Instantiate virtual args in function-call usage

[Jeffrey A Law]

 In message <20001231153802.A26613@redhat.com>you write:
  > On Sat, Dec 30, 2000 at 06:46:10AM -0200, Alexandre Oliva wrote:
  > > I have noted that not all REG_NOTES are updated either :-(
  > 
  > Really?  Example?  Looks to me like they are all handled at the
  > end of eliminate_regs_in_insn.
Really :-)  I've got a case of the same problem on the PA.

jeff