There have recently been issued some patches for various race conditions in the amdgpu driver to fix the frequent ‘page flip timed out’ errors/screen freezes like
and I’m asking myself how I can find out when those arrive in 6.12.
I’m not even sure which kernel they have been added to (6.16/6.17rc) as of right now. I don’t want to bother the amdgpu driver folks in their gitlab with such a beginner question and thought there must be some kind of procedure/website/changelog I could check.
That is a difficult one …
According to my understand - and I may be wrong …
Changes - fixes - that kind of stuff - may not be back ported at all; it depends on the source of the targeted kernel if it can be incorporated in automated way.
All that happens at the upstream repository on kernel.org.
I know that @philm patches the kernel to work with various hardware projects and that may include fixing issues in early stages - but I have no idea what motivates such patches.
For such information you would have to keep an eye on core · GitLab
Thanks for mentioning (despite the small typo in the URL 
). I think I need to check the changelog for longterm: 6.12.XX on kernel.org to confirm if an amdgpu patch has been added to the stable kernel.
One of the AMD developers (Mario Limonciello) mentioned in one of the Gitlab issues, that anyone can nominate a patch for stable:
If you’re not familiar with the rules for the stable kernel here they are:
Everything you ever wanted to know about Linux -stable releases — The Linux Kernel documentationBut at a high level - anyone can nominate a patch for stable as long as it follows those guidelines. If you’re sure that these patches solve it, please feel free to nominate!
So that’s where I’m coming from and what lets me hope those VRR and ‘race condition’ fixes end up in the stable kernel. According to a bunch of comments in the AMD Gitlab they do indeed fix a lot of the crashes.
I did the [Stable Update] 2025-08-11 yesterday and can confirm that with updated Manjaro Zetar 25.0.7, amdgpu firmware 20250808-1, Mesa 25.1.7-1, 6.12.41-1 kernel, … I still get my regular screen freezes and ‘flip_done’ timed out errors in the journal. So no fix yet 
and therefore I’m hoping those patches upstream will ‘trickle down’ eventually to fix the issue.
Did you try the kernel parameter amdgpu.dcdebugmask=0x10?
I didn’t. It disables PSR (Panel Self Refresh) DC_DISABLE_PSR which increases power draw. There are also multiple reports [1], [2], [3] … in AMD’s gitlab that it doesn’t work.
I only have one freeze a day so I didn’t think it’s worth the extra cost atm.
On the other hand I’ve got a normal desktop system and not a laptop. It’s just that I don’t know how much more power draw we’re talking about. The (family) PC is running most of the time and power is expensive over here in germany. If we’re talking about dozens of additional watts (absolutely no idea if this is reasonable at all), then it would get costly over time. That’s why I shied away from disabling PSR 'til now.
Does anyone know how much extra watts it’ll take on average when disabling PSR?