Efficient on-the-fly data race detection in multithreaded C++ programs

Teodorescu

2007 IEEE 13th International Symposium on High Performance Computer Architecture

2007

144

The emergence of multicore architectures will lead to an increase in the use of multithreaded applications that are prone to synchronization bugs, such as data races. Software solutions for detecting data races generally incur large overheads. Hardware support for race detection can significantly reduce that overhead. However, all existing hardware proposals for race detection are based on the happensbefore algorithm which is sensitive to thread interleaving and cannot detect races that are not exposed during the monitored run. The lockset algorithm addresses this limitation. Unfortunately, due to the challenging issues such as storing the lockset information and performing complex set operations, so far it has been implemented only in software with 10-30 times performance hit. This paper proposes the first hardware implementation (called HARD) of the lockset algorithm to exploit the race detection capability of this algorithm with minimal overhead. HARD efficiently stores lock sets in hardware bloom filters and converts the expensive set operations into fast bitwise logic operations with negligible overhead. We evaluate HARD using six SPLASH-2 applications with 60 randomly injected bugs. Our results show that HARD can detect 54 out of 60 tested bugs, 20% more than happens-before, with only 0.1-2.6% of execution overhead. We also show our hardware design is cost-effective by comparing with the ideal lockset implementation, which would require a large amount of hardware resources.

Section: False Positive Pruning For Barriersmentioning

confidence: 99%

“…Other studies [21,25,36] have proposed combinations of lockset and happens-before. The hybrid algorithm in [21] uses happens-before to reduce the false positives generated by using lockset alone, but still preserves the coverage of the lockset technique.…”

Section: Race Detectionmentioning

confidence: 99%

HARD: Hardware-Assisted Lockset-based Race Detection

Teodorescu

2007 IEEE 13th International Symposium on High Performance Computer Architecture

2007

144

Electronic Notes in Theoretical Computer Science

“…Hence, it detects data races on a subset of shared locations that are reported by the lockset approach and misses some real data races. Therefore, Multi-Race [13] tries to improve the data race detection capabilities by combining enhanced versions of Lock-set and DJIT into a common framework.…”

Section: Dynamic Methodsmentioning

confidence: 99%

Fault Detection in Multi-Threaded C++ Server Applications

Mühlenfeld

Wotawa

2007

Due to increasing demands in processing power on the one hand, but the physical limit on CPU clock speed on the other hand, multi-threaded programming is becoming more important in current applications. Unfortunately, multi-threaded programs are prone to programming mistakes that result in hard to find defects, mainly race-conditions and deadlocks. The need for tools that help finding these faults is immanent, but currently available tools are either difficult to use because of the need for annotations, unable to cope with more than a few 10 kLOC, or issue too many false warnings. This paper describes experiments with the freely available tool Helgrind and results obtained by using it for debugging a server application comprising 500 kLOC. We present improvements to the runtime analysis of C++ programs that result in a dramatic reduction of false warnings.

Journal of Parallel and Distributed Computing

“…This work was inspired by [23], which improves the performance of Lockset and Djit. However, the work in [23] could not detect witnesses for traces which were not explored by the dynamic algorithm.…”

Section: Dynamic Toolsmentioning

confidence: 99%

“…Because dataraces are a cause for concern, many works have dealt with the problem of detecting them. Most works rely on static [9,14,33] or dynamic [20,23,26] analysis to find dataraces, while others use model checking [11], or a combination of techniques [8,16] for better results.…”

Section: Introductionmentioning

confidence: 99%

Scaling model checking of dataraces using dynamic information

Shacham

Sagiv

Schuster

2007

Self Cite

Dataraces in multithreaded programs often indicate severe bugs and can cause unexpected behaviors when different thread interleavings are executed. Because dataraces are a cause for concern, many works have dealt with the problem of detecting them. Works based on dynamic techniques either report errors only for dataraces that occur in the current interleaving, which limits their usefulness, or produce many spurious dataraces. Works based on model checking search exhaustively for dataraces and thus can reveal even those that occur in rarely executed paths. However, the applicability of model checking is limited because the large number of thread interleavings in realistic multithreaded programs causes state space explosion. In this work, we combine the two techniques in a hybrid scheme which overcomes these difficulties and enjoys the advantages of both worlds. Our hybrid technique succeeds in providing thread interleavings that prove the existence of dataraces in realistic programs. The programs we experimented with cannot be checked using either an ordinary industrial strength model checker or bounded model checking.