How much memory does the code the blog poster described leak? How does it respond to edge cases and invalid input? Finally, is it portable beyond one specific OS? Beyond POSIX or Windows-based systems? Those questions are especially pertinent in C.
I debugged a memcopy (taken from Linux! 10 years ago) on a RISC processor - it had 12 bugs in a dozen lines of code. Not designed to run on RISC but shows how hard it can be to get this stuff right.