https://bugs.kde.org/show_bug.cgi?id=527822
Bug ID: 527822
Summary: Plasmashell fails after prolonged screen lock with
external displays: "global wl_output is unavailable",
followed by shutdown timeout
Classification: Plasma
Product: plasmashell
Version First 6.7.5
Reported In:
Platform: Fedora RPMs
OS: Linux
Status: REPORTED
Severity: crash
Priority: NOR
Component: generic-crash
Assignee: [email protected]
Reporter: [email protected]
Target Milestone: 1.0
## Description
I recently encountered a Plasma desktop failure after leaving my session locked
for a prolonged period while using an external monitor connected directly to my
HP laptop through USB-C, without an external dock or adapter on that
connection.
The most useful recorded incident occurred on October 26, 2026, at 09:10:42
(UTC+08:00). Shortly after external-display disconnection and reconnection
events, plasmashell reported a Wayland protocol error:
```text
wl_registry#2: error 0: global wl_output (78) is unavailable
```
KWin simultaneously reported a client communication error for the plasmashell
process. During a subsequent reboot, plasmashell did not terminate within 40
seconds, so systemd sent SIGABRT. This final abort was caused by a stop timeout
and should not be mistaken for the initial failure.
## Steps preceding the problem
1. Use a Plasma Wayland session with an external monitor directly connected
through USB-C.
2. Lock the session and leave it locked for a prolonged period.
3. Return to the computer and wake/unlock the session.
4. Observe that the Plasma desktop no longer operates normally.
This describes the usage pattern, not a verified minimal reproducer. I have not
established the minimum lock duration or reproduction frequency. The retained
session also included suspend/resume cycles, so a lock-only trigger has not
been isolated.
## Observed result
Plasmashell encounters a Wayland output-global error after display-state
changes. It subsequently fails to terminate cleanly during reboot and is
aborted by systemd after a timeout.
## Expected result
The Plasma desktop should remain functional across screen locking, monitor
sleep/wake, and output removal/reconnection. Plasmashell should also terminate
cleanly during logout or reboot.
## Relevant logs
All timestamps below are local time, UTC+08:00. Hostnames have been omitted.
```text
2026-10-26 09:10:31 kernel:
usb 3-2: USB disconnect, device number 24
usb 3-2.1: USB disconnect, device number 25
2026-10-26 09:10:33 org_kde_powerdevil[2142]:
DDCA_EVENT_DISPLAY_DISCONNECTED, card1-HDMI-A-1
2026-10-26 09:10:35 org_kde_powerdevil[2142]:
DDCA_EVENT_DISPLAY_CONNECTED, card1-HDMI-A-1
2026-10-26 09:10:42 kwin_wayland_wrapper[1923]:
error in client communication (pid 2097)
2026-10-26 09:10:42 plasmashell[2097]:
wl_registry#2: error 0: global wl_output (78) is unavailable
QSocketNotifier: Socket notifiers cannot be enabled or disabled from
another thread
QSocketNotifier: Socket notifiers cannot be enabled or disabled from
another thread
2026-10-26 09:13:00 systemd-logind[1117]:
The system will reboot now!
2026-10-26 09:13:00 systemd[1637]:
Stopping plasma-plasmashell.service - KDE Plasma Workspace...
2026-10-26 09:13:40 systemd[1637]:
plasma-plasmashell.service: State 'stop-sigterm' timed out. Aborting.
plasma-plasmashell.service: Killing process 2097 (plasmashell) with signal
SIGABRT.
plasma-plasmashell.service: Main process exited, code=dumped, status=6/ABRT
plasma-plasmashell.service: Failed with result 'timeout'.
```
The DDC events explicitly identify HDMI-A-1. The logs do not establish which
physical display corresponds to Wayland global 78, whether the USB-C display
was removed, or whether the USB disconnection came from a particular monitor.
They also do not distinguish physical unplugging from power cycling or link
renegotiation.
No corresponding retained core dump for the October 26 plasmashell process
appeared in `coredumpctl list`.
## System information
| Item | Details
|
| ----------------------- |
---------------------------------------------------------- |
| Operating system | Fedora Linux 44, KDE Plasma Desktop Edition, x86_64
|
| Session | Wayland
|
| Plasma | 6.7.5 (`plasma-workspace-6.7.5-1.fc44`)
|
| KWin | 6.7.5 (`kwin-6.7.5-1.fc44`)
|
| Qt Base | 6.11.2 (`qt6-qtbase-6.11.2-3.fc44`)
|
| Qt Wayland | 6.11.2 (`qt6-qtwayland-6.11.2-1.fc44`)
|
| KDE Frameworks packages | KCoreAddons and KGuiAddons 6.31.0
|
| Mesa DRI drivers | 26.2.5
|
| Kernel | 7.2.11-200.fc44.x86_64, verified in the affected
boot |
| Laptop | HP ProBook 440 14 inch G10 Notebook PC
|
| CPU | Intel Core i5-1340P
|
| GPU | Intel Raptor Lake-P / Iris Xe Graphics, PCI ID
`8086:a7a0` |
| Kernel graphics driver | `i915`
|
| BIOS | HP V72 Ver. 01.12.04, dated 2026-05-07
|
## Display configuration
The following configuration was collected after reboot using DRM sysfs EDID and
`kscreen-doctor`. It is the current configuration, not a historical snapshot of
the failure.
| Output | EDID identity | Current mode
| Orientation |
| -------- | ------------------------------------------- |
---------------------- | ------------------ |
| DP-1 | ViewSonic / VSC VX2762-2K, product `0x623c` | 2560×1440 at 100 Hz
| Portrait |
| HDMI-A-1 | HKC MG24Q, product `0x2420` | 2560×1440 at 143.91
Hz | Landscape, primary |
| eDP-1 | LGD internal panel, product `0x0763` | 1920×1080 at 60.02
Hz | Landscape |
All three outputs are currently enabled at scale 1. HDR is disabled or
unsupported, and VRR is reported as incapable.
A direct USB-C connection is confirmed, but the mapping between that cable and
the listed external monitors, and whether all three outputs were enabled during
the incident, have not yet been confirmed.
## USB-C information
[HP's official specification for another ProBook 440 G10
configuration](https://www.hp.com/id-en/shop/hp-probook-440-14-inch-g10-notebook-pc-wolf-pro-security-edition-8c415pa.html)
lists two USB-C ports supporting 10 Gbps USB signaling, USB Power Delivery and
DisplayPort 2.1. This is model-family documentation for a different SKU, not a
measurement of this machine's negotiated connection.
Local PCI enumeration includes an Intel Raptor Lake-P Thunderbolt 4 USB
Controller (`8086:a71e`). This controller name alone does not establish
Thunderbolt support on the physical port or indicate that the monitor uses a
Thunderbolt connection.
`/sys/class/typec` and `/sys/bus/thunderbolt/devices` contain no entries. The
negotiated DisplayPort rate/lane count, USB-C alt mode, PD contract, monitor
USB-C specification and cable rating are therefore unknown.
## Additional shutdown-hang evidence
A separate incident on October 24 produced a retained core dump after systemd
aborted plasmashell for exceeding its stop timeout. The main thread was waiting
in:
```text
WaylandClipboard::~WaylandClipboard()
QThread::wait()
pthread join / futex wait
```
The clipboard thread was in:
```text
ClipboardThread::run()
QRecursiveMutex::tryLock()
QBasicMutex::lockInternal()
```
These frames involve `libKF6GuiAddons`. The selected stacks are available in
`earlier-shutdown-stack.txt`. They demonstrate an exit-time wait, not the
initial October 26 protocol failure. I have not established that the two
incidents share a root cause; the shutdown hang may warrant a separate report.
## Attachments
- `relevant-journal.txt`: selected October 26 display, Wayland protocol and
stop-timeout events.
- `earlier-shutdown-stack.txt`: selected threads from the separate October 24
exit-time hang.
Hardware serial numbers, raw EDID, authentication logs and process-memory dumps
are not included.
Please reassign this report if the initial protocol failure belongs to KWin,
QtWayland or another component.
--
You are receiving this mail because:
You are watching all bug changes.
