Concurrency bugs in multithreaded software: modeling and analysis using Petri nets

Abstract

In this paper, we apply discrete-event system techniques to model and analyze the execution of concurrent software. The problem of interest is deadlock avoidance in shared-memory multithreaded programs. We employ Petri nets to systematically model multithreaded programs with lock acquisition and release operations. We define a new class of Petri nets, called Gadara nets, that arises from this modeling process. We investigate a set of important properties of Gadara nets, such as liveness, reversibility, and linear separability. We propose efficient algorithms for the verification of liveness of Gadara nets, and report experimental results on their performance. We also present modeling examples of real-world programs. The results in this paper lay the foundations for the development of effective control synthesis algorithms for Gadara nets.

Eliminating Concurrency Bugs in Multithreaded Software: An Approach Based on Control of Petri Nets

“…The special class of RAS Petri nets that arises in this context is called Gadara nets. The reader is referred to [27] for the formal definition of Gadara nets and for a treatment of their analytical properties in the context of multithreaded programs that use mutually-exclusive locks to control access to shared data. The deadlocks caused by the use of mutuallyexclusive locks are circular-mutex-wait deadlocks, where threads in a set are waiting for one another and none can proceed.…”

“…The deadlocks caused by the use of mutuallyexclusive locks are circular-mutex-wait deadlocks, where threads in a set are waiting for one another and none can proceed. Avoidance of circular-mutex-wait deadlocks in multithreaded programs is mapped to the problem of liveness in Gadara nets in [27]. Liveness here refers to the property that every transition is eventually firable from any reachable state of the net.…”

“…Due to the structure of Gadara nets, liveness is equivalent to reversibility, i.e., the initial state must be reachable from every reachable state. Algorithms based on solving Mixed Integer Linear Programs (MILP) are presented in [27] for determining if liveness holds or not. The algorithms exploit the structural properties of Gadara nets, in terms of certain classes of siphons.…”

“…This solution corresponds to a partition of the set of reachable states of the Gadara net into legal and illegal states. It is shown in [27] that this partition can be done using a set of linear inequalities on the states; in other words, the set of legal states is linearly separable. In [30], the problem of finding the minimum number of linear inequalities for effecting the desired separation of the state space is solved using concepts and techniques borrowed from classification theory.…”

“…This approach does not guarantee at the outset that the number of control places will be minimized. However, it leverages the structural properties associated with liveness analysis in Gadara nets from [27] in the context of an iterative scheme that eliminates illegal states by eliminating so-called resourceinduced empty siphons [20,27]. ICOG employs siphon analysis, coupled with SBPI, as well as a book-keeping mechanism to ensure convergence in a finite number of iterations.…”