OpenCL program causes steady stream of GPU fault detected errors

Submitted by Jolan Luff

Assigned to Default DRI bug account

Link to original bug (#96906)

Description

Hi,

I have an OpenCL program which causes a steady stream of errors:

kernel: amdgpu 0000:01:00.0: GPU fault detected: 147 0x03f84801
kernel: VM fault (0x01, vmid 6) at page 135290494, read from 'TC7' (0x54433700) (132)
kernel: amdgpu 0000:05:00.0: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x0C084001
kernel: amdgpu 0000:05:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x08105E7E
kernel: amdgpu 0000:05:00.0: GPU fault detected: 147 0x03f08401
kernel: VM fault (0x01, vmid 3) at page 135290494, read from 'TC2' (0x54433200) (200)
kernel: amdgpu 0000:04:00.0: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x060C8001
kernel: amdgpu 0000:04:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x08105E7E
kernel: amdgpu 0000:04:00.0: GPU fault detected: 147 0x03f0c801
kernel: VM fault (0x01, vmid 6) at page 135290495, read from 'TC1' (0x54433100) (4)
kernel: amdgpu 0000:05:00.0: VM_CONTEXT1_PROTECTION_FAULT_STATUS 0x0C004001
kernel: amdgpu 0000:05:00.0: VM_CONTEXT1_PROTECTION_FAULT_ADDR 0x08105E7F

AMDGPU-Pro: 16.30.3.306809
Kernel: 4.7.0-rc6-mainline
GPU: RX480