Making DidFail Succeed: Enhancing the CERT Static Taint Analyzer for Android App Sets
March 2015 • Technical Report
Jonathan Burket, Lori Flynn, Will Klieber, Jonathan Lim, Wei Shen, William Snavely
In this report, the authors describe how the DidFail tool was enhanced to improve its effectiveness.
Software Engineering Institute
CMU/SEI Report Number
DOI (Digital Object Identifier):10.1184/R1/6575201.v1
This report describes recent significant enhancements to DidFail (Droid Intent Data Flow Analysis for Information Leakage), the CERT static taint analyzer for sets of Android apps. In addition to improving the analyzer itself, the enhancements include a new testing framework, new test apps, and test results. A framework for testing the DidFail analyzer, including a setup for cloud-based testing was developed and instrumented to measure performance. Cloud-based testing enables the parallel use of powerful, commercially available virtual machines to speed up testing. DidFail was also modified to use the most current version of FlowDroid and Soot, increasing its success rate from 18% to 68% on our test set of real-world apps. Analytical features were added for more types of components and shared static fields and new apps developed to test these features. The improved DidFail analyzer and the cloud-based testing framework were used to test the new apps and additional apps from the Google Play store.