Browse Prior Art Database

Synchronizing Error Reporting and Recovery

IP.com Disclosure Number: IPCOM000061204D
Original Publication Date: 1986-Jul-01
Included in the Prior Art Database: 2005-Mar-09
Document File: 1 page(s) / 12K

Publishing Venue

IBM

Related People

Scott, EK: AUTHOR [+2]

Abstract

Two loosely-coupled programmed units, such as a host processor and a programmable peripheral control unit, are synchronized in the error detection and recovery for peripheral devices attached to the control unit. In certain conditions, if the control unit detects an error and logs it within the control unit, subsequent events can cause a loss of the reported error. Error reporting to the host is delayed until the host missing-interrupt handler (MIH) detects the problem such that the peripheral subsystem error is never lost. In cached peripheral subsystems, a peripheral device works with a semiconductor cache. Data transfers occur between the device and the cache. During such asynchronous transfers, an error condition may occur in the device to prevent further activity of the device in the subsystem.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 60% of the total text.

Page 1 of 1

Synchronizing Error Reporting and Recovery

Two loosely-coupled programmed units, such as a host processor and a programmable peripheral control unit, are synchronized in the error detection and recovery for peripheral devices attached to the control unit. In certain conditions, if the control unit detects an error and logs it within the control unit, subsequent events can cause a loss of the reported error. Error reporting to the host is delayed until the host missing-interrupt handler (MIH) detects the problem such that the peripheral subsystem error is never lost. In cached peripheral subsystems, a peripheral device works with a semiconductor cache. Data transfers occur between the device and the cache. During such asynchronous transfers, an error condition may occur in the device to prevent further activity of the device in the subsystem. Such a condition is often colloquially referred to as a "hung device". Later, a host processor needs to access some data from the peripheral device. If when the data requested by the host processor is not resident in the cache, then the host requires a logical and physical connection to the device storing the data. Then during such device election, the control unit attempts to complete the selection of the device and finds the device busy (a hung device always appears as being busy) and issues a device busy to the host. The host processor MIH receives the device-busy signal and sets a timer for receiving a device end, i.e., t...