LLVMVF: A Generic Approach for Verification of Multicore Software

Proliferation of multicore hardware boosted the need for verification of multicore software that is running on these hardware. Multicore software demands new verification techniques different from the ones used for sequential software. Many optimized compiler frameworks are arising to address the complexities of multicore software. Among these compilers, Low Level Virtual Machine (LLVM) is especially gaining popularity because i) has a universal front-end that allows to read in many different input languages, ii) aggressive optimizations to improve code performance and quality, and iii) a well-defined intermediate bytecode representation, called LLVM IR, that allows a unified intermediate representation. In this work, we present a novel framework, called LLVM Verification Framework (LLVMVF), implemented in a purely functional language for verification of multicore software. To our knowledge, this is the first verification framework using the LLVM bytecode representation for multicore software. We present an SMT-based Bounded Model Checker backend of LLVMVF and perform initial experiments on multicore software using Pthreads library. Furthermore, we compare our results with an existing multicore software verification tool.     

基于国产自主CPU的SEP0611平台中Dalvik虚拟机的移植和优化

Dalvik是Google公司自己设计用于Android平台的Java虚拟机,Andriod系统不支持所有构架的特性使将andriod系统移植到基于国产内核Unity的Soc的SoC SEP0611上有着大量的工作.对Dalvik虚拟机平台相关性和无关性进行分析,使对虚拟机的移植有了基本认识和实现基础,同时介绍了我们自主研发的基于国产自主内核的SoC,将与基于国产内核Unity架构平台相关的JNI(Java Native Interface)机制进行分析,找到了移植的关键环节本地调用桥(Callbridge),同时对Dalvik虚拟机优化的进行了分析与研究,找到了解释器优化的最好途径,即用汇编重写解释器,并初步探讨了JIT( Just In Time)编译器的实现方法.通过实现本地调用桥,实现了Dalvik虚拟机在SEP0611国产自主SoC上的移植.     

Hardware Acceleration of Red-Black Tree Management and Application to Just-In-Time Compilation

Due to the everlasting consumer demand for more complex applications, embedded systems have evolved both in terms of complexity and heterogeneity. The architecture of such systems often includes several kinds of different computing resources (DSPs, GPUs, etc.). As a consequence, software designers are facing significant performance and portability issues to target these devices. Software relies more and more on virtualization technologies to maximize portability of applications. In order to balance portability and performance, most virtualization technologies leverage Just-in-time (JIT) compilation to provide runtime optimized code from portable one. Nevertheless, the efficiency of JIT compilation depends on the ability to compensate its overhead with execution speedups of generated code. While most research efforts focus on limiting overhead of JIT compilation phases by reducing their occurrences, this paper investigates opportunities of speeding up JIT compilation itself. We first present a performance analysis of different JIT compilation technologies in order to identify hardware and software optimization opportunities. Second, we propose a solution based on a dedicated processor with specialized instructions for critical functions of JIT compilers. These specialized instructions provide an average 5× speedup on manipulations of associative arrays and dynamic memory allocation. Based on the LLVM framework, we show a 15% overall speedup on code generator’s execution time. Because our specialized instructions are hidden behind standard libraries, we also argue that these instructions may be transparently reused for a wider range of applications.     

基于程序分析的Android应用恶意行为检测

随着Android智能手机的普及,手机应用的安全隐患也日益凸显。提出了一种基于程序分析的Android应用程序检测方法,用于检测Android应用程序中的恶意行为。通过预处理剔除不存在恶意行为的应用程序。对通过预处理阶段的应用程序,模拟执行应用程序中的字节码指令,构建出函数的摘要信息,最终在构建的函数摘要上使污点传播算法,检测应用程序中的恶意行为。实验结果表明,该方法可以有效检测出Android应用程序的恶意行为,具有较高的实用性。     

Markov decision process (MDP) framework for software power optimization using call profiles on mobile phones

We present an optimization framework for delay-tolerant data applications on mobile phones based on the Markov decision process (MDP). This process maximizes an application specific reward or utility metric, specified by the user, while still meeting a talk-time constraint, under limited resources such as battery life. This approach is novel for two reasons. First, it is user profile driven, which means that the user’s history is an input to help predict and reserve resources for future talk-time. It is also dynamic: an application will adapt its behavior to current phone conditions such as battery level or time before the next recharge period. We propose efficient techniques to solve the optimization problem based on dynamic programming and illustrate how it can be used to optimize realistic applications. We also present a heuristic based on the MDP framework that performs well and is highly scalable for multiple applications. This approach is demonstrated using two applications: Email and Twitter synchronization with different priorities. We present experimental results based on Google’s Android platform running on an Android Develepor Phone 1 (HTC Dream) mobile phone.     

Design and analysis of simple irregular LDPC codes with finite‐lengths and their application in CDMA systems

In this paper, we design and optimize simple irregular low‐density parity‐check (LDPC) codes for finite‐length applications where the asymptotic noise threshold of the channel cannot play as a dominant optimization factor. Our design procedure is based on some observations resulted from analytical study of these codes. Although we present our design procedure for some specified rates but it can generally be used for any rate. Specifically, we design a simple irregular LDPC code for IS‐95 and compare its performance with the other reported codes 1 - 3 for this application. Our results show a 3.7‐fold increase in the capacity at bit error rate (BER) equal to 10⁻⁵ compared to the low‐rate orthogonal convolutional codes and 1.2 times increase compared to a high performance LDPC code of Reference 3 . Copyright © 2004 John Wiley & Sons, Ltd.     

Automatic data aggregation for recursively modeled NFV services

Network function virtualization (NFV) allows to model network services as graphs interconnecting virtual network functions (VNFs), which may include nested VNFs, modeled as subgraphs of VNFs or end points. To query the performance data of network services modeled as abstract and high‐level graphs is challenging because of the recursivity of the NFV architecture and the elasticity and dynamicity provided by the NFV infrastructure. We propose to use Datalog, a declarative logic programming language, to build a framework that supports efficient data aggregation for performance metrics of recursively modeled network services. We present our recursive query language for automatic and flexible decomposition and aggregation of NFV performance metrics and describe example use cases for both compute and network metrics. We also describe the design and implementation of a proof‐of‐concept query engine using the language. Our performance evaluation shows that the total query latency is dominated by retrieval times of performance metrics from infrastructure databases, and the time for execution and automatic decomposition of high‐level queries by the query engine itself increases linearly with the size of the service graph up to 1000 nodes. Hence, our evaluations show that the query engine scales well, bounded mainly by the limited execution capabilities of our test environment. Furthermore, it can handle multiple concurrent queries up to the concurrency limits of the backend database in use. The proposed query language and engine are thus effective in recursively retrieving performance metrics of NFV environments supporting large‐scale service graphs and large numbers of query requests.     

AndroScope: An Insightful Performance Analyzer for All Software Layers of the Android‐Based Systems

Android has become the most popular platform for mobile devices. However, Android still has critical performance issues, such as “application not responding” errors and hiccups resulting from garbage collection. Many phone vendors have tried to resolve the problems by characterizing and improving the performance. However, there are few insightful performance analysis tools for the Android‐based systems. This paper presents AndroScope, which is a performance analysis tool for both the Android platform (Dalvik virtual machine, core libraries, Android libraries, and even Linux kernels) and its applications. To the best of our knowledge, this is the first tool to collect and analyze performance data from all the software layers of the Android‐based systems. AndroScope offers a trace mechanism to collect such deep and wide performance data as hardware performance counters, time, and memory usage. In addition, the tool includes TraceBridge, which is a middleware for the fast handling of mass logs. Moreover, AndroScope offers an integrated graphical user interface with the Android software development kit to display a great volume of the detailed performance data.     

基于 Android 的 Java MIDP应用方案研究

随着数字电视的发展,在Android平台上兼容原有的中间件业务的需求越来越迫切,而且在基于浏览器兼容的业务方面已有成熟的方案,所以除了浏览器业务之外,广电还有一部分具有高质量、高附加值的J2ME（CLDC）的应用迫切与平台兼容,但基于J2ME的Java应用程序不能直接运行于Android平台之上,因此针对该问题进行研究,结合Android 平台的特点,以及Android 平台Dalvik虚拟机与J2ME虚拟机的异同和两者的优势,将JVM移植到Android系统并进行适配,通过浏览器进行调用和展现,最终给出在Android平台上兼容原有的中间件J2 ME体系的Java应用程序的解决方案。     

Phased Scheduling for Resource-Constrained Mobile Devices in Mobile Cloud Computing

Mobile cloud computing combines wireless access service and cloud computing to improve the performance of mobile applications. Mobile cloud computing can balance the application distribution between the mobile device and the cloud, in order to achieve faster interactions, battery savings and better resource utilization. To support mobile cloud computing, the paper proposes a phased scheduling model of mobile cloud such that mobile device’s users experience lower interaction times and extended battery life. The phased scheduling optimization is solved by two subproblems: mobile device’s batch application optimization and mobile device’s job level optimization. At the first stage, the mobile cloud global scheduling optimization implements the allocation of the cloud resources to the mobile device’s batch applications. At the second stage, mobile device’s job level optimization adjusts the cloud resource usages to optimize the utility of single mobile device’s application. In the simulations, compared with other algorithm, our proposed mobile cloud phased scheduling algorithms achieve the better performance with acceptable overhead.     

一种有效保护Java字节码的方法

JavaTM是开发Web网络应用的主要编程语言之一。由于Java语言自身的特点,它的字节码文件很容易被反编译,所以它的安全问题越来越受到重视。结合信息隐藏与加密技术,提出了一种保护Java字节码的新方法。该方法利用Shamir门限分存方案,将加密密钥打散,隐藏在加密的Java字节码文件中。能有效保护Java字节码文件抵抗反编译,并且很好地支持Java应用程序的升级,通过大量测试数据证明,该方法没有对原Java程序的运行效率产生很大影响。     

基于Android的数据访问技术研究

数据的存储与访问是Android应用程序设计过程中需要解决的基本问题,随着应用需求的不断增加,采用合理的数据访问技术将有效提高项目的开发效率、优化系统性能.文中通过深入阐述Android系统的数据存储访问机制,提出了文件读写、数据库存取和数据共享等3种数据访问技术的基本实现方法.最后结合各个方法的优缺点分析它们的应用场景.     

DJFS: Providing Highly Reliable and High‐Performance File System with Small‐Sized NVRAM

File systems and applications try to implement their own update protocols to guarantee data consistency, which is one of the most crucial aspects of computing systems. However, we found that the storage devices are substantially under‐utilized when preserving data consistency because they generate massive storage write traffic with many disk cache flush operations and force‐unit‐access (FUA) commands. In this paper, we present DJFS (Delta‐Journaling File System) that provides both a high level of performance and data consistency for different applications. We made three technical contributions to achieve our goal. First, to remove all storage accesses with disk cache flush operations and FUA commands, DJFS uses small‐sized NVRAM for a file system journal. Second, to reduce the access latency and space requirements of NVRAM, DJFS attempts to journal compress the differences in the modified blocks. Finally, to relieve explicit checkpointing overhead, DJFS aggressively reflects the checkpoint transactions to file system area in the unit of the specified region. Our evaluation on TPC‐C SQLite benchmark shows that, using our novel optimization schemes, DJFS outperforms Ext4 by up to 64.2 times with only 128 MB of NVRAM.     

Performance evaluation of throughput in optical burst switching

Optical burst switching (OBS) provides a promising solution to utilize the huge terahertz bandwidth of optical wavelength division multiplexing (WDM) transmission technology. To exploit this bandwidth, several reservation schemes have been proposed that include just‐in‐time (JIT) signaling, just‐enough‐time (JET) signaling and burst segmentation (BS). It is necessary to investigate the performance of these schemes under the same constraints for a prescribed OBS application. Accordingly, in this paper, we analyzed and compared the performance of JIT, JET and BS techniques under various scenarios such as network size, delay variation and load variation in an OBS network using various performance metrics, such as the offset time and switch configuration time. Also, the performance of the network under various switching delays was also investigated. The modified BS reservation scheme has been found to yield significantly better performance and better throughput compared with the JIT and JET reservation schemes. Test results show that the effect of varying loads as well as delays significantly impacts the performance of the OBS network. The results presented in this paper are expected to lead further performance improvements in OBS networks using the BS reservation scheme. Copyright © 2010 John Wiley & Sons, Ltd.     

Cross-stratum resource reservation (CSRR) algorithm for deadline-driven applications in datacenter networks

Nowadays, datacenters interconnected with optical networks have become the fundamental infrastructure to accommodate high-performance datacenter applications. Lacking of layer interaction between datacenters and networks during service provisioning, many end user applications cannot efficiently utilize the network capabilities, nor can they achieve the desired quality of service objectives. In response to these challenges, cross-stratum optimization has been studied to optimize the computing and network resources utilization from a united view. Meanwhile, the type of network applications becomes diverse. Compared to traditional immediate reservation (IR) request, a new type of request called advance reservation (AR) has recently been gaining attention for optical networks. IR’s start time of data transmission is assumed to be immediate, while AR request typically specifies the earliest time or the deadline of data transmission. According to the time features of AR request, AR applications should be scheduled in time dimension. Thus, it is both important and challenging to reserve computing and network resources in efficient manners. In this study, a cross-stratum resource model considering time dimension is set up, and a cross-stratum resource reservation (CSRR) algorithm is proposed to schedule AR applications and to reserve cross-stratum resource. Simulation results show that CSRR can reduce the failure rates of AR applications and improve the resource consumption ratio significantly.     

Multi‐granular,multi‐purpose and multi‐Gb/s monitoring on off‐the‐shelf systems

As an attempt to make network managers’ life easier, we present M³Omon , a system architecture that helps to develop monitoring applications and perform network diagnosis. M³Omon behaves as an intermediate layer between the traffic and monitoring applications that provides advanced features, high performance and low cost. Such advanced features leverage a multi‐granular and multi‐purpose approach to the monitoring problem. Multi‐granular monitoring provides answers to tasks that use traffic aggregates to identify an event, and requires either flow records or packet data or even both to understand it and, eventually, take convenient countermeasures. M³Omon provides a simple API to access traffic simultaneously at several different granularities, i.e. packet‐level, flow‐level and aggregate statistics. The multi‐purposed design of M³Omon allows not only performing tasks in parallel that are specifically targeted to different traffic‐related purposes (e.g. traffic classification and intrusion detection) but also sharing granularities between applications, e.g. several concurrent applications fed from flow records that are provided by M³Omon . Finally, the low‐cost characteristic is brought by off‐the‐shelf systems (the combination of open‐source software and commodity hardware) and the high performance is achieved thanks to modifications in the standard NIC driver, low‐level hardware interaction, efficient memory management and programming optimization. Copyright © 2014 John Wiley & Sons, Ltd.     

基于Android的流媒体播放器的设计与实现

针对Android自带多媒体库解码效率相对较低、支持视频文件格式少的问题,基于目前主流的Android系统及开源跨平台的音视频流方案FFmpeg,提出一种流媒体播放器的设计方案,并给出了方案实现步骤。通过在Linux环境下对集录制、转换音/视频编码功能为一体的完整的开源解决方案FFmpeg进行裁剪和优化,编译并移植到Android平台上,实现了远程视频监控和多种格式本地视频文件的播放。实验结果表明,Android流媒体播放器播放效果良好,可嵌入到其它Android应用软件中,应用前景广泛。