https://gcc.gnu.org/bugzilla/show_bug.cgi?id=127562
Bug ID: 127562
Summary: [Missed optimization] SLP fails to coalesce adjacent
member movs
Product: gcc
Version: 16.2.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c++
Assignee: unassigned at gcc dot gnu.org
Reporter: ofekshilon at gmail dot com
Target Milestone: ---
Missed optimization I see in assignments and copy ctors:
```cpp
struct IntWrapper { int t; };
struct S { IntWrapper a; IntWrapper b; };
// SLP succeeds
void f1(S& s1, const S& s2 ) {
s1 = s2;
}
// SLP fails
void f2(S& s1, const S& s2 ) {
s1.a = s2.a;
s1.b = s2.b;
}
// SLP succeeds again
void f3(S& s1, const S& s2 ) {
s1.a.t = s2.a.t;
s1.b.t = s2.b.t;
}
```
In the 'succeeds' case both mov's are unified to a single one:
mov rax, QWORD PTR [rsi]
mov QWORD PTR [rdi], rax
In the 'fails' cases they aren't:
mov eax, DWORD PTR [rsi]
mov DWORD PTR [rdi], eax
mov eax, DWORD PTR [rsi+4]
mov DWORD PTR [rdi+4], eax
Godbolt link: https://compiler-explorer.com/z/b8fazYn1s
