Re: pci and pcie device-tree binding - range No cells

On Wed, Dec 12, 2012 at 10:49 AM, Grant Likely wrote:

On Wed, Dec 12, 2012 at 10:37 AM, Michal Simek <monstr@monstr.eu<mailto:m=

onstr@monstr.eu>> wrote:

quoted

On 12/10/2012 10:41 PM, Grant Likely wrote:

quoted

drivers/pci/pci-of.c would be good. I'd also accept drivers/of/pci.c
which might actually be a good idea in the short term so that it gets
appropriate supervision while being generalized before being moved int=

o

quoted

quoted

the pci directory.

Ben: Are you willing to move that ppc code to this location?
It is probably not good idea that I should do it when I even don't have
hardware available for testing (Asking someone else).

=20
You're a clever guy, you are more than capable of crafting the patch,
even if you can't test on hardware. :-)
=20
I refactored most of the OF support code without having access to most
of the affected hardware. Once I got the changes out there for review
I also asked for spot testing before getting it into linux-next for
even more testing.

I've been working on a relatively architecture agnostic PCI host bridge dri=
ver
and also wanted to avoid duplicating more generic DT parsing code for PCI
bindings.

I've ended up with a patch which provides an iterator for returning resourc=
es
based on the the typical 'ranges' binding. This has ended up living in
drivers/of/address.c. I originally started out in drivers/of/pci.c and
drivers/pci/pci-of.c but found there were good (and static) implementations=
 in
drivers/of/address.c which can be reused (e.g. of_bus_pci_get_flags,
bus->count_cells).

I'm not just ready to post it - but can do before early next week if you ca=
n
wait.

Andrew Murray