Response-Time Analysis for Limited-Preemptive Self-Suspending and Event-Driven Delay-Induced Tasks

Heterogeneous computing platforms running highly parallelized applications are becoming increasingly common in real-time embedded systems. This demands for expressive task models that can capture parallelism, precedence constraints and self-suspending behavior caused by work-offloading on coprocessors, access to shared resources or synchronization between tasks running on different (heterogeneous) cores. The Event-Driven Delay-induced (EDD) task model was specifically designed to address these needs. Yet, to date, a single schedulability test for the EDD task model exists, and that test is limited to the analysis of fully-preemptive partitioned scheduling.In this work, we provide the first worst-case response time analysis for limited-preemptive EDD tasks that are globally scheduled on a multicore platform. Moreover, our evaluation results show that our analysis also outperforms the state-of-the-art response time analyses for both limited-preemptive DAG tasks and self-suspending tasks.