New TLBleed vulnerability could leak cryptography keys
Target: Intel’s processors.
Attack Vector: Exploitation of Intel’s Hyper-Threading technology and processor caches to leak data.
Summary: A vulnerability in Intel’s processors has proved to be exploitable by a malware with the aim of extracting encryption keys and sensitive information from applications. The code used for the malware is capable of extracting a secret 256-bit key from another program while it performs a signing operation with libgcrypt’s Curve 25519 EdDSA implementation. Each key was obtained using brute force and machine-learning software.
Risk assessment summary: This threat has been assessed as 3c MODERATE. If successful, TLBleed can leverage flaws in protection of the CPU’s translation lookaside buffer, which can be exploited to extract cryptography keys from another running program with a minimum 98% success rate. However, it is to be considered that malware would need to be running, or a malicious user logged into the system, to exploit it. Additionally, the attack is not currently running in the wild. However, it can be seen as alarming for public cloud users, as other guest instances on the same hardware could attempt to use this to exfiltrate data from threads running in other cores.