摘要 |
A method for race detection in parallel software, including performing an execution of software under test (SUT), the SUT having an expected outcome. The method further includes detecting a race condition in the SUT responsive to a first interleaving generated during the execution, replaying the first interleaving to a start-point of the race condition, and imposing a second interleaving at the start-point of the race condition. The method also includes determining a second-execution outcome of the SUT responsive to the second interleaving, and comparing the expected outcome with the second-execution outcome.
|