On Tue, Apr 03, 2007 at 07:06:29PM -0700, Mark A. Greer wrote:

On Sat, Mar 24, 2007 at 10:37:12AM +1100, David Gibson wrote:

quoted

On Fri, Mar 23, 2007 at 10:30:56AM -0500, Scott Wood wrote:

quoted

On Fri, Mar 23, 2007 at 04:54:42PM +1100, David Gibson wrote:

quoted

I don't think you should need a vmlinux_alloc for this platform.

Without it, the kernel has to fit in 4MiB.  As others pointed out, that
can be too small if an initramfs is used.  Unlike 8xx (which was what
caused me to stick the kernel at 0 in previous patches), 83xx platforms
should have plenty of memory above the wrapper.

quoted

Or we could get the wrapper script to base the link/load address on
the vmlinux size.

That might be handy for some but it won't work for everyone.  Some
platforms require a specific link/load address.  We can provide what
you said as a default but allow it to be overridden by platform code
somehow.

Sure.  The wrapper already does platform dependent things, so I was
assuming it would only do this for some platforms.

quoted

quoted

quoted

The more complicated reason is that looking ahead to when we're using
libfdt instead of flatdevtree.c, we may need an extra step that
prepares the device tree for read/write access (with libfdt, an
fdt_open_into()).  That won't happen until after platform_init(), but
will be before .fixups().

OK.

Actually, to be clearer here: this reason actually degenerates to the
first; I want the "ft_open" step to go after console_open so we can
get error messages from it.

The problem is,  the console device is specified in the dt so you have
to ft_open before you console_open.

Sorry, I should clarify.  In the context of libfdt, I was thinking
*read* access to the fdt can happen at any point - right from
platform_init().  ft_open() would mark the point at which write access
to the tree can begin.

-- 
David Gibson			| I'll have my music baroque, and my code
david AT gibson.dropbear.id.au	| minimalist, thank you.  NOT _the_ _other_
				| _way_ _around_!
http://www.ozlabs.org/~dgibson