On Tue, 26 Dec 2017, Jason Gerecke wrote:
> Touch toggle softkeys send a '1' while pressed and a '0' while released,
> requring the kernel to keep track of wether touch should be enabled or
> disabled. The code does not handle the state transitions properly,
> however. If the key is pressed repeatedly, the following four states
> of states are cycled through (assuming touch starts out enabled):
>
> Press: shared->is_touch_on => 0, SW_MUTE_DEVICE => 1
> Release: shared->is_touch_on => 0, SW_MUTE_DEVICE => 1
> Press: shared->is_touch_on => 1, SW_MUTE_DEVICE => 0
> Release: shared->is_touch_on => 1, SW_MUTE_DEVICE => 1
>
> The hardware always properly enables/disables touch when the key is
> pressed but applications that listen for SW_MUTE_DEVICE events to provide
> feedback about the state will only ever show touch as being enabled while
> the key is held, and only every-other time. This sequence occurs because
> the fallthrough WACOM_HID_WD_TOUCHONOFF case is always handled, and it
> uses the value of the *local* is_touch_on variable as the value to
> report to userspace. The local value is equal to the shared value when
> the button is pressed, but equal to zero when the button is released.
>
> Reporting the shared value to userspace fixes this problem, but the
> fallthrough case needs to update the shared value in an incompatible
> way (which is why the local variable was introduced in the first place).
> To work around this, we just handle both cases in a single block of code
> and update the shared variable as appropriate.
>
> Fixes: d793ff8187 ("HID: wacom: generic: support touch on/off softkey")
> Cc: <stable@xxxxxxxxxxxxxxx> # v4.12+
> Signed-off-by: Jason Gerecke <jason.gerecke@xxxxxxxxx>
> Reviewed-by: Aaron Skomra <aaron.skomra@xxxxxxxxx>
> Tested-by: Aaron Skomra <aaron.skomra@xxxxxxxxx>
Applied to for-4.16/wacom.
--
Jiri Kosina
SUSE Labs