[REPORT] synaptics_intertouch on ThinkPad T14 Gen 2 (LEN2072)

[REPORT] synaptics_intertouch on ThinkPad T14 Gen 2 (LEN2072)

[Nick Green]

Hi all,

Following up on my previous report regarding synaptics_intertouch on the ThinkPad T14 Gen 2 (PNP: LEN2072).

After further daily-use testing, please DO NOT add LEN2072 to the smbus_pnp_ids whitelist.

Key findings:
1. Under psmouse.synaptics_intertouch=1 (RMI4/SMBus), the touchpad exhibits intermittent tap-to-click failures during natural use.
2. Low-level event debugging via 'libinput measure touchpad-pressure' confirmed that missed taps produce zero evdev events at the kernel level—momentary tap interrupts are being dropped by the SMBus controller before reaching userspace, while continuous swiping and pointer motion transmit normally.

The default PS/2 driver via the Embedded Controller remains necessary on this platform for reliable, zero-drop tap detection.

Thanks,
Nick

Re: [REPORT] synaptics_intertouch on ThinkPad T14 Gen 2 (LEN2072)

[Dmitry Torokhov]

Hi Nick,

On Sat, Aug 29, 2026 at 08:44:23PM +0000, Nick Green wrote:
> Following up on my previous report regarding synaptics_intertouch on the ThinkPad T14 Gen 2 (PNP: LEN2072).
>
> After further daily-use testing, please DO NOT add LEN2072 to the smbus_pnp_ids whitelist.
>
> Key findings:
> 1. Under psmouse.synaptics_intertouch=1 (RMI4/SMBus), the touchpad exhibits intermittent tap-to-click failures during natural use.
> 2. Low-level event debugging via 'libinput measure touchpad-pressure' confirmed that missed taps produce zero evdev events at the kernel level—momentary tap interrupts are being dropped by the SMBus controller before reaching userspace, while continuous swiping and pointer motion transmit normally.

Dmitry Mikushin (CCed) recently tracked down a performance issue with
Synaptics TM3471 (LEN2072) over SMBus where rmi_f12 was reading the
entire 79-byte F12 data packet on every attention interrupt (requiring
three SMBus block reads and taking ~17.4 ms per frame, which capped the
report rate at ~43 Hz), even though only the first 32 bytes (Data1) are
used.

Could you please test his patch series (now in the input next branch)
that limits the F12 read to Data1 and enables SMBus for LEN2072, and
let us know if it resolves the missed taps on your T14 Gen 2:

https://lore.kernel.org/linux-input/20261026142648.2567233-1-dmitry@kernelgen.org/

Thanks.

-- 
Dmitry