Geração de casos de teste para linguagens com aritimética de ponteiro

Abstract

Software testing and debugging are hard tasks. In general, it is necessary that the execution flow reaches a function in order to test it. Much effort was employed in the development of techniques to automatically generate test cases to analyze functions of interest. However, such techniques have limitations. One which catches our attention is the fact that, from the best of our knowledge, no current technique relates arrays received by functions with their sizes. Thus, existing techniques may generate test cases which lead to invalid memory accesses that would not happen in a real execution of the program. We present two static analyses for inference of array sizes, as well as an algorithm for generation of test cases capable of generating safe tests by using such analyses. The analyses were able to bind sizes for 34.6% of the arrays received by functions of the SPEC CPU 2006 benchmark suite. We also show that our technique is useful for the automatic execution of interest functions.