Re: [PATCH v2 3/3] platform/x86: thinkpad_acpi: Add support for Lenovo lap sensor

[Hans de Goede <hdegoede@redhat.com>]

Hi,

On 10/20/20 2:15 AM, Mark Pearson wrote:
> Use input device event support for notifying userspace of lap mode sensor
> state changes.
> 
> Signed-off-by: Mark Pearson <markpearson@lenovo.com>
> ---
>  drivers/platform/x86/thinkpad_acpi.c | 75 ++++++++++++++++++++++------
>  1 file changed, 59 insertions(+), 16 deletions(-)
> 
> diff --git a/drivers/platform/x86/thinkpad_acpi.c b/drivers/platform/x86/thinkpad_acpi.c
> index 5ddf2775fb06..c20b9902270b 100644
> --- a/drivers/platform/x86/thinkpad_acpi.c
> +++ b/drivers/platform/x86/thinkpad_acpi.c
> @@ -4013,7 +4013,7 @@ static bool hotkey_notify_usrevent(const u32 hkey,
>  }
>  
>  static void thermal_dump_all_sensors(void);
> -static void proxsensor_refresh(void);
> +static void proxsensor_refresh(bool palm, bool lap);
>  
>  static bool hotkey_notify_6xxx(const u32 hkey,
>  				 bool *send_acpi_ev,
> @@ -4081,7 +4081,7 @@ static bool hotkey_notify_6xxx(const u32 hkey,
>  	case TP_HKEY_EV_PALM_DETECTED:
>  	case TP_HKEY_EV_PALM_UNDETECTED:
>  		/* palm detected  - pass on to event handler */
> -		proxsensor_refresh();
> +		proxsensor_refresh(true /* palm */, false /* lap */);
>  		return true;
>  
>  	default:
> @@ -9929,6 +9929,23 @@ static struct ibm_struct dytc_driver_data = {
>  struct input_dev *tpacpi_sw_dev;
>  bool has_palmsensor;
>  bool palmsensor_state;
> +bool has_lapsensor;
> +bool lapsensor_state;

Again, drop the global _state caching, it is not necessary.

> +
> +static int lapsensor_get(bool *present, bool *state)
> +{
> +	acpi_handle dytc_handle;
> +	int output;
> +
> +	*present = false;
> +	if (ACPI_FAILURE(acpi_get_handle(hkey_handle, "DYTC", &dytc_handle)))
> +		return -ENODEV;
> +	if (!acpi_evalf(dytc_handle, &output, NULL, "dd", DYTC_CMD_GET))
> +		return -EIO;
> +	*present = true; /*If we get his far, we have lapmode support*/
> +	*state = output & BIT(DYTC_GET_LAPMODE_BIT) ? true : false;
> +	return 0;
> +}
>  
>  static int palmsensor_get(bool *present, bool *state)
>  {
> @@ -9945,36 +9962,56 @@ static int palmsensor_get(bool *present, bool *state)
>  	return 0;
>  }
>  
> -static void proxsensor_refresh(void)
> +static void proxsensor_refresh(bool palm, bool lap)

There is zero shared code between the palm ==true and the
lap ==true paths, please just make this 2 separate functions.

And then I guess rename the original proxsensor_refresh to
palmsensor_refresh (note please do this in the 2/3 patch)
and add a new lapsensor_refresh

>  {
>  	bool new_state;
>  	int err;
>  
> -	if (has_palmsensor) {
> +	if (palm && has_palmsensor) {
>  		err = palmsensor_get(&has_palmsensor, &new_state);
> -		if (err)
> -			return;

And then you can also keep the if (err) return; construct, which
is a bit cleaner (more common used) solution vs the !err way
of handling errors.

> -		if (new_state != palmsensor_state) {
> +		if (!err && (new_state != palmsensor_state)) {
>  			input_report_switch(tpacpi_sw_dev, SW_PALMREST_PROXIMITY, new_state);
>  			input_sync(tpacpi_sw_dev);
>  			palmsensor_state = new_state;
>  		}
>  	}
> +
> +	if (lap && has_lapsensor) {
> +		err = lapsensor_get(&has_lapsensor, &new_state);
> +		if (!err && (new_state != lapsensor_state)) {
> +			input_report_switch(tpacpi_sw_dev, SW_LAP_PROXIMITY, new_state);
> +			input_sync(tpacpi_sw_dev);
> +			lapsensor_state = new_state;

Same as with the other patch there is no need for the
new_state != lapsensor_state check, the input core does this for you
turning reporting the same state twice into a no-op.

> +		}
> +	}
>  }
>  
>  static int tpacpi_proxsensor_init(struct ibm_init_struct *iibm)
>  {
> -	int palm_err;
> +	int palm_err, lap_err, err;
>  
> +	/* Make sure globals are set to a sensible initial value */
> +	has_palmsensor = false;
> +	has_lapsensor = false;
>  	palm_err = palmsensor_get(&has_palmsensor, &palmsensor_state);
> +	lap_err = lapsensor_get(&has_lapsensor, &lapsensor_state);
> +
>  	/* If support isn't available (ENODEV) then don't return an error */
> -	if (palm_err == -ENODEV)
> +	if ((palm_err == -ENODEV) && (lap_err == -ENODEV))
>  		return 0;

return 1, see comment on previous patch.

## begin block ###
> -	/* For all other errors we can flag the failure */
> +	/* If both sensors error out - return an error */
> +	if (palm_err && lap_err)
> +		return palm_err ? palm_err : lap_err;
> +	/*
> +	 * If just one sensor not available, we still want the input device,
> +	 * so just flag it and carry on
> +	 */
>  	if (palm_err)
> -		return palm_err;
> +		pr_info("Palm sensor returned error %d", palm_err);
> +	if (lap_err)
> +		pr_info("Lap sensor returned error %d", lap_err);
### end block ###

thinkpad_acpi will typically error out completely on non -ENODEV
errors and the palmsensor code from patch 2/3 also does that.
Note that returning an error from a module/sub-driver's init() is
fatal (causes the module to not load), so before this change the
palmsensor_get call failing with a non -ENODEV error was fatal.
This may seem a bit harsh, but it is how error handling in
thinkpad_acpi has worked so far, so lets be consistent here.

Also if now only 1 of the 2 sensors is available you will log
the -ENODEV error.

So this whole block should be replaced with something like this:

	if (palm_err && palm_err != ENODEV)
		return palm_err;

	if (lap_err && lap_err != ENODEV)
		return lap_err;

>  
> -	if (has_palmsensor) {
> +	if (has_palmsensor || has_lapsensor) {
>  		tpacpi_sw_dev = input_allocate_device();
>  		if (!tpacpi_sw_dev)
>  			return -ENOMEM;
> @@ -9990,10 +10027,14 @@ static int tpacpi_proxsensor_init(struct ibm_init_struct *iibm)
>  			input_set_capability(tpacpi_sw_dev, EV_SW, SW_PALMREST_PROXIMITY);
>  			input_report_switch(tpacpi_sw_dev, SW_PALMREST_PROXIMITY, palmsensor_state);
>  		}
> -		palm_err = input_register_device(tpacpi_sw_dev);
> -		if (palm_err) {
> +		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 (err) {
>  			input_free_device(tpacpi_sw_dev);
> -			return palm_err;
> +			return err;
>  		}
>  	}
>  	return 0;
> @@ -10057,8 +10098,10 @@ static void tpacpi_driver_event(const unsigned int hkey_event)
>  		mutex_unlock(&kbdlight_mutex);
>  	}
>  
> -	if (hkey_event == TP_HKEY_EV_THM_CSM_COMPLETED)
> +	if (hkey_event == TP_HKEY_EV_THM_CSM_COMPLETED) {
>  		dytc_lapmode_refresh();
> +		proxsensor_refresh(false /* palm */, true /* lap */);
> +	}
>  
>  }
>  
> 

Otherwise this looks good to me,

Regards,

Hans