Bug 167990 - Appearance theme extension removal restart destructor fails with skia Vulkan rendering on NVidia GPU
Summary: Appearance theme extension removal restart destructor fails with skia Vulkan ...
Status: UNCONFIRMED
Alias: None
Product: LibreOffice
Classification: Unclassified
Component: graphics stack (show other bugs)
Version:
(earliest affected)
25.8.0.0 alpha0+
Hardware: x86-64 (AMD64) Windows (All)
: medium normal
Assignee: Not Assigned
URL:
Whiteboard:
Keywords: haveBacktrace
Depends on:
Blocks: UI-Theming Skia 164970
  Show dependency tree / graph
 
Reported: 2025-08-17 17:02 UTC by V Stuart Foote
Modified: 2025-08-18 06:33 UTC (History)
2 users (show)

See Also:
Crash report or crash signature:

Attachments
WinDbg stack trace of pointer crash at exit destructor with skia Vulkan on nVidia GPU (15.46 KB, text/plain)
2025-08-17 17:02 UTC, V Stuart Foote 		Details
View All Add an attachment (proposed patch, testcase, etc.)

Note You need to log in before you can comment on or make changes to this bug.
Description V Stuart Foote 2025-08-17 17:02:35 UTC 
Created attachment 202366 [details]
WinDbg stack trace of pointer crash at exit destructor with skia Vulkan on nVidia GPU

While testing for bug 167987 "Don't ask for a restart on removal of an unused theme" noticed that LO would not actually restart.

I had skia / Vulkan rendering enabled.

Attached process to WinDbg with symbols, the stack trace of the crash during LO detach for the appearance theme removal is attached--affecting the nVidia Vulkan driver support:

skialo!skwindow::internal::VulkanWindowContext::`dynamic atexit destructor for 'fGlobalShared'

When I disabled Vulkan and went to software raster based rendering no issue.

Stack trace with symbols attached.


Version: 25.8.0.2 (X86_64)
Build ID: 80a8bc2ef75d415a197e282da0ebf917315d5e24
CPU threads: 28; OS: Windows 11 X86_64 (build 26100); UI render: Skia/Vulkan; VCL: win
Locale: en-US (en_US); UI: en-US
Calc: CL threaded
Comment 1 V Stuart Foote 2025-08-17 17:10:45 UTC 
Guess that was more the too late action

skialo!GrVkPrimaryCommandBuffer::forceSync

skia\src\gpu\ganesh\vk\GrVkCommandBuffer.cpp @ 698
Comment 2 Noel Grandin 2025-08-17 19:38:27 UTC 
I am guessing that some window is not being properly destructed before process shutdown, and that is keeping the fGlobalShared data structure around after the point in time when it should have been shut down.

Other than that, no idea