nixos/hardware/nvidia: relax conditions eager loading nvidia-uvm

[GaetanLepage]

Things done

Fixes contingent CUDA issues with open driver:
#334180

cc @SomeoneSerge @ConnorBaker @Atry @eljamm

• Built on platform(s)
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• For non-Linux: Is sandboxing enabled in nix.conf? (See Nix manual)
  • sandbox = relaxed
  • sandbox = true
• Tested, as applicable:
  • NixOS test(s) (look inside nixos/tests)
  • and/or package tests
  • or, for functions and "core" functionality, tests in lib/tests or pkgs/test
  • made sure NixOS tests are linked to the relevant packages
• Tested compilation of all packages that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage
• Tested basic functionality of all binary files (usually in ./result/bin/)
• 25.05 Release Notes (or backporting 24.11 and 25.05 Release notes)
  • (Package updates) Added a release notes entry if the change is major or breaking
  • (Module updates) Added a release notes entry if the change is significant
  • (Module addition) Added a release notes entry if adding a new NixOS module
• Fits CONTRIBUTING.md.

---

Add a 👍 reaction to pull requests you find important.

[GaetanLepage]

I confirm that this has fixed my use case (with mistral-rs).
Let's hope it does not break any other use cases.

[eljamm]

LGTM, thanks.

[GaetanLepage]

nixpkgs-review result

Generated using nixpkgs-review.

Command: nixpkgs-review pr 398355

---

x86_64-linux

⏩ 2 packages blacklisted:

• nixos-install-tools
• tests.nixos-functions.nixos-test

---

aarch64-linux

⏩ 2 packages blacklisted:

• nixos-install-tools
• tests.nixos-functions.nixos-test

[SomeoneSerge]

@GaetanLepage now that we remembered https://github.com/NixOS/nixpkgs/pull/334340/files, could you please rewrite the commit message to cite the previous commit and to explain that we're making the decision to eager-load the module for all users of the open driver because otherwise we (evidently) still get confused by the error on headless configurations. If we encounter ordering issues again (loading nvidia-uvm too early) we follow up by rolling out a more granular solution. One way to achieve that might be to go forward refactoring hardware.nvidia and disentangling it from services.xserver

[GaetanLepage]

Done, feel free to suggest a better wording.