Non-intrusive hardware acceleration for dynamic binary translation in embedded systems

Abstract

This article describes a non-intrusive hardware acceleration approach for Dynamic Binary Translation (DBT) in modern resource-constrained embedded systems, detailing its motivation, design decisions and overall architecture. It was deployed and tested on DBTOR, an in-house DBT system that targets constrained embedded systems. The performed evaluations demonstrate the feasibility of the proposed method in handling condition code (CC) flags, peripheral remapping and interrupt support, by running legacy MCS-51 code on a modern Arm v7-M architecture (Cortex-M3) that resorts field-programmable gate array (FPGA) technology for acceleration purposes.