For such a class of Petri nets, the relationship between a multi-step look-ahead deadlock avoidance control method and the structure of the model is established and expanded in a mathematical way. This paper mainly focuses on a class of Petri nets namely the system of simple sequential processes with resources, which contains some special resource places. In the framework of Petri nets, resourcetransition circuits and siphons are often used to characterize and derive a deadlock control policy for an AMS. The deadlock control problem in automated manufacturing systems (AMSs) has received much attention in recent years due to the flexibility of an AMS. Experimental studies demonstrate the efficiency of the proposed approach. In this case, given a state vector, its reachability can be decided by an algebraic method. When an explicit and complete enumeration of reachable states is impossible (due to the limited memory and storage space), we propose the exact number of reachable states. Given an initial resource configuration of a D/SU-RAS, a complete state enumeration can be obtained through pure algebraic operations on this configuration with the pre-computed an initial basis state space. The attention of this work is restricted to a special class of resource allocation systems: Disjunctive/Single-unit (D/SU)-RASs. However, they are stalled by the computational complexity of a large state space that grows exponentially with respect to the size of an underlying RAS. We discuss several special structures that lead to computationally tractable safety characteristics.Ī deluge of approaches has been proposed to implement and deploy a computationally effective and maximally permissive liveness-enforcing supervisor for sequential resource allocation systems (RASs).

The objective of this article is to settle this question by formally establishing the NP-completeness of safety for this class and investigating the boundary between the hard and easy cases. Recent results, however, indicate that safety is often computationally easy. Manufacturing researchers have generally accepted that safety is computationally hard, and numerous suboptimal deadlock avoidance solutions have been proposed for this class. Each request is for a single unit of a single resource and is accompanied by a promise to release the previously allocated resource. For this class, the resource needs of each process are expressed as a well-defined sequence. This work investigates the complexity of safe resource allocation for a class of systems relevant in automated manufacturing. In these cases, establishing the underlying complexity of the safety problem is essential for developing the best deadlock avoidance approach. Although safety is established as NP-complete for certain broad resource allocation classes, newly emerging resource allocation scenarios often exhibit unique features not considered in previous work. The key issue in deadlock avoidance is whether or not a given resource allocation state is safe: that is, whether or not there exists a sequence of resource allocations that completes all processes. See page 204 in the OS slides.Deadlock is a major problem for systems that allocate resources in real time. Remember how Linux deal with this situation? It dynamically promote the priority of those processes. The low priority processes could be starved to death if there are always some high priority processes running. You know, in the system there are lots of processes competing for one CPU.

As long as you can be sure there is no starvation, I don't mind. Well, what if I really hate the writer being favored? Hmm, I smell racism. What if the situation is not so obvious? Shall we favor the writer or the reader? We still favor the writer. Meanwhile in reality, we usually just ignore it as long as our program is fairly robust because it costs too much to do deadlock detection and recovery.Ībout the r/w problem, the solutions favor the writer thread because we clearly know it tends to be starved to death if there are lots of reader threads coming in.

Talking about deadlock, it's good to have it as a research topic.

Deadlock avoidance