[Soot-list] setEntryPoints and CallGraph

[Ondrej Lhotak]

On Thu, Jul 28, 2005 at 03:11:35PM -0400, Saswat Anand wrote:
> public class Test12
> {
>     public Test12(){}
>     public static void main(String[] arg)
>     { System.out.println( "hello world" ); }
> }
> 
> Output:
>         this := @this: Test12
>         specialinvoke this.<java.lang.Object: void <init>()>()
>         return
> 
>         arg := @parameter0: java.lang.String[]
>         $r0 = <java.lang.System: java.io.PrintStream out>
>                 CLINIT edge: $r0 = <java.lang.System: java.io.PrintStream
> out> in <Test12: void main(java.lang.String[])> ==> <java.lang.System:
> void <clinit>()>
>                 CLINIT edge: $r0 = <java.lang.System: java.io.PrintStream
> out> in <Test12: void main(java.lang.String[])> ==> <java.lang.Object:
> void <clinit>()>
>         virtualinvoke $r0.<java.io.PrintStream: void
> println(java.lang.String)>("hello world")
>         return
> 
> I have two questions about the above output:
> 1- Why there is no edge out of println stmt? (commenting the above
> setEntryPoints hack does produce a edge out of this stmt)

The field System.out starts out null. It is initialized to something
non-null by the entry point
java.lang.System: void initializeSystemClass()
which you have removed. So, the points-to analysis finds that $r0 can
only be null, and invoking println on null will not invoke any method.

> 2- Why there is no edge out of Object.<init> stmt? (commenting hack does
> not change this).

The Test12 constructor (in which the call occurs) is unreachable in the
call graph, so no call edges are constructed for it.

Ondrej

> 
> Could anybody explain to me.
> 
> Thanks
> Saswat
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 			-Saswat Anand (saswat at cc.gatech.edu)
> _______________________________________________
> Soot-list mailing list
> Soot-list at sable.mcgill.ca
> http://mailman.cs.mcgill.ca/mailman/listinfo/soot-list
>