Method for disk I/O transfer

• US 4,888,691 A
• Filed: 03/09/1988
• Issued: 12/19/1989
• Est. Priority Date: 03/09/1988
• Status: Expired due to Fees

• Current Assignee: Computervision Corp. (PTC Inc.)
• Original Assignee: Prime Computer, Inc. (DR Holdings, Inc. of Delaware)
• Inventors: Mednick, Elliot H., O'Brien, Kevin J., Sybel, Randall T., Waxman, David M., Spatara, Joseph M., George, Paul L.
• Primary Examiner(s): Zache, Raulfe B.
• Application Number:
  US07/166,515
• Time in Patent Office: 650 Days
• Field of Search: 364/200, 364/300, 364/900
• US Class Current: 714/15
• CPC Class Codes: G06F 2003/0697 device management, e.g. handlers, drivers, I/O schedulers
• CPC Class Codes: G06F 3/0601 Interfaces specially adapted for storage systems

Abstract

A disk control system offloads to the disk controller much of the overhead associated with disk operations and makes the CPU available for other work. A command block that fully specifies a user request for a disk operation is forwarded to the disk memory unit. The command block contains a unique identifier for tracking of user requests. User requests are executed by the disk memory unit in an order that is most efficient for the disk drive system. The status of a user request is communicated to the CPU via an interrupt and a status block containing the unique identifier. The status block indicates status conditions such as command read, completion and DMA channel request. The disk driver contains a work queue for user requests that have not been forwarded to the disk memory unit and a pending queue for user requests that are awaiting completion by the disk memory unit. By manipulation of the work queues and pending queues, the disk controller can be automatically reinitialized when an error occurs. The disk driver monitors the time that each user request is on the pending queue in order to detect failures of the disk memory unit.

Claims

1. In a computer system including a central processing unit (CPU) having a main memory, and a disk memory unit interconnected to the CPU, a method for executing user requests for work by the disk memory unit comprising the steps of:

for each user request received, placing a queue block representative of the user request on a work queue;

constructing a command block for each queue block placed on the work queue, said command block fully defining the user request;

forwarding each command block from the CPU to the disk memory unit for execution;

after each command block has been forwarded to the disk memory unit for execution, transferring the queue block that is representative of the user request from the work queue to a pending queue;

executing the user requests represented by each command block in an order determined by said disk memory unit; and

removing each queue block from the pending queue after the disk memory unit has completed the user request.

2. A method as defined in claim 1 wherein said user requests are received in a first order and further including the step of assigning a unique identifier to each user request, whereby said user requests are uniquely identified when they are executed in an order different from said first order.

3. A method as defined in claim 1 further including the steps of:

constructing a status block indicative of a completed user request, and

forwarding the status block to the CPU to initiate removal of the queue block that is representative of the completed user request from the pending queue.

4. A method as defined in claim 1 wherein said user requests are received in a first order and wherein the step of forwarding each command block from the CPU to the disk memory unit is performed in said first order.

5. A method as defined in claim 1 further including the steps of:

monitoring time that each queue block has been on the pending queue and generating an error condition when any queue block has been on the pending queue for longer than a predetermined time, and

reinitializing the disk memory unit and resuming operation in response to said error condition.

6. A method as defined in claim 5 wherein the step of monitoring time that each queue block has been on the pending queue includes the steps of assigning a maximum allowed completion time to each queue block on the pending queue, and generating said error condition when said maximum allowed completion time is exceeded for any queue block.

7. A method as defined in claim 6 wherein the work queue and the pending queue each have a head and a tail and wherein the step of reinitializing the disk memory unit and resuming operation includes the step of appending queue blocks and in the pending queue to the head of the work queue and initializing the pending queue to zero.

8. A method as defined in claim 1 further including the steps of:

constructing a status block indicative of a status of a user request that is pending in the disk memory unit,

identifying the queue block, the command block and the status block for each user request by a unique identifier, and

forwarding the status block from the disk memory unit to the CPU.

9. A method as defined in claim 8 further including the steps of:

determining that an unrecoverable error has occurred in the disk memory unit,

notifying the CPU via an interrupt and the status block that an unrecoverable error has occurred, and

automatically reinitializing the disk memory unit and resuming operation in response to said unrecoverable error.

10. A method as defined in claim 9 wherein the step of reinitializing the disk memory unit and resuming operation includes the steps of:

converting from an intelligent mode of operation of said disk memory unit to a degraded mode of operation,

downloading a disk control program for said intelligent mode of operation from said CPU to said disk memory unit, and

resuming operation in said intelligent mode of operation.

11. A method as defined in claim 10 wherein the work queue and the pending queue each have a head and a tail and wherein the step of reinitializing the disk memory unit and resuming operation further includes the steps of appending queue blocks in the pending queue to the head of the work queue and initializing the pending queue to zero.

12. A method as defined in claim 8 wherein the step of constructing a command block includes the step of writing command information to a specified location in main memory and wherein the step of forwarding each command block from the CPU to the disk memory unit includes the step of notifying the disk memory unit to read the command information from the specified location in main memory.

13. A method as defined in claim 12 wherein the step of constructing a status block includes the step of writing status information to a specified location in main memory and wherein the step of forwarding the status block from the disk memory unit to the CPU includes the step of notifying the CPU via an interrupt to read the status block from the specified location in main memory.

14. A method as defined in claim 13 including the step of notifying the CPU via an interrupt and the status block that a specified command block has been read from main memory.

15. A method as defined in claim 14 further including the steps of:

checking the work queue for an additional user request each time the CPU is notified that a specified command block has been read, and

when an additional user request is found on the work queue, notifying the CPU to process the user request and forward a command block representative of the additional user request to the disk memory unit, whereby user requests are quickly forwarded to the disk memory unit and the CPU is made available for other work.

16. A method as defined in claim 13 including the step of notifying the CPU via an interrupt and the status block that a specified user request has been completed.

17. A method as defined in claim 13 including the step of requesting the CPU via an interrupt and the status block to set up a data transfer channel when the disk memory unit is ready for transfer of data to or from the disk memory unit.

18. A method as defined in claim 17 including the step of notifying the CPU via an interrupt and the status block that data has been transferred to the disk memory unit.

19. A method as defined in claim 13 including the step of notifying the CPU via an interrupt and the status block that an error has occurred in the execution of a specified user request.

20. A method as defined in claim 13 including the step of notifying the CPU via an interrupt and the status block that an error unrelated to a user request has occurred.

21. A method as defined in claim 13 wherein the step of causing the CPU to read the status block includes the step of notifying the CPU by a vectored interrupt.

22. In a computer system including a central processing unit (CPU) having a main memory, and a disk memory unit interconnected to the CPU, a method for executing user requests for work by the disk memory unit comprising the steps of:

constructing for each user request a command block that fully specifies the user request;

assigning to each command block a unique identifier;

forwarding each command block from the CPU to the disk memory unit;

after forwarding each command block to the disk memory unit, resuming normal operation of the CPU;

p1 executing the user request specified by each command block in an order determined by said disk memory unit;

constructing a status block representative of a status of a user request that is pending in the disk memory unit, said status block containing the unique identifier; and

initiating an interrupt and forwarding the status block from the disk memory unit to the CPU.

23. A method as defined in claim 22 further including the steps of:

maintaining in the CPU a record of each command block forwarded to the disk memory unit and a time that the command block was forwarded to the disk memory unit,

deleting the record of each command block when the disk memory unit notifies the CPU that the user request was completed, and

generating an error condition when a period of time since any command block was forwarded to the disk memory unit exceeds a predetermined value.

24. A method as defined in claim 23 further including the step of reinitializing the disk memory unit and resuming operation in response to said error condition.

25. A method as defined in claim 24 wherein the step of maintaining a record of each command block forwarded to the disk memory unit includes the step of placing a queue block representative of the user request on a pending queue when the command block is forwarded to the disk memory unit, the queue block containing the time that the command block was forwarded to the disk memory unit.

26. A method as defined in claim 22 wherein the step of constructing a command block includes the step of writing command information to a specified location in main memory and wherein the step of forwarding each command block from the CPU to the disk memory unit includes the step of notifying the disk memory unit to read command information from the specified location in main memory.

27. A method as defined in claim 26 wherein the step of constructing a status block includes the step of writing status information to a specified location in main memory and wherein the step of forwarding the status block from the disk memory unit to the CPU includes the step of notifying the CPU via an interrupt to read the status block from the specified location in main memory.

28. A method as defined in claim 27 including the step of notifying the CPU via an interrupt and the status block that a specified command block has been read from main memory.

29. A method as defined in claim 27 including the step of notifying the CPU via an interrupt and the status block that a specified user request has been completed.

30. A method as defined in claim 27 including the step of requesting the CPU via an interrupt and the status block to set up a data transfer channel when the disk memory unit is ready for transfer of data to or from the disk memory unit.

31. A method as defined in claim 30 including the step of notifying the CPU via an interrupt and the status block that data has been transferred to the disk memory unit.

32. A method as defined in claim 27 including the step of notifying the CPU via an interrupt and the status block that an error has occurred in executing a specified user request.

33. A method as defined in claim 22 further including the step of notifying the CPU that a predetermined event unrelated to a user request has occurred in the disk memory unit.

34. A method as defined in claim 22 further including the steps of:

determining that an unrecoverable error has occurred in the disk memory unit,

notifying the CPU via an interrupt and the status block that an unrecoverable error has occurred, and

automatically reinitializing the disk memory unit and resuming operation in response to said unrecoverable error.

35. A method as defined in claim 34 wherein the step of reinitializing the disk memory unit and resuming operation includes the steps of:

converting from an intelligent mode of operation of said disk memory unit to a degraded mode of operation,

downloading a disk control program for said intelligent mode of operation from said CPU to said disk memory unit, and

resuming operation in said intelligent mode of operation.

37. A method as defined in claim 22 further including the steps of:

making an entry in a trace table for each command block,

making an entry in the trace table for each interrupt by the disk memory unit, and

making an entry in the trace table for each error condition, whereby the trace table contains a running record of disk memory unit operations that can be accessed for diagnostic purposes.

38. A method as defined in claim 22 wherein the step of notifying the CPU of the status of a specified user request includes the steps of:

writing the status block to a specified location in main memory,

setting a controller interrupt flag in the CPU, and

notifying a disk semaphore in the CPU that a status block has been written into main memory.

36. In a computer system including a central processing unit (CPU) and a disk memory unit interconnected to the CPU, a method for executing user requests for work by the disk memory unit comprising the steps of:

constructing for each user request a command block that fully specifies the user request, each command block being assigned a unique identifier number;

forwarding each command block from the CPU to the disk memory unit for execution in an order determined by said disk memory unit;

after forwarding each command block to the disk memory unit, resuming normal operation of the CPU;

requesting the CPU to set up a data transfer channel each time the disk memory unit is ready for transfer of data to or from the disk memory unit;

transferring data to or from the disk memory unit under control of the disk memory unit; and

notifying the CPU each time that a user request has been completed.

39. In a computer system including a central processing unit (CPU) having a main memory and a disk memory unit interconnected to the CPU, a method for executing user requests for work by the disk memory unit comprising the steps of:

for each user request received, constructing a command block fully defining the user request;

forwarding each command block from the CPU to the disk memory unit for execution;

executing the user requests represented by each command block in an order determined by said disk memory unit;

notifying the CPU that a specified user request has been completed;

maintaining in the CPU a record of each command block forwarded to the disk memory unit and a time that the command block was forwarded to the disk memory unit;

deleting the record of each command block when the disk memory unit notifies the CPU that the user request was completed;

generating an error condition when a period of time since any command block was forwarded to the disk memory exceeds a predetermined maximum value; and

automatically reinitializing the disk memory unit and resuming operation in response to said error condition.

40. A method as defined in claim 39 wherein the step of reinitializing the disk memory unit and resuming operation includes the steps of:

converting from an intelligent mode of operation of said disk memory unit to a degraded mode of operation,

downloading a disk control program for said intelligent mode of operation from said CPU to said disk memory unit, and

resuming operation in said intelligent mode of operation.

41. A method as defined in claim 40 wherein the step of maintaining a record of each command block forwarded to the disk memory unit includes the step of placing a queue block representative of the user request on a pending queue when the command block is forwarded to the disk memory unit, the queue block containing the time that the command block was forwarded to the disk memory unit.

42. In a computer system including a central processing unit (CPU) having a main memory and a disk memory unit interconnected to the CPU, a method for executing user requests for work by the disk memory unit comprising the steps of:

for each user request received, constructing a command block fully defining the user request;

forwarding each command block from the CPU to the disk memory unit for execution;

executing the user requests represented by each command block in an order determined by said disk memory unit;

notifying the CPU that a specified user request has been completed;

determining when an unrecoverable error has occurred in the disk memory unit;

notifying the CPU that an unrecoverable error has occurred; and

automatically reinitializing the disk memory unit and resuming operation in response to said unrecoverable error.

43. A method as defined in claim 42 wherein the step of reinitializing the disk memory unit and resuming operation includes the steps of:

converting from an intelligent mode of operation of said disk memory unit to a degraded mode of operation,

downloading a disk control program for said intelligent mode of operation from said CPU to said disk memory unit, and

resuming operation in said intelligent mode of operation.

44. In a computer system including a central processing unit (CPU) having a main memory, and a disk memory unit interconnected to the CPU, a method for executing user requests for work by the disk memory unit comprising the steps of:

for each user request, forwarding a command block that fully specifies the user request from the CPU to the disk memory unit, said command block containing a unique identifier;

resuming normal operation of the CPU until notified by the disk memory unit of a status of a specified user request;

executing the user request specified by the command block at a time determined by the disk memory unit; and

notifying the CPU of the status of the specified user request via an interrupt and a status block containing said unique identifier.

45. A method as defined in claim 44 further including the steps of:

notifying the CPU via an interrupt and the status block containing said unique identifier to set up a data transfer channel when the disk memory unit is ready for transfer and data, and

transferring data to or from the disk memory unit under control of the disk memory unit.

46. A method as defined in claim 45 including the step of notifying the CPU via an interrupt and the status block that the specified command block has been read from main memory.

47. A method as defined in claim 45 including the step of notifying the CPU via an interrupt and the status block that the specified user request has been completed.

48. A method as defined in claim 45 including the step of notifying the CPU via an interrupt and the status block that data has been transferred to the disk memory unit.

49. A method as defined in claim 45 including the step of notifying the CPU via an interrupt and the status block that an error has occurred in executing the specified user request.

50. A method as defined in claim 45 further including the step of notifying the CPU that a predetermined event unrelated to a user request has occurred in the disk memory unit.

51. A method as defined in claim 44 wherein the step of forwarding a command block includes the steps of writing command information to a specified location in main memory and notifying the disk memory unit to read the command information from the specified location in main memory.

52. A method as defined in claim 44 wherein the step of notifying the CPU includes the steps of writing status information to a specified location in main memory and notifying the CPU via an interrupt to read the status block from the specified location in main memory.

53. In a computer system including a central processing unit (CPU) having a main memory, and a disk memory unit interconnected to the CPU, a method for executing user requests for work by the disk memory unit comprising the steps of:

notifying the disk memory unit that a command block representative of a user request is available in the CPU;

accessing the command block for execution by the disk memory unit; and

notifying the CPU when the user request represented by the command block has been completed.