Method and system for detecting excessive interrupt processing for a processor
AssigneeHewlett-Packard Development Company, L.P.
InventorsRandall F. Wright, Jerry A. Hoemann
Life of this patent
Estimating the term…
Abstract
A method and system is provided for detecting excessive interrupt processing for a processor. The method includes the operation of defining an interrupt processing period during which measuring of interrupts for a processor takes place. The amounts of time being spent by the processor in an interrupt context can then be measured during the interrupt processing period. A further operation is detecting an interrupt storm occurring for the processor based on the amounts of time spent by the processor in interrupt context during the interrupt processing period.
Claims26
A method for detecting excessive interrupt processing for a processor, comprising:
- defining an interrupt processing period during which measuring of interrupts for a processor takes place;
- measuring amounts of time being spent by the processor in an interrupt context during the interrupt processing period by assigning different interrupt priorities to device level interrupts and to interrupt routines for measuring time in the interrupt context;
- and
- detecting when an interrupt storm occurs for the processor based on the amounts of time spent by the processor in interrupt context during the interrupt processing period, wherein the interrupt storm occurs when the processor spends more than a specified percentage of processing time in the interrupt context.
A method for detecting excessive interrupt processing for a processor, comprising the steps of:
- defining an interrupt processing period during within which interrupts are measured for a processor;
- assigning interrupt routines a different priority than device interrupts to correctly measure amounts of time being spent by the processor in an interrupt context during the interrupt processing period;
- computing a smoothed average using the amounts of time spent in the interrupt context by the processor during the interrupt processing period;
- determining when an interrupt storm is occurring for the processor by comparing the smoothed average to a pre-determined interrupt storm threshold.
A system for detecting excessive interrupt processing overhead for a processing system, comprising:
- an operating system kernel configured to schedule processes;
- a processor configured to execute processes received from the operating system kernel and receiving interrupts, wherein the processor switches to interrupt context when in an interrupt is received;
- a measurement module configured to measure amounts of time spent by the processor in interrupt context over an interrupt processing period by assigning different interrupt priorities to device level interrupts and to interrupt routines for measuring time in the interrupt context;
- a smooth scaling module configured to compute a smoothed average of measured amounts of time spent in interrupt context by the processor;
- and
- a comparison module for comparing the smoothed average received from the smoothing scaling module with a pre-determined interrupt storm threshold to identify an interrupt storm.
A method for detecting excessive interrupt processing overhead for an operating system kernel, comprising the steps of:
- defining a processing period over which interrupts will be measured for an operating system kernel;
- measuring amounts of time being spent by the operating system kernel in an interrupt context by assigning different interrupt priorities to device level interrupts and to interrupt routines for measuring time in the interrupt context;
- computing a smoothed average of amounts of time being spent in interrupt context by the operating system kernel;
- determining when an interrupt storm is occurring for a operating system kernel by comparing the smoothed average to a pre-determined interrupt storm threshold.
A system for detecting excessive interrupt processing overhead for a processing system, comprising:
- an operating system kernel means configured to schedule processes;
- a processor means for executing processes received from the operating system kernel means and receiving interrupts, wherein the processor means switches to interrupt context when in an interrupt is received;
- a measurement means that assigns different interrupt priorities to device level interrupts than to interrupt routines for measuring amounts of time spent by the processor means in interrupt context over an interrupt processing period;
- a smooth scaling means for computing a smoothed average of measured amounts of time spent in interrupt context by the processor means;
- and
- a comparison means for comparing the smoothed average received from the smooth scaling means with a pre-determined interrupt storm threshold to identify an interrupt storm.
An article of manufacture, comprising:
- a computer usable medium having computer readable program code embodied therein for detecting excessive interrupt processing for a processor, the computer readable program code means in the article of manufacture comprising:
- computer readable program code configured for defining an interrupt processing period during which measuring of interrupts for a processor takes place;
- computer readable program code configured for measuring amounts of time being spent by the processor in an interrupt context during the interrupt processing period by assigning different interrupt priorities to device level interrupts and to interrupt routines for measuring time in the interrupt context;
- and
- computer readable program code configured for detecting when an interrupt storm occurs for the processor based on the amounts of time spent by the processor in interrupt context during the interrupt processing period, wherein the interrupt storm occurs when the processor spends more than a specified percentage of processing time in the interrupt context.
Dependencies are inferred from the claim text.