nvidia.ko for kernel 2.6.25.10-86.fc9.i686 is not found

[gmangual]

Hi, everything was working find last night. Today I just turned on the computer and saw a message saying that nvidia.ko for the kernel 2.6.25.10-86.fc9.i686 is not found. It seems that a kernel update damage the working driver. I tried to updated the driver using yum update kmod-nvidia but nothing was found and I have the livna repo. Now GL Desktop (Gnome Compiz) can't be enabled because I get a blank screen :(

I have a Nvidia Fx5200 AGP8x video card.This is what I get when running rpm -qa |grep nvidia

xorg-x11-drv-nvidia-173.14.09-1.lvn9.i386
kmod-nvidia-2.6.25.9-76.fc9.i686-173.14.09-2.lvn9.i686
xorg-x11-drv-nvidia-libs-173.14.09-1.lvn9.i386
kmod-nvidia-173.14.09-2.lvn9.i686

Any idea of what can i do to solve this? or I just need to keep waiting for an nvidia driver update for the kernel 2.6.25.10-86.fc9.i686 to appear someday using yum?

THANKS!

[MheAd]

It happens sometimes.
It will show up on the repository soon.

[gmangual]

Thank you for your comment. That's what I thought. :(

[Rbb_e]

I'm experiencing the exact same problem.
So I take it anyone else effected by this will just have to wait?
If so when it is fixed will I have to do anything like "yum update kmod-nvidia" also how long does this generally take to be fixed?

Also sorry if this is seen as thread-hijacking or anything I just thought it better to ask here instead of creating another topic for the same problem.

[Finalzone]

yum install akmod-nvidiaUse that command, reboot the system and voilà.

[Rbb_e]

I try doing that and I get errors about it being unable to connect.

[anonymouse]

It's now working for me. Actually I had the opposite problem - the new version of kmod-nvidia was found but the latest kernel was not!

Anyway if you try updating again now it should work.

[Rbb_e]

I tried it and I got "could not retireve mirrorlist http://mirror.fedoraprojectblahblah error was
[error 4] IOerror: <urlopen error(-3,'temporary failure blah blah

I had to write it down by hand and that got tired pretty quickly. Any idea on how I can fix this? It's a completely different message then I was getting before when I tried and I got a screen full of errors now i'm only getting the one.

[Dan]

Sounds like a moirror is still updating. Give it about 30 minutes and try again.

[Rbb_e]

Thanks for the help.
I'll try again in a little while and hopefully it'll be fixed =] If not i'll be back tomorrow haha

[brebs]

An alternative to akmod is to run nvidia's own installer automatically after booting into a new kernel - add this to /etc/rc.local
# Setup, for nvidia module checking
ARCH=$(uname -i)
LIBDIR=lib64
# pkg2 includes 32-bit compatibility libs - /usr/lib/libGL.so
PKG=pkg2
if [ "${ARCH}" != "x86_64" ] ; then
# Is 32-bit
ARCH=x86
LIBDIR=lib
PKG=pkg1
fi

INSTALL_NVIDIA=
/sbin/modprobe nvidia
if [ $? != 0 ] || ! [ -L "/usr/${LIBDIR}/xorg/modules/extensions/libglx.so" ] ; then
INSTALL_NVIDIA=y
else
# Check opengl headers, because nvidia's headers break the
# compilation of xorg-x11-server.
# mesa-libGL-devel provides /usr/include/GL/gl.h
# http://koji.fedoraproject.org/koji/packageinfo?packageID=184
grep GL_DEPTH_STENCIL_MESA /usr/include/GL/gl.h > /dev/null
[ $? != 0 ] && INSTALL_NVIDIA=y
fi

if [ -n "${INSTALL_NVIDIA}" ] ; then
# Releases: http://www.nvnews.net/vbulletin/showthread.php?t=122606
VER=180.22
# $ rpm -qf /usr/lib64/xorg/modules/extensions/libglx.so
# xorg-x11-server-Xorg-1.4.99.905-2.20080702.fc9.x86_64
# Posted at http://www.fedoraforum.org/forum/showthread.php?p=1047697#post1047697
# find /lib/modules/ -name nvidia.ko -delete
# yum install kernel-devel kernel-headers glibc-headers
# Info: sh NVIDIA-Linux-x86_64-173.08-pkg1.run -A
# wget ftp://download.nvidia.com/XFree86/Linux-x86_64/173.14.05/NVIDIA-Linux-x86_64-173.14.05-pkg2.run
# Populates /usr/lib64/tls/
# Seems to delete /usr/lib64/xorg/modules/extensions/libGLcore.so
# Also creates /usr/lib64/tls/libnvidia-tls.so.1
mkdir -p /root/nvidia
cd /root/nvidia
[ -e NVIDIA-Linux-${ARCH}-${VER}-${PKG}.run ] || wget ftp://download.nvidia.com/XFree86/Linux-${ARCH}/${VER}/NVIDIA-Linux-${ARCH}-${VER}-${PKG}.run || wget http://us.download.nvidia.com/XFree86/Linux-${ARCH}/${VER}/NVIDIA-Linux-${ARCH}-${VER}-${PKG}.run
# Run nvidia's installer, without overwriting e.g. /usr/include/GL/gl.h
sh /root/nvidia/NVIDIA-Linux-${ARCH}-${VER}-${PKG}.run -a -q -s -N --no-opengl-headers
/sbin/modprobe nvidia
fi

Also put in /etc/modprobe.conf for /dev/nvidia* permissions (http://forums.fedoraforum.org/showthread.php?t=201392)
# 100 is users group in /etc/group
options nvidia NVreg_DeviceFileMode=0660 NVreg_DeviceFileUID=0 NVreg_DeviceFileGID=100 NVreg_ModifyDeviceFiles=1

For further info on Nvidia's installer, see FUD (http://www.fedoraforum.org/forum/showthread.php?t=200654) and bug (http://www.nvnews.net/vbulletin/showthread.php?t=120377).

[gmangual]

yum install akmod-nvidiaUse that command, reboot the system and voilà.
Hi. I tried yum install akmod-nvidia and some drivers where installed. After a reboot the system found the nvidia.ko driver but never started X. I had to log in as root in console mode and then delete the file from (/lib/modules/2.6.25.10-86.fc9.i686/extra) :( After another reboot I could log in with the graphical interface, but still have the problem.

[Finalzone]

What does
cat /var/log/Xorg.0.log say? It will be a good idea to attach that file and ask Firewing1 about the issue.

[leigh123linux]

An alternative to akmod is to run nvidia's own installer automatically after booting into a new kernel, e.g. on x86_64 put this in /etc/rc.local
/sbin/modprobe nvidia
if [ $? != 0 ] || ! [ -L /usr/lib64/xorg/modules/extensions/libglx.so ] ; then
# Posted at http://www.fedoraforum.org/forum/showthread.php?p=1047697#post1047697
VER=173.14.12
# find /lib/modules/ -name nvidia.ko -delete
# yum install kernel-devel kernel-headers glibc-headers
# Info: sh NVIDIA-Linux-x86_64-173.08-pkg2.run -A
# wget ftp://download.nvidia.com/XFree86/Linux-x86_64/173.14.05/NVIDIA-Linux-x86_64-173.14.05-pkg2.run
# Populates /usr/lib64/tls/
# Seems to delete /usr/lib64/xorg/modules/extensions/libGLcore.so
mkdir -p /root/nvidia
cd /root/nvidia
[ -e NVIDIA-Linux-x86_64-${VER}-pkg2.run ] || wget http://us.download.nvidia.com/XFree86/Linux-x86_64/${VER}/NVIDIA-Linux-x86_64-${VER}-pkg2.run
sh NVIDIA-Linux-x86_64-${VER}-pkg2.run -a -q -s -N
/sbin/modprobe nvidia
fi


With this method X would break every time the xorg-x11-server-Xorg package was updated !

i.e it would overwrite some of the nvidia files ( the nvidia run file overwrites some of the fedora xorg files )

[brebs]

With this method X would break
Big deal. Just rerun the nvidia installer. Such an obvious, intuitive and easy fix ;)

Anyway, you will notice my checking & automatically fixing line, since you chose to quote ALL of my code, causing me to wear out the scroll button on my mouse:
[ -L /usr/lib64/xorg/modules/extensions/libglx.so ]

There's a ton of legacy FUD about the nvidia installer. It's actually not the child of Satan that it's made out to be. In fact, it actually just works. Without having to go through the pain barrier of figuring out what on Earth kmod/akmod packages do.

[Lazy-buntu]

I'm having the exact same problem as the first post.

I tried:
yum -y install kmod-nvidia

Output:
Loaded plugins: refresh-packagekit
Setting up Install Process
Parsing package install arguments
Package kmod-nvidia-173.14.12-3.lvn9.i686 already installed and latest version
Nothing to do


Do we wait for an update later today or what?

[Lazy-buntu]

bump for nvidia.ko for kernel 2.6.25.10-86.fc9.i686 is not found

still not solved

[Lazy-buntu]

Wait for an update, someone have a work around, or can someone tell us how to revert to the previous kernel (if that's the problem)?

[Hlingler]

Wait for an update, someone have a work around, or can someone tell us how to revert to the previous kernel (if that's the problem)?Please search the forums, since this question has been asked and answered many times recently, for example: http://forums.fedoraforum.org/forum/showthread.php?t=196750

V

[Lazy-buntu]

This thread was started before that thread, so they should have searched harder :p

Anyway, I reinstalled akmod nvidia and restarted my computer 2x and it fixed the problem.

[grantschoep]

Basically, The Nvidia driver needs to go hand in hand with the kernel driver. I'm pointing to the livna repositorys as well. Understand how it happens.

However, Is there any way to tell yum or something to not update the kernel if it doesn't find a matching nvidia.ko driver?

I suppose I could just do a "yum update" and not select yes to install if I didn't see it.

Hmm, I could probably script that. Any one use a method so they don't bump their kernel over the latest nvidia.ko?

[wvn]

yum -y install akmod-nvidia

[fcjunkie]

Please search the forums, since this question has been asked and answered many times recently, for example: http://forums.fedoraforum.org/forum/showthread.php?t=196750

V

Really? I've searched the forums high and low and can't find which method to use to install the rpms for my Quadro NV 135M. It's a laptop GPU currently still sold in the Dell D630 with Linux supported by NVIDIA, but the how-to here for the fusion rpms is incomplete and doesn't correlate properly to the supported GPU list it refers to. That said, it looks like a good guide and once I get a couple of questions answered, I trust it will work just fine, but my point is that the answer isn't there today, and I understand the frustration of others on this topic. I wish once in a while that someone would actually make sure the answer is available before emitting the cloddish "search, grasshopper" patterned response.

NVIDIA and ATI support in Fedora is getting worse (and is now basically non-existent) with every release. Those are the two most popular graphics engines and Fedora makes adding support for them anything but intuitive. I think these types of issues should be addressed in F11.