Generic Soft-Error Detection and Correction for Concurrent Data Structures

Abstract

Recent studies indicate that transient memory errors (soft errors) have become a relevant source of system failures. This paper presents a generic software-based fault-tolerance mechanism that transparently recovers from memory errors in object-oriented program data structures. The main benefits are the flexibility to choose from an extensible toolbox of easily pluggable error detection and correction schemes, such as Hamming and CRC codes. This is achieved by a combination of aspect-oriented and generative programming techniques. Furthermore, we present a wait-free synchronization algorithm for error detection in data structures that are used concurrently by multiple threads of control. We give a formal correctness proof and show the excellent scalability of our approach in a multiprocessor environment. In a case study, we present our experiences with selectively hardening the eCos operating system and its benchmark suite. We explore the trade-off between resiliency and performance by choosing only the most vulnerable data structures for error recovery. Thereby, the total number of system failures, manifesting as silent data corruptions and crashes, is reduced by 69.14 % at a negligible runtime overhead of 0.36 %.

Untitled

“…Consequently, the project is following a proven method of collecting, tabulating, and analyzing the background that has been obtained and proving its validity directly in the field in which the research fact is being presented (Borchert, Schirmeier & Spinczyk, 2017).…”

Dependability Aspects in Configurable Embedded Operating Systems

“…These works have been previously published in conference proceedings and journals [8,29,36], and are presented here in a summarized manner. Section 6 concludes the chapter and summarizes the results of the DanceOS project, which was funded by the German Research Foundation (DFG) over a period of 6 years as part of the priority program SPP 1500 "Dependable Embedded Systems" [26] ( Fig.…”

“…Our experience with the embedded operating system eCos shows that OS kernel data structures are highly susceptible to soft errors in main memory [8]. Several kernel data structures, such as the process scheduler, persist during the whole OS uptime, which increases the chance of being hit by a random soft error.…”

“…By defining such generic pieces of advice, AspectC++ enables a modular implementation of the GOP mechanism, completely separated from the remaining source code. More advice definitions exist in the complete GOP implementation, for instance, covering static data members, non-blocking synchronization, or virtual-function pointers [8].…”

“…We evaluate the five EDM/ERM variants on eCos 3.0 with a subset of the benchmark and test programs that are bundled with eCos itself, namely those 19 implemented in C++ and using threads (omitting CLOCK1 and CLOCKTRUTH due to their extremely long runtime). More details on the benchmarks can be found in previous work [8]. Because eCos currently does not support x64, all benchmarks are compiled for i386 with the GNU C++ compiler (GCC Debian 4.7.2-5), and eCos is set up with its default configuration.…”