CP with Architectural State Lookup for Functional Test Generation

Constraint programming (CP) is a powerful technology that has proved very useful in random functional test generation. On the other hand the complexity of the test generation process prevents it from being represented and solved as a single constraint satisfaction problem (CSP). Developers of test generation tools have to decompose the problem into a set of smaller tasks and to make CP work in interaction with other technologies, e.g. architectural simulator. The task of turning these foreign technologies into an integral part of the CP search algorithm raises multiple challenges. We propose a method for accomplishing this task by means of so called custom constraints, which are based on demons and custom propagation algorithms. We implemented the method in a framework of ILOG Solver and tested it on a use case of the memory address translation mechanism for protected mode of IA-32 processor