Memory corruption is one of the most commonsoftware failures. For sequential software and multi-tasking software with synchronized data accesses, it hasbeen shown that program faults causing memory cor-ruption can be detected by analyzing the relations be-tween defines and uses of variables (DU-based testing).However, such methods are insufficient in preemptivesystems, since they lack the ability to detect inter-taskshared variable dependencies. In this paper, we proposethe use of a system level shared variable DU analy-sis of preemptive multi-tasking real-time software. Byderiving temporal attributes of each access to shareddata using WCET analysis, and combining this infor-mation with the real-time schedule information, ourmethod also detects inter-task shared variable depen-dencies. The paper also describes how we extended theSWEET tool to derive these temporal attributes.
展开▼