Proof-carrying code is a framework for the mechanical verification of safety properties of machine language programs, but the problem arises of quis custodiat ipsos custodes---who will verify the verifier itself? Foundational proof-carrying code is verification from the smallest possible set of axioms, using the simplest possible verifier and the smallest possible runtime system. I will describe many of the mathematical and engineering problems to be solved in the construction of a foundational proof-carrying code system.

A type system withlin earity is useful for checking software protocols and resource man agemen t at compile time. Lin#"$"M y provides powerfulreason#)M about state chan ges, but at the price of restriction son aliasin g. The hard division between lin ear an dn on lin ear types forces the programmer to make a trade-off between checkin g a protocol on an object an# aliasin# the object. ost on erous is the restriction that an y type with alin ear component must itself be linear. Because of this, checking a protocolon an object imposes aliasin g restriction s on an y data structure that directly orin directly points to the object. We propose an#$ type system that reduces these restrictions with the adoption and focus constructs. Adoption safely allows a programmer to alias objects on which she is checkin g protocols, and focus allows the reverse. A programmer can alias data structures that point to linear objects and use focus for safe access to those objects. We discuss how we implemented these ideas in the Vault programming language.

Linear type systems permit programmers to deallocate or explicitly recycle memory, but they are severly restricted by the fact that they admit no aliasing. This paper describes a pseudo-linear type system that allows a degree of aliasing and memory reuse as well as the ability to define complex recursive data structures. Our type system can encode conventional linear data structures such as linear lists and trees as well as more sophisticated data structures including cyclic and doubly-linked lists and trees. In the latter cases, our type system is expressive enough to represent pointer aliasing and yet safely permit destructive operations such as object deallocation. We demonstrate the flexibility of our type system by encoding two common compiler optimizations: destination-passing style and Deutsch-Schorr-Waite or "link-reversal" traversal algorithms.

Today's mainstream object-oriented compilers and tools do not support declaring and statically checking simple pre- and postconditions on methods and invariants on object representations. The main technical problem preventing static verification is reasoning about the sharing relationships among objects as well as where object invariants should hold. We have developed a programming model of typestates for objects with a sound modular checking algorithm. The programming model handles typical aspects of object-oriented programs such as downcasting, virtual dispatch, direct calls, and subclassing. The model also permits subclasses to extend the interpretation of typestates and to introduce additional typestates. We handle aliasing by adapting our previous work on practical linear types developed in the context of the Vault system. We have implemented these ideas in a tool called Fugue for specifying and checking typestates on Microsoft .NET-based programs.

We develop an explicit two level system that allows programmers to reason about the behavior of effectful programs. The first level is an ordinary ML-style type system, which confers standard properties on program behavior. The second level is a conservative extension of the first that uses a logic of type refinements to check more precise properties of program behavior. Our logic is a fragment of intuitionistic linear logic, which gives programmers the ability to reason locally about changes of program state. We provide a generic resource semantics for our logic as well as a sound, decidable, syntactic refinement-checking system. We also prove that refinements give rise to an optimization principle for programs. Finally, we illustrate the power of our system through a number of examples.

We observe that the principal typing property of a type system is the enabling technology for modularity and separate compilation [10]. We use this technology to formulate a modular and polyvariant closure analysis, based on the rank 2 intersection types annotated with control-flow information. Modularity manifests itself in a syntax-directed, annotated-type inference algorithm that can analyse program fragments containing free variables: a principal typing property is used to formalise it. Polyvariance manifests itself in the separation of different behaviours of the same function at its different uses: this is formalised via the rank 2 intersection types. As the rank 2 intersection type discipline types at least all (core) ML programs, our analysis can be used in the separate compilation of such programs. 1

Machine We have described the type constructor language of CL and the typing rules for the main term-level constructs. In fact, the previous section contains all of the ACM Transactions on Programming Languages and Systems, Vol. TBD, No. TDB, Month Year. 20 D. Walker, K. Crary, and G. Morrisett #; #;# # h at r : # # # # f : Type #; ## # ; #{f :# f , x 1 :# 1 , . . . , xn :# n}; C # e # # f = #[# # ].(C, # 1 , . . . , #n ) # 0 at r f, x 1 , . . . , xn ## Dom(#) # #; #;# # fix f[# # ](C, x 1 :# 1 , . . . , xn :# n ).e at r : # f (h-fix) #; #;# # v i : # i (for 1 # i # n) # # r : Rgn #; #;# # #v 1 , . . . , vn # at r : ## 1 , . . . , #n # at r (h-tuple) #; #;# # h at r : # # # # # # = # : Type #; #;# # h at r : # (h-eq) #; #;# # v : # #; #;# # x : # (#(x) = #) (v-var) #; #;# # i : int (v-int) #; #;# # v : #[#:#, # # ].(C, # 1 , . . . , #n ) # 0 at r # # c : # #; #;# # v[c] : (#[# # ].(C, # 1 , . . . , #n ) # 0)[c/#] at r (v-type) #; #;# # v : #[# # C ## , # # ].(C # , # 1 , . . . , #n ) # 0 at r # # C # C ## #; #;# # v[C] : (#[# # ].(C # , # 1 , . . . , #n ) # 0)[C/#] at r (v-sub) #; #;# # v : # # # # # # = # : Type #; #;# # v : # (v-eq) Fig. 6. Capability static semantics: Heap and word values. information programmers or compilers require to write type-safe programs in CL. However, in order to prove a type soundness result in the style of Wright and Felleisen [Wright and Felleisen 1994], we must be able to type check programs at every step during their evaluation. In this section, we give the static semantics of the run-time values that are not normally manipulated by programmers, but are nevertheless necessary to prove our soundness result. At first, the formal definition ...

We present a new approach for supporting user-defined type refinements, which augment existing types to specify and check additional invariants of interest to programmers. We provide an expressive language in which users define new refinements and associated type rules. These rules are automatically incorporated by an extensible typechecker during static typechecking of programs. Separately, a soundness checker automatically proves that each refinement’s type rules ensure the intended invariant, for all possible programs. We have formalized our approach and have instantiated it as a framework for adding new type qualifiers to C programs. We have used this framework to define and automatically prove sound a host of type qualifiers of different sorts, including pos and neg for integers,tainted anduntainted for strings, andnonnull and unique for pointers, and we have applied our qualifiers to ensure important invariants on open-source C programs.

In prior work [15] we studied a language construct restrict that allows programmers to specify that certain pointers are not aliased to other pointers used within a lexical scope. Among other applications, programming with these constructs helps program analysis tools locally recover strong updates, which can improve the tracking of state in flow-sensitive analyses. In this paper we continue the study of restrict and introduce the construct confine. We present a type and effect system for checking the correctness of these annotations, and we develop efficient constraint-based algorithms implementing these type checking systems. To make it easier to use restrict and confine in practice, we show how to automatically infer such annotations without programmer assistance. In experiments on locking in 589 Linux device drivers, confine inference can automatically recover strong updates to eliminate 95% of the type errors resulting from weak updates.

Scientists and engineers must ensure that the equations and formulae which they use are dimensionally consistent, but existing programming languages treat all numeric values as dimensionless. This thesis investigates the extension of programming languages to support the notion of physical dimension. A type system is presented similar to that of the programming language ML but extended with polymorphic dimension types. An algorithm which infers most general dimension types automatically is then described and proved correct. The semantics of the language is given by a translation into an explicitlytyped language in which dimensions are passed as arguments to functions. The operational semantics of this language is specified in the usual way by an evaluation relation defined by a set of rules. This is used to show that if a program is well-typed then no dimension errors can occur during its evaluation. More abstract properties of the language are investigated using a denotational semantics: these include a notion of invariance under changes in the units of measure used, analogous to parametricity in the polymorphic lambda calculus. Finally the dissertation is summarised and many possible directions for future research in dimension types and related type systems are described. i ii