From: Chédotal Julien <rjdi@wanadoo.fr>
Subject: [moonlight-qt] CRITICAL ASSERT: m_QtWindow != nullptr - Crash on game start with AMDGPU
To: ports@openbsd.org
Date: Sat, 15 Nov 2025 15:11:55 +0100

Dear OpenBSD ports maintainers,

I am reporting a critical crash affecting the |moonlight-qt| package 
when attempting to start a streaming session.


      1. System Information

  *

    *Operating System:* OpenBSD*| 7.8|*

  *

    *Kernel:**| OpenBSD i7.my.domain 7.8 GENERIC.MP#54 amd64|*

  *

    *Package Name/Version:**| moonlight-qt-6.1.0|*

  *

    *Graphics Driver:* |amdgpu(4) -> amdgpu0 at pci2 dev 0 function 0
    "ATI Polaris 12" rev 0xc7|

  *

    *Window Manager:**| xfce4|*


      2. Problem Description and Behavior

The application launches successfully, and the client is able to *pair 
with the host PC* and *display the list of available games*.

The crash occurs at the precise moment a game is selected, which is when 
the client attempts to initialize the video/audio decoding pipeline to 
start the stream. The session fails immediately before the full-screen 
window appears.

The program fails with the following *Qt Fatal Assertion* in the console:

|00:00:27 - Qt Fatal: ASSERT: "m_QtWindow != nullptr in file 
/usr/obj/ports/moonlight-qt-6.1.0/app/streaming/session.cpp, line 1269 
Abort trap (core dumped) |

his error indicates that the application's streaming session logic is 
attempting to access the main Qt streaming window object (|m_QtWindow|), 
but this object is null (|nullptr|), causing an immediate fatal error 
before the streaming view can be rendered. This points to a failure in 
the creation or initialization of the window object during the 
transition from the UI to the streaming session.


      3. Steps Taken to Troubleshoot

I have attempted the following initial troubleshooting steps, but the 
issue persists:

 1.

    *Explicitly tested the XCB platform backend*: Launching with
    |QT_QPA_PLATFORM=xcb moonlight-qt| did *not* resolve the crash.

 2.

    *Tested with Software Decoding*: Forcing software decoding via the
    client's options did *not* prevent the crash.

 3.

    *Tested in Windowed Mode*: Attempting to launch the stream in
    windowed mode (instead of full screen) did *not* prevent the crash.

Thank you for your time and assistance in maintaining the ports collection.

Best regards,

Chédotal Julien - rjdi@wanadoo.fr