This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [RTL, ColdFire 24/63] Add support for a MODE_INDEX_REG_CLASS macro
On Tue, 2007-02-27 at 16:50 +0000, Richard Sandiford wrote:
> Jeffrey Law <law@redhat.com> writes:
> > On Thu, 2007-02-22 at 19:08 -0800, Mark Mitchell wrote:
> >> I'm not quite sure if I know exactly whose court the ball is in at this
> >> point. If I understand correctly, Jeff is still unhappy with the patch,
> >> and Richard still thinks it's the right thing.
> > It's in my court. Current todo is to look at what Bernd did and why
> > and ponder how it plays with what Richard wants to do.
> >
> > I'd also like to see (and this is a new request) some code which
> > triggers these problems. I have a couple theories that I'd like to
> > poke at, but without triggering code that's going to be tough to do.
>
> OK. I can't remember off-hand which testcases failed without this,
> so here's a testcase specifically for it:
>
> --------------------------------------------------------------------
> void foo (float *x);
> float bar (void)
> {
> float x[0x5000];
> foo (x);
> return x[0x4000] + x[0];
> }
> --------------------------------------------------------------------
>
> Compile with -O2 -fomit-frame-pointer -mcpu=5485.
That fails nicely. Thanks.
Jeff