This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] -ffinite-math-only A-A
On 8/22/07, Christian BRUEL <christian.bruel@st.com> wrote:
>
> > -Inf - -Inf is Nan, Inf - -Inf would be Inf and Inf - Inf is NaN.
> > This A-A optimisation would give 0.0 so the !HONOR_INFINITY is required
> > in all the cases.
> >
> rereading myself, Inf - -Inf is not A - A so this case doesn't exist.
>
> Geert and you are right only !HONOR_NANS is required.
>
> >>
> >> Geert is right with that only !HONOR_NANS is required).
> >>
> > !NONOR_INFINITY was already checked, I didn't add it. So maybe it was
> > missing in its wiki ? CCing him.
> >
>
> patch simpified as attached,
>
> >
> >> Richard.
> >>
> >
> > Best Regards,
> >
> > Christian
> >
>
This patch is ok with a ChangeLog, a testcase and proper bootstrap/testing and
> Index: fold-const.c
> ===================================================================
> --- fold-const.c (revision 127666)
> +++ fold-const.c (working copy)
> @@ -10149,10 +10149,7 @@
> Also note that operand_equal_p is always false if an operand
> is volatile. */
>
> - if ((! FLOAT_TYPE_P (type)
> - || (flag_unsafe_math_optimizations
> - && !HONOR_NANS (TYPE_MODE (type))
> - && !HONOR_INFINITIES (TYPE_MODE (type))))
> + if ((! FLOAT_TYPE_P (type) || !HONOR_NANS (TYPE_MODE (type)))
removal of the extra parens
> && operand_equal_p (arg0, arg1, 0))
> return fold_convert (type, integer_zero_node);
>
> Index: simplify-rtx.c
> ===================================================================
> --- simplify-rtx.c (revision 127666)
> +++ simplify-rtx.c (working copy)
> @@ -1773,10 +1773,7 @@
> -ffinite-math-only. */
> if (rtx_equal_p (trueop0, trueop1)
> && ! side_effects_p (op0)
> - && (! FLOAT_MODE_P (mode)
> - || (flag_unsafe_math_optimizations
> - && !HONOR_NANS (mode)
> - && !HONOR_INFINITIES (mode))))
> + && (! FLOAT_MODE_P (mode) || !HONOR_NANS (mode)))
> return CONST0_RTX (mode);
and consistent white-space for ! FLOAT_MODE_P and !HONOR_NANS.
Thanks,
Richard.