Overview

Introduction

The Kernel Developer’s Guide describes how to create a Kernel Application. A Kernel Application is a Standalone Application that can be extended (statically or dynamically) to install, run, and control the execution of new applications called Sandboxed Applications.

The intended audience of this document are Java developers and system architects who plan to design and build their own Kernel.

Here is a non-exhaustive list of the activities to be done by Kernel Developers:

  • Integrating the Kernel Application with a VEE Port to produce a Multi-Sandbox Executable and Virtual Device

  • Defining the set of APIs that will be exposed to Applications, optionally by maintaining a custom Runtime Environment

  • Managing lifecycles of applications (deciding when to install, start, stop and uninstall them)

  • Defining and applying permissions on system resources (rules & policies)

  • Managing connectivity

  • Controlling and monitoring resources

This document takes as prerequisite that a VEE Port is available for the target device (see VEE Porting Guide). This document also assumes that the reader is familiar with the development and deployment of Applications (see Application Developer Guide) and specifics of developing Sandboxed Applications (see Sandboxed Application).

Terms and Definitions

A Multi-Sandbox VEE Port is a VEE Port with the Multi-Sandbox capability of the Core Engine enabled (see the chapter Multi-Sandbox of the VEE Porting Guide). A Multi-Sandbox Executable can only be built with a Multi-Sandbox VEE Port.

A Virtual Device is the Multi-Sandbox Executable counterpart for developing a Sandboxed Application. It provides the Kernel functional simulation part. Usually it also provides a mean to directly deploy a Sandboxed Application on the target device running the Multi-Sandbox Executable (this is called Local Deployment).

Overall Architecture

Kernel Boundary Overview

Kernel Boundary Overview

Input and Output Artifacts

Kernel Input and Output Artifacts

Kernel Input and Output Artifacts

Kernel Build Flow

The following describes the Kernel build flow.

Kernel Build Flow

Kernel Build Flow

The Virtual Device builder performs the following steps:

  • Remove the embedded part of the VEE Port (including MEJ32).

  • Append Add-On Libraries and pre-installed Applications into the runtime classpath. See Kernel Module Configuration section for specifying the dependencies.

  • Add a custom license allowing Virtual Device redistribution.

  • Generate the Runtime Environment from the Kernel APIs.

Kernel Implementation Libraries

Kernel implementations must cover the following topics:

  • The kernel entry point implementation, that deals with configuring the different policies, registering kernel services and converters, and starting applications.

  • The storage infrastructure implementation: mapping the Storage service on an actual data storage implementation. There are multiple implementations of the data storage, provided in different artifacts that will be detailed in dedicated sections.

  • The applications management infrastructure: how application code is stored in memory and how the lifecycle of the code is implemented. Again, this has multiple alternative implementations, and the right module must be selected at build time to cover the specific Kernel needs.

  • The simulation support: how the Virtual Device implementation reflects the Executable implementation, with the help of specific artifacts.

  • The Kernel API definition: not all the classes and methods used to implement the Kernel Application are actually exposed to the Sandboxed Applications. There are some artifacts available that expose some of the libraries to the applications, these ones can be picked when the Kernel is assembled.

  • The Kernel types conversion and other KF-related utilities: Kernel types instances owned by one application can be transferred to another application through a Shared Interface. For that to be possible, a conversion proxy must be registered for this kernel type.

  • Tools libraries: tools that plug into the SDK, extending them with features that are specific to the Kernel, like deployment of an application, a management console, …

Getting started

Get familiar with Kernel development by following the training Get Started with Kernel.