After a retry-fault happens, the fault handler will modify the UTCL2 to set PTE bits to force a no-retry-fault. This will cause the wave to enter the trap handler.
Change-Id: I177102891f715068f15605957ff23b0cab862603 Signed-off-by: Alex Sierra <[email protected]> --- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 13 +++++-------- 1 file changed, 5 insertions(+), 8 deletions(-) diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 3c0bd6472a46..e4d1a8fc97d5 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -3167,7 +3167,8 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, unsigned int pasid, uint64_t addr) { struct amdgpu_bo *root; - uint64_t value, flags; + uint64_t value = 0; + uint64_t flags; struct amdgpu_vm *vm; long r; @@ -3200,13 +3201,9 @@ bool amdgpu_vm_handle_fault(struct amdgpu_device *adev, unsigned int pasid, AMDGPU_PTE_SYSTEM; if (amdgpu_vm_fault_stop == AMDGPU_VM_FAULT_STOP_NEVER) { - /* Redirect the access to the dummy page */ - value = adev->dummy_page_addr; - flags |= AMDGPU_PTE_EXECUTABLE | AMDGPU_PTE_READABLE | - AMDGPU_PTE_WRITEABLE; - } else { - /* Let the hw retry silently on the PTE */ - value = 0; + /* Setting PTE flags to trigger a no-retry-fault */ + flags = AMDGPU_PTE_EXECUTABLE | AMDGPU_PDE_PTE | + AMDGPU_PTE_TF; } r = amdgpu_vm_bo_update_mapping(adev, vm, true, NULL, addr, addr + 1, -- 2.17.1 _______________________________________________ amd-gfx mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/amd-gfx
