Formalizing an Ssa-based Compiler for Verified Advanced Program Transformations

FORMALIZING AN SSA-BASED COMPILER FOR VERIFIED ADVANCED PROGRAMTRANSFORMATIONSJianzhou ZhaoSupervisor: Steve Zdancewic Compilers are not always correct due to the complexity of language semantics and transformation algo-rithms, the trade-offs between compilation speed and verifiability, etc. The bugs of compilers can underminethe source-level verification efforts (such as type systems, static analysis, and formal proofs) and producetarget programs with different meaning from source programs. Researchers have used mechanized prooftools to implement verified compilers that are guaranteed to preserve program semantics and proved to bemore robust than ad-hoc non-verified compilers.The goal of the dissertation is to make a step towards verifying an industrial strength modern compiler—LLVM, which has a typed, SSA-based, and general-purpose intermediate representation, therefore allowingmore advanced program transformations than existing approaches. The dissertation formally defines thesequential semantics of the LLVM intermediate representation with its type system, SSA properties, memorymodel, and operational semantics. To design and reason about program transformations in the LLVM IR,we provide tools for interacting with the LLVM infrastructure and metatheory for SSA properties, memorysafety, dynamic semantics, and control-flow-graphs. Based on the tools and metatheory, the dissertationimplements verified and extractable applications for LLVM that include an interpreter for the LLVM IR, atransformation for enforcing memory safety, translation validators for local optimizations, and verified SSAconstruction transformation.This dissertation shows that formal models of SSA-based compiler intermediate representations canbe used to verify low-level program transformations, thereby enabling the construction of high-assurancecompiler passes.