Mageia forum

Hallo,

after the latest Mesa upgrade today I have problems with hardware acceleration:

Code: Select all

glxgears
_amdgpu_device_initialize: amdgpu_query_gpu_info_init failed
amdgpu: amdgpu_device_initialize failed.
glx: failed to create dri3 screen
failed to load driver: radeonsi
_amdgpu_device_initialize: amdgpu_query_gpu_info_init failed
amdgpu: amdgpu_device_initialize failed.
glx: failed to create dri2 screen
failed to load driver: radeonsi
_amdgpu_device_initialize: amdgpu_query_gpu_info_init failed
radv/amdgpu: failed to initialize device.
_amdgpu_device_initialize: amdgpu_query_gpu_info_init failed
radv/amdgpu: failed to initialize device.
_amdgpu_device_initialize: amdgpu_query_gpu_info_init failed
amdgpu: amdgpu_device_initialize failed.
glx: failed to create dri3 screen
failed to load driver: radeonsi
_amdgpu_device_initialize: amdgpu_query_gpu_info_init failed
amdgpu: amdgpu_device_initialize failed.
glx: failed to create dri2 screen
failed to load driver: radeonsi
X connection to :0 broken (explicit kill or server shutdown).


It seems that the driver radeonsi will not be loaded anymore.
How can I try to fix this problem?

I solved the problem:

1. Downloaded lib64dri-drivers-24.2.8-1.mga10.tainted.x86_64.rpm
2. urpmi /home/denis/lib64dri-drivers-24.2.8-1.mga10.tainted.x86_64.rpm --force --replacefiles --allow-nodeps --replacepkgs

Good, and thank you for the heads up.

You could have checked updates testing first, where there are 24.2.8-1.mga9
No bug opened to officially test it yet.

EDIT: Hmm we released mesa 24.2.7-1.mga9 to updates 2024-11-22. Is this the version you first updated to today?

According the forum category and the package name, this is about Cauldron...Cauldron does not use testing repository.

OOPS yes, I did not notice this is Cauldron forum.
Denis, can you confirm you run Cauldron?

lib64dri-drivers-24.2.8-1.mga10.tainted.x86_64.rpm is in /mageia/distrib/cauldron/x86_64/media/tainted/release/ since 241129.
So that package needed to forced for some reason.
Did that package get installed (but somehow failed) or did you have an elder version still installed before you forced this one?

I can confirm, I'm running Cauldron.

My problems are not fixed at all. I'm not able to run freerdp, xfreerdp or remmina with rdp plugin anymore.

Code: Select all

(org.remmina.Remmina:8468): LIBDBUSMENU-GLIB-WARNING **: 11:30:27.702: Unable to get session bus: The connection is closed
[11:30:32:786] [8468:00002145] [WARN][com.freerdp.settings] - [freerdp_settings_client_monitors_check_primary_and_origin]: Monitor configuration empty.
[11:30:32:851] [8468:00002145] [INFO][com.freerdp.primitives] - [primitives_init_opencl_context]: openCL: using platform=Clover device=AMD Radeon Graphics (radeonsi, renoir, LLVM 19.1.3, DRM 3.54, 6.6.62-desktop-1.mga10)
amdgpu: The CS has cancelled because the context is lost. This context is innocent.
Abgebrochen (Speicherabzug geschrieben)


Something is definitely wrong with with amdgpu and openCL. The problem started after last Mesa updates.

Well you downgraded only one package of the whole mesa stack, and the rest of the mesa stack is newer, what do you expect ?

after downgrading of the whole Mesa stack to mag9 everything is working well again..

Might be time for a bug report.

https://wiki.mageia.org/en/How_to_report_a_bug_properly

https://bugs.mageia.org/

I have also thought about a bug report. But my problem is that I can only say something is wrong with mesa.
Unfortunately I'm not able to describe the problem more precisely...

edit:
during the mean time I filed a bug report: https://bugs.mageia.org/show_bug.cgi?id=33833

Well, you have the error messages from glxgears, and in the other case the segfault from freerdp, that should be sufficient