Subject: Re: Is the kernel designed to return?
To: Todd Vierling <email@example.com>
From: Bill Studenmund <firstname.lastname@example.org>
Date: 01/14/2002 11:53:36
On Mon, 14 Jan 2002, Todd Vierling wrote:
> On Mon, 14 Jan 2002, Bill Studenmund wrote:
> : Also, if we want to use the Apple booters (which would be nice), we have
> : to be able to quiesce (sp?) OF (or run with it quiesced). To be able to
> : run w/o OF callbacks is most of the same problem as running on NuBus
> : boxes. So if we solve it for one, we solve it for the other. :-)
> OFW callbacks are inherently *useful* -- going on a crusade to eliminate OFW
> usage on "all" machines is not the best idea. Among other things, it's a
> very important way to get bootstrap console and device enumeration support
> on a machine with a new device tree layout.
> With that said, your implicit assumption about how a port for NuBus machines
> would be merged ("by not using OFW at all") is not quite valid. 8-)
No, it's not invalid. :-) Your assumption of what I was thinking is not
what I was thinking. Probably as I hadn't said what I was thining. :-) My
thoughts were not to remove all OF calls, but to let us work w/o OF. All
of the OF_ routines would check to see if we really have OF running. If
so, we call OF. If not, we either return an error, or fake an answer. The
latter is for when we're looking at the device tree; the kernel was handed
a fake device tree on entry, and we look in it.
So NuBus machines would always not have OF around, while pci machines
would either have or not have OF depending on boot method.