AN-Encoding Compiler: Building Safety-Critical Systems with Commodity Hardware

Abstract

Abstract. In the future, we expect commodity hardware to be used in safety-critical applications. However, in the future commodity hardware is expected to become less reliable and more susceptible to soft errors because of decreasing feature size and reduced power supply. Thus, software-implemented approaches to deal with unreliable hardware will be needed. To simplify the handling of value failures, we provide failure virtualization in the sense that we transform arbitrary value failures caused by erroneous execution into fail-stop failures. The latter ones are easier to handle. Therefore, we use the arithmetic AN-code because it provides very good error detection capabilities. Arithmetic codes are suitable for the protection of commodity hardware because guarantees can be provided independent of the executing hardware. This paper presents the encoding compiler EC-AN which applies AN-encoding to arbitrary programs. According to our knowledge, this is the first in software implemented complete AN-encoding. Former encoding compilers either encode only small parts of applications or trade-off safety to enable complete AN-encoding.

Operating system support for redundant multithreading

“…Reis et al integrated control-flow checking and error detection into one compiler, SWIFT [34]. SWIFT still had a couple of vulnerabilities that were addressed by Fetzer et al [11] using arithmetic coding techniques.…”

“…In contrast to the application layer, we have full control over the software stack running in the critical core. This allows us to make use of potentially expensive techniques available at the software level such as applying basic block signatures [26], operand encoding [11], and compiler-inserted assertions [45].…”

Generative software-based memory error detection and correction for operating system data structures

“…Surprisingly, in spite of their impact on total system resiliency and -compared to the rest of the system -their very small memory footprint, state-of-the-art OS kernels are not equipped with softwarebased protection against memory errors: An efficient softwarebased fault-tolerance technique would offer an enormous potential to reduce system failures. Unfortunately, most earlier attempts to apply software-based memory protection suffer from excessive runtime overhead, ranging between 30 % and 260 % [13], [14], [15], [16]. These studies only address userlevel applications; such extreme performance degradations are considered unacceptable for the OS layer, especially in the case of general-purpose OS.…”

“…Fetzer et al [16] use arithmetic AN-encoding of memory (among other methods) to detect errors by essentially doubling the storage space for encoded values. Even at this high level of redundancy, recovery is unaddressed.…”

Automated application of fault tolerance mechanisms in a component-based system

“…To reduce costs and make soft error tolerance available to a wider range of application scenarios, software-based fault handling methods for control flow [15,16,14,17,29,22,6] or memory protection [23] were developed over the recent years. Lately, Chang et al [21] proposed SWIFT-R, an approach that triplicates machine instructions and uses a majority voter in combination with encoding techniques.…”