DEBIAN-CVE-2026-46223

See a problem?
Please try reporting it to the source first.
Source
https://security-tracker.debian.org/tracker/CVE-2026-46223
Import Source
https://storage.googleapis.com/debian-osv/debian-cve-osv/DEBIAN-CVE-2026-46223.json
JSON Data
https://api.osv.dev/v1/vulns/DEBIAN-CVE-2026-46223
Upstream
Published
2026-05-28T10:16:37Z
Modified
2026-09-14T17:03:14Z
Severity
  • 5.5 (Medium) CVSS_V3 - CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H CVSS Calculator
Summary
[none]
Details

In the Linux kernel, the following vulnerability has been resolved: cgroup: Defer css percpu_ref kill on rmdir until cgroup is depopulated A chain of commits going back to v7.0 reworked rmdir to satisfy the controller invariant that a subsystem's ->css_offline() must not run while tasks are still doing kernel-side work in the cgroup. [1] d245698d727a ("cgroup: Defer task cgroup unlink until after the task is done switching out") [2] a72f73c4dd9b ("cgroup: Don't expose dead tasks in cgroup") [3] 1b164b876c36 ("cgroup: Wait for dying tasks to leave on rmdir") [4] 4c56a8ac6869 ("cgroup: Fix cgroup_drain_dying() testing the wrong condition") [5] 13e786b64bd3 ("cgroup: Increment nr_dying_subsys_* from rmdir context") [1] moved task cset unlink from do_exit() to finish_task_switch() so a task's cset link drops only after the task has fully stopped scheduling. That made tasks past exit_signals() linger on cset->tasks until their final context switch, which led to a series of problems as what userspace expected to see after rmdir diverged from what the kernel needs to wait for. [2]-[5] tried to bridge that divergence: [2] filtered the exiting tasks from cgroup.procs; [3] had rmdir(2) sleep in TASK_UNINTERRUPTIBLE for them; [4] fixed the wait's condition; [5] made nr_dying_subsys_* visible synchronously. The cgroup_drain_dying() wait in [3] turned out to be a dead end. When the rmdir caller is also the reaper of a zombie that pins a pidns teardown (e.g. host PID 1 systemd reaping orphan pids that were re-parented to it during the same teardown), rmdir blocks in TASK_UNINTERRUPTIBLE waiting for those pids to free, the pids can't free because PID 1 is the reaper and it's stuck in rmdir, and the system A-A deadlocks. No internal lock ordering breaks this; the wait itself is the bug. The css killing side that drove the original reorder, however, can be made cleanly asynchronous: ->css_offline() is already async, run from css_killed_work_fn() driven by percpu_ref_kill_and_confirm(). The fix is to make that chain start only after all tasks have left the cgroup. rmdir's user-visible side then returns as soon as cgroup.procs and friends are empty, while ->css_offline() still runs only after the cgroup is fully drained. Verified by the original reproducer (pidns teardown + zombie reaper, runs under vng) which hangs vanilla and succeeds here, and by per-commit deterministic repros for [2], [3], [4], [5] with a boot parameter that widens the post-exit_signals() window so each state is reliably reachable. Some stress tests on top of that. cgroup_apply_control_disable() has the same shape of pre-existing race: when a controller is disabled via subtree_control, kill_css() ran synchronously while tasks past exit_signals() could still be linked to the cgroup's csets, and ->css_offline() could fire before they drained. This patch preserves the existing synchronous behavior at that call site (kill_css_sync() + kill_css_finish() back-to-back) and a follow-up patch will defer kill_css_finish() there using a per-css trigger. This seems like the right approach and I don't see problems with it. The changes are somewhat invasive but not excessively so, so backporting to -stable should be okay. If something does turn out to be wrong, the fallback is to revert the entire chain ([1]-[5]) and rework in the development branch instead. v2: Pin cgrp across the deferred destroy work with explicit cgroup_get()/cgroup_put() around queue_work() and the work_fn. v1 wasn't actually broken (ordered cgroup_offline_wq + queue_work order in cgroup_task_dead() saved it) but the explicit ref removes the dependency on those non-obvious invariants. Also note the pre-existing cgroup_apply_control_disable() race in the description; a follow-up will defer kill_css_finish() there.

References

Affected packages

Debian:14 / linux

Package

Name
linux
Purl
pkg:deb/debian/linux?arch=source&distro=forky

Affected ranges

Type
ECOSYSTEM
Events
Introduced
0 Unknown introduced version / All previous versions are affected
Fixed
7.0.9-1

Affected versions

6.*
6.12.38-1
6.12.41-1
6.12.43-1~bpo12+1
6.12.43-1
6.12.48-1
6.12.57-1~bpo12+1
6.12.57-1
6.12.63-1~bpo12+1
6.12.63-1
6.12.69-1~bpo12+1
6.12.69-1
6.12.73-1~bpo12+1
6.12.73-1
6.12.74-1
6.12.74-2~bpo12+1
6.12.74-2
6.12.85-1~bpo12+1
6.12.85-1
6.12.86-1~bpo12+1
6.12.86-1
6.12.88-1~bpo12+1
6.12.88-1
6.12.90-1~bpo12+1
6.12.90-1
6.12.90-2~bpo12+1
6.12.90-2
6.12.94-1~bpo12+1
6.12.94-1
6.12.95-1~bpo12+1
6.12.95-1
6.12.96-1
6.12.100-1
6.12.101-1
6.12.105-1
6.12.107-1
6.13~rc6-1~exp1
6.13~rc7-1~exp1
6.13.2-1~exp1
6.13.3-1~exp1
6.13.4-1~exp1
6.13.5-1~exp1
6.13.6-1~exp1
6.13.7-1~exp1
6.13.8-1~exp1
6.13.9-1~exp1
6.13.10-1~exp1
6.13.11-1~exp1
6.14.3-1~exp1
6.14.5-1~exp1
6.14.6-1~exp1
6.15~rc7-1~exp1
6.15-1~exp1
6.15.1-1~exp1
6.15.2-1~exp1
6.15.3-1~exp1
6.15.4-1~exp1
6.15.5-1~exp1
6.15.6-1~exp1
6.16~rc7-1~exp1
6.16-1~exp1
6.16.1-1~exp1
6.16.3-1~bpo13+1
6.16.3-1
6.16.5-1
6.16.6-1
6.16.7-1
6.16.8-1
6.16.9-1
6.16.10-1
6.16.11-1
6.16.12-1~bpo13+1
6.16.12-1
6.16.12-2
6.17.2-1~exp1
6.17.5-1~exp1
6.17.6-1
6.17.7-1
6.17.7-2
6.17.8-1~bpo13+1
6.17.8-1
6.17.9-1
6.17.10-1
6.17.11-1
6.17.12-1
6.17.13-1~bpo13+1
6.17.13-1
6.18~rc4-1~exp1
6.18~rc4-1~exp2
6.18~rc5-1~exp1
6.18~rc6-1~exp1
6.18~rc7-1~exp1
6.18.1-1~exp1
6.18.2-1~exp1
6.18.3-1
6.18.5-1~bpo13+1
6.18.5-1
6.18.8-1
6.18.9-1~bpo13+1
6.18.9-1
6.18.10-1
6.18.12-1~bpo13+1
6.18.12-1
6.18.13-1
6.18.14-1
6.18.15-1~bpo13+1
6.18.15-1
6.19~rc4-1~exp1
6.19~rc5-1~exp1
6.19~rc6-1~exp1
6.19~rc7-1~exp1
6.19~rc8-1~exp1
6.19-1~exp1
6.19.2-1~exp1
6.19.3-1~exp1
6.19.4-1~exp1
6.19.5-1~exp1
6.19.6-1
6.19.6-2~bpo13+1
6.19.6-2
6.19.8-1~bpo13+1
6.19.8-1
6.19.10-1~bpo13+1
6.19.10-1
6.19.11-1~bpo13+1
6.19.11-1
6.19.12-1
6.19.13-1~bpo13+1
6.19.13-1
6.19.14-1~bpo13+1
6.19.14-1
7.*
7.0-1~exp1
7.0.1-1~exp1
7.0.3-1
7.0.4-1~bpo13+1
7.0.4-1
7.0.7-1~bpo13+1
7.0.7-1
7.0.9-1~bpo13+1

Ecosystem specific 

{
    "urgency": "not yet assigned"
}

Database specific

source 
"https://storage.googleapis.com/debian-osv/debian-cve-osv/DEBIAN-CVE-2026-46223.json"