Utilization of Compiler Information in Decompilation

This thesis deals with the tool for decompilation of binary code into a higher level language, which is being developed as a part of the Lissom project. In this thesis, we present a method to improve efficiency of extraction of information from binary files, based on the detected compiler. Namely the extraction of parameters, return types and names of function from symbol names via the so-called demangling. As a part of this thesis, differences between mangling schemes of different compilers are compared. Afterwards, the implementation of the demangler is described and finally, demangler is tested and demangling success rate is compared for mangled names created by different compilers. The subsequent task is to propose and implement better integration of tools in preprocessing and front-end parts of the Lissom decompiler and optimize parameter distribution among them