Array type abstraction reconstruction in binary code

Reconstruction of array abstractions plays a crucial role in vulnerability detection, reverse engineering and malicious code analyses. This paper presents a method for reconstructing multi-dimensional arrays. The loop semantics is extracted and transformed to the relative FOREACH formula set. Five FOREACH formula regulations are presented to calculate the dimensions, size, multi-dimension array bounds, basic element size and scale of each dimension to reconstruct the array. The results are compared to 15 programs in CoreUtils tools which are compiled by the gcc compiler. The present method can reconstruct 16.3% more files than the IDA Pro Hex-ray plugin.