I was checking the status of some GTK bots and I noticed that we had >600!!! at-spi2-bus-launcher processes with their corresponding dbus-daemon processes I think the problem is that we unconditionally launch them in _start_accessibility_daemons() in run-gtk-tests
Created attachment 210470 [details] Patch proposal (In reply to comment #0) > I was checking the status of some GTK bots and I noticed that we had >600!!! at-spi2-bus-launcher processes with their corresponding dbus-daemon processes > > I think the problem is that we unconditionally launch them in _start_accessibility_daemons() in run-gtk-tests Actually, this should not be an issue since we already take care of terminating the process that we start in _tear_down_testing_environment(). The problem was that, as silly as it might sound, we were using the wrong variable name (_ally_bus_launcher) to keep a reference to that process, so later on we were not terminating it properly (as we were expecting _spi_bus_launcher in _tear_down_testing_environment()).
Comment on attachment 210470 [details] Patch proposal Clearing flags on attachment: 210470 Committed r155606: <http://trac.webkit.org/changeset/155606>
All reviewed patches have been landed. Closing bug.