drm/amdgpu: Implement virt req_ras_err_count

Enable RAS late init  if VF RAS Telemetry is supported.

When enabled, the VF can use this interface to query total
RAS error counts from the host.

The VF FB access may abruptly end due to a fatal error,
therefore the VF must cache and sanitize the input.

The Host allows 15 Telemetry messages every 60 seconds, afterwhich
the host will ignore any more in-coming telemetry messages. The VF will
rate limit its msg calling to once every 5 seconds (12 times in 60 seconds).
While the VF is rate limited, it will continue to report the last
good cached data.

v2: Flip generate report & update statistics order for VF

Signed-off-by: Victor Skvortsov <victor.skvortsov@amd.com>
Acked-by: Tao Zhou <tao.zhou1@amd.com>
Reviewed-by: Zhigang Luo <zhigang.luo@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

1 files changed, 65 insertions, 7 deletions

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
index ffc4a1ced672..1bc95b0cdbb8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ras.c
@@ -1214,6 +1214,42 @@ static void amdgpu_ras_error_generate_report(struct amdgpu_device *adev,
 	}
 }
 
+static void amdgpu_ras_virt_error_generate_report(struct amdgpu_device *adev,
+						  struct ras_query_if *query_if,
+						  struct ras_err_data *err_data,
+						  struct ras_query_context *qctx)
+{
+	unsigned long new_ue, new_ce, new_de;
+	struct ras_manager *obj = amdgpu_ras_find_obj(adev, &query_if->head);
+	const char *blk_name = get_ras_block_str(&query_if->head);
+	u64 event_id = qctx->evid.event_id;
+
+	new_ce = err_data->ce_count - obj->err_data.ce_count;
+	new_ue = err_data->ue_count - obj->err_data.ue_count;
+	new_de = err_data->de_count - obj->err_data.de_count;
+
+	if (new_ce) {
+		RAS_EVENT_LOG(adev, event_id, "%lu correctable hardware errors "
+			      "detected in %s block\n",
+			      new_ce,
+			      blk_name);
+	}
+
+	if (new_ue) {
+		RAS_EVENT_LOG(adev, event_id, "%lu uncorrectable hardware errors "
+			      "detected in %s block\n",
+			      new_ue,
+			      blk_name);
+	}
+
+	if (new_de) {
+		RAS_EVENT_LOG(adev, event_id, "%lu deferred hardware errors "
+			      "detected in %s block\n",
+			      new_de,
+			      blk_name);
+	}
+}
+
 static void amdgpu_rasmgr_error_data_statistic_update(struct ras_manager *obj, struct ras_err_data *err_data)
 {
 	struct ras_err_node *err_node;
@@ -1237,6 +1273,15 @@ static void amdgpu_rasmgr_error_data_statistic_update(struct ras_manager *obj, s
 	}
 }
 
+static void amdgpu_ras_mgr_virt_error_data_statistics_update(struct ras_manager *obj,
+							     struct ras_err_data *err_data)
+{
+	/* Host reports absolute counts */
+	obj->err_data.ue_count = err_data->ue_count;
+	obj->err_data.ce_count = err_data->ce_count;
+	obj->err_data.de_count = err_data->de_count;
+}
+
 static struct ras_manager *get_ras_manager(struct amdgpu_device *adev, enum amdgpu_ras_block blk)
 {
 	struct ras_common_if head;
@@ -1323,7 +1368,9 @@ static int amdgpu_ras_query_error_status_helper(struct amdgpu_device *adev,
 	if (error_query_mode == AMDGPU_RAS_INVALID_ERROR_QUERY)
 		return -EINVAL;
 
-	if (error_query_mode == AMDGPU_RAS_DIRECT_ERROR_QUERY) {
+	if (error_query_mode == AMDGPU_RAS_VIRT_ERROR_COUNT_QUERY) {
+		return amdgpu_virt_req_ras_err_count(adev, blk, err_data);
+	} else if (error_query_mode == AMDGPU_RAS_DIRECT_ERROR_QUERY) {
 		if (info->head.block == AMDGPU_RAS_BLOCK__UMC) {
 			amdgpu_ras_get_ecc_info(adev, err_data);
 		} else {
@@ -1405,14 +1452,22 @@ static int amdgpu_ras_query_error_status_with_event(struct amdgpu_device *adev,
 	if (ret)
 		goto out_fini_err_data;
 
-	amdgpu_rasmgr_error_data_statistic_update(obj, &err_data);
+	if (error_query_mode != AMDGPU_RAS_VIRT_ERROR_COUNT_QUERY) {
+		amdgpu_rasmgr_error_data_statistic_update(obj, &err_data);
+		amdgpu_ras_error_generate_report(adev, info, &err_data, &qctx);
+	} else {
+		/* Host provides absolute error counts. First generate the report
+		 * using the previous VF internal count against new host count.
+		 * Then Update VF internal count.
+		 */
+		amdgpu_ras_virt_error_generate_report(adev, info, &err_data, &qctx);
+		amdgpu_ras_mgr_virt_error_data_statistics_update(obj, &err_data);
+	}
 
 	info->ue_count = obj->err_data.ue_count;
 	info->ce_count = obj->err_data.ce_count;
 	info->de_count = obj->err_data.de_count;
 
-	amdgpu_ras_error_generate_report(adev, info, &err_data, &qctx);
-
 out_fini_err_data:
 	amdgpu_ras_error_data_fini(&err_data);
 
@@ -3930,7 +3985,7 @@ int amdgpu_ras_late_init(struct amdgpu_device *adev)
 	}
 
 	/* Guest side doesn't need init ras feature */
-	if (amdgpu_sriov_vf(adev))
+	if (amdgpu_sriov_vf(adev) && !amdgpu_sriov_ras_telemetry_en(adev))
 		return 0;
 
 	list_for_each_entry_safe(node, tmp, &adev->ras_list, node) {
@@ -4397,11 +4452,14 @@ bool amdgpu_ras_get_error_query_mode(struct amdgpu_device *adev,
 		return false;
 	}
 
-	if ((smu_funcs && smu_funcs->set_debug_mode) || (mca_funcs && mca_funcs->mca_set_debug_mode))
+	if (amdgpu_sriov_vf(adev)) {
+		*error_query_mode = AMDGPU_RAS_VIRT_ERROR_COUNT_QUERY;
+	} else if ((smu_funcs && smu_funcs->set_debug_mode) || (mca_funcs && mca_funcs->mca_set_debug_mode)) {
 		*error_query_mode =
 			(con->is_aca_debug_mode) ? AMDGPU_RAS_DIRECT_ERROR_QUERY : AMDGPU_RAS_FIRMWARE_ERROR_QUERY;
-	else
+	} else {
 		*error_query_mode = AMDGPU_RAS_DIRECT_ERROR_QUERY;
+	}
 
 	return true;
 }