How to compile hand-made PTX source?

Hi everybody,
I’m newby, apologize if my question is very simple, but I fail to find answer in manuals.

I have CUDA kernel written in PTX ans calling program in C.
What is NVCC command line to get executable?

I understand, that PTX should be “fatbined” into C source and then linked with C calling program.
But how?
And what about .entry name conventions?

Ubuntu, SDK 2.1

(Just now I’ve disable OPENCC with --opencc-options -### , so, my hand-made .PTX is taken, but it looks like dirty hack ;) )

Who knows, please, show compile line or Makefile rules for assembler CUDA programming.