drm/amd/pm: correct the gpu metrics version

For V1_0 and V1_1, they come with different size. Misuse may cause
out of memory access.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Reviewed-by: Kevin Wang <kevin1.wang@amd.com>
Reviewed-by: Lijo Lazar <lijo.lazar@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>

drivers/gpu/drm/amd/pm/swsmu/smu13/aldebaran_ppt.c

@@ -206,7 +206,7 @@ static int aldebaran_tables_init(struct smu_context *smu)
 		return -ENOMEM;
 	smu_table->metrics_time = 0;
 
-	smu_table->gpu_metrics_table_size = sizeof(struct gpu_metrics_v1_0);
+	smu_table->gpu_metrics_table_size = sizeof(struct gpu_metrics_v1_1);
 	smu_table->gpu_metrics_table = kzalloc(smu_table->gpu_metrics_table_size, GFP_KERNEL);
 	if (!smu_table->gpu_metrics_table) {
 		kfree(smu_table->metrics_table);