Re: [External] Using IIO to export laptop palm-sensor and lap-mode info to userspace?

[Mark Pearson <markpearson@lenovo.com>]

On 2020-10-14 4:16 a.m., Hans de Goede wrote:
> Hi Mark,
> 
> On 10/13/20 11:59 PM, Mark Pearson wrote:
>> Hi
>>
>> On 2020-10-12 8:13 a.m., Hans de Goede wrote:
>>> Hi,
>>>
>>> On 10/9/20 4:19 AM, Jeff LaBundy wrote:
>>>> Hi Hans,
>>>>
>> <snip>
>>>>>
>>>>>> I just wanted to chime in and confirm that we do have at least one
>>>>>> precedent for these being in input (keyboard/iqs62x-keys) and not
>>>>>> iio so I agree with Jonathan here. My argument is that we want to
>>>>>> signal binary events (user grabbed onto or let go of the handset)
>>>>>> rather than deliver continuous data.
>>>>>
>>>>> I was curious what keycode you are using for this, but I see
>>>>> that the keycodes come from devicetree, so I guess I should
>>>>> just ask: what keycode are you using for this ?
>>>>
>>>> The idea here was that a vendor might implement their own daemon
>>>> that interprets any keycode of their choice, hence leaving the
>>>> keycodes assignable via devicetree.
>>>>
>>>> This particular device also acts as a capacitive/inductive button
>>>> sensor, and these applications were the primary motivation for it
>>>> landing in input with its status bits mapped to keycodes.
>>>>
>>>> I don't think there are any keycodes that exist today that would
>>>> universally work for this application. The couple that seem most
>>>> closely related (e.g. KEY_WLAN or KEY_RFKILL) are typically used
>>>> for disabling the adapter entirely or for airplane mode (please
>>>> correct me if I'm wrong).
>>>
>>> You're right (aka not wrong), KEY_WLAN and KEY_RFKILL are used to
>>> toggle wireless radios on/off and re-using them for some SAR
>>> purpose would lead to nothing but confusion. We really need to
>>> define some standard *new* event-codes for this, such as e.g.
>>> the proposed SW_LAP_PROXIMITY and SW_PALMREST_PROXIMITY.
>>>
>>>> To that end, I'm keen to see how this interface unfolds because
>>>> SAR detection tends to be an available mode of operation for
>>>> several of the capacitive touch devices I've been working with.
>>>
>>> I guess that for touchscreens at least (which are on the front),
>>> using the existing SW_FRONT_PROXIMITY would make the most sense.
>>>
>>
>> I've been looking at implementing this and I'm missing something - and 
>> I think it's probably something obvious so hoping someone can short 
>> cut me to the answer. Hope it's OK to do that in this thread (I 
>> removed the linux-iio list as I'm assuming they won't be interested)
>>
>> I've added the new event codes to input-event-codes.h and updated 
>> mode_devicetable.h
>>
>> In the thinkpad_acpi.c driver I initialise the device:
>>
>>     tpacpi_sw_dev = input_allocate_device();
>>     if (!tpacpi_sw_dev)
>>             return -ENOMEM;
>>     tpacpi_sw_dev->name = "Thinkpad proximity switches";
>>     tpacpi_sw_dev->phys = TPACPI_DRVR_NAME "/input1";
>>     tpacpi_sw_dev->id.bustype = BUS_HOST;
>>     tpacpi_sw_dev->id.vendor = thinkpad_id.vendor;
>>     tpacpi_sw_dev->id.product = TPACPI_HKEY_INPUT_PRODUCT;
>>     tpacpi_sw_dev->id.version = TPACPI_HKEY_INPUT_VERSION;
>>     tpacpi_sw_dev->dev.parent = &tpacpi_pdev->dev;
>>
>>     if (has_palmsensor) {
>>        input_set_capability(tpacpi_sw_dev, EV_SW, SW_PALMREST_PROXIMITY);
>>        input_report_switch(tpacpi_sw_dev,SW_PALMREST_PROXIMITY, 
>> palmsensor_state);
>>     }
>>
>>     if (has_lapsensor) {
>>          input_set_capability(tpacpi_sw_dev, EV_SW, SW_LAP_PROXIMITY);
>>          input_report_switch(tpacpi_sw_dev, SW_LAP_PROXIMITY, 
>> lapsensor_state);
>>     }
>>     err = input_register_device(tpacpi_sw_dev);
>>
>> If the sensor triggers I update the inputdevice with:
>>     input_report_switch(tpacpi_sw_dev, SW_PALMREST_PROXIMITY, new_state);
>>     input_sync(tpacpi_sw_dev);
>> <similar for lapmode>
>>
>> However I'm not seeing the change when I look under evtest, though I 
>> do see the new sensors show up:
>>
>>     [banther@localhost linux]$ sudo evtest
>>     No device specified, trying to scan all of /dev/input/event*
>>     Available devices:
>>     /dev/input/event0:    Sleep Button
>>     /dev/input/event1:    Lid Switch
>>     /dev/input/event2:    Power Button
>>     /dev/input/event3:    AT Translated Set 2 keyboard
>>     /dev/input/event4:    TPPS/2 Elan TrackPoint
>>     /dev/input/event5:    SYNA8004:00 06CB:CD8B Mouse
>>     /dev/input/event6:    SYNA8004:00 06CB:CD8B Touchpad
>>     /dev/input/event7:    Video Bus
>>     /dev/input/event8:    Thinkpad proximity switches
>>     /dev/input/event9:    PC Speaker
>>     /dev/input/event10:    Integrated Camera: Integrated C
>>     /dev/input/event11:    sof-hda-dsp Headset Jack
>>     /dev/input/event12:    sof-hda-dsp Mic
>>     /dev/input/event13:    sof-hda-dsp Headphone
>>     /dev/input/event14:    sof-hda-dsp HDMI/DP,pcm=3
>>     /dev/input/event15:    sof-hda-dsp HDMI/DP,pcm=4
>>     /dev/input/event16:    sof-hda-dsp HDMI/DP,pcm=5
>>     /dev/input/event17:    ThinkPad Extra Buttons
>>     Select the device event number [0-17]: 8
>>     Input driver version is 1.0.1
>>     Input device ID: bus 0x19 vendor 0x17aa product 0x5054 version 0x4101
>>     Input device name: "Thinkpad proximity switches"
>>     Supported events:
>>       Event type 0 (EV_SYN)
>>       Event type 5 (EV_SW)
>>         Event code 17 (?) state 0
>>         Event code 18 (?) state 0
>>     Properties:
>>     Testing ... (interrupt to exit)
>>
>> The state for both sensors is supposed to be 1.
>> I did download and rebuild evtest and fixed the (?), but haven't 
>> figured out why the state is wrong. It seemed related to the number of 
>> keys which I found odd.
>>
>> Any suggestions from what I'm missing, or have done wrong, or where I 
>> should dig next? What's the recommended way of testing my implementation?
> 
> A couple suggestions:
> 
> 1. What Jeff said, add a printf to make sure that the 
> input_report_switch + sync actually get called.
> 2. Did you rebuild your entire kernel after adding the new SW_ 
> definitions to input-event-codes.h ?  The core may very well be using a 
> check for SW_MAX / SW_CNT
> 3. Did you rebuild the evtest tool against the latest headers, including 
> an updated SW_MAX / SW_CNT.
> 
> To rule out issues with evtest still using an old SW_MAX somewhere you 
> could (as a hack) replace the 2 new SW_... codes with 2 existing ones 
> and see if that makes things work.
> 

Looks like rebuilding the entire kernel did the trick - guess I needed 
to update the input driver (which now I write that does seem 
obvious....sigh)

I'm seeing the events show up in evtest which is very cool

Thanks to all for the comments and help - much appreciated. I'll work on 
getting a patch out for this soon for review

Mark