Full Regular Temporal Property Verification as Dynamic Program Execution

Abstract

Verification of programs in code-level has attracted more and more attentions and considerable progress has been made in this area. The early research is limited to the verification of safety properties [1], [2], [3]. To do so, assertions are required to be instrumented in the program to be verified in advance, and the reachability of error labels is checked afterwards. In this way, safety properties can be verified. However, verification of other temporal properties such as liveness cannot be supported. Model checking temporal properties without executing code [4], [5], [6], [7] considers all possible behaviors, which makes a small program have a very large state-space. Thus, it is difficult to apply the approach to verifying programs in large scale. In addition, the approach is poor in accuracy and often produces unsound results. Runtime verification is pursued to check whether one or a few execution traces satisfy a given property by monitoring the execution of a system [8], [9]. Although it avoids the state-explosion problem, extracting information from a running system and sending them to monitors may generate a large runtime overhead. In a word, the techniques available are far more from mature in the field of temporal property verification of programs in code-level.