Pls run evemu-describe against the device and attach the output here, thanks.

Created attachment 114099 [details]
evemu-describe of the device (/dev/input/event4)

what's your evdev version?

zypper se -si evdev:

S | Name               | Version     | Arch   | Repository         
--+--------------------+-------------+--------+--------------------
i | libevdev2          | 1.2-2.2.2   | x86_64 | openSUSE-13.2      
i | xf86-input-evdev   | 2.9.0-3.1.4 | x86_64 | openSUSE-13.2

alright, this device is a fake MT device and evdev doesn't handle this correctly.
 
http://www.freedesktop.org/software/libevdev/doc/latest/group__mt.html has a bit of an explanation.

http://patchwork.freedesktop.org/patch/44409/
http://patchwork.freedesktop.org/patch/44410/
http://patchwork.freedesktop.org/patch/44411/
http://patchwork.freedesktop.org/patch/44412/
http://patchwork.freedesktop.org/patch/44413/

not a complete fix, but at least it quietens the current issue

commit 38e107a39fb4a0b630ee5adb5870c91dbc27abde
Author: Peter Hutterer <peter.hutterer@who-t.net>
Date:   Wed Mar 11 14:58:04 2015 +1000

    Disable axis faking and MT event processing on fake MT devices

*** Bug 86924 has been marked as a duplicate of this bug. ***

I am seeing what looks like the same issue, except it is accompanied by poor quality sounds that seems to degrade with other keyboard, mouse, or window events.

This is with Debian Jessie, I have tried Debian Sid package xserver-xorg-input-evdev package.

I tried building from the git repository on January 15th (and overwrote /usr/lib/xorg/modules/input/evdev_drv.so)

Neither of these changed the behaviour, the error is reported and the degrading sound behaviour continues (not seen with Wheezy or Windows 10).

Are these changes confirmed to fix this issue? e.g. should I open a new bug.

if you still get the error, please file a new bug and attach the evemu-describe output there. This bug was fixed, but a different device may trigger a similar issue in a different way.

The sound quality going down isn't related, at least not directly. The error messages is for a known case and we handle it, the only thing the driver does is print the backtrace and continue (so it's not a crash either). Maybe there's something like abrt running that spots the backtrace and then takes CPU time away? Either way, new bug please, thanks.

Could you please tell me whether it got fixed in version 2.9.2? I got the following versions updated from the originally reported ones:

S | Name             | Type    | Version   | Arch   | Repository         
--+------------------+---------+-----------+----------------------------
i | libevdev2        | package | 1.4.4-1.1 | x86_64 | openSUSE-Leap_42.1
i | xf86-input-evdev | package | 2.9.2-5.2 | x86_64 | openSUSE-Leap_42.1

And the following backtrace still happens, but now with additional lines about axis faking:

[...]
[   325.358] (**) Option "config_info" "udev:/sys/devices/pci0000:00/0000:00:1a.7/usb7/7-5/7-5.3/7-5.3:1.3/0003:046D:0A29.0006/input/input7/event4"
[   325.358] (II) XINPUT: Adding extended input device "Logitech Logitech Wireless Headset" (type: KEYBOARD, id 10)
[   325.358] (**) Option "xkb_rules" "evdev"
[   325.358] (**) Option "xkb_model" "pc105"
[   325.358] (**) Option "xkb_layout" "hu"
[   325.358] (**) Option "xkb_options" "terminate:ctrl_alt_bksp"
[   325.358] (II) evdev: Logitech Logitech Wireless Headset: Faking axis ABS_X.
[   325.358] (II) evdev: Logitech Logitech Wireless Headset: Faking axis ABS_Y.
[   325.358] (II) evdev: Logitech Logitech Wireless Headset: Faking axis ABS_PRESSURE.
[   325.358] (II) evdev: Logitech Logitech Wireless Headset: Faking axis ABS_DISTANCE.
[   325.358] (EE) BUG: triggered 'if (axnum >= dev->valuator->numAxes)'
[   325.358] (EE) BUG: exevents.c:2087 in InitValuatorAxisStruct()
[   325.358] (EE)
[   325.358] (EE) Backtrace:
[   325.358] (EE) 0: /usr/bin/X (xorg_backtrace+0x48) [0x58b0c8]
[   325.358] (EE) 1: /usr/bin/X (InitValuatorAxisStruct+0x68) [0x51f748]
[   325.358] (EE) 2: /usr/lib64/xorg/modules/input/evdev_drv.so (0x7f044a6ca000+0x5479) [0x7f044a6cf479]
[   325.358] (EE) 3: /usr/lib64/xorg/modules/input/evdev_drv.so (0x7f044a6ca000+0x590f) [0x7f044a6cf90f]
[   325.358] (EE) 4: /usr/lib64/xorg/modules/input/evdev_drv.so (0x7f044a6ca000+0x763b) [0x7f044a6d163b]
[   325.358] (EE) 5: /usr/bin/X (ActivateDevice+0x37) [0x432d97]
[   325.358] (EE) 6: /usr/bin/X (0x400000+0x88fa6) [0x488fa6]
[   325.358] (EE) 7: /usr/bin/X (0x400000+0x9e9db) [0x49e9db]
[   325.359] (EE) 8: /usr/bin/X (0x400000+0x9efa3) [0x49efa3]
[   325.359] (EE) 9: /usr/bin/X (config_init+0x9) [0x49d9f9]
[   325.359] (EE) 10: /usr/bin/X (InitInput+0xab) [0x47cc0b]
[   325.359] (EE) 11: /usr/bin/X (0x400000+0x40f79) [0x440f79]
[   325.359] (EE) 12: /lib64/libc.so.6 (__libc_start_main+0xf5) [0x7f04508e3b05]
[   325.359] (EE) 13: /usr/bin/X (0x400000+0x2c5de) [0x42c5de]
[   325.359] (EE)

no, the fixes are on the master branch only (and in evdev 2.10.0), not on the 2.9 branch.