5 ms·

The best respons is already a reality: iPhone 6 with Secure Enclave - or making it impossible to unlock the phone with a hack and with brute force. I hope Goog

The best respons is already a reality: iPhone 6 with Secure Enclave - or making it impossible to unlock the phone with a hack and with brute force. I hope Google and Microsoft follow suit, and make this a requirement for future versions of Android and Windows Phone.

Aissen 11y ago
Nope, Secure Enclave is updateable, even Apple said so.
- 4ad 11y ago
  Do we know what can the software running on the secure enclave do exactly? If the software itself doesn't have access to the master key (a.i. can't read it directly), and if the slow authentication is inherent from running the algorithm on the chip, rather than a having been slowed down artificially, then even an update to the secure enclave would not compromise the security of the system. Do we know exactly what the secure enclave software can do and what it can't do?
  - Aissen 11y ago
    It does not matter for the case at hand, which is FBI wanting to bruteforce the code without 10-try erase.
    - 4ad 11y ago
      Yes, it does not matter for the case at hand, but it is very important in general, and it has been brought into attention up the thread.
  - 542458 11y ago
    > if the slow authentication is inherent from running the algorithm on the chip, rather than a having been slowed down artificially There is no known cryptographic algorithm that provides inherently increasing times. The slowdown is entirely artificial, and it is believed that an update could remove it.
    - 4ad 11y ago
      I was under the impression that the algorithm (running on the secure enclave?) takes at least 200ms. Yes, obviously longer times are artificial, but my question is whether the 200ms baseline is artificial or not. Of course this doesn't really matter at all if modified software could just read the key from the chip. That is the most interesting question. So, can it? Or can the software only provide data to the chip, to do the algorithm in hardware, but the software can't read the key burned (?) into the chip.
      - 542458 11y ago
        The 200ms hash baseline (on all up-to-date iPhones) is not artificial, but the increasing time is. Unfortunately, I don't believe the answers to your other questions are known to anybody other than Apple right now.
      - pfg 11y ago
        Their hash algorithm takes ~80ms per attempt. All crypto happens on a dedicated AES engine. Files are encrypted with a key derived from UID (device key which cannot be extracted via firmware) and passphrase.
        4ad 11y ago
        So if the key can't be extracted via the firmware, someone updating the firmware, even with malicious intent, should not be a problem, am I correct?
        pfg 11y ago
        With a sufficiently complex passphrase, it's not a problem. With a typical 6-digit PIN, disabling the artificial delay and auto-wipe (which is possible via firmware updates) is all you need for a successful brute-force attack.
        4ad 11y ago
        Thanks for the clarifications.

Aissen 11y ago

Nope, Secure Enclave is updateable, even Apple said so.

4ad 11y ago
Do we know what can the software running on the secure enclave do exactly? If the software itself doesn't have access to the master key (a.i. can't read it directly), and if the slow authentication is inherent from running the algorithm on the chip, rather than a having been slowed down artificially, then even an update to the secure enclave would not compromise the security of the system. Do we know exactly what the secure enclave software can do and what it can't do?
- Aissen 11y ago
  It does not matter for the case at hand, which is FBI wanting to bruteforce the code without 10-try erase.
  - 4ad 11y ago
    Yes, it does not matter for the case at hand, but it is very important in general, and it has been brought into attention up the thread.
- 542458 11y ago
  > if the slow authentication is inherent from running the algorithm on the chip, rather than a having been slowed down artificially There is no known cryptographic algorithm that provides inherently increasing times. The slowdown is entirely artificial, and it is believed that an update could remove it.
  - 4ad 11y ago
    I was under the impression that the algorithm (running on the secure enclave?) takes at least 200ms. Yes, obviously longer times are artificial, but my question is whether the 200ms baseline is artificial or not. Of course this doesn't really matter at all if modified software could just read the key from the chip. That is the most interesting question. So, can it? Or can the software only provide data to the chip, to do the algorithm in hardware, but the software can't read the key burned (?) into the chip.
    - 542458 11y ago
      The 200ms hash baseline (on all up-to-date iPhones) is not artificial, but the increasing time is. Unfortunately, I don't believe the answers to your other questions are known to anybody other than Apple right now.
    - pfg 11y ago
      Their hash algorithm takes ~80ms per attempt. All crypto happens on a dedicated AES engine. Files are encrypted with a key derived from UID (device key which cannot be extracted via firmware) and passphrase.
      - 4ad 11y ago
        So if the key can't be extracted via the firmware, someone updating the firmware, even with malicious intent, should not be a problem, am I correct?
        pfg 11y ago
        With a sufficiently complex passphrase, it's not a problem. With a typical 6-digit PIN, disabling the artificial delay and auto-wipe (which is possible via firmware updates) is all you need for a successful brute-force attack.
        4ad 11y ago
        Thanks for the clarifications.