Re: [PATCH] HID: multitouch: reclassify HTIX5288 to WIN_8_FORCE_MULTI_INPUT_NSMU

On Sat, 11 Jul 2026, Xianglin Lin wrote:

> Commit b5e65ae557da ("HID: multitouch: Add quirk for Hantick 5288
> touchpad") assigned MT_CLS_NSMU to the HTIX5288 (0911:5288).  This was
> necessary because the device sometimes fails to send touch release
> signals when transitioning from >=2 fingers to <2 fingers, and
> MT_QUIRK_NOT_SEEN_MEANS_UP fixes stuck touches by treating missing
> contacts as released.
> 
> However, MT_CLS_NSMU only carries MT_QUIRK_NOT_SEEN_MEANS_UP.  It
> lacks MT_QUIRK_CONTACT_CNT_ACCURATE and MT_QUIRK_IGNORE_DUPLICATES.
> As a result, after a two-finger scroll finger lift, the device still
> reports stale coordinates from the released contact in subsequent
> frames, and the driver overwrites the remaining active slot with
> those frozen coordinates.  The remaining finger appears stuck at the
> lift position until all fingers are lifted.
> 
> This was confirmed via evtest on Arch Linux 7.1.3: after
> TRACKING_ID=-1 for the released slot, every subsequent frame contained
> duplicate position pairs -- the real moving finger's coordinates
> followed by the lifted finger's frozen position, both attributed to
> the active slot.
> 
> Reclassify the device to MT_CLS_WIN_8_FORCE_MULTI_INPUT_NSMU
> (0x0018), which preserves the original MT_QUIRK_NOT_SEEN_MEANS_UP
> fix while adding the necessary Win8 quirks (CONTACT_CNT_ACCURATE,
> IGNORE_DUPLICATES), preventing stale coordinate contamination.
> 
> The additional FORCE_MULTI_INPUT flag is harmless here: it separates
> the mouse and touchpad collections into distinct input devices,
> which is the standard behavior libinput already expects.
> 
> Fixes: b5e65ae557da ("HID: multitouch: Add quirk for Hantick 5288 touchpad")
> Signed-off-by: Xianglin Lin <[email protected]>

Applied, thanks.

-- 
Jiri Kosina
SUSE Labs