Guest editors’ introduction

The emergence of a new computer architecture is a rare event. Designing a new instruction-set architecture (ISA) is very expensive, and it is often necessary to recompile several operating systems and a large number of application programs. Binary translation attempts to alleviate much of this effort and the associated risk: recompilation is replaced by automatic conversion of code from the legacy ISA to the new one. Even with modern programming languages, language semantics are typically not defined tightly enough to make recompilation a transparent process. However, the semantics of binary code is usually well defined, facilitating automatic and transparent translation. Binary translation has been practiced for many years. But only with the recent increases in processing power has it become possible to utilize translation fully. In the last few years the field has expanded, including the development of a number of new binary translation systems and research projects. The field has matured significantly, evolving a set of common concepts. We begin with a tutorial on binary translation, followed by three papers written by practitioners, each demonstrating a different approach: