Geometric Encoding for Context Sensitive Points-to Analysis in WALA Framework

MPhil Thesis Defence


Title: "Geometric Encoding for Context Sensitive Points-to Analysis in WALA 
Framework"

By

Mr. Fan YU


Abstract

WALA framework is a static analysis framework for Java bytecodes which has 
capabilities such as data ow analysis, class hierarchy analysis, bytecode 
instrumen- tation and pointer analysis. It provides a ow-insensitive 
Anderson-style pointer analysis with on-the- y call graph construction. 
ContextSelector is the core element of WALA's pointer analysis which determines 
how call graph construction clone methods based on context. However, it does 
not provide an efficient and scalable context-sensitive pointer analysis. In 
this thesis, we propose to implement a context-sensitive pointer analysis with 
a scalable technique called geometric encoding in the WALA framework. Geometric 
Encoding is a encoding technique which can evaluate contexts of pointer 
information in a compressed form. The technique is capable of increasing the 
scalability of context-sensitive pointer analysis by greatly reducing the 
redundancy of calling contexts. The evaluation results shows that our pointer 
analysis is efficient and effective.


Date:			Monday, 27 August 2018

Time:			2:30pm - 4:30pm

Venue:			Room 3494
 			Lifts 25/26

Committee Members:	Dr. Charles Zhang (Supervisor)
 			Dr. Wei Wang (Chairperson)
 			Dr. Tao Wang


**** ALL are Welcome ****