[4.5] Better support for widening multiplies

[Steven Bosscher]

On Wed, Feb 18, 2009 at 11:38 AM, Bernd Schmidt <bernds_cb1@t-online.de> wrote:
> Steven Bosscher wrote:
>> If it's the right thing in the wrong place, then IMO there is always a
>> point to try and fit it in the right place. Doing the right thing in
>> the wrong place is how things like regmove and reload came to be.
>
> I still fail to see how you can seriously argue that fwprop is the wrong
> place.  If you care about see.c, feel free to fix it.

Because in fwprop we always propagated only from single-DEF. You
don't. So you are doing something completely different from what
fwprop is doing right now. I am not arguing that fwprop is the wrong
place.  I am arguing that it may not be the right place. It is sad and
unfortunate (but typical) that you don't even take my suggestion
seriously to look at see.c and see if it can easily be extended to do
what you want to do. But that's the last what I'll say about this.
Feel free to do whatever you think is right.

Gr.
Steven