Browse Prior Art Database

Partial Data Page Write Detection

IP.com Disclosure Number: IPCOM000045689D
Original Publication Date: 1983-Apr-01
Included in the Prior Art Database: 2005-Feb-07
Document File: 1 page(s) / 12K

Publishing Venue

IBM

Related People

Crus, RA: AUTHOR [+3]

Abstract

This article describes a method for operating a computing apparatus to detect, upon reading a logical page from working storage, a write error occurring during the previous write of the logical page to working storage, comprising the steps of: establishing check bits at the beginning and end of a logical page to be written on to working storage, the check bits having a predetermined correspondence (say, equal) , and inverting the check bits prior to each write; writing the logical page in to working storage; reading the logical page from working storage; and testing the check bits. noting an error if the predetermined correspondence does not exist.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 72% of the total text.

Page 1 of 1

Partial Data Page Write Detection

This article describes a method for operating a computing apparatus to detect, upon reading a logical page from working storage, a write error occurring during the previous write of the logical page to working storage, comprising the steps of: establishing check bits at the beginning and end of a logical

page to be written on to working storage, the check bits

having a predetermined correspondence (say, equal) , and

inverting the check bits prior to each write;

writing the logical page in to working storage;

reading the logical page from working storage; and

testing the check bits. noting an error if the predetermined

correspondence does not exist.

A typical subsystem stores/manages data in "logical" pages of, say, 4K or 32K bytes. The mapping of these pages to physical blocks on secondary storage is dependent on the page size and on the physical characteristics of the secondary storage device. Always in the case of the 32K page size and, in some cases, of the 4K page size, a single page will be mapped to multiple physical blocks.

If a write operation fails (e.g., because of a system crash) in such a way that the first n physical blocks of a page are written but the last m are not, then it must be possible to detect this condition upon subsequent read of the page from secondary storage.

The technique used is to initialize a bit in the first byte of the page and a bit in the last byte of the page to the same value when a page is f...