Re: [PATCH v2] psmouse - focaltech pass finger width to userspace

[Dmitry Tunin <hanipouspilot@gmail.com>]

> 18.04.2015 16:35, Dmitry Tunin пишет:
>> Focaltech touchpads report finger width in packet[5] of absolute packet.
>> Range for width in raw format is 0x10 - 0x70. Second half-byte is always 0.
>> 0xff is reported, when a large contact area is detected.
>> This can be handled in userspace.
>>
>> Signed-off-by: Dmitry Tunin <hanipouspilot@gmail.com>
>> ---
>>  drivers/input/mouse/focaltech.c | 9 +++++++++
>>  1 file changed, 9 insertions(+)
>>
>> diff --git a/drivers/input/mouse/focaltech.c b/drivers/input/mouse/focaltech.c
>> index 23d2594..25e4bb3 100644
>> --- a/drivers/input/mouse/focaltech.c
>> +++ b/drivers/input/mouse/focaltech.c
>> @@ -90,6 +90,12 @@ struct focaltech_finger_state {
>>  	 */
>>  	unsigned int x;
>>  	unsigned int y;
>> +
>> +	/*
>> +	* Finger width 0-7 and 15 for 'latching'
>> +	* 15 value stays until the finger is released
>> +	*/
>> +	unsigned int width;
>>  };
>>  
>>  /*
>> @@ -137,6 +143,7 @@ static void focaltech_report_state(struct psmouse *psmouse)
>>  			input_report_abs(dev, ABS_MT_POSITION_X, clamped_x);
>>  			input_report_abs(dev, ABS_MT_POSITION_Y,
>>  					 priv->y_max - clamped_y);
>> +			input_report_abs(dev, ABS_TOOL_WIDTH, finger->width);
>>  		}
>>  	}
>>  	input_mt_report_pointer_emulation(dev, true);
>> @@ -187,6 +194,7 @@ static void focaltech_process_abs_packet(struct psmouse *psmouse,
>>  
>>  	state->fingers[finger].x = ((packet[1] & 0xf) << 8) | packet[2];
>>  	state->fingers[finger].y = (packet[3] << 8) | packet[4];
>> +	state->fingers[finger].width = packet[5] >> 4;
>>  	state->fingers[finger].valid = true;
>>  }
>>  
>> @@ -331,6 +339,7 @@ static void focaltech_set_input_params(struct psmouse *psmouse)
>>  	__set_bit(EV_ABS, dev->evbit);
>>  	input_set_abs_params(dev, ABS_MT_POSITION_X, 0, priv->x_max, 0, 0);
>>  	input_set_abs_params(dev, ABS_MT_POSITION_Y, 0, priv->y_max, 0, 0);
>> +	input_set_abs_params(dev, ABS_TOOL_WIDTH, 0, 15, 0, 0);
>>  	input_mt_init_slots(dev, 5, INPUT_MT_POINTER);
>>  	__set_bit(INPUT_PROP_BUTTONPAD, dev->propbit);
>>  }
>>
> After some testing I discovered that this looking quite harmless patch leads to a regression.
> At least on a weak laptop X200MA with xorg-synaptics.
> 
> If I scroll with two fingers and then lift them and put them down again, the scrolled text jerks further.
> I think it is because of limited bandwidth of PS/2. When I increase amount of data, it starts behaving worse.
> Relative packets do not report finger width either way, so we do not need to report it each time.
> If general idea is approved, then it can be done another way, so finger width information is sent, when it is really
> appropriate. 
> 

Userspace was confused by this patch. When reported type 0x3 'untouch' packet with width reported, xorg-synaptics just ignored that packet.
That's why scrolling was 'continued' with new co-ordinates. That appeared as a cursor jump.
This is fixed in v 3. I had to use finger count to make sure that width is reported only for absolute packets that contain
this info.
It is possible to make a separate 'report_abs_state()' and call it in case absolute packet is processed. But the result should be same.
Number of active fingers is reported by touch packets either way.
--
To unsubscribe from this list: send the line "unsubscribe linux-input" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html