Citation Context

...4, 45], deterministic replay support [24, 41], etc. As mentioned in Section 1, previous works have also investigated hardware and coherence protocol supports for data race detection. Examples include =-=[23, 28, 32]-=-, almost all of which implement the happens-before race detection algorithm by combining with cache coherence protocols in distributed shared memory systems. Most recently, ReEnact [31] employs advanc...

Citation Context

...r pattern in the Collections class, the developers did not override the containsAll method to make it thread-safe. 6. Related Work A large body of research focuses on dynamic or static race detection =-=[41, 35, 21, 43, 10, 14]-=-. Type based techniques [20, 5, 6], which require programmer annotations, have been used to reduce the race detection problem to a type checking problem. Since annotation writing creates significant o...

Citation Context

...ization addressed. Some works address the fork-join synchronization model [8, 21, 29]. Others focus on lock-based synchronization (two-party synchronization) [7, 35, 36], barrier-only synchronization =-=[33]-=-, or both two-party and multi-party synchronization primitives [16, 30]. Post/Wait synchronization is addressed in [13, 24]. An additional criterion is the method used to activate the data race detect...

Citation Context

... programs, both in academic circles and in industry. Progress has been made in many domains and it seems that a high quality solution must contain components from many of them. Work on race detection =-=[47]-=- [48] [33] [41] [21] [1] [2] has been going on for a long time. Race detection tools suffers from the dual problem of having too many false alarms as well as not identifying some of the races. A diffe...

Citation Context

...and PathExpander [13]. Other researchers have proposed architectures that assist in low-overhead collection of execution traces to be used for postmortem bug detection [34, 18, 35, 17]. A few studies =-=[15, 24, 29, 27, 26]-=- have looked at hardware support for race detection. Interestingly, all proposals focused on variations of the happens-before algorithm. Min and Choi [15] use cache coherence protocol events to filter...

Citation Context

...tic detection of dataraces in multithreaded programs can be extremely valuable. As a result, there has been a substantial amount of past work in building tools for analysis and detection of dataraces =-=[19, 16, 29, 21, 27, 34, 42, 28, 37, 20, 35]-=-. Dyamic datarace detectio tools detect dataraces at runtime for a specific program execution [19, 16, 29, 21, 27, 35]. Static datarace aalysis tools, on the other hand, consider the space of all poss...

Citation Context

...er, this property is difficult to prove because all possible runs of a program must be taken into account. A more common approach is to test the program with a tool that detects data races at runtime =-=[3, 4, 6, 8, 10, 16, 19, 20, 21]-=-. If no data races are found in test runs, the program is assumed to be properly labeled. Unfortunately, a synchronization defect may be manifested through data races only under conditions that are no...

Citation Context

...consistency systems, false detections may occur (see discussion in [1]). Similarly, pinpointing the exact location of code where the first data race occured becomes more difficult. Richards and Larus =-=[20]-=- used a fine-grain dsm to get access information and track accesses between barriers. Since they support only barrier synchronization, only some of the data races can be detected online. The others ar...