APPLICATION-DRIVEN SHARED DEVICE QUEUE POLLING IN A VIRTUALIZED COMPUTING ENVIRONMENT

• US 20130139156A1
• Filed: 11/30/2011
• Published: 05/30/2013
• Est. Priority Date: 11/30/2011
• Status: Active Grant

• Current Assignee: Red Hat Israel Ltd. (International Business Machines Corporation)
• Original Assignee: Red Hat Israel Ltd. (International Business Machines Corporation)
• Inventors: Tsirkin, Michael
• Granted Patent: US 9,009,702 B2
• Time in Patent Office: Days
• US Class Current: 718/1
• CPC Class Codes: G06F 2009/45583 Memory management, e.g. access or allocation
• CPC Class Codes: G06F 2209/548 Queue
• CPC Class Codes: G06F 9/45558 Hypervisor-specific management and integration aspects
• CPC Class Codes: G06F 9/544 Buffers; Shared memory; Pipes

Abstract

Methods and systems for application-driven polling of shared device queues are provided. One or more applications running in non-virtualized or virtualized computing environments may be adapted to enable methods for polling shared device queues. Applications adapted to operate in a polling mode may transmit a request to initiate polling of shared device queues, wherein operating in the polling mode disables corresponding device interrupts. Applications adapted to operate in a polling mode may be regulated by one or more predefined threshold limitations.

Claims

1. A computer-implemented method for application-driven polling in a virtualized computing environment, said method comprising:

engaging a hypervisor to process a request at a host operating system, said request received from an application, identifying one or more sockets associated with said application, and providing instructions to poll for messages;

in response to said request, determining if at least one message, having an association with at least one of said sockets associated with said application, is stored in a memory space associated with said host operating system, wherein said at least one message when stored in said memory space is to be passed to said application via said hypervisor;

identifying one or more shared device queues of at least one device having previously stored messages for at least one of said sockets associated with said application;

transmitting instructions to said at least one device to enable polling of said identified one or more shared device queues;

in response to said instructions to enable polling, determining if at least one message is stored in said identified one or more shared device queues; and

determining if said at least one message has an association with at least one of said sockets associated with said application, wherein said at least one message when stored in said identified one or more shared device queues is to be passed to said application via said hypervisor.

2. The computer-implemented method of claim 1, further comprising defining a threshold value associated with said polling of said identified one or more shared device queues, wherein said polling of said identified one or more shared device queues is disabled when a determination is made that said threshold value has been met or exceeded.

3. The computer-implemented method of claim 2, wherein instructions are generated at said host operating system, or at said hypervisor and communicated to said host operating system, and transmitted to said at least one device upon determining said threshold value has been met or exceeded to disable said polling of said identified one or more shared device queues.

4. The computer-implemented method of claim 2, wherein said request received at said host operating system further comprises identifying said threshold value.

5. The computer-implemented method of claim 2, wherein said threshold value is representative of a time period permitted for polling said identified one or more shared device queues.

6. The computer-implemented method of claim 2, wherein said threshold value is representative of a data allowance limit permitted for polling said identified one or more shared device queues.

7. The computer-implemented method of claim 2, wherein said threshold value is representative of a time period in which a message intended for a socket, not included in said one or more sockets associated with said application, is permitted to remain in said identified one or more shared device queues during said polling.

8. The computer-implemented method of claim 2, wherein said threshold value is representative of a total number of messages for different sockets, not included in said one or more sockets associated with said application, permitted in said identified one or more shared device queues during said polling.

9. The computer-implemented method of claim 1, wherein said polling of said identified one or more shared device queues for messages follows a determination that there are no messages for said one or more sockets associated with said application in said allocated memory space.

10. The computer-implemented method of claim 1, wherein said polling of said identified one or more shared device queues is disabled upon a determination made by a guest operating system to stop running the application.

11. The computer-implemented method of claim 1, wherein said polling of said identified one or more shared device queues is disabled upon a determination made by said host operating system to stop running at least a part of said hypervisor, said part of said hypervisor being responsible for running said application, or said part of said hypervisor being responsible for passing messages from said shared device queues to said application.

12. A computer system, comprising:

a memory;

a processing device, coupled to said memory; and

an operating system hosted by said computer system, having access to said memory and use of said processing device, said host operating system being in communication, via a hypervisor, with a guest operating system of a virtual machine, said host operating system configured to;

engage a hypervisor to process a request at a host operating system, said request received from an application, identifying one or more sockets associated with said application, and providing instructions to poll for messages;

in response to said request, determine if at least one message, having an association with at least one of said sockets associated with said application, is stored in a memory space associated with said host operating system, wherein said at least one message when stored in said memory space is to be passed to said application via said hypervisor;

identify one or more shared device queues of at least one device having previously stored messages for at least one of said sockets associated with said application;

transmit instructions to said at least one device to enable polling of said identified one or more shared device queues;

in response to said instructions to enable polling, determine if at least one message is stored in said identified one or more shared device queues; and

determine if said at least one message has an association with at least one of said sockets associated with said application, wherein said at least one message when stored in said identified one or more shared device queues is to be passed to said application via said hypervisor.

13. The computer system of claim 12, wherein said host operating system is further configured to define a threshold value associated with said polling of said identified one or more shared device queues, wherein said polling of said identified one or more shared device queues is disabled when a determination is made that said threshold value has been met or exceeded.

14. The computer system of claim 13, wherein instructions are generated at said host operating system, or at said hypervisor and communicated to said host operating system, and transmitted to said at least one device upon determining said threshold value has been met or exceeded to disable said polling of said identified one or more shared device queues.

15. The computer system of claim 13, wherein said request received at said host operating system further comprises identifying said threshold value.

16. The computer system of claim 13, wherein said threshold value is representative of a time period permitted for polling said identified one or more shared device queues.

17. The computer system of claim 13, wherein said threshold value is representative of a data allowance limit permitted for polling said identified one or more shared device queues.

18. The computer system of claim 13, wherein said threshold value is representative of a time period in which a message intended for a socket, not included in said one or more sockets associated with said application, is permitted to remain in said identified one or more shared device queues during said polling.

19. The computer system of claim 13, wherein said threshold value is representative of a total number of messages for different sockets, not included in said one or more sockets associated with said application, permitted in said identified one or more shared device queues during said polling.

20. The computer system of claim 12, wherein said polling of said identified one or more shared device queues for messages follows a determination that there are no messages for said one or more sockets associated with said application in said allocated memory space.

21. The computer system of claim 12, wherein said polling of said identified one or more shared device queues is disabled upon a determination made by a guest operating system to stop running the application.

22. The computer system of claim 12, wherein said polling of said identified one or more shared device queues is disabled upon a determination made by said host operating system to stop running at least a part of said hypervisor, said part of said hypervisor being responsible for running said application, or said part of said hypervisor being responsible for passing messages from said shared device queues to said application.

23. A non-transitory computer-readable storage medium programmed to include instructions that, when executed by a processing device, cause the processing device to perform a method for application-driven polling in a virtualized computing environment, said method comprising:

engaging a hypervisor to process a request at a host operating system, said request received from an application, identifying one or more sockets associated with said application, and providing instructions to poll for messages;

in response to said request, determining if at least one message, having an association with at least one of said sockets associated with said application, is stored in a memory space associated with said host operating system, wherein said at least one message when stored in said memory space is to be passed to said application via said hypervisor;

identifying one or more shared device queues of at least one device having previously stored messages for at least one of said sockets associated with said application;

transmitting instructions to said at least one device to enable polling of said identified one or more shared device queues;

in response to said instructions to enable polling, determining if at least one message is stored in said identified one or more shared device queues; and

determining if said at least one message has an association with at least one of said sockets associated with said application, wherein said at least one message when stored in said identified one or more shared device queues is to be passed to said application via said hypervisor.

24. The non-transitory computer-readable storage medium of claim 23, further comprising defining a threshold value associated with said polling of said identified one or more shared device queues, wherein said polling of said identified one or more shared device queues is disabled when a determination is made that said threshold value has been met or exceeded.

25. The non-transitory computer-readable storage medium of claim 24, wherein instructions are generated at said host operating system, or at said hypervisor and communicated to said host operating system, and transmitted to said at least one device upon determining said threshold value has been met or exceeded to disable said polling of said identified one or more shared device queues.

26. The non-transitory computer-readable storage medium of claim 24, wherein said request received at said host operating system further comprises identifying said threshold value.

27. The non-transitory computer-readable storage medium of claim 24, wherein said threshold value is representative of a time period permitted for polling said identified one or more shared device queues.

28. The non-transitory computer-readable storage medium of claim 24, wherein said threshold value is representative of a data allowance limit permitted for polling said identified one or more shared device queues.

29. The non-transitory computer-readable storage medium of claim 24, wherein said threshold value is representative of a time period in which a message intended for a socket, not included in said one or more sockets associated with said application, is permitted to remain in said identified one or more shared device queues during said polling.

30. The non-transitory computer-readable storage medium of claim 24, wherein said threshold value is representative of a total number of messages for different sockets, not included in said one or more sockets associated with said application, permitted in said identified one or more shared device queues during said polling.

31. The non-transitory computer-readable storage medium of claim 23, wherein said polling of said identified one or more shared device queues for messages follows a determination that there are no messages for said one or more sockets associated with said application in said allocated memory space.

32. The non-transitory computer-readable storage medium of claim 23, wherein said polling of said identified one or more shared device queues is disabled upon a determination made by a guest operating system to stop running the application.

33. The non-transitory computer-readable storage medium of claim 23, wherein said polling of said identified one or more shared device queues is disabled upon a determination made by said host operating system to stop running at least a part of said hypervisor, said part of said hypervisor being responsible for running said application, or said part of said hypervisor being responsible for passing messages from said shared device queues to said application.