Static detection of leaks in polymorphic containers Export

@inproceedings{citeulike:3839090, abstract = {This paper presents the first practical static analysis tool that can find memory leaks and double deletions of objects held in polymorphic containers. This is especially important since most dynamically allocated objects are stored in containers.The tool is based on the concept of object ownership: every object has one and only one owning pointer. The owning pointer holds the exclusive right and obligation to either delete the object or to transfer the obligation. This paper presents a new type system that allows different instances of a polymorphic container to hold different types of elements, and to independently own or not own their elements.Our tool is sound: it will report all potential memory leaks and multiple deletions of pointers in a program. Our system automatically identifies the container implementation routines in an application. The user provides a short specification on the container structure and ownership constraints for these routines. The system then solves for the ownership constraints flow- and context-sensitively, and reports inconsistencies in ownership constraints as potential memory leaks and double deletions.We applied our tool to a suite of five large open-source and commercial C and C++ applications totaling one million lines of code. The tool successfully identified memory leaks in these programs and found double deletions of objects that could lead to program failures or security vulnerabilities.}, address = {New York, NY, USA}, author = {Heine, David L. and Lam, Monica S.}, booktitle = {ICSE '06: Proceedings of the 28th international conference on Software engineering}, citeulike-article-id = {3839090}, citeulike-linkout-0 = {http://portal.acm.org/citation.cfm?id=1134285.1134321}, citeulike-linkout-1 = {http://dx.doi.org/10.1145/1134285.1134321}, doi = {10.1145/1134285.1134321}, isbn = {1-59593-375-1}, keywords = {deos, leak, program-analysis}, location = {Shanghai, China}, pages = {252--261}, posted-at = {2008-12-31 13:37:06}, priority = {0}, publisher = {ACM}, title = {Static detection of leaks in polymorphic containers}, url = {http://dx.doi.org/10.1145/1134285.1134321}, year = {2006} }

TY - CONF ID - citeulike:3839090 L3 - citeulike-article-id:3839090 N2 - This paper presents the first practical static analysis tool that can find memory leaks and double deletions of objects held in polymorphic containers. This is especially important since most dynamically allocated objects are stored in containers.The tool is based on the concept of object ownership: every object has one and only one owning pointer. The owning pointer holds the exclusive right and obligation to either delete the object or to transfer the obligation. This paper presents a new type system that allows different instances of a polymorphic container to hold different types of elements, and to independently own or not own their elements.Our tool is sound: it will report all potential memory leaks and multiple deletions of pointers in a program. Our system automatically identifies the container implementation routines in an application. The user provides a short specification on the container structure and ownership constraints for these routines. The system then solves for the ownership constraints flow- and context-sensitively, and reports inconsistencies in ownership constraints as potential memory leaks and double deletions.We applied our tool to a suite of five large open-source and commercial C and C++ applications totaling one million lines of code. The tool successfully identified memory leaks in these programs and found double deletions of objects that could lead to program failures or security vulnerabilities. CY - Shanghai, China SN - 1-59593-375-1 AD - New York, NY, USA EP - 261 TI - Static detection of leaks in polymorphic containers PB - ACM T2 - ICSE '06: Proceedings of the 28th international conference on Software engineering SP - 252 KW - deos KW - leak KW - program-analysis AU - Heine, David AU - Lam, Monica PY - 2006/// UR - http://dx.doi.org/10.1145/1134285.1134321 ER -