Markus Koßmann wrote:
Am Samstag, 24. Februar 2018, 18:53:38 schrieb Peter Suetterlin:
Any ideas ?
Not sure what to make from the above - looks as if it had been compiled against the wrong kernel? Well before I noticed that problem I booted into runlvel 3 and forced a reinstallation of the nvidia rpms with zypper in --force. So I'am quite sure that is was build against the kernel, where it failed to load.
If you look into /lib/modules for nvidia.ko you will find: /lib/modules # find . -name nvidia.ko | xargs ls -l -rw-r--r-- 1 root root 29440208 Feb 24 18:01 ./4.15.2-1- default/updates/nvidia.ko lrwxrwxrwx 1 root root 47 Feb 19 18:22 ./4.15.3-1-default/weak- updates/updates/nvidia.ko -> /lib/modules/4.15.2-1-default/updates/nvidia.ko lrwxrwxrwx 1 root root 47 Feb 21 20:10 ./4.15.4-1-default/weak- updates/updates/nvidia.ko -> /lib/modules/4.15.2-1-default/updates/nvidia.ko
So it seems to be compiled into the 4.15.2 module tree and then linked into the other module trees
Not sure if you solved the issue or are following the bugreport. The problem is/was a kmp compile script that is triggered when a new kernel-default-devel gets installed. That script (you can check it with rpm -q --triggers nvidia-gfxG04-kmp-default) does compile the nvidia modules against the new kernel, but places them in the old location, which is hardcoded in the script... I attach a script that I used to recompile my modules for the kernels I have. ./compile_nvidia_kmp.sh # this will compile against the latest installed # kernel, but place the modules properly ./compile_nvidia_kmp.sh 4.15.2-1-default # modules for older kernel HTH