Citation Context

...exhibited by the program. 3.1. Actual Program Executions The first part of our model is simply a notation for representing an execution of a shared-memory parallel program on a sequentially consistent=-=[14] process-=-or + . A program execution, P, is a triple, ��E, T , D ��, where E is a finite set of events, and T (the temporal ordering relation) and D (the shared-data dependence relation) are relations #...

Citation Context

...stein's conditionssstate that atomic execution is guaranteed if shared variables that are read and modified by the critical section are not modified by any other concurrently executing section of code=-=[3]-=-. A violation of these conditions has typically been called a data race[1, 4, 5, 11, 16, 17, 20, 22] or access anomaly[6-8, 18]. We prefer the term data race. Figure 1 shows an example program for whi...

Citation Context

...n them has not been previously recognized. Because no consistent terminology has appeared, several terms have been used with different intended meanings, such as access anomaly[6-8, 12, 18], data race=-=[1, 4, 5, 11, 16, 20, 22]-=-, critical race[13], harmful shared-memory access[24], race condition[10, 26], or just race[2, 9, 17]. This paper explores the nature of race conditions and uncovers some previously hidden issues rega...

Citation Context

...ost/Wait synchronization without Clear operations); detection is NP-hard for more powerful types of synchronization (such as semaphores). 1 To be consistent with the fault tolerant research community =-=[25]-=-, a failure occurs when a program’s external behavior differs from its specification, and a fault is its algorithmic cause (although we use the term bug). 2 There is some controversy over terminology ...

Citation Context

...n them has not been previously recognized. Because no consistent terminology has appeared, several terms have been used with different intended meanings, such as access anomaly[6-8, 12, 18], data race=-=[1, 4, 5, 11, 16, 20, 22]-=-, critical race[13], harmful shared-memory access[24], race condition[10, 26], or just race[2, 9, 17]. This paper explores the nature of race conditions and uncovers some previously hidden issues rega...

Citation Context

...n them has not been previously recognized. Because no consistent terminology has appeared, several terms have been used with different intended meanings, such as access anomaly[6-8, 12, 18], data race=-=[1, 4, 5, 11, 16, 20, 22]-=-, critical race[13], harmful shared-memory access[24], race condition[10, 26], or just race[2, 9, 17]. This paper explores the nature of race conditions and uncovers some previously hidden issues rega...

Citation Context

...sed with different intended meanings, such as access anomaly[6-8, 12, 18], data race[1, 4, 5, 11, 16, 20, 22], critical race[13], harmful shared-memory access[24], race condition[10, 26], or just race=-=[2, 9, 17]-=-. This paper explores the nature of race conditions and uncovers some previously hidden issues regarding the accuracy and complexity of dynamic race detection. We present the following results. (1) Tw...

Citation Context

...sed with different intended meanings, such as access anomaly[6-8, 12, 18], data race[1, 4, 5, 11, 16, 20, 22], critical race[13], harmful shared-memory access[24], race condition[10, 26], or just race=-=[2, 9, 17]-=-. This paper explores the nature of race conditions and uncovers some previously hidden issues regarding the accuracy and complexity of dynamic race detection. We present the following results. (1) Tw...

Citation Context

...t the distinction between them has not been previously recognized. Because no consistent terminology has appeared, several terms have been used with different intended meanings, such as access anomaly=-=[6-8, 12, 18]-=-, data race[1, 4, 5, 11, 16, 20, 22], critical race[13], harmful shared-memory access[24], race condition[10, 26], or just race[2, 9, 17]. This paper explores the nature of race conditions and uncover...

Citation Context

...n them has not been previously recognized. Because no consistent terminology has appeared, several terms have been used with different intended meanings, such as access anomaly[6-8, 12, 18], data race=-=[1, 4, 5, 11, 16, 20, 22]-=-, critical race[13], harmful shared-memory access[24], race condition[10, 26], or just race[2, 9, 17]. This paper explores the nature of race conditions and uncovers some previously hidden issues rega...

Citation Context

...perties. In this section, we briefly overview our model for reasoning about race conditions that was first presented in an earlier paper[22] and that is based on Lamport's theory of concurrent systems=-=[15]-=-. Our model consists of two parts: one to represent the actual behavior exhibited by the program and the other to represent potential behaviors possibly exhibited by the program. 3.1. Actual Program E...

Citation Context

...rform the same events as P, it must also exhibit the same shared-data dependences as P; when general races occur (whether or not they are considered failures), F SYNC may contain infeasible executions=-=[19, 22]-=-. As discussed later, the existence of such infeasible executions impacts the accuracy of races reported by methods that analyze only explicit synchronization. Nevertheless, we will see that this noti...

Citation Context

...eral terms have been used with different intended meanings, such as access anomaly[6-8, 12, 18], data race[1, 4, 5, 11, 16, 20, 22], critical race[13], harmful shared-memory access[24], race condition=-=[10, 26]-=-, or just race[2, 9, 17]. This paper explores the nature of race conditions and uncovers some previously hidden issues regarding the accuracy and complexity of dynamic race detection. We present the f...

Citation Context

...t the distinction between them has not been previously recognized. Because no consistent terminology has appeared, several terms have been used with different intended meanings, such as access anomaly=-=[6-8, 12, 18]-=-, data race[1, 4, 5, 11, 16, 20, 22], critical race[13], harmful shared-memory access[24], race condition[10, 26], or just race[2, 9, 17]. This paper explores the nature of race conditions and uncover...

Citation Context

...ection An important aspect of debugging involves determining whether portions of the execution are race-free in the sense that they are unaffected by incorrect or inconsistent dats produced by a race =-=[5, 20]-=-. For example, a programmer browsing an execution trace might focus only on portions of the trace recorded before any races occurred. These portions of the trace contain events that are guaranteed to ...

Citation Context

...eral terms have been used with different intended meanings, such as access anomaly[6-8, 12, 18], data race[1, 4, 5, 11, 16, 20, 22], critical race[13], harmful shared-memory access[24], race condition=-=[10, 26]-=-, or just race[2, 9, 17]. This paper explores the nature of race conditions and uncovers some previously hidden issues regarding the accuracy and complexity of dynamic race detection. We present the f...

Citation Context

...sed with different intended meanings, such as access anomaly[6-8, 12, 18], data race[1, 4, 5, 11, 16, 20, 22], critical race[13], harmful shared-memory access[24], race condition[10, 26], or just race=-=[2, 9, 17]-=-. This paper explores the nature of race conditions and uncovers some previously hidden issues regarding the accuracy and complexity of dynamic race detection. We present the following results. (1) Tw...

Citation Context

...gy has appeared, several terms have been used with different intended meanings, such as access anomaly[6-8, 12, 18], data race[1, 4, 5, 11, 16, 20, 22], critical race[13], harmful shared-memory access=-=[24]-=-, race condition[10, 26], or just race[2, 9, 17]. This paper explores the nature of race conditions and uncovers some previously hidden issues regarding the accuracy and complexity of dynamic race det...

Citation Context

...menting mutual exclusion. However, for more powerful types of synchronization (such as semaphores), conservative approximations that locate a superset of the apparent general races have been proposed =-=[10, 11]-=-. Such methods provide a means of debugging general races in the programs using such synchronization, but have the potential of misleading the programmer with potentially large numbers of spurious rac...

Citation Context

...n them has not been previously recognized. Because no consistent terminology has appeared, several terms have been used with different intended meanings, such as access anomaly[6-8, 12, 18], data race=-=[1, 4, 5, 11, 16, 20, 22]-=-, critical race[13], harmful shared-memory access[24], race condition[10, 26], or just race[2, 9, 17]. This paper explores the nature of race conditions and uncovers some previously hidden issues rega...

Citation Context

...% b) is a member of FDI~~. We have proven that deciding these membership problems is NP-hard (no matter what type of synchronization the program uses) and that locating feasible races is also NP-hard =-=[19, 21]-=-. It is therefore an intractable problem to locate precisely the race conditions exhibited by an execution of the program. This result suggests that, in practice, we must settle for an approximation, ...

Citation Context

...zation incapable of implementing mutual exclusion (such as Post/Wait style synchronization without Clear operations); all apparent races can thus be efficiently detected in executions of such programs=-=[23]-=-. It is important to note that data races are not of interest for such programs since weaker synchronization cannot implement critical sections. Exhaustively locating all apparent data races is theref...

Citation Context

...rties. In this section, we briefly overview our model for reasoning about race conditions that was first presented in an earlier paper [22] and that is based on Lamport’s theory of concurrent systems =-=[15]-=-. Our model consists of two parts: one to represent the actual behavior exhibited by the program and the other to represent potential behaviors possibly exhibited by the program. 3.1 Actual Program Ex...

Citation Context

...y over terminology that is the most descriptive. In place of data race and general race, atomicity race and determinacy race have also been suggested. references "can potentially execute concurre=-=ntly"[6] or have n-=-o "guaranteed run-time ordering"[10]. Our work is novel since we explicitly characterize two different types of race conditions using a formal model and explore their properties. The distinc...

Citation Context

...n them has not been previously recognized. Because no consistent terminology has appeared, several terms have been used with different intended meanings, such as access anomaly[6-8, 12, 18], data race=-=[1, 4, 5, 11, 16, 20, 22]-=-, critical race[13], harmful shared-memory access[24], race condition[10, 26], or just race[2, 9, 17]. This paper explores the nature of race conditions and uncovers some previously hidden issues rega...

Citation Context

...d. Because no consistent terminology has appeared, several terms have been used with different intended meanings, such as access anomaly[6-8, 12, 18], data race[1, 4, 5, 11, 16, 20, 22], critical race=-=[13]-=-, harmful shared-memory access[24], race condition[10, 26], or just race[2, 9, 17]. This paper explores the nature of race conditions and uncovers some previously hidden issues regarding the accuracy ...

Citation Context

...program bugs. For example, race conditions have only been defined as occurring when two shared-memory #################################### + To be consistent with the fault tolerant research community=-=[25]-=-, a failure occurs when a program's external behavior differs from its specification, and a fault is its algorithmic cause (although we use the term bug). # There is some controversy over terminology ...