Re: [PATCH] Fix register elimination bug in reload

[Ian Lance Taylor <ian at airs dot com>]

Ranjit Mathew <rmathew@gmail.com> writes:

> Ulrich Weigand wrote:
> >   
> > ! static bool
> >   verify_initial_elim_offsets (void)
> >   {
> >     HOST_WIDE_INT t;
> >   
> > +   if (!num_eliminable)
> > +     return true;
> > + 
> >   #ifdef ELIMINABLE_REGS
> >     struct elim_table *ep;
> 
> You have introduced a statement before a
> variable declaration. It doesn't stop
> bootstrap for me, but I do see a warning
> about it. (I wonder why bootstrap didn't
> fail though - isn't -Werror the default
> these days?)

I checked in a patch for this problem last night.

Bootstrap didn't fail because of this in Makefile.in:

# These files need -Wno-error because the gimplifier triggers hard to fix
# warnings when converting to GIMPLE form.  The warnings are triggered because
# moving the condition into the loop prevents the loop optimizer from
# recognizing that the loop will always be executed at least once.  We need
# a new loop optimizer.
reload1.o-warn = -Wno-error

This has been there since May 2004.  I haven't seen those warnings
myself.  Do we still need it?

Ian