltpackage. You will need to install both kernel-module-nvidia and nvidia-glx at the same time. If yum does not install packages due to gpgcheck, disable it on /etc/yum.conf
6. Reboot your X.

Using that method will save you more tweakling as that package will automatically replace "nv" by "nvidia" inside /etc/X11/xorg.conf

ok, i have yumed the fedora-rpmdevtools, but have not managed to get the next step, fedora-buildtree to work -> i get a command not found error
any help would be great

also, when attempting to insatll the drivers using hte .run file form NVidia, the file repedably failed to find the kernal source -> even though i know i yumed it twice

[Finalzone]

ok, i have yumed the fedora-rpmdevtools, but have not managed to get the next step, fedora-buildtree to work -> i get a command not found error
any help would be great
My mistake. the right command is fedora-buildrpmtree. The post is edited to reflect the correction.