On Mon, Feb 10, 2020 at 01:13:55PM -0500, Cole Robinson wrote: > I attempted to review some patches on Friday and started hitting > occasional crashes via eventtest.c. Long story short it's a glib bug: > > https://gitlab.gnome.org/GNOME/glib/merge_requests/1358 > > It's a ref counting issue caused when g_source_remove > (virEventRemoveHandle) is called from one thread, while the main loop is > in a particular state in a different thread. > > The way vireventglib is implemented means every user initiated > g_source_remove is likely called from a different thread so we risk > hitting this. Not sure how likely it is in realworld usage, > vireventtest.c is pretty pathologic in this area. We could change > vireventglib.c to do the final source_unref from the idle callback which > would avoid the problem
I think libvirtd itself is fairly pathological wrt event loop too, so we ought to try to workaround it IMHO. Regards, Daniel -- |: https://berrange.com -o- https://www.flickr.com/photos/dberrange :| |: https://libvirt.org -o- https://fstop138.berrange.com :| |: https://entangle-photo.org -o- https://www.instagram.com/dberrange :|
