Revised patch to generate correct dwarf2 debugging info

[Jeffrey A Law]

  In message < 39C21C6E.304B2CD6@redhat.com >you write:
  > I have revised the patch to correct the generation of dwarf2 debugging
  > informartion based on the comments I received on the previous version.
  > This patch allows gcc to generate correct debugging information for
  > parameters passed to inlined functions that are constant float,
  > double, or long long values.
  > 
  > 2000-09-14  Will Cohen  <wcohen@redhat.com>
  > 
  > 	* dwarf2out.c (add_const_value_attribute): Changed array into a
  > 	xmalloced object, so it exists after function exit.
  > 	(free_AT): Added case for dw_val_class_float to free allocated
  > 	memory.
  > 
  > Is this okay to checkin?
Approved.  Please install!

Thanks,
jeff