Using Linux in Embedded and Real-Time Systems

When you need an embedded operating system, Linux is a good place to start. Here's why.

Intelligent dedicated systems and appliances used in interface, monitoring, communications and control applications increasingly demand the services of a sophisticated, state-of-the-art operating system. Many such systems require advanced capabilities such as high resolution and user-friendly graphical user interfaces (GUIs), TCP/IP connectivity, substitution of reliable (and low-power) flash memory solid-state disk for conventional disk drives, support for 32-bit ultra-high-speed CPUs, the use of large memory arrays, and seemingly infinite capacity storage devices including CD-ROMs and hard disks. This is not the stuff of yesteryear's “stand-alone” code, “roll-your-own” kernels or “plain old DOS”. No, those days are gone—forever!

Then, too, consider the rapidly accelerating pace of hardware and chipset innovation, accompanied by extremely rapid obsolescence of the older devices. Combine these two, and you can see why it has become an enormous challenge for commercial RTOS (real-time operating system) vendors to keep up with the constant churning of hardware devices. Supporting the newest devices in a timely manner—even just to stay clear of the unrelenting steamroller of chipset obsolescence—takes a large and constant resource commitment. If it's a struggle for the commercial RTOS vendors to keep up, going it alone by writing stand-alone code or a roll-your-own kernel certainly makes no sense. With the options narrowing, embedded system developers find themselves faced with a dilemma:

  • On one hand, today's highly sophisticated and empowered intelligent embedded systems—based on the newest chips and hardware capabilities—demand nothing less than the power, sophistication and currency of support provided by a popular high-end operating system like Windows.

  • On the other hand, embedded systems demand extremely high reliability (for non-stop, unattended operation) plus the ability to customize the OS to match an application's unique requirements.

Here's the quandary: general-purpose desktop operating systems (like Windows) aren't well-suited to the unique needs of appliance-like embedded systems. However, commercial RTOS, while designed to satisfy the reliability and configuration flexibility requirements of embedded applications, are increasingly less desirable due to their lack of standardization and their inability to keep pace with the rapid evolution of technology.

The Alternative

Fortunately, an exciting alternative has emerged: open-source Linux. Linux offers powerful and sophisticated system management facilities, a rich cadre of device support, a superb reputation for reliability and robustness and extensive documentation. Best of all (say system developers), Linux is available with source code at no charge.

Unlike Windows, Linux is inherently modular and can be scaled easily into compact configurations—barely larger than DOS—that can even fit on a single floppy. What's more, since Linux source code is freely available, it's possible to customize the OS according to unique embedded system requirements. It's not surprising, then, that Linux has created a new OS development and support paradigm in which thousands of developers continually contribute to a constantly evolving Linux code base. In addition, dozens of Linux-oriented software companies have sprung up, eager to support the needs of developers building a wide range of applications, ranging from factory automation to intelligent appliances.

Which Linux?

Because Linux is openly and freely available in source form, many variations and configurations are available. There are implementations specifically for “thin server” or “firewall” applications, small footprint versions and real-time enhanced versions. There are also Linux ports for non-x86 CPUs, including PowerPC, RISC, 68xxx and microcontrollers.

How do you decide which distribution to use? First, realize that all Linux distributions are variations on the same theme—that is, they are collections of the same basic components, including the Linux kernel, command shells (command processors) and many common utilities. The differences tend to center around which of the many hundreds of Linux utilities have been included, what extras are included on the distribution CD and how the installation process is managed. Distributions may include either open-source or proprietary software such as, for example, StarOffice. Even though Linux is free, purchasing a “commercial” Linux distribution can have many advantages, including development tools, useful utilities and support. The many companies which are now building businesses around distributing and supporting Linux are busy investing in developing tools and services to differentiate their Linux offerings from the rest of the pack. Some of the special capabilities being developed include:

  • Installation tools to automate and simplify the process of generating a Linux configuration that is tuned to a specific target's hardware setup. These can save weeks or even months of development effort.

  • A variety of Windows-like GUIs that vary in size, appearance, features and capabilities to support a wide range of embedded requirements.

  • Support for the specific needs of various embedded and real-time computing platforms and environments (e.g., special CompactPCI system features).

Yet despite this great diversity, all Linux implementations share a common core, including kernel, drivers, several popular GUIs and utilities.