fix(session): wait for lock screen to render before suspending

[Aitor42]

Summary

This PR ensures that the deferred suspend command is executed only after all active lock screen surfaces have successfully drawn and committed their first frame.

• Overrode render() in LockSurface to track when m_firstFrameRendered is set to true and trigger a render callback.
• Added allSurfacesReady() in LockScreen to verify if all output surfaces have completed their first render.
• Updated runAfterSessionLocked and handleLocked to defer execution of the pending suspend callback until allSurfacesReady() is true.

Motivation

This resolves a race condition where the system suspends before the lock screen has finished rendering. When resuming, the unlocked desktop could briefly flash or remain visible, presenting a security vulnerability and visual glitch.

Type of Change

• Bug fix
• New feature
• Breaking change
• Refactoring
• Build / packaging

Related Issue

Closes #2962

Testing

Built the daemon and ran the entire test suite:
meson compile -C build
meson test -C build
All tests pass successfully.

Manual Coverage

• Tested on Niri
• Tested on Hyprland
• Tested on Sway
• Tested on another compositor:
• Tested with different bar positions and density settings
• Tested at different interface scaling values
• Tested with multiple monitors

Checklist

• This PR is ready for review, or it is marked as Draft.
• I read and followed the relevant guidance in CONTRIBUTING.md.
• I ran just format with clang-format v22+ installed, or this PR has no code changes.
• I ran the relevant build or test commands, or explained why they were not run.
• I self-reviewed the changes.
• I checked for new warnings or errors.
• I will update end-user documentation after merge, or this PR does not change user-facing configuration or behavior.
• I added or updated assets/translations/en.json, or this PR adds no new user-facing strings.
• I did not edit non-English translation files unless this PR is explicitly for translation tooling, an import/export sync, or a maintainer-requested locale change.
• I used the existing canonical names for config keys, IPC names, paths, and identifiers.

Additional Notes

If a lock surface fails to configure or render due to driver glitches or compositor-specific edge cases, the suspend sequence could get indefinitely postponed. I sugest adding a 2-3 second fallback timer to trigger suspend regardless of render status.

[Ly-sec]

Hey @Aitor42 , thanks for tackling this, the direction looks good to me.

The race is real: handleLocked is compositor state, not "pixels are on screen." Waiting for a first render() on every LockSurface before running the pending suspend callback is a sensible fix for #2962 .

I'd want a timeout fallback before merge though. If a surface never renders, suspend could hang forever. A 2-3s timer that warns and runs the callback anyway is a reasonable tradeoff.

Minor nits: the flush-when-ready logic is duplicated in three places (a tryFlushPendingAfterLocked() helper would help), and firstFrameRendered could reset on reconfigure for accuracy mid-lock. Neither is a blocker.

Manual testing is fine here. Nice work overall.

[Aitor42]

Hey @Aitor42 , thanks for tackling this, the direction looks good to me.

The race is real: handleLocked is compositor state, not "pixels are on screen." Waiting for a first render() on every LockSurface before running the pending suspend callback is a sensible fix for #2962 .

I'd want a timeout fallback before merge though. If a surface never renders, suspend could hang forever. A 2-3s timer that warns and runs the callback anyway is a reasonable tradeoff.

Minor nits: the flush-when-ready logic is duplicated in three places (a tryFlushPendingAfterLocked() helper would help), and firstFrameRendered could reset on reconfigure for accuracy mid-lock. Neither is a blocker.

Manual testing is fine here. Nice work overall.

Thanks for the feedback! I added the timeout to 3 seconds and the tryFlushPendingAfterLocked() function. For the firstFrameRendered rest I only change it when a real geometry change occurs, preventing unnecessary re-blocking of the flush during non-sizing configure events.

Let me know if you see something else.