On Fri, 2008-05-02 at 19:35 -0500, Matthew Woehlke wrote: >Uh... it sounds like what you really want is to build with msoft-float? > > ...from The Fine Manpage: > -msoft-float > Do not use floating-point coprocessor instructions. > Implement floating-point calculations using library calls > instead. > > This changes the calling convention and way of parameter passing comapred to using the hardfloat abi and breaks binaries we need to support but to which we have no source. that's why the wish to tackle this problem at the gcc internals levels. Sivan
