This is still wrong -- we should say "an argument that is a null pointer value". We don't care how the null pointer value was written in the source code. But maybe fixing that is a lost cause. Thoughts?

PTAL, will upload screenshot shortly.

New screenshots:

@zygoloid But neither NULL nor nullptr is a null pointer value. The parameter has a null pointer value because it is initialized by a null-pointer-constant argument, but the argument need not be a null pointer value.

@timsong-cpp: yeah... I'm not sure. Even the wording "an mbstate_t* argument" is odd. It suggests that the argument already has pointer type. Meh, maybe that's good enough.