Search

Exception handling is a well known mechanism to react on errors or exceptional events during the runtime of a program. In this paper we discuss the reliability and efficiency demands that real-time systems make upon exception handling. The reliability requirement means that no runtime error should occur due to the wrong use of the exception handling commands in a program. We use static type checking to guarantee the correct typing of the exception handling constructs. Data flow analysis is required to check whether an appropriate handler exists for each exception. The efficiency requirement means that the execution of all statements being involved in exception handling should produce as little runtime overhead as possible. We describe how the two most time-critical steps in exception handling, the determination of the associated handler and the scheduling, can be implemented efficiently by using specific runtime data structures and by making use of data flow analysis.