Whelp, this about wraps it up for memcpy_s(). It looks like we need to go back to the drawing board and find a new secure function to copy memory in C.
The C programming language that run almost all of the world's compute infrastructure has both the feature and the flaw that it doesn't double-check memory. Thus, bugs in one part of the code can overwrite memory used in other parts.
It's a necessary feature for some "systems" programming, but is also dangerous for other tasks. Language like Rust solve this by being inherently "safe" for most accesses, allowing "unsafe" access only when needed.
Microsoft has addressed this by replacing 'memcpy()' (memory copies) with a safer version 'memcpy_s()' that double-checks where memory is being copied into.
It does vastly improve safety, but at the same time, it only solve specific problems. Here, it was a stupid replacement that shouldn't have been done.
Normally, memcpy_s() guards against run over the end of the buffer. Here, the problem is figuring out where the buffer starts, the "offset" from the start of the buffer.
What you really want is a function that checks the starting offset hasn't gone past the end of the buffer as well, like:
memcpy_offset_s(buf, offset, max, src, length).
memcpy_offset_s(buf, offset, max, src, length).
