| Abstract |
In general computing systems, a job (process/task) may suspend itself whilst it is waiting for some activity to complete, \eg, an accelerator to return required data or results from the offloaded computation. For real-time embedded systems, such self-suspension can cause substantial performance/schedulability degradation. This has led to the investigation of the impact of self-suspension behaviour on timing predictability, with many results reported since 1990. This paper reviews the design and analysis of scheduling algorithms and schedulability tests for self-suspending tasks in real-time systems. We report that a number of these existing approaches are flawed. As a result, we provide (1) a systematic description of how self-suspending tasks can be handled in both soft and hard real-time systems; (2) an explanation of the existing misconceptions and their potential remedies; (3) an assessment of the influence of such flawed analysis on partitioned multiprocessor fixed-priority scheduling when tasks synchronize access to shared resources; and (4) a computational complexity analysis for different self-suspension task models. In summary, this paper provides a state-of-art review of existing real-time analysis of self-suspending tasks to provide a correct platform on which future research can be built.
|
