Re: [PATCH 2/2] ARM: DT: tegra: pinctrl suspend resume hook

[Stephen Warren <swarren@wwwdotorg.org>]

On 04/28/2013 07:42 AM, Bibek Basu wrote:
>> -----Original Message-----
>> From: Thierry Reding [mailto:thierry.reding@avionic-design.de]
>> Sent: Saturday, April 27, 2013 1:19 AM
>> To: Bibek Basu
>> Cc: Rafael J. Wysocki; linus.walleij@linaro.org; swarren@wwwdotorg.org;
>> linux-tegra@vger.kernel.org; linux-kernel@vger.kernel.org; linux-
>> pm@vger.kernel.org
>> Subject: Re: [PATCH 2/2] ARM: DT: tegra: pinctrl suspend resume hook
>>
>> * PGP Signed by an unknown key
>>
>> On Tue, Apr 23, 2013 at 11:39:02PM +0530, Bibek Basu wrote:
>>> Make pinmux as the first node of tegra DT so as to achieve pinctrl as
>>> last device to suspend and first device to resume.
>>
>> I don't think this is a good idea. For one it encodes Linux specific
>> implementation details within the device tree. Furthermore I seem to
>> remember that the device tree makes no guarantees that the order of nodes
>> in the blob is the same as the order in the DTS.
>>
>> That said I don't have a good alternative. Perhaps one solution would be to
>> use suspend_late() and resume_early() for the pinmux driver since they will
>> be called after and before the suspend() and resume() callbacks of all other
>> devices, respectively. Other than that I wasn't able to find anything about
>> forcing a particular order during suspend and resume.
>>
>> Cc'ing Rafael and the linux-pm mailing list, maybe somebody among them
>> knows a proper solution.
>
> Stephen actually suggested me earlier to add dummy pinmux state on all dependent devices node. And that will make sure that dependency of driver prevails.
> I tried that but I was not getting the desired result.
> Can anyone tell me correct syntax for adding dummy state?

I agree with Thierry that this patch is not the correct approach.

A dummy pinctrl state would look something like the following in each
affected device node:

                pinctrl-names = "default";
                pinctrl-0 = <>;