[elrepo] Failure to install kmod-nvidia on CentOS 7.4 (kernel symbol(?) requirements wrong)

[Frank Thommen]

Dear all,

I'm currently trying to install kmod-nvidia from ELRepo on a CentOS 7.4 
system (kernel 3.10.0-514.el7.x86_64).  The installation fails with lots 
of required kernel symbols or symbol addresses(?) not being correct.  E.g.:

Error: Package: kmod-nvidia-390.87-1.el7_5.elrepo.x86_64 (elrepo)
             Requires: kernel(drm_atomic_helper_connector_destroy_state) 
= 0x2085f9c7
             Installed: kernel-3.10.0-514.el7.x86_64 (@anaconda)
                 kernel(drm_atomic_helper_connector_destroy_state) = 
0xd2e7890b
             Installed: kernel-3.10.0-693.21.1.el7.x86_64 (@updates-dkfz)
                 kernel(drm_atomic_helper_connector_destroy_state) = 
0xca876bf4
             [...]


The full listing can be found on https://pastebin.com/QGqaaQq9.  The 
repository names are those of our local repo mirrors.  The same errors 
appear when trying to install kmod-nvidia-390.77 and kmod-nvidia-390.67.

Could it be a conflict with other installed packages from nVidia?  CUDA 
10.x is installed and the system already contains the following nvidia 
packages:

$ yum list installed  *nvidia*
Loaded plugins: fastestmirror
Loading mirror speeds from cached hostfile
   * elrepo: dfw.mirror.rackspace.com
Installed Packages
dkms-nvidia.x86_64                                  3:410.48-1.el7 
               @cuda
nvidia-detect.x86_64                                390.87-1.el7.elrepo 
                    @elrepo
nvidia-driver.x86_64                                3:410.48-1.el7 
               @cuda
nvidia-driver-NVML.x86_64                           3:410.48-1.el7 
               @cuda
nvidia-driver-NvFBCOpenGL.x86_64                    3:410.48-1.el7 
               @cuda
nvidia-driver-cuda.x86_64                           3:410.48-1.el7 
               @cuda
nvidia-driver-cuda-libs.x86_64                      3:410.48-1.el7 
               @cuda
nvidia-driver-devel.x86_64                          3:410.48-1.el7 
               @cuda
nvidia-driver-libs.x86_64                           3:410.48-1.el7 
               @cuda
nvidia-libXNVCtrl.x86_64                            3:410.48-1.el7 
               @cuda
nvidia-libXNVCtrl-devel.x86_64                      3:410.48-1.el7 
               @cuda
nvidia-modprobe.x86_64                              3:410.48-1.el7 
               @cuda
nvidia-persistenced.x86_64                          3:410.48-1.el7 
               @cuda
nvidia-settings.x86_64                              3:410.48-1.el7 
               @cuda
nvidia-xconfig.x86_64                               3:410.48-1.el7 
               @cuda
$


I haven't had such an error in all the years that I use ELRepo.  Can 
someone enlighten me what is going wrong and how it can be fixed?

Thank you very much in advance
frank