nixos/ipu6: Don't build out-of-tree driver for kernels that have it

[shlevy]

Description of changes

The ipu6 drivers are in-tree as of Linux 6.10, so we can stop building them for newer kernels

Things done

• 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/)
• 24.11 Release Notes (or backporting 23.11 and 24.05 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.

[flokli]

Do we want to remove them from the attrset alltogether too?

I can still try to build linuxPackages_latest.ipu6-drivers, and it throws a lot of build errors. Should probably only be available for kernels that don't have it in-tree.

pkgs/top-level/linux-kernels.nix seems to have a bunch of if lib.versionOlder kernel.version "x.y" then callPackage ../os-specific/linux/something { } else null;.

[shlevy]

@flokli Done

[github-actions]

Successfully created backport PR for release-24.05:

• [Backport release-24.05] nixos/ipu6: Don't build out-of-tree driver for kernels that have it #332240

[flokli]

@shlevy I was just about to test this out, but unfortunately v4l2-relayd-ipu6.service doesn't seem to come up again:

× v4l2-relayd-ipu6.service - Streaming relay for v4l2loopback using GStreamer
     Loaded: loaded (/etc/systemd/system/v4l2-relayd-ipu6.service; enabled; preset: enabled)
     Active: failed (Result: exit-code) since Sun 2024-08-04 16:25:50 EEST; 1min 58s ago
 Invocation: 0e2c6407e7fb490b964a120dc6c3215a
    Process: 69115 ExecStartPre=/nix/store/qhbqvknp0b1a330f4kbkjnq3smij1hbb-unit-script-v4l2-relayd-ipu6-pre-start/bin/v4l2-relayd-ipu6-pre-start (code=exited, status=1/FAILURE)
    Process: 69126 ExecStopPost=/nix/store/zv3qyvvkv82a9dg59261yjzm2c8jrr3i-unit-script-v4l2-relayd-ipu6-post-stop/bin/v4l2-relayd-ipu6-post-stop (code=exited, status=1/FAILURE)

Aug 04 16:25:50 x1 systemd[1]: v4l2-relayd-ipu6.service: Scheduled restart job, restart counter is at 5.
Aug 04 16:25:50 x1 systemd[1]: v4l2-relayd-ipu6.service: Start request repeated too quickly.
Aug 04 16:25:50 x1 systemd[1]: v4l2-relayd-ipu6.service: Failed with result 'exit-code'.
Aug 04 16:25:50 x1 systemd[1]: Failed to start Streaming relay for v4l2loopback using GStreamer.

Looking at the logs, maybe the naming did change, or this is otherwise incompatible? How do I test this?

[shlevy]

Ack, I’m traveling without my personal ipu6 laptop this week, will look into this when I get back.

[matejc]

For me it fails with:

Aug 13 16:10:15 nixcode v4l2-relayd-ipu6-pre-start[15383]: failed to create device: Inappropriate ioctl for device

when pre-start is doing:

v4l2loopback-ctl add -x 1 ....

I can also confirm that this produces the same error:

sudo v4l2loopback-ctl add

I have tried to unload and load the v4l2loopback the module and still same issue. Other observations: I do have /dev/video0 to /dev/video32 and all in between for some reason. Also... I am running on kernel 6.10.3

[matejc]

We might need to update v4l2loopback https://github.com/umlaeute/v4l2loopback/tags
Our one is 12.7 which is two years back and there has been 3 versions released since then

[flokli]

Sent a bump for v4l2loopback #334573.

With this, at least the service doesn't crash anymore. If I select the loopback device in the browser, I see the following in the logs:

Aug 14 12:06:41 x1 pipewire[3035]: spa.v4l2: '/dev/video0' VIDIOC_REQBUFS Invalid argument
Aug 14 12:06:41 x1 pipewire[3035]: pw.port: 0x55625008def0: negotiate buffers on node: -22 (Invalid argument)
Aug 14 12:06:41 x1 pipewire[3035]: pw.port: 0x55625008def0: state ready -> error (can't negotiate buffers on port)
Aug 14 12:06:41 x1 pipewire[3035]: pw.link: (195.0.0 -> 351.0.0) allocating -> error (can't negotiate buffers on port) (error-ready)

The camera still doesn't work, but it feels like a step into the right direction.