Full system crash when hardware video decoding on AMD

Hi! I’ve got a Framework 13 AMD Ryzen 7640U with Radeon 760M Graphics. I’m on Linux Mint Cinnamon v22.3 with Kernel v7.0.0-28 and BIOS v3.20.

Note: I originally wrote this post thinking this was related to web browsers, but I did some more testing and found out more, and have fully updated this post.

When attempting to use hardware decoding specifically for video, the display turns off and the system becomes unresponsive. Strangely sometimes the Caps Lock key light can still be toggled by pressing the key, but regardless, no key combinations seem to work, I’ve tried variations of CTRL+ALT+F1-12 and ALT+F1-12 with no luck.

This happens in both web browsers and MPV with hardware video decoding enabled. There is no issue using software video decoding, and there is no issue decoding videos on my Windows 11 install with hardware decoding.

I was actually having this exact issue on Kubuntu as well before I distro-hopped to Linux Mint. This issue also happens with Linux Mint on kernel versions 6.17.0-40 and 6.14.0-37.

I checked the Journal for errors and consistently found the following:

amdgpu 0000:c1:00.0: amdgpu: ring vcn_unified_0 timeout, signaled seq=32, emitted seq=34
amdgpu 0000:c1:00.0: amdgpu:  Process RDD Process pid 2317 thread browser 4 :cs0 pid 2803
amdgpu 0000:c1:00.0: amdgpu: Starting vcn_unified_0 ring reset
amdgpu 0000:c1:00.0: amdgpu: Ring vcn_unified_0 reset failed
amdgpu 0000:c1:00.0: [drm:amdgpu_ring_test_helper [amdgpu]] *ERROR* ring vcn_unified_0 test failed (-110)
amdgpu 0000:c1:00.0: amdgpu: resume of IP block <vcn_v4_0> failed -110
amdgpu 0000:c1:00.0: amdgpu: GPU Recovery Failed: -110
amdgpu 0000:c1:00.0: amdgpu: ring vcn_unified_0 timeout, signaled seq=34, emitted seq=34
amdgpu 0000:c1:00.0: amdgpu:  Process RDD Process pid 2317 thread browser 4 :cs0 pid 2803
amdgpu 0000:c1:00.0: amdgpu: Starting vcn_unified_0 ring reset
amdgpu 0000:c1:00.0: amdgpu: Ring vcn_unified_0 reset failed

While I’ve found various posts from the past few years of users experiencing various video-related issues on Linux, I haven’t been able to find any with an error regarding VCN. I don’t really know where to go from here, so any help would be much appreciated, thank you.

I did some more testing and while I originally thought this was an issue exclusive to web browsers, I found that it’s hardware video decoding as a whole. I’ve since updated my post to reflect my findings. Any help would still be really appreciated, this issue makes using my Framework Laptop really frustrating.

I had nothing but issues with 7.0.0-28, even on a VM guest machine (sometimes hangs on boot). I hadn’t run into your specific issue, but I didn’t run that kernel very long, so maybe it would’ve happened to me too, if I kept using it.

I reverted to 6.17.0-40 and haven’t had any issues. I’m just going to skip 7.0.0-28.

If you didn’t remove the older kernel, you can just boot into it (SHIFT while booting to pick older kernel), and then once booted, run Update Manager and remove the 7.0.0-28. It’ll show up again in the update list. I right clicked on it and made it ignore that specific update.

If you already deleted the older kernel, you can re-install it from Update Manager, which is what I had to do on a couple of machines I have.

Hey, thanks for your reply. Unfortunately I’ve already tried both Kernels v6.17.0-40 and 6.14.0-37 and the issue still persists. I downloaded and installed both and selected them in GRUB before booting, and I made sure to check the system information to verify I actually booted with a different kernel. I’m honestly quite surprised that it didn’t solve my issue.

I’d also like to update and say that I tried CachyOS on a Live USB and it didn’t have any issues, even with hardware decoding with VAAPI through MPV.

1 Like

Kernel 7.1 fixed a bunch of amd gpu glitches. Cachy is already on that version. Could be a while before Mint gets it unless you set up a ppa.

Hi, thank you for your reply. I just tried Kernel 7.1.5 through Mainline and unfortunately the issue still persists. Is there another piece of software that could be the cause?

The question has to be ‘what else changed?’

I’ve got a history of querying the AMD firmware package because it’s a hidden portion of Linux’s amdgpu in-kernel drivers. What edition of linux-firmware was in the CachyOS live USB and what’s in your Linux Mint (Cinnamon) install?

2 Likes

Hmm interesting. On Linux Mint it’s 20240318.git3b128b60-0ubuntu2.27 and on CachyOS it’s 1:20260622-1? I used apt policy and pacman -Qi respectively.

Thank you @Kenny_Lewis! You sent me on the right track. I tried a desperate hail mary approach: I downloaded the latest version of the amdgpu files from the linux-firmware repository and simply swapped them with the ones on my install and it fixed the issue! I had no idea about linux-firmware, thank you so much!