[Bug tree-optimization/86367] FRE1 tree pass deletes code in gcc.target/powerpc/nan128-1.c when long double is IEEE 128

[wschmidt at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=86367

--- Comment #8 from Bill Schmidt <wschmidt at gcc dot gnu.org> ---
That makes sense -- we already have a NaN rather than an SNaN by the time we
hit the Ealias pass.