backport r151691 and r151729 to gcc-4.4 branch

[Nathan Froyd]

On Wed, Jun 02, 2010 at 01:33:21PM +0930, Alan Modra wrote:
> On Tue, Jun 01, 2010 at 01:20:18PM -0400, David Edelsohn wrote:
> > The patch is a little larger and more invasive than I would like at
> > this stage in GCC 4.4 release cycle.  Is this really better than
> > directing users at GCC 4.5?
> 
> I think so.  Many people avoid a .0 release, and 4.4 -Os is in a sorry
> state without this patch.  The backport wasn't at all complicated.

I can't remember if the patch below makes a difference on its own or if
it needs the patch that Jakub and I wrote--does applying it make -Os
better?

2010-02-05  Nathan Froyd  <froydnj@codesourcery.com>
	
	* config/rs6000/rs6000.c (rs6000_override_options): Invert check
	for rs6000_gen_cell_microcode.

-Nathan