Least-restrictive enforcement of the Chinese wall security policy

The Chinese Wall security policy states that information from objects that are to be confidential from one another should not flow to a subject. It addresses conflict of interest, and was first articulated in the well-cited work of Brewer and Nash, which proposes also an enforcement mechanism for the policy. Work subsequent to theirs has observed that their enforcement mechanism is overly restrictive -- authorization states in which the policy is not violated may be rendered unreachable. We present two sets of novel results in this context. In one, we present an enforcement mechanism for the policy that is simple and efficient, and least-restrictive -- an authorization state is reachable if and only if it does not violate the policy. In our enforcement mechanism, the actions of a subject can constrain the prospective actions of another, a trade-off that we show every enforcement mechanism that is least-restrictive must incur. Our other set of results is that the enforcement mechanism of Brewer-Nash is even more restrictive than previous work establishes. Specifically, we show: (1) what is called the *-rule is overspecified in that one of its sub-rules implies the other, and, (2) if a subject is authorized to write to an object that contains confidential information, then all objects that contain confidential information must belong to the same conflict of interest class. Our work sheds new light on what is generally considered to be important work in information security.