Re: [PATCH 0/2] power button repeat input events

[Nishanth Menon <nm@ti.com>]

Denis Karpov wrote, on 03/14/2011 08:05 PM:
> On Mon, Mar 14, 2011 at 05:10:22PM +0530, ext Menon, Nishanth wrote:
>> Date: Mon, 14 Mar 2011 17:10:22 +0530
>> From: "ext Menon, Nishanth"<nm@ti.com>
>> To: Denis Karpov<ext-denis.2.karpov@nokia.com>
>> CC: tony@atomide.com, linux-omap@vger.kernel.org
>> Subject: Re: [PATCH 0/2] power button repeat input events
>> In-Reply-To:<20110314111327.GA10563@smart.research.nokia.com>
>>
>> On Mon, Mar 14, 2011 at 16:43, Denis Karpov
>> <ext-denis.2.karpov@nokia.com>  wrote:
>>>
>>> On Fri, Mar 11, 2011 at 10:25:27PM +0530, ext Nishanth Menon wrote:
>>>>
>>>> Denis Karpov wrote, on 03/11/2011 10:07 PM:
>>>>> Hi,
>>>>>
>>>>> this patchset adds optional generation of repeat input events for
>>>>> twl4030 power button. It adds optional platform data for
>>>>> twl4030-pwrbutton and initializes it for RX51/N900 (default
>>>>> behavior kept - repeated input events disabled).
>>>>>
>>>>> We are working on an OMAP3 with TWL5031 MFD. In one of the use
>>>>> scenarios userspace wants to receive repeated input events from
>>>>> long press of power button.
>>>>
>>>> dumb question - is'nt this a userspace responsibility as far as power
>>>> button is concerned? what is the benefit of providing repeat events
>>>> like a generic keypad for the power button?
>>>
>>> In our case userspace needs this to track the continuous button press
>>> of power button.
>> thanks for your explanation, but a follow on dumb question:
>> include/linux/input.h
>> #define KEY_POWER               116     /* SC System Power Down */
>>
>> in twl-powerbutton.c, I see:
>> input_report_key(pwr, KEY_POWER, value&  PWR_PWRON_IRQ);
>> where value is the value from the register status, should'nt we get
>> two events? one on press and one on release after say 10 mins(assuming
>> person keeps it pressed for 10 mins?
>>
>> Is there different behavior we intend to implement in userspace which
>> differentiates 1 KEY_POWER event Vs 10 KEY_POWER event?
>
> No, but due to specifics of userspace sw components involved it seems
> not to be possible to reliably save the state for power button to detect
> long keypresses.
>
> I thought that this change if implemented as optional feature and
> localized to scpecific devices/boards would not hurt.
> twl-keypad.c implements exactly same behavior for keypad keys, why not
> to do this for the power key ?
Are'nt the key events different? KEY_POWER is meant to shutdown the 
system if my reading is right (other keys need repeatability) - I mean 
trying to add a feature because userspace handling is flawed in some 
manner makes me wonder if we are moving the code to kernelspace because 
it is convenient?

I have no debates about just the board in question using this, but just 
worried if we are taking a hack for userspace in kernel given that no 
other board or other userspace variants had a necessity for this change 
previously.

-- 
Regards,
Nishanth Menon