SANER 2017

2017 IEEE 24th International Conference on Software Analysis, Evolution, and Reengineering (SANER), February 20-24, 2017, Klagenfurt, Austria

Desktop Layout

Patterns and Optimization
Early Research Achievements
Cloud-Based Parallel Concolic Execution
Ting Chen, Youzheng Feng, Xiapu Luo, Xiaodong Lin, and Xiaosong Zhang
(University of Electronic Science and Technology of China, China; Hong Kong Polytechnic University, China; University of Ontario Institute of Technology, Canada)
Abstract: Path explosion is one of the biggest challenges hindering the wide application of concolic execution. Although several parallel approaches have been proposed to accelerate concolic execution, they neither scale well nor properly handle resource fluctuations and node failures, which often happen in practice. In this paper, we propose a novel approach, named PACCI, which parallelizes concolic execution and adapts to the drastic changes of computing resources by leveraging cloud infrastructures. PACCI tailors concolic execution to the MapReduce programming model and takes into account the features of cloud infrastructures. In particular, we tackle several challenging issues, such as making the exploration of different program paths independently and constructing an extensible path exploration module to support the prioritization of test inputs from a global perspective. Preliminary experimental results show that PACCI is scalable (e.g., gaining about 20X speedup using 24 nodes) and its efficiency declines slightly about 5% and 6.1% under resource fluctuations and node failures, respectively.

Authors:


Time stamp: 2020-02-28T22:24:00+01:00