Towards a Theoretically-Backed and Practical Framework for Selective Object-Sensitive(…)
summary
A practical selective object-sensitive pointer-analysis framework named Moon identifies which heap objects actually improve precision by reducing the problem to graph-reachability on a Pointer Flow Graph and then efficiently traversing an optimized Variable Flow Graph. Implementation on the CHIN framework evaluates 30 Java programs and selects only ~1.6%–2.9% of objects, yielding 37.2× and 382.0× speedups for 2- and 3-object sensitivity with negligible precision loss of 0.1% and 0.2%.
AI-written summary · about
chapters (9)
- 0:00Selective object-sensitive analysis
- 1:01Point analysis and context sensitivity
- 2:27Selective context sensitivity benefits
- 3:09Theoretical framework for precision objects
- 7:42Precision objects definition and approximations
- 9:05Moon tool implementation and evaluation
- 9:49Comparison with object-sensitive analysis
- 11:34Summary and conclusion
- 11:55Q&A
related talks
24mReturn of CFA: Call-Site Sensitivity Can Be Superior to Object Sensitivity Even
15mTAILCHECK: A Lightweight Heap Overflow Detection Mechanism with Page Protection and...
57mInvited Talk: Bridging Safety and Performance
24mOscar: A Practical Page-Permissions-Based Scheme...
24mMoonShine: Optimizing OS Fuzzer Seed Selection with Trace Distillation
18mA Cocktail Approach to Practical Call Graph Construction
14mSelectively-Amortized Resource Bounding
01h 09mThere's Treasure Everywhere