what is dirtyclone fr?

DirtyClone is a proof of concept exploit for a security flaw in the Linux kernel, tracked as CVE-2026-43503. It is written entirely in Python and demonstrates how a regular, unprivileged user on an affected Linux system could gain full root access. The README describes it as an independent, non-official Python reimplementation of a technique originally researched by JFrog Security Research, and calls it the fourth vulnerability in a related family that also includes earlier bugs known as DirtyPipe and DirtyFrag. According to the README, the underlying issue involves how the Linux kernel handles copied network socket buffers when a specific netfilter feature is combined with an encrypted networking protocol. Because of a flag handling mistake, data that should stay isolated can end up written directly into memory that is shared with files on disk, which is the page cache the operating system uses to speed up file access. The README rates the severity of this bug as high, states it affects a narrow range of Linux kernel release candidates before a specific fix commit, and notes it only works when a particular kernel setting for unprivileged user namespaces is turned on. The proof of concept itself is described as pure Python with no compilation step required. Once run, it is said to add a new user with root level permissions to the system's password file and then open an interactive root shell for the attacker. The README describes the code as clean and commented. The project includes a clear disclaimer stating it is meant only for educational and security research purposes, and that using it against systems the user does not own or have explicit permission to test is illegal. It links to the CVE record and the kernel commit that eventually fixed the issue, though a link to the original research writeup is not yet filled in.