Hacker News new | past | comments | ask | show | jobs | submit login
“modern phone encryption sort of sucks” (twitter.com/matthew_d_green)
52 points by justicz on Dec 23, 2020 | hide | past | favorite | 8 comments



I wonder to the degree this paradigm affects the development of some of the upcoming mobile Linux distros for devices like the Pinephone and Librem 5. I know that support for full disk encryption recently got added to Mobian[1] (which leveraged work done by postmarketOS), I believe through the calamares[2] installer. I know that full-disk encryption is less secure after boot (which I think this thread addresses), but I'm under the impression the method they're using is basically on-par with actual full-on "this is a computer used for x purpose"-level encryption.

Unless that impression is wrong (possible), does that mean that a) Android and iOS devices are now fundamentally less secure than these new distros with disk-level encryption enabled or b) that encryption is really just a hard problem and this isn't actually unique to phones, it's just that police have more opportunities to try to break into phones than they do into laptops, so it gets more attention?

[1] https://gitlab.com/mobian1/issues/-/issues/50

[2] https://github.com/calamares/calamares


What do you mean by «actual full-on "this is a computer used for x purpose"-level encryption» ?

Both iOS and Android use hardware full disk encryption and RAM encryption.

Keys for all files are in RAM, keys for RAM are in Secure Processor, so you can’t dump data from RAM chips via direct electrical connection and see keys for files.


I think this answers my question, in that I wasn't sure if the encryption methods used by modern phones is sub-par compared to other computing devices like laptops or servers.

At least based on yours and sibling comments, the title "modern phone encryption kind of sucks" is really more that encryption on modern hardware in general is somewhat fallible, and that phones are just the canary in the coalmine that demonstrates it.


I totally don't understand what you mean. Full disk encryption isn't new in the iOS or Android world. iOS for examples supported it since iOS 4. But full disk encryption doesn't protect against buggy OS. No one really expected to take apart a phone, extract its flash, and be able to read its contents that way.


I'm no expert on encryption, but this article made it sound like several of the areas that made iOS less than secure weren't bugs, but specific design choices in terms of what they allow developers to do, which often make things more vulnerable.

I guess I was asking if, in a world where someone could (feasibly, if not easily) take one of the existing mobile Linux distros and modify it to make it more secure than Android or iOS, or if they're already sort of at the limit there and barring a fundamental advance in computer security across the board, this is just how things are?


Can anyone clarify if temporarily disabling FaceID (holding volume + side button) puts the device back into the BFU state?


No, at least based on the info given in the thread. Apps need access to unencrypted data to perform background tasks and disabling FaceID in this way doesn’t have any effect on those tasks; otherwise it would be a confusing user experience.


ouch thats pretty serious... So basically you need to power down an iPhone to get full security. Doesn't that take like 10 seconds and a swipe. Hardly reassuring if you need to lock it for sure in a hurry.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: