[Soot-list] How to have whole-program analysis run from non-main() methods

[Eric Bodden]

Yes, actually that did help, thanks. It appears that the method
soot.jimple.toolkits.callgraph.CallGraphPack.internalApply() is
missing a crucial check to avoid overriding the entry points you set.

Can you try replacing...

   	        if(!options.implicit_entry()) {
	            Scene.v().setEntryPoints(EntryPoints.v().application());
	        }
	        if( options.all_reachable() ) {
	            List<SootMethod> entryPoints = new ArrayList<SootMethod>();
	            entryPoints.addAll( EntryPoints.v().all() );
	            entryPoints.addAll(
EntryPoints.v().methodsOfApplicationClasses() );
	            Scene.v().setEntryPoints( entryPoints );
	        }

...by...

        if(!Scene.v().hasCustomEntryPoints()) {
            //original code
        }

? Does this help? If yes, then I can commit this change to the repository.

Eric

On 12 December 2011 17:20, Jiaqi Tan <tanjiaqi at gmail.com> wrote:
> Hi Eric,
>
> Here is the stack trace:
> Exception in thread "main" java.lang.RuntimeException: There is no
> main class set!
>        at soot.Scene.getMainClass(Scene.java:160)
>        at soot.EntryPoints.application(EntryPoints.java:65)
>        at soot.EntryPoints.all(EntryPoints.java:95)
>        at soot.jimple.toolkits.callgraph.CallGraphPack.internalApply(CallGraphPack.java:43)
>        at soot.Pack.apply(Pack.java:114)
>        at soot.PackManager.runWholeProgramPacks(PackManager.java:408)
>        at soot.PackManager.runPacks(PackManager.java:329)
>
> Does this help?
>
> Thanks,
> Jiaqi
>
> On Mon, Dec 12, 2011 at 11:16 PM, Eric Bodden <eric.bodden at ec-spride.de> wrote:
>> Hi Jiaqi.
>>
>> It's surprising that you see "There is no main class set!". This
>> should not actually happen when custom entry points are defined. Could
>> you use a debugger to tell me in which context this message is raised,
>> i.e., in which context the method soot.Scene.getMainClass() is called?
>>
>> Cheers,
>> Eric
>>
>> On 12 December 2011 14:13, Jiaqi Tan <tanjiaqi at gmail.com> wrote:
>>> Hi Eric,
>>>
>>> This is what I am currently doing:
>>>
>>> 1. Retrieve all methods from class to use as current entry-points
>>> (begin exploring call-graph from one class of interest) using the
>>> following:
>>>
>>> Options.v().set_allow_phantom_refs(true);
>>> PhaseOptions.v().setPhaseOption("jb", "enabled:true");
>>> Scene.v().addBasicClass(className, SootClass.BODIES);
>>> PackManager.v().runPacks();
>>> SootClass c = Scene.v().loadClassAndSupport(className);
>>>
>>> 2. Set entry points:
>>> List<SootMethod> entry = c.getMethods();
>>> G.reset();
>>> Options.v().set_whole_program(true); // same as "-w"
>>> Options.v().set_allow_phantom_refs(true);
>>> Options.v().set_app(true); // set application mode
>>> Options.v().set_debug(true);
>>> PhaseOptions.v().setPhaseOption("cg.cha", "enabled:true"); // true or off
>>> Scene.v().setEntryPoints(entry);
>>> Scene.v().loadNecessaryClasses();
>>> PackManager.v().runPacks();
>>>
>>> 3a. When I run with this, I get an error message that the main-class is not set:
>>> Exception in thread "main" java.lang.RuntimeException: There is no
>>> main class set!
>>>
>>> 3b. When I run it with "Scene.v().setMainClass(clazz);" added before
>>> setting the entry-points, where clazz is the SootClass returned from
>>> above, I get this error instead:
>>> Exception in thread "main" java.lang.RuntimeException: Main-class has
>>> no main method!
>>>
>>> I'm not sure if I've missed something.
>>>
>>> Thanks!
>>> Jiaqi
>>>
>>>
>>> On Fri, Dec 9, 2011 at 5:16 PM, Eric Bodden <eric.bodden at ec-spride.de> wrote:
>>>> Hello.
>>>>
>>>> Does this thread answer your questions?
>>>> http://www.sable.mcgill.ca/pipermail/soot-list/2011-July/003799.html
>>>>
>>>> Eric
>>>>
>>>> On 8 December 2011 15:27, Jiaqi Tan <tanjiaqi at gmail.com> wrote:
>>>>> Hi,
>>>>>
>>>>> I am trying to explore the whole-program call-graph of a program
>>>>> completely by flooding it, including from non-main() functions (even
>>>>> if this will create disconnected components in the call-graph).
>>>>>
>>>>> However, it seems that the current Soot whole-program analysis only
>>>>> treats functions named as "main", or with a single String[] argument,
>>>>> as main functions.
>>>>>
>>>>> Is there any particular restriction that requires the whole-program
>>>>> analysis to begin from a main(String[]) function (i.e. it is just
>>>>> impossible to begin whole-program exploration from a non-main()
>>>>> function)? Or is it just that an arbitrary exploration was never
>>>>> developed?
>>>>>
>>>>> Help would be much appreciated :)
>>>>>
>>>>> Thanks!
>>>>> Jiaqi Tan
>>>>> _______________________________________________
>>>>> Soot-list mailing list
>>>>> Soot-list at sable.mcgill.ca
>>>>> http://mailman.cs.mcgill.ca/mailman/listinfo/soot-list
>>>>
>>>>
>>>>
>>>> --
>>>> Eric Bodden, Ph.D., http://bodden.de/
>>>> Head of Secure Software Engineering Group at EC SPRIDE
>>>> Principal Investigator in Secure Services at CASED
>>>> Tel: +49 6151 16-75422    Fax: +49 6151 16-72051
>>>> Room 3.2.14, Mornewegstr. 30, 64293 Darmstadt
>>
>>
>>
>> --
>> Eric Bodden, Ph.D., http://bodden.de/
>> Head of Secure Software Engineering Group at EC SPRIDE
>> Principal Investigator in Secure Services at CASED
>> Tel: +49 6151 16-75422    Fax: +49 6151 16-72051
>> Room 3.2.14, Mornewegstr. 30, 64293 Darmstadt



-- 
Eric Bodden, Ph.D., http://bodden.de/
Head of Secure Software Engineering Group at EC SPRIDE
Principal Investigator in Secure Services at CASED
Tel: +49 6151 16-75422    Fax: +49 6151 16-72051
Room 3.2.14, Mornewegstr. 30, 64293 Darmstadt