[PATCH] bcm5974-0.62: Compatibility mode reverted, BTN_TOUCH event added

[PATCH] bcm5974-0.62: Compatibility mode reverted, BTN_TOUCH event added

[Henrik Rydberg]

The mousedev driver requires the use of BTN_TOUCH events to process
ABS_X and ABS_Y events properly, which is what is needed for the
bcm5974-based apple computers to have a functional pointer out-of-the-box.
This patch adds the BTN_TOUCH events to bcm5974.

Signed-off-by: Henrik Rydberg <rydberg@euromail.se>
---
 drivers/input/mouse/bcm5974.c |    2 ++
 1 files changed, 2 insertions(+), 0 deletions(-)

diff --git a/drivers/input/mouse/bcm5974.c b/drivers/input/mouse/bcm5974.c
index ae78bb8..cafbb71 100644
--- a/drivers/input/mouse/bcm5974.c
+++ b/drivers/input/mouse/bcm5974.c
@@ -248,6 +248,7 @@ static void setup_events_to_report(struct input_dev *input_dev,
 				0, cfg->y.dim, cfg->y.fuzz, 0);

 	__set_bit(EV_KEY, input_dev->evbit);
+	__set_bit(BTN_TOUCH, input_dev->keybit);
 	__set_bit(BTN_TOOL_FINGER, input_dev->keybit);
 	__set_bit(BTN_TOOL_DOUBLETAP, input_dev->keybit);
 	__set_bit(BTN_TOOL_TRIPLETAP, input_dev->keybit);
@@ -296,6 +297,7 @@ static int report_tp_state(struct bcm5974 *dev, int size)

 	input_report_abs(input, ABS_PRESSURE, int2bound(&c->p, p));

+	input_report_key(input, BTN_TOUCH, p > 0);
 	input_report_key(input, BTN_TOOL_FINGER, n == 1);
 	input_report_key(input, BTN_TOOL_DOUBLETAP, n == 2);
 	input_report_key(input, BTN_TOOL_TRIPLETAP, n > 2);
-- 
1.5.4.3

Re: [PATCH] bcm5974-0.62: Compatibility mode reverted, BTN_TOUCH event added

[Dmitry Torokhov]

On Tue, Sep 02, 2008 at 10:05:56PM +0200, Henrik Rydberg wrote:
> @@ -296,6 +297,7 @@ static int report_tp_state(struct bcm5974 *dev, int size)
> 
>  	input_report_abs(input, ABS_PRESSURE, int2bound(&c->p, p));
> 
> +	input_report_key(input, BTN_TOUCH, p > 0);

Just relying on p will probably make it too sensitive, have you tried
adding hysteresis, like synaptics driver does?

-- 
Dmitry

Re: [PATCH] bcm5974-0.62: Compatibility mode reverted, BTN_TOUCH event added

[Henrik Rydberg]

Dmitry Torokhov wrote:
> On Tue, Sep 02, 2008 at 10:05:56PM +0200, Henrik Rydberg wrote:
>> @@ -296,6 +297,7 @@ static int report_tp_state(struct bcm5974 *dev, int size)
>>
>>  	input_report_abs(input, ABS_PRESSURE, int2bound(&c->p, p));
>>
>> +	input_report_key(input, BTN_TOUCH, p > 0);
> 
> Just relying on p will probably make it too sensitive, have you tried
> adding hysteresis, like synaptics driver does?
> 

I have, and the problem is somewhat more complicated when dealing with
multiple fingers. The simple patch shown about actually works very well
in practice. However, things can always improve. Yesterday I sent a
second patch, bcm5974-0.63, which addressed this issue. However, the
quality of the mouse movement has regressed a little in that patch.
I would like to send yet another patch today, bcm5974-0.64, I hope this
does not cause too much confusion.

Regarding mousedev, I noticed that the tap functionality does not distinguish
between a tap (on/off localized in time and space) and a quick drag (on/off
localized in time only).

Re: [PATCH] bcm5974-0.62: Compatibility mode reverted, BTN_TOUCH event added

[Dmitry Torokhov]

On Wed, Sep 03, 2008 at 09:55:16PM +0200, Henrik Rydberg wrote:
> Dmitry Torokhov wrote:
> > On Tue, Sep 02, 2008 at 10:05:56PM +0200, Henrik Rydberg wrote:
> >> @@ -296,6 +297,7 @@ static int report_tp_state(struct bcm5974 *dev, int size)
> >>
> >>  	input_report_abs(input, ABS_PRESSURE, int2bound(&c->p, p));
> >>
> >> +	input_report_key(input, BTN_TOUCH, p > 0);
> > 
> > Just relying on p will probably make it too sensitive, have you tried
> > adding hysteresis, like synaptics driver does?
> > 
> 
> I have, and the problem is somewhat more complicated when dealing with
> multiple fingers. The simple patch shown about actually works very well
> in practice. However, things can always improve. Yesterday I sent a
> second patch, bcm5974-0.63, which addressed this issue. However, the
> quality of the mouse movement has regressed a little in that patch.
> I would like to send yet another patch today, bcm5974-0.64, I hope this
> does not cause too much confusion.
> 

That is fine.

> Regarding mousedev, I noticed that the tap functionality does not distinguish
> between a tap (on/off localized in time and space) and a quick drag (on/off
> localized in time only).
> 

Yes. Mousedev is a stop-gap measure and is to be used only while user
installs proper driver. It does not need to be perfect.

-- 
Dmitry