Definition and Implementation of a Points-To Analysis for C-like Languages Export

@article{citeulike:3383238, abstract = {The points-to problem is the problem of determining the possible run-time targets of pointer variables and is usually considered part of the more general aliasing problem, which consists in establishing whether and when different expressions can refer to the same memory address. Aliasing information is essential to every tool that needs to reason about the semantics of programs. However, due to well-known undecidability results, for all interesting languages that admit aliasing, the exact solution of nontrivial aliasing problems is not generally computable. This work focuses on approximated solutions to this problem by presenting a store-based, flow-sensitive points-to analysis, for applications in the field of automated software verification. In contrast to software testing procedures, which heuristically check the program against a finite set of executions, the methods considered in this work are static analyses, where the computed results are valid for all the possible executions of the analyzed program. We present a simplified programming language and its execution model; then an approximated execution model is developed using the ideas of abstract interpretation theory. Finally, the soundness of the approximation is formally proved. The aim of developing a realistic points-to analysis is pursued by presenting some extensions to the initial simplified model and discussing the correctness of their formulation. This work contains original contributions to the issue of points-to analysis, as it provides a formulation of a filter operation on the points-to abstract domain and a formal proof of the soundness of the defined abstract operations: these, as far as we now, are lacking from the previous literature.}, archivePrefix = {arXiv}, author = {Soffia, Stefano}, citeulike-article-id = {3383238}, citeulike-linkout-0 = {http://arxiv.org/abs/0810.0753}, citeulike-linkout-1 = {http://arxiv.org/pdf/0810.0753}, day = {4}, eprint = {0810.0753}, keywords = {abstract-interpretation, pointer-analysis, program-analysis, shape-analysis}, month = {Oct}, posted-at = {2009-05-21 15:36:54}, priority = {0}, title = {Definition and Implementation of a Points-To Analysis for C-like Languages}, url = {http://arxiv.org/abs/0810.0753}, year = {2008} }

TY - JOUR ID - citeulike:3383238 L3 - citeulike-article-id:3383238 N2 - The points-to problem is the problem of determining the possible run-time targets of pointer variables and is usually considered part of the more general aliasing problem, which consists in establishing whether and when different expressions can refer to the same memory address. Aliasing information is essential to every tool that needs to reason about the semantics of programs. However, due to well-known undecidability results, for all interesting languages that admit aliasing, the exact solution of nontrivial aliasing problems is not generally computable. This work focuses on approximated solutions to this problem by presenting a store-based, flow-sensitive points-to analysis, for applications in the field of automated software verification. In contrast to software testing procedures, which heuristically check the program against a finite set of executions, the methods considered in this work are static analyses, where the computed results are valid for all the possible executions of the analyzed program. We present a simplified programming language and its execution model; then an approximated execution model is developed using the ideas of abstract interpretation theory. Finally, the soundness of the approximation is formally proved. The aim of developing a realistic points-to analysis is pursued by presenting some extensions to the initial simplified model and discussing the correctness of their formulation. This work contains original contributions to the issue of points-to analysis, as it provides a formulation of a filter operation on the points-to abstract domain and a formal proof of the soundness of the defined abstract operations: these, as far as we now, are lacking from the previous literature. TI - Definition and Implementation of a Points-To Analysis for C-like Languages KW - abstract-interpretation KW - pointer-analysis KW - program-analysis KW - shape-analysis AU - Soffia, Stefano PY - 2008/Oct/04/ UR - http://arxiv.org/abs/0810.0753 ER -