hardware.nvidia.powerManagement.enable causing system suspend not to work

[LDprg]

Describe the bug

When I am setting hardware.nvidia.powerManagement.enable = true my nvidia gpu wont boot after suspend or hibarnation.

Steps To Reproduce

1. Set hardware.nvidia.powerManagement.enable = true
2. Apply config and reboot
3. Suspend Laptop
4. Try to start from suspend

Expected behavior

Well my displaymanager to start after suspend and my laptop not to crash

Additional context

I have a Asus TUF Gaming Laptop with an Ryzen 7 and Nvidia RTX 3060.

Notify maintainers

Metadata

Please run nix-shell -p nix-info --run "nix-info -m" and paste the result.

[user@system:~]$ nix-shell -p nix-info --run "nix-info -m"
 - system: `"x86_64-linux"`
 - host os: `Linux 6.1.51, NixOS, 23.11 (Tapir), 23.11.20230908.db9208a`
 - multi-user?: `yes`
 - sandbox: `yes`
 - version: `nix-env (Nix) 2.17.0`
 - channels(root): `"nixos"`
 - nixpkgs: `/nix/store/9incrwprw2s90minyy0gpmwbxmrkcx1v-source`

Neofetch

[user@system:~]$  neofetch
          ▗▄▄▄       ▗▄▄▄▄    ▄▄▄▖            ld@LD-Laptop 
          ▜███▙       ▜███▙  ▟███▛            ------------ 
           ▜███▙       ▜███▙▟███▛             OS: NixOS 23.11.20230908.db9208a (Tapir) x86_64 
            ▜███▙       ▜██████▛              Host: ASUSTeK COMPUTER INC. FA506QM 
     ▟█████████████████▙ ▜████▛     ▟▙        Kernel: 6.1.51 
    ▟███████████████████▙ ▜███▙    ▟██▙       Uptime: 10 mins 
           ▄▄▄▄▖           ▜███▙  ▟███▛       Packages: 2007 (nix-system), 940 (nix-user), 18 (flatpak) 
          ▟███▛             ▜██▛ ▟███▛        Shell: fish 3.6.1 
         ▟███▛               ▜▛ ▟███▛         Resolution: 1920x1080 
▟███████████▛                  ▟██████████▙   DE: Plasma 5.27.7 
▜██████████▛                  ▟███████████▛   WM: KWin 
      ▟███▛ ▟▙               ▟███▛            Icons: Nordic-darker [GTK2/3] 
     ▟███▛ ▟██▙             ▟███▛             Terminal: kitty 
    ▟███▛  ▜███▙           ▝▀▀▀▀              Terminal Font: monospace 10.0 
    ▜██▛    ▜███▙ ▜██████████████████▛        CPU: AMD Ryzen 7 5800H with Radeon Graphics (16) @ 4.463GHz 
     ▜▛     ▟████▙ ▜████████████████▛         GPU: NVIDIA GeForce RTX 3060 Mobile / Max-Q 
           ▟██████▙       ▜███▙               GPU: AMD ATI Radeon Vega Series / Radeon Vega Mobile Series 
          ▟███▛▜███▙       ▜███▙              Memory: 3618MiB / 31507MiB 
         ▟███▛  ▜███▙       ▜███▙
         ▝▀▀▀    ▀▀▀▀▘       ▀▀▀▘                                     

[LDprg]

Hybernation is also not working with powerManagment enabled!

[FrankelJb]

I'm experiencing the same problem. meta output below. Happy to share any other outputs that can help to troubleshoot

› nix-shell -p nix-info --run "nix-info -m"
 - system: `"x86_64-linux"`
 - host os: `Linux 6.1.51, NixOS, 23.11 (Tapir), 23.11.20230908.db9208a`
 - multi-user?: `yes`
 - sandbox: `yes`
 - version: `nix-env (Nix) 2.17.0`
 - nixpkgs: `/nix/store/9incrwprw2s90minyy0gpmwbxmrkcx1v-source`

[pluiedev]

I have the same issue, and it's keeping me from switching to Wayland permanently — I can confirm that suspend, sleep and hibernate work perfectly when I switch to X11 within the same user session, so it looks like it's definitely a Wayland + Nvidia issue. (Tested on a RTX 3060 Mobile)

[profiluefter]

Same issue, am having it for a few months already (so it's not recent). If it's enabled my GPU also doesn't start at all and wayland stays blank on KDE Plasma. However if I disable it it just results in corrupted graphics after suspending which isn't much better. (Also I can screenshot it so I'm pretty sure it's not my faulty GPU)

Edit: GPUs I had this issue with: GTX 1080 Ti, GTX 1050 Ti

[edmundmiller]

We should probably start including GPUs in these comments 🫠

Personally: 2070 SUPER 545.29.02

[benoitdr]

Same here, but only when 2nd monitor (hdmi) is connected

nix-shell -p nix-info --run "nix-info -m"
 - system: `"x86_64-linux"`
 - host os: `Linux 6.1.69, NixOS, 23.11 (Tapir), 23.11.2413.32f63574c85f`
 - multi-user?: `yes`
 - sandbox: `yes`
 - version: `nix-env (Nix) 2.18.1`
 - channels(root): `"home-manager-23.11.tar.gz, nixos-23.11"`
 - nixpkgs: `/nix/var/nix/profiles/per-user/root/channels/nixos`

with NVIDIA RTX A4000 Laptop GPU

[LDprg]

I made a new discovery! The problem only occurs when using the open kernel module for NVIDIA, which make kind of sense.

Maybe there should be a check that open and powerManagement.enable are not enabled at the same time.

Edit: Hibernation working too

[programotuojes]

The problem only occurs when using the open kernel module for NVIDIA

Thanks to you, I also made a discovery. Despite not using the open kernel module, I experienced the same symptoms. Turns out it was due to me using an older driver version - 535.129.03 (referenced through config.boot.kernelPackages.nvidiaPackages.production). The newer one (545.29.02) works fine. I made this change, because on the newer version screen brightness control does not work. I guess I'll just have to choose the lesser of the two devils.

Maybe there should be a check that open and powerManagement.enable are not enabled at the same time.

This bug occurs when using Plasma with Wayland. A check would be overkill IMO, though it would be nice to have it at least mentioned in the option's description. But even that's probably not worth it, as per https://github.com/NVIDIA/open-gpu-kernel-modules/issues/472#issuecomment-1918464628, the fix is being actively worked on. Hopefully it's fixed in the near future.

[LDprg]

@programotuojes are you using kde plasma 5 or 6? I got no problems using kde 6 and latest drivers.

[pwaller]

@LDprg

I made a new discovery! The problem only occurs when using the open kernel module for NVIDIA, which make kind of sense.

Unfortuantely on my GeForce 1080, I have nix eval .#nixosConfigurations.machinename.config.hardware.nvidia.open set to false;

With hardware.nvidia.powerManagement.enable = true:

• On wayland I get machine lockups from suspend
• (Xorg untested)

With hardware.nvidia.powerManagement.enable = false:

• With wayland I get significant, persistent graphical corruption on resume
• Xorg seems to work; suspend/resume works and no graphical corruption.

[LDprg]

@pwaller Try with and without the open nvidia kernel drivers. They should have fixed the open ones (afaik) with the 555 driver.

[pwaller]

I think my hardware is too old for the open drivers, it's a 10xx series where the open driver started at the ~20xx~ (edit: 16xx) series.