Masters Theses

Date of Award

8-2018

Degree Type

Thesis

Degree Name

Master of Science

Major

Computer Science

Major Professor

Michael R. Jantz

Committee Members

Micah Beck, James S. Plank

Abstract

Calling context is widely used in software engineering areas such as profiling, debugging and event logging. It can also enhance some dynamic analysis such as data race detection. To obtain the calling context at runtime, current approaches either perform expensive stack walking to recover contexts or instrument the application and dynamically encode the context into an integer. The current encoding schemes are either not fully precise, or have high instrumentation and detection overhead, and scalability issue for large and highly recursive applications.We propose slot-based calling context encoding (SCCE), which consists of a scalable encoding for acyclic contexts and an efficient encoding for cyclic contexts. Evaluating with CPU 2006 benchmark suite, we show that our acyclic encoding is scalable, has very low instrumentation overhead, and an acceptable detection overhead. We also show that our cyclic encoding also has lower instrumentation and detection overhead than the state-of-the-art approach by significantly reducing the number of bytes pushed and checked for cyclic contexts.

Download
Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS