install nvidia driver success but execute `modprobe nvidia` get error

[soulseen]

I try to install the NVIDIA driver in VM with Debian9 distro, after installing successful, I get an error when executing command modprobe nvidia, The output:

modprobe: ERROR: ../libkmod/libkmod.c:514 lookup_builtin_file() could not open builtin file '/lib/modules/5.10.25-nvidia-gpu/modules.builtin.bin'

Environment
OS：debian9 VM
uname -r： 5.10.25-nvidia-gpu
Driver Version: 470.94

Further information
I checked the file /lib/modules/5.10.25-nvidia-gpu/modules.builtin.bin, its size is zero,

-rw-r--r-- 1 root root    0 Apr 26 08:41 modules.builtin.bin

[aaronp24]

I think that file is normally produced by depmod -a. Can you please try running that command as root and see if you get a non-empty file?

[soulseen]

hi @aaronp24, thanks for your reply. I executed depmod -a but get an warning as follow:

root@kata3-84bf96886c-qgnvj:/# depmod -a
depmod: WARNING: could not open modules.order at /lib/modules/5.10.25-nvidia-gpu: No such file or directory
depmod: WARNING: could not open modules.builtin at /lib/modules/5.10.25-nvidia-gpu: No such file or directory

here is which files in dir /lib/modules/5.10.25-nvidia-gpu , seems like don't have modules.order and modules.builtin

root@kata3-84bf96886c-qgnvj:/# ls -alh /lib/modules/5.10.25-nvidia-gpu
total 44K
drwxr-xr-x 3 root root 4.0K Apr 27 01:43 .
drwxr-xr-x 3 root root 4.0K Apr 26 07:35 ..
lrwxrwxrwx 1 root root   35 Jan 19 16:18 build -> /usr/src/kernels/5.10.25-nvidia-gpu
-rw-r--r-- 1 root root  271 Apr 27 01:43 modules.alias
-rw-r--r-- 1 root root  213 Apr 27 01:43 modules.alias.bin
-rw-r--r-- 1 root root    0 Apr 27 01:43 modules.builtin.bin
-rw-r--r-- 1 root root  141 Apr 27 01:43 modules.dep
-rw-r--r-- 1 root root  305 Apr 27 01:43 modules.dep.bin
-rw-r--r-- 1 root root    0 Apr 27 01:43 modules.devname
-rw-r--r-- 1 root root   55 Apr 27 01:43 modules.softdep
-rw-r--r-- 1 root root 3.7K Apr 27 01:43 modules.symbols
-rw-r--r-- 1 root root 4.3K Apr 27 01:43 modules.symbols.bin
lrwxrwxrwx 1 root root   35 Jan 19 16:18 source -> /usr/src/kernels/5.10.25-nvidia-gpu
drwxr-xr-x 2 root root 4.0K Apr 26 08:41 video

I don't know which difference between VM and Bare-Metal about the NVIDIA driver installer?

[aaronp24]

Where did that kernel come from? Does dpkg -S /lib/modules/5.10.25-nvidia-gpu/modules.builtin report a package that should have provided that file?

[soulseen]

I followed this doc https://github.com/kata-containers/kata-containers/blob/main/docs/use-cases/Nvidia-GPU-passthrough-and-Kata.md#install-nvidia-driver-in-kata-containers, and install the kernel by run rpm -ivh kernel-devel-5.10.25_nvidia_gpu-4.x86_64.rpm

[soulseen]

run dpkg -S /lib/modules/5.10.25-nvidia-gpu/modules.builtin, and get as follow:

root@kata3-84bf96886c-qgnvj:~/kata-containers/kata# dpkg -S /lib/modules/5.10.25-nvidia-gpu/modules.builtin
dpkg-query: no path found matching pattern /lib/modules/5.10.25-nvidia-gpu/modules.builtin