Translation validation

Abstract

Abstract. We present the notion of translation validation as a new approach to the verification of translators (compilers, code generators). Rather than proving in advance that the compiler always produces a target code which correctly implements the source code (compiler verification), each individual translation (i.e. a run of the compiler) is followed by a validation phase which verifies that the target code produced on this run correctly implements the submitted source program. Several ingredients are necessary to set up the -fully automatic -translation validation process, among which are:1. A common semantic framework for the representation of the source code and the generated target code. 2. A formalization of the notion of "correct implementation" as a refinement relation. 3. A syntactic simulation-based proof method which allows to automatically verify that one model of the semantic framework, representing the produced target code, correctly implements another model which represents the source. These, and other ingredients are elaborated in this paper, in which we illustrate the new approach in a most challenging case. We consider a translation (compilation) from the synchronous multi-clock data-flow language SIGNAL to asynchronous (sequential) C-code.

Translation Validation of High-Level Synthesis

“…Our work is related to translation validation [19], [33], [42], [46], [47], [53], [54], relational approaches to reasoning about programs [5], [7], [15], [28], [35], CSP refinement checking [13], [27], [39], [51], and HLS verification [2], [14], [31], [34], [41]. We now discuss each area in more detail.…”

“…Relational approaches are a common tool for reasoning about programs, and they have been used for a variety of verification tasks, including model checking [7], [15], translation validation [42], [46], and reasoning about optimizations once and for all [5], [35]. In this context, our work is inspired by Josephs's approach [28] for proving refinements.…”

“…We achieve this by capturing the possibly infinite state space of data using formulas and using a theorem prover to reason about these formulas. Although, this technique is well known and has been used in dataflow analysis [16], [20], model checking [4], [8], [9], [24], and translation validation [42], [46]. The use of this technique in the context of checking CSP refinements appears to be novel.…”

“…Although this approach does not guarantee that the HLS process is bug free, it does guarantee that any errors in translation will be caught when the particular steps of HLS are performed, preventing such errors from propagating any further in the hardware fabrication process. This approach to verification, called translation validation, has previously been applied with success in the context of optimizing compilers [19], [42], [46], [47], [53].…”

Formal Verification of a Memory Model for C-Like Imperative Languages

“…The translation validation approach [18,19,10,20,21] aims at validating every run of the compiler, producing a formal proof that the produced target code is a correct implementation of the source code. This approach is based on program checking and on static analysis.…”