Cyber Risks · Google Project Zero

MTE As Implemented, Part 3: The Kernel

Summary: Google Project Zero's Mark Brand published the third part of a series on ARM's Memory Tagging Extensions, examining how MTE-based mitigations hold up in the kernel context after earlier posts covered hardware limitations and user-mode implications. The post states that since Spectre, standard memory-tagging approaches cannot serve as a hard probabilistic mitigation wherever an attacker can construct a speculative side-channel, and that enforcing a panic on kernel tag-check failure would be essential for any mitigation based on async or asymmetric MTE modes.

Why it matters: The analysis concludes that known-tag bypasses are a fundamental weakness that must be accounted for, and that a kernel panic on tag-check failure - the enforcement step async MTE mitigations depend on - may itself be considered a stability problem.

Read the original at Google Project Zero ↗

Topics
Vulnerabilities & Patching
Related