安卓系統上多重細度且具回溯功能之垂直分析工具

Abstract

Android作業系統近年來在行動裝置上越來越受歡迎。在Google play商店裡也有超過百萬個支援的應用程式。目前每一代軟硬體推陳出新的間隔越來越短。這造成了許多相容性以及最佳化的問題。然而現今常見的效能分析工具卻不能跨越使用者空間、核心空間，以及Dalvik虛擬機器三者間的障礙。這意味著開發者無法找出在不同系統層級間的瓶頸或是呼叫關係。本論文在Oprofile的基礎上研製了三個基本模組，稱為RGB module來解決這些問題。並且可以提供多重細度且具回溯功能的垂直分析能力。

Android is becoming more and more popular on mobile devices in recent years. There are over a million various apps supported in the Google play store. Moreover, intervals between hardware and software generations are becoming shorter. This makes compatibility and optimization issues ever more challenging. However, existing profiling tools cannot across the information barriers between user space, kernel space and Dalvik virtual machine. That means developers cannot find the bottleneck or recreate call-graph between different system layers. This thesis proposes a set of three basic components called RGB module based on Oprofile in order to solve these issues and provide multi-granular vertical profiling with backtrace abilities.