Crash: qFatal "Tried to show lockscreen surfaces without active lock" in WlSessionLock::updateSurfaces

[Paritsingla7]

What caused the crash

quickshell aborts (SIGABRT) via qFatal() in WlSessionLock::updateSurfaces when the
locked property is set to true from QML while the underlying native session lock
is not yet active.

Crash

quickshell-git 0.3.0.r20.g28771c7-2 (Qt 6.11.2), running as the Omarchy shell.
Two identical crashes on the same session, ~35 minutes apart, both with the exact
same stack trace.

Symbolized backtrace (debuginfod, gdb):

WlSessionLock::updateSurfaces (this=..., show=true, old=0x0)
    at src/wayland/session_lock.cpp:106
106     qFatal() << "Tried to show lockscreen surfaces without active lock";
WlSessionLock::qt_metacall (QMetaObject::WriteProperty) -> moc_session_lock.cpp:175
QQmlPropertyPrivate::write(...)
QV4::QObjectWrapper::setProperty / setQmlProperty / virtualPut
QV4::Lookup::setterFallback
... (QML property write triggered from JS) ...
QQmlBoundSignalExpression::evaluate
QQmlTimer::ticked / QQmlTimer::event
QCoreApplication::notifyInternal2 -> sendPostedEvents -> g_main_context_iteration
QCoreApplication::exec
qs::launch::launch

old=0x0 at the abort site confirms there was no active native lock object when
show=true was requested.

Trigger (QML side)

The QML that flips the property is a WlSessionLock user, e.g.:

WlSessionLock {
    id: sessionLock
    locked: false
    ...
}
// elsewhere:
sessionLock.locked = true

Setting locked = true here reaches updateSurfaces(true, ...) in the native
object while, apparently, the actual Wayland ext-session-lock-v1 object hadn't
been (re)established yet — plausibly right after quickshell restarted itself
from a prior crash, where the QML-level lock state and the native lock object's
lifecycle can get out of sync.

Expected behavior

A QML-level locked = true write racing ahead of the native lock's actual
readiness should not be a fatal, aborting condition — at minimum it should be
recoverable (queue/retry, or ignore with a warning) rather than calling qFatal()
and killing the process.

Environment

• quickshell-git 0.3.0.r20.g28771c7-2
• Qt 6.11.2 (qt6-base, qt6-declarative)
• Compositor: Hyprland, via Omarchy's wayland-wm@hyprland.desktop.service
• Two crashes same session: 2026-08-22 14:28 IST and 15:03 IST, identical stack
  both times (confirmed via coredumpctl + gdb with Arch's debuginfod server)
• No OOM, no relevant package update at the crash times (last quickshell/Qt
  update was ~14 hours prior)

Reproduction

Not yet isolated to a minimal repro — happened twice organically during normal
use, both immediately reproducing the identical fatal path. Happy to provide the
core dump or the two quickshell crash log directories under
~/.cache/quickshell/crashes/ if useful.

Report file

report.txt

Configuration

No response

Backtrace

No response