Prediction Of Coverage Of Expensive Concurrency Metrics Using Cheaper Metrics

Testing of concurrent programs is difficult since the scheduling non-determinism requires one to test a huge number of different thread interleavings. Moreover, a simple repetition of test executions will typically examine similar interleavings only. One popular way how to deal with this problem is to use the noise injection approach, which is, however, parametrized with many parameters whose suitable values are difficult to find. To find such values, one needs to run many experiments and use some metric to evaluate them. Measuring the achieved coverage can, however, slow down the experiments. To minimize this problem, we show that there are correlations between metrics of different cost and that one can find a suitable test and noise setting to maximize coverage under a costly metrics by experiments with a cheaper metrics.