A knowledge-based debugger for real-time software systems based on a noninterference testing architecture

Abstract

On the basis of a general mechanism for reproducing the execution behavior of real-time software systems, a new approach is suggested for a knowledge-based debugger for real-time software systems. A description is given of the use of noninterference architecture to collect the program execution history of a target system without interfering with the execution of the target system. Thus, the timing constraints of a real-time software system can be preserved. To eliminate the redundant information in the collected execution history, a postprocessing mechanism is introduced to organize the necessary information for testing and debugging. The redundant data, which are useless for debugging certain types of error, are removed, and only the required data proceed to the debugging phase. A knowledge-based debugging tool is then used to assist users in localizing errors. Examples are used to illustrate the approach in detecting synchronization and timing errors of real-time software systems.