• Thanks, this patch works for me on 5.6.0-0.rc5.git0.2.fc32.x86_64 #1

    We are still trying to fix this one https://bugzilla.redhat.com/show_bug.cgi?id=1806257#c3 but still no luck

    Edited by Artem Silenkov
  • Working here for me on rc6 as well without major changes. I'm running dkms to build so I drop the leading /kernel/ from the paths and run the patch as root in /usr/src/nvidia-440.64/ Thanks for this Isaak!

    edit: Working on rc7 here too!

    Edited by Zachariah Anderson
  • I was desperately looking for a solution to get my ASUS TUF Gaming Notebook to work (Ryzen 3750H with NVIDIA RTX 2060). Thank you very much for your great work! Finally, kernel 5.6-RC7 works with the RC7 patch (available also here on Gitlab, forget the link). Have to struggle with xorg.conf (as usual) but kernel boots with mode setting and NVIDIA driver loaded. Thx a lot, guys!!! UPDATE: Topic solved with the help of some people from the NVIDIA developer forum: My ASUS TUF Gaming Notebook comes with Ryzen 3750H and RTX 2060 as stated above. Just wanted to have the NVIDIA X11 driver alone (and not the AMDGPU driver) working in X11 (my plain Debian 10 'buster' does not come with any bumblebee or prime). Works now with the 'modesetting driver trick' I did not know before. You can look here:

    https://forums.developer.nvidia.com/t/amd-gpu-and-nvidia-rtx-2060-together-in-a-notebook-problems/117027/10

    Thanks again for the great driver work.

    Edited by Ingo Kubbilun
  • Tested this patch with 5.6.0 kernel, works fine, except for the refcount_t underflow issue: As with 5.5.x series, if I load nvidia-drm with modeset=1, I still hit this bug (as mentioned before): https://bugzilla.redhat.com/show_bug.cgi?id=1806257

  • I should mention that e.g. PRIME Synchronization and such actually seems to work with 5.6.0 even if you get that bug. So I don't know if the bug matters that much...

  • @madsl Yeah, it shouldn't affect stuff much beyond the kernel spitting out that underflow warning. It's as Steve in the RedHat tracker pointed out, that it's a refcounted object somewhere that's being double free'd/refcount_dec'd. I'm not any good at debugging these kinds of faults (or debug in general, honestly) especially so since the three things I suspected were apparently a bust.

    Still, as I can't test that functionality myself, it's nice to hear that it's working as intended (albeit with that refcount annoyance present).

  • Note that this also works on the Nvidia Beta Drivers (440.66.08 is the latest). Thank you so much for the patch!

Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment