Re: [PATCH 3/3] input: add multitouch slot support to w8001.
From: Ping Cheng
Date: Mon Aug 23 2010 - 10:12:05 EST
On Mon, Aug 23, 2010 at 10:18 AM, Henrik Rydberg <rydberg@xxxxxxxxxxx> wrote:
> Hi Peter,
>
> thanks for the patches. Comments inline.
>
>> Some serial wacom devices support two-finger touch. Test for this during
>
>> init and parse the touch packets accordingly. Touch packets are
>> processed using Protocol B (MT Slots).
>>
>> Note: there are several wacom versions that do touch but not two-finger
>> touch. These are not catered for here, touch events for these are simply
>> discarded.
>>
>> Signed-off-by: Peter Hutterer <peter.hutterer@xxxxxxxxx>
>> CC: Henrik Rydberg <rydberg@xxxxxxxxxxx>
>> ---
>> drivers/input/touchscreen/wacom_w8001.c | 99 ++++++++++++++++++++++++++++++-
>> 1 files changed, 96 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/input/touchscreen/wacom_w8001.c b/drivers/input/touchscreen/wacom_w8001.c
>> index c302cc3..a38a3aa 100644
>> --- a/drivers/input/touchscreen/wacom_w8001.c
>> +++ b/drivers/input/touchscreen/wacom_w8001.c
>> @@ -60,6 +60,17 @@ struct w8001_coord {
>> u8 tilt_y;
>> };
>>
>> +/* touch data packet */
>> +struct w8001_touch {
>> + u8 f1_status;
>> + u16 f1_x;
>> + u16 f1_y;
>> + /* only some tablets have 2FG info */
>> + u8 f2_status;
>> + u16 f2_x;
>> + u16 f2_y;
>> +};
>
>
> No pressure information from this device?
No useful pressure/capacity available for this device.
>
>> +
>> /* touch query reply packet */
>> struct w8001_touch_query {
>> u8 panel_res;
>> @@ -85,8 +96,18 @@ struct w8001 {
>> char phys[32];
>> int type;
>> unsigned int pktlen;
>> + unsigned char tracking_id[2];
>> };
>>
>> +static int get_next_tracking_id(void)
>> +{
>> + static unsigned char next_tracking_id;
>> + next_tracking_id = (next_tracking_id + 1) % 256;
>> + if (next_tracking_id == 0)
>> + next_tracking_id = 1;
>> + return next_tracking_id;
>> +}
>
>
> Zero is part of the valid range from 2.6.36. Can be implemented simply as
> (tracking_id++ & 0xfff), see recent MT slots patches.
>
>> +
>> static void parse_data(u8 *data, struct w8001_coord *coord)
>> {
>> memset(coord, 0, sizeof(*coord));
>> @@ -111,6 +132,26 @@ static void parse_data(u8 *data, struct w8001_coord *coord)
>> coord->tilt_y = data[8] & 0x7F;
>> }
>>
>> +static void parse_touch(u8 *data,
>> + unsigned int pktlen, struct w8001_touch *touch)
>> +{
>> + memset(touch, 0, sizeof(*touch));
>> +
>> + touch->f1_status = data[0] & 0x1;
>> + touch->f1_x = data[1] << 7;
>> + touch->f1_x |= data[2];
>> + touch->f1_y = data[3] << 7;
>> + touch->f1_y |= data[4];
>
>
> What are data[5] and data[6]? Why is f1_x/y split into several lines?
>
>> +
>> + if (pktlen >= W8001_PKTLEN_TOUCH2FG) {
>> + touch->f2_status = (data[0] & 0x2) >> 1;
>> + touch->f2_x = data[7] << 7;
>> + touch->f2_x |= data[8];
>> + touch->f2_y = data[9] << 7;
>> + touch->f2_y |= data[10];
>> + }
>
>
> What are data[11] and data[12]?
>
> Since all needed information is available here, there seems to be little reason
> for the w8001_touch structure. Why not complete the report in this function?
>
>> +}
>> +
>> static void parse_touchquery(u8 *data, struct w8001_touch_query *query)
>> {
>> memset(query, 0, sizeof(*query));
>> @@ -128,6 +169,46 @@ static void parse_touchquery(u8 *data, struct w8001_touch_query *query)
>> query->y |= (data[2] >> 3) & 0x3;
>> }
>>
>> +static void w8001_mt_event(struct input_dev *dev,
>> + int slot, int tid, int x, int y)
>> +{
>> + input_mt_slot(dev, slot);
>> + if (tid != 0) {
>> + input_report_abs(dev, ABS_MT_POSITION_X, x);
>> + input_report_abs(dev, ABS_MT_POSITION_Y, y);
>> + }
>> + input_report_abs(dev, ABS_MT_TRACKING_ID, tid);
>> + input_report_abs(dev, ABS_MT_TOOL_TYPE, MT_TOOL_FINGER);
>> + input_mt_sync(dev);
>> +}
>
>> +
>> +static void w8001_track_fingers(struct w8001 *w8001, struct w8001_touch *touch)
>> +{
>> + struct input_dev *dev = w8001->dev;
>> +
>> + if (touch->f1_status) {
>> + if (!w8001->tracking_id[0])
>> + w8001->tracking_id[0] = get_next_tracking_id();
>> + w8001_mt_event(dev, 0, w8001->tracking_id[0],
>> + touch->f1_x, touch->f1_y);
>> + } else if (w8001->tracking_id[0]) {
>> + w8001->tracking_id[0] = 0;
>> + w8001_mt_event(dev, 0, 0, touch->f1_x, touch->f1_y);
>> + }
>
>> +
>> + if (touch->f2_status) {
>> + if (!w8001->tracking_id[1])
>> + w8001->tracking_id[1] = get_next_tracking_id();
>> + w8001_mt_event(dev, 1, w8001->tracking_id[1],
>> + touch->f2_x, touch->f2_y);
>> + } else if (w8001->tracking_id[1]) {
>> + w8001->tracking_id[1] = 0;
>> + w8001_mt_event(dev, 1, 0, touch->f2_x, touch->f2_y);
>> + }
>> +
>> + input_sync(dev);
>> +}
>
>
> Apparently, f1 and f2 represent the slots themselves, i.e., the device uses
> slots internally. Please simplify the logic accordingly. There is an example in
> the recent patch for the Bamboo Touch.
>
>> +
>> static void report_pen_events(struct w8001 *w8001, struct w8001_coord *coord)
>> {
>> struct input_dev *dev = w8001->dev;
>> @@ -172,6 +253,7 @@ static irqreturn_t w8001_interrupt(struct serio *serio,
>> {
>> struct w8001 *w8001 = serio_get_drvdata(serio);
>> struct w8001_coord coord;
>> + struct w8001_touch touch;
>> unsigned char tmp;
>>
>> w8001->data[w8001->idx] = data;
>> @@ -217,10 +299,11 @@ static irqreturn_t w8001_interrupt(struct serio *serio,
>> complete(&w8001->cmd_done);
>> break;
>>
>> + /* 2 finger touch packet */
>> case W8001_PKTLEN_TOUCH2FG - 1:
>> - /* ignore two-finger touch packet. */
>> - if (w8001->pktlen == w8001->idx)
>> - w8001->idx = 0;
>> + w8001->idx = 0;
>> + parse_touch(w8001->data, w8001->pktlen, &touch);
>> + w8001_track_fingers(w8001, &touch);
>> break;
>> }
>>
>> @@ -282,6 +365,16 @@ static int w8001_setup(struct w8001 *w8001)
>> break;
>> case 5:
>> w8001->pktlen = W8001_PKTLEN_TOUCH2FG;
>> +
>> + input_mt_create_slots(dev, 2);
>> + input_set_abs_params(dev, ABS_MT_TRACKING_ID,
>> + 0, 255, 0, 0);
>> + input_set_abs_params(dev, ABS_MT_POSITION_X,
>> + 0, touch.x, 0, 0);
>> + input_set_abs_params(dev, ABS_MT_POSITION_Y,
>> + 0, touch.y, 0, 0);
>> + input_set_abs_params(dev, ABS_MT_TOOL_TYPE,
>> + 0, 0, 0, 0);
>> break;
>> }
>> }
>
>
> No information about signal-to-noise ratio (fuzz) for this device?
No.
Ping
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/