[Soot-list] ForwardBranchedFlowAnalysis gets stuck in endless loop

[Eric Bodden]

Thanks for the detailed description, Daniel.

Do you have a little test case that exposes the problem? That would
help us checking if a possible patch really works.

Also, I would like to make sure this is really a problem of the flow
analysis framework and not a problem of some implementation of some
concrete analysis abstraction.

Best wishes,
Eric

On 28 August 2011 05:34, Daniel Popescu <dpopescu at usc.edu> wrote:
> ForwardBranchedFlowAnalysis can get stuck in an endless loop because
> ForwardBranchedFlowAnalysis does not correctly update the working list
> changedUnits.
>
> The endless loop happens whenever ForwardBranchedFlowAnalysis analyzes
> a method that contains a loop. Every time ForwardBranchedFlowAnalysis
> has finished analyzing a unit, it compares the two sets afterFlow and
> previousAfterFlow. The set afterFlow will contain the data values from
> all previously analyzed units. However, the set previousAfterFlow will
> only contain the data values from units that
> ForwardBranchedFlowAnalysis analyzed more than once.
>
> The endless loop happens because ForwardBranchedFlowAnalysis only adds
> the successors of a changed unit to the working list. Units that are
> outside of the loop are not analyzed more than once.
>
> If we would also add the changed unit itself to the working list, the
> analysis would not get stuck.
>
> Daniel
> _______________________________________________
> Soot-list mailing list
> Soot-list at sable.mcgill.ca
> http://mailman.cs.mcgill.ca/mailman/listinfo/soot-list
>



-- 
Dr. Eric Bodden, http://bodden.de/
Principal Investigator in Secure Services at CASED
Coordinator of the CASED Advisory Board of Study Affairs
PostDoc at Software Technology Group, Technische Universität Darmstadt
Tel: +49 6151 16-5478    Fax: +49 6151 16-5410
Mailing Address: S2|02 A209, Hochschulstraße 10, 64289 Darmstadt