Bug 471559

Summary:	Wayland plasma session does not get along with egl-wayland 1.1.12
Product:	[I don't know] kde	Reporter:	Max <thesourcehim>
Component:	general	Assignee:	Unassigned bugs mailing-list <unassigned-bugs>
Status:	RESOLVED UPSTREAM
Severity:	normal	CC:	ekurzinger, techsav
Priority:	NOR
Version:	unspecified
Target Milestone:	---
Platform:	Fedora RPMs
OS:	Linux
Latest Commit:		Version Fixed In:

Description Max 2023-06-29 07:18:37 UTC

SUMMARY
After an update of egl-wayland to 1.1.12 native wayland GL applications like glmark2-wayland and glmark2-es2-wayland can not run (using prime render offload, NVIDIA GeForce RTX 3060 laptop, driver 535.54.03), fail with "error 7: importing the supplied dmabufs failed". Reverting back to 1.1.11 solves the issue, but also those apps wotk just fine under gnome wayland session with 1.1.12.
I reported this problem on egl-wayland git and they suspect this is a bug of KDE wayland compositor reporting something wrong. Please see this discussion: https://github.com/NVIDIA/egl-wayland/issues/83


STEPS TO REPRODUCE
1. Try to use prime render offload on nvidia card with egl-wayland 1.1.12 with a native wayland GL app
2. 
3. 

OBSERVED RESULT
Native wayland GL applications fail with "error 7: importing the supplied dmabufs failed"

EXPECTED RESULT
Those apps should work

SOFTWARE/OS VERSIONS
Operating System: Fedora Linux 38
KDE Plasma Version: 5.27.6
KDE Frameworks Version: 5.107.0
Qt Version: 5.15.10
Kernel Version: 6.3.9-200.fc38.x86_64 (64-bit)
Graphics Platform: Wayland
Processors: 16 × Intel® Core™ i7-10870H CPU @ 2.20GHz
Memory: 15.4 ГиБ of RAM
Graphics Processor: Mesa Intel® UHD Graphics
Manufacturer: Dell Inc.
Product Name: Dell G15 5510

ADDITIONAL INFORMATION

Comment 1 Rean 2023-06-30 05:34:47 UTC

I have a few OpenGL games installed that run native Wayland, and I also have MPV when run using render offloading that also stops working under Plasma on EGL 1.1.12, but they all run perfectly fine under Gnome on EGL 1.1.12. I'm on Arch Linux.

Comment 2 Erik Kurzinger 2023-07-12 19:45:32 UTC

I've determined that this is a bug in the NVIDIA EGL library. Kwin is not doing anything wrong, we're just not correctly processing the dma-buf feedback that it sends us.

Comment 3 Max 2023-07-19 08:28:56 UTC

Fixed on the egl-wayland side by recent commits