Black Hat: Apple Outlines iOS Security
Apple takes the stage at the hacker conference to demonstrate how security has been baked into iOS devices, from the hardware on up.
LAS VEGAS. Security vulnerabilities in Apple products have been frequent topics of discussion at Black Hat conferences for many years, but this year's event marks the first time that Apple has sent an employee to give a presentation at the hacker confab.
In a packed session today, Apple Platform Security Manager Dallas De Ately explained in great detail how iOS is built for security from the ground up.
"iOS is different than Mac," De Atley said. "The baseband components are always on for notifications, the device never fully goes to sleep, and the phone is always connected."
De Atley added that with iOS there is always networking infrastructure that is aware of the device and can send it data as well.
"The phone knows a lot about how we live our lives and it has a lot of personal data," De Atley said. "We thought about all these things when building the iOS operating system."
The security of iOS starts with a Secure Boot model. De Atley explained that secure boot allows the device to trust the kernel it is running. He noted that the iOS kernel is the ultimate authority for allowing things to run and secure boot is how the device and the users can trust the phone.
Secure Boot is built right into the Apple A5 processor that powers iOS devices. The boot ROM on the A5 has a fingerprint of the root Certificate Authority for Apple. De Atley explained that this allows the boot ROM to validate all the components on an IOS device. Apple digitally signs each piece of the firmware so that a chain of trust can be established as the device is booting.
De Atley explained that iOS also include personalization protection such that a downgrade attack cannot be executed. Apple routinely patches iOS for issues and De Atley noted that over 80 percent of customers are on the latest version now. In a downgrade attack, an attacker tries to force an older version of firmware to run. That can't happen on iOS, according to De Atley: The secure boot model prevents it.
App Code Signing
Moving up the stack from the kernel to userspace, all executable app code is verified and signed by Apple.
"This fundamentally represents our first line of defence on the device," De Atley said. "There is an entire class of issues that we sidestep by ensuring that the applications are verified."
Apple has also limited the attack surface on iOS by not including unnecessary tools like having a root shell. Additionally, iOS separates the operating system from the user data.
"We physically separated the user's data so we can focus on protecting the user's data and not worry about it mingling with the OS," De Atley said.
Going even further, all third-party apps live in a sandboxed container. De Atley explained that the container is randomly assigned every time the application is run. That means that apps are not hardcoded in terms of where they live on the device, providing an additional layer of protection.
All user data on iOS is cryptographically signed by way of features built into the Apple A-series processors. De Atley stressed that every file is encrypted with a 256 bit key.
As an additonal layer of protection, when a user chooses to delete data from an iOS device, Apple has developed a new mechanism to securely delete NAND Flash storage, called effaceable storage.
"It lets us address specific regions on a flash drive and then reliably erase data," De Atley said.
Even at the basic level of an attacker trying to get access to iOS by way of hacking a users passcode, Apple has security. De Atley said that iOS has mitigations such that a brute force attack against the passcode cannot be executed.
De Atley's talk ended with a summary of his main points, and then much to the audience's surprise, he did not take questions. Instead, Black Hat event staff whisked him off the stage out a side door.