[Bug tree-optimization/71842] Optimize FMA when some arguments are simple constants

[joseph at codesourcery dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71842

--- Comment #2 from joseph at codesourcery dot com <joseph at codesourcery dot com> ---
On Mon, 11 Jul 2016, glisse at gcc dot gnu.org wrote:

> double f(double x, double y){ return fma(x, y, 0.); }

Would require -fno-signed-zeroes to be valid to optimize to x*y.

> double g(double x, double y){ return fma(0., x, y); }

Would require -ffinite-math-only -fno-signed-zeroes to be valid to 
optimize to y.

Also the cases that might overflow / underflow / involve domain errors 
from infinities have issues with errno setting, though GCC currently 
ignores the possibility of fma setting errno.