Huion Kamvas 16 Gen 3 cannot exceed 16-bit X coordinates

[higginsdragon]

• Device name: Huion Kamvas 16 (Gen 3)

• Device model identifier: GS1563

• libwacom version: master branch

• I understand that libwacom does not affect whether the device works (see Troubleshooting)

Bug description

I am not sure if this is something in libwacom, seems like it might be more of a udev-hid-bpf issue, just trying to track it down, and would be more than happy to file a bug there as long as I knew what I was looking for specifically. Might even be deeper than that.

I commented on this briefly in #881, but since it's not directly related to the Kavmas 13 opened this issue for better tracking.

The problem seems to be that udev-hid-bpf may not handle horizontal scale beyond 16-bits. The size of the Kamvas 16 is 69920. I wrote a .c testing file for the Kamvas 16 in udev-hid-bpf (master branch) and after compiling most everything works fine, the buttons, dials, etc. The stylus works too, with this one problem.

From hid-reporter:

E: 000001.294037 14 08 80 23 fe e0 12 00 00 00 00 0a 0b 03 00
# ReportID: 8 / Tip Switch: 0 | Barrel Switch: 0 | Secondary Barrel Switch: 0  0 | # | In Range: 1 | X:  65137 | Y:   4841 | Tip Pressure:      0 | # | X Tilt:   10 | Y Tilt:   11 
E: 000001.298038 14 08 80 71 fe e9 12 00 00 00 00 0a 0b 03 00
# ReportID: 8 / Tip Switch: 0 | Barrel Switch: 0 | Secondary Barrel Switch: 0  0 | # | In Range: 1 | X:  65215 | Y:   4850 | Tip Pressure:      0 | # | X Tilt:   10 | Y Tilt:   11 
E: 000001.301037 14 08 80 bf fe f2 12 00 00 00 00 0a 0b 03 00
# ReportID: 8 / Tip Switch: 0 | Barrel Switch: 0 | Secondary Barrel Switch: 0  0 | # | In Range: 1 | X:  65294 | Y:   4859 | Tip Pressure:      0 | # | X Tilt:   10 | Y Tilt:   11 
E: 000001.306018 14 08 80 0e ff fb 12 00 00 00 00 0a 0b 03 00
# ReportID: 8 / Tip Switch: 0 | Barrel Switch: 0 | Secondary Barrel Switch: 0  0 | # | In Range: 1 | X:  65373 | Y:   4867 | Tip Pressure:      0 | # | X Tilt:   10 | Y Tilt:   11 
E: 000001.309049 14 08 80 5d ff 03 13 00 00 00 00 0a 0b 03 00
# ReportID: 8 / Tip Switch: 0 | Barrel Switch: 0 | Secondary Barrel Switch: 0  0 | # | In Range: 1 | X:  65451 | Y:   4875 | Tip Pressure:      0 | # | X Tilt:   10 | Y Tilt:   11 
E: 000001.313035 14 08 80 ab ff 0b 13 00 00 00 00 0a 0b 03 00
# ReportID: 8 / Tip Switch: 0 | Barrel Switch: 0 | Secondary Barrel Switch: 0  0 | # | In Range: 1 | X:  65527 | Y:   4883 | Tip Pressure:      0 | # | X Tilt:   10 | Y Tilt:   11 
E: 000001.317036 14 08 80 f7 ff 13 13 00 00 00 00 0a 0b 03 00
# ReportID: 8 / Tip Switch: 0 | Barrel Switch: 0 | Secondary Barrel Switch: 0  0 | # | In Range: 1 | X:     67 | Y:   4891 | Tip Pressure:      0 | # | X Tilt:   10 | Y Tilt:   11 
E: 000001.321026 14 08 80 43 00 1b 13 00 00 01 00 0a 0b 03 00
# ReportID: 8 / Tip Switch: 0 | Barrel Switch: 0 | Secondary Barrel Switch: 0  0 | # | In Range: 1 | X:    142 | Y:   4899 | Tip Pressure:      0 | # | X Tilt:   10 | Y Tilt:   11 
E: 000001.324034 14 08 80 8e 00 23 13 00 00 01 00 0a 0b 03 00
# ReportID: 8 / Tip Switch: 0 | Barrel Switch: 0 | Secondary Barrel Switch: 0  0 | # | In Range: 1 | X:    217 | Y:   4907 | Tip Pressure:      0 | # | X Tilt:   10 | Y Tilt:   11 

From the magic bytes I got from huion-switcher:

* 	HUION_FIRMWARE_ID="HUION_M22d_241101"
 * 	HUION_MAGIC_BYTES="1403201101ac9900ff3fd81305080080083c4010"
 *
 * 	MAGIC BYTES
 * 	          [LogicalMaximum, X   ] [LogicalMaximum, Y   ] [LogicalMaximum, Pressure] [  LPI]
 * 	    14 03 [            20 11 01] [            ac 99 00] [                   ff 3f] [d8 13] 05 08 00 80 08 3c 40 10

The LogicalMaximum is actually a 24-bit value, though in the other files it's referenced as a 16-bit value. In this case, it all checks out.

LogicalMaximum is 69920 which when divided by the LPI (5080 LPI == 200 lines/mm) is 349.6mm which matches the Kamvas 16 specs at 350mm width.

Just changing the values in udev-hid-bpf from LogicalMaximum_i16 to LogicalMaximum_i32 results in the hid-reporter output above. As I suspected, it's rolling over at 65535.

I am looking for the place to fix this, not sure if it's in libwacom, udev-hid-bpf, or even lower than that at the udev or kernel level?

Or maybe I'm missing something obvious, which is also a high possibility. Just looking to be pointed in the right direction.

[higginsdragon]

I did a USB trace to see if I can figure out more of what's going on. This is the data returned as I cross over the 16-bit boundary:

125	0.204971	1.2.1	host	USB	78	URB_INTERRUPT in
0040   08 80 e9 ff 4e 91 00 00 00 00 11 f9 03 00         ....N.........
                               --

127	0.208968	1.2.1	host	USB	78	URB_INTERRUPT in
0040   08 80 03 00 53 91 00 00 01 00 11 f9 03 00         ....S.........
                               --

So byte 8 changes to 01 once it crosses the threshold, so the device is definitely sending the correct data.

Probably something in the interpretation especially considering this is a Huion device.

[github-actions]

Hi, there. I'm a bot and have been asked to have a look at this.

libwacom is a bunch of text files with a C API, and the information in those text files
is static fixed information that we cannot get from the device (e.g. the model name or which
button will produce which event code). It does not affect functionality directly, if the
tablet does not work this is either an issue with the kernel drivers, xf86-input-wacom
(if using X) or libinput/the desktop environment (if using Wayland). Please file the
issue with the component that most likely applies.

See also https://github.com/linuxwacom/libwacom/wiki/Troubleshooting
See also https://github.com/linuxwacom/libwacom/wiki/Adding-a-new-device

---

This is an automated comment created by a bot. Responding to the bot or mentioning it won't have any effect.

[whot]

Oh, if this tablet is one that is enabled via udev-hid-bpf this may be a fault in the BPF program