When a privilege fault occurs in a user queue context, the GPU may be left in a hang state. To handle this gracefully and recover the GPU, schedule a user queue reset via amdgpu_reset_domain_schedule.
Signed-off-by: Jesse Zhang <[email protected]> --- drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c index 6cd16f016c37..4162d30947ed 100644 --- a/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c +++ b/drivers/gpu/drm/amd/amdgpu/gfx_v12_0.c @@ -46,6 +46,7 @@ #include "mes_v12_0.h" #include "mes_userqueue.h" #include "amdgpu_userq_fence.h" +#include "amdgpu_reset.h" #define GFX12_NUM_GFX_RINGS 1 #define GFX12_MEC_HPD_SIZE 2048 @@ -5030,6 +5031,10 @@ static void gfx_v12_0_handle_priv_fault(struct amdgpu_device *adev, BUG(); break; } + } else { + /*TODO: schedule a user queue reset work */ + amdgpu_reset_domain_schedule(adev->reset_domain, + &adev->userq_reset_work); } } -- 2.49.0
