Automatic Determination of May/Must Set Usage in Data-Flow Analysis

Abstract

Data-flow analysis is a common technique to gather program information for use in transformations such as register allocation, dead-code elimination, common subexpression elimination, scheduling, and others. Tools for generating data-flow analysis implementations remove the need for implementers to explicitly write code that iterates over statements in a program, but still require them to implement details regarding the effects of aliasing, side effects, arrays, and user-defined structures. This paper presents the DFAGen Tool, which generates implementations for locally separable (e.g. bit-vector) data-flow analyses that are pointer, side-effect, and aggregate cognizant from an analysis specification that assumes only scalars. Analysis specifications are typically seven lines long and similar to those in standard compiler textbooks. The main contribution of this work is the automatic determination of may and must set usage within automatically generated data-flow analysis implementations.

May/must analysis and the DFAGen data-flow analysis generator

“…We first introduced the may/must analysis and the DFAGen tool in [16]. This paper advances the previous research by (1) extending the class of analyses specifiable in DFAGen to include nonseparable analyses, (2) introducing a method for retargeting DFAGen's output to different compiler infrastructures, and (3) describing how may/ must analysis could be extended if new operators were added to the analysis specification language.…”

“…May/ must analysis is one of the main contributions of this research. Our prior paper [16] introduced the concept of may/must analysis for locally separable analysis using a strict transfer function format. We extend that previous work by presenting may/must analysis as a table driven algorithm, showing how may/must analysis can be extended for new operations including examples of how this is done for conditional operators.…”

Final Report for Colorado State University

“…As part of the OpenAnalysis project, we developed a Data-Flow Analysis Generator tool (DFAGen), which enables analysis writers to generate analyses for separable and nonseparable data-flow analyses that are pointer, aggregate, and side-effect cognizant from a specification that assumes only scalars [12,13]. By hiding the compiler-specific details behind predefined set definitions, the analysis specifications for the DFAGen tool are typically less than ten lines long and similar to those in standard compiler textbooks.…”

“…a public domain suite of software for program analysis (i.e. OpenAnalysis, UseOA-ROSE, UseOA-Open64) that is currently being used in software tools being developed at Argonne, 2. the development of linearity analysis for use in improving the performance of the adjoint code generated by automatic differentiation tools, 3. a domain-specific programming language based on set building notation for specifying dataflow analyses at a high-level of abstraction with a compiler that generates the full data-flow analysis implementation, 4. theory and a prototype implementation for performing data-flow program analysis of MPI programs, 5. and the development of a parallel, network protocol simulation framework (which turns out to be similar to data-flow analysis frameworks) in collaboration with networking researchers at CSU.The above research contributions have resulted in 10 conference and workshop publications [17,16,4,14,6,8,9,12,13,5], 2 journal papers [19,11], a master's thesis [10], a conference paper still in submission [2], and a journal paper under preparation [15]. At Colorado State University, this project supported one graduate student for four years, provided some summer salary for the PI, and provided some hourly support for an undergraduate research assistant.…”

“…The above research contributions have resulted in 10 conference and workshop publications [17,16,4,14,6,8,9,12,13,5], 2 journal papers [19,11], a master's thesis [10], a conference paper still in submission [2], and a journal paper under preparation [15]. At Colorado State University, this project supported one graduate student for four years, provided some summer salary for the PI, and provided some hourly support for an undergraduate research assistant.…”