Skip to contentSkip to conversation

input: mousedev: Emulate right and middle buttons for single-button touchpads

PATCH In review Henrik Rydberg 2 participants 6 messages Sep 5, 2008 – Sep 11, 2008

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 2/2 input: mousedev: Emulate right and middle buttons for single-button touchpads

On most linux-based systems, the right and middle buttons are used extensively. On computers lacking either of those buttons, such as the Apple Macbooks, emulation is needed. This patch adds emulation of the missing buttons via two-fingers-and-click and three-fingers- and-click operations, which has recently become standard, both in MacOS and in the Xorg synaptics driver.

---
 drivers/input/mousedev.c |   31 +++++++++++++++++++++++++++++++
 1 files changed, 31 insertions(+), 0 deletions(-)
drivers/input/mousedev.c +31 -0
index 27e1d73..a1e929f 100644
@@ -53,6 +53,14 @@ static unsigned tap_move = 20;
 module_param(tap_move, uint, 0644);
 MODULE_PARM_DESC(tap_move, "Tap distance for touchpads in absolute mode (res)");
 
+static unsigned two_finger_code = BTN_RIGHT;
+module_param(two_finger_code, uint, 0644);
+MODULE_PARM_DESC(two_finger_code, "Button code of two-finger click");
+
+static unsigned three_finger_code = BTN_MIDDLE;
+module_param(three_finger_code, uint, 0644);
+MODULE_PARM_DESC(three_finger_code, "Button code of three-finger click");
+
 struct mousedev_hw_data {
 	int dx, dy, dz;
 	int x, y;
@@ -81,6 +89,7 @@ struct mousedev {
 	int frac_dx, frac_dy;
 	unsigned long touch;
 	int touch_dx, touch_dy;
+	int double_tap, triple_tap;
 };
 
 enum mousedev_emul {
@@ -225,11 +234,31 @@ static void mousedev_rel_event(struct mousedev *mousedev,
 	}
 }
 
+/* multi-finger emulation of non-existing buttons */
+static unsigned int get_button_code(const struct mousedev *mousedev)
+{
+	unsigned int code = BTN_LEFT;
+	if (mousedev->double_tap)
+		code = two_finger_code;
+	if (mousedev->triple_tap)
+		code = three_finger_code;
+	if (test_bit(code, mousedev->handle.dev->keybit))
+		code = BTN_LEFT;
+	return code;
+}
+
 static void mousedev_key_event(struct mousedev *mousedev,
 				unsigned int code, int value)
 {
 	int index;
 
+	if (code == BTN_TOOL_DOUBLETAP)
+		mousedev->double_tap = value;
+	if (code == BTN_TOOL_TRIPLETAP)
+		mousedev->triple_tap = value;
+	if (code == BTN_LEFT)
+		code = get_button_code(mousedev);
+
 	switch (code) {
 
 	case BTN_TOUCH:
@@ -327,6 +356,8 @@ static void mousedev_touchpad_touch(struct mousedev *mousedev, int value)
 {
 	if (!value) {
 		if (mousedev->touch &&
+		    !mousedev->double_tap &&
+		    !mousedev->triple_tap &&
 		    abs(mousedev->touch_dx) < tap_move &&
 		    abs(mousedev->touch_dy) < tap_move &&
 		    time_before(jiffies,
Signature · 1 line
1.5.4.3
maintainer
#1 lore raw

Hi Henrik,

Quoted 7 lines · Henrik Rydberg

On most linux-based systems, the right and middle buttons are used extensively. On computers lacking either of those buttons, such as the Apple Macbooks, emulation is needed. This patch adds emulation of the missing buttons via two-fingers-and-click and three-fingers- and-click operations, which has recently become standard, both in MacOS and in the Xorg synaptics driver.

The standard way of emulating right and middle button presses for "button-deficient" devices is to use Macintosh emulation (MAC_EMUMOUSEBTN) that works not only for touchpads but for regular mice as well.

Overall I don't think we should enhance mousedev for scenarios where proper solution is to install Synaptics X driver. This also goes for your other patch improving tapping in mousedev.

Signature · 1 line
Dmitry
#2 lore raw

Greetings Dmitry,

Quoted 9 lines · Henrik Rydberg
Quoted 7 lines

On most linux-based systems, the right and middle buttons are used extensively. On computers lacking either of those buttons, such as the Apple Macbooks, emulation is needed. This patch adds emulation of the missing buttons via two-fingers-and-click and three-fingers- and-click operations, which has recently become standard, both in MacOS and in the Xorg synaptics driver.

Quoted 4 lines

The standard way of emulating right and middle button presses for "button-deficient" devices is to use Macintosh emulation (MAC_EMUMOUSEBTN) that works not only for touchpads but for regular mice as well.

I have a hard time believing that in five years from now, people will still consider pressing a key on the keyboard the standard way to emulate a right click. Nevertheless, I can see the point in moving away from user-land decisions in the kernel. I will not pursue this one further, although it is a rather neat piece of functionality.

Quoted 3 lines

Overall I don't think we should enhance mousedev for scenarios where proper solution is to install Synaptics X driver. This also goes for your other patch improving tapping in mousedev.

Regarding the tapping patch; it is addressing the correctness of already implemented functionality. I can see nothing wrong with that.

Henrik

maintainer
#3 lore raw
Quoted 11 lines · Henrik Rydberg

Greetings Dmitry,

Quoted 9 lines · Henrik Rydberg
Quoted 7 lines

On most linux-based systems, the right and middle buttons are used extensively. On computers lacking either of those buttons, such as the Apple Macbooks, emulation is needed. This patch adds emulation of the missing buttons via two-fingers-and-click and three-fingers- and-click operations, which has recently become standard, both in MacOS and in the Xorg synaptics driver.

Quoted 8 lines
Quoted 4 lines

The standard way of emulating right and middle button presses for "button-deficient" devices is to use Macintosh emulation (MAC_EMUMOUSEBTN) that works not only for touchpads but for regular mice as well.

I have a hard time believing that in five years from now, people will still consider pressing a key on the keyboard the standard way to emulate a right click.

Well, what other options do you have if your device has only one button and nothing else (i.e. it's a mouse, not a touchpad)? Anyway, I hope in 5 years distros will switch to pure evdev-based interfaces and won't use mousedev at all.

Quoted 11 lines

Nevertheless, I can see the point in moving
away from user-land decisions in the kernel. I will not pursue this one further, although it is a rather neat piece of functionality.

Quoted 3 lines

Overall I don't think we should enhance mousedev for scenarios where proper solution is to install Synaptics X driver. This also goes for your other patch improving tapping in mousedev.

Regarding the tapping patch; it is addressing the correctness of already implemented functionality. I can see nothing wrong with that.

My position that since the isssue can be "fixed" by simply installing the proper X driver I'd rather not include it. Mousedev purpose is to provide good enough emulation in absence of a native driver and that's it. Also, the default value would not work well on high-resiolution devices (like Synaptics) and would require user fiddling with kernel parameters - not worth it in my opinion.

Signature · 1 line
Dmitry
#4 lore raw
Quoted 10 lines
Quoted 3 lines

Regarding the tapping patch; it is addressing the correctness of already implemented functionality. I can see nothing wrong with that.

My position that since the isssue can be "fixed" by simply installing the proper X driver I'd rather not include it. Mousedev purpose is to provide good enough emulation in absence of a native driver and that's it. Also, the default value would not work well on high-resiolution devices (like Synaptics) and would require user fiddling with kernel parameters - not worth it in my opinion.

The parameter could be made dimensionless and the distance test scaled with the size of the device; the argument invalidates the current patch, but not the idea itself. Maybe the patch will appear in acceptable form at some point, but for now, I rest my case.

Henrik

#5 lore raw
Quoted 10 lines · Henrik Rydberg
Quoted 9 lines
Quoted 3 lines

Regarding the tapping patch; it is addressing the correctness of already implemented functionality. I can see nothing wrong with that.

My position that since the isssue can be "fixed" by simply installing the proper X driver I'd rather not include it. Mousedev purpose is to provide good enough emulation in absence of a native driver and that's it. Also, the default value would not work well on high-resiolution devices (like Synaptics) and would require user fiddling with kernel parameters - not worth it in my opinion.

Quoted 4 lines

The parameter could be made dimensionless and the distance test scaled with the size of the device; the argument invalidates the current patch, but not the idea itself. Maybe the patch will appear in acceptable form at some point, but for now, I rest my case.

Actually, the tap_move parameter is already dimensionless - the mousedev_touchpad functions emulates movements on an imaginary screen of resolution 256x256. Thus, with the following description change

-MODULE_PARM_DESC(tap_move, "Tap distance for touchpads in absolute mode (res)"); +MODULE_PARM_DESC(tap_move, "Tap move radius (256 * fraction of screen)");

it should be clear that the parameter does not depend on device resolution. It could even be made a constant in the code, if that is preferred.

Henrik