Installation guide

2 Installing OS-9 V2.4.5 OS-9 V2.4 on EUROCOM-17

10 Software Manual

2.5 Floating Point Support

This release provides soft- and hardware floating point instructions.

The following instructions are supported by software:

facos, fasin, fatan, fetox, fint, fintrz, flog10, flogn, fmovecr,

fsin, fcos, ftan and ftentox.

The following instructions are supported in hardware by the 68040:

fabs, fadd, fbcc, fcmp, fdbcc, fdiff, fmove, fmovem, fmul, fneg, fnop,

frestore, fsave, fscc, fsqrt, fsub, ftrapcc and ftst.

There is no support for fcosh, fsinh and ftanh.

In order to use the supported functions:

1. Add the modules fpu040 and math881 to the bootlist.

2. Make sure that fpu is included in the extension module list in the init

module.

The init module of this release includes a correct extension module list.

3. Remake the bootfile and reboot the system.

4. Compile code with option '-k=2f' to use inline floating point, or option

'-x' use the math trap handler.

RESTRICTIONS:

1. The fpu emulation requires the D50 mask or better of the MC68040,

f-line exceptions will occur if mask is too low.

2. At least edition 109 of the kernel is necessary for FPU support. This

release includes kernel edition equal or greater than 135.