[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Another problem with egcs-1.0.2-prerelease


>I just checked and egcs 1.0.2 defines both __mcf5200__ and __mc68000__ when
>you specify -m5200 at the command line. I'd recommend you use __mcf5200__
>and not m5200 to identify this processor core.
Hmm.. I'm sure I checked __mc68000__...must have missed something. I've
finally had a look at Eric's patches for m68k.h and I'll modify my code to
match..I think my home mail program ate the attachment or something...
Looking at the number of instances of #ifdef m5200 in my code it's perhaps
time I did a bit of a rewrite and used a new #define M68K_IS_COLDFIRE_ARCH
rather than some explicit processor reference which was/is dumb.

>I think if this MAC register is part of the machine context then it's
>likely the gcc code generator will need to know about it too, and so
>support for the 5300 needs to go into gcc before you can to cross any
>RTEMS fences.

I hope it doesn't get supported in gcc...getting solid ColdFire support has
proved a significant problem. The MAC instructions will need to be added to
binutils though but I think I'll let somebody with a chip deal with that

Thanks for the pointers,


David Fiddes, CALM Software Production Officer
Department of Mathematics, Heriot-Watt University, Edinburgh
email D.J.Fiddes at hw.ac.uk - Tel: +44 131-451-3251