Installing on a VM with OS RHEL 7.9 an upgrade from driver version 460 to 510. GPU passthrough is enabled in the VM. The machine is a server machine equipped with T4s.
Tried installing using .run files downloaded from the driver downloads, .rpms for RHEL specifically (nvidia-driver-local-repo-rhel7-510.73.08-1.0-1.x86.rpm), and both installs successfully but with errors in logging:
(dmesg)
nvidia: loading out-of-tree module taints kernel
nvidia-nvlink: Nvlink-core is being initialized, major device number 240
....
Initialized nvidia-drm 0.0.0 20160202 for 0000:00:03:0 on minor 1
..loads 3 more GPUs...
nvidia 0000:00:03:0: irq 33 for MSI/MSI-X
nvidia 0000:00:03:0: irq 34 for MSI/MSI-X
nvidia 0000:00:03:0: irq 35 for MSI/MSI-X
nvidia 0000:00:03:0: irq 36 for MSI/MSI-X
nvidia 0000:00:03:0: irq 37 for MSI/MSI-X
nvidia 0000:00:03:0: irq 38 for MSI/MSI-X
nvidia 0000:00:04:0: irq 33 for MSI/MSI-X
nvidia 0000:00:04:0: irq 34 for MSI/MSI-X
.....
[journalctl]
..intertwined with dmesg logs about MSI/MSI-X....
kernel: KVRM: GPU: 0000:00:05: rm_init_adapter failed. device minor number 2
kernel: KVRM: GPU 0000:00:05: RmInitAdapter failed! (0x63:0x55:2344)
..
kernel: KVRM: GPU: 0000:00:05: rm_init_adapter failed. device minor number 2
kernel: KVRM: GPU 0000:00:05: RmInitAdapter failed! (0x63:0x55:2344)
Nvidia GPUs seen from inside the VM:
lshw | grep nvidia
configuration: depth=32 driver=nvidia latency=0 mode=1024x768 visual=truecolor xres=1024 yres=768
configuration: driver=nvidia latency=0
...
nvidia-smi
No devices were found.