Interprocedural dataflow analysis in an executable optimizer

Abstract

Interprocedural dataflow information enables link-time and post-link-time optimizers to perform analyses and code transformations that are not possible in a traditional compiler. This paper describes the interprocedural dataflow analysis techniques used by Spike, a post-linktime optimizer for Alpha/NT executables. Spike uses dataflow analysis to su mmarize the register definitions, uses, and kills that occur external to each routine, allowing Spike to perform a variety of optimizations that require interprocedural dataflow information. Because Spike is designed to optimize large PC applications, the time required to perform interprocedural dataflow analysis could potentially be unacceptably long, limiting Spike's effectiveness and applicability.To decrease dataflow analysis time, Spike uses a compact representation of a program's intraprocedural and interprocedural control flow that efficiently summarizes the register definitions and uses that occur in the program. Experimental results are presented for the SPEC95 integer benchmarks and eight large PC applications.The results show that the compact representation allows Spike to compute interprocedural dataflow information in less than 2 seconds for each of the SPEC95 integer benchmarks. Even for the largest PC application containing over 1.7 million instructions in 340 thousand basic blocks, interprocedural dataflow analysis requires just 12 seconds.

Static program analysis of embedded executable assembly code

“…Not as much attention has been levied on analyzing machine code. Several researchers have looked at link time optimization [16,17,18,19,21], where machine code has to be analyzed, however, as pointed out by Debray et al [8], they are "limited to fairly simple local analysis." There is a wealth of literature on pointer analysis [1,2,3,22,25,26,27,28,8,5,6,7].…”

Architecture-Level Dependence Analysis for Software Systems

“…This type of analysis was originally applied to compiler optimization [2] in order to determine safe code restructuring. The development of efficient and precise dependence analysis algorithms continues as an active area of research among compiler designers (e.g., [4,9,13]). The application of dependence analysis techniques to aid program understanding and impact analysis has been widely studied (e.g., [19,21,23]).…”

Link-Time Path-Sensitive Memory Redundancy Elimination

“…Alias analysis and register liveness analysis are well-known data-flow problems already described in the literature [1,23]. However, such analyses are typically formulated in terms of source-level constructs, and do not handle features typically encountered in executable programs [17,11,24,13]. Also, LRE requires of a careful cost-benefit analysis, since we assume that the cost of inserting moves will be lower than that of the redundant load.…”

“…While a number of systems have been described for optimization of executable code [28,7,17,25], to the best of our knowledge, elimination of redundant memory operations in these systems is limited to fairly simple removal.…”

“…Optimizations performed at link time or directly applied to final program executables [28,17,7,25] have received increased attention in recent years, due to two main reasons. First, large programs tend to be compiled using separate compilation, that is, one or a few files at a time.…”