https://gcc.gnu.org/bugzilla/show_bug.cgi?id=111309
--- Comment #2 from Jakub Jelinek <jakub at gcc dot gnu.org> --- For clz/ctz/parity/popcount/clrsb/ffs, it should be quite easy to implement them, the primary questions are, what the builtin name should be (because __builtin_clz etc. is already taken for unsigned int argument (resp. int for clrsb), so we want some suffix either for all type-generic _BitInt types, or make it type-generic completely and support all integral types in there. And second question for clz/ctz is whether we should preserve the UB behavior for 0 argument, or define return value even for those. For larger _BitInt it will be implemented using loops and so will need to take 0 limbs into account anyway, so the advantage of invoking UB on it is smaller, and for users testing whether _BitInt(32768) is non-zero is already quite expensive if they wanted to have well-defined result in that case.
