NOTE: This is still a work in progress project, suitable for: developers, contributors and testers. None of the provided tests have been extensively evaluated as of January 2017.
The LISA project provides a toolkit that supports regression testing and interactive analysis of Linux kernel behavior. LISA stands for Linux Integrated/Interactive System Analysis. LISA's goal is to help Linux kernel developers to measure the impact of modifications in core parts of the kernel. The focus is on the scheduler (e.g. EAS), power management and thermal frameworks. However LISA is generic and can be used for other purposes too.
LISA has a "host"/"target" model. LISA itself runs on a host machine, and uses the devlib toolkit to interact with the target via SSH, ADB or telnet. LISA is flexible with regard to the target OS; its only expectation is a Linux kernel-based system. Android, GNU/Linux and busybox style systems have all been used.
LISA provides features to describe workloads (notably using rt-app) and run them on targets. It can collect trace files from the target OS (e.g. systrace and ftrace traces), parse them via the TRAPpy framework. These traces can then be parsed and analysed in order to examine detailed target behaviour during the workload's execution.
Some LISA features may require modifying the target OS. For example, in order to collect ftrace files the target kernel must have CONFIG_DYNAMIC_FTRACE enabled.
There are two "entry points" for running LISA:
-
Via the Jupyter/IPython notebook framework. This allows LISA to be used interactively and supports visualisation of trace data. Some notebooks are provided with example and ready-made LISA use-cases.
-
Via the automated test framework. This framework allows the development of automated pass/fail regression tests for kernel behaviour. The BART toolkit provides additional domain-specific test assertions for this use-case. LISA provides some ready-made automated tests under the
tests/directory.
The main goals of LISA are:
- Support study of existing behaviours (i.e. "how does PELT work?")
- Support analysis of new code being developed (i.e. "what is the impact on existing code?")
- Get insights on what's not working and possibly chase down why
- Share reproducible experiments by means of a common language that:
- is flexible enough to reproduce the same experiment on different targets
- simplifies generation and execution of well defined workloads
- defines a set of metrics to evaluate kernel behaviours
- enables kernel developers to easily post process data to produce statistics and plots
This project is licensed under Apache-2.0.
This project includes some third-party code under other open source licenses. For more information, see lisa/tools/LICENSE.*
Contributions are accepted under Apache-2.0. Only submit contributions where you have authored all of the code. If you do this on work time make sure your employer is cool with this.