Towards collaborative resource sharing under real-time conditions in multitasking and multicore environments

Today's embedded system designs demand for an ever increasing integration density of various services on common platforms. Especially within highly dynamic environments where severe real-time demands must be met, sharing exclusive resources among these concurrently running subsystems is a hard compositional problem. Classic approaches suffer from the fact that tasks or cores are not aware about their mutual influences and existing resource conflicts, and thus cannot collaborate efficiently in critical situations. We eliminate this flaw, and present the novel DynamicHinting method for sharing exclusive resources on-demand among prioritized tasks on both common and different cores: Hints will be issued by the resource manager(s) to indicate e.g. priority inversions, and to provide blocking tasks with the time, knowledge, and CPU power to resolve the conflicts in time.