drm/amdgpu: add UTCL2 RAS poison query for Aldebaran (v2)

Add help functions to query and reset RAS UTCL2 poison status.

v2: implement it on amdgpu side and kfd only calls it.

Signed-off-by: Tao Zhou <[e-mail]>
Reviewed-by: Hawking Zhang <[e-mail]>
Signed-off-by: Alex Deucher <[e-mail]>

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c

···
 	else if (reset)
 		amdgpu_amdkfd_gpu_reset(adev);
 }
+
+bool amdgpu_amdkfd_ras_query_utcl2_poison_status(struct amdgpu_device *adev)
+{
+	if (adev->gfx.ras->query_utcl2_poison_status)
+		return adev->gfx.ras->query_utcl2_poison_status(adev);
+	else
+		return false;
+}

drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h

···
 bool amdgpu_amdkfd_bo_mapped_to_dev(struct amdgpu_device *adev, struct kgd_mem *mem);
 void amdgpu_amdkfd_block_mmu_notifications(void *p);
 int amdgpu_amdkfd_criu_resume(void *p);
+bool amdgpu_amdkfd_ras_query_utcl2_poison_status(struct amdgpu_device *adev);
 
 #if IS_ENABLED(CONFIG_HSA_AMD)
 void amdgpu_amdkfd_gpuvm_init_mem_limits(void);

drivers/gpu/drm/amd/amdgpu/amdgpu_gfx.h

···
 struct amdgpu_gfx_ras {
 	struct amdgpu_ras_block_object  ras_block;
 	void (*enable_watchdog_timer)(struct amdgpu_device *adev);
+	bool (*query_utcl2_poison_status)(struct amdgpu_device *adev);
 };
 
 struct amdgpu_gfx_funcs {

drivers/gpu/drm/amd/amdgpu/gfx_v9_4_2.c

···
 	mutex_unlock(&adev->grbm_idx_mutex);
 }
 
+static bool gfx_v9_4_2_query_uctl2_poison_status(struct amdgpu_device *adev)
+{
+	u32 status = 0;
+	struct amdgpu_vmhub *hub;
+
+	hub = &adev->vmhub[AMDGPU_GFXHUB_0];
+	status = RREG32(hub->vm_l2_pro_fault_status);
+	/* reset page fault status */
+	WREG32_P(hub->vm_l2_pro_fault_cntl, 1, ~1);
+
+	return REG_GET_FIELD(status, VM_L2_PROTECTION_FAULT_STATUS, FED);
+}
+
 struct amdgpu_ras_block_hw_ops  gfx_v9_4_2_ras_ops = {
 		.ras_error_inject = &gfx_v9_4_2_ras_error_inject,
 		.query_ras_error_count = &gfx_v9_4_2_query_ras_error_count,
···
 		.hw_ops = &gfx_v9_4_2_ras_ops,
 	},
 	.enable_watchdog_timer = &gfx_v9_4_2_enable_watchdog_timer,
+	.query_utcl2_poison_status = gfx_v9_4_2_query_uctl2_poison_status,
 };