## System details

Lenovo ThinkPad X260 (Intel i7-6600U, Intel HD 520/i915, Validity VFS5011 fingerprint reader 138a:0017); Omarchy 4.0.4-1; Linux 7.2.5-3-omarchy; Hyprland 0.56.2-2; quickshell 0.3.1-1; Qt 6.11.2; fprintd 1.94.5-2; libfprint-git 1:1.94.100.r10.g6f9479c-1; PAM 1.7.2-2.

## What's wrong?

The stock Omarchy shell at `/usr/share/omarchy/shell/shell.qml` has crashed with SIGSEGV three times during fingerprint lock authentication on the same boot:

```text
2026-10-02 00:56:53 EDT  PID 3999355
2026-10-04 19:08:14 EDT  PID 1721871
2026-10-05 08:38:20 EDT  PID 2987514
```

Each crash occurred while an `omarchy-lock-fingerprint` PAM conversation was active and waiting for, or timing out on, the fingerprint reader. In the October 5 occurrence, the Validity VFS5011 reader disconnected and re-enumerated shortly before authentication resumed; Quickshell crashed about 15 seconds after the new fingerprint prompt.

The three cores vary within the same Qt QML lifecycle/garbage-collection area:

```text
Oct 2:
QV4::Heap::Object::markObjects
QV4::MarkStack::drain
QV4::GCStateMachine::transition

Oct 4:
libQt6Qml
QV4::MarkStack::drain
QV4::GCStateMachine::transition

Oct 5:
QQmlData::destroyed(QObject*)
QObject::~QObject()
QQuickItem::~QQuickItem()
QObjectPrivate::deleteChildren()
QObject::event(QEvent*)
QCoreApplicationPrivate::sendPostedEvents(...)
```

The October 5 coredump reports `si_code: SI_TKILL` because Quickshell's crash handler re-raises the original SIGSEGV. The faulting path beneath the handler is the Qt/QML object-destruction stack above.

After each crash, Quickshell restarted itself and the replacement shell logged:

```text
lock-stranded: recovering
lock-requested
lock-pending: screen-stabilizing
secure=true
```

Fingerprint PAM retries then frequently failed because fprintd reported `Device was already claimed`.

No OOM condition was present: the journal contains no OOM-kill evidence, and several GiB of RAM plus ample swap were available. No relevant Quickshell or Qt package update occurred immediately before the crashes. No user data was lost; the shell restarted automatically and recovered the lock.

This looks like stale or corrupted QML object state exposed during lock/fingerprint lifecycle changes, but the available symbols do not identify the exact QML object, so I am not asserting a more specific root cause.

## Related issues checked

- #7645 covers a fingerprint-related crash in `PamSubprocess::conversation()` with `libjemalloc`/`pam_fprintd` frames. These cores do not have that stack; they fail in Qt QML GC/object destruction.
- #11824 has a closely matching `QQmlData::destroyed()` teardown stack, but its documented trigger is monitor add/remove. No monitor event preceded these three crashes; the common event here is active fingerprint lock authentication.
- #8973 concerns cross-window `QQuickItem` reuse at lock time and logs `Cannot use same item on different windows`; those warnings are absent here.

Full Quickshell reports remain available locally under `~/.cache/quickshell/crashes/`, and the October 5 systemd core is still present.

Filed by GPT-5.6-Sol via Hermes Agent.