Browse Prior Art Database

Buffering Trace Information

IP.com Disclosure Number: IPCOM000115965D
Original Publication Date: 1995-Jul-01
Included in the Prior Art Database: 2005-Mar-30
Document File: 2 page(s) / 38K

Publishing Venue

IBM

Related People

Butler, N: AUTHOR [+3]

Abstract

Many processes, device drivers, and kernel extensions log trace information into a circular trace buffer when a system is running.

This text was extracted from an ASCII text file.
This is the abbreviated version, containing approximately 84% of the total text.

Buffering Trace Information

      Many processes, device drivers, and kernel extensions log trace
information into a circular trace buffer when a system is running.

      On a busy system there will be lots of data going into the
trace buffer.  Since the trace buffer is circular, if a particular
event occurs, the trace must be turned off if the event is to be
examined, otherwise the event will be overwritten.

      Normally, the command line is used for turning on and off
system trace, which means that the user needs to sit waiting for the
event to occur, then use the command line to turn off the system
trace.  Typically, an event is associated with an error being logged
in the application.

      In accordance with the solution disclosed here, all errors are
channelled through a single point in the application, which is a
program that processes errors.

      The program includes a list of error numbers.  When a listed
error is detected, the program freezes the system trace.

      A second list may associate with each error number in the error
list a delay value to indicate the delay between when the error
arrives, and when the program should freeze the system trace, for
those situations in which the trace of system behavior after the
error occurs needs to be captured.

      A configuration file containing these 2 lists will be read at
startup or on demand by the program to which all errors are directed.

      Alternatively, and more...