Schedulability Aware WCET-Optimization of Periodic Preemptive Hard Real-Time Multitasking Systems

Abstract

In hard real-time multitasking systems, applying WCEToriented code optimizations to individual tasks may not lead to optimal results with regard to the system's schedulability. We propose an approach based on Integer-Linear Programming which is able to perform schedulability aware code optimizations for periodic task sets with fixed priorities. We evaluate our approach by using a static instruction SPM optimization for the Infineon TriCore microcontroller.

Code Optimization of Periodic Preemptive Hard Real-Time Multitasking Systems

“…As shown in [16], the use of EDF for the ILP approach presented in section IV implies that changing the ILP's objective function to min u and therefore minimizing the overall system load is sufficient to optimize the task set with regard to the system's schedulability. Alternatively, eq.…”

“…However, the system will now be schedulable using RMS. We propose the following approach for fixedpriority systems: The approach was previously outlined in [16] in a rudimentary way. Although using the same underlying approach, [16] could not precisely model the number of preemptions of a task by higher priority tasks in the ILP formulation.…”

“…We propose the following approach for fixedpriority systems: The approach was previously outlined in [16] in a rudimentary way. Although using the same underlying approach, [16] could not precisely model the number of preemptions of a task by higher priority tasks in the ILP formulation. This leads to the necessity of solving the formulated ILP twice and to the possibility of still over-approximating the number of preemptions.…”

“…They can therefore not be used to reliably improve the schedulability of a multitasking system. An ILP-based approach which builds up on [14] was proposed in [16] to perform a schedulability-aware optimization of periodic hard real-time multitasking systems. However, this approach needs to solve the proposed ILP twice for fixed-priority systems, resulting in computational overhead and suboptimal results.…”

“…However, this approach needs to solve the proposed ILP twice for fixed-priority systems, resulting in computational overhead and suboptimal results. In this paper, we improve on [16]. Our proposed approach only needs to solve the ILP once and removes the need risk of suboptimal results due to an overapproximated number of preemptions.…”