A novel resource management technique for deadlock-free systems.

Abstract

Deadlock in a shared resource system is a well-known problem. It has been extensively studied and recently a new class of resource reservation technique is researched upon for deadlock free resource management. This class of technique reserves a portion of the resources. The unreserved resources are freely allocated to any process demanding it. When the unreserved resources are not sufficient for a process demand the reserve pool resources are used such that the process completes and releases all the resources it is holding. This paper presents a new resource reservation technique resource driven DFRR. This technique estimates the optimal number of resources needed for a deadlock free resource reservation policy. The correctness is proved in the form of theorem 1. The theorem 2, suggests the resource reservation with minimal resources. The overhead of the resource pool estimation is $O\left(n\right)$ and that of resource management is $O\left(m\right)$ which is optimal for any deadlock handling technique. The effectiveness of the proposed technique is shown in the form of examples and simulation results.