call graph for __encodeFloat,__encodeDouble

I have been sponging off the current GMP-interface with regard to  
__encode/__decode Float/Double in rts/StgPrimFloat.c so this has not  
come up before: how is __encodeFloat/Double integrated into Haskell?   
There is no reference to it in rts/PrimOps.cmm, no '#include' for any  
of the Haskell files (say, libraries/base/GHC/Num.lhs), and no  
references in compiler/CodeGen/CgPrimOp.hs or compiler/NativeGen/*.   
This is the only function I have not been able to follow all the way  
through so if any of you have an idea of where it goes, I'd be  
grateful for some pointers.

Thanks,
Pete