Multicore processors are increasingly used in real-time embedded systems. Better utilization of hard real-time systems requires accurate scheduling and synchronization analysis. In this paper, we characterize the major synchronization penalties arising from partitioned fixed priority scheduling for hard real-time tasks on multicore platform, including transitive remote preemption, multiple remote blocking, and multiple priority inversions. Subsequently, we propose a new response time analysis by improving the approach to bound task blocking time. The key idea of this approach is to classify the total blocking time into (i) direct blocking, including local and remote blocking, and transitive remote preemption; and (ii) multiple local interference which is incurred by multiple priority inversion. Simulation results indicate that the proposed approach produces less pessimistic results in task blocking time, and better schedulability performance.
Timing analysis of mixed-criticality hard real-time applications implemented on distributed partitioned architectures