Verifying observational robustness against a c11-style memory model

We study the problem of verifying robustness concurrent programs against a C11-style memory model that includes relaxed accesses and release/acquire fences, show this verification can be reduced to standard reachability under sequential consistency. further observe existing notions do not allow use speculative reads as in sequence lock mechanism, introduce novel "observational robustness" property fills gap. In turn, we how soundly check for observational robustness. have implemented our method applied it several challenging algorithms, demonstrating applicability approach. To best knowledge, is first programming language concurrency fences.