[Bug middle-end/127785] [15/16/17 Regression] !__builtin_isunordered (a, b) & (a != b) folded to LTGT raises FE_INVALID for a quiet NaN

[pinskia at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=127785

Drea Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|!__builtin_isunordered (a,  |[15/16/17 Regression]
                   |b) & (a != b) folded to     |!__builtin_isunordered (a,
                   |LTGT raises FE_INVALID for  |b) & (a != b) folded to
                   |a quiet NaN                 |LTGT raises FE_INVALID for
                   |                            |a quiet NaN
   Last reconfirmed|                            |2026-10-08
             Status|UNCONFIRMED                 |NEW
     Ever confirmed|0                           |1
   Target Milestone|---                         |15.4

--- Comment #1 from Drea Pinski <pinskia at gcc dot gnu.org> ---
I think we can just remove the pattern:
```
(simplify
 (bit_and:c (ordered @0 @1) (ne @0 @1))
 (ltgt @0 @1))
```

As it should be correctly handled by this pattern:
```
/* Optimize (a CMP b) &| (a CMP b)
   using the helper combine_comparisons function.  */
(for bitop (bit_and bit_ior bit_xor simple_comparison)
 (for cmp1 (tcc_comparison)
  (for cmp2 (tcc_comparison)
   (simplify
    (bitop (cmp1 @0 @1) (cmp2 @0 @1))
    (with rescmpcode (tcc_comparison)
      {
        tree_code rescmpcode;
        tree res;
        bool honor_nans = HONOR_NANS (@0);
        rescmpcode = combine_comparisons (bitop, cmp1, cmp2,
                                          type, honor_nans, &res);
      }
     (if (rescmpcode == INTEGER_CST)
      { res; }
      (if (TREE_CODE_CLASS (rescmpcode) == tcc_comparison)
       (rescmpcode @0 @1))))))))
```