Proposed resolution:

This wording is relative to N5054.

1. Modify [c.math.abs] as indicated:

   constexpr floating-point-type abs(floating-point-type x);
   

   -4- ReturnsEffects: The absolute value of `x` Equivalent to: `return fabs(x);`.

   See also: ISO/IEC 9899:2024, 7.12.8.3, 7.24.7.1

[c.math.abs] states that std::abs(floating-point-type x):

Returns: The absolute value of `x`.

What is the absolute value of NaN? What happens when passing signaling NaN? What about infinity? Does it produce positive zero when given negative zero? What about floating-point exceptions? When is it a constant expression? Nobody knows.

In practice, all major standard libraries forward to `std::fabs`.