This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: [tree-ssa] Insert on edge comment
- From: law at redhat dot com
- To: Andrew MacLeod <amacleod at redhat dot com>
- Cc: Chris Lattner <sabre at nondot dot org>, vmakarov at redhat dot com, matz at suse dot de, gcc mailing list <gcc at gcc dot gnu dot org>
- Date: Tue, 10 Jun 2003 10:42:08 -0600
- Subject: Re: [tree-ssa] Insert on edge comment
- Reply-to: law at redhat dot com
In message <1055262334.6414.178.camel@p4>, Andrew MacLeod writes:
>We don't have the equivilent of a conditional goto do in GIMPLE do we?
>We use a COND_EXPR with a GOTO in the THEN branch right?
>
>we need a conditional goto/branch of some sort in order to turn our
>LOOP, COND equivilent non-container code with goto's.. So we need
>something like
> GOTO_IF_NONZERO (var) label
> GOTO_IF_ZERO (var) label
>
>We could then lower from containers whenever we felt the need, either at
>the beginning before loop optimizations or afterwards at the same time
>we lower ARRAY and COMPONENT refs. Current optimizations would then
>work on either GIMPLE or lowered GIMPLE, once they are taught about
>these opcodes.
>
>Does that make any sense? It doesnt seem like a lot of work...
I was referring to dropping the nested structure entirely. Fixing
GOTOs, SWITCHES or any of them piecemeal isn't terribly difficult,
but taken as a whole it gets more complex.
Jeff