Hi and thanks for your answer!
It is not that I know a better way to perform the division, but since my code only divides once it ends up a bit silly to implement the 20 instructions in that lib_fdiv_simd function using a call. Especially since this is the only call in my code. It is just much easier to have totally self-contained modules as I need to load and unload this code from external flash etc.
And earlier versions of VisualDSP did in fact inline divisions, which is why I am asking.