The breakdown of Dennard's transistor scaling has driven computing systems toward application-specific accelerators, which can provide orders-of-magnitude improvements in performance and energy efficiency over general-purpose processors. To enable the radical departures from conventional approaches that heterogeneous systems entail, research infrastructure must be able to model processors, memory and accelerators, as well as system-level changes---such as operating system or instruction set architecture (ISA) innovations---that might be needed to realize the accelerators' potential. Unfortunately, existing simulation tools that can support such system-level research are limited by the lack of fast, scalable machine emulators to drive execution. To fill this need, in this dissertation we first present a novel machine emulator design based on dynamic binary translation that makes the following improvements over the state of the art: it scales on multicore hosts while remaining memory efficient, correctly handles cross-ISA differences in atomic instruction semantics, leverages the host floating point (FP) unit to speed up FP emulation without sacrificing correctness, and can be efficiently instrumented to---among other possible uses---drive the execution of a full-system, cross-ISA simulator with support for accelerators. We then demonstrate the utility of machine emulation for studying heterogeneous systems by leveraging it to make two additional contributions. First, we quantify the trade-offs in different coupling models for on-chip accelerators. Second, we present a technique to reuse the private memories of on-chip accelerators when they are otherwise inactive to expand the system's last-level cache, thereby reducing the opportunity cost of the accelerators' integration.

Typescript (photocopy). / Digitized by Kansas Correctional Industries

Hardware and software do not operate in isolation. Neither should they be regarded as such when securing systems. To seamlessly facilitate computing, they have to communicate via interfaces. Besides characterizing the means by which software can harness the exposed functionalities of hardware, these hardware-software interfaces define the degree and granularity of control and access that software possesses to the lower layers of the system stack. These mechanisms provide a rich source of hardware assistive technologies that can be tapped to enhance security as a full-system property. On the flip side, given the level of access software has to these hardware features, security-oblivious designs of hardware and their interfaces can expose systems to new vulnerabilities. Evidently, these hardware-software interfaces represent a crucial focal area in systems for the formulation, review and refinement of security measures. This dissertation advances the thesis that security as a full-system property can be improved by examining and leveraging the interworking of hardware and software. It advocates a full-system approach in architecture design by demonstrating how unanticipated ways in which hardware and software co-operate can induce unintended computing behavior and pose security risks. It develops novel techniques to repurpose commodity hardware support to create new defense primitives that exploit the synergy between hardware and software. It shows how commodity hardware-software interfaces play an instrumental role in security with the hardware's well-positioned access to runtime information. All these interface-oriented design principles, as this dissertation demonstrates, are widely applicable and practical as the highlighted three case studies span the three primary stages of a typical security attack, namely the act of inducing unintended system behavior, exploiting vulnerability to achieve initial system control, and executing malicious code for nefarious goals. First, the dissertation begins by scrutinizing the design of energy management mechanisms, a prevalent class of hardware-software interfaces found in almost all commodity systems. It shows, for the first time, that as we pursue increasingly aggressive cooperative hardware-software mechanisms to improve energy efficiency, doing so with no regard for security can create serious vulnerabilities. This dissertation highlights a multitude of issues in the current designs of energy management mechanisms. It further demonstrates how, with fine-grained software-based control of underlying hardware voltage and frequency regulators, attackers can exploit these issues to induce unintended computing behavior. It shows that beyond causing unintended system behavior, abusing these interfaces in security-oblivious energy management designs can violate all three key security properties in spite of hardware-enforced isolation: confidentiality (extracting AES keys), integrity (loading self-signed code), and clearly, availability (freezing the device). Second, the dissertation addresses an advanced class of dynamic code reuse exploits that rely on memory disclosure vulnerabilities to construct their initial payload code at runtime. This class of exploits bypasses even the finest-grained randomization-based defenses. While the concept of execute-only memory in existing defenses works well, it cannot be applied effectively in closed-source systems where perfect disassembly of compiled binaries is not possible. To tackle this problem, this dissertation first introduces the Destructive Code Read primitive---a defense technique that randomizes executable memory as it is being read as data---as a means to thwart memory disclosure exploits as well as to sidestep the problem of imperfect binary disassembly in COTS systems. It leverages the virtualization assistive hardware feature to timely mediate read operations into executable memory, thereby significantly lowering the cost of deploying the Destructive Code Read defense primitive. Tapping into the unique strengths of functionality closer to the hardware layer of the system stack, it extends the benefits of execute-only memory defenses to COTS systems. Finally, the dissertation builds on the insight that hardware, being the lowest part of the system stack, is uniquely positioned to augment traditionally software-only techniques. Besides being more performant and energy-efficient, hardware offers extensive visibility into code execution at the software layers. This dissertation shows that these hardware characteristics offer unprecedented insights into code execution, both benign and malicious. It demonstrates that the interaction of hardware and software can be modeled as microarchitectural events, which can in turn be leveraged to detect anomalous malicious code execution in the latter stages of a security attack. Using assistive debugging hardware features to efficiently audit these events, it further develops novel techniques to make sense of the noisy and lower-level microarchitectural events to detect in-flight shellcode execution and full-fledged anomalous malicious programs.

Accurately predicting program behaviors (e.g., memory locality, method calling frequency) is fundamental for program optimizations and runtime adaptations. Despite decades of remarkable progress, prior studies have not systematically exploited the use of program inputs, a deciding factor of program behaviors, to help in program dynamic optimizations. Triggered by the strong and predictive correlations between program inputs and program behaviors that recent studies have uncovered, the dissertation work aims to bring program inputs into the focus of program behavior analysis and program dynamic optimization, cultivating a new paradigm named input-centric program behavior analysis and dynamic optimization.;The new optimization paradigm consists of three components, forming a three-layer pyramid. at the base is program input characterization, a component for resolving the complexity in program raw inputs and extracting important features. In the middle is input-behavior modeling, a component for recognizing and modeling the correlations between characterized input features and program behaviors. These two components constitute input-centric program behavior analysis, which (ideally) is able to predict the large-scope behaviors of a program's execution as soon as the execution starts. The top layer is input-centric adaptation, which capitalizes on the novel opportunities created by the first two components to facilitate proactive adaptation for program optimizations.;This dissertation aims to develop this paradigm in two stages. In the first stage, we concentrate on exploring the implications of program inputs for program behaviors and dynamic optimization. We construct the basic input-centric optimization framework based on of line training to realize the basic functionalities of the three major components of the paradigm. For the second stage, we focus on making the paradigm practical by addressing multi-facet issues in handling input complexities, transparent training data collection, predictive model evolvement across production runs. The techniques proposed in this stage together cultivate a lifelong continuous optimization scheme with cross-input adaptivity.;Fundamentally the new optimization paradigm provides a brand new solution for program dynamic optimization. The techniques proposed in the dissertation together resolve the adaptivity-proactivity dilemma that has been limiting the effectiveness of existing optimization techniques. its benefits are demonstrated through proactive dynamic optimizations in Jikes RVM and version selection using IBM XL C Compiler, yielding significant performance improvement on a set of Java and C/C++ programs. It may open new opportunities for a broad range of runtime optimizations and adaptations. The evaluation results on both Java and C/C++ applications demonstrate the new paradigm is promising in advancing the current state of program optimizations.

Thesis (M.S.)--University of Missouri--Rolla, 2007. / Vita. The entire thesis text is included in file. Title from title screen of thesis/dissertation PDF file (viewed April 11, 2007) Includes bibliographical references (p. 54-55).

Direct interaction in virtual environments can be realized using relatively simple hardware, such as standard webcams and monitors. The result is a large gap between the stimuli existing in real-world interactions and those provided in the virtual environment. This leads to reduced efficiency and effectiveness when performing tasks. Conceivably these missing stimuli might be supplied through a visual modality, using sensory substitution. This work suggests a display technique that attempts to usefully and non-detrimentally employ sensory substitution to display proximity, tactile, and force information. We solve three problems with existing feedback mechanisms. Attempting to add information to existing visuals, we need to balance: not occluding the existing visual output; not causing the user to look away from the existing visual output, or otherwise distracting the user; and displaying as much new information as possible. We assume the user interacts with a virtual environment consisting of a manually controlled probe and a set of surfaces. Our solution is a pseudo-shadow: a shadow-like projection of the user's probe onto the surface being explored or manipulated. Instead of drawing the probe, we only draw the pseudo-shadow, and use it as a canvas on which to add other information. Static information is displayed by varying the parameters of a procedural texture rendered in the pseudo-shadow. The probe velocity and probe-surface distance modify this texture to convey dynamic information. Much of the computation occurs on the GPU, so the pseudo-shadow renders quickly enough for real-time interaction. As a result, this work contains three contributions: a simple collision detection and handling mechanism that can generalize to distance-based force fields; a way to display content during probe-surface interaction that reduces occlusion and spatial distraction; and a way to visually convey small-scale tactile texture.

Dissertation (Ph.D.)--Auburn University, 2009. / Abstract. Includes bibliographic references (p.78-83).

Thesis (Ph. D.)--University of Texas at Austin, 2003. / Vita. Includes bibliographical references. Available also from UMI Company.

Thesis (M.S. in Modeling, Virtual Environments and Simulation)--Naval Postgraduate School, December 2003. / Thesis advisor(s): Michael Zyda, Bret Michael. Includes bibliographical references (p. 111-114). Also available online.

No description available.