Real-Time Operating System Patent Awarded to Lynx Real-Time Systems, Inc. Unique interrupt approach a breakthrough for embedded applications in multi-media, ATM, and other high-end applications areas

READ OUR PATENTED I/O WHITE PAPER

SAN JOSÉ, Calif., March 18, 1996—Lynx Real-Time Systems, Inc. has been awarded a patent for a key part of its real-time operating system (RTOS) architecture, the company announced today.

The invention provides a unique approach to handling interrupts. Interrupts are the basis for an operating systems (OS) to schedule different computer activities. The new technology will enable programmers to develop embedded applications that are both complex and also meet real-time requirements. The ability to meet both of these needs is critical for high-end multimedia and ATM-based communications applications.

"Before we created this architecture, design engineers could not economically write complex, embedded applications while still meeting real-time processing needs," said Inder M. Singh, President and CEO of Lynx Real-Time Systems, Inc. "Due in part to this patent, we're seeing many companies develop software more quickly and speed up time-to-market of increasingly robust applications by outsourcing their RTOS needs to LynxOS, an off-the-shelf operating system that incorporates this technology."

Growing needs for real-time and complex embedded applications

The price of 32-bit processors has decreased dramatically over the past 12-to-16 months, spawning a new growth in embedded applications, particularly in communications applications such as in network bridges and routers, digital telephone switches, multi-media systems, and highly graphical video-conferencing products. According to Hambrecht & Quist forecasts, unit sales of 32-bit microcontrollers will double annually until 1998, giving off-the-shelf real-time operating systems companies like Lynx Real-Time Systems excellent growth potential.

One stumbling block for many of these complex applications has been the need to meet "hard" real-time needs. Hard real-time refers to applications which, if not met consistently and reliably, result in lost productivity, profits, or in some cases, lives:

• Most anyone who has tried playing digital video has been victim of intermittent real-time failures, seriously damaging quality of service.
• Factories can lose tens of thousands of dollars when they experience relatively short system shut-downs based on automation programs whose complex, real-time applications fail to perform properly.
• Aerospace and automobile systems failures can be devastating, resulting in losses of millions of dollars and potentially, lost lives.

"In the past, it was very difficult—if not impossible—to build complex systems, make use of modern architectures and capabilities, and still retain real-time performance, without disabling certain devices within the computing system," said Mitchell Bunnell, Technical Founder at Lynx Real-Time Systems, Inc. and inventor of the patented technology. "The problem with the old technology was that it didn't take into account the extraordinary demands made by several interrupting devices, such as network devices on a single system."

What it does, and how it works

The patented invention is generally related to interrupt handling control. Following is a brief explanation of what interrupts are, how they have been dealt with in the past and the problems with traditional methods, and finally how this patented technology solves those problems.

All modern CPUs allow external devices (usually I/O devices) to suspend the current program sequence in order to execute a special sequence of instructions to "handle" the device.

This request to change the execution sequence is called an interrupt. The special sequence of code written to execute when an interrupt occurs is the interrupt handler. Traditionally, interrupt handlers execute to completion unless preempted by a higher priority interrupt. If any device is requesting an interrupt, the CPU will execute an interrupt handler. User programs are not usually directly involved with interrupts. That is a job left up to the operating system.

In a real-time system, certain tasks must execute and complete certain functions with a time deadline after an external event or periodically at some rate. In a traditional environment, even these real-times tasks will get no chance to run until no devices are requesting an interrupt. There is no bound placed by the kernel on the CPU time stolen from real-time tasks by interrupt handling in a traditional system. If too much time is stolen, though, tasks will miss their time deadlines.

Failure to finish the desired function before the deadline could cause a failure of the system. For instance, in a multi-media system, failure of a display task to update the display within a screen frame time will cause a glitch on the display. In a data acquisition system, failure to read all the A/D samples in the input buffer before the buffer overflows will cause a loss of data. This may cause all the collected data to be useless. A real-time kernel must schedule tasks so that all critical tasks meet their deadlines. This is a difficult job if interrupts are stealing too much time from these critical tasks and is a major cause of intermittent failure of real-time systems.

The patented invention schedules interrupt handling at normal task level priorities, not interrupt priorities. This means real-time tasks can execute even if devices are requesting interrupts. There is also a bound of only a few microseconds placed on the amount of CPU time that can be stolen from a time critical task during one cycle of its execution by a device that is accessed only by less time critical tasks. By scheduling interrupt handling better, this invention allows a computer to execute more time-critical tasks simultaneously and allows the use of more external devices and types of devices than the traditional interrupt mechanism without risking intermittent failures.

Proprietary patent supports POSIX, UNIX standards

The patented technology is proprietary to Lynx, although it has been incorporated into Lynx's real-time operating systems which complies to UNIX® and POSIX® standards.

No other real-time operating system matches the degree of LynxOS' compliance to POSIX. IEEE's POSIX standard plays a key role in leading the embedded, real-time world toward open systems. It defines the industry's standard application programming interface (API) for UNIX, and provides the first standard API for real-time embedded applications.

Lynx Real-Time Systems, Inc. is a worldwide supplier of real-time operating systems and development tools. Its products shorten product development cycles and lower costs for software engineers who design mid-range to high-end embedded applications. By committing to POSIX, UNIX, and other industry standards, the company enables customers to leverage their investments in software development and education. Founded in 1988 as a privately-held company, Lynx has offices in the U.S., Europe and Japan, and worldwide headquarters in San José, Calif.

LynxOS and the Lynx logo are trademarks of Lynx Real-Time Systems, Inc. All other product names, trademarks, and registered trademarks are the property of their respective holders.

LynuxWorks press contacts

Kirsten Long Director, Marketing Communications LynuxWorks, Inc. USA +1 408-979-4404	Barbara Stewart Patterson & Associates USA 480-488-6909
Hamid Mirab Managing Director EMEA LynuxWorks Ltd. United Kingdom +44 208-906-9506	Peter van der Sluijs Neesham Public Relations Europe +44 1442-879222