No abstract

Proof-Carrying Code (PCC) is a general framework for

Today's software systems often use many different computation features and span different abstraction levels (e.g., user code and runtime-system code). To build foundational certified systems, it is hard to have a single verification system supporting all computation features. In this paper we present an open framework for foundational proof-carrying code (FPCC). It allows program modules to be specified and certified separately using different type systems or program logics. Certified modules (i.e., code and proof) can be linked together to build fully certified systems. The framework supports modular verification and proof reuse. It is also expressive enough so that invariants established in specific verification systems are preserved even when they are embedded into our framework. Our work presents the first FPCC framework that systematically supports interoperation between different verification systems. It is fully mechanized in the Coq proof assistant with machine-checkable soundness proof.

Runtime stacks are critical components of any modern softwarethey are used to implement powerful control structures such as function call/return, stack cutting and unwinding, coroutines, and thread context switch. Stack operations, however, are very hard to reason about: there are no known formal specifications for certifying C-style setjmp/longjmp, stack cutting and unwinding, or weak continuations (in C--). In many proof-carrying code (PCC) systems, return code pointers and exception handlers are treated as general first-class functions (as in continuation-passing style) even though both should have more limited scopes.In this paper we show that stack-based control abstractions follow a much simpler pattern than general first-class code pointers. We present a simple but flexible Hoare-style framework for modular verification of assembly code with all kinds of stackbased control abstractions, including function call/return, tail call, setjmp/longjmp, weak continuation, stack cutting, stack unwinding, multi-return function call, coroutines, and thread context switch. Instead of presenting a specific logic for each control structure, we develop all reasoning systems as instances of a generic framework. This allows program modules and their proofs developed in different PCC systems to be linked together. Our system is fully mechanized. We give the complete soundness proof and a full verification of several examples in the Coq proof assistant.

Abstract. Formal, modular, and mechanized verification of realistic systems code is desirable but challenging. Verification of machine context management (a basis of multi-tasking) is one representative example. With context operations occurring hundreds to thousands of times per second on every computer, their correctness deserves careful examination. Given the small and stable code bases, it is a common misunderstanding that the context management code is suitable for informal scrutiny and testing. Unfortunately, after being extensively studied and used for decades, it still proves to be a common source of bugs and confusion. Yet its verification remains difficult due to the machine-level detail, irregular patterns of control flows, and rich application scenarios. This paper reports our experience applying XCAP-a recent theoretical verification framework-to certify a realistic x86 implementation of machine context management. XCAP supports expressive and modular logical specifications, but has only previously been applied on simple idealized machine and code. By applying the XCAP theory to an x86 machine model, building libraries of common proof tactics and lemmas, composing specifications for the context data structures and routines, and proving that the code behave accordingly, we achieved the first formal, modular, and mechanized verification of realistic x86 context management code. Our proofs are fully mechanized in the Coq proof assistant. Our certified library code runs on stock hardware and can be linked with other certified systems and application code. Our technique applies to other variants or extensions of context management (e.g., more complex context, different platforms), provides a solid basis for further verification of thread implementation and concurrent programs, and illustrates how to achieve formal, modular, and mechanized verification of realistic systems code.

Runtime stacks are critical components of any modern softwarethey are used to implement powerful control structures such as function call/return, stack cutting and unwinding, coroutines, and thread context switch. Stack operations, however, are very hard to reason about: there are no known formal specifications for certifying C-style setjmp/longjmp, stack cutting and unwinding, or weak continuations (in C--). In many proof-carrying code (PCC) systems, return code pointers and exception handlers are treated as general first-class functions (as in continuation-passing style) even though both should have more limited scopes.In this paper we show that stack-based control abstractions follow a much simpler pattern than general first-class code pointers. We present a simple but flexible Hoare-style framework for modular verification of assembly code with all kinds of stackbased control abstractions, including function call/return, tail call, setjmp/longjmp, weak continuation, stack cutting, stack unwinding, multi-return function call, coroutines, and thread context switch. Instead of presenting a specific logic for each control structure, we develop all reasoning systems as instances of a generic framework. This allows program modules and their proofs developed in different PCC systems to be linked together. Our system is fully mechanized. We give the complete soundness proof and a full verification of several examples in the Coq proof assistant.

Proof-Carrying Code (PCC) is a general framework for

No abstract