https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88712
Bug ID: 88712
Summary: Optimization: mov edx, 0 not replaced with xor edx,
edx in this case
Product: gcc
Version: 9.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: rtl-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: matt at godbolt dot org
Target Milestone: ---
The code:
---snip
int func(int val, const int *ptr)
{
int res = val + 1234;
if (res == *ptr)
{
res = 0;
}
return res;
}
---
generates the following ASM on all version of GCC back to 4.9.x:
---
func(int, int const*):
lea eax, [rdi+1234]
mov edx, 0
cmp DWORD PTR [rsi], eax
cmove eax, edx
ret
---
The `mov edx, 0` is surprising to me. All the other compilers I tested (see
https://godbolt.org/z/Nt9pKp for more details) use the common `xor edx, edx`
(or `xor eax, eax`) idiom for zeroing edx.
Is this a missed optimization in the case of a cmov being generated, or am I
missing something subtle?
