Using devices trees on X86

[Grant Likely]

On Mon, Jun 28, 2010 at 10:57 AM, Stephen Neuendorffer
<stephen.neuendorffer at xilinx.com> wrote:
> Grant,
>
> OK, here's the current set of patches...  The end goal of which is to be
> able to parse a device
> tree and generate a set of struct device objects on any architecture,
> but specifically X86.
> I can't see that much of this is anywhere near
> final form, hence the lack of signed-off-by lines.  It does compile on
> X86, although it's not
> clear that I actually get any devices discovered behind the PCI bus yet.
> I'll summarize a couple
> of issues, that you can see in somewhat concrete form in the patches:

Thanks Stephen.

> 1) no _ALIGN on x86, which you've already dealt with
> 2) config OF is currently implemented in the architecture code.  This
> should be non-architecture dependent and selected by the arches that
> need it.

Agreed.  I've been thinking the same thing.  Just haven't gotten
around to it yet.

> 3) Theres a bunch of support functions that drivers/of requires.  Some
> are just stubs and some seem to actually
> do something non-trivial (like the SMP affinity mappings).  Probably all
> of this should get stubs with weak symbols so that the arches don't have
> to do anything special, especially since I expect pushback from adding
> this architecture code "Just for device trees".

Yes.  I expect to slowly push these instances down and get rid of them
before I have a final set of core ARM device tree patches ready for
merging.

> Comments greatly appreciated, in particular if you have
> likely-to-be-easy-to-get-accepted suggestions for 3), or feel like
> carefully solving 2) in
> a way which doesn't bork the existing of-based arches.

:-)

Thanks for the patches.  I'm traveling at the moment, but I'll dig
into them when I get back to Calgary.

g.