Library functions identification in binary code by using graph isomorphism testings

Library functions identification is a key technique in reverse engineering. Discontinuity and polymorphism of inline and optimized library functions in binary code create a difficult challenge for library functions identification. To solve this problem, a novel approach is developed to identify library functions. First, we introduce execution dependence graphs (EDGs) to describe the behavior characteristics of binary code. Then, by finding similar EDG subgraphs in target functions, we identify both full and inline library functions. Experimental results from the prototype tool show that the proposed method is not only capable of identifying inline functions but is also more efficient and precise than the current methods for identifying full library functions.