Call Graph Construction of C Code: Haven't We Solved This Problem Yet?

PhD Qualifying Examination


Title: "Call Graph Construction of C Code: Haven't We Solved This Problem Yet?"

by

Mr. Yuandao CAI


Abstract:

After decades of research, unfortunately, call graph construction is still 
either imprecise or inefficient when scaling up to the ever-growing complexity 
of C-based modern software. For example, recent work shows that the 
state-of-the-art field-, flow-, and context-sensitive demand-driven pointer 
analysis is incapable of constructing a call graph for the GCC project (135 
thousand lines of code) with a 64 GB memory budget. In this survey, we 
investigate and evaluate two common approaches to call graph construction: type 
analyses and pointer analyses. Our results show that existing approaches suffer 
severely from a scalability-precision trade-off, creating significant obstacles 
to adoption in practice. We expect that our survey can shed light on the 
problems behind existing approaches, motivating future research to mitigate 
them.


Date:			Monday, 12 December 2022

Time:                  	3:00pm - 5:00pm

Venue:                  Room 5566
                         Lifts 27/28

Committee Members:	Dr. Charles Zhang (Supervisor)
  			Prof. Huamin Qu (Chairperson)
 			Dr. Wei Wang
 			Prof. Ke Yi


**** ALL are Welcome ****