Context-bounded verification of thread pools

Thread pooling is a common programming idiom in which fixed set of worker threads are maintained to execute tasks concurrently. The workers repeatedly pick and them completion. Each task sequential, with possibly recursive code, communicate over shared memory. Executing can lead more new being spawned. We consider the safety verification problem for thread-pooled programs. parameterize two parameters: size thread pool as well number context switches each task. determines running how many times be swapped out while executing single task---like problems multithreaded programs, bounding important decidability. show that thread-pooled, context-bounded, Boolean programs EXPSPACE-complete, even if bound given binary. Our main result, EXPSPACE upper bound, derived using sequence succinct encoding techniques independent language-theoretic interest. In particular, we polynomial-time construction downward closures languages accepted by pushdown automata doubly nondeterministic finite automata. While there explicit exponential lower bounds on accepting closure, our result shows these compressed. significantly reduces computational power: contrast, only provided binary, but no pooling, becomes 3EXPSPACE-complete. Given high complexity related involving binary parameters, relatively low thread-pooling comes surprise.