argXtract: Deriving IoT Security Configurations via Automated Static Analysis of Stripped ARM Binaries

TL;DR

argXtract is an open-source static analysis tool that automatically extracts security configuration details from stripped ARM Cortex-M IoT firmware, revealing widespread security weaknesses across real-world device datasets.

Contribution

The paper introduces argXtract, a novel tool for automated static analysis of stripped ARM IoT firmware, enabling large-scale security configuration assessment.

Findings

Widespread security and privacy weaknesses found in IoT firmware.

argXtract successfully analyzes 243 Bluetooth LE binaries.

Automated analysis reveals minimal security protections in real-world devices.

Abstract

Recent high-profile attacks on the Internet of Things (IoT) have brought to the forefront the vulnerability of "smart" devices, and have resulted in numerous IoT-focused security analyses. Many of the attacks had weak device configuration as the root cause. One potential source of rich and definitive information about the configuration of an IoT device is the device's firmware. However, firmware analysis is complex and automated firmware analyses have thus far been confined to devices with more traditional operating systems such as Linux or VxWorks. Most IoT peripherals, due to lacking traditional operating systems and implementing a wide variety of communication technologies, have only been the subject of smaller-scale analyses. Peripheral firmware analysis is further complicated by the fact that such firmware files are predominantly available as stripped binaries, without the ELF headers and symbol tables that would simplify reverse engineering. In this paper, we present argXtract, an open-source automated static analysis tool, which extracts security-relevant configuration information from stripped IoT peripheral firmware. Specifically, we focus on binaries that target the ARM Cortex-M architecture, due to its growing popularity among IoT peripherals. argXtract overcomes the challenges associated with stripped Cortex-M analysis and is able to retrieve arguments to security-relevant supervisor and function calls, enabling automated bulk analysis of firmware files. We demonstrate this via three real-world case studies. The largest case study covers a dataset of 243 Bluetooth Low Energy binaries targeting Nordic Semiconductor chipsets, while the other two focus on Nordic ANT and STMicroelectronics BlueNRG binaries. The results reveal widespread lack of security and privacy controls in IoT, such as minimal or no protection for data, fixed passkeys and trackable device addresses.