[Bug rtl-optimization/123070] [16 Regression] pr90178.c -m32 starts to fail since r16-5975 (make_forwarders_with_degenerate_phis )

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=123070

--- Comment #16 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Andrew Pinski from comment #15)
> Created attachment 63115 [details]
> Another testcase
> 
> This one shows on aarch64 we have an extra move for some reason (even
> without my change) but not on x86_64.

I take that back about the extra move not on x86_64. It is there, I must have
been looking into something different. Anyways this shows it is not related to
return/argument passing. That is the main thing I wanted to figure out first.

There does seems like some overlapping live range now but I am still trying to
figure out why it was not before.