On Mon, Jan 29, 2024 at 08:47:48AM -0800, Bjorn Andersson wrote:
> The touchscreen present on some SKUs of Lenovo Thinkpad X13s is never
> detected by Linux. Power is applied and the device is brought out of
> reset using the pinconfig in DeviceTree, but the read-test in
> __i2c_hid_core_probe() fails to access the device, which result in probe
> being aborted.
>
> Some users have reported success after rebinding the device.
>
> Looking to the ACPI tables, there's a 5ms after-power and a 200ms
> after-reset delay. The power-supply is shared with other components, so
> this is active all the way through boot. The reset GPIO, on the other
> hand, is low (reset asserted) at boot, so this is first deasserted by
> the implicit application of the pinconf state.
>
> This means the time between reset deassert and __i2c_hid_core_probe() is
> significantly below the value documented in the ACPI tables.
>
> As the I2C HID binding and driver support specifying a reset gpio,
> replace the pinconf-based scheme to pull the device out of reset. Then
> specify the after-reset time.
>
> The shared power rail is currently always on, but in case this ever
> change, the after-power delay is added as well, to not violate the
> power-on to reset-deassert timing requirement.
>
> Fixes: 32c231385ed4 ("arm64: dts: qcom: sc8280xp: add Lenovo Thinkpad X13s devicetree")
> Tested-by: Daniel Thompson <daniel.thompson@xxxxxxxxxx>
For the sake of completeness (and to make sure I don't forget if anyone
asks me in the future) I also (successfully) tested with the output
high removed from the touchscreen pinctrl.
Daniel.
PS Strictly speaking I tested v2 and am replying to v3... but the DT
should be the same for both versions ;-).
