[LLVMdev] GSoC 2012 Ideas - Alias Analysis

[Douglas do Couto Teixeira]

Hi guys,

I'm an undergraduate computer science student and I've been working
with pointer analysis this semester under the orientation of professor
Calvin Lin, at The University of Texas at Austin. I'm interested on
helping to develop LLVM's alias analysis infrastructure. I have two
ideas for a possible proposal and I'd like to know if you have
interest on any of them.

My first idea is to improve LLVM's existing alias analysis
infrastructure. I saw on the open projects page that there are a
couple of things that need to be done on LLVM's Pointer and Alias
Analysis[1]. However, some things seem to be solved
(http://llvm.org/PR1604 and http://llvm.org/PR452) and it made me a
little confused because I don't know exactly what should be addressed
by a possible proposal.

My second idea is to implement a pointer analysis algorithm in LLVM
using Value Flow Graph, as described by the paper "Boosting the
Performance of Flow-sensitive Points-to Analysis using Value Flow"[2].

Thus, I'd like to know whether or not you think improving LLVM's alias
analysis infrastructure using one of those two suggestions would be a
reasonable thing to do as a GSoC.

[1] http://llvm.org/OpenProjects.html#pointeranalysis
[2] http://labs.oracle.com/docs/2011/150-299/2011-0189/esec053-li.pdf

Best,

Douglas