Gertrude Levine

Defining deadlock

Deadlock has been widely studied in many fields of computer science, notably in communications, database, and operating systems. Perhaps because (at least one state called) deadlock is tractable, there exists extensive literature on the subject. Deadlock was apparently decisively defined over thirty years ago, with its characteristics and handlers. Yet, much of the literature remains inconsistent in its treatment of this anomaly.A more precise definition is clearly needed to distinguish between the different states that are termed deadlock. A classification of dead states is required to distinguish the causes and appropriate handlers for each. We introduce a model to structure our research.

Defining defects, errors, and service degradations

The study of defects is a principal topic of software systems, af-fecting all phases of a systems lifecycle. Defects are the cause of errors and service degradations. Unresolved errors cause failures. If defects cannot be prevented effectively, then error control me-chanisms must be evaluated. We introduce a model to distinguish between defects, errors, and service degradations. A two-dimensional classification scheme is developed for defects, defined by the types of process interaction and software corruption that are involved. A third dimension is added to this taxonomy for defects that cause service degradation, based on the deviations in service quality that are tolerated. We investigate the role of service degradation in error prevention.

THE CONTROL OF STARVATION

A generalized discussion of the anomaly of starvation is presented. Starvation is defined to be an infinite or unbounded delay, and classified in terms of its cause and the methods used for its control. A model and a set of control mechanisms are developed. The model, appropriate for various buffered and unbuffered resource systems, allows an ordering of requests by time and by stages of their lifetimes. We investigate the behavior and control of requests during starvation and during various anomalies of conflict.

Controlling deadlock in Ada

Adas concurrency constructs provide synchronization and mutual exclusion mechanisms to control resource allocation. These constructs, if improperly used, can cause indefinite or infinite postponement of completion of service.Hembold and Luckham [3] and Cheng, Araki, and Ushijima [2] identify classes of these anomalies in order to aid in their detection. In this article we continue their discussion of the classification and control of these unbounded delays.

Priority inversion with fungible resources

Priority inversion occurs when the execution of a task is unnecessarily delayed by the dispatch of a lower priority task. This anomaly can result in failure if it causes the delay of hard real-time tasks. Priority inheritance protocols have been developed to limit priority inversions during competition over shared resources. Such methods are designed for individually identifiable resource units. Other approaches are indicated when shared resources contain interchangeable resource units. Current practices with resource deadlock, a scheduling anomaly that has many characteristics in common with priority inversion, provide insight into the control of priority inversion with these fungible resources.

Kernel Ada to unify hardware and software design

1. ABSTRACT This paper is a call to SIGAda members to make a determined thrust to broaden Ada usage in the commercial world. More specifically, we wish to exploit an opportunity in the Electronic Design Automation industry (EDA) to use (a kernel of) Ada as a hardware description language (HDL) for the design and programming of todays System On a Chip (SOC). The Ada subset referred to in this paper is intended for the EDA domain to solve practical problems as an HDL, and, in addition, as the interfacing programming language used for testing and simulations (a market currently dominated by C/C++). Simplicity can be a significant aid in penetrating the market of users and makers of EDA tools. This community consists mostly of electrical engineers and tool manufacturers who would have both a cultural orientation and a commercial interest in a simple kernel of Ada. We use examples to illustrate the appropriateness of Kernel Ada for the development and testing of reusable hardware components. 1.1

The Game of Life with Ada tasks

The Game of Life is a popular programming project, both in computer science courses and on the World Wide Web. It is commonly implemented with a bounded array of values representing the cells of life. This project provides an appealing visual representation of cells living or dying over repeated generations.An Ada implementation using tasks allows a number of interesting features. State information can be retained between calls. A generic declaration defines a conceptually unbounded grid, which a distributed implementation with tasks can potentially implement. Tasking communication provides an exercise in both inter-cell communication and in deadlock prevention. Lastly, the independence of cell communication can allow for relaxing the synchronization of cell generations.

Computer security with service degradations

Computer systems invest substantial resources in securing their service. Costs of failure prevention are balanced against those of detection and recovery, even though recovery mechanisms can cause greater degradations of services. Most computer users consider degradations of service to be insidious and injurious. Yet, service degradations commonly assist in both the prevention and the curtailment of failures. Their toleration enables service continuation following fault activation, during which degradations are monitored for symptoms of security breaches.

Signaling from within Interrupt Handlers Reconsidered

In M. Ben-Aris solution the two tasks synchronize partially through global variables . M. Ben-Ari recognizes that his solution contains a minor bug in that the fast arrival of a secon d interrupt will suspend the Interrupt Handler for a cycle of execution of the Interrupt Task . In our correspondence [3], he also agrees that his solution is not good Ada style, and tha t his global variables require qualification by pragma SHARED (which, however, makes hi s generalization invalid and is not supported by all compilers) .

Incorporating Service Degradations into a Security Plan

Computer systems invest substantial resources in securing their services. Costs of failure prevention are balanced against those of detection and recovery, both of which typically involve substantial degradations. Most computer users consider degradations of service to be insidious and injurious. Yet, degradations commonly assist in both the prevention and the curtailment of failures. Their toleration enables service continuation following fault activation, during which degradations are monitored for symptoms of security breaches. A systematic approach to degradations should be incorporated into security plans. We propose a template in which service degradations are classified according to their usage and their impact on services within the different stages of a systems lifecycle. Such standardization will promote system security.