NVIDIA on 11.2: kernel header file ‘/usr/src/linux/include/linux/version.h’ does not exist
October 18, 2009 around 6am (openSUSE)If you compiled the NVIDIA driver on openSUSE 11.2, while you had all the relevant packages installed (kernel-source, kernel-default-devel if you have the default kernel), you might have been getting this error:
The kernel header file '/usr/src/linux/include/linux/version.h' does not exist.
The most likely reason for this is that the kernel source files in '/usr/src/linux' have not been
configured.
Well, despite the NVIDIA run file looking for the files in the wrong place, you can fix this error by using the more recent NVIDIA driver: NVIDIA-Linux-x86-185.18.36-pkg1.run
Thanks to all those who helped in #suse
Powered by FireStats
cd /usr/src/linux
make cloneconfig
make prepare
http://en.opensuse.org/NVIDIA/The_hard_way#Prepare_kernel_source
Cheers
-J
That way isn’t recommended as no module should ever use /usr/src/linux It should always be using /lib/modules/….build/
Just install kernel-desktop-devel.
kernel-desktop-devel _was_ installed.
Great !!!, thanks a lot cyberorg, it works on my openSUSE 11.2 with custom kernel.
kernel-desktop-devel _was_ installed
ilan ver sahibinden good news
thank you sir
I tried this with 11.3 but it doesn’t work. Tried:
make oldconfig prepare;
#didn’t work
make clean; make cloneconfig prepare;
#didn’t work
I’ve run:
zypper in -t pattern devel_kernel;
I need to install a CUDA driver which I got from here:
http://www.nvidia.com/object/cuda_get_stage.html
“no module should ever use /usr/src/linux”, and yet NVidia does just this. NVidia engineers the devices and writes all the driver code. What do you do exactly? Please, don’t make stuff up just because it fits your existing perception of reality.
It’s stuff like this that drives people away from linux. This should not be a problem. Maybe a bug deep in the driver, yes, but stopped dead at installation? Why even write the driver if you are going to write a half ass installer that fails at the first sign of trouble?
Please ignore my comments above. The issue I’m seeing is “Unable to determine the version of the kernel sources located in” and the problem is unrelated.
I got the 2.0 beta driver from here:
http://www.nvidia.com/object/cuda_get_stage.html