Scalable Dynamic Partial Order Reduction

Šimša, Jiří; Bryant, Randal E.; Gibson, Garth A.; Hickey, Jason

doi:10.1007/978-3-642-35632-2_4

Cited by 6 publications

(4 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In our study, we use a schedule limit instead of a time limit; it is worth noting that PCT and controlled random scheduling are both trivially parallelizable, and that systematic techniques can also be parallelized with additional effort [Simsa et al 2013].…”

Section: Related Workmentioning

confidence: 99%

Concurrency Testing Using Controlled Schedulers

Thomson

Donaldson

Betts

2016

ACM Trans. Parallel Comput.

View full text Add to dashboard Cite

We present an independent empirical study on concurrency testing using controlled schedulers. We have gathered 49 buggy concurrent software benchmarks, drawn from public code bases, which we call SCTBench. We applied a modified version of an existing concurrency testing tool to SCTBench, testing five controlled scheduling techniques: depth-first search, preemption bounding, delay bounding, a controlled random scheduler, and probabilistic concurrency testing (PCT). We attempt to answer several research questions, including: Which technique performs the best, in terms of bug finding ability? How effective are the two main schedule bounding techniques, preemption bounding and delay bounding, at finding bugs? What challenges are associated with applying concurrency testing techniques to existing code? Can we classify certain benchmarks as trivial or non-trivial? Overall, we found that PCT (with parameter d=3) was the most effective technique in terms of bug finding; it found all the bugs found by the other techniques, plus an additional three, and it missed only one bug. Surprisingly, we found that the "naïve" controlled random scheduler, which randomly chooses one thread to execute at each scheduling point, performs well, finding more bugs than preemption bounding and just two fewer bugs than delay bounding. Our findings confirm that delay bounding is superior to preemption bounding and schedule bounding is superior to an unbounded depth-first search. The majority of bugs in SCTBench can be exposed using a small schedule bound (1-2), supporting previous claims, although one benchmark requires 5 preemptions. We found that the need to remove nondeterminism and control all synchronization (as is required for systematic concurrency testing) can be nontrivial. There were 8 distinct programs that could not easily be included in out study, such as those that perform network and inter-process communication. We report various properties about the benchmarks tested, such as the fact that the bugs in 18 benchmarks were exposed 50% of the time when using random scheduling. We note that future work should not use the benchmarks that we classify as trivial when presenting new techniques, other than as a minimum baseline. We have made SCTBench and our tools publicly available for reproducibility and use in future work.

show abstract

Section: Related Workmentioning

confidence: 99%

Concurrency Testing Using Controlled Schedulers

Thomson

Donaldson

Betts

2016

ACM Trans. Parallel Comput.

View full text Add to dashboard Cite

show abstract

“…Stateless model checking: Analysis of programs under weak memory models, in particular C11, has been an active area of research. Several prior investigations have focused on stateless model checking (SMC) with DPOR [5] as the basis to analyze (SC) [19], [20], [4], [21], [22], weak architectural memory models such as TSO,PSO [6], [7] and Power [23]. An SMC technique for a superset of architectural memory models (including MCA model) was recently proposed in [14].…”

Section: Related Workmentioning

confidence: 99%

Dynamic Verification of C/C++11 Concurrency over Multi Copy Atomics

Singh,

Sharma,

Sharma

2021

Preprint

View full text Add to dashboard Cite

We investigate the problem of runtime analysis of C11 programs under Multi-Copy-Atomic semantics (MCA ). Under MCA , one can analyze program outcomes solely through interleaving and reordering of thread events. As a result, obtaining intuitive explanations of program outcomes becomes straightforward. Newer versions of ARM (ARMv8 and later), Alpha, and Intel's x-86 support MCA . Our tests reveal that state-of-the-art dynamic verification techniques that analyze program executions under the C11 memory model generate safety property violations that can be interpreted as false alarms under MCA semantics. Sorting the true from false violations puts an undesirable burden on the user. In this work, we provide a dynamic verification technique (MoCA) to analyze C11 program executions which are permitted under the MCA model. We design a happens-before relation and introduce coherence rules to capture precisely those C11 program executions which are allowed under the MCA model. MoCA's exploration of the state-space is based on the state-of-the-art dynamic verification algorithm, source-DPOR. Our experiments validate that MoCA captures all coherent C11 program executions, and is precise for the MCA model. 1 ARMv8 calls its model other-MCA [8]. The difference with MCA is of terminology, not semantics (as clarified by [9]).

show abstract

“…Previous work from ours, dealing with DPOR algorithms, are the papers by Kastenberg and Rensink , Bruintjes , Lauterburg et al , and Simsa et al . In the work by Kastenberg and Rensink a DPOR algorithm that relies on the use of probe sets instead of persistent or sleep sets is presented.…”

Section: Related Workmentioning

confidence: 99%

“…Our work attacks this problem by using the new notion of stable actor and providing new heuristics to select the actor to be executed, which in most cases allows determining uniquely the actor to be selected. The paper by Simsa et al improves the previous work by presenting the first distributed systematic testing of concurrent programs that scales. It parallelizes DPOR to deal with distributed and parallel programs.…”

Section: Related Workmentioning

confidence: 99%

Systematic testing of actor systems

Albert

Arenas

Gómez-Zamalloa

2018

Software Testing Verif & Rel

View full text Add to dashboard Cite

Summary Testing concurrent systems requires exploring all possible nondeterministic interleavings that the concurrent execution may have, as any of the interleavings may reveal the erroneous behavior. In testing of actor systems, we can distinguish 2 sources of nondeterminism: (1) actor selection, the order in which actors are explored, and (2) task selection, the order in which the tasks within each actor are explored. This article provides new strategies and heuristics for pruning redundant state‐exploration when testing actor systems by reducing the amount of unnecessary nondeterminism of both types. Furthermore, we extend these techniques to handle synchronization primitives that allow awaiting for the completion of an asynchronous task. We report on an implementation and experimental evaluation of the proposed techniques in SYCO, a testing tool for actor‐based concurrency.

show abstract

Scalable Dynamic Partial Order Reduction

Cited by 6 publications

References 16 publications

Concurrency Testing Using Controlled Schedulers

Concurrency Testing Using Controlled Schedulers

Dynamic Verification of C/C++11 Concurrency over Multi Copy Atomics

Systematic testing of actor systems

Contact Info

Product

Resources

About