What does --entry take in CUDA's PTX JIT compiler?

Question

NVIDIA's CUDA offers a PTX compilation library. One of the supported JIT compilation options for PTX code using the library is

--entry entry,... (-e)

which the documentation describes as:

Specify the entry functions for which code must be generated.

Entry function names for this option must be specified in the mangled name.

How do you "specify in the mangled name"? Is this telling us we need to specify mangled names, or does it mean something else?

Answer 1

This sentence:

must be specified in the mangled name.

should have read "must be specified in the mangled form (in which they appear in the PTX source)".

_{I'm planning on exposing this functionality in my API wrappers and will make sure to properly reflect this in method/member name(s) and Doyxgen comments.}