Description

In the Linux kernel, the following vulnerability has been resolved: libbpf: Reject non-exclusive metadata maps in the signed loader The loader verifies map->sha against the metadata hash in its instructions. map->sha is calculated when BPF_OBJ_GET_INFO_BY_FD is called on the frozen map. While the map is frozen, the /signed loader/ must also ensure the map is exclusive, as, without exclusivity (which a hostile host could just omit when loading the loader), another BPF program with map access can mutate the contents afterwards, so the check passes on stale data. With the extra check as part of the signed loader, it now refuses to move on with map->sha validation if the host set it up wrongly.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "7.1",
  "pubDate": "2026-08-28T08:16:52.793Z",
  "pubdate": "2026-08-28T08:16:52.793Z",
  "executiveSummary": "A security vulnerability in the Linux kernel's libbpf signed loader allows for the potential bypass of integrity checks on BPF metadata maps.\nThe vulnerability arises from the lack of mandatory exclusivity checks on frozen BPF maps during the signed loading process.\nThis flaw enables a hostile host or an adjacent BPF program to mutate map contents after initial validation, effectively rendering the map->sha integrity verification based on stale data.\nSuccessful exploitation allows an attacker to manipulate map contents that were intended to be immutable, potentially leading to unauthorized BPF program behavior or privilege escalation.\nThe vulnerability affects environments utilizing signed loaders to enforce policy or security constraints on BPF program execution.\nRemediation requires updating the libbpf loader to enforce map exclusivity as a prerequisite for metadata hash validation, ensuring the integrity of the BPF object lifecycle.",
  "technicalDetails": "The root cause of this vulnerability is an insufficient validation logic within the libbpf signed loader. Specifically, the loader performs integrity verification of BPF metadata maps by comparing the map->sha (calculated via BPF_OBJ_GET_INFO_BY_FD on a frozen map) against a predefined metadata hash stored within the BPF instructions. However, the loader previously failed to enforce the 'exclusive' attribute on the map during this verification phase.\nIn the context of the Linux BPF subsystem, a 'frozen' map status is intended to prevent further modifications. However, if a map is not marked as 'exclusive,' the current design allows other processes or BPF programs with map access to interact with the map even after it has been frozen. An attacker acting as a hostile host can intentionally omit the exclusivity requirement when loading the malicious BPF object.\nThe attack flow follows a precise sequence: First, the attacker initiates the loading of a BPF program with an associated metadata map. During this process, the attacker ensures the map is not configured with the 'exclusive' flag, even while the map is technically in a 'frozen' state. Second, the libbpf signed loader calculates the map->sha value based on the current map state to verify its authenticity. Third, because the map lacks exclusivity, the attacker uses an alternative, unauthorized BPF program or a malicious host-level operation to modify the contents of the map after the loader's initial integrity check has passed.\nConsequently, the signed loader proceeds under the false assumption that the map remains in its verified state. The integrity check operates on stale or 'time-of-check-to-time-of-use' (TOCTOU) data. This allows for the injection of arbitrary data or malicious configurations into the map, which the BPF program then executes or utilizes. This bypasses the security guarantees provided by the signed loader mechanism, as the integrity of the configuration is no longer anchored to the cryptographically signed metadata. The lack of exclusivity verification creates a TOCTOU window where the BPF subsystem's expected security posture is subverted by subsequent, unauthorized memory mutation of the BPF map structure."
}