Browse Prior Art Database

Mechanism for Managing Tables for High Speed Recording

IP.com Disclosure Number: IPCOM000045120D
Original Publication Date: 1983-Feb-01
Included in the Prior Art Database: 2005-Feb-06
Document File: 2 page(s) / 14K

Publishing Venue

IBM

Related People

Levin, AL: AUTHOR [+2]

Abstract

In some computer architectures, a trace facility is provided for automatically noting, in a storage table, information about selected events relevant to program execution. Data may collect in this table at such a high but random rate that it becomes impractical to develop a table size that guarantees the opportunity to record the contents of each table on an external medium. In addition, the processing environment at the time a table fills is such that a device output request cannot practically be issued. The following describes a mechanism for managing these problems.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 54% of the total text.

Page 1 of 2

Mechanism for Managing Tables for High Speed Recording

In some computer architectures, a trace facility is provided for automatically noting, in a storage table, information about selected events relevant to program execution. Data may collect in this table at such a high but random rate that it becomes impractical to develop a table size that guarantees the opportunity to record the contents of each table on an external medium. In addition, the processing environment at the time a table fills is such that a device output request cannot practically be issued. The following describes a mechanism for managing these problems.

Two or more tables are reserved for use by the trace facility. The size of each table, and the number of tables, is established based on the expected table fill rate and an estimated acceptable loss level. A control element is defined for each table creating a wrapping chain: BACKPTR FOREPTR TABLE LIND TOD1 TOD2 BACKPTR Identifies the previous control element in the chain (the last element in the chain, if this is the first).

FORBPTR Identifies the next control element in the chain (the first element in the chain, if this is the last).

TABLE Identifies the first location for trace information LIND storage. A counter incremented by 1 whenever the associated table is selected as the current trace table

(i.e., lost data indicator).

TOD1 Storage for a clock reference when a table is selected as the current trace table.

TOD2 Storage for a clock reference when the table saturates.

When the control chain is created, all LIND fields are at 0. When the trace facility is set into operation, the first control element is obtained. A sequence "A" is initiated where LIND is incremented by 1; a clock reference is stored in TOD1; TOD2 is set to 0. The trace facility is initialized to use TABLE. When the trace table saturates, a program interrupt occurs, a clock reference is stored in TOD2, the next control element is obtained, and the process is repeated from A above.

A master pointer is maintained by the external medium recording service (EMRSS). This pointer identifies the control element to be referenced next. A waiting interval is determined based on estimated fill and "acceptable loss" rates. When an interval elaps...