Re: [tree-ssa] RFC: Making control flow more explicit

[Jason Merrill <jason at redhat dot com>]

On Thu, 07 Aug 2003 15:43:23 -0400, Diego Novillo <dnovillo@redhat.com> wrote:

> On Thu, 2003-08-07 at 14:45, Jason Merrill wrote:
>
>> MWHIRL lowers all control flow to gotos, possibly conditional or computed.
>> This seems to be what you're looking for.  We probably want to define a
>> lower GIMPLE to match, and do the lowering somewhere in the middle of
>> optimization.
>> 
> Right.  And this is where we are currently having conflicts.  We are
> essentially doing ORC's OPT phase in HWHIRL.  The effect of exposing
> more control flow would take us to MWHIRL and build the SSA form there. 
> We are now building the SSA form in HWHIRL.

Is that a problem?  What would be the benefit of delaying the move to SSA
form?

> We probably do want to define a lower GIMPLE form.  Syntactically it
> would be no different.  It would only accept fewer tree nodes.  Perhaps
> we could also code the IL level in the trees to know what level we're
> at?

I don't think that makes sense; replacing a leaf node could invalidate such
flag on a containing node.  Rather, I think we should have global state
indicating what level we're at, and enforce the relevant restrictions.

Jason