Intellectual property protection is a major concern for fabless IC designers. Among the proposed protection means, the active ones are preventing counterfeiting and over-usage to occur in the first place. One of the solution to implement an active design data protection scheme is to modify the combinational logic. Several methods are available, called logic encryption, logic obfuscation, logic masking or logic locking. A formal framework is first provided for these notions. We clearly define those four types of logic modification, and give didactic examples. Then, a new method to achieve logic locking is presented. This method, based on graph analysis, allows to select the insertion sites for the extra gates orders of magnitudes faster than existing techniques. We give experimental results following from a practical implementation and discuss design considerations about integration in an overall, more robust, protection scheme. We also consider existing attacks and propose some countermeasures.