CN112486702B - Global message queue implementation method based on multi-core multi-processor parallel system - Google Patents

Abstract

The invention discloses a global message queue implementation method based on a multi-core multi-processor parallel system, wherein each processing core of the multi-core multi-processor parallel system carries out system initialization, and a section of shared storage space is mapped through SRIO (serial number input/output) for storing a global message queue and a global message queue name table; creating and initializing a global message queue from a global message queue buffer pool corresponding to a processing core to which a thread belongs, and filling global message queue information into a global message queue name table; when a thread sends a message to the global message queue or receives a message from the global message queue, the global message queue resource is managed and controlled through the global semaphore, so that the transmission of the message is realized. And the global semaphore and the shared storage area are utilized to control the message transfer among threads by the global semaphore, so that the high-speed communication requirement among multi-core multi-processors is met, the method has the characteristics of rapidness and high efficiency, and the work of application developers can be simplified to a great extent.

Description

Implementation method of global message queue based on multi-core multi-processor parallel system

Technical field

The invention relates to a global message queue implementation method, specifically a global message queue implementation method based on a multi-core multi-processor parallel system, and belongs to the field of embedded computers.

Background technique

With the development of computer systems and the increasing demand for real-time and parallelism in field applications, multi-core and multi-processor parallel systems have become an important trend in the development of embedded computer systems.

As the number of cores and processor nodes in multi-core and multi-processor parallel systems increases, compared with the traditional multi-core processor message communication method, the message communication method of multi-core and multi-processor parallel systems is more complex. It not only requires the implementation of multi-core processor chip In addition to inter-core communication, it is also necessary to implement multi-processor inter-chip communication. In other words, message communication has become one of the bottlenecks affecting the performance of multi-core and multi-processor parallel systems. Therefore, efficient communication between processing cores and processors has become a key issue in multi-core and multi-processor parallel systems.

In addition, when communicating between processing cores and between processors, their threads must mutually access hardware resources on the multi-core processor. Otherwise, problems such as access data errors due to resource contention will occur, which will lead to inter-processing cores. Communication between processors failed.

Contents of the invention

The purpose of the present invention is to provide a global message queue implementation method based on a multi-core multi-processor parallel system in order to solve the above problems. This method uses global semaphores and shared storage areas to transfer messages between threads to global signals. Volume control. Any thread that uses the global message queue can perform operations related to the global message queue only after acquiring the global semaphore. This meets the high-speed communication needs between multi-core and multi-processors. It is fast, efficient, and safe, and can be used to a great extent. Simplify the work of application developers.

The present invention achieves the above objects through the following technical solutions: a global message queue implementation method based on a multi-core multi-processor parallel system. The global message queue implementation method includes:

S1) Each processing core of the multi-core multi-processor parallel system initializes the system, and maps a shared storage space through SRIO to store the global message queue and global message queue name table;

S2) Create and initialize the global message queue from the global message queue buffer pool corresponding to the processing core to which the thread belongs, and fill in the global message queue information into the global message queue name table;

S3) When a thread sends a message to or receives a message from the global message queue, the global message queue resources are managed and controlled through the global semaphore to realize message transmission.

Preferably, the multi-core multi-processor parallel system has at least one processor node; the processor node has at least one processing core; and the processor nodes or processing cores support SRIO bus interconnection.

Preferably, in step S1), the system initialization process includes:

S11) Each processing core initializes SRIO, and maps a shared storage space through SRIO to store the global message queue and global message queue name table;

S12) Select any processing core as the main processing core, create and initialize a shared global message queue name table, which is used to record all created global message queues;

S13) Create a global message queue buffer pool and a message buffer pool in the shared storage space mapped by each processing core.

Preferably, in step S12), the content of the global message queue name table includes a global semaphore that controls mutually exclusive access to the name table, the number of all created global message queues in the name table, and all created global message queue information;

The global message queue information includes name, type, processing core, message attributes, number of opens, data queue, idle queue, and global semaphore;

Among them, the message attributes include the maximum number of messages, message size, message identification and the current number of messages. The global semaphore includes a global semaphore used to control data queue access and a global semaphore used to control idle queue access.

Preferably, in step S13), the global message queue buffer pool is used to allocate a global message queue, and the message buffer pool is used to allocate an idle queue of the global message queue;

Among them, the global message queue buffer pool and the message buffer pool are two-way linked lists with header pointers. The linked list headers are placed on their respective processing cores, and mutually exclusive access can be managed with spin locks or global semaphores.

Preferably, in step S2), when the global message queue is initialized, the value of the global semaphore controlling the data queue is initialized to 0, and the value of the global semaphore controlling the idle queue is initialized to the maximum number of messages the message queue is allowed to carry.

Preferably, in step S3), the operation of the thread sending messages to the global message queue specifically includes:

S31) Obtain the global semaphore that controls the idle queue in the message queue;

S32) If the global semaphore controlling the idle queue is 0, indicating that the message queue has reached the maximum number of messages and the global message queue is full, the thread will be blocked through the global semaphore;

S33) If the global semaphore controlling the idle queue is not 0, remove a piece of free area from the idle queue, copy the message to the free area, then join the data queue according to the message priority, modify the message attributes, and release the control data queue Global semaphore.

Preferably, in step S3), the operation of the thread receiving messages from the message queue specifically includes:

S34) Obtain the global semaphore that controls the data queue in the message queue;

S35) If the global semaphore controlling the data queue is 0, indicating that the message queue is empty, the thread will be blocked through the global semaphore;

S36) If the global semaphore controlling the data queue is not 0, according to the message priority and the "FIFO" principle, remove a message from the data queue, modify the message attributes, then copy the message, add the message to the idle queue, and release it Global semaphore that controls the idle queue.

The beneficial effects of the present invention are: the global message queue implementation method based on a multi-core multi-processor parallel system uses global semaphores and shared storage areas to control the message transfer between threads by global semaphores, which satisfies the needs of multi-core and multi-processors. It is fast and efficient, which can greatly simplify the work of application developers.

Description of the drawings

Figure 1 is a structural block diagram of a multi-core multi-processor parallel system applied in an embodiment of the present invention;

Figure 2 shows the principle of a global message queue implementation method for a multi-core multi-processor parallel system applied in the embodiment of the present invention;

Figure 3 is a system initialization flow chart applied in the embodiment of the present invention;

Figure 4 is a global message queue name table structure applied in the embodiment of the present invention;

Figure 5 is a flow chart of sending messages through the global message queue applied in the embodiment of the present invention;

Figure 6 is a flow chart of global message queue receiving messages applied in the embodiment of the present invention.

Detailed ways

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

A global message queue implementation method based on a multi-core multi-processor parallel system. The global message queue implementation method includes:

S1) Each processing core of the multi-core multi-processor parallel system initializes the system, and maps a shared storage space through SRIO to store the global message queue and global message queue name table.

In step S1), the system initialization process includes:

S11) Each processing core initializes SRIO, and maps a shared storage space through SRIO to store the global message queue and global message queue name table;

S12) Select any processing core as the main processing core, create and initialize a shared global message queue name table, which is used to record all created global message queues;

The content of the global message queue name table includes a global semaphore that controls mutually exclusive access to the name table, the number of all created global message queues in the name table, and all created global message queue information;

The global message queue information includes name, type, processing core, message attributes, number of opens, data queue, idle queue, and global semaphore;

Among them, the message attributes include the maximum number of messages, message size, message identification and the current number of messages. The global semaphore includes a global semaphore used to control data queue access and a global semaphore used to control idle queue access.

S13) Create a global message queue buffer pool and a message buffer pool in the shared storage space mapped by each processing core.

The global message queue buffer pool is used to allocate the global message queue, and the message buffer pool is used to allocate the idle queue of the global message queue;

Among them, the global message queue buffer pool and the message buffer pool are two-way linked lists with header pointers. The linked list headers are placed on their respective processing cores, and mutually exclusive access can be managed with spin locks or global semaphores.

S2) Create and initialize the global message queue from the global message queue buffer pool corresponding to the processing core to which the thread belongs, and fill the global message queue information into the global message queue name table.

When the global message queue is initialized, the value of the global semaphore that controls the data queue is initialized to 0, and the value of the global semaphore that controls the idle queue is initialized to the maximum number of messages that the message queue is allowed to carry.

S3) When a thread sends a message to or receives a message from the global message queue, the global message queue resources are managed and controlled through the global semaphore to realize message transmission.

Among them, the operation of the thread sending messages to the global message queue specifically includes:

S31) Obtain the global semaphore that controls the idle queue in the message queue;

S32) If the global semaphore controlling the idle queue is 0, indicating that the message queue has reached the maximum number of messages and the global message queue is full, the thread will be blocked through the global semaphore;

S33) If the global semaphore controlling the idle queue is not 0, remove a piece of free area from the idle queue, copy the message to the free area, then join the data queue according to the message priority, modify the message attributes, and release the control data queue Global semaphore.

The operation of the thread receiving messages from the message queue specifically includes:

S34) Obtain the global semaphore that controls the data queue in the message queue;

S35) If the global semaphore controlling the data queue is 0, indicating that the message queue is empty, the thread will be blocked through the global semaphore;

S36) If the global semaphore controlling the data queue is not 0, according to the message priority and the "FIFO" principle, remove a message from the data queue, modify the message attributes, then copy the message, add the message to the idle queue, and release it Global semaphore that controls the idle queue.

The multi-core multi-processor parallel system has at least one processor node; the processor node has at least one processing core; and the processor nodes or processing cores support SRIO bus interconnection.

Example

It should be noted:

1) As shown in Figure 1, this embodiment is implemented on an embedded multi-core multi-processor parallel system;

2) The embedded multi-core multi-processor parallel system used in this embodiment includes four processing boards, namely S ₀ , S ₁ , S ₂ , and S ₃ . Among them, the processing boards S0, S1, and S ₂ include two MPC8641D dual cores. processor and an SRIO switching device. The processing board S ₃ contains an MPC8641D dual-core processor and an SRIO switching device. Each MPC8641D processor contains two e600 processing cores. The processing cores are C ₀ , C ₁ ,..., C _i ,..., C ₁₃ , the processing core C ₀ is selected as the main processing core;

3) There is a thread A on the processing core C ₀ for sending messages; there is a thread B on the processing core C ₁₃ for receiving messages;

4) The MPC8641D processors used in this embodiment all support SRIO bus interconnection, and each MPC8641D dual-core processor is connected through an SRIO switching device;

5) This embodiment is a multi-core multi-processor parallel system based on SRIO.

As shown in Figure 2, a method for implementing a global message queue in a multi-core multi-processor parallel system in this embodiment includes the following steps:

S1) Each e600 processing core of the MPC8641D multi-core multi-processor parallel system performs system initialization and maps a shared storage space through SRIO to store the global message queue and global message queue name table.

S2) Each e600 processing core creates and initializes a global message queue from the global message queue buffer pool corresponding to the processing core to which the thread belongs, and fills the global message queue information into the global message queue name table.

S3) When a thread sends a message to or receives a message from the global message queue, the global message queue resources are managed and controlled through the global semaphore to realize message transmission.

As shown in Figure 3, in step S1, the system initialization process specifically includes:

S11) Each e600 processing core initializes SRIO, and maps a shared storage space through SRIO to store the global message queue and global message queue name table;

S12) Select the processing core C ₀ as the main processing core, create and initialize a shared global message queue name table, which is used to record all created global message queues;

S13) Create a global message queue buffer pool and a message buffer pool in the shared storage space mapped by each e600 processing core.

Among them, in step S11, each e600 processing core maps a space with a length of 4M bytes from the local address space to the SRIO address space as a shared memory for all processing cores of the MPC8641D multi-core multi-processor parallel system to access. The first address of the shared memory space is in sequence. It is 0xA4000000, 0xA4400000,…, 0xA4000000+i*0x400000,…, 0xA4000000+13*0x400000.

As shown in Figure 4, in step S12, the contents of the global message queue name table include the global semaphore that controls mutually exclusive access to the name table, the number of all created global message queues in the name table, and all created global message queue information. Among them, the global message queue information mainly includes name, type, processing core, message attributes, open times, data queue, idle queue, global semaphore, etc. The message attributes include the maximum number of messages, message size, message identifier and current number of messages. Global semaphores mainly include global semaphores used to control data queue access and global semaphores used to control idle queue access.

In step S13, the global message queue buffer pool is used to allocate a global message queue, and the message buffer pool is used to allocate an idle queue of the global message queue. Among them, the global message queue buffer pool and the message buffer pool are two-way linked lists with header pointers. The linked list headers are placed on their respective processing cores, and mutually exclusive access can be managed with spin locks or global semaphores.

In step S2, when the global message queue is initialized, the value of the global semaphore controlling the data queue is initialized to 0, and the value of the global semaphore controlling the idle queue is initialized to the maximum number of messages the message queue is allowed to carry.

As shown in Figure 5, in step S3, the operation of thread A to send a message to the global message queue specifically includes:

S31) Obtain the global semaphore that controls the idle queue in the message queue;

S32) If the global semaphore controlling the idle queue is 0, indicating that the message queue has reached the maximum number of messages and the global message queue is full, the thread will be blocked through the global semaphore;

S33) If the global semaphore controlling the idle queue is not 0, remove a piece of free area from the idle queue, copy the message to the free area, then join the data queue according to the message priority, modify the message attributes, and release the control data queue Global semaphore.

As shown in Figure 6, in step S3, the operation of thread B to receive messages from the message queue specifically includes:

S34) Obtain the global semaphore that controls the data queue in the message queue;

S35) If the global semaphore controlling the data queue is 0, indicating that the message queue is empty, the thread will be blocked through the global semaphore;

S36) If the global semaphore controlling the data queue is not 0, according to the message priority and "FIFO" or other principles, remove a message from the data queue, modify the message attributes, then copy the message, and add the message to the idle queue , release the global semaphore controlling the idle queue.

It is obvious to those skilled in the art that the present invention is not limited to the details of the above-described exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, the embodiments should be regarded as illustrative and non-restrictive from any point of view, and the scope of the present invention is defined by the appended claims rather than the above description, and it is therefore intended that all claims falling within the claims All changes within the meaning and scope of equivalent elements are included in the present invention. Any reference signs in the claims shall not be construed as limiting the claim in question.

In addition, it should be understood that although this specification is described in terms of implementations, not each implementation only contains an independent technical solution. This description of the specification is only for the sake of clarity, and those skilled in the art should take the specification as a whole. , the technical solutions in each embodiment can also be appropriately combined to form other implementations that can be understood by those skilled in the art.

Claims (2)

1. A global message queue implementation method based on a multi-core multi-processor parallel system, characterized in that the global message queue implementation method includes: S1) Each processing core of the multi-core multi-processor parallel system initializes the system, and maps a shared storage space through SRIO to store the global message queue and global message queue name table; In step S1), the system initialization process includes: S11) Each processing core initializes SRIO, and maps a shared storage space through SRIO to store the global message queue and global message queue name table; S12) Select any processing core as the main processing core, create and initialize a shared global message queue name table, which is used to record all created global message queues; In step S12), the content of the global message queue name table includes a global semaphore that controls mutually exclusive access to the name table, the number of all created global message queues in the name table, and all created global message queue information; The global message queue information includes name, type, processing core, message attributes, number of opens, data queue, idle queue, and global semaphore; Among them, the message attributes include the maximum number of messages, message size, message identification and the current number of messages. The global semaphore includes a global semaphore used to control data queue access and a global semaphore used to control idle queue access; S13) Create a global message queue buffer pool and message buffer pool in the shared storage space mapped by each processing core; In step S13), the global message queue buffer pool is used to allocate the global message queue, and the message buffer pool is used to allocate the idle queue of the global message queue; Among them, the global message queue buffer pool and message buffer pool are two-way linked lists with header pointers. The linked list headers are placed on their respective processing cores, and mutually exclusive access can be managed by spin locks or global semaphores; S2) Create and initialize the global message queue from the global message queue buffer pool corresponding to the processing core to which the thread belongs, and fill the global message queue information into the global message queue name table; In step S2), when the global message queue is initialized, the value of the global semaphore that controls the data queue is initialized to 0, and the value of the global semaphore that controls the idle queue is initialized to the maximum number of messages that the message queue is allowed to carry; S3) When a thread sends a message to or receives a message from the global message queue, the global message queue resources are managed and controlled through the global semaphore to realize message transmission; In step S3), the operation of the thread sending messages to the global message queue specifically includes: S31) Obtain the global semaphore that controls the idle queue in the message queue; S32) If the global semaphore controlling the idle queue is 0, indicating that the message queue has reached the maximum number of messages and the global message queue is full, the thread will be blocked through the global semaphore; S33) If the global semaphore controlling the idle queue is not 0, remove a free area from the idle queue, copy the message to the free area, then join the data queue according to the message priority, modify the message attributes, and release the control data queue global semaphore; In step S3), the operation of the thread receiving messages from the message queue specifically includes: S34) Obtain the global semaphore that controls the data queue in the message queue; S35) If the global semaphore controlling the data queue is 0, indicating that the message queue is empty, the thread will be blocked through the global semaphore; S36) If the global semaphore controlling the data queue is not 0, according to the message priority and the "FIFO" principle, remove a message from the data queue, modify the message attributes, then copy the message, add the message to the idle queue, and release it Global semaphore that controls the idle queue. 2. The global message queue implementation method based on a multi-core multi-processor parallel system according to claim 1, characterized in that: the number of processor nodes of the multi-core multi-processor parallel system is at least 1; the processor node There is at least one processing core; and the processor nodes or processing cores support SRIO bus interconnection.