Fast and memory protected asynchronous message scheme in a multi-process and multi-thread environment

• US 7,549,151 B2
• Filed: 06/03/2005
• Issued: 06/16/2009
• Est. Priority Date: 02/14/2005
• Status: Active Grant

• Current Assignee: Malikie Innovations Limited (Key Patent Innovations Limited)
• Original Assignee: QNX Software Systems Company (Blackberry Limited)
• Inventors: Marineau-Mes, Sebastien, Zhou, Hao, Belair, Steve, van der Veen, Peter, Kathail, Pradeep
• Primary Examiner(s): Ho; Andy
• Application Number:
  US11/145,105
• Publication Number:
  US 20060182137A1
• Time in Patent Office: 1,474 Days
• Field of Search: 719/313, 719/314, 719/318, 719/330
• US Class Current: 719/313
• CPC Class Codes: G06F 2209/547 Messaging middleware
• CPC Class Codes: G06F 9/544 Buffers; Shared memory; Pipes
• CPC Class Codes: G06F 9/546 Message passing systems or structures, e.g. queues

Abstract

An asynchronous message passing mechanism that allows for multiple messages to be batched for delivery between processes, while allowing for full memory protection during data transfers and a lockless mechanism for speeding up queue operation and queuing and delivering messages simultaneously.

Claims

1. A method of asynchronously passing messages in an operating system, the method comprising:

generating, by a consumer process in communication with the operating system, a channel for receiving asynchronous messages;

generating, by a producer process, a connection for sending asynchronous messages;

sending messages, by the producer process, to the connection;

joining the channel and the connection;

determining the occurrence of a triggering condition; and

passing messages from the connection to the channel after determining the occurrence of the triggering condition.

2. The method of claim 1, where the operating system includes a kernel module, the method further comprising joining the channel and the connection within the kernel module.

3. The method of claim 2, where messages are passed, by the kernel module, directly from a sender memory space associated with the producer process to a receiver memory space associated with the consumer process.

4. The method of claim 2, where sending messages to the connection comprises adding the messages to a send messages queue.

5. The method of claim 4, where the send messages queue is shared between the kernel module and the producer process.

6. The method of claim 5, where the queue is shared in a lockless manner such that the producer process may send messages to the connection while the kernel module is passing messages to the channel.

7. The method of claim 6, where the lockless method comprises:

allowing only the kernel module to modify a head pointer to the beginning of a list of messages to be sent from the send messages queue; and

allowing only the producer program to modify a tail pointer to the end of the list of messages to be sent from of the send messages queue and a free pointer to the start of free buffer space associated with the send messages queue.

8. The method of claim 1, further comprising:

establishing a synchronous communication path between the producer process and the consumer process.

9. The method of claim 1, where the triggering condition is a predetermined number of messages in send messages queue.

10. The method of claim 1, where the triggering condition is an expiration of a time period.

11. An operating system operable to asynchronously pass messages from a producer program to a consumer program, the operating system comprising:

an asynchronous messaging library in communication with the producer program and the consumer program and operable to provide a channel to the consumer program and a connection to the producer program, the connection operable to receive messages from the producer program and add them to a messages queue, the channel operable to queue messages for the consumer program; and

a kernel module in communication with the asynchronous messaging library and operable to pass messages from the producer program to the consumer program.

12. The operating system of claim 11, where messages are queued in the connection until a triggering event occurs.

13. The operating system of claim 12, where the triggering event is a predetermined number of messages in send messages queue.

14. The operating system of claim 12, where the triggering condition is an expiration of a time period.

15. The operating system of claim 11, where the messages are passed directly from a sender memory space associated with the producer process to a receiver memory space associated with the consumer process by the kernel module.

16. The operating system of claim 11, the messages queue is shared in a lockless manner such that the producer process may send messages to the connection while the kernel module is passing messages to the channel.

17. An asynchronous message passing architecture in communication with an operating system including a kernel module, the asynchronous message passing architecture operable to asynchronously pass messages from a sender to a receiver, the operating system comprising:

an asynchronous messaging library in communication with the sender and the receiver and operable to provide a channel to the receiver and a connection to the sender, the connection operable to receive messages from the sender and add them to a messages queue, the channel operable to queue messages for the receiver; and

where the kernel module is in communication with the asynchronous messaging library and operable to pass messages directly from the memory space associated with the sender to memory space associated with the receiver upon the occurrence of a triggering event.

18. The asynchronous message passing architecture of claim 17, the messages queue is shared in a lockless manner such that the sender may send messages to the connection while the kernel module is passing messages to the receiver.

19. The asynchronous message passing architecture of claim 18, where the lockless manner comprises:

allowing only the kernel module to modify a head pointer to the beginning of a list of messages to be sent from the messages queue.

20. The asynchronous message passing architecture of claim 19, where the lockless manner further comprises:

allowing only the asynchronous messaging library to modify a tail pointer to the end of the list of messages to be sent from of the messages queue and a free pointer to the start of free buffer space associated with the messages queue.