https://gcc.gnu.org/bugzilla/show_bug.cgi?id=125586
Bug ID: 125586
Summary: [x86-64] Redundant movs around mul carry-chains
Product: gcc
Version: 16.1.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: tree-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: chfast at gmail dot com
Target Milestone: ---
In the snippet coming from multi-word by word multiplication there are
redundant register movs.
using u64 = unsigned long;
using u128 = unsigned __int128;
void mul(u64 t[3], const u64 x[2], u64 y) {
const u128 p0 = (u128)x[0] * y, p1 = (u128)x[1] * y;
t[0] = (u64)p0;
unsigned long long c;
t[1] = __builtin_addcll((u64)p1, (u64)(p0 >> 64), 0, &c);
t[2] = __builtin_addcll((u64)(p1 >> 64), 0, c, &c);
}
mov rcx, rdx
mov rax, rdx
mul QWORD PTR [rsi]
mov r8, rax
mov r9, rdx
mov rax, rcx
mul QWORD PTR [rsi+8]
mov rcx, r9
mov QWORD PTR [rdi], r8
add rcx, rax
mov rax, rdx
adc rax, 0
mov QWORD PTR [rdi+8], rcx
mov QWORD PTR [rdi+16], rax
ret
https://godbolt.org/z/jPx8GGnnM
