Re: Correct ELF machine for coldfire?

[Jonathan S. Shapiro]

On Fri, 2008-02-01 at 10:40 +0000, Nick Clifton wrote:
> > The GNU tools never generate EM_COLDFIRE.  I think some non-GNU tools
> > do.
> 
> Ok, but is the EM_COLDFIRE number the correct one to use ?  (ie are the GNU 
> tools wrong ?)  It would appear so given the name, but maybe it is an 
> unofficial number.

Nick:

I suspect that this is a huge can of worms. First, it will get everyone
entangled in the "should m68k be split" discussion. Second, there will
be compatibility issues. Third, doing a split won't really benefit the
target, because the dingbats at FreeScale have made some incompatible
and conflicting choices within the Coldfire line.

Concretely, this means that splitting the architecture won't absolve the
developer of the need to specify the target CPU in any case, and from an
expressiveness perspective there just isn't that big a difference
between "its a coldfire" and "its an m68k+isa_b CPU".

Looking over our kernel, I'm not convinced (for the moment) that there
is anything CPU dependent at this level that can't be handled
sufficiently with #define's supplied from the BSP.

shap