Towards automatic debugging of computer programs

Hiralal Agrawal, Purdue University

Abstract

Programmers spend considerable time debugging code. Symbolic debuggers provide some help but the task still remains complex and difficult. Other than breakpoints and tracing, these tools provide little high level help. Programmers must perform many tasks manually that the tools could perform automatically, such as finding which statements in the program affect the value of an output variable under a given testcase, what was the value of a given variable when the control last reached a given program location, and what does the program do differently under one testcase that it does not do under another. If the debugging tools provided explicit support for such tasks, the whole debugging process would be automated to a large extent. In this dissertation, we propose a new debugging paradigm that easily lends itself to automation. Two tasks in this paradigm translate into techniques called dynamic program slicing and execution backtracking. We discuss what these techniques are and how they can be automated. We present ways to obtain accurate dynamic slices of programs that may involve unconstrained pointers and composite variables. Dynamic slicing algorithms spanning a range of time-space-accuracy trade-offs are presented. We also propose ways in which multiple dynamic slices may be combined to provide further fault localization information. A new space-efficient approach to execution backtracking called "structured backtracking" is also proposed. Our experiment with the above techniques has also resulted in development of a prototype tool, S scPYDER, that explicitely supports them.

Degree

Ph.D.

Advisors

Spafford, Purdue University.

Subject Area

Computer science

COinS