Profmig: A framework for flexible migration of program profiles across software versions

Offline program profiling is costly, especially when software update is frequent. In this paper, we initiate a systematic exploration in cross-version program profile migration, which tries to effectively reuse the valid part of the behavior profiles of an old version of a software for a new version. We explore the effects imposed on profile reusability by the various factors in program behaviors, profile formats, and impact analysis, and introduce ProfMig, a framework for flexible migrations of various profiles. We demonstrate the effectiveness of the techniques on migrating loop trip-count profiles and dynamic call graphs. The migration saves significant (48-67% on average) profiling time with less than 10% accuracy compromised for most programs.

[1]  Gregg Rothermel,et al.  Whole program path-based dynamic impact analysis , 2003, 25th International Conference on Software Engineering, 2003. Proceedings..

[2]  Yutao Zhong,et al.  Predicting whole-program locality through reuse distance analysis , 2003, PLDI '03.

[3]  Youfeng Wu,et al.  Efficient discovery of regular stride patterns in irregular programs and its use in compiler prefetching , 2002, PLDI '02.

[4]  Perry Cheng,et al.  The garbage collection advantage: improving program locality , 2004, OOPSLA.

[5]  Michael D. Bond,et al.  PACER: proportional detection of data races , 2010, PLDI '10.

[6]  Zheng Wang,et al.  BMAT - A Binary Matching Tool for Stale Profile Propagation , 2000, J. Instr. Level Parallelism.

[7]  David Chenho Kung,et al.  Change impact identification in object oriented software maintenance , 1994, Proceedings 1994 International Conference on Software Maintenance.

[8]  Xiangyu Zhang,et al.  Matching execution histories of program versions , 2005, ESEC/FSE-13.

[9]  Milind Kulkarni,et al.  Accelerating multicore reuse distance analysis with sampling and parallelization , 2010, 2010 19th International Conference on Parallel Architectures and Compilation Techniques (PACT).

[10]  Brad Calder,et al.  Value profiling , 1997, Proceedings of 30th Annual International Symposium on Microarchitecture.

[11]  Chris Lattner,et al.  LLVM: AN INFRASTRUCTURE FOR MULTI-STAGE OPTIMIZATION , 2000 .

[12]  Alessandro Orso,et al.  Leveraging field data for impact analysis and regression testing , 2003, ESEC/FSE-11.

[13]  Chen Ding,et al.  Locality approximation using time , 2007, POPL '07.

[14]  Shan Lu,et al.  Instrumentation and sampling strategies for cooperative concurrency bug isolation , 2010, OOPSLA.

[15]  Frank Tip,et al.  Chianti: a tool for change impact analysis of java programs , 2004, OOPSLA.

[16]  Shuvendu K. Lahiri,et al.  SYMDIFF: A Language-Agnostic Semantic Diff Tool for Imperative Programs , 2012, CAV.

[17]  Susan Horwitz,et al.  Identifying the semantic and textual differences between two versions of a program , 1990, PLDI '90.

[18]  Frank Tip,et al.  A survey of program slicing techniques , 1994, J. Program. Lang..

[19]  J. Larus.  Whole program paths , 1999, PLDI '99.

[20]  Xiaofeng Gao,et al.  Profile-guided proactive garbage collection for locality optimization , 2006, PLDI '06.