[Illustar0]

What does the pull request do?

Improves automatic DRM card detection.

Avalonia documentation states:

// Avalonia auto-detects the output card.
// To specify one explicitly: card: "/dev/dri/card1"
return builder.StartLinuxDrm(args, card: null, options: new DrmOutputOptions

However, the previous detection simply opened the first DRM card returned by the file system.

What is the current behavior?

When no card path is specified, Avalonia opens the first matching /dev/dri/cardN, even if it does not expose usable KMS resources.

What is the updated/expected behavior with this PR?

Avalonia checks DRM cards in numeric order and selects the first card exposing KMS resources with at least one CRTC, encoder, and connector.
Unsuitable cards are closed and skipped.

How was the solution implemented (if it's not obvious)?

Card names are validated and sorted numerically. Each candidate is opened and checked with drmModeGetResources.

Checklist

• Added unit tests (if possible)?
• Added XML documentation to any related classes?
• Consider submitting a PR to https://github.com/AvaloniaUI/avalonia-docs with user documentation

Breaking changes

None.

Obsoletions / Deprecations

None.

Fixed issues

None.

[avaloniaui-bot]

You can test this PR using the following package version. 12.2.999-cibuild0068093-alpha. (feed url: https://nuget-feed-all.avaloniaui.net/v3/index.json) [PRBUILDID]

[cla-avalonia]

• All contributors have signed the CLA.

[Illustar0]

@cla-avalonia agree

[Illustar0]

On second thought, I don't think using null to mean “auto-detect a DRM card” is ideal API design. It would be clearer to expose auto-detection through a separate method and have the DrmCard constructor accept only an explicit, non-nullable device path.

That would be a breaking change, though, so it would probably need to be considered separately from this fix.

[MrJul]

A minor change is needed but overall the PR looks good.

[MrJul]

On second thought, I don't think using null to mean “auto-detect a DRM card” is ideal API design. It would be clearer to expose auto-detection through a separate method and have the DrmCard constructor accept only an explicit, non-nullable device path.

I agree that the current design isn't the best, especially as a constructor.
Feel free to open an issue with a proposal so we can review it.

[avaloniaui-bot]

You can test this PR using the following package version. 12.2.999-cibuild0068137-alpha. (feed url: https://nuget-feed-all.avaloniaui.net/v3/index.json) [PRBUILDID]

[MrJul]

LGTM!

[avaloniaui-bot]

You can test this PR using the following package version. 12.2.999-cibuild0068145-alpha. (feed url: https://nuget-feed-all.avaloniaui.net/v3/index.json) [PRBUILDID]

[Illustar0]

@MrJul Apologies for the bother, it seems the CI failure might be a flaky test unrelated to my commits. Could you please give it a kick and re-run the CI?