This version is no longer maintained A new and more efficient version of Boomerang can be found here.
Boomerang is a demand-driven context and flow-sensitive pointer analysis for Java. It is built on top of Soot and Heros.
The key novelties of Boomerang are the query format and the client-driven context-resolution.
Query format: Boomerang delivers rich query information: For a given variable and statement, Boomerang returns the points-to set and additionally all pointers visible in the current method that point to the allocation sites in the points-to set. Pointers in Boomerang are abstracted as access graphs. Access graphs are used to support field accesses.
Client-driven context-resolution: Each query of Boomerang can be limited to a context dictated by the client - the analysis Boomerang is integrated into. Instead of merging points-to information across all possible call sites of a method, Boomerang allows to filter context such that the queries output is with respect to a given calling context. This improves precision when integrating into a context-sensitive analysis, such as a taint or typestate analysis.
This project is an eclipse project and can be imported into your workspace. It depends on the git repositories Soot, Heros and Jasmin. These are included as git submodules to ease the synchronisation process and avoid compile time errors. To clone a compilable version all submodulues must be available as well. This is taken care of by the use of the --recursive option of git clone.
git clone --recursive git@github.com:uasys/boomerang.git
After the checkout, the root folder of Boomerang will contain a libs folder with the three submodule repositories. Each of them is an eclipse project. Also import these projects as existing Java projects into your eclipse workspace. Boomerang then has all it dependencies and is ready to be used.
We have prepared a couple of examples on how to use Boomerang. They can be found here.
Boomerang is released under LGPL - see LICENSE.txt for details.
Boomerang has been developed by Johannes Späth, Lisa Nguyen Quang Do, Karim Ali and Eric Bodden.