Un vérificateur de modèle peut produire une trace de contreexemple, pour un programme erroné, qui est souvent difficile à exploiter pour localiser les erreurs dans le code source. Dans ma thèse, nous avons proposé un algorithme de localisation d'erreurs à partir de contreexemples, nommé LocFaults, combinant les approches de Bounded Model Checking (BMC) avec un problème de satisfaction de contraintes (CSP). Cet algorithme analyse les chemins du CFG (Control Flow Graph) du programme erroné pour calculer les sous-ensembles d'instructions suspectes permettant de corriger le programme. En effet, nous générons un système de contraintes pour les chemins du graphe de flot de contrôle pour lesquels au plus k instructions conditionnelles peuvent être erronées. Ensuite, nous calculons les MCSs (Minimal Correction Sets) de taille limitée sur chacun de ces chemins. La suppression de l'un de ces ensembles de contraintes donne un sous-ensemble satisfiable maximal, en d'autres termes, un sous-ensemble maximal de contraintes satisfaisant la postcondition. Pour calculer les MCSs, nous étendons l'algorithme générique proposé par Liffiton et Sakallah dans le but de traiter des programmes avec des instructions numériques plus efficacement. Cette approche a été évaluée expérimentalement sur des programmes académiques et réalistes. / A model checker can produce a trace of counter-example for erroneous program, which is often difficult to exploit to locate errors in source code. In my thesis, we proposed an error localization algorithm from counter-examples, named LocFaults, combining approaches of Bounded Model-Checking (BMC) with constraint satisfaction problem (CSP). This algorithm analyzes the paths of CFG (Control Flow Graph) of the erroneous program to calculate the subsets of suspicious instructions to correct the program. Indeed, we generate a system of constraints for paths of control flow graph for which at most k conditional statements can be wrong. Then we calculate the MCSs (Minimal Correction Sets) of limited size on each of these paths. Removal of one of these sets of constraints gives a maximal satisfiable subset, in other words, a maximal subset of constraints satisfying the postcondition. To calculate the MCSs, we extend the generic algorithm proposed by Liffiton and Sakallah in order to deal with programs with numerical instructions more efficiently. This approach has been experimentally evaluated on a set of academic and realistic programs.

La programmation par contraintes offre des langages et des outils permettant de résoudre des problèmes à forte combinatoire et à la complexité élevée tels que ceux qui existent en vérification de programmes. Dans cette thèse nous résolvons deux familles de problèmes de la vérification de programmes. Dans chaque cas de figure nous commençons par une étude formelle du problème avant de proposer des modèles en contraintes puis de réaliser des expérimentations. La première contribution concerne un langage réactif synchrone représentable par une algèbre de diagramme de blocs. Les programmes utilisent des flux infinis et modélisent des systèmes temps réel. Nous proposons un modèle en contraintes muni d’une nouvelle contrainte globale ainsi que ses algorithmes de filtrage inspirés de l’interprétation abstraite. Cette contrainte permet de calculer des sur-approximations des valeurs des flux des diagrammes de blocs. Nous évaluons notre processus de vérification sur le langage FAUST, qui est un langage dédié à la génération de flux audio. La seconde contribution concerne les systèmes probabilistes représentés par des chaînes de Markov à intervalles paramétrés, un formalisme de spécification qui étend les chaînes de Markov. Nous proposons des modèles en contraintes pour vérifier des propriétés qualitatives et quantitatives. Nos modèles dans le cas qualitatif améliorent l’état de l’art tandis que ceux dans le cas quantitatif sont les premiers proposés à ce jour. Nous avons implémenté nos modèles en contraintes en problèmes de programmation linéaire en nombres entiers et en problèmes de satisfaction modulo des théories. Les expériences sont réalisées à partir d’un jeu d’essais de la bibliothèque PRISM. / Constraint programming offers efficient languages andtools for solving combinatorial and computationally hard problems such as the ones proposed in program verification. In this thesis, we tackle two families of program verification problems using constraint programming.In both contexts, we first propose a formal evaluation of our contributions before realizing some experiments.The first contribution is about a synchronous reactive language, represented by a block-diagram algebra. Such programs operate on infinite streams and model real-time processes. We propose a constraint model together with a new global constraint. Our new filtering algorithm is inspired from Abstract Interpretation. It computes over-approximations of the infinite stream values computed by the block-diagrams. We evaluated our verification process on the FAUST language (a language for processing real-time audio streams) and we tested it on examples from the FAUST standard library. The second contribution considers probabilistic processes represented by Parametric Interval Markov Chains, a specification formalism that extends Markov Chains. We propose constraint models for checking qualitative and quantitative reachability properties. Our models for the qualitative case improve the state of the art models, while for the quantitative case our models are the first ones. We implemented and evaluated our verification constraint models as mixed integer linear programs and satisfiability modulo theory programs. Experiments have been realized on a PRISM based benchmark.

Computer systems are used in almost all aspects of our lives and our dependency on them keeps on increasing. When computer systems are used to handle critical tasks, any software failure can cause severe human and/or material losses. Therefore, for such applications, it is important to detect software errors at an early stage of software development. Furthermore, the growing use of concurrent and distributed programs exponentially increases the complexity of computer systems, making the problem of detecting software errors even harder (if not impossible). This calls for defining systematic and efficient techniques to evaluate the safety and the correctness of programs. The aim of Model-Checking is to analyze automatically whether a given program satisfies its specification. Early applications of Model-Checking were restricted to systems whose behaviors can be captured by finite graphs, so called finite-state systems. Since many computer systems cannot be modeled as finite-state machines, there has been a growing interest in extending the applicability of Model-Checking to infinite-state systems. The goal of this thesis is to extend the applicability of Model Checking for three instances of infinite-state systems: Ad-Hoc Networks, Dynamic Register Automata and Multi Pushdown Systems. Each one of these instances models challenging types of networks of communicating processes. In both Ad-Hoc Networks and Dynamic Register Automata, communication is carried through message passing. In each type of network, a graph topology models the communication links between processes in the network. The graph topology is static in the case of Ad-Hoc Networks while it is dynamic in the case of Dynamic Register Automata. The number of processes in both types of networks is unbounded. Finally, we consider Multi Pushdown Systems, a model used to study the behaviors of concurrent programs composed of sequential recursive sequential programs communicating through a shared memory.

Cette thèse présente une formalisation en Coq des langages et des algorithmes fondamentaux portant sur les bases de données. Ainsi, ce fourni des spécifications formelles issues des deux approches différentes pour la définition des modèles de données: une basée sur l’algèbre et l'autre basée sur la logique.A ce titre, une première contribution de cette thèse est le développement d'une bibliothèque Coq pour le modèle relationnel. Cette bibliothèque contient les modélisations de l’algèbre relationnelle et des requêtes conjonctives. Il contient aussi une mécanisation des contraintes d'intégrité et de leurs procédures d'inférence. Nous modélisons deux types de contraintes: les dépendances, qui sont parmi les plus courantes: les dépendances fonctionnelles et les dépendances multivaluées, ainsi que leurs axiomatisations correspondantes. Nous prouvons formellement la correction de leurs algorithmes d'inférence et, pour le cas de dépendances fonctionnelles, aussi la complétude.Ces types de dépendances sont des instances de contraintes plus générales : les dépendances génératrices d'égalité (equality generating dependencies, EGD) et, respectivement, les dépendances génératrices de tuples (tuple generating dependencies, TGD), qui appartiennent a une classe encore plus large des dépendances générales (general dependencies). Nous modélisons ces dernières et leur procédure d'inférence, i.e, "the chase", pour lequel nous établissons la correction. Enfin, on prouve formellement les théorèmes principaux des bases de données, c'est-à-dire, les équivalences algébriques, la théorème de l' homomorphisme et la minimisation des requêtes conjonctives.Une deuxième contribution consiste dans le développement d'une bibliothèque Coq/ssreflect pour la programmation logique, restreinte au cas du Datalog. Dans le cadre de ce travail, nous donnons la première mécanisations d'un moteur Datalog standard et de son extension avec la négation. La bibliothèque comprend une formalisation de leur sémantique en theorie des modelés ainsi que de leur sémantique par point fixe, implémentée par une procédure d'évaluation stratifiée. La bibliothèque est complétée par les preuves de correction, de terminaison et de complétude correspondantes. Cette plateforme ouvre la voie a la certification d' applications centrées données. / This thesis presents a formalization of fundamental database theories and algorithms. This furthers the maturing state of the art in formal specification development in the database field, with contributions stemming from two foundational approches to database models: relational and logic based.As such, a first contribution is a Coq library for the relational model. This contains a mechanization of integrity constraints and of their inference procedures. We model two of the most common dependencies, namely functional and multivalued, together with their corresponding axiomatizations. We prove soundness of their inference algorithms and, for the case of functional ones, also completeness. These types of dependencies are instances of equality and, respectively, tuple generating dependencies, which fall under the yet wider class of general dependencies. We model these and their inference procedure,i.e, the chase, for which we establish soundness.A second contribution consists of a Coq/Ssreflect library for logic programming in the Datalog setting. As part of this work, we give (one of the) first mechanizations of the standard Datalog language and of its extension with negation. The library includes a formalization of their model theoretical semantics and of their fixpoint semantics, implemented through bottom-up and, respectively, through stratified evaluation procedures. This is complete with the corresponding soundness, termination and completeness proofs. In this context, we also construct a preliminary framework for dealing with stratified programs. This work paves the way towards the certification of data-centric applications.

Cette thèse s'inscrit dans la démarche de preuve de programmes à l'aide de vérification déductive. La vérification déductive consiste à produire, à partir des sources d'un programme, c'est-à-dire ce qu'il fait, et de sa spécification, c'est-à-dire ce qu'il est sensé faire, une conjecture qui si elle est vraie alors le programme et sa spécification concordent. On utilise principalement des démonstrateurs automatiques pour montrer la validité de ces formules. Quand ons'intéresse à la preuve de programmes qui utilisent des structures de données allouées en mémoire, il est élégant et efficace de spécifier son programme en utilisant la logique de séparation qui est apparu il y a une dizaine d'année. Cela implique de prouver des conjectures comportant les connectives de la logique de séparation, or les démonstrateurs automatiques ont surtout fait des progrès dans la logique du premier ordre qui ne les contient pas.Ce travail de thèse propose des techniques pour que les idées de la logique de séparation puissent apparaître dans les spécifications tout en conservant la possibilité d'utiliser des démonstrateurs pour la logique du premier ordre. Cependant les conjectures que l'ont produit ne sont pas dans la même logique du premier ordre que celles des démonstrateurs. Pour permettre une plus grande automatisation, ce travail de thèse a également défini de nouvelles conversions entre la logique polymorphe du premier ordre et la logique multi-sortée dupremier ordre utilisé par la plupart des démonstrateurs.La première partie a donné lieu à une implémentation dans l'outil Jessie, la seconde a donné lieu à une participation conséquente à l'écriture de l'outil Why3 et particulièrement dans l'architecture et écriture des transformations qui implémentent ces simplifications et conversions. / This thesis comes within the domain of proofs of programs by deductive verification. The deductive verification generates from a program source and its specification a mathematical formula whose validity proves that the program follows its specification. The program source describes what the program does and its specification represents what the program should do. The validity of the formula is mainly verified by automatic provers. During the last ten years separation logic has shown to be an elegant way to deal with programs which use data-structures with pointers. However it requires a specific logical language, provers, and specific reasoning techniques.This thesis introduces a technique to express ideas from separation logic in the traditional framework of deductive verification. Unfortunately the mathematical formulas produced are not in the same first-order logic than the ones of provers. Thus this work defines new conversions between the polymorphic first-order logic and the many-sorted logic used by most proves.The first part of this thesis leads to an implementation in the Jessietool. The second part results in an important participation to the writing of the Why3 tool, in particular in the architecture and writing of the transformations which implement these conversions.

L'analyse statique de programme a pour but de prouver automatiquement qu'un programme vérifie certaines propriétés. L'interprétation abstraite est un cadre théorique permettant de calculer des invariants de programme. Ces invariants sont des propriétés sur les variables du programme vraies pour toute exécution. La précision des invariants calculés dépend de nombreux paramètres, en particulier du domaine abstrait et de l'ordre d'itération utilisés pendant le calcul d'invariants. Dans cette thèse, nous proposons plusieurs extensions de cette méthode qui améliorent la précision de l'analyse.Habituellement, l'interprétation abstraite consiste en un calcul de point fixe d'un opérateur obtenu après convergence d'une séquence ascendante, utilisant un opérateur appelé élargissement. Le point fixe obtenu est alors un invariant. Il est ensuite possible d'améliorer cet invariant via une séquence descendante sans élargissement. Nous proposons une méthode pour améliorer un point fixe après la séquence descendante, en recommençant une nouvelle séquence depuis une valeur initiale choisie judiscieusement. L'interprétation abstraite peut égalementêtre rendue plus précise en distinguant tous les chemins d'exécution du programme, au prix d'une explosion exponentielle de la complexité. Le problème de satisfiabilité modulo théorie (SMT), dont les techniques de résolution ont été grandement améliorée cette décennie, permettent de représenter ces ensembles de chemins implicitement. Nous proposons d'utiliser cette représentation implicite à base de SMT et de les appliquer à des ordres d'itération de l'état de l'art pour obtenir des analyses plus précises.Nous proposons ensuite de coupler SMT et interprétation abstraite au sein de nouveaux algorithmes appelés Modular Path Focusing et Property-Guided Path Focusing, qui calculent des résumés de boucles et de fonctions de façon modulaire, guidés par des traces d'erreur. Notre technique a différents usages: elle permet de montrer qu'un état d'erreur est inatteignable, mais également d'inférer des préconditions aux boucles et aux fonctions.Nous appliquons nos méthodes d'analyse statique à l'estimation du temps d'exécution pire cas (WCET). Dans un premier temps, nous présentons la façon d'exprimer ce problème via optimisation modulo théorie, et pourquoi un encodage naturel du problème en SMT génère des formules trop difficiles pour l'ensemble des solveurs actuels. Nous proposons un moyen simple et efficace de réduire considérablement le temps de calcul des solveurs SMT en ajoutant aux formules certaines propriétés impliquées obtenues par analyse statique. Enfin, nous présentons l'implémentation de Pagai, un nouvel analyseur statique pour LLVM, qui calcule des invariants numériques grâce aux différentes méthodes décrites dans cette thèse. Nous avons comparé les différentes techniques implémentées sur des programmes open-source et des benchmarks utilisés par la communauté. / Static program analysis aims at automatically determining whether a program satisfies some particular properties. For this purpose, abstract interpretation is a framework that enables the computation of invariants, i.e. properties on the variables that always hold for any program execution. The precision of these invariants depends on many parameters, in particular the abstract domain, and the iteration strategy for computing these invariants. In this thesis, we propose several improvements on the abstract interpretation framework that enhance the overall precision of the analysis.Usually, abstract interpretation consists in computing an ascending sequence with widening, which converges towards a fixpoint which is a program invariant; then computing a descending sequence of correct solutions without widening. We describe and experiment with a method to improve a fixpoint after its computation, by starting again a new ascending/descending sequence with a smarter starting value. Abstract interpretation can also be made more precise by distinguishing paths inside loops, at the expense of possibly exponential complexity. Satisfiability modulo theories (SMT), whose efficiency has been considerably improved in the last decade, allows sparse representations of paths and sets of paths. We propose to combine this SMT representation of paths with various state-of-the-art iteration strategies to further improve the overall precision of the analysis.We propose a second coupling between abstract interpretation and SMT in a program verification framework called Modular Path Focusing, that computes function and loop summaries by abstract interpretation in a modular fashion, guided by error paths obtained with SMT. Our framework can be used for various purposes: it can prove the unreachability of certain error program states, but can also synthesize function/loop preconditions for which these error states are unreachable.We then describe an application of static analysis and SMT to the estimation of program worst-case execution time (WCET). We first present how to express WCET as an optimization modulo theory problem, and show that natural encodings into SMT yield formulas intractable for all current production-grade solvers. We propose an efficient way to considerably reduce the computation time of the SMT-solvers by conjoining to the formulas well chosen summaries of program portions obtained by static analysis.We finally describe the design and the implementation of Pagai,a new static analyzer working over the LLVM compiler infrastructure,which computes numerical inductive invariants using the various techniques described in this thesis.Because of the non-monotonicity of the results of abstract interpretation with widening operators, it is difficult to conclude that some abstraction is more precise than another based on theoretical local precision results. We thus conducted extensive comparisons between our new techniques and previous ones, on a variety of open-source packages and benchmarks used in the community.

[pt] Atualmente, a maioria dos programas para computadores é obtida através da WEB. Como muitas vezes a procedência são fontes desconhecidas, é preciso se certificar de que o código se comporta como o esperado. A solução ideal seria verificar o código contra uma especificação de políticas de segurança ,contudo, isso pode consumir muito tempo.Uma outra alternativa é fazer com que o próprio código prove ser seguro. O conceito de proof-carryng code (PCC)é baseado nessa idéia : um programa carrega consigo uma prova de sua conformidade com certas políticas de segurança. Ou seja ,ele carrega uma prova a respeito de propriedades do próprio código. Portanto, os mesmos métodos froamsi usados para a verificação de programs podem se utilizados para esta tecnolgia. Considerando este fato,neste trabalho é estudado como cálculo de Hoare, em método formal para realizar a verificação de programas, aplicado a códigos-fonte escritos em uma linguagem de programação imperativa, pode ser útil á tecnica de PCC. Conseqüentemente, são pesquisados métodos para a geração de provas de correção de programas utilizando o método citado, para tornar possível a geração de provas de segurança para PCC utilizando o cálculo de Hoare. / [en] Nowdays most computer programs are obtained from the WEB. Since their source is usually unknown, it is necessary to be sure that the code of the program behaves as expected.The ideal solution would be verify the code against a specification of safety policies.However, this can take too much time.Another approach is making the code itself prove that it is safe. The concept os proof-carryng code (PCC) is based on this idea: a program carries a proof of its conformity with certain safety policies. That is , it carries a proof cencerning properties related to the code itself. Therefore, the same formal methods employed in formal verification of programs can be used in this tecnology. Due to this fact, in this work it is studied how Hoare logic applied to source codes written in an imperative programming language, which is a formal methods are researched to generate proofs of program correctness using the method explained, so that it can be possible to generate PCC safety programs with Hoare logic.

<p>Computer programs keep finding their ways into new safety-critical applications, while at the same time growing more complex. This calls for new and better methods to verify the correctness of software. We focus on one approach to verifying systems, namely that of <i>model checking</i>. At first, we investigate two categories of problems related to model checking: <i>games</i> and <i>stochastic infinite-state systems</i>. In the end, we join these two lines of research, by studying <i>stochastic infinite-state games</i>.</p><p>Game theory has been used in verification for a long time. We focus on finite-state 2-player parity and limit-average (mean payoff) games. These problems have applications in model checking for the <i>μ</i>-calculus, one of the most expressive logics for programs. We give a simplified proof of memoryless determinacy. The proof applies <i>both</i> to parity and limit-average games. Moreover, we suggest a strategy improvement algorithm for limit-average games. The algorithm is discrete and strongly subexponential.</p><p>We also consider probabilistic infinite-state systems (Markov chains) induced by three types of models. <i>Lossy channel systems (LCS)</i> have been used to model processes that communicate over an unreliable medium. <i>Petri nets</i> model systems with unboundedly many parallel processes. <i>Noisy Turing machines</i> can model computers where the memory may be corrupted in a stochastic manner. We introduce the notion of <i>eagerness</i> and prove that all these systems are eager. We give a scheme to approximate the value of a reward function defined on paths. Eagerness allows us to prove that the scheme terminates. For probabilistic LCS, we also give an algorithm that approximates the limit-average reward. This quantity describes the long-run behavior of the system.</p><p>Finally, we investigate Büchi games on probabilistic LCS. Such games can be used to model a malicious cracker trying to break a network protocol. We give an algorithm to solve these games.</p>

Computer programs keep finding their ways into new safety-critical applications, while at the same time growing more complex. This calls for new and better methods to verify the correctness of software. We focus on one approach to verifying systems, namely that of model checking. At first, we investigate two categories of problems related to model checking: games and stochastic infinite-state systems. In the end, we join these two lines of research, by studying stochastic infinite-state games. Game theory has been used in verification for a long time. We focus on finite-state 2-player parity and limit-average (mean payoff) games. These problems have applications in model checking for the μ-calculus, one of the most expressive logics for programs. We give a simplified proof of memoryless determinacy. The proof applies both to parity and limit-average games. Moreover, we suggest a strategy improvement algorithm for limit-average games. The algorithm is discrete and strongly subexponential. We also consider probabilistic infinite-state systems (Markov chains) induced by three types of models. Lossy channel systems (LCS) have been used to model processes that communicate over an unreliable medium. Petri nets model systems with unboundedly many parallel processes. Noisy Turing machines can model computers where the memory may be corrupted in a stochastic manner. We introduce the notion of eagerness and prove that all these systems are eager. We give a scheme to approximate the value of a reward function defined on paths. Eagerness allows us to prove that the scheme terminates. For probabilistic LCS, we also give an algorithm that approximates the limit-average reward. This quantity describes the long-run behavior of the system. Finally, we investigate Büchi games on probabilistic LCS. Such games can be used to model a malicious cracker trying to break a network protocol. We give an algorithm to solve these games.

No description available.