DirtyClone Linux Kernel Local Privilege Escalation
Score: 8/10

DirtyClone Linux Kernel Local Privilege Escalation

A high-severity Linux kernel vulnerability (DirtyClone) allows local attackers to achieve root access by exploiting improper shared memory flag propagation in the networking stack.

Executive Summary

DirtyClone, identified as CVE-2026-43503, is a high-severity (CVSS 8.8) local privilege escalation vulnerability affecting the Linux kernel's networking subsystem. It is part of the 'DirtyFrag' family of flaws, characterized by the kernel's failure to maintain proper metadata (the SKBFL_SHARED_FRAG flag) when cloning network packet fragments. This failure causes the kernel to treat read-only, file-backed memory as private writable data, effectively bypassing Copy-on-Write (CoW) protections.

Technically, an attacker can load a privileged binary into memory and use a local IPsec tunnel to overwrite in-memory login checks without modifying the original binary on disk. This makes the attack highly stealthy, as file-integrity tools and reboots will not show signs of permanent disk tampering. The exploit has been confirmed to work on major distributions including Debian, Ubuntu, and Fedora that allow unprivileged user namespaces.

This vulnerability is particularly critical for multi-tenant environments, container hosts, and CI/CD runners where untrusted users might be granted the ability to create namespaces. Organizations should prioritize patching kernels to the fixed versions (e.g., v7.1-rc5 or backported LTS versions) immediately to prevent unauthorized root access.

Key Details

Threat Name

DirtyClone (CVE-2026-43503)

Affects

Linux kernel, Debian, Ubuntu, Fedora, Kubernetes clusters, Linux kernel algif_aead module, Linux kernel IPsec ESP, RxRPC, Linux kernel skb_try_coalesce(), Linux Kernel (net/skbuff), Linux 5.10 before 5.10.257, Linux 5.15 before 5.15.208, Linux 6.1 before 6.1.174, Linux 6.6 before 6.6.141

Adversary

—

MITRE Techniques

Malware/Tools

Miasma, RedLine, DirtyClone, DirtyFrag, Copy Fail, Fragnesia

Report Score

8out of 10
Quality Score
Good
IOC Quality4
TTP Details9
Detection Guidance6
Enterprise Relevance8
Clarity & Structure9
Technical Depth9

Sources