[Cryptography] 9999 keys for this one iPhone

Tony Arcieri bascule at gmail.com
Mon Feb 29 22:07:55 EST 2016


On Mon, Feb 29, 2016 at 6:22 PM, Tom Mitchell <mitch at niftyegg.com> wrote:

> The limited set of keys is generated from three things: the device unique
> id,
> Apple's secret key same on all like phones and the user controlled  PIN.
>
> Once the binary contents of the encrypted file system are pulled
> from the flash device.  Should the device be connected to a computer
> or phone network it will reset, reboot reload and operate with the same
> internal secrets it has always had.
> Setting the four number PIN  will start encrypting the phone in one of the
> 10000
> possible ways that this one phone can use in this mode.
>
> Is there any component of the encrypted data that can validate
> or dismiss the key and move on to the next?
>

In a modern iPhone with a "Secure Payments Enclave" (SPE), no. There is no
instruction to exfiltrate the UID key from the SPE which is needed to
derive the root key. It can perhaps be extracted through a decapping attack
on the SoC, unless Apple used one of these:

https://en.wikipedia.org/wiki/Physical_unclonable_function

Going forward it seems Apple is intent upon building a phone that moves all
the key derivation functionality into hardware and will most likely use
something like ^^^ to prevent exfiltration of the "UID key" /
device-specific master key.

Note that none of this applies to the FBI case involving the iPhone 5c,
which does not have an SPE, however the phone in the San Bernadino case
does (iPhone 5s I believe)

Failing to exfiltrate this key, the rate at which the key can be brute
force is bounded by mechanisms like exponential backoff and
wipe-after-n-attempts. Also note: all that needs to be wiped is the master
encryption key. Once destroyed backups become useless.

-- 
Tony Arcieri
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.metzdowd.com/pipermail/cryptography/attachments/20160229/a7a2b9de/attachment.html>


More information about the cryptography mailing list