[Bug target/65847] SSE2 code for adding two structs is much worse at -O3 than at -O2

[vincent-gcc at vinc17 dot net]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65847

--- Comment #2 from Vincent Lefèvre <vincent-gcc at vinc17 dot net> ---
I've just found the same issue. The code is a bit different (here, AFAIK, this
is AVX), but I assume that the cause is the same.

With -O2:

foo:
.LFB0:
        .cfi_startproc
        vaddsd  %xmm3, %xmm1, %xmm1
        vaddsd  %xmm2, %xmm0, %xmm0
        ret
        .cfi_endproc

With -O3:

foo:
.LFB0:
        .cfi_startproc
        vmovq   %xmm0, -40(%rsp)
        vmovq   %xmm1, -32(%rsp)
        vmovapd -40(%rsp), %xmm5
        vmovq   %xmm2, -24(%rsp)
        vmovq   %xmm3, -16(%rsp)
        vaddpd  -24(%rsp), %xmm5, %xmm4
        vmovaps %xmm4, -40(%rsp)
        vmovsd  -32(%rsp), %xmm1
        vmovsd  -40(%rsp), %xmm0
        ret
        .cfi_endproc

Tested with:

gcc (Debian 20190102-1) 9.0.0 20190102 (experimental) [trunk revision 267505]