I delted my last post because I thought I had it figured out.... Now it turns out I am more confused...

rules.mak includes this line:

rules.mk:TARGET_SOFT_FLOAT:=-msoft-float

but compiles don't show soft-float being used anywhere...

./staging_dir_mipsel/bin/mipsel-linux-gcc -dumpspecs

doesn't show soft-float as a default option...

So it seems that soft-float is not being used.  Is there a reason for this?  Should I try to recompile the whole toolchain for soft-float?

Thanks,

--Yan