[UBUNTU 22.04] OS guest boot issues on 9p filesystem

[bugproxy]

=== Reported by <email address hidden> - 2024-05-13 03:53:01 ===

---Problem Description---
OS guest boot issues on 9p filesystem due to unix domain sockets open failure

Contact Information = <email address hidden>

Machine Type = 3931-7G4

---uname output---
5.15.0-92-generic #102-Ubuntu SMP Wed Jan 10 09:35:24 UTC 2024 s390x s390x s390x GNU/Linux

---Steps to Reproduce---
 #!/bin/bash

# Cleanup target dir
[ -d ./target ] && rm -rf target
mkdir target

# Add configuration updates
mkdir -p ./target/etc/initramfs-tools/
echo 9p >> ./target/etc/initramfs-tools/modules
echo 9pnet_virtio >> ./target/etc/initramfs-tools/modules

# Add the test script
cat > ./target/test_init << EOF
#!/bin/bash

echo "Test for unix domain sockets"

nc -Ul /socket &
sleep 1
echo "Sockets work" | nc -UN /socket || echo "Sockets fail"

echo o > /proc/sysrq-trigger
sleep 999
EOF
chmod 700 ./target/test_init

# Create an Ubuntu 23.10 around it
echo "Creating Ubuntu target OS"
debootstrap --variant=minbase\
            --include=udev,kmod,initramfs-tools,systemd,netcat-openbsd,linux-image-generic \
            --exclude=man,bash-completion \
            mantic ./target > /dev/null || exit 1

# Run the test in 9p forwarded filesystem
echo "Running OS in qemu"
qemu-system-s390x \
  -m 8192 \
  -smp 4 \
  -nodefaults -nographic -no-reboot -no-user-config \
  -kernel ./target/boot/vmlinuz \
  -initrd ./target/boot/initrd.img \
  -append 'root=fsRoot rw rootfstype=9p rootflags=trans=virtio,version=9p2000.L,msize=512000,cache=mmap,posixacl console=ttysclp0 init=/test_init quiet' \
  -fsdev local,security_model=passthrough,multidevs=remap,id=fsdev-fsRoot,path=./target \
  -device virtio-9p-pci,id=fsRoot,fsdev=fsdev-fsRoot,mount_tag=fsRoot \
  -device virtio-serial-ccw -device sclpconsole,chardev=console \
  -chardev stdio,id=console,signal=off

---Debugger---
A debugger is not configured

Userspace rpm: qemu-(current).deb

Userspace tool common name: qemu

Userspace tool obtained from project website: na

The userspace tool has the following bit modes: both

*Additional Instructions for <email address hidden>:
-Attach ltrace and strace of userspace application.

[bugproxy (bugproxy)]

------- Comment From <email address hidden> 2024-05-13 06:24 EDT-------
This bug is also described at:
https://gitlab.com/qemu-project/qemu/-/issues/2337
created in the qemu project bugtracker.

[bugproxy (bugproxy)]

------- Comment From <email address hidden> 2024-05-13 06:57 EDT-------
This Bug is the result of the fix to:
CVE-2023-2861: Prohibit opening any special file directly on host

I also opened a Bug in the qemu bugtracker
https://gitlab.com/qemu-project/qemu/-/issues/2337

The containers fail because syslog cannot open its unix domain socket on the filesystem.
We tracked the change that provokes this error to a CVE change in qemu that forbids opening of special files to
prevent exposing data from the host. Special files should be handled by the guest os.
Unix domain socket files are also special files, and they are handled by the guest OS in their entirety, and the 9p server in qemu assigns them individual inodes so they are safe to open. But they must be opened so their fd can be passed to the appropriate connect() or bind() function so the OS can use them.
Socket files don't have a traditional read or write functionality, they are mere representatives for a local address.
There is no convention for where domain socket files should go, so there is no easy fix by just creating a tmpfs somewhere.
We also see other workloads and services failing for not being able to open their local socket files.

The analysis of CVE-2023-2861 in detail reveals
- opening of device files through the 9p server directly grants access to read/write functions of those device files. Also device files can be created in-place anywhere.
- opening of FIFOs is somewhat unsafe as long as there are possible collisions that could expose host data using read/write.
- opening of sockets is safe because the 9p server protects the revealed inode and provides no way to connect the file to a socket.

[Sergio Durigan Junior (sergiodj)]

Thank you for the report.

Given that this is an upstream regression and there is a related upstream bug about it, I believe it's best to wait for their input/feedback before moving forward.