https://gcc.gnu.org/bugzilla/show_bug.cgi?id=97156
Bug ID: 97156
Summary: Missed optimization [x86-64] tzcnt unnecessarily zeros
out destination register
Product: gcc
Version: 10.2.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: c
Assignee: unassigned at gcc dot gnu.org
Reporter: goldstein.w.n at gmail dot com
Target Milestone: ---
See: https://gcc.godbolt.org/z/Y591MW
void __attribute__((noinline))
tzncnt_not_just_return(uint64_t v) {
for (; v; ) {
uint64_t i;
i = _tzcnt_u64(v);
v &= (v - 1);
bench_do_not_optimize_out(i);
}
bench_flush_all_pending();
}
compiles to
tzncnt_not_just_return(unsigned long):
jmp .L11
.L6:
xor eax, eax
tzcnt rax, rdi
blsr rdi, rdi
.L11:
test rdi, rdi
jne .L6
ret
the "xor eax, eax" isnt needed.
