[PATCH] Fix a fallout of PR14179 fix (3.3/3.4/4.0 regression)

[Jakub Jelinek]

On Mon, Dec 27, 2004 at 02:46:52PM -0800, Richard Henderson wrote:
> On Mon, Dec 27, 2004 at 10:43:07AM -0500, Jakub Jelinek wrote:
> >   /* Size types *are* sign extended.  */
> >   sign_extended_type = (!TYPE_UNSIGNED (TREE_TYPE (t))
> >                         || (TREE_CODE (TREE_TYPE (t)) == INTEGER_TYPE
> >                             && TYPE_IS_SIZETYPE (TREE_TYPE (t))));
> 
> Boggle.  I wonder what breaks, if anything if this bit
> of bogosity is removed?

I guess the original intent of
http://gcc.gnu.org/ml/gcc-patches/2000-10/msg00663.html
was to get overflows detected on signed types AND sizetype and this
sign-extended representation of sizetype was just a side-effect?

I'll see what breaks with this removed.  That would be suitable
just for HEAD if it works, right?

	Jakub