CVE-2026-45933

Linux · Kernel

An incorrect register state propagation in the Linux kernel's BPF verifier can lead to security policy bypasses.

Executive summary

A logic error in the Linux kernel BPF verifier could allow an attacker to bypass security restrictions.

Vulnerability

The issue exists in sync_linked_regs(), where the BPF verifier incorrectly copies register IDs. This can lead to the propagation of incorrect bounds information, potentially allowing a malicious BPF program to read or write memory outside of its permitted range.

Business impact

This vulnerability is highly significant as it targets the BPF subsystem, often used for security monitoring and packet filtering. Successful exploitation could lead to privilege escalation or complete system compromise, justifying the 7.8 CVSS score.

Remediation

Immediate Action: Apply the vendor-provided kernel patch that corrects the BPF register ID synchronization.

Proactive Monitoring: Review BPF program loading logs and monitor for anomalous kernel behavior or unexpected BPF program crashes.

Compensating Controls: Restrict access to the bpf() system call to highly privileged users (e.g., via unprivileged_bpf_disabled sysctl).

Exploitation status

Public Exploit Available: false

Analyst recommendation

This is a critical patch for any system that utilizes BPF features. Given the potential for privilege escalation, administrators must prioritize this update to maintain the integrity of kernel-level security filters.