Verifying embedded software functionality: fault localization, metrics and directed testing
EETimes (8/26/2012 3:16 PM EDT)
Editors Note: In the third in a four part series Abhik Roychoudhury, author of Embedded Systems and software validation, discusses the pros and cons of metric base fault localization and directed testing for assessing software functionality.
So far, in Part 1 and Part 2 in this series, we have presented the dynamic slicing method, which is fully formal and requires examination of the control/data dependencies in an execution trace.
But, the difficulties in using it include (a) time and space overheads for storing/analyzing program traces and (b) potentially large slice sizes. In the preceding, we examined methods to deal with the second problem - comprehension of large slices.
However, we still have to grapple with the time and space overheads of dynamic slicing. As observed earlier, state-of-the-art dynamic slicing tools employ various tricks such as online compaction of the execution trace and program dependence analysis on the compact trace (without decompressing it).
Nevertheless, the time and space overheads for large real-life programs is still substantial, and the quest for lightweight methods remains. In this part in the series, we will discuss a class of such lightweight methods. In the following we use the terms execution trace and execution run interchangeably. Indeed, the existing literature on software debugging also uses these two terms interchangeably. Before proceeding any further, let us first give an illustrative example.
Click here to read more ...