On Oct 09 2015 or thereabouts, Mika Westerberg wrote:

On Thu, Oct 08, 2015 at 06:10:23PM -0700, Andrew Duggan wrote:

quoted

No, this touchpad doesn't have HID_QUIRK_NO_INIT_REPORTS set and
usbhid/hid-core.c is calling usbhid_init_reports() which is retrieving
feature 8. But, I did some additional testing and I wasn't able to reproduce
this issue on other systems. The failure occurs on a Dell XPS 13 9343, but
the exact same touchpad and the exact same kernel (I booted both systems off
an external HD to ensure that the kernel and OS are identical) I didn't see
a problem. I also tried a different USB PTP touchpad which didn't have an
issue. Also, I tried returning in mt_get_feature() instead of reading the
report when the report id is 8. When I did that everything worked.

So as of right now, it looks like a single non production USB PTP touchpad
fails on one particular system. But, all other tests have been successful.

Let me know if you have any other suggestions for me to test. But, it seems
like this failure might be fairly isolated.

Thanks a lot for testing all this.

If the failure is limited to a single non-production touchpad then, I
suppose we should be fine. As long as there are no such systems out
there in the wild.

One additional thing we could check in mt_get_feature() is if the device
is connected via I2C and only in that case fetch the feature in
question.

Jiri, Benjamin, what do you think?

I think we are fine with the current patch. The windows driver does
retrieve the important features during the initialization so with this
patch we are closer than ever than what it does.

If (and only if) we ever encounter a problem later on, we can always
make this patch only for i2c devices, but I would still try to match the
Windows driver as much as possible.

So my reviewed-by still applies.

Cheers,
Benjamin