[Bug target/124948] wrong code with wrapping __atomic_add_fetch() on _BitInt(17) on riscv

[law at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124948

--- Comment #4 from Jeffrey A. Law <law at gcc dot gnu.org> ---
Kito, doesn't this testcase show that the bitfield_ext_full is wrong for
risc-v.  Essnetially with that setting we will take the 17 bit field out of
memory in SI and we'll have garbage in those upper 15 bits.

It seems like bitfield_ext_none is the right settting based on this testcase,
but this is very much out of any area I have experience with.