The second one is effectively a nop though. The fact that it’s xor rax, rax rather than xor eax, eax is also interesting as it’s one byte longer for exactly th

The second one is effectively a nop though. The fact that it’s xor rax, rax rather than xor eax, eax is also interesting as it’s one byte longer for exactly the same effect (modifying the bottom 32 bits of a register clears the upper 32 bits). It makes me think there’s something weird going on other than compiler stupidity. I’d be interested in seeing the code it was compiled from.