Stacking-based Context-sensitive Points-to Analysis for Java

• Item type: Trans(1)
• 公開日: 2010-03-16
• タイトル: Stacking-based Context-sensitive Points-to Analysis for Java
• タイトル / 言語: en
• タイトル: Stacking-based Context-sensitive Points-to Analysis for Java
• 言語: eng
• キーワード / 主題Scheme: Other
• キーワード / 主題: 発表概要
• 資源タイプ / 資源タイプ識別子: http://purl.org/coar/resource_type/c_6501
• 資源タイプ: journal article
• 著者所属: Japan Advanced Institute of Science and Technology
• 著者所属: Japan Advanced Institute of Science and Technology
• 著者所属(英): en
• 著者所属(英): Japan Advanced Institute of Science and Technology
• 著者所属(英): en
• 著者所属(英): Japan Advanced Institute of Science and Technology
• 著者名: Xin, Li; Mizuhito, Ogawa
• 著者名(英): Xin, Li; Mizuhito, Ogawa
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: A points-to analysis is a prerequiste to context-sensitive program analyses for Java, in which difficulty lies in mutual dependency between a call graph generation and a points-to analysis result. Existing practical context-sensitive points-to analyses are mostly cloning-based, which have an inherent limit to handle recursive procedure calls and are hard to scale under deep cloning. We present a stacking-based context-sensitive points-to analysis for Java, by encoding the analysis as weighted pushdown model checking. Sound abstraction and proper modeling of a Java program are proposed. Further, localizing each analysis at an iteration on a partial program brings a substantial acceleration, which finally leads to an incremental algorithm. Our empirical study shows that, the incremental analysis scales well to Java benchmarks of significant size, whereas some of them cannot be handled by non-incremental ones.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: A points-to analysis is a prerequiste to context-sensitive program analyses for Java, in which difficulty lies in mutual dependency between a call graph generation and a points-to analysis result. Existing practical context-sensitive points-to analyses are mostly cloning-based, which have an inherent limit to handle recursive procedure calls and are hard to scale under deep cloning. We present a stacking-based context-sensitive points-to analysis for Java, by encoding the analysis as weighted pushdown model checking. Sound abstraction and proper modeling of a Java program are proposed. Further, localizing each analysis at an iteration on a partial program brings a substantial acceleration, which finally leads to an incremental algorithm. Our empirical study shows that, the incremental analysis scales well to Java benchmarks of significant size, whereas some of them cannot be handled by non-incremental ones.
• 書誌レコードID / 収録物識別子タイプ: NCID
• 書誌レコードID / 収録物識別子: AA11464814
• 書誌情報: 情報処理学会論文誌プログラミング（PRO）; 巻 3, 号 2, p. 55-55, 発行日 2010-03-16
• ISSN / 収録物識別子タイプ: ISSN
• ISSN / 収録物識別子: 1882-7802
• 出版者 / 言語: ja
• 出版者: 情報処理学会