[Soot-list] Exceptions in IFDS Analysis

[Marc-André Laverdière-Papineau]

Hello,

This is something that Eric helped me understand, because it takes a 
good understanding of Soot's implementation to see the point.

Excluded classes have no real impact unless you activate that option. If 
you do turn it on, the excluded classes' methods have empty bodies. So 
that takes less memory, and any IFDS analysis ends up returning quickly 
to the application code. Otherwise, you'll spend a ton of time analyzing 
the JDK.

So, unless that is what you are trying to do, the option makes a lot of 
sense.

Marc-André Laverdière-Papineau
Doctorant - PhD Candidate

On 13-04-05 06:04 PM, Henddher Pedroza wrote:
> Thanks Marc-Andre.
>
> No, I am not using that option. Should I?
>
> Is soot analyzing and loading all classes and their respective methods?
> (I thought all java* and other "standard" packages will be excluded "automatically")
>
> I am using "-w", "-pp" and "-process-dir"
>
> - Henddher
>
> On Apr 5, 2013, at 4:58 PM, Marc-André Laverdière-Papineau <marc-andre.laverdiere-papineau at polymtl.ca> wrote:
>
>> Hello,
>>
>> This is weird.
>> For such a small project, I don't see why you'd need that much memory.
>> Are you using -no-bodies-for-excluded?
>>
>> I used to have analyses that blew up my memory until I turned that
>> option on.
>>
>> Marc-André Laverdière-Papineau
>> Doctorant - PhD Candidate
>>
>> On 13-04-05 04:44 PM, Henddher Pedroza wrote:
>>> I was getting strange exceptions also, and after letting it run pass the exceptions, it ended up with OOE.
>>> I bumped up the heap size to 4000M and it seems to be ok now.
>>>
>>> -Xmx4000m
>>>
>>> I am actually running IDE Main + IFDSUninitializedVariables on a dummy stand-alone java app with 5 classes (~5K uncompressed)
>>>
>>> But, I don't know :(
>>>
>>> On Apr 5, 2013, at 3:35 PM, Marc-André Laverdière-Papineau <marc-andre.laverdiere-papineau at polymtl.ca> wrote:
>>>
>>>> Hello,
>>>>
>>>> I have been getting this problem only recently, and I am not sure what I
>>>> am doing wrong.
>>>>
>>>> I am getting exceptions like these:
>>>>
>>>> Exception in thread "pool-2-thread-1" java.lang.NullPointerException
>>>> 	at
>>>> com.google.common.base.Preconditions.checkNotNull(Preconditions.java:191)
>>>> 	at com.google.common.cache.LocalCache.get(LocalCache.java:3989)
>>>> 	at com.google.common.cache.LocalCache.getOrLoad(LocalCache.java:3994)
>>>> 	at
>>>> com.google.common.cache.LocalCache$LocalLoadingCache.get(LocalCache.java:4878)
>>>> 	at
>>>> com.google.common.cache.LocalCache$LocalLoadingCache.getUnchecked(LocalCache.java:4884)
>>>> 	at
>>>> soot.jimple.toolkits.ide.icfg.JimpleBasedInterproceduralCFG.getOrCreateUnitGraph(JimpleBasedInterproceduralCFG.java:174)
>>>> 	at
>>>> soot.jimple.toolkits.ide.icfg.JimpleBasedInterproceduralCFG.isExitStmt(JimpleBasedInterproceduralCFG.java:199)
>>>>
>>>> and
>>>>
>>>> Exception in thread "main" java.lang.NullPointerException
>>>> 	at
>>>> soot.jimple.toolkits.ide.icfg.JimpleBasedInterproceduralCFG.getMethodOf(JimpleBasedInterproceduralCFG.java:163)
>>>> 	at
>>>> soot.jimple.toolkits.ide.icfg.JimpleBasedInterproceduralCFG.getMethodOf(JimpleBasedInterproceduralCFG.java:65)
>>>> 	at heros.solver.IDESolver.setVal(IDESolver.java:576)
>>>>
>>>> Digging a bit, I am seeing that unitToOwner returns null in this chunk
>>>> of code, when u is an IdentityStmt.
>>>>
>>>> 	@Override
>>>> 	public boolean isExitStmt(Unit u) {
>>>> 		Body body = unitToOwner.get(u);
>>>> 		DirectedGraph<Unit> unitGraph = getOrCreateUnitGraph(body);
>>>> 		return unitGraph.getTails().contains(u);
>>>> 	}
>>>>
>>>> Even when I breakpoint and look down in the trace, I am unable to
>>>> determine where it is coming from, or why is it that a given method is
>>>> analyzed or not. You can say that troubleshooting is hard :(
>>>>
>>>> Has anybody else ever dealt with this? How did you fix it?
>>>>
>>>> P.S. I am running with some exclusions and -no-bodies-for-excluded
>>>> P.P.S I tried feeding all the units in the application classes in
>>>> unitToOwner and I'm still hitting this problem.
>>>>
>>>> Regards,
>>>>
>>>> --
>>>> Marc-André Laverdière-Papineau
>>>> Doctorant - PhD Candidate
>>>> _______________________________________________
>>>> Soot-list mailing list
>>>> Soot-list at sable.mcgill.ca
>>>> http://mailman.cs.mcgill.ca/mailman/listinfo/soot-list
>>>
>>>
>> _______________________________________________
>> Soot-list mailing list
>> Soot-list at sable.mcgill.ca
>> http://mailman.cs.mcgill.ca/mailman/listinfo/soot-list
>
>