Evaluating Static Worst-Case Execution-Time Analysis for a Commercial Real-Time Operating System

The worst case execution time (WCET) of a task is a key component in the design and development of hard real-time systems. Malfunctional real-time systems could cause an aeroplane to crash or the anti-lock braking system in a car to stop working. Static WCET analysis is a method to derive WCET estimates of programs. Such analysis obtains a WCET estimate without executing the program, instead relying on models of possible program executions and models of the hardware. In this thesis WCET estimates have been derived on an industrial real-time operating system code with a commercial state-of-the art WCET analysis tool. The goal was to investigate if WCET analysis tools are suited for this type of code. The analyses have been performed on selected system calls and on regions where interrupts are disabled. Our results indicate that static WCET analysis is a feasible method for deriving WCET estimates for real-time operating system code, with more or less intervention by the user. For all analysed code parts of the real-time operating system we were able to obtain WCET estimates. Our results show that the WCET of system calls are not always fixed but could depend on the current state of the operating system. These things are by nature hard to derive statically, and often require much manual intervention and detailed system knowledge. The regions where interrupts are disabled are easier to analyse automatically, since they are usually short and have simple code structures.