https://gcc.gnu.org/bugzilla/show_bug.cgi?id=60181
--- Comment #10 from Marc Glisse <glisse at gcc dot gnu.org> --- Flags like -ftrapping-math can prevent gcc from folding at compile-time when the result is infinite (or maybe it always refuses to fold in that case). In your example, gcc generates a runtime call to __muldc3 in main. The behavior with -fno-trapping-math is a bit surprising. CCP folds the division but not the multiplication (not even with -ffast-math)). CPLXLOWER turns it into __muldc3, and DOM2 folds that to a (bad?) constant. So in some sense, when gcc fails to fold to a constant, then we go to the runtime behavior, and gcc can still fold that one to a constant (not as good as what it could get in the first round).
