Re: [PATCH] MFD: TWL: add power off functionality

[Igor Grinberg <grinberg@compulab.co.il>]

On 12/04/11 23:43, NeilBrown wrote:
> On Sun, 4 Dec 2011 21:58:59 +1100 NeilBrown <neilb@suse.de> wrote:
> 
>> On Sun, 04 Dec 2011 11:56:44 +0200 Igor Grinberg <grinberg@compulab.co.il>
>> wrote:
>>
>>> Hi Neil,
>>>
>>> On 12/03/11 03:35, NeilBrown wrote:
>>>> On Sun, 27 Nov 2011 11:42:17 +0200 Igor Grinberg <grinberg@compulab.co.il>
>>>> wrote:
>>>>
>>>>> ping!
>>>>
>>>> pong ...
>>>>
>>>>
>>>> Hi,
>>>>  I've been trying this patch out on my GTA04 with 3.2-rc4 and it doesn't
>>>>  work :-(
>>>
>>> Probably, v3.2-rc4 is not the best to try things out...
>>> This patch is based on v3.1, can you try v3.1, so at least we can
>>> check the that the patch itself has no problems?
>>> Also, CC'ing linux-omap.
>>
>> I think I'll be able to give 3.1 a try - I'll let you know.
>>
> 
> Yes, works fine with 3.1

Good to hear!

> 
> 
> I've managed to find the problem.
> 
> commit af8db1508f2c9f3b6e633e2d2d906c6557c617f9
> Author: Peter Chen <peter.chen@freescale.com>
> Date:   Tue Nov 15 21:52:29 2011 +0100
> 
>     PM / driver core: disable device's runtime PM during shutdown
>  
> ....
> 
> @@ -1742,6 +1743,8 @@ void device_shutdown(void)
>                  */
>                 list_del_init(&dev->kobj.entry);
>                 spin_unlock(&devices_kset->list_lock);
> +               /* Disable all device's runtime power management */
> +               pm_runtime_disable(dev);
>  
>                 if (dev->bus && dev->bus->shutdown) {
>                         dev_dbg(dev, "shutdown\n");
> 
> 
> 
> Removing this call allows power-off to work.
> 
> It seems that omap_i2c.1 is normally in runtime suspend.
> omap_i2c_xfer wakes it up, performs the xfer, then puts it back to sleep.
> 
> So this pm_runtime_disable is called while the device is asleep, so it stays
> asleep.  omap_i2c_xfer cannot wake it up and so cannot xfer anything.

Good job on that investigation.
There were several discussions runtime pm related at the kernel summit,
but I failed to see the impact of that on the issue...

> 
> I'll start a new thread including the people responsible for that patch.

Hopefully, it will get us a solution.


-- 
Regards,
Igor.