[Bug tree-optimization/128306] `a - a` for real types where a is not a NaN can be assumed to be >= 0

[jsm28 at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=128306

--- Comment #2 from Joseph S. Myers <jsm28 at gcc dot gnu.org> ---
Inf - Inf is NaN with "invalid" raised.

In FE_DOWNWARD rounding mode, a - a is -0.0 for finite a.