https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87744
--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
(In reply to Jonathan Wakely from comment #1)
> Created attachment 57345 [details]
> Use 64-bit integers to do 128-bit arithmetic
>
> This patch defines a custom type that implements the necessary 128-bit
> arithmetic for linear_congruential_engine<uint64_t, ...> without __int128.
>
> It's an order of magnitude slower than using __int128 natively on x86_64,
> but it gives the right results.
if (__builtin_uaddll_overflow(__l._M_lo, __c, &__l._M_lo))
__l._M_hi++;
Why not just
__l._M_hi += __builtin_uaddll_overflow(__l._M_lo, __c, &__l._M_lo);
and similarly for subtraction?
Is the reason for using the clang compat builtins instead of
__builtin_{add,sub,mul}_overflow the compatibility with clang versions which
don't
support these?
