Automated testing of LLVM programs with complex input data structures

Symbolic execution is a widely used approach for automatic regression test generation and bug vulnerability finding. The main goal of this paper to present practical symbolic execution-based LLVM programs with complex input data structures. based on the well-known idea lazy initialization, which frees user from providing constraints structures manually. Thus, it provides us fully even program. Two initialization improvements are proposed segmented memory models: one timestamps type information. implemented in KLEE virtual machine platform tested real C — lists, binomial heaps, AVL trees, red-black binary tries.