Termination and Cost Analysis of Loops with Concurrent Interleavings

By following a rely-guarantee style of reasoning, we present a novel termination analysis for concurrent programs that, in order to prove termination of a considered loop, makes the assumption that the “shared-data that is involved in the termination proof of the loop is modified a finite number of times”. In a subsequent step, it proves that this assumption holds in all code whose execution might interleave with such loop. At the core of the analysis, we use a may-happen-in-parallel analysis to restrict the set of program points whose execution can interleave with the considered loop. Interestingly, the same kind of reasoning can be applied to infer upper bounds on the number of iterations of loops with concurrent interleavings. To the best of our knowledge, this is the first method to automatically bound the cost of such kind of loops.