System doesn't fully wake from sleep when VM are in use

[Alessio]

Hi friends.

Fedora 41 fully updated.
If I have running VMs (using virt-manager or GNOME Boxes) and I suspend
the system, when I wake it up, the screen stays black and the keyboard
is unresponsive (the caps lock led doesn't turn on/off).
BTW the system looks somewhat alive, indeed I can ping it and
eventually connect with ssh.

If the VMs are off, the system can go to sleep and wake up normally.
I can reproduce it on different machines.

Do you have the chance to reproduce this issue?

Ciao,
A.

[Alessio]

It looks like the issue is here in "frozen" user session

$ systemctl status session-4.scope 
● session-4.scope - Session 4 of User alessio
     Loaded: loaded (/run/systemd/transient/session-4.scope; transient)
  Transient: yes
     Active: active (running) (frozen-by-parent) since Wed 2024-10-23
12:37:55 CEST; 8min ago

$ systemctl status user-1000.slice
● user-1000.slice - User Slice of UID 1000
     Loaded: loaded
    Drop-In: /usr/lib/systemd/system/user-.slice.d
             └─10-defaults.conf
             /etc/systemd/system.control/user-1000.slice.d
             └─50-CPUWeight.conf, 50-IOWeight.conf, 50-MemoryLow.conf,
50-MemoryMin.conf
     Active: active (freezing-by-parent) since Wed 2024-10-23 11:10:41
CEST; 1h 37min ago

$ systemctl status user.slice
● user.slice - User and Session Slice
     Loaded: loaded (/usr/lib/systemd/system/user.slice; static)
    Drop-In: /etc/systemd/system.control/user.slice.d
             └─50-MemoryLow.conf, 50-MemoryMin.conf
     Active: active (freezing) since Wed 2024-10-23 10:54:25 CEST; 1h
53min ago

Indeed, issuing "sudo systemctl thaw user.slice" via ssh, the GDM login
screen reappears.

Ciao,
A.

On Wed, 2024-10-23 at 08:41 +0000, Alessio wrote:
> Hi friends.
>
> Fedora 41 fully updated.
> If I have running VMs (using virt-manager or GNOME Boxes) and I
> suspend
> the system, when I wake it up, the screen stays black and the
> keyboard
> is unresponsive (the caps lock led doesn't turn on/off).
> BTW the system looks somewhat alive, indeed I can ping it and
> eventually connect with ssh.
>
> If the VMs are off, the system can go to sleep and wake up normally.
> I can reproduce it on different machines.
>
> Do you have the chance to reproduce this issue?
>
> Ciao,
> A.

[Kamil Paral]

On Wed, Oct 23, 2024 at 12:52 PM Alessio  wrote:
> Indeed, issuing "sudo systemctl thaw user.slice" via ssh, the GDM login
> screen reappears.

Oh, fun. I can reproduce this when I'm using user-session virtual machines
(gnome-boxes or virt-manager). It doesn't happen when system-sessions VMs
are used.
That's a bummer. I'll create a bug report and post it here.

[Kamil Paral]

On Wed, Oct 23, 2024 at 1:24 PM Kamil Paral  wrote:
> Oh, fun. I can reproduce this when I'm using user-session virtual machines
> (gnome-boxes or virt-manager). It doesn't happen when system-sessions VMs
> are used.
> That's a bummer. I'll create a bug report and post it here.

Here we go:
https://bugzilla.redhat.com/show_bug.cgi?id=2321268

[Kamil Paral]

On Wed, Oct 23, 2024 at 2:01 PM Kamil Paral  wrote:
> On Wed, Oct 23, 2024 at 1:24 PM Kamil Paral  wrote:
> > Oh, fun. I can reproduce this when I'm using user-session virtual
> > machines (gnome-boxes or virt-manager). It doesn't happen when
> > system-sessions VMs are used.
> > That's a bummer. I'll create a bug report and post it here.
>
> Here we go:
> https://bugzilla.redhat.com/show_bug.cgi?id=2321268

I'd love to propose this as a release blocker, because I think it's quite
nasty, but unfortunately we don't have any suspend-related release
criterion, intentionally, because most suspend bugs are very hardware
specific. But perhaps it would make sense to have some generic suspend
criterion, as we currently have for power off, reboot, log in, etc, that
would only apply if it happened across all or a wide range of hardware. If
people were really in favor of this, I can draft something.