Fortran alias analysis on tree-ssa - a proposal.
Paul Brook
paul@codesourcery.com
Mon Dec 22 16:20:00 GMT 2003
On Monday 22 December 2003 3:11 pm, Toon Moene wrote:
> Discussion with Daniel Berlin led me to believe that the following is
> the correct treatment of Fortran alias analysis on the tree-ssa branch.
I don't really know how the alias analysis works, but it sounds right to me.
There is certainly a lot of scope for improving the alias analysis on fortran
code. I haven't touched alias analysis so far, which presumably means we're
either assuming everything can alias, or using C aliasing rules.
Paul
More information about the Gcc-patches
mailing list