On Aug 17 2024, Aditya Garg wrote:
> From: Kerem Karabay <kekrby@xxxxxxxxx>
>
> This is needed to support Apple Touch Bars, where the contact ID is
> contained in fields with the HID_DG_TRANSDUCER_INDEX usage.
>
> Signed-off-by: Kerem Karabay <kekrby@xxxxxxxxx>
> Signed-off-by: Aditya Garg <gargaditya08@xxxxxxxx>
> ---
> drivers/hid/hid-multitouch.c | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/hid/hid-multitouch.c b/drivers/hid/hid-multitouch.c
> index 56fc78841..3e92789ed 100644
> --- a/drivers/hid/hid-multitouch.c
> +++ b/drivers/hid/hid-multitouch.c
> @@ -635,7 +635,9 @@ static struct mt_report_data *mt_allocate_report_data(struct mt_device *td,
>
> if (field->logical == HID_DG_FINGER || td->hdev->group != HID_GROUP_MULTITOUCH_WIN_8) {
> for (n = 0; n < field->report_count; n++) {
> - if (field->usage[n].hid == HID_DG_CONTACTID) {
> + unsigned int hid = field->usage[n].hid;
> +
> + if (hid == HID_DG_CONTACTID || hid == HID_DG_TRANSDUCER_INDEX) {
I'm not super happy about this. The HID spec specifically mentions
CONTACTID to be related to multitouch, when TRANSDUCER is not
specifically for multitouch. I would rather have this hidden behind a
quirk, because I don't think this is standard (the device comes from an
environment where both the hardware and the software stack is
controlled, which already gave some fun decisions from Apple).
Cheers,
Benjamin
> rdata->is_mt_collection = true;
> break;
> }
> @@ -814,6 +816,7 @@ static int mt_touch_input_mapping(struct hid_device *hdev, struct hid_input *hi,
> MT_STORE_FIELD(tip_state);
> return 1;
> case HID_DG_CONTACTID:
> + case HID_DG_TRANSDUCER_INDEX:
> MT_STORE_FIELD(contactid);
> app->touches_by_report++;
> return 1;
> --
> 2.43.0
>
