Lectures

Lectures

For daily activities, please see the Schedule page.


Introduction to Verification with Spark 2014

Lecturer: Yannick Moy (AdaCore)

SPARK 2014 is the latest evolution of the SPARK programming language. Based on Ada 2012, it encompasses a rich subset of the language and augments it with further language contracts designed to support advanced static verification techniques.

This lecture will provide a broad overview of the SPARK 2014 language: the subset of Ada 2012 which it includes and the additional constructs which it adds to support verification. The morning session will cover basic features and static analysis techniques. This will be followed by an introduction to one of the most novel features of the language - the dual nature of the contracts as both executable and mathematical statements and how these can be verified by testing, proof, or a combination of the two.

In the afternoon, the tutorial will look at more advanced language features that are essential for industrial-scale specification, such as abstraction and refinement. We will also look at more advanced verification techniques supported by the language such as information flow analysis and proving the absence of run-time exceptions (divide-by-zero, numeric overflow). Finally, we will touch on some of the more advanced proof techniques that may be required.


Bakar Kiasan – A Certifying Program Verifier for High-Integrity Systems

Lecturers: Robby (Kansas State University) and John Hatcliff (Kansas State University)

Software-dependent critical systems that impact daily life are rapidly increasing in number, size, and complexity. Unfortunately, inadequate software and systems engineering can lead to accidents that cause economic disaster, injuries or death. There is a growing reliance on development and verification tools to reduce costs, better manage complexity, and to increase confidence in these systems. Recent standards for critical systems have an increased emphasis on characterizing the requirements of tools used in the certification context and the particular certification obligations that these tools can discharge. In the case of avionics, DO-333 explicitly addresses the role/use of formal methods tools, e.g., allowing formal verification of code compliance to procedure contracts to replace unit testing.

This lecture presents the next generation Bakar Kiasan verifier in the Sireum platform (sireum.org). Bakar Kiasan is a certifying symbolic execution-based program verifier for high-integrity critical systems written using the Spark 2014 safe subset of Ada 2012. It checks conformance of Spark programs against their contracts and produce evidence for its analysis results. For example, it produces counter-examples and test cases for contract violations. In addition, Bakar Kiasan produces test cases to illustrate contract satisfactions as well as machine-checkable proofs for its contract verifications. The lecture will present foundations of symbolic execution, discuss how evidence can be generated from symbolic execution-based analysis, and will provide hands-on exercises using Bakar Kiasan to verify Spark 2014 program units.


SCCE: Variability and Evolution through Integrated Domain, Data and Process Modeling

Lecturers: Bernhard Steffen (TU Dortmund), Tiziana Margaria (University of Potsdam),

                  Stefan Naujokat (TU Dortmund), and Johannes Neubauer (TU Dortmund)

Domain-specific tool support is the key towards opening development responsibility for a wider public without dedicated programming knowledge. This is, for example, important in areas like business process modelling or scientific workflows. The intention behind the tools and methods presented in this lecture lies in providing domain experts with appropriate (non-programming) tools, so that they can solve problems within their domainin in a non-technical way and even construct domain-specific software products without dedicated programming knowledge.

In the first session of this lecture we will introduce the Dynamic Web Application (DyWA) framework that supports the prototype-driven development of process-oriented web applications. It is a generic, fully functional prototype that accompanies the development right from the beginning, offering a web-based definition facility for application domains in terms of type schemata that captures the data types and their associations. The session will provide hands-on experience on domain modeling with DyWA which automatically generates the (complex) modeled data structures together with their fully executable access functionality.

The second session focuses on ‘breathing life’ into the DyWA domain models via the process modeling environment jABC4. In particular, it will illustrate the power of prototype-driven process development throughout the whole system lifecylce, with a focus on system evolution and system migration.

The third session presents Cinco, a particularly powerful application of our process modeling framework based on the Eclipse ecosystem. Cinco is a meta tooling suite for generating domain-specific graphical development tools in terms of fully operative Eclipse products. It can be regarded as a domain-specific instance of generic meta modeling frameworks such as Eclipse EMF/Graphiti which radically eases their use through its domain-specific focus. Cinco is an application of our process modeling and design framework jABC4 as all the required transformations, be they model-to-model or model-to-code, are modeled with jABC4 so that they can also be developed without specific programming knowledge. Please note that Cinco is a prime example for bootstrapping. It is a tool for delevoping domain-specific development tools that is build with a domain-specific development tool (jABC4). Indeed, the next version of jABC4, jABC5 will be constructed in this very fashion. In the hands-on tutorial part of this session the participants will use Cinco to create a full graphical DSL modeling tool of their choice (e.g., for State Charts, Petri Nets, BPMN, Timed Automata, or even architectural models or industrial plant layouts) consisting of an abstract specification describing its structure and appearance, accompanied by a code generator that is modeled in jABC4 using modeling components generated from the very same specification.


Introduction to SMT and SMT-based Model Checking

Lecturer: Cesare Tinelli (University of Iowa)

Many problems in computer science, in particular in formal methods, can be reduced to checking the satisfiability of a formula in some logic. Several of these problems can be naturally formulated as Satisfiability Modulo Theories (SMT) problems which are about checking the satisfiability of first-order formulas with respect to some logical theory T of interest. SMT differs from general automated deduction in that the background theory T need not be finitely or even first-order axiomatizable, and specialized inference methods are used for each theory. By being theory-specific and restricting their language to certain classes of formulas, these specialized methods can be implemented in solvers that are more efficient in practice than general-purpose theorem provers. SMT solvers have been used successfully in several application ares including hardware verification, equivalence checking, bounded and unbounded model checking, predicate abstraction, static analysis, automated test case generation, extended static checking, and type checking.

The first session of this lecture will start with an overview of SMT and its applications, followed by exercises using SMT solvers supporting the standard SMT-LIB input format, with a focus on the CVC4 solver. CVC4 is a widely-used open-source SMT solver supporting a rich set of logical theories including the theory of arrays, bit vectors, linear integer and real arithmetic, algebraic data types and strings.

The second session will give a brief overview of model checking techniques relying on SMT solvers, and show how to use the SMT-based Kind 2 model checker to prove invariant properties of transition systems. Kind 2 checks multiple properties simultaneously by running in parallel several model checking engines based on such techniques as bounded model checking (BMC), k-induction, IC3/PDR, and automatic invariant generation. The engines cooperate by exchanging information about the proven or disproven properties as well as auxiliary invariants. Examples and exercises will use transition systems and properties specified in the synchronous data flow language Lustre.