[uClibc] -msoft-float bug on mipsel solved

[David Wuertele]

I found a bug in the gcc bugzilla:

  http://gcc.gnu.org/bugzilla/show_bug.cgi?id=8765

This described behavior where the libgcc.a ABI doesn't match the
arguments that programs pass in FP regs.  The solution was to compile
libgcc.a with -msoft-float.  You can do this by adding -msoft-float to
the TARGET_LIBGCC2_CFLAGS variable in
toolchain_build_mipsel/gcc-3.3/gcc/config/t-linux before building the
toolchain.

Once I built a system like this, all my -msoft-float programs ran like
there was never any problem!

Dave