drm/amdgpu: Conditionally reset RAS counters on boot

Only clear RAS error counters if perestent EDC harvesting is not supported
Reviewed-by: Hawking Zhang <Hawking.Zhang@amd.com>
Signed-off-by: John Clements <john.clements@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

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

@@ -607,7 +607,6 @@ int amdgpu_gfx_ras_late_init(struct amdgpu_device *adev)
 	struct ras_ih_if ih_info = {
 		.cb = amdgpu_gfx_process_ras_data_cb,
 	};
-	struct ras_query_if info = { 0 };
 
 	if (!adev->gfx.ras_if) {
 		adev->gfx.ras_if = kmalloc(sizeof(struct ras_common_if), GFP_KERNEL);
@@ -625,12 +624,8 @@ int amdgpu_gfx_ras_late_init(struct amdgpu_device *adev)
 		goto free;
 
 	if (amdgpu_ras_is_supported(adev, adev->gfx.ras_if->block)) {
-		if (adev->gmc.xgmi.connected_to_cpu) {
-			info.head = *adev->gfx.ras_if;
-			amdgpu_ras_query_error_status(adev, &info);
-		} else {
+		if (!amdgpu_persistent_edc_harvesting_supported(adev))
 			amdgpu_ras_reset_error_status(adev, AMDGPU_RAS_BLOCK__GFX);
-		}
 
 		r = amdgpu_irq_get(adev, &adev->gfx.cp_ecc_error_irq, 0);
 		if (r)

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

@@ -2194,7 +2194,7 @@ int amdgpu_ras_init(struct amdgpu_device *adev)
 	return r;
 }
 
-static int amdgpu_persistent_edc_harvesting_supported(struct amdgpu_device *adev)
+int amdgpu_persistent_edc_harvesting_supported(struct amdgpu_device *adev)
 {
 	if (adev->gmc.xgmi.connected_to_cpu)
 		return 1;

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

@@ -625,4 +625,7 @@ void amdgpu_ras_set_error_query_ready(struct amdgpu_device *adev, bool ready);
 bool amdgpu_ras_need_emergency_restart(struct amdgpu_device *adev);
 
 void amdgpu_release_ras_context(struct amdgpu_device *adev);
+
+int amdgpu_persistent_edc_harvesting_supported(struct amdgpu_device *adev);
+
 #endif

drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c

@@ -1268,6 +1268,7 @@ static int gmc_v9_0_late_init(void *handle)
 		}
 	}
 
+	if (!amdgpu_persistent_edc_harvesting_supported(adev)) {
 		if (adev->mmhub.ras_funcs &&
 		    adev->mmhub.ras_funcs->reset_ras_error_count)
 			adev->mmhub.ras_funcs->reset_ras_error_count(adev);
@@ -1275,6 +1276,7 @@ static int gmc_v9_0_late_init(void *handle)
 		if (adev->hdp.ras_funcs &&
 		    adev->hdp.ras_funcs->reset_ras_error_count)
 			adev->hdp.ras_funcs->reset_ras_error_count(adev);
+	}
 
 	r = amdgpu_gmc_ras_late_init(adev);
 	if (r)