Region-based memory management (RBMM)is a form of compile-time memory management,well-known from the functional programming world.In this paper we describe our work on implementing RBMMfor the logic programming language Mercury.One interesting point about Mercury is that it is designed withstrong type, mode, and determinism systems.These systems not only provideMercury programmers with several direct software engineering benefits,such as self-documenting code and clear program logic,but also give language implementorsa lot of information that is useful for program analyses.In this work, we make use of this information to develop program analysesthat determine the distribution of data into regionsand transform Mercury programs by inserting into themthe necessary region operations.We prove the correctness of our program analyses and transformation.To execute the annotated programs, we have implemented runtime supportthat tackles the two main challenges posed by backtracking.First, backtracking can require regionsremoved during forward execution to be ``resurrected'';and second,any memory allocated during a computation that has been backtracked overmust be recovered promptly and without waitingfor the regions involved to come to the end of their life.We describe in detail our solution of both these problems.We study in detail how our RBMM system performson a selection of benchmark programs,including some well-known difficult cases for RBMM.Even with these difficult cases, our RBMM-enabled Mercury systemobtains clearly faster runtimes for 15 out of 18 benchmarkscompared to the base Mercury system with its Boehm runtime garbage collector,with an average runtime speedup of 24%,and an average reduction in memory requirements of 95%.In fact, our system achieves optimal memory consumption in some programs.
展开▼
