On Wed, Oct 3, 2018 at 3:46 AM Joseph Myers <jos...@codesourcery.com> wrote: > We already have -fsignaling-nans, if implementing the semantics of some > operation in a way that works with signaling NaNs is more expensive than > implementing them in a way that doesn't handle signaling NaNs correctly. > The glibc tests in question use -fsignaling-nans.
Thanks. I see that the rs6000.md port has a pattern that checks HONOR_SNANS which looks like the right way to solve this problem. I will try modifying Andrew's patch to use that and check against the glibc testsuite. Jim
