I’m on KDE Plasma 5.21.4 and can confirm the system crashes since I did a update a few weeks ago. dmesg has the same error message:
[ 1067.734921] amdgpu 0000:06:00.0: amdgpu: [gfxhub0] retry page fault (src_id:0 ring:0 vmid:6 pasid:32779, for process Web Content pid 11147 thread firefox:cs0 pid 11584)
[ 1067.734930] amdgpu 0000:06:00.0: amdgpu: in page starting at address 0x0000800109401000 from client 27
[ 1067.734934] amdgpu 0000:06:00.0: amdgpu: VM_L2_PROTECTION_FAULT_STATUS:0x00641051
[ 1067.734937] amdgpu 0000:06:00.0: amdgpu: Faulty UTCL2 client ID: TCP (0x8)
[ 1067.734940] amdgpu 0000:06:00.0: amdgpu: MORE_FAULTS: 0x1
[ 1067.734942] amdgpu 0000:06:00.0: amdgpu: WALKER_ERROR: 0x0
[ 1067.734944] amdgpu 0000:06:00.0: amdgpu: PERMISSION_FAULTS: 0x5
[ 1067.734947] amdgpu 0000:06:00.0: amdgpu: MAPPING_ERROR: 0x0
[ 1067.734949] amdgpu 0000:06:00.0: amdgpu: RW: 0x1
@fkfd is right, this commit should fix this error.
BUT this fix is NOT in the 5.11 release, if you check the tag v5.11 in the repo. It’s only fixed on tag v5.12-rc7 and upwards. So you need to be running at least the experimental Kernel 5.12.rc7 to have this fix running.
I’m running on Kernel 5.12.rc7 since 5 hours and did a bit of load testing with 8 video sources running at the same time and it seems the system crash doesn’t occur anymore, so this is fixed!
For those wondering how to switch kernel:
- Open the Manjaro Kernel manager GUI (windows key → search “Kernel”)
- Click install for the experimental 5.12.rc Kernel
- Reboot your PC and in Grub, choose “Advanced Options for Manjaro” and pick Manjaro 5.12
Once booted, run uname -a to verify that the new Kernel is running.