In addition to spectrum scarcity, unfair and insufficient channel contention resolution has become one of the major bottlenecks for a good throughput in multi-hop wireless ad-hoc networks (MHAHNs). In this paper, we propose an optimal cognitive MAC protocol for MHAHNs under opportunistic spectrum access (OSA) approach in which channel contention resolution is done on the basis of interference-dependent random access addressing both social welfare maximization and energy efficiency. We show that, in OSA-based MHAHNs, contention control at MAC layer has an interaction relationship with power control at physical layer and congestion control at transport layer. Studying their mutual effect on such an OSA-based MHAHNs' performance motivates an optimal cognitive MAC framework which is shown as an NP-hard problem. To solve the problem, we introduce some auxiliary variables which are interpreted as interference weights and develop a distributed solution, which has been proved to converge to the global optimum.