Comment 8 for bug 2068738

Revision history for this message
Lin Manfu (linmanfu) wrote (last edit ): Re: Kernel update 5.15.0-112 might cause severe problems with specific AMD GPUs

Attached is an annotated copy of my system log with the relevant update from 107 to 112, the failed boots with 112, a successful boot with the last good kernel 107, and a boot with 112 using the workaround. Search for BOOT in capital letters to find the start of each boot. I hope this might be helpful.

I am no expert, but perhaps these lines might be relevant, as they only seem to occur on the failed boots with 112:

Jun 8 19:17:56 paul kernel: [ 3.190269] amdgpu: Topology: Add APU node [0x15d8:0x1002]
Jun 8 19:17:56 paul kernel: [ 3.190273] kfd kfd: amdgpu: added device 1002:15d8
Jun 8 19:17:56 paul kernel: [ 3.190280] kfd kfd: amdgpu: Failed to resume IOMMU for device 1002:15d8
Jun 8 19:17:56 paul kernel: [ 3.190287] amdgpu 0000:27:00.0: amdgpu: amdgpu_device_ip_init failed
Jun 8 19:17:56 paul kernel: [ 3.190292] amdgpu 0000:27:00.0: amdgpu: Fatal error during GPU init
Jun 8 19:17:56 paul kernel: [ 3.190296] amdgpu 0000:27:00.0: amdgpu: amdgpu: finishing device.