Ubuntu 20.04 — runc — multiple vulnerabilities (5 CVEs) — patch and remediation guide

📖 ~4 min read  •  Source: Ubuntu Security Notice USN-6619-1

Related CVEs: CVE-2024-21626 CVE-2023-25809 CVE-2023-27561 CVE-2023-28642 CVE-2021-30465

Upstream summary: Rory McNamara discovered that runC did not properly manage internal file
descriptor while managing containers. An attacker could possibly use this
issue to obtain sensitive information or bypass container restrictions.

Table of contents

1. Symptom & Impact
2. Environment & Reproduction
3. Root Cause Analysis
4. Quick Triage
5. Step-by-Step Diagnosis
6. Solution – Primary Fix
7. Solution – Alternative Approaches
8. Verification & Acceptance Criteria
9. Rollback Plan
10. Prevention & Hardening
11. Related Errors & Cross-Refs
12. References & Further Reading

Symptom & Impact

runc runs when an image is pulled, unpacked, inspected or a container is started, so a defect here is normally reached through untrusted image content or a registry response rather than inbound traffic to a listening port. On Ubuntu 20.04 that shows up as escapes from the container’s namespaces, files written outside the intended root during buildah or podman build, mount and capability handling that grants more than was requested, and images produced before the update carrying the defect onward to every host that pulls them. runc is container tooling — command-line binaries invoked per operation, though a node runtime such as cri-o does run as a unit — so restart steps apply to that runtime and to containers still running the old code, not to runc itself.

Environment & Reproduction

Reproduction targets Ubuntu 20.04 (focal). Confirm release and installed package:

lsb_release -a
cat /etc/os-release
dpkg -l runc | tail -2
apt-cache policy runc
uname -r

Exercise the workload that uses runc while collecting:

needrestart -b -r l  # (pkg: needrestart) run as root; machine-readable. NEEDRESTART-SVC = units still mapping deleted libs, NEEDRESTART-SESS = login sessions, NEEDRESTART-KSTA = kernel state. -r l never restarts anything
needrestart -v -l -r l  # human-readable, libraries only (-l); names each PID and the obsolete file it still maps
checkrestart -v  # (pkg: debian-goodies) older alternative; add -p to report only deleted files owned by a dpkg package, -m for machine-readable
sudo journalctl -xe --no-pager | tail -200
sudo tail -200 /var/log/apt/history.log
sudo tail -200 /var/log/kern.log | grep -i apparmor

Root Cause Analysis

Root cause is documented in Ubuntu Security Notice USN-6619-1. Canonical security maintainers shipped fixes in the corresponding runc update for Ubuntu 20.04; running an outdated build leaves the host exposed to the failure modes described in the advisory. Fixes land in the focal-security pocket of the main archive. Correlate apt history with the journal:

grep -A2 -B2 runc /var/log/apt/history.log
zgrep -A2 -B2 runc /var/log/apt/history.log.*.gz 2>/dev/null
cat /proc/sys/kernel/tainted   # non-zero = tainted kernel / out-of-tree modules

Quick Triage

Run these on Ubuntu 20.04 to capture the current state of runc:

dpkg -l runc | tail -1                  # installed version
dpkg -V runc                             # verify shipped files
sudo apt update && apt list --upgradable 2>/dev/null | grep -i security
dpkg-query -W -f='${Package} ${Version}\n' runc; gcc --version | head -1; ld --version | head -1; as --version | head -1  # record the FIXED toolchain version you are rebuilding with (ld/as come from binutils)
ccache -C && ccache -s  # (pkg: ccache) MUST clear the cache, or the rebuild silently reuses object files produced by the vulnerable compiler and you ship the bug again
make clean; cmake --build build --target clean; ninja -C build -t clean; cargo clean; go clean -cache  # drop stale objects before rebuilding (use whichever applies to the project)
sudo ufw status verbose 2>/dev/null | head -20
sudo aa-status 2>/dev/null | head -20
# If runc ships a service unit (unit/job name often differs from pkg name, e.g.
# bind9→named, apache2→apache2, postgresql-NN→postgresql@NN-main):
dpkg-query -L runc 2>/dev/null | grep -E '^/etc/init\.d/[^/]+$'  # legacy SysV init script; its basename is the service name for 'service <name> restart' on non-systemd hosts
systemctl list-unit-files --no-pager --type=service,socket,timer | grep -iF -- 'runc'  # fuzzy name match; -F is required because package names contain regex metacharacters ('+' in libstdc++6/g++-13 makes -E fail to match at all)

Step-by-Step Diagnosis

1. List failing services.

   systemctl --failed --no-pager

2. Tail the journal / syslog for runc.

   sudo journalctl -xe -f --no-pager

3. Inspect UFW (Uncomplicated Firewall) state.

   sudo ufw status numbered
   sudo ufw show added
   sudo iptables -L -n -v | head -30

4. Surface AppArmor denials and switch the profile to complain mode if needed.

   sudo journalctl -k 2>/dev/null | grep -i 'apparmor="DENIED"' | tail -30
   sudo aa-status
   # /etc/apparmor.d/usr.bin.runc or usr.sbin.runc — inspect first
   sudo aa-complain /etc/apparmor.d/usr.bin.runc 2>/dev/null || true

5. Verify runc integrity and reinstall if anything is altered.

   sudo dpkg -V runc
   sudo debsums -c runc 2>/dev/null
   sudo apt install --reinstall -y runc

6. Correlate findings with /var/log/apt/history.log, /var/log/dpkg.log, and Ubuntu Security Notice USN-6619-1 to pin the change that introduced the regression.

Solution – Primary Fix

Apply the corrective apt transaction referenced by Ubuntu Security Notice USN-6619-1, then reload the affected service:

sudo apt update
sudo apt -y install --only-upgrade runc
sudo systemctl daemon-reload
# Service name may differ from pkg name; check first:
dpkg -l runc | tail -1            # confirm new version

For kernel / glibc / systemd / openssl advisories a reboot (or Livepatch) is required:

sudo apt install -y needrestart
sudo needrestart -r l       # list units that need restart
sudo systemctl reboot       # or: sudo shutdown -r now
# Livepatch (Ubuntu Pro) avoids reboot for many kernel CVEs:
sudo canonical-livepatch status
sudo canonical-livepatch refresh

Solution – Alternative Approaches

If the primary upgrade is not viable, pick from these:

• Hold the package so apt cannot upgrade it:

  sudo apt-mark hold runc
  apt-mark showhold | grep runc
  # Release the hold later with:
  sudo apt-mark unhold runc

• Pin a known-good version via apt preferences:

  # /etc/apt/preferences.d/runc.pref
  Package: runc
  Pin: version <good-version>
  Pin-Priority: 1001

• Downgrade to an older version if a regression is suspected:

  apt-cache madison runc
  sudo apt install --allow-downgrades -y runc=<older-version>

• Investigate AppArmor blocking the new binary; switch to complain briefly, capture denials, then re-enforce:

  sudo aa-complain /etc/apparmor.d/usr.bin.runc 2>/dev/null
  # reproduce the failure
  sudo journalctl -k | grep apparmor | tail
  sudo aa-enforce /etc/apparmor.d/usr.bin.runc 2>/dev/null

• Apply Canonical Livepatch (Ubuntu Pro) to land kernel fixes without reboot:

  sudo canonical-livepatch status
  sudo canonical-livepatch refresh

• Take only the security pocket update and defer the full point-release upgrade:

  sudo apt -y install --only-upgrade -t focal-security runc

Verification & Acceptance Criteria

All of these should pass after the fix is applied:

dpkg -l runc | tail -1                                  # expected fixed version
apt list --upgradable 2>/dev/null | grep -i security || echo OK
sudo ufw status numbered | head
sudo aa-status 2>/dev/null | head -5

The conditions described in the advisory must no longer be reported for runc across two consecutive runs.

Rollback Plan

Capture state before any change:

apt list --installed 2>/dev/null > /root/apt-pre.txt
dpkg --get-selections > /root/dpkg-pre.txt
# ZFS-on-root (Ubuntu 20.04+ default installer option):
sudo zfs snapshot rpool/ROOT/ubuntu@pre-runc
# LVM-on-root:
sudo lvcreate -L 4G -s -n root_pre_patch /dev/<vg>/<root-lv>

To revert:

sudo apt install --allow-downgrades -y runc=<old-version>
sudo systemctl daemon-reload
# Kernel rollback: pick the prior kernel from the GRUB menu, then:
sudo systemctl reboot
# ZFS rollback (rolls the whole root dataset):
sudo zfs rollback -r rpool/ROOT/ubuntu@pre-runc

Prevention & Hardening

Reduce the chance of this recurring on Ubuntu 20.04 (focal):

• Enable scheduled security updates via unattended-upgrades:

  sudo apt install -y unattended-upgrades update-notifier-common
  sudo dpkg-reconfigure -plow unattended-upgrades
  # /etc/apt/apt.conf.d/50unattended-upgrades:
  Unattended-Upgrade::Allowed-Origins { "${distro_id}:${distro_codename}-security"; };

• Install needrestart so services restart automatically after library upgrades:

  sudo apt install -y needrestart
  # /etc/needrestart/needrestart.conf -> $nrconf{restart} = 'a';

• Attach Ubuntu Pro to enable Livepatch and extended security coverage:

  sudo pro attach <token>
  sudo pro enable livepatch

• Subscribe to ubuntu-security-announce and watch ubuntu.com/security/cves.

• Monitor file integrity with debsums and AIDE:

  sudo apt install -y debsums aide
  sudo debsums -ca
  sudo aideinit && sudo mv /var/lib/aide/aide.db.new /var/lib/aide/aide.db
  sudo aide --check

• For estate-wide patching, manage with Canonical Landscape:

  sudo apt install -y landscape-client
  sudo landscape-config

• Keep AppArmor profiles in enforce mode and apply CIS Ubuntu Linux Benchmark hardening.

Related Errors & Cross-Refs

Issues that commonly surface alongside a runc update: apt lock contention, broken dpkg state, systemd ordering cycles, AppArmor denials, and UFW rule drift. Useful triage:

sudo dpkg --configure -a
sudo apt --fix-broken install
systemd-analyze critical-chain
sudo journalctl -k 2>/dev/null | grep -i apparmor | tail
cat /proc/sys/kernel/tainted

View all ubuntu-20-04 tutorials on the Tutorials Hub →

Browse all common problems & solutions on the Tutorials Hub.

References & Further Reading

Primary reference: Ubuntu Security Notice USN-6619-1. Manual pages useful on Ubuntu 20.04:

man apt
man apt-get
man apt-mark
man dpkg
man systemctl
man journalctl
man ufw
man apparmor
man aa-status
man unattended-upgrades
man canonical-livepatch
man pro

Other resources: Ubuntu Security Notices, Ubuntu CVE Tracker, Ubuntu upgrade notes, and per-package notes in /usr/share/doc/runc/ for components implicated in this advisory.