On Thu, Sep 8, 2022 at 9:27 AM Richard Biener
<richard.guenther@gmail.com> wrote:
>
>
>
> > Am 08.09.2022 um 08:28 schrieb Aldy Hernandez <aldyh@redhat.com>:
> >
> > This is what I have in mind for the fpclassify-like methods on the
> > current implementation.  I'll get to the removal of the tristates after
> > Cauldron.
> >
> > As you mentioned, isnormal is kinda tricky, and likely to be confusing
> > for the user.  We can revisit it if it's important.
>
> Yeah.
>
> > ?? I assume maybe_inf() does not return true for NAN ??
>
> Only maybe_nan and known_nan return true for NaN.
>
> >
> > Also, what are your thoughts on signbit?  Perhaps a method returning
> > true if we're sure, and the actual signbit result as a reference?
> >
> >    bool known_signbit (const frange &r, int &signbit);
>
> That works for me.  You could look at

ok.  settled on:
+  bool known_signbit (bool &signbit) const;

It even cleaned up the __builtin_signbit folding kludge because now we
can just bail on NANs from known_signbit until they're properly
handled.

> Tree_expr_nonnegative_p to see if that’s good enough to be used there.  Not sure if -0. is negative or if only x for which x < 0. counts as such..  But yes, access to the sign bit is useful.

We're definitely keeping track of signed zeros.  That was the whole
point of the sign bit.  I'll be removing it also, as it's a tristate.

tree_expr_nonnegative_p could definitely be made to make use of
it...especially because we have global ranges.  I'll look into it.

>
> > How does this look?  I must say, the uses look much cleaner.
>
> Yes, I like it.

Attached is the patch I'm pushing.

Tested on x86-64 Linux.

Thanks.
Aldy

>
> Richard
>
> > Aldy
> >
> > gcc/ChangeLog:
> >
> >    * gimple-range-fold.cc
> >    (fold_using_range::range_of_builtin_int_call): Use fpclassify like API.
> >    * range-op-float.cc (finite_operand_p): Same.
> >    (finite_operands_p): Same.
> >    (foperator_lt::fold_range): Same.
> >    (foperator_le::fold_range): Same.
> >    (foperator_gt::fold_range): Same.
> >    (foperator_ge::fold_range): Same.
> >    (foperator_unordered::fold_range): Same.
> >    (foperator_unordered::op1_range): Same.
> >    (foperator_ordered::fold_range): Same.
> >    * value-range.cc (frange::set_nan): Same.
> >    (frange::set_signbit): Same.
> >    (frange::union_): Same.
> >    (frange::intersect): Same.
> >    (frange::operator==): Same.
> >    (frange::singleton_p): Same.
> >    (frange::verify_range): Same.
> >    (range_tests_nan): Same.
> >    (range_tests_floats): Same.
> >    * value-range.h(frange::known_finite): New.
> >    (frange::maybe_inf): New.
> >    (frange::known_inf): New.
> >    (frange::maybe_nan): New.
> >    (frange::known_nan): New.
> > ---
> > gcc/gimple-range-fold.cc |  2 +-
> > gcc/range-op-float.cc    | 26 ++++++++-----------
> > gcc/value-range.cc       | 37 ++++++++++++++-------------
> > gcc/value-range.h        | 54 +++++++++++++++++++++++++++++++++++++++-
> > 4 files changed, 83 insertions(+), 36 deletions(-)
> >
> > diff --git a/gcc/gimple-range-fold.cc b/gcc/gimple-range-fold.cc
> > index c9c7a2ccc70..47a1f49eb36 100644
> > --- a/gcc/gimple-range-fold.cc
> > +++ b/gcc/gimple-range-fold.cc
> > @@ -1031,7 +1031,7 @@ fold_using_range::range_of_builtin_int_call (irange &r, gcall *call,
> >      {
> >        if (tmp.get_signbit ().varying_p ()
> >        // FIXME: We don't support signed NANs yet.
> > -        || !tmp.get_nan ().no_p ())
> > +        || tmp.maybe_nan ())
> >          return false;
> >        if (tmp.get_signbit ().yes_p ())
> >          r.set_nonzero (type);
> > diff --git a/gcc/range-op-float.cc b/gcc/range-op-float.cc
> > index 5fbbaa1fb36..0f928b6c098 100644
> > --- a/gcc/range-op-float.cc
> > +++ b/gcc/range-op-float.cc
> > @@ -167,7 +167,7 @@ frange_set_nan (frange &r, tree type)
> > static inline bool
> > finite_operand_p (const frange &op1)
> > {
> > -  return flag_finite_math_only || op1.get_nan ().no_p ();
> > +  return flag_finite_math_only || !op1.maybe_nan ();
> > }
> >
> > // Return TRUE if OP1 and OP2 are known to be free of NANs.
> > @@ -175,9 +175,7 @@ finite_operand_p (const frange &op1)
> > static inline bool
> > finite_operands_p (const frange &op1, const frange &op2)
> > {
> > -  return (flag_finite_math_only
> > -      || (op1.get_nan ().no_p ()
> > -          && op2.get_nan ().no_p ()));
> > +  return flag_finite_math_only || (!op1.maybe_nan () && !op2.maybe_nan ());
> > }
> >
> > // Floating version of relop_early_resolve that takes into account NAN
> > @@ -546,7 +544,7 @@ foperator_lt::fold_range (irange &r, tree type,
> >       else
> >    r = range_true_and_false (type);
> >     }
> > -  else if (op1.get_nan ().yes_p () || op2.get_nan ().yes_p ())
> > +  else if (op1.known_nan () || op2.known_nan ())
> >     r = range_false (type);
> >   else
> >     r = range_true_and_false (type);
> > @@ -648,7 +646,7 @@ foperator_le::fold_range (irange &r, tree type,
> >       else
> >    r = range_true_and_false (type);
> >     }
> > -  else if (op1.get_nan ().yes_p () || op2.get_nan ().yes_p ())
> > +  else if (op1.known_nan () || op2.known_nan ())
> >     r = range_false (type);
> >   else
> >     r = range_true_and_false (type);
> > @@ -742,7 +740,7 @@ foperator_gt::fold_range (irange &r, tree type,
> >       else
> >    r = range_true_and_false (type);
> >     }
> > -  else if (op1.get_nan ().yes_p () || op2.get_nan ().yes_p ())
> > +  else if (op1.known_nan () || op2.known_nan ())
> >     r = range_false (type);
> >   else
> >     r = range_true_and_false (type);
> > @@ -844,7 +842,7 @@ foperator_ge::fold_range (irange &r, tree type,
> >       else
> >    r = range_true_and_false (type);
> >     }
> > -  else if (op1.get_nan ().yes_p () || op2.get_nan ().yes_p ())
> > +  else if (op1.known_nan () || op2.known_nan ())
> >     r = range_false (type);
> >   else
> >     r = range_true_and_false (type);
> > @@ -927,10 +925,10 @@ foperator_unordered::fold_range (irange &r, tree type,
> >                 relation_kind) const
> > {
> >   // UNORDERED is TRUE if either operand is a NAN.
> > -  if (op1.get_nan ().yes_p () || op2.get_nan ().yes_p ())
> > +  if (op1.known_nan () || op2.known_nan ())
> >     r = range_true (type);
> >   // UNORDERED is FALSE if neither operand is a NAN.
> > -  else if (op1.get_nan ().no_p () && op2.get_nan ().no_p ())
> > +  else if (!op1.maybe_nan () && !op2.maybe_nan ())
> >     r = range_false (type);
> >   else
> >     r = range_true_and_false (type);
> > @@ -949,7 +947,7 @@ foperator_unordered::op1_range (frange &r, tree type,
> >       r.set_varying (type);
> >       // Since at least one operand must be NAN, if one of them is
> >       // not, the other must be.
> > -      if (op2.get_nan ().no_p ())
> > +      if (!op2.maybe_nan ())
> >    frange_set_nan (r, type);
> >       break;
> >
> > @@ -993,11 +991,9 @@ foperator_ordered::fold_range (irange &r, tree type,
> >                   const frange &op1, const frange &op2,
> >                   relation_kind) const
> > {
> > -  // ORDERED is TRUE if neither operand is a NAN.
> > -  if (op1.get_nan ().no_p () && op2.get_nan ().no_p ())
> > +  if (!op1.maybe_nan () && !op2.maybe_nan ())
> >     r = range_true (type);
> > -  // ORDERED is FALSE if either operand is a NAN.
> > -  else if (op1.get_nan ().yes_p () || op2.get_nan ().yes_p ())
> > +  else if (op1.known_nan () || op2.known_nan ())
> >     r = range_false (type);
> >   else
> >     r = range_true_and_false (type);
> > diff --git a/gcc/value-range.cc b/gcc/value-range.cc
> > index c3f668a811a..364919ca5c6 100644
> > --- a/gcc/value-range.cc
> > +++ b/gcc/value-range.cc
> > @@ -274,7 +274,7 @@ frange::set_nan (fp_prop::kind k)
> > {
> >   if (k == fp_prop::YES)
> >     {
> > -      if (get_nan ().no_p ())
> > +      if (!maybe_nan ())
> >    {
> >      set_undefined ();
> >      return;
> > @@ -284,7 +284,7 @@ frange::set_nan (fp_prop::kind k)
> >       return;
> >     }
> >
> > -  if (k == fp_prop::NO && get_nan ().yes_p ())
> > +  if (k == fp_prop::NO && known_nan ())
> >     {
> >       set_undefined ();
> >       return;
> > @@ -308,7 +308,7 @@ frange::set_signbit (fp_prop::kind k)
> >   gcc_checking_assert (m_type);
> >
> >   // No additional adjustments are needed for a NAN.
> > -  if (get_nan ().yes_p ())
> > +  if (known_nan ())
> >     {
> >       m_props.set_signbit (k);
> >       return;
> > @@ -467,7 +467,7 @@ frange::union_ (const vrange &v)
> >
> >   // If one side has a NAN, the union is the other side, plus the union
> >   // of the properties and the possibility of a NAN.
> > -  if (get_nan ().yes_p ())
> > +  if (known_nan ())
> >     {
> >       frange_props save = m_props;
> >       *this = r;
> > @@ -478,7 +478,7 @@ frange::union_ (const vrange &v)
> >    verify_range ();
> >       return true;
> >     }
> > -  if (r.get_nan ().yes_p ())
> > +  if (r.known_nan ())
> >     {
> >       m_props.union_ (r.m_props);
> >       set_nan (fp_prop::VARYING);
> > @@ -525,7 +525,7 @@ frange::intersect (const vrange &v)
> >
> >   // If two NANs are not exactly the same, drop to an unknown NAN,
> >   // otherwise there's nothing to do.
> > -  if (get_nan ().yes_p () && r.get_nan ().yes_p ())
> > +  if (known_nan () && r.known_nan ())
> >     {
> >       if (m_props == r.m_props)
> >    return false;
> > @@ -534,7 +534,7 @@ frange::intersect (const vrange &v)
> >       return true;
> >     }
> >   // ?? Perhaps the intersection of a NAN and anything is a NAN ??.
> > -  if (get_nan ().yes_p () || r.get_nan ().yes_p ())
> > +  if (known_nan () || r.known_nan ())
> >     {
> >       set_varying (m_type);
> >       return true;
> > @@ -590,8 +590,7 @@ frange::operator== (const frange &src) const
> >       if (varying_p ())
> >    return types_compatible_p (m_type, src.m_type);
> >
> > -      if (m_props.get_nan ().yes_p ()
> > -      || src.m_props.get_nan ().yes_p ())
> > +      if (known_nan () || src.known_nan ())
> >    return false;
> >
> >       return (real_identical (&m_min, &src.m_min)
> > @@ -644,7 +643,7 @@ frange::singleton_p (tree *result) const
> >   if (m_kind == VR_RANGE && real_identical (&m_min, &m_max))
> >     {
> >       // Return false for any singleton that may be a NAN.
> > -      if (HONOR_NANS (m_type) && !get_nan ().no_p ())
> > +      if (HONOR_NANS (m_type) && maybe_nan ())
> >    return false;
> >
> >       // Return the appropriate zero if known.
> > @@ -701,7 +700,7 @@ frange::verify_range ()
> >     {
> >       // If either is a NAN, both must be a NAN.
> >       gcc_checking_assert (real_identical (&m_min, &m_max));
> > -      gcc_checking_assert (get_nan ().yes_p ());
> > +      gcc_checking_assert (known_nan ());
> >     }
> >   else
> >     // Make sure we don't have swapped ranges.
> > @@ -710,7 +709,7 @@ frange::verify_range ()
> >   // If we're absolutely sure we have a NAN, the endpoints should
> >   // reflect this, otherwise we'd have more than one way to represent
> >   // a NAN.
> > -  if (m_props.get_nan ().yes_p ())
> > +  if (known_nan ())
> >     {
> >       gcc_checking_assert (real_isnan (&m_min));
> >       gcc_checking_assert (real_isnan (&m_max));
> > @@ -3637,7 +3636,7 @@ range_tests_nan ()
> >   ASSERT_FALSE (r0 == r0);
> >   ASSERT_TRUE (r0 != r0);
> >
> > -  // [5,6] U NAN is [5,6] with an unknown NAN bit.
> > +  // [5,6] U NAN.
> >   r0 = frange_float ("5", "6");
> >   r0.set_nan (fp_prop::NO);
> >   r1 = frange_nan (float_type_node);
> > @@ -3646,7 +3645,7 @@ range_tests_nan ()
> >   real_from_string (&r, "6");
> >   ASSERT_TRUE (real_identical (&q, &r0.lower_bound ()));
> >   ASSERT_TRUE (real_identical (&r, &r0.upper_bound ()));
> > -  ASSERT_TRUE (r0.get_nan ().varying_p ());
> > +  ASSERT_TRUE (r0.maybe_nan ());
> >
> >   // NAN U NAN = NAN
> >   r0 = frange_nan (float_type_node);
> > @@ -3654,7 +3653,7 @@ range_tests_nan ()
> >   r0.union_ (r1);
> >   ASSERT_TRUE (real_isnan (&r0.lower_bound ()));
> >   ASSERT_TRUE (real_isnan (&r1.upper_bound ()));
> > -  ASSERT_TRUE (r0.get_nan ().yes_p ());
> > +  ASSERT_TRUE (r0.known_nan ());
> >
> >   // [INF, INF] ^ NAN = VARYING
> >   r0 = frange_nan (float_type_node);
> > @@ -3666,18 +3665,18 @@ range_tests_nan ()
> >   r0 = frange_nan (float_type_node);
> >   r1 = frange_nan (float_type_node);
> >   r0.intersect (r1);
> > -  ASSERT_TRUE (r0.get_nan ().yes_p ());
> > +  ASSERT_TRUE (r0.known_nan ());
> >
> >   // VARYING ^ NAN = NAN.
> >   r0 = frange_nan (float_type_node);
> >   r1.set_varying (float_type_node);
> >   r0.intersect (r1);
> > -  ASSERT_TRUE (r0.get_nan ().yes_p ());
> > +  ASSERT_TRUE (r0.known_nan ());
> >
> >   // Setting the NAN bit to yes, forces to range to [NAN, NAN].
> >   r0.set_varying (float_type_node);
> >   r0.set_nan (fp_prop::YES);
> > -  ASSERT_TRUE (r0.get_nan ().yes_p ());
> > +  ASSERT_TRUE (r0.known_nan ());
> >   ASSERT_TRUE (real_isnan (&r0.lower_bound ()));
> >   ASSERT_TRUE (real_isnan (&r0.upper_bound ()));
> > }
> > @@ -3795,7 +3794,7 @@ range_tests_floats ()
> >   // A range of [-INF,+INF] is actually VARYING if no other properties
> >   // are set.
> >   r0 = frange_float ("-Inf", "+Inf");
> > -  if (r0.get_nan ().varying_p ())
> > +  if (r0.maybe_nan ())
> >     ASSERT_TRUE (r0.varying_p ());
> >   // ...unless it has some special property...
> >   r0.set_nan (fp_prop::NO);
> > diff --git a/gcc/value-range.h b/gcc/value-range.h
> > index 645dc76c33a..e426225eabf 100644
> > --- a/gcc/value-range.h
> > +++ b/gcc/value-range.h
> > @@ -330,6 +330,7 @@ private:
> > class frange : public vrange
> > {
> >   friend class frange_storage_slot;
> > +  friend class vrange_printer;
> > public:
> >   frange ();
> >   frange (const frange &);
> > @@ -366,12 +367,19 @@ public:
> >   const REAL_VALUE_TYPE &lower_bound () const;
> >   const REAL_VALUE_TYPE &upper_bound () const;
> >
> > +  // fpclassify like API
> > +  bool known_finite () const;
> > +  bool maybe_inf () const;
> > +  bool known_inf () const;
> > +  bool maybe_nan () const;
> > +  bool known_nan () const;
> > +
> >   // Accessors for FP properties.
> > -  fp_prop get_nan () const { return m_props.get_nan (); }
> >   void set_nan (fp_prop::kind f);
> >   fp_prop get_signbit () const { return m_props.get_signbit (); }
> >   void set_signbit (fp_prop::kind);
> > private:
> > +  fp_prop get_nan () const { return m_props.get_nan (); }
> >   void verify_range ();
> >   bool normalize_kind ();
> >
> > @@ -1187,4 +1195,48 @@ frange_nan (tree type)
> >   return frange (type, r, r);
> > }
> >
> > +// Return TRUE if range is known to be finite.
> > +
> > +inline bool
> > +frange::known_finite () const
> > +{
> > +  if (undefined_p () || varying_p () || m_kind == VR_ANTI_RANGE)
> > +    return false;
> > +  return (!real_isnan (&m_min)
> > +      && !real_isinf (&m_min)
> > +      && !real_isinf (&m_max));
> > +}
> > +
> > +// Return TRUE if range may be infinite.
> > +
> > +inline bool
> > +frange::maybe_inf () const
> > +{
> > +  if (undefined_p () || m_kind == VR_ANTI_RANGE)
> > +    return false;
> > +  if (varying_p ())
> > +    return true;
> > +  return real_isinf (&m_min) || real_isinf (&m_max);
> > +}
> > +
> > +inline bool
> > +frange::known_inf () const
> > +{
> > +  return (m_kind == VR_RANGE
> > +      && real_identical (&m_min, &m_max)
> > +      && real_isinf (&m_min));
> > +}
> > +
> > +inline bool
> > +frange::maybe_nan () const
> > +{
> > +  return !get_nan ().no_p ();
> > +}
> > +
> > +inline bool
> > +frange::known_nan () const
> > +{
> > +  return get_nan ().yes_p ();
> > +}
> > +
> > #endif // GCC_VALUE_RANGE_H
> > --
> > 2.37.1
> >
>