Happened again today. Got no such hang since 04.04.2018
>From Debian testing:
Linux kernel: 4.16.5-1
Mesa: 18.0.3-1
xserver-xorg-video-amdgpu: 18.0.1-1
libdrm2: 2.4.91-2
(Nearly) current firmware: PRIME-B350M-A-ASUS-4008.CAP

[ 1085.521781] amdgpu 0000:08:00.0: GPU fault detected: 146 0x0bf8440c
[ 1085.521789] amdgpu 0000:08:00.0:   VM_CONTEXT1_PROTECTION_FAULT_ADDR  
0x0010037F
[ 1085.521792] amdgpu 0000:08:00.0:   VM_CONTEXT1_PROTECTION_FAULT_STATUS
0x0204400C
[ 1085.521798] amdgpu 0000:08:00.0: VM fault (0x0c, vmid 1) at page 1049471,
read from 'TC1' (0x54433100) (68)
[ 1095.659133] [drm:amdgpu_job_timedout [amdgpu]] *ERROR* ring gfx timeout,
last signaled seq=39661, last emitted seq=39663
[ 1095.659146] [drm] IP block:gfx_v8_0 is hung!
[ 1095.659194] [drm] GPU recovery disabled.

Is there anything else I could collect next time or try?