CN100424675C - Information communication method and apparatus of multiprocessor system - Google Patents
Information communication method and apparatus of multiprocessor system Download PDFInfo
- Publication number
- CN100424675C CN100424675C CNB2005101127481A CN200510112748A CN100424675C CN 100424675 C CN100424675 C CN 100424675C CN B2005101127481 A CNB2005101127481 A CN B2005101127481A CN 200510112748 A CN200510112748 A CN 200510112748A CN 100424675 C CN100424675 C CN 100424675C
- Authority
- CN
- China
- Prior art keywords
- sfu
- message
- functional unit
- software functional
- address
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Abstract
The invention relates to message communicating method used between multiprocessor system software functional units SFU. It includes the following steps: message enrolling for the SFU; building the coincidence relation between the target SFU address and the drive module; source SFU transmitting the message to the target SFU; when the target SFU is not in the same processor with the source SFU, using the drive module corresponding with the target SFU address to transmit message. The invention also supplies SFU message communicating device, processing message communicating method and device. It can greatly increase the portability of the middle and up layers software in multiprocessor system, reduce its designing complexity at the same time.
Description
Technical field
The present invention relates to multicomputer system, relate in particular to information communication method between the process and between the SFU software functional unit and device in the multicomputer system.
Background technology
In multicomputer system, need between each processor to coordinate its calculation process process mutually, finish the work with cooperation.A plurality of processes are all arranged on each processor usually, and for the Coordination Treatment process, being between the process on the same processor and being between the process between the different processor all needs to carry out data transmission, and data are transmitted between process with the form of message usually.
In the prior art, the message queue that normally adopts operating system to provide between the process on the same processor carries out message communicating; Message communicating on the different processor between the process then need depend on the transmission channel that hardware provides on the one hand, as network interface, serial line interface, HPI (Host Port Interface, host interface) bus, DPRAM (the Dual Port RandomAccess Memory that shares between processor, double-port RAM) or the like, on the other hand need be by means of the phase driver module that provides for each hardware corridor, to realize sending to the purpose process on the purpose processor function of message and the message that originating process was sent from the hardware corridor reception sources processor by hardware corridor.
For example, the hardware syndeton of a multicomputer system as shown in Figure 1, processor A is connected with processor B by serial ports, is connected with processor C by the HPI bus, the hardware corridor between processor B and the processor C is for sharing DPRAM.Process A1 and A2 are arranged on processor A, process B1 and B2 are arranged on the processor B, process C1, C2 and C3 are arranged on the processor C.Like this, between process A1 and the process A2, between process B1 and the process B2, the message queue that need utilize operating system to provide of the message communicating between process C1 and the process C2 carries out; And the driver module that provides for serial ports need be provided for process on the processor A and the communication between the process on the processor B, the driver module that provides for the HPI bus need be provided in communication between the process on process on the processor A and the processor C, and the driver module that provides for DPRAM need be provided the communication between the process on process on the processor B and the processor C.
As seen, there are following two problems in such inter-process messages communication means:
First application software design is complicated: adopted distinct mode owing to be on the same processor communication between the process and be on the different processor communication between the process in multicomputer system, so, originating process in the communication must know clearly very that whether the purpose process of communicating by letter with oneself is positioned on the same processor, decides both sides' communication to adopt which type of mode with this; If communicating pair is in different processors, then also must knows the physical channel that intercommunication relies on, and need call the corresponding driving module and communicate.This make the applied software development of multicomputer system than single processor system application software complexity a lot, to developer's requirement than higher.
It two is that application software is portable poor: owing to the inter-process messages communication that is positioned on the different processor in the multicomputer system need be called and the corresponding driver module of specific hardware corridor, so in case change has taken place in the hardware platform of multiprocessing system, for example original communication port is a serial ports between two processors, after hardware platform correcting, changed the HPI bus into, the part that then relates to the inter-process messages communication of these two processors in the application software code just need be made amendment, and changes into and calls HPI bus driver module from calling the serial port drive module originally.Greatly reduce the portability of application software on different hardware platforms like this.
Summary of the invention
The present invention will solve be in the application software of multicomputer system in the prior art to be positioned on the same processor be positioned at different processor on inter-process messages communicate by letter and need carry out different modes, and relate on the different processor problem that the portability of the application software of message communicating between process depends on hardware platform.
Information communication method in the multicomputer system of the present invention between SFU software functional unit SFU, wherein SFU software functional unit SFU is the communication entity that is under the jurisdiction of process, may further comprise the steps:
SFU software functional unit SFU comprises the information registration of its address, can be learnt the processor at this SFU place by described address;
Set up that the purpose SFU software functional unit SFU address on same processor and message do not arrive the corresponding relation of the employed driver module of this purpose SFU software functional unit SFU with source software functional unit SFU;
Source SFU sends message according to the address of purpose SFU software functional unit SFU to it;
When purpose SFU and source software functional unit SFU are not on same processor, use the driver module of this purpose SFU address correspondence to carry out message and send.
Preferably, when between the processor a plurality of hardware corridor being arranged, the corresponding relation of setting up purpose SFU software functional unit SFU address and driver module is specially: set up SFU address, source, and this source SFU not in the purpose SFU software functional unit SFU address on the same processor, message arrives corresponding relation between the employed driver module three of this purpose SFU from this source software functional unit SFU.
Preferably, described method also comprises: when the driver module of purpose SFU software functional unit SFU place processor receives described message, send message to it according to the address of purpose SFU in the described message on same processor.
Preferably, described method also comprises: set up the corresponding relation between SFU software functional unit SFU address and this SFU software functional unit SFU place process;
Described purpose SFU software functional unit SFU on same processor sends message and is specially:
Message is sent to the message queue of purpose SFU software functional unit SFU place process;
Receive the address of the process of message, give purpose SFU message distribution according to purpose SFU software functional unit SFU in the message.
Preferably, described method also comprises: for message queue of each process binding receives message for SFU software functional unit SFU in this process.
Preferably, the log-on message of SFU also comprises the message processing function of this SFU software functional unit SFU;
The described process that receives message is specially to purpose SFU software functional unit SFU message distribution:
Barrage reception sends to the message of described process on the message queue of described process;
According to the message processing function of purpose SFU software functional unit SFU address search purpose SFU software functional unit SFU registration in the message, call this message processing function and handle the message that receives.
Preferably, the address of described SFU software functional unit SFU comprises the processor flag at this SFU software functional unit SFU place and the sign of this SFU software functional unit SFU, wherein processor flag has uniqueness in multicomputer system, and SFU software functional unit SFU is identified on the processor at this SFU place has uniqueness.
The present invention also provides the message communicating device between SFU software functional unit SFU in a kind of multicomputer system, comprises SFU software functional unit SFU, SFU communication management module, SFU routing module, SFU information module and driver module group, wherein:
SFU software functional unit SFU information module is used for storing the SFU information that comprises the SFU address, can be learnt the processor at this SFU place by described address;
SFU software functional unit SFU routing module is used for storing source software functional unit SFU address on this processor, receive purpose SFU software functional unit SFU address on other processors of message, send a message to the corresponding relation of the employed driver module of this purpose SFU software functional unit SFU;
The driver module group comprises at least one driver module, be used for other processors on corresponding driver module carry out message communicating;
The SFU communication management module is used for sending and reception message, when purpose SFU software functional unit SFU address is on other processors in the transmission instruction of SFU software functional unit SFU, find corresponding driver module by the described corresponding relation of storing in the SFU software functional unit SFU routing module, transmission of messages is sent to described driver module.
Preferably, described device also comprises the local communication module of SFU, is used for proceeding to the communication of purpose SFU on this processor;
Purpose SFU software functional unit SFU address is on this processor in the transmission instruction of SFU software functional unit SFU, and SFU software functional unit SFU communication management module will send instruction and transfer to the local communication module of SFU software functional unit SFU.
Preferably, described SFU software functional unit SFU information module comprises SFU address module and SFU Registering modules, wherein:
The SFU address module is used for storing the SFU address, and the SFU address comprises processor flag and SFU software functional unit SFU unique identification on the processor under it at SFU software functional unit SFU place;
The SFU Registering modules is used for storing the log-on message of SFU, comprises the process identification (PID) at SFU software functional unit SFU place.
Preferably, described driver module group comprises sending module group and receiver module group, and wherein the sending module group comprises that at least one sends driver module, and the SFU software functional unit SFU that is used on other processors sends message; The receiver module group comprises that at least one receives driver module, be used for receiving the message that sends to SFU software functional unit SFU on this processor and with this transmission of messages to SFU software functional unit SFU communication management module;
The SFU communication management module with this transmission of messages to the local communication module of SFU software functional unit SFU, the local communication module of SFU software functional unit SFU is searched the process at purpose SFU software functional unit SFU place from SFU software functional unit SFU Registering modules, message is distributed to purpose SFU software functional unit SFU by described process.
Information communication method in the multicomputer system of the present invention between process may further comprise the steps:
The address of setting process comprises the processor flag at this process place in the described address;
Set up that the purpose process address on same processor and message do not arrive the corresponding relation of the employed driver module of this purpose process with originating process;
Originating process sends message according to the address of purpose process to it;
When purpose process and originating process are not on same processor, use the driver module of this purpose process address correspondence to carry out message and send.
Preferably, when between the processor a plurality of hardware corridor being arranged, the corresponding relation of setting up purpose process address and driver module is specially: set up the originating process address, and this originating process do not arrive corresponding relation between the employed driver module three of this purpose process from this originating process in the purpose process address on the same processor, message.
Preferably, described method also comprises: after the driver module on the processor of purpose process place receives described message, according to the address of purpose process message is sent to the purpose process.
Preferably, described method also comprises:
Judge that according to the originating process address that sends message and purpose process address originating process and purpose process are whether on same processor;
When purpose process and originating process are on same processor, use the message queue of operating system to send message to the purpose process.
The present invention also provides the message communicating device between process in a kind of multicomputer system, comprises process, process communication administration module, process address module, process routing module and driver module group, wherein:
The process address module is used for the address information of storage process, comprises the processor flag at process place;
The process routing module is used for storing originating process sign on this processor, receive purpose process address on other processors of message, send a message to the corresponding relation of the employed driver module of this purpose process;
The driver module group comprises at least one driver module, be used for other processors on corresponding driver module carry out message communicating;
The process communication administration module is used for sending and reception message, when the purpose process address is on other processors in the transmission instruction of process, find corresponding driver module by the described corresponding relation of storing in the process routing module, transmission of messages to described module is sent.
Preferably, described device also comprises the local communication module of process, is used for proceeding to the communication of purpose process on this processor;
The purpose process address is on this processor in the transmission instruction of process, and the process communication administration module will send instruction and transfer to the local communication module of process.
Preferably, described driver module group comprises sending module group and receiver module group, and wherein the sending module group comprises that at least one sends driver module, and the process that is used on other processors sends message; The receiver module group comprises that at least one receives driver module, be used for receiving the message that sends to process on this processor and with this transmission of messages to the process communication administration module;
The process communication administration module sends to the purpose process by the message that the local communication module of process will receive the driver module reception.
Pass through the processor at its place of sign in the address of communication main body among the present invention, safeguard the corresponding relation of purpose communication main body and the driver module that is called on other processors for the sources traffic main body, for providing the address that only needs purpose communication main body, application software can carry out the unified platform of message communicating, and according to carrying out on the same processor or the message communicating between processor in the address of purpose communication main body, thereby, reduced the design complexities of application software for the inter-process messages communication of application software provides unified implementation; And shielded the difference of hardware corridor between processor to application software, strengthened the portability of application software;
And then the present invention also provides unified interface to the driver module that receives message, has further strengthened portability of the present invention.
Description of drawings
Fig. 1 is a multiprocessor syndeton exemplary plot in the multicomputer system;
Fig. 2 is the process flow diagram of inter-process messages communication means of the present invention;
Fig. 3 is the structural drawing of inter-process messages communicator of the present invention;
Fig. 4 is for using the hierarchical diagram that the present invention carries out the multicomputer system of interprocess communication;
Fig. 5 is the process flow diagram of information communication method between SFU of the present invention (Software Function Unit, SFU software functional unit);
Fig. 6 is the structural drawing of message communicating device between SFU of the present invention;
Fig. 7 is for using the hierarchical diagram that the present invention carries out the multicomputer system of communicating by letter between SFU.
Embodiment
For unified inter-process messages communication mode is provided to application software, need to application software shield on the same processor with different processor on inter-process messages communicate by letter different, and the correlativity of interprocess communication on the different processor and hardware platform.
The flow process of inter-process messages communication means as shown in Figure 2 in the multicomputer system provided by the invention.At step S210, set the address of process, in the address, comprise the processor flag at this process place.The address of process is used for identifying uniquely each process in the multicomputer system, can find corresponding process definitely according to the address of process on certain processor.
Each process all has unique process ID (sign) on the processor at its place in the prior art, recommends to constitute in the following way the address of process among the present invention:
The processor ID at process place;
The process ID of process on this processor;
Other information.
At step S220, set up purpose process address that is positioned on other processors and the corresponding relation that arrives this purpose driver module that process is used for originating process.
When having only a hardware corridor to communicate by letter between per two processors with inter-process messages thereon, can set up the corresponding relation that purpose process address and message arrive the employed driver module of this purpose process for the processor at originating process place in this step, also can set up other processors at purpose process place and the corresponding relation that message arrives the employed driver module of this processor.
When having between the processor when surpassing one hardware corridor, the process on the processor can select different hardware corridors to carry out message communicating.At this moment, can set up the corresponding relation between purpose process on originating process, other processors, message arrives the employed driver module of this purpose process from this originating process the three, specify different hardware corridors so that the user can be different interprocess communication.
In order further to realize irrelevant with hardware platform, can make driver module that unified message transmission interface is provided.
At step S230, originating process sends message according to the address of purpose process to the purpose process.In other words, for originating process, when the purpose process sends message, only need to get final product as parameter with the address of purpose process and the message that will send, be positioned on which processor and needn't distinguish the purpose process, also needn't distinguish the hardware corridor between the processor at the processor at originating process place and purpose process place.
At step S240, judge whether purpose process and originating process are positioned on the same processor, if then the message queue that provides by operating system sends message to the purpose process; If not, execution in step S250 then.
Owing to all comprise the processor at process place in the address of originating process and purpose process, only need the processor flag in two process addresses of contrast to know promptly whether originating process and purpose process are positioned on the same processor.The message queue that inter-process messages communication on the same processor adopts operating system to provide is realized,, repeated no more herein with identical in the prior art.
At step S250, search the driver module that has corresponding relation with this address according to the purpose process address, use this driver module to send message to the purpose process.
At step S260, after the driver module on the processor of purpose process place receives the message that sends to the purpose process, message is sent to the purpose process according to the purpose process address in the message.
The method that driver module sends message is identical with the method that originating process sends message, and this moment, by the message queue that operating system provides, message was sent to the purpose process because driver module and purpose process are positioned on the same processor.
In the present invention, can comprise two parts in the message of interprocess communication, message header and message body, active process address and purpose process address in the message header, message body is used for carrying the concrete data of originating process and purpose process communication.
The device of using above-mentioned inter-process messages communication means as shown in Figure 3, process 310 is connected with the local communication module 360 of process address module 350, process routing module 320, process communication administration module 330 and process respectively, process communication module 330 is connected with driver module group 340 with process routing module 320, process address module 350, the local communication module 360 of process, wherein driver module group 340 comprises sending module group 341 and receiver module group 342, all is connected to process communication administration module 330.
Storing the address information of process on this processor in the process address module 350.The processor flag and the process ID of process on this processor that comprise the process place in the address information.
In process routing module 320, storing originating process sign on this processor, with this processor on this originating process carry out purpose process address on other processors of message communicating, and with the corresponding relation that sends a message to the employed driver module of this purpose process address from this originating process.
When having only a hardware corridor to communicate by letter between per two processors with inter-process messages thereon, can be in process routing module 320 storage purpose process address and message arrive the corresponding relation of the employed driver module of this purpose process, or other processors at purpose process place and message arrive the corresponding relation of the employed driver module of this processor.
When having between the processor when surpassing one hardware corridor, can in process routing module 320, store originating process address, purpose process address and message and arrive corresponding relation between the employed driver module three of this purpose process from this originating process.
When process 310 sends message, send the transmission instruction to process communication administration module 330, comprise purpose process address and the message that is sent in the transmission instruction.Process communication administration module 330 judges whether purpose process and originating process 310 are positioned on the same processor, if on same processor, then will send instruction and transfer to the local communication module 360 of process; If not on same processor,, and transmission of messages transmission driver module of this driver module to the driver module group 340 sent then to process routing module 320 inquiry and originating process 310 and the corresponding driver modules of purpose process address.
When certain the reception driver module in the receiver module group 342 receives the message of the process of sending to 310, this message is passed to process communication administration module 330.Process communication administration module 330 to the local communication module 360 of process, sends to purpose process 310 by it with message with this transmission of messages.
As seen, the local communication module 360 of process is used for finishing the message communicating between process on this processor.
In the present invention, the message that is used for interprocess communication recommends to adopt following form: message comprises message header and message body, wherein comprises originating process address and purpose process address in the message header, and message body is used for carrying the data of interprocess communication.
After implementing the present invention, carry out interprocess communication multicomputer system hierarchy as shown in Figure 4.The present invention is in upper application software and be employed in the prior art between the driver module set that software directly calls and increased one deck multicomputer system software platform (MPSSP, Multiple ProcessorSystem Software Platform), by the process communication administration module that comprises among the MPSSP, the process address module, the local communication module of process routing module and process has been isolated driver module set and upper application software, for upper application software provides Message Call Interface unified between process, also provide unified message sink interface simultaneously for the driver module set.Upper application software can be called corresponding driver module by MPSSP by the purpose process address, by the multicomputer system hardware platform message is sent to the purpose process.
It below is the specific implementation of a kind of recommendation of inter-process messages communication among the present invention.
The address of a process comprises the numbering and the process ID of this process on this processor of this process place processor.
The process routing module will be safeguarded originating process ID, purpose process address, driver module three's corresponding relation.Corresponding relation with this three in the implementation of recommending is divided into the realization of two steps, and the first step realizes the corresponding relation of originating process ID, purpose process address, driving numbering, and the realization of second step drives the corresponding relation of numbering with driver module.According to originating process ID and purpose process address, can find the driver module that originating process and purpose process are carried out the required use of inter-processor communication like this.
The suggested design of realizing first step corresponding relation is to safeguard a routing rule entity structure chained list for each process on this processor, and the routing rule entity structure should comprise following member at least:
The address of purpose process;
Drive numbering;
Point to the pointer of next bar routing rule entity structure.
When needs during according to originating process ID and purpose process address query driven numbering, at first find the routing rule entity structure chained list of this process according to originating process ID, search for this chained list then, find the routing rule entity corresponding, drive numbering thereby find with the purpose process.
The process routing module should provide the interface that adds routing rule to upper application software, is used for upper application software and adds a route to MPSSP, and a kind of recommendation prototype that routing rule adds function is:
MPSSP_PRO_REGISTER_RULE (the purpose process address,
The numbering that drives)
The suggested design of realizing the second step corresponding relation is to adopt to drive the entity structure array, drives entity structure and should comprise following member at least:
The transmission interface function pointer of driver module.
Be numbered the index that drives the entity structure array with driving, the process communication administration module can find the transmission interface function of corresponding driver module.
The process routing module should provide the interface of registration to driver module, is used for to driver module of MPSSP registration, and a kind of recommendation prototype that drives the registration interface function is:
MPSSP_PRO_REGISTER_DD (drive numbering,
The transmission interface function pointer of driver module)
The process message transmission interface that the process communication administration module will provide to upper application software.The recommendation prototype of process message transmission interface function is:
MPSSP_PRO_SEND (message pointer)
The message of message pointed comprises message header and message body, comprises originating process address and purpose process address in the message header.
The flow process of process message transmission interface is as follows:
A. according to the purpose process address, judge whether the purpose process is on this processor,, carry out B, otherwise carry out C if at this processor;
B. the purpose process is on this processor: the message queue message that the call operation system provides sends function, and message is sent to the purpose process, and message sends successfully and returns;
C. the purpose process is not on this processor: the routing rule query interface that the calling process routing module provides, find the driving numbering that sends a message to the required use of purpose process, according to driving numbering, find the transmission interface function pointer of this driver module, the transmission interface function of calling driver module sends message by hardware corridor, send successfully and return.
Driver module is after the message that the process on other processors of receiving sends, and the process message transmission interface function that the calling process communication management module provides is sent to purpose process on this processor with message.
In the prior art, the main body of message communicating is a process.In actual applications, may comprise a plurality of functional units in the process, each functional unit is realized difference in functionality (as the message that sends to this process is carried out different processing), and process decides which functional unit of execution according to the message of concrete implementation status or reception.When multicomputer system ran on multi job mode, the functional unit that process is moved and the functional unit of other processes usually had relevance.A kind of manifestation mode of this relevance is the message that sends to another process when the functional unit a of certain process of execution, needs the functional unit b of another process to handle.At this moment, the communication between two processes is actually the communication between functional unit a and the functional unit b, so the main body of message communicating also can be the functional unit of process.Among the present invention, this functional unit as communication entity is called SFU (Software Function Unit, SFU software functional unit).As seen, SFU is a kind of unit littler than process.A SFU is under the jurisdiction of and only is under the jurisdiction of a process, and one in-process can a plurality of SFU; The unique communication entity of having represented in the SFU communication of SFU.
The invention provides the information communication method between a kind of SFU, its flow process is shown in 5.At step S510, SFU carries out the information registration.Before SFU is as the communication main body, need register to system.The address that this SFU must be arranged in the log-on message can also comprise initialization function and the message processing function of this SFU.
The address of SFU is used for representing the identity of SFU, therefore should have uniqueness in total system, also needs simultaneously can learn from its address which processor this SFU is positioned on.The user can oneself set the expression mode of SFU address, for the ease of the communication between SFU on each processor, the present invention is recommended in sign and the SFU sign of this SFU on the processor of place that comprises SFU place processor in the address of SFU, wherein being identified in the whole multicomputer system of processor should have uniqueness, and should have uniqueness on the processor that is identified at its place of each SFU.
The initialization function of SFU can be called when SFU applies for the registration of, and is used for finishing the initial work of this SFU, as the application of the initialization of global variable, resource requirement and initialization or the like.The message processing function of SFU is used for the message that sends to this SFU is handled.
At step S520, set up the corresponding relation between SFU address and this SFU place process.Among the present invention, the message that sends to SFU arrives this SFU by the process at this SFU place, is undertaken by the address of purpose SFU and send message, sets up the corresponding relation of SFU address and its place process so that SFU can correctly receive message.
At step S530, foundation is positioned at the purpose SFU address on the different processor with source SFU and sends the corresponding relation of the employed driver module of message to this purpose SFU.
When having only hardware corridor between per two processors with between SFU thereon during message communicating, can set up purpose SFU address and message arrives the corresponding relation of the employed driver module of this purpose SFU for the processor at SFU place, source in this step, or set up other processors at purpose SFU place and the corresponding relation that message arrives the employed driver module of this processor.
When having between the processor when surpassing one hardware corridor, the SFU on the processor can select different hardware corridors to carry out message communicating.At this moment, can set up the corresponding relation between purpose SFU on source SFU, other processors, message arrives the employed driver module of this purpose SFU from this source SFU the three, so that the user can be that different hardware corridors is specified in communication between different SFU.
The present invention's suggestion proposes unified interface prototype to driver module, makes each driver module according to the interface that this interface prototype gives information and sends, and can further realize the independence with hardware platform like this.
At step S540, for each process is bound a message queue so that the SFU in this process receives message from this message queue.
In the prior art, two kinds of modes that receive message by message queue are arranged.First kind is own message queue of each process binding, when a process is given other processes transmission message, in fact is that message has been sent to the message queue that the purpose process is bound; Each process receives the message that sends to oneself from the message queue of oneself binding.Second kind is that message queue does not depend on the process existence as an independent entity, and this moment, a plurality of processes can receive message from same message queue.
In the SFU message communicating of the present invention, require the message queue of one of each process binding oneself.For the operating system that first kind of mode of employing carried out message sink, do not need to carry out this step; And for the operating system that receives message with the second way, as use very wide VxWorks in the Embedded Application field, then should carry out this step, be message queue of each process creation, and limit each process and can only receive message, thereby construct communication environment according to the invention from the own message queue of being bound.
At step S550, source SFU sends message according to the address of purpose SFU to purpose SFU.For source SFU, when purpose SFU sends message, only need to get final product as parameter with the address of purpose SFU and the message that will send, be positioned on which processor and needn't comprehend purpose SFU, which kind of hardware corridor also needn't comprehend between the processor at the processor at SFU place, source and purpose SFU place is.
Among the present invention, the message of communicating by letter between SFU can comprise message header and message body, wherein comprise the SFU address, source and the purpose SFU address of message in the message header, and message body is used for carrying the concrete data of communicating by letter between two SFU.
At step S560, judge purpose SFU and source SFU whether on same processor, if not, execution in step S570 then; If then the message queue that provides by operating system carries out local message communication, searches the process at purpose SFU place, message is sent in the message queue of purpose SFU place process binding, give purpose SFU by the process at purpose SFU place with message distribution.The process at purpose SFU place can realize giving purpose SFU with message distribution by the message processing function that calls the SFU registration.
Owing to, only need the processor flag at contrast source SFU and purpose SFU place can learn whether two SFU are positioned on the same processor by the processor that can learn its place in the address of SFU.
At step S570, according to SFU address, source and purpose SFU address, search the driver module of in step S530, setting up corresponding relation with it, use this driver module to send message to purpose SFU.
At step S580, when driver module receives the message that sends to purpose SFU, send message to purpose SFU according to the purpose SFU address in the message.
The driver module that receives message is identical with the method that source SFU sends message to the method for purpose SFU transmission message, be positioned at same processor on because receive the driver module and the purpose SFU of message this moment, by the message queue that operating system provides, message is sent to the process at purpose SFU place.
At step S590, the process at purpose SFU place is to purpose SFU dispatch messages.The process at purpose SFU place barrage reception on the message queue of binding sends to the message of this process; When the process at purpose SFU place receives message from the message queue of binding after, according to the purpose SFU address in the message, find the message processing function of this SFU registration, call this this message processing function and handle the message that receives, thereby realize the distribution of message to purpose SFU.
The device of using information communication method between above-mentioned SFU as shown in Figure 6, SFU 610 is connected with the local communication module 660 of SFU information module 650, SFU routing module 620, SFU communication management module 630 and SFU respectively, SFU communication module 630 is connected with driver module group 640 with SFU routing module 620, SFU information module 650, the local communication module 660 of SFU, wherein driver module group 640 comprises sending module group 641 and receiver module group 642, all is connected to SFU communication management module 630; And SFU information module 650 comprises SFU address module 651 and SFU Registering modules 652, all is connected with SFU communication management module 630 with SFU 610.
In SFU routing module 620, storing the SFU address, source on this processor, carry out purpose SFU address on other processors of message communicating and the corresponding relation that sends a message to the employed driver module in this purpose SFU address with this source SFU on this processor.
When having only hardware corridor between per two processors with between SFU thereon during message communicating, can be in SFU routing module 620 storage purpose SFU address and message arrive the corresponding relation of the employed driver module of this purpose SFU, or other processors at storage purpose SFU place and message arrive the corresponding relation of the employed driver module of this processor.
When having between the processor when surpassing one hardware corridor, can in SFU routing module 620, store SFU address, source, purpose SFU address and message and arrive corresponding relation between the employed driver module three of this purpose SFU from this source SFU.
When SFU 610 sends message, send the transmission instruction to SFU communication management module 630, comprise purpose SFU address and the message that is sent in the transmission instruction.SFU communication management module 630 judges whether purpose SFU and source SFU 310 are positioned on the same processor, if on same processor, then will send instruction and transfer to the local communication module 660 of SFU; The local communication module 660 of SFU inquires the process at purpose SFU place from SFU Registering modules 652, message is sent to the message queue of purpose SFU place process; Distribute the messages to purpose SFU by purpose SFU place process.If source SFU 310 and purpose SFU be not on same processor, then to SFU routing module 620 inquiry and the corresponding driver modules of source SFU 610 and purpose SFU address, and transmission of messages transmission driver module of this driver module to the driver module group 640 sent.
When certain the reception driver module in the receiver module group 642 receives the message that sends to SFU 610, this message is passed to SFU communication management module 630.SFU communication management module 630 will send instruction and transfer to the local communication module 660 of SFU; The local communication module 660 of SFU inquires the process at purpose SFU 610 places from SFU Registering modules 652, message is sent to the message queue of purpose SFU 610 place processes; Distribute the messages to SFU 310 by purpose SFU 610 place processes.
The local communication module 660 of SFU is used for finishing the message communicating between SFU on this processor.
After implementing the present invention, the hierarchy that carries out the multicomputer system of communicating by letter between SFU as shown in Figure 7.As seen, the present invention is in upper application software and be employed in the prior art between the driver module set that software directly calls and increased one deck multicomputer system software platform MPSSP, driver module set and upper application software have been isolated by the local communication module of the SFU communication management module, SFU information module, SFU routing module and the SFU that comprise among the MPSSP, for upper application software provides Message Call Interface unified between SFU, also provide unified message sink interface simultaneously for the driver module set.Upper application software can be called corresponding driver module by MPSSP by purpose SFU address, by the multicomputer system hardware platform message is sent to purpose SFU.
Below be the specific implementation of a kind of recommendation of message communicating between SFU among the present invention.
The address of a SFU comprises the numbering and the numbering of this SFU on this processor of this SFU place processor.
The SFU information module will be safeguarded the corresponding relation of message processing function of initialization function, the SFU of process identification (PID), the SFU of SFU address on this processor, SFU place process.The implementation of recommending is to adopt SFU entity structure array, and the SFU entity structure should comprise following member at least:
The process identification (PID) of process under the SFU;
SFU initialization function pointer;
SFU message processing function pointer.
With the latter half of SFU address, promptly the index that be numbered SFU entity structure array of SFU on this processor promptly can find the SFU entity structure corresponding to certain SFU on this processor.
The SFU information module should provide the interface of registration SFU to upper application software, is used for upper application software to SFU of MPSSP registration, and a kind of recommendation prototype of SFU registration function is:
MPSSP_REGISTER_SFU (address designator of SFU,
The Process identifier of process under the SFU,
The function pointer of the initialization function of SFU,
The function pointer of the message processing function of SFU)
The SFU routing module will be safeguarded SFU address, source, purpose SFU address, driver module three's corresponding relation.Corresponding relation with this three in the implementation of recommending is divided into the realization of two steps, and the first step realizes the corresponding relation of SFU address, source, purpose SFU address, driving numbering, and the realization of second step drives the corresponding relation of numbering with driver module.According to SFU address, source and purpose SFU address, can find the driver module that source SFU and purpose SFU carry out the required use of inter-processor communication like this.
The suggested design of realizing first step corresponding relation is that the routing rule entity structure should comprise following member at least for each SFU on this processor safeguards a routing rule entity structure chained list:
The address of purpose SFU;
Drive numbering;
Point to the pointer of next bar routing rule entity structure.
When needs drive numbering according to SFU address, source and purpose SFU address lookup, at first find the routing rule entity structure chained list of this SFU according to the numbering of source SFU on this processor, search for this chained list then, find the routing rule entity corresponding, drive numbering thereby find with purpose SFU.
The SFU routing module should provide the interface that adds routing rule to upper application software, is used for upper application software and adds a route to MPSSP, and a kind of recommendation prototype that routing rule adds function is:
MPSSP_REGISTER_RULE (address of purpose SFU,
The numbering that drives)
The suggested design of realizing the second step corresponding relation is to adopt to drive the entity structure array, drives entity structure and should comprise following member at least:
The transmission interface function pointer of driver module.
Be numbered the index that drives the entity structure array with driving, the SFU communication management module can find the transmission interface function of corresponding driver module.
The SFU routing module should provide the interface of registration to driver module, is used for to driver module of MPSSP registration, and a kind of recommendation prototype that drives the registration interface function is:
MPSSP_REGISTER_DD (drive numbering,
The transmission interface function pointer of driver module)
The SFU message transmission interface that the SFU communication management module will provide to upper application software.The recommendation prototype of SFU message transmission interface function is:
MPSSP_SEND (message pointer)
The message of message pointed comprises message header and message body, comprises SFU address, source and purpose SFU address in the message header.
The flow process of SFU message transmission interface is as follows:
A. according to purpose SFU address, judge whether purpose SFU is on this processor,, carry out b, otherwise carry out c if at this processor;
B. purpose SFU is on this processor, call the query interface that the SFU information module provides, find the process identification (PID) of purpose SFU place process, the message queue message that the call operation system provides sends function, message is sent to the purpose process, and message sends successfully and returns;
C. purpose SFU is not on this processor, call the routing rule query interface that the SFU routing module provides, find the driving numbering that sends a message to the required use of purpose SFU, according to driving numbering, find the transmission interface function pointer of this driver module, the transmission interface function of calling driver module sends message by hardware corridor, send successfully and return.
Driver module calls the SFU message transmission interface function that the SFU communication management module provides after receiving the message that SFU on other processors sends, message is sent to purpose SFU on this processor.
The present invention has improved the multicomputer system portability of software at the middle and upper levels to a great extent, has reduced the multicomputer system design complexities of software at the middle and upper levels simultaneously to a great extent.
Above-described embodiment of the present invention does not constitute the qualification to protection domain of the present invention.Any any modification of being done within the spirit and principles in the present invention, be equal to and replace and improvement etc., all should be included within the claim protection domain of the present invention.
Claims (18)
1. the information communication method between SFU software functional unit SFU in the multicomputer system, wherein SFU software functional unit SFU is the communication entity that is under the jurisdiction of process, it is characterized in that, may further comprise the steps:
SFU software functional unit SFU comprises the information registration of its address, can be learnt the processor at this SFU place by described address;
Set up that the purpose SFU software functional unit SFU address on same processor and message do not arrive the corresponding relation of the employed driver module of this purpose SFU software functional unit SFU with source software functional unit SFU;
Source software functional unit SFU sends message according to the address of purpose SFU software functional unit SFU to it;
When purpose SFU software functional unit SFU and source software functional unit SFU are not on same processor, use the driver module of this purpose SFU address correspondence to carry out message and send.
2. according to the described information communication method of claim 1, it is characterized in that: when between the processor a plurality of hardware corridor being arranged, the corresponding relation of setting up purpose SFU software functional unit SFU address and driver module is specially: set up source software functional unit SFU address, and this source software functional unit SFU do not arrive corresponding relation between the employed driver module three of this purpose SFU from this source SFU in the purpose SFU software functional unit SFU address on the same processor, message.
3. according to the described information communication method of claim 2, it is characterized in that, described method also comprises: when the driver module of purpose SFU software functional unit SFU place processor receives described message, send message to it according to the address of purpose SFU software functional unit SFU in the described message on same processor.
4. according to the described information communication method of claim 3, it is characterized in that described method also comprises: set up the corresponding relation between SFU software functional unit SFU address and this SFU software functional unit SFU place process;
Described purpose SFU software functional unit SFU on same processor sends message and is specially:
Message is sent to the message queue of purpose SFU software functional unit SFU place process;
Receive the address of the process of message, give purpose SFU message distribution according to purpose SFU software functional unit SFU in the message.
5. according to the described information communication method of claim 4, it is characterized in that described method also comprises: for message queue of each process binding receives message for SFU software functional unit SFU in this process.
6. according to the described information communication method of claim 5, it is characterized in that: the log-on message of SFU software functional unit SFU also comprises the message processing function of this SFU software functional unit SFU;
The described process that receives message is specially to purpose SFU software functional unit SFU message distribution:
Barrage reception sends to the message of described process on the message queue of described process;
According to the message processing function of purpose SFU software functional unit SFU address search purpose SFU software functional unit SFU registration in the message, call this message processing function and handle the message that receives.
7. according to any described information communication method of claim 1 to 6, it is characterized in that: the address of described SFU software functional unit SFU comprises the processor flag at this SFU software functional unit SFU place and the sign of this SFU software functional unit SFU, wherein processor flag has uniqueness in multicomputer system, and SFU software functional unit SFU is identified on the processor at this SFU place has uniqueness.
8. the message communicating device between SFU software functional unit SFU in the multicomputer system, it is characterized in that, comprise SFU software functional unit SFU, SFU software functional unit SFU communication management module, SFU software functional unit SFU routing module, SFU information module and driver module group, wherein:
SFU software functional unit SFU information module is used for storing the SFU software functional unit SFU information that comprises SFU software functional unit SFU address, can be learnt the processor at this SFU software functional unit SFU place by described address;
SFU software functional unit SFU routing module is used for storing source software functional unit SFU address on this processor, receive purpose SFU software functional unit SFU address on other processors of message, send a message to the corresponding relation of the employed driver module of this purpose SFU software functional unit SFU;
The driver module group comprises at least one driver module, be used for other processors on corresponding driver module carry out message communicating;
SFU software functional unit SFU communication management module is used for sending and reception message, when purpose SFU software functional unit SFU address is on other processors in the transmission instruction of SFU software functional unit SFU, find corresponding driver module by the described corresponding relation of storing in the SFU software functional unit SFU routing module, transmission of messages is sent to described driver module.
9. according to the described message communicating device of claim 8, it is characterized in that described device also comprises the local communication module of SFU, is used for proceeding to the communication of purpose SFU software functional unit SFU on this processor;
Purpose SFU software functional unit SFU address is on this processor in the transmission instruction of SFU, and SFU software functional unit SFU communication management module will send instruction and transfer to the local communication module of SFU software functional unit SFU.
10. according to the described message communicating device of claim 9, it is characterized in that described SFU software functional unit SFU information module comprises SFU software functional unit SFU address module and SFU software functional unit SFU Registering modules, wherein:
SFU software functional unit SFU address module is used for storing the SFU address, and SFU software functional unit SFU address comprises processor flag and SFU software functional unit SFU unique identification on the processor under it at SFU software functional unit SFU place;
The SFU Registering modules is used for storing the log-on message of SFU, comprises the process identification (PID) at SFU software functional unit SFU place.
11. according to the described message communicating device of claim 10, it is characterized in that, described driver module group comprises sending module group and receiver module group, and wherein the sending module group comprises that at least one sends driver module, and the SFU software functional unit SFU that is used on other processors sends message; The receiver module group comprises that at least one receives driver module, be used for receiving the message that sends to SFU software functional unit SFU on this processor and with this transmission of messages to SFU software functional unit SFU communication management module;
SFU software functional unit SFU communication management module with this transmission of messages to the local communication module of SFU software functional unit SFU, the local communication module of SFU software functional unit SFU is searched the process at purpose SFU software functional unit SFU place from SFU software functional unit SFU Registering modules, message is distributed to purpose SFU by described process.
12. the information communication method in the multicomputer system between process is characterized in that, may further comprise the steps:
The address of setting process comprises the processor flag at this process place in the described address;
Set up that the purpose process address on same processor and message do not arrive the corresponding relation of the employed driver module of this purpose process with originating process;
Originating process sends message according to the address of purpose process to it;
When purpose process and originating process are not on same processor, use the driver module of this purpose process address correspondence to carry out message and send.
13. according to the information communication method between process in the described multicomputer system of claim 12, it is characterized in that: when between the processor a plurality of hardware corridor being arranged, the corresponding relation of setting up purpose process address and driver module is specially: set up the originating process address, and this originating process do not arrive corresponding relation between the employed driver module three of this purpose process from this originating process in the purpose process address on the same processor, message.
14. according to the information communication method between process in the described multicomputer system of claim 13, it is characterized in that, described method also comprises: after the driver module on the processor of purpose process place receives described message, according to the address of purpose process message is sent to the purpose process.
15., it is characterized in that described method also comprises according to the information communication method between process in any described multicomputer system of claim 12 to 14:
Judge that according to the originating process address that sends message and purpose process address originating process and purpose process are whether on same processor;
When purpose process and originating process are on same processor, use the message queue of operating system to send message to the purpose process.
16. the message communicating device in the multicomputer system between process is characterized in that, comprises process, process communication administration module, process address module, process routing module and driver module group, wherein:
The process address module is used for the address information of storage process, comprises the processor flag at process place;
The process routing module is used for storing originating process sign on this processor, receive purpose process address on other processors of message, send a message to the corresponding relation of the employed driver module of this purpose process;
The driver module group comprises at least one driver module, be used for other processors on corresponding driver module carry out message communicating;
The process communication administration module is used for sending and reception message, when the purpose process address is on other processors in the transmission instruction of process, find corresponding driver module by the described corresponding relation of storing in the process routing module, transmission of messages to described driver module is sent.
17., it is characterized in that described device also comprises the local communication module of process, is used for proceeding to the communication of purpose process on this processor according to the message communicating device between process in the described multicomputer system of claim 16;
The purpose process address is on this processor in the transmission instruction of process, and the process communication administration module will send instruction and transfer to the local communication module of process.
18. according to the message communicating device between process in the described multicomputer system of claim 17, it is characterized in that, described driver module group comprises sending module group and receiver module group, wherein the sending module group comprises that at least one sends driver module, and the process that is used on other processors sends message; The receiver module group comprises that at least one receives driver module, be used for receiving the message that sends to process on this processor and with this transmission of messages to the process communication administration module;
The process communication administration module sends to the purpose process by the message that the local communication module of process will receive the driver module reception.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005101127481A CN100424675C (en) | 2005-10-12 | 2005-10-12 | Information communication method and apparatus of multiprocessor system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2005101127481A CN100424675C (en) | 2005-10-12 | 2005-10-12 | Information communication method and apparatus of multiprocessor system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1949206A CN1949206A (en) | 2007-04-18 |
| CN100424675C true CN100424675C (en) | 2008-10-08 |
Family
ID=38018726
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2005101127481A Active CN100424675C (en) | 2005-10-12 | 2005-10-12 | Information communication method and apparatus of multiprocessor system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100424675C (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2014082421A1 (en) * | 2012-11-30 | 2014-06-05 | 华为技术有限公司 | Method, device and system for inter-process communication of many-core processor |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2012159305A1 (en) * | 2011-06-28 | 2012-11-29 | 华为技术有限公司 | Distributed multi-process communication method and device |
| CN102541644B (en) * | 2011-12-30 | 2014-12-24 | 深圳万兴信息科技股份有限公司 | Method and device for message driving based on state machine |
| CN102867018B (en) * | 2012-07-27 | 2015-04-22 | 北大方正集团有限公司 | Method for analogue signal communication between threads in database system |
| CN103995739B (en) * | 2014-04-24 | 2017-12-19 | 深圳中微电科技有限公司 | The processor system of shared drive mapping page produces the method and device of process number |
| CN106155975A (en) * | 2015-04-07 | 2016-11-23 | 深圳市蓝海华腾技术股份有限公司 | The drive circuit of CAN mouth interconnection between multiple processors |
| CN110572348B (en) * | 2018-06-06 | 2020-11-13 | 大唐移动通信设备有限公司 | Message communication method and device between network elements |
| CN108810645B (en) * | 2018-06-28 | 2021-01-01 | 武汉斗鱼网络科技有限公司 | Bullet screen message distribution method, device, equipment and storage medium |
| CN111045801A (en) * | 2019-11-19 | 2020-04-21 | 蓝箭航天空间科技股份有限公司 | Unified inter-task communication method supporting multiple processors and rocket communication method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6157960A (en) * | 1997-05-07 | 2000-12-05 | International Business Machines Corporation | Technique for programmatically creating distributed object programs |
| CN1288201A (en) * | 1999-09-09 | 2001-03-21 | 上海贝尔有限公司 | Communication interface of master-slave type processor system |
| US20030182355A1 (en) * | 2002-03-20 | 2003-09-25 | Nec Corporation | Parallel processing system by OS for single processor |
| CN1670721A (en) * | 2004-03-17 | 2005-09-21 | 日本电气株式会社 | Inter-processor communication system in parallel processing system by os for single processors and program thereof |
-
2005
- 2005-10-12 CN CNB2005101127481A patent/CN100424675C/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6157960A (en) * | 1997-05-07 | 2000-12-05 | International Business Machines Corporation | Technique for programmatically creating distributed object programs |
| CN1288201A (en) * | 1999-09-09 | 2001-03-21 | 上海贝尔有限公司 | Communication interface of master-slave type processor system |
| US20030182355A1 (en) * | 2002-03-20 | 2003-09-25 | Nec Corporation | Parallel processing system by OS for single processor |
| CN1670721A (en) * | 2004-03-17 | 2005-09-21 | 日本电气株式会社 | Inter-processor communication system in parallel processing system by os for single processors and program thereof |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2014082421A1 (en) * | 2012-11-30 | 2014-06-05 | 华为技术有限公司 | Method, device and system for inter-process communication of many-core processor |
| US9678813B2 (en) | 2012-11-30 | 2017-06-13 | Huawei Technologies Co., Ltd. | Method, apparatus, and system for mutual communication between processes of many-core processor |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1949206A (en) | 2007-04-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100424675C (en) | Information communication method and apparatus of multiprocessor system | |
| CA1272809A (en) | Dynamically partitionable parallel processors | |
| US6854021B1 (en) | Communications between partitions within a logically partitioned computer | |
| EP0169455B1 (en) | Non-unique names for broadcast messages | |
| US5245608A (en) | Logical grouping of layer entities in a layered communication architecture | |
| US20020124108A1 (en) | Secure multiprotocol interface | |
| CN103368848A (en) | Information processing apparatus, arithmetic device, and information transferring method | |
| CN100552623C (en) | Be used between system processor and coprocessor, carrying out the system and method that message transmits | |
| CN103942178A (en) | Communication method between real-time operating system and non-real-time operating system on multi-core processor | |
| CN104123265A (en) | Method and system for communication among multiple kernels | |
| CN110278161A (en) | Message diversion method, apparatus and system based on User space protocol stack | |
| CN101771724B (en) | Heterogeneous distributed information integration method, device and system | |
| CN109802951A (en) | A kind of message forwarding method, equipment and storage equipment, program product | |
| CN101931605B (en) | Inter-board communication control device and method for transmitting and receiving inter-board communication information | |
| Von Bochmann | Concepts for distributed systems design | |
| JPS5962245A (en) | Local area network | |
| CN115328645A (en) | Computing task scheduling method, computing task scheduling device and electronic equipment | |
| CN104123252A (en) | Multi-channel direct memory access controller and control method thereof | |
| CN102054213A (en) | Information integration method, device and system | |
| CN112506676A (en) | Inter-process data transmission method, computer device and storage medium | |
| CN109218156A (en) | A kind of management method of Dynamic link library data | |
| US5528768A (en) | Multiprocessor communication system having a paritioned main memory where individual processors write to exclusive portions of the main memory and read from the entire main memory | |
| CN116405555B (en) | Data transmission method, routing node, processing unit and system on chip | |
| US7426531B1 (en) | Distributed electronic data processing | |
| Singhvi et al. | Prophet: a contextual information system framework |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |