System and method for dynamically determining a thread to perform work using a lockable resource

• US 20060225078A1
• Filed: 03/30/2005
• Published: 10/05/2006
• Est. Priority Date: 03/30/2005
• Status: Active Grant

• Current Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Anderson, Eric A.
• Granted Patent: US 7,797,704 B2
• Time in Patent Office: Days
• US Class Current: 718/104
• CPC Class Codes: G06F 9/4843 by program, e.g. task dispatcher, supervisor, operating system
• CPC Class Codes: G06F 9/526 Mutual exclusion algorithms

Abstract

According to at least one embodiment, a method comprises a first thread desiring to perform work using a resource. The first thread determines whether another thread holds a lock on the resource, and if another thread (“blocking thread”) holds a lock on the resource, the work is transferred to the blocking thread for performance of the work by the blocking thread using the resource.

Claims

1. A method comprising:

a first thread desiring to perform work using a resource;

the first thread determining whether another thread holds a lock on the resource; and

if another thread (“

blocking thread”

) holds a lock on the resource, transferring the work to the blocking thread for performance of the work by the blocking thread using the resource.

2. The method of claim 1 wherein said transferring comprises:

said first thread placing a description of the work in a pending queue associated with the resource.

3. The method of claim 2 further comprising:

after placing a description of the work in said pending queue, the thread performs other work without waiting for said work described in the pending queue to be performed.

4. The method of claim 2 wherein said transferring further comprises:

the blocking thread assigning to itself the work described in the pending queue for performance thereof using the resource.

5. The method of claim 1 further comprising:

the blocking thread using the resource to perform some first work, while the blocking thread holds said lock on the resource; and

the blocking thread then determining whether any work is pending in a pending queue associated with the resource before unlocking the resource.

6. The method of claim 5 further comprising:

said first thread placing a description of the work in the pending queue; and

the blocking thread assigning to itself the work described in the pending queue for performance thereof using the resource.

7. A method comprising:

attempting, by a thread that desires to use a resource for performing work, trylock on the resource;

if said trylock is unsuccessful, the thread placing a description of the work in a pending queue associated with the resource, and a thread that holds a lock on the resource performs the work.

8. The method of claim 7 further comprising:

if said trylock is successful, said thread holds a lock on the resource and performs the work.

9. The method of claim 7 further comprising:

after placing a description of the work (“

pending work”

) in said pending queue, the thread that placed the description in the pending queue performs other work without awaiting for said pending work to be performed.

10. A method comprising:

attempting, by a thread that desires to use a resource for performing work, trylock on the resource;

if said trylock is successful, said thread holds a lock on the resource and performs the work;

if said trylock is unsuccessful, said thread places a description of the work in a pending queue associated with the resource; and

after a thread holding a lock on the resource performs work using the resource, the thread holding the lock performs any work identified in the pending queue before unlocking the resource.

11. The method of claim 10 further comprising:

after placing a description of the work in said pending queue, the thread is available to perform other work.

12. The method of claim 10 further comprising:

after placing a description of the work (“

pending work”

) in said pending queue, the thread performs other work without waiting for said pending work to be performed.

13. A system comprising:

processor for executing a plurality of threads;

a lockable resource usable by said plurality of threads;

said threads each operable to dynamically determine a thread to use the lockable resource for performing work desired by a thread, wherein a first thread desiring to perform work using the lockable resource is operable to determine whether a lock is available for the resource;

wherein if said lock is determined available for the lockable resource, the first thread is operable to hold the lock on the lockable resource and use the lockable resource to perform the work; and

wherein if said lock is determined not available for the lockable resource, the first thread is operable to place a description of the work it desires to perform in a pending queue associated with the resource.

14. The system of claim 13 wherein after a thread that holds a lock on the resource performs work using the resource, the thread that holds the lock performs any work described in the pending queue before unlocking the resource.

15. The system of claim 13 wherein if said lock is determined not available for the lockable resource, then after placing a description of the work (“

pending work”

) in said pending queue, the first thread performs other work without awaiting for said pending work to be performed.

16. The system of claim 13 wherein if said lock is not available to the first thread for the lockable resource because the lock is held by a second thread, the second thread performs the work described in the pending queue.

17. A multi-threaded system comprising:

a lockable resource that is usable by a plurality of threads, said lockable resource comprising a first lockable means for performing work of a thread and a second lockable means for queuing pending work desired to be performed using the first lockable means;

means for determining, upon a thread desiring to perform work using the first lockable means of said lockable resource, whether another thread holds a lock on the first lockable means;

means for placing a description of the work in said second lockable means for performance of the work by said another thread that holds said lock on the first lockable means, if determined by the determining means that said another thread holds said lock on the first lockable means.

18. The multi-threaded system of claim 17 wherein said means for determining comprises:

said thread desiring to perform said work attempting trylock on the first lockable means.

19. The multi-threaded system of claim 17 wherein said means for placing comprises:

said thread desiring to perform said work acquiring a lock on said second lockable means and placing said description of the work in said second lockable means.

20. The multi-threaded system of claim 17 further comprising:

wherein said thread desiring to perform said work performs other work after placing said description of the work in said second lockable means.

21. The multi-threaded system of claim 20 further comprising:

said thread desiring to perform said work performs said other work before said another thread that holds said lock on the first lockable means performs said work described by the description in said second lockable means.

22. A system comprising:

a plurality of threads;

a partitioned queue that comprises a plurality of partitions;

each thread operable to add a work description in the partitioned queue and remove a work description from the partitioned queue; and

each thread further operable to dynamically determine which of the plurality of partitions to use for adding said work description in the partitioned queue or removing said work description from the partitioned queue, thereby dynamically determining which of said plurality of threads is to perform work described by a work description.

23. The system of claim 22 further comprising:

a collection of work descriptions distributed across said partitioned queue before said plurality of threads begin execution.