Software emulation of directed exceptions in a multithreading processor

• US 20060161421A1
• Filed: 12/20/2005
• Published: 07/20/2006
• Est. Priority Date: 08/28/2003
• Status: Active Grant

• Current Assignee: MIPS Tech, LLC
• Original Assignee: MIPS Technologies Incorporated (Wave Computing, Inc.)
• Inventors: Kissell, Kevin D.
• Granted Patent: US 7,849,297 B2
• Time in Patent Office: Days
• US Class Current: 703/26
• CPC Class Codes: G06F 9/4812 by interrupt, e.g. masked
• CPC Class Codes: G06F 9/4881 Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues

Abstract

A multithreading microprocessor has a plurality of thread contexts (TCs) each including sufficient state, such as general purpose registers and program counter, to execute a separate thread of execution as one of a plurality of symmetric processors controlled by a multiprocessor operating system. However, the microprocessor hardware does not support the ability for one TC to direct an exception to another TC, i.e., to specify to which of the other TCs the exception is directed. A first thread running on a first TC of the operating system executes architected instructions to halt a second thread (either user or kernel thread) running on a second TC, save state of the second TC, write the second TC state to emulate an exception—including writing a restart register with the address of an exception handler, and unhalt the second TC to execute the exception hander.

Claims

1. A method for performing software emulation of a directed exception in a multithreading processor running a multiprocessor operating system, the processor having a plurality of thread contexts each comprising storage elements that describe a state of execution of a respective plurality of threads running on the respective plurality of thread contexts, the method comprising:

writing, by a first thread running on a first thread context, to a second thread context to cause a second thread running on the second thread context to stop running;

writing, by the first thread, to the second thread context an address of an exception handler of the operating system; and

writing, by the first thread, to the second thread context to cause the exception handler to commence running at the address on the second thread context.

2. The method as recited in claim 1, further comprising:

processing, by the exception handler, an inter-processor interrupt message, after said writing to cause the exception handler to commence running.

3. The method as recited in claim 2, further comprising:

writing, by the first thread, the inter-processor interrupt message to a memory coupled to the multithreading processor, prior to said writing to cause the exception handler to commence running.

4. The method as recited in claim 2, further comprising:

saving, by the exception handler, general purpose registers of the second thread context to a memory coupled to the multithreading processor, prior to said processing the inter-processor interrupt message.

5. The method as recited in claim 2, further comprising:

setting, by the exception handler, an execution privilege level of the second thread context to the operating system privilege level, prior to said processing the inter-processor interrupt message.

6. The method as recited in claim 2, further comprising:

disabling, by the exception handler, interrupts from being serviced by the second thread context, prior to said processing the inter-processor interrupt message.

7. The method as recited in claim 1, further comprising:

writing, by the first thread, to the second thread context to set an execution privilege level of the exception handler to the operating system privilege level, prior to said writing to cause the exception handler to commence running.

8. The method as recited in claim 1, further comprising:

writing, by the first thread, to the second thread context to disable interrupts from being serviced by the second thread context, prior to said writing to cause the exception handler to commence running.

9. The method as recited in claim 1, further comprising:

reading, by the first thread, from the second thread context a restart address of the second thread.

10. The method as recited in claim 9, wherein the restart address is an address of an instruction of the second thread at which the second thread recommences running on the second thread context after the exception handler returns.

11. The method as recited in claim 9, wherein said restart address is an address of an instruction of the second thread interrupted by said writing to cause the second thread to stop running.

12. The method as recited in claim 9, wherein said restart address is an address of a branch instruction immediately preceding an instruction of the second thread interrupted by said writing to cause the second thread to stop running, if the interrupted instruction is in a branch delay slot of the branch instruction.

13. The method as recited in claim 9, further comprising:

writing, by the first thread, the restart address to a memory coupled to the multithreading processor.

14. The method as recited in claim 13, wherein said writing the restart address to a memory comprises:

writing, by the first thread, the restart address to a stack frame in the memory allocated by the first thread for the exception handler.

15. The method as recited in claim 9, further comprising:

returning to the restart address, after the exception handler completes.

16. The method as recited in claim 15, further comprising:

reading, by the exception handler, the restart address from the memory, after said writing to cause the exception handler to commence running; and

writing, by the exception handler, the restart address to a MIPS Privileged Resource Architecture (PRA) Coprocessor 0 Exception Program Counter (EPC) register of a MIPS Multithreading Application-Specific Extension Virtual Processing Element (VPE) comprising the second thread context, prior to said returning to the restart address.

17. The method as recited in claim 16, wherein the VPE comprises at least the first and second thread contexts.

18. The method as recited in claim 17, wherein the VPE comprises an exception domain for the first and second thread contexts.

19. The method as recited in claim 17, wherein the VPE comprises one or more interrupt requests, wherein the one or more interrupt requests are non-directable specifically to one of the first and second thread contexts.

20. The method as recited in claim 19, wherein the one or more interrupt requests comprise one or more hardware interrupt inputs.

21. The method as recited in claim 19, wherein the one or more interrupt requests comprise one or more software interrupt requests.

22. The method as recited in claim 17, wherein the VPE comprises one or more interrupt requests, wherein the one or more interrupt requests are shared by the first and second thread contexts.

23. The method as recited in claim 22, wherein the VPE is configured to select one of the first and second thread contexts to handle one of the interrupt requests during a first instance and to select the other of the first and second thread contexts to handle another of the interrupt requests during a second instance.

24. The method as recited in claim 17, wherein the VPE comprises one or more interrupt requests, wherein the one or more interrupt requests are independent of a number of the first and second thread contexts and other of the plurality of thread contexts of the multithreading processor comprised in the VPE.

25. The method as recited in claim 1, further comprising:

setting, by the exception handler, a MIPS Privileged Resource Architecture (PRA) Coprocessor 0 Status register EXL flag, after said writing to cause the exception handler to commence running.

26. The method as recited in claim 25, wherein the Coprocessor 0 Status register EXL flag being set causes a MIPS Multithreading Application-Specific Extension (MT ASE) Virtual Processing Element (VPE) comprising the first and second thread contexts to have the operating system execution privilege level and to have interrupts disabled.

27. The method as recited in claim 25, further comprising:

reading, by the first thread, a MT ASE TCStatus register of the second thread context, prior to said writing to cause the exception handler to commence running; and

restoring, by the exception handler, the TCStatus register of the second thread context, after said setting the EXL flag.

28. The method as recited in claim 27, further comprising:

setting, by the exception handler, an interrupt exempt (IXMT) bit of the TCStatus register of the second thread context, after said restoring the TCStatus register of the second thread context.

29. The method as recited in claim 27, further comprising:

setting, by the exception handler, to kernel mode a thread kernel/supervisor/user (TKSU) state field of the TCStatus register of the second thread context, after said restoring the TCStatus register of the second thread context.

30. The method as recited in claim 1, further comprising:

reading, by the first thread, from the second thread context an execution privilege level of the second thread, after said writing to cause the second thread context to stop running.

31. The method as recited in claim 30, wherein said reading the execution privilege level of the second thread comprises the first thread executing an instruction of an architected instruction set of the multithreading processor.

32. The method as recited in claim 31, wherein the architected instruction comprises a MIPS Multithreading Application-Specific Extension (MT ASE) MFTR instruction.

33. The method as recited in claim 30, further comprising:

writing, by the first thread, the execution privilege level of the second thread, to a memory coupled to the multithreading processor, after said reading the execution privilege level.

34. The method as recited in claim 33, further comprising:

restoring the execution privilege level from the memory, upon completion of the exception handler and return to running the second thread on the second thread context.

35. The method as recited in claim 1, wherein said writing the address of the exception handler comprises the first thread executing an instruction of an architected instruction set of the multithreading processor.

36. The method as recited in claim 35, wherein the architected instruction comprises a MIPS Multithreading Application-Specific Extension (MT ASE) MTTR instruction.

37. The method as recited in claim 35, wherein said architected instruction is only executable at the operating system execution privilege level.

38. The method as recited in claim 1, wherein the exception handler is configured to handle an architected exception of the multithreading processor.

39. A multiprocessor computer system, comprising:

a microprocessor, comprising at least first and second thread contexts, each comprising;

a restart register; and

a control register, for controlling whether said thread context is halted from execution;

a memory, coupled to said microprocessor, for storing a first thread of execution for execution on said first thread context; and

a multiprocessing operating system (MP OS), stored in said memory, comprising a second thread of execution for execution on said second thread context configured;

to write to said control register of said first thread context to halt execution of said first thread;

to write to said restart register of said first thread context an address of an exception handler of said MP OS; and

to write to said control register of said first thread context to commence execution of said exception handler on said first thread context at said address.

40. The system as recited in claim 39, wherein said first thread comprises a user thread executing on said first thread context at a user execution privilege level.

41. The system as recited in claim 39, wherein said MP OS comprises said first thread, and said first thread executes on said first thread context at an execution privilege level of said MP OS.

42. The system as recited in claim 39, wherein said MP OS is configured to program each of said at least first and second thread contexts as a separate processor of the multiprocessor system.

43. The system as recited in claim 39, wherein each of said at least first and second thread contexts further comprises:

a status register, for storing an execution privilege level of said thread.

44. The system as recited in claim 43, wherein said second thread is further configured:

to write to said status register of said second thread context for setting said second thread context to an execution privilege level of said MP OS, after writing to said control register of said first thread context to halt execution of said first thread, and prior to writing to said control register of said first thread context to commence execution of said exception handler.

45. The system as recited in claim 44, wherein said second thread is further configured:

to read a contents of said status register of said second thread context, after writing to said control register of said first thread context to halt execution of said first thread; and

to write said contents to said memory, prior to writing to said control register of said first thread context to commence execution of said exception handler.

46. The system as recited in claim 45, wherein said second thread executing on said second thread context reads said contents of said status register of said first thread context via an instruction included in an instruction set architecture of said microprocessor.

47. The system as recited in claim 46, wherein said instruction included in said instruction set architecture of said microprocessor comprises a MIPS instruction set architecture (ISA) MFTR instruction.

48. The system as recited in claim 39, wherein each of said at least first and second thread contexts further comprises:

a status register, for storing an interrupt exemption control indicator of said thread.

49. The system as recited in claim 48, wherein said second thread is further configured:

to write to said status register of said second thread context for exempting said second thread context from servicing interrupts, after writing to said control register of said first thread context to halt execution of said first thread, and prior to writing to said control register of said first thread context to commence execution of said exception handler.

50. The system as recited in claim 39, wherein said second thread executing on said second thread context writes to said first thread context via an instruction included in an instruction set architecture of said microprocessor.

51. The system as recited in claim 50, wherein said instruction included in said instruction set architecture of said microprocessor comprises a MIPS instruction set architecture (ISA) MTTR instruction.

52. The system as recited in claim 39, wherein said microprocessor further comprises:

an input, for indicating a request to interrupt execution by the microprocessor and to service said request, wherein said request is non-specific about which of said at least first and second thread contexts is to be interrupted and is to service said request.

53. The system as recited in claim 52, wherein said microprocessor is configured to select one of said at least first and second thread contexts to be interrupted and to service said request, wherein said microprocessor does not select a same one of said at least first and second thread contexts to be interrupted and to service said request every time said request is indicated on said input.

54. The system as recited in claim 53, wherein said microprocessor is configured to suspend execution of all but said one of said at least first and second thread contexts selected to service said interrupt request.

55. The system as recited in claim 54, wherein said one of said at least first and second thread contexts selected to service said interrupt request is configured:

to write to said control register of a non-selected one of said at least first and second thread contexts to halt execution thereof;

to write to said restart register of said non-selected one of said at least first and second thread contexts said address of said exception handler of said MP OS; and

to write to said control register of said non-selected one of said at least first and second thread contexts to commence execution of said exception handler on said non-selected one of said at least first and second thread contexts at said address.

56. The system as recited in claim 39, wherein said microprocessor further comprises:

a plurality of thread contexts, in addition to said at least first and second thread contexts;

a plurality of virtual processing elements (VPEs), wherein a first of said plurality of VPEs comprises said at least first and second thread contexts, and wherein each of the other of said plurality of VPEs comprises a mutually exclusive set of at least one of said plurality of thread contexts;

an input, for indicating a request specifying, from among said plurality of VPEs, said one of said plurality of VPEs comprising said at least first and second thread contexts to interrupt execution thereof and to service said request, rather than specifying one of said at least first and second thread contexts to interrupt execution thereof and to service said interrupt request.

57. The system as recited in claim 56, wherein said first VPE is configured to select one of said at least first and second thread contexts to interrupt execution thereof and to service said request, wherein said first VPE does not select a same one of said at least first and second thread contexts to interrupt execution thereof and to service said request every time said request is indicated on said input.

58. The system as recited in claim 57, wherein each of said VPEs further comprises:

a exception program counter register, for storing an address of an instruction to which said VPE transfers control in response to execution of a return from exception instruction.

59. The system as recited in claim 58, wherein said second thread is configured:

to read a restart address from said restart register of said first thread context, after writing to said control register of said first thread context to halt execution of said first thread; and

to write said restart address to said memory.

60. The system as recited in claim 59, wherein said exception handler is configured:

to read said restart address from said memory; and

to write said restart address to said exception program counter.

61. The system as recited in claim 57, wherein each of said VPEs further comprises:

a status register, for storing an exception level indicator indicating whether said VPE is currently servicing an exception.

62. The system as recited in claim 61, wherein said exception handler is configured:

to write said status register to indicate said exception level indicator to indicate said VPE is currently servicing an exception.

63. The system as recited in claim 39, wherein said second thread is configured:

to read a contents of said restart register of said first thread context, after writing to said control register of said first thread context to halt execution of said first thread; and

to write said contents to said memory.

64. The system as recited in claim 39, wherein each of said at least first and second thread contexts further comprises:

a set of general purpose registers;

wherein said exception handler is further configured;

to read a contents of said general purpose registers; and

to write said contents to said memory.

65. The system as recited in claim 39, wherein a computer program product comprising a computer usable medium having computer readable program code causes the system, wherein said computer program product is for use with a computing device.

66. The system as recited in claim 39, wherein a computer data signal embodied in a transmission medium comprising computer-readable program code provides the system.

67. A computer program product for use with a computing device, the computer program product comprising:

a computer usable medium, having computer readable program code embodied in said medium, for causing a method for performing software emulation of a directed exception in a multithreading processor running a multiprocessor operating system, the processor having a plurality of thread contexts each comprising storage elements that describe a state of execution of a respective plurality of threads running on the respective plurality of thread contexts, said computer readable program code comprising;

first program code for providing a step of writing, by a first thread running on a first thread context, to a second thread context to cause a second thread running on the second thread context to stop running;

second program code for providing a step of writing, by the first thread, to the second thread context an address of an exception handler of the operating system; and

third program code for providing a step of writing, by the first thread, to the second thread context to cause the exception handler to commence running at the address on the second thread context.

68. The computer program product of claim 67, wherein said computer readable program code further comprises:

fourth program code for providing a step of writing, by the first thread, to the second thread context to set an execution privilege level of the exception handler to the operating system privilege level, prior to said writing to cause the exception handler to commence running.

69. The computer program product of claim 67, wherein said computer readable program code further comprises:

fourth program code for providing a step of writing, by the first thread, to the second thread context to disable interrupts from being serviced by the second thread context, prior to said writing to cause the exception handler to commence running.

70. A method for providing operating system software for performing emulation of a directed exception in a multithreading processor running a multiprocessor operating system, the processor having a plurality of thread contexts each comprising storage elements that describe a state of execution of a respective plurality of threads running on the respective plurality of thread contexts, the method comprising:

providing computer-readable program code describing the operating system software, the program code comprising;

first program code for providing a step of writing, by a first thread running on a first thread context, to a second thread context to cause a second thread running on the second thread context to stop running;

second program code for providing a step of writing, by the first thread, to the second thread context an address of an exception handler of the operating system; and

third program code for providing a step of writing, by the first thread, to the second thread context to cause the exception handler to commence running at the address on the second thread context; and

transmitting the computer-readable program code as a computer data signal on a network.