On Fri, Sep 14, 2018 at 12:05 PM Hanno Zulla <abos@xxxxxxxx> wrote:
>
> Hi Benjamin,
>
> [ Sending this with cc to kt.liao in the hope that he can chime in.
> This is about a touchpad with USB id 04f3:074a on the Medion S2218
> that works as multitouch on Win10, but only shows up as a
> single-click mouse in Linux. ]
>
>
> "udevadm info -e" does not return any ACPI id for this touchpad, so
> this isn't a case like commit 6f88a6439da5d94de334a341503bc2c7f4a7ea7f.
>
>
> I had ignorantly tried adding the USB id to hid-elan.c with
> { HID_USB_DEVICE(USB_VENDOR_ID_ELAN, 0x074a) }
> but that didn't work. hid-elan.c responds with
> "Input device is not registred" (sic).
>
>
> So per your request I ran Win10 on an emulated qemu box (it's
> amazing that the Z3735F supports kvm, but boy is it sloooow) and
> got usb passthrough and capture going.
>
> The touchpad hails in Windows as "Elan Smart-Pad" connected to USB.
> The driver calls itself "ELAN 11.7.11.14_X86_WHQL".
>
> The Elan-branded support application to configure the touchpad in
> Win10 does not give any additional hardware revision information
> about the device.
>
>
> I've tried to make sense of the .pcap file, but my knowledge of the
> USB HID protocol is still limited. The capture shows that the device
> keeps sending its HID report descriptor upon request, but for each
> request in my capture, it sends the same one, although I used it as
> a multitouch device during the capture session.
>
> So there is no obvious mode switch to a different HID descriptor,
> at least it isn't visible to me when looking at it in wireshark.
>
> Let me know where I shall send the USB .pcap wireshark file. I assume
> that you rather want it sent to a bug tracker issue or direct mail.
Can you open a bug on https://bugzilla.kernel.org/, CC me and attach
the capture there?
Also, in the capture, please make sure to record some multitouch
actions that are handled as multitouch actions by Windows 10.
Cheers,
Benjamin
>
> Kind regards & thanks,
>
> Hanno
