# \[PATCH\] bcm5974: report ABS\_MT\_PRESSURE

- List: linux-input
- 2 messages, 1 author
- Page: https://ratatoskr.run/linux-input/2017/12/8930713/t

## \[PATCH\] bcm5974: report ABS\_MT\_PRESSURE

- From: Ryan Hendrickson \<hidden\>
- Date: 2017-12-01 17:19:58 
- Message: https://ratatoskr.run/linux-input/2017/12/8930713.md

```text
Report ABS_MT_PRESSURE in the bcm5974 driver with the same data used for
ABS_PRESSURE.

Some touchpad improvements in recent versions of libinput, such as
pressure-based palm detection, will only work if ABS_MT_PRESSURE is
reported.

Signed-off-by: Ryan Hendrickson <redacted>

---

Please treat all of this with an extra grain of salt; this is my first real
foray into both evdev/libinput internals and kernel development. That said,
I have been running this patch on my daily-use MacBookPro5,1 for several
months, so I have reason to believe it's stable. It might not be the
‘right’ way to do this though; I mostly just copied the corresponding code
for ABS_PRESSURE, which from this[1] blog post I've inferred should have
the same values and range.

[1]: http://who-t.blogspot.com/2016/09/understanding-evdev.html

Patch was originally submitted as bug 197909[2], which can be closed if
this is accepted.

[2]: https://bugzilla.kernel.org/show_bug.cgi?id=197909

--- linux-4.13/drivers/input/mouse/bcm5974.c.orig	2017-12-01 11:36:35.258890243 -0500
+++ linux-4.13/drivers/input/mouse/bcm5974.c	2017-12-01 11:48:55.187342210 -0500
@@ -548,6 +548,8 @@ static void setup_events_to_report(struc
 	set_abs(input_dev, ABS_MT_POSITION_X, &cfg->x);
 	set_abs(input_dev, ABS_MT_POSITION_Y, &cfg->y);

+	set_abs(input_dev, ABS_MT_PRESSURE, &cfg->p);
+
 	__set_bit(EV_KEY, input_dev->evbit);
 	__set_bit(BTN_LEFT, input_dev->keybit);

@@ -575,10 +577,14 @@ static int report_bt_state(struct bcm597
 	return 0;
 }

-static void report_finger_data(struct input_dev *input, int slot,
+static void report_finger_data(struct input_dev *input,
+			       const struct bcm5974_config *cfg,
+			       int slot,
 			       const struct input_mt_pos *pos,
 			       const struct tp_finger *f)
 {
+	int abs_p;
+
 	input_mt_slot(input, slot);
 	input_mt_report_slot_state(input, MT_TOOL_FINGER, true);

@@ -594,6 +600,9 @@ static void report_finger_data(struct in
 			 MAX_FINGER_ORIENTATION - raw2int(f->orientation));
 	input_report_abs(input, ABS_MT_POSITION_X, pos->x);
 	input_report_abs(input, ABS_MT_POSITION_Y, pos->y);
+
+	abs_p = clamp_val(256 * raw2int(f->touch_major) / cfg->p.max, 0, 255);
+	input_report_abs(input, ABS_MT_PRESSURE, abs_p);
 }

 static void report_synaptics_data(struct input_dev *input,
@@ -640,7 +649,7 @@ static int report_tp_state(struct bcm597
 	input_mt_assign_slots(input, dev->slots, dev->pos, n, 0);

 	for (i = 0; i < n; i++)
-		report_finger_data(input, dev->slots[i],
+		report_finger_data(input, c, dev->slots[i],
 				   &dev->pos[i], dev->index[i]);

 	input_mt_sync_frame(input);
```

## Re: \[PATCH\] bcm5974: report ABS\_MT\_PRESSURE

- From: Ryan Hendrickson \<hidden\>
- Date: 2017-12-15 01:38:06 
- Message: https://ratatoskr.run/linux-input/2017/12/8930739.md

```text
Hi, gentle ping? No rush on the review but it would be nice to know if I followed the right submission procedure and wasn't mistaken for a spammer.

At 2017-12-01 12:13-0500, Ryan Hendrickson [ref] sent:

> Report ABS_MT_PRESSURE in the bcm5974 driver with the same data used for
> ABS_PRESSURE.
>
> Some touchpad improvements in recent versions of libinput, such as
> pressure-based palm detection, will only work if ABS_MT_PRESSURE is
> reported.
>
> Signed-off-by: Ryan Hendrickson <redacted>
>
> ---
>
> Please treat all of this with an extra grain of salt; this is my first real
> foray into both evdev/libinput internals and kernel development. That said,
> I have been running this patch on my daily-use MacBookPro5,1 for several
> months, so I have reason to believe it's stable. It might not be the
> ‘right’ way to do this though; I mostly just copied the corresponding code
> for ABS_PRESSURE, which from this[1] blog post I've inferred should have
> the same values and range.
>
> [1]: http://who-t.blogspot.com/2016/09/understanding-evdev.html
>
> Patch was originally submitted as bug 197909[2], which can be closed if
> this is accepted.
>
> [2]: https://bugzilla.kernel.org/show_bug.cgi?id=197909
>
> --- linux-4.13/drivers/input/mouse/bcm5974.c.orig	2017-12-01 11:36:35.258890243 -0500
> +++ linux-4.13/drivers/input/mouse/bcm5974.c	2017-12-01 11:48:55.187342210 -0500
> @@ -548,6 +548,8 @@ static void setup_events_to_report(struc
> 	 set_abs(input_dev, ABS_MT_POSITION_X, &cfg->x);
> 	 set_abs(input_dev, ABS_MT_POSITION_Y, &cfg->y);
>
> +	set_abs(input_dev, ABS_MT_PRESSURE, &cfg->p);
> +
> 	 __set_bit(EV_KEY, input_dev->evbit);
> 	 __set_bit(BTN_LEFT, input_dev->keybit);
>
> @@ -575,10 +577,14 @@ static int report_bt_state(struct bcm597
>  	return 0;
>  }
>
> -static void report_finger_data(struct input_dev *input, int slot,
> +static void report_finger_data(struct input_dev *input,
> +			       const struct bcm5974_config *cfg,
> +			       int slot,
> 			        const struct input_mt_pos *pos,
> 			        const struct tp_finger *f)
> {
> +	int abs_p;
> +
> 	 input_mt_slot(input, slot);
> 	 input_mt_report_slot_state(input, MT_TOOL_FINGER, true);
>
> @@ -594,6 +600,9 @@ static void report_finger_data(struct in
> 	 		 MAX_FINGER_ORIENTATION - raw2int(f->orientation));
> 	 input_report_abs(input, ABS_MT_POSITION_X, pos->x);
> 	 input_report_abs(input, ABS_MT_POSITION_Y, pos->y);
> +
> +	abs_p = clamp_val(256 * raw2int(f->touch_major) / cfg->p.max, 0, 255);
> +	input_report_abs(input, ABS_MT_PRESSURE, abs_p);
>  }
>
> static void report_synaptics_data(struct input_dev *input,
> @@ -640,7 +649,7 @@ static int report_tp_state(struct bcm597
> 	 input_mt_assign_slots(input, dev->slots, dev->pos, n, 0);
>
> 	for (i = 0; i < n; i++)
> -		report_finger_data(input, dev->slots[i],
> +		report_finger_data(input, c, dev->slots[i],
> 				    &dev->pos[i], dev->index[i]);
>
> 	input_mt_sync_frame(input);
>
```
