Programmable hardware event monitoring method

• US 6,134,676 A
• Filed: 04/30/1998
• Issued: 10/17/2000
• Est. Priority Date: 04/30/1998
• Status: Expired due to Term

• Current Assignee: Intel Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: VanHuben, Gary Alan, Ford, III, Carl B.
• Primary Examiner(s): Beausoliel, Jr., Robert W.
• Assistant Examiner(s): Elisca, Pierre E
• Application Number:
  US09/070,145
• Time in Patent Office: 901 Days
• Field of Search: 714/39, 714/45, 714/47, 714/37, 714/38, 364/264, 364/264.1, 364/264.4
• US Class Current: 714/39
• CPC Class Codes: G06F 11/3466 Performance evaluation by tracing or monitoring
• CPC Class Codes: G06F 2201/86 Event-based monitoring
• CPC Class Codes: G06F 2201/88 Monitoring involving counting
• CPC Class Codes: G06F 2201/885 Monitoring specific for caches

Abstract

A system for monitoring hardware events in a computer system implements a hardware event monitor of control registers and has programmable generic fields which can be internally or externally programmed for monitoring events ranging from simple operations to complex event sequences. Using programmed criteria incorporated into the hardware event monitor to events within said computer system are monitored by initiating successive compares of programmed criteria with processing events. The hardware event monitor can trigger external actions upon successful detection of compares indicating that all criteria programmed into the hardware event monitor have been achieved. None of the bits within the Control Registers are mutually exclusive which allows multiple scenarios to be programmed. The hardware event monitor and system trace controls act as a single unified entity with remote programming of said hardware event monitor and trace controls over a UBUS to permit capturing and logging problem debugging and instrumentation data for use in remote system administration, technical support assistance, field and customer engineering applications, and performance analysis, and hardware error injection for recovery and diagnostic testing, and enabling dither to break resource deadlocks.

Claims

1. A method for monitoring hardware events in a computer system, comprising the steps of:

implementing a hardware event monitor of control registers and having programmable generic fields which can be internally or externally programmed for monitoring events ranging from simple operations to complex event sequences and

using programmed criteria incorporated into said hardware event monitor to monitor events within said computer system by initiating successive compares of programmed criteria with processing events, and

using said hardware event monitor to trigger external actions upon successful detection of compares indicating that all criteria programmed into the hardware event monitor have been achieved, and wherein

all aspects of the event monitor are replicated to form two or more Compare Stations, each of which can operate independently as disjoint functional units or can be chained together to act as a single entity; and

each of said Control Stations can monitor all fields, of any and all pipelines or other resources of said computer system, each Compare Station providing an independent compare output signal governed by a unique set of Control Registers for that Compare Station, and using a unique or shared event counter, whereby said compare output signals of each station can be used to trigger the same external actions and can be used to trigger different external actions and can be further combined using AND/OR/Invert (AOI) logic to create "super compare" functions such as, but not limited to;

space="preserve" listing-type="equation">(F1x*F2x)+(F1y*F2y)

space="preserve" listing-type="equation">(F1x+F2x)*(F1y+F2y).

2. A method according to claim 1, wherein programmable generic fields are masked with corresponding field masks to expand the scope of the compares to perform operations, including one or more matches selected from a group consisting of:

an exact match against a field, wildcard matches against portions of a field, or exclusion of unwanted fields from the compare, or a combination thereof.

3. A method according to claim 1, wherein a programmable fields has control bits used for providing information, and said control bits are

used to decode a portion of said field or a completely different field to discern the type of information currently residing in the field,

matched against corresponding control bits which exist as part of said field or existing in a separate field, and upon detecting a match, allow the contents of said field to participate in the compare.

4. A method according to claim 3, wherein one or more multistage pipelines of said computer system are monitored by controls permitting the compare to be reported in any, or all, stages of the pipeline;

and by compare logic for the first stage of the pipeline such that a resulting compare is captured in a one-cycle staging register and pipelined once for each stage, wherein the output of each staging register is combined with the corresponding control bits to determine whether the compare should be reported on that cycle.

5. A method according to claim 4, wherein is included controls for selecting which pipeline should participate in a compare;

and controls permitting compares to be reported only if the operation is valid in a corresponding pipeline stage.

6. A method according to claim 1, wherein said computer system provides a Compare Type switch to control whether the fields are AND'"'"'d or OR'"'"'d together to form an overall compare, and for combining individual field compares together according to the Compare Type Switch using Field Include Bits to perform an appropriate boolean field function using a one-cycle delay to reuse gate-level building blocks for testability.

7. A method according to claim 1, wherein compares can be made against the inverse of a multi-bit field.

8. A method according to claim 1, wherein event monitoring is delayed until after resetting or rebooting (IML) of the system to prevent unwanted compares during an IML phase.

9. A method according to claim 1, wherein the hardware event monitor'"'"'s Control Registers and fields can be programmed over a Microcode Bus (UBUS) permitting microcode patches or other similar means to be applied while the systems clocks are running.

10. A method according to claim 9, wherein said Control Registers and fields are directly manipulated by the Operating System, Application Software, Operator'"'"'s Service Console, and by loading hardware registers through external means to allow the hardware event monitor to be programmed.

11. A method according to claim 10, wherein said hardware event monitor is remotely programmed by delivering the desired contents of the Control Registers and fields via Local Area or Wide Area Networks, telephone connections, the Internet or through other conduits, including upgradable BIOS and Digital Signal Processor code patches.

12. A method according to claim 1, wherein is included controls permitting the event monitor to suppress reporting of compares until after "N" successful compares have occurred;

and controls permitting the event monitor to disable reporting further compares after "N" successful compares have occurred.

13. A method according to claim 1, wherein all aspects of the event monitor are replicated to form two or more Compare Stations, each of which can operate independently as disjoint functional units or can be chained together to act as a single entity;

and each of said Control Stations can monitor all fields, of any and all pipelines or other resources of said computer system, each Compare Station providing an independent compare output signal governed by a unique set of Control Registers for that Compare Station, and using a unique or shared event counter, whereby said compare output signals of each station

can be used to trigger the same external actions

can be used to trigger different external actions

can be further combined using AND/OR/Invert (AOI) logic to create "super compare" functions such as, but not limited to,

space="preserve" listing-type="equation">(F1x*F2x)+(F1y*F2y)

space="preserve" listing-type="equation">(F1x+F2x)*(F1y+F2y).

14. A method according to claim 13, wherein the output of one Compare Station is used to enable another Compare Station to begin monitoring for the same or different events, and said first Compare Station can be used to disable another Compare Station, thus permitting complex event sequences to be monitored.

15. A method according to claim 14, wherein compare output signals from the hardware event monitor are used to trigger external actions for one or more of actions selected from the group of gathering problem debug information using trace arrays, instrumentation sampling for performance analysis, hardware error injection for recovery and diagnostic testing, enabling dither to break resource deadlocks.

16. A method according to claim 15, wherein said hardware event monitor operates independently of trace controls such that said event monitor targets one event while trace controls gather data for an unrelated event.

17. A method according to claim 1, wherein said hardware event monitor operates independently of trace controls and as an extension to system trace controls in which event monitoring can be enabled or disabled when trace gathering starts, trace gathering stops, or compares can only be reported when the trace is actually running.

18. A method according to claim 17, wherein trace gathering can be started, stopped, and advanced by a programmable number of cycles whenever successful compares are reported permitting said hardware event monitor and trace controls to act as a single unified entity such that all modes of operation available to both functions independently can be combined to accomplish monitoring and data gathering applications including tracing state machines, capturing small snapshots of multiple passes of a single event, and trapping events within subroutines invoked by a particular driver program.

19. A method according to claim 18, wherein said hardware event monitor and trace controls act as a single unified entity with remote programming of said hardware event monitor and trace controls over a UBUS to permit capturing and logging problem debugging and instrumentation data for use in remote system administration, technical support assistance, field and customer engineering applications, and performance analysis.

20. A method according to claim 1, wherein none of the bits within said Control Registers are mutually exclusive.