Hi,

I'm submitting this patch for approval to commit.

fp.cxx and fp.h have existed in the sid/component/cgen-cpu directory for 
a while, but as far as I can tell, they have not been used. I'm working 
on an internal port which requires simulation of accurate floating point 
support, and so, I've cleaned up these files and created two new source 
files, cgen-fpu.cxx and cgen-fpu.h, in sid/component/cgen-cpu which add 
support for the semantic code which CGEN already generates to the 
cgen-cpu components.

The two new sources are analogous to sim/common/cgen-accfp.c and 
implement the fpu as a virtual class defining operations on objects of 
type f as defined in fp.cxx and fp.h. Not all operations have been 
implemented (only the ones I needed), but the remaining operations have 
at least been prototyped and implementation should be straight forward 
as the need for them arises (similar to cgen-ops.h).

The changes to fp.cxx and fp.h were mainly to virtualize the class in 
order to allow for implementation-defined and non-standard behaviors 
(tested by my internal port). Some clean up was also done and several 
bugs were fixed.

Dave