JP2002057703A - Data conversion processing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a data switching processing device that can flexibly core with multi-protocols and ensure a sufficient transmission efficiency. SOLUTION: The data switching processing device outputs data with various protocols received from input ports Pi1-Pi5 to prescribed output ports Po1-Po5. A parallel processing processor 30 receives the received data via an input multiplexer 20. A packet generating section 31 generates packets including the data. The head of the packets includes information denoting the protocol of the data. Processor elements 33 apply processing to each packet depending on each protocol in order to confirm a transmission destination or the like of each packet. Finally, the information denoting an output port depending on the transmission destination is added to each packet and the resulting packets are outputted to an output multiplexer 40. The output multiplexer 40 outputs the data in each packet to the instructed ports.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data exchange processing apparatus, and more particularly to a data exchange apparatus capable of performing exchange processing on data transmitted by a plurality of different communication protocols using a data driven parallel processing processor. It relates to a processing device.

[0002]

2. Description of the Related Art In recent years, with the spread of computer networks, demands for various data exchange processing devices such as switching hubs and routers have been rapidly increasing. A basic problem required of such a data exchange processing device is an exchange process for distributing data input to a plurality of input ports to a plurality of output ports and outputting the data.
The data flowing on the computer network is handled in units of one unit, and the data exchange processing device functions to control the flow of data in units. Generally, processing executed in the data exchange processing device is divided into segment processing and routing processing. The segment process is a process of performing an error check, a life check, and the like on each data unit that has entered the input port, and the routing process outputs a data unit that has been subjected to the segment process to an output port connected to a correct destination. Processing.
Whichever processing is performed, it is necessary to analyze the inside of the data unit and make a predetermined judgment.

[0003] As described above, the data exchange processing device needs to perform an analysis process and various judgment processes on an input data unit, and therefore needs to incorporate a processor unit. A conventional general data exchange processing device incorporates a general-purpose Neumann-type processor and a dedicated ASIC processor, and executes these processes. Recently, a data exchange processing device using a non-Neumann-type parallel processor has been proposed.

[0004]

Recently, with the diversification of computer networks, various communication protocols have become widespread, and a network configuration in which a plurality of communication protocols are mixed has become common.
For this reason, the demand for a data exchange processing device that supports a plurality of communication protocols (so-called multiprotocol) is increasing.

The most basic method for supporting such a multi-protocol is to provide a plurality of hardware corresponding to each protocol. According to this method, since dedicated hardware performs processing for each protocol, the load on software is not increased, but there is a problem of lack of flexibility. For example, a request to change the protocol of one port to another protocol has to be dealt with by a major exchange of hardware. As a software method for supporting multi-protocols, there is a so-called encapsulation.
Is generally used. In this method, a method is used in which the entire data unit described in the first protocol is completely encapsulated and wrapped in the second protocol so that the data unit can be apparently handled as data of the second protocol. However, the processing for performing encapsulation and the processing for extracting the original data unit from the capsule are complicated, so that the load on software is increased and the transmission efficiency is inevitably reduced.

Accordingly, an object of the present invention is to provide a data exchange processing apparatus which can flexibly cope with a multi-protocol and which can secure a sufficient transmission efficiency.

[0007]

According to a first aspect of the present invention, there is provided a data exchange processing device for performing a data exchange process between a plurality of input ports and a plurality of output ports. A plurality of data input interfaces connected to each other for inputting a protocol data unit having a predetermined communication protocol, and connected to each output port for outputting a protocol data unit having a predetermined communication protocol; A plurality of data output interfaces, a data driven parallel processing processor for executing parallel processing on a given plurality of data packets, and a protocol data unit input by each data input interface are provided to the parallel processing processor. The input multiplexer and the output from the parallel processor An output multiplexer for providing a protocol data unit to a predetermined data output interface, each data input interface having a function of externally inputting a protocol data unit having a specific protocol, and Interface has a function of outputting a protocol data unit having a specific protocol to the outside, and the input multiplexer provides information indicating the protocol of the protocol data unit when giving each protocol data unit to the parallel processing processor. To the parallel processor, the parallel processor generates a data packet to be processed, and the packet generator according to a specific program loaded in advance. A plurality of processor elements for performing a predetermined process and an I / O router having a function of sending each data packet to a predetermined destination are configured. When information indicating a protocol is given, an auxiliary field containing information indicating the protocol, a generation number field containing a generation number assigned to be unique in the parallel processing processor, A data packet having an instruction field containing the first execution instruction to be determined and a data field containing the protocol data unit is generated.
The I / O router executes processing to be given to the / O router, and transfers the given data packet to a specific processor element or outputs the given data packet based on an instruction contained in an instruction field in each data packet. The processor element executes processing for outputting to the multiplexer, and each processor element executes an instruction in an instruction field when a data packet is transferred from the I / O router, and according to the execution result and a protocol in an auxiliary field. A new data packet in which the contents of the data field and the instruction field are rewritten by using the program, and providing the new data packet to the I / O router as necessary. The data in the data field of the data packet output from the O router Extracted as Tayunitto, which is obtained so as to execute processing for outputting a predetermined data output interface.

(2) A second aspect of the present invention is the above-mentioned first aspect.
In the data exchange processing device according to the aspect, the parallel processing processor analyzes the transmission destination address of the protocol data unit included in the data field of the data packet to be processed, thereby obtaining the correct output port for the protocol data unit. Is performed, and the information indicating the output port is written in the auxiliary field of the data packet output from the I / O router. By referring to the data packet, a process of determining a data output interface to output the protocol data unit extracted from the data packet is performed.

(3) A third aspect of the present invention is the above-described first aspect.
In the data exchange processing device according to the aspect, the parallel processing processor analyzes the transmission destination address of the protocol data unit included in the data field of the data packet to be processed, thereby obtaining the correct output port for the protocol data unit. To generate a port designating data packet containing information for specifying a correct output port for a protocol data unit included in a data packet with a specific generation number, and output the data packet to an output multiplexer. And the output multiplexer refers to the information of the port instruction data packet output from the I / O router, thereby extracting the protocol data unit extracted from the data packet with the specific generation number. For data output to be output It is obtained to perform the process of determining the interface.

(4) The fourth aspect of the present invention is the above-mentioned second aspect.
Or in the data exchange processing device according to the third aspect,
A processor element is provided among a plurality of processor elements for performing an arithmetic operation for obtaining a specific output port corresponding to a specific destination address based on a table indicating a correspondence relationship between a destination address and an output port. Things.

(5) A fifth aspect of the present invention is the above-mentioned first aspect.
In the data exchange processing device according to the fourth to fourth aspects, the parallel processing processor analyzes the data in the protocol data unit included in the data field of the data packet to be processed, so that the life of the protocol data unit is reduced. When it is confirmed that the data packet is exhausted, a process of deleting the data packet without outputting it to the output multiplexer is performed.

(6) The sixth aspect of the present invention is the above-mentioned first aspect.
In the data exchange processing device according to any one of the fourth to fourth aspects, the parallel processing processor analyzes data in the protocol data unit included in the data field of the data packet to be processed, thereby generating an error related to the protocol data unit. When it is confirmed that there is an abnormality in the consistency of the check code, a process of deleting the data packet without outputting it to the output multiplexer is performed.

(7) The seventh aspect of the present invention is the above-mentioned first aspect.
In the data exchange processing device according to the fourth to fourth aspects, the parallel processor analyzes the data in the protocol data unit included in the data field of the data packet to be processed, and When it is confirmed that the data packet is destined for the destination, the data packet is not output to the output multiplexer, and the process of receiving the transmission target data included in the protocol data unit as the data destined for itself is performed. Things.

(8) An eighth aspect of the present invention is the above-mentioned first aspect.
In the data exchange processing device according to the seventh to seventh aspects, the parallel processing processor performs a process of converting a communication protocol of a protocol data unit included in a data field of a data packet to be processed into another communication protocol. It was made.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below based on an embodiment shown in the drawings.

§1. The data exchange processing device according to the present invention
Basic Configuration FIG. 1 is a block diagram showing a basic configuration of a data exchange processing device according to one embodiment of the present invention. Here, for convenience of explanation, a data exchange processing device having a simple configuration for performing data exchange processing between five input ports Pi1 to Pi5 and five output ports Po1 to Po5 will be described. In practical use, it is general to configure a device for performing a switching process between a larger number of input / output ports. A feature of the data exchange processing device according to the present invention is that it supports a multi-protocol. Here, "Ethern
Although an example corresponding to two types of protocols, that is, an “et” protocol and an “ATM” protocol is shown, it is needless to say that the present invention can support a larger number of protocols. For example, the present invention is applicable to protocols such as IPv4, IPv6, and MPLS. In the illustrated example, the input ports Pi1, Pi2, and Pi3 are ports corresponding to the “Ethernet” protocol.
Pi5 is a port corresponding to the "ATM" protocol. The output ports Po1, Po2, and Po3 are "Et"
The output ports Po4 and Po5 are ports compatible with the "ATM" protocol.

The main components of this data exchange processing device are a data input interface 11 to 15, an input multiplexer 20, a data driven parallel processing processor 30, an output multiplexer 40, and a data output interface 51 to 55. It is. In the parallel processing processor 30, a packet generator 31 and an I / O router 3
2 and a number of nanoprocessor elements 33.

Data input interfaces 11 to 15
Is connected to each of the input ports Pi1 to Pi5 and has a function of inputting a protocol data unit PDU having a predetermined communication protocol. Specifically, each interface is configured by an interface board equipped with hardware dedicated to a specific protocol, and each of the interfaces 11 to 13 has hardware for receiving a signal of the “Ethernet” protocol, Each of the interfaces 14 and 15 has hardware for receiving a signal of the "ATM" protocol. Similarly, the data output interfaces 51 to 55 are connected to the output ports Po1 to Po5, respectively, and have a function of outputting a protocol data unit PDU having a predetermined communication protocol. Specifically, each interface is configured by an interface board equipped with hardware dedicated to a specific protocol.
53 have hardware for transmitting a signal of the "Ethernet (registered trademark)" protocol, and the interfaces 54 and 55 both have hardware for transmitting a signal of the "ATM" protocol.

The input multiplexer 20 has a function of providing the protocol data unit PDU input by the interfaces 11 to 15 to the packet generator 31 in the parallel processor 30. Conversely, the output multiplexer 40 outputs the protocol data unit PDU output from the parallel processing processor 30 (actually, as described later, as a part of the data packet DP).
To one of the data output interfaces 51 to 55. Which interface to output is determined by the arithmetic processing in the parallel processing processor 30. Parallel processing processor 30
Is a data-driven parallel processor that executes parallel processing on a given plurality of data packets, and its specific operation will be described later.

Data input interfaces 11 to 15
Has a function of externally inputting a protocol data unit PDU having a specific protocol as described above. The format of the protocol data unit PDU is different for each individual protocol. The upper part of FIG. 2 shows an example of the format of a general protocol data unit PDU. In this example, the protocol data unit PDU includes “identification information”, “destination address”, “source address”, “data length”, “transmission target data”, and “ECC”. The “transmission target data” is the original transmission target data (in some cases, a higher-layer protocol data unit is accommodated as “transmission target data”),
The “data length” contains information indicating the length of the “transmission target data”. The “destination address” and the “source address” are address data indicating the source and destination of the “data to be transmitted”, for example, “Ethe
In the case of the “rnet” protocol, the MAC address of each device is accommodated. The first "identification information" contains information indicating the protocol of the protocol data unit PDU, the information indicating the version thereof, the information indicating the lifetime, and the like, and the "ECC" at the end stores an error check code.

Actually, the exact format of the protocol data unit PDU shown in the upper part of FIG. 2 differs for each protocol, and the electrical characteristics of signals in the physical layer also differ for each protocol. Is different. Data input interface 11-1
Reference numeral 5 denotes an interface board equipped with hardware dedicated to a specific protocol as described above, and converts signals input from the respective input ports Pi1 to Pi5 into a protocol data unit PDU having a predetermined format. Input multiplexer 20
Has the function of sending to The input multiplexer 20
In this way, a process of giving the protocol data unit PDU sent from each of the data input interfaces 11 to 15 to the packet generation unit 31 in the parallel processor 30 is performed. At this time, information indicating the protocol of each protocol data unit PDU is performed. Also communicate with them. The input multiplexer 20 previously stores information indicating which data input interface corresponds to which protocol, and the input multiplexer 20 refers to this information and refers to each protocol data unit PDU.
Then, a process of adding information indicating the protocol to the packet generation unit 31 is performed. For example, regarding the protocol data unit PDU sent from the data input interface 11, "Ethern"
After adding information indicating that the protocol is the “et” protocol, a process of sending the packet to the packet generation unit 31 is performed.

As shown in FIG. 1, the parallel processing processor 30 generates a data packet DP to be processed in the parallel processing processor 30 and sends each data packet DP to a predetermined destination. It has an I / O router 32 having a sending function, and a plurality of nanoprocessor elements 33 for executing a predetermined process on a data packet DP in accordance with a specific program loaded in advance. In the parallel processor 30, all data to be processed are data packets D.
Handled in the form of P. Therefore, the input multiplexer 20
The protocol data unit PDU sent to the parallel processing processor 30 from is converted into the data packet DP format in the packet generator 31.

In the present invention, the data packet DP handled by the data driven parallel processor 30 is described in a format as shown in the lower part of FIG. That is, the data packet DP includes an auxiliary field, a generation number field, an instruction field, and a data field. The data field is a field for storing data to be processed in the parallel processing processor 30. In the present invention, the protocol data unit PDU is directly stored in the data field. The instruction field contains the first execution instruction for the data packet DP. Generally, in a data driven parallel processing processor, the content of "going to which processor element, receiving what kind of instruction there, and then going to which processor element depending on the result" is written in the instruction field as an execution instruction. Written.

The generation number field contains a generation number for specifying the data packet DP. This generation number needs to be assigned so as to be unique within the parallel processing processor 30, and a large number of data packets DP simultaneously existing in the parallel processing processor 30 at a certain point in time should not be duplicated. A unique generation number is assigned to each. The auxiliary field contains information indicating the protocol of the protocol data unit PDU contained in the data field of the data packet DP. As described above, the information indicating the protocol of the protocol data unit PDU is described at the top as identification information (see FIG. 2).
), The auxiliary field contains the same information. In this embodiment, the information indicating the output port is also stored in the auxiliary field, which will be described later.

As described above, by referring to the identification information in the protocol data unit PDU contained in the data field, it is possible to recognize the protocol of the protocol data unit PDU. The reason why the information indicating the protocol is stored in the auxiliary field is that the protocol can be recognized without referring to the data in the data field. As described above, the format of the protocol data unit PDU shown in the upper part of FIG. 2 differs for each protocol. Therefore, it is not possible to recognize from what byte to what byte from the beginning the identification information indicating the protocol is described unless the protocol is known. In other words, the identification information indicating the protocol in the protocol data unit PDU is "key in a locked box". "A key is required to open the box, but the key is the box. In. " The information indicating the protocol contained in the auxiliary field of the data packet DP functions as a so-called "another key outside the box". In short, if the information indicating the protocol is stored in the auxiliary field provided separately from the data field, the protocol can be specified without looking inside the data field.

§2. The data exchange processing device according to the present invention
Processing Operation Next, in order to make the relationship between the protocol data unit PDU and the data packet DP shown in FIG. 2 easier to understand, a case where a predetermined protocol data unit PDU is sent to the input port Pi1 will be described. The processing operation of the generation unit 31 will be described. As described above, the data of the “Ethernet” protocol is sent to the input port Pi1. The data sent in this way is a protocol data unit PDU,
The data is sent from the data input interface 11 to the input multiplexer 20 and is supplied to the packet generator 31. At this time, the input multiplexer 20 determines that the protocol data unit PDU is "Etherne".
t "protocol is transmitted to the packet generation unit 31 (as described above, the input multiplexer 20 transmits the protocol data unit PD
Since U is data sent from the data input interface 11, the protocol is "Eth
ernet ").

Thus, the packet generator 31 stores the protocol data unit PDU and the protocol "E
information which is "thernet". Therefore, the packet generation unit 31 first creates a data field containing the entire protocol data unit PDU as it is. Subsequently, an instruction field containing a predetermined execution instruction is added thereto. The execution instruction is an initial instruction that is executed first in the parallel processing processor 30. “The first instruction goes to which processor element, receives what instruction, and then, depending on the result, Go "is the execution instruction. What kind of initial command is stored in the command field is determined in advance. Next, a generation number field containing a predetermined generation number and an auxiliary field containing information indicating a protocol are added.
A data packet DP having a format as shown in the lower part of FIG. As described above, the generation number needs to be unique within the parallel processing processor 30. For example, a serial number or the like is sequentially added. As the information indicating the protocol to be accommodated in the auxiliary field, the information transmitted from the input multiplexer 20 may be used as it is. Note that the information indicating the output port has not yet been written in the auxiliary field at this time.

The initial command stored in the command field in the packet generator 31 may be changed for each protocol as needed. For example,
In the case of the protocol data unit PDU of the “Ethernet” protocol, the first initial command is used and “AT
In the case of the protocol data unit PDU of the "M" protocol, it is possible to make a setting such as the second initial command. In this case, the initial destination and the content of processing to be executed differ depending on the protocol of the data packet DP. Of course, "Etherne
In the case of the "t" protocol and in the case of the "ATM" protocol, exactly the same common initial command can be provided. As described later, in the nanoprocessor element 33, a different program can be selected for each protocol. Therefore, even if the same common initial instruction is given, the processing content executed by each nanoprocessor element 33 is determined by the protocol. It becomes possible to change every time.

Thus, the data packet DP generated by the packet generator 31 is sent to the I / O router 32. Thus, the data input interface 1
Each time a protocol data unit PDU arrives at 1 to 15, the packet generator 31 generates a data packet DP including each protocol data unit PDU and sends it to the I / O router 32. I / O
The router 32 performs a process of transferring the given data packet DP to a specific processor element 33 or outputting the data packet DP to an output multiplexer 40 based on an instruction contained in an instruction field in each data packet DP. . That is, the instructions in the instruction field include:
"Go to the next specific nanoprocessor element 33" or "Go to the output multiplexer 40 (when all processing for the data packet DP is completed)"
Is specified, the I / O router 32
Performs the process of distributing each data packet DP according to this destination.

On the other hand, a predetermined program is loaded on each nanoprocessor element 33 in advance. Normal,
Each of the nanoprocessor elements constituting the data driven parallel processing processor has a built-in RAM, and a predetermined program is loaded in the RAM at the time of startup. When the data packet DP is transferred from the I / O router 32, each nanoprocessor element 33 executes a specific process by using the loaded program. At this time, a program to be used for processing can be selected according to the protocol in the auxiliary field. For example, as a program to be loaded into the RAM, "Eth
ernet "protocol, and the" AT
The second routine for the “M” protocol is prepared, and the protocol in the auxiliary field of the data packet DP transferred from the I / O router 32 is “Eth
ernet ", the first routine is selected, and if it is" ATM ", the second routine is selected, and processing using each selected routine is performed. Good.

In the nanoprocessor element 33, when the execution of the instruction in the instruction field and the predetermined program processing using the result for the predetermined data packet DP are completed, the contents of the data field and the instruction field are rewritten. Thus, a new data packet DP is generated. This new data packet DP
May be inherited the same generation number as the original data packet DP, or may be generated as a completely new data packet having a different generation number. Thus, the destination of the new data packet DP generated in the nanoprocessor element 33 is also determined by the contents of the instruction field. Therefore, there is a case where the instruction stays in the same nanoprocessor element 33 as it is and a new instruction is executed, or a case where the instruction is returned to the I / O router 32 and then transferred to another nanoprocessor element 33. There is also.

FIG. 3 is a block diagram showing a basic internal configuration of a nanoprocessor element used in a general data driven parallel processor. First, I /
Data packet DP transferred from O-router 32
Is sent to the waiting storage unit MM via the external input unit IN, and stays in the waiting storage unit MM until conditions necessary for executing a predetermined logical operation indicated by the instruction in the instruction field are satisfied. (For example, when performing a logical operation with another data packet DP, the queue storage unit MM waits until the data packet DP of the other party arrives.) When the necessary conditions are satisfied, the data packet DP in the waiting storage unit MM is
The data is sent to the arithmetic processing unit ALU, and a predetermined logical operation is performed on the data in the data field based on the instruction in the instruction field. The “data-driven type” means that the process is executed when necessary data is prepared in the waiting storage unit MM. The result after the logical operation in the arithmetic processing unit ALU is sent to the program storage unit PM. R in the program storage unit PM
As described above, the AM is loaded with a predetermined program in advance, and the processing specified by this program is executed based on the result of the logical operation.

The program in the program storage unit PM is
For example, a process of writing a new instruction in the instruction field of the data packet DP in which the content of the data field has been rewritten by a logical operation is performed. The instruction field updated by this program includes the next destination of the data packet DP (may be another nanoprocessor element 33, the same nanoprocessor element 33, or the output multiplexer 40). Is specified, and an instruction such as what kind of logical operation is to be performed at the next destination and as a result, where to go further are written.
As described above, different programs can be loaded in the program storage unit PM depending on the protocol. If a plurality of programs have been loaded into the program storage unit PM, the program for the corresponding protocol is selectively executed by referring to the information indicating the protocol written in the auxiliary field of the data packet DP. .

The program in the program storage unit PM is
If necessary, a process of generating a completely new data packet DP can be performed. Such processing can be said to be the same processing as the data packet DP generation processing performed by the packet generation unit 31. For example, a new data field is generated based on the result of the logical operation performed by the arithmetic processing unit ALU, a predetermined instruction field is added thereto, and a generation number field containing a unique new generation number is added. However, if an auxiliary field containing information indicating a predetermined protocol is added, a new data packet DP can be generated.

Thus, data packet D whose contents of the instruction field are updated in program storage unit PM
P or the data packet DP newly generated in the program storage unit PM is sent to the external output unit OUT, and is output therefrom to the I / O router 32 (the destination is another nanoprocessor element 33). Alternatively, in the case of the output multiplexer 40, the signal is sent to the external input unit IN (the destination is the same nanoprocessor element 33).
in the case of). The data packet DP output to the I / O router 32 is further transferred to another destination nanoprocessor element 33 or output multiplexer 40. On the other hand, the data packet DP sent to the external input unit IN is again in the waiting state in the waiting storage unit MM.

The internal configuration and operation of a general nanoprocessor element 33 have been briefly described above with reference to FIG. 3, but as shown in FIG. Many such nanoprocessor elements 33 are provided, and a unique program is loaded for each nanoprocessor element 33. Many data packets DP generated by the packet generation unit 31 are transmitted through the I / O router 32
The data is transferred to various nanoprocessor elements 33, where predetermined processing is performed. That is, in the nanoprocessor element 33, the data field and the instruction field of the data packet DP are updated, or a completely new data packet DP is generated. It will be executed in parallel in the nanoprocessor element 33. Such a data driven parallel processing processor 30 itself is already known, and a description of the basic operation principle of this processor will be omitted.

§3. Processing Performed by Parallel Processors In a general data exchange processing device, usually, segment processing and routing processing are performed. In the segment processing, the received data is analyzed to check for errors, check the lifetime (a limit value of the number of times of passing through the data exchange processing device determined for each protocol), and check whether the data is addressed to itself. If it is confirmed that an error has occurred, the service life has expired, or the data has been addressed to itself, it is inappropriate to transmit the data to the outside. It is determined that there is, and the routing process is not executed. The routing process is a process of determining from which output port the received data is to be transmitted, and is performed only on data determined to be appropriate for transmission to the outside as a result of the segment process.

The processing performed by the parallel processor 30 in the data exchange processing device according to the present invention also corresponds to the above-described segment processing and routing processing. For example, the segment processing and the routing processing for the protocol data unit PDU having the format as shown in the upper part of FIG. 2 are performed by performing a logical operation for separately extracting the data of each part constituting the format, and then performing the data operation of each part. Can be performed by performing a logical operation of comparing with a specific reference data.

For example, the life can be checked by extracting the life value stored at a specific position in the identification information and checking whether or not this value has reached zero. If the lifetime value is 0, the protocol data unit PDU no longer needs to be transmitted to the outside. Therefore, if the data packet DP including such a protocol data unit PDU is subjected to a process of extinguishing it as it is. Good. If the life value is greater than 0, the life value is updated by, for example, 1 and then transmitted to the outside (in this case, the life value is reduced by passing through the exchange processing device). If the initial life value at the transmission source is set to 5, the life will be exhausted by passing through the exchange processing device 5 times).

An error check can be performed in a similar manner. For example, from the data constituting the format as shown in the upper part of FIG. 2, a part of data which is a source of generating an error check code ECC is extracted, and the data is subjected to a calculation based on a predetermined algorithm to perform an error check. A check code ECC may be generated, and a process of checking the consistency with the error check code ECC previously included at the end of the protocol data unit PDU may be performed. If it is confirmed that an error exists in the consistency of the error check code, the data packet DP is processed to be deleted immediately.

Checking whether or not the transmitted protocol data unit PDU is addressed to itself can be performed in a similar manner. For example, if the destination address is extracted from each data constituting the format as shown in the upper part of FIG. 2 and is compared with its own address, the protocol data unit PD
It is possible to recognize whether U is addressed to itself. If it is confirmed that the protocol data unit PDU is addressed to itself, a predetermined process for receiving the transmission target data included in the data packet PDU as data addressed to itself is performed. do it. If the destination address is addressed to the group including itself, after performing processing to receive the transmission target data as the data addressed to itself, it further transfers this to another destination in the group. What is necessary is just to perform a process.

When the above-described segment processing is executed, the protocol data unit PDU is divided into portions. Therefore, as described in §4, the data packet DP including the same protocol data unit PDU is actually used. , A segment process is performed on the copy generated by the copy process, and the data packet DP including the original protocol data unit PDU is left as it is.

After performing the segment processing as described above,
The routing process is performed on the protocol data unit PDU determined to be transmitted to the outside. This routing process can also be executed basically in the same manner as the segment process described above. That is, if the destination address is taken out from each data constituting the format as shown in the upper part of FIG. 2, the final destination of the protocol data unit PDU can be recognized. Accordingly, an operation for specifying a correct output port for the protocol data unit PDU may be performed. Specifically, a table indicating the correspondence between the destination address and the output port is prepared in the plurality of nanoprocessor elements 33, and this table is used in the nanoprocessor element 33 by using this table. An arithmetic process for finding a specific output port corresponding to a specific destination address may be performed. The determined output port is the original data packet DP containing the protocol data unit PDU.
The data is written in the auxiliary field of (the data packet DP that is the source of the duplication used for the segment processing).

When the data packet DP in which the information indicating the output port is written in the auxiliary field is obtained by the routing process, the data packet DP is output to the output multiplexer 40. Output multiplexer 40
Performs a process of extracting data in the data field of the data packet DP output from the I / O router 32 as a protocol data unit PDU, and outputting this to any of the data output interfaces 51 to 55. Which interface to output to can be determined by referring to the information indicating the output port written in the auxiliary field of the data packet DP.

§4. In nanoprocessor elements
Example of Logical Operation As described in §3 above, for example, segment processing and routing processing for a protocol data unit PDU having a format as shown in the upper part of FIG. After performing a logical operation to be separately taken out, it can be performed by performing a logical operation of comparing data of each part with specific reference data. Here, an example of such a logical operation performed in the nanoprocessor element is shown in FIG.
This will be described with reference to the flowchart of FIG.

The flowchart of FIG. 4 shows an example of a logical operation for extracting identification information, ECC, source address, and destination address from the protocol data unit PDU having the format shown in the upper part of FIG. Is shown. In addition, since this flowchart is a flowchart showing the processing of the parallel processing processor, each step is not executed in numerical order, and a plurality of steps are simultaneously executed in a plurality of nanoprocessor elements in parallel. .

First, a protocol data unit PDU having the format shown in the upper part of FIG. 2 is input. Then, in step S1, a process of copying this is performed. Specifically, the same protocol data unit P
Two sets of data packets DP containing the DU in the data field are generated. These two sets of data packets DP have different contents of the generation number field and the instruction field, but have the same data field contents (protocol data unit PDU).
Subsequently, each of the two sets of data packets DP
Duplicated in steps S2 and S3, at this point, four sets of duplicate data packets DP with the same data fields are created. However,
These four sets of duplicated data packets DP are duplicates used to extract identification information, ECC, source address, and destination address, respectively. Actually, the original data packet DP and, as necessary, the original data packet DP Several more copies have been made.

Now, these four sets of duplicate data packets DP
, A shift operation is performed in steps S4 to S7, respectively. The shift amount in the shift operation at this time differs for each step. For example, FIG.
In the case of extracting the identification information located at the head of the protocol data unit PDU having the format shown in (1), the position from the right end of the identification information is L1 bits. Therefore, if the rotation is shifted to the right by L1 bits, the identification information can be moved to the rightmost side as shown in FIG. The shift operation performed in step S4 of FIG. 4 is an operation of rotating the protocol data unit PDU by L1 bits to the right as described above. The identification information position L1 is data indicating the shift amount at this time.

After all, the shift operation in step S4 is performed by the first data packet DP1 including the protocol data unit PDU to be shifted in the data field and the second data packet DP2 including the identification information position L1 in the data field (this Is generated in any one of the nanoprocessor elements 33). That is, the first storage in the waiting storage unit MM in the nanoprocessor element 33 shown in FIG.
Data packet DP1 and second data packet DP2
At the time when both are sent to the logic processing unit ALU, the protocol data unit PDU contained in the data field of the first data packet DP1 is
The rotation shift operation is performed by the amount indicated by the identification information position L1 accommodated in the data field of the second data packet DP2. The same applies to the shift operation performed in steps S5 to S7. That is, by performing a shift operation based on the ECC position L2, the transmission source address position L3, and the transmission destination address position L4, a process of moving the ECC, the transmission source address, and the transmission destination address to the rightmost side is performed.

The AND operation of steps S8 to S11 is performed on the result of the shift operation thus obtained. The AND operation is a process of calculating a logical product with a predetermined mask to replace a portion other than necessary data with bit 0. For example, the result shown in FIG. 5 (b) is obtained by the shift operation in step S4. In order to extract only the identification information from the result, FIG.
In step S8, an AND operation is performed with the identification information mask M1 shown in (c) (data in which only the portion corresponding to the identification information in FIG. 5B is bit 1 and the other portions are bit 0). For example, as shown in FIG. 5D, only the target identification information can be extracted as valid data.

As a result, the AND operation in step S8 includes a first data packet DP1 including the result of the shift operation in step S4 (the data in FIG. 5B) in a data field;
This is a logical operation performed using the second data packet DP2 (which is generated in any of the nanoprocessor elements 33) including the identification information mask M1 (FIG. 5C) in the data field. Step S9
The same applies to the AND operations performed in steps S11 to S11. As a result of these operations, the ECC, the source address, and the destination address are extracted as valid data.

The processing for extracting the identification information, the ECC, the source address, and the destination address from the protocol data unit PDU having the format shown in the upper part of FIG. 2 has been described above. In the segment processing such as checking whether or not the data is addressed to itself, a logical operation may be performed to compare each of the extracted data with predetermined reference data. In the routing processing, the correspondence between the destination address and the output port may be determined. A logical operation may be performed to determine a correct output port to be output using a table indicating the relationship.

Each shift operation and each AND operation shown in FIG.
Which nanoprocessor element performs the operation depends on the program loaded into each nanoprocessor element. As described in the block diagram of FIG. 3, the calculation result in the calculation processing unit ALU is sent to the program storage unit PM. This program storage unit P
M is loaded with a predetermined program in advance, and based on this program, a new instruction field is written, and it is indicated which nanoprocessor element to go next and what instruction to execute. become.
Therefore, for example, the shift operation in step S4 and the AND operation in step S8 shown in FIG. 4 may be executed by the same nanoprocessor element (in this case, the result of the shift operation in step S4 is again Input to the input unit IN), and may be executed by another nanoprocessor element.

§5. Some Modifications Finally, some modifications of the data exchange processing device according to the present invention will be briefly described below.

(1) First, in the above embodiment, information indicating an output port obtained as a result of the routing process is written in the auxiliary field of the data packet DP, and the output multiplexer 40 writes the information in this auxiliary field. Referring to each protocol data unit PDU
Has been sent to the data output interface corresponding to the predetermined output port, but the information indicating the output port is not necessarily the protocol data unit PD.
It is not necessary to write the auxiliary field in the data packet DP containing U. For example, a port indication data packet PDP for transmitting information indicating an output port is newly generated, and the output port of a specific protocol data unit PDU is notified to output multiplexer 40 by the port indication data packet PDP. It is also possible.

As described above, the parallel processor 30
Can analyze the transmission destination address of the protocol data unit PDU included in the data field of the data packet DP to be processed, thereby performing the process of specifying the correct output port for the protocol data unit PDU. . Therefore, for example, if this processing can recognize that the correct output port of the protocol data unit PDU in the data packet DP to which the generation number 1001 is assigned is the second output port Po2, A port instruction data packet PDP containing information indicating that the output port of the protocol data unit PDU included in the data packet DP is the output port Po2 in the data field (the generation number of the data packet PDP is, for example, 2001; ) Is performed. When both the data packet DP with the generation number 1001 and the port specification data packet PDP with the generation number 2001 are prepared, the output multiplexer 40 changes the data field of the port specification data packet PDP. If it is referred to, it is possible to recognize that the output port of the protocol data unit PDU included in the data packet DP of the generation number 1001 is the output port Po2, so that the data packet DP with the generation number 1001 A process of outputting the protocol data unit PDU extracted from the second data output interface 12 to the second data output interface 12 can be performed.

(2) In the above embodiment, the exchange processing of outputting a protocol data unit PDU coming from an input port from a predetermined output port while maintaining the protocol has been described. Thus, the protocol conversion processing can be performed inside the parallel processing processor 30. When the communication protocol changes, the internal structure (order and length of each data, etc.) of the protocol data unit PDU shown in the upper part of FIG. 2 changes. However, such a change in the internal structure occurs in the nanoprocessor element. This is possible by repeatedly executing a predetermined calculation process. As described above, when the protocol of the protocol data unit PDU contained in the data field of the data packet DP is converted, information indicating the protocol of the auxiliary field of the data packet DP is also updated.

(3) In the above embodiments, one protocol data unit PDU is converted into one data packet D
Although contained in the data field of P, depending on the protocol, one protocol data unit PDU may be divided into a plurality of data units and divided and accommodated in the data fields of a plurality of data packets DP. Usually, there is a predetermined upper limit on the length of the data packet DP that can be handled in the parallel processing processor 30. Therefore,
In the case of a protocol in which the total data length of one protocol data unit PDU is considerably long, it is necessary to divide and accommodate the data packets into a plurality of data packets DP.

(4) The plurality of nanoprocessor elements 33 shown in the parallel processor 30 of FIG. 1 are all drawn as elements having the same hardware configuration. It is preferable to configure the parallel processing processor 30 by a set of a plurality of types of nanoprocessor elements suitable for the following. In particular,
Integer type nanoprocessor element suitable for performing integer logical operation and conditional branch judgment, generation number operation nanoprocessor element suitable for generation number operation (read / write, update, etc.), initialization and reference processing for various tables Nano processor element for table operation suitable for etc.
If a plurality of types of nanoprocessor elements are prepared, smooth processing becomes possible.

(5) The upper and lower parts of FIG. 2 show examples of the format of the protocol data unit PDU and the data packet DP, respectively.
The present invention is not limited to such a format. For example, as a format of a data packet DP for a general parallel processing processor, a color field for adding expandability is often defined in addition to a data field, an instruction field, and a generation number field. The auxiliary field in the present invention can be realized, for example, by using a part of the color field. In short, the auxiliary field used in the present invention is any field provided separately from the data field, the command field, and the generation number field that constitute the data packet DP.
Any field is acceptable.

[0061]

As described above, according to the present invention, it is possible to realize a data exchange processing apparatus which can flexibly cope with a multi-protocol and can secure a sufficient transmission efficiency. That is, since a parallel processing processor is used in the central part for performing the switching processing and the parallel processing is performed for each data packet, very efficient processing becomes possible. Each data packet is provided with an auxiliary field, in which information indicating a protocol is stored, so that processing using a program suitable for each protocol becomes possible. Further, even when the protocol for a specific port is changed or when a new protocol is adopted, it is possible to flexibly cope with this by loading a new program into the processor element.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a basic configuration of a data exchange processing device according to an embodiment of the present invention.

FIG. 2 is a diagram showing an example of a format of a protocol data unit PDU used in the present invention and an example of a format of a data packet DP including the protocol data unit PDU.

FIG. 3 is a block diagram showing a basic internal configuration of a nanoprocessor element used in a general data driven parallel processing processor.

FIG. 4 shows an identification information and an EC from a protocol data unit PDU having the format shown in the upper part of FIG.
9 is a flowchart illustrating an example of a logical operation for extracting C, a source address, and a destination address, respectively.

FIG. 5 is a diagram showing data used in the logical operation shown in FIG.

[Explanation of symbols]

 11-15 Data input interface 20 Input multiplexer 30 Parallel processor 31 Packet generator 32 I / O router 33 Nanoprocessor element 40 Output multiplexer 51-55 Data output interface ALU Operation processing unit DP: Data packet ECC: Error check code IN: External input unit L1 to L4: Information indicating the position of each data M1 to M4: Mask information for extracting each data MM: Waiting storage unit OUT: External output unit Pi1 Pi5: Input port PM: Program storage unit Po1 to Po5: Output port PDU: Protocol data unit

 ──────────────────────────────────────────────────続 き Continued on the front page F term (reference) 5B001 AA03 AB01 AC01 AC04 AD06 AE02 5B089 GA31 KA05 KB02 KF05 KF06 5K030 GA03 HA08 HD03 JA01 JA05 KA13

Claims (8)

[Claims] An apparatus for performing a data exchange process between a plurality of input ports and a plurality of output ports, the apparatus being connected to each of the input ports and receiving an input of a protocol data unit having a predetermined communication protocol. A plurality of data input interfaces, each of which is connected to each output port and outputs a protocol data unit having a predetermined communication protocol, and a plurality of data output interfaces which output a given data packet. A data-driven parallel processor that performs parallel processing by using the data input interface; an input multiplexer that provides the protocol data unit input by the data input interface to the parallel processor; and a protocol data unit that is output from the parallel processor. Output the specified data And an output multiplexer that provides a data interface to the data interface.The data input interface has a function of externally inputting a protocol data unit having a specific protocol, and the data output interface has a specific function. The input multiplexer has a function of outputting a protocol data unit having a protocol to the outside, and the input multiplexer supplies information indicating a protocol of the protocol data unit to the parallel processing processor when providing each protocol data unit to the parallel processing processor. The parallel processing processor includes a packet generation unit that generates a data packet to be processed, and a plurality of processors that execute predetermined processing on the data packet in accordance with a specific program loaded in advance. And processor elements, having a function to send each data packet to a predetermined destination I /
An O-router, the packet generator, when a protocol data unit and information indicating the protocol are given from the input multiplexer, an auxiliary field containing information indicating the protocol, Generates a data packet including a generation number field containing a generation number assigned to be unique, an instruction field containing the first execution instruction determined according to the protocol, and a data field containing the protocol data unit. The I / O router has a function of giving each of the given data packets to a specific processor element based on an instruction contained in an instruction field in each of the data packets. Forward to
Alternatively, each processor element has a function of outputting to an output multiplexer, and when a data packet is transferred from the I / O router, each processor element executes an instruction in an instruction field, and executes the execution result and an auxiliary field in an auxiliary field. The output multiplexer has a function of generating a new data packet in which the contents of the data field and the instruction field are rewritten by using a program according to a protocol, and providing the new data packet to the I / O router as necessary. Has a function of extracting data in a data field of a data packet output from the I / O router as a protocol data unit and outputting the data to a predetermined data output interface. apparatus. 2. The data exchange processing device according to claim 1, wherein the parallel processing processor analyzes a transmission destination address of a protocol data unit included in a data field of a data packet to be processed. It has a function of performing an operation of specifying a correct output port for the protocol data unit, and performing a process of writing information indicating this output port to an auxiliary field. An output multiplexer outputs a data packet of the data packet output from the I / O router. A data exchange processing device for determining a data output interface to output a protocol data unit extracted from a data packet by referring to information indicating an output port written in an auxiliary field. 3. The data exchange processing device according to claim 1, wherein the parallel processing processor analyzes a transmission destination address of a protocol data unit included in a data field of a data packet to be processed. A port designating data packet that performs an operation to specify a correct output port for a protocol data unit and includes information for specifying a correct output port for a protocol data unit included in a data packet with a specific generation number And outputs the same to an output multiplexer. The output multiplexer refers to the information of the port designating data packet output from the I / O router to add the specific generation number. Protocol data unit extracted from the Data exchange apparatus characterized by determining a data output interface to be output. 4. The data exchange processing device according to claim 2, wherein a specific destination address is set in a plurality of processor elements based on a table indicating a correspondence between a destination address and an output port. A data exchange processing device provided with a processor element for performing an arithmetic process for obtaining a corresponding specific output port. 5. The data exchange processing device according to claim 1, wherein the parallel processor analyzes data in a protocol data unit included in a data field of a data packet to be processed. Thus, when it is confirmed that the life of the protocol data unit has expired, the data exchange processing device performs a process of deleting the data packet without outputting the data packet to the output multiplexer. 6. The data exchange processing device according to claim 1, wherein the parallel processing processor analyzes data in a protocol data unit included in a data field of a data packet to be processed. When it is confirmed that there is an error in the consistency of the error check code related to the protocol data unit, the data packet is deleted without being output to the output multiplexer. Processing equipment. 7. The data exchange processing device according to claim 1, wherein the parallel processor analyzes data in a protocol data unit included in a data field of a data packet to be processed. Accordingly, if it is confirmed that the protocol data unit is addressed to itself, the data packet is not output to the output multiplexer, and the transmission target data included in the protocol data unit is sent to the data addressed to itself. A data exchange processing device, which performs a process of receiving as a data. 8. The data exchange processing device according to claim 1, wherein the parallel processor separates a communication protocol of a protocol data unit included in a data field of a data packet to be processed. A data exchange processing device having a function of performing a process of converting to a communication protocol.