FAST RESOURCE RECOVERY AFTER THREAD CRASH

• US 20110173483A1
• Filed: 01/14/2010
• Published: 07/14/2011
• Est. Priority Date: 01/14/2010
• Status: Active Grant

• Current Assignee: Juniper Networks Incorporated
• Original Assignee: Juniper Networks Incorporated
• Inventors: LYNN, Michael
• Granted Patent: US 8,365,014 B2
• Time in Patent Office: Days
• US Class Current: 714/2
• CPC Class Codes: G06F 11/0715 in a system implementing multitasking multitasking per se G06F9/46
• CPC Class Codes: G06F 11/0793 Remedial or corrective actions recovery from an exception in an instruction pipeline G06F9/3861; by retry G06F11/1402; for recovering from a failure of a protocol instance or entity H04L69/40
• CPC Class Codes: G06F 11/1402 Saving, restoring, recovering or retrying
• CPC Class Codes: G06F 9/5016 the resource being the memory

Abstract

A resource recovery system may maintain a counter in memory that indicates a number of times one or more threads of execution, which use shared resources, have crashed. The system may associate a first value of the counter with a resource allocated to a thread of the one or more threads, and may set an indicator associated with the thread to indicate whether the thread has crashed. The system may determine whether to re-allocate the resource to the thread based on the first value of the counter associated with the resource and based on the indicator associated with the thread.

Claims

1. A method implemented by a resource recovery system, the method comprising:

maintaining a counter in memory that indicates a number of times one or more threads have crashed;

determining, by the resource recovery system, whether at least one of the one or more threads has crashed; and

using, based on the determination, the counter to determine whether to re-allocate resources to the one or more threads for thread crash recovery.

2. The method of claim 1, further comprising:

incrementing the counter if at least one of the one or more threads has crashed; and

using the incremented counter to determine whether to re-allocate the resources to the one or more threads for thread crash recovery.

3. The method of claim 1, further comprising:

performing a resource recovery process if at least one of the one or more threads has crashed.

4. The method of claim 3, further comprising:

storing data associated with each of the resources in an array as each of the resources is allocated to the one or more threads.

5. The method of claim 4, further comprising:

storing a value of the counter in association with the data in the array when each of the resources is allocated to the one or more threads.

6. The method of claim 5, where performing the resource recovery process comprises:

comparing the value of the counter stored in association with the data in the array with another value of the counter obtained when the resource recovery process is occurring; and

determining whether to re-allocate each resource of the resources based on the comparison.

7. A resource recovery system, comprising:

a memory storing a data array;

a thread crash counter to maintain a count of a number of times one or more threads of execution have crashed; and

a resource recovery management unit to;

determine whether at least one of the one or more threads has crashed, and

use the counter to determine whether to perform a resource recovery process when at least one of the one or more threads has crashed.

8. The system of claim 7, where the resource recovery management unit is further to:

increment the thread crash counter if at least one of the one or more threads has crashed.

9. The system of claim 7, where the resource recovery management unit is further to:

store data associated with each of the resources in the data array as each of the resources is allocated to the one or more threads.

10. The system of claim 9, where the resource recovery management unit is further to:

store a value of the counter in association with the data in the data array when each of the resources is allocated to the one or more threads.

11. The system of claim 10, where, when performing the resource recovery process, the resource recovery management unit is to:

compare the value of the counter stored in associated with the data in the array with a value of the counter obtained when the resource recovery process is occurring, and

determine whether to re-allocate each resource of the resources based on the comparison.

12. A method implemented by a resource recovery system, comprising:

maintaining a counter in memory that indicates a number of times one or more threads of execution, which use shared resources, have crashed;

associating, by the resource recovery system, a first value of the counter with a resource allocated to a thread of the one or more threads;

setting an indicator associated with the thread to indicate whether the thread has crashed; and

determining whether to re-allocate the resource to the thread based on the first value of the counter associated with the resource and based on the indicator associated with the thread.

13. The method of claim 12, where determining whether to re-allocate the resource to the thread comprises:

obtaining a second value of the counter;

determining if the second value of the counter is greater than the first value of the counter;

determining if the indicator associated with the thread indicates that the thread has crashed; and

re-allocating the resource to the thread if the second value of the counter is greater than the first value of the counter and if the indicator indicates that the thread has crashed.

14. The method of claim 13, further comprising:

clearing the indicator associated with the thread, to indicate that the thread is no longer in a crashed state, when re-allocating the resource to the thread.

15. The method of claim 13, further comprising:

obtaining the first value of the counter when the resource is initially allocated to the thread,

where the second value of the counter is obtained subsequent to when the resource is initially allocated to the thread.

16. The method of claim 15, where the second value of the counter is obtained after any thread of the one or more threads has crashed.

17. A system, comprising:

a thread crash counter that indicates a number of times one or more threads, which use shared resources, have crashed; and

a resource recovery management unit to;

obtain a first value of the counter,

associate the first value of the counter with a resource allocated to a thread of the one or more threads,

set an indicator associated with the thread to indicate whether the thread has crashed,

obtain a second value of the counter, where the second value of the counter is obtained at a different time than the first value of the counter is obtained,

determine if the second value of the counter is greater than the first value of the counter,

determine if the indicator associated with the thread indicates that the thread has crashed, and

re-allocate the resource to the thread if the second value of the counter is greater than the first value of the counter and if the indicator indicates that the thread has crashed.

18. The system of claim 17, where, when re-allocating the resource to the thread, the resource recovery management unit is further to:

clear the indicator associated with the thread to indicate that the thread is no longer in a crashed state.

19. The system of claim 17, where the resource recovery management unit is further to:

obtain the first value of the counter when the resource is initially allocated to the thread,

where the second value of the counter is obtained subsequent to when the resource is initially allocated to the thread.

20. The system of claim 19, where the second value of the counter is obtained after any thread of the one or more threads has crashed.

21. A resource recovery system, comprising:

means for maintaining a counter that indicates a number of times one or more threads, which use shared resources, have crashed;

means for obtaining a first value of the counter when a resource of the shared resources is initially allocated to a thread of the one or more threads;

means for associating the first value of the counter with the resource;

means for setting an indicator associated with the thread to indicate whether the thread has crashed; and

means for determining whether to re-allocate the resource to the thread based on the first value of the counter associated with the resource and based on the indicator associated with the thread.