Shared resources in a data processing apparatus for executing a plurality of threads

• US 10,528,350 B2
• Filed: 07/28/2015
• Issued: 01/07/2020
• Est. Priority Date: 09/08/2014
• Status: Active Grant

• Current Assignee: Arm Limited (SoftBank Group Corp.)
• Original Assignee: Arm Limited (SoftBank Group Corp.)
• Inventors: Holm, Rune, Mansell, David Hennah
• Primary Examiner(s): Kim, Sisley N
• Application Number:
  US15/505,714
• Publication Number:
  US 20170286107A1
• Time in Patent Office: 1,624 Days
• CPC Class Codes: G06F 9/3009 Thread control instructions
• CPC Class Codes: G06F 9/524 Deadlock detection or avoidance
• CPC Class Codes: G06F 9/526 Mutual exclusion algorithms

Abstract

A data processing apparatus (100) executes threads and includes a general program counter (PC) (120) identifying an instruction to be executed for at least a subset of the threads. Each thread has a thread PC (184). The subset of threads has at least one lock parameter (188, 500-504) for tracking exclusive access to shared resources. In response to a first instruction executed for a thread, the processor (160) modifies the at least one lock parameter (188), (500-504) to indicate that the thread has gained exclusive access to the shared resource. In response to a second instruction, the processor modifies the at least one lock parameter (188, 500-504) to indicate that the thread no longer has exclusive access. A selector (110) selects one of the subset of threads based on the at least one lock parameter (188, 500-504) and sets the general PC (120) to the thread PC (184) of the selected thread.

Claims

1. A data processing apparatus configured to execute a plurality of threads, the data processing apparatus comprising:

a general program counter to identify one instruction in a stream of instructions to be executed for at least a subset of the plurality of threads, wherein each thread in the subset has an associated thread program counter to identify one instruction in the stream of instructions,

selector circuitry configured to select a selected thread of the subset of threads and to set the general program counter to the thread program counter associated with the selected thread; and

a processor configured to execute an instruction identified by the general program counter for one or more of the subset of threads including the selected thread,

wherein the subset of threads is associated with at least one lock parameter for tracking which of the subset of threads has exclusive access to a shared resource;

wherein the processor is configured to modify the at least one lock parameter to indicate that a thread has gained exclusive access to the shared resource in response to a first instruction executed for that thread, and to modify the at least one lock parameter associated with the thread to indicate that the thread no longer has exclusive access to the shared resource in response to a second instruction executed for that thread; and

wherein the selector circuitry is configured to select the selected thread based on the at least one lock parameter.

2. A data processing apparatus according to claim 1,

wherein the processor is configured to grant a thread exclusive access to the shared resource in response to a locking sequence comprising at least one locking instruction executed for that thread.

3. A data processing apparatus according to claim 2,

wherein the locking sequence does not comprise the first instruction.

4. A data processing apparatus according to claim 1,

wherein the processor is configured to relinquish exclusive access to the shared resource for a thread in response to an unlocking sequence comprising at least one unlocking instruction executed for that thread.

5. A data processing apparatus according to claim 4,

wherein the unlocking sequence does not comprise the second instruction.

6. A data processing apparatus according to claim 1,

wherein the selector circuitry is configured to select, as said selected thread, a first thread indicated by the at least one lock parameter as having exclusive access to the shared resource in preference to a second thread indicated by the at least one lock parameter as not having exclusive access to the shared resource.

7. A data processing apparatus according to claim 1,

wherein the selector circuitry is configured to select said selected thread based on a function call depth parameter associated with each thread of the subset of threads.

8. A data processing apparatus according to claim 1,

wherein the selector circuitry is configured to select said selected thread based on the thread program counter associated with each thread of the subset of threads.

9. A data processing apparatus according to claim 1,

wherein each thread of the subset has a corresponding lock parameter, and the processor is configured to modify the lock parameter corresponding to a thread to indicate that the thread has gained exclusive access to the shared resource in response to a first instruction executed for that thread, and to modify the lock parameter corresponding to the thread to indicate that the thread no longer has exclusive access to the shared resource in response to a second instruction executed for that thread.

10. A data processing apparatus according to claim 9,

wherein the processor is configured to permit a first thread to have exclusive access to a first shared resource while a second thread has exclusive access to a second shared resource.

11. A data processing apparatus according to claim 9,

wherein the selector circuitry is configured to select a first set of threads based on the corresponding lock parameter of each thread of the subset of threads;

wherein the selector circuitry is configured to select, from among the first set of threads, a second set of threads based on a function call depth parameter;

wherein the selector circuitry is configured to select said selected thread from among said second set of threads, based on the thread program counters associated with the threads in the second set of threads.

12. A data processing apparatus according to claim 1,

wherein the at least one lock parameter comprises a shared lock parameter shared between the subset of threads and a lock owner parameter indicating which of the subset of threads has exclusive access to a shared resource.

13. A data processing apparatus according to claim 12,

wherein the shared lock parameter comprises a lock flag indicating whether any of the subset of threads have exclusive access to a shared resource.

14. A data processing apparatus according to claim 12,

wherein the shared lock parameter comprises a lock count indicating a number of resources to which the thread indicated by the lock owner parameter has exclusive access.

15. A data processing apparatus according to claim 12,

wherein if a first thread has exclusive access to a first shared resource, the processor is configured to prevent a second thread having exclusive access to a second shared resource.

16. A data processing apparatus according to claim 12,

wherein if the at least one lock parameter indicates that a thread has exclusive access to a shared resource, then the selector is configured to select the thread indicated by the lock owner parameter as said selected thread;

and

if the at least one lock parameter indicates that no thread has exclusive access to a shared resource, then the selector is configured to select, from among the subset of threads, a further subset of threads based on a function call depth parameter, and to select said selected thread from among said further subset of threads based on the thread program counters associated with the threads in said further subset of threads.

17. A data processing apparatus according to claim 1,

wherein the processor is configured to increment the at least one lock parameter in response to the first instruction;

and

wherein the processor is configured to decrement the at least one lock parameter in response to the second instruction.

18. A data processing apparatus according to claim 17,

wherein the at least one lock parameter indicates the number of shared resources to which a thread has exclusive access.

19. A data processing apparatus according to claim 1,

wherein the one or more of the subset of threads comprise the threads having one or more associated thread parameters that match corresponding thread parameters of said selected thread, the one or more associated thread parameters including at least the thread program counter.

20. A data processing apparatus according to claim 1,

wherein the stream of instructions comprises micro-operations.

21. A data processing apparatus according to claim 1,

wherein the first instruction and the second instruction are encoded as a NOOP instruction in at least one predefined instruction set.

22. A data processing apparatus according to claim 1,

wherein the processor is configured to perform a further operation in response to at least one of the first and second instructions.

23. A data processing apparatus according to claim 22,

wherein the further operation is one or more of:

an operation that is part of a locking sequence for gaining exclusive access to the shared resource, an operation that is part of an unlocking sequence for relinquishing exclusive access to the shared resource, an operation for accessing the shared resource, an operation for determining whether the thread has exclusive access to the shared resource, and a compare and exchange operation.

24. A data processing apparatus according to claim 1,

wherein the shared resource comprises data in a memory.

25. A data processing apparatus according to claim 24,

wherein the processor is configured to issue a memory barrier operation to the memory prior to relinquishing exclusive access to the shared resource.

26. A data processing method for executing a stream of instructions for a subset of a plurality of threads, wherein each thread in the subset has an associated thread program counter for identifying one instruction in the stream of instructions, the data processing method comprising the steps:

selecting a selected thread of the subset of threads, and setting a general program counter to the thread program counter associated with the selected thread, wherein the general program counter identifies one instruction in a stream of instructions to be executed for the subset of threads; and

executing an instruction identified by the general program counter for one or more of the subset of threads including the selected thread,

wherein the subset of threads is associated with at least one lock parameter for tracking which of the subset of threads has exclusive access to a shared resource;

wherein the at least one lock parameter is modified to indicate that a thread has gained exclusive access to the shared resource in response to a first instruction executed for that thread, and is modified to indicate that the thread no longer has exclusive access to the shared resource in response to a second instruction executed for that thread; and

wherein the selected thread is selected based on the at least one lock parameter.

27. A data processing apparatus configured to execute a plurality of threads, the data processing apparatus comprising:

a general program counter means for identifying one instruction in a stream of instructions to be executed for at least a subset of the plurality of threads, wherein each thread in the subset has an associated thread program counter means for identifying one instruction in the stream of instructions,

a selection means for selecting a selected thread of the subset of threads and for setting the general program counter means to the thread program counter means associated with the selected thread; and

a processor means for executing an instruction identified by the general program counter means for one or more of the subset of threads including the selected thread,

wherein the subset of threads is associated with at least one lock parameter for tracking which of the subset of threads has exclusive access to a shared resource;

wherein the processor means is for modifying the at least one lock parameter to indicate that a thread has gained exclusive access to the shared resource in response to a first instruction executed for that thread, and for modifying the at least one lock parameter to indicate that the thread no longer has exclusive access to the shared resource in response to a second instruction executed for that thread; and

wherein the selection means is for selecting the selected thread based on the at least one lock parameter.