Re: [PATCH v2 6/6] HID: uclogic: actually invert the in-range bit for huion tablets only

[Nikolai Kondrashov <spbnick@gmail.com>]

On 03/03/2015 12:25 AM, Benjamin Tissoires wrote:
> On Feb 28 2015 or thereabouts, Nikolai Kondrashov wrote:
>> On 02/26/2015 08:57 PM, Benjamin Tissoires wrote:
>>> This hack is only needed for Huion tablets. It does not seem to have
>>> any effect on the other tablets handled by this device right now, but
>>> it's better to check for the product id sooner than discovering that
>>> we have messed up one tablet later.
>>>
>>> Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
>>> ---
>>>
>>> New in v2
>>>
>>>   drivers/hid/hid-uclogic.c | 3 ++-
>>>   1 file changed, 2 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/drivers/hid/hid-uclogic.c b/drivers/hid/hid-uclogic.c
>>> index f44e72b..5f5c16e 100644
>>> --- a/drivers/hid/hid-uclogic.c
>>> +++ b/drivers/hid/hid-uclogic.c
>>> @@ -930,7 +930,8 @@ static int uclogic_raw_event(struct hid_device *hdev, struct hid_report *report,
>>>   	struct usb_interface *intf = to_usb_interface(hdev->dev.parent);
>>>
>>>   	/* If this is a pen input report */
>>> -	if (intf->cur_altsetting->desc.bInterfaceNumber == 0 &&
>>> +	if (hdev->product == USB_DEVICE_ID_HUION_TABLET &&
>>> +	    intf->cur_altsetting->desc.bInterfaceNumber == 0 &&
>>>   	    report->type == HID_INPUT_REPORT &&
>>>   	    report->id == 0x07 && size >= 2)
>>>   		/* Invert the in-range bit */
>>
>> Right, I completely forgot about this. That's why it's important to put these
>> things out for testing.
>
> yes, and no. A thorough review can help detect these while it's a pain
> to ask each and every users to test every possible device :)

Perhaps :)

>> However, this should be inverted not just for Huion tablets, but rather for
>> all "enabled" tablets. The UC-Logic TWHA60 v3 inverts that bit as well when
>> enabled, IIRC. We can filter enabled tablets by drvdata->rdesc != NULL, or add
>> an explicit flag.
>
> I like the flag. It will remove the dependency to usb in raw_event()
> which is always a plus.

Great!

> Speaking of which. I started looking at these dependencies in the hid
> subtree (like every year, in spring, these dependencies itches me).
> I think we can find a preventive solution which would allow to reinject
> through uhid any device even if the driver wants to talk to the usb
> layer. This will require some more effort on my side but will allow us
> to have at least some regressions tests in hid-uclogic too.

This sounds good. I was thinking about record/replay on HID level over uhid,
similar to evdev level, but as usual didn't have time to even think it through
(it's not my day job).

Nick