If you disable the mouse pointer from within ThinLinc, then the entire session crashes with this in the log: > (EE) > (EE) Backtrace: > (EE) 0: /opt/thinlinc/libexec/Xvnc (xorg_backtrace+0x41) [0x76d921] > (EE) 1: /opt/thinlinc/libexec/Xvnc (0x400000+0x370f29) [0x770f29] > (EE) 2: /lib64/libc.so.6 (0x7fd160482000+0x1a090) [0x7fd16049c090] > (EE) 3: /opt/thinlinc/libexec/Xvnc (miPointerGetPosition+0x87) [0x75d927] > (EE) 4: /opt/thinlinc/libexec/Xvnc (vncGetPointerPos+0x32) [0x593312] > (EE) 5: /opt/thinlinc/libexec/Xvnc (_ZN14XserverDesktop12blockHandlerEPi+0xea) [0x591a7a] > (EE) 6: /opt/thinlinc/libexec/Xvnc (vncCallBlockHandlers+0x27) [0x5864b7] > (EE) 7: /opt/thinlinc/libexec/Xvnc (BlockHandler+0x56) [0x723706] > (EE) 8: /opt/thinlinc/libexec/Xvnc (WaitForSomething+0xc5) [0x76b9d5] > (EE) 9: /opt/thinlinc/libexec/Xvnc (Dispatch+0xac) [0x71ecec] > (EE) 10: /opt/thinlinc/libexec/Xvnc (dix_main+0x388) [0x722dd8] > (EE) 11: /lib64/libc.so.6 (0x7fd160482000+0x3248) [0x7fd160485248] > (EE) 12: /lib64/libc.so.6 (__libc_start_main+0x8b) [0x7fd16048530b] > (EE) 13: /opt/thinlinc/libexec/Xvnc (0x400000+0xc4220) [0x4c4220] > (EE) > (EE) Segmentation fault at address 0x2c > (EE) > Fatal server error: > (EE) Caught signal 11 (Segmentation fault). Server aborting > (EE) This can easily be provoked by doing: > $ xinput set-int-prop 6 117 8 0 (this changes the property "enabled" for the VNC pointer device to 0) It is also available in the GUI in (at least) Xfce, where they have a checkbox for it. This is a regression from 4.17.0, where this crash cannot be observed. This issue was originally reported upstream: https://github.com/TigerVNC/tigervnc/pull/1894
It is unclear what caused this regression. It seems to have something to do with floating devices. But the bug cannot be triggered on 4.17.0 by setting the pointer device to be floating. This is the most likely candidate I can find, given the range of commits between the broken and working state: https://gitlab.freedesktop.org/xorg/xserver/-/commit/5c4816afa7722ea47d1a7dea983a953e7b454d26
That was a red herring. A git bisect showed that the issue is indeed connected to disabling the device. Specifically: https://gitlab.freedesktop.org/xorg/xserver/-/commit/8b75ec34dfbe435cd3a17e64138e22a37395a6d8
