Building a Secure Bootloader: Considerations and Recommendations

As the number of so-called Internet-of-Things is continuously rising and the urge for regular hardware updates has never been higher, the security of our smart objects all around us is often skipped due to lack of development time. By default, these devices are equipped with a primitive bootloader, which is executing code located at a designated address. Within this document, we are describing a simple but worthwhile validation mechanism, which is realized by using digital signature algorithm included in the bootloader. As a result, we are making sure that no malicious code or malware can be executed and the running application is originating from a trusted and known developer. Nevertheless, even the strongest cryptographic system cannot ensure the highest possible security, if the device itself is prone to fault injection attacks. A simple alteration of the voltage or clock speed is sufficient in order to unsettle the device's usual and intended behaviour and cause instruction skipping among others. At worst, this could result in bypassing of authentication mechanisms, one of the most common targets of an attack. With the aid of randomly-chosen delays in instruction processing as well as multiple use of authentication mechanisms, we are describing straightforward mechanisms in order to decrease the level of success of common fault injections. Moreover, a list of recommendations is given in this document suitable for situations, where resources are limited and no additional hardware can be added.This work was partially supported by the Eurostars-2 joint programme with co-funding from the European Union Horizon 2020 research and innovation programme under the grant agreement E11897 RESCURE "Retrofit Security for Critical infrastructures". Furthermore, this work was also partially supported by the Austrian Ministry for Transport, Innovation and Technology under the framework of "IKT der Zukunft" with the FFG grant agreement project 865233