Browse Prior Art Database

Enhanced Dynamic Watchdog for Embedded Applications

IP.com Disclosure Number: IPCOM000019997D
Original Publication Date: 2003-Oct-16
Included in the Prior Art Database: 2003-Oct-16
Document File: 2 page(s) / 45K

Publishing Venue

IBM

Abstract

This invention disclosure describes a new method for recovering an embedded system that has become unresponsive. The solutions available today provide limited coverage in both time and monitoring ability. The disclosure provides a method to expand the coverage area, providing a more flexible and effective solution.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 1 of 2

Enhanced Dynamic Watchdog for Embedded Applications

   The basic problem is how to recover from an embedded system, without an operator console, when the software becomes unresponsive. A common solution is to use a user space application, the watchdog, which interfaces with a hardware timer. When enabled, the hardware timer triggers periodically after a fixed interval. If the watchdog has not reset the timer trigger before the next time interval, the embedded system is restarted.

     This approach addresses the lack of coverage area for today's watchdog devices. Most watchdog devices are used to monitor the ability to run user space applications. If a special user space application has started, and fails to reset to the timer device within the specific time-out period, the machine is restarted. If the application dies or exits, the behavior of the watchdog device is specified by the CONFIG_WATCHDOG_NOWAYOUT configuration option. If it is set, the device keeps running and will restart the machine if another instance of the application is not started within the amount of remaining time. If the configuration option is not set, the watchdog device is disabled.

     The major limitation of these devices is that they do not begin the monitoring period until a user space application has opened the device for writing. This means that if that application is never started, the watchdog device remains disabled, rendering its services useless.

     This invention provides a way to dynamically monitor both user space applications and the servicing of kernel space functionality. Also, the watchdog device is started with the kernel initcalls, so the device provides the maximum possible coverage.

     The flexible service processor (FSP) hardware has been implemented with a unit check timer (UCT). This is used by service processor code to detect hang or infinite loop conditions. If the UCT has been enabled and the programmed amount of time has elapsed, a low priority interrupt is generated. If this interrupt bit is not cleared and a second mount of the programmed time has elapsed, then a critical interrupt is generated. If this interrupt bit is not cleared and a third amount of the pr...