Variable timeout method in a missing-interrupt-handler for I/O requests issued by the same operating system

• US 5,768,620 A
• Filed: 04/09/1996
• Issued: 06/16/1998
• Est. Priority Date: 04/09/1996
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Ripberger, Richard Anthony, Yudenfriend, Harry Morris, Johnson, Gavin Stuart, Urbanejo, Luis Ricardo
• Primary Examiner(s): Lee, Thomas C.
• Assistant Examiner(s): Huang, Po C.
• Application Number:
  US08/631,689
• Time in Patent Office: 798 Days
• Field of Search: 371/3, 371/48, 371/62, 395/182.04, 395/182.14, 395/183.19, 395/184.01, 395/185.08, 395/553, 395/825, 395/835, 395/857, 395/868, 395/883
• US Class Current: 710/15
• CPC Class Codes: G06F 13/24 using interrupt G06F13/32 takes precedence

Abstract

Missing interrupt handler (MIH) internal software features support a variable MIH timeout for I/O requests issued by an operating system (OS), when the same OS is involved with both an executing I/O request and a waiting I/O request. The OS varies its MIH timeout period without a signal from any I/O entity to prevent a false indication of a potential failure in a current I/O device operation. If a current I/O request has not completed when the OS senses the end of a primary MIH timeout period, started when issuing that request, the OS then scans the I/O program of that I/O request for any contained long-running command. (Most I/O requests complete during their primary MIH timeout period.) If a long command is found, the OS extends the MIH timeout period from the primary MIH timeout period to a long MIH timeout period, The latter gives the I/O device more time to complete its operation before the OS indicates it has a potential I/O error condition. But if the OS does not detect any long-running command in its scan of the I/O program, the OS does not extend the primary MIH timeout period, and then the OS declares a potential I/O error condition for the current I/O device operation and invokes a conventional I/O error recovery program, such as retrying that I/O request for a number of times until it runs without error or until a permanent error condition is determined.

Claims

1. A method of avoiding false detection of missing I/O interrupt signals by an operating system (OS) for use in a computer arrangement having at least one OS and a missing interrupt handler (MIH) process for detecting a potentially failed input/output (I/O) device operation for an I/O request issued by the OS to an I/O subsystem, the MIH process comprising the steps of:

sending by the OS to the I/O subsystem a first I/O request and a second I/O request, the first I/O request associated with a first I/O program containing at least one command for controlling an I/O device, the second I/O request associated with a second I/O program containing at least one command for controlling the I/O device,

operating the I/O device under control of the first I/O program for the first I/O request, delaying operation of the second I/O request in the I/O subsystem because of the I/O device currently operating for the first I/O request,

measuring MIH time for a primary MIH timeout period starting when the second I/O request is made to wait by the I/O subsystem,

signalling a device-completion interrupt to the OS when and if the I/O device successfully completes operation for the first I/O request,

terminating the measurement step if the device-completion interrupt for the first I/O request is received by the OS before measuring the end of the primary MIH timeout period,

but scanning the first I/O program for a long command requiring the I/O device to operate for an excessive amount of time if no device-end interrupt is received by the OS before measuring the end of the primary MIH timeout period,

continuing measuring the MIH time for a secondary MIH timeout period if the scanning step finds the long command in the first I/O program, and

terminating the continuing measuring step if the device-completion interrupt for the first I/O request is received by the OS before the measuring step indicates the end of the secondary MIH timeout period.

2. An MIH process as defined in claim 1, further comprising the step of:

enabling the second I/O program to operate the I/O device for the second I/O request in response to the step of terminating the continuing measuring step.

3. An MIH process as defined in claim 1, further comprising the step of:

indicating a MIH potential error condition for the operation of the I/O device if the continuing measuring step indicates the end of the secondary MIH timeout period when the OS has not received the device-completion interrupt.

4. An MIH process as defined in claim 3, further comprising the step of:

invoking an I/O error recovery program for the first I/O request if the MIH potential error condition is indicated by the indicating step.

5. An MIH process as defined in claim 4, further comprising the step of:

retrying the first I/O request in response to the step of invoking the I/O error recovery program for the first I/O request.

6. An MIH process as defined in claim 1, further comprising the step of:

releasing I/O resources used by the first I/O request at least by the time the continuing measuring step indicates an end to the secondary MIH timeout period.

7. An MIH process as defined in claim 1, in which the computer arrangement has a plurality of OSs, the method further comprising the steps of:

setting a busy indicator in an I/O controller when any I/O program is accepted by the I/O controller for operation of the I/O device,

detecting in the I/O controller when operation time for the I/O device for a current request exceeds a predetermined threshold time in relation to a waiting I/O request from the OS,

setting a long-busy indicator in the I/O controller when the threshold time is exceeded,

signalling a long-busy interrupt by the I/O controller to each OS having an I/O request waiting in the I/O subsystem to indicate to each OS that a current I/O request is operating a long time on the device,

receiving by each OS the long-busy interrupt signal before an end of the primary MIH timeout period is measured,

and continuing measuring by each OS the MIH time for the secondary MIH timeout period in response to the long busy interrupt signal being received prior to the end of the primary MIH timeout period.