Previously we're passing llvm::Function& into M68kCCState to lower
arguments in fastcc. However, that reference might not be available if
it's a library call and we only need its argument types. Therefore,
now we're simply passing a list of argument llvm::Type-s.
This fixes PR-50752.
The origin bug will only be triggered if we're using fastcc for a libcall. However, M68k is not using
fastcc for any of the libcall, so we never really trigger the bug. Therefore it makes little sense to create
a test that calls a library function using fastcc. And the existing tests already cover cases like normal fastcc
and normal libcall.
(semi off topic) - should this be 'const M68kCCState &CCInfo' ?