Re: [PATCH] [PR68097] frange::set_nonnegative should not contain -NAN.

[Richard Biener <richard.guenther@gmail.com>]

On Mon, Sep 19, 2022 at 9:59 AM Aldy Hernandez <aldyh@redhat.com> wrote:
>
> ISTM that a specifically nonnegative range should not contain -NAN,
> otherwise signbit_p() would return false, because we'd be unsure of the
> sign.
>
> Do y'all agree?

what tree_expr_nonnegative_p actually means isn't 100% clear.  For REAL_CST
it actually looks at the sign-bit but we have

(simplify
 /* copysign(x,y) -> fabs(x) if y is nonnegative.  */
 (COPYSIGN_ALL @0 tree_expr_nonnegative_p@1)
 (abs @0))

is abs (@0) OK for sNaNs and -NaN/+NaN?

And we have

/* Convert abs[u] (X)  where X is nonnegative -> (X).  */
(simplify
 (abs tree_expr_nonnegative_p@0)
 @0)

where at least sNaN -> qNaN would be dropped?

And of course

(simplify
 /* signbit(x) -> 0 if x is nonnegative.  */
 (SIGNBIT tree_expr_nonnegative_p@0)
 { integer_zero_node; })

that is, is tree_expr_nonnegative_p actually tree_expr_sign or
does tree_expr_nonnegative (x) mean x >= (typeof(X)) 0
or !(x < (typeof(X))0)?

That said, 'set_nonnegative' could be interpreted to be without
NaNs?

Richard.

>         PR 68097/tree-optimization
>
> gcc/ChangeLog:
>
>         * value-range.cc (frange::set_nonnegative): Set +NAN.
>         (range_tests_signed_zeros): New test.
>         * value-range.h (frange::update_nan): New overload to set NAN sign.
> ---
>  gcc/value-range.cc |  9 +++++++++
>  gcc/value-range.h  | 14 ++++++++++++++
>  2 files changed, 23 insertions(+)
>
> diff --git a/gcc/value-range.cc b/gcc/value-range.cc
> index 67d5d7fa90f..e432ec8b525 100644
> --- a/gcc/value-range.cc
> +++ b/gcc/value-range.cc
> @@ -752,6 +752,10 @@ void
>  frange::set_nonnegative (tree type)
>  {
>    set (type, dconst0, dconstinf);
> +
> +  // Set +NAN as the only possibility.
> +  if (HONOR_NANS (type))
> +    update_nan (/*sign=*/0);
>  }
>
>  // Here we copy between any two irange's.  The ranges can be legacy or
> @@ -3800,6 +3804,11 @@ range_tests_signed_zeros ()
>    r1.update_nan ();
>    r0.intersect (r1);
>    ASSERT_TRUE (r0.known_isnan ());
> +
> +  r0.set_nonnegative (float_type_node);
> +  ASSERT_TRUE (r0.signbit_p (signbit) && !signbit);
> +  if (HONOR_NANS (float_type_node))
> +    ASSERT_TRUE (r0.maybe_isnan ());
>  }
>
>  static void
> diff --git a/gcc/value-range.h b/gcc/value-range.h
> index 3a401f3e4e2..5b261d4f46a 100644
> --- a/gcc/value-range.h
> +++ b/gcc/value-range.h
> @@ -312,6 +312,7 @@ public:
>    const REAL_VALUE_TYPE &lower_bound () const;
>    const REAL_VALUE_TYPE &upper_bound () const;
>    void update_nan ();
> +  void update_nan (bool sign);
>    void clear_nan ();
>
>    // fpclassify like API
> @@ -1098,6 +1099,19 @@ frange::update_nan ()
>      verify_range ();
>  }
>
> +// Like above, but set the sign of the NAN.
> +
> +inline void
> +frange::update_nan (bool sign)
> +{
> +  gcc_checking_assert (!undefined_p ());
> +  m_pos_nan = !sign;
> +  m_neg_nan = sign;
> +  normalize_kind ();
> +  if (flag_checking)
> +    verify_range ();
> +}
> +
>  // Clear the NAN bit and adjust the range.
>
>  inline void
> --
> 2.37.1
>