Hi,

the latest user-mode-linux breaks fakemachine and thus debos. Reverting
to 6.10um1 fixes it:

$ fakemachine -b uml --show-boot
[..]
systemd[1]: systemd-remount-fs.service: Main process exited, code=exited, status=1/FAILURE
systemd[1]: systemd-remount-fs.service: Failed with result 'exit-code'.
systemd-journald[763]: Collecting audit messages is disabled.
systemd[1]: Failed to start systemd-remount-fs.service - Remount Root and Kernel File Systems.
[FAILED] Failed to start systemd-remount-fs.service - Remount Root and Kernel File Systems.
[..]
         Mounting scratch.mount - /scratch...
[FAILED] Failed to mount etc-alternatives.mount - /etc/alternatives.
See 'systemctl status etc-alternatives.mount' for details.
[DEPEND] Dependency failed for local-fs.target - Local File Systems.
[FAILED] Failed to mount etc-dbus\x2d1.mount - /etc/dbus-1.
See 'systemctl status "etc-dbus\\x2d1.mount"' for details.
[FAILED] Failed to mount etc-java.mount - /etc/java.
See 'systemctl status etc-java.mount' for details.
[FAILED] Failed to mount etc-java\x2d21\x2dopenjdk.mount - /etc/java-21-openjdk.
See 'systemctl status "etc-java\\x2d21\\x2dopenjdk.mount"' for details.
[FAILED] Failed to mount etc-ssl.mount - /etc/ssl.
See 'systemctl status etc-ssl.mount' for details.
[FAILED] Failed to mount run-fakemachine.mount - /run/fakemachine.
See 'systemctl status run-fakemachine.mount' for details.
[FAILED] Failed to mount etc-ca\x2dcertificates.mount - /etc/ca-certificates.
See 'systemctl status "etc-ca\\x2dcertificates.mount"' for details.
[..]
You are in emergency mode. After logging in, type "journalctl -xb" to view
system logs, "systemctl reboot" to reboot, or "exit"
to continue bootup.
Give root password for maintenance
(or press Control-D to continue):

Cheers Jochen

I'd need more information. I'm presuming it isn't happening on just UML ?

In your logs, there's not much information other than the mount failures
from within.

It'd be helpful if you could dig some more details

s3nt fr0m a $martph0ne, excuse typ0s

I'd need more information. I'm presuming it isn't happening on just UML ?

In your logs, there's not much information other than the mount failures
from within.

It'd be helpful if you could dig some more details

s3nt fr0m a $martph0ne, excuse typ0s

No, "fakemachine -b kvm" works fine.  It's just UML.

(I see this too, and it broke part of debusine's CI the day that the new
user-mode-linux landed in testing.)

Jochen dug out https://pastebin.mozilla.org/iRqZcQLd from fakemachine,
although I'm not sure exactly how he did that.  It looks as though
virtfs mounts are broken with the new UML.

Hi Ritesh,

* Colin Watson <cjwatson@debian.org> [2024-10-29 12:09]:

I think it is hostfs and I guess it is related to the new mount API
discussed here:

https://lore.kernel.org/all/CANP3RGceNzwdb7w=vPf5=7BCid5HVQDmz1K5kC9JG42+HVAh_g@mail.gmail.com/

A minimal reproduce is:

$ mmdebstrap --variant=essential --include=init --customize-hook='passwd -d --root "$1" root' unstable unstable.ext4
$ linux ubd0=unstable.ext4 systemd.unit=rescue.target

# inside UML

Press Enter for maintenance
(or press Control-D to continue):
root@host:~# mount none /mnt/host -t hostfs
mount: /mnt/host: fsconfig system call failed: Bad address.
        dmesg(1) may have more information after failed mount system call.
root@host:~# mount none /mnt/home -t hostfs -o /home
mount: /mnt/home: fsconfig system call failed: hostfs: Unknown parameter '/home'.
        dmesg(1) may have more information after failed mount system call.

Cheers Jochen

Control: tags -1 +confirmed
Thank you. I can confirm the same on my local setup as well.


```
[  OK  ] Finished systemd-journal-flush.service - Flush Journal to
Persistent Storage.
[FAILED] Failed to mount usr-lib-modules.mount - /usr/lib/modules.
See 'systemctl status usr-lib-modules.mount' for details.
[DEPEND] Dependency failed for local-fs.target - Local File Systems.
[FAILED] Failed to mount var-tmp-Debian\x2dBuild.mount -
/var/tmp/Debian-Build.
See 'systemctl status "var-tmp-Debian\\x2dBuild.mount"' for details.
```

and

```
Oct 30 05:57:41 uml systemd[1]: Mounting usr-lib-modules.mount -
/usr/lib/modules...
Oct 30 05:57:42 uml systemd[1]: usr-lib-modules.mount: Mount process
exited, code=exited, status=32/n/a
Oct 30 05:57:42 uml mount[814]: mount: /usr/lib/modules: fsconfig
system call failed: hostfs: Unknown parameter '/usr/lib/uml/modules'.
Oct 30 05:57:42 uml mount[814]:        dmesg(1) may have more
information after failed mount system call.
Oct 30 05:57:42 uml systemd[1]: usr-lib-modules.mount: Failed with
result 'exit-code'.
Oct 30 05:57:42 uml systemd[1]: Failed to mount usr-lib-modules.mount -
/usr/lib/modules.

```

I will check on what is the path ahead.

Thanks,
Ritesh

(By the way, why don't we make the version number for the user-mode-linux binary package reflect the version of linux-source-* package that was used at build-time, so that rebuilds of user-mode-linux against newer linux-source-* packages will be identified as upgrades? The version numbers of binary packages need not have any relation to their source packages. The status quo is also troublesome for bug reports like this, because '7.1um1' isn't a useful identifier for the kernel used. If there's not a particular reason for this I can try a merge request.)

I ran into this issue in a quite different context, having built user-mode-linux against 7.1.8-1~bpo13+1 on Trixie and wanting to give it a try to investigate a kernel issue. A succinct statement and a patch were proposed in 2024 at https://lore.kernel.org/all/4a8ae74e-2f90-4da8-9511-325ca6f67aa6@huawei.com/ but it doesn't seem like anything happened to it and it was just forgotten. Here's a clear manifestation: /usr/share/doc/user-mode-linux/README.Debian includes the conventional advice of

Note that the subdirectory of the host is specified as a "keyless" mount option, and the source device name (the word "hostfs" here) is not used but ignored. This doesn't actually work anymore though:

It's buried very deep in the mailing list thread at https://lore.kernel.org/linux-um/2fb819b00563afbeb69c156b463dd41335c430b6.camel@sipsolutions.net/ but I found that using a key name of "hostfs" makes everything work:
(correct!)

So if this apparent regression is not going to be fixed in the kernel, being a couple years old at this point, we need to update documentation to use the "hostfs=" key. We ought to do this in README.Debian, but also no other documentation about UML (Rob Landley's site, kernel.org UML HowTo v2, ...) has addressed this yet. This is why it's been so hard for me to find.

Thank you for the bug report. You are right. I entirely missed to update
the readme file. Will do it in the next update

s3nt fr0m a $martph0ne, excuse typ0s