Bug#812396: nvidia-kernel-dkms: fail to load nvidia-uvm.

Andreas Beckmann anbe at debian.org
Sat Jan 23 15:36:17 UTC 2016

On 2016-01-23 10:24, Pascal Obry wrote:
> $ sudo modprobe nvidia-uvm
> modprobe: ERROR: could not insert 'nvidia_current_uvm': Invalid argument
> modprobe: ERROR: ../libkmod/libkmod-module.c:977 command_do() Error running install command for nvidia_uvm
> modprobe: ERROR: could not insert 'nvidia_uvm': Operation not permitted
> This issue was first seen in nvidia 346.96 and now in 349.16.

These were the missing bits:

> [  103.476413] nvidia_uvm: no symbol version for nvUvmInterfaceChannelDestroy
> [  103.476417] nvidia_uvm: Unknown symbol nvUvmInterfaceChannelDestroy (err -22)
> [  103.476424] nvidia_uvm: no symbol version for nvUvmInterfaceQueryCaps

I had fixed this for 346.96, but apparently the fix only worked for the
nvidia-kernel-source package (which is much better suited for automatic
testing than dkms ...).


More information about the pkg-nvidia-devel mailing list