software floating point & machine descriptions

[Eric Fisher]

Hi, gcc

I have to send the new mail again for the software floating point problem.
I need more details about it.  1) If I use software floating point, does I need
implement float mode insns in md file? Such as movsf, movdf. 2)How to
generate correct object files of libgcc2 of floating point operations? Such
as _floatdidf. 

Thank you very much.

Eric.