Method and apparatus for fault recovery within a computing system

• US 4,703,481 A
• Filed: 08/16/1985
• Issued: 10/27/1987
• Est. Priority Date: 08/16/1985
• Status: Expired due to Term

• Current Assignee: Hewlett-Packard Company (HP Inc.)
• Original Assignee: Hewlett-Packard Company (HP Inc.)
• Inventors: Fremont, Michael J.
• Primary Examiner(s): Atkinson, Charles E.
• Application Number:
  US06/766,212
• Time in Patent Office: 802 Days
• Field of Search: 371/12, 371/16, 371/19, 364/200
• US Class Current: 714/17
• CPC Class Codes: G06F 11/1407 Checkpointing the instruction stream

Abstract

A method and apparatus are presented which enable recovery from a fault occurrence within a computing system which occurs during an original execution of machine instructions. The computing system is reset to a previously recorded system state, and a re-execution of machine instructions is initiated. To repeat an asynchronous event at a point in the re-execution of machine instructions identical to a point at which the repeating event was handled in the original execution of machine instructions, machine instructions executed and re-executed are counted. During the re-execution of machine instructions, the computing system performs operations upon identical inputs, at identical points, that the computing system did during the original execution of machine instructions. The computing system is thereby restored to a final system state identical to a specified system state existing before fault occurrence.

Claims

1. A method of preparing for recovery from a fault detected within a computing system, said method comprising the steps of:

recording a checkpoint system state of the computing system at a first checkpoint;

counting machine instructions executed after recording the checkpoint system state; and

,

logging determining event information for a handling of a determining event.

2. A method as in claim 1, further comprising the step of:

logging non-recurring inputs.

3. A method as in claim 2, wherein the determining event information includes a count of a number of machine instructions executed between an identified instruction point and the occurrence of a determining event.

4. A method as in claim 3, wherein an interval between the first checkpoint and a second checkpoint is made programmable, enabling a trade-off to be made between computing system performance and mean time required to perform a fault recovery.

5. A method of recovering from a fault occurrence within a computing system, said method comprising the steps of:

restoring a system state of the computing system to a previously recorded last checkpoint system state;

re-executing machine instructions, originally executed in an execution sequence, in a re-execution sequence, wherein the re-execution sequence is substantially similar to the execution sequence; and

,

repeating determining events, originally handled in an order in the execution sequence, in an identical order in the re-execution sequence.

6. A method as in claim 5, wherein determining events are repeated by re-handling recurring events and simulating non-recurring events.

7. A method as in claim 6, wherein machine instructions are re-executed in an order identical to an order in which machine instructions were originally executed.

8. A method as in claim 7, further comprising the step of:

repeating each determining event which was handled after recording the checkpoint system state and before the fault occurrence.

9. A method as in claim 8, further comprising the step of:

repeating each determining event at an instruction point in the re-execution sequence identical to an instruction point in the execution sequence at which each determining event originally was handled.

10. A method as in claim 9, wherein the step of repeating each determining event at an instruction point in the re-execution sequence identical to an instruction point in the execution sequence at which each determining event originally was handled is accomplished by counting machine instructions re-executed and using a previously logged count of machine instructions.

11. A method as in claim 10, further comprising the step of:

abandoning the re-execution sequence if an instruction point, identical to an instruction point at which the fault occurrence took place, has been reached.

12. A method as in claim 10, further comprising the step of:

abandoning the re-execution sequence if an instruction point, identical to an instruction point at which the fault occurrence was detected, has been reached.

13. A method as in claim 10, further comprising the step of:

abandoning the re-execution sequence if an instruction point has been reached at which determining event information indicates that a determining event should be repeated, but no such determining event has been initiated in the re-execution sequence.

14. A method as in claim 10, further comprising the step of:

abandoning the re-execution sequence if an instruction point has been reached at which a determining event has been initiated and should be handled, but no corresponding determining event information has been logged in the execution sequence.

15. A method as in claim 10, wherein the step of repeating each determining event at an instruction point in the re-execution sequence identical to an instruction point in the execution sequence at which each determining event originally was handled is accomplished by distinguishing but not immediately handling an early recurrence of an early recurring event.

16. A method as in claim 10, wherein the step of repeating each determining event at an instruction point in the re-execution sequence identical to an instruction point in the execution sequence at which each determining event originally was handled is accomplished by waiting for a recurrence of a late recurring event.

17. A method as in claim 10, wherein the step of repeating each determining event at an instruction point in the re-execution sequence identical to an instruction point in the execution sequence at which each determining event originally was handled is accomplished by using a previously logged non-recurring input to simulate a non-recurring event whose original handling involved an inputting of the non-recurring input.

18. A method as in claim 10, wherein the step of repeating each determining event at an instruction point in the re-execution sequence identical to an instruction point in the execution sequence at which each determining event originally was handled is accomplished by generating a trap at each instruction point in the re-execution sequence identical to an instruction point in the execution sequence at which each determining event originally was handled.

19. A method as in claim 10, further comprising the step of:

bypassing re-execution of a non-determining event.

20. A method as in claim 19, wherein the step of bypassing re-execution of a non-determining event is accomplished by not counting machine instructions executed as part of an original occurrence of the non-determining event.

21. A method of preparing for and recovering from a fault detected within a computing system, said method comprising the steps of:

counting machine instructions executed after recording a last checkpoint system state;

logging determining event information for a handling of a determining event;

restoring a system state of the computing system to a previously recorded last checkpoint system state;

re-executing machine instructions, originally executed in an execution sequence, in a re-execution sequence, wherein the re-execution sequence is substantially similar to the execution sequence; and

,

repeating determining events, originally handled in an order in the execution sequence, in an identical order in the re-execution sequence.

22. An apparatus useful for recovering from faults within a computing system, said apparatus comprising:

counting means for counting machine instructions executed;

reading means for reading a value of the counting means; and

,

logging means for logging determining event information.

23. An apparatus as in claim 22, further comprising:

handling means for handling an occurrence and re-handling a recurrence of a determining event;

simulating means for simulating a recurrence of a non-recurring event;

recording means for recording information necessary to restore the computing system to a checkpoint system state; and

,

restoring means for restoring the computing system to a previously recorded checkpoint system state.

24. An apparatus as in claim 23, further comprising:

trap generation means for generating a trap when the counting means has counted a first number of machine instructions; and

,

trap handling means for handling a trap generated when the counting means has counted the first number of machine instructions.

25. An apparatus as in claim 24, further comprising:

enabling means for enabling the counting means; and

, disabling means for disabling the counting means.

26. An apparatus as in claim 25, further comprising:

writing means for writing a new value into the counting means.

27. An apparatus for recovering from faults within a computing system comprising:

means for counting executed machine instruction; and

means for recording determining event information during an execution of a sequence of machine instructions, the recording means including means for recording a value of the counting means.

28. The apparatus as in claim 27 further comprising means for selectively enabling and disabling the counting means.

29. The apparatus as in claim 27 further comprising means for setting the counting means to a prescribed value.

30. The apparatus as in claim 27 further comprising means for generating a trap when the counting means contains a prescribed value.

31. The apparatus as in claim 30 further comprising means for processing the trap.

32. The apparatus as in claim 27 wherein the recording means includes means for recording information necessary for restoring the computing system to a prescribed state and further comprising:

means for restoring the computing system to the prescribed state.

33. The apparatus as in claim 32 further comprising means for generating a trap when the counting means contains a prescribed value.

34. The apparatus as in claim 33 further comprising means, responsive to the trap generating means, for processing a recurrence of a determining event during a re-execution of the sequence of machine instructions.

35. The apparatus as in claim 34 further comprising means, responsive to the trap generating means, for simulating a recurrence of a non-recurring event during a re-execution of the sequence of machine instructions.