[Bug middle-end/106805] Undue optimisation of floating-point comparisons

[glisse at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=106805

--- Comment #2 from Marc Glisse <glisse at gcc dot gnu.org> ---
A problematic optimization pointed in the discussion:

 (simplify
  (cmp @0 REAL_CST@1)
[...]
   (if (REAL_VALUE_ISNAN (TREE_REAL_CST (@1))
        && !tree_expr_signaling_nan_p (@1)
        && !tree_expr_maybe_signaling_nan_p (@0))
    { constant_boolean_node (cmp == NE_EXPR, type); })