[PULL] [PATCH 0/4] arm: at91: gpio fix

[olof@lixom.net (Olof Johansson)]

On Wed, Jul 18, 2012 at 3:57 AM, Jean-Christophe PLAGNIOL-VILLARD
<plagnioj@jcrosoft.com> wrote:
> On 22:41 Tue 17 Jul     , Olof Johansson wrote:
>> On Tue, Jul 17, 2012 at 10:23 PM, Jean-Christophe PLAGNIOL-VILLARD
>> <plagnioj@jcrosoft.com> wrote:
>> > On 20:54 Tue 17 Jul     , Olof Johansson wrote:
>> >> Hi Jean-Christophe,
>> >>
>> >> Comments below. As I mentioned in the SPEAr pull request just now, we
>> >> are very close to final 3.5, so only fixes to bugs should go in.
>> >> Please help me out a bit below.
>> >>
>> >> On Sun, Jul 15, 2012 at 6:16 AM, Jean-Christophe PLAGNIOL-VILLARD
>> >> <plagnioj@jcrosoft.com> wrote:
>> >>
>> >> > Jean-Christophe PLAGNIOL-VILLARD (4):
>> >> >       ARM: at91: gpio: implement request
>> >> >       ARM: at91: gpio: implement gpio_free
>> >>
>> >> Are the two above regressions, or just general improvement? Looks like
>> >> the latter to me?
>> > On non dt I agree but on DT the gpio are not mux in the kernel.
>> > Today it work if the bootloader mux it as gpio so the kernel work onther wise BUG
>> >
>> > and this is the case today
>>
>> Ok, thanks -- that makes sense. Care to roll those two patches into
>> one and document that in the commit message?
> no so much honestly as you wish
>>
>> >>
>> >> >       at91: regroup gpio and pinctrl under a simple-bus
>> >>
>> >> This looks like a cleanup patch to me. I'm also not sure that this is
>> >> appropriate use of a simple-bus. The patch seems to have been posted
>> >> for review the same day as you sent the pull request, and not cc:d to
>> >> devicetree-discuss.
>> > forget to cc the devicetree but on at91 the pintcrl and gpio use the same
>> > registers so you must you the simple-bus
>> >
>> > As I've a huge patch series that I'm finishing to fix after found this bug on
>> > the gpio it really help to do not rebase it
>>
>> Defining the same device twice to have two drivers bind to it doesn't
>> seem like the right solution here. That patch needs an acked-by from
>> Rob Herring or Grant Likely before we pick it up. Existing device
>> trees don't include the pinctrl nodes, so this is 3.6 material as far
>> as I can tell.
> that's why I'm rewriting the pinctrl and gpio as one driver only
>
> but we need the simple bus to describe it as done on other SoC Imx as example
>
> so Rob can we have the ack to avoid to re-write about 20 patches that touch
> the dts of all the at91 socs


I'm sending in the last batch of fixes for 3.5 without this branch in
it. It's also getting too late to get them into 3.6 at this time since
the merge window is about to open and all code should already be in
arm-soc.

Jean-Christophe, when you've had a chance to clean up this branch and
get it ready for upstream merging, please mark the patches for stable
as appropriate to get them into 3.5.x.


-Olof