<html><body><div style="font-family: arial, helvetica, sans-serif; font-size: 12pt; color: #000000"><div>Hello list,</div><div><br data-mce-bogus="1"></div><div>I'm struggling deploying Rocky with vGPU using nvidia drivers.</div><div>Has anyone experienced the issues loading nvidia modules?</div><div><br data-mce-bogus="1"></div><div>I'm talking about hypervisor part of the setup. There are two modules provided by nvidia. One loads correctly it's the nvidia.ko one.</div><div>The other however does not. The module is called nvidia-vgpu-vfio.ko</div><div><br data-mce-bogus="1"></div><div>I'm trying to load it and it seems that 7.6 kernel is no longer compatible with it</div><div><br></div><div><div>modprobe nvidia-vgpu-vfio</div><div>modprobe: ERROR: could not insert 'nvidia_vgpu_vfio': <span style="color: rgb(255, 0, 0);" data-mce-style="color: #ff0000;">Invalid argument</span></div></div><div><br></div><div>dmesg shows this:</div><div><div><span style="color: rgb(255, 0, 0);" data-mce-style="color: #ff0000;"><strong>nvidia_vgpu_vfio: disagrees about version of symbol vfio_pin_pages</strong></span></div><div>nvidia_vgpu_vfio: Unknown symbol vfio_pin_pages (err -22)</div><div>nvidia_vgpu_vfio: disagrees about version of symbol vfio_unpin_pages</div><div>nvidia_vgpu_vfio: Unknown symbol vfio_unpin_pages (err -22)</div><div>nvidia_vgpu_vfio: disagrees about version of symbol vfio_register_notifier</div><div>nvidia_vgpu_vfio: Unknown symbol vfio_register_notifier (err -22)</div><div>nvidia_vgpu_vfio: disagrees about version of symbol vfio_unregister_notifier</div><div>nvidia_vgpu_vfio: Unknown symbol vfio_unregister_notifier (err -22)</div></div><div><br></div><div><div>modinfo nvidia-vgpu-vfio</div><div>filename: /lib/modules/3.10.0-957.27.2.el7.x86_64/weak-updates/nvidia-vgpu-vfio.ko</div><div>version: 430.27</div><div>supported: external</div><div>license: MIT</div><div>rhelversion: 7.6</div><div>srcversion: 0A179A61A02AD500D05FB1A</div><div>alias: pci:v000010DEd00000E00sv*sd*bc04sc80i00*</div><div>alias: pci:v000010DEd*sv*sd*bc03sc02i00*</div><div>alias: pci:v000010DEd*sv*sd*bc03sc00i00*</div><div>depends: nvidia,mdev,vfio</div><div>vermagic: 3.10.0-940.el7.x86_64 SMP mod_unload modversions </div></div><div><br></div><div><br data-mce-bogus="1"></div><div>My guess is that somewhere along the rhel/centos 7.6 lifecycle vfio module changed the vfio module and broke the compatibility.</div><div><br data-mce-bogus="1"></div><div>Nvidia provides those modules built against the BETA 7.6 release and assume weak-modules to make it work.</div><div>Somehow it does not.</div><div><br data-mce-bogus="1"></div><div>Anybody got any suggestions how to handle this? I'm working on it with nvidia enterprise support but maybe one of you got there first?</div><div><br data-mce-bogus="1"></div><div>best regards</div><div><br data-mce-bogus="1"></div><div data-marker="__SIG_POST__">-- <br></div><div>Piotr Baranowski</div></div></body></html>