为何memcmp实现未针对指针相等（q==p）的场景做短路优化？

I’ve definitely wondered about this too—when you pass the exact same pointer to both arguments of memcmp, it seems like a no-brainer to just return 0 immediately, right? After all, if both pointers point to the exact same memory location, every byte being compared has to match. It feels like such an obvious early-exit optimization.

But when I dug into the implementations of glibc and FreeBSD LibC, I found neither one includes this check. I also looked through the C99 open standard draft, and there’s no explicit requirement for this behavior either.

A few reasons might explain why mainstream libraries skip this optimization:

• Rare real-world use case: Most calls to memcmp are comparing distinct memory blocks. Adding a pointer equality check would introduce an extra branch that’s rarely taken, which could actually add overhead to the far more common path.
• Standard doesn’t mandate it: Since the C standard doesn’t require this optimization, library maintainers prioritize optimizing the scenarios that come up most often rather than edge cases.
• Zero-length edge case already handled: If the length parameter is 0, memcmp returns 0 regardless of pointer values anyway. For non-zero lengths, identical pointers are such an unusual scenario that the cost of adding the check outweighs the tiny benefit.

内容的提问来源于stack exchange，提问作者Frames Catherine White