Title :
Generating Efficient Distributed Deadlock Avoidance Controllers
Author :
Sánchez, César ; Sipma, Henny B. ; Manna, Zohar
Author_Institution :
Dept. of Comput. Sci., Stanford Univ., CA
Abstract :
General solutions to deadlock avoidance in distributed systems are considered impractical due to the high communication overhead. In previous work we showed that practical solutions exist when all possible sequences of resource requests are known a priori in the form of call graphs; in this case protocols can be constructed that involve no communication. These run-time protocols make use of annotations of the call graph that are computed statically based on the structure of the call graph. If the annotations are acyclic, then deadlocks are unreachable. This paper focuses on the computation of these annotations. We first show that our algorithm for computing acyclic annotations is complete: every optimal annotation can be generated. We then show that, given a cyclic annotation and a fixed set of resources, checking whether deadlocks are reachable is NP-complete. Finally, we consider the problem of computing minimal annotations that satisfy given constraints on the number of available resources. We show that the problem is NP-complete in the general case, but that it can be solved in polynomial time if the only restrictions are that the number of certain resources is 1, that is, these resources are binary semaphores.
Keywords :
computational complexity; distributed processing; graph theory; resource allocation; system recovery; NP-complete; acyclic annotation; binary semaphore; call graph; deadlock avoidance controller; distributed system; polynomial time; resource allocation; run-time protocol; Communication system control; Computer science; Control systems; Distributed control; Embedded system; Polynomials; Protocols; Resource management; Runtime; System recovery; Deadlock Avoidance; Distributed Algorithms; Scheduling;
Conference_Titel :
Parallel and Distributed Processing Symposium, 2007. IPDPS 2007. IEEE International
Conference_Location :
Long Beach, CA
Print_ISBN :
1-4244-0910-1
Electronic_ISBN :
1-4244-0910-1
DOI :
10.1109/IPDPS.2007.370357