Efficient statistical debugging via hierarchical instrumentation.

ABSTRACT Debugging is known to be a notoriously painstaking and time-consuming task. As one major family of automated debugging, statistical debugging approaches have been well investigated over the past decade to assist in debugging. All these approaches instrument the entire buggy program to produce execution profiles for debugging. Consequently, they often incur hefty instrumentation and analysis cost. However, as in fact major part of the program code is error-free, full-scale program instrumentation is wasteful and unnecessary. In this doctoral research, a novel hierarchical instrumentation (HI) technique is devised to perform selective instrumentation so as to make statistical debugging more efficient while upholding the debugging effectiveness. We apply HI to two different categories of statistical debugging: in-house and cooperative debugging. The experiments validate that HI can greatly improve the efficiency of debugging.