> 	I hit a snag with CodeWarrior 9 trying to compile this from the
> m68k/endian.h header file:
> 
> typedef u_int32_t	in_addr_t;
> typedef u_int16_t	in_port_t;
> 
> __BEGIN_DECLS
> in_addr_t	htonl __P((in_addr_t)) __attribute__((__const__));
> in_port_t	htons __P((in_port_t)) __attribute__((__const__));
> in_addr_t	ntohl __P((in_addr_t)) __attribute__((__const__));
> in_port_t	ntohs __P((in_port_t)) __attribute__((__const__));
> __END_DECLS

Did it snag on the __attribute__()?  If so,

#define __attribute__(x)	/* ignore me */

should be OK, as long as there's a definition of htonl somewhere else.
On the m68k, those functions are no-ops.

> 	Good idea. I think we still need a lot of the stuff from the
> MacOS (i.e. all of the environment variables that the Booter passes to
> the kernel), but once that information has been produced for a particular
> Mac setup, it could be written onto the disk for the boot loader to use.

Some of the env. variables are informational or designed for
configuration options, debugging, etc.  Things that would be nice
to have, but that are not exactly required to get booted.  Actually,
it would be a good idea to document these somewhere.  I'd like to
trim down the number of _required_ values in the booter/kernel
interface as much as we can.

-allen