Automated Display Kernel Validation

Abstract

Number of portable devices had increased not in percentage but in multiples in last decade and Android is the most widely used operating system for handheld devices. Linux Kernel which is at the heart of the Android allows user to leverage the full potential of the hardware. Display is one of the most crucial components in these consumer electronic devices as it has a huge impact on user experience. Frame data seen on the display is received from processor and display drivers are needed for communication between the display and display engine. Display drivers for Android operating system running on Intel platforms are available in Linux kernel and are improved and developed everyday. Validation of display kernel is different from other components as it is a visual entity and not just a computational entity. Manual validation would take hours and hours and even then it wouldn't be as accurate due to limitations of human eye. Even if some standard test cases are developed, it wouldn't be sufficient as there can be innumerable use case scenarios for a user to use the device which can vary from high end gaming or watching 4k movies on the same device to using it for social networking. The automation framework developed here can be used to validate various display features over different types of displays. Different APIs are developed for validating various features irrespective to the display connected to the chipset. This framework is generic in the sense that as soon as a new platform comes up, few configuration files need to be added to it and framework is ready to run validation of display driver on the new chipset. Python is used to automate this validation framework. To validate a feature, same API is used in different scenarios created by python to validate that feature in various possible configurations. A set of all such test cases written in python make up a test suite that can itself be automated. With a single trigger, all test can start execution one after the other without any human intervention. The best part is target need not be at your desk. It can be any device from a pool of devices located at any point on the globe. PAGE Tool is used to identify test cases to verify a patch submitted. For this being done manually so far, validation owner had to analyze the patch and plan out test cases to verify a patch. PAGE Tool on the other hand automatically analyzes a patch and identifies the most significant test cases by providing patch id as the input. It also recommends test cases to check regression. GRAD Tool, on the contrary, grades validation framework. It tell validation engineer which part of driver code is not validated by any test cases. It will encourage validation engineer to develop stricter test cases. This can help in developing even better quality display driver.