Efficent Machine-Code Test-Program Induction

Technology advances allow integrating on a single chip entire system, including memories and peripherals. The test of these devices is becoming a major issue for manufacturing industries. This paper presents a methodology for inducing testprograms similar to genetic programming. However, it includes the ability to explicitly specify registers and resorts to directed acyclic graphs instead of trees. Moreover, it exploits a database containing the assembly-level semantic associated to each graph node. This approach is extremely efficient and versatile: candidate solutions are translated into source-code programs allowing millions of evaluations per second. The proposed approach is extremely versatile: the macro library allows easily changing target processor and environment. The approach was verified on three processors with different instruction sets, different formalisms and different conventions. A complete set of experiments on a test function are also reported for the SPARC processor.