AMD GPU glitches with kernel > 6.11

Hello dear Manjaro friends!

Since upgrading to the 6.11 kernel, I have been experiencing occasional graphical glitches on my all-AMD Lenovo ThinkPad Z16 Gen 1 laptop.

I can’t capture them with something like OBS, so I made a little video with my phone: https://youtu.be/94FUiyGhWzo

The glitches have not bothered me too much so far. But since I tried kernel 6.12, they have risen to an annoying level, even worse with 6.13rc.

The glitches appear in windowed and fullscreen applications (it does not only happen in Firefox, all kinds of applications are affected).

My system:

  • Manjaro Linux rolling x86_64
  • ThinkPad Z16 Gen 1
  • 6.11.0-3-rt7-MANJARO
  • GNOME 47.2
  • Mutter (Wayland)
  • AMD Ryzen 5 PRO 6650H (12) @ 4.56 GHz
  • AMD Radeon 680M [Integrated]
  • 16GB memory
  • 2TB SSD with ext4

Oh, I am also running the latest 0.1.70 firmware, updated with fwupdmgr update.

Are any of you experiencing something similar?

Might be the Real Time Kernel, you could try the standard one?

2 Likes

Why are you using a real time kernel? Try the normal one. If older kernels show a better performance I would use them and wait until your issue is fixed on newer kernels. Even the older LTS kernels are updated often, no reason not to use them if they perform better for your hardware.

2 Likes

I have also tried the standard one, and I have the general feeling that there are fewer glitches, but there are still some. :thinking:
I use the rt kernel because I mainly make music on Linux. Although I have to admit that I have not really noticed any performance gains for my use case with the rt kernel.

try with older kernels, the 6.6, and the 6.1…
try with xorg …

PSR (Panel Self Refresh) Maybe?

I started getting full hangs beginning with 6.11 and ~might~ have stability again on 6.12.1 with amdgpu.dcdebugmask=0x200 boot option on this zen3 system.

I have a longer detailed post on PSR on one of the update announcement threads…

Hi @cscs

Please forgive my ignorance, but when I try adding this flag to my grub config and updating I get an error.

/etc/default/grub: line 66: amdgpu.dcdebugmask=0x200: command not found

Hmm. Did the upgrade do something to my graphics drivers?

If I run glxgears on the display that uses my 7800x3d’s onboard graphics it works, if I run it on the display connected to my 7900 it closes straight away.
I use both since there were power issues in the past with driving two monitors from the 7900. No idea if that’s still the case, but it’s been working well since I changed it.

I guess I’ll try a game on the main display and see if that works.

Cheers,
Westy

Edit:
glxgears closing straight away on both displays now.
The info flag seems to display my main card though, so that’s good:

> glxgears -info
Running synchronized to the vertical refresh.  The framerate should be
approximately the same as the monitor refresh rate.
GL_RENDERER   = AMD Radeon RX 7900 XTX (radeonsi, navi31, LLVM 18.1.8, DRM 3.54, 6.6.63-1-MANJARO)
GL_VERSION    = 4.6 (Compatibility Profile) Mesa 24.2.7-arch1.1
GL_VENDOR     = AMD
GL_EXTENSIONS = *snip*

Edit2: Games are running fine, so that’s great!
But yeah, I’m back on kernel 6.6 for now, since get horrible graphics hanging in 6.11 and 6.12, hence me trying your fix. I guess it’s not a bad thing to be on the LTS kernel, but I believe .11 or .12 have performance updates for AMD processors.

Line 66?

Where/how did you add it?

Grub boot options go in the GRUB_CMDLINE_LINUX line. Example if you have no other options;

GRUB_CMDLINE_LINUX="amdgpu.dcdebugmask=0x200"

Which in my file is line 7.

And afterwards you must run sudo update-grub and reboot.

Yeah, I just discovered the fact it needed to be in the command line (from here: https://askubuntu.com/questions/1437259/how-do-you-set-amdgpu-options), although added it to GRUB_CMDLINE_LINUX_DEFAULT.

Before I just added it at the end, since literally had no idea!
I’ll move it the non-default setting now and give it a go.

Thanks for the help!

Technically either of those works … but I tend to keep the actual system defaults on the DEFAULT one and my own extras on the other.

Please also keep in mind that it is only one of possible variables.
It will only disable the newest PSR_SU (or PSR 2.0), but as shown at the previous link, there are various combinations to disable all of PSR and/or other related options.

That is even if PSR is the culprit - this option may not suffice.

With PSR being a power-saving feature we will generally want to only disable as much of it as is necessary.

Well, I’m back on 6.11 and the graphics haven’t crashed on the main display yet, so looking promising!

Is there a way of verifying the flag is set?
Will there be a time in the future where it is no longer needed? How will be know?

Yeah, I saw that there are a lot of options in your linked post. I’ll see how I get on with this one I guess. I have a 7900 GTX, if that helps determine what I might need.

Cheers

You should be able to see it from

cat /proc/cmdline

We can also observe them at the top of a robust enough inxi, ex

inxi -Farz
1 Like

Yeah, I see it in both.

Excellent. Thank you so much for your help!

1 Like

This topic was automatically closed 3 days after the last reply. New replies are no longer allowed.