Study And Implement On Basic Data Type Analysis In Static Binary Translation

With the rapidly development of computer architecture, many excellent old softwares are confronted with being abandoned. At the same time, it needs to rapidly migrate the old software to the new architecture when we study and develop a new architecture that can prevail and be widely used in short time. So the binary translation and decompilation techniques were emerged with the times require. The decompilation technique has the significant meanings in software maintenance, software development and debug, finding virus and bugs fields, and so on.Data type analysis is a difficult task in binary translation and decompilation. In this paper, firstly we introduce the ITA system that is a static binary translation system. Then the technique of data type analysis in binary translation is proposed based the ITA system plat. We introduce the optimization and data flow analysis techniques of compilation into binary translation. Data type information will be obtained through analyzing instruction semantic and HRTL representation of the procedure. The main subject of this paper includes:The variables of procedure are renamed. The register operands will be transformed the variables when the instruction semantic is transformed HRTL representation. One register may be used by many data pieces which have different type properties because of the limit of register number and the optimization tactic of compiler. So, many data pieces which may have different type properties and have no relations with each other can be denoted by one variable in HRTL representation. It is necessary to rename the variables of procedure HRTL representation through that data flow analysis and set partition theory.The variable types are deduced according to type deducing rules in the locations where it is defined and used. These types compose a set. The type deducing rules are important, the rules are more complete and the information of the variable type is more accurate.The final variable type is obtained according to lattice theory. The relations of the basic types in binary translation can be represented by a lattice. We can obtain the final variable type through comparing the type set of the variable that was deduced in the second step with the type lattice. The final variable type will be as the declared variable type when the HRTL representation of the procedure is transformed to the C representation.Experimental results show that the proposed technique is feasible, effective and correct in static binary translation.