"Copy Invite Link" Causes Complete System Freeze on Linux

[acasais]

Summary

Clicking the "Copy Invite Link" button in a Zoom meeting causes a complete system freeze on Linux (Wayland/Sway), requiring a hard power-off. The issue is 100% reproducible.

Environment

• Zoom Version: 6.6.11 (6052)
• OS: Manjaro Linux (rolling)
• Kernel: 6.6.107-1-MANJARO
• CPU: 13th Gen Intel Core i7-1355U
• Desktop: Sway 1.11 (Wayland compositor)
• Display Server: Wayland (also tested with XWayland - same result)
• GPU: Intel Iris Xe Graphics (Raptor Lake-P), Mesa 25.2.3-arch1.2
• Audio: PipeWire 1.4.8 + WirePlumber 0.5.11
• Memory: 16GB

Steps to Reproduce

1. Launch Zoom on Linux with Sway/Wayland
2. Create a new meeting or join an existing one
3. Click "Copy Invite Link" (or any clipboard button)
4. Result: Entire system freezes immediately - mouse, keyboard, everything unresponsive
5. Only recovery is holding power button for hard shutdown

Expected Behavior

Invite link should be copied to clipboard without affecting system stability.

Actual Behavior

Complete system freeze requiring hard power-off.

Diagnostic Information

Kernel Logs (from journalctl -b -1 after reboot)

Note: These logs are from crash #2 (XWayland mode). Crash #1 (native Wayland) showed identical RT throttling pattern but logs were overwritten by subsequent reboot.

Dez 17 10:38:25 aplane kernel: warning: `zoom' uses wireless extensions which will stop working for Wi-Fi 7 hardware; use nl80211
Dez 17 10:56:11 aplane kernel: sched: RT throttling activated
Dez 17 10:58:10 aplane kernel: zoom[38499]: segfault at 10 ip 000055c8ff8cfe88 sp 00007ffebe0dd790 error 4 in zoom[55c8fb5ae000+c55e000] likely on CPU 11 (core 15, socket 0)
Dez 17 10:58:10 aplane systemd-coredump[41000]: Process 38499 (zoom) of user 1000 terminated abnormally with signal 11/SEGV, processing...

Timeline:

• 10:56:11 - RT throttling activated (system freezes)
• 10:58:10 - Zoom segfaults (triggered by forced power-off, system was unrecoverable)

Zoom Logs (zoom_stdout_stderr.log)

OpenGL/Qt errors before crash:

QOpenGLFramebufferObject: Framebuffer incomplete attachment.
QOpenGLFramebufferObject: Framebuffer incomplete, missing attachment.
Qt Quick Layouts: Detected recursive rearrange. Aborting after two iterations.

X11/D-Bus connection failures during crash:

[38570:38625:1217/105809.584582:ERROR:ui/gfx/x/connection.cc:64] X connection error received.
The X11 connection broke (error 1). Did the X11 server die?

Massive QML cascade failure:

qrc:/qml/PreMeeting.qml:104: TypeError: Cannot read property 'workflowEnable' of null
qrc:/qml/ChatToastControl.qml:20: TypeError: Cannot call method 'getChatToastModel' of null
qrc:/qml/DragBtnChat.qml:90: TypeError: Cannot read property 'unreadMessageCount' of null
[... 20+ more TypeError lines ...]

Fatal D-Bus error:

[38517:38543:1217/105810.926481:FATAL:dbus/bus.cc:1248] D-Bus connection was disconnected. Aborting.

Workarounds Attempted (All Failed)

Workaround	Command	Result
Force XWayland	QT_QPA_PLATFORM=xcb zoom	Still freezes
Disable GPU sandbox	zoom --disable-gpu-sandbox	Still freezes

Analysis

The crash sequence appears to be:

1. Clipboard operation triggers something in Zoom's Qt/CEF code
2. X11 connection breaks - "Did the X11 server die?" (even under Wayland/XWayland)
3. QML cascade failure - 20+ TypeErrors as UI objects become null
4. RT throttling activates - A real-time priority thread deadlocks the scheduler
5. System freezes - Compositor becomes unresponsive
6. D-Bus disconnects - FATAL error, process aborts
7. Zoom segfaults - Only after forced power-off initiated

The root cause appears to be in Zoom's clipboard handling code, which somehow triggers a cascade that takes down the X11 connection and starves the RT scheduler.

Related Issues

• sway#8810 - "sway freezes when I start zoom too soon" - similar Qt5 segfault pattern
• Zoom Community: Multiple reports of "zoom freezes when stopping screen sharing on Linux"

Additional Context

• This same system had similar clipboard freeze issues in August 2025 with Zoom 6.5.7
• Previous workarounds (XWayland mode) worked then, but no longer work with 6.6.11
• A single application crash should not freeze the entire system, but Zoom manages to do this consistently
• The gap between RT throttling and segfault reflects time spent attempting recovery before forced power-off

Reproducibility

This issue is 100% reproducible on my system. I triggered it twice today (2025-12-17):

1. First crash: Running Zoom normally (native Wayland) → clicked "Copy Invite Link" → system freeze
2. Second crash: Running with `QT_QPA_PLATFORM=xcb zoom` (forced XWayland) → clicked "Copy Invite Link" → system freeze again

Both crashes required hard power-off. The XWayland workaround does not help.

If you need additional logs, different debug output, or specific diagnostics, please let me know and I can reproduce with whatever logging you need.