[PATCH 06/14] mfd: pxa-w1: MFD driver for PXA 1wire control + DS1WM chip

[PATCH 06/14] mfd: pxa-w1: MFD driver for PXA 1wire control + DS1WM chip

[Haojian Zhuang]

On Tue, Apr 26, 2011 at 5:48 PM, Samuel Ortiz <sameo@linux.intel.com> wrote:
> Hi Haojian,
>
> On Mon, Apr 18, 2011 at 10:04:03PM +0800, Haojian Zhuang wrote:
>> This driver provides registers and IRQ of PXA3xx chips to the ds1wm driver.
> I see why this is needed, but I really think the ds1wm driver should be
> converted to a regular platform device. The MFD driver below is just adding a
> useless middle layer between your platform code and the ds1wm one.
>
I'm OK to use the regulator platform device for ds1wm driver except
for one concern.
ds1wm driver is designed for mfd cell driver. I need to update it as
regulator platform
driver first. cell->enable() / cell->disable() is used to enable clock
for ds1wm driver.
So I need to change the API from cell->enable/disable to clk_enable()
/ clk_disable().

But the key issue is that there's no common structure for clkdev. Now
clkdev is only
designed for deeply machine depend.

[PATCH 06/14] mfd: pxa-w1: MFD driver for PXA 1wire control + DS1WM chip

[Russell King - ARM Linux]

On Sat, May 14, 2011 at 10:21:25PM +0800, Haojian Zhuang wrote:
> But the key issue is that there's no common structure for clkdev. Now
> clkdev is only designed for deeply machine depend.
> 
> >From my view is that we need a common structure for clkdev.

You mean struct clk.

As I said on our call on Thursday, everyone knows this, and it's
something that has been worked on for the last year.  Concensus had
been reached and everyone had settled on a proposal, and just when
we were almost ready to merge it, the concensus fell apart again.

As a result of that, a common struct clk has been delayed - it's not
going to be in place for 2.6.40 (three months) but maybe we can reach
concensus again for the .41 merge window.

As I also said, this issue was talked about in a session at the Linaro
conference, and it will be interesting to hear what the result of that
session was.  At the moment, I'm entirely out of the loop on many things
that were discussed there (except for a message from tglx.)

[PATCH 06/14] mfd: pxa-w1: MFD driver for PXA 1wire control + DS1WM chip

[Mark Brown]

On Sat, May 14, 2011 at 08:40:17PM +0100, Russell King - ARM Linux wrote:

> As I also said, this issue was talked about in a session at the Linaro
> conference, and it will be interesting to hear what the result of that
> session was.  At the moment, I'm entirely out of the loop on many things
> that were discussed there (except for a message from tglx.)

Everything should have had some live noted minutes linked off this site:

   http://summit.ubuntu.com/uds-o/

probably everything interesting is on this page:

   http://summit.ubuntu.com/uds-o/track/linaro-kernel/

(in etherpad, you want the icon that looks a bit like a notepad on the
left of the box in the talk).  They won't always be accurate as they
were done live and not 100% reviewed but might be helpful.

Several people had actions to bring things to the list so I'd expect to
see some more digested mails hitting in the next week or two.

[PATCH 06/14] mfd: pxa-w1: MFD driver for PXA 1wire control + DS1WM chip

[Samuel Ortiz]

Hi Haojian,

On Sat, May 14, 2011 at 10:21:25PM +0800, Haojian Zhuang wrote:
> On Tue, Apr 26, 2011 at 5:48 PM, Samuel Ortiz <sameo@linux.intel.com> wrote:
> > Hi Haojian,
> >
> > On Mon, Apr 18, 2011 at 10:04:03PM +0800, Haojian Zhuang wrote:
> >> This driver provides registers and IRQ of PXA3xx chips to the ds1wm driver.
> > I see why this is needed, but I really think the ds1wm driver should be
> > converted to a regular platform device. The MFD driver below is just adding a
> > useless middle layer between your platform code and the ds1wm one.
> >
> I'm OK to use the regulator platform device for ds1wm driver except
> for one concern.
> ds1wm driver is designed for mfd cell driver. I need to update it as
> regulator platform
> driver first. cell->enable() / cell->disable() is used to enable clock
> for ds1wm driver.
> So I need to change the API from cell->enable/disable to clk_enable()
> / clk_disable().
> 
> But the key issue is that there's no common structure for clkdev. Now
> clkdev is only
> designed for deeply machine depend.
> 
> From my view is that we need a common structure for clkdev. Then mfd driver can
> get benefit from this. Some client driver can be written as regulator
> platform device.
> And mfd driver can use them seamless. What's your opinion?
I may be missing your point, but what is preventing you from calling the clock
API from drivers/w1/masters/ds1wm.c, the same way you're calling it from this
patch ?

Cheers,
Samuel.

-- 
Intel Open Source Technology Centre
http://oss.intel.com/