[Spice-devel] help debugging: Assertion `!xcb_xlib_threads_sequence_lost' failed.

[Daniel P. Berrangé]

On Thu, Jan 23, 2020 at 11:54:50AM -0500, Cole Robinson wrote:
> Hi all,
> 
> For the past year+ we have had a steady trickle of virt-manager X11
> threading crashes. This is the f31 bug I've duped all the other ones
> too, but there were a handful that were auto closed by Fedora
> end-of-life bug cleanup: https://bugzilla.redhat.com/show_bug.cgi?id=1756065
> 
> The message that goes along with this is:
> 
> [xcb] Unknown sequence number while processing queue
> [xcb] Most likely this is a multi-threaded client and XInitThreads has
> not been called
> [xcb] Aborting, sorry about that.
> remote-viewer: xcb_io.c:263: poll_for_event: Assertion
> `!xcb_xlib_threads_sequence_lost' failed.
> Aborted (core dumped)
> 
> That may not apply to all the poll_for_error() bugs but it's most of the
> ones I have dug into.
> 
> I figured there was some threading issue in virt-manager and I was
> mostly ignoring the bugs until I found time to try and diagnose. I'm not
> really finding anything in virt-manager though, and some similar issues
> make me think maybe this is spice-gtk or something lower.
> 
> This user can reproduce the same error with remote-viewer + spice + X11
> ssh forwarding: https://bugzilla.redhat.com/show_bug.cgi?id=1758384
> 
> This user can reproduce the virt-manager crash multiple times a day, and
> it's always when interacting with a spice-gtk console:
> https://bugzilla.redhat.com/show_bug.cgi?id=1792576
> 
> Anyone have any ideas how to debug this further? What spice-gtk
> debugging info would be helpful?

FWIW if you're trying to reproduce this yourself, the key factor here
is that they are using X11, not Wayland. Given that Fedora defaults
to Wayland this likely explains why only certain people are seeing
it.

The error message strongly suggests that there are multiple threads
making X11 calls at the same time. This in turn means multiple threads
are making GTK calls at the same time, and this is expressly forbidden
by GTK - GTK APIs are only permitted to be invoked from the main thread
which runs the event loop.

It does look like a probable spice issue given the common crash in
remote-viewer, as aside from GTK itself and GTK-VNC, that's the only
UI library both apps use.

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 :|