This is the mail archive of the
java@gcc.gnu.org
mailing list for the Java project.
Re: implementing escape analysis
Paul Biggar wrote:
> A question that remains is whether or not it will need to be implemented
> at both ends. In particular, there were questions of StringBuffer
> optimizations which seem to require the data being available at the
> front-end.
>
You will probably need to talk to the front end at some point. It may
even be the case that we need to add new langhooks and/or have it
preserve information it's now tossing away.
The actual structural analysis is simplified if done in the middle-end,
because you've got the CFG, use-def information, potential escape sites
identified, etc.
What you will need from the FE are rules regarding type system,
attributes at call sites (for special library calls) and such. I don't
think that should be too hard to get.
> Another question is whether, if it is to be done at the back-end,
> somebody else is already working on it, in gcc. Or it may be done
> before. I found this post when looking through the gcc archives:
> http://gcc.gnu.org/ml/gcc/2004-06/msg01774.html
> I can't find any evidence of what came of it though.
>
No. Nothing came out of that.