Achieving low grace period latencies despite energy efficiency

• US 9,389,925 B2
• Filed: 12/03/2013
• Issued: 07/12/2016
• Est. Priority Date: 12/03/2013
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: McKenney, Paul E.
• Primary Examiner(s): Cao, Chun
• Assistant Examiner(s): Derose, Volvick
• Application Number:
  US14/095,541
• Publication Number:
  US 20150153817A1
• Time in Patent Office: 952 Days
• Field of Search: 713/323
• US Class Current: 1/1
• CPC Class Codes: G06F 1/3203 Power management, i.e. event-based initiation of a power-saving mode
• CPC Class Codes: G06F 9/5094 where the allocation takes into account power or heat criteria power management in computers in general G06F1/3203; thermal management in computers in general G06F1/206
• CPC Class Codes: Y02D 10/00 Energy efficient computing, e.g. low power processors, power management or thermal management

Abstract

A technique for achieving low grace-period latencies in an energy efficient environment in which processors with Read-Copy Update (RCU) callbacks are allowed to enter low power states. In an example embodiment, for each processor that has RCU callbacks, different grace period numbers are assigned to different groups of the processor'"'"'s RCU callbacks. New grace periods are periodically started and old grace periods are periodically ended. As old grace periods end, groups of RCU callbacks having corresponding assigned grace period numbers are invoked.

Claims

1. In a computer system having one or more processors operatively coupled to one or more memory devices, a method for achieving low grace-period latencies in an energy efficient environment in which processors with Read-Copy Update (RCU) callbacks are allowed to enter low power states, comprising:

periodically generating RCU callbacks as part of shared data element update Operations, said callbacks representing requests for deferred destruction of shared data elements following an RCU grace period;

maintaining said RCU callbacks on a per-processor basis;

for each of said one or more processors that maintains RCU callbacks, assigning different RCU grace period numbers to different groups of said processor'"'"'s RCU callbacks;

periodically starting new RCU grace periods and ending old RCU grace periods; and

as said old RCU grace periods end, invoking said groups of RCU callbacks having corresponding assigned RCU grace period numbers.

2. The method of claim 1, wherein said groups of RCU callbacks are maintained on sublists of RCU callback lists associated with said one or more processors, and wherein said grace period number assigning comprises a callback accelerate operation wherein said RCU grace period numbers are assigned to newly arrived groups of RCU callbacks, and wherein said newly arrived groups of RCU callbacks are placed on selected sublists according to said assigned RCU grace period numbers.

3. The method of claim 2, wherein said RCU callback lists include a done-sublist for RCU callbacks that are ready to be invoked, and wherein said method further includes performing a callback advance operation for each of said one or more processors that is not offline or designated as a no-callbacks processor, and having RCU callbacks that are waiting for a not-yet-completed RCU grace period, said callback advance operation including advancing to said done-sublist any of said groups of RCU callbacks whose assigned RCU grace period numbers correspond to RCU grace periods that have ended, said callback advance operation further including performing said callback accelerate operation.

4. The method of claim 3, wherein said RCU callback lists each comprise plural sublists, and wherein said callback advance operation further includes adjusting sublists other than said done-sublist to maintain said RCU callback lists in a consistent state without holes therein.

5. The method of claim 3, wherein said callback advance operation is attempted as said one or more processors prepare for entry into a low power state, said callback accelerate operation is attempted on entry into said low power state, and said callback advance operation is attempted again on exit from said low power state.

6. The method of claim 1, wherein said grace period number assigning includes recording future RCU grace periods needed by said one or more processors so that said future RCU grace periods can be initiated without waking said one or more processors if they are in a low power state.

7. The method of claim 6, further including using said future grace period recording to offload callback invocation from specially designated ones of said one or more processors.

8. A multiprocessor system, comprising:

two or more processors;

a memory coupled to said processors, said memory including a computer useable medium tangibly embodying at least one program of instructions executable by said processors to implement a read-copy update (RCU) subsystem and to perform operations for achieving low grace-period latencies in an energy efficient environment in which processors with Read-Copy Update (RCU) callbacks are allowed to enter low power states, said operations comprising;

periodically generating RCU callbacks as part of shared data element update Operations, said callbacks representing requests for deferred destruction of shared data elements following an RCU grace period;

maintaining said RCU callbacks on a per-processor basis;

for each of said one or more processors that maintains said RCU callbacks, assigning different RCU grace period numbers to different groups of said processor'"'"'s RCU callbacks;

periodically starting new RCU grace periods and ending old RCU grace periods; and

as said old RCU grace periods end, invoking said groups of RCU callbacks having corresponding assigned RCU grace period numbers.

9. The system of claim 8, wherein said groups of RCU callbacks are maintained on sublists of RCU callback lists associated with said one or more processors, and wherein said grace period number assigning comprises a callback accelerate operation wherein said RCU grace period numbers are assigned to newly arrived groups of RCU callbacks, and wherein said newly arrived groups of RCU callbacks are placed on selected sublists according to said assigned RCU grace period numbers.

10. The system of claim 9, wherein said RCU callback lists include a done-sublist for RCU callbacks that are ready to be invoked, and wherein said method further includes performing a callback advance operation for each of said one or more processors that is not offline or designated as a no-callbacks processor, and having RCU callbacks that are waiting for a not-yet-completed RCU grace period, said callback advance operation including advancing to said done-sublist any of said groups of RCU callbacks whose assigned RCU grace period numbers correspond to RCU grace periods that have ended, said callback advance operation further including performing said callback accelerate operation.

11. The system of claim 10, wherein said RCU callback lists each comprise plural sublists, and wherein said callback advance operation further includes adjusting sublists other than said done-sublist to maintain said RCU callback lists in a consistent state without holes therein.

12. The system of claim 10, wherein said callback advance operation is attempted as said one or more processors prepare for entry into a low power state, said callback accelerate operation is attempted on entry into said low power state, and said callback advance operation is attempted again on exit from said low power state.

13. The system of claim 8, wherein said grace period number assigning includes recording future RCU grace periods needed by said one or more processors so that said future RCU grace periods can be initiated without waking said one or more processors if they are in a low power state.

14. The system of claim 13, wherein said operations further include using said future grace period recording to offload callback invocation from specially designated ones of said one or more processors.

15. A computer program product, comprising:

one or more machine-readable non-transitory data storage media;

program instructions provided by said one or more data storage media for programming a multiprocessor data processing platform to implement a read-copy update (RCU) subsystem and to perform operations for achieving low grace-period latencies in an energy efficient environment in which processors with Read-Copy Update (RCU) callbacks are allowed to enter low power states, said operations comprising;

periodically generating RCU callbacks as part of shared data element update Operations, said callbacks representing requests for deferred destruction of shared data elements following an RCU grace period;

maintaining said RCU callbacks on a per-processor basis;

for each of said one or more processors that maintains said RCU callbacks, assigning different RCU grace period numbers to different groups of said processor'"'"'s RCU callbacks;

periodically starting new RCU grace periods and ending old RCU grace periods; and

as said old RCU grace periods end, invoking said groups of RCU callbacks having corresponding assigned RCU grace period numbers.

16. The computer program product of claim 15, wherein said groups of RCU callbacks are maintained on sublists of RCU callback lists associated with said one or more processors, and wherein said grace period number assigning comprises a callback accelerate operation wherein said RCU grace period numbers are assigned to newly arrived groups of RCU callbacks, and wherein said newly arrived groups of RCU callbacks are placed on selected sublists according to said assigned RCU grace period numbers.

17. The computer program product of claim 16, wherein said RCU callback lists include a done-sublist for RCU callbacks that are ready to be invoked, and wherein said method further includes performing a callback advance operation for each of said one or more processors that is not offline or designated as a no-callbacks processor, and having RCU callbacks that are waiting for a not-yet-completed RCU grace period, said callback advance operation including advancing to said done-sublist any of said groups of RCU callbacks whose assigned grace period numbers correspond to RCU grace periods that have ended, said callback advance operation further including performing said callback accelerate operation.

18. The computer program product of claim 17, wherein said callback advance operation is attempted as said one or more processors prepare for entry into a low power state, said callback accelerate operation is attempted on entry into said low power state, and said callback advance operation is attempted again on exit from said low power state.

19. The computer program product of claim 15, wherein said grace period number assigning includes recording future RCU grace periods needed by said one or more processors so that said future RCU grace periods can be initiated without waking said one or more processors if they are in a low power state.

20. The computer program product of claim 19, wherein said operations further include using said future grace period recording to offload callback invocation from specially designated ones of said one or more processors.