"CheckDroid: A Tool for Automated Detection of Bad Practices in Android Applications using Taint Analysis" by S. Yovine, G. Winniczuk MobileSoft'17, Buenos Aires, Argentina, 2017.

1. CheckDroid: A Tool for
Automated Detection of Bad
Practices in Android
Applications using Taint Analysis
S. Yovine, G. Winniczuk
CONICET-Universidad de Buenos Aires
syovine@dc.uba.ar, gonzalonet@gmail.com
S. Yovine, G. Winniczuk MOBILESoft 2017 May 22, 2017 1 / 8

2. Recommended practices
S. Yovine, G. Winniczuk MOBILESoft 2017 May 22, 2017 2 / 8

3. Guidelines are often not respected
S. Yovine, G. Winniczuk MOBILESoft 2017 May 22, 2017 3 / 8

4. Non-respect of guidelines ...
... could produce
bad user
experiences
and
unexpected faults
S. Yovine, G. Winniczuk MOBILESoft 2017 May 22, 2017 4 / 8

5. Categories of guidelines
Performance
Long running tasks should execute in worker threads
Memory
References to objects associated with a Context
should not be stored in static variables
User interface
UI objects must not be manipulated by a worker thread
S. Yovine, G. Winniczuk MOBILESoft 2017 May 22, 2017 5 / 8

6. Original idea
Map a guideline into a path in the code
Use taint analysis to seek the path
S. Yovine, G. Winniczuk MOBILESoft 2017 May 22, 2017 6 / 8

7. CheckDroid
.apk .xml
InstrumentationSoot aapt
instrumented .apk origins/targets
AnalysisFlowDroid
result
CHECKDROID ARCHITECTURE
S. Yovine, G. Winniczuk MOBILESoft 2017 May 22, 2017 7 / 8

8. Checkdroid experience report
18 applications developed by undergrads
32 occurrences of bad practices
50% of the applications involved
26 violations concerned performance
6 involved memory recommendations
3 non-respected guidelines
found in BA Subte app
S. Yovine, G. Winniczuk MOBILESoft 2017 May 22, 2017 8 / 8