Communication Deadlocks Research Articles

MPI-RCDD: A Framework for MPI Runtime Communication Deadlock Detection

The message passing interface (MPI) has become a de facto standard for programming models of high-performance computing, but its rich and flexible interface semantics makes the program easy to generate communication deadlock, which seriously affects the usability of the system. However, the existing detection tools for MPI communication deadlock are not scalable enough to adapt to the continuous expansion of system scale. In this context, we propose a framework for MPI runtime communication deadlock detection, namely MPI-RCDD, which contains three kinds of main mechanisms. Firstly, MPI-RCDD has a message logging protocol that is associated with deadlock detection to ensure that the communication messages required for deadlock analysis are not lost. Secondly, it uses the asynchronous processing thread provided by the MPI to implement the transfer of dependencies between processes, so that multiple processes can participate in deadlock detection simultaneously, thus alleviating the performance bottleneck problem of centralized analysis. In addition, it uses an AND⊕OR model based algorithm named AODA to perform deadlock analysis work. The AODA algorithm combines the advantages of both timeout-based and dependency-based deadlock analysis approaches, and allows the processes in the timeout state to search for a deadlock circle or knot in the process of dependency transfer. Further, the AODA algorithm cannot lead to false positives and can represent the source of the deadlock accurately. The experimental results on typical MPI communication deadlock benchmarks such as Umpire Test Suit demonstrate the capability of MPI-RCDD. Additionally, the experiments on the NPB benchmarks obtain the satisfying performance cost, which show that the MPI-RCDD has strong scalability.

VANDALISME DALAM PEMANFAATAN LAYANAN INFORMASI PERPUSTAKAAN

The purpose of this research is to know the factors that cause vandalism in the library. The research object discusses the importance of preventing vandalism as well as understanding the causes. The scope of research is focused on the basis for the creation of library services, so that a positive image of the library can be realized. The historical conceptual application of thought is critically analyzed realistically, in research methodit is called a historical approach and primary data is the primary source of information. The data collection method is library research. The research analysis is descriptive qualitative and content analysis to analyze the meaning of all the notions of deviant behavior, the factors that cause vandalism and its application to library institutions. Utilization of observation as a complement to support theoretical construction. The results of the study are the factors that cause vandalism in the library are: (1). Factors from users consist of; (a). Disappointment with library services, (b). There is an intention, (c). There is an opportunity, (d). Lack of awareness of library users, (e). Stress. (2). Factors from the library consist of; (a). Loose supervision, (b). Unprofessional Officers. (3). Environmental factors consist of; (a) The economic conditions of library users, (b). Social factors of library users, (c). The library room management is incorrect. (4). Communication deadlock. Keyword : Vandalism, Library Collection; Deviant Behavior

Translating active objects into colored Petri nets for communication analysis

Actor-based languages attract attention for their ability to scale to highly parallel architectures. Active objects combine the asynchronous communication of actors with object-oriented programming by means of asynchronous method calls and synchronization on futures. However, the combination of asynchronous calls and synchronization may introduce communication cycles which lead to a form of communication deadlock and livelock. This paper addresses such communication deadlocks and livelocks for ABS, a formally defined active object language which additionally supports cooperative scheduling to express complex distributed control flow, using first-class futures and explicit process release points. Our approach is based on a translation of the semantics of ABS into colored Petri nets, such that a program and its state correspond to a marking of this net. We prove the soundness of this translation and demonstrate by example how the implementation of this net can be used to analyze ABS programs with respect to communication deadlocks and livelocks.

The Implementation of a Communication Deadlock Analysis Based on the Theory of Deadlock Analysis for the Actor Model

We present a unique approach for communication deadlock analysis for actor-model which has an under-approximated analysis result. Our analysis detects narrowly defined communication deadlocks by finding a cyclic dependency relation in a novel dependency graph called the slave dependency graph. The slave dependency graph is based on a new relationship between Actors, slave dependency, defined by us. After that, we implement this theory in Soot, an analysis tool for Java, and use it to analyze actor-based Java program realized by Akka, a Java library that allows actor-based programming. We argue that our analysis can detect a specific kind of communication deadlock with the precise result, but has many limitations.

Specification and Verification in Integrated Model of Distributed Systems (IMDS)

Distributed systems, such as the Internet of Things (IoT) and cloud computing, are becoming popular. This requires modeling that reflects the natural characteristics of such systems: the locality of independent components, the autonomy of their decisions, and asynchronous communication. Automated verification of deadlocks and distributed termination supports rapid development. Existing techniques do not reflect some features of distribution. Most formalisms are synchronous and/or use some kind of global state, both of which are unrealistic. No model supports the communication duality that allows the integration of a remote procedure call and client-server paradigm into a single, uniform model. The majority of model checkers refer to total deadlocks. Usually, they do not distinguish between communication deadlocks from resource deadlocks and deadlocks from distributed termination. Some verification mechanisms check partial deadlocks at the expense of restricting the structure of the system being verified. The paper presents an original formalism for the modeling and verification of distributed systems. The Integrated Model of Distributed Systems (IMDS) defines a distributed system as two sets: states and messages, and the relationship of the “actions” between these sets. Communication duality provides projections on servers and on traveling agents, but the uniform specification of the verified system is preserved. General temporal formulas over IMDS, independent of the structure of the verified system, allow automated verification. These formulas distinguish between deadlocks and distributed termination, and between communication deadlocks and resource deadlocks. Partial deadlocks and partial termination can be checked. The Dedan tool was developed using IMDS formalism.

Precise Predictive Analysis for Discovering Communication Deadlocks in MPI Programs

The Message Passing Interface (MPI) is the standard API for parallelization in high-performance and scientific computing. Communication deadlocks are a frequent problem in MPI programs, and this article addresses the problem of discovering such deadlocks. We begin by showing that if an MPI program is single path, the problem of discovering communication deadlocks is NP-complete. We then present a novel propositional encoding scheme that captures the existence of communication deadlocks. The encoding is based on modeling executions with partial orders and implemented in a tool called MOPPER . The tool executes an MPI program, collects the trace, builds a formula from the trace using the propositional encoding scheme, and checks its satisfiability. Finally, we present experimental results that quantify the benefit of the approach in comparison to other analyzers and demonstrate that it offers a scalable solution for single-path programs.

Communication and Resource Deadlock Analysis Using IMDS Formalism and Model Checking

Modern static deadlock detection techniques deal with the global properties of the verified systems, using methods that explore the state space. Local features, like partial deadlocks or individual process terminations, are not easily expressed or checked by such methods. Also the distinction between communication deadlocks and resource deadlocks, common in dynamic waits-for methods, cannot be addressed or verified by static methods. An Integrated Model of Distributed Systems (IMDS) is proposed which specifies distributed systems as sets of servers’ states, sets of messages and sets of actions. The message passing/resource sharing dualism of distributed systems is provided by projections: on servers (server view) and on agents (agent view), yet the uniform specification of a verified system is preserved. A progress of computation is defined in terms of actions which change (local) states and generate messages. Distributed actions do not depend on global states and are independent from one another. Therefore, local features of subsystems can be easily described in IMDS. Communication and resource deadlocks can be handled separately and total and partial deadlocks and terminations can be distinguished from each other. Integration of IMDS with model checking is outlined and temporal formulas for deadlock and termination checking are discussed.

Breaking the climate change communication deadlock

Climate change communication is trapped between the norms that govern scientific practice and the need to engage the public. Overcoming this tension requires new societal institutions where the science and politics of climate change can co-exist.

Promoting Harmony during Disagreements: A Portrait of Adult Discussions in the Indonesian Context

This paper presents different strategies applied by adult students when expressing disagreements during discussions. The use of a variety of strategies by the students show that conveying disagreements is not an easy task for the academia living in a collectivist society. Responses to the disagreements are also varied, ranging from communication deadlock to partial acceptance, support, offer of an account, support from another participant, neutralizer from another participant, and acceptance. The analysis of the data shows that, in the presence of disagreements, learning is mostly facilitated by disagreements expressed through peer acknowledgment and vulnerability assertion.

Self-Stabilizing Symmetry Breaking in Constant Space

We investigate the problem of self-stabilizing round-robin token management on a bidirectional ring of identical processors. Each processor is an asynchronous probabilistic finite state (i.e., constant space) machine which sends and receives constant-size messages and whose state transition is triggered by the receipt of a message. We also show that this problem is equivalent to symmetry breaking (i.e., leader election). Wejustify and suggest a two-layer (hardware and software) solution to the token management problem: The subproblem of reducing an arbitrary but nonzero number of tokens (in an otherwise arbitrary initial system state) to exactly one token (and a legal system state) is solved in hardware and takes only small polynomial time. The detection of a complete lack of tokens (communication deadlock) is done by a software clock. In high-speed networks the hardware layer can be implemented using fast universal switches (i.e., finite state machines) independent of the size of the network. We note that randomization is essential, since Dijkstra showed that for arbitrary rings the subproblem does not have a deterministic solution (regardless of the computational power of the identical processors). The use of the software layer (deadlock detection) in our solution is minimized.

Concurrency Model for Distributed Systems

Concurrent behaviors in distributed systems are generally difficult to specify and analyze. In this paper, we introduce a concurrency model that comprises concurrent events, concurrency blocks, and global states. The model is attractive in that it is simple and yet allows for the definition and calculation of concurrency measures, communication deadlocks and data races, as well as supporting concurrent system specifications at a high-level abstraction that is useful for understanding concurrent behaviors. An efficient algorithm based on the concept of concurrency blocks for computing concurrency measures is provided. To demonstrate the practicality of the model, concurrency measures of asynchronous communications and synchronous communications are compared. Brief discussions of how the model can be applied to the formal description techniques Estelle, SDL, and LOTOS for reasoning and classification of concurrency behaviors are given.

Communication deadlock in muti-agent systems

This paper presents the communication deadlock problem in multi-agent systems. A model of extended deterministic finite-state machine is stated here to formalize the problem and discuss its properties, A sufficient condition is achieved that communication deadlock will never occur if the corresponding extended deterministic finite-state machine is an acyclic graph. Then two algorithms are discussed, one is a simple but efficient algorithm, and the other considers global optimization according to costs.

Achieving fault-tolerant multicast in injured wormhole-routed tori and meshes based on Euler path construction

Recently, wormhole routers with multidestination capability have been proposed to support fast multicast in a multicomputer network. To avoid communication deadlock, existing results have proposed to construct a Hamilton path, Euler path, trip, or their variants in the network, perhaps with some degree of support of virtual channels. In this paper, we identify that a network which is itself Eulerian or is Eulerian after some links are removed, can enjoy the multidestination capability without support of virtual channels. From this definition, we then develop several techniques to achieve fault-tolerant multicast in a torus/mesh of any dimension with regular fault patterns (such as single node, block, L-shape, T-shape, +-shape, U-shape, and H-shape) and even irregular fault patterns. The result improves over existing results on the requirement of support of virtual channels and fault-tolerant capability. Simulation results on tori are presented.

Resource Bounds for Self-Stabilizing Message-Driven Protocols

Self-stabilizing message-driven protocols are defined and discussed. The class weak exclusion that contains many natural tasks such as $\ell$-exclusion and token passing is defined, and it is shown that in any execution of any self-stabilizing protocol for a task in this class, the configuration size must grow at least in a logarithmic rate. This last lower bound is valid even if the system is supported by a time-out mechanism that prevents communication deadlocks. Then we present three self-stabilizing message-driven protocols for token passing. The rate of growth of configuration size for all three protocols matches the aforementioned lower bound. Our protocols are presented for two-processor systems but can be easily adapted to rings of arbitrary size. Our results have an interesting interpretation in terms of automata theory.

RESTRICTION-FREE ADAPTIVE WORMHOLE ROUTING IN MULTICOMPUTER NETWORKS

The adaptive routing approach has been expected as a promising way to improve network performance by utilizing available network bandwidth. Previous adaptive routing strategies in wormhole-routed multicomputer networks restrict the routing of messages by the routing algorithm to prevent deadlock. This results in low degree of adaptivity and low utilization of physical or virtual channels. In this paper, we examine the possibility of performing restriction-free adaptive routing in wormhole-routed networks as an approach to further improving the performance of these networks. A new flow control policy, called message cutting-in, is proposed, and two adaptive routing strategies are presented. Freedom of communication deadlock is achieved by the proposed flow control policy. The proposed adaptive routing strategies do not restrict routing and maximally utilize the physical and virtual channels. Simulation results show that the restriction-free adaptive routing approach is promising from the fact that it has the lowest latency and highest throughput depending on the number of virtual channels per physical channel and patterns of message traffic.

TOPOLOGY DEPENDENCE IN LATTICE SIMULATIONS OF NON-LINEAR PDES ON A MIMD COMPUTER

We tested the parallelization of explicit schemes for the solution of non-linear classical field theories of complex scalar fields which are capable of simulating hadronic collisions. Our attention focused on collisions in a fractional model with a particularly rich inelastic spectrum of final states. Relativistic collisions of all types were performed by computer on large lattices (64 to 256 sites per dimension). The stability and accuracy of the objects were tested by the use of two other methods of solutions: Pseudo-spectral and semi-implicit. Parallelization of the Fortran code on a 64-transputer MIMD Volvox machine revealed, for certain topologies, communication deadlock and less-than-optimum routing strategies when the number of transputers used was less than the maximum. The observed speedup, for N transputers in an appropriate topology, is shown to scale approximately as N, but the overall gain in execution speed, for physically interesting problems, is a modest 2–3 when compared to state-of-the-art workstations.

A new deadlock prevention scheme for nonminimal adaptive wormhole routing

In multicomputer networks, the adaptive routing has been expected as a promising way to improve network performance by utilizing available network bandwidth. Previous adaptive routing algorithms in wormhole-routed multicomputer networks restrict the routing of messages to prevent deadlocks, and the routing restriction results in low degree of adaptiveness and low utilization of communication channels. In this paper, we examine the possibility of performing restriction-free, nonminimal adaptive routing in wormhole-routed networks as an approach to further improving the performance of these networks. A new flow control policy, called message cutting, is proposed, and two adaptive routing strategies are presented. Freedom of communication deadlock is achieved by the proposed flow control policy. The proposed adaptive routing strategies do not restrict routing and maximally utilize the physical and virtual channels. Simulation results show that the restriction-free adaptive routing approach is promising from the fact that it has the lowest latency and highest throughput depending on the number of virtual channels per physical channel and patterns of message traffic.

DPDP: An interactive debugger for parallel and distributed processing

This article presents the debugger for parallel and distributed processing (DPDP) system. First, the processing environment is defined; then, several types of bugs that may occur within that environment are investigated. DPDP is designed to provide an adequate tool for parallel and distributed debugging. Accordingly, it enables the programmer to monitor, detect, and correct a system interactively. The system design is based on a formal event-action model that exploits a trace file to record the events, whereas the actions are determined according to a syntactic analysis of the underlying events. This work also includes the design and implementation of the DPDP system on a local area network. DPDP is available in C language and is used to debug parallel and distributed applications that are written in C. Four case studies are considered to evaluate its capabilities. The study includes experiments for finding race conditions, debugging of sequencing errors, and detecting both resource and communication deadlocks.

Adaptive Deadlock-Free Packet Routeing in Transputer-Based Multiprocessor Interconnection Networks

This paper presents a deadlock-free routeing algorithm for multiprocessor interconnection networks based on store-andforward (S/F) communication. The adaptive nature of the method proposed encourages using light traffic paths. Furthermore it has the properties of avoiding blocked communication (deadlock), reducing communication delay time between source and destination, using efficiently message buffers as network resources and being able to control communication traffic flow from each processor of the network. The routeing algorithm has been implemented on a 64node transputer network (T-Rack) configured as a number of well known topologies to evaluate the behaviour of the algorithm and some performance figures have been derived.

Distributed algorithm for communication deadlock detection

In a distributed system, a set of processes may sometimes get involved in a communication deadlock. This is a situation where each process in the group waits for some process to communicate with it, but no other process is attempting communication. A new algorithm for deadlock detection in the communication model is presented. The proposed scheme is based on a new approach to the problem of distributed knot detection, which appears to be more efficient and simpler than existing ones. The resulting deadlock detection algorithm has the feature that when it terminates, the initiator knows all the processes that have caused the deadlock and may subsequently proceed to deadlock resolution.