BuildNot yet confirmed elsewhere1 publisher2 min readPublished
Docker and Kubernetes default seccomp profiles allow the AF_UNIX socket call behind CVE-2026-80521
DepthFirst published a working Ubuntu 26.04 exploit for CVE-2026-80521 on 2026-09-22, before distribution kernels carried a fix, according to a dev.to post. The flaw sits in AF_UNIX, which default Docker and Kubernetes seccomp profiles permit.
The Engineer · Build desk
47 days: the age of the upstream fix on 2026-09-22, when DepthFirst published the exploit. Kernel 7.2 carries the fix. The vulnerability is rated 7.8; the post says that rating looks low for multi-tenant build systems.
| Figure | Value | Note | Claim |
|---|---|---|---|
| Age of upstream fix when DepthFirst published exploit | 47 days | 16 | |
| Kernel with the fix | 7.2 kernel version | 8 | |
| Vulnerability rating | 7.8 rating | Low for multi-tenant builds, post says | 17 |
What happened
- The defect is in the code that tracks edges in the AF_UNIX socket connection graph, and the upstream fix references unlinking scc_entry in unix_del_edge().
- The upstream fix landed on 2026-08-06; kernels 7.2 and 7.1.10 carry it.
- When the exploit became public, no distribution patch existed for the Ubuntu 22.04, 24.04 and 26.04 LTS kernels or for their AWS, Azure and GCP variants.
- The exploit shapes the heap with repeated socket operations, uses the dangling reference to read and write kernel memory, and lands a write that escalates to root on the host.
Why it matters
- constraint Systemd, container runtimes, databases and logging daemons all use AF_UNIX, so a profile that denied it fleet-wide would break them.
- exposure Any code that can run in a container can take the first step: a compromised application, a malicious dependency, a build job or a scheduled task.
- decision Teams whose workloads must keep AF_UNIX and cannot wait for a kernel patch have to move them behind a hypervisor boundary such as Firecracker or Kata Containers.
According to the dev.to post, the isolation argument rests on three layers. Namespaces and cgroups partition and limit resources, and a kernel memory bug is confined by neither, because what it reaches is kernel memory [13]. Seccomp is the layer that limits which kernel interfaces a process can reach at all, and it had to allow this syscall family for ordinary software to run [14].
The post adds that the kernel structures involved sit outside the namespaces a container defines [12]. The escape does not depend on a misconfigured volume mount or a privileged flag. None of the three exploit steps needs a capability a normal container is denied [11].
Patching means waiting on the distributor. The post says no user-space change removes the underlying bug and tells readers to track the distribution advisory, because the affected cloud kernels are the ones that matter in practice [18]. The upstream fix was 47 days old on the day the exploit went public [16].
The vulnerability is rated 7.8 [7]. The post calls that a fair reading if the container runs untrusted code under an account that cannot do much. Multi-tenant build systems, notebook platforms and function-as-a-service runtimes make exactly that assumption, it says, and the rating looks low for them [17].
Where no patch has arrived, the post's interim answer is a custom seccomp profile that denies socket(AF_UNIX). It removes the trigger, costs nothing in a workload that never uses the socket family, and breaks workloads that need local IPC. The post puts it on the highest-risk workloads, not fleet-wide [19]. In our view that is the right scope. We would run the profile in staging first, since the failure mode of a wrong guess is a service that stops starting.
Both the exploit details and the patch status rest on that single post.
What to watch
- Distribution advisories and patched kernels for Ubuntu 22.04, 24.04 and 26.04, including the AWS, Azure and GCP variants.
- DepthFirst's own write-up, which would confirm the exploit details and show whether it works against other distributions.
- Any change to the default Docker or Kubernetes seccomp profile, or guidance from either project on AF_UNIX.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence40
- Adoption
- Insufficient
- Hype gap
- Insufficient
- Incentives
- Insufficient
- Confidence45
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
CVE-2026-80521 is a Linux kernel use-after-free in the AF_UNIX subsystem that lets an unprivileged process inside a container reach the host.
- [2]
DepthFirst published a working exploit for Ubuntu 26.04 on 2026-09-22.
- [3]
Distribution kernels, including the ones cloud providers ship, were still without a fix when the exploit was published.
- [4]
The default seccomp profiles in Docker and Kubernetes permit socket creation for AF_UNIX, because blocking socket(AF_UNIX) outright breaks most of userspace.
- [5]
AF_UNIX sockets are used by systemd, by container runtimes, by databases, by logging daemons and by almost everything that needs to talk to something else on the same machine.
- [6]
The defect is in the edge-tracking code that maintains the socket's connection graph; the upstream fix references 'af_unix: Unlink scc_entry in unix_del_edge()'.
- [8]
The upstream fix landed on 2026-08-06 and is present in kernels 7.2 and 7.1.10.
- [9]
Ubuntu 22.04, 24.04 and 26.04 LTS kernels, including the variants built for AWS, Azure and GCP, did not have a distribution patch when the exploit became public.
- [10]
The exploit runs in three parts: a process in the container triggers the edge removal sequence through repeated AF_UNIX socket operations to shape the heap; the dangling reference is used to read and write kernel memory; and because allocator state is predictable after shaping, the write lands on a structure used to escalate to root on the host.
- [11]
None of the three exploit parts requires a capability a normal container is denied.
- [12]
The kernel structures involved sit outside the namespaces a container defines, so the escape does not depend on a misconfigured volume mount or a privileged flag.
- [13]
Namespaces and cgroups partition and limit resources, and a kernel memory bug is not confined by either, because the resource being reached is kernel memory rather than a namespaced object.
- [14]
Seccomp constrains which kernel interfaces are reachable at all, and it had to allow the syscall family in question for ordinary software to function.
- [15]
The first exploit step is available to any process that can execute code, including a compromised application, a malicious dependency, a build job and a scheduled task.
- [16]
The upstream fix was 47 days old on 2026-09-22, the day DepthFirst published the exploit.
- [17]
A 7.8 is a fair reading if the container is assumed to run untrusted code under an account that cannot do much; multi-tenant build systems, notebook platforms and function-as-a-service runtimes make that assumption, which makes the rating look low for those environments.
- [18]
There is no user-space change that removes the underlying bug and the upstream fix is the only complete answer; readers should track the distribution advisory, because the affected cloud kernels are the ones that matter in practice.
- [19]
A custom seccomp profile that denies socket(AF_UNIX) removes the trigger at the cost of breaking workloads that need local IPC, so it belongs on the highest-risk workloads rather than fleet-wide; denying AF_UNIX in a workload that does not use it costs nothing.
- [20]
Firecracker and Kata Containers put a hypervisor between the workload and the host kernel, and are the post's answer for workloads that must keep AF_UNIX and cannot wait.
Sources
1 independent publisher whose own reporting we read for this story.
- dev.toCVE-2026-80521: An AF_UNIX Use-After-Free That Crosses the Container Boundary
1 article · October 13, 2026
Topics and entities
Follow any of these and your For You feed starts watching them — no settings page required.
Topics
- Cloud kernel patchingFollow
- Container SecurityFollow
- Seccomp sandboxingFollow
- Linux Kernel VulnerabilitiesFollow
Entities
- CVE-2026-80521Follow
- Linux kernelFollow
- AF_UNIX socketsFollow
- DockerFollow
- KubernetesFollow
- seccompFollow
- UbuntuFollow
- DepthFirstFollow
- FirecrackerFollow
- Kata ContainersFollow