On Sun, Oct 07, 2001 at 10:45:58PM +0200, Ignatios Souvatzis wrote:
> Hi,
> 
> as Amiga should eventually move to use the machine-independent drivers
> for siop / osiop (instead of its own amiga/dev/siop.c and amiga/dev/siop2.c),
> I've started reading them.
> 
> I notice that the siop register offset definitions only apply to little endian
> bus modes of the siop chips, e.g. PCI, while the chips in the Amiga (53c770,
> 53c710) are connected in big endian chip mode.
> 
> I wonder what to do about this - if it was only for the sake of Amiga, using
> a special bus_space that does translations should work fine. But are there
> any other non-PCI siop out there in NetBSD land, that would need a generic
> solution? Or would they do the same bus_space abusing?
> 
> (I think there are for the old siop (53x700/710), but I forgot which
> architecture is affected.) 
> 
> Short outline of such a proposal:
> 
> siop_softc gets a new field: sc_address_xor
> 
> siop_common gets two new functions, maybe defined as a macro or inline for 
> efficiency:
> 
> siop_write_byte(sc, off, value)  {
> 	bus_space_write_1(sc->tag, sc->handle, off & sc->sc_address_xor, val)
> }
> 
> siop_read_byte(sc, off)  {
> 	return bus_space_read_1(sc->tag, sc->handle, off  & sc->sc_address_xor);
> }
> 
> Note the xor with sort of a variable value, which adds a new delay. Hiding the
> operation inside bus_space access functions, where needed, will avoid this 
> delay at least for PCI.)

of course, I didn't notice that there's also a read_2 access in the code.
But this doesn't really add anything new to the problem.

Regards,
	-is