This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Inserting a GIMPLE node in a SSA form representation pass
- From: Zdenek Dvorak <rakdver at atrey dot karlin dot mff dot cuni dot cz>
- To: seanatpurdue at hotmail dot com
- Cc: Diego Novillo <dnovillo at redhat dot com>, gcc at gcc dot gnu dot org, dberlin at dberlin dot org
- Date: Tue, 18 Jul 2006 16:31:16 +0200
- Subject: Re: Inserting a GIMPLE node in a SSA form representation pass
- References: <44BCED1C.6050202@dberlin.org>
Hello,
> Diego Novillo wrote:
> > sean yang wrote on 07/17/06 22:16:
> >
> >> Can someone give a brief introduction? Hopefully my question is clear:)
> >>
> > When you insert a brand new GIMPLE statement, the symbols referenced in
> > it must be rewritten into SSA form. See the section 'Preserving the SSA
> > form' in the GCC internals manual.
> >
>
> Also, IIRC, tree-ssa-lim ends up rewriting those statements into SSA
> later on.
yes, in move_computations. But note that this is not the recommended way --
in fact, it is in my TODO list of things to fix (looong way down, after
many more important problems). In your case, you just need to ensure
that the arguments of the call_expr are ssa names (or constants).
Zdenek