04 March 2013 Karim Elatov

I updated my machine to Fedora 18 and I had the Nvidia driver installed (check out this previous post for the Nvidia setup), but after I rebooted, X wouldn’t start. Checking out the /var/log/Xorg.0.log file, I saw the following:

[ 44.607] (EE) NVIDIA: Failed to load the NVIDIA kernel module. Please check your
[ 44.607] (EE) NVIDIA: system's kernel log for additional error messages.
[ 44.607] (II) UnloadModule: "nvidia"
[ 44.607] (II) Unloading nvidia
[ 44.607] (EE) Failed to load module "nvidia" (module-specific error, 0)
[ 44.607] (EE) No drivers available.

Next, checking out the Nvidia log file /var/log/nvidia-installer.log, I saw the following:

test -e include/generated/autoconf.h -a -e include/config/auto.conf || ( \
echo; \
echo " ERROR: Kernel configuration is invalid."; \
echo " include/generated/autoconf.h or include/config/auto.conf are
missing.";\

Checking out which package provides those files, I saw the following:

moxz:~> yum provides "*include/generated/autoconf.h"
Loaded plugins: remove-with-leaves
adobe-linux-i386/filelists | 140 kB 00:00:00
fedora/filelists_db | 22 MB 00:00:24
google-chrome/filelists | 1.1 kB 00:00:00
rpmfusion-free/filelists_db | 277 kB 00:00:00
rpmfusion-free-updates/filelists_db | 88 kB 00:00:00
rpmfusion-nonfree/filelists_db | 184 kB 00:00:00
rpmfusion-nonfree-updates/filelists_db | 30 kB 00:00:00
updates/filelists_db | 9.2 MB 00:00:08
kernel-PAE-devel-3.6.10-4.fc18.i686 : Development package for building kernel modules to match the PAE kernel
Repo : fedora
Matched from:
Filename : /usr/src/kernels/3.6.10-4.fc18.i686.PAE/include/generated/autoconf.h
kernel-PAEdebug-devel-3.6.10-4.fc18.i686 : Development package for building kernel modules to match the PAEdebug kernel
Repo : fedora
Matched from:
Filename : /usr/src/kernels/3.6.10-4.fc18.i686.PAEdebug/include/generated/autoconf.h

It looks like those files are provided by the kernel-devel package. Checking out my kernel packages:

moxz:~>rpm -qa | grep kernel
kernel-tools-3.6.10-4.fc18.i686
kernel-devel-3.3.4-5.fc17.i686
kernel-3.3.4-5.fc17.i686
kernel-headers-3.6.10-4.fc18.i686
kernel-devel-3.6.10-4.fc18.i686
kernel-tools-libs-3.6.10-4.fc18.i686
kernel-3.7.8-202.fc18.i686

I have the kernel-devel packages installed but not for my current kernel. My kernel was the following:

moxz:~>uname -r
3.7.8-202.fc18.i686

and here were the kernel-devel packages:

moxz:~>rpm -qa | grep kernel-devel
kernel-devel-3.3.4-5.fc17.i686
kernel-devel-3.6.10-4.fc18.i686

I tried a yum update:

moxz:~>sudo yum update kernel-devel
Loaded plugins: remove-with-leaves
Nothing to do

But it looks like it’s not in the YUM repositories. I found the package on http://rpm.pbone.net and downloaded it. I then installed it manually:

moxz:~>sudo rpm -ivh kernel-devel-3.7.8-202.fc18.i686.rpm
Preparing... ################################# [100%]
Updating / installing...
1:kernel-devel-3.7.8-202.fc18 ################################# [100%]

Now checking over my kernel-devel packages, I saw the following:

moxz:~>rpm -qa | grep kernel-devel
kernel-devel-3.3.4-5.fc17.i686
kernel-devel-3.7.8-202.fc18.i686
kernel-devel-3.6.10-4.fc18.i686

That looks better. Now we can restart the akmods service and that will compile the new kernel module:

moxz:~>sudo service akmods restart
Redirecting to /bin/systemctl restart akmods.service

No errors were shown. Now checking out the status:

moxz:~>sudo service akmods status
Redirecting to /bin/systemctl status akmods.service
akmods.service - Builds and install new kmods from akmod packages
     Loaded: loaded (/usr/lib/systemd/system/akmods.service; enabled)
     Active: active (exited) since Sun 2013-02-24 08:44:29 PST; 26s ago
     Process: 13889 ExecStart=/usr/sbin/akmods --from-init (code=exited, status=0/SUCCESS)
Feb 24 08:43:18 moxz systemd[1]: Starting Builds and install new kmods from akmod packages...
Feb 24 08:43:18 moxz akmods[13889]: Checking kmods exist for 3.7.8-202.fc18.i686[ OK ]
Feb 24 08:43:18 moxz runuser[13916]: pam_unix(runuser:session): session opened for user akmods by (uid=0)
Feb 24 08:44:15 moxz runuser[13916]: pam_unix(runuser:session): session closed for user akmods
Feb 24 08:44:29 moxz akmods[13889]: Building and installing nvidia-kmod[ OK ]
Feb 24 08:44:29 moxz systemd[1]: Started Builds and install new kmods from akmod packages.

We can see that the nvidia-kmod kernel module compiled fine. Now checking out for last installed RPM, I see this:

moxz:~>rpm -qa --last | head -1
kmod-nvidia-3.7.8-202.fc18.i686-304.64-6.fc18.i686 Mon 25 Feb 2013 09:16:38 AM PS

Then checking out that RPM package:

moxz:~>rpm -qi kmod-nvidia-3.7.8-202.fc18.i686-304.64-6.fc18.i686
Name : kmod-nvidia-3.7.8-202.fc18.i686
Epoch : 1
Version : 304.64
Release : 6.fc18
Architecture: i686
Install Date: Sun 24 Feb 2013 08:44:18 AM PST
Group : System Environment/Kernel
Size : 11593588
License : Redistributable, no modification permitted
Signature : (none)
Source RPM : nvidia-kmod-304.64-6.fc18.src.rpm
Build Date : Sun 24 Feb 2013 08:44:09 AM PST
Build Host : moxz
Relocations : (not relocatable)
URL : http://www.nvidia.com/
Summary : nvidia kernel module(s) for 3.7.8-202.fc18.i686
Description :
This package provides the nvidia kernel modules built for the Linux
kernel 3.7.8-202.fc18.i686 for the i686 family of processors.

The “Build Host” is the hostname of my machine :) Checking out what files the package installed, we see this:

moxz:~>rpm -ql kmod-nvidia-3.7.8-202.fc18.i686-304.64-6.fc18.i686
/usr/lib/modules/3.7.8-202.fc18.i686/extra/nvidia
/usr/lib/modules/3.7.8-202.fc18.i686/extra/nvidia/nvidia.ko

Lastly restarting X:

moxz:~>sudo init 3
moxz:~>sudo init 5

or with this command:

moxz:~>sudo service gdm restart

My X Server started up without any issues and we see the kernel module loaded as well:

moxz:~>lsmod | grep ^nvidia
nvidia              10257972  40

Update: 4/20/13 After some time I realized that my kernel was never getting updated. Looking into it, I discovered that I had excluded the kernel package from being installed via yum. I had the following option set:

moxz:~>grep exclude /etc/yum.repos.d/fedora-updates.repo
exclude=kernel*

After I removed that line, kernel updates started to work. That was the source of my problem all along.


blog comments powered by Disqus