https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123094
--- Comment #14 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Robin Dapp from comment #13)
> (In reply to Andrew Pinski from comment #12)
> > Will psuh the testcase tomorrow. and yes, the fix is r16-7941.
>
> Interesting.
Actually it does make sense based on comment #7.
```
(insn 21 20 22 6 (set (reg:V2DI 32 v0)
(const_vector:V2DI [
(const_int 1 [0x1])
(const_int 0 [0])
])) "t.c":22:3 1338 {*aarch64_simd_movv2di}
(nil))
...
(insn 54 53 55 6 (set (subreg:DI (reg:OI 128 [ _4 ]) 0)
(reg:DI 32 v0)) "t.c":24:10 104 {*movdi_aarch64}
(nil))
```
As mentioned in r16-7941 we deal with the vec_select of 0 as the same as lower
part. Here we have the "lowerpart" of the hard register.
>
> > Robin,
> > What do you think about backporting r16-7941 to GCC 15?
>
> Sure, will do that. I'll take that as an OK for the backport.
Yes
