[PATCH v3 2/2] documentation: Fix pinctrl documentation for Meson8 / Meson8b

[Rob Herring]

On Wed, Mar 23, 2016 at 05:04:19PM +0100, Carlo Caione wrote:
> On Wed, Mar 23, 2016 at 4:20 PM, Rob Herring <robh at kernel.org> wrote:
> > On Wed, Mar 23, 2016 at 10:14:00AM +0100, Carlo Caione wrote:
> >> From: Carlo Caione <carlo at endlessm.com>
> >>
> >> Fix pin controller documentation introducing the new compatibles for
> >> the pinctrl drivers specific for aobus / cbus.
> >>
> >> This is needed because we have changed the pin controller driver: we
> >> have now a single specialized pinctrl driver / compatible for each bus
> >> the controller is attached to, instead of one single driver dealing with
> >> all the controllers we have on different buses.
> >
> > Aren't you breaking compatibility with old DTs here? If so, you need to
> > be clear that you are and why you think that is okay.
> 
> Rob,
> It's a bit too late to worry about breaking compatibility since the
> driver changes are already landed in mainline and currently Meson8 and
> Meson8b platforms are broken because of this.
> You can read the whole discussion here [1] and here [2]. Driver and DT
> changes were supposed to go in together but a bit of general
> misunderstanding caused this issue.

My comment was the commit message needs to be clear that you are 
breaking compatibility. That was true before part of this went in. 
 
> We decided to break compatibility with the old DTs since the the
> platform support is still in the really early stage so I really don't
> think this is going to cause any problem and we have a lot of good
> reasons to split the pinctrl driver. First of all it makes sense from
> the hardware prospective, since we actually have two different pin
> controllers on two different buses. Splitting the driver allows us to
> introduce in the DTS both CBUS and AOBUS as simple buses. We need a
> clear description of the two buses in the DTS since several devices
> have a different register mapping depending on which bus they are
> attached to. Also as you can read here [3] we want to map the whole
> CBUS as a syscon device to be able to access several registers
> scattered inside CBUS.

>From a quick glance, you mainly needed to keep the old compatible string 
in the driver and just ignore ao-bank. Then only what depended on 
ao-bank would break. Maybe that's less broken than completely breaking 
the pinctrl driver...

Anyway, merge the fix:

Acked-by: Rob Herring <robh at kernel.org>

Rob