CVE-2026-72111

Published: August 18, 2026Last modified: August 18, 2026

Description

In the Linux kernel, the following vulnerability has been resolved: bpf: Reset register bounds before narrowing retval range in check_mem_access() When the BPF verifier processes a context load of an LSM hook return value, it calls __mark_reg_s32_range() to narrow the register to the hook's valid range. However, __mark_reg_s32_range() intersects the new range with the register's existing bounds using max_t()/min_t() rather than replacing them. If the destination register carries stale bounds from a prior instruction (e.g. BPF_MOV64_IMM), the intersection can produce a range narrower than reality. The verifier then believes it knows the register's exact value, while at runtime the actual hook return value is loaded, creating a verifier/runtime mismatch that can be used to bypass BPF memory safety checks. The else branch already calls mark_reg_unknown() to reset register state before any narrowing. Apply the same reset in the is_retval path so stale bounds are cleared before __mark_reg_s32_range() intersects.

Severity score breakdown

Status

References

• https://git.kernel.org/stable/c/0993dc5fc619c0b25ab1310cb11d65e78351c0fe
• https://git.kernel.org/stable/c/5a55f9aecc08990940e70f0c7048a80850c5a16a
• https://git.kernel.org/stable/c/5e0b273e0a62cc04ec338c7b502797c66c2ed42a
• https://git.kernel.org/stable/c/bde92f65042ec14389782dd223f706bf6b59ce5d