[Bug middle-end/118070] wrong code with _BitInt() in a bitfield vs scalar_storage_order attribute

[ebotcazou at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=118070

--- Comment #8 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
> But the rest still applies, we'd need to define what the other endianity
> _BitInt ABI is for each target supporting _BitInt.  And given that different
> architectures can have different _BitInt layout ABIs, SSO wouldn't really
> achieve data exchangeability between those.

What leeway do you have on the layout of _BitInts though?  Are the constraints
coming from straightforward equivalences for power-of-2 lengths not sufficient
to prescribe one (almost) entirely, if you put aside PDP-11 nonsense?