Screen freeze before the login screen (if the third party software option was selected)

[BertN45]

[ Workaround ]

1. Install *without* the third-party option.

2. sudo apt install ubuntu-restricted-extras

3. sudo apt remove gstreamer1.0-vaapi

[ Original Description ]

The problem started to appear 2 weeks ago after one of my daily updates. It also happens after I re-install the system from a freshly downloaded ISO file. The system freezes at the moment, it starts the gnome display manager.

I file the bug-report after a partial upgrade just before I will reboot the system. If needed I can reboot in recovery mode, since the gnome display manager is not started in that case.

I tried the continue button without effect and also "apt upgrade" and "apt dist-upgrade" do have the same effect

I run the VM from the Host-OpenZFS file system, so I can restore to before the update. Recently I use a Virtualbox snapshot to restore the system to before the update/upgrade.

If you need log files, let me know.

ProblemType: Bug
DistroRelease: Ubuntu 22.10
Package: systemd 251.4-1ubuntu1
ProcVersionSignature: Ubuntu 5.15.0-27.28-generic 5.15.30
Uname: Linux 5.15.0-27-generic x86_64
ApportVersion: 2.23.0-0ubuntu1
Architecture: amd64
CasperMD5CheckResult: pass
CurrentDesktop: ubuntu:GNOME
Date: Thu Sep 1 19:42:04 2022
InstallationDate: Installed on 2022-05-28 (96 days ago)
InstallationMedia: Ubuntu 22.10 "Kinetic Kudu" - Alpha amd64 (20220526)
Lsusb:
 Bus 002 Device 001: ID 1d6b:0003 Linux Foundation 3.0 root hub
 Bus 001 Device 002: ID 80ee:0021 VirtualBox USB Tablet
 Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
Lsusb-t:
 /: Bus 02.Port 1: Dev 1, Class=root_hub, Driver=xhci_hcd/6p, 5000M
 /: Bus 01.Port 1: Dev 1, Class=root_hub, Driver=xhci_hcd/8p, 480M
     |__ Port 1: Dev 2, If 0, Class=Human Interface Device, Driver=usbhid, 12M
MachineType: innotek GmbH VirtualBox
ProcKernelCmdLine: BOOT_IMAGE=/boot/vmlinuz-5.15.0-27-generic root=UUID=9eaf84b0-b69b-4760-80b5-310e3a258bda ro quiet splash
RebootRequiredPkgs: Error: path contained symlinks.
SourcePackage: systemd
UpgradeStatus: No upgrade log present (probably fresh install)
dmi.bios.date: 12/01/2006
dmi.bios.vendor: innotek GmbH
dmi.bios.version: VirtualBox
dmi.board.name: VirtualBox
dmi.board.vendor: Oracle Corporation
dmi.board.version: 1.2
dmi.chassis.type: 1
dmi.chassis.vendor: Oracle Corporation
dmi.modalias: dmi:bvninnotekGmbH:bvrVirtualBox:bd12/01/2006:svninnotekGmbH:pnVirtualBox:pvr1.2:rvnOracleCorporation:rnVirtualBox:rvr1.2:cvnOracleCorporation:ct1:cvr:sku:
dmi.product.family: Virtual Machine
dmi.product.name: VirtualBox
dmi.product.version: 1.2
dmi.sys.vendor: innotek GmbH

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[BertN45 (lammert-nijhof)]

This bug is still present and has nothing to do with bug /1988473

[BertN45 (lammert-nijhof)]

I never used the package 6.1.38-dfsg-3~ubuntu1.20.04.1 or the 6.1.34.and 6.1.36 version of it.

I always load the virtualbox package directly from virtualbox.org, so I used virtualbox-6.1_6.1.34-150636.1~Ubuntu~eoan_amd64; virtualbox-6.1_6.1.36-152435~Ubuntu~jammy_amd64 and virtualbox-6.1_6.1.38-153438~Ubuntu~jammy_amd64

I never had any problem with these Oracle vbox packages not in Ubuntu 22.04; Xubuntu 22.04; Ubuntu Studio 20.04; Ubuntu 16.04 ESM nor in one of the other favors and remixes of 22.04 and 20.04.

The problem does not show up as a crash, but the screen always freezes just before the login screen should appear. It only freezes in Ubuntu 22.10 and not in Ubuntu Mate 22.10

[BertN45 (lammert-nijhof)]

Today I installed Ubuntu Unity 22.10 and that one has exactly the same issue. It will freeze after it starts the gnome display manager. In this case, it did run without the Virtualbox Guest Additions.

[Daniel van Vugt (vanvugt)]

Please attach a system log (from journalctl) showing the time at which the screen frozen.

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[BertN45 (lammert-nijhof)]

The program crashes when I add syslog as an annex: so here are the last lines:
2210 dbus-daemon[1090]: [session uid=128 pid=1090] Activating via systemd: service name='org.freedesktop.Tracker3.Miner.Files' unit='tracker-miner-fs-3.service' requested by ':1.6' (uid=128 pid=1086 comm="/usr/libexec/tracker-extract-3" label="unconfined")
Sep 19 09:27:04 VM-Ubuntu-2210 systemd[1076]: snap.snapd-desktop-integration.snapd-desktop-integration.service: Scheduled restart job, restart counter is at 1.
Sep 19 09:27:04 VM-Ubuntu-2210 systemd[1076]: Stopped Service for snap application snapd-desktop-integration.snapd-desktop-integration.
Sep 19 09:27:04 VM-Ubuntu-2210 systemd[1076]: Started Service for snap application snapd-desktop-integration.snapd-desktop-integration.
Sep 19 09:27:04 VM-Ubuntu-2210 systemd[1076]: Starting Tracker file system data miner...
Sep 19 09:27:04 VM-Ubuntu-2210 tracker-miner-f[1174]: Unable to get XDG user directory path for special directory &DOCUMENTS. Ignoring this location.
Sep 19 09:27:04 VM-Ubuntu-2210 tracker-miner-f[1174]: Unable to get XDG user directory path for special directory &MUSIC. Ignoring this location.
Sep 19 09:27:04 VM-Ubuntu-2210 tracker-miner-f[1174]: Unable to get XDG user directory path for special directory &PICTURES. Ignoring this location.
Sep 19 09:27:04 VM-Ubuntu-2210 tracker-miner-f[1174]: Unable to get XDG user directory path for special directory &VIDEOS. Ignoring this location.
Sep 19 09:27:04 VM-Ubuntu-2210 tracker-miner-f[1174]: Unable to get XDG user directory path for special directory &DOWNLOAD. Ignoring this location.
Sep 19 09:27:04 VM-Ubuntu-2210 tracker-miner-f[1174]: Unable to get XDG user directory path for special directory &DOCUMENTS. Ignoring this location.
Sep 19 09:27:04 VM-Ubuntu-2210 tracker-miner-f[1174]: Unable to get XDG user directory path for special directory &MUSIC. Ignoring this location.
Sep 19 09:27:04 VM-Ubuntu-2210 tracker-miner-f[1174]: Unable to get XDG user directory path for special directory &PICTURES. Ignoring this location.
Sep 19 09:27:04 VM-Ubuntu-2210 tracker-miner-f[1174]: Unable to get XDG user directory path for special directory &VIDEOS. Ignoring this location.
Sep 19 09:27:04 VM-Ubuntu-2210 kernel: [ 8.012561] audit: type=1400 audit(1663594024.724:50): apparmor="DENIED" operation="capable" profile="/usr/lib/snapd/snap-confine" pid=1173 comm="snap-confine" capability=12 capname="net_admin"
Sep 19 09:27:04 VM-Ubuntu-2210 kernel: [ 8.012568] audit: type=1400 audit(1663594024.724:51): apparmor="DENIED" operation="capable" profile="/usr/lib/snapd/snap-confine" pid=1173 comm="snap-confine" capability=38 capname="perfmon"
Sep 19 09:27:04 VM-Ubuntu-2210 snapd-desktop-integration.snapd-desktop-integration[1173]: Sorry, home directories outside of /home are not currently supported.
Sep 19 09:27:04 VM-Ubuntu-2210 snapd-desktop-integration.snapd-desktop-integration[1173]: See https://forum.snapcraft.io/t/11209 for details.
Sep 19 09:27:04 VM-Ubuntu-2210 systemd[1076]: snap.snapd-desktop-integration.snapd-desktop-integration.service: Main process exited, code=exited, status=1/FAILURE
Sep 19 09:27:04 VM-Ubuntu-2210 systemd[1076]: snap.snapd-desktop-integra...

[Daniel van Vugt (vanvugt)]

Thanks. The only relevant problem I can see there is in the kernel graphics driver:

[drm:vmw_msg_ioctl [vmwgfx]] *ERROR* Failed to open channel.

[Ubuntu Kernel Bot (ubuntu-kernel-bot)]

This change was made by a bot.

[Daniel van Vugt (vanvugt)]

Please use one of these commands to provide a full log.

If you have not rebooted since the freeze:

  journalctl -b0 > journal.txt

Or after you have rebooted:

  journalctl -b-1 > prevboot.txt

and attach the resulting text file here.

[BertN45 (lammert-nijhof)]

Again the program crashes, if I add the text file. So I attach the last lines:

sep 21 16:08:33 VM-Ubuntu-2210 systemd[1082]: snap.snapd-desktop-integration.snapd-desktop-integration.service: Scheduled restart job, restart counter is at 4.
sep 21 16:08:33 VM-Ubuntu-2210 systemd[1082]: Starting Virtual filesystem service - disk device monitor...
sep 21 16:08:33 VM-Ubuntu-2210 systemd[1082]: Stopped Service for snap application snapd-desktop-integration.snapd-desktop-integration.
sep 21 16:08:33 VM-Ubuntu-2210 systemd[1082]: Started Service for snap application snapd-desktop-integration.snapd-desktop-integration.
sep 21 16:08:33 VM-Ubuntu-2210 snapd-desktop-integration.snapd-desktop-integration[1262]: Sorry, home directories outside of /home are not currently supported.
sep 21 16:08:33 VM-Ubuntu-2210 snapd-desktop-integration.snapd-desktop-integration[1262]: See https://forum.snapcraft.io/t/11209 for details.
sep 21 16:08:33 VM-Ubuntu-2210 systemd[1082]: snap.snapd-desktop-integration.snapd-desktop-integration.service: Main process exited, code=exited, status=1/FAILURE
sep 21 16:08:33 VM-Ubuntu-2210 systemd[1082]: snap.snapd-desktop-integration.snapd-desktop-integration.service: Failed with result 'exit-code'.
sep 21 16:08:33 VM-Ubuntu-2210 dbus-daemon[1102]: [session uid=128 pid=1102] Successfully activated service 'org.gtk.vfs.UDisks2VolumeMonitor'
sep 21 16:08:33 VM-Ubuntu-2210 systemd[1082]: Started Virtual filesystem service - disk device monitor.
sep 21 16:08:33 VM-Ubuntu-2210 dbus-daemon[1102]: [session uid=128 pid=1102] Activating via systemd: service name='org.gtk.vfs.GoaVolumeMonitor' unit='gvfs-goa-volume-monitor.service' requested by ':1.15' (uid=128 pid=1203 comm="/usr/libexec/tracker-miner-fs-3" label="unconfined")
sep 21 16:08:33 VM-Ubuntu-2210 systemd[1082]: Starting Virtual filesystem service - GNOME Online Accounts monitor...
sep 21 16:08:33 VM-Ubuntu-2210 dbus-daemon[1102]: [session uid=128 pid=1102] Activating service name='org.gnome.OnlineAccounts' requested by ':1.17' (uid=128 pid=1288 comm="/usr/libexec/gvfs-goa-volume-monitor" label="unconfined")
sep 21 16:08:34 VM-Ubuntu-2210 /usr/libexec/gdm-wayland-session[1115]: dbus-daemon[1115]: [session uid=128 pid=1115] Activating service name='org.a11y.Bus' requested by ':1.4' (uid=128 pid=1163 comm="/usr/bin/gnome-shell" label="unconfined")
sep 21 16:08:34 VM-Ubuntu-2210 /usr/libexec/gdm-wayland-session[1115]: dbus-daemon[1115]: [session uid=128 pid=1115] Successfully activated service 'org.a11y.Bus'
sep 21 16:08:34 VM-Ubuntu-2210 goa-daemon[1292]: goa-daemon version 3.45.2 starting
sep 21 16:08:34 VM-Ubuntu-2210 dbus-daemon[1102]: [session uid=128 pid=1102] Activating service name='org.gnome.Identity' requested by ':1.18' (uid=128 pid=1292 comm="/usr/libexec/goa-daemon" label="unconfined")
sep 21 16:08:34 VM-Ubuntu-2210 dbus-daemon[1102]: [session uid=128 pid=1102] Successfully activated service 'org.gnome.OnlineAccounts'
sep 21 16:08:34 VM-Ubuntu-2210 dbus-daemon[1102]: [session uid=128 pid=1102] Successfully activated service 'org.gnome.Identity'
sep 21 16:08:34 VM-Ubuntu-2210 dbus-daemon[1102]: [session uid=128 pid=1102] Successfully activated service 'org.gtk.vfs.GoaVolume...

[BertN45 (lammert-nijhof)]

Another try to send the journalctl file, using Email.

On Tue, 2022-09-20 at 10:30 +0000, Daniel van Vugt wrote:
> Please use one of these commands to provide a full log.
>
> If you have not rebooted since the freeze:
>
>   journalctl -b0 > journal.txt
>
> Or after you have rebooted:
>
>   journalctl -b-1 > prevboot.txt
>
> and attach the resulting text file here.
>
>
> ** Changed in: linux (Ubuntu)
>        Status: Confirmed => Incomplete
>

[Daniel van Vugt (vanvugt)]

Thanks. I'm going to keep assuming this is the problem:

sep 21 16:08:34 VM-Ubuntu-2210 kernel: [drm:vmw_msg_ioctl [vmwgfx]] *ERROR* Failed to open channel.
sep 21 16:08:34 VM-Ubuntu-2210 kernel: [drm:vmw_msg_ioctl [vmwgfx]] *ERROR* Failed to open channel.

which would be either a bug in the guest's kernel driver, or a bug in the guest additions if they're replacing the vmwgfx kernel driver, or a bug in the host hypervisor.

[Ubuntu Kernel Bot (ubuntu-kernel-bot)]

This change was made by a bot.

[BertN45 (lammert-nijhof)]

I don't expect it is a problem in Virtualbox as you assume, since I use exactly the same VBox release in Ubuntu 22.04 LTS; Ubuntu 20.04 LTS and Fedora 36. I only have a problem with the 22.10 systems using the latest gdm3. I've tried Ubuntu Mate 22.10 and that one worked fine with Virtualbox 6.1.3x.

In the begin of Ubuntu 22.10 I had no issues and everything worked fine, it went wrong after an update of the development edition, somewhere in August. I expect that gdm3 changed the way, in which they start/check the driver, so I would be looking for an incompatibility between gdm3 and Virtualbox 6.1.3x, introduced by a change in gdm3.

[Daniel van Vugt (vanvugt)]

gdm3 does not do any graphics so that's unlikely to be related. Even the login screen is done by gnome-shell/mutter, which accesses the virtual screen through the vmwgfx kernel driver.

[Daniel van Vugt (vanvugt)]

Although your Lspci.txt suggests it's not relevant here, if you have any VMs set to use a non-default virtual graphics adapter then you will hit bug 1990256 which sounds similar. But while your 'lspci -k' is showing 'vmwgfx' that won't be the issue here.

[BertN45 (lammert-nijhof)]

I did read that gdm3 did take care of the login screen, maybe it is wrong or incomplete, but whatever I have news:

I changed /etc/gdm3/custom.conf and disabled wayland. Both systems Ubuntu 22.10 and Ubuntu Unity 22.10 work now with Xorg without any issues.

[Daniel van Vugt (vanvugt)]

gdm3 launches the login screen app but does not do any graphics itself. So if the display or input are frozen on the login screen then it's either gnome-shell, mutter, or the kernel.

[BertN45 (lammert-nijhof)]

Well a summary: The Ubuntu 22.04 LTS VM works file and it also works fine with the Ubuntu 22.10 VM on Xorg after disabling Wayland.
If I use Wayland it freezes and does not display the login screen. The last message, I see flashing by during the start-up, is about starting the gnome display manager. The system shuts down normally, if I use Virtualbox to initiate the shut down.

I see exactly the same behavior with Ubuntu Unity 22.04 and 22.10.

In recovery mode the system will start and it also presents a graphical interface but at a resolution of 800x600. The driver displayed in the settings utility is the "software renderer". Safe mode I assume.

The problem is independent of the used hardware, I checked and I have the same issue with the same Ubuntu 22.10 VM on my 2011 laptop with an i5-2520M (2C4T) and 8GB DDR3 :) There I still run Virtualbox 6.1.36 both in host and VM, Ubuntu itself is up-to-date on both host and in the VM.

[Daniel van Vugt (vanvugt)]

OK, if the problem is just Wayland then please try adding to /etc/environment:

  MUTTER_DEBUG_ENABLE_ATOMIC_KMS=0
  MUTTER_DEBUG_FORCE_KMS_MODE=simple

and re-enable Wayland then reboot the VM. Does it work after that?

P.S. What OS is the host?

[BertN45 (lammert-nijhof)]

That change did not help, the display still did freeze.
The content of /etc/environment was:

PATH="/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin"
MUTTER_DEBUG_ENABLE_ATOMIC_KMS=0
MUTTER_DEBUG_FORCE_KMS_MODE=simple

The host OS is a minimal install of Ubuntu 22.04 LTS and for the VMs I use OpenZFS as storage. For these VMs the hardware is a 512GB Silicon Power nvme-SSD (3400/2300MB/s).

[Daniel van Vugt (vanvugt)]

In the guest VM, what output do you get from:

  find /lib/modules -name vmwgfx.ko

?

[BertN45 (lammert-nijhof)]

I did the command in the xorg session, since in the Wayland session I have no display.

bertadmin@VM-Ubuntu-2210:~$ find /lib/modules -name vmwgfx.ko
/lib/modules/5.15.0-27-generic/kernel/drivers/gpu/drm/vmwgfx/vmwgfx.ko
/lib/modules/5.19.0-15-generic/kernel/drivers/gpu/drm/vmwgfx/vmwgfx.ko
bertadmin@VM-Ubuntu-2210:~$

[Daniel van Vugt (vanvugt)]

The log in comment #13 suggests Wayland was working before those error messages from the vmwgfx kernel driver... Can you check the VM isn't just booting into the wrong virtual terminal? I've seen that happen occasionally.

Try Ctrl+Alt+F1 in the VM or running: sudo chvt 1

[BertN45 (lammert-nijhof)]

I both cases I get the login screen of the Host and not the one of the VM :)

Even Right-Ctrl followed by Ctrl+Alt+F1 gives the Host screen.

I've tried in Xorg the following command:

bertadmin@VM-Ubuntu-2210:~$ ps $(pgrep Xorg)
    PID TTY STAT TIME COMMAND
   1749 tty2 Sl+ 0:16 /usr/lib/xorg/Xorg vt2 -displayfd 3 -auth /run/user

[Marian Rainer-Harbach (marianrh)]

I can confirm the behavior as reported by BertN45:
Setting MUTTER_DEBUG_ENABLE_ATOMIC_KMS=0 and MUTTER_DEBUG_FORCE_KMS_MODE=simple in /etc/environment does not help, but disabling Wayland in /etc/gdm3/custom.conf does help.

I also tried both Ctrl-Alt-F1 and chvt 1 when the screen is frozen, neither changed anything.

[Daniel van Vugt (vanvugt)]

I've just installed 22.10 in VirtualBox and still can't reproduce this bug. What changes from the default VirtualBox settings might be causing it?

[Marian Rainer-Harbach (marianrh)]

Hi Daniel, I played around with VirtualBox settings and Ubuntu installer settings: The problem seems to occur only when "Install third-party software for graphics and Wi-Fi hardware and additional media formats" is selected in the Ubuntu installer.

If the option is _not_ selected, the installed system boots normally. And indeed, after then installing the package ubuntu-restricted-extras, the problem occurs on the next boot.

[BertN45 (lammert-nijhof)]

A screenshot of my settings in Virtualbox are in the attachment. If needed I can also send the vbox file. Note that the first thing I do, if I run into display/video issues, is switching off 3D acceleration, but that did not help this time.

[Daniel van Vugt (vanvugt)]

Thanks. Indeed it seems the workaround is to just avoid selecting the third party software option during install.

[Daniel van Vugt (vanvugt)]

Seems the login screen (gnome-shell) is frozen in:

#0 0x00007ffba4f10ff6 in __ppoll (fds=0x5604764fe680, nfds=3, timeout=<optimised out>, sigmask=0x0)
    at ../sysdeps/unix/sysv/linux/ppoll.c:42
#1 0x00007ffb81f46db4 in gst_poll_wait () from /lib/x86_64-linux-gnu/libgstreamer-1.0.so.0
#2 0x00007ffb81f4c1ce in ?? () from /lib/x86_64-linux-gnu/libgstreamer-1.0.so.0
#3 0x00007ffb81f4d739 in ?? () from /lib/x86_64-linux-gnu/libgstreamer-1.0.so.0
#4 0x00007ffb81f588f3 in gst_update_registry () from /lib/x86_64-linux-gnu/libgstreamer-1.0.so.0
#5 0x00007ffb81eeadda in ?? () from /lib/x86_64-linux-gnu/libgstreamer-1.0.so.0
#6 0x00007ffba6153501 in g_option_context_parse () from /lib/x86_64-linux-gnu/libglib-2.0.so.0
#7 0x00007ffb81ee2877 in gst_init_check () from /lib/x86_64-linux-gnu/libgstreamer-1.0.so.0

[Daniel van Vugt (vanvugt)]

Here's the minimal fix required:

  sudo apt remove gstreamer1.0-vaapi

[Daniel van Vugt (vanvugt)]

I think the problem is gnome-shell waiting for this command (which never finishes):

/usr/lib/x86_64-linux-gnu/gstreamer1.0/gstreamer-1.0/gst-plugin-scanner -l /usr/bin/gnome-shell

because it's waiting for an X server (which does exist as Xwayland):

#0 0x00007f76ea710ec4 in __GI___poll (fds=0x7ffed6d30280, nfds=1, timeout=-1) at ../sysdeps/unix/sysv/linux/poll.c:29
#1 0x00007f76e9c78e23 in ?? () from /lib/x86_64-linux-gnu/libxcb.so.1
#2 0x00007f76e9c7c522 in xcb_connect_to_fd () from /lib/x86_64-linux-gnu/libxcb.so.1
#3 0x00007f76e9c7cedc in xcb_connect_to_display_with_auth_info () from /lib/x86_64-linux-gnu/libxcb.so.1
#4 0x00007f76e9e57eea in _XConnectXCB () from /lib/x86_64-linux-gnu/libX11.so.6
#5 0x00007f76e9e47d17 in XOpenDisplay () from /lib/x86_64-linux-gnu/libX11.so.6
#6 0x00007f76ea37c537 in ?? () from /usr/lib/x86_64-linux-gnu/gstreamer-1.0/libgstvaapi.so
#7 0x00007f76ea353d28 in ?? () from /usr/lib/x86_64-linux-gnu/gstreamer-1.0/libgstvaapi.so
#8 0x00007f76ea3181f7 in ?? () from /usr/lib/x86_64-linux-gnu/gstreamer-1.0/libgstvaapi.so
#9 0x00007f76eaae85c5 in ?? () from /lib/x86_64-linux-gnu/libgstreamer-1.0.so.0
#10 0x00007f76eaaef84d in ?? () from /lib/x86_64-linux-gnu/libgstreamer-1.0.so.0
#11 0x00007f76eaaf08f4 in ?? () from /lib/x86_64-linux-gnu/libgstreamer-1.0.so.0
#12 0x00007f76eaaf19b8 in _gst_plugin_loader_client_run () from /lib/x86_64-linux-gnu/libgstreamer-1.0.so.0
#13 0x000055c6eaeb61dc in ?? ()
#14 0x00007f76ea623510 in __libc_start_call_main (main=main@entry=0x55c6eaeb6120, argc=argc@entry=3,
    argv=argv@entry=0x7ffed6d30c98) at ../sysdeps/nptl/libc_start_call_main.h:58
#15 0x00007f76ea6235c9 in __libc_start_main_impl (main=0x55c6eaeb6120, argc=3, argv=0x7ffed6d30c98, init=<optimised out>,
    fini=<optimised out>, rtld_fini=<optimised out>, stack_end=0x7ffed6d30c88) at ../csu/libc-start.c:381
#16 0x000055c6eaeb6235 in _start ()

[Daniel van Vugt (vanvugt)]

Looks like a bug in libx11 or libxcb because I don't think it's meant to be possible for XOpenDisplay() to hang.

[Daniel van Vugt (vanvugt)]

Dropped to Medium because I don't think there's any useful reason to tick the third party software option in virtual machines. You're not missing out on anything that a virtual machine could otherwise do.

[Marian Rainer-Harbach (marianrh)]

I'm using VirtualBox to test upcoming Ubuntu versions using the exact same config and packages that will later be used on several physical machines. So it's pretty important to me to install the third-party packages in VMs as well. But yes, I can live with the workaround for now.

[Daniel van Vugt (vanvugt)]

That makes sense, and I have done the same in the past. Although this bug highlights the fact that those third-party packages behave differently and exercise different code paths in virtual vs real machines.

[BertN45 (lammert-nijhof)]

I moved all my applications to virtual machines, I use 6 VMs for that purpose. One VM is dealing with all multimedia apps and I need all codecs there. That is, why the Host OS is a minimal install of Ubuntu 22.04 LTS, so without any Apps except Firefox.

I can live with my Xorg workaround for 2 reasons:
- Normally I prefer LTS releases and I used the other releases only a few times in the last 10 years for better drivers or new ZFS functionality;
- I can use Xorg instead of Wayland, since I don't really need 22.10 and for now I installed it mainly out of curiosity and trying to ensure my use case is still OK;

Note, that all YouTubers promote to install that third-party software.

[Daniel van Vugt (vanvugt)]

Yes people should be able to promote a feature advertised during install...

While the bug remains open, a workaround is to:

1. Install *without* the third-party option.

2. sudo apt install ubuntu-restricted-extras

3. sudo apt remove gstreamer1.0-vaapi

[Daniel van Vugt (vanvugt)]

gnome-shell is just an innocent victim here. It's unlikely we would need or want to do a workaround in that.

[Daniel van Vugt (vanvugt)]

Here's the full freeze location (in gst-plugin-scanner) with debug symbols:

#0 0x00007ff635310ec4 in __GI___poll (fds=fds@entry=0x7fff2f5e7790,
    nfds=nfds@entry=1, timeout=timeout@entry=-1)
    at ../sysdeps/unix/sysv/linux/poll.c:29
#1 0x00007ff63479ae23 in poll (__timeout=-1, __nfds=1, __fds=0x7fff2f5e7790)
    at /usr/include/x86_64-linux-gnu/bits/poll2.h:39
#2 read_block (len=8, buf=0x5633edcd0b60, fd=6) at ../../src/xcb_in.c:394
#3 _xcb_in_read_block (c=c@entry=0x5633edcd1f90, buf=0x5633edcd0b60,
    len=len@entry=8) at ../../src/xcb_in.c:1087
#4 0x00007ff63479e522 in read_setup (c=0x5633edcd1f90)
    at ../../src/xcb_conn.c:180
#5 xcb_connect_to_fd (fd=fd@entry=6, auth_info=auth_info@entry=0x7fff2f5e78d0)
    at ../../src/xcb_conn.c:384
#6 0x00007ff63479eedc in xcb_connect_to_display_with_auth_info (
    displayname=displayname@entry=0x5633edcd0a80 ":1024", auth=auth@entry=0x0,
    screenp=screenp@entry=0x0) at ../../src/xcb_util.c:536
#7 0x00007ff63479f06e in xcb_connect (
    displayname=displayname@entry=0x5633edcd0a80 ":1024",
    screenp=screenp@entry=0x0) at ../../src/xcb_util.c:493
#8 0x00007ff634979eea in _XConnectXCB (dpy=0x5633edcd0d20,
    display=0x5633edcd0a80 ":1024", screenp=0x7fff2f5e7aec)
    at ../../src/xcb_disp.c:78
#9 0x00007ff634969d17 in XOpenDisplay (display=0x5633edcd0a80 ":1024")
    at ../../src/OpenDis.c:129
#10 0x00007ff634e9e537 in ?? ()
   from /usr/lib/x86_64-linux-gnu/gstreamer-1.0/libgstvaapi.so
#11 0x00007ff634e75d28 in ?? ()
   from /usr/lib/x86_64-linux-gnu/gstreamer-1.0/libgstvaapi.so
#12 0x00007ff634e3a1f7 in ?? ()
   from /usr/lib/x86_64-linux-gnu/gstreamer-1.0/libgstvaapi.so
#13 0x00007ff6356095c5 in ?? ()
   from /lib/x86_64-linux-gnu/libgstreamer-1.0.so.0
#14 0x00007ff63561084d in ?? ()
   from /lib/x86_64-linux-gnu/libgstreamer-1.0.so.0
#15 0x00007ff6356118f4 in ?? ()
   from /lib/x86_64-linux-gnu/libgstreamer-1.0.so.0
#16 0x00007ff6356129b8 in _gst_plugin_loader_client_run ()
   from /lib/x86_64-linux-gnu/libgstreamer-1.0.so.0
#17 0x00005633ecaf01dc in ?? ()
#18 0x00007ff635223510 in __libc_start_call_main (
    main=main@entry=0x5633ecaf0120, argc=argc@entry=3,
    argv=argv@entry=0x7fff2f5e81a8)
    at ../sysdeps/nptl/libc_start_call_main.h:58
#19 0x00007ff6352235c9 in __libc_start_main_impl (main=0x5633ecaf0120, argc=3,
    argv=0x7fff2f5e81a8, init=<optimised out>, fini=<optimised out>,
    rtld_fini=<optimised out>, stack_end=0x7fff2f5e8198)
    at ../csu/libc-start.c:381
#20 0x00005633ecaf0235 in _start ()

[Mlocik97 (mlocik97)]

Same issue, some additional info: https://bugs.launchpad.net/ubuntu/+source/gstreamer-vaapi/+bug/1991577

[Mlocik97 (mlocik97)]

I use one installation in both physical and VM environments, then I require it to work.
I also often clone systems from VM to HW and viceversa.

[Daniel van Vugt (vanvugt)]

If I modify gstreamer-vaapi to ignore X11 on Wayland, then it freezes in Wayland instead !?

#0 0x00007f5ec1d10ec4 in __GI___poll (fds=0x7fff30df9e50, nfds=1, timeout=-1)
    at ../sysdeps/unix/sysv/linux/poll.c:29
#1 0x00007f5ec14c3e0c in wl_display_dispatch_queue ()
   from /lib/x86_64-linux-gnu/libwayland-client.so.0
#2 0x00007f5ec14c4f1f in wl_display_roundtrip_queue ()
   from /lib/x86_64-linux-gnu/libwayland-client.so.0
#3 0x00007f5ec1a4253b in gst_vaapi_display_wayland_setup (
    display=<optimised out>)
    at ../gst-libs/gst/vaapi/gstvaapidisplay_wayland.c:190
#4 0x00007f5ec1a0dec8 in gst_vaapi_display_create (data=0x0,
    init_type=GST_VAAPI_DISPLAY_INIT_FROM_DISPLAY_NAME, display=0x562df9389100)
    at ../gst-libs/gst/vaapi/gstvaapidisplay.c:958
#5 gst_vaapi_display_config (display=0x562df9389100,
    init_type=GST_VAAPI_DISPLAY_INIT_FROM_DISPLAY_NAME, init_value=0x0)
    at ../gst-libs/gst/vaapi/gstvaapidisplay.c:1265
#6 0x00007f5ec19d3317 in gst_vaapi_create_display (
    display_name=<optimised out>, display_type=<optimised out>)
    at ../gst/vaapi/gstvaapipluginutil.c:110
#7 gst_vaapi_create_test_display () at ../gst/vaapi/gstvaapipluginutil.c:929
#8 gst_vaapi_create_test_display () at ../gst/vaapi/gstvaapipluginutil.c:915
#9 plugin_init (plugin=0x562df9373190) at ../gst/vaapi/gstvaapi.c:191

[Daniel van Vugt (vanvugt)]

Fix proposed: https://salsa.debian.org/gnome-team/gnome-shell/-/merge_requests/66

[Daniel van Vugt (vanvugt)]

I don't think this bug should only occur on VirtualBox. One reason we've focussed on VirtualBox might just be that ubuntu-restricted-* fails to install on other systems, which is a different bug.

Evidence suggesting non-virtual systems also have the bug:

 * Bug 1992285
 * https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/5710
 * https://gitlab.gnome.org/GNOME/gnome-shell/-/issues/5956

[Launchpad Janitor (janitor)]

This bug was fixed in the package gnome-shell - 43.0-1ubuntu2

---------------
gnome-shell (43.0-1ubuntu2) kinetic; urgency=medium

  [ Daniel van Vugt ]
  * Avoid deadlocking in gst_init_check() on startup (LP: #1988488)

  [ Jeremy Bicha ]
  * debian/source_gnome-shell.py: Don't try to report nonexistent mouse
    gsettings schema
  * Cherry-pick 4 patches from the future gnome-shell 43.1 to fix various issues

 -- Jeremy Bicha <email address hidden> Wed, 12 Oct 2022 11:25:50 -0400

[BertN45 (lammert-nijhof)]

In my system the problem has been solved, The past weeks I logged in from xorg to bypass the login screen freeze. But now I login from Wayland again and everything works fine both in Ubuntu 22.10 and Ubuntu Unity 22.10

[Daniel van Vugt (vanvugt)]

It seems the problem was Wayland + gnome-shell specific because gnome-shell is the display server and effectively single threaded. gnome-shell deadlocks on itself if it spawns a client synchronously and expects to wait for that client to finish. The client never finishes because the server isn't replying while it's waiting for the client to finish.

[emptythevoid (emptythevoid)]

This appears to be fixed in Daily (tested 10/15/22)