System and method for implementing shared locks between kernel and user space for synchronize access without using a system call to the kernel

• US 9,128,786 B2
• Filed: 11/22/2011
• Issued: 09/08/2015
• Est. Priority Date: 11/22/2011
• Status: Active Grant

• Current Assignee: Futurewei Technologies Incorporated (Huawei Investment & Holding Co., Ltd.)
• Original Assignee: Futurewei Technologies Incorporated (Huawei Investment & Holding Co., Ltd.)
• Inventors: Dong, Xuesong, Yin, Hongtao, Stewart, Randall, Hu, Shihui, Li, Fengkai
• Primary Examiner(s): An, Meng
• Assistant Examiner(s): GHAFFARI, ABU Z
• Application Number:
  US13/302,918
• Publication Number:
  US 20130132627A1
• Time in Patent Office: 1,386 Days
• Field of Search: 718/102, 718/107, 711/147, 711/152, 711/167, 711/163, 710/200
• US Class Current: 1/1
• CPC Class Codes: G06F 9/4881 Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
• CPC Class Codes: G06F 9/52 Program synchronisation; Mutual exclusion, e.g. by means of semaphores
• CPC Class Codes: G06F 9/526 Mutual exclusion algorithms
• CPC Class Codes: G06F 9/545 where tasks reside in different layers, e.g. user- and kernel-space

Abstract

An apparatus comprising one or more processors configured to implement a plurality of operations for an operating system (OS) platform including a kernel and a user application, one or more shared resource blocks by the kernel and the user application, and one or more shared locks by the kernel and the user application corresponding to the shared resource blocks, wherein the user application is configured to synchronize accesses to the shared resource blocks between a user thread and a kernel thread by directly accessing the locks without using a system call to the kernel.

Claims

1. An apparatus comprising:

one or more processors configured to implement a plurality of operations for an operating system (OS) platform including a kernel and a user application;

one or more shared physical memory blocks that are accessible to both a thread of the kernel and a user thread of the user application; and

one or more shared locks that are manageable by both the kernel thread and the user thread,

wherein each of the shared locks corresponds to at least one of the shared physical memory blocks,

wherein the user application is configured to synchronize accesses to the shared physical memory blocks between the user thread and the kernel thread by accessing the locks via the user thread without using a system call to the kernel,

wherein the kernel thread is configured to set at least one of the shared locks prior to accessing the corresponding shared physical memory block to lock out access by the user thread,

wherein the user thread is configured to set at least one of the shared locks prior to accessing the corresponding shared physical memory block to lock out access by the kernel thread,

wherein the user application shares the shared physical memory blocks by mapping physical memory block data from a kernel space into a user space and operating on the mapped data,

wherein the operations for the OS include a plurality of kernel threads and a plurality of user threads that are categorized according to priority,

wherein the kernel threads that have higher priority are allowed to remove the locks set by the user threads that have lower priority by terminating the user threads, and

wherein the shared locks are initialized in a kernel space and used by the user threads and the kernel threads.

2. The apparatus of claim 1, wherein the OS platform is a multi-processing network routing OS platform, and wherein the user application is a network routing function.

3. The apparatus of claim 1, wherein the shared locks comprise at least one of a plurality of mutually exclusive locks, read/write locks, and spin locks.

4. The apparatus of claim 1, wherein the shared physical memory blocks comprise a zero-copy socket that is shared between the kernel and the user application.

5. The apparatus of claim 1, wherein the one or more shared physical memory blocks comprise a send buffer and a receive buffer, wherein the one or more shared locks comprise a send buffer lock and a receive buffer lock, wherein the user thread is a producer user thread that is configured to access the send buffer and manage the send buffer lock, wherein the producer user thread is not configured to access the receive buffer, and wherein the producer user thread is not configured to manage the receive buffer lock.

6. The apparatus of claim 5, wherein the kernel thread is a consumer kernel thread that is configured to access the send buffer and manage the send buffer lock, wherein the consumer kernel thread is not configured to access the receive buffer, and wherein the consumer kernel thread is not configured to manage the receive buffer lock.

7. An apparatus comprising:

one or more processors configured to implement a plurality of operations for an operating system (OS) platform including a kernel and a user application;

one or more shared physical memory blocks that are accessible to both a thread of the kernel and a user thread of the user application; and

one or more shared locks that are directly manageable by both the kernel thread and the user thread,

wherein each of the shared locks corresponds to at least one of the shared physical memory blocks,

wherein the user application is configured to synchronize accesses to the shared physical memory blocks between the user thread and the kernel thread by directly accessing the locks via the user thread without using a system call to the kernel,

wherein the kernel thread is configured to set at least one of the shared locks prior to accessing the corresponding shared physical memory block to lock out access by the user thread,

wherein the user thread is configured to set at least one of the shared locks prior to accessing the corresponding shared physical memory block to lock out access by the kernel thread,

wherein the one or more shared physical memory blocks comprise a send buffer and a receive buffer,

wherein the one or more shared locks comprise a send buffer lock and a receive buffer lock,

wherein the user thread is a consumer user thread, wherein the kernel thread is a producer kernel thread,

wherein the user application is associated with the consumer user thread, wherein the kernel is associated with the producer kernel thread,

wherein the receive buffer is shared by the consumer user thread and the producer kernel thread,

wherein the consumer user thread is configured to access the receive buffer and manage the receive buffer lock,

wherein the consumer user thread is not configured to access the send buffer, wherein the consumer user thread is not configured to manage the send buffer lock,

wherein the producer kernel thread is configured to access the receive buffer and manage the receive buffer lock,

wherein the producer kernel thread is not configured to access the send buffer, and

wherein the producer kernel thread is not configured to manage the send buffer lock.