This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: [tree-ssa] AST optimizer in C++?
- From: Chris Lattner <sabre at nondot dot org>
- To: Diego Novillo <dnovillo at redhat dot com>
- Cc: Pop Sébastian <pop at gauvain dot u-strasbg dot fr>,<gcc at gcc dot gnu dot org>
- Date: Mon, 26 Aug 2002 10:39:01 -0500 (CDT)
- Subject: Re: [tree-ssa] AST optimizer in C++?
On Sun, 25 Aug 2002, Diego Novillo wrote:
> On Sun, 25 Aug 2002, Chris Lattner wrote:
> > > The approach we use now does not interfere with the underlying
> > > IR. What it does is overlay a web of def-use/use-def pointers on
> >
> > Is there a reason that you'd like to have a SIMPLE form that is not in
> > SSA?
> Only that nobody has bothered implementing it. We'd need to
> represent PHI nodes in the IR and teach the rest of the compiler
> how to deal with them (RTL expanders, tree manipulation, etc).
> In principle it would be a rather mechanic, albeit large, patch.
In principle, you could get away from teaching all of the backends and
their RTL expanders about PHI nodes by simply simulating them by putting
an assignment in the appropriate predecessor blocks. Performing UnSSA on
a list-of-instructions representation is really quite simple and could be
performed at the same time as RTL expansion.
-Chris
http://llvm.cs.uiuc.edu/
http://www.nondot.org/~sabre/Projects/