Efficient May Happen in Parallel Analysis for Async-Finish Parallelism

Abstract

Abstract. For concurrent and parallel languages, the may-happen-inparallel (MHP) decision problem asks, given two actions in the program, if there is an execution in which they can execute in parallel. Closely related, the MHP computation problem asks, given a program, which pairs of statements may happen in parallel. MHP analysis is the basis for many program analysis problems, such as data race detection and determinism checking, and researchers have devised MHP analyses for a variety of programming models. We present algorithms for static MHP analysis of a storeless abstraction of X10-like languages that have async-finish parallelism and procedures. For a program of size n, our first algorithm solves the MHP decision problem in O(n) time, via a reduction to constrained dynamic pushdown networks (CDPNs). Our second algorithm solves the MHP computation problem in O(n · max(n, k)) time, where k is a statically determined upper bound on the number of pairs that may happen in parallel. The second algorithm first runs a type-based analysis that produces a set of candidate pairs, and then it runs the decision procedure on each of those pairs. For programs without recursion, the type-based analysis is exact and gives an output-sensitive algorithm for the MHP computation problem, while for recursive programs, the type-based analysis may produce spurious pairs that the decision procedure will then remove. Our experiments on a large suite of X10 benchmarks suggest that our approach scales well. Our experiments also show that while k is O(n 2 ) in the worst case, k is often O(n) in practice.

May-Happen-in-Parallel Analysis for Asynchronous Programs with Inter-Procedural Synchronization

“…Moreover, since these paths originate in the same node 11, and the first edges use different future variables, we conclude that (58,35) is an indirect MHP pair. Similarly, since 11 ; 38 ∈ G 1 and 11 ; 50 ∈ G 1 we conclude that (11,38) and (11,50) are direct MHP pairs. However, in this case (38,50) is not an indirect MHP pair because the first edges of these paths use the same future variable w. Indeed, the calls to g and k appear in different branches of an if statement.…”

“…2. Since 11 ; 35 ∈ G 1 and 11 ; 58 ∈ G 1 we conclude that (11,58) and (11,35) are direct MHP pairs. Moreover, since these paths originate in the same node 11, and the first edges use different future variables, we conclude that (58,35) is an indirect MHP pair.…”

“…It also includes conditional and loop constructs, however, conditions in these constructs are simply non-deterministic choices. Developing the analysis at such abstract level is convenient [11], since the actual computations are simply ignored in the analysis and what is actually tracked is the control flow that originates from asynchronously calling methods and synchronizing with their termination. Our implementation, however, is done for the full concurrent object-oriented language ABS [10] (see Sec.…”

“…Existing MHP analyses for asynchronous programs [3,11,1] lose all information when future variables are used as parameters, as they do not handle inter-procedural synchronization. As a consequence, existing analysis for more advanced properties [9,4] that rely on the MHP relations do all lose the associated analysis information on such futures.…”

Region-Based May-Happen-in-Parallel Analysis for C Programs

“…However, such flexible non-lexically-scoped parallelism poses a major challenge to scalable and precise MHP analysis. The techniques developed for structured languages [5], [6], [7], [8] cannot be directly applied to unstructured languages. In C programs, for example, a thread may outlive its spawning thread or can be joined partially along one program path (a partial join) or indirectly in one of its child threads (a nested join).…”