Re:Cyber Attack on Iran's Nuclear Facility
by mhh__ on Sunday April 11, 2021 @06:12PM
(#26773955)
> Stuxnet is just a worse version of the normal software operating machinery and it used a non-standard install mechanism.
No it's not. Stuxnet was specifically tuned to ruin the centrifuges under the noses of the presumably well-trained operators and scientists using them. If the PLC suddenly stopped or sped up to the point that it triggered some kind of failsafe then the game is up.
> Extracting secrets from processor caches based on timing is far more sophisticated and it didn’t even require HUMINT.
Implementing spectre is really not that hard. Also, the secrets aren't extracted from the caches (in the archetypal spectre implementation at least, the vulnerabilities are popping up all over the place): Spectre is an attack against the branch predictor and speculative execution - the cache is just the side channel to exfiltrate data (specifically having each byte value map to a cacheline which you can then time the latency of).
No it's not. Stuxnet was specifically tuned to ruin the centrifuges under the noses of the presumably well-trained operators and scientists using them. If the PLC suddenly stopped or sped up to the point that it triggered some kind of failsafe then the game is up.
> Extracting secrets from processor caches based on timing is far more sophisticated and it didn’t even require HUMINT.
Implementing spectre is really not that hard. Also, the secrets aren't extracted from the caches (in the archetypal spectre implementation at least, the vulnerabilities are popping up all over the place): Spectre is an attack against the branch predictor and speculative execution - the cache is just the side channel to exfiltrate data (specifically having each byte value map to a cacheline which you can then time the latency of).