[Soot-list] Doubt in Pointers Analysis Result
Saswat Anand
saswat at cc.gatech.edu
Fri Oct 29 19:02:31 EDT 2004
Well, the cause of the bug that I reported is in PASideEffectTester.java.
if( G.v().Union_factory == null ){
G.v().Union_factory = new UnionFactory() {
public Union newUnion() {
return FullObjectSet.v();
}
};
}
instead of 'FullObjectSet.v()', if it returns 'new
MemoryEfficientRasUnion()' the bug disappears. I dont know why? (I found
it by comparing this file with SideEffectTagger.java which did not give
that bug). I would appreciate an explanation.
Apart from that, there is another bug in PASideEffectTester, although it
does not affect correctness. In readSet, writeSet, reachingObjects, the
caching mechanism (ie strore a local copy instead of querying
sideeffect analysis repeatedly) does not work in all cases -- when
reachingObjects or RWset returned by side effect analysis is null.
Saswat
More information about the Soot-list
mailing list