JPH04270536A - Packet processing unit - Google Patents

Abstract

PURPOSE:To eliminate the need for detailed management of a value able to be set in a field of a generation number/color number by setting various information used for processing a hierarchical protocol to each field as it is. CONSTITUTION:A sub-field 3 distinguishes inter-layers in a stack of each layer protocol formed hierarchically, a sub-field 4 distinguishes a service access point(SAP) for the inter-layers and a sub-field 5 distinguishes a connection end point(CEP) in the SAP. A sub-field 6 distinguishes the transmission reception to the CEP and a sub-field 7 distinguishes the order of primitive informed to the CEP. Then at least one sub-field among the sub-fields 3-7 is provided with a forming means 100 forming at least one field to a packet 1 and a processing means 200 revising the setting of the formed field and progressing the packet processing according to the revised value.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a packet processing device such as a data-driven computer that processes a communication processing program such as a hierarchical protocol program in parallel using data-driven execution.

0002

2. Description of the Related Art FIG. 6 is a diagram showing the structure of a packet used in a conventional data-driven computer disclosed in, for example, Japanese Unexamined Patent Publication No. 1-106229. In the figure, 31 indicates where the packet will go next. Destination flag indicating whether to
is an operand flag indicating whether the instruction is a unary operation or a binary operation, 33 is the generation number/color number, 34 is the destination node number, 35 is the instruction, and 36 is the data on either the left or right side of the binary operation instruction. 37 is first data, 38 is second data, and 39 is a packet composed of these.

FIGS. 2 and 4 are diagrams for explaining a conventional example and an embodiment of the present invention to be described later. FIGS. 2 and 4 each show an example of an internal configuration in a layered protocol. 9 is {N} layer, 10 is {N+1
}It is a layer. SAP also provides services, access,
and CEP indicates the connection end point, 11a is the service access point-1 of the {N} layer, {N} SAP-1, and 11b is the {N} layer.
} layer service access point-2 {
N}SAP-2, 12a is the connection end point-1 of the {N} layer.{N}CEP-1, 12b is the connection end point-2 of the {N} layer.{N}CEP-2, and so on. , 12c is {N}CEP-3, 12d is {N
}CEP-4, 13a is {N+1} SAP-1, 13b
is {N+1} SAP-2, 14a is {N+1} CEP
-1, 14b is {N+1}CEP-2, 14c is {N+
1}CEP-3, 14d is {N+1}CEP-4, 15
a is connection -1 in the {N+1} layer protocol, 15b is connection -2 in the {N+1} layer protocol, and 15c is connection -3 in the {N+1} layer protocol.

Next, the operation will be explained. For example, Figure 2
Consider the case where a program with a layered protocol is processed according to the basic reference model of OSI (Open Systems Interconnection) shown in FIG. {N+1}SAP-
{N+1}CEP-1(14a) in 1(13a)
When issuing a request/response primitive to a packet, the generation number/color number 33 of the packet is set to a value up to the maximum value that can be set in that field, according to a predetermined rule. And {N+1} SAP13a
Since there are a plurality of {N+1} CEPs 14a and 14b within the network, the correspondence between the values set in this way and the connections to be processed is managed using a correspondence table or the like. Also, {N+1}C of {N+1} layer 10
When EP-1 (14a) receives an instruction/confirmation primitive from a certain {N}CEP-1 (12a) in {N}SAP-1 (11a), it also receives the instruction/confirmation primitive in CEP units according to the predetermined rules. A certain value is assigned to the generation number/color number 33 of each packet, and the correspondence between that value and the connection to be processed is also managed using a correspondence table or the like. In other words, for {N+1} primitives processed inside layer 10, {N+1} inside SAP 13a, 13b
1} Request CEP14a, 14b, 14c as a unit/
Values are assigned to the generation number/color number 33 in a system that allows response primitives to be easily controlled in order.

[0005]In addition, independently of this, the instruction/confirmation primitive is also {N}CE in {N}SAP 11a, 11b.
Values are assigned to the generation number/color number 33 in a system that facilitates order control using P12a, 12b, and 12c as units. In this case, the value set in the request/response primitive and the value set in the instruction/response primitive should not match, and the assigned value and {N+1}
The correspondence between the connections 15a and 15b in the layer 10 and the value assignment status are managed.

On the other hand, as shown in FIG. 4, when connections 15a and 15b are demultiplexed in {N+1} layer 10, the generation number assigned to one connection in {N} layer 9 /The value of color number 33 is {N+1} multiple connections 15a in layer 10,
15b, and in order to set the generation number/color number 33 whose order can be controlled for each connection, {N+1} connection 1 in layer 10
The generation number/color number 33 needs to be able to distinguish between 5a and 15b. Therefore, when performing demultiplexing processing, some of the bits that make up the generation number/color number 33 are used only for the purpose of distinguishing connections, and when processing primitives, By not using this, demultiplexing processing is realized.

[0007]

[Problems to be Solved by the Invention] Since the conventional packet processing device is executed as described above, it is difficult to use generation numbers/color numbers with a limited bit width provided for executing numerical calculation programs. When applying the field to a communication processing program, if each layer, each connection, and even each primitive are different, it becomes necessary to set different values to the generation number/color number field in the packet. Therefore, one value must be assigned to this field in a format that takes into account various information in such communication processing. Moreover, the assignment status of these values (assigned or unassigned) must also be managed. In addition, we decided on an assignment rule that makes it easy to implement order control so that memory access processing is executed in the order in which each primitive is received for each connection, and then assigns it to each connection. It is necessary to manage the correspondence with the values that are specified. Furthermore, if you take demultiplexing processing into consideration and allocate bits for this purpose in the generation number/color number/field, the number of values that can be expressed only with the remaining bits will be reduced at the stage of executing primitive processing. There were problems such as insufficient values to allocate.

[0008] This first invention was made to solve the above-mentioned problems, and is based on various information (layers, connections, primitives) and generation numbers/numbers in communication processing.
There is no need to manage the correspondence with the value assigned to the color number field, and the second invention also eliminates the need to manage the correspondence relationship with the value assigned to the color number field. This is to ensure that there is no shortage of expressible values. That is, the object is to obtain a packet processing device that specifies information such as the hierarchical structure and connection setting status in communication processing as a value as is.

[0009]

[Means for Solving the Problems] A packet processing device according to the present invention increases the number of bits of a field that can specify a color, divides this field into several subfields, and adds a layered protocol to these subfields. This allows some information in processing to be set as is.

[0010] That is, the first invention is based on a subfield that distinguishes between layers in a stack of layered protocols, and a service access point (
a subfield that distinguishes a connection end point (CEP) within that SAP;
Forming means for forming at least one field into a packet out of subfields that distinguish transmission and reception to the CEP and subfields that distinguish the order of primitives notified to the CEP, and a forming means that sets a value in the formed field. A processing means is provided for changing settings and proceeding with packet processing based on the changed values.

[0011] The second invention also provides forming means for forming in advance in a packet a subfield that distinguishes a connection within the layer where demultiplexing is performed, and a subfield that is demultiplexed into the formed field. It is provided with a processing means for setting and utilizing a value for identifying a connection.

0012

[Operation] The packet processing device according to the first invention sets various information used when processing a layered protocol in each field with the same value, so that generation number/color number can be set as is. This eliminates the need for detailed management of the values that can be set in fields. Also, because the number of bits in the color specification field has increased,
The number of values that can be expressed in this field increases, and it is possible to generate values that can sufficiently distinguish between parts that can be processed in parallel when executing a layered protocol program.

[0013] The packet processing device in the second invention includes:
Since a field is provided in advance to distinguish connections that are being demultiplexed, other fields are not overwhelmed.

[0014]

[Example] Example 1. An embodiment of the present invention will be described below with reference to the drawings. Figure 1
100 is a forming means, 200 is a processing means, 1 is a packet to which the present invention is applied, 2 is a color field whose field width is expanded according to the present invention, and 3 is for distinguishing between layers in a layered protocol structure. {N} subfield of layer identifier; 4 is subfield of {N} SAP identifier for distinguishing service access points serving the {N+1} layer; 5 is subfield of {N}
}{N} to distinguish between connection endpoints in SAP
CEP identifier subfield, 6 is a sending/receiving identification subfield for distinguishing the flow direction at {N}CEP, and 7 is an order identification subfield for distinguishing the order of primitives notified to {N}CEP. field, 8 is {N
+1} layer for distinguishing connections in the {N+1} layer protocol when demultiplexing processing is performed within the {N+1} layer. This is a subfield of the connection identifier. These subfields are those formed within the color field of the packet by the forming means 100.

FIGS. 2 and 4 are diagrams showing the internal structure of a layered protocol, and since it has already been explained in the conventional example, the explanation will be omitted here. Furthermore, FIGS. 3 and 5 show values set in color field 2 when various primitives are notified. Among these, the arrows shown in the order column and connection identifier column in FIG.
This indicates that the value to be set will be changed midway through. The value changes from the value on the left side of the arrow to the value on the right side.

Next, the operation of data-driven execution in the protocol configuration shown in FIG. 2 will be described using packet 1 having the format shown in FIG. Note that this data drive has a data storage function. This operation explains the operation of the processing means 200, and the processing is based on the fact that the value of color field 2 in packet 1 is reset at any time between each layer by the processing means in each layer. be. First, in the configuration of FIG. 2, {N+1}SAP-1
One {N+1}CEP-1(14a) in (13a)
) is notified of the {N+1} connection setup request primitive, the content of the color field 2 of packet 1 is set to the value 16 in the first line shown in FIG. 3, and the {N+1} layer 10 processing is executed. The request primitive to {N} layer 9 caused by this {N+1} layer 10 processing is {N} CEP in {N} SAP-1 (11a)
-1 (12a), at that point the color field 2 is newly set in the {N} layer 9.

Next, the primitive containing the confirmation for the {N+1} connection setup request is sent to {N}SAP-1(1
If notified to {N}CEP-1 (12a) in 1a), the value 17 in the second row of FIG.
{N+1} layer 10 processing is executed. Then, the {N+1} connection setting confirmation primitive notified to the upper layer by the processing of this {N+1} layer 10 is {N+1} of the {N+1} SAP-1 (13a).
1} When it arrives at CEP-1 (14a), the upper layer receives it and uses the processing means of the upper layer to color and process it.
Reset field 2.

[0018] When {N+1} three data transmission request primitives are notified from {N+1} SAP-1 (13a) to {N+1} CEP-1 (14a) upon completion of the connection setup within the {N+1} layer 10. The values of color field 2 set to 1 are shown in lines 18, 19, and 20 in the third to fifth lines of FIG. Also, {N}SAP-1(
11a) for {N}CEP-1 (12a) {N+
1} The value of color field 2 that is set when a primitive including data reception is notified is shown in the sixth line 21 of FIG. In FIG. 3, the value of the connection identifier is always set to 0, as will be explained later in Example 2, because the configuration shown in FIG. 2 does not perform demultiplexing processing. Moreover, the order value in FIG. 3 is
Sending and receiving are treated independently, and the order in which they are notified is shown.

On the other hand, the processing means 200 in the configuration of FIG.
The operation will be explained below. {N+1}SAP-1
{N+1}CEP-1 (14a) in (13a)
+1} When the connection setup request primitive is notified, the content of color field 2 is set to the value 22 in the first row shown in FIG. 5, and {N+1} layer 10 processing is executed. Similarly, {N+1}{N in SAP-1 (13a)
+1} When the {N+1} connection setup request primitive is notified to CEP-2 (14b), the content of color field 2 is set to the value 23 in the second line shown in FIG. 5, and {N+1} layer 10 processing is performed. I do. These {N+1
}{N} layer 9 caused by layer 10 processing
When request primitives to {N} are sequentially notified to {N} CEP-4 (12d) of {N} SAP-1 (11a), at that point a new color field 2 is set in {N} layer 9 processing. Try again.

Next, two primitives containing confirmations for the two {N+1} connection setup requests are {N}SA
When notified to {N}CEP-4 (12d) in P-1 (11a), the third line 24 and the fourth line 2 shown in FIG.
Set the value of 5. However, at this point, the value of the order subfield and the value of the connection identifier subfield are the values on the left side of the arrow, and {N+1} layer 1
When the processing means 200 determines that demultiplexing is occurring while proceeding with the processing within 0, both values are changed to the values on the right side of the arrow. And this {N+1} layer 1
0 processing, the two connection setup confirmation primitives are {N+1}C of {N+1}SAP-1 (13a)
EP-1 (14a) and {N+1}CEP-2 (14b)
At the stage where the color field 2 is notified, the processing means of the upper layer sets a new value in each color field 2.

[0021] {N+1} With the completion of the connection setting in layer 10, {N+1} SAP-1 (13a)
{N+1}CEP-1 (14a) and {N+1}CEP
The values of color field 2 that are set when the {N+1} data transmission request primitive for -2 (14b) is notified are shown in the 5th line 26, the 6th line 27, and the 7th line 2 in Figure 5.
8. Also, {N} of {N} SAP-1 (11a)
The values of color field 2 that are set when a primitive including {N+1} data reception is notified to CEP-4 (12d) are shown in the 8th line 29 and the 9th line 3 in Figure 5.
0. However, at this point, the value of the order subfield and the value of the connection identifier subfield are the values on the left side of the arrow, and demultiplexing occurs at the stage when the processing means in the {N+1} layer 10 proceeds with the processing. When it is determined that the value on the left side of the arrow is correct, the value on the left side of the arrow is changed to the value on the right side of the arrow.

Example 2. In the first embodiment, the color field has six subfields, but the six subfields do not need to be included in the color field. Furthermore, the six fields do not need to be consecutive and may be located at separate locations. Furthermore, 6
It is not necessary to have all three fields; it is sufficient that at least one field is independent from the other fields. In the first invention, any one of the first five fields among these six fields is provided in advance, and in the second invention, in particular, a connection identification field is provided. In addition, in the above-mentioned Example 1, only the case of demultiplexing was explained, but if demultiplexing is the case where one layer is viewed from another layer, conversely, one layer can be viewed from the other layer. Since this is multiplexing, it can also be applied to multiplexing.

[0023]

As described above, according to the present invention, various information in the processing of the layered protocol can be directly set in fields in the packet format, so that parallelism for each layer and parallelism for each connection can be achieved. , it has the effect of allowing data-driven parallel processing to be easily and naturally executed without the programmer having to consciously manage the parallelism of each primitive in detail in correspondence with generation numbers/color numbers. .

[Brief explanation of the drawing]

FIG. 1 is a diagram showing a packet processing device for a data-driven computer according to an embodiment of the present invention.

FIG. 2 is a diagram showing the internal structure of a layered protocol.

FIG. 3 is a diagram showing values set in each field.

FIG. 4 is a diagram showing the internal structure of a layered protocol.

FIG. 5 is a diagram showing values set in each field.

FIG. 6 is a diagram showing a conventional packet format.

[Explanation of symbols]

1 packet 2 Color field 3 {N} layer identifier 4 {N} SAP identifier 5 {N}CEP identifier 6 Flag indicating sending or receiving 7. Order

Claims (2)

[Claims] Claim 1: A packet processing device having the following elements: (a) a forming means for forming a packet by providing in advance at least one field among the following fields; (a1) distinguishing between layers by layered protocol processing; (a2) an access point identification field for distinguishing access points between layers, (a
3) A connection end point identification field that distinguishes a connection end point within an access point; (a4) a transmission/reception field that distinguishes sending and receiving events to and from a connection end point;
a5) an order identification field for controlling the order of events; (b) processing means for changing the setting of the field value of the packet formed by the forming means and processing the packet according to the changed value. [Claim 2] A packet processing device having the following elements: (a) Forming a packet by providing in advance a connection identification field in the packet to identify the connection in the layer where connections are being multiplexed or demultiplexed; (b) processing means for identifying a multiplexed or demultiplexed connection in a layer performing connection multiplexing or demultiplexing, and setting the identified value in a connection identification field;