3 ms·
Note that PAC is broken on all M1-based Apple devices, so the most secure device on the list may be the A15-based iPad Mini, https://9to5mac.com/2022/06/10/pacm
by walterbell 4y ago
Note that PAC is broken on all M1-based Apple devices, so the most secure device on the list may be the A15-based iPad Mini, https://9to5mac.com/2022/06/10/pacman-m1-chip/ https://9to5mac.com/2022/06/10/pacman-m1-chip/
Apple needs to provide a way to test and report the integrity of the radio baseband firmware on a device.
4G LTE is susceptible to IMSI catchers / fake cellular base stations, https://news.ycombinator.com/item?id=32237621 https://news.ycombinator.com/item?id=32237621
After installing E2EE messaging app, you can block all network traffic to Apple, then whitelist the Apple notification servers, https://support.apple.com/en-us/HT203609 https://support.apple.com/en-us/HT203609 and iOS security update servers, https://support.apple.com/en-us/HT210060 https://support.apple.com/en-us/HT210060
- tptacek 4y agoWhy does Apple need to provide a way to test and report the integrity of baseband firmware? I'm not saying it wouldn't be a good thing; I suppose it would be. But people believe very weird things about iPhone/iPad basebands.
- walterbell 4y agoMay 2022, https://dl.acm.org/doi/abs/10.1145/3507657.3528547 https://dl.acm.org/doi/abs/10.1145/3507657.3528547 > On recent iPhones, Bluetooth, Near Field Communication (NFC), and Ultra-wideband (UWB) keep running after power off, and all three wireless chips have direct access to the secure element. As a practical example what this means to security, we demonstrate the possibility to load malware onto a Bluetooth chip that is executed while the iPhone is off. LTE baseband issues, not specific to Apple, https://gist.github.com/adaburrows/fda8711e468858fc5ace98daf2d73148 https://gist.github.com/adaburrows/fda8711e468858fc5ace98daf...
- tptacek 4y agoThe paper lays this out: there are baseband features that use the Secure Element, and talk to it over I2C, because it stores secrets for things like (I guess?) car keys. That doesn't mean those chips can DMA things into and out of the Secure Element. For clarity: the Secure Element is the payment chip in the phone; it's not the SEP, the "Secure Enclave". The Enclave's memory is, in addition, hardware-encrypted. The cellular baseband on an iPhone is an HSIC peripheral.
- walterbell 4y agoDefense in Depth is good (TM). It's still worthwhile to deny malware from foothold on iPhone radios, e.g. an official process to verify baseband integrity and if necessary, restore official firmware without waiting for the next baseband security update.
- tptacek 4y agoI agree, more transparency is good. Baseband transparency would be a good thing. But even more important is a sound architecture. Lots of people believe that 3rd-party firmware running on iPhones can read and write any memory on the phone (and also that "DMA" means "god mode", like it did in 1997). The fact is that Apple (and Google, for its flagship phones) spend a fortune on internal and external design and implementation expertise that all starts from the premise of "code we didn't write is suspicious as fuck" and works back from there to a secure design. There Will Be Bugs! I just hope people recognize that they haven't outguessed Apple and Google simply by noticing that third-party firmware developers ship blobs instead of source repos. It's a hugely interesting space for security development and research, and I wince when I feel like it's getting oversimplified. Which you didn't do! It's just a hobbyhorse.
- als0 4y agoDoes the A15 have PAC? If not, then why is it _more_ secure?
- walterbell 4y agoA12 onward has PAC. P0 review of PAC on the A12, https://googleprojectzero.blogspot.com/2019/02/examining-pointer-authentication-on.html https://googleprojectzero.blogspot.com/2019/02/examining-poi...
- als0 4y agoI haven't seen anything that suggests the A15 is not vulnerable to PACMAN. Do you have a reference?
- walterbell 4y agoSince M2 is now shipping and is based on A15, we'll find out soon enough.
- olliej 4y agoPAC is not broken, any more than bounds checking is broken on intel hardware. PAC has a weakness, you still need to be able to exploit that weakness, before you get to the point of being able to make use of it. in addition PACMAN requires that you have vastly more control of execution than the prior cache attacks that could be trivially exploited. So claiming that PAC is broken is nonsense. As far as the baseband, what are you talking about? What do you want to be able to do to “verify” the already signed update images? If you really tinfoil hat level believe that the baseband of your device has been compromised, why do you think any query to that fully compromised device wouldn’t simply return a copy of the original baseband image?