Skip to content

display: add DRM helpers and fix core_auth skip handling #372

Open
smuppand wants to merge 2 commits intoqualcomm-linux:mainfrom
smuppand:core-auth
Open

display: add DRM helpers and fix core_auth skip handling #372
smuppand wants to merge 2 commits intoqualcomm-linux:mainfrom
smuppand:core-auth

Conversation

@smuppand
Copy link
Copy Markdown
Contributor

@smuppand smuppand commented Mar 28, 2026

This PR adds reusable display helpers in lib_display.sh and updates the core_auth runner to report skip-only cases correctly. Also addressed the issue #327

Summary

  • add get_drm_primary_node()
  • add weston_cleanup_stale_sockets()
  • source lib_display.sh from core_auth/run.sh
  • skip core_auth when no DRM primary node is present
  • map IGT rc=77 to SKIP instead of FAIL

This fixes incorrect FAIL results in headless or display-disconnected LAVA runs while keeping normal manual runs unchanged.

Results with these changes:
https://lava.infra.foundries.io/scheduler/job/166573
https://lava.infra.foundries.io/scheduler/job/166574

Update Runner/utils/lib_display.sh to add two reusable display helpers:

- get_drm_primary_node()
  Returns the first available DRM primary node under /dev/dri/card* so
  display and IGT tests can detect primary DRM availability consistently.

- weston_cleanup_stale_sockets()
  Removes stale Wayland socket files only when Weston is not running,
  using best-effort cleanup for common Weston runtime locations.

These helpers improve reuse across display-related tests and remove
runtime noise from Weston stop/start paths without changing normal
execution flow.

Signed-off-by: Srikanth Muppandam <smuppand@qti.qualcomm.com>
@smuppand smuppand requested review from abbajaj806 and vnarapar March 28, 2026 15:37
@smuppand
Copy link
Copy Markdown
Contributor Author

@vsalipal - Please review this as well. I was unable to add you as an assignee, possibly because you are not part of the qualcomm-linux group.

@mwasilew
Copy link
Copy Markdown
Contributor

I don't like this patch. It seems to be masking the root cause of the issue instead of fixing it.

@smuppand
Copy link
Copy Markdown
Contributor Author

I don't like this patch. It seems to be masking the root cause of the issue instead of fixing it.

Thanks for the feedback.

The rc=77 / skip-only handling is intended as a result-classification fix, but the new pre-skip on missing DRM primary node can indeed be seen as masking the deeper issue.

@smuppand
Copy link
Copy Markdown
Contributor Author

I don't like this patch. It seems to be masking the root cause of the issue instead of fixing it.

I’ve updated the patch to remove the proactive DRM-node skip. The runner now only logs which DRM primary node is being used when present, or warns if none is found, and still lets core_auth run and determine the actual outcome itself.

So the patch stays focused on:

  • fixing the missing display helper path
  • classifying actual IGT skip outcomes correctly
  • avoiding a runner-side precondition that could mask the underlying issue

Update the core_auth runner to preserve IGT skip semantics without
adding a runner-side skip gate.

Changes in this update:
- use get_drm_primary_node() only for diagnostic logging
- log the DRM primary node when present
- warn and continue when no DRM primary node is found
- treat IGT rc=77 as SKIP instead of FAIL
- keep other non-zero return codes as FAIL

Signed-off-by: Srikanth Muppandam <smuppand@qti.qualcomm.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants