[Soot-list] Question regarding inter-procedural analysis

[Társis Toledo]

Dear SOOT mailing list,

We are a group of researchers at the UFPE (Universidade Federal de
Pernambuco) University in Recife, Brazil and the IT University of
Copenhagen, Denmark - and we are experimenting with adapting/transfering
"Data-Flow Analysis" to a new way of programming.

Anyways, we are interested in building on SOOT since a lot of things have
already been dealt with in sensible ways - and it provides a lot of good
infra-structure.

We have looked at the "Survival Guide" from Aarhus and played around with
the SOOT implementation.  It seems that it is *very* easy to use SOOT for
*intra*-procedural analysis (indeed, we have already used SOOT to analyze
simple programs, *intra*-procedurally).

However, it is still not clear to us /how/ to use SOOT to make
*inter*-procedural analysis.  From what we can tell, the fixed-point
iteration seems to be hard-wired to work only on one function body (as
opposed to one big Control Flow Graph produced from the entire program).

Our question is:
Is there a(n easy) way to generate one big CFG (for the entire program) and
have the fixed-point iteration run on that (using our own lattice and
transfer functions)? And if so, how?

Any insights or pointers would be appreciated.

Cheers,
/Márcio Ribeiro, Társis Toledo, and Claus Brabrand.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.cs.mcgill.ca/pipermail/soot-list/attachments/20101105/dfa132b5/attachment.html