expanding addrs for initializers

[Richard Guenther]

On 9/19/05, DJ Delorie <dj@redhat.com> wrote:
>
> > Alternately, are we sure that we're not in fact catching bugs?
>
> Like I said, I'm surprised it works for the test case in the PR
> (23799).

I want to start the clock for reverting of (part of) this patch.  There is
really no reason to make this an internal_error, a warning should be
sufficient.

Please take action on fixing the fallout you caused.

Thanks,
Richard.