# Shell crash: "FATAL: Tried to show lockscreen surfaces without active lock" when the shell restarts while a lock is pending

## What happened

While my machine was idle with a lock in progress, the Omarchy shell (quickshell child instance, PID 206773) aborted itself with SIGABRT after hitting quickshell's fatal session-lock check:

```
FATAL: Tried to show lockscreen surfaces without active lock
```

The parent quickshell instance (crash handler) relaunched the shell about a second later, and the restarted instance recovered the lock successfully (`secure=true` ~3 s after the crash). So the impact was a brief shell blip mid-lock, but it is a real crash with a coredump.

## Timeline (BST, 2026-09-06, from the user journal and file mtimes)

```
11:45:51  idle-monitor: idle -> screensaver launched (machine idle, not locked)
11:47:11  omarchy lock: lock-stranded: recovering
          omarchy lock: lock-requested
          omarchy lock: lock-pending: screen-stabilizing     <- lock requested, waiting
11:48:08  ~/.config/omarchy/shell.json modified (mtime 11:48:08.990)
11:48:08  new shell instance starts; duplicate-handler warnings show the previous
          instance was still registered:
          WARN scene: QML IpcHandler ... Handler was registered but will not be used
          because another handler is registered for target omarchy.agents
          (same for omarchy.bluetooth, omarchy.network, omarchy.audio, ... )
11:48:09  FATAL: Tried to show lockscreen surfaces without active lock
11:48:09  ERROR: Quickshell has crashed under pid 206773
11:48:09  Quickshell has been restarted
11:48:12  omarchy lock: secure=true                          <- lock engages fine on retry
```

## Backtrace (coredumpctl, key frames)

Trigger: a QML `Timer` fired and its handler wrote a property that shows lockscreen surfaces before the ext-session-lock was active:

```
#6  abort
#7  libQt6Core.so.6 (qt_message_fatal)
#8  QDebug::~QDebug()                      <- fatal message emitted here
#9  quickshell + 0x832b3                    <- qFatal site (unsymbolized)
#10 quickshell + 0x2240c2                   <- property setter (unsymbolized)
#12 QQmlPropertyPrivate::write(...)
#13 QV4::QObjectWrapper::setProperty(...)
#16 QV4::Lookup::setterFallback(...)
#25 QQmlJavaScriptExpression::evaluate
#26 QQmlBoundSignalExpression::evaluate
#28 QQmlNotifier::emitNotify
#30 QQmlTimer::ticked()                     <- the timer that fired
#32 QCoreApplication::notifyInternal2
```

The two quickshell-owned frames could not be symbolized (no debug symbols via debuginfod); the fatal string itself is compiled into `/usr/bin/quickshell`. Quickshell's own crash report (`~/.cache/quickshell/crashes/<run-id>/report.txt`) records the same stack and log tail.

## Expected behavior

The shell's lock-recovery path (`lock-stranded: recovering` / `lock-pending: screen-stabilizing`) should not make lockscreen surfaces visible until the session lock is actually active. The restarted instance does this correctly, so the happy path exists — the recovery path during a shell restart appears to race the lock state.

## Reproduce (best guess)

1. Let the machine go idle far enough that a lock is requested and sitting in `lock-pending: screen-stabilizing` (screensaver active).
2. While in that state, trigger a shell restart/reload (in my case `~/.config/omarchy/shell.json` was rewritten at 11:48:08.990, which coincides with the new instance starting).
3. The recovering instance's timer flips the lockscreen visible before the session lock engages -> qFatal -> SIGABRT.

What rewrote `shell.json` at that moment I can't tell from the evidence (the machine was idle; an assistant/miner automation was active around that time), so step 2's exact trigger is inferred — the mtime/restart correlation and the duplicate-instance warnings are what's established. The two coexisting shell instances during the restart may be what strands the lock state in the first place.

## System details

- Omarchy 4.0.2-1 (`omarchy version`)
- quickshell 0.3.1-1 (Arch package, RelWithDebInfo), Qt 6.11.2
- Hyprland 0.56.2-1, kernel 7.1.9-arch1-2
- GPU: Intel i915 (8086:9a78), Wayland session
- Shell config: default `/usr/share/omarchy/shell`, user plugins: `kelvinlee.netbird`, `wearable.assistant`
- No OOM kill, no package updates around the crash (quickshell installed 2026-08-31, unchanged since)
- One-off occurrence so far; `omarchy debug` is not available on this install (4.0.2) to attach fuller diagnostics

Filed by GLM (glm-5.3) via opencode.