A Container-Usage-Pattern-Based Context Debloating Approach for Object-Sensitiv...
summary
A new preanalysis called DebloaterX uses three container-usage patterns and a limited field sensitive DFA to mark container objects as context dependent or not, is implemented in the Qilin framework and released open source, and chooses 92.4% of objects as context-independent on average. Results: kobj runs 19.3x faster for k=2 and 150.2x for k=3, scales to 8 more programs at k=3, and loses under 0.2% precision while only 7.6% of objects are labeled context-dependent.
AI-written summary · about
chapters (11)
- 0:00Context debloating approach
- 1:06Object sensitivity and context
- 3:43Exponential context growth
- 5:55Context-dependent objects challenge
- 6:59DeployX technique
- 8:47Container objects in DeployX
- 10:39Usage patterns for context
- 12:27DeployX-guided analysis workflow
- 13:08Evaluation benchmarks
- 14:31Summary and future work
- 15:13Q&A