Browse Prior Art Database

Maintenance Procedure for a Storage Controller in a Multiprocessing System with Nonstore through Local Storage Buffers

IP.com Disclosure Number: IPCOM000077915D
Original Publication Date: 1972-Oct-01
Included in the Prior Art Database: 2005-Feb-25
Document File: 3 page(s) / 52K

Publishing Venue

IBM

Related People

Carroll, JT: AUTHOR [+3]

Abstract

The device shown in Fig. 1 allows continuous availability of distributed memory system during some machine checks in a Processor/ Storage Controller. The device operates a multiprocessing system composed in part of groups of processing elements attached to storage control units. Valid blocks of main storage are distributed among the processing elements in local storage buffers (caches). Consequently, main memory accesses must be screened across all caches in the system, to assure that the most recently updated copy of a data block is retrieved. This is called broadcasting. When an error condition prevents broadcasting in any processing element attached to any storage control unit in the system, all main memory accesses must be halted until broadcasting can be restored.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 55% of the total text.

Page 1 of 3

Maintenance Procedure for a Storage Controller in a Multiprocessing System with Nonstore through Local Storage Buffers

The device shown in Fig. 1 allows continuous availability of distributed memory system during some machine checks in a Processor/ Storage Controller. The device operates a multiprocessing system composed in part of groups of processing elements attached to storage control units. Valid blocks of main storage are distributed among the processing elements in local storage buffers (caches). Consequently, main memory accesses must be screened across all caches in the system, to assure that the most recently updated copy of a data block is retrieved. This is called broadcasting. When an error condition prevents broadcasting in any processing element attached to any storage control unit in the system, all main memory accesses must be halted until broadcasting can be restored.

The storage control unit (called the Quad Storage Control or QSC) has three logical areas (Fig. 1):
1) Request Selection Logic (RSL) which selects requests for main

storage accesses.
2) Broadcast Logic (ECL) which broadcasts to attached processors

all system requests for main storage.
3) Data Return Logic (DRL) which controls the orderly return of

data from main storage to attached units.

Error Handling Procedure - The error handling procedure described here has the following features:
1) Allows some continued broadcasting to caches under its control,

after detection of error condi...