GTX 1060 nvidia-current 375.20 fails to load on debian 4.8.0-2-amd64 kernel but succeeds under exact...

i’ve got the source code for the nvidia 375.20 driver installed via the process of “apt-get install nvidia-kernel-dkms” which (as you probably are aware of) takes the standard nvidia source download and packages it up for auto-building from source, on debian.

i carried out some tests on an Aorus X3plusV6 which, when running with the standard debian 3.16 kernel that comes with debian 8.6.0 showed that there’s nothing wrong with 3.16: optirun glxinfo loaded things up, i got a log file /var/log/Xorg.8.log as you would expect… all good.

however, due to the machine being insanely recent, the 3.16 i915 intel kernel driver doesn’t recognise the intel skylake v6 hardware, so i was forced to upgrade to debian 4.8.0-2-amd64.

here’s where the i915 driver worked perfectly… but ironically even just doing “modprobe nvidia-current” came up with the following error which DID NOT OCCUR ON THE 3.16 KERNEL WITH ABSOLUTELY ZERO OTHER SOFTWARE CHANGES OTHER THAN WHAT IS NEEDED TO BUILD NVIDIA 375.20 FOR 4.8.0:

Dec 10 18:59:02 localhost kernel: [ 52.261529] nvidia: module license ‘NVIDIA’ taints kernel.
Dec 10 18:59:02 localhost kernel: [ 52.261532] Disabling lock debugging due to kernel taint
Dec 10 18:59:02 localhost kernel: [ 52.279069] nvidia 0000:01:00.0: enabling device (0000 → 0003)
Dec 10 18:59:02 localhost kernel: [ 52.279263] vgaarb: device changed decodes: PCI:0000:01:00.0,olddecodes=io+mem,decodes=none:owns=none
Dec 10 18:59:02 localhost kernel: [ 52.279301] NVRM: The NVIDIA GPU 0000:01:00.0 (PCI ID: 10de:1c20)
Dec 10 18:59:02 localhost kernel: [ 52.279301] NVRM: installed in this system is not supported by the 375.20
Dec 10 18:59:02 localhost kernel: [ 52.279301] NVRM: NVIDIA Linux driver release. Please see ‘Appendix
Dec 10 18:59:02 localhost kernel: [ 52.279301] NVRM: A - Supported NVIDIA GPU Products’ in this release’s
Dec 10 18:59:02 localhost kernel: [ 52.279301] NVRM: README, available on the Linux driver download page
Dec 10 18:59:02 localhost kernel: [ 52.279301] NVRM: at www.nvidia.com.
Dec 10 18:59:02 localhost kernel: [ 52.279416] nvidia: probe of 0000:01:00.0 failed with error -1
Dec 10 18:59:02 localhost kernel: [ 52.279466] nvidia-nvlink: Nvlink Core is being initialized, major device number 243
Dec 10 18:59:02 localhost kernel: [ 52.279493] NVRM: The NVIDIA probe routine failed for 1 device(s).
Dec 10 18:59:02 localhost kernel: [ 52.279495] NVRM: None of the NVIDIA graphics adapters were initialized!
Dec 10 18:59:02 localhost kernel: [ 52.279497] nvidia-nvlink: Unregistered the Nvlink Core, major device number 243

if the full source code was available i could investigate further… but it’s not. greatly appreciated the nvidia team getting this sorted out, promptly. buying a machine for $USD 2,500 including an extremely expensive high-end 1060 graphics card by nvidia and finding that stuff doesn’t work and i can’t fix it myself because you - nvidia - see fit to keep components proprietary - doesn’t go down very well.

http://www.nvidia.com/download/driverResults.aspx/111596/en-us

Supported Products

GeForce 10 Series:
NVIDIA TITAN X (Pascal), GeForce GTX 1080, GeForce GTX 1070, >>>>GeForce GTX 1060<<<<, GeForce GTX 1050 Ti, GeForce GTX 1050

it’s supposed to be a GTX 1060 in this laptop (which i saw somewhere in some debug output). [UPDATE: checked lspci and compared to the PCI id of 1bc0 and it’s a 1060… which is still supposed to be supported].

i’m running the output of nvidia-bug-report.sh but after close inspection there’s no way i’m posting it publicly, so i will email it to linux-bugs@nvidia.com. it actually doesn’t help much because it forgets to include the bumblebee setup (which luckily is the default config supplied, no modifications required).