CVS-19981209: Patch for "unsigned HOST_WIDE_INT" on Solaris2.6

[Jeffrey A Law]

  In message < 199812172225.RAA06021@caip.rutgers.edu >you write:
  > 	Paul, I included a patch below to move the HOST_WIDE_INT
  > definition from cccp.c to system.h and change the macro defined to be
  > HOST_WIDEST_INT instead.  Using this, I think the next step is for you
  > to change the affected uses of HOST_WIDE_INT in cccp.c to HOST_WIDEST_INT. 
Agreed.  Having Paul identify the places where we need HOST_WIDEST_INT is
definitely better than us guessing :-)

  > 	Once that is done, we can use hwint.h everywhere and nuke
  > HOST_WIDE_INT definitions from machmode.h and other places it appears. 
Agreed.

  > 	BTW, Dave, why does cpplib.h include machmode.h and use the
  > HOST_WIDE_INT definition from there?  Doesn't libcpp.a need the possibly
  > larger intmax_t form to evaluate long long quantities (same as cccp.c?)
Sounds like a bug to me.  As far as I know, cpplib & cccp should behavior in
the same manner as far as this issue is concerned

jeff