I'm using Gnome's included VNC server (vino-server) to allow remote access from VNC (usually realvnc on windows). Occationally after disconnecting, vino-server CPU usage on the remote (vino-server) machine goes to 100%. It doesn't happen all of the time, and it's difficult to duplicate. There are no error messages in the process. Killing vino-server fixes the issue until next usage. While I couldn't find a related bug in the gentoo forums, I did see the same issue on an ubuntu forums: https://launchpad.net/distros/ubuntu/+source/vino/+bug/31037 There is a suggested patch that fixes the problem by not including clients "on hold" in the GTK main loop. I am weary to use this patch on my system, since it is a different distribution and I have limited experience with patches. I have experienced this bug for over a year, through many system updates. I still find the vino-server process eating 100% CPU occationally after disconnect. Can the patch be adapted for gentoo? Is is the same issue?
Created attachment 98383 [details, diff] Patch found on an ubuntu forum (author Gary Coady)
GNOME Upstream bug for this is http://bugzilla.gnome.org/show_bug.cgi?id=332011 (forwarded from the Ubuntu one). I have experienced the problem myself too. It happens after you get the password wrong. As a workaround I have killed the vino-server over ssh (while VNC is tunnelled through ssh anyway for me), restarted it over ssh (background + disown if necessary) and tried to get the password right on a new try, to not have to do it all again :) Upstream seems to find the patch suitable for commit after identation fixes, but I'm not sure what's up with their responsiveness (on that bug, and last vino release being 2.3.5...) I think we should try the patch out and go with it if it's fine.
The GNOME release team picked the package up, committed gentoo provided 3 patches and the 100% CPU bug fixing patch and released 2.16.0. I have added it to the tree in ~arch, and tested that it fixes the CPU hogging problem after invalid password for me.