Bug 32150 - Display not working with kernel 5.15.122 (and 120) -desktop and Radeon-R5 Graphics (x86_64)
Summary: Display not working with kernel 5.15.122 (and 120) -desktop and Radeon-R5 Gra...
Status: NEW
Alias: None
Product: Mageia
Classification: Unclassified
Component: RPM Packages (show other bugs)
Version: 8
Hardware: x86_64 Linux
Priority: Normal critical
Target Milestone: ---
Assignee: Kernel and Drivers maintainers
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2023-07-31 21:50 CEST by Jean-Yves LE GOFF
Modified: 2023-08-05 21:21 CEST (History)
1 user (show)

See Also:
Source RPM: kernel-desktop-latest-5.15.122-1.mga8.x86_64.rpm
CVE:
Status comment:


Attachments
journal file (17.72 KB, application/x-xz)
2023-08-02 21:34 CEST, Jean-Yves LE GOFF
Details
dmesg copied via another machine (75.62 KB, text/plain)
2023-08-03 21:36 CEST, Jean-Yves LE GOFF
Details

Description Jean-Yves LE GOFF 2023-07-31 21:50:24 CEST
Description of problem:
With kernel 5.15.122, display is not working as soon as login screen occurs.

Version-Release number of selected component (if applicable):
kernel-desktop-5.15.122-1.mga8.x86_64.rpm

How reproducible:
Same occurs with kernel 5.15.120

Steps to Reproduce:
1.launch mageia 8 x86_64 with kernel 5.15.122 
2.
3.

Works fine with kernel 5.15.117

Hardware concerned: Acer Aspire 3 (A315-21-96U1) - AMD A9-9225 - Radeon R5
Comment 1 Lewis Smith 2023-08-02 20:53:49 CEST
Sorry for your problem.

To summarise the graphics, please post the output from (if you have inxi):
 $ inxi -G

Can you please attach to this bug a compressed system journal done from a virtual console (Ctrl/Alt/Fn) as soon as the problem shows:
 $ journalctl --no-hostname -b > journal.txt
 $ xz journal.txt
After the second command, you should have journal.txt.xz, this is the file to attach.

CC: (none) => lewyssmith

Comment 2 Jean-Yves LE GOFF 2023-08-02 21:34:03 CEST
Created attachment 13930 [details]
journal file

with 5.15.122 kernel, display is not working.
Comment 3 Jean-Yves LE GOFF 2023-08-02 21:37:05 CEST
Here is the result of the inxi -G command in a working session :

Graphics:  Device-1: AMD Stoney [Radeon R2/R3/R4/R5 Graphics] driver: amdgpu v: kernel 
           Device-2: Quanta VGA WebCam type: USB driver: uvcvideo 
           Display: x11 server: Mageia X.org 1.20.14 driver: amdgpu note: n/a (using device driver) 
           resolution: 1366x768~60Hz 
           OpenGL: renderer: AMD STONEY (DRM 3.42.0 5.15.117-desktop-2.mga8 LLVM 11.0.1) 
           v: 4.5 Mesa 21.3.8 


Note: as soon as the problem occurs, virtual consoles are unavailable and the display is not working correctly (moving lines as if the screen was an analogical screen )
Comment 4 Thomas Backlund 2023-08-03 20:01:28 CEST
from log:
Command line: BOOT_IMAGE=/boot/vmlinuz-5.15.122-desktop-1.mga8 root=UUID=9e71225e-c6d4-439b-8cb0-d55473b84b71 ro failsafe

you are booting in failsafe mode
Comment 5 Jean-Yves LE GOFF 2023-08-03 20:16:47 CEST
this is the only way to access to the CLI
If I boot in normal mode, the screen does not function well and I can't see anything.
So I did this: 
1- start with 5.15.122 -> malfunction -> power off
2- boot with failsafe mode -> save jounal

The last working kernel (on this hardware) is 5.15.117.
If I boot with 5.15.117 it's working. It does not with 5.15.120 nor 5.15.122
Comment 6 Thomas Backlund 2023-08-03 20:50:20 CEST
yeah, but booting in failsafe can hide the bug too...

do you have another computer ?
if so, try to ssh into the problematic system and grab the logs / dmesg output
Dave Hodgins 2023-08-03 20:51:13 CEST

CC: (none) => davidwhodgins

Comment 7 Jean-Yves LE GOFF 2023-08-03 21:36:42 CEST
Created attachment 13932 [details]
dmesg copied via another machine

Ok, I found an raspberry Pi and grab the dmesg output while having the screen flickering
Comment 8 Thomas Backlund 2023-08-03 21:44:55 CEST
Ah , you are probably triggering the AMD Stoney bug... it was supposed to be fixad....
Comment 9 Jean-Yves LE GOFF 2023-08-03 21:55:59 CEST
Is there a workaround ?
it scares me a bit before doing the upcoming mageia 9 upgrade
The system worked well until kernel 5.15.117. 
Now I'm stuck with kernel 5.15.117
Comment 10 Thomas Backlund 2023-08-03 22:12:23 CEST
its already fixed in 6.x series kernels, 
but I suspect the problematic patches has been backported upstream to 5.15 series kernels but not the follow-up fixes... 

I'll dig through the 5.15 branch to see...
Comment 11 Jean-Yves LE GOFF 2023-08-03 22:27:15 CEST
In the mageia history of the  5.15.122 package
after the 5.15.117, there is : 

- Revert "drm/amd/display: edp do not add non-edid timings"

could this is the cause ? I suspect a frequency problem in the display
Comment 12 Thomas Backlund 2023-08-03 22:31:27 CEST

my suspicion was correct, the broken patchset landed in upstream 5.15.119 :(

sorry for missing that :/ 

I'll fix it up...
Comment 13 Jean-Yves LE GOFF 2023-08-03 22:44:44 CEST
Thank you  !
Comment 14 Thomas Backlund 2023-08-03 23:18:08 CEST
I've submitted a kernel-5.15.124-2.mga8 to Mga8 Core Updates Testing that should fix this issue.

it will take several hours before it's built and uploaded to the mirrors
Comment 15 Lewis Smith 2023-08-05 21:21:10 CEST
Thank you Thomas for investigating this, and Jean-Yves for your helpful co-operation.

CC: lewyssmith => (none)
Assignee: bugsquad => kernel


Note You need to log in before you can comment on or make changes to this bug.