BEAMJIT is a tracing just-in-time compiling runtime for the Erlang programming language. The core parts of BEAMJIT are synthesized from the C source code of BEAM, the reference Erlang abstract machine. The source code for BEAM's instructions is extracted automatically from BEAM's emulator loop. A tracing version of the abstract machine, as well as a code generator are synthesized. BEAMJIT uses the LLVM toolkit for optimization and native code emission. The automatic synthesis process greatly reduces the amount of manual work required to maintain a just-in-time compiler as it automatically tracks the BEAM system. The performance is evaluated with HiPE's, the Erlang ahead-of-time native compiler, benchmark suite. For most benchmarks BEAMJIT delivers a performance improvement compared to BEAM, although in some cases, with known causes, it fails to deliver a performance boost. BEAMJIT does not yet match the performance of HiPE mainly because it does not yet implement Erlang specific optimizations such as boxing/unboxing elimination and a deep understanding of BIFs. Despite this BEAMJIT, for some benchmarks, reduces the runtime with up to 40\%.

Compiler back-ends generate assembly code by solving three main tasks: instruction selection, register allocation and instruction scheduling. We introduce constraint models and solving techniques for these code generation tasks and describe how the models can be composed to generate code in unison. The use of constraint programming, a technique to model and solve combinatorial problems, makes code generation simple, flexible, robust and potentially optimal.

This paper introduces a constraint model and solving techniques for code generation in a compiler back-end. It contributes a new model for global register allocation that combines several advanced aspects: multiple register banks (subsuming spilling to memory), coalescing, and packing. The model is extended to include instruction scheduling and bundling. The paper introduces a decomposition scheme exploiting the underlying program structure and exhibiting robust behavior for functions with thousands of instructions. Evaluation shows that code quality is on par with LLVM, a state-of-the-art compiler infrastructure. The paper makes important contributions to the applicability of constraint programming as well as compiler construction: essential concepts are unified in a high-level model that can be solved by readily available modern solvers. This is a significant step towards basing code generation entirely on a high-level model and by this facilitates the construction of correct, simple, flexible, robust, and high-quality code generators.

This paper introduces a constraint model and solving techniques for code generation in a compiler back-end. It contributes a new model for global register allocation that combines several advanced aspects: multiple register banks (subsuming spilling to memory), coalescing, and packing. The model is extended to include instruction scheduling and bundling. The paper introduces a decomposition scheme exploiting the underlying program structure and exhibiting robust behavior for functions with thousands of instructions. Evaluation shows that code quality is on par with LLVM, a state-of-the-art compiler infrastructure. The paper makes important contributions to the applicability of constraint programming as well as compiler construction: essential concepts are unified in a high-level model that can be solved by readily available modern solvers. This is a significant step towards basing code generation entirely on a high-level model and by this facilitates the construction of correct, simple, flexible, robust, and high-quality code generators.

This report presents an algorithm for efficiently simulating view synchrony, including failure-atomic total-order multicast in a discrete-time event simulator. In this report we show how a view synchrony implementation tailored to a simulated environment removes the need for third party middleware and detailed network simulation, thus reducing the complexity of a test environment. An additional advantage is that simulated view synchrony can generate all timing behaviours allowed by the model instead of just those exhibited by a particular view synchrony implementation.

This thesis presents the design, implementation, and evaluation of Flow Java, a programming language for the implementation of concurrent programs. Flow Java adds powerful programming abstractions for automatic synchronization of concurrent programs to Java. The abstractions added are single assignment variables (logic variables) and futures (read-only views of logic variables). The added abstractions conservatively extend Java with respect to types, parameter passing, and concurrency. Futures support secure concurrent abstractions and are essential for seamless integration of single assignment variables into Java. These abstractions allow for simple and concise implementation of high-level concurrent programming abstractions. Flow Java is implemented as a moderate extension to the GNU gcj/libjava Java compiler and runtime environment. The extension is not specific to a particular implementation, it could easily be incorporated into other Java implementations. The thesis presents three implementation strategies for single assignment variables. One strategy uses forwarding and dereferencing while the two others are variants of Taylor's scheme. Taylor's scheme represents logic variables as a circular list. The thesis presents a new adaptation of Taylor's scheme to a concurrent language using operating system threads. The Flow Java system is evaluated using standard Java benchmarks. Evaluation shows that in most cases the overhead incurred by the extensions is between 10% and 50%. For some pathological cases the runtime increases by up to 150%. Concurrent programs making use of Flow Java's automatic synchronization, generally perform as good as corresponding Java programs. In some cases Flow Java programs outperform Java programs by as much as 33%.

Logic variables pioneered by (concurrent) logic and concurrent constraint programming are powerful mechanisms for automatically synchronizing concurrent computations. They support a declarative model of concurrency that avoids explicitly suspending and resuming computations. This paper presents Flow Java which conservatively extends Java with single assignment variables and futures as variants of logic variables. The extension is conservative with respect to object-orientation, types, parameter passing, and concurrency in Java. Futures support secure concurrent abstractions and are essential for seamless integration of single assignment variables into Java. We show how Flow Java supports the construction of simple and concise concurrent programming abstractions. We present how to moderately extend compilation and the runtime architecture of an existing Java implementation for Flow Java. Evaluation using standard Java benchmarks shows that in most cases the overhead is between 10% and 40%. For some pathological cases the runtime increases by up to 75%.