https://gcc.gnu.org/bugzilla/show_bug.cgi?id=128265
--- Comment #6 from Drea Pinski <pinskia at gcc dot gnu.org> --- (In reply to Drea Pinski from comment #5) > I think there are 2 ways of handling this. > 1) just reject (delay) conditional reads from bool and non mode precision > integral types. I suspect there will be some issue with say long double on > x87 too. 2) for conditional reads from bool that delayed (and not unconditionally done), then rewrite it so there will be defined. I think I am going with 1 because it is easiest to backport. Since this has not been seen so far without a fuzzer the less of optimizations here is not going to be a big deal (I think).
