CVE-2023-54153

What it actually does

Failure to call ext4_quota_off_umount() on error from ext4_mark_recovery_complete() during __ext4_fill_super leaves quota structures allocated via v2_read_file_info and dquot_load_quota_sb, producing a kmemleak of quota file metadata.

Why it matters to the business

Repeated mount attempts on ext4 volumes with quotas enabled under error conditions can slowly exhaust kernel memory, degrading stability and availability of Linux systems that host or serve such file systems.

How an attack unfolds

An attacker with local access triggers repeated failed ext4 mounts (for example by supplying a corrupted journal or conflicting feature flags) after quotas have been enabled; the missing quota release leaves unreferenced objects that accumulate until memory pressure or OOM occurs.

How to fix it

Upgrade the Linux kernel to a version containing commit c327b83c59ee938792a0300df646efac39c7d6a7 or later; rebuild and reboot with the patched kernel. Where immediate patching is impossible, avoid enabling quotas on ext4 volumes that may encounter mount-recovery errors.

How to confirm the fix worked

After upgrade, reproduce the original failing-mount scenario and confirm that kmemleak no longer reports the v2_read_file_info allocation; verify the presence of the ext4_quota_off_umount call in the failed_mount10 error path via kernel source or binary inspection.

Typical attack path

1. Obtain local shell or container access on a Linux host using ext4 with quotas
2. Craft or locate an ext4 image that causes ext4_mark_recovery_complete to return an error after quotas are loaded
3. Invoke mount in a loop or via automated tooling to trigger the error path repeatedly
4. Monitor slab or kmemleak reports for growth of quota-related allocations

How to detect it

Search kernel logs and kmemleak reports for unreferenced objects whose backtrace includes v2_read_file_info, dquot_load_quota_sb and ext4_enable_quotas; correlate with mount syscalls that return errors on ext4 file systems.

What your security team should do

• Prioritize kernel updates on all systems that mount ext4 volumes with quotas enabled
• Add kernel command-line or fstab options to disable quotas during troubleshooting mounts until patches are applied
• Enable kmemleak or KASAN in test environments to detect similar quota-handling regressions
• Review downstream package maintainers for inclusion of the referenced commit in distribution kernels