A hardware security team is evaluating defences for an HSM that performs ECDSA signing. Which statement about fault injection attacks is TRUE?
- AFault injection is a purely software attack that exploits exception handlers to leak the key through the operating system log.
- BFault injection is a subtype of brute force where the attacker tries random keys until one decrypts the ciphertext.
- CFault injection induces transient errors during a cryptographic computation so the faulty output can be analysed to recover key material. Correct
- DFault injection is mitigated entirely by increasing the symmetric key length, because longer keys make glitches statistically harmless.
Why A is wrong: Fault injection is fundamentally a physical attack on the device using voltage, clock, electromagnetic, or optical disturbance. Confusing it with software exception abuse loses the distinguishing feature of the attack class.
Why B is wrong: Brute force exhausts the key space by trial. Fault injection does not search keys; it perturbs a single signing or decryption operation so the faulty result mathematically reveals key bits.
Why C is correct: Differential fault analysis and similar techniques deliberately glitch clock, voltage, or laser pulses to corrupt an intermediate value, then compare faulty and correct outputs to extract bits of the private key.
Why D is wrong: Key length is irrelevant to fault attacks because the maths exploits the difference between faulty and correct outputs, not the size of the key space. Real countermeasures are redundant computation, sensors, and shielding.