wakeup source - power management

wakeup source - power management

[Ran Shalit]

Hello

What does it mean device without wakeup source in terms of runtime PM ?

Thanks,
Ran

wakeup source - power management

[AYAN KUMAR HALDER]

On Sat, Sep 6, 2014 at 9:56 PM, Ran Shalit <ranshalit@gmail.com> wrote:
> Hello
>
> What does it mean device without wakeup source in terms of runtime PM ?
>
Can you give some source/reference where you came across this terminology.
"device without wakeup source"

Regards,
Ayan Kumar Halder

wakeup source - power management

[Ran Shalit]

On Sun, Sep 7, 2014 at 11:14 PM, AYAN KUMAR HALDER <ayankumarh@gmail.com>
wrote:
> On Sat, Sep 6, 2014 at 9:56 PM, Ran Shalit <ranshalit@gmail.com> wrote:
>> Hello
>>
>> What does it mean device without wakeup source in terms of runtime PM ?
>>
> Can you give some source/reference where you came across this terminology.
> "device without wakeup source"
>
> Regards,
> Ayan Kumar Halder

Hi,

You can find it here:
http://elinux.org/images/1/18/Elc2011_damm.pdf

Thanks, Ran
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.kernelnewbies.org/pipermail/kernelnewbies/attachments/20140907/83b65693/attachment.html 

wakeup source - power management

[AYAN KUMAR HALDER]

On Mon, Sep 8, 2014 at 1:50 AM, Ran Shalit <ranshalit@gmail.com> wrote:
>
>
> On Sun, Sep 7, 2014 at 11:14 PM, AYAN KUMAR HALDER <ayankumarh@gmail.com>
> wrote:
>> On Sat, Sep 6, 2014 at 9:56 PM, Ran Shalit <ranshalit@gmail.com> wrote:
>>> Hello
>>>
>>> What does it mean device without wakeup source in terms of runtime PM ?
>>>
>> Can you give some source/reference where you came across this terminology.
>> "device without wakeup source"
>>
>> Regards,
>> Ayan Kumar Halder
>
> Hi,
>
> You can find it here:
> http://elinux.org/images/1/18/Elc2011_damm.pdf

If I am not mistaken, it takes about device without wakeup source in
Suspend to Ram
which is system power management and not runtime power management.
In this respect it means, the device cannot wake-up the system from
suspend state.
Such devices are completely powered down/clock disabled in suspend state.
Devices which have wakeup source like RTC are not powered down completely
so as to generate a wakeup interrupt. This would try to wakeup the system.

Sorry, I am not aware about "device without wakeup source" in terms of
runtime pm.

Regards,
Ayan Kumar Halder

>
> Thanks, Ran

wakeup source - power management

[Ran Shalit]

On Sun, Sep 7, 2014 at 11:36 PM, AYAN KUMAR HALDER <ayankumarh@gmail.com> wrote:
> On Mon, Sep 8, 2014 at 1:50 AM, Ran Shalit <ranshalit@gmail.com> wrote:
>>
>>
>> On Sun, Sep 7, 2014 at 11:14 PM, AYAN KUMAR HALDER <ayankumarh@gmail.com>
>> wrote:
>>> On Sat, Sep 6, 2014 at 9:56 PM, Ran Shalit <ranshalit@gmail.com> wrote:
>>>> Hello
>>>>
>>>> What does it mean device without wakeup source in terms of runtime PM ?
>>>>
>>> Can you give some source/reference where you came across this terminology.
>>> "device without wakeup source"

> If I am not mistaken, it takes about device without wakeup source in
> Suspend to Ram
> which is system power management and not runtime power management.
> In this respect it means, the device cannot wake-up the system from
> suspend state.
> Such devices are completely powered down/clock disabled in suspend state.
> Devices which have wakeup source like RTC are not powered down completely
> so as to generate a wakeup interrupt. This would try to wakeup the system.
>
> Sorry, I am not aware about "device without wakeup source" in terms of
> runtime pm.
>

Hi,

How should a wakeup source for a device be configured in code.
I could not find it documented anywhere.

Thanks very much,
Ran

wakeup source - power management

[AYAN KUMAR HALDER]

>
> How should a wakeup source for a device be configured in code.
> I could not find it documented anywhere.
>
Check out drivers/input/keyboard/samsung-keypad.c. It is configured as
a wakeup source:-
1. device_init_wakeup() - configuring the device to be of wakeup type
2. enable_irq_wake() - configuring the interrupt to be of wakeup type.

True, I could not find the documentation for the same.

Regards,
Ayan Kumar Halder

wakeup source - power management

[Ran Shalit]

On Mon, Sep 8, 2014 at 9:43 PM, AYAN KUMAR HALDER <ayankumarh@gmail.com> wrote:
>>
>> How should a wakeup source for a device be configured in code.
>> I could not find it documented anywhere.
>>
> Check out drivers/input/keyboard/samsung-keypad.c. It is configured as
> a wakeup source:-
> 1. device_init_wakeup() - configuring the device to be of wakeup type
> 2. enable_irq_wake() - configuring the interrupt to be of wakeup type.
>
> True, I could not find the documentation for the same.
>
> Regards,
> Ayan Kumar Halder

Hi Ayan,

This issue is the point I'm get difficulties with and would appreciate
your clarifications.
Is the wakeup source relevant only for suspend/resume , or also for
runtime PM ? for example, is the keybaord example , which you linked,
relevant for runtime PM too ?
If yes - is the keybaord interrupt wakeup only the keyboard driver,
and nothing else, including the cpu ?

Thanks,
Ran

wakeup source - power management

[AYAN KUMAR HALDER]

> Is the wakeup source relevant only for suspend/resume , or also for
> runtime PM ? for example, is the keybaord example , which you linked,
> relevant for runtime PM too ?
Refer to my reply in the other thread.

> If yes - is the keybaord interrupt wakeup only the keyboard driver,
> and nothing else, including the cpu ?
In your case, the keyboard interrupt being the wakeup source, wakes up the
entire system

wakeup source - power management

[Ran Shalit]

On Tue, Sep 9, 2014 at 8:48 PM, AYAN KUMAR HALDER <ayankumarh@gmail.com> wrote:
>> Is the wakeup source relevant only for suspend/resume , or also for
>> runtime PM ? for example, is the keybaord example , which you linked,
>> relevant for runtime PM too ?
> Refer to my reply in the other thread.
>
>> If yes - is the keybaord interrupt wakeup only the keyboard driver,
>> and nothing else, including the cpu ?
> In your case, the keyboard interrupt being the wakeup source, wakes up the
> entire system

Though the wakeup source is more relevant to suspend/resume and not to
runtime PM, as you've answered to my other thread question, it is
still worth to add that wakeup source exist as part of platform HW
(for example in OMAP35x, I deal with). It makes me think that the
given requirements to make all PM automatic without any userspace
application dealing with it, was irresponsible requirement. There must
be some process or thread who will need to decide when to suspend some
devices, and the cpu will wait for wakeup source to decide when to
resume. I think this will be a better approach. What do you think?