Month: December 2017

I swear Apple fanboys are some of the dumbest people on the planet. Quite a few of them have been saying, “If an attacker as physical access, it’s game over anyways,” as if that statement makes the root user exploit recently discovered in High Sierra a nonissue.

At one time that statement was true. However, today physical access is not necessarily game over. Look at all of the trouble the Federal Bureau of Investigations (FBI) has been having with accessing iOS devices. The security model of iOS actually takes physical access into account as part of its threat modeling and has mechanisms to preserve the integrity of the data contained on the device. iOS requires all code to be signed before it will install or run it, which makes it difficult, although far from impossible, to insert malicious software onto iOS devices. But more importantly iOS encrypts all of the data stored in flash memory by default. Fully encrypted disks protect against physical access by both preventing an attacker from getting any usable data from a disk and also by preventing them from altering the data on the disk (such as writing malware directly to the disk).

macOS has a boot mode called single user mode, which boots the computer to a root command prompt. However, if a firmware password is set, single user mode cannot be started without entering the firmware password. The firmware password can be reset on machines with removable RAM (resetting the password requires changing the amount of RAM connected to the mainboard) but most of Apple’s modern computers, some iMacs being the exception, have RAM modules that are soldered to the mainboard.

Physical access is especially dangerous because it allows an attacker to insert malicious hardware, such as a key logger, that would allow them to record everything you type, including your passwords. However, that kind of attack requires some amount of sophistication and time (at least if you want the malicious hardware to be difficult to detect), which is where the real problem with High Sierra’s root exploit comes in. The root exploit required no sophistication whatsoever. Gaining root access only required physical access (or remote access if certain services were enabled) to an unlocked Mac for a few seconds. So long as an attacker had enough time to open System Preferences, click one of the lock icons, and type in “root” for the user name a few times they had complete access to the machine (from there they could turn on remote access capabilities to maintain their access).

Attempting to write off this exploit as a nonissue because it requires physical access requires willful ignorance of both modern security features that defend against attackers with physical access and the concept of severity (an attack that requires no sophistication can be far more severe than a time consuming sophisticated attack under certain threat models).