Research On Global EDF Schedulability For DAG Task Model In Multi-Core Platform

Real-time systems are widely used in various fields of social production because of their outstanding real-time property and reliability.With the continuous maturity of the Internet of Things technology and the gradual advancement of Industry 4.0,the development of real-time systems is undergoing qualitative changes.Faced with increasingly complex real-world requirements,real-time systems are confronting high-performance,high-integration real-world requirements while ensuring their high-response and high-reliability characteristics.On the other hand,due to the strict constraints of the application environment on real-time systems’ energy consumption,volume and other factors,the development of modern real-time systems face various pressures.The rapid advancement of multi-processor platforms provides new ideas for the development and design of real-time systems.In order to ensure the real-time property and predictability of real-time systems,real-time scheduling algorithms play a significant role.The global EDF(Earliest Deadline First)scheduling algorithm has attracted the attention of many scholars because of its excellent scheduling performance,and is widely used in industry as well.However,in the current research on global EDF schedulability analysis,a large amount of research work is based on the sequential programming model.With the widespread use of multi-core processors,the programming paradigm is moving to a fine-grained thread-level paradigm.This paper mainly studies the schedulability problem of the global EDF scheduling algorithm when using DAG(Directed Acyclic Graph)task model which supports thread-level parallelism in multi-core hard real-time systems.The main research contents of this paper are as follows.(1)Research the global EDF schedulability problem when ignoring the self-suspension state of threads in the DAG task.This study proposed a global EDF schedulability test algorithm based on the graph structure.By correcting the execution window of the nodes in the task and considering the DAG structure characteristics of the task,a schedulability test algorithm with higher computational accuracy for inter-task interference and intra-task interference was proposed.The experimental results show that the proposed algorithm makes the schedulability test result more in line with the actual situation while ensuring a higher recognition rate.(2)Research the global EDF schedulability problem when explicitly considering the self-suspension state of threads in the DAG task.This study proposed a global EDF schedulability test algorithm that explicitly considers the thread self-suspension state in DAG tasks.By extending the task map structure multiple times,the workload and the suspended load of the task were quantitatively studied respectively when analyzing inter-task interference and intra-task interference,then,a schedulability test algorithm with higher recognition rate was proposed.