Oberon Community Platform Forum

Development => General => Topic started by: aubrey on January 24, 2008, 05:41:35 AM

Title: Cross Compiler Investigation Intel workplace to Sparc target
Post by: aubrey on January 24, 2008, 05:41:35 AM
I did some preliminary work to see if I can cross compile on a Windows machine.

I installed the Linz V4 environment on an XP machine, and made a directory that holds the Solaris sources from a backup CD around 1997.  This environment does have a custom Oberon.Hex file with a modification to deal with an issue with temporary file names.

I compiled Compiler and Sparc (the decoder) and the modules that they use above Oberon producing an Intel binary of the SPARC compiler.

I switched to another directory and put all the 1997 .Sym files into it, and I compiled the SPARC sources with the SPARC compiler.

Many .Sym files that have floating point numbers caused a trap.  Most .Sym files were accepted by the compiler, and no new .Sym file was produced.  The compiler emitted .Obj files.

SPARC.Decode fails with a trap on these .Obj files.
Hex.Open reveals that the values are essentially the same in the files, but bytes are exchanged in some of the numbers (Byte sex or Endian troubles)

My actual interest is in the MIPS processor, I just happened to have the SPARC files available to do the initial exploration.  I do have the OPP.Mod file for the SPARC compiler, but this file is not completely compatible with the MIPS compiler.  I had thought it might be, from the tag line in it: 

    MODULE OPP;   (* Copyright (c) ETH Z