Probabilistic Total Store Ordering

Abstract We present Probabilistic Total Store Ordering (PTSO) – a probabilistic extension of the classical TSO semantics. For given (finite-state) program, operational semantics PTSO induces an infinite-state Markov chain. resolve inherent non-determinism due to process schedulings and memory updates according probability distributions. provide comprehensive set results showing decidability several properties for PTSO, namely (i) Almost-Sure (Repeated) Reachability : whether run, starting from initial configuration, almost surely visits (resp. repeatedly visits) target configurations. (ii) Almost-Never run never target. (iii) Approximate Quantitative approximate, up arbitrary degree precision, measure runs that start configuration (repeatedly) visit (iv) Expected Average Cost: expected average cost derive our through nontrivial combination theory (infinite-state) chains, theories decisive eager specific techniques combinatorics, as well as, complexity (non-probabilistic) As far we know, this is first work considers verification programs running on weak models.