Re: [DIGImend-devel] New Monoprice Tablet using UC-Logic 5543:0781

[Nikolai K. <sp...@gm...>]

Hi Jerry,

First of all, thanks for a comprehensive report and for isolating the
problem.

On 08/23/2012 07:11 AM, Jerry Siebe wrote:
> On Mon, Aug 20, 2012 at 10:49 PM, Jerry Siebe<jer...@gm...>  wrote:
>> I spent some time with the tablet today and found some bad behaviors.
>> This is using your 3.6 patch applied to against a 3.5.2 kernel (as
>> patched by Gentoo).
>>
>> First, having the tablet plugged in when computer boots up left the
>> tablet in a bad state.  It would detect pen and frame buttons (unsure
>> about pressure, didn't think to check) but would not detect position
>> of the pen.  The LED on the tablet usually lights up with any
>> activity, including the pen being close to the surface.  It was not
>> lighting at all except when a button (including the tip) was pressed.
>> Cursor on screen also showed no motion.  I've found it in this state
>> after booting up twice.  Unplugging it and plugging it back in fixes
>> it.
>>
>>  From dmesg:
>> [    5.339450] input: HA60S as
>> /devices/pci0000:00/0000:00:12.2/usb1/1-4/1-4.1/1-4.1:1.0/input/input5
>> [    5.339647] uclogic 0003:5543:0781.0004: input,hidraw3: USB HID
>> v1.10 Device [HA60S] on usb-0000:00:12.2-4.1/input0
>> [    5.340017] input: HA60S as
>> /devices/pci0000:00/0000:00:12.2/usb1/1-4/1-4.1/1-4.1:1.1/input/input6
>> [    5.340056] usb 1-4.1: link qh8-0e01/ffff88012b0e9d80 start 2 [1/2 us]
>> [    5.340223] uclogic 0003:5543:0781.0005: input,hidraw4: USB HID
>> v1.10 Keyboard [HA60S] on usb-0000:00:12.2-4.1/input1
>>
>> Starting Xorg
>> [   82.341530] usb 1-4.1: link qh4-0e01/ffff88012b319880 start 3 [1/2 us]
>>
>> Then when trying to use it I get this:
>> [ 5986.741810] usb 1-4.1: unlink qh4-0e01/ffff88012b319880 start 3 [1/2 us]
>> [ 5986.742146] ehci_hcd 0000:00:12.2: detected XactErr len 0/8 retry 1
>> ...
>> [ 5986.748635] ehci_hcd 0000:00:12.2: detected XactErr len 0/8 retry 27
>> [ 5986.748770] usb 1-4.1: unlink qh8-0e01/ffff88012b0e9d80 start 2 [1/2 us]
>> [ 5986.749009] ehci_hcd 0000:00:12.2: detected XactErr len 0/8 retry 28
>> ...
>> [ 5986.749759] ehci_hcd 0000:00:12.2: detected XactErr len 0/8 retry 31
>> [ 5986.750008] ehci_hcd 0000:00:12.2: devpath 4.1 ep0out 3strikes
>> [ 5986.750016] usb 1-4: clear tt buffer port 1, a5 ep0 t00080248
>> [ 5986.750072] ehci_hcd 0000:00:12.2: reused qh ffff88012b319880 schedule
>> [ 5986.750080] usb 1-4.1: link qh4-0e01/ffff88012b319880 start 3 [1/2 us]
>> [ 5986.750093] uclogic 0003:5543:0781.0004: can't reset device,
>> 0000:00:12.2-4.1/input0, status -71
>> [ 5986.750509] ehci_hcd 0000:00:12.2: detected XactErr len 0/8 retry 1
>> ...
>> and on and on.

> I've now isolated it to having the tablet plugged in during a cold
> boot that it ends up in that bad state.  Once it's in that state,
> rebooting does not fix it, however unplugging and plugging it back in
> does.  A dozen reboots into different kernel builds earlier never left
> it in that bad state.  It wasn't until I remembered that the power had
> gone out (twice) the evening I was doing my prior testing that I
> thought to test a cold boot.
>
> I don't know where to even begin trying to debug this, and now that I
> know how to avoid it I'm not worried about it. However, if someone
> else is interested in this issue and willing to guide me, I'm willing
> to help.

This might be spontaneous compatibility/proprietary mode switching we've
seen with UC-Logic Tablet WP1700U. Some discussion of the problem can be
found in this thread:

http://sourceforge.net/mailarchive/forum.php?thread_name=4FFDD5B9.2060606%40gmail.com&forum_name=digimend-devel

Could you please try running usbhid-dump and capturing pen movement, when
you reproduce this problem again?

If this is verified, then I'll finally need to figure out UC-Logic
proprietary mode switching method.

>> Second bad behavior was during about twenty minutes of use. Three
>> times I got spurious lines while drawing. Twice the lines jumped to
>> the 0,0 corner, and once it jumped only vertically to y=0.  There is
>> nothing in dmesg at the times this happened.

> As for the occasional spurious line that shoots off to the 0.0
> position, I was able to replicate it in 3.5.0 as well, though it took
> 12 minutes of continuously having the stylus down.  It seems to happen
> more frequently in 3.5.2, where I can observe it within a few minutes.
>   It has happened in both MyPaint and Gimp.

It is possible that this is a timing issue and depends on the system load.
I suspect an X.org or xf86-input-evdev bug. I have something similar
reported in this bug:

http://sourceforge.net/tracker/?func=detail&aid=3557017&group_id=233297&atid=1089171

I'm yet to investigate it properly.

> It seems I can't both draw with the tablet and use usbhid-dump at the
> same time, and I don't know what else to do to provide more
> information.

You can use debugfs to capture input from your tablet while having it sent
to userspace also. Look into /sys/kernel/debug/hid.

Sincerely,
Nick