Experimental Questions
- What are the differences among the variants of inclusion based pointer analysis in terms of time and memory consumption?
- How much overhead is introduced by context sensitivity in flow sensitive analysis?
Variants of Andersen Algorithm
Executed command:
./wpa --help
./wpa -hander ../../testcase/spec2000bc/CFP2000/mesa.clang.bc
Testing subject: mesa.clang.bc (82kLoc)
Table 1
Flow Sensitive Inclusion based Analysis
Executed command:
./dvf -cxt -stat=true ../../testcase/spec2000bc/CFP2000/ammp.clang.bc 1 > /home/chengpeng/codebase/SVF/output/dvf/ammp.clang_cxt.txt 2>&1
- dfs: demand-driven flow sensitive analysis
- cxt: demand-driven context and flow sensitive analysis
Subject: ammp.clang.bc(13kLoc)
Table 2:
Table 3
Findings
- Flow insensitive pointer analysis